NBS SPECIAL PUBLICATION 493 U.S. DEPARTMENT OF COMMERCE/ National Bureau of Standards NEUTRON STANDARDS AND APPLICATIONS PROCEEDINGS OF A SYMPOSIUM Digitized by the Internet Archive in 2013 http://archive.org/details/neutronstandardsOOinte CfS. /£> • V93 Neutron Standards and Applications Proceedings of the International Specialists Symposium on Neutron Standards and Applications Held at the National Bureau of Standards Gaithersburg, MD, March 28-31, 1977 Edited by CD. Bowman and A.D. Carlson Center for Radiation- Research Institute for Basic Standards National Bureau of Standards Washington, D.C. 20234 H.O. Liskien Central Bureau of Nuclear Measurements Commission of European Communities B-2440 Geel Steenweg Naar Retie BELGIUM and L. Stewart Los Alamos Scientific Laboratory Los Alamos, NM 87545 Sponsored by National Bureau of Standards U.S. Department of Commerce Central Bureau of Nuclear Measurements Commission of European Communities U.S. Energy Research and Development Administration Electric Power Research Institute (U.S.A.) American Nuclear Society American Physical Society Nuclear Energy Agency (Europe) International Union of Pure and Applied Physics with the Cooperation of the International Atomic Energy Agency V U.S. DEPARTMENT OF COMMERCE, Juanita M. Kreps, Secretary Dr. Sidney Harman, Under Secretary Jordan J. Baruch, Assistant Secretary for Science and Technology NATIONAL BUREAU OF STANDARDS, Ernest Ambler, Acting Director Issued October 1977 Library of Congress Cataloging in Publication Data International Specialists Symposium on Neutron Standards and Applications, National Bureau of Standards, 1977. Neutron standards and applications. (NBS special publication ; 493) Supt.ofDocs.no.: C13.10:493 1. Neutrons— Standards— Congresses. I. Bowman, Charles D., 1935- II. United States. National Bureau of Standards. III. Title. IV. Series: United States. National Bureau of Standards. Special publication ; 493 QC100.U57 no. 493 [QC793.5.N4622] 602M s 77-14317 [539.7'213'021] National Bureau of Standards Special Publication 493 Nat. Bur. Stand. (U.S.), Spec. Publ. 493, 379 pages (Oct. 1977) CODEN: XNBSAV For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 - Price $8.60 Stock No. 003-003-01847-3 PREFACE These are the proceedings of an International Specialists' Symposium on Neutron Standards and Applica- tions held at the National Bureau of Standards, Gaithers- burg, Md. from March 28-31, 1977. The field of neutron standards now spreads across a rather broad range of subjects including absolute neutron source strength, differential and integral standards for reaction cross sections, and standards for neutron dosimetry for nuclear reactor core and containment, personnel protection, cancer therapy, and fission reactor design. The purpose of the Symposium was to bring together workers from this spectrum of fields for the purpose of reviewing the present status, tying the different branches more closely together, and plan- ning for a more coordinated international program in the future. In order to achieve the broad and even coverage of the field, the papers were presented by invitation only although the conference was open to anyone. In spite of this and its billing as a symposium for special- ists, 142 registrants were recorded with 34 participants from 13 foreign countries. The symposium closed with a summary session which was recorded and is reproduced in full in these proceedings for those interested in comments and review of the full conference. The papers are printed in the proceedings as they were received from the authors and in the order in which they were presented in the sessions. For con- venience we have preserved the conference notation for the sessions. To speed the publication of the proceed- ings, all papers were submitted by the authors in camera-ready form. We are greatly indebted to the authors and all those who assisted in the preparation of the manuscripts. Their efforts have made it possible to get the proceedings in print much more rapidly than would otherwise be the case. To make the proceedings more useful we include, as well as a table of contents, an author index, a list of participants, and a CINDA index of subject matter. We would like to thank Dr. Charles Dunford of Brookhaven National Laboratory for the preparation of the CINDA index. When commercial equipment, instruments and mate- rials are mentioned or identified in this proceedings it is intended only to adequately specify experimental procedure. In no case does such identification imply recommendation or endorsement by the National Bureau of Standards, nor does it imply that the material or equipment identified is necessarily the best available for the purpose. We wish to express our appreciation for the financial support from the U. S. Energy Research and Development Administration, The Central Bureau of Nuclear Measurements and the National Bureau of Standards which made the publication possible. The Editors gratefully acknowledge the assistance of the National Bureau of Standards Office of Technical Publications in the preparation of these proceedings and of Mrs. Sara Torrence and Mrs. Lynn Boggs of the Office of Information Activities for help in the arrangements for the Symposium. We are appreciative also of the advice and suggestions of Dr. Bryan Patrick, Harwell and of the excellent secretarial assistance of Mrs. Julia Marks, Mrs. Gracie Wood, and Mrs. Sarah Stewart. C. D. Bowman and A. D. Carlson NBS H. 0. Liskien CBNM L. Stewart LASL III PROGRAM COMMITTEE M. R. Bhat, BNL J. B. Czirr, LLL F. Gabbard, Univ. of Kentucky W. W. Havens, Jr., Columbia University B. R. Leonard, Jr. , BNW H. Li ski en, CBNM R. W. Peelle, ORNL A. B. Smith, ANL P. G. Young, LASL C. D. Bowman, NBS, Chairman IV ABSTRACT These proceedings contain forty-seven papers, which were presented at the International Specialists Symposium on Neutron Standards and Applications held at the National Bureau of Standards on March 28-31, 1977. The topics addressed at the Symposium include light-element cross section standards, capture and fission cross section standards, integral neutron standards, flux measuring techniques, and medical and personnel dosimetry. Key words: Cross section standards; dosimetry; fission; flux; measuring techniques; neutrons; standards. TABLE OF CONTENTS Contents Page Preface Ill Program Committee IV Abstract V Table of Contents VI Papers: Session A through I 1 -346 Summary Session 347 -359 List of Participants 360 -363 Author Index 3g4 Citation Index 365 .370 SESSION A: INTRODUCTORY REMARKS Chairman: C. D. Bowman A. 1. Welcome by Ernest Ambler 1 A. 2. Remarks by H. Liskien 2 SESSION B: THE 6 Li(n,a) STANDARD* Chairman: P. G. Young B. 1. Survey of Recent Experiments for the 7 Li-System by H.-H. Knitter 3 - 9 B. 2. Angular Anisotropy in the b Li(n,a) 3 H Reaction Below 100 keV by J. A. Harvey and I. G. Schroder 10 - 13 B. 3. Experimental Data Base for the Li-7 System by H. Derrien and L. Edvardson 14 - 29 B. 4. R-Matrix Analysis of the 7 Li System by G. M. Hale 30 - 36 B. 5. Special Problems with 6 Li Glasses by G. P. Lamaze 37 - 42 B. 6. Instruments for Use of 6 Li as a Standard by L. W. Weston 43 - 46 SESSION C: LIGHT ELEMENT STANDARDS OTHER THAN 6 Li Chairman: S. W. Cierjacks C. 1. Experiments and Theory for Differential n-p Scattering by C. A. Uttley 47 - 53 C. 2. Use of the n,p Scattering Reaction for Neutron Flux Measurements by J. B. Czirr 54 - 60 C. 3. Surface Barrier Spectrometers for Calibration of Fast Neutrons in MeV Range by 0. P. Joneja, R. V. Srikantaiah, M. R. Phiske, J. S. Coachman and M. P. Navalkar 61 - 66 C. 4. Review of 10 B(n,a) 7 Li Cross Section Measurements in the Energy Range From 10 keV to 1 MeV by E. Wattecamps 67 - 84 C. 5. Instruments for Use of 10 B as a Standard by A. D. Carlson 85 - 92 C. 6. Evaluation and Use of Carbon as a Standard by J. C. Lachkar 93 -100 SESSION D: MEDICAL AND PERSONNEL NEUTRON DOSIMETRY Chairman: J. L. Fowler D. 1. Need for Improved Standards in Neutron Personnel Dosimetry by John A. Auxier 101 -105 D. 2. Standards in Medical Neutron Dosimetry by J. J. Broerse 106 -114 D. 3. NBS Facilities for Standardization of Neutron Dosimetry From 0.001 to 14 MeV by 0. A. Wasson 115 -120 D. 4. International Neutron Dosimetry Intercomparisons by R. S. Caswell 121 -127 The session on 6 Li was planned for the program committee by the Nuclear Energy Agency Nuclear Data Committee as a consequence of its strong interest in this standard and its associated uncertainties. VI SESSION E: FISSION REACTOR INTEGRAL NEUTRON STANDARDS Chairman: A. B. Smith E. 1. Reactor Core Dosimetry Standards by Willem L. Zijp 128 - 136 E. 2. Standards for Dosimetry Beyond the Core by Frank J. Rahn, Karl E. Stahlkopf, T. U. Marston, Raymond Gold and James H. Roberts 137 - 145 E. 3. Fission Yields: Measurement Techniques and Data Status by W. J. Maeck 146 - 155 E. 4. Fission Reaction Rate Standards and Applications by J. Grundl and C. Eisenhauer .... 156 - 164 SESSION F: FISSION STANDARDS I Chairman: J. A. Harvey F. 1. Utility and Use of Neutron Capture Cross Section Standards and the Status of the Au(n,Y) Standard by A. Paulsen 165 - 169 F. 2. Remarks on the 2200 m/s and 20°C Maxwell ian Neutron Data for U-233, U-235, Pu-239 and Pu-241 by H. D. Lemmel 170 - 173 F. 3. An Assessment of the "Thermal Normalization Technique" for Measurement of Neutron Cross Section vs Energy by R. W. Peelle and G. de Saussure 174 - 181 252 F. 4. Review of v For Cf and Thermal Neutron Fission by J. W. Boldeman 182 - 193 252 F. 5. Measurement of the Cf Spontaneous Fission Neutron Spectrum by M. V. Blinov, V. A. Vitenko, and V. T. Touse 194 - 197 F. 6. Prompt Fission Neutron Spectra by Leona Stewart and Charles M. Eisenhauer 198 - 205 F. 7. On Quantitative Sample Preparation of Some Heavy Elements by A. H. Jaffey 206 - 211 SESSION G: CROSS SECTION - INDEPENDENT FLUX MEASUREMENT TECHNIQUES Chairman: R. C. Block G. 1. Black and Grey Neutron Detectors by F. Gabbard 212 - 220 G. 2. Associated Particle Methods by Michael M. Meier 221 - 226 G. 3. Associated Gamma-Ray Technique for Neutron Fluence Measurements by J. D. Brandenberger 227 - 233 G. 4. Associated Activity Method by K. K. Sekharan 234 - 236 G. 5. Accuracies and Corrections in Neutron Bath Techniques by E. J. Axton 237 - 243 G. 6. International Comparison of Flux Density Measurements for Monoenergetic Fast Neutrons by V. D. Huynh 244 - 249 G. 7. Calibration and Use of Filtered Beams by R. B. Schwartz 250 - 254 45 56 G. 8. Much Ado About Nothing: Deep Minima in Sc and Fe Total Neutron Cross Sections by R. E. Chrien, H. I. Liou, R. C. Block, U. N. Singh, and K. Kobayashi 255 - 260 SESSION H: FISSION STANDARDS II Chairman: G. Bartholomew H. 1. The U-235 Neutron Fission Cross Section From 0.1 to 20.0 MeV by W. P. Poenitz 261 - 268 H. 2. Propagation of Uncertainties in Fission Cross Section Standards in the Interpretation and Utilization of Critical Benchmark Measurements by C. R. Weisbin and R. W. Peelle. . 269 - 277 H. 3. 237 Np and 238 U as Possible Standards for the MeV Region by S. Cierjacks 278 - 289 H. 4. Standard Integral Measurement Facilities by A. Fabry 290 - 298 H. 5. Integral Measurement Results in Standard Fields by D. M. Gilliam 299 - 303 H. 6. Absolute Fission Cross Section Measurements Using Fixed Energy Neutron Sources by G. F. Knoll 304 - 309 VII H. 7. Impact of ENDF Standards on Fast Reactors by Ugo Farinelli 310 - 312 poc ?^$ ?^7 H. 8. Absolute U, U, Np Fast Neutron Fission Cross Section Measurements by V. M. Adamov, B. M. Alexandrov, I. D. Alkhazov, L. V. Drapchinsky, S. S. Kovalenko, 0. I. Kostochkin, G. Yu. Kudriavzev, L. Z. Malkin, K. A. Petrzhak, L. A. Pleskachevsky, A. V. Fomichev, V. I. Shapakov 313 - 318 SESSION I: SPECIAL TOPICS Chairman: R. F. Taschek I. 1. Neutron Energy Standards by G. D. James 319 - 328 I. 2. Neutron Transport Calculations for the Intermediate-Energy Standard Neutron Field (ISNF) at the National Bureau of Standards by C. M. Eisenhauer, J. A. Grundl and A. Fabry 329 - 334 I. 3. Standardization of Fast Pulse Reactor Dosimetry by A. H. Kazi , E. D. McGarry and D. M. Gilliam 335 - 341 I. 4. Dosimetry Standards for Neutrons above 10 MeV by H. H. Barschall 342 - 346 SUMMARY SESSION 347 . 359 List of Particioants 360 _ 353 Author Index 354 Citation Index .._ „_„ 365 - 370 VIII INTRODUCTORY REMARKS Session A: Dr. Ernest Ambler, Acting Director National Bureau of Standards It is indeed a pleasure for me to welcome you to the National Bureau of Standards for this Symposium. I understand that 15 countries are represented and that about one-third of you are from outside of the United States. I strongly believe that almost every aspect of basic standards requires close international cooperation and collaboration and that such interactions will be established and strengthened at this meeting. I might add that here at NBS we place emphasis on the international aspect of all our work. I am proud of the strong line of communication that we maintain around the world through conference sponsorship and participation, laboratory visits, corroborative activities and committee work in international organizations in setting standards. Dr. Charles Bowman's fine service as chairman of this Symposium is just one example - an example that I am very proud of - of our activity in this area. The international flavor and significance of improved neutron standards are clearly reflected in the endorsements which this Symposium has received. I note the endorsements of four international organizations - The Central Bureau of Nuclear Measurements, The Nuclear Energy Agency (Europe), The International Union of Pure of Applied Physics, and the International Atomic Energy Agency. Within the United States, the Symposium has been endorsed by four divisions of the Energy Research and Development Administration, the industry-operated Electric Power Research Institute, the American Physical Society, the American Nuclear Society, and of course by the National Bureau of Standards. These endorsements reflect the importance and diversity of the field of neutron standards. For energy production by fission, a knowledge of neutron cross sections to high accuracy is central. In fusion-energy sources, at least in the near term, nearly all of the energy is carried by neutrons. Fuel for both fission and fusion systems is bred using neutrons. Many medical specialists believe that high-energy neutrons provide an effective modality for cancer therapy. Neutrons are now commonly used in activation analysis and radiography. The need for safe working conditions, in all of these fields, requires accurate neutron personnel dosimetry. The National Bureau of Standards has long been aware of the need for standards in the field of neutron based technology. Our program goes back more than 20 years to the NBS initiative in developing the radium-beryllium neutron source standard widely known as NBS-I. Increased efforts in this field during the late fifties and early sixties led to the first symposium in this country on the subject of neutron standards. Sponsored by the European-American Nuclear Data Committee, the meeting took place in 1970 at the Argonne National Laboratory. This was followed by the IAEA-sponsored panel which met in Vienna, Austria in 1972 and further delineated the challenges in this field. Here at NBS we were increasingly recognizing the importance of this field. While Director of the Institute of Basic Standards of NBS, I strongly encouraged a growing program in neutron standards within our Center for Radiation Research. Our work has broadened into a program in both differential and integral neutron standards carried out using such major facilities as our 3-MeV positive ion Van de Graaff accelerator, our 10-MW reactor, our 100-MeV electron linac, and several other specialized facilities. Our program now spans nine decades of neutron energy from subthermal to 20 MeV. We can count eight calibrated neutron fields which are available as flux standards for outside use. We have on-going, active programs in support of the full spectrum of neutron applications including fission nuclear energy programs, fusion energy, medical and personnel neutron dosimetry. I hope that you will be able to get an opportunity to visit our experiments and facilities in spite of the busy schedule of the Symposium. I am justifiably proud of our commitment at NBS to neutron standards, but I strongly believe that our program should be a partnership with other neutron standards activities both within and outside the U.S. Neutron work in general is perhaps the most technically difficult area of work in ionizing radiation. Progress in neutron standards often comes slowly and after much intercomparison and corroboration. Therefore, I believe that we all recognize a unique opportunity here to advance the field. Let us hope that some problems will be laid to rest, that progress in international coordination and collaboration can be made in others, and that the new problems in neutron standard will be recognized. You now face a very full four-day schedule. Let me again welcome you here and wish you much success. 1 Dr. H. Liskien, Chairman INDC Subcommittee on Standard Reference Data There are still many scientific/technical fields where essential effort has to be devoted to the change of units and the harmonization of standards. This is due to regionally independent approaches in the past with results which are insufficient for the international interchanges of today. In contrast to this, a modern field like neutron technology - and especially the data aspect of it - has been an international enterprise nearly from the beginning. Today we have WRENDA, an international request list for neutron data, we have CINDA, an international index to literature containing information on microscopic neutron data, and we have under the EXFOR agreement an international network for the compilation, exchange and retrieval of experimental neutron data. It is therefore not at all astonishing that also the task to establish a set of neutron data standards has seen an international approach. In a month's time it will be ten years since the IAEA convened a panel on "Nuclear Standards for Neutron Measurements" in Brussels. In 1970, the European-American Nuclear Data Committee organized a symposium on "Neutron Standards and Flux Normalization", while the 2nd IAEA panel on "Neutron Standard Reference Data" was held in Vienna four and a half years ago. In this sense we are here celebrating the opening of the fourth international meeting on this subject. In such a meeting, we try to compare critically our established set of standards with the real needs, to assess the progress in methods and accuracy and, to identify weaknesses to be eliminated in the future. The progress itself is of course achieved between the meetings in our home laboratories by improving detectors, performing accurate measurements and evaluating best values. But also this work should see more international cooperation. In this respect two good examples are taken from the recent past: l)the first round of comparing results on fluence determi- nations between various laboratories organized by the Bureau International des Poids et Mesures and 2)the work of an INDC ad-hoc group to establish a set of neutron energy standards. I think it would be most fruitful if this symposium could initiate more such common attacks on crucial points in the field of "Neutron Standards and Applications". SURVEY OF RECENT EXPERIMENTS FOR THE Li-SYSTEM H. -H. Knitter Central Bureau for Nuclear Measurements B-2440 Geel, Belgium Recent experiments on reactions relevant to the Li-system are described. It concerns the reactions 6 Li(n,t) 4 He, 6 Li(n,n) 6 Li, 4 He(t,n) 6 Li and 4 He(t,t) 4 He, for which differential cross sections v 2 \ ^^ > ^-*-^C!!^~*-~-»- — <*T 1 — Od 01 0.6 07 Fig. 7 03 0.4 as Neutron Energy [MeV] Neutron total, integrated elastic and (n,t) cross sections of &Li versus incident neutron ener- gy 1 **. Aandiare data of ref. 22 and 23 respec- tively. 4 4 The He(t,t) He scattering experiments 7 This entrance channel to the Li-system consists of a spin-zero and a spin-l/2 particle and therefore the interpretation of the data is rather easy. Earlier studies of the triton - a scattering process have con- tributed considerably to the understanding of the ?Li- system" ' L0 • L '. Recent measurements did not come to my knowledge. There are unpublished data in the triton energy range from 7 to 14 MeV from the Los Alamos tandem Van de Graaff facility displayed in gra- phical form in the work of Hale ° on the R-matrix ana- lysis of light element standards. However, these data will be published soon24 under ref. 29. Experiments using polarized projectiles Studies of polarization effects can provide infor- mation of a basic nature, which can be obtained only indirectly or sometimes not at all using traditional ex- perimental methods-^. This was demonstrated alrea- dy in the first nuclear polarization experiment which was made by Heusinkveld and Freier in 195 1 at the University of Minnesota 3 ' . This 4He-proton double scattering experiment gave the level sequence for the P-state doublet of the ^Li* nucleus, an information which could not be obtained by conventional scattering ■a? experiments . The polarization power of Li or neutrons was measured at the electron Linac of the Yale University ' . Fig. 8 shows the schematics of the experimen- tal set-up. A beam of unpolarized neutrons is produ- ced at the (y ,n)-target of the Linac. The neutrons be- come partially polarized by the first scattering on car- bon. In a second scattering the analysing power A(@) of °Li and of other nuclei were measured in the ener- gy range from 2 to 5 MeV. The polarization of the in- cident neutrons was obtained from a separate scatte- ring experiment in which the polarizer and the analy- zer consisted of the same material of spin zero nuclei, in which case the analyzing power is equal to the po- larization power-^. The solenoid which is positioned in the flight path served to precess the neutron spin, scattering cross section data of Lane et al. 22 and their own polarization results. The differential scattering cross section for unpolarized neutrons 0(6) and the po- larization can be given in form of expansions 0(6) = \ 2 Y B L P L (cos 6) a(e) . p(6)=* 2 Ic l p[(cos 9) L where P L (cos 9) and PJ! (cos 9) are the Legendre and associated Legendre polynomials. The Bl's as func- tions of the collision matrix are derived by Blatt and Biedenharn 3 -' with the phase correction of Huby36 anc j the Cl's are given as function of the collision matrix by Simon and Welton 3 '. The analysing power for the 4 He(t,t)4He reaction was measured by a team of the Los Alamos tandem Van de Graaff facility^ 8 > 3 9 a t centre-of-mass angles » 0.0 7.0 8.0 9.0 10.0 11.0 I2.0 13 14 D Triton Energy (MeV) Fig. 9 : Excitation function for 4 He(t, t) 4 He analyzing power 38 Graakilc Sc«lt.r«i (Cylindrical Shall On a-la.s I cm Ihlct) Oiyaaa Seallarar • (C,IIMr leal Shall 7.S c» «!«.» ten IB let I Plaallc Sclarlllatart - (It.Scm ala.olSca Ihlct) Fig. 8 Experimental arrangement for measuring the analyzing power in neutron scattering experi- ments at the Yale-Univer sity32_ and the analyzing power could be obtained from the count rate ratios with and without magnetic field and from the spin precession angle. Firk et al. ^3 , 34 made a R-matrix analysis for the 7 Li system, using the TIME PICK-OFF PROTON BEAM T Li TARGET X 10 AMP BEAM MONITOR a INTEGRATOR Li TARGET <■''' \ -\\--' 12 3 4 SCALE INCHES PRE-AMP Fig. 10 : A plan view of the experimental arrangement to measure the analyzing power of the 6 Li(n,t) 4 He reaction. of 49. 6° and 123° in the energy range from 7 to 14 Me V In addition angular dependence of the polarization po- wer at 8. 79 MeV, 10. 82 MeV and 12. 25 MeV was ob- tained. The polarized tritons were produced with the Los Alamos Lambshift polarizing ion source for tri- tium*™. At the target of the accelerator a beam cur- rent of 80 nA with a polarization of 0. 8 has been ob- tained. Fig. 9 shows a part of the results. The reso- nance near 8.8 MeV is of particular interest, since it corresponds to the resonance near 240 keV in the °Li(n,t) 4 He reaction. The analyzing powers of the Li(n,t) 4 He reaction was measured by Karim and Overley 41 with the Uni- versity of Oregon pulsed Van de Graaff accelerator in the neutron energy range from 0.2 to 1.4 MeV. Fig. 10 shows the experimental arrangement. With a thick metalic lithium target polarized neutrons are produ- ced in the energy range between 0. 2 and 1. 4 MeV at the angles of + 50 degrees with respect to the proton beam. The a and triton particles were detected in a solid state detector and two dimensional spectra were recorded in energy and flight time. This allowed to determine the neutron energy and to make a particle identification. The results compare also favourably ■with previous results 4 "- at two energies near the 240 keV resonance. Conclusion Recent measurements of the Li(n,t) 4 He cross section were done in the incident neutron energy range from some keV to several hundreds of ke V > 3 and from 2 to 14 MeV 4 >5. Neutron total cross section mea- surements were made' ' > '■" and are being made' ' in the neutron energy range from 10 eV to 10 MeV, where special emphasis has been paid to the region of the 250 keV resonance. Recent neutron scattering cross section measurements were performed in different ex- periments from 0. 2 to 3. MeVl°, from 4.0 to 7.5 MeV 20 and from 7. 5 to 14 MeV 21 . The gap between 2 and 3 MeV is being closed' ' such that the whole ener- gy range from 200 keV to 14 MeV is being covered by recent scattering experiments. Zero and 180° degree centre-of-mass cross sections of the reaction 'H( a, ^Li) are measured and will be available soon 1 -'*. Similar is the situation with 4 He(t,t) 4 He scattering experiments ^4,29, Also analyzing power measurements for the 6 Li(n,n) 6 Li, 4 He(t,t) 4 He and 6 Li(n,t) 4 He were done in the laboratory energy ranges from 2 to 5 MeV 33 > 34 , from 7 to 14 MeV 38 ' 39 and from 0. 2 to 1.4 MeV 41 respectively. All this recent experimental results re- present a large and valuable body of data for under- standing of the 'Li-system. References 1. F. C. Barker, Phil. Mag. 2 (1957) 780 2. G. P. Lamaze , O. A. Was son , R. A. Schrack and A. D. Carlson , Proc. of the Int. Conf. on the Inter- actions of Neutrons with Nuclei, Lowell, Massachu- setts, 6-9 July, 1976, Vol.2, p. 1341 3. D. B. Gayther , AERE-Harwell , private communica- tion 12. 1. 1977, will be published as report AERE- 8556 (1977) 4. CM. Bartle , Proc. of the Conf. on Neutron Cross Sections and Technology , Washington D. C. ,1975, Vol. 2, p. 688, NBS Special Publication 425 5. C. M. Bartle, D.W. Gibbie and C. Hollas, Proc. of the Int. Conf. on the Interactions of Neutrons with Nuclei, Lowell, Massachusetts, 6-9 July, 1976, Vol. 2, p. 1342 6. O. A. Wasson.Proc. of the NEANDC/NEACRP Spe- cialists Meeting on Fast Fission Cross Sections, 28-30 June 1976, held at Argonne National Labora- tory, ANL-76-90 p. 183 7. M.G.Sowerby, B. H. Patrick and D. S. Mather , Ann. Nucl. Sci. Eng. 1 (1974) 409 8. C. M. Bartle and P. A. Quin, Nucl. Instr. Meth. 121 (1974) 119 9. E. D. Pendlebury, AWRE 0-60/64 (1964) 10. P. J. Clements and I. C. Rickard, AERE-R 7075 (1972) 11. J. C. Overley.R. M. Sealock and D. H. Ehlers, Nucl. Phys. A 221 (1974) 573 12. E. Rosario-Garcia and R. E. Benenson, Nucl. Phys. A 275 (1977) 453 13. F. S. Goulding, D.A.Landis, J. Cerny III and R. H. Pehl, IEEE Trans. Nucl. Sci. 11_ (1964) 388 14. H. Bichsel and C. Tschalaer, UCRL-17663 (1967) 15. I.G.Schroder, E. D. McGarry, G.de Leeuw-Gierts and S. de Leeuw, Proc. of the Conf. on Neutron Cross Sections and Technology, Washington D. C. , 1975, Vol. 1 , p. 240 16. S. Raman, N. W. Hill, J. Halperin and J. A. Harvey , Proc. of the Int. Conf. on Interactions of Neutrons with Nuclei, Lowell, Massachusetts, 6-9 July, 1976, Vol. 2, p. 1340 17. J. A. Harvey and N. W. Hill, Proc. of the Conf. on Neutron Cross Sections and Technology, Washington D. C. , 1975, Vol. 1 , p. 244 18. H. -H. Knitter , C. Budtz- J^rgensen, M. Mailly and R. Vogt, Report EUR-5726e (1977) 19. A.B.Smith, private communication January 24 , 1977 Argonne National Laboratory 20. R.O.Lane, R. M. White and H. D. Knox, Proc. of the Int. Conf. on the Interactions of Neutrons with Nu- n clei, Lowell, Massachusetts, 6-9 July, 1976, Vol. 2, p. 1043 2 1. E.G. Bilpuch, D.H.Epperson, D.W.Glasgow, S. G. Glendinning, C.R.Gould, H.H.Hogue, P.W.Li- sowski, C.E.Nelson, H. W. Newson, F. O. Purser , L. S. Seagondollar , T. Tornow and P. von Behren, Proc. of the Int. Conf. on the Interactions of Neu- trons with Nuclei , Lowell, Massachusetts , 6-9 July, 1976, Vol. 2, p. 1309 22. R. O. Lane , A. S. Langsdorf Jr. , J. E. Monahan and A.J. Elwyn, Report-ANL-6l 72 , and Ann. Phys. j_2 (1961) 135 23. H. B.Willard, J. D. Bair, J. D. Kington and H. O. Cohn Phys. Rev. |0J_ (1956) 765 24. N.Jarmie, Los Alamos Scientific Laboratory, pri- vate communication, January 10, 1977 25. T. A. Tombrello and L. S. Senhouse , Phys. Rev. 129 (1963) 2252 26. R. J. Spiger and T. A. Tombrello, Phys. Rev. 1 63 (1967) 964 27. M.Ivanovich, P. G. Young and G. G. Ohlsen , Nucl. Phys. A110 (1968) 441 28. G.H.Hale, Proc. of the Conf. on Neutron Gross Sections and Technology, Washington D. C. , 1975, Vol. 1 , p. 302 29. R. A. Hardekopf, N. F. Jarmie, G.G. Ohlsen, R. V. Poore, R. F. Haglund Jr. , R. E. Brown, P. A. Schmelzbach, D.B.Anderson, D. M. Stupin and P. A. Lovoi, to be published LA-6188 (1977) 30. F. W. K. Firk, Proc. of the Int. Conf. on the Inter- actions of Neutrons with Nuclei, Lowell, Massa- chusetts, July 6-9, 1976, p. 389 31. M. Heusinkveld and G. Freier, Phys. Rev. 85_ (1952) 80 32. H. Faissner, Ergebnisse der Exakten Naturwissen- schaften 22 (1959) 180 33. R.J.Holt, F.W.K. Firk, G. T. Hickey and R. Nath Nucl. Phys. A 237 (1975) 111 34. F. W. K. Firk, J.E.Bond, G. T. Hickey, R.J.Holt, R. Nath and H. L. Schuttz , Proc. of the Conf. on Neutron Cross Sections and Technology, Washing- ton D. C. , 1975, Vol. 2, p. 875, NBS Special pu- blication 425 35. J. M. Blatt and L. C. Biedenharn, Rev. Mod. Phys. 24 (1952) 258 36. R.Huby, Proc. Phys. Soc. 67A (1954) 1103 37. A. Simon and T. A. Welton, Phys. Rev. 90 (1953) 1036 38. R. A. Hardekopf, N. Jarmie, G.G. Ohlsen and R. V. Poore, Proc. of the Fourth International Sympo- sium on Polarization Phenomena in Nuclear Reac- tions, Zilrich, Switzerland, 25-29 August, 1975 39. R. A. Hardekopf , G.G. Ohlsen, R. V. Poore and N. Jarmie, Proc. of the Fourth International Symposium on Polarization Phenomena in Nuclear Reactions, Zurich, Switzerland, 25-29 August, 1975 40. R. A. Hardekopf, G.G. Ohlsen, R. V. Poore and N. Jarmie, Phys. Rev. C13 (1975) 2127 41 . M. Karim and J. C. Overley , Proc. of the Conf. on Neutron Cross Sections and Technology, Washing- ton D. C, 1975, Vol. 2, p. 788, NBS Special pu- blication 425 42. H. Wfiffler and G. Walch, Direct Interactions and Nuclear Reaction Mechanisms, Gordon and Breach, New York, 1964, p. 633 ANGULAR ANISOTROPY IN THE 6 Li(n,a) 3 H REACTION BELOW TOO keV J. A. Harvey Oak Ridge National Laboratory Oak Ridge, Tennessee 37830 and I. G. Schroder National Bureau of Standards Washington, D.C. 20234 The data on the differential (n,a) cross section of Li has been reviewed below a neutron energy of 100 keV. Measurements in this region, compared to that above 100 keV, have been few. Only recently has interest increased as large and unsuspected anisotropies have been observed in this energy region. Thus, at 25 keV an anisotropy amounting to 67% in the forward- to-backward direction has been observed; while in the region between 1 eV and 10 keV measurements indicate that in the forward- to-backward 66° cone the asymmetry has an energy d epende nce, in the laboratory system, which can be expressed analytically as 1 + 0.0055 /E n (eV) . These angular anisotropies seem to arise from interference between the large p- wave resonance at 247 keV and many s-wave resonances which account for the large 1/v (n,a) thermal cross section. New and detailed measurements of the differential (n,a) cross sec- tion are needed not only to be able to account analytically for these interference effects but also because it is necessary to consider these anisotropies when the ^Li(n,a)3H cross section is used as a standard. (Angular distribution; fast neutrons; linac; reactor filtered beam; Li(n,a)T; standard) Introduction The 6i_i(n,a)3H reaction has been increasingly used as a standard in partial cross section measure- ments, for flux normalization, for neutron detection and in neutron spectroscopy. The suitability of this reaction as a standard rests in its Q-value (+ 4.784 MeV) and large thermal cross section (940 barns); in the smooth 1/v behaviour of its cross section to within 1% below 10 keV, and that above this energy and up to 100 keV the cross section varies smoothly though no longer following the 1/v behaviour. This is due to a large p-wave resonance (5/2 - ) at 247 keV which dominates the region between 100 and 500 keV. The 6|.i(n,a)3H reaction has been extensively studied both experimentally! and theoretically.! Most of these studies, however, have centered in the region above 100 keV. Below this energy very little data exists. In particular, differential (n,a) mea- surements in the region below 100 keV have been few and mainly unpublished. 2 5 3 This is regrettable as large and unsuspected anisotropies have been recently observed in this energy region. The present review will try to give a summary of these recent experi- mental results and some of their implications. 100 (' 1 1 1 J II 1 1 1 ^ ___T I I I ' 1 on . I 1 r . SO E n =20keV 1 J I- 40 - - 20 1 1 1 1 1 , i 1 1 Fig. 1 Differential cross section at 20 keV from the data of reference 2. Solid curve is theoretical fit of Mahaux and Robaye.4 Early Measurements The first attempt to measure the 6|_i(n,a) 3 H differential cross section below 100 keV was made by G. de Leeuw-Gierts in 1963? using ultra fine grain nuclear emulsions loaded with ^Li at an incident neu- tron energy of 20 keV. The angular distribution ob- tained is shown on Fig. 1 together with a theoretical fit arising from a single-level S-matrix analysis made by Mahaux and Robaye.4 This experiment was followed by that of Beets et al.3 who used a thick tritium target (60 keV) and measured the (n,a) angu- lar distribution in the energy region between 60 and 100 keV. Two different experimental set-ups were used. The first consisted of a circular multiplate nuclear emulsion chamber with a central 6|_i target. The second experiment used the same type of 6|_i loaded fine grain emulsion as in the 20 keV experi- ment. The angular distributions obtained from the two arrangements are shown in Fig. 2 together with a theoretical fit from the S-matrix analysis of Mahaux and Robaye. 4 The discrepancies between the two distributions can be attributed to the slightly differ- ent energy composition of the neutron beams, the non- uniformity of the epicadmium neutrons in the emulsion experiment and the presence of neutrons > 150 keV that could not be properly accounted for in the camera ar- rangement. Furthermore, both the 20 and 80 keV experi- ments using loaded emulsions were subject to rather large thermal backgrounds of ~ 20%. Angular Anisotropy at 25 keV and 2 keV Using Filtered Beams The installation of both a 25 keV iron-filtered 5 and a 2 keV scandium-filtered^ beam facility at the NBS reactor, the fact that unexplained results were being obtained with 6|_i semiconductor spectrometers in the region below 100 keV,7 and the paucity of data on the behaviour of the differential (n,a) cross section in this region prompted the study of the angular aniso- tropy of the 6 Li(n,a) 3 H reaction at 25 keV^ and to later look for similar effects at 2 keV. 10 CHANNEL NUMBER 200 300 cos9 CM Fig. 2. Differential cross section at 80 keV ob- tained by Beets et al.3 The circles repre- sent the values obtained with the &Li loaded nuclear emulsion, while the dots represent the values obtained with the emulsion cham- ber. Solid curve is from fit of Mahaux and Rob aye. 4 To measure the angular anisotropy in the ^Li(n,a)^H reaction at 25 keV a surface-barrier de- tector was coated with 80 yg/cm 2 of ^LiF (front face) and used as a 2tt detector. This detector was placed in two different angular positions with respect to the neutron beam: front face at 90° and back face 90° to the beam. The pulse-height distribution of both ct- particles and tritons were recorded for these two positions (Fig. 3). In a subsequent set of measure- ments a second surface barrier detector was placed coaxially with the first (at 90° to the beam) and in such a way as to subtend a 45° cone. Coincidence measurements that simultaneously recorded the energy distribution in both detectors allowed a measure of the angular asymmetry in the backward-to-forward 45° cone (Fig. 4). Lastly, thermal calibration runs were performed before and after the 25 keV measurements to calibrate the system and to insure that spurious aniso- tropics were not introduced electronically. 8 Subsequently a similar set of experiments were performed with the same detectors and electronic sys- tem at the 2 keV filtered beam facility. The results obtained together with those at 25 keV are summarized in Table I. TABLE I Ratio 25 keV 2 keV R (180°) 1.67 ± 0.11 1.07 ± 0.06 R (45°) 1.96 ± 0.06 1.24 ± 0.02 1500 2000 ENERGY, keV Fig. 3. Experimental pulse-height distribution ob- tained in the forward 2tt measurements made at 25 keV. (Reference 8) Angular Distribution of the ^Li(n,g)^H Reaction at 25 keV The results obtained at 25 keV using the 2tt de- tector were complemented by a set of two other measure- ments: front face at 45° to the neutron beam and back face 45° to the neutron beam. From these four 2tt mea- surements and from the coincidence experiments a coarse-grided angular distribution was obtained. (The details of the method are explained in reference 8.) The results are shown in Fig. 5. This figure shows the angular distribution in both the laboratory and center- of-mass system. The two dashed lines in these distri- butions represent the errors arising from the uncer- tainty in the determination of the background (see Fig. 3). The dotted lines in the laboratory distribution correspond to an isotropic distribution in the center- of-mass system. Angular Anisotropy in the Li(n,a) H Reaction in the 1 eV to 10 keV Energy Region dependence of the me determination ot the energy dependence ot tf angular anisotropy of the ^Li(n,a)3H reaction in the energy region between 1 eV and 10 keV was performed at ORELA using a thin (101 yg/cm 2 ) 6 LiF target. 9 The measurements were made with a diffused-junction sili- con detector which subtended an average angle of 66° 11 ENERGY, keV 2000 2400 300 350 CHANNEL NUMBER Fig. 4. Composite picture showing the pulse-height distribution obtained in the coincidence experiment of Schroder et al.8 performed at 25 keV. to the lithium target. The detector resolution was such that both the triton and alpha groups were well resolved so that an accurate ratio of their intensi- ties could be obtained as a function of energy. In the 1 eV to 10 keV energy region the ORNL results give an energy dependence for the asymmetry in the forward- to- backward 66° cone (in the laboratory system) of the form A = 1 + 0.0055 /E^" where E n is the neutron energy in electron-volts. This energy dependence is shown graphically in Fig. 6. Conclusions The differential (n,ct) cross section of ^Li in the region above 100 keV has been extensively studied, most recently by Overley et alJO No such detailed work exists yet below 100 keV. Thus all one can say is that the existence of the angular anisotropy below 100 keV seems to arise from interference between the large p-wave (5/2") resonance at 247 keV and the many s-wave resonances which account for the large 1/v (n,a) thermal cross section. The theoretical analysis of the ^Li(n,a) cross section which ranges from the single-level, single- channel R-matrix and S-matrix calculations of Mahaux and Robaye4 to the multilevel, multichannel R-matrix computations of Hale^l does not have the input which is available above 100 keV. In order to fill this gap it is necessary to obtain detailed angular I.OO- 1 0.75 1. . . . 0.50 ■ ' - 0.25 1 1 1 45 90 0-lab 135 180 0.75 0.50 0.25 n 1 1 1 45 90 8-c.m. 135 180 Fig. 5. Angular anisotropy of the Li(n,a) H reaction at 25 keV in the lab and center- of-mass systems (reference 8). distributions in the energy range from 1 to 100 keV. Below this energy the differential (n,a) cross section should be of the form a + b cose and therefore the ORNL data as it stands seems adequate. At present only a few experiments are contem- plated. The filtered beam work at NBS using solid state detectors will be completed at 2 keV and a measurement made at 144 keV (silicon filtered beam) J2 Lastly, two groups, one from the University of Michi- gan^ and one from HEDL^4 are planning to perform a series of differential measurements using track-etch detectors. Bibliography 1. See papers on Li in this conference and references therein. 2. G. de Leeuw-Gierts , PhD Thesis, Faculte de Sciences, Universite Libre de Bruxelles (1968). 12 20 40 60 7E n (eV) Fig. 6. Asymmetry in the forward- to-backward 66° cone (in the laboratory system) of the 6|_i(n,a)3H reaction between 1 eV and 10 keV. Data of Harvey et al.9 3. C. Beets, 6. de Leeuw-Gierts , S. de Leeuw, Y. Baudinet-Robinet, G. Robaye, and L. Winand, Nucl. Phys., 69, 145 (1965). 4. C. Mahaux and G. Robaye, Nucl. Phys., 74, 161 (1965). 5. E. D. McGarry and I. G. Schroder, Proceedings of a Conference on Nuclear Cross Sections and Tech- nology, Washington, D.C. March 3-7, 1975. NBS Special Publication 425, Vol. 1, p. 116. 6. I. G. Schroder, R. B. Schwartz, and E. D. McGarry, Proceedings of a Conference on Nuclear Cross Sections and Technology, Washington, D.C. March 3-7, 1975. NBS Special Publication 425, Vol. 1, p. 89. 7. G. de Leeuw-Gierts and S. de Leeuw (private com- munication). 8. I. G. Schroder, E. D. McGarry, G. de Leeuw-Gierts, and S. de Leeuw. Proceedings of a Conference on Nuclear Cross Sections and Technology, Washington, D.C. March 3-7, 1975. NBS Special Publication 425, Vol. 1, p. 240. 9. J. A. Harvey, J. Halperin, N. W. Hill and S. Raman, ORNL/TM-5450, p. 14, May 1976. 10. J. C. Overley, R. M. Sealock, and D. H. Ehlers, Nucl. Phys., A221 , 573 (1974). 11. G. M. Hale, Proceedings of a Conference on Nuclear Cross Sections and Technology, Washington, D.C. March 3-7, 1975. NBS Special Publication 425, Vol. 1, p. 302. 12. R. B. Schwartz, this conference. 13. John C. Engdahl, private communication. 14. Raymond Gold, private communication. 13 EXPERIMENTAL DATA BASE FOR THE Li -7 SYSTEM H. Derrien and L. Edvardson Centre de Compilation de Donnees Neutroniques de l'Agence pour l'Energie Nucleaire (OCDE) B.P. No. 9 91190 GIF SUR YVETTE FRANCE A review of the experimental data available for the Li -7 system is given, including reactions induced by neutrons and those induced by He-4 on H-3 or bv H-3 on He-4. For reactions induced by neutrons only, the energy range up to about 5 MeV has been considered. Some recommendations are given concerning possible future measurements and the validity of the recent evaluations for the Li -f^n, a) cross section. (Elastic scattering; 6 Li(n,a); 6 Li total; 7 Li system; neutrons; review of measurements) . Introduction The most recent international meetings on neutron standards were held in October, 1970 at Argonne (org- anised by Argonne National Laboratory and sponsored by the EANDC), and in Vienna in November, 1972, organised by the IAEA. The problem of the Li-6(n,a) cross section was examined during these meetings. A com- plete review of the Li -6 neutron cross section in the energy range from Thermal to 1,7 MeV was presented by C.A. Uttley at the Argonne meeting (1) for all the data available in 1970, but no such review was presen- ted at Vienna in 1972. Since this date, a great effort has been made to try to explain the large discrepancies existing in the Li-6(n,a) cross section, particularly in the vicinity of the 250 keV resonance and beyond. The investigations have been made in the following directions : 1. The energy calibration . The important differences in the location of the resonance peak do not permit a direct comparison of the cross section from several measurements. The problem of the energy standard has been studied in several laboratories by the time-of- flight method using linacs and cyclotrons (Columbia, Harwell, Geel , Karlsruhe and Saclay). The results of these investigations will be presented by G.D. James at this meeting. 2. The content of Li -6 in the Lithium glasses . The content given by the manufacturers has been question- ed for some glasses used in absolute measurements; precise measurements have been made, for instance, by a non-destructive method : measurement of the neutron transmission of the glasses and determination of the Li -6 content from the 1/v variation of the total cross section at low energy. 3. Further measurements of the Li-6(n,a) cross section" ! The results have not been any more encourag- ing; for instance, a new method used by Friesenhahn in 1974 led to a cross section which is in disagreement with all the previous results (61). 4. Analysis of the experimental data using the nuclear reaction choice has to be mental results; is to try to pred using a formalism the Li-6(n,a) cro results considere experimental resu problem was envis and C.A. Uttley, section at the re obtained from the theories. At the present stage, a made amongst the disparate experi- a way in which to make this choice ict the cross section values by which permits the calculation of ss section from other experimental d as accurate or from a set of Its as diverse as possible. This aged as early as 1969 by K.M. Diment (2) who calculated the (n,a) cross sonance by using the parameters scattering and total cross sections. We know that the cross sections obtained at the peak of the resonance was about 1 barn higher than the values measured by Schwarz (46). At this time, the one-level Breit-Wigner formula used by Diment and Uttley was suspected : for instance, it does not take into account the interference due to a 5/2 level which lies below the neutron threshold. As a consequence, a solution to the problem which could be considered with high confidence, should be obtained by using a complete R-matrix formalism which takes into account the interferences between all the levels of same spin in all the open channels. This kind of analysis has recently been performed by G.M. Hale et al . (3) for the ENDF/B data files and the results are presented at this session by the author. In fact, in the R-matrix formalism, one must consider all the reactions leading to Li -7 compound nucleus and all the possibilities of de-excitation of this system, i.e., all the possible entrance and exit channels. A complete compilation of the Li -7 levels and reaction channels has recently been performed by F. Ajzenberg-Selove (4); the data in Figure 1 have been extracted from this compilation; they show the 5/2" levels at 6.68 and 7.47 MeV excitation energy (the neutron binding energy is 7.2506 MeV). The nuclear reactions energetically available for our purpose are the following : Li -6 total neutron Li-6(n,y)Li-7 Li-6(n,n)Li-6 Li-6(n,a)t a-t scattering t(a,n)Li-6 The reaction Li-6(n,y)Li-7 has very little importance; the corresponding cross section is smaller than 50 mb at thermal energy and is neglig- ible at higher energies and then the neutron absorp- tion cross section is nearly equal to the (n, a ) cross section. The a-t scattering is of great importance because it corresponds to the exit channel of the Li-6(n,a)t reaction. G.M. Hale et al . have particular- ly shown that the calculated cross sections for Li-6(n,a)t are very sensitive to the a-t scattering angular distributions. The same remark applies to the a (t,n)Li-6 reaction which is the inverse reaction of Li-6(n, a )t. In this paper, we will review the experimental data available for the reactions listed above, except for the Li-6(n,y)Li-7 reaction. The neutron data are available in the EXF0R international files on request from the 4 Neutron Data Compilation Centres (NNCSC, Brookhaven; CCDN, Saclay; NDS, Vienna, 14 and CJD, Obninsk). The references given at the end of this review are sorted following the corresponding reactions. For the neutron entrance channel reactions, some details concerning the experiments are also given and the references are coded according to the procedure used in CINDA 76/77. For the other reactions, the references and some mixed references are given in the usual manner. a-t Scattering and t( a ,n)Li-6 Reactions The experimental data available correspond to the references 8-i3. They include differential cross section, angular distribution and polarization measurements. Some important theoretical analyses have been done on a-t scattering and the results have been published in the references 14-17; we think that these works should not be ignored when trying to solve the Li-6(n,a) problem by a nuclear reaction formalism. The differential cross sections and angular dis- tributions for a-t scattering have been measured by M. Ivanovich et al . (9) in the 3-11 MeV a energy range; by R.J. Spiger et al . (10) for a energies between 4 and 18 MeV and by L.S. Chuang (11) for triton bombarding energy in the range 2-3 MeV. Phase shift analyses have been performed by these authors and parameters for the Li -7 levels have been obtained. The polarization of tritons scattered by He-4 have been measured by P.W. Keaton et al . (8) for incident triton energy equal to 6.0; 6.9; 9.5; 10.2 and 10.25 MeV and for several angles; concerning the phase shift analyses, their results are in agreement with those of Spiger et al . More recently, R.A. Harde Kopf et al . (12), at Los Alamos, have obtained the excitation function and angular distribution for the He-4(~t,t)He-4 analysing power in the triton energy range 7-14 MeV for 8 incident energies; they have also measured the angular distribution of the He-4(t,t)He-4 scattering for 6 energies and 17 angles in the energy range 8.23-12.00 MeV; the results have been used by Hale et al . for an R-matrix analysis. Several theoretical analyses were performed on the above-mentioned data by different authors. The resonating method has been used by R.E. Brown et al . (14) to calculate the phase shifts and the results are in good agreement with the experimental results of Spiger and those of Ivanovich. An optical poten- tial analysis was performed by V.G. Neudatchin (16), which happens to be as good as the resonating method when applied to the results of Spiger or Ivanovich. The shell treatment of separable interactions made by L.M. Kuznetsova (15) is also in agreement with the experimental results of Spiger. An R-matrix analysis was also performed by R.C. Baker (17) on the Spiger data. The interesting feature of these theoretical analyses is that no inconsistencies in the experi- mental data have been pointed out. Concerning the H-3( a,n)Li-6 reac results are known. Firstly, the meas Spiger et al. (10) for a incident ene with the dominant effect of the 7.47 ance; the results were normalized to Li(n,a)t results (inverse reaction), results of R.E. Brown et al . (13) not the cross sections have been obtai energies from 11.3 to 12.0 MeV (range 3.9 MeV for the inverse reaction); 2 is expected and the results could be tance for solving the problem of the reaction. tion, only two urement of rgy 11.0-12.4 MeV MeV Li -7 reson- the Schwarz Secondly, the yet published : ned at 15 a of En=0.08 to to 5% accuracy of great impor- Li-6(n,a) Concerning the a-t scattering, we have not compared the 3 most important results, i.e., Spiger et al . , Ivanovitch et al . and Harde Kopf et al . A direct com- parison is not easily possible because the measure- ments were not done at the same angles and energies. The R-matrix analysis should be applied simultaneously to the three sets of data to verify their consistency. The Total Neutron Cross Section The total cross section is generally obtained from transmission measurements using samples of appropriate thicknesses. If the resolution is good enough, which is always the case in the Li -6 measure- ments, the transmission is an absolute measurement of the total cross section : the measurement of the incident and transmitted neutron flux is done with the same detector in the same geometry. If the Li -6 content of the sample is well known, the total cross section is generally obtained with an accuracy better than 2% and can be considered as a basic measurement for testing the coherence of partial cross sections which are more difficult to measure with high accuracy. A precise knowledge of the total cross section is particularly important at low energies in order to determine the 1/v variation of the absorption cross section, i.e., the (n,a) cross section and the ther- mal values can also be obtained by extrapolating the 1/v law. The references 18-26 correspond to the total cross section measurements performed between 1954 and 1976 in the energy range from 70 eV to about 10 MeV. One notes that there is only one measurement at low energy : that of Diment and Uttley (24) for neutron energies above 72 eV. However, J. Harvey (25) announced, at the 1975 Washington conference, a measurement from 10 eV neutron energy; unfortunately, the lower part of the results have not been published and the data present in the EXF0R files begin at 24.6 keV. There is no measurement at very low energy, and the only way in which to obtain the total cross section below 72 eV is to extrapolate the 1/v law obtained by Diment and Uttley, i.e.: a T (barn) = (149.5±0.3)//E~ + (0.70+0.01); however, the extrapolation gives a thermal absorption cross section equal to (940±2) barn, in agreement with direct measurements at thermal energy. The experimental total cross sections are com- pared in Figures 2 and 3 over the whole energy range. The data of Diment and Uttley' are present in the total energy range and are used in Table I as an element of comparison. In Figure 2, the energies have been adjusted to obtain the peak of the reson- ance at 246 keV for all sets of data; for a given set, the corrections are constant in relative values. However, this method of adjustment could be inexact. For instance, in the case of a linac time-of-flight measurement, the part of the error due to a poor evaluation of the so-called t increases relatively when the time-of-flight decreases. One example of this kind of effect is given in Figure 4 which shows the superposition of the results of Diment and Uttley (chosen as reference data) and those of Meadows (23); in Meadows' results the resonance seems to be wider than in that of Diment's results. This is probably due to an under-estimation of the energy correction in the high energy range of Meadow's data (relative to Diment's data). Table I shows that, with the exception of the old values of Johnson (18) and those of Farrell (19), the measured cross sections agree within 3% over the resonance. However, in the low energy part of the 15 resonance, the values of Harvey and Meadows have a tendency to be higher than those of Diment and Uttley. Again, if one accepts the values of Johnson, the agreement is rather good between the different mea- surements in the high energy range (0.5 to 10 MeV); but there is a tendency for the Diment and Uttley values to be lower than the others. A careful evalu- ation of the data should lead to an evaluated total cross section with an accuracy better than 3%; in particular, one should reach an accuracy of 2% at the peak of the resonance. Neutron Elastic Scattering The experimental results available in the experi- mental data files correspond to references 27-35. These data contain angular distributions, polariza- tion measurements and integrated cross sections. The integrated cross sections are plotted in Figure 5 with the Hale et al . R-matrix evaluation. Above about 0.5 MeV no severe discrepancies exist between the different results. But the 250 keV resonance is not well defined and it is difficult to evaluate the en- ergy of the maximum a(n,n) cross section. There are only 9 points between 200 and 500 keV in the recent results of Knitter (35); 14 points between 100 and 500 keV in Lane's (28) data and only 3 points in Villard's data (27); the values are generally given with a large error bar in the resonance. At low energy (from 1 to 100 keV), there is only one mea- surement by Asami and Moxon (32) giving a constant o(n,n) cross section equal to (0. 752±0.043) b up to about 50 keV. This value is about 0.05 b higher than the one obtained in the Hale et al . R-matrix analysis (0.71 b at thermal energy). The thermal value obtain- ed by Uttley et al . (1) by analysing the Diment and Uttley total cross section is equal to 0.724 barn. The Li-6(n,ct)t Reaction The experimental data available, including an- gular distribution and polarization measurements cor- respond to the references 36-69. Everyone knows that the main feature of the data is the important discre- pancies existing in the experimental results for all energy ranges above about 100 keV. Between 1950 and 1960, nine measurements were performed with a discre- pancy of about 25% at the peak of the 250 keV reson- ance; 7 measurements between 1960 and 1970 give approximately the same discrepancy; 13 measurements have been performed since 1970 and the inconsistency is still 25% in this period of time. If we consider all the measurements between 1950 and 1977, the dif- ference between the lower and higher values in the peak of the resonance is about 40% relative to the mean value. Apparently no improvement has been obtained in 27 years, and the evident conclusion drawn by the unbiased observer would be that we do not know how to measure the Li-6(n,a) cross section with high accuracy in the vicinity of the 250 keV resonance and at higher energies. Thus the question to be answered is : Is it possible and reasonable to use the Li-6(n,a) cross section as a standard above 100 keV neutron energy? Perhaps at the end of this session we will be able to answer this question. The thermal absorption cross section has been measured by Meadows (53) using the pulsed neutron method. He obtained a value of (936±4) barn. From the measurement of the total cross section of a Li p S0 4 sample, using the BR2 chopper facility at Geel (Belgium), Becker and Deruytter (51) obtained the value of (944 ±19) barn for the absorption. In this experiment the accuracy was limited by the difficulty in obtaining the accurate value of the Li -6 content of the sample. The value obtained by Uttley and Diment by extrapolating the 1/v law is equal to (939+2) barn. The value proposed by Hale et al . from the R-mat- rix evaluation is equal to 935.89 barn. One can see that there is an agreement within less than 1% in the different values proposed for the (n,a) cross section at thermal . The 1/v law established by Diment and Uttley from the total cross section measurement has been confirmed to be valid by Coates' et al . (59) measure- ment with an approximation of ±1% up to 10 keV. Unfortunately, there are no other measurements at low energy to confirm the Harwell results. However, the Hale et al . evaluation is also in agreement with Coates' results as shown in Figure 6. Figure 7 illustrates the large discrepancies existing at the 250 keV resonance for some selected results. It is not necessary to comment on this Figure . In Figure tf, we have shown the most recent measurements and evaluations : 1) the new values of Fort et al . (69) which corres- pond to the earlier Helsinki values corrected by a factor of 1.117 due to the re-evaluation of the Li -6 content of the glasses (by transmission measurement at the Saclay Linac (5) ) . 2) the values obtained by Lamaze et al . (65) measurement relative to (n,p) scattering; from a 3) the values obtained by Gayther et al . (67) from a measurement relative to U-235(n,f) cross section using the U-235(n,f) standard of the Sowerby evalu- ation; 4) the R-matrix evaluation by Hale et al . ; 5) the value calculated by Knitter (68) from a fit of his measured total and scattering cross sections using a single level Breit-Wigner formula. The Lamaze data have been included in the Hale et al . R-matrix analysis and it is worth mentioning that the discrepancy between the experimental and calculated values is still 3% at the peak of the resonance, whereas the evaluation of Knitter is in agreement with the values of Lamaze. We have also plotted, in Figure 8, the data from Poenitz (60) and Coates which are not too far from the above results at the peak of the resonances. The integral cross section between 0.100 and 0.500 MeV corresponding to some sets of data in Figure 8 are given in the following Table : Area in b-MeV Lamaze 0.481 Gayther 77 0.492 Knitter 77 0.488 Hale 0.486 Poenitz 74 0.471 One can see that : 1) there is only a 0.4% difference between the area of the resonance calculated from the Knitter and Hale et al . values (Knitter's area is 0.4% higher than Hale's, whereas Knitter's values are 2.5% lower at the peak of the resonance); 2) the area calculated from the Poenitz values is only 3% smaller than the one calculated from Hale et al . whereas the peak cross section is about 10% smaller; perhaps there is an effect of resolution in the results of Poenitz; 16 3) the resonance in Fort's results seems to be wider than in the others. It is difficult to know if this is due to an effect of resolution or to the correction on the energy scale. However, the Fort measurement is the only one which is purely absolute and it is comforting to see that his final value agrees, at the peak of the resonance, with the recent measurements and the recent evaluations. Figure 9 shows the dispersion of the results in the energy range 0.5 - 5 MeV. A series of six dif- ferent measurements corresponds to high values of the cross section and one measurement gives very low cross sections (about two times smaller). Between these extremes, in the energy range 0.5-1 MeV, we find in quite good agreement the values of Poenitz, the recent values of Lamaze and Gayther and the Hale et al . evaluation. It is also important to note that Stephany et al. (63) have tried to resolve the discre- pancy in the MeV energy range by an absolute cross section measurement at 964 keV. They found a value of 0.356 b which falls into the high series of values mentioned above. However, in the last ERDA Progress Report (7.b) the authors recognize that this value is 12 or 15% too high and they are now investigating a new method for a more accurate absolute measurement at 964 keV and at other available energies. Conclusion In this review we have not tried to examine or criticise the measurement methods and the possible sources of error in view of eliminating some results which could be doubtful. This was done by C.A. Uttley et al . at the 1970 Argonne meeting for the measure- ments before 1970 and will probably be done at this meeting by Knitter for the most recent measurements. We have only tried to gather the maximum of experi- mental data in the energy range up to about 5 MeV; these data are available to any evaluator who wishes to make a definitive point on the Li-6(n,a) cross section. However, from a simple examination of the data, it is possible to make some recommendations for the future : 1. There is a lack of measurements in some of the energy ranges for some reactions : (a) the total neutron cross section has not been measured at low energy; a recommendation for such a measurement was also made by Uttley at the 1970 ANL meeting; the Harvey measurement has not fulfilled the recommendation because the results in the low energy range have not been published. So, the 1/v law is still established only from the Harwell measurement; (b) according to Deruytter (7) more measurements need to be done at thermal energy; although the few results available agree within better than 1%, these results cannot be trusted with more than 2% accuracy (problems of the Li -6 content in the samples used); c) a better definition of the resonance in the elastic scattering needs to be obtained; that was also a recommendation of Uttley at the 1970 ANL meeting; only one measurement has been done, by Knitter et al., since the ANL meeting. 2. Concerning the (n,a) reaction, a lot of experi- ments have been performed since 1970, and we have al- ready pointed out that the discrepancies are still unacceptable. We should not recommend that more experiments be performed unless the reason for the discrepancies has been clearly established and the new experiments could be undertaken with a high degree of confidence. The attempt of Stephany et al . to resolve the discrepancy in the MeV region is an example of what could be done : they realized that the cross sec- tion value at 964 keV announced at the last Washington Conference is 12 to 15% too high due to the under- estimation of some experimental effect (7.b) and they are investigating a new method which will be used only if the preliminary results are sufficiently encouraging. 3. The problem of energy calibration should also be examined carefully. According to the calculation of Hale et al . and Knitter et al . there is a difference of about 4 keV between the maximum values in the total and the (n,a) cross section for the resonance near 240-250 keV. This difference, which is a function of the resonance parameters and the channel radii, needs to be verified experimentally. On the other hand, energy adjustment of the data for a point per point comparison cannot be done precisely if one ignores how to perform the correction; this means that each author should give the law of variation of the error on energy versus the energy, if possible. 4. Nevertheless, we should not be too pessimistic at the present time. The R-matrix evaluation done by Hale et al . has shed a supplement of light on to the Li-6(n,a) cross section, by using the a-t scattering data; the parameters of this reaction channel should be the same as those of the exit channel of the Li-6(n,a)t reaction. However, Knitter has also shown that the parameters obtained from a single level Breit-Wigner analysis of the total and scattering cross section give, for the Li-6(n,a)t cross section, values which are in quite good agreement with the evaluation of Hale et al . Consequently, it appears that the interferences between the two 5/2" levels in Li -7 are not very strong. It is comforting to note the consistency which now exists between the evalu- ations and the very recent sets of experimental data (Lamaze, Gayther, Knitter), including the re-evaluated data of Fort et al . , and a recommendation should be made to use those sets of evaluated and experimental data as a basis for the definition of the Li-6(n,«) standard cross section. Acknowledgement The authors wish to thank Miss Lynette Truran for her help in correcting the English and the material realisation of this document. Mixed References 1. C.A. Uttley et al . , Neutron Standards and Normaliz- ation, p. 80, ANL, 21-23 October, 1970. 2. K.M. Diment et al . , AERE-PR/NP 16 (1969), p. 3 3. G.M. Hale et al . (a) Nuclear Cross Sections & Technology, Washington, March, 1975, p. 302; (b) LA-6518-MS; (c) private communication. 4. F. Ajzenberg-Selove et al . Nucl . Phys. A 227 (1974), 54. 5. H. Derrien, not published. M. Moxon, private communication. 6. E. Fort et al . , to be published. 7. A.J. Deruytter, private communication. 7.b J.C. Engdahl et al . , BNL-NCS-2151, p. 143 17 References for a-t Entrance Channel 13. R.E. Brown et al . , BNL-NCS-21501 (1976) 8. P.W. Keaton, et al . , Phys. Rev. Let. 20 (1968) 1392 9. M. Ivanovich et al . , Nucl. Phys. A 110 (1968) 441 10. R.J. Spiger and T.A. Tombrello, Phys. Rev. 163 (1970) 964 11. L.S. Chuang, Nucl. Phys. A 174 (1971) 399 12. R.A. Harde Kopf et al . , LA-6188 (1977) 14. R.E. Brown and Y.C. Tang, Phys. Rev. 176 (1968) 1235 15. L.M. Kuznetsova et al . , Sov. Jour. Nucl. Phys. 13 (1971) 394 16. V.G. Neudatchin et al . , Lettere al Nuovo Cimento 5 (1972) 834 17. F.C. Barker, Aust. Jour, of Phys. 25 (1972) 341 References for Total cross section measurements No. Author (*) Reference v ' Year Energy Range Machine No. of Points 18 J0HNS0N+ PR 96 985 54 33-4150 keV VDG 232 19 FARRELL+ 68 WASH 153 68 44-646 keV VDG 125 20 HIBD0N+ 68 WASH 159 68 10-1360 keV VDG 704 21 F0STER+ PR/C 3 576 71 2-15 MeV VDG 237 22 G0ULDING+ GOULDING 72 72 0.7-30 MeV Linac 507 23 MEAD0WS+ NSE 48 221 72 0.1-15 MeV VDG 987 24 UTTLEY+ UTTLEY 74 74 0.07-7000 keV Linac 3260 25 HARVEY+ 75 WASH 244 75 0.025-10 MeV Linac 185 26 KNITTER+ EUR-5726e 77 0.084-3 MeV VDG 233 References for elastic scattering measurements No. Author Reference v ; Year Energy Range No. of Integrated values 27 VILLARD+ PR 101 765 56 210-300 keV Only da/de. 28 LANE+ AP 12 135 61 0.05-2.25 MeV 25 29 BATCHEL0R+ NP 47 385 63 3.35-4.64 MeV 6 30 KNITTER+ EUR-3454e 67 1.0-2.3 MeV 14 31 H0PKINS+ NP/A 107 139 68 4.8-7.5 MeV 32 ASAMI+ 70HELS 153 70 1.0-110 keV 1 33 DEMANINS+ INFN-73 2 73 2.0-4.7 MeV 8 34 H0LT+ NP/A 237 111 75 2.0-5.0 MeV Analyzing power 35 KNITTER+ EUR-5726e 77 0.1-3.0 MeV 40 No. Reference (*) Year Author References for Li-6(n,a)t Measurements Energy Range Comments No. of points 36 ANL-4515 37 PR 90 1049 50 BLAIR+ 142-624 keV 53 DARLINGTON+ 200-600 keV relative to U-235(n,f) 13 only angular distribution 1 18 References for Li-6(n#)t Measurements Cont/d No. Reference » Year Author Energy Range Comments to. of points 38 PR 95 117 39 DOK 111 791 40 PR 103 741 41 PR 112 926 42 PR 114 1580 43 PR 114 201 54 56 56 58 59 59 44 PR 115 1707 59 45 ZN/A 15 200 60 46 NP 63 593 65 47 AF 29 45 48 66 WASH 763 49 JNE 21 271 50 ZFK-130 143 51 70 ANL 125 52 JNE 24 323 53 NSE 40 12 54 70 HELS 1 253 55 71 KNOX 2 611 56 AERE-7075 57 EANDC(E)-148 72 58 NP/A 221 573 74 59 COATES 74 74 WEDDELL+ G0RL0V+ RIBE. KERN+ BAME+ GABBARD+ 1.5 and 2.0 MeV 9.1-730 keV 0.88-6.52 MeV 12.6-17.9 MeV 9.0-342 keV 0.025-4.07 MeV MURRAY+ 1.20-7.93 MeV BORMAN. 2.5 and 14.1 MeV SCHWARZ+ 2.9-588 keV 65 C0NDE+ 100 keV 66 BARRY. 25-100 keV 67 C0X+ 10.7-102 keV 67 RENDIC+ 2.7 and 14.4 MeV 70 BECKER+ 0.0253 eV 70 S0WERBY+ 10 eV - 74 keV 70 MEAD0WS+ 0.0253 eV 70 F0RT+ 82-517 keV 71 PHERS0N+ 10-287 keV 72 CLEMENT+ 0.16-3.9 MeV F0RT+ 0.021-1.7 MeV 0VERLEY+ 0.1-1.8 MeV C0ATES+ 1.04-327 keV Nuc. emulsion, normalized 2 0.42b at 0.6 MeV absolute 24 absolute 10 absolute 23 Relative to U5(n,f) 29 also do/de . Absolute value determined 123 at 0.255 and 0.600 MeV relative to U8(n,f) 11 relative to I-127(n,2n) 2 Relative to H(n,p) for E>150 keV normalized 945 b at thermal absolute 1 Normalized 950 b at 0.025 eV; 3 standard U5(n,f)=577b at 0.025 eV absolute (absorption measur.) 7 Diff. sig. int. relative to 2 H(n,p) absorption = (944±19) barns 1 relative to B-10(n,a) 86 absorption = (936±4) barns 1 absolute 40 Peak normalized at 2.8 b; 31 flux calibration with Pu-Be Normalized between 0.3-0.5 MeV 68 on previous Uttley, Coates, Fort. absolute diff. sig. int. 118 25 Normalized at 149.5//E in the 159 range 1.5-10 keV 60 ZP 268 359 74 POENITZ. 91 keV - 1.50 MeV Normalized to absolute meas. by the same author 67 61 INT 7011 74 FRIESENHAHN+ 1.03 keV - 1.7 MeV Relative to (n,p) scatt.; normalized 148.9//E in 3.5-4.5 331 keV 62 BAP 20 163 75 BARTLE. 2.16-9.66 MeV absolute 24 63 75 WASH 236 75 STEPHANY+ 964 keV absolute 1 64 75 WASH 240 75 SHR0DER+ 25 keV angular anisotropy 65 LAMAZE 76 76 LAMAZE+ 3.34-945 keV relative to (n,p) scattering 88 66 76 LOWELL 1340 76 RAMAN+ 0.5 eV - 25 keV angular anisotropy 19 No. Reference (*) Year Author References for Li6(n,g)t Measurements Cont/d Energy Range Comments No. of points 67 GAYTHER 77 68 EUR-57262 69 FORT (*) 77 GAYTHER+ 3-809 keV 77 KNITTER+ 85-500 keV 77 F0RT+ 82-517 keV Relative to U-235(n,f); U5(n,f) from Sowerby evaluation calculated from a,. anda(n,n) Helsinki values re-evaluated See the codes for references in CINDA 76/77; the name followed by the year corresponds to a private communication. TABLE I COMPARISON BETWEEN THE Li -6 EXPERIMENTAL TOTAL CROSS SECTIONS The values given are the ratio to the DIMENT and UTTLEY values ENERGY RANGE JOHNSON FARREL HIBDON FOSTER MEADOWS GOULDING KNITTER HARVEY (MeV) 0RL.1954 DKE,1968 ANL,1968 BNWJ971 ANL.1972 RPI.1972 GEL, 1974 0RL.1975 0.06-0.08 1.018 1.00 1.022 0.08-0.10 1.047 1.018 1.000 1.059 0.10-0.12 0.968 1.003 1.065 0.974 1.000 0.12-0.50 0.967 0.979 1.003 1.033 1.020 1.007 0.50-1.0 1.042 1.053 0.989 1.0-1.5 1.143 1.037 1.060 1.033 1.5-2.0 1.031 1.059 1.043 2.0-2.5 1.025 1.041 1.016 2.5-3.0 1.073 1.054 1.060 1.033 3.0-3.5 1.005 1.010 1.024 1.006 3.5-4.0 0.979 0.996 1.017 0.986 4.0-4.5 1.002 1.020 1.013 4.5-5.0 0.995 1.010 0.993 20 11.84 Vi + t-d 1-17 348 N *He+ 4 He-p 2.4668 He + 1 Iffiftxww t wwww 11.25 3/i.Vi )9;/7?# »Hm» ,V£iM,. ?,M/ w 7 467 jfl-H. 6 68^ r^w * i< ,?o;.','/? 4 633 h 047761 7 Li J T =3/27 T-l/2 JS- 3<1 12.91 « ^Li + 20 9.61 8 He+d *Li + n 7 2506 T 2791 °B+n-a inV i 0862 ^T 'Be e^ / ' -11202 Jf Be + p- He Figure 1 Energy levels of the Li -7 System (From F. Ajzenberg-Selove (4) ) 21 l»j o i- o u in _l □ I— Q > Z h- m — uj cr a :r I Z^UdD ai O (1) CM (NdbBI but) IS 22 far vitro; I CD O >u > — o a > I— UJ Q U1 UJ UJ (- _>_» Zi- > -HDJl/)Ct 2 t~ o o o a ^t 3 u ~) u. I CD c QL 2 era IT •^Mb, > 3 sfi2r* LU O ^j •-UJ jS^' w*' ~ W " «$£ •*&•*. $« Jf JT ***** V a o ^>v... 'J-s. *•**•••• %.: . ;>* * V. '' * * ' & ' « ' . ' ./_ » . • • » * ' . • '• ,; * •/••''' * •^ *•• \ id '•'• • \ •' . ** • . * j . % n. • • * * -*'. : - * o a -o c ra >- o (NSbai tHJSIS 24 l/> O Q CH -Z. — I Qi LU i— i UJ i>i < h Z I lu s: _i i— «* o _i iii O OQ + —i — i — r i i ^ - i- — i < • X % 4-* V {/ *r 1 1 1 1 1 1 1 ' -~-~i?)_ i "~^— ^ ° 1 1 1 1 1 1 1 1 1 1 1 © - .i.i / i i i i 1 . V^' i « i o -o o c .— < (D 3 o o aj lu c _i ■i- cr d in >- DC 3 CD UJ UJ d K. QL 3 U3 ' O O T U IX Li CX U U & IT. (J U. 2 §5 s_ ai c ai INSHBI ttUIDIS 26 ■i i ""i i r l*J* VNODJi I jl I n Ft I n M I ti n 1 1 1 1 i fci n i . ii oaa-i u a i/i a. cr :r m i o M d d z cr a. a ui u d '*j o a i/i 3»- -J ©Ui » iaao-L ' Qoatrcrzii CLUU -lU^I iNdb^i bJSib 28 ! -1 nit.' Z z in a cr — _iuii UJ UJ - NJUJIId-iUl Ui E^UICDQtO: — o era -icrDcru I L' U U E D U o 5 OJ CD SZ +J (NMb9) UU3IS 29 7 * R-MATRIX ANALYSIS OF THE Li SYSTEM G. M. Hale Los Alamos Scientific Laboratory, University of California Theoretical Division Los Alamos, New Mexico 87545 ABSTRACT We describe a multichannel, multilevel, R-matrix analysis of reactions in the 7 Li system which was used to provide the ENDF/B-V neutron cross sections for 6 Li at low energies. Resonance parameters obtained from the R-matrix levels are presented. Various features of the data are interpreted in terms of these resonances. 7 fi ( Li System; Li(n,t); R-Matrix; resonance parameters; standard) Introduction Discrepant measurements is a problem one almost always faces in the evaluation of neutron cross sec- tions, but in the case of a standard cross section, the problem intensifies the difficulty of obtaining evaluated cross-section values of the desired accuracy. However, one knows from general considerations of nu- clear reaction theory that the standard cross section is linked by unitarity to data for other reactions in the same compound system, and thus one has the pros- pect of reducing the uncertainty of the evaluated standard cross section by analyzing it simultaneously with these other data in a unitary framework. R-matrix theory is a particularly appropriate unitary frame- work for such analyses, as was discussed in Ref. 2. We have used multichannel, multilevel R-matrix analyses of reactions in the Li system to provide evaluated cross sections for the neutron- induced reac- tions on Li at low energies for versions IV and V of the Evaluated Nuclear Data Eile (ENDF/B) . At the time of the version IV analysis, 2 ' 3 the data set dominating the fit was the neutron total cross section measure- ment from Harwell, both because of its quoted preci- sion and number of points. The resulting Li(n,t) cross section (the "standard" in this system) was not much different from that obtained in Version III, which had been largely based on the Harwell measurement. The value of the calculated 6 Li(n,t) cross section at the peak of the 240 keV resonance was 3.5 b, some 16% higher than the value indicated by a contemporary group of direct measurements. However, we noticed, as had the Harwell group, that unitary constraints forced the total cross section unacceptably high above Diment's measurements in the peak when the (n,t) cross section was lowered to better agree with the direct measurements. It is emphasized that this discrepancy between the cross section measurements was indicated by attempts to fit them unitarily. The common practice of evaluating the total and (n,t) cross sections separately, then obtaining the elastic cross section by subtraction, would not necessarily have signaled a problem. ence on Nuclear Cross Sections and Technology two years ago, indicated the effect of the new data was to lower the calculated (n,t) cross section in the peak of the resonance (to 3.4 b) and raise the calculated peak total cross section (to 11.1 b). This trend has continued in subsequent analyses, including that used recently to provide ENDF/B-V neutron cross sections for Li at en- ergies below 2 MeV. The Version V analysis will be described briefly in the following section, indicating the types of data in- cluded, and the fits obtained to representative data sets from each reaction. The third section presents resonance parameters corresponding to R-matrix levels required for the fit, and the concluding section dis- cusses the interpretation of specific features in the data for reactions in the 7 Li system in terms of these resonances. Version V Analysis The 3 arrangement channels considered in this anal- ysis were t + "He, n + 6 Li, and n + 6 Li* (2.18). Vari- ous quantities specifying the channel configuration are given in Table I. TABLE I Channel Configuration for Li Analysis Arrange- ment Channel Radius 4.02 fm max Channel Spins (s) t + "He 5 1/2 n + 6 Li 4.20 fm 1 3/2, 1/2 n + 6 Li* 4.50 fm 1 7/2, 5/2 Data for all possible reactions were considered at tri- ton energies below 14 MeV, and at neutron energies be- low 2 MeV. Although specific references to all the data included will not be given, the different types of data analyzed for each reaction are listed in Table II. We noticed in performing the Version IV analysis that in addition to this strong unitary link to the to- tal cross section in the region of the resonance, the (n,t) cross section was also quite sensitive to the differential cross section for t + a elastic scattering. At that time, however, measurements of this cross sec- tion had not been made with accuracy comparable to that of the Harwell total cross section measurements. Pre- cise measurements of the t + a differential cross sec- tion were later made at Los Alamos, and these were in- corporated in the analysis, along with new measurements of the total cross section from Oak Ridge. Prelimi- nary results of this analysis, reported at the Confer- TABLE II Types of Data Included in Li Analysis Inte- Differ- Total Reaction grated ential Cross Cross Cross Cross Polar- Reaction Section Section Section Section ization 1, He(t,t)"He X X G Li(n,t)"He X X 6 Li(n,n) 6 Li X X X a + t ^ J. 6 T A V X °Li *Work performed under the auspices of the United States Energy Research and Development Administration. 30 A He(t,t) 4 He o(6) 8„. = lso* CM J Ivan (1967), Ja (1976) A _i f v 1 4 \ L / i ■"-— , ■ 1 ■ J Ns , 4 He(t,t) 4 He 0(6) E t = 8.790 MeV | Jarmie (1976) =j ^ 1 1 1 — , i 1 ^- — — L*6 K INET1C ENERCr IN MEV CENTE" 0E MASS ANCLE 4 He(t,t) 4 He A(6) « = *9.6° Hardekopf (1976 e n £ ) 4 He(t,t) 4 He A(9) E,. = 8.785 MeV f\ t t Hardekopf (1976) ,■ m. y s LAB KINETIC ENERGY IN MEV CENTER OF MASS ANCLE 4 4 Fig. 1. Calculated and measured observables for He(t,t) He. The cross section data are from Refs. 9 and 12; the analyzing power data are from Ref . 13. The data were fitted in the usual least-squares sense by the R-matrix analysis code EDA. 11 Renormalizations and energy shifts were allowed for some of the data sets; in these cases, the deviations from the original experi- mental scales contributed to the overall x °f tne fit. The resulting fits to some of the t + a elastic scattering measurements are shown in Fig. 1. Anomalies occurring in the differential cross section and analy- zing power excitation curves (left side of the figure) at triton energies of approximately 3.9, 7.3, 8.8, and 12.4 MeV correspond to the first 4 resonances in Li above the t + a threshold. Parameters for these reso- nances, which have total angular momentum and parity p _ (J ) assignments of 7/2 , 5/2 , 5/2 , and 7/2 , respec- tively, are given in the next section. The third reso- p nance is the J = 5/2 level which shows up prominantly in the neutron cross sections at E = 250 keV. Angular distributions of the cross section and analyzing power are shown on the right side of the figure at triton en- ergies close to this resonance. The low-energy measure- ments of the cross-section excitation are those of Ivan- ovich, while those at overlapping energies and higher are of Jarmie. The calculated curve follows the newer, more precise Jarmie data in the region of the overlap. The differential cross section data shown are also those of Jarmie, 9.13 while the analyzing power excitation and angular distributions were measured by Hardekopf. Figure 2 compares the R-matrix calculation with selected data for the 6 Li(n,t)'*He reaction. On the left are shown recent measurements of the 6 Li(n,t) cross section over the 5/2 resonance at center-of-mass angles of and 180°. These measurements became avail- able after the Version V analysis was completed, so that the curves represent a prediction, rather than a fit. The right side of the figure shows the fits to Overley's measurements 15 of the 6 Li(n,t) angular distributions at E = .1 and 1.0 MeV. The pronounced asymmetry in the n cross section evident at 100 keV persists down to ener- gies as low as 25 keV. These low-energy asymmetry effects in the cross section are well reproduced by the calculations, and can be explained in terms of resonance interference, as discussed in the concluding section. Representative fits to the 6 Li(n,n) 6 Li angular dis- tributions measured by Lane and Knitter are shown in Fig. 3. The Lane angular distributions required substantial renormalizations at energies near the peak of the 240 keV resonance, with the result that our cal- culated integrated elastic cross section lies somewhat above Lane's points over the peak. This agrees with recent results of Knitter for the elastic cross sec- 31 A "1 | | | / \ 6 Li(n,t) 4 He a (9) _ 9 CM = °° | Brown (1976) \ / \ p \ / \ / \ / \ i \^ 1 1 1 1 E = .100 MeV | Overley (1974) - \ — r - 4. LAB KINETIC ENERGY IN MEV CENTER OF MASS ANCLE f\ \ f X C Li(n,t) H He a (9) 6 CM = 180 ° f Brown (1976) \ \ \ \ J \ % ^ LAB KINETIC ENERGY CENTER OE MASS ANCLE 6 4 Fig. 2. Calculated and measured observables for Li(n,t) He. Data for the excitation curves are from Ref. 14. Data for the angular distributions are from Ref . 15. tion over the resonance. Calculated values of the pola- rization (not shown) for elastically scattered neutrons also agree well with measurements by Lane. Figure 4 shows the fits to the total cross sections of primary interest in this discussion, the neutron total and Li(n,t) integrated cross section. Of course, the latter cross section is the standard and is, at that, only recommended for use at energies below the 240 keV resonance. But because of the strong influence of the resonance on the 6 Li(n,t) cross section as it begins to deviate from 1/v behavior (at about 20 keV) , and because of the close unitary connection (already mentioned) , it has with the neutron total cross section, an isolated discussion of the standard cross section would not suffice. We see that the calculated 6 Li(n,t) cross section is generally in very good agreement with recent measure- ments made at the National Bureau of Standards 21 (NBS) up to ~600 keV. In the region of the resonance, the cal- culated cross section has come down by roughly half the difference between the Version IV results and the earlier direct measurements, 5 7 and still exceeds the NBS data by about 2.5% in the peak of the resonance. The data shown are those that were used in the analysis. Correc- tions to the data since that time have raised the exper- imental values in the minimum around 90 keV to agree even better with the calculation, but have reduced some- what the experimental values in the peak. The largest uncertainty of the calculated cross section in the standards region (below 150 keV) occurs in the vicinity of the 90 keV minimum. At lower energies, the thermal value of 935.9 b is in excellent agreement with Meadows' value, and the cross section is quite consistent with Sowerby's ratio measurements below 80 keV, when con- sidered in conjunction with the Version V B(n,a) cross section. The calculated total cross section agrees well with measurements of Diment 1 * and Harvey, except for a syste- matic tendency to overshoot the experimental values in the peak of the resonance and undershoot them in the preceeding minimum. These differences are of the order of 2-6%, and may be within the bounds of realistic un- certainties on the total cross sections in these regions, if not within the stated errors. The energy scale of these total cross section measurements ' essentially determined the position of the 5/2 resonance, with the calculated peak total cross section occurring at 245 keV, and the calculated peak (n,t) cross section at 240 keV, in agreement with most of the measurements using time-of-f light neutron energy determination. 32 1 I I 1 1 1 1 1 6 Li(n,n) 6 Li a(6) E = .118 MeV n f Lane (1960) , i ' i — ■ -f — t » 1 -* =- 1 f= "■r • / \"t • _/Q\ Li(n,n) Li a(6) E n = .530 MeV | Lane (1960) CENTER OF MASS ANGLE CENTER OF MASS ANCLE T 1 ^ iy t\ i/ ' I \ > 6 Li(n,n) 6 Li a(8) E = .240 Mei n J Lane (1960, / 1 1 i | 6 L:.(n,n) 6 Li o(6) E =1.0 MeV n f Knitter (1967) ^~t T "-~^ i- — i — i-^ CENTER OF MASS ANCLE CENTER OF MASS ANCLE Fig. 3. Calculated and measured angular distributions for Li(n,n) Li. The data are from Refs. 17 and 18. Resonance Parameters Since experimental data for nuclear reactions are most directly related to matrices (such as the S-matrix) of amplitude ratios for asymptotic wavefunctions , it is appropriate that resonance parameters for asymptotic measurements be related to the poles and residues of such matrices. The R-matrix is not, of course, an asymptotic quantity, and its poles and residues depend on the boundary conditions imposed at the nuclear sur- face to define the R-matrix, and upon the channel radii which define the nuclear surface. However, one can always transform the R-matrix to an asymptotic form having poles and residues which, although energy- dependent, no longer depend on boundary conditions at the nuclear surface, or upon radial distances outside that surface. where the g and E are known, complex (energy-dependent) functions of the parameters Y-i>E, °f tne original R-ma- A A trix. If, for some complex energy E , E (E ) then E is a pole of R and thus, of the S-matrix as L well. The resonant energy and total width associated with the pole are E R = Re(E Q ) r - -2 Im(E ), We choose the asymptotic form to be the S-matrix so that our specification of resonance parameters cor- responds with the prescription given by Humblet. 25 The transformed R-matrix in that case has the form °l-E T w E -E y while the partial widths are given by r = c W^ P (E ) c V (E o ) "P (E ) c 33 10" 2 3 4 5 6 10 9 10" 2 f 4 5 6 1 8 9 10° < NEUTRON ENERGY . MEV 6 4 Fig. 4. The Li(n,t) He integrated cross section 6 and n + Li total cross section for E be- n tween .01 and 2 MeV. The solid curves are the ENDF/B-V results, and the dashed curves are ENDF/B-IV. Data for the upper curve are from Refs. 5, 21, and 23; data for the lower curves are from Refs. 4 and 10. with Re(P ) P c (E o ) " -77^ where P is the product of the channel radius with the channel wave number, and is the channel outgoing- spherical-wave function, all evaluated at the complex energy E . Resonance parameters obtained in this way from the Version V R-matrix parameters are presented in Table III, along with widths and resonance positions from F. Selove's latest compilation for 7 Li. 26 The first 4 resonances listed are those visible in the excitation curves shown in Fig. 1. The agreement of our calcu- lated parameters for those resonances with values taken from the compilation 26 is satisfactory, although not always within the errors assigned by Selove. The last two resonances lie above the highest ener- gy at which experimental data were included, and their parameters are therefore quite uncertain. They do cor- 2 7 respond roughly with the levels found by Holt et al. in their analysis of n + Li elastic polarizations at neutron energies between 2 and 5 MeV, except that our calculated widths are somewhat narrower. It should be mentioned in this connection that the prescription we use can result in resonance parameters that are quite different from those obtained from the usual expressions 28 relating R-matrix parameters to external widths and resonance positions at real ener- gies. This is particularly true at higher excitation energies, where the complex pole prescription tends to give smaller widths which appear to correspond more closely with the widths of the experimentally observed anomolies. We also note that the complex poles and partial widths defined above are formally radius-inde- pendent in the external region, while those resulting from the usual R-matrix relations are not. Conclusions This analysis indicates that many of the features observed in the measurements for reactions in the Li system can be interpreted in terms of the resonances identified. In addition to the obvious structure in the cross sections and polarizations due to the pres- ence of the first 4 levels above the t + a threshold, one can ascribe the behavior of the low-energy 6 Li(n,t) cross section to broader, more distant states. Near- P , + background levels having J = 1/2 (one of which may be associated with the broad structure at E =16.8 MeV in 7 , + X Li) , and the broad 3/2 resonance tentatively identi- fied in this analysis seem to be mainly responsible for the large 1/v integrated cross sections at low energies. Interference terms, arising from the presence of the prominant 5/2 resonance at 7.46 MeV excitation , + energy and of the 3/2 level higher up, cause most of the asymmetry in the differential cross sections at low . + energies, with interference between the 1/2 and nega- 2 tive parity levels having non-zero P widths contrib- uting to a lesser extent. More recent extensions of 2 9 this analysis to somewhat higher energies indicate that the 3/2 level is responsible for the broad struc- ture seen in both the neutron elastic and total cross sections at E ~ 3.5 MeV, and at the same time, produces n a "shoulder" in the 6 Li(n,t) integrated cross section at E -.2.2 MeV. n Recent measurements for reactions in this system appear to be approaching unitary consistency. At this stage, however, the precise experimental determination of the total cross section below and over the resonance seems to be most elusive, both in terms of the magnitude of the cross section near the peak, and the energy of the peak. New measurements of the total cross section by Knitter 19 and by Smith 30 agree well in magnitude with the Version V calculations, but not in energy scale. Questions about the 6 Li(n,t) cross section persist in relation to ratio measurements. While Gayther's 31 new ratio measurement of 235 U(n,f) relative to 6 Li(n,t) appears quite consistent with the Version V results, Macklin's 32 ratio of 197 Au(n,y) relative to 6 Li(n,t) does not. The resolution of these differences will doubtless require further (but hopefully, small) changes in the neutron cross sections for Li. We feel, however, that the utility of this approach, and the importance of analyzing standard cross sections in a unitarily consistent way with other data from the same system, have been demonstrated by the Version V results. 34 TABLE III Resonance Parameters for the Li System „ b 7/2" 5/2" 5/2" 7/2" 3/2" 4.672 6.596 7.455 9.568 State Partial Width 3 Total Width 3 Total Width 4. 633+. 008 6. 675+. 054 7. 467+. 004 9. 61+. 81 9.853 10.25+.1 .071 .071 .093+. 008 .944 .945 • 875 -:i .001 .023 .077 .089+. 007 .054 .294 .412 .118 .182° 1.171° 1.4+.1 .969 3/2" 11.279 .020 1.360 2.517 1.157 a. Values (in MeV) derived from the Version V R-matrix parameters. b. Values (in MeV) tabulated in Ref . 26. c. Values given for resonances above the range of the analysis, which are therefore quite uncertain. Acknowledgments This analysis is an outgrowth of work begun by D. C. Dodder and K. Witte. The effort still benefits from collaboration with them on many aspects of the problem. I am grateful to N. Jarmie, R. Brown, L. Stewart, and P. Young for their help in collecting and reviewing the experimental data. References 1. E. P. Wigner and L. Eisenbud, Phys . Rev. 72_, 29 (1947), and A. M. Lane and R. G. Thomas, Rev. Mod. Phys. 30, 257 (1958). 2. G. M. Hale, "R-Matrix Analysis of the Light Element Standards," Proceedings of a Conference on Nuclear 12. Cross Sections and Technology, Vol. 1, 302 (1975). 3. G. M. Hale, L. Stewart, and P. G. Young, Los Ala- 13. mos Scientific Laboratory report LA-6518-MS (1976). 4. K. M. Diment and C. A. Uttley, Harwell report AERE- PR/NP 15, p 12 (1969). 5. W. P. Poenitz, Z. Phys. 268, 359 (1974). 6. W. Fort and J. P. Marquette, "Experimental Methods Used at Cadarache to Determine the ^Li(n,a)T Cross Section between 200 keV and 1700 keV," Proceedings of a Panel on Neutron Standard Reference Data, No- vember 20-24 (1972), IAEA, Vienna. Note: Fort expects to renormalize his data upward by ~11%. 7. M. S. Coates, G. J. Hunt, and C. A. Uttley, "Meas- urements of the Relative 6Li(n,a) Cross Sections in the Energy Range 1 keV to 7500 keV," Neutron Stand- ards Reference Data, IAEA, Vienna, p. 105 (1974). 8. R. J. Spiger and T. A. Tombrello, Phys. Rev. 163 , 964 (1970). 9. N. Jarmie et al . , Bull. Am. Phys. Soc. 20, 596 (1975) 10. J. A. Harvey and N. W. Hill, Proc . Conf . on Nuclear Cross Sections and Technology, Vol. I, 244 (1975). 11. D. C. Dodder, K. Witte, and G. M. Hale, "The LASL Energy-Dependent Analysis Code EDA," unpublished. M. Ivanovich, P. G. Young, and G. G. Ohlsen, Nucl. Phys. A110 , 441 (1968). R. A. Hardekopf et al . , Los Alamos Scientific Labor- atory report LA-6188 (1977). 14. R. E. Brown, G. G. Ohlen, R. F. Haglund , Jr., and N. Jarmie, LA-UR-77-407 (1977) (Submitted to Phys. Rev. C). 15. J. C. Overley, R. M. Sealock, and D. H. Ehlers, Nucl. Phys. A221 , 573 (1974). 16. I. G. Schroder, E. D. McGarry, G. de Leeuw-Gierts , and S. de Leeuw, Proc. Conf. on Nuclear Cross Sec- tions and Technology, Vol. I, 240 (1975). 35 IT. R. 0. Lane, Ann. Phys. 12, 135(l96l). 18. H. H. Knitter and A. M. Coppola, EANDC report (E) 57(U) (1967). 19. H. H. Knitter, C. Budtz-Jorgensen, M. Mailly, and R. Vogt, CBNM-VG (1976). 20. R. 0. Lane, A. J. Elwyn, and A. Langsdorf, Jr., Phys. Rev. 136, B 1710 (I96U). 25. J. Humblet and L. Rosenfeld, Nucl. Phys. 26, 529 (1961). 26. F. Selove and T. Lauritsen, Nucl. Phys. A227 , 5k (197*0. 27. R. J. Holt, F. W. K. Flrk, G. T. Hickey, and R. Nath, Nucl. Phys. A237 , HI (1975). 28. See, for instance, Lane and Thomas (Ref. l), p. 295- 21. G. P. Lamaze, 0. A. Wasson, R. A. Schrack, and A. D. Carlson, Proc . International Conference on the Interactions of Neutrons with Nuclei, Vol. 2, 13Ul (1976). 22. J. W. Meadows, Neutron Standards and Flux Normali- zation (AEC23), 129 (1971). 23. M. G. Sowerby, B. H. Patrick, C. A. Uttley, and K. M. Diment, J. Nucl. Energy 2h_, 323 (1970). 2h. G. M. Hale, Nuclear Theory in Neutron Nuclear Data Evaluation, Vol. II (IAEA-190) 1 (1976). 29. G. M. Hale and D. C. Dodder, Proc. International Conf. on the Interactions of Neutrons with Nuclei. Vol. 2, 1U59 (1976). 30. A. B. Smith, P. Guenther, D. Havel, and J. F. Whalen, ANL/NDM-29 (1977). 31. D. B. Gayther, AERE-R 8556 (1977). 32. R. L. Macklin, J. P. Halperin, and R. R. Winters, Phys. Rev. Cll, 1270 (1975). 36 SPECIAL PROBLEMS WITH 6 Li GLASSES G. P. Lamaze National Bureau of Standards Washington, D.C. 20234 Properties of Ce activated Li loaded glass scintillators are discussed as well as their applications as neutron detectors. Three special problems that may arise in their use for neutron detection are non-uniformity of the 6 Li content, multiple scattering, and after pulsing of the photomultiplier . These problems and their consequences are discussed as well as some possible solutions. (glass scintillators; 6 Li(n,a)T; Monte Carlo; photomultipliers) Introduction Before I discuss the problems with the use of °Li glass, let me point out some of its advantages. As we have seen in the earlier talks, the 6 Li(n,a)T reaction has a very large cross section with a value of ^40 b at thermal and a very nice 1/V behavior up to about 10 keV. Figure 1 shows a result of a recent NBS measure- ment. The straight line is the ENDF/B-V evaluation that Gerry Hale talked about previously. One can see that at 10 keV, the deviation from 1/V is less than 1% and at 30 keV it is already about 6% and rising rapidly. Figure 2 compares the NBS results with ENDF/B-V in the region of 2 to 800 keV. Not only does the 6 Li(n,a) reaction have a large cross section (^3. 15b at the peak) but it also has a very high Q-value (^4.79 MeV) . This combination of features has led many investigators to search for a fast scintillator to take advantage of these two attractions. Early investigations 3 centered on 6 LiI(Eu), but the Iodine introduces a great deal of structure in the multiple scattering; neutron detection; response function and 6 I,i.I(Eu) is also reported to under- go changes in its scintillation properties as it ages. In 1958, Ginther and Schulman 1 * reported on Cerium acti- vated glass scintillators. In that same year, Voitovetskii, Tolmacheva, and Arsaev 5 reported the first 6 Li loaded, Ce activated glass scintillators suitable for neutron detection. Much work has been done in the ensuing years and a wide variety of glasses are now commercially available. In fact, to get in the neutron detection business today is pretty simple; all you need is a photomultiplier, a piece of 6 Li glass and a can of beer. Figure 3 shows a schematic of the mounting scheme used at NBS with a "typical" pulse height spectrum. The glass is mounted into an aluminum 7 oz . beer can and viewed edge on by a photomultiplier about 2 cm away. No light pipe or optical coupling compound is used in order to minimize the scattering material in the beam. The simplicity and low cost of this apparatus makes it a very attractive neutron detector. 6.! I PRESENT MEASUREMENT — ENDF-B/V > 5.9 ^ 904b \*J"^ • 902c *"'Voi ^^"^ DISTRIBIT10N 902a • . V^^ FROM FIC. 2. PLT.SE KE1CHT Fig. 6. Analysis of scintillator performance, activation by thermal neutrons from Am/Be moderated source. 6.15 wtl ,Li SAMPLE TEMPERATURE (DEGREES KELVIN) Fig. 7. Variation of 6 Li(n,oc) peak pulse height with temperature for medium lithium glass. 39 luminescent centers. This increased efficiency of energy transfer also improves the resolution as shown in Figure 8. For most applications, the experimenter won't bother controlling the temperature; however, one should be aware of this characteristic and take proper precautions against large temperature variations during an experiment. This also presents the experimenter with the interesting option of cooling the photomultiplier while heating the Li glass to maximize the system performance. Fig. 8. The effect of temperature on pulse height statistics . The next major problem I would like to discuss is the non-uniformity of the Li glasses. This can be a serious problem for many measurements, especially certain types of Van de Graaff measurements where the neutron flux might not be uniform over the face of the scintillator. Figure 9 shows the Li content of a 9 mm Harwell glass as measured by Moxon. The content was obtained by measuring the transmission of the glass as a function of energy and position. By fitting the results to a formula of a T = a + b//E, the 1/V and constant components of the cross section could be determined. The 1/V component (i.e., 6 Li) of the glass seems to be very uniform except at the edges where some depletion of the Li has occurred. The con- stant component on the other hand does seem to have some variation in density across the diameter. Moxon concludes that this glass as well as a 9 mm thick Cadarache glass were very uniform in Li content, except at the edges. He speculates that the Li might be leached out in the cutting and the polishing of individual pieces. Tests performed by the Analytical Chemistry Section at Harwell indicate that Li could easily be leached out if the glass were to come into contact with slightly acidic water solutions. Figure 10 shows the result of transmission measure- ments on a 0.5 mm thick piece of NE 912 using the NBS Reactor thermal column. As you traverse the glass from one edge to the other, a 4% variation in 6 Li content is observed. GLASS SCINTILLATOR /V COMPONENT CONSTANT COMPONENT POSITION AND MAGNITUOE OF CROSS SECTION COMPONENT GIVEN BY FITTED SHAPE FITTED SHAPE ^ Fig distr 10 20 30 10 50 60 70 80 POSITION Imml (NOTE CHANGE OF SCALE AT 20 AND 601 9. A least square fit to one of the spatial ibution measurements for the AERE glass. Figure 11 shows the transmission of the same glass with a traversal made at 90 to the first traversal. In this direction the glass seems to be fairly uniform. The non-uniformity of a particular glass sample raises several experimental problems. First, the 5 Li content cannot be determined by chemical analysis of a part of the glass, or by transmission of part of the glass. In fact, the absolute determination of the °Li content may be impossible to obtain with any precision. Even in relative measurements, the experimenter must be careful to always illuminate the same portion of the glass and take care that the neutron flux is uniform over the used portion. T TRANSMISSION OF U GLASS RELATIVE TO TRANSMISSION AT CENTER 11 fl^ IJ* flfl I -1.0 -0.5 +0.5 + 10 + 1.5 +2 DISTANCE , cm Fig. 10. Transmission of a 0.5 mm piece of NE912 as measured with thermal neutrons. i i i r TRANSMISSION OF Li GLASS RELATIVE TO TRANSMISSION AT CENTER - { T H jH- r +H-i Tr -2 -15 -1.0 -05 0.0 + 5 + 1.0 +1.5 +2 DISTANCE, cm Fig. 11. Transmission of the same glass as Figure 10 but with a traversal perpendicular to the first tra- versal. Let me now turn to the one unavoidable problem found in any experiment with °Li glass: multiple scattering. NE912 which has one of the highest per- centages of 5 Li contains 21% silicon, 55% oxygen, 22.5% 6 Li, 1% 7 Li, and 0.5% cerium. Scattering reso- nances in these elements can cause significant errors in the measure of the neutron flux. Figure 12 shows the multiple scattering correction factor for a 5.08 cm diameter and 0.5 mm thick NE912 scintillator. With the exception of a very narrow Si resonance at 55 keV, the correction is very flat from 2 keV to 250 keV. At that point, the influence of the 6 Li resonance is seen. At 440 keV, the effect of the oxygen resonance is very dramatic and requires a rather large correction for even a very thin glass. As the glass thickness is increased, the corrections become even larger, as seen in Figure 13. The correction is shown for three repre- sentative energies. For an 8 mm thick glass, a "V 15% 40 correction must be made at 250 keV and a ^5% correction must be made at 440 keV. The error bars shown are only the statistical errors for the MonteCarlo calcu- lation. There are obvious systematic errors involved, such as the actual composition of the glass and the accuracies of the scattering cross sections and angular distributions used in the program. Clearly, high pre- cision measurements in the 400 to 500 keV region with Li glass detectors are almost impossible due to the multiple scattering problem. T I I I I I I I I I I I I I | 1 1 — I I I 1 1 I MULTIPLE SCATTERING CORRECTION FACTOR. PER INCIDENT NEUTRON J ' I I 1 I I 10 " 100 NEUTRON ENERGY, keV ' I I I I I Fig. 12. Multiple scattering correction factor for a 5.08 cm diameter by 0.05 cm thick piece of NE912. MULTIPLE SCATTERING CORRECTION vs. 6 Li GLASS THICKNESS 10 keV 250 keV 440 keV .0 1.0 2 4.0 80 THICKNESS , mm Fig. 13. Variation of the multiple scattering correction factor as a function of glass thickness. Another problem in the use of Li glass in Linac T-o-F experiments is the y-flash. The glasses are very sensitive to y-radiation and this can lead to total saturation of the photo multiplier and amplifiers as well as after pulsing of the photomultiplier . The large electron pulse in the photomultiplier ionizes residual gases in the tube. These ions are accelerated back to the photocathode and will excite further bursts of electrons. These signals come from 0.4 uS to 60 yS after the y-flash and can look like valid events. At NBS we solve this problem by offgating the photo- multiplier during the y-flasK* This has the double feature of keeping the photo multiplier from saturating as well as preventing the after pulsing. The after pulse suppression works by pulsing a fine mesh grid that is placed over the front face of the photomultiplier tube. During the y-flash a +300 volt pulse is placed on the grid, suppressing the photoelectrons and preventing them from striking the first dynode. Figure 14 shows the after pulses after a large light source hits the photocathode. The upper scope trace shows two after pulse groups at about 0.6 and 1.0 pS after the light flash. The lower trace shows the effect of off gating; the light flash has been reduced by a factor of about 15 and the after pulses have been completely eliminated. Figure 15 shows the after pulsing at long times. The after pulsing peaks at about 40yS after the flash and gradually dies away so that no after pulsing is observed after about 100 uS. The after pulse suppression system completely eliminates this long term component also. Although this system can solve the problem of the after pulsing of the photomultiplier, the glass scintillator itself seems to after pulse after a strong Y - flash. Additionally the experimenter is faced with the y-decay of the neutron producing targets. These effects make neutron detection at short times after the y-flash very difficult and seems to indicate that two parameter data taking (i.e., time-of-f light and pulse height) may be necessary to separate y-rays from neutron events. 500 mV/div LIGHT FLASH i AFTER PULSES mV/div 200 ns/div Fig. 14. Observed phototube output. Top trace is with no after-pulse suppression; bottom trace is with 300 volt pulse applied to wire mesh. LIGHT FLASH AFTER PULSES " t 50^/j '"*¥! m ■»,•■■ -- ■■>/ is***- ft&K ■futfsr- 1- Ml, 1 V m k._ Fig. 15. Observed phototube output for long times (20 yS/div) . When gating is applied, these pulses are completely suppressed. 41 To summarize, the work of Spowart demonstrates that both improved scintillation properties and in- creased efficiency of e Li glass scintillators is pos- sible. Above 200 keV, uncertainties in the 6 Li (n,a)T cross section and multiple scattering corrections reduce the desirability of 6 Li glass as a flux monitor. Below 200 keV, the cross section is relatively well known, y-flash and after pulsing is not a problem (except at short flight paths) , and the multiple scattering corrections are smooth and relatively small. If nonuniformity of the scintillator is not a problem or if a very uniform scintillator is being used, 6 Li glass makes a very nice flux monitor below 200 keV. Acknowledgements I would like to thank R. B. Schwartz for his assistance in the transmission measurements of the NBS 6 Li glasses and M. C. Moxon for advance copies of his results and figures. References 1. G. P. Lamaze, R. A. Schrack, 0. A. Wasson, To be published. 2. G. Hale, See these proceedings. 3. James Schenck, Nature 171( 1953) 518. 4. R. J. Ginther and J. H. Schulman, I.R.E. Trans. NS-5U958) 92. 5. Voitovetskii, Tolmacheva, and Arsaev, Atomnaya En. 6 (1959) 321. 6. A. R. Spowart, Nucl. Instr. and Meth. 135 (1976) 441. 7. A. R. Spowart, Nucl. Instr. and Meth. 140 (1977) 19. 8. M. C. Moxon, J. D. Downes, and D.A.J. Endacott, UKAEA Report, AERE-R8409 , Harwell, Oxfordshire (1976). 9. G. P. Lamaze, J. K. Whittaker, R. A. Schrack and 0. A. Wasson, Nucl. Instr. and Meth. 123 (1975) 403. 42 A review is given the measurement of gaseous ionization tor is the most ve the reaction produ ray sensitivity th used to detect the of instruments to angular distributi INSTRUMENTS FOR USE OF 6 Li AS A STANDARD* L. W. Weston Oak Ridge National Laboratory Oak Ridge, Tennessee 37830 of the instruments which use the 6 Li(n,a) 3 H reaction neutron flux. These instruments consist of scintill detectors, and external particle detectors. The Li rsatile because of minimized effects of the angular d cts and simplicity. The gaseous ionization detector an does the scintillation detector. Surface-barrier reaction products when a neutron spectrometer is des use this reaction must consider gamma-ray sensitivity on of the reaction, and the range of the triton in th as a standard in ation detectors, glass scintilla- istribution of has lower gamma- diodes may be ired. A choice effects of the e detector. (Instruments: 6 Li(n,a) 3 H, scintillation detectors, gaseous ionization detectors, external particle detectors) Introduction Instruments which use the 6 Li(n,a) 3 H reaction for the measurement of neutron flux must use the advan- tages and minimize the disadvantages of this reaction. The Q value of the reaction is 4.78 MeV, which is high enough to make the reaction relatively easily detect- able for all incoming neutron energies; however, this high Q value makes the reaction difficult for use as a neutron spectrometer except for very high resolution detectors such as silicon surface-barrier detectors. The cross section of the reaction is large with rela- tively smooth variations except near the large reso- nance at about 244 keV. This p-wave resonance makes the reaction difficult to use as a standard cross section in the neutron energy range from about 150 to 350 keV and causes the angular distribution to be anisotropic in the center-of-mass system above a few eV neutron energy. The tritons show forward peaking of about 5% at 100 eV, 18% at 1000 eV, 55% at 10 keV, and anisotropics which do not vary monotonically at higher neutron energies. 1 " 3 Any detector using this reaction for the measurement of neutron flux must be independ- ent of this angular distribution or must be understood well enough that corrections can be made. The energy of the alpha particle from the reaction is 2.05 MeV for zero incoming neutron energy. The range of the alpha particle in Argon at atmospheric pressure is 1.1 cm for zero energy neutrons and 2.0 cm for 1-MeV neutron absorption. These ranges create no problem in most detectors. The energy of the emitted triton is 2.73 MeV for zero incoming neutron energy. The cor- responding ranges of the triton in Argon are 6 cm and 11 cm for zero and 1 MeV neutron absorption and this rather long range creates problems and limitations in most detectors using this reaction. There are three basic methods by which to detect the 6 Li(n,a) reaction: 1. 6 Li embedded in a scintil- lation detector; 2. a gaseous ionization detector; and 3. external charged-particle detectors such as silicon surface-barrier detectors. There is no convenient gaseous form of Li such as in the case of 10 BF 3 . At the present time the scintillation detector is the most popular because of high efficiency, accurate time determination of the event, insensitivity to angular distribution, and low mass in the neutron beam. The advantages and disadvantages of the three methods of detection will be discussed. Scintillation Detectors There are three practica detectors available which may These are the 6 Li loaded glas layer sandwiched between two tillators, and 6 LiI scintilla lator is most popular because able and easily used. The 6 L is not practical for the meas because the iodine has a high with many large resonances detector developed by Dabbs e which is relatively unproven. 1 forms of scintillation use the 6 Li reaction. s scintillator, a Li F very thin plastic scin- tors. The glass scintil- it is commercially avail - il crystal scintillator 4 ' 5 urement of neutron flux capture cross section The plastic sandwich t al . 6 is an innovation The advantages of the Li gla that it has a high efficiency, fa perturb the neutron beam to only principal disadvantage of the Li sensitivity to gamma-rays. Becau tivity, the optimum thickness for is about 0.5 mm if the correspond acceptable. At this thickness th the surface is <10% and the gamma mized. If the glass is made thin of the reaction products is lost the glass because of the rather 1 ton. If an appreciable fraction lost from the surface of the glas obtains a peak in the pulse-heigh well above most gamma-ray induced of the tritons are lost from the becomes sensitive to the angular reaction products. ss scintil lator are st timing, and may a small extent. The glass detector is its se of the gamma sensi- a Li glass detector ing efficiency is e loss of tritons from sensitivity is mini- ner, a large fraction from the surface of ong range of the tri- of the tritons is s, one no longer t spectrum which is pulses. Also if part glass, the efficiency distribution of the Since Li glass can be obtained 7 with a concentra- tion of 6 Li of up to 7.3%, the efficiency of 1/2-mm glass can be as high as 93% for thermal neutrons and 0.46% for 1-keV neutrons. For higher efficiency the glass is available up to 2.54 cm in thickness. Such thick detectors of Li glass are used for measurements such as total cross sections; however, one pays a price in gamma-ray sensitivity. For this reason the volume of glass scintillitor should always be kept at a min- imum for a given application. The gamma-ray sensitivity of Li glass has been discussed in detail elsewhere. 4 ' 8 The maximum value of the peak created by the 6 Li(n,a) 3 H reaction corresponds to an equivalent electron of 1.6 MeV energy created by 43 a gamma ray. The pulse-height resolution (FWHM) of the peak is 20-30%. Gamma rays which deposit less than about 1 MeV energy in the scintillator can be effec- tively discriminated against by means of pulse-height selection. With an electron linear accelerator, the use of Li glass usually requires the equivalent of 2 cm or more of Pb in the beam to reduce the effects of the gamma flash for neutron flight times of less than about 10 usee. In time-of-f light experiments the timing of a detector is important. From Li glass a time resolu- tion of about 2 ns can be obtained. This is a good match to many other detectors such as fission chambers and capture-gamma-ray detectors and thus is an attrac- tive feature of the Li glass detector. The mounting of Li glass scintillators so they can be viewed by photomul tipl iers is not very critical. In most applications it is preferable to mount the PM tubes outside the neutron beam. Such a mounting mini- mizes neutron scattering corrections and gamma-flash problems if a LINAC is used. Tests at Oak Ridge indi- cated that coupling the Li glass optically at the edges to two PM tubes gives equivalent results to mounting the Li glass in a light reflecting box with no direct light coupling. Both methods 9 ' 10 of mount- ing Li glass are used at ORELA in Oak Ridge. The plastic sandwich detector has most of the characteristics of the Li glass detector. This detec- tor, as used by J. W. T. Dabbs, 6 consists of a layer of 96 ygms of Li F per cm 2 which is vacuum evaporated on a 0.01-cm thick NE-110 plastic scintillator. A similar layer of plastic scintillator is placed over the Li F layer. The Li F layer is 2 cm in diameter and the plastic 2.5 cm in diameter. The sandwich is viewed by two photomul tipl iers in a light-reflecting box. The principal advantage of this detector is a faster recovery from a "gamma flash" such as is char- acteristic of a LINAC. The plastic does not exhibit long-term phosphorescence as does Li glass, which can exhibit a pulse of several microseconds in width under such conditions. Another advantage is the smooth cross section of the plastic components. There are no resonances in silicon and oxygen to contend with. Be- cause the reaction products must be allowed to escape from the Li F layer, it is limited to about 200 pg/cm 2 . The maximum efficiency of a single detector is thus limited to about 0.45% for thermal neutrons and 2.3 x 10" 3 % for 1-keV neutrons. A disadvantage of this detector is the light output is lower than Li glass so that electronic noise is more of a problem. Also, the "gamma flash" response of the detector is greater than that of Li glass even though the recovery is faster. This is not a serious problem for beams of the order of 2 cm in diameter, but presents electronic recovery problems for beams of greater dimensions. The hydro- gen in the plastic of the detector scatters about 2.2% of the neutron beam and a correction must be applied for neutron scattering in the hydrogen and carbon and subsequent absorption in the 6 Li. Gaseous Ionization Detectors The general advantage of gaseous ionization de- tectors is low gamma-ray sensitivity. In applications where the gamma-ray intensity accompanying the neutron flux cannot be effectively reduced to tolerable levels this can be very important. The 6 Li reaction has not been used as extensively in gaseous ionization chambers because of the popularity of the use of the 10 B(n,a) reaction. Chambers using BF 3 gas have been used exten- sively in the form of cylindrical proportional counters as well as parallel plate ionization chambers. Since the 6 Li(n,a) reaction is becoming more popular as a standard relative to the 10 B(n,a) reaction, there may be increased use of the Li reaction in gaseous ioniza- tion counters. The 6 Li(n,a) reaction has the advantage of a less complex cross section above about 80 keV and the absence of two groups of alpha particles as occurs in the 10 B reaction. The gaseous ionization detector must detect the alpha and triton from the 6 Li reaction. Since the range of the alpha particle in Argon is from 1 . 1 cm for thermal neutrons to 2.0 cm for 1-MeV neutrons, a detector of about 2 cm thickness will transfer all the energy of the alpha particle to the gas. Unfortunately the triton range is correspondingly about 6 and 11 cm, which is too long for normal thickness chambers because of electron drift time and voltage gradient problems. An example of a gaseous ionization chamber using 6 Li is that of Friesenhahn et al. 11 which was a gridded ion chamber with a spacing of 2 cm between the grid and the Li metal coated plate. The pulse-height spectrum was complex as can be seen in Figure 1, because the tritons which were emitted at 90° with respect to the foil gave a much larger pulse than those emitted at 0°. This effect was caused by the fact that the triton range was long compared to the spacing of the chamber. Examples 1.0 2 0.1 1 1 — Li RESPONSE 200-300 keV • MEASURED CALCULATED — ALPHA TRITIUM sb90° _l_ RT-06361 2 3 4 IONIZATION ENERGY (MeV) Figure 1. Calculated versus measured 6 Li gridded ion chamber response. Chamber constructed by Friesenhan et al . * l 44 of non-gridded parallel plate iomz those used by Macklin 12 and Poenitz gridded chambers the pulse-height s broad because of the dependence on tion of the particle. With these c tant that either the lowest energy counted or that corrections be made counted; otherwise the efficiency o comes sensitive to changes in the a of the 6 Li reaction versus neutron ation chambers are 13 With non- pectrum is very the track orienta- hambers it is impor- triton pulses are for the tritons not f the chamber be- ngular distribution energy. With present day electronics the time resolution of a gaseous parallel plate ionization chamber is limited to a minimum of about 10 ns with about 30 ns being achieved with relative ease. This limitation is imposed by the low energy release of the reaction as compared to fission fragments and the long range of the reaction products. The efficiency of a single gaseous ionization detector is limited by the thickness of the Li layer in the chamber. Since 200 ug/cm 2 of Li F is about the limit, this efficiency is about 2.3 x 10" 3 % for 1-keV neutrons. The effective area of such a chamber may be many square centimeters. Proportional counters with amplification of the ionizing events in the gas could be constructed, but would have no obvious advantages. External Particle Detectors There are tors which hav these systems barrier diodes the 6 Li reacti placed very cl diode 1 - 2 . 11 *- 16 placed outside geometry. 17 " 19 close to the d the other side two forms of external particle detec- e been used with the 6 Li reaction. Both used commercially available surface- to detect the reaction products from on. One system is when the Li layer is ose to the surface of the silicon and the other is when the diode is the neutron beam in low solid angle When the Li layer is placed on or very iode, a second diode may be placed on to form a sandwich detector. The use of surface-barrier detectors has two advantages: very good energy resolution of the reac- tion products and high stopping power. Surface- barrier detectors are capable of timing resolution of about 25 ns. The 6 Li sandwich detector has one char- acteristic which is unique relative to the other dis- cussed detectors. This characteristic is that the energy resolution of the surface-barrier detectors is sufficient that the detector can determine neutron energy as well as the occurrence of an absorption and thus can be used as a neutron spectrometer. The uncertainty in neutron energy determination (FWHM) is about 300 keV. 14 The efficiency of the Li sandwich detectors is comparable to other detectors using a layer of Li F; however, the area is restricted to a few cm 2 in this case because of the size of available surface-barrier detectors. The sandwich detector has a disadvantage which has prohibited its use except in special cases. The surface-barrier detectors must be very close to the Li F layer. Thus there is a relatively large mass of silicon in the neutron beam and high backgrounds are caused by neutron-induced charged-particle reactions in the silicon and gold of the detectors. This pro- blem is severe for neutrons of greater than about 5 MeV energy. Figure 2 illustrates pulse-height spectra obtained at neutron energies of 1.99 MeV and 14 MeV. Proposals 20 have been made to construct such a spectrometer using silicon enriched in 30 Si to con- struct the surface-barrier detectors. This would reduce the charged-particle backgrounds by more than an order of magnitude; however, this proposal has not 500 UNCLASSIFIED ORNL-LR-DWG 48797 1.99-Mev NEUTRONS | COUNTER NO 1 L^ RECOIL PROTON BACKGROUND w 300 f 1 2 < I O 0.28 Mev- c/1 z o 200 o P- SLOW -NEUTRON PEAK — •f I ' \/> *„jr~*\+r»> ^4 U^* " ^ • \J* 10 a 30 40 50 60 70 80 90 100 HO 120 130 140 150 PULSE HEIGHT UNCLASSIFIED ORNL-LR-DWG 51263RI Id 1U 10' 10 n 1 ^Ay \3 ITH Li 6 F LAYER 1 U- W r> v4 f 1 i 1 / / / 14.7 -Mev NE UTRONS y \ n It 1 i 1 j 1 i 1 0.45 Mev- — 4 -* 1 I _L 1 \/ i * I \ 1 20 40 60 80 100 PULSE HEIGHT 120 140 160 Figure 2. Pulse-height spectra for Li sandwich detector 14 using surface-barrier diodes. been carried out because of the expense. Because of the high background from charged-particle reactions, the 6 Li sandwich detector will probably be retained for specialized use. Surface-barrier detectors outside the neutron beam to detect the 6 Li reaction have been used by several experimentalists. 17-19 This method has the disadvantages of very low efficiency and sensitivity 45 to the angular distribution of the reaction. Because of these disadvantages this form of detection is suit- able only when there is a very high intensity of neu- trons and other means of detection are not suitable. Summary The characteristics of the various detectors of the 6 Li(n,a) 3 H reaction have been discussed. Even though it is far from an ideal detector because of its gamma-ray sensitivity and the large resonance in 6 Li at about 244 keV, the 6 Li glass scintillator is widely applicable and a good choice of a detector for the measurement of neutron flux. Ionization chambers may present an attractive choice when the gamma-ray sensitivity of the glass detector is prohibitive. The surface-barrier sandwich may be attractive when a determination of the interacting neutron energy is desirable and the intensity of neutrons above about 5 MeV is low. The plastic sandwich detector requires further test and evaluation. The other forms of detectors are desirable only for specialized cases. References *Research sponsored by the Energy Research and Development Administration under contract with the Union Carbide Corporation. 1. S. Raman, N. W. Hill, J. Halperin and J. A. Harvey, "Angular Anisotropy in the 6 Li(n,a) 3 H Reaction in the eV and keV Energy Region," Proa. Int. Conf. on Interactions of Neutrons with Nuclei, Lowell, Mass., p. 1340, C0NF-760715-72, Tech. Info. Center, ERDA (1976). 2. I. Schroder et al., Proc. Conf. Nuclear Cross Sections and Technology, p. 240, National Bureau of Standards Special Publication 425, Vol. I, U.S. Gov. Printing Office, Washington, D.C. (1975). 3. D. I. Garber et al., "Angular Distributions in Neutron-Induced Reactions," BNL 400, Third Edition, Vol. I, p. 3-6-1, Clearinghouse for Federal Scientific and Technical Information, NBS, U.S. Dept. Commerce, Springfield, Va. (1970). 4. F. D. Brooks, "Neutron Time-of-Fl ight Methods," p. 339, European Atomic Energy Community, Brussels (1961). 5. R. B. Murray, IRE Trans, on Nucl . Sci . , p. 159, Vol. NS-5, No. 3 (1958). 6. J. W. T. Dabbs et al., Phys. Div. Ann. Prog. Rept., Dec. 31, 1975, 0RNL-5137, p. 3 (1975). 7. Nuclear Enterprises Inc., Scintillator Division, Scintillator Catalogue, 935 Terminal Way, San Carlos, Calif. 94070. 8. G. G. Slaughter, F. W. K. Firk, and R. J. Ginther, "Neutron Time-of-Fl ight Methods," p. 425, European Atomic Energy Community, Brussels (1961). 9. R. L. Macklin, N. W. Hill, and B. J. Allen, Nucl. Inst, and Methods 96, 509 (1971). 10. R. B. Perez et al., Nucl. Sci. Eng. 55, 203 (1974). 11. S. J. Friesenhahn et al., "The (n,a) Cross Sections of 6 Li and 10 B Between 1 and 1500 keV," INTEL-RT 7011-001, Intelcom Rad. Tech., P.O. Box 80817, San Diego, Calif. (1974); see also GULF-RT-A12210, Gulf Radiation Tech., San Diego, Calif. (1972); see also Proc. Conf. Nuclear Cross Sections and Technology, p. 232, NBS Special Publication 425, Vol. I, U.S. Gov. Printing Office, Washington, D.C. (1975). 12. R. L. Macklin and N. W. Hill, private communication. 13. W. P. Poenitz, private communication. 14. T. A. Love et al . , Nuclear Electronics I (IAEA Vienna, 1962) p. 415. 15. I. C. Rickard, Nucl. Inst. Methods 105, 397 (1972). 16. P. J. Clements et al., Nucl. Inst. Methods |25, 61 (1975). 17. V. V. Verbinski and M. S. Bokhari , Nucl. Inst. Methods 46, 309 (1967). 18. J. R. Lemley, G. A. Keyworth, and B. C. Diven, Nucl. Sci. Eng. 43, 281 (1971). 19. C. Wagemans and A. J. Deruytter, Annals of Nuclear Energy, Vol. 3, p. 437 (1976). 20. T. A. Love, private communication. 46 EXPERIMENTS AND THEORY FOR DIFFERENTIAL n-p SCATTERING C. A. Uttley Atomic Energy Research Establishment Harwell The present status of the n-p differential scattering cross section is presented over the energy range below 30 MeV. This energy range covers the application of this cross section for the flux or relative flux spectrum measurements which are used to produce differential cross section data for fission and fusion reactor systems. Recent neutron-proton scatter- ing experiments between 20 and 30 MeV have improved the isospin-zero phase shifts, particularly 6( P>)» which largely determine the anisotropy and the asymmetry about tt/2 in neutron-proton scattering below 20 MeV. [Nuclear Reactions np scattering E < 30 MeV; experimental a(6), calculated a(9), phase shift analyses, model predictions]. Introduction The differential neutron-proton scattering cross- section is used as a standard relative to which other elastic cross sections are measured in the MeV region of neutron energies. It is also the primary cross section for neutron flux or flux spectrum measure- ments above about 0.5 MeV. In these applications of the cross section a knowledge of the angular distribu- tion in both hemispheres is required. The detection of proton recoils from hydrogeneous radiators to determine, for example, the neutron spect- rum from a linac target, requires the angular distri- bution at backward angles in the centre-of-mass , and neutron flux measurements using a proton recoil telescope usually utilize the cross section near 180 . The angular distribution predominantly in the forward hemisphere is needed on the other hand for relative scattering cross section measurements, and for the determination of the relative response of organic scintillators to neutron energy by scattering mono- energetic neutrons from hydrogenous samples. Over the last few years most neutron-proton scat- tering experiments below 30 MeV have been made in the energy range 20-30 MeV using Van de Graaff or Tandem accelerators and the J H(d,n) He reaction as the neutron source. Indeed, apart from total cross sections and some differential scattering data at 14.1 MeV few n-p measurements exist below 20 MeV compared with proton-proton scattering experiments. Thus the determination of the neutron-proton differential scat- tering at energies below 30 MeV has relied n n (2) calculation using the phase shifts obtained from continuous energy phase shift analyses of all nucleon-nucleon scattering data up to about 350 MeV. The isospin-one phases obtained from these analyses are accurate and unambiguous due to the accuracy of proton- proton experiments and the variety of observables which have been measured. The relatively inaccurate and sparse neutron-proton scattering data prior to 1970, however, resulted in difficulties in achieving a unique set of isospin-zero phases, especially below 80 MeV, which were theoretically acceptable unless (3) some constraint was imposed on the analysis. A phase shift analysis of nucleon-nucleon scatter- ing data, which includes several new neutron-proton scattering experiments, has recently been carried out in the energy range 20-30 MeV. These new data comprise neutron-proton polarization measurements at 21.1 MeV and 21.6 MeV and neutron-proton angular distribution measurements at 24 and 27.2 MeV as well (9) as cross section measurements at two angles at 24 MeV. Between them the measurements have considerably improved the accuracy of the data set of neutron-proton scatter- ing observables available in this energy range, and the effect has been to provide a unique set of isospin- zero phases for partial waves 1 = 0,1,2 which is independent of model predictions or other constraints. Angular Distribution and Differential Scattering Cross Section Measurements Two different experimental techniques are necessary to measure the angular distribution of neutron-proton scattering over a wide angular range in the centre-of- mass (CM.) system for energies below 50 MeV. Recoil protons from a thin polyethylene radiator exposed to an incident neutron beam can be detected with a counter i ^o , , r o telescope over an angular range to more than 45 before the recoils have insufficient energy to be distinguished from increasing background events. This angular range corresponds to the whole of the backward hemisphere for n-p scattering in the CM. system. In order to extend the angular distribution to more forward angles the neutrons scattered by a hydrogenous target must be detected. This technique enables the angular distribution to be measured from some limiting forward angle to beyond 90 . Thus an angular region of overlap exists between the two methods which enables the data to be normalized. The conversion of the relative cross section onto an absolute scale is usually achieved by fitting a Legendre polynomial expansion to the relative data and normalizing to the total cross section which has been measured to an accuracy of better than 1% below 30 MeV. However phase shift analysts prefer their own freedom to normalise relative data. Detection of recoil protons The most recent angular distribution measurements employ a counter telescope in which one or more thin solid state transmission detectors are placed between the hydrogenous radiator and a detector, usually a plastic scintillator or cesium iodide crystal, sufficiently thick to stop the most energetic recoil protons. The advantages of fast rise times and fast coincidence circuits were exploited by RothenbergC ' 0) to measure the angular distribution of neutron-proton scattering at 24 MeV to an accuracy of better than 2% between 164 and 89 in the centre-of-mass. A schematic diagram of Rothenberg's counter telescope is shown in fig. 1 . The hydrogenous radiator is a polyethylene foil mounted on a thin platinum backing and supported on a target wheel which also has positions for a 47 TARGET WHEEL POLYETHYLENE FOIL PHOTO MULTIPLIER Fig. Cs I CRYSTAL Proton recoil detector for angular distri- bution measurements. platinum blank, and a carbon target which can be centred on the telescope axis for background measure- ments. A modified version employing only one AE detector and with the Csl crystal replaced by an NE102 scintillator as the total energy detector was used by Burrows to measure the angular distributions at 24 and 27.2 MeV over the angular range 158 to 7 1 . Burrows also made use of a particle identification programe to select the recoil protons which allowed his measurements to be extended into the forward hemisphere. The 24 MeV data of Rothenberg and Burrows, converted into the C. of M. were fitted by least squares to a two parameter expression A + A_P„(Cos 6) and normalized to the same total cross section of 397 mb. The values of a(180/o(90) obtained by Rothenberg and by Burrows were 1.146 + 0.017 and 1.135 + 0.014 respectively. Montgomery et al . have recently reported an angular distribution measurement at 25.8 MeV in which backward angle data between 90 and 178 were obtained using a similar AE - E telescope. The half-angle subtended by the telescope at the polyethylene foil is smaller (2.55 ) in this measurement than in the case of Rothenberg (3.3 ) and of Burrows (6.2 ) so that the correction to the angle of the telescope axis to obtain the mean laboratory proton recoil angle is also less. Few attempts have been made to measure the cross section at or near 180 using counter telescopes due to the difficulty of measuring the neutron flux incident on the polyethylene target with an accuracy comparable to that of the angular distribution data. An exception was the measurement of Shirato and Saitoh^'-' at 14.1 MeV in which the flux was determined by the associated particle method. (12) Recently Drosg has reported measurements which give information on the 180 neutron-proton scattering cross section in the energy range 20 - 30 MeV. In these measurements, Drosg compared the zero-degree 3 4 H(d,n) He cross sections at 5 deuteron energies between 6 and 11 MeV (E = 23.1 to 28.5 MeV) obtained by time-of- n flight and an NE213 liquid scintillator with those obtained with a proton recoil counter telescope. The efficiency of the scintillator over this neutron energy range was calibrated relative to the differential cross 3 4 section of the H(d,n) He reaction at 13.36 MeV between neutron emission angles 6(n) of 29 and 90 . The latter cross section was obtained from charged particle measure- 4 ments of the D(t, He)n reaction at 20 MeV. The zero- 34 . degree H(d,n) He cross sections using the proton (2) recoil telescope and the Hopkins and Breit (YALE) calculation of the 180 neutron-proton cross sections were (5.7 + 3.3)% lower than those obtained with the calibrated time-of-f light system. In another experiment using both detectors, Drosg compared the zero-degree yield of 11.23 MeV neutrons 3 3 from the H(p,n) He reaction with that of 25.3 MeV 3 4 neutrons from the H(d,n) He reaction. Thus only the relative efficiencies of the detectors are required, and an independent check of the relative efficiencies of the scintillator was carried out at these two energies 3 4 4 using the H(d,n) He and D(t,n) Ho reactions. A comparison of the telescope yields at 11.23 MeV and 25.3 MeV with those for the scintillator at these energies determines the ratio 0(1 80 , 11.23)/o(180 , 25.3) of the neutron-proton scattering cross sections. The measured ratio of 2.22 + 0.06 is (6.7 + 2.9)% higher than the value of 2.08 from the Hopkins and Breit (YALE) calculation and supports the previous experiment. Detection of Scattered neutrons Three recent experiments have been reported at energies below 30 MeV in which the angular distribution of neutrons scattered from an incident monoenergetic beam by a hydrogenous target has been measured. In two of these measurements, by Master son (9) at 24 MeV and by (13) Cookson et al . at 27.3 MeV, the monoenergetic 3 4 neutrons were produced by the H(d,n) He reaction at zero-degrees, while in the measurement by Montgomery et al . at 25.8 MeV using the Davis cyclotron the incident neutron beam was selected by time-of-f light . /|_ _| 0013mm (100mm | Nl WIN00W Fig. Experimental layout for angular distribution measurements by detecting scattered neutrons. The principle of the experiments is illustrated in Fig. 2 which refers specifically to the measurement of Cookson et al . Neutrons in the incident beam are scattered by protons in the small target scintillator and are detected at a laboratory angle t|i by a larger scintillator placed about lm from the target. A scattered event is defined by a coincidence between pulses from the two detectors occurring within a narrow time interval ~ 100 nsec. A second organic scintillator at a fixed angle to the incident beam is used as a monitor. Each event is characterized by three parameters: the scattered neutron time-of-f light and the pulse amplitudes from the target scintillator and neutron detector. The first of these parameters is required to separate scattered neutrons arising from the primary beam from those of lower energies produced in the source and from Y -ra y s - The pulse amplitude distributions help to identify the type of 48 events producing peaks in the time-of-f light spectra for the different scattering angles. Two potentially important sources of error in the relative cross section measurements by this method are: (a) scattered events due to fast recoil protons produced near the edge of the target and reaching the neutron detector; (b) scattered events due to neutron reactions in carbon in the target scintillator. The effect of fast recoil protons can be eliminated by placing a thin counter in anticoincidence between the target and neutron detectors as in the experiment of Montgomery et al . Alternatively a thin aluminium plate shielding the neutron detector will suffice as indicated in Fig. 2. The reactions in carbon which can simulate a neutron-proton scattered event are 12 12 12 C(n,n')3a and C(n,n'y) C. For all situations when the first reaction gives neutrons which could fall within the time-of-f light limits of the n-p elastically scattered peak, the light output of the alpha particles is lower than the discriminator level on the target scintillator and so the events are not recorded. The second reaction is significant, however, and must be investigated experimentally. T J27.3 („-, p) SCATTERING 17* SCATTERING ANGLE Oid.n) 1_ NEUTRON I - SCATTERING-J s* .•<*/ y Fig. 60 70 80 90 CHANNEL NUMBER 3 Time-of-f light spectrum of scattered events at a laboratory angle of 1 7 . (Cookson et al . ) A time-of-f light spectrum is shown in fig. 3 for the scattering of 27.3 MeV neutrons through a laboratory angle of 1 7 in the experiment of Cookson et al. The small peak to the left (low energy end) of 1 2 the main peak was ascribed to the C(n,n'y) reaction in which the 4.44 MeV y-vay is detected in the target and the inelastic neutron in the neutron detector. This assignment was confirmed in a separate experiment in which the NE102A target scintillator was replaced by an NE213 cell of similar dimensions used with an (n-y) pulse shape discrimination circuit . A peak was observed at the same position in the TOF spectrum, independent of scattering angle, when y-ray events in the target scintillator were selected. A correction for the effect of this reaction is necessary because the inelastic neutrons are not resolved from the main elastic peaks at intermediate scattering angles. Calculated corrections are also necessary for: (a) the loss of events due to scattering collisions close to the walls of the target in which the recoil protons lose too little energy to be detected, (b) multiple scattering in the target in which the net contribution is calculated for the attenuation of neutrons initially scattered at the correct angle and for those scattered into the detector by subsequent collisions in carbon or hydrogen. Scattering events from carbon atoms alone are not detected. The size of these corrections in the experiment of Cookson et al . is shown in Table 1 . Relative efficiency of the neutron detector The wide energy range of the scattered neutrons to be detected requires that a measurement of the efficiency of the detector scintillator be made. In the experiments of Montgomery et al . and of Cookson et al . , the efficiency measurements over the required energy range were carried out using the associated particle method which gives the absolute efficiency for neutron detection. Although the absolute efficiency is not required even for absolute cross section measure- ments, this method has the essential feature of being independent of other cross section data. The associated particle technique adopted by Cookson et al (14) utilized the 3 H(p,n) 3 He, 2 H(d,n) 3 He 3 4 and H(d,n) He reactions in order to measure the efficiency of a 10 cm. diameter by 2.54 cm thick NE102A. detector over the energy range 5-25 MeV which corresponds to the scattering of 27.3 MeV neutrons oo over a laboratory angular range from 17 to 57.9 , and f^^^^- - Tlpji) 3 Hl ' ', ' Dld.nl Ht Tld.nl'H. STANTON MODIFIED 8Y McNAUGHTON »t al UCD NE102A ABSOLUTE EFFICIENCY FOR 116 MeV ELECTFION BIAS i i > i i NEUTRON ENERGY. M. Fig. 4 The absolute efficiency of a 1 cm diameter by 2.54 cm thick NEI02A plastic scintillator for a bias of 1.16 MeV electron energy. allowing for an angular spread of scattering events at each detector angle of + 6.6 . The efficiency of their detector is shown in Fig. 4 for the 1.16 MeV electron energy bias used in their angular distribution measure- ment. The curve which is compared with the measured data is the efficiency predicted by the Monte Carlo code of Stanton in which some of the input data for the carbon reaction cross sections and the electron-proton light output relationship have been modified by McNaughton et al . The agreement between measured and predicted efficiencies observed by the Davis group for their 7.1 cm diameter by 15.2 cm long plastic scintillators, used in the angular distribution measure- ment at 25.8 MeV, is similar to that indicated in Fig. 4. In the latter case, the difference in the angular distribution between using the relative efficiency predicted by the Stanton code and that given by a freehand curve through the experimental efficiency points is small except for the energy region round 11 MeV, where an additional uncertainty was included to the data of Cookson et al. for the two largest scatter- ing angles in Table 1 . 49 Table 1 Corrections to angular distribution data in which scattered neutrons are detected (Cookson et al . ) Lab Scattering angle (degrees) proton bias on target MeV Loss of proton recoils 12 C(n,n' Y ) Net relative multiple scattering C-of-mass Relativistic Correction 17 0.5 1.001 0.950 1.000 0.989 26.6 2.3 1.0) 1 0.974 1 .001 0.992 33.2 4.0 1.023 0.979 1 .002 0.995 39.2 5.4 1 .041 0.985 1 .003 0.998 45.0 6.5 1.058 0.990 1 .005 0.999 50.8 7.8 1.08 0.995 1.006 1 .004 57.9 7.8 1 .10 0.996 1.007 1 .01 1 Differential neutron-proton cross sections To measure the absolute cross section by detecting the scattered neutrons requires the measurement of the neutron flux incident on the target scintillator using the same neutron detector. It is achieved by removing the target scintillator and measuring the incident neutron flux relative to unit charge collected on the neutron source. Thus the scattering cross section at laboratory angle i[i depends only on the ratio of the neutron detector efficiencies at incident energy E and 2 ° at scattered neutron energy E cos \j), and not on the absolute detector efficiency. The absolute 24 MeV neutron-proton scattering cross sections at laboratory angles of 19.5 and 25.1 (9) were measured by Masterson to an accuracy of better than 2% using this procedure. It is applicable only when a detector is employed capable of discriminating between neutrons and gamma-rays since, unless a pulsed accelerator is used, no time-of-f light discrimination is available for the zero-degree flux measurement. In principle the accuracy of the differential cross section will be worse than the angular distribution data due to the additional uncertainty in the absolute correction for multiple scattering and flux attenua- tion in the target. Normalization of relative cross section data It is common practice for angular distribution data to be normalized to the total cross section by fitting a legendre polynomial expansion n £ a. P. (cos 6) to the data after a relativistic l l o conversion to the centre-of-mass system. However, little is to be gained by this procedure if the data are to be included in a phase shift analysis since all measurements are then separately normalized to the most recent total cross section data. A comparison can be made with existing evaluations of n-p differential scattering based on different phase shift analyses by normalizing the relative data to unity at 90 . Such a comparison is shown in fig. 5 of the 27.3 MeV data of Cookson et al . and the 27.2 MeV proton recoil data of Burrows with the predicted values of a (6) /o(90) using the Yale phase shifts and those from the parametrization of Binstock. The latter is based on calculations from phase shifts predicted by the Bryan-Gersten( ' 7) RELATIVE NEUTRON -PROTON SCATTERING AT 27<3 MeV HOPKINS & BREIT I YALE PHASES) [f\--$r~ $ COOKSON etal SCINT. TARGET 3 BURROWS. PROTON RECOIL I I I I 1 8 6 4 2 -2 -4 -6 -8 -1 COS 9 C Fig. 5 Relative scattering cross section at 27.3 MeV compared with predictions from a model calcul- ation (Binstock) and from phase shift analyses. model. The data appear to be in better agreement with the Yale phase shifts due largely to the datum at the most backward angle. DIFFERENTIAL NEUTRON - PROTON SCATTERING CROSS SECTION AT 27 3 MeV cos 9 C Fig. 6 50 Differential scattering cross sections at 27.3 MeV from relative data normalized to total cross section and compared with predictions from phase shift analyses. The angular distribution measurement of Burrows was included in the data set used in the phase shift analysis of Bohannan et al. referred to in the Introduction. The differential scattering cross section resulting from the normalization of Burrows data in this analysis is shown in fig. 6 along with the cross sections predicted by Bohannan at 27.3 MeV (the effect of 0.1 MeV difference in energy < 5% in cross section) . The Livermore (constrained) prediction is also shown in fig. 6. Phase Shift Analyses The continuous energy phase shift analyses of (3) MacGregor, Arndt and Wright (Livermore) and of (4) Seaman et al . (Yale) were made to nucleon-nucleon scattering data available prior to 1970 and form the basis for the evaluation of the H(n,n)H scattering (2) observables up to 30 MeV by Hopkins and Breit. The simultaneous fitting of all p-p and n-p scattering observables over a wide energy range gave the energy dependence of all the significant phase parameters which improved the determination of the lower partial wave phases of importance below 30 MeV. The analysis by the Livermore group without any imposed constraints on the data showed that the isospin-zero phases were relatively poorly determined due to an inadequate set of n-p scattering observables and particularly differential scattering measurements at energies below 100 MeV. They observed a strong 3 3 correlation between the S - D. mixing parameter, . 1 E . , and the singlet p-wave phase shift, o( P,), such that when the former was constrained to be positive at low energies, as suggested by its relationship to the deuteron quadrupole moment , the value of 6( P ) decreased to a value more in accord with model predictions. The Livermore group also demonstrated that the solution obtained by forcing an exact fit to the ratio o(165 )/a(8Q ) at 24 MeV measured by Rothenberg resulted in a positive £1 at low energies and an equally acceptable value of 6( P.). The solution arising from this experimental constraint is shown in Table 2. The Livermore group also reported a single energy phase shift analysis at 25 MeV which included all p-p and n-p data in the energy range 20-30 MeV. The phase parameters obtained from this search are also shown in Table 2. Evidently the accuracy of the angular distribution data of Rothenberg at 24 MeV is not sufficient to outweigh the earlier neutron-proton angular distribution data included in the analysis, since the sign of C. is still negative and the value 1 ' of 6( P ) larger than for the Yale and Livermore constrained set. Recently Bohannan, Burt and Signell have published the results of a phase shift analysis of p-p and n-p data in the energy range 20-30 MeV which include the results of several n-p scattering experi- ments made since the Livermore analysis. The precision of the new data has meant that many of the earlier measurements included in the Livermore analysis at 25 MeV could be removed from the data set because they had no significant effect on the deduced phase parameters. The new high precision data now available and used by Bohannon et al . consists of neutron-proton polariza- tion measurements at 21.1 MeV by Morris et atl.(6) and at 21.6 MeV by Jones and Brooks in addition to the angular distribution measurements of Burrows at 24 MeV (8) and 27.2 MeV sections at 24 MeV of Masterson and the differential scattering cross (9) The analysis includes charge splitting of all the p-p and n-p isospin-cne phase shifts and is not confined to separate p-p and n-p values of S . In fact the charge splitting of S was included as a separate parameter. The low orbital angular momentum phase parameters obtained from this analysis are shown in Table 2 although calculated higher partial wave phase shifts were also included. A comparison of the phase parameters listed in Table 2 shows that: (a) the isospin-cne phases, largely determined by p-p scattering experiments, are in good agreement; (b) discrepancies in the isospin-zero phases are largely confined to 6( P.) and e and arise from the inadequacy of the earlier n-p differential scatter- ing data. In the case of e no neutron-proton scattering observable has been measured which is parti- cularly sensitive to this parameter, and the range of the values of £ noted in Table 2 has a negligible influence on the neutron-proton differential scattering cross section below 30 MeV compared with that for 6( P,). The significance of 6( P ) below 20 MeV (3) It was suggested by Macgregor et al. " that the neutron-proton scattering observable most sensitive to 6( P ) which could readily be measured was the differ- ential scattering cross section. This was confirmed in a sensitivity analysis carried out at 50 MeV by (19) Binstock and Bryan. The importance of this phase parameter to n-p differential scattering at low energies is shown in fig. 7(a). The summed contribution from the three triplet S-P terms to the coefficient of P (cos 9) in the n-p differential scattering cross section decreases more rapidly with energy than the singlet S-P term. The calculation was carried out using the tabulated phase shifts of Seaman et al . down to 10 MeV and the constrained set of Macgregor et al . below 10 MeV for partial waves £ = 0,1,2. It is clear that most of the coefficient below 20 MeV, and therefore the asymmetry in scattering, is determined by 6( P.)- The reason for the suppression of the triplet S-P terms is indicated in Fig. 7(b). In addition to a cancella- tion due to the sign of one ^p phase being opposite, 3 3 3 . the individual S- P terms are reduced by S. passing through tt/2 near 18 MeV. The importance of the recent neutron-proton scattering experiments between 20 and 30 MeV is that, when included in the data set for a phase shift analysis by Bohannon et al . , they produce unambiguous isospin- zero phases with no external constraints. The accurate neutron-proton polarization measurements of Jones and Brooks at 21.1 MeV over the C.of M. angular range 50 to 170 have determined the spin- 3 3 3 3 orbit combination of the D phases (9 D. +5 D--14 D-.) 51 / a r r o,i,2 TOTAL COEFFICIENT FOR t = 0,1.2 •^V p i ENERGY MeV 10 15 20 ENERGY MeV 25 _J 30 Fig. 7 (a) Components of coefficient A of P (cos 6) in differential n-p scattering b elow 30 MeV; (b) energy dependence of individual triplet S-P components of A. . Table 2 Phase shifts in degrees at 25 MeV from analyses of (p,p) and (n,p) scattering data Phase Shift Livermore (Experimental) Livermore (Constrained) Livermore (Single Energy) Yale Bohannon et al . T = 1 's (n,p) o 3 P \ \ \ 52.96 + 0.62 8.21 +0.11 -5.08 + 0.03 2.59 + 0.04 0.72 + 0.01 52.31 + 0.62 8.21 +0.11 -5.08 + 0.03 2.59 + 0.04 0.72 + 0.01 48.84 + 1 .75 8.52 + 0.31 -5.04 + 0.15 2.54 + 0.08 0.74 + 0.03 50.51 + 0.01 8.15 + 0.17 -4.96 + 0.06 2.63 + 0.05 0.915 + 0.017 51.0+ 1 .5 7.95 + 0.68 -5.06 + 0.16 2.64 + 0. 12 0.751 + 0.029 T = \ 79.48 + 0.38 -1 .85 + 0.39 -0.68 + 0.42 -2.60 + 0.09 4.16 + 0.05 0.18 + 0.03 79.18 + 0.37 -4.61 + 0.08 1.29 + 0.32 -2.42 + 0.08 4.08 + 0.05 0.23 + 0.03 84.49 + 2.70 -4.0 +0.69 -0.34 + 0.731 -3.21 + 0. 18 80. 19 + 0. 14 -4.90 + 0.48 1.82 + 0.49 -3.00 + 0.42 3.99 + 0.42 0. 14 + 0.20 81.0 + 1 .6 -5.18 + 0.47 1.03 + 0.58 -2.91 + 0.09 3.89 + 0.10 0.038 + 0.023 from the shape of P(9) with angle. The differential nucleon data near 50 MeV to determine the value of . n,P 3 1 . . scattering data also involve the separate D phases and 6( P.) at this energy and to compare it with the two types of measurements together evidently produce , . , . _. , , .. . Jr . , ,. „,•„,,„ i c r ,i„ s predictions of various meson-theoretical and a solution yielding a unique value of o('P,). , ... , . „, c . , . .. . 3 6 1 phenomenological models. They find from their Model predictions of 6( V .) n- h i a i ( 19) (20) (21 ) . . . Binstock and co-workers have carried out an extensive analysis of experimental nucleon- phase shift analysis that the value of 6( P.) depends strongly on the n-p differential scattering data included in the data set, with less significant changes in the other isospin-zero phase parameters. Bryan and 52 Binstock have compared the values of 6( Pj) from their analyses with those predicted by models in which the coupling constants for it and heavier meson exchange have been obtained by fitting nucleon-nucleon data from to 450 MeV. One such model is that of Bryan and Gersten referred to earlier as the model used by Binstock to parametrize the total, differential scattering and polarization for (n-p) elastic scattering above 25 MeV. The predicted values of 6( P.) at 50 MeV only agree with those deduced from analysis when selected n-p differential scattering data are included. The value of 6( P ) at 25 MeV predicted by the Bryan-Gersten model is -6.25 which is more than two standard deviations smaller than the value of (-5.18 + 0.47) from the analysis of Bohannon in Table 2. This accounts at least in part for the increased asymmetry observed in fig. 5 for the Binstock angular distribution compared with those for the Yale and Livermore (constrained) phase shifts which have values of <5 ( P ) closer to Bohannon's. It 1 1 should be pointed out, however, that the 6( P ) of Bohannon et al. at 25 MeV is in agreement with the value of -5.12 predicted by the phenomenological (22) potential model of Hamada and Johnston. The latter model has been recommended by Lomon and (23) Wilson as giving phase parameters which best represent the n-p scattering observables at low energies. The phase shifts below 5 MeV vary with energy as k , where £ is the orbital angular momentum and k is the neutron wave number. This threshold behaviour was used in the evaluation of Hopkins and Breit. Lomon and Wilson have discussed the deviation from the threshold behaviour which occurs with increasing energy due to the influence of the long range part of the interaction potential arising from one-pion-exchange (OPEP) . They have found from a comparison of several models with experimental data that the Hamada - Johnston potential, which includes OPEP and is adjusted to reproduce the deuteron data (binding energy, triplet scattering length and quadrupole moment) , gives the best agreement at low energies. Discussion The phase shift analysis of Bohannon et al . centered at 25 MeV neutron energy and motivated by recent accurate n-p polarization data can be updated shortly by the inclusion of the angular distribution data of McNaughton et al. at 25.8 MeV and those of Cookson et al. at 27.3 MeV. A comparison with existing evaluations at 25 MeV may indicate the need to recalcul- ate the observables 0(9) .0 and P(6) up to 30 MeV in a convenient format similar to that adopted by Hopkins and Breit. In this event, the energy dependence of the phase shifts prescribed by the Hamada- Johns ton potential could be used and normalized if necessary to those of the phase shift analyses. One problem posed by the measurements of Drosg is the accuracy with which the 180 neutron-proton scattering cross section can be specified, since this limits the accuracy of flux measurements above 20 MeV using proton recoil counter telescopes. The 180 cross sections in mb/sr at 27.3 MeV predicted from available phase shift analyses are: Yale 31.27, Livermore (constrained) 31.16, Bohannon 31.68 and Binstock 32.35, representing a total spread of 3.7%. The value 53 indicated by Drosg's measurements is (5.7 + 3.3)% lower than the Yale and Livermore (constrained) predictions and about two standard deviations lower than the arithmetic mean value of the predicted cross (24) sections. However Drosg has shown that the Yale and Livermore predictions of a(180 ) over the energy range 8 to 16 MeV, when applied to counter telescope measurements of the excitation function of the 2 3 H(d,n) He reaction of other workers, agree very well with his measurement using the calibrated time-of- flight system discussed earlier. Acknowledgements Much of the material presented in this paper was produced by my colleagues Drs. J. A. Cookson (Harwell), J.L. Fowler (ORNL) , M. Hussain (Dacca University) and R.B. Schwartz (NBS Washington) to whom I am indebted. Communications with Dr. P. Signell (Michigan State University) and Dr. M. Drosg (Vienna University) are gratefully acknowledged. References Shirato, S. and Saitoh, K. Japan 36, 331 (1974). Hopkins, J.C. and Breit, G. A9, 137 (1971). Macgregor, M.H. , Arndt, R.A Phys. Rev. 182, 1714 (1969), Seaman, R.E., Friedman, K.A. (1) (2) (3) (4) (5) (6) (7) <8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) Journal of Phys. Soc. Nucl. Data Tables and Wright, R.H. Jreit, G. , Haracz, Rev. 165, R.D., Holt, J.M. and Pratkash, A. Phys. 1579 (1968). Bohannon, G.E., Burt, T. and Signell, P. Phys. Rev. C13, 1816 (1976). Morris, C.L. , O'Malley, T.K. , Hay, J.W. and Thornton, S.T. Phys. Rev. C9, 924 (1974). Jones, D.T.L. and Brooks, F.D. Nucl. Phys. A222, 79 (1974). Burrows, T.W. Phys. Rec. C7, 1306 (1973). Masterson, T.G. Phys. Rev. C6, 690 (1972). Rothenberg, L.N. Phys. Rev. CI, 1226 (1970). Montgomery, T.C., Bonner, B.E., Brady, F.P., Broste, W.B. and McNaughton, M.W. Annual Report, Crocker Nucl. Lab. UCD-CNL786. Drosg, M. Proc. of Int. Conf. on the Int. of Neutrons with Nuclei, P. 1383, Lowell 1976. Cookson, J. A., Hussain, M. , Fowler, J.L., Schwartz, R.B. and Uttley, C.A. INDC(UK)-28U also Fowler, J.L. et al. Bulletin of the Amer. Phys. Soc. 22 , 53 (1977). Cookson, J. A., Hussain, M. , Uttley, C.A. , Fowler, J.L. and Schwartz, R.B. Proc. of Conf. on Nuclear Cross Sections and Technology, Washington, 1975, ed. R.A. Schrack and CD. Bowman. McNaughton, M.W., et al. Nucl. Instr. and Meth. 116, 25 (1974) and ref . 1 1 . Binstock, J. Phys. Rev. C 10, 19 (1974). Bryan, R. and Gersten, A. Phys. Rev. D 6, 341 (1972). Signell, P. Advances in Nuclear Physics, ed. H. Baramger and E. Vogt (Plenum) 1969 vol. 2. Binstock, J. and Bryan, R. Phys. Rev. D9 2528 (1974) . Arndt, R.A., Binstock, J. and Bryan, R. Phys. Rev. D8, 1397 (1973). Bryan, R. and Binstock, J. Phys. Rev. D10, 72 (1974). Hamada, T. and Johnston, I.D. Nucl. Phys. 34, 382 (1962). Lomon, E. and Wilson, R. Phys. Rev. C, 1329 (1974), Drosg, M. Private communication. USE OF THE n,p SCATTERING REACTION FOR NEUTRON FLUX MEASUREMENTS J. B. Czirr Lawrence Livermore Laboratory University of California Livermore, California 94550 Several contemporary proton-recoil detectors are described and compared. These de- tectors have been used for neutron-spectrum measurements over various portions of the 10-keV-to-20-MeV energy range. Several factors which limit the accuracy of the results are compared quantitatively. General suggestions are given for setting and using standard cross sections and for future developments using the n,p scattering reaction. (Detectors; flux; neutrons; proton-recoil; scattering; standards) Introduction For a neutron-induced reaction to be considered as a first-class standard, we require that the cross section be known to better than ±1%. Even with this requirement met, experimental difficulties may pre- clude the use of the reaction in certain energy re- gions of interest. Such considerations have led to the use of the ^Li(n,a) and 10|3(n,a) or 10B(n,ay) re- actions as standards for neutron energies below 10 or 20 keV. Above these energies, high-precision flux- measurement techniques (involving a neutron cross section) make use of then,p scattering reaction. Other techniques of comparable accuracy involve total- ly absorbing detectors or associated-particle systems. Primary-Cross-Section Genesis The present paper will desc neutron flux measurements in the ergy range which use the n,p sea Figure 1 gives an example of the to obtain the fission cross sect oaper describes methods of obtai Figure 1. Implicit in this figu that the n,p scattering reaction everyday use by those interested monitors. Contemporary Proton ribe techniques for 10-keV-to-20-MeV en- ttering reaction, use of this reaction ion of 239 Pu. This ning the rate R] of re is the suggestion is not suitable for in off-the-shelf flux Detectors A classification scheme is given in Table 1 for five contemporary proton-recoil detectors and Figure 2 shows the geometrical arrangement for each system.! -5 Primary-Cross-Section Genesis Primary-Cross-Section Utilization VMeOJurement \ < r S . Secondary-Cross-Section Genesis Secondary- Cross-Section Utilization Tertiary-Cross-Section Genesis Tertiary-Cross-Section Utilization U Fission Chamber ' VMeasurement/ Pu Fission Chamber 239 r Fig. 1. Schematic outline of the use of standard cross sections to obtain a f ( Pu). * Work performed under the auspices of the U.S. Energy Research and Development Administration. W-7405-Eng-48 54 FLUX DETECTORS COLUMATEDV ^ COLLIMATOR v\^ p> I NEUTRON BEAM \ \ \ \ \ f" . I L agfta RADIATOR FOILS KFK Proton-recoil telescope 0.25 mm SS cap 0.25mm SS cap . LASL Proton-recoil detector Si(L1) semiconductor detector shield Collimated neutron beam 6.35-cm diam -12.7-cm diam — u LLL Proton-recoil detector Protons 0.88 mm SS \ NBS Hydrogen-filled PROPORTIONAL COUNTER A 5.1 cm _J_ -60 cm INCIDENT NEUTRON SILICON SURFACE BARRIER CHARGEO PARTICLE DETECTOR w I60cm TO PREAMPLIFIER ^=? ORNL Proton-recoil detector POLYETHYLEN ALPHA CALIBRAT Fig. 2. Geometrical arrangements for five contem- porary proton-recoil -detector systems. 55 It is interesting that none of these modern detectors rely on ,a precise neutron-energy-measurement capa- bility. This feature permits an important increase in proton-detection efficiency. It is also to be noted that none of the five systems match the ideal detector listed in the last column of Table I. A quantitative comparison of several detectors is shown in Table II. Each of these detectors was chosen to perform with a particular accelerator and they are not necessarily to be compared with each other for use with a single neutron source. For this reason, the information must be used judiciously as a starting point for future developments or for evaluating past results. Fortunately, most of these systems were used to obtain the fission cross section of 235u, a circum- stance which allows us to make valid comparisons. For this purpose, we show in Figure 3 the total error quoted for several 235u fission-crossrsection measure- ments as a function of neutron energy J -4 Because the error in measuring the fission rate is usually smaller than that of the flux measurement, these results serve as a rough figure of merit for the several proton de- tectors. Limitations on Accuracy I will attempt to evaluate those factors which limit the accuracy of the various detector systems. Eight such factors have been identified and are listed in Table III. 1 . Cross Section Errors The n,p total -cross-section error is less than 1% below 30 MeV and will be ignored. Only the effect of uncertainties in the angular distributions will be compared in Table III. The listed percentages are the published uncertainty in oh (9) at the appropriate angle. 6 The numbers in parentheses refer to the neu- tron energy at which the uncertainty applies. 2. Background The typical percentage background over the useful energy range is listed. If it can be assumed that the fractional error in the background is the same for each system (±10% of the background, for example), then a relative uncertainty can be estimated from Table III. 3. Neutron-Energy Resolution Because of the widely different energy ranges and accelerators involved, only a crude characterization of energy resolution will be given. The values of AE/E are listed near the midrange of the appropriate energy region for each system. In most cases, the cross- section accuracy illustrated in Figure 3 was achieved by summing over several high-resolution time-of-fl ight channels. For these cases, the practical resolution is taken to be the fractional separation between adjacent published energies. For the NBS proportional counter, a timing uncertainty of 0.6 ysec was used and for the ORNL system, a beam width of 50 nsec was the limiting factor. 4. Counting Rate I will assume that each investigator was free to place his proton detector at the optimum flight-path distance. The appropriate counting rates can then be computed directly from Table II. The resulting rates are listed in Table III, where only the thick-foil rates are included. The high rate for the LASL de- tector is a result of the short flight path (10 cm.) employed with a monoenergetic neutron source. A KFK LLL j i i i 1 1 1 i i ■ ■ ■ ■ i ■ i i__i i i i i i 1 1 1 10 10 10 10 Neutron Energy (eV) 10 Fig. 3. Figure o f merit for four contemporary systems vs. neutron energy, Heutron energy is measured bv time of flight or is inferred from the source reaction. 56 further efficiency factor of approximately 10-2 must be applied to this system because of the necessity of ob- taining each data point separately. The rate in par- entheses reflects this added factor. 5. Pulse-Height Extrapolation An error is portion of the pu height for two of threshold for the 1.1 keV and this less than 0.5% fo This estimate is threshold energy per ion pair (W) 10% in W have bee by Bennett and Yu the necessity of below 20 keV. incurred in extrapolating the measured lse-height distribution to zero pulse the detector types. The electronic NBS proportional counter is set at results in an extrapolation error of neutron energies above 10 keV.4 based on a 5% uncertainty in the and a constant value for energy loss above 1.1 keV. However, variations of n observed in the l-to-10-keV region le. 7 These latter results would imply a small correction to the NBS data The low-energy KFK telescope relies on calculated efficiencies to obtain the effect of a 300 keV elec- tronic threshold. The calculated efficiency drops rapidly below 2.5 MeV and limits the accuracy below this energy. 1 A ±5% uncertainty in the calculated ef- ficiency seems reasonable for energies below 2.5 MeV. Above this energy the efficiency is more slowly vary- ing and less affected by bias uncertainties. 6. Spurious Reactions A portion of the observed background rates is due to neutron-induced reactions in the proton detec- tors. Although an absolute calculation of these rates would be difficult, it is instructive to compare relative rates. For this purpose, we will list the ratio of the neutron flux to the proton flux impinging on the various proton detectors. All ratios were cal- culated at an incident neutron energy of 1 MeV. It is evident that there is a poor correlation between fn/fp and the observed background rates. 7. and 8. Unwanted Neutron Scattering The fraction of neutrons which strike the neutron "radiator" after a prior collision is listed as "Inscatter" in Table III. These estimates are based upon calculations and indicate a negligible error from this source. An estimate of the uncertainties caused by scat- tering materials in the neutron beam may be obtained from the row labeled "Outscatter." The listed quanti- ties are the total atoms/cm 2 of material in front of the proton radiator. In addition to the eight sources of error dis- cussed above, there are several effects which are not routinely reported on a quantitative basis. These in- clude 1) effective-volume changes for gas-filled counters and 2) timing errors caused by variations in pulse height at the input of fixed-level discrimi- nators. In the absence of published data concerning the magnitude of these effects, a quantitative error estimate is impossible. General Rules for Setting Standards I would like to summarize five personal observa- tions concerning the demanding business of setting standards for world-wide use. 1. Do not push your luck. (Avoid strong state- ments about accuracy near the ends of your useful range. ) 2. Never allow your systematic errors to exceed your random uncertainty. (Statistical precision is often achieved at high cost by incurring large correc- tions in the process. ) 3. Do not feel obligated to build "the universal detector" when measuring standards. (Tailor the de- tector to your source since it need not be used else- where. ) 4. It is easy to measure the signal --the hard part is measuring the background. (Think ahead as to what can be changed to obtain the background rate.) 5. Avoid extrapolations into unmeasured regions. (Keep the signal away from zero pulse height.) These statements may not be ratified by everyone involved in standards work, but I feel that they would go a long way toward preventing "down stream pollu- tion." Suggestions for Using Standards What can the measurer of a tertiary cross section (in the spirit of Fig. 1) do to take full advantage of contemporary standard cross sections? Include a measurement of a standard cross section with a reaction similar to the desired quantity. For example, when measuring the fission cross section of 239p u relative to 10B(n,a) below 10 keV, include a measurement of o"f(235(j), to provide a "parallel" standard in addition to using the l^B "vertical" standard as a spectrum monitor. This added informa- tion can be very useful even at energies where rapid fluctuations in the standard cross section ( 235(j) preclude its use as a spectrum monitor. Future Developments I will now propose two new detector systems, one a direct descendant of the LLL system described abover and the other a more general detector type not cur- rently in use in standards work. One possible disadvantage of the current LLL system is the presence of a massive lead shield in the central portion of the neutron beam. Any detector- such as a fission chamber—which is placed a short distance in front of this shield would be subject to sizable backscattering corrections. An alternate geometry with an order of magnitude greater efficiency is illustrated in Figure 4. A thin proton radiator is placed a few cm in front of a 19-cm-diameter photo- multiplier (PM) tube, with the entire assembly con- tained in a vacuum chamber. Protons are detected in a 1-mm-thick 'Li-glass scintillator which covers only an outer ring on the PM tube face. A massive collimator prevents exposure of the scintillator to the incident beam but exposes the full radiator area. The spacing may be adjusted so that the minimum proton energy is approximately 50% of the incident neutron energy and the fraction of recoil protons detected is 14%. If (n,a) or (n,p) reactions in the glass scintillator present background problems, the newly-developed BaF2(Ce) scinti llators^ may be used instead, with an order of magnitude reduction in the troublesome cross sections. This sytem is listed as number 6 in Table II and should be useful throughout the l-to-20-MeV range. Turning now to entirely new systems, we notice that the very convenient organic scintillators have never been used--in a manner which takes advantage of the accurate n,p cross section—for standards-quality 57 spectrum measurements. This unfortunate circumstance is probably based on the striking non-linearity in the pulse-height vs proton energy^ and in large surface losses for thin scintillators. I would like to sug- gest that the second effect could be eliminated by backing a thin^.S mm)plastic scintillator with enough ^Li-glass scintillator to stop maximum-energy protons (^1.5 mm). The resulting pulse-height spec- trum for mono-energetic neutrons would be similar to that observed in proportional counters, for the top 90% of the spectrum. Monte Carlo calculations indi- cate that the light output from recoil carbon nuclei and alpha particles [from 12c(n,a) reactions] are confined to less than 10% of that from protons, at a fixed incident-neutron energy. 10 It seems reasonable that with appropriate effort, the pulse-height spec- trum could be extrapolated to zero with an accuracy approaching 1-2% of the total area. Scattering from the photomul tipl ier tube can be eliminated by viewing the scintillators edge on. Table II, number 7, lists the characteristics of this detector. These proposals do not describe finished products, but may suggest fruitful avenues for investigation. Undoubtedly, other schemes will arise to permit con- tinued future use of the all-important n,p reaction. Collimator Proton Radiator Li Glass 9 cm diameter Photomultiplier Tube Fiq. 4. Geometrical arrangement for proposed proton- recoil detector. Appendix A brief historical survey will be given for proton-recoil detectors which predate the systems de- scribed in Table I. Four basic types have been em- ployed and the geneology of each will be traced back a generation or two. 1. Gas-filled proportional counters The NBS counter is similar to the detector de- scribed by Bennett and Yule in Ref. 7, a system in use for many years at ANL. The NBS system achieved im- proved timing resolution by restricting the incident neutron beam to the central one-fourth of the detector volume. 2 . Mul ti pi e-component proton tel escopes The high-energy KFK system followed historically the early proportional -counter telescopes described in Ref. 11. Improved timing resolution was achieved by using gas scintillators. 3. Single-component in-beam detectors The LASL detector is similar to the system de- scribed by Johnson, '' with solid-state detectors re- placing the inorganic scintillators. This technique was also suggested by E. Pfletschinger and employed by Kappeler and Frbhner at KFK J 2 4. Single-component out-of-beam detectors The improved shielding arrangement of the ORNL system was first employed by Jaszczak and Macklin.5 The LLL detector took advantage of this concept and in- cluded an improved proton-detection efficiency. 1 References 1. I. Schouky, S. Cierjacks, P. Brotz, D. Grb'schel , B. Leugers, Proceedings of the Conference on Nuclear Cross Sections and Technology, Washington, D.C., March 3-7, 1975, p. 277. 2. 3. 6. 10. 11, 12. D. M. Barton, B. C. Diven, G. E. Hansen, G. A. Jarvis, P. G. Koontz, R. K. Smith, Nucl . Sci. Engin. 60, 369 (1976). G. S. Sidhu and J. B. Czirr, Nucl. Instr. and Methods, 120, 251 (1974). 0. A. Wasson, Proceedings of the NEANDC/NEACRP Specialists Meeting on Fast Neutron Fission Cross Sections of U-233, U-235, U-238 and Pu-239, Argonne National Laboratory, June 28-30, 1976, p. 183. L. Macklin, Rev. Sci. Instr., R. J. Jaszczak and R. 42, 240 (1971). J. C. Hopkins and G. Breit, Nuclear Data Tables, A9, 137 (1971). E. F. Bennett and T. J. Yule, Proceedings of the Conference on Neutron Standards and Flux Normali- zation, Argonne National Laboratory, October 21-23, 1970, p. 385. J. B. Czirr and E. Catalano, "Cerium-activated BaF2 and CaF2 Scintillators," to be published in Nucl. Instr. and Methods. J. B. Czirr, D. R. Nygren and C. D. Zafirators, Nucl. Instr. and Methods, 31_, 226 (1964). R. Batchelor, W. B. Gilboy, J. B. Parker and J. H. Towle, Nucl. Instr. and Methods, 1_3, 70 (1961). C. H. Johnson, Chapter II. C in Fast-Neutron Physics , Part I, edited by J. B. Marion and J. L. Fowler (Interscience Publishers, Inc. N.Y., 1960), p. 247. F. Kappeler and F. H. Frohner, Proceedings of the Conference on Nuclear Data for Reactors, Helsinki, June 15-19, 1970, p. 221. 58 Table I Proton-Detector Classification 1 KFK 2 LASL 3 LLL 4 NBS 5 ORNL Ideal Proton detector in neutron beam YES YES YES Neutron energy measure- ment YES Extrapolation to zero proton pulse height YES YES © Solid radiator YES YES YES YES © Multiple-element detector YES © © 100% proton efficiency © YES YES Ref. Detector Code 1 . KFK - Telescope 2. LASL - One-element telescope 3. LLL - One-element telescc pe 4. NBS - Proportional counte >r 5. ORNL - One-element telescope Table II Detector System Comparison H Atoms Proton Neutrons Flight Radiator Proton System 2 per cm Eff. min max Path Area min max (X10 20 ) (%) (MeV) (MeV) (m) (Cm 2 ) {%) l-KFK(LOW) 0.4 % 60 2 6 57 75 " (HIGH) 0.8 % 3 5 30 57 75 ^10 2-LASL 0.41-2.2 1.0 1 >15 0.10 3.1 99 3-LLL 0.26-2.8 1.7 1 >15 60 95 60 4-NBS 65 100 0.005 0.8 200 4.9 5-ORNL 0.16-2.2 0.26 0.2 > 6 40 18 75 6-Proposed 0.3-3.0 14 -v- 1 >15 -- 75 50 7-Proposed 5-10 100 -v 0.1 >15 -- 100 ©= NO 59 Table III Sources of Error 1 KFK (LOW) (HIGH) ^ (all E n ) 0.6% (5MeV) 2 3 4 5 LASL LLL NBS ORNL 1) Error in a (e) ^0 0.6% 0.6% 0.4% 0.3% H (all E n ) (5MeV) (5MeV) (5MeV) (all E n ) (5MeV) 1.1% 1.1% 0.3% 0.1% (20MeV) (20MeV) (20MeV) (20MeV) 30% 6% (3MeV) 1.0 * 2 Si ? 21 2) Background 3% 1% 10% 3% 3% 3) Neutron-energy resol jtion 3% 2% 3% 4% 2.5% Energy (3MeV) (15MeV) (3MeV) (3MeV) (0.1 MeV) 4) 5) Counting rate (relat Pulse-height-extrapo error ive) lation 85 5% (<2MeV) 8 10 5 (1000) 20 1200 < 0.5% 6) Mp 10* 105 10 5 2 36 Detector material 85% Ar 15% N 2 85% Ar 15% N 2 Si Si 0.8%C 99%H 7) Inscatter (Typical ) <0.1% <0 . 1 % 2% 1% 1/2% 8) 2 Outscatter (Atom/cm Material ) negligible negligible negligible 21 7x10^' Al 8xl0 21 11x10 Al 60 SURFACE BARRIER SPECTROMETERS FOR CALIBRATION OF FAST NEUTRONS IN MeV RANGE O. P. Joneja, R„ V. Srikantaiah , M. R. Phiske, J, S. Coachman and M. P. Navalkar Neutron Physics Section Bhabha Atomic Research Centre Trombay, Bombay 400 085 India At present there are very few methods available for calibrating fast neutrons in the MeV range. In the present paper methods employing Li sandwich and proton recoil spectro- meters using surface barrier detectors for calibration of neutron energies and fluxes have been described. The results obtained with mono-energetic neutron sources in the energy range of 1-4 MeV are given. The accuracies for energy and flux calibrations are also discussed. (Energy and flux calibration, fast neutrons, isotropic neutrons, Li sandwich surface barrier spectrometer, proton recoil surface barrier spectrometer) Introduction At present there are very few methods avai- lable for calibration of neutrons from 1 MeV to 4 MeV, both in terms of energy and flux. The stand- ard methods such as time of flight or proton recoil spectrometers using gas for ionization involve either elaborate instrumentation or strong neutron sources or unfolding procedure which have to take into account wall effects or end corrections. Surface barrier spectrometers, on the other hand, are free from some of the above disadvantages and cover a wide range of energy. Moreover sandwich spectro- meters such as Li° or He-^, can be used for calibrat- ions with both beam and isotropic neutron sources. This aspect of calibration in isotropic neutron distri- bution as exist in reactors or moderating assemblies is very important since the other methods cannot be used in such a situation. In this paper, the use of surface barrier spectrometers for calibrating neutrons in the energy range of 1. to 4 MeV is dis- cussed with some of the experimental results. Description of the Spectrometers and the Experiment A Li sandwich spectrometer using two surface barrier detectors (250 mm active area with a depletion depth of about 400 microns) in coincident mode was assembled J_ 1_/. A block diagram of the electronics set-up is shown in Fig. (1). A thin layer of Li°F with thickness of 1 50yHgm/cm^ was deposited on a VYNS film to act as radiator. This method was prefered to vacuum deposition of Li°F directly on the surface barrier detector since the same detecting head could be used for background corrections. A similar type of proton recoil spectrometer using single surface barrier detector and hydrogenous radiator (200yt«gm/cm ) deposited on VYNS film was assembled /_2_/. The spectrometers were subjected to mono- energetic neutrons in the energy range of 1. MeV to 4 MeV obtained from p-t reaction using 5. 5 MeV Van de Graaff accelerator. The output of the accelerator was monitored by a calibrated long counter. Results and Discussions a) Li Sandwich Spectrometer A typical response of a sandwich Li" spectro- meter for neutron energy of 2. 23 MeV is shown in Fig. (2). The time integrated counts in a channel 'i corresponding to energy E is given by p. where M. Total number of Li atoms n \]S)~ Incident flux of neutrons of energy E falling on the radiator Normalised resolution function * Technical Physics Division, BARC \» (z\ = Total efficiency In order to calculate both energy and flux of the incident neutrons falling on the radiator, it is necessary to calibrate channel number and efficiency ' || ' which is energy dependent. The method of doing such a calibration involving both experimental and theoretical calculations is explained in APPENDIX. The results of energy calibration which is linear with channel number is given in Fig. (3), thereby showing that neutrons of unknown energy can be calibrated. 61 The time integrated incident neutron flux can be obtained using eqn. (1) if ^ (E) can be calculated. In the present case, the method of calculations for •I (E) which is given in APPENDIX was tested using four neutron energies which gave same time integra- ted flux within errors of 10%. b) Proton Recoil Surface Barrier Spectrometer A typical response of surface barrier spectro- meter for neutron energy of 2. 73 MeV is shown in Fig. (4). The proton recoil spectrum is given by A 2H>-. N T (cCe) to u z o or h- o LU _J LU Li_ O < or o < Q it: o o _J CD $ wQ ! 3<_l 2l< oz < *■ < STROBED SINGLE CHANNEL ANALYSER > . STROBED SINGLE CHANNEL ANALYSER j t I i t " CC Of omg LU 8 s ? o a. en o ooy 0. to > z en x o o a: £ o 1 i l! I i i i UJ Q * UJ c . > uj a > uja. p !^ z ►=!3 2 o d< ad< < u_ branching ratio R = a (n,a,) / ( a (n,a ) +o (n,a,)) and a ' n the o (n,a) ratio of °Li to B. The related experimental data have been compiled also since any subsequent evaluation and recommendation for o (n,a) or a (n,a,) datamust be con - sistent with the present knowledge of some related cross-sections which might be even more accurately known, for instance o tnt or the a(n,a) ratio of Li to io B . Some data sets of a(n,a) or a(n,a,) are deduced from measurements of R and o (n,a ) or from cr(n,a) ratio of ^Li to 10 B and a(n,a) of °Li . In an attempt to deal with independent observations related cross-section data are added to this compilation, but illustrated in separate plots. Many reviews and evaluations have already been made. See for instance [IRVING D.C. ( 1 967) , DERUYTTER A.J. (1967), GUBERNATOR K. (1968), SOWERBY M.G. ( 1 970) , STEWART L. (1972), HALE G.M. (1976) and LISKIEN H. (1976)] The 2200 m/s values of a (n, a) and R are 3835 ± 5 barn [DERUYTTER A.J. (1973)] and0.y3692+ 0.00006 [ DERUYTTER A. J, (1967)] and therefore o(n,a 1 ) = 3593 ± 5 barn. From ther- mal energy up to 1 keV the recommended fit of o(n,a ) of SOWERBY M.G. et al (1970) is claimed to be accurate within ±1%, within ± 2% up to 10 keV and within ± 3% up to 100 keV. According to SOWERBY M.G. (1965) the branch- ing ratio below 150 keV is constant 0.935 ± 0.005. Thus, below 10 keV the 10 B(n,a)/Li reaction is a well estab- lished and accepted standard. Although the cross-section above 10 keV is not particularly smooth and large and although a (n,n) of hydrogen is an accepted standard, boron 10 is often used in detectors even in that higher energy range because of the positive Q-value and ease of application in various detectors. To get well overlapping results in broad energy ranges be- tween 10 B(n,a)6Li and H(n,n)H normalised data, the o(n,a) cross-section should be known to within less or equal 2 per cent up to 1 MeV. The emphasis of the present review is put on the energy range from 10 keV to 1 MeV where most of the recent contributions deal with, and where the requested accuracy is still not achieved. A typical request for a(n,a) data in the range of energy from 10 keV to 1 MeV is 2% and similar requests have been formulated by eleven laboratories with unanimous priority I in WRENDA 76/77. Presentation and Discussion of Cross-Section Data General Information The status of experimental data for each cross- section type is drawn in Fig. 1 to II. The essential features or particularities of each measurement are briefly summarized in Tables 1 to 7 in Annex 1 . The compilation deals with about 5000 pairs of energy and cross-section values. Most of the data have been ob- tained on magnetic tape from the CCDN, Centre de Com- pilation de Donnees Neutroniques , Paris; some data were taken from publications and in some very few cases numerical data have been extracted from published graphs. All data were put on cards in the same format in a single set of units [eV, barn]. Some data sets with high resolution but poor statistical accuracy were summed into groups with improved statistical accuracy. Plots were made at the CBNM computer with the code ANGELA. The scale of the plots was adapted individually to get a clear picture even if a drawing comprises thousand data points. Together with the experimental data a continuous curve is drawn which is the ENDF/B-IV evaluated data, see MAGURNO B. A. (1975), made available by CCDN, and IRVINGS D.C. (1967) evaluated data for the branching ratio R. This compilation, together with some qualitative arguments on accuracies is part of CBNM's contribution to the INDC and NEANDC meetings on standards and dis- crepancies. An investigation of assessable accuracies of fits on the basis of a quantitative analysis is in progress now for some selected cross-section types and began with the compilation of an error file. 67 Data of o tot Data from 100 eV to 20 MeV are illustrated in Fig. 2. Detailed plots of small energy regions are given in Fig. 1, 3 and 4. A brief characterisation of the measurements and some comments are listed in Table 1 of Annex 1 . The measurements of MOORING F. (1966) and DIMENTK. (1967) agree well and they are the basic data for the evaluation of this cross-section. Measurements prior to those suffer from poor accuracy of sample composition. The data of SPENCER R. (1 973) were obtained with .2 ns/m resolution from 90 keV to 420 keV and do not show a suspected narrow resonance. Spencer's data are shown in their original high resolution fashion, namely 538 points, in Fig. 3, but in condensed fashion, with im- proved statistical accuracy, in Fig. 2. The promising data of AUCHAMPAUGH G. (1976) are claimed to have an uncertainty less than 1.5%. Below 10 keV DIMENT's data and fit are accurate to within 0.5%, and fulfil the requested accuracy. From 10 keV up to 400 keV new data of SPENCER R. (1973) make it senseful to investigate fits up to 400 keV to get a recommendable curve with properly defined accuracies. Additional measurements are needed from 400 keV to 1.5 MeV to fulfil the 1% request up to 1 MeV, (see WRENDA 1977, CASWELL R.S. n° 691016) since the avail- able data sets are scarce, DIMENT K. (1967) and BOCKELMAN C. (1951), and differ by 10% in that particu- lar energy range. Above 1.5 MeV five resonances are apparent and the ENDF/B-IV evaluation does not fit too well the experimental data as it is illustrated in Fig. 4. Data of g(n,a), g(n,aQ and o(n,ao) These data are illustrated in Fig. 5, 5 bis, 6 and 7. Characteristics of the measurements with some com- ments are listed in Tables 2, 3 and 4 of Annex 1. Below 10 keV the recommended data of SOWERBY M.G.(1970), with an accuracy of 2% at 10 keV, are commonly accepted. Above 10 keV much effort is still devoted to accurate measurements and many different methods have been and are still applied. Recent experiments by FRIESENHAHN J. (1974) and SEALOCK R. (1976) have been performed with refined and new techniques but their data did not solve the problems as such. By detailed consideration of Fig . 5 and in particular Fig. 5 bis, in the range of energy from 30 keV to 1 MeV, the data can be split into three groups: "high", "low" and "medium". The "low" group comprises the results of DAVIS E. (1961), BICHSEL H. (1957) and MACKLIN R. (1968), the "high" group com- prises the data of FRIESENHAHN J. (1974), BOGART D. (1968), COX S. X1966), MOORING F. (1966) and BILPUCH E. (1960), and the "medium" group comprises data of SOWERBY M.G. (1970), SEALOCK R. (1975) and ENDF/B-IV, HALE G. (1973). Handdrawn curves through the "low" and through the "high" group, with all data considered of equal weight, yield two distinct curves which differ by 10, 23, 30 and 25 per cent at 100, 250, 500 and 750 keV respective- ly. The data of SEALOCK R. (1975) coincide with the "low" group at 200 and above 600 keV, but with the "high" group at about 400 keV. The data of SOWERBY M.G. (1970) are not the result of an independent observation. The '"B(n,a)'Li cross- section was deduced from a measured ratio of (n, ) of "Li to ' ^B and 6Li(n,a) T cross-section which is not numerically documented. As a matter of test the author of this paper has used the well documented and accurate ratio measurements of SOWERBY M.G. (1970) together with 6 Li(n,a)T data of the ENDF/B-IV evaluation. It turns out that this B(n,a) Li cross-section is 2, 3.4, 3.3, 7.5 and 15.4 per cent higher than the recommended data of Sowerby at 20, 40, 60, 80 and 100 keV respectively. This and previous arguments illustrate that cr(n,a) of ' "B still does not have the status of a standard cross-section with an accuracy of 2% as requested in the energy range from 40 keV to 1 MeV. The peculiar "( n ,a) data of ENDF/B-IV above 1 MeV (see Fig. 5) are to be explained by the evaluation of <7(n,a0 data in Fig. 6. Above 1 MeV there are only two measurements, DAVIS E. (1961) and NELLIS D. (1969), the shape is similar but the amplitude differs by a factor 1.4. Evaluators have preferred the a(n,a,) data of NELLIS and have renormalized the data of DAVIS. The latter did also measure U relative to o(n,a) of 6 Li and to '0b, by WAGEMANS C. (1971), claimed to be accurate within ± 3%, are in favour of ratios that are 9% lower at 30 keV than the presently recommended ratios. Below 1 keV the ratios are well established, from 1 keV to 100 keV the accuracy gradually decreases with increasing energy and hardly gets± 3% at 100 keV. Above 100 keV experimental data are too scarce to draw significant conclusions. Many different measurements of R have been made, as illustrated in Fig. 10 and Table 6. Below 100 keV the branching ratio might be defined to within 1% and the emphasis of this discussion is put on the less accurate data from 100 to 1000 keV. The uncertainty of PETREE B. (1951) data is dif- ficult to estimate, though the energy dependence of R is smooth and consistent with more recent data. The data of BICHSEL H. (1952) and DAVIS E. (1961) suffer from large uncertainties of 10 to 20%. Data of SOWERBYM. (1965) scatter but nevertheless indicate lower values than any other experiment. The first experiment with surface barrier semi-conductor detectors , see MACKLIN R. (1967), did separate more clearly a from a, than any former measurement with BF^-proportional counting de- vices. The absolute error claimed by MACKLIN R. is 0.003 which is extremely small and never achieved so far. MACKLIN' s data should heavily determine recommend- able values but a smooth handdrawn curve through the bulk of the data does hardly fit likewise MACKLIN 's value at 500 keV and the LAMAZE G. (1974) value at 790 keV which is claimed to within a standard deviation of 0.03. The data of FRIESENHAHN J. (1974) have been deduced from separate measurements of ff(n,a,) and a(n,a). The results scatter and are systematically high as compared to others. The recent measurements by SEALOCK R. (1976) are made with a complex but promising method which essentially relies on surface barrier semi- conductor detectors. The experiment was not primarily designed for R determination. The data suffer from poor statistical accuracy, though a general trend towards low values is present. A critical appraisal of R-values from 1 00 to lOOOkeV shows that 3 to 5% accuracy for a recommendable curve might hardly be achieved with existing data. Conclusions The compilation of o(n,a) and a(n,a,) cross-section data shows that data requests from 10 keV to 1 MeV are not fulfilled. To define recommendable data with as- sessable accuracies a new evaluation is recommended. The evaluation ought to take full account of and be consistent with related and available accurate cross- section data, such as branching ratios and a(n,a) ratios of 6 Li to 10 B. It is questionable whether new measurements of o from 400 keV to 1 MeV would reduce the error margins of o (n,a) . New measurements of a(n,a,) or a(n,a) are inevitable if one has to achieve 2% accuracy up to 1 MeV. The large scattering of present a(n,a) data and the common use of a(n,a,) for flux determination in cross-section measurements tends to recommend primarily new a(n,a ) and branching ratio measurements. Cross-section ratio measurements above 10 keV of a(n,a) of °Li to B with a single detector are re- commended. Evaluations of B and °Li cross-section data should yield data sets which are consistent with the measured ratio of o(n,a) of °Li to B. Acknowledgement I should like to thank Dr. C. BASTIAN, Mr. R. BUYL and Mr. J. VAN GILS for providing and running the plotting programs and Mrs. P. DAEMS-LUYPAERTS for her diligent help in preparing the drawings. Data of o (n,a) of Li to 10„ As seen in Fig. 11 and Table 7 there are five ratio measurements and one calculated ratio. The latter was obtained from a(n,a) data of 6 Li and ] °B of ENDF/B-IV. The measurement of SOWERBY M. (1970) from 1 eV to 80 keV illustrates the 1 /v range and the departure from 1 /v from 500 eV on. Particular attention must be drawn to these measurements since they are the basis of all recommended '^B(n,a)'Li cross-section sets below 100 keV. See also comments on a(n,a) in previous para- graph . The data of BERGMAN A. (1961) are claimed to be within 0.4% accuracy and are systematically lower than SOWERBY 's data. The difference amounts to 3% at 30 keV. The data of FRIESENHAHN S. scatter considerably in the range of energy from 1 keV to 100 keV. Above 100 keV there is no other measurement to compare with, and a shift. of 12 keV is observed between their results and the ENDF/B-IV data. The data of PEREZ (1974) were de- duced by the author of this paper from published flux 69 REFERENCES ASAMI A., J. Nucl. Energy 24 (1970) 85 AUCHAMPAUGH G.F. et al . , to be published BECKER R. et al . , Phys.Rev. K)2, n° 5 (1956) 1384 BERGMAN A. A. et al., Soviet Phy . JETP L3_> n ° 5 ( ' 961 ) 895 BICHSEL H. et al . , Phys.Rev. L£8, n° 4 (1957) 1025 BILPUCH E., Ann. Phys . j_0 (1960) 455 BOCKELMAN C, Phys.Rev. 80, n° 6 (1950) 1011 BOGART D. et al . , Nucl. Phys. A J_25 (1969) 463 COATES M.S. et al . , Conf. Vienna "Neutron Standards and Reference Data" IAEA-STI/PUB/37 1 , p. 129,(1972) COOK C.F. et al., Phys.Rev. 94, n° 3 (1954) 651 C00KS0N J. A., Nucl. Phys. A ^46 (1970) 417 COON J.H. et al., Phys.Rev. 88, n° 3 (1952) 562 COX S.A. et al., J. Nucl .Energy 2J_ (1967) 271 DAVIS E.A. et al . , Nucl. Phys. 2_7 (1961) 448 DE PANGHER J. et al . , Rept. BNWL-260 (1966) DERUYTTER A. et al., J. Nucl. Energy 2J_ (1967) DERUYTTER A.J. et al. J. Nucl .Energy 2_7 (1973) 645 DIMENT K.M., Rept. AERE-R-5224 (1967) FOSSAN D.B. et al . , Phys.Rev. L23, n° 1 (1961) 209 FRIESENHAHN S.J. et al . , Rept. GULF-RT-A 12210 (1972) FRIESENHAHN S.J. et al., Rept. INTEL-RT 7011-001 (1974) GUBERNATOR K. et al . , Rept. EUR 3950 e (1968) HALE G.M. et al . , see MAGURNO B.A. (1975) HALE G.M. et al., Rept. LA-6518-MS or NEANDC(US)-200/U (1976) HAUSLADEN S.L., Dr. Thesis C00-1717-5, Ohio University HIBDON C. et al., Rept. ANL-4552 (1950) 6 HOPKINS J.L., see MAGURNO B.A. , p. 73 IRVING D.C., Rept. 0RNL-TM-1872 (1967) LAMAZE G.P. et al., Nucl.Sci. and Eng. 56 (1975) 94 LANE R.O. et al . , Phys.Rev. C4n" 2 (1971) 380 see also HAUSLADEN S.L. LISKIEN H., Conf ."Interactions of Neutrons with Nuclei" Lowell (1976) 1110 MACKLIN R.L. et al . , Phys.Rev. j_6_5 n° 4 (1968) 1147 MAGURNO B.A., Rept. BNL-NCS-50464 or NEANDC(US)-1 96/L or INDC(US)-73L or NEACRP-L-148 (1975) MOORING F.P. et al . , Rept. ANL-6877 (1964) MOORING F.P. et al., Nucl. Phys. 82 (1966) 16 NELLIS D.O. et al., Phys.Rev. C j_, n° 3 (1969) 847 NERES0N, Rept. LA-1655 (1954) PEREZ R.B. et al . , Nucl.Sci. and Eng. 55_ (1974) 203 PETREE B. et al . , Phys.Rev. i8_3, n° 6 (1951) 1148 PORTER D. et al . , Rept. AWRE 45/70 (1970) ROHRER, see BILPUCH E. (1960) SCHRACK R. et al . , priv. communication and Rept. ERDA/NDC 3/U (1976) 148 SEALOCK R.M. et al., Phys.Rev. C 13 n° 6 (1976) 2149 SOWERBY M.G., J. Nucl .Energy A/B 20 (1966) 135 SOWERBY M.G. et al., J. Nucl .Energy 24_ (1970) 323 SPENCER R. et al . , Rept. KFK 1518 and EANDC(E) 1 47"AL" (1973) STEWART L., Conf. Neutron Standard Reference Data IAEA-Vienna (1972) 149 TESCH K., Nucl. Phys. 37 (1962) 412 TSUKADA K. et al., J.Phys.Soc. of Jpn j8^ n° 5 (1962) 610 VALKOVIC V. et al . , Phys.Rev. K3_£ (1965) 331 VAUCHER B. et al . , Helv. Phys .Acta 4J3_ (1970) 237 WAGEMANS C. et al . , Ann. Nucl .Energy _3 (1976) 437 WILLARD H.B., Phys.Rev. 98 n" 3 (1954) 669 WRENDA World Request List for Nuclear Data INDC(SEC)- 55/URSF(1976) 70 a 4000- 2000 20 ENERGY [meV] Fig.1 cytot * rom 5meV to 100 meV. — 10 b 8 7 6 5 BOCKELMflN 1951 COON 1952 COOK 1954 NERESON 1954 BECKER 1956 ROHRER 1960 FOSSRN 1961 TSUKRDR 1963 MOORING 1966 DIMENT 1967 COOKSON 1970 PORTER 1970 SPENCER 1973 ENDF/B IV 1975 _J I I l_ 100 eV 200 400 IkeV 2 4 Fig. 2 Ctot f rom 100 eV to 20 MeV 10 keV 20 40 ENERGY 100 keV 200 400 1MeV 10 MeV 71 9.00 h 8.00 7.00 600 i 2E 5.00 tf 4.50 4.00 3.50 300 -1 1 1 1 1 1 - 1 1 — 1—1 1 1 1 1 1" -i 1 — i — t—r- -i — i — i — i — i i i i i — t- ii ii + BOCKELMAN 1951 X ROHRER 1960 * MOORING 1966 Q DIMENT 1967 X SPENCER 1973 -J — I — I — L_ _1 I I I I I— -J l_ _1 I I ' ' ' ' I I I I I l_ 10keV 20 40 60 80 100 keV ENERGY 200 400 600 800 IMeV Fig.3 o-jo, from 10keV to 1MeV. 2.60 240 2 20 200 *> 180 160 140- \ + + + BOCKELMRN 1951 * ♦ \ ♦/ ♦ COON 1952 \ + e> COOK 195U **♦ *\ / V X NERESON l95^ * \ + + f\ 1 + \t» Y BECKER 1956 + \ ++ / ++ V * / ° x X FOSSflN 1961 \ ii * t & TSUKflDR 1963 * +\ + **+ * T+ A + * + / i ° A C00KS0N 1970 Y PORTER 1970 + ^^ + \ ++ ♦/ +r \ ENDF/B IV 1975 \y y * ' i Vy ■ * ♦V y^a V ww + fn V . \i A */ lM * A '/ xlhI * Y w + v* ffli I v % • ■ X I f xx Ay i V ia ™>» n i hi Y Y YY »^ JT ^xTl 1 I Y I I XXI X^xiW.. • i tFwb — *3~~^__ I m — -^. I x ♦ • - J 1_ Y II V I Y 1 1 1 ' ' ' 1 12 14 16 18 2 Fig.* Cto, from 1MeV to 20MeV 3 4 5 6 ENERGY [MeV] > 72 8 9 10 12 14 16 18 20 10 1 — r — i — i 1 — i — i — i — n — i 1 1 1 — i — i — i 1 1 — i — t— n r 1 I T 1 Til! ii iii - 7 ^W^©*)* i 4 © & tfW^Sg J ■ t 1 I 2 r i 1 > > 0.7 O + X BICHSEL BILPUCH 1957 1960 °#© *+ i © o c* ft DRV I 5 1961 *,° A b 0.4 Y MOORING 1965 fi.° fiflr*n\ 0.2 0.1 0.07 Y X & & X MOORING COX MHCKLIN BOGRRT SOWERBY FRIESENHHHN SEHLOCK 1966 1967 1968 1969 1970 1974 1976 V.°0 \ \ \ V ^ ■ + ■ — V 004 ENDF/B IV 1975 ii iii lOkeV 20 40 IkeV 2 4 Fig.5 o'Cn.ao) + o'Cn.a,) from 1keV to 20MeV 100 keV 200 400 ENERGY »• IMeV 10MeV 20 - * 1 ■ . ■ 6 ■ 4 °^ f4k ^kj© 5 5&©Jbx : 2 ? ©v © © * © ° © o ©* * © -O . ""n **^\* ft* V , B. " + ^"~\ o c" b + BICHSEL 1957 + 1 X BILPUCH 1960 * ^"i. x ©* + ^^s, © ° O o es. I DHVIS 1961 c 0.8 Y MOORING 1965 * ft ^^v ° b Y MOORING 1966 ■ * ■ V \ » 0.6 X COX 1967 "* * ***\ ° A MHCKLIN 1968 A 0.4 & BOGRRT 1969 « S0WERBY 1970 ■X. o FRIESENHflHN 1974 v © s SERLOCK 1976 * V» z » ; — ENDF/B IV 1975 V X ■V" X" 0.2 ■ ♦V 10keV Fig.5 20 40 60 80 100 keV ENERGY bis cr"(n,a ) + o'(n J a 1 ) from 10keV to 1MeV. 200 400 600 800 1MeV 73 10 8 + DRV IS 1961 6 o MRCKLIN 1967 X NELLIS 1969 : 4 ^^k£t* Y CORTES 1972 ^y^^xxXjj X FRIESENHRHN 1 97U '. 2 t£ x* X SCHRRCK 1976 i i ■ ^fe~ A SERLOCK 1976 1 ■ ENDF/B IV 1975 0.8 ♦t*Sv£x . 2[ 06 ■ j 0.4 ; * * A" X* ■ d ■ »! K x ■ £ ■ \ * - b ■ 0.2 A* K ■ * £*\ ■ • V +*+ \ *A 0.1 ■ \ t *>» V v \ 0.08 • * %* , ** \ 0.06 . xv M 0.04 • + \ 0.02 ; y\y\ 0.01 I I I 1 1 1 1 1 1 1 1 1 i ill! i 1 1 1 1 1 1 + i i i i i * ' i i—i i i i ■ i * IkeV 2 4 lOkeV 20 40 Fig.6 tfd-i.a,) from 1keV to 20MeV. 100 keV 200 400 ENERGY »• IMeV lOMeV 20 — ! — i — i — r— r - 20 10 + DAVIS 1961 x MRCKLIN 1968 N SERLOCK 1976 - ENDF/B IV 1975 IkeV 2 4 lOkeV 20 40 Fig.7 o-Cn.ao) from 1keV to 20 MeV 100 keV 200 400 ENERGY ► 10 MeV 20 74 0001 0.0001- 100 eV 200 i.00 )keV 20 i.0 100 keV 200 400 ENERGY » Fig.8 Cross-sections: o"tot , o'(n,a ) + o / (n,a ] ), crXn.a,) and cCn.ao) Continuous curves are ENDF/B IV 1975 evaluation 75 P i i i i r T--i ■ ■ j TT 1 T 1 ■ -T T '" -I r 1 — i — i 1 r- -I — 1 — I— 1 — -i 1 1 1 — i — i 1 1~~T iJJ T T ° 1 1 r— i 1 — i — i r ! Ill — 35 /» ^ 5 X T 30 /x X . 28 x X\ ' 26 2A 22 ♦ ♦ . • ♦ * « ♦ g • 4 ♦ X x\ M\ x\ 20 X \ " 18 *\ x* U " 16 - o NILLflRD 1954 'Ax 1 - 1.4 + WILLflRD T-fl TESCH MOORING 1954 1962 1966 %" \ * X " 12 " Y FISflMI PORTER COOKSON 1970 1970 1970 " 1.0 VflUCHER LANE ENDF/B IV 1970 1971 1975 1 1 1 ■ ii*i ■ i i i ■ i IkeV 2 U lOkeV 20 40 100 keV 200 400 ENERGY > Fig.9 Elastic scattering cross-section from 500eV to 20MeV. IMeV 10MeV 20 100 090 080 ■ i » . * * . " 'l'<,i' ■ ' »« . > <» l * t\ m . i , -^ r ^_ Ji .\ PETREE 1951 BICHSEL 1951 OHVIS 1961 MRCKLIN 1965 SOWERBT 1965 MflCKLIN 1967 OERUTTTER 1967 FRIESENHfiHN 1974 LRMflZE 1975 SERLOCK 1976 IRVING 1967 6 lOkeV 20 40 ENERGY ► lOOeV 200 IM*v 200 400 WMeV Fig K) Branching ratio o'(n,a))/o'(n,ao) ♦ C(n,OL|) from K)eV to KDMeV 76 28 A* 20 x BERGMRN 1961 ■ 1.5 '. o SOWERBY 1970 [ x PEREZ 1974 + FRIESENHflHN 1974 * NflGEMflNS 1976 L r / * 1 1 09 „ 0.8 — ENDFB/IV 1975 1 1 i \ V r \ r \ t \ a I \* i *\ f o' ti. o „ 06 o < 05 : I ' y \ j \ '■ 04 1 * \ V 03 /*-* .* * * * • • \ + *.*»*•„*►•.«.**•. % * \ %** ♦ \s~ * "* ■ •*' * -W ,y y^ttTTw. - - ; , 02 lOeV 20 40 lOOeV 200 400 IkeV 2 Fig 11 Ratio of 6 Li(n,a)T to ,0 B(n,a) 7 Li from lOeV to 20MeV WkeV 20 40 ENERGY » 100 keV 200 77 ANNEX 1 . 1CL Table 1. Summary of a cross-section measurements of B in keV and low MeV energy range Investigator Name Date - Laboratory Data Energy min.-max. Number of points BOCKELMAN C. 1951 - WIS 20 keV - 3.4 MeV 197 COON J. 1952 - LAS 14 MeV 1 COOK C. 1954 - RIC 14.1 MeV - 1 8 . Me^ 7 NERESON 1954 - LAS 2 . 8 MeV -9.7 MeV 26 BECKER R. 1956 - WIS 4.4 MeV-8.6 MeV 66 ROHRER 1960 - DKE 3 keV - 82 keV 41 FOSSAN A. 1961 - WIS 3 . 3 MeV - 1 5 MeV 138 TSUKADA K. 1963 - JAE 3 . 2 MeV -5.1 MeV 59 MOORING F. 1966 - ANL 10 keV - 500 keV 55 Source DIMENT K. 1967 - HAR C00KS0N J. 1969 - ALD PORTER D. 1970 - ALD SPENCER R. 1973 - KFK AUCHAMPAUGH G. 1976 - LAS 76 eV - 953 keV 83 9.72 MeV 1 2 . Mev -4.8 MeV 7 90 keV - 420 keV 538 . 5 MeV - 1 1 MeV Experiment - spectrum - sample resolution Comments V.d.G. - mono - AE =20 keV determination isotopic composition might be in error D(T,n)a - mono - AE = 40 keV accuracy of sample composition not docu- mented V.d.G. D(T,n)a - 8 energies - AE < 40 keV accuracy of sample composition not docu- mented. Reactor - fission spec, -continuous E=10% p. recoil spectrom. - sample at least 98% pure V.d.G. mono - E = 40 keV, . ... n ' , . . no indication on accuracy of determination cross-sections about 15% greater than those of BARSCHALL-1946 of sample composition V.d.G. - mono - AE =30 keV 5% impurities of uncertain composition V.d.G. - mono - AE < 20 keV - 93% enriched B from ORNL no indication on accuracy of composition V.d.G. - mono - AE =10 keV same experiment yields also o jo •K.- e t 1 Pa 11 t0t composition of samples known to within 0.01 atom-percent. Linac - white - 0.5 ns/m sample ' °B content (93.0 ± 0.2)% corrections made for impurities V.d.G. - mono - AE m 40 keV elast .& inelastic scatt . angular cross section is measured V.d.G. - mono - AE = 30 keV - elastic and inelast .scatt . cross section measure- ment relative to H(n,n)H B-sample composition given to within 0.01 % V.d.G. - white - 0.2 ns/m. Up to 5% uncertainty in o by devia- _• • i • i i tot tions in chemical analyser. V.d.G. - white - 25 ps/m - agreement with ENDF/B-IV better than 1%. Previous- ly unknown resonances in ' ,. are re- i , • i i tot . , , , vealed - numerical data not available 1 .47 + 0.03 barns accuracy 0.03 barns Statistical accuracy ±3% at 3 MeV, ± 8% at 13 MeV statistical accuracy single point about 5%. unpublished, but see BLLPUCH E. . I960 tot 642 E~" +2.43 < 3% statistical accuracy statistical error < 2% overall error estimated < 3% accuracy 5% o-o < 100 mb abs na below 500 keV -% tot fit yields = (630.3 ±3.1)E (1 .95 ±0.10) for E < 10 keV and o (DIMENT) - o (MOORING) tot y n,n yields o — E abs up to 300 keV a = 1597 mb whereas FOSSAN measured 1430 mb integration over angle and known efficiency of calibrated scintillator yields a =0 , + o tot el. non-el. agreement with BOCKELMAN C. 1951 - MOORING F. 1966 and DIMENT K. 1967, if chemical analysis of IMF Frankfurt is taken. No narrow resonance. No indication of narrow resonance structure. Error< 1 .5% yet but expected soon. HALE G. Nov. 73 - BNL Evaluation ENDF/B-IV to 1 MeV R-matrix calculation plus o data of DIMENT K. 67. to 20 MeV smooth curve through DIMENT K.-67, BOCKELMAN C.-51, TSUKADA K.-62, FOSSAN D.-61, COON J. -52, and COOK C.-54. 78 Table 2. Summary of B(n,a +a,) Li cross-section measurements in keV and low MeV energy range Investigator Name Date - Laboratory BICHSEL H. 1957 - RIC BILPUCH E. 1960 - DKE Data Energy min.-max. Number of points 20 keV - 5 MeV 3 keV - 70 keV 32 Experiment Source - spectrum - detector flux shape - normalisation V.d.G. mono - BF, tubes assumed flat - norm, at 20 keV on ' /v long counter at 20 1 from thermal 4010 b. no experimental determination of o v n,a Comments Long counter not documented but see later BOGART D.-68 accuracy: 25% of ROHRER R. minus tot of HIBDON C. DAVIS E. 1961 - RIC MOORING F. 1964 - ANL MOORING F. 1966 - ANL COX S. 1966 - ALD DIMENT K. 1967 - HAR MACKLIN R. 1968 - ORL BOGART D. 1968 - LRC SOWERBY M. 1970 - HAR FRIESENHAHN J. 1974 - IRT SEALOCK R. 1975 - ORE 200 keV - 8 MeV 100 1 I keV - 77 keV 10 keV - 500 keV 55 1 1 keV - 250 keV 12 100 eV - 10 MeV 83 30 keV - 500 keV 6 30 keV - 800 keV 27 10 eV - 80 keV 33 1 keV -1.5 MeV 152 200 keV -1.2 MeV 21 + o^p < 100 mb, thus o n a =o abs hu:- n7 -67 V.d.G. - mono - BF3 grid ion. Accuracy: 20 - 30% E< 4.5 MeV long counter, E > 4.5 MeV known Li(p ,n)yield - long counter abs . by Pu-Be source not well documented in ANL-6877, but pro- see Mooring F.-66 bably preliminary results of type in MOORING F.-66 V.d.G. - mono - a and ratio a over Upper limit for a ,. + a ■ £°t . . nn , "' L o measured with transmission sample and scattering detector. No flux shape no calibration needed. V.d.G. - mono - spherical shell trans- mission method - no flux shape - no calib. needed Linac - white - o tot measured data for o n „ given, but ob- tained by calculation from a tot of DIMENT K. 1967 minus a n n of MOORING F. - 1966. accuracy: 2% small standard deviation and particular type of method used make results attractive. a tot _ a abs fits wel1 wit;ih a of LANE R. n,n accuracy : 10% V.d.G. - n,a determination by inverse reaction 'Li(a,n) lu B - long counter "4ff graphite sphere" - additional °n a. I °i\ a rat i° measurement with SBSC. V.d.G. - mono - BF3 - "Precision long counter" see DE PANGHER J.; norm, at 80 keV with ' /v from thermal 3840 Linac - white - ratio measured o of 6 Li(n,a) to ,0 B(n,a) Linac - white- B ion.ch. but also BF3 - flux shape by CH4 prop, counter, norm, at 4 keV by ' /v from thermal 3843,8 V.d.G. - white - SBSC-known Li(p,n) yield gives absolute energy dependent flux and other quantities absolute also, therefore absolute cross-sections calibration of BICHSEL H.-57 long counter by "precision long counter" yields revised BICHSEL data in good agreement. The a n Q of B is deduced from measured a(n,a) ratio of ^B to Li and by assuming a n a of ^Li to be known. The latter de- duced from " tot ~ a n n but no numerical data given. Cross- section given by an equation and uncertainty at 1, 10, 100 and 200 keV respectively is 1 , 2, 3 and 5%. cross-checks ion ch. with BFo-counter results deviate by 20%. detailed measurements of n,a„ ; n,o.i ; n,a„+a,; a (t?) and similar measurements for °Li(n,o)t. Accuracy ± 12%, but a n at 350 keV is about 1.8 times value of DAVIS E. or MACKLIN R. HALE G. Nov. 1973 - BNL Evaluation ENDF/B-IV n,a< n.ttj o is sum of a and a n,a n,Qo n,a, - below 1 MeV R-matrix calculation and experimental data of MACKLIN-68, DAVIS-61 and VAN DER ZWAN-72 - 1 MeV to 20 MeV based on DAVIS-67, but above 2 MeV renormalised by 1.4 - below 1 MeV R-matrix calculation and experimental data of FRIESENHAHN-72 - 1 MeV to 20 MeV smooth curve through DAVIS-61 and NELLIS-70 , data of DAVIS-61 above 2 MeV renormalised by 1.4 79 Table 3. Summary of B(n,a,7) Li cross-section measurements in keV and low MeV energy range Investigator Name Date - Laboratory Data Energy min.max. number of points Experiment Source - spectrum - detector flux shape - normalisation Comments DAVIS E. 1961-RIC 200 keV - 2 MeV V.d.G. - mono - AE>25 keV - BF grid 82 ion ch . - E<4.5 MeV long counter E>4.5 MeV known 7 Li(p,n) yield Long counter absolute by Pu Be-source error: 20 - 30% MACKLIN R. 1968 - ORL 30 keV - 800 keV V.d.G. - n, a o determination by inverse 7 reaction Li(a,n)'"B - long counter "4tf graphite sphere" - additionally a n a / a n a, measurement by SBSC. n,a and n',a, normalised at 30 keV on '/v from thermal 40 keV 31 Nal and/or GeLi - flux shape and flux by long counter - 7-ray efficiency also known! 6% COATES M. 1972 - HAR FRIESENHAHN J. 1974 - IRT SCHRACK R. 1976 - NBS 1 keV - 300 keV 63 1 keV - 1.5 MeV 56 5 keV - 600 keV 36 Linac - white - 1 ns/m - B2O3 and Nal - flux shape by spherical B-vaseline long counter - normalisation to SOWERBY at 1 keV Linac - white - Ae = 3 keV at 250 keV 'OB slab Ge(Li) flux shape methane proport. counter - normalised on /v at 4 keV from 3601,7 at thermal. 10„ o lower than values of Sowerby: 2% at 10 keV and up to 10% lower at 150 keV error on shape of o n a (E) is 2.5 to 4.6% from 'l IceV to 1 MeV. - Data of report FRIESENHAHN 1972 are superseded by rept. FRIESENHAHN 1974. In both reports c(n,a,) data are identical. Linac - white - 0.6 ns/m - ' "B slab Nal results Nal and Ge(Li) agree and Ge(Li) - hydrogen proportional counter within 5% - most accurate relative units normalised at 4 keV on results with Ge(Li) error 3% 3459,8 at thermal. between 8 and 400 keV, 5% be- tween 5 and 700 keV. SEALOCK R. 1976 - ORU . 2 MeV -1.2 MeV 21 V.d.G. - white - SBSC - absolute data by known ?Li(p,n) yield and known geo- metry - angular distribution also measured . error ranges from 5% at .2 MeV to 10% at 1.1 MeV. HALE G. et al , 1975 - BNL Evaluation ENDF/B - IV below 1 MeV - calculated from R-matrix parameters and experimental data of FRIESENHAHN 1972. 1 to 20 MeV - smooth curve through measurements of DAVIS 1961 and NELLIS 1970 with smooth extrapolation from 15 to 20 MeV. The data of DAVIS 1961 above approximately 2 MeV were renormalised by a factor of 1.4. 80 Table 4. Summary of B(n,a ) Li cross-section measurements in keV and low MeV energy range Investigator Name Date - Laboratory Data Energy min.-max. number of points Experiment Source - spectrum - detector flux shape - normalisation Comments DAVIS E. 1961 - RIC AE>25 keV - BF„ 200 keV - 8 MeV V.d.G. - mono 82 ion. chamber - E>4.5 MeV known ^Li(p,n) yield - long counter abs. by Pu-Be source 3 grid E<4.5 MeV long counter error: 20 - 30% MACKLIN R. 1968 - ORL 30 keV - 800 keV 7 V.d.G. - n,a determination by inverse reaction 'Li(a,n)'^B - long counter "4tf graphite sphere" and O a /a _ ratio measurement by SBSC. n,a and n,a, normalised at 30 keV on 1/v from o(n,a) equal to 3843.2 b and use of own measured ratio o(n,a ) to a (n,a,) equal to 0.0689± 0.006. standard deviation estimates 2 a 2.5%. Small error and type of method used makes results attractive. SEAL0CK R. 1976 - ORL 2 MeV - 1.2 MeV V.d.G. - white - SBSC - absolute data 21 by known 'Li(p,n) yield and known geo- metry - angular distribution also measured. error: 200 keV - 15%, 600 keV - 5%, 1.1 MeV - 9%. HALE G. 1975 - BNL Evaluation ENDF/B-IV below 1 MeV calculated from R-matrix analysis and experimental n,a data input for the fit were those of MACKLIN 1968 and DAVIS 1961. In addition, the angular distributions of VAN DER ZWAN 1972 for the inverse reaction were included in the analysis. from 1 to 20 MeV based on DAVIS-61 measurements with smooth extrapolation from 8 to 20 MeV, DAVIS-61 measurement above approx. 2 MeV was normalised by a factor of 1 .4. 81 Table 5. Summary of measured elastic scattering cross-section data in keV and low MeV energy range Investigator Name Date - Laboratory WILLARD H.B. 1955 - ORL Data Energy min.-max. number of points 0.5, 1 and 1.5 MeV + angular distri- bution measured from 30° to 120° at 9 angles additionally 3 points given for a Experiment Source - Spectrum - Resolution Detector V.d.G. - mono - hydrogen proportional counter - E dependent efficiency de- termined by calib. with long counter results claimed in absolute units Comments o n deduced by integration of measured do (i?) /dtJ which over-all error is ±15% a n n va lues deviate by 16% from er tot - a ?bg ; might _ be due to non-uniform density of the samples. tot abs TESCH K. 1962 - AAC 14 MeV - 1 point angular distrib. from 20 to 155° at 15 angles H (d,n)a - mono - associated a-particle technique: NE 102 and Pilot chemicals scintillator - absolute efficiency known by previous associated particle technique in non-scattered beam c n deduced by integration of measured do(#)d 10 UJ (X 10 20 40 60 PULSE HEIGHT CHANNEL COLLIMATOR FOR l0 B SAMPLE No I DETECTOR COLLIMATOR FOR CROSS SECTION MEASUREMENT DETECTOR CROSS SECTION MEASUREMENT DETECTOR Figure 3. Simplified experimental setup for implement- ing the 10 B(n,a)y) cross section. Figure 4. Pulse-height distributions for 510-keV neu- trons for a Nal detector system used in implementing the 10 B(n,ai7) cross section. The upper curve was obtained with a 10 B sample. The lower curve was obtained with a carbon scatterer. The arrow labeled 203 keV shows the position of the (n,n'j gamma-ray from iodine. The arrow labeled 421 keV shows the position of a peak which is a mixture of 438-keV gamma- rays from sodium and 417-keV gamma-rays from iodine. The use of Ge(Li) detectors significantly simpli- fies this problem as a result of the intrinsically better resolution compared with Nal and since the gamma- rays from inelastic scattering in germanium are well separated from the 478-keV line. Fig. 5 shows measure- ments by Orphan 3 of the pulse-height distribution for a Ge(Li) detector for neutron energies near 1 MeV. The 596-keV gamma-rays from 7l *Ge and the 695-keV gamma- rays from 72 Ge are well resolved from the 478-keV gamma-ray peak. The Ge(Li) detector technique does suffer from reduced efficiency due to limitations in the size of Ge(Li) detectors. Both Nal and Ge(Li) detectors have good time resolution, the Ge(Li) detector being somewhat better, so they can be used conveniently for white source neutron time-of -flight measurements. Throughout the useful region for this cross section (up to <300 keV), the systematic errors are probably ^1% 86 150 Id z < 100 i o \ CO 3 o u 50 "1 1 1 r 804 S E n < 1001 keV Ml I i i i in a) 600 800 200 400 600 CHANNEL NUMBER Figure 5. The pulse height distribution for a Ge(Li) detector used in implementing the I0 B(n,a 1 7) cross section for neutron energies from 804 to 1001 keV. for the Ge(Li) detector and somewhat larger for Nal de- tectors. These systematic errors are primarily a re- sult of uncertainties in the background and small un- certainties in the self shielding and multiple scatter- ing in the sample. In addition to these systematic errors which are appropriate to relative measurements, a solid angle and detector efficiency uncertainty must be included for absolute measurements. A convenient method for determining the product of solid angle and detector efficiency involves replacing the 10 B sample with a disk of equal area containing a uniform cali- brated source of 7 Be. 7 Be decays by electron capture to produce the 478-keV excited state of 7 Li . 10 B(n,a o + a!» 7 Li This reaction is implemented by employing detectors which will detect either the a or aj particles (and also possibly the 7 Li nuclei). The energies of these particles change with neutron energy thus causing some complication in the interpretation of data where the neutron energy is a significant fraction of the Q value. Historically this has limited the use of this cross section at higher neutron energies. The cross section has been implemented with proportional counters, ioniza- tion chambers, solid-state detectors and boron scintil- lators. The use of each of these detectors will be discussed below. Proportional Counters The simplicity of these detectors has led to ex- tensive use of these counters in neutron experiments. Fig. 6 shows a typical simplified experimental set-up. Typically the counter gas employed is 10 BF 3 . Neutrons incident upon the counter can interact with 10 B and the reaction products, 7 Li and an a-particle, will produce ionization in the counter as they come to rest. Consider an event which occurs near the center of the counter. Then for low enough neutron energies the re- action products lose all their energy before striking the walls. The total ionization will be closely pro- portional to the total energy of the reaction products, TlJ Figure 6. Simplified experimental set-up for the use of a 10 BF 3 gas proportional counter. (Q + E ). Thus, aside from resolution effects, a sharp peak should be observed for the lc B(n ,a ) 7 Li reaction and a separate peak for the 10 B(n,a 1Y ) 7 Li reaction. These two peaks will be separated in energy by 478 keV since the 7 Li gamma ray will not generally be detected. If this entire counter were being irradiated with neutrons, some of the reaction products would strike the walls of the counter before losing all their energy by ionization. This "wall effect" gives rise to a low energy tail in the pulse-height distribution. In Fig. 6 the neutron beam is collimated to a size smaller than the diameter of the counter. If the beam diameter is small enough (and the counter dimensions are great enough), the wall effect can be removed. The reduction in counting rate is yery large for a moderate size counter if the collimator is designed to entirely eliminate the wall effect. The collimator also removes the background resulting from direct beam neutrons striking the walls of the counter. An added benefit for linac measurements is that the gamma flash is significantly reduced by this collimator since much of this problem is a result of photons which interact with the walls of the counter. The time jitter of proportional counters is worse than that of the other detectors under discussion. This results from the fact that multiplication in these counters only occurs very near the central anode wire. So electrons must drift in a moderate field (i.e. slow- electron velocity) until they are essentially at the center of the counter. Thus the time between the reaction (and formation of ionization) and the formation of the signal at the anode varies from % zero for ionization concentrated near the wire to a maximum for ionization concentrated near the walls of the counter. The collimator which was described previously has the further advantage of reducing the time jitter by re- ducing the effective radius of the counter (note the effective radius is somewhat larger than the diameter of the collimator due to the reaction products which are directed toward the walls of the counter). The timing of the counter can also be improved by employing various gas mixtures. The proportional counter necessarily will have effects associated with distortion of the field near the ends of the counter (end effect) however, with modern counter geometries having reasonable lengths, this effect is small, aL/L ^1-2% (uncertainty in length/ length). In Fig. 7, the energy spectrum from a 5-cm 87 8000 6000 uj 4000 cr UU 0- co \- g 2000 — O o 1 ~T i — i — | — r~ i ^°B(n,a, l0 B(n,o, y) ? Li l0 B(n,a ) 7 Li '*, ^ 800 1600 2400 ENERGY DEPOSITED, keV Figure 7. Pulse height distribution for a 10 BF 3 gas proportional counter for thermal neutrons. diameter proportional counter exposed to thermal neutrons is shown. The two groups associated with 10 B(n,a,7) 7 Li and 10 B(n,a o ) 7 Li are well resolved. This resolution (^5%) was made possible by operating the counter with 10% 10 BF 3 and 90% argon. The argon was used to reduce wall effects by increasing the stopping power of the gas. It also reduced the time jitter of the counter. The long tail at low energies is a result en a 20 40 60 20 40 60 PULSE HEIGHT, ARBITRARY UNITS Figure 8. Pulse height distributions for 10 BF3 gas proportional counters for fillings of 20, 30, 40 and 60 cm Hg pressure for thermal neutrons. of wall and end effects. The low 10 B content in this detector limits its usefulness. Fig. 8 shows the effect of increasing the pressure of 10 BF 3 in a gas proportional counter. 5 The larger the pressure the poorer the separation of the two groups. This results from electron attachment to the 10 BF 3 molecules. However it is not necessary to resolve the groups. All that is needed is the total number of counts above a bias which is appreciably above the noise pulses. Thus useful counters can be made with % one atmosphere of i0 BF 3 . However the higher pressures lead to poorer time jitter (See Ref. 6). The rather poor timing is perhaps the worst limitation in the use of these counters. As the neutron energy increases the wall effect can become important. Also at high neutron energies, pulses from l0 B nuclei which have been scattered by neutrons are large enough so that the discrimination level must be increased. Thus a possible systematic error in the spectrum fraction can be introduced. The pulse-height distribution for a gas proportional counter with one atmosphere of 1 °BF 3 for neutron energies of about 500 keV is shown in Fig. 9. The low-energy pulses in 450 =. E_ =- 550 keV '* »V*v. r*v . *> isf MticuT [keV] Figure 9. Pulse height distribution for a one atmosphere 10 BF 3 gas proportional counter for neutrons from 450 to 550 keV. this Fig. are 10 B recoils. These recoils are present in all *°B counters which detect charged particles. The total systematic errors (mainly from room return neutrons, spectrum fraction and end window scattering) are in the 1 to 3% range for this type of detector. This detector is most easily implemented in a cross- section measurement as the counter located further from the source so that transmission corrections for the constituents of the counter need not be applied to the cross-section measuring counter. In white source experiments this reduces the effect of the time jitter of the counter also. 88 1 1 -1600 V (FRISCH GRID) -2700 V |'°B FILM) -1800 V (FRISCH GRID | Figure 10. Schematic layout of the Frisch gridded ion chamber used by Friesenhahn. Ionization Chambers Ionization chambers containing 10 BF 3 or solid 10 B deposits can be fabricated so that scattering from the chamber is small and the transmission corrections for the constituents of the chamber are easily calculated. A l0 BF 3 ungridded gas ionization chamber has been used by Weston? for low-neutron energies. A 20% 10 BF 3 + 80% argon gas mixture is used to reduce the range of the reaction products and to improve the timing of the detector. A timing of 35 ns has been obtained. Sum- ming of the reaction products occurs but for ungrid- ded chambers the pulse size depends on the track orientations of the a particle and 7 Li ion. Thus the pulse height distribution is very broad and the chamber is limited in usefulness to low-neutron energies where the pulse height distribution is essentially constant with neutron energy (i.e., the spectrum fraction is approximately constant). The use of gridded chambers removes this problem. Fig. 10 shows the layout for a detector with Frisch grids used by Friesenhahn.-^ In this chamber a self supporting l0 B film was employed which is a colloidal suspension of 10 B in a thin plastic film. The thickness of the film is % 200 ug/cm 2 . This is thin enough so that both reaction products can escape from the film. The, spacing between the film and the Frisch grid is 2 cm which is slightly greater than the range of the most energetic reaction products for the counting gas (10% C0 2 + 90% A). The electrons are drawn through the Frisch grid to the collector, all moving through the same potential difference to the collector. A fast pulse proportional to the ionization appears between the grid and the collector. Thus by summing the pulses at the two collectors shown here, a pulse proportional to the Q value of the reac- tion plus the neutron energy is produced. The princi- pal disadvantage of this detector is the low counting rate due to the thin 10 B deposits. To increase the counting rate, Friesenhahn employed eleven modules of this type separated by suppressor grids to prevent cross talk between adjacent modules. The complete ion chamber is more than 1 m long and the frames are 25 cm x 25 cm. It is the largest ion chamber of which I am aware. Fig. 11 shows the sum pulse height distribution for this detector for neutron energies from 1 to 10 keV. The greatest difficulty with the use of this detector is the determination of the spectrum fraction. At low- 1 1 1 1 1 1 _ o .* * 1000 800 • • • • 600 o °o o • 400 *> -oV. 1 1 1 **(**%fmw 1 1 60 80 CHANNEL NUMBER Figure 11. The sum pulse height distribution for the gridded ion chamber designed by Friesenhahn for neutron energies from 1 - 10 keV. 10* 47- MeV a-PARTICL -s — 0.84 -MeV 7 Li- PARTICLES m B{n,a r ) 7 Li* [478 5 ,0 B(/i.ov) T LiT478l.evl . 1 A tf«:t 2 ijl 1 keV — — *-i 2.6*/ 38 ki -4 PULSER j] — 4< V — 35keV eq— 1— - 1 01 -MeV 7 Li- I PARTICLES ,0 B(/.,al 7 U /I — i - 1.78 -MeV a- PARTICLES '°B(*, a ) 7 L, F= 5 i j 1 1 1 2 <(0 Z l (| 1 1 l [ - 1 1 i H — £ i — i — ° 4 U r 2 in' J I 1-7^ J \\ 1 1 - — r — u 1 — 1 u -Q- 2 10° -L, U- I ■ " - , 5 200 300 PULSE HEIGHT CHANNEL soo Figure 12. The pulse height distribution of the gridded ion chamber designed by Peelle for neutron energies less than 2 keV. neutron energies, as in this Fig., the fraction of the events below a reasonable pulse-height bias is quite small. However, at high neutron energies the bias must be increased to eliminate proton recoils from hydrogen in the plastic binder. In principal the spectrum fraction can be calculated however some of the input data which are needed are not known accurately enough, such as dE/dx of the film at low ion energies. The calculation must take into account such effects as the neutron kinetic energy, the branching ratio and the angular distribution of the reaction products. The spectrum fraction correction was 15% at 700 keV. The systematic uncertainty with this detector is % 2%. 89 In Fig. 12 the pulse-height distribution for a gridded 10 B ionization chamber designed by Peelle is shown. This chamber was designed to be used with gridded fission chambers for cross section measurements on a linac. The timing of this detector should be ^ 30 ns. This chamber contains thin deposits of 10 B evaporated onto a mylar backing. Only one of the reaction products is then observed except at very high neutron energies where the incoming momentum of the neutron in a fraction of the events permits both reac- tion products to go in the forward direction. Four peaks are observed since there is no summing. These spectra were obtained for neutron energies below 2 keV. In normal practice the pulse-height bias would be placed just above the 7 Li particles. The pulse height resolution is seen to be excellent. At high neutron energies, however, the spectra are more complicated than those shown in Fig. 12 as a result of kinematic effects. Then the separation between the a particles and 7 Li ions is not as clear and corrections must be made for the a particles lost below the bias. This is not the most important complication, however. The low count rate at high-neutron energies due to the thin film is the real problem. Solid-State Detectors The reaction products from the 10 B(n,a) 7 Li reac- tion can also be detected with any charged-particle detector. Solid-state detectors are convenient and the demands on these detectors are quite minimal com- pared with the state of the art. The depletion layer need not be very thick since the maximum energies of the reaction products are relatively low. The energy resolution of the detector can be fairly poor since the thickness of the 10 B film will dictate the resolu- tion of the system. It is preferable however to have large area detectors to increase the solid angle for counting rate considerations. errors in the cross section measurement. This chamber is evacuated so that the reaction products do not lose energy when passing from the foil to the detector. For the geometry of Fig. 13 the reaction products must pass through a thicker deposit of material than the areal density before they are detected in the solid-state detector. This worsens the pulse-height resolution of the system relative to an ion chamber of equivalent areal density. Also the solid angle must be determined with the solid-state detector system whereas an ion chamber detects all reaction products which are emitted from the foil which have energies greater than the bias energy. It is then obvious that the counting rate will be higher for the ion chamber than for a solid-state detector system of equivalent areal density. The limited angular range over which particles are detected for the solid state detector has one virtue. The kinematic effects which worsen the separation of the particle groups are reduced if the range of angles for which particles are detected is 1 i mi ted . C- -J SOLID STATE DETECTOR for FISSION FRAGMENTS a. 4000- l 1 3000- < m rr 2000- O i t- / < / z I 5 / <£ } 1000- CO / Q j ,1/ \ 1 U DEPOSIT Cz ■f2 -BACKING MATERIAL B DEPOSIT SOLID STATE DETECTOR for l0 B(n,a) 7 Li EVENTS Figure 13. The detector geometry employed by Wagemans and Deruytter for a measurement of the 235 U fission cross section relative to 10 B(n,a) 7 Li. Fig. 13 shows the q detector geometry employed by Wagemans and Deruytter for measurements of the 235 U fission cross section relative to the 10 B(n,a) 7 Li standard. The solid-state detectors are shielded from the direct neutron beam by a collimator. Fundamentally this is the same geometry as that involved in a double ionization chamber set-up. The nearly equivalent environments for the two films reduce the systematic 200 CHANNEL NUMBER Figure 14. Pulse height distribution for the solid- state detector system employed by Wagemans and Deruytter. In Fig. 14 the pulse height distribution observed by Wagemans and Deruytter is shown. This distribution should be compared with that observed by Peelle (Fig. 12). The resolution of the solid-state detector system is much worse due to the greater effective thickness of the 10 B film. The measurements of Wagemans and Deruytter only extend to 30 keV so the shape of the pulse height distribution should be essentially the same for the full energy region. The timing with solid-state detectors is very good (in the ns region). In particular, for this experiment it was not a limitation to the time resolution. The solid angle for the solid state detector can be deduced by calculation or by replacing the lu B film with a uniform calibrated alpha source having the same area as the 10 B deposit. In the Wagemans experiment the systematic errors in the use of the 10 B(n,a) 7 Li reaction are small , £ 1%. 90 1Q B Scintillators In the late 1950's and early 1 960 ' s a considerable effort was expended looking for a satisfactory method for implementing scintillators containing 10 B. Good timing should be possible with scintillator-photo- multiplier tube combinations. The kinetic energy of the reaction products would be summed and should pro- duce a convenient response function. The objective was to obtain high concentrations of 10 B so that high efficiency could be achieved. For some uses high con- centrations of 10 B would allow smaller scintillator thicknesses so that the multiple scattering corrections (and therefore their uncertainties) could be reduced. Scintillators which are loaded with 10 B compounds unfortunately poison the scintillator so that the greater the concentration of 10 B the lower the pulse height from 10 B(n,a) 7 Li reactions. In Fig. 15 the z =) 10 > t- < 0.8 _J Ul (E £ 6 0. ^\_ °0A I "Vz B CONO, Figure 15. Dependence of the light output and light absorption on boron concentration for a boron-loaded plastic scintillator. light output as a function of boron concentration for a plastic scintillator is shown. To obtain usable efficiencies the pulses were so small that photo- multiplier tube noise was an important factor. Various techniques were employed to handle these problems. For example, Bollinger 11 coupled two photomultipl ier tubes to one scintillator and by requiring a coincidence between the pulses from these tubes, he significantly reduced the noise rate. Also cooled photomultipl ier tubes have been employed to reduce the noise rate. Even pulse-shape discrimination has been used; this has the virtue of reducing noise and discriminating against Y-ray events in the scintillator. Unfortunately these methods are complicated and/ or awkward to implement. Thomas 12 fabricated a 10 B loaded liquid scintillator assembly coupled to a single photomultiplier tube which is usable without any of the special refinements mentioned above. The pulse- height distribution for that detector is shown in Fig 16. The pulse-height resolution is about 50% and photomultiplier tube noise is still a small problem; but, this represents the best results obtained to date with 1U B scintillators without the special techniques mentioned above. Though in the past 10 glass and plastic scintill ferent experimental techni little work has been done Some of the problems noted of non-uniform efficiency photomultiplier tubes empl that with the improvements and electronics in general possibility of improved sc this general area could yi B has been loaded into liquid, ators and a number of dif- ques have been tried, very recently on 10 B scintillators, by Bollinger 11 were a result of photocathodes for the oyed. It seems reasonable in photomultiplier tubes in the past 15 years and the inti 11 ators, more work in eld a satisfactory detector. THERMAL NEUTRONS 20 40 ELECTRON ENERGY (keV) Figure 16. Pulse height distribution of a boron loaded liquid scintillator for thermal neutrons. Conclusion A number of techniques are available at the present time for implementing the 10 B(n,a) 7 Li cross sections to an accuracy satisfactory for most cross section measurements. Further developmental work may improve the utilization of this standard. Improvement of the standard cross sections themselves is needed at the higher neutron energies, however, the rapid fall of the cross sections above ^500 keV may introduce fundamental limitations to the use of this cross section for some experiments. 91 References 4. M. S. Coates, G. J. Hunt, C. A. Uttley and E. R. Rae, Proc. Symp. Neutron Standards and Flux Normalization, CONF-701002, p. 401, U.S.A.E.C. (1971). R. A. Schrack, G. P. Lamaze, and 0. A. Wasson, to be published. S. J. Friesenhahn, A. D. Carlson, V. J. Orphan and M. P. Fricke, "Measurements of the 10 B(n,a lY ) and 10 B(n,a) Cross Sections," U.S.A.E.C. Report Gulf-RT-A12210 (1972). G. P. Lamaze, A. D. Carlson and M. M. Meier, Nucl. Sci. Eng. 56 (1975) 94. 5. I. L. Fowler, Rev. Sci. Instrum. 34 (1963) 731. 6. 0. K. Harling, Nucl. Instrum. Methods 34 (1965) 141. 7. L. W. Weston, private communication (1977). 8. R. W. Peelle, L. W. Weston, R. W. Ingle, J. H. Todd and F. E. Gillespie, Neutron Phys. Div. Ann. Progr. Rep. May 31, 1972, ORNL-4800, p. 10 (1972). and R. W. Peelle, private communication (1977). 9. C. Wagemans and A. J. Deruytter, Annals of Nucl. Energy 3_ (1976) 437. 10. G. I. Anisimova, L. S. Danelyan, A. F. Zhigach, V. R. Lazarenko, V. N. Siryatskaya and P. Z. Sorokin, Pribory i Tekhnika, Eksperimenta 1_ (1969) 49. 11. L. M. Bollinger and G. E. Thomas, Rev. Sci. Instrum. 28 (1957) 489 and L. M. Bollinger and G. E. Thomas, Nucl. Instrum. Methods 17_ (1962) 97. 12. G. E. Thomas, Nucl. Instrum. Methods 17 (1962) 137. 92 EVALUATION AND USE OF CARBON AS A STANDARD . J.C. Lachkar Service de Physique Nucleaire - Centre d 'Etudes de Bruyeres-le-Chatel B.P. n° 561 - 92542 Montrouge-Cedex, France . Available data on the carbon total and elastic scattering cross sections are reviewed. Major emphasis is placed on two neutron- energy regions below 5 MeV and between 8.5 and 15 MeV. The overall consistency of the recommended values has been established with the aid of previ- ously performed theoretical analyses. It is concluded that carbon elastic data below 5 MeV can be adopted as a standard except at the location of resonances. It is also suggested that the present need for high energy neutron standards could be satisfied by carbon data. (Standard ; evaluation ; carbon ; total cross section ; elastic scattering cross section ; R-function analysis optical model) . Introduction Carbon cross-section data are of interest in a number of areas of applied physics. Detailed knowledge of differential cross sections for neutrons scattered by carbon is required for calculations of neutron transport in materials for fusion and fission reactors. These needs have justified a large number of measure- ments, the earliest ones being performed 30 years ago. Since reasonably accurate sets of data were available, neutron data on carbon have been proposed as standards, garbon is a good candidate since it is a light nucleus and then the level density of the compound n + carbon system is small ; also the threshold for inelastic scat- tering is high. In 1970, at the Argonne Symposium on Neutron Standards and Flux Normalization, it was sug- gested that carbon is suitable as a transmission stan- dard up to 1.5 MeV but is marginal in angular distri- bution applications ' 2 . Since that time new measure- ments and more complete analyses have been performed in several laboratories ; consequently it has been conjec- tured that , by including all the presently available experimental informations on the total cross section, differential elastic cross sections and polarization in a R-matrix fit, a complete and consistent set of evalu- ated data could be generated up to 5 MeV 3 » "* . At higher energies, where the presence of resonances in the 13 C compound nucleus and competing reactions do not allow carbon to be considered as a standard scatterer, dif- ferential scattering cross sections for carbon are still of interest. In scattering experiments, absolute normalization is generally made by replacing the sample by a polyethylene scatterer : moreover, emitted neu- trons usually are detected using plastic scintillators. Thus neutron data for carbon, are needed for those two parts of the measurements. This need is more and more pronounced as the energy increases since the relative importance of the neutron data of carbon to the (n,p) scattering data increases with energy as shown in fig.1. In addition it is likely that a measurement of scattering from a secondary standard such as carbon, from which good samples are easy to prepare, would allow easier comparison of data obtained at various laboratories . Also, neutron induced reactions on 12 C produce mainly one Y" ra y corresponding to the de-excitation of its first excited level at k.kk MeV. This line may be a good candidate to determine the detector efficiency at high energy or to be a reference for Y -ra y produc- tion cross sections for neutron induced reactions at high energy 5 . Finally the neutron cross sections for carbon exhibit several well defined and sharp resonances and thus carbon may be considered as an energy-scale stan- dard in both white and monoenergetic source measure- ments 3 ' 6 . Here, we shaj.1 review the data on the carbon total and elastic scattering cross sections. We shall Fig.l. Energy dependence of the carbon total and elas- tic cross sections compared to the H(n,p) scattering cross section. concentrate on the most recent available data , mainly in the low-energy range. Contributions from 13 C also will be considered along with radiative capture cross sections. We shall then discuss the theoretical models used in the analysis of the data. By comparing theore- tical values to the experimental ones we shall give so- me error estimates. Neutron-induced reactions on carbon The Q-values and thresholds of the neutron-indu- ced reactions corresponding to the n + C system are derived from ref. The "C(n,Y) reaction has a Q value of U.9I+7 MeV. The threshold energy for the inelastic scattering is 1+.812 MeV. Natural carbon consists of 98.892$ of 12 C and 1.108$ C, so we have to consider the contribution of the 13 C content. Neutron data for 1 3 C have to be taken into account if a precision of 0.5$ or better is requi- red. The (n,y) reaction for 13 C has a Q value of 8.177 MeV and the threshold energy for the inelastic scatte- ring is 3.323 MeV. Analysis of the data The analysis of the data will be divided into three parts. In the first one, from the thermal value up to 2.0-MeV neutron energy, the total cross section for C exhibits no resonance structure. The second part extends from 2.0 MeV up to k. 8 MeV which corresponds to the threshold of inelastic scattering for 12 C. The 93 third one is up to 15 MeV. For each part, total and differential elastic scattering cross sections will be discussed. From to 2 MeV neutron energy . The total cross section for natural carbon has been measured and evaluated by many groups ; we have analysed most of these studies and used the data to propose evaluated total cross sections 8 . Our recom- mended values are shown in fig. 2. The most complete data considered in our analysis are listed in table 1 . They can be seen in fig. 3 which plots versus neutron energy the percent difference between each data set and our recommended values. The data of Cier jacks et al. 9 from KFK are characterized by a good energy resolution and a precise energy calibration ; however they have been found to be higher by about ]%, On the other hand, the data from RPI 10 seem to be low by about 1.5$. Finally nearly all the data from NBS ll , Harwell l 2 , ORNL 13 and ANL lk lie in a band of total width appro- ximately 1% of the average cross section. The thermal value of the total cross section has been evaluated by Leonard et al. 15 and Story et al. 16 These two evaluations were based on all the available carbon thermal cross section measurements from the year 19^-6 to 1970. Their adopted values are given in table 2. More recently, Mughabghab and Garber 17 have proposed the value of ^.750 ± 0,02 barn. This value is very consistent with that ( U . 756 ± 0.036 b) deduced by li from their data above 1 keV by as- Heaton et al. suming that the total cross section is constant bet- ween 10 keV and thermal energy. Our recommended value, ^.728 ± 0.008 b, is consistent with all the previous ones . t 1 — 1 1 1 1 r Thermal value t r ,2 c- iJ-, U L-, L_ l_ l_ l_. L_ U L, — l_- — L_ . , jffi 10* ioJ io* 10-" io 1 io 1 io 7 io 1 10 ' 10 « io 1 10' E.(.V Fig. 2. Recommended carbon total cross section below 2 MeV. Between 10 eV and 2.0 MeV, the variation of the total cross section with neutron energy has been ex- pressed by a polynomial in energy which was fitted to the averaged data by minimizing x^ are given by the expression: The proposed values a(E) = H. 725 - 3.251 E + 1.316 E 2 - 0.227 E 3 (barn) where E is the neutron energy in MeV. The accuracy of the total cross section is believed to be 0.5$ below 1 keV and less then 1 % up to 2 MeV. Such an analytical expansion does not take into account resonances in C which will be discussed below. Various and detailed measurements on differen- tial elastic cross sections have been reported below 2 MeV. Most of them, listed in table 3, have been pre- sented or discussed at the Argonne Symposium on Neu- tron Standards and Flux Normalization in 1970. All these data have been also considered to propose TABLE ] Available total cross section data. Authors Year Lab Energy-range (MeV) Remarks Ref Bockelman 1951 WIS 1.25-3.35 exp.data 20 Fossan 1961 WIS 3.30-16 11 23 Uttley 1968 HAR 7. 10 _5 -1.5 •• 12 Cierjacks 1968 HFK 0.3 - 30 II 9 Clements 1970 RPI 0.9-30 '• 10 Foster 1971 BNW 2.3- 15 I! 22 Nishimura 1971 JAE 10" 2 -0.3 tt 46 Perey 1972 ORL 0.2 -20 II 13 Meadows 1970 ANL 0.5- 1 .5 It 14 Holt 1975 ANL 1.5 -5.0 » 18 He a ton 1975 NBS I0~ 3 -15 11 11 Auchampaugh 1976 LAS 1.5- 14 prelimi- nary data 35 Francis Nishimura 1970 1971 KAP JAE IO" 10 - ,5 42 evalua- ted data 11 1 46 Perey 1974 ORL 10- |0 -20 ENDFB/IV 19 Lachkar 1975 BRC 10- 10 -20 evalua- ted data 8 "i 1 1 r T 1 1 1 r— sNBS eKFK *HAR ©RPI *0Rl -ani LU s- -2 -3 © © © * ¥ b * © * I® * * s .© •3 n *» a l » *». . « -' * * • ■$* ■■ *» • * ©S a H n. v 1 •"? B©»© • ML *• _L J_ _L 02 0.4 0.6 Oi 10 12 E (MeV) 1.4 16 1.8 10 Fig. 3. Ratios of carbon a total experimental data to recommended curve. NBS data are from ref. 11 , KFK from ref. 9 , HAR from ref. ll , RPI from ref. 10 J ORL from ref. l 3 and ANL from ref. l * . TABLE 2 Thermal neutron total cross section for carbon Authors Year Recommended Orn for thermal energy Ref Leonard 1970 4.730 ± 0.008 b 15 Story 1970 4.723 ± 0.0125 b 16 Mughabghab 1973 4.750 ± 0.02 b 17 Lachkar 1975 4.728 ± 0.008 b 8 94 TABLE 3 Summary of neutron elastic scattering measurements from 0.05 MeV up to 15 MeV. Authors Year Lab Ref Neutron-Energy range (MeV) Number of measured energies Measurement angles (deg) Remarks Lane Lane 1961 1969 ANL ANL 47 42 0.05 -2.20 0.10-2,0 56 30 25-145 22 - 145 Ang. Distr. Ang. Distr. Ahmed Wills Holt Meier Galati Knox 1970 1958 1975 1954 1972 1973 GEL ORL ANL LAU KTY OHO 48 27 18 26 24 39 0.5 -2.0 1.45- 4.10 1.8 -4.0 2.2 -3.8 3.0 -7.0 2.63 31 12 12 8 32 1 20-160 35- 145 20- 160 35 - 145 15 - 160 17-118 Pol. Ang. Distr. Ang. Distr. Ang. Distr. Ang. Distr. Fasoli 1973 PAD 25 2.1 -4.7 29 20- 160 Pol. Ang. Distr. Total X Perey 1969 ORL 28 4.5- 8.5 5.2- 8.7 13 40 15 - 140 15- 140 sect . Ang. Distr. Velkley 1973 ABD 29 7.0- 9.0 5 10- 150 •' Haouat 19/4 BRC 30 8.5 - 1 1.0 |8.0- 14.5 4 14 30- 160 10- 160 11 Glasgow 1976 DKE 31 1 9- 15 14 25-160 " evaluated data in ENDF/B IV . The new results since that time are from ANL where Holt et al . : e have mea- sured angular distributions at 1.8 and 2.0 MeV. Their data have been found consistent with the recommended angular distributions from ENDF/B IV except at forward angles. This deviation might suggest that some changes would be necessary in the existing evaluations at least above 1 . 8 MeV . Above 0.1 eV the integrated elastic scattering cross section can be taken equal to the total cross section. This results in a (n,y) cross section that is too small. The deviations between the integrated elas- tic cross section, determined from polynomial fits to the measured angular distributions, and the recommended total cross section are less than the experimental un- certainties which are about 3%. Differential cross sec- tion are generally quoted with a precision of less then 5%. From 2- to 4.8-MeV neutron energy . At the neutron energy of 2.077 ± 0.002 MeV, the 12 C + n reaction reaches the resonance at 6,863 MeV in 13 C ; the total width of this resonance is 6 keV and the value of the total cross section at the peak is 6.020 b taken from ref. 13 . The narrowness of this resonance implies that a good energy resolution is nee- ded in the vicinity of 2.077 MeV. This may be a severe constraint in the extension of the use of carbon as a standard above 2 MeV. Between 2.1 and 2.7 MeV, the total cross sec- tion show no resonance (fig.lt) and, here, carbon still could be used as a standard. In this energy range,three sets of data from NBS ll , 0RNL 13 and the older ones from the University of Wisconsin 20 have been found very consistent. The accuracy is of the same order as below 2 MeV, i.e. not greater than \% . Recently reporr ted data from ANL ! 8 were also found to be in good agreement. The major interest in these measurements lies in a very good energy resolution which was obtai- ned with the aid of monoenergetic source techniques. At E n = 2.816 ± 0.00U MeV, most of the recent data exhibit a maximum associated with the resonance at 7-5^5 MeV in 13 C. This resonance was ignored in the evaluation file ENDF/B III but was included in the ENDF/B TV file. The total width of less than 5 keV re- ported by Ajzenberg-Selove 21 for this resonance is consistent with the data of ref. 9 > 11 » 13 » 18 , Fig. 4. Comparison of neutron total cross sections of carbon between 2 and 5 MeV (a - ref.*, b = ref. 20 ' 2 , a = ref. ll , d = ref. 13 , the evaluated curve is from ref. e ). Large discrepancies appear at the 3.05 MeV in- terference dip due to differences in the resolution and energy calibration precision. Our adopted values were based on the data of KFK 9 and 0RNL ' 3 . The re- sults of Foster and Glasgow 22 present a significant shift to lower energies and those from Fossan et al. 23 are subtantially shifted to higher energies. Above 3.05 MeV up to k.Q MeV no other fine structure is ex- pected and only some smooth variations such as the broad 3. 58 MeV resonance and the 1+.261 MeV one are en- countered. The proposed values, obtained by smoothing the composite results of ref. i ' 11 ' 13 in the energy range 2.8U-U.8 MeV are given with 2% accuracy. Three recent sets of elastic scattering data are available from the University of Kentucky 2 \ from the Uni- versity of Padoua 2 5 and from ANL 1 8 . They are higher accuracy repetitions of older ones from Meier et al. 26 and Wills et al. 27 . Average relative uncertainties of the measured dif- ferential cross sections vary from 5 to 7 %. The adcoted values are then quoted with an accuracy of less than \ %. The angle-integrated cross sections deviate from .the recommended total cross sections by 3 to 5 %• 95 High resolution neutron scattering experiments are in progress at ORELA from 2 to 3.2 MeV but the data are not yet available . From 4.8 to 15 MeV . Between U.8 and 8.5 MeV neutron energies as many as 12 resonances in 12 C are reached, some of them being good candidates for energy-scale standards. But, for our purposes, they make this energy range of limited value. In addition two exit channels are open by the inelastic scattering and the (n,a) reaction. The data from KFK 9 , NBS ! ORNL 13 and University of Wisconsin 23 , which are in good agreement for the resonances ener- gies have been considered. Between the sharp resonances the recommended total cross section values are very close to those of Heaton et al . ll and the uncertainty is about 2%. The values at the resonances are taken from Per ey et al . 13 and the accuracy is better than h%. Above k.8 MeV and below 8.5 MeV, elastic angular distributions a (9) have been extensively measured main- ly by Galati et al, n up to T MeV, by Fcrey et al. 28 from h.5 to 8.7 MeV and by Velkley et al. 29 above 7.2 MeV. The data expressed in the center of mass system were fitted to a Legendre polynomial expansion using the least squares method. The coefficients are plotted ver- sus energy in fig.5» they show large variations at the location of the various structures observed in the to- tal cross section. Fig. 5. Recommended Fg Legendre coefficients for elas- tic scattering below 15 MeV. The coefficients are defined by : da dQ. znt (v -"irl^ v h p i (cos CM ). 06 Fi 04 ■ • i •.:' '• 2 li 0.0 \-if ' 06 F2 04 '/■ ■ ;. : \ i ' 02 . ■'. \ K J\ 00 A ■ . ^ Fa 04 It r \ ■ 00 '""jsi/ 02 F 4 0.1 -; r . 00 ....y~^ F 5 006 004 \ ■. . .. 000 .'/■''■ Fe 004 002 ! 000 4 8 12 E n (MeV) In the region extending from 8.5 to 15 MeV the resonances reached in the 13 C compound system become broader than in the lower energy range . Moreover they are generally overlapping so that the cross^ section variations with neutron energy are rather smooth. The data from KFK 9 , NBS 1T , ORNL 13 , BNW 22 and from the University Of Wisconsin 23 are in satisfactory agree- ment. Nearly all of them lie in a band of total width approximately Q% of the average cross section. The recommended values are very close to those from ref. 13 and they are given with h% accuracy. In this energy range, differential elastic scattering cross sections are well known. Evaluated data have been mainly based upon the time of flight measurements performed at Bruyeres-le-Chatel by Haouat et al. 30 between 8 to 1^.5 MeV. Two independant sets of data were obtained in two measurements periods and they were found consistent with each other. The norma- lization used was the n-p scattering cross section near 0°. Glasgow et al . 31 have measured the elastic angular distributions between 9 and 15 MeV. Their data are in overall good agreement with those from ref. All these data compare favorably with the previous measure- ments below 9 MeV from Perey et al. 28 and Velkley et al. . The integrated elastic cross sections are given with an accuracy varying from 6 to 8%. The adopted va- lues are then taken within a global uncertainty of less than 1% , The comparison of these values with those from ENDF/B III and ENDF/B IV shows significant devia- tions by about 7 to 15$, The recommended values from the previous evaluations are too high at about 11.0 MeV and systematically low between 12 and 1 h . 5 MeV. The evaluated Legendre coefficients of the elas- tic angular distributions have been deduced with a rea- sonable accuracy due to the overall consistency of the available data. The degree of confidence in the evalua- ted data was increased when they were compared to the data of Bucher et al . 32 from forward angle elastic scattering measurements and those of Morgan et al . 33 from backward angle measurements made at ORELA. Reactions in 13. Competing reactions Wow we have to consider the contribution of the 1.1$ of 13 C present in natural carbon. The existing data of Cohn et al . 3 " from 0.112 to 22.8 MeV and those of Auchampaugh et al . 35 from 1 . 5 to 1 k MeV show that the cross section of this iso- tope is very close to the corresponding one for C except in the vicinity of the ''C resonances observed at neutron energies of 0.153, 1.751, 2.1+32 and 2.U5U MeV 17 . The first one at 153 keV has been observed in the total cross section measured by Heaton et al. L1 using natural carbon sample. This is mainly due to the large value of the total cross section (a = 21 ± 2 b) and to the small width of the resonance ( r = 3.7 ± 0.7 keV). This example shows clearly that good 13 C data, including differential elastic cross sections which have never been measured, are needed for a good knowledge of carbon data. Radiative capture in carbon The recommended values for the (n,y) cross sec- tion for thermal neutrons are for the two isotopes of natural carbon : 12 C a(n,y) t h = 3.36 ± 0.3 mb 13 C a(n,y)th = °-9 ± 0-2 mb. These recommended values taken from ref._ 17 have been strongly influenced by the data of Jurney and Motz 36 . At low energy, i.e. E n < 100 keV, the (n,y) cross section has been assumed to vary with neu- tron energy according to the 1 /v law. The radiative capture is the inverse of the nuclear photo-effect ; by applying the reciprocity 96 theorem, the (n,y) cross sections can be deduced from the (y _ n) data. The excitation function for the 13 C (y, n) reaction has been measured by Cook 3? ; using these data we have obtained the (n,y) cross sections from 0.2 to 20 MeV 8 . The radiative capture cross sec- tion from thermal energy up to 20 MeV is very low and does not exceed 0.07$ of the total cross section. Theoretical analysis In order to confirm the overall consistency of the evaluated values deduced from the review of all the available experimental data a theoretical analysis is required. Such an analysis would lead to accurate knowledge of the scattering phase shifts. The phase-shift analysis of differential cross sections can be performed in a rather simple way since the ground- state spin of 12 C is + and then the chan- nel spin has only the value S = 1/2 + .In this formalism, the center of mass differential cross section can be written as 3e : where t =1 da dfi (9) = %' imQ P/(cos 9) (-£ + l)e .st If, I") -i&£ f . = sin 9Y sin fy + 4 " sin i<5$ sin &{ - e i<5 ^ in 6[j. &l = 6 {£, J =1 ± 1/2) is the phase-shift of the par- tial wave of total angular momentum J =£± 1/2. Pj(cos 9) and P'^(cos 9). are the 2 t)a - order Legendre polynomial and its derivative, \ is the incident wave length. For E below k.8 MeV when only the elastic scat- tering channel is open, if the radiative capture pro- cess is supposed negligible, the phase shift can be expressed as the sum of two real quantities, the first one being the contribution from hard-sphere scattering, the second being due to scattering from compound - nucleus resonances. When nonelastic channels are open the phase shifts become complex quantities. Phase-shift analyses have been carried out by several groups. Some sets of parameters were obtained by fitting a limited set of measured angular distri- bution data 26 ' 27 9 an d the others were obtained by simultaneously fitting the total cross section, the differential elastic scattering cross sections and po- larization data 1 8 ' 2 "*' 2 5 ' 39 ' over a large energy range. In addition to direct fits, R-matrix calcula- tions and coupled channel calculations were used to derive the phase shifts. R-matrix calculations . In the' low energy part below U . 8 MeV, extensive R-matrix calculations have been performed at Ohio University 39 , at Yale University University of Padoua 25 . Some other calculations are in progress at LASL and at ORNL, In this energy range, the application of R-matrix theory is relatively easy since the R-matrix reduces to a R function. This func- tion has an explicit energy dependence given by : and ANL l 8 , and at R 2 ■ + V ^\lj is the reduced width and E.« is the level energy. The sum over X denotes the number of levels which are explicity considered, while R^j is the contribution to the R-function from distant levels. This last contribu- tion has been treated either as a constant R* ^ ) or expanded as follows l 8 : r- = R * J + B f . e. C J o 1 .R£j All the levels in 13 Cupto an excitation energy of about 10 MeV except the ones at and at 7-5^5 .MeV have been considered in the various analyses. The scat- tering in carbon below 2-MeV neutron energy, is strongly influenced by the -1.86-MeV, 1/2 + , state. Two other bound states have been included specifically : the -1.09-MeV, 5/2 + , state in the calculation of ref. 39 and the -1 .27 -MeV, 3/2~, state in the calculations of ref. ltl and 18 . The two strongly interfering 3/2 + states at 2.95 and 3.58 MeV influence both the cross sections and polarizations. The R-function is defined from the boundary condition for the radial part of the wave function ug(r) at the channel radius a : 1 du. dr + B. ) / R„ . The Bj, quantity is the boundary condition fac- tor chosen to cancel the level shift factor at the resonance energy. In the different analyses, two distinct values were adopted for the channel radius a. Holt et al . "* 1 have taken a equal to the interaction radius which is U.61 fm. Then they determined the reduced width of the S!/ 2 hound state at - 1.86 MeV by fitting the expected cross section to the data below 2 MeV. They found Yq 1/2 = 0.69 MeV. The p 3 / 2 bound state reduced width was deduced from polarization data. On the other hand, Lane et al. '* 2 have assumed that low-energy data are dominated by the contribution from only the above men- tioned Sj/2 bound state. The s-wave phase shift was obtained from the analysis of the data for a(9) below 0.8 MeV. The deduced parameters were a = 3.72 fm , E Q 1/2 = ~ 6.0 MeV, Yq 1 / 2 = k-0 MeV. The last one is large and close to the single particle estimate which is U.87 MeV t|2 . In this analysis other fits also led to higher values of a as large as U.9 fm. The comparison of the adopted R-function para- meters derived from these analyses is given in table k. Although they are different, up to 3 MeV they give equally good fits to all the data. The corresponding phase shifts are given by the relation : S^CE) = -0^arctan[p^ R^/ MSj-B^R^ . where TEST AND FUNDAMENTAL PHYSICS 1 1 UAAMCT / 1 1 / / ' 1 / POSITIVE ION BEAM FROM 3 MV VAN DE GRAAFF Fig. 2 Van de Graaff Experimental Area MOVABLE SHIELD BLACK PRECISION DETECTOR NEUTRON BEAM COLLIMATOR MONITOR 4.5° HALF ANGLE a LONG COUNTER Fig. 3 Standard Neutron Beam Line beam direction. Tests have shown that the back angle detector is the better arrangement. The secondary- monitors are calibrated by means of the Black Detector using a pulsed proton beam. Continuous beam operation is used for dosimetry measurements in order to increase the neutron flux by a factor of 5. Black Detector Neutron Flux Monitor Our initial experiments have been limited to the neutron energy region from 200 keV to 2 MeV. The ' 7 3/3 neutron sources were the Li(p,n) Be and H(p,n) He reactions. The neutron source is surrounded by a large neutron shield in order to reduce room back- ground. The shield is mounted on an air table in order to produce easy access to the end of the beam pipe for target charges. The neutron beam is collimated to a cone with a 4.5 half angle by means of a removable lithium loaded polyethylene insert. This angle pro- vides a 19 cm diameter neutron beam at a distance of 120 cm from the target. The primary neutron flux monitor is the Black Detector located in a shield approximately 6 meters from the target. The beam size incident on this monitor is determined by a precision machined collimator 30 cm long with a 4.445 - 0.010 cm diameter cylindrical hole. This shield is also mounted on an air table for easy movement. The proton beam energy is calibrated by measuring the Li(p,n) Be and 3 H(p,n) 3 He thresholds and checked by measurements of carbon transmission near the 2077 keV resonance. The neutron flux at the dosimeter position, r, is determined from the yield of the Black Detector by the expression The Black Detector is a cylindrical shaped plastic scintillator viewed by a single photomul tiplier tube as is shown in Fig. 4. The scintillator is 12.5 cm in diameter with a length of 15 cm and a 5 cm diam- eter reentrant hole 2.5 cm deep. The purpose is to completely absorb the incident neutron in the scintil- lator so that the light output is proportional to incident neutron energy. Hence the name, Black Detec- tor, as christened by Poenitz . A detailed presenta- tion of the operation and calibration of this monitor is given in another contribution to this symposium by M. M. Meier <> The neutron detection efficiency is measured by means of the associated particle method and also cal- culated by means of a Monte-Carlo type program. The results for 500 keV and 750 keV neutrons are shown in Fig. 5. The points represent the measurement while the solid curves are from the calculation. Since the two methods agree within - 1%, the efficiency is taken from calculation for intermediate energies. The detector efficiency is greater than 907„ for neutron energies between 200 keV and 1 MeV. cp(r) = R BD r 3 " e T(R - r) A where ^(r) is the neutron flux at distance r R is the distance from source to exit of collimator BD is the Black Detector counting rate e is the Black Detector efficiency A is the effective area of the collimator T(R - r) is the air transmission between the Black Detector and the position r. For thick dosimeters which absorb a large frac- tion of the beam, it is necessary to use secondary monitors placed outside the direct beam for neutron flux monitoring. The monitors include a long counter placed at 90 to the beam direction and a BF3 counter located inside the shield at approximately 180 to the — LIGHT PHOTON O RECOIL PROTON O NEUTRON PHOTO CATHODE ANODE NEUTRON BEAM °- " PLASTIC SCINTILLATOR NE110 PHOTOMULTIPLIER TUBE RCA 9854 Fig. 4 Black Detector Operation 116 4.000 560 keV 250 keV { ] - CALCULATED RESULTS ' EXPERIMENTAL RESULTS - CALCULATED RESULTS ■ EXPERIMENTAL RESULTS I BIAS CHANNELS 20 40 60 80 100 120 140 160 CHANNEL NUMBER 8 I I I I I I I I I BLACK DETECTOR I i/j « NEUTRON TIME ol FLIGHT z I En ■ 560 lieV => I TARGET CURRENT ■ 07 MA _ >- (C 26 r- m cr < ; TARGET CURRENT ■ — _l LtJ 2 4 Z < I / — v. / - \- §2 o o — /a - ^^ — - -* J i /, I I i B I I I I - 1 1 1 1 i ~- I2cm^ ! - 8F 3 \ - EC Ul '/////, m0.6 V//A 04 Fig. 200 400 600 800 1000 NEUTRON ENERGY, keV 7 Ratio of observed neutron flux to incident flux for moderating cylinder The next measurements were done with larger mod- erating cylinders from two different manufacturers as shown in Fig. 8. These are the Andersson-Braun type detectors which are supposed to have a uniform dose equivalent response from thermal to 10 MeV neutron energies. The detectors are composed of polyethlyene cylinders 21 cm in diameter and a length of 24 cm with a small BF3 proportional counter in the middle. The moderated flux is tailored by means of a boron absorb- er with holes in order to produce a response which approximates the neutron dose equivalent in tissue. The neutron beam was incident on the front, flat end of the counter as is shown in the figure. Both detec- tors were calibrated by an Am-Be neutron source at a distance of 1.5 m using a dose equivalent to flux con- stant of 3.49 x 10 rem/(n/cm ) as given by Nachtigall . Fig. 8 10 15 20 25 DISTANCE, cm Diagram of larger moderating neutron rem counter OBSERVED TO CALCULATED RESPONSE 1.50 1 1 1 1 DETECTOR 1 • DETECTOR 2 1.25 I 1.00 " \ - 0.75 i \ i I 0.50 _ 0.25 n l 1 - 250 500 750 1000 NEUTRON ENERGY, keV Fig. 9 Ratio of observed to calculated response for two rem counters The results of the measurements are shown in Fig. 9. The ratio of the observed dose equivalent to that calculated from the incident neutron fluence, is plot- ted as a function of neutron energy for the two detec- tors. The response of both detectors follows the same general shape, although they differ in absolute cali- bration by 25%. Two values are shown for each detector at 250 keV. These result from different extrapolations of the neutron fluence to dose equivalent relationship at that energy. 118 Figure 10 shows the dependence of dose equiva- lent on neutron fluence as a function of neutron energy. The points connected by the solid lines are from the 1971 report by the National Council on Ra- diation Protection and Measurements while those con- nected with the dashed line are from the 1957 NBS Handbook No. 63 9 . These recent results differ by nearly 507. from the earlier results which were in use at the time of the Andersson-Braun detector invention and the Am-Be source calibration. Since no values are given for 250 keV, the value at this energy is obtained by extrapolation from the points at 100 and 500 keV. Due to the rapid variation in this region, the value obtained is, of course, de- pendent on the method of extrapolation and can easily vary by a factor of two. The circle on the graph is a result of logarithmic extrapolation which pro- duces a better fit to the measurements as is shown by the lower datum points at 250 keV in Fig. 9. The higher values result from a linear extrapolation. Thus logarithmic extrapolation should be used in the 100 to 500 keV interval for the data in the NCRP report and not linear extrapolation as recommended in the report. FLUENCE TO DOSE EQUIVALENT CONVERSION 1 "1 1 1 1 1 III 1 1 1 1 II 1 1 1 . • TABLE 2 NCRP 38 (1971) - - o EXTRAPOLATION A NBS HANDBOOK 63 (19571 " - A- \ v -\ - <£ ■* \ ^- - v -a- ■* •*— a-A ,,l 11,1 II 0.1 1.0 10 NEUTRON ENERGY , MeV Part of the difference between observed and incident dose equivalent in Fig. 9 is due to the use of different values of the neutron fluence to dose equivalent conversion factor for the detector cal- ibration and the incident fluence. The older 1957 results were used for the former while the 1971 results were used for the latter. Figure 11 shows the results obtained if the older fluence to dose equivalent conversion is used for both the calibration and the fluence measurement. For detector No. 1 the response agrees with that predicted from the incident fluence while the response of detector No. 2 is ap- proximately 257. low. Since the error bars indicate the statistical errors only, the 257. difference be- tween the responses of the two similar detectors is an indication of the systematic errors involved in neutron dose equivalent measurements with typical neutron dose meters. Measurements with Nuclear Test Films The Nuclear track film used in dosimeters has an energy threshold at approximately 500 keV and an exposure threshold of approximately 100 mrem. As a check of these limits, film dosimeters were exposed to 250 keV, 500 keV, and 1 MeV beams with exposures rang- ing from 7 to 120 mrem. The films were exposed in groups of six in both free air and while mounted on the chest of a plastic phantom. The films were then treated in the usual manner by the processing labora- tory. The processing laboratory scanned the film visually and observed no exposure for the 250 keV and 500 keV beams. This is as expected for a 500 keV energy thresh- old. Since the results for 1 MeV neutrons were the same, within statistical error, for both the phanton and free air film backing, only the results observed with the phantom are shown in Table 2. The first col- umn indicates the incident dose equivalent in units of millirem determined from the fluence measurement, the second column lists the total number of observed tracks in each group of 6 films, while the final column lists the exposure deduced by the processing laboratory. The listed error is only that from the finite number of observed tracks. Fig. 10 Fluence to Dose-equivalent conversion factor DETECTOR RESPONSE 2 ' O 0.50 > K Ul CO CD 0.25 O ° DETECTOR I • DETECTOR 2 NBS HANDBOOK 63 (1957) 250 500 750 1000 NEUTRON ENERGY, keV Fig. 11 Ratio of observed to calculated dose equivalent for two rem counters. Conversion factor is taken from ref. 9 The observed dose agrees, within the ± 127. statis- tical error, with the exposure at the 110 mrem level but is systematically high for the lower exposures. These results confirm the observation that the typical film dosimeters used in personnel monitoring do not function for neutron energies below approximately 500 keV and for exposure levels less than approximately 100 mrem. Different type monitors are required for these regions Table 2. Film irradiation results for 1 MeV neutrons Incident Dose Equiva mrem lent 0b served T ra< ks Do Observed se Equivalent mrem 7.7 36 56 i 9 15.4 28 43 i 8 30.3 43 66 i 10 110 77 119 ± 14 119 Summary A calibrated neutron fluence beam is available at the NBS Van de Graaff with an accuracy of - 27. in the energy region from 200 keV to 1-0 MeV with low back- ground. This large sized beam is ideal for testing of instruments at low flux levels which is becoming more important as permissable personnel dose levels are lowered. Examples of the use of this beam for checking the response of typical radiation monitoring equipment, including the Andersson-Braun type monitor, were given. This facility along with the other neu- tron sources available at NBS offers a unique oppor- tunity for standardization of neutron dosimetry for neutron energies below 15 MeV» References 1. R. B. Schwartz, Proceedings of International Specialists Symposium on Neutron Standards and Applications, Paper G7 , (1977). 2c D. M. Gilliam, Op. Cit„, Paper H4. 3. C. M. Eisenhauer, Op. Cit., Paper 12. 4. W. P. Poenitz, Nucl. Instr. and Meth. 109, 413 (1973). 5. M. M. Meier, Proceedings of International Specialists Symposium nn Neutron Standards and Applications, Paper G2, (1977). 6. "Protection Against Neutron Radiation", National Council on Radiation Protection and Measurements Report Number 38, Table 2, (NCRP Publications, Washington, D. C, 1971). 7. I. 0. Andersson and J. Braun, "A Neutron rem Counter With Uniform Sensitivity from 0.025 eV to 10 MeV", Neutron Dosimetry, Vol 2 , p. 87 (Inter- national Atomic Energy Agency, Vienna, 1963), Proceedings of a Symposium, Harwell, 1962. 8. Do Nachtigall, "Average and Effective Energies, Fluence-Dose Conversion Factors and Quality Factors of the Neutron Spectra of Some (a,n) Sources", Health Physics _13, 213 (1967). 9. "Protection Against Neutron Radiation up to 30 Million Electron Volts", National Bureau of Standards Handbook 63, (Superintendent of Documents, Washington, D. C, 1957) o 120 INTERNATIONAL NEUTRON DOSIMETRY INTERCOMPARISONS R. S. Caswell National Bureau of Standards Washington, D.C. 20234 Three recent international neutron dosimetry intercomparisons are discussed: the International Neutron Dosimetry Incomparison (INDI) sponsored by the International Commission on Radiation Units and Measurements (ICRU); the European Neutron Dosimetry Intercomparison Project (ENDIP) sponsored by EURATOM; and the intercomparison carried out by the centers doing neutron radiotherapy. Physical dosimetry to an accuracy of two or three percent is desired in order to achieve a generally-accepted 5% accuracy in dose to the tumor. In general it is found that 3% accuracy has not been achieved by the intercomparison participants; however, the radiotherapy centers agree on an arbitrary (but not absolutely known) scale within this uncertainty. (Standards, medical, neutron dosimetry, intercomparisons, radiation effects). Introduction Perhaps the first question one might ask is: Why do intercomparisons at all? Some of the chief reasons are: (1) to verify the international measurement scale; (2) to obtain information on systematic errors; (3) to compare measurement methods or instruments; (4) to com- pare corrections used by different laboratories; (5) to evaluate the state-of-the-art in measurement; and (6) to motivate scientists to make better measurements. In reference to the first reason above, it is wery important that 500 rads of absorbed dose to a tumor be the same in one hospital as in another so that com- parison of clinical results is on a uniform basis, and one hospital may benefit from the experience of another hospital. It is also important to a national standards laboratory to verify that errors have not been made in the laboratory's own absolute measurements—comparison with one or more other standards laboratories gives some degree of assurance that no blunder has been made, and that the international measurement system is uniform, even though it may not be known absolutely to the ac- curacy to which uniformity can be established. If national standards laboratories do not offer standards for a particular kind of measurement, then intercom- parisons represent a way of testing the uniformity of those institutions that constitute the international measurement system. Evaluation of systematic errors, unlike random errors for which there is a good theory, is very dif- ficult and is usually done by an estimate or guess. Intercomparison provides a way for getting at systematic errors. If all laboratories using independent methods get the same result we gain confidence that the system- atic errors of each method are small. On the other hand, if one laboratory using a particular method is 20% higher than all other laboratories which are in agreement to, say, 3%, we tend to look for errors in this one method or in this laboratory's procedures. A scientist participating in an intercomparison to some extent has his vulnerabilities exposed. He is therefore likely to try to do the best possible job of measurement to make sure that his numbers stand up well in the comparison wi/th his peers. Three international neutron dosimetry intercom- parisons will be discussed here, although there have been a number of others: (1) The INDI (International Neutron Dosimetry Intercomparison) comparison was sponsored by the International Commission on Radiation Units and Measurements (ICRU) and was carried out by Leon Goodman and collaborators at the RARAF Accelerator facility at Brookhaven National Laboratory, in 1973 with 14 international participating groups. ' (2) The ENDIP (European Neutron Dosimetry Intercomparison Project) comparison was sponsored by EURATOM and carried out at TNO Rijswijk, Netherlands for dosimetry for neu- tron therapy, and at GSF Neuherberg (Munich) for neutron protection, in 1 975. 4,5,6 The Radiotherapy Centers' intercomparisons have been carried out on a bilateral basis with participants from two laboratories performing measurements at a common radiotherapy facility. '»^ INDI Comparison The physical arrangement for the INDI intercomparison is shown in Figure 1. The ion beam from the accelerator I ACCELERATOR ^BEAM TARGET ~\ 5cm DEPTH. 10cm DEPTH- 20cm DEPTH- PRECISION LONG COUNTER'' -|z: m Ocm REE — 5cm (MONITOR ION CHAMBER) 20 cm FRONT FACE -, 30cm(KERMA) -U. PHANTOM J ^(REMOVABLE) 30cm CUBE ALIGNMENT''' TELESCOPE (TRANSLATABLE) ■200 cm FRONT FACE ALIGNMENT JT" TELESCOPE (FIXED) 0° INTERCOMPARISON ARRANGEMENT, TOP VIEW Fig. 1. Arrangement for the INDI intercomparison. 121 was incident upon the target where the neutrons were produced. At 5 cm downstream from the target was a transmission ionization chamber which monitored the total radiation (predominantly neutrons) in the experiments. Air measurements were carried out at 30 cm from the target. A 30 cm x 30 cm acrylic box filled with distilled water served as a phantom and could be located with its front face 20 cm from the neutron- producing target. Water was used, rather than tissue- equivalent liquid, to avoid problems of change of composition with time. Multiple monitoring was carried out, including a Precision Long Counter along the beam line at 200 cm, and 3 BF-, counters at various depths in moderators, which werS sensitive to changes in neutron spectrum. Positioning was done optically using an alignment telescope with such accuracy that errors due to position could be neglected. A photograph of the experimental area is shown in Figure 2. Fig. 2. Photograph of the experimental area used for the INDI intercomparison at the Brookhaven National Laboratory RARAF facility. The principal monitor for the experiment is shown in Figure 3 and is a parallel-plate tissue-equivalent-wall CAVITY WALLS TISSUE EQUIVALENT PLASTIC (MUSCLE) STEM, TISSUE EQUIVALENT PLASTIC (MUSCLE ELECTROMETER CONNECTOR, ALUMINUM transmission chamber. The principal gamma-ray monitor is a GM counter^, also shown in Figure 3, and is located below the beam line at a horizational distance of 15 cm. A photograph of the transmission ionization chamber is shown in Figure 4. 10 CENTIMETERS Fig. 3. Transmission ionization chamber monitor and location of phantom and GM counter used as a gamma-ray monitor in the INDI comparison. Fig. 4. Photograph of transmission ionization chamber monitor used in INDI experiment. Measurements were carried out by participating groups for 11 measurement conditions. Five measurement condi-.-. tions were in air: 15.1 MeV, 5.5 MeV, 2.1 MeV, 0.67 MeV and for Cf-252 neutrons. In addition measurements were made at three depths in the phantom for each of the two higher neutron energies, 15.1 and 5.5 MeV. The measure- ment conditions and reactions are summarized in Table 1. A summary of the INDI results in air is given in Figure 5. Note the larger spread in the results at 0.67 MeV where the dose was low. Results are grouped tightly at the two higher energies. At the lower energies the tendency of the C ? H. counter and the C2H4 ionization chamber to read low is evident. This is believed to be due to the necessity of applying a conducting film, usually graphite, to the polyethylene to make the chamber wall conducting. Secondary particles ejected by neutrons may be absorbed in this conducting coating of the chamber wall. The tissue-equivalent ionization chambers and acetylene-polystyrene equivalent chambers (C^H,,) are intrinsically conducting, so this problem does not exist. The silicon diode and the precision long counter are calibrated independently, but do not yield results very different from those of the tissue-equivalent ionization chambers. A more detailed look at the measurements at 15.1 MeV in air is given in Figure 6. Note that measurements were made over a period of roughly a year, requiring very good Van de Graaff accelerator radiation field stability. The stability of the calibration fields used in the experiment is believed to be good to 2%. Note that RARAF made mea- surements at 4 times during the procedures, and these all agree within a spread of 2%. One can see that the mea- surements are generally in agreement with the uncer- tainty quoted by the experimentalist. However, one should not be content with this situation, because the absolute uncertainties are relatively large, typically 122 Table 1. Neutron energies and kerma rates Reaction Neutron energy MeV Percent energy spread (±) Nominal maximim total tissue kerma rate in air at 30 cm Gy h _1 (100 rad h" 1 Approximate percent gamma-ray tissue kerma, relative to total tissue kerma in free air T(d,nrHe D(d,n) 3 He T(p,n) 3 He T(p,n) 3 He fission 15.1 4 5.5 7 2.1 5 0.67 20 252 Cf spectrum 0.40 0.80 0.20 0.05 0.06 3 5 5 5 5 40 About 2 mg. i i i i I 0.67 I VW$ > I I "1 I I I I I I I I TE V C 2 H 4 D C 2 H 4 Clr • C;H 2 O SI DIODE o PLC « + F L £?^0 • NEUTRON ENERGY/ MeV Fig. 5. Summary of INDI comparison results in air. Fig. 6. Summary of INDI measurements at 15.1 MeV from December 1972 to November 1973. Solid circles, solid squares, erect hollow triangles and open circles all refer to tissue-equivalent ionization chambers with different systems for measuring the gamma-ray component of the mixed neutron and gamma-ray radiation field. Inverted open triangle is for C2H2 ionization chamber, erect and inverted solid triangles are for C2H4 ioniza- tion chambers. Open square is for the C2H4 proportional counter, diamond is for the Precision Long Counter, and the circle with dot in the center is for the Si diode. 1% or 8%. Since many of the measurers used similar systems, for example the tissue-equivalent ionization chamber, one might expect that these methods would agree to much better than the quoted uncertainty. However, this is not the result of the present comparison. Figure 7 shows the results of the measurement of gamma-rays in the presence of neutrons, expressed as a percentage of the neutron kerma in air. Several dif- ferent systems of gamma-ray measurement were used, the GM counters and the film having generally the lowest neutron sensitivity. Some negative results are shown particularly at 15.1 MeV which indicate that the neutron sensitivity assumed for the gamma-ray detector was probably too high. Note that for Cf-252 (for which one should read the right hand scale on the figure) the gamma rays are a much larger fraction of the neutron dose than in the other cases, and the agreement of the various measurement methods for the gamma rays is in fact much better. Measurements at 15.1 MeV in air and in phantom are summarized in Figure 8. Note that this is not a depth dose curve, since the phantom is not present during the measurements in air. A tendency for the polyethylene-ethylene dosimeters to read low in the lower neutron energy spectra associated with depth in the phantom is evident. Finally, in Figure 9, are shown measurements in air and phantom at 5.5 MeV. Notice the spread in results at 20 cm depth, where the dose rates are rather low, making it difficult to make accurate measurements with the ionization chambers (which 123 1 1 1 1 1 1 • GM COUNTERS 1 1 1 1 1 I ■ Mg-a A c-co 2 Al-Ar CHAMBERS CHAMBERS ♦ TLD V FILM 15.1 ~0.67 MeV 2.1 MeV i " 2 Cf i 5.5 MeV J ♦ MeV — ■ 58 56 A • 54 A • ♦ 52 ■ 50 ■ A >• ■■ ■ B ■ i A A A 48 4b X ♦ ■ * 44 A ^ A . _ 4: :♦. ■ A ■■ A A 1 1 I 1 1 A 1 1 1 1 1 1 i> 3° i i r 5.5 MtV NEUTRONS IN AIR AND PHANTOM — * TE V _ f C 2 H, O C 2 H 4 Clr • 2 H 2 _ Si Diode o PLC ♦ b o — — — — — 10'!.. 15.1 MeV K N 5/11 3/11 3/11 K ( t 5/11 3/11 3/11 5.25 MeV K N 7/10 2/10 1/10 So. 8/10 2/10 0/10 2.1 MeV K N 9/11 0/11 2/11 K ( ( 9/11 1/11 1/11 0.57 MeV K N 9/11 0/11 2/11 K ,„, 9/11 1/11 1/11 252 Cf neutrons K N 7/8 1/8 0/8 So, 8/8 0/8 0/8 TABLE 3 VALUES OF ABSORBED DOSE AND KERMA (RELATIVE TO THE MEAN) AS DETERMINED BY THREE PARTICIPANTS AT GSF AND TNO FOR 15 MeV NEUTRONS Conclusion The Radiotherapy Centers, following the intercomparisons, are now largely on the same measurement scale which is well-known relatively, but not absolutely. It is the author's belief that more efforts in the laboratory are needed to correct possible systematic errors and to under- stand these consistent differences that appear from laboratory to laboratory throughout many different mea- surement conditions, before further international com- parisons will be worthwhile. However, following an effort at eliminating systematic errors, international comparison would be justified to see if the attempted corrections of systematic errors have in fact worked. In view of the importance of this problem, it seems appropriate that the National Standards Laboratories should be developing standards and providing calibration services for neutron dosimetry. Such calibrations could follow three largely independent routes to see if the same calibration is achieved following each route. These routes are: (1) calibration of neutron dosimeters with known fluences of monoenergetic neutrons using kerma factors to obtain the dose (a subject very dependent on the work being discussed at this conference); (2) using a gamma-ray calibration of an ionization chamber and appropriate ratios of W, stopping power and kerma factors to determine the dose; and (3) use of a tissue-equivalent calorimeter with appropriate corrections for calorimetric defect to determine the dose or kerma in the radiation field. Condi Mom TNO GSFM CENF GSF, free In oir K N 0.97 1 08 1 .17 Sot 0.97 1 08 1 .17 TNO, free in oir K N 0.99 1 05 1.05 So, 1 .00 1 06 1.05 TNO, 5 cm depth D N 0.94 1 06 1 .11 D ,o. 0.94 1 08 1 .10 TNO, 10 cm depth D N 0.94 1 05 1 .11 D ,o, 0.94 1 07 1 .10 TNO, 20 cm depth D N 0.91 1 05 1 .17 D .o. 0.94 1 07 1 .14 125 Table 4. Intercomparison results: ratios of measurements with respect to TAMVEC measurements during the Spring of 1974.8 a Data suspicious owing to leakage problems, b MRC excludes TL gamma component c MRC excludes 6% gamma component d MRC excludes ~97° gamma component Participants MRC TAMVEC U of W TAMVEC NRL TAMVEC U of W TAMVEC NRL TAMVEC Location MRC U of W NRL TAMVEC TAMVEC Beam 16 MeV d+Be 21.5 MeV d+Be 35 MeV d+Be 50 MeV d+Be 50 MeV d+Be Tissue kerma in air 0.89 b 1.04 a 0.99 0.99 1.00 Dose at depth 0.86 c @ d max 0.86 d @ 10 cm 1.01 @ d max 1.02 (? 10 cm 0.99 (3 d max 0.99 @10 cm 0.99 (3 2 cm 0.99 (310 cm 0.99 (3 2 cm 0.98 (310 cm Photon Calibration 1.01 (8 MeV) 1.01 ( 60 Co) 1.03 (137 CS ) 1.00 (60 Co ) 0.99 ( 60 Co) Table 5. Intercomparison results: ratios of measurements with respect to TAMVEC measurements during the Autumn of 1975 and February 1976. Participants CHHRI TAMVEC MRC TAMVEC NRL TAMVEC Louvain TAMVEC Chiba TAMVEC Chiba TAMVEC Location CHHRI MRC NRL Louvain TAMVEC TAMVEC Beam 15 MeV d+T 16 MeV d+Be 35 MeV d+Be 50 MeV d+Be 30 MeV d+Be 16 MeV d+Be Tissue kerma in air 0.95 a 1.01 1.00 1.01 1.01 Dose at depth 0.97 (3 1 cm 0.96 (3 5 cm 0.97 (310 cm 1.02 (3 2 cm 1.02 (3 10 cm 0.97 (3 5 cm 0.96 (310 cm 0.97 (315 cm 0.96 (320 cm Photon Calibration 1.00 (137 CS ) 1.00 ( 60 Co) 0.99 b (60 Co ) a Choice of parameters account for 4% of the 57., difference (see text) . b Ratio of calbration made in Japan to calibration made at M.D. Anderson Hospital. 126 References 1. ICRU Report 29, An International Neutron Dosimetry Intercomparison , International Commission of Radi- ation Units and Measurements, Washington, D.C. 20014. U.S.A., to be published (1977). 2. R. S. Caswell, L. J. Goodman, and R. D. Colvett, International Intercomparison of Neutron Dosimetry, in Radiation Research, Biomedical, Chemical, and Physical Perspectives , 0. F. Nygaard, H. I. Adler, and W. K. Sinclair, eds., Academic Press, New York (1975), p. 532. 3. L. J. Goodman, R. D. Colvett, and R. S. Caswell, An International Neutron Dosimetry Intercomparison, Proceedings of the Second Symposium on Neutron Dosimetry in Biology and Medicine , Commission of the European Communities, Luxembourg (1975), p. 627. 4. J. J. Broerse, G. Burger, and M. Coppola, Basic Physical Data for Neutron Dosimetry , J. J. Broerse, ed. , Commission of the European Communities, Luxembourg (1976), EUR 5629e, p. 257. 5- Basic Physical Data for Neutron Dosi metry, (op. cit.), p. 243~ ~ 6- Basic Physical Data for Neutron D osimetry, (op. cit.), p. 249. 7. A. R. Smith, P. R. Almond, J. B. Smathers, V. A. Otte, F. H. Attix, R. B. Theus, P. Wootton, H. Bichsel, J. Eenmaa, D. Williams, D. K. Bewley, and C. J. Parnell, Medical Physics, 2, 195 (1975). 8 - Basic Physical Data for Neutron Dos imetry, (op. cit.), p. 267": 9. E. B. Wagner and G. S. Hurst, Health Physics 5, 20 (1961). - 10. Basic Physical Data for Neutron Dosim etry, (op. cit.), p. 2T9~ 127 REACTOR CORE DOSIMETRY STANDARDS Willem L. Zijp Netherlands Energy Research Foundation, ECN 1755-ZG Petten (NH) , The Netherlands Reactor neutron metrology serves to determine directly flux densities, fluences, spectra, and indirectly effects like burn-up, depletion and displacements. There are tendencies to require an accuracy of 2 to 5%. This gives requirements for the accuracy of nuclear data, of which the cross section data are most important. Average fission neutron cross sections for many reactions of interest are at present not accu- rate enough, owing to inadequacy of the spectral cross section data and to inadequacy of the knowledge of the fission neutron spectrum of 235 U above about 8 MeV. More experiments in benchmark fields, performed in interlaboratory experiments and in interna- tional collaboration are necessary to arrive at accuracies specified in reactor development programs. (Comparative evaluation; cross sections; fission spectra; integrals; neutrons; radioactivation; resonance integrals; spectral functions) Purpose of in-core reactor metrology Reactor neutron metrology performed inside re- search and power reactors provides numerical informa- tion on flux density values, fluence values and spec- trum data for many purposes. This information is of the primary type. Information which can be derived from these source data are e.g. burn-up values, dis- placed atoms, helium production, and transmutation rates. For providing these secondary data one often needs full information of the primary type, and a theoreti- cal model for the way of interaction under considera- tion. Since often the interaction of interest (e.g. displacements) is a spectrum dependent function, one then needs spectrum information. The interest in in-core metrology data can be categorized with respect to the materials. 1. Fuel material: Of importance is here the fission rate density (i.e. the number of fissions per unit volume) and its distribution axially (vertically) along the fuel rods (or plates), and radially (hori- zontally) within these rods (or plates), and its dis- tribution over a cluster of fuel rods (or a fuel as- sembly) , and from cluster to cluster (or assembly to assembly) . Flux density mapping over the fuel core is needed to: - check the assumed flux density pattern macroscopi- cally over the fuel region; - check the assumed flux density pattern microscopi- cally in a restricted region; - locate the positions with peak values of the flux density (detection of possible hot spots); - calculate the heat dissipation in specified fuel regions . Localized flux density measurements are required to provide information on local flux density values (e.g. for radionuclide production, in irradiation facilities, or any other accessible location). Apart from these interests, from the point of view of smooth and optimized reactor operation, there are the interests from the point of view of research and de- velopment of new types of fuels. Here measurements are desired to determine the changes in the characteristics of fuel samples (due to irra- diation under severe test conditions). 2. Fuel cladding material: Of main importance is here the determination of radiation damage effects in fuel cladding after long irradiations, i.e. after large in- cident neutron fluences. Results of irradiation of test samples are used to pre- dict the radiation damage occurring when the sample material is irradiated at an other location (often a longer irradiation in a smaller flux density) in a com- parable neutron spectrum. 3. Structural materials: This category comprises re- flector materials (such as graphite in a HTGR) and ma- terials for reactor tanks and reactor pressure vessels. Here one needs to know how long the materials can be positioned at a certain location in a reactor of given type, before deleterious radiation induced property changes prohibit further use. Nowadays much money and effort is spent to reactor pres- sure vessel surveillance programs, since the useful lifetime of such a vessel is an economic issue of first importance. The damage effects in graphite and steels are mainly induced by fast neutrons, i.e. neutrons with energies larger than say 10 keV. Since thermal nuclear reactors are being sold by many vendors to utility companies, there is a need for a good surveillance metrology and for national and inter- national standards for execution of such a dosimetry at an internationally accepted level. Accuracies required In practical applications of reactor neutron me- trology the neutron spectrum is not a goal in itself; it serves as a tool to calculate integral effects (re- action rates, burn-up, radiation damage effects) which in general are not directly measurable. The accuracies in neutron metrology are set by the ac- curacies for the integral quantities of final interest. The 1973 Consultants Meeting 1 mentioned that in special cases like fuel irradiations and graphite irradiations in high temperature gas cooled reactors, accuracies to 5% or better may be needed. The following remarks are based on the conclusions of the IAEA Consultants Meeting in 1976 . For most appli- cations at present the reached accuracies are in the range from 5 to 10%. In some cases, required accuracies have recently been re-evaluated to more stringent spe- cifications, partly also as consequence of improved understanding of the damage functions. These require- ments are reflected in target accuracies to be set for neutron field determination for the three categories of benchmark fields discussed further on. At the 1975 Petten Symposium these accuracy require- ments were stated 3 to be in the range from 2 to 5% for fast breeder reactor, and somewhat less stringent for light water reactors and controlled thermonuclear reac- tors . Present state-of-the art accuracies are estimated to be in the range of 2 to 30%. The 2 to 5% goal objective may be considered ambitious for some applications, it is nevertheless reasonable. 128 At least on the long term most reactor fuel and ma- terials development programs will not accept an uncer- tainty larger than 5%. In order to achieve such an ac- curacy routinely, however, it is necessary to work towards a better level of accuracy, namely 2 to 5%. Role of nuclear data Threshold reactions commonly used to determine fast neutron fluences are 58 Ni(n,p), 54 Fe(n,p), 1 * 6 Ti(n,p), 63 Cu(n,a). If the neutron spectrum is un- known, one cannot define effective cross sections to arrive at fluences of neutrons with energies above 0.1 or 1 MeV (or any other energy bound) . One can only de- fine an equivalent fission neutron fluence, by using a cross section averaged over the fission neutron spec- trum. The accuracy of the average cross section values de- termine directly the accuracy of the equivalent fis- sion neutron fluences. Activation detectors with cross section curves approximating a step function with the step at 1 or 0. 1 MeV are not readily available. The most interesting reaction in this respect is 93 Nb(n,n'), but there are still some practical problems with this reaction 4 . It has been recognized that for radiation damage studies reporting of equivalent fis- sion neutron fluences only is often not sufficient. More information is required, either in the form of spectrum information, or in the form of displacements per atom, calculated using an assumed spectrum and an accepted damage model. Neutron spectrum measurements using activation techni- ques and unfolding programs have been reviewed else- where 5 . For this purpose accurate energy dependent cross section data for a series of detectors must be available. Displacements can easily be calculated when so-called damage cross sections are available, derived from such an accepted displacement model . Damage cross sections At an IAEA Specialists Meeting on Irradiation Damage Units, held in Harwell, November 2-3, 1976 the present status of damage cross sections was discussed. Compa- rison has shown that two sets of damage energy cross sections for Fe, Cr and Ni (and hence steels) based on the UKNDF file and the ENDF/B-IV file agree within adequate accuracy, when applied to fission reactor spectra. The agreement for Ni and Fe is within a few per cent; the agreement for Cr is somewhat poorer, but the dis- crepancy is negligible in applications to stainless steels . Direct integral measurements of damage effects in gra- phite are possible with the so-called GAMIN detectors . These detectors consist essentially of a small gra- phite cylinder between two nickel activation detectors. By determining after irradiation the change in electri- cal resistivity in the graphite cylinder, and the in- cident neutron fluence on the nickel foil, one can de- rive an index $G/*Ni (=graphite damage f luence/nickel fluence). Using the GAMIN technique the damage-to- activation ratio has been determined experimentally in many research reactors in the Euratom Community. Available experience show that within roughly 10% these measured integral effects agree with calculated effects using damage cross sections and available spectrum in- formation. Thermal and intermediate energy regions Much attention has been paid in the past few years to develop neutron metrology methods related to fast re- actor neutron spectra. Apart from this development work there remain still pro- plems related to improvement of techniques for measure- ments of thermal and intermediate neutrons with capture reactions (self shielding effects; flux perturbation effects; resonance structure studies; gamma sensitivity of self powered neutron detectors; influence of cadmium or boron carbide covers of activation foils). Also in these studies the cross section data play an important role. Fast energy region The main uncertainties in nuclear data are in general related to the fast neutron cross section data. At the 1975 Petten Symposium Paulsen and Magurno con- cluded that the situation in the field of spectral cross section data for reactor radiation measurements is still unsatisfactory with respect to accuracy and completeness. As reasons for this unsatisfactory situa- tion the 1973 Consultants Meeting recognized the fol- lowing reasons: a. The lack of agreement for a limited set of reactions on which all measuring efforts should be concentrated; b. The failure to concentrate the differential measure- ments on the most sensitive energy region for dosi- metry purposes; c. The lack of a sufficient number of laboratories equipped with accelerators which can produce mono- energetic neutrons in the 6 to 12 MeV region, and which are used for neutron measurements. Since then much attention was paid to the development of a consistent set of cross section data and to the set-up of benchmark experiments. Reference set of cross sections The ENDF/B-IV dosimetry file > is now generally avail- able and has been adopted for international comparisons as a reference cross section set. All users are reques- ted to communicate their experience with this data file, so that future improvements aid to the aim of arriving at one generally accepted, internally consistent and extended dosimetry data file. The dosimetry reactions have been classified in two ca- tegories . Category I reactions are defined as reac- tions : a. for which the energy dependent cross sections are well known over their response ranges in standard neutron fields; b. for which calculated reaction rates in the standard neutron fields are consistent with the measured reaction rates. The following reactions belong to category I: 197 Au(n,y) 19S Au, 239 Pu(n.f), 237 Np(n,f), 238 U(n,f), 56 Fe(n,p) 56 Mn, 27 Al(n,a) 21+ Na, 63 Cu(n,2n) 62 Cu and 58 Ni(n,2n) 57 Ni. (Remark: for the (n,2n) reactions with very high threshold energies, accuracies of about 10% are presently acceptable). A number of other reactions are considered category I candidates: 59 Co(n,Y) 60 Co, 238 U(n,y) 239 U, 115 In(n,n') 115 Irfa, 58 Ni(n,p) 58 Co, 32 S(n,p) 32 P, 5 ' t Fe(n,p) 5 ' t Mn and 59 Co(n,a) 56 Mn. All other reactions used for neutron metrology are category II reactions. Decay data Apart from neutron cross section data, also other nu- clear data play an important role in neutron detection: half-lives, decay schemes, gamma abundances, fission product yields etc. 129 Role of benchmark fields The 1976 Consultants Meeting 2 identified three types of benchmark, neutron fields for reactor dosime- try: 1. Standard field: A permanent and reproducible neu- tron flux intensity, energy spectra and angular flux density distributions characterized to state-of-the art accuracy (Examples: thermal Maxwellian spectrum; epithermal 1 /E spectrum; Cf spontaneous fission neutron spectrum) . 2. Reference field: A permanent and reproducible neu- tron field, less well characterized than a standard field, but accepted as a measurement reference by a community of users. (Examples: U thermal fission neutron spectrum; the spectra in facilities like ZZ, ISNF, Big Ten, Tapiro, CFRMF; shielding benchmarks using a Fe-block or a Na-block) . 3. Controlled environment field: A neutron field, phy- sically well-defined and with some spectrum definition, employed for a restricted set of validation experiments. (Examples: HFIR; BSR; BR-2 Cd loops; HFR; reactor pres- sure vessel mock-ups; cores of ECEL, EBR-II) . In reactor neutron metrology benchmark fields serve three general objectives : - validation and/or calibration of experimental tech- niques; - validation and/or improvement of cross section data and other nuclear data for proper application of experimental techniques; - validation and/or improvement of analytical methods needed to extrapolate dosimetry data from a moni- toring or surveillance position to the location of interest. The highest priority is given to the second purpose, which includes more specifically : - establishment of high accuracy, consistent cross sec- tion data for basic integral detectors; - spectrum dependence of fission yields; - validation of resonance and threshold detectors for use in neutron fields where differential microscopic data may be inadequate; - activation detectors for which energy dependent cross section data are uncertain. Fission spectrum representation For the calculation of equivalent fission neutron fluences one needs the values for the cross sections of the activation (or fission) detectors used, and in particular the cross section averaged over a fission neutron spectrum. For the fission neutron spectrum of 235 U several ana- lytical representations have been used in the past years . In the following expressions, which have been norma- lized to a value of unity, E denotes the neutron ener- gy, expressed in MeV: - the formula proposed by Watt 10 Xj(E) = 0.48395 exp (-E) . sinh/2T - the formula proposed by Cranberg, Frye etal. 11 X 2 (E) = 0.45274 exp(-E/0. 965) . sinh/2. 29E" - the formula of Maxwellian type proposed by Leachmair X 3 (E) = 0.76985 exp(-E/l . 29) . /IT - the modified Watt-Cranberg formula proposed by Wood 3 X 4 (E) = 0.5827 exp(-0.992E).sinh(1.27^) The IAEA Consultants Meeting on prompt fission neutron spectra , held in 1971, concluded that a simple Maxwellian form does not satisfactorily fit all ob- served fission spectra. It was felt then that for the present a purely numeri- cal representation of experimental results woiild be best. Magurno and Ozer 7 tested the data on the ENDF/B-IV dosimetry file also by calculating spectrum averaged cross section using the Maxwellian spectrum function: X 5 (E) = 0.770./E'.exp(-E/T) using T = 1.29 MeV and T = 1.32 MeV. Recently Grundl and Eisenhauer 1 5 » 16 from the National Bureau of Standards made a new evaluation, based on 16 documented differential spectrometry measurements of the thermal neutron induced 35 U fission neutron spec- trum, and of the 252 Cf spontaneous fission neutron spec- trum. Their results can be described in three forms: - A reference Maxwellian representation, obtained from a weighted least squares fit in the energy range from 0.25 MeV to 8 MeV. For 235 U : M(E) = 0.7501 ./E\exp(-1 .50E/1 .97) . For 252 Cf: M(E) = .6672. v^.exp(-l .50E/2. 13) . - A seven-group spectrum of adjusted Maxwellian seg- ments, which fit the data over all energies. Estimated uncertainties are 1% to 4% for both spectra between 0.25 and 8 MeV and between 5 and 15% outside this energy range . - A continuous line segment correction to the reference Maxwellian, which establishes a final fit to the ex- perimental data: X(E) = u(E).M(E) Below 6 MeV the correction function u(E) is linear, above 6 MeV it is exponential. The correction func- tions for the two spectra are as follows: energy interval u(E) for 235 U u(E) for 252 Cf (in MeV) to 0.25 1+0.800E-0.153 1+1 .200E-0.237 0.25 to 0.8 1-0.140E+0.082 1-0.140E+0.098 0.8 to 1.5 1+0.040E-0.062 1+0. 024E-0. 0332 1.5 to 6.0 1+0.010E-0.027 1+0. 0006E+0. 0037 6.0 to °° 1 .043{exp-0.06x 1 .0 exp{-0.03x ' (E-6.0)/1.043} (E-0.60)/1.0} Similar representations have been tried for the spon- taneous fission neutron spectrum of Cf: - A formula proposed by Knitter et al , using a Max- wellian function with an average energy =2 . 1 3 MeV; - A more complicated function used by Green , based on a detailed evaporation model, and yielding an average energy =2.105 MeV. The choice of the representation of the fission spec- trum may not be so important for activation reactions with low thresholds, but it becomes important when re- actions with very high threshold (say about 10 MeV) are considered. As can be seen from table 1 the different representa- tions of the 235 U fission neutron spectrum give clearly different results for reactions with very high thres- hold energy. The Euratom Working Group on Reactor Dosi- metry noted that the fission neutron spectrum in the energy region above 8 MeV is only known with an accuracy of the order of 25%. For the application of reactions with high thresholds and for the prediction of helium production by (n,a) reactions the knowledge of the fission neutron spectrum should be improved. Quality of integral cross section data Integral experiments to determine a Q (the 2200 m/s value) , I (the resonance integral) and (the cross section averaged over the fission neutron spectrum) have been performed by experienced people in recognized la- boratories . 130 The values for o , I and <0f> which can be derived from the ENDF/B-IV dosimetry file have been compared in tables 2, 3, 4 and 5 with evaluated or experimental data in recent compilations. For the fission neutron spectra of 235 U and 252 Cf the NBS evaluations of Grundl and Eisenhauer 1 5 ' 1 6 were used. A rough estimation on the quality can be based on three aspects: the uncertainty in the measurement, and the accuracy (or bias) and the consistency of the eva- luated values. As a measure for the experimental un- certainty serves the quoted fractional error, v. As a measure of the accuracy serves the absolute value of the fractional difference A between (evaluated) expe- rimental error and the calculated value from the cross section file. As a measure of the consistency between (evaluated) experimental value and the calculated va- lue serves the ratio of the fractional difference and its stated fractional error v. The following indications are used in the tables: category uncertainty accuracy consistency ++ < v < 2% < A < 2% < A/v < 1 + 2% < v < 4% 2% < A < 4% 1 < A/v < 2 4% < v < 6% 4% < A < 6% 2 < A/v < 3 - 6% < v < 8% 6% < A < 8% 3 < A/v < 4 — 10% < v 10% < A 5 < A/v The data for the fission neutron spectra are taken from recent reviews by Fabry et al ' . The normalization adopted involves a so-called flux transfer, using the 239 Pu(n,f) reaction and the NBS 252 Cf source. This californium source was chosen be- cause of its availability and its well known source strength (error 1.1%). The 239 Pu(n,f) reaction was chosen because of its relatively flat shape in the energy range of interest and its well known cross sec- tion. It has been concluded by Fabry, McElroy et al that integral cross section data for dosimetry reactions as measured in standard and reference benchmark neutron fields depart from computed ones, not only because of differential energy cross section inadequacies, but also because the spectral shapes characteristic of these benchmarks are usually inaccurate in the energy ranges not covered or poorly covered by differential neutron spectrum techniques, e.g.: - below =250 keV and above ~\0 MeV for the fission neutron spectra of 235 U and 252 Cf; - below -10 keV and above =2 MeV for 11, CFRMF , BIG- TEN. The same authors conclude that even in the well covered energy ranges, the reliability remains questionable, as is presently the case for the 235 U fission neutron spectrum between 3 and 6 MeV, and for LFRMF between 100 and 400 keV. The only benchmark whose spectral shape appears to be accurately established between =0.25 and =10 MeV is the 252 Cf neutron spectrum. The inconsistencies ob- served for some facilities like LFRMF and BIG- 10 are mostly attributed to inaccurate spectral computations resulting from the inadequate 235 U fission spectrum, and the inelastic scattering cross section data in ENDF/B-IV. This effect is less pronounced for 11, be- cause the spectral characterization from =10 keV up to =2 MeV mostly relies on differential spectrometry meas- urements and not on computations . The 1976 Consultants Meeting recommended that efforts should be made to remove inconsistencies between in- tegral measurements and differential evaluations at least as concerns the 235 U fission spectrum, the £X type facilities and the ISNF, and the cross sections for 58 Ni(n,p), 235 U(n,f), 59 Co(n,Y), 115 In(n,n'), 5L, Fe(n,p), * 03 Rh(n,n' ) , so as to qualify them as stand- ard spectra and category I reactions respectively. Improvement of consistencies The consistency between measured and calculated reaction rates may be influenced by various effects or procedures : - The group structure chosen for presentation of the final spectrum; - The discontinuity or extrapolation at the lower and upper energy bounds of the spectrum; - The group structure of the cross section libraries, and especially the detailed structure in the reson- nance region (e.g. important for (n,y) reactions); - The adjustments sometimes made to a spectrum from reactor physics calculations to obtain a better fit for experimental data; - The accuracy and precision of the experimentally de- termined reaction rates; - The perturbation of the neutron field by the presence of one or more activation and fission detectors, or their encapsulations; - The uncertainty in the self shielding effect in the activation and fission detector applied. The present situation of the cross sections for reactor radiation measurements can be improved by a series of actions : - Application of recommended evaluated cross section libraries, both in reactor physics calculations and in spectrum unfolding; - Application of a reference group structure and appli- cation of recommended procedures for arriving at other group structures or a series of point values; - Recalculation of neutron spectra using well esta- blished reactor physics computer programs under well defined and improved conditions; - Selection of a few well known spectra (serving as benchmark spectra) with different shapes; - Careful definition of the neutron spectra in the benchmark facilities; - Accurate determination of experimental activities, using enlarged series of detectors; - Adoption of agreed procedures for adjustment and ex- trapolation of neutron spectra based on reactor phy- sics calculations; - Intercalibration of counting equipment, based on dis- tribution of calibrated radionuclide samples; - Adoption of agreed nuclear data (half-lives, decay schemes, gamma abundances, fission product yields, etc.) . It will be clear that several parallel actions are ne- cessary to arrive at the desired accuracies within a few years . Current international efforts The actions mentioned above are being discussed in several international meetings (the IAEA Consultants Meetings in 1973 and 1976; the regular meetings of the Euratom Working Group on Reactor Dosimetry; the annual meetings of the IAEA Working Group on Reactor Radiation Measurements; the IAEA Specialists Meeting on Radiation Damage Units in 1976; the ASTM-Euratom symposium on Reactor Dosimetry, held in 1975 at Petten, and the next one scheduled for October 1977 in the USA). Owing to intensive international contacts the following progress results can be mentioned: - the acceptance and the availability of the ENDF/B-IV dosimetry file as a reference data set for reporting and comparing obtained results; - the acceptance of a list of dosimetry reactions in categories I and II; - the acceptance of the IAEA program on Benchmark Neu- tron Fields Applications for Reactor Dosimetry; - the recommendation for international interlaboratory studies like the Interlaboratory Reaction Rate Pror 131 gram in the USA; the definition of a few selected standard neutron fields and revision of the list of category I reac- tions ; the acceptance of a few radiation damage models for the calculation of the number of displaced atoms; the recommendation by the Euratom Working Group to apply the BURLIB 100 groups structure (very similar to the DLC 100 groups structure) as reference both for shielding and reactor metrology work; the exchange of technical information at the open biannual ASTM-Euratom symposia on Reactor Dosimetry; the quick exchange of items of interest through the Euratom Newsletter on Reactor Radiation Metrology; the intercomparison of some promising neutron spec- trum unfolding codes; the IAEA intercomparison of computer programs for analyzing Ge(Li) gamma ray spectra; the international intercomparison of gamma emission rates of Eu sources. Conclusions 1 . Different reactor development programs require ac- curacies in integral data (fission rates, burn-up, fluences, damage effects) of 2 to 5%. 2. An internationally accepted starting point for data treatment in neutron metrology work is the ENDF/B IV dosimetry file, which has been made available in a world-wide scale through the four nuclear data centres (Brookhaven, Saclay, Vienna, Obninsk). 3. Appreciable discrepancies exist between measured and calculated average cross section values in the U fission neutron spectrum. The representation of the high energy tail (above x8 MeV) of this spectrum needs further study. 4. Dosimetry reactions in categories I and II, and classes of benchmark neutron fields have recently been reviewed by the IAEA 1976 Consultants Meeting. The ac- cepted tables constitute the basis for a critical ana- lysis of integral measurement data which is or becomes available. The benchmark field approach can serve to detect dis- crepancies and inconsistencies, to arrive at a better quality of cross section data, and to improve spectrum determinations by the unfolding technique. Many integral experiments in several benchmark fields, based on interlaboratory and international cooperation are needed to arrive at the requested accuracies of 2 to 5%. References 1. VLASOV, M. ; DUNFORD, C. : "Proceedings of a Consul- tants' meeting on nuclear data for reactor neutron dosimetry", held in Vienna, 10-12 September 1973. INDC(NDS)-56/U (IAEA, Vienna, 1974). 2. VLASOV, M. : "IAEA Consultants 'Meeting on integral cross section measurements in standard neutron fields", held in Vienna 15-19 November 1976. Summary report, conclusions and recommendations, Report INDC(NDS)-81/L+D0S (IAEA, Vienna, 1977). 3. McELROY, W.N.; BENNETT, R.A. ; JOHNSON, D.L.; DUDEY, N.D. : "Neutron environmental characterization re- quirements for reactor fuels and materials develop- ment and surveillance programs" Proc. 1st ASTM-Euratom Symposium on Reactor Dosi- metry, Petten, 22-26 September 1975. HEDL-SA-91 1 . 4. LLORET, R. : "Application de la reaction 93 Nb(n,n') a la dosimetrie des irradiations de materiaux" Proc. 1st ASIM-Euratom Symposium on Reactor Dosime- try, Petten, 22-26 September 1975. 5. ZIJP, W.L.: "Review of activation methods for the determination of neutron flux spectra" Report RCN-241 (Netherlands Energy Research Found., Petten, 1976). Also proc. 1st ASTM-Euratom Symposium on Reactor Dosimetry, Petten, September 22-26 1975. 6. PAULSEN, A.; MAGURNO, B.A. : "Differential neutron data for reactor dosimetry" Proc. 1st ASTM-Euratom Symposium on Reactor Dosime- try, Petten, 22-26 September 1975. 7. MAGURNO, B.A. ; OZER, 0.: "ENDF/B-IV file for dosi- metry applications" Nuclear Technology 25 (1975), 376. 8. MAGURNO, B.A. : "ENDF/B-IV dosimetry file" BNL-NCS-50446 (April 1975). 9. GRUNDL, J.; EISENHAUER, C. : "Benchmark neutron fields for reactor dosimetry" Paper presented at IAEA Consultants 'Meeting on Integral Cross Section Measurements in Standard Neu- tron Fields for Reactor Dosimetry, Vienna, November 15-19, 1976. 10. WATT, B.E.: "Energy spectrum of neutrons from ther- mal fissions of 23 ^U" Phys. Rev. 87 (1952), 1037. 11. CRANBERG, L. ; FRYE , G.; NERESON, N. ; ROSEN, L. : "Fission neutron spectrum of 235 u" Phys. Rev. J03 (1956), 662. 12. LEACHMAN, R.B.: "Determination of fission quantities of importance to reactors" Proc. Int. Conf. on Peaceful Uses of Atomic Energy, Geneva, 1955, Vol. 2 (1956), 193. 13. WOOD, J.: "The fission neutron spectrum of 235 U" J. Nuclear Energy 27_ (1973), 591-595. 14. "Prompt fission neutron spectra" Proceeding of a Consultants 'Meeting on prompt fis- sion neutron spectra (IAEA, Vienna, 1972). 15. GRUNDL, J. A.; EISENHAUER, CM.: "Fission spectrum neutrons for cross section validation and neutron flux transfer" Conf. on Nuclear Cross Sections and Technology, Washington D.C. (March 1975). 16. GRUNDL, J. A.; EISENHAUER, CM.: "Fission rate meas- urements for materials neutron dosimetry in reactor environments" Proc. 1st ASTM-Euratom Symposium on Reactor Dosime- try, Petten, 22-26 September 1975. 17. KNITTER, H.H. ; PAULSEN, A.; LISKIEN, H. ; ISLAM, M.M. : "Measurements of the neutron energy spectrum of the spontaneous fission of 252 Cf" Atomkernenergie 22 (1973), 84. 18. GREEN, L.; MITCHELL, J. A.; STEEN, N.M. : "The cali- fornium 252 fission neutron spectrum from 0.5 to 13 MeV" Nucl. Sci. Eng. 50 (1973), 257. 19. FABRY, A.; CEULEMANS , H. ; VAN DE PLAS , P.; McELROY, W.N.; LIPPINCOTT, E.P.: "Reactor dosimetry integral reaction rate data in LMFBR benchmark and standard neutron fields: status, accuracy and implications" Paper presented at 1st ASTM-Euratom Symposium on Reactor Dosimetry, Petten, September 22-26, 1975. 20. FABRY, A.; McELROY, W.N. ; KELLOGG, L.S.; LIPPINCOTT, E.P.; GRUNDL, J. A.; GILLIAM, D.M.; HANSEN, G.E.: "Review of microscopic integral cross section data in fundamental reactor dosimetry benchmark neutron fields" 132 Paper presented at IAEA Consultants 'Meeting on Integral Cross Section Measurements in Standard Neutron Fields, Vienna, November 15-19, 1976. 21. SHER, R. : "2200 m/s neutron activation cross sec- tions" Contribution to "Handbook on Nuclear Activation Cross Sections" Technical Reports Series No 156 (IAEA, Vienna, 1974) 22. MUGHABGHAB, S.F. ; GARBER, D.I. J "Neutron cross sections, Vol. I, resonance parameters" BNL-325, Third edition, Vol. I (NNCS-BNL, June, 1973). 23. BIGHAM, C.B.: "The slow-neutron fission cross sec- tions of the common fissile nuclides (revised 1975)" Nucl. Sci. Eng. 59 (1976), 50-52. 24. ZIJP, W.L.; NOLTHENIUS, H.J.: "Comparison of inte- gral cross section values of several cross section libraries in the SAND-II format" Report ECN-2 (Netherlands Energy Research Founda- tion, Petten, 1976). 25. ALBINSSON, H. : "Infinite-dilution resonance inte- grals" Contribution to "Handbook, on Nuclear Activation Cross Sections" Technical Reports Series No 156 (IAEA, Vienna, 1974). 26. SMITH, D.L.: "Evaluation of the x x 5 In(n,n' ) * l 5 In m reaction for the ENDF/B-IV dosimetry file" Report ANL/NDM-26 (December 1976). 27. PHILIS, C.j BERSILLON, 0.; SMITH, D. ; SMITH, A.: "Evaluated (n,p) cross sections of Ti, Ti and Report ANL/NDM-27 (January 1977). 28. CANCE, M. ; GENTHON, J.P. ; MICAUD, G.; SALON, L. : "Le detecteur neutronique au graphite G. A. M.I.N. Etudes et applications a la dosimetrie des dom- mages radio-induits" Report CEA-N-1823 (Saclay, January 1975). 29. GENTHON, J.P.; HASENCLEVER, B.W.; SCHNEIDER, W. ; MAS, P.; WRIGHT, S.B.; ZIJP, W.L.: "Recommenda- tions on the measurement of irradiation received by the structural materials of reactors" EUR-5274 e,n,d,f (C.E.C., Luxembourg, 1975). 133 Table 1: Influence of representation of fission neutron spectrum of 235 U on average cross sections Based on cross sections from the ENDF/B-IV dosimetry file, and data reported by Fabry et al |l9J,|20|. Cross section values are given in fm 2 (=10- 30 m ) . reaction effective threshold energy (in MeV) integral measurement a m (in fm 2 ) ratio o c /am Maxwellian =1 .97 MeV NBS-eval. =1 .98 MeV WAtt =2.00 MeV 237 Np(n,f)FP 58 Ni(n,p) 58 Co 54 Fe(n,p) 51+ Mn 27 Al(n,p) 27 Mg 56 Fe(n,p) 56 Mn 59 Co(n,a) 56 Mn 27 Al(n,a) 2lt Na 127 I(n,2n) 126 I 55 Mn(n,2n) 51 *Mn 58 Ni(n,2n) 57 Ni 0, 2, 3, 4, 6, 6, 7, 10, 1 1 13, 131.2 10.85 7.97 0.386 0.1035 0.0143 0.0705 0.105 0.0244 5.77xl0 -4 1.006 0.926 0.965 1.078 1.081 1.140 1.106 1.499 1.426 0.776 1.006 0.936 0.975 1.067 1.017 1.035 0.983 1 .130 1.004 0.489 1.019 0.947 0.984 1.062 1.000 1.021 0.970 1.094 0.951 0.440 Table 2: Comparison of 2200 m/s cross section values Cross section values are expressed in units of 100 fm (in 10 m) reaction compilation value, °m uncer- tainty calculated value, a c ENDF/B-III |24| °~c/°m accu- racy consis- tency IAEA-TR-156 I 21 1 BNL-325 |22| BIGHAM [-23 1 (in %) 6 Li(n,ct) 3 H 940±4 0.4 ++ 923 0.982 ++ 10 B(n,a) 7 Li 3837±9 0.2 ++ 3770 0.983 ++ — 23 Na(n, Y ) 2It Na 0.528±0.005 0.530±0.005 0.9 ++ 0.524 0.989 ++ + 1+5 Sc(n, Y ) lt6 Sc 25±2 26.5 ±1.0 3.7 + 25.9 0.977 + ++ 58 Fe(n, Y ) 59 Fe 1.14 ±0.05 1.15±0.02 1.7 ++ 1.16 1.009 ++ ++ 59 Co(n, Y ) 60 Co 37.5 ±0.2 37.2 ±0.2 0.5 ++ 36.6 0.984 ++ - 63 Cu(n,Y) 64 Cu 4.4 ±0.2 4.5 ±0.1 2.2 + 4.42 0.982 + ++ 115 In(n,Y) 116 In m 161 ±5 157 ±15 9.6 -- 164 1.045 ++ 197 Au(n, Y ) 198 Au 98.8 ±0.3 98.8 ±0.3 98.7 0.2 0.2 ++ 97.1 0.983 ++ — 232 Th(n, Y ) 233 *h 7.4 ±0.1 7.40±0.08 1.1 ++ 7.26 0.981 ++ + 235 U(n,f)FP 580 ±2 582.2+1 .3 576.9 3.4 0.6 ++ 573 0.993 ++ + 237 Np(n,f)FP 0.019±0.003 15.8 — 0.0163 0.857 — ++ 238 U(n, Y ) 239 U 2.720±0.025 2.70 ±0.02 0.7 ++ 2.65 0.981 ++ 239 Pu(n,f)FP 742 ±3 742.5±3.0 742.8 4.4 0.6 ++ 730 0.983 ++ ' Table 3: Comparison of cross sections, averaged over a 1/E neutron spectrum Values of resonance integrals refer to a cadmium cut-off equal to 0.5 eV, and are expressed in units of 100 fm (10 -28 m 2 ). reaction compilation value, a m uncer taint y calculated value, a c ENDF/B-IV |8| ,|22 ac/°m >> CJ cd u 3 O O CO 1 CO p> ■H CJ m a 3 = 1 .98 MeV. c) Based on a recent evaluation by C. Philis et al . | 27 | . d) Cross section data not present in ENDF/B-IV dosimetry file; listed value has been taken from SAND-II cross section file. 135 Table 5: Comparison of cross sections, averaged over the fission neutron spectrum of 252 Cf Cross section values are expressed in units of fm (=10"" iu m ) . Calculated values refer to the ENDF/B-IV dosimetry file and the NBS spectrum evaluation. Table is based on data reported by Fabry et al > |l9|,|20| reaction effective threshold integral measurement uncertainty calculated value a c (in fm 2 ) °c /a m accu- consis- (in %) (in MeV) a ) o m (in fm 2 ) racy tency 115 In(n,Y) 116 In ln 12.53 3.4 + 13.03 1.040 + + 197 Au(n, Y ) 198 Au 7.99 3.6 + 7.99 1.000 ++ ++ 235 U(n,f)FP 120.3 2.5 + 124.1 1.033 + + 239 Pu(n,f)FP 180.4 2.5 + 178.9 0.992 ++ ++ 237 Np(n,f)FP 0.6 133.2 2.8 + 135.1 1.014 ++ ++ 103 Rh(n,n') 103 Rhm 0.8 75.7 7.0 - 17.55 b) 0.886 b) 115 In(n,n') 115 Inm 1.2 19.8 2.5 + — — 238 U(n,f)FP 1.5 32.0 2.8 + 31.54 0.986 ++ ++ ' t7 Ti(n,p) 1+7 Sc 2.2 1.89 2.1 + t 2.422 ' 1.261 . 1.281° ; — — 58 Ni(n,p) 58 Co 2.8 1 1.8 2.5 + 11.50 0.975 + ++ 5l *Fe(n,p) 5lt Mn 3.1 8.46 2.4 + 8.91 1.053 ^TKn.p^Sc 3.9 1.38 2.2 + | '• 252 c> i 1.381 ' 0.907 , 1.001 CJ ++ ++ 27 Al(n,p) 27 Mg 4.4 0.51 9.8 — 0.514 1.008 ++ ++ 56 Fe(n,p) 56 Mn 6.0 0.145 2.4 + 0.1475 1.017 + + ++ 27 Al(n,a) 21+ Na 7.2 0.1006 2.2 + 0.1059 1.053 - lt8 Ti(n,p) 48 Sc 7.6 0.042 2.4 + r 0.265 . ' 0.0446 c; 0.630 . 1.062° ; — o 55 Mn(n,2n) 51+ Mn 11.6 0.058 10.3 — 0.0528 0.910 - ++ 59 Co(n,2n) 58 Co 0.057 10.5 — 0.0379 0.665 — - 63 Cu(n,2n) 62 Cu 12.4 0.030 10.0 — 0.02415 0.715 — — a) Threshold values are valid for a 235 U fission neutron spectrum. b) Based on a recent evaluation by D.L. Smith ] 26 | . c) Based on a recent evaluation by C. Philis et al | 27 | . 136 STANDARDS FOR DOSIMETRY BEYOND THE CORE Frank J. Rahn , Karl E. Stahlkopf and T> II. Marston Electric Power Research Institute Palo Alto, CA 94304 and Raymond Gold and James H. Roberts* Hanford Engineering Development Laboratory Richland, WA 99352 Introduction The need for well understood, standardized neutron dosimetry techniques is increasingly evident for several applications in thermal and fast reactors beyond the core region. This need for dosimetry comes mainly from separate but related interests, namely: pressure vessel surveillance, materials testing, design and shielding requirements. The needs of the controlled thenno-nuclear program are addressed in another session of this symposium. Objectives of Dosimetry The problem of dosimetry for use in reactors (damage studies, irradiation experiments, shield assessment, reactor studies) is a very critical one that may influence the development of nuclear power. The self- consistency of calculated and dosimetr ically derived flux and spectrum is of prime importance to the development of validated techniques for design, testing and licensing purposes. Dosimetry measurements serve the following objectives: 1) Surveillance of the safety and integrity of the pressure vessel of a nuclear steam supply (NSS) system. 2) Knowledge of the neutron fluence and spectrum for correlation with changes in properties in a materials test program, 3) Validation and/or calibration of experimental techniques, 4) Verification of analytical methods for predicting fluxes far from the core (in terms of a neutron's mean free path) for shielding purposes, sensitivity analyses, channel theory, etc. There are many additional uses of dosimetry results. Among these are: 1) choosing between two or more data sets, 2) identifying discrepancies and 3) improving (i.e., adjusting) cross sections. These latter uses are generally controversial, and split the engineering and physics community. One hopes to identify discrepancies between integral and differential data and, if necessary, make adjustments in a self-consistent way. It is of prime importance to obtain the sensitivity of the integral results to uncertainties in the nuclear data. To reach these objectives it is important that the cross section data for dosimetry reactions be integrally consistent and well-known. Benchmark neutron fields are now classified according to guidelines first Symposium (1975): established at the Petten Standard: a permanent and reproducible neutron field with neutron flux intensity, energy spectra and angular flux distributions characterized to state-of-the-art accuracy. The main characterizations must be verified by interlaboratory measurements and calculations. Reference: a permanent and reproducible neutron field, less well-characterized than a standard, but accepted as a measurement reference by a community of users. Controlled Environment: a neutron field, physically well-defined and with some spectrum definition, employed for a restricted set of validation experiments. Dosimetry and Benchmark Experiments A conclusion of the recent IAEA Consultants' Meeting of Integral Cross Section Measurements in Standard Neutron Fields is that spectrum averaged cross section data as measured in standard and reference benchmark neutron fields differ from computed ones, not only because of absolute flux and cross section inadequancies , but also because of uncertainties in observed spectra in most of these benchmarks. Uncertainties in spectral measurements stem from limitations in the energy range coverage in differential neutron spectra techniques. In controlled environments, a combination of calculations, neutron differential spectrometry and integral measurements will be required for the determination of the spectra. The development of improved unfolding codes to handle the results of these techniques is desired. The success of such techniques, however, will depend on the availability of data and analytical methods for handling errors and their correlation, i.e., not only cross section data files but also for spectrometry and reaction rate data. The total flux and spectral shape uncertainties for the important energy regions are currently estimated in the +5 to 15% (l 5 MeV. The state-of-the-art determination of spectrum requires the combined use of calculations, good unfolding techniques, and data from differential spectrometry and integral measurements. *0n leave from Macalester College, St. Paul, MN 55105. 137 Dosimetry Requirements for Design and Shielding Dosimetry requirements for reactor applications can be quite stringent. The nature of the problems can be quite diverse and complex and is often related to specific designs. Situations can arise in which the uncertainties in calculations can lead to unacceptable constraints. If in an LWR, for example, the diameter of the pressure vessel is increased to mitigate radiation embrittlement of the steel, then the effects of dosimetry uncertainties can have a rather dramatic effect on the cost of the plant. If, in an existing plant, the dosimetry has unduly large uncertainties, then licensing and safety considerations may result in premature plant life and/or operating constraints. Many other examples exist. For instance, possible radiation effects on support plates and grids, fuel assemblies, and other core internals, especially in the design of fast reactors, can impact on the size of the pressure vessel required, pressure drop across the core, and in-vessel configuration. Good dosimetry can also lead to the optimum location of low and intermediate power flux monitors and instrumentation. There is presently a fairly disparate range of opinions as to the accuracies required for the dosimetry measurements. In part, this is due to the lack of consensus on the quantification of the design/experimental criteria and the diversity of the problems encountered. In addition to knowledge of the magnitude of the flux in some region outside the core, spectral information is usually required. This is true, for instance, of activation rates in control rod, drive mechanisms or for pressure vessel irradiation where both the flux and the spectrum (as related to either the activation cross section or some pseudo-damage cross sections) are required. Some typical target accuracies {2a) required in power and experimental reactors are: 1) Radiation Heating +20% a) thermal shields b) support plate c) control rods 2) Displacement Rates +15% a) reflector region b) support plate c) internal shields 3) Instrument & Control Tubes +15% 4) Activation +30% a) structural members b) coolant and coolant circuit c) components 5) Radiation Streaming +30% Dosimetry Requirements in Operating Reactors The reference experimental measurements planned on operating reactors generally fall into the category of controlled enviro-mnent fields. An integral part of a dosimetry program is the comparison of the data with results from computer codes so as to produce a validated analytic method of estimating the flux and spectrum for streaming, damage, or activation purposes. There are also a series of computational benchmarks available which are intended to check the methodology, data and adequacy of the models used to calculate quantities of interest. The purpose of reactor dosimetry programs outside the core is to obtain high accuracy and precision quantities which can be related to physical quantities of interest and which can be used to validate the 1 2 calculational methodology. This mandates that the neutron flux and spectrum be measured as the intermediary between the radiation exiting the core and the various quantities to be determined. One of the greatest needs for good dosimetry is for use in correlating material effects occurring in the pressure vessel of LWR's under irradiation. The requirements for dosimetric measurements on operating and experimental reactors are: (a) A relatively complete flux and spectrum mapping starting at the edge of the core through the outside edge of the pressure vessel. This should include detectors in the coolant gap radially outward from the core, and in a sufficient number of axial and azimuthal positions within the coolant region in order to assure complete knowledge of the neutron field at the pressure vessel. In addition, detectors should be positioned on both sides of the pressure vessel to measure the transmission and the change in spectrum in the steel. Such complete knowledge is required to validate the computational methods and to establish a self-consistent set of experimental points, (b) The principal energy range of interest for dosimetry purposes is the range 0.1 < E < 10 MeV. It is the neutrons in this energy range which are the greatest contributors to radiation embrittlement in steels. However, information on the spectra in the energy region 1 <_ E <^ 100 keV is important for shielding problems due to the high transmission of neutrons through steel and sodium for these energies. The most important location for spectral measurements in this energy range is at the exit surface of the pressure vessel. At this point the usual calculational approach is to interface the 2D finite difference transport calculations used inside the pressure vessel with the 3D Monte Carlo and albedo scattering techniques used in the reactor cavity region. Flux and spectral measurements, then, in the 1 to 100 keV range are necessary to verify the deep penetration predictions and establish the source for the Monte Carlo simulations. Thermal flux measurements are usually required for activation considerations, and for corrections of the various detector responses to the fast flux. Using an LWR as an example for fast neutrons, the water gap is approximately 7 mean free paths (mfp) wide. For neutrons much below 1 MeV, the concept of spatial distances in terms of mean free paths is somewhat misleading, because the neutron density of these energies is due to high energy neutrons, undergoing collision and rapidly thermalizing . The pressure vessel is about 1.5 mfp wide for BWR's and 2.0 mfp wide for PWR's. At the inside edge of the pressure vessel, the expected flux (< 0.1 Mev) is approximately 3.6 x 10 n/cm /sec for a BWR and 1.4 x 10 " n/cm 2 /sec for a PWR, for plants producing 3400 MWth and 3600 MWth, respectively. (c) Practical considerations which determine the proper balance between desired accuracy and what is actually obtainable on commercial operating reactors. On such plants, operational considerations, accessibility and instrument locatability require very precise planning to 138 Table 1 Energy Range thermal 0.414 eV - 1.0 keV 1.0 keV - 100 keV 0.1 MeV - 1.0 MeV 1.0 MeV - 4.5 MeV 4.5 MeV - 10 MeV Standard** Radiation Field >4% (4%) >10% (10%) Accuracy Required* (210% (6%) >10% (10%) >10% (>20%) >20% (30%) ~* >10% (20%) >6% (>20%) >6% (20%) 1 >6% (20%) >10% (>20%) >10% (30%) *the first number is what is ultimately needed, the second is what is believed to be currently achievable with state-of-the-art techniques for measuring integral flux over the energy interval. **for standard and reference fields, the accuracy is specified for LWR, LFMBR and CTR requirements, reflecting the overall requirements for all of these systems. Table 2 COMPARISON OF SELECTED DIFFERENTIAL REACTOR NEUTRON SPECTROSCOPY METHODS Method (n,p) Emulsions ° Collimated Source Non-Col lima ted Source (n,p) Proportional Counters 6 Li(n,t) 4 He Time-of-Flight (TOF) ° H (d,n) H e Source LINAC Source 3x10 3x10 1 x 10" 1 x 10" 5 xlO* 5 x 10" 20 10 2.5 6.5 0.2 Resolution Fair Accuracy % (la) Good Fair Good 5-10% 0.3 2 MeV, the direction of the proton velocity can be determined with close to 100% certainty. This confidence factor decreases as proton tracks get shorter . For neutrons of energy 250 keV, nuclear emulsions loaded with glass specks and containing Li can provide some knowledge of the anisotropy in neutron flux. For example, in the Li(n,rt) t reaction the sum of the ranges of the alpha particle and triton is significantly longer if the triton goes in a forward instead of a backward cone. Thus, there will be a strong correlation between the direction of the long tracks and the neutron direction. The Li(n,oO t cross section resonance at 250 keV significantly increases the number of triton-alpha pairs and thereby provides improved statistical accuracies of angular observations in the vicinity of the resonance. In hazarding one final projection for possible future directions of differential reactor neutron spectroscopy, one cannot ignore the virtual explosion of activity and developments which has occurred in the field of Solid State Track Recorders (SSTR) over the 27 last decade or so, In particular, the ability to extract relevant physical data such as mass, charge and energy from the shape of tracks formed in SSTR is a striking advance. This recent development, which is commonly referred to as track profile analysis, can only be considered to be in the very formative stages. This ability coupled with the vast improvements of c* -particle sensitive SSTR, such as Cellulose Acetate Buterate (CAB) and Cellulose Nitrate (CN), augurs for the evolution of SSTR differential neutron spectroscopy methods. Advanced SSTR techniques for observing angular flux anisotropy should also become possible. Such methods would, of course, be direct descendants of emulsion techniques and thereby automatically possess many advantages. However, in contrast with emulsions, the insensitivity of SSTR to electrons would provide an enormous advantage for reactor applications. Moreover, the reduced (^--particle range also implies improved high energy sensitivity with less attendant stress on the need for finite-size corrections. The Importance of Good Dosimetry - A Specific Example To put the value of good dosimetry into perspective, let us consider a specific example, relating to the materials test program and radiation embrittlement in LWR's. There are presently 53 commercial nuclear reactors operating in the United States which account for 10% of our nation's electricity requirements. Commercial nuclear power has been generating electricity in this country since 1959 with a safety record unmatched by any other heavy industry. Since the commissioning of Dresden I, the first commercial nuclear power plant, no fatalities and no substantial personnel contamination have resulted from the operation of the nuclear steam supply systems for electrical generation. One of the primary reasons for this excellent operating record is the very conservative design utilized in the building and operation of these plants . 141 A conservative design is necessary to account for unknowns in materials properties, operating stresses, and operating environments. As these factors are better defined through development of refined analytical techniques and measurements made on operating systems, it is possible to reduce overconservat ism and still maintain safety, reliability, and availability. Overly conservative design or regulatory action can result in substantial losses to the economy. If we assume a 7 mill/kWh difference in base loaded coal and nuclear electricity generation rates, and further assume that all electricity from nuclear outages is replaced by base loaded coal generation capacity, each percent of lost nuclear capacity represents a $16.6 million increase in cost of electricity per year. If each lost percent in nuclear capacity were replaced by oil the cost to the economy would be about $70 million per year. The preceding remarks were presented to emphasize the economic consequences of overconservative definition of design and operating margins. The nuclear industry is presently faced with potential premature retirement or long forced outage for thermal anneal of a number of PWR plants due to irradiation embrittlement of reactor pressure vessel steels. The remainder of this paper will discuss the present regulatory and code requirements for irradiated materials properties and suggest research necessary to accurately define the proper irradiated materials properties requirements. The Regulatory Position on Radiation Embrittlement The prevention of brittle fracture of the pressure boundary of nuclear systems is assured by compliance with toughness requirements set forth in 10 CFR Part 50, Appendices G and H. Recently the Nuclear Regulatory Commission has issued Regulatory Guide 1.99, which describes procedures for predicting the effect of copper and phosphorous on transition temperature shift and upper shelf Charpy energy decrease of irradiated pressure vessel materials. The minimum upper shelf energy during the life of the plant is set at 50 ft-lb to insure sufficient toughness to prevent low energy ductile tearing. If the upper shelf level falls below the 50 ft-lb limit, the following actions may be required for continued licensability: 1. Extensive testing and analysis of existing surveillance specimens, where available. 2. Generation of irradiated fracture toughness data on relevant steels. 3. Extensive fracture mechanics analysis of the reactor vessel . 4. Extended outage to comply with nondestructive examination requirements. 5. In extreme cases, an in situ anneal with consequent prolonged outage or premature plant retirement could result. It should be noted that such consequences could also result from inadequate data, even though the vessel may in reality be within safe limits. When credible surveillance data from a reactor are not available, prediction of radiation damage must be done in accordance with procedures given in Regulatory Guide 1.99. Unfortunately, many installations do not have adequate surveillance data and thus must rely on Regulatory Guide 1.99 predictive methodology. The reasons for the lack of proper data include: 1. The materials incorporated in some vessel surveillance programs were selected on the basis of unirradiated materials properties and not on sensitivity to radiation embrittlement. This situation occurred because the deleterious effects of copper and phosphorous were not known at the time of the materials selection. Thus, these reactors have not included those radiation- sensitive materials, which will be controlling, in their surveillance programs 2. The surveillance capsules in some reactors were removed because of fretting fatigue damage resulting from flow-induced vibration. These plants presently have no active surveillance dosimetry programs. Predictive Methodology of Regulatory Guide-1.99 The curves for prediction of transition temperature shift and shift in Charpy shelf energy are shown in Figure 1. These curves are based in large part, upon the data compiled by Bush for all reported reactor surveillance capsules tested through 1973. Serious questions have been raised about the appropriateness of the utilization of these curves for prediction of irradiated properties. Specific objections include: 1. The data base is incomplete and sometimes inapplicable. The inclusion of low exposure temperature Yankee Rowe surveillance data (irradiated at 450-525°F) results in prediction of greater than normal embrittlement. The trend curves should be developed, rather than simply available data. statistically bounding the 3. Weld metal, base metal, and heat affected zone are unique materials (with different responses to irradiation), but the Guide requires that a single curve be used to predict property shifts for all materials. The data base used to develop the shift curves includes data of doubtful accuracy. Source of possible errors include: i) Accuracy of dosimetry, particularly of older data, is in question. Estimates of possible error range as high as +60%. 32 b) Calibration information for test machines is not available from much of the data; thus, testing errors could be significant. The data base includes information from A302-B steel irradiations. This data may not be applicable for prediction of property shift for A533-B, Class 1 steel used in present-day vessels . 3 3 Hawthorne has recently finished a preliminary assessment of the degree of conservatism in NRC Regulatory Guide 1.99. His data for A533-B plate and weld show the Guide to be extremely 142 \N\\\\ < \\\\\ \ 5 V\\\\ \ i Yv\\ \ \ ° W\\\ i~ \ »\ \ \ * \ ^ v\\\ x \ * v\\\ \ \ °. °. ^\ \\ * \ o o \\ » \ V\ m a. a. V\ N \ ° — 1 V^l 1 i V\ H _i \ o i_ n Q. 3 ii i Q- i y II 1 3 1 * £ o © v> O O 3 Tf CO < o o o LT) 03 c o o Q- D u •a < OQ < u OJ °- o o o 2^ 2d 33-B 10% 13% 14% 2 pop X CM O a X Q. a 3 400 300 Q 200 100 10' 1 1 I I I I I BASE MSTAL WELD HA2 A302 • ■ ♦ AS33 a ♦ I I 1 I 10' 10 18 10 19 FLUENCE n/cm 2 (E>1Mev) Figure 3. Data Used to Develop NRC Reg. Guide 1.99 io- 10- 143 conservative in the prediction of upper shelf Charpy energy drop at low fluence. A summary of these data is shown in Figure 2, Consultants' Meeting on Integral Cross Sections Measurements in Standard Neutron Fields", Vienna, November 15-19, 1976. Because of the aforementioned problems and inaccuracies inherent in the present trend curves of Regulatory Guide 1.99, a new trend curve analysis should be formulated. To demonstrate the impact of less than optimum dosimetry data, refer to Figure 3. This figure shows the correlation of a fracture toughness property as measured by the Charpy shift of various pressure vessel steels versus fluence. The large scatter in the data is evident. This data base impacts directly on the licensing of nuclear reactors in the U.S. The fluence limits for pressure vessel steels with various Cu and P content as given in Reg. Guide 1.99 is shown superimposed. The benefit of more precise and accurate dosimetry as well as the correlation of the data with spectral indices is evident. Conclusion Dosimetry beyond the core is an essential part of our reactor technology. In the materials test program, it is the link relating changes in the micro- and macroscopic material properties to neutron fluence and spectra. For pressure vessel surveillance programs, dosimetry is needed for the interpretation of actual or projected damage effects throughout reactor life and extrapolation from the surveillance testing positions to the pressure vessel wall itself. The dosimetry for shielding is a result of various problems receiving general attention for all types of reactors. These shielding requirements have given rise to interrelated programs of evaluation and benchmark experiments which coordinate closely with dosimetry benchmark programs. References M. Vlasov, "Proceedings of a Consultants' Meeting on Integral Cross Section Measurements in Standard Neutron Fields", Vienna, November 15-19, 1976, Report INDC(NDS )-81/L, IAEA (1977). R. Odette, et al . , "Application of Advanced Irradiation Analysis Methods to Light Water Reactor Pressure Vessel Test and Surveillance Programs", Proc. of the ASTM 8th International Symposium on the Effects of Radiation on Structural Materials, St. Louis, Mo., May 1976. E. M. Oblow, "Reactor Cross Section Sensitivity Studies Using Transport Theory", ORNL-TM-4437, April (1974). M. L. Williams and W. W. Engle, Jr., N.S.E 62_, 92 (1977). U. Farinelli, "Proc. of the Specialists' Meeting on Sensitivity Studies and Shielding Benchmarks", page 25, OECD, Paris, 7-10 October (1975). J. Grundle and C. Eisenhauer, "Proceedings of a Consultants' Meeting on Integral Cross Section Measurements in Standard Neutron Fields, IAEA, Vienna, November 15-19, 1976. Proceedings of the First International ASTM-EURATOM Symposium on Reactor Dosimetry, Petten, The Netherlands, September 22-26, 1975. w. N. McElroy , e t al , "Proceedings of a 9. M. Vlasov, op. cit. 10. A. F. Avery and S. D. Lympany, "Proc. of the Specialists' Meeting on Sensitivity and Shielding Benchmarks", OECD, Paris, October 7-10, 1975. 11. V. Herrnberger, et al . , op. cit. 12. G. Hehn and P. Stiller, "Prediction of Radiation Damage in Structural Materials Outside the Reactor Core", Paper 32, Proc. of 1st ASTM-EURATOM Symposium, Petten, The Netherlands, September 22-25, 1975. 13. Nuclear Technology, 25_, 1975. 14. Proc. of a Consultants' Meeting on Nuclear Data for Reactor Neutron Dosimetry, IAEA, Vienna, September 10-12, 1973. 15. Proc. of the ASTM-EURATOM Symposium on Reactor Dosimetry, Petten, The Netherlands, September 1975. 16. Private Communication, W. McElroy. 17. W. N. McElroy, et al . , Special Series, Nucl, Techn. 25, 177-422 (1975). 18. G. DeLeeuw, "In-Pile Neutron Spectroscopy: Status", CEN/SCK Report (1976). 19. R. Gold, "Neutron Spectrometry for Reactor Applications: Status, Limitations and Future Directions", HEDL-SA 901 (1972") and in Ref. 2. 20. H. Farrar, E. P. Lippincott, and W. N. McElroy, "Dosimetry for Fluence Applications", HEDL-SA 939 (1975), and in Ref. 2 . 21. J. H. Roberts, "Absolute Flux Measurements of Anisotropic Neutron Spectra with Proton Recoil Tracks in Nuclear Emulsions", Rev. Sci. Instr. 28, 677 (1957). 22. R. Gold, "Energy Spectrum of Fast Cosmic Ray Neutrons Near Sea Level", Phys. Rev. 165, 1406 (1968). 23. T. Iijima and S. Momoto, "Measurements of Anisotropic Fast Neutron Spectrum with Nuclear Emulsions", Nucl. Sci, and Eng. 22, 102 (1965). 24. L. Rosen, "Nuclear Emulsion Techniques for the Measurement of Neutron Energy Spectra", Nucleonics U_, 7, 32 (1953). 25. J. H. Roberts and A. N. Behkami, "Measurements of Anisotropic Fast-Neutron Spectra with Nuclear Emulsion Techniques", Nuclear Applications 4, 182 (1967). 26. J. H, Roberts and F. E. Kinney, "Measurements of Neutron Spectra by Use of Nuclear Emulsions Loaded with Li Glass Specks", Rev. Scien. Instr. 28, 510 (1957). 27. R. L. Fleischer, P. B. Price, and R. M. Walker, Nuclear Tracks in Solids, University of California Press, Berkeley, California (1975). 144 28. "Operating Unit Status Report." U.S. Nuclear Regulatory Commission, NUREG-0020-1 , August 1976. 29. Title 10 - Chapter 1, Part 50, "Licensing of Production and Utilization Facilities", Federal Register, Vol. 38, No. 130, July 1973. 30. Regulatory Guide 1.99, "Effect of Residual Elements on Predicted Radiation Damage to Reactor Vessel Materials", U.S. Nuclear Regulatory Commission, July 1975. 31. S. H. Bush, "Structural Materials for Nuclear Power Plants", ASTM Journal of Testing and Evaluation, Vol. II, No. 6, November 1974. 32. R. Odette, N. Dudey, W. McElroy, R. Wullaert, A. Fabry, "Application of Advanced Radiation Analysis Methods to LWR Pressure Vessel and Surveillance Programs", ASTM 8th International Symposium on Effects of Radiation on Structural Materials, St. Louis, Mo., May 1976. 33. J. R. Hawthorne, "Radiation Effects on Vessel Steels and Welds with Varied Residual Element Content and Embrittlement Relief by Post irradiation Annealing", Fourth Water Reactor Safety Research Information Meeting, Nuclear Regulatory Commission, September 1976. 145 FISSION YIELDS: MEASUREMENT TECHNIQUES AND DATA STATUS W. J. Maeck Idaho National Engineering Laboratory Allied Chemical Corporation Idaho Falls, Idaho 83401 Techniques for the measurement of absolute and relative fission yields are reviewed. The preferred techniques are isotope dilution mass spectrometric measurement of the individual fission products, and the heavy element mass difference or the fission product summation method to establish the number of fissions. The accuracy of most thermal fission yields appears adequate, and the status of the various yield compilations is reviewed. For fast fission yields, the data are not nearly as well established. For many heavy nuclides, fast fission yield data are nearly nonexistent. Because fast fission yields change with neutron energy, it is imperative that fast yield data be evaluated as a function of neutron energy to generate the most complete and accurate compilation. (Fission yield measurement techniques, counting, mass spec- trometry; fission yields, thermal, fast, absolute, relative; data compilations; fission yields versus neutron energy) INTRODUCTION AND HISTORY Fission product yield data constitute one of the most important sets of basic information in the nuc- lear industry. Fission yield data are important in the measurement of nuclear fuel burnup; safeguards; neutron dosimetry and flux measurements; nuclear physics calculations; reactor design and operation (both from the physics and engineering standpoint); decay heat studies; shielding calculations; fuel handling and reprocessing; and environmental studies. The measurement of fission yields is as old as the discovery of fission itself. Hahn and Strassman performed the first yield measurements in their initial discovery of the fission process. Fission yield studies in the early 1940 's established the asymetric mode of thermal fission; however, the shape of the mass yield curve was not well defined because of the rather large uncertainties associated with the data. In the 1950' s, the shape of the mass yield 235 curves for thermal fission, especially for U, became reasonably well defined and the existence of "fine structure" was established. Similar data for 233 239 U and Pu thermal fission became available in the early 1960's. Up until the early 1960's the majority of the fission yield data had been obtained using radio- chemical techniques. In the mid 1960's the demand for improved and new yield data increased. Radio- chemical data lacked the desired accuracy because of uncertainties in the decay schemes, half -lives, and in-pile and out-pile decay corrections. These factors also tended to limit the applicability of the tech- nique. A more accurate technique for the measure- ment of fission yields is mass spectrometry because concentration measurements for the stable isotopes using the isotope dilution technique are more accu- rate than counting of radioactive isotopes, and several isotopes of the same element can be measured simultaneously. Thus, several laboratories became involved in fission yield measurements using mass spectrometric techniques with the most extensive work 2 being done at Idaho Falls . The stable and long-lived isotopes of Rb, Kr, Sr, Zr, Mo, Ru, Cd, Xe, Cs, Ba, La, Ce, Nd, Sm, Eu, and Gd can now be routinely de- termined using the isotope dilution mass spectro- metric technique. Up to 1970, most nuclear energy laboratories had emphasized thermal fission yield measurements and over 500 literature references are available. The shift from thermal yield measurement to fast yield measurement started about 1970. Present fast fission yield measurements being produced are based on radiochemical measurements for samples irradiated in low-flux fast reactor critical assemblies, radio- chemical measurements for samples irradiated with beams of monoenergetic neutrons, and mass spectro- metric measurements on samples irradiated in high flux fast reactors such as EBR-II. In this latter phase, the most extensive efforts are again being conducted at the Idaho National Engineering Laboratory (INEL) 3 4 near Idaho Falls ' . TYPES OF FISSION YIELD MEASUREMENTS Fission yield data basically can be divided into two types, relative fission yields and absolute fission yields. Depending upon the specific appli- cation of the data, each have their merits; however, the most fundamental needs are for absolute data. Unfortunately, these are the most time consuming and difficult to obtain. In the following discussion, fission yields refer to cumulative or end-member chain yields after delayed neutron emission. Absolute Fission Yields Fission yield is defined by the following re- lationship: % FY^ x 100 where FY is the fission yield of a given nuclide, x ' x, N , is the number of atoms of the given nuclide x formed, and F is the number of fissions that occurred. In the measurement of absolute yields, the most difficult measurement is the number of fissions. Generally, the number of the selected fission product atoms can be determined with smaller uncertainties. Measurement of number of fissions - At least four techniques have been used to establish the number of fission events. These are: 1) heavy element mass difference, 2) fission product summation, 3) fission counting, and 4) calculational methods. A discussion 146 of each of these techniques and their associated un- certainties follows: 1. The number of fissions, F, based on the heavy element mass technique is derived from the relationship, U e (u+u cp ) where U° is the accurately determined number of heavy element atoms being irradiated, U is the number of atoms of the heavy element remaining after the irradiation, and U c _ is the number of atoms of the heavy element which has been converted to another element by a capture or decay process. In this method, an amount of the well characterized (chemically and isotopically) heavy element is accurately weighed into the irradiation capsule and the capsule is sealed. Follow- ing irradiation, the capsule and its contents are dissolved and the amount of the heavy element and its isotopic composition are accurately measured, usually by mass spec- trometry. Also measured, is the amount of other elements formed by capture and/or decay processes. Heavy element mass difference is potentially the most accurate method for establishing the number of fissions, provided that the loss (ie, burnup) of the target element is 10% or greater. Using isotope dilution mass spectrometry, the pre- and post-irradiation content of the sample can be determined to at least 0.1%; however, even with a 0.1% absolute uncertainty, the error in the number of fissions is ^1.5% relative for a burnup of 10% [(100+0.1) - (90±0.1) = 10±0.14]. Therefore, irradiation to 20% - 40% burnup is preferred, such that the number of fission can be determined to better than 1%. The sources of error in this method are: a) the characterization (absolute number of atoms) of the heavy element target, b) the accuracy of the mass spectrometric analysis, including the spike isotope cali- bration, c) incomplete dissolution of the sample, and d) contamination. To minimize errors which might arise from changes in the response of the mass spectrometer or from un- known changes in the isotope dilution spike isotope concentration, it has been strongly recommended that an archive sample of the heavy element be retained and analyzed when the post-irradiation heavy element analysis is performed. A major advantage of the fission product sum- 2 ation technique , compared to the heavy ele- ment mass difference method for determining the number of fission is that it is especial- ly applicable to samples having low burnup (1-10%). The summation technique is based on the fact that the sum of all of the fission products in one of the peaks in the mass yield curve is equal to the number of fissions. The success of this technique is directly dependent upon the fraction of the selected mass peak which is measured. The preferred peak for measurement is the heavy mass peak because the elements in this peak are more easily measured. The preferred measurement technique for the individual fission product atoms is isotope dilution mass spectrometry. In our laboratory, approximately 90% of the 4 fission product atoms in the heavy mass peak (isotopes of Xe, Cs, Ba, La, Ce, Nd, and Sm) are measured to an uncertainty of better than 1% relative. The number of atoms of the unmeasured isotopes is obtained by interpolation and extrapolation, and an un- certainty of 10-20% relative is assigned to these values. This procedure only slightly increases the uncertainty in the total number of fissions. This technique is cap- able of establishing the number of fissions to an uncertainty of 1.0-1.5%. A requirement of this method is that the number of fission events should approach ] 8 10 such that sufficient quantities of fission products are available for measure- ment and the amount of natural contamination is minimized. The principal sources of error in this tech- nique are: a) measurement of the individual fission product atoms, including the uncer- tainties in the concentration of the various spike isotopes, b) incomplete dissolution of the target nuclide and all fission product nuclides, c) introduction of fission product element natural contamination from reagents and equipment, d) leakage of fission gases from the target assembly during and after irradiation, and e) incomplete collection of fission gases. 3. Two methods are used to measure the number of fission events directly. These involve use of fission counting chambers and solid- state track recorders (SSTR's). However, these methods do not measure the number of fissions which have occurred in the sample which is measured for fission products. In both of these methods two target foils are used. Both targets are adjacent and irradiated in the same neutron flux. One of the targets is a bare thin foil from which the fission fragments escape and are counted, and the other is a heavy target, usually wrapped in aluminum foil to contain the fission fragments. The heavy target is analyzed for the desired fission products. The number of fissions occurring in the thin foil is counted, and the number of fissions occurring in the heavy target is calculated from the relative amount of the target nu- clide in each foil. When fission chambers (ionization chambers) are used, the number of fissions occurring in the thin foil is established by direct counting of the fission fragments escaping from the foil. In the SSTR technique the thin foil is irradiated in direct contact with a material such as mica. The recoil fission fragments produce structural damage in the mica such that when the mica is etched with hydrofluoric acid, the damage is revealed in the form of individual tracks. The tracks are counted under a microscope, and the number of fissions is calculated by multiplying the number of tracks by an optical efficiency factor. An excellent detailed discussion of these techniques and their associated sources of error has been 147 presented by Grundel, Gilliam, Dudey, and Popek . The use of these techniques is limited to low fluence and hence, a low number of fission events in the heavy foil. The heavy foil is dissolved and the desired fission products are separated and analyzed by count- ing, or the heavy foil is counted directly, preferably using a well calibrated high resolution Ge(Li) detector system. Accuracy in the measurement of certain nuclides is improved by counting at different time periods following the irradiation. Because the number of fission events is low, the use of mass spectrometry to determine the indiv- idual fission product abundances for fission yield measurements is not practical. The two major sources of error in both tech- niques are: the measurement of the number of target atoms in the thin foil and the re- lationship between the number of recorded events and the number of fissions. Other common sources of error are: a) production of fission events from other nuclides in the target material, b) contamination in the foil backing or in the materials of con- struction, c) fission fragment adsorption in the thin foil deposit), and d) fission fragment scattering. Sources of error distinct to the fission chamber technique are: a) efficiency and calibration factors for the chamber, b) correction for dead time losses in the fission rate measurement, c) background counting corrections, and d) scattering corrections. Based on very well controlled experiments, the number of fission events determined by use of a fission counting chamber can be established with an uncer- tainty of vL.5% . The major distinct source of error for SSTR's is the optical efficiency factor for con- verting the number of tracks to fission. Other distinct sources of error are the counting of the number of fission tracks, the etching of the mica plate, and the identification of the tracks which result from fission fragments. Currently, the estimated uncertainty for the determination of the number of fissions by this technique ,„6 is ^3% . 4. The number of fissions also can be calcu- lated using some measured quantity and an assumed value for a given factor, such as a cross section or the capture-to-fission ratio, a. For example, the number of 235 U fissions can be calculated from the measured pre- and post-irradiation isotopic 235 composition of U and an assumed value 235 for a for U. Because a varies with neutron energy and temperature, its true value is always in question. Other calcu- lational methods based on neutron fluxes and cross sections also are subject to significant uncertainties. This is the least accurate of the various method dis- cussed. Measurement of fission product atoms - The oldest method for measuring fission product concentrations is based on counting of a given radioactive nuclide. Briefly, the irradiated target is dissolved, the de- sired nuclide is isolated by chemical means, and the activity of the separated product is measured. All of the early yield data were based on beta counting, because the gamma/disintegration values were unknown. In some cases, especially where the activity level is very low, beta counting is still used today. Precise gamma ray spectrometry has minimized the need for exhaustive chemical separations because discreet gamma rays can be associated with a given nuclide. In some instances, no chemical separations are performed and the irradiated target is counted directly for selected radionuclides. The preferred counting technique is high resolution gamma ray spec- trometry. Counting techniques are especially useful for the measurement of short-lived nuclides, or where the number of fissions is low. Sources of error in converting gamma ray spectrometric data to atoms of the partic- ular fission product are: a) the method for subtraction of the baseline or background activity, b) detector efficiency, c) the accuracy of the stan- dards if a comparison counting technique is used, d) the nuclear constants, specifically the photon-to- disintegration ratios and the half-life values, e) the mass adsorption of the gamma rays by the material being counted, f) random coincidence summing, and g) interferences from other gamma rays. In a limited number of cases, an uncertainty of 1-2% can be attained for the determination of the number of fission product atoms using high resolution gamma ray spectrometry. Uncertainties in the range of 3-10% are more common. The most accurate method for the measurement of the concentration of stable and long-lived nuclides is isotope dilution mass spectrometry. In this tech- nique, a known number of spike isotope atoms, pref- erably of an isotope not formed in the fission pro- cess, is added to a sample of the dissolved target and chemical identity of the spike isotope and sample is effected by chemical means. That element is sepa- rated chemically and the isotopic composition is measured using a mass spectrometer. The ratio of the spike isotope peak to the other mass peaks of that ele- ment is measured and the number of atoms of the other isotopes is determined relative to the number of spike isotope atoms added. A particular advantage of this technique is that quantitative recovery in the sepa- ration procedure is not required. It is imperative, however, that isotopic exchange be achieved. The irradiated sample should have experienced 1 8 about 10 fissions so that a reasonable amount of material is produced for measurement. Because the ionization efficiency in the mass spectrometer varies from element to element, the minimum amount of a fission product required for a reliable mass spectro- metric measurement also varies. For example, 100 nanograms of Rb or Cs on the mass spectrometer fila- ment is more than adequate, while several micrograms of Ru are ideal for analysis. 148 Because the accuracy of any isotope dilution technique is controlled by the accuracy of the spike isotopes, great care must be taken in their prepa- ration and standardization. In many cases, the occurrence of natural contamination can be corrected because one or more of the naturally occurring iso- topes is not formed in fission. For those cases where all of the naturally occurring isotopes of the element also are fission product isotopes (eg, Rb, Cs) significant errors can occur. Spectral inter- ference from an isotope of another element can also occur; however, this problem can usually be minimized by improved chemical separation techniques. Other factors which must be considered are mass fractiona- tion, especially for the light elements, from the mass spectrometer filament, and mass discrimination corrections. For most elements, the isotopic composition and atom abundances can be determined to an uncer- tainty of 0.25%. Other techniques, although limited in scope, which have been used to measure selected fission pro- 99 2 duct concentrations are spectrophotometry for Tc , X-ray analysis for one or more rare earth elements , 2 vapor phase chromatography for the fission gases , Q and volumetric determination of the fission gases . Summary - Absolute Fission Yields - The pre- ferred methods for the measurement of absolute fission yields is the isotope dilution mass spectrometric measurements of the individual fission products coupled with the heavy element mass difference or the fission product summation technique to establish the number of fissions. Based on state-of-the-art, more yield data with smaller uncertainties can be obtained from a well designed irradiation using this technique than from the others which have been discussed. A particular item of concern, however, is that while irradiations to high burnup can result in a more accurate value for the number of fissions, signifi- cant corrections for neutron capture on some of the fission product nuclides are required. This is especially true for irradiations conducted in thermal reactors. Although correction factors can be applied when sufficient information concerning the irradi- ation is available (for example, see Ref. 9) the existence of significant unknown cross sections can lead to biased fission yield data . Because of this, it is believed that the more correct approach may be to irradiate a larger target to a low burnup, 1% or less, and to determine the number of fissions by the summation technique using isotope dilution mass spectrometry . Yields with uncertainties in the 1-2% range are attainable using this approach. generally displaced one or two Z units from the end- member of the chain. Thus, the yield may have a lower value than that determined on the stable end- member of the chain because of the independent yield contribution from the other non-measured members of the chain. This difference may be as large as 5% relative. Relative Fission Yields Relative fission yields, which constitute a major fraction of the reported yield data, are yields which have been measured relative to some other nuclide or normalized to some value. The basic assum- ption is that once the yield of one fission product is accurately known, from absolute yield measurements, is can be used as a monitor of the total number of fissions, or other yields can be measured relative to it. Nuclides which have been used as reference stan- , , 95„ 97 v 99 M 137„ 140 B , 148 H , dards are Zr, Zr, Mo, Cs, Ba, and Nd. The bulk of the relative yield data is associated with radiochemical measurement. Several variations of treating the data on a relative basis have been proposed over the years, but the one receiving the most attention is the R-value method. This method is based on the assumption that if the yield values for a given fissionable isotope, 235 usually U, are well known, that uncertainties in the radiochemical measurement, especially those associated with decay schemes and counting geometry, can be minimized. In practice, samples of two dif- ferent fissionable isotopes are irradiated simultan- eously, and each is chemically treated to obtain the same nuclides from each source. The reference 140 monitor nuclides, for example Ba, from each target isotope and the unknown from each source are then counted as close in time as possible. Assuming that the yield of the reference nuclide is known, the yield of the unknown can be determined. Various ex- perimental techniques and prior references are given in a review by von Gunten The major obvious systematic source of error in any type of relative yield measurement is the accuracy of the reference standard. Because absolute thermal yield values are now reasonably well known, relative yield measurements for thermal yields is a viable op- tion. It is especially useful in the measurement of the valley nuclides and those on the extreme wings of the mass yield curve where absolute mass spectrometric yields have not been determined. The application of this technique to fast yield measurements suffers from the fact that the yields are a function of the energy of the incident neutrons; hence, relative fast fission yield data must be carefully evaluated with respect to the assumed yield of the reference nuclide. Recent improvements in fission chamber counting coupled with direct gamma-ray spectrometry measure- ments have resulted in a limited number of absolute yields being reported with uncertainties of ^2.5% This approach should be most valuable in the measure- ment of yields where irradiations are conducted in low flux critical assemblies or when mono energetic neutron beams are used. A feature of any counting technique compared to the mass spectrometric techniques, is that in most cases, the yield of the nuclide being determined is FISSION YIELDS - STATUS AND NEEDS To date, approximately 1000 literature references relative to all types of fission yield measurements are in existence. Recognizing that it would be im- possible for each user to evaluate and select the best value from this mass of data, several individuals have taken up the task of evaluating and compiling these data. As could probably be predicted, differ- ences occur in these evaluated data sets which exceed the error assignment attached by the various evalu- ators. In attempting to clarify the situation, the 149 most used compilations of thermal and fast fission 13 14 yields were reviewed by Walker and Cuninghame , respectively, for the 1973 IAEA Panel Meeting on Fission Product Nuclear Data . The current status of the yield data compilations and data requirements for thermal and fast yields are discussed below. Thermal Fission Yields The review, evaluation, and compilation of data from nearly a thousand individual literature refer- ences is a monumental task. Several individuals have undertaken this task with varying degrees of vigor and continuity. Of these, the most well documented and referenced are the works of Meek and Rider , E.A.C. Crouch , W. H. Walker , Lammer and Eden , and the U.S. Evaluated Nuclear Data File (ENDF/B-1V task force group. 19 All of the compilations are based on some version of pooling and weighting techniques, plus the per- sonal opinion and preferences of the individual evaluators. Briefly, Meek and Rider attempt to in- clude all reported data in a computerized data library and then assign various weighting functions to the reported data based upon the method of measurement. Crouch , whose work is sponsored by the U.K. Chemical Nuclear Data Committee, also uses a computer based library but tends to limit the input to well docu- mented data. The errors are assigned by Crouch. 9 Walker's compilation is based primarily on mass spectrometric yield data, except where mass spectro- metric data are not available and then radiochemical 1 8 data are used. Lammer 's and Eden's compilation is not as extensive as the others and is primarily based on mass spectrometric data and the evaluation tech- 9 nlque is similar to that of Walker . Version IV of of the ENDF/B file is basically the same as the Meek and Rider compilation. For a more detailed discussion of these data sets and a comparison study of the data sets, it is strongly recommended that the reader consult Walker's 13 excellent review presented at the 1973 IAEA Bologna Fission Product Nuclear Data Conference. Since that time, the primary emphasis in the U.S. has been an intensified effort to upgrade and update the ENDF/B data file. To this end, a task force group which includes, among others, B. F. Rider and W. H. Walker, has been formed under the auspices of U.S. ERDA to develop a preferred fission yield data file. This file is to be continually updated and Version V should be available for presentation at the Second IAEA Advisory Group Meeting on Fission Product Nuclear Data to be held in Petten, Netherlands, September 1977. Concurrent with the U.S. effort, the UK fission yield data file is in a continual upgrade under Crouch's direction. Rider and Crouch are con- tinually exchanging information with respect to literature references. The update status of the Lammer and Eden compilation is uncertain at this time. A review of the current status of the various thermal fission yield compilations is being prepared by J. G. Cuninghame, UK, for presentation at the Petten Conference mentioned above. At the Bologna IAEA meeting , it was concluded that the thermal yield data, except for certain iso- lated cases, generally satisfied the user's required accuracies, and no new extensive remeasurement pro- gram was justified. The exceptions are detailed in Volume II of Reference 15. In 1975, we had the opportunity to remeasure a 235 limited number of thermal fission yields for U and 239 Pu. Although to date, only relative yields have 20 been measured, the initial results of this study show for several isotopes that there are large differ- ences compared to previous measurement made in this 2 laboratory and compared to data in current fission yield compilations . Particularly significant are 1 38 14% higher values for Ba and 8.5% higher values 239 for the Xe isotopes for Pu thermal fission. The 239 major impact is that all of the Pu fission yields will have to be adjusted to preserve mass balance. 235 Only minor differences were observed for U thermal fission. It is expected that new absolute thermal 235 yield data for about 40 different nuclides for U 239 and Pu will be produced when this measurement pro- gram is completed in 1978. Fast Fission Yields Fission yield data for fast neutrons are not nearly as well developed as they are for thermal neutrons. Based on the fast fission yield compil- 14 ations produced through 1973 and Cuninghame 's review , the data are fragmentary and usually carry large uncertainties. The most complete fast yield compil- 235 239 ations are for U and Pu which were produced by Meek and Rider and Crouch . The fast yield data 232^ 233,, 237„ 238 TI . 241 A -, ■ ■ ^ A for Th, U, Np, U, and Am are limited and generally have large errors. No reliable fast yield data are given for the higher isotopes of plutonium. This lack of data stems from several sources. Primary, is the lack of available irradiation facil- ities. Because fast reactor fission cross sections are low, prolonged high flux irradiations are re- quired to generate a sufficient number of fission product atoms for mass spectrometric measurements. If this is not possible, irradiations can be ma'de in experimental low power critical assemblies; however, normally only a limited number of radiochemical measurements can be obtained. Another consideration is that accurate fast yields are more difficult to measure than thermal yields, because all of the heavy nuclides undergo fast fission, while thermal fission is only signifi- 233,, 235,, 239 D , 241 D _ , . cant for U, U, Pu, and Pu. This is especially important in the measurement of fast fission yields for the even-even and odd-odd heavy ,232 m ^ 238 IT 237„ 240,, 242 D , nuclides ( Th, U, Np, Pu, Pu, and 241 Am) , because all have a high neutron energy threshold cross section for fission, a relatively high capture-to-fission ratio, and in some cases, a long-lived capture product which fissions with any energy neutron. For example, in a long-term fast 150 238 240 T neutron irradiation of U or Pu, a large fraction 239 241 of the fissions will result from Pu and Pu, respectively. While corrections for the fission from the capture product can be made if the fission yields for the fissioning capture product are well known, this correction can result in significant errors in the fission yields for the principal target nuclide. Before going further, the term "fast fission yield" should be addressed. It is at best a relative term because it is well known that fast fission yields can vary with neutron energy. This one item has made it most difficult for the evaluators and compilers of "fast yields" to produce accurate com- pilations, because it prohibits use of the pooling techniques which have been applied to thermal yield data. Unfortunately, because most reported experimen- tal fast yield data can not be associated with a known neutron spectrum, the compilers to date have had no choice other than to pool the data. Thus, it is not unreasonable that fast yield compilations show significant discontinuities for adjacent yields because the data were probably obtained from irradi- ations conducted in significantly different neutron spectra. It is appreciated that the early worker probably had little regard for the neutron spectrum effect on yields and that just obtaining fast yield data was of primary concern; however, it is strongly emphasized that all new measurements be associated with a known neutron spectrum. The effect of neutron energy on fission yields is shown in Figure 1. In this presentation, the ratios 22 235 of the fast to thermal yields for U are plotted as a function of mass number. The general result of fissioning with increasing energy neutrons is an increase in the yields on the wings of the mass yield curve, an increase in the valley yields, and a small depression in the peak yields. This effect appears to change systematically as the energy of the fissioning neutron increases. The discontinuities in the plotted data may be due to certain fine structure effects or the result of inaccuracies in 22 data source An extensive fast yield measurement program is currently being conducted in our laboratory at the Idaho National Engineering Laboratory. Mass spectro- metric yields are being measured for over 40 nuclides , 233.. 235., 238 IT 237 H 239_ 241 D 242 D for U, U, U, Np, Pu, Pu, Pu, and 241 2 4 Am which were irradiated in EBR-II Yield data 233 235 238 239 for U, U, U, and Pu and neutron spectral 4 240 data were published in 1975 , and data for Pu, 0A1 0/0 r \~] Pu, Pu, and Np should be available by the end of 1977. Of primary concern in this program are the fast fission yields for the isotopes of Nd which are used for monitors for the determination of burnup in fast _1 < a. < 10 •*• • ••• 1.0 0.8 -L ••• •• l( «m«* # ••••%*• ••*•%• • ^ 75 80 90 100 I 10 120 130 140 150 160 MASS NUMBER Fig. 1 Uranium-235 Fast to Thermal Fission Yield Ratios. 151 TABLE I. RELATIVE Nd ISOTOPIC COMPOSITIONS AND ISOTOPIC RATIOS FOR 235 U FISSIONING s< f f(f f Spectral Index, Reactor, and Reference 8 5 Thermal 10 EBR-II 5 EBR-II 4 23 EBR-II 5 OSIRIS 24 EBR-I 4 Axial Bl. Row- 8 Row- 4 Core SI tf f Itff 8 8 0.0 0.003 0.02 0.035 0.06 0.089 0.12 Isotopes iA3 Nd 0.3923 0.3898 0.3891 0.3870 0.3840 0.3836 0.3826 145 Nd 0.2583 0.2589 0.2563 0.2544 0.2545 0.2546 0.2544 146 Nd 0.1964 0.1975 0.1969 0.1981 0.1985 0.1962 0.1973 148 Nd 0.1101 0.1110 0.1126 0.1133 0.1149 0.1167 0.1161 150.,, Nd 0.0429 0.0429 0.0450 0.0472 0.0481 0.0489 0.0496 Isotopic Ratios 150/143 0.1094 0.1101 0.1157 0.1220 0.1253 0.1275 0.1296 150/145 0.1661 0.1657 0.1756 0.1855 0.1890 0.1920 0.1950 148/143 0.2807 0.2848 0.2894 0.2928 0.2992 0.3042 0.3034 For U, the change in the Nd/ breeder reactor fuels. In the process of evaluating our Nd fast yield data and comparing it to data re- ported by other workers, the noted differences were considered to be too large to be attributed to meas- urement error. In lieu of comparing absolute yield values, we evaluated the relative isotopic data because systematic errors in the measured number of fissions used to calculate the absolute yields could be eliminated. A comparison of the relative Nd isotopic data is given in the top half of Table I. When the data are ordered by spectrum hardness, system- atic changes in the Nd isotopic composition are apparent. To amplify these changes, selected isotopic ratios were calculated and are given in the bottom half of Table I. 143 Nd ratio between the thermal neutron values and the very hard EBR-I spectrum values is ^20%. As a result of the trends shown in Table I, several different forms of neutron energy indices were considered to provide a means for the possible correlation of yields with energy. Included were mean neutron energy, median neutron energy, mean and median neutron energy for fission of a given heavy isotope, fraction of neutrons in a given energy range, and the ratio of cross section values for two selected 238 fissioning isotopes. Of these the ratio U(n,f)/ 235 U(n,f) was selected because more yield data could be associated with this index than any of the others. This ratio gives a good indication of spectral hardness because ^95% of the neutron energy .response for o o o U(n,f) is above ^1.4 MeV and ^-95% of the response 0.25 for 235 U(n,f) is below 2 to 3 MeV. .04 .06 SPECTRAL .08 0.10 0.12 INDEX The change in the isotopic ratio, Nd/ Nd, 235 239 for those U and Pu data which could be associ- 238 23 5 ated with the spectral index, U(n,f)/ U(n,f), is shown in Figure 2. The correlation of the iso- topic composition of Nd with the selected spectral index is clearly demonstrated. Fig. 2 Change in Isotopic Ratio of 150 Nd/ 143 Nd with Neutron Energy for 235 U and 239 Pu Fast Fission. 152 To extend the correlation of isotope ratios with neutron energy to other fission products, the exten- 235 sive U fission yield data obtained in the author's 2 10 laboratory for thermal fission ' , fast fission in 4 row-8 of EBR-II , and fast fission in the core of 2 EBR-I were evaluated. The respective spectral index values for these irradiations are zero, 0.02, and 0.124. Because little data are available between index values of 0.02 and 0.124, the correlations are not as accurate as those for the Nd isotopes. Isotopic ratios were calculated for the various fission product elements using the isotope whose fission yield changed least with neutron energy as the reference isotope. For example, for the Kr Of- isotopes, Kr showed little change with neutron energy and was selected as the reference isotope. + 18 +10 + 8 CO O < bJ -8 3 z +8 < -8 +8 85/87 RUBIDIUM STRONTIUM _ , 90/88 1 ZIRCONIUM r ^^^___^ 92/96_ 94/96s 91/96 93/96' 0.04 0.08 SPECTRAL INDEX 0.12 -8 1 97/95\ 98/95 100/95^ — MOLYBDENUM 1 1 0.04 0.08 SPECTRAL INDEX 0.12 Fig. 3 Change in Fission Product Isotopic Ratios with Neutron Energy for the Isotopes of Krypton, Rubidium, Strontium, Zirconium, Molybdenum, and Ruthenium. 153 + 18 1 1 ^-> 131/136 + 10 - 4 — /•"■' 132/136 ~ U XENON ~ J34/I36 +60r 144/146 143/146 0.04 SPECTRAL 0.08 INDEX 0.12 Fig. 4 Change in Fission Product Isotopic Ratios with Neutron Energy for the Isotopes of Xenon, Cesium, Neodymlum, and Samarium. The percent changes in the ratios, relative to the thermal values are presented in Figures 3 and 4. The isotopic ratio changes in the light mass peak are reflected by counter isotopic ratio changes in the heavy mass peak. For example, the change of 15 to 20% for Kr isotopes 83 to 86 on light wing of the light mass peak is reflected by a like change in the mass region 148 to 151 on the heavy wing of the heavy mass peak. A second example is that the 86 to 88 mass region and its reflected 146 to 148 mass region, both change little. Another example, is similar changes for the light Xe isotopes and the light Ru isotopes. Little significant changes are seen for those isotopes on the peaks of each wing of the mass yield curve. This study, of the changes in yields with neutron energy will be continued with the goal being the ability to predict a given yield for any neutron energy. To aid in the effort to quan- tify the changes in yields with neutron energy, it is imperative that every effort be made to define the neutron spectrum associated with fast fission 0.04 0.08 SPECTRAL INDEX 0.12 yields being determined. For experiments currently in progress, the workers should attempt to obtain this information from the reactor physicist, or preferably, irradiate specific samples to define the neutron spectrum. For new experiments, the worker should make every effort to include a series of spectrum monitors in the irradiation assembly. The compilers of fast fission yield data must use only well documented data and evaluate each set of yield data with respect to the neutron energy. REFERENCES 1. Hahn, 0., Strassman, F., Naturwiss. 2]_, 11,89 (1939). 2. Lisman, F. L. , Abernathey, R. M. , Maeck, W. J., Rein, J. E., Nucl. Sci. Eng. 42, 191 (1970). 3. Maeck, W. J., Larsen, R. P., Rein, J. E. , USAEC Report, TID-26209 (1973). 4. Maeck, W. J., Editor, U.S. ERDA Report, ICP-1050-1 (1975). 154 5. Larsen, R. P., Argonne National Laboratory, Private Communication, 1973. 6. Grundal, J. A., Gilliam, D. M. , Dudey, N. D. , Popek, R. J., Nucl. Tech. 2_5 (2), 237 (1975). 7. Larsen, R. P., Schablaske, R. V., Oldham, R. D. , Meyer, R. J., Homa, M. I., U.S. AEC Report, LA-4407, (1967). 8. Arrol, W. J., Chackett, K. F., Epstein, S. , Can. J. Res, 27B, 757 (1949). 9. Walker, W. H. , AECL Rept. 3037-11, (1973). 10. Maeck, W. J., U.S. ERDA Rept. ICP-1092 (1976). 11. Dudey, N.D., Popek, R. J., Greenwood, R. C, Helmer, R. G. , Kellogg, L. S., Zimmer, W. H. , Nucl. Tech. 25 (2), 294 (1975). 12. von Gunten, H. R. , BNES Proceeding of the International Conference on Chemical Nuclear Data, Canterbury, UK (1971). 13. Walker, W. H. , Review Paper 11a, IAEA-169 (1973). 14. Cuninghame, J. G. , Review Paper lib, IAEA-169 (1973). 15. IAEA Proceedings of a Panel on Fission Product Nuclear Data, Bologna, Italy, IAEA-169 (1973). 16. Meek, M. E. , Rider, B. F. , General Electric Co. Rept. NED0-12154 (1972). Update NED0-12154-1 (1974). 17. Crouch, E.A.C., Paper IAEA/SM-170/94 Symposium on Applications of Nuclear Data, Paris, (March 1973). 18. Lammer, M. , Eder, 0. J., Paper IAEA/Sm-170/13, Symposium on Applications of Nuclear Data, Paris (March 1973). 19. ENDF/B-IV Fission Yield Data File, National Nuclear Data Center, Brookhaven National Laboratory, USA. 20. Maeck, W. J., Emel, W. A., Delmore, J. E., Duce, F. A., Dickerson, L. L. , Keller, J. H. , Tromp, R. L. , U.S. ERDA Rept. ICP-1092 (1976). 21. Crouch, E.A.C., UKAEA Rept., AERE-R 7394 (1973). 22. Rider, B. F. , General Electric Co., Private Communication (1977). 23. Sinclair, V. M. , Davies, W. , International Conference on Chemical Nuclear Data, BNES, Univ. of Kent, Canterbury, England (1971). 24. Robin, M. , Bouchard, J., Darrouzet, M. , Frejaville, G. , Lucas, M. , Prost Marechal, F. , ibid. 155 FISSION REACTION RATE STANDARDS AND APPLICATIONS J. Grundl and C. Eisenhauer National Bureau of Standards Washington, D.C. 20234 Fission rate measurements in and around prototype and power reactors of all kinds, as well as in the criticals of reactor physics are vital elements in understanding nuclear energy generation rates, neutron transport, and the integrity of materials exposed to reactor radiation fields. Standardization and interlaboratory refer- encing for this historic measurement activity have improved significantly since the last neutron standards symposium in 1970. This advancement will be summarized along with a general orientation and description of fission detector response characteristics and interpretation. For the last, necessary analytic formulations and a brief treatment of error propagation are included. Also included is an up- dated look at observed versus predicted fission cross sections for fission spectrum neutrons and the related fast criticals of reactor physics. (Cross sections; fission; neutron reactions; neutron spectrum; reactor fuels; reactor materials) Introduction and Summary The requirement to measure isotopic fission rates appears in a wide variety of technical-engineering problems associated with radiation effects in materials and reactor design and management. Looking only at physical changes in materials exposed to neutrons the array is bewildering. Some examples are swelling of reactor fuel claddings, fuel matrix rearrangements, integrity of pressure vessel weldments, activation of reactor service equipment, vaporization of liquid rocket propellant, and gain changes in transistors. In the management of these radiation effects problems, fission rate measurements are often used indirectly to derive neutron fluence and spectrum information. Di- rect application of fission rate measurements are also important and they arise most often, and with a longer history, in the area reactor design and nuclear fuels management. Examples are verification and guidance of reactor physics calculations; optimization of fission power gradients; relative fission rates among fission- able isotopes competing for neutrons in a reactor core; determination of fuel burn-up and burn-in rates in all types of power reactors including nuclear weapons. Techniques of fission rate measurements are also diverse. Among active fission detectors there exist fission ionization chambers as small as a pencil tip and as large as a bread box; passive fission activa- tion detectors can be as thin as a needle or as large as a sheet of paper. Among destructive analysis tech- niques, mass spectrometry for example, will measure total fissions in almost anything that can be dis- solved. Of primary concern perhaps for a neutron standards symposium, are fission rates that give rise to funda- mental integral cross sections. These are the obser- vables that stand as constraints for the evaluation of differential cross section data. A single review of integral results at the previous Neutron Standards Symposium in 1970 is matched at this meeting with a half dozen papers on integral measurement techniques, facilities, and results. The remarks in 1970 regarding differential and integral measurement as "areas of effort that do not always communicate so well," is now almost outdated. 1 The need by a much harassed nuclear energy industry for consistent design and operation parameters overwhelms the vestiges of mutual disre- spect. Today, on one hand, an observed fission- spectrum-averaged fission cross section contributes to the setting of the 235 U fission cross section scale for ENDF/B-V, and on the other, reactor dosimetry spec- trum characterizations routinely use the new ENDF/B dosimetry cross section file with only mild threats to "adjust." Measurement Standards Accuracy requirements for many of the direct applications of fission rates can be severe: as low as ± (2-5)% (la) for the determination of reactor fuel burn- up or in checks of reactor physics calculations. Neu- tron flux and fluence estimates for managing radiation effects problems typically are less severe and de- pendence upon neutron transport calculations with less experimental verification is common. The latter is true also, because in situ neutron flux measurements can be difficult for radiation effects surveillance. However, when optimum material performance is sought, and high-investment engineering or safety decisions are involved, stiff accuracy requirements appear. In this case, a measurement program in support of calcu- lation assumes recognized importance. This is particu- larly true when critically evaluated spectrum charac- terization is sought and fission detectors coupled with other types of neutron energy sensitive activation de- tectors are employed (i.e., the familiar multiple foil spectrum unfolding techniques). Standards and intermeasurement reference for all of this varied and decades-long effort of fission rate measurement are of two kinds : 1) artifacts , such as standard solutions or fissionable deposits; and 2) well- characterized neutron fields, such as thermal equi- librium, monoenergetic, and fission neutron spectra. The latter are most important when fission rates are used to derive neutron field characteristics. Not all fission rates are standardized in the man- ner just stated, most in fact, are not. Measurements which depend upon extraneous instrument calibrations and various types of nuclear data are common (e.g., fission yields and alpha decay constants) . Until re- cently, there existed no central repository of mass assayed fissionable deposits generally available for fission rate measurement calibration or referencing. Considering the vital role fission rates play in all branches of radiation effects and nuclear energy de- velopment, the deficiencies in standardization and interlaboratory comparisons is surprising. The intro- ductory paper by R. Taschek at the 1970 Symposium re- ferred to this odd circumstance in relation to the use of fission rate measurements as a standard for neutron flux determinations with accelerators. 156 Improvements Since 1970 After 1970 notable concern and some improvements in referencing fission rate measurements have taken place. Recall first the operative distinctions set forth in References 1 and 3. Mascroscopic Data . Experimental results from arrangements in which a dominating feature is neutron transport, i.e., multiple encounters of neutrons with nuclei which produce a series of velocity changes, and in important instances, new sources of neutrons. Microscopic Data . Experimental results, generally energy dependent cross sections, from arrangements in which single encounters of neutrons with nuclei pre- dominate. Integral Measurements . Measurements that require for interpretation an integration of pointwise micro- scopic data over energy intervals and/or the neutron flux over space intervals that are not small compared to the range of interest. Differential Measurements . Measurements for which energy or space integrations are not necessary for in- terpretation, only for corrections to data. (1) In the area of macroscopic data from integral measurements, organized efforts are underway in the U.S. to achieve interlaboratory consistency in reactor physics and in reactor materials dosimetry . 1+ ~ 6 Inter- nationally, the IAEA has initiated a program of identi- fying benchmark neutron fields and evaluating their application. A Consultant's Meeting on Integral Cross Section Measurements in Standard Neutron Fields for Reactor Dosimetry was convened in November, 1976. To prepare for this meeting a compendium of benchmark neu- tron fields was prepared within the context of the following characteristics: 9 1. Simple and well-defined geometry; 2. Adequate neutron fluence and stable flux density; 3. Reproducible and accurately characterized neutron spectra based on spectrum measure- ments and/or reliable calculations; 4. Sustained availability for measurements. (2) Both integral and differential measurements have benefitted from the development of a permanently available set of reference and working fissionable de- posits at the National Bureau of Standards. 8 Isotopes presently represented in the set include 2 U, 239 Pu, 238 U, and 23V Np, with 21,0 ' 21,1 p u t o be added within the next year. Traceability of fission rate measurements to these deposits may be established by direct alpha or fission comparison counting of deposit pairs, or by exposure in a relevant neutron field of a complete fission detector to be calibrated along with an NBS double fission ionization chamber operating with an NBS mass assayed deposit. 8 (3) The situation regarding observed versus pre- dicted fission-spectrum-averaged fission cross sections is better now than it was in 1970. ' New absolute cross section measurements have been performed and the dis- crepancies with the differential microscopic data are no longer in excess of ten percent. 10-12 And most important for fast neutron standardization there exists now an evaluation of the fission neutron spectrum in- cluding uncertainty estimates based on all documented differential spectrometry. 9 ' 13 The stubborn discrep- ancies surrounding fission rate ratios in simple macro- scopic systems like the Big Ten Fast Metal Critical Assembly also have changed drastically following the inclusion of long-awaited changes in inelastic scatter- ing cross sections into neutron transport calculations/ This report will review briefly reference these various improvements standardization and the associated ad measurement. Preceding this and comp portion of the paper will be an orien fission detector measurements and the spectrum characterizations. This is fully, and no more controversial than the occasion of a symposium devoted t ards and applications. and largely by in fission rate vances in applied rising the major tation regarding ir use for neutron appropriate, hope- is beneficial on o neutron stand- Neutron Spectrum Response The neutron energy dependence of fission detection falls neatly into two categories: 1) full-energy-range detectors with generally flat fission cross sections in the MeV range and increasing fission cross sections down to thermal equilibrium energies; and 2) threshold detectors with smooth cross sections that rise rather sharply in the energy range 0.5 to 2 MeV. There is not much distinction within these two categories and a small array of fission detectors are sufficient for illustra- tion. Four are chosen for this review: 35 U(n,f), 239 Pu(n,f), 237 Np(n,f), and 23e U(n,f). Spectrum re- sponses of these detectors will be summarized as an introduction to quantitative methods of interpretation. Spectrum Characteristics Five neutron spectra representative of much of nuclear technology are displayed in Figure 1. Included are two spectra associated with differential cross sec- tion verification and detector calibrations (the 235 U fission spectrum and the one-dimensional Intermediate- Energy Standard Neutron Field (ISNF)), two fast-breeder related spectra (the FTR and EBR-II core) , and a typi- cal light-water related spectrum (LWR pressure vessel environment) . Two ordinate scales are employed for this display: log [E(|)(E)] from 1 MeV down to the cadmium cut-off and log [(E)//E] versus a linear energy scale above 1 MeV. They were chosen in order to show the full spectrum energy range below 1 MeV, along with fission spectrum components as linear slopes proportional to average spectrum energy above 1 MeV. The neutron transport calculations represented in Figure 1 are from multi- group computations normalized toJt(>(E)dE = 1 above 0.4 eV. As a first order approximation of the true shape of the spectra for this plot, the histogram spectra from the calculations were transformed to discontinuous linear segments which approximate slopes and preserve group fluxes. Fission neutrons represent the top of the energy spectrum for fission reactors and their environments. Interestingly enough, the fission neutron energy dis- tribution remains discernible in a number of these fis- sion reactor environments. Measurements with threshold detectors have established that the fission spectrum component is well preserved above ^1.5 MeV in fast metal criticals of U and Pu, and also in a zoned- core critical assembly with a spectrum typical of a fast breeder. 5 ' 11 * Reactor calculations appear to verify this spectrum characteristic showing it to be maintained in the cores of the U.S. Fast Test Reactor (FTR) and in EBR-II. Even for the LWR core shown in Figure 1 the fission spectrum component persists down to below 2 MeV. If transport calculations at LWR pressure vessels are to be believed, the spectrum there also resemble a fis- sion spectrum shape over a significant part of the MeV 157 energy range. In fact, spectrum-averaged cross sec- tions for U(n,f) and Np(n,f), truncated at thres- hold, for all of these spectra are within 10% of the fission spectrum value - see Eq. (1) below. Below the fission spectrum range a transition energy region exists between 10 keV and 1 MeV for the reactor spectra shown in Figure 1. This is the energy region where breeding occurs in fast reactors and re- laxation to a 1/E equilibrium spectrum occurs in the light-water power reactor systems. This is a diffi- cult energy range for neutron flux measurements, and for calculations as well, particularly in the decade above 10 keV. Below 10 keV the four energy decades down to thermal are characterized by a neutron slowing down spectrum, E(j)(E) 'v constant. E, MeV Fig. 1. Representative neutron spectra for materials dosimetry applications. For energies less than 1 MeV the quantity log [E«(E)] is plotted against log E. For energies greater than 1 MeV, the quantity log [»(E)//E] is plotted against E. Fission Detector Response Fission detection for the purpose of estimating neutron flux intensity and spectrum belongs to a class of integral measurements which in principle are simple to perform and interpret. As with all integral mea- surements, however, accuracy of measurement and the best possible understanding of detector response is essential if quantitative results are to be obtained. For two of the neutron fields shown in Figure 1, fis- sion detector responses will be shown in cumulative spectrum response plots, Figures 2 and 3. In such plots the integral spectrum / a(E)(E)dE, appear to- gether. Energy-dependent cross sections for the dis- play are from ENDF/B-IV. For the 235 U fission spectrum in Figure 2, the ordinate gives the fraction of the response which is above the corresponding abscissa energy. The threshold fission reactions 37 Np and 238 U are seen to provide effective complementary coverage for the bulk of the spectrum. Near the 2 MeV average energy of the fission spectrum, the U response fraction is ^ 0.8 while for 237 Np it has reached only *v< 0.5. Both threshold re- actions cut off sharply providing a well-defined re- sponse range. The wide-energy-response fission detec- tors, 239 Pu, and 235 U, follow closely the fission spectrum shape. Thus, they are good total flux monitors for fission spectra, and in fact, Pu is a good total flux monitor for all fast neutron spectra which do not have a significant spectrum component below ^ 10 keV. 1.00 0.95 0.9 Ld > O 0.8 < 0.4 0.2 10 235 U FISSION SPECTRUM I I 111 10' 10 MeV E. NEUTRON ENERGY Fig. 2. Fraction of neutron flux and fission detector response above energy E in a 235 U fission neutron spectrum. ° ^_ipL_r^ Ul 0.8 O 2 0.6 o H O < 0.1 1 ITrTTTj r-t-W-UUll 1 N N FTR CORE 239. V^FLUX a7 i rrmj i i i inn mill i \\\ i i mill i i i inn! i i N \ \ i iiinl^ a f'S.jNTiii 2 I0" s K3' I0 n 10" MeV KT E, NEUTRON ENERGY Fig. 3, Fraction of neutron flux and fission detector response above energy E in the FTR breeder core neutron spectrum. Fission detector responses for the typical breeder reactor core spectrum of FTR are plotted in Figure 3. The 238 U response cut-off remains sharp, a result which remains in need of complete verification. Some recent measurements are helpful in this regard. 15 The comple- mentary coverage of 2 3 7 Np and 238 U is even better than for fission spectra. The cumulative fraction 0.9 for 238 U corresponds to a cumulative fraction of less than 158 0.4 for 237 Np. A notable subthreshold tail for 237 Np extends to energies well below the 0.95 response frac- tion energy. This is a significant issue for spectrum characterizations and especially for checks of reactor physics calculations. The difficulty is that no satis- factory agreement exists regarding subthreshold fission for 237 Np. 16 The response functions for 235 U and 239 Pu in Figure 3 fall below the spectrum because of the ris- ing cross sections which shift the detector responses to lower energies. Almost 10% of their response lies below 2 keV where transport calculation places less than 2% of the spectrum. Concerns for neutron self- shielding begin for activation fission foils when sig- nificant response below 10 keV occurs. In spite of this and similar measurement problems, the 239 Pu(n,f) detector is still an attractive total flux monitor. The maximum departure of the flux curve from Pu response curve in Figure 3 is 15% and the average 239 Pu(n,f) cross section for this particular breeder core spectrum differs by less than 1% from the fission spectrum value. The possibility of neutron flux trans- fer which takes advantage of these small cross section differences is discussed in the next section. The graphical displays just given indicate the general features of measured fission detector responses and their significance. In order to initiate quantita- tive interpretation of these observed fission rates, calculated spectrum-averaged fission cross sections must be available. Obtaining such computed cross sec- tions is not always a straight-forward procedure be- cause of the coarseness of multigroup spectra from reactor physics computations and because of rapidly changing cross sections in some energy regions. Never- theless, a consistent set of calculated cross sections is essential for detector interpretation if arbitrary computational biases are not to render careful measure- ment worthless. Some Principles of Neutron Field Characterization The characterization of neutron fields in reactor physics criticals, and in and around the great variety of materials testing, prototype, and power reactors, depends very largely on passive integral detectors. Of these, fission detectors are the most essential. Among passive fission detectors, fission product activation is the dominant technique assisted in special situations by fission track recorders. Fission ionization chambers when applicable are important active detector comple- ments. Neutron flux spectrum or intensity information may be derived from integral detectors based on absolute detection efficiencies and absolute fission cross sec- tions, or upon a calibration carried out by exposure in a benchmark neutron field. A benchmark for calibration is a well-characterized neutron field which provides an adequate fluence of neutrons for detector calibration and which exhibits a spectrum that is relevant and better known than the spectra under study. The choice between absolute detection methods and neutron field calibration is often one of convenience and custom more than an explicit evaluation of the relative effort and the errors involved. Absolute detection methods have taken precedence historically, a fact attributable in part to the minimal development and poor availability of well-characterized neutron fields for purposes of calibration. The latter situ- ation has improved, although the recognition of it is delayed. For this reason the outline below of neutron field characterization with integral detectors will emphasize benchmark field calibration. The recognition that absolute measurements are also important is not set aside. Such complements are proper and "keep half an eye" on the claims of proponents of particular bench- mark fields. Response ratios among a set of integral detectors exposed to a benchmark field provide first and foremost a test of detector reaction cross sections over the energy range of the benchmark spectrum. If observed to predicted detector response ratios agree, field charac- terization may proceed with confidence and minimized uncertainties. If they disagree by more than the abso- lute detection errors and the assigned benchmark spec- trum uncertainties, allowed adjustment of the cross sections may be justified. Alternatively, for detectors with reliable cross sections, allowed adjustments of some benchmark spectra may be undertaken. All in all, consistency of benchmark detector response should be established (forced somewhat, if necessary) in order to achieve unambiguous spectrum characterization in the neutron field under investigation. Similarly, if the detection techniques employed in the benchmark exposure are the same as — or calibrated relative to — the techniques used in the field charac- terization exposure, observed and predicted ratios for the benchmark may be brought into agreement by ad hoc adjustment of the overall detection efficiency. Estab- lishing detection efficiencies in this way removes a number of systematic errors associated with the detec- tion scheme. Examples are errors of absolute cross section scales, activation counter calibrations, and nuclear parameters including branching ratios and fis- sion yields. This error correlation in turn allows for a wider choice of integral detector types and of acti- vation detection methods. Beyond the matter of adjusting reaction cross sections, benchmark field spectra, or detection ef- ficiencies — an issue for which agreement on systematics does not yet exist among experimenters — the benchmark exposure provides a basis for setting the accuracy and confidence for the entire spectrum characterization procedure. Often enough, it will point to inadequacies in cross sections or weaknesses in the experimental de- tection scheme. Brief Formulations Some analytic expressions for observed and derived integral detector responses are needed to move from the general assertions just given to specific applications. As noted, activation detectors are most typical for spectrum investigations, and therefore the formulations below are in terminology applicable for them. Modi- fications required for other types of integral detectors involve for the most part changes in time integrating quantities, e.g., f lux-f luence , decay constants, etc., and do not affect the principles of integral detector calibration. Spectrum and Cross Section Definitions (nv) ; (nvt) o o E t ) a(E) O(E) total energy- integrated flux and fluence, respectively. neutron spectrum normalized to unity. fraction of spectrum above neutron energy E t : detector reaction rate cross section vs. energy. a • s(E), where is the absolute cross section scale factor, and s(E) the cross section shape normalized to unity over a relevant benchmark spectrum, ty, (E) : 159 / s(E)i(j (E)dE = 1 spectrum-averaged cross section: CO a = a J s(E)ip(E)dE o(>E ) = spectrum- averaged cross section truncated at E : CO CO 5(>E t ) = O q f s(V)4>mdE/j (KE)dE (1) [g(E)^(E)] a detector response function R = G(A,t) i *<>y P ljj(>E ) a (> Ep ) W(>E J(nvt) J (5) This formulation has the advantage that it contains the cross section O (>E D ) truncated at an energy E appropriate for the reaction, and the fluence truncated at another energy appropriate for dosimetry. The factor 4>(>E p ) EJ can be rewritten as E p ) r J F iKE)dE f *(E)dE (6) E = truncation energy for defining a detector energy response range. For percentile P, the truncation energy is defined by 1 f 0(E)iKE)dE a J E P (2) where E for thii E p (P=l) Observed Reaction Rate (P=0.5) = median energy, and paper E (P=0) = 20 MeV, =0.4 eV? The departure of this quantity from unity is determined by the fraction of the spectrum between E and E , the OR energy region where the detector does not respond. Thus, for two different spectra with similar shapes above E but dissimilar between E and E , the truncated cross sections a (>E ) would be about equal and the dissimilarities would be expressed by the spectrum component ratio in eq . (6). Uncertainty in Spectrum Average Cross Section Spectrum and cross section errors are estimated in multigroup formats. The spectrum-average cross section as a discrete summation, e • u(A, N, Br, Y, I, . . .) D (3) = observed activation detector disintegration rate in disintegrations per second (dps) at end of irradiation. = observed gamma counting rate after neutron field exposure in counts per second (c/s). = gamma counting efficiency. = composite factor for converting gamma coun- ting rate of a detector to disintegration rate: decay constant (A), effective number of detector atoms (N) , branching ratio (Br), fission yield (Y) , Y - ctg losses and activa- tion interference (I) . Derived Reaction Rate G(A,t) (nvt) (4) R = derived reaction rate (dps) from fluence (nvt) received in a neutron field with spectrum ip(E) . G(A,t) = activation decay rate factor. At the end of an irradiation at constant flux and duration t, G(A,t) = [1 - exp (-At)]/t. o E s. v. AE.; o. O.ll 11 o s. o 1 (7) is subject to an error propagation which must account for the normalization of the neutron spectrum: 6a oj :(^A (f\ * EU Spectral Index Calibrations t)$1 *'**' m The physical datum for spectrum characterization with integral detectors is a set of measured reaction rate ratios. Consideration of a single ratio suffices for an outline of calibration principles. For two detectors a and 8, the observed disintegration rate ratio in a neutron field under study, R /R R > obtained from a count rate ratio, D /D R , according to eq. (3), can be set equal to the reaction rate ratio according to eq. (4) in order to obtain an observed cross section ratio or spectral index: The neutron fluence greater than a given energy [(nvt) • (|;(>E )] is often used as a dosimetry para- meter. For example, E = 1 MeV) in U.S. guides for surveillance of reactor pressure vessel fluence. Since detector reactions have various minimum response energies, it is useful to express a derived reaction rate in terms of a neutron fluence greater than E Q , and a detector cross section truncated at percentile P. This can be done by expressing O in eq. (4) in terms of P using Eqs. (1 and 2): R Thus, G(A,t) N • - O (>E p ) E p ) (nvt). [e. Ml. ) [G • N] _a _^ \ 3 8 [G • N] ( (9) [a /o R ] = observed spectral index for the study spectrum. The spectral index expected on the basis of some presumed spectrum for the neutron field under study, (e.g., measured or computed, "apriori input", or derived from other integral results) , is 160 s , exp . (°q) / S ^ E ^exp (E)dE ' 'a J o (°ol / s e< E >W E)dE (10) Vp = presumed spectrum of field under study, • s(E) = energy dependent detector cross section. The double ratio of observed to expected spectral indexes (eq. 9 and 10) is a measure of the adequacy of \\) to describe the study spectrum. The departure from unity of the double ratio provides a basis for adjusting ^ exp in two energy groups corresponding to the response ranges of the two detectors. With or without the help of unfolding codes (useful when more than a few detectors are employed) , the uncertainty in the spectrum adjustment will include errors from all of the factors in eqs. (9) and (10): 6D, 6e , 6y(y, N, Br, Y, I ...), 6N, 6G, &a , & Y os The propagation of these errors, further emphasized in the common case of detector pairs with large over- lapping response ranges, can easily reduce a spectrum adjustment to one of qualitative significance. Calibration of a spectral index in a benchmark neutron field can change the uncertainty dramatically. The similarity of the benchmark and study spectrum is not of primary importance, a substantial overlap of detector response ranges in the two spectra is suffi- cient to achieve most of the accuracy improvement. An observed spectral index obtained in a benchmark field irradiation may be set in ratio to the observed spectral index from the study field. Using eq. (9), the double ratio dosimetry to benchmark is, [ VVs [ VVs [ VVb [ Wb The spectral index can be specified for the benchmark from its known spectrum, \L), . and the expression, b [D /D fi ] (°o) / s a (E >V E)dE L a 3 s x a J o tVVb ( a ol / s 3< E >V E > dE (ll) «& J o This is the observed spectral index for the study field calibrated by means of a benchmark field exposure. When this observed index is compared to the de- rived spectral index for the study spectrum, the un- certainty will involve only the gamma counting rates (c/s), the benchmark spectrum, and partially, the shapes of the detector cross sections: 6D, 6s(E) partially, 6\\) (E) . For all situations, the detector calibration by exposure to neutrons in a benchmark field removes the errors due to detector efficiency <5Ey, the dps conversion factor 6y, and absolute cross sections scale 6o o . If the detector response functions for the study and benchmark spectra, [s(E)E ), and neutron fluences above a given energy [e o ) (nvt). becomes E y°° 1- J E Q )dE E / E / E /°° 1- / \p (E)dE// ty (>E )dE E S 7 E S ° L o / o °k( >E J _b P_ a (>E ) s P • TE o ) • (nvt) 1 (13) where [iK>E ) * (nvt) ] = neutron fluence greater than E , tJj(>E ) = fraction of normalized spectrum above E , o 0(>E p ) computed spectrum-averaged cross section truncated at percentile energy E near the lower limit of the detector response range - see Eq. (1). 161 The integrals in the brackets of eq. (13) give the spectrum fraction between E and E p , an energy region where the detector does not respond. This spectrum fraction for the benchmark field will be known; for the dosimetry study spectrum, however, this fraction contri- butes an irreducible uncertainty which is common to any fluence measurement technique employing integral detectors. In multiple foil techniques, using unfolding codes for example, the corresponding problem is referred to as the energy range of poor detector coverage. The cross section ratio in eq. (13) for appropriate detectors will be near unity. The absolute cross section scale cancels as with spectral index calibra- tions, and the remaining uncertainty in flux transfer due to cross section shape errors, <5s(E), propagates more nearly on the departure from unity of the cross section ratio rather than on the ratio itself. Thus, a ± 10% cross section shape error would affect a flux transfer involving a cross section ratio of 1.1 by about ± 1%. An exception to this occurs when the detector response ranges for the benchmark and dosimetry fields are very different. In this case spectrum uncertainties, presumably dominated by the study field, will propagate into the flux transfer according to the last term of eq. (8). This term, as applied to (>E_) in eq. (13), will not involve a sum over energies below E where o. = 0. P i Fission Rate Measurement Standards Fissionable Deposits It is not difficult to achieve electronic pulse registration of a fission fragment as it emerges from a thin deposit of fissionable material with an absolute detection efficiency of better than 99.8%. Devices are simple to construct: ionization chambers and surface barrier detectors are examples. Thus, accuracy of fission rate measurements is primarily concerned with fissionable deposit mass determinations and the absorp- tion of fission fragments in the deposit. The latter problem can be made small in many cases by obtaining effective masses of thicker working deposits (e.g., 0.4 to ~ 1 mg/cm^) relative to thin deposits (e.g., ^ 0.05 mg/cm2) by means of low-geometry alpha counting or back-to-back fission counting. Well-studied and permanently maintained fissionable deposits then become the basic artifact standard for fission rate measure- ments. A set of permanently maintained reference fission- able deposits generally available for interlaboratory comparisons has been established at NBS. Character- istics of the deposits including interim errors for principle isotopic masses are given in Table I. The TABLE I. REFERENCE FISSIONABLE DEPOSITS AT NBS PRINCIPLE ISOTOPE: 239 pu " 5 u 238 U 237 K, Np NATURAL DEPLETED DEPOSIT THICKNESS (MG/CH 2 ) 0.083 0.20 0.18 0.20 0.093 ISOTOPIC PURITY (AT 2) 99.11 99.75 99.275 99.899 100 (NOMINAL) INTERIM ERROR ASSIGNMENT FOR PRINCIPLE ISOTOPE MASS (10) il.22 ±1 . 2% ±1.52 *1.5Z n.72 STANDARD DEVIATION OF INTER- LABORATORY COMPARISONS Europe (1974-76) U.0X 10.8? ±0.82 — — Vs. (1972-75) to. b'J. ±0. IX — — — ALL DEPOSITS ARE 1.27 CM DIAMETER OXIDE DEPOSITS ON POLISHED PLATINUM DISKS (19.0 MM DIA. X 0.13 MM THK) Normalized fission rate measurements at the sicma sigma facility (mol, belcium) involvinc nbs (u.s.), gfk (germany) and rcn (netherlands) each using their own fission ionization chambers to record fission rates. b informal fissionable deposit comparisons involving lasl, anl. ornl, and bcnm (geel, belgium), the first two laboratories provided results for both 239p u and 235u. imposed requirement for a systematic program of redundant mass assay, not yet complete, results in relatively high error assignments for these reference deposits. Some interlaboratory comparisons with NBS deposits have been performed. The last two lines of Table I indicate the standard deviations associated with some of these efforts: (1) a formal fission rate intercomparison involving two laboratories besides NBS; and (2) an informal grouping of results from four laboratories that have participated in joint measurements with NBS. The interlaboratory results suggest that the interim mass errors for the NBS reference deposits are conservative. Natural Neutron Fields Turning from artifact standards to well-character- ized neutron fields as a reference for fission rate measurements, a natural focus is the upper and lower bounds of fission-driven reactor systems, namely fission neutron spectra and thermal neutron distributions. The latter are commonly employed for precise intercomparisons of fissile deposits or of fission rate detectors con- taining fissile materials. In addition, very useful cross checks among isotope species may be carried out at thermal based on well-studied nuclear parameters. 1. Mass ratios among fissile isotopes, e.g. 239 Pu: 235 U: 233 U, based on thermal fission cross sections. For high confidence levels, the comparison should be carried out also with monoenergetic neutrons near 0.025 eV in order to assure that the required non-l/v correction (e.g., Westcott g-factor) is correctly known for the thermal Maxwellian employed. 2. Mass ratio between 2 3 5 U and 238 U based on abundance of 235 U in natural uranium. A certified natural uranium base material should be used (e.g., NBS-SRM 950a, Colorado Plateau ore, 0.719 + 0.0007 at.% 235 U) because the nominal natural abundance may differ slightly depending upon origin of the uranium ore. High thermal fission cross sections for fissile isotopes along with the general availability of intense thermal flux intensities also makes these fields convenient for many interlaboratory comparisons and detector perform- ance investigations, and for the mass assay of very light fissionable deposits, in the nanogram range for example. Fission spectrum neutrons are the complement of thermal neutrons in the cycle of nuclear energy genera- tion. Fission rate measurements in fission spectra are important for validating and normalizing differential fission cross sections, for measuring nuclear parameters needed for certain types of absolute fission rate detectors, and for direct neutron calibration of detectors used for reactor spectrum characterizations. The latter includes the so-called flux transfer tech- nique outlined in the previous sections. Fission-spectrum-averaged, fission cross section ratios have been particularly important for the issue of standardization and were the subject of the integral measurement paper given at the last Neutron Standards Symposium in 1970. ' Measurement results obtained since 1970 and a review of the situation regarding observation vs. prediction are summarized in Table II. Considerable progress in measurement has occurred with the advent of clean 252 Cf fission spectrum fluxes. Also, there are new 2 3 5 U fission spectrum measurements which employ large cavities and a traverse technique for establishing the contribution of wall-return neutrons 12 . Measured ratios among U, U, Np (and for 239 Pu as well but not reported here) are accurate now to about + 2% (la), and the fission rate 162 FISSION CROSS SECTION RATIOS FOR 235(1 AND " 2 Cf FISSION NEUTRONS ° f ( 238 U)/° f r 35 U) ° f ( 238 U)/c f (Np) x( 235 u) x( 252 c« x( 23S u) xC 252 cf) Observed 0.254 1 2.2% 0.266 * 1.6% 0.270 £ 4.6% 0.260 l 6% 0.26S i 4.8% ±3.2% 0.230 t 3.0% 0.240 ± 2.2% 0.238 1 3Jj% Fission chbrs. 197S (Refs. 10.11) 1972 (Ref. 3) Activation 1968 (Ref. 1) Track recorders 1967 (Ref. 1) Total spread ±1.8% Obs. /Predicted 1.06S t 2. St 1.047 ± 2.0% 1.13 1.20 1.026 ± 3.0% 1.031 1 2.2% 1.11 •1976 1972 1967-68 •Predicted values are obtained by convolution of ENDF/B-IV cross sections and evalu- ated fission spectra from Ref. 9. Spectrum uncertainties are propagated according to Eq. 8. 252 TABLE III FISSION-SPECTRUM - AVERAGED FISSION CROSS SECTIONS Isotope Observed *Obs./Pred. U235 1205 ± 2.2% 0.971 ± 2.2% U238 321 ± 2.4% 1.018 ± 2.6% Pu239 1808 ± 2.2% 1.010 ± 2.2% Np237 1332 + 2.6% 0.986 ± 2.7% ^Predicted values are obtained by convolution of ENDF/B-IV cross sections and the evaluated 252cf fission spectrum from Ref. 9. Spectrum uncertainties are propagated according to Eq. 8. component of the measurement has been subject to inter- laboratory verification — see Table I. The observed- to-predicted ratios in Table II carry uncertainties for the first time based on an evaluation of fission neutron spectrum measurements that includes uncertainty assign- ments. The spectrum uncertainty, adds little to the total uncertainty. The departures from unity for 1976 values vary between 1.026 to 1.065 with uncertainties between +2.0% to + 3.0%. These discrepancies with the ENDF/B-IV evaluation are statistically significant but still in much better agreement than has been the case in previous years when they were the cause of much concern and speculation. The relative consistency of the integral measurement since 1967 indicates that the changes have occurred primarily in the differential microscopic cross section data. Should integral ratio measurements be accepted uncritically without the associated absolute determin- ations as a systematic validation? Ratio data has been a feature of integral reaction rate measurements almost without exception; and, as has been noted on occasion, it is a feature much less common for differential microscopic data. Attention must be called, therefore, to the absolute fission-spectrum-averaged, fission cross sections that now exist. Published results for Cf spontaneous fission neutrons provide integral microscopic data, namely fission cross sections, which are independent of any other nuclear cross section and to first order are independent of any other nuclear parameter as well. Results are given in Table III. The comparison with ENDF/B-IV predicted values include a propagation of fission spectrum uncertainties so that departures from unity are a direct check of the differ- ential-microscopic fission cross sections chosen for ENDF/B-IV. The agreement is generally good. Critical Metal Spheres Closely related to fission spectra are the unreflected critical spheres of 2 U and 239 Pu metal. Lady Godiva ( 235 U) and Jezebel ( 239 Pu), constructed more than two decades ago, sustain a strong fission spectrum component, some 60% for Lady Godiva and 80% for Jezebel. The remainder of the spectrum is largely determined by inelastic scattering. Fission cross section ratios for these systems provide vital checks of inelastic para- meters presently on the evaluated files. They are also an example of integral macroscopic measurements nearly equivalent to their differential microscopic counterpart since inelastically scattered and fission neutrons are indistinguishable components in both types of experi- ments. Fission cross section ratios for the 2 3 5 U and 239 Pu critical metal spheres are presented in Table IV. Ratio measurement results listed in the upper section of the Table were obtained separately with fission chambers and with activation detectors, both calibrated by means of exposures to monoenergetic neutrons with energies close to 2 . 5 MeV. 1 "* Uncertainties for the 1967 measure- ments on Godiva and Jezebel were +2.5% (fiss. chbr.) and + 3% (activation) exclusive of fission cross section errors at the calibration energies. This is an early example neutron field calibration of fission rates undertaken in order to reduce errors. The errors given for average values, + 2.5% for all four ratios, do not include fission cross section uncertainties at the calibration energy. The actual values in the Table have been adjusted to be consistent ENDF/B-IV cross section ratios at the calibration energies. The 1976 absolute values for BIGTEN were obtained with high resolution double fission chambers and include, as do the Godiva and Jezebel results, small corrections for cavity flux pertubations. Observed-to-ENDF/B-IV predicted ratios are relatively good for the uranium systems, Lady Godiva and BIGTEN. The average deviation from unity for the four obs/pred. ratios, 1.003, 1.043 (Godiva), and 0.976, 1.031 (BIGTEN), TABLE IV. FISSION CROSS SECTION RATIOS FOR CRITICAL METAL SPHERES V 238 u>/o f ( 235 u> o f ( 238 U)/o £ (Np) 235„ (GODIVA) 239 P U (JEZEBEL) 10% 235 U-902 238 U (BIG TEN)* 235 u (GODIVA) 239 Pu (JEZEBEL) I0Z 235 U-90! 238 U (BIG TEN)* MONOENERGETIC CALIBRATION (1967)** ABSOLUTE (19 76) MONOENERGETIC CALIBRATION (1967)** ABSOLUTE (1976) FISSION CHAMBERS ACTIVATION AV. 0.166 4 0.164 0.165 12. 5Z 0.215 3 0.218, 0.217 ±2.5% 0.0372 1.0006 0.196 0.199 0.198 ±2.57. 0.2190 0.229 0.224 ±2.51 0.117 2 ±.0026 0.0372 ±1.7* 0.1172 ±2.2* OBSERVED 1.003 1.12 0.976 1.043 1.084 1.031 PRED. (ENDF/ B-IV) FIS JION SPECTRUM CAL IBRATION FISSION CHAMBERS (19 76) ACTIVATION (1967) 0.617 13.25: 0.803 ±3.21 0.14711.42 0.816 ±2.07. 0.907 tl.SI 0.510 ±1.21 OBSERVED 0.887 1.03, 0.918 0.964 0.993 1.007 PRED. (ENDF/ B-IV) *A LARGE CYLINDRICAL CRITICAL ASSEMBLY WITH A HOMOGENEOUS METAL CENTRAL REGION. THE CENTRAL SPECTRUM IS LITTLE EFFECTED BY A DISTANT REFLECTOR AND CYLINDER-SHAPED BOUNDARIES. ••DETECTORS CALIBRATED WITH MONENERGETIC NEUTRONS: 2.43 MeV (FISS. CHBR.) 2 75 MeV (ACTIVATION). <14) VALUES REPORTED IN 1967 ADJUSTED TO CORKESPOND TO ENDF/B-IV CROSS SECTION RATIOS AT THESE ENERCIES. 163 is 2.5% with a standard deviation of + 1.7. This compares with an observed-to-predicted average deviation from unity of 6.0% for Lady Godiva reported in 1970 when calculations were based on the earliest version of ENDF/B. The departures from unity for the 'Pu critical sphere are large, 1.12 and 1.084, far beyond the uncertainties of the measurements. This is indica- tive of inadequate 3 9 Pu inelastic transfer cross sections in the calculations if the detector fission cross sections and the Pu fission spectrum are correctly known. A more proper neutron field for calibration of cross section ratio measurements intended to check neutron transport calculations is the fission spectrum. It is the strong fission spectrum component in the critical metal spheres that computation transforms into the respective spectra of these systems. A fission spectrum calibration, therefore, circumvents fission cross section scale errors and fission spectrum uncertainties. The lower section of Table IV shows results of ratios based on experimental calibration with fission neutrons. For Godiva and Jezebel, this was carried out by the activation method alone and the 235 U and 239 Pu fission spectra experiments were performed in difficult geometry . ]1) However, the same results, derived in an alternative manner by taking the values based on mono- energetic calibration (upper section of Table IV) and dividing by the 1975 3 U fission spectrum ratios (Table II), do not differ by more than 4%. The BIGTEN measurements carried out in 1976 were performed with fissionable deposits related by alpha and fission comparison counting to those used in the 1975 U cavity fission spectrum measurements. The resulting spectral indexes for BIGTEN relative to the fission spectrum driving source (0.147 + 1.4% and 0.510 + 1.2%) are particularly accurate checks of neutron energy transfer for reactor physics. The observed-to-predicted ratios now show a very different pattern from those in the upper section of Table IV. The 239 Pu system now is in agreement with calculation and the uranium systems are not. The obs/pred. ratios all are uniformly lower, a result consistent with but not fully accounted for by the latest obs. /predicted ratios for fission spectra - see 1976 values in Table II. 10. 11. 12. Pinter, M. , Scholtyssek, W. , Fehsenfeld, P., Van der Camp, H. A. J., Quaadvliet, W. H. J., Fabry, A., DeLeeuw, G. and S., Cops, F. , Grundl, J., Gilliam, D. , and Eisenhauer, C. , "Interlaboratory Comparison of Absolute Fission Rate and Uranium-238 Capture Rate Measurements in the M0L-Z£ Secondary Intermediate-Energy Standard Neutron Field," Conf. on Nucl. Cross Sections and Technology, Washington, D.C. (March 1975). McElroy, W. N. and Kellogg, L. S. , "Fuels and Materials Fast-Reactor Dosimetry Data Development and Testing," Nuclear Technology 25, 180-223 (1975). McElroy, W. N. , ed. ILRR Progress Reports, Nos. 9-11, HEDL-TME 75-130 (1973-1977). Vlasov, M. , "IAEA Program on Benchmark Neutron Fields Applications for Reactor Dosimetry," INDC(SEC)-54/L+D0S. (July, 1976). Grundl, J. A., Gilliam, D. M. , Dudey, N. D. and Popek, R. J., "Measurement of Absolute Fission Rates," Nuclear Technology 2J5, 237-257 (1975). Grundl, J., Eisenhauer, C. , "Benchmark Neutron Fields for Reactor Dosimetry," Consultants Meeting on Integral Cross Section Measurements in Standard Neutron Fields for Reactor Dosimetry, IAEA (Nov. 1976) . Manuscript copies available from authors at National Bureau of Standards. Heaton II, H. T. , Grundl, J. A., Spiegel Jr., V. Gilliam, D. M. , and Eisenhauer, C, "Absolute 235 U Fission Cross Section for 2 Cf Spontaneous Fission Neutrons," Conf. on Nucl. Cross Sections and Technology, Washington, D.C. (1975). Gilliam, D. M. , Eisenhauer, C. , Heaton II, H. T., and Grundl, J. A., "Fission Cross SEction Ratios in the 252 Cf Neutron Spectrum ( 235 U: 23e U: 239 Pu: 237 Np)," Conf. on Nucl. Cross Sections and Tech- nology, Washington, D.C. (1975). Fabry, A., Grundl, J., and Eisenhauer, C, "Funda- mental Integral Cross Section Ratio Measurements in the Thermal-Neutron-Induced Uranium-235 Fission Neutron Spectrum," Conf. on Nucl. Cross Sections and Technology, Washington, D.C. (March, 1975). It is not the purpose of this paper to evaluate 13. the implications of the results displayed in Table IV. It is a purpose however to illustrate how reference neutron field calibrations of experiments designed to check reactor spectrum computations may reveal features not otherwise apparent. The critical metal spheres are particularly apt for this purpose because they involve 14. little or no complexities of geometry. This ambiguity for many reactor physics criticals can, and probably has, obscured the desirability of reference neutron field calibrations. Grundl, J. A., and Eisenhauer, C. M. , "Fission Spectrum Neutrons for Cross Section Validation and Neutron Flux Transfer," Conf. on Nucl. Cross Sections and Technology, Washington, D.C. (March 1975). Grundl, J. A., and Hansen, G. E. , "Measurement of Average Cross Section Ratios in Fundamental Fast- Neutron Spectra," in Nuclear Data for Reactors, Vol. I, pp. 321-336, IAEA, Vienna (1967). Also see Nucl. Sci. and Eng., 8_, 598-607 (1960). References 1. Grundl, J. A. "Fission-Neutron Spectra: Macroscopic and Integral Results," Proc. Symp. on Neutron Standards and Flux Normalization, P. 417, Conf- 701002, ANL (August, 1971). 2. Taschek, R. F. , ibid. 1. 3. Grundl, J. A. "Brief Review of Integral Measure- ments with Fission Spectrum Neutrons," and "Measure- 15. 16. ment of Av. Fiss. Cross-Section Ratio, 235^/2 38 U For U and Pu Fission Neutrons;" Prompt Fission Neutron Spectra, Panel Proceedings Series, IAEA (1972). 164 Behrens, J. W. , Carlson, G. W. , and Bauer, R. W. , "Neutron-Induced Fission Cross Sections of U, 231 *U, 23e U and 238 U with Respect to 235 U," Conf. on Nucl. Cross Sections and Technology, Washington, D.C. (1975). See also M. S. Coates, D. B. Gayther and N. J. Pattenden, "A Measurement of the B U7 U Fission Cross Section Ratio," same conference. Private communication from ENDF/B-IV cross section evaluators (1975). UTILITY AND USE OF NEUTRON CAPTURE CROSS SECTION STANDARDS AND THE STATUS OF THE Au(n, v) STANDARD A. Paulsen Central Bureau for Nuclear Measurements B-2440 Geel, Belgium The main application of a neutron capture cross section standard should be found in ratio mea- surements using the prompt y-ray detection method. A review of neutron capture cross section measurements in the last six years shows that the Au(n,y) standard is increasingly used for this purpose. But the majority of all measurements is still based on other normalization methods than ratio measurements, although the accuracy established for the Au(n,y) cross section be- low 3 MeV competes now well with that of other normalization methods. On the other hand this accuracy has scarcely reached the accuracy of necessary corrections associated with prompt y-ray detection measurements below 3 MeV neutron energy and is completely insufficient above 3 MeV. Below 200 keV the cross section fluctuations due to level statistics in the compound sys- tem ''°Au are seriously disturbing measurements aiming at high accuracy and high resolution. (Review ; neutron capture cross section; measurement; normalization; reference standard; accu- racy; Au(n,y) cross section; cross section fluctuations) Introduction A standard cross section for neutron capture mea- surements is of quite different importance in the three neutron energy regions : -thermal and resonance region -unresolved resonance region -high energy region. Thermal and resonance region In the thermal and resonance region many thermal cross sections and resonance parameters (e. g. for 55 59 103 , 107,109 1]3,115 T 197 a . , Mn, Co, Rh, Ag, In, Au) ha- ve been determined to a rather high degree of accura- cy 1. These data have been often used to normalize neu- tron capture cross section measurements in the ther- mal or resonance region. Therefore, with good rea- sons these thermal capture cross sections and reso- nance parameters could be named "capture standards". Something similar is valid for certain capture reso- nance integrals (e. g. for Mn and A u) which have been used as reference standards for the measurement of resonance integrals. The relative large amount of existing data is responsible for today's relative small interest in thermal and resonance capture standards. Unresolved resonance region A standard cross section in its unresolved reso- nance region is in principle of no use unless the mea- surement averages over a sufficiently broad energy interval to arrive at a smooth cross section shape. This implies that one capture reaction chosen as a re- ference standard can never be suited for the whole neutron energy range. High energy region Due to the required smoothness of the cross sec- tion curve a capture standard cross section is of suffi- cient versatility in application only in the high energy region. Therefore the main region of interest in the Au(n,y) reaction, which is up to now the only one re- cognized as capture standard^, can be only in the energy region above about 100 keV. The Measurement of Cross Section Ratios In the thin sample approximation the reaction cross section a can be calculated from the number of counts C of a capture detector with efficiency e, the number of sample atoms N and the neutron fluence density $ according to Fig. 1. In a similar way the neutron flu- ence density can be deduced from a known standard c ross section a . C : number of counts e : detection efficiency 0* = N : number of atoms : neutron fluence density I Co index 'o' corresponds £o ' No ' Oo *° standard reaction J . C • Eq ' No . standard used to o = — — — On -0 Cn'EN "° measure neutron fluence if E=E & CN Co C N standard used to perform ratio measurement (e is unknown, 4> can not be evaluated) Fig. 1 : Explanation of the term "ratio measurement". 165 If the standard reaction requires another detector than the capture reaction to be measured, then the efficiencies will not cancel from the equation for a . In this situation the measurement of the neutron fluence density could also be done with any other suitable stan- dard which has not at all to be a capture cross section. In this case all detector efficiencies have to be known and the neutron fluence density can be calculated. However, if the standard reaction can be detected in completely the same way than the capture reaction under study, then due to e= G the standard is used for a ratio measurement for which no efficiency has to be known and $ can not be evaluated. A neutron capture standard is therefore characterized by the fact that it permits capture cross section determinations by ratio measurements. The Prompt Gamma-Ray Detection Method Neglecting here measurement techniques based on neutron absorption (shell transmission and pile reacti- vity measurements) only the detection of the prompt capture y-rays fulfils approximately the condition for ratio measurements, whereas the activation method is faced with characteristic radiations changing comple- tely from radioisotope to radioisotope. For the prompt y-ray detection method the condition e= e is only ap- proximately fulfilled because the capture v-ray spectra (and possibly also the angular distributions) are chan- ging from isotope to isotope. This has to be corrected for. It is a principal question whether the prompt y-ray detection method measures the full capture cross sec- tion also at higher neutron energies. To discriminate against y-rays from inelastic scattering one is obliged to use only the upper part (E n < Ey) of the capture spectra. Therefore cascade passages through unbound states could be missed. For this reason it is interes- ting to compare prompt y-ray detection results CT„ amma with activation results CJ ac ti v . • This is done for Au(n,y) in the 1 to 3 Me V energy range-' ' ^' -* in Fig. 2 and for several reactions at 14 MeV in Fig. 3. For the latter only activation results were taken into account which passed a critical examination in view of corrections for scattered neutrons of degraded energy. Within to- day's experimental accuracy there is no indication that capture y-ray cascades are passing partially through unbound states. II so 40 30- 20 tt fl* ,97 Au(n,y)' 98 Au *f > M • Pbnitz i A Lindner et at o Paulsen et al. 9 10 20 ~r 3 E n (MeV) ► Fig. 2 : Comparison of prompt y -detection (full symbols) and activation (open symbols) results for Au(n,y). Fig. 3 : Ratio of activation results CJ ac tiv to prompt y-ray detection results CJg amma at ]4 MeV. Actually used Normalization Methods of Neutron Capture Measurements A review was made concerning the used standards and normalization methods for capture cross section measurements in the neutron energy range above 20 keV by the prompt v-ray detection technique. This re- view covers all measurements for which numerical re- sults have been compiled at one of the four cooperating Neutron Data Centers between 1971 and 1976. The re- sult is shown in Table ] where the measurements are counted per studied isotope or element (Au excluded). The number of ratio measurements relative to gold (at least at one neutron energy) is increasing from the 71/73 to the 74/76 period, indicating that the gold stan- dard is now indeed increasingly used. All the ratio measurements relative to gold were performed at elec- trostatic accelerators, whereas at linear accelerators a normalization in the eV range by saturated resonan- ces is preferred. However, an additional normalization at the high energy end (E n > 100 keV) by a gold ratio measurement could probably improve the accuracy of the results produced at linear accelerators. It has to be kept in mind that these ratio measurements can be renormalized at any time. 166 Table 1: Normalization Methods of Neutron Capture Measurements* Period Separ Determ of e: and 4> Satur Reson •♦rd Therm Cross Sect Ratio Measurements Total Au In Ag 1971-1973 1974-1976 30 7 15 23 4 7 10 23 10 3 72 60 » per isotope or element prompt v-ray detection method only tor neutron energies above 20 keV only Comparison of Normalization Methods Most probably the preference given to other nor- malization methods than the gold ratio measurement is due to the fact that experimentalists believe to assess better accuracies that way. Therefore a comparison is made between the estimated accuracy of the Au(n,y) cross section and that one of other normalization me- thods used for measurements above 20 keV neutron energy. This comparison is shown in Table 2. The un- certainties are estimated at the confidence level of one standard deviation and contain the principal uncertain- ty of the neutron fluence and the y-ray detector effi- ciency. The total error was obtained from the indivi- dual uncertainty components by quadratic summation. The table contains further the upper energy limit for the validity of the quoted error and the reference to the corresponding publication. Table 2: Uncertainties of Normalization Methods Method e a bs or (e $)abs *abs Or 4>re4 Capture Standard Total Uncertainty Method AccRef Energy Limil Method Ace Ret Energy Limit Cross Section Ace Ref Energy Limit Separ Determ of e and ^adioac Sources Ip.YlRe- actions 2*1271 10*1161 IMeV IS MeV Asspart Long C H (n,n) n5 U(n,f, 2* (91 27. 00) 3* (91 47. 112) IS MeV 1 MeV 15 MeV 6 MeV - - - 3-107. Determ of (€■♦) Satur Reso- nances 1-2* 111,22) 100 eV 6 Liln,t) "B ln,a) "Bln.al • H ln,n) ^hjln.f) 47. 03) 37. 113) 47. 113) 37. (W 01 MeV 01 MeV 1 MeV 6 MeV - - - 3 - 57. Normaliz at Therm Energy - - Therm Cross Section 1-27. ID 0025eV 3- 57. Ratio Measurem with Gold Au ln,v) Cross Section 47.129) 57. (18) 02-35 MeV 0.1-3 MeV 4- 5* The comparison of these uncertainties with today's accuracy of the gold capture cross section (last line of Table 2) is not at all unfavourable for the latter. But it is also not yet convincing to give preference to ratio measurements with gold. However, the accuracy of the gold capture cross section could finally surpass that one of the competing normalization methods, be- cause many measurements and all experimental mea- suring techniques for neutron capture can principally contribute to it. Up to now measurements carried out with high ef- fort and using a normalization method according to the first two lines of Table 2 can be still superior to gold ratio measurements with respect to resulting accura- cy. Here the influence of the simultaneously progres- sing standards 6Li( n ,t) ■ 10 B(n,a) and 235 U(n, f) has to be mentioned. Of course, this comparison is only valid for measurements in or ranging up to the indicated energy range for the Au(n, y) cross section. Comparable measurements belonging to the third line of Table 2 are nearly not existing : there is no experimental tech- nique to cover neutron energies from thermal energy up to several hundreds of keV (slowing-down time mea- surements just reach 100 keV).The relative simplicity of ratio measurements and the readjustibility of their results justify further efforts for an amelioration of the accuracy of the gold capture cross section. Uncertainties of Ratio Measurements There are certain sources of error inherent to all capture cross section measurements performed by the prompt y-ray detection technique. It is interesting to compare the accuracy of the gold capture cross sec- tion with these 'inherent' uncertainties. This is done in Table 3 by comparing uncertainties which have been quoted in literature as average values. Table 3: Uncertainties of Ratio Measurements Source of Uncertainty Pulse Height Weighting Techn Moxon-Rae Detectors Liquid Scint Tanks Spectrometers NallTD, Ge(Li), Pair Variation of y-foy Spectrum ♦17. Ref 8,15 13% Ref 19,20 ±57. Ref 21,22 ♦107. Ref 16,17 Multiple Scattering ± 2-37. Ref 22, 23 Neutron Scatt into y -detector tor>-W 2 i 0.5 -37. Ret 24, 2S Background Subtraction ± 1-37. Ref 23, 24 Au (n,y) Cross Section ± 4-57. Ref 18,29 There is a detector-dependent error due to varia- tions of the capture v-ray spectra. This error is mini- mized for detectors having the least spectrometric response and is a maximum for the true spectrometers. The pulse height weighting technique in conjunction with C^ F^ o r C^D^ y-detectors results in a remarkable inde- pendence from the shape ot tne y-spectrum. It is the clear predominance of the Compton y- scattering pro- cess in these detectors with its smooth and relative weak energy dependence which is responsible for this independence as well as for a good reliability of effi- ciency calculations. This type of detector should be al- so applicable in the MeV neutron energy range because the spectrum fraction above threshold can be well cal- culated. For the pulse height weighting technique and for Moxon-Rae detectors there are difficulties in the data evaluation if the sample is a compound of elements or a mixture of isotopes with unknown capture cross sections and strongly differing neutron binding ener- gies. Uncertainties originating from these difficulties have not been considered in Table 3. For tank measure- ments the spectrum fraction uncertainty is astonishing- ly high due to the relative high thresholds which have to be used for background reasons with such detectors. Cross section measurements performed with spectro- meters are mainly done in the high MeV range where the accuracy oftheAu(n,y) cross section is anyway 167 not yet defined. Variations of the y-ray angular distri- butions with increasing neutron energy seem to be ne- gligibly small . Uncertainties due to corrections for multiple scat- tering, response of v-detector to scattered neutrons and general background can vary considerably and are depending on several experimental conditions. The ave- rage uncertainties of these corrections quoted in Table 3 are smaller than the normalization and y-detection uncertainties. In conclusion Table 3 demonstrates that further efforts for an improvement of the accuracy of the gold standard seem to be justified. Status of Au(n,y) Standard Below 0. 2 MeV Recent measurements' ' 'have confirmed that considerable cross section fluctuations are extending 200 Fig. 4 : The relative neutron energy resolution neces - sary for the observation of +_ 3 and +5 % re- lative cross section fluctuation for the neutron capture in Au and Ta (ace. to ref. 28). up to 200 keV. These fluctuations are the high energy extension of the unresolved resonance region and are caused by the statistical distribution of the capturing levels in the '°'Au + n system. The amount of obser- vable fluctuations does not only depend on the neutron energy but also on the energy resolution. Fig. 4 shows the relative neutron energy resolution which would permit the observation of + 3 and + 5 % deviations for capture in gold and tantalum as a function of neutron energy. These data have been produced'-" by Monte Carlo calculation assuming level distances and redu- ced level widths according to a Wigner and a Porter- Thomas distribution, respectively. Fig. 4 demonstra- tes that due to the higher level density Ta(n,y) would be in this respect a much better standard in the 50 to 200 keV energy range. As these fluctuations could strongly influence high resolution ratio measurements ENDF/B-V^ will no longer support the gold standard below 200 keV. Furtheron one can read from Fig. 4 that gold activation measurements with Sb-Be neutron sources at 22. 8+1.3 keV can deviate up to 5 % from the smooth average cross section curve. B etween 0. 2 and 3 MeV In this energy region the accuracy of the Au(n , y) cross section is estimated to be + 5 % 18 for the ENDF/B-IV data or + 4 % 2 ^ for the ENDF/B-V data in preparation. Above 3 MeV In the energy region above 3 MeV the available Au(n,y) cross sections are very scarce. Fig. 5 shows all available cross section data together with the ENDF/B-IV curve and the proposed ENDF/B-V data. It is in this energy region from 3 to 14 MeV where new Au(n, y) cross section measurements would be ve- ry valuable in view of checking the validity of interpo- lations by model calculations. io-| ,97 Au(n, Y ) ,98 Au a Pbnitz 3 ■ Lindner et al/" 7 Paulsen et al o Miskel et al . M » Johnsrud et al ' o Drake et al 32 • Schwerer et al ENDF/B - IV ENDF/B-V (Preliminary) ■V (Preliminary) I -0--I-- 9 10 n E n (MeV) ► 12 13 14 15 Fig. 5 : Cross section data for the neutron capture in Au above 3 MeV. Conclusions Since the introduction of a capture standard cross section is aiming at the performance of ratio measu- rements, more measurements of that type should be done, especially at the high energy limit of linear ac- celerator facilities. Results of ratio measurements can be easily renormalized afterwards. But already 168 today the accuracy of the gold standard can well com- pete with other normalization methods in the 0. 2 to 3 MeV energy range. Also the problem of the influen- ce of cross section fluctuations seems to be clarified. With respect to these cross section fluctuations the Ta(n,y) reaction is a much more suited standard in the 50 to 200 keV energy range. Above 3. 5 MeV neu- tron energy the existing cross section data are com- pletely insufficient for any reference purpose. This reflects of course the data needs for fission reactors. Further measurements in this energy range could help to extend considerably the applicability of this standard for developments in fusion technology and nuclear safeguards . References S. F. Mughabghab and D. I. Garber, Neutron Cross Sections, Volume I, Resonance Parameters , Re- port BNL 325 Third Edition (1973) Proc. Neutron Standard Reference Data, Vienna 1972, Recommendations p. 360 17. M. Stamatelatos , B. Lawergren and L. J. Lidof sky . Nucl. Sci. Eng. 5J_ (1973) 113 197 18. A. B. Smith, The Contemporary Status of the Au (n,y) Cross Section, status report to the 8th INDC Meeting, Vienna October 1975 19. M. C. Moxon, thesis London University 1968 20. R. L. Macklin, J. H. Gibbons and T. Inada.Nucl. Phys. 43 (1963) 353 21. D.Kompe, Nucl. Phys. A133 (1969) 513 22. R. E. Chrien, Neutron Capture Cross SectionMea- surement Techniques . Proc. Nucl. Cross Sections and Technology, Washington D. C. 1 975 , NBS Spec. Publ. 425, Vol. I (1975) 139 23. R. C. Block and R. W. Hockenbury. The Precision of Tank Capture Cross Section Measurements, Proc. Neutron Standards and Flux Normalization Argonne 1970, AEC Symposium Series 2_3 (1971) 355 3. W. P. Ponitz, Nucl. Sci. Eng. 57 (1975) 300 4. M.Lindner, R. J. Nagle and J. H. Landrum , Nucl. Sci. Eng. 59 (1976) 381 24. M. C. Moxon, D. B. Gayther and M. G. Sowerby, NEANDC Topical Conf. on Capture Cross Section Measurements, Report AERE-R 8082, Harwell 1975, p. 12 A. Paulsen, R. Widera and H. Liskien , Atomkernener- gie 26 (1975) 80 25. R. L. Macklin and B. J. Allen, Nucl. Instr. Meth. 91 (1971) 565 6. M. Wagner and H. Warhanek, to be published 26 7. M.Budnar, F.Cvelbar, A.Kikar, R.Martincic, M. Potokar and V. Ivkovic , Measurements of 14 MeV Neutron Radiative Capture y- R ay Spectra and 27. Integrated Cross Sections in Sc , Y, Pr, and Ho, Report INDC(YUG)-5/L (1977) 28 8. C. Le Rigoleur, A.Arnaud, Proc. NEACRP/NEANDC Specialists Meeting Karlsruhe May 1973, Report KFK 2046/NEANDC-U-98, p. 33 29. 9. A.Paulsen, H. Liskien and M. Cosack, Nucl. Instr. Meth. 1_0J3 (1972) 103 30. 10. J. L. Leroy, J. L. Huet and J. Gentil, Nucl. Instr. Meth. 88 (1970) 1 31. 11. R. L. Macklin, J. Halperin and R. R. Winters , Phys. Rev. CU (1975) 1270 32. 12. M.Lindner, R. J. Nagle and J. H. Landrum , Report UCRL-75838 (1974) 33. 13. H. Liskien, Neutron Standards and their Applica- tion, Proc. Intern. Conf. on the Interaction of Neu- 34. trons with Nuclei, Lowell 1976, p. 1110 14. V.A.Konshin, Review and Evaluation of the 235 U Fission Cross Section in the 0. 1 keV to 15 MeV Energy Range, Report INDC(CCP) -94/U (1976) D.M.Drake, E. D. A rthur and I. Halpern , Proc. Sec. Intern. Symp. on Neutron Capture Gamma-Ray Spectroscopy, Petten 1974, p. 227 C. Le Rigoleur, A. Arnaud and J. Taste, Report CEA-N-1662 (1973) H. Liskien and H. Weigmann, Annals Nucl. Energy in press S. F. Mughabghab, Informal Report BNL-NCS-2 1 774 (1976) J.A.Miskel, K. V. Marsh. M. Lindner and R. J. Nagle, Phys. Rev. L28 (1962) 2717 A. E. Johnsrud, M. G. Silbert and H. G. Barshall, Phys. Rev. JJ_6 (1959) 927 D. Drake, I. Bergqvist and D. K. McDaniels , Phys. Lett. 36B (1971) 557 O.Schwerer, M. Winkler-Rohat sch , H. Warhanek and G. Winkler, Nucl. Phys. A264 (1976) 105 R. M. Lessler, WRENDA 76/77 World Request List for Nuclear Data, Report INDC(SEC) -55 /URSF (1976) 15. J. B. Czirr and M. L. Stelts, Nucl. Sci. Eng. 52 (1973) 299 16. I. Bergqvist, D. M. Drake and D. K. McDaniels , Nucl. Phys. A191 (1972) 641 169 REMARKS ON THE 2200 m/s and 20° C MAXWELLIAN NEUTRON DATA FOR U-233, U-235, Pu-239 and Pu-241 + H.D. Lemmel Nuclear Data Section International Atomic Energy Agency A-1011 Vienna, Austria Abstract ; Attention is drawn to the still existing systematic discrepancy between experimental cross-sections for 2200 m/s neutrons and those for a 20 C Maxwellian neutron spectrum for U-235 and U-233. (Keywords: Neutron nuclear data evaluation. Fission standards. U-233, U-235, Pu-239, Pu-241, thermal neutron cross-sections, fission-neutron yields,, Pu-239 half-life.) Introduction In 1965 [1], 1969 [2] and 1975 [3] the IAEA Nuclear Data Section, in cooperation with external specialists, published consistent sets of recommended best values of the thermal neutron data of the main fissile nuclides. The method of evaluation was a multi-parameter least-squares fit- of all available experimental data, after reviewing and, where feasible, reassessing the authors' quoted values and errors, usually in consult- ation with the authors or other experts. In the present paper I would like to demonstrate which discrepancies still exist for these important data. existed in two distinct groups around 3.7 and around 3.8 respectively. We then concluded that , considering the Chalk River a value, a high value of p (Cf-252) should be correct rather than a lower one. - Soon after, this conclusion turned out to be wrong. This wrong prediction of 1969 was a good illustration of the limited value of a least— squares fit, when systematic discrepancies are present, and in particular when several different sources of discrepancies interfere within the same fit. Today, this limitation is no longer as serious as previously, but it seems advisable to remember that the results of a least-squares fit are subject to certain limitations which I shall discuss briefly. Discrepancies Treatment of correlated data Although the thermal neutron data of the fissile nuclides are basic parameters and reference standards for the data at higher energies as well as for data of other nuclides, it has not yet been possible to issue a final set of recommended values, due to the existence of systematic uncertainties. In 1965 the situation looked still fine. The accuracy of experimental data was not yet as good as today, and possible discrepancies did not significant- ly exceed the experimental errors. In 1969 it became obvious that there were, among the experimental data, some unexplained discrepancies. For example, experimental U-235 fission cross-section data were discrepant; data for the neutron-yield per fission, v, were discrepant; the mean energy values of the fission neutron spectra, which influence the 13 values, were uncertain; the Westcott g-factors and their temperature dependence were uncertain; half- life values, which influence the fission cross-section values, were discrepant; and some of them, for example the experimental values of the Pu-239 half-life, were consistent but turned out later to be all together wrong. In particular the neutron yield data, v and n , appeared to be inconsistent with the Chalk River irradiation experiments which yielded a capture-to- fission cross-section ratio a with high accuracy. The equation 1 + 8 = v t /$ (where the sign denotes the 20 C Maxwellian spectrum average) was not fulfilled within the error limits of the experimental data. At that time, the weakest parameter within this equation seemed to be v, since all I) values had been measured relative to the spontaneous T3 of Cf-252, of which experimental values The thermal cross-sections and neutron yield data for one fissile isotope are all interrelated, and the data of different fissile isotopes are also inter- related due to ratio measurements. Thus there are about 200 interrelated experimental input data, there- of about 50 independent variables, which are simultaneously treated in the least-squares fit. The main problem in such a fit is the treatment of experimental input data which have correlated error sources. It would lead to wrong results if such cor- related data would be treated in the fit in the same way as uncorrelated data. Several methods were used to take care of such correlations. In 1965 a mathematical procedure [1] was used by which certain types of data correlations can be re- presented in the fit by appropriate down-weighting of the correlated input data and their ratios. This method applies when, e.g., the fission cross-section is measured in one experiment for three nuclides; the three results are then correlated due to the uncertain- ties of all those corrections which they have in common. For a number of input data this method was still applied, but in 1975 an additional, more flexible method was introduced, by which error sources common to several input data are formulated in the fit as independent parameters. For example, most measurements of the fission- neutron yield p depend on assumptions about the energy- spectrum of fission neutrons. Therefore, the mean energy values "E of the fission neutrons of each of the nuclides considered, were treated as independent para- meters in the fit, thus adjusting automatically the experimental T> data for changes in the values of the mean fission neutron energies "S. t Since the author was unable to attend, this paper was summarized by 170 R. Leonard, Jr. at the Symposium. Another example are the fission cross-section measurements which mostly depend on half-life values, when the sample assay was done by alpha-counting. Therefore, the half-lives for the nuclides considered were entered as independent parameters in the fit, thus adjusting automatically the experimental fission cross-section data for changes in the half-life values. A special problem is the treatment of the Westcott g-factors, which relate the cross-sections for mono- energetic neutrons at the reference velocity of 2200 m/s to the cross-sections for a thermal Maxwellian neutron spectrum. The g-factors are strongly dependent on the exact cross-section curve shape a (E) in the thermal neutron energy range, in particular in the not very well known range below 0.0253 eV [3]» In the fit, experimental values of 2200 m/s cross- sections, of Maxwellian average cross-sections correct- ed to a spectrum temperature of 20 C, and g-factor values as obtained from curve shape studies, are all entered as parameters to be fitted. However, these data show a correlation which is difficult to formulate. Firstly, the absorption cross-section curve o (E) is derived from measurements of the total cross-section. Thus, the g-factor computed from the curve o (E) is dependent on the scattering cross-section assumed. Secondly, if in the course of the least-squares fitting procedure, the value of the absorption cross-section d (0.0253 eV) is adjusted, then this adjustment will affect the cross-section curve-shape a (E) and, as a consequence, will change the g-factor g . Consequently, the absorption cross-section, the scattering cross-section, and the g-factor for absorp- tion are correlated in the fit. Whereas in the earlier IAEA evaluations this correlation had been ignored, it had been considered, at least in a good approximation, in the 1975 evaluation. These examples are mentioned here, in order to illustrate that the more important complications in- herent in the least-squares fitting method have been taken care of, and that the results of the least-squares fit should therefore be reliable. The improved fitting procedure by B. R. Leonard Jr. et al et al deviate partly by more than a standard deviation from the 1975 IAEA recommended values. However, this deviation is only an expression of the still existing disturbing discrepancies among the existing experimen- tal data. The last (?) discrepancy to be solved Despite of the more powerful fitting method developed by B.R. Leonard, the less powerful IAEA fit still serves a good purpose, because it may give us a hint about the source of the, perhaps last, disturb- ing discrepancy which remains to be solved for the ■ thermal fission data. In 1969 the results of the least-squares fit were somewhat unsatisfactory, since a number of different sources of discrepancies were interfering, thus making the interpretation of the results difficult. It was obvious that there were disturbing discrepancies among the experimental data, but it was not evident which of the many input data may be responsible for the dis- crepancies. In the 1975 evaluation, the situation had so much improved, that one could well localize the origin of the discrepancies encountered. The one of the discre- pancies has meanwhile found its solution, after a new more reliable value of the Pu-239 half-life has been established in a number of parallel experiments [5]» Hence it seems that there is only one discrepancy left which will hopefully be the last one: the system- atic discrepancy between 2200 m/s cross-sections and thermal Maxwellian cross-sections for the uranium isotopes. This is illustrated in Table 1. The data given in Table 1 were obtained from least-squares fits of exactly the same input data as in 1975 [3] except for two items: a lower value of the Pu-239 half-life (24130 + 50 years) was tentative- ly adopted, which seems to be the consensus of several independent experiments being performed at present [5]? and the final values of J.R. Smith's 9 experiment [6] were used instead of the somewhat lower values reported in 1974 [7]. Nevertheless, this way of a least-squares analysis leaves the evaluator in a difficult situation. If the g-factors are adjusted in the fit, as a result of simultaneous fitting of monoenergetic and spectrum average cross-sections, the evaluator is faced with the problem , that he has to construct, subsequently, a best cross-section curve which exactly reproduces the g-factor resulting from the least-squares fit. Obvious- ly, this is a tedious, if not impossible job. B.R. Leonard Jr. et al, therefore, achieved a con- siderable progress in the evaluation method of the thermal cross-sections, by considering in the least squares fit not only the monoenergetic and thermal Maxwellian experimental cross-sections but also the entire low-energy cross-section curves o(E) in a for- mulation using multilevel resonance parameters [4]. The results for U-235 were published in February 1976, and I am not informed at the moment when writing this, whether this work continues for the other fissile nuclides. But it would certainly be most useful, if this work by B.R. Leonard et al could be continued as a simultaneous fit for the four main fissile nuclides. The recommended U-235 data obtained by B.R. Leonard When trying to localize the origin of the dis- crepancies encountered, several least-squares fits were made with different subsets of the experimental data, and the internal consistency of each subset was studied. As a result it became obvious that one can devide the experimental data into two subsets, where each subset has a very good internal consistency, but both subsets are inconsistent with each other. The one subset comprises all monoenergetic 2200 m/s data together with the v data, see column 1 Table 1. The other subset comprises all data measured in a thermal Maxwellian neutron spectrum, see column 3 Table 1. Both of these subsets show an internal consistency which is striking and by far better than can be sta- tistically expected. (Within each of both subsets only 4% of the input data deviate from the fitted value by more than the quoted, or re-assessed, experimental error. ) Column 2 in Table 1 shows the thermal Maxwellian data as deduced from the 2200 m/s data as given in column 1 using g-factor values as determined from curve-shape studies (see Table 2 column 1). 171 1. 2. 3. 4. Pit of 20°C Maxwellian Pit of Difference experimental data deduced experimental between 2200 m/s data from col 1. with 20°C Maxwellian columns incl. v data 20°C g-factors of Table 2 ool.1. data 2. and 3. U-233 o a 573.8 + 1.8 573.1 + 2.1 574.3 + 3.7 +0.8 °f 533.2 + 3.0 531.4 + 3.1 526.9 + 3.4 zkl (0.9/o) a Y 40.6 +2.5 41.7 + 2.7 47.4 + 0.4 ±5*1 d3/ ) a O.O76 + 0.005 0.0/9 + 0.005 0.090 + 0.001 +0.012 (15/.) 7 2.294 _+, 0.009 2.289 j- 0.010 2.296 J; 0.019 +0.007 *t 2.469 + 0.008 U-235 o a 680.0 j- 1.8 665.1 ± 2.0 663.6 j; 4.5 -1.5 d f 583. 1+ 1.9 574.9 + 2.0 566.0 ± 3.8 :&2 (1.5/c) °v 91.9 + 2.3 90.3 ± 2.3 97.5 ± 0.8 ±L1 (8/0) a O.156 + 0.004 0.157 ± 0.004 0.172 ± 0.001 +0.015 (10/0) 7 2.079 ± 0.003 2.077 +. 0.003 2.090 ± 0.016 +0.013 *t 2.404 + 0.006 Pu-239 a a 1014.1 +4.3 1091.7 ± 7.0 1095.5 + 7.5 +3.8 °f 748.1 ± 2 « 8 787.8 +, 3.7 787.8 ± 5.3 0. ° Y 265.9 ± 4.1 304.O + 7.1 307.7 ± 2.6 +3.7 a 0.355 + 0.006 O.386 + 0.010 0.391 ± 0.002 +0.005 7 2.110 + 0.003 2.064 ± 0.014 2.060 ± 0.019 -0.004 *>t 2.860 j; 0.009 Pu-241 a 1377. + 13. 1431. + 14. 1432. ± 13. + 1. °f 1023. ± 11. 1069. ± 13. 1059. i 10. -10. 355. + 8. 362. jt 11. 373. ± 3. + 11. a 0.347 ± 0.009 0.339 ± 0.012 0.352 ± 0.008 +0.013 (4/o) ? 2.165 + 0.013 2.177 ± 0.019 2.192 ± 0.032 +0.015 V t 2.915 + 0.010 Cf-252 v 3.740 + 0.009 ._ _ Table 1. Discrepancies between experimental 2200 m/s cross-sections and experimental 20 C Maxwellian cross-sections. When comparing the directly measured 20 C Maxwel- lian data with those derived from monoenergetic data and g-factors, one finds: Consistent are also the absorption cross-sections, where the discrepancy contributions from capture resp. fission seem to cancel. 1. Discrepancies exist for the uranium isotopes and, to a much lesser extent, for Pu-241. The data for Pu-239 are consistent. (Some authors, e.g. [4] t quote that the data for Pu-239 are also inconsistent. However, this is correct only for the Pu-239/U-235 cross-section ratios, where however the discrepan- cies seem to come rather from U-235 than from Pu-239. ) 2. For the uranium isotopes, systematic discrepancies exist for the fission and capture cross-sections and their ratio a, but not for the neutron yield 9 • 3. The order of magnitude of the discrepancies is 1-2/ for the uranium fission cross-sections and 8 - 15/ for the uranium capture cross-sections and the ratio a. These discrepancies are well known since long. But their origin is still unknown. Since our 1969 evaluation all important experi- mental data have been carefully remeasured or re- analyzed: the fission cross-sections, the half-lives involved, the scattering cross-sections, the Chalk 172 River thermal irradiation data, the fission-neutron yields 9 and y, the fission-neutron spectra, etc. Although one can never be sure of unexpected surprises, the striking consistency of the experimental data within the two subsets suggests that all these values are reliable within their quoted errors. The only area which has not been reinvestigated with comparable scrutiny, is the lowest energy region around and below 0.01 eV, which significantly contri- butes to the 20 C Maxwellian cross— sections. In 1975 [3], I thought that the lowest-energy cross-section curve shapes and thus the Westcott g-factors may be responsible for the discrepancies, but the analysis of B.R. Leonard et al [4] showed that the g-factors as obtained from the curve shapes, are rather accurately known, at least for U-235» not so much for U-233. However, a re- examination or re- measurement of the few existing data in the lowest energy region still seems to be advisable. 1 = 1. 2. g = J o/e 1 Maxw dE/o M g = 3/o o U-233 absorption 0.999 + 0.003 [3] 1.001 fission 0.9965 ± 0.002 [8] (or 1.000 for different extra- polation of o(E) to zero energy) 0.988 capture (1.03 ± 0.02) 1.17 U-235 absorption (0.973 + 0.002) 0.976 fission 0.9775 + 0.0015 [4] 0.963 capture 0.982 ± 0.002 [4] 1.06 Pu-239 absorption 1.077 ± 0.005 [10] 1.081 fission 1.053 + 0.003 [9] 1.053 capture (1.14 + 0.02) 1.16 Pu-241 absorption 1.039 + 0.003 '[10] 1.039 fission 1.045 + 0.006 [[11] 1.035 capture (1.02 ± 0.02) 1.05 Table 2. 20 C Westcott g-factors col. 1: g-factors calculated from the curve shape o(E) according to g= Jo(E) /EMaxw(E) dE/o {e The values in () were deduced from the other two g-factors quoted for the same isotope using cross-sections of col. 1 in Table 1. col. 2: g-factors calculated from 20°C Maxwellian and 2200 m/s experimental cross-sections accord- ing to g = 3 / o o using the values from columns 3. resp. 1. in Table 1. It is however evident, that the lowest-energy curve shapes and thus the g-factors derived from them, cannot be responsible for the full amount of the dis- crepancies. The g-factors g = 3/o o obtained from comparison of experimental data 3 in a 20 C Maxwellian neutron spectrum with a for 2200 m/s neutrons, cannot be brought into agreement with the g-factors obtained from curve shapes g= I a(E)-JE Maxwellian (E) dE/o TF This is illustrated in Table 2. One must suspect that there exists a still unknown physical effect in the lowest energy range. Scientists at NBS are considering whether this unknown effect may be related to phonon transitions between cold neutrons and the sample [12], Not know- ing any details of these considerations, I believe that they may go into the right direction, although I cannot judge whether such effects may give a quantitative ex- planation of the discrepancies encountered. I can only conclude that the energy range around 0.01 eV will re- quire further investigations before the thermal cross- sections of the fissile nuclides can be established as reliable standards. References [1] Westcott C.H. , Ekberg K. , Hanna G.C., Pattenden N.S., Sanatani S. , Attree P.M., Atomic Energy Review j 2 (1965) 3. [2] Hanna G. C«, Westcott C.H., Lemmel H.D., Leonard Jr. B.R. , Story J. S., Attree P.M., Atomic Energy Review J_ (1969) No. 4 p. 3. [3] Lemmel H„D. , Proceedings of the Conference on Nuclear Cross-Sections and Technology, Washington D.C., 3-7 March 1975, NBS Special Publication 425 (Oct, 1975) Vol. 1 p. 236. Note that the more detailed report announced as Lemmel H.D. , Axton E.J. , Deruytter A.J., Leonard Jr. B.R. , Story J.S., INDC(NDS)-64 has not yet been issued. [4] Leonard Jr. B.R. , Kottwitz D.A. , Thompson J.K. , EPRI/NP-167 (Feb. 1976). [5] Vaninbroukx R. , ANL/ND-77-1 (Nov. 1976) p. 29. [6] Smith J.R. , Proceedings of the Conference on Nuclear Cross-Sections and Technology, Washington D.C., 3-7 March 1975, NBS Special Publication 425 (Oct. 1975) Vol. 1 p. 262. [7] Smith J.Ro, USNDC-11 (June 1974) 12. [8] Primarily based on Steen N.M., WAPD-TM-1052 (Sept. 1973). [9] Deruytter A.J., Becker W. , Annals of Nucl. Sci, and Eng. 1 (1974) 311, and Deruytter A.J., Wagemans "5. , J. of Nucl. En. 26 ( 1 97 2 ) 293. [10] Westcott C.H. AECL-3255 (April 1969), value in- creased due to revised scattering cross-section. [11] Lemmel H.D., Westcott C.H., J. of Nucl. En. _21 (1967) 417 » corrected for revised scattering cross-section [12] Bowman C.D. , private communication 1977. 173 AN ASSESSMENT OF THE "THERMAL NORMALIZATION TECHNIQUE" FOR MEASUREMENT OF NEUTRON CROSS SECTIONS Vs ENERGY* R. W. Peelle and G. de Saussure Oak Ridge National Laboratory Oak Ridge, Tennessee 37830 Refined knowledge of the thermal neutron cross sections of the fissile nuclides and of the (n,ct) reaction standards, together with the reasonably well-known energy dependence of the latter, have permitted resonance-region and low-keV fissile nuclide cross sections to be based on these standards together with count-rate ratios observed as a function of energy using a pulsed "white" source. As one evaluates cross sections for energies above 20 keV, optimum results require combination of cross section shape measurements with all available absolute measurements. The assumptions of the "thermal normalization method" are reviewed and an opinion is given of the status of some of the standards required for its use. The complications which may limit the accuracy of results using the method are listed and examples are given. For the 35 U(n,f) cross section, the option is discussed of defining resonance-region fis- sion integrals as standards. The area of the ^9 eV resonances in this nuclide may be known to one percent accuracy, but at present the fission integral from 0.1 to 1.0 keV is known to no better than about two percent. This uncertainty is based on the scatter among inde- pendent results, and has not been reduced by the most recent measurements. This uncertainty now limits the accuracy attainable for the 235 U(n,f) cross section below about 50 keV. Suggestions are given to indicate how future detailed work might overcome past sources of error. [ °B(n,a); cross section; 6 Li(n,a); neutron; normalization; resonance; shape; standards; thermal; 235 U(n,f)] Introduction Since the earliest days of neutron cross section measurements using pulsed "white" neutron sources, workers have utilized the simple (1/v) shape of the (n,a) cross sections of 6 Li and 10 B to exploit the relatively well known absorption cross sections of the thermal energy range. 1 Through the last decades our knowledge of the requisite data has become successively more refined, and the existence of more powerful sources has allowed this "thermal normalization method" to be extended over an extremely broad energy range of ^8 decades. In this talk the authors explore the defi- nition of the method, identify some practical problems which have prevented this method from dominating all others, show some measure of the success so far ob- tained, and suggest where the cross section community might labor in its quest to exploit this technique to the extent many have dreamed should be possible. This talk is pertinent to the present meeting because the thermal normalization technique provides a prime re- quirement for the shapes of the standard (n,a) reac- tions and for the 2.2 km/s cross section standards. For 235 U(n,f), the technique is used to aid the defi- nition of a secondary standard of great utility, so many of the examples below are from work on that nuclide. The recent papers of Leonard and of Bhat give forceful examples of the effort required to apply the method to this nuclide in a more correct way than earlier was possible for a present author. Carlson and Czirr have discussed the status of the thermal normalization for this cross section from a somewhat different point of view. One reason for the ascendancy of the technique is that with it a linac or other white-source laboratory can provide an in-house cross section result which is complete over a very large energy range. All this can be done without direct measurements of sample mass which would otherwise generate a minimum uncertainty of 1-2 percent. Therefore, many cross section sets are normalized at thermal energies, although segments of the results could as well have been normalized in another way; therefore, more of the experimental literature seems to be based on this method than in fact is the case. From a more global point of view, which requires that all existing data be combined, one sees at once that the use of a normalization based on thermal cross sections is conceptually separate from the use of rela- tive measurements of count ratios between a counter based on the unknown and one based on a cross section of known shape. (This distinction is nicely emphasized by evaluation techniques such as those of Poenitz 6 in which shape and normalization decisions are separated.) In fact, nearly all sets of cross sections vs energy, in whatever energy range, are based on count ratios to the output of a detector with smooth and presumably known energy response, and it is a matter of detail whether or not this response follows the values of a known cross section. One can expect a typical synthe- sized result from a linac lab to consist of one or more segments of relative o(E) data, the lowest segment reaching the thermal energy region and providing one normalization. This normalization has peculiar status only because it may be tied directly to applications through thermal critical experiments and when it pro- vides a normalization with smaller uncertainty than others available to an evaluator. Other means of abso- lute normalization are available, especially in the MeV range, so that for sufficiently high energy some abso- lute measure other than the thermal normalization is likely to dominate. Whether this crossover of relative importance occurs at 20 keV or at 1 MeV or above will depend on the nuclide and the opportunities provided to the experimenters at the several laboratories. However, in every case a thermal normalization can be applied, it is important and could not be ignored below ^10 keV even if ±1% absolute normalization data were available at scattered energy points above 20 keV. In the special case of 235 U(n,f), one may be tempted in evaluating the standard cross section to ignore data based on normalization at thermal energies because the presence of fluctuations prevents this fis- sion standard from being recommended for use at ener- gies below 200 keV, and because there is such a wealth of experimental absolute data at the higher energies. Such a reaction would be hasty for three reasons: 1) the weight of measurements based largely on normali- zation at thermal energies is far from null, and inde- pendent methods must always be compared for mutual 174 consistency if we are ever to achieve the correct cross sections, 2) average cross sections of "standards" quality are needed in the region of fluctuations if one is to take advantage of clean integral fission-rate experiments in which the lower-energy neutrons play a significant role, and 3) when the approximate shape of the flux is known and is smooth, one may make use of the 235 U fission rate to fix the normalization of the flux even if one cannot be sure enough of energy scale alignment to use directly the ratio observed to 235 U(n,f) within each experimentally defined energy cell. The question of the importance of 235 U(n,f) below 0.2 MeV is actually moot because this cross sec- tion must anyhow be known well because of its direct practical applications. Table 1. Evaluations of 2.2 km/s Fission Cross Sections for Uranium Have Varied By More Than the Uncertainties Quoted 'U 3 U 'Pu Reference 530.6 ± 1.9 580.2+1.8 741.6 ± 3.1 Hanna (IAEA) 1969 585.7 + 1.8 742.5 ± 3.1 De Volpi, 1971 b 526.3 ± 0.8 577.5 ± 1.1 Steen, 1972° 533.7 ± 2.7 585.7 ± 2.3 742.0 ± 4.2 ENDF/B-IV, 1973 d 529.9 ± 1.4 583.5 + 1.3 744.0 ± 2.5 Lemmel, 1975 e 583.5 ± 1.7 Leonard, 1976* ;f What the Thermal Normalization Method Requires For now, assume a restricted definition of the thermal normalization method. The method requires use of a standard cross section known well as a function of energy and at thermal energy (2.2 km/s), and it may be applied to an unknown whose cross section is also well known in the thermal energy region. Figure 1 exhibits the standard relation between unknown, known, and ob- served quantities, and quotes the restrictions on the method's applicability. One must previously have sub- tracted any backgrounds from the observed rates. Understood departures from the required energy inde- pendence of flux ratio or detector efficiencies may be corrected. The beauty of the technique is that neither neutron fluxes nor sample masses or areal densities appear in the equation of Figure 1. The Thermal Normalization Method Requires No Mass Measurements and Only Relative Count Rates vs_ Energy If R . (E) = Reaction Rate Ratio x-sample — __ x/s standard sample /°s (E) \/ R x/s (E) a„(E) = a„(th) L ,.uJ ( „ ,^ | , if x v ' o (th)/\R , (th) x/s ratio of flux between two samples is not a function of energy. Count-rate ratios may be used for R X / <; ( E ) whenever efficiency for detecting reactions is also not a function of energy. G. C. Hanna et al . , Atomic Energy Review ]_, No. 4, p. 3 (1969). A. de Volpi, Conference on Neutron Cross Sections and Technology, CONF-710301 (Vol. 2), p. 564 (1971). C N. M. Steen, WAPD-TM-1052 (Sept. 1972). J. R. Stehn, BNL, private communication to Cross Sec- tion Evaluation Working Group members on final fit Q4 (Dec. 13, 1973). H. D. Lemmel, Conference on Nuclear Cross Sections and Technology, NBS Special Publication 425, p. 286 (1975). B. R. Leonard, D. A. Kottwitz, and J. K. Thompson, "Evaluation of the Neutron Cross Sections of 235 U in the Thermal Energy Region," Report EPRI-NP-167-Project 512 (1976). Table 2. Evaluated ENDF/B Cross Sections for the Li and "B(n,a) Reactions. (Values of C v eV/(a /0.0253) are Tabulated) o 'Li(n,q) i(n,tt) E(kev) III IV III IV V 0.1 0.999 1.001 0.999 0.995 0.994 0.994 1.0 0.987 0.998 0.987 0.985 0.982 0.983 10.0 1.000 1.009 1.004 0.963 0.965 0.963 30.0 1.027 1.054 1.042 0.958 0.970 0.963 50.0 1.078 1.121 1.Q99 0.971 0.984 0.972 100.0 1.379 1.460 1.386 1.035 1.016 1.000 d in barns o 940.3 940.0 935.9 3836.5 3836.5 3836.6 Figure 1. To sense whether the necessary standards data are available to apply this method, one may look at an example consisting of the thermal cross sections of the fissile materials and the evaluated results for the light-element (n,a) reactions. Table 1 compares several evaluations of the 2.2 km/s fission cross sec- tions of some important materials. Evaluations will differ until data inconsistencies are removed. The sometimes-quoted uncertainties of ^0.2 percent may be too ambitious, but one senses that half-percent uncer- tainties suffice for fission. Achieving such an accu- racy in a single absolute measurement involving sample masses involves very great care. 7 The 2.2 km/s cross sections of °Li and °B are typically quoted to 0.4 and 0.2 percent, 8 and Table 2 exhibits the stability of the evaluated energy dependence of these cross sections through the last three ENDF/B evaluations. Since mea- sured data for these cross sections at any one energy Values for Versions III and IV were obtained from the files of the National Neutron Cross Section Center at BNL. The ENDF/B-III standards are also described in BNL 17188 (ENDF-179), M. K. Drake, editor (1972). Ver- sion IV was documented by Hale, Stewart, and Young in LA-6518-MS (1976). The preliminary evaluation for Ver- sion V was obtained from G. M. Hale et al., LASL (1976) and also included in the Minutes of the CSEWG Normali- zation and Standards Subcommittee, May 17-20, 1976. in the tens of keV region spread widely (10 percent), the stability shown could be illusory if surprises develop in proper representation of the data using reaction theory. Though complete evaluated uncertain- ty information for the energy dependence of these cross sections has not been given, one can see why confidence in the cross sections through 10 keV to 1 percent has been gained from the stability of the result and the closeness to a (1/v) response. (These authors are con- cerned whether one can at present have confidence in 175 either reaction cross section at 50 keV to 2 percent or at 100 keV to 3 percent, but the doubt arises from casual rather than detailed studies.) These (n,a) reactions and the °B(n,ay) reaction have been used as standards to much higher energies, so it seems important that work continue to establish the underlying cross sections. Fortunately, other flux detectors based on the hydrogen cross section become available above about 10 keV, but counters based on 6 Li and 10 B can have much better time response than the hydrogen proportional counter which becomes useful at this energy. As additional smooth-response detector systems, not applicable at thermal energies, become available at energies above 10 keV, the simple thermal- normalization technique discussed above gradually loses its pre-eminence. In weighing the data or planning an experiment, one must balance the niceness of a fresh detector system against the need to internormalize results if the new system can no longer "see" the thermal-energy guidepost. Of course, the thermal normalization method need not depend on an (n,a) reaction, and need not depend on a cross-section shape standard at all if a detector of calculable response with adequate time resolution can be used over a broad energy range down to thermal ener- gies. Experience with the Thermal Normalization Technique for Z3bl U(n,f) While thermal normalization may be employed by a single group to give results covering a broad energy range, its most important application is in cross sec- tion evaluation. An evaluator quickly finds that there are few individual measurements which span the region of interest without significant changes of apparatus, so one must combine the results from a number of indi- vidual experiments covering the energy range in patch- work fashion. Moreover, each reported data set may be referred to a different value of the 2.2 km/s cross section using a different technique for establishing this normalization, and be based on a different shape for the standard (n,a) reaction used to determine the energy dependence of the flux. With great effort, and the help of the data center system, it is possible to unravel the variations of data treatment. Leonard has recently described his efforts to renormalize on a common basis a number of measurements covering portions of the energy range below 1 eV; most of the examples in this section depend on his work. Through such consistency studies it is sometimes possi- ble to uncover systematic difficulties not suspected by the original authors. The tables of Ref 2 imply renor- malization requirements ranging from 0.3 to 1.5 per- cent, not counting the effect of differences in the 2.2 km/s cross section used. For reasons given by Deruytter and Wagemans and suggested in the next section, and because some of the more careful measurements do not in any case reach to energies above the resonance region, it is convenient to define the fission integral of the resonances near 9 eV as an intermediate standard. Table 3 gives the data for this integral using two popular choices for the energy bounds. Most of the listed data follow the renormalizations given by Leonard and the uncertain- ties on these values reported by Bhat, 3 but the first o line came directly from Ref 10 renormalized to a 583.5 b, and the Gwin data on the last line came from newly analyzed data sets. 11 The uncertainties on the Deruytter and Wagemans data were taken from Ref 10, and the uncertainties given on the last two data sets were assigned by this author without much analysis. The weighted average integrals at the bottom of the table indicate that the decision, whether to include Leonard's re-evaluation of Ref 10 or the original results, alters the output average by as much as the uncertainty assign- ed by expanding the propagated output uncertainty to make x /df =1. In the presence of the observed incon- sistency and conflict, it is difficult to accept the small 0.6 percent output uncertainty given. Pending more detailed uncertainty analysis, and including an uncertainty for the 2.2 km/s value, the overall uncer- tainty on this integral is judged to be about 1 percent. Based on more study, Leonard has estimated a 2.8% uncer- tainty in the 7.8-11.0 eV integral. 2 Table 3. DATA ARE INCONSISTENT FOR THE FISSION AREA" OF THE % 9 eV RESONANCES IN 235 l). Reference Deruytter & Wagemans (1971). Deruytter, per Leonard (1976). Czirr & Sidhu (1976) . Private communication (Feb. 1977). Leonard (June 1976) gives 224 b-eV for this value and 241 b-eV for Czirr's 7.8-11.0 eV integral. Gwin et al . (1976), per Leonard (1976). de Saussure e_t al. (1966) , per Leonard (1976) . Bowman (1966), per Leonard (1976). Shore & Sailor (1958), per Leonard (1976). Gwin (1977). Private communication. These two integrals come from separate new measurements . O dE ' 7.4 (barn-eV) I °f dE J 7.8 (barn-eV) 221 ± 2 (226 ± 2) 227 + 2° 238 ± 2 (243 ± 2) 219 ± 225 ± 234 ± 216 ± 226 + 236 241 252 0(7.4-10.0) = 223.7 ± 1.5(225.2 ± 1.4) b-eV, X 2 = 11.8(9.5) c a. Normalized to 0° = 583.5 b. Table follows that of Leonard (1976). b. Uncertainties as reported by Bhat (1976). c. Values in parentheses use the Deruytter data renormalized by Leonard using only the data above 0.21 eV. The energy intervals used in Table 3 which are commonly chosen for study are plausible ones, but if normalization or the relative behavior of various data sets are to be considered seriously, it is highly desir- able to compare the areas of two or more resonances. In this way shape differences can be sensed which might place an evaluation procedure at an appropriate level of doubt. If the fission integral of the ^9 eV resonances is considered given, then the chosen value can be used to normalize experiments which did not reach to thermal energies or for some reason should not be trusted in that regime. Table 4 shows most of the independent data for the 0.1-1.0 keV 235 (n,f) fission integral which can be normalized at thermal energy or in this resonance. Table 4. THE MOST MODERN VALUES OF THE 0.1-1.0 keV 2 3 5 U FISSION INTEGRAL DIFFER BY 7 PERCENT. Details depend somewhat on evaluation technique. Listed by Bhat Listed by Wagemans Original Author 12.4 11.8 11.4 12.2 11.8 12.3 11.8 11.5 12.3 T = (11.9 + .2) b-keV de Saussure (1967) Gwin (1976) Czirr and Sidhu (1976) Wagemans (1976) Wasson (1976) a M. R. Bhat, ANL-76-90, p. 307 (1976). Normalized to a£ - 583.5 b and using the ENDF/B-V (n,a) shapes. Wagemans and Wasson data were normalized to 1(7.8-11) ■ 241.2 b-eV. C. Wagemans and A. J. Deruytter, Annals of Nucl. Energy 3^, 437 (1976). This table is based entirely on data and standard shapes utilized by the respective authors. 176 The data sets of Wagemans and of Wasson were nor- malized in the resonance while the other three sets retain the thermal normalization. The left column lists the results as renormalized by Bhat, while the right column lists the values given in the original papers. Bhat's reworked values are to be preferred because they were obtained in a consistent way, but in this particular case the distinction is not great com- pared to the large scatter of results. The scatter will appear little less than disastrous for anyone hav- ing strong hope for early use of thermal normalizations for precision work; the range of values is 6 to 8 per- cent depending on whether one includes only the newer measurements which were optimized for the goal of obtaining precise fission cross sections. In fact there seems to be no reason to downweight the older results, since the uncertainty based on the scatter of the last three values is just as large as that based on all five. Simply put, the data base to date does not define the 0.1 to 1 keV fission integral, relative to the thermal cross section, to better than about 2 percent. The particular energy interval displayed in Table 4 was chosen for convenience, but experience shows that such comparisons for normalization purposes are needlessly confused if comparison intervals much nar- rower than two lethargy units are used. As noted by Carlson and Czirr, it is important that a normaliza- tion interval be chosen above ^0.2 keV because below this energy the shapes of the various measurements seem more discrepant. Also, as treated in the evalua- tion of Bhat, many additional measurements are avail- able above about 0.1 keV . It is puzzling that prior to the experiments of Refs 12-14 so few measurements spanned the range from 7 eV to 1 keV. From the scatter indicated in Tables 3 and 4, and from the additional similar difficulties encountered at higher energies, » one must conclude that the application of the thermal normalization technique to U(n,f) through the resonance and on into the keV region is more difficult than has been assumed. More care will be required than has already been expended. Much concern is relieved by recognizing that many of the studies were not optimized for the fission cross section alone, but as seen above there is important scatter among the most recent measurements performed with only this goal in mind. Considering the disper- sion among results and uncertainties in the shapes of the (n,a) standards above 30-50 keV, it is remarkable that an evaluation task force was able to determine that an alteration as small as 1 percent to the results obtained by this method for the 0.01 to 0.2 MeV region would be all that would be necessary to bring the results into better agreement with experiments per- formed using monoenergetic sources and other normaliza- tion techniques. 3 Complexities in Application of the Thermal Normalization Method Given the apparent adequacy at least through 10 keV of the data which underlie the use of the thermal normalization method, why might various investigators obtain conflicting results? Figure 2 lists some possibilities . The ratio data proving the consistency of the °Li and 10 B(n,a) standards are quite limited, the work of Sowerby et al. being most quoted though some indica- tions are given in papers by Perez et al. and by Wagemans and Deruytter. 2 Since the ratio of these cross sections is presumably easier to measure than the absolute value of either one, and since any incon- sistency between these standards is causing serious confusion, a significant effort would be worthwhile to establish beyond doubt the ratios of these cross sections . Why D on' t All Authors Get the Same Answers ? • The (n,a) standards may be inconsistent. • Efficiencies of detectors for flux and unknown may not be proportional to reaction rates. • Conflicting requirements on sample thickness, flight- path length, source pulse rate, and run time inhibit optimization for accuracy. • Assessment of counting backgrounds and resolution functions is beset with hazards. • Time pressure often curtails complete evaluation and intercomparison of results. Figure 2. Detectors for the light element (n,a) reactions are usually designed to have uniform efficiency for the reaction products, independent of neutron energy at least to first order. Not often, however, does one see a detailed demonstration that this assumption is valid for a particular detector. The reaction Q-values are sufficiently large that one would not expect signifi- cant effects on detector performance from reaction kinematics for energies up to at least 10 keV, but at higher energies the energy dependence should be worked out for each detector design. Nonisotropy of reaction cross sections in the cm. system also requires consid- eration; we have seen from the work of Raman et al. that a measurable fore-aft asymmetry is confirmed even for 10 eV neutrons when the detector subtends ir solid angle.' The observed energy dependence of this asym- metry to 10 keV had the form 1 + 0.005x E(eV) , so the effect is strong. This finding again suggests that for precision work the energy dependence of flux detec- tors based on the (n,a) reactions should be analyzed with considerable care. The reason that a thermal normalization must so often be carried out through successive renormalization of partially overlapping data sets is that the experi- menter using a pulsed white source cannot simulta- neously optimize all the experimental variables which affect data accuracy; it is mandatory to compromise other values as one extends the energy range of an individual data run to minimize the number of succes- sive normalizations. Gwin et al. , in their paper describing very broad-range experiments, discuss some of the problems to be handled in such measurements. One set of compromises involves sample thickness: to obtain adequate counting rate one may use a sample so thick that self-absorption is serious in the unknown or in the flux detector, and in some cases a thick sample in a fission ion chamber may complicate timing and make less sure a constant efficiency for the detec- tion of fission products. The sample-thickness con- flict might be resolved by establishing precision aver- age cross sections using a thin sample, while filling in the short-range detailed energy dependence using a detector with more material; a disadvantage of this solution would be that backgrounds are more difficult to assess using the black resonance technique when the counting rates are low. 177 Assuming that the repetition rate and the thick- ness of anti-overlap filters may be changed at will when using a pulsed white neutron source, another set of compromises involves choice of flight-path length and neutron pulse repetition rate. Short flight paths (5-20 m) minimize overlap problems but may not permit satisfactory time resolution for flux detectors and may place any effect of "gamma flash", or of the electrical noise associated with neutron pulse production, too close in time to the interesting data. Assuming the flight-path length has been chosen, Figures 3 and 4 illustrate the shape of the count rate from a (1/v) flux detector with two different arrangements to reduce the low-energy sensitivity enough to avoid time-frame overlap. In Figure 3 the experiment could cover ener- gies to about one-half electron volt, but there is need 0.2 0.5 1 2 5 10 20 50 100 200 500 1000 NEUTRON ENERGY (eV) 10j flux monitor when a cadmium filter was used to eliminate time-frame overlap. The cobalt filter was used for background estimation. From Ref 19. to work hard to correct for the effect of the cadmium resonances which may be resolved to different extents in detectors for the flux and for the unknown. For Figure 4 a boron filter was used to cut the flux off at *<£-)= 7,2 f a,6 r 56/ ^- 2.7 Al 5.6 keV 35.2 9" No 2.65 keV S'(f) = 0.6«S df) 96 AND . - HOheV 5 |f) SPECTRUM WITH ,0 8 ONLY ^^^ ^10^^^ \ 1 \ '• ['■; ' - / / 5"(f) SPECTRUM WITH RESONANCE ; i ■ I / / FILTERS AND I0 6 :'• ENERGY (eVI Neutron Energy Spectrum with and without the Resononce Filters Figure 4. Time spectra in a B-based flux monitor when a boron filter was used to avoid overlap. The lower, curve illustrates the use of resonance filters to help estimate backgrounds. From Ref 20. ^1 eV, but the counting rate at about 4 eV is only one percent of what it would have been if a clean cutoff- mechanism had been available to eliminate all neutrons with energies below 1 eV. If more boron is used, a faster repetition rate is possible and vice versa ; in the case of Figure 4 the net counting rate was opti- mized at about 3 keV, and counting rates could have been improved by decreasing boron thickness and repeti- tion rate had the experimenters wished to improve per- formance for energies below 3 keV. A point of this discussion is that lowering repetition rates to increase the energy range of an experiment may increase counting rates through a large share of the dynamic range. Backgrounds in pulsed neutron experiments are usu- ally determined by inserting into the beam "filters" designed to remove all the neutrons at the resonance energies of these absorbers. If the filter can be thin, one assumes that the filter does not affect the shape of the overall spectrum of higher-energy neutrons which normally cause all the troublesome backgrounds. Between the well-spaced energies of the filter resonances, the experimenter must interpolate a shape based on auxil- liary data or an assumption of smoothness. Rarely has it seemed possible to understand the shape of the back- ground based on calculation of the transport of beam neutrons through the regions surrounding the neutron source and detector. Figures 3 and 4 illustrated this background estimation method for boron-based flux detec- tors. Note that the backgrounds seem very low. In Figure 4, where a boron filter was used to roll off the flux, one senses that the background ratio was not so favorable at lower energy since the boron filter atten- uated relatively less the higher-energy neutrons induc- ing the background. Figure 5 from an earlier experi- ment 19 shows that measured relative backgrounds have not always been so low; one can see that the average cross sections and the valleys between resonances would have been markedly affected by an error in the assumed » .;< •'. i;':'-; :v. ■■■ -> V_J /V; V; ; ■ Mi j| |j N U I ASSUMED BACKGROUND Figure 5. An example of background determina- tion in a fission cross section measurement, from Ref 19. Coincidences were recorded between a fis- sion chamber and a liquid scintillator tank which surrounded it. background shape. Figure 6 is a more favorable example at higher energy where the background level appears to be about 1 percent. When the background level estimated by the above technique is not small, more detailed analysis is required. Gayther et al. 2 * have suggested one way to 178 proceed using data obtained using several filter thick- nesses. The difficulties arise partly because one struggles to differentiate sharply between a neutron- induced background and a "tail" on the experimental The small discrepancy on the low energy side of the peak was subsequently explained by taking into account neutron diffusion in the thick 6 Li-glass detector and the skew shape expected for the time distribution of neutrons leaking from the target moderator. 23 1000 t 100 0.01 0.001 10 20 NEUTRON ENERGY (keV) 50 100 Figure 6. A background determination in the keV energy range using the black resonance tech- nique, from Ref 16. resolution function. The only difference is one of degree. Figure 7, though from a transmission experi- ment, shows the effect of a close-range asymmetry in a resonance chosen by Olsen 22 to facilitate analysis. 1.0 0.8 O « 0.6 0.4 0.2 1 1 1 1 1 r NEED FOR COMPLEX DETECTOR RESOLUTION FUNCTION — P 3/2 RESONANCE 206 Pb, 0.021 AT/A NORMAL RESOLUTION FUNCTION ASSUMED _L 3340 3350 3360 NEUTRON ENERGY (eV) 3370 Figure 7. Discrepancy between observed and expected shapes of a transmission dip when the skew distribution of neutrons from a moderator and detector distortions were neglected. As the thickness of a resonance filter is in- creased, one typically obtains evidence that the flux at a given energy is affected a little by that at near- by energies far beyond the nominal system resolution width. Figure 8 illustrates this effect. The third- thinnest filter would be computed "black", with less than 0.001 transmission, yet increasing the sample 801 811 821 TIME CHANNEL Figure 8. Neutron intensity near 6.7 eV observed through a broad range of 238 U absorbers. From the work of Ref 24. thickness continued to reduce the counting rate near the minimum as the energy region of low transmission broadened. At any time after the neutron burst there are neutrons present at low intensity from a very broad energy range. The intensity and spectral distribution of these neutrons depends on the source, collimator, and detector construction and their environs. Improv- ing accuracy of measurement through background reduc- tion depends on identifying the sources of these off- energy neutrons, reducing their intensity, and learning to correct more precisely for the presence of those which cannot be removed. Given the difficulty of know- ing the precise content of neutron beams, it seems preferable that the neutron flux shape be measured at the same time and place as the count spectrum from the sample under study. Figure 9 is adapted from a very old comparison among preliminary sets of U(n,f) data in the reso- nance region. Much of the discrepancy was ascribed to detection of neutrons scattered from aluminum and concrete structures in the beam near the detector, and we believe that subsequent final data did not suffer so much from this difficulty. Some of the error sources discussed above relate to counter design, some to beam production and collima- tion, and some may be inherent to pulsed-source methods. 179 Most have not been investigated and minimized with all the tenacity and ingenuity which would be possible, partly because practical experiments are beset by re- current crises originating in electrical noise, counter decay, electronic malfunction, and sometimes the need to suffer suboptimal beam conditions to meet the needs of other experimenters. technical work has really been completed. The require- ments on final reporting are substantial if evaluators are to have all the needed information. 2 ° The main standard cross section efforts called for in this paper are listed in Figure 11; substantial progress toward these goals has already been made in this decade. ORNL-DWG 77-7107 s / -5c^ — v \ 1 1 W 1 A - — \\ u\ — — \ \ I _ - \ -7 • 1 1 \ - - 1 - \ ^^, I 1 [ 1 1 1 2 3 NEUTRON ENERGY (eV) Figure 9. Two sets of preliminary observa- tions of the U(n,f) cross section, illustrat- ing effects ascribed to return of neutrons scattered from structures in the beam. How Precise Extrapolation from Thermal Cross Sections May Yet be Achieved Experience shows that accuracy improvements are iterative, and depend on improved sources and instru- ments as well as the broad shoulders of those who have labored before. Figure 10 lists some major areas of effort which require attention. In the second point MORE WORK ON STANDARDS IS REQUIRED TO ENABLE THE FULL POTENTIAL OF THE THERMAL NORMALIZATION METHOD • Perform measurements to prove compatibility of the Li(n,a), 10 B(n,a), and 10 B(n,aY) reactions. • Press hard to reach 100 keV with (n,a) cross sections accurate to one percent. • For 235 U(n,f), try harder to establish two resonance- region fission integrals to 0.5 percent accuracy. A standard integral over a broad region between 0.1 and 1 keV is probably also needed. • Continue work to achieve thermal-neutron cross section standards with credible 0.3 percent uncertainties. (Data inconsistencies now confuse uncertainty analysis.) Figure 11. With improved techniques and underlying standards, the goal of precise extrapolation to high energies based on thermal-region normalization can yet be reached. This goal, qualified by the recognition that absolute measurements at energies above ^20 keV will also be important, indeed must be achieved if there is to be a 235 U fission standard "for all seasons" and a knowledge of other cross sections sufficiently precise to satisfy all needs of developing technologies. Acknowledgements WE KNOW SOME DIRECTIONS TO LOOK FOR IMPROVING ACCURACY USING THE THERMAL NORMALIZATION METHOD • Improved analysis of backgrounds and resolution functions will be needed for most detectors. • To cover 7-8 energy decades, important overlap of several data sets will often be required. Full evaluation of results through this range is required to establish credibility of normalization. • Hasty exhibition of final results must be resisted, if haste will prevent full analysis and exhibition of data uncertainties and correlations. The authors gratefully acknowledge discussions on this topic with colleagues the world over, and partic- ularly with R. Gwin, L. W. Weston, R. B. Perez, and D. K. Olsen. We are indebted to J. Gentry and E. Plemons for expedited production of the typescript. References *Research sponsored by the Energy Research and Development Administration under contract with the Union Carbide Corporation. 1. For example, J. F. Raffle and B. T. Price, Proceed - ings of the International Conference on the Peace- ful Uses of Atomic Energy , paper P/422, Vol 4 p 187 (1955). Equally, papers by B. R. Leonard, ibid , paper P/589, p 193 (1955); V. L. Sailor, ibid , paper P/586, p 199 (1955). Figure 10. we mean by evaluation no results on the same cros evaluation including the partial cross sections, against the release and but the third point indi are more concerned about pressures to issue final t only the intercomparison of s section but thorough joint total cross section and other It is customary to advise use of preliminary results, cates that the present authors the apparently increasing results before the necessary 4. B. R. Leonard, Jr., in ANL-76-90, Proceedings of the NEANDC/NEACRP Specialists Meeting on Fast Neutron Fission Cross Sections of 23%, " b U, 2 3 8i and "^Pu, p 281 (1976). M. R. Bhat, in ANL-76-90, op_. cit . , p 307 (1976). Also, related private communications (1976). R. W. Peelle, ORNL-4955, An Evaluation for ENDF/B - IV of the Neutron Cross Sections for 23b U from 82 eV to 25 keV, (1976). 180 5. C. W. Carlson and J. B. Czirr, in ANL-76-90, op.cit. p 258 (1976). 6. W. P. Poenitz and P. Guenther, in ANL-76-90, op.cit. p 154 (1976). 7. A. J. Deruytter, in CONF-701002, Neutron Standards and Flux Normalization , p 221 (1971). 8. S. F. Mughabghab and D. I. Garber, BNL 325 Third Edition, Vol 1 (1973). 9. G. M. Hale, NBS Special Publication 425, Proceed - ings of a Conference on Nuclear Cross Sections and Technology , p 302 (1975). See also G. M. Hale, in proceedings of December 1975 Trieste IAEA con- sultant's meeting on Use of Nuclear Theory in Neutron Nuclear Data Evaluation . Lane, Hausladen, and Monahan, in CONF-701002, o£.cit. p 107 (1971). Uttley, Sowerby, Patrick, and Rae, in CONF-701002, op . cit . p 80 and p 151 (1971). 10. A. J. Deruytter and C. Wagemans , J. Nucl. Energy 25, 263 (1971). 11. R. Gwin, ORNL, private communication (1977). 12. C. Wagemans and A. J. Deruytter, Annals of Nucl. Energy 3, 437 (1976). 13. 0. A. Wasson, in ANL-76-90, op. cit . p 183 (1976). 14. J. B. Czirr and G. S. Sidhu, Nucl. Sci. Engr. 60, 383 (1976). See also Ref 5. 15. Sowerby, Patrick, Uttley, and Diment, in CONF-701002, op.cit. p 151 (1971). 16. Perez, de Saussure, Silver, Ingle, and Weaver, Nucl. Sci. Eng. 55, 206 (1974). 17. Raman, Hill, Halperin, and Harvey, CONF-760715-P2 , International Conference on the Interactions of Neutrons with Nuclei , p 1340 (1976). 18. Gwin, Silver, Ingle, and Weaver, Nucl. Sci. Eng. 59, 79 (1976). . See particularly sections IV. A. and IV. D. 19. G. de Saussure et al . , ORNL-TM-1804 (1967). Also, with A. Lottin, Proc. Conf . Nucl. Data for Reactors , Paris, 2, 233 (1967). (The values given in the ORNL report are to be preferred.) 20. de Saussure, Silver, Perez, Ingle, and Weaver, Nucl. Sci. Eng. 51, 385 (1973). 21. D. B. Gayther et al . , Proc. of a Panel on Neutron Standard Reference Data , p 201 (1974). IAEA, Vienna. Also, D. B. Gayther, AERE Harwell, private communication (1972) . 22. D. K. Olsen, ORNL, private communication (1977). 23. A. Michaudon, J. Nucl Energy Parts A/B, 17., 165 (1963). 24. D. K. Olsen et al . , Nucl. Sci. Eng. 62_, 479 (1977). 25. F. D. Brooks, EANDC(UK)28 (1964). 26. R. W. Peelle, in ANL-76-90, op_. cit . , p 421 (1976). 181 REVIEW OF V FOR 252 Cf AND THERMAL NEUTRON FISSION J. w. Boldeman Australian Atomic Energy Commission Research Establishment Private Mail Bag, Sutherland, NSW 2232, Australia A review is presented of absolute measurements of v for the spontaneous fission of 252 Cf and of relative measurements for thermal neutron induced fission of 233,235^ ant j 239,241 pu _ Tne discussion includes the consideration of a number of sources of revision that have been suggested for some of the measurements. No evidence is found in the revised data of any experiment-dependent systematic error. A set of recommended values is given. (Neutron standards, v, 252 Cf, 233 ' 235 U (n,f ) , 239 ' 2itl Pu (n,f ) ) 1. Introduction Absolute values of v, the average number of neutrons emitted per fission*, for the thermal neutron fission of 233 U, 235 U, 239 Pu and 2kl Vu have been requested by reactor designers to accuracies approach- ing 1/4 to 1 /2%. For experimental reasons, the most convenient way of performing such measurements is rela- tive to a spontaneously fissioning standard and, for a number of years now, v for the spontaneous fission of 252 Cf has been the accepted standard. Consequently, an equivalent or better level of accuracy is required for this standard. In view of the precision which has been requested, the presence of a number of discrepancies in the measurements over the last 10 years or so has been the cause of considerable concern. Historically, the difficulties began following the publication of v for 252 Cf as measured by the Boron Pile 1 '. The value obtained [3 . 713±0.015j differed from the average of the two liquid scintillator determinations 2 / 3 ' [3.793±0.023] by an amount uncomfortably larger than the comparative error. Subsequently, preliminary values of v for 252 Cf obtained using manganese sulphate baths'*' 5 ' for the neutron counting also supported a value lower than the liquid scintillator average. Alternatively, indirect values of v for 252 Cf calculated using the 2200 ms" 1 constants (»3.78) were in agreement with the liquid scintillator values. The 1969 IAEA review 6 ' of the 2200 ms constants, therefore, recommended compromise figures in which both v and n values were shifted slightly from their experimental averages. For example, the average experimental value of v for 252 Cf of 3.743±0.016 compared with the recommended value of 3.765±0.012. However, as the precision of the MnSO^ bath deter- minations of v for 252 Cf improved, the support for the lower values increased. A third liquid scintillator measurement ' obtained an intermediate value. Further- more, it became apparent that some minor corrections were required for the two early liquid scintillator measurements ' . Thus the 1972 Neutron Standards Reference Panel was able to conclude that within the direct measurements the spread of the different values was statistically acceptable. The 1972 Panel recommended a value of 3.733±0.008 for v for 252 Cf. The implications of this recommendation were either that there existed some error in the other 2200 ms -1 para- meters or, alternatively, the error assignments were too optimistic. In particular, the n values used to provide indirect values of v were questioned. Subse- quent revisions 9,10,11 ' 12 ' of two n measurements ' ' for 233 U and 235 U have failed to attribute the discrepancy to any specific factor. Although Leonard and Lemmel ' have questioned the a measurements, the entire explanation does not seem to lie here either. 15) *In this review, the symbol v will be used for the average prompt neutron emission, while v will be used for the total neutron emission, including the delayed neutrons . The unresolved nature of this discrepancy has stimulated considerable activity in the revision of previous v measurements. At least two new absolute measurements are in progress , , in addition to a re-examination 19) of the v ratios, all of which aim to achieve high accuracy. In the present report, all documented v measurements of reasonable accuracy have been reviewed, including subsequent revisions to these measurements and consideration is given to further corrections that are now appropriate. In section 2, absolute measurements of v for the spontaneous fission of 252 Cf are examined, while section 3 is devoted to the v ratios. A set of recommended values is presented in section 4 . 2 . Absolute v Measurements Absolute v measurements may conveniently be sub- divided into two categories : (a) Delayed coincidence experiments (e.g. liquid scintillators and Boron Pile) in which a neutron counting gate is opened for a finite time after each fission event. The technique to a first approximation is independent of the absolute fission rate. (b) Direct measurements (e.g. manganese sulphate baths) in which the absolute fission and neutron emission rates are compared from separate determinations. Liquid Scintillator Measurements The large liquid scintillator technique was first developed for the measurement of v by Diven et al. 20 ' The neutron detector consists of a large liquid scintil- lator [lOO-lOOO SL in volume] which is loaded with a high neutron capture cross section material such as gadolinium or cadmium. A fission counter containing the appropriate target is placed at the centre of a tube which runs axially through the scintillator tank and allows entry and exit of a neutron beam. Neutrons produced by fission in this counter enter the scintilla- tor, are moderated there and, after a mean lifetime generally of the order of 10 us, are captured by the gadolinium or cadmium. The capture gamma rays so produced cause scintillations which may be observed by photomultiplier tubes mounted on the outside of the scintillator tank. By this method a multiplicity of neutrons produced in the fission event may be counted individually. Excellent discrimination against back- ground radiation can be obtained by gating the output of the photomultiplier tubes with the fission pulse and only counting scintillation pulses for several neutron lifetimes. The neutron detection efficiency of the liquid scintillator is proportional to the probability of neutron capture in the hydrogen and gadolinium or cadmium loading (and the structural materials), e c , and to the probability of the subsequent detection of the capture gamma rays, £y . The efficiency is a function 82 of the original neutron source energy (mostly because of the variation in leakage) and, because of the asymmetry introduced by the axial tube, it is also a function of the neutron emission angle with respect to the scintillator axis 8 . To determine the neutron detection efficiency of the scintillator for 252 Cf spontaneous fission neutrons requires a knowledge of e c (E n ,6) and £ Y (E n ,6) for E n = 0-15 MeV. These func- tions cannot be determined entirely experimentally. The normal procedure of calibration in the past has been as follows: (i) The probability of neutron capture in the scintillator as a function of neutron energy and emission angle e c (E n ,6) has been calcu- lated with a Monte Carlo code. (ii) It was assumed that e Y is independent of E n and 6. Y (iii) By measuring the probability of neutron detection of neutrons of specific energy and with a particular emission angle into the scintillator, a value of e was obtained. (iv) The calibration of the scintillator efficiency was checked by repeating the efficiency measurement for a number of other values of E n and 9 . Four sources of revision have been considered in the present review. (a) Fission Neutron Spectra Because of the neutron energy dependence of the scintillator tanks, it is necessary to make regular revision of all measurements to take advantage of the improvement in the precision of fission neutron spectra. Unfortunately, recommended values for the fission neutron spectra from Johansson 21 'were not available for this review, which used the data listed in Table 1. For neutron fission of 235 U and 239 Pu the data were taken from the evaluation of Adams . The spectrum for neutron fission of 233 U was taken to be Maxwellian with an average energy 1.02 5 times that for 235 U from the review of Smith ' . The neutron spectrum for neutron fission of 2Itl Pu was related to that for 239 Pu \2k) and v values from using the expression of Terrell^ the present paper. Because the experimental data for 252 Cf show a wide distribution of values , the spectrum normally assumed, namely Maxwellian with E = 2.15 MeV, has been retained. TABLE 1 FISSION NEUTRON SPECTRA Reaction Shape TAB (MeV) (MeV) (MeV) 233 U(n,f) Maxwellian 1.37 235 U(n,f) Watt 0.9878 2.1893 239 Pu(n,f) Watt 0.9723 2.7005 2Itl Pu(n,f) Maxwellian 1.40 252 Cf(s,p) Maxwellian 1.43 Watt spectrum: P (E) = C exp (-E/A) sinh/BE Maxwellian: P (E) = C>^E exp (-E/T) (b) Delayed Gamma Rays from Fission Because of the gamma ray sensitivity of the liquid scintillators, the delayed gamma rays which arise from the decay of isomeric states among the fission products contribute to the apparent neutron count rate. Table 2 lists the delayed gamma ray cascades which can make a significant contribution. Some gamma ray yields have not been obtained experimentally. Fortunately, in all cases (bracketed yields in Table 2) , it has been poss- ible to estimate the missing yield using relative fission fragment yields of the identified parent from the experimental data of Unik et al . ' Attention is drawn, in particular, to the isomer with a half life of 162 ns, where the yield in the spontaneous fission of Cf is fairly large and the estimated yields for 23 ->u and 239 Pu show that it can make a serious contribution to v measurements. This contribution can be minimised in future measurements by introducing a delay of at least 500 ns before initiating neutron counting. No experimental data are available for neutron fission of either 233 U or 2 ^ 1 Pu. For 23 U, I have used the 235 U yield data and for 21+1 Pu, the 239 Pu data. TABLE 2 DELAYED GAMMA RAY DATA T ^ Energy of Cascade (MeV) Parent % Yield per Fission (us) 252 Cf (sp) 23 5u 239 pu .020 0.614 A=100 0.34 (0.93) (0.85) .100 1.723 A=134 0.22 (0.51) (0.49) .162 1.692 134 Te 1.15 (2.7) (2.5) .62 1.505 A=135 0.28 (0.51) (0.51) 3 .1 1.891 A=137 or 136 Xe 0.60 0.63 1.30 26 .7 1.710 0.38 0.45 0.73 54 .0 1.110 93 Rb 0.50 0.85 0.72 80 .0 1.710 0.64 0.32 0.46 Data taken from refs. 7,25-28) (c) Improved Monte Carlo Calculations All liquid scintillation measurements of v are dependent on the accuracy of the calculation of the neutron leakage. Improved input data, and particularly the ability to specify more exactly the geometry of the detector systems, have afforded some recent improvement. (d) Dependence of the Capture Gamma Detection Efficiency on E and 8 It has normally been assumed that the capture detection efficiency is independent of the original properties of the neutron. This assumption was questioned at the 1972 Panel Meeting. Since then, Poitou and Signarbieux °' have investigated the validity of this assumption by also following the history of the capture gamma rays in a Monte Carlo calculation. They find a small variation in neutron capture detection efficiency with neutron energy which depends on the size of the scintillator. More specific calculations have been made by Ullo ' with particular reference to the liquid scintillator measurement 7) Particular application of these four sources to individual liquid scintillator measurements follows. 183 Spencer 18 ^ A new absolute determination is in progress at Oak Ridge National Laboratory. No details or data are yet available. 7 1 Boldeman ' The liquid scintillator in this measurement was spherical, 76 cm in diameter, and contained a loading of 0.5% by weight gadolinium. The axial hole through the scintillator had a diameter of 7.62 cm. The absolute calibration of the system was made as follows. (a) The relative neutron capture efficiency of the scintillator was calculated as a function of neutron energy and angle of emission with respect to the beam tube, using a Monte Carlo method. (b) The calculated efficiency values were normalised at the low energy end of the fission neutron spectrum in a measurement in which 2 MeV neutrons were scattered from a hydrogen gas target. Essentially, the normalisation was based on the detection efficiency for 0-1 MeV neutrons emitted at angles of 90-45 respectively, with respect to the scintillator axis. (c) It was assumed that there was no energy dependence of the neutron capture gamma ray detection efficiency. (d) The shape and absolute calibration of the energy dependence of the efficiency curve were checked in a second experiment in which 16 MeV neutrons were scattered from a poly- thene target. The detection efficiency of the system was verified within 1.5% to 8.73 MeV. In the original work, the fission neutron spectrum assumed for 252 Cf was the same as the present and therefore no correction is necessary. The effects of delayed gamma rays from fission were also adequately taken into account. An extensive re-analysis of items (c) and (d) has been carried out by Ullo 3 * ) . Most of the structural details of the scintillator were included in a Monte Carlo calculation of the neutron energy dependence of the scintillator, in which the histories of the capture gamma rays were also followed. The variation in leak- age with neutron energy and angle from the original calculation was confirmed. However, an energy dependence for the probability of neutron capture detection was determined. The relative probability of neutron capture detection for isotropic 252 Cf spontaneous fission neutrons to that for isotropic 0-1 MeV neutrons was computed to be 0.99 72. Furthermore, the relative probability of capture detection of 0-1 MeV isotropic neutrons to those of the calibration, 0-1 MeV between 90° and 45°, was estimated to be 0.9997. Thus this work estimates a total difference of 0.31% between the calibration neutrons and those for 5 Cf , compared with a +0.10% correction in the original paper. For the present paper, the analysis of Ullo 3 ' has been accepted and a further correction of +0.21% has been applied to the measured value. A further modification suggested by Ullo concerns the reliability of the estimate of the effect on neutron leakage caused by the axial hole. His somewhat pessimistic contribution of 0.3% to the experimental error compares with a value of 0.1% used in the original paper. A compromise figure of 0.2% has been used in this review. Table 3 lists the revised corrections for this measurement and all identified sources of error. The final value obtained for the prompt neutron emission is 3.74610.016 and 3.755±0.016 for the total neutron emission after the addition of the delayed neutron fraction. TABLE 3 CORRECTIONS TO EXPERIMENTAL DATA AND SOURCES OF ERROR FOR REF . Effect Correction (%) Contribution to Accuracy (%) Statistical accuracy Dead-time correction: 0.24 (a) 252 Cf (b) Low energy proton recoil (c) Relative Delayed gamma rays Fission neutron spectra: (a) Accuracy of E (b) Accuracy of energy calibration French Effect Effect of hole through scintillator on neutron leakage Background error in proton recoil counter Variation in neutron capture detector efficiency for 252 Cf fission neutron and calibration neutrons +1.107 +0.279 +0.828 0.10 -0.28 0.07 0.12 0.17 -0.10 0.10 0.20 (0.10) a) -0.33 0.10 +0.31 +0.05 +0.05 a) 0.10 0.05 0.05 a) b) Value used in original paper Two separate effects considered in original paper 2) Asplund-Nilsson et al .' In this experiment the liquid scintillator was 60 cm in diameter with a 6 cm diameter axial hole. The tank contained 110 I of liquid scintillator with a Cd/H atom ratio of 0.002. The absolute calibration of the neutron detector efficiency was carried out as follows : (i) (ii) For four incident energies (3.0, 4.3, 4.5 and 14.9 MeV) neutrons were scattered from an anthracene crystal at the centre of the scintillator and the neutron detection probability was determined for 28 values of (E n ,6). The highest energy for which the efficiency was determined was 10.75 MeV. It was assumed that there was no angular dependence for the neutron detection efficiency. Thus the detection efficiency for a 252 Cf fission neutron spectrum (T = 1.40 MeV assumed) was obtained by integrating over the measured energy dependence. 184 (iii) The efficiency figure so obtained was corrected for the effect of the axial hole using a Monte Carlo calculation. This measurement was down-weighted in the review by Axton ' partly because assumption (ii) was shown to be inappropriate. However, because the neutron detection efficiency was measured for such a large number of neutron energies and emission angles, it seemed reasonable to try to correct this deficiency in the original analysis. For the exact scintillator geometry, the probab- ility of neutron capture in both hydrogen and cadmium was calculated using the code of ref. 7) to be 0.9035 for an isotropic source of 252 Cf spontaneous fission neutrons [E = 2.15 Mev] . For each of the 28 efficiency measurements at a specific neutron energy and emission angle the probability of capture was also determined. Comparison of the experimental data with the calculated leakage, yields 28 values for the probability of neutron capture detection. The relevant data are listed in Table 4. TABLE 4 CALIBRATION DATA FROM ASPLUND-NILSSON ET AL ,. 2) Scattered Neutron Energy (MeV) Scattered Angle (degrees) Detection Probability (%) Calculated Capture (%) Effective e T (%) 0.24 73.6 72.6+1.5 99.25 73.1 0.65 62.3 74.7+1.2 98.92 75.5 1.07 53.3 74.8+1.1 97.75 76.5 0.15 79.5 70.9+1.5 99.46 71.3 0.50 70.5 72.9±1.4 98.97 73.7 0.97 62.3 73.1±1.5 98.34 74.3 1.40 56.1 73.4±1.5 96.79 75.8 1.86 50.0 71.5±1.4 94.01 76.0 2.43 42.7 72.0+1.4 90.18 79.8 2.75 38.6 67.3+1.2 87.30 76.9 2.93 36.2 67.7+1.4 88.15 76.8 0.21 77.2 73.2+1.6 99.27 73.7 0.58 68.5 73.9+1.6 99.00 74.6 1.00 61.2 72.7+1.7 98.19 74.0 1.82 49.4 72.1±1.6 94.79 76.1 3.17 30.8 65.4±2.0 83.75 78.1 0.48 79.7 71.611.7 99.21 72.2 1.65 70.6 72.3+1.7 96.02 75.3 1.79 69.7 72.8±1.8 95.48 76.2 2.18 67.5 70.2±1.6 93.22 75.3 2.49 65.9 72.411.8 91.49 79.1 3.01 63.3 69.911.7 88.86 78.7 3.46 61.2 66.812.1 88.47 75.5 4.27 57.6 62.911.8 84.51 74.4 5.42 52.8 57.511.7 74.47 77.2 6.69 47.9 51.511.8 68.04 75.7 7.94 43.1 51.311.7 66.67 76.9 10.75 31.9 40.111.6 48.83 82.1 It will be noted that the capture detection probability is fairly constant, although there is a suggestion of a small drop at low neutron energies, while the value at 10.75 MeV seems slightly large. An average value for the capture detection efficiency was obtained by weighting the numbers in Table 4 according to a fission neutron spectrum. The value obtained, 0.7595, when combined with the probability of neutron capture gives a value of 0.6862 for the overall neutron detection efficiency. Since the publication of the original paper, it has been reported 3 ' that a correction of -0.610.3% is necessary in this experiment to account for the French Effect ) . This is the bias that can be introduced into v measurements by the use of a coincidence between the fission counter pulse and a scintillator pulse from fission gamma ray detection and neutron induced proton recoils, to initiate the neutron counting gate. There are many experimental advantages in utilising this coincidence in liquid scintillator measurements. It was assumed that those fissions without a coincident pulse had the same average neutron emission as those with a coincident pulse. In the original identifica- tion of this problem, the data suggested a possible correction of the order of 1% or so. Subsequent work has shown the French Effect to be very small and it is only in this experiment that the effect is really significant. No correction was applied in the original paper for the contribution to the neutron count rate of the delayed gamma rays from fission. This was estimated for the 1972 Panel Meeting to be -0.2010.20% 34) This correction has been re-evaluated. The threshold on the neutron detector was set at the equivalent of approxi- mately 600 keV from gamma ray sources. Furthermore, the neutron counting gate began 160 ns after fission. With these parameters, the revised correction is -0.4310.2%. The final revised value from this experiment for the prompt neutron emission is 3.78310.040 and 3.792+0.040 for the total neutron emission. A full list of the experimental errors is given in Table 5. TABLE 5 SOURCES OF ERROR ( ASPLUND-NILSSON ET AL . 2 ' } Cause Error Statistical accuracy ( 252 Cf) 0.15 Statistical accuracy calibration 0.30 Corrections for proton recoil 0.20 Dead time correction 0.30 Delayed gamma rays 0.20 Fission neutron spectra (10.03 in E) 0.15 Accuracy of energy calibration 0.75 French Effect 0. 30 Effect of hole through scintillator 0.30 Variation in neutron capture 0.20 detection efficiency Hopkins and Diven 3 ^ This experiment used a cylindrical liquid scintillator, 1 m long and 1 m diameter. The 1000 I of liquid scintillator had a cadmium to hydrogen atom ratio of 0.002. For the absolute calibration of this detector, 3.9 MeV neutrons were scattered in a plastic 185 scintillator located at the centre of the 7 cm diameter axial tube. Neutrons within the energy range 0-1.3 MeV at an average emission angle of 72 were selected for the actual calibration by the bias on the plastic scintillator. The energy dependence of the neutron leakage was based, as before, on a Monte Carlo calcula- tion. In a second experiment to confirm the calculated energy dependence, 14.5 MeV neutrons were scattered in the plastic scintillator. The efficiency scale was confirmed for two energy groups, 0-2 MeV and 6-8 MeV. A number of queries regarding the accuracy of the calculated leakage probabilities were raised by Axton ' at the 1972 Panel Meeting. The authors subsequently confirmed the relative accuracy of their leakage calculation. Prior to the present paper, the leakage calculations were checked using the Monte Carlo code of ref . 7) . Our calculations would require the v value from this experiment to be adjusted downwards by 0.15%. This adjustment has not been made, as our calculations included none of the structural details of the scintillator. The original value of 3.771±0.031 was obtained using Bonner's measurement of the 252 Cf fission neutron spectrum. An adjustment of +0.11% is required to correct the efficiency for the assumed fission neutron spectrum of Table 1. No correction was applied in the original work for the effect of delayed gamma rays from fission. The threshold was set at approximately 1 MeV and it would appear that the neutron gate began approximately 300 ns after fission. Thus, relative to experiment 7), the efficiency for detection of delayed gamma rays compared to that for neutron capture gamma rays will be smaller. On the other hand, the earlier neutron gate increased the contribution. A reasonable estimate would appear to be 0.2±0.1%. There are unfortunately no data on which to base a correction for the variation in neutron capture detection between 252 Cf spontaneous fission neutrons and those used in the calibration (0-1.3 MeV neutrons). Poitou and Signarbieux 30) indicate that the variation of gamma ray detection with neutron energy is flatter for a 50 cm radius scintillator than for the 38 cm radius scintillator in experiment 7). Therefore, the effect should be smaller than the total effect of +0.31% calculated by Ullo 31 ' for experiment 7). Of course, the higher threshold in experiment 3) increases the effect relative to 7) , although this will be counter-balanced by the higher gamma ray energy emitted in cadmium capture versus that in gadolinium. In fact, if all the gamma ray energy is deposited in the scintillator, it is possible to conceive of a negative magnitude for this effect. Because of these considera- tions, it has been decided to make no correction at all. However, a contribution of 0.1% has been included in the experimental error of v for the spontaneous fission of 252 Cf. The final value obtained from the revision of this experiment is 3.768±0.031 or 3.777±0.031 for the total neutron emission. The Boron Pile (Colvin and Sowerby 1 ') In this experiment, the neutron detector was a 220 cm cube of graphite surrounded by a 35 cm thick reflector of graphite. The pile contained a lattice of 240 BF3 counters to detect the thermalised neutrons. The energy dependence of the leakage from the pile and therefore the relative efficiency of the system, was calculated by Pendlebury 36 ' using the Carlson S n technique (Fig. 1). For the calculation, the pile was idealised to a homogeneous sphere of graphite, 10 B and aluminium, with aluminium being used to simulate the copper bodies of the BF3 counters. The calculated energy dependence was normalised using four measure- ments in which the absolute detection efficiency was obtained for four different essentially monoenergetic sources. For the Boron Pile the reaction used to produce the neutrons was the photo-disintegration of the deuteron. Four different product neutron energies were obtained using four different gamma ray energies. The data are listed in Table 6. It will be noted that the calculated efficiency curve does not reproduce particu- larly well the variation of the calibration values. LU o0«9 u_ LU > LU 0-8 • • • • • • • ULL0 40) AXTON 8) PENDLEBURY 36) 1 1 01 23456789 10 NEUTRON ENERGY (MeV) Fig. 1: Efficiency calculations for the Boron Pile TABLE 6 CALIBRATION MEASUREMENTS FOR BORON PILE 1) Measurement Neutron Effic- Energy iency Error (MeV) (%) D(y,n)p with Th C" y-rays D(Y,n)p with 24 Na Y~ r ays D(y,n)p with 19 F (p,a,y) y-rays 2.0 D(y,n)p with 27 A1 (p,y) Y~rays Ra-Y-Be k Source calibrated by N.P.L. MnSOi, bath 38 ' 19 6440 60 265 6457 21 2 6433 25 4 9 6500 1 24 25 6475 74 Because the v value obtained, 3.713±0.015, was so different from existing measurements, the Boron Pile was subjected to close scrutiny. Colvin et al. have answered a number of criticisms that have been made. Furthermore, a number of independent checks of the efficiency of the Boron Pile were made. For example, an independent measurement was made using a Ra-y-Be source calibrated by N.P.L. using a MnSOi, bath 38 '. This value, effectively for neutrons at 250 keV, has been included in Table 6. Some recent objections have been voiced by 186 Leonard 15) (a) (b) The error assigned to the correction for the proportion of neutrons detected outside the gate period should be increased from the original assignment of 0.1%. The proportion of neutrons outside the 4 ms gate period is given by Colvin et al. 37 ' as 4.3±0.3%. From the data in Table V of ref . 37) , the correc- tion would appear to be at least as accurate as this. Thus, for the comparison of the Ra-y-Be source measurement of the neutron detection efficiency with that from the D(y,n) reaction, this correction must be applied and the full error in the correction included in the list of errors. However, for the comparison of 252 Cf fission neutrons and D(y,n) reaction neutrons, it is the relative difference in the correction which is important. It is difficult to see how this could be any larger than the error of 0.1% assigned for this effect. It should be noted that a similar effect exists in all liquid scintillator measurements (1-2% loss outside the gate). In these cases, no contribution to the experimental error has been included. Loss of neutrons in the Boron Pile caused by Cu(n,p), Cu(n,a) and C(n,a) reactions was neglected in the efficiency calculations. Certainly the first two reactions were ignored; however Sowerby ' estimates their effect as less than 0.1%. The effect of the third reaction has also been estimated by Sowerby to be 0.15±0.05%. It is thought that the effect of this reaction was included in the original calculations; however it has not been possible to verify this. Because the measured values of the efficiency calculations do not fit the calculated energy dependence, some subjective judgement is involved in choosing the best method of normalisation of the curve. In fact, it is fairly obvious that a recalculation of the energy dependence of the neutron detection efficiency of the Boron Pile is required, especially when the age of the original calcu- lation is considered. A new calculation would also eliminate any objection on account of (b) above. For the 1972 Panel Meeting, Axton 8 ' recalculated the energy dependence of the Boron Pile using a Monte Carlo calculation. His calculated efficiency curve is shown in Fig. 1, together with the original calculation. Use of this efficiency curve, rather than the Pendle- bury calculation, leads to a value for the detection efficiency of the Boron Pile for 252 Cf spontaneous fission neutrons of 0.64145 versus the original value of 0.6428. Also shown in Fig. 1 is a more recent Monte Carlo calculation of the relative neutron energy dependence of the Boron Pile neutron detection effi- ciency by Ullo °' . This work has not yet been published, but some details are available. For neutron transport the energy range from 5 kev to 10 MeV was spanned by 3000 energy points and inelastic scattering was handled implicitly. The angular distribution for elastic scattering in carbon was described with four Legendre components in the centre of mass. The cross sections for Cu(n,p), Cu(n,a) and C(n,a) were included in the calculation. Finally, some structural details of the Boron Pile were included. The calculated efficiency curve is also shown in Fig. 1. (c) There is a most unfortunate lack of agreement between the three different calculations. The two Monte Carlo calculations give the most similar dependence. In fact, it is likely that the difference between them can be partly attributed to slightly incomplete data sets employed in the calculation of S) The difference between the original efficiency Axton curve and that of Ullo is quite marked. The latter calculation does not show the rapid rise in efficiency at low energies, but shows a much more dramatic fall in efficiency at high neutron energies than does the Pendlebury calculation. The overall effect of the use of the Ullo efficiency curve is to increase the measured value of v. For this review the Ullo efficiency curve has been used for two reasons. Firstly, the input data set was considerably more comprehensive. Secondly, in the calculation of the energy dependence of the liquid scintillator of ref. 7) the Ullo code produced values in substantial agreement with two other Monte Carlo calculations '' and in agreement with the experimental data. However, it is important to resolve this question of the energy dependence of the Boron Pile as soon as possible. For the normalisation of the calculated efficiency curve , a weighted value has been obtained from the experimental values at 0.190, 0.265, 2.0 and 4.9 MeV from the D(y,n) measurements, plus the value at 0.250 MeV from the Ra-y-Be source measurement. The weighted efficiency obtained for a 252 Cf fission neutron spectrum is 0.637910.0010. The error here is the direct experi- mental error. Because of the failure of the experi- mental data to reproduce satisfactorily the calculated energy dependence of the efficiency curve, some contri- bution to the experimental error is necessary. Unfortunately, this contribution is somewhat subjective. A value of ±0.3% has been used which compares with an effective value of 0.25% used by Colvin and Sowerby 1) Colvin and Sowerby have indicated that they are prepared to accept, provisionally, the recalculated efficiency curve for the Boron Pile, pending an examina- tion of the details of the calculation. The revised (provisional) value of v for 252 Cf from the Boron Pile is therefore 3.732±0.016 and 3.741±0.016 for the total neutron emission. All contributions to the experimental error are listed in Table 7. TABLE 7 SOURCES OF ERROR (BORON PILE 1 ') Cause Contribution to Experi- mental Error (%) Statistical accuracy of calibration Statistical accuracy 252 Cf Anisotropy in pile efficiency Beam modulation Electrical pick-up Neutrons after gate Fission spectrum (error in E) Lack of agreement of calculated and experiment shape of energy dependence of efficiency a) 0.15 0.11 0.20 0.05 0.03 0.10 0.05 0.30 Includes contribution from error in correction for loss of neutrons caused by chamber materials 187 l - One objection that might be levelled at any revision of the Boron Pile v value for Cf is that the old value was in agreement with values obtained using calibrated neutron sources to determine the efficiency of the pile. Colvin et al 37) lists these measurements in Table II of that reference. Three different neutron sources were employed, namely, the AERE and AWRE 240 Pu sources and a Ra-y-Be source. The latter source, which was calibrated using the N.P.L. MnS(\ bath, has already been included in the absolute normalisation of the efficiency scale. 240 For the calibration of the Boron Pile using the Pu sources, it is necessary to make a correction for the energy difference between 25 Cf spontaneous fission neutrons and those for spontaneous fission of 240 Pu if the efficiency curve of Ullo is appropriate. A correction of 0.25% is required in this case. For these source calibrations three independent efficiency values can be derived for the Boron Pile based on three independent methods of source calibration, namely, (a) the revised original calibration *' of the AERE 240 Pu source; (b) the AWRE oil bath 42 ' calibration of the AWRE 2i+0r Pu source; (c) the calibration of the AWRE using the N.P.L. MnSOit bath. 240 Pu source After applying the 0.25% correction referred to above, the efficiency of the Boron Pile for 252 Cf fission neutrons derived using the source calibrations (a) and (b) are - 6417 S'mn! "^ -■ ^d 0.6396 ±0 - 0006 random ±0.0104 systematic ±0.0087 systematic' The average of these two values leads to a value of v for 252 Cf of 3.726±0.039. The error was derived assum- ing the original errors were independent. This is probably not the case, and the actual error may be fractionally larger. It is evident that the value of v derived using these two^source calibrations is con- sistent with the revised v value from the Boron Pile experiment itself. Finally, the value of v for Cf derived using Pu source the N.P.L. calibrated value for the AWRE 2<+0 is 3.697±0.034. The difference between this value and the revised Boron Pile value (slightly larger than one standard deviation) is not sufficient to raise any objections to the present revision. This third source value is more appropriately incorporated with the N.P.L. band measurements. Manganese Sulphate Bath Measurements The absolute measurement of neutron source strengths with manganese sulphate baths will be dis- cussed in considerable detail at this conference by Axton . Consequently, few of the details of the actual neutron counting will be given here and more attention will be devoted to the fission counting. l 7) Bozorqmanesh ' A new MnSOi, bath determination of v for the spont- aneous fission of 252 Cf has apparently been completed at the University of Michigan. No details have been published as yet. The reported value, 3.744±0.023 for the total neutron emission, has been accepted for this review, together with the nominated error. Measurements based on N.P.L. MnSO u bath 5 ' 8 ' 1 * 1 * >'* 5) Two separate absolute determinations 8 ' 1 * 5 of v for 2 52 Cf have been made in which the neutrons were counted 188 in the N.P.L. MnSOit bath, but the method of fission counting varied. Each is discussed separately below. Axton et al . 5 ' 8 ' In this experiment a series of thin 52 Cf spontan- eous fission sources were aliquotted from a stock solu- tion of californium chloride and the absolute fission rates were counted in a pill box type gas flow propor- tional counter. The method employed to provide a satisfactory correction for fragment losses in the foils was discussed in detail in ref . 5) . The neutron emission rate of the remaining californium chloride solution was then determined in the N.P.L. MnSOi^ bath. This process was repeated a number of times for four separate californium samples. A total of 20 separate neutron sources and 200 fission sources were prepared. The final value for the total neutron emission obtained in this experiment was 3.725±0.019 where the experi- mental error includes only the error on the fission counting. White and Axton 1 * 5 ' For this determination, the fission and neutron emission rates were obtained for the same source. The neutron emission rate was measured in the N.P.L. MnSO^ bath and the fission rate was determined using a low geometry fission counter at Harwell. At the 1972 Panel Meeting, Axton 8 ' revised the accuracy of the fission counting. For the present review the revised value is retained. The value for v obtained was 3.797±0.038 where the error, as before, includes only that for the fission counting. Other N.P.L. MnSOi t bath dependent measurements A number of other measurements of v for 5 Cf are generally included in the list of N.P.L. dependent measurements. One of them was the calibration of the Boron Pile with a Ra-y-Be source and the AWRE 21 *°Pu source, both of whose activities were calibrated in the N.P.L. MnSOit bath. In the present report, the first has been included as part of the absolute calibration of the efficiency curve for the Boron Pile. Only the second source measurement, revised previously to 3.697±0.034, is included here with the N.P.L. dependent measurements . Another addition was a measurement from Moat et al. ' in which the neutron emission rate of a californium sample was determined using two cylindrical wax detectors of different dimensions, each containing several BF3 counters for the detection of thermalised neutrons. The efficiency of the detector was determined using the Harwell 21 *°Pu source which was calibrated using the N.P.L. MnSO^ bath, the Boron Pile and the Aldermaston oil bath. The original value of 3.77±0.07 for v for 252 Cf was corrected by Fieldhouse et al. 1 * 1 ' to 3.685±0.040 in a revision of the Harwell 21+0 Pu source and then to 3.727±0.056 by Hanna et al. 6 ' to account for an improved estimate of the difference in the 21 *"pu and 252 Cf fission neutron spectra. Because of the very strong dependence of this measurement on the "softened" 240 Pu spontaneous fission neutron spectrum, it has not been included in the final list of values . Average value from N.P.L . The average of the three N.P.L. dependent measure- ments is 3.731±0.015 for v for 252 Cf. The error here includes only that from the fission counting. The error for the neutron source calibration in the N.F.L. bath was given by Axton error becomes ±0.020. I) as 0.33%. Thus the total A small revision to the N.P.L. value of v for 252 Cf has been recommended by Smith et al. 4 *"' follow- ing their accurate measurement of the Ou/a.^ ratio. They used the variable manganese concentration technique developed by Axton et al . ' , but for the neutrons employed a Bragg reflected beam of approximately 0.02 eV. Thus corrections required for the *°0(n,a) C and 32 S(n,p) 32 P are eliminated, together with any dependence on resonance capture data for 55 Mn. Recent experimental work 48 ' can be used to predict considerable intermediate structure in the Mn(n,y) cross section and specifically a strong valence component which could lead to a radiative width for the 2.375 keV resonance as large as 2 eV. The value they obtained, 0.02503±0.23% , compares with the revised value from Axton et al. 41 *' of 0.02495±0. 35%. Use of the Smith et al. value would lead to a positive adjustment of 0.15% in the N.P.L. v value. For the purposes of this review, the N.P.L. v value has provisionally been adjusted by +0.15% to 3.737±0.020. De Volpi and Porges 4 ' 49 /50) For this experiment the experimental procedure was as follows. Three separate fission counters were prepared in which the californium sources, all yielding approximately 10 8 neutrons/s, were deposited in three different ways on one electrode of a fast ionisation chamber with a plate spacing of 1-2 mm and operated in the current mode. The neutron emission rates were measured by inserting the fission chambers in the A.N.L. manganese sulphate bath. Subsequently, the absolute fission rates for each of the three fission counters were determined in a coincidence system in which the fission fragments were counted in coincidence with prompt fission neutron detection in a Hornyak button^* ' . The coincidences were recorded as a function of the angle (6) between the fission counter normal and the neutron detection axis. The fission rate was given by NF/C where N is the neutron count rate, F is the fission rate and C the coincidence rate. The angular anisotropy of the effective fission rate was removed by averaging the four measurements at 6 = ±37.4 and ±79.2. Although three different counters were employed, the final value of v was essentially based upon one counter (parallel plate) which had a fission fragment detection efficiency >99%. In addition to the coincidence method, the fission rate was also measured in a variety of other ways, including low geometry counting and simple extra- polation of the bias curve. Despite the considerable care that has been taken in arriving at the absolute fission rates, the estimated error for the third fission counter, 0.13%, appears optimistic. From Table 6 of ref. 4), the two sources of error for fission counter 3, namely the error of 0.13 for counter 3 in particular and that for the error common to the three detectors, appear not to have been added. Furthermore, two of the entries in the common errors, those for neutron detection efficiency dependence on v and fission detection efficiency dependence on v, could each be increased to 0.1%. In addition, there does not appear to be any component added for the distribution in nodal values listed in Table III of ref. 4) . The error has provision- ally been expanded to 0.23%. This compares to an effective fission counting error in the summed N.P.L. measurements of 0.42%. Axton 8 ' in his 1972 review recommended the intro- duction of an additional 0.5% error into the neutron source calibration in the MnSOit bath to account for aliquotting errors. De Volpi, in a private communica- tion to the 1972 Panel Meeting, answered criticisms on this score, and the 0.5% error has therefore not been included in the list of errors in this review. However, from the text of ref. 8) and De Volpi ' s reply, there appears to be a number of small unresolved differences between A.N.L. and N.P.L. Finally, it should be noted that the measured value of o H /a Mn from A.N.L. of 0.2531±0. 00003 differs significantly from the measure- ment of 0.02503±0. 00006 by Smith et al. 47 '. The explanation proposed by the authors ' to explain their high value, impurities in the solution, could introduce an unknown correction as would any densimetric error. No error has been introduced here on this account. The final value obtained using the A.N.L. MnSO^ bath is therefore 3.729±0.017. Summary of Absolute v Measurements for 252 <:f The revised values of v for the spontaneous fission of 252 Cf from the eight measurements are listed in Table 8. The average of these measurements, weighted according to the experimental error, is 3.745±0.008. The errors are not all independent and the accuracy should be expanded to ±0.010 to include various common errors. This revised value represents an increase of 0.32% in the value recommended by the 1972 Panel Meeting. TABLE 8 RE-EVALUATED VALUES OF V FOR 252 Cf Measurement Total Neutron Emission Original Total Neutron Emission Revised Liquid Scintillator Measurements : Boldeman 7) Asplund-Nilsson et al . Hopkins and Diven 3 ' Boron Pile : Colvin and Sowerby 1 ' 2) Fieldhouse et al. Hi) Manganese Sulphate Baths : 17) Bozorgmanesh ' Axton et al. 8,h5) De Volpi and Porges 1+9,50) 3.747±0.015 3.808±0.034 3.780±0.031 3.713±0.015 3.728+0.019 3.729±0.015 3.755±0.016 3.792+0.040 3.777±0.031 3.741±0.016 3.726±0.039 3.744±0.023 3.737±0.020 3.729±0.017 The most important question to decide is whether the weighted error represents the true error of the summed eight measurements, or whether there is any evidence within the final numbers for a possible experiment-dependent systematic error. Of the eight measurements, only that from Asplund-Nilsson et al. ' lies more than one standard deviation from the mean and certainly the external error of ±0.006 is less than the internal error. The weighted average of the three liquid scintillator measurements is 3.763±0.014 which is approximately one and a half standard deviations of the comparison displaced from the average of the three manganese sulphate bath measurements [3 . 735±0.01lJ . Alternatively, the average of the four gated measure- ments, 3.754±0.010, compares satisfactorily with the four source measurements. Thus any suggestion in the original data of an experiment dependent systematic error has now disappeared. Therefore it is considered that the error of ±0.010 represents a legitimate estimate of the accuracy of the standard. The recommended value for the total neutron emission from the spontaneous fission of 25 Cf is 3.745±0.010 189 3. The v Ratios Most previous reviews of the ratios of neutron emission for neutron fission of 233 U, 235 U, 239 pu arK i Pu to that for the standard, spontaneous fission of Cf, e.g. refs. 6 and 16) have concluded that the different measurements were in satisfactory agreement. However, it will always be necessary to effect those small revisions that become apparent with the general improvement in nuclear data. The principal revisions made in this paper include those for improvements in the measured fission neutron spectra (Table 1) and in the delayed gamma ray from fission data (Table 2) . The correction, consider- ed for the standard, to account for variation in the neutron capture detection efficiency with emitted neutron source energy can be safely ignored as it will be at least an order of magnitude smaller here (i.e. <0.01%) . A new correction that must be considered for all of the v ratio measurements arises from the resonance dependence of v. None of the measurements that are generally included in the evaluation of the v ratios were made with monoenergetic neutrons and thus the measured value has some small sensitivity to the exact shape of the incident neutron spectrum. Since the existing experimental data on the resonance dependence of v is neither complete for the four important isotopes nor always consistent, a brief review is included here of the data and the likely character of the effects which are contributing to the variation. Resonance Dependence of v Work on the resonance dependence of v was stimu- lated by the conflicting evidence presented at the second IAEA Symposium on the Physics and Chemistry of Fission. For the neutron fission of 239 Pu, Weinstein en) — et al. ' found that v values for 20 resolved resonances below 100 eV fell into two groups strongly correlated with the resonance spin. The average value of v for resonances with spin J = was 3% higher than the average for resonances with spin J = 1. A smaller resonance effect was noted for 235 U. Ryabov et al.-* 3 ' observed the opposite effect. For resonances in 239 Pu they found v for J = 1 resonances was 5% larger than that for J = resonances. A subsequent experiment by Weston and Todd 5 ' confirmed neither experiment. They found that v for the two spin classes was similar within /i+%. However, they observed a fluctuation in the different v values outside the statistical accuracy of the measurements. Two later measurements »' in substantial agreement with ref. 54) gave rise to an explanation 5 '' of most of the fluctuations in v values in terms of the (n,yf) process. Essentially, they found that the v values for the 23 °Pu resonances were inversely correlated with the fission widths. Fairly clearly, v is smaller for the (n,yf) reaction than for direct fission. Furthermore, because the (n,yf) process is a multichannel process, its width is fairly constant. Thus v for resonances with small fission widths should also be small because of the increased relative importance in these cases of the (n,yf) process. The relevance of the resonance energy dependence of v to the evaluation of the thermal value will be apparent from Fig. 2 from Leonard 5 ' where the v energy dependence near thermal neutron energies has been plotted from the data of refs. 52,59-61). The resonance at 0.296 eV has a considerable impact on the energy dependence. The value of v for this resonance is approximately 0.029 neutrons less than that for thermal fission. From the data in ref. 56), 0.011 of this depression can be expected because of the (n,yf) process. The remainder, 0.018, is similar in magnitude to the difference, 0.014±0.007, found by Frehaut and Shackleton DD ' for the difference in v for J = and J = 1 resonances with the effects of the (n,yf) process removed. Two effects can be expected to contribute to this difference. Cowan et al. 62 ' have shown that the symmetric fission yields for J = 1 + resonances are about 1/3 those for J = . Since symmetric fission fragments emit more neutrons 2 ' , v for J = + resonances should be larger than that for J = 1 + . However, this effect is likely to be extremely small. For 235 U, Howe et al. 63 ' have found no correlation between resonance v values and the measured mass asymmetry 61 ) . A second contribution arises because of the considerable difference in the fission barrier for J = + compound states relative to that for J = 1 + states . It has been shown that the fine structure in the v p (E n ) dependence between 0-1 MeV for neutron fission of 2 33^64) can ^e explained if the collective energy at the fission saddle point is weakly coupled to the nuclear degrees of freedom at scission, and appears predominantly in the fission fragment kinetic energies. The fission barrier for J = 1 com- pound states (Kir = 1 + ) is displaced by at least 1.2 MeV with respect to the ground state fission band (Kir = ) . Thus one might expect the average total fission fragment kinetic energy for the fission of a J = 1 + compound state to be as much as 1.2 MeV larger than that for the fission of a J = + compound state. From energy con- servation, this is equivalent to a difference of approximately 0.15 neutrons. In fact, the measured effect is very much less than this. The reduction is readily explained. The measurements of Bach et al. ' have shown that the inner peak of the fission barrier is the higher for the 2 ^°Pu compound nucleus. Thus the difference in the average collective energy for the two J states is a property of the first barrier. It is well established that mixing of K states occurs in crossing the intermediate well and the outer peak of the fission barrier 66) Thus only a small component of the original effect survives. Possible resonance effects can now be evaluated for neutron fission of the other fissile nuclei. For 233 U/ the (n,yf) process should have a width similar to that in the neutron fission of 239 Pu. Here, however, there are fully open fission channels for both J = 2 and J = 3 + compound states and the (n,yf) process for both spin populations is only a minor effect. Similarly, it can be expected that any effect caused by the variation in the mass distribution is extremely small. However, using the data of ref. 64) it can be shown that v ( J = 2 + )-v(J = 3 + ) should be approximately 0.009. Thus it is expected that there should be a difference of <0.0045 between v at 0.0253 eV and that for the resonance at 0.19 eV where the spin has not been identified. For the neutron fission of 235 U, the fission barrier for both the 3 and 4~ spin states is consider- ably higher relative to the neutron binding energy than that for neutron fission of either 233 U or 239 Pu. Thus on these grounds alone, the width for the (n,-yf) process is extremely small. As indicated previously, Howe et al. 63 ' have shown that v is also uncorrelated with the resonance mass asymmetry. The data from ref. 64) can also be used to show that collective effects give rise to a difference in v for the two spin states of less than 0.001. Similar arguments can be used to show that resonance effects in v for neutron fission of 2 *Pu are also very small, in agreement with the experimental data from Simon and Frehaut 6 7) . Thus the only case where care must be exercised is in the evaluation of v for thermal neutron fission of 239 Pu. 190 v Ratios In the consideration of the v values for thermal neutron fission there are a number of different approaches that may be adopted. In all cases, the values have been measured relative to 252 Cf, but for a number of experiments, e.g. 1 and 7) in particular, it is possible to eliminate the standard altogether and refer the neutron counting rate for thermal fission directly to the absolute calibration. This procedure has been adopted in the past because of the apparent disagreement between different absolute measurements for 252 Cf. However, since the revised v data for 252 Cf show a very acceptable level of consistency, it is more appropriate to evaluate the ratios and refer these values to the average experimental value of v for 252 Cf. A brief discussion follows of the corrections that have been made for specific experiments. Boldeman and Dalton 68 ' The adjustments applied to the measured v ratios are listed in Table 9. For the evaluation of the delayed gamma ray correction, the measured delay of 585 ns between the fission event and the opening of the neutron counting gate was used. The listed delayed gamma ray corrections are those relative to the correc- y R 9 tion for Cf. The correction for the resonance dependence of v corrects the measured result to the equivalent 2200 ms value. For the only case where this correction was necessary, neutron fission of 239 Pu, the correction was based on the data of Fig. 2 and a calculated neutron leakage spectrum from a well thermalised graphite system at 300 K. The estimated correction is +0.10%. TABLE 9 REVISION OF v RATIOS 2 3 3 r 2 3 5, 239 Pu 241 Pu Boldeman and Dalton 68 ' Fission spectra Delayed y-rays Resonance dependence Mather et al. 69 / 70) Fission spectra Delayed y-rays Colvin and Sowerby ' Fission spectra Resonance dependence Cond€ 71) Fission spectra Delayed y-rays 3) Hopkins and Diven Fission spectra Delayed y-rays +0.27 +0.21 +0.21 +0.21 -0.16 -0.16 -0.43 -0.43 +0.10 -0.04 -0.12 +0.01 -0.20 -0.20 -0.54 -0.06 -0.06 -0.03 -0.03 +0.10 +0.06 -0.25 -0.08 -0.10 -0.10 -0.10 -0.30 Mather et al . 69 > 70 > The liquid scintillator used in these experiments was identical with that of ref . 7) . The adjustments for the fission neutron spectra differences were based on the data from this experiment. There is not sufficient data in the original papers to make an accurate estimate of the contribution of the delayed gamma rays from fission. However, the correction will be slightly larger than that for ref. 7) because of the earlier (undefined) opening of the neutron counting gate. Corrections 1.25 times those applied for ref. 68) have been used. The corrections are listed in Table 9. V 0-05 0-1 ENERCY lev) Fig. 2: Energy dependence of v for 239 Pu from Leonard 58 ' Colvin and Sowerby ' A very small correction for fission neutron spectra differences is necessary if the Ullo efficiency curve is used. The thermal v measurements were made using a beam of thermal neutrons from GLEEP. For 239 Pu an identical correction to that applied in ref. 68) has been used for the resonance variation of v. Conde 7 1) The fission neutron spectra difference correction has been revised using the calculated energy dependence referred to in the section dealing with the absolute measurement of Asplund-Nilsson 2 ' . The delayed gamma ray correction is also based on the data from this section. Hopkins and Diven 3 ) The adjustments applied in this experiment for fission spectra differences were based on the assump- tion that the original corrections were derived usin< ? 3 S the fission neutron spectra measurements of Bonner 03 The thermal neutron measurements were made using a 400 keV neutron beam moderated by a polyethylene plug in the neutron collimator. A correction was applied for fission by neutrons above a cadmium filter cut-off. The correction for the resonance dependence of v is therefore extremely small. Comparison of Ratios Table 10 lists the revised ratios from the above experiments. The listed ratios are those for prompt neutron emission. The consistency noted in previous evaluations still exists and there is no evidence with- in the sets to suggest any systematic error. 4 . Recommended Values Table 11 lists the recommended values and their errors from the revision above. The delayed neutron components have been taken from Lemmel 16) . The conclusion to be drawn from this review of existing v data is that there is no evidence within the data of any significant systematic error. 191 TABLE 10 V RATIOS -P Experiment 233 235 U 239 Pu 2"+l Pu Boldeman and Dalton 68 ' Mather et al. 69 » 70) Colvin and Sowerby 1 ' Cond^ 71) Hopkins and Diven 3 ' 0.6593+0.0015 0.6388±0 . 0012 . 7662±0. 0021 0.7771+0.0019 0.668 ±0.008 0.6357±0.0032 0.771 ±0.009 0.6547±0.0039 0.6392±0 . 0029 . 7598±0 . 0051 0.7746+0.0071 0.6388±0.0053 0.6546±0.0058 .6418±0 .0053 0. 7485±0.0074 Average 0.6587±0.0013 .6386±0. 0010 . 7647±0 . 0018 . 7769±0 .0018 TABLE 11 RECOMMENDED V VALUES Isotope 233 U 2.461±0.008 2.468±0.008 2.386±0.007 2.402±0.007 Pu 2.857±0.010 2.863+0.010 Pu 2.902±0.010 2.918±0.010 Cf 3.736±0.010 3.745±0.010 235r 239 241 252 Acknowledgements I wish to acknowledge the contributions to this report of many authors too numerous to mention. I am particularly indebted to J. J. Ullo, B. R. Leonard Jr., E. J. Axton and J. R. Smith for permission to use unpublished data. References 1) D. W. Colvin and M. G. Sowerby, Proc. IAEA Symp. Physics and Chemistry of Fission, Salzburg (1965) Vol. 2, p. 25. 2) I. Asplund-Nilsson, H. Condi* and N. Starfelt, Nucl. Sci. and Eng. 16 (1963) 124. 3) J. C. Hopkins and B. C. Diven, Nucl. Phys . 48 (1963) 433. 4) A. De Volpi and K. G. Porges , Phys. Rev. Cl (1970) 683. 5) E. J. Axton, A. G. Bardell and B. N. Audric, J. Nucl. Energy A/B 23_ (1969) 457. 6) G. C. Hanna, C. H. Westcott, H. D. Lemmel , B. R. Leonard Jr., J. S. Story and P. M. Attree, Atomic Energy Review 7_ (1969) No. 4, p. 3. 7) J. W. Boldeman, Nucl. Sci. and Eng. 55_ (1974) 188. 8) E. J. Axton, Proc. IAEA Panel Neutron Standard Reference Data, Vienna (1972) p. 261. 9) J. R. Smith, Proc. Conf. Nuclear Cross Sections and Technology, Washington (1975) Vol. 1, p. 262. 10) J. A. Mitchell and G. J. Emert, Proc. 3rd Conf. Neutron Cross Sections and Technology, Knoxville (1971) , Vol. 2, p. 605. 11) J. J. Ullo and M. Goldsmith, Nucl. Sci. and Eng. 60 (1976) 239. 12) M. Goldsmith and J. J. Ullo, Nucl. Sci. and Eng. 60 (1976) 251. 13) J. R. Smith, S. D. Reeder and R. G. Fluharty (1966) Phillips Petroleum IDO-17083. 14) R. L. Macklin, G. de Saussure, J. D. Kington and W. S. Lyon, Nucl. Sci. and Eng. 8 (1960) 210. 15) B. R. Leonard Jr., Proc. Conf. Nuclear Cross Sections and Technology, Washington (1975) Vol. 1, p. 281. 16) H. D. Lemmel, Proc. Conf. Nuclear Cross Sections and Technology, Washington (1975) Vol. 1, 286. 17) H. Bozorgmanesh, private communication to B. R. Leonard Jr. 18) R. R. Spencer, private communication. 19) R. Gwin, private communication. 20) B. C. Diven, H. C. Marlin, R. F. Taschek and J. Terrell, Phys. Rev. 101 (1956) 1012. 21) P. I. Johansson, these proceedings. 22) J. M. Adams, NEANDC (E) -172 , (1976). 23) A. B. Smith, Proc. Meeting Prompt Fission Neutron Spectra, Vienna (1971) p. 3. 24) J. Terrell, Symp. Physics and Chemistry of Fission, Salzburg (1965) Vol. 2, p. 3. 25) F. W. Guy, Laurence Radiation Laboratory report, UCRL-50810 (1970) . 26) N. N. Ajitanand, Nucl. Phys. 164 (1971) 300. 27) R. B. Walton and R. E. Sund, Phys. Rev. 178 (1969) 1894. 28) J. W. Gruter, K. Sisternick, P. Armbruster, J. Eidens and H. Lawn, Phys. Rev. Lett. 33B (1970) 474. 29) J. P. Unik, J. E. Gindler, L. E. Glendenin, K. F. Flynn, A. Gorski and R. K. Sjoblom, Proc. 3rd. IAEA Symp. Physics and Chemistry of Fission, Vienna (1973) Vol. 2, p. 19. 30) J. Poitou and C. Signarbieux, Nucl. Instr. and Methods 114_ (1974) 113. 31) J. J. Ullo, Bettis Atomic Power Laboratory report WAPD-TM-1232 (1976) . 32) H. Conde, J. Hansen and L. Widen, Proc. Conf. Neutron Standards arid Flux Normalisation, Argonne (1970) 481. 33) M. Soleilhac et al. - refer D. W. Colvin, Proc. -Conf . Neutron Data- for Reactors, Helsinki (1970) Vol. 2, p. 195. 34) H. Cond^ , Proc. Panel Neutron Standard Reference Data, Vienna (1972) p. 277. 35) T. M. Bonner, Nucl. Phys. 2_3 (1961) 116. 36) E. Pendlebury, private communication to Colvin and Sowerby i5 192 37) 38) 39) 40) 41) 42) 43) 44) 45) 46) 47) 48) 49) 50) 51) 52) 53) 54) D. W. Colvin, M. G. Sowerby and R. I. MacDonald, Proc. Conf. Nuclear Data for Reactors, Paris (1966) Vol. 1, p. 307. E. J. Axton, private communication to Colvin et al."~ 37) M. G. Sowerby, private communication. J. J. Ullo, private communication. P. Fieldhouse, E. R. Culliford, D. S. Mather, D. W. Colvin, R. I. MacDonald and M. G. Sowerby, J. Nucl. Energy A/B £0 (1966) 549. P. Fieldhouse, E. R. Culliford and D. S. Mather, (1966) , private communication to 37) . E. J. Axton, Proc. this conference. E. J. Axton, P. Cross and J. C. Robertson, J. Nucl. Energy A/B 19^ (1965) 409. P. White and E. J. Axton, J. Nucl. Energy 22 (1968) 73. A. Moat, D. S. Mather and M. H. McTaggart, J. Nucl. Energy A/B 15 (1961) 102. J. R. Smith, S. D. Reeder, B. M. Cawsey and V. Spiegel, private communication. J. R. Bird, B. J. Allen, J. W. Boldeman, M.- J. Kenny and A. R. de L. Musgrove , Proc. Int. Conf. Interaction of Neutrons with Nuclei, Lowell (1976) Vol. 1, p. 77. A. De Volpi and K. G. Porges, Metrologia 5_ (1969) 128. A. De Volpi, J. Nucl. Energy 26_ (1972) 75. W. F. Hornyak, Rev. Sci. Instr. 23. (1952) 264. S. Weinstein, R. L. Reed and R. C. Block, Proc. 2nd IAEA Symp. Physics and Chemistry of Fission, Vienna (1969) p. 447. Yu. V. Ryabov, So Don Sirk, N. Chikov and N. Yaneva, Sov. J. Nucl. Phys. 1£ (1972) 519. L. W. Weston and J. H. Todd, Proc. 3rd Conf. Neutron Cross Sections and Technology, Knoxville (1971) Vol. 2, p. 861. 55) 56) 57) 58) 59) 60) 61) 62) D. Shackleton, J. Trochon, J. Frehaut and M. Le Baro, Phys. Lett. 42B (1972) 344. J. Frehaut and D. Shackleton, Proc. 3rd IAEA Symp. Physics and Chemistry of Fission, Vienna (1972) Vol. 2, p. 201. Yu. V. Ryabov, J. Trochon, D. Shackleton and J. Frehaut, Nucl. Phys. A216 (1973) 395. B. R. Leonard Jr. (1976), private communication. B. R. Leonard Jr., E. J. Seppi and W. J. Friesen, HW-4344. W. Hockenbury, private communication to R. Leonard Jr. R. Gwin, R. Ingle, R. R. Spencer, J. H. Todd and L. W. Weston, private communication to B. R. Leonard Jr. G. A. Cowan, B. P. Bayhurst, R. J. Prestwood, J. S. Gilmore and G. W. Knobeloch, Phys. Rev. 144 (1966) 979. Howe, T. W. Phillips and C. D. Bowman, Rev. C13 (1976) 195. 63) R. E Phys 64) J. W Nucl 65) B. B J. D Boldeman, W. K. Bertram and R. L. Phys. 265 (1976) 337. Walsh, Bach, 0. Hansen, H. C. Britt and Garrett, Proc. 3rd IAEA Symp. Physics and Chemistry of Fission, Rochester (1973) Vol. 1, p. 25. 66) V. M. Strutinsky and H. C. Pauli, Proc. 2nd IAEA Symp. Physics and Chemistry of Fission, Vienna (1969) p. 155. 67) G. Simon and J. Frehaut, Proc. 3rd National Soviet Conf. Neutron Physics, Kiev (1975) Vol. 5, p. 337. 68) J. W. Boldeman and A. W. Dalton, AAEC/E172 (1967). 69) D. S. Mather, P. Fieldhouse and A. Moat, Phys. Rev. 123 (1964) 1403. 70) D. S. Mather, P. Fieldhouse and A. Moat, Nucl. Phys. 66 (1965) 149. 71) H. Conde, Ark. f. Fysik 29 (1965) 293. 193 MEASUREMENT OF THE "^Cf SPONTANEOUS FISSION NEUTRON SPECTRUM M. V. Blinov, V. A. Vitenko, V. T. Touse V. G. Khlopin Radium Institute Leningrad, USSR The neutron energy spectrum of 252 Cf spontaneous fission has been measured by the time- of-flight method using two neutron detectors -- a 6 LiI(Eu) crystal and a fast ionization chamber with 235 U layers. The influence of scattered neutrons and fission neutron emission time on the spectrum shape was studied. In the range from 10 keV to 7 MeV, the neutron energy spectrum is satisfactorily described by a Maxwellian distribution within the experimental errors with T = 1.41 ± 0.03 MeV. p en (Fission neutron spectrum; neutron detectors; neutron standard; time-of-fl ight; Cf) Introduction It is known that the neu spontaneous fission of 25 Cf standard. Such important rol trum demands high precision o over the whole energy region 10 - 15 MeV. The experimenta interval above 1 MeV are in a measurement accuracy is desir the neutron energies below 1 cant difference between the d between the results attain 30 makes much more precise measu tron energy spectrum of has been recommended as a e of this neutron spec- f its shape measurement from several keV up to 1 data for the energy greement,l"5 however the ed to be improved. As to MeV, there is a signifi- ata; the discrepancies 40% and more. 4 " 10 That rements necessary. The difficulties of measurements in this low energy region are first of all due to the absence of suitable detectors, which could be used in a wide energy region, including the keV range. Plastic and liquid scintillators permit neutron spectra measure- ments only in a limited part of this region. In the past few years, lithium scintillation glasses have come into use for neutron detection at energies below 1-2 MeV. But they present some serious difficulties in the experiment. Hence we have developed a tech- nique of "Lil(Eu) crystal use for time-of-fl ight neutron spectrum measurements. This crystal was used earlier as a neutron detector in experiments which did not demand high time resolution. The ionization chamber with 23 U layers is a neutron detector with smooth spectral efficiency and practically unlimited energy sensitivity region. In order to use it in this experiment, it was necessary to solve the problem of ensuring the necessary efficiency and high time reso- lution simultaneously. Such a chamber has not yet been used for fission neutron spectrum measurements. 252 In this report we present the results of Cf fission neutron spectrum measurements, done by time- of-fl ight method using a ^Lil(Eu) crystal, and also the preliminary data on spectrum shape, which was measured with a multiplate ' 23 ^U ionization chamber. Thus, we have used two detectors, which detect neutrons by means of °Li(n,a) and ^^[i{n,f) reactions. Both reactions are standards. Experimental Method and Instrumentation The time-of-fl ight method was chosen, since at present it gives the highest accuracy in neutron energy measurements. 252 The fragments of Cf spontaneous fission were detected by a gas scintillation counter or by a fast ionization chamber. Californium layers were deposited on platinum backings 0.2 mm thick; 0.3 x 10^, 0.6 x 10 and 1.2 x 10 5 fissions per second occurred in the layers. A ^Lil(Eu) crystal was used for neutron detection below 2 MeV. This crystal was chosen because of its higher y-equi valent in comparison with that for lithium glass (3.5 MeV as compared to 0.8 MeV). This reduces considerably the danger of spec- trum distortion due to y-quanta detection. Besides, the lithium glasses contain large amounts of oxygen and silicon, total neutron cross sections of which exhibit large resonances. The main difficulty of the work with lithium crystals is their large decay time (about 1.2 microseconds), which even at their high light yield, leads to the emission of a small number of photons per nanosecond time interval. Thus, it was necessary to select crystals according to their light output and photomul tipl iers according to their quantum yield of photocathode. In order to obtain a high time resolution, the electronics were constructed, taking into account the specifics of light emission features of 6|_iI(Eu) crystal. A short description of the procedure for the use of this crystal in a time- of-fl ight spectrometer was given in our publication. 11 The time resolution of 1.5 ns was obtained for low- energy neutrons. It must be noted, though, that the large decay time of the crystal in case of insuffi- cient quality of electronics adjustment may lead to the erroneous spectrum -- intensity rise for low neutron energies. The correct adjustment was con- trolled by the symmetry of a gamma peak. The second detector was the fission chamber. Its low efficiency requires the use of multilayer con- struction. In our chamber ten layers of uranous- uranic oxide ( 2 -"U), 10 cm in diameter and deposited on both Sides of the electrodes were mounted (the total 235y am ount was 1.5 grams). The distance between electrodes was 3 mm. The chamber was filled with methane to atmospheric pressure. All the chamber electrodes were connected to the same fast current amplifier. The time resolution of the spectrometer with this chamber was about 2 ns. Investigation of the Effects Caused by Scattered Neutrons Scattered neutrons usually cause considerable difficulties at low neutron energies, when continuous neutron spectra are being measured. We paid special attention to this problem and made several control experimental setups. At first we tried fission frag- ment detectors and neutron detectors of the types currently used in similar experiments. We found the scattering effects were rather large, which could give a neutron intensity increase at the low energy part of the spectra about 30-50% or more. Then we measured neutron spectra, using detector setups of different weight. The constructions, which caused no noticeable spectrum deformation within experimental errors, were chosen in that way. Two spectra, which show the detector scattering mass influence on the spectrum shape, are presented in Fig. 1. The gas counter after several modifications was made of a steel tube (diameter 40 mm, wall thickness 194 3 10 a— y-peaK ♦ • > » . ./% 1 ft !i f X A 40 ■ V- . v" i ♦ V '"*'".• 1 i \ " x v " i i v^- ♦ / \ "^ ♦ + \x 10 ■ ■ j X k , ► ^ 1 ♦ > 1 111! 10 20 30 40 50 60 70 t n: pro c Figure 1. Time-of-f light spectra of Cf neutrons, measured with the Lil(Eu) crystal on 25 cm flight path in two different setups: (o) - Setup with usual neutron scattering probability (the gas counter with quartz window, the californium layer 6 cm from the window, model 30 PMT in the neutron detector). (+) - The modified setup with significantly lower neutron scattering probability (the gas counter without a quartz window, the californium layer 15 cm from the photocathode, model 71 miniature PMT in the neutron detector). 0.2 mm, length 150 mm) without the usual quartz window. It was mounted directly on the photomultiplier (PMT). The californium layer was placed at the far end of the tube. The scattering effects of PMT mass and of crystal protective capsule were observed for 6|_il(Eu) detector. Multiple scattering in the crystal more than 6 mm thick caused a noticeable change in spectrum shape. For mea- surements a crystal 4 mm thick and 15 mm in diameter was used. It was glued directly to a miniature model 71 PMT. The crystal container was made of aluminum 0.5 mm thick. The last modification of multilayer uranium chamber was made of a brass foil 0.2 mm thick. Uranium was deposited on thin nickel foil. The measurements taken to decrease (or remove) the scattering masses of neutron and fission fragment detectors permitted a decrease in scattering effects (more than 10 times in comparison with currently used experimental setups). On Fission Neutron Emission Time 252 In Cf fission neutron spectrum measurements, it is necessary to take into account the neutron emission time in case it is more than 1 x lO"* 4 second. Such intervals between fission events and the emission of some part of the neutrons are possible because of large fission fragment angular momentum. A time interval of 10~13 to 10~1* sec is large enough to allow the fragment to slow down noticeably in the backing, on which the californium layer is deposited. In such a case the energies of neutrons, emitted by a slowed-down and by a free-moving fragment, may differ considerably. Due to this fact the backing material and thickness may influence the spectrum shape. In Refs. 12 and 13 it has been shown that the upper limit of neutron emission time in 235y fission induced by thermal neutrons is about 4 x 10~ 14 to 10"13 seconds. But in the case of the 252Qf S p 6 ctrum shape measure- ment, which must be used as a standard, a much more accurate determination of emission time must be made, to permit a clearer interpretation of results. 195 We made a neutrons, which vacuum and in a made in wide ne 7 MeV. It was are emitted ear event. The neu after fission e total number of precise comparison of fission spectra of were emitted by fragments, moving in dense medium.^ The measurements were utron energy region from 20 keV to found that spontaneous fission neutrons lier than 1 x 10"^ sec after fission trons emitted about 1Q-13 sec and more vent are less than 2 to 3% from the neutrons. To determine the influence of iodine and of 7|_i , which were present in our detector, and also the influence of our crystal capsule, the same measure- ments were repeated with a 'Lil(Eu) crystal of the same size and light output. The background was found to be 3% for 1 MeV, 2% for 0.5 MeV and 0.3% for 0.25 MeV. Measurement runs with ^Lil(Eu) were done round- the-clock for one month and for 15 days with a uranium chamber. Neutron Spectra Measurements 252 Cf fission neutron spectra were measured at flight distances of 12.5, 25, 37.5 and 50 cm with a 6|_iI(Eu) crystal and at 25 and 50 cm with a uranium chamber. The results obtained at four flight distances with a lithium crystal were compared and found to be identical within the experimental errors. A small discrepancy was observed only in the region of the 240-keV peak, especially for 12.5 cm flight path, due to the influence of energy resolution. These data contradict the results of the work, 9 where it was found that the neutron spectra for various flight distances differ significantly in the low energy part of the spectra. The authors of Ref. 9 came to the conclusion that below 1 MeV a considerable yield of neutrons with emission times 10"^ to 10"^ sec exists, and that at 100 keV the yield of degraded neutrons is comparable to the yield of prompt ones. Our experi- ments^ did not lead to such a conclusion even at 10% level of the value, published by the authors of Ref. 9. Fig. 2 shows the experimental time-of-fl ight distribu- tion, obtained with ^Lil(Eu) crystal, measured in one of the runs with a 25 cm flight path. It can be seen that for a 25 cm path a complete separation of the gamma-peak from neutron distribution is observed, the gamma-peak full width being 1.2 ns at half maximum and 5 ns at 0.1% of its height. This is evidence of the absence of any "tailing" effects of the gamma-peak in the region of neutron flight times. The intensity ratio of the 240-keV peak to the valley between the peak and the rest of the neutron distribution was 2.4 to 2.5, which also characterizes the time resolution obtained. In Refs. 8 and 16 "fine structure" of the 252cf fission neutron spectra was reported for a wide energy region. We did not find it below 1 MeV in this experiment, and also in the measurements with stylbene crystal for neutron energies between 1 and 5 MeV. ui as M5 (us mo EjMeVj Discussion of Experimental Results Fig. 3 presents the energy spectrum of neutrons from 252cf fission, which was measured by us with 6 LiI(Eu) crystal . 252, Figure 2. Time-of-fl ight """ Cf fission neutron spectrum [ 6 Li I (Eu ) neutron detector, flight path 25 cm]. EjMeV) 252 Figure 3. Energy spectrum of Cf fission neutrons (o) - Data obtained with 6 LiI(Eu) detector at 12.5, 25 and 37.5 cm flight paths. The smooth curve is a Maxwell ian distribution with T = 1.38 MeV. The errors plotted are statistical , only. In calculating crystal efficiency we used the estimated values of the 6u(n,a) cross sections from the ENDF/B-IV file. No correction was made for multiple scattering in the crystal. In Fig. 3 a Maxwell distribution curve for T = 1.38 MeV is also plotted. The experimental points lie within ±5% from this curve. Taking into account the errors in the 6|_i(n,a) reaction cross sections, one should note that there is satisfactory agreement between the Maxwell distribution and the experimental data. Our preliminary data, after correction for new values of cross sections, agree with the present data within the experimental errors. Further precision spectrum measurements using the ^Li(n,a) reaction depend greatly on more accurate cross section data for this reaction. In the calculation of the spectra, obtained with the uranium chamber, we used the estimated fission cross section for 235[j f rom R e f, \i w j ne energy spectrum measured in this way is presented in Fig. 4. The results of these preliminary measurements in a wide energy interval can be approximated by a Maxwell distribution with T = 1.41 ± 0.03 MeV, which is in good agreement with the data obtained by the lithium technique. Thus, we may say that our experimental data agree in the region above 1 MeV with the data. 2 " 5 In the low energy region (below 1 MeV) we did not find any noticeable neutron excess over the Maxwell ian, which does not confirm earlier Refs. 6, 8, 9 and 18. Consi- dering the importance of knowledge of the precise 196 References 01 0.2 0.3 O.i, 0.5 0.6 0.7 *" 10 - 12 3 4 5 6 7 En(MeV) 252 Figure 4. Energy spectrum of Cf fission neutrons. Points - experimental data obtained with uranium chamber at flight paths 25 and 50 cm. Continuous line - Maxwellian distribu- tion with T = 1.41 MeV. 252 shape of the Cf spontaneous fission neutron spec- trum as an international standard, we plan to continue the measurements of this spectrum with higher accuracy and in a wider energy region. 252 Along with the measurements of Cf fission neutron spectrum at the Radium Institute, the deter- mination of the average number of neutrons per fission, v, is being performed in the spontaneous fission of 252cf. The work is being done in the laboratory, headed by K. A. Petrzhak. rmination of neutron of cal ifornium source 5 microgramm ^^ Cf poration on stainless ssion rate is measured th a small level of -fragment spectrum in ing calculated. The ganese sulphate bath ion is pumped con- k 1 m in diameter calibrate the counting -up was made. At present the value of v t ( 252 Cf) = 3.738 is obtained with estimated total error about 0.5%. The work is still in progress. Acknowledgments The authors wish to thank S. S. Kovalenko and M. A. Bak for discussion of the results, and also 0. I. Batenkov, I. T. Krisyuk, N. M. Kazarinov, A. S. Veshchikov, V. I. Yourevich, B. M. Alexandrov, G. G. Verbitskaya for their help at different stages of this work. The method of separate dete emission rate and fission events is used. The samples of about have been prepared by vacuum eva steel backings. The absolute fi by a surface barrier detector wi low energy "tail" of the fission low geometry, the solid angle be neutron yield is measured by man techniques. The manganese solut tinuously from the spherical tan through the counting system. To system, a 4ire-Y-coincidence set 10. 11. 12. 13. 14. 15. 16. 17. 18. A. B. Smith, Prompt Fission Neutron Spectra (Proc. Conf. Meeting Vienna 1972) IAEA, Vienna (1972), 3. G. V. Kotelnikova, B. D. Kuz'minov, G. N. Lovchikova, 0. A. Sal'nikov et al . , Neitronnaya fizika (Proc. 3rd National Conference on Neutron Physics, Kiev, USSR, 1975) Moskwa (1976), 5, 109. L. Green, I. A. Mitschel, N. N. Steen, Nucl . Sci. Eng. 50 (1973), 257. H. H. Knitter, A. Paulsen, H. Liskien, M. M. Islam, Atomkernenergie 22_(2) (1973), 87. H. Werle, H. Bluhm, I. Nucl. Energy 26 (1972), 165. J. W. Meadows, Phys. Rev. 157 (1967), 1076. L. Eki, D. Kluge, A. Laitaj, P. P. Dyachenko, B. D. Kuz'minov, Atomnaya Energia 33_ (1972), 784. Yu. S. Zamyatnin, N. I. Krochkin, A. K. Mel'nikov, V. N. Nefedov, Nucl. Data for Reactors (Proc. Int. Conf. Helsinki 1970) IAEA, Vienna 2_ (1970), 183. V. N. Nefedov, B. I. Starostov, Neitronnaya fizika (Proc. 2nd Nat. Conf. on Neutron Physics, Kiev, USSR, 1973) Obninsk 4 (1974), 163. 0. I. Batenkov, M. V. Blinov, V. A. Vitenko, 1. T. Krisjuk, V. T. Touse, Neitronnaya fizika (Proc. 3rd Nat. Conf. on Neutron Physics, Kiev, USSR, 1975) Moskwa 5 (1976), 114. M. V. Blinov, V. A. Vitenko, Nietronnaya fizika (Proc. 3rd Nat. Conf. on Neutron Physics, Kiev, USSR, 1975) Moskwa 6 (1976), 252. Yu. S. Zamyatnin, Phyzika delenia tyazhelykh jader (Fission Physics of Heavy Nuclei) Moskwa (1957), 75. I. S. Fraser, Phys. Rev. (1952), 536. M. V. Blinov, V. A. Vitenko, I. T. Krisjuk, Neitronnaya fizika (Proc. 3rd Nat. Conf. on Neutron Physics, Kiev, USSR, 1975) Moskwa 5_ (1976), 131. M. V. Blinov, V. A. Vitenko, I. T. Krisjuk, DAN 224 (1975), 802. V. N. Nefedov, Preprint, NIAR (Atomic Reactor Research Institute, Melekess) (1969), 52. G. V. Antsipov, A. R. Benderskii, V. A. Kon'shin et al., Jadernye konstanty (Nuclear constants) 20 (1975), 3. A. M. Andreichuk, V. A. Korostylev, B. G. Basova, V. N. Nefedov et al . , Neitronnaya fizika (Proc. 3rd Nat. Conf. on Neutron Physics, Kiev, USSR, 1975) Moskwa 5 (1976), 120. 197 PROMPT FISSION NEUTRON SPECTRA *,t Leona Stewart University of California Los Alamos Scientific Laboratory Los Alamos, NM 87545 and Charles M. Eisenhauer National Bureau of Standards Washington, D.C. 20234 Recent measurements of the spectra of prompt neutrons emitted from neutron-induced fission in U and Pu are reviewed. Results are discussed in terms of departures of the data from a simple Watt representation. An evaluation of the neutron spectrum from neutron- induced fission in 235 U and spontaneous fission in 252 Cf, made by NBS in 1975, is also reviewed. Recent measurements on the fission spectra of 235 U and 239 Pu seem to indicate a harder energy spectrum than indicated by the earlier data available for the NBS evalu- ation. Possible reasons for this trend in the experimental data are discussed. (Calif ornium-252; Maxwellian spectrum; plutonium-239; prompt fission spectrum; uranium-235; Watt spectrum) I. Introduction The prompt neutrons emitted in the fission process have been studied for more than three decades. It is interesting to note, however, that the prompt neutron spectra for many important fissile and fertile isotopes have not been measured and the data available on other isotopes show significant discrepancies. Recent experi- mental measurements of the spectra of emitted neutrons have reduced some of these discrepancies. However, since experiments have been limited to incident energies of less than 2 MeV, the dependence of the spectra upon incident neutron energy for neutron-induced„f ission is still virtually unknown, even for U and Pu. 239 harder than for Pu. However, the difference in average energy between 23 ^U and 2 ^ 2 Cf is only ~ 5%, and all of the spectrum determinations available today are "similar" in shape. Figure 1 shows a typical time-of- flight spectrum by Johansson for 23 ^U neutron- induced fission. Also shown are best fits of his data to the following forms : Maxwellian Distribution: cf> E (E') ~ JZ' exp(-E'/T); E' = f T Watt Distribution: The differential fission spectrum (E')dE' , where o(E') is the cross section for a reaction such as U(n,f), often used for reactor dosimetry studies. Apparent discrepancies between differential fission spectral measurements and integral cross sections and/or reaction-rate ratios have existed for several years. These discrepancies were discussed at a Consultants' Meeting in August 1971 and recommenda- tions for measurements were made. In order to better define the differential spectrum F (E') for neutron- induced fission, a Specialists Meeting was called at Harwell in April 1975 to present and discuss recent experimental data from Cadarache (France) , Geel (Belgium) , Harwell (United Kingdom) and Studsvik (Sweden) on U and 23 'Pu. These results are described in the next section. 252 The spectrum of Cf fission neutrons is very ? ^ S similar to that for neutron- induced fission of iJJ U and 23 ^Pu. Furthermore, small near-point sources of " Cf can be used as absolute flux standards. Therefore, integral detectors with broad energy responses, such as the the flux in a Pu(n,f) fission reaction, can be used to relate 235 U fission environment to an absolute flux determination using a 2 -> 2 Cf fission source. Historically, both integral and differential measurements predicted a "harder" spectrum, that is 5 39 higher average neutron energy, for Pu than for For 252 2 ^U. Cf, the spontaneous fission spectrum is even VE') E' exp(-AE') sinh(i/BE'); -2 + Ji- 2A 4A 2 In general, good fits to either of these analytic functions are obtained from all of the experimental data available, except for small systematic deviations which will be described in more detail in Sections II and III. For more than a decade, however, the Maxwel- lian with T = 1.29 to 1.30 MeV was most often the recommended representation for low-energy neutron- induced fission on 235 U. A complete review of the measurement and theoreti- cal contributions on fission spectra cannot be given here — this subject, alone, could command an entire symposium. Instead, only a few of the more recent contributions to the field are discussed. II. Neutron-Induced Fission: 235,, , 239 D U and Pu: Since the Consultants' Meeting on "Prompt Fission Spectra" in 1971, much effort has been expended in studies on the differential and integral spectra for several important nuclei. t. Work performed partially under the auspices of the United States Energy Research and Development Adminis- tration. This paper was presented in place of a paper by P. I. Johansson of Sweden, who was unable to attend the Symposium. 198 10 8 6 4- 2- 10- 8 6 4- 2- LU -2 3. I0 Z -e- 8 6 10 r3 10" 1 1 1 1 1 1 — 235 u r\ o JOHANSSON \ N o v o \ o — \ _ o \ — o - \ o - \ o \ o — \ — o. — V \ MAXWELLIAN (E = 2.00) — — WATT -~"\ - (E = 2.02) - \\ o\ \\ 1 1 | | \ 1 1 4- 8 E', MeV 10 12 Figure 1. Comparison of Measured and Calculated Spectra of Emitted Neutrons from Neutron-Induced Fission in 235, The experiments described at the Specialists Meeting at Harwell are listed in Table I along with other pertinent information. All of the experiments were performed using time-of-f light , and Monte Carlo corrections were applied for scattering and secondary fission in the scattering samples. These corrections, which were not applied in most of the earlier experiments, tend to harden the spectra; that is give slightly higher average energy. In addition, all but one of the newer measurements record the neutron spectrum out to energies greater than 13 MeV. As a result, the spectrum can now be predicted with more confidence for neutrons emitted with energies up to 10 MeV. 199 235 239 TABLE I. Experimental Data for Neutron- Induced Fission of U and Pu Target Incident Neutron Energy (keV) No. Pts. Range of Emitted Neutrons (MeV) Authors 235 T 10 - 58 + 33 400 - 39 < 520 + 20 100 180 530 + 32 2070 96 91 90 62 0.5 - 13.946 0.575 - 6.87 0.625 - 15.629 0.225 - 14.4 D. Abramson Cadarache C. Lavelaine M. M. Islam (Dacca), Geel H. H. Knitter J. M. Adams Harwell P. I. Johansson (Studsvik) P. I. Johansson Studsvik B. Holmqvist T. Wiedling L. Jeki (Budapest) 239 Pu 10 - 58 215 + 32 100 180 530 + 32 2070 95 183 60 0.55 - 14.253 0.28 - 13.87 0.325 - 14.4 D. Abramson C. Lavelaine H. -H. Knitter P. I. Johansson B. Holmqvist T. Wiedling L. Jeki (Budapest) Cadarache Geel Studsvik Many attempts have been made to obtain analytic representations of the fission spectra in order to satisfy the requests of the reactor designer. The two most often recommended are mentioned in the Introduc- tion, namely the Maxwellian and the Watt distributions. Furthermore, since it is impossible to compare the experiments point-by-point, ratios of the experimental data to analytic expressions permit one to compare trends in the data. 235 The U data shown in Figure 2 are plotted as ratios of the experimental spectra to a Watt distri- bution with A = 1.0193; B = 2.3075; and E' = 2.027 MeV. 739 The J7 Pu experimental data are presented in the same manner in Figure 3 with A = 1.0364; B = 2.8728; and E' = 2.116 for the Watt distribution. The excess of neutrons below one MeV in the Cadarache results are said to be due to instrumental effects. Figures 2 and 3 were obtained by averaging over groups of emitted neutron energies for each set of data listed in Table I. Data for all incident energies measured by Johansson et al were combined in one set. It should be noted that the experiments outlined in Table I also included a study of the angular depen- dence of the fission spectra. For example, the Geel spectra of ^-"u were observed at 45°, 90°, and 130°. Cadarache made measurements on both -'- > U and ^"Pu with acceptance angles of + 30° and + 60°. All other spectra were observed at 90°. These data indicated that the shape of the observed spectra were not dependent upon the angle of the emitted neutrons. Since the Maxwellian shape has been the recommended distribution for fission-reactor analyses for many years, the Johansson data on ^3->jj were chosen to compare the Watt and Maxwellian parametrizations. The ratios of the 2 + experimental points to these analytic representations' are plotted in Figure 4. While the Cadarache and Geel results do not show as clearly the preference for the Watt distribution over the Maxwellian as do the Studsvik (Figure 4) and Harwell data, all of the measurements individually and collectively indicate slightly better agreement with a Watt distribution. Therefore, the Proceedings of the Specialist Meeting include recom- mended Watt distributions for 235 U and 239 252, Pu. Consider- ing these results and recent data on Cf, perhaps the Watt distribution should be the recommended parametri- zation for all important isotopes on ENDF/B. III. Comparison with NBS Evaluations In 1975 Grundl and Eisenhauer"* described their evaluation of differential neutron fission spectra for 235 9S9 U and J Cf , based on all available experimental data for these two isotopes. Tabulated data were obtained and least-squares fits were made to a Maxwellian form by consistent procedures for each data set. The goodness- of-fit was generally of the quality of the U spectrum shown in Figure 1. However, the average energy deduced for each fit varied by as much as 10% - considerably greater than the quoted uncertainties of less than 5%. The main purpose of the analytic representation was to obtain a consistent procedure for normalizing the data. 935 757 Reference Maxwellians for "-"U and ^ J ^Cf were determined from a weighted average of individual data sets for each isotope. Seven energy intervals were chosen and depar- tures of the data from the reference Maxwellian shape were averaged. Uncertainties in the small but systematic + It is interesting to note that the "best fit" to the Johansson data using the Watt representation gives an average energy of 2.020 MeV while the Maxwellian fit gives 2.001 MeV. These average energies were obtained by fitting over limited energy regions while Johansson obtained the same average energy (2.027 MeV) for both shapes when fits were made over the entire energy range. 200 I.I 1.0 0.9 o o CADARACHE cifrxfccRfcqO ° ^oocP oo o ° o oooo° ° o I.I 1.0 GEEL x x V* x, x x X x x x x x x yxY Xxx x x UJ DC X W uj 0.9 235 u Ql X UJ UJ I.I 1.0 0.9 HARWELL %. -^ o°°o o0 O^OqOq OC^xPX) o o o I.I .0 A STUDSVIK xx v wV x x x x Xxx' x x x x x x x x xxx x x x x X 0.9 4 , 6 E, MeV 8 10 Figure 2. Ratio of Experimental Data for Neutron-Induced Fission in 235 U to a Reference Watt Spectrum 2 with A = 1.0193, B = 2.3075 and E* = 2.027 MeV. 201 1.0 0.9 CADARACHE "^V, o\o o° rt oo « ° 00 o ° o o o o u: UJ K ^*+ LU I.I -nmir* -e- \ H 1.0 OL X UJ *^N> uj 0.9 »_^ -9- — iv — GEEL x x *x^ XxX VxX y xx 239 Pu I.I .0 0.9 ^VfeOs STUDSVIK ^Q^O -o ° o o o 1 4 6 E', MeV 8 10 239 2 Figure 3. Ratio of Experimental Data for Neutron-Induced Fission in Pu to a Reference Watt Spectrum with A = 1.0364, 2.8728, and E 1 = 2.116 MeV. departures, based on the spread of the data, were calcu- lated and found to be statistically significant. The NBS reference Maxwellian 1 * , together with departures from this Maxwellian and assigned uncertainties, constituted the evaluation. Specification of the departures in seven energy intervals had the disadvantage that the evaluated spectra were discontinuous. Therefore, subsequent to the published evaluation, the same average departures of the data from the Maxwellian were reanalyzed in finer energy groups. This procedure resulted in a continuous analytic representation of the fission spectrum. This NBS representation, which is called the "segment-adjusted Maxwellian," was described in a recent paper at the Vienna Consultants Meeting 5 . The departures of the NBS evaluated spectra from a Maxwellian_with E = 1.97 MeV for 235y an( j f rom a Maxwel- lian with E = 2.13 MeV for 252 Cf are shown in Figures 5 and 6. Some of the departures of the measurement of Green 6 for the 252q£ spectrum shown in Figure 6 give an indication of variations from the mean values. The average energies of the evaluated spectra are 1.98 MeV 235tt -«j o 1 1 Mo» f^,v- 252,- for 3 U and 2.12 MeV for -Cf. Both figures show a deficiency of neutrons, compared with the Maxwellian shape, above 6 MeV and below 0.3 MeV. This behavior, which is consistent with a Watt distribution, suggests that the reference spectrum is closer to a Watt rather than to a Maxwellian. However an integral measurement reported by McElroy et al in a '-"U fission spectrum indicates that the 202 LU ■5- 1. 10 1.05 0.0 Q. X UJ 0.95^ LU - 0.90 0.85 0.80 oo 235 o o u *xx** ° x x x x °o X \Q^ XX &**>' x x o o ftp * x *x x x 6 * o o STUDSVIK o MAX x WATT 4 6 E', MeV 8 10 Figure 4. Comparison of Ratios of Measured Spectrum of Neutrons Emitted from Neutron-Induced Fission of Analytic Fits of a Maxwellian and a Watt Shape. For the Maxwellian; E' = 2.001 MeV. For the A = 1.0172, B = 2.2553 amd E' = 2.020 MeV. Ti \ to Watt; UJ z> 1- < I -^r g. <\ a / \ — — — - ~ / \ ^- — — < z LU u / " DC UJ -10 / 235 u Y -20 1 1 1 1 i iin 10 E.MeV Figure 5. Percent Departures of the NBS Evaluated Spectrum of Neutrons Emitted from Neutron-Induced Fisssion in U. Horizontal lines are average departures of the data in 28 energy groups; linear segments are piecewise continuous approximations to these departures. 203 E'MeV Figure 6. Percent departures of the NBS Evaluated Spectrum for Neutrons Emitted from Spontaneous Fission of Cf. Solid horizontal lines are average departures of the data in 28 energy groups; dashed horizontal lines are departure of Green's data; linear segments are piecewise continuous approximations to the average departures. UJ r* i.i UJ -e- .0 < > UJ 0.9 - -e- 1 1 1 1 1 235 u ~ ^ "— L_ 1— ~~i — r _L JL _L _L _L 6 E' , MeV 8 10 n o c n Figure 7. Ratio of NBS Evaluated Spectrum for Neutron-Induced Fission in U to a reference Watt spectrum with A = 1.0193, B = 2.3075 and E' = 2.027 MeV. spectrum at about 10 MeV may be closer to a Maxwellian than a Watt. Finally, both figures indicate an excess of neutrons around 0.5 MeV, which is inconsistent with either of the representations discussed here. 235 For U, the apparent excess of neutrons has been eliminated in more recent data by scattering corrections in the 2 35jj sample. For example, when Knitter's correction for this effect was applied to Johansson's 204 data, the average energy of the experimental spectrum increased from 1.99 MeV to 2.02 MeV. Since Johansson's data represented only one of six experiments included in the NBS evaluation, the impact of this change on the NBS evaluation is small, increasing the average energy of the evaluated spectrum of U fission neutrons from 1.98 MeV to 1.99 MeV. It is suggested that the average energy would increase still further if the older data sets of Cranberg and Rosen 8 , Barnard et al and Werle and Bluhm 10 were corrected for scattering effects in the U sample. However, Johansson's sample was consider- ably thicker than many of those used in earlier work. Therefore the effect of scattering in the sample would probably be much less than the effect calculated for Johansson's data. In order to permit„a direct comparison, the NBS- evaluated spectrum for U is plotted in Figure 7 as ratios to the same reference Watt spectrum assumed in Figure 2. In general, the magnitude of the ratios shown in Figure 7 is comparable to those shown in Figure 2. However, the ratios are systematically greater than unity between 0.2 and 1 MeV and less than unity between 1 and 10 MeV. This follows from the fact that the average energy of the evaluated spectrum (E = 1.98 MeV) is lower than that of the reference spectrum (E = 2.03 MeV) obtained from the most recent measurements. For 252, Cf, the excess of neutrons in the region around 0.5 MeV is still not understood. Green 6 has formulated a model based on experimentally determined parameters which predicts an excess of neutrons over the Maxwellian below about 0.75 Mev. On the other hand, in the previous paper presented at this Symposium, Blinov noted that an apparent component of low-energy neutrons disappeared when the thickness of the Li-glass detector was reduced. His statistics are not good enough, however, to determine the excess over a Maxwel- lian to better than 5%. IV. Summary and Conclusions Recent experiments have served to establish with greater confidence the fission neutron spectrum for low- energy neutron- induced fission of U and Pu and for the spontaneous fission of Cf. Average energies of emitted spectra inferred from experiments on neutron induced fission have tended to increase with time. Since the scattering corrections 3 applied to the recent measurements slightly harden all of the spectra, better agreement with the older data might be obtained if such corrections had been applied. While these data, when considered separately, show a slight increase in the average energy of the emitted neutron, the highest incident neutron energy investigated was only 2.07 MeV. Therefore, the trend is indistinguishable from a fission spectrum independent of incident neutron energy and the dependence of the shape of the prompt fission spectrum upon the incident neutron energy remains an enigma. REFERENCES "Prompt Fission Neutron Spectra", Proceedings of a Consultants' Meeting on Prompt Neutron Spectra; organized by the International Atomic Energy Agency and held in Vienna, 25-27 August 1971, IAEA, Vienna, 1972. B. H. Armitage and M. G. Sowerby, ed , "Inelastic Scattering and Fission Neutron Spectra," AERE-R 8636, Proceedings of a Specialist Meeting; sponsored by the Joint Euratom Nuclear Data and Reactor Physics Committee and held at AERE, Harwell, April 14-16, 1975, Harwell, U.K., January 1977. M. M. Islam and H. -H. Knitter, "The Energy Spectrum of Prompt Neutrons from the Fission of Uranium-235 by 0.40 MeV Neutrons", Nuc . Sci. Eng. 50, 108 (1973). J. A. Grundl and C. M. Eisenhauer, Proc. Conf. Neutron Cross Sections and Technology, Washington, D.C. (1975). NBS Special Publication 425, p. 250. Edited by R. A. Schrack and C. D. Bowman. J. Grundl and C. Eisenhauer, "Benchmark Neutron Fields for Reactor Dosimetry," IAEA Consultant's Meeting on Integral Cross Section Measurements in Standard Neutron Fields for Reactor Neutron Dosimetry; International Atomic Energy Agency, Vienna, Nov. 15-19, 1976 (to be published). L. Green, J. A. Mitchell, and N. M. Steen, "The Calif ornium-252 Fission Neutron Spectrum from 0.5 to 13 MeV," Nuc. Sci. Eng. 50, 257 (1973). W. N. McElroy, R. Gold, E. P. Lippincott, A. Fabry, and J. H. Roberts, "Spectral Characterization by Combining Neutron Spectroscopy, Analytical Calcu- lations and Integral Measurements", Consultants Meeting on Integral Cross-Section Measurements, IAEA, Vienna, Nov. 15-19, 1976. L. Cranberg, G. Frye , N. Nereson, and L. Rosen, "Fission Neutron Spectrum of U" , Phys. Rev. 103 , 662 (1956). E. Barnard, A. T. G. Ferguson, W. R. McMurray, and I. J. van Heerden, "Time-of -Flight Measurements of Neutron Spectra from the Fission of U, Pu", Nuc. Phys. 71, 228 (1965). U, and 10. H. Werle and H. Bluhm, ."Fission-Neutron Measurements of U, Pu, and Cf", Jour. Nuc. Energy 26 , 165 (1972). The fission spectral shape is considered to be well determined over the range of emitted neutron energies from about 1-8 MeV. Below one MeV, some experimentalists see a large excess of neutrons over the Watt or the Maxwellian shape; while above 8 MeV, the statistical errors are often large due to the extremely low flux in that region. In many reactor physics applications it could be important to know about an excess of low- energy neutrons. 205 ON QUANTITATIVE SAMPLE PREPARATION OF SOME HEAVY ELEMENTS* A. H. Jaffey Chemistry Division, Argonne National Laboratory, Argonne, Illinois 60439 A discussion is given of some techniques that have been useful in quantitatively preparing and analyzing samples used in the half-life determinations of some plutonium and uranium isotopes. Application of these methods to the preparation of uranium and plutonium samples used in neutron experiments is discussed. (Absolute counting, aliquotting, a-activity, analysis (plutonium, uranium), isotope dilution, sample preparation, specific activity) Introduction In an experiment which measures such properties of heavy element isotopes as fission cross-sections and the like, it is usually necessary to know the sample's content of the isotope of interest with some accuracy. For some experiments, for example those involving fission fragment counting, it is also de- sirable to have thin uniform deposits of the isotope. Unfortunately, the two desiderata are often partially incompatible. I shall not review the various methods that may be applied to this problem. A cursory examination of the literature indicates that there have been many papers on this and related subjects, and there have even been a number of conferences. I would like to describe some techniques that my colleagues and I have used extensively in a series of half-life measurements of some uranium and plutonium isotopes. These methods may be directly used for preparing certain kinds of samples for fission and neutron measurements, and, using the same principles, the methods may also be modified for preparing a wider range of sample types. Measurement of Heavy Element Half-Lives by Specific Activity After a period of relative dormancy, there has been a flurry of activity in this field in the last five or six years. Our group has measured and pub- lished results on 238 U, 235 U, 236 U, and 233 U, 1 " 3 and on 238 Pu and 239 Pu. l+ ' 5 The techniques used have gone through several stages of development, but they have some common features throughout. In these experiments, we used absolute alpha counting to measure the alpha particle emission rate from known masses of the isotope considered. We have examined many published experi- mental reports on the results of absolute counting as applied to half-life measurements. When two experi- menters, each making absolute measurements to 0.2% error or less, get results differing by over 1%, it is clear that some "absolute" measurement is less than absolute. In examining our own work, we checked all the possible sources of error that we could think of; there nevertheless remained the possibility that some important point was overlooked. However, in our last measurement, 5 which involved 239 Pu, the half-life was evaluated by two methods: one, using our usual abso- lute counting technique; the other, using isotopic dilution (mass spectrometric) to determine the mass of 235 U grown into a known mass of 239 Pu in a given time. That the two results agreed within a statistical error <0.1% gave us confidence that our absolute counting method was rightly named. This method stands on three legs: (1) measure- ment of the mass of the element used (e.g., uranium or plutonium), (2) accurate dilution and aliquotting for sample preparation, and (3) counting the samples in an alpha counter of precisely known geometry factor. Mass Measurement of Plutonium and Uranium Samples Since it is desired to transform weights to moles or inversely, a mass spectrometric analysis is neces- sary to provide the correct atomic weight of the ele- ment. In most cases, one isotope dominates, so only a modest accuracy is needed in the analysis. For accurate mass measurement of good accuracy (in the tenth percent range) two primary methods are use- ful : (a) preparation of a compound of known compo- sition and purity or (b) oxidation-reduction titration. A secondary method is that of mass spectrometric iso- tope dilution; secondary, in the sense that the spike itself must ultimately be standardized by a primary method. The compound preparation technique has been widely used in the half-life field. Among the compounds used for uranium have been the pure metal and the oxide U 3 8 . A wider variety has been used for plutonium, including the pure metal, the dioxide Pu0 2 , the di- sulfate Pu(S0i t ) 2 , cesium plutonium chloride Cs 2 PuCl 6 , and the halides PuF 3 , PuCl 3 and PuBr 3 . For accurate determination of the element's weight from the weight of the compound, it is necessary to know that the com- pound is really stoichiometric, i.e., that the materi- al as formed actually contains the ratio of atoms de- scribed by the simple chemical formula (e.g., that, to a high decree of accuracy, prepared PuCl 3 indeed has three times as many chlorine atoms as plutonium atoms)- The metal is simplest in that no stoichiometry is in- volved, but it is harder to make without impurities, which tend to be mostly oxides or materials coming from the reduction process or from the crucible material . For uranium it has been found that heating the oxide carefully at 900°C in air for at least one hour yields a composition very close to U 3 8 . Hence, the NBS finds it feasible to provide natural uranium oxide with a warranteed composition as a primary standard. There has been less satisfaction with the stoichio- metry of the various plutonium compounds, hence the wide variety of compounds used. With suitable experience and use of a large mass of metal, the impurities in metal formation can be ade- quately segregated. Plutonium is harder to make, one of the reasons being the handling problem caused by its high radioactive toxicity. For the same reason, it is harder to analyze for small amounts of impuri- ties, the methods used generally requiring volatili- zation (e.g., optical spectra). Nevertheless, the problems have been sufficiently well solved so that the Bureau also offers uranium and plutonium metal samples of warranteed purity. No plutonium compounds are so offered. The U content in the metal is known better than the Pu content, the NBS setting a 1 o error of 0.009% for U and 0.025% for Pu. 206 Since both plutonium and uranium form well-defined and stable (+6) and (+4) states in water solution, it is feasible to measure the amount of oxidant or reduct- ant required to change the element completely from one form to the other. For example, the method we used for analyzing plutonium 6 involved oxidizing it completely to Pu(VI) and then reducing it quantitatively with standardized ferrous ion solution according to the reaction: Pu(VI) + 2Fe(II) ■* Pu(IV) + 2Fe(III) The concentration of Fe(II) is determined by oxidizing it with pure standard potassium dichromate, whose com- position is warranteed by NBS. In the titration, the point of exact equivalence is observed as a sudden change in the current passing through the solution (hence, an amperometric endpoint). Uranium can be completely reduced to U( IV) and then quantitatively oxidized with standard dichromate, with an endpoint detected through a sudden potential change. 7 The NBS standard uranium or plutonium can be used as the pri- mary standard instead of potassium dichromate; such a procedure serves to protect against possible uncertain- ties in the stoichiometry of the endpoint. Experience, however, shows that this uncertainty is very small for uranium (<0.02%) and quite small for plutonium (^0.05%). The uranium and plutonium provided by NBS are excellent chemical standards, but the isotopic com- position is usually not that desired by the experimen- ter. The titration method, can then serve as a means of transferring knowledge about the exact mass of an NBS sample to the experimenter's material. The titration method for uranium can be made more precise than that for plutonium. For uranium titra- tions, the standard deviation per sample can be <0.02%, whereas that for plutonium titrations is ^0.05%. For most purposes, such precision suffices. Such accu- racies are attainable only through the use of relative- ly large samples. The precision values described were attained when 60 mg U and 20 mg Pu samples were used, and smaller samples yield poorer precision. It may also be noted that other precision titration methods are available for both uranium and plutonium, but we do not have experience with their use. Aliquotting, Dilution and Sample Spreading After the solution concentration has been evaluat- ed, sample preparation requires that a known fraction of the solution be appropriately transferred to a back- ing plate. Frequently the amount of isotope needed for the experiment is so small that an intermediate dilu- tion is required. This was, indeed necessary for the half-life measurements of such highly radioactive nuclides as 239 Pu or 233 U. That an intermediate dilution might be necessary is due to the fact that accuracy requirements set a lower limit on the sample size which may be trans- ferred. In our experiments, we use a Mettler semi- micro single-pan balance which could weigh to 100 g and had a random weighing error of O.02 mg. Since all weight measurements were made by difference, the expected error was 0.02 /2 s 0.03 mg. We required our minimum sample size to be ^200 mg, which should yield ^0.015% net weighing error. If, for example, there were 400 g of solution, then 200 mg would represent 1/2000 of the total amount. If the sample required was larger than this limit, then the sample could be prepared by direct transfer from the stock solution. However, for samples which were a smaller fraction of the stock solution, an intermediate dilution would be needed. In the 239 Pu measurement, for example, requirements for weighing out titration samples resulted in a concentration of ^40 mg Pu/g solution. Since counting samples were to contain 3-4 yg Pu, intermediate dilution was necessary, and by a factor of ^2000, i.e., diluting the trans- ferred volume (aliquot) containing ^8 mg Pu with 400 g of acid solution. Where dilution is required, we have found it feasible to dilute only once, by choosing the dilution volume as large as necessary. The technique used for accurately transferring an aliquot to a dilution flask is the same as that used for sample transfer to a backing plate. It involves the use of a polyethylene pycnometer of the type des- cribed briefly by Janet Merritt in Ref. 8 and more fully in Ref. 9. The pycnometer is made from a small polyethylene vial whose top is pulled out in the form of a long, thin snout. Solution is sucked up into the vial, the vial is weighed, some drops are delivered into a dilution bottle or onto a plate (as in Fig. 1), and the vial is reweighed. The weight delivered in the aliquot is the difference in vial weights. The technique is not quite this simple; some refinements are necessary for good results, and these are des- cribed in Refs. 4 and 9. After some practice an operator can achieve quite good precision in the ali- quotting process. POLYETHYLENE SOLUTION DROP Fig. 1. Use of Po The snout on a polyethyle long capillary, the vial The vial is carefully han into the stock solution, capillary is wiped, the e delivered with pressure, jig, and with no free-fal splashing) when sample pi figure. Reweighing the v with great accuracy. DISK lyethylene Pycnometer ne vial is pulled out into a cleaned, dried and weighed, died, the capillary inserted and liquid drawn up. The nd cut off. Droplets are preferably with a mechanical 1 (with its potential of ates are prepared as in the ial gives the aliquot weight The data of Table I and II show examples, the first representing the results in an earlier phase of developing the technique. Table I shows the counts of samples taken from a dilution bottle and prepared as in Fig. 1. The final result represents the activity concentration of the original stock solution. Table II corresponds to a different stock solution. s_ is the usual estimate of the standard deviation per sample and e a v is the average counting error. We may take 6 2 = s 2 - e| v , where 6 represents the error of 207 Table I. Aliquotting Precision - 1 Dis/min Ct per g soln error No. Ct/min Min xlO' 7 E i 1 26996.5 400 523.883 ±0.159 2 31768.9 400 523.441 0.147 3 34580.1 1000 523.917 0.089 4 30425.8 900 523.441 0.100 5 28880.3 1000 523.547 0.097 6 39130.4 400 523.658 0.132 7 28887.3 1000 523.652 0.097 8 29806.8 1000 523.453 0.096 s = 192; e av = 0.115 Table II. Aliqt otting Precision - 2 Dis/min Ct per g soln error No. Ct/min Min xlO" 7 ei 1 40534.3 3800 419.978 ±0.034 2 48980.3 400 419.821 0.095 3 34436.1 900 419.913 0.075 4 41628.5 400 419.814 0.103 5 38192.6 1000 420.042 0.068 6 42040.3 400 419.830 0.105 uranium up to 500 ug/cm 2 in thickness. Plutonium samples have not been made to such thickness. s = 0.095; e av = 0.080 aliquotting, sample preparation and positioning it in the counter. In Table 1,6= 0.153 (0.029%); in Table II, 6 = 0.051 (0.012%). For sample plates prepared as in Fig. 1, the distribution of the activity over the place is made more uniform by evaporating with a spreading agent such as tetraethylene glycol. 1 * Better uniformity, how- ever, is achievable by electrodeposition. We have used the method of molecular plating, 10 a high voltage deposition from an isopropyl alcohol solution, using the plating cell in Fig. 2. In this case, the aliquot Fig. 2. Molecular Plating Cell A: Aluminum sample plate and cathode. B: Teflon wall. C: Base plate. D: Isopropyl alcohol solu- tion. E: Platinum anode. is delivered into a temporary transfer tube or direct- ly into the plating cell. The deposition gives quite a uniform distribution, the range of density variation being up to 20%, but generally <10% (for example, see Ref. 1). Such uniformity makes practical the deposi- tion of thicker samples than can be usefully prepared by the method of Fig. 1. We have prepared samples of The molecular deposition tive, part of the element rema alcohol. The amount, however, fraction of the total, and it counted for by washing the pla the isopropyl alcohol and depo counting with an a-counter. 1 position yields more consisten sidual being generally <1%; wi of <1% are achievable, but 2 o Absolute Alpha Counting method is not quantita- ining in the isopropyl is generally a small can be accurately ac- ting cell, evaporating siting the residue for We have gotten higher de- tly with uranium, the re- th plutonium, residuals r 3% are not uncommon. Various methods can be used for absolute counting, including coincidence techniques, liquid scintillation counting and 4tt proportional counting. We have used the defined geometry technique, in which the a-particle source is placed at a known distance from an aperture of known dimensions. On the assumption of isotropic emergence from the source, the fraction of particles passing through the aperture is the same as the frac- tional solid angle subtended by the aperture at the source. The major factor limiting the accuracy of this assumption lies in the scattering of a-particles either from the sample mount or from the counter walls. With suitable care in the counter's construction, a negli- gible fraction of the scattered a-particles reach the aperture. The limit on the accuracy of the method is then set by the accuracy of measuring the dimensions which define the subtended solid angle. In our half-life work we have used IGAC, 1 an in- termediate geometry alpha counter (Fig. 3). With a \ m////MM///////////^ f Fig. 3. The Intermediate Geometry a-Counter (IGAC) Operated with flowing argon (10% CH 4 ) gas filling en- tire chamber, 35 to 50 torr pressure (depending upon a-energy). Geometry defined by accurately machined and measured circular aperture K (with 0.025 mm thick edge) and by measured distance between aperture plane and sample surface H. Proportional counter chamber above thin window L (0.6 mg/cm 2 , plastic with evaporated gold layer) with wires M spanning circular area. relative solid angle G ^1/11, it is intermediate in solid angle between the 2tt counter (G ^0.5) and the low geometry counter, for which G is generally <0.01. By reducing 'G wel 1 below 0.5, we avoid the well-known back-scattering from the sample mount and severe sample self-absorption at oblique angles. Further, the large 208 aperture makes for accurate dimensional measurement and negligible slit-edge scattering, and the relative- ly large value of G allows counting of low specific activity isotopes. At a given distance from the aper- ture plane (Fig. 3), the relative solid angle G sub- tended by an elementary area dS_ on the source is a function G(r) of the distance r^ from the axis (defined by the circular aperture) to dS. 11 The average rela- tive solid angle for a particular sample is evaluated by counting the sample in a 2ir counter with a series of collimators. Each collimator contains annularly- distributed orifices at a given r-value, hence corre- sponds to the relative solid angle G(r). 1 ' 11 By counting the same sample in both counters, we have checked the IGAC geometry factor against that of a low geometry counter (G ^1/100) with a 2 cm aperture and a surface barrier detector; the two counters agreed within counting statistics (0.05%). One of the most important features of IGAC is that it has a very flat plateau. This means that the pulses at levels below the discriminator level consti- tute <0.005% of the total, even for quite thick sam- ples. Figure 4 shows a pulse distribution for a rela- tively thick 238 U sample, Fig. 5 for a thinner sample. In both cases, the number of pulses smaller than the pulse selector level is negligible. 1 1 1 1 ' 1 ' 1 i 1 i 1 2 IOOOO 4 000 +10 -10 1 1 1 1 2 Z 1 000 1 ' , 1 400 100 40 = \ / 10 4 I l 1 1 1 1 30 40 CHANNEL NUMBER Fig. 4. Pulse Height Distribution from IGAC Proportional Counter for 238 U Source Discriminator setting is A. Inset shows the plateau region BB_' on expanded scale. Subtracting background, the net counts over the whole plateau is -4.6 + 10.6 counts/1000 min, compared to 2 x 10 5 counts/1000 min for the sample. It is common plateaus to zero extrapolated coun geometry factor, ciency. It has b extrapolation to a sloping plateau generally too low bution (as in Fig diagnostic. Net sloping plateau) such samples are ascribe entry of due to poor samp! practice to extrapolate a-counter pulse height and to assume that the ting rate, when corrected for the corresponds to 100% counting effi- een our experience that when an zero pulse height is necessary due to , then the extrapolated result is In using IGAC, the pulse distri- s. 4 or 5) is routinely used as a counts in the plateau region (i.e., a usually correspond to low results, so automatically discarded. We usually small pulses into the distribution as e spreading, with some clumping. A continuum of small pulses has also been observ- ed with liquid scintillation a-counting and with a faulty surface-barrier detector used with a low 30 40 CHANNEL NUMBER Fig. 5. Pulse Distribution from a Thinner and More Active Source Discriminator setting at A. Within counting error, the background accounts for all counts between B_ and £, so the plateau is flat in this region. Even without sub- tracting background, the counts of pulses smaller than A (extrapolating to zero pulse height) is <2 x 10" 5 of the total number of counts. geometry counter. For the liquid scintillator, the pulse distribution showed a-particle pulses down to the noise level, and extrapolation to zero pulse-height yielded results 0.15-0.20% lower than that derived from IGAC. Replacement of the surface barrier detector did not sensibly change the peak half-width, but decreased the number of small pulses and restored the correct counting rate. This effect may have been due to a small defective region in the large area detector. While IGAC (or other counters) may be used to determine the disintegration rate of an a-emitting source, the experimenter generally wishes to determine the rate for a specific isotope. Auxiliary information is then required, usually supplied by a mass spectro- metric analysis and an a-pulse analysis with a solid state detector. In certain cases, however, the pulse analysis does not help, because it cannot discriminate between certain nuclide pairs which have overlapping a-energies. The most important of these are: ( 240 Pu, 239 Pu), ( 238 Pu, 2 ^Am) and ( 233 U, 234 U). The members of the pair must be distinguished by the mass spectro- metry analysis or by other radiometric methods, such as detection of the prolific 21+1 Am 60 keV y-ray with a Ge(Li) detector. With the mass spectrometric and pulse analysis data, the fraction of the measured disintegra- tion rate ascribable to the nuclide of interest may be evaluated. When some component in the sample gives rise to relatively short-lived decays, it is important that the disintegration rate measurement be made at close to the same time that the auxiliary measurements are made. The most common cases in which such problems arise are those in which samples of uranium contain 232 U (73.6 yr) or samples of plutonium contain 241 Pu (14.5 yr). The latter decays to 2ltl Am (432 yr) and the former de- cays to 228 Th (RdTh, 1.91 yr) and then in turn to an entire chain of very short-lived decays giving four more a-particles. Depending upon the fraction of 232 U (or 21+1 Pu) present, neglecting the possibility of these activity-growths can, and, indeed, has, caused serious errors. Consider the following examples: 1. A sample of 233 U contains 0.0005% 232 U; hence, the 232 U provides 1.072% of the total uranium a-activity. Just after the uranium has been chemically separated from thorium, the 233 U a-activity is 98.928% 209 of the total activity, a fact that is readily deter- mined by a-pulse analysis. Immediately after such separation, the thorium daughter starts growing and the subsequent decay chain is in almost immediate equilibrium with the 228 Th. Then, as a function of time (T yr) after chemical separation, the a-activity is expressed as A(T) = A [1 + 5(0.01072)(l-e"°- 363 T )], where A is the activity of the sample just after pu- rification. For short times, approximately, A(T) = A [1 + 0.01946 T] o The 233 U a-activity is a decreasing fraction of the total activity. Thus, in six months, th'e sample ac- tivity has increased by about 0.9%, so the 233 U frac- tion has correspondingly decreased. Clearly, for longer times or larger concentrations of 232 U, the growth of a-activity is more serious. 2. A plutonium sample contains 97.1% 239 Pu, 2.5% 21+0 Pu and 0.4% 241 Pu. After separation from Am, the sample activity is A and after T yr, it is A(T) = A [1 + 0.010 T] Analysis by Isotope Dilution Into a sample we add a known amou After the two isoto the measurement of mi nation of the amo of measurement is c added isotope is th commonly carried ou an A/B mole ratio, A/B a-activity rati containing isotope A of an element, nt of isotope B_ of the same element, pes are chemically equilibrated, the ratio of A to B yields a deter- unt of A in the sample. This type ailed isotope dilution and the e spike. The procedure is most t mass spectrometrically, yielding or by pulse analysis, yielding an o. Either me correct for in measurement, accuracy (0.1% The mass spect inherent mass a valid correc peak has a tai hence gives th sity. thod is val id on herent isotopic Such correction or better) requ rometric method discrimination; tion for the fac 1 which overlaps e lower peak an ly when care is taken to biases in the ratio may be made, but high ires a skilled operator, requires calibration for pulse analysis requires t that the high energy the low energy peak, apparent higher inten- In both cases, we tested the correction methods by analyzing known artificial mixtures. Mass spectro- metrically, samples made from mixed aliquots of titrat- ed 233 U, 235 U and 238 U were tested. For pulse analy- sis, we analyzed samples prepared from standardized solutions of pure 238 Pu and 240 Pu. The half-life measurements were made when the calibration techniques were found to be satisfactory. The first experiment was made somewhat more com- plex because a 238 U spike was used. The ubiquitous presence of natural uranium contamination in reagents, dust, glassware, etc., required that super-clean tech- niques be used in all handling and chemical manipula- tions. It would be possible to avoid such meticulous handling by using only one spike (usually 233 U) and to calibrate the mass discrimination bias with a separate source containing a known mass ratio (generally, an NBS 235 U/ 238 U source). However, one then loses an in- ternal calibration based upon spike ratios taken under the identical conditions used for the experimental ratio. For the bias calibration with external source to be valid, it must be shown that the bias calibration of the mass spectrometer remains constant to the ac- curacy required over a number of succeeding runs. Mass spectrometric isotope dilution analysis of plutonium isotopes may be similarly used. Enriched 242 Pu is most commonly used for plutonium, because it is generally in low concentration in the plutonium samples of most interest. The General Sample Problem The techniques described above have the virtue that they allow preparation of samples containing known amounts of material, when analyzed solutions of the desired isotopes are available. Samples prepared as in Fig. 1 are suitable when small masses are needed or for heavier samples, when sample self-absorption is not important. Molecular plating may be used when heavier samples are needed and uniformity is important. When heavier samples of good uniformity are need- ed, other preparation techniques must be used. These, unfortunately, are generally not quantitative. For example, the well-known electrospraying technique can be used to prepare thick, quite uniform samples, but quantitative deposition is not possible. In many cases, however, it is feasible to calibrate the samples, either by counting or by destructive analysis. Recent examples from our work: 1. In a determination of the 239 Pu half-life, 5 we measured the amount of 235 U grown in a given time. Solutions of 233 U and 238 U were standardized as des- cribed above, and precise aliquotted spikes were added to the 239 Pu solution. Uranium was extracted and the ( 233 U, 235 U, 238 U) mixture was mass-spectrometrically analyzed. The linear bias was evaluated by comparing the observed 233y/238u mo i e ra tio to the known spike ratio. 2. In an evaluati measured the amount of known a-activity of 2h2 a-activity of 2U0 Pu was was extracted, and the was determined by pulse potential error source the low energy tail of MeV) from the apparent (5.159, 5.114, 5.01 MeV was derived from pulse on of the 238 Pu half-life, 4 we 38 Pu grown from the d Cm during a given time added as a spike, the 240 Pu /238p u a .activity analysis. The most s arose from the need to the 238 Pu peaks (5.491 intensity of the 240 Pu ) . A "template" for s analysis of a pure 238 ecay of a A known plutonium ratio erious subtract , 5.448 peaks ubtraction Pu sample. If the samples are not too thick, they may be counted either in a low or intermediate geometry counter, depending upon specific activity. In either case, the quality of the plateau must be carefully examined, as mentioned above. If the isotopic half- lives and the mole- or activity-ratios are known, the isotopic masses can be calculated. If the samples are not good enough for direct counting, they may be dissolved after the experiment is completed. Analysis may then be carried out either by a-counting an aliquot as described above or by analyz- ing through one or the other form of isotope dilution. References ''Work performed under the auspices of the Division of Physical Research of the U. S. Energy Research and Development Administration. A. H. Jaffey, K. F. Flynn, L. E. Glendenin, W. C. Bentley and A. M. Essling, Phys. Rev. C, 4, 1889 (1971). 210 K. F. Flynn, A. H. Jaffey, W. C. Bentley and A. M. Essling, J. Inorg. Nucl. Chem. , 34, 1121 (1972). 3 A. H. Jaffey, K. F. Flynn, W. C. Bentley and J. 0. Karttunen, Phys. Rev. C, 9, 1991 (1974). 4 H. Diamond, W. C. Bentley, A. H. Jaffey and K. F. Flynn, Phys. Rev. C, 15, March (1977). In press. 5 A. H. Jaffey, H. Diamond, W. C. Bentley, K. F. Flynn, D. J. Rokop, A. M. Essling and J. Williams (to be published) . 6 C. A. Seils, Jr., R. J. Meyer and R. P. Larsen, Anal. Chem., 35, 1673 (1963). A. R. Eberle, M. W. Lerner, G. C. Goldbeck and C. J. Rodden, "Titvimetvie Determination of Uranium in Product Fuel and Scrap Materials after Ferrous Ion Reduction in Phosphorous Acid; Manual and Automatic Titration," Proc. Sympos. on Safeguards and Techniques (Karlsruhe, July 6-10, 1970), Vol. II, p. 27, IAEA, 1970. Also New Brunswick Laboratory Report NBL-252 (1970). o Janet S. Merritt, Nucl. Instrum. Methods, 112 , 325 (1973). 9 J. S. Merritt and J. G. V. Taylor, "Gravimetric Sampling in the Standardization of Solutions of Radionuclides," Atomic Energy of Canada, Ltd. Report AECL-2679, 1967 (unpublished). W. Parker, H. Bildstein and N. Getoff, Nucl. Instr. Methods, 26, 55 (1964). 11 A. H. Jaffey, Rev. Sci. Instr., 25, 349 (1954). 211 BLACK AND GREY NEUTRON DETECTORS F. Gabbard University of Kentucky Lexington, Kentucky 40506 Recent progress in the development and use of "black" and "grey' detectors is reviewed. Such detectors are widely used for counting neutrons in (p,n) and (a,n) experiments and in neutron cross section measurements. Accuracy of each detector is stressed. (Review; black & grey neutron detectors; neutron flux measurement.) Introduction Comprehensive early reviews of methods of neutron flux measurement and associated standard cross sections have been given by Barschall, et al. Larsson 2 and Perry 3 . Ad- vances in neutron flux measurement techniques have been discussed in Conferences on Neutron Cross Sections and Technology in 1968 and 1970. ' 5 Reviews were given at these confer- ences by Batchelor, Gibbons'* and Landon.^ The symposium on Neutron Standards and Flux Nor- malization, held at the Argonne National Laboratory in October, 1970, treated, rather extensively, the whole subject of neutron flux measurement as practiced at that time. Further work has been reported at the Conference on Nuclear Cross Sections and Technology' held at the National Bureau of Standards in 1975 and at the International Conference on the Inter- actions of Neutrons with Nuclei held at the University of Lowell in July, 1976. Papers at the Lowell Conference by H. Liskien and B. Zeitnitz deal with detectors and standards. The basic techniques for neutron flux measurement are well known and advances have been largely in the refinement of these methods toward easier use and higher accuracy. There are some five (5) basic methods in current use for the measurement of neutron flux. These are : (1) Associated particle methods; (2) Proton recoil methods; (3) Neutron moderation methods; (4) Associated radioactivity methods; and (5) Methods of reaction cross section comparison. One obvious point to be emphasized is that there is no "best" method for measuring neutron flux which works at all energies and in all geometries; the "best" technique de- pends on the neutron energy and the geometri- cal configuration of source and detector. Since the advent of time-of-f light methods of neutron spectroscopy, the time response of fast neutron detectors has become an important consideration in neutron-flux- standards applications. Properties of an IDEAL neutron detector include the following: (1) Variable detection sensitivity; (2) An efficiency independent of neutron energy, i.e., a "flat" response; (3) Insensitivity to background radia- tions; and (4) Fast time response. The advantage of variable detection sensitivity is that different experiments require the use of different intensity levels in the neutron flux. Fast time response is important in time-of-f light applications. Such an ideal detector does not exist and it appears unlikely that any single detec- tor can have the optimum characteristic in all of these properties. Nevertheless, these are important criteria for selection of a real detector. The objective here is a review of recent developments in the use of "black" and "grey" neutron detectors for the measurement of neu- tronflux. The detectors which I shall describe are of two types; proton-recoil detectors and moderated neutron detectors. The recoil de- tectors use liquid or plastic scintillators as the active medium and moderated detectors utilize hydrocarbons, LiH, or carbon as the moderating medium. Selection of detectors for description was based on personal knowledge of recent de- velopment and/or new uses. Among the "black" or "grey" neutron detectors or classes of detectors which will not be discussed in any detail are the Large Liquid Scintillators and the Activated Bath Detectors. These are cov- ered by Dr. Axton in this session and by others at other sessions of the Symposium. The time period covered in this review is roughly from 1970 until now. During this time there have been no startling breakthroughs in the neutron counting game; however, there has been steady progress on all fronts. The Macklin Sphere First, I want to recall, or call as the case may be, your attention to the Graphite Sphere Detector, ^ or Macklin Sphere designed and built by R.L. Macklin at the Oak Ridge National Laboratory. Although this detector is not new, there have been important develop- ments in its use. This detector is a sphere of reactor grade graphite with a radius of 78 cm. Neutrons originating at the center are moderated in the graphite and counted by eight (8) BF^ counters near the surface of the sphere. The detector is covered with cadmium to reduce the effect of room scattered neutrons. Figure 1 shows a picture of the detector with Cleland Johnson in the laboratory at Oak Ridge. An attractive feature of this detector is that its simple construction makes it fea- sible to calculate its relative efficiency as a function of neutron energy through applica- tion of age-diffusion theory. Figure 2 shows the calculated efficienty of the detector as a function of neutron energy. To be noted is the 212 Fig. 1. The Macklin Graphite Sphere neutron Detector with C.H. Johnson at the Oak Ridge National Laboratory. (0.003149 ± 0.000011). A local source was made by placing 492.2 mg of radium in a Be can. This source was compared to NBS II and is used as a local calibration source in the continued use of the Macklin Sphere. Johnson estimates that recent measurements of (p,n) cross sec- tions are accurate to better than 1%.10 In addition to its continued importance for measurement of neutron production cross sections, the Macklin Sphere as a prototype has good potential as a secondary standard. Source comparison with this detector can be made quick and precise since the BF-, detectors make measurements of neutron emission rate simple and in "real" time as compared to Acti- vated Bath methods. The efficiency can be raised to about 3% through use of -^BF, coun- ters in place of the natural BF^ detectors. The response curve for the detector is well understood. The response can be quantitata- tively checked with high precision in the keV neutron energy range through the use of the associated activity method with 7 LI(p,n) 7 Be, 51 V(p,n) 51 Cr and 57 Fe (p,n) 57 Co as neutron sources. I also want to mention in passing a seconddetector designed by R.L. Macklin and Q "I 5 co-workers ' at Oak Ridge, This detector can be made flat through appropriate tuning of the position of BF, counters in the moder- ator assembly. This detector has an effici- ency of ^30% and will be useful in applica- tions where very high sensitivity is needed. The Poenitz Detectors 1. The Grey Detector (10 si' 05 I i i °° :030 NEUTRON ENERGY(Mov) Fig. 2. The calculated efficiency of the Macklin Sphere as a function of neutron energy. flat response c region. This fe ticularly well s total neutron yi the center. The sively for neutr measurements and recently by John for highly accur sections for neu (a,n) reactions. urve at low energy in the keV ature makes the detector par- uited for use in measuring elds from sources placed at detector has been used exten- on production cross-section continues to be used, most son, Bair and co-workers ate measurements of cross trons produced in (p,n) and Bair, Johnson and co-workers have re- cently recalibrated the Macklin Sphere using NBS II 11 as the calibrating source. At the time of calibration, the detector efficiency for Ra-Be neutrons was measured to be An important group of black and grey neutron detectors has been developed and used by Poenitz. 13 ' 15 The first of these detec- tors which Poenitz has called the "Grey Neu- tron Detector" consists of a homogeneous hydro- geneous material such as water, paraffin, or polyethylene with an entrance channel for a collimated neutron beam. The neutrons are thermalized in the moderator and subsequently captured in the moderator material. n^e cap- ture Y~ rays are detected at the surface of the moderator. Figure 3 shows a schematic diagram of a typical experimental setup for the use of this detector. Wl YM COLLIMATOR NEUTRON J— BEAM NTRANCE CHANNEL Nol- DETECTOR eV-RANGE keV-RANGE MeV-RANGE NEUTRON ENERGY Fig. Schematic diagram of the Grey Neutron Detector and a comparison of the ef- ficiency response curve (b) with that of a manganese bath (a) . 213 Poenitz has calculated the relative ef- ficiency for the "Grey Detector" which exhib- its a flat energy response over a wide range of energies. Comparisons of the relative ef- ficency of this detector with that of a manga- nese bath are shown in the lower part of Figure 3. The relative efficiency of the "Grey Detector" was measured by two different meth- ods, 13 the associated activity method and a flux integration technique. Results showed agreement with calculated efficiency to within 2% for neutron energies below 1 MeV and showed somewhat larger deviations toward higher neu- tron energy. 14 uses the same operating principle, B-vaseline detector is a "qrey" The detector reported by Coates, et al i.e. the 'grey" detector. In this latter case, the 478 keV Y-rays from the l^B (n,ay) 7 Li were detected. The time re- sponse of this detector is about 0.7 u sec which is sufficiently fast for work with linac sources . MULTIPLIERS SCINTILLATOR m COLLIMATOR NEUTRON BEAM MULTIPLIERS ENERGY SPECTRA ENERGY SPECTRA Fig, Schematic comparison of a convention- al scintillator detector (left) and the Black Neutron Detector (right) . The energy range for which the Grey Neu- tron Detector appears to be most useful is the keV range. This detector has been used by Poenitz in several experiments for measure- ment of relative cross sections. 13 , 16 The pre- cision of these measurements ranges around (2- 3)% for keV neutrons. 2 . The Black Neutron Detector The "Black Neutron Detector" is a fast time-response detector (designed by W.P. Poenitz) for absolute neutron flux measure- ments. 5 ' 1 ' The detector consists of a hydro- geneous scintillator material in the shape of a cube, cylinder or sphere. The neutron beam enters the detector system through a channel which terminates near the center of the detec- tor volume. The light produced in the scin- tillation material is detected by several photo-multiplier tubes. A neutron entering the detector undergoes several successive col- lisions and loses most of its energy to hydro- gen or carbon nuclear recoil energy. The com- parison in the concept and energy response of a conventional scintillation detector and the Black Neutron Detector is illustrated in Fig- ure 4. A very important difference is in the light pulse spectra in the two detectors. The energy spectrum for the Black Neutron Detector allows accurate extrapolation to zero pulse height because most of the energy of all neu- trons is lost in the detector. Pulses counted above a low-set bias will account for almost all (90-95)% of the neutrons losing energy in the detector. 1 7 A monte-carlo computer code, Carlo Black, for the purpose of evaluating a Black Neutron Detector of cylindrical shape was used by Poenitz. The parameters in the evaluation were the length of the cylinder, H, the radius of the cylinder, R, and the length of the re- entrant hole, H c , the radius of the re-entrant hole, R c , and the detection threshold, E . The prototype detector is filled with a scintilla- tor having atomic densities of hydrogen and carbon of N and N , respectively. The cal- culated efficiency of this detector is shown in Figure 5. The structure in the calculated Fig. NEUTRON ENERGY (M«v) The calculated efficiency of the Black Neutron Detector versus energy The calculation was done with Carlo- Black with H = 40 cm, R 15 cm, R„ = 1.26 cm, 4.2 cm" 10^2 /' 13 cm, H c = E c = 2 00 key. N„ = N H - 7.04 x 10 22 r efficiency curve is due to scattering reson- ances in carbon and gives fluctuations of 1% or so. Figure 6 shows examples of calculated energy spectra at neutron energies of 1 MeV, 2.5 MeV, and 4 MeV. The important features are the low background counting rates at pulse heights below 20. Figure 7 shows experi- mental spectra obtained with the Black Neutron Detector. The two spectra at 1.5 MeV and 2.5 Mev show the desirable features of low back- ground and good peak definition. The energy resolution spreadinq of the detector system causes the shape of the spectra to be smoother than the calculated curves of Figure 6. The Black Neutron Detector has high sen- sitivity, flat response as a function of ener- gy and a fast time response (^4-5 ns) . This detector with a well known energy response can be used in time-of-f light experiments with pulsed or associated particle-tagged sources. 214 1 1 A 1 ___l Mev _.__ 2 5 MeV 4 MeV ///"" 4 Mev / /~T — 2 5 Mev X - /• 1 Mev 1 1 20 40 60 80 RELATIVE PULSE HEIGHT BLACK DETECTOR OPERATION "* LIGHT PHOTON O RECOIL PROTON O NEUTRON PHOTO CATHODE ANODE / NEUTRON BEAM O— PLASTIC SCINTILLATOR NE no PHOTOMULTIPLIER TUBE RCA 8854 Fig. 6. Calculated (Carlo-Black) pulse-height spectra evaluated for three different monoenergetic neutron beams. Fig. 8. Schematic of the NBS Black Detector showing the scintillator and the elec- tron multiplier. i TV'frv \ / Hufufft 1 ^ / Fig. 7. Experimental pulse-height spectra for 1.5 and 2.5 - MeV neutrons. The detector, designed for use with collimated beams, must be carefully shielded against stray neutrons and 7-radiation. Absolute measure- ments^ with the Black Neutron Detector have been made to an accuracy of ^2%. 3. The NBS Black Detector Fig. 9 shows the efficiency of this de- tector as a function of neutron energy for three different bias settings. It is seen that the response curves are very nearly flat on the energy interval between 200 keV and 1000 keV. Id 60 "< r 1 r -1 1 1 r ^— - 15 keV CUTOFF 50 keV CUTOFF 100 keV CUTOFF 400 600 NEUTRON ENERGY, k«V 1000 Fig. 9. Calculated efficiencies of the NBS Black Detector for different lower level cutoffs. Calculations were done with the computer program Carlo- Black. Lamaze, Meier and Wasson at the Na- tional Bureau of Standards have constructed a detector of a plastic scintillator after the Black Neutron Detector design of W.P. Poenitzv This detector consists of a cylinder of NE 110 plastic scintillator 12.7 cm in diameter by 17.78 cm long with a re-entrant hole having a diameter of 2.54 cm and a depth of 5.08 cm. The scintillator is optically coupled to an RCA 8854 photomultiplier tube. A schematic diagram of the detector is shown in Figure 8 . 17 The program Carlo-Black was used to provide design criteria for the NBS Black De- tector. At low neutron energies (<250 keV) , only a small number of photons are produced in the scintillator and only a fraction of these pro- duce photoelectrons. Therefore, the photo- electron statistics are important in the deter- mination of the response below 300 keV or so. This effect is illustrated in Figure 10 which shows comparison of a calculated and experimen- tal spectrum at E R = 250 keV and 560 keV. The calculated curves were obtained using Carlo- Black 17 as modified by M. Meier to calculate the spectral distributions at the two neutron energies. The modification is the explicit treatment of the Poisson statistics when N, 215 i i 560 keV I I I I I I r CALCULATED RESULTS I ooooo EXPERIMENTAL RESULTS 4,000 - 250 ! UJ 2,000 - - i ' \ /^\ 1,000 V \ l°° I \ 20 40 60 80 100 120 140 160 CHANNEL NUMBER Fig. 10. Comparison of calculated and experi- mental recoil spectra from the NBS Black Detector at incident neutron energies of 250 keV and 560 keV. Calculations were done with Carlo- Black as modified by M.M. Meier for use with small N in the Poisson sta- tistics. the number of photoelectrons, is small. The result of Meier that the photoelectron pro- duction at low energies corresponds to about 15 keV/photoelectron is consistent with Cran- berg's20 findings. The NBS Black Detector has an efficiency near 100% and has very fast time response (^5 ns) . As is the case with the Black Neu- tron Detector built by Poenitz, this detector is designed for use with a collimator or beam defining device and must be well shielded. The detector is being used in both the Linac and Van de Graaff laboratories at NBS. The Kentucky Polyethylene Sphere The nuclear physics g sity of Kentucky has develo use in the keV region. Thi briefly described in the Co Cross Sections and Technolo 1975. 7 ' 21 This detector wa marily for the accurate mea neutron production cross se tor consists of a polyethyl radius of 30 cm in which 8 radially installed. A pict tor is shown in Figure 11. designed so that it may be roup at the Univer- ped a detector for s detector has been nference on Neutron gy held at NBS in s constructed pri- surement of total ctions. The detec- ene sphere with a 1°BF3 counters are ure of this detec- The detector is opened up for in- Fig. 11. The Kentucky polyethylene sphere detector. The two hemispheres are drawn apart. In operation the counter is closed. sertion of targets into a chamber at the center of the detector. The charged-particle beam from a Van de Graaff accelerator can be passed through the target and the resulting neutrons counted. The counter is covered with 1 mm of cadmium to reduce neutron background from the room. The measured i keV to 2.5 curve is s ciency mea 1.35 MeV . and (0.56 ciency of ity method surements. to compare dual 51 Cr efficiency of this detector has been n the neutron energy range from 3 MeV. The resulting energy response hown in Figure 12. Absolute effi- surements were made near 300 keV and The results gave (0.55 ± 0.017)% ± 0.016)% respectively for the effi- the detector. The associated activ- was used in these efficiency mea- A 35 cm Ge(Li) detector was used the gamma-ray yield from the resi- and 57 Co produced in the calibrating 0.8 0.6 0.4 0.2 Absolute measurements Relative measurements Sphere Counter Efficiency for neutrons from Pu-Be Source =0. 47%' jf-*-H-J — l | i- -*-*-*-*■ ***n^ 0.5 1.0 1.5 2.0 Average Neutron Energy (MeV) 2.5 Fig. 12. The absolute efficiency of the poly- ethylene sphere counter as deter- mined by the associated activity method. 216 targets of V and J Fe, with standard sources of ^Cr and Co obtained from the National Bureau of Standards. Accuracy of the standards was ± 1.7%. The estimated accuracy of our mea- surements was ± 3%. Relative efficiency was measured using the 'Li(p,n)'Be reaction. In the energy range below an average of 1.5 MeV, the relative neutron yield as a function of neutron energy was determined using the 'Be decay as observed from observation of the 478 keV Y~ rav from 'Li. 2 7 Figure 13 shows a comparison of the Li (p,n)'Be excitation function as measured by Gibbons and Macklin 23 with the cross section measured with the polyethylene sphere detector. Agreement is very good up to a proton energy of 4.2 MeV. Toward higher energies, the yield observed with the polyethylene detector falls off relative to the yield observed with the graphite sphere counter. This fall-off of ef- ficiency of the polyethylene counter relative to the graphite sphere is shown as a function - 1 1 i i i i i i i i l 1 l i i i i i l i l"i it i i i i i i 800 - T U(p,n) T Be Reaction ERROR ± 4% - 600 - — Daio by Macklin and Gibbons • This work — Ratio of the front to the rear counts - 400 - _ _^=^. - '"' ■■•'"'•". 200 ; i i ,i, ,i, 2.0 3.0 4.0 5.0 INCIDENT PROTON ENERGY (MeV) Fig. 13, Comparison of the total (p,n) cross section as determined with the poly- ethylene sphere and by Gibbons and Macklin in the Macklin Sphere. of neutron energy in Figure 12. The decrease in counting efficiency for the polyethylene sphere is quite rapid above 2 MeV. The falloff of efficiency toward higher energies is primarily a geometrical effect re- sulting from the radial placement of the ^BF, counters; this means that the fraction of the volume filled with detector decreases inverse- ly as the square of the distance from the de- tector center. The average thermalization dis- tance increases with neutron energy and the spatial distribution of thermal neutrons spreads radially. This latter effect partially compen- sates for the detector geometry. The key to the energy response (Fig. 12) is the placement of the °BF3 counters in their radial channels. Figure 14 shows counting rate as a function of counter position for three different average neutron energies. The three lines come very near to intersecting at the same point. If the counters are placed at 1.2 cm, the appar- ent difference in efficiency is about 1% for the three energies shown. Figure 12 shows that the flat response extends nearly to 2 MeV. In addition to reasonably high sensiti- vity and flat response, the polyethylene sphere is quite insensitive to gamma-ray background. It has moderate sensitivity to neutrons coming 5 20 a (0 3 o x: H 0.5 1.0 1.5 2.0 2.5 Position (cms) Fig. 14. The counting rate in the polyethyl- ene counter as a function of counter position for three different aver- age neutron energies. The detectors were placed at 1.2 cm for operation. from surroundings. The time response is a few microseconds. Now I want to turn briefly to the perfor- mance of the detector as a device for measur- ing total neutron production cross sections from nuclear reactions, such as (p,n)and (a,n), of interest in stellar nucleosynthesis and nuclear reactor design problems. The detector and the y~ray counter, Ge (Li), were used to measure the neutron yield for y- ray count 21 from ^Cr for the 5 V(p,n)51cr re- action at incident proton energies of 3 and 5 MeV. This ratio was constant to a precision of 0.5% indicating that a precise neutron count is given by the polyethylene sphere even though the ground state neutrons at a proton bombard- ing energy of 5 MeV is 3.4 MeV (Q = -1.534); and, the sphere efficiency is decreasing with energy above a neutron energy of 2 MeV. This precise result is obtained primarily because the neu- tron spectrum emitted in (p,n) and (a,n) re- actions roughly follows a lumpy Maxwellian dis- tribution. The lumps correspond to indivi- dual excited states and groups of states. Such a spectrum is illustrated schematically by the dashed curve in Fig. 15. This dashed curve was derived from the time-of-f light spectrum shown, corrected for scintillator efficiency, and the assumption that the peak of the Maxwellian falls at about 0.5 MeV. 24 The time-of-f light spec- trum was taken with a 1.3 cm plastic scintilla- tor. Fewer than 10% of the neutrons emitted have energies higher than 2 MeV. The energy range in which the efficiency of the polyethylene sphere is flat can be in- creased by the insertion of trimmer counters into the sphere. For the present applications , this has not been necessary. 217 E n (MeV) ■4 .6 .9 |.3 2.2 3.4 i i i 1 T 5l V(p,n) 5l Cr , 1 3000 Ep 0b = 502 MeV n 5 CO e lob ■ 30- n. 1- Ij -z. o o 2000 i "4 n, ,n, j 1000 "il n U X 111 ii il / 1 1 I.I , 1 1 300 400 CHANNEL Fig. 15. Time-Qf-f light spectrum for neutrons from V(p,n) Cr. The dashed curve is a schematic representation of the neutron spectrum emitted at a pro- ton bombarding energy of 5 MeV. The LLL Li Detector A neutron detector for time-of-f light measurements in the 1 keV to 1 MeV energy 25 range has been designed by Czirr and Shosa' at the Lawrence Livermore Laboratory. The design concept is an outgrowth of the detector designs of W.P. Poenitz. -> Neutrons incident through a channel are moderated and captured in a 4 m of compressed LiH and the capture distribution is sampled with thin slabs of °Li glass scintillator. The mean capture time is approximately 100 nsec. Figure 16 shows a schematic diagram of this detector. The outer dimensions of the detector were determined primarily by the available lengths of Li-glass slabs. Slabs 165 mm long can be reliably produced so the detector length was fixed at 320 mm. The design provides for use of two pieces of glass to sample each longitudinal detection zone. Photomultiplier tubes view the glass slabs end on. The radius of the sampled region used in the design model was 200 mm. The glass Fig. 16. Schematic diagram of the totally absorbing detector showing end-on and side views. slabs were taken to be 2 mm in an effort to compromise between requirements of strength and good counting statistics and the require- ment of small flux perturbations. The glass slabs were separated by 10 mm in the model. The radius of the entrance hole was chosen to be 10 mm in a deuterated-polyethylene, (CD_) scatterer with a radius of 14 mm. The optimum depth of the re-entrant hole in the (CD-) scatterer was found to be 115 mm. The TARNP Monte Carlo transport code 26 which access the LLL cross-section library was used for the design calculations. Figure 17 shows the calculated efficiency for the model detector. The efficiency is flat over the energy range from 1 keV to 1 MeV. The error flags are an estimate of the computation- al uncertainty at each point. The error in the shape of the efficiency curve is estimated as 0.7%. The computational accuracy is about 2% as shown in Fig. 17. 1.30 = 1.15 g 1.05 Fig. 17. I0< I0» INCIDENT NEUTRON ENERGY (eV) Calculated efficiency of the LLL totally absorbing Time-of-Flight neutron detector. Calculation was done with computer program TARNP. io» 0.8 0.6 0.4 - 0.2 10 100 TIME (NANOSECONDS) 1000 Fig, Time response of LLL detector as a function of incident neutron energy. 218 The time response of the detector is shown in Fig. 18. Essentially all of the neutrons are captured in 1 usee and about half are cap- tured during the first 100 nsec. This time response is quite satisfactory for long flight paths such as encountered at Linacs. This detector is novel and exhibits good characteristics. It should be well suited for use at linac neutron sources. The materials required for its fabrication tend to be expen- sive. by Langsdorf , ° one very important region in a collimator is the throat which is the most nar- row part where neutrons are strongly scattered. The heavy shielding material near the center of the collimator (Fig. 19) serves two purposes. It is useful to have strong attenuation in the collimator throat; and, this material serves as a detector shield for the 2.225 MeV y-rays produced by the neutron capture in hydrogen. The intensity of such y-rays is large near the neutron source in hydrogen containing collima- tors. A Word About Collimators The black and grey neutron detectors which have been discussed are capable of ultimate accuracy of better than 1% for measurement of neutrons originating in a target at the center. Usage of these detectors with collimated neu- tron beams has been extensive. 13-18 In the first use mode, the neutrons detected are the primary nuclear reaction products; in the sec- ond, neutrons produced in a neutron source are collimated and caused to react with a target. The neutron flux is measured with the black or grey neutron detector. These detectors can accurately count all neutrons which reach the central region of the counter near the end of the re-entrant channel. One of the largest potential sources of error is the perturbation of the neutron "beam" by the material between the neutron source and the detector. When col- limation is used this perturbation takes two forms; first, the resolution function or energy definition is broadened and made less precise through energy losses in scattering from air, the target and the collimator material; and, second, a general background may be produced in the collimator materials which affects ac- curacy. An example of this is the production of 2.225 MeV y-rays in hydrogen containing col- limators which will affect detectors sensitive to y-rays. Most of these y-rays are emitted after the neutrons are moderated. 27 Spencer and Woolf provide important guidelines for the construction of neutron col- limators. Langdorf's work remains as the most complete treatment of neutron collimation. Figure 19 shows a collimator for use at a Van de Graaff accelerator. As has been emphasized j _J Li 2 Co, and Paraffin I Copper or Tungsten Fig. 19. A neutron collimator designed for keV neutrons. Tests of a collimator 1 m long at the Uni- versity of Kentucky by Cochran 2 ^ have shown that excellent geometrical definition of the neutron beam can be achieved for neutrons with energies in the range of 250 - 750 keV. This work also shows that fewer than 5% of the neu- trons in the collimated beam have energies differing from the primary beam by more than 25 keV. With proper design, it should be pos- sible to further reduce this fraction. Detec- tors such as the Black Neutron Detector and the NBS Black Detector with fast time response reject all but a fraction of a percent of the wrong-energy neutrons by the time-of-f light discrimination. Even so these "background" neutrons remain a problem since their effect will depend on the target and reaction studied. Further work on the transmission and beam quality of neutrons passing through such col- limators is needed. Summary As counting devices for neutrons in the energy range between 1 - 1000 keV, the black and grey neutron detectors show high potential for ultimate accuracy. High accuracy levels can also be achieved in the MeV energy range. Such detectors are useful in two operational modes. First, they may be used to detect neu- trons originating in their center. Second, they may be used to count the neutrons in a collimated beam. With current techniques, the accuracy of such detectors can be made better than 1% for neutrons originating at the center of the de- tector. The accuracy limit with a collimated beam of neutrons is less well established. It is known that long, thin collimators provide neutron beams of high quality; both with re- spect to their geometrical definition and with respect to energy definition. Ultimate accu- racy of measurement of the flux of collimated neutrons in the keV energy range with current techniques is estimated at 2%. In the neutron energy range between 1 keV and 2 50 keV, the moderated detectors even with their corresponding slow time response are among the best detectors for measurement of neutron flux. Acknowledgments I want to express gratitude for the oppor- tunity to make this presentation. Thanks go to J.B. Czirr, C.H. Johnson, R.L. Macklin,M.M. Meier, and W.P. Poenitz without whose help the review would not have been possible. Special thanks go to W.P. Poenitz. 219 *Work supported in part by the National Science Foundation and U.S.E.R.D.A. References 1. H.H. Barschall, L. Rosen, R.F. Tascheck and J.H. Williams, Rev. Mod. Phys . 24_, 1 (1952) . 2. K.E. Larsson, Ariv. for Fysik 9, 287 (1954). 3. J.E. Perry, Jr., Fast Neutron Physics , Part I, J.B. Marion and J.L. Fowler, eds.; Interscience Publisher, New York, p. 623 (1960) . 4. R. Batcherlor, in Proceedings of the Con- ference on Neutron Cross Sections and Technology NBS Special Pub. 299, Vol. 1, Wash., D.C. (1968), p. 89; J.H. Gibbons, Ibid. p. 111. 5. H.H. Landon , in Proceedings of the Con- ference on Neutron Cross Sections and Technology, Knoxville, Tennessee, U.S. A.E.C., Conf. 710301, p. 528 (1971). 6. Proceedings of the Symposium on Neutron Standards and Flux Normalization, October 21-23, 1970. Available as CONF-701002 from National Technical Information Ser- vice, U.S. Department of Commerce, Spring- field, Virginia 22161. 7. Proceedings of the Conference on Nuclear Cross Sections and Technology, National Bureau of Standards, NBS Special Publica- tion 425 (1975). Available from U.S. Government Printing Office, Washington, D.C. 20402. 8. Proceedings of the International Confer- ence on the Interactions of Neutrons with Nuclei, University of Lowell, Lowell, Massachusetts (1976). CONF-760715-P2 . 9. R.L. Macklin, Nucl. Instr. 1, 335 (1957). 10. R.L. Macklin and J.H. Gibbons, Phys. Rev. 109 , 105 (1958); J.H. Gibbons and R.L. Macklin, Phys. Rev. 114 , 571 (1959); C.H. Johnson and R.L. Kernell, Phys. Rev. C 2_, 639 (1970); C.H. Johnson, J.K. Bair, CM. Jones, S.K. Penny, and D.W. Smith, Phys. Rev. C 15_, 196 (1977). 11. C.H. Johnson, Private Communication, ( 1977 ). 12. R.L. Macklin, F.M. Glass, J. Halperin, R.T. Roseberry, H.W. Schmitt, R.W. Stough- ton, and M. Tobias, Nucl. Instr. and Meth. 102, 181 (1972). 13. W.P. Poenitz, Nucl. Inst, and Meth. 58 , 39 (1968); W.P. Poenitz et al., J. Nucl. Energy 22, 505 (1968); H.O. Menlove and W.P. Poenitz, Nucl. Sci. Eng. 3_3, 24 (1968); W.P. Poenitz, Nucl. Instr. and Meth. 72, 120 (1969) . 14. M.S. Coates, G.J. Hunt, C.A. Uttley, and E.R. Rae, Reference 6, p. 401 (1970). 15. W.P. Poenitz, "Neutron Standard Reference Data", International Atomic Energy Agency, Vienna (1974); W.P. Poenitz, Nucl. Inst. and Meth. 109, 413 (1973). 16. W.P. Poenitz, Nucl. Sci. and Eng. 53 , 370 (1974). 17. W.P. Poenitz, ANL-7915, Argonne National Laboratory (1972). 18. G.P. Lamaze, M.M. Meier, and A.O. Wasson, Reference 7, p. 73 (1975). 19. M.M. Meier, Private communication (1977). 20. L. Wishart, R. Plattner and L. Cranberg, Nucl. Instr. and Meth. 57_, 237 (1967). 21. K.K. Sekharan, H. Laumer, B.D. Kern, and F. Gabbard, Nucl. Inst, and Meth. 133 , 253 (1976). 22. M. Mutterer, Reference 6, p. 452 (1970). 23. J.H. Gibbons and R.L. Macklin, Phys. Rev. 114 , 571 (1959). 24. J.M. Blatt and V.F. Weisskopf, Theoretical Nuclear Physics , John Wiley, New York (1952) , p. 365 ff . 25. J.B. Czirr and D.W. Shosa, " A Totally - Absorbing Time-of-Flight Neutron Detec- tor . " Lawrence Livermore Laboratory, Livermore, California 94550. Preprint UCRL-79089. 26. Ernest F. Plechaty and John R. Kimlinger, TARNP : A Coupled neutron-photo Monte Carlo transport code, UCRL-50440, Vol.14 (1976) . 27. L.V. Spencer and S. Woolf, Nucl. Instr. and Meth. 97_, 567 (1971) . 28. Alexander Langsdorf, Jr., in Fast Neutron Physics , Part I , J.L. Fowler and J.B. Marion, eds. (Interscience Publishers, Inc., New York, (1960) p. 721. 29. J.L. Cochran, M.S. Thesis, University of Kentucky (unpublished) . 220 ASSOCIATED PARTICLE METHODS Michael M. Meier Developments in the associated particle method for the last ten years are summarized with emphasis on the reactions 3 H(d,n) 4 He, 2 H(d,n) He and H(p,n) He. Recent progress on the associated particle calibration of a black detector in the energy range 250 to 1000 keV is reported. Current accuracies for the time uncorrelated and time correlated approaches are noted and prospects for future improvement are estimated. protons) (calibration; efficiency; heutronsj neutron beams; neutron flux; H, H, He, He, Introduction H(d,n) He Reaction The associated particle technique (APT) has been used as a monitor of neutron flux for about thirty years and has been discussed in several reviews of neutron flux monitoring techniques. This survey includes major developments in the method utilizing the 3 H(d,n) 4 He, 2 H(dn) 3 He and 3 H(p,n) 3 He reaction from 1970 to the present. These reactions have been used for flux monitoring at energies between about 200 keV and 25 MeV. The APT has been used in two rather different ways and there are distinct advantages associated with each. In the first, the charged particle reaction products are collimated and counted in a detector, and the neutron flux is then calculated for the correspon- ding neutron angle. This method has the advantage that the electronics can be rather simple, needing only to identify the charged particles with which the neu- trons are associated. Also, if the neutron flux does not change rapidly with angle, or if its angular behav- ior is well known, then the neutron detector or sample can subtend a large solid angle. This approach is also useful when the neutron detection system is not amenable to fast timing techniques or when high count rates associated with large solid angle are necessary. Essential to the success of the method is accuracy in measuring the effective solid angles for both the neutron detector and the charged particle collimator. Angle measurement is also critical, since the differ- ential ratio of solid angles enters into the calcula- tion and may be a rapidly varying function of angle. If the neutron detector subtends an appreciably larger solid angle than the one associated with charged par- ticle detection a correction dependent on the differ- ential cross section may be necessary. A further problem is the correction for background neutrons which are not associated with the direct flux. The second technique correlates the detected neutron with the charged particle of interest by coincidence or time of flight methods. One is restric- ted obviously to detectors which have fast response, although in principle they need only be fast relative to the data acquisition rate. The neutron detector must also be larger than the associated neutron cone, which normally implies that the extent of this beam must be experimentally determined. This usually restricts the size of the charged particle solid angle and with it the associated neutron production rate. It is also clear that this is a cali- bration of only a part of the detector and uniformity may be a problem. Advantages which accrue are the elimination of geometry determination and suppression of neutron backgrounds by coincidence or time of flight. Once the experiment is properly set up there is a minimum of calculation and correction to the data. The efficiency of a detector is simply the ratio of the correlated rate in the neutron detector to total associated particle counts. In the following, these two methods will be referred to as the " time-uncorre- lated" and ' time correlated" methods respectively. Several properties of this reaction make it at- tractive to use for absolute flux monitoring. The + 17.6 MeV Q value provides a source of 14 MeV neutrons and associated 4 MeV alpha particles which are there- fore well separated in energy from scattered deuterons at bombarding energies of only a few hundred keV. The attractiveness of this technique to a laboratory with a low energy accelerator is enhanced by the 110 keV resonance with a peak value of about 400 mb sr ± . Also other (d, charged particle) reactions in target components, e.g. titanium or zirconium and backing materials, which would provide background for the alpha particles are strongly suppressed by the Coulomb barri- er. There are also serious difficulties encountered in the use of low bombarding energies and thick tritiated targets. First, the identification of alpha particles is complicated by the presence of an intense Coulomb scattered deuteron background which is inverse- ly proportional to Ej. If detected, these deuterons can cause saturation of electronics, baseline shifts due to pulse pileup and other problems associated with high rates. One simple technique 4 to eliminate this problem has been to interpose a metal foil between target and detector. The foil thickness is selected to be slightly greater than the deuteron range, permitting the alpha particles to pass through with slightly in- creased straggling. Another problem arising from Coulomb scattering is the broadening of the neutron cone beyond the limits kinematically defined by the alpha detector. At low bombarding energies this effect is largest and predom- inately due to deuteron scattering. For experiments in which the neutron beam must be smaller than the detec- tor, this effect must be taken into account. Another set of problems arises because of the time dependence of the tritium distribution in a solid target. For these targets, where the total beam power is dissipated in the target and substrate, tritium is depleted from the surface layer and the effective mean deuteron energy is decreased. For a fixed charged particle detection angle, the associated neutron angle increases (see, for example Figure 2, Ref. 4) and thereby causes misalignment of the neutron detector relative to the beam center. This problem is not usu- ally important for a time uncorrelated experiment, since the laboratory (d,n) cross section varies by less than a percent for a typical five degree change in neutron angle. Finally, for time uncorrelated experiments, back- ground neutrons may result from scattering in the tar- get and beam line structures and from production via (d,n) reactions. In particular, drive in deuterons are always a source of background for experiments using a solid target. References 5, 6 and 7 are recent examples of the APT with low energy deutrons for measurement of fission cross sections, (n, charged particle) spectro- metry and for evaluation of collimetor design. If one wishes to extend the energy range of neu- trons away from 14 MeV the use of higher bombarding 221 energies becomes necessary. Although the Coulomb scat- tering effects are thereby reduced, other difficulties become important. Specifically, beam power is usually increased, accelerating the tritium loss and the conse- quent neutron angle change. Also, other (d, charged particle) and (d,n) reactions contaminate the alpha particle and neutron spectrum respectively. A unique scheme employed by Cookson et al. 8 utilizes a tritium gas target chamber to obtain neutron energies between 14 and 25 MeV. Many experimental difficulties are eliminated by utilizing such a system. First, problems associated with the target backing disappear. The Coulomb scattering, neutron beam broadening and neu- tron angle migration are all essentially eliminated. Background in the charged particle spectrum is reduced since the alpha particle collimation shields the sur- face barrier detector from reactions in the entrance and exit foils and from the beam dump where (d charged particle) reaction products would be produced. This apparatus has been used to obtain an absolute cali- bration for the efficiency of a 2.5 cm thick by 10 cm diameter NE-213 scintillator. Neutron background orig- inating in foils and beam defining slits is small and easily corrected since the experiment uses the time- of-flight method for obtaining the detector efficiency. s H(d,n) 3 He Reaction Many of the remarks concerning the H(d,n) reac- tion at low bombarding energies apply to the 2 H(d,n) reaction as well. For acceleration voltages lower than 500 keV monoenergetic neutrons spanning the energy range 2 to 3.5 MeV can be produced with appreciable yield. In at least two important respects this is a less favorable reaction. First, the competing s H(d,p) H(Q = 4.033 MeV) complicates the spectrum of 3 He from the (d,n) reaction (Q = 3.269 MeV), the triton and He differing i n energy by only a couple hundred keV for energies below 500 keV bombarding energy. Second, the 3 He energy is usually less than 1 MeV, giving rise to Coulomb scattering problems for the re- coiling particles as well as the bombarding deutrons. Apart from the degradation of energy resolution in the He peak, this effect also contributes to the cone broadening and migration. Finally, the cross section for this reaction is only 20 mb sr" 1 at 500 keV and increases with increasing energy, becoming 90 mb sr" at 10 MeV bombarding energy. Use of a deuterium gas target for this reaction has been employed with the chamber discussed above. The benefits of a gas target and higher bombarding energies are the same as those discussed for the H(d,n) reaction plus the fact that the yield increases with increasing energy. A transmission surface barrier detector is used in this experiment in order to detect the He particles in the presence of tritons with almost identical energy. The thickness of the detector is selected to stop He particles and allow the tritons to pass through, losing only a fraction of their energy. Breakup and other reactions that produce neutrons are not a problem since the experiment time correlates charged particles and neutrons. Applications of the APT have been made at bom- barding energies below 500 keV by employing electro- static 9 and magnetic 10 analysis of the reaction prod- ucts to eliminate scattered deuterons and tritons. Foils were used to stop the scattered deuterons for an experiment 11 at 150 keV bombarding energy although background in the charged particle spectrum is a dif- ficult problem under these conditions. Bartle and Quin have used a surface barrier detec- tor 'E - E pair. 18 The AE detector has a thickness equal to the range of the 3 He particles. Tritons and elastically scattered deuterons are transmitted and detected in the E detector which produces a veto for data accumulation. To reduce the Coulomb scattering a deuterated polyethylene target 0.5 pm thick is used in a transmission position. Target rotation extends the lifetime to about 100 hrs with a 0.5 pA beam collimated to a 2 mm diameter. Background in the charged particle spectrum becomes a problem when the 12 C(d, a ) 10 B(g.s.) alpha particles begin to interfere with the 3 He peak. This background can be reproduced by using a carbon film of similar thickness in place of the polyethylene. A simple stripping technique is then sufficient to separate the 3 He peak. This system has been used to produce calibrated neutron beams from 2 to 10 MeV. Other applications of the s H(d,n) reaction include a calibration 13 of a neutron detector in the energy range 25 to 60 MeV. Also, a novel approach to the time uncor related method has been used by Ryves and Sharma. 14 Here the H(d,p) protons are monitored at an angle where (d,p) and (d,n) cross sections are compara- ble. 3 H(p,n) 3 He This reaction has been used extensively in the past decade, its attractiveness stemming largely from the possibility of monitoring flux at energies as low as 100 keV. To obtain energies this low the associated He particles must be detected at angles on the order of 10 , where the ratio of Coulomb to (p,n) cross sec- tion is 10 s . It is necessary to cope with these condi- tions in order to obtain He particles with sufficient energy to escape the tritiated target and be detected unambiguously. Under these "favorable" conditions the energy spread of 1 MeV He's is 157„ for a 100 ug/cm 2 titanium layer. Again, the neutron cone is broadened, this time almost exclusively on account of the recoil particle scattering. Apart from the large elastic proton background, inelastic protons from the target and collimating surfaces and tritons from elastic scattering H(p,t) are present in the charged particle spectrum. Fort and co-workers " have used this technique for measurements in the range 100 to 500 keV. Transmission targets 200 u,g/cm thick are used and by employing se- quential electrostatic and magnetic fields the separa- tion of He particles from background is very clearly accomplished. The solid angle for this technique is rather small, limiting the count rate, but providing a consequent advantage. The apparatus can be turned to accept a very limited band of He energies and in this way actually select those He particles which have undergone a minimum of Coulomb scattering in the tar- get. By operating in this fashion the angular and energy spread of the neutron cone can be minimized. This system has been used to calibrate the effi- ciency of a Li - loaded glass scintillator. The mea- surement was complicated by the necessity for making a large multiple scattering correction in the glass to extract the Li cross section. This calculation in- cluded a Monte Carlo calculation of the Coulomb scatter- ing of the He and its effects on the energy spread and angular broadening of the associated neutron beam. The largest uncertainty in this measurement was for the conversion of the efficiency as measured at a distance of 5 cm from a point source to that which would result in a parallel beam. Liskien, Paulsen and co-workers have com- bined an electrostatic deflector and a pulsed beam time of flight system to produce calibrated beams between 250 keV and 1 MeV. The deflection is used to suppress the elastic proton rate by sweeping the 222 3 He into an offset surface barrier detector. A gate is generated by particles which have the source to de- tector flight time appropriate for a 3 He particle. This logic signal is then used to generate gated pulse height spectra for the surface barrier detector. Experiments done with this apparatus have been done using the time uncorrelated approach. As previ- ously mentioned, the calculation of flux, for this case involves determination of the effective solid angle for charged particle collimation, solid angle of neu- tron detection and (dfi 3ll /dO ). That careful measure- He n ment of these quantities is essential is illustrated by the fact that (d0 3 /dfi ) varies by 57. for a one degree change in the 3 He detection angle. The use of deflection complicates the time uncorrelated measure- ment in two ways. First, assurance must be obtained -a ++ that all the He which are collimated into the deflection system are, in fact, detected by the solid state detector. This can be ascertained by varying the deflection voltage and operating on the plateau of counting rate. Also, since neutral and singly-ionized 3 He are associated with the flux at the neutron detec- tor, a correction of order 10% must be made. The lat- ter can be partially checked by experiment. By vary- ing the deflecting field and using a 2 mm collimator Paulsen, Liskien and Cosack have measured the rela- _1_ _J L tive 3 He to 3 He yields with their associated par- ticle system. These measurements agree well with calculation and give an added measure of confidence to the correction. Excellent agreement was obtained using this method to calibrate a proton recoil propor- tional counter. Leroy has used the associated particle apparatus described earlier in the time uncorrelated way. By properly tuning the electrostatic - magnetic analyser, it is possible to eliminate to first order the effects of spatial spread of the 3 He" 1 "" 1 " beam which is due to its energy dispersion. Because of this, the effective solid angle of the 3 He beam can be increased and a larger corresponding solid angle of neutron flux can be measured. Count rates a factor of 50 to 100 higher than with the time correlated technique are possible with this method. The system has been used to check the calibration of a long counter which had been pre- viously calibrated with a MnS0 4 bath. The calibrations are in good agreement within the - 27. error bars. At NBS we have used a setup similiar to Liskien's, employing pulsed beam and electrostatic deflection to identify the 3 He particles. Two chambers for detection of recoil particles at 10°and 25 are used to cover the range of neutron energies 250 keV to 900 keV. The Coulomb scattering is not severe at 25 , and no elec- trostatic deflection is used there. A two parameter data analysis system is now used for obtaining the He data, providing an easy to use method for background determination. Figure 1 shows such a spectrum from an experiment where electrostatic deflection is used. Each of the 32 time channels has a width of about 7ns and is comprised of a 128 channel pulse height spec- trum. The background under the peak is about 5% of the 3 He rate and can be very accurately determined by gating the spectrum with a logic signal generated by neutron detection. This method provides a powerful technique for determining the 3 He lineshape. It is especially useful in cases where no deflection is used and the 3 He peak can be incompletely resolved from tritons from proton elastic scattering, T(p,T). CHARGED PARTICLE SPECTRUM 2 2 MeV Figure 1. Charged particle spectrum for 2.2 MeV protons. The resolved group is 3 He from 3 H(p,n) 3 He The system has been used to calibrate a black de- tector " which has a response calculable by Monte Carlo technique. For the calibration the detector is placed at the appropriate angle with the front face 11 cm from the target. The anode signal from the photomultiplier is supplied to two linear gates which are gated by logic signals generated by He events. The first gate operates normally, opening for 200 ns to permit the neutron event to be analysed when an associated He event occurs. The second gate opens one microsecond earlier corresponding to the preceding proton pulse and thereby measures accidental background that is not physically correlated with true 3 He events. After subtracting this background, the reduced spectrum is normalized in area and channel width so as to be directly comparable to the Monte Carlo calcula- tion. The program used is "CARLO BLACK" 20 by W. P. Poenitz modified to include the effects of Poisson statistics for photoelectron production at the photo- cathode. Figure 2 shows data taken at 250, 500, 700 and 880 keV and the corresponding Monte Carlo calcula- tions. The data were all obtained under the same con- ditions and in particular no gains or bias voltages were changed in the neutron electronics. In reducing the data, the same bin width conversion factor was used for all energies so that the agreement between experi- ment and calculation could be compared as a function of energy. The Monte Carlo generated response functions are shown as smooth curves for each energy in the fig- ure. Two curves are shown for each energy, correspon- ding to levels of photoelectron production which differ by 50% for a given light output. These comparisons are very encouraging in that calculation and experiment qualitatively agree for a single set of parameters for all energies. The spread- ing of the peak seems to be well described by a single Poisson parameter between the two shown here and the peak amplitude shifts as calculated. This latter indicates that the light tables used are valid in a relative sense over this energy region. The efficien- cies for the detector when a bias corresponding to channel 20 is chosen are compared for calibration and calculation in Figure 3. The agreement seems accept- able at the 2.57. level. The black detector is now being used as a flux monitor in absolute cross section measurements and for dosimetry at the NBS Van de Graaff. Our confidence in the system is enhanced by the understanding we have obtained in bringing experiment and calculation into the detailed agreement shown above. 223 2000 LU BLACK DETECTOR CALIBRATION ASSOCIATED PARTICLE RESULTS ° 250 keV a 500 keV o 700 keV • 880 keV 1000 — MONTE CARLO CALCULATIONS 22 5 keV/ PHOTOELECTRON 34 keV/ PHOTOELECTRON Figure 2. Experimental and calculated response functions for the black detector Conclusion 0.96 0.94 > o z UJ y 0.92 O U 0.90 0.88 0.86 1 1 ' ' ' ' 1 " : > E - — BLACK DETECTOR x EFFICIENCY ■ ; - ■ - x CARLY CALCULATION J ° ASSOCIATED PARTICLE CALIBRATION - 1 1 1 1 I 1 i — Figure 3. 100 300 500 700 900 En (keV) Measured and calculated detector efficiencies for the black, detector It is perhaps worthwhile to summarize by reviewing the present APT accuracy and to speculate on future improvements. The best accuracies obtainable with the time uncorrelated method have been on the order of 3%. A major limitation of this method, as mentioned previ- ously, is the difficulty in determining the solid an- gles for neutron and charged particle detectors. Un- certainties for the latter are quoted to be 0.57„. A neutron detector in a divergent beam with finite source dimension may be even more difficult. Without some breakthrough in this area it is unlikely that accura- cies much better than 17„ overall will be obtained. For the time correlated method the accuracy is limited by the details of the particular reaction con- sidered. For 3 H(d,n) Cance and Grenier estimate their uncertainty in the APT to be 0.1%, only dependent on counting statistics of the alpha particles. This im- pressive accuracy is a good bit better than can be expected for sample assay or efficiency calculation in the near future. It is probable that experimenters will therefore attack these latter problems rather than improve this accuracy level. The 3 H(d,n) reaction with a gas target has the po- tential for a very high ultimate accuracy. He identi- fication is quite unambiguous with detectors that trans- mit the competing tritons. Also, the Coulomb scat- tering which disperses the neutron beam is due only to the tritium and not to any intervening foils. The accuracies obtained in a scintillator calibration have 224 been about 2Y„, due mostly to uncertainties in extending the calibration over the complete surface of the detec- tor. Presumably the accuracy could be considerably improved in the calibration of a detector with a re- entrant hole or in a cross section measurement where sample assay and uniformity were well known. The 3 H(p,n) reaction has a current accuracy of 1 - 37. using the time correlated approach. Some of this uncertainty is due to problems not inherent to APT, but to sample and geometric difficulties. A good example is the non-trivial problem of converting an efficiency in a divergent beam back to the normal parallel beam situation. APT measurements will, as a rule, have such detector or sample problems and it would be inappropri- ate to attempt a listing of all of them here. There are two areas of concern specific to the APT which are common to all 3 H(p,n) applications (and to some extent the other two reactions) and which now limit our use of the method. First is the identification of charged particles. This has been done with an accuracy of 0.17, using very sophisticated deflection systems but at a large sacrifice in count rate. At NBS we now cope with a 5% background known to 20% accuracy and can obtain with some effort 17. ± 0.57.. The technique of using the neutrons to gate the charged particle spectrum and thereby determine the lineshape has been mentioned above. We plan to combine this method with a computer data collection system that utilizes "tagging" for very accurate determination of background. It is likely that we can obtain better than 0.27. accuracy in charged particle background in this way. A second problem is somewhat less tractable. Fig- ure 4 shows the angular profile of our neutron beam for two different targets with a 3 He energy of 1.2 MeV 10" 10 10' 10 I - IOO M g/cm 2 35 M g/cm 120 130 NEUTRON ANGLE 140 DEGREES 150 Figure 4, Spatial profile of associated neutron beam for two targets of different thickness where the Coulomb scattering is rather severe. A prob- lem that ultimately limits accuracy is the question of how many associated neutrons are found outside the cone defined by a given detector or sample. Such neutrons arise from scattering by the chamber walls and by cone broadening due to Coulomb scattering. The former are usually calculable, although for our 0.8 mm aluminum wall the correction is on the order of one percent. One way to experimentally check on the Coulomb scattering problem is to perform the measurement with differing target thicknesses. We have checked our black detector calibration and do not find differences at the 27. level for the targets represented in the figure. Tuning the electrostatic analyser to select 3 He ' s which have undergone different amounts of Coulomb scattering might also provide some measure of this effect. It is clear that both the above problems could be diminished by the development of new targets, for example, tritium gas with very thin windows or tritiat- ed polyethylene. References 1. H. H. Barschall, L. Rosen, R. F. Tascek and J. H. Williams, Rev. Mod. Phys . 24, 1 (1952). 2. J. E. Perry, Fast Neutron Physics, Vol. 1, J. B. Marion and J. L. Fowler, Interscience Publisher, New York (1960). 3. R. Batchelor, Neutron Cross Sections and Technol- ogy, NBS Special Pub. 299, Vol. 1, Washington, D. C. (1968). 4. H. Marshak, A. C. B. Richardson and T. Tamura, Phys. Rev., L50, No. 3, 996-1010 (1966). 5. M. Cance and G. Grenier, Trans. Am. Nucl. Soc. 22 , 664-665, Nov. 1975. 6. C. Sellem, I. D. Perroud and J. F. Loude , Nuclear Instruments and Methods 128 , 495 (1975). 7. K. M. Jones, E. P. Cytackiand C. S. Kelsey, Phys. Med. Biol., 20 (1), 131-135 (1975). 8. J. A. Cookson, M. Hussian, C. A. Uttley, J. L. Fowler and R. B. Schwartz, Neutron Cross Sections and Technology, Vol. 1, pp. 66-68 (1975). 9. R. B. Galloway and A. Waheed, Nucl. Instr. and Methods jL28, 505 (1975). 10. P. B. Johnson, J. E. Callaghan, C. M. Bartle and N. G. Chapman, Nucl. Instr. and Methods, 100 , 141-148 (1972). 11. A. Greil and K. Triitzschler, Kernenergie, 12, 68-70 (1969). 12. C. M. Bartle and P. A. Quin, Nucl. Instr. and Methods J_21 , 119-127 (1974). 13. R. A. J. Riddle, G. H. Harrison, P. G. Rose, M. J. Saltmarsh, Nucl. Instr. and Methods 121 , 445 (1974). 14. T. B. Ryves and D. Sharma, Nucl. Instr. and Methods _L28, 455 (1975). 15. E. Fort, Nuclear Data for Reactors (Proc. Conf. Helsinki: 1970) 1_, IAEA, Vienna (1970); E. Fort, J. L. Leroy and J. P. Marquette, Nucl. Instr. and Methods 85, 115-123 (1970). 225 16. H. Liskien and A. Paulsen, Nucl. Instr. and Methods 69, 70-76 (1969). 17. A. Paulsen, R. Wldera, A. Berlin and A. Trapani, Nucl. Instr. and Methods 9_1> 589-593 (1971). 18o A. Paulsen, H. Liskien and M. Cosack, Nucl. Instr. and Methods U15, 103-107 (1972). 19. G. P. Lamaze, M. M. Meier and 0. A. Wasson, Nuclear Cross Sections and Technology, Vol. 1, pp. 73-74 (1975). 20. W. P. Poenitz, ANL-7915, Argonne National Laboratory (1972). 21. M. M. Meier, A. D. Carlson and G. P. Lamaze, Nuclear Cross Sections and Technology, Vol. 1, 75-77 (1975). 22. J. L. Leroy, I. Szabo and J. Y. Tocquer, Neutron Standard Reference Data, IAEA, pp. 63-73 (1972). 226 ASSOCIATED GAMMA-RAY TECHNIQUE FOR NEUTRON FLUENCE MEASUREMENTS by J. D. Brandenberger University of California Los Alamos Scientific Laboratory Los Alamos, New Mexico 87545 The use and development of the 7 Li (p,n -^J 7 Be reaction as an example of the associated gamma-ray technique for neutron fluence and detector efficiency measurements are described. Present limits on energy range and accuracy are stated for this method, and current extensions of this work are discussed. (Neutron fluence; detector efficiency; cross sections) Introduction Associated gamma-ray techniques have been used ex- tensively in nuclear physics. Two examples are the gamma-gamma coincidence method that has been used for decades in analyzing gamma-ray cascades of excited nuclei in nuclear spectroscopic studies, and coinci- dent measurement of annihilation radiation to study the interaction of positrons with matter. Several years ago, with the advent of high resolution Ge(Li) detectors, it became convenient and enhanced experi- mental precision to monitor particle induced neutron producing reactions by a gamma-ray associated with a neutron group. With respect to neutron measurements, the 431-keV gamma-ray from the Li(p,n,y) Be reaction has been used at Lowell University to monitor the relative fluence from the 7 Li(p,n ) 7 Be reaction. 1 While this gamma ray is not actually associated with the n neutrons, the principle is similar because at a given proton energy, the branching ratio of n-]/n neutrons is constant. The closest example to the technique described here is the use by Presser and Bass in determining both the Li (p,n, y) 7 Be to 7 7 Li(p,p y)'Li branching ratio and the integrated cross sections by observing the 431- and 478-keV gamma rays, respectively. That work, however, re- quired proton fluence and target thickness measure- ments in order to obtain the absolute total reaction cross sections. The present work uses associated gamma-ray techniques in conjunction with a neutron source reaction to measure angular distributions which for simplicity and accuracy can be used to measure the neutron fluence and detector efficien- cies while avoiding target thickness and proton beam current measurements. Use of the Method For the moment, let us assume that a set of 7 Li (p,n.Y) 7 Be angular distributions as a function of proton energy are known with good precision either on an absolute or relative scale. The measurement of these angular distributions must be made before the associated gamma-ray method can be used with this reaction. This subject is left for later discussion because of its critical relation to the accuracy of the method. Figure 1 shows the spectroscopic information in- volved in the example reaction. The n, neutron leaves 7 Be in the l/2~, 431-keV level, which promptly decays to the 3/2 ground state. Because of the spin values involved, the decay of the level is by the isotropic emission of the 431-keV gamma ray, and the ratio of the number of these gamma-rays to n, neu- trons is unity from threshold until the 4.5-MeV level is reached. Thus, from £ E n < 4 MeV one may moni- tor the n, neutrons by observing the 431-keV gamma ray. If the absolute efficiency of the gamma-ray de- tector has been determined for the 431-keV gamma ray by comparison to gamma-ray standards that are now available from, for example, the National Bureau of Standards (NBS) , the absolute n^^ neutron fluence may be determined at each proton energy and detector angle. A high-resolution gamma-ray detector must be used in order to resolve the 431-keV gamma rays. If the objective is to determine the absolute fluence of n^ neutrons, the above measurements are all that is necessary. To calibrate a neutron detector with this reaction, it is necessary to make time-of-f light measurements of the nj neutrons because of the presence of n Q neutrons, gamma rays, and room-scattered background. Subsequently, a second detector may be calibrated for monoenergic neutrons by direct comparison with the first if only monoenergetic neutrons are present. The method was developed to measure detector efficiencies as a function of energy with a combination of ease and precision that has not previously been possible in this energy region (E < 4.1 MeV). The ultimate motivation, of course, is the measurement of neutron differential cross sections normalized to the n-p cross section at laboratory angles of about 40°. Figure 2 shows a set of 7 Li(p,n 1 ) 7 Be angular dis- tributions, 3 which are considered for the present to be precise. We can use these angular distributions, or we can use the information 3 in Fig. 3 to generate angular distributions from interpolated polynominal coefficient ratios. Each angular distribution can be expressed as i. = I = i measurable max P' w(9,e ) = y ft) A. (6,E )P„ (cosS) I p I A. = A where A„/A is known, and I o w(9,e ) P max / A \ a ( E ) y f-i) The total number of n-^ neutrons N emitted by the source is obtained by dividing the number of detected 431-keV gamma rays, N Y , by the gamma-ray detector's absolute efficiency, c . Thus, N = N Y /e Y or N = Kfw(6,EJdR = N /e . J P y Y The fluence at a particular energy and angle is KW F(6,E ) = N kw(8,e )/n 227 5/2" 7480 (5/2') 6 U+n S560 s7/?r 4630 in o o o I/2 - " 477.4 (5/2") 7190 5/2" 6510 5606.4 7/2" 6 L i 4500 3/2" '7xl0" l4 s 0.0; 1/2" 431 2.7kI0" I3 s 3/2" ' 0.0 'Be 53.28 d 9000 - 8000 7000 6000 5000 - 4000 3000 2000 1400 1200 1000 - 800 - 600 - 400 - 200 Fig. 1 The level and decay schemes of Li and Be. 228 _> "«♦- en m c o »IIW X5 CO a 3 c < m c Q. -J 7 UCp.^)'Be E p (MeV) 0, cm Fig. 2 7 7 The Li(p,n ) Be angular distributions from E =3.1 MeV to E = 4.9 MeV. Units are relative 1 P P 229 0.0 -0.2 -0.4 o < CD 1 r <> 0.4 0.2 0.0 0.2 0.4 i -i 0.0 0.2 0.4 -0.6 • • " • ■ I ' I ' ■ • • • a J i I i 1 i L 1 I ' — I — r A /A l_j_J__j L ^> A 2 /A * • ■ • I • * • 7 Li( p,n,) 7 Be ■ Elbakr et al • Present work Z i I i I i 1 i L J I L <> A, /A o • • i • ■ i i i i . i i i i i . 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 ' .0 Ep (MeV) Fig. 3 The Legendre Polynominal ratios that fit the angular distributions of Fig. 2. The dots are from Ref. 3 and the squares from Elbakr et al . 230 where fi d and S d are the solid angle subtended by the detector and its effective area, respectively. Thus, we simply have F(6,E ) = N /e P Y Y W(9,E ) P_ 4tt A i/ o where the distance from the source is L. Note that £ is a property of the gamma-ray detector that has been measured, L appears as the 1/R dependence of the flux, and W/A is determined by the known ratios of the polynominal coefficients. The only quantities measured for each fluence determination are N v and 6, which are extremely easy to determine. Y count, assuring that each point on an angular distri- bution is monitored to a constant integral fluence of n, neutrons. If an efficiency curve is found which fits both the data points of each angular distribu- tion and each of the integral cross sections, it is assumed to be correct. This assumes that the effi- ciency curve is smooth or that enough data are taken to establish any fine structure in the efficiency curve. The precision of the efficiency curve is taken to be not much greater, say 1.5 times greater, than the standard deviation of the data points from the assumed efficiency curve. There is no obvious means to determine a valid statistical uncertainty in the usual sense for this method. To bring us to a final result, the absolute ef- ficiency of a neutron detector whose yield is YfE^,) is e = Y(E n )/F(9,Ep)xS for a detector with effective (or nominal) area S, and the relative efficiency is e , = Y(E )/N xW(6,E ) . rel n Y P For determining Y and N y , it is assumed that correc- tions for attenuations by the target, target backing, and target assembly have been made appropriately as the experimental arrangement requires. To obtain a detector efficiency in the range of applicability of the method with the Li(p,n-,Y) Be reaction, i.e. to 4 MeV, typically two or three angular distributions of the 7 Li(p,n 1 Y) Be reaction are necessary. This can be seen in Fig. 4, where the angular distribu- tions denoted by A, F, and L have overlapped nicely in energy to give an efficiency curve. By knowing the efficiency e, both elastic and inelastic neutron cross sections can be normalized to the n-p cross sections. If the absolute efficiency is determined as outlined, (p,n) , (a,n), (d,n) and in general all neutron source reactions below 4 MeV can be deter- mined provided target thicknesses and appropriate particle currents are known. It is important to realize (1) that the standard deviation of the points from a smooth curve (and probably also the correct efficiency curve) is <2%, which is probably indica- tive of the accuracy as well as precision of the results, and (2) that the results are independent of detailed knowledge of (a) the gamma-ray detector's position and effective area, (b) the proton current or target thickness, or (c) the neutron detector's area. Present Status Figures 2 and 3 indicate the sole source of com- pleted measurements using the associated gamma-ray method. These are from Ref. 3. A statistical anal- ysis indicates that the standard deviation of the set of data points from the assumed efficiency curve de- rived by the fitting and normalizing procedure is 1.9% in the region above E n = 0.4 MeV. Many alter- nate efficiency curves have been considered and have been rejected as being erroneous or at least inferior to that shown. Thus, considering the average and individual errors as 1.5 times the standard deviation over most of the energy range gives an uncertainty of about 3% from 0.6 MeV to 2.75 MeV. The higher energy region, where only one angular distribution is repre- sented, may have slightly larger errors, but this an- gular distribution is required to be compatible with a large number of points from several angular distri- butions at lower energies and is consistent with the behavior of efficiency curves well past the maximum. Further Developments in Energy Range and Accuracy Further work is underway by Ron Harper, E. C. Hagen, B. D. Kern and the author to extend the energy range and accuracy of the method. Data have been taken, and continue to be taken by Harper and Kern, to extend the range from 70 keV to 4.1 MeV with an accuracy of 1-2%. The difficulties are (1) main- taining a very low detector threshold at a constant value, (2) maintaining appropriately thin targets for energies near the neutron threshold that are uniform, and (3) integrating the n-^ neutron time-of-flight peaks in a consistent and precise manner. These dif- ficulties may become real obstacles as the work is extended to lower energies. Conclusions Method of Establishment of the Primary Standard To establish the standard for neutron flux deter- mination using the Li(p,n-,Y) Be reaction, it has been shown in the previous section that a set of an- gular distributions is required for use with the as- sociated gamma-ray method. These angular distribu- tions may be relative with respect to angle, but each angular distribution must have the same normalization factor to the absolute value. Such a set of angular distributions must be established concurrently with the neutron detector's efficiency by measuring a large set of such angular distributions and demanding that the gross redundancy in measurements at the same or very nearby neutron energies be consistent with a single detector efficiency curve. The method is out- lined in detail by Brandenberger et al. 3 In prin- ciple the establishment of these angular distribu- tions is simple, in practice it is very time con- suming. Each point on all the angular distributions is normalized to a constant associated gamma-ray It appears that the present work is accurate to — 3% for energies above 600 keV. Such accuracy as indicated by the data cannot be accepted without equally rigorous substantiation by independent ob- servers. In any case, the method, per se , opens the possibility of measuring differential cross sections in the 0.1 to 4.1 MeV energy range to accuracies heretofore not attained, say 1.5 to 3%. Several other uses quickly come to mind. It should be pos- sible to: 1. further develop secondary standards such as the C(n,n)C reactions, 2. make more precise measurements of neutron-pro- ducing reactions, such as the T(p,n) He reaction, using a calibrated detector, 231 A0N3OIJJ3 3AI1VT38 Fig. The efficiency curve of the neutron detector used in Ref. 3. Each set of letters represents a 12 or 13 point angular distribution at a given E . 232 3. calibrate other detectors including counters of 3. J. D. Brandenberger , F. D. Snyder, J. D. Dawson, the McKibben long counter type, the moderating sphere T. W. Burrows, and F. D. McDaniel, Nucl. Instr. type, 3 He and 4 He counters, etc., as well as scintil- Methods 138 (1976) 321. lation counters operated in the time-of -flight mode. 4. S. A. Elbakr, I. J. Van Heerden, W. J. McDonald, References and G. C. Neilson, Nucl. Instr. and Methods 105 (1972) 519. 1. A. Mittler, Lowell University, private communi- cation, April 1973. 2. G. Presser and R. Bass, Nucl. Phys. A182 (1971) 321. 233 ASSOCIATED ACTIVITY METHOD K. K. Sekharan Cyclotron Institute Texas A&M University College Station, Texas 77843 A brief description of the associated activity method is given and its application for the calibration of flat response neutron detectors in three laboratories is described. The uncertainties involved in the associated activity method have been discussed and suggestion: for improving the accuracy of the efficiency for neutron detection have been made. (Description of Method; Possible Improvements; Uncertainties in Efficiency) Introduction A neutron has to be detected always by some nuclear process such as B(n,a) Li reaction or (n,p) scattering. As the nuclear processes are energy de- pendent the efficiency of the neutron detector for detection of neutrons also tend to be energy depen- dent. Hence the efficiency of a neutron detector has to be determined experimentally and/or by theoretical calculations as a function of the neutron energy. The "Associated Activity Method" is a precise experi- mental method for determining the efficiency of a neutron detector. A survey of the literature shows that this method has been employed for detector cali- bration in at least three laboratories. Recently, K. K. Sekharan, H. Laumer, F. Gabbard and B. D. Kern have determined the efficiency of a spherically shaped 4tt neutron detector using associated activity method. They have used 51 V(p,n) 51 Cr and 57 Fe(p,n) 57 Co reactions for the determination of the absolute effi- ciency of the detector, w. P. Poenitz has calculated 2 the efficiency of the grey neutron detector by cal- culating the intensity of the 2.2 MeV capture gamma rays. However, he has used the associated activity 3 technique to determine the relative efficiency of this detector as a function of neutron energy. Another group to use this technique to calibrate their long counter was J. M. Adams, A. T. Ferguson, 4 51 and C. D. Mckenzie. They have also used Cr and 57 Co for the activity measurement. Taschek and 5 Hemmendinger employed the associated activity me- thod to measure the total cross section of Li(p,n)Be reaction. Description of the Method There are many (p,n) reactions in which the re- sidual nucleus decays back, at least partly, to an excited state of the target nucleus by positron emis- sion or electron capture which results in the emis- sion of a gamma ray when the nucleus is de-excited to the ground state. Some of these residual nuclei de- cay with long half lives whereas others are short lived. As one residual nucleus is produced for e\/ery neutron produced during the (p,n) reaction the gamma rays associated with the decay of the residual nu- cleus have a definite relationship to the total num- ber of neutrons produced in the nuclear reaction. Hence it is possible to determine the efficiency of a neutron detector by counting the neutrons (with the detector whose efficiency is to be determined) for the entire length of time for which the target is bom- barded with the proton beam and then determining the gamma ray activity of the bombarded target. From the ratio of the neutron yield to the gamma ray yield the efficiency of the neutron detector can be determined. It turns out that the (p,n) reactions which can be conveniently used for the associated activity me- thod are 7 Li(p,n) 7 Be, 51 V(p,n) 51 Cr and 57 Fe(p,n) 57 Co although possibilities of employing some of the other (p,n) reactions cannot be ruled out. Available in- formation 6,7 about the residual nuclei of the above ( Be, Cr and Co) are tabulated in Table reactions 1. It can be observed from the table that the half lives of these residual nuclei are very long compared to the time required to activate the target by proton bombardment or the time required to measure the gam- ma ray activity. Appropriate corrections can always be applied for the finite time required for the proton bombardment of the target, the counting of the gamma rays from the residual nucleus and also for the time elapsed between proton bombardment and gamma ray counting. Table 1 Half lives of residual nuclei and the associated gamma ray energies. Residual Nucleus Half Life (days) Activation Time (hours) Associated Gamma Ray Energy (keV) 7 Be 51 Cr 57 Co 53 27.704±.002 270.9 +.6 2 3 to 4 6 477.4 319.8 121.94 136.31 However, one should remember that the charged pa accelerators are not a source of constant curren Therefore, the rate of build up of the radioacti sidual nucleus will not be constant and it is ad able to make the bombarding time as small as pos compared to the half life of the residual nucleu brief description of the techniques employed in three laboratories in calibrating the detectors given below. Li(p,n) Be reaction was used to determine relative efficiency of the 4tt neutron detector University of Kentucky in the neutron energy ran 30 keV to about 1600 keV. Neutron yield was mea from Li F targets at several energies using a fre target for proton bombardment lasting one to two at each energy. 478 keV gamma ray yield was mea rticle t. ve re- vis- sible s. A the is the at the ge sured sh hours sured 234 from each target immediately after the proton bombard- ment. For the absolute efficiency calibration a va- nadium metal target was bombarded with protons cor- responding to an average neutron energy of 300 keV. 57 Similarly, an isotopically enriched Fe target ( Fe ? 0o evaporated on to a carbon backing) was bom- barded with protons to produce neutrons of average energy 1350 keV. The gamma ray detector was cali- 51 57 brated using Cr and Co sources (standard sources supplied by the National Bureau of Standards) which enabled the determination of the absolute efficiency of the neutron detector at 300 and 1350 keV. The relative efficiency in the energy range 30 keV to 1600 keV was then normalized to the absolute effi- ciencies at 300 and 1350 keV to obtain the efficiency for the entire energy range. The efficiency of the detector as a function of the neutron energy is shown in Fig. 1. The efficiency of the detector for neu- trons form a Po-Be source was found to be 0.47%. Li(p,n) Be reaction cross section was measured with this detector and compared to the cross section pub- Q lished by R. L. Macklin and J. H. Gibbons. In Fig. 2, the dots are the present measurements and the dashed lines are cross section data published by Macklin and Gibbons. The detector is being used ex- tensively for measurement of (p, n) and (a, n) cross sections, some of which are important for astrophysi- cal calculations. 3 The grey neutron detector is applicable in the neutron energy range thermal to several MeV. Fast neutrons incident at the center of the detector, are slowed down in a moderating medium such as water or paraffin and a fraction of these neutrons are cap- tured in the hydrogen medium emitting 2.2 MeV gamma rays which are detected by a Nal (Tl) detector mounted at the surface of the moderating medium. As- suming that the detector is large enough to prevent leakage of neutrons from the moderating medium the efficiency of the detector as a function of neutron energy has been calculated taking into account the gamma ray absorption in the medium. The relative efficiency of the grey neutron detector was deter- mined by the associated activity method and compared to the calculated efficiency in the energy range 30 to 1000 keV. The 7 Li(p,n) 7 Be and the 51 V(p,n) 51 Cr I | i r r- Li (p.n) Be Reaction ■ i i i r 1 ' ' o 600 1 Error t 6% 8 400 ; i - cr 1 4 --rt • : *.•*** 200 '■,],,, i . . INCIDENT PROTON ENERGY (MeV) Fig. 2. Li(p, n) Be reaction cross section. Dashed lines are cross sections published by Macklin and Gibbons. reactions were the neutron sources for the calibra- tion. 7 51 The activities of Be and Cr in the irradiated targets were counted by measuring the yield in the photopeak of the 478 and 320 keV gamma rays 7 51 from Be and Cr respectively. The results are shown in Fig. 3. The performance of the detector was also checked by measuring the Li(p,n) Be cross sec- o tion. The grey neutron detector has been used for measurement of cross sections of neutron producing 9 10 reactions important for fast reactor calculations. ' The conventionally determined efficiency of the long counter was compared to the efficiency determined by the associated activity method by J. M. Adams, 4 A. T. G. Ferguson and C. D. McKenzie in the neutron energy range 50 to 1300 keV. They measured the angu- 51 51 lar distribution of the neutrons from V(p,n) Cr 57 \57 and Fe(p,n) Co reactions and compared the neutron yields from these reactions to the activities of the bombarded targets. A Nal (Tl) detector was used to measure the yield in the photopeak of the 320 keV 57 gamma ray from Fe. They observed deviations in the efficiency of the long counter from that determined by the conventional method. AVERAGE NEUTRON ENERGY (kiV) JO 60 IJJ 2«5 320 430 620 910 993 1190 1360 1 1 1 1 1 1 1 1 1 1 SPHERE COUNTER ■ Atno'ute Mtoiu 1 • Reiai>«« M»i« ■™»" - " - -i— ]-! «'I I ' | i i i 1 - " _ EMicltncr lor Pv-8« NcutrOflt ■ -ATV. " - " i i iii 1 " INCIDENT PROTON ENERGY i r ASSOCIATED ACTIVITY TECHNIQUE o v 5l (p,n) Cr 51 a Li 7 (p,n)8e 7 FLUX INTEGRATION TECHNIQUE • H'ln.ylH 2 / V 5l (n,y) V 52 r~NOR NORMALIZATION POINT J I L 200 400 600 BOO 1000 1200 1100 NEUTRON ENERGY. keV Fig. 1. The absolute efficiency of the 4tt neutron detector. The error flags are an estimate of the standard error at each point. Fig. 3. The ratio of the experimental efficiency to the theoretically calculated efficiency of the grey neutron detector. 235 Uncertainties in the Determination of the Efficiency of the DetectdF The important factors in the determination of the efficiency of the detector using the associ- ated activity method are (1) the knowledge of the half life of the radioactive target whose activity is to be measured after bombarding the target with charged particles and (2) the knowledge of the strength of the sources used to calibrate the gamma ray detector. The beauty of the associated activity method is that the accuracy of the efficiency deter- mination is independent of the target uniformity and errors in the target thickness determination which are, in general, two difficult factors in efficiency calibration. Even the absolute efficiency of the gamma ray detector need not be known. One needs to know only the relative efficiency of the particular geometry in which gamma ray counting from the bom- barded targets and the standard sources are done. However, one should exercise great caution to avoid uncertainties arising from geometrical factors in the placement of bombarded targets and calibrated gamma ray sources. The easiest method to determine the absolute number of gamma rays from the irradiated targets is to compare the gamma ray yield from the target to that of the standard sources of the same radioactive element. As several days may be elapsed between the time of the determination of the strength of the standard sources and the determination of the rela- tive efficiency of the gamma ray detector used for gamma ray activity measurement, the half lives of the standard sources must be known as accurately as pos- sible. The half lives of 7 Be, 51 Cr and 57 Co have been reported extensively in the literature. ' Various methods have been used to determine the half lives of these nuclei. Any improvement in the ac- curacy of the values of the half lives of these or similar radioactive nuclei will be useful for improving the accuracy of the efficiency of the neu- tron detector. Hence it is necessary to obtain con- sistent values for the half lives of those radio- active nuclei from several different methods. Improvements in the accuracy of the knowledge of the strength of the gamma ray sources employed for the calibration work will also reduce the overall un- certainty in the efficiency of the neutron detector. The effect of strongly anisotropic angular dis- tribution of neutrons on the efficiency of the flat 1 3 response 4tt neutron detector has been studied ' and found to be less than 1% in the case of the strongly forward peaked resonance at 2.28 MeV in the Li(p, n) Be reaction. However, for the development of a neutron detector whose efficiency is known to an accuracy of about 1%, the effect of the angular dis- tribution on the detector should not be neglected. Possible Improvements in the Efficiency Determination The efficiency of the above mentioned detectors as a function of the neutron energy is known to about 1300 keV. It will be useful to determine the effi- ciency of the detectors for somewhat higher energies though complications may arise due to flux leakage from the moderating medium. A Monte Carlo type cal- culation can be used for this purpose. In fact, W. P. Poenitz has calculated the efficiency of a 12 black neutron detector using Monte Carlo method. A practical method for extending the energy range over which the efficiency of the 4ir neutron detector is known, will be to compare the calculated efficiency to the experimentally determined effi- ciency in the energy range 30 to 1300 keV and then extend the calculation to higher energies. Good agreement between the calculated and experimental efficiencies in the energy region 30 to 1300 keV will be adequate justification for using calculated effi- ciencies at higher energies. The Monte Carlo method can also be used to investigate the effect of strongly anisotropic angular distributions of a nuclear reaction on the efficiency of the 4tt detector. Thus better knowledge of the half lives, improvements in the uncertainty of the source strength of standard sources used in associated activity method and theo- retical calculation of the neutron efficiency are ma- jor factors which will improve the accuracy of the neutron detector efficiency. References K. K. Sekharan, H. Laumer and F. Gabbard, Proc. Int. Conf. Nuc. Cross Sec. Tech., NBS SP 425, 108(1975); K. K. Sekharan, Ph.D. Dissertation, Unpublished; K. K. Sekharan, H. Laumer, B. D. Kern and F. Gabbard, Nucl . Instr. and Meth., 133, 253(1976). "W. P. Poenitz, Nucl. Instr. and Meth., 58, 39(1968). 3 W. P. Poenitz, Nucl. Instr. and Meth., 72, 120(1969) J. M. Adams, A. T. G. Ferguson and C. D. Mckenzie, Activation Technique for the Absolute Calibration of a Long Counter, A. E. R. E.- R. 6429(1970) . 5 E. F. Taschek and A. Hemmendinger, Phys. Rev. 74 , 373(1948). Nuclear Decay Data for Selected Radionuclides, ed. M. J. Martin, 0RNL - 5114(1976). E. De Roost and F. Lagoutine, Atomic Energy Review, 11, 642(1973). 8 R. L. Macklin and J. H. Gibbons, Phys. Rev., 114, 571 (1959). g H. 0. Menlove and W. P. Poenitz. Nucl. Sci. Eng., 33 , 24(1964). 10 W. P. Poenitz, J. Nucl. Energy, 22, 505(1968). J. H. Gibbons and H. W. Newson, Fast Neutron Physics Part I, ed. J. B. Marion and J. L. Fowler, Inter- science Publishers Inc., New York, 164(1960). 12 W. P. Poenitz, Nucl. Instr. and Meth., 109, 413 (1973). 236 ACCURACIES AND CORRECTIONS IN NEUTRON BATH TECHNIQUES E.J. Axton National Physical Laboratory Teddington, Middlesex, United Kingdom The corrections and currently attainable accuracy in the neutron source emission rate meas- urements with the manganese sulphate bath technique are discussed in detail, followed by a review of other bath techniques and their adaptation for neutron flux measurement. 1 . Introduction Over the last 25 years much has been written about the use of bath techniques for neutron measure- ments and it would be impossible to review all this work in the space of 15 minutes. Probably of greatest interest is the use of manganese sulphate baths to measure neutron source strengths, particularly with respect to fission yield measurements. Therefore, in spite of the title of this session, it is proposed to begin by assessing the ultimate accuracy attainable with an ideal system and if there is time left to discuss the limitations of practical systems, and some other applications of the bath technique. 2. Design Figure 1 is a schematic diagram of a typical system. This shows some of the features which are essential to the attainment of high accuracy. Some of the design features are necessarily compromises. For example the bath has to be large to reduce neutron escape but the specific activity goes down with the cube of the radius. The presence of the central cavity increases the neutron escape but reduces the capture of neutrons returning to the source. Spherical geometry maximises the specific activity whilst mini- mising the escape. It also eases the problems of calculation and measurement of the corrections. The bath should not be shielded unnecessarily, but if it is shielded provision should be made for the insertion of detectors to assess the neutron escape. Provision should also be made to measure the thermal neutron flux at the cavity boundary. The solution should have a high concentration to minimize the effect of uncer- tainties in the cross section ratios, and should be circulated through the activation detectors so that they can be operated whilst the source is in the bath in order to improve the counting statistics. Two independent detection systems should be employed so that a change or failure in one of them is instantly detected. A further advantage is that twice as much data is acquired. The bath detectors are calibrated by the insertion of calibrated samples of 5%n which have been measured by the 4^3 y coincidence method. A very stable high pressure ionisation chamber system is essential to check the consistency of the coincidence counting. Standard sources in fixed geometry should be used regularly to check the stability of the bath detectors. The concentration of the solution should be measured regularly by at least two methods. Two bath sizes should be used to check the validity of the evaluated corrections. Finally a standard neutron source should be measured periodically to check the whole system. 3. Source strength determination The source strength Q is derived from the fol- lowing equation: A The fraction f of neutrons captured by manganese is NH 1 + f E (1 - 0)(1 - S)(1 - L) "W 1 + G r s) Wln ff Mn (1 + G r S) Where A is the saturation manganese counting rate measured by the bath detector system with efficiency E, is the fraction of neutrons which are captured in the n,a and n,p reactions in sulphur and oxygen, S is the fraction of neutrons recaptured by the source, and L is the fractional neutron escape from the boundaries of the bath. 0jj, o"g and O"^ are -the thermal neutron capture cross sections of hydrogen, sulphur, and manganese respectively, and NH/NMn is the hydrogen to manganese atom ratio in the solution, s is the normalised above l/v resonance activation integral of manganese, r is the epithermal flux par- ameter averaged over the bath, and G is the resonance self shielding factor. In the following sections the components of this equation will be examined in detail. 3. 1 Measurement of bath efficiency E The difficulties associated with this apparently simple operation are almost invariably underestimated as there are so many things which can go wrong. Basically a quantity of radioactive 5°Mn of high purity is divided into portions by weight, some of which are used to activate the baths and others are measured in the 4it3y coincidence equipment to determine the specific activity. Differences in the specific activity of these solutions can be caused by sticking of activity to glassware and by evaporation of water during the weighing of small liquid samples either in the bath loading or counting operations. In absolute measurement of 5°Mh solutions, an international compari- son revealed differences in apparent activity depending on whether a liquid scintillator or gas flow pro- portional counter is used for the 3 channel. In the proportional counter the wrong result can be obtained if the plastic source mount becomes non-conducting. If a liquid scintillator is used it is necessary to check experimentally that no significant after pulsing occurs. Both methods require a correction for the fact that the 3 counter is not 100$ efficient (the decay scheme correction). In the case of the liquid scintillator this correction is larger and more difficult to derive. Unsuspected radioactive contamination can occur in which case the measurement will be time dependent. Inevitably the time will come when the bath efficiency will appear to have changed. This could be due to a fault in the bath counting or in the coincidence count- ing, or in the dispensation of the sample or its purity, or to a simple mistake. A third party is then required to arbitrate. This is the role of the high pressure ionisation chamber referred to in section 2. The cali- bration figure for the ion chamber based on the coinci- dence equipment should remain constant, a typical standard deviation for a large number of calibrations being about ± 0.15$. Only if this reproducibility is 237 Small spherical bath Neutron escape monitor Large spherical bath Pump Spherical cavity Stirrer Cavity flux monitor Dual detector system Fig. 1. Schematic arrangement for a typical bath system. achieved, and the resultant calibration figure is in agreement with that obtained using a standard from an appropriate national laboratory can there he confidence in the measurements, and their relationship to those of other laboratories throughout the world. For the actual bath efficiency measurements, if precautions are taken to prevent loss of activity on the way e.g. along the walls of entrance pipes, or by splashing, and no electronic drifts or faults have occurred a typical standard deviation of ± 0.3% can be obtained and sufficient repetitions of the calibration may be car- ried out to reduce the standard error of the mean to an acceptable level. Ultimately one is left with an additional systematic uncertainty ~ ±0.2% for the hath calibration. This is the current level of agree- ment obtained by national laboratories in international comparisons, de Volpi^ ^ ' and is attributable in part to the difficulties of weighing small quantities of liquids. 3.2 The H/Mn atom ratio This may be determined by three methods. (1) The volumetric method by titration with EDTA. (2) The gravimetric method by drying and weighing weighed samples of solution hoth as the mono- hydrate after drying at 100 °C and as the dehydrate after drying at 300 °C. (3) Determination of solution density. The third method is not absolute but has to be calibrated by one of the first two, hoth of which are capahle of reproducibility ~ 0.05% standard deviation. In the absence of impurities the two methods should give identical results. In the presence of impurities the volumetric method should give the correct specific manganese concentration provided there are no other titratable ions present, whilst the difference between the results of the two methods gives a guide to the impurity level. Needless to say, impurities should he avoided wherever possible. The H/Mn ratio can he determined sufficiently accurately to make a negligible contribution to the overall uncertainty. 3.3 The manganese to hydrogen thermal neutron cross section ratio This is the dominant factor in the computation of the fraction of neutrons captured by manganese. The ratio of the individual cross sections taken from Mughabghab and GarberC 2 ) is 0.02496 ± 1.62%. However the ratio can he determined more acurately by variation of the concentration of the solution. Equation (1) is rearranged as follows to form a straight line equation. NH E (1 - 0)(1 - S)(1 - L) i + 1 0~ Mn Q (1 + G r s) NMn 1 + Mn (1 + G r s) Measurements are made for a variety of values of NH/NMn from say 30 to 300 and the source strength is derived from the intercept of the line and the cross section ratio from the ratio of the slope to the inter- cept. All the quantities on the left hand side are concentration dependent and therefore have to be 238 evaluated for each concentration. By suitable choice of source energy and the bath size can be made zero and L very small. Nevertheless L can vary by a factor of 2 over the range of concentration used so provision must be made to measure and calculate L at each point if it is significant. A typical uncer- tainty in the cross section ratio which can be obtained by this method would be ± 0.3$ to 0.4$. The NPL value is 0.02495 ± 0.34$ standard errorU). A measurement in progress^) indicates agreement between this value and the BNL 325 3rd edition value. A further value of 0.02531 ± 0.12$ has been published, but this is qualified as an effective ratio for the particular bath and solution used to measure itV5j. The sign of the change of efficiency with concentration can be positive or negative depending on the type and mode of operation of the bath counters. In principle a different bath and counting system could be used for each point provided the appropriate corrections were made. Thus if the same source were used, all avail- able data could be analysed in one global fit. 3.4 The effect of the impurities The effects of the impurities are rather complex. Impurities which absorb neutrons and produce measurable activity would introduce a time dependent bias to the result which would be very difficult to interpret. Assuming that no measurable activity is produced, the linearity of the fit should be unaffected if the impurity concentration is proportional either to that of the Mn or to that of the H. Neglect of an impurity in the supply of the solid MnSO* would pro- duce a low value for the H/Mn cross secxion ratio, but if the atom ratio was determined by a gravimetric method it too would be too low and this would tend to cancel the error in the cross section ratio. However, an impurity in the water would produce the correct source strength but a high value of the cross section ratio. The latter situation is unlikely in practice because the initial concentrated solution is probably made up from Mn SO* 4H2O, in other words 25$ of the water is already present, and an impurity in the added water would produce a non— linear situation. 4. Corrections 4. 1 Fast neutron capture in oxygen and sulphur This correction is both concentration and spectrum dependant. Its evaluation is almost entirely depend- ant on calculations. The oxygen loss in water for a-,., Ra Be source was measured by de Trover and Tavernier as (2.25 ± 0.3)$. Ryves and Hardem ') using the same type of source obtained (1.69 ± 0.25)$ in water and (3.O5 ± 0.3)$ for the oxygen and sulphur loss in concen- trated manganese sulphate solution. Agreement is obtained between the NPL manganese bath and the AERE boron pile(°)9) if a correction of (3.0 ± 0.5)$ is applied to the bath measurements. This information could be interpreted as another measurement of the correction. Unfortunately the Ra Be source is not a very good source for this experiment from the point of view of comparison with calculations as one is dependent on assumptions regarding the intensity of the low energy component in the spectrum, which is fairly large and known to vary from source to source. Finally one should mention the experimental method developed by de VolpiwJ whereby the correction is derived from the change in slope of the line obtained in a dilution experiment. Unfortunately the method is not very sensitive. Nearly 80$ of the oxygen effect in a con- centrated solution is due to the oxygen in water. Thus the change in the slope registers only the sulphur effect plus 20$ of the oxygen effect, which varies from about 1.4$ for AmBe source down to 0.35$ for a californium spectrum. Thus a change in the slope of only 1 to 4 times the uncertainty in the slope can be expected. Calculations have been published for a number of source spectra by Ryves and Hardenw) and by LouwrierV 10J based on standard slowing down theory and by Murpheyv ^ ' J using Monte Carlo techniques. In the case of Ryves and Harden, and Murphey the oxygen cross sections used are now thought to be too high particu- larly in the energy range above 7 MEV. Likewise the oxygen cross sections used by Louwrier are thought to be too low. Table 1 shows a comparison of calcu- lations based on both methods for a number of source spectra using cross sections obtained from the data centre at Saclay at the end of 1972 and are believed to be up to date. The calculations are for concentrated manganese sulphate solution with NH/NMn = 30. Table 1 Oxygen and Sulphur correction ($) Source Mean E Monte Carlo Slowing down theory Am Be 4.46 3.09 3.17 Ra Be 3.94 2.20 2.45 Cf-252 2.09 (ET=i.39) a 0.53 0.73 Cf-252 2.15 (ET=1.43) 0.62 0.79 Am B 2.76 0.49 O.64 a Maxwellian temperature in MeV. The agreement seems to become progressively worse as the neutron energy decreases. The only other cal- culations known to have been performed with up to date cross sections are those of Ullo and Goldsmithl 1 2, 13 ) in their re-appraisal of the n measurements. Table 2 compares their results with recent calculations at NPL. Table 2 Axton Ullo Macklin Steen S 0+S s 0+S 0+S 0+S Smith et al. NH/NMn= 77.7 0.30 0.09 0.39 0.19 0.03 0.22 Macklin et al. NH/NMn= 192.6 0.24 0.04 0.28 0.25 So it appears that even calculations by the same technique, using the same cross section data do not agree. Whatever cross section data is used the oxygen effect should follow the 0/H atom ratio, and the sulphur effect should follow the S/H atom ratio. The figures on the right of the table do not appear to do this. There is no way of assessing the absolute val- idity of these results other than by agreement between different calculations. These programs are necess- arily complex and it seems almost impossible to be certain that no hidden faults exist in them. It would be a good idea to hold an international comparison of Monte Carlo calculations to resolve the point. 4. 2 Neutron Escape As with the oxygen and sulphur loss correction the escape correction is dependent on the concentration of the solution and on the source spectrum. In addition it depends on the size and geometry of both the bath and the central cavity. It is also varied by the 239 A M e ^6D to presence of shielding materials. There are so many- variables that it is difficult to produce a comprehen- sive correction set which would suit all occasions, and it is equally difficult to compare corrections which have been evaluated for different systems, authors have used expressions of the type evaluate their escape corrections using constants derived for other systems by other authors. Whilst such expressions give a good estimate of the order of magnitude of the correction, they are not transferable when the geometry of the bath or the central cavity are changed, or if the concentration of the solution or the shielding is changed. On the other hand the correct- ion may be checked by measurement. In principle the escape correction can be derived from measurements in different bath sizes and extrapolated to infinite radiusOU This is not very practicable, but the use of two bath sizes provides a valuable bench mark for testing measurement and calculation procedures. For day to day monitoring at NPL a flat response detector (long counter) is used to monitor the neutron flux at the surface of the bath, which is then integrated over the bath surface. The counter efficiency is chosen empirically for each source spectrum to give agreement between the two bath sizes. Although there are many reasons why the long counter would appear to be unsuit- able for this purpose it seems to give satisfactory results for the two baths. Table 3 compares these measurements with Monte Carlo calculations at NPL which are only in a preliminary state. The fast neutron program records fast neutron escape and records the position of neutrons which become thermalised. Shell sources of thermal neutrons so produced can be used to determine thermal escape either by diffusion calcu- lations (M/C + Diff) or by a Monte Carlo program (M/C). The programs are still being developed so that the results must be regarded as provisional. Agree- ment becomes progressively worse as the source energy decreases, and the thermal fraction increases. Part of the problem is due to the uncertainty in the source spectrum. Some Am B and Am Li sources are believed to be contaminated with Be. At the present state of the art, a systematic uncertainty of ± 10% of the correction for an Am Be source and 20% for a Californium source seems reasonable. Table 3 Neutron escape from an unshielded sphere with a 8.9 cm. dia. cavity and NH/NMn = 30 Table 4 Comparison of neutron escape calculations with Ullo and Goldsmith evaluations 98 cm. tank Total leakage (%) Thermal fraction Meas. M/C M/C + Diff. Meas. M/C M/C + Diff. 1.46 1.49 1.41 17 18 13 Am Be Ra Be 1.15 1.14 1.07 9 15 10 Am B 0.33 0.33 0.34 35 28 19 Cf (ET=1.39) 0.30 O.32 0.30 23 22 17 Cf (ET=1.43) O.42 O.38 20 12 Am P 0.05 0.05 0.05 19 25 cm. tank 23.2 23.4 21.2 21 12 20 Am Be Ra Be 19.7 18.8 15.8 22 22 13 Am h 14.6 29 Cf (ET=1.39) 10.0 12.1 10.0 32 34 20 Smith et al bath Macklin Total escape M/C M/C + Diff. Ullo 0.16 0.28 0.13 0.21 0.23 ± 0.11 0.26 ± 0.03 4.3 Source capture of moderated neutrons This correction is both concentration and source spectrum dependent. In all cases it is necessary to know accurately the amount and geometry of each material in the source assembly .and the appropriate reaction cross sections. There are two approaches to this problem. One can derive the cavity boundary thermal and epithermal neutron flux in terms of the Wescott parameters either by measurement with foils or by diffusion calculations*. 14J or by Monte Carlo calcu- lations. It is then necessary to evaluate an effec- tive cross section for the source assembly with appropriate flux depression and self shielding corrections for each region of the source assembly. Spiegel and Murphey^^J describe a method of treating the problem. Alternately sufficient detail can be programmed into the Monte Carlo calculation to repro- duce the source interactions directly. In either case a convenient test of the calculation is the ability to determine the cavity boundary flux, which can be measured to about ± 2% with foils. Table 5 compares the result of Monte Carlo and diffusion calculations with sub-cadmium thermal neutron flux measurements by Bardell^"/ using gold foils. Again the agreement gradually worsens as the energy decreases. The low value for Am B is due to flux depression in the absence of which these results would lie on a smooth curve as a function of neutron energy. Table 5 Thermal neutron flux at cavity boundary for unit source (fo) Measured with gold foils Monte Carlo/ Diffusion calculations Am Be 0.139 0.134 Ra Be 0.180 0.178 Am B 0.140 0.128 Cf (ET=1.39) 0.223 0.218 Cf (ET=1.43) 0.223 0.214 Am F 0.249 0.208 Smith et al HH/NMn=77 ET= 1.323 0.175 Macklin et al NH/NMn=192 ET= 1.323 0.204 240 4.4 The manganese resonance correction This correction can he obtained hy deriving an effective cross section for manganese in terms of the Wescott^^' parameters G, r f and s. Calculations "by this method have been published by Axton and Ryves' 1 K Alternately one could enter the detailed manganese resonance data directly into a Monte Carlo calculation . Both methods depend on knowledge of the manga- nese activation integral. If the resonance parameters are put into a Monte Carlo calculation it is necessary to choose the y width so as to give the preferred resonance integral. The value given in BNL 325 edition 3 is 8 barns. Recent measurements at NPL in terms of the gold resonance integral^'? ' indicate a value of about 1.6 barns. There are some much higher values elsewhere in the literature. The derived source strength is not very sensitive to the precise value of the integral. With concentrated solutions (H/Mn =30) a change of 10$ in the integral would alter the source strength by about 0.07$. Table 6 gives examples of the magnitude of the correction. The difference between the Axton, and Ullo and Goldsmith values is almost entirely accounted for by the difference in the value of the resonance integral used. Ullo and Goldsmith used 9«15 barn. The details of their calculation are not given in their paper. Table 6 Manganese resonance corrections Table 7 Typical estimates of uncertainties NPL Smith et al Macklin et al NH/NMn NPL Ullo and Goldsmith 30 77.7 192.6 1.0071 1.0083 1.0093 1.0100 1.0111 4.5 Photoneutron production In some cases a small correction is necessary to allow for photoneutron production from deuterium in the solution. This correction is usually negligible with light water baths except in the case of high energy high intensity y emitting sources such as Na Be. 5. Uncertainties Provided all the necessary precautions are taken and all the checks* are successful, typical uncer- tainties would be estimated as shown in Table 7. Uncertainties are evaluated for two concentrations and two source spectra. The figures shown are for a 98 cm. diameter bath with an 8.9 cm. diameter cavity. The random uncertainties are quoted at the 68$ confidence level. Five measurements of the bath efficiency, and five irradiations of the bath by the source are assumed. The existence of the second counting channel, in addition to providing internal consistency checks, also doubles the amount of data for analysis. It is assumed in all cases that the source is sufficiently strong to provide adequate counting statistical accuracy and that the background uncertainty is negli- gible even for the dilute solution. *see Appendix Systematic Uncertainties ($) NH/NMn=30 Am Be Cf NH/NMn Am Be = 192 Cf Bath efficiency 0.2 0.2 0.2 0.2 Cross section ratios V°Mn 0.145 0.145 0.287 0.287 ^Afa 0.122 0.122 0.039 0.039 Manganese resonance 0.07 0.07 0.09 0.09 Oxygen and sulphur 0.3 0.1 0.18 0.060 Leakage 0.15 0.03 0.20 0.04 Source capture 0.015 0.020 0.02 0.03 Random uncertainties 0. 13 0.13 Bath efficiency Source activation of bath 0. 10 0.10 6. Other baths used for source strength measurements 6.1 Noyce et al\ ®' introduced heavy water into their manganese bath in order to reduce the cross section dependence. It was necessary to introduce a pro- portion of light water into the mixture to prevent excessive neutron escape. In addition to the usual corrections, a correction for the photoneutron pro- duction in the bath was necessary. To minimise this correction an intermediary Sb-Be source was cali- brated instead of the NBS standard Ra Be photoneutron source. In a subsequent comparison measurement ± 1$ accuracy was achieved for the standard source. 6.2 de Troyer and Tavernier\6) Jaritsina et al(2l) ) Van der Eijkv22J g^ MichikawaV23; used a water bath in which the neutron density as a function of distance from the source was integrated numerically. The neutron density at each point was measured by gold foil activation. The necessary corrections are similar to those described for the manganese bath. The method is time consuming and less amenable to automation but pro- vides a valuable check on MnSO* bath calibrations. In principle it might be possible to develop the method to the same accuracy. Present uncertainty in the gold/hydrogen cross section ratio is ± 0.7$. 6.3 Fieldhouse et alv 2 4j used a cylindrical bath in which the neutron density as a function of distance was determined from BF-> counter measurements. The system was calibrated by associated particle counting of a particles from the T(dn) reaction. The experiment did not achieve its aimed accuracy of ± 1$. 6.4 Smith et al( 2 5) and Macklin et al( 26 ) used cylindrical manganese sulphate baths for their tj measurements. In these experiments only ratio's of source strength are involved so none of the uncer- tainties associated with the absolute calibration of the bath detectors are relevant. On the other hand the problem of calculating the correction for neutron 241 absorption and multiplication in the complex source assembly are correspondingly greater and it seems diffi- cult to believe that these effects can be calculated to the accuracy quoted. For example in the Goldsmith and Ullov - ^} re-evaluation of the Oak Ridge measurement, the correction quoted for cadmium absorption is (1.85 ± 0.04)$. 7. The use of baths with accelerators 7.1 Scott\27J used a spherical manganese sulphate bath to determine the thick target yield from the 'Li(p,n) reaction, the source in this case being the accelerator target situated at the centre of the bath. In addition to the usual corrections it was necessary to allow, in the continuous flow detector system, for time dependent fluctuations in the neutron emission rate. Overall accuracy of 2-3$ was achieved. 7.2 Poenitz'^".) used a spherical manganese sulphate bath to measure the neutron flux at zero degrees at energies close to the threshold of the endothermic reactions 'Li(p,n) and T(p,n). By careful control of the accelerator energy the neutron production was limited to a forward cone of half-angle 9 degrees, thereby removing the need to collimate the bath with bulk shielding. The largest contribution to the uncertainty in the neutron flux was ± 1.3$ due to neutron escape. 7.3 The use of 3.8 minute. -^V speeds up the measure- ment cycle considerably compared with 155 minute 5°m However, the short half-life introduces problems in the absolute measurement of the activity and its comparison with other laboratories. Furthermore, the uncertainty in the V/H cross section ratio of ± 1% enters into the source evaluation with greater weight as the V cross section is only 5 barns. For these reasons, if an absolute system is required it is preferable to calibrate the system with standard neutron sources. 7.4 Robertson et al\^9/ used a collimated vanadyl sulphate bath to measure the relative zero degree neutron yield from the 'Li(p,n) reaction. It was necessary to shield the bath from room scattered neutrons and to admit the direct neutrons into the bath through a collimator. An important feature of the experiment was the demonstration by Monte Carlo calcu- lation supported by inverse square measurements that the collimator could be taken as having, for each energy, a fixed entrance aperture at a fixed position over the energy range considered. The system was used to measure relative neutron flux with a standard deviation of ± 2$. 7.5 Leroy et al\^ / used a small shielded and colli- mated manganese sulphate bath to measure the neutron flux from an accelerator in the energy range from 10 keV to 1 MeV. The bath detector system was calibrated by means of a Ra Be neutron source which had been calibrated by BIPM. Neutron capture in the neutron source assembly and neutron escape into the reflector were obtained by Monte Carlo calculations as was the effective solid angle subtended by the bath aperture. An additional correction was necessary for air attenuation of the neutrons and the overall precision was reported to be + 1.8$. Above 1 MeV the neutron escape fraction became unacceptably large. 8. Conclusion With sufficient care, the manganese sulphate bath can be used to achieve an uncertainty estimated at + 0.35$ for a Californium neutron source at the 68$ confidence level or + 0.7$ at the 99$ confidence level. Adapted for the measurement of neutron flux density, typical uncertainties of + 1.8-2.0$ at the 68$ confidence level have been achieved with bath detectors. 9. References 1. de Volpi, A., 1973, Metrologia, 6, 65. 2. Mubhabghab, S.L. and Garber, D.I., 1973, BNL 325, Edition 3. 3. Axton, E.J., Cross, P., and Robertson, J.C., 1965, J. Nucl. Energy, _]£, 409. 4. Smith, J.R., 1977. Private communication. 5. de Volpi, A. and Porges, K.G., 1969, Metrologia, 5_, 128. 6. de Troyer, A. and Tavernier, G.C., 1954 t Acad. Roy. Belg. Bull. Class. Sci. , 4_0, 150. 7. Ryves, T.B. and Harden, D. , 1965, J. Nucl. Energy a/b, 12, 607. 8. Colvin, D.W. and Sowerby, M.G. , 1965, Physics and Chemistry of Fission, (Proc. Symp. Salzburg, 1965), 2, 25. IAEA Vienna. 9. Colvin, D.W. , Sowerby, M.G. and Macdonald, R.I., 1967, Nuclear Data for Reactors, (Proc. Symp. Paris, 1966), ]_, 307. IAEA Vienna. 10. Louwrier, P.F. , 1964, PhD Thesis, Institute for Nuclear Physics Research (IK0) Rototype, Amsterdam. 11. Murphey, W.M. , 1965, Nucl. Inst. Meth. , 37., 13. 12. Ullo, J.J. and Goldsmith, M. , 1976, Nucl. Sci. Eng., 60, 239. 13. Goldsmith, M. and Ullo, J.J., 1976, Nucl. Sci. Eng., 60, 251. 14. Ryves, T.B. , 1964, J. Nucl. Energy A/B, J_8, 49. 15. Spiegel, V. Jr. and Murphey, W.M. , 1971, Metrologia, 7, 34. 16. Bardell, A.G., 1977, to be published. 17. Wescott, C.H. , Walker, W.H. and Alexander, T.K. , 1958, Proc. 2nd. Conf. The Peaceful Uses of Atomic Energy, Geneva. 15/P/202 United Nations N.Y. 18. Axton, E.J. and Ryves, T.B. , 1967, J. Nucl. Energy, 21_, 543. 19. Ryves, T.B. and Axton, E.J., to be published. 20. Noyce, R.H. , Mosburg, E.R. Jr., Garfinkel, S.B. and Caswell, R.S., 1963, J. Nucl. Energy A/B, 11, 313. 21. Jaratsina, I. A., Constantinos, A. A. and Forminikh, V.I., 1964, Energie Atomique, 1_6, 253. Jaratsina, I. A., Andreev, O.L. , Katchine, A.E. and Stukov, S.M. , 1964, Energie Atomique, _l6. t 2 55. 22. Van der Eijk, see Naggiar, V., 1967, Metrologia, 3, 51. 242 23. Michikawa, T. , Teranishi, E., Tomimasu, T. and Inoue, Y, 1959» Bull. Electrotechnical Laboratory, Japan, 23, 223. 24. Pieldhouse, P., Culliford, E.R. and Mather, D.S. , 1967, J. Nucl. Energy, 2J_, 131. 25. Smith, J.R. , Reeder, S.D. and Pluharty, R.G., 1966, IDO 17083, Phillips Petroleum. 26. Macklin, R.L. , de Saussure, G. , Kington, J.D. and Lyon, W.S., 196O, Nucl. Sci. Eng. , 8, 210. 27. Scott, M.C. and Ashraf Atta, M. , 1973, Nucl. Inst. Meth. , 106 , 465. 28. Poenitz, W. , 1966, J. Nucl. Energy, 20, 825. 29. Robertson, J.C., Ryves, T.B. and Hunt, J.B. , 1972, J. Nucl. Energy, 26, 507. 30. Leroy, J.L., Huet, J.L. and Gentil, J., Nucl. Inst. Meth., 88, 1. 31. Davy, D.R., 1966, J. Nucl. Energy, 20, 277. Appendix The following internal checks give valuable insight into the behaviour of the system and can save time being wasted on abortive irradiations. If any check fails the cause should be ascertained and rectified. 1. The product of bath efficiency (E) and total volume of bath and detector system should be constant for all bath sizes. 2. Fixed geometry y source checks on the bath detectors should yield a constant count rate for each channel. 3. Results should be processed for both growth (source in bath) measurements and decay (source removed) measurements, growth/decay ratios should be constant and equal to unity for both channels. Channel 1/channel 2 ratios should be the same for growth, decay, and efficiency measurements. Deviations in the early stages of growth or near the changeover point immediately reveal faults due to change in pumping speed, timing errors or inadequate mixing. 243 INTERNATIONAL COMPARISON OF FLUX DENSITY MEASUREMENTS FOR MONOENERGETIC FAST NEUTRONS V.D. Huynh Bureau International des Poids et Mesures Pavilion de Breteuil, F-923 1 SEVRES, France An international flux density intercomparison of fast neutrons has been organized by the Bureau International des Poids et Mesures during the past three years. Nine laboratories have carried out measurements. Three neutron energies were selected (250 keV, 2.5 MeV and 14.8 MeV), to which two optional energies were added (565 keV and 2.2 MeV). A polyethylene sphere with a small BFo counter at the center was used as a transfer instrument at all energies except for 14.8 MeV. A He proportional counter was used at the two lower energies as the second transfer instrument. At 14.8 MeV a fission chamber ( U) and the iron foil activation method using - ) °Fe(n,p)- ) °Mn reaction were used. The results concerning the sensitivity of each transfer instrument measured by all the participating laboratories are summarized. (International comparison; fast neutron flux density) Introduction At the request of Section III (Mesures neutroniques) of the Comite Consultatif pour les Etalons de Mesure des Rayonnements lonisants, the Bureau International des Poids et Mesures decided in 1972 to orqanize an inter- _ .... .,, . , ... . c c , , 3 Conditions or the international comparison of Mux density measurements Table 1 (conditions or rne inrcmurionui curnpu i iiun or riux aensiry measurements national comparison of flux density measurements for f or monoenergetic fast neutrons (neutron energies, transfer instruments monoenergetic fast neutrons. Nine laboratories have and participating laboratories) carried out measurements. Table 1 lists the neutron i i r • ill Neutron energies chosen, the transfer instruments used and the energies Transfer instruments Participating laboratories* participating laboratories with their responsible physi- cists. V.D. Huynh of the BIPM took part in the measu- , polyethylene sphere + BF 3 fAl C 'ptb N nb B S CMN ' NPL ' rements performed in all the participating laboratories 2 50 keV ) in order to control the good functioning of the transfer 3,, NRC, CEN, NPL, ETL I II r I He COunter otd MIC instruments and to ensure the homogeneity ot the com- Mts < p*"" panSOn * 565 keV l polyethylene sphere + BF ) NRC, CEN, NPL, ETL, (optional) 3 PTB, NBS Three neutron energies were selected, namely ' counter 250 keV, 2.50 MeV and 14.8 MeV, to which two op- ^°^ polyethylene sphere + BF., CEN, BCMN, NPL, PTB tional energies (565 keV and 2.20 MeV) were added . , ., uu yci vie g iuiiqiv ' ur- (optional) v i l r 3 -> ' ' (K.W.GeigerandL.VanderZwan) Am-Be source which Can be introduced into the sphere PTB . Physikalisch-Technische Bundesanstalt, Braunschweig, and placed near the counter. Federal Republic of Germany (R . Jahr and M. Cosack). 244 3. Fission chamber ( U) The fission chamber was constructed and supplied by NBS. It consists in fact of two chambers called chamber T ("top") and chamber B ("bottom") (Fig. 3), The supports are two platinum foils set back to back and having each a 238y deposit of 1 mg/cm^ over an area of 1 .3 cm' . The gas circulating through the chamber is pure methane. Figure 1 - BIPM polyethylene sphere with BF„ counter (all dimensions in mm) P - polyethylene sphere, 203 F - polyethylene sheath containing the BF_ counter, type 0,2NE3/1 F (LMT) polyethylene plug which can be removed to place the source active part of the counter, ,0 10, I 30 polyethylene plug containing the source 100 mCi Am-Be(a,n) source, 2.2 x 10 5 n/s 2 . H e proportional counter 3 The He counter was supplied by N RC . It has an active length of 10.2 cm and a diameter of 2.44 cm. It is filled with helium-3 at a pressure of 10° Pa. Figure 2 gives an example of the response of the counter to 565 keV neutrons. In the comparison the bias was set at a pulse height which corresponds to 764 keV + (0.6 x E ) , n where E is the neutron energy considered (in keV). thermal peak 764 keV (764*E„)keV bias * (764+ 0.6En)keV - '•V-.-li-s-x.-— *■"»■'• V channel number Figure 2 - He counter spectrum for E = 565 keV n 25 :'■'.'. aluminium 0Z2 teflon Figure 3 - Fission chamber constructed by NBS (all dimensions in mm) a - fission chamber: T - top chamber B - bottom chamber EC - collecting electrodes ^ , D - ^ J °U deposits, 1 mg/cm , surface 1 .3 cm on platinum foils (ground potential) b - envelope of the fission chamber: P - position of the deposits E - entrance of methane S - exit Figure 4 - Response of the fission chamber 245 Figure 4 shows the extrapolation method used for the low energy counts of the spectrum: a straight line is drawn between the point corresponding to the minimum of the "valley" and the one corresponding to zero am- plitude ("true zero") of the multichannel analyzer. 4. Iron foil activation method Iron foils are irradiated in the 14.8 MeV neutron beam and the induced "saturated" (3 activity according to the °Fe(n, p)^°Mn reaction is measured. This com- parison was organised by E.J. Axton (NPL). Before the real comparison started, ° u Co foils were circulated among the participating laboratories to enable them to compare the results of (3 counting measurements. In the light of the agreement obtained the method may be considered va I id . Determination of the sensitivity of the transfer instruments Let us recall that the comparison consisted in measuring the sensitivity £• of each instrument defined as the quotient of the count rate of the detector to the neutron flux density. In practice, all laboratories measured the neutron flux density by means of a quan- tity V n which is the number of neutrons emitted by unit of solid angle in the direction of the detector; therefore, it is easier to compare directly the ratio SI = N c /V n for a chosen source-detector distance d, where N_ is the number of counts of the transfer ins- trument. This gives the relation £ = ^ x d 2 . For the sake of convenience, £~ will be called the sensitivity of the transfer instrument in what follows. a difference turned up in all laboratories (Table 3), Moreover, this difference is more important in the cases where the contribution of scattered neutrons is larger. One can see in Fig. 5 that for the 250 keV energy there are two groups of results, one for the laboratories having a higher correction for scattered neutrons and the other for those having a smaller correction (Table 2a). In addition, it happens that the group with the larger correction obtains also a higher value for the sensitivity of the Bonner sphere. Consequently, if for each laboratory we correct the contribution of scattered neutrons by means of the value determined by the shadow bar, the discrepancy between the two groups of laboratories should be reduced. 2 As a matter of fact, the 1/d law for the deter- mination of the contribution of scattered neutrons is valid only if the variation of the scatter with distance is known or constant. Since this is not true, we should use the shadow bar. Table 4 summarizes the results obtained in the laboratories which used a 50 cm long shadow bar to estimate the contribution of scattered neutrons for the Bonner sphere. In conclusion, we can say that, on the one hand, a great effort is still necessary to improve the absolute measurement of flux density for reaching an accuracy of 1% and, on the other hand, a new suitable transfer instrument has to be found which has a reasonable efficiency and a low sensitivity to scattered neutrons and y rays. Results The results obtained in the various participating laboratories are summarized in Tables 2a, b, c, d, e and f. The Bonner sphere, the ^He counter and the fission chamber were placed at 1 .50 m, 50 cm and 10 cm from the target, respectively. The ]/d* law was used to estimate the contribution of scattered neutrons for the Bonner sphere. The statistical and systematic uncertainties given in the tables were obtained by a quadratic sum of the various contributions as given by each laboratory. These results are also indicated in Figures 5 to 12 . It appears that there exists in general a reasonable agreement among the results (within 5% or better), except for 250 keV (Bonner sphere) where the discre- pancy ranges from 10 to 20%. At 250 keV the correc- tions for the contributions of scattered neutrons are inconsistent: the 1/d^ law was used by all laboratories to estimate the contribution of scattered neutrons, but some laboratories (NPL, ETL and PTB) used also a shadow bar which gave practically the same correction, except for the cases where the contribution of scattered neutrons was high, in particular for 250 keV where Table 2a Results of the comparison for the 250 keV neutron energy Laboratory and "Absolute" Correction for scattered Sensitivity Uncertainty transfer detector* neutrons [Jcounts/(n/sr)j syst . stat. (lcrj instrument (%) (%) (%> / NRC PC 11.3 5.71 x 10" 6 3.5 1.0 CEN DC 6.9 4.81 x 10" 6 2.3 0.4 CO I S £ BCMN PC 20.7 5.31 x 10" 6 2.8 0.5 + ° 1 „"2 / NPL LC 9.1 4.71 x 10" 6 3.2 0.6 ■1 o 1 ETL PC 10.2 5.24 x I0~ 6 2.5 0.6 -6 PTB PC 6.4 4.71 x 10 3.1 0.4 \ NBS BD 1.0 4.76 x 10" 6 3.7 0.8 / NRC PC - 3.35 x 10" 6 3.2 1.7 a 1 CEN DC - 3.05 x 10" 6 2.2 0.1 1 S } NPL LC _ 2.90 x 10" 6 3.1 0.5 < 0,0 J ETL PC 3.05 x 10" 6 2.5 0.4 V *- 1 ■> X ° [ PTB PC - 3.09 x 10~ 6 3.1 0.6 \ NBS BD - 3.20 x 10" 6 3.6 0.8 PC - recoil proton proportional counter DC - directional counter LC - long counter BD - block detector 246 Results of the comparison for the 565 keV neutron energy Laboratory "Absolute" de tec tor* Correction for scattered neutrons Sensitivity [_counts/(n/sr)^ Uncertainty transfer syst . stat.(l (%) (%) / NRC PC 8.9 5.96 x 10" 6 1.9 0.4 u." CEN DC 3.5 6.21 x 10" 6 2.2 0.1 . »— 1»- 6.5, 10" 6 coun»i/(n/ir) Figure 7 - E = 565 keV, sphere + BF counter at 1 . 50 m Figure 10- E = 2 . 50 Me V. sphere + BF„ counter at 1 . 50 n J n r 3 248 CEN c Table 3 Contribution of scattered neutrons (%) (polyethylene sphere + BF counter at i .50 H >- 7.0 x 10 " counts/(n/ s r) 250 keV 565 keV 2.2 MeV 2.50 MeV Laboratory (a) (b) (a) (b) (a) (b) (a) (b) BIPM - - - - 15.8 17.8 NPL 9.1 11.2 6.0 6.3 6.0 6.2 6.5 5.8 ETL 10.2 17.1 9.0 11 .2 - 48.5 58.1 PTB 6.4 9.3 4.6 4.8 4.0 5.3 3.6 5.2 NBS 1.0 1.0 1 .0 1 .0 - - a) according to the 1/d law b) measured with a shadow bar Table 4 Figure 1 1 - E = 14.8 MeV, fission chamber at 10 ct n Results of the comparison. Polyethylene sphere + BF_ counter at 1 .50 (Correction for scattered neutrons measured with a shadow bar) 3.2 H I* Fe toil 47 I * 3.4x10" c 3 □C Fe foil 45 I 1 1 Fe foil 48 I ■ I I ■ J 3.4,10 * 3.0 3.4x10 ountyWc™ ) H 1% Fe foil 49 I ■ I H i% Fe foil 57 — I— 3.0 Figure 12 - E = 14.8 MeV, Fe(n,p) Mn reaction n Neutron ene rgy (MeV) 0.250 0.565 2.20 2.50 Laboratory NPL ETL PTB NBS NPL ETL PTB NBS NPL PTB BIPM NPL ETL PTB Correction for scattered neutrons (%) 11 .2 17.1 9.3 1 .0 6.3 11 .2 4.8 1 .0 6.2 5.3 17.8 5.8 58.1 5.2 Sensitivity Qcounts/(n/sr)j 4.60 x 10 4.84 x 10" 4.56 x 10" 4.76 x 10" 5.86 x 10 5.70 x 10 5.83 x 10" 5.97 x 10" 7.22 x 10 7.14 x 10" 6.99 x 10" 7.02 x 10" 6.57 x 10" 7.01 x 10" -6 Uncertainty syst. stot.(l100 barns. 10 Physically, the scatterer is a 3 mm thick Mn-Al alloy containing 57 atomic percent manganese. This design eliminates unwanted core neutrons and gammas. The advantage of this design can be inferred from Table II, Lab. 2 keV Flux 7. of Other Energy Neutrons y-Intensity (mR/hr) BNL 3.5 x 10 6 257. 340 MTR 5 x 10 s 407. ~ 1 NBS 2.6 x 10 5 3% 8 which compares 2 keV beams. We note that the NBS beam has a considerably lower intensity than the other two, but also a much lower fraction of higher energy neutrons and a much lower gamma ray background than the BNL beam. For our purposes, this is a very advantageous trade-off. (Greenwood and Chrien, 3 BNL, describe a scandium-cobalt- titanium filter which they estimate would have 9% higher energy neutrons, and a 2 keV flux of 2 x 10 6 neutrons/ (cm 2 -sec) . This filter does not appear to have been actually used for any experiments, however.) The spec- trum of the NBS 2 keV beam is shown in Fig. 2. These data were taken with a 1 atm. hydrogen proton-recoil counter. The solid line shows the spectrum obtained with a 110 cm long Sc filter; the dashed curve shows the effect of adding one cm of Ti to the Sc. Even without the Ti, £ ^ « I I i i i l | I I — i M M 1 1 SCANDIUM FILTERED NEUTRON BEAM 110 cm SCANDIUM PLUS Icm TITANIUM II I II 100 200 NEUTRON ENERGY. keV Fig. 2. Neutron spectrum through scandium filter. The solid curve represents the spectrum with a 110 cm scandium filter alone, and the dashed curve the spectrum with the addition of 1 cm of titanium. the main secondary peak at 29 keV has an area of only 3% of the 2 keV peak. The addition of the Ti reduces this peak (as well as the ones at 7, 15, and 40 keV) by a factor of about 2-1/2, at the cost of only 17% of the 2 keV peak. 251 In terms of total flux, the higher energy contami- nants amount to approximately 6% of the 2 keV flux with- out the titanium; the addition of the titanium reduces their contribution to 3% of the 2 keV flux. Before leaving the matter of 2 keV beams, it must be pointed out that the value of the cross section minimum in scandium of 50 mb originally reported 1 ' 11 is almost certainly wrong. A very careful combined BNL-RPI measurement 12 gives a total cross at the minimum of 700 mb, rather than 50. This higher value is consistent with a very rough transmission measurement made with the NBS filter, and suggests that the optimum filter thickness for a scandium filter may be rather thinner than heretofore believed. Table III. Comparison of 144 keV Filtered Beams Lab. MTR MURR NBS 144 keV Flux 2.5 4.5 10' 10 6 10 s y-Intensi ty (mR/hr) 500 450 40 Our silicon filtered 144 keV spectrum is shown in Fig. 4. 24 keV Beam The NBS 24 keV beam is produced by an iron-aluminum filter in another through tube, using a graphite scat- terer. Both iron and aluminum have broad cross section minima near 24 keV, but their other minima do not, in general, overlap. Hence, it is not too difficult to produce reasonably clean 24 keV beams by judicious choice of the thickness of iron and aluminum, and since iron is a good gamma-ray absorber, these beams generally do not have serious gamma ray background problems. In this instance, therefore, the use of the through tube does not seem to be any great advantage. Our iron- aluminum 24 keV spectrum is shown in Fig. 3. The magni- tude of the higher energy components is typical of the other 24 keV beams listed in Table I. IRON-ALUMINUM FILTERED BEAM 25 cm IRON 36 cm ALUMINUM RELATIVE AREAS ENERGY % OF FLU 24.3 keV 74 135 280 364 97.0% 0.23 1.6 1.0 0.16 /"\ s w Fig. 50 ioo 150 ~2CQ 250 3bO- - 350 40C NEUTRON FNERGY , keV 3. Neutron spectrum through the iron-aluminum filter. 144 keV Beam Other than the 144 keV window, the only window in silicon is at 54 keV, and the effect of this window can easily be minimized by an additional thin titanium filter. The main problem with silicon-filtered beams is the Y - ray background. In our case, the silicon filter is on the other side of the through tube from the iron- aluminum filter and, hence views the same graphite scatterer. Table III compares the 144 keV beams; we see that the use of the through tube-plus-scatterer greatly reduces the Y _ t> a ckground, but at the expense of a loss in neutron intensity. x 100 2 O cc h- => UJ LU > LU 50 10 l — r "I — i — r SILICON FILTERED NEUTRON 8EAM 136 cm. SILICON 2 cm. TITANIUM 144 keV - 98.5% 54 keV - 1.5% |l 1 I I I I I 1 I I I ll I 50 !00 150 NEUTRON ENERGY , keV 200 Fig. 4. Neutron spectrum through the silicon filter. Beam Intensity Measurements The 2 keV and 24 keV beam intensities are determined by counting with a one-atmosphere 10 BF 3 counter. We use standard commercial counters, 5 cm diameter, varying in sensitive length from 6 to 33 cm, and employing a thin ceramic end window. The beam is brought in through the center of the ceramic window, parallel to the center wire. Since the beams are 2-1/2 cm diameter or less, there is minimum wall effect, and it is only necessary to know the 10 B density and the sensitive length to de- termine the counter efficiency. The 10 B density was determined by measuring the transmission of the counter along a diameter, using beams of 3.86 and 54.4 millivolt neutrons produced by NBS crystal spectrometers. The transmission was measured relative to an identical, but empty, counter to take into account the effect of the 0.9 mm thick stainless steel walls. At these low neutron energies, the 10 B absorption dominates the transmission, and thus the 10 B content can be determined quite accu- rately. The sensitive length of the counter was deter- mined by scanning along its length with a finely colli- mated thermal beam. The small correction for losses in the 2 mm ceramic window was determined by an explicit measurement using a dummy window. We thus end up with counters whose absolute efficiencies are known in terms 252 of the 10 B cross section and the explicitly measured physical properties of the counter. Measurements of the 2 keV beam intensity, using 3 counters of lengths 6.4, 31, and 33 cm, showed a spread in values of less than + 1%, which suggests at least internal consistency in our use of these counters. In the absence, however, of any cross checks on these measurements (e.g., foil activation) we feel that we can only quote an accuracy of ± 5% for the 2 keV beam intensity. This technique is better suited to the 2 keV than the 25 keV beam, since the 10 B(n,a) cross section is ^ 3-1/2 times higher at the lower energy. However, a proton recoil counter can also be used for the 25 keV beam, as well as for the 144 keV beam. In this case, we use a counter which is physically similar to the 31 cm long BF3 counter, but filled with hydrogen to an accurately known pressure. Since the hydrogen cross section is very well known, the area under the proton recoil spectrum is a direct measure of the beam intensity. The 25 keV beam intensity measurements performed with the two different types of detectors differed by 7%. We consider this to be satisfactory agreement at this stage, although we are working to understand, and reduce, this difference. The flux quoted in Table I is the mean of these two values and ± 7% is probably a reasonable measure of the uncertainty in this flux, as well as for the uncertainty in the 144 keV flux. Use of the Beams for Calibration importance of this can be judged from the albedo dosimeter calibration curve; without the NBS 2 keV datum, there would simply be no way of knowing the low energy response of the dosimeter. Future Development We feel that "Phase I" of the NBS filtered beam facility is complete: we have three clean beams whose intensities are well enough known for the initial uses to which they are being put. "Phase II" will consist first, of optimizing the filter configurations, and, second improving the calibrations. We feel that the 2 keV beam intensity can probably be increased with no appreciable loss in beam purity, and that the 24 keV beam purity can be further improved. More careful work on the corrections associated with the counter calibra- tions (e.g., end effects, backgrounds, etc.) together with alternate calibration methods (foil activation, counting with standard fission chambers) will reduce the uncertainty in our intensity measurements. We feel that accuracies in the 2%-3% range are a reasonable goal for the near future. Conclusion At the conclusion of his talk at the 1970 Neutron Standards Symposium, Orval Simpson predicted that "... one fine day a reactor will offer standard neutron sources of 2, 24.5, and 144 keV produced from filtered neutron beams." We should like to suggest that perhaps that fine day has arrived. The first extensive use of the beams for dosimeter calibrations was by Dale Hankins of Livermore. Some of his results are shown in Fig. 5. 13 The circles are the -KT~1 — I I I I 1 l | 1 1 — I I II I l| I I I I I I I I """ -^ ^ALBEDO DOSIMETER REMMETER ond DOSIMETER CALIBRATION DE HANKINS. UCRL- 78307 (19771 °}lll CYCLOGRAPH )n.bs filtered beams I I I-+-I+I4J Fif 10 100 NEUTRON ENERGY, keV 5. Energy response of 9~inch sphere and of albedo dosimeter. data obtained with the L.L.L. cyclograph; the squares are the data taken with the NBS beams. While the ordi- nate is in arbitrary units, there has been no normaliza- tion between the two sets of data. The line is an eye- guide only. We note that the NBS 144 keV point is in excellent agreement with the data taken with the cyclo- graph. The 30 keV cyclograph data represent the low energy limit of that device; the beam quality is rather poor at this energy and, for some of Hankins ' runs , there was a suspicion that the data were systematically some- what low. 14 This is borne out by comparison of the 30 keV cyclograph and 25 keV NBS datum for the 9 inch sphere calibration. The 2 keV NBS data then allow the calibra- tions to be extended a decade lower in energy. The Acknowledgements The importance of using the resonant scatterer technique to clean up the 2 keV beam was first pointed out by Ivan Schroder, and we are grateful for his con- tinued advice, assistance, and close collaboration. It is a pleasure to thank E. D. McGarry and C. Heimbach for making the proton recoil spectra measure- ments shown in Figs, 2, 3, and 4. References 1. 0. D. Simpson, J. R. Smith, and J. W. Rogers, Proc. Svmp. Neutron Standards and Flux Normalization, Argonne, Illinois (1970). CONF-701002, p. 362. 2. F. Y. Tsang and R. M. Brugger, private communication. 3. R. C. Greenwood and R. E. Chrien, Nucl. Instr. and Meth. JL_38, 125 (1976). 4. E. Kondaiah, R. P. Anand, and D. Bhattacharya, Proc. Int. Conf. Photonuclear Reactions and Appli- cations, Asilomar, California (1973), CONF-730301, paper 2D11-1. 5. R. C. Block, Y. Fujita, K. Kobayashi, and T. Oovaki, J. Nucl. Sci. and Technology, 1_2, (1975). 6. R. C. Block, N. N. Kaushal, and R. W. Hockenbury, Proc. Nat. Topical Meeting on New Develop. Reactor Phys. and Shielding, CONF-72091, 1107 (1972). 7. F. Y. Tsang and R. M. Brugger, Nucl. Instr. and Meth. 134, 441 (1976). 8. J. R. Harvey and S. Beynon, Proc. First Symposium on Neutron Dosimetry in Biology and Medicine, Neuherberg/Munchen (Germany), (1972) p. 955. 253 9. I. G. Schroder, R. B. Schwartz, and E. D. McGarry, Proc. Conf. Neutron Cross Sections and Technology, Washington, D.C. (1975). NBS Special Publication 425, p. 89; E. D. McGarry and I. G. Schroder, loc . cit., p. 116. 10. M. D. Goldberg, S. F. Mughabghab, B. A. Magurno, and V. M. May, "Neutron Cross Sections, Vol. IIA, Z=21-40," BNL 325, Second Edition, Supplement No. 2 (Physics TID-4500) Feb. 1966. 11. W. L. Wilson, "Neutron Cross Section Measurements and Gamma Ray Studies of l45 Sc;" M. S. Thesis, University of Idaho Graduate School, Dec. 1966 (unpublished. ) 12. H. I. Liou, R. E. Chrien, K. Kobayashi, and R. C. Block, Bull. Am. Phys . Soc 22^, 53 (1977). 13. D. E. Hankins, UCRL-78307 (1977). 14. D. E. Hankins, private communication. 254 MUCH ADO ABOUT NOTHING: 45 56 * DEEP MINIMA IN SC AND FE TOTAL NEUTRON CROSS SECTIONS R. E. Chrien and H. I. Liou Brookhaven National Laboratory, Upton, New York 11973 R. C. Block and U. N. Singh Gaerttner Linac Laboratory, Rensselaer Polytechnic Institute, Troy, New York 12181 and K. Kobayashi Kyoto University Research Reactor Institute, Kumatori-cho , Osaka-fu, Japan The deep minima in - > Sc and 56p e neutron total cross sections have been measured at the Gaerttner Linac Laboratory by using thick, ultra-pure samples in transmission experiments. The samples are used to produce quasi-monoenergetic beams at the BNL High Flux Beam Reactor. For the 45Sc minimum near 2.05 keV we obtain °" tota i= 0.71 + 0.03 barns, in sharp contrast to a previously reported value of ~ 0.05 barns. The 56Fe measurement was carried out with a 6 kg, 68.58-cm-long sample of 99. 87% isotopically pure sample of 56p e ; a minimum cross sec- tion of 0.0085 + 0.004 barns at 24.39 keV is inferred. This may be compared to a value of 0.420 barns for natural iron. (Neutron total cross sections, Sc measured from 0.4 to 22 keV; deduced neutron resonance parameters; ^"Fe mea- sured from 0.4 to 1000 keV) I . Introduction Many elements exhibit deep neutron total s-wave cross section minima due to the interference between resonance and potential scattering amplitudes. These minima are of considerable interest in shielding appli- cations and in the design of transmission filters for the production of quasi-monoenergetic neutron beams. For the latter application both steady state re- actors^>2>3 and pulsed sources^'-* have been used. The use of such filters for dosimetry measurements has been discussed by Schwartz^ at this meeting. Scandium and iron-aluminum filters have been also rather widely used for capture Y-ray and cross section studies. The empirical optimization of scandium and iron filters has been discussed by Greenwood and Chrien. A sum- mary of known filter facilities has recently been prepared by Tsang and Brugger . In spite of the wide applicability, the total neutron cross sections of ^^Sc and ->"Fe , in particular, are rather ly poorly known. In the installation of the HFBR scandium filter, it became obvious that the flux measured at this filter could not be reconciled with the accepted cross sections. Furthermore no accurate measurement of 56p e total cross sections has been published. For these reasons we carried out experi- ments to establish the cross sections at the 2.05 and 24.4 keV minima in 45 Sc and 56 Fe . Thick metallic samples of Sc (99.9% pure) and separated Fe (enriched to 99.87% in 56p e ) were obtained from the HFBR Tailored Beam Facility. Neutron transmission measurements were carried out at RPI ' s Gaerttner Linac Laboratory, and the data subsequently were analyzed at BNL. II. Method A full description of the RPI Linac Facility has been reported, and only a brief description is given here . The standard water-cooled Ta and CH2-moderated neutron TOF target and the 10B-NaI neutron detector at the 28.32-m flight path were used for these mea- surements. The linac was operated at a repetition rate of 500 sec" 1 , an electron energy of ~ 70 MeV, a peak electron current of * 1 A, and an electron pulse width of either 19, 35, or 66 ns . The counting data were recorded vs. TOF with the 31.25-ns TOF clock interfaced to the PDP-7 on-line computer. The transmission samples were cycled automatically into and out of the neutron beam by the programmed com- puter, and a cycle was repeated every 10 to 20 minutes to average out neutron source intensity fluctuations. The composition of the ^5g c anc j 56p e satn pi es are listed in Table 1. The ^Sc samples were prepared by Table 1 Sample Properties (A) Scandium Samples Sample No. Dimensions (cm") 1/N (barn/atom) 1 2.54 diameter x 10.2 2.738 2 2.54 diameter x 15.2 1.844 3 2.54 diameter x 30.5 0.922 4 2.54 diameter x ^0.2 152.5 5 2.54 diameter x 0.55 50.8 6 2.54 diameter x 2.4 11.7 Impurity Atom Pc r Cent H 0. 49 0. 18 Ta 0. 037 (B) 56 Fe Sample Dimensions (cm) 1/K (barn/atom) 3.22 ! x 3.85 x 68.6 0.175 Isotope Wt . Per Cent 5 *Fe 56 Fe 0.05 99.87 57 Fe 58 Fe 0.07 0.006 Impurity Atom Per Cent Impurity Atom Per Cent .617 Cr .011 Cu .120 W .010 H .059 Ni .005 C .020 Si .002 N .012 P .002 vacuum sublimation onto a cooled Ta plate, and the sublimed metallic material was subsequently pressed into steel containers 2.54 cm in diameter. The H and impurities were determined by the vacuum fusion method, and the heavier elements by mass spectrometry, 255 for both samples, at the Ames Laboratory Analytical Center. In addition the Ta content of the Sc was checked by emission spectrography , by resonance x-ray f luorence , by neutron activation, and by neutron transmission. This wide variety of methods for Ta analysis was necessitated by the discovery of a sharply inhomogeneous distribution of Ta impurity throughout the bulk of the Sc filter. Both activa- tion and transmission methods were able to sum over large sections of the filter and thus produce a more reasonable average impurity content than was the case for the other techniques, which were applied to small samples of the filter material. The 5oFe sample was prepared at Oak Ridge National Laboratory from electromagnetically enriched iron in the form of iron oxide. The metallic sample was obtained by reducing the oxide in a hydrogen atmosphere. The isotopes and impurities listed in Table 1 were determined from measurements of the iron oxide and metal respectively. In the final analysis, these cross section mea- surements are only as reliable as the impurity content determination. At the Sc minimum, the impurity cor- rection is about 112 mb out of a measured 822 mb ; while in the ^"7e, the correction is about 51.5 mb out of a measured ~ 60 mb. In each case ENDF/B IV evaluated cross sections were used in the correction. RELATIVE NEUTRON INTENSITY FOR 2 , 8 50 40 30 20 UJ z z < X o £E UJ o_ UJ < g: o z o <_> UJ > |i 1 1 1 1 i i i i i i i — i — i — \ — i r 45 Two sets of Sc transmission measurements were carried out. In one measurement the 10.2-cm-long scandium sample No. 1 was placed in the neutron beam to produce a T0F- filtered spectrum of neutrons which is peaked near the interference minimum in scandium. This removes most of the neutrons from the beam and results in a very low background of neutrons with energies far from the minimum. Then samples No. 2 and 3 were cycled into the filtered beam to obtain an accurate measurement of the cross section near the interference minimum. The other measurement was a conventional transmission measurement. The 10.2-cm- long scandium filter was removed from the beam and samples No. 4, 5 and 6 were cycled into the beam. This latter measurement enabled the cross section to be determined near the peaks of the resonance as well as near the minima. For the ^°Fe measurements a 20.3-cm-long filter of pure Armoo iron was placed in the neutron beam to produce a TOF-filtered spectrum peaked near the iron minimum. The 68.6-cm-long ^°Fe sample was then cycled into and out of the filtered neutron beam. The fil- tering effect in iron is illustrated in Fig. 1 near the 24 keV minimum. Here the *■ B-Nal detector rela- tive counting rate is shown for Armco iron filters varying in thickness from 5.1 cm(2") to 50.8 cm(20") . For this experiment the 20.3 cm(8") thickness was used, and from Fig. 1 we see that the peak transmission through the filter is 48% and that most of the neutrons at energies several keV away from the peak are removed from the beam. , 14", AND 20" IRON FILTERS (KeV) 10 2" (84% TRANSMISSION) 8" (48% TRANSMISSION) 500 600 700 CHANNEL NUMBER Fig. 1: 10. The " B-Nal detector counting rate vs. T0F with Armco iron filters 5.1-cm(2"), 20 .3-cm(8 M ) , 35 .6-cm(14"') and 50.8-cm(20") thick. The peak transmission through each filter is shown in parentheses. 256 The TOF counting data were corrected for dead- time losses in the electronics and for background, and the neutron transmission was then determined. The total cross section 0" t was obtained from the neutron transmission with the following equation •(1/N) In T (N . a , )/N air air (1) where N is the thickness of the scandium or 56p e sample, T is the neutron transmission, N a i r is the thickness of air displaced by the sample, and o a ± r is the neutron total cross section of the air. The cross section of air was obtained from the oxygen and nitrogen cross sections plotted in BNL-325. III. Results A. 45 Sc The 4 5 Sc total cross section obtained from equation (1) is plotted in Fig. 2 over the neutron energy range from ~ 0.4 to 22 keV. This plot is a "blend" of all the Sc data, where the data near the peaks in the cross section are from the thinnest sample and the data near the deep minima are from the thickest sample. The data have been corrected for the presence of the contaminants listed in Table 1. in the higher energy ^Sc minima, and thus the length of the ^5g c filter should be limited to enhance the transmitted neutron flux near 2 keV relative to that transmitted at higher energies. The Sc total cross section has been fit by an approximate R-matrix formalism, 13 and the solid curve through the experimental points in Fig. 2 is a "best fit" to the data. The curve has been reso- lution broadened; the resolution FWHM is approximately 3 channels near the 2 keV minimum. In Table 2 are listed the resonance parameters derived from this fit. T able 2 4! Sc Re sonance Parameters -wave level parameters : r Y = 0.4 eV n (eV) V 1 n (eV) J 3 E (eV) 11575 r n _ievi 290 J -500 4.o (r n °) 4 -220 0.67 (T n °> 4 14525 20 3 3295 75 3 14740 26 4 4330 340 4 15560 28 4 6684 130 3 15850 5 3 8023 145 4 18580 32 3 9092 300 3 18870 62 4 0625 10 3 20500 80 4 0735 6 4 20780 710 3 p-wave level parameters ie\Q &E n (eV) 460.6 0.0022 1060.4 0.0050 7377.0 0.4 7458.0 0.4 7548.0 0.25 Resonance parameters for -'Sc derived from shape fits to the total cross section. This fit was determined by the following procedure: (a) The spins of the positive energy resonances were obtained by fitting the peak cross sections and the shape of the interference between resonances. Fig. 2: The neutron total cross section of ^^Sc . The experimental data are shown as solid points with error bars (standard deviations deter- mined from the counting statistics'* . The solid curve is a resolution-broadened "R- matrix" fit to the data using the resonance parameters listed in Table 2. The measured minimum cross section of 45g c near 2 keV is (0.71 + 0.03)b, where the error is derived from a combination of the counting statistics and the uncertainties in the H and corrections. The mini- mum occurs at an energy of (2.05 + 0.02) keV . This minimum cross section of (0.71 + 0.03)b is in serious disagreement with earlier measurements reported by Wilson^ of ~ 0.05 b, and it is also in disagreement with the evaluated minimum cross section^ of ~ 0. which was based on these older measurements. Our result of 0.71 b has serious implications for the effective use of expensive scandium for filtered beams. The 0.71 b cross section at 2.05 keV is comparable to, or larger than, the cross sections 55 b (b) Negative energy levels were introduced to fit the cross sections at thermal and in the low energy region. (c) The neutron widths were obtained for all the resonances such that (i - ) the calculated R-matrix cross section curve produced an acceptable overall fit to the data, and (ii) the R-matrix minimum total cross section near the 2 keV minimum equaled the observed value of 0.71 b. (d) A single radiation width was then determined for all the resonances such that the thermal capture cross section resulting from the sum of contributions from all the resonances equaled the evaluated value^ of (26.5 + 1.0) b. The best fits to the data are obtained when the J=3 channel spin contributes significantly to thermal capture. The "best fit" parameters listed in Table 2 produce a thermal capture cross section which has approximately equal contributions from J=3 and J=4 channel spins. Thermal neutron capture v-ray spectral measure- ments by Bolotin favor a significant J=3 channel spin contribution. He observed a primary gamma-ray transi- tion to the 1" state in Sc at an excitation energy 257 of 142 keV . Thermal capture in scandium consists of a mixture of capture into 3" or 4~ states and Bolotin's observed transition strength of 1.3 gammas per 1000 captures indicates that this gamma ray is an E2 transition from a 3" to a 1" state. The par- tial radiation width for this E2 transition can be calculated from the observed transition strength and the resonance parameters deduced from the R-matrix fit to the total cross section. The E2 width cal- culated from the "best fit" parameters in Table 2 is about 6 times larger than the typical E2 width observed in this mass and energy range, and this is reasonable considering the fluctuations of the observed E2 widths. However, when the E2 width is determined from the R- matrix parameters which produce predominantly J=4 channel spin thermal capture, the E2 width is about 500 times larger than the typical E2 width. Such a 500 times larger E2 width is very un- likely, and thus Bolotin's measurements favor thermal capture which has a significant J=3 channel spin con- tribution. We have independently confirmed Bolotin's observation in a separate experiment to be reported elsewhere . CHANNEL NUMBER A polarized neutron experiment by Roubeau et al,^ claims to have measured the difference in scattering lengths at thermal between the (I + 1/2) J=4 and (I - 1/2) J=3 spin states. They report a value of (a+ - a.) = + 1.2 x 10" . cm. The implication of their result is that a J=4 state dominates thermal scattering. This result is inconsistent with our present result, since it would mandate a deep minimum near 2 keV, as a result of interference between the dominant J=4 resonance at 4.330 keV and a strong J=4 bound state. An attempt to fit our data subject to this constraint results not only in a poor fit, but requires absurd R-matrix parameters, (e.g., Rj- 1,/Rj+i, ~ 8) . We conclude that the experiment of Roubeau et al. is in error. Fig. 3: The B-Nal detector counts vs. TOF channel number for a 20.3-cm(8") Fe filter placed in the neutron beam. The TOF channel width is 31.25 ns. 56 Fe The B-Nal detector counts per TOF channel are plotted in Fig. 3 for the 20.3-cm (8") Armco iron fil- ter in the neutron beam, and in Fig. 4 for the 20.3- cm Armco iron filter plus the 68.6-cm (27") 56p e sample. The 68.6-cm ^ Fe sample produces a quasi- monoenergetic peak of neutrons which peaks near an energy of 24 keV. The total cross section of ^"Fe near the 24 keV minimum is plotted in Fig. 5. For iron both impurity cross section corrections and resolution corrections are very important, which was not the case for scandium. The total correction for impurities amounted to 51 mb in the region of the minimum; and we infer a minimum observed cross section of ~ 15 mb at an energy of 24.39 + 0.04. This cross section is, however, seriously affected by our resolution function, which has a FWHM of about 200eV at the minimum (~ 2-1 channels). Using the parameters of Pandey and Gargl" we fitted a series of resolution broadened cross sec- tion curves to the minimum, using the radiation width of the major s-wave resonance near 27 keV as a parameter. We find the best fit for a radiation width of about 2.2 eV , which provides an independent mea- surement of that important parameter. We also include a p-wave potential scattering contribution in this calculation. From the best fit, we infer a minimum cross section of 8.5 + 4 mb , of which 1.3 mb is due to £=1 scattering, and the balance is due to the (n,Y) reaction. Fig. 4: The B-Nal detector counts vs. TOF channel number for a 20.3-cm(8") Fe filter plus a 68.8-cm(27") 56p e sample placed in the neutron beam. The TOF channel width is 31.25 ns . 258 /VVV-r- ; ■ -: =~= r. : i ^ -' _- : d r -»• -■ .:::- z --- / T ~ ..-■ ----- - : : ::" E:: _--.: h HS « ==. EE Er lr — /I Y-\- EE ~ "T- :-■ ttt; ~ z EE -■.-. S: eh ■~ Er -~ 7 — - EE ™7 4_ :- t:. feF T :--;f i? X : '.-'- Hr. ~.= rjif: 7.::= ■.. :. ,: -- ■ : :- . 1: 3 — -t- P ::r - ~ ? i I i \ = = ' EE z= 4r -■'- Eri £= EE ~ ? \ 2 . If Ti" Tt • • ! ill — - zz — EE ii~ ^ -e -5 =3 \ V — : ■]\ . . ^ 1 1 ; ... M i Hi: tV — — — / r* ~_ —■_ — ! i ' ' l ' 1 1 1 ill / P 1 ' : i ; ♦ --*Y •— I I 1 1 ; . ■ i ! 1 l!" i 1 . : « 1 | ] 1 : * i : : i f IOQ-" '? q J . < L_ j 7 - k * - rJ ^ e - — ■ - ;:::: L-- k\ -f :r _- - : -- ,£ - " lZ — - EE I~ --■ E: — ■ -.-■- — — — . .-.-: ' "'\ :_:■ — ; ~ rl~ TiE - ; 1> - - : - - ' EE ^ ;: :::: — .... ~z --__ .i=i zzz - EE EE :E = EE ii . \ :: 1f :--. :: - "- r = "£ ~ :::: : : :: Ik" - — - — — — — — ~ L 1 ^ =; zi'r. ~: ==; ,:~ .— --.-- n\ /// :■:: Z^ ■=-: - *':•"."• T :" n - 3. ._ 1 - ±r__. — " zzz. Er ~ - ■ — , . £rr „_ ___ :E — 1 * : ::- i — ZZl. "ZZ- ZZi r^E \ ft-nf- i 1 " ■fH p |H' ThE IfL- — ~_ /7 Z± Oi JV 2 1 i \ i / - ~j.Z rz — : -- ~ ~- EE — il i! 'ii fill" ■ ■ T r,p 7 — 1_ — — — #/> Q n ii ' ' © — /0 ii 1 1 1 1 9 == y-7 v . ^i y 8 '* -/> w*^ 7 6. '-'■ 5 " r 4 ~ ix—i : = E^ — — - — '■■ = — V a*' fa ^ el ^ :— -L-'Jr — — — — = — ^z — EE 2 " ti r^r ■ry ffft h ■ ■44 -RE tfff — : ■if: — : - Z^L — -Et r— -^+^ ■ i I x — ; 1 1 : TTlt t ■ - tH+ ! ...j , t ll 1!7| 1 1 lin 1 in fj-n in : r ^ A/ £ ?s •5 7- S :i/ E — - — — ti ------ — ■ — 1 1 ■in in !|i i 22,5 23,0 23,5 24-0 2V-.5 25.0 Fig. 5: The neutron total cross section of Fe near the 24.37 keV minimum. The solid curves are resolution-broadened "R-matrix" calculations using the resonance parameters shown in the figure. 259 The minimum cross section measured for this -""Fe sample is almost two orders of magnitude smaller than the "^ 420 mb cross section measured for elemental Fe . ' This results in a much more in- tense beam of quasi-monoenergetic ~ 24 keV neutrons from this filter than from a filter of Fe of the same length. For example, for the 68.6-cm long filter with 1/N = 0.175 b/a, the transmission through the ^"Fe filter at 24.4 keV is 72% (allowing for impurities), whereas the transmission through the same thickness of Fe is only 9%. Thus a quasi-monoenergetic beam of ~ 24 keV neutrons can be obtained with a 56jre filter which has excellent transmission through the 24.4 keV minimum. The filter can be used in very thick con- figurations to reduce unwanted fast neutrons and gamma rays . Summary The neutron total cross section has been measured for 4og c and 56p e w ith particular emphasis upon mea- suring the cross section in the minima. The 4-5gc cross section has a prominent minimum at (2.05 + 0.02) keV which is (0.71 + 0.03) b. This cross section is an order of magnitude larger than estimated from earlier measurements, and this has serious implications in the design of a 45sc filter for reactors. Although the design of a ^Sc filter system depends upon the appli- cation of the system (e.g., for neutron capture spec- tra, dosimetry, etc.), this higher cross section of 0.71 b should lead to the selection of a thinner Sc filter than one based on the former ~ 0.05 b value. The Fe cross section has a prominent minimum at 24.4 keV which is (8.5 + 4.0) mb . This is con- siderably smaller than the ~ 420 mb minimum in ele- mental iron, and thus thick filters of 56Fe can pro- vide intense quasi-monoenergetic beams of ~- 24 keV neutrons with a very small contamination of gamma rays and fast neutrons. y Z . M. Bartolome, R. W. Hockenbury, W. R. Moyer , J. R. Tatrczuk and R. C. Block, Nucl. Sci. and Eng . 37, 137 (1969) . 10 D. I. Garber and R. R. Kinsey, BNL-325, 3rd Ed., Vol. 2, (1976). U W. L. Wilson, M.S. Thesis (Univ. of Idaho, 1966), unpublished . "B . A. Magurno and S. F. Mughabghab , Proc . Conf. on Nuclear Cross Sections and Technology, NBS Spec. Pub. 425, Vol. 1, 357 (1975) . 13 R. G. Thomas, Phys . Rev. 97, 224 (1955); F.W.K. Firk, J. E. Lynn, M. C. Moxon, Proc. Phys. Soc . J32, 477 (1963). 14 H. Bolotin, Phys. Rev. 168, 1317 (1968). 1 5 P. Roubeau, A. Abragam, G. L. Bacchella, H. Glaetti, A. Malinovski, P. Meriel, J. Piesvaux and M. Pinot, Phys. Rev. Lett. 33, 102 (1974). M. S. Pandey, J. B. Garg, J. A. Harvey and W. M. Good, Proc. of Conf. on Nuclear Cross Sections and Tech., NBS Special Publ. 425, Vol. 2,748 (1975). 17j. A. Harvey, Proc. of New Devel. in Reactor Phys. and Shielding, CONF-720901, Book 2, 1075 (1972). 18 K. A. Alfieri, R. C. Block and P. J. Turinsky, Nucl. Sci. and Eng. 51, 25 (1973). 16 Acknowledgements The authors wish to acknowledge the valuable contribution of B. Beaudry in arranging for the vacuum fusion and mass spectrographs analyses of the samples at Ames Laboratory. Valuable discussions were held with Larry Passell, Harvey Marshak, Wally Koehler, Marty Blume and Said Mughabghab. The able assistance of Frederick Paffrath in the sample preparation is much appreciated. References Work supported by the Energy Research and Development Administration . 0. D. Simpson and L. G. Miller, Nucl. Instr. and Meth. 61, 245 (1968); and U.S. Atomic Energy Commission Report I N-1218 (1968). R. B. Schwartz, contribution to this conference. 3r . C. Greenwood and R. E. Chrien, Nucl. Instr. and Meth. n8, 125 (1976) . 4 R. C. Block, N. N. Kaushal and R. W. Hockenbury, Proc. of New Devel. in Reactor Phys. and Shielding, CONF-720901, Book 2, 1107 (1972). 5r. C. Block, Y. Fujita, K. Kobayashi and T. Oosaki, J. of Nucl. Sci. and Tech. 11, 1 (1975); Y. Fujita, K. Kobayashi, T. Oosaki and R. C. Block, Nucl. Phys. A258 , 1 (1976) . R. B. Schwartz, contribution to this conference. F. Y. Tsang and R. M. Brugger, private communication, University of Missouri. 8 R. W. Hockenbury, Z. M. Bartolome, J. R. Tatarczuk, W. R. Moyer and R. C. Block, Phys. Rev. 1_78, 1746 (1969) . 260 THE U-235 NEUTRON FISSION CROSS SECTION FROM 0.1 TO 20.0 MEV W. P. Poenitz Argonne National Laboratory, Argonne, Illinois 60439 The status of the U-235 fast neutron fission cross section is discussed based primarily on material contributed and considered at the NEANDC/NEACRP Specialists Meeting on Fast Neutron Fission Cross Sections held at Argonne National Laboratory in June 1976. However, some newer measurements and evaluations are discussed as well. Specifically, recent measurements at ANL over the energy range 0.2-8.2 MeV, using several BND's, are reported. Data from the last 10 years are found to be in good agreement with an evaluated average of these data. Suggested problem areas are in- vestigated in terms of their actual significance. It is found that the presently suggested version of ENDF/B-V for the U-235 (n,f) cross section does not represent the data base well and a reconsideration is recommended. (U-235 (n,f), 0.1-20.0 MeV, Status, Data, Evaluations) Introduction An NEANDC/NEACRP Specialists Meeting on Fast Neutron Fission Cross Sections of U-233, U-235, U-238 and Pu-239 was held at Argonne National Laboratory in June, 1976 1 . One of the major sessions dealt with the absolute U-235 cross sections. A workshop on absolute cross sections considered essentially U-235 and presented conclusions and recommendations. The material presented at this meeting, and considered by the workshop, forms the base for the present in- vestigations. However, additional data were included and further extensive analysis was carried out. This has led to the present status report and recommenda- tions which go beyond those of the above meeting. Recent Experiments Most of the experiments considered here have been described in detail in the literature, and some of the detectors were discussed in other sessions of this symposium. Thus, the present account will be extremely short. WaAion ('76 ANL, Re^.2) This was a very important and extensive contribution on U-235 (n,f) at the '76 ANL Meeting. Though the actual normalization of the data was in the 7.8-11.0 eV interval, the experiment follows a suggestion by C. Bowman 32 that all cross sections be measured as shape data and then normalized at thermal energy. This would utilize the well known thermal cross sec- tion. The measurements were carried out at the NBS- Linac using the time-of-flight technique and a "white neutron-source" spectrum. First, the ratio U-235 (n,f ) /Li-6(n,a) was measured from 7.8 eV to 30 keV using a multiple layer ionization chamber and a Li- glass detector. Secondly, the ratio of U-235(n,f)/ H(n,n) was measured with the same ionization chamber and a recoil detector. The U-235 (n,f) cross section was obtained from the U5/Li-measurements using an R- Matrix fit 33 for the Li-6 cross section and normal- izing in the 7.8-11.0 eV interval to an absolute value reported by Deruytter and Wagemans 3 . The U-235 (n,f) cross section obtained from these measurements in the 10-20 keV interval was then used to normalize the cross section obtained from the U5/H-measurements. The two sets of data differed in the remaining over- lap-range below 10 keV by ^ 5% but agreed rather well in the 20-30 keV range (better than 1%). The quoted uncertainty of the results is typically 1.6%, not including normalization uncertainties. lea.qe^ oX al. ('76 ANL, Re.f.4) The shape of the ratio U-235 (n,f)/H(n,n) was mea- sured between 1.2 MeV and 20 MeV. A multiple gas- scintillation chamber was used and both fission frag- ments were detected in coincidence. The neutron flux was measured with a special proton recoil telescope. Both detectors were positioned at a 59 m flight-path in the white neutron-source spectrum obtained from the KFK-Cyclotron. The systematic uncertainties of these measurements are between 4 and 8%. Szabo and MaAqu&tte. ('76 ANL, Re/(.5) An extension of previous measurements to the 2.3-5.5 MeV energy range was reported. The efficiency of the directional neutron monitor was determined relative to the D(d,n) reaction cross section and normalized below 2.2 MeV where it was known from two previous independ- ent calibrations. The paper also contains a listing of all previously reported data in its finalized form. BaAton zt al. ['76 ANL, R&^.6) A summary of the previously reported data was presented and some time was devoted to considering the possibil- ity of an energy shift around 6 MeV between the data from LASL and LLL. It was shown that the suggestion of an energy shift results from insufficient consider- ation of statistical uncertainties. The experimenters have in the meantime extensively investigated their energy scale calibration using threshold reactions, resonances in carbon, and the A£(p,y) reaction. The largest difference which could be found was 20 keV, but under more realistic operation conditions the estimated uncertainty would be much less. U. of, klicltlqan {'ANL, Re^.7] These measurements will be reported later in this session. Four values were obtained using calibrated photo neutron sources and one value was obtained with a Cf-252 source. The uncertainties are ^ 2% for the former and 1.6% for the latter. Cance. and GnanldK ['76 ANL, Rzj.Z) An absolute value was obtained at 14 MeV using the associated particle technique and the T(d,n)a-reaction. The experiment was carried out in time-of-flight and in coincidence with the associated a-particle. The result has an uncertainty of 1.9%. Htaton eX al. ['76 ANL, Re^.9) This experiment will be reported later in this session. A value was obtained for a calibrated Cf-252 source with 2.2% uncertainty. ZhuAavlav eX al. ('76 Lowell, Re^.IO) Several values were measured relative to the B-10(n,a) cross section using filtered reactor neutron beams. One of the values falls in the energy range considered here and was obtained at 144 keV using a Si-filter. 261 VoernXz (Re^.ni Previous measurements with a Black Neutron Detector were extended to lower and higher energies using various sized detectors. The basic experimental set- up was similar to that previously reported. An ionization chamber with two U-235 samples back-to- back was placed 'v 8 cm from the neutron source (Li-7 (p,n) below 4.5 MeV and D(d,n) above). The neutron flux was measured with a BND positioned behind a collimator at a distance of "v 5 m. At low energies, a small BND with only one photo-multiplier was used and the modification of the code "Carlo Black" by G. Lamaze from NBS for the Poisson statistics of photo electrons was utilized. At higher energies, a 16 multiplier-detector was used. An additional shape measurement with improved resolution was made by de- tecting the higher energy (> 6 MeV) Y -ravs associated with the fission process in a metallic sample with a large-liquid scintillator. The energy resolution of this experiment was ^ 6 keV in the 200-300 keV range. Data were grouped in 10-keV intervals and normalized using the absolute values. The latter have a typical uncertainty of 2%, but are larger at higher energies. Data Consideration Ge.ne.nal Oveiv-im It was noted at the '76 ANL Meeting that all more recent data can be found in a ± 3% band. Though statistics or systematic errors cause some points to scatter outside this band, the general systematic trends are with few exceptions within this band. Some. Spo-d^ic Ptoblemi As a result of the '76 ANL Workshop on absolute cross sections, it was stated that the ±3% band would have to be replaced with a ±5% band in a local region from 200 to 400 keV. Another notable differ- ence appears to be between the data by Szabo and other sets above 3 MeV. Fig. 1 shows the relevant data sets. The difference at 280 keV between the data by Wasson and those by Szabo 5 and by Poenitz is about 13%. The systematic difference between the data by Szabo 5 and by Barton et al. 6 and Czirr and Sidhu 12 is about 6% around 4 MeV. - i 1 1 1 1 U-235 HI * o III II SZABO UASS0N U. MICHIGAN FOENITZ BARTON CZIRR I l MM s s — i $- £k- ifaMHi/ z _ s _ ■ T ^r^' ^f X i _L 1 1 1 Til ii ll Sim 100 1.00 EN/MEV 10.0 Fig. 1. Comparison of some data in the 0.1-10 MeV energy range. The difference between the data by Wasson and those by Poenitz and by Szabo is about 13% at 280 keV. The difference between the data by Szabo and those by Barton et al. and by Czirr and Sidhu is about 6% at 4 MeV. 1 I I I I I I II I I TTTT WASSON SZABO T - POENITZ 77 BND J I I I I r» 100 1. 00 EN/MEV 10. Fig. 2. Comparison of the data by Szabo with new re- sults by Poenitz. Fig. 2 urements whi areas under be construed be only "\< 4% measurements Wasson data, ratio of the keV interval shows the results from the new ANL meas- ch support the data by Szabo in both question. Structure around 270 keV may from the new data set but the dip would and ^ 10 keV wide based on the ANL compared with "V 9% and ^ 30 keV for the A better comparison can be made for the 250-300 keV interval vs. the 200 - 250 Poenitz (13) GND '72 .958 Szabo (5) '70- '73 .967 Gayther (14) '75 .940 Wasson (2) '76 .907 Poenitz (11) BND '77 .955 Fig. 3. Comparison of fluctuation in the U-235 (n,f) cross section obtained in the 50-400 keV energy range. 262 The applicable uncertainties for these values are unknown but presumably are tf> and Vcuta. Composition Many problems, including some discussed during the '76 ANL Meeting, appear to be based on a poor under- standing of the result of an experiment on the part of "data interpreters". This situation is most vividly demonstrated in the still continuing practice of displaying or discussing differences in data with- out displaying or taking into account the uncertain- ties of these data sets or the uncertainty in the knowledge of the considered quantity. It is impera- tive to judge the significance of existing differ- ences by relating them to their uncertainties. Another important criterion for determining the signif- icance of a suggested problem is the magnitude of its practical implication. For further investigation of the status of U-235 (n,f), an evaluated cross section is desirable. The comparison of individual data sets and conclusions about possible problem areas should be done using such an evaluated "best value set". For practical purpose we restrict the data base here to values re- ported since the experiment by White 16 (y 12 years ago), thus all discussion is restricted, by defini- tion, to this data base. Following standard proce- dures and previously described techniques 17 an eval- uated "best value set" was determined. It represents practically a weighted average of the data base (which was, with one exception, not altered). An extensive display of existing data sets was made in the supplements of the proceedings of the '76 ANL Meeting 1 . Here we consider instead the quantity a j " a ~. D/U = /(Aa. ) z + (Aa ) I ev where a ± Aa is an individual measured cross sec- tion an3 a ± Aa the evaluated "best data set" cross section at the same energy. In other words, D/U represents the deviation of a measured value from the best value in terms of the uncertainties of both, the measurement and the knowledge of the measured quantity. The above definition of D/U permits us to define criteria for the existence or non-existence of significant problems: 1. Fluctuations of D/U values should be expected to follow a normal distribution. This applies to both, individual point-scatter and larger range, systematic variations. Fluctuations must be expected for statistical and for many systematic uncertainties. 2. Average D/U-values should be within ± 1. The cause is most likely an error of the normalization which should not exceed the estimated total uncertain- ty. Vcuta CompcuiLbon Figs. 4 and 5 show the D/U values for the data which contributed to the U-235 (n,f) evaluation. The shape data by Czirr and Sidhu 12 , and by Leugers et al. 4 , were normalized to the absolute 14 MeV value by Cance and Grenier 8 . The shape data by Gayther et al. 14 were normalized to the absolute values from the U. of Michigan . We can conclude that, for the high energy range (E>1 MeV) , no significant problem exists for U-235 as a whole nor for individual data sets. How- ever, it may be noted that all data but those by Czirr and Sidhu 12 tend to suggest lower values between 2 and 5 MeV. It should also be noted that the sug- gestion of an energy shift of the data by Barton et al. cannot be supported by the present analysis as significant. At lower energies (< 1 MeV) the data set by Kaeppeler 18 fails to follow anything close to a normal distribution. The 100 keV-averaged data by Wasson 2 still reflect the influence of the structure previously discussed but, in general, follow the shape measured by Gayther et al. , Poenitz 1>13 and Szabo . The data by Wasson appear to be inconsistent with the evaluated result as they are systematically lower by a D/U of 2 or more. However, a proper normalization and accounting of normalization uncertainties (which will be discussed later) shift these data into a satisfactory D/U range. In Table 1, the average deviations of all abso- lute values which contributed to the normalization of the evaluated cross section are listed. The average uncertainty and the D/U values are also listed. Again, there is no indication of any major problem after adjustments of the data by Wasson were made. The uncertainty of the normalization of U-235 (n,f), based on the values given in the Table, is 0.6%. This uncertainty covers the unweighted average normalization factor. A check for a dependence of the results on the time of the measurements, which in a similar analysis was shown at the '70 ANL Symposium 1 ", may still indicate some trend, however, exclusion of pre-1975 data would change the average normalization by no more than its uncertainty. TABLE I. Average Difference of Absolute Cross Section Values from the Evaluated Data Set of U-235(n,f). All Data Since and Including the Experiment by White. Dependent Data were Lumped Together. Average Average Deviation Uncertainty * First Author Ref. _% __% D/U . Smith 22 -7.5 6.3 -1.3 White 16 +2.6 2.6 +0.9 Kaeppeler 18 +3.0 3.0 +1.0 Kuks 23 +5.5 3.6 +1.4 Poenitz 13 -1.0 3.6 -0.3 (W/0 BND) Szabo 5 -0.2 3.0 -0.1 Barton 6 +0.4 1.5 +0.2 U. Michigan 7 -0.2 2.0 -0.1 Cance 8 -1.5 1.9 -0.8 Adamov 24 +3.6 1.6 +2.0 Heaton 25 -1.3 2.2 -0.6 (Wasson 2 -4.0 1.7 -2.3) not used Wasson See text -2.8 3.8 -0.9 (revised) Poenitz (BND) See text -1.3 2.4 -0.5 Data rel.BlO 10,15 -3.2 3.1 -1.0 Data rel.Li6 26 -4.8 5.2 -1.0 Values of D/U Probability. < ± 0.7 are expected with 50% 263 o t CM O CM I 2 =r •< ►— Oil bJ O z 3 1 1 1 1 1 BARTON I 1 1 1 MM 1 1 1 II II II MM 1 1 1 1 1 II II — CM O I— CM < I UJ o r\j o (M t — 1 • 1 1 1 1 CANCE CZIRR 1 1 1 Mill 1 1 Mill 1 1 1 1 II 1 II i i 1 1 1 n - <3> LEUGERS — • CANCE j. CM 1 100 1.00 EN/P1EV 10.0 20. Fig. 4. Comparison of experimental data with an evaluated cross section. The difference relative to the uncertainty is shown. 264 C\J o — WHITE <^ O O o CM o i O O — O ?€>- o <^ *^&?<£ o POENITZ ^4 GNQ X BND A. ACT. t— • CM << h- Ckl UJ O u z 3 z OJ o 1 ^-« »— < > a* a CM t CM O CM 1 1 1 1 M M 1 • 1 1 1 1 • INI U. MICHIGAN' GAfTHER -*«» <• O • 1 , 1 Mil 1 1 1 II MM 1 1 1 t o 1 1 1 • Mill WASSON • 1 X WASSON — _x • X • X • X 1* X • 1 1 *x $> ••i 1 1 II 1 1 1 1 KAEPPELER 1 1 1 II l l l l l ll ll l ll _ * POENITZ 77 BND 1 1 1 1 II 1 II 1 II II 1 II 1 1 1 1 1 '. 100 1.00 EN/NEV 10. 20.0 Fig. 5. Comparison of experimental data with an evaluated cross section. The difference relative to the uncertainty is shown. 265 The only data which were changed for the present evaluation were those by Wasson. The original normalization, which was based on one specific experi- mental value, was replaced by the normalization to an average value obtained from a large number of experi- ments for the 7.8 eV-11.0 eV interval (obtained by the '76 ANL Workshop 27 ). The uncertainty for the normal- ization consists of several components which are listed in Table 2. TABLE II. Accounting of Uncertainties Associated with the Wasson Data Above 100 keV. Uncertainty Source Re f . % 1. Average Stated Uncertainty (Statistical + Systematic) 1.6 2. Added Uncertainties Related to the Normalizatic Uncertainty of Norm. Integral 27 Statistical, 7.8-11.0 eV 2 Statistical, 10-20 keV, U5/Li 2 Statistical, 10-20 keV, U5/H 2 Systematic, U5/Li, eV-keV 2 Systematic, U5/H, 10-20 keV 2 Li-6 Cross Section, eV-keV 34 H Cross Section, Tens to 34 Hundreds keV Systematic Discrepancy in Overlap Range 2.4 0.8 0.8 0.6 1.0 1.6 0.5 0.7 None Added Total Uncertainty (1. and 2.) 3.8 % Compcuvl&cm with OtheA Evataatloni In Fig. 6 two other evaluations are compared with the present evaluation result. The recent eval- uation by Konshin 20 represents the data base very well. The proposed ENDF/B-V 21 is generally lower be- low 1 MeV and higher above 1 MeV than the existing data base would require. The deviation is substan- tial below 1 MeV and caused in part by placing un- justified emphasis on a preliminary data set. It was specifically suggested by the '76 ANL Workshop that the use of preliminary data should be restricted or avoided. The upper part of the figure shows a com- parison with the ±3% band considered at the '76 ANL Meeting. practical importance beyond their implications in high resolution differential measurements 31 . The knowledge of U-235 (n,f) is backed by a sufficient number of independent techniques and results. Re.quAJiejm£wti fan FitfuAe UeaiuAemeitfi, Choi.ce. o{ Technique* A sensitivity study based on the present level of un- certainty and the number of independent techniques and experiments involved, lead to the requirement thac any new measurement should have an uncertainty of less than 2%. Any difference which is found in respect to the average of presently available data should be backed up by an absolute D/U of at least 2 in order to be significant. Table 3 summarizes the status of present techniques for obtaining absolute values above 100 keV. The table suggests that the preferable techniques will be those involving the determination of absolute masses and efficiencies rather than the reliance on normalization at low energies. The pres- ent status and expectations for future improvements of techniques indicate that improvements in the cross section can only be expected from a large number of independent measurements. TABLE III. Present Status of Experimental Uncertain- ties for Absolute Values Above 100 keV. Technique, Reference Quoted Uncertainty I 3.6 2.5 2.3 1.7 3.6 Improvemnt % Calibrated Bath, Absolute Mass Determination Poenitz(13) 1972,74 Szabo(5) 1970,73 Heaton(25) 1976 Davis (7) 1976 Associated Activity Poenitz(13) 1972,74 Associated Particle, Absolute Mass Determination Szabo(5) 1970,73 2.5 Kuks(23) 1973 3.8 Cance(8) 1976 1.9 Hydrogen Recoil, Absolute Mass Determination White(16) 1965 2.3 Kaeppeler(18) 1972 2.1 Barton(6) 1972,76 1.5 Conclusions Changes oj the U-235 [n, |Q Cuobi Section At the panel discussion of the 1970 Helsinki Confer- ence, R. F. Taschek pointed out 28 that one can draw a fine line between the data by White 16 and other English 9 and LASL measurements" which would be con- sistent with all but one data set. The presently evaluated average cross section is ^ 7-8% lower in the hundreds of keV range and a further lowering by ^ 1% might be expected. StoutuA oj{ U-235 [n.jj) Above. TOO keV The fact that no data exist which conflict with the weighted evaluated data set for U-235 (n,f) suggests that the evaluated uncertainty represents a reasonable estimate of the real uncertainty. The evaluated un- certainty is less than 2% below 2 MeV, less than 3% below 8 MeV, and increases to 5% at 20 MeV. Fluctua- tions superimposed on the average cross section are probably less well established but should have little Black Neutron Detector, Mass by ct-Counting Poenitz(ll) 1977 2.0 Low Energy Normalization Wasson 1976 3.8 1.5 2.8 The Question ofa the Need jon TutuAe U-235 Me.a&uA.ejme.nt& Nuclear data request lists usually state a 1% uncer- tainty requirement for U-235 (n,f) up to 14 MeV. In the light of these requests and the fact that the present uncertainty is ^ 2-3%, the need for continued measurements is obvious. However, economics and the considerations mentioned above suggest that we look also at the limitations involved in the application of the U-235 (n,f) cross section at its present uncer- tainty level. The major application is its use to 266 r\) o CM I J****®**^ C> O o^o^ oo<^<$<$§^ i 00 $ eval. konshin J 1 I I 1 1 1 II 1 1 ENDF/B-V 1 1 1 1 1 II 1 1 1 1 1 1 1 II o (SSsg^tf***^ *«w^ «¥V — 1 J L 1 1 1 1 1 II — 1.00 EN/MEV 10.0 20.0 Fig. 6. Comparison of other evaluated cross sections with the present re- sult. The difference relative to the uncertainty is shown. convert measured ratios to cross sections. Such ratios, their present uncertainty and importance are Ratio Importance to U-235(n,f) Uncertainty for k eff Pu-239(n,f) ^3% High U-238(n,f) < 2% Medium U-238(n,y) •}; 5% Medium U-233(n,f) - 3% High Th-232(n,f) Medium Th-232(n,Y) Medium Pu-2A0(n,f) Low Pu-2Al(n,f) Low Others This indicates that for U-238(n,f) and Pu-240, Pu-241 (n,f) the knowledge of U-235 (n,f) is sufficiently precise; and for other values, the uncertainty of the ratios is probably an equal or dominant factor. The workshop at the '76 ANL meeting considered the question whether absolute measurements should be made for Pu-239 (n,f) directly, rather than measuring U-235 (n,f) and the ratio Pu9/U5. No conclusion was reached at that meeting, but considerations mentioned above sug- gest that absolute measurements of Pu-239 (n,f) are more profitable for the near future than further absolute U-235 (n,f) measurements. Rz-zvaluation o£ U-235{nA] A re-evaluation of U-235 (n,f) for ENDF/B-V is recom- mended. This is based not so much on any new measure- ment but on the fact that ENDF/B-V does not represent the existing data base. The true cross section is un- known and the presently suggested version of ENDF/B-V may actually be closer to the true values than the weighted average of the data is. However, such assump- tion is purely speculation. ACKNOWLEDGEMENT This work was supported by the U.S. search and Development Administration. Energy Re- 267 References 1. Proc. of NEANDC/NEACRP Specialists Meeting on Fast Neutron Fission Cross Sections, Argonne National Laboratory, ANL-76-90 and Supplement, (1976). 2. 0. A. Wasson, p. 183 of Ref.l, (1976). 3. A. J. Deruytter and C. Wagemans, J. Nucl. Energy _25, 263 (1971). 4. B. Leugers et al. , p. 246 of Ref.l, (1976). 5. I. Szabo and J. P. Marquette, p. 208 of Ref.l, (1976) ; also see other references cited therein. 6. D. M. Barton et al., p. 173 of Ref.l, (1976) , Nucl. Sci. Eng. 60, 369 (1976). 7. M. C. Davis et al. , p. 225 of Ref.l, (1976). 8. M. Cance and G. Grenier, p. 237 of Ref.l, (1976). 9. H. T. Heaton II et al. , p. 333 of Ref.l, (1976). 10. K. D. Zhuravlev et al. , Proc. of Int. Conf. on Interactions of Neutrons with Nuclei, U.Lowell, ERDA, CONF-760715 (1976). 11. W. P. Poenitz, to be submitted to Nucl. Sci. Eng. , (1977). 12. J. B. Czirr and G. S. Sidhu, Nucl. Sci. Eng. 57, 18 (1975), Nucl. Sci. Eng .58, 371 (1975). 13. W. P. Poenitz, Nucl. Sci. Eng. 53, 370, (1974). 14. D. B. Gayther, Proc. of Conf. on Nuclear Cross Sections and Technology, Washington, NBS Spec. Pub. 425, (1975). 15. R. Gwin et al. , Nucl. Sci. Eng. 59, 79, (1976). 16. P. H. White, Nucl. Energy A/B 19 , 325, (1965). 17. W. P. Poenitz and P. Guenther, p. 154 of Ref.l, (1976). 18. F. Kaeppeler, Proc. of Panel on Neutron Standard Reference Data, Vienna, IAEA, p. 213 (1974). Appendix 1- Evalaatzd VaXa The evaluation of a best value cross section should include all absolutely measured cross sections and all ratios in an "consistent data set" evaluation proce- dure (Ref.19). It also should extend to average cross sections at lower energies. Such evaluation is in progress and will be published elsewhere. The evalua- tion for the present purpose was restricted to recent data, to energies above 100 keV and to U-235(n,f), thus the validity of the values given below is re- stricted as well. E/MeV a/b E/MeV a/b E/MeV a/b 0.100 1.587 0.900 1.178 5.800 1.050 0.125 1.510 0.940 1.215 5.900 1.073 0.150 1.454 0.960 1.230 6.000 1.096 0.175 1.412 1.000 1.226 6.200 1.180 0.200 1.346 1.050 1.209 6.400 1.255 0.225 1.326 1.100 1.210 6.800 1.484 0.250 1.297 1.200 1.213 7.000 1.550 0.275 1.274 1.400 1.221 7.400 1.671 0.300 1.256 1.500 1.233 7.800 1.742 0.350 1.225 1.700 1.254 8.000 1.758 0.400 1.200 1.900 1.271 8.500 1.758 0.450 1.178 2.000 1.274 9.000 1.744 0.500 1.156 2.250 1.263 10.000 1.734 0.550 1.149 2.500 1.242 12.000 1.741 0.600 1.143 2.750 1.220 13.000 1.892 0.650 1.135 3.000 1.206 14.000 2.054 0.700 1.130 3.500 1.160 15.000 2.111 0.750 1.128 4.000 1.147 16.000 2.117 0.800 1.129 4.500 1.110 17.000 2.051 0.820 1.134 5.000 1.079 18.000 2.004 0.850 1.147 5.500 1.039 20.000 1.967 AppeMcU. x II: CiiXe.iia (, 01 V/U- LimiX&tlonh Int. Conf. on Nuclear Data for IAEA Vol.11, Panel Discussions, , G. Ferguson, Proc. Phys. Soc. 19. W. P. Poenitz, Proc. of Symposium on Neutron Standards and Flux Normalization, Argonne National Laboratory, AEC Symp. Series 23, p. 331, (1970). 20. V. A. Konshin, Review and Evaluation of the U-235 Fission Cross Section, Intern. Nucl. Data Committee, INDC(CCP)-94/U, (1976). 21. M. R. Bhat, p. 307 of Ref.l, (1976). 22. R. K. Smith et al., Bull. Am. Phys. Soc. 2, 196 (K4), 5704, revised by G.E. Hansen et al. (1969). 23. I.M. Kuks et al., Conf. on Neutron Physics, Kiev, Vol. 4, 18, (1973). 24. W. M. Adamov et al., Conf. on Neutron Physics, Kiev, Vol. 4, 21, (1973). 25. H. T. Heaton II, p. 333 of Ref.l, (1976). 26. J. B. Czirr and G. S. Sidhu, Nucl. Sci. Eng. 60, 383 (1976). 27. B. R. Leonard and 0. A. Wasson, p. 452 of Ref.l, (1976). 28. R. F. Taschek, Sec. Reactors, Helsinki, p. 929, (1970). 29. W. D. Allen and A.T A70 , 573 (1957). 30. R. L. Henkel, LA-2122 (1957), B.C. Diven, Phys. Rev. 105, 1350, (1957). 31. C. D. Bowman et al. , p. 270 of Ref.l, (1976). 32. CD. Bowman, Proc. of Symp. on Neutron Standards and Flux Normalization, Argonne National Labora- tory, p. 246, AFC Symp. Series 23, (1970). 33. G. R. Hale, priv. communication to G.Lamaze (1976). 34. G. R. Hale and L. Stewart, private communication (1977). and comparing economic or technical models) . This ap- proach is presently under investigation and the results will be published elsewhere. The two criteria stated in the text are crude approximations based on the more commonly expected behavior of experimental variables. Appendix III: Racmt ANL ENV ReMitti, The recent experimental values shown in Fig. 2 are listed below. Until their final publication these val- ues must be considered preliminary, though no change is anticipated at the present time. E/MeV a/b Aa/mb E/MeV a/b Aa/mb E/MeV o/b Aa/mb Establishing criteria for the existence or nonexistence of problems of any cross section can be obtained by simulation techniques (as for example used in testing 0.215 1.342 50 0.929 1.158 35 3.756 1.096 25 0.225 1.311 38 0.969 1.214 25 3.955 1.095 27 0.235 1.309 34 0.988 1.203 25 4.153 1.069 29 0.245 1.297 29 1.088 1.201 38 4.351 1.068 30 0.255 1.294 27 1.138 1.209 25 4.449 1.065 31 0.265 1.283 27 1.185 1.209 29 4.396 1.083 23 0.275 1.225 26 1.236 1.199 29 4.618 1.075. 22 " 0.285 1.246 27 1.288 1.221 26 5.179 1.028* 22 ' 0.295 1.257 27 1.336 1.217 26 5.536 1.026 24 0.305 1.255 26 1.388 1.190 25 5.618 1.033 28 0.193 1.354 37 1.433 1.225 26 5.714 1.049 27 0.213 1.342 30 1.486 1.208 26 5.809 1.032 27 0.233 1.304 26 1.578 1.247 26 5.898 1.073 30 0.253 1.297 26 1.778 1.245 29 6.006 1.087 33 0.271 1.261 23 1.978 1.262 28 6.107 1.154 32 0.293 1.254 28 2.178 1.262 28 6.185 1.161 31 0.684 1.138 23 2.222 1.286 28 6.384 1.260 36 0.735 1.144 23 2.472 1.224 28 6.579 1.387 30 0.793 1.119 22 2.718 1.201 26 6.803 1.515 35 0.814 1.126 22 2.968 1.150 26 7.025 1.541 37 0.835 1.151 28 3.162 1.134 26 7.478 1.713 54 0.850 1.135 23 3.262 1.137 25 7.875 1.795 59 0.871 1.151 24 3.360 1.134 26 8.275 1.793 62 0.893 1.155 24 3.463 1.133 26 0.909 1.154 24 3.658 1.093 27 268 PROPAGATION OF UNCERTAINTIES IN FISSION CROSS SECTION STANDARDS IN THE INTERPRETATION AND UTILIZATION OF CRITICAL BENCHMARK MEASUREMENTS 3 > b C. R. Weisbin and R. W. Peelle Oak Ridge National Laboratory Oak Ridge, Tennessee 37830 USA This work explores the constraints imposed on the 23b U(n,f) standard (proposed ENDF/B version V) by information deduced from clean integral measurements and demonstrates how uncertainties in fission cross section standards propagate in an uncertainty analysis and interpretation of those experiments. The question of what a significant improvement in the accuracy of the 235 U(n,f) standard would accomplish is addressed in the limited context of analyses of GODIVA and JEZEBEL measurements . The CSEWG integral benchmark results and uncertainties were updated in accordance with more recent information. Sensitivity coefficients were developed and used to estimate calculated re- sults which should be obtained using the subsequent release of 238 U(n,f), 235 U(n,f), and 239 Pu(n,f) at version V status. Covariance files were evaluated and processed for all important cross sec- tions with the sole exception being inelastic scattering for all levels and the continuum. Uncertainties due to the 235 U(n,f) standard were estimated to comprise more than half of the calculated uncertainty for criticality and ( 28 0f / 25 afY. spectral index in JEZEBEL as well as GODIVA; though the JEZEBEL assembly contained no 23 'U. We are not able at this time, to pre- dict criticality or \ 28 a c / 28 af) c to anywhere near the accuracy obtained by direct measurements, and therefore the integral results are significant to our analysis capability. Inclusion of integral information from GODIVA and JEZEBEL in an adjustment procedure was effective in reconciling all parameters other than^8 af /25 a A measurement in JEZEBEL for which current cal- culation and measurement are in disagreement. The adjustment procedure made changes of less than one standard deviation in the cross sections for 235 U(n,f), 235 U(n,y), 238 U(n,f), 238 U()1,y), and 239 Pu(n,f) including an increase of ^1.5% for the 235 U(n,f) cross section above 1.3 MeV. This specific adjustment result could change with inclusion of inelastic covariance files and must be viewed cautiously at this time. (criticals, cross section, ENDF/B, fission, standards, uncertainties) I. Introduction It is well known that if uncertainties in fast reactor performance parameters were to be developed solely from consideration of uncertainties in dif- ferential nuclear data, and even assuming all methods approximations to be negligible, the resulting un- certainties in predicted fast reactor performance would still be significantly larger than those re- quired by designers. 1 Thus, in addition to the customary calculation/experiment comparison, data adjustment schemes have been developed 2 h which can incorporate information from both evaluated dif- ferential data and relatively clean geometry, fast integral data. The latter generally lacks energy resolution but provides bounds within which the reactor performance predictions should lie. In general, the cross section adjustment schemes mini- mize a quadratic function of the weighted difference between the measured and computed values of the performance parameters and between the reference and adjusted cross section values. The fitting procedure provides an estimate of the accuracy of the adjusted cross sections and of the reactor properties calculated using them. 2 " 4 Although the standard deviations of the adjusted cross sections may not be significantly smaller than those assumed for the un- adjusted cross sections, performance parameter pre- dictions can be significantly improved because of the large effect the adjustment has in altering the off- diagonal_elements of the cross section covariance matrix. 2 " 4 It is important to note, however, that the adjusted covariance matrices depend upon the initial estimated accuracy for both the differential and integral measurements. Adjustment of the nuclear data base, and associ- ated uncertainty analysis, for improvement of per- formance parameter prediction has sometimes been received with considerable skepticism. In part, this was because adjustments have been made without detailed consideration of the differential data co- variance information. 3 This was not at all a simple error of omission. The evaluation of differential cross section covariance matrices is a formidable task 5 involving correlations of evaluated cross sec- tions with cross sections in other energy ranges and reaction types. Evaluated data which is derived (e.g., by deducing the elastic cross section from the total and non-elastic data) or evaluated through measure- ments relative to standard cross sections introduce additional complexity. Although the level of effort required is considerable, the rational assessment of uncertainty information for ENDF/B files was considered of high urgency 6 to improve upon existing analyses by permitting systematic sensitivity investigations to propagate uncertainties in a credible fashion. In a recent paper, we presented the first results of our uncertainty analysis for fast reactor bench- marks. 7 The FORSS system 7 was employed to compute sensitivities 8 in 126 energy groups using transport theory. Multi group covariance files were developed for 238 U( nj f), 238 U(n, Y ), 239 Pu(n,f), 239 Pu(n, 7 ), and 239 Pu(v) using evaluations generated at ORNL 9 and formats and procedures established for the ENDF/B Research performed at Oak Ridge National Laboratory for Union Carbide Corp. under contract with U. S. Energy .Research and Development Administration. Notice: By acceptance of this article, the publisher or recipient acknowledges the U. S. Government's right to retain a non-exclusive, royalty-free license in and to any copyright covering the article. 269 system 10 for the definition and processing of evalu- ated and correlated energy dependent uncertainty information. Using the measurements in ZPR-6/7 for k and central 238 U capture/ 239 Pu fission with assigned uncorrelated one standard deviation (la) uncertainties of 1 and 2 percent resulted in (la) predictions for the same parameters of 0.8% and 1.8% when the integral data was included in a cross section adjustment procedure which made changes of less than one standard deviation to the basic multigroup file. This technique was sub- sequently applied 11 to a realistic two-dimensional model of an LMFBR, including several reaction rate ratios measured in ZPR-6/7 and ZPR-3/56B. The (la) uncertainties in predicted multiplication factor and breeding ratio of a 1200 MWe LMFBR were reduced from 3% and 7.2% (differential nuclear data only) to 0.9% and 3.2% after inclusion of the information from CSEWG benchmark assemblies. In all cases, the cross section covariance files referred to the average cross sections in an infinitely dilute situation. The only cross section reaction correlation considered was for 239 Pu(n,f) and 239 Pu(n,Y) as developed from measurements of their ratio. Later work 12 in the analysis of the TRX-2 thermal lattice required covariance file evalua- tions for the four low energy resonances of 238 U (i.e., covariance of r n and Ty), and the thermal region. How- ever, even with this expanding data base, none of the studies above gave detailed consideration to the effects of correlated uncertainties resulting from measurements relative to standards. (It should be noted that a more comprehensive covariance file is expected with the 1978 release of ENDF/B-V.) A file of cross section uncertainty information for use in reactor performance uncertainty analysis should take into account the propagated effects of uncertainties in the standard cross sections used. 13 In particular, this applies to the 235 U(n,f) stan- dard above a few hundred keV; the 239 Pu(n,f) and 238 U(n,f) cross section measurements, among others, are often made relative to 235 U(n,f). The purpose of this paper is to explore the constraints imposed on the 235 U(n,f) standard by information deduced from clean integral measurements and to demonstrate how uncertainties in fission cross section standards pro- pagate in an uncertainty analysis and interpretation of those experiments. The question of what a significant improvement in the accuracy of the 235 U(n,f) standard would accomplish is addressed in the limited context of analyses of GODIVA and JEZEBEL measurements. II . Data Testing Assemblies Description The CSEWG Benchmark Specif i cations 11+ contain a recommended calculational model as well as pertinent experimental results for the GODIVA and JEZEBEL assemblies. 15 ' 16 The GODIVA model is a bare sphere of enriched uranium metal, having a core radius of 8.74 cm and atomic densities of .045/. 002498/ .000392 atoms/barn- cm for 235 U/ 238 U/ 231 *U, respectively. Similarly, the calculational model for JEZEBEL is a bare sphere of Plutonium metal, having a core radius of 6.385 cm and atomic densities of .03705/. 001751/. 00011 7 atoms/barn-cm for 239 Pu/ 2l+0 Pu/ 21tl Pu, respectively. The experimental results for the performance param- eters studied are given in Table I; the bracketed quan- tities refer to reaction rate ratios measured at the center of the sphere and are effective microscopic quantities, densities divided out. It should be noted that the experimental values for the ( 28 af/ 25 af\ ratios have been modified from the CSEWG recommendations (0.205 for JEZEBEL and 0.156 for GODIVA) in accordance with the recent recommendations of Hirons. 17 The uncertainties in the 49 af/ 25 af Table I. Accurate Measurements and Associated Uncertainties Have Been Reported for GODIVA and JEZEBEL GODIVA JEZEBEL k 1.00+0.003 < 28 o f / 25 o f ) c 0.16+0.005 a <* 9 o f / 25 o f ) c 1.42+0.071 b ( 28 o c / 28 o f ) c 0.47+0.02 k 1.00+0.003 < 28 o f / 25 o f ) c 0.21+0.008 3 <" 9 o f / 25 o f ) ( . 1.49+0.075 b The experimental values for the ( 28 o f / 25 oi ratios have been modi- fied from the CSEWG recommendations (0.205 for JEZEBEL and 0.156 for GODIVA) in accordance with recent recommendations of Hirons. 17 The uncertainties in the ( M o f / 2S o f ) ratios have been increased 18 from ^2% to ^5% (la) due to revised uncertainty estimates -associa- ted with the deduction of the number of fissions through the e- counting of "Mo. Macroscopic central reaction rates, denoted by < > , have been divided by respective densities; i.e., all ratios are "microscopic" reaction rate ratios, o is the capture (n,y) cross, section; a, is the fission (n,f) cross section. have been increased from ^2% to ^5% (la) due to revised uncertainty estimates 18 associated with the deduction of the number of fissions through e-counting of 99 Mo. III. Analyses of GODIVA and JEZEBEL: Generation of Sensitivity Coefficients The cross section library employed in these calcu- lations was a 126 group processed MINX/SPHINX library 19 generated from ENDF/B-IV data at 0RNL. Group constants were developed simultaneously in CCCC, 20 AMPX, 21 and MATXS 22 formats to allow for self shielding, prepara- tion of user libraries for transport codes and creation of a data base for the F0RSS sensitivity profile modules. The ANISN code 23 was used to generate regular and generalized fluxes and adjoints in the S 16 , P 3 , 40 spatial mesh interval description provided in the CSEWG specifications 14 and to compute the desired responses. The volume integrated flux spectrum for GODIVA is illustrated in Fig. 1. Previously determined correction factors 24 to account for higher order trans- port approximations were applied and the results are VOLUME INTEGRATED FORWRRD FLUXES FOR GOOIVB ~i m m — m — r~n — rr FLUX 6.7036E 00 i~n — r 126 GROUPS / / I I I J i l I J Ml III III III I Fo 1 * "o 2 Fig. 1. The Flux Spectrum in GODIVA Is Applic- able for Testing 235 U(n,f) Cross Sections in the Fission Source Energy Range. 270 listed in Table II along with results of other labor- atories participating in the CSEWG data testing effort 24 and the experimental uncertainties from Table I. It is important to note that the revised recommendations for the / 28 af/ 25 oAexperimental results 17 change previous GODIVA overprexliction into reasonable agreement but makes the JEZEBEL discrepancy even worse. The plutonium fission ratio is underpredicted in the JEZEBEL assembly consistent with underprediction of critical ity there. This somewhat complicated series of results is not the final picture. With the coming issue of ENDF/B-V, several of the principal cross sections are due for revision. We assess this effect in a subsequent section by making use of the sensitivity profiles developed from ENDF/B-IV as discussed below. Table II. Calculation/Experiment Data Testing with ENDF/B-IV Indicates Discrepancies Particularly for JEZEBEL Table IV. Relative Sensitivity of JEZEBEL Performance Parameters to Various Cross Section Reaction Types < 2 V 25 °f»c <*V"v e ORNL Experimental Uncertainties (*: 1°) Reaction Relative n b Relative Sensiti vi ty b n Relative Sensitivity Sensi tivi ty u Reactio Reactio 1,9 v 0.967 2 S 1.0 "o, 1.011 "»f 0.729 "o f -1.0 "°f -1.0 "\, n 0.082 "»„.-■ ■ d -0.192 "\ n' ,d -0.022 40 ^ 0.031 4 9- n,n' ,c -0.151 h \ n' ,c -0.014 ""n.n-.d 0.027 "S 0.101 H \ n -0.011 ". f 0.023 " 9 o n.n -0.078 l,9 o n,n' ,c 0.012 " 9 V 0.054 h \ -0.008 c n,n' n,n ,d >c 0.012 -0.010 -0.007 JEZEBEL k eff 0.992 0.996 0.992 0.3 < 2 v 2 W 0.908 0.922 0.905 3.8 ("V'S/C/E 0.934 0.936 GODIVA 0.933 5.0 k eff 1.005 1.007 1.003 0.3 K^Vc/E 1.031 1.035 1.031 3.1 <^°f/ 25 °f)c/E 0.971 0.970 0.972 5.0 < 2 V 28 °f>C/E 0.937 0.954 4.2 a This is percent change in response per percent change in cross section uniformly over all energy. b c f is fission [(n,f) MM 8] o is capture [(n,y) MT=102] o , is total discrete inelastic when x=d (MT=4); discrete inelastic to n,n ,x level n when x=n (MT=51-90); total continuum cross section when x=c (HT=91 ) . The fission spectrum temperature is not included in this list; otherwise rank- ing of reactions is in order of importance. C/E is the ratio of calculation to experiment for the quantity in brackets. The (C/E)'s for ( 28 o,/ 25 o f ) are relative to the experi- mental values in Table I which updates the CSEWG recommendations changing previous GODIVA overprediction into reasonable agreement but making the JEZEBEL discrepancy even worse. Energy dependent sensitivity profiles were calcu- lated with the JULIET module 7 of the FORSS system. Space- energy-and angle integrated sensitivities for each of the performance parameters in GODIVA and JEZEBEL are presented in Tables III and IV for many of the cross sections of interest. Comprehensive libraries 8 of energy dependent coefficients in a computer retrievable for- mat 22 ' 8 have been documented and released for distribu- tion by RSIC and NNCSC. As examples, Fig. 2 illustrates the sensitivity the 235 U fissio sensitivity pro with respect to dashed line is 235 U(n,f) at hi the 238 U(n,f) c peaks near Z38 U below, competes Figure 4 presen in GODIVA with tion. Increase direct effect o cross section a parameter defin profile of k for n cross section, file for the 238 U the 235 U fission positive, i .e. , an gh energies effect ross section, thus tio [the solid lin (n,f) threshold an only with the 238 ts the sensitivity respect to the 23 ~ s in 238 U(n,f) cle n this performance ppears explicitly i tion . The net co GODIVA with respect to Figure 3 presents the capture/fission ratio cross section. The increase in ively competes with increasing the e is negative; it d at all energies U capture rate]. of { 28 a f / 25 aA c U fission cross sec- arly have a positive parameter since the in the performance ntribution to the flux O2-16-77C0DIVA AREA- 6.S985E-01 Table III. Relati to \ ve Sensltlv arlous Cros ty of GODIVA Performance Paran Section Reaction Types Kters 1 <*■', < 2 \)c CV"« > c (' '.',.. 2 Vc Reaction Relative Sensitivity* Reaction b Relative Sensitivit Reaction" Sens tivity" Reac „ n b Relative Ser.sitwity "I 0.983 *8 a 0.998 ->o f 1 00 2.„ r 0.998 °t 0.660 c f -0.819 "•f "° 979 °f -0.,997 2S 0.113 -0.269 ""n.n'.d -« 031 2S n . .d 0.38S 2 "n.n'.d 0.062 25_ n,n^ -0.164 ""n n' c "° 013 25 n "f -0.266 "°c -0.037 zs o -0.087 "°n n "° on 2S n , iC 0.230 ""n.n'.c 0.014 »» 0.068 ♦0 0059 2S °n n 0.126 28" 0.0097 2 5 0.034 »v -0.096 2S a n.n',2 0.0083 ""n.n'.d -0.026 Zi a -0.0S4 2< *v 0.0083 2 So n.n',2 -0.009 Z8 °n n - id 0.037 29 0.0073 ""n.n'.c -0.009 2S °n n ! ,2 0.016 "f 0.0068 "°n.n -0.006 26 a -. ,c 0.013 n.f O.00S7 :9 o " .3 0.009 0.007 10 ° co JO" 3 - z - LU 5 ; This 1s percent change In response per percent change 1n cros b o f Is fission [(n.f) HT-18] s sect Ifor mly over all energy. c c is capture [(n, T ) HT-102] °n n' x ls toUl discrete Inelastic when x-d (HT=4); discrete inela stlc o le vel n when x-n (MT=S1-90); total contlnuim cross section when x=c (HT-91). The fission spectrun temperature is not Included in this list; of importance. other wise ..H ng of reactions is in order 5 10'' .-I Mini I i i i i nil , i I i i i i i in iTF 5 s id"' s 10 ° s 10 ENERGY (HEV) Fig. 2. Relative Sensitivity Per Unit Lethargy of in Section. k e ff in GODIVA with Respect to the 235 U(n,f) Cross" :tior 271 02-22-77 GODIVB U-238 CRPT / 0-238 FI5 U-23S FI5 I fiREfl- -2.6S70F.-01 10 °p Fig. 3. f)c Cross Section. <28 0c/2 . 10" 2 s 10" ENERGY (MEV) in GODIVA with Respect to the ; Lethargy 235 U(n,n ie present study, we've taken McKnight's five group itimated cross section changes for 235 U(n,f) and the estimated cross section changes" 239 Pu(n,f) and have projected the impact on all of the GODIVA and JEZEBEL performance parameters con- sidered in this work. The cross section changes proposed were +1.44, -0.24, -2.57, -3.32, -2.13 percent change in 25 af relative to ENDF/B-IV for 3.679-10.000, 1.353-3.679, 0.498-1.353, 0.183-0.498, 0.067-0.183 MeV energy regions. Similarly, the per cent changes for 49 af in the same energy ranges were -1.07, +0.83, +1.02, -0.30, and -2.13, respectively. In addition to McKnight's tabulation 25 we have in- cluded the proposed changes associated with 238 U(n,f) for ENDF/B-V. For the three highest energy groups, we estimate -2.5, -3.5, and +12% respectively (high- est energy group first). As a result of lowering the 235 U(n,f), the 238 U(n,f), and raising (slightly) the 239 Pu(n,f) cross section, the results on the whole are brought into much better agreement. Critical ity, for both systems, is underpredicted as is the ( 28 °f/ 25a f) c ratio in JEZEBEL. Other performance parameters are close to or within one standard deviation of the experimental value. Table V summarizes the pro- jected calculated results (ENDF/B-V 235 U, 238 U, and 239 Pu fission) and measurements for GODIVA and JEZEBEL. Table V. ENDF/B-IV, Projected ENDF/B-V and Measured GODIVA and JEZEBEL Performance 02-22-77 GGOIVR U-238 FIS / 1>-23S FIS U-236 FJS 1*0 BR£fl- 9.9783E-01 IU >- CO 5 ,' "(_ u_ IE [' i UJ 10 i 1 ! I — , J •- t — t 5 t .' ZD - rr - i UJ , J IX. 10 •a - J y- I— > - J t — i ,J h- ,1 0") in ■3 L_ I .' iij U) " f 1 ,1 UJ i U_ i i i i nil i i i mill i i i i 1 1 nl i i Ji i 1 1 nl i 1 1 1 1 1 1 1 1C "I s 10" 3 s 10" J s 10"' 5 10° s 10 ENERGY (MEV1 Fig. 4. Relative Sensitivity Per Unit Lethargy of the V 28a f/ 25 °f)c in GODIVA with Respect to the 238 U Fission Cross Section. effect is essentially negligible with additional fis- sion neutrons increasing both relative 238 U and 235 U fission rates in roughly the same proportion. IV. Changes to GODIVA and JEZEBEL Calculated Performance Associated with ENDF/B Re-Evaluation to Version V At the May, 1976, ENDF/B-V uators made "first cut" estimat might be made in the cross sect isotopes [ 235 U(n,f), 238 U(n, Y ), V relative to ENDF/B Version IV committee then made projections sensitivity coefficients of the reactor criticals analysis (e.g sequently, McKnight and Poenitz question with alternative evalu 1976 ANL Fast Neutron Fission C ENDF/B-IV "ENDF/B-V" (Updating 238 U, 235 U, 239 Pu Fission) Experimental Uncertainty JEZEBEL k eff 0.9920 0.9944 < 2 V 25 °f>C/E 0.905 0.892 < M V 25 °f>C/E 0.933 GODIVA 0.949 K eff < 2 V 2 S>C/E <*V 25 °f>C/E < 2 V 2 S>C/E 1.0033 1.031 0.972 0.954 0.9937 1.010 0.990 0.991 0.3 3.8 5.0 0.3 3.1 5.0 4.2 V. Evaluated "Pointwise" Co variance Files Task Force meeting, eval- ions for what changes ions for the principal 239 Pu(n,f)J for Version The Data Testing Sub- based upon available possible impact on fast . , ref. 7, p. 71). Sub- 25 re-examined this ations discussed at the ross Section Meeting. For 272 Covariance files have been evaluated for 2 238 U(n, Y ), 235 U(v), 238 U(n,f)/ 235 U(n,f) ratio, n' ,1st), 238 U(n, Y ), 238 U(v), 239 Pu(n,f)/ 235 U(n, 239 Pu(n, Y )/ 239 Pu(n,f) ratio, 239 Pu(v), 240 Pu(n, 21+0 Pu(v), 241 Pu(n,f), and 2ttl Pu(n, Y ). The eval were based primarily on "external" methods of a which examine the scatter among existing data s These sets are assumed to represent fairly the tistical ensemble of hypothetical sets of measu which could have been obtained in the experimen form our present data base. The "pointwise" co files were represented on convenient energy gri believed to be adequately fine to reproduce the range behavior important for estimation of unce in integral quantities. 35 U(n,f), 238 U(n, f) ratio Y). uations nalysis ets. sta- rements ts which variance ds broad rtainties The source of data for new uncertainty files evalu- ated for 28 af/ 25 af and 1 * 9 af/ 25 of was the recent com- pilation prepared by W. Poenitz. 26 In the method used, weights were assigned to each experiment which were estimated to reflect the reciprocal variance of a typical point from the data set. Since uncertainties in most sets are a function of energy, an overall judge- ment was used. The evaluated ratio for 28 o^/ 25 o^ W as taken to be the proposed evaluated fission ratio for ENDF/B-V; 27 i.e., the ratio was used which, when mul- tiplied by the 235 U fission cross section for ENDF/B-V gives the proposed 238 U fission cross section for ENDF/B-V. For the tt9 a^/ 25 o f fission ratio, the ENDF/B-V evaluated ratio was not in hand early enough so the evaluated ratio had to be taken from ENDF/B-IV. Only in these two recent fission ratio evaluations was a distinction made between the ensemble of hypothetical measurements and the ensemble of evaluations based upon these measurements. Accounting for this apparently straightforward difference introduced calculational com- plexities because of the varied pattern of measurements made by individual authors and because of unequal weights assigned. In practice, these files for the ratios were group averaged and subsequently combined with multigroup covariance files based upon existing 235 U data 12 (properly weighted). In general, the remaining files were obtained as indicated in the SUR report, 9 essentially applying the definition of the covariance matrix directly to com- piled sets of experimental cross sections. Additional information relating to the pointwise covariance files can be found in ref. 7. In cases for which only a few data sets exist or could be compiled, the ensemble vari- ances were statistically poorly determined by the small sample; however, variance fluctuations over small energy regions may be unimportant after averaging over the assembly spectrum. It is also important to recognize that such "external" covariance evaluation methods do not include any systematic bias; for example, the possibility that the 239 Pu half life be uncertain to 2% could systematically affect al "absolute" ratio measurements requiring exact foi weights. Finally, in all cases, the uncertainty refer to the infinitely dilute average cross sect S, v Of <£„■/ c <*V 2 S>c K eff < 2 V 25 °f>c <*V 2 S>c < 2 V 2 S>c ENDF/B-V 'Calculation" Uncerta All Nui inty :lear 1%: la) Data Uncertainty (%: 1o) [ 235 U(n,f) Assumed Known Exactly] JEZEBEL 0.9944 1.5 . 0.5 0.188 1.3 0.3 1.414 GODIVA 0.4 0.3 0.9937 1.7 1.0 0.163 1.5 0.7 1.401 0.4 0.3 0.466 14.3 14.0 in discrete and continuum inelastic scattering could add significantly to the projected uncertainties in Table 12. This is particularly true for the thres- hold reaction rate ratios, { 28 °f/ 25 af) c anc ' / 28 a c / 28 of^, and is also important for the uncertain- ties in k e ff. In addition to the lack of covariance file data for inelastic interactions in 235 U and 239 Pu, covariance files were not constructed for the fission spectrum shape. This might also have made a signif- icant contribution. Clearly, the uncertainties listed in Table 12 are probably lower estimates. However, the numbers presented are themselves quite interest- ing. The most startling result obtained is probably the 0.4% uncertainty for the ( 1+9 af/ 25 af) c ratios, partic- ularly in view of the ^3% uncertainties for each of the reactions (Table VI and Table X). However, there is good reason for this low number. Recall that the 239 Pu(n,f) was evaluated through a ratio measurement relative to 235 U(n,f). Thus, the direct component of the 235 U(n,f) uncertainty vanishes to first order. The large number of rather independent measurements tend to reduce the estimated uncertainty on the evaluated ratio to MD.5%. In addition, combining uncorrected data over the spectrum of the assembly reduces the uncer- tainty further. For the(* 28 a.f/ 25 a.f) c ratio, the situation is not as clean. As shown in Table III, the indirect effect of 235 U(n,f) on the GODIVA ( 28 a f / 25 o f ) c spectral index is consTderable. [The 235 u(n,f) uncertainty would can- cel if the sensitivity for 238 U(n,f) and 235 U(n,f) were equal and of opposite sign.] In addition to the direct effect, which would correspond to a sensitivity of -1, there is an approximate +0.2 indirect effect so that the sensitivity to 238 U(n,f) [0.998] is not equal in magnitude to the sensitivity to 235 U(n,f) [-0.819]. Thus, a component of the 235 U(n,f) covari- ance remains. In the case of JEZEBEL where the sensitivities to 238 U(n,f) and 235 U(n,f) are +1 and -1 respectively (and thus, the 235 U contribution should cancel ) there is a reasonably large sensitivity to 239 Pu(n,f) which in turn depends strongly on 235 U(n,f). The resulting uncertainty in <^ 28 af/ 25 of) c in JEZEBEL is then comprised largely of the uncertainty in 238[j/235(j fission ratio (small) and the uncertainties arising from 239 Pu from 235 U(n,f). The nuclear data uncertainties in k e +-f, approximately 1-2 percent, depend on. many factors but 235 U(n,f) is important. Finally \ 28a c / 28 °^ c cannot be computed with high con- fidence due primarily to the large uncertainties assigned to 235 U capture at fission spectrum energies. 274 The third column of Table 12 indicates that elim- inating all uncertainty in the 235 U(n,f) cross section could significantly improve our confidence in calculated critical ity and ( 28 of/ 25 °-f\ c • This remark must be qualified until such time as uncertainties in 235 U and 239 Pu inelastic and in imprecision in the fission spec- trum temperature have been included in the analysis. Hence, the significance of increased precision in the 235 J(n,f) cross section on our analysis capability for GODIVA and JEZEBEL is limited by the lack of covariance file components for other cross sections in the system and, of course, on the reliability of our estimates of uncertainties in cross sections evaluated for this analysis. Inclusion of the measurements (and their tainties) in GODIVA and JEZEBEL, listed in Tab in a data adjustment scheme 2 which included si taneously the seven integral parameters led to new set of calculated results (and uncertainti listed in Table 13. The integral measurements a pronounced ejffect^gn N the resulting uncertain since in these cases 3 °f)< Derm for k e ff and (~° a d assigned integral experiment uncertainties are siderably smaller than those projected to be d nuclear data. It is interesting to note that optimal adjustment could do little to resolve Table 13. The Adjusted Data Set Combines Information from Both the Differential and Integral Data uncer- le I, mul- the es) have ties the con- ue to the the ENDF/B-V "Calculation" Adjusted Data Set Reported Measurement Uncerti on Cal( Usinq linty (%: lo) :ulated Result Reported Measurement Adjusted Set JEZEBEL k eff 0.9944 0.9999 0.3 < 2 V"°f>C/E \ V "WE 0.892 0.949 GODIVA 0.901 0.950 1.1 0.3 eff 0.9937 0.9999 0.3 < 2 V"°f>C/E <*V 2 S>C/E ?V 2 S>C/E 1.010 0.990 0.991 1.019 0.991 0.998 1.2 0.3 4.0 ( 28 a f / 25 a.p\ in JEZEBEL; increases in this ratio tended to disturb the current agreement of the same ratio for GODIVA. (This might not have been true had the 239 Pu and/or 238 U inelastic files or the covariance of the fission spectrum shape' been included.) Since the experimental ratio of ratios [( 28 of/ 25 afJ G0D T VA / ( 28 cf/^ 5 of)jEZEBEL] is more surely well known," the discrepancy in one of these ratios would appear to be real and of necessity be related to the calculated flux spectrum. The adjustment was characterized by a x 2 /degree of freedom of 1.55; there is a ^20% probability that x 2 /degree of freedom is at least that large if the assigned integral and differential errors are correct. This minor overall inconsistency cannot be given much weight until uncertainties are included for all important quantities. The actual adjustments made in this study are less significant since the adjustment is very sensi- tive to additional degrees of freedom which would be provided with the inclusion of the fissile material inelastic covariances (which may be difficult to estimate) and the uncertainty on the fission spectrum shape. For the record, however, only the cross sec- tions for 235 U(n,f), 235 U(n, Y ), 238 U(n,f), 238 U(n, Y ), and 239 Pu(n,f) were modified in any significant way, and at most, these adjustments were 0.4-0.6 of one standard deviation. This particular adjustment would suggest that the 235 U(n,f) cross section above ^1.3 MeV be increased by %1.5%; below 500 keV an additional 0.4% reduction was suggested. These suggested adjust- ments cannot be divorced from associated adjustments to the other four cross section type reactions; the adjustments must yet be confirmed by review and exten- sion of the current covariance files before use in associated analysis. Conclusions Uncertainties due to the 235 U(n,f) standard were estimated to comprise more than half of the calculated uncertainty for critical ity and \ 28 of/ 25 °f) c spectral index in JEZEBEL as well as GODIVA, though the JEZEBEL assembly actually contains no 235 U. We are not able, at this time, to predict criticality or ( 28 o c / 28 af) c to anywhere near the accuracy obtained by direct mea- surements, and therefore the integral results are significant to our analysis capability and, particu- larly in the case of criticality, could eventually lead to improvement in our knowledge of the 235 U(n,f) cross section. Inclusion of integral information from GODIVA and JEZEBEL in an adjustment procedure was effective in reconciling all parameters other than the <^ 28 af/ 25 af)c measurement in JEZEBEL. The adjustment procedure made changes of less than one standard deviation for the 239 Pu and 235 U fission and capture cross sections including an increase of ^1.5% for the "ENDF/B-V" 235 U(n,f) cross section above 1.3 MeV. This specific adjustment, as well as the strength of the JEZEBEL( 28 of/ 25 of) c discrepancy, could change with inclusion of inelastic covariance files and must be viewed cautiously at this time. Acknowledgements The authors take pleasure in acknowledging the significant contributions of R. Q. Wright and J. L. Lucius in the generation of cross sections, transport results, and sensitivity profiles. E. M. Oblow and J. H. Marable offered valuable insight on the validity of the sensitivity profiles. The guidance of F. G. Perey in the generation of multigroup covariance files was essential as was the efforts of J. D. Drischler in processing of such data and in carrying forth the data adjustment calculated. The authors appreciate the advice of G. Hansen with respect to the measured un- certainties in GODIVA and JEZEBEL. Many of the original covariance files generated by F. Defillipo were developed from information generated by R. B. Perez, G. deSaussure, R. Gwin, and L. Weston. The authors gratefully appreciate the careful review of this paper performed by J. H. Marable, G. deSaussure, and E. M. Oblow. Timely information important to this study per- taining to the proposed ENDF/B-V evaluations was graciously provided by L. Stewart and W. Poenitz prior to its formal general release. Finally, this paper ^ would not be nearly as readable without the careful typing and organization provided by Virginia Glidewell and would probably not exist at all without the financial and moral support of P. B. Hemmig and F. C. Maienschein. 275 References 1. See, for example, the review of E. Kiefhaber, "Evaluation of Integral Physics Experiments in Fast Zero Power Facilities," Advances in Nuclear Science and Technology , Vol .8 (1975) . 2. A. Gandini, "Nuclear Data and Integral Measure- ments Correlation for Fast Reactors," Parts I, II, and III, RT/FI(73)5, RT/FI(73)22, and RT/FI(74)3 (1973 and 1974). 3. C. G. Campbell and J. L. Rowlands, "The Relation- ship of Microscopic and Integral Data," Nuclear Data for Reactors, Proc. Int. Conf. 2nd, Vol. II, p. 391 (June 1970 )• 4. J. B. Dragt, J. W. M. Dekker, H. Gruppelaer, and A. J. Janssen, "Methods of Adjustment and Error Evaluation of Neutron Capture Cross Sections: Ap- plication to Fission Products," Nucl . Sci. Eng., 62, 117-129 (1977). 5. S. M. Zaritsky, M. N. Nikovaev, and M. F. Troyanov, "Nuclear Data Requirements for the Calculation of Fast Reactors," INDC(CCP)-17U, Int. Nucl. Data Committee, IAEA Nucl. Data Sect., Karntner Ring ll.A-1010, Vienna (1971). 6. 7. W. H. Hannum, "Projections for the Future of Reactor Physics," Luncheon address at the ANS National Topical Meeting on New Developments in Reactor Physics and Shielding (September 1972). C. R. Weisbin, J. H. Marable, J. L. Lucius, E. M. Oblow, F. R. Mynatt, R. W. Peelle, and F. G. Perey, "Application of FORSS Sensitivity and Uncertain- ty Methodology to Fast Reactor Benchmark Analysis," ORNL/TM-5563 (ENDF-236) (December 1976). J. H. Marable, J. L. Lucius, and C. R. Weisbin, "Compilation of Sensitivity Profiles for Several CSEWG Fast Reactor Benchmarks," 0RNL-5262 (ENDF- 234) (March 1977). F. C. Difillipo, "SUR, A Covariance Files," ORNL/ data in this report were with G. deSaussure, R. B Weston, and R. W. Peelle the procedures described Perey, G. deSaussure, an Data Covariance Files fo - Examples for 235 U and Physics, Design, and Eco Program to Generate Error TM-5223 (March 1976). The derived from discussions . Perez, R. Gwin, L. W. ; the principles behind are discussed in F. G. d R. B. Perez, "Estimated r Evaluated Cross Sections 238 U," Advanced Reactors : nomics , Edited by Kallfelz 10. and Karam, Pergaraon Press, p. 578 (1975). F. Perey, Format Modifications 73-7, minutes of the CSEWG Meeting, December 1973 (Enclosures 6 and 12), S. Pearlstein, Editor, Brookhaven Na- tional Laboratory; see also, F. G. Perey, "Estimated Uncertainties in Nuclear Data - An Approach," Proceedings of a Conference on Nuclear Cross Section and Technology, Vol. II, edited by R. A. Schrack and C. D. Bowman, p. 842 (October 1975). 11. J. H. Marable and C. R. Weisbin, "Performance Parameter Uncertainties for a Large LMFBR," Trans. Am. Nucl. Soc, June 1977 (to be published). 12. E. T. Tomlinson, G. deSaussure, and C. R. Weisbin, "Sensitivity Analysis of TRX-2 Lattice Param- eters with Emphasis on Epi thermal 238 U Capture," Research Projecy 612 (EPRI) Final Report (March 1977). 13. R. W. Peelle, "Uncertainties and Correlations in Evaluated Data Sets Induced by Use of Standard Cross Sections," Nuclear Cross Sections and Technology, Vol. II, p. 173, edited by R. A. Schrack and C. D. Bowman, Proceedings of a Conference, Washington, D.C. (March 1975). 14. "ENDF-202 Cross Section Evaluation Working Group Benchmark Specifications," BNL-19302 (ENDF-202) (November 1974). 15. G. E. Hansen and H. C. Paxton, "Reevaluated Critical Specifications of Some LASL Fast Neutron Systems," LA-4208 (1969). 16. G. E. Hansen, "Status of Computational and Ex- perimental Correlations for Los Alamos Fast Neutron Critical Assemblies," Proc. of Seminar on Physics of Fast and Intermediate Reactors, Vol. I, IAEA, Vienna (1962). 17. T. J. Hirons and R. E. Seamon, "Calculations of Fast Critical Assemblies Using ENDF/B-IV Cross Section Data," Trans. Am. Nucl. Soc, 22, 722 (1975). 18. G. Hansen, private communication (March 1977). 19. C. R. Weisbin, R. W. Roussin, J. E. White, and R. Q. Wright, "Specifications for Pseudo- Composition Independent Fine-Group and Composi- tion-Dependent Fine- and Broad-Group LMFBR Neutron-Gamma Libraries at ORNL," ORNL-TM-5142 (ENDF-224) (December 1975); see also, J. E. White, R. Q. Wright, L. R. Williams, and C. R. Weisbin, "Data Testing of the 126/36 Neutron-Gamma ENDF/B-IV Coupled Library for LMFBR Core and Shield Analysis," Trans. Am. Nucl. Soc, Z3, 507 (June 1976). 20. B. M. Carmichael, "Standard Interface Files and Procedures for Reactor Physics Codes, Version III,' LA-5486-MS (February 1974). 21. N. M. Greene, J. L. Lucius, L. M. Petrie, W. E. Ford, III, J. E. White, and R. Q. Wright, "AMPX: A Modular Code System for Generating Coupled Multigroup Neutron-Gamma Libraries from ENDF/B," ORNL/TM-3706 (March 1976). 22. Committee on Computer Code Coordination, Working Group Meeting, Los Alamos Scientific Laboratory (May 1976). 23. W. W. Engle, Jr., "A User's Manual for ANISN, A One-Dimensional Discrete Ordinates Transport Code with Anisotropic Scattering," K-1693, Computing Technology Center, Oak Ridge Gaseous Diffusion Plant (1967). 24. E. M. Bohn, R. Maerker, B. A. Magurno, F. J. McCrossen, and R. E. Schenter, editors, "Benchmark Testing of ENDF/B-IV," ENDF-230, Vol. I (March 1976). 25. R. D. Mcknight and W. P. Poenitz, "The Effect of 235 U(n,f) and 239 Pu(n,f) Changes on CF-252 and ZPR6-6A and -7 Averages," ANL memorandum to participants on the CSEWG Meeting. Oct. 27-28, 1976, illused October 27, 1976. 26. W. P. Poenitz and A. B. Smith, Editors, "Pro- ceedings of the NEANDC/NEACRP Specialists Meet- ing on Fast Neutron Fission Cross Sections of 233 U, 235 U, 238 U, and 239 Pu," ANL-76-90 (June 1976). 276 27. W. Poenitz, Argonne National Laboratory, private communication (March 1977). 28. C. R. Weisbin, E. M. Oblow, J. Ching, J. E. White, R. Q. Wright, and J. D. Drischler, "Cross Section and Method Uncertainties: The Application of Sensitivity Analysis to Study Their Relationship in Radiation Transport Benchmark Problems," ORNL- TM-4847 (ENDF-218) (August 1975); see also, RSIC Code Collection PSR-93. 29. F. G. Perey, "Final Draft of the Formats Manual for Files 31, 32, and 33 for ENDF/B-V," letter to R. J. LaBauve, March 1, 1977. 277 237 Np AND 238 U AS POSSIBLE STANDARDS FOR THE MeV REGION S. Cierjacks Institut fur Angewandte Kernphysik Kernforschungszentrum Karlsruhe, F.R. Germany The aspects of using the fission cross sections of 237 Np and 238 U as possible stan- dards in the MeV region are considered. In comparison to other neutron standards their application is particularly advantageous for experiments involving white-source techniques. Major distortions in fast neutron measurements due to frame-overlap problems and contribu- tions from slow neutrons events can be avoided by spectrum cut-off at threshold energies. The present data basis for both nuclei is discussed and critically examined. Some suggestions are made of how to achieve an ultimate accuracy of 2 % with measurements employing 237 Np or 238 U as secondary standards. ( 237 Np, 238 U(n,f) as neutron standards, summary of experimental results, E = 0.1 - 20 MeV) Introduction Recent developments in fast reactor design, fast reactor safety, waste disposal projects and in fusion reactor technology have turned the interest in accurate neutron data to a substantial extent towards neutron cross sections in the MeV region. As experienced large- ly in the eV and the keV-range absolute flux measure- ments necessary for absolute cross section determina- tions are difficult to perform, so that the use of a cross section standard turnes out to be most preferable also in this range. Among the presently internationally recommended six standard cross section of H(n,n), 6 Li (n,a), 10 B (n,a), 12 C (n,n), 197 Au (n, Y ), and 235 U (n,f) there are only two, H (n,n) and 235 U (n,f), which are suitable for general application in the MeV re- gion. For the other four reactions either a resonance structure of increasing complexity occurs or cross sec- tions drop-off very rapidly with increasing energy ; both facts excluding them as useful standards. The use of the two remaining neutron reactions, though accept- able in principle, is also not fully satisfactory in this range. The difficulty for the application of the H(n,n) cross section is mainly connected with the fact, that the use of pure hydrogen is difficult and that the use of solid radiator foils creates either threshold problems and/or background problems if no telescope- like flux counters are used, or suffer from counting statistics due to the low efficiency of counter tele- scopes. Involving the fission cross section of 235 U instead, causes complications due to the appearance of structure in the cross section up to the several hun- dred keV range or due to background problems occurring from small impurities of slow neutrons. The latter fact is mainly of importance in experiments with white neu- tron sources and high pulse repetition rates. In the present paper, therefore, two possible new standards for the MeV region are proposed. In connection with such a proposal the recent assessments in determining accurate cross sections for these nuclei are discussed. Possible Standards in the MeV-Region In general an ideal standard reaction should meet the following requirements: i. The cross sections of the reactions should be large and accurately known over the energy interval of interest. ii.The cross section should smoothly vary with neutron energy. iti .Good samples should easily be preparable, target material with the required chemical and isotopic purity ought to be readily available. iv. A suitable neutron detector based on the standard reaction should have good overall properties, such as high efficiency, high stability, low sensity against other than neutron irradiations. For use with time-of-flight devices also a good timing characteristic is important. In addition the above general requirements another pro- perty becomes highly desirably if a standard is to be used in the MeV range and/or with white neutron sources and broad continuous energy spectra. v. The reference cross section process should be a threshold reaction, since MeV cross sections are ty- pically small compared with eV- but also keV- cross sections. Small admixtures of slow-neutrons thus may introduce large distortions from small im- purities of slow neutrons. Another favourable aspect in this context is that such a standard could di- rectly be used in connection with the threshold method. The latter has favourably been employed in a number of recent fission cross section ratio mea- surements 1 ) . In relation to our criterion (i) fission cross sec- tions are reasonably high compared with other partial cross sections in the MeV-range. From the large number of fissile isotopes only a small number of nuclei is of interest for our purpose, since highly radioactive ma- terial and isotopes occuring with low abundance in the isotopic mixture are to beexcluderi from the considera- tion. On this basis there remain except from 235 U mainly three nuclei and their fission reactions: 232 Th, 238 U and 237 Np. From these 232 Th turned out to be less suitable in view of the pronounced structure effects ob- served recently at Saclay in most of the energy range between 1-2 MeV as a consequence of isomeric fission in this nucleus. In how far structure effects to a les- ser extent play also a role in the fission cross section of the remaining two isotopes will be discussed in the next paragraph. Considering the criterion iii, 237 Np and 238 U favourably fulfill this requirement. The mono- isotopic element 237 Np is nowadays widely used in medi- cine as an energy source for heart pace-makers. Thus 237 Np material with high chemical purity is easily available. Likewisely highly enriched 2 8 U can easily be obtained from nuclear industry. Suitable fission foils and fission detectors are standard production of various laboratories in different countries. From the viewpoint of detector availability, fis- sion reactions have several advantages in the sense mentioned under item iv: Gas scintillation chambers as well as ionization chambers have high efficiencies, good stability and good timing characteristics to be used for measurements in the 1 nsec range. Their sen- 278 sitivity against Y _ra y s and charged particles can be largely reduced. Since suitable detectors can be run in approximately 2n or 4 it -geometry , their sensitivity against anisotropy effects is small as long as suffi- ciently thin fission foils are employed. Status of Present Data 238 U Fission Cross Sections Ratios Relative to 235 U. Structure and Fluctuations in 7 Np, S U As mentioned in the previous paragraph a pro- nounced structure in the fission cross section can occur in nuclei exhibiting subthreshold neutron fis- sion. To a lesser extent fluctuations can also be ob- served as a consequence of statistical fluctuations of average level spacings and level widths in an energy range of strongly overlapping resonances. Struc- ture of the first type can be caused by vibrational levels in the second wall of the double humped fis- sion barrier, where the total width might be smaller than the spacing of class-!I states. Since the damping width increases rapidly with excitation energy, gross structure .effects should be minimum for nuclei having about the same barrier height for the first and the second well with a deep minimum within between. An in- spection of the systematics of fission barrier para- meters - e.g. that given by Michaudon 2 ) - shows that the above condition is best fulfilled for the uranium isotopes and neptunium. The thorium isotopes have about the same height for the inner and outer well, but only a rather shallow minimum within between. For nuclei starting from plutonium, the height of the outer barrier is systematically lower than that of the inner one; this effect increases with increasing target mass. Structure effects connected with isomeric fission should therefore be minimum in the range of 238 U and 237 Np. Fluctuation due to statistical effects in the matrix elements determining the neutron reaction can also be observed provided the fission cross section is measured in a sufficiently high resolution experiment. As demonstrated by Bowman 3 ) such random fluctuations can be predicted with resonance parameters obtained from measurement in the resolved resonance region. Bowman has also shown that the well documented fluctu- ations in 235 U in the several hundred keV region can be well understood in terms of Adler-Adler 1 * > or R-ma- trix resonance parameters. For 237 Np and 238 U very high resolution measure- ments have recently been carried out at the Saclay laboratory. The result for 237 Np obtained by Plattard et al. 5 ) with a resolution of 0.3 ns/m in the range from 25 keV - 2 MeV is shown in Fig. 1. The data ex- hibit over the whole energy range a structureless fis- sion cross section. All oscillations are purely due to counting statistics of the measurement. The data show also that no intermediate resonance occurs near 0.3 MeV as assumed by Brown et al. 6 ). In Fig. 2 a new unpublished measurement for 23e U of Blons and cowor- kers 7 ) is shown. This measurement which was also car- ried out at the Saclay linac with a 0.3 ns/m resolu- tion reveals substantial fluctuation in the whole ener- gy region from threshold to ^ 4 MeV. Unexpectedly large fluctuation amplitudes exceeding partially 10 % of the average fission cross sections are being ob- served. These fluctuations might have a significant in- fluence on many of the past and future measurements in this range because the resolution and point spaces be- come important factors in the measurements. If this observed structure turns out to be real it becomes doubtful to some extent, whether 238 U would be a good standard, since one of the major requirements, that a standard should possess a smooth excitation function, is no longer fulfilled. The status of the cross section ratio relative to 235 U has only recently been reviewed at the 1976 ANL Specialists Meeting on Fast Neutron Fission Cross Section of U and 239 Pu 6 ). Since only little new infor- mations was obtained in addition during the meantime, mainly a brief summary of the conclusions from the Argonne Meeting will be given here. Present data avail- ableforthis nucleus have been summarized in the Pro- ceedings of the Meeting 8 ). In Fig. 3 the more recent data sets obtaind since - say about 1970 - are shown. Some new data not available at the ANL Meeting have been included. Fig. 4 shows a selection of most of the older data available from CCDN Saclay. In these graphs seve- ral relative measurements have been renormalized at 2.5 MeV to the corresponding Pbnitz 9 ) value. With respect to the ratio data the Working Group on Ratios concluded at the time of the Argonne Meeting, that a large number of experiments can be brought now into what one would consider an excellent agreement in the energy region from threshold to 10 MeV. This is achieved after some energy changes and after some adjustments for possible mass changes. It was expected by the Group that a new evaluation of the ratio data would yield ratio numbers of as good as 2 %. Above 10 MeV data becomes more sparce. There appear to be discrepancies in the white source data. Here are presently two cyclotron and two linac measurements extending to 20 MeV and beyond. Between cyclotron and linac measurements there occurs a diver- gence above ^10 MeV, showing increasingly higher ra- tios for the cyclotron measurements with increasing energy. The difference approaches about 10 % at 20 MeV. This discrepancy is not yet well understood, in parti- cular since the agreement below 10 MeV is so well in the same measurements. One possible source of the dis- crepancy might be due to the use of both ion chambers and gas scintillation chambers in cyclotron and linac measurements . Absolute Fission Cross Section of J U Figs presently Centre at are shown all old c figures a curve of and absol surements tron ener shows tha agreement 13 MeV, a 2 MeV. . 5 and 6 compare available from th Saclay. In Fig. 5 , while Fig. 6 con ross section measu re shown together KEDAK 3 10 ). No adj ute values were ma were arbitrarily gies. A more detai t a large portion over most of the n exception being the fission c e NEA Nuclear the more rec tains mainly rements. The with the newe ustments of e de, although normalized at led inspectio of the data i energy range the threshold ross sections Cross Section ent data sets a collection of data in both st evaluated nergy scales several mea- particular neu- n of the data s in rather good between 2 and region below Also after adjustment of energies due to the ob- vious energy shifts in several data sets there seem to exist further significant differences in the overall shape in the threshold region. Excluding the threshold region it can be expected that a simultaneous evalua- tion of ratio and shape measurements would produce fis- sion cross sections with an accuracy of better than 3 %. in the range below about 10 MeV. Above this ener- gy fission cross section measurements are more scarce than ratio determinations relative tO' 235 U. Excluding discrepant very old measurements of Katase and keeping in mind that the measurements of Wilson were not inten- 279 ded to be precise determinations of the 235 U fission cross section, there exist only two cyclotron measure- ments in addition which agree rather favourably with each other within 5 % (compare Figs. 5 and 6). unfor- tunately, however, this agreement is misleading since no linac measurements are available in this range. As long as we do not know the sources for the discrepancy between the linac and the cyclotron measurements in the ratio determinations we cannot exclude that the absolute cross sections have systematic uncertainties of as much as 10 % between 10-15 MeV as well as the ratios 4 ". 237 Np Fission Cross Sections Ratios Relative to 235 U. There have only very few ratio measurements of 237 Np relative to 235 U been made in the past. Only two data sets are presently available from CCDN-Saclay. These values together with a new measurement of Behrens et al. 11 ) from the Lawrence Livermore Laboratory are shown in Fig. 7. The latter data which have been ob- tained with the threshold method agree favourably with the 1967 White data by better than 2 % below 10 MeV. The additional data point given for 14 MeV deviates from the LLL results by as much as 10 %. The third data set of Stein turns out to be on average 4 % higher than Behrens' ratios. Despite the general good quality of the Livermore results for fission cross section ratios, on- ly restricted conclusions might be drawn from these 237 Np ratio measurements. To judge the present data ba- sis, one has to rely also on the numerous absolute fis- sion cross section determinations. Absolute Fission Cross Sections of 237 Np In contrast to the few ratio measurements a large number of absolute or shape measurements exists for 237 Np. The latter fact reflects its importance as a threshold detector. Existing data contained in the CCDN library are shown in Figs. 8 and 9. As in the case of 235 U the data are plotted separately with respect to new and old results. Here the END-FBI\_curve is inclu- ded in both graphes as a reference for interrelations of measurements in the figures. It should be noted that the fission data are given on a double logarithmic scale. No adjustments of nnssible enerqy and mass changes were made .This spread is mainly due to the fact, that some rather discrepant 14 and 2-3 MeV values exist for this nucleus. The absolute fission cross section mea- sured in the 14 MeV range are listed in Table I. On a first view these data appear to have very large scat- tering. But, if we exclude the 1960 Pankratov results but take his new 1963 values instead, and remove all clu uata with unassigned uncertainties (Rago, Henkel), there remains only a severe discrepancy with Iyer's value. All remaining determination are then consistent with the newest and most accurate Of- value of Adamov (2.43 + 0.043 b). A similar, only slightly better situ- ation exists in the 2-3 MeV region (Table II). If we exclude for the same reasons Henkel 's measurements and the very old Klema data, all remaining results are + i>lote added in proof: at the time of the Conference a preliminary evaluation of Pbnitz 12 ) became available, which is consistent with above accuracy estimates. A comparison of "absolute" and (U8/U5)*U5(ENDF-B/IV) re- sults showed that a consistent-data-fit would have lowered the present U5 (ENDF-B/V) data between 2-13 MeV. This would support the data of Hansen et al . and the recent data of Szabo (both above 3 MeV). The difference between the "two" sets could also be resolved by lower U8/U5 values in the above energy range. This assumption would support the ratio measurements of Stein and recent data by Cance. Table I. Absolute fission cross sections of 237 Np in the 14-15 MeV range Author, Lab. Year E p (MeV) CT(b) Flux Stand. Henkel, LAS 1957 14.0 2.6 no information Pankratov, KUR 1960 14.5 15.0 2.6 + 5 % 2.62+ 5 % o>(7) Pankratov, KUR 1963 14.0 14.7 2.4 + 5 % 2.54+ 5 % a f (7) White, ALD 1966 1967 14.1 2.33+ 0.1 o>(5) Pago, NRD 1968 14.0 14.2 2.31+ ? 2.27+ ? a f (8) Protopopov, 1958 14.6 2.4 + 0.2 USSR Iyer, TRM 1969 14.1 2.98+ 0.3 a f (8) Adamov, LEN 1977 14.8 2.43+ 0.047 ass. particle cons is tent with the most accurate values of Jiacoletti of 1.59 + 2 %. Even Plattard's high value is not discre- pant, considering the large error in the absolute cross section determination. Due to the scatter in the norma- lization cross section data in Fig. 9 deviate in a broad band of ± 10%. But also if one restricts consi- deration to more recent data sets as given in Fig. 8 data scattering exceeds bv far the 3 % limit necessary to justify its use as a fast neutron standard cross section. Table II. Absolute fission cross sections of 237 Np in the 2-3 MeV range Author, Lab. Year E (MeV) o(b) Flux. Stand. Klema, LAS 1948 2.5 3.0 Henkel, LAS 1957 2.0-2.5 2.5-3.0 Schmitt, 0RL 1959 2.82 White, ALD 1966 2.25 1967 Grundl,LAS 1967 2.0-2.5 2.5-3.0 Jiacoletti, 1967 2.75 LAS Brown, LAS 1970 2.0-2.5 Kobayashi, 1973 3.5 KT0 1.45+3 % o f (5),a f (8), 1.48+3 % o T f {9) 1.48 no inform . 1.45 1.65+10 % oy(8) 1.67+0.1 a f (5) 1.652+0.05 o>(8) 1.575+0.05 1.59 +2 % o f (5) 1.62 +10 % a f (5) Davey 1.65 +0.17 Inll5 Plattard, SAC 1976 2.0-2.1 1.79+12 % a f (5) Proposal for Additional Measurements It has been demonstrated that the availability of 237 Np and 23 ll as additional neutron standards for the MeV range would be desirable. Furthermore it was shown that the fission cross sections for these data are not yet sufficiently known to carry out precise flux mea- surements in the accuracy region of < 2 t. In order to reach this long-term goal, further experimental and 280 evaluational effort is necessary. In this context the following measurements are suggested: i. Ratio measurements for Np relative to 235 U over the whole energy region from threshold to 20 MeV employing the threshold method. Resolutions might be moderate of the order of 5 %. A high accuracy for the absolute neutron energy of 1 % should be obtained. Useful measurements must aim a statis- tical accuracy of better than 2 %. Measurements can rather easily be carried out with a cyc- lotron or some of the existing linacs. ii. Additional absolute measurements of the fission cross section of 237 Np at-^14 MeV and at an energy point between 2-3 MeV. The measurements at both energies would serve as a cross check of good normalizations at either one or the other energy point. iii. Accurate shape measurements for 237 Np and 238 U between about 10 and 20 MeV with moderate resolu- tion but good statistics to establish a reliable shape curve in this range. iv. Extended high resolution cross section measurements for 238 U above 4 MeV and for 237 Np above 2 MeV with resolutions better than 0.3 ns/m. Measurements should be made relative to a smooth cross section such as H(n,n) and should extend to an energy for U at which the percent standard deviation of the cross section fluctuation decreases below 2 %. For Np the measurements should cover at least the plateau region between about 2-5 MeV. For the absolute values a 10 % accuracy should be adequate. Such investigations are necessary for U to inves- tigate the existence and the extent of the cross section fluctuations in the few MeV range. For Np such measurements serve to verify that the smooth shape of the fission cross section curve continues above 2 MeV as would be most likely. Such measurements can in principle by carried out with most of the white-source neutron facilities. Summary To summarize, the 237 Np (n,f) and the 238 U(n,f ) re- actions have several favourable characteristics which justify their use as fission standards for fast neutron cross section measurements. An inspection of the 238 U (n,f) data indicate that the numbers might already be as accurate as 3 % in the range from threshold to 10 MeV. But care must be taken with respect to fluctuations at least below 4 MeV. From this point of view 237 Np is much more favour- able, since it experiences a smooth behaviour as demon- strated in a recent very high resolution measurement. Unfortunately the fission data for this nucleus are not yet accurate enough, to permit precise cross section determinations. Above 10 MeV the data situation is un- satisfactory for both, Np and U. In this range, which gains increasing importance in connection with fusion data needs further experimental effort is necessary to achieve the ultimately desired accuracy of s 2 %. Acknowledgements The author is indebted to Dr. L. Edvardson from the NEA Neutron Data Compilation Centre at Saclay for supplying renormalized cross section data for 2i U and for providing graphs for almost all cross sections shown in this paper. The cooperation of Drs. J. Behrens. J. Blons, M. Cance, G. Grenier and S. Plattard of readily providing their data for this survey partially even before publication is gratefully acknowledged. The author wishes also to thank Dr. Derrien from CCDN Saclay for his assistance in the collection of lackincdata sets. References 1. J.W. Behrens, G.W. Carlson, NSE, to be published, Preprint UCRL-78704. 2. A. Michaudon, Neutrons and Fission, Invited Paper Int. Conf. on the Interactions of Neutrons with Nuclei, Lowell, Mass., July 1976. 3. CD. Bowman, Nuclear Data for Reactors, Conf. Proc. Helsinki, IAEA, 1970, Paper CN-26/41. 4. D.B. Adler, F.F. Adler, USAEC Report, ANL-6792, p. 695, 1963. 5. S. Plattard, J. Blons, D. Paya, Nucl . Sci.Eng.6^ (1976) 477. 6. W.K. Brown, D.R.Dixon, D.M.Drake, Nucl. Phys.A156( 1970) 609. 7. J. Blons, private communication. 8. Proc. NEADNC/NEACRP Specialists Meeting on Fast Neu- tron Fission Cross Sections, ANL 28-30 June, 1976, ed. by W.P. Ponitz and A.B. Smith. 9. W.P. Ponitz, J. Nucl. Energy 26, (1972) 683. 10. B. Goel , F. Weller, KEDAK-Evaluation File, KFK-Report 2386, Part II. 11. J.W. Behrens, J.W. Magana, C.J. Browne, Report, UCID-17370 (1977). 12. W.P. Ponitz, private communication, 1977. 13.V.M. Adamov et al . , Contribution to this Symposium. Data Basis Used in the Graphs Figs. 3,4 U-238 / U-235 Name, Symbol C0ATES Har 75 CIERJACKS KFK 76 BEHRENS LLL 75 DIFILIPP0 0RL 76 FURS0V FEI 73 C0NDE ULL 76 P0ENITZ ANL 72 P0ENITZ ANL 72 P0ENITZ ANL 72 P0ENITZ ANL 72 P0ENITZ ANL 72 MEADOWS ANL 75 MEADOWS ANL 72 MEADOWS ANL 72 MEADOWS ANL 75 CANCE BRC 76 CANCL BRC 75 CANCE BRC 77 WHITE ALD 67 Name, Symbol PANKRAT0V KUR 60 Reference Source PANKRAT0V KUR 63 75 Wash GB 7 CCDN, energy V0R0TNIK0V KUR 71 scale revised KALININ KUR 58 76 ANL 2 KFK Listing ADAMS ALD 61 75 Wash 2 591 CCDN NETTER SAC 56 76 ANL 0RL CCDN GRUNDL LAS 67 73 Kiev 3 73 FEI CCDN LAMPHERE ORL 56 76 ANL UPP CCDN KATASE KYO 61 JNE 26 483 CCDN BATCHEL0R ALD 65 JNE 26 483 CCDN JNE 26 483 CCDN Fig. 7 Np-237/U-235 JNE 26 483 CCDN JNE 26 483 CCDN BEHRENS LRL 76 Meadows 75 D CCDN WHITE ALD 67 NSE 49 310 CCDN STEIN LAS 68 NSE 49 310 CCDN Meadows 75 D CCDN Fig. 8,9 Np- 237 76 ANL 1 BRC Paper 75 Kiev 1 BRC CCDN ENDFB/IV Grenier P.C. 77 Listing KOBAYASHI KTO 73 JNE 21 671 Paper KOBAYASHI KTO 73 Reference So' AE 9 399 CCDN AE 14 177 CCDN 71 MOSCOW CCDN 58 GEN 16 136 CCDN JNAB 1485 CCDN Wetter PC CCDN NSE 30 39 CCDN PR 104 1654 CCDN KATASE PC 61 CCDN NP 65, 236 CCDN 77 UCID-17370 Report JNE 21, 671 CCDN 68 Wash 1 627 CCDN EANDC(j; EANDC(j; -26 -26 CCDN CCDN CCDN 281 Figs. 5,6 U-238 KEDAK 3 Ged. KFK-2386( 77)Report ALKHAZOV RI 74 VOROTNIKOV KUR 71 YFI 20 9 CCDN VOROTNIKOV KUR 75 75 Kiev FEI CCDN CIERJACKS KFK 76 76 ANL 2 KFK Listing BLONS SAC 77 BLONS PC 77 Computer cards WILSON LAS 67 WASH 1074 CCDN GRUNDL NBS 72 ANS 15 945 CCDN FLEROV USSR 58 AE 5 657 CCDN JIACOLETTI LAS 70 GRUNDL LAS 67 PLATTARD SAC 75 KALININ KUR 58 SCHMITT ORL 59 BROWN LAS 70 RAGO NRD 68 PROTOPOPOV USSR 58 HOCHBERG KUR 59 PANKRATOV KUR 60 WHITE ALD.66 HENKEL LAS 57 IYER TRM 69 KLEMA LAS 47 NSE 48 412 CCDN NSE 30 39 CCDN NSE 61, 477 Journal 58 GEN 16 136 CCDN PR 116 1575 CCDN NP A 156 609 CCDN HP 14 595 CCDN AE 4 190 CCDN LA-1495, 52 CCDN AE 9 399 CCDN EANDC(UK)-77 Report LA-1495, 52 CCDN BARC-474, 1 CCDN PR 72 88 CCDN 0.1 237 Np(n,f) Fig. I Fission cross section of 237 Np between 0.1 and 2 MeV Ref. 5) Jffl fiWy^ 1 * 7W 2^ S£ E (MeV) 0.1 0.2 0.3 0.5 1 282 i ub i | i n m i| ' iii |iiiiiiiii|iiiiiiiii;'i mm iii|iiiiiiiii|iii |iiiiiiiii|iiiiiiiii|iiii i 2 2 2 2 2 2 UJ U UJ Ld Ld Z) O O O O O QL Q. Q. Q. Q_ Q_ U >>••>>• oQO< c LlI (nc CZ )*P/(n flC ,)*p 8ez ; 283 0) c LU co ftaaj uiaaDDui n ■D T> iC ^ — L0 to -1 ID N. cn * :*: ct U3 CK (i IX 10 > > 3 u cn o n in > > -J) o * a a cr ry 'J 3 o o i£ :*: _j en _j Ui V n It CO i-i T- cr QL CD > it j: z z — a: _i lu hi hi ^. o t a •— >~ z lOUOI U) 1 U' aa^^oo- z; h- Z Q_ a < ) () duizzqlq; j (THJE S- uuacroaa auictd (i 1 (1 i:u da>>^: crz(j_i x. Uj U o» ►< < imilili lIlilllUilLllllllll 1 1 1 1 1 1 I 1 1 1 1 1 J I (q) >P 286 'I p 'tii|iiTitT'iT H 'i nM )i|i'iiiiii'|TTT'TTiT'[i n 'i['i'|ii') M '[T'iii'ii'i' n i'i m i'| mM i n r TTfTi P TTi H | nHHM T[rp'lH Mr [ rnM iTiT|riTi n i M |i Q_ xt ♦ _J UJ to _J D If) rr 10 (T Z UJ UJ z o; ►- I — UJ 1 CD 3 if) > c UJ I i- in ct -Ilh-IJD in in ui _i a co cr a: _i _j cr \ ^ >- o -z. o >- a co co a: o zd a 2 o o — a: ac _i LU *: it ~) CO CJ> CL > c UJ CO o (q)*o 288 o 1 "I T T 1 T o o o o o o o • < oT o 1 o o / CD in o o \ Of OT ID (DOT lT< LTl _ in in lo cl rv o do: didi/1 iv •> \ LH_>rvOO ZiLD OTJ- 1j Ci ID > * into >XOlO o >dct in — aoau o_i_iEcr Zi-_iQ.OCt HI lX-J W, CQ — h- ZaUi CT j(- \ Z: — Z. OCDQtUIUI ex o_i:tol>ouz -z:ujuj z a* u c<: a a c t x lu >- _j Lu *: in n, q: q. n; q. 3i- *: ♦"/I c LU _ o o (q)'fl 289 STANDARD INTEGRAL MEASUREMENT FACILITIES A. Fabry C.E.N./S.C.K., Mol, B-2400 Belgium The usefulness of integral measurements in standard and reference neutron fields is examined in terms of the validation of microscopic differential-energy neutron fis- sion cross section standards needed for fission reactor technology. This synthesis encompasses a summary description of the identified neutron fields and of the status of their spectral characterization, discussion of the corrections and uncertainties involved in such experiments, and an appraisal of the accuracy of integral fission cross sections, in particular at the light of interlaboratory comparisons. The signi- ficance of such integral measurements to the testing and improvement of evaluated nuclear data is illustrated by a limited confrontation with the ENDF/B IV cross section file. (Cross sections, ENDF/B, fission, integral measurements, standard neutron fields) 1 . Introduction The aim of this paper is to examine how far integral measurements in standard neutron fields may contribute to obtain accurate microscopic dif- ferential-energy neutron fission cross section standards. Other important applications of stan- dard neutron fields to fission reactor technology are discussed elsewhere - " - ^. A standard radiation field is tentatively defined- 7 as a permanent , stable and reproducible radiation field (neutron or gamma or mixed) that is characterized to state-of-the-art accuracy in terms of flux intensity and energy spectra, and spatial and angular flux distribution. Important field quantities must be verified by interlaboratory measurement. The concepts of microscopic and macroscopic integral measurements have been delineated satis- factorily in reference". In particular, a micro- scopic integral neutron cross section o is simply the convolution of a microscopic differential-ener- gy neutron cross section o(E) and of an energy distributed neutron field or spectrum, $(E). All quantities of relevance to the design, operation and safety of nuclear reactors are macroscopic integral ones : critical masses or enrichments, reactivity worths, fuel pin power and burn up, damage in structural-components, ...; all directly relate to integral reaction rates, e.g. microscopic integral cross sections of various types, in dif- ferent neutron spectra. Because differential-ener- gy neutron cross sections are not known well enough to afford sufficiently accurate predictions of reactor macroscopic properties, it has been current practice to adjust them within their uncertain- ties' i°»9i10 so as to match the results of a variety of microscopic and macroscopic integral measurements - critical masses, fission rate ratios, central reactivity worths, material bucklings, ... - per- formed in zero-power "clean" "" critical or expo- nential assemblies and even mock-up^"" of actual reactors. Differential neutron spectrum measure- ments' 12 , mostly by time-of-f light , "Li(n,a) and proton recoil techniques, are also done in such "controlled environments" (terminology as defined in- 7 ); they are sometimes taken into account ^ in the automated cross section adjustment procedures. Neutron spectra are predominantly sensitive to elastic and inelastic scattering cross sections; such nuclear data are often tested in separate neutron spectrometry experiments on simple source- driven one-material macroscopic arrangments ' -* . All in all, this semi-empirical approach lead6 to adjusted multigroup cross section sets asso- ciated to appropriate reactor physics computational codes. It is not generally claimed that such ad- justed cross sections are more accurate than unad- justed ones but they have proven successful and necessary in establishing in a timely manner design parameters for specific demonstration power plants such as LMFBRs in the 300 MWe range 1 ?. In the long term however, the nuclear indus- try will have to keep a flexible perspective rela- tively to various competitive reactor and fuel cycle options. Adjusted cross section sets deve- loped in the frame of such specific projects as the LMFBR need thus to be validated for wider applica- bility; this is because the type of approach brief- ly evocated above may introduce various fictitious correlations between the many differential nuclear data and the integral observables simultaneously analyzed : though the accuracy of predicted reactor integral quantities is improved, this is likely to result of compensating errors which are propagated in a consistent way. As far as fission cross section standards are concerned, accuracy require- ments are and will remain severe, whatever the nuclear strategy. The potential shortcomings of cross section adjustment procedures can in this case be largely alleviated if accurate integral data sensitive only to fission cross sections are considered and given adequate weight. Integral microscopic fission cross section measurements in standard neutron fields serve this purpose. Such measurements provide straightforward, relevant and general integral constraints that must be satisfied by differential data. In this paper, standard neutron fields are identified, the status of their spectral characte- rization is briefly discussed as well as the quali- ty of integral fission cross section measurements. 290 Some ratios of integral fission cross sections are largely insensitive to neutron spectral shapes and can be considered as standard integral observables, even if not obtained in standard neutron fields : this review consequently encompasses also some reference^ neutron fields less well characterized than the standard ones. Finally a quick outlook at selected integral results in the various consi- dered facilities reveals a few interesting trends relatively to the ENDF/B IV cross section file. 2. Identification of Standard Neutron Fields The definition of standard neutron fields gi- ven in the introduction does not specify what is meant by a state-of-the-art accuracy characteriza- tion in terms of the space, angle and energy depen- dent absolute flux spectrum *(E, r, S> a) . To a large degree, this vague qualification stems from the fact that the accuracy level at which a neutron field may be considered as a standard depends on the application and energy range considered. For instance, if the field is used to cali- brate or validate an instrument aimed at measuring scalar reactor neutron spectra with an energy reso- lution of 5% to a requested accuracy of +_ 5% in the energy range 10 keV - 1 MeV, there is no need for the standard to be characterized much better than according to these specifications; in this example nevertheless, the accuracy of the flux spectrum above 1 MeV and below 10 keV should be sufficient to allow adequate corrections for the eventual sen- sitivity of the instrument to higher and lower energy neutrons; furthermore, the gamma dose to neutron flux ratio should not be excessive (typical- ly, 10 rad hr'vcm'^sec constitutes an upper bound). Thus, a number of more or less subtle con- siderations are involved in deciding whether a par- ticular permanent, stable and reproducible neutron field is indeed a standard in terms of a given application. For the integral testing and potential impro- vement of differential-energy cross sections, it is the shapes of the cross sections and of the flux spectrum as a function of energy that determine the sensitivity of the integral cross sections to flux spectral uncertainties. The less rapid is the variation of a cross section, or flux spectrum, or both, with energy over the reaction response range (e.g. over the energy range contributing to 90% or so of the reaction rate) and the less stringent is the accuracy with which the spectral shape must be known. Although a +_ 1 - 2% accuracy goal for dif- ferential-energy fission cross sections means as well a +_ 1 - 2% target for the integral fission rate measurements and for the determination of the total absolute flux in the standard neutron fields, the corresponding needed accuracy for the flux spectral characterization is generally much less severe, of the order of +_ 5% over the energy res- ponse range; this is still a difficult target to meet, but it is manageable provided the standard flux spectrum displays little or no energy structure. A neutron field can thus be standard relatively to a given type of nuclear reaction and not relatively to another : for example, the thermal -neutron in- duced uranium-235 fission neutron spectrum is pre- sently a standard for the validation of fission cross sections, but it is not for the validation of (n, 2n) cross sections of relevance to fusion reactor technology. In the context of this paper - fission cross section standards needed for fission reactor tech- nology - five types of neutron fields only have been identified as standard. They are, in order of decreasing spectral average energy : 1. 2. 3. the californium-252 spontaneous fission neutron spectrum *g2 the thermal-neutron induced uranium-235 fission neutron spectrum, ^ ? . the intermediate energy standard neutron _fields, ISNF, a family of boron-10 tailored, carbon mo- derated neutron spectra driven by fission neu- trons ; the spectral hardness is variable within some limited energy range and the gross spectral shapes are comparable to LMFBRs 1/E or near /E shape neutron spectra Maxwellian thermal neutron spectra 3. Appraisal of Standard Neutron Fields Calif ornium-252 Fission Spectrum Facilities Intense spontaneous fission sources of vide a close approximation to an isolat tron source. Typically, this involves small aluminium-steel capsule containi of CfpO.SO, and associated, lightweigh instruments suspended in a low scatter (a large room or outdoor). Such facil been developed in recent years by a fe most noticeably NBS (USA)'8 j p TB (Germ IEP (Hungary )20. Integral fission cro measurements in the ^-J^-Zl fission neut have been performed at NBS^' 1 22 ^ an level suitable for standard integral d Cf pro- ed point neu- simply a ng a few mgr t detecting environment ities have w laboratories, any) ° and ss section ron spectrum accuracy ata. The absolute source strength is determined to + 1 %\"' by total absorption techniques (the man- ganese sulfate bath at NBS) ; current international comparisons of ^P^Cf source strengths support so far this accuracy assessment. The total absolute free field neutron flux intensity at carefully establis- hed distances from such source is deemed known to within + 1- k %. The error component due to distance assessment is typically as low as + 0- 6 % ; this is achieved"^ by simultaneous exposure of two nearly identical de- tectors on opposite sides of, and equidistant from, the source : the first-order distance error is then associated with the separation of the detectors. The maximum error component related to intrinsic neutron field perturbation by scattering in the source capsule and supports is + 0.7 %• (a) In all neutron fields discussed in this paper, $(E, r, Q) is separable and space-angle variations are well understood. (b) All uncertainties in this paper are quoted for a 68.3 % confidence interval (1c), 291 The fraction of the neutrons that undergo one ine- lastic scatter in the source capsule is indeed as low as (0.7 + 0.3) %. The only other component of error in such measure- ments relates to neutron return from the environ- ment : albedo from boundaries, irradiation support structures, source capsule and air scatter. These corrections and their uncertainties are very small : for a source-to-detector distance of 5 cm, (1.4 + 0.6) % for 2 - 5 - ? U fission with cadmium enclo- sure and (0.4 + 0.7) % for 58 U fission. The uncertainties in the determination of absolute fission rates themselves are discussed in a subse- quent section. ? _P Integral fission cross sections in the Cf fis- sion neutron spectrum are reported with an overall combined uncertainty of + 2.2 - 2.8 %, depending on the isotope. The spectral characterization of the califor- nium-252 spontaneous fission neutron spectrum is excellent over the bulk of the energy range. A recent evaluation*^ f eight documented spectrome- try measurements indicates an accuracy of + 1 to 2 % between 250 keV and 8 MeV. This is by"far the best known distributed energy neutron field to-day. If spectral uncertainties are weighted by differen- tial-energy cross sections, they can be expressed as uncertainty components of the corresponding in- tegral cross sections : for ^35jj anc i OOy fission, these components are 0.24 % and 0.95 % respective- ly *. This is a satisfactory situation from the viewpoint of the differential-energy fission cross section validation by integral measurements. Uranium-235 Fission Spectrum Facilities Integral cross section measurements in the thermal neutron induced uranium-235 fission neutron spec- trum represent an effort covering almost two de- k ?zL PS p£ cades ' ' -'' ° , but few fission cross section data have been reported and most of them are ra- tios. This is because the generation of pure uranium-235 fission neutron spectra is not straight- forward and the determination of the absolute flux is difficult. Four facilities are in operation to-day : at.KYOTO University (Japan) 2 ?, LJUBLANA (Yugoslavia)' ^28 NBS (USA) 29 and CEN-SCK (Belgium)^. Background responses in the two first ones prevent their ap- plication to non-threshold fission cross section work. The CEN-SCK facility is the most versatile and extensively investigated one ; it will thus be briefly discussed. This facility involves a one-meter diameter spheri- cal cavity within the vertical graphite thermal column of the BR1 reactor at MOL. Two types of arrangments have been developed to produce fission neutron spectra : 1. central coaxial source disc assemblies31 for passive measurements (solid state track re- corders, activation foils) : two bare discs of 2 35u, typically diameter 19 x 0.1 mm, surround coaxially a cadmium box containing the samples to be exposed (Fig. 1) ; this type of arrang- ment is also used at NBS (30 cm diameter gra- phite cavity) 2. central cylindrical source shell30 concentric to an extruded cadmium sleeve extending from TABLE I. CORRECTION FACTORS FOR BACKGROUND RESPONSES AND NEUTRON FIELD PERTURBATIONS IN THE CAVITY URANIUM-235 FISSION SPECTRUM NEUTRON FIELD MEASUREMENTS : STANDARD CYLINDRICAL SOURCE SHELL TYPE OF CORRECTION NUCLIDE TYPICAL ASSOCIATED UNCERTAINTY 235 u 2 >h 2 ^9„ Pu 241 Pu 237 Np 238 u Wall return background Photof ission , epither- mal and thermal neu- tron penetration (a) Impurity isotopes Cadmium sleeve perturbation Instrumental perturbation 0.8636 0.9907 0.999 0.991 1.000 0.7810 0.9939 0.997 0.991 1.000 0.8923 0.9840 0.999 0.991 1.000 0.7995 0. 9939 1.000 0.991 1.000 0.9975 0.9967 1.000 1.001 1.003 0.9978 0.9935 0.998 1.010 1.006 + 0.1 to + 0.3 % + 0. 1 % + 0.1 % + 1.0 % + 0.3 % Net correction 0.8472 0.7668 0.8693 0.7875 0. 9982 1 . 0056 + 1.05 % to + 1.10 % (a) For typical fissionable deposits 292 cavity bottom to reactor top : this allows fis- sion rate traverses with a variety of active (fission chambers) and passive detectors. Arrangement 1 is currently applied for absolute cross section measurements. The source strength is deri- ved 2 ? from 1 ^°Ba- La absolute radiometric coun- ting of the uranium-235 discs, using literature values for the fission yield and the number of neutrons per fission ; the accuracy is of + 2 %. The flux follows by straighf orward geometry cal- culations^. This arrangement has been chosen for international radiometric integral fission cross section measurements for 2 35u and 2 3°U. Selected high quality sources and samples will be assembled and exposed at CEN-SCK by end 1977, will be trans- ferred to IAEA for dismantling, inspection and geo- metry control, and will be subsequently shipped for counting by various laboratories in Europe, the U.S. and Japan. Table I presents, for the standard cylindrical source shell (diameter 33 x 77 x 0.1 mm 3>g) i n arrangement 2, an inventory of all corrections an un- certainties involved in cross section measurements for six important fissionable isotopes33 (excluding uncertainties in the determination of absolute fis- sion rates, to be discussed separately). The do- minating correction is for non-threshold reactions and relates to the background responses to neutrons returning from the cavity walls ; these fractional responses, of 15 to 20 % in the table, can be varied easily by modifying the diameter and height of the shell source, in a range as large as from 2 % (using miniature fission chambers-? ) to 100 %. These responses can be measured directly by moving the instruments from cavity center up to half its radius : this is because the fission flux decays rapidly with distance to the source while the ca- vity wall return neutron flux in the energy range of response for non-threshold fissionable isotopes (< 1 keV) displays the remarkable property of a relatively great space-angle unif ormity3^. Indeed, the return neutron spatial and angular distribution within the cavity would be perfectly flat if the angular flux at the vacuum-moderator boundary was perfectly isotropic. The actual patterns can be parametrically vizualized by considering the reci- procal configuration : a central point detector and a source shell of increasing radius R in a cavity of arbitrary radius R . Figure 2 displays the re- lative central wall return neutron flux ratio (£, P )/0 (E,0) as a function of the reduced w s w source radius p = R /R for three cavity sizes and various_neutron energy groups of approximate mid energy E. The short history of neutron cross section techno- logy presents here an amusing paradox : years ago, measurers of differential-energy cross sections, alert to struggle with room return neutrons in their experiments, tended to feel that integral measurers applying fission neutron sources in mo- derating cavities might well mishandle their cor- rections for wall return backgrounds ; to-day, integral measurers do not hesitate to use such background neutrons as one of the primary spectral components of intermediate-energy standard neutron fields, as outlined in the next section. The dominating uncertainties of the corrections needed in • > - > M fission spectrum integral experiments are actually the ones due to intrinsic neutron field perturbation by the materials close to the source and detectors, as shown by Table I. 2"5 5 The spectral characterization of the -'-'U thermal fission neutron spectrum is not comparable in qua- lity to the californium case. The previously con- sidered evaluation 3 f spectrometry data lists uncertainties of + 3 % to + 5 % in the energy range 250 keV - 8 MeV, and these propagate into computed integral cross section errors of 0.2^ % and 1.7 % for -^U and •* U fission respectively, e.g. about twice the 2 5 2 Cf error f or 238u fi Ss i on . This eva- luation however does not include finite sample size corrections which are important in some spectrome- try experiment&35. Although the long-standing in- consistencies" between differential and integral data in the thermal neutron induced 2 35u and ^39p u fission neutron spectra tend to be generally resol- ved - ? , there are still some discrepancies affecting nuclear reactions of importance to reactor dosime- try. These suggest 2 °'3 ^37 a slightly harder 2 35y fission spectrum than presently evaluated*^ and close to the fits of pulsed Van de Graaffs spectro- metry measuremente35. Nevertheless, the impact of these differences is small insofar as the valida- tion of fission cross sections is concerned and >2\5 raay be considered a good standard to this par- ticular respect ; the fact that it is a less accu- rate standard than >^2 is balanceu by its greater practical importance as basic source of neutrons in chain reacting systems, an argument whose weight is obvious at the light of the introductory section. Intermediate-energy Standard Neutr o n Fields The concept of primary intermediate-energy standard neutron field, contemplated since about a decade, has become reality in 1976 only when the ISNF-1 assembly was put into operation at NBS (USA), as the result of a cooperation between NBS, CEN-SCK (Belgium) and LOS ALAMOS SCIENTIFIC LABORATORY (USA). The first ISNF fission cross section mea- surements are reported at this conference-? ; the NBS-ISNF facility is also described and analyzed to some extent39. i n this section, a brief and some- what general discussion of the concept and its physics is done. ISNF does not designate a specific standard neutron field, but a family of standards derived from the same fundamental principle : controlled moderation and absorption of fission spectrum neutrons in a one-dimensional macroscopic arrangment of materials with differential-energy cross sections known to the accuracy of primary standards - at least over the energy range where they significantly partici- pate to the mechanisms of spectral generation. The last condition is crucial to the definition of primary intermediate-energy standard neutron fields, as opposed to secondary ones (exemplified by ££). This condition bears upon a theoretical predictabi- lity argument : it is required that the ISNF space, angle and energy neutron flux density be fully and accurately characterized by solution of the sta- tionary Boltzm a nn linear integro-dif f erential trans- port equation ; neutron spectrometry techniques are consequently not mandatory for spectral characteri- zation purposes, but can be validated usefully by measurements in the ISNF and contribute to the certification of the fields. Other conditions of ISNF definition such as adequate simulation of LMFBR neutron spectra, reproducibility, ... are outlined elsewhere 2 . A parametric investigation has resulted into some specialization of the con- cept : selection of boron-10 and carbon as major 293 (« R. ) , centered iq within a gra- material constituants of the systems, and restric- tion to the class of spherical geometry arrangements specified by Fig. 3. Thus the ISNF fields involve a fission spectrum source shell of radius R£ con- centric to a boron-10 shell (the absorber) of outer radius R, (< Rg) and thickness y. in a spherical cavity of radius R ( phite moderator (the reflector). In practice, the infinite reflector thickness is 50 cm, but 35 cm is satisfactory for most applica- tions, and the source shell can be replaced by a few adequately distributed point sources, either thin uranium-235 discs driven by thermal neutrons or calif ornium-252 capsules. As illustrated by Fig. h, ISNF neutron fluxes (full line) are formed as the superposition of fission neutrons (dotted line) and cavity wall return neutrons (dashed line minus dotted line, not shown) tailored by the boron-10 absorber. This super- position can be formulated semi-analytically in terms of reduced radii and thicknesses P A = Ra/ r C' P'S - Rs/ R p and °A = *a/ r c* Consider first a con- figuration in which the boron shell is replaced by a thin cadmium absorber, the so-called ISNF/CV stan- dard field-^. The central scalar neutron flux spectrum $(E,pg,Rc) f° r a fission shell source of unit total strength may be expressed as : *+TTR, (E,P S ,R C ) X(E) v w (E 'V g(E ' p s\TTlV? where ,(E) is the normalized fission spectrum, vy(E,Rc) is the wall return neutron flux spectrum for a central point fission source in a cavity of radius R c normalized so that J 0# 5 e v t ' ; y(E,R c )dE = 1 and the function g (E,Pg) is displayed on figure 2. The constant Y and the relaxation length K are characteristic of the moderator (11.37 and 8.75 cm respectively for carbon). This equation quantifies the superposition of an uncollided source spectrum and the collided component associated to the re- flector, these two basic spectral shapes being weighted by the intensity scaling factors 1/i-ig and Y Q /(1 + A/R c ) 2 . This relates the ISNF to the pro- perties of carbon wall return neutrons, which are accurately predictable by discrete-ordinates trans pling C as (i A /ps) = 0.99 to 1.0 for P A /p g = 0.5 to zero : this is a simple shadowing effect related to the decrease of the first collision source density in the carbon wall caused by the intensity resca- ling of the uncollided flux ; b) the absorber-re- flector coupling C AR ((- A ,T A (E, 6 A ) ) , that varies smoothly with energy between ~ 0.7 and unity, de- pending on p a (for p A <: 0.15, the deviation from unity is less than 5 % over the whole energy range); this expresses the fact that the probability of neutron absorption by the boron-10 shell is en- hanced by multiple cavity crossings of the collided component neutrons ; this may also be seen as a shadowing effect, but the important point is that it is governed by the intrinsic shell transmission itself. Summarizing, the ISNF physics is dominated by simple and well understood mechanisms acting in a largely decoupled way and amenable to almost text- book formulation : streaming of virgin fission neutrons in vacuum, reflective scattering of neu- trons from a point source at the center of a sphe- rical moderating cavity and transmission of neutrons by a spherical absorbing shell. 1/E and near 1/E Standard Neutron Fields The slowing-down of fission neutrons from point sources homogeneously distributed in a non- absorbing infinite moderator generates a collided neutron spectral flux that varies as the inverse of the neutron energy above the thermalization range. Such idealized way of creating 1/E neu- tron fields has unfortunately no direct implementa- tion able to compete with californium facilities when they meet the challenge of materializing a free field point source fission flux i Measurers not of resonance integrals, e.g. of *Qr G (E) 2£-, only have to be careful in assessing what their cadmium cut-off energy E^ actually is^' , how ac- curate their neutron self-shielding corrections are - to list only two major uncertainty components _ but most importantly, they have to ascertain that their integral measurements are indeed performed in a pure 1/E standard field, e.g. place realistic t theory approximations-' and MONTE CARLO methods, bounds on errors due to spectral shape deviations por The parametric survey calculations for the ISNF have shown that detailed design criteria are optimized if pg =- 1 and P^/pe <- 0.5- Under these conditions, insertion of the boron-10 absorber modifies the above equation as follows : P S from the ideal law. Such deviations may be small,, but it does not matter if they are not, provided they are accurately known. Carbon cavity wall return neutron flux spectra do not obey the 1/E law, w - w (E,R c )g(E,p, Yo (1+A/R C ) jJV^Wp? WW^V* The function T a (E,6a) is the intrinsic neutron transmission of the boron-10 shell in infinite va- cuum ; deep physical insight has been brought into this type of macroscopic neutron interaction by the famous analytical work of Bethe, Beyster and Carter ho It is this tailoring exponential -like term that drastically shapes the neutron spectrum at intermediate and low energies while the other terms essentially result in relatively small res- calings of the intensity factors for the two super- posed spectral components as identified by the pre- vious equation. For the uncollided flux, the res- caling is a transmission factor T A (6 A ) °" 0.97, al- most energy-independent from a practical standpoint. For the collided flux, the rescaling involves two coupling functions : a) the source-absorber cou- but generally speaking, and most certainly in terms of integral fission cross sections, the deviations are excessively well known, e.g. ISNF/CV type fields as defined in the previous section are relevant. They do not seem necessary from the standpoint of this review because fission resonance integra]^^ are usually considered as well known for standard isotopes and they also agree generally with dif- ferential data. This section could thus have been limited to such statement. But even keeping out of mind the various technological applications of ISNF/CV standard fields, it seem6 valuable to re- port here the integral-versus-dif f erential cross section comparison displayed in Table II ; ~ ^ and 25 (a) 1.505 (+2.2%) 1.049 4.30 (+3.0%) 0.964 3.94 (+2.0%) 0.939 0.985 JEZEBEL 1.49 ( + 5.0%) (f) 1.068 4.59 (+3.0%) (e) 0.950 4.74 (+3.2%) (e) 0.917 O.98O GODIVA (b) 1.42 (+5.0%) (f) 1.030 5.19 (+3.0%) (e) 0.985 6.21 (+3.2%) (e) 1.033 1.064 EE (C) 1.167 (+2.0%) 1.025 6.92 (+2.5%) (0.926) ( S > 17.8 (+1.5%) 0.948 0.972 CFRMF (b) 1.145 (+1.5%) 1.025 7.29 (+2.3%) 0.967 20.6 (+1.4%) 0.993 1.018 BIG-10 (b) 1.199 (+1.5%) 1.023 8.52 (+2.3%) 0.965 26.8 (+1.7%) 1.014 1.037 $(E) for computation : NBS evaluation 4(E) for computation : transport theory, ENDF/B-IV cross sections (c) 2 ^(E) for computation : reference ( A \ ^^ All experimental data taken from reference , unless otherwise indicated (e) 49 Reference . The data relate to monoenergetic calibrations at ~ 2.5 MeV where absolute ratio measure- ments were performed for the normalization ; they are backed by interlaboratory comparisons in Flat- top (f) D - 50 Reference (k) In the energy range of major response, the neutron spectrum is to be significantly revised 5. Integral data testing of selected ENDF/BIV fission cross sections 239 235 The Pu/ U fission cross section ratio belongs to the category of data qualified in section 1 of standard observables for reference neutron fields. This is maybe true also for the 2 ^ 7 Np/ 2 3 8 U ratio, but not for the 23 5u/ 2 3ou ra tio. Nevertheless, all three ratios are gathered in Table IV for an array of standard and reference neutron fields ; the Los Alamos bare metal criti- cal spheres GODIVA and JEZEBEL are well known and the other reference fields are the ones investi- gated by the already mentionned ILRR program. The ENDF/BIV cross section file was used to obtain the computed integral values. This table represents a state-of-the art confrontation between differen- tial and integral fundamental fission cross sec- tion data and illustrates clearly the major and most interesting trends. This deserves an exten- sive discussion which lies outside the scope of the present paper. References 1. A. FABRY, P. VANDEPLAS - Fast Reactor Physics I, 389, IAEA (1968) 2. A. FABRY, G. and S. DE LEEUW - Nucl. Techn. 25, 3*+9 (1975) 3. J. A. GRUNDL, C. EISENHAUER - "Benchmark Neutron 4. 5. 9- 10. 11. 12. Fields for Reactor Dosimetry" in IAEA Consul- tants' Meeting on Integral Cross Section Mea- surements in Standard Neutron Fields for Reactor Neutron Dosimetry, November 15-19 (1976) W.N. McELROY et al. - Special Issue, Nucl. Techn. 25 n° 2 (Feb. 1975) J. A. GRUNDL, A. FABRY - "Report of Workshop Session on Reactor Dosimetry Benchmarks" in First ASTM-EURATOM Symp. on Reactor Dosimetry, Petten, Netherlands (1975) J. A. GRUNDL - Proc. Symp. on Neutron Standards and Flux Normalization, p. 417, Conf-701002 (1971) The Physics of Fast Reactor Operation and De- sign, Proc. Int. Conf. BNES, June 24-26 (1969) Proc. IAEA Helsinki Conf. "Nuclear Data for Reactors" (1970) Third Conf. Neutron Cross Sections and Technolo- gy, CONF-7IO30I (197D Proc. Int. Symp. on Physics of Fast Reactors, Tokyo, October 16-19 ( 1 973 ) W.G. DAVEY, W.C. REDMAN - Techniques in Fast Reactor Critical Experiments. Gordon and Breach Science Publ. , Y.A. KAZANSKIJ, A. A. VAN'KOV, E.I. INYUTIN - Atom. En. Rev. !£, 807 (1975) 296 13. 14. 15. 16. 17- 18. 19. 20. 21. 22. 23- 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. E.D. PENDLEBURY - CONF-710301, p. 1 (197D - Nucl. Sci. Eng. 62, in Report BLG 495, p. 1-44 Nucl. Sci. Eng. A. PARVEZ, M. BECKER 571 (1977) G. and S. DE LEEUW - (1974) H. BLUHM, G. FIEG, H. WERLE 54. 300 (1974) J.L. ROWLANDS et al. - ibid. _10' 11 33 J. A. GRUNDL et al. - Nucl. Techn. 32, 315 (1977) W.G. ALBERTS et al. - NBS Spec. Publ. 425, I, 273 (1975) M. BUCZKO et al. - Int. Symp. on Calif ornium- 252 Utilization, Paris, April 26-28 (1976) H.T. HEATON II et al. - NBS Spec. Publ. 425, I, 266 (1975) D.M. GILLIAM et al. 270 (1975) NBS Spec. Publ. 425, I, NBS Spec. Publ. J. A. GRUNDL, C. EISENHAUER 425 , I, 250 (1975) A. FABRY - Report BLG 465 (1972) A. FABRY - Report NEACRP-L-140, also in AERE- R8636 (1977) A. FABRY et al. - "Reactor Dosimetry Integral Reaction Rate Data in LMFBR Benchmark and Stan- dard Neutron Fields : Status, Accuracy and Im- plications" ibid. _5 I. KIMURA et al. (1973) M. NAJZER et al. - Proc. Reactors, II, 571 (1970) J. Nucl. Sci. Tech. _10i 57 1 * Conf. Nuclear Data for J. A. GRUNDL - Proc. of "Prompt Fission Neutron Spectra" p. 107, IAEA (1972) A. FABRY, J. A. GRUNDL, C. EISENHAUER - NBS Spec. Publ. 425, I, 254 (1975) A. FABRY, K.H.CZOCK - Report IAEA/RL/27 (1974) J. A. GRUNDL - Nucl. Sci. Eng. 3J., 191 (1968) A. FABRY, I. GIRLEA - "Quality control and cali- bration of miniature fission chambers by expo- sure to standard neutron fields. Application to the measurement of fundamental integral cross section ratios" ibid. 3 Trans. ANS 15, 940 34. A. FABRY, J.D. JENKINS (1972) 235 23Q 35. J.M. ADAMS - "Comparison of ^U and >7 Pu Fast Neutron Fission Spectra" Appendix A in AERE- R8636 (1977) 36. M.F. VLASOV, A. FABRY, W.N. McELROY - Proc. Int. Conf. on Interactions of Neutrons with Nuclei, Lowell, Mass. (1976) 37. A. FABRY et al. - "Review of Microscopic Inte- gral Cross Section Data in Fundamental Reactor Dosimetry Benchmark Neutron Fields" ibid. 3_ « 38. D.M. GILLIAM - Integral Measurement Results in Standard Fields" this conf. 39- C.M. EISENHAUER et al. - "Transport Theory Ap- plications in Neutron Standards" this conf. 40. H.A. BETHE, J.R. BEYSTER, R.E. CARTER - J. Nucl. En. 3, 207 ; 3, 273 ; 4, 3 ; 4, 147 (1956) 41. S. PEARLSTEIN, E.V. WEINSTOCK - Nucl. Sci. Eng. 29, 28 (1967) 42. S.F. MUGHABGHAB, D.I. GARBER - Report BNL 325 ( 1 973 ) 43. H.D. LEMMEL - NBS Spec. Publ. 425, I, 290 (1975) 44. P.I. AMUNDSON et al. - Report ANL 7320, 6?9 (1966) 45. M. DARROUZET et al. "Studies of Unit k Lattices in Metallic Uranium Assemblies ZEBRA 8H, SNEAK 8, ERMINE and HARMONIE UK" - ibid. 10 46. M. PINTER et al. - NBS Spec. Publ. 425, I, 258 (1975) 47. D. GILLIAM, National Bureau of Standards - Private communication (1976) 48. J. A. GRUNDL et al. - Nucl. Techn. 25, 237 (1975) 49. J. A. GRUNDL, G.E. HANSEN - Proc. Nuclear Data for Reactors, vol.1, 321, IAEA (1967) 50. H. ALTER et al. - Report ENDF-202 (1974) 297 Lock spring Fig. 1. TYPICAL SOURCE -DETECTOR ASSEMBLY FOR ACTIVATION MEASUREMENTS IN THER- MAL-NEUTRON INDUCED FISSION NEUTRON SPECTRA o luJ |LtJ I.I d l0 id > 5 0.9 S |6 P 3 40 GROUP DISCRETE -OROINATES COMPUTATIONS CAVITY DIAM. 2R, 30cn 6 CAVITY DIAM. 2R C =50cm • CAVITY DIAM. 2R £ =IOOem Rj ■ RADIUS OF SHELL SOURCE F s APPROXIMATE GROUP MID-ENERGY O 0.5 1.0 REDUCED SOURCE RADIUS p = R s /R c Fig 2. SPATIAL DISTRIBUTION OF CARBON WALL RETURN NEUTRON FLUXES NEUTRON ENERGY. iV Fig.4. NEUTRON SPECTRUM COMPONENTS OF THE ISNF Fig. 3. GEOMETRICAL PARAMETERS OF THE ISNF 298 INTEGRAL MEASUREMENT RESULTS IN STANDARD FIELDS D. M. Gilliam National Bureau of Standards Washington, D.C. 20234 Measured spectrum-averaged fission cross sections are reported for several benchmark fast neutron fields. For the Cf-252 spectrum, absolute cross section measurements at NBS and the University of Michigan are compared. Cross section ratios (relative to Pu-239) are reported for U-235, U-238, and Np-237 for the following fields: U-235 fission spectrum, BIG-10, CFRMF, SIGMA SIGMA, and the Intermediate-Energy Standard Neutron Field (ISNF) at NBS. Fission product yields measured by the Interlaboratory LMFBR Reaction Rate Program ( ILRR) are reported in two categories: (1) Consensus Yields from thorough interlaboratory studies and (2) Subsidiary Yields for isotopes studied less intensively (usually by a single labora- tory). All measured yields were determined by Ge(Li) counting of fission activation foils, with specific fission rates determined by counting fissions from a separate light deposit in an ionization chamber. The fission product yields for the CFRMF and BIG-10 fields are reported first separately and then combined, to provide a single set of yields for a Fast Reactor Spec- trum. (Nuclear data; neutrons; standard fields; cross sections; fission yields) Introduction The preceding paper by A. Fabry has discussed facilities employed in standardization of measurement of neutron doses for reactor fuels and materials re- search. The present paper summarizes results from tests in many of these facilities in which there has been a direct NBS involvement in the past five years. Most of the NBS involvement in this area has been through the USERDA Interlaboratory LMFBR Reaction Rates (ILRR) program, centered at the Hanford Engineer- ing Development Laboratory. Spectrum-Averaged Fission Cross Sections Absolute Cross Sections The availability of small Cf-252 neutron sources and manganese bath facilities has made it possible to determine the neutron flux in a fission-spectrum neu- tron field so that absolute spectrum-averaged cross section measurements can be made. Table I shows the ABSOLUTE FISSION CROSS SECTIONS IN A Cf-2S2 NEUTRON FIELD Cmb) U-235 U-238 Pu-239 Np-237 NBS [Refs. 1,2) 1205 t 27 319 ± 8 1808 i 41 1332 t 37 University of Michigan (Ref. 3) 1215 * 17 " 1790 * 34 " Average of Measurements 1210 t 2.0% 319 t 2.5% 1800 ±2.2% 1332 ±2.8% Calculated Spectrum-Averaged Cross Section* 1241 31S 1789 1351 Measured/ Calculated 0.975 1.013 1.006 0.986 •Calculated cross sections were derived from ENDF/B-IV data and the neutron energy spectrum given in the NBS evaluation (Ref. 4). results of absolute fission cross section measurements made in Cf-252 neutron fields at NBS and at the Uni- versity of Michigan. The results for U-235 and Pu-239 agree within 1.0%, well within the total uncertainty estimates of either laboratory. At both laboratories, the neutron irradiations were performed with two counters positioned symmetrically about the source in order to avoid sensitivity to the positioning of the highly radioactive source. The calibration of the manganese baths at both laboratories is based on the same standard neutron sources, NBS-I and NBS-II. The mass assay of fissionable deposits at the two labora- tories have been intercompared by both fission and alpha counting methods; and in the case of Pu-239, the mass of the University of Michigan deposit was based almost entirely on comparison with the NBS reference deposit by alpha counting. The good agreement of the results is not so surprising in view of the several areas of collaboration between the two laboratories, and yet there are also significant areas of indepen- dence in the separate measurements. The fission count- ing methods were entirely different. At NBS miniature fission ionization chambers were employed; while at the University of Michigan, track-etch detectors were used. The scattering perturbations from the source capsule, the experiment support structures, and the laboratory walls were all very different at the two sites. The averages of the experimental results are compared with calculated values in Table I, also. The calculated values for U-235 and Pu-239 are not very sensitive to the assumed 'neutron energy spectrum. Thus the discrepancy of 2.5% for U-235 is primarily a disagreement of the integral measurements with the ENDF/B-IV data. Since the total estimated uncertainty in the integral result for U-235 is 2.0%, the disagree- ment cannot be considered a strong one. The measured cross sections for U-238, Pu-239, and Np-237 agree with calculated values within 1.47-, indicating a gen- erally satisfactory situation in terms of both the ENDF/B data and the evaluated neutron energy spectrum for Cf-252. Cross Section Ratios The Intermediate-Energy Standard Neutron Field (ISNF) Table II presents the results of initial measure- ments in the Intermediate-Energy Standard Neutron Field (ISNF) which has recently become operational at NBS. The fission cross section ratios for U-235, U-238, and Pu-239 were measured by back- to-back fis- sion counting in an NBS double fission ionization chamber at the center of the ISNF facility. _The most accurate measurement of the ratio a f (U-238)/5 f (U-235) 299 was achieved by use of a natural enrichment uranium sample which allowed the deposit mass ratio and coun- ter efficiencies to be determined by observing the fission rate ratio in a thermal neutron beam prior to TABLE II ISNF MEASUREMENTS: FISSION CROSS SECTION RATIOS a f (U-238)/a f (U-23S) Comments on Separate Measurements and Averages 0.0914 ± 0.7% 0.0921 ± 2.0% 0.0925 ± 3.0% Natural uranium vs. 99.75% enriched U-235 1600/1 depleted uranium vs. 99.75% enriched U-235 Thick U-238 deposit, 250,000/1 depletion; same U-235 0.0918 ± 1.1%* 0.0923 ± 1.1%* Average in situ without correction for scattering in the detector Corrected average for unperturbed ISNF field {E) 10" COMPUTED PHOTONEUTRON EMISSION SPECTRA (Normalized J^(E)dEM) r-vn r 1 h Na-CD 2 10" Ga-CD 2 P. j j ! ru-t! j J -n r [j " ! i i i I i i li I I i i I l LJ- all energy 6 0.1 0.2 0.3 0.4 0.5 0.6 OT OS 0.9 1.0 1.1 1.2 NEUTRON ENERGY (MeV) Fig. 2 Neutron spectra produced by several photoneutron sources as calculated by Monte Carlo. Scale: Approx. 1:1 Shutter Path Fig. 3 Dual foil "compensated beam" geometry used for fission cross section measurements. 308 1.6 1.5 1.4 in -z. rr < m — 1.3 \$ 1.2 1.173 1.1 THE APPARENT 235 U(n,f) INTEGRAL CROSS SECTIONS OVER PHOTONEUTRON SPECTRA 1.508 265 keV 770 keV a=1.173 20 40 60 80 [SOURCE-DETECTOR DISTANCE] 2 [CM 2 ] Fig. 4 Plots of the apparent fission cross section vs. the square of the source-target spacing showing extrapolation to elim- inate the contribution of room-return neutrons. 309 IMPACT OF ENDF STANDARDS ON FAST REACTORS Ugo Farinelli Comitato Nazionale Energia Nucleare C.S.N. Casaccia, Roma, Italy ENDF/B is widely used throughout the world by fast reactor designers, either directly or as a basis for adjustment procedures. Neutron standards have at least potentially a value that goes beyond their intrinsic function in the production of evaluated nuclear data files; they could be used as standards also for integral measurements, especially in connection with benchmark experiments, and could provide a reference set for adjustment procedures. The conditions under which this new use of the standards would be possible are briefly reviewed in the paper. (Adjustment; benchmark; cross-sections; fast reactors; integral experiments; standards) Introduction When was the im first reac please don just NONE, perhaps no scrutiny a be amply q report on tive or to interpret I was asked pact of ENDF tion was to ' t misunders Cooler thin t the case nd that even ualified. In this subject invite prov it in this s to prepar standard answer by tand me, king conv that the a negati any case is eithe ocation econd way e a paper s on fast a four 1 the four inced me question ve answer asking r meant t and I dec discussing what reactors, my etter word - letter word was that this was deserved closer would have to a European to o be provoca- ided to The question does of course not concern ENDF/B as a whole: anybody connected with fast reactor design, in any part of the world, will share grateful appreciation and due respect for the virtues of ENDF/B-IV in dealing with fast reactor problems. If ENDF standards help measure and evaluate nuclear data and prepare END Files, God bless them. But this is an indirect and somewhat trivial consideration; the real problem is, does the concept of neutron standards - in the sense meant in this Conference - have a more direct bearing on the real world, I apologize for the professional bias, I mean on the macroscopic world to which nuclear reactors belong? I believe that it does - or at least that it might. But before discussing these possible impacts, at least from a European point of view, I have to venture on very slippery ground and refer to a subject which is very touchy on this side of the Atlantic: cross section adjustment, a procedure that may meet here with an ad- verse environment. The Utopia of the IDAlists In Europe we do not believe in the Happy Utopia of IDA - the Ideal Differential Approach. According to the tenets of this faith, there shall come a day (the Day of the Final Evaluation) in which all the relevant cross sections will be known to any desirable detail and degree of accuracy. By using the highly sophisti- cated calculational methods then available on the Utoputer (Utopian Computer) it will be possible to obtain results as accurate as needed, and an estimate of the errors to go with. Although tremendous progress has been made in nuclear data measurement and evalua- tion, notwithstanding the development of powerful codes, despite the fact that new breeds of bigger and faster computers appear every day on the shelves, I do not think that IDA is any closer now than it was some years ago /!/. There are mainly two reasons behind this. The first is that reactors are not waiting for IDA to be built. Light Water Reactors are now designed and operated with methods and data that make IDAlists frown their brows. It is an upside-down world that of commercial reactor design, where diffusion performs better than transport and old data better than new ones. Still, reactors are built and do work, although basic data and methods are still unable to account for the observed temperature coefficient, to name but one. Adjusting to Adjustment A similar situation applies to fast reactors. Here the different time-scale, and emphasis, in Europe and in the United States may account at least partially for the difference in philosophical approach. With two demo reactors in operation and advanced work on the commercial plants, hardware is obviously attracting most of the attention and of the effort, and little space is left for fundamental research. Our reactor physicists are hardly pressed for numbers that have to go into the design, and they would become highly unpopular should answer something like "Wait a minute, I need a better cross section for hafnium, let me file it into WRENDA". We have to make use of anything that is available, which often means integral experiments. In this way one makes mistakes (who doesn't anyway?) but one also makes reactors. Incidentally, something like this might conceiv- ably repeat itself for fusion reactors in a not too far future, with an exchange of roles, due to the more relaxed time-scale and fundamental approach prevailing in Europe, as compared with the faster and more technologically-oriented CTR program in the : Uhited States. The second reservation on IDA is more basic. As time goes, the amount of detail that is available on nuclear data grows tremendously, and to the reactor designer who has to use them the situation may well seem to run out of hand. We had difficulties in adjusting to the idea of 5 000 energy points for the elastic cross section of iron, when Francis Perey comes up and tells us that this is oversimplified, the cross section is much more complex and structured than that, and the amount of detail available (and needed) is one order of magnitude greater, or more. At this point of course we are unable to deal with such a hyperfine structure and the first thing we do is straightforward summation to bring the number of points down to an amenable few thousands at most. Why should then one go to such a resolution in the first place if all this information is going to be thrown away? Somebody might compare this situation with that of the designer of a capacitor for routine circuitry who is given a detailed account of the electronic levels, including Lamb shifts, of his dielectric to come out with his capacity. 310 A Hundred Ways to Cook a Reactor How does all this connect with the problem of standards? Just a moment of patience. I will first dis- close to you a Party Secret: the Party of Adjusters is not monolitic in Europe as it would appear from the outside. There are hidden contrasts and diverging points of view. Incidentally, our intelligence reports speak of similar divergences in the Non-Adjusters' Party of North America, but this is a different question. European adjusters exhibit a very wide range of behavioural schemes, the extremes of which can be exemplified by the Italian Consistent Approach on one side and the Formulaire a la Francaise on the other (the two extremes co-existing more or less peacefully within the same joint fast reactor program). The Consistent Approach 111 tries to take into account all the information available from the differential measurements and the theoretical models by using the results of integral experiments only to make corrections that are consistent with the original information, and therefore preserve all the physical meaning. This method implies the use of calculations for neutron diffusion or transport that are sophisticated enough not to introduce any bias. In this case, I believe one can really talk of cross-section improvement rather than adjustment (or "fudging"). The other extreme - the French formulaire or Book of Recipees /3/ - follows a route close to that used for thermal reactors. It boldly faces the fact that errors are likely to be introduced by the calculational methods commonly used in design as much as by the nuclear data, and uses integral experiments - as well as a limited set of sophisticated calculations with basic data - to generate multigroup cross-sections that are applicable to a certain range of compositions and problems to yield, by low-cost, simple calculational methods, a final result that has an accuracy estimated to be adequate for the design needs. In this context, there have been two different cross section sets for sodium according to the amount of steel or other materials present with the sodium /4/. Obviously in this case you cannot talk about improvement of cross- sections - the only thing you can say in favour of this approach is that it appears that you can design, build and operate fast reactors with it. Standards as Fixed Points... but How Fixed? This distinction of a variety of positions in the adjusters' front is important in order to discuss the role of neutron standards It is clear that brute-force approaches correcting for limitations in calculational methods at the same time as for uncertainties in the nuclear data have little, if any, interest in neutron standards. On the other hand, standards may play a_ preeminent role in the more fundamental methods. The role of standards with respect to these methods could be twofold. On one side, they could be standards for integral cross sections or reaction rates much in the same way as they are for differential measurements. On the other side, they could provide a basis for the evaluation of errors in the nuclear data, a process which is essential for sensitivity studies and uncertainty determination, and for any adjustment procedure. Let us consider these two aspects separately. The idea of establishing a set of international standards for integral measurements has been proposed for instance by the IAEA Consultants on Reactor Dosimetry Cross Sections /5, 6/. According to the recommendations of this group the cross sections of interest for reactor dosimetry should be subdivided into two categories. Category 1 reactions are a limited set with better known cross sections, covering all the energy range of interest, and for which differential and integral measurements in the standard neutron fields are consistent within acceptable margins. Category 2 includes all the other reactions important for reactor dosimetry. The basic idea is that Category 1 cross sections should be extracted from differential measurements and evaluation, and that integral experiments in benchmark fields should only assess the validity of these evaluations - much along the classical lines of CSWG and ENDF/B in the United States; for Category 2 reactions, on the other hand, a more "European"approach is suggested. The original draft conclusions of the IAEA Consultants predicted that for most Category 2 reactions "the improvement of cross sections is expected to derive essentially from integral measurements in benchmark fields". Strong opposition from US differential ists prompted a milder formulation of improvements being "expected to derive from a combination of integral and differential measurements which should yield internally consistent data". What if they don't? Whichever the formulation, Category 1 reactions hold a preeminent position in the procedure, and it is suggested that integral experiments in benchmark fields are aimed at measuring the ratio between Category 2 and Category 1 average cross sections in a given neutron field rather than the absolute values of the cross sections themselves. In this sense, Category 1 reactions have the role of standards. I believe that this approach is fruitful and represents a reasonable compromise between differentialists' and integralists' points of view and that it could be usefully extended to other areas of reactor physics (and not only reactor physics: think for instance of damage cross sections) where integral experiments play an important role. The situation would be still more promising if the same standards could be used in diffe- rential and in integral experiments. This would make it much easier to reach consistency. Unfortunately, this seems impossible at least at the present time and for various reasons. The first reason is that most, if not all, of ENDF standards are very difficult to use in an integral experiment. In most such experiments, only activation measurements are made. This rules out the use of Li, B, C and H. Even fission standards are used only in- directly with activation detectors, since energy- dependent yields must be known or assumed. The situation could be improved by the development of sufficiently precise and standardized techniques such as alpha-sensitive solid state track detectors, or helium production assay, but at the moment these are not simple, reproducible and precise enough to be used in connection with standards. Even if it were possible to measure these standards in integral experiments, it would certainly not be possible to restrict their use to the energy range for which they are defined as standards in differential measurements. Finally, in my personal appreciation, the word "standard" carries with it some implication of stability; one would very much like to measure a length with a ruler that does not stretch and shrink during the operation. The desire to improve continuously the quality of the standards, however, implies frequent revisions of the standard cross sections, in contrast with the stability. This contrast is particularly evident for the U-235 fission cross section: the extreme direct usefulness of this cross section in 311 reactor calculations (as distinct from its use as a standard) on one hand results in more frequent and more accurate measurements of this cross section, on the other hand it pushes the use of new values as soon as they become available; or in other cases it encourages the adjustment procedures in order to force agreement with integral results. I know of many calculations carried out now that use recent values for this cross section, that will no doubt be reflected in ENDF/B-V but that could not be considered the standard values at the present time. The Role of Standards in Error Evaluation Standards also have a crucial role in the assessment of uncertainties in the integral results of calculations and in all adjustment procedures. I think this aspect has been covered in detail in the preceding paper HI . I just want to stress the importance of an accurate assessment of the errors and of the covariance of errors in any consistent adjustment procedure. The splitting of this error in the part pertinent to the measurement relative to the standard, and in the part inherent to the standard itself, greatly helps in arriving at a consistent formulation of the errors and the correlations. References /!/ U.FARINELLI, "The Role of Integral Experiments in the Production of Nuclear Data for Reactor Core and Shield Design and for Irradiation Experiments", in Nuclear Data in Science and Technology, Vol.11, p. 103-119, I.A.E.A., Vienna (1973) HI A.GANDINI, M.PETILLI and M.SALVATORES, "Nuclear Data and Integral Measurements Correlation for Fast Reactors. Statistical Formulation and Analysis of Method: The Consistent Approach", International Symposium on Physics of Fast Reactors, Vol.11, 612, Tokyo (1973) 13/ J.C.MOUGNIOT et al . , "Review of Neutronic Considera- tions arising from Studies for a Proposed 1200 MWe Fast Reactor", ibid. Vol.1, paper A 44 /4/ P.BOUTEAU et al., "Formulaire de Propagation de Neutrons dans les Milieux Acier-Sodium pour les Protections de la Filiere Rapide", Proceedings of the Specialists' Meeting on Sensitivity Studies and Shielding Benchmarks, p. 148, Nuclear Energy Agency, Paris (1975) /5/ Proceedings of a Consultants' Meeting on Nuclear Data for Reactor Neutron Dosimetry held at Vienna, 10-12 September 1973; I.A.E.A., Nuclear Data Section INDC (NDS)-56/6, Vienna (1973) i /§>/ Proceedings of a Consultants' Meeting on Integral Cross-Section Measurements in Standard Neutron Fields for Reactor Dosimetry held at Vienna, 15-19 November 1976, I.A.E.A., Nuclear Data Section (to be published) HI C.R.WEISBIN, "Propagation of Uncertainties in Fission Cross Section Standards in the Interpreta- tion and Utilization of Critical Benchmark Measurements", this Symposium. 312 ABSOLUTE 235 U, 238 U, Np FAST NEUTRON ^FISSION CROSS- SECTION MEASUREMENTS* Vo Mo Adamov, Bo M Alexandrov, Io Do Alkhazov, L» Vo Drapchinsky, So So Kovalenko, Oo I. Kostochkin, Go Yuo Kudriavzev,. L. Z„ Malkin, K„ A„ Petrzhak, Lo Ao Pleskachevsky, A Vo Fomichev, V Io Shapakov Vo Go Khlopin Radium Institute, Leningrad, USSR The fission cross sections of 2 U, 3 U and 3 Np for Cf fission spectrum neutrons and 14 8-MeV neutrons have been absolutely measuredo Coincidence method fission fragment - associated particle was used. Measurement accuracy is better than 2%o Error sources are discussedo (Fission cross sections: absolute measurements; 235.. 238., 237 . neutrons; U; U; Np) Cf; fission spectrum neutrons; 14.8-MeV Absolute measurements of induced fission cross c 2 3 6„ 238., , 237., . . 252„ r c . sections of U, U and Np for both Cf fission spectrum neutrons and 14 8-MeV neutrons have been made. The method of coincidence between fission events in the target of the nuclide and the particles associated with neutrons, inducing fissions, was usedo For the fission spectrum neutron measurements, the neutron source was 2 Cf, and the associated particles were 2 3 Cf sponta- neous fission fragmentSo Each 3j3 Cf fission fragment corresponds to v neutrons - a value known to a precision of tenths of 1%. For the measurements on 14o8-MeV neutrons, the T(d,n) He reaction from a neutron genera- tor was used and the associated particles were a-parti- cleso The main advantages of the coincidence method are as follows: there is no need for either direct neutron flux determination or 100% efficiency of associated particle detection; the influence of slowed-down and scattered neutrons can be minimizedo Additionally, in the case of 14o8-MeV neutron measurements, effects due to neutrons from other reactions are eliminated, and there is no need to determine the solid angles and other geometrical factors« In the fission spectrum neutron measurements, the experimental set-up was used, where a flat target of a fissionable nuclide was irradiated by neutrons from a s Cf source of the smallest possible dimension. Source and target were fixed 4 mm apart with an ac- curacy - 10 microns. The mass of structural materials was negligible (Fig. 1 and 2). The double ionization current pulse chamber was used as a fission fragment detectoro The following expression was derived for calcula- tion of fission cross sections for the Cf fission neutrons : C f = 2j[ A R 3 N P" o GLN (1) T Cf where: fission cross section of the measured nuclide, number of fissions of the measured nuclide, atomic weight of the measured nuclide, Avogadro constant, NUCLIDE MEASURED FRAGMENT / FRAGMENT L>7 ^ """ Figo lo Experimental geometry in the cross-section measurements for 3 3 Cf fission-spectrum neutrons. Work supported by the International Atomic Energy Agency (Research Contract No. 1718/RB) This paper was not presented orally at the Conference. 313 'Cf source is deposited on the lower backing (not shown) o Fig. 2. Assembly for holding the 2D Cf source and fissionable nuclide target. Cf - mass of fissionable nuclide, - 252 Cf prompt neutrons average number, - number of 2 Cf fission events, - geometrical factor, - fissionable nuclide layer radius, - the correction for overlapping in the 262 f fission fragments channel. The neutron flux is introduced in Eq (1) through the value of " ( 352 Cf) and the geometrical factor which takes into account the mutual position of the Cf neutron source and the target. An exact calculation of the geometrical factor requires: (a) the fissile layer radius, the distance be- tween 253 Cf layer and that of the fissile nuclide, and the thicknesses of the backing material have to be accurately determined, the layers of 252 Cf and of the fissile nu- clide have to be strictly parallel; (b) the layer of fissile nuclide be uniform to better than 17„; (c) the backings be as thin as possible and their surfaces - mirror-polished; (d) the mass of structural materials be mini- mized in the vicinity of the 2 3 Cf source and target. For present experimental geometry 6-I 1+ I +I 3 (2) where: the main term, taking into account fissions by neutrons, not scattered in the backings, in case of a "point" 353 Cf source; the correction term, taking into account the Cf source real size; I, - the correction term, taking into account fissions, induced by neutrons, scattered in the backings „ Expressions for I.. , I„» I, are multidimensional integrals and were calculated by numerical integra- tion. The factors in the geometrical error estima- tion were made by varying of formulae parameters in their error limits. The experimental geometry in case of neutron generator is shown in Figo 3. Associated a-particles were detected by a thin plastic scintillator,. The detection angle was 165 The fission-fragment detector was an ionization current pulse chamber. As mentioned above, there was no direct neutron flux determination. The associated alphas were detected in the solid angle defined by the input diaphragm of the detector. Provided the two main geometrical constraints are met, 1) the solid angle of the cone of neutrons, associated with the detected alphas, is sufficiently smaller than that subtended by the fissionable deposit, and 2) the fission target nonuniformity is less than TARGET NUCLIDE Fig. 3. Experimental geometry in 14.8-MeV neutron fission cross-section measurements. 314 17.; the induced fission cross section, for 14.8 MeV neutrons, a is determined from the expression: f ~ N N (1 - (3)(1 - 6) n a (3) where : - number of coincidences (without random ones) , - number of associated alphas, - number of fissile nuclei per 1 cm 2 , - correction for undetected fission fragments in 2n-geometry , - neutron flux attenuation correction. The fission-fragment detection efficiency decrease in 27c-geometry is due to counting plateau slope of ionization chamber and fission fragment absorption in the target layer. The plateau slope was determined using fragment pulse-height spectrum analysis. The fission fragment losses were calculated, taking into account fission anisotropy. The neutron flux attenuation due to scattering and absorption on structural materials was determined both experimentally (the materials thicknesses being doubled) and by calculation in the same way as in the preceding case . To ensure the main geometrical constraints, the ratio N /N was measured at the different distances c a between tritium and fission targets. These ratios were found to be constant within the errors limits for the distances used. Two special channels for time and pulse-height spectra analysis were included in electronic equipment to distinguish the true coincidences and to improve the pules-height extrapolation to zero. The fission targets were prepared by high frequen- cy sputtering. The deposit nonuniformi ty had been checked by means of a-counting scanning and was found to be less than 17.. The masses of fissionable deposits were determined by a-counting in a 2it-ionization chamber, and with a low geometry surface-barrier detec- tor. All the dimensions for low geometry were deter- mined by optical methods; errors were found to be a few microns (the solid angle calculation accuracy being 0.27.). Moreover, the solid angle calculation was checked experimentally with an 2 Am standard source, the latter had been calibrated by the a-y coincidence method to an accuracy 0.17.. Isotopic analysis was carred out for all the fis- sionable deposits, using a semiconductor spectrometer with energy resolution 27 keV. For 2 U, which had impurities undistinguishable with a-spectrometry , mass- spectrometric analysis was carried out. the results of other authors. To compare the data obtained with differential measurement results, numerical calculations of U and 2 U cross sections had been made for 2=2 Cf fis- sion-spectrum neutrons, compiled differential 2 U and 2 U cross sections data being taken from ref. 4. The fission neutron spectrum had been approximated by a Maxwellian distribution with parameter, T = 1406 keV. Calculated values obtained are in a good agreement with our data. One should keep in mind, however, in this comparison, that the accuracy of compiled data is much less than that of our results. References 1. Second IAEA Panel Meeting on Neutron Standard Reference Data, 20-24 November, 1972, IAEA, Vienna, 1972, IIIC 3. 2. A. H. Jaffey, K. F. Flynn, L. E. Glendenin, W. H. Bentley, A. M. Essling, Phys . Rev. C4, 1889 (1971). 3. F. P. Brauer, R. V. Stromatt, J. D. Ludwick, F. P. Roberts and W. L. Lion, J. Inorg. Nucl. Chem. 1_2, 234 (1960). 4. I. Laigner, J. J. Schmidt, D. Woll, Tables of evaluated cross sections for fast reactor materials, Karlsruhe, 1968. 5. H. T. Heaton II, J. A. Grundl, V. Spiegel, Jr., D. M. Gilliam, and C. Eisenhauer, Nuclear Cross Sections and Technology, Proc. of a Conference, Washington, D. C. , March 3-7, 1975. Vol. I, p. 266. 6. J. A. Grundl, V. Spiegel, Jr., C. Eisenhauer, Trans. Amer. Nucl. Soc. L5, 945 (1972). 7. J. B. Czirr, G. B. Sidhu, Nucl. Sci. Eng. _57, 18 (1975). 8. M. G. Sowerby, B. H. Patrick, Annals of Nucl. Sci. and Eng. _1, 409 (1974). 9. M. Cance, G. Grenier, Proc. of the NEANDC/ NEACRP Specialists Meeting on Fast Neutron Fission Cross Sections of U-233, U-235, U-238 and Pu-239, June 28-30, 1976, at Argonne National Laboratory, p. 237. 10. W. Hart, UKAEA, ANS(S)R, 169 (1969). 11. V. M. Adamov, L. V. Drapchinsky, G. Yu. Kudriavzev, S. S. Kovalenko, K. A. Petrzhak, L. A. Pleskachevsky , A. M. Sokolov, Preprint Radievogo Instituta, RI - 52, Leningrade, 1976. The following a-decay half-lives were used to calculate deposits masses: 1/2 ( 33S U) (7.0381 1/2 ( 238 U) = (4.4683 0.0048) 0.0029) 10° 10 9 T 1/2 ( 23? Np) = (2.11 ± 0.01) x 10 6 y. The sources and values of errors, associated with fission cross sections determinations, are listed in Tables 1 and 2. The fission cross sections, determined in the present work, are listed in Table 3 along with 315 Table 1. The errors of data, associated with determination of fission cross sections for 2B2 Cf fission spectrum neutrons (%) Error Sources Nuclides 23S U 338 U 337., Np v ( 2SS Cf) 0.35 0.35 0.35 Geometrical factor 0.71 0.71 0.71 Mass deter- Solid angle 0.30 0.30 mination Statistics 0.35 0.45 at low Part of measured geometry isotope a-activity 0.41 0.10 Half-life 0.20 0.47 Statistics 0.55 Mass deter- Extrapolation to zero mination pulse height 0.51 in 2it - Absorption and geometry scattering correction Part of measured isotope a-activity 0.30 0.30 Statistics 0.80 1.11 0.95 Number of Extrapolation to zero fissions pulse height 0.40 0.37 0.41 of isotope Correction for measured absorption in layer Fissions in other 0.30 0.25 0.28 isotopes 0.30 0.14 Total fission cross section error 1.44 1.68 1.59 2 U half-life error is taken into account in determination of a part of the measured isotope a-activity. 316 Table 2. The errors of data, associated with determination of 14.8 MeV neutron induced fission cross sections (7„) Nuc lides Error sources 236 U 838 U Np Mass deter- Solid angle 0.35 0.35 mination Statistics 0.35 0.45 at low Part of measured geometry isotope a-activity 0.41 0.10 Half-life 0.20 0.47 Statistics 0.55 Mass deter- Extrapolation to zero mination pulse height 0.51 in 2ti - Absorption and geometry scattering correction Part of measured isotope a-activity 0.30 0.30 Statistics 1.0 1.0 1.4 Number of random - coincidences 0.20 0.20 0.20 Neutron flux Number of attenuation 1.0 1.0 1.0 fissions Extrapolation to zero pulse height 0.40 0.20 0.30 Correction for absorption in layer 0.30 0.30 0.30 Fissions in other isotopes 0.20 0.12 Number of associated particles 0.05 0.05 0.05 Total fission cross section error 1.60 1.70 1.94 317 Table 3. Results of absolute fission cross section measurements (and comparison with other authors) Nuclides Fission cross sections (mb) 2 Cf fission spectrum neutrons 14.8 MeV neutrons This work Other authors This work Other authors Experiment Calculation S35 U 1266 - 19 1281 1204 - 29 2188 - 37 2075 - 40 (14.6 MeV) 2191 - 40 (14.6 MeV) 8 2063 * 39 (14.6 MeV) 9 338 u 347 i 6 352 324 - 14 6 1207 ± 20 1193 - 20 (14.6 MeV) 8 1149 - 25 (14.6 MeV) 9 a37 Np 1442 - 23 - - 2430 - 47 2360 * 90 (14.1 MeV) 10 2500 - 70 (14.0 MeV) 10 318 NEUTRON ENERGY STANDARDS G. D. James UKAEA, Atomic Energy Research Establishment, Harwell, England. In recent years there have been several examples of discrepancy between the neutron energy scales from different spectrometers. Some of the work undertaken to review and improve neutron energy determination is noted and some suggestions on how errors can be reduced are listed. The view advocated by Youden that the only worthwhile estimates of systematic error are those made experimentally is presented. Comparison of energy determinations for a few resonances show that at best resonance energies can be quoted to an accuracy of about one in 10 000. A list of ' J i narrow resonances, over the energy range 0.6eV to 12.1MeV, which should prove suitable as energy standards is given. At present, not all the energies listed are known to the highest accuracy attainable. (Accurate neutron energy determination, energy standards) 1. Introduction Neutron energy standards are required to help ensure that all neutron spectrometers produce data on energy scales that agree to within the estimated errors of measurements. Discrepancies in neutron energy scales, of iwhich there have been several examples in recent years, present additional problems for evaluators and compilers, for the users of data, for example those who wish to use transmission data for the calculation of neutron shielding, and for analysts who wish to undertake a combined study of partial and total cross section data. The work required to revise and correct energy scales could be saved if a set of accurately measured resonance energies became available. These could then be checked on each spectrometer, preferably during each experiment. At least one such list has been published^ ' but improvements in neutron spectrometer resolution demand that this should be revised or supplemented by much narrower resonances. As energy measurements become progressively more accurate, several examples of energy discrepancies have arisen some of which have been successfully removed but some of which are still outstanding. A discrepancy of 20keV at 6MeV between measurements made using the 2 H(d,n) reaction compared with those made using the 3H(p,n) reaction was explained by Davis and Noda^ 2 '' in terms of a field dependent calibration factor. More recently, a discrepancy of 25keV at 1.5MeV in the energy scale of 2 3 8 U/ 2 35q' fission cross section measurements as measured on white source spectrometers(3-°) wa s removed when inadequacies in the method of calibrating one of the spectrometers by the resonance technique were revealed and the results were revised using a direct calibration in terms of measured flight path length^?'. Measurements of the same ratio'"' using a mono- energetic Van de Graaff neutron source give results which are about 20keV above the mean of the broad spectrum measurements. A discrepancy of 7keV, again indicating that data from mono-energetic sources tend to lie higher than white sources, is presented by data on the energy of the peak of the resonance in the total cross section of 6 Li nea r 250keV. The data available on this cross section are presented in sect. 5.1. This resonance is very broad and neither it nor the 23ou threshold are suitable for accurate energy comparison. Nevertheless they easily reveal discrepancies which amount to about 1%. Within the group of white source spectrometers the discrepancies are not so large. Bockhoff et al.(9) have shown, however, that the energies of 2 5°U resonances below 2.7keV as measured at Geel and Columbia ( 1 0) differ by 0.1%. This difference is almost independent of energy and is equivalent to a 10cm discrepancy in the lengths of the 200m flight paths used in each experi- ment. Tnis is far greater than the errors of measurement which are about 1cm or less. As a framework for considering how to improve the measurements and comparison of neutron energies, the methods in use on three spectrometers and the random and systematic errors in each of the methods are briefly reviewed in sect. 2. Some considerations which can help to reduce inaccuracies are described in sect. 5 and the important concept of making experi- mental measurements of systematic errors, so ably advocated by Youden'' ', is presented in sect. k. Examples of the accuracies which are achieved in energy measurements at present are discussed in sect. 5 where the results available from transmission measurements for 6Li , 23Na, 23oU and 12c are discussed. In sect. 6 a list of forty- one resonances over the energy range 0.6eV to 12.1MeV is presented. This list was drawn from examples of narrow resonances kindly suggested to me by members of an INDC sub-group on neutron energies*. It does not yet carry the imprimatur of the sub-group but it is likely that the list finally adopted will contain most if not all the resonances given in sect. 6. The establishment of a recognised list of resonances is important in that it will encourage the measurement and intercomparison of the energies of these resonances by scientists who recognise the need to establish accurate energy scales for their spectro- meters. The conclusions that can be drawn from the studies presented in this paper are discussed in sect. 7. 2. Neutron resonance energy determination Two methods of neutron energy determination are in use; those based on neutron time-of- flight and those based on the uses of mono-energetic charged particles to produce mono-energetic neutrons. The *J. Boldeman, F. Corvi, J. A. Harvey, G.D. James, J. Lachkar, A.B. Smith and F. Voss 319 former require pulsed sources and spectrometers based o.i electron linear accelerators, cyclotrons, synchrocyclotrons and pulsed Van de Graaff machines are in operation. The latter method is based exclusively on Van de Graaffmachines which alone can produce charged particle beams with energies known to sufficient accuracy. In this section the methods used to determine energy on three spectro- meters which are typical of the range of instruments in use are presented so that the relative magnitudes of random and systematic errors can be more readily appreciated. 2. 1 Neutron energy determination at ORELA In the ORELA neutron time-of-flight spectrometer, short bursts of V+OMeV electrons of minimum width 5ns strike a water cooled tantalum target to produce bursts of fast neutrons which are moderated in water to produce a pulsed source of neutrons covering the energy range from several MeV down to the Maxwellian spectrum at thermal neutron energy. Neutrons are detected by a detector set at a known distance from the source. Part of the interval between the time when the neutrons are produced and the time when neutrons are detected is measured and recorded by a time digitizer in units of timing channel width which have a minimum value of 2ns. Another part of the interval is determined by the delay between the electron pulse and the opening of the first timing channel - the so called initial delay. As an example(12) to illustrate the contribution of systematic and random errors, in an experiment to measure the transmissions at a path length of 155^1 -10mm, the following equations are used to derive neutron energy E c and the relativistically corrected energy Er E r = E R = 72. 2977 (l55'+'-+1- 10mm) t-(2695 ± 2ns) 2 E (1 + 1.5963 10" 7 3 ) c c Ka) Kb) Here 2695±2ns is the initial delay and t is the part of the flight time recorded on the time digitizer. Examples of the results obtained for a few selected resonances in Pb,Al,S, Na and 25ojj are given in table 1 which gives the contribution to the quoted uncertainty made by dE c jj, the error in locating the peak of the resonance, dEt the error in energy due to the 2ns uncertainty in the initial delay and dE^ the error in energy due to the 10mm uncertainty in the neutron flight path. Below the Al resonance at 5903. 5eV the error is dominated by the uncertainty in flight path length. Above this energy the error is dominated by the error in locating the peak of the resonance by the fitting programme SIOB. 2.2 Neutron energy determination on the Harwell Synchrocyclotron In the Harwell synchrocyclotron neutron time- of-flight spectrometer, pulses of protons are accelerated to an energy of I'+OMeV and deflected downwards to strike a 2cm thick tungsten target to produce fast neutrons by spallation. The salient features of the target and detector system are illustrated in fig. 1. Fast neutrons from the proton target reach a 2cm thick beryllium faced water moderator tank from which a pulse of moderated neutrons travels towards the detector. rable 1 ORELA 150m Neutron Ene rgies Isotope Resonance dE c h dE t dE L BNL-325^ 1 3) Energy (eV) (eV) (eV) (eV) (eV) 206 Pb 3357.'+ ±0.5 0.1 0.07 o.i+3 3360 ±10 2 ?A1 5903-5 ±0.8 0.3 0.16 0.76 5903 ± 8 32S 30378 ±6 if 2 h 30350 ±90 2 ?Na 53191 ±27 26 k 7 53150 ±30 206^ 71191 ±18 T+ 6 9 71000 ±100 32 S 97512 ±28 22 10 13 96600 ±500 32 S 112186 ±33 27 12 1'+ 1 1 1 i+00 ±500 2 ?A1 257-3 ±0.3* 276 lf1 33 257.5 ±1.8 238u 1^19.76*0,18 0.0*t 0.02 0.18 1^19.5 ±0.3 238 D 2^89.18±0.32 0.03 0.0'+ 0.32 2^+88. k ±0.7 'keV Since late 1973 transmission measurements have generally been carried out using either a 2.5cm thick Li-glass detector at 50m or using a 1.2cm thick NE110 detector at 100m. The 50m measurements give results over the energy range 100eV to 100keV and the 100m measurements give results over the energy range 10keV to 10MeV. During 197*+ the distances between reference marks at 1m, km, 13m, ktym and 98m from the neutron source were measured to better than 0.5mm accuracy by a tellurometer*. The equations used to determine neutron energy are set oat in Appendix 1 where the symbols used have the following meaning. L is the distance from the face of the moderator to the point in the detector where a neutron interacts to give a detectable pulse. It is made up of the moderator face to detector face distance P and the distance D traversed inside the detector. The neutron time-of- flight is the difference between the time no when a neutron leaves the face of the moderator and the time t n when it is detected. Using t^, the time when Y- rays and fast neutrons leave the proton target, and ty when the Y-flash is detected, t can be re-stated in terms oft-i, the recorded time between the detected Y-flash in channel Xo and a neutron in channel X, t2 the T-ray time-of-flight over a distance Ly, t-z the transit time of fast neutrons from the proton target to the water moderator and tu the neutron slowing down time delay. The channel width is denoted by w, Dy is distance travelled by Y-rays in the detector and Sy is the distance travelled by Y-rays to reach the face of the moderator from which they emerge. In calculating the transit timeof fast neutrons it is assumed that detected neutrons of energy E are produced by neutrons of energy 2E. The slowing down time delay is calculated from the formulae of Groenewold and Groendijk^ . In an experiment to determine the energy of the peak of the resonance near 250keV in the total cross section of Li carried out in 197^+, the Li-glass detector was used, exceptionally, at 100m. In the same experiment the energies of three peak cross sections in 23fta and 2'A1 were also determined. A careful assessment of the errors arising in the measurement have been madeC 1 5) and are presented in table 2. " Tellurometer (U. K. ) Ltd., Roebuck Rd. , Chessington, Surrey, England K19 1RQ Tellurometer-U.S.A.,89 Marcus Blvd, Hauppauge,NY11787 320 Table 2 Harwell Synchrocyclotron 100m Neutron Energies Isotope Al-27 Li-6 Al-27 Na-23 Resonance Energy (keV) 119-753±0.042 242.71 -0.^ 257-16 ±0.13 299-19 ±0.12 dE cn (eV) 3 330 90 15 dE t (eV) 7^ 97 dE L (eV) 28 58 62 72 2.3 Energy measurements at the Argonne Fast Neutron Generator . Recently, Meadows* 1 "-' has undertaken a close scrutiny of the techniques involved in energy determination at the Argonne FNG when used as a source of mono-energetic neutrons. Yield curves and neutron energy spectra were calculated for some (p,n) reactions commonly used as neutron sources for energy calibration purposes. These calculations take into account the energy spread of the incident proton beam and the statistical nature of the proton energy loss. Meadows shows that when thresholds are observed by detecting neutrons at O deg. to the proton beam direction, the best results are obtained by plotting the square of the yield against proton energy and extrapolating to zero yield. A linear plot of neutron yield against proton energy can be in error by 1 - 2keV if the energy spread of the proton beam is large. A calibration was established by locating the 7 Li(p,n) 7 Be threshold (188O.6O * 0.07keV) and the 11 B(p,nr 1 C threshold (3016.4 ± 1.6keV). Using this calibration a measurement of a carbon resonance gave the value for a mean over five target thicknesses of 2078.2 - 2.8keV. This error allows for systematic uncertainties. The error in the mean derived from the spread of the five readings is 0.67keV. The calibration was also confirmed by a neutron time-of- f light measurement which at En= 2.7739 ± 0.0046M e V gives Ep = 4.4624 i 0.0046MeV to be compared with the value Ep = 4.466 t 0.004MeV derived from the analys- ing magnet calibration. 3. Redaction of uncertainties Several techniques are available which can be used to reduce uncertainties in some of the quantities involved in the determination of neutron energy. Five of these are discussed in this section. 3. 1 Effect of resolution on s-wave resonance energies In total cross section and transmission measure- ments, s-wave resonances show a marked asymmetry caused by resonance-potential interference which causes a minimum on the low energy side of the resonances. With worsening energy resolution, the energy at the observed peak of the cross section, Em, shifts to higher energy and the energy difference between the peak and the interference minimum, observed at E m , increases. This effect is illustrated for the 299keV resonance in Na. in fig. 2 which is taken from a paper by Derrien*' ' ' •'. Derrien shows that provided both F44 and F^ are known, the effects of spectrometer resolution can be allowed for to derive E r , the energy at the resonance peak observed with perfect resolution. Values of E r derived from six measurements of different resolution are illustrated in fig. 3. The mean of the six values of E r is 298.65 - 0.32keV. Even after carrying out the correction, however, only the three measurements with the best energy resolution agree within the error in the mean. Taking these three values only, the mean value of E r is 298.550 - 0.032keV which corresponds to an error of one in 365O. This effect of energy resolution on asymmetric s-wave resonances makes them less suitable for accurate energy comparison than symmetric resonances. 3-2 Cumulative probability plot Fitting programmes such as SIOB in use at Oak RidgedS) are no t always available for the determin- ation of the position of the peak of a resonance in energy or time. When the observed shape of a resonance is Gaussian, or at least close to symmetric, the error in deducing the position of the resonance peak can be considerably reduced by the use of a cumulative probability plot (ogive curve) in which, after the subtraction of a suitable background representing the neighbouring value of the data, the data (cross section, transmission or observed counts per timing channel) are treated as probability values and plotted on probability graph paper against timing channel number. The linear portion of the graph is fitted by the least squares method to derive the timing channel corresponding to the centre of the resonance at 50% probability. Half a channel must be added to the result obtained because of the binning of data into channels. This technique readily allows peak positions to be estimated to within an error of 0.1 channels or less and an example is shown in fig. 4. The extent to which the resonance shape is not normal is immediately apparent. Only the linear portion of the curve is fitted and the error derived reflects any departure from normality. 3-3 The &L/&t method In deriving neutron energies by the direct method set out in sect. 2.2 and in Appendix 1, it will be seen that several of the quantities involved can only be estimated with relatively large uncertainties* These quantities are the distance D traversed by a neutron within the detector before detection, the energy, transit distance and transit time, t-j, of the fast neutrons which generate the neutrons under consideration, the slowing down time delay th and any differences, caused by differences in pulse shape, between the detection of neutrons and coincident gamma rays. It is shown in Appendix 1 that all these quantities can be removed from the equations used to determine energy if the neutron flight times t 1 and t 2 are measured with the same detector at two path lengths L-| and L 2 . However, a disadvantage of the method is that the effective path length is reduced to AL = L-p L 2 . In principle the method allows (t x + t^ ) to be measured but only at the expense of re-introducing the poorly known quantities D, L^1 , L^2 and£. However, by careful design, such as the use of thin detectors to reduce D, improved measurements of (t^ + t^) could be made. 3.4 H Method This method, described by Meadows^ , for reducing the error in determining the energy at the threshold of a (p,n) reaction has been described in sect. 2.3- 3.5 Effect of counting statistics on the energy uncertainty In their measurement of the Li total cross section peak energy, the statistical quality of the data was not good but James et al.^ 1 5/ were a bi e to 321 derive accurately the error in determining the peak energy caused by the errors in the counts per timing channel. This was done using random number (Monte Carlo) techniques to vary each measured transmission datum by an amount controlled by the normal law of errors and by the standard deviation of each datum. The data set obtained in this way was fitted and another estimate of peak cross section energy obtained. This process was repeated ten times to get a measure of the spread of the values derived. In this numerical experiment, uniform pseudo-random variates were con- verted to normal variates using the simple but exact Box-Miiller transformation'-' 1 "''. It was found that even for the broad resonance near 250keV in "Li the peak could be located to an error of 0.33keV. 4. Experimental determination of systematic errors When a constant has been measured in several independent studies, it becomes necessary to under- take some data evaluation with the aim of determining a best value and of setting some limit to the error in this best value. In a series of papers which are models of clarity, W.J. Youden'- ' has shown that the aim should be to establish 'enduring values' which are best values with an error range within which future best values will lie. By examining fifteen values of the astronomical unit measured between 1895 and 1961 and 21 measured values of the velocity of light, Youden shows that, as so often happens, the data differ from one another by more than would be expected from the ascribed estimates of uncertainty. He attributes this to a poor assessment of systematic errors and argues that the only worthwhile estimates of system- atic errors are those which are made experimentally. To achieve this it should be noted that when an experiment is done at another laboratory everything gets changed whereas an investigator at one laboratory makes only minor changes to his equipment. Youden argues elegantly for making a direct experi- mental assessment of systematic errors by planning experiments in which everything which could make a difference to the experiment, and even a few things which are regarded as not affecting the measurement, is changed. Often the things or quantities changed will have only two descriptions or values but the benefit in bringing systematic errors to light will be immense. The time devoted to the experiments need not increase inordinately. Clearly, more will be learnt by making six changes and accumulating data for a sixth of the time after each change than if no changes are made. Youden also clearly shows that if the changes are properly planned a large reduction in the error in the mean can be achieved by changing more than one thing at a time. Many papers on the way such experiments should be planned are now available, some in the collection of Ku(20). 5. Comparison of data for certain resonances In this section the data available on single resonances in Li , Na and C and five resonances in 2 -?"U are presented. Average values derived by taking the best value from each laboratory are given and used to produce a comparison factor G which equals the mean value divided by the error in the mean. The resonances selected differ markedly from one another. The resonance near 250keV in Li is broad and will not be used as an energy standard but it is interesting to note the accuracy achieved for the quoted peak cross section energy. This resonance also shows the importance of measuring a well defined quantity such as the energy at the maximum of the cross section rather than the value of the 'resonance energy' which appears in the theoretical expression for the cross section. The resonances in Na and 2 3"U are s-wave resonances, which in the case of Na have been corrected by Derrien for the effects of instrumental resolution, whereas the resonance in carbon has 1=2 and is symmetric. 5. 1 The resonance in "Li total cross section near 250keV Values of the energy at the maximum of the "Li total cross section near 250keV are available from four white source time-of- flight measurements and from four mono-energetic measurements on Van de Graaff accelerators. The data are presented in table 3 and illustrated in fig. 5 in chronological order. The data of Harvey and Bockhoff have been reanalysed using the formulae used initially by Uttley and then by James so that there is no differ- ence in the method of analysis for the four time-of- flight values quoted. The average of these values is 244.0 ± 0.5keV corresponding to 3 = 488. This contrasts with the four mono-kinetic beam Van de Graaff measurements all of which lie at or above 244keV. In fairness to the early measurers it must be stated that the energy at the peak was not a prime consideration in these papers and no errors are quoted on the values given. The errors of ilkeV given in the table are judgements made by experienced workers. The value of 0.33keV given by James et al. ^ ? is derived from a careful assessment of all the errors involved in their measurement including, as described in sect. 3-5 above, the effect of limited counts per timing channel on the energy derived. It is likely that an experimental determination of systematic errors as advocated in sect, 4 above would lead to an upward revision of this error. Recently a value, given preliminarily as 242 ± 2keV, has been measured by time-of- flight on a Van de Graaff in an experiment which allows the succeeding gamma flash to fall at the peak of the "Li resonance. If confirmed, this would reduce the average of all measurements made by the time-of- flight method of 243.60 - 0.58keV. Table 3 Neutron energy at the maximum of the "Li total cross section Author Hibdon & Mooring^ 21 ) Farell & Pineo^ 22 ) Meadows & Whalen' 2 ^) Uttley(24) James et al. (15) Harvey et al.(- 2 5.> Harvey (James) Bockhoff et al. ^9) BCckhoff (James) Knitter^ 2 ") Average of values marked Year 1968 1968 1972 1975 1975 1975 1975 1975 1975 1976 E(max) (keV) 246 250.6 252.5 *243.5 + 1 ♦242.71 ± 0.33 246 ± 1 ♦244.8 ± 1 245.0 ± 1 ♦245.0 ± 1 247-0 ± 5 244.0 ± 0.5 5.2 The resonance in 2 ^Na at 298.550 ± 0.082keV As discussed in sect. 3'1i the data available on the observed peak energy Em for the s-wave resonance near 299keV in 2 ^Na have been corrected by Derrien^ '' for the effects of instrumental resolution. The values of Em and the values of E r which would be observed with perfect resolution are illustrated in 322 fig. 3 and listed in table 4. The mean of the three values with the best energy resolution corresponds to ■G = 3640 Table 4 Observed and corrected peak energy values for the 298.55keV resonance in Na Origin %(keV) E r (keV) Sac lay II 298.94 ± 0.37 *298.39 1 0.37 Karlsruhe 299.26 ± 0.20 »298.66 ± 0.20 Columbia 298.5 - 1.0 297.35 i 1.0 Harwell S.C. 299.2 ± 0.2 ♦298.6 ± 0.34 Cadarache 302 i 4 299.4 i 4.0 Saclay I 303 ± 3 299.5 ± 3.0 Average 298.65 ± 0.32 Average of values marked * 298.550 i 0.082 5.3 Five resonances in 238 U It is suggested in sect. 6 that resonances in 238u could be selected for use as standard over the energy range 6eV to 3keV. The data available on five of these resonances are presented in table 6 . All the measurements made at the Harwell synchro- cyclotron used a Li-glass detector. A 50m measure- ment was made in August 1976. For other reasons, the flight path was then raised by 3cm and the source geometry was altered so that the water moderator stood in front of the proton target instead of under- neath it. Energy measurements of 2 -? 3 U resonances were then made, in November 1976, both at 50m and at 100m. Results were also derived using the Al/At method discussed in sect. 3.3- The AL/At method is based on the same data as the two other results obtained in November and the correct method of deriving the best value and an error in this best value remains to be formulated. For the present, the four results obtained for each resonance are regarded as independent and the average value is given. However, the error quoted in the average value is that for an individual reading derived from the spread of the measurements. Results from Oak Ridge for three of the resonances were not available to me when the table was prepared. An unweighted average from all laboratories together with an error in the mean derived from the spread is also given. The agreement between Harwell and Oak Ridge is well within the quoted errors and all the Harwell values are within the quoted range of the average from all laboratories. The discrepancy between Geel and Columbia has already been aoted^ '. The results for the resonance at 2489.47 * 0.5eV are illustrated in fig. 6. 5-4 The resonance in carbon at 2078.05 - 0.32keV The data available on the energy of the resonance in the total cross section of carbon near 2078keV are presented in tables . Two measurements were made on the Harwell synchrocyclotron in August 1976 one using a Li-glass detector at 50m and one using an NS110 detector at 100m. The value obtained from the measurements by the AL/At method are also given although, with two dissimilar detectors, the distances D travelled by neutrons within the detectors are not eliminated from the equations. Source distance uncertainties are, of course, still eliminated. The measurements of Heaton et al. and of Meadows were made on Van de Graaff accelerators by the mono-energetic beam technique. All the other measurements are made by neutron time-of- flight. The excellent agreement between the two methods is clear. The three central values are quoted more precisely than the others. Taken alone these values have an average of 2078.11 t 0.l6keV corresponding to G greater than 10 000. Table 5 Measured peak energy for the resonance in carbon at 2078.05 * 0.32keV Reference Harwell 50m " 100m " AL/At " Average* Davis & Noda(?L, Heaton et al. KZ ° ) James Meadows ^ 16) Perey et al.^ 2 9) Bockhoff et al. (9) Cierjacks et al.^50; Average** Date Aug 1976 Aug 1976 Aug 1976 1969 1975 1977 1977 1972 1976 1968 Energy - keV 2079.2 ± 1.1 2078.31 ± 0.44 2077.45 ± 0.84 2078.33 ± O.89 2079 ± 3 2079 ± 3 2078.33 ± O.89 2078.2 ± 2.8 2077.8 ± 1.5 2077 ± 1 2077 ± 1 2078.05 ± 0.32 * Error in individual reading derived from spread of the data ** Error in the mean derived from spread of the data 6. Resonances for use as energy standards a selected list To compare the energy scales of different spectrometers and thereby help to establish accurate energy standards it is necessary that those concerned with this task should all make measurements on the same set of resonances. To promote this development the INDC set up a sub-group with the task of producing a list of suitable resonances. Seventy-six narrow resonances were suggested in the energy range 0.6eV to 12.1MeV. This list is probably too long and I have reduced it by selecting about six resonances in each decade which have the smallest value of (P + A )/£. Here P is the resonance width, A is the Doppler width and E is the resonance energy. The reduced list of forty-three resonances is presented in table 7. The resonances are all in the total cross section of the eleven elements listed. Apart from sodium and iridium, all the elements are commonly occurring, readily available and easily fabricated into suitable transmission samples. The energies given are listed as nominal energies to emphasise the fact that they do not represent evaluated data. Nevertheless the best values of energy available are given wherever possible. No errors are given for the U-238 resonances of ref. (12) but it is known that the errors are likely to be of the order of 1 in 5000. The distribution of the resonances listed is shown in fig. 7- It will be seen that the largest gaps on a logarithmic energy scale are between 0.6eV and oeV and between 6keV and 30keV. It may be necessary to augment the list within these ranges. As more data become available more stringent evaluation of the energies of the resonances listed in table 7 will be possible. Evaluators may then reasonably demand that the only data to be considered are those for which full details of all steps in the 323 Origin Harwell 50m Aug 1976 " 50m Nov 1976 " 100m Nov 1976 " AL/At Nov 1976 Harwell average* Geel^~~ Oak Ridge Kiage .(277 Columbia ^°^ Average** G Table 6 Measured peak energies for five resonances in 238{j 145.634 145.578 1-+5.593 145.606 0.037 0.051 0.033 0.071 463.23 ±0.12 462.93 ±0.15 463.09 ± 0.12 463.24 ± 0.25 Energy - eV 708.44 ± 0.19 708.62 ± 0.27 708.09 ± 0.13 707.59 ± 0.29 145.603 i 0.024 463.12 ± 0.14 708.18 - 0.45 145.68 ± 0.10 145.57 - 0.15 463.62 ± 0.20 708.59 ± 0.25 462. 0.4 707.9 0.4 145.617 4412 0.033 463.13 i 0.24 708.22 - 0.20 1929 3541 1420.12 - 0.045 2490.16 ± 0.40 1419.56 + 0.35 2489.96 ± 1.1 1419.80 ± 0.34 2489.26 ± O.79 1420.02 ± 0.46 2488.61 ± 1.54 1419.88 i 0.25 2489.50 ± 0.71 1419.76 ± 0.19 2489.18 ± 0.43 1420.7 ± 0.3 2490.8 ± 0.4 1419.2 ± 0.3 2488.4 ± 0.7 1419.88 - 0.32 2489.47 4437 4978 0.5 •Error quoted here is that in an individual reading derived from the spread of four values '*Error in the mean derived from the spread is quoted energy determination are published. At present, such a demand would mean that almost no resonance energy data could be evaluated. 7. Conclusion This report has reviewed the methods of neutron energy determination in such a way as to indicate the sources of error ) and methods whereby some of the errors can be reduced have been described. The errors quoted on published energy values are often derived from reasoned judgements by experienced scientists. Better estimates of the systematic errors would clearly be derived by adopting the suggestion made by Youden that as many experimental components as possible should be deliberately varied preferably more than one at a time in a planned way. From a comparison of the data available on certain resonances it appears that at best energy measure- ments from different laboratories agree to a fraction approaching one in 10 000. Furthermore, the work of Meadows has shown that with the development of care- ful methods there is no discrepancy between mono- energetic and white source measurements. A list of forty-three narrow resonances suitable for energy comparison and calibration has been drawn up which may encourage experimentalists to measure the energies of the resonances listed both as a means of improving energy standards and also as a means of keeping a continual check on the energy scales of their spectrometers. The confirmation of at least one energy from a list such as that in sect. 7 should be encouraged whenever changes are made which could result in a changed energy scale. Very few papers have been published giving the full details of energy determination which could reasonably be demanded by an evaluator. It is hoped that this situation will change and that soon energy values from fully documented published sources will be available for all the resonances used in energy int ere ompari sons. Acknowledgements I am grateful to the members of the INDC sub- group on Neutron Energy Calibration, J. Boldeman, F. Voss, F. Corvi, J. A. Harvey, J. Lachkar and A. B. Smith, for their invaluable advice, for their suggestions on suitable narrow resonances and for many of the energy values quoted in this paper. However, the sub-group has had no opportunity to comment on the reduced list presented in sect. 7 and it does not carry their imprimatur. Both J. A. Harvey and K. H. BSckhoff readily sent me their data on; the total cross section of D Li so that all white source measurements could be analysed by the same method. The help given by D. K. Olsen in selecting suitable resonances in ^*° T j is greatly appreciated. His list was, however, augmented slightly and the responsibility for this rests with me. It is a pleasure to acknowledge the interest taken in this work by Basil Rose who in the course of lively discussions suggested the use of the ogive curve method and stimulated the development of the AL/At method. Experiments on the synchrocyclotron enjoy the unstinting support of Colin Whitehead and the active participation of P. H. Bowen, A. D. Gadd, D. B. Syme and I. L. Watkins. Many of the energy calculations were carried out by A. D. Gadd. References 1. S. Wynchank, F. Rahn, H. S. Carmada, G. Hacken, M. Slagowitz, H. I. Liou , J. Rainwater and W. W. Havens, Jr. Nucl. Sci. & Engng.5J. (1973) 119 2. J. C. Davis and F. T. Noda. Nucl. Phys. A13^ (1969) 361 3. M. S. Coates, D. B. Gayther and N. J. Pattenden. Proc. 4th Conf. on Neutron Cross Sections and Technology, Schrack and Bowman (Eds.) NBS Spec. Pub. 425 Vol. 2 (1976) 568 4 J. W. Behrens and G. W. Carlson. Proc. NEAWDC/ NEACRP Specialist Meeting on Fast Neutron Fission Cross Sections, ANL-76-90 (ANL, Argonne, 197b) 47 5. S. Cier jacks, B. Leugers, K. Kari , B. Brotz, D. Erbe, D. Grbschel, G. Schmalz and F. Voss. Ibid. 94 324 Table 7 Narrow resonances suitable for use as energy standards Energy Range (eV) 0.1-1 1 - 10 10 - 100 Isotope Nominal Energy Order* Reference 100 - 1000 1000 - 10 000 (keV) 10 - 100 100 1000 1000 - 10 000 10 000 - 100 000 Ir-191 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 U-238 Pb-206 U-238 U-238 U-238 Al-27 S-32 Na-23 Si-28 Pb-206 Fe-56 S-32 S-32 Fe-56 S-32 S-32 0-16 C-f2 C-12 0-16 C-12 C-12 0.6551 6.672 10.236 20.8b4 36.671 66.015 80.729 145.617 189.64 311.18 397.58 i+63.18 619-95 708.22 905.03 1419.88 1473.8 2489.47 2672.2 3360 3458.1 4512.0 5650.6 5903 - (keV) 30.378 ± 53.191 ± 67-73 ± 71.191 T 90.134 97.512 ± 112.186 i t 0.0014 - 0.033 ± 0.25 ± 0.24 0.20 0.32 ± 0.5 ±10 006 027 02 018 016 028 033 266.347 ± 0.053 412.3 * 818.7 1651 2078 2818 3211.1 6293 12100 - 2 4 1.5 5 100 'Order of (P+&)/E within an Energy Range 13 12 12 12 12 12 12 Table 5 12 13 12 Table 5 12 Table 5 12 Table 5 12 Table 5 12 13 12 12 12 13 12 12 31 12 32 12 12 32 31 31 13 33 13 13 13 6. P. A. R. Evans, G. B. Huxtable and G. D. James, Ibid. 149 7. M. S. Coates - private communication 1975 8. J. W. Meadows, Ibid. 73 9. K. H. Bockhoff, F. Corvi and A. Dufrasne, NEANDC(E) 172 "U" Vol. Ill- EURATOM (1976) 17 10. F. J. Rahn, H. Camarda, G. Hacken, W. W. Havens, Jr., H. I. Liou, J. Rainwater, M. Slagowitz and S. Wynchank, Phys. Rev. C6 (1972) 1354 11. W. J. Youden, Technometrics ^4 (1972) 1 12. D. A. Olsen, Unpublished data communicated by J. A. Harvey 13. S. F. Mughabghab and D. I. Garber, BNL325, Third Ed. Vol. 1 (1973) 14. H. J. Groenewold and H. Groendijk, Physica _13_ (1947) 141 15. G. D. James, D. B. Syme, P. H. Bowen, P. E. Dolley, I. L. Watkins and M. King, Report AERE-R7919 (1975) 16. J. W. Meadows, Report ANL/NDM-25 (1977) 17. H. Derrien, Report NEANDC(E)-l6 i +-L 18. G. de Saussure, D. K. Olsen and R. B. Perez, (unpublished) 19. G. E. P. Box and M. E. Muller, Ann. Math. Statistics 29 (1958) 610 20. H. H. Ku, NBS Special Pub. 300 (1969) 21. C. T. Hibdon and F. P. Mooring, Neutron Cross Sections and Technology, NBS Special Publication 299 Vol. 1 (1968) 159 22. J. A. Farell and W. F. E. Pineo, Ibid. 153 23. J. W. Meadows and J. F. Whalen, Nuc. Sci. & Engng. 4l_ (1970) 351, 48 (1972) 221 24. C. A. Uttley - private communication 25. J. A. Harvey and N. W. Hill, Nuclear Cross Sections and Technology, NBS Special Publication 425 (U.S. Dept. of Commerce, NBS, Washington, 1975) 244 26. H. H. Knitter . C. Budtz-Jjzfrgensen, M. Mailly and R. Vogt Report EUR-5726e (1977) 27. F. Corvi - private communication and ref. (9) 28. H. T. Heaton II, J. L. Menke , R. A. Schrack and R. B. Schwartz, Nucl. Sci. & Engng. 56 (1975) 27 29. F. G. Perey, T. A. Love and W. E. Kinney, Report ORNL-4823 (ENDF-178). The value quoted is derived from the data described in this report. 30. S. W. Cierjacks et al. Report KFK-1000 (1< The value communicated by F. Voss is based on the data in this report. 31. J. A. Harvey - private communication and Report ORNL/TM-5618 32. Unpublished results from the Harwell synchro- cyclotron to be regarded as preliminary. 33- H. Weigmann, R. L. Macklin and J. A. Harvey, Phys. Rev. C 14 (1976) 1328, with errors provided by J. A. Harvey - private communication. 325 App endix 1 DETERMINATION OF NEUTRON ENERGY DIRECT METHOD L = P + D (tn - t ) + (t - t ) - (t - t ) Y Y Y n y^ fc l + fc 2 (x - x ) w + L /O. 29979 - (fast transit time o y + slowing down time delay) L = P + D + S Y Y Y Fast Neutron Source l 9o \ p t,\ *— Moderated Source D ~X > tn Detector Fig. 1 Arrangement of target and detector in the Harwell synchrocyclotron neutron time-of- flight spectrometer illustrating some of the symbols used in Appendix 1. = L/(t x 0.29979) E = 469776.3 (0 + 0.756 ) AL/At METHOD Perform experiment at L and L L l " P l + ° t l = (X 1 " X oi +6,w + L/0. 29979 - (t 3 + t 4 ) L 2 = P 2 + D t 2 = (X 2 " X 02 + 6)w + L 2 /°. 29979 - (t 3 + t^) L = P + D + S • L = P„ + D + S Yl 1 y Y Y2 2 y Y AL = P L - P 2 At = (X 1 - X Q1 ) w - (X - X Q2 ) w + (P - P 2 )/0. 29979 3 = AL/ (At x 0.29979) E = 469776.3 (g 2 + 0.75g 4 ) Note: 6 D D S and t_ + t. are removed Y Y 3 4 .8 1 9 .SACLAY II 1 ' SACLAY I > # i* 1 • z i * 9 o * • * • *< ' * * *. z < i- » * * .4 - H .2 ." v t 290 300 310 ENERGY keV Fig. 2 The transmission of sodium near the resonance at 299keV measured with good resolution, Saclay II, and poor resolution, Saclay I, as presented by Derrien^* ' '• - It illustrates the effect of resolution on the observed energy of the transmission maximum and minimum. 326 Sodium Resonance at 298-65 ±0-32 keV 6 Li Resonance at 2440 ±0-5 keV Saclay I Cadarache Karlsruhe Harwell Saclay II Columbia Average Hibd on & Mooring Farell & Pineo Meadows &Whalen Uttley James Harvey Harvey -James Bockhoff Bockhoff -James Knitter 295 300 E - keV Fig. 3 The energy of the sodium resonance at 298.65 - 0.32keV as measured in six labora- tories (X) and the values obtained by Derrien(l7) after correcting for the effect of spectrometer resolution (0). 240 Fig. 5 Cumulative Probability U-238 Resonance at 2489eV N = 13754-99±0-12 Average of TOF values 250 E - keV 6 T Energies at the peak of the °Li total cross section near 250keV. (James) denotes results obtained after fitting the data by the formulae used by Uttley and by James. The average of results obtained by time-of-flight experiments is Zhk.O ± 0.5keV U-238 Resonance at 2489-47 ±0-50eV 13760^ N 13750 (a) (b) 50m Aug '76 50m 100m Nov '76 dL/dt Average Bockhoff James Olsen Rahn 10 50 90 PROBABILITY - % 99 2487 2490 Fig. k Cumulative probability distribution for the transmission of a resonance in U-238 at 2^89eV. A least squares fit to the data plotted on probability graph paper enables the resonance timing channel N to be located to 0. 12 units. Average Fig. E -eV An illustration of the data for the resonance in the total cross section of 2 ^°U at 2^89.47 ± 0.5eV presented in table 6 327 Energy Distribution of Selected Resonances j i i i in 1 1 i i i n 1 1 ii ii iii i i ii i E-eV U U U UUPbUAl Si No Si R> f. S \\\\\ LLl 1_ E-keV S h>S SOCOC C 1 U I I I I I I I 'I I I II I I I I , III E-MeV Fig. 7 Energy distribution of narrow resonances selected as suitable for use as standards. The largest relative gaps are between 0.6eV and 6eV and also between 6keV and 30keV. 328 NEUTRON TRANSPORT CALCULATIONS FOR THE INTERMEDIATE-ENERGY STANDARD NEUTRON FIELD (ISNF) AT THE NATIONAL BUREAU OF STANDARDS C. M. Eisenhauer and J. A. Grundl National Bureau of Standards Washington, D.C. 20234 and A. Fabry C.E.N. - S.C.K. 2400 Mol, Belgium The intermediate-energy standard neutron field (ISNF) established in the thermal column of the NBS reactor is designed to produce a benchmark neutron spectrum concentrated between 10 keV and 3 MeV. Design principles and physical description of the spherically-symmetric system are summarized. Neutron transport calculations of the ISNF spectrum at the center of the facility by the discrete-ordinates method are discussed and results are given for 40-group and 240-group computations. The sensitivity of the calculated spectrum to variations in important physical parameters such as material densities and carbon and boron-10 cross sections is explored. (Benchmark spectrum; discrete ordinates; intermediate energy; measurement assurance; neutron standard; reaction rates) Motivation and Design Principles The Intermediate-energy Standard Neutron Field (ISNF) is an irradiation facility at NBS designed to produce a strong component of neutrons in the energy range of interest for fast neutron reaction rate technology. It is intended for use in the following measurement activities. 1. Measurement of integral fission and capture cross sections. 2. Long-term measurement assurance for neutron reaction rates required for the design and operation of power reactors. 3. Calibration of passive neutron detectors used in materials neutron dosimetry to characterize neutron fields associated with radiation effects testing. 4. Validation of neutron spectrometers and other energy-sensitive neutron detectors. The ISNF arrangement is simple: a spherical cavity in the thermal column of the NBS reactor, a thin shell of B lightly supported at the cavity center, and fission-source disks of U placed symmetrically around the periphery of the cavity. The ISNF field at the center of the cavity consists of a fission-neutron component and a component of neutrons returned from le graphite, both of which are attenuated by the B shell. The general properties of the neutron field at the center of the ISNF are as follows: i&! -7x10 ~ 0.5 x ^ 2% over 4 cm n cm 0.5 x 10 14 n cm Total neutron flux Typical maximum fluence Neutron field gradient Spectrum: Average energy 1.0 MeV Median energy 0.56 MeV Energy range 90% between 8 keV and 3.5 sec 2 MeV Neutron transport in the ISNF is determined almost entirely by the " J U fission neutron spectrum and by neutron scattering in carbon and neutron absorption in boron-10. In particular, the neutron spectrum at the center of the facility is governed by the kinematics of elastic scattering in carbon and by two accurately- known cross sections: ° tot f° r carbon and ° a k s f° r boron-10. Moreover, only the energy dependence of the boron-10 absorption cross section is critical because the neutron transmission of the shells is checked experimentally with a 2-keV monoenergetic neutron beam. The one-dimensional geometry of ISNF 1 permits calculation of the spectrum with an accuracy limited only by the uncertainties of input cross sections and densities of materials, and allows investigation of (1) sensitivity of the spectrum to variations in system parameters and (2) perturbation of the spectrum due to extraneous structural materials. Physical Description of Facility The initial ISNF facility has been set up in the graphite thermal column of the NBS reactor. A square opening, 30 cm x 30 cm provides access to the center of the graphite column. Split graphite blocks enclosing the ISNF cavity and containing cylindrical access penetrations may be inserted through the biological shield of the reactor and into the thermal column opening . A detailed schematic cross section of the ISNF arrangement within the cavity is shown in Figure 1. Three levels of access are available: (1) instruments and/or irradiation samples may be inserted or removed through a 5 cm dia. cylindrical channel; (2) the boron-10 shell and the graphite pieces to which it is attached slide in a 20 cm dia. cylindrical access; and (3) the entire 30 cm x 30 cm cavity block may be with- drawn or inserted with the reactor at full power. Four one of the eight ZJJ U fission source disks are indicated in symmetric array near the surface of the cavity in Fig. 1. Each disk is made of enriched uranium metal 16 mm dia. x 0.15 mm thick (93% enrichment, 0.476 or 0.511 gm each) and positioned at 1 cm from the surface of the 30 cm dia. cavity. Eight disks are used in normal operations producing a total fission neutron source strength of about 6 x 10-'--'- sec . Figure 1. Intermediate-Energy Standard Neutron Field arrangement at the center of the NBS Reactor thermal column. 329 The crucial component of the ISNF is the boron-10 shell and serious difficulties were encountered in its fabrication. Two shells of different thickness, formed as stepped hemispheres at the Los Alamos Scientific Laboratory [1] by techniques of powder metallurgy, are now available. The shells are 14.26 cm outside diameter. The thick and thin shells have thicknesses of 1.293 cm and 0.638 cm. The mean boron-10 thickness of each shell is 1.26 and 0.61 g/cm . Fabrication of the shells required that a significant amount of aluminum be added to the 95%-enriched boron-10 metal powder. TABLE I Constituent aluminum boron-10 boron-11 carbon silicon iron lead uranium Chemical and Isotopic Analysis of Boron and Aluminum Powders [1] Concentration (weight%) Boron-10 Aluminum 99.65 93 58 4 87 6 4 3 1 15 0.15 0.2 (99.3% 238 U, 0.6% 234 U) Table I presents an analysis of the boron and aluminum metal powders employed. Materials other than B in and around the shell, including a protective shell of 0.7 mm thick spun aluminum metal, add up to 29 atom percent of aluminum and 4.8 atom percent of boron-11. Rings and support rods for mounting the shell in the cavity introduce another 1.4 atom percent of aluminum into the cavity. The uniformity and absolute absorber thickness of the shells is determined by means of neutron transmission measurements in the 2 keV filtered beam at the NBS reactor. Transmission measurements in the 2 keV beam for several orientations of the shell indicate that the thickness is uniform to better than 1%. The flanged cadmium shield indicated in Figure 1, which fits over the stems of instruments placed inside the shell, prevents streaming of thermal neutrons through the hole in the shell access plug. Such stream- ing, particularly by epicadmium wall-return neutrons, is a characteristic problem for any driven, one-dimensional neutron field. Careful experimental investigation is required to demonstrate that the problem is under control. Shell access plugs without a hole are avail- able to investigate this effect with passive detectors which do not involve instrument stems. An inventory of ISNF components and their physical properties is given in Table II. Included are some characteristic nuclear parameters appropriate for the fission neutron transport problem. TABLE II Inventory and Physical Properties of Principal ISNF Components 2. Graphite Thermal Column — dimensions: — graphite density: — cavity diameter: Boron-10 Shells [1] — dimensions: inside diameter: outside diameter: thickness 1.4 x 1.3 x 0.94 m" 1.71 g/cm3 29.8 cm Thick 11 . 68 cm 14.26 cm 1.293 cm — densities: boron-10 boron aluminum 0.975 g/cm: 1.025 g/cm; 0.992 g/cm- Thin 12.99 cm 14.26 cm 0.638 cm 0.954 g/cm; 1.002 g/cm: 0.971 g/cm -thickness in mean free paths E t for neutrons (thick shell only; for thin shell multiply by 0.49) 1 MeV 0.1 MeV 25 keV 2.7 keV absorption scattering .025 .30 .145 .33 .29 .20 235, .23 3. Fission Source Disks (93.5% Enriched U Metal) — dimensions: 16 mm dia. x 0.15 mm thick total fission neutron source strength (8 disks, reactor at 10 MW) a in 11 "! ~ 6 x 10 sec Calculation of Central Flux Spectrum (ISNF-1) The main method for calculating the flux in the ISNF facility has been the discrete ordinates method. We have used the ANISNW version of this code which includes revisions by the Westinghouse Astro-nuclear Laboratory [2] . In making one dimensional discrete ordinates calculations of the ISNF system, certain schematizations must be made. Figure 2 gives the parameters which were used in the calculations for the thick shell. The linear diagram at the top shows the radial coordinates of ISNF-1. The 10 B-AX, shell has inner and outer radii Vacuum -A£ A£ Vacuum Carbon I ) IHIIIIIUIIi r(cm) 10, 5(838 7 .'131' Uf 2005 .0582 at/(b-cm); N AX. .0602 at/(b-cm) 13 '. 9 14.0 Source N = 0.0860 at/(b-cm) N A£ = .0216; N = .0038 Adjustments made in Shell Composition to preserve Total Mass for Calculation in Spherical Geometry ISNF-1 (thick 10 B shell) AX, density = .0021 x .9916 = .9838 g/cm 3 (1.27 g/cnu) 10 B density = .9747 x .9916 = .9665 g/cm 3 (1.25 g/cm ) Other = .0762 x .9916 = .0756 g/cm (0.10 g/cm ) Total density 2.0259 g/cm Volume = 4/3 tt [(7.131) - (5.838) ] = 685.5 cm Total Mass = 1388.8 g Protective AX Shell Thickness t = 0.0695 cm Inner Radius = 7.131 cm , Density = 2.70 g/cnC Mass = 4/3 it [(7.2005)- (7.131) ] 2.70 = 121 g Figure 2. Parameters used in the one-dimensional discrete ordinates calculation of the ISNF-1 configuration. 330 of 5.838 cm and 7.131 cm. This is surrounded by an A£ protective shell with an outer radius of 7.2005 cm. The source was assumed to be distributed in a 1 mm annulus with inner radius of 13.9 cm. This assumption is justified by the fact that the field at the center of a spherically symmetric system is the same for a spherical shell source as for a number of point or thin disk sources placed at the same radius and having the same total source strength. The radius of the cavity is 14.92 cm and the carbon was assumed to extend out to a radius of 65 cm. Calculations show that neutrons reflected from graphite beyond 65 cm would add less than 0.2% to the flux above 8 eV. The density of the various components in the B-A& shell were adjusted from those provided by Los Alamos Scientific Laboratory in order to preserve the total mass (1389 g) of the two B-A£ hemispheres plus the heaviest of 3 access plugs. The outer radius of the protective A£ shell was adjusted to maintain the measured mass of 121 g. Calculations with 40-group ENDF/B-III Cross Sections Cross Sections The cross sections used in calculations at NBS are based on ENDF/B-III data averaged in a 40-group energy structure. This library was derived from a 100-group set (the GAM-II structure) weighted by fine-group fluxes determined from one dimensional discrete ordinates calculations and averaged over spatial zones appropriate to the ISNF geometry. Results The three calculated spectra of interest at the center of the ISNF are shown in Figure 3. The quantity plotted is E(j)(E) vs the log of the energy E. Thus the relative area under each histogram for any energy interval is proportional to the relative number of neutrons in that interval. The dotted histogram shows the "-ty fission neutron flux at the center of the ISNF. It dominates the ISNF spectrum above a few MeV. 10 ' 10" 10 10 NEUTRON ENERGY E.MeV Fig. 3 Calculated central flux lethargy spectra for the intermediate-energy standard neutron field (ISNF-1) , the fission neutron source spectrum, and the cavity spectrum with no 10 B shell (ISNF/CV) . The dashed curve shows the central spectrum (designated ISNF/CV) resulting from the addition of neutrons returned from the graphite. The spectrum below 1 MeV is greatly enhanced by the neutrons slowed down in the graphite, and below ~ 1 keV the spectrum extends smoothly down to thermal energies with a near-l/E behavior. This spectrum is realized for experiments by replacing the ISNF 10 B shell with a cadmium shell. The solid line shows the spectrum at the center of the ISNF facility, with the thick 10 B shell in place (designated ISNF-1). Here, neutrons below ~ 1 keV have been attenuated severely by absorption in -^B. The corresponding group fluxes - ) U/ J U cross section ratio that is about 10% different from predictions of ENDF/B-IV tabulations. These measurements are discussed in another paper at this conference [4]. Similar discrepancies have been found for other fast reactor spectra. Sensitivity Studies Since the spectrum at the center of the ISNF is established by means of calculation, it is essential to assess the sensitivity of both the spectrum and integral detector responses to variations in physical and geo- metric parameters of the system. B Absorption The most important parameter for determining the shape of the ISNF spectrum below about 10 keV is the absorber thickness of l^B in the B-Ai shell. A given uncertainty in either the 10,: cross section a(b/at) or the mass thickness X(at/b) produces a larger uncertainty in the flux at low energies than at high energies because of the exponential nature of trans- mission of neutrons through the shell. In the calculation of ISNF-1 we have used a mass thickness of 1.25 g/cm 2 of 10 B and ENDF/B-III absorption cross sections. An uncertainty of 4% in the product of cross section and thickness produces an uncertainty of about 3% in the flux between 1 and 10 keV. With the 2 keV beam-transmission measurement as verification of shell thickness in mean free paths, the uncertainty in this quantity is judged to be less than + 2%. Graphite Density Present measured values of the graphite density exist only for removable thermal-column stringers, because the bulk of the thermal column cannot be . disassembled. Results vary from 1.67 to 1.71 g/cm , while the nominal value used in ISNF calculations is 1.71 g/cm 3 . When the graphite density is increased by 3%, the flux in the cavity increases gradually from 0.1% at energies above 6 MeV to 2.7% at epithermal energies. This can be interpreted as a nearly uniform 2.7% increment in the cavity-return component of the flux. The corresponding increment in integral response 238 U of a 235 U fission foil is 1.4%, while that of a foil is 0.5%. Since a density change is equivalent to a change in total cross section, the error in the total cross section of carbon, presently set at + 0.5%, would account for proportionately smaller uncertainties in flux and detector response. 331 Group 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 TABLE III. Group Fluxes c(>(E) AE at Center of ISNF pper Energy Fission ISNF/CV Bound (eV) Spectrum (no 10 B shell) 1.4918+07 1.03250-5 1.08436-5 6.0653+06 4.47391-5 5.39036-5 3.6788+06 4.68987-5 7.01945-5 2.7253+06 3.64821-5 7.34228-5 2.2313+06 9.21736-5 2.00399-4 1.3534+06 3.22965-5 8.00682-5 1.1080+06 4.04492-5 1.17128-4 8.2085+05 2.18288-5 7.32631-5 6.7206+05 2.61317-5 1.03554-4 4.9787+05 1.34049-5 6.35855-5 4.0762+05 1.51932-5 8.79786-5 3.0197+05 1.04694-5 7.86573-5 2.2371+05 4.84693-6 4.75933-5 1.8316+05 5.01832-6 6.50756-5 1.3569+05 2.27986-6 3.94921-5 1.1109+05 2.03273-6 4.57084-5 8.6517+04 1.39449-6 4.20754-5 6.7380+04 9.57525-7 3.90805-5 5.2475+04 6.54398-7 3.65123-5 4.0868+04 4.51031-7 3.43059-5 3.1828+04 3.08527-7 3.23951-5 2.4788+04 2.12484-7 3.07273-5 1.9305+04 1.45919-7 2.92581-5 1.5034+04 9.99573-8 2.79549-5 1.1709+04 6.87815-8 2.67891-5 9.1188+03 7.96820-8 5.02952-5 5.5308+03 3.75963-8 4.69476-5 3.3546+03 1.77700-8 4.40797-5 2.0347+03 8.38826-9 4.15821-5 1.2341+03 3.96618-9 3.93778-5 7.4852+02 1.87239-9 3.74039-5 4.5400+02 8.84187-10 3.56270-5 2.7536+02 4.17368-10 3.40149-5 1.6702+02 1.97399-10 3.25422-5 1.0130+02 9.31675-11 3.11880-5 6.1442+01 4.40192-11 2.99415-5 3.7266+01 3.06191-11 5.70116-5 1.3710+01 4.64120-12 2.67281-5 8.3153+00 4.10660-12 1.51313-4 4.1400-01 ISNF-1 (thick 10 B shell) 9.81571-6 4.89605-5 6.30993-5 6.66930-5 1.86659-4 7.60595-5 1.10525-4 6.86174-5 9.41721-5 5.70434-5 8.00351-5 7.08074-5 4.16005-5 5.39625-5 3.39017-5 3.39677-5 3.54945-5 3.02092-5 2.62202-5 2.02771-5 2.32677-5 1.98570-5 1.76938-5 1.57295-5 1.40237-5 2.28631-5 1.78851-5 1.31942-5 9.28015-6 6.11757-6 3.68933-6 1.97961-6 9.11946-7 3.42626-7 9.82393-8 1.95149-8 1.43028-9 6.66496-12 2.32821-11 Reaction 239 T TABLE IV Spectral Indexes for ISNF and Fission Spectra 235 Pu(n,f) 238 237 U (n,f) Np(n,f) 0(x )/a( 235 U) U Fission ISNF-1 1.435 1.113 0.238 0.0843 1.064 0.480 ISNF/CV 1.205 0.0111 0.0635 Source Radius The flux at the center varies somewhat with source radius. The main sensitivity occurs at high energies where uncollided neutrons predominate. Flux changes at these energies are largely associated with the inverse-r^ dependence. The variation below about 800 keV is generally less than 0. 6% when the source radius of 13.95 cm is varied by + 2 mm. The normal position of the sources is 1 cm from the cavity surface with an estimated uncertainty of less than 2 mm. Scattering Anisotropy in Graphite The effect of the angular distribution of scat- tering in graphite was examined by repeating the ISNF calculation with truncated Pn scattering cross sections. The ratios of fluxes calculated by trun- cating after P rather than P.,, as in the usual ISNF calculations, indicate that the Pi and Pj components are important in determining the intensity of the scalar flux at the center. The ratio of fluxes with truncation after P vs. after P, remains constant at about 1.06 at low energies and does not reach 1.10 until a neutron energy of a few MeV is reached. The ENDF/B-III Legendre harmonic coefficients for elastic scatter in carbon, in particular the Pi component between 0.35 MeV and 1.35 MeV, which are responsible for lowering the flux of neutrons returned to the center of the ISNF, compared with that obtained for isotropic scatter, have not changed significantly in the ENDF/B-IV tabulation. For example, the elastic scatter in the backward direction O(-l) changed by only 1% at an energy of 2 MeV. Therefore difference in the ISNF central flux due to differences in scattering cross sections for carbon in ENDF/B-III and ENDF/B-IV are not expected to exceed 1%. 332 Aluminum in Boron Shells 10„ The fabrication of the B shells required that a significant amount of aluminum be included with the l^B. In order to assess its effect a calculation of the spectrum at the center of the ISNF was performed with all aluminum removed. Fluxes with and without aluminum generally differ by less than 4% with the exception of energy groups where aluminum resonances scatter neutrons preferentially into an adjacent group. Variations of the spectrum due to uncertainties in the amount of aluminum actually present in the ISNF would be much smaller. The effect will be discussed in more detail in the next section. Table V presents a summary of the effects of varying the parameters discussed above. TABLE V. Summary of Sensitivity Studies Percent Effect Variation Group Fluxes Total Flux Cross Section o(n,f) Parameter 1-10 keV 10-100 keV 0.1-1.0 MeV 0.4 eV-18 MeV 235 u 238 u B Thickness + 4% - 3.3 - 1.4 - 0.5 - 0.8 - 0.9 + 0.8 Graphite Density + 3% + 2.4 + 1.9 + 1.2 + 1.1 + 0.3 - 0.4 Source Radius + 2 mm - 0.3 + 0.2 + 0.0 - 0.2 - 0.1 - 0.6 Angular Distribution of Scatter in Carbon P., ■*■ P 3 o + 6.2 + 6.7 + 9.5 + 8.3 - 0.6 - 0.3 Aluminum in 10 B Shell 29 ■*■ atom% + 1.9 + 1.8 + 0.1 + 1.0 + 0.2 + 1.2 Cross Sections and [ENDF/B-III + IVJ + 1.1 + 0.4 + 1.5 + 0.7 - 0.3 - 1.2 Group Structure [ 40 ■*■ 240 J Results for 240-group Calculations Two important limitations of the 40-group calcu- lations are the accuracy of the ENDF/B-III cross sections and the coarse structure of the energy groups in which the cross sections and the resultant spectra are tabulated. In order to investigate these limita- tions calculations were performed by LaBauve and Muir [5], using ENDF/B-IV cross sections in a 240-group energy structure. They used exactly the same input parameters for dimensions and material concentrations as used in the 40-group ENDF/B-III based calculation. Table VI shows a comparison of the two calculated spectra in terms of the integral responses predicted for the total flux and several fission detectors. The largest discrepancy is 1.5% for a 238jj fi ss i n detector. This is a rather remarkable invariance since the 40-group calculation contains only 4 groups in the energy range 2 38 in which 90% of the U response occurs. One reason for the good agreement may be that the integral responses are calculated with the DETAN code, which transforms the histogram flux spectrum into a series of linear segments with interpolated slope in flux-vs. -lethargy. This comparison of coarse and fine group structure is mainly a check on the group-averaging techniques used in discrete ordinates calculations, since the difference in the total cross sections of boron and carbon cross sections for ENDF/B-III and ENDF/B-IV is negligibly small. Significant differences in the angular distribution of scattering in carbon exist in the two data files, but they occur above 5 MeV where the impact on the ISNF spectrum is negligible. Figure 4 shows the 240-group central ISNF-1 flux spectrum calculated by LaBauve and Muir. It is similar to the 40-group spectrum shown in Figure 3, except that the resonance structure due to the aluminum in the B shell is now resolved. The effect of this structure TABLE VI Comparison of Calculated Integral Cross Sections for 40-Group and 240-Group ISNF Spectra o(b) ENDF/B-III ENDF/B-IV Percent NBS 40-Group LASL 240-Group Difference Flux .0013749 .0013850 + .7% 235 U(n,f) 1.6362 1.6313 - .5% 238 U(n,f) 0.13798 0.13632 - 1.5% 239 ZJy Pu(n,f) 1.8176 1.8219 + 0.2% 237 Np(n.f) 0.7834 0.7855 + 0.3% ISNF CENTRAL SPECTRUM 2W-GR0UP CALCULATION Fig. 4 Calculated Central flux lethargy spectrum for ISNF using 240-group cross sections. 333 i on reaction rates whose energy dependence is smoothly varying in this region or for which only a small part of the response occurs in this region will be small. For example, the effect of the aluminum in the shell on the total flux is 1%; the effect on integral cross sections for wide-energy-range detectors such as 235 U(n,f) and 239 Pu(n,f) is less than 0.2%. The effect ? 3 A on threshold integral detectors such as U(n,f) epends generally on the threshold energy. For U(n,f). for which 95% of the response lies above 1.5 MeV, the effect of the shift in spectrum due to downscatter is 0.8%. The effect of the aluminum on the flux spectrum can be interpreted almost entirely as a redistribution of elastically scattered neutrons in a localized energy range. In order to demonstrate this hypothesis, the detailed behavior of the ISNF spectrum near an aluminum resonance was calculated in a simple single scatter analysis. This was done by assuming that the effect of the aluminum could be approximated by concentrating it in an aluminum shell with an equivalent number of atoms per square cm placed just inside the boron shell. The scattering problem then is treated as resonance- energy removal of neutrons directed radially inward and inscatter of the others with a spectrum degraded in energy by the average logarithmic decrement for elastic scatter (E, = 0.072). The expression for the ratio of the spectra with, and without, aluminum estimated by this procedure is ISNF co/Afc( ISNF-1) ISNF u)/oAJ£(ISNF/NA) = 1 - (1 e ) + e T . J e£ l Conclusions The spectrum of the ISNF at NBS has been thoroughly studied by means of discrete ordinates transport calcu- lations. Sensitivity of the spectrum and integral detector responses to variations of all significant physical parameters of the system has been investigated. For parameter variations considered realistic, spectrum changes in a coarse energy-group description are mostly less than 1% and never more than a few percent. A quantitative assessment of the accuracy of the neutron spectrum calculated for the ISNF will be based on the sensitivity studies described in this paper, combined with a final estimate of uncertainties associated with the corresponding physical and nuclear parameters. The initial goal is to achieve a coarse (about 40 groups) specification of the spectrum that is accurate to better than + 5% in the energy range 0.4 keV to 8 MeV. The lower bound corresponds to an energy below which the response of a 1/v-detector is less than 5% and the upper bound to the limit of good fission spectrum spectrometry data [6] . Acknowledgements We wish to thank H. Scheinberg and J. Kostacopoulos for their efforts in the extremely difficult task of fabricating the boron-10 shell, and R. LaBauve and D. Muir for their 240-group discrete ordinates calcu- lations of the ISNF spectrum. References Figure 5 shows the results of applying this single scat- tering analysis to the prominent resonance at 35 keV. [The region from 20 keV to 50 keV accounts for 6.5% of the ISNF flux.] In this figure we see the histogram representation of the smoothly varying ISNF/NA spectrum, the negative removal contribution and the positive inscatter contribution. The figure shows that the spectrum "ISNF(CALC)" obtained by single scatter analysis gives a very good representation of the ISNF-1 spectrum calculated by discrete ordinates method. The differences indicated by cross hatching in figure 5 account for about 1% of the ISNF flux between 20 keV and 50 keV. [1] H. Scheinberg, Los Alamos Scientific Laboratory (Private Communication) . [2] R. G. Saltesz, "Revised WANL ANISN Program User's Manual", Westinghouse Astronuclear Laboratory Report WANL-TMI-1967, April, 1969. [3] C. Eisenhauer and A. Fabry, "DETAN 74: Computer Code for Calculating Detector Responses in Reactor Neutron Spectra", available upon request from the National Bureau of Standards. [4] D. Gilliam, et al., this conference. I.2.I0" 4 [5] R. LaBauve and D. Muir, Los Alamos Scientific Laboratory (Private Communication) . [6] J. A. Grundl and C. M. Eisenhauer "Fission Spectrum Neutrons for Cross Section Validation and Neutron Flux Transfer", NBS Special Publication 425, October 1975. Fig. 5. Comparison of discrete-ordinates spectrum (ISNF-1) with spectrum (ISNF(CALC)) calculated by considering effect due to removal and in- scatter in aluminum on slowly varying spectrum (ISNF/NA) for no aluminum. Hatched area indicated difference of two spectra. 334 STANDARDIZATION OF FAST PULSE REACTOR DOSIMETRY A. H. Kazi Army Pulse Radiation Facility, Aberdeen Proving Ground, Maryland 21005 E. D. McGarry Harry Diamond Laboratories, Adelphi, Maryland 20783 D. M. Gill iam National Bureau of Standards, Washington, D.C. 2023** A dosimetry method, developed at the National Bureau of Standards and known as the Flux Transfer Technique, is proposed for accurately determining the total fast flux in the vicinity of a fast-pulse reactor or bare-critical assembly. The method is to determine the fast flux from the comparison of free-field 239p u fission-rate measurements made at the reactor facility to calibration measurements made in a standard 2 ->2r_f ne utron flux. Use of the technique at the Army Pulse Radiation Facility shows that total fluxes can be measured in and near the reactor to a determinable accuracy of -5%. For comparison, but with less accuracy of -8% , the same total fluxes were verified using 23/Np ; 23 i *y > 236y anc | 23o|j The technique has several important advantages. It uses a recognized standard neutron source for calibration. Accurate fission rates are measured and compared with dual-isotope fission chambers that are easily calibrated. The method is independent of errors in foil masses. The method does not require the use of an unfolding code and the effects of cross section errors are lessened because of the need to evaluate only cross section ratios. The method is simple and is readily amenable to absolute error analysis and inter laboratory ca 1 ibrat ions. (Pulse reactor calibration, neutron flux standard, dosimetry, Ca 1 i forn i um-252 , radiation effects) Introduct ion Fast Pulse Reactors Fast pulse reactors are bare, all-metal fuel critical assemblies that are operated either at steady state power or at super-prompt critical ity to produce near-fission-spectrum neutrons in a microsecond time frame. ^ At present there are six such reactors operating in the US. They serve as sources of neutrons for a wide variety of DOD, ERDA and NASA programs, and are used extensively for TREE tests. There are two non-US reactors which fall into this class: The French reactor CALIBAN5 and a fast-pulse reactor in the USSR." These reactors have the least complex physical structures of all the operating nuclear reactors, and their small sizes and lack of moderators enable them to be positioned far from neutron reflecting or moderating surfaces. Also, many of these reactors have simple geometry in-core irradiation facilities or "glory holes". All these conditions facilitate calculating spectra for direct comparison with measurement. Since most current fast pulse reactors are made of the same uranium-molybdenum fuel alloy, the in-core and free-field leakage spectra of these assemblies are expected to be nearly iden- tical. Neutron Dosimetry Requirements There are three key neutron dosimetry require- ments for fast-pulse reactors: (1) accurate measure- ment of the total neutron fluence that is delivered to a particular location; (2) high-resolution measure- ments of the neutron spectrum; (3) known energy dependence of the response functions characteristic of the particular phenomena being investigated. The first two requirements are fundamental and the first is integrally related to the second through the spectrum-averaged cross sections of the neutron detectors. However, the third requirement encompasses more than neutron dosimetry, as discussed so far, and includes the response functions of the experiments of interest. Typical examples are neutron dose in tissue for biological experiments and atom-displacement damage effects in semiconductor electronics for experiments involving Transient Radiation Effects in Electronics (TREE experiments). For such experiments a spectrum measurement or calculation is required to determine a given response per unit neutron exposure. This is obtained by integrating the response function over the spectrum. For fast-pulse reactors a few reference spectrum determinations are adequate since for many response functions of practical interest the neutron f 1 uence-to-response conversion factors are fairly weak functions of variations in fast-pulse reactor neutron spectra. This was confirmed in a sensitivity analysis performed at the Army Pulse Radiation Facility (APRF) , which examined the sensitivity of various neutron response functions to realistic variations in neutron spectra encountered at fast pulse reactors.'' Therefore, the key output characteristic of these reactors is the total neutron fluence which must be accurately determined for each exposure at each location. Present dosimetry is such that absolute errors in the fluence determinations are so dependent upon complex procedures and calibrations that accuracies are unknown. Furthermore, different dosimetry tech- niques give results that vary by 501 for the same expos u re. °' ' ® Obviously there is need for a standard calibration procedure. Such a standardization technique is described herein. The method was developed by the National Bureau of Standards (NBS) , and has been used at APRF'2 to measure total fluxes to an accuracy of -5%. 335 Flux Transfer Technique Fission Chambers Basic Method 239 c Basically, the method uses a Pu fission chamber as a flux-measurement standard after it has been cali- brated against a ^52Qf source. The neutron-energy spectrum of a properly encapsulated *"£f deposit is the most accurately known neutron reference source avai lable. '3, 1** The fission cross section of 239p u is energy independent to within tS% between 10 keV and 5 MeV.5 In the absence of extraneous moderators, the fraction of neutrons with energies below 10 keV in and near a fast-pulse reactor is insignificant. Furthermore, greater than 97% of the neutrons in a fast-pulse reactor spectrum, like those in a 252 Cf spectrum, have energies between 10 keV and 5 MeV. All these factors, taken together, are the basis for the standardization technique. Detai Is of the Method Let " r 1 ' refer to a reactor spectrum and "s" refer to the standard 252r,f spectrum. For a given isotope the ratio of the fission rates, F r /F s is: F a r _ _r_ r F ~ a ' s s s where (o r /o s ) is the ratio of the spectrum-averaged fission cross sections for the isotope of interest in the reactor and 252r,f spectra, and is the flux. Note that the mass of the isotope has cancelled out of the ratio. Now let K = ■=?■ s F s The total flux, r , at the reactor is therefore 9 ^ / derived from the known '' Cf flux, S , by F • — r o r Note that the equation for r only involves the ratio of the spectrum-averaged cross sections. This ratio is more accurately known than the absolute value of either cross section, because errors in cross-section magnitudes tend to cancel from the ratio. Since the 239p u cross section is nearly flat (energy independent) over the energy range of interest, the ratio is very close to unity, typically \ .Ok for a fast pulse reactor. This says that the 239p u Flux Transfer Technique is insensitive to the shape (energy dependence) of the spectrum between 10 keV and 5 MeV. However, the cross section of 239p u j s certainly not energy independent much below 10 keV. The method must therefore account for the presence of low-energy neutrons caused by structural or other extraneous material near the reactor. Herein, the technique is expanded to provide for several capabilities. First, by virtue of their dua 1 -detector construction, the fission chambers can provide accurate ratios of various isotopic fission rates. Such ratios give information about the spectrum and are most useful in detecting spectral changes. Furthermore, the chambers may be covered with cadmium or boron-10 to demonstrate that the bare-chamber response has not been compro- mised by low-energy neutrons. The fission rates are measured with dual fission chambers and a triple-sealer pulse-processing system developed by NBS. The chambers have been described in detail '' and have been extensively evaluated at a variety of radiation sources .' °> ' 7 Figure 1 shows a fission chamber with and without a boron-10 cover. Figure 2 shows an enlarged cross-sectional view of the chamber construction. To reduce neutron absorption and scattering effects, the electrodes and structural elements are made of aluminum and insulators are made of a hydrogen-free polymer (Teflon). Five isotopes were used for the present measure- ments: 239p u; 237 Np; 23^. 236 U; and 238y A) , deposits were 12.7mm-dia oxides vacuum evaporated on polished nickel or platinum disks of 19.1mm-dia and 0.33mm thickness. Isotope data and foil masses are summarized in Table I. As seen in Fig. 2 the NBS double-fission chamber contains two independent fas t- ion i zat ion chambers. The backings of the thin-fission deposits are back to back so that the deposits are only one millimeter apart. The chambers are operated at about 135 volts with con- tinuous methane-gas flow. The electrical signals from the chambers have the voltage pulse height shown in Fig. 3- Note the very broad fission-fragment peak that is well separated by a low valley from the noise and alpha background, near zero pulse height. In the absence of undue electronic noise, the magnitude of the valley region is proportional to the thickness of the fission foils. The absolute efficiency of each fission chamber is near 100% for foils that are thin compared to the range of a fission fragment. Small corrections are described in Ref. 11. Cal ibrat ion Use was made of the NBS y Cf Fission-Spectrum Irradiation Facility. '•' ^ This 252r,f neutron source is a O.l-cm' deposit and singly encapsulated in a 0.3^-cmJ steel container. The location of the *"cf deposit in the capsule is known to within -0.5mm relative to the surface. The source strength in February 1976 was 3-3x109 n/s - 1 . 3% - The *52cf neutron source strength was determined' ° >' 9 with a manganous sulfate bath relative to the internationally compared radium-beryllium photoneutron source, NBS- 1 , presently known to tl . \%. For calibration, the fission chambers were mounted, two at a time, 106-mm apart on a light aluminum frame on opposite sides of the "2r,f source. This is shown schematically in Fig. k . By placing the source between two similar detectors in carefully controlled geometry, positioning errors were held to within 1.5% of the source-to-detector distance and, therefore, represent only 3% uncertainty in the 252r,f calibration flux at the fission chamber. Repositioning of the source relative to the chambers can be accomplished to -0.5mm. Prior to irradiation, all important distances can be measured without the source in place. Chamber-to- chamber distances are determined optically to t0.5mm on a machinist's bench. Fast Pulse Reactor Dosimetry Evaluation The Army Pulse Radiation Facility (APRF) The APRF reactor is representative of current generation fast-pulse, or as they are often called, fast-burst reactors. It consists of an unmoderated, 336 unreflected cylinder 22.6cm in diameter by 20cm high, made of 901 uranium (93-21 2 35n) and 10% molybdenum. The reactor is suspended from a transporter so that it can be located at floor level or as high as 1 km above a borated concrete floor anywhere within, or 30m outside, a 30m dia by 26m high, low-scattering weather shield. Fig. 5 shows the reactor at 6m above the floor inside this reactor building. The APRF can be operated with two in-core irradiation facilities, called glory holes. The smaller glory hole is 38mm in useable diameter and 20cm long. The larger glory hole is 106mm in useable diameter and also 20cm long. When using the larger glory hole, the reactor is operated with reflector control. Total Flux Measurements 239 P The fluxes determined from ■'-'Pu fission-rate measurements are summarized in Table II. The errors quoted in the table result from a 1.3% error in the 2 ^ 2 Cf source strength, a 2.5% error in APRF power determination, a 3-3 to 3-9% combined error in the 2 5 2 Cf calibration factors and APRF fission-rate measurements, and a 0.71 error in effective cross- section ratio. This last error was determined from a sensitivity analysis in which the 2 ->^Cf spectrum was shifted by ±30 keV and the APRF spectra by -0.1 MeV and the corresponding effects on the spectrum- weighted cross-section ratios were calculated. The resultant error of 0.7% for 2 39p u \ s small, because the J "Tu fission cross section is flat over the energy range of interest. As mentioned, this was the principal reason for selecting 23 °Pu as the flux-standardization material. Flux Measurements With Other Fissionable Isotopes asurements were also made using 2 ~'Np, and - 1 U. The average fluxes obtained at Flux mea; 23 / * U) 236u and -->-y_ The average . the three locations were 1.22x10? - 7-3%, 1.00x10° ± 7-3% and 7-91xl0 7 t 7.3% neutrons/cm 2 -sec-watt respectively. The estimated errors in these measure- ments are based on data similar to the 2 39pu data, except that the cross-section ratio error is estimated at about t(>%. This larger error is because approxi- mately k0% of the neutrons in the APRF spectrum are in the energy range 600 keV to 1 . 5 MeV where the cross sections of these other fissionable isotopes are not energy independent like 2 3°Pu. Indeed, these other isotopes are nearly threshold functions with the 23 'Np and 2 ^u thresholds near 600 keV, the 2 3 6 u threshold near 0.95 MeV and the 238 U threshold near 1.5 MeV. However, the total fluxes measured with all isotopes, including iJ - 7 Pu, agree to well within the estimated errors. This lends credence to the validity of the basic flux transfer data given in Table II. The implication is that the spectrum-averaged cross- section ratios and hence the calculated APRF spectra are satisfactory for use with detectors with threshold- type response functions. Effects of Cadmium and 'Ob^C Shields Fission-rate measurements were made with a 0.8mm thick cadmium-covered fission chamber containing 2 3°p u and Z 3/Np at all three locations. The cadmium cover should eliminate any thermal neutron response of the 239 Pu. The measured fission rates differed by less than 0.2% from bare-chamber results, indicating the absence of any detectable thermal -neutron contribution. Measurements were also made with a boron-covered 239p u /237n p chamber (2.2 g/cm 2 thickness of ,0 B). This boron shield should effectively remove all neutrons below 1 keV. The measured 2 39p u /23/Np fission-rate ratios were 2% below the bare chamber results. These data confirm that the APRF spectra at the selected locations have a very small low-energy (< 1 keV) component. Additional verification of this fact is given in a later section. Fission-rate measurements with boron-covered 237 Np in the 106mm dia glory hole are essentially a measure- ment of the attenuation of the flux due to the boron. This was determined to be 1^.1%. Spectral Indices Table IN ^compares measured^and calculated spectral indices Because of ferent isotopes col 1 ect ion of system, one the same time ndices were measured at of the spectral t the 2 52 C f are 1 .35 - 2.5% the clo in each f i ss ion obta i ns as meas determi APRF an i nd i ces source . and k.] C III L.UIIIUO I C3 IUCCJSUICU anu LdlLUIdLCU ^pcLLI dl for 2 3°Pu relative to 2 ^'Np and ^^'Hp relative at three APRF exposure locations se, fixed position of the two dif fission chamber and simultaneous rates on a dual-sealer counting accurate fission-rate ratios at ured fluxes. The APRF spectral i ned from the fission-rate ratios d two independent determinations for 239 Pu/ 2 37n p and 237 Np/ 23 °U a The indices for 2 52cf neutrons 6 t 2.7%, respectively. 20 The data of Table III show that transport-calculated spectra and ENDF/B-IV cross sections give cross section ratios reasonably consistent with experimental data. The sensitivity of the ratios is, to first order, pro- portional to the difference in the low-energy thresholds of the two isotopes in the ratio. Thus agreement is expected to be better for the 2 3/Np/ 23 U ratio, with 600 keV and 1.5 MeV thresholds, than for the 2 39p u / 2 37 Np; which is primarily sensitive to spectrum behavior below 0.6 MeV. The 2 37n p /238u measured and calculated results are within the uncertainties of the measured results. The errors on the calculated ratios are not known as they involve absolute uncertainties in the cross sections as well as spectra. The disagreement between measurement and calculation for the 2 39pu/ 2 37Np ratio is not due to thermal neutrons, as indicated by measurements with cadmium and boron covers. For leakage-spectrum measurements, one could argue that the difference is due to the presence of neutrons scattered from external core-support structures. However, because the in-core ratios are also higher than calculated and not subject to spectral perturbations from external structures, the discrepancy is more likely to result from the ca 1 cul at ional model. Further Application The evaluations, thus far, indicate that the pro- cedures, when calibrated against a recognized-standard 2 5 2 Cf source, can provide a practical and acceptable technique for accurately determining total fluence and spectral indices at fast-pulse reactors. The fission chambers are not presently designed to operate in pulsed irradiations; however, there is no evidence of spectral differences between pulsed and steady-state reactor operations. In practice, fast-neutron pulsed fluence levels are determined from the 3 P beta radio- activity induced in sulfur. Because the sulfur reaction has a high-energy threshold, sulfur-monitored result is usually reported 2 ' as fluence with energy >3 MeV. The sulfur fluence is then multiplied by a previously determined (>10 keV/>3 MeV) flux ratio to yield the total fluence. In the absence of extraneous moderator, the number of neutrons with energies below 10 keV is insignificant. The required flux ratio is determined using a number of techniques including transport calculations, spectrum measurements, fission- foil analysis and spectral unfolding. The resulting flux ratios vary considerably, from 6.7 to 10. 337 This spread is not due to real differences, in reactor spectra, but is the result of differences in measure- ment techniques. It is evident that there is also a need for a standard calibration procedure for sulfur- fluence monitoring. Sulfur Fluence Measurements For steady-state irradiations, 2 - > Cf can be used to provide a flux-transfer calibration for sulfur by the fol lowi ng : Jt = C r r where 0t r is the fluence in spectrum "r" and C refers to the respective sulfur responses (e.g., counts/time on the beta counter used to measure the sulfur activity). The cross section ratio a s /a r refers to the ratio of the sulfur cross section integrated over the entire energy range of the spectrum. A separate ^ Cf calibration of the sulfur resulted in fluxes of 8.0x10' n/cm 2 *s'w, 1.0x10° n/cm •S'w and 1.2x10' n/cm 2 -s*w, respectively, for the large- and small-glory hole and reactor-surface locations. There is very good agreement between these results and the total fluxes reported in Table II. The estimated error on the sulfur-monitored fluxes is ±7% and is composed of ±}>.k% from the sulfur dosimetry, -2.5% from power level normalization, and -(>% on the sulfur cross section ratio. The good agreement between the sulfur-flux values and the flux-transfer data suggest that the >3 MeV portions of the calculated APRF spectra are quite good. Spectrum Measurement Further confirmation of the quality of the calcu- lated leakage spectrum is given by results of proton- recoil measurements, recently completed" and shown in Fig. 6. The uncertainties in the measured results in the 200 keV to 2 MeV range are approximately ±81. Uncertainties below 200 keV increase from about -\SZ at 100 keV to ^35% at the low-energy end. Assuming the calculated results (solid curve in Fig. 6) above 2 MeV, the measured-to-calculated total flux ratio above 30 keV is only 1.035. There is 21% more measured flux in the 30 keV to 200 keV interval but only 1% of the total flux is below 200 keV. This measured difference only accounts for half of the total difference in the previously discussed z 39p u / 2 3'Np spectral index. >3 MeV Sulfur Dosimetry Although it is not necessary for the sul f ur-f 1 uence measurement technique, it is of interest to extend the procedure to sulfur measurements of >3 MeV fluence to better evaluate the past history'^ of the 10 keV/3 MeV controversy . Based on present knowledge of the 2 -> 2 Cf spectrum, the fraction of 252 Cf flux >3 MeV is 0.237- 7 The present total -fl uence sulfur-calibration results are corrected to reflect cross sections >3 MeV in the 2 "cf anc | /\PRF spectra. Such adjustments yield (>10 keV/>3 MeV) flux ratios for both glory holes and the leakage spectra of 8.8 +0.6 and 1 .h ± 0.5, respect i vely . Conclusions • 252 standard Cf source, can provide a practical, acceptable technique for accurately determining total fluence and spectral indices at fast pulse reactors. By calibrating with the 2 -> 2 Cf source accurately positioned between two high-quality, thin- deposit fission chambers, the uncertainty in the 2 ^ 2 Cf calibration flux at the deposits can be held to -2%. Subsequent measurements at the APRF, with 2 '°Pu fission deposits, show that free-field fluence at a fast-pulse reactor can be calibrated to an absolute accuracy of t-5%. Additional experiments with cadmium and '^B/,C covers confirm the applicability of the Flux Transfer Technique for the situation where the neutron flux is free of low-energy neutrons. Also these results show that when the spectrum under investigation is similar to a fission spectrum, a direct neutron-flux transfer can be undertaken. For other classes of spectra, computed fission-spectrum-averaged cross sections are required to obtain a flux in the spectrum under investi- gation. In both cases, the neutron Flux-Transfer Technique relaxes the requirements to establish absolute activation detector efficiencies and uncertainties associated with absolute cross section scales . Comparisons of the Flux Transfer Technique were made with 2 ^'Np, several uranium isotopes and the 3 2 S(n,p)32p reaction all calibrated against 2 5 2 Cf. Total fluxes from all measurements agree to within -7%. The technique provides several other capabilities. First, by virtue of their dual-deposit construction, the fission chambers give accurate ratios of various isotope fission rates at specified locations near reactors. Such ratios provide information about the neutron-energy spectrum and are most useful in detect- ing spectral changes that result from perturbations by structural materials or experiments. Second, the method may be used to establish secondary standari- zation of activation foils, which are suitable for neutron dosimetry during pulsed reactor operations. Such secondary standardization can be used to calibrate data obtained with unfolding codes. Reactor neutron dosimetry includes questions of both the magnitude (total fluence) and the shape of the neutron spectrum. Sensitivity analyses, performed at APRF, indicate that for many applications the question of total fluence is paramount. This problem is satisfactorily solved with good accuracy by the present flux transfer technique. Since essentially all fast pulse reactors are similar in composition and con- struction, the method is proposed as a means of inter- calibrating all fast-pulse reactor dosimetry. It has the tremendous advantage of being an easy task to repeat, thereby permitting periodic restandard i zat ion and i nter 1 aboratory comparison. The method is con- ceptually simple, accurate and readily amenable to absolute error analysis. Acknowl edgments The evaluations at APRF indicate that the pro- cedures, when calibrated against a recognized Credit is due to D portion of the calibra Mr. Thomas Wright for and computer analyses, forming the sulfur cal Mr. Charles Eisenhauer section data and calcu shifts. We appreciate fabrication of fission ware and equally appre ions given by Dr. Jame support of the APRF st r. V. Spiegel for doing a large tion work at the NBS 2 5 2 Cf source, assistance with data collection and Lt George Davis for per- ibrations. We also thank for obtaining ENDF/B cross ating the effects of energy Mr. Robert Dallatore's careful chambers and irradiation hard- ciate the many helpful suggest- s Grundl and the enthusiastic aff in operating the reactor. 338 References 1. Proceedings of the National Topical Meeting on Fast Burst Reactors, Albuquerque, N.M., January 1969, CONF-690102 (December 1 969) . 2. J. T. Mihalczo, "Superprompt-Cr i t ica 1 Behavior of an Unmoderated , Unreflected Uranium-Molybdenum Alloy Reactor," Nuclear Science and Engineering 16:291-298 (July 1963). 13. J. A. Grundl ej^. a_l_. , "A Cal i forn i um-252 Fission Spectrum Irradiation Facility for Neutron Reaction Rate Measurements," Nuclear Technology 32 (3) (March 1977). 14. H. T. Heaton II, et. aj_. , "Absolute 235 U Fission Cross Section for - 2"->'Cf Spontaneous Fission Neutrons," Proc. of Conference, Nuclear Cross Sections and Technology, Vol I, NBS Special Publication 425:266 (October 1975). 3. A. H. Kazi, "Fast-Pulse Reactor Operation with Reflector Control and a 1 06mm Diameter Glory Hole," Nuclear Science and Engineering, 60 : 62-73 (May 1976). A. R. W. Klingensmith et_. a_K , "TREE Simulation Facilities," DNA-2432H, Defense Nuclear Agency (1973). 5. J. Cortella et^. aj_. , "Operational Characteristics of the CALIBAN Fast Pulse Reactor," pp 157-170, Proceedings of US/Japan Seminar on Fast Pulse Reactors, University of Tokyo, Takai, Ibaraki, Japan (January 1976). 6. G. G. Doroshenko e_t. aj_. , "Spectra of Fast Neutrons from a Pulsed Reactor," Atomnaya Energiya, 40 : 6, pp 460-464 (June 1976). 7. T. A. Dunn, A. H. Kazi, J. Saccenti, "Fluence-to- Dose Conversion Factors for APRF Fast Pulse Reactor Neutron Spectra," USA Ballistic Research Laboratory Report 1832 (September 1975). 8. A. H. Kazi, T. A. Dunn and J. C. Saccenti, "Sensitivity of Fl uence-to-Dose Conversion to Changes in Fast Pulse Reactor Spectra," Proc. First ASTM-EURATOM Symposium on Reactor Dosimetry, Petten, Netherlands (September 1975). 9. F. N. Coppage, "The Influence of Dosimetry on Damage Equivalence Ratios," IEEE Trans. Nuc. Sci. NS-22, 6:2336 (December 1975). 10. J. L. Meason et. aj_. , "The Neutron Spectral Dis- tribution from a Godiva-Type Critical Assembly," IEEE Trans. Nuc. Sci. NS-22, 6:2330 (December 1975). 11. J. A. Grundl, D. M. Gilliam, N. D. Dudey and R. J. Popek, "Measurement of Absolute Fission Rates," Nuclear Technology 25_: 237-257 (February 1975). 15. J. A. Grundl and C. M. Eisenhauer, "Fission Spectrum Neutrons for Cross Section Validation and Neutron Flux Transfer," Proc. of Conference, Nuclear Cross Sections and Technology, Vol I, NBS Special Pub 1 i cat ion 425 : 250 (October 1975). 16. J. A. Grundl and J. W. Rogers, "Absolute Fission Chamber Measurements in CFRMF," LMFBR Reaction Rate and Dosimetry 7th Progress Report, HEDL-TME 73-31, Hanford Engineering Development Laboratory (1973). 17. D. M. Gilliam, "Integral Measurement Results in Standard Fields," paper #4. This conference. 18. R. H. Noyce, E. R. Mosburg, Jr., J. B. Garfinkel and R. S. Caswell, "Absolute Calibration of the National Bureau of Standards Photoneutron Source III Absorption in a Heavy Water Solution of Manganous Sulfate," Reactor Sci. Technol . (J. Nucl. Energy, Part A/B) _I_7: 31 3 (1963). 19. V. Spiegel, Jr. and W. M. Murphy, "Calibration of Thermal Neutron Absorption in Cylindrical and Spherical Neutron Sources," Metrologia, _7 : 34 (1971). 20. D. M. Gilliam et. a_l_. , "Fission Cross Section Ratios in the *52cf Neutron Spectrum," Proc. of Conference, Nuclear Cross Sections and Technology, Vol I, NBS Special Publication 425:270 (October 1975) . 21. T. A. Dunn, "APRF Reactor 3 MeV (Sulfur) Threshold Neutron Fluence Measurements," USA Ballistic Research Laboratories Report 1 786 (May 1975). 22. E. D. McGarry, C. R. Heimbach, A. H. Kazi and G. W. Morrison, "Fast Pulse Reactor Neutron Spectrum Measurement and Calculation," IEEE Trans. Nuc. Sci. NS-24 (1977). To be published. 12. E. D. McGarry, A. H. Kazi, G. S. Davis and D. M. Gilliam, "Absolute Neutron-Flux Measure- ments at Fast Pulse Reactors with Calibration against Cal i forni um-252, " IEEE Trans. Nucl. Sci., NS-23, 6:2002 (December 1976). 339 ISOTOPIC COMPOSITION AND NOMINAL MASSES OF FISSIONABLE DEPOSITS TYPE OF DEPOSIT MASS (MICROGRAMS) Pu-239 Np-237 Np-237 U-234 U-236 U-238 427 265 170 30 37 175 Pu-239 99.976 Np-235 - 0.004 U-234 99.887 0.002 - PERCENTAGE OF Pu-240 0.002 Np-236 - 0.005 U-235 0.064 0.005 ISOTOPES Pu-241 0.005 Np-237 99.98 99.991 U-236 0.035 99.990 0.001 Pu-242 0.005 (Pu-239) 0.02 - U-238 0.014 0.003 99.999 S=r GAS INFLOW CZ3 ALUMINUM ■■ TEFLON BACK TO BACK FISSION DEPOSITS FIG. 2. CROSS-SECTIONAL VIEW OF NBS DUAL FISSION CHAMBER CONSTRUCTION FIG. I. NBS FISSION CHAMBER WITH AND WITHOUT BORON-IO COVER I I T o HEAVY DEPOSIT l96 M g/cm 2 • LIGHT DEPOSIT 4 > .q / /cm 2 I I I I I I I I I I I I PULSE HEIGHT (CHANNELS FIG. 3. FISSION FRAGMENT VOLTAGE PULSE HEIGHT DISTRIBUTION FIG. A. GEOMETRY FOR FISSION CHAMBER CALIBRATION 340 TABLE II TOTAL APRF FLUXES MEASURED WITH PU-239 USING THE FLUX TRANSFER TECHNIQUE EXPOSURE LOCATION NEUTRONS /CM 2 -SEC-WATT 67 MM FROM CORE SURFACE 1.22 XIO 7 « 4.47. CENTER OF CORE INSIDE 38 MM GLORY HOLE 1.04 XIO 8 * 4.4% CENTER OF CORE INSIDE 106 MM GLORY HOLE 8.30 XIO 7 ' 4.9% FIG. 5. APRF REACTOR POSITIONED 6 M ABOVE BORATED CONCRETE FLOOR FOR CORE SURFACE LEAKAGE FLUX MEASUREMENT TABLE III ■ COMPARISON OF MEASURED AND CALCULATED SPECTRAL INDICES FOR THE APRF REACTOR ISOTOPE RATIO SPECTRUM AVERAGED CROSS SECTION RATIOS FOR: 67-MM FROM CORE SURFACE 38-MM GLORY HOLE 106 MM GLORY HOLE MEASURED CALCULATED MEASURED MEASURED CALCULATED MEASURED MEASURED CALCULATED MEASURED 237 NP 5.39 ±3.4% 1.017 5.72 + 3.4% 1.000 5.77 + 3.4% 0.991 238 u 239 PU 237 NP 1.69 ± 3.2% 0.929 1.78 + 3.2% 0.916 1.81 + 3.2% 0.901 1 8 I 1 ' 1 1 I I I I | 1 1 1 1 6 /••/••.A ' .^^^ * **V« ~ 4 ../•' ...•••'* jf i\ - >- HI z 111 >\ 1- z 13 tr 1 Ul a. — x 8 D -i "" 6 LU 4 - 2 i i i i i i t i 1 i i \ i 0.01 0.02 0.04 0.06 0.08 0.1 02 0.4 06 0.8 1 2 4 6 8 10 NEUTRON ENERGY (MeV) FIG. 6. MEASURED AND CALCULATED APRF NEUTRON LEAKAGE SPECTRUM 341 DOSIMETRY STANDARDS FOR NEUTRONS ABOVE 10 MeV H.H. Barschall University of Wisconsin, Madison, Wisconsin 53706 Dosimetry of neutrons in the energy range 10-50 MeV is needed for applications in radiation damage studies and in biomedical work. Dose determinations use either a fluence measure- ment or the Bragg-Gray principle. Better knowledge of activation cross sections, kerma factors, and energy per ion pair is needed to reduce uncertainties in dosimetry. (Activation; Bragg-Gray; dosimetry; energy per ion pair; fluence; kerma factor; neutrons) Introduction This report discusses dosimetry standards for neu- trons in the energy range 10-50 MeV with special em- phasis on 14-MeV neutrons. Neutron dosimetry in this energy range is of importance for applications to studies of radiation damage in various materials and to biomedical work. The effect of 14 -MeV neutrons on materials is un- der intense study for the design of fusion reactors, in particular for the choice of a material for the first wall of the reactor. l Radiation damage problems are expected to be much more severe in fusion reactors than in fission reactors because of the much larger re- action cross sections for the neutrons of higher energy. Production of hydrogen and helium by nuclear reactions is likely to produce swelling, blistering, and embritt- lement, and generally new elements produced. in trans- mutations may affect the mechanical properties of the materials adversely. In the biomedical area the interest in dosimetry used to be restricted to applications in radiation pro- tection. More recently the interest has shifted to radiotherapy 2 where much higher accuracy is needed. If the dose administered to a tumor is 5% too low, the cure rate is greatly reduced, while a 51 too large dose may result in severe side effects. Definitions Both material scientists and radiobiologists wish to correlate observed effects and radiation dose. Dose is defined 3 as the energy imparted by ionizing radia- tion to the matter in a volume element, divided by the mass in that volume element. Dose differs in general from a similar quantity, kerma(K), which is defined 3 as the sum of the initial kinetic energies of all the charged particles liberated by indirectly ionizing par- ticles in a volume element divided by the mass in that volume element. For neutrons kerma and dose differ primarily near the surface of the irradiated material within a distance of the order of the range of the secondary charged particles in the material. Both kerma and dose are measured in grays (1 Gy = 1 J/kg) , al- though the older unit, the rad, is still more widely used (100 rad = 1 Gy) . Another quantity related to the dose is the neutron fluence $. The neutron fluence is defined as the track length of all the neutrons in a volume element divided by the volume, and it determines the number of nuclear reactions in that volume. Kerma and fluence are related by the kerma factor, which is defined as the kerma per unit fluence. The kerma fac- tor depends only on the nuclear constants of the mater- ial. For a single element of atomic weight A K/ A v i i l where N is Avogadro's number and a. is the partial cross section for producing charged particles of aver- age kinetic energy E.. Neither dose, nor kerma, nor fluence are, however, good measures of either the radiation damage or of the biological effects induced by neutrons. As mentioned before, the radiation damage is strongly influenced by the gas production and transmutation cross sections which vary rapidly with neutron energy. When a mater- ial scientist says that a sample has been exposed to a certain neutron dose, he usually means fluence, not dose. If one looks at the Dosimetry File 1 * published by the National Neutron Cross Section Center, one finds only neutron cross sections which might help to deter- mine fluences, but there is no information in the Dosimetry File on how to get dose; in fact, the word dose appears only in the title. On the other hand, engineers who design fusion reactors need to know the dose in calculations of both nuclear heating and shielding. For biomedical studies the dose is the quantity which is always quoted. In order to take into account the biological effect of the dose the radiobiologist multiplies the dose by the Relative Biological Effec- tiveness (RBE) . Measurement of Fluence Neutron fluence can be measured most easily for a parallel beam of monoenergetic neutrons, for example, at a distance of, say, 50 cm from a small source of DT neutrons. In radiation damage studies the needed flu- ence is so high that small samples have to be placed as close to the neutron source as possible. In this geometry the neutrons are neither monoenergetic nor monodirectional , since the neutron energy varies with direction of emission, and the sample has a diameter equal to, or smaller than, the size of the source. In such experiments thin foils of an activation detector serve to determine neutron fluence and are placed around the sample that is under study. The principal requirements for the activation de- tector are that it must have a half life longer than the time of bombardment, and that the resulting activ- ity can be counted absolutely. In addition, the reac- tion that produces the activity should have a thresh- old not too far below the neutron energy being studied, and it should have a cross section that varies slowly with neutron energy in the energy range of interest. Three reactions which are frequently used are listed 5 in Table I. The (n,2n) reactions have thresholds around 10 MeV, the (n,p) reaction is exoergic, but has an effective threshold near 2 MeV. The Nb(n,2n) re- action is preferred for 14-MeV neutrons. The absolute counting of the resulting activities does not introduce as much uncertainty as the lack of knowledge of the reaction cross sections. The only neutron cross section that is accurately known for fast neutrons is the n-p scattering cross section, and most absolute reaction cross section measurements are comparisons with the n-p scattering cross section. The best known reaction cross section is probably the fission cross section of 235 U. Fig. 1, taken from a report on a specialists' meeting, 6 shows 342 TABLE I Activation Reactions used in Dosimetry Reaction Q(MeV) T 1/2 (davs) yOfeV) 58 Ni(n,p) 58 Co 0.39 70.8 0.81 59 Co(n,2n) 58 Co -10.5 70.8 0.81 93 Nb(n,2n) 92 Nb m -8.95 10.2 0.93 Fig. 1. Fission cross of 235 U between 8 MeV and 22 MeV. This figure is taken from ref. 6. that measurements of this cross section at different laboratories differ by more than 101 although the ex- perimenters quote much smaller errors. Activation cross sections are not as well known as the 235 U fis- sion cross section; measurements at different labora- tories often differ by an order of magnitude. The ac- tivation cross section that is reported to be best known for neutrons around 14 MeV is that for the re- action 27 Al(n,a) 21 *Na. Fig. 2 shows some values of this cross section as well as the ENDF/B-IV evaluation given in the dosimetry file. 1 * The measurement quoted with the smallest uncertainty (less than 1%) is the absolute determination carried out in 1970 by Vonach et al . 7 by the associated particle method. This measurement is, however, ignored in the 1975 dosimetry file, and the ENDF/B-IV evaluation differs by 5% from this presumably most accurate measurement. In most practical situations in which a foil serves to measure fluence in radiation damage studies, the variation of the activation cross section with neutron energy may introduce more uncertainty in the fluence determination than the error in the cross section at one energy. Fig. 3 shows the activation cross section as a function of neutron energy for the three reactions mentioned earlier. Experimental data on the Nb(n,2n) cross section are shown in Fig. 4. The two most recent extensive measurements 8 ' 9 around 14 MeV differ by 101. There are no measurements available above 20 MeV. Even for D-T neutrons the energy spread is typi- cally 1 MeV, and the energy distribution within this spread is usually poorly known and varies over the sample. Hence there is substantial uncertainty in the average neutron energy. If deuterons on Li or Be pro- duce the neutrons, the energy spread is of the order of the bombarding energy. Hence activation of foils gives only a very rough measure of fluence. By using several reactions with different thresholds the contribution of 130 120 IOO — 13.5 14.0 14.5 15.0 E n (MeV) Fig. 2. Cross section of the reaction 27 Al(n,a) 21 *Mg for neutrons between 13.5 and 15.0 MeV. Only experimental results which are quoted with small uncertainties are shown. The solid curve is the ENDF/B-IV evaluation given in reference 4. Additional measurements may be found in this reference. 750 - 500 b 250 1 1 /""V 9 Co (n,2n) 58 Co 58 Ni in,p) 5B r „ / \ \- 93 Nb(n, \ \ \ In) 92n ''Nb / 1 \ \ '--._ "-- ._ 20 E n (MeVl 30 40 Fig. 3. Activation cross sections of three frequently used detectors as a function of neutron en- ergy from ref. 5. The solid curves follow evaluated data sets, the dashed portion is an extrapolation based on theoretical con- siderations. neutrons in different energy ranges can be estimated. Dose and Kerma Determinations If the fluence in a sample has been determined, the kerma can be calculated if the kerma factor is known. Again the only nuclide for which the kerma factor is accurately known is 'H, since the only 343 500 400 300 93... , >92„,, m Nb(n,2n) Nb O LIVERMORE (1972) X GEEL (1970) • ZAGREB (1962) 13 14 16 E n (MeV) Fig. 4. Measured cross sections of the reaction 93 Nb(n,2n) 92 Nb m between 13 and 16 MeV. The most extensive recent measurements are given in ref. 8 and 9. charged particles that are produced are recoil protons and their angular distribution is close to isotropic in the CM system. The kerma factors for 2 H and ''He are also well known, but they are of smaller practical importance. For other nuclides the nuclear data needed for calculating kerma factors are not well known for neutrons above 10 MeV energy. Above 20 MeV there are hardly any measurements of the needed nu- clear data, and kerma factors are calculated from nu- clear models. Abdou has developed a computer code 10 for calculating kerma factors from the ENDF/B data file. Because of their importance to the biomedical ap- plications the kerma factors of the principal con- stituents of tissue, H, C, N, and 0, have been calcu- lated most carefully. The most widely used values are the kerma factors published by Bach and Caswell 11 in 1968. These have been revised recently by Caswell, Coyne and Randolph 12 on the basis of more recent nu- clear data. Table II gives a comparison of the kerma TABLE II Kerma Factors for 15. 5 -MeV Neutrons -9 2 (in 10 rad cm ) Bach, Caswell 47.5 (1968) Caswell, Coyne, 47.1 Randolph (1976) Howe rt on 47.1 (1976) 2.36 1.39 2.13 3.02 2.68 2.07 2.17 2.90 1.48 factors of H, C, N, and at 15.5 MeV according to the two evaluations by the NBS group, and also according to a recent evaluation by Howerton 13 at Lawrence Liv- ermore Laboratory. The differences between these three evaluations indicate the uncertainties in the kerma factors for the most carefully studied nuclides. Bragg- Gray Approach to Dosimetry In biological work dose in tissue is usually measured with a tissue -equivalent (TE) ionization chamber. According to the Bragg- Gray theory of cavity ionization, the ionization produced in a small gas- filled cavity in a solid material is related to the dose D at the position of the cavity by D = Q WS, wall TO" where Q is the collected charge, W the gas energy required to produce an ion pair in the gas, M the mass of the gas in the chamber, and S the mass stopping power. TE chambers were originally developed for measuring y dose. The walls of the chamber con- sist of TE plastic, usually Shonka A-150 plastic, 1! * in which most of the oxygen in tissue is replaced by carbon. Such plastic is, however, not exactly tissue equivalent for neutrons, since C and have very dif- ferent kerma factors for neutrons and the ratio of the two kerma factors varies with neutron energy by an or- der of magnitude. For 15-MeV neutrons the difference between tissue and TE plastic introduces a 5% correc- tion. TE chambers are usually calibrated with y-rays. Such a calibration is not generally applicable for neutrons because W and S are likely to be different for the electrons produced by y-rays, and for the pro- tons, a particles, and 12 C and 16 recoils, produced by neutrons. In the past data on the dependence of W on the mass and energy of the ionizing particle have been contradictory. During the last year two groups 1 5 » ie have obtained new results which clearly indicate that W increases with increasing mass. There is also good evidence that for heavy ions W increases at low energies. Some of the recent re- sults obtained for H + , He + , and C in TE gas are shown in Fig. 5. Except for low energy C ions these w (eV) 45 \ 1 TE 1 GAS 1 1 1 o & • * a FONTENAY -AUX ■ BROOKHAVEN • ROSES " rA t " - 1 I A c + _ 40 T * i } I * — 35 "li } i i 1 A H, + - _ L- h J + H - -\ t \ •• • » • • " 30 f> H 1 1 1 ■ 1 i 1 Fig. 5. 600 800 ENERGY (keV) Results of two recent determinations of the average energy loss per ion pair as a func- tion of particle energy in methane-based tissue-equivalent gas for H" 1 ", He + , and C ions. The data are from references 15 and 16. 344 data are consistent to within 3%. A calculation of an average value of W for the different type particles produced by energetic neutrons will, however, have an appreciably larger error. Although there is also substantial uncertainty in the knowledge of the ratio of the mass stopping powers in solids and gases, the uncertainty has probably rel- atively little effect on the neutron calibration, be- cause some of the errors will tend to cancel in taking the ratio of two ratios. Comparison of Dose Measurement Based on Fluence and on Bragg- Gray Approach Neutron dose in tissue can be measured by two in- dependent methods, i.e., by measuring fluence or by measuring charge in a TE chamber. It is interesting to compare these two procedures for monoenergetic neu- trons for which fluence can be measured fairly accu- rately. In an experiment " 7 performed at the Rotating Target Neutron Source at Lawrence Livermore Laboratory a 1 -cm 3 -volume commercially obtained TE chamber was exposed to 15-MeV neutrons. Neutron fluence was measured both by foil activation and by proton recoil counters. Fluences were converted to dose using the kerma factors of ref. 12. Charge collected in the chamber was converted to dose on the basis of the y-ray calibration with corrections for the difference in W for y rays and neutrons. In separate experiments both air and TE gas served as chamber fillings. For air there is a small correction for the difference in stopping powers for electrons and protons. Table III TABLE III T-E Ion Chamber Conversion Factor (rad/nC) (20°C, 760 Torr) Chamber Gas Radiation Method Material Air T-E Gas 60^ Co y s 3.19 2.78 Bragg- Gray A-150 3.45 2.90 15-MeV Approach Muscle 3.22 2.71 Neutrons Neutron fluence A-150 3.31 2.81 based calibration Muscle 3.09 2.62 lists the results of the comparison both for dose in muscle and A-150 plastic. The two procedures yield calibrations which differ by only 4%, which is much closer than would be expected on the basis of the un- certainties in the atomic and nuclear constants in- volved in the calculations. Continuous Spectrum Neutron Sources Since it is easier to obtain high neutron inten- sities by producing broad spectra than from the DT re- action, currently most neutron radiotherapy sources use Be targets bombarded by either protons or deu- terons. For the materials applications there is a plan to build a large facility in which the reactions of deuterons with a thick lithium target serve as a high intensity neutron source. These sources produce a continuous neutron spectrum extending from very low neutron energies to the bombarding energy. Neutron dosimetry for such sources is very much more difficult than for D-T sources, and there is an almost complete lack of the required nuclear data. There are hardly any of the relevant cross sections known for neutron energies above 15 MeV so that neither the fluence nor the kerma can be determined if the bombarding energy is above 15 MeV. In the biomedical application a tissue equivalent dosimeter yields fairly reliable information, but for the materials application there are large uncertain- ties. Since the sample is placed near the target, there is a wide variation of both neutron spectrum and intensity even over a small sample. The Li target has to be so thick compared to the distance to the sample that there is a significant difference in distance be- tween source and sample for the more energetic neu- trons, which are produced farther from the sample, from that for the lower energy neutrons, which are produced close to the sample. Hence the neutron spec- trum at the sample is not the same as that observed at a distance from the sample by, say, time-of- flight measurements. In addition, the neutron spectrum varies rapidly with angle of emission with respect to the incident deuterons ; the average neutron energy de- creases rapidly with increasing angle. Before measure- ments with a D-Li neutron source can be interpreted meaningfully, more measurements of the neutron inten- sity and spectra are needed as a function of deuteron energy and emission angle. Furthermore, the charged- particle production cross sections in the materials to be studied and the activation cross section of the dosimeter materials need to be measured as a function of neutron energy up to the highest neutron energy that the source produces. The uncertainty in the dose determination for radiotherapy with continuous-spectrum neutron sources is not as serious a problem because one does not know the RBE accurately either. Radiobiological experi- ments measure the product of dose and RBE. Uncer- tainties in dose produce corresponding uncertainties in RBE. While a better knowledge of the two factors separately would be of considerable interest, it is not essential for the therapy applications. Experi- ments are underway, however, which should improve the knowledge of kerma factors for tissue for neutron en- ergies above 14 MeV. The results of such experiments will help in the understanding of the dependence of RBE as a function of average neutron energy. References 1. Proceedings of the International Conference on Radiation Test Facilities for the CTR Surface and Materials Program. Report ANL/CTR-75-4 (1975). 2. H.H. Barschall, Am. Scientist M_, 668 (1976). 3. International Commission on Radiation Units and Measurements, Report 19 (1971). 4. ENDF/B-IV Dosimetry File, Report BNL-NCS-50446 (1975). 5. M.J. Saltmarsh, et al . , Report ORNL/TM-5696 (1976) 6. Proceedings of the NEANDC/NEACRP Specialists Meeting. Report ANL-76-90 (1976). 7. H. Vonach, et al., Z. Physik 237, 155 (1970). 8. A. Paulsen and R. Widera, Z. Phys. 238 , 23 (1970). 9. D.R. Nethaway, Nucl. Phys. A190 , 635 (1972). 345 10. M.A. Abdou and C.W. Maynard, Nucl. Sci. Eng. 56, 360 (1975). 11. R.L. Bach and R.S. Caswell, Rad. Res. 35, 1 (1968). 12. R.S. Caswell, J.J. Coyne, and M.L. Randolph, Proc. Workshop on Physical Data for Neutron Dosimetry, Rijswijk, The Netherlands, EUR 5629e, Luxembourg, 1976, p. 69. 13. R.J. Howerton, private communication. E.F. Plechaty, D.E. Cullen, R.J. Howerton, and J.R. Kimlinger, Lawrence Livermore Laboratory Report UCRL-50400, Vol. 16 (1976). 14. F.R. Shonka, J.E. Rose, and G. Failla, Proc. Second U.N. Conf. on Peaceful Uses of Atomic Energy, 21, 184 (1958). 15. N. Rohrig and R.D. Colvett, Rad. Res. 67, 613 (1976). 16. M. Chemtob, B. Lavigne, J. Chary, V.D. Nguyen, N. Parmentier, J. P. Noel, and C. Fiche, Phys. Med. Biol, (to be published) 17. H.H. Barschall and E. Goldberg, Medical Physics 4 (to be published). 346 SUMMARY SESSION: W. W. Havens, Jr., Chairman PANEL MEMBERS: R. Caswell, V. Farinelli, H. Liskien, H. Motz, R. Peelle, and L. Stewart Introductory Remarks : Dr. Bowman We now come to the final session - the summary of the Symposium. Bill Havens will be chairing this session and he has been charged with maintaining a tight schedule. Since we plan to incorporate the record of this session into the proceedings, I am required by federal government regulations to remind you that the discussion of this conference is being recorded. Now, I'll turn this session over to Bill Havens. Prof. Havens : In the interest of time, we are not going to follow the format listed on the program. We are going to incorporate the summaries of the discussion groups into the appropriate portion of the summary of the program. And the format will be that the leader of the discussion group for the particular workshop will be allowed five minutes at maximum to present the sum- mary of the workshop. Then Alan Smith will summarize that portion of the conference. Then we will go to the panel for discussion of that particular portion of the conference. Alan Smith and I put together a list of times which would be devoted to the particular subject matter of the conference, and we'll try to adhere rather rigidly to the allotted times. I don't know what I will do if somebody doesn't stop when I get up to cut them off but, hopefully, forcible means will not be necessary. So in the interest of time we'll start immediately. The first session was on light nuclei and the nearest workshop to that was the one Bryan Patrick chaired, the Workshop on Li and B. So, I'll call on Bryan Patrick to give a five-minute summary of the workshop. Dr. Patrick : I wonder if I'm allowed to take, as a foreigner, the Fifth Amendment to get around this re- cording business. No? We had a fairly lively discus- sion on b Li and 1( ^B in which a large number of people participated, and I hope this isn't too disjointed a representation of what went on. The use of a large body of nuclear physics infor- mation in the form of angular-distribution, polarization and inverse-reaction data in addition to the direct measurements (a , a , and a ) in the R-matrix n,t n,ct n,n calculation of Hale on the 7 Li system, although not a new idea, is a welcome development which should be encouraged. Most of the recent experimental work on b Li seems to have concentrated in the region of the 250-keV resonance. It's encouraging to note that the latest (n,a) measurements agree to the order of 3 percent with each other and that they also agree with the R-matrix calculation of Hale. However, we should not put too much emphasis on this. There have been many measurements over that resonance, and there has been much disagreement. The reasons for these problems are still not fully understood, and we should hesitate to conclude that the cross section is known to even 3 percent in that region. Before proceeding with further measurements using 6 Li glasses, we really need to know why the disagreements exist. Perhaps some of them are due to the problems of 6 Li concentrations and distri- bution in scintillation glasses that were discussed during the symposium. This is a situation that for a standard like 6 Li, which is used so much, is very disturbing, and I think really unacceptable if we're going to use this cross section as a primary standard. On the credit side, the agreement between the R-matrix calculation and the inverse reaction data of Brown et al. at 6^0° and 180°, which were completed after the calculation was done and which were therefore not included as input to the calculation, is to be noted. We spent some time considering the question, "Is 6 Li a suitable primary standard?" There are people who put forward the view that, with all the problems we have had measuring this cross section, it is not a good primary standard. It was concluded that for essentially, practical reasons, although for other reasons too, it is a good primary standard up to ^150 keV. Above that energy it is questionable. Now the problems with e Li and 1 °B are bound up with detector technology. There have been few real advances in recent years, and this is an area which clearly needs improvement if higher accuracies are to be achieved in the cross sections. It was felt that we're probably fairly close to the limit of accuracy using present s Li glass techniques for measuring the cross section and maybe it is time that lu B was investigated. Measurements have concentrated on 6 Li recently and 10 B has not had much attention, and perhaps it's time to look at I0 B again, particularly to cover the region above ^150 keV where 6 Li becomes a poor standard, to V500 keV where hydrogen takes over. This region between is probably the most important problem at present. There was also some suggestion that d He detectors should be considered again, and maybe a fresh look would prove to be fruitful. If you want some measurements, which are thought to be the most desirable, we've made a shopping list, mainly from Gerry Hale. The 5 Li(n,a) and total cross sections at thermal energy are still important. We need to determine these more accurately if we are to know the 1/v part of the cross section better. In this connection, the levels in 7 Li responsible for the low- energy 6 Ll(n,a) cross section have not yet been clearly identified. This is not just an academic problem as their position determines the sign of the deviation from 1/v of that component. The minimum at about 80 or 90 keV in the region below the resonance is another important region where accurate (n,a) and a measure- ments would help to define the cross section'over a wider energy interval and again the cross sections at the peak of the resonance are still relatively poorly known and must be more accurately defined. Elastic scattering over the resonance and angular distributions at low energies are also important . °Li above 500 keV to 1 MeV is used as a standard by certain groups and that's a region where the R-matrix calculation may be somewhat in error. And finally, it's felt that the Li(n,a) to °B(n,a) cross section ratio measurements above 50 keV, extending up to at least several hundred keV would also be extremely useful. Thank you. Prof. Havens: We'll have Alan's summary. Dr. Alan Smith : I think before I start on the light nuclei which will include B, H, C and Li, I should out- line my position here. I am somewhat awed by standing before you. I try to avoid making a precise measure- ment if at all possible, and there are those who say I never do. I think that maybe the best you can expect from me is an unbiased opinion, and one good sound reason is that the ultimate in unbiased judgment is complete ignorance in some cases. But I also think 347 I bring to it a look which maybe isn't as close. And I bring also some age, which is not anything but my biolog- ical credit; but I have gone to more of these standard conferences than have been mentioned in this proceedings . Somebody forgot the Oxford conference of 1963, he didn't count back that far. Well, that starts me off because I was reminded this morning by Bob Block that, at the '63 conference, Bob Batchelor had a little audit done by the Queen's exchequer, and his conclusion at that time was that one million pounds had been spent on the 6 Li(n,a) cross section to that date. We're still at it, and the pound has been devaluated, but even so I think it must be up to about 8 or 10 million dollars at least, and the question which I have now that really bothers me is, "Where are we, what do we want, and how much more effort is warranted?" I looked at the new results, I counted crudely, and I see only a few new total cross section measure- ments that are really relevant to standards; maybe two. I see two new (n,a) measurements, and I believe they were both relative. I see two scattering measurements, several angular distributions of the a particle, and one study of inverse reactions. That's apparently in the last 10 years. I asked myself what the impact has been, and I think that the major effect is that we have perhaps a better knowledge of the (n,a) cross section below 150 keV. Beyond that I don't know. I concur with the Workshop's report that this is where we should focus our attention. But looking at it from the pragmatic point of view, I asked myself why we're making all these measurements at higher energies. And I asked myself another question. Looking back in history, have we any evidence that perhaps we know it a good deal better than we think we do? I look back, for example, at an early measurement made 10 years ago by two English gentlemen who did a very good job using a simple-minded theory, and they came up with an (n,a) cross section. Bob Peelle plotted it on a slide here the other day. As near as I can figure out, it's within 1-1/4 percent to 1-1/2 percent of the current ENDF/B-V. So I wonder what we have gotten from these measurements. Have we got more confidence and to what degree? I have an uncertain feeling about how we improve that situation further. Now, I am a little concerned at some of the state- ments that we see for what our needs are. You can look at some of the compilations of requests and you can see that you want the (n,a) cross section of 10 B to 1/2 per- cent from to 3 MeV, and I think that's not a very rational request; I think that's what the workshop said. I think you may need it to that accuracy to 150 keV, but I think we should focus on what the range of interest is. Some of you know that some years ago I wrote a memorandum extolling the virtues of physical inter- pretations and theoretical extrapolations for nuclear data purposes and some resistance to pragmatic adjust- ments. You also may know I took my knocks for that memo and I'm still black and blue in some places. So therefore I really welcome the outstanding work at Los Alamos to use established theory in a long-term way of knitting together this process. But I am a 1 ittle concerned that apparently, if I understood correctly, after 10 years we are still looking for the positive parity state in the / Li system. That, I think, was Ernie Rae's search, too, in 1970. I also wonder about how far we extrapolate our fit. It is, I think, a parameterization and a theoretically sound one, if that's all it is. I was interested particularly in Jack Harvey's statement, and his very nice experiments, and those here of Ivan SchrSder at the Bureau, where we now have a different reaction mechanism proposed to account for the (n,a) angular distributions. I don't know, I'm not a theorist, but it bothers me that maybe we shouldn't get too hung up on one type of theory until we're sure we really understand it. And I think then if you take that approach maybe we shouldn't worry too much about the exact energy of that resonance. And I am certainly a guilty party for probing around in that thing, too. I wonder if I should continue. I would be happy to do so, in fact I'd love to tickle some toes, but I would like some guidance. (Granted by Havens) I too share the workshop's opinion that 10 B is probably a more promising material. One of the reasons for this is that in my happy life I have not dealt either with B or Li, so after hearing Larry Weston's excellent talk on the difficulties with the Li detection system, 10 B I decided I never wanted to get involved. Maybe is easier. Somebody else mentioned something about the total cross section of 10 B. I think Bob Block needs that for correction procedures, and I think he's correct certainly. But that again reminds me of a statement made by Bob Batchelor, and I think I do have a positive suggestion there. Bob Batchelor in 1963 recommended the old shell transmission experiment as a way of getting the (n,a) cross section. And in fact some of the results were shown here. They didn't come out very good, but I think that was a mistake of the times in two ways: the correction procedures were poorly done and we didn't know the elastic- scattering angular distribution. Maybe it's worthwhile doing that one again, now using the calculated distri- butions which we have from our R-matrix theory. Maybe Batchelor 's recommendation of 15 years ago is still a good one. The question I guess is "Do we have those Batchelor spheres in the AWRE vaults?" There are two other light element cross sections, the hydrogen and carbon, that I would like to say a word about. Lovely experiments on (n-p) scattering from Harwell and their visitors; I was much impressed. I'm not a specialist in that area certainly, and it's been I guess five years since I even looked at that type of reaction at even lower energy. But I looked at the curves and so forth and I guess my question is, "Excellent and beautiful though the results are, do they really change our assumption of the Breit-Hopkins expression of the standard?" I didn't see a quantita- tive statement. Is there real doubt in that basic and most elemental standard? I'd like to see an answer to that question. I also think that we better give a little harder look at this n-p cross section at lower energies, and this is partly what the workshop said, but I'd go lower. And this impacts upon the 6 Li. I am impressed by Mr. Czirr's detector which is said to be flat to, I believe, 1 keV to a percent or so, and that I think takes much of the crunch out of the need for standards such as (n,a) of Li and B from say 150 to 500 or 600 keV. That was a very impressive detector. My final light element comment has to do with carbon. I don't have any real serious objections, in fact none whatsoever on the paper presented. I partic- ularly though wish to quote again the remark of one man from the audience who, yesterday I believe it was, said that, "If you are measuring angular distribu- tions you better damn well measure carbon; if you can't get that you're in trouble". I strongly support that comment. I might also suggest that in my view the total and differential scattering cross section of carbon, probably from 1.8 MeV on down, is known to about 2 percent and that is very nearly as good as the hydrogen. That's also an energy standard which, we heard today, is well established. I will extend my remarks also from scattering from carbon to total 348 cross sections. And some of you may know that total cross sections are a mess in many areas. I would also suggest that those who measure total cross sections, even though self-normalizing, use carbon as a reference check point. Prof. Havens : I would like to call on some of the panel members to comment on the summary and the work- shop report. Lee Stewart. Lee Stewart : Let me make one remark on hydrogen. Remember that Ugo Farinelli suggested the one thing he didn't like about ENDF standards was that we change them too often. Hydrogen has not changed since ENDF/B- III and yet I come to this symposium and I find very few people who use it; the answer is that hydrogen is too hard to implement. Well my comment, having been in this business for many, many years (26 to be exact), is we've done the easy experiments, now let's start and do the hard ones, because we have been measuring cross sections to standards other than hydrogen for many years. Hydrogen is the only standard in the MeV range that is both smooth and precisely known compared to all other standards; there is no structure. So let's do use it. The only other comment I'll make refers to Alan's comment that we do not need further measurements in 6 Li. One problem has hurt us for a long time; there are two measurements on 6 Li elastic scattering at low energies. They differ by 30 to 40 percent; one comes from Harwell and one comes from Argonne, and we have never been able to resolve that 30 to 40 percent dis- crepancy. Therefore we cannot subtract the elastic from the total to get the (n,a). Thank you. Prof. Havens : Unfortunately that's all the time we have allowed for this particular subject. The others will have to go more rapidly. Yes, it's a railroad iob! The next session was Capture Standards, Fission Para- meters, and Thermal Standards. And since there was no particular workshop which was associated with this, I'll call on Alan Smith to give the summary at this time. Dr. Smith : Well I have ten minutes to cover this collection, and I will move very fast and try to catch up a bit. There was really only one capture standard discussed in detail and that was gold. It has its strong shortcomings in structure at low energies, and the results are poor at higher energies, but is really fairly well known, 4 percent or so from 200 keV to 3 MeV, and it was said there was no discrepancy between the direct detection and activation measurements. That, I think, is a change from the past. I guess my real problem with this general capture area is, "What happened to the other capture standards, Ho and Ta?" They have been proposed. Another question is, "Do we know some of the applied capture processes even better than our standard, and if so, should we neglect the standard? For example, "Is 238 U capture, in fact, better known than that of gold?" If I look in the same energy region between the ENDF/B-IV and ENDF/B-V, the changes are less than 4 percent. I think then I should pass on to some of the other things that go under the general heading of fission physics, the fission neutron spectra. Here we had some new interesting results. One has been a question that's been in my mind for a long time. What is the energy distribution at very low energies? The Russian results seem to show it's a close fit to a Maxwellian. This gives me a bit of trouble in the context of some of the integral experiments which seem to show quite the opposite, an abundance of neutrons in the low-energy tail. I think that one must stress very importantly the acceptance of the 252 Cf fission spectrum as the standard fission spectrum as it already is accepted as a standard for v. I think it is regrettable that in some past meetings the emphasis was put on 235 U and 239 Pu for pragmatic purposes. The basic standard should have been emphasized instead. Once you have that, the ratio obviously suffices for other isotopes. Thermal standards I will not discuss to any great extent, the main authority on the subject unfortunately had to leave. I would only point out that there's one associated problem that bears on this and a number of other aspects of fission, and that's the foil assay in fabrication. I think that I have seen an order of magnitude difference in accuracy uncertainties in our standard foils. The Plutonium Subcommittee is working in the 0.1 percent range, while the rest of us are still at 1 percent. I think there's a transfer of technology problem here that_will impact upon our pre- cise thermal constants, our v, and a number of other areas. I'm embarrassed to say that the speaker from the Plutonium Subcommittee comes from my own laboratory, and my foils aren't anywhere near as good as his. Very carefully discussed was v for Cf. I think the thing that disturbs me a bit here is, "Are we at the end of the line with present techniques?" There was one new measurement as I understood it in the last 10 years. The rest are reassessments and recorrections . I understand another one which will combine n and v is in progress or planned. That might be welcomed, but I question whether we should continue those type of measurements with those existing techniques if we can't get a real breakthrough. And one thing that comes to mind, in which somebody with much more knowledge of the field should look at. You have an awfully intense and sensitive thermal column in the Lucas Heights reactor, it's an Argonaut reactor. What can you do with a few precise fissile foil assays a la the Plutonium Subcommittee and a good pile oscillator in that sensitive thermal column? Is that a new method? I know, for example, that very high accuracies are claimed for weighting of flux distributions in fast criticals. Perhaps this is some new technique that can be explored, maybe it's been looked at. Prof. Havens: I understand that Dr. Boldeman had a rump session on v, and I'd like to call on him. Could you take the microphone since this session is being recorded and give a very brief statement of the results of the discussions on v? Dr. Boldeman : Richard Smith with the eta bath is measuring v and, having seen the things that he's planning to do and the information that he's got already, I'd be very surprised in fact if he doesn't turn up with a particularly nice result, whatever that happens to be. Bo Leonard suggested to me at one stage that when we combine the manganese sulphate bath measurements and the liquid scintillator measurements we should recognize that there are common errors in each of these. Measure- ments of a given type therefore should be first averaged together and then the systematic error applied to the average. The results from the different sets of measurements should then be averaged with weights according to their errors to get the final result. Well, I thought I would try that, so I did a bit of a quick calculation this morning. Using the conventional approach the result that I got was 3.745, + 0.010. I expanded the error to this value because I thought there are lots of things like the uncertainties in the fission-neutron spectrum which really occur in every measurement. You might remember Ted Axton showed a slide for a manganese sulphate bath where, for average energies of 1.39 or 1.43 MeV; there's a 0.1 percent 349 change. So even in the manganese sulphate baths the spectrum matters. We expanded this error then to that. So then I thought we would assume that manganese sulphate baths have a common error and liquid scintillators have a common error. I therefore averaged all liquid scintil- lator measurements together and separately averaged all manganese sulphate baths together. These two values were then averaged with weights given by their uncertainties to the boron pile and the Fieldhouse measurements. The result now became 3.748 +0.01. So it doesn't strike me that anything you do with the measurements or the way which you write them will make very much difference. If I were to guess, as Lee Stewart sometimes thinks it's reasonable to do, I reckon the final answer is about 3.75. Despite the very acceptable level of agreement between the different absolute measurement of v for 252 Cf, the question of a possible systematic difference between the MnSoi, bath determinations and those obtained using liquid scintillators, and therefore the real accuracy of the mean of the measurements, continues to concern some evaluators. It should be reemphasized that of the eight absolute measurements only one lies more than one standard deviation away from the mean. Furthermore, the averages of the liquid scintillator and MnSo^ bath values differ by only 1-1/2 standard deviations of the comparison. Consequently, their fears are unfounded and there is no evidence to postulate an experimental technique-dependent error. I'd like to take up that thought of oscillating things in and out of reactors; maybe somebody can do something about that. Prof. Havens : I'd like to then call on Dr. Liskien to comment on the summary and Dr. Boldeman's results. Dr. Liskien : I think I have nothing to comment on this, but I would like to remind the Symposium that we had in this session and this group of subjects two potential newcomers as standards, namely 238 U and Np. If a person asks me why we do need these cross sections, the answer is of course yes we want to conveniently measure fluxes. Then he will certainly answer yes, OK, so I understand you need one cross section for the whole energy range, but in fact I have to explain that we need perhaps more. Now comes the point that is suggested that we add again at least one of the two. Of course we all know the reason is that we don't want to have a thermal response in the standard. But my real point is that when I remember correctly that the interest for such a standard is coming from the reactor in-pile dosimetry, and I wonder if this is also true now for 235 U, if what is really wanted is not the total fission cross section for these isotopes, but, in fact, the partial production cross section for the production of a convenient fission product because many of these measurements are done not on-line by counting fission fragments but by off-line activity measurements. I would like to see in the future this point clarified; really, what is needed there? Prof. Havens : Any other member of the panel like to comment on what is needed? Dr. Peelle. Dr. Peelle: I'd like to comment in terms of another question which has concerned me. In my reading of many papers on the subject of the neutron spectrum from fission, I've never seen a really good analysis of what properties of the spectrum are actually required. Is the average energy enough? What parameters, what moments have sensitivity in the applications? And if we knew that, we might know better how to deal with, how to parameterize, and how to seek better accuracy in fission-spectrum measurements. Prof. Havens : Would you like to comment on that? Dr. Farinelli : I might make a comment on this last question. Well it depends very much on the problem. For instance, the high-energy tail of the fission spectrum is very important when you use threshold detectors with a high threshold. And why do you use these detectors? For instance, because you are interested in high-energy neutrons, if you have to simulate CTR conditions in a fast reactor. So this may explain why the status of the fission spectrum which was acceptable at high energies up to a certain time ago, perhaps a couple of years ago, is no longer satisfactory, and it's certainly one application for which you would need better resolution in the high- energy tail. Prof. Havens : We have about one more minute. Lee. Lee Stewart : Another remark is that Jud Hardy did some calculations in a thermal reactor and using the same average energy but just changing the shape, from a Maxwellian to a Watt, he got differences of 0.3 per- cent in k. So if you harden the spectrum, of course, you get a change in k. Just measuring the average energy in some configurations is certainly not enough; we need to know the shape and whether the shapes are different among the different isotopes. Prof. Havens : I'm sure the problem is different for shielding than it is for k and therefore the answer to the question is you want to know as much detail as pos- sible in the final analysis. The next topic for which there is a workshop, which is closely related, that we divided the conference up into was Flux Intercomparisons. I will call on Dr. Axton to present the results of the workshop he conducted on the future of BIPM flux intercomparisons. Mr. Axton : First let me state what the object of the inter comparison is and the criterion for judging whether or not it was a success. The object of the intercompar- ison is to identify systematic errors in absolute measurements of neutron flux density. There are two types of measurements involved in this comparison; firstly there are the absolute measurements of flux density made by different methods at different labora- tories, and secondly there -are the comparison measure- ments between those absolute measurements which are made directly as part of the comparison. The criterion for success is that the errors of comparison must be small compared with the absolute errors, otherwise you know nothing about the systematic errors in the abso- lute measurements. The intercomparison falls into ten subgroups because there were five intercomparison energies and there were two transfer methods for each energy. The conclusions of the discussion group were as follows: (1) Some transfer methods were good and other transfer methods were not so good; (2) When the transfer method is good the results are encouraging. Flux measurements by different methods agree and flux measurements at different laboratories agree. For example, associated-particle techniques and proton telescope techniques agree at 14 MeV. At 2.5 MeV, associated particle counting, the stilbene crystal spectrometer, proton telescope, the manganese bath and the proton proportional counter all give very good results. When the transfer method is bad, then the information we obtain is "that the transfer method is bad." Conclusion (3) is that we should carry on and improve the comparisons, in other words, do some more. (4)Future comparisons should satisfy both dosimetry interests and cross section measurement 350 interests. Conclusion (5): The energy range should be extended down to 144 keV to include reactors, and this energy is still feasible with Van de Graaffs. The range should also be extended up to 50 MeV for cross-section interests. Conclusion (6): There is a need to improve transfer methods. Conclusion (7): Cross section measurements should be included as transfer methods. This would allow participation from the white spectrum linacs and also those from the Van de Graaffs. There is need for a study period to iron out the problems involved in this particular aspect of the comparison. Thank you. Prof. Havens : Alan, will you summarize this part of the comment of conference? Dr. Smith : I only have a couple of remarks on this one. First of all, I'm happy to see that the concept of a transfer basic physical standard of cross section has entered the program. One of my problems with this type of measurement over many years is that the detectors tend to be environmentally associated or instrumentally associated, and therefore they're not suitable for many places. The second thing is that I am impressed by the real good agreement. I think it's quite good, and I think it was Caswell who pointed out that the similar dosimetry comparison was almost identical. I think they did quite a good job. Prof. Havens : Dr. Caswell, will you comment? Dr. Caswell : I would like to make a couple of comments, one of which might involve some interaction with the audience. The one question that the people who organized this have wondered about is that this has been essentially an intercomparison between Van de Graaffs and Cockcrof t-Waltons. The white source machines have not been involved although there were earlier some expressed desires to have them involved. I think that people who are involved should be very interested to know to what extent the white-source people would want to participate and what is the mechanism envisioned. A very precise measurement of a specific cross section, preferably one that is not well known, could play the role of an intercomparison in this way. Secondly, we have used the word "BIPM", and I'm not sure everyone in this audience knows what organization this is; but I thought I'd take a minute to explain. When you take an elementary physics course, in the old textbooks they would have a picture of the standard meter bar that sat in some institution near Paris, and also of the world standard kilogram that also sat in some institution near Paris. This institution is, in English, The International Bureau of Weights and Measures, in French the Bureau Inter- nationale des Poids et Mesures, and it was set up originally by the Treaty of the Metre in 1875, at that time as the keeper of the artifact standards like the meter bar, the kilogram, and so on. There's only one artifact standard left. For example, the meter bar is passe; it's now a wavelength of light, and the one that's left is the kilogram. Incidentally, there have been some interesting troubles recently discovered in the transfer of information on the unit of mass where the effect of the density of the air^ on the balance was wrong. So even that field hasn't settled down. Prof. Havens : Would any other panel member like to comment on the flux intercomparisons? No one. I'll proceed to the next sub-category, Personnel Dosimetry, Biomedical Needs. Prof. Barschall, would you give the results of the workshop on that subject? Prof. Barschall : For biomedical applications, the quantity which one wants to know is the absorbed dose. Let me repeat from this morning what we mean by absorbed dose: the energy absorbed per unit mass. There are two types of instruments which can be used to measure directly absorbed dose. One is a tissue-equivalent ionization chamber and one is a calorimeter. There was agreement among those working in the area of personnel protection and those working in the area of therapy, that it would be exr.'emely important to be able to have an ion chamber or calorimeter calibrated at the National Bureau of Standards and certified by the National Bureau of Standards as to the dose calibration for neutrons of these instruments. There was agreement that calibration should be available for two neutron energies; one in the energy range of the fission neutrons, and this might be done with a Cf source, but there's also a need for a calibration at higher energy. For this purpose, the most suitable source appears to be a 14 MeV d-T neutron source. I should like to emphasize once more that the interest is in the calibration in terms of dose per Coulomb rather than in terms of fluence per Coulomb, which is why perhaps the National Bureau of Standards might prefer to make the calibration. Now since actually most centers, especially those active in therapy, use continuous neutron spectra, if not fis- sion spectra, they'll have to make corrections for the energy dependence of the kerma factor in order to apply the calibration which they might get from the Bureau to their particular spectrum. For this reason, there's an important need for a better knowledge of the energy dependence of the kerma factors, particularly of carbon, nitrogen and oxygen. It is not sufficient to know just the cross sections for various kinds of processes; what is more important to know is the charged-particle spectra which are the spectra of the charged particles produced by various kinds of processes, and so this is the quantity which should be measured. And this informa- tion is not only needed to evaluate the biological effects of the neutron; it is also needed to calculate the effective value of W, the energy per ion pair, for the mixture of charged particles that is produced by a given neutron spectrum. High priorities should be given to determining the charged-particle spectra for the 14-MeV neutrons, not only because the instrument should be calibrated at this energy, but also because it would be most desirable to tie down the kerma factors at a particular energy in the neutron energy range of interest in therapy. A knowledge of the kerma factor is also important to calculate the dose delivered in therapy to various organs which have different chemical composition. Prof. Havens : Alan, would you comment on this? Dr. Smith : I only have a few brief remarks because we had agreed that Randy is far more cognizant of this field than I ever hope to be. I, however, have had occasion to observe the Bureau's performance in this area for many years and I must say I am impressed by the magnitude of this problem and the impact on society and I don't think that's what comes through to people out here. It came through to me, sitting here watching the Bureau trying to handle it and its a very difficult problem and I think the problems in personnel protection for radiation dosimetry are fundamentally ones of public implementation, technology transfer and assur- ance and these are administrative management social problems but they are enormous in scope and I think this should be recognized. I don't think that there is really a basic standard need for that one. On the bio-med dosimetry problem I had some troubles, really. I was awestruck by the accuracies that are quoted. I am just glad that I'm not in a tumor hospital with someone probing into my innards with a 5% accuracy. What bothers me is, "Has anybody measured a kerma to 351 5% in an ideal laboratory situation of a dummy torso?" Can it be done with the current technology much less in a clinical application? I have trouble again with this debate that apparently is going on, what can we do with standards and how are we going to clinically utilize them? I think there is a great deal of ambiguity there and I have one final other remark. I think that from the point of view of my interests we should be rather specific in what we want. I saw these differences in calculated kermas which if I understood correctly were attributed to differences in carbon cross sections used at Livermore and the Bureau of Standards. What differences in the carbon cross section? What specifically is causing the trouble? I would like to see quite definitive lists if I could, of say the ten most wanted things to characterize the 14-MeV biological dosimetry problem. I agree the cross sections were bad. They've got to be determined and stated in the sense the physicists understand, not values like biological effective something or other but I would like to see a list. What is the specific set of ten cross sections? Then maybe I could do something about it. Prof. Havens : Right now, Randy, would you care to comment on these? Dr. Caswell: Well, first I'll send Alan a list in a few days and then I will go back to some observations I wanted to make earlier. First, I have had the good fortune I think to be at four out of five neutron standards meetings mentioned. I think this is the first time that any physicist or other representatives of the biomedical community have really been in attendance so I think we should recognize we are just beginning a dialogue between the neutron standards people and the biomedical people who are concerned with neutrons. We need to learn to speak each other's language and I would like to congratulate Prof. Barschall on the choice of elements that he presented which are really the essential guts you need to know to go from one to the other. So, his paper and Dr. Broerse's paper were both very nice from the standpoint of general introductions to the field. On the cross section question, I think that although there are problems in the values of the cross sections from time to time, the much bigger problem is that ENDF/B is an incomplete set. In general, it does not give you the charged-particle energies which are what you need for an energy deposition calculation. It is much more likely to give you neutrons than gamma rays so one way of playing the game is to take the energy available, subtract off the gamma-ray energy, subtract off an escaping or scattered neutron energy from a reaction and then say the rest has gone to charged particles. Well, how well you can do this depends on whether you are given both the neutron and gamma ray energy so for the kind of application we are talking about the biggest problem is really definitive secondary charged- particle energies. The elements in which the cross sections are needed are in carbon, nitrogen, and oxygen. An example that has come up several times in neutron dosimetry meetings is: There is a lz C(n,n'3a) reaction which is known and measured. You would think therefore there might be an 16 0(n,n'4a) measurement and in fact it's got something like a 14- or 15-MeV threshold. To my knowledge, no one has ever measured it even once at one energy so that we are really in an area where there are lots of cross sections that are not known. I guess I'll make two more comments and then shut up. One is that in the personnel monitoring area I think accuracy is required which is easy for the neutron standards people. There is still an energy gap between thermal and 2 keV where there are lots of neutrons that people are interested in and where Auxier showed that successive intercomparisons improved performance. I think in his case that's true; it may not always be true in successive intercomparisons. It may be just that the field is getting better and inter- comparison had nothing to do with the agreement but I think that the case shown by Auxier is clearly one where the existence of the intercomparisons improved the measurement performance in the field. Prof. Havens : Any other member of the panel like to comment on it? Lee Stewart : At Los Alamos, we have had some interest and expertise in calculating kerma factors and have found, in addition to the problems in the ENDF/B cross sections, problems in the calculational methods. Phil Young, who is in the audience, can perhaps tell you how not to calculate kerma factors. As far as the data files are concerned, we have found that some ENDF materials make very good refrigerators in that the calculated kermas are often quite negative for many of the individual reactions. I have asked other people about their calculations and the Livermore Group, for example, told me they just set a negative kerma to zero, wherever negativity occurs in the reaction cal- culation. Therefore, some of our problems are in the ENDF/B representation of the data and others may be traced to the calculational methods employed. Prof. Havens : Shall I call on Phil Young to tell us how not to calculate kermas? Dr. Young : Well, we looked at several elements in ENDF and tried to estimate what the kermas were by doing what Randy suggested. Taking differences between total energy, subtracting out neutron energy and gamma ray energy, and frequently we would get things like negative numbers. Energy is not conserved accurately enough in the files to get kermas by subtraction techniques for many of the elements, mainly for medium and heavy nuclei. That was the conclusion. Dr. Peelle: I have formed an impression based on looking a little bit at the NASA work on this question some years ago as well as what's happened more recently. I think the personnel dosimetry problem, as one goes to higher neutron energies rising above successive reaction thresholds, is a very nice application of nuclear physics and related phenomena — related fields to some practical matter and a very difficult one especially as one gets above 15 or 20 MeV where I believe to do it right requires physics we haven't even figured out yet. So it's a challenging problem; its not just a matter of throwing off and transferring the information we really have to solve a practical problem. It's going to require some deep thought and is a very interesting as well as an important problem. Prof. Havens : Any other comments? Dr. Caswell : There were remarks made in some of the talks that the standards laboratories are not offering calibration services for biomedical application and therapy. I would like to say they are beginning now. NBS has been doing it for a little while. PTB and NBS are sort of in the status of beginning to offer known monoenergetic fluences for calibration. I think this is a step forward. I think doing the business for the therapy dosimeters at the high accuracy and following multiple methods to get the same answer has not been done and I might comment that we have two proposals which we had going for two years now neither of which has gotten anywhere to do that. Prof. Havens : Well, we'll proceed now to the next sub- ject which is "Benchmarks, Core Dosimetry"; Chuck Weisbin, would you report on the results of your work- shop? 352 Dr. Weisbin : Let me apologize in advance to my colleagues if I misinterpret or leave out any of the comments. The discussion was quite lively on the inter- action between differential and integral data. There seemed to be in our workshop essentially unanimous agreement that integral data must be considered in the process of cross-section standards definition. In order to achieve this worthwhile goal, important procedural issues must be addressed including which integral data, that is, what constitutes a standard integral benchmark and how is it to be included in the evaluation process. The methodology available for addressing the subject is fairly demanding in that it requires uncertainty infor- mation including correlations for both the integral and differential data and there I don't just mean standards, I mean all differential data. It expects quantitative estimates of possible methods bias as well as computed sensitivities of integral data to the various ENDF/B cross sections and their associated uncertainty files. Will they both represent only the differential data or will they both have to represent all available infor- mation? The characterization and definition of what is an uncertainty file may be difficult. It is clear that the methodology is becoming available but significant advances must be made in order to provide more credible input. The methods developed must be sophisticated enough to provide specific recommendations to an evaluator, for example, in distinguishing between modifications to the cross-section normalization and shape. In developing such methodology and data the very practical question is raised as to what kind of time frame would be required for an endeavor of this scope and indeed is it worth it. The process of assuring compatibility is clearly iterative and desperately needs documentation to improve best the communication between people, the real barrier, rather than funda- mentally conflicting data. A strong driving force, perhaps much too strong in the direction of reconcilia- tion of differential and integral data, would be to insist that a standard can only be so designated when it is judged to be consistent with all relevant informa- tion — both differential and integral. Prof. Havens : Thank you. That may rule out all standards. Alan, would you summarize this? Dr. Smith: I would like to back up a bit to the bench- mark fields particularly those under the auspices, I believe, of the Inter-Laboratory LMFBR Reaction Rates (ILRR) program. I think this has been one of the best I correlated national data efforts where people have really gotten together and through cooperation have come up with some very fine results. I heartily concur with what has just been said that these benchmarks in the real world of microscopic data are an essential part of the whole process. I don't think it's a goal in itself and I am worried that we have really not a measure of the sensitivity of the benchmark measurement in such things as the ISNF facility. It came out in the last paper this morning. There were some crude sensitivities, but it seems to me that's the first place to apply the sensitivity analysis. That's a first step toward the integral test and I also quarrel with an opinion which I've heard, and it wasn't expressed at this meeting, but I think I should express the view. A benchmark is a benchmark that tests some underlying principle; it is defined so in the dictionary. I think that you should never conceive of a benchmark as a method to measure microscopic data and I think that is something that is done periodically and I've had some words on that in the past and couldn't leave the opportunity go. I think too that I could not, in this vein, quarrel with the Farinelli syndrome which is I guess to, "build it and try and fly it". If a cavity and testing it all out in the benchmarks is going to be the way to an engineering goal, then I don't think I am in a position to challenge but when you say that the benchmark is go- ing to give you microscopic basic information, that it won't do. I was impressed by the interest in core dosimetry both at and in the core and beyond the core but I was left a little bit with the feeling that the accuracies that are required are really not all that great. That the people that are involved in that work, particularly the commercial people — their needs are mostly met. There are some "fine structure" things remaining but there are no gross holes. Quite a different area was the fission product, the burnup control, the analysis of core performance, and beautiful magnetic-spectrometry results from Idaho, and the excellent correlation of the chemical yield work here. Clearly the mass fields are changing violently with energy and I don't think it was fully appreciated what the impact of these, which I think really are proper fission-product standards, might be. I seem to recall that the way of monitoring the basic number of fissions in some of the best delayed- neutron experiments in the world was a counting of merely a hundred products. If that is true and they varied with energy, I question how good those delayed- neutron yields are now in view of these recent spectrom- etry results. The impact of those I think was very great. I think the Cf fission spectrum, as I mentioned earlier, should be accepted as a standard but, I got the general feeling that the averaged ratio rates given in various standard fission spectra, are running about two to three percent for Cf and that's probably not good enough. We are looking for another factor of two. Now, again with the previous speaker, I concur with his error-file concept. It's a wonderful world, but there are several things that begin to bother me a little bit. I wonder if the sensitive things are not the standards. It seems to me that in some of the examples shown yesterday, the fission cross sections, which were the standards in the Godiva and Jezebel assemblies, they were not sensitive really. The other holes were those holes of uncertainty where we're guessing. What is the inelastic cross section of 235 U? Well I'd like to know and I guess some other people would like to know and that is a very sensitive item and how do I place an error on that aside from saying I don't know. I think that you've got to be a bit careful on how you specify these error files. I also am a bit surprised that somebody hasn't given more emphasis to where you assign these uncer- tainties. Certainly the measurers and evaluators have simply got to give a great deal more attention to their error definition. Discrepancies are rampant throughout; people are just not representing things completely. It is a little like Ben Diven says, I guess, you correct it and put an error on everything you can think of but what do you do about the things you can't think of. Well you can't think of every- thing so people get ultra conservative but I think too that you might give consideration, particularly in the benchmark area, rather than assigning your discrepancies or your uncertainties and correlations to the microscopic data, follow the English course, as I understand it, and work with the correlated error files on the multlgroup cross-section sets. Prof. Havens : Dr. Farinelli would you comment on the subject? 353 Dr. Farinelli : Well, I think Randy Caswell mentioned before that this was the first Standards Meeting in which there was an attendance and the presence of bio- medical people. To a certain extent I think it was the first meeting in which there was a fairly consistent representation of integral people as well. My opinion is that dif ferentialists and integralists are much closer today than they were some years ago and that we are now starting to speak a common language. We seem to be able to understand and the fact that we have lively discussions is probably a representation of the fact that we have at least a common understanding - a common language. Otherwise, we couldn't even discuss. I think it is very important that we develop a common basis for understanding. It's quite clear that we must have some common way of looking, for instance, at errors, at error correlations if we want to assess what is the actual reliability of the data when they are applied to practical calculations. I don't think it is the best procedure that dif ferentialists and evaluators just well take their own responsibility to a certain point and then they leave the whole question to the others and say, "Well it's your business from now on." I think there should be some sort of hybridization of integral and differen- tialists to evaluate the final errors and so on and there are some other areas which we have not had time to mention but which are certainly interface areas and one of them I think is the area of how to make group cross sections, how to process the nuclear data. The way in which you process nuclear data involves both the way in which you have represented the nuclear data to start with and the way in which you are going to use the group cross sections and this is an in-between area which I think may be the cause of some of the discrepancy that we still observe. Prof. Havens : I guess we have come full circle; when the reactor business first started it was the measurers who were the cross-section evaluators as well as the reactor theorists. In the Fermi group back in Columbia in 1940-1941, and then as time went on, we developed various specialities and were so interested in our specialities that we didn't talk to our nearest neighbor. We then find out that it is impractical and then come back again, but in the original days of the Manhattan project there was no need for communication because the people that did the measurements also used the measurements. They made them because they had to have them immediately to proceed to the next step, so I'm surprised sometimes at what the graduate students in both physics and nuclear engineering know and I'm also surprised sometimes at what they don't know. Would any other member of the panel care to comment on this subject? Lee Stewart : I just might answer for some of the people here who have done the work. In CSEWG we have made many, many code comparisons to insure that we are processing the data in the same way or at least getting almost the same answers and I mean very close, depending on the methods that we use. There has been a great deal of work in this area in this country in CSEWG. Prof. Havens : Any further comments from the panel? Dr. Peelle : I have heard Chuck Weisbin give a talk on the subject of methods uncertainties and I'm sure Dr. Farinelli has made talks on that subject. Maybe we could hear just one comment which relates to this thing very clearly. How big are the uncertainties liable to be in calculated quantities, using the transport theory? Dr. Weisbin : I don't know, Bob, if you're asking for a number or just a qualitative statement. Number one, they are not negligible. That's the first thing. I think I concur with Lee's comment that they are getting better, and there really are two types. There's the type of methods bias that we can know and pin down, for example P., versus P., or 50 groups versus 100 groups versus 200 groups. There are harder methods biases like taking a three-dimensional real problem and modeling it in 2-D, and saying what is the error due to our -modeling when that is the best we can do. So now I go back to Al Smith ' s comment what do you do about the errors that you don't know about and of course those are harder. I would say that in general if you wanted a number though; well, let's take the number, k. I would say that methods errors today in the calculation of criticality are not less than three tenths of a percent. Prof. Havens : Do you agree? Dr. Farinelli : I concur. Prof. Havens : I remember one time when there were about five measurements of the 235 U cross section at thermal, all of which were good to two percent and all of which varied by about 15%. They were poor measure- ments because the errors had been incorrectly deter- mined. Prof. Havens : We'll proceed on to the next subject which is rather a difficult subject to bring together, "Techniques and Methods." However, there was a work- shop which is close to that although not exactly, that's the first on the list of the workshops, "Establishing Neutron Energy Standards," and would Dorien James please give a summary of that panel's discussion? Dr. James : Perhaps I could say again that the work of the INDC subgroup in selecting a list of narrow resonances simply for use as energy standards could be regarded as complete and the list is given in Table 7 in my paper. The next stage is to evaluate the data available on all these resonances and establish the best values, and it is hoped that workers on each spectrometer will now and in the future publish their best values for these resonances. Mughabghab and Bhat have suggested to me that the list of best values should be published in the next addition of BNL-325. I've agreed to bring these results together. Now Joe Fowler suggested that someone should write a review article giving all the details of all the measurements involved in deriving the best values. He twisted my arm into doing this. He threatened to tell Basil Rose, if I didn't do it. You must wipe that out, Charlie. I guess you've got an expletive delete button on these federal tapes that you are producing. Now this would, I hope counter Alan Smith's strong objection to using any data as a standard which is not fully documented. I agree with that sentiment but I think that if we demand of all the measurers that they fully document their results, then maybe we would have to wait a long time. It looks from the evidence that white sources are now in fairly good agreement except perhaps the Geel-Columbia 238 U discrepancy which I mentioned this morning. The care- ful work of Meadows has shown that white sources and monokinetic sources can be brought into agreement also. Now many white sources have certain substances either always in the beam, for instance, ORELA bind the 10 B filter with sulfur so they always have sulfur in the beam and on the synchrocyclotron we use background filters so that our measurements can have silicon oxide in each run. These enable certain resonances on the list to be continually checked and it is suggested that the values obtained for these resonances should be included in the publication of any data. Anytime you 354 publish data you should also report your value for the resonances on the list, just to confirm that you are still checking. Finally, there might be a slight problem with poorer spectrometers because they may not see all the narrow resonances on the list, but in time of course they will be able to use the energies of broader resonances given by the laboratories which do quote their values for the narrow resonances. Prof . Havens : Thank you. Alan, would you put together the summary? Dr. Smith : I regret the action Dorien has taken because if he is going to write a comprehensive review with all the errors, I'm not going to see a good friend for 18 months, I guess. You're out of business, but I think it is certainly a proper thing to do and I encourage it. I think at the higher energies which I'm familiar with, the whole problem has essentially gone away. I think that the remarks this morning that there are discrepancies but with an exception of perhaps one fission cross section, for most practical reasons, they are negligible. What really bothers me is that for this example, and I think also one dealing with the fis- sion neutron spectrum, considerable effort has been expended because of one or two measurements which turned out to be a little in error. I think one should be careful before you immediately take some action that you carefully review any standard measurement and try and find out if there is a problem or something. It is kind of a double-edged weapon. There was a great deal of effort put on fission cross sections and I include fission spectra and I think that was very warranted; we didn't know much. We know a lot more now. But in fact the disaster that we thought was there from some preliminary estimates was not there. I think there is a word of caution not to get carried away when somebody has a preliminary result on some of these standards that goes somewhere different. Take a good look at them, before everyone takes off and makes some measurements . I would like to give justice particularly to all the people who talked about instruments. If I have any competence, it is didling with instruments, and I realize how difficult and how complex these things are. It was a very diverse discussion here with many contri- butions, and I can't begin to do justice to them. There were many of them that were very elegant and many of them were carried to very high degrees of accuracy; but if I recall, only two of the instruments were not described at the last standards conference. Really what is talked about is refinements - very hard to get and very well done but not a really new instrument with two exceptions which are really modifications of other schemes. One that I mentioned before was the flat counter from Livermore which has a tremendous potential for getting over this gap between the (n,a) resonance and one MeV. I think one of the great advances in the instrument area in the last 10 years which I never would have believed 10 or 15 years ago, is the ability to use the hydrogen recoil counter reliably at low energies. Just a technological achievement, but if somebody would have told me in 1970 that you could monitor a beam in a linac with precision at 50 keV, I would have laughed. It's done and I think that it is done well. I think certainly that the sharp improvement in reference fields is a real key to the dosimetry problem particularly the very high purity fields which have the peculiar problems for dosimetry that they have here at the Bureau. I am not sure their impact was made clear; you may not find those so valuable for some research purposes, but it is clear that they are essential to the proper calibration of dose. I think we still have a big hole - the one thing which I have always been looking for - a flat detector with a high sensitivity. It's the crux of a lot of problems and I don't see it coming. For example, how do you solve the delayed-neutron issue which has been a very serious bottleneck for a number of years. Prof. Havens : Thank you. That reminds me of one time a navy Admiral was trying to describe everything that was desirable in an anti-submarine detector, and I think it was Panofsky who got up and said, "Admiral, we can only use nature we can't coerce it." Seems to me that Alan is trying to coerce nature into getting the ideal detector. I don't know how we can do that. Dr. Smith : I think that we should point out that one of our medical people wanted a spectrometer which would go from one kilovolt to 20 MeV with an absolutely determined response so they have their problems too. Prof. Havens : We all have problems, but we can only use nature, we still can't coerce it. Dr. Motz would you take the lead on this? Dr. Motz : Several other comments have been made about detectors and the fact that perhaps boron detectors deserve more attention and Lee has mentioned that hydrogen recoil isn't used as extensively as one would expect. Another detector that I think has hardly been mentioned is the 3 He counter. It has some disadvan- tages but it certainly has a good predictable response. It's a question of sensitivity and packaging but certainly one that deserves some attention as a standard reference. I am very pleased to hear that Dorien is going to document the energy standards and I wish him luck. I hope he can come up with the enduring values that he was trying to explain to us during his talk. I would be especially interested in seeing that, Dorien, because I have some degree of scepticism about the high degree of accuracy that is claimed in the final results. I think that it does take a careful analysis to convince one that they are really known as well as you were indicating. I wanted to make one comment about high-energy neutron sources. With respect to the presentation on associated-particle methods, it was mentioned that a lot of the defects which occur in neutron sources and associated-particle work are really associated with the solid targets that are used and many of these problems go away when one uses a gas target. There are in fact two recent reports out about purity of 8-14 MeV neutron sources considering most of the parameters that one could change such as beam stops and collimators defining beams and things of this sort. This work has been done at Los Alamos both for the t-P and the p-T sources and compared with d-D source and the second report is from Bruyeres-le-Chatel in France, similarly for p-T sources. Prof. Havens : Would any other member of the panel like to comment on this subject? Dr. Liskien : In the field of standards one has to use any occasion to check consistencies and I see a broad field where this is not fully exploited and I mean linac measurements which typically use shape monitors and then go from the eV to keV region to normalize, using hydrogen telescopes or similar devices as shape monitors. I could imagine that it must not be too difficult to go one step further from a shape monitor to absolute recoil counting and thereby provide additional information. I think nobody should be afraid that other inconsistencies could come out when you have two methods of normalization. Another point 355 is that I think in fact we have not really seen many new techniques in this symposium. I think one, not a new method, but at least a way of applying a technique has been demonstrated and I think it should be mentioned, I mean associated-particle counting, not the particle itself, via its activity or by counting but really by its gamma-deexcitation as Dr. Brandenberger has demon- strated. I think we should see the people that are doing absolute flux determination, and inquire what they really can do with this method with other sources. Prof. Havens : Any further comments from members of the panel? Dr. Peelle : I would like to just accent, another time the use of diverse techniques. We have heard in many contexts, that this is the key to long term success in the standards' field. One example is Dr. Knoll's paper in which he described the work he and his students have done at a few energies with y-n sources. That's not a new idea but it appears that it is being pushed in a logical way to give highly credible results which seem very independent of most of the other work that is done, and therefore very valuable. Prof. Havens : Lee Stewart do you have any comments? Lee Stewart : If I may just mention a situation at Los Alamos many years ago; a young scientist started working with the n-p cross section and he accordingly calibrated his detector. Then he decided to measure the hydrogen cross section with this calibrated detector and he was quite discouraged when, at some angles he could not reproduce the hydrogen cross section better than approximately 15%. Naturally, he went back to the drawing board. Therefore, I believe the business of inter comparison is extremely important. Prof. Havens : Any comments from the audience? Then we will proceed to a subject which has been controver- sial ever since this field started; that's the 235 U fission both at low and high energy. Now Dr. Cier jacks was the leader of a discussion group on the 23 % cross section. Would you give the summary, Dr. Cier jacks? Dr. Cier jacks : The fission cross section of 235 U is in quite good shape presently, nevertheless there are a few problems left. As a result of the discussions we had in the working group, we came up with a few observations and recommendations which I should like to read to you. Thermal Value: The group felt uncomfortable about the fact that evaluations gave about 0.3% accuracy for the actual thermal value while deviations are somewhat larger than the quoted values. Coming now to the situation of the 9-eV resonance region. Here we felt that more evaluations and more measurements are needed. The data in hand when published and evaluated could well result in a +1% accuracy of the fission integral in that resonance region. Let's go on to the remaining eV range; there we have two statements. First, another normalization point in the resonance region should be chosen to allow for a cross check of the data. A suitable resonance range would perhaps be between 33 and 41 or between 41 and 60 eV. Second, an additional reference band would be desirable in the range of per- haps 0.3 to 1 keV. With respect to the keV range, the following recommendation was made. Absolute white- source measurements with adequate accuracy ought to be encouraged in that region. Finally we had a recom- mendation for the MeV region which might not neces- sarily be limited to 235 U. Consideration should be given to fission reference cross sections in the range from about 10 to 40 MeV. Prof. Havens : Alan, could you give your comments on this subject? Dr. Smith : As many of you know, eight or nine months ago there was a special workshop conference at Argonne on fast fission cross sections of U-235, U-238, and Pu-239. All the available data were reviewed. There is a book over there in the coffee room. One of the outcomes was that certain areas were identified as being critical, other areas were in good shape, and we hoped that as a result some work could be done on those selected problem areas. In fact some was, and some is still in progress. I am pleased to say that I think that was an effective result of that conference, and some of it is reported here. I think as a result of this work, depending on how you want to view your various interpretations, the fission cross section of 235 U from a few keV to less than 2 MeV is known to about 2 percent. Up to 8 MeV I would estimate 3 per- cent, at 20 MeV I would estimate 5 percent. That puts it within the requirement, for example, of the bio- medical people. One of the issues at the conference was structure in the fission cross section, it's there. There were examples shown at this meeting where careful search shows that it isn't particularly correlated in some cases, in some cases it is. I think the issue is not really important at the present time. I think that now the thing to do is to continue the work, which the previous speaker emphasized, to try and get a normal- ization of the white source measurements using an absolute calibration. I am very leery of this low- energy extrapolation. I think that the consistency with this and the higher-energy normalization has got to be achieved. A gap exists, and until you can get across it, I think you've got a real problem. The work is particularly going on here at the Bureau in that area, and I guess other places, and I suggest that these people get together and they keep in better contact with one another and that they attempt to put together a preliminary evaluation of these results in a better correlated manner. I agree with the opinion that was quoted at the conference at Argonne that I don't think any formal evaluation should use unfinal- ized or unpublished results. I think preliminary results are out. I'll extend that statement a bit farther, and I think I already made it today, that no evaluation should be distributed for final use or considered final until it is fully documented too. And I think a lot of the problems that are involved now are the interpretations of evaluations, how you want to interpret the data, and it's all rolling around in a ball of wax. Now I think with this approach there are some other serious concerns. I think that you might, as I think it was expressed at this conference, wonder whether you need this cross section any better until you can master the ratio problem. I looked at those ratios again last night, and well, depending upon who you want to believe, it's one to three percent uncer- tainty in the ratio of 235 U to the other fission cross sections. It certainly is not an order of magnitude better and maybe not even a factor of two, or maybe even worse. I think before you push too hard on the problem of the absolute 35 U cross section, if it is correct to two percent over most of this range, you'd better master the ratio problem. I also think that one should give far more attention to the weighting of the absolute source measurements; I believe it was Bob Peelle who just mentioned them here a minute ago. I think that there is an awful tendency in fission cross section measure- ments, both ratios and absolute ones, to look at the mass of points and to draw a line through them, and to ignore these highly precise and quite independent check points. I think that's a mistake that has been made in a number of contexts, and it's acute in this case. Only four or five points which appear to be 356 excellently done, and they get lost in the wilderness; but they're very essential normalization points. The same thing tends to be true of white-source versus monoenergetic-source machines. I think that I would then like to go on to the problem of the high-energy fission standards. I certainly subscribe to the statements made by Siegfried Cier jacks a minute ago. Yes, we need a high energy fission standard from 14 MeV to 20 MeV, with perhaps 2 or 3 percent accuracy. What bothers me I think is the problem of structure in the 238 U cross section. The 238 U cross section has structure, I'm sure, and some very excellent measurements show it. Apparently it is not so prominent at all in 237 Np. But I think before we change standards one should be very careful to confirm the exact magnitude and nature of this structure. Furthermore, I had a little inspection of the data I could find last night, and I concluded that with any reasonable working experimental average you in fact could not be perturbed by that structure, and I am a little afraid that this is going to be a reflection of a syndrome, and I'm sure people will take exception to this, but a syndrome for resolutionmanship, as Henry Newson used to call it. In many of these things, let's not be possessed with the structure. I think Paulsen was right; gold's got structure, but when used with judgment, it's a good capture standard, and the same may be true for 238 U fission. I think, too, that we should, of course, in this high energy range, before we put a big crunch on a new standard which does not have a direct application such as 238 U, always remember that this is a region, if no- where else, where hydrogen should work. Do we really have a need for another reaction standard in this area? Well, that's where I'd like to stop on fission cross sections. I'd like to make one more remark which is applicable to fission cross sections, but more generally applicable than that. It is aimed at our evaluators who, I must confess as my operating budget gets smaller and smaller and as I tend to spend more time in my office punching square holes in long cards, I seem to be joining. There seems to be an awful rigidity to evaluations. I don't think there's a sin in being smarter today than yesterday and that the result of new information or new thoughts ought to be incorporated into an evaluation - even a standard. I think we should be more adaptable in adjusting to changing conditions and changing knowledge. I guess that's in a direct oppo- sition to the way the world is going, but I seem to have this feeling that we're awfully rigid in what we're doing. Even if somebody comes up today with a new standard result, it's going to be four years, if we're lucky, before anybody gets it out of their private files and used. And I think that's a very serious problem. There's a need for a great deal more responsiveness . Prof. Havens : Thank you . Dr. Peelle : There were many remarks in the last few days about the problem of normalizing cross sections near thermal energy, in addition to my own. One point I didn't make during my talk but which Dr. Bhat made at Argonne was, that in spite of the great discomfort with some of the intermediate results that one looks at in making such a normalization effort in 235 U, by the time the CSEWG Task Force with its imperfect understanding of everything got to about 100 keV, the seeming discrepancy on the average with the mono- energetic sources was of the order of one or two per- cent which is well within the uncertainty of the measurements. So perhaps the central limit theorem or law of averages has helped out. Perhaps it has not been so bad as it could possibly have been with the uncertainties along the way. We were challenged to come up with an uncertainty on the request for a one percent 235 U cross section. Based on the evaluation of the statistics of the various request lists I've seen over the years, the answer is 1+0.6 percent accuracy without any statement whatso- ever as to what that means. I think it must mean, since most of the uses are somewhat integral in nature, that this kind of accuracy is needed in the integral over a substantial energy region including at least a few lethargy units. Now how we verify that as a desired accuracy, I don't really know. Most of those numbers were derived by people a decade or more ago that I didn't talk with. Since that time, in the case of 235 U, the cross section I think on the average has changed by several times that amount. But until we get stability somewhat near that value, I don't think we need to worry too much about the precise number. But how do we know how well we must do? That's a question we'll have to face within a few years as the standards field becomes more mature. It seems to me that the work that Chuck Weisbin reported on, although of preliminary nature in a sense because all the necessary inputs that he said he needed were not present, does start to show in a fairly vigorous way what the impact of a change in a standard or an uncertainty in a standard really is. In the analysis of integral benchmarks, which are dear to the heart of the applications people, changes at the 1% level seem to have a significant impact, even in an integral experiment that contained no uranium. So the first tentative systematic study that I've seen suggests that there is a definite impact of knowing a standard to high accuracy perhaps better than the present know- ledge. I think that since 1970 there's been a dramatic improvement in the 235 U fission cross section. Dr. Poenitz suggested that indeed over some broad band it's something like 7 percent in seven years. Now I don't think it will do that in the next seven years, but who's to be sure, especially if we achieve the flexibil- ity in using our knowledge that Dr. Smith asked us to show. Prof. Havens : Would any other member of the panel like to comment on the 235 U cross section? Lee Stewart. Lee Stewart : Well I suppose that you can plot data and, if you know what you want to show, you plot it that way to show it. But I believe I would find it very difficult to accept a 2 percent error on 235 U above a few keV, remembering the work that we've done for sometime now in CSEWG. In addition, if I remember correctly, and if people have not changed their data, the structure that we put into version IV in 235 U was based on a comparison of four separate experiments; one from Harwell, one by Charlie Bowman from Livermore, one from Los Alamos by Lemley, and Reg Gwin's from Oak Ridge. The structure did line up. We did find that they did not identically line up, but saw the same peaks and valleys in all four experiments. The main reason, however, we chose to put in the structure is because there are wide minima, there are 15 percent deviations, and therefore a person doing a monoener- getic experiment will measure a cross section depending upon his resolution, if you assume that those data are correct. Therefore I believe that we do need the structure in the cross section. Prof. Havens: Further comments? 357 Dr. Poenitz: When structure shows up, I always take the practical point of view; I ask, what is the effect in a calculation? For example, what is the effect on a calculation using the 238 U capture cross section when the structure differs by an order of magnitude? While the structure may be interesting, I don't believe it has any practical implication, and the calculations agree with this . The major comment I wanted to make is in regard to the question of the hydrogen cross section as a standard. First of all, there is no standard cross section anyway because standards can only exist for basic units, like length and time. I noticed you have the derived quanti- ties, and so each new measurement necessarily changes that so-called standard; we should probably call it a reference. From that it follows immediately that there can be no difference among the standards. There is no primary standard. The argument that the hydrogen cross section, for example, is the best known, is a true argument or true statement, but it does not mean that the measurement using the hydrogen cross section is pre- ferable to any other one because the cross section is not the only quantity involved. The efficiency of the detectors is the second one. So the absolute measure- ment of the fission cross section may bring us a much better reference cross section than the use of the hydrogen would permit us to do. As a matter of fact the question comes up now in the high energy range where we don't know the angular distribution of the hydrogen so well. It may be much better to use say a fission cross section where we have a very high precision in the efficiency of the counter. Prof. Havens : Anyone care to comment on that? Dr. Peelle : A small piece of it I'd like to comment on. It's true that we did one study for the unresolved resonance region in 238 U, looking at fast reactors, and found the fluctuations which have been seen there with indeed much less resolution than is available could not be shown to have an impact of any importance relative to the error in the average cross section. However, I think that one has to have a little reserva- tion about that study and recognize that in any such question about the importance of fluctuations you have to ask about the particular situation. In general, I think we've gotten in trouble with our reactor engineer friends several times by having given them the impres- sion that cross sections are smooth. They believed the graphs they saw in the book. Even though we know they weren't smooth, they didn't know they weren't smooth. It may be best to represent them to the extent that we know them. Prof. Havens : Thank you. Dr. Smith : What does Ugo Farinelli think about that? Dr. Farinelli : This is I think a good example of the influence of the averaging procedure. It depends on whether you make the average from very detailed structure in the differential measurements, or whether it's the measurement itself that makes a certain average. I have always thought that it would be conceivable to make measurements that measured directly the average in the sense that it was to be used, especially when the detail is so great that even if theoretically you have the methods to weigh the structure differently according to the composition, which should be done, practically you don't do it because of the complexity. If you have say 100,000 points, you are not going to weight this over any spectrum; you're simply going to make a very straightforward average, independent of composition. So I think a measurement can yield the same result. Well, it may not be perfectly correct, but this is going to be the same value that is used by the integral man who is doing the calculations. So I agree that in principle the man who makes the measurements would like to give something which is perfect and then it's the responsibility of the man who uses his data if he fudges with them. But from a practical point of view I don't see the difference. Dr. Peelle : I did make a quick effort to check that point on the question of thick-target transmission and self-indication measurements for 238 U, and to compare that to what one sees in a processed cross section taking into account what are thought to be the fluctua- tions. Even though the two studies are based on the same phenomena of the self-protection in the resonances, I could not demonstrate a direct correlation for either' of those two types of measurements and the processed cross sections which are now found. Not only was there none, but I could not show there should be. I think I may have erred, but sometimes it will be necessary for the calculational expert to take the details in some form or another and deal with them. Prof. Havens : Francis Perey. Dr. Perey : I'm very much interested about the state- ment regarding structure details. As you know I've struggled with the problem, but you tell me that I must average the cross section in the measurement pro- cess or the evaluation process and give you a smaller number of points. We've tried to do that for years and never succeeded. I don't know how you're going to homogenize your calculations, whether you're going to use a couple of mm or one mm of fuel cladding or one meter of iron in your shield. Which average should I use to put into the file when I have structure, when I know the structure exists? Prof. Havens : Do you want to answer that Ugo? Dr. Farinelli : Yes. You don't know what to do, but you should not think that people who make the calcula- tions know what to do. It's not really conceivable that somebody making the calculations for the reactor uses a different cross section if he moves from one point of the reactor to the other. Do you really think that somebody is going to take a different cross section for iron or nickel in the cladding or in the follower of the control rod, in the core or in just a little way off? I don't think so. It's common to have just one cross section for iron, at least in the core. You may have two if you go to the blanket or to the shield, but certainly in the core you have just one cross section. How are you going to calculate it? With an average composition over all the core or over a system of cores which are not too different from each other. From one point to the other there will be differences, but the differences will not be reflected in any set of cross sections used. So the problem is there. Prof. Havens : Lee Stewart. Lee Stewart : Let's get back to 235 U. It is a standard, and I dislike penalizing a monoenergetic experimentalist who does an experiment with very good resolution and therefore measures a cross section that is 10 percent lower than anybody else who does one with very poor resolution. And for that reason alone, if none other, I think we should be willing to put in the structure in j5 U that has been measured by at least four people. That was my primary reason for thinking that the struc- ture should go in. It did bring in line, by the way, a lot of older experiments. It did show us where some of the problems are. Prof. Havens: Chuck Weisbin. 358 Dr. Weisbin : I think this one study that we're talking about here in intermediate structure unfortunately can go back to something we did. This was one study, one case, not a large sampling. I think, however, in the future years we are going to see advanced techniques which can make use of the structure, Ugo, in the sense of the work that Red Cullen is doing on probability tables, where he tabulates the probability of a given cross section versus that cross section in a given energy range. This is a very efficient representation that can make use of the structure so that if evaluators did the smoothing ahead of time, and just gave us a single average, we could take advantage of that method- ology. However, if we do have the access to the structure, we can put it in an efficient fashion, and I think we will find out we get different answers, at least in some cases, not yet shown, admittedly. Prof. Havens : This problem of how much detail you should put in the cross section and then what you can use and what you assume and how you average has existed, I'm sure, ever since nuclear energy has existed in the form of reactors, and if you had seen some of the averages that Hans Bethe made in 1943 and 1944 you would be really flabergasted at some of the assumptions that went into the formulae. I remember very well when I first showed Neils Bohr a resonance in 235 U he said, "This can't happen". So, there are some problems with the detail you put in the cross section and the theo- retical models, not only of the nucleus but of the reactor as well, and you're limited to what you can do in a practical sense. Dr . Bowman : I think that's an interesting story because what you're saying is that as years have gone by and calculational techniques via computers have improved, that we've been able to handle more and more structure. It seems like a lot of the discussion here this afternoon assumes that we are at the end of that calculational development and calculational capability. But, if I read the latest issue of Science where they talk about what's coming in the speed of computers and so forth and assume that that's going to be implemented in the nuclear power industry, it would seem to me that there's quite a likelihood that we will develop these techniques. Maybe the technique that Chuck was referring to a minute ago is just the next of a series of steps into the future. Prof. Havens : Well the conference will be brought to a close, but before we do that I'd like to thank the panel members and Alan Smith for summarizing the con- ference. I'd also like to take this opportunity to thank Charlie Bowman for organizing the program, which, when he first said we're going to spend four days on standards, I said, "Oh, my God, not that." But cer- tainly the interest of this conference has held up a lot more than many of the meetings that I have attended. The fraction of the audience here on Thursday afternoon to that here on Monday morning is very much higher than it is in most technical meetings that occur. The stan- dards, someone said, only are needed after you've made several measurements and after a field has been pretty well developed. I think that's true if you only have one measurement. Then you don't need a standard, the one measurement is the standard. I don't want to get into a debate with Poenitz about standards because we have talked about standard reference data and there are various definitions of standards. I will suggest that you read the definitions of standards in some of the literature put out by the National Bureau of Standards, and you'll be surprised at what some of the standards are. Some of them are arbitrary, some of them are natural, in that they are physical measurements, but they vary all over the lot depending on the field that you're working in. But I think that Charlie and the Bureau have done an outstanding job. It's the first conference I have attended here at the Bureau, and I think that it has been a very well run conference, and the Bureau has excellent facilities which we've all enjoyed. I think we should give a vote of thanks to the Bureau, and we certainly appreciate the activity in this field that has been demonstrated here at this conference. Thank you, Charlie. I now declare this meeting adjourned. Prof. Havens : I think with the development of reactor design and reactor calculations over the years, the capability of computers has very often been the limitation of what you can do; and as that's improved, the calculational accuracy of what could be predicted has also improved and will continue to improve. How- ever, I was asked to try to bring this session to a close shortly. Before we finish the session, I'd like to ask Alan Smith if he has anything further to say. Dr. Smith : I do. I am brought back to Farinelli's point of view. I really think the greatest value of standards is the all-important ability to calculate the performance of the system within a measurable range. And if you can't do that, you could never expect to make the extrapolation of the calculation to an unmeasurable range which you'll have to do to prove acceptance in safety. So I think there's been a change in the values in my mind of standards from primarily economic now to safety and public acceptance. Now it may change back after you've passed a period to economic grounds, but I think that's the key. Your calculational ability is the thing that's important now. It's your ability to predict reliably and show that you can do it. Without that, I don't think your technology is going to fly. 359 LIST OF PARTICIPANTS Alberts, W. G. Physikalisch-Technische Bundesans talt 100 Bundesallee D-3300 Braunschweig WEST GERMANY August, Leon S. Naval Research Laboratory- Code 6643 Washington, D. C. 20375 Auxier, John A. Health Physics Div., ORNL Route 1, Jones Road Lenoir City, Tennessee 37771 Awschalom, Miguel Fermilab P. 0. Box 500 Batavia, Illinois 60510 Ax ton, E. J. National Physical Laboratory Teddington Middlesex TWII OLW UNITED KINGDOM Barschall, H. University of Wisconsin Engineering Research Building Madison, Wisconsin 53706 Bartholomew, G. A. Atomic Energy of Canada Limited Chalk River Nuclear Laboratories Chalk River Ontario, CANADA K0J 1J0 Bartlett, William T. Battelle, Pacific Northwest Labs. Battelle Boulevard Richland, Washington 99352 Bastian, C. CEC, CBNM Euratom Steenweg naar Retie B-2440 GEEL BELGIUM Behrens, J. W. Lawrence Livermore Laboratory P. 0. Box 808 Livermore, California 94550 Benjamin, Richard W. E. I. du Pont de Nemours and Co. Savannah River Laboratory Aiken, South Carolina 29841 Bergqvist, I Dept. of Physics, University of Lund 14 Solvegatan S-22362 LUND SWEDEN Bhat, Mulki R. Brookhaven National Laboratory T197C Upton, New York 11973 Blinov, Mickail Chlopin Radium Institut Leningrad USSR Block, Robert Rensselaer Poly. Institute Troy, New York 12181 Boldeman, John William AAEC Private Mailbag Southerland N.S.W., AUSTRALIA Bowman, Charles D. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Brandenberger, Jerry D. Los Alamos Scientific Laboratory Box 1663 Los Alamos, Nex Mexico 87 545 Broerse, J. J. Radiobiological Institute TNO 151 Lange Kleiweg NL-RIJSWIJK THE NETHERLANDS Cance , Michel^ Commissariat a 1 Energie Atomique B.P. n°561 F-92542 MONTROUGE CEDEX FRANCE Carlson, Allan D. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Carter, Robert E. Institute for Resource Mgmnt. Inc. 9512 Edgeley Road Bethesda, Maryland 20014 Caswell, Randall S. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Chemtob, M. Dept. de Protection SPS/LDS Centre d' Etudes Nucle aires de Fontenay-Aux-Roses B.P. n° 6 F-92260 FONTENAY-AUX-ROSES, FRANCE Chrien, Robert E. Brookhaven National Laboratory Physics Department, Bldg. 510A Upton, New York 11973 Cier jacks, Siegfried W. D-75 Karlsruhe, Weberstrasse Karlsruhe 1 West Germanv Cosack, Martin Physik Tech. Bundesans talt Bundesalle 100 33 Brannschweiz WEST GERMANY Czirr, John Be Lawrence Livermore Laboratory P. 0. Box 808, L-280 Livermore, California 94550 Davis, Mark C. Dept. of Engrg. Sciences The Technological Institute Northwestern University 2145 Sheridan Avenue Evanston, Illinois 60201 Derrien, Herve CCDN/NEA OECD B. P. n° 9 F-91190 GIF-SUR-YVETTE FRANCE Dierckx, Ronald P.O. Box 1663 Los Alamos, New Mexico 87545 Diven, B. C. Los Alamos Scientific Laboratory P. 0. Box 1663, MS 442 Los Alamos, New Mexico 87 545 Duvall, Kenneth C. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Eisenhauer, Charles M. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Engdahl, John C. Department of Physics University of Michigan Ann Arbor, Michigan 48109 Fabry, Albert CEN-SCK (Belgium) 1900 Stevens Drive 214 Richland, Washington 99352 Farinelli, Ugo CNEN-CSN CASACCIA 00060 Santa Maria Di Ualeria Roma, ITALY Farrar, Harry IV Atomics International 8900 DeSoto Avenue Canoga Park, California 91304 Flemming, Dale M. Battelle, Pacific Northwest Labs. Battelle Boulevard Richland, Washington 99352 Fleming, Ronald F. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Fowler, Joseph L. Oak Ridge National Laboratory Oak Ridge, Tennessee 37830 Fuchs , Jean Eric 60 G rue Albert Garry LIMEIL-BREVANNES, FRANCE 94450 360 Gabbard, Fletcher Department of Physics & Astronomy University of Kentucky Lexington, Kentucky 40506 Geiger, Klaus W. National Research Council of Canada Division of Physics Ottawa, Ontario CANADA K1A 0R6 Gilliam, David M. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Goldman, David T. Institute for Basic Standards National Bureau of Standards Washington, D. C. 20234 Goodman, Leon J. Radiological Research Accelerator Facility, Columbia University Bldg. 902B, Brookhaven National Lab. Upton, New York 11973 Green, Lawrence Bettis-Wes tinghouse P. 0. Box 79 West Mifflin, Pennsylvania 15220 Greenwood, Lawrence Argonne National Lab. 9700 South Cass Avenue Argonne, Illinois 60439 Grundl, James A. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Hajnal, Ferenc U.S. ERDA Health & Safety Lab. 376 Hudson Street New York, New York 10014 Hale, Gerald M. Los Alamos Scientific Laboratory P.O. Box 1663 Los Alamos, New Mexico 87544 Harrison, George H. Radiobiology Section Univ. of Md . School of Medicine 660 W. Redwood Street Baltimore, Maryland 21201 Harvey, J. A. Oak Ridge National Laboratory P. 0. Box X Oak Ridge, Tennessee 37830 Havens, William W. Jr. Columbia University 520 West 120 Street New York, New York 10027 Heaton, Henry T. , II Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Hemming, P. B. U.S. ERDA Washington, D. C. 20545 Hill, N. W. Los Alamos Scientific Lab. P.O. Box 1663 Los Alamos, New Mexico 87545 Howe r ton, R. J. Lawrence Livermore Lab. Box 808, L-71 Livermore, California 94550 Huynh, V. D. Bureau International des Poids et Mesures Pavilion de Breteuil F-9 2310 SEVRES FRANCE Islam, M. Mizanul Bangladesh Atomic Energy Commission Atomic Energy Centre P. 0. Box 164, Dacca-2 Bangladesh Jaffey, Arthur H. Argonne National Laboratory 9700 S. Cass Avenue Argonne, Illinois 60439 James, G. D. Atomic Energy Res. Est. HARWELL, Didcot, Oxon 0X11 Ora UNITED KINGDOM Johnsen, S. W. Crocker Nuclear Lab. University of Calif., Davis Davis, California 95616 Kazi, A. H. Army Pulse Radiation Facility STEAP-RF, Bldg. 860 Aberdeen Proving Ground, Md . 21005 Kellie, J. D. Glasgow University Kelvin Laboratory N.E.L. East Kilbride, Lanarkshire SCOTLAND Knitter, H.-H. CEC,CBNM Euratom Steenweg naar Re tie B-2440 Geel Belgium Knoll, Glenn F. 2266 GGBL University of Michigan Ann Arbor, Michigan 48105 Knox, Harold Physics Department Accelerator Laboratory Ohio University Athens, Ohio 45701 Kuchnir, Franca T. Dept. Radiology, Box 442 University of Chicago 950 E. 59th Street Chicago, Illinois 60637 LaBauve, Raphael J. Los Alamos Scientific Laboratory MS-243, P. 0. Box 1663 Los Alamos, New Mexico 87545 Lachkar, Jean C. Commissariat a l'Energie Atomique Centre d' Etudes de Bruyeres-le-Chatel B. P n° 561 92542 Montrouge Cedex FRANCE Lamaze , George P. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Lane, Raymond 0. Physics Dept., Accelerator Lab. Ohio University Athens, Ohio 45701 Leiss, James E. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Leonard, Bowen R. Jr. Battelle - Northwest P. 0. Box 999 Richland, Washington 99352 Lippincott, E. P. Westinghouse Hanford Company P. 0. Box 1970 Richland, Washington 99352 Liskien, H. CEC, CBNM Euratom Steenweg naar Re tie B-2440 Geel BELGIUM Lisowski, Paul W. Triangle Universities Nuclear Lab. Dept. of Physics, Nuclear Bldg. Duke University Duke Station Durham, North Carolina 27706 Lone, M. A. Chalk River Nuclear Labs. Chalk River, Ontario K0J 1J0 CANADA Lurie, Norman IRT Corporation 7650 Convoy Court San Diego, California 92111 McDonald, Joseph C. Sloan-Kettering Institute 1275 York Avenue New York, New York 10021 McElroy, William N. Westinghouse Hanford Company P. 0. Box 1970 Richland, Washington 99352 McGarry, Emmert Dale Harry Diamond Lab 2800 Powder Mill Road Adelphi, Maryland 20738 361 Maeck, William J. Allied Chemical Corp., INEL 550 2nd Street Idaho Falls, Idaho 83401 Magurno, Benjamin A. Brookhaven National Lab. T-197 Upton, New York 11973 Marshak, Harvey- Heat Division National Bureau of Standards Washington, D. C. 20234 Meier, Michael M. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Moazed, Cyrus Harry Diamond Laboratories 2800 Powder Mill Road Adelphi, Maryland 20783 Moore, Michael S. Los Alamos Scientific Lab. P.O. Box 1663 Los Alamos, New Mexico 87 545 Morgan, W. C. Bat telle -Northwest Battelle Boulevard Richland, Washington 99352 Motz, Henry T. Los Alamos Scientific Lab. Box 1663 Los Alamos, New Mexico 87 544 Mughabghab, Said F. Brookhaven National Lab. Upton, Long Island New York 11973 Navalkar, M. P. Bhabha Atomic Research Centre Neutron Physics Section Trombay Bombay, Maharasmtra INDIA 400 085 Nelson, Charles E. Triangle Universities Nuclear Lab. Duke University Durham, North Carolina 27706 Nelson, Ronald 0. Los Alamos Scientific Lab. P.O. Box 1663 Los Alamos, New Mexico 87 544 Nilsson, Leif Tandem Accelerator Lab. August Sodermans vag 70 Uppsala, SWEDEN Otroshenko, Gennady Kurchatov Institute on the Atomic Energy Moscow, USSR Ozer, Odelli Electric Power Research Institute P. 0. Box 10412 Palo Alto, California 94303 Patrick, Bryan H. U. K, Atomic Energy Authority Harwell, Didcot Oxon. 0X11 ORA ENGLAND Paulsen, A. CEC, CBNM Euratom Steenweg naar Re tie B-2440 Geel BELGIUM Peelle, Robert W. Oak Ridge National Laboratory Building 6010 Oak Ridge, Tennessee 37830 Perey, F. G. Oak Ridge National Laboratory P. 0. Box X Oak Ridge, Tennessee 37830 Peters, Gerald J. Naval Surface Weapons Center White Oak Laboratory Code WR41, Building 2-013 Silver Spring, Maryland 20910 Phillips, Leigh F. Brookhaven National Laboratory Building 535 Upton, New York 11973 Poenitz, W. P. Argonne National Laboratory 9700 S. Cass Avenue Argonne, Illinois 60439 Pullen, David Dept. of Physics University of Lowell 1 University Avenue Lowell, Massachusetts 01854 Qazi, M. Nuruddin Pinstech, P. 0. Nilore Rawalpindi, PAKISTAN Quam, W. M. E.G.&G. P. 0. Box 98 Goleta, California 93017 Rahn, Frank J. Electric Power Research Inst. Box 10412 Palo Alto, California 94303 Renner, Cleide Oak Ridge National Laboratory P. 0. Box X Oak Ridge, Tennessee 37830 Riel, Gordon K. Naval Surface Weapons Center Nuclear Branch White Oak Silver Spring, Maryland 20910 Robertson, James Craig Dept. of Nuclear Engineering University of Michigan Ann Arbor, Michigan 48105 Rogers, J. W. E.G.&G. Idaho P. 0. Box 1625 Idaho Falls, Idaho 83401 Rogosa, George L. Division of Physical Research, J- 309 U. S. Energy Research and Development Administration Washington, D. C. 20545 Rohrig, Norman Brookhaven National Laboratory Building 902B Upton, New York 11973 Rosen, Mervine Naval Research Lab. Washington, D. C. 20370 Saltmarsh, M. J. Oak Ridge National Laboratory P. 0. Box X Oak Ridge, Tennessee 37830 Schell, Michael C. Mid-Atlantic Neutron Ther. 6104 Breezewood Ct. # 304 Greenbelt, Maryland 20770 Schrack, Roald A. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Schwartz, Robert B. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Sekharan, K. K. Cyclotron Institute Texas A&M University College Station, Texas 77840 Shapiro, Philip Naval Research Laboratory Code 6643 Washington, D. C. 20375 Shneiderov, Anatol The Polycultural Inst, of America 1673 Columbia Road, N. W. , Unit 309 Washington, D. C. 20009 Slater, William E. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Smith, Alan B. Argonne National Laboratory 9700 South Cass Avenue Argonne, Illinois 60439 Smith, Donald L. Argonne National Laboratory 9700 South Cass Avenue Argonne, Illinois 60439 362 Smith, J. Richard E.G.&G. Idaho, INEL Box 1625 Idaho Falls, Idaho 83401 Spiegel, Valentin Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Stewart, Leona Los Alamos Scientific Laboratory P. 0. Box 1663 Los Alamos, New Mexico 87545 Taschek, Richard F. Los Alamos Scientific Lab. P. 0. Box 1663 Los Alamos, New Mexico 87545 Theus, Richard B. Naval Research Lab. Washington, D. C. 20375 Twining, Bruce Division of Magnetic Fusion Energy U. S. Energy Research & Dev. Admin. Washington, D. C. 20545 Uttley, C. A. Atomic Energy Res. Est. Harwell Didcot, Oxon 0X11 0RA United Kingdom White, Roger M. Accelerator Laboratory Ohio University Athens, Ohio 45701 Yockey, Hubert P. Army Pulse Rad. Facility Aberdeen Proving Ground Aberdeen, Maryland 21005 Yolken, H. Thomas Nuclear Safeguards Meas., IBS National Bureau of Standards Washington, D. C. 20234 Young, Phillip G. Los Alamos Scientific Laboratory Los Alamos, New Mexico 87544 Zeitnitz, B. Universitat Bochum Postfach 102148 D-4630 BOCHUM 1 WEST GERMANY Zijp, W. L. Netherlands Energy Dev. Agency 3 Westerduinweg NL-PETTEN THE NETHERLANDS Vlasov, Mercury F. Nuclear Data Section International Atomic Energy Agency P. 0. Box 590 Vienna, AUSTRIA Wang, Joseph Naval Surface Weapons Center White Oak Laboratory Code WR-Y1 White Oak Silver Spring, Maryland 20910 Wasson, Oren A. Center for Radiation Research National Bureau of Standards Washington, D. C. 20234 Wattecamps, E. CEC, CBNM Euratom Steenweg naar Retie B-2440 Geel BELGIUM Weisbin, Charles R. Oak Ridge National Laboratory P. 0. Box X Oak Ridge, Tennessee 37830 Weston, Lawrence W. Oak Ridge National Laboratory Bldg. 6010, P. 0. Box X Oak Ridge, Tennessee 37830 Whetstone, Stanley L. U. S. Energy Research & Dev. Admin. MS J-309 Washington, D. C. 20545 363 AUTHOR INDEX Adamov, V. M. -- p. 313 Weisbin, C. R. -- p. 269 Alexandres, B. M. — p. 313 Weston, L. W. -- p. 43 Alkhazov, I. D. -- p. 313 Vitenko, V. A. - p. 194 Ambler, E. -- p. 1 Zl JP> w - L - ~ P- 128 Auxier, J. A. -- p. 101 Axton, E. J. -- p. 237 Barschall , H. H. — p. 342 Blinov, M. V. -- p. 194 Block, R. C. -- p. 255 Boldeman, J. W. -- p. 182 Brandenberger, J. D. -- p. 234 Broerse, J. J. -- p. 106 Carlson, A. D. -- p. 85 Caswell, R. S. -- p. 121 Chrien, R. E. -- p. 255 Cierjacks, S. -- p. 278 Coachman, J. S. — p. 61 Czirr, J. B. -- p. 54 Derrien, H. — p. 14 De Saussure, G. --p. 174 Drapchinsky, L. V. -- p. 313 Edvardson, L. -- p. 14 Eisenhauer, C. M. -- pp. 156, 198, 329 Fabry, A. -- pp. 290, 329 Farinelli , U. -- p. 310 Fomichev, A. V. -- p. 313 Gabbard, F. — p. 212 Gilliam, D. M. -- pp. 299, 335 Gold, R. — p. 137 Grundl , J. A. -- pp. 156, 329 Hale, G. M. — p. 30 Harvey, J. A. -- p. 10 Huynh, V. D. — p. 244 Jaffey, A. H. -- p. 206 James, G. D. -- p. 319 Joneja, 0. P. — p. 61 Kazi, A. H. -- p. 342 Knitter, H.-H. -- p. 3 Knoll, G. F. — p. 304 Kobayashi , K. -- p. 255 Kostochkin, 0. I. -- p. 313 Kovalenko, S. S. — p. 313 Kudriavzev, G. Yu. -- p. 313 Lachkar, J. C. -- p. 93 Lamaze, G. P. -- p. 37 Lernnel , H. D. -- p. 170 Liou, H. I. — p. 255 Liskien, H. -- p. 2 Maeck, W. J. — p. 146 Malkin, L. Z. — p. 313 Marston, T. V. — p. 137 McGarry, E. D. -- p. 342 Meier, M. -- p. 221 Navalkar, M. P. -- p. 61 Paulsen, A. -- p. 165 Peele, R. W. — pp. 174, 269 Petrzhak, K. A. -- p. 313 Phiske, M. R. -- p. 61 Pleskachevsky, L. A. -- p. 313 Poenitz, W. P. -- p. 261 Rahn, F. J. -- p. 137 Roberts, J. H. -- p. 137 Schroder, I. G. -- p. 10 Schwartz, R. B. -- p. 250 Sekharan, K. K. -- p. 234 Shapakov, V. I. -- p. 313 Singh, U. N. -- p. 255 Srikantaiah, R. V. -- p. 61 Stahlkopf, K. E. — p. 137 Stewart, L. -- p. 198 Touse, V. T. -- p. 194 Uttley, C. A. -- p. 47 wasson, 0. A. -- p. 115 Wattecamps, E. -- p. 67 364 ELEMENT QUANTITY S A TYPE ENERGY MIN MAX DOCUMENTATION LAB REF VOL PAGE DATE COMMENTS H 001 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 LI 006 B 010 B 010 B 010 B 010 B 010 B 010 B 010 C 012 012 012 012 012 013 013 016 NA 023 NA 023 NA 023 DIFF ELASTIC EVALUATION TOTAL XSECT TOTAL XSECT ELASTIC SCAT ELASTIC SCAT DIFF ELASTIC DIFF ELASTIC POLARIZATION POLARIZATION RES INT ABS N, TRITON N, TRITON N, TRITON N, TRITON N, TRITON RESON PARAMS TOTAL XSECT ELASTIC SCAT DIFF ELASTIC RES INT ABS N, ALPHA REAC N, ALPHA REAC N, ALPHA REAC TOTAL XSECT DIFF ELASTIC POLARIZATION N, GAMMA RESON PARAMS TOTAL XSECT N, GAMMA RESON PARAMS RES INT ABS N, GAMMA RESON PARAMS REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- REVW- EVAL- COMP- COMP- COMP- REVW- COMP- REVW- REVW- EVAL- EVAL- REVW- REVW- EVAL- REVW- REVW- EVAL- REVW- REVW- EVAL- CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF CONF 20+7 10-5 10 + 70 + 1 10 + 5 10 + 3 20 + 5 10 + 3 20+5 20 + 6 50-1 30 + 3 10 + 25-2 25-2 10 + 2 24 + 5 10 + 2 40+2 40+2 50-1 10 + 4 25-2 10+2 25-2 50 + 4 10 + 5 25-2 10 + 7 10 + 5 25-2 10 + 6 50-1 25-2 10 + 4 30 + 7 20 + 7 10 + 7 1 + 7 70 + 5 75 + 6 14 + 7 50 + 6 14 + 7 50 + 6 1 4 + 7 10 + 5 18 + 7 10 + 5 25 + 5 20 + 7 20 + 7 20 + 7 20 + 7 10 + 5 15 + 7 15 + 7 26 + 6 20 + 7 10 + 8 23 + 7 20 + 7 10 + 7 10 + 5 77NBS 47 477 HAR UTTLEY . GRPHS , TBLS . EXPT CFD Th CALCS. 77NBS 30 477 LAS H A LE+R-MATR1X ENDFB5 . GRPHS . CFD EXPTS 77NBS 3 477 GEL KNITTER. 3 MEAS CFD. TOT CFD SEL GRPH. 77NBS 10 477 SAC DER RIEN+EXPTS CFD . TbL , GRPHS . REF TBL. 77NBS 3 477 GEL KN 1TTER . I NTEG DIFF CFD TOT CS GRPH 77NBS 10 477 SAC DERRIEN+EXPTS CFD GRPh.REF TBL. 77NBS 3 477 GEL KNITTER. 4 MEAS CFD.ANG D1ST.LEG COFS 77NBS 10 477 SAC DERRIEN + ANG DISTR . NDG . TbL OF REFS. 77NBS 3 477 GEL KN ITTER . AN A L , POL POWER MEAS. CFD. NDG 77NBS 10 477 SAC DERR1EN+ANAL PWft.NDG.TbL OF KEFS. 77NBS 128 477 RCN ZIJP.TBL.1/E SPEC. FOR TOT HE PRODUCT 77NBS 3 477 GEL KNITTER. 1NV NDG . INTEG , DIFF CS CFD. 77NBS 10 477 ORL HARVEY + ANG A NISOTROP Y . DIFF GRPHS. 77NBS 10 477 SAC DERRIEN+1NV NDG. DIRECT GRPHS. REF TbL 77NBS 128 477 RCN Z UP . TBL . COMPARISON OF 2200M/S CS. 77NBS 174 477 ORL PEELLE+THR NORM TECH STUDY . TbLS , GRPH 77NBS 319 477 HAR JAMES. ALL FOR 250KEV RES CFD. TBL. 77NBS 67 477 GEL WATTEC AMPS . GRPHS , TBLS . TBL REFS GVN 77NBS 67 477 GEL W ATTECAMPS . GRPHS , TBLS . TBL REFS GVN 77NBS 67 477 GEL W ATTECAMPS . GRPHS , TBLS . TBL REFS GVN 77NBS 128 477 RCN ZIJP.TbL.I/E SPEC. FOR TOT HE PRODUCT 77NBS 67 477 GEL WATTEC AMPS . GND+ 1 ST EXC CS GRPHS, TBLS 77NBS 128 477 RCN Z UP . TbL . COMPARISON OF 2200M/S CS. 77NBS 174 477 ORL PEELLE+THR NORM TECH STUDY . TbLS , GRPH 77NBS 93 477 BRC LACHKA R . GRPHS , TBLS PHASE SHIFT ANAL. 77NBS 93 477 BRC L ACHK AR . G RPHS , TBLS PhASE SHIFT ANAL. 77NBS 93 477 BRC LACHKA R . TBLS . CC , PHASE SHIFT ANAL. 77NBS 93 477 BRC LACHKAR.THR VAL GVN.1NV STUDY ALSO. 77NBS 319 477 HAH JAMES. NARROW RES FOR E STANDARD GVN. 77NBS 93 477 BRC LACHKAR.CS CONTRIBUTION TO NATURAL C 77NBS 93 477 BRC LAChKAR.THR VAL GVN.1NV STUDY ALSO. 77NBS 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. 77NBS 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. 77NBS 128 477 RCN Z UP . TBL . COMPARISON OF 2200M/S CS. 77NBS 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. 365 ELEMENT QUANTITY S A TYPE ENERGY MIN MAX DOCUMENTATION LAB REF VOL PAGE DATE COMMENTS MG 024 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD AL 027 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. AL 027 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD AL 027 N, ALPHA REAC REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. AL 027 N, ALPHA REAC REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD AL 027 RESON PARAMS EVAL-CONF 10+0 10+4 77NBS 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. SI 028 RESON PARAMS EVAL-CONF 10+4 10+5 77NBS 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. P 031 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD S 032 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD S 032 RESON PARAMS EVAL-CONF 10+4 10+6 77NBS 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. SC 045 TOTAL XSECT EXPT-CONF 40+2 22+4 77NBS 255 477 RPI CHRIEN+LINAC . TRNS . GRPHS . CFD . SC 045 RES INT ABS REVW-CONF 50-1 77NBS 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. SC 045 N, GAMMA REVW-CONF 25-2 77NBS 128 477 RCN Z I JP . TBL . COMPARISON OF 2200M/S CS. SC 045 RESON PARAMS EXPT-CONF -5+3 21+4 77NBS 255 477 RPI CHRIEN+S WAVE WN,J,WG.P WAVE G»WN. TI NXN REACTION REVW-CONF FISS 77NBS 128 477 RCN Z I JP . Tl ( N , X ) 4 6SC . U235 FISS SPEC AVG. TI N, PROTON REVW-CONF FISS 77NBS 128 477 RCN Z I JP . TI ( N , X ) 4 6SC . U235 FISS SPEC AVG. TI N,N PROTON REVW-CONF FISS 77NBS 128 477 RCN Z I JP . TI ( N , X ) 4 6SC . U235 FISS SPEC AVG. TI 046 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD TI 047 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD TI 047 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD TI 048 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD TI 048 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC.CFD MN 055 N2N REACTION REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. MN 055 N2N REACTION REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 F ISS SPEC . CFD FE 054 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. FE 054 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD FE 054 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD FE 056 TOTAL XSECT EXPT-CONF 40+2 10+4 77NBS 255 477 RPI CHRIEN+L IN AC . TRNS . GRPHS . CFD . FE 056 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. FE 056 N, PROTON REVW-CONF FISS 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD FE 056 RESON PARAMS EVAL-CONF 10+4 10+6 77NBS 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. FE 058 RES INT ABS REVW-CONF 50-1 77NBS 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. FE 058 N, GAMMA REVW-CONF 25-2 77NBS 128 477 RCN Z I JP . TBL . COMPARISON OF 2200M/S CS. CO 059 RES INT ABS REVW-CONF 50-1 77NBS 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. CO 059 N, GAMMA REVW-CONF 25-2 77NBS 128 477 RCN Z I JP . TBL . COMP A RISON OF 2200M/S CS. 366 ELEMENT QUANTITY S A TYPE ENERGY DOCUMENTATION LAB MIN MAX REF VOL PAGE DATE COMMENTS CO 059 N2N REACTION REVW -CONF FISS 77NBS CO 059 N, ALPHA REAC REVW -CONF FISS 77NBS NI 058 N2N REACTION REVW -CONF FISS 77NBS NI 058 N, PROTON REVW -CONF FISS 77NBS NI 058 N, PROTON REVW -CONF FISS 77NBS CU 063 RES INT ABS REVW -CONF 50-1 77NBS CU 063 N, GAMMA REVW. -CONF 25-2 77NBS CU 063 N, GAMMA REVW- -CONF FISS 77NBS CU 063 N2N REACTION REVW. -CONF FISS 77NBS CU 063 N2N REACTION REVW- -CONF FISS 77NBS CU 063 N, ALPHA REAC REVW- -CONF FISS 77NBS ZN 064 N, PROTON REVW- -CONF FISS 77NBS ZR 090 N2N REACTION REVW- -CONF FISS 77NBS NB 093 N2N REACTION REVW- -CONF FISS 77NBS RH 103 TOT INELASTI REVW- -CONF FISS 77NBS RH 103 TOT INELASTI REVW- -CONF FISS 77NBS IN 115 TOT INELASTI REVW- ■CONF FISS 77NBS IN 115 TOT INELASTI REVW- -CONF FISS 77NBS IN 115 RES INT ABS REVW- -CONF 50-1 77NBS IN 115 N, GAMMA REVW- -CONF 25-2 77NBS IN 115 N, GAMMA REVW- -CONF FISS 77NBS IN 115 N, GAMMA REVW- -CONF FISS 77NBS I 127 N2N REACTION REVW- -CONF FISS 77NBS IR 191 RES.iN PARAMS EVAL- -CONF 10-1 10+0 77NBS AU 197 RES INT ABS REVW- -CONF 50-1 77NBS AU 197 N, GAMMA REVW- -CONF 25-2 77NBS AU 197 N, GAMMA REVW- ■CONF FISS 77NBS AU 197 N, GAMMA REVW- ■CONF FISS 77NBS AU 197 N, GAMMA REVW- ■CONF 10+5 14+7 77NBS PB 206 RESON PARAMS EVAL- ■CONF 10+0 10+5 77NBS TH 232 N, GAMMA REVW- •CONF 25-2 77NBS TH 232 RES INT CAPT REVW- •CONF 50-1 77NBS TH 232 FISSION REVW- ■CONF FISS 77NBS U 233 ABSORPTION EVAL- •CONF 25-2 77NBS U 233 N, GAMMA EVAL- -CONF 25-2 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC.CFD 128 477 RCN ZIJP.CS TbL.U235 FISS SPEC MDLS CFD. 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. 128 477 RCN Z I JP . TbL . COMP AR ISON OF 2200M/S CS. 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD. 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD. 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. 128 477 RCN Z I JP . TBL . COMP ARISON OF 2200 M/S CS. 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. 128 477 RCN Z UP . TBL . COMP ARISON OF 2200 M/S CS. 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 165 477 GEL PAULSEN . NORM . GRPhS . STANDARD STATUS. 319 477 HAR JAMES. NARROW RES FOR E STANDARD GVN. 128 477 RCN Z 1 JP . TBL . COMP ARISON OF 2200 M/S CS. 128 477 RCN ZIJP.TBL.AVG OVER 1/E SPECTRUM. 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 170 477 IAE LEMMEL. 2200M/S VAL VS MAXW VAL.TBL. 170 477 IAE LEMMEL. 2200M/S VAL VS MAXW VAL.TBL. 367 ELEMENT QUANTITY S A TYPE ENERGY DOCUMENTATION LAB MIN MAX REF VOL PAGE DATE COMMENTS u 233 FISSION EVAL-CONF 25-2 77NBS 170 477 u 233 FISSION EVAL-CONF 25-2 77NBS 170 477 u 233 FISSION EVAL-CONF 25-2 77NBS 182 477 u 233 FISSION EXPT-CONF NDG 77NBS 304 477 u 233 ALPHA EVAL-CONF 25-2 77NBS 170 477 u 233 ETA EVAL-CONF 25-2 77NBS 170 477 u 233 NUBAR, (NU) EVAL-CONF 25-2 77NBS 170 477 u 233 FISS YIELD REVW-CONF FAST 77NBS 146 477 u 233 FISS YIELD REVW-CONF FISS 77NBS 299 477 u 235 ABSORPTION EVAL-CONF 25-2 77NBS 170 477 u 235 N, GAMMA EVAL-CONF 25-2 77NBS 170 477 u 235 FISSION REVW-CONF 25-2 77NBS 128 477 u 235 FISSION REVW-CONF FISS 77NBS 126 477 u 235 FISSION REVW-CONF FISS 77NBS 128 477 u 235 FISSION REVW-CONF FISS 77NBS 156 477 u 235 FISSION REVW-CONF 25-2 45 + 77NBS 174 477 u 235 FISSION EVAL-CONF 25-2 77NBS 182 477 u 235 FISSION EVAL-CONF 10 + 5 20 + 7 77NBS 261 477 u 235 FISSION REVW-CONF FISS 77NBS 299 477 u 235 FISSION REVW-CONF FISS 77NBS 299 477 u 235 FISSION EXPT-CONF 14 + 5 96 + 5 77NBS 304 477 u 235 FISSION EXPT-CONF FISS 77NBS 313 477 u 235 FISSION EXPT-CONF 15+7 77NBS 313 477 u 235 RES INT FISS REVW-CONF 50-1 77NBS 128 477 u 235 RES INT FISS REVW-CONF 70 + 10 + 3 77NBS 174 477 u 235 ALPHA EVAL-CONF 25-2 77NBS 170 477 u 235 ETA EVAL-CONF 25-2 77NBS 170 477 u 235 NUBAR, (NU) EVAL-CONF 25-2 77NBS 170 477 u 235 SPECT FISS N REVW-CONF 10-3 10 + 7 77NBS 156 477 u 235 SPECT FISS N REVW-CONF 10 + 4 21+6 77NBS 198 477 u 235 FISS YIELD REVW-CONF 25-2 77NBS 146 477 u 235 FISS YIELD REVW-CONF FAST 77NBS 146 477 u 238 N, GAMMA REVW-CONF 25-2 77NBS 128 477 u 238 RES INT CAPT REVW-CONF 50-1 77NBS 128 477 u 238 FISSION REVW-CONF FISS 77NBS 128 477 IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. 1AE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. AUA BOLDEMAN.RE-EVAL.TBL. RECOMMENDED VAL MHG KNOLL. PHOTO N SOURCE TBD.NDG IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. INL MAECK.MASS SPEC YLDS MEAS TBD NBS GILLIAM. ILRR RESULTS . TBLS . IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. RCN ZIJP.TBL. COMPARISON OF 2200 M/S CS. RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD NBS GRUNDL+U235.CF252 FISS SPEC. TBLS. ORL PEELLE+THR NORM TECH STUDY . TBLS , GRPH AUA BOLDEMAN. RE-EVAL.TBL. RECOMMENDED VAL ANL POENITZ. TBL, GRPHS. MUST UPDATE ENDF5B NBS GILLIAM. CF252 SPEC AVG CS.TBL.CFD. NBS GILLIAM. REL PU239.DATA FOR 5 FIELDS MHG KNOLL. PHOTO N SOURCE . 4ES . TBL RI ADAM0V+CF252 SPEC . TBL . ABSL CS.CFD RI ADAMOV+ABSL CS.TBL.CFD OTH. RCN ZIJP.TBL. AVG OVER 1/E SPECTRUM. ORL PEELLE + 2E R ANGES . 7- 1 OE V , . 1 - 1 KE V . TBLS IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. NBS GRUNDL+DET SPEC RESPONSE . GRPHS . LAS STEWART+GRPH, TBLS. CFD WATT, MAXW SPEC INL MAECK.REL YLDS MEAS IN PROGRESS. TBC INL MAECK. GRPHS. MASS SPEC YLDS MEAS TBC RCN ZIJP.TBL. COMPARISON OF 2200 M/S CS. RCN ZIJP.TBL. AVG OVER 1/E SPECTRUM. RCN ZIJP.CS AVG OVER U235 FISS SPEC.CFD 368 ELEMENT QUANTITY S A TYPE ENERGY DOCUMENTATION LAB MIN MAX REF VOL PAGE DATE COMMENTS u 238 FISSION REVW- -CONF FISS 77NBS u 238 FISSION REVW- -CONF FISS 77NBS u 238 FISSION EXPT- -CONF 10 + 5 20 + 7 77NBS u 238 FISSION REVW- -CONF FISS 77NBS u 238 FISSION REVW- -CONF FISS 77NBS u 238 FISSION EXPT- -CONF FISS 77NBS u 238 FISSION EXPT- -CONF 15 + 7 77NBS u 238 SPECT FISS N REVW- -CONF + 6 60 + 6 77NBS u 238 FISS YIELD REVW- -CONF FAST 77NBS u 238 FISS YIELD REVW- -CONF FISS 77NBS u 238 RESON PARAMS EVAL- -CONF 10 + 10 + 4 77NBS NP 237 FISSION REVW- -CONF FISS 77NBS NP 237 FISSION REVW- -CONF 25-2 77NBS NP 237 FISSION REVW- -CONF FISS 77NBS NP 237 FISSION REVW- -CONF FISS 77NBS NP 237 FISSION REVW- -CONF FISS 77NBS NP 237 FISSION EXPT- -CONF 10 + 5 20 + 7 77NBS NP 237 FISSION REVW- -CONF FISS 77NBS NP 237 FISSION REVW- -CONF FISS 77NBS NP 237 FISSION EXPT- -CONF NDG 77NBS NP 237 FISSION EXPT- -CONF FISS 77NBS NP 237 FISSION EXPT- -CONF 15 + 7 77NBS NP 237 SPECT FISS N REVW- -CONF + 5 40 + 6 77NBS NP 237 FISS YIELD REVW- -CONF FAST 77NBS PU 239 ABSORPTION EVAL- -CONF 25-2 77NBS PU 239 N, GAMMA EVAL- -CONF 25-2 77NBS PU 239 FISSION REVW- -CONF 25-2 77NBS PU 239 FISSION REVW- -CONF FISS 77NBS PU 239 FISSION REVW- -CONF FISS 77NBS PU 239 FISSION REVW- -CONF FISS 77NBS PU 239 FISSION EVAL- -CONF 25-2 77NBS PU 239 FISSION EVAL- -CONF 25-2 77NBS PU 239 FISSION REVW- -CONF FISS 77NBS PU 239 FISSION EXPT- -CONF 1 4 + 5 96 + 5 77NBS PU 239 RES INT FISS REVW- -CONF 50-1 77NBS 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 156 477 NBS GR UNDL + U 2 35 , CF252 FISS SPEC.TBLS. 278 477 KFK C I ER JACKS . POSSIBLE ST ANDARD . DATA GVN 299 477 NBS G ILLI AM . CF252 SPEC AVG CS. TBL. CFD. 299 477 NBS GILLIAM. REL PU239.DATA FOR 5 FIELDS 313 477 RI ADAMOV+CF252 SPEC . TBL . ABSL CS.CFD 313 477 Rl ADAMOV+ABSL CS. TBL. CFD OTH. 156 477 NBS GRUNDL+DET SPEC RESPONSE . GRPHS . 146 477 INL MAECK.MASS SPEC YLDS MEAS TBD 299 477 NBS GILLIAM. ILRR RESULTS . TBLS . 319 477 HAft JAMES. NARROW RES FOR E STANDARD GVN. 128 477 RCN ZIJP.CS TBL.U235 FISS SPEC MDLS CFD. 128 477 RCN Z I JP . TBL . COMPARISON OF 2200 M/S CS. 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC. CFD 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 156 477 NBS GR UNDL+U 2 35 , CF252 FISS SPEC.TBLS. 278 4?7 KFK C IER JACKS . POSS IBLE ST ANDA RD . D ATA GVN 299 477 NBS G ILLI AM . CF252 SPEC AVG CS. TBL. CFD. 299 477 NBS GILLIAM. REL PU239-DATA FOR 5 FIELDS 304 477 MHG KNOLL. PHOTO N SOURCE TBD. NDG 313 477 RI ADAMOV+CF252 SPEC . TBL . ABSL CS.CFD 313 477 RI ADAMOV+ABSL CS. TBL. CFD OTH. 156 477 NBS GRUNDL+DET SPEC RESPONSE . GRPHS . 146 477 INL MAECK.MASS SPEC YLDS MEAS TBD 170 477 IAE LEMMEL.2200M/S VAL VS MAXW VAL.TBL. 170 477 IAE LEMMEL. 2200M/S VAL VS MAXW VAL.TBL. 128 477 RCN ZIJP. TBL. COMPARISON OF 2200 M/S CS. 128 477 RCN ZIJP.CS AVG OVER U235 FISS SPEC. CFD 128 477 RCN ZIJP.CS AVG OVER CF252 FISS SPEC CFD 156 477 NBS GRUNDL + U235 ,CF252 FISS SPEC.TBLS. 170 477 IAE LEMMEL. 2200M/S VAL VS MAXW VAL.TBL. 182 477 AUA BOLDEMAN. RE-EVAL. TBL. RECOMMENDED VAL 299 477 NBS GILLIAM. CF252 SPEC AVG CS. TBL. CFD. 304 477 MHG KNOLL. PHOTO N SOURCE . 4ES . TBL 128 477 RCN ZIJP. TBL. AVG OVER 1/E SPECTRUM. 369 ELEMENT QUANTITY S A TYPE ENERGY MIN MAX DOCUMENTATION LAB REF VOL PAGE DATE COMMENTS PU 239 ALPHA EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2200M/S VAL VS MAXW VAL.TBL. PU 239 ETA EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2200M/S VAL VS MAXW VAL.TBL. PU 239 NUBAR.(NU) EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2 200M/S VAL VS MAXW VAL.TBL. PU 239 SPECT FISS N REVW-CONF 10-3 30+6 77NBS 156 477 NBS GRUNDL+DET SPEC RESPONSE . GRPHS . PU 239 SPECT FISS N REVW-CONF 10+4 21+6 77NBS 198 477 LAS STEW ART+GRPH , TBLS . CFD WATT, MAXW SPEC PU 239 FISS YIELD REVW-CONF 25-2 77NBS 146 477 INL MAECK.REL YLDS MEAS IN PROGRESS. TBC PU 239 FISS YIELD REVW-CONF FAST 77NBS 146 477 INL MAECK . GRPHS . MASS SPEC YLDS MEAS TBC. PU 239 FISS YIELD REVW-CONF FISS 77NBS 299 477 NBS GILLIAM. ILRR RESULTS . TBLS . PU 240 FISS YIELD REVW-CONF FAST 77NBS 146 477 INL MAECK. MASS SPEC YLDS MEAS TBD. PU 241 ABSORPTION EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2200M/S VAL VS MAXW VAL.TBL. PU 241 N, GAMMA EVAL-CONF '25-2 77NBS 170 477 IAE LEMMEL . 2200M/S VAL VS MAXW VAL.TBL. PU 241 FISSION EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2200M/S VAL VS MAXW VAL.TBL. PU 241 FISSION EVAL-CONF 25-2 77NBS 182 477 AUA BOLDEMAN . RE-E VAL . TBL . RECOMMENDED VAL PU 241 ALPHA EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2200M/S VAL VS MAXW VAL.TBL. PU 241 ETA EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2 200M/S VAL VS MAXW VAL.TBL. PU 241 NUBAR.(NU) EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL . 2 200M/S VAL VS MAXW VAL.TBL. PU 241 FISS YIELD REVW-CONF FAST 77NBS 146 477 INL MAECK. MASS SPEC YLDS MEAS TBD. PU 242 FISS YIELD REVW-CONF FAST 77NBS 146 477 INL MAECK. MASS SPEC YLDS MEAS TBD. AM 241 FISS YIELD REVW-CONF FAST 77NBS 146 477 INL MAECK. MASS SPEC YLDS MEAS TBD. CF 252 NUBAR.(NU) EVAL-CONF 25-2 77NBS 170 477 IAE LEMMEL. VAL FOR 2200M/S DATA FIT. CF 252 NUBAR.(NU) EVAL-CONF SPON 77NBS 182 477 AUA BOLDEMAN . RE-E VAL . TBL . RECOMMENDED VAL CF 252 NUBAR.(NU) EXPT-CONF SPON 77NBS 194 477 RI BL1NOV+PRELIM VAL GVN.MEAS TBC. CF 252 SPECT FISS N EXPT-CONF SPON 77NBS 194 477 RI BLINOV + TOF.2 DETS . GRPHS . MAXW . TBC . CF 252 SPECT FISS N REVW-CONF SPON 77NBS 198 477 LAS STEW ART+GRPH . CFD WATT, MAXW SPEC. 370 NBS-114A (REV. 7-73) U.S. DEPT. OF COMM. BIBLIOGRAPHIC DATA SHEET 1. PUBLICATION OR REPORT NO. NBS SP-493 2. Gov't Accession No. 3. Recipient's Accession No. 4. TITLE AND SUBTITLE NEUTRON STANDARDS AND APPLICATIONS Proceedings of the International Specialists Symposium on Neutron Standards and Applications, Held at the National Bureau of Standards, Gaithersburg, MD, March 28-31, 1977 5. Publication Date October 1977 6. Performing Organization Code 7. AXTOCTO Editors: C. D. Bowman, A. D. Carlson, H. 0. Li ski en and L. Stewart 8. Performing Organ. Report No. 9. PERFORMING ORGANIZATION NAME AND ADDRESS NATIONAL BUREAU OF STANDARDS DEPARTMENT OF COMMERCE WASHINGTON, D.C. 20234 10. Project/Task/Work Unit No. 11. Contract/Grant No. 12. Sponsoring Organization Name and Complete Address (Street, City State, ZIP) National Bureau of Standards; Central Bureau of Nuclear Measurements; U.S. Energy Research & Development Adm. ; Electric Power Research Institute (U.S.A.); American Nuclear Society; American Physical Society; Nuclear Energy Agency (Europe); International Union of Pure & Applied Physics; with the cooperation of the International Atomic Energy Agency 13. Type of Report & Period Covered 14. Sponsoring Agency Code 15. SUPPLEMENTARY NOTES Library of Congress Catalog Card Number; 77-14317 16. ABSTRACT (A 2 00- word or less factual summary of most significant information. If document includes a significant bibliography or literature survey, mention it here.) These proceedings contain forty-seven papers, which were presented at the International Specialists Symposium on Neutron Standards and Applications held at the National Bureau of Standards on March 28-31, 1977. The topics addressed at the Symposium include light-element cross section standards, capture and fission cross section standards, integral neutron standards, flux measuring techniques, and medical and personnel dosimetry. 17. KEY WORDS (six to twelve entries; alphabetical order; capitalize only the first letter of the first key word unless a proper name; separated by semicolons) Cross section standards; dosimetry; fission; flux; measuring techniques; neutrons; standards. 18. AVAILABILITY [%] Unlimited 1 | For Official Distribution. Do Not Release to NTIS be I Order From Sup. of Doc, U.S. Government Printing Office Washington, D.C. 20402, SD Cat. No. C1^10;Z | .93 | ! Order From National Technical Information Service (NTIS) Springfield, Virginia 22151 19. SECURITY CLASS (THIS REPORT) UNCLASSIFIED 20. SECURITY CLASS (THIS PAGE) UNCLASSIFIED 21. NO. OF PAGES 379 22. Price $8.50 USCOMM.DC 29042-P74 it US. GOVERNMENT PRINTING OFFICE : 1977 O— 246-327