■MM n&i SPECIAL PUBLICATION *• U.S. DEPARTMENT OF COMMERCE / National Bureau of Standards Procedures Used at National Bureau of Sla fo Determine Selected Trac In Biological and Botanical Materials / V < CO DIGESTION mmmmm ' . / \ '<- g— ■ S O O £ <*; PURIFICATION ■■" • „ i ) V lam* \i«a< At o C/3. /O ■ V9a o a Procedures Used at the National Bureau of Standards To Determine Selected Trace Elements in Biological and Botanical Materials Edited by: R. Mavrodineanu Analytical Chemistry Division Institute for Materials Research National Bureau of Standards Washington, D.C. 20234 Sponsored by the U.S. Environmental Protection Agency and Partially supported by the U.S. Energy Research and Development Administration ^ ° F *o. 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 November 1977 Library of Congress Catalog Card Number: 77-608301 National Bureau of Standards Special Publication 492 Nat. Bur. Stand. (U.S.), Spec. Publ. 492, 295 pages (Nov. 1977) CODEN: XNBSAV U.S. GOVERNMENT PRINTING OFFICE WASHINGTON: 1977 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 (Order by SD Catalog No. C13.10:492). Stock No. 003-0O3-O1858-9. Price $7.50 (Add 25 percent additional for other than U.S. mailing). FOREWORD In the act of Congress which provides the basic authority of the National Bureau of Standards (NBS) , an essential func- tion of the Bureau is the "preparation and distribution of standard samples such as those used in chemical analysis . . . ' and "the development of methods of chemical analysis . . .". At the Bureau these two functions are closely joined in the Standard Reference Materials (SRM's) program. In this program the desired characteristics of a quantity of material, sufficiently large to make it available to a large number of users over a reasonably long time, is determined. By using these SRM's to calibrate instruments, to test the accuracy of newly developed methods, or in a routine program to monitor laboratory results, the quality of analytical data can be assured. More than nine hundred and fifty SRM's are available now from NBS for use in science and industry. These SRM's cover a large variety of applications and include inorganic and organic compounds certified for their chemical composition (major, minor, and/or trace elements), or for physical properties such as, for example, their optical, thermal, electrical, or radioactive characteristics. A significant number of these SRM's are certified through the measurements performed in the Analytical Chemistry Division of the NBS Institute for Materials Research. These measurements require that the systematic and random errors of the procedures used to determine the particular element or compound desired be sufficiently well known to state the concentration of that analyte on the certificate within a required uncertainty level. It is one of the major responsi- bilities of the Division to select suitable procedures from the literature or, if necessary, to develop new or modified procedures to insure that the necessary accuracy is obtained during the certification process. 111 This publication contains a detailed description of the procedures used at NBS to determine twenty-five trace elements in biological and botanical matrices. They were used to determine the trace element content of spinach, orchard leaves, pine needles, tomato leaves, bovine liver, blood serum, etc. The dissemination of these procedures, and their careful use, should permit other analysts to perform measurements with an accuracy comparable to those obtained at NBS. This volume was prepared at the request of the U. S. Environmental Protection Agency who also partially supported the development of the procedures and/or SRM's. The U. S. Energy Research and Development Administration also supported a portion of the work described in this report. This support is gratefully acknowledged. Philip D. LaFleur, Chief Analytical Chemistry Division July 1977 IV LIST OF CONTRIBUTORS J. R. Baldwin, D. A. Becker, R. W. Burke, B. I. Diamondstone, T. E. Gills, S. H. Harrison, G. J. Lutz, E. J. Maienthal, R. Mavrodineanu, L. T. McClendon, T. J. Murphy, P. J. Paulsen, T. C. Rains, H. L. Rook, and S. A. Wicks Analytical Chemistry Division Institute for Materials Research National Bureau of Standards Washington, D. C. 20234 TABLE OF CONTENTS PAGE ABSTRACT 1 INTRODUCTION by R. Mavrodineanu 2 SECTION I. Sample Preparation T. J. Murphy 5 SECTION II. NEUTRON ACTIVATION ANALYSIS A. Determination of Arsenic, Antimony and Copper in Biological and Botanical Materials Using Neutron Activation Analysis T. E. Gills 9 B. Determination of Platinum in . Biological and Botanical Materials Using Neutron Activation Analysis D. A. Becker and T. E. Gills 12 C. Determination of Vanadium and Manganese in Biological and Botanical Materials by Neutron Activation Analysis S. H. Harrison 15 D. Radiochemical Separation for the Determination of Mercury and Selenium in Biological and Botanical Materials Using Neutron Activation Analysis H. L. Rook, T. E. Gills and G. J. Lutz .... 20 E. Determination of Chromium in Biological and Botanical Materials Using Neutron Activation Analysis L. T. McClendon 23 F. Rapid, Quantitative Separation for the Determination of Selenium Using Neutron Activation Analysis H. L. Rook 26 SECTION III. SPARK SOURCE MASS SPECTROMETRIC ISOTOPE DILUTION Determination of Cadmium, Copper, Iron, Lead, Mercury, Molybdenum, Nickel, Selenium, Silver, Tellurium, Thallium, and Zinc P. J. Paulsen 33 via. PAGE SECTION IV. ATOMIC ABSORPTION AND FLAME EMISSION SPECTROMETRY A. Determination of Beryllium, Cadmium, Copper, Magnesium, Manganese, Mercury, Nickel, and Zinc by Atomic Absorption, and of Calcium, Potassium, Rubidium, and Sodium by Flame Emission Spectrometry, in Biological and Botanical Materials . T. C. Rains 49 B. Determination of Calcium, Potassium, and Sodium in Flour by Emission Flame Spectrometry J. R. Baldwin and R. Mavrodineanu 66 SECTION V. MOLECULAR ABSORPTION SPECTROMETRY Procedures for the Determination of Arsenic, Copper, and Nickel by Molecular Absorption Spectrometry R. W. Burke and B. I. Diamondstone 73 SECTION VI. FLUORESCENCE SPECTROMETRY Determination of Beryllium by Fluorescence Spectrometry S. A. Wicks and R. W. Burke 85 SECTION VII. POLAROGRAPHY Analysis of Botanical Standard Reference Materials by Cathode Ray Polarography E. June Maienthal 91 ACKNOWLEDGEMENT 106 APPENDIX 108 V-Lll LIST OF TABLES TABLE NO. PAGE SECTION II 1. Selenium contents of Standard Reference Materials, ppm 31 SECTION III 1. Trace elements (yg/g) in Coal, SRM 1632, by- spark source mass spectrometry isotope dilution 47 2. Trace elements (yg/g) in Beef Liver, SRM 1577, by spark source mass spectrometry isotope dilution 47 3. Trace elements (yg/g) in Spinach, SRM 1570, by spark source mass spectrometry isotope dilution 48 3b. Cd and Cu in spinach (Lot 6-2), additional sampling 48 SECTION IV 1. Determination of sodium, potassium, and calcium (yg/g) in a wheat flour sample on an "as received "basis 70 2. Determination of sodium, potassium, and calcium (yg/g) i n a rice flour sample on an "as received"basis 71 SECTION V 1. Spectrophotometric determination of arsenic in tissue samples by the arsine-silver diethyldithiocarbamate procedure 78 2. Spectrophotometric determination of copper in three new NBS botanical SRM's 82 SECTION VI 1. Beryllium content of three SRM's as determined by the acetylacetone extraction 2-hydroxy-3- naphthoic acid spectrofluorometric method. . . 89 IX TABLE NO. PAGE SECTION VII 1. Polarographic determination of iron (acid soluble) in orchard leaves 98 2. Polarographic determination of aluminum (acid- soluble) in orchard leaves 99 3. Polarographic determination of iron (total) in orchard leaves 100 4. Polarographic determination of aluminum (total) in orchard leaves 101 5. Polarographic determination of bismuth and lead in orchard leaves 102 6. Polarographic determination of nickel in orchard leaves 103 LIST OF FIGURES FIGURE NO. SECTION II 1. Sample Combustion Tube and Mercury Trap 27 PROCEDURES USED AT THE NATIONAL BUREAU OF STANDARDS TO DETERMINE SELECTED TRACE ELEMENTS IN BIOLOGICAL AND BOTANICAL MATERIALS R . Mavrodineanu, Editor Abstract This volume consists of thirteen papers describing the analytical procedures selected at the National Bureau of Standards (NBS) for the determination of Ag, Al , As, Be, Bi, Ca, Cd, Cr, Cu, Fe, Hg, K, Mg, Mn, Mo, Na, Ni, Pb , Pt , Sb , Se, Te, Tl, V, and Zn in biological and botanical materials. These procedures are used at the present time for the certi- fication of various substances issued by NBS as Standard Reference Materials, and belong to the following analytical disciplines: sample preparation, neutron activation analysis, spark source mass spectrometric isotope dilution, atomic absorption and flame emission spectrometry, molecular absorp- tion spectrometry, fluorescence spectrometry, and polarography Further details on the analytical methods including sample preparation, purity of reagents, and problems associated with blanks are given in sixteen additional papers which are reproduced in the Appendix to this volume. Key Words: Analytical disciplines; analytical procedures; atomic absorption spectrometry; biological materials: blanks; botanical materials; chemical species; flame emission spectrometry; fluorescence spectrometry; molecular absorption spectrometry; neutron activation analysis; polarography; pure reagents; sample preparation; spark source mass spectrometric isotope dilution; tissues. INTRODUCTION This work consists of a collection of analytical proce- dures used in the Analytical Chemistry Division, Institute for Materials Research, of the National Bureau of Standards (NBS) , for the determination of trace levels of Ag, Al , As, Be, Bi, Ca, Cd, Cr, Cu, Fe, Hg , K, Mg, Mn, Mo, Na, Ni, Pb , Pt, Sb , Se, Te, Tl, V, and Zn in biological and botanical materials. These procedures were critically selected or adapted, and often specially developed, by the scientific staff members of the Analytical Chemistry Division to provide measurements with the best obtainable accuracy. They were considered to be most appropriate for the analysis and certi- fication of various Standard Reference Materials (SRM's) issued by NBS such as SRM 1570, Trace Elements in Spinach; SRM 1571, Orchard Leaves; and SRM 1577, Bovine Liver. The description of these procedures is given here with sufficient detail to permit the analyst to use them as a protocol for routine analyses in his laboratory. They are assembled according to the analytical disci- plines involved in the measuring process: I- Sample Preparation II- Neutron Activation Analysis III- Spark Source Mass Spectrometric Isotope Dilution IV- Atomic Absorption and Flame Emission Spectrometry V- Molecular Absorption Spectrometry VI- Fluorescence Spectrometry VII- Polarography As a general rule, and unless otherwise indicated, all the acids and distilled water used in the measurements are of high-purity and were prepared at NBS. A detailed description of the methods and instrumentation used to produce these reagents, including the analytical techniques used to test their purity, is given in a paper reproduced as Appendix 1. The important subject of blanks, a determining factor common to all analytical measurements, is examined in Appen- dix 2, which also includes a detailed description of the environmental conditions necessary to control this essential factor, and to insure a maximum protection against contami- nation. An evaluation of the lyophilization method for precon- centration and drying of various analytical samples is presented in Appendix 3. Also included in the Appendix are reprints of thirteen papers which provide further relevant and more detailed information on the analytical procedures included in this publication. In order to describe adequately materials and experi- mental procedures discussed in this work, it was occasionally necessary to identify commercial products by manufacturer's name or label. In no instance does such identification imply endorsement by NBS or that the particular product or equip- ment are necessarily the best available for that purpose. The analytical procedures assembled in this publication are believed to provide the best measuring capabilities available at this time. They are, however, continuously being revised, improved, or replaced by more accurate ones. Hence, it is recommended that the analyst who desires to be informed of these advancements maintain a contact with the scientific staff of the Analytical Chemistry Division at NBS. January, 1977 R. Mavrodineanu SECTION I Sample Preparation T. J. Murphy 1. Drying All analytical results for the tissue Standard Reference Materials should be expressed on a dry weight basis. For non-volatile elements, drying may be performed on the sample to be analyzed, but for volatile elements, such as mercury and selenium, drying should be done on separate samples other than the samples for analysis. The only animal tissue certified to date as a Standard Reference Material is SRM 1577, Bovine Liver. This material was lyophilized during preparation, so freeze-drying is the recommended drying method. * The material should be dried by lyophilization for at least 24 hours using a cold trap at or below-50 °C and at a pressure not greater than 30 Pa (0.2 mm Hg) . Even though this material has been previously lyophi- lized, drying should be done on separate samples for volatile elements such as mercury and selenium. To date, we have worked on four botanical SRM's, spinach, pine needles, tomato leaves, and orchard leaves. All contain varying amounts of water. The removal of this water is not a simple problem since additional volatiles can also be lost. Desiccator drying over P 2 5 or Mg(C10i f ) 2 is not satisfac- tory as the rate of moisture loss is too slow. Oven drying at 105 °C and vacuum oven drying at 75 °C are also unsatis- factory since both methods cause large losses of volatiles other than water. Oven drying for two hours at 75 °C, vacuum drying for 24 hours at room temperature, and freeze-drying (lyophiliza- 1 An evaluation of the lyophilization procedure is given in Appendix 3. tion) for 24 hours are all acceptable methods. The loss of drying by these different methods showed that all three methods agreed to about 0.2 percent. For volatile elements, drying should be performed on separate samples and results corrected to a dry basis. In general, the samples were dried in the following manner: a suitable sample (usually about 2 g) was placed in a weighing bottle (with cover) . The bottle and contents were weighed and transferred uncovered to an oven, vacuum oven or freeze-drier as desired. After leaving the samples for the desired time under the desired conditions, the bottles and contents were placed in a desiccator and covered. After equilibration, the weighing bottles and contents were again weighed and the moisture loss calculated. For the non- volatile elements the samples were used directly for analyses. For the volatile elements (Hg, Se, As, etc.) separate samples were taken for analysis and corrected to dry weight from the calculated moisture loss. 2. Wet Ashing The basic wet ashing system employed in the decomposition of tissue samples involved the use of nitric-hydrofluoric- perchloric acids. The acids and water used in these dissolu- tions were prepared by sub-boiling distillation. The preparation and analyses of these acids and water are described in a paper included as Appendix 1. Covered beaker chemistry was used for all elements except the volatile elements such as Hg, As, Se, and Cr. For the non-volatile elements such as Pb, Cu, Ni, and Cd the following procedure was used. A suitable sample (usually 1 g) was transferred to a Teflon FEP beaker and treated with 5 ml nitric acid, 1 ml hydrofluoric acid and 2 ml perchloric acid. The beaker was covered with a Teflon cover and placed on a hot plate at low heat. After the action of nitric acid slowed, the temperature was raised to about 110 °C and the beaker and contents were heated overnight. The cover was then removed and washed with water. The solution was heated to fumes of perchloric acid and the heating was continued until fumes of perchloric acid were no longer noted. After removing the beaker from the hot plate and allow- ing it to cool, the sides of the beaker were washed with water and the resulting solution was again heated until fumes of perchloric acid were no longer noted. For the analysis of samples where perchloric acid could not be used (Cr, for example), sulfuric acid was substituted and quartz instead of Teflon beakers were used. A suitable sample (usually 1 g) was transferred to a quartz beaker and treated with 5 ml of H 2 and 10 ml HN0 3 . The beaker was covered with a quartz watch glass and the solution was digested overnight on a hot plate. Five ml of sulfuric acid were added and the solution was evaporated to fumes of sulfuric acid. The beaker was then covered and nitric acid was added dropwise to the hot solution to destroy the residual organic matter. The solution was then transferred to a Teflon FEP beaker and any remaining siliceous material was washed into the Teflon beaker. To remove silica, 1 ml of hydrofluoric acid was added and the solution was evaporated to fumes of sulfuric acid. For the volatile elements, dissolution was effected under reflux conditions. The system consisted of a 125 ml quartz flask fitted with a quartz condenser having a 24/40 standard taper joint. The condenser was filled with quartz beads. A suitable sample (usually about 1 g) was transferred to the flask and the sides washed with a few ml of water. The condenser was then attached and cold water (-vlO °C) was circulated through the condenser jacket. Ten grams of HNO3 were then added through the top of the condenser and the solution was allowed to digest for about four hours while 7 the temperature was gradually increased. Nine grains of sulfuric acid (or perchloric acid) were then slowly added through the top of the condenser and the temperature was increased. The water was drained from the condenser jacket and the solution was allowed to concentrate and reflux overnight. This treatment results in a clear solution, except for undis- solved siliceous matter. When possible, this was dissolved by transferring the solution to a Teflon beaker and adding a small amount of hydrofluoric acid. All operations involving sample preparation were per- formed in Class 100 environmental conditions to reduce environmental contamination. Whenever possible all analyt- ical operations were carried-out in the same controlled conditions. A thorough discussion of these conditions, in- dispensable to minimize sample contamination, and to insure the production of accurate measurements, will be found in Appendix 2. NOTE: The symbol for milliliters used throughout this manuscript was ml instead of mL. SECTION II NEUTRON ACTIVATION ANALYSIS Determination of Arsenic, Antimony, Chromium, Copper, Manganese, Mercury, Platinum, Selenium, and Vanadium A. Determination of Arsenic, Antimony and Copper in Biological and Botanical Materials Using Neutron Activation Analysis T. E. Gills 1. Objective The object is to provide an effective procedure for isolating and determining arsenic, antimony and copper using neutron activation analysis, in biological and botanical materials . 2. Principle This procedure involved the neutron irradiation of a biological or botanical sample and subsequent solubilization of the radioactive material by nitric and perchloric acids. After obtaining the irradiated material in a soluble form, the radioactive solutions were made to 6M HCld* and passed through 6M HClOi* equilibrated inorganic ion exchange columns of cuprous chloride and tin dioxide. Copper and antimony were absorbed on the cuprous chloride while arsenic was absorbed on the tin dioxide. Sodium, the major interference in the analysis of biological materials, was eliminated by passing through to the eluate fraction. 3. Instrumentation Pneumatic tube RT-3 of the NBS Reactor (NBSR) and a Ge(Li) detector system coupled to a multichannel analyzer were used. For further details see Appendices 4 and 5. 9 4. Reagents Cu standard: 100 mg Cu metal (99.99% pure) dissolved in high purity HN0 3 and diluted to 100 ml with high purity H 2 0. As standard: 150 mg As 2 3 (99.95% pure) dissolved in high purity HC1 and diluted to 100 ml with high purity H 2 0. Sb standard: 278 mg antimony potassium tartrate (reagent grade) dissolved in a minimum amount of high purity HC1-HN0 3 and diluted to 100 ml with high purity H 2 0. Carriers: Cu, As, Sb - 1.0 mg/ml (reagent grade oxides and chlorides) . Na carrier: 1 mg/ml (reagent grade chloride). Acids: HClOi,, HN0 3 , HF (ACS reagent grade). Inorganic Ion Exchangers: Cuprous chloride, tin dioxide. 2 NBS SRM's: Orchard Leaves (1571), Bovine Liver (1577). 5. Sample Preparation and Analysis Sample sizes of approximately 250 mg are placed in clean polyethylene microcentrifuge tubes and heat sealed. Standards consisting of solutions of copper, arsenic, antimony and ^250 mg of SRM 1571 Orchard Leaves are treated in the same manner. Standards and samples are irradiated for 30 minutes in the pneumatic tube RT-3 of the NBSR in a flux of ^6xl0 13 n*cm 2 s l . After 24 hours decay, samples and standards are post-weighed and/or pipetted into 125 ml boiling flasks which have been fitted with air condensers. The irradiated materials are then dissolved in 1 : 3 mixtures of HClOif- HN0 3 . 2 These inorganic ion exchangers were obtained from Carlo Erba, Division of Industrial Chemicals. Via C. Imbonati, 24, Milano, 20159, Italy. 10 If silica containing material is present, 3 to 4 drops of HF may be added. After dissolution, 10 ml of 6M HCIO^ is added and the solution is passed through 3-cm columns of cuprous chloride and tin dioxide respectively. Copper and antimony are absorbed on the cuprous chloride column while arsenic is absorbed on the tin dioxide column. After the columns are washed, the contents are transferred separately into 45 mm petri dishes and counted. 6. Procedure Weigh the irradiated samples into 125 ml boiling flasks. Add 2 ml of Cu, As, Sb and Na carriers. Add 15 ml of the HClOi^-HNOs mixture (1:3) and heat until the sample is dis- solved. Continue heating until the volume is reduced to ^1 ml. Allow sample to cool. Place each inorganic exchanger in a polyethylene exchange column; a height of 3 cm is sufficient for sample sizes of 1 gram. Mount the columns in series with the cuprous chloride column at the top. Wash each column three times with 5 ml of 6M HC1CU . Discard wash. Add 5 ml of 6M HClOi* to the cooled sample and transfer to the cuprous chloride column. Allow the sample solution to pass through both columns. Wash both columns three times with 5 ml portions of 6M HClOit . Discard all washings. Transfer separately the cuprous chloride and tin dioxide columns into 45 mm petri dishes and count each fraction. Remarks The use of the inorganic ion exchangers as outlined in the above procedure provides a rapid and efficient separa- tion for trace amounts of copper, arsenic and antimony from the matrix radioactivities. The technique also allows the analyst to process a series of samples simultaneously. 11 Calculation If the induced activities are measured in an identical manner, comparing counting rates in the sample and standard will give the weight of the desired element in the sample by the following simple relation: Total cts/min Standard _ wt (yg) Standard Total cts/min Sample wt (yg) Sample Example : - -I _ (Total cts/min Sample) (wt (ug) Standard) r " ! .-, Total cts/min Standard (yg element) c i (SxlO^cpm) (10 yg) nn Sample = ^ - t -^ J - = 20 yg (yg element) (2 . SxlO^cpm) B. Determination of Platinum in Biological and Botanical Materials Using Neutron Activation Analysis. D. A. Becker and T. E. Gills 1. Objective To provide a simple and selective separation procedure for determining trace quantities of platinum in tissues, sediments and waters by neutron activation analysis (NAA) . 2. Principle The most sensitive and useful nuclear reactions for quan- titation of platinum involves isotopes of low gamma energies which are difficult to detect in an activated tissue, sediment or sometimes aqueous matrix. This procedure is based on the electro-chemical theory of spontaneous deposition of an ele- ment onto a less noble element. The nuclear reaction used is 198 Pt(n,y) 199 Pt£ 199 Au. The end product of the reaction, 12 199 Au, has a 3.15 day half -life and y _ra y energies of 158 keV (37 percent abundant) and 208 keV (8 percent abundant) which are more suitable parameters for the detection. 3. Instrumentation Pneumatic tube RT-3 of the NBS Reactor, large volume Ge(Li) detector coupled to a multi-channel analyzer (see Appendices 4 and 5) . 4. Reagents Pt standard: NBS SRM 680 - 1 mg is dissolved in a mini- mum volume of high purity aqua regia (hot) Aqua regia: l+4+l-HN0 3 :HC1 :H 2 Pt carrier: 5 mg Pt/ml Na carrier: 1 mg Na/ml Ca carrier: 1 mg Ca/ml Silver: metal powder 5. Sample Preparation and Analysis Approximately 250 mg of the sample is weighed into a clean polyethylene micro vial and heat sealed. The platinum standard solution is treated in the same manner. Sample and standard are placed in an irradiation "rabbit" and irradiated for one hour in the NBS Reactor in a thermal neutron flux of ^6xl0 13 n»cm 2 »s 1 . After a 24 hour decay period to reduce short half-life matrix radioactivities, the samples are opened and transferred to 100 ml beakers containing carriers of elements of interest, and dissolved in aqua regia. After dissolution and evaporation to 2 ml the samples are washed into a 50 ml centrifuge tube with 24 ml of H2O which contains finely divided silver powder. The mixture is shaken to allow for spontaneous reduction of the platinum. The mix- ture is filtered through a glass fiber-filter pad to retain the silver metal powder upon which the platinum had been 13 deposited. The filter pad is mounted and counted in a stan- dard geometry. The platinum standard solution is treated in the same manner as the sample. 6. Procedure Transfer each irradiated material to a 100 ml beaker. Add 1 ml of Pt, Na, and Ca carriers. Add 10 ml of aqua regia and heat until the sample has dissolved and the volume has been reduced to 2 ml. Let the sample cool. Place 50 mg of finely divided silver powder into a 50 ml round bottom centrifuge tube (with polyethylene-lined cap) and add 24 ml of H 2 0. Transfer the sample solution to the centrifuge tube and shake for 10 minutes. Filter the mixture through a 25 mm glass -fiber filter pad. The silver metal powder containing the platinum is retained on the filter pad. Discard the solution. Place the filter pad in a 45 mm petri dish and count on a Ge(Li) detection system. Remarks This procedure is rapid, quantitative and selective for platinum. The sodium and calcium carriers are needed as hold-back carriers for the decontamination of the Pt from the matrix. Both photopeaks of * 9 9 Au (158 keV and 208 keV) can be used individually for quantitation. Calculation If the induced activities are measured in an identical manner, comparing counting rates in the sample and standard will give the weight of the desired element in the sample by the following simple relation: 14 Total cts/min Standard _ wt (yg) Standard Total cts/min Sample ' wt (yg) Sample 2.5x10" cpm m 1 yg 5x10** cpm X = 2 yg (yg element) C. Determination of Vanadium and Manganese in Biological and Botanical Materials by Neutron Activation Analysis. S. H. Harrison 1. Objective Determination of vanadium and manganese at yg/g levels in biological and botanical materials using Instrumental Neutron Activation Analysis (INAA) . 2. Principle Instrumental Neutron Activation Analysis is applied to samples for vanadium and manganese determinations utilizing the thermal neutron reactions 51 V(n,y) 52 V and 5 5 Mn(n, y) 5 6 Mn. Measurement of the radiation emitted and comparison to a known standard made radioactive under the same conditions enable quantitative determination of the vanadium and manganese in the sample. 3. Instrumentation For instrumental analysis of complex matrices such as biological or botanical materials, a high resolution Ge(Li) semiconductor detector, preferably a resolution less than 2.2 keV for the 1332 keV gamma of 60 Co is used in conjunction with a multichannel analyzer of at least 2048 channels. 15 4. Reagents The standard solutions are made of high purity metals or oxides dissolved in purified reagents (see Appendix 1) . Distilled Water Nitric acid, diluted 50 percent for dissolution of standards (V) . Acetic acid, reagent grade, diluted 50 percent for dissolution of standards (Mn) . 5. Sample Preparation and Analysis For this INAA procedure, careful sample preparation is important. Precleaned plastic containers and instruments are used whenever possible for sample handling, but nickel spatulas and stainless steel knives are acceptable. A sample size of about 300 mg of dried tissue is ade- quate for analyses. For irradiation, the sample is pelle- tized to provide a small uniform sample. This pellet is then sealed in a small precleaned polyethylene bag, and the sample is sealed in a second bag for double encapsulation. Primary standards for vanadium and manganese analyses are made by dissolving pure metals in a suitable acid medium. Solution standards are prepared to approximate the tissue sample sizes. Tissues are irradiated for about five minutes in a flux of lxl0 13 n»cm 2 s l . After irradiation, the tissue samples are transferred to new containers for counting. After decay times of from 2 to 12 minutes the samples are counted for five minutes live time ( e.g. for a matrix such as sediment it may be necessary to wait longer for some of the 28 A1 activity to decay). Since 52 V decays with a half-life of 3.75 minutes, it is helpful if the total activity and decay rate of the standard approximate that of the sample. 16 Manganese can be determined in the same count as vanadium, however, a lower uncertainty may be obtained if a second count is taken after short-lived isotopes have decayed away. Thirty minutes to two hours is a reasonable decay time for this second count. 6. Procedure The sample- standard comparator technique is used for quantitative determination of vanadium and manganese. The predominante y _ra y emitted by 52 V has an energy of 1434 keV. There are no commonly observed interferences at this energy. The predominante y -ra y for 56 Mn has an energy of 847 keV. The half -life of 56 Mn is 154.6 minutes. There is. a commonly observed y-ray interference with the 847 keV peak. This interference is caused by the 843 keV y - ray of 27 Mg, which has a half -life of 9.46 minutes. Two choices are available for avoiding this interference: The second most prominent Y (1811 keV) of 56 Mn may be used for quantification or one can wait until the 9.46-minute 27 Mg has decayed. In using the second alternative, it would be impossible to determine V and Mn in the same count. 7. Calculation of V in Tissue Vanadium was chosen for the calculation example because of the special problems in measuring short-lived isotopes. The commonly used activation analysis equation is: A = A o e" At (1) where X = the decay constant, t = the decay time, A » the activity of the isotope at that decay time and A Q ■ the activity of the isotope when t = o. 17 When using this equation, A is assumed to be equal to A, the average activity of the isotope during the live time counting interval, At». This is a valid assumption only At when TZ < 0.5 (Tj is the half-life of the isotope of interest) In the case of vanadium, At» = 5 min. Tj = 3.75 min, and At I rp — = 1.33 not less than 0.5. Therefore, for the calculation of vanadium, it is necessary to use the following equation: XAAt. A = o -Xti -Xt 2 e -e (2) where ti t 2 At the decay time at the start of the count, the decay time at the end of the count and the clock time counting interval. Example Sample wt Standard wt 297. 7 mg sample 0.4022 mg V/g soln x 0.1000 g soln = 0.04022 mgV A (sample) ti (sample) t 2 (sample) At (sample) A (sample) /mg sediment 1965 cts/min 12.00 min 17.80 min 5.80 min = clock time counting interval , 6.601 cts/mg sediment min A (standard) ti (standard) = *2 (standard) = At (standard) c A (standard) /mg V = 12,867 cts/min 12.00 min 17 . 77 min 5.77 min = clock time counting interval . 3.199xl0 5 cts/mg V min 18 T, C 52 V) = 3.75 min At > 0.5, therefore equation 2 is used to calculate A T ( 52 V) -2 XAAt c , .69 3 A = where a = ■= — ^-^ « — o , . , . 3./5 mm e -Xt!_ e -At 2 After substituting the time and activity parameters in equation 2 for the sample and then the standard, the fol flowing values for A are obtained: A (sample) /mg sample = 98.82 A Q (standard) /mg V = 4.777xlO G A (sample) 98.82 cts/mg sed. A Q (standard) 4.777xl0 6 cts/mg V sample or cone. = 20.68 ug V/g sample 2. 068x10" 5 mg V/mg 19 D. Radiochemical Separation for the Determination of Mercury and SElenium in Biological and Botanical Materials Using Neutron Activation Analysis. H. L. Rook, T. E. Gills and G. J. Lutz 1. Objective The objective is to determine mercury and selenium at concentration levels as low as 10 ppb using neutron activa- tion coupled to a selective radiochemical separation procedure. 2. Principle The principle of the procedure is the irradiation of an unknown sample along with standards in an intense thermal neutron flux for a period of several hours inducing the reaction 1 9 6 Hg(n, y) x 9 7 Hg and 7 ^Se (n,y) 7 5 Se. The mercury and selenium activities are subsequently separated from the matrix radioactivities by combusting the sample in a stream of oxygen and trapping the volatile elements in a liquid nitrogen cooled trap. 3. Instrumentation Apparatus and Reagents Instrumentation and apparatus required include: a tube combustion furnace, a quartz tube combustion assembly, a high resolution gamma-ray detector, a pulse height analyzer, precipitation and filtering equipment. For further details see Appendix 4 and 5. Reagents required are tank oxygen, mercury oxide, con- centrated nitric acid, ammonium hydroxide, liquid nitrogen, and thioacetamide. The chemicals are of reagent grade. 4. Procedure The samples are weighed and encapsulated in quartz tubes for irradiation. If expected levels of mercury are very low, encapsulation is done in a Class 100 clean 20 room. 3 The quartz encapsulated samples are subsequently packaged along with encapsulations of mercury solution stan- dards in polyethylene rabbits and irradiated in the NBS Reactor in a flux of 6xl0 13 n/cm 2 s for approximately six hours. After irradiation, the individual quartz ampoules are rinsed with cone. HN0 3 and water. The samples are then immersed in liquid nitrogen for a few minutes and opened by scoring the vial with a file and pressing at the scratch mark. The samples are then transferred along with approximately ten milligrams of HgO carrier into porcelain combustion boats (1x10 cm) . The combustion boat is then inserted into the combustion assembly. An oxygen stream of 20 cm 3 /min is allowed to flow over the sample. If the sample is a biologi- cal material, it is allowed to burn freely after igniting with a torch. After combustion, the tube is heated to about 800 °C for 20 minutes. Portions of the tube not covered by the furnace are flamed down with a gas oxygen torch to further distill mercury into a liquid nitrogen cooled con- denser. The assembly is then allowed to cool and the distilled mercury in the condenser is transferred with nitric acid and water into a 4-oz polyethylene bottle. The solution con- taining the mercury and selenium is diluted to a selected standard volume and shaken to homogenize the solution. Standards can be carried through the procedure or an aliquot can be transferred to a 4-oz bottle and diluted. All samples and standard solutions are counted on a Ge(Li) detector in conjunction with a multichannel pulse height analyzer. The 68 and 77 keV peaks of * 9 7 Hg and the 265 keV line of 75 Se are used in the evaluation. An improvement in sensitivity can be achieved by reduc- ing the volume of the counting sample by precipitating the mercury as a sulfide. Selenium coprecipitates with the HgS . See Appendix 2. 21 Bromine- 82 can be removed during the combustion step with a heated silver wool trap. Silver wool also removes selenium so the procedure is applicable when only mercury is to be determined. Bromine-82 can also be removed during the precipitation step by the addition of hold-back carrier. 5. Calculation The concentration of mercury and selenium is calculated using the standard-comparator method: Total cts/min Standard Wt (yg) Standard Total cts/min Sample Wt (yg) Sample Example Standard: Counts per min » 9068 Wt (yg) » 1.53 Sample: Counts per min = 1831 Wt (yg) = X 9068 cts/min m 1.53 1831 cts/min X X = .31 yg Note: For further information see Appendix 6 22 E. Determination of Chromium in Biological and Botanical Materials Using Neutron Activation Analysis. L. T. McClendon 1. Objective The objective is to provide a simple and accurate pro- cedure for determining chromium in biological and botanical materials using neutron activation analysis. 2. Principle This procedure is designed for a rapid, selective deter- mination of trace quantities of chromium in tissues, by separating it from the matrix activity produced when these materials have been irradiated in a nuclear reactor. Experi- ments have shown that a volatile form of chromium is present in some tissues so that dissolution of the sample in a closed system and entrapment of the vapor fumes in a HNO3 or HC1 solution is recommended. 3. Instrumentation Pneumatic tube RT-3 or RT-4 of the NBS Reactor, Ge(Li) detector system or Nal detector system. Further details will be found in Appendices 4 and 5. 4. Reagents Cr standard: 0.5000 g Cr metal (99.9% pure) dis- solved in high purity HC1 and diluted to 1000 ml with high purity H 2 0. Cr carrier: 1 mg Cr/ml, (CrCl 3 '10 H 2 reagent grade) dissolved in HC1, 1.5M. H 2 0: High Purity HN0 3 : Cone, 6M HC10-: 70-72% Ce 2 (SCM3: 101 (w/v) in 3M H2SO4 23 Tribenzylamine: \% (w/v) in CHC1 3 NBS SRM: Orchard Leaves (1571). 5. Sample Preparation and Analysis Approximately 250 mg of the sample is placed in a clean quartz vial or polyethylene tubing and sealed. The chromium standard solution and Orchard Leaves O250 mg) are treated similarly. Samples and standards are placed in an irradi- ation "rabbit" and irradiated for an appropriate period of time (1 to 6 h) in RT-3 or RT-4 of the NBS Reactor. After a sufficient decay period is allowed to minimize some of the matrix activity, the sample container is washed with 1:1 HNO3 to remove any exterior contamination. If Cr is >2 yg, the Cr concentration can be determined instrumentally without chemical treatment. If the concentration is below 1 ppm a radiochemical separation may be needed. Thus the sample is transferred to a 50-ml Erlenmeyer flask for "open" or "closed" flask dissolution. The chromium standard solution and com- parative standards (Orchard Leaves, Coal, etc.) are treated in the same manner as samples. 6. Procedures a. Instrumental Analysis If the sample and standards were preweighed into a con- tainer that was suitable for both irradiation and counting, the following operation may be omitted. Weigh the sample into a suitable counting vessel. Calibrate the detection system for assaying the 320 keV gamma ray emitted by Cr-51 and count sample for a suitable time period. b. Radiochemical Separation Weigh the sample into a 50 ml boiling or distilling flask. Add 1 ml of Cr carrier, 5 ml cone. HN0 3 and 3 ml of cone. HClOit. Place on hot plate and heat using a medium 24 temperature setting until sample has dissolved and Cr 3 has been oxidized to Cr e . To the hot solution add 4 ml of hot eerie sulfate solution and heat for five minutes. Let sample cool. Transfer the sample to a 50-ml extraction tube equipped with a polyethylene-lined top using 15 ml of 1 . 5M HC1. Mix well. Add 10 ml of tribenzylamine solution to the extraction vessel. Agitate for five minutes. Let stand until phases have separated, take a 5-ml aliquot of the organic phase and count on a gamma detection system. Transfer the trap solution to a suitable counting vessel and count if a "closed system dissolution" was used. Remarks This solvent extraction procedure is selective for Cr VI which makes it advantageous to dissolve the sample in an oxidizing media. Ceric sulfate is added to the dissolved sample to maintain the chromium in solution as Cr VI. If gold, antimony, or mercury are present in large quantities, they may also extract under the conditions described. The gamma-ray energies of these three elements are sufficiently different from the 320 keV gamma ray emitted by Cr-51 so that no interference will occur. Calculation The amount of chromium in a sample is calculated using the comparator technique, since the measured counting rate using a standard measuring set-up is proportional to the total activity: Wt (yg/g) of Cr in Sample Counts /min of Cr in Sample Wt (yg/g) of Cr in Standard Counts/min of Cr in Standard 25 Example : Cr in Orchard Leaves (O.L.) 1 yg of Cr standard yields 10.4xl0 3 cpm; count rate of Cr in 1 g of O.L. yields 26.8xl0 3 cpm Cr in 1 g O.L. = 26. 8x1 3 cpm 1 yg 10. 4x10 3 cpm yg Cr in O.L. = 26 - 8xl ° 3 c ? m 10. 4x10 3 cpm Cr in O.L. = 2.577 yg/g sample. F. Rapid, Quantitative Separation for the Determination of Selenium Using Neutron Activation Analysis. 4 H. L. Rook 1. Objective The analytical procedure described in this section was developed to permit the quantitative determination of selenium in biological and botanical materials, at the part per million and sub-part per million levels, using neutron activation analysis (NAA) and with a minimum of chemical manipulations. The physical separation method used minimizes the errors associated with chemical separations, and, in addition, permits the simultaneous determination of mercury. 2. Principle The method is based on the irradiation of the analytical sample in a nuclear reactor followed by the separation of h The material used in this section was taken from a work published by the author in Analytical Chemistry, Vol. 44, No. 7, p. 1276 to 1278 (1972) . 26 selenium by thermal volatilization in an oxygen stream. The selenium separated in this manner is transferred quantita- tively to an adequate container, and the activity of the sample is determined by counting with a Ge(Li) detector and 2048-channel pulse-height analyzer. 3. Procedure Apparatus. The sample combustion train (Figure 1) was made from an outer 19/38 standard taper Vycor 5 joint which had been extended to a total length of 30 cm. The condenser was made from an inner 19/38 standard taper Vycor joint, extended to 25 cm, which was inserted through a liquid nitrogen cold trap made from a small polyurethane acid shipping container. FLOWMETER GAS FLOW VALVE SAMPLE\ QBaaggBD-— 1 QUARTZ' WOOL ^JOINT OXYGEN INLET Figure 1. Sample Combustion Tube and Mercury Trap. Reagents. Standard solutions were made by dissolving a weighed quantity of high purity elemental selenium in con- centrated high purity nitric acid and diluting to volume with In order to specify procedures adequately, it has been necessary to identify commercial materials in this report. In no case does such identification imply recommendation or endorsement by the National Bureau of Standards, nor does it imply that the material identified is necessarily the best available for the purpose. 27 distilled water. The selenium concentration used was 0.5910 mg/ml , which was dilute enough to prevent any significant neutron self-adsorption, but large enough to negate loss due to wall adsorption of selenium in the storage container. Procedure. For analysis, 0.5-gram samples were encap- sulated in cleaned quartz tubes. Standard solutions were encapsulated in a like manner. Five sample tubes were arranged in a concentric ring with a standard in the center in a polyethylene rabbit and irradiated in pneumatic transfer facility RT-3 of the National Bureau of Standards Reactor [1] for two to six hours at a thermal neutron flux of approxi- mately 6xl0 13 n*cm 2 «s l . Flux monitors were not used, as the radial flux variation is less than ±2 percent in RT-3. The samples were allowed to decay 5 days to minimize personnel radiation exposures. The quartz tubes were washed in 1:1 nitric acid, cooled to liquid nitrogen temperature, and broken ^5 mm from the top. The samples were weighed into a ceramic combustion boat, and 2.5-mg of mercuric oxide carrier added. Since selenium oxides are difficult to dissolve in mineral acids, inactive selenium carrier was not used. Mercury carrier was adequate to eliminate adsorption of selenium on the walls of the Vycor condenser and polyethylene counting bottle through the formation of mercury selenide. This was readily soluble in nitric acid, enabling the dissolution procedure developed for the mercury separation to be used. The boat was inserted into the combustion chamber, an oxygen flow of ^30 cc/min was passed over the sample and the sample was then ignited with an oxygen-gas torch. The sample was allowed to burn freely with the downstream quartz wool plug heated to red heat to ensure complete combustion of volatile hydrocarbons and carbon produced during the sample burn. After combustion ceased, the ash was heated to approximately 1000 °C for five minutes and the entire combustion tube heated down to the 28 standard taper joint to drive all volatile compounds of the sample into the condenser. The combustion tube was allowed to cool for five minutes, the liquid nitrogen cold trap was removed, and the condenser was placed joint-end down into a 4-oz polyethylene bottle. The condenser was warmed to room temperature, and the sele- nium and mercury carrier washed into the bottle with 10 ml of concentrated nitric acid and three 10-ml portions of dis- tilled water. The total volume of solution was adjusted to 50 ml and the sample was counted using a 60-cc Ge(Li) detector in conjunction with a 2048-channel pulse-height analyzer. The 265 keV gamma ray of selenium-75 was used for the deter- mination, as the 136 keV gamma ray suffers from interferences from mercury-197m. After irradiation, the standard was frozen in liquid nitrogen, broken in a beaker, allowed to melt, and decanted into a 4-oz polyethylene bottle. The broken quartz vial and beaker were washed with six 5-ml portions of distilled water and the final volume was adjusted to 50 ml. The specific activity at the end of the irradiation was compared to that of the sample for quantitative results. 4. Discussion In this work, the method for the separation of volatile elements, originally developed for the determination of mercury, was studied and adapted to the selenium problem [2] . Four significant improvements over currently used procedures have been attained: the procedure has a minimum of chemical manipulations, thereby minimizing technique related errors; the radio-selenium separation is quantitative, thus elimina- ting errors involved in the determination of chemical yields ; the procedure is based on a physical, rather than a chemical separation allowing the same technique to be used on a wide variety of matrices; and the determination of mercury may be conducted simultaneously using this separation procedure. 29 The method was tested using radiotracer experiments and excellent agreement was obtained by independent analysis using isotope-dilution spark-source mass spectrometry. The procedure has been subsequently used to determine the sele- nium concentrations of three new Standard Reference Materials being issued by the National Bureau of Standards. Results The method was tested for complete recovery of selenium by burning non-irradiated samples of biological material with known quantities of selenium-75 nitrate solutions added. The recovery selenium-75 was brought to 50-ml volume and compared to the activity of a like quantity of tracer diluted directly to 50-ml volume and counted in a standard configuration. The results of these tests indicated a complete recovery of selenium-75 tracer. A total of five separations was carried out with a mean recovery of 99.1 percent and a relative standard deviation of 1.1 percent. Using the analytical procedure developed, the selenium concentration in three new Standard Reference Materials (SRM) being issued by the National Bureau of Standards was determined. These materials are: SRM 1630 Mercury in Coal SRM 1577 Bovine Liver SRM 1571 Orchard Leaves The method described was used to determine the selenium contents in these materials with an accuracy in the mean results of better than ±10 percent relative at the 95 percent confidence level. The results were in excellent agreement with independent analyses using isotope-dilution spark- source mass spectrometry. The individual results are given in Table I. 30 Table I. Selenium contents of Standard Reference Materials, ppm. Results Obtained by Neutron Activation Analysis Using the Combustion Technique SRM 1630 SRM 1571 SRM 157/ Mercury Orchard Leaves Bovine Liver in Coal 0.079 1.08 2.16 0.084 1.05 2.03 0.083 1.16 2.14 0.075 1.13 2.08 0.074 1.12 2.08 0.069 1.16 2.05 0.085 1.17 2.14 0.075 1.11 -- 0.087 Mean 0.079 1.12 2.11 2s ±0.012 ±0.08 0.09 Results Obtained by Isotope-Dilution Spark-Source Mass Spectrometry 0.08 ± 0.02 1.11 ± 0.04 N.D. Not determined. The results indicate that the described separation affords a degree of confidence in the final results not obtain able using the nondestructive activation analysis technique. Yet with the simple one-step combustion separation, samples may be processed in a minimum of time (approximately 15 minutes each), maintaining the process-time advantage of nondestructive activation analysis when large numbers of samples must be analyzed. The technique has the added 31 advantage that the mercury content, which is often of great interest in biological materials, may be determined simul- taneously. 5. References [1] LaFleur, P. D. and Becker, D. A., Eds., NBS (U.S.) Tech Note 548, 10 (1970) . [2] Rook, H. L., Gills, T. E., and LaFleur, P. D. , Anal. Chem. 44, 1114 (1972). 32 SECTION III SPARK SOURCE MASS SPECTROMETRIC ISOTOPE DILUTION Determination of Cadmium, Copper, Iron, Lead, Mercury, Molybdenum, Nickel, Selenium, Silver, Tellurium, Thallium, and Zinc. P. J. Paulsen 1. Introduction The spark source mass spectrometer (SSMS) can detect all elements in the periodic table at detection limits down to a few ng/g. In actual samples there will be interferences between elements, however all elements are inherently de- tectable with similar sensitivity. Quantitative results will require concentrations significantly higher than these "detection limit" values. Isotope dilution (ID) analysis can, in theory, be applied to any element having two or more stable isotopes. If one considers the various practical aspects of SSMS analysis, approximately 50 elements are amenable to isotope dilution SSMS analysis. Our laboratory has had occasion to perform SSMS ID analysis on a total of 36 different elements and a wide variety of matrices during the last 10 years. Isotope dilution SSMS has multi-element capabilities. As many as 28 elements have been measured simultaneously by isotope dilution in a single sample while doing a survey type analysis (designed for low detection limits in pure materials, but not designed for the best precision). In a precise quantitative analysis of a NBS SRM for certification, up to 11 elements have been determined simultaneously from a single sample. Using the element copper as an example, the general procedure for mass spectrometric isotope dilution analysis is as follows: 33 1. A known amount of 65 Cu isotope (spike) is added to a weighed amount of sample to be analyzed. 2. The 65 Cu spike is equilibrated (mixed) physically and chemically with the natural copper in the sample. Equilibration is usually obtained by dis- solution of the sample in the appropriate acids. 3. The equilibrated copper is separated from the sample in a form suitable for mass spectrometric measure- ments . 4. The altered 63 Cu/ 65 Cu isotopic ratio is measured mass spectrometrically . The concentration of copper is then calculated using the formula: ppm m (A -BR) WK v sp sr (by weight) = -jj (g R . ff where W = Spike weight in yg ; M = sample weight in grams K = Ratio of natural atomic weight to atomic weight of spike R = Measured altered isotopic ratio a/b A and B = Abundances of isotopes a and b in spike sp sp F F A and B = Abundances of isotopes a and b in natural From this equation it is seen that it is the ratio (R) of the two Cu isotopes that is required to calculate the copper con- centration. The total amount of copper is not measured. It is therefore not necessary to either recover 100 percent of the copper in the sample or know recovery efficiency in order to obtain quantitative results. Therein lies the biggest advantage of using isotope dilution analysis. 34 Once equilibrium is obtained between the "spike" and "natural" isotopes of an element, procedures for separation can be adopted which optimize such factors as group separa- tions, low blanks, clean separations, etc., without the restraint of having a controlled, known recovery. Efficient recovery methods are, of course, desirable since the sensi- tivity of the method is dependent on recovery. 2. General Considerations for Multi-Element SSMS Isotope Dilution a. Concentration Range Covered Quantitative analysis good to about ±5 percent can be obtained if from 1x10 6 to 1x10 ' g of an element can be loaded into the instrument for sparking. Smaller amounts have been measured on occasion, however, the precision suf- fers proportionally as the concentration is lowered. For a 1-g sample this gives quantitative values at the 1 to 0.1 ppm level. Lower concentrations can be determined at this precision if the sample size is increased, however, in any system there is a limit to either the sample available or the amount of the sample that can be effectively handled. For some of the more common elements the system blank may be the limiting factor and it may increase with the sample size depending on its source. There is no upper concentra- tion limit on the use of isotope dilution SSMS, however at high concentrations there are usually other techniques of analysis which are more cost effective and more precise than the ±5 percent obtained with SSMS photoplate detection. Be- cause of the value of the separated isotopes ($1 to $10/mg + time involved in preparation) , spikes are held to less than 1 mg. Samples with high concentrations are dissolved, ali- quoted and then spiked in order to conserve spikes. 35 b. Spike Levels A given amount of spike will cover only a limited range of concentrations, the range will depend on the natural iso- topic ratio, the purity of the separated "spike" isotope and the precision and accuracy with which the altered isotopic ratio can be measured. A valid analysis cannot be performed when the measured altered isotopic ratio differs from either the natural or "spike" ratio by only the measurement uncer- tainty (± percent standard deviation or ± accuracy) . With photographic plate detection the plate uniformity limits precisions to ±5 percent and accuracy (for ratios ^1) depends on obtaining a good plate calibration curve. The best accuracy with the photoplate is usually obtained if the amount of spike added is adjusted to give a measured isotopic ratio of 1 (R=2 if the natural ratio =1) . This procedure minimizes the errors associated with photoplate calibration and also gives the maximum number of measurable ratios for a given situation. If electronic detection SSMS is used, the accuracy is not dependent on having R=l and a wider concentration range can be covered with a given spike. The effective range is still limited by the measurement precision and the isotopic ratio of the spike and natural element . c. Preconcentration The need for obtaining physical and chemical mixing of the "spike" and natural isotopes for a solid sample is met by dissolving the sample in acids. This step both dilutes the sample and adds extraneous anions. The high absolute sensitivity of the SSMS makes it impractical to consume more than a total of about 10 mg of material in a normal analysis. Some form of preconcentration is therefore usually necessary in order to reach this consumable size and hence obtain maximum sensitivity. The ideal preconcentration procedure 36 would yield only those elements to be analyzed and no others. This would give the best analytical sensitivity while mini- mizing interference problems. The presence of a large excess of an element not being analyzed in the SSMS sample can cause interference problems. This is especially true if it is an unwanted cation which carries along an equivalent anion from the acids used in dissolution - double trouble. Isotope dilution SSMS is a practical method mainly be- cause a number of elements can be determined simultaneously. This means that any preconcentration method used needs to provide group separations. In some types of samples this can be obtained simply by excluding the major components of the sample. A number of preconcentration procedures have been used for SSMS isotope dilution. These include ion ex- change, solvent extraction, electrodeposition, precipitation with and without carriers, and high temperature volatiliza- tion. Electrodeposition onto high purity gold wires (photo- plate detection) is an especially attractive method since it has a very high rejection factor for organics and anions originating from the acids used in sample dissolution. The SSMS readily detects organic molecules. If organic material from either the sample or the reagents used in pre- concentration is present in the sample loaded into the SSMS, interferences can result. d. Interferences As in all analytical methods it is the identification of interferences that enables the proper action to be taken which avoids errors in analysis. There are three general types of interferences in SSMS ID analysis. 1. Isobars - adjacent Z elements with isotopes having the same nominal masses: Interferences of this type can always be identified and it occurs with only a few ele- ments. In some cases proper selection of the isotopes measured may minimize the magnitude of the interference, 37 and/or allow correction for the interference. Chemical pre- concentration procedures could also be used to remove the undesired element. In any case, this type of interference could not go undetected. 2. Multicharged ions of high mass falling at the same nominal mass as the singly charged ion of the desired element: The SSMS disperses ions according to their mass to charge ratio (m/e) . The energy available for ionization in the spark is so high that all elements can be multiply ion- ized. This can result in interferences when two elements of different masses and different charge states have the same nominal mass to charge ratio. The relative intensities of the +1, + 2, +3, etc., ions of an element follows a predict- able pattern (intensity drops off by a factor of ^3 for each additional charge) . This type of interference is therefore readily recognized by a visual examination of the photoplate Again, in many instances, interferences can be avoided or corrected for by measuring the proper isotopes of both ele- ments. In some cases the interference can be further mini- mized or avoided if a multiply charged ion of the desired element can be measured. Again this type of interference would not go undetected. 3. Interference by compounds of the same nominal mass as the element being analyzed: Hydrocarbons fit this classification. The resolution of any SSMS using photoplate detection is adequate to resolve and identify the presence of a hydrocarbon line with the same nominal mass as an element line. Quantitative measurements can be made on the resolved element lines providing the hydrocarbon line is not too intense. In this case the sample preconcentration method could be modified to lower hydrocarbons or in some situations (explained later) the interference can be avoided by measuring a multiply charged ion of the element. 38 Other types of compound lines will usually not have a large enough mass difference from an element of the same nomi- nal mass to be resolved into separate lines. The major diffi- culty with this type of interference is that the intensity of compound interferences cannot be as easily predicted from other information on the photographic plate. However, since the SSMS can detect all elements there is good information on what elements are available as starting material to form the com- pounds. The stability of a compound in the form of a positive ion (the species being detected) under the conditions that exist in the spark source is very small. It is only the ex- tremely high sensitivity of the SSMS combined with the presence of elements at high concentrations that allows the detection of such ions. Under most conditions these interferences occur only for elements being determined in the low ppm range. When checking for potential interferences a list of possible compounds can be obtained by following these gen- eral rules : 1. Consider only the major elements (as detected by SSMS) . 2. All samples, regardless of source, have adequate 0, H, (and possibly carbon) available for forming compounds . 3. Anions of the acid used in sample dissolution can appear in all possible forms, i.e. , C1CK gives CI , + + + CIO , C10 2 , C10 3 , plus "cation" combinations thereof. 4. Group la, lb, 2a, and 2b elements have a tendency to form dimers, trimers, etc., and these can also pick up oxygen and hydrogen, or anion fragments. 5. The compounds that cause interferences need to exist as ions for less than 1 ys; during their lifetime (as a detected ion) no collisions occur with other ions, molecules, or any surface. The combinations that will occur therefore may appear to be strange when viewed from a classical chemical background. 39 The presence of compounds can be detected by their generation of abnormal isotopic ratios and of lines which do not correspond to any element lines. If the compound has an element with more than one isotope it may allow correc- tions to be made. Abnormal lines that can be identified give good clues to the possibility of a related compound. For instance, if a sample (with high NaClCU content) shows lines which are identified as NaC10 2 then the spectrum should also be checked for NaClot, NaC10 + , Clot, Clo£, etc. One approach to this type of interference problem is to measure the isotopic ratios for all elements being analyzed in an unspiked sample. If the ratios found are the natural ratios of the elements it can be assumed that no interfer- ences exist. This approach requires good control over the sample preconcentration process. There is a simple method of avoiding errors of analysis from compound interference that will apply to most but not all elements. The method is to measure the isotopic ratios for elements on a + 2, +3, or +4 ion instead of the +1 ion. The stability of a compound with a + 2 or +3 charge on it is so low that such species cannot be detected. The +2 and +3 ions of an element on the other hand are only a factor of ^3 and ^10 times less intense than the +1 ion. There will be a loss in sensitivity in such an analysis and the method cannot be used much below 1 ]ig. The pre- cision of isotope ratio measurements using the photoplate is the same for all charge states (limited only by plate uniformity) so the precision does not suffer. There is one major qualification that prevents use of multiply charged ions for all elements. Both measured isotopes of the multi- ply charged ion of the element must fall at fractional mass/ charge (m/e) values if that charge state is to be usable. This is necessary since the unit m/e positions are very likely to have either a +1 element line or possibly a 40 compound line. This is especially true at the lower m/e region of the plate where most of these types of measure- ments would be made. Interference by compound lines is usually only signif- icant in the analysis of low ppm concentration elements. The interference can be minimized by keeping elements not being analyzed out of the sample, and interference can be eliminated if a multiply charged ion can be measured. The major problem is being able to spot the existence of this type of interference. 3. Analysis of Organic Materials Using SSMS Isotope Dilution Most of the SSMS ID analysis at NBS has been for the certification of Standard Reference Materials (SRM's). The procedures adopted have emphasized accuracy and precision over other factors such as speed and turn around time. Since a wide range of analytical techniques are applied to a SRM certification, the SSMS ID analysis stressed only those elements where it would be most effective. In general, most elements above 100 to 500 ppm would be analyzed by other methods to a better precision than the ±5 percent obtained by SSMS photoplate detection. More elements could have been analyzed by SSMS ID than were reported if it had been necessary. The organic materials (not all SRM's) that have been analyzed by SSMS ID are: orchard leaves, beef liver, coal, tomato leaves, pine needles, spinach, tuna fish, urine, wheat, rice, fuel oil, and river bottom extract. The ele- ments determined for NBS organic SRM certification are: Pb , Tl, Hg, Te, Cd, Ag, Mo, Se, Zn, Cu, Ni, and Fe. A total of 35 elements have been analyzed for SRM certification if other types of SRM's are included. 41 The chemical preconcentration methods used in the analysis of the organic materials include: electrodeposi- tion onto high purity gold wires, ion exchange, vaporization of volatile elements, and co-precipitation. Our analysis of coal, beef liver, and spinach SRM's will be used as examples for SSMS ID analysis. A detailed description of the instrumentation and tech- niques used to perform the actual analysis by SSMS ID will be found in the four papers reproduced in the Appendices 7 through 10. Although the chemical species discussed in those papers are contained in a metallic matrix, the same analytical procedures are applicable to organic matrices. 4. Preliminary Analysis The preliminary analysis serves two functions in a SRM analysis. First, several group separation techniques can be tested to see which system recovers the greatest number of elements of interest. Second, the concentration levels of the different elements needs to be determined in order to adjust the spike level for each element for optimum precision and accuracy in the actual analysis. If no information is available on the sample, it is spiked with 100 ppm, 10 ppm, and 1 ppm of a master spike solution containing "spikes" of 28 elements. This range of spike levels can be obtained by changing both sample and spike amounts, but it is best to keep the spike above 1 yg. Analysis of these three samples would give reasonably accurate results over a concentration range from ^500 ppm down to ^0.1 ppm. If an element not in the spike is detected, its concen- tration can be estimated by using one of 28 spiked elements as an internal standard. In all of the SSMS ID analyses performed to date the preconcentration procedure has been selected to give the maximum number of desired elements from a single sample processing. Two independent procedures (two SSMS samples) would have doubled both the chemical and the 42 SSMS workload but would not have doubled the number of elements determined. What has been just referred to as a preliminary analy- sis for our NBS SRM SSMS ID analysis would, in fact, provide adequate accuracy and precision for "final" analysis of many types of analyses. Such a system would be much more effi- cient than the one we feel obligated to use on NBS SRM's. 5. Procedure Used for Analysis of Coal, Beef Liver, and Spinach Spiking Weighed samples of the organic material were spiked with proper amounts of each element's "spike" in order to obtain an equilibrated ratio (R) of 1. Silver has a natural ratio of 1, it is spiked to give a ratio of 2. The amounts of spike required were computed using the results of the preliminary analysis. The spike elements can be added one element at a time, however, our preferred method is to mix all the spikes together in the proper pro- portion and then make just one addition. In either case, spike calibration samples are prepared at the same time using the same pipet setting and spike solutions as used on the samples. Known amounts of the pure natural elements are also added to these spike calibration samples. The natural solutions are obtained by quantitative dilutions of a solu- tion prepared from a known amount of the pure natural element. SSMS analysis of the spike calibration sample gives an accurate determination of the amount of spike actually added to the samples. "Blanks" are also spiked at the same time as the samples using only 5 to 10 percent as much spike as was used on the sample (but not less than 0.1 yg) . This level of spiking can safely measure the blanks down to less than 1 percent of the value being determined in the analysis. The blank samples are then treated in an identical manner as the real samples. 43 Spike Equilibration Equilibration is obtained by the acid dissolution of the samples. These samples were dissolved by a stepwise treatment with HNO 3 , HC1CU , and HF (no HF for beef liver). The amounts of reagents required were determined during the preliminary analysis. A detailed discussion of the procedures used for the dissolution of samples contained in an organic matrix is given in Section I, "Sample Preparation" by T . J. Murphy of this work. The analyst should refer to this section when such matrices are under consideration. 6. Preconcentration All three samples were separated-preconcentrated by electrodeposition onto high purity gold wire cathodes. This technique is used often since it has a number of advantages. The method is simple and uses no reagents beyond those needed for sample dissolution. This results in low analysis blanks. The gold wires are loaded directly into the SSMS for analysis and give high elemental sensitivity with photo- plate detection. Electrodeposition gives good rejection of the alkalies and alkaline earths (high in these samples) which are seldom analyzed by SSMS ID. The elements are recovered in a reduced form and therefore no extraneous anions or cations are added to the sample being sparked. Following dissolution (equilibration) the samples were evaporated to dryness and redissolved in 10 ml of distilled water. They were plated for three hours or more at voltages from 2.5 to 3.0 V. The elements recovered depended on both the plating voltage and the acid content of the plating solution. Zinc recovery is very sensitive to the presence of small amounts of acid in the plating solution. Unless the sample is dried carefully, zinc will not plate. The six-nines pure gold wire used for electrodeposition is about 44 1/2 mm diameter. The wire is plated only on the tip for a distance of about 2 mm. 7. Sample Sparking Two cathode wires are plated during deposition. These are loaded so that their 2 mm of plating overlap and that the sparking will occur along this 2 mm length. Two photo- plates are exposed using 1/2 of the plated surface for each. A graded series of 20 exposures are made on each plate starting from lxlO" 12 coulombs and going up to 1x10 8 coulombs. The electrodes are moved so as to spark new sam- ple on each exposure. The gold wire becomes warm while making the 10" 8 exposures on the first plate. This outgasses the sample appreciably and the second plate exposed has sig- nificantly lower hydrocarbon content (interference) . Some elements (Cd, Se) may also be volatilized and show lower intensities on the second plate. 8. Photoplate Analysis The plates are carefully examined for any of the types of interferences discussed previously. Based on this infor- mation, appropriate lines and charge states are selected for each element for the densitometric measurement of the altered isotopic ratio. Even though electrodeposition gives the "cleanest" possible spectrum, significant amounts of the sample's major cations and anions (from acid) are retained in the plated sample by occlusion, adsorption and absorption. The procedures used on these samples result in perchlorate being the major anion in the final solutions. Perchlorate fragments showed up in all three organic SRM's. The SSMS spectrum show lines for Cl + , C10 + , Cioj, and Clot. The CIO gives lines at m/e 51 and 53, these would interfere with SSMS ID analysis of V and Cr respectively if they had been determined in any of the samples (they were not) . The 45 C10 2 gives lines at m/e of 67 and 69 (two lines are due to the fact that CI has two isotopes) . Zinc is spiked with 67 Zn and therefore C10 2 can interfere with the analysis of Zn. Fortunately, the amount of C10 2 interference can be determined by measuring the line at 69. Also the measurement of the 67 Zn/ 6lt Zn ratio can be made at Zn 3 (if enough Zn is recovered to give a measurable +3 zinc line) where zinc is free from CIO2 interference. Even if interferences are not detected, as many dif- ferent charge states of an element are measured as is practical. This gives the maximum number of ratios (we stop at <"20/element) and improves the accuracy of the analysis. (A multiply charged ion is used only if both isotopes fall off unit mass and have no other type of interference) . The +1 data for an element is calculated separately to determine if it gives significantly different results than that from the multiply charged ions. If it is different, it is assumed to be caused by an undetected interference and the +1 data are discarded. Results The results of the analysis of coal, beef liver and spinach are given in tables 1, 2, and 3. 46 Table 1. Trace elements (yg/g) in Coal, SRM 1632, by spark source mass spectrometry isotope dilution. Pb Ag Cd Se Zn Cu Tl Bottle 358 1 22.6 0.041 0.15 3.02 32 17.7 0.642 2 31.6 0.065 0.14 2.88 36 20.1 0.648 Bottle 292 1 27.1 0.057 0.17 3.02 18.4 2 41 17.6 0.603 3 30.8 0.22 36 19.3 0.532 4 28.7 0.12 3.03 44 38 17.2 0.532 Average 28 0.07 a 0.17 3.0 18 0.59 Recommended value < 0.1. Table 2. Trace elements (yg/g) in Beef Liver, SRM 1577, by spark source mass spectrometry isotope dilution. Pb Cd Ag Mo Se Zn Cu Tl 1 0.41 0.32 0.058 3.25 1.13 126 181 0.048 1 0.41 0.29 1.12 128 183 0.046 2 0.38 0.26 0.060 3.08 1.12 131 184 0.050 2 0.33 0.27 0.081 1.07 182 0.051 3 0.36 0.27 0.054 3.34 1.12 130 184 0.047 3 0.33 0.27 0.077 3.24 1.12 130 197 0.050 4 0.33 0.28 0.061 3.24 1.10 121 183 0.045 6 0.34 0.26 0.067 1.12 120 181 0.043 Average 0.36 0.28 0.06 3.23 1.11 126 184 0.047 47 Table 3. Trace elements (yg/g) in Spinach, SRM 1570, by- spark source mass spectrometry isotope dilution Sample No. Lot No. 304 6-2 305-1 13-2 305-2 13-2 306 9-1 Average 1.12 0.031 1.67 11.5 49.5 Pb Tl Cd Cu Zn 1.15 0.028 2.11 12.18 49.0 1.09 0.037 1.54 11.13 50.0 1.10 a 1.53 11.25 50.2 1.15 0.029 1.50 11.36 48.7 Insufficient data obtained for sample 305-2. Table 3b. Cd and Cu in spinach (Lot 6-2) , additional sampling. Element yg/g Cd 2.32 b Cu 12.54 Original sample depleted; additional material of the same lot was analyzed. The close agreement for the two independent runs on Lot 6-2 for Cd and Cu indicates that the Cd is inhomogeneous in the sample and should not be certified. 48 SECTION IV ATOMIC ABSORPTION AND FLAME EMISSION SPECTROMETRY A. Determination of Beryllium, Cadmium, Copper, Magnesium, Manganese, Mercury, Nickel, and Zinc by Atomic Absorption, and of Calcium, Potassium, Rubidium, and Sodium by Flame Emission Spectrometry in Biological and Botanical Materials. T. C. Rains 1. Objective The procedures described in this section were developed and applied to the quantitative determination of traces of beryllium, cadmium, copper, magnesium, manganese, mercury, nickel, and zinc, using atomic absorption spectrometry (AAS) , and calcium, potassium, rubidium and sodium, using flame emission spectrometry (FES) , in a variety of biological and botanical materials. The uncertainty of these methods are 2 to 10 percent, and improvements in the accuracy of the measurements can be attained if a number of precautions are taken. Instrumental parameters must be optimized for each element; standard solutions should be prepared to contain the same chemical species as are present in the unknown sample. Also, the optimum precision and accuracy are attained by AES and FES when the technique of bracketing is used [1-4] . 2. Principle The absorption of radiant energy by ground state atoms in the gaseous state is the basis of AAS. When radiation of proper wavelength passes through a vapor containing ground state atoms, some of this radiation will be absorbed by ex- citation of particular atoms (3) . The magnitude of the absorption signal is directly related to the number of these ground state atoms found in 49 the optical path. Ground state atoms necessary for the measurements are produced from the sample material, by vaporization of the material introduced in an adequate excitation source followed by dissociation of the molecular species into neutral atoms. Normally these steps are carried out by a nebulizer associated with a flame or by a nonflame excitation device [5,6]. In flame emission spectrometry a combustion flame is used to provide the energy necessary to atomize and to ex- cite the atoms introduced into the flame. The solution con- taining the element of concern is converted into small droplets by a nebulization process. The small droplets enter the flame where they are dried, vaporized, and dis- sociated. The dissociated species are then excited to emit characteristic radiation having an intensity proportional to the number of excited particles. The emission spectra produced by various chemical species excited in combustion flames is relatively simple, and the discrete monochromatic radiations can be easily separated by a monochromator and measured by a photoelectric device with a minimum of radiation interferences [1] . In certain circumstances, the concomitant or flame gases can produce atomic lines or overlapping band structures which may interfere with the measurements. To overcome these difficulties in FES, it is recommended that repetitive optical scanning in the derivative mode be used. This technique was described by Snelleman et al . [7]. Further discussions on the principles of AAS and FES will be found in references 1 to 6. 3. Procedure 1. Instrumentation A number of commercial instruments are available to perform measurements by AAS and FES, and it is assumed that the analyst is familiar with his particular instrument. It 50 is not possible in this section to give detailed instructions concerning specific instrumental parameters to assure sta- bility, linearity, adequate flame conditions, etc. for all the available commercial instruments. For AAS, the Perkin Elmer Models 403 and 603 were used in this work. 1 For FES, the instrumentation described by Snellman et al . (7) was used. 2 . Reagents All standard stock solutions were prepared from high purity metals or salts 2 as described in reference 2, Chapter 13 reproduced in Appendix 14. All other solutions were prepared from ACS reagent grade chemicals. All acids were NBS purified [8] . 3. Sample Preparation and Analysis It is difficult to cover the many problems associated with sampling the variety of plant and animal tissues. The sample, of course, must be representative, and it must be protected from contamination and deterioration prior to analysis [9, see also Appendix 2] . It must be determined at the start if the analysis is to be reported on the "as received" basis or after drying to a constant weight by one of several methods [1,9]. 1 In order to describe adequately the materials and experi- mental procedures, it was occasionally necessary to identify commercial products by manufacturer's name or label. In no instance does such identification imply endorsement by the National Bureau of Standards, nor does it imply that the particular product or equipment is necessarily the best available for that purpose. 2 Editor's Note - The weights given in this reference were calculated on the assumption that the compounds are 100 percent pure. 51 a. Sample Preparation and Analysis for Na , K, Rb , and Ca by Flame Emission Spectrometry. The following procedure is applicable for the dis- solution of a variety of tissues such as NBS Orchard Leaves, SRM 1571, and Bovine Liver, SRM 1577, for the determination of the elements mentioned above: Transfer a 2 to 5-g sample of tissue to a 100 ml Teflon beaker and add 25 ml of dilute nitric acid 1:1. Digest on a hot plate for one hour, and cool and rinse the sides of the beaker with water. Add 10 ml of HNO3 and 5 ml of HF and evaporate to approximately 10 ml. If the sample contains black particles of unreacted carbon, repeat the addition of 10 ml of HNO3. Add 10 ml of HC1CU , heat to strong fumes of perchloric acid, cool and wash the sides of the beaker and continue the digestion until the sample volume is reduced to approximately 2 ml . In some analyses where the concentration of the analyte is less than 1 ug/g, separation and preconcentration are required, therefore, the volume of samples should be evaporated to near dryness to remove excess acid. Transfer the solution to a 100-ml volumetric flask and dilute to volume with water. Proceed with the FES determination as described for each individual element. 4. Determination of Sodium by FES Prepare a standard stock solution of sodium from NBS SRM 919, Sodium Chloride, as described in reference [2] (see Appendix 14). Prepare a series of standard working solutions from the standard stock solution containing 1.0, 2.0, 3.0, 4.0 and 5.0 yg Na/ml in . IN HC1CK . Transfer duplicate 52 test portions of the sample to 100-ml volumetric flasks. The aliquot taken will depend upon the concentration of sodium in the sample. The final concentration should be 2.0 to 3.0 yg Na/ml in the final volume of 100 ml. To one volumetric flask, add 100 yg of Na and dilute the solution to volume. Label this sample as spiked with 1.0 yg Na/ml. Start the FES instrument [7], set the wavelength control to 589.0 nm, and, install an oxy-hydrogen burner (total consumption) in the instrument. Set the tank regulators and flow rates for the oxygen and hydrogen as recommended by the burner manufacturer. Light the oxy-hydrogen burner and allow the burner and the spectrometer to stabilize for 15 minutes while aspirating water. Nebulize the 5.0 yg Na/ml or higher standard into the flame and adjust the amplifier gain until the emission signal is reading 90 percent transmission or 90 divisions on the strip chart recorder. Nebulize water after each sample or standard and record the percent transmission. Nebulize the 1.0, 2.0, 3.0, 4.0 and 5.0 yg Na/ml into the flame and record the percent transmissions. Repeat the measure- ments until the percent transmissions of the standards do not deviate by more than ±0.5 percent. Nebulize the unknown solutions and record the percent transmissions. Repeat the nebulization of water between all samples. Nebulize a standard solution after each 3 to 5 unknown solutions. (Note: If the FES in- strument drifts, the standard solutions should be repeated more often.) Determine the net percent transmission for the standards and 53 unknowns by substracting the background and water readings from each standard and unknown. Calculate the concentration of the unknown sample as follows: Calculation Set up a calibration curve using a least square fit on a hand calculator. (Note: If a hand calculator or computer is not available, plot on rectilinear graph paper the percent transmission corrected for the background, as the ordinate, and the concentration of the standards, expressed in micrograms of analyte per milliliter, as the abscissa. The calibration curve should pass through the origin) . Determine the concentra- tion of the unknowns from the calibration curve. If the recovery of the standard addition is not 100 percent, correct the concentration by the following equation: 3 C = y-x where: x = yg/ml of analyte in the unknown as determined from the calibration curve. s = amount of standard added, yg/ml in final volume. y = yg/ml of analyte found in unknown with standard added as determined from the calibration curve. C = concentration yg/ml. 3 The calculation procedure used here is also applied to the succeeding determinations described in this section. 54 Calculate the analyte concentration of the original sample as follows: let W = weight of solid sample, g, Vs = volume of liquid in which W is dissolved, ml , V = test portion of liquid sample or aliquot of Vs , ml, v., V~ • • • V v. n J-14.-.C 12 nr Z = dilution factor = 12 n where: V-. = volume to which V is first diluted, ml, a 1 = aliquot of V,, ml, V~ = volume to which a 1 is diluted, ml, a = aliquot of V~ , ml, etc. to V and a , 2 n 2 n n Vr = final volume on which measurement is made, ml. C = analyte concentration of Vr, yg/ml ; then analyte in the original sample, oz-v yg/g = vw S 5. Determination of Potassium by FES Prepare a standard stock solution of potassium as described in reference [2], and prepare a series of standard working solutions from the standard stock solution containing 1.0, 2.0, 3.0, 4.0 and 5.0 yg K/ml in . IN HCIO^. Proceed as directed in paragraph 4 (page 52) except set the wavelength control to 766.5 nm. Calculations Proceed as directed in Calculations, page 54. 55 6. Determination of Rubidium by FES Prepare a standard stock solution of rubidium as described in reference [2] , and prepare a series of standard working solutions from the standard stock solution containing 0.1, 0.2, 0.3, 0.4, and 0.5 yg Rb/ml in . IN HC1CK which also contains 100 yg K/ml . Transfer duplicate test portions of the sample to 10-ml volumetric flasks. (Note: Select an aliquot of the sample that will provide a final concentration of 0.2 yg Rb/ml.) To one volumetric flask, add 1.0 yg of Rb and dilute the solution to volume. Start the FES instrument [7], set the wavelength control to 780.0 nm and proceed as directed in item 4 (page 52) . Calculation Proceed as directed in Calculations, page 54. 7. Determination of Calcium by FES Prepare a standard stock solution of calcium as described in reference [2] using NBS SRM 915, Calcium Carbonate. Prepare a series of working solutions of calcium containing each 1000 yg K/ml in 0.1 N HC1CK . The concentration of the standards will depend upon the type of tissue to be analyzed. (Note: For NBS-Bovine Liver, standard solutions containing 0.2, 0.4, 0.6, 0.8 and 1.0 yg Ca/ml were prepared. For NBS-Orchard Leaves, standard solutions containing 2.0, 4.0, 0.6, 0.8 and 10.0 yg Ca/ml were prepared.) Select an aliquot of the unknown and spike with an appropriate concentration of calcium as described in item 4 (page 52). (Note: A solution of K must be added to the unknown to provide a final concentration of 1000 yg K/ml.) Start the FES 56 instrument [7] , set the wavelength control to 422.7 nm and install a premix laminar flow burner with a 5-cm slot burner head. Set the tank regu- lator and flow rates for air, nitrous oxide, and acetylene as recommended by the manufacturer and light the air-acetylene flame. Switch to nitrous oxide-acetylene, while observing all safety pre- cautions recommended by the manufacturer for nitrous oxide-acetylene flame. Proceed as directed in paragraph 4 (page 52) . Calculations Proceed as directed in Calculations, page 54. b. Sample Preparation and Analysis for Be, Cd, Cu, Mg, Ni, Mn, and Zn by Atomic Absorption Spectrometry. The sample preparation procedure described in paragraph 3, page 51, is used for the dissolution of a variety of tissues such as orchard leaves (SRM 1571) , and bovine liver (SRM 1577) for the determination by AAS of the elements mentioned above. 8. Determination of Beryllium by Nonflame AAS Prepare a standard stock solution of beryllium as described in reference [2] , and prepare a series of standard working solutions from the standard stock solution containing 0.005, 0.010, 0.015, 0.020 and 0.025 yg Be/ml in . IN HN0 3 . If the unknown solu- tions contain more than 10 yg Ca/ml , add an equal amount of calcium to the beryllium standard working solutions. Transfer duplicate test portions of sample to 100 ml Teflon beakers, evaporate to near dryness. Add 4 ml of . 2N HN0 3 , and transfer to 10-ml volumetric flasks. Select an aliquot that will provide a final concentration of Be in the range 57 of 0.010 to 0.015 yg Be/ml . To one volumetric flask, add 0.05 yg of Be and dilute the unknown solution to volume. Start the instrument (Perkin-Elmer Model 603 equipped with the HGA-2100 graphite furnace) as described in the instruction manual [12] . Turn on the beryllium hollow cathode lamp, adjust the lamp current to 14 mA, and set the wavelength to 234.9 nm using a slit width of 0.7 nm. Turn on the D~ arc lamp, and allow the hollow cathode and D~ arc lamps to warm for 15 minutes, then balance the two beams. Set the HGA-2100 furnace at the following instrumental conditions: dry for 30 seconds at 100 °C; char for 40 seconds at 700 °C; atomize for 7 seconds at 2800 °C. Set the instrumental condition as follows: peak height integration time for 5 seconds; scale expan- sion for 1.0. Using an Eppendorf pipette, transfer 0.005 to 0.025 yg Be/ml, establish a calibration curve and repeat the calibration curve until the absorbances are within ±5 percent. Determine the net absorbances of the unknown and standard solutions. (Note: to increase the precision of the analysis, analyze each sample in triplicate.) Calculation Proceed as directed in Calculations, page 54, except plot absorbance instead of transmission versus concentration. 9. Determination of Cadmium by Nonflame AAS Prepare a standard stock solution of cadmium as described in reference [2] and prepare a series of standard working solutions containing 0.002, 0.004, 0.006, 0.008 and 0.010 yg Cd/ml in . IN HN0 3 . Pro- ceed as directed in paragraph 8 (page 57) except for the following instrumental conditions: for the 58 Perkin Elmer 603 instrument set the cadmium hollow cathode lamp at 8 mA and wavelength at 2 28.8 nm; set the HGA-2100 furnace: dry for 30 seconds at 100 °C; char for 40 seconds at 250 °C; atomize for 7 seconds at 2100 °C. Calculations Proceed as directed in Calculations, page 54, except plot absorbance instead of transmission versus concentration. 10. Determination of Copper by AAS Prepare a standard stock solution of copper as de- scribed in reference [2] , and prepare a series of working solutions of 0.5, 1.0, 2.0, 3.0 and 4.0 yg Cu/ml in . IN HCIO^ that contain 500 yg Na/ml . (Note: For plant material, potassium should be sub- stituted for sodium.) Transfer duplicate test portions of the sample to 25-ml volumetric flasks. Select an aliquot of the sample that will provide a final concentration of 1.0 to 2.0 yg Cu/ml. To one volumetric flask add 25 yg of Cu and dilute the unknown solution to volume. Attach a premixed air-acetylene burner having a three slot burner head to the Perkin-Elmer 603 AAS unit as described in the instruction manual [12] , and turn on the unit containing a copper hollow cathode lamp. Set the hollow cathode lamp current at 15 mA, set wave- length to 324.7 nm, and use a slit width of 0.2 nm. Turn on the D ? arc lamp, allow the lamp to warm for 10 minutes, and balance the two beams. Set the instru- mental conditions as follows: mode-concentration, integration time for 5 seconds, scale expansion 2.0. Turn on the air-acetylene flame and allow the burner to warm for 15 minutes while nebulizing water. Con- 59 struct a calibration curve by nebulizing the series of analyte standards and measuring their absorbance. Always nebulize water before and after taking a measurement of the standards or unknown solutions. Obtain a net absorbance by subtracting any reading obtained for the water from the readings of the standard solutions. Repeat the calibration curve until the absorbances are within ±2 percent. Measure the absorbance of the unknown including a reagent blank. (Note: To increase the precision and accuracy of the analysis, measure a low standard, unknown, and then a higher standard.) Obtain the net absorbance by substracting any reagent blank and/or readings obtained for the water from the readings of the standard solution. Calculations Proceed as directed in Calculations, page 54, except plot absorbance instead of percent transmission versus concentration. 11. Determination of Magnesium by AAS Prepare a standard stock solution of magnesium as described in reference [2] , and prepare a series of working solutions of 0.1, 0.2, 0.3, 0.4, and 0.5 yg Mg/ml in . IN HCIO^ that contains 0.1 per- cent lanthanum solution. (Note: Magnesium concen- trations of 0.5 to 2.5 yg/ml can be used for many types of plant tissues, but due to the nonlinearity of the calibration curve, the unknown samples must be closely bracketed with standard solutions.) Transfer duplicate test portions of the sample to 25- or 100-ml volumetric flasks and select an aliquot of the sample that will provide a final concentration of 0.2 yg Mg/ml. Add 0.1 percent lanthanum solution 60 to all samples. To one volumetric flask, add a concentration of magnesium that will provide a final concentration of 0.1 yg Mg/ml . Proceed as directed in paragraph 10, (page 59), except set the wavelength to 285.2 nm and the magnesium hollow cathode lamp at 15 mA. Calculations Proceed as directed in Calculations, page 54, except plot absorbance instead of percent transmission versus concentration. 12. Determination of Manganese by AAS Prepare a standard stock solution of manganese as described in reference [2] , and prepare a series of working solutions of 0.5, 1.0, 1.5, 2.0 and 2.5 yg Mn/ml in 0.1 N HC1CK that contain 300 yg K/ml and 20 yg Ca/ml . (Note: For some animal tissues, the manganese concentration may be a factor of 10 lower. In this case the analyst should use the nonflame technique with Mn standard solutions of 0.005 to 0.025 yg/ml.) Transfer duplicate test portions of the sample to 25-ml volumetric flasks and select an aliquot of the sample that will provide a final con- centration of 1.0 to 1.5 yg Mn/ml. Add a known concentration of manganese to one of the samples as described previously. Proceed as directed in paragraph 10 (page 59) except set the wavelength to 279.5 nm and the manganese hollow cathode lamp at 18 mA. Calculations Proceed as directed in Calculations, page 54, except plot absorbance instead of percent transmission versus concentration. 61 13. Determination of Zinc by AAS Prepare a standard stock solution of zinc as de- scribed in reference [2] , and prepare a series of working solutions of 0.1, 0.2, 0.3, 0.4 and 0.5 yg Zn/ml in . IN HC1CK that contain 300 yg K/ml and 200 yg Ca/ml. Transfer duplicate test portions of samples to 25-ml volumetric flasks and select an aliquot of the sample that will provide a final con- centration of 0.2 to 0.3 yg Zn/ml. Add a known con- centration of zinc to one of the samples as described previously, and proceed as directed in paragraph 10 (page 59) except set the wavelength to 213.8 nm and the zinc hollow cathode lamp at 10 mA. Calculations Proceed as directed in Calculations, page 54, except plot absorbance instead of percent transmission versus concentration. 14. Determination of Mercury by Cold Vapor AAS Sample Preparation The following procedure is applicable for the dissolu- tion of a variety of tissue samples such as NBS-SRM Orchard Leaves and Bovine Liver for the determination of mercury. The equipment used in this procedure is described by Rains and Menis [11] . Transfer a 2 to 3-g sample to a digestion flask, add 25 ml of 18N H 2 S0 4 and 20 ml of 7N HN0 3 and connect the Allihn condenser. Using a heating mantle, gently heat the flask for one hour at 150 °C. Cool the flask, rinse the inside of the Allihn condenser and add 20 ml of HN0 3 - HC1CU (1+1) through the con- denser. Drain the water from the condenser and con- tinue heating the flask until dense fumes of perchloric acid appear. Cool the flask and rinse the inside of 62 the Allihn condenser with 5 ml of water. If the sample shows a tendency to char, add an additional 10 ml of HN0 3 and reheat to fumes of perchloric acid. Continue heating the flask until dense fumes of H 2 S0if appear, cool, and rinse the condenser with 1:10 HNO3. Heat to boiling, cool, and rinse the condenser again. Transfer the solution to a 100-ml volumetric flask, add 5 drops of one percent solution of K 2 Cr 2 7 and dilute to volume. Proceed with the determination of mercury by the cold vapor method as described in the next paragraph. Determination Prepare a standard stock solution of mercury as de- scribed in reference [2] , and prepare a series of standard working solutions that contain 0.005, 0.010, 0.015 and 0.020 yg Hg/ml in . 1M HN0 3 . Add five drops of one percent K 2 Cr 2 07 and dilute to volume. Allow samples and standards to digest one hour after the addition of K 2 Cr 2 07 before proceeding with the analysis. Prepare "reducing" and "diluting" solutions as described in Appendix 15. Turn on the Perkin-Elmer 403 AAS unit as described in reference [12], insert the mercury hollow cathode lamp, and set the current at 10 mA. (Note: Any commercial hollow cathode lamp will give adequate sensitivity; however, vapor discharge lamps are not recommended due to severe line broadening.) Attach a quartz absorption cell (10 cm x 0.8 cm) to the burner head. This absorption cell can be constructed from a 0.8 cm internal diameter and 10 cm long Vycor tubing with gas inlet and outlet ports and quartz windows. The quartz windows are secured to the Vycor tubing with epoxy cement. The cell is wrapped with a heating tape and attached to the burner head. Turn on the 63 D~ arc lamp and allow it to warm up for 15 minutes, then balance the hollow cathode and D~ arc lamps as recommended in the instruction manual. Heat the absorption cell to 200 °C with the aid of a variable autotransformer as directed in Appendix 15. Attach the reducing cell and set argon gas flow at 0.3 1/min. Transfer an aliquot of standard mercury working solution to the reducing cell. (Note: A mercury- concentration of 0.2 yg will give an absorbance of 0.800.) Dilute to 25 ml with diluting solution, and turn on recorder. Add 20 ml of reducing solution and immediately close the system. Record maximum absorbance. (See comments under calibration in Appendix 15.) Obtain a net absorbance by substrac- ting the value obtained for the reagent blank from the standard solution. Repeat the calibration curve until the absorbance values are within ±2 percent. Transfer an aliquot of unknown sample that has a mercury concentration of 0.C1 to 0.1 yg into the absorption cell. Proceed as directed above. Bracket the unknown sample with standard samples since the mercury calibration curve is nonlinear. Calculation Proceed as directed in Calculations, page 54, except plot absorbance instead of percent transmission versus concentration. 4. References [1] Dean, J. A., and Rains, T. C, eds., Flame Emission and Atomic Absorption Spectrometry, Vol. 3_, Elements and Matrices, (Marcel Dekker, New York, 1975). [2] Dean, J. A., and Rains, T. C. , eds., Flame Emission and Atomic Absorption Spectrometry, Vol. 2_, Components and Techniques, (Marcel Dekker, New York, 1971). Chap 64 ter 13, Standard Solutions for Flame Spectrometry is reproduced as Appendix 14. [3] Dean, J. A., and Rains, T. C, eds . , Flame Emission and Atomic Absorption Spectrometry, Vol. 1_, Theory, (Marcel Dekker, New York, 1969). [4] Mavrodineanu, R. , ed. , Analytical Flame Spectroscopy, (Macmillan, London; Springer, New York, 1970). [5] Rains, T. C, Chemical Aspects of Atomic Absorption, ASTM STP 443,, pp. 19-36, American Society for Testing and Materials, 1969 (see Appendix 11). [6] Rains, T. C. , Atomic Absorption. Spectrometry - General Consideration for the Application of Experimental Tech- niques, ASTM STP 564, pp. 50-67, American Society for Testing and Materials, 1974 (see Appendix 12). [7] Snelleman, W. , Rains, T. C, Yee, K. W. , Cooks, H. D., and Menis, 0., Anal. Chem. , 4_2, 394 (1970) (see Appendix 13) . [8] Kuehner, E. C, Alvarez, R. , Paulsen, P. J., and Murphy, T. J., Anal. Chem., 44, 2050 (1972) (see Appendix 1) . [9] Scott, R. 0. and Ure, A. M. , Proc. Soc. Anal. Chem., December, 1972, pp. 288-293. [10] Christian, G. D. , and Feldman, F. J., Atomic Absorption Spectroscopy: Application in Agriculture, Biology, and Medicine, (Wiley- Interscience , New York, 1970). [11] Rains, T. C, and Menis, 0., JAOAC, 55_, 1339 (1972) (see Appendix 15). [12] Analytical Methods for Atomic Absorption Spectrometry, Perkin-Elmer, Norwalk, CT , March 1973, Reviewed September, 1974, and March, 1976. 65 B. Determination of Calcium, Potassium, and Sodium in Flour by Emission Flame Spectrometry. J. R. Baldwin and R. Mavrodineanu 1. Objective The purpose of the procedure described in this paper is to determine calcium, potassium, and sodium in flour using emission flame spectrometry. 2. Principle Emission flame spectrometry is based on the well known principle that chemical species, when properly energized, may re-emit the absorbed energy as radiations. These radiations are specific for every emitting species and their intensities are proportional to the number of particles, atoms, ions, molecules, or free radicals, par- ticipating in the radiation process. These two properties are at the foundation of qualitative and quantitative spectral analyses [1] In the present case, the excitation energy is supplied by a pre-mixed acetylene-air combustion flame into which the analytical sample is introduced, in a constant and reproducible manner, as a suspension of fine droplets using compressed air and a pneumatic sprayer. 3. Procedure Instrumentation The equipment needed to perform the analysis consists in principle of the analytical flame, with its sample supply- ing system, an adequate spectrometer followed by a photo- detector, and a data acquisition and presentation system. A detailed description of the equipment assembled and used for this purpose at NBS, is given in reference [3], which is re- produced in Appendix 16. A pre-mixed acetylene-air flame 66 with pneumatic sprayer was used in this work. Further infor- mation on the instrumentation and methodology used in general for Na, K, and Ca determinations by emission flame spectro- scopy will be found in reference [2] . 4. Reagents Distilled water Nitric acid Perchloric acid Lanthanum perchlorate, 10 mg La/ml, made by dissolving 11.727 g La 2 3 , reagent grade, in 83 ml of perchloric acid, diluted to 1 liter with distilled water. Sodium, potassium and calcium stock standard solutions, each containing 1000 yg/ml. These solutions were made by dissolving for: Sodium : 2.5447 g sodium chloride 99.9 percent pure, SRM 919, in 1000 ml distilled water to which 5 -ml nitric acid were added; Potassium : 1.9087 g potassium chloride 99.9 percent pure, SRM 918, in 1000 ml distilled water to which 5-ml nitric acid were added; Calcium : 2.4997 g calcium carbonate 99.9 percent pure, SRM 915, in 1000 ml distilled water to which 5 ml nitric acid has been added. This basic standard solution as well as the composite diluted solutions were kept in polyethylene bottles. Composite Standard Solutions: Suitable dilutions of the above stock solutions are made with distilled water as indicated below to produce the standards needed to calibrate the emission flame spectrom- eter : 67 Composite Solution Na yg/ml K Ca 1 0.35 50 8 2 0.50 80 10 3 0.70 100 16 4 1.00 120 24 These composite standard solutions were used in the deter- mination of Na and K. For the determination of Ca, a lanthanum solution was added to the same concentrations of Na, K, and Ca, to pro- duce solutions containing a final concentration of 1000 yg/ml La each. 5. Sample Preparation The weight loss at 85 °C is determined in separate 1-g flour samples taken in duplicate. Each sample is weighed separately into a glass weighing bottle provided with a cover. The bottles are placed uncovered in an oven at 85 °C for 20 hours, then placed in a desiccator. After cooling, they are weighed with the cover on. For the analysis, 5 g of flour sample is weighed in five replications at the same time the samples for weight loss are taken. The weighing is made to the nearest 0.1 mg . Each sample is transferred to a 250-ml Teflon beaker provided with a Teflon cover, and 10 ml of distilled water is added to each sample. The samples are treated with a mixture of 30 ml of HN0 3 and 5 ml of HC1CU on a hot plate until all organic matter is destroyed. During this time, HN0 3 is added, periodically, in 5 ml portions, as needed to remove all organic matter, and the walls of the beaker are washed periodically with a minimum of water. Finally, the walls of the beaker are washed for the last time, and the contents 68 of the beaker is evaporated to about 1 ml. This operation can take approximately one day, and depends on the type of the flour. To this residue, 10 ml of water are added, the solution is transferred in a 50-ml polypropylene volumetric flask, and diluted to mark with water. This solution is used for the determination of Na and K. Calcium is determined in a similar solution to which lanthanum was added to minimize the interfering action of phosphorous. Thus, 5 ml of the lanthanum stock solution are added to the sample to provide a final concentration of 1000 ug/ml La. A corresponding amount of lanthanum stock solution is added to each standard solution to provide a final concentration of 1000 ug/ml La. The spectrometer is calibrated in the usual manner, before and after measuring the sample solutions, and the o o measurements are made at 5889.9 A for sodium; 7664.9 A for o potassium; and 6214 A for calcium (CaOH molecular band maximum) . The samples are measured each in six replications and their content of Na, K, and Ca is found by interpolation on the standard calibration curves which are prepared by plotting as the abscissa the concentrations of the standard solutions (ug/ml) and the corresponding photocurrent values (volts on the digital voltmeter) as the ordinate. Calculation and Results The amount of element contained by the flour is expressed as ug/g and is referred to the material on an "as received" basis. An example is given in Table 1 for a wheat flour sample and in Table 2 for a rice flour including the percent loss at 85 °C. 69 Table 1. Determination of sodium, potassium, and calcium (yg/g) i n a wheat flour sample on an "as received" basis. Sample No. 1 2 3 4 5 Sodium Potassium Calcium 0.695 116 16.44 0.695 115 16.33 0.658 109 16.22 0.650 105 16.77 0.670 108 15.34 Average yg/g in Aliquot 0.674 110.6 16.22 Standard Error (SE) 0.01 2.1 0.24 Percent SE 1.4 1.9 1.5 Average yg/g . „ - 6.71 1102.0 161.6 in Sample Loss: Twenty hours at 85 °C; 8.6 percent. 70 Table 2. Determination of sodium,potassium, and calcium (yg/g) i- n a rice flour sample on an "as received" basis. Sample No. 1 2 3 4 5 Average yg/g in Aliquot Standard Error (SE) Average yg/g in Sample . odium Potassium Calcium 0.430 91 11.55 0.425 92 11.10 0.420 86 10.45 0.465 92 11.10 94 11.78 0.435 91 11.20 0.01 1.3 0.23 4.32 905.0 111.6 Loss: Twenty hours at 85 °C; 8.3 percent. 6. References [1] Mavrodineanu, R. , and Boiteux, H. , Flame Spectroscopy, (John Wiley, New York, 1965). [2] Mavrodineanu, R. , Bibliography on Flame Spectroscopy, Analytical Applications, 1800-1966. NBS Misc. Publ. 281, (1967), continued from 1966 to 1968 in Analytical Flame Spectroscopy, Selected Topics, R. Mavrodineanu, ed. (MacMillan - Springer, London, New York, 1970). [3] Mavrodineanu, R. , Discussion of Some Experimental and Fundamental Conditions in Analytical Flame Spectroscopy, Revue du GAMS (Paris) 1971 , No. 3, 39-60; reproduced in Appendix 16. 71 SECTION V MOLECULAR ABSORPTION SPECTROMETRY Procedures for the Determination of Arsenic, Copper, and Nickel by Molecular Absorption Spectrometry R. W. Burke and B. I. Diamondstone 1. Objective The procedures to be described have been used in the Analytical Chemistry Division for several years for the routine determination of trace concentrations of arsenic, copper, and nickel in a variety of tissue samples. Through appropriate selection of experimental conditions, analyses for these three elements can be readily performed at the 0.01 part-per-million (ppm) level. Immediately following are detailed descriptions of the procedures currently used and recommended for the determin- ation of these trace elements in biological matrices. To demonstrate the broad applicability of these methods, typical results are given for the analyses of some recently issued and pending Standard Reference Materials (SRM's). 2. Principle The determination of arsenic is based on the formation of a deep red color when arsine is reacted with a pyridine solution of silver diethyldithiocarbamate ; copper is deter- mined as the yellow complex of Cu(II) - dibenzyldithio- carbamate; and nickel is determined as a-furildioxime chelate. 3. Spectrophotometric Procedure for Arsenic Samples are wet-ashed in a mixture of high purity nitric and perchloric acids prepared at NBS (Appendix 1) . To eliminate volatilization losses all dissolutions are performed under reflux in a Bethge apparatus. The arsenic (V) formed 73 is reduced to arsine (AsH 3 ) in a special generator flask and the latter is absorbed in a pyridine solution of silver diethyldithiocarbamate . The deep red color which develops, and whose exact nature is not completely understood, is measured spectrophotometrically and compared to a series of known arsenic standards. 4. Apparatus Bethge Dissolution Apparatus. Available from G. Frederick Smith Chemical Company, Columbus, Ohio. 1 Arsine Generator/Absorption Tube. Commercially avail- able from Arthur H. Thomas or Fisher Scientific. Spectrophotometer. Any stable single or double beam instrument having reproducible cuvette positioning and a thermostatable sample compartment which will accept 10 mm rectangular cuvettes. Readability and reproducibility should be 0.001 absorbance or better. Cuvettes. Rectangular, semi-micro, thick-walled, of glass or quartz construction fitted with Teflon stoppers and having a light path of 10 mm and an internal width of 4 mm. Shaker-Water Bath. A Dubnoff type or its equivalent. 5. Reagents Silver Diethyldithiocarbamate, 0.5 percent. Dissolve 1 g of silver diethyldithiocarbamate in 200 ml of pyridine. Store in an amber bottle. Potassium Iodide, 15 percent. Dissolve 15 g of KI in 100 ml of distilled water and store in an amber bottle. In order to describe materials and experimental procedures adequately it was occasionally necessary to identify com- mercial products by manufacturer's name or label. In no instance does such identification imply endorsement by the National Bureau of Standards, nor does it imply that the particular product or equipment is necessarily the best available for that purpose. 74 Stannous Chloride, 21 percent. Dissolve 25 g of SnCl 2 *2H 2 in 100 ml of concentrated hydrochloric acid. Arsenic(V) Standard Stock Solution, 100 ppm. Dissolve 0.1320 g of As 2 3 in 10 ml of water containing 1 g of NaOH. Oxidize As(III) to As (V) by addition of 2 ml of saturated bromine water. Warm on a hot plate or steam bath to remove excess bromine. Cool and dilute to 1 liter with distilled water. This solution contains 100 yg As (V) /ml . Arsenic (V) Working Standard, 1 ppm. Prepare 1 yg As (V) /ml working standard by diluting 10 ml of the standard stock solution to 1 liter with distilled water. Prepare fresh weekly. Zinc. Use granulated zinc that is 20 mesh or less and whose arsenic content does not exceed 0.02 ppm. 6. Analysis Procedure Dissolution. Depending upon the expected arsenic con- tent, weigh a 1 to 5 g sample of dried 2 tissue and transfer to the round bottom flask of the Bethge apparatus. Add 30 ml of a 1:1 mixture of high purity nitric and perchloric acid and connect the flask to the reflux condenser. With cold tap water running through the condenser, gradually raise the temperature of the flask and contents to 120 °C by use of a variable transformer to control the power supplied to the heating mantle. The stopcock should remain open and the temperature maintained near 120 °C until all of the sample has dissolved. Close the stopcock and complete the oxidation by slowly raising the temperature to 210 °C. Initially no increase in transformer power will be necessary since the temperature will rise as the nitric acid distills up the side arm and condenses above the stopcock. Additional heat may be needed near the end of the oxidation to attain 2 See Section 1, Sample Preparation, page 5 for recommended drying procedures. 75 the 210 °C reading. Normally, 15 to 20 minutes are required to go from the 120 °C setting to 210 °C. At the latter temperature, the sample should be clear and exhibit a yellow or pale orange color. Lower the heating mantle, allow the flask and contents to cool and drain the nitric acid back into the sample solution. Wash down the inside of the Bethge apparatus with distilled water and transfer the contents of the flask to a 100-ml Teflon beaker. If silica is present add 0.5 to 1 ml of hydrofluoric acid. Evaporate to strong fumes of perchloric acid. Cool, add 5 ml of distilled water and evaporate to approximately 2 ml . If aliquots are to be taken for arsenic analysis, transfer the sample to a 25-ml volumetric flask and dilute to volume with distilled water. Arsine Generation and Collection. Transfer the entire sample or a suitable aliquot (As ^0.1 ]ig) to the generator flask and add sufficient concentrated HC1 to give a total acid volume of 10 ml. Add 35 ml of water, 2 ml of 15 per- cent KI solution and 1 ml of 21 percent SnCl2 solution. Mix and allow to stand at room temperature for 15 minutes in order to reduce As (V) to As(III). Cool the assembled generator in an ice bath for 30 minutes and transfer to a 25 °C shaker bath. Add 4.00 ml (Note A) of the pyridine solution of silver diethyldithiocarbamate to the collection tube. Disconnect the collector tube and add 2.5 g of zinc to the generator flask, using a wide-mouth funnel to keep zinc off the walls of the ground glass joint. Immediately connect the collection tube to the flask by clamping the ball joint together. Allow one hour for the evolution of Note A. For determining arsenic in the range of 0.1 to 0.5 uj the amount of absorber solution used is reduced from 4.00 to 2.00 ml. In the latter instance, glass beads approximately 1 mm in diameter are used to produce the same length of col- umn of absorber solution as was present in the normal run. 76 arsine to be completed. To avoid leaks, both the standard taper joint of the flask and the ball joint should be lubricated. "Nonaq" stopcock grease, available from Fisher Scientific, has proven satisfactory for this purpose. Disconnect the collection tube and, through either a rocking motion or by use of a disposable Pasteur-type pipette, rinse the walls of the tube with the absorber solution. Any residue must be redissolved and the pyridine solution thoroughly mixed. Transfer an aliquot to the 10-mm cuvette and measure the absorbance at 535 nm against a reagent blank prepared in the same way as the sample. Determine the amount of arsenic present by comparing the measured absorbance with the absorbances of a series of arsenic standards (0.5, 1, 2, 4, and 6 yg As/4.00 ml pyridine reagent or 0.1, 0.2, 0.3, 0.4 and 0.5 yg As/2.00 ml pyridine reagent - see Note A). Both systems obey Beer's law and the slopes of the two calibration curves are in the exact ratio of 1:2. The amount of arsenic present in the tissue sample is given by the relationship: A _ • yg Arsenic Found Arsenic, ppm = ^^ * t^- * ^ r Sample Wt . , g Sample Absorbance, A ' PP Slope of Calibration Curve (A /yg) x Sample Wt . , g 7. Results The spectrophotometric arsine-silver diethyldithio- carbamate procedure for arsenic has been applied to the analysis of a number of NBS SRM's. Some typical results are given in Table 1. 77 Table 1. Spectrophotometric determination of arsenic in tissue samples by the arsine-silver diethyldithiocarbamate procedure . Sample NBS SRM No Orchard Leaves 1571 Wheat Flour Rice Flour Spinach Pine Needles Tomato Leaves 1567 1861 1570 1575 1573 Sample Wt . , g Arsenic, pp 0.5 10.2; 10.3; 9.9; 10.1 5.0 0.03; 0.02; 0.03; 0.03; 0.03; 0.02 5.0 0.09; 0.10; 0.10; 0.08; 0.08; 0.09 2.0 0.17; 0.18; 0.16; 0.18 2.0 0.22; 0.20; 0.20; 0.22 2.0 0.30; 0.29; 0.28; 0.29 Currently available from the NBS Office of Standard Reference Materials. In preparation. Remarks Antimony is the only potential interference in the spectrophotometric determination of arsenic in tissue by the arsine-silver diethyldithiocarbamate procedure. Upon reduc- tion with zinc, stibene (SbH 3 ) is evolved which reacts simi- larly to arsine (AsH 3 ) when absorbed in a pyridine solution of silver diethyldithiocarbamate. Fortunately, the absorb- ance maxima of the two products are sufficiently different (X„„ for Sb = 505 nm; X for As = 535 nm) and the pres- max ' max J ^ ence of antimony can be checked by measurement of the 505/535 absorbance ratio. If only arsenic is present, this 78 ratio should not exceed 0.80. If this ratio is exceeded, simultaneous equations can be used to correct for antimony- interference since the absorbance of the colored arsenic and antimony products are additive at these two wavelengths. No corrections were necessary for any of the tissue samples analyzed to date. 8. Spectrophotometric Procedure for Copper This procedure is based on the spectrophotometric measurement of the yellow complex formed in moderately strong acidic media between copper(II) and dibenzyldithio- carbamate after its isolation and preconcentration by extrac- tion into a non-polar organic solvent. The selectivity and acid stability of the carbamate reagent is enhanced by using it in the form of its zinc chelate which is dissolved directly in the carbon tetrachloride extracting solvent. When this solution is equilibrated with an aqueous solution containing copper(II), the corresponding copper complex is formed and extracted almost instantaneously as a result of its chelate exchange reaction with zinc. 9. Apparatus Spectrophotometer. Any stable single or double beam instrument having reproducible cuvette positioning and a sample compartment provided with temperature control which will accept 10 mm rectangular cuvettes. Readability and reproducibility should be 0.001 absorbance or better. Cuvettes. Rectangular, semi-micro, thick-walled, of glass or quartz construction, fitted with Teflon stoppers and having a light path of 10 mm and an internal width of 4 mm. Separatory Funnels. 60-ml capacity with Teflon stop- pers and stopcocks. Stems are shortened to approximately 10 mm and the ends fire-polished. 79 10. Reagents Zinc Dibenzyldithiocarbamate , 0.01 percent. Dissolve 20 mg of zinc dibenzyldithiocarbamate in 200 ml of carbon tetrachloride. Store in an amber bottle. Copper Standard Stock Solution, 100 ppm. Dissolve 0.100 g of copper metal in 5 ml of 1:1 nitric acid. Evaporate to dryness and dissolve residue in IN hydrochloric acid. Transfer to a 1 liter volumetric flask and dilute to the mark with IN hydrochloric acid. Copper Working Standard, 2 ppm. Dilute 5.00 ml of the standard copper stock solution to the mark with . IN hydro- chloric acid in a calibrated 250-ml volumetric flask. 11. Analysis Procedure Wet-ash 1 to 5 g of dried tissue in a 1:1 mixture of nitric and perchloric acids. Use 15 ml of the mixed acids for a 1-g sample and increase proportionately to 30 ml for 1 to 5-g of tissue. Dissolution may be carried out in a Bethge apparatus, using the conditions previously described for arsenic, or in a Teflon beaker. If a beaker is used, it should be covered with a Teflon lid until all reaction with the nitric acid has ceased. The cover is removed and rinsed with distilled water and the solution is heated to fumes of perchloric acid. Fuming is continued until approximately one-half of the perchloric acid is removed. The sample is cooled, the walls of the beaker are rinsed with distilled water and the fuming is repeated. When the sample volume has been reduced to 1 to 2 ml the beaker is removed from heat, cooled and reserved for copper analysis. If the total sample is to be used, it is transferred directly to a 60 ml separatory funnel and diluted to 25 ml with distilled water. If an aliquot is taken, the sample is first transferred to a 25-ml volumetric flask. In either instance the sample weight or aliquot size should be chosen so as to contain 1 80 to 10 yg of copper and a final acid concentration of 0.1 to IN. Add 5.00 ml of the carbon tetrachloride solution of zinc dibenzyldithiocarbamate and extract the copper by shaking vigorously for one minute. While the phases are clearing, insert a loosely-fitting paper plug in the stem of the separatory funnel. Drain a portion of the carbon tetra- chloride solution containing the yellow copper complex into a 10-mm cuvette and measure the absorbance of this solution at 435 nm against a reagent blank carried through the entire procedure. Determine the amount of copper present by comparing the net absorbance with the absorbances of known amounts of copper taken through the extraction procedure. Beer's law is obeyed over the range 1 to 10 yg of copper. When absorbance is plotted as a function of micrograms of copper added, the slope of the resulting calibration curve should be approximately 0.051 absorbance units per microgram of copper. 12. Results The results obtained for copper in a series of new NBS botanical SRM's are summarized in Table 2. All values are reported on a dried basis, i.e., samples were heated in an air oven for one hour at 85 °C. Typical weight losses were of the order of five percent. Remarks Bismuth and thallium(III) are potential interferences in the spectrophotometric zinc dibenzyldithiocarbamate pro- cedure for copper. However, neither of these elements has been found in sufficiently high concentrations to cause difficulty in any of the tissue samples analyzed. 81 Table 2. Spectre-photometric determination of copper in three new NBS botanical SRM's. Sample NBS SRM No. Sample Wt . , g Copper, ppm Spinach 1570 0.5 12.03; 12.08 12.07 Pine Needles 1575 1.0 2.94; 2.93; 2.95 Tomato Leaves 1573 0.5 10.81; 10.79 10.83 Available from the NBS Office of Standard Reference Materials . 13. Spectrophotometric Procedure for Nickel The preferred reagent for the spectrophotometric deter- mination of nickel in tissues is a-furildioxime. It is superior to dimethylglyoxime because the wavelength of maximum absorbance of its nickel chelate lies in the visible region of the spectrum. Nickel forms, with an excess of a-furildioxime, an insoluble chelate which can be quantitatively extracted [1] from aqueous media into chloroform in the pH range of 8.5 to 9.4. This extraction can be made very selective because the interference of cobalt and copper can be eliminated by backwashing the chloroform extract with ammonia. Citrate or tartrate is used to prevent the hydrolysis and subsequent interference of such common elements as iron and aluminum. 14. Apparatus Spectrophotometer. Any stable single or double beam instrument having a sample compartment provided with temperature control which will accept 50-mm cuvettes. Readability and reproducibility should be 0.001 absorbance or better. 82 Cuvettes. 50-mm cylindrical or rectangular cuvettes having a maximum volume of 15 ml and equipped with snugly- fitting Teflon stoppers or covers. Separatory Funnels. 60-ml capacity and fitted with Teflon stoppers and stopcocks. Stems are shortened to approximately 10 mm and fire-polished. pH Meter. Equipped with individual or a combination glass-calomel electrode. 15. Reagents a-Furildioxime, 1 percent. Dissolve 1 g of a-furil- dioxime in 100 ml of ethanol. Nickel Standard Stock Solution, 100 ppm. Dissolve 0.100 g of nickel metal in 5 ml of 1:1 nitric acid. Evaporate to dryness and dissolve residue in . IN hydrochloric acid. Transfer to a 1-liter volumetric flask and dilute to the mark with . IN hydrochloric acid. Nickel Working Standard, 1 ppm. Dilute 2.00 ml of the standard nickel stock solution to the mark with . IN hydro- chloric acid in a calibrated 200-ml volumetric flask. Ammonium Citrate, 10 g/100 ml. Dissolve 10 g of ammonium citrate in distilled water and dilute to 100 ml. 16. Analysis Procedure Wet-ash, in a mixture of nitric and perchloric acids, a sufficiently large sample to give a minimum nickel content of 0.5 yg (see copper procedure for details). Dilute the acidic sample, or an appropriate aliquot, to 20 ml with dis- tilled water and add, with mixing, 10 ml of ammonium citrate solution and 1.0 ml of alcoholic 0.1 percent a-furildioxime reagent. Adjust pH to 9.0 ± 0.3 with 1:1 NH^OH and extract with 10.00 ml of chloroform. Repeat the extraction with a second 10.00 ml portion of chloroform and backwash the combined extracts with 10 ml of 1:40 NHi+OH. Place a loosely- fitting paper plug in the shortened stem of the separatory 83 funnel and drain the chloroform extract into a 50-mm cuvette. Measure its absorbance at 435 nm against a reagent blank carried through the entire dissolution and extraction procedures. Determine the micrograms of nickel present from a calibration curve prepared from 0.50, 1.00, 2.00, 3.00 and 4.00 yg nickel standards. 17. Results Table 3 shows some typical results for the determination of nickel in Orchard Leaves, NBS SRM 1571. Table 3. Spectrophotometric determination of nickel in Orchard Leaves, NBS SRM 1571. Sample Wt., g Nickel, ppm 2 1.35; 1.41; 1.37; 1.33 Remarks The use of the backwash step in the extraction procedure effectively removes the interference of copper and cobalt. Tests show that at least 50 yg of copper and 2 yg of cobalt can be tolerated in the determination of 1 \ig of nickel. 18. Reference [1] Peshkova, V. M. , Bochkova, V. M. , and Lazareva, L. I., Spectrophotometric Determination of Traces of Nickel in Pure Indium and Aluminum, J. Anal. Chem. USSR (Eng. Trans.) 15, 699-702 (1960). Original Article, Zhur. Analit. Khim. 15. 610-613 (1960). 84 SECTION VI FLUORESCENCE SPECTROMETRY Determination of Beryllium by Fluorescence Spectrometry S. A. Wicks and R. W. Burke 1. Objective The procedure described in this section is used in the Analytical Chemistry Division to determine beryllium in tissue samples at the submicrogram level by fluorescence spectrometry. 2. Principle The classical procedure for the determination of beryl- lium by fluorescence spectrometry is based on its reaction with morin. This method enjoys considerable popularity because of its high sensitivity. However, it suffers from two disadvantages: (a) the reagent is unstable, being particularly prone to oxidation even by atmospheric oxygen, and (b) the method lacks selectivity. These disadvantages necessitate the use of a carefully controlled, and hence tedious, procedure in order to obtain satisfactory results. In our experience, a more useful method for determining submicrogram amounts of beryllium, especially in tissue samples, utilizes the fluorescence of the beryllium 2-hydroxy-3-naphthoic acid complex in water near pH 7.5. This method has been described by Kirkbright, West and Woodward [1] . We have examined this method in detail and have subsequently made three modifications which significantly improve its practicality . (a) The need for silicone-coating the flasks to prevent absorption losses of beryllium has been eliminated by developing the complex under weakly acidic media (pH 2 to 3). The pH is then increased to 7.5 and the fluorescence measured at the end of five minutes. 85 (b) Phenol Red is used as an internal visual indicator for adjusting pH and (c) Preseparation of beryllium by extraction with acetylacetone [2] is used to improve selectivity and sensitivity. 3. Apparatus Spectrof luorometer . Equipped with stabilized xenon source, variable slits and excitation and emission mono- chromators . 4. Reagents Beryllium Stock Solution, 100 ppm. Dissolve 0.100 g of beryllium metal in 10 ml of 1:2 hydrochloric acid. Add 10 ml of perchloric acid and evaporate to HClOit fumes. Cool, transfer to a 1-liter volumetric flask and dilute to the mark with distilled water. Beryllium Working Standard A, 1 ppm. To 10 ml of the beryllium stock solution, add 10 ml of perchloric acid and dilute to the mark in a 1-liter volumetric flask with dis- tilled water. Prepare fresh daily. Beryllium Working Standard B, 0.1 ppm. Dilute 10 ml of beryllium standard A and 1 ml of perchloric acid to the mark, in a 100 ml volumetric, with distilled water. Prepare fresh daily. 2-Hydroxy-3-Naphthoic Acid, 2 x 10 ^M. Dissolve 38 mg of 2-hydroxy-3-naphthoic acid in 5 ml of 1:10 ammonium hydroxide, transfer to a 1-liter volumetric flask and dilute to the mark with distilled water. Masking Agent, Calcium-CDTA (cyclohexanediamine- tetraacetic acid). Prepare the following solutions: A. CaCl 2 , 0.1M. Dissolve 7.35 g of CaCl 2 -2H 2 in 500 ml of distilled water. 86 B. CDTA, 0.1M. Dissolve 3.46 g CDTA in 90 ml of distilled water by addition of 1:10 ammonium hydroxide. Adjust to pH 7.5 and dilute to 100 ml with distilled water. Combine 25 ml of . 1M calcium solution and 6 ml of . 1M CDTA in a 100-ml volumetric flask and dilute to volume with distilled water. Buffer Solution, pH 7.5. Dilute 50 ml of concentrated ammonium hydroxide to 300 ml with distilled water. Add sufficient glacial acetic acid to adjust pH to 7.5 and dilute to 1 liter with distilled water. Ethylenediaminetetraacetic Acid, Disodium Salt Solution (EDTA) , 10 Percent. Dissolve 10 g of the disodium salt in 100 ml of distilled water. Acetylacetone, 5 Percent. Transfer 5 ml of acetylace- tone to a 100 ml volumetric flask and dilute to the mark with distilled water. Phenol Red Indicator, 0.1 Percent. Dissolve 100 mg of Phenol Red in 100 ml of distilled water. Quinine Sulfate Standard, 5 ppm. Dissolve 5 mg of quinine sulfate dihydrate in 1 liter of . 5M sulfuric acid. 5. Analysis Procedure Wet-ash 1 to 5 g of dried tissue in a 1:1 mixture of nitric and perchloric acid in a platinum dish. Use 10 ml of the acid mixture for a 1-g sample and increase proportionately to 30 ml for sample weights in the range 1 to 5 g . If silica is present, add 1 ml of hydrofluoric acid. Cover the dish with a platinum lid and digest on a hot plate until oxides of nitrogen are no longer evolved. Cool, remove and wash the cover and the sides of the dish with 10 ml of distilled water. Heat to fumes of perchloric acid and continue fuming until volume has been reduced to approximately 1 ml. Dilute to 50 ml and add 2.0 ml of 10 percent EDTA 87 solution. Add 2 to 3 drops of Phenol Red and adjust pH to 7 to 8 with 1:10 NHi+OH (until indicator just turns pink). Add 5 ml of five percent acetylacetone and readjust pH to 7 to 8. Allow the solution to stand for five minutes and then extract it with three 5-ml portions of chloroform. Combine the extracts in a platinum dish and cover with 10 ml of water, 2 ml of concentrated nitric acid and 2 ml of concen- trated perchloric acid. Evaporate off the chloroform on a steam or water bath, transfer the dish to a hotplate and continue the evaporation to dryness. Add 2 ml each of nitric and perchloric acid and repeat evaporation to dryness. Dissolve the residue in 10 ml of . IN perchloric acid and transfer to a 100-ml volumetric flask. Dilute to approximately 70 ml with distilled water and add 5.0 ml of the calcium CDTA solution and 2 to 3 drops of Phenol Red solution. Adjust to pH 7 to 8 with 6N NH^OH. Add 5.0 ml of pH 7.5 buffer and 5.0 ml of 2-hydroxy-3-naphthoic acid reagent. Dilute to volume with distilled water and mix. Transfer to a 10-mm cuvette and, after exactly five minutes, excite the beryllium complex at 373 nm and measure its fluorescence at 460 nm. Use a 0.5 ppm quinine sulfate solution in . IN sulfuric acid to correct for short term fluctuations in instrument sensitivity. This solution should be prepared fresh daily by ten-fold dilution of the stock quinine sulfate with distilled water. Determine the amount of beryllium present by comparing the measured fluorescence with that of a series of beryllium standards covering the range 0.05-0. 5 yg Be. 6. Results Table 1 summarizes the results obtained for beryllium in three SRM's which are currently available through the NBS Office of Standard Reference Materials. Table 1. Beryllium content of three SRM's as determined by the acetylacetone extraction- 2-hydroxy-3-naphthoic acid spectrof luorometric method. Sample Orchard Leaves Bovine Liver Coal SRM No. 1571 1577 1632 Beryllium, ppm 0.026; 0.025; 0.025; 0.026 0.021; 0.013; 0.020; 0.015 1.52; 1.46; 1.50 Remarks Some preliminary experiments carried out during the analysis of the orchard leaves suggest strongly that the preparative wet oxidation should be performed in platinum rather than Teflon. When 100-ng spike additions were made to 1-g samples of orchard leaves and the dissolution sub- sequently performed in 100 ml Teflon beakers, the recovery of the added beryllium was low and non-reproducible, usually in the range of 10 to 40 percent. The reason for this behavior is not understood at present. Recoveries from borosilicate glass container and platinum, however, were much superior to those observed from Teflon. Spike recoveries from platinum through the entire analysis procedure were 95 percent or better. 7. References [1] Kirkbright, G. F. , West, T. S. and Woodward, C, Spectrof luorometric Determination of Submicrogram Amounts of Aluminum and Beryllium with 2-Hydroxy- 3-Naphthoic Acid, Anal. Chem. 3_7, 137-143 (1965). [2] Adam, J. A., Booth, E. and Strickland, J. D. H. , The Determination of Microgram Amounts of Beryllium Using Acetylacetone, Anal. Chim. Acta 6, 462-471 (1952). 89 SECTION VII PO LARO GRAPH Y Determination of Aluminum, Bismuth, Cadmium, Copper, Iron, Lead, and Nickel. Analysis of Botanical Standard Reference Materials by Cathode Ray Polarography 1 E. June Maienthal 1. Objective The procedure described in this section was developed and applied to the quantitative determination of traces of aluminum, bismuth, cadmium, copper, iron, lead, and nickel in botanical material by cathode ray polarography. 2. Principle The principle of polarography is based on the unique characterization of the voltage - current curves, or polaro- grams , produced when a microelectrode is immersed in an electrochemically active solution. The abscissa, or voltage parameter, of the polarogram identifies the reducible or oxidizable chemical element present in a complex solution. The ordinate, or current parameter of the polarogram gives an information on the quantity of a particular element present in the solution. The method is specific and sensitive, and requires only small volumes of solution for the analysis. The material used in this section was taken from a work published by the author in the Journal of the Association of Official Analytical Chemistry, Vol. 55, No. 5, p. 1109 to 1113 (1972). 91 3. Procedure Instrumentation: The Davis Differential Cathode Ray polarograph was used in performing the measurements described in this section. Reagents: All solution are aqueous unless otherwise specified. (a) Cupferron solution — 6 percent. (b) Sodium acetate solution — 2M. (c) Sodium acetate (2M) + EDTA (disodium salt) solution (0.1M) . (d) Solochrome Violet RS solution — 0.1 percent. (e) Sodium tartrate solution — 20 percent. (f) Potassium cyanide solution — 20 percent. (g) Sodium diethyldithiocarbamate solution — 0.1 percent (h) Sodium citrate solution — 10 percent. (i) Hydroxylamine HC1 solution — 20 percent. (j) Dimethylglyoxime solution — one percent in alcohol. Iron and Aluminum, Acid-Soluble. Transfer 250 mg sample which has been dried 24 hours at 90 °C to 50 ml Erlenmeyer flask. Add 5 ml HN0 3 , 1 ml HC1, and 1 ml per- chloric acid; cover and heat cautiously until vigorous frothing ceases. Treat blank and standard solutions for the calibration curve similarly. Digest and fume to perchloric fumes, adding more HNO3 if some organic material remains, and evaporate to dryness. Dissolve residue in 1 ml HC1 and evaporate to dryness at <100 °C. Dissolve residue in 1 ml perchloric acid (1 + 1) , transfer to 125 ml separatory funnel and dilute to ca 20-25 ml. Add 1 ml of six percent cupferron solution which has been filtered and cooled in ice and extract with four 10 ml portions of cold CHCI3, adding a few drops of cupferron before each extraction. (If a reddish brown precipitate is formed more cupferron is needed.) Reserve combined organic extracts for determination of iron. Add 10 ml 2M sodium acetate solution to aqueous layer, add 92 1 ml six percent cupferron solution, wait 1-2 minutes and extract with 15, 10, and 5 ml portions of CHC1 3 . Reserve combined organic extracts for determination of aluminum. (a) Iron. Let CHC1 3 extract evaporate to dryness in stream of filtered air and destroy residual organic material by evaporation to dryness with a few ml HN0 3 and 1 ml perchloric acid. Add 1 ml HC1 and evaporate to dryness at <100 °C. Dissolve residue in 0.4 ml H2SO4 (1+1); then add 5 ml 2M sodium acetate +0.1M EDTA . (disodium salt) solution and adjust pH to ca 4 with NH^OH. Dilute to 10.0 ml. Transfer portion of solution to polarographic cell, deaerate with nitrogen, and measure iron reduction peak subtractively with differential cathode ray polarograph at ca —0.2V vs. mercury pool anode, using background blank in second cell. (b) Aluminum. Remove CHCI3, destroy organic material, and evaporate with HC1 as described above for iron. Dissolve residue in 1 ml perchloric acid (1+1) and 5 ml of 2M sodium acetate solution. Add 5 ml 0.1 percent Solochrome Violet RS solution. Dilute to 50.0 ml and transfer to polyethylene bottle fitted with cap. Heat in water bath five minutes at 55-70 °C, cool, transfer to polarographic cell, deaerate, and measure aluminum-dye complex reduction peak with cathode ray polarograph at ca -0.65V vs mercury pool anode. For aluminum, better peak resolution is obtained if blank signal is measured separately and then subtracted from that of s amp 1 e . Iron and Aluminum— Total . Proceed as above except filter the solution remaining after sample dissolution through Whatman No. 42 paper. Wash with HC1 (1+9). Transfer residue to platinum crucible, dry, and ignite over burner. Add two drops perchloric acid and 10 drops hydrofluoric acid, and evaporate to dryness. Fuse residue in small amount of sodium carbonate. Add water and then HC1 dropwise to neutralize sodium carbonate and transfer solution to 93 reserved filtrate which has been evaporated to dryness. Proceed as described above for determination of iron and aluminum. Lead and Bismuth. Wet-ash 1 g samples of dried material and evaporate as already described, except that ca 5-10 ml more of HN0 3 may be necessary to destroy all organic material. Treat blank and standard solutions for calibration curve similarly. Dissolve residue in 10 drops HC1 and ca 10 ml water. Add ca 100 mg ascorbic acid to reduce iron and transfer solution to 125 ml separatory funnel. Add 10 ml 20 percent sodium tartrate solution, 30 ml HN 4 0H, 10 ml 20 percent potassium cyanide solution, and 10 ml 0.1 percent sodium diethyldithiocarbamate . Extract lead and bismuth complexes with 15, 10, and 5 ml portions of CHCI3. Evaporate CHCI3 extracts and destroy organic material as already described. Dissolve residues in 0.5 ml HN0 3 (5+95) and 2.0 ml 20 percent sodium tartrate solution. Deaerate solution and measure polarographically as already described. The bismuth reduction peak occurs at ca -0.5V and lead at ca -0.7V vs. mercury pool anode. Nickel. Wet-ash 1 g samples of dried material and evaporate as described for lead and bismuth. Treat blank and standard solutions for calibration curve similarly. Transfer solution to 30 ml separatory funnel with water, keeping final volume <10 ml. Add 1 ml 10 percent sodium citrate solution, 2 ml 20 percent hydroxylamine HC1, and 1 ml one percent dimethylglyoxime in alcohol. Make solution ammoniacal to methyl red plus three drops excess. Extract nickel complex with 15, 10, and 5 ml portions of CHC1 3 . Evaporate CHC1 3 and destroy organic material as already described. Dissolve residue in 0.2 ml H 2 SCU (1+1) and a small amount of water. Add 1 ml pyridine, dilute solution to 5.0 ml, and measure nickel polarographically as already described at ca -0.9V vs. mercury pool anode. 94 4. Discussion The need for certified botanical standards both in research and in industrial and field applications is leading to the development and production of a number of new types of Standard Reference Materials by the National Bureau of Standards. Workers in the botanical and agricultural areas have long needed standards to assist in trace element deter- minations in leaves and plants. These standards would also be useful in the analysis of environmental samples such as air particulates and contaminated vegetation, and for calibra tion of much of the automated equipment now being used in these fields. Efforts have first been directed toward orchard, alfalfa, citrus, and tomato leaves, aspen chips, and pine needles. The first of these botanical standards to be prepared, analyzed, and certified was Orchard Leaves SRM 1571 [1]. This standard was prepared from fruit leaves, mostly peach, with some apple and cherry leaves, gathered from an orchard near Lansing, Michigan. In order to certify any given element in a Standard Reference Material, two completely independent methods or an accepted referee type method by two different labora- tories must be used. The methods must be both sensitive and accurate. The most frequently used methods of analysis by workers in botanical areas have included spectrophotometry, atomic absorption, emission spectroscopy, and classical wet chemi- cal methods. Polarography has been utilized very little. Some of the polarographic methods have been described by Nangniot [2,3] . It is difficult to understand why polarography has not been more widely used when the sensitivity, accuracy, and resolution of the technique are considered. Differential cathode ray polarography has been applied successfully in 95 this laboratory for the past few years for the determination of about 30 elements in many different matrices such as steels and alloys [4-10], food [9,10], environmental samples [5,7,9-11], and high purity reagents [6,7,9,10] covering concentration ranges from 3 x 10" 9 to 100 percent. The reliability of these polarographic determinations has been verified through interlaboratory comparisons with many other techniques . The proposed scheme was first to determine polarographi- cally iron and aluminum in the orchard leaves; later lead, bismuth, copper, nickel, and cadmium were included. Typical fruit leaf compositions range from about 100 to 300 ppm for both iron and aluminum, so sensitivity would be no problem. Previous work here has resulted in the deter- minations of as little as 0.1 yg iron or aluminum. In these cases the limitation on sensitivity was not instrumental but rather due to the blank coming from the reagents and the environment. For the iron and aluminum levels in the orchard leaves, the blank would probably be low enough with respect to the samples to be of no consequence. Iron could be determined polarographically with no separations, but aluminum would have to be separated. Cup- ferron extraction at pH 4.5 has been used often here for separation of aluminum from the matrix [6] . This involves a preliminary extraction of iron cupferrate from an acid solution, so it seemed advantageous to reserve the extracted iron for its subsequent polarographic determination. The best supporting electrolyte for the polarographic determination of aluminum is obtained by forming an aluminum- dye complex with Superchrome Garnet Y [8,12] or Solochrome Violet RS [12-15]. Iron could be determined with fewest interferences in an EDTA-sodium acetate supporting electrolyte. Prior to the extractions, the organic material must first be removed by wet ashing. After this step a small amount of insoluble siliceous residue, which undoubtedly 96 contained iron and aluminum, was always present; bismuth, cadmium, copper, lead, and nickel are not present in measurable amounts in this type of residue. This was probably the result of external sand or soil contamination of the fruit leaves and its content would have no relation to the actual leaf composition. Some nondestructive tech- niques, however, would only determine the total iron or aluminum, so it seemed advisable to determine separately the acid-soluble portion and then the total iron and aluminum on a separate sample by treating the residue with hydrofluoric acid, fusing it with sodium carbonate, and adding the fused residue to the original solution. For lead and bismuth, a simplification of a procedure for their simultaneous determination, used in this laboratory for steels and cast iron [5,16], appeared feasible, since the orchard leaf matrix is much simpler for polarography than that of steels. For the determination of nickel, a polarographic method developed for its determination as a trace element in Glass SRM 614 [1,10] appeared to be equally feasible, since the nickel in the orchard leaves was expected to be at about the 1 ppm level. The procedure described in this section was also used to determine iron, nickel, lead, and cadmium in tomato leaves and spinach Standard Reference Materials. 5. Results The results for acid-soluble iron are shown in Table 1. The results for iron on the first set were not reported because of the extreme variation. The values on this set ranged from 195 to 359 ppm which is a far greater variation than has ever occurred before in this laboratory and could only be ascribed to gross and erratic environmental con- tamination or sample inhomogeneity . The normal blank for this procedure was in the vicinity of 0.3 ppm. The overall 97 average value of 262 ppm is in good agreement with a value of 278 ppm later obtained by isotope-dilution, spark-source mass spectrometry [R. Alvarez, P. J. Paulsen, and C. W. Mueller (1970)], and the recommended value which appears on the provisional certificate [1] is 270 ppm. Table 1. Polarographic determination of iron (acid soluble) in orchard leaves. Fe, ppm - - - S amp 1 e Set II Set III 2 260 258 9 257 282 10 263 272 11 257 259 12 257 255 13 254 253 15 258 282 19 262 267 Average 259 256 s , ppm 3.0 11.7 Pooled average 262 Average s 9.1 a HCl-HNOs-HClCU. The results obtained for aluminum on the same samples are shown in Table 2. No other values for acid-soluble aluminum have been reported, hence this value has not been certified. 98 Table 2. Polarographic determination of aluminum (acid-soluble) a in orchard leaves. - Al , ppm - Sample Set I Set II Set III 2 226 lost 218 9 226 222 211 10 228 241 224 11 225 223 225 12 220 226 204 13 225 229 222 15 223 226 222 19 230 228 225 Average 225 228 219 s , ppm 3.0 6.3 7.6 Pooled average 224 Average s 6.7 a HCl-HNOa-HClO.,. The results for total iron determined on new samples are shown in Table 3. From comparison with the acid-soluble iron value, it is seen that the residue contains about 50 ppm of the total sample equivalent of iron. Although isotope- dilution spark-source mass spectrometry analysis was not used for total iron, the insoluble residue was examined by semiquantitative emission spectroscopy [V. C. Stewart (1970)] and was found to contain iron in the 0.1-1 percent range (estimated to be about 0.5%). This corresponds to about 25 ppm in the total sample, which, added to the mass spectrometric value of 278 ppm, gives a total iron value of 303 ppm, in good agreement with the polarographic value of 312 ppm. 99 Table 3. Polarographic determination of iron (total) in orchard leaves. - - - Fe, ppm - - - Set II Set III 322 297 323 314 336 295 317 310 320 310 305 295 318 317 316 300 320 305 8.7 9.0 312 11.4 a HCl-HNOs-HClO^-HF, Na 2 C0 3 fusion. The total aluminum was then measured and the results are shown in Table 4. From comparison with the acid-soluble aluminum value of 224 ppm (Table 2) , it is seen that the residue contains 123 ppm of the total sample equivalent of aluminum. The results obtained so far are not in good agreement and aluminum has not been certified. A value of 420 ppm was obtained by activation analysis [T. E. Gills (1970)]. In a round-robin study [P. D. LaFleur and T. W. Mears (1971)], values ranging from 99 to 401 ppm were reported by 11 non-NBS optical emission spectroscopy laboratories . Sample 2 9 10 11 12 13 15 19 Average s , ppm Pooled average Average s 100 Table 4. Polarographic determination of aluminum a (total) in orchard leaves. - - - Al , ppm - - Sample Set II Set III 2 359 344 9 340 349 10 340 341 11 340 348 12 345 348 13 341 350 15 346 350 19 350 368 Average 345 350 s , ppm 6.7 8.0 Pooled < average 347 Average s 7 5 a HCI-HNO3-HCICU-HF, Na 2 C0 3 . The results for lead and bismuth are shown in Table 5. The lead average of 44.6 ppm was considerably higher than the anticipated value of about 5 to 10 ppm; however these results were later confirmed by two other techniques. A value of 43 ppm was obtained, using isotope-dilution, spark- source mass spectrometry, and a value of 45 ppm, using photon activation analysis [G. J. Lutz (1971)]. Another set was analyzed by the polarographic procedure, utilizing the hydrofluoric acid treatment, in order to verify that no lead was in the siliceous residue. An average of 44 ppm was obtained, thus the residue contained an insignificant amount of lead. 101 Table 5. Polarographic determination of bismuth and lead in orchard leaves. Sample Bi, ppm Pb , ppm 2 0.11, 0.10 44.2, 41.8 9 0.10 46.3 10 0.13, 0.15 44.2, 44.8 11 0.11 44.2 12 0.11, 0.11 43.6, 44.2 13 0.12 43.0 15 0.09, 0.10 47.9, 47.9 19 0.09, 0.10 43.6, 44.2 Average 0.11 44.6 s , ppm 0.02 1.7 Each sample from bottle 15 gave significantly higher lead values than did those from the other bottles; this indicated that some lead inhomogeneity might exist. Lead has now been certified at 45 ppm, but no other values have been reported for bismuth, hence it is not certified. The results for nickel are shown in Table 6. The average of 1.4 ppm is in good agreement with an average of 1.2 ppm obtained by isotope-dilution, spark-source mass spectrometry. Nickel has not been certified at 1.3 ppm [1] . 102 Table 6. Polarographic determination of nickel in orchard leaves. Sample Ni , ppm 2 1.4 9 1.5 10 1.2 11 1.5 12 1.3 15 1.5 19 1.5 Average 1.4 s , ppm . 1 The investigation was extended to determine cadmium. Cadmium is also extracted by carbamate along with lead and bismuth. The peaks were very small and poorly defined for a 1 g sample but gave approximate values of 0.1 ppm. It was decided to take larger samples and try to determine cadmium, copper, lead, and bismuth simultaneously. By omitting the potassium cyanide during the extraction, copper would also be extracted. Five gram samples were taken through the procedure and bismuth, copper, lead, and cadmium were meas- ured simultaneously in a 10 percent hydroxylamine hydro- chloride supporting electrolyte. An average value of 0.11 ppm was obtained for bismuth which is in perfect agreement with the 0.11 obtained earlier. A value of 11.5 ppm copper was obtained which agrees with the 12 ppm value by neutron activation analysis [D. A. Becker (1971)] and isotope-dilution, spark-source mass spectrometry, and copper has been certified at 12 ppm. A value of 44.5 ppm lead was found which is also in good agreement with the previous values. The cadmium content was found to be 103 0.10 ppm which is in good agreement with the 0.11 ppm obtained by atomic absorption [T. A. Rains (1971)]. Both the polarographic methods described here and the new Standard Reference Material which has now been certified for 18 elements with information given for nine others should be very useful in many kinds of analytical problems, particu- larly those concerned with trace metals in foods, plants, and other environmental samples. 6. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] References Catalog of Standard Reference Materials, NBS Special Publication 260, July 1970 (and July 1971 supplement), Office of Standard Reference Materials, National Bureau of Standards, Washington, D. C. (available from U.S. Superintendent of Documents, U.S. Government Printing Office, Washington, D. C. 20402). Nangniot, P., J. Electroanal. Chem. 7_, 50-59 (1964). Nangniot, P., J. Polarogr. Soc. 11, 42-52 (1965). Maienthal, E. J. and Taylor, J. K. , Anal. Chem. 37_, 1516-1519 (1965). , J. , NBS (U.S Maienthal , (1965) . Maienthal , (1966). Maienthal , (1967). Maienthal, (1968) . Maienthal, (1969) . Maienthal , (1970). J. , NBS (U.S. J. , NBS (U.S. J. , NBS (U.S. J. , NBS (U.S. J. , NBS (U.S. Tech. Note 273, 8-13 Tech. Note 403, 22-40 Tech. Note 425, 23-41 Tech. Note 455, 19-40 Tech. Note 505, 17-36 Tech. Note 545, 41-69 104 [11] Maienthal, E. J. and Taylor, J. K. , in Advances in Chemistry Series, Vol. 73, "Trace Inorganics in Water," American Chemical Society, Washington, D. C. pp. 172- 182 (1968). [12] Florence, T. M. and Izard, D. B., Anal. Chim. Acta 2_5, 386-389 (1961). [13] Willard, H. H. and Dean J. A., Anal. Chem. 2_2, 1264- 1267 (1950). [14] Perkins, M. and Reynolds, G. F. , Anal. Chim. Acta, 18^, 616-631 (1958). [15] Rooney, R. C, Analyst, 8_3, 546 (1958). [16] Rooney, R. C, Analyst, 83, 83-88 (1958). 105 ACKNOWLEDGMENT The editor acknowledges the permission received from the various scientific journals to reproduce the papers contained in the Appendix. Thus: Appendices 1, 3, 6, 7, 9, and 13 were reprinted by permission of Analytical Chemistry; Appendix 5 was reprinted by permis- sion of the Journal of Radioanalytical Chemistry; Appendix 8 was reprinted by permission of Spectrochimica Acta; Appendix 10 was reprinted by permission of Applied Spectroscopy; Appendices 11 and 12 were reprinted by permission of the American Society for Testing and Materials; Appendix 13 was reprinted by permission of Marcel Dekker, Inc.; Appendix 15 was reprinted by permission of the Journal of the Associa- tion of Official Analytical Chemists; Appendix 16 was reprinted by permission of the Revue du GAMS. The editor further acknowledges with gratitude the assistance received during the preparation of this work from the following organizations and persons: -The Human Studies Laboratory of the Environmental Protection Agency for encouragement, and support given under Contract No. EPA AG-DS-0568, which made this publication possible; -The Office of Standard Reference Materials of the Institute for Materials Research at NBS, J. P. Cali, Chief, for continuous support throughout various phases of the experimental work; -Philip D. LaFleur, Chief, Analytical Chemistry Division, Institute for Materials Research at NBS, for advice and critical review of this publication; -H. L. Rook and T. E. Gills, Neutron Activation Analysis Section, Analytical Chemistry Division at NBS, for continuous guidance and help during the preparation of this work; 106 -J. R. DeVoe, Chief, Special Analytical Instrumentation Section, and T. J. Murphy, Analytical Spectrometry Section at NBS , and J. I. Shultz, ASTM Research Associate at NBS, for critically reviewing the manuscript; ■Mrs. Rosemary Maddock, Technical Information Service, Institute for Materials Research at NBS, for editorial assistance and for ascertaining that the terminology used in this work conforms to NBS style and format; •The Analytical Chemistry Division Text Editing Facility for typing the manuscript. 107 APPENDIX This Appendix contains sixteen reproductions from papers describing various phases of the analytical procedures assembled in this volume. It was intended to provide, in a convenient manner, additional relevant scientific information which could not be included in the main body of this work. It is hoped that this information will be useful to the analyst. The titles and names of the authors of the papers contained in this Appendix are listed below: 1. Production and Analysis of Special High-Purity Acids Purified by Sub-Boiling Distillation, by E. C. Kuehner, R. Alvarez, P. J. Paulsen, and T. J. Murphy. 2. The Role of the Analytical Blank in Accurate Trace Analysis, by T. J. Murphy. 3. Evaluation of Lyophilization' for the Preconcentration of Natural Water Samples Prior to Neutron Activation Analysis, by S. H. Harrison and P. D. LaFleur. 4. Activation Analysis with the Nuclear Reactor, by P. D. LaFleur and D. A. Becker. 5. Characterization of a Nuclear Reactor for Neutron Activation Analysis, by D. A. Becker and P. D. LaFleur. 6. Method for Determination of Mercury in Biological Materials by Neutron Activation Analysis, by H. L. Rook, T. E. Gills, and P. D. LaFleur. 7. Simultaneous Determination of Trace Elements in Platinum by Isotope Dilution and Spark Source Mass Spectrometry, by R. Alvarez, P. J. Paulsen, and D. E. Kelleher. 8. Determination of Trace Elements in Zinc by Isotope Dilu- tion Spark Source Mass Spectrometry, by P. J. Paulsen, R. Alvarez, and D. E. Kelleher. 108 9. Spark Source Mass Spectrographic Analysis of Ingot Iron for Silver, Copper, Molybdenum, and Nickel by Isotope Dilution and for Cobalt by an Internal Standard Technique, by P. J. Paulsen, R. Alvarez, and C. W. Mueller. 10. Trace Element Determinations in a Low-Alloy Steel Stan- dard Reference Material by Isotope Dilution, Spark Source Mass Spectrometry, by P. J. Paulsen, R. Alvarez, and C. W. Mueller. 11. Chemical Aspects of Atomic Absorption, by T. C. Rains. 12. Atomic Absorption Spectrometry - General Considerations for the Application of Experimental Techniques, by T. C. Rains. 13. Flame Emission Spectrometry with Repetitive Optical Scanning in the Derivative Mode, by W. Snelleman, T. C. Rains, K. W. Yee, H. D. Cook, and 0. Menis. 14. Standard Solutions for Flame Spectrometry, by J. A. Dean and T. C. Rains. 15. Determination of Submicrogram Amounts of Mercury in Standard Reference Materials by Flameless Atomic Absorption Spectrometry, by T. C. Rains and 0. Menis. 16. Discussion of Some Experimental and Fundamental Condi- tions in Analytical Flame Spectroscopy, by R. Mavrodineanu. 109 APPENDIX I Production and Analysis of Special High-Purity Acids Purified by Sub-Boiling Distillation Edwin C. Kuehner, Robert Alvarez, Paul J. Paulsen, and Thomas J. Murphy Analytical Chemistry Division, Institute for Materials Research, National Bureau of Standards, Washington, D.C. 20234 Sub-boiling distillation from pure quartz or Teflon (Du Pont) stills has been investigated for the production of high-purity inorganic acids and water. Nitric, hy- drochloric, hydrofluoric, perchloric, and sulfuric acids produced by this method contained significantly lower cationic impurities than high-purity acids from com- mercial sources. A complete system, including the Class 100 environment, production, and storage of these high-purity reagents is described. A method based on spark source mass spectrographic isotope dilution analysis has been developed for the simul- taneous determination of 17 elements in these mate- rials. Results of the analyses of both the acids puri- fied by sub-boiling distillation and the ACS reagent grade acids used as starting materials are reported. The sum of the common impurity elements determined in the purified acids ranged from 2.3 ppb in nitric acid to 27 ppb in sulfuric acid. No element in any of the purified acids exceeded 10 ppb and most were well below the 1-ppb level. The Analytical Chemistry Division at the National Bureau of Standards has become increasingly involved with the analysis of samples and Standard Reference Materials (SRM's) requiring the determination of elements at the low parts per million (ppm, 10~ 6 g/g) to the parts per billion (ppb, 10~ 9 g/g) concentration range. This is illustrated by recent analyses which include lead at 3.31 ppm in lunar samples (7), uranium at 72.1 ppb in Trace Elements in Glass (SRM 616) (2), nickel at 1.3 ppm in Orchard Leaves (SRM 1571) (2), and strontium at 0.14 ppm in Bovine Liver (SRM 1577) (2). The low level of these elements puts stringent requirements on the purity of reagents used in the analytical procedure. Inorganic acid purity is of particular importance because of the relatively large amount of these acids required for sample dissolution and other chemical operations. These procedures can require quantities of acids in excess of ten times the sample weight. To take full advantage of the sensitivity and accuracy of an analytical technique, the reagent blank should be held to no more than a few per cent of the amount being determined. Therefore, the accurate determination of an element near the 1-ppm level will depend on the availability of acids containing no more than 1 ppb of the element. High-purity inorganic acids have been available for some time from commercial sources. Although these acids are satisfactory for many trace element determinations, they are not always adequate for low level trace work either through a lack of purity or high upper limit specifications. [Emission (1) 1. L. Barnes, B. S. Carpenter, E. L. Garner, J. W. Gramlich, E. C. Kuehner, L. A. Machlan, E. J. Maienthal, J. R. Moody, L. J. Moore, T. J. Murphy, P. J. Paulsen, K. M. Sappenfield, and W. R. Shields, "Isotopic Abundance Ratios and Concen- trations of Selected Elements in Apollo 14 Samples," Proc. Apollo 14 Lunar Sci. Conf., Geochim. Cosmochim. Acta Suppl. 3, 2, MIT Press, Cambridge, Mass., in press. (2) Office of Standard Reference Materials, NBS Spec. Publ. 260 (1970). spectrography (J), the technique usually applied to the analyses of high-purity acids, lacks sensitivity causing upper limits to be set at 1 ppb or higher.] The determination of the isotopic composition and concentration of lead in lunar samples can be cited as a recent example. A reagent blank of 340 ng Pb was calculated from the producer's values for lead in commercial high-purity acids and the volume of these acids required to dissolve one gram of lunar rock and separate the lead from other constituents. Since the expected Pb con- centration was only a few ppm, the introduction of this quantity of external lead (of a different isotopic composition) would have precluded a reliable determination of the Pb present either as to amount or isotopic composition. Using acids purified by sub-boiling distillation, the total blank was determined to be 5 ng for the actual analysis. Consequently, because of this experience and other problems associated with the determinations of trace metals in a variety of materials, the Analytical Chemistry Division at NBS recently set up facilities for the "in-house" purification, analyses, and dis- tribution of inorganic acids. Sub-boiling distillation was selected as the method of purification for the inorganic acids. No single purification procedure is capable of removing all classes of impurities from these acids. Since the trace element program at NBS is mainly concerned with trace metals, the technique which ap- peared to be most efficient in removing metallic or cationic impurities, sub-boiling distillation, was studied. In sub- boiling distillation, infrared radiators vaporize the surface without boiling the liquid in the vaporizer compartment. The vapor is condensed on a tapered cold finger and the distillate is collected in a suitable container. Sub-boiling distillation was selected over conventional or boiling dis- tillation since studies have shown that in the latter method significant contamination of the distillate occurs from creeping of the unrectified liquid and entrainment of particulates in the vapor stream formed during bubble rupture (4). Sub- boiling distillation completely eliminates the entrainment problem since no bubbles are formed. Creeping of un- rectified liquid from the vaporizer is minimized by the position of the cold finger type condenser. It should be pointed out that while this is an extremely efficient still for the separation of impurities of low vapor pressure such as metal ions, it offers little purification from impurities of high vapor pressure such as organic matter or many of the anions. Commercial quartz sub-boiling stills which were designed for the production of high-purity water have been used for the production of high-purity mineral acids of low-level lead content at the Carnegie Institution of Washington (5). Similar sub-boiling quartz stills (Quartz Products Corp., Plainfield, N.J.) were installed at NBS for the production of (3) N. A. Kershner, E. F. Joy, and A. J. Barnard, Jr., Appl. Spec- trosc, 25, 542 (1971). (4) R. C. Hughes, P. C. Miirau, and G. Gundersen, Anal. Chem., 43, 691 (1971), and references cited therein. (5) K. D. Burrhus and S. R. Hart, ibid.. 44, 432 (1972). Reprinted from ANALYTICAL CHEMISTRY, Vol. 44, Page 2050, October 1972 American Chemical Society 110 hydrochloric, nitric, perchloric, and sulfuric acids as well as water. An all Teflon (Du Pont) sub-boiling still was designed and constructed at NBS for the production of high-purity hydrofluoric acid. Efficient purification by the still only partially solves the problem of producing and distributing pure acids. Airborne particulate contamination and container contamination must also be minimized to ensure a high quality product. Air- borne particulate contamination can be virtually eliminated by enclosing the still and distillate containers in a Class 100 clean air chamber. Teflon FEP bottles which have been vigorously cleaned with nitric and hydrochloric acids are used as containers for the purified acids and high-purity quartz is used for water. To evaluate the purity of the acids and water produced by this process, an analytical method based on isotope dilution- spark source mass spectrometry (SSMS) was developed. This method has limits of detection as low as 0.01 ppb for common impurity elements. Seventeen elements were determined simultaneously in each of the high-purity acids and water pro- duced by sub-boiling distillation and the ACS grade acids used as starting materials. APPARATUS FOR THE SUB-BOILING DISTILLATION OF HIGH-PURITY ACIDS AND WATER Quartz Sub-Boiling Still. The commercially available sub- boiling still shown in Figure 1 is made of quartz. This type still is used for the production of HC1, HN0 3 , HC10 4 , H 2 S0 4 , and H 2 0. Heating of the liquid being distilled is done by a pair of infrared radiators positioned on both sides of the con- denser. These elements, inside quartz tubes, heat the sur- face of the liquid and evaporate it without causing it to boil. This positioning also serves to heat the walls above the liquid tending to keep them dry which minimizes creep of liquid along the walls between the liquid reservoir and the condensing cold finger. The condenser is tilted downward toward the distillate outlet to allow the condensed liquid to flow to the tip above the outlet. The still is fed by a 6-8 pound bottle of ACS reagent grade acid through a liquid level control which maintains the liquid to just below the overflow height. Approximately 400 to 500 ml of liquid are thus maintained inside the still at all times. A three-way stopcock on the liquid level control is used to drain the still after the consump- tion of each bottle of ACS reagent grade feed acid. All parts of the liquid feed system are made of Teflon. Teflon Sub-Boiling Still. An all-Teflon (Du Pont) sub- boiling still having all the essential features of the quartz still was designed and constructed for the production of hydro- fluoric acid. This still, shown in Figure 2, was fabricated starting with a commercial 2-liter Teflon bottle. The heat- ers, condenser, acid inlet, and the overflow were inserted into what was originally the bottom of the bottle and the distillate outlet was inserted at the bottle cap. For the heaters, Teflon rods are machined to form 19-mm o.d. closed-end tubes. Heating coils inside glass tubes are then inserted into the heater tubes. The maximum operating temperature of the still is limited by the softening point of the Teflon around the heater. The glass tubing serves to support the heating tube. The cold finger condenser is similarily machined into a 25- mm o.d. tube from solid rod. Both the condenser and heater tubes are threaded at the open ends to fit into conversion fittings which are in turn threaded into tapped holes in the bottom of the bottle. The tapped hole in the condenser fitting is machined at an angle such that the condenser is tilted downward toward the distillate outlet. Because HF and other liquids do not adhere readily to Teflon, the dis- tillate drips off instead of flowing down along the condenser. Therefore, a trough is secured under the condenser to catch the HF and direct it to the distillate outlet. A piece of tubing Figure 1. Pure quartz sub-boiling still FEED BOTTLE Figure 2. All-Teflon sub-boiling still inserted through the bottle cap using commercial Teflon fittings serves as a collector for the distillate. The tube is provided with an umbrella at the end to protect the collection bottle from possible fall-out contamination. The acid feed and overflow tubes are attached with Teflon fittings threaded into the bottle bottom. Teflon tape is used when necessary on the threads of the fittings to ensure a liquid tight seal. Clean Air Chamber. Each sub-boiling still is housed in a clean air chamber (Environmental Air Control, Inc., Annan- dale, Va.) to protect the distillation process from external par- ticulate contamination. All parts on the clean air side of the chamber including the HEPA filter frame and diffuser are constructed of aluminum or plastic. The chambers meet Class 100 specifications — that is, they remove 99.97% of all particulate matter larger than 0.3 /im. The air flow through the chamber is adjusted to match the exhaust rate through a plenum assembly at the back-bottom of the unit. By this means, the acid purification operation is protected from room contaminates and at the same time, acid fumes are minimized in the room. Containers for Reagent Storage. Acid distillates are col- lected and stored in Teflon-FEP bottles. These bottles are cleaned by soaking for 24 hours first in hot 8M HN0 3 and then in hot 6M HC1 for 24 hours. They are then rinsed with sub-boiling distilled water and finally with the acid being stored. The containers are reused for the same acid after first being rinsed with distilled water and then rinsed with the acid. Water from the sub-boiling still is stored in quartz con- tainers which are cleaned in the same manner as the Teflon bottles. Ill Figure 3. Clean environment evaporator and transfer system APPARATUS AND REAGENTS FOR ANALYSIS OF ACIDS AND WATER Enriched Isotope Spike Solutions. The individual isotopic spikes for the elements being determined were on hand as 1 mg/ml solutions. Compatible spiking elements were com- bined to give 50 M8/ml concentrations of each element in a pair of master spiking solutions. One solution contained the HC1 soluble elements, the other solution the HN0 3 soluble elements. Clean Air Exhaust Hood. In the analysis of the acids, evaporations were performed on a hot plate in a Class 100 exhausted fume hood. The basic unit is similar to the clean chamber except that the air is exhausted down through a grill of Teflon that serves as a "floor" for the hood and then out through a standard acid fume exhaust system. Teflon is used extensively throughout those parts of the hood exposed to acid fumes. Evaporator and Liquid Transfer System. The apparatus, shown in Figure 3, provides a clean environment for evaporat- ing a spiked sample and transferring the residual volume onto high-purity gold wires. Nitrogen flowing through a 0.1 -^m cellulose filter provides a clean atmosphere, sweeps out evap- orating acid, and prevents entry of particulate contamination. A pure quartz pipet inserted off center in a Teflon stopper at the top is used to transfer the last few drops of evaporating acid onto pure gold wires positioned from the side joint. PROCEDURE FOR ANALYSIS OF ACIDS AND WATER BY ISOTOPE DILUTION SSMS The isotope dilution analyses were performed on the acids and water produced by sub-boiling distillation and on the ACS reagent grade acids used as starting materials. In addition, we have also included results obtained in previous analyses of commercial high-purity acids. This variety in the type and source of acids analyzed has resulted in some changes of the procedures used. Hydrofluoric Acid. Due to its reactivity with quartz and glassware, HF was handled exclusively in Teflon-FEP con- tainers, otherwise it was treated the same as the other acids from a similar source. Water from the Sub-Boiling Still. Because the purity of the sub-boiling distilled water was higher than that of the acids, the water was preconcentrated before spiking. Two liters of water were evaporated to approximately 100 ml in the original storage container before being removed and spiked. The water was not spiked in the storage container because of our reluctance to contaminate the container with the enriched spike isotopes. The spiked 100-ml water sample was then handled in the same manner as the acids purified by sub- boiling distillation. Sample Spiking. All samples were spiked using 1 fxg/ml solutions freshly diluted from the 50 £ig/ml master spike solu- tions. Acids purified by sub-boiling distillation were all spiked for a nominal concentration of 1 ppb; nitric and hydro- chloric acids were also spiked for 0.1 ppb. The nominal 1-ppb spike contained 100 ng of each spike element in 100 grams of acid (1 ng/g); the 0.1 ppb spike had 30 ng/300 grams of acid. The 1 ng/g spike gives an altered isotopic ratio of approximately one at a concentration of 1 ppb for most of the impurity elements. Concentrations up to ten times higher and more than ten times lower than the spiked for value can be determined from the measured ratios, but with greater uncertainty. The ACS reagent grade acids used as starting materials for the sub-boiling distillation were spiked at 1 ppb, 10 ppb, and 100 ppb in order to cover the expected concentration range for the different impurity elements. An element was only measured for the sample where the spike came closest to the actual concentration. One half of the spiked sample was used for each analysis. The commercial high-purity acids analyzed previously were spiked for many but not all of the elements surveyed in the ACS reagent grade and sub-boiling still acids. One hundred- gram samples were spiked for concentrations in the 1- to 100- ppb range. Because of the higher level of impurities only 20 grams of the 100-gram sample were required for the analyses. Sample Evaporation and Transfer to Pure Gold Wires. The clean environment evaporator shown in Figure 3 was used to vaporize commercial high-purity acids in a standard laboratory fume hood prior to our obtaining a Class 100 clean air fume hood. A single sample at a time was evap- orated to a few tenths of a milliliter and then transferred with the pure quartz pipet to the tip of "six nines grade" gold wires and dried. The ACS reagent grade acids and the acids purified by sub- boiling distillation were evaporated in the Class 100 fume hood in open quartz and/or Teflon beakers on a hot plate. The clean environment evaporation dish was used however to transfer the last few drops of the spiked sample to the gold wires for drying. The gold wires were then heated to 425 °C for 15 minutes to drive off hydrocarbons and occluded acid. At this point partial loss due to incomplete transfer or loss on heating can be ignored inasmuch as the quantitative analytical data are a function of the altered isotopic ratio and the only requirement is that enough of each element be retained for an adequate measurement. Spark Source Mass Spectrographic Determination of Altered Isotopic Ratios. The pair of gold wires was mounted in the spark source such that several millimeters of the wire ends, which were coated with the spiked sample residue, would be overlapping and parallel to each other. When a source vacuum of 1 X 10~ 7 Torr was reached, a graded series of exposures was made. The electrodes were moved relative to each other during sparking in order to spark new spiked sample residue on each exposure. One half of the surface of each electrode was sampled during this first graded series of exposures. A second identical series of graded exposures was then made on the remainder of the sample. A single photographic plate was used for each sample. The photographic plates were processed using the bleach and internal image developer for reducing plate fog developed by Cavard (6). The photoplates were examined visually to select exposures which would give optimum sensitivity for each element deter- mined and for evidence of interferences at the same nominal mass such as unresolved doublets, abnormally wide lines, or isotopic ratios which varied from one exposure to the next. If an interference was indicated for the +1 lines of an ele- ment, the less sensitive +2 or +3 lines were considered for measurement. The selected exposures for each element were then densitometered measuring both the spike and natural isotopes on each exposure. From four to six exposures were measured for each element if available. (6) A. Cavard in "Advances in Mass Spectrometry," E. Kendrick, Ed., The Institute of Petroleum, London, 1968, pp 419-29. 112 Calculation of Impurity Concentrations. The amounts of impurities present in each acid were computed from the sample weight, the weight of each spike added, and the measured altered isotopic ratios using the usual formula for isotopic dilution analysis as described in earlier papers (7, 8). Im- purity concentrations were calculated for each isotopic ratio measured for an element. These values were then averaged. This system gives a more realistic evaluation of the uncertainty in the results than using an averaged ratio to calculate one concentration value. This is especially critical when the measured (altered) isotopic ratio is close to that of either the spike or natural element. When only the spike isotope for an element was detected, an upper limit of concentration was computed by substituting twice the noise level of the photographic plate at the position of the natural isotope for the intensity of the natural isotope. The concentration of mononuclidic sodium was estimated by assuming it has the same sensitivity as potassium and mea- suring intensity-areas of the 23 Na compared to that of either the 39 K or the 41 K isotope of the same exposure. The results of these analyses of the acids and water are presented in a later section describing the operating procedure used for the production of each of the acids by sub-boiling distillation. DISCUSSION OF ANALYSIS PROCEDURE The acids purified in the sub-boiling still and the ACS reagent grade starting materials were analyzed for 17 elements simultaneously by the spark source. The elements measured represent elements found as impurities in reagents, elements commonly determined in many types of analytical samples, and elements from different groups of the periodic table. It was felt that this selection of elements would give a representa- tive view of the overall reagent purity. The analysis technique itself is subject to blank problems which cannot be properly evaluated. Care was taken during the analysis to prevent contamination of the samples from containers and the external environment. Based upon the lowest values found in repeated analysis of the acids produced by sub-boiling distillation, estimates can be made of the probable blank contributions. In general any element reported near the 0.1 -ppb level or below may have a con- siderable blank contribution and the value should therefore be considered as an upper limit. Concentrations signifi- cantly above this level represent actual reagent impurities with uncertainties ranging from ±10 to 30% depending on the value of the altered ratio. Interfering lines having the same nominal mass as either the spike or natural abundance isotope of an element being determined were sometimes encountered during these analyses. These interferences are normally most prevalent for the lower mass elements and for the more impure acids. The three major sources of these lines are hydrocarbons, anions, and anion fragments, and various molecular combinations of the major impurities with themselves and the acid anion. Heat- ing of the sample to 425 °C before sparking and the heating that occurs during sparking greatly reduced interference from the hydrocarbon and anion lines. Examples of anion fragment interferences are 37 C1 16 + with the 53 Cr+ spike and 36 C1 16 2 + with the 67 Zn+ spike in perchloric acid analysis, and 32 S l6 2 with natural 64 Zn + in sulfuric acid. The perchloric acid interferences can be evalua- ted by monitoring the line from the other chlorine isotope. (7) R. Alvarez, P. J. Paulsen, and D. E. Kelleher, Anal. Chem., 41,955(1969). (8) P. J. Paulsen, R. Alvarez, and D. E. Kelleher, Spectrochim. Acta, 24B, 535 (1969). Figure 4. Laboratory for production of high-purity acids ACS reagent grade HC10 4 with high levels of Na and K showed lines for: NaO+, NaK+, Na 2 +, K 2 +, Na 2 Cl+, K 2 C1+, Na 2 C10+, K 2 C10+, Na 2 C10 2 +, etc. Some of these lines represented potential interferences; however, this type of interference was not significant for the acids purified by sub- boiling distillation since their cationic impurities were always at a very low level. Measurement of the isotopic ratio of an element at the +2 or +3 charge state will eliminate all three types of interfering lines since these molecular species fragment rather than form a multiply-charged ion. When such interferences existed, Ag, Cu, and K isotope ratios were measured on +2 ions and Zn, Ni, Cr, and Ca isotope ratios were measured on +3 ions. PRODUCTION AND ANALYSIS OF HIGH-PURITY ACIDS AND WATER Figure 4 shows the laboratory setup for the production of high-purity acids. Each quartz sub-boiling still and dis- tillate container is housed in a separate Class 100 clean air unit to prevent particulate contamination during distillation. Any extraneous acid vapor produced is exhausted through the balanced plenum assembly which serves four clean air units. The Teflon sub-boiling still is housed in a separate clean air chamber and extraneous acid vapors are exhausted by a standard laboratory hood. The ACS reagent grade acid to be purified is fed to each still from a container just outside the clean air chamber through a liquid level control which main- tains the pre-set liquid level. In this manner, acid is con- stantly fed to the still as distillate is removed. Since im- purities are concentrated in the pot liquid, the still is drained after the consumption of each 6- to 8-pound feed bottle. After distillation, each acid was allowed to stand in its Teflon-FEP container for at least two weeks before analysis to allow for any impurities in or on the container walls to react with the acid. 113 Table I. Hydrochloric Acid Impurity concentrations ppb by wt (ng/g) Table HI. Perchloric Acid Impurity concentrations ppb by wt (ng/g) Acid from Acid from sub-boiling ACS reagent Commercial sub-boiling ACS reagent Commercial still grade high purity still grade acid pure acid Pb 0.07 0.5 <1 Pb 0.2 2 16 Tl 0.01 0.1 Tl 0.1 0.1 Ba 0.04 2 Ba 0.1 >1000 16' Te 0.01 0.1 Te 0.05 0.05 Sn 0.05 0.07 <6 Sn 0.3 0.3 3400 ppb 2 100 ppb Table II. Nitric Acid Impurity concentrations ppb by wt (ng/g) Acid from sub-boiling ACS reagent Commercial still grade acid high purity Pb 0.02 0.2 0.3 Tl 0.2 Ba 0.01 8 Te 0.01 0.1 Sn 0.01 0.1 1 In 0.01 Cd 0.01 o!i 0.2 Ag 0.1 0.03 0.1 Sr 0.01 2 Se 0.09 0.2 Zn 0.04 4 8 Cu 0.04 20 4 Ni 0.05 20 3 Fe 0.3 24 55 Cr 0.05 6 130 Ca 0.2 30 30 K 0.2 10 11 Mg 0.1 13 Na 1 80 2 2.3 ppb 2 220 ppb 2 240 ppb The purified acids were then analyzed by the isotope dilution mass spectrometric method described previously. These analyses characterize the acid as delivered to the user rather than as produced, since this is what is of interest to the analyst. The Teflon-FEP bottles will be used repeatedly over again to contain the same high-purity acid. It is reasoned that the container should become cleaner with use and cause less con- tamination to future lots of acids. Any acid not used within a few months will be replaced with freshly distilled reagent. Hydrochloric Acid. Hydrochloric acid and water form a constant boiling mixture at 6/V hydrochloric acid. Initial experiments were with starting acid of this concentration. Further experiments showed that the starting concentration could be increased to 10/V HC1 without causing bubble for- mation and that the distilled product was also 107V. How- ever, concentrated 12A r acid could not be used because of bubble formation on heating. Apparently, acid of higher concentration than constant boiling 6N can be produced be- cause the azeotropic mixture that condenses on the cold finger absorbs HC1 from the HCl-rich vapor to reach the concen- tration of the acid in the pot. ACS reagent grade hydrochloric acid diluted to ION or 31 weight per cent with high-purity water was used as the feed acid. The still heaters were adjusted to 225 W of power. About 2 liters of 10/V high-purity hydrochloric acid were pro- duced per day under these conditions. Table I shows the analysis of the high-purity acid produced by sub-boiling distillation, the starting ACS reagent grade acid, and a lot of commercial high-purity acid. A summation of these impurity elements shows that the sub-boiling dis- tilled acid contained 6.2 ppb, the ACS reagent grade contained 820 ppb, and commercial high-purity 70 ppb. The only element found in the sub-boiling distilled acid at levels higher than 1 ppb was iron at 3 ppb. The other elements were at the sub-ppb level, generally lower than 0.1 ppb. Nitric Acid. The sub-boiling distillation of concentrated 70% ACS reagent grade nitric acid required a low heater temperature to prevent the HN0 3 vapor in the proximity of the heater from disproportionating to form nitrogen dioxide with resulting discoloration of the distillate. Experiments showed that at a heater power setting of 107 W, little or no discoloration of the distillate took place. As a result, only about 500 ml per day of nitric acid was produced. Titration showed the distillate to be 70% HN0 3 , the same concentra- tion as the starting acid. Table II shows the results of the analysis of the sub-boiling distilled acid, ACS grade starting acid, and a lot of commercial high-purity acid. The totals of the impurity elements deter- mined were 2.3 ppb for the sub-boiling distilled acid, 220 ppb for the ACS starting acid, and 240 ppb for the commercial high-purity acid. The sub-boiling distilled nitric acid was the purest of the acids produced. No element was found at a concentration greater than 1 ppb and only sodium was detected at that level, and this accounts for almost half of the total impurities found. Most elements were in the 0.05- to 0.01 -ppb range. Perchloric Acid. Initial experiments on the sub-boiling distillation of concentrated 70% perchloric (HC10 4 -2H 2 0) showed that high heater temperatures caused the perchloric 114 Table IV. Sulfuric Acid Impurity concentrations ppb by wt (ng/g) Table V. Hydrofluoric Acid Impurity concentration ppb by wt (ng/g) Acid from ACS reagent Acid from ACS reagent sub-boiling still grade acid sub-boiling still grade acid Pb 0.6 0.5 Pb 0.05 0.8 n 0.1 0.1 Tl 0.1 0.2 Ba 0.3 0.2 Ba 0.1 0.5 Te 0.1 0.1 Te 0.05 0.1 Sn 0.2 0.6 Sn 0.05 11 Cd 0.3 0.2 Cd 0.03 2 Ag 0.3 0.6 Ag 0.05 0.1 Sr 0.3 0.4 Sr 0.1 0.5 Zn 0.5 2 Zn 0.2 4 Cu 0.2 6 Cu 0.2 3 Ni 0.2 0.5 Ni 0.3 12 Fe 7 6 Fe 0.6 110 Cr 0.2 0.2 Cr 5 20 Ca 2 123 Ca 5 14 K 4 9 K 1 28 Mg 2 4 Mg 2 10 Na 9 50 Na 2 100 2 27 ppb 2 200 ppb 2 17 ppb 2 320 ppb acid in the still to turn bright yellow, probably because of the formation of chlorine dioxide. At lower power settings, little or no color was noted and the distillate was colorless. Another problem with this distillation was that in the earlier work, crystals would occasionally build up on the cold finger con- denser. This material was probably perchloric acid mono- hydrate (HC10 4 H 2 0) which has a melting point of 50 °C. Crystal formation could be controlled by changing either the flow rate of the condenser water or the heater temperature. However, this problem was noted only in the earlier work. For the last six months, no crystal formation on the cold finger has been noted even when the still is operated con- tinuously for weeks. The power setting used for the pro- duction of purified perchloric acid was 240 W, and the pro- duction rate was about 600 ml of perchloric acid per day. Titration has shown that the distillate is 70 wt % of HCIO4, the same as the starting ACS grade acid. Table III shows the results of the analyses of the sub- boiling distilled acid, the starting ACS grade acid, and a lot of commercial high-purity acid. The total of the impurity elements found was 16 ppb for the sub-boiling distilled acid, 3400 ppb for the starting acid, and 100 ppb for the commercial high-purity acid. These analyses demonstrate the efficiency of sub-boiling distillation when the concentration of impurity elements in the purified product is compared to the starting acid. The concentration of barium which was greater than 1000 ppb was reduced to 0.1 ppb, a purification factor of greater than 10,000. Another example is calcium which was reduced to 0.2 ppb in the purified acid from 760 ppb in the starting acid, a factor of 3,800. It should also be noted that chromium which was present at 10 ppb in the starting acid was not significantly changed in the purified acid. This is probably due to the fact that chromium can be volatilized from hot perchloric acid solutions as chromyl chloride, Cr0 2 Cl 2 , so little, if any, purification can be expected by sub- boiling distillation. Chromium accounts for over half of the total impurities found in the purified acid. Sulfuric Acid. Initial experiments with the sub-boiling distillation of sulfuric acid showed that, because of the density of H2SO4, the acid that condensed on the cold finger tended to drop off before reaching the collection tube. Another problem was that even at the highest power setting, the dis- tillation rate was extremely slow. Modification of the still Table VI. Water Impurity concentrations ppb by wt (ng/g) Pb Tl Ba Te Sn Cd Ag Sr Zn Cu Ni Fe Cf Ca K Mg Na Water from sub-boiling still 0.008 0.01 0.01 0.004 0.02 0.005 0.002 0.002 0.04 0.01 0.02 0.05 0.02 0.08 0.09 0.09 0.06 2 0.5 ppb by increasing the angle of incline of the cold finger condenser solved the condensate problem, and the rate of distillation was increased by using a platinum foil reflector around the still. At the maximum power setting of 427 W, the rate of distillation is now about 300 ml per day. Titration has shown that the purified acid is 96 wt % H 2 S0 4 , the same as the starting ACS grade acid. Table IV shows the results of the analyses of the sub- boiling distilled acid and the starting sulfuric acid. The total of the impurities found in the starting acid was 200 ppb, the lowest of any of the ACS grade acids, and the purified acid contained 27 ppb, the highest of the purified acids. Sodium was the principal contaminant in the high-purity acid at 9 ppb with iron right behind it at 7 ppb. These two elements account for over half the total impurities found in the purified sulfuric acid. Hydrofluoric Acid. Hydrofluoric acid was purified in the all-Teflon sub-boiling still shown in Figure 2 which was designed and built at NBS. Hydrofluoric acid and water form a constant boiling mixture at 36% HF. Because of the previous experience with hydrochloric acid, the sub-boiling distillation of concentrated 48% HF was attempted. Titra- tion showed that the distilled product was also 48 % HF. Ap - 115 Table VII. Summary of Experimental Conditions for Producing High-Purity Acids and Water by Sub-Boiling Distillation Concen- tration of Concen- still feed tration of Production acid, % Still heater distillate, rate, Reagent by wt power, W % by wt ml/24 hr HC1 31 225 31 2,000 HN0 3 70 107 70 500 HCIO4 70 240 70 600 H 2 SO, 96 427 96 300 HF 48 164 48 300 H z O 200 4,000 parently, as in the case of hydrochloric acid, the azeotropic mixture that condenses on the cold finger absorbs HF from the HF rich vapor to reach the same concentration as the acid in the still. Since the heaters of the Teflon still for the purification of hydrofluoric acid are enclosed in Teflon tubes, the temperature of the heater had to be kept below the softening point of Teflon. As a result the maximum power setting for this still was 164 W which produced a distillation rate of 300 ml per 24 hour day. Table V shows the results of the analyses of the starting ACS grade HF and the purified product. The total of the impurities in the purified acid is 17 ppb compared to 320 ppb in the starting ACS grade acid. The two principal impurities in the purified acid are Ca and Cr, both at 5 ppb, and they account for over half the total impurity found. The only other elements found at a concentration of greater than 1 ppb are K at 1 ppb, Mg at 2 ppb, and Na at 2 ppb. Water. Water has been included in this study because it is the most commonly used chemical reagent and because sub- boiling distillation has been used to produce high-purity water both here at NBS and elsewhere. Initially, water of higher purity than the distilled water from the laboratory distribu- tion system was obtained by re-distillation in a commercial, continuous-feed still. This still had a tin-coated evaporator and baffle and was equipped with a quartz condenser. This unit was designed to disengage C0 2 and other gases from the distillate and vent them off. Water produced by this still was used as the feed water to the sub-boiling still. The en- tire assembly of feed apparatus, quartz sub-boiling still, and quartz collector was contained in a Class 100 clean air cham- ber. The distillation rate at a power setting of 200 W was 4 liters per day. The analysis of the sub-boiling purified water is given in Table VI. This water contained a maximum of 0.5 ppb total of the 1 7 elements determined. No element was detected at a concentration of greater than 0.1 ppb. Since no correction for an analysis blank could be made, the values must be regarded as upper limits for the purity of the water. The water may in fact be more pure than the analysis indicates. The optimum conditions for the production of each high- purity acid and water by sub-boiling distillation described above are summarized in Table VII. The impurity elements determined in the acids and water (Tables I-VI) are tabulated in order of decreasing mass. This reveals that the major impurities for the acids purified by sub-boiling distillation are all low mass elements. None of the NBS purified acids have an impurity level as high as 1 ppb for elements above iron in mass. Elements from iron on down sometimes exceed 1 ppb but none exceeded 10 ppb. A number of the major impurities found in the ACS reagent grade and commercial high-purity acids are elements common to their glass storage containers. It would appear that the means of storing the acids are as responsible for the acid purity as the purification method. It should be noted that of all the acids analyzed, only the ACS grade perchloric acid showed an element (Ba) that exceeded the 1-ppm level. Received for review March 27, 1972. Accepted May 30, 1972. Certain commercial instruments or chemicals are mentioned in this paper; this does not imply recommendation or endorsement by the National Bureau of Standards. 116 APPENDIX II REPRINTED IN PART FROM NATIONAL BUREAU OF STANDARDS SPECIAL PUBLICATION 422. Accuracy in Trace Analysis: Sampling, Sample Handling, and Analysis, Proceedings of the 7th IMR Symposium, Held October 7-1 1 , 1974, Gaithersburg, Md. (Issued August 1976). THE ROLE OF THE ANALYTICAL BLANK IN ACCURATE TRACE ANALYSIS Thomas J. Murphy Analytical Chemistry Division Institute for Materials Research National Bureau of Standards Washington, D.C. 20234 U.SA. The inability to control the analytical blank, i.e., contamination from all sources external to the sample, has seriously affected the accuracy of low level trace determinations. Most of the sources of the blank are variable and it is this variability that determines the uncertainty of the blank correction and, there- fore, the lower limit of trace element concentration that can be determined with reliability. To improve both the accuracy and lower limit of trace deter- minations, it is imperative to control the variability of the analytical blank. The only practical way to accomplish this is to reduce the size of the blank itself by controlling the sources of the blank. The analytical blank is composed of contamination from four principal sources, namely: the environment the analysis is performed in, the reagents used in the analysis, the apparatus used, and the analyst performing the analy- sis. Environmental contamination is caused by particulates and gaseous com- pounds in the ambient air. Methods for reducing the blank from this source by the use of isolation chambers or "absolute" filters are described. While commercially available high-purity reagents have helped the analyst to control the blank from reagents, they are frequently not low enough in trace element concentration and must be purified before use. Methods for the preparation of high-purity reagents, especially water and the mineral acids, are reviewed. Contamination from beakers, containers, mortars and other apparatus can seriously affect the blank. Techniques for controlling this type of contamina- tion are described with emphasis on the purity of materials. The last source of contamination is that caused by the analyst during the analysis due to carelessness or poor technique. Suggestions are given to aid the analyst in this regard. Keywords: Accuracy; analysts; apparatus; blank; contamination; environ- ment; purity; reagents. 117 Accuracy in Trace Analysis I. Introduction The analytical blank may be considered the "Achilles' heel" of trace analysis. As the Greek warrior of Homer's Iliad had his vulnerable point, so also does trace analysis in that the size and variability of the analytical blank may render useless the information from the analysis. First, let me define what is meant by the analytical blank. The analytical blank is simply the contamination by the element or compound being determined from all sources external to the sample. I am not including in this defini- tion the so-called instrumental blank or noise level which really refers to the sensitivity or level of detection of the analytical method. Also not in- cluded in this definition is the so-called negative blank, that is loss of an element during analysis by physical processes such as adsorption. This is really a separate subject unto itself. Obviously, this definition includes both inorganic and organic contamination, but my remarks will deal mainly with trace inorganic cation contamination. Modern methods of analysis have lowered the threshold of determining trace elements to the low parts per billion (ng/g) for many elements but the inability to control the analytical blank has seriously affected the accuracy of these methods. The blank is a problem common to all trace element techniques with the exception of certain nuclear techniques such as neutron activation analysis and fission track analysis. Let's look at the effect of the blank on the accuracy of a trace analysis. During the analysis, the analyst normally carries "blanks" thru all the steps of the analysis and makes a "blank-correction" of the analysis based on the estimation of the contamination from these blank determinations. Figure 1 shows the effect on accuracy of a 10 ng blank that exhibits a variability of ± 5 ng. This represents a very low blank with a reasonable uncertainty. It is not uncommon, even in very careful work to have blanks an order of magnitude higher. Notice that at the microgram level, the ef- fect on accuracy of a blank of this size is small, but that it increases rapidly until at the 0.01 Aig or 10 ng level, a 50 percent uncertainty would be introduced by the blank. If the blank were an order of magnitude higher with the same percentage variability, a determination at the 0. 1 /Ag level would have a 50 percent uncertainty due to the blank alone. It is the variability or the uncertainty in the blank correction [ 1 ] and not the absolute value of the blank that affects the accuracy of the analy- sis. As the variability becomes a significant percentage of the sample amount, more and more determinations of the blank are necessary so that the variability can be estimated with some degree of reliability. It is a common mistake of analysts to base the blank correction on a single 118 Murphy 0.01 ■ 10 20 30 40 PERCENT UNCERTAINTY Figure 1. Effect of a 10 ± 5 ng blank on accuracy of analysis. determination and have no information on the approximate variability or effect on accuracy. One might argue that, since it is the variability and not the absolute value of the blank that affects accuracy, the way to increase accuracy is to have large blanks but with reduced variability. This is certainly not true in the real world of trace analysis. Firstly, it is inherently inaccurate to base a result on the difference between two signals or numbers of relatively the same size. But secondly, and more importantly, real blanks, large as well as small, do exhibit marked variability since most of the sources of the blank are not constant. The only practical way to lower the variability and increase the accuracy of the blank correction is to reduce the size of the blank. It is obvious from figure 1 that as one goes to lower levels of trace concentrations the blank becomes more and more of a factor. Figure 2 119 + 40 r + 20 -20 -40 Accuracy in Trace Analysis COMPARISON of INTERLABORATORY RESULTS for an AQUEOUS STANDARD and WHOLE BLOOD Aqueous Solution 140 ng Pb/g Porcine Blood 30 ng Pb/g 75 60 456 109 453 9 •0||4 62 1234567 1234567 LABORATORY NUMBER Figure 2. Comparison of interlaboratory results for an aqueous standard and whole blood. represents two parts of a study to develop a referee method for lead in blood by atomic absorption spectrophotometry. The left portion represents the results from different laboratories for the analysis of an acidified aqueous standard. No chemistry was necessary and the results fluctuate, not very accurately, around the "base-line" value. But notice what happens when the same laboratories analyze a real blood sample as shown on the right portion. 1 The level of lead was a factor of five lower than the standard and some chemistry was necessary. The results varied from 30 to 450 percent high, in all probability due to lack of control of the analytical blank. This is but one example of the fact that as one goes to lower trace con- centrations, the blank becomes the dominant factor on accuracy and determines the lower limit of trace element concentration that can be determined with reliability. It is imperative to control and reduce the analytical blank if improvements in precision and accuracy of trace deter- minations are to be made. This is for porcine blood which contains much lower lead than human blood. 120 Murphy II. Sources of Contamination How does one go about lowering a blank? The sources of the blank must be considered and steps taken to reduce the contamination from each source. The analytical blank is composed of contamination from four principal sources, namely, the environment the analysis is performed in, the re- agents used in the analysis, the apparatus used, and the analyst perform- ing the analysis. Of these factors only the contamination from reagents is relatively constant while the contamination from the other factors is varia- ble and determines the variability of the analytical blank. A. Environment Let us consider environmental contamination. Fall-out or absorption of particulates or gases from the laboratory air can cause significant and variable amounts of contamination of the sample or a solution of the sam- ple. As much as 200 fxg of particulates per m 3 of air have been reported in the air of an analytical laboratory [2] and it was shown to contain large amounts of calcium, silicon, iron, aluminum, sodium, magnesium potassi- um, thallium, copper, manganese and lesser amounts of several other ele- ments. Analysts in Dortmund, Germany, were not able to lower iron blanks lower than 0.04 ^tg/ml because 20 tons of ferric oxide were dispersed into the air from nearby plants [3 ] . Table 1 shows the lead par- ticulate concentrations from three different locations. The lead concentra- tion from downtown St. Louis, 18.84 /Mg/m 3 , was much higher than rural Missouri, 0.77 /xg/m 3 , [4] due to automotive exhaust pollution. Our Table 1. Examples of lead concentration in air Site Lead concentration Mg/m 3 Downtown, St. Louis, Missouri 8 18.84 Rural Park, Southeastern, Missouri* 0.77 Laboratory air, NBS, Gaithersburg, Maryland b 0.4 a See reference [4]. b See reference [5]. 121 Accuracy in Trace Analysis laboratory air at NBS, Gaithersburg, Maryland [5] compares favorably to rural air but still contains too much lead for low level work. Wide varia- tions in the particulate content of air is to be expected. U seller [6] has stated that in a typical rural area airborne particulate counts larger than 0.5 /urn of 1, 400, 000/ m 3 are not unusual and that in metropolitan areas counts of 5 3, 3 00 ,000/ m 3 are normal. The use of evaporation chambers has been one approach to reducing airborne contamination. Evaporations are performed in a closed system under filtered air or nitrogen. Figure 3 shows some different evaporation chambers from the literature. Chamber (a) was designed by Thiers [7,8] . Filtered air is passed over the solution and out the bottom while the sam- ple is heated from below with a hot-plate and above with a heat lamp. Chamber (b) is a German design by Koch [9] in which filtered air or nitrogen is passed over the sample solution. A heated quartz dome IR lamp Pyrex W^ Filtered air in •Out Hot plate (a) Quartz surface evaporator Glass Heater (b) Borosilicate glass Teflon , Filtered N, in II ■ IA I (c) Drain Out ' ■!■■ Filtered N, in Stainless steel 7 id) Figure 3. Evaporation chambers from the literature, reference 35. Reprinted with permis- sion of John Wiley & Sons, Inc. 122 Murphy prevents condensation of liquids. The last two chambers, (c) and (d), were used by Chow and McKinney [10] for the analysis of high-purity HC1 (Glass) and HF (Teflon). Figure 4 is a drawing of an evaporation chamber that has been used by Alvarez at NBS [11]. The solution is evaporated by heating from above and below the chamber while filtered nitrogen passes over the solution until only a drop remains. The drop is then drawn up into the capillary and loaded onto gold wire for analysis by spark source mass spectrometry. Table 2 shows the results of Chow and McKinney [10] for the analysis of 500 ml samples of high-purity hydrochloric acid which was evaporated Figure 4. Clean environment evaporator and transfer system. Table 2. Lead contamination of labware after exposure to air* Beaker Condition Laboratory Time (days) Lead M(g) Teflon Open Ordinary 8 4.07 2.32 Nitrogen-flushed container Open Ordinary Pure air 8 8 1.13 0.44 Nitrogen-flushed container Pure air 8 0.18 .13 Borosilicate glass Nitrogen-flushed container Nitrogen-flushed container Pure air Pure air 1 1 0.02 .03 • See reference [10]. 123 Accuracy in Trace Analysis slowly under different conditions. When the HC1 was evaporated over a period of 8 days from an open Teflon beaker, the lead contamination was high and variable. Using a clean air chamber under nitrogen reduced the lead contamination only by a factor of about three. Both of these evapora- tions were performed in a normal fume hood. (Since the volume of air moving across an area in a fume hood is many times greater than the air moving across the same area in the laboratory, the contamination is greater in the hood than elsewhere in the laboratory.) When the same evaporations were performed in a clean air laboratory in which dusts and aerosols were removed by electrostatic precipitation and filtration, the lead contamination was reduced by a factor of 10 when the HC1 was evaporated open and by a factor of 20 when it was evaporated in an evaporation chamber flushed with nitrogen. The major development to date for providing particulate free air is the high efficiency particulate air filter, better known as the HEPA filter [12]. This filter was developed during World War II for the Manhattan Project to remove fissionable particulates from air and to this date remains the most efficient means of air filtration. The HEPA filter has an efficiency of 99.97 percent for 0.3 /xm particles and are routinely em- ployed for supplying air that meets the Class 1 00 specification [13], that is less than 100 particles per cubic foot of air larger than 0.5 /xm. These fil- ters, coupled with the laminar airflow principle [14] in which the "clean" air is moved unidirectionally, have been used for the last decade to pro- vide clean particle-free work areas. Figure 5 is a view of the clean room located in the Analytical Chemistry Division at the National Bureau of Standards. The HEPA filters are located on the far wall. In fact the entire wall is composed of HEPA filters causing a wall of clean air to pass unidirectionally down the room. This room is really a particulate-free room and cannot be used as a clean laboratory since it was not designed for acid vapor. Figure 6 is a cut-away view of the clean laboratory also located in the Analytical Chemistry Division at NBS. This room was designed for trace inorganic analysis and can be used for chemical operations such as acid dissolutions and evaporations. All of the air entering the room is passed thru prefilters and then thru the HEPA filters located over the center work-bench of the room. The Class 100 clean air is directed unidirectionally down and out across the bench to prevent particles from entering the area. Plexiglas dividers prevent cross-contamination. The air is recirculated thru the filters by drawing air at the base of the outside walls. In addition to the "clean" area under the center bench, there are several Class 100 hoods that provide clean work areas on the outside bench. Acid dissolutions and evaporations are performed in the two 124 Murphy Figure 5. Particulate free clean room -National Bureau of Standards. hoods at the rear of the room. These are Class 100 clean air hoods lined with sheet Teflon and are so designed that all air 1 2 cm from the front is exhausted while a curtain of air moves out the front to prevent particu- lates from entering the hood. Figure 7 is a schematic of a laboratory used at Bell Telephone Labora- tories for trace analysis [15]. The inlet air to the room is non-laminar flow clean air which has been filtered thru HEPA filters. The air to the clean hood is further purified by recirculation thru HEPA filters and the room is kept under positive pressure. Using this room, no significant con- tamination of water samples could be detected by spark source mass spec- trometry or activation anaysis. Table 3 shows a comparison of particulates in the air of an ordinary laboratory, a clean room, and a Class 100 hood, all at NBS [5 ] . Both the clean room and Class 1 00 hood air showed a dramatic reduction in con- tamination. Lead concentrations were reduced by a factor of over 1000, iron by a factor of 200 and the other elements, which were low to start with, by a factor of 1 0. While HEPA filters are a powerful tool in reducing particulate con- tamination, they are not the complete answer to environmental con- 125 Accuracy in Trace Analysis Q. a S £ >% O X> CO 3 u 126 Murphy WALL CABMET CHEM BENCH FILTER Figure 7. Schematic of clean laboratory — Bell Telephone Laboratories, reference 22. Reprinted with permission of Analytical Chemistry. Table 3. Particulate concentration in laboratory air* Concentration, Mg/m 3 Iron Copper Lead Cadmium Ordinary Laboratory 0.2 0.02 0.4 0.002 Clean Room .001 .002 .0002 n.d. Clean Hood .0009 .007 .0003 0.0002 s See reference [5]. tamination. Figure 8 shows the particle size of some common contami- nants. Since HEP A filters are efficient for the removal of particles 0.3 /xm or larger, particles as small as bacteria are removed, but some common particulates such as tobacco smoke are not efficiently removed since they are smaller than 0.3 /xm. Obviously gases, both organic and inorganic, are not removed. To remove these contaminants, the air must be passed thru absorbants and scrubbed, which is an expensive operation. Most labora- tories either live with the small amount of contamination that results from clean room operation or resort to the evaporation chamber technique in the clean room to further reduce environmental blanks. 127 Accuracy in Trace Analysis POLLEN I < >- _i u. t/> LU s 3 a. 2 y 5 LU I Q. LU >- LU z < £ 3 X >- CO LU _l CO CO > - 1 1 1 Mill BACTERIA to (— LO z> Q LO o t— < O z —J 1— I— UJ LO - 1 1 1 1 1 1 O < o z 1— 1— UJ LO z 3 - PL, - o 128 Murphy B. Reagents The chemical reagents used during the course of an analysis are an im- portant source of the analytical blank. The first order of priority for a trace laboratory in regard to reagents is a readily available supply of ex- tremely high-purity water. There are several commercial systems on the market for producing "conductivity water"; that is, water whose conduc- tivity approaches the theoretical conductivity of 0.055 micromho/cm at 25 °C [15,16]. These systems are based on the use of mixed-bed or monobed ion exchangers, in which cation and anion exchange resins are mixed in a single unit, and sub-micron filtration. Measured conductivity of deionized water has been reported as low as 0.05 micromho/cm (tem- perature not specified) [16]. However, only ionized compounds are de- tected by conductivity measurements and soluble unionized compounds or particulates are not detected. Both soluble organic materials and par- ticulates are generated by ion-exchange columns. Studies at the Philips' Laboratories [17] have shown that deionized water with a conductivity of 0.06 micromho/cm contained up to 100 ppb of cation contamination. They postulated that the detected impurities were present in particulate or colloidal form and were not appreciably ionized. Another possibility is that the cations form unionized complexes with dissolved resin material. In fact there is some literature evidence [18-20] to show that small amounts of heavy metals can be masked in deionized water and lost to polarographic analysis due to chelation by soluble resin. Whatever the reason, there is ample evidence in the literature [17] to prove that con- ductivity is not a reliable measure of trace element concentration in water. Dr. Ralph Thiers [8] of Harvard wrote in 1957, "Since non-ionic materials are not held by ion-exchange resins, and since traces of nitrogeneous compounds are leached off the resin by water, probably the best treatment for obtaining purest water in all ways is ion-exchange fol- lowed by distillation from fused quartz." Nothing has happened in the last two decades to change the validity of that statement and this position has been supported by other investigators [21-23]. The system used by our group at NBS consists of taking the "building distilled" water as it is supplied to the laboratory and redistilling it in a conventional still equipped with a quartz condensing system. The distil- late from this still is fed to a quartz "sub-boiling" still (Quartz Products Corp., Plainfield, N.J.) where the water is again distilled, but by a non- boiling technique which will be described later. The "building distilled" water consists of ordinary city water that has been passed thru a mixed- 129 Accuracy in Trace Analysis bed deionizer, distilled in a tin-lined still, stored in a tin-lined tank and dis- tributed thru tin pipes to the laboratory. Table 4 shows the results of the analysis of the "sub-boiling" distilled water and the "building distilled" water by spark source isotope dilution mass spectrometry (SS-IDMS). There was better than a tenfold reduction in impurities from "building distilled" water which is really good quality water for a high- volume operation. Copper at 3 ppb is the principal impu- rity in this water. The analysis of the "sub-boiling" distilled water, which totaled 0.5 ppb for the elements determined, really represents an upper limit for most impurities since no attempt was made to correct for a blank during the evaporation of the water. This water compares favorably with the best water reported in the literature. While it is one problem to produce high quality distilled water, it is quite another to keep it pure be- fore use. We have been using quartz for storage and this analysis is for water stored in quartz for 2 weeks. We are currently studying the storage of water in Teflon. Studies in the literature have shown that the trace element content of water stored in Teflon or polyethylene increased by a factor of 5 to 10 for some elements after 30 days storage [22]. As a general rule, distilled water should be used as soon as possible after preparation. Table 4. Impurity concentration in distilled water Sub-boiling Building distilled distilled (ng/g) (ng/g) Pb 0.008 0.1 Tl .01 <.01 Ba .01 .004 Te .004 .02 Sn .02 .2 Cd .005 .03 Ag .002 .005 Sr .002 .002 Zn .04 1 Cu .01 3 Ni .02 0.1 Fe .05 .4 Cr .02 — Ca .08 1 K .09 0.06 Mg .09 2 Na .06 0.1 Total Impurity 0.5 ppb 8.0 ppb 130 Murphy The next most important group of reagents for the trace analyst are the mineral acids. In recent years high-purity acids have been available from commercial suppliers. While these acids may prove quite satisfactory for a number of trace applications, the concentrations of some trace elements are too high for low level trace analyses so the analyst will have to purify his own acids in these cases. High-purity hydrochloric acid, hydrobromic acid, and hydrofluoric acid have been produced by saturating pure water with the pure gaseous compound [7,8,24-30] or by isopiestic or isother- mal distillation [31,32]. The two-bottle sub-boiling method of Mattinson [33] has been used to produce high-purity HC1 and HF [34]. Nitric acid, perchloric acid and sulfuric acid are usually purified by repeated distillations [35]. We have been purifying these acids by "sub-boiling" distillation from quartz or Teflon stills [ 1 1 ,36] . Figure 9 is a schematic of a quartz "sub- boiling still which is used for the production of high-purity water and the mineral acids with the obvious exception of hydrofluoric acid. The still is fed by a 6- to 8-lb bottle of ACS Reagent Grade acid through a liquid level control. Heating of the liquid being distilled is accomplished by a pair of infrared radiators positioned on both sides of the condenser. The surface of the liquid is heated and evaporated without causing the liquid to boil. This positioning of the heaters also serves to heat the walls above the liquid, tending to keep them dry which minimizes creep of FEED BOTTLE iNOENSER OVERFLOW Figure 9. Pure quartz sub-boiling still. 131 Accuracy in Trace Analysis liquid between the cold finger condenser and the liquid reservior. The cold finger condenser is tilted downward to allow the condensed liquid to flow to the tip above the outlet. The distillate is caught in Teflon FEP bot- tles which have been rigorously cleaned with nitric and hydrochloric acids. Figure 10 is a schematic for the all Teflon sub-boiling still [ 1 1 ] . It was fabricated from a 2-liter TFE Teflon bottle and is designed on the same principal as the quartz still except that a trough of Teflon was hung from the cold finger to catch the drops of hydrofluoric acid dropping from the condenser. Each "sub-boiling" still is housed in a clean air chamber, that is, a Class 100 hood, to protect the distillation process from external contamination. Figure 1 1 is a view of the laboratory for producing these high-purity acids. We produce enough high-purity acid in this facility to supply the en- tire Analytical Chemistry Division and some other laboratories with spe- cial needs. Table 5 shows the results of the analysis of sub-boiling distilled hydrochloric acid, the starting ACS reagent grade acid, and a lot of com- mercial high-purity acid. A summation of these impurity elements shows that the sub-boiling distilled acid contained 6.2 ppb, the ACS reagent grade 820 ppb, and the commercial high purity, 70 ppb. The only element found in the sub-boiling distilled acid at a level higher than 1 ppb was iron FEED BOTTLE DRAIN Figure 10. All-Teflon sub-boiling still. 132 Murphy Figure 1 1 . Laboratory for production of high-purity acids. 133 Accuracy in Trace Analysis Table 5. Impurity concentration in hydrochloric acid Sub- boiling ACS reagent Commercial distilled grade acid high purity (ng/g) (ng/g) (ng/g) Pb 0.07 0.5 <1 Tl .01 .1 — Ba .04 2 — Te .01 0.1 — Sn .05 .07 <6 In .01 — — Cd .02 .03 0.5 Ag .03 .05 .2 Sr .01 .05 — Zn .2 2 4 Cu .1 4 1 Ni .2 6 3 Fe 3 20 7 Cr 0.3 2 0.3 Ca .06 70 24 K .5 200 10 Mg .6 10 20 Na 1 500 — Total impurity 6.2 ppb 820 ppb 70 ppb at 3 ppb. The other elements were at the sub-ppb level, generally lower than 0. 1 ppb. The results for the analyses of sub-boiling distilled nitric acid, the ACS reagent grade acid, and a lot of commercial high-purity acid are shown in table 6. The totals of the impurity elements determined were 2.3 ppb for the sub-boiling distilled acid, 220 ppb for the ACS starting acid, and 240 ppb for the commercial high-purity acid. No element was found in the sub-boiling distilled acid a concentration greater than 1 ppb and only sodi- um was detected at that level. Most of the elements were in the 0.05 to 0.01 ppb range. Similar analyses for perchloric acid are shown in table 7. These analyses demonstrate the efficiency of sub-boiling distillation when the concentrations of impurity elements are compared to the starting acid. The concentration of barium which was greater than 1000 ppb was reduced to 0.1 ppb, a purification factor of greater than 10,000. Another example is calcium which was reduced to 0.2 ppb from 760 ppb in the starting acid, a factor of 3,800. It should also be noted that chromium which was present at 10 ppb in the starting acid was not significantly changed in the purified acid probably due to the formation of volatile 134 Murphy Table 6. Impurity concentration in nitric acid Sub-boiling ACS reagent Commercial distilled grade acid high purity (ng/g) (ng/g) (ng/g) Pb 0.02 0.2 0.3 Tl — .2 — Ba .01 8 — Te .01 0.1 — Sn .01 .1 1 In .01 — — Cd .01 0.1 0.2 Ag .1 .03 .1 Sr .01 2 — Se .09 0.2 — Zn .04 4 8 Cu .04 20 4 Ni .05 20 3 Fe .3 24 55 Cr .05 6 130 Ca .2 30 30 K .2 10 11 Mg .1 13 — Na 1 80 — Total impurity 2.3 ppb 220 ppb 240 ppb Table 7. Impurity concentration in perchloric acid Sub-boiling ACS reagent Commercial distilled grade acid high purity (ng/g) (ng/g) (ng/g) Pb 0.2 2 16 Tl .1 0.1 — Ba .1 >1000 10 Te .05 0.05 — Sn .3 .03 <1 Cd .05 0.1 4 Ag .1 0.1 0.5 Sr .02 14 — Zn .1 7 17 Cu .1 11 3 Ni .5 8 0.5 Fe 2 330 10 Cr 9 10 18 Ca 0.2 760 7 K .6 200 9 Mg .2 500 4 Na 2 600 — Total impurity 16 ppb > 3400 ppb 100 ppb 135 Accuracy in Trace Analysis Table 8. Impurity concentration in sulfuric acid Sub-boiling ACS reagent distilled grade acid (ng/g) (ng/g) Pb 0.6 0.5 Tl .1 .1 Ba .3 .2 Te .1 .1 Sn .2 .6 Cd .3 .2 Ag .3 .6 Sr .3 .4 Zn .5 2 Cu .2 6 Ni .2 0.5 Fe 7 6 Cr 0.2 0.2 Ca 2 123 K 4 9 Mg 2 4 Na 9 50 Total impurity 27ppb 200 ppb chromyl chloride, Cr0 2 Cl 2 . Chromium accounts for over half of the total impurities of 1 6 ppb found in the purified acid. The analyses of sub-boiling distilled sulfuric acid, shown in table 8, showed that it contained 27 ppb of the elements determined. Sodium and iron were the principal contaminants, accounting for over half of the total impurities. Hydrofluoric acid was purified using the all-Teflon still. Table 9 shows the results of the analyses of the starting ACS grade acid and the purified product. The total of the impurities in the sub-boiling distilled acid was 1 7 ppb compared to 320 ppb in the starting acid. The two principal impurities in the purified acid were calcium and chromium at 5 ppb, which accounted for over half the total impurities found. The situation for salts and other reagents is not as favorable as for the acids. Again, there are special high-purity grades of some of these re- agents available from commercial sources which are a major improvement over ACS reagent grade materials. However, there are relatively few of these reagents available and they often contain impurities at a level too high for use in low level trace analysis [22,23,37]. Hume [38] has stated in a recent article that natural sea water contains much lower concentra- tions of many heavy metals than artificial sea water made up from the 136 Murphy Table 9. Impurity concentration in hydrofluoric acid Sub-boiling ACS reagent distilled grade acid (ng/g) ' (ng/g) Pb 0.05 0.8 Tl .1 .2 Ba .1 .5 Te .05 .1 Sn .05 11 Cd .03 2 Ag .05 0.1 Sr .1 .5 Zn .2 4 Cu .2 3 Ni .3 12 Fe .6 110 Cr 5 20 Ca 5 14 K 1 28 Mg 2 10 Na 2 100 Total impurity 17ppb 320 ppb purest available reagent chemicals. An example of a high-purity, commer- cially available reagent that has met a need is phosphorus pentoxide. The specifications for the ACS reagent grade would allow for up to 100 ppm of lead. One supplier of high purity reagents has a limit of 0. 1 ppb for lead in this material and it has found wide use in the silica gel method for lead by isotope dilution mass spectrometry [39]. On the other hand, the specifications for high-purity sodium acetate allow a maximum of 0.05 ppm of lead. A total of 3 g of sodium acetate is sometimes used during the sample processing for the polarographic determination of lead, so the lead contamination could be as high as 150 ng Pb from this reagent alone. In cases like this the analyst will have to purify his reagents before use. The purification method used will depend on the nature of the reagent and the impurities to be removed. Mercury cathode electrolysis, extraction with dithizone or cupferron, ion-exchange, and crystallization have all found use in this regard [ 7 ,8 ,2 1 ,3 5 ] . C. Apparatus The third important source of contamination is that due to apparatus that comes into direct contact with the sample or sample solution, such as beakers hottles, filters, mortars, etc. 137 Accuracy in Trace Analysis Table 10 shows the results of Vasileyeskaya [23] for the analysis of high-purity hydrofluoric acid, hydrochloric acid, and nitric acid after evaporation in Teflon, platinum, and quartz containers. The results show higher trace element concentrations when the evaporations were per- formed in either platinum or quartz than in Teflon with the evaporation in quartz showing the highest values. Note the increase in aluminum, iron, calcium, and magnesium when the evaporations were performed in quartz compared to Teflon. Table 1 1 shows the data compiled by Hetherington, Stephenson, and Witherburn [40] for the maximum recorded impurity levels in various types of fused silica or quartz. There are two main types of vitreous silica, translucent or opaque quartz, and transparent quartz. Translucent quartz has the highest trace element concentrations and should be avoided. Transparent Types I and II quartz are made from naturally occurring quartz crystals or sands. Type I is made by electric melting and Type II by flame melting. Type II quartz has a somewhat lower trace element content because some impurities are partially volatilized in the flame. Type III quartz is a synthetic quartz made by vapor phase hydrolysis of pure silicon compounds such as SiCLj. It has a much lower trace element content than natural quartz with the exception of chloride which runs about 50 ppm and it contains about 0. 1 percent hydroxide. There is yet another type of transparent quartz, Type IV, which is also a synthetic quartz made by the oxidation of SiCL» and electrical fusion. It has about the same trace element content as Type III, is virtually free from hydrox- ide but has a chloride content of several hundred ppm. Notice that the principal impurities in Types I and II quartz are aluminum, calcium, iron, and magnesium, the same elements that increased when the pure acids were evaporated in quartz. Obviously synthetic quartz should be used whenever possible. One problem is that it is usually not available in the form of laboratory ware such as beakers, dishes, or crucibles. Table 12 shows the trace element concentrations of some container materials from the literature [7,48,53,54]. While variations of impurity content from batch to batch are to be expected, the compilation is useful for comparative purposes. Borosilicate glass can seriously contaminate solutions [ 8 ] and should be avoided unless the analyst is sure that the contamination of the element being determined is negligible such as for rhenium. In addition to these elements contamination from arsenic and lead have been reported. The quartz reported here is a Type 1 quartz made from the melting of pure domestic quartz sand which is claimed to be purer than the imported quartz crystal. Both polyethylene and Teflon are purer than natural quartz. The polyethylene reported here is produced by 138 a 5 s s i i -<3 s: I 4 5 I <3 ■s: c ZZ 2 od zzz .O ,* «o q« u « QQQ do ZHo ZZZ V © -« Z° N H © © — «"> Tt Tf IT) Q d w Z - — V V u e 1 V E 9 2°~ o M Z H Z -1 a < H c BO 2 ^ (S ,« © Q ** O 28 ^SS mO tNO CN © © 2 E 8 J IS 3 HI Uh E .- 3 N "S3 * a E _ 3 N §•£ S "5.2 3 s /n. Listy 55, 389 (1961). [27 ] Stegeman, H. , Z. A nal. Chem. 154, 267 ( 1 957). [28 ] Tiltqn, G. R., Patterson, C, Brown, H., Ingham, M., Hayden, R., Hess, D., and Lar- sen, E., Jr., Bull. Geol. Soc.Amer. 66, 1 131 (1955). [29] Tatsumoto, Mitsunobu,/*W. Chem. 41,2088 (1969). [30] Lancet, M. S. and Huey, J. M.,Anal. Chem. 46, 1360 (1974). [3 1 ] Irving, H. and Cox, J. J., Analyst 87, 207 (1962). [32] Kwestroo,W. and Visser,J.,/4/ia/yj/90,297 (1965). [33] Martinson, J. M.,/JW.C7i20 Sc 0.31 118 Th 0.088 2.5 Cr 5.4 0.51 Fe 530 30 Co 1.2 10 Sr 440 >300 Ag 0.24 7.7 Zn 12.3 3.2 Ba <35 Sb 0.64 6.0 Se 0.33 10 Na" 0.53 mg/g Cl° 0.84 mg/g Mn" 0.23 Mg/g " The results for Na, CI, and Mn were obtained on a liquid sample and are presented for informational purposes only. results because the water vapor which is frozen in the trap is not uniformly distributed on the glass surface but accu- mulates at a small area around the level of the liquid nitro- gen. If the cryogenic flask has a tight fitting cover, the ice in the trap will accumulate at the level of the cover rather than the level of the liquid nitrogen. Since such a small portion of the total capacity of the trap is used to freeze out the water vapor, the ice condensing from a large sample may collect until it fills the available space, thus blocking the path to the sample chamber. A standard size trap (used in the system described) which has 13-mm o.d. tubing and a 38-mm diameter body has to be changed twice when freeze drying a 100-g sample. The use of a Dry Ice-acetone mixture as a coolant does not alleviate the problem. To avoid the inconvenience of changing traps every 24 hours, an improved freeze-drying system has been devel- oped. This system is fitted with 75-mm traps which provide adequate area to freeze dry at least two 100-g water sam- ples consecutively before changing traps. Other improved features of this system are glass O-ring joints for all con- nections and Teflon valves in place of greased stop cocks. With these changes, contamination originating from con- 150 struction materials of the freeze dryer itself are minimized. The entire new system, with nine chambers for samples is shown in Figure 1. An expanded diagram of one of the units is shown in Figure 2. Since this freeze drier was installed in a Class 100 clean room facility, contamination of the dry residue from laboratory air particulates is minimized (15). Under normal laboratory conditions, it is recommended that air which is bled through inlet valves be filtered. It was previously stated that one of the advantages of this method of freeze drying water was the elimination of the need to transfer the dry residue. The bag containing the residue after freeze drying may be heat sealed, and then folded to make a small sample (1 cm X 2 cm X 0.5 cm) ready for irradiation and counting. The results of a typical analysis of dissolved species in water by INAA following the described preconcentration procedure appear in Table IV. This sample was taken from the Back River, just east of Baltimore, Md. The ratio of trace element content in the sample to the average trace el- ement content in a polyethylene bag is sufficiently high for most elements to result in reliable data, particularly con- sidering errors due to the sampling procedure alone. Work carried out in this laboratory to determine the variations in trace element concentration attributable to sampling tech- niques, suggest that such errors may easily exceed ±20%. The use of neutron activation for the analysis of trace el- ements in the aquatic environment is now well established. The multielement analyses of fresh to moderately saline water is greatly aided by the ability to handle relatively large samples with a minimum of pre-irradiation manipula- tion, as reported here. In addition, the use of freeze drying does not preclude subsequent post-irradiation chemical separations for elements that cannot be observed instru- mentally. LITERATURE CITED (1) R. H. Filby, W. A. Halter, and K. R. Shah, J. Radioanal. Chem., 5, 277 (1970). (2) H. D. Buenafama, (Conf-72 1069-3), 1972, 10 pp. Dep NTIS. (3) W. H. Zoller and G. E. Gordon, Anal. Chem., 42, 257 (1970). (4) A. J. De Groot, K. H. Zschuppe, M. De Bruin, J. P. W. Houtman, and P. Amin Singgih, "Modern Trends in Activation Analysis", J. R. DeVoe and P. D. LaFleur, Ed., Vol. I, 1969, p 62. (5) E. L. Kanabrocki, L. F. Case, L. Graham, T. Fields, Y. T. Oester, and E. Kaplan, J. Nucl. Med., 9, 478 (1968). (6) T. Mamuro, Y. Matsuda, A. Mizohaia, T. Matsunami, and T. Takeuchi, "Annual Report of the Radiation Center, Osaka Prefecture", Vol. II, 1970, p 23 (7) D. E. Robertson, "Pacific Northwest Laboratory Annual Report for 1972 to the USAEC Division of Biomedical and Environmental Research", Vol. II, part 2, p 24. (8) D. E. Becknell, R. H. Marsh, and W. Allie, Jr., Anal. Chem., 43, 1230 (1971). (9) S. Nagatsuka, H. Suzuki, and K. Nakajima, Radioisotopes, 20, 305 (1971). (10) T. Fujinaga, Y. Kusaka, M. Koyama, H. Tsuji, T. Mitsuji, S. Imai, J. Okuda, T. Takamatsu, and T. Ozaki, J. Radioanal. Chem., 13, 301 (1973). (11) D. F. Schutz and K. K. Turekian, Geochim. Cosmochim. Acta, 29, 259 (1965). (12) A. P. Grimanis, G. Pantazis, C. Papadopoulos, and N. Tsanas, "Third United Nations International Conference on the Peaceful Uses of Atomic Energy", 1964. (13) P. Schramel and K. Samshal, Int. J. Environ. Studies, 5, 37 (1973). (14) W. F. Merritt, "Proceedings of International Conference on Identification and Measurement of Environmental Pollutants", 1971, p 358. (15) T. J. Murphy, "Proceedings of 7th Materials Research Symposium", Na- tional Bureau of Standards, 1974. RECEIVED for review November 21, 1974. Accepted March 26, 1975. Certain commercial equipment, instruments, or materials are identified in this paper in order to adequately specify the 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. 151 APPENDIX IV REPRINTED IN PART FROM NBS TECHNICAL NOTE 548 ISSUED DECEMBER 1970 Nat. Bur. Stand. (U.S.), Tech. Note 548, 164 pages (Dec. 1970) CODEN: NBTNA Activation Analysis Section: Summary of Activities, July 1969 to June 1970 Philip D. LaFleur and Donald A. Becker, Editors Activation Analysis Section Analytical Chemistry Division Institute for Materials Research National Bureau of Standards Washington, D.C. 20234 *<"*.„ o« s 152 3. ACTIVATION ANALYSIS WITH THE NUCLEAR REACTOR A. Introduction During this year we have continued our efforts to attain a high degree of versatility in activation analysis and have spent much of our time in developing a greater expertise in nondestructive activation analysis, particularly coupled with computer data reduction. We intend, of course, to continue our efforts in obtaining the highest possible versatility in both destructive and nondestructive activation analysis and in the investigation of parameters affecting accuracy and precision in activation analysis. We have spent a great deal of time in characterizing the two pneumatic tube facilities which became available this year. All four of the originally installed facilities are now available and have been very carefully studied in terms of temperature effects, relative neutron fluxes and to a more limited extent on neutron energy spectra. We have maintained a very satisfactory relationship with the Reactor Operations Section and with the Health Physics Section. A greater portion of our activities this year than in past years has been spent in service analyses for organizations outside the Bureau of Standards and for other groups within the Bureau of Standards. Particularly notable are our efforts in the analysis of lasers and silicon chips. We are presently beginning a service analysis program for the Department of Agriculture, Involving the determination of mercury in wheat and peas and with the National Institutes of Health in the determination of iodine in blood serum samples. It is our hope that contacts with other governmental organizations in the Washington area, such as these, may continue and that interesting cooperative research programs may develop there- from. 153 B. Facilities 1 . Characterization of the NBSR for Neutron Activation Analysis Since the last report [1] much additional information has been obtained on the irradiation characteristics of the NBSR. This information has included thermal neutron flux variations within the irradiation areas, neutron energy distribution in the various irradiation positions, tempera- ture effects, and measurements of the gas production from several different biological materials. These different types of measurements will be treated separately here, with the techniques used presented along with the results obtained For comparative purposes, all results obtained to date will be included here. The current status of the NBSR irradiation facilities are given in table 1. The operational facilities include Table 1. NBSR irradiation facilities. Pneumatic Facilities Ident . Position RT-1 Core gap edge RT-2 In core RT-3 Core gap, D„0 RT-4 DpO reflector RT-V Vertical thimble RT-C } [Vertical thimble (RT-V-2) j (Converter) J-4 RT-T Thermal column Manual Access Facilities Vertical thimble In core, G-2 Vertical thimble Reflector, V-4 Access Attitude Diameter Status Horiz. 1.5 in. Operational Horiz. 1.5 in. Operational Horiz. 1.5 in. Operational Horiz. 1.5 in. Operational Vert. 0.75 in. Under const. Vert. 0.75 in. Under develop Vert. 0.75 in. Under const. Vert. 1.5 in. Under const. Vert. 1.5 in. Operational Vert. 1 in. Planning (7 Positions) 154 the four pneumatic tubes (fig. 1) and the manually accessed Long Term Irradiation Facility (LTIF) located in the G-2 v vertical thimble. The program of characterization has included measurements on all of these operational facilities, but with particular emphasis on the pneumatic facilities which are used for the majority of the activation analysis irradiations . a. Pneumatic Tube Irradiation Facilities The four operational pneumatic tube facilities have diameters of 1 1/2 inches, are powered by C0 ? , and termin in significantly different neutron environments (fig. 1). TANK WALL RT-1 RT- 2 D 2° Figure 1 RT- 3 Positions of the pneumatic tube terminals in the NBSR. (Terminology: The proximal end of the rabbit is that end closest to the reactor core vertical centerline; the distal end is that furthest from the core centerline. ) 155 Linear polyethylene rabbits, which have a capacity of approxi- mately 42 cm (2.6 cm dia x 8 cm long) are used in these faci- lities [1]. (1) Neutron Flux Measurements Initial measurements in the pneumatic tubes were for the thermal neutron flux. Of interest was the absolute thermal neu- tron flux, and the variation of this neutron flux within each irradiation position. Some measurements also have been made of the neutron energy distribution characteristics within each facility. ( 2 ) Thermal (Sub-cadmium) Neutron Flux The thermal (sub-cadmium) neutron flux measurements were made using the thin foil technique [2]. Two different detector foils, gold and copper, were used for these measure- ments. Copper foils were preferred, due to the absence of significant resonance peaks in its neutron cross section (fig. 2). Some measurements were also made on gold foils, COPPER GOLD COBALT NEUTRON ENERGY (EV) Figure 2. Neutron cross sections of copper, gold and cobalt as a function of neutron energy. 156 since gold is often used in other laboratories. However, since gold has significant neutron self-absorption problems, slight variations in foil thickness can lead to appreciable errors in flux determinations . The technique used consisted in irradiating a thin foil (0.005 in Cu foil; 0.001 in Au foil) of the element of interest in the appropriate facility, then counting the radioactive species formed on a large volume Ge(Li) detector. This detector has been calibrated for absolute photopeak counting efficiency using NBS calibrated radioactive sources . Neutron fluxes were then calculated from nuclear constants. The thermal (sub-cadmium) neutron flux was obtained after correction for epicadmium contribution to the foil activity. For copper, this correction did not exceed 7%, however, for gold this correction was in excess of 25%. No neutron self- shielding corrections were made. The results of the flux determinations are found in table 2 . It is apparent that the thermal neutron flux in tubes RT-1, RT-2 and RT-3 are very similar, with a signi- ficantly lower flux in RT-4. The gold foils gave higher flux numbers (relative to those from Cu measurements) for the tubes RT-1, RT-2 and RT-3, where a larger epicadmium neutron flux exists, but a value very close to that for copper in RT-4, where the neutrons are highly thermalized. Repeated determinations of these flux measurements gave results agreeing well with the original values. Table 2. Experimental neutron flux-NBSR pneumatic facilities Detector Foil Gold Copper Sub-Cadmium Neutron Flux-Proximal End i -2 -In ( n. cm sec ) RT-1 13 7.0x10 J ° 5.6xl0 13 RT-2 RT-3 RT-4 6.9xl0 13 5.8xl0 13 1.4xl0 13 6.3xl0 13 4.9xl0 13 1.3xl0 13 157 ( 3) Neutron Flux Variation Measurements In order to obtain accurate and precise results in neutron activation analysis using a nuclear reactor, the geometrical variation of the flux within each irradiation position must be accurately evaluated. Previous studies of the NRL Nuclear Reactor [3] revealed serious neutron flux variations within the sample irradiation area, and thus em- phasized the need for such measurements. The flux measurements for the NBSR were made on the inside dimensions of the polyethylene rabbits for the 1 1/2 in. pneumatic tube system (RT-1 to RT-4). Since the irradi- ation space available for each facility approximated a right circular cylinder 2.6 cm in diameter and 8 cm long, both the circumferential flux distributions' and the axial flux distributions were measured. The procedure developed used 3/16 in. disks of copper foil or 20 mil copper wire with gross gamma counting of the copper-64 radioactivity using an auto- matic sample changer. The copper activity was checked for extraneous radioactivities by counting on a 47cc Ge(Li) detec- tor and 4096 channel PHA and by following the decay character- istics of the foil on the gross gamma counter for over 8 half-lives. Both procedures revealed no detectable radio- activity in either the wire or the foil other than copper-64. All samples from a particular type of measurement were counted under identical conditions. The results obtained for the circumferential flux distri- bution of the pneumatic facilities are found in table 3- In all cases copper foils were used. These results are also shown graphically in figure 3- The foil placement dia- gram for the circumferential flux measurements is shown in figure 4. It can be readily seen from these results that there is no significant circumferential flux variation at any plane perpendicular to the axis of the right circular cylinder comprising the rabbit irradiation area. 158 Table 3. Circumferential flux distribution in NBSR pneumatic tube facilities. Pneumatic Facili Tube ty Location in Rabbit No. of Monitors RT-1 RT-1 RT-1 Proximal Middle Distal 7 7 7 RT-2 RT-2 RT-2 Proximal Middle Distal 7 6 7 RT-3 RT-3 RT-3 Proximal Middle Distal 7 6 7 RT-4 RT-4 RT-4 Proximal Middle Distal 7 7 6 Relative Neutron Flux 1.00010.003 0.947±0.004 0.871±0.007 1.000±0.005 0.902±0.002 0.874±0.003 1.000±0.004 0.928±0.003 0.829±0.003 1.000±0.010 0.787±0.007 0.603±O.Ol4 a ^Normalized to proximal end = 1.000. Uncertainties are ts//n at 95$ confidence level. POSITION RT-1 RT-1 RT-3 RT-4 Figure 3. Diagramatic representation of the circumfer- ential flux distribution in the NBSR pneumatic tubes at three positions inside the rabbit. 159 Figure 4. Copper foil placement for circumferential flux measurements in the NBSR pneumatic tubes . The axial flux measurements were made using the copper wire. A single length of the wire was centered along the rabbit axis and held securely in position with polyethylene foam. After irradiation, the wire was cut into 1 cm sections, weighed, and counted. The results, shown in figures 5 and 6, reveal a serious neutron flux dropoff along the axis of the rabbit. The magnitude of this dropoff over the length of the rabbit ranged from approximately 11$ for RT-1 1.00 X D .80 Z £ .60 D iu Z uj .40 > < -1 UJ K .20 ■._RT-4 I 2 3 4 5 6 7 8 DISTANCE FROM PROXIMAL END (CM) Figure 5. Axial flux distribution in the NBSR pneumatic tubes 160 2 3 4 5 6 DISTANCE FROM PROXIMAL END (CM) ure 6. Axial flux dropoff inside the rabbits (normal- ized to proximal end =1.00 for each pneumatic tube ) . over H0% for RT-4. The flux dropoff was linear in all umatic facilities except RT-2. ( 4 ) Neutron Energy Distribution Measurements Some preliminary neutron energy distribution measure- ts have been made, consisting almost entirely of cadmium io measurements . These cadmium ratios were made at both s of all four pneumatic facilities. Three different mental foils were used, 1 mil thick gold foil, 5 mil copper 1, and 2 mil cobalt foil. Interlocking 20 mil cadmium ers were used for the cadmium shielded measurements. The results of these measurements are found in table The values for gold and cobalt were obtained primarily 161 Table 4. Cadmium ratios for the NBSR pneumatic tube facility. Experimental Cadmium Ratios Pneumatic Facility Location In Rabbit Au(Cd) Cu(Cd) Co(Cd) RT-1 RT-1 Proximal Distal 7-3 10 46 64 29 RT-2 RT-2 Proximal Distal 3.7 4.0 15.7 15.8 11 RT-3 RT-3 Proximal Distal 10.3 16.5 65 80.5 42 RT-4 RT-4 Proximal Distal 87 111 536 582 395 a For elemental foils using 20 mil cadmium covers. for intercomparisons with other laboratories, since these two elements are the ones generally used for flux and cadmium ratio measuarements . However, as mentioned previously, the large resonance neutron cross sections (fig. 2) casts some doubt on usefulness of measurements using these foils. The copper foils, which behave as true "infinitely thin" foils throughout the neutron energy range seen in a nuclear reactor, should yield results which can be compared directly to neutron fluxes above and below the cadmium cutoff. Additional efforts will be expended in the direction of neutron energy measurements in these facilities by the use of threshold detectors. These threshold detectors yield integrated neutron fluxes above their reaction thresholds, and will be used to plot diagrams of the integral neutron flux vs. neutron energy similar to the work done by us at the NRL nuclear reactor [3]. (5) Temperature Measurements An extensive series of measurements were made in the pneumatic facilities using stick-on type temperature sensors. 162 These sensors were described previously [1] as were the various techniques employed and some preliminary results . Essentially all of the measurements have been completed at this time, and the results obtained are presented in tables 5, 6, 7 and 8. The equilibrium temperatures were lower than those expected, and therefore maximum irradiation times are based on factors other than temperature, e.g. radiation Table 5. Radiation damage and temperature measurements at 10 MW in RT-1 . Material and Configuration Empty rabbit Quartz ampoule, empty Lead in quartz ampoule Iron in quartz ampoule Length of Wt.(g) Irrad . 6 h Results Radiation damage not excessive^ 3 ■ 5g 20 m 170±10°F 3 ,6g Pb . 5g Quartz 20 m 225±5°F 3 . 6g Fe 5g Quartz 20 m <180°F ,20 minutes required to reach equilibrium temperature In this and succeeding tables, this phrase means the rabbit is not so brittle as to break upon deceleration in receiver Table 6. Radiation damage and temperature measurements at 10 MW in RT-2. Length of Materials and Configuration Wt . (g) Irrad. Results Empty rabbit - 20 m 155±5°F Empty 2 dram poly vial 3-25g poly 10 m 205±5°F Empty heat sealed sections 0.7g poly n n of poly tubing (total) 1U m ^ U ^ * 163 Table 7« Radiation damage and temperature measurements at 10 MW in RT-3 . Material and Configuration Empty rabbit Empty rabbit Lead pieces (6) in sealed quartz ampoule 'Quartz block Quartz block Lead and quartz - lead (single piece) in sealed quartz ampoule Empty rabbit, containing two 2/5 dram polyvials, with polyvial packing Le Wt.(g) I ngth of rrad . 20 m Results - <110°P — 8 h Radiation damage not excessive. 0.6g Pb 3-7g Quartz 4 h 195±5°F 6.0g 15 m 170±10°P 7.0g 20 m 190±10°F l.Og Pb 3.2g Quartz 10 m >220°F 8 h Radiation damage not excessive to rabbit or polyvials . Polyfoam crumbly . Table 8. Radiation damage and temperature measurements at 10 MW in RT-4. Material and Configuration Wt . (g) Empty rabbit Length of Irrad . 16 h Results Radiation damage not excessive Empty rabbit Lead in sealed quartz ampoule 0.6 g Pb 3.5 g quartz 16 m 15 m 105±5°P <130°P damage to the sample carrier. Irradiations of up to 8 hours are made routinely in the polyethylene rabbits at a thermal 13 -2 -1 neutron flux of 5x10 n.cm sec 164 Initially, reactor operations would permit temperatures of up to 200°F in the rabbit facilities. However, a few tests were made of the melting point of these rabbits and the effect of irradiation on the melting point. The results of these tests are given in table 9- As a result of these tests, the maximum permissable temperature in the pneumatic facilities was raised to 240°F. Table 9. Heat resistance of polyethylene rabbits: effect of irradiation time a . Length of Effect of Temperature (°F) Irradiation 13 230°F 240°F 260°F 280°F none none none slight completely softening melted 10 sec none none slight mostly softening melted 1 min none none slight completely softening deformed 10 min none none slight transparent softening and starting to deform 3. Melting point of conventional polyethylene is ^250°F; high b density polyethylene ^270°F. Rabbits irradiated in RT-1. ( 6) Pressure Measurements When the request was made for the irradiation of freeze dried biological materials in the NBSR, a number of questions were raised which could not be definitively answered. One of these questions involved the danger of explosion of a quartz ampoule due to the pressure generated by gaseous decomposition products of the material under irradiation. To effectively answer this question, and to establish criteria for the safe irradiation of a variety of aqueous solutions, 165 organic materials, and biological materials, a series of pres- sure measurements was initiated. The pressure testing system used to test irradiated samples is shown in figure 7. It consists of an appropriate sized rubber balloon, a section of polyethylene tubing which will fit inside the balloon but is still large enough to hold the irradiated quartz ampoule, and a gas-tight clamp for the open end of the balloon. To make a pressure measurement, an irradiated sample (inside the quartz ampoule) is inserted into the tubing and the balloon, the end of the balloon clamped shut and the ampoule struck smartly with a hammer. The polyethylene tubing prevents the shattered quartz from puncturing the balloon, and the pressure inside the ampoule estimated from the degree of expansion of the balloon. Tests with a hypodermic syringe (less needle) snugly fitted to the balloon opening indicated that 1/2 cm of gas could be readily detected. The quartz ampoules being tested had volumes of 1 cm or greater, therefore excess pressures of 1 atm could most certainly be measured. Since even poor quality glass blowing produces ampoules which can readily withstand 2 atm excess pressure, and good quality quartz glass blowing can produce ampoules which will withstand internal pressures in excess of 5 atm excess pressure, a preliminary limit for testing purposes of 1 1/2 atm excess pressure was established. RUBBER BALLOON CLAMP POLYETHYLENE TUBING Figure 7- Device used to measure pressures generated from irradiated samples . 166 The results of these pressure tests are found in table 10. Prom these results it can be seen that freeze-dried biological materials produce only small quantities of gaseous products upon irradiation in the pneumatic facilities RT-1, RT-3 and RT-4. For example, 1/2 gram of freeze-dried orchard leaves, irradiated for 6 hours in RT-3 at a neutron flux of 13 -2 -1 3 5x10 n.cm sec produced only about 1 cm of gaseous de- composition products (which produced approximately 1 atm of excess pressure in the ampoule size used) . These conditions of excess pressure are well within the tentative limits set up for these tests. Table 10. Pressure measurements on neutron irradiated samples . Sample Length Wt (g) of Irrad. Facility Results' Orchard leaves 0.25 freeze dried Alfalfa, freeze 0.25 dried Blood, freeze 0.25 dried Liver, freeze 0.25 dried Water, in heat- sealed poly- ethylene Orchard leaves, freeze dried Orchard Leaves, freeze dried 2.0 2 h 2 h 2 h 2 h 5 m RT-1 Small pressure buildup RT-3 No significant pressure buildup RT-3 No significant pressure buildup RT-3 No significant buildup RT-3 No significant pressure buildup 0.50 8 h RT-4 Small pressure buildup 0.50 6 h RT-3 Small pressure buildup a 3 Limit of detection ^1/2 cm J ; "small pressure buildup" estimated to be 1/2 - 1 cm3 . 167 b . Long Term Irradiation Facility (LTIF) The flux measurements made on the LTIF Vertical Thimble (see [4] for details of construction) included only neutron flux measurements and cadmium ratio determinations . Since this facility has its own recirculating DpO cooling system, the temperature measurements were not immediately essential. The measurement of pressures due to sample decomposition have not yet been made on this facility, at least partly due to the probable applicability of those measurements made with the pneumatic tubes. The flux variation measurements on this facility were made using two different techniques . For the vertical flux measurements a copper wire inside of sealed polyethylene tubing was quickly inserted the entire length of the vertical irradiation facility and irradiated for 30 m at a power level of 10 KW . At the end of the irradiation time, the copper wire was quickly removed from the reactor, the polyethylene tubing removed, and the copper wire cut into small pieces for gross y counting of the 12. 9h copper-64 radioactivity. At the lower end, where the activity level was relatively high, the wire was cut into 1/2 inch lengths . As the induced radio- activity decreased in the upper portions of the wire (above the core area), sections were increased to 1 inch, and finally to 6 inches at a point considerably above the upper core area. After this vertical flux diagram was obtained, the piece having the highest specific activity (from the core gap area) was taken for an absolute flux determination. Circumferential flux measurements were made during irradiations for the cadmium ratios. These measurements were taken using suitably weighted polyethylene rabbits at the end of a nylon fishing line. The copper wire was wrapped around the rabbit in which the bare copper foil was irradiated. The cadmium covered copper foil had its separate irradiation. This set of two irradiations was necessary for each area measured. Four areas were measured this way, as shown in 168 figure 8 and table 11. The results of these measurements are shown in figure 8 and tables 11 and 12. Figure 8 shows the vertical distribution of the total neutron flux in the LTIF, and demonstrates the effectiveness of the core gap area as a region of maximum thermal neutron flux. The maximum neutron NEUTRON FLUX DISTRIBUTION IN VERTICAL THIMBLE G-2 ^/ / UPPER CORE I \ -V CORE GAP LOWER CORE ,V ^T 100 90 80 70 60 50 40 30 20 10 PERCENT OF MAXIMUM FLUX 70 65 60 55 50 45 40 35 30 25 20 15 10 5 HEIGHT (inches) Figure 8. Vertical distribution of the total neutron flux in the NBSR LTIF. 169 Table 11. Cadmium ratios in the LTIF. Cadmium Ratio Area H eight (in. ) 0.5 (Copper) Below core 84.1 Lower core 11 32.8 Core gap 19.5 54.9 Above core 56 1830 Table 12. Circumferential flux variation in the NBSR LTIF irradiation facility. Location in LTIF Area Height (in) Location in Rabbit No Mon . of itors Relative . Neutron Flux On bottom On bottom On bottom 0.5 1.8 3.0 Bottom Middle Top 9 9 9 1.000±0.010 c 1.122±0.008 1.238±0.009 Lower core Lower core Lower core 11.5 12.8 14.0 Bottom Middle Top 9 9 9 1.000±0.018 1.055±0.026 1.088±0.015 Core gap Core gap Core gap 20.0 21.3 22.5 Bottom Middle Top 9 9 9 1.000±0.021 0.997±O.Ol6 0.980±O.Ol8 Upper core Upper core Upper core 28.5 29.8 31.0 Bottom Middle Top 9 9 9 1.000±0.08l 0.837±0.170 0.725±0.092 Above core Above core Above core 56.5 57.8 59.0 Bottom Middle Top 9 9 9 1.00010.011 0.885±0.013 0.774±0.011 L Measured from bottom of polyethylene rabbit (see figure 8) . Normalized to bottom = 1.00. Uncertainties are ts//n at 9' confidence level. 170 flux for 10 MW operation was found to be approximately 14 -2 -1 1.1x10 n.cm sec . The cadmium ratios are given in table 11. The core gap cadmium ratio of 55 is significantly higher than the core area, which has a cadmium ratio of approximately 33. All of these cadmium ratios are somewhat higher than would normally be expected, due to the highly thermalizing effect of the D ? filled thimble. The copper cadmium ratio of 1830 at a hight of 56 inches in the thimble 12 -2 -1 (with a 10 MW neutron flux of ^5x10 n.cm sec ) suggests the potential use of this facility for obtaining a highly thermalized neutron environment. The results of the circum- ferential flux distributions in the LTIF reveal a small but significant variation in the neutron flux from side-.to-side in the thimble. This variation is strongly dependent upon the area chosen for the irradiation, but usually is less than 10$ over the outer diameter of the polyethylene rabbits used for the measurements . c . Conclusions The NBSR has been characterized for many of its irradi- ation parameters, both for the pneumatic facilities RT-1, RT-2, RT-3 and RT-4, and for the LTIF (G-2 Vertical Thimble). While some additional measurements remain to be made, the irradi- ation facilities available for use are revealed as a set of extremely useful and convenient facilities under which a wide variety of irradiation environments can be obtained. (D. A. Becker and P. D. LaPleur) 171 APPENDIX V REPRINTED FROM JOURNAL OF RADIOANALYTICAL CHEMISTRY, Vol. 19, 1974, pp. 149-157. CHARACTERIZATION OF A NUCLEAR REACTOR FOR NEUTRON ACTIVATION ANALYSIS D. A. BECKER, P. D. LaFLEUR Activation Analysis Section, Analytical Chemistry Division, National Bureau of Standards, Washington, D. C. 20234 (USA) Ouelques caracteristiques du reacteur nucleaire NBSR utilise pour l'analyse par activation neutronique out 6te 6valuees; tes techniques de mesure deVeioppees sont d6crites. La discussion porte sur les caracteristiques suivantes: (1) le flux de neutrons thermiques absorbes par le cad- mium, (2) la distribution energetique des neutrons mesuree grace aux rapports cadmium et a des detecteurs a seuil. L'exces de pression dans les echantillons engendre Irradiation. Introduction Optimum utilization of a nuclear reactor for neutron activation analysis requires a care- ful and complete characterization of all the available irradiation facilities. This information is important both for reactor and personnel safety, and for complete control of the irradi- ation parameters affecting the precision and accuracy of the technique. When the NBS Nuclear Reactor (NBSR) started up several years ago, an extensive program of characteri- zation was initiated. During this reactor characterization a number of techniques and pro- cedures were developed which were felt to be of interest to other activation analysis labo- ratories. The design and operating characteristics of the NBSR have been reported previously. 1 ' 2 These reports also contain some of the preliminary results on the characterization studies. Due to space limitations, this paper will be limited to new or improved techniques devel- oped or results obtained during these characterization studies. The are presented in the hope that other activation analysis laboratories will find them useful in the evaluation and characterization of their own nuclear reactors. The techniques discussed below are the thermal (sub-cadmium) neutron flux determina- tion, neutron energy distribution measurements using cadmium ratios and threshold foil- detectors, and the determination of excess sample pressures generated during irradiation. J. Radioanal. Chem. 19 (1974) 172 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION Thermal (sub-cadmium) neutron fluxes In all measurements reported here, the thermal (sub -cadmium) neutron flux was deter- mined utilizing the thin foil technique, in which the measurement is made with minimum disturbance of the surrounding neutron environment. 3 For most of these measurements, reagent grade copper foil or wire was used. The advantages of this material include easy availability at adequate purities, low cost, and lack of high resonance neutron cross section peaks. Some measurements were also made with 0.0025 cm thick gold foils, but the neu- tron self-shielding occurring due to the very high resonance neutron cross sections made these results subject to large corrections. The technique used consisted in irradiating a thin foil or wire of the element of interest in the appropriate f acuity, then counting the radio- active species formed on a large volume Ge(Li) detector at a reproducible counting geom- etry. This detector had been calibrated for absolute photopeak counting efficiency at this geometry using NBS calibrated radioactive point sources. Neutron fluxes were then calculated from the nuclear datas shown in Table 1 . For the flux determinations using copper foils, care must be taken to insure total annihilation of the positrons in close prox- imity to the detector foil. In this case, covering the foil with approximately 0.1cm of alu- Table 1 Data for nuclear calculations 3 Effective Gam- Sam- Nuclear Isotopic Oeff threshold E7 ma-ray Half- Detector efficiency Sample ple reaction abundance (Oth). energy, abun- life form pu- MeV dance rity^ 63 Cu(n,7) 64Cu .691 (4.5b) thermal 0.511 0.38 12.8 h 0.00161 Cu foil 99.9+ ,97 Au(n,7)i98 Au 1.00 (98.8b) thermal 0.412 0.95 2.70 d 0.00200 Au foil 99.99+ 5S Ni(n,p) ssco 0.678 420 mb b 2.9 b 0.810 0.99 71.3 d 0.00096 Ni wire 99.98 S4 Fe(n,p) 54 Mn 0.0584 610 mb b 4.2 b 0.835 1.00 303 d 0.00093 Fe wire e 46 Ti(n,p) 46 Sc 0.0799 230 mb C 5.5 C 0.889 1.120 1.00 1.00 83.9 d 0.00087 0.00068 Ti wire e 24 Mg(n,p)24Na 0.786 60 mb b 6.3 b 1.369 1.120 1.00 1.00 15.0 h 0.00054 0.000235 Mg foil e 56 Fe(n,p) s6 Mn 0.917 110 mb b 7.5 b 8.4^ 0.847 0.99 2.58 h 0.000915 Fe wire e 27 Al(n,a) 24 Na 1.00 130 mb b 1.369 1.00 15.0 h 0.00054 Al wire 99.9+ 2.754 1.00 0.000235 Nuclear data taken from Ref. 4 unless otherwise noted. Data taken from Ref. s Data taken from Ref. 6 Experimentally determined using NBS calibrated radioactive sources; 47 cm 3 Ge(Li) detector at a distance of 5.0 cm. e See text. J. Radioanal. Chem. 19 (1974 J 173 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION minum was sufficient. Also it must be noted that the positron decay fraction of 19% for copper-64 leads to a 0.5 1 1 MeV photon abundance of 38% for the nuclear calculation. In all cases, the thermal (sub-cadmium) neutron flux was obtained after correction both for epicadmium contribution to the foil activity and for neutron self-shielding. For copper, the epicadmium correction did hot exceed 7% in the worst case (RT-2), however, for gold this correction was much greater. The results of these flux measurements for the pneumatic tubes are found in Table 2. This table includes values obtained using both the copper foils and the gold foils. It is appar- ent that the thermal neutron flux in tubes RT-1 , RT-2 and RT-3 is very similar, with a significantly lower flux in RT-4. As expected, the thermal column tube RT - T had a neu- tron flux two orders of magnitude lower. The gold foils gave higher flux numbers (relative to those from Cu measurements) for the tubes RT-1 , RT-2 and RT-3, where a larger epicadmium neutron flux exists, but a value very close to that for copper in RT-4, where the neutrons are highly thermalized. Several redeterminations of these flux measurements gave results agreeing well with the original values. A later, independent flux determination using a mul- tiple foil activation iterative method to evaluate the reactor neutron flux spectra, 5 produced a thermal neutron flux value for RT-4of 1.33-10 13 n-crn" 2 -sec' 1 , 7 in good agreement with the 1 .3* 10 13 n*cm" 2 -sec" 1 reported here using the copper foil and simple calculation from nuclear constants. Table 2 Experimental neutron flux-NBSR pneumatic facilities Maximum sub-cadmium neutron flux Detector n- cm" 2 -sec" 1 foil RT-1 RT-2 RT-3 RT-4 RT-T Copper 5.9-10 13 5.6-10 13 4.9-10 13 1.3-10 13 1.6-10 11 Gold 7.0-10 13 6.9-10 13 5.8-10 13 1.4-10 13 — Neutron energy distribution measurements Preliminary neutron energy distribution measurements consisted almost "entirely of cad- mium ratio measurements. 2 These cadmium ratios were made in all five pneumatic facilities. A total of three different elemental foils were used. They were 0.01 27 cm thick copper foil, 0.0025 cm thick gold foil, and 0.005 1 cm thick cobalt foil. Interlocking 0.05 1 cm cad- mium covers were used for the cadmium shielded measurements. The results of these /. Radioanal. Chem. 19 (1974) 174 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION Table 3 Cadmium ratios for the NBSR pneumatic tube facilities Pneumatic Location in rabbit Experimental cadmium ratios facility Cu(Cd) Au(Cd) Co(Cd) RT-1 Proximal 46 7.3 29 RT-1 Distal 64 10 - RT-2 Proximal 15.7 3.7 11 RT-2 Distal 15.8 4.0 - RT-3 Proximal 65 10.3 42 RT-3 Distal 80 16.5 - pT-4 Proximal 540 87 390 RT-4 Distal 580 111 - RT-T Bottom 3400 - - For elemental foils using 20 mil cadmium covers, where subcadmium + epicadmium neutrons Bare foil activity at T zero Cd Ratio = epicadmium neutrons Cd covered foil activity at T zero measurements are found in Table 3. The values for gold and cobalt were obtained primarily for intercomparisons with other laboratories, since these two elements are the ones gener- ally used for flux and cadmium ratio measurements. However, as mentioned previously, the large resonance neutron cross sections casts some doubt on the usefulness of measure- ments using these foils. The copper foils, which behave more closely to true infinitely thin" foils throughout the neutron energy range seen in a nuclear reactor, yield results which should be more representative of the neutron fluxes above and below the cadmium cut-off. The copper cadmium ratios for the pneumatic tube facilities (Table 3) range from less than 16 for RT- 2 to over 3000 for RT - T, the thermal column rabbit. Variations in the cadmium ratio from end-to-end of the irradiation container (rabbit) were especially large for both RT - 1 and RT - 3. More recently, a program has been initiated to extensively characterize the fast neutron spectra present in all the NBSR irradiation facihties. Besides the useful information this would inevitably provide for NBSR users, it was felt that an evaluation of suitable threshold foils in the well characterized NBSR facihties would assist other activation analysis labora- tories in their own reactor evaluation programs. /. Radioanal. Chem. 19 (1974) 175 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION At present six different nuclear reactions have been used to determine the integrated fast neutron flux, with threshold energies of 2.9 MeV to 8.4 MeV. These threshold foils and wires were irradiated in the pneumatic tubes RT - 1 through RT- 4, for periods of 15 seconds to 30 minutes. In all cases interlocking 0.102 cm thick cadmium covers were used to reduce the thermal neutron activation. In order to establish that no significant attenuation of fast neutrons occurred through the use of the cadmium covers, several experiments were made using elemental nickel. The nuclear reactions used were 64 Ni(n, 7) 6s Ni and s8 Ni(n,p) S8 Co, for thermal and fast flux measurements respectively. The results are shown in Table 4, and indicate that above the 2.9 MeV threshold for the fast neutron reaction, no significant neutron attenuation occurred due to the cadmium covers. This result is not complicated by cobalt-58 burn- up problems, since the conditions employed here result in less than 1% burn-up for a bare nickel foil. 9 The nuclear reactions and sample materials used for the neutron energy distributions as well as the nuclear data used for the calculations are found in Table 1 . After each ir- radiation, the radioisotope of interest was counted using the same caUb rated 47 cm 3 Ge(Li) semiconductor detector used for the thermal neutron flux determinations, and the fast flux calculated from the nuclear constants and threshold values given. Copper foil flux monitors were included with each irradiation at the opposite end of the rabbit from the cadmium to normalize the values obtained for each irradiation position. It was determined experimentally that the thermal neutron flux depression 8 cm distant from 6.5 grams of cadmium was only 2.0±0.5%. The purties of the foils and wires used are also given in Table 1 if known. Whenever the product nuclide could be formed by an (n, 7) nuclear reaction, the radioactivity produced both with and without a cadmium cover was observed. In all cases, with a simple post-irradiation wash of the sample material, no interference was noted from the (n,7) product. Initial work with the iron reaction forming manganese-56 did indeed re- Table 4 Variation of neutron flux with cadmium thickness using elemental nickel Wire No. Cadmium thickness Relative fast flux a Relative thermal flux 1 2 2 bare .020 in .040 in 1.00 0.981 0.987 1.000 .0033 .0011 a Absolute value 1.00 = 5.21 -10 9 n-cm" 2 -sec 1 ; integral neutron flux above 2.9 MeV for RT-4. /. Radioanal. Chem.19 (1974) 176 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION suit in a significant interference from, the (n,7) product, even with a cadmium cover. However, a sample of extremely pure iron wire was obtained which did not produce an interference, and this material was used for the threshold reactions. Analysis of the iron wire later revealed a concentration of only about 2 ppb manganese. The resulting integral neutron flux spectra for the four pneumatic tubes RT - 1 through RT - 4 are shown in Fig. 1 , along with a spectrum of a pure fission neutron flux. 10 The data points obtained fall very close to a straight line, with a slope similar to that of the fission spectrum. The fast flux levels observed for each of the pneumatic facilities are close to those expected, showing increasing thermalization of the neutron flux with distance from the core area. For comparison purposes, the only other fast neutron data available for the NBSR is that of McELROY et al., for RT - 4 from thermal up to 6 MeV. 8 His data show a very sim- ilar spectrum shape % with an absolute magnitude somewhat lower. At present, only data points for neutron energies greater than 2.9 MeV have been determined. Additional work is in progress to lower this to 1.0 MeV, using the U5 In(n,n') 115m In nuclear reaction, and even lower with other nuclear reactions. With the use of additional suitable threshold Unmoderated fission spectrum 01 234567891011 12 13 K Neutron energy, MeV Fig. 1 . Integral fast neutron flux for the NBSR pneumatic tube facilities /. Radioanal Chem. 19 (1974) 111 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION reactions, it should soon become quite simple to make 3 to 8 irradiations with readily ob- tainable materials, and obtain a reasonably good approximation of the fast neutron spectrum in any reactor faculty . Pressure measurements With the increased emphasis on the analysis of trace elements in organic materials, aqueous solutions, and freeze dried biological materials in the NBSR, a number of ques- tions were raised which could not be definitively answered. One of these questions in- volved the danger of explosion of a quartz ampoule due to the pressure generated by gaseous decomposition products of the material under irradiation. To effectively answer this question, and to establish criteria for the safe irradiation of such materials, a series of pressure measurements was initiated. The pressure testing system used to test irradiat- ed samples shown in i Fig. 2. It consists of an appropriate sized rubber balloon, a section of polyethylene or other tubing which will fit inside the balloon but is still large enough to hold the irradiated quartz ampoule, and a gas-tight clamp for the open end of the balloon. To make a pressure measurement, an irradiated sample (inside the quartz ampoule) is inserted into the tubing and the balloon, the end of the balloon clamped shut, and the Clamp Polyet- hylene tubing Sample Tjiyi 1 1 1 1 1 fy^T ■'I'lil.l Quartz ampoule Fig. 2. Irradiated sample pressure test apparatus using rubber balloon /. Radioanal Chem. 19 (1974) 178 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION ampoule struck smartly with a hammer. The polyethylene tubing prevents the shattering quartz from puncturing the balloon, and the pressure inside the ampoule estimated from the degree of expansion of the balloon. Tests with a hypodermic cyringe (less needle) snugly fitted to the balloon opening indicated that 0.5 cm 3 of gas could be readily detected. The quartz ampoules being tested had volumes of 1 cm 3 or greater, therefore excess pressures of 1 atm could most certainly be measured. Since even poor quality glass blowing produces ampoules which can readily withstand 2 atm excess pressure, and good quality quartz glass blowing can produce ampoules which will withstand internal pressures of more than 5 atm excess pressure, a preliminary limit for testing purposes of 1 .5 atm excess pressure was estab- lished. From the results of these pressure tests it can be seen that freeze-dried biological materials produce only small quantities of gaseous products upon irradiation in the pneu- matic facilities RT - 1, RT- 3 and RT- 4. For example, 1/2 gram of freeze-dried orchard leaves, irradiated for 6 hours in RT- 3 at a neutron flux of 5*10 13 n-cm" 2 'sec' 1 , pro- duced only about 1 cm 3 of gaseous decomposition products (which produced approxi- mately 1 atm of excess pressure in the ampoule size used). These conditions of excess pressure are well within the tentative limits set up for these tests, and in many cases permit irradiations of these materials in heat sealed polyethylene instead of quartz. 2 A modification of the balloon technique was recently developed, which permits a more quantitative measurement of the excess pressure in irradiated samples. This modi- fication uses heavy walled rubber tubing as the fracture chamber, with the volume change determined using a calibrated water manometer. Using this modified system, a large number of various materials are in the process of being checked for pressure gener- ation during irradiation. These results, when completed, will be reported in the literature. Conclusion In conclusion, the NBSR has been characterized for many of its irradiation param- eters,someof which are reported here. While additional measurements remain to be made, especially in the area of threshold foil measurements of neutron flux energies, the irradi- ation facilities available for use are revealed as a set of extremely useful and convenient facilities under which a wide variety of irradiation environments can be obtained. In addition, it is hoped that the procedures and techniques developed will be found to be useful in the characterization of other irradiation facilities. * The authors wish to acknowledge very substantial assistance of the NBS Reactor Operations staff in the performance of this work. They also wish to acknowledge the very helpful and able assistance of a summer intern, Mr. R. SPERBURG, for much of the threshold foil work reported here. /. Radioanal. Chem. 19 (1974) 179 D. A. BECKER, P. D. LaFLEUR: CHARACTERIZATION References 1. P. D. LaFLEUR, Editor, NBS Tech. Note 508, 1969. 2. P. D. LaFLEUR, D. A. BECKER, Editors, NBS Tech. Note 548, 1970. 3. ASTM Publication E-262-65T, ..Measuring Thermal Neutron Flux by Radioactivation Tech- niques," published by the American Society for Testing and Materials (ASTM), Philadelphia, Pa., 1965. 4. C. M. LEDERER, J. M. HOLLANDER, I. PERLMAN, Table of Isotopes, 6th ed. Wiley, 1967. 5. J. MOTEFF, Nucleonics, 20 (December 1962) 56. 6. Data taken from literature of Reactor Experiments, Inc., San Carlos, California. Not further identified as to source. 7. W. N. McELROY, J. A. ULSETH, S., BERG, G. GIAS, T. B. CROCKETT, ..Proceedings, Neu- tron Cross Sections and Technology," edited by D. T. Goldman, NBS Special Publication 299, Volume 1, 1968 p. 235. 8. W. N. McELROY, . L. S. KELLOGG, W. Y. MATSUMOTO, J. L. JACKSON, Hanford Eng. Develop. Lab. Progress Report, p. HEDL-1, June-July, 1971. 9. ASTM Publication E-264-65T, „Measuring Fast-Neutron Flux by Radioactivation of Nickel,' published by the American Society for Testing and Materials (ASTM), Philadelphia, Pa., 1965. 10. B. E. WATT, Phys. Rev., 87 (1952) 1037. /. Radioanal. Chem. 19 (1974) 180 APPENDIX VI Method for Determination of Mercury in Biological Materials by Neutron Activation Analysis Harry L. Rook, Thomas E. Gills, and Philip D. LaFleur Activation Analysis Section, Analytical Chemistry Division, National Bureau of Standards, Washington, D.C. 20234 A method is described for the determination of mer- cury in a variety of matrices. Three significant im- provements over currently used procedures have been achieved: The procedure has a minimum of chemical manipulations, thereby minimizing tech- nique related errors; the radiomercury separation is quantitative, thus eliminating unnecessary errors involved in the determination of chemical yields; the procedure is based on a simple combustion separation of radiomercury allowing the same technique to be used on a wide variety of matrices. The method was tested using radiotracer experiments and by analyzing samples of flour and coal which had been indepen- dently analyzed. The procedure has been subse- quently used to determine the mercury concentration of two new Standard Reference Materials being offered by the National Bureau of Standards. In recent years, mercury contamination of the environment has been recognized as a significant health problem. Because of the severe toxicity of some organic mercury compounds, notably methyl mercury, levels of less than 1 part per mil- lion (ppm) must be routinely determined to ensure the safety of human and animal life. Because of high sensitivity, neutron activation is an ideal technique for this analytical problem. Nondestructive neutron activation analysis has been reported for the determination of mercury above the 0.1 ppm level (1-4). However, both useful isotopes of mercury have low energy gamma emissions, usually necessitating a chemical separation of mercury from the activated matrix. Most separation methods reported in the literature involve acid dissolution followed by a chemical separation specific for mercury, coupled with Nal(Tl) counting (5-22). These (1) L. A. Rancitelli, J. A. Cooper, and R. W. Perkins, "Proceedings of the 1968 International Conference, Modern Trends in Ac- tivation Analysis," Nat. Bur. Stand. Spec. Publ., 312, 101 (1969). (2) W. A. Haller, R. Filby, L. A. Rancitelli, and J. A. Cooper, ibid., p 177. (3) R. H. Filby, A. L. Davis, K. R. Shah, and W. H. Haller, Mikrochim. Acta, 1970, 1130. (4) V. P. Guinn and R. Kishore, Paper presented at ANS Topical Meeting, Nuclear Methods in Environmental Research, Colum- bia, Mo., 1971. (5) K. Ljunggren and T. Westermark, Pure Appl. Chem., 1, 127 O960). (6) T. Westermark and B. Sjostrand, Int. J. Appl. Radiat. Isotopes, 9, 1 (1960). (7) C. Kellershohn, D. Comar, and C. LePoec, J. Lab. Clin. Med., 66, 168 (July 1965). (8) H. Hamaguchi, R. Kuroda, and K. Hosohara, J. At. Energy Soc. Jap., 2, 317(1960). (9) C. K. Kim and J. Silverman, Anal. Chem., 37, 1616 (1965). (10) D. Comar, C. LePoec, M. Joly, and C. Kellershohn, Bull. Soc. Chim. Fr., 1, 56 (1962). (11) H. Smith, Anal. Chem., 35, 635 (1963). (12) B. Sjostrand, il.d., 36, 814 (1964). (13) K. Samsahl, ibid., 39, 1480(1967). (14) H. A. Das, J. G. VanRaaphorst, D. Hoede, and J. Zunderhuis, Int. J. Appl. Radiat. Isotopes, 17, 252 (1966). (15) H. D. Livingston, H. Smith, and N. Stojanovic, Talanta, 14, 505 (1967). (16) T. H. Handley, Anal. Chem., 36, 153 (1964). (17) R. Kuroda, T. Kiriyama, and K. Ishida, Anal. Chim. Acta, 40, 305 (1968). methods were usually developed for a specific matrix type and are not easily adapted to changed composition. This type of activation analysis is often time-consuming, requires yield corrections, and gives results which are often technique- dependent. Recent separation methods reported use combustion tech- niques with mercury absorbed either on selenium paper (23) or in various solutions (24). In each case, however, the separated mercury was again required to be radiochemically pure. In this work, a rapid, quantitative separation of volatile elements was carried out followed by counting using a high efficiency, high resolution Ge(Li) detector and gamma ray spectrometer. Using this method, samples of flour and coal containing from 0.015 to 5 ppm were analyzed to test the accuracy of the procedure. Samples of Standard Reference Material 1571 and 1577, Orchard Leaves and Beef Liver, were also analyzed for mercury, and concentrations of 0.155 ppm and 0.0145 ppm, respectively, were found. It is of interest to note that excellent agreement was obtained later by flameless atomic absorption spectrometry and isotope dilution spark source mass spectrometry. EXPERIMENTAL Apparatus and Reagents. The sample combustion train (Figure 1) was made from an outer 19-38 standard taper Vycor joint which had been extended to a total length of 30 cm. The mercury condenser was made from an inner 19/38 standard taper Vycor joint, extended to 25 cm, which was in- serted through a liquid nitrogen cold trap made from a small polyurethane acid shipping container. Standard solutions were made by dissolving a weighed quantity of triply distilled metallic mercury in 1 : 1 high purity nitric acid and diluting to volume. The concentration used was 1.0065 mg/ml which was dilute enough to prevent any significant neutron self absorption, but concentrated enough to negate loss due to wall absorption of mercury in the storage container. Procedure. For analysis, 0.5-gram samples were en- capsulated in cleaned, heat treated quartz tubes. Standard solutions were encapsulated in a like manner. Five sample tubes were arranged in a concentric ring with a standard in the center in a polyethylene rabbit and irradiated in the pneumatic transfer facility RT-3 of the NBS Reactor for 2 to 4 hours at a thermal neutron flux of approximately 6 X 10"n cm -2 sec -1 . Flux monitors were not used as the radial flux variation is less (18) F. Kukula and M. Kirvanek, Isotopenpraxis, 4, 57 (1968). (19) G. V. S. Rayuda, B. Tiefenbach and R. E. Jervis, Trans. Amer. Nucl. Soc, 11, 54 (1968). (20) O. Johansen and E. Steinnes, Int. J. Appl. Radiat. Isotopes, 20,715(1969). (21) B. R. Petersen, Dan. Kemi., 49 (11), 171 (1968). (22) E. Steinnes and O. Johansen, Nord. Hyg. Tidskr., 50(2), 71 (1969). (23) L. Kosta and A. R. Byrne, Talanta, 16, 1297 (1969). (24) R. R. Ruch and H. J. Gluskoter, Environ. Geol. Notes, 43, (Feb. 1971). Reprinted from ANALYTICAL CHEMISTRY, Vol. 44, Page 1 114, June 1972 American Chemical Society i o i than ±2% in RT-3. The samples were allowed to decay 3 days to minimize personnel radiation exposures. The quartz tubes were washed in 1 : 1 nitric acid, cooled to liquid nitrogen temperature, and broken 5 mm from the top. The samples were weighed into a ceramic combustion boat, and 25 mg of mercuric oxide carrier added. The boat was inserted into the combustion chamber, an oxygen flow of ~30 cc/min passed over the sample and the sample ignited with an oxygen-gas torch.- The sample was allowed to burn freely with the downstream quartz wool plug heated to red heat to ensure complete combustion of volatile hydrocarbons and carbon produced during the sample burn. After combustion ceased, the ash was heated to approximately 800 °C for 5 minutes and the entire combustion tube flamed down to the standard taper joint to drive all volatile components of the sample into the condenser. The combustion tube was allowed to cool for 5 minutes, the liquid nitrogen cold trap removed, and the condenser placed joint end down into a 4-oz polyethylene bottle. The con- denser was warmed to room temperature and the mercury washed into the bottle with 10 ml of concentrated nitric acid and three 10-ml portions of distilled water. The total volume of solution was adjusted to 50 ml and counted on a 60-cc Ge(Li) detector. The 77-keV gamma ray of mercury-197 was used for the determination. Primary standards were made by sealing 1-ml aliquots of the previously mentioned standard solution in quartz vials and irradiating them in the same configuration as that of the sam- ple. After irradiation, the standard was frozen in liquid nitro- gen, broken in a beaker, allowed to melt, and decanted into a 1-liter volumetric flask. The broken quartz vial and beaker were washed with six 5-ml portions of distilled water and the final volume adjusted to 1 liter. One milliliter of the dilute solution was pipetted into a 4-oz polyethylene bottle and di- luted to 50 ml. The specific activity at the end of irradiation was compared to that of the sample for quantitative results. A method of additions analysis also was carried out to deter- mine if mercury was lost during analysis. This procedure is described in a following section. Bromine Separation. The only isotope to interfere after the combustion separation is bromine-82 and then only when its concentration is at least a factor of 100 times that of mer- cury. To eliminate this interference, either of two courses of action may be employed. Bromine-82 has a half-life of 35.3 hours while mercury-197 has a half-life of 65 hours. Thus, allowing the sample to decay 10 to 1 5 days optimizes the reduc- tion of bromine activity while maintaining usually adequate mercury sensitivity. If decay does not give the necessary im- provement of the mercury signal relative to bromine, a silver bromide separation can be carried out with quantitative re- covery of mercury-197 due to the relatively high solubility of mercury bromide in warm water. The procedure is as fol- lows: The mercury-197 plus carrier was washed from the con- denser into a 40-ml centrifuge tube with 10 ml of concentrated HN0 3 followed by four 5-ml portions of distilled water. The solution was warmed to 40 °C in a hot sand bath, 2 mg of bromide carrier were added, and bromide was precipitated with AgN0 3 . The solution and precipitate are stirred at 40 °C to ensure complete dissolution of HgBr 2 and precipita- tion of AgBr. The solution was centrifuged, decanted into a 4-oz polyethylene bottle, and the walls of the tube were washed down with 3 ml of 2/V HN0 3 . The precipitate was washed twice with 2-ml portions of warm 2/V HN0 3 followed by centrifugation. The wash solutions were added to the original solution. The final solution was adjusted to 50-ml total volume and counted as before. A reduction to less than 5 % of the initial bromine activity was obtained with quantitative retention of mercury-197 in the solution. Method Evaluation. The method was first tested for com- plete recovery of mercury by burning non-irradiated samples of orchard leaves with mercury-203 tracer added. The re- a, INLET SAMPLE 4a M 1 ppm), samples were analyzed nondestructively by irradiation followed by direct counting of the mercury-197. These samples were then analyzed by the described combustion technique and the results compared. The analyses were carried out on the following samples: a flour sample prepared by the Inter- national Atomic Energy Agency which had been doped with approximately 5 ppm of mercury ; a sample of flour contain- ing approximately 0.01 ppm of mercury analyzed by the U.S. Food and Drug Administration by both atomic absorption and activation analysis and by the Food and Drug Directorate Table IV. Mercury in Beef Liver Hg, ppm 0.0140 0.0141 0.0118 0.0140 0.0173 0.0137 0.0152 0.0138 0.0169 Mean = 0.0145. 2 s = 0.0034. ts/Vn = 0.0013. of Canada using atomic absorption. The results of these analyses are given in Table I. With the procedure giving satisfactory results, an intensive analytical program was carried out on a new Standard Ref- erence Material — SRM 1571, Orchard Leaves. The intent was to determine the mercury content using two or more non-related analytical techniques with an accuracy in the final results of better than ±10% at a 95% confidence level. A total of 1 1 samples from three different bottles was analyzed by the procedure described. The samples were found to contain 0.155 ± 0.006 ppm of mercury (Table II). The absolute value of the mercury concentration was checked by carrying out a method of additions analysis on the orchard leaves. Four 40-gram samples of the base leaves were spiked with standard mercury solutions to give samples containing approximately 0.0, 0.5, 0.5, and 1.0 ng added mercury per gram of leaf, respectively. The samples were freeze dried, tumbled for four hours, and mixed in a ball mill for a minimum of 6 hours. Two extra samples were pre- pared in the same manner and spiked with mercury-203 tracer to ascertain that no losses occurred during freeze drying and that the mixing was uniform. The tracer experiments in- dicated complete mixing with an inhomogeneity of less than ±1%. Aliquots of each group were analyzed by the described procedure and the base concentration of mercury in the orchard leaves was determined by fitting a least squares regression to the individual results. The least squares fit gave a base concentration of 0.161 ± 0.010 ppm of mercury. The individual results are given in Table HI. Table III. Method of Addition of Hg in Orchard Leaves Hg calculated by [ PPM Hg"|_r PPM Hg "1 Found J LCalculatedJ Sample Hg added, ppm Hg found, ppm least squares, ppm 5-1 0.151 0.161 -0.0096 5-2 0.156 0.161 -0.0046 5-3 0.169 0.161 +0.0084 5-4 154 0.161 -0.0066 5-5 0.154 0.161 -0.0066 1-1 0.495 0.699 0.667 +0.0316 1-2 0.495 665 0.667 -0.0023 4-1 0.485 0.621 0.657 -0.0361 4-2 0.485 0.702 0.657 +0.0439 2-1 1.012 1.168 1.196 -0.285 2-2 1.012 1.207 1.196 +0.0104 Y = A + B x. A = 0.1606 ± 0.910. B = 1.024 ± 0.0201. Base mercury content = 0.161 ± 0.010. 183 The mercury concentration in the SRM Orchard Leaves also was determined using flameless atomic absorption spec- trometry and isotope dilution spark source mass spectrom- etry. The results obtained by these techniques were 0.160 ± 0.012 and 0.145 ± 0.022 ppm of mercury, respectively. A second new Standard Reference Material, Beef Liver, with a considerably lower mercury content was also analyzed by this technique. Nine samples were analyzed using the bromine separation. The samples contained 0.0145 ± 0.0034 ppm of mercury. The individual results are given in Table IV. The mercury concentration in the SRM Beef Liver was determined using flameless atomic absorption spectrometry. The results obtained by this technique were 0.0160 ± 0.0012 ppm of mercury. ACKNOWLEDGMENT The cooperation of T. C. Rains, P. J. Paulsen, and Robert Alvarez is gratefully acknowledged. Their investigations and analyses gave special credence to this work . The Authors would also like to thank H. P. Yule for his help in the data reduction. Received for review October 26, 1971. Accepted February 7, 1972. In order to specify procedures adequately, it has been necessary to identify commercial materials and equipment in this report. 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. 184 APPENDIX VII Simultaneous Determination of Trace Elements in Platinum by Isotope Dilution and Spark Source Mass Spectrometry Robert Alvarez, Paul J. Paulsen, and Daniel E. Kelleher Spectrochemical Analysis Section, National Bureau of Standards, Washington, D. C. 20234 An isotope dilution technique using the spark source mass spectrograph was investigated for applicability to simultaneous multi-element trace determinations. Isotopically altered Ag, Cu, Ni, Pb, and Pd were sepa- rated from a 250-mg sample of platinum by cation exchange, and electrodeposited onto gold electrodes for sparking in the spectrograph. Electrodeposition provided the spiked analytes in an adherent form, relatively free of anions and organic residue from the ion exchange resins, both of which could complicate the spectrum. Contamination, the principal limitation of the method, was minimized by isothermally distilling the reagent acids and by isolating the operations from the laboratory environment. After determining the isotope ratios, the concentrations were computed, ranging from 0.5 ppm (/ig/gram) for Ni to 14 ppm for Pb. Improved precision and accuracy was obtained compared to conventional spark source analysis. The results were compared with those of other methods and the platinum material was issued by NBS as Standard Reference Material 681. Stable isotope dilution analyses are usually performed using mass spectrometers with thermal ionization sources. These methods depend on the chemical form of the element for optimum results and are generally not suitable for simul- taneous multi-element trace determinations. This is particu- larly true if elements having high ionization potentials are to be included in the analysis. Webster (/) presents an excellent description of the general method including applications. The spark source mass spectrograph has been employed by Leipziger (2) to determine copper in Nickel Oxide NBS Standard Reference Material (SRM) 673 and antimony in Tin SRM 835 using isotope dilution techniques. The average values obtained, 21 and 121 ppm, respectively, agreed with the NBS certificate values. Because the high voltage spark source ionizes all elements with approximately the same efficiency, isotope dilution spark source mass spectrometry (ID-SSMS) should enable the simultaneous determination of many trace elements in a sam- ple. The object of this investigation was to apply this method to determine five trace elements in a sample of platinum. Isotopically altered Ag, Cu, Ni, Pb, and Pd were separated from the platinum matrix by cation exchange and eluted with HC1 into an electrolysis cell. After neutralizing the acid, the spiked analytes were electrodeposited onto gold wires and sparked in the spectrograph. The mass spectra were photo- graphically recorded, isotopic ratios determined, and the results calculated. After comparing these results with those of other analytical methods, this platinum material, previously examined for homogeneity, was subsequently issued by the National Bureau of Standards as SRM 681. The authors have also developed an ID-SSMS method for determining trace elements in each of two NBS zinc standard reference materials (J). Compared to this zinc method, the (1) R. K. Webster in "Advances in Mass Spectrometry," J. D. Wal- dron, Ed., Pergamon Press, New York, N. Y., 1959, pp 103-119. (2) F. D. Leipziger, Anal. C'hem., 37, 171 (1965). (3) P. J. Paulsen, R. Alvarez, and D. E. Kelleher, Spectrochim. Acta, in press. PLATINUM WIRE ANODE. ANODE COMPARTMENT POROUS VYCOR ROD -GOL0 WIRE CATHODES Figure 1. Cell for acid neutralization and electrodeposition platinum procedure required a greater number of analytical operations using special techniques and apparatus to reduce contamination to an acceptable level. EXPERIMENTAL Reagents. Deionized water is made immediately before use by passing distilled water from the laboratory distribution system through a mixed bed resin column. HC1 and HN0 3 are prepared by isothermal distillation. Aqueous ammonia is prepared by bubbling NH 3 into deionized H 2 0, after passing the gas into a gas washing bottle. Evaporator for Sample Solutions. A borosilicate glass chamber, similar to the one described by Thiers (4), provides a clean environment for evaporating solutions. The ap- paratus is fabricated from a crystallizing dish 150 mm in diameter by 75 mm high, and a culture dish of the same diameter. A lateral displacement of the coincident edges of the two units is prevented by three lengths of glass rod that are fused to the outside of the culture dish (lower unit) and project upward. Filtered nitrogen is introduced through a tube sealed at the center of the crystallizing dish, and escapes from between the edges of the modified dishes to prevent entry of particulate contaminants. The gas is filtered by a cellulose filter which has a mean pore size of 0.1 /zm and is supported by porous polypropylene disks in a Nylon filter holder. Glove Box. A fiber glass glove box is employed to reduce airborne contamination during ion exchange and miscellane- ous operations. Ion Exchange Column. The column consists of an 8-mm i.d. borosilicate glass tubing drawn to a fine tip, containing a resin bed 4 cm in height. The resin has a 100 to 200 mesh size and is of an analytical-grade, sulfonic acid type with an 8% divinylbenzene crosslinkage. Before use, the column is (4) R. E. Thiers in "Trace Analysis," John H. Yoe and H. J. Koch, Eds., John Wiley and Sons, Inc., New York, N. Y., 1957, p 637. Reprinted from ANALYTICAL CHEMISTRY, Vol. 41, Page 955, June 1969 185 Table I. Natural and Spike Isotopic Abundance of Isotopes Isotopic abundance Natural abundance, in spike, Isotope atomic per cent atomic per cent 107 Ag 51.83 0.9 109 Ag 48.17 99.1 63 Cu 69.17 0.30 "Cu 30.83 99.7 M Ni 67.88 0.38 6 2 N j 3.66 99.02 108 Pc J 26.71 10.6 nopd 11.81 87.5 206p(ja — 92.15 J0ep o a — 1.255 ° Lead exhibits isotopic variability in nature and is therefore not listed. Table II. Trace Element Concentrations in Platinum Standard Reference Material 681 Determined by Isotope Dilution Analysis Element Ag Cu Pd Pb Ni Concentration in ppm (Mg/gram) 2.09 5.04 5.92 13.6 0.46 No. of samples 7 6 4 6 4 95% Confidence limits in ppm for the concentration" ±0.15 ±0.54 ±0.32 ±0.9 ±0.06 • These confidence limits are based upon tS-; where S- is an esti- mate of the uncertainty in the average concentration and is based upon the observed variation between the k samples. The t is Student's t factor for 95% confidence and k — 1 degrees of freedom. pretreated with 50 ml of 6A/ HC1 and washed with 5 ml of water. Neutralization and Electrolysis Cell. The cell shown in Figure 1 was designed for neutralization of the acid eluate and subsequent electrodeposition of the isotopically-altered trace elements. The components are fabricated of polytetra- fluoroethylene (PTFE) except as otherwise indicated. The cell body has a 25-mm i.d. and a volume of 13 ml. For the neutralization step, the cylindrical compartment, shown suspended by a reentrant centerpost threaded into the lid, is filled with 1 ml of aqueous NH 3 . This assembly is removed prior to electrodeposition. Two high-purity gold wires, inserted into a removable PTFE plug, serve as the cathode during electrodeposition. When this plug is fitted tightly into the body of the cell, the 0.7-mm diameter cathode wires project 2 mm into the cell body. An isolated anode compart- ment is formed by inserting a 3.5-mm diameter by 6-mm long porous Vycor (5) rod into an undersize hole at the side of the cell and connecting this rod to a glass reservoir with plastic tubing. A platinum wire makes an electrical contact with diluted HN0 3 added to this compartment as an electrolyte. During electrodeposition, the cells are housed in a plastic enclosure under a slight positive pressure of nitrogen to minimize airborne contamination. Enriched Isotope ("Spike") Solutions. With the exception of ^Pb, the enriched isotope materials in amounts equivalent to 10 mg of the element were obtained from the Oak Ridge National Laboratory. Solutions containing one mg per ml of 109 Ag, 65 Cu, 62 Ni, and 110 Pd were prepared by weighing each material to ±0.02 mg, dissolving it in the appropriate acid, and diluting the solution to volume in a 10-ml volumetric flask. The possible presence of impurities in these materials (5) R. A. Durst, J. Chem. Educ., 43, 437 (1966). and the small sample weights used made it advisable to verify the solution concentrations. However, because these solu- tions were diluted to add the required submicrogram additions of the spikes to the sample solution, the concentrations of the diluted spike solutions were determined by isotope dilution as follows: Measured volumes containing 5 Mg of each spike were mixed with solutions containing known amounts of the normal elements. The pure clement additions were com- puted from Equation 1 to yield isotope ratios approximating one for all elements except Ag (see Calculations and Results). The solution was electrolyzed for one hour at a current of 2 mA in the previously described cell. The gold wire elec- trodes were then sparked, the isotope ratios were determined, and the spike weights were computed from Equation 1. A reliable 206 Pb solution could be prepared by weighing 100.0 mg of the NBS radiogenic lead isotopic standard, SRM 983 (6), dissolving it in HNO3, and diluting to volume with water in a 100-ml volumetric flask. Mass Spectrograph. A double-focusing Mattauch-Herzog design, spark source instrument (Consolidated Electrodynam- ics Model 21-110) that had been modified to give 22 exposures per plate is operated at an ion accelerator voltage of 24 kV. Microphotometer Analog Computer. A microphotometer having a precision screw (Jarrell-Ash Model 23-500) and equipped with an analog computer to measure peak areas (7) is used. A function generator, which is the basic com- ponent of the analog computer, provides an output pro- portional to relative ion intensities when the input is a voltage proportional to per cent absorption. An integrator follows the function generator to give signals proportional to peak intensity areas as the line is scanned. Procedure. A 250-mg sample of platinum was treated with warm aqua regia for several minutes and rinsed in water to remove possible surface contamination. The sample was weighed into a 25-ml quartz erlenmeyer flask and heated with 3 ml of HC1 and 1 ml of HNO3 until dissolved. After adding measured volumes of the spike solutions, the platinum solution was transferred to a 20-ml low form, quartz crucible and evaporated almost to dryness in the chamber described. The resulting salts were dissolved in 5 ml of H 2 and the solution was passed through the cation exchange column using the full flow rate of the column. During this step, the isotopically altered trace elements were adsorbed on the resin and the chloroplatinate solution was the effluent. The column was washed with 4 ml of H 2 0, and eluted with 5 ml of 3M HC1 into the electrolysis cell after the washings had been discarded. To neutralize the acid eluate with a minimum of contamination, aqueous ammonia was pipetted into the cylindrical compartment and the cell was closed with its tightly fitted lid. This distillation step yielded a small solution volume that was favorable for the electrodeposition of the analytes. After removal of the cylindrical compart- ment, the acidity of the solution was adjusted to read approxi- mately 6 with indicating pH paper. The acidity is not critical. The solution was then electrolyzed for 5 hours at a current of 2 mA, while stirring at approximately 200 rpm. After with- drawing the solution, the gold electrodes were rinsed in water and transferred to the mass spectrograph. A graded series of exposures was made with the pair of electrodes at integrated beam currents of 0.3 X 10 -10 to 1 X 10~ 7 coulomb to obtain optimum intensities for the isotope lines of the various amounts of electrodeposited elements. By moving one electrode so as to expose a fresh surface, the electrodeposit was consumed more effectively and a greater signal was obtained. The ion-sensitive photo- graphic plates (Ilford Q-2), were developed in a high contrast (6) E. J. Catanzaro, T. J. Murphy, W. R. Shields, and E. L. Garner, J. Res. Natl. Bur. Stand., 72A, 261 (1968). (7) P. J. Paulsen and P. E. Branch, NBS Technical Note 401, Washington, D. C, Sept. 1966. 186 Element Ag Cu Pd Pb Ni Isotope dilution 2.1 5.0 5.9 13.6 0.46 Table III. Comparison of Results by Isotope Dilution with Other Analytical Methods for Platinum Standard Reference Material 681 Concentration in ppm Gig/gram) Activation Optical emission spectrochemical' analysis" Polarography 6 1.8 5.0 5.6 5.2 10.3 Lab A 1.4 6.3 5.0 8.5 <1 LabB 2.0 3.0 6.1 15.0 N.D.<< " Activation analyses by Michel Crambes and Gilbert W. Smith, NBS Analytical Chemistry Division. 6 Polarographic analyses by E. June Maienthal, NBS Analytical Chemistry Division. c Spectrochemical analyses by outside laboratories cooperating with NBS. d N.D. Not Detected. LabC 2 2 4.0 10 0.8 LabD 12.3 0.44 developer (Kodak D-19) for 2'/ 2 min, placed in a short stop solution for 10 sec, fixed for 1 min, washed 15 min in tap water, and rinsed in distilled water. The intensity areas of the lines for the isotopes listed in Table I were measured and their ratios were calculated. From 6 to 20 ratio measure- ments were made for each element depending on the avail- ability of lines having suitable intensity. To correct the determination for contamination, the entire analytical procedure was performed using the same sample size of high-purity platinum, SRM 680. A direct mass spectrographic comparison of SRM 680 with SRM 681 had indicated that the elements of interest except for Ni were substantially lower in SRM 680 than in the material being analyzed. Consequently, to obtain favorable intensity ratios, the SRM 680 solution was spiked with one tenth of the amount added to SRM 681 and both materials were analyzed concurrently. Use of the higher-purity matrix to evaluate contamination, which did not exceed 10% of the reported values, yielded more reproducible results than were obtained by spiking the reagents alone. For Ni however, this latter procedure had to be used. CALCULATIONS AND RESULTS The concentrations were computed from Equation 1. C = WK(A, B SP R) M (BR - A) 0) where C is the concentration in ppm (/ig/gram), W is the weight of isotopically enriched material ("spike") added in fig, M is the weight of sample in grams, A and B are the natural abundances of the analyte isotopes a and b, A sv and B sp are the abundances of isotopes a and b in the spike, R is the measured altered ratio of isotope a to isotope b, and K is the ratio of the natural atomic weight to the atomic weight of the spike. In the method, it is desirable to alter the isotopic composi- tions of the analytes to obtain isotope ratios, R, approximating one whenever possible. This minimizes the effect of photo- graphic emulsion calibration errors, and could be done for all analytes except Ag. For that element, 109 Ag spike was added to change the ia9 Ag/ 107 Ag ratio from its natural value of 0.93 to 2. To obtain the desired R values, the spike additions were computed from Equation 1 using the elemental concen- trations estimated from a direct sparking of the platinum sample. This photographically recorded spectrum was also used to determine the 206 Pb and 208 Pb natural abundances in the platinum material. These determinations were necessary because of the isotopic variability of Pb in nature (6). For the other analytes, an experimental determination of the natural abundances of the isotopes was not required ; and the values listed in Table I, supplied by the NBS Analytical Mass Spectrometry Section, were used. The isotopic analyses provided by the Oak Ridge National Laboratory for its isotopically enriched materials and by Catanzaro et al., (6) for the radiogenic lead isotopic standard, SRM 983, are also listed in this table. These isotope analyses are of sufficient accuracy that errors from this source do not significantly affect our results. In all, seven samples were run, the first three using experi- mental conditions that did not satisfactorily preconcentrate all of the elements. However, using the experimental condi- tions described under Procedure, the last four samples yielded all of the analytes for simultaneous determination. The elemental concentrations obtained for the first three samples ase also included in the results. The determination of Ag was satisfactory for all of the samples. Table II lists the average concentrations obtained, the number of samples, and the 95 % confidence limits. A comparison of the isotope dilution results with the results obtained by other analytical methods is given in Table III. Except for Pb, the agreement among methods having high precision was satisfactory. DISCUSSION AND CONCLUSIONS Unlike most trace analytical methods, isotope dilution analysis does not require analyzed or synthesized standards of the same matrix material for reliable determinations. As applied to spark source mass spectrometry, the simplest procedure consists of evaporating a portion of an isotopically spiked solution of the sample on a suitable substrate for sparking in the spectrometer. Compared to direct sparking of the sample, poorer limits of detection will be obtained. Moreover, a trace element may be undeterminable if the only isotope lines suitable for measurement are interfered with by isotope lines of the matrix elements or complex ionized species of the matrix and anion fragments. For example, copper could not be determined in platinum by this direct-evaporation technique because one of its two stable isotopes, 65 Cu + , is interfered by ln ¥t*+. To obtain improved limits of detection and remove these interferences, the isotopically spiked analytes can be separated from the matrix before sparking. A preconcentration pro- cedure should be selected that minimizes contamination which is the principal limitation of these methods. These proce- dures can be relatively simple or incompletely characterized because determinations by isotope dilution methods are unaffected by analyte losses once equilibrium has been established with the added spikes. However, to obtain isotope lines of sufficient intensity for precise ratio measure- ment, a high recovery of the spiked analytes is desirable. 187 Concentration limits of detection comparable and even superior to those obtained by the direct sparking technique are possible if contamination can be reduced to an acceptable value. Although the experimental procedure should be designed to achieve a low method blank, its variability is an important factor affecting the precision of the results. There- fore, the least number of manipulations is desirable and should be performed in a clean environment to reduce spurious con- tamination. In the platinum method, contamination can occur during sample dissolution, evaporation, ion exchange, and transfer of the spiked analytes to the substrate for spark- ing. Gold is a suitable substrate for it does not react with the solution being electrolyzed ; it is mononuclidic thus producing a simple spectrum; and it is obtainable in high purity. The gold wire that was used has the composition of NBS Standard Reference Material 685-W listed in the Certificate of Analysis. For the dissolution step, minimal quantities of isothermally distilled acids are used. The contribution to the total blank from the reagents is relatively constant and can be corrected for. The spiked platinum sample solution is evaporated in a quartz vessel surrounded by filtered nitrogen. The ion exchange step is performed in a glove box using methyl- methacrylate rods and holders to support the ion exchange columns. The last stage, during which the submicrogram amounts of the analytes are being transferred to the wire sub- strate for sparking, is particularly sensitive to spurious con- tamination. Although the eluate solution can be evaporated to a small volume and pipetted onto the gold wires, it is preferable to electrodeposit the analytes. This procedure yielded an adherent form for sparking, relatively free of anions and organic residue from the resin, both of which would complicate the spectrum. Moreover, electrodeposition required less handling and, therefore, was less subject to contamination. The electrolysis cell design incorporated a replaceable porous Vycor plug to prevent cross-contamina- tion from the eluate solutions. To determine elements that are not ordinarily electrodepos- ited, evaporation of the eluate would have been required. In this case, the electrodeposition step could be followed by evaporation and possible ignition of the residue on fresh electrodes. Ignition would remove the organic residue, one of the disadvantages of this evaporation procedure. In conclusion, isotope dilution techniques provide spark source mass spectrometry with methods for independently analyzing homogeneous materials that can serve as standards for the direct sparking comparison procedures. In general, these standards of the matrix either are not available or often have values that are not as precise as required. Although ID-SSMS is not a rapid method (for the procedure developed, four samples required approximately 8 days to run), its capability of simultaneous multi-element determinations and applicability to most of the elements having two or more stable isotopes, makes it very attractive for reliable trace element determinations. Received for review March 3, 1969. Accepted April 4, 1969. Presented in part before the Division of Analytical Chemistry, 155th National Meeting, ACS, San Francisco, Calif., April 1, 1968. Certain commercial materials and equipment are identified in this paper in order to adequately specify the 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 avail- able for the purpose. 188 APPENDIX VIII Reprinted from SPECTROCHIMICA ACTA, Vol. 24B, pp. 535-544. Pergamon Press 1969, Printed in Northern Ireland. Determination of trace elements in zinc by isotope dilution spark source mass spectrometry* Paul J. Paulsen, Robert Alvarez and Daniel E. Kelleher Spectrochemical Analysis Section, Analytical Chemistry Division, National Bureau of Standards, Washington, D.C. 20234, U.S.A. (Received 11 April 1969) Abstract — The applicability of the isotope dilution technique in spark source mass spectrometry has been investigated for the determination of Pb, Cu, Cd, Ag, Tl, and Sn in zinc metal. In the procedure developed, a solution of the zinc is spiked with known amounts of stable isotopes of the elements to be determined, and the elements are electrodeposited onto gold wires that subsequently serve as electrodes for the mass spectrographic isotopic analysis. Chemical operations, including isothermal distillation of reagents, dissolution of the zinc, and the electro- deposition step are conducted in a closed system to minimize contamination. The results of the analysis of two NBS zinc standard reference materials include determinations ranging from 11 ppm (11 /zg/g) for Pb to as low as 0.02 ppm (20 ng/g) for Sn. It is concluded that this method is of general utility for simultaneous multi-element trace analysis and is especially applicable in providing much needed standardization in spark source mass spectrometry. Introduction Spark source mass spectrometry (SSMS) is potentially capable of determining all elements in a sample at ppbf concentrations. In practice this capability is limited by specific interferences and background generated by the matrix being analyzed, instrument memory, and instrument background for gases in the source (C, N, 0, CO, hydrocarbons, etc.). The accuracy of analysis depends on the availability of suitable reference standards for each material to be analyzed. If standards are not available, the analysis is based on estimated relative sensitivity factors for the elements; this type of analysis will have an uncertainty of a factor of three or more. If a standard is available [1] or if the sample is amenable to the preparation of synthetic standards [2], accuracies to within 5-10% can be obtained under favorable conditions. For many materials, however, standards of the same matrix are not readily available and, for high-purity metals, synthetic standards are not practical. This situation points up the need both for standards at the ppm level suitable for SSMS and for additional techniques that can give accurate analyses without comparison to a reference standard. Stable isotope dilution analysis using SSMS to measure isotopic ratios offers the possibility of meeting these needs. With this technique the sample to be analyzed is mixed with a known amount of a separated stable isotope (spike) of the element to be determined. The spike must have a known isotopic composition significantly different from that of the element in the sample. Measurement of the * Contribution of the National Bureau of Standards, not subject to copyright. f In this paper the abbreviations ppm and ppb represent respectively, parts per million by weight (,ag/g) and parts per billion by weight (ng/g). [1] R. K. Skogerboe, A. T. Kashtjba and G. H. Morrison, Anal. Chem. 40, 1096 (1968). [2] G. D. Nicholls, A. L. Graham, E. Williams and M. Wood, Anal. Chem. 39, 584 (1967). 189 Paul J. Paulsen, Robert Alvarez and Daniel E. Kelleher resulting ratio of isotopes in the mixture by mass spectrometry, enables the cal- culation of the amount of the element present in the original sample. Analysis by stable isotope dilution requires no reference standards, and is specific and sen- sitive. It can be applied to most elements having two or more stable isotopes. Thermal ion source mass spectrometry has been applied to stable isotopic dilution analysis for many years and is a widely accepted means of analysis [3]. In applying the method to SSMS the procedure is basically the same. However, the generation of ions by a high voltage r.f. spark combined with photographic plate detection of ions gives the SSMS considerably different capabilities in terms of sensitivity and element coverage than the thermal ion instruments in which ions are detected electronically. Stable isotope dilution analysis with SSMS has been applied by Leipziger [4] to the determination of Cu in NBS 673 nickel oxide and Sb in NBS 835 tin standard reference materials. The object of this work has been to develop a procedure to apply the techniques of preconcentration, combined with stable isotope dilution and SSMS, to the determination of ppm and lower level trace elements. Of particular interest is its application to the analysis of Standard Reference Materials (SRM) which could serve as composition standards for SSMS. Analysis of zinc samples proposed as NBS standard reference materials, SRM 728 and SRM 683, offered an opportunity for comparison of results of SSMS isotope dilution with those of other trace analytical techniques. Experimental Reagents Distilled water from the building distribution system was passed through a mixed bed ion exchanger prior to use. Hydrochloric acid was isothermally distilled just before it was used to dissolve the zinc samples. Open beakers of A.C.S. reagent grade HC1 were placed in a sealed container overnight with a quartz dish containing deionized water. This procedure yielded approximately 9M HC1 in the quartz container when 3 vol. of HC1 were used for 1 vol. of water. The separated stable isotopic spikes used in these experiments were obtained from Oak Ridge National Laboratory. The material selected for spiking an element is always an enriched sample of the isotope that has a low natural abundance. When an element has more than two isotopes, the selection of the spiking isotope is based on freedom from interferences from isotopes of other elements, its abundance with respect to the major isotope of the element, the available isotopic purity of the enriched isotope, and its cost. Enriched 206 Pb, 203 T1, 117 Sn, m Cd, 109 Ag, and 65 Cu were obtained in 10 mg quantities and stock spike solutions were prepared at concentrations of approximately 1 mg/ml using accurately weighed samples. A.C.S. reagent grade HC1 and HN0 3 were used to prepare the spike solutions. Subsequent dilutions of these spike stock solutions with distilled water were so large that impurities from these reagents could not contribute significantly to the analysis blanks. [3] R. K. Webster, Advances in Mass Spectrometry (Edited by J. D. Waldron), pp. 103-119. Pergamon Press (1959). [4] F. D. Leipziger, Anal. Chem. 37, 171 (1965). 190 Determination of trace elements in zinc The diluted stock solutions employed for analysis were assayed by reverse spiking to determine the concentration of the element. In this operation, solutions containing known amounts of the pure natural elements were mixed with precisely measured volumes of the spike solutions, and the altered isotopic ratios were measured. The procedures and calculations were similar to those for the zinc analysis described under Procedure. Analysis of the diluted solutions avoids errors that might occur during the dilution step. Electrodeposition cell Figure 1 shows the electrodeposition cell used to preconcentrate the trace elements in zinc. The tetrafluoroethylene polymer (Teflon) cell had been employed previously [5] to electrodeposit silver onto the surface of a spectroscopic graphite electrode. For the present work the graphite electrode was replaced with a plug of the same diameter through which two high purity (6-9's grade) gold wires, 0.63 mm dia., were inserted to serve as cathodes. The anode compartment con- sists of a 15 mm length of porous Vycor sealed to Vycor tubing. The compartment contains dilute HN0 3 and a high purity platinum wire as anode. The electro- deposition cell is protected from external contamination by placing it inside a small plastic box having a flow of filtered nitrogen gas, and mounting this box and its external stirring motor inside a portable clean box through which filtered air is passed. Clean environments for all stages of sample handling are provided by either clean boxes or covered containers flushed with filtered nitrogen. Spark source mass spectrograph A double-focussing Mattauch-Herzog design, spark source instrument (modi- fied Consolidated Electrodynamics Model 21-110 [6]) was operated at an ion accelerator voltage of 24 kV. The ion source pumping system is a combination of an ion pump and a titanium sublimation pump. The instrument has been modified to give 22 exposures per plate. Microphotometer A microphotometer (Jarrell-Ash Model 23-500) having a precision screw had been equipped in our laboratories with an analog computer to measure peak areas [7]. The analog computer is basically a function generator the output of which can be adjusted to conform to the plate response curve to ions. Thus when its input is a voltage proportional to percent absorption profile of a line image from the photo- graphic plate, its output is a voltage proportional to the intensity profile of the ions that generated the image on the plate. The function generator is followed by an [5] R. Alvarez, Electrodeposition of trace constituents as a preconcentration technique for optical emission spectroscopy — deposition of silver from zinc solution. Symp. Trace Characterization, National Bureau of Standards, October 3-7 (1966). [6] P. J. Paulsen and P. E. Branch, Instrument modifications. NBS Tech. Note 286, pp. 2-6 (1966); NBS Tech. Note 401, pp. 70-73 (1966); NBS Tech. Note 422, pp. 50-55 (1968); P.J. Paulsen and D. E. Kelleheb, NBS Tech. Note 452, pp. 45-47 (1968). [7] P. J. Paulsen and P. E. Branch, Electronic analog computer for measuring intensity areas pf peaks. NBS Tech. Note 401, pp. 62-70 (1966). 191 Paot J. Paulsen, Robert Alvarez and Daniel E. Kelleher Stirrer- Platinum anode a Anode /^"'compartment Gold cathodes- lin I cm Fig. 1. Electrodeposition cell. integrator which measures the peak area under the intensity profile. The area thus recorded is proportional to the total number of ions generating the line image. Preliminary experiments Since most of the trace elements of interest in the zinc samples are more noble than zinc, electrodeposition was selected as a means of preconcentration. Electro- deposition of a neutralized zinc solution obtained by dissolution of the zinc in HN0 3 yielded a deposit which, when analyzed by SSMS, proved to be mostly zinc containing increased concentrations of the trace elements Ag and Cu. Electrodeposition of a sample obtained by dissolution of the zinc in HC1 yielded a zinc deposit with Pb, Tl, Sn, Cd, Ag, and Cu highly enriched over their concentrations in the original zinc sample. For this reason HC1 was selected as the solvent. Procedure One gram samples of zinc were etched lightly ( < 1 % removal) in dilute HC1 to remove surface impurities, rinsed in distilled water, dried, and weighed. The samples were then dissolved in a minimum volume of HC1 (about 4 ml HC1 to 1 g of zinc) that had been purified by isothermal distillation. Measured quantities of the separated stable isotopes were added to the zinc samples while they were in the early stages of 192 Determination of trace elements in zinc dissolution. The solution was placed in the electrodeposition cell where the more noble impurities were electrodeposited onto two high purity gold cathodes by passage of a 2 mA current for 2-4 hr. At the end of the electrodeposition period the sample solution was removed and the gold cathodes were washed with distilled water with the cell potential still applied. The gold cathodes were then loaded into the spark source mass spectrograph. Two plates (Ilford Q-2) were exposed by sparking half the available plated surface area of the gold cathodes for each plate, using a graded series of exposures starting with an integrated beam current of 0.3 x 10 -10 coulombs and increasing to 1 x 10 -7 coulombs. The electrode surfaces were moved during the exposures to spark as much of the deposit as possible. The combination of electrode movement and increasing exposure levels insured isotope ratio measurements over a wide range of impurity concentration levels as well as optimum usage of all the available sample for elements present in low amounts. The plates were developed in high contrast developer (Kodak D-19) for 2\ min, placed in short stop solution for 10 sec, fixed for 1 min, washed 15 min in tap water and rinsed in distilled water. One plate was selected for isotope ratio measurements based on the level of line intensities and the relative freedom from fog and scratches on the plate. The isotope ratios measured were Pb 208/206*, Tl 205/203*, Sn 120/117*, Cd 112/111*, Ag 109*/107, and Cu 65*/63 (*indicates the stable isotopic spike). In general the ratios of the singly charged ions were measured ; however, ratios of multiply charged ions also were measured when appropriate to obtain additional data. The formula for calculating concentrations is shown in equation (1). . W K (A av - B av R) ppm (by weight) = - — _ - (1) In this equation W is the weight of the stable isotopic spike in //g, M is the weight of the sample in grams, R is the measured "altered" isotope ratio (isotope a/isotope b), K is the ratio of natural atomic weight/atomic weight of the spike, A and B are the natural abundances (atomic fraction) of isotopes a and b, and .4 8p , B 8p are the abun- dances (atomic fraction) of isotopes a, b in the spike. This is a general equation which applies to elements with two and more isotopes. Calculation of concentration from the measured isotopic ratios normally requires two additional measured quantities: (a) the weight of the sample and (b) the weight of the separated stable isotope added. The Oak Ridge values for the isotope abun- dances in the enriched spike, which accompany each spike purchased, and table values of natural abundances were used in calculation. The one exception was for lead, the abundance of which varies in nature [8]. The natural isotopic abundance of the lead in the zinc was measured from SSMS spectra of the original zinc sample. When the natural ratio of the major isotope to the minor isotope (the spiking isotope) is approximately two or larger the amount of spike isotope added was adjusted to give a final ratio in the spiked sample of close to one. This procedure minimizes the errors in the determination of isotope ratios that may arise from errors in the photographic plate calibration. If the natural ratio is close to one, as is the [8] R. D. Russell and R. M. Farquhar, Lead Isotopes in Geology. Interscience (1960). 193 Paul J. Paulsen, Robert Alvarez and Daniel E. Kelleher case for silver, the element was spiked to give a ratio of approximately two. An initial spike of the sample was made based on the results of a normal SSMS analysis of the sample. The concentrations found from this spiking were used to calculate the exact amount of spikes to add in order to obtain the optimum isotope ratios for measurement. Reagent and system blanks were evaluated by adding stable isotopic spikes to a zinc sample of much higher purity and following the same analysis procedure. All elements except tin gave blanks too small to affect the results. An evaluation of the tin blank by this technique was not possible since a direct analysis of the higher purity zinc by the spark source showed it to have almost the same tin content as the samples being analyzed. Blank determinations using the same procedures and volumes of reagents as in a normal determination but without the dissolution of a zinc sample gave inconclusive results. The time required for carrying out this procedure in practice has been estimated. We assume that reagents, spike solutions, and apparatus have been readied for the analysis and a group of 4—6 samples is to be run. The time required is 2-3 days per sample for the six listed elements, or less than one-half day per element for each sample. Results Table 1 gives the results of analysis of the two ends of two zinc bars for silver and tin. The two sets of values listed are from completely independent runs starting with new spiking solutions. After the first set of results for tin was obtained (run 1) there was some concern that sorption losses could have occurred from the solution used to make the tin 117 spike addition. This solution contained 0.04 ^g/mloftin 117 and 1 ml of this solution was added to provide only 4 x 10~ 8 g of tin 117 to the zinc sample. In the second set of samples (run 2) two modifications were made in order to minimize any possible sorption losses. First, high purity zinc was added to the tin 117 spike to serve as a carrier with 100 parts zinc to 1 part tin. This ratio was selected so that the impurities from this zinc sample would be too low to cause interferences with any of the elements being determined. Second, the tin spike additions were made from a 10-fold higher concentration of tin, 0.4 fig /ml, using only 0.1 ml. All spikes were added to the samples immediately after being diluted from the 0.1 mg/ml working solutions. The agreement between the two sets of tin results indicated that sorption losses were not a problem for either spike concentra- tion. This was further confirmed when the 0.4 /ig/ml solution was analyzed by reverse spiking with pure tin and found to be 93% of its original concentration 1 1 days after it was used to spike the samples. Three lots of zinc in the form of shot (SRM 728) and both ends of two zinc bars (SRM 683) were analyzed for Pb, Tl, Sn, Cd, Ag, and Cu. Both the shot and the bars were found to be homogeneous for all elements tested within the precision limits of the method. Therefore, the data for each standard were pooled to give an average value. The analyses of both the zinc bar and the zinc shot by isotope dilution SSMS are given in Table 2. Determinations of Pb, Cd, Cu, and Ag by other techniques are also listed for comparison. The close agreement between the isotope dilution results 194 Determination of trace elements in zinc Table 1. Independent duplicate determinations of Ag and Sn in zinc SRM 683 (bar form) Bar 1 Bar 2 End A EndB (ppm by End A wt.) End B Silver Run 1 1.18 1.06 1.13 1.02 Run 2 1.14 1.18 1.33 1.25 Average 1.16 1.12 1.23 1.14 Tin Run 1 0.019 0.018 0.013 0.016 Run 2 0.023 0.019 0.014 0.016 Average 0.021 0.019 0.014 0.016 Table 2. Comparison of results by isotopic dilution and other analytical techniques for zinc standard reference materials 683 and 728 SRM 683 SRM 728 Isotope dilution Other Isotope dilution Other SSMS techniques (ppm SSMS by wt.) techniques Lead 11.2 ± 0.5* 10.8f 11.3 ± 0.4 10.8f Copper 5.6 ± 0.3 6.4f 5.5 ± 0.5 5.8f Cadmium 1.17 ± 0.09 1.12f 1.13 ± 0.07 1.13f Silver 1.16 ± 0.12 1.46$ 1.16 ± 0.05 1.09$ Thallium 0.17 2 ± 0.015 0.16 3 ± 0.01 2 Tin 0.017 ± 0.001 0.021 ± 0.009 * The ± values are the 95 % confidence limits for the limiting mean expressed in ppm. This is equal to t 95 S\/n, where S is the standard deviation of the mean and n is the number of independent determinations, ranging from 6 to 8, and t is student's t for 95 % confidence. f Polarography, E. J. Maienthal (NBS). % Neutron activation analysis, B. A. Thompson and D. A. Becker (NBS). for the two standards for all six elements arises from the fact that both were produced from a common lot of zinc. This agreement is also a measure of the precision of the isotope dilution technique. Good agreement was found with the other methods of analysis for all determinations except the Cu and Ag in SRM 683. These differences on Cu and Ag are not excessive considering the low concentrations involved. It should be noted that isotope dilution SSMS provided values for six of the seven elements that were certified for these two standards. 195 Paul J. Paulsen, Robert Alvarez and Daniel E. Kelleher Discussion Preconcentration Measurement of the isotopic ratios for the impurities in the zinc without the electrodeposition-preconcentration step is a possibility. Direct evaporation of the zinc chloride solution onto high purity gold wires might serve this purpose ; however, the amount of sample on the surface of the gold wires would be relatively small compared to that available when zinc metal is sparked directly. Larger samples could be loaded but this would probably involve pelletizing large amounts of zinc chloride with a high purity conducting powder (Ag, Au, graphite, etc.) ; this binder would in effect dilute the zinc chloride sample. Either system of loading the sample for SSMS measurement of isotopic ratios would result in poorer sensitivity for the isotope dilution analysis than would exist in a direct analysis of the metal by SSMS. It is estimated that sample evaporation onto gold wires would give detection limits of about 1-10 ppm for isotope dilution as compared to 1-10 ppb found for a direct SSMS analysis. The chloride ions introduced in the dissolution step act to dilute the zinc sample as well as to introduce additional lines into the mass spectrum. An example of the complications that these additional lines could cause was observed when the electro- deposition procedures were being investigated. Zinc plated electrodes which were not washed with distilled water retained sufficient zinc chloride on their surfaces to contribute significantly to the spectrum. The mass spectra of these samples contained lines for ZnCl+, ZnOCl+, and ZnC10H+, as well as C1+. The Z11OCI+ lines were intense enough to interfere with the tin lines which have the same nominal mass. Washing the electrodeposit with distilled water removed this interference. Electro- deposition thus served two functions: it removed the chloride interferences and it increased the sensitivity of the method. Preconcentration in part compensates for the sensitivity lost because of the sample dissolution which is necessary to perform the spiking step. Preconcentration of a 10 ppb impurity from a lg sample into a consumable sample size of about 10 mg or less would provide 1 x 10~ 8 g of the impurity. This is 1000 times the 1 x 10 -11 g that gives a just detectable line and should give sufficient signal for measuring isotopic ratios on several exposures with reasonable precision. These figures would predict quantitative determinations down to 10 ppb for most elements with preconcentration, provided of course that reagent and system blanks did not set a higher limit. In our experiments a comparison of the line intensities obtained from direct SSMS analysis of the zinc metal to those from the samples preconcentrated onto the gold electrodes indicates that electrodeposition results in a 1000 fold increase in the trace element concentrations in the zinc deposit. From the intensities of the lines obtained for the different elements reported it has been con- cluded that all six elements could have been determined at the 20 ppb level that was found for tin, and with comparable precision. The electrodeposition technique applied to the preconcentration of the impurities in the zinc represents one of the simplest possible separation methods. No chemical other than the easily purified HC1 required to dissolve the zinc sample was used. Little physical handling was involved in the process. These conditions plus the fact that all the impurity elements determined were separated from a single sample 196 Determination of trace elements in zinc simultaneously should keep the sample contamination (blank level) to a minimum. Modifications of this method should be applicable to preconcentration from most active metals. More complex preconcentration procedures have also been used successfully for isotope dilutions SSMS analysis. In a separate investigation [9] five elements were pre concentrated from a platinum matrix using a cation exchange column to separate these impurities from the platinum complexed as a chloride anion. All five elements were determined simultaneously with the concentration ranging from 0.5 to 14 ppm by weight. The six trace elements determined in zinc represent a range of more than 500 (Sn to Pb) in concentration for elements that could be simultaneously determined from a single sample. There is of course no upper limit on the concentrations that could be covered by isotope dilution, and it should therefore be possible to simul- taneously determine elements from a single sample over a concentration range from the ppb level to the per cent level. The ppb to ppm range, however, will encompass the levels where this technique should prove most useful. Comparison to other mass spectrometric methods SSMS analysis by direct comparison to a standard reference material must be carefully controlled in order for the precision of the measurements to approach the photographic plate limiting value of 3-5% relative standard deviation. The pre- cision is normally affected by the variations in the amount of the impurity elements reaching the plate from one exposure to the next. These variations can arise from inhomogeneities in the sample on a micro scale, changes in the point of sparking with respect to the "optical" axis of the instrument, distance between sample electrodes, rf and dc voltages of the instrument, etc. None of the factors however, have any detectable effect on the measured isotopic ratio for an element ; but only affect the absolute amount of the impurity measured. The precision of the analysis by stable isotope dilution SSMS is therefore independent of the forementioned instrumental factors and is controlled only by the limitations of the photographic plate. Isotope dilution analysis could therefore be expected to give better precision than that of direct analysis by SSMS. The comparison of spark source to thermal ion source mass spectrometry for analysis by stable isotope dilution is of interest. Thermal ion instruments can measure isotopic ratios of many elements with precisions of 0.01% relative standard deviation while the photographic plates used by SSMS limits its precision to about 3% at best. In an actual analysis the difference in precision will be less than indicated by the ability to measure isotope ratios since the precision of a determination will also be affected by other factors such as sample homogeneity relative to the amount of sample dissolved for spiking, the accuracy of spike additions, reproducibility of blanks, etc. The sensitivity of SSMS is relatively independent of the ionization potential of an element. Sensitivity varies markedly with ionization potential for a thermal ion source; its sensitivity is substantially better than the SSMS for low ionization potential elements, but is much poorer for the elements with high ioniz- ation potentials. [9] R. Alvarez, P. J. Paulsen and D. E. Kelleher, Anal. Chem., 41, 955 (1969). 197 Paul J. Paulsen, Robert Alvarez and Daniel E. Kelleher Thus, the spark source can extend the range of elements determined by stable isotope dilution to elements not attainable with the thermal ion source instrument, but would not be competitive for the other elements if precisions of better than 3-5% are needed. If the analysis requirements or blank problems make this precision acceptable, the ability to analyze for many elements simultaneously, over wide ranges in concentration, gives the spark source a definite advantage. Summary Isotope dilution spark source mass spectrographic analysis when combined with preconcentration techniques has yielded reliable determinations, down to the 20 ppb level. A comparison of isotope dilution determinations to those of polarography and neutron activation has shown good agreement for the trace elements. This technique gives absolute values without the use of a reference standard and therefore should be particularly applicable to the analysis of standard reference materials that are needed for direct trace analysis by SSMS. The recording of the entire mass range by SSMS on each exposure makes it possible to determine a large number of elements simultaneously from a single sample. This not only represents a saving of time for the analysis but also allows for simple methods of preconcentration which should minimize reagent blanks. The nearly constant detection sensitivity of SSMS for all elements would predict that most elements should be measureable at the 20 ppb level found for tin, providing, of course, that preconcentration can be accomplished and blanks do not set higher limits of detection. This should be particularly important for elements with high ionization potentials where, in general, isotope dilution analysis using thermal ion sources is not possible. The preconcentration procedure adopted here for the determination of trace elements in zinc should be applicable to most other active metal matrices. 198 APPENDIX IX Spark Source Mass Spectrographic Analysis of Ingot Iron for Silver, Copper, Molybdenum, and Nickel by Isotope Dilution and for Cobalt by an Internal Standard Technique Paul J. Paulsen, Robert Alvarez and Charles W. Mueller Spectrochemical Analysis Section, Analytical Chemistry Division, National Bureau of Standards, Washington, D. C. 20234 A number of trace elements are known to significantly affect the properties of ferrous materials. As compositions are being changed to meet the need of new applications, addi- tional trace elements are becoming important. The increased interest in these trace elements makes it desirable to provide more reliable methods for their determination by using new analytical instrumentation and techniques. Analysis by isotope dilution spark source mass spectrometry (ID-SSMS) enables many trace elements in a matrix to be determined simultaneously with high sensitivity and precision (/, 2). Moreover, the methods do not require a quantitative recovery or even a determination of the recovery. Because of the in- herent high sensitivity of the spark source mass spectrograph, simplified separation procedures that have low recoveries but are relatively contamination-free can be used. Contamina- tion is often the limiting factor in reliable trace determinations. The object of this investigation was to develop an ID-SSMS method for simultaneously determining trace amounts of silver and other elements in ingot iron, a matrix having a greater number of elements at higher concentrations than the high-purity platinum (/) and zinc (2) analyzed previously. In this method, silver, copper, molybdenum, nickel, and co- balt were satisfactorily electrodeposited onto gold wire elec- trodes after extracting the iron from a solution of ingot iron, NBS-SRM 1165. The aqueous layer had been neutralized before electrolysis. By spiking the sample solution with known amounts of enriched isotope solutions, the elements Ag, Cu, Mo, and Ni could be determined. Co, a mono- nuclidic element, was also determinable because of its chem- ical and electrochemical similarity to Ni. EXPERIMENTAL Reagents. Deionized water was prepared by passing distilled water from the laboratory distribution system through a mixed bed resin column. Hydrochloric and nitric acids, previously prepared by isothermal distillation and stored in quartz flasks, were used for sample dissolution. Ammonia gas was bubbled into a gas washing bottle and then into deionized water to produce aqueous ammonia. Reagent grade isopropyl ether, with a reported boiling range of 67.8 to 69.0 °C and a nonvolatile matter content of 0.003%, was shaken before use with an equal volume of 1M HC1 in a separatory funnel. Enriched isotope materials, in amounts equivalent to 10 mg of the element, were obtained from the Oak Ridge National Laboratory. Spike solutions contain- ing one mg per ml of 109 Ag, 65 Cu, e2 Ni, and "Mo were pre- pared by weighing each material to ±0.02 mg, dissolving it in nitric acid, adding aqueous NH 3 to the Mo solution, and diluting each solution to volume in a 10-ml volumetric flask. Standard solutions of natural Ag, Cu, Ni, Mo, and Co were (1) R. Alvarez, P. J. Paulsen, and D. E. Kelleher, Anal. Chem., 41,955(1969). (2) P. J. Paulsen, R. Alvarez, and D. E. Kelleher, Spectrochim. Acta, 24B, 535 (1969). similarly prepared by weighing 100 mg of each high-purity metal to ±0.05 mg, dissolving in ACS reagent grade nitric acid, adding aqueous NH 3 to the Mo solution, and diluting the solutions to volume in 100-ml volumetric flasks. Evaporator for Sample Solutions. The borosilicate glass chamber, which uses filtered nitrogen to prevent entry of particulate contaminants and provides a clean environment, has been described previously (/). It is similar to that de- scribed by Thiers (J). Electrolysis Cell. The polytetrafluoroethylene cell, which has two 0.7-mm diameter high-purity gold wire cathodes and an isolated anode compartment, has been described in detail previously (/). Mass Spectrograph. The double-focusing Mattauch- Herzog design, Spark Source Mass Spectrograph and the microphotometer with its analog computer have been de- scribed previously (/). Procedure. Approximately 100-mg samples of ingot iron, NBS-SRM 1165, were weighed to 0.1 mg and introduced into 30-ml borosilicate beakers. The 109 Ag, 65 Cu, and 62 Ni solutions were diluted and combined to make a multiple- spike solution having elemental concentrations ranging from 7 /xg/ml for Cu, to 0.3 ;ug/ml for Ag. The "Mo solution was diluted to 3 jig/mX. Immediately after preparing these solutions, 2 ml of 1M HC1 were added to each of the beakers. While the samples were dissolving, 1 ml of the multiple- spike solution and 100 (A of the diluted "Mo solution were pipetted into the beakers. In addition, for future verification of the elemental concentrations in the multiple and diluted "Mo spike solutions, aliquots of these solutions were com- bined with solutions containing known amounts of the natural elements. The Co solution was diluted to 50 jug/ml and 250 /il of this solution were pipetted into half of the sample beakers. The covered beakers were heated to dissolve the samples and several drops of HN0 3 were added. The solu- tions were then evaporated to dryness in the clean-environ- ment chamber and the resulting residues were dissolved by adding 9 ml of 1M HC1 to each beaker and warming. The solutions were cooled in an ice bath and introduced into 60-ml separatory funnels. Each beaker was rinsed twice with 0.5 ml of 1M HC1 and the washings were added to the appropriate separatory funnel. Ten milliliters of cold iso- propyl ether was added to each separatory funnel and the iron was extracted into the ether layer by shaking for one minute. The aqueous layer containing the isotopically altered analytes was withdrawn into the original beaker used for sample dissolution. The extraction step was repeated using the same ether which had been back-extracted with two 10-ml portions of water to remove the iron and shaken with 1M HC1. The aqueous layer was then transferred to the original beaker, and evaporated to 2-3 ml in the clean- environment chamber. Aqueous NH 3 was added until a precipitate formed from the residual iron remaining in solu- tion. The solution was then heated to remove excess NH 3 , transferred to the electrolysis cell, and electrolyzed for 3 hrs (3) R. E. Thiers in "Trace Analysis," John H. Yoe and H. J. Koch, Eds., John Wiley and Sons, Inc., New York, N. Y., 1957, Chap. 24. Reprinted from ANALYTICAL CHEMISTRY, Vol. 42, Page 673, May 1970 199 Table I. Average Concentrations in Ingot Iron NBS-SRM 1165 Determined by Spark Source Mass Spectrometry Compared to NBS Certified Values Concentration ( % by weight) Element Method This work" 0.00050 ± 0.00003 NBS certified values'" Ag Isotope dilution Cu Isotope 0.0172 ±0.0008 0.019 dilution Ni Isotope 0.0247 ±0.0009 0.026 dilution Mo Isotope 0.0042 ±0.0003 0.005 dilution Co Standard 0.0092 ±0.0013 0.008 addition " Uncertainty is expressed as 95% confidence limits on the limiting mean t .96S/\/n- For Ag, Cu, Ni, and Mo, the value includes uncertainties associated with the variations among the individual samples and the analysis of the spike solution. For Co, it includes the variations in the concentrations computed from Equation 2 and the variations in determining Ri. 6 These values are considered reliable to ± 1 in the last digit. c Not certified. at 4 mA while stirring at 200 rpm. After withdrawing the solution, the gold electrodes were rinsed by introducing ap- proximately 7 ml of water into the cell and stirring. The pair of electrodes was sparked in the mass spectro- graph and a graded series of exposures was made from 3 X 10 -13 to 1 X 10~ 8 coulomb. The ion-sensitive photo- graphic plates (Ilford Q-2) were processed using the bleach and internal image developer MK-7 described by Cavard (4). They were placed in the dichromate bleach for 4 minutes, rinsed in water for '/s min, developed for 7 min, fixed for 1 min, and washed for 15 min. The intensity areas of the lines on the photographic plate for the elements were mea- sured and the W9 Ag/ I07 Ag, 65 Cu/ 68 Cu, 62 Ni/ 60 Ni, and "Mo/ 96 Mo ratios were determined for all samples. The "Co/^Ni ratios were also determined for samples with added Co, and the 59 Co/ 60 Ni ratios determined for samples without the Co addition. From 6 to 14 ratio measurements were made for each element depending on the availability of lines having suitable intensity. To correct the determinations for contamination, the entire analytical procedure was performed under the same experimental conditions using the same quantities of reagents and 5 mg of high-purity iron. To obtain more favorable isotope ratios, the iron was spiked with one-tenth of the amounts added to the ingot iron samples and both materials were analyzed concurrently. The blanks were found not to exceed 2 % of the reported values. To verify the elemental concentrations in the multiple- spike and diluted "Mo solution, aliquots of the solutions that had been reserved for this purpose were pipetted into the polytetrafluoroethylene electrolysis cell. After electrolyzing the stirred solutions for 3 hours at a current of 4 mA, the gold wire electrodes were sparked in the mass spectrograph. Except for Mo, the altered isotope ratios were measured and the spike weights computed from Equation 1. The isotope lines of Mo were too weak to be measured and therefore an evaporation procedure was used. Approximately 0.5 ml of the combined diluted "Mo and natural Mo solution was evaporated to one tenth of the volume in a quartz crucible. The evaporated solution was pipetted onto gold wires and dried. After sparking the wires in the mass spectrograph, the "Mo spike weight was computed from the measured isotope ratios. (4) A. Cavard in "Advances in Mass Spectrometry," E. Kendrick, Ed., The Institute of Petroleum, London, 1968, pp 419-429. CALCULATIONS AND RESULTS The concentrations of Ag, Cu, Mo, and Ni were computed from Equation 1 C = WK(A, P - B sp R) M (BR - A) (1) were C is the concentration in jug/g, W is the weight of isoto- pically enriched material (spike) added in y.%, M is the weight of sample in grams, A and B are the natural abundances of the analyte isotopes a and b, A sp and B SP are the abundances of isotopes a and b in the spike, R is the measured altered ratio of isotope a to isotope b, and AT is the ratio of the natural atomic weight to the atomic weight of the spike. The spike isotopes provided by Oak Ridge National Labora- tory were all highly enriched; the abundances of the isotopes 109 Ag, 65 Cu, "Mo, and 62 Ni being 99.1, 99.7, 92.8, and 99.0 atomic per cent, respectively. The quoted accuracy of the abundances was such that errors from this source do not sig- nificantly affect our results. To minimize the effect of photographic emulsion calibration errors, it is desirable in determining R to measure isotope lines that have approximately the same intensity. This was done for all elements except silver. Silver has only two naturally occurring isotopes, I09 Ag and 107 Ag, which have almost the same abundance. Therefore, an amount of 109 Ag spike was added to change the value of the 109 Ag/ 107 Ag ratio to 2. Table I lists the average concentrations for Ag, Cu, Mo, and Ni obtained by isotope dilution, the 95% confidence limits, and a comparison of the results with the NBS certified values. The results were derived by running ten samples, in two groups of five. Separate spike solutions were prepared and added to each group. The individual results from each group did not vary significantly and therefore the data could be pooled for statistical evaluation. In addition to the elements being determined by isotope dilution, cobalt, a mononuclidic element, was also found to be electrodeposited on the gold electrodes. Because of the similar chemical and electrochemical behavior of cobalt and nickel, it was postulated that both elements would be precon- centrated on the electrodes with a constant relative recovery. To test this hypothesis, the ratios of the 69 Co to 60 Ni line in- tensities were measured on each exposure together with the altered isotope ratios used for the isotope dilution determina- tions of Ag, Cu, Mo, and Ni. The ratios were found to be constant among the samples. We concluded that under these experimental conditions, Ni would serve as an internal stan- dard to monitor the recovery of Co which could vary from sample to sample. Co would then be determinable by a standard addition of Co. The ratio of the 69 Co to 68 Ni line intensities were determined for each sample having a known and constant amount of Co added. By using the more abun- dant 68 Ni instead of the 60 Ni, a 69 Co/ 68 Ni ratio approximating unity could be obtained by adding the appropriate amount of Co. An average 69 Co/ 6 °Ni ratio, R u of 1.04 ± 0.05 was obtained for seven samples without added cobalt. The cobalt content, Co(s), of a sample was computed from the following equation: Co(a) /J,««Ni CoCs) = A 2 58 Ni - /?! 6 »Ni where Ri is the measured 69 Co/ 58 Ni ratio for a sample with added cobalt, Co (a) is the added cobalt, and 60 Ni, 68 Ni are the nickel contents including the small amounts added with the 62 Ni spike. All concentrations are in /ig-atom/gram of ingot iron. The average of six determinations of Co, the 95 % 200 confidence limits, and the NBS certified value are given in Table I. Recovery Experiments. One of the advantages of the isotopic dilution technique of analysis is that only enough spiked sample required for an accurate isotope ratio deter- mination need be recovered from the original sample. A knowledge of the recovery efficiencies is not required for computation of results of analysis. The efficiency of re- covery of a preconcentration technique however, is of interest because the amount of recovery will effect the detection limits for the method. Experiments were performed in order to determine the recovery for the procedures up to and includ- ing the electrodeposition step. A set of samples was run using the previously described experimental procedure but omitting the spike additions. For comparison, electrodeposition periods of one hour were also used in addition to the 3-hour periods used in the procedure. The material recovered on the high-purity gold cathodes was redissolved in 0.2 ml of warm HNO3, spiked with the stable isotopes, and neutralized with aqueous ammonia. The re- covered trace elements were then electrodeposited onto the same gold electrodes. The electrodes were sparked in the SSMS and the amounts of material recovered on the elec- trodes were computed from Equation 1 . This procedure gave results for Ag, Cu, and Ni; however, Mo could not be de- tected in this second plating. Increasing the plating time from one to three hours increased the total recovery for Ag, Cu, and Ni from 20 to 40%. The high sensitivity of the spark source mass spectrograph method is illustrated by realizing that only 0.2 jug of silver on the gold electrodes was more than adequate for the determination of silver. Received for review December 2, 1969. 24, 1970. Accepted February 201 APPENDIX X REPRINTED FROM APPLIED SPECTROSCOPY, Vol. 30, No. 1, 1976, pp. 42-46. Trace Element Determinations in a Low-Alloy Steel Standard Reference Material by Isotope Dilution, Spark Source Mass Spectrometry* PAUL J. PAULSEN, ROBERT ALVAREZ, and CHARLES W. MUELLER Institute for Materials Research, National Bureau of Standards, Washington, D.C. A stable isotope dilution procedure using the spark source mass spectrograph was developed for the simultaneous de- Received 16 May 1975; revision received 10 September 1975. * Certain commercial materials and equipment are identified in this paper in order to adequately specify the experimental pro- cedure. In no case does such identification imply recommenda- tion or endorsement by the National Bureau of Standards, nor does it imply that the material or equipment identified is neces- sarily the best available for the purpose. Volume 30, Number 1, 1976 202 termination of cerium, copper, neodymium, selenium, sil- ver, tellurium, and zirconium in a low-alloy steel material. Except for copper, the elements were present at trace con- centrations that are difficult to determine by most analyti- cal methods. In the procedure, known amounts of the iso- topically enriched elements were added to 100-mg samples, which were dissolved in perchloric and hydrofluoric acids. After evaporation of the solutions almost to dryness, dis- solution of the residues, and addition of hydrofluoric acid APPLIED SPECTROSCOPY to complex the iron, the resulting solutions were electro- lyzed in polytetrafluoroethylene cells. The electrodeposit on high purity, gold cathode wires was sparked in the mass spectrograph and the isotope ratios of the isotopically equilibrated elements were measured. The concentrations, ranging from 0.041% for copper to 3.1 v-%1% (ppm) for neo- dymium, were calculated from the general isotope dilution equation. The low -alloy steel is available from NBS as Standard Reference Material 1261. Index Headings: Analysis, for trace elements; Mass spec- * — — — »nv', spark source; Methods, analytical. INTRODUCTION A set of low-alloy steel Standard Reference Materials (SRMs) was issued by the National Bureau of Standards in 1957 primarily for use in optical emission and x-ray spectrometric methods of analysis. At that time, several trace elements present in this set were not certified either because they were at extremely low concentrations or be- cause they were generally regarded as having an innocuous effect on the physical properties of steel. Consequently, the anticipated benefits of providing certified values for these elements would not have justified the cost of de- veloping analytical methods having the requisite relia- bility. 3ince this set of SRMs was issued, new applications of ferrous products, mainly for the nuclear and aerospace industries, have necessitated more stringent specification limits and quality control determinations of additional trace elements. Coinciding with an increased interest in the effects of trace elements on the properties of ferrous materials has been the increased use of spark source mass spectrometry as a powerful tool for estimating the con- centration of nearly every element in a sample down to a few nanograms per gram (parts per billion). However for reliable quantitative determinations, SRMs were needed having certified values for the trace elements of interest and having concentrations of the major constituents, simi- lar to those of the samples being analyzed. A new set of iron and low-alloy steel SRMs was made in the late 1960's to replace the earlier set. The SRMs were fabricated in the form of rods, 3.2 mm in diameter by 51 mm long, for spark source mass spectrometry and micro- chemical techniques as well as in the form of 32-mm diam- eter discs for x-ray and optical emission spectrometry. This replacement set was intended to provide concentra- tion ranges for approximately 40 elements. Certification of the trace elements in these materials required sensitive and reliable methods. Moreover, for the efficient characterization of these materials with respect to trace elements, analytical methods that would yield results for a number of elements simultaneously were de- sirable. Isotope dilution procedures using the spark source mass spectrograph (ID-SSMS) has been found to meet these requirements. 13 The object of this investigation was to develop an ID- SSMS method for determining a number of analytically difficult trace elements such as selenium, tellurium, and neodymium in a low-alloy steel material SRM. In a previ- ous application of an ID-SSMS method to a ferrous ma- terial, copper, molybdenum, nickel, and silver had been determined in an ingot iron, SRM 1165. 3 Although the low-alloy steel material was a more complex matrix than the ingot iron, a relatively simple ID-SSMS procedure was developed which enabled a number of chemically diverse elements to be determined simultaneously. I. EXPERIMENTAL A. Reagents and Special Solutions. Deionized water was prepared by passing distilled water from the labora- tory distribution system through a mixed-bed, resin col- imn. Commercially produced, high purity, perchloric and lydrofluoric acids were used for sample dissolution; similar dgh purity sulfuric, nitric, and hydrochloric acids were ised for preparing solutions of the enriched isotope ma- terials (spikes) and natural elements. The purity of the perchloric, nitric, and hydrochloric acids had been evalu- ated previously by an ID-SSMS procedure developed by the authors. 4 Materials, isotopically enriched with respect to U5 Nd, 142 Ce, I25 Te, 109 Ag, 91 Zr, 82 Se, and M Cu, were ob- tained from the Oak Ridge National Laboratory. The isotopic abundance, which was the highest available at the time of purchase, was at least 89 atomic percent. Each spike solution was prepared to contain 1 mg/ml of the element by weighing the spike material to ±0.02 mg, dis- solving it in the appropriate acid, and diluting the solution to volume in a 10-ml volumetric flask. A standard solution of each natural element under investigation was prepared by weighing 100 mg of the high purity metal to ±0.05 mg, dissolving it in the appropriate acid, and diluting the solution to volume in a 100-ml volumetric flask. For cerium and neodymium, the equivalent weight of oxide was used. B. Evaporator for the Sample Solution. The evap- orator is similar in function and design to the borosilicate glass unit previously described 15 except that the principal components are constructed of polytetrafluoroethylene TFE) and translucent fluorinated copolymer (TFE- EP). The 14.5-cm diameter TFE base is 13 mm thick nd has four cylindrical cavities into which 30-ml TFE- FEP beakers fit snugly. The 10-mm deep cavities are 90° apart with centers on a 3.5-cm diameter radius. A 12.5-cm diameter by 9.0-cm high cover of TFE-FEP fits tightly into a cylindrical groove in the base. Near the top of the cover are two 8-mm diameter TFE tubes, which are press- fit into holes approximately 180° apart. Filtered air enters the apparatus through one of the tubes to remove the acid vapors and prevent entry of particulate con- taminants. C. Electrolysis Cell. The TFE electrolysis cell consists of two parts: a cylindrical body and a base. The cell body has a 33-mm height, a 17-mm inside diameter, and a 3-mm wall thickness. The base makes a liquid-tight fit into the cell body and has four holes, 90° apart, on a 12-mm diam- eter circle. The holes are drilled undersize for inserting 0.9-mm diameter high purity gold wire electrodes. The gold wire had been examined by direct sparking in the mass spectrograph for the presence of impurities that could affect the measurements. None were found. The overall purity was determined to be comparable to that of the high purity gold, SRM 685. 6 The ends of the four wires are dipped in aqua regia, rinsed in water, and ad- 203 APPLIED SPECTROSCOPY justed to project approximately 2 mm into the cavity of the assembled cell. Two opposite wires serve as the cathode ; the other two as the anode. During electrodeposi- tion, the solutions in the cells are stirred with TFE stirrers at approximately 200 rpm. The cells are housed in a plastic enclosure under a slight positive pressure of nitrogen to minimize airborne contamination. D. Mass Spectrograph. A double-focusing, Mattauch- Herzog design, spark source instrument (Consolidated Electrodynamics model 21-110) that had been modified to give 22 exposures per plate is operated at an ion accelera- tor voltage of 24 kV. The ion-source sample chamber is evacuated with a high-speed, high-conductance ion pump having added titanium sublimation. A pressure of 7 X 10 -5 Pa (5 X 10~ 7 Torr) can be obtained in approximately 20 min after loading a sample into the chamber. Less than 60 min is required to attain a pressure in the range of 10 -6 Pa (10 -8 Torr). The pressure is measured using an ion gauge mounted directly in the ion source chamber. The chamber containing a sample from the isotope dilution procedure is evacuated to a pressure of less than 1 X 10~ 5 Pa (1 X 10 -7 Torr) before sparking. The high conductance and speed of the pumping system maintains pressures in the 10 -5 Pa (10 -7 Torr) range during sparking even when relatively large amounts of gas are being released. E. Microphotometer-Analog Computer. A micro- photometer having a precision screw (Jarrell Ash model 23-500) and equipped with an analog computer to measure peak areas 7 is used. A function generator, which is the basic component of the analog computer, provides an output proportional to relative intensities when the input is a voltage proportional to percent absorption. An in- tegrator follows the function generator to give signals pro- portional to peak intensity areas as the line is scanned. F. Procedure. A preliminary analysis of the steel sam- ple was performed during the initial development of the isotope dilution procedure. The elemental concentrations determined were used to calculate the amount of spike isotope necessary to obtain an optimum altered isotopic ratio, which for the elements of interest except silver is 1 . An altered isotopic ratio of approximately unity is desira- ble because it will minimize the effects of any errors in the calibration curve for the photographic plate and will usually yield the maximum number of measurable isotopic ratios. Silver, which has a natural isotopic ratio of ap- proximately one, was spiked to yield an altered isotopic ratio of 2. In the final procedure, a 100-mg sample was weighed to ±0.1 mg and transferred to a 30-ml TFE-FEP beaker. The beaker was covered with a TFE lid. Aliquots of spike solutions having compatible ions were combined and di- luted in volumetric flasks. Immediately after preparing these solutions, 2 ml of 6 M HC10 4 and 0.5 ml of 26 M HF were added to the sample. While the sample was dis- solving, known volumes of the spike solutions were added using calibrated plastic pipets. At the same time, aliquots of the spike solutions were mixed with solutions containing known amounts of the natural elements for future verifi- cation of the spike concentrations. The beaker was then heated to completely dissolve the sample, transferred to the evaporator, and heated further until HC10 4 fumes were observed. After adding 0.5 ml of 26 M HF, the contents of the beaker were heated al- most to dryness. The residue was dissolved and the iron was complexed by adding 2 ml of H 2 and 0.5 ml of 26 M HF, and heating the beaker. The solution was then elec- trolyzed for 4 h at a constant current of 6 mA while being stirred at 200 rpm. The solution was withdrawn and the electrodes were rinsed in water. The electrodeposit on the ends of the gold wire elec- trodes was sparked in the mass spectrograph and a graded series of exposures was made from 3 X 10 -13 to 1 X 10 -8 C. The ion-sensitive photographic plates (Ilford Q-2) were processed using the bleach and internal image de- veloper MK-7 described by Cavard. 8 They were placed in the dichromate bleach for 4 min, rinsed in water for J^ min, developed for 7 min, fixed for 1 min, and washed for 15 min. For all elements except zirconium, intensity areas of both measured isotopes of an element were determined on each suitable exposure using the microphotometer-analog computer. Maximum peak intensities, corrected for back- ground, were measured for zirconium. The isotopic masses measured for each element are: Ag 107, 109; Ce 140, 142; Cu 63, 65; Nd 144, 145; Se 80, 82; Te 126, 125; and Zr 90, 91. The second mass listed for each element is the spike isotope ; the first mass is usually the isotope of high- est abundance in the naturally occurring element. In general, the ratios of the singly charged ions were measured, but ratios of the multiply charged ions were also measured when appropriate to obtain additional data or to avoid interferences existing at the singly charged lines. The individual elemental concentrations were calculated from the measured altered isotopic ratios (R) using Eq. (1). C = WKU.p - B ip R) M(BR - A) (1) where C is the concentration in micrograms per gram, W is the weight of isotopically enriched material (spike) added in micrograms, M is the weight of sample in grams, A and B are the natural abundances of the analyte iso- topes a and b, A ap and B ap are the abundances of isotopes a and b in the spike, R is the measured altered ratio of isotope a to isotope b, and K is the ratio of the natural atomic weight to the atomic weight of the spike. The spike concentrations in the solutions that had been added to the samples were verified by preconcentrating the isotopically equilibrated elements from the solution that had been reserved for this purpose, determining the isotopic ratios, and calculating W in Eq. (1). Two precon- centration procedures were used: electrolysis using the experimental procedure described for the sample, and evaporation of separate aliquots of the solution onto high purity gold wires after which the wires were ignited at 450°C for 30 min. The evaporation procedure enabled us to obtain data for elements that had not been electrode- posited satisfactorily. The method blank was determined by performing all of the steps in the procedure using the same quantities of acids but smaller quantities of the spikes. Volume 30, Number 1, 1976 204 [I. DISCUSSION AND RESULTS The graded series of exposures used in the analysis was selected to obtain the optimum sensitivity for elements at concentration levels down to 1 /xg/g (ppm). Among other considerations, the selection of isotope lines and charge states was based on the required sensitivity and potential interferences. Some of the specific considerations for the determined elements are listed below. 1. Copper. Because of the high copper concentration, 0.041 %, in the sample, the exposures produced Cu + lines, most of which were too dark to be measured reliably. Therefore, the Cu 2+ and Cu 4+ isotopic ratios were also measured to obtain an adequate number of ratios. For these charge states, both M Cu and M Cu isotopes occurred at fractional mass/charge positions, and therefore were free of interferences from +1 ions occupying unit mass positions in these m/e regions. 2. Selenium. Copper-containing samples normally yield lines at mass positions 79 and 81 corresponding to [CuO] + , and less intense lines at 80 and 82 corresponding to [CuOH] + . If [CuOH]+ lines exist, they can interfere with the determination of the isotopic ratio, 82 Se/ 80 Se, when the singly charged ions are used. This potential interference was indicated by faint lines of [CuO] + being present at mass positions 79 and 81. Because molecular species, such as [CuOH] + will not form stable multiply charged ions, the +3 charge states of the selenium isotopes were meas- ured to obtain the isotopic ratio. Consequently, these measurements, at fractional m/e positions, were also free of interferences by any molecular species. Concentrations calculated from the Se + and Se 3+ isotopic ratios were in agreement, indicating that [CuOH] + did not in fact sig- nificantly interfere with the Se + measurements. 3. Zirconium. Each of the zirconium isotope lines had a faint line on the high mass side, which was attributed to hydrocarbons and could not be resolved when measuring the customary intensity areas*. Because the contribution of the hydrocarbon line to the center of each Zr isotope line was essentially zero, the maximum peak intensity, corrected for background, was measured for each isotope line. Measurements of maximum peak intensity are not as sensitive as those of intensity area. However, the elec- trodeposited Zr yielded a signal of sufficient intensity for satisfactory precision with either mode of measurement. 4. Silver. Both the +1 and +2 ions were measured for TABLE I. Selenium concentration in sample 1, NBS-SRM 1261, low-alloy steel calculated from replicate exposures. TABLE II. Average selenium concentrations determined in test samples of NBS-SRM 1261, low -alloy steel. Exposure No. 8S Se intensity 80 Se intensity Concentration in % area area by wt 6 713 825 0.003598 7 603 710 0.003682 8 387 470 0.003838 9 336 377 0.003455 10 378 479 0.004064 11 367 440 0.003772 12 333 402 0.003822 Average U. 00375 Relative standard deviation 5.2% 95% confidence limits of the mean 0.00018 Sample Concentration in % by wt Average 95% confidence limits of a single determination 0.00375 0.00350 0.00360 0.00373 0.00370 0.00352 0.00366 0.0036 0.0002 TABLE III. Results obtained by ID-SSMS for seven ele- ments determined simultaneously in NBS-SRM 1261, low- alloy steel. Element Average concentration in % by wt 95 % confidence limits of a single determination No. of samples Copper 0.041 ±0.0020 7 Zirconium 0.0071 ±0.0011 o Selenium 0.0036 ±0.0002 7 Cerium 0.0015 ±0.0001 7 Tellurium 0.0007 ±0.0002 5 Silver 0.00036 ±0.00004 5 Neodymium 0.00031 ±0.00002 6 the 107 Ag and 109 Ag isotopes to obtain an adequate number of isotopic ratios. 5. Tellurium. The singly charged ionr were used to determine the isotopic ratios from which the concentration was calculated. 6. Cerium. The 142 Ce isotope line used in determining the isotopic ratio also has a contribution from 142 Nd present in the sample. This contribution was calculated for each exposure from a measured I44 Nd intensity and the natural isotopic ratio, 142 Nd/ 144 Nd. This correction of the intensity area at mass position 142 was always less than 10% of the total area because of the low concentration of Nd compared to Ce in the sample and the relatively low abundance of the 142 Nd isotope. 7. Neodymium. Singly charged ions were used to deter- mine the altered isotopic ratio. A computer program was developed, incorporating Eq. (1), that provides a print-out of concentration values for each pair of intensity area measurements without printing the values of R used in the calculations. The program also gives an average concentration and a statistical analysis of the data. As an example of the data, Table I shows the 82 Se and 80 Se intensity areas that were measured for a series of ex- posures, and the selenium concentrations calculated for each pair of intensity areas. The average concentration of selenium in samples 1, the relative standard deviation, and the 95% confidence limits of the mean are also shown. Table II shows the average selenium concentration in each test sample, and the 95% confidence limits of a single determination. Table III shows the results obtained by ID-SSMS, for 205 APPLIED SPECTROSCOPY TABLE IV. Comparision of results by ID-SSMS with other methods of analysis for NBS-SRM 1261, low -alloy steel. Average cor i- Element centration Analytical method in % by wt Copper 0.041 ID-SSMS 0.043 Neutron activation 0.042 Isotope dilution, thermal source mass spectrometry 0.041 Spectrophotometry Zirconium 0.0071 ID-SSMS 0.009 Gravimetry 0.008 Spectrophotometry Selenium 0.0036 ID-SSMS 0.003 Neutron activation Cerium 0.0015 ID-SSMS 0.001 Spectrophotometry 0.0014 Isotope dilution, thermal source mass spectrometry Tellurium 0.0007 ID-SSMS 0.0005 Polarography Silver 0.00036 ID-SSMS 0.00040 Atomic absorption Neodymium 0.00031 ID-SSMS 0.00028 Isotope dilution, thermal source mass spectrometry seven elements in NBS-SRM 1261, low-alloy steel. The table lists the average concentrations of the elements in percentage by weight, the 95 % confidence limits of a single determination, and the number of samples for which re- sults were obtained. The experimental conditions used for the first two samples did not preconcentrate all of the elements satisfactorily, but the final conditions described under "Procedure" yielded all of the analytes for simul- taneous determination of the seven elements. Table IV shows a comparison of the results obtained by ID-SSMS with other methods of analysis. Considering the difficulties in determining most of these trace elements, the agreement of results is considered satisfactory. Most other methods are specific for a single element and often involve complicated chemical separations and manipulations. III. CONCLUSIONS The major sources of error in trace element determina- tions are usually losses of the elements and adventitious contamination, the blank. The use of isotope dilution in the procedure that was developed minimizes errors due to trace element losses. The trace elements are brought to equilibrium with the spikes at an early stage of the pro- cedure, after which the concentrations obtained are inde- pendent of losses. Although the blank is not considered a problem in the determination of these elements, small quantities of reagents and few handling operations ensure that the blank does not contribute significantly to the method error. 1. R. Alvarez, P. J. Paulsen, and D. E. Kelleher, Anal. Chem. 41, 955 (1969). 2. P. J. Paulsen, R. Alvarez, and D. E. Kelleher, Spectrochim. Acta 24B, 535 (1969). 3. P. J. Paulsen, R. Alvarez, and C. W. Mueller, Anal. Chem. 42, 673 (1970). 4. R. Alvarez, P. J. Paulsen, and C. W. Mueller, NBS Technical Note 542,70, (1970). 5. R. E. Thiers, in Trace Analysis, John H. Yoe and H. J. Koch, Eds., (Wiley, New York, 1957), Chap. 24, p. 637. 6. Catalog of Standard Reference Materials, NBS Special Publ. 260, April 1973, Office of Standard Reference Materials, National Bureau of Standards, Washington, D.C. (available from U.S. Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402). 7. P. J. Paulsen and P. E. Branch, NBS Technical Note 401. (1966). 8. A. Cavard, in Advance in Mass Spectrometry, E. Kendrick, Ed., (The Institute of Petroleum, London, 1968), pp. 419-429. Volume 30, Number 1, 1976 206 APPLIED SPECTROSCOPY APPENDIX XI Reprint from Special Technical Publication 443 American Society for Testing and Materials 1916 Race Street, Philadelphia, Pa. 19103 1969 T. C. Rains 1 Chemical Aspects of Atomic Absorption REFERENCE: Rains, T. C, "Chemical Aspects of Atomic Absorption," Atomic Absorption Spectroscopy, ASTM STP 443, American Society for Testing and Materials, 1969, pp. 19-36. ABSTRACT: The phenomenal growth of atomic absorption spectrometry in the past twelve years can be attributed to many factors such as the high sensitivity of the method for a large number of elements, advances in overcoming interferences, saving of time required for analysis, and the rapid improvement of instrumentation. The major interferences in atomic absorption are of a chemical nature; however, physical interferences which are associated with the chemical matrix often are present. To overcome chemical interferences and to eliminate or minimize their influences from phosphate, aluminum, titanium, and any other elements producing nonvolatile compounds, releasing agents, or high-temperature flames are used. Also, the use of a separation and preconcentration technique often becomes necessary when determining elements at the subtrace level. A list of organic ligands and solvents used in solvent extraction for absorption is presented as a means of preconcentration and removal of chemical interferences. KEY WORDS: atomic absorption, detection limits, electrodeless discharge, nebulizer burner, interferences, releasing agents, organic ligands, solvents, analyzing, tests Atomic absorption spectrometry (AAS) has made phenomenal growth since its introduction by Walsh in 1955 [l]. 2 This growth can be attributed to many factors such as the high sensitivity of the method for a large number of elements, advances in overcoming interferences, speed of analysis, and the rapid improvement in instrumentation. Today the analytical chemist has a variety of atomic absorption instruments and components from which to choose. These include different light sources, nebulizer burners, oxidant-fuel systems, optical arrangements, and associated electronics. All of these parameters must be considered in any discussion of the chemical aspects of AAS, and for that reason a brief discussion of these components also is presented. The sensitivity and precision obtained by AAS are of concern to the analyst; however, the requirements for sample preparation prior to making an Research chemist, Analytical Chemistry Div., National Bureau of Standards, Washington, D. C. 20234. 2 The italic numbers in brackets refer to the list of references appended to this paper. 207 ATOMIC ABSORPTION SPECTROSCOPY absorption measurement can be of equal importance. Where the sample is a solution, the analysis, often times, can be performed on the sample as received. The major interferences in absorption are of a chemical nature. However, physical interferences often are present and are associated with the chemical matrix. Scattered light which appears as molecular absorption may be due to high solids or large droplets produced by the nebulizer. This type of interference is very pronounced in the far ultraviolet region of the spectrum as exemplified by arsenic and selenium. Also, molecular fragments such as CH, C2, and CN do exist and can contribute to the background as well as to scattered light. Molecular absorption also is observed from the OH bands over the wavelength region of 2800 to 3500 A. Sensitivity and Detection Limit The terms "sensitivity" and "detection limits" have led to some confusion in the field of atomic absorption. Sensitivity is defined as the concentration of an analyte in solution that produces an absorption of 1 percent. Detection limits is not defined so readily. A number of papers have appeared which discuss this issue [2-4 ] . Winefordner and co-workers [5-6 ] have derived theoretical equations for the calculation of the detection limit based on instrumental and spectral constants. In general the term is defined as the concentration, micrograms per milliliter, that produces an absorption equal to twice the magnitude of the fluctuations of the blank or background noise. The major problem is the selection of a blank. Usually, the blank is distilled water which really cannot be considered a true blank in reporting a detection limit in a sample. Table 1 presents the detection limits obtained with a variety of instruments and under various conditions [7-9] . Instrumental Components The instrumental components are best illustrated by a schematic diagram, as shown in Fig. 1. The basic components are the light source, nebulizer burner, monochromator, photodetector, and readout system. Light Source To attain maximum absorption the light source should emit radiation of the desired wavelength that is sharp, intense, and stable. There are many types of line sources used in absorption work; however, the most commonly used source is a hollow-cathode lamp. Another source which fulfills the first two requirements is the electrodeless discharge lamp. The lamp developed by Meggers and Westfall [10] is an excellent source for use in absorption. In conjunction with the lamp various types of power sources and cavities are used [11-13]. The purpose of the cavity is to transfer the power from the microwave generator, such as a medical diathermy unit, to the inert gas within the lamp. The charged inert gas ions then bombard the metal present producing an excited metallic atom which emits its characteristic radiation. 208 RAINS ON CHEMICAL ASPECTS TABLE 1 -Detection limits reported by atomic absorption spectrometry. Element Wave- length, Detection Limit, Mg/ml Element Wave- length, Detection Limit, Mg/ml Ag 3281 Al 3093 As 1937 Au 2676 B 2497 Ba 5536 Be 2349 Bi 2231 Ca 4227 Cb 4059 Cd 2288 Co 2364 Cr 3579 Cs 8521 Cu 3248 Dy 4211 Er 4008 Eu 4594 Fe 2483 Ga 2874 Gd 3684 Ge 2651 Hf 3072 Hg 2537 Ho 4163 In 3039 Ir 2850 K 7665 La 3928 Li 6708 Lu 3312 Mg 2852 Mn 2795 0.005 fl 0.F 0.2c 0.01* 6 01* 0.003* 0.02 0.003* 20* 0.001<* 0.007* 0.005 0.05* 0.005 0.2* 0.2* 0.2* 0.01 0.07 4* 2* 15* 0.2£ 0.3* 0.05 4* 0.005* 8* 0.005 50* 0.0005 0.005 Mo 3133 0.1 Na 5890 0.005 Nd 4634 2* Ni 2320 0.01C Pb 2170 0.0K 2833 0.02 Pd 2476 0.5 Pr 4951 10* Pt 2659 0.5 Rb 7800 0.005-' Re 3460 1.5* Rh 3435 0.03 Ru 3499 0.3 Sb 2176 0.2 Sc 3912 0.2* Se 1961 0.5 Si 2516 0.1* Sm 4297 5* Sn 2246 0.1^ Sr 4607 0.01 Ta 2715 6* Tb 4326 2* Te 2143 0.3 Ti 3643 0.1* Tl 2768 0.2 Tm 4094 0.1* U 3585 30* V 3184 0.02* W 4008 3* Y 4077 0.3* Yb 3988 0.04 Zn 2139 0.002 Zr 3601 5* a Laminar flow burner with preheater. * Nitrous oxide-acetylene. c Argon (entrained air)-hydrogen. ^Quartz tube. e High-brightness lamp. /Osram lamp. £ Mercury germicidal lamp. One of the more effective cavities which meets the requirement for absorption is the Evenson cavity [14] shown in Fig. 2. To match the resonance frequency and impedance of the source, the cavity is provided with a coupling slide and tuning screw. To facilitate tuning a RF power meter is placed between the source and cavity. When the resonance frequency of the cavity is tuned to the frequency of the diathermy unit and the impedance is then matched, the power reflected from the cavity is at a minimum. 209 ATOMIC ABSORPTION SPECTROSCOPY CZD--A-" LIGHT CHOPPER SOURCE NEBULIZER- FLAME POWER SUPPLY I -H^- j PHOTODETECTOR 1 M0N0CHR0MAT0R LOCK-IN AMPLIFIER DIGITAL READ-OUT RECORDER FIG. 1 -Schematic diagram of atomic absorption instrument. FIG. 2-Evenson cavity and lamp. The major objection to the electrodeless discharge lamp as a source for absorption measurements is the instability or drift of the source. The effect of intensity and stability of lead lamps have been studied as a function of argon pressure as shown in Table 2. Lamps prepared for arsenic, bismuth, copper, lithium, sodium, and zinc have exhibited similar characteristics. TABLE 2-Effect of argon pressure in lead electrodeless discharge lamps. Argon Pressure, torr <0.1 0.2 0.4 1.0 2.0 3.0 Intensity, arbitrary units 250 1100 200 1 1 Drift, %/min < 1 1 < 1 5 5 210 RAINS ON CHEMICAL ASPECTS A comparison of the intensity of electrodeless discharge lamps and hollow-cathode lamps is given in Table 3. In nearly every case, the electrodeless lamp was more intense than the hollow-cathode lamp, but in no case was the absorption greater (see Fig. 3). The lower sensitivity for the electrodeless lamp was attributed to self reversal. Measurement of the line width of the sodium lamp with an interferometer showed that self reversal occurred, as shown in Table 4, for the sodium resonance line at 5890 A when the power of the RF generator exceeded 15 W. To eliminate the line reversal of sodium, the RF generator is operated at a power of 10 W, and the air to the lamp is preheated 180 to 200 C. This approach is found to increase the intensity of the lamps without any loss of absorption. TABLE 1— Comparison of intensity of electrodeless discharge and hollow-cathode lamps. Electrodeless Element Wavelength, Hollow Cathode Current, mA I Discharge Power, W I Ag .... 3280 12 1100 30 1000 Bi .... 3068 10 400 30 200 Cu . . . . . 3247 12 2300 30 8400 Fe 2483 30 200 30 200 Hg 2537 30 5400 Na 5890 15 640 20 900 Pb 2833 12 96 30 1100 Zn 2138 12 14 30 530 4 8 12 16 20 CONCENTRATION, ^g /ml FIG. 3-Comparison of sensitivities of hollow-cathode and electrodeless discharge lamps. Nebulizer-Burner System The nebulizer-burner system is probably the most important component in any absorption measurement. It is the function of this system to convert the analyte in the sample solution to atomic vapor. This process includes the 211 ATOMIC ABSORPTION SPECTROSCOPY TABLE A-Absorbance for sodium with electrodeless discharge, lamp. Burner, triflame-air/C2H2 Wavelength, X: 5890 Na, /ig/ml: 1.0 RF Generator, W Line Width, A a Absorbance 5 ~ ~ 0.170 6 0.035 0.165 10 0.04 0.160 15 0.06 0.130 20 0.08 0.100 40 0.020 ^Interferometer measurement, resolution: 0.012 A. following steps: (1) dispersion of solution in the form of small droplets, (2) evaporation to dryness, (3) decomposition of salt or oxide, and (4) formation of atomic vapor. The extent to which these occur will depend upon the matrix, type of nebulizer burner, oxidant, and fuel gases. The most commonly used system for the production of atomic vapor is the premixed gas or laminar flow burner. The solution is sprayed into a spray chamber where the coarse and fine droplets are separated. The fine droplets are mixed with the flame gases; they then pass into the flame where they are vaporized. The major advantages of the laminar flow burner over the diffusion or total consumption burner in absorption are increased sensitivity and a minimum of interference. Since only atoms in the ground state will absorb energy at the resonance wavelength, the flame should be of sufficient temperature to dissociate the molecules, but not so high as to produce ions. Many elements in relatively cool flames such as argon (entrained air)- hydrogen or air -propane are fully dissociated, others, such as the alkaline earth metals, require air-acetylene flame. Aluminum, silicon, and titanium require a fuel-rich nitrous oxide-acetylene flame. Also, the length of time that the atoms remain in the observed area of the flame will depend upon the gas velocity. The flame front propagation rate for an oxygen-acetylene flame in a diffusion burner is very high (approximately 1200 cm/s); therefore, the residence time of the atoms is very short. However, the flame velocity for the air-acetylene or nitrous oxide-acetylene flame is 160 or 180 cm/s, respective- ly. The decrease in flame velocity will permit a longer residence time for the atoms in the observed area and give an increase in sensitivity with a minimum number of interferences. The temperature and burning velocities for various gas mixtures are given in Table 5 [15-1 7] . Another method for producing atomic vapor is by use of a high- temperature furnace. Woodriff and Ramelow [18] report the use of a high-temperature graphite-tube resistance furnace capable of attaining a temperature of 3000 C. The solution is aspirated with argon into the furnace 212 RAINS ON CHEMICAL ASPECTS TABLE 5 -Temperature and burning velocities of oxidant- fuel gases used in absorption. Temperature Burning Velocity, Oxidant Fuel Measured, deg C cm/s Air-natural gas 1700 to 1900 55 AU--C2H2 2125 to 2400 160 Air-H2 2000 to 2050 320 to 440 02-natural gas 2740 O2-C2H2 3060 to 3135 1100 O2-H2 2550 to 2700 2000 N2O-C2H2 2867 180 N2O-H2 2607 380 where the free atoms are produced. This system was reported to be useful for eliminating those interferences due to refractory elements. The coupling of an ultrasonic nebulizer with a high-temperature furnace undoubtedly would provide optimum conditions for the determination of trace elements in the presence of refractory oxides. Optical Arrangement and Associated Electronics Of equal importance in atomic absorption are the optical arrangement, detection system, and the overall electronic components. These components are essential and must be evaluated in any discussion of atomic absorption. The monochromator should have sufficient resolution to separate the spectral line under investigation from other spectral lines. Any emitted line not resolved by the monochromator, whether from the filler gas or from any impurity in the metal in the hollow-cathode lamp, can produce a severe loss of sensitivity. Therefore, a spectrum of the desired region should be prepared for each source of radiation. There are some special instances where a narrow slit width is required to avoid this type of interference. To detect the radiation a multiplier phototube with its associated power supply and measuring system is essential for high sensitivity and precision. Because the detection limit is affected directly by the stability of the measuring circuit, the multiplier phototube, power supply, amplifier, and readout system must be sensitive and stable. The most commonly used phototubes for absorption as well as emission are given in Table 6. Interferences Chemical interferences occur in varying degrees for all elements in atomic absorption spectrometry. A chemical interference is defined as any reaction which affects the analyte or the element to be determined. This may occur in the liquid, solid, or gaseous phase of the flame. Any cation or anion present in the sample which will form compounds with the analyte and are not dissociated easily will reduce the absorption. An example of this type of interference which usually is cited is the effect of aluminum, phosphate, 213 ATOMIC ABSORPTION SPECTROSCOPY TABLE 6 -Multiplier phototubes commonly used in absorption measurements. Multiplier Spectral Phototube Response Range, A R-106 S-19 1600 to 6500 RCA-1P21 SA 3000 to 6000 RCA-1P28 S-5 2000 to 6500 RCA-6217 S-10 3300 to 8000 RCA-7102 S-l 4000 to 10000 EMI-6256 S-13 1650 to 6700 EMI-9558Q S-20 1650 to 8500 FW-118 S-l 3200 to 10000 Dumont6291 S-ll 3200 to 6000 NOTE- R: Hamamatsu TV Co., Ltd., Hamamatsu-City, Japan. RCA: Radio Corporation of America, Harrison, N. J. EMI: E.M.I. Electronics Ltd., Middlesex, England. FW: International Telephone and Telegraph Corporation, Fort Wayne, Ind. Dumont. Allen B. Dumont Laboratories, Inc., Clifton, N. J. titanium, or silica on the determination of alkaline-earth metals. The effects of aluminum and phosphate on magnesium for the total consumption and laminar flow burners are given in Figs. 4 and 5. The mechanism of this interference is discussed in great detail by many investigators [19-21]. In general, most of the workers agree that the suppression is caused by an interferent which affects the vaporization rate of the alkaline-earth element such as magnesium from the solid to the gaseous state. To eliminate or control chemical interferences in atomic absorption, it is essential that various parameters be optimized. These include the type of burner (laminar flow or total consumption), oxidant fuel, flame temperature, flame region, and sample medium. The premixed gas or laminar flow burner is used primarily in absorption. This system suffers less from chemical interferences than the diffusion or total consumption burner because smaller droplets are introduced into the flame and the burning velocity of the oxidant-fuel mixture is low. The flame temperature and the flame region play a major role in the extent of the interference. With a fuel-rich nitrous oxide-acetylene flame the chemical interference of aluminum, titanium, or silicon on the alkaline-earth metals is removed. However, this high-temperature flame usually introduces another type of interference which is due to ionization of the analyte. The alkali and alkaline-earth metals are especially susceptible to ionization, but this is controlled by the addition of a cation having a similar or lower ionization potential than that of the analyte. Another way to eliminate the suppression of the analyte due to formation of involatile compounds is by the addition of releasing or protective chelating agents. Releasing or protective chelating agents are defined as substances 214 RAINS ON CHEMICAL ASPECTS IOU 80 - 60 - 40 A- BURNER-LAMINAR FLOW HOT MODE \a 20 Mg, /ig/ml-O.l i i i i 100 80 60 40 20 B- BURNER-PRE MIX Mg, /ig/ml-l.O 00 80 ^\b 60 40 c- BURNER-TOTAL CONSUMPTION C 20 - Mg,/ig /ml-l.O " i i i 1 0.001 0.01 0.1 1.0 PHOSPHATE, M FIG. A -Effect of phosphate on magnesium in absorption for three different burners: A and B-air-acetylene flame; C-oxy gen-hydrogen, (a) H3PO4 (b) glycerol, 10 (v/v)% plus HClOa, 0.1 M (c) lanthanum, 1 mg/ml (d) (NHtiifiPO* (3) (NHtitflPOs plus glycerol, 10 (v/v)% and HCIO4, 0.1 M. 215 ATOMIC ABSORPTION SPECTROSCOPY 100 80 60 - 40 20 i ! BURNER -TOTAL CONSUMPTION Mg,^g/ml-l BURNER-LAMINAR FLOW HOT MODE Mg,^g/ml -0.1 1.0 10 ALUMINUM, Mg/ml FIG. 5-Effect of aluminum on magnesium in absorption with and without releasing agents: (a) aluminum (b) glycerol, 10 (v/v)% plus HCIO4, 0.1 M (c) lanthanum, 1 mg/ml (d) EDTA, 0.05 M (e) lanthanum, 1 mg/ml; glycerol, 10 (v/v)%and HCIO4, 0.1 M. which, when added in sufficient quantity in the presence of an interferent, will restore the absorption of the analyte to its original value. The manner in which these chelates and releasing agents prevent condensed phase type of interference or an interference that occurs in the solid state is not understood clearly. Compounds such as ethylenediaminetetraacetic acid (EDTA) are known to form strong complexes with the alkaline-earth metals, and will preferentially complex the alkaline-earth metals even in the presence of phosphate or aluminum. However, the similar action of the simple sugars and glycerol has not been explained. A list of releasing and protective chelating agents is given in Table 7. The effects of phosphate in the analysis for magnesium by atomic absorption with and without releasing and protective chelating agents for three different burner systems are given in Fig. 4. The suppression of magnesium is far greater with the total consumption burner (Fig. AC) than with the laminar flow burner. This is due to the high-burning velocity of the oxygen -hydrogen flame and the larger droplet size produced by this type of burner. Also, a solution 10 percent in glycerol and 0.1 -M perchloric acid (HCIO4) was more effective than lanthanum in releasing magnesium. With the laminar flow burner, phosphate as orthophosphate (H3PO4) is not a major interferent; however, ammonium orthophosphate (NH^HPC^) or H3PO4 adjusted to a pH of 4 to 5 interferes strongly. In a solution containing glycerol [10 volume/volume (v/v) percent] and 0.1 M in perchloric acid, 216 RAINS ON CHEMICAL ASPECTS TABLE 7 -Releasing and protective chelating agents used in absorption. Reagent Type Interferent Analyte Ref La releaser Al, Si, PO^TSO^ 2 " Mg, Ca 22,23 Sr releaser Al, B, Te, Se, NO3 Mg Ca 2 4, 23, 25 P04 3 7S0 4 2_ 2 _ Ba 26 Mg releaser Al, Si, P0 4 7S0 4 Ca 2? Ca • • • releaser Al, P04 3 ~ Mg, Sr 27 Glycerol, HCIO4 • • • protector [Al, Fe,Th, rare earths, [Mg,Ca, Sr,Ba] 28,29 Si,B,Cr,Ti,P04 S04 2 "] NH4CI protector Al Na,Cr 30 NH4CI protector Sr,Ca,Ba,P04 3 7 S04 2- Mo 31 Ethylene glycol .... protector PO4 Ca 29 Mannitol protector PO4 Ca 29 Dextrose, sucrose . . . protector PO4 Ca, Sr 29 EDTA chelate [Se,Te, B, Al, Si, Mg, Ca 26 N0 3 1 ~P0 4 3 ~S04 2 ~ 8-hydroxyquinoline . chelate Al Mg, Ca 32 recovery of 1 /Lzg/ml of magnesium was complete in the presence of 0.05 -M phosphate. By increasing the acidity to a pH of 0.5, complete recovery was obtained up to 0.5 -M phosphate. Lanthanum is used frequently as a releasing agent for the alkaline-earth metals in the presence of aluminum or phosphate. The releasing action of lanthanum has been postulated to be due to the high heat of formation of lanthanum phosphate [22] . Dinnin [33] suggested that the releasing action is a mass action effect, Ax+By^Ay+Bx where : A = analyte, B = releasing agent, x = anion which forms a refractory compound, and y = anion which forms an easily dissociated compound. As the concentration of B is increased, the reaction will be forced toward Bx, thus leaving Ay free to dissociate at a lower temperature. This reaction alone will not explain why some cations will act as releasing agents and others will not. Since the process which occurs within the flame gases is rate controlled, as indicated by studies with low-velocity flames and variable heights within the flame, the correct explanation is most probably a combination of the two mechanisms. Lanthanum was found to be more effective than glycerol or EDTA for the determination of magnesium or calcium in the presence of aluminum, as shown in Fig. 5. However, a weight ratio of 10:1 for lanthanum-to- 217 ATOMIC ABSORPTION SPECTROSCOPY aluminum was required to provide complete recoveries of magnesium or calcium. The high concentration of lanthanum required to release the alkaline-earth metals in the presence of aluminum will often produce burner clogging. If, however, glycerol is used in conjunction with the lanthanum, a lanthanum-to-aluminum ratio of only 2:1 is necessary for complete release of alkaline-earth metals. Also, the addition of a 10 percent solution of glycerol to the standards, as well as to the unknown, has the advantage of improving the burning characteristics and controlling the viscosity, thus increasing the accuracy and precision of the method. Rubeska and Moldan [21] have observed that the releasing effect of lanthanum is facilitated by the addition of 8-hydroxyquinoline. Instead of requiring a 1 percent solution of lanthanum to overcome the suppression of aluminum, a 0.2 percent solution of lanthanum in 1 percent 8-hydroxyquinoline suffices. This corresponds to a lanthanum/aluminum molar ratio of 2. Similar results have been reported using a combination of EDTA and lanthanum [34] . Role of Organic Solvent The sensitivity and detection limit can be improved with organic solvents. The increase in sensitivity is not as great in absorption as reported for flame emission; however, the possibility of concentrating the analyte by solvent extraction techniques permits the determination of many trace elements. Also, chemical, as well as physical interferences, often can be removed and higher accuracy obtained. In absorption the addition of miscible organic solvents such as alcohols, ketones, and organic acids to an aqueous solution is of limited value. Usually the increase in sensitivity is less than fivefold, and the analyte is diluted inevitably by the addition of the solvent. By the judicious selection of an organic ligand and solvent, many trace elements are removed from the sample matrix and then determined by atomic absorption spectrometry. Table 8 lists many of the organic ligands and solvents used in emission and absorption spectrometry. Ammonium pyrrolidine dithiocarbamate (APDC), diethylammonium diethyldithiocarbamate (DDDC), thenoyltrifluoroacetone (TTA), and dithizone are used extensively in chelate extractions for absorption as well as flame emission. Many of these organic ligands are made selective by controlling the acidity and the use of masking agents. The chelates formed can be extracted readily with a variety of solvents. The most frequently used organic solvent is methyl isobutyl ketone (MIBK). Benzene and kerosene are inferior to MIBK because they produce a large luminous and smoky flame. Carbon tetrachloride or chloroform can be used; however, the enhancement is considerably less than with MIBK. Apart from the fact that the solvent should extract the chelate, it should burn and provide a stable flame. In addition, the physical properties of the solvent such as viscosity, surface tension, boding point, and mutual solubility in an aqueous medium must be taken into account. It should not produce toxic products during combustion or give a high background in the flame. 218 RAINS ON CHEMICAL ASPECTS TABLE ^-Organic ligands and solvents used in solvent extraction for atomic absorption and flame emission spectrometry. Organic Ligand Solvent Element Material Ref HAA .... APDC .... MIBK MIBK Be Co,Cu,Fe, Pb,Zn Ni, air sample sea water 35 36 Cu,Zn Pb, Bi, Hg, Co,Cu,Pb, Ni.Sb Cd.Fe plants, soils urine KC1 37 38 39 Mn,Ni Cd,Tl,Hg blood, urine 40 Cupferron . . . MIBK Al V Mg alloys,limestone 41 42 DDC DDDC .... MIBK CHCI3 CHCI3 Te Cu, Ni, Mn Cu,Mn Al alloys 43 44 45 Dithizone . . . Oxine CHCI3 MIBK Butanol As Hg Mg Mo cast iron, Se natural water,brine Cb and Ta 46 47 48 49 3-methyl-l-butanol MIBK Ca Ca, Mg, Fe Ni H3PO4 Al 50 51 Naphthenic Acid .... MIBK Ca 52 Salicylic acid, di-n- butylamine. TTA MIBK MIBK Ag Cu zinc 53 8 Al Fe,Cu,Pb,Bi rare earths Sb,Fe,Mo,Se, Sn, Au, In, Pb,Cb minerals,bronzes, Zn-base alloys 41 54 55 NOTE- HAA = acetylacetone APDC = ammonium pyrrolidine dithiocarbamate Cupferron = N-mtrosophenylhydroxylamine DDC = sodium diethyldithiocarbamate DDDC = diethylammonium diethyldithiocarbamate Dithizone = diphenyltluocarbazone Oxine = 8-hydroxyquinoline TTA = 2-thcnoyltrifluoroacetone The major objection to MIBK is the increased solubility in a highly acid medium. By using a 3:1 mixture of MIBK-cyclohexane the solubility in the aqueous phase is decreased without any significant change in the extraction of the various chelates or burning characteristics of the ketone. 219 ATOMIC ABSORPTION SPECTROSCOPY Preparation of Standards, Samples, and Evaluation of Data The standard stock solutions are prepared best in hydrochloric or nitric acids from high-purity metals or nonhygroscopic reagent grade chemicals. If possible, sulfuric or phosphoric acids should be avoided as they produce an adverse effect on many elements. The stock solutions are prepared generally at concentrations of 1.000, 5.000, or 10.000 mg/ml. The standards are prepared by an appropriate dilution from the stock solution as needed. The final anion concentrations of the standards and samples should be approxi- mately the same. The final accuracy of the analysis may depend upon the care in the preparation of the standard solutions. In atomic absorption a curve is prepared to cover the appropriate concentration range. Usually, this means the preparation of standards which produce an absorption of to 80 percent. The correct method for plotting data derived from an atomic absorption instrument equipped with a linear readout system is to convert percent absorption to absorbance and plot the absorbance on linear coordinates against concentration. The following relationship is used to convert absorption values to absorbance: absorbance = log (1 00/%T) = 2 - log %T where %T = 100 — % absorption. Frequently, the curve is nonlinear especially at high absorption values. These curves can be used; however, the number of standards should be increased especially in the nonlinear portion of the curve. If the nonlinear portion of the curve is used it is essential that the concentration and the absorbance be very carefully read. The sample should be prepared by a standard laboratory procedure. Many of the high refractory oxides and silicates are dissolved with hydrofluoric acid in a Teflon lined bomb [56]. By this technique the silicates and aluminates are dissolved within 4 h at a temperature of 120 to 140 C. The excess fluoride is removed by fuming the sample with HCIO4. This procedure eliminates the need for a basic fusion to dissolve the refractory oxides and silicates which contribute to high concentration of undesirable salts. The final choice and concentration of acid, such as HCIO4, hydrochloric acid (HC1) or nitric acid (HNO3) are important since the acid concentration may affect the rate of atomization and the production of atomic vapor. Frequently, a separation or preconcentration step is required before making the final measurement. Ion-exchange resins are helpful in separating trace elements from the matrix; however, ion-exchange procedures have a tendency to be time consuming. Preconcentration by coprecipitation is usually fast and very effective. An excellent example is the coprecipitation of barium with lead chromate. Usually, a solvent extraction procedure is the best method for separation and preconcentration. To check interferences and to determine the need for separations or the use of releasing agents, the method of standard additions can be utilized. 220 RAINS ON CHEMICAL ASPECTS While there are only two basic methods (single and multiple) for standard addition, the method has a variety of names such as the standard addition method, mixture, incremental addition, autostandarization, and method of additions. The multiple addition method generally is preferred to the single addition; however, if improperly used the multiple method can lead to a serious error. The curve must be linear, and corrections must be made for any absorption which results from scattered light or reagents. This method requires the transfer of three to five test portions of the sample to volumetric flasks. One flask is diluted to volume, and incremental amounts of the standard solution containing 1 to 4 times the concentration of the analyte are added to the remaining flasks and then diluted to volume. The absorbance is measured and plotted against the concentration on linear graph paper. The concentration is determined graphically by extrapolating the linear curve through the Y axis until the line intersects the X or concentration axis. The concentration of the unknown is determined from the point of intersection [57\. In the single addition method duplicate aliquots of the sample are transferred to flasks, A and B. An aliquot of a standard from a stock solution is added to B and both solutions are then diluted to volume. The concentrations of the standard and the analyte should be approximately equal. If there is a suppression or enhancement due to the matrix, the standard addition will be affected in a similar manner. The absorptions of Samples A and B are determined, and their concentrations are determined from a calibration curve. If the recovery was either high or low, the true concentration is determined by the following equation [58] . ■=C Y-X where: X =Mg per ml of analyte in unknown (A) as determined by the calibration curve, S = standard added, Mg per ml, Y = Mg per ml of analyte found in unknown (B) containing the standard addition as determined by the calibration curve, and C = corrected concentration of analyte, ^g per ml. Advantages of the single addition over the multiple addition include: (1) only two test portions are required, (2) the method may serve as a duplicate determination when there is no interference, and (3) it is not essential to prepare the standards for the calibration curve in the same matrix as in the unknown sample. Acknowledgments The author is indebted to O. Menis and J. I. Shultz for their assistance in the preparation of this manuscript. 221 ATOMIC ABSORPTION SPECTROSCOPY References \1 [2 [3 \4 [5 [6 [7 [* [9 [10 [11 [12 [13 [14 [15 [16 117 [18 [19 [20 [21 [22 Walsh, A., 'The Application of Atomic Absorption Spectra to Chemical Analysis," Spectrochimica Acta, Vol. 7, 1955, pp. 108-117. Ramrrez-Munoz, J., "Qualitative and Quantitative Sensitivity in Flame Photometry," Talanta, Vol, 13, 1966, pp. 87-101, Kaiser, H., "Zur Definition der Nachweisgrenze, der Garantiegrenze und der dabei benutzten Begriffe," Zeitschrift fuer Analytische Chemie,\o\. 216, 1966, pp 80-94. Slavin, W., Sprague, S., and Manning, D. C, "Detection Limits in Analytical Atomic Absorption Spectrophotometry," Atomic Absorption Newsletter, No. 18, Feb. 1964. Winefordner, J. D. and Vickers, T. J., "Calculation of the Limit of Detectability in Atomic Absorption Flame Spectrometry," Analytical Chemistry, Vol. 36, 1964, pp. 1947-1954. St. John, P. A., McCarthy, W. J., and Winefordner, J. D., "A Statistical Method for Evaluation of Limiting Detectable Sample Concentrations," A nalytical Chemistry, Vol. 39, 1967, pp. 1495-1497. Slavin, W., "Atomic Absorption Spectroscopy-A Critical Review," Applied Spectroscopy, Vol. 20, 1966, pp. 281-288. Menis, O., ed., "Analytical Coordination Chemistry: Titrimetry, Gravimetry, Flame Photometry, Spectrophotometry, Gas Evolution, and Isotopic Preparation, July 1965 to June 1966," Technical Note 402, National Bureau of Standards, U.S. Department of Commerce, 1967. Koirtyohann, S. R., "Recent Developments in Atomic Absorption and Flame Emission Spectroscopy," Atomic Absorption Newsletter, Vol. 6, 1967, pp. 77-84. Meggers, W. F. and Westfall, F. 0., "Lamps and Wavelengths of Mercury 198," Journal of Research of the National Bureau of Standards, Vol. 44, 1950, pp. 447455. Ivanov, N. P. et al, "Electrodeless Tubes with High-Frequency Excitation of the Spectrum of In, Ga, Bi, Sb, Tl, Pb, Mg, Ca, and Cu as a Radiation Source in Atomic Absorption Analysis," Journal of Analytical Chemistry of USSR, Vol. 21, 1966, pp. 998-1000. Tomkins, F. S. and Fred, M., "Electrodeless Discharge Tubes Containing Rare Earth and Heavy Elements Halide," Journal of the Optical Society of America, Vol.47, 1957, pp. 1087-1091. Mavrodineanu, R. and Hughes, R. C, "Excitation in Radio-Frequency Dis- charges," Spectrochimica Acta, Vol. 19, 1963, pp. 1309-1317. Fehsenfeld, F. C, Evenson, K. M., and Broida, H. P., "Microwave Discharge Cavities Operating at 2450 MHz," Review of Scientific Instruments, Vol. 36, 1965, pp. 294-298. Mavrodineanu, R. and Boiteux, H., Flame Spectroscopy, Wiley, New York, 1965. Kniseley, R. N., "Flames for Atomic Absorption and Emission Spectrometry," chapter in Flame Emission and Atomic-Absorption Spectrometry, Dean, J. A. and Rains, T. C, eds. ; Dekker, New York, 1969. Willis, J. B., "Some Properties of Nitrous Oxide Supported Flames," Seventh National Meeting, Society for Applied Spectroscopy, 13-17, May 1968. Woodriff, R. and Ramelow, G., "Atomic Absorption Spectroscopy with a High-Temperature Furnace," XIII Colloquium Spectroscopicum Internationale Ottawa, Adam Hilger, London, 1968, pp. 283-284. Gibson, J. H., Grossman, W. E. L., and Cooke, W. D., "Excitation Process in Flame Spectrometry "Analytical Chemistry, Vol. 35, 1963, pp. 266-277. Fuskushima, S., "Mechanism and Elimination of Interferences in Flame Photometry," Mikrochimica Acta, 1959, pp. 596-618. Rubeska, I. and Moldan, B., 'The Mechanism of Interference Effects and Their Elimination in the Determination of Alkaline Earth Metals by Flame Photometry ," Analytica Chimica Acta, Vol. 37, 1967, pp. 421428. Yofe, J. and Finkelstein, R., "Elimination of Interferences in Flame Photometric Determination of Calcium in the Presence of Phosphate and Sulphate," Analytica Chimica Acta, Vol. 19, 1958, pp. 166-173. 222 RAINS ON CHEMICAL ASPECTS [23] Dickson, R. E. and Johnson, C. M., "Interferences Associated with the Determination of Calcium by Atomic Absorption," Applied Spectroscopy., Vol. 20, 1966, pp. 214-218. [24] Elwell, W„ T. and Gidley, J. A. F., Atomic Absorption Spectrophotometry, 2nd ed., Pergamon, London, 1966. [25] David, D. J., 'The Determination of Exchangeable Sodium, Potassium, Calcium, and Magnesium in Soil by Atomic Absorption Spectrophotometry "Analyst, Vol. 85, 1960, pp. 495-503. [26] Ramakrishna, T. V., Robinson, J. W., and West, P. W., "The Determination of Calcium and Magnesium by Atomic Absorption Spectroscopy," Analytica Chimica Acta, Vol. 36, 1966, pp. 57-64. [27] Rocchiccioli, C. and Townshend, A., "Some Interferences in the Atomic Absorption Spectrophotometry of Calcium," Analytica Chimica Acta, Vol. 41, 1968, pp. 93-98. [28] Rains, T. C, Zittel, H. E., and Ferguson, M., "Flame Spectrophotometric Determination of Micro Concentrations of Strontium in Calcareous Material," Analytical Chemistry, Vol. 34, 1962, pp. 778-781. [29] Rains, T. C, Zittel, H. E., and Ferguson, M., "Elimination of Anionic Interferences in Flame Spectrophotometric Determination of Calcium," Talanta, Vol. 10, 1963, pp. 367-374. [30] Barnes, L., Jr., "Determination of Chromium in Low Alloy Steels by Atomic Absorption Spectrometry," A nalytical Chemistry, Vol. 38, 1966, pp. 1083-1085. [31] David, D. J., 'The Suppression of Some Interferences in the Determination of Molybdenum by Atomic-Absorption Spectroscopy in an Air-Acetylene Flame," Analyst, Vol. 93, 1968, pp. 79-82. [32] Marshall, G. B. and West, T. S., "Determination of Traces of Calcium, Magnesium, Iron, and Nickel in Aluminum Salts by Atomic-Absorption Spectrophotometry with a Microwave-Excited Source and Hollow-Cathode Lamps," Talanta, Vol. 14, 1967, pp. 823-831. [33] Dinnin, J. I., "Releasing Effects in Flame Photometry Determination of Calcium," Analytical Chemistry, Vol. 32, 1960, pp. 1475-1480. [34] Adams, P. B. and Passmore, W. O., "Critical Factors in the Determination of the Alkaline Earth Elements in Glass by Atomic Absorption Spectrometry," Analytical Chemistry, Vol. 38, 1966, pp. 630-633. [35] Bokowski, D. L., "Rapid Determination of Beryllium by a Direct Reading Atomic Absorption Spectrophotometer," Atomic Absorption Newsletter, Vol. 6, 1967, pp. 97-100. [36] Brooks, R. R., Presley, B. J., and Kaplan, I. R., "APDC-MIBK Extraction System for the Determination of Trace Elements in Saline Waters by Atomic-Absorption Spectrophotometry," Talanta, Vol. 14, 1967, pp. 809-816. [37] Allan, J. E., 'The Determination of Copper by Atomic Absorption Spectro- photometry," Spectrochimica Acta, Vol. 17, 1961, pp. 459-466. [38] Willis, J. B., "Determination of Lead and Other Heavy Metals in Urine by Atomic Absorption Spectroscopy, "Analytical Chemistry, Vol. 34, 1962, pp. 614-617. [39] Sprague, S. and Slavin, W., "Determination of Very Small Amounts of Copper and Lead in KC1 by Organic Extraction and Atomic Absorption Spectrophotometry," Atomic Absorption Newsletter, No. 20, May, 1964. [40] Berman, E., "Determination of Cadmium, Thallium and Mercury in Biological Materials by Atomic Absorption," /Iromz'c Absorption Newsletter, Vol. 6, 1967, pp. 57-60. [41] Eshelman, H. C. et al, "Extraction and Flame Spectrophotometric Determination of Aluminum," Analytical Chemistry, Vol. 31, 1959, pp. 183-187. [42] Sachdev, S. L., Robinson, J. W., and West, P. W., "Determination of Vanadium by Atomic Absorption Spectrometry," Analytica Chimica Acta, Vol. 37, 1967, pp. 12-19. [43] Dean, J. A. and Simms, J. C, "Flame Spectrometric Determination of Tellurium," Analytical Chemistry, Vol. 35, 1963, pp. 699-702. [44] Dean, J. A. and Cain, C, "Flame Spectrophotometric Determination of Copper, Nickel, and Manganese in Aluminum-Base Alloys," Analytical Chemistry, Vol. 29, 1957, pp. 530-532. 223 ATOMIC ABSORPTION SPECTROSCOPY [45] Wyatt, P. F., "Diethylammonium Diethyldithiocarbamate for the Separation and Determination of Small Amounts of Metals," Analyst, Vol. 78, 1953, pp. 656-661. [46] Menis, O. and Rains, T. C, "Determination of Arsenic by Atomic Absorption Spectrometry with an Electrodeless Discharge Lamp as a Source of '. Radiation"," Analytical Chemistry, submitted for publication. [47] Feldman, C. and Dhumward, R., "An Atomic Absorption Tube for Use with an Atomizer-Burner: Application to the Determination of Mercury," U.S. Atomic Energy Commission TID-7655, 1962, pp. 379-389. [48] Suzuki, M., Yanagisawa, M., and Takeucki, T., "Determination of Magnesium by Atomic, Absorption Spectrophotometry after Extraction," Talanta, Vol. 12, 1965, pp. 989-996. [49] Kirkbright, G. F., Peters, M. K., and West, T. S., "Determination of Small Amounts of Molybdenum in Niobium and Tantalum by Atomic-Absorption Spectroscopy in a Nitrous Oxide-Acetylene Flame," Analyst, Vol. 91, 1966, pp. 705-708. [50] Yanagisawa, M., Suzuki, M., and Takeuchi, "Determination of Traces of Calcium in Phosphoric Acid and Its Salts by Atomic Absorption Spectrometry," Talanta, Vol. 14, 1967, pp. 933-936. [51] Marshall, G. B. and West, T. S., "Determination of Traces of Calcium, Magnesium, Iron, and Nickel in Aluminum Salts by Atomic Absorption Spectrophotometry with a Microwave-Excited Source and Hollow-Cathode Lamps," Talanta, Vol. 14, 1967, pp. 823-831. [52] Szego, L. and Calvert, J. M., "Naphthenic Acid Extraction of Calcium for Determination by Atomic Absorption Spectrophotometry," Analytica Chimica Acta, Vol. 42, 1968, pp. 1-6. [53] Belcher, R., Dagnall, R. M., and West, T. S., "An Examination of the Atomic Absorption Spectroscopy of Silver," Talanta, Vol. 11, 1964, pp. 1257-1263. [54] Rains, T. C, House, H. P., and Menis, O., "Flame Spectra of Sc, Y, and Rare-Earth Elements" Analytica Chimica Acta, Vol. 22, 1960, pp. 315-327. [55] Menis, O. and Rains, T. C, "Sensitivity, Detection Limits, Precision and Accuracy in Flame Emission and Atomic Absorption Spectrometry," chapter in Analytical Flame Spectroscopy, Selected Topics, Mavrodineanu, R., ed., Centrix, Eindhoven, 1969. [56] Bernas, B., "A New Method for Decomposition and Comprehensive Analysis of Silicates by Atomic Absorption Spectrometry," Analytical Chemistry, Vol. 40, 1968, pp. 1682-1688. [57] Willard, H. H., Merritt, L. L., Jr., and Dean, J. A., Instrumental Methods of Analysis, 4th ed., Van Nostrand, Princeton, N. J., 1965, pp. 342-344. [58] Dean, J. A., Flame Photometry, McGraw-Hill, New York, 1960. 224 APPENDIX XII Reprint from Soecial Technical Publication 564 American Society for Testing and Materials 1916 Race Street, Philadelphia, Pa. 19103 1974 T. C. Rains 1 Atomic Absorption Spectrometry General Considerations for the Application of Experimental Techniques REFERENCE: Rains, T. C, "Atomic Absorption Spectrometry— General Con- siderations for the Application of Experimental Techniques," Some Fundamentals of Analytical Chemistry, ASTM STP 564, American Society for Testing and Ma- terials, 1974, pp. 50-67. ABSTRACT: In applying atomic absorption spectrometry to an analytical problem, the analyst must have an understanding of the basic instrumental components. The essential components of an atomic absorption instrument consist of a primary source of radiation, a means of producing atomic vapor from the analyte, wave- length selection, signal detection, and readout. While several types of radiation sources are available, hollow cathode lamps are the primary choice for most work- ers in the field. The production of atomic vapor of the analyte is probably the most important parameter in atomic absorption spectrometry. The selection of the oxidant-fuel or nonflame method of producing neutral atoms will depend upon the concentration of analyte and the matrices. While the analyst has a wide choice of methods at his disposal, an understanding of the problems related to the production of atomic vapor is essential to obtain the maximum precision and accuracy. Also, inter- ferences, sample preparation, and methods for evaluation of data are discussed. KEY WORDS: chemical analysis, accuracy, atomic spectroscopy, evaluation, inter- ferences, light sources, precision, atomic vapor Atomic absorption spectrometry (AAS) is a unique anaytical technique for the determination of metallic elements. To take advantage of the method, it is essential that the analyst become familiar with the basic components of his instrument and have a working knowledge of the chemical processes ocurring in the production of atomic vapor. The results obtained by AAS, for the most part, are dependent upon the ability of the analyst to apply many of the experimental techniques which have been 'Research chemist, Analytical Chemistry Division, National Bureau of Standards, Wash- ington, D.C. 20234. 225 RAINS ON ATOMIC ABSORPTION SPECTROMETRY developed over the years. The principles and practices discussed here are not the only techniques that can be applied but are intended only to act as guidelines. Essential Features of Atomic Absorption Spectrometry Sources of Radiation The basic components of all atomic absorption spectrometers are depicted in Fig. 1. The function of the primary source of radiation in AAS Primary Source Production of Wavelength Radiation of Radiation (HCL) -* Atomic Vapor (Flame or Non-Flame) — > Isolator ( M on oc h rom ator or Filter) -* Detector (Multiplier Phototube) -* Readout (Digital or Recorder) FIG. 1 — Essential features of an atomic absorption spectrometer. is to generate radiation at the wavelength at which absorption is to be measured. This source should radiate the wavelength of the resonance lines of the analyte without interference from other spectral lines. The resonance lines should be sharp with very low background. At the present state of the art, a number of light sources are commercially available which fulfill these requirements in AAS, for example, hollow cathode, vapor discharge, and electrodeless discharge lamps. Hollow cathode lamps (HCL) are available for some 60 elements and, in general, are satisfactory for most AAS work. Although the operation of hollow cathode lamps is extremely simple, one may sometimes obtain better results by deviating from the recommended conditions. The choice of operating conditions depends to a great extent on whether the operator is searching for a maximum precision at concentrations near the detection limit or over a wide analytical range. To determine the optimum conditions, it is necessary to examine the effect of absorbance of an analyte by varying the lamp current and the spectral bandpass of the monochromator. The principal effect of increasing lamp current is to increase the light intensity; however, this may lead to an increase in line width as a result of self-absorption or Doppler broadening. These effects are discussed in great detail by Butler and Brink [I]. 2 In general, the best sensitivity is obtained with a low current; however, a poor signal-to-noise ratio may be experienced because of the resultant low intensity. As a result one should balance the benefits of maximum signal-to-noise against sensitivity. Also, the spectrum produced by the lamp should be scanned over the area of analytical importance and all lines identified. This serves as a warning of possible spectral interference due to the carrier gas or impurities in 2 The italic numbers in brackets refer to the list of references appended to this paper. 226 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY cathode material and as a means of correcting light scatter with a nonabsorbing line. Batchelor [2] has reviewed the present state of development of the HCL. The major emphasis by manufacturers has been on improvement in stability and greater light intensity by electrode design and control of the carrier gas pressure. Other radiation sources used include vapor discharge lamps and electrodeless discharge lamps. Vapor discharge lamps are known for their high intensity; however, self-absorption is inherently present in these lamps which limits their sensitivity. Electrodeless discharge lamps (EDL) were originally described by Jackson [3]. Meggers and Westfall [4] reported the construction of an EDL using 198 Hg as a spectroscopic source. With the single element isotope, the lamp gave very sharp spectral lines which were measured to eight signifi- cant figures. The subsequent development of lamps using metallic halides [5-9] has made available light sources for wavelength calibration, atomic absorption, and fluorescence spectrometry. While EDL's are known for their high spectral intensity, they lack reproducibility and ease of operation. The difficulty encountered with lamp drift has been overcome by controlled heating of the lamp [10,11]. More recently two commercial manufacturers have made available EDL's having power supplies which operated at radio-frequencies of 29 and 951 mHz. These lamps for a few elements, for example, arsenic and selenium, produce relatively high emission signals with excellent stability. However, the limiting factor for these sources is the number of elements which can be excited at these frequencies. Sources of Atomic Vapor Production Historically a flame was the original means of producing atomic vapor for AAS. The flame is still the basic source for the vast majority of AAS measurements, and it will probably remain so for the foreseeable future. While a lot of development work is in progress with nonflame techniques (graphite furnace, carbon rod, and tantalum strip) they are not expected to replace the flame except where additional sensitivity is needed. There are two basic type of nebulizer burners. When the liquid is sprayed into a mixing chamber and the droplets are mixed with the ..combustion gas, the process is called indirect nebulization and is referred to as a premixed chamber or laminar flow burner system. In the second type the aerosol passes directly into the flame where the fuel and support gas are mixed. This is the so-called turbulent (sprayer) burner. Of the two types of nebulizer-burners, the premixed system is used generally in all commercial systems. Schematics of each type are given by Herrmann [12]. The aerosol produced by an indirect nebulizer is mixed in the spray chamber with the combustible gas before entering the burner. This spray 227 RAINS ON ATOMIC ABSORPTION SPECTROMETRY chamber also performs another function by removing the large droplets before they reach the flame. To achieve optimum sensitivity in AAS the spray chamber should produce uniform fine aerosol particles (less than 10 (iin in diameter) which are mixed with the support gas. Normally, the efficiency of this system is less than 10 percent. The tvoical spray chamber used in AAS has varied very little from the original design used by Walsh [13]. This can be attributed to one of two causes, either the original design left nothing to be desired or there has been a surprising lack of initiative by investigators. Various ways to improve the efficiency of atomization in a flame have been proposed. One is the so-called "sample boat" system. In this system, a few microliters of the sample are placed in a small tantalum boat. The boat is mounted on a sliding device that allows it to be pushed into the flame in a reproducible manner. The sample is first dried in the boat near the flame. After drying, the boat is inserted into the flame, and the metal content is rapidly atomized to form a transient atomic vapor cloud above the boat. The radiation source is focused just above the top of the boat. A sharp peak of absorbance versus time is recorded, and the height is proportional to the concentration of metal vaporized. This technique is most successful for the more volatile elements. The major disadvantage of the system is the inability to reproducibly dry the sample before introduc- tion it into the flame. Delves [14] modified the sampling boat technique for the determination of blood lead by vaporizing samples from nickel crucibles into a quartz absorption tube. Blood lead results were superior with this system. Also, there were fewer interferences with increased sensitivity. An entirely different concept of introducing the analyte into the flame is that of hydride generation. In this method, the sample in solution is reacted in a flask with stannous chloride and potassium iodide solutions and finally with zinc metal [15] or sodium borohydride [16] to produce hydrides such as AsH 3 , H 2 Se, and SbH 3 . The gas is swept from the solution with argon carrier gas into an argon (entrained air)-hydrogen flame in which the hydride is readily decomposed to give the metal (atomic vapor). The sensitivity from this technique is much greater than for direct aspiration of the solution into a flame. For arsenic the sensitivity is 1 ng/ml/1 percent absorption. In summary, the major advantages of the flame as a means of production of atomic vapor are: (a) simplicity of the technique, (b) speed with which a determination can be made, and (c) relative little maintenance of system. The major disadvantages are: (a) a relative large quantity of sample is required for a determination, (b) a hostile environment is created for the production of ground state atoms, (c) large quantities of oxidant and fuel gases are required, and (d) sensitivity is limited for many elements. 228 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY Nonflame Absorption Devices A relatively new technique of producing atomic vapor is with nonflame devices. These methods differ from flames in that an environment is provided which is more conducive to production of atoms in their ground state. Also, a relatively small quantity of the sample (1 to 50 \A) is necessary for an analysis which gives these methods improved sensitivity over flame methods. However, high precision is more difficult to attain. Nonflame methods can be grouped into categories such as: (a) cathodic sputtering, (b) graphite furnace or carbon rod devices, (c) tantalum ribbon, and (d) cold vapor. Cathodic sputtering was first described by Gatehouse and Walsh [17\. It consists of making the sample the cathode in a demountable hollow cathode lamp. Atomization of the material of the cathode is achieved by the process of cathodic sputtering. A major disadvantage of the technique is the need to prepare a cathode from the sample which tends to restrict it to metallic samples. Dogan et al [18] has demonstrated the feasibility of mixing a nonmetallic sample with a metallic powder and pressing the mixture into pellets. These pellets are buffed and used as the cathode in a demountable lamp. This technique has a potential advantage over other nonflame devices, since it can be used in conjunction with a vacuum monochromator to determine those elements such as carbon, sulphur, phosphorus, and the halides which emit in the far ultraviolet region of the spectrum. As of this date, only limited work has been reported on this technique, but it does possess some interesting analytical possibilities. The first furnace device proposed for AAS was that described by L'vov [19]. In his early work, a solution was placed on the end of the carbon electrode and evaporated to dryness. The sample was then vaporized by a d-c arc into the carbon furnace containing a tantalum liner heated with an a-c current. With this device very impressive absolute detection limits were reported. Typical detection limits ranged from 10" 12 to 10" 14 g of the analyte. However, the analytical capabilities of this device were limited be- cause of the power requirements and poor precision. Massmann [20] constructed a somewhat simpler graphite furnace. In his design, the sample was placed in the graphite tube through the hole in the top, and then the tube was thermally heated to vaporize the sample. Comparison of sensitivities reported by L'vov and Massmann devices are summarized in Table 1. L'vov has shown that the sensitivity is inversely proportional to the square of the diameter of the graphite tube if small amounts of the sample are analyzed. The sensitivity and detection limits in these methods are expressed in absolute mass rather than concentration since the solvent is evaporated before atomization. Modifications to the graphite furnace have been made by Woodriff and Ramelow [21]. West [22] described a slightly different approach at the 229 RAINS ON ATOMIC ABSORPTION SPECTROMETRY TABLE 1 — Sensitivities obtained by atomic absorption with graphite cells. Wavelength, A Sensivity, weight/ 1% absorption Element L'vov Furnace Massmann Furnace Ag 3281 0.8 pg Al 3093 7-7 pg As 1890 0.2 ng Ba 5535 41.0 pg Bi 3068 26.0 pg° 0.2 ng Cd 2288 0.5 pg 2.0 pg Co 3527 200.0 pg* Cr 4254 41.0 pg Cs 4555 3.3 ng° Cu 3248 1.8 pg 9.0 pg Fe 2967/2483 17.0 pg 20.0 pg Hg 2537 170.0 pg" 200.0 pg In 3039 3.0 pg 200.0 pg K 4044 960.0 pg Li 6708 20.0 pg Mg 2852 0.5 pg Mn 2795 0.5 pg" 0.8 pg Mo 3133 11.0 pg a Na 5890 6.0 pg Ni 3525 180.0 pg° 70.0 pg Pb 2833 15.0 pg* 10.0 pg Rb 4202 1.5 ng* Sb 2311 16.0 pg* 0.1 ng Se 1961 1.5 ng Sr 4607 8.2 pg 10.0 pg Te 2259 290.0 pg* Tl 2768 40.0 pg Ti 4982 2.0 ng Zn 2139 0.2 pg* 0.8 pg "Analyses were made at a pressure between 3 and 6 atm. Note: pg = picogram (10" J2 g). ng = nanogram (10" 9 g). International Conference on Atomic Absorption Spectroscopy in Sheffield, England. He used an electrically heated carbon filament or graphite rod as the atomizer. In this device, 1 to 5 pd of the sample was placed on the carbon filament mounted horizontally between two metal supports. To prevent oxidation of the carbon filament, it was placed in a glass cell through which argon flowed. The carbon filament was heated slowly to vaporize the solvent, and a current of 100 A at 7 V was then passed through the filament to vaporize the analyte and produce a cloud of atomic vapor. Absorption was measured by passing the radiation from a hollow cathode lamp just above the filament. At the present, two types of graphite rod or furnace devices are commercially available. These instruments have been redesigned from the 230 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY original designs of Massmann and West for compactness and ease of operation. Both of these devices produce excellent sensitivity for a number of elements. Regardless of the devices selected by the analyst, it is important that he observe certain precautions. The sample (5 to 100 fA depending upon device) must be transferred to the graphite cell in a reproducible manner. Then the sample is dried with very gentle heating. If the heating is too rapid, vigorous boiling and spattering will result. The drying time will vary with different sample types, but it should be long enough to dry the sample completely. If volatile organic and inorganic compounds are present, the sample must be heated at a higher temperature to remove these compounds. This step is frequently referred to as the ashing stage. The selection of temperature and time parameters for the ashing stage is vital for good precision. If the temperature is too low or the time too short, some of the matrix will not be removed and will cause light scattering during the atomization step. On the other hand, if the ashing temperature is too high, some of the more volatile metals will be lost. Usually a setting can be selected so that the matrix is removed but the analyte remains. For some samples the ashing step may have to be repeated several times to completely remove the volatile matrix. The final stage is atomization. The temperature of furnace and time required for complete vaporization will depend upon the analyte and matrix. As a rule, the lowest possible temperature is advisable to increase the life of the graphite rod or tube. To increase the life of these rods or tubes, 3 to 5 percent methane is introduced to the inert gas sweep during the atomization state. Interferences encountered with graphite rods or tubes can be classified as physical, chemical, and interelement in nature. Physical interferences are more pronounced with graphite rods or tubes than in most flame systems. Light scatter due to incomplete volatization of organic or inorganic compounds is of major importance. To minimize this type of interference, the time of ashing should be increased. However, if it persists, background measurements should be made with a continuum light source or at a nearby nonabsorbing line and then substracted from the absorbance value obtained for the analyte. Three possible mechanisms account for chemical interferences. The analyte may be lost by the formation of a volatile compound, by occlusion in a nonvolatile matrix, and by the formation of carbides. As an example, lead chloride which boils at 950° C has a considerable vapor pressure at lower temperatures and can be lost in the ashing step. Therefore, it is frequently necessary to add a reagent which will react with the analyte to produce a less volatile species. Hydrogen peroxide has been observed to minimize the losses of lead in the presence of chloride. Atomic arsenic is volatilized readily from aqueous residues, but arsenic in a metal matrix is volatilized slowly. This is an example of an element forming an occlusion 231 RAINS ON ATOMIC ABSORPTION SPECTROMETRY to produce a less volatile compound. The net result is that the rate of volatilization will vary, and this can lead to erroneous results, especially if peak heights are being measured. Another major interference is from the reaction of the furnace material to form nonvolatile metal carbides. Although carbide formation does occur, it can be controlled with the use of a pyrolytic coating on the inner surface of the rod or tube. Also the continued use of methane during the atomization stage should greatly help to maintain the pyrolytic coating for many hours of operation. The interelement effect is more pronounced in the flameless techniques than in tlame systems. It occurs when the analyte reacts with another element which may produce a suppression at certain atomic ratios and an enhancement under certain other conditions. This type of interference can sometimes be corrected by the standard addition method; however, the best technique is to prepare the standards in a similar matrix as the unknowns [23]. Feldman [24] described a slightly different system in which the carbon filament is replaced by a tantalum strip or ribbon. The major advantage of the tantalum strip over the graphite furnace or rod is the power requirements; however, temperatures high enough to excite the more refractory elements are not obtained with the tantalum strip. A comparison of absolute detection limits for the three systems is given in Table 2. Cold Vapor Method The earliest application of AAS extends many years before the first paper by Walsh [13]. Mercury has an appreciable vapor pressure at room temperature, and the volatile mercury vapor is capable of absorbing at the resonance lines of mercury. This technique has been employed for years by geologists and industrial hygienists for monitoring mercury levels in the at- mosphere. The first use of the technique was described by Poluektov et al in 1964 [25]. In 1968, Hatch and Ott [26] described a similar method. It is well known that mercury and many of its compounds are highly toxic, but the wide distribution of mercury in our environment was not fully realized until our analytical capability was developed to detect mercury at trace and subtrace levels. Since the early publication by Hatch and Ott, there have been numerous publications. A review of the techniques has been made by Manning [27]. Rains and Menis [28] modified the Hatch and Ott method by wrapping the absorption cell with a heating tape and by heating the absorption cell to 200° C. This modification eliminated the need for a drying tube prior to the absorption cell and improved the precision of analysis. Since losses of mercury at the nanogram level may occur due to absorption on the wall of the vessel or to its high vapor pressure, it is essential that an oxidant such as K 2 Cr 2 O y be added to standards and unknowns when these solutions are prepared. Feldman [29] 232 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY TABLE 2 — Detection limits for flameless atomic absorption spectrometry. Detection Limit, g Element Graphite Furnace Carbon Rod* Tantalum Ribbon c Ag 2 x l0-» 2 x l0-» 4 x 10-" Al 3 x 10"" 3 x 10"" 3 x lO"" As 1 x lO-io 2x lO"' Au 8 x 10-" 1 x lO-" 1 x lO"" Be 9 x 10-» 3 x 10-" Bi 1 x lO"" 7 x 10"" Ca 3 x 10" u Cd 1 x lO' 13 1 x 10" 13 2 x 10-" Co 4 x 10"" 6 x 10" 12 4 x lO"" Cr 5 x 10" 12 5 x lO"" 8 x 10-" Cs 2 x 10-» Cu 1 x 10"" 7 x 10" 12 1 x 10"" Eu 1 x 10" 10 Fe 3 x l0-» 2 3 x lO"" Ga 2 x lO"" Hg 1 x lO' 10 5 x l0-i° K 9 x l0-» U 5 x lO' 12 Mg 6 x lO"" 1 x 10-" Mn 1 x 10"" 5 x lO"" 5 x 10-" Mo 4 x 10-" Na 1 x 10"" Ni 1 x lO" 11 1 x lO"" 2 x 10"" Pb 6 x lO" 12 5 x lO-" 7 x 10-" Pd 2 x lO-" Pt 2 x 10-i° Rb 6 x lO"" Sb 2 x NT" 3 x lO"" 9 x l0-i° Se 1 x lO-»° Si 8 x 10-" Sn 6 x 10-» 2 x 10-i° Sr 5 x I0" 12 Tl 1 x lO"" 3 x 10-" V 1 x lO"" 4 x lO-io Zn 6 x 10-" 8 x lO"" 5 x lO-" a Perkin-Elmer Corporation brochure RE 67210, 1972, "High-Sensitivity Sampling Systems tor Atomic Absorption." ^Varian Techtron Carbon Rod Atomizer, Model 63 brochure, April 1972. c Instrumentation Laboratory, Inc., "Sensitivities and Detection Limits in I. L. Model 355 Flameless Sampler," June 1971. studied the effect of 5 percent (volume per volume) HN0 3 plus 0.01 percent Cr 2 7 2 " on 0.1 to 10 ng Hg/ml in glass and found the solutions to remain stable for as long as five months. Optical Design Commercial instruments for AAS are of two basic designs — single beam and double beam. In a single-beam system the source (HCL) is 233 RAINS ON ATOMIC ABSORPTION SPECTROMETRY focused by means of two lenses through the flame and onto the entrance slit of the monochromator. In a double-beam system the light from the source is interrupted periodically, and diverted around the flame, and re- combined before being focused on the entrance slit by a series of mirrors, lenses, and rotating mirrors. The chief advantage of a double-beam system is that the intensity of the resonance line is periodically monitored even while a sample is in the flame. Thus, any source noise, drift, or in- stability may be easily corrected electronically. However HCL's have been improved to the extent that the flame is the limiting factor in precision in AAS. In AAS, the monochromator is an essential part of the system. The monochromator must isolate the analytical line from the various other lines emitted by the source. Failure to resolve the analytical line from all spectral irradiation will result in a loss of sensitivity and nonlinear calibra- tion curve. Another important component of an optical system is the slits to the monochromator. It is desirable that the slit widths be variable as they control the resolution of the monochromator. Normally, we operate at a slit width which gives the desired resolution from any adjacent lines. For some elements the minimum slit width fails to give the desired resolution of the analytical line, and, in that case, a tradeoff is made usually between spectral resolution and sensitivity of the analyte. The effect of slit width on resolution (spectral bandwidth) of the nickel HCL is shown in Fig. 2. The most sensitive AAS ground state resonant line of nickel is located at 2320 A. However, it is very close to a nonresonant line at 2321.4 A, and only by means of a high resolving monochromator can these two lines be com- pletely resolved. The effect of spectral bandwidth on the absorbance of nickel is shown in Fig. 3. Electronic and Readout The basic electronics most often used in AAS consist of a power supply for the HCL, a photosensitive detector and its necessary power supply, an a-c amplifier, and readout device. The choice of a HCL/power supply will depend upon the limit and accuracy of the desired detectability. It will also depend upon the type of instrument that is being used. For precise single-beam measurement, a well-regulated current supply providing 0.1 percent or better current regulation is desirable. With a double-beam system the requirements are less stringent, and a regulated power supply of 1 percent is adequate for most purposes. The multiplier phototube is by far the most widely used detector in AAS. A list of the most commonly used multiplier phototubes is given by Rains [30]. These devices require voltages ranging from under —500 to — 2000 V, most being between —500 and —1500 V. The voltages are 234 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY SPECTRAL BAND- WIDTH, % WAVELENGTH, I FIG. 2 — Spectra of nickel hollow cathode lamp for different spectral bandwidths of mono- chromator. 235 S ON ATOMIC ABSORPTION SPECTROMETRY 1.6 r- SPECTRAL BAND- WIDTH, I 0.3 0.9 1.8 3.6 10 40 50 20 30 Ni, fig/ml FIG. 3 — Effect of spectral bandwidth on absorbance of nickel at 2320 A negative with respect to ground, and the positive terminal is usually at ground potential. The power supply for these tubes should provide outputs of up to — 2000 V at several milliamperes of current regulated to within 0.001 percent against line voltage and load current changes. Also, it 236 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY should provide a stability of better than 0.01 percent over periods of hours with a-c ripple components of less than 1 mV. The signal produced by the multiplier phototube is usually too weak to drive indicators and requires amplification. It is essential for the optimum signal-to-noise ratio (S/N) that the amplifier have high gain stability and low noise. The principles, type, and operation of amplifiers are discussed in detail by Veillon [31]. For readout devices, meters and recorders have been popular. Digital readout devices are gaining in popularity and may be considered essential for highly precise work. Advantages are that operator bias in making readings is eliminated, and since the digital device employs a decimal-to- binary or BCD converter on the output, the signal can be fed to a printer or tape punch for subsequent computation on a computer. Preliminary Instrumental Operations After the analyst has a basic understanding of the components of his instrument, he is ready to perform the analysis. Prior to the actual determination, there are several basic operations. The primary light source must be "turned on" with optimum operating current and allowed to "warm up." The time required for the HCL to stabilize will vary with different lamps, but, as a general rule, the operator should allow 15 min. Next, a resonance line of the analyte is selected. Usually the manufacturer of the instrument will provide a listing of the most sensitive analytical lines. Then the wavelength dial of the monochromator is set to that wavelength and adjusted until maximum intensity is obtained on the readout device. For this operation the prior discussion on slit widths should be considered for optimum operating conditions. The selection of oxidant-fuel and burner is made if a flame system is being used. There are several gas combinations available (see Table 3), but TABLE 3 — Temperatures and burning velocities of common premixed flames. Burning Velocity, Temperature, Gas Mixture cm/s °C Air-methane 70 1875 to 1950 Air-propane 82 1925 Air-acetylene 160 2125 to 2400 Air-hydrogen 320 2050 to 2115 Nitrous oxide- icetylene 160 to 680 2600 to 2955 greater than 90 percent of all AAS measurements are made using an air-acetylene or nitrous oxide-acetylene flame. Selecting the optimum conditions for a given analysis is somewhat difficult, and usually one must 237 RAINS ON ATOMIC ABSORPTION SPECTROMETRY resort to experimentation with various gases and flow ratios or consult the literature for data on a specific problem. Sample Preparation The samples suitable for analysis by AAS cover a wide range of materials and concentrations. The analyte may be present as a trace or as a major constituent. The amount and nature of sample preparation required will depend on a number of variables, such as the type of sample, the elements to be determined, and the precision and accuracy desired. Therefore, no general method can be described for the preparation of all samples. Solution samples are nebulized directly into the flame and the measurement taken. Solid samples are dissolved in hydrochloric, nitric, or perchloric acids. Sulfuric acid may be substituted for perchloric acid with certain samples; however, sulfates are undesirable and should be avoided, if possible. For the more refractory materials, a fusion with lithium metaborate [32] or with a high pressure "bomb" vessel [33] may be required. Fluxes for fusions must be completely free of trace inorganic ions. They may also add high concentrations of solids to the solution which may result in clogging of the burner. The advantage of a high pressure vessel is that the sample is dissolved in mineral acids which later can be removed by fuming. However, these devices have a tendency to leak, especially after long periods of use, and some of the volatile elements may be lost. After the sample has been dissolved, a separation or preconcentration may be required. As a rule, interferences can be corrected without a separation, but preconcentration steps may be required when the sensitivity of the analyte is insufficient for the sample concentration. Preconcentration is accomplished by liquid -liquid extraction, evaporation, ion-exchange, or coprecipitation. These methods are discussed in detail by Ellis [34]. Calibration and Data Evaluation Before discussing methods used for preparing a standard curve for a particular element, we should consider some factors which may offset the accuracy of the results. Impurities in the reagents can decrease the practical detection limit or if contamination or impurities occur at random concentrations, correction is difficult and accuracy may falter. Therefore, high-purity reagents and a clean environment are essential for the highest precision and accuracy in AAS. Also, cleanliness of glassware is of the utmost importance in trace analysis. Cleaning of glassware with dichromate- sulfuric acid cleaning solution is a common means of obtaining grease-free walls; however, this glassware should not be used for chromium analyses. The boiling of glassware in aqua regia followed by several rinses with ultra 238 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY pure distilled water will result in essentially a metal-free surface. In spite of precautions taken to prevent contamination, it is very difficult to completely eliminate all traces of impurities. Therefore, it is essential that sample and reagent blanks be processed along with the sample. Preparation of Standard Solution Standard stock solutions should be prepared from high-purity metals or salts of known stoichiometry. Michaelis [35] has compiled a comprehen- sive listing of suppliers of high puritv materials for virtually every element. A standard stock solution should contain at least 1.000 mg/ml of the analyte in an appropriate acid to prevent hydrolysis. Working solutions should be prepared daily from this stock solution. Procedures for the preparation and storage of standard solutions are given by Dean and Rains [36]. Preparation of Calibration Curve There are several ways to prepare a standard calibration curve. The simplest method is to prepare a series of standard aqueous solutions of the analyte to cover the desired concentration range and to measure their absorbance at an appropriate wavelength. Failure to select the appropriate concentration range could lead to a nonlinear calibration curve. Solutions of the unknown sample are then measured, and their concentrations are determined from a calibration graph. This is often the least reliable way because of matrix or inter-element effects. Therefore, possible interferen- ces must be considered in the preparation of a calibration curve. Interferences Interference effects which occur in flames and nonflame systems can be classified as physical, chemical, inter-element, and ionization. Physical interferences occur when components of the sample alter their physical properties such as viscosity, surface tension, and density. To minimize these effects it is essential that the physical properties of standards and samples be similar. The standards should contain about the same acid concentration as the samples. As a general rule, the acid con- centration should not exceed 5 percent, but the preferred acid concentration for most elements is 0.1 to 1 percent. High salt concentrations would change the viscosity and density of the sample. To compensate for these effects, it is advisable, whenever possible, to add an equal concentration of the major ions of the sample to the standards. Chemical interferences originate during the desolvation of the analyte in the flame or nonflame cell forming a compound which volatilizes at a different rate than the standard solutions. This type of interferent may be 239 RAINS ON ATOMIC ABSORPTION SPECTROMETRY overcome with a releasing agent or protective chelate added to sample and standards. Lanthanum or strontium are the most frequently used, although calcium, magnesium, and rare earth elements have been used. It is essential that the releasing agent be free of the analyte or high blanks will result. Inter-element interferences are not well understood. For example, the absorption of nickel can be suppressed or enhanced by iron depending upon the iron concentration and instrumental conditions. Also, titanium has been observed to enhance the absorption of aluminum and vanadium, and aluminum will enhance the absorption of vanadium. To prevent this inter-element effect, the interfering element should be added to standard solutions in approximately the same concentration as the sample. Ionization interference occurs when the analyte is ionized in the flame. In the cooler flames, ionization is not a serious problem. With high temperature flames such as nitrous oxide-acetylene, it is a serious problem, especially for those elements which have a low ionization potential. If ionization cannot be avoided by the choice of flame, an easily ionizable element is added to standard and samples to suppress ioniza- tion, As a rule, 500 to 1000 /ug/ml of potassium or cesium is sufficient to suppress the ionization of most analytes. Other elements have been used but a higher concentration may be required. Accuracy of Method After optimization of all instrumental parameters and consideration of possible interferences, the analyst must evaluate the reliability of the data or accuracy of the method. Grant [37] has defined accuracy as "a quantitative measure of the variability associated with the relation of analytical results with what is assumed to be the true value." The obvious way of checking the accuracy of an AAS method is to compare the analytical results with those obtained by an "accepted method." This must be done with discretion as an "accepted method" may have a bias. Another approach is to evaluate the method using Standard Reference Materials which have been certified by two or more independent analytical techniques. If similar material of known composition is not available, a check of the accuracy of the method can be made by the following techniques. (1) The dilution method is useful for revealing interferences. In this method the aliquots or dilution should cover as wide a range of concentrations as practical. (2) The accuracy can be checked by recovery experiments. It is advisable to add approximately the same concentration of analyte as was originally found in the test sample. A complete recovery of the added test portion, however, does not always assure an accurate value. (3) The method of standard addition can be used to check for possible chemical interferences. (4) The presence of light scatter or molecular absorption should be checked using a nonabsorbing line or a continuum source. (5) 240 SOME FUNDAMENTALS OF ANALYTICAL CHEMISTRY And, finally, if in doubt, the analyte can be separated from the matrix by liquid -liquid extraction, coprecipitation or ion-exchange techniques. While this final approach will add considerable time to the analysis, it may be essential to prove the accuracy of the method. References [/] Butler, L. R. P. and Brink, J. A. in Flane Emission and Atomic Absorption Spectrom- etry, Vol. 2, Dean, J. A. and Rains, T. C, eds., Dekker, New York, 1971, pp. 21-56. [2] Batchelor, E. H., Spectrovision, Vol. 25, 1971, pp. 7-15. [3] Jackson, D. A., Proceedings of the Royal Society (London), Vol. A121, 1928, pp. 432- 447. [4] Meggers, W. F. and Westfall, F. O., Journal of Research of the National Bureau of Standards, Vol. 44, 1950, pp. 447-455. [5] Metters, W, F, and Stanley, R. W., Journal of Research of the National Bureau of Standards. Vol. 61, 1958, pp. 95-103. [6] Fehsenfeld, F. C, Evenson, K. M., and Broida, H. P., Review of Scientific Instru- ments, Vol. 36, 1965, pp. 294-298. [7] Tomkin, F. S. and Fred, M., Journal of the Optical Society of America, Vol. 47, 1957, pp. 1087-1091. [8] Zacha, K. E., Bratzel, M. P., Winefordner, J. D., and Mansfield, J. M., Jr., Analytical Chemistry, Vol. 40, 1968, pp. 1733-1736. [9] Dagnall, R. M. and West, T. S., Applied Optics, Vol. 7, 1968, pp. 1287-1294. [10] Menis, O., ed., "Analytical Coordination Chemistry Section: Summary of Activities, July 1968 to June 1969," Technical Note 504, National Bureau of Standards, U.S. De- partment of Commerce, 1969, pp. 1-2. [//] Browner, R. F., Rietta, M. E., and Winefordner, J. D., "Improvement in EDL Radiant Flux and Stability by Precise Temperature Control," presented at Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, 6-10 March 1972, paper number 136. [12] Herrmann, R. in Flame Emission and Atomic Absorption Spectrometry, Vol. 2, Dean, J. A. and Rains, T. C, eds., Dekker, New York, 1971, pp. 57-93. [13] Walsh, A., Spectrochimica Acta, Vol. 7, 1955, pp. 108-117. [14] Delves, H. T., Analyst, Vol. 95, 1970, pp. 431-438. [15] Manning, D. C, Atomic Absorption Newsletter, Vol. 10, 1971, pp. 123-124. [16] Yamamoto, Y., Kumamaru, T., Hayashi, Y., and Tsujino, R., Analytical Letters, Vol. 5, 1972, pp. 419-424. [17] Gatehouse, B. M. and Walsh, A., Spectrochimica Acta. Vol. 16, 1960, pp. 602-604. [18] Dogan, M., Laqua, K., and Massmann, H., Spectrochimica Acta, Vol. 26B, 1971, pp. 631-649. [19] L'vov, B. V., Inzhenerno-Fizicheskii Zhurnal, Vol. 2, 1959, pp. 44-52. [20] Massmann, H., Spectrochimica Acta, Vol. 23B, 1968, pp. 215-226. [21] Woodriff, R. and Ramelow, G., Spectrochimica Acta, Vol. 23B, 1968, pp. 665-671. [22] West, T. S., I UP AC Atomic Absorption Spectroscopy, Dagnall, R. M. and Kirkbright, G. F., eds., Butterworth, London, 1970, pp. 99-126. [23] Schrenk, W. G. in Flame Emission and Atomic Absorption Spectrometry, Vol. 2, Dean, J. A. and Rains, T. C, eds., Dekker, New York, 1971, pp. 303-325. [24] Feldman, F. J., "A Non-flame Sampling Technique for Atomic Absorption with Auto- matic Background Correction," presented at Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, 1-6 March 1970, paper num- ber 49. [25] Poluektov, N. S., Vitkun, R. A., and Zelyukova, Yu. Y., Zhurnal Analiticheskoi Khimii, Vol. 19, 1964, pp. 937-942. [26] Hatch, W. R. and Ott, W. L., Analytical Chemistry, Vol. 40, 1968, pp. 2085-2087. [27] Manning, D. C, Atomic Absorption Newsletter, Vol. 9, 1970, pp. 97-99. [28] Rains, T. C. and Menis, O., Journal of the Association of Official Analytical Chemists, Vol. 55, 1972, pp. 1339-1344. 241 APPENDIX XIII REPRINTED BY PERMISSION OF ANALYTICAL CHEMISTRY Reprinted from ANALYTICAL CHEMISTRY, Vol. 42, Page 394, March 1970 Flame Emission Spectrometry with Repetitive Optical Scanning in the Derivative Mode W. Snelleman, 1 T. C. Rains, K. W. Yee, 2 H. D. Cook, 2 and O. Menis Analytical Chemistry Division, National Bureau of Standards, Washington, D. C. A flame emission spectrometer using a rapid repetitive scan of a narrow wavelength region has been de- veloped. By this method of wavelength scanning the second derivative of the output intensity is measured. The use of this approach to minimizing spectral inter- ference in matrices and the use of microsamples greatly enhance the potentialities for flame emission spectrometry, and minimize the need for a mono- chromator of high resolving power. A quartz plate, made to vibrate at 145 Hz, is mounted behind the entrance slit of the monochromator. The ac amplifier is synchronized with the oscillations of the quartz plate. When the amplifier is tuned to twice the fre- quency of vibration, the second derivative of the spectrum is obtained. This permits the measurement of weak line spectra nested in or on a broad band or continuum. It is demonstrated that spectral inter- ference due to CaOH bands and/or a continuum are minimized in the measurement of barium. The elim- ination of interferences from bands and flame struc- ture led to an improvement in detection limits of alkali and alkaline earth elements in the presence of many matrix ions. An analysis can be performed with 50 jul of solution which makes it applicable to biochem- ical and air pollution studies. The measurement of the radiant intensity of an atomic line which is located on a bandhead arising from the concomitant or flame gases is of concern to the flame emission spectros- copist. Frequently, an atomic line cannot be used because of interferences from some overlapping band structure. Two typical examples are the CaOH bands (5430-6220 A), which interfere with the Na 5890 A and Ba 5536 A lines, and the MgOH bands (3600-4000^ A) which contribute to high back- ground for the Fe 3720 A and Ru 3727 A lines. Also, the OH band system, covering the region of 2800 to 4000 A, limits the measurement of the radiant intensity of many atomic lines in flame emission spectrometry. These interferences have dis- couraged many workers in the field of flame emission. How- ever, Buell (7) in using a high resolution monochromator has shown that many atomic lines in this region can be used. 1 Present address, Fysisch Laboratorium, Rijks-Universiteit Utrecht, The Netherlands. 2 Measurement Engineering Division, National Bureau of Stan- dards, Washington, D. C. (1) B. E. Buell in "Flame Emission and Atomic Absorption Spec- trometery," J. A. Dean and T. C. Rains, Eds., Dekker, New York, N. Y., 1969. A technique which permits the measurement of weak line spectra nested in or on a broad band or a continuum resulted in an improvement in detection limits. In addition, micro- liter samples can conveniently be taken for analyses. The newly designed optical system permits a rapid repetitive scan of a narrow wavelength region. This scan is synchronized with the ac amplifier. With the ac amplifier tuned to the same frequency as the vibrating quartz plate, the signal pro- duced is the first derivative of the emission intensity with re- spect to wavelength. At twice the frequency, the second derivative is obtained. This mode of operation permits the measurement of weak line spectra without the interference from background radiation or broad band spectra. When operating at high frequencies (145 Hz), the apparent signal-to- noise ratio is improved by eliminating the low frequency flicker noise of the flame. This technique was applied to the determination of lithium in 50 /d of solution containing high concentrations of sodium, and to the determination of barium in a calcium matrix. Derivative spectroscopy has been described by several authors (2-4). Giese and French (5) used it to detect low in- tensity bands by measuring the first derivative of the trans- mission curve with respect to wavelength. Under certain conditions, a recording of the first or second derivative of an absorption spectrum can give increased resolution over a normal spectrum. In magnetic resonance spectroscopy the first derivative and occasionally the second derivative are re- corded to increase resolution of the spectrum (6). There are various ways for obtaining the derivative of ab- sorption lines or bands with respect to wavelength. Balslev (7) vibrated the exit slit of his spectrometer and synchronously detected the signal at the vibrating frequency while Gilgore et al. used optical wobblers (5). Snelleman (9) was able to in- (2) A. Perregaux and G. Ascarelli, Appl. Opt., 7, 2031 (1968). (3) F. R. Stauffer and H. Sakai, ibid., 7, 61 (1968). (4) E. C. Olson and C. D. Alway, Anal. Chem., 32, 370 (1960). (5) A. T. Giese and C. S. French, Appl. Spectrosc, 9, 78 (1955). (6) C. H. Townes and A. L. Schwawlow, "Microwave Spectros- copy," McGraw-Hill, New York, N. Y., 1955, Chapters Hand 17. (7) I. Balslev, Phys. Rev., 143, 636 (1966). (8) A. Gilgore, P. J. Stoller, and A. Fowler, Rev. Sci. lustrum., 38, 1535(1967). (9) W. Snelleman, Spectrochem. Acta, 23B, 403 (1968). 242 Figure 1. Displacement of an oblique ray laterally and parallel to its initial direction crease the sensitivity of the measurement of copper by atomic absorption using a continuum source and optically scanning part of the continuum. It was accomplished by placing in the optical path a vibrating mirror which was mounted on the movement of a milliammeter. While this arrangement clearly demonstrated the feasibility of optical scanning in atomic ab- sorption, it had certain limitations because of alignment difficulties. The current investigation led to the development of a system utilizing an oscillating quartz plate which pro- vided a stable and easy to operate apparatus. Also, it was demonstrated that this technique had special applications in flame emission spectrometry. EXPERIMENTAL Apparatus. General Instrumentation. The instru- mental system used in this study consists of a 0.75m Czerny- Turner mount monochromator, f/6.5 aperture, curved slits> exchangeable gratings blazed for 3000 and 7500 A with 1180 grooves/mm; an oxygen-hydrogen flame with a total con- sumption burner and an air-acetylene flame with a premixed nebulizer burner; gas flow meters and pressure regulators; an end-on multiplier phototube, EMI 9558AQ, with 44-mm diameter photocathode and S-20 spectral response; a to 2100 V multiplier phototube power supply, 0-30 mA, 0.001% regulation (line and load), ±0.25% accuracy; a phase sensitive lock-in amplifier with full-scale sensitivity ranges of 100 nV to 500 mV, and a 10-mV strip-chart recorder, 0.2 sec full-scale response time. Quartz Plate. A quartz plate 12 X 14 X 2 mm serving as the displacement element was mounted on a rotating me- chanical oscillator with a variable amplitude of up to 6° peak to peak. The light beam, in passing through the plate (as shown in Figure 1), will be displaced laterally (perpen- dicular to the axis) and parallel to its initial direction when measured normal to the ray, and is given by the following relationship: A _Lray = t • cos a (tan a — tan a') ~ [ta{n - l)/»] where t = the thickness of the plate, a = radians in air, a' = radians in the plate, and n = index of refraction of the material. The image shift is approximately proportional to the thickness of the quartz plate and the angle of rotation. In this experiment with a peak to peak amplitude of 5°, the perpendicular shift was 132 n using a mca „ = 0° which cor- responded to about 1.5 A of the spectrum. A larger dis- placement can be obtained by using a thicker quartz plate and by increasing the angle of rotation. However, to avoid Figure 2. Displacement of an image along the optical axis HOLE K TOP DISC V,jDIA. 4 SLOTS DO'APART^ 0010 ■ WIDE X^DEEP ALHIC0 X PERJUKENT MAGNET- ^x^'x'/is '^X^'CUT IN EACH COIL CORE 4 LAYERS SILICON STEEL 0.014 * THICK . . STAND OFF- %DIAXTL BRASS CLEARACE HOLE FOR NO. 2 -56 SCREW 4STRIPS, OOIO'T. PHOSPHOR BRONZE SHIN STOCK 5 /64«l 5 /|5 BRASS DISC V 1 *. '|6 T 4 SLOTS 90* APART 0.010" X 5 / H DRILL #42 HOLE IN CENTER BRASS ROO Figure 3. Mechanical oscillator for optical scan an off-set of the wavelength calibration of the monochro- mator and light losses due to increased reflection at large tilt angles, an angle of a = 0° was selected. The ray in addition to being displaced laterally and parallel to its initial direction will be shifted longitudinally. The displacement of the image from the exit slit by the quartz plate, as shown in Figure 2, is given by the following relation- ship (10): A | U, s = t (n - l)/n ~ (//3 for common crown glass quartz) Because the image was defocused at the exit slit due to the change in optical path length, the quartz plate was ground to a slight convex curvature corresponding to a focal length of 12 m. (10) W. Meuller-Herget, Philips Laboratories, Briarcliff Manor, N. Y., personal communication, 1969. 243 I EXIT SLIT Figure 4. Location of quartz plate within the optical system Figure 5. Electrical circuit for mechanical oscillator Design of High Frequency Wavelength Oscillator. The mechanical oscillator shown in Figure 3 consists of two brass disks connected by four narrow phosphor bronze strips cut from 0.010" thick shim stock. The strips are soldered at each end into slots cut radially 90° apart at the edges of the disks. One disk is rigidly attached to the end of a rod which provides a large mass for an inertial reference. The other disk will oscillate in a rotational mode about the axis. The spring structure allows a large angular amplitude while retaining a high stiffness for other vibrational modes. The quartz plate is mounted on edge in a channel cut in this disk. The frequency of oscillation is determined by the stiffness of the spring strips and the moment of inertia of the top disk, including any attachments. The Q of the device is approximately 300. The torque required to overcome the losses is produced by the force between current carrying coils and the ends of a permanent-magnet bar cemented to the underside of the moving disk. The angle between the magnet and the drive core is 45°. The actual core is in the shape of a cross. The two halves of the drive coil are wound on opposite arms of the cross, and an identical pick-up coil is wound on the remaining two arms. The signal induced in the pick-up coil is amplified and fed back to the drive coil. The system oscillates at a frequency determined by the mechanical resonance. Operation at the mechanical reso- nance is necessary to obtain sufficient angular displacement with a reasonable driving power, and to ensure a constant phase relation between the oscillator motion and its driving voltage. The lower disk holds the bottom ends of the spring strips rigidly. This construction simplified the assembly of the delicate parts. To assemble the springs without strain, a jig was used to position the disks the desired distance apart and perpendicular to the axis. The strips were soldered into the slots using a minimum amount of heat and solder. The core was mounted on a standoff which also served as a nut for attaching the bottom disk to the rod. The magnet was cut to size with a thin abrasive-filled cutting disk. The slots in each end provide clearance for the spring strips. The magnet and quartz plate must be assembled after the coil Figure 6. Flame emission spectrum of 10 /jg/ml of barium (A) Mechanical disk chopper, (B) Optical scan-first derivative,! (C) Optical scan-second derivative assembly has been installed. The quartz plate is placed within the optical path as shown in Figure 4. Associated Electronic Controls. The electrical circuit for the oscillator, shown in Figure 5, consists of the magnet, pick-up coil, impedance step-up transformer 77, limiting amplifier circuit of ARI, variable gain amplifier circuit of AR2, impedance stepdown transformer T2, and the drive coil. The drive and pick-up coils are oriented to eliminate any direct magnetic coupling. The varying magnetic field of the moving permanent magnet induces a voltage in the pick-up coil. This voltage is applied to the input of amplifier ARI through transformer 77. Amplifier ARI is connected as a gain of 101 amplifier with output voltage limiting. For all useful oscillation amplitudes, the peak output voltage is approximately ±0.5 V. The limited voltage is amplified by the adjustable gain circuit of AR2. Transistors Ql and Q2 form a complementary-emitter-follower output stage to in- crease the available output current. At maximum gain, the output voltage is nearly ±10 V peak. Reducing the gain lowers the output voltage and the oscillation amplitude. The output voltage is applied to the drive coil through trans- former T2. The 100 il resistor in series with the primary increases the dc resistance seen by the amplifier. The offset voltages of ARI and AR2 should be adjusted for less than ± 100 mV at the emitter of Ql. An auxiliary winding of T2 provides an output equal in amplitude and 180° out of phase with the voltage from Ql and Q2. These two voltages are half-wave rectified and are added to provide sufficient signal at twice the oscillation frequency to drive the ac-coupled tuned amplifier in the reference channel of the lock-in ampli- fier used. A reference output at the frequency of oscillation is obtained from the output of ARI. The capacitors across the feedback resistors of ARI and AR2 are selected for zero phase shift around the loop at the mechanical resonant fre- quency. The loop may be opened between the two amplifiers and AR2 driven by a frequency-stable audio oscillator at the resonant frequency. The feedback capacitors should then be adjusted for zero phase shift between the input voltage and the output voltage from ARI. Alternatively, with the system oscillating, the feedback capacitors may be adjusted for maximum voltage at the secondary of Tl. Reagents. All solutions used were prepared from high purity metals or salts. 244 Operating Characteristics. The spectrum of the resonance line of barium, obtained in an oxygen-hydrogen flame with a total consumption burner, is shown in Figure 6 A. The wavelength region of the barium line at 5536 A was scanned in the normal mode of a rate of 5 A/min with a monochro- mator having a spectral band width of 0.6 A. The light signal was chopped at 87 Hz using a mechanical disk chopper. The mechanical disk chopper was then removed from the optical system and the quartz plate was placed behind the entrance slit within the optical path (Figure 4). By means of the drive coil, the quartz plate was oscillated at 145 Hz. The synchronous lock-in amplifier was tuned to this fre- quency. As shown in Figure 6fl, a wavelength scan of the 5536 A barium resonance line was repeated. The resulting spectrum was the first derivative with the peak corresponding to the zero crossing of the X axis. While the first derivative has been used in spectroscopy (5) for identifying overlapping band structures, a scan of the wavelength region is required and offers no advantage in flame emission over the mechanical disk chopper. The experiment was repeated, with the lock-in amplifier turned to twice the frequency of the vibrating quartz plate. The signal (Figure 6C) which is the second derivative with respect to wavelength is proportional to the concen- tration of the analyte. The effect of calcium on barium was investigated to evaluate the use of the optical scan in the second derivative mode for the measurement of the intensity of a line nested on a broad continuum. Operating in the normal mode with a mechanical disk chopper the wavelength region of 5500 to 5560 A was scanned at a rate of 25 A/min with 4 /zg/ml of barium in the presence of 20 Mg/ml of calcium in an oxygen-hydrogen flame. The spectrum is shown in Figure 1A. The spectrum produced with the vibrating quartz plate within the optical path is shown in Figure IB. The re- corded data shown in Figure 8 were obtained by setting on the peak intensity of the barium line and repetitively scanning a narrow wavelength region. By measuring the emission intensity in the second derivative mode, the signal is propor- tional to the analyte concentration. However, to obtain the maximum signal, the amplitude of the optical scan is ad- justed with 5K fi variable resistor, shown in Figure 5. RESULTS AND DISCUSSION In flame emission spectrometry, the barium resonance line at 5536 A suffers serious spectral interferences from calcium, magnesium, and sodium (11, 12). Also in atomic absorption, the CaOH band system with its peak at 5540 A will interfere in the determination of barium in an air-acetylene flame (13, 14). As shown in Figure 1A, barium can be determined in the presence of calcium by flame emission spectrometry provided the calcium to barium ratio does not exceed 10:1. For this purpose the wavelength region of 5530 to 5540 A must be scanned and the proper background correction applied. By operating with a smaller spectral band width, the calcium band structure is reduced; however, to obtain the desired sensitivity for barium, a spectral band width of 1.0 to 1.5 A is usually necessary. Under these conditions the high back- ground emission from CaOH band structure becomes a limit- ing factor and zero offset is required to shift the spectrum on the recorder scale. Also, in making a wavelength scan of this region, two to five milliliters of solution are required. (11) J. A. Dean, "Flame Photometry," McGraw-Hill, New York, N. Y., 1960, p 211. (12) J. A. Dean, J. C. Burger, T. C. Rains, and H. E. Zittel, Anal. Chem., 33, 1722 (1961). (13) S. R. Koirtyohann and E. E. Pickett, ibid., 38, 585 (1966). (14) G. K. Billings, Ph.D. Thesis, Rice University, Houston, Texas, 1963. 5520 5540 WAVELENGTH, A Figure 7. Flame emission spectrum of 4 jug/ml of barium in the presence of 20 /ig/ml of calcium in an oxygen-hydrogen flame with (A) a mechanical disk chopper and (B) optical scan and derivative mode Ca blank Figure 8. Effect of calcium on barium in flame emission spec- trometry using optical scanning and derivative mode With the optical scan and taking the second derivative, the high background emission due to the CaOH band structure is eliminated as shown in Figure IB. The intensity of the barium line is measured from the dotted base line to the peak height, and is approximately two-thirds the height obtained with the disk chopper. The continuum of unresolved band structure with a nomi- nally zero curvature which lies under the analytical line is not detected. The amount of band structure or continuum which can be tolerated with the optical scanning technique will de- pend upon the optical system and the photodetector. The major limiting factors are the cathode loading and fatigue of photodetector. Under the optical conditions used the radiant intensity of analytical lines that were 0.1 % of the relative in- tensity of broad band structure have been measured without interference. However, narrow band peaks still appear as a slight deviation from the background. The unique advantage of this technique is shown in the study of the effect of various concentrations of calcium on the 245 Table I. Detection Limits by Flame Emission Spectrometry with Optical Scanning in the Presence of Matrix Ions Burner-diffusion Oxidant-fuel-oxygen-hydrogen Sample size-0.25 ml ng Matrix, element Detection limit," ng/ml 1 mg/ml Li K Ca Li 0.2 3.0 Na 0.02 0.08 0.5 K 0.02 3.0 Rb 0.02 1.0" 3.0 Mg 0.02 0.1 Ca 0.02 0.03 Fe 0.02 0.2 3.0 10"' grams " Detection limit is defined as 2a of background. 6 Rubidium free of potassium not available. ' Magnesium free of calcium not available. measurement of the barium emission line. As shown in Figure 8, by setting on the resonance line of barium and measuring the peak height of the second derivative with re- spect to wavelength, barium was determined in the pres- ence of a 100-fold concentration of calcium with an air- acetylene flame. By scanning a narrow wavelength region of one A an appropriate background correction is made. While in this case the peak height measurement was obtained with 2 ml of solution and a time constant of 1 sec, the analysis can be performed with a fraction of this sample. With a higher temperature flame, such as oxygen-hydrogen or nitrous oxide-acetylene with an appropriate burner, the calcium to barium ratio can be extended to 1000:1. A higher weight ratio can be tolerated in the presence of magnesium or sodium. The correction for background even in the presence of 1000 Aig of calcium is easily applied, and the need for extensive zero offset in the electrical circuit is eliminated. The detection limits determined for lithium, potassium, and calcium in the presence of 1000 /zg/ml of various matrix elements are given in Table I. The continuum or unresolved band structure produced by the matrix element did not show a shift in the base line. The detection limit of potassium in rubidium could not be extended below 1 ng because potassium- free rubidium was not available. The detection limits which are generally reported in the literature are usually based on pure water solution. These values can be very misleading when applied to samples con- taining other cations. Generally, for a complex matrix the optical scan led to an improvement in the detection limits of an analyte of 1 to 2 orders of magnitude over the normal wave- length scan with a mechanical chopper. However, in a pure (15) V. A. Fassel and D. W. Golightly, Anal. Chem., 39, 466 (1967). (16) E. E. Pickett and S. R. Koirtyohann, Spectrochim. Acta, 23B, 235 (1968). uu u U Li, ^g'rnl Figure 9. Lithium calibration results of flame emission spectrometry using optical scanning and derivative mode water solution and with an oxygen-hydrogen flame, the de- tection limits of the two methods were identical and cor- responded to the values reported in the literature (75, 76). Another unique feature of the technique is the amount of sample required for analysis. Normally with the mechanical disk chopper, three to five milliliters of solution are required for a scan of the wavelength region of interest and for the base line recording for measuring the peak height. By setting on the peak of a line with the optical scan and measuring the second derivative of the output signal, a fraction of a milliliter of solution is only required. In this manner, with a total consumption burner and a small bore capillary, lithium was determined in a sodium matrix. The data in Figure 9 show replicate determinations of lithium in 50 \A of solution. The precision in this method was 2% of the full scale. In this case the response of the electronics was the limiting factor in determining the volume of the solution needed. With a faster electronic detection system or an integrating system, this could be reduced further. In flame emission spectrometry this technique, in many cases, will eliminate the need for a high resolution spectrom- eter to minimize spectral interferences. Because the optical scan is applicable to microliter samples, it should be of great assistance in the analysis of biochemical materials and air pollution particulates. Also, it is being extended to atomic absorption spectrometry with a continuum light source. Received for review September 15, 1969. Accepted January 9, 1970. 246 APPENDIX XIV REPRINTED BY PERMISSION OF MARCEL DEKKER, INC. 13 Standard Solutions for Flame Spectrometry John A. Dean DEPARTMENT OF CHEMISTRY UNIVERSITY OF TENNESSEE KNOXVILLE, TENNESSEE and Theodore C. Rains ANALYTICAL CHEMISTRY DIVISION NATIONAL BUREAU OF STANDARDS WASHINGTON, D.C. I. Introduction ...... A. Sources of Standard Reference Materials . B. Types of Standard Reference Materials Available II. Aqueous Stock Solutions .... A. Relation of Concentration Units III. Specialized Stock Solutions .... IV. Standards for Trace Elements in Petroleum Products Appendix 1 : Standard Stock Solutions . Appendix 2 : Special Aqueous Stock Solutions Appendix 3: Standard Reference Materials in Lubricating Oil References 327 328 328 330 331 331 331 332 336 337 338 I. Introduction In all evaluation methods one must have available standard samples and high-purity materials for each element being determined, often in the same matrix as the sample. Quite often several standards must be prepared for a given element, one for each matrix or system that contains the test element. To assist the individual chemist who must prepare these standard solutions, this chapter will outline the preparation of various individual stock solutions and suggest sources where high-purity materials may be obtained or from whom standard samples may be purchased. An excellent report on available standard reference materials and high-purity materials has been compiled by Michaelis (7). Copyright © 1971 by Marcel Dekker, Inc. NO PART of this work may be reproduced or utilized in any form by any means, electronic or mechanical, including xerography, photocopying, microfilm, and recording, or by any information storage and retrieval system, without the written permission of the publisher. 247 JOHN A. DEAN AND THEODORE C. RAINS A. Sources of Standard Reference Materials The National Bureau of Standards (U.S.) offers a wide selection of standard alloys, steels, ores, glasses, and miscellaneous types of materials in the form of chips, turnings, powders, or granules. Full information concerning the values that are certified, the weights, and prices is contained in Ref. (2). The Bureau of Analysed Samples, Ltd. (Newham Hall, Middlesborough, Yorkshire, England) also offers a wide selection of standard samples. Many of these standards may be obtained through the Jarrell-Ash Division of Fisher Scientific Company (590 Lincoln Street, Waltham, Mass. 02154) or from Spex Industries, Inc. (3880 Park Avenue, Metuchen, N.J. 08841). A series of 20 ceramic powders are supplied by Applied Research Laboratories (9545 Wentworth Street, Sunland, Calif. 91040) for the analysis of barium, calcium, magnesium, and strontium titanates, and titanium dioxide. The standards have a variation of composition to cover the range of impurity constituents normally found in ceramic materials. In addition to the various standard samples, reference samples are also available. Generally compositions are less accurately known. Smith and Underwood (1023 Troy Court, Troy, Mich. 48084) supply over 1000 different analyzed materials, primarily for student use in universities and colleges. B. Types of Standard Reference Materials Available Three aluminum-base alloys prepared in the form of turnings are available from the National Bureau of Standards (NBS), and seven chip samples are issued by the Bureau of Analysed Samples (BAS). These samples possess certificates of analysis for Cu, Fe, Mg, Mn, Pb, Si, Ti, and Zn in all cases; individual samples may contain also Cr, Ga, Ni, Sb, Sn, and V. Two cobalt-base alloys are issued by NBS and contain Ni, Mn, Si, Fe, Cr, Mo, W, Nb, and Cu. Nine copper-base alloys are available from NBS and five from BAS, all in chip form. In addition to Cu, Zn, Sn, Pb, Ni, and Fe, certain standards also contain Al, As, Ag, Co, Mn, Sb, Si, S, P, and Mg. Six standards in the form of turnings from high-purity copper are available from Bundensanstalt fiir Materialpriifung (Berlin-Dahlem, Unter den 248 13 STANDARD SOLUTIONS FOR FLAME SPECTROMETRY Eichen 87, Berlin, Germany). These standards contain less than 0.07% of Pb, Sn, Bi, As, Sb, Fe, and Ni. Over 120 standards for iron-base alloys are available from NBS and BAS in the form of chips. These include cast iron, ingot iron, B.O.H. and A.O.H. steels, plain-carbon steels, mild steels, and alloy steels. The mild steels and alloy steels contain accurately determined traces of a number of residual elements (W, Co, Ti, As, Sn, Al, Zr, Pb, Nb, Ta, B, and Ag) in addition to the usual constituents: Mn, P, S, Si, Cu, Ni, Cr, V, and Mo. Two lead-base standards are issued by NBS and one by BAS; these contain Sn. Sb, Bi, Cu, As, and traces of Ag and Ni. One magnesium-base alloy is available from NBS and one from BAS. The former contains Al, Zn, Mn, Si, Cu, Pb, Fe, and Ni; the latter contains total rare earths, Zn, Zr, Cu plus traces of Mn, Ni, and Fe. The NBS issues three primary standards of nickel oxide (powder form) and four nickel-base alloys (Monel, nickel-base casting alloy, Ni-Cr alloy, Waspaloy). The BAS supplies one standard (Nimonec 90). One tin-base alloy is available from NBS and from BAS. The standard contains Sb, Cu, As, Pb and traces of Bi, Ag, Ni, and Fe. Only one zinc-base alloy is available from NBS. It contains Al, Cu, Mg, and traces of Fe, Mn, Pb, Ni, Sn, and Cd. One zirconium-base alloy, Zircaloy 2, is issued by NBS. It contains Sn, Fe, Cr, and traces of Ni, Mn, and Cu. Steel-making alloy standards from NBS and BAS include refined silicon, ferrovanadium, ferrochromium, spiegeleisen, ferrophosphorus, ferroboron, and calcium molybdate. A set of six cement standards is available from the NBS with these constituents certified: Si0 2 , A1 2 3 , Fe 2 3 , Ti0 2 , P 2 O s , CaO, SrO, MgO, S0 3 , Mn 2 3 , Na 2 and K 2 0. Miscellaneous standard samples issued by the NBS include: three alumina refractories, two silica refractories, one chrome refractory, a bauxite, a plastic clay, a glass sand, four glasses, and argillaceous lime- stone, a silica brick, a burned magnesite, titanium dioxide, and silicon carbide. These are supplemented by three basic slags, a silica brick, a firebrick, and a sillimanite from the BAS. Among ore samples available are six iron ores from the BAS. A phos- phate rock, a Mesabi iron ore, three lithium ores (Spodumene, Petalite, and Lepidolite), one manganese ore, two tin ores, and one zinc ore are available from the NBS. Except for the phosphate rock and Mesabi iron ore, the NBS ores are certified only for the component of interest. 249 JOHN A. DEAN AND THEODORE C. RAINS A syenite rock sample and a sulfide ore are also available (Department of Geological Sciences, McGill University, Montreal, P. Q., Canada). Four horticultural standards of apple, cherry, citrus, and peach leaves have been prepared (Horticultural Department, Michigan State University, East Lansing, Mich.). II. Aqueous Stock Solutions High-purity materials (99.9 % or greater) can be obtained from several sources: Alpha Metals, Inc. (60 Water Street, Jersey City, N.J.), Johnson Matthey & Co., Ltd. (Hatton Garden, London, E.C.I, England), Spex Industries, Inc. (Section I.A.), and various chemical supply houses, for example, Baker & Adamson, J. T. Baker & Co., Fisher Scientific Co., Mallinckrodt Chemical Works, and Merck and Co. A comprehensive listing of suppliers of high-purity materials for virtually every element is available (/). In selecting a suitable compound or material for use as a standard, several criteria must be met: (1) It must be available commercially in a high degree of purity and meet A.C.S. specifications or a similar "official" certification. Established methods should be available for confirming its purity (3). (2) The compound should not be hygroscopic, nor should it be efflorescent. It should be stable under the conditions specified for drying. (3) The composition of the selected material must be known; if not, the final stock solution will have to be standardized. (4) Metallic and other contaminants present in these standard materials must not be in amounts greater than that of the trace element to be determined in any particular sample, otherwise mixed standards could not be prepared. Procedures which provide standard stock solutions containing 1000 ^ag/ml of the element of interest are listed in Appendix 1. Precautions outlined in Chapters 10 and 1 1 (Section II. E) should be reviewed. Molarity of the reagents used is as follows: Reagent Molarity Reagent Molarity Acetic acid 17.4 HN0 3 (69%) 15.4 HC1(38%) 12.0 H 3 P0 4 (85%) 14.7 HClO 4 (70%) 11.6 H 2 S0 4 (94%) 17.6 HF(45%) 25.7 Ammonia (27 %) 14.3 250 13 STANDARD SOLUTIONS FOR FLAME SPECTROMETRY Many stock solutions should be slightly acidic to prevent hydrolysis. Unless stated otherwise, final dilution should be made to 1000 ml with deionized water — ordinary distilled water which has been passed through a mixed-bed ion-exchange resin such as Amberlite MB-3, then freshly boiled and cooled to room temperature. Solutions are stored in poly- ethylene containers unless a special containment material is specified. A crystal of thymol in each stock solution will inhibit bacterial growth. A. Relation of Concentration Units The interrelationship between various expressions for concentrations is: 1 jUg/ml = 1 mg/liter = 1 ppm (w/v) 1 meq/liter = (weight in grams per equivalent/ 1000) in 1 liter 1 mg atom per liter = (atomic weight in grams/1000) in 1 liter 1 jMg/ml = (1 meq/liter)(1000/weight in grams per equivalent) HI. Specialized Stock Solutions Solutions for clinical, agronomic, and certain other special situations are listed in Appendix 2. These solutions are prepared in the same manner as described in Section II. IV. Standards for Trace Elements in Petroleum Products Stable, oil-soluble compounds are available from the National Bureau of Standards (U.S.) as certified standards for 24 elements in petroleum products. Appendix 3 lists the compounds; all are solids. Directions for the preparation of individual compounds have been published (5) for those who may wish to prepare their own materials. Some of the com- pounds are fairly hygroscopic. As a precautionary measure, bottles con- taining the standard material should be stored, tightly closed, in desiccators containing P 2 O s , and that, before use, the required amount of a compound (plus a slight excess) should be dried over P 2 5 in an open container for the length of time recommended in Appendix 3, column 4. To prepare an oil solution containing one or several elements of interest, a solubilizing agent is prepared by placing 5 ml of 2-ethylhexanoic acid, 4 ml of 6-methyl-2,4-heptanedione, and 2 ml of xylene in a weighed 200-ml flask. The calculated weight of standard sample (Appendix 3, 251 JOHN A. DEAN AND THEODORE C. RAINS column 5) is carefully transferred to the flask. The suspension is gently heated on a hot plate, with swirling, until the salt has dissolved to a clear solution. The salts are added, one at a time, and completely dissolved before the next compound is added. After each addition, the mixture is gently heated, tilted, and rotated in order to wash down any particles adhering to the inside of the flask. After the last compound has been added and dissolved, 2 ml of bis(2-ethylhexyl)amine is added. The beaker is now washed with six 15-ml portions of lubricating oil, and these rinsings are all transferred to the flask. This last step is omitted if the sample is transferred directly to the flask from the sample container, and not weighed into an intermediate beaker. The oil solution is now allowed to cool to room temperature, and cool lubicating oil is added to the flask until the weight of the solution is 100 ± 0.5 g. The solution is reheated to 85°C, allowed to cool to room temperature, transferred to a storage bottle, and tightly stoppered. Compounds of silver and mercury are sensitive to prolonged heating and so, they should be added last to any mixture. For the other compounds, the order of addition is immaterial. Appendix 1: Standard Stock Solutions* Element Procedure Aluminum Dissolve 1.000 g Al wire in minimum amount of 2 M HCI; dilute to volume. Antimony Dissolve 1 .000 g Sb in (1) 10 ml HN0 3 plus 5 ml HC1, and dilute to volume when dissolution is complete; or (2) 18 ml HBr plus 2 ml . liquid Br 2 , when dissolution is complete add 10 ml HC10 4 , heat in a well-ventilated hood while swirling until white fumes appear and continue for several minutes to expel all HBr, then cool and dilute to volume. Arsenic Dissolve 1.3203 g of As 2 3 in 3 ml 8 M HC1 and dilute to volume; or treat the oxide with 2 g NaOH and 20 ml water, after dissolution dilute to 200 ml, neutralize with HCI (pH meter), and dilute to volume. Barium (1) Dissolve 1.7787 g BaCl 2 -2H 2 (fresh crystals) in water and dilute to volume. (2) Dissolve 1.516 g BaCl 2 (dried at 250°C for 2 hr) in water and dilute to volume. (3) Treat 1 .4367 g BaC0 3 with 300 ml water, slowly add 10 ml of HCI and, after the C0 2 is released by swirling, dilute to volume. * 1000 A'g/ml as the element in a final volume of 1 liter unless stated otherwise. 252 13 STANDARD SOLUTIONS FOR FLAME SPECTROMETRY Appendix 1: (Cont.) Element Procedure Beryllium Bismuth Boron Bromine Cadmium Calcium Cerium Cesium Chlorine Chromium Cobalt Copper Dysprosium Erbium Europium Fluorine Gadolinium Gallium Germanium' (1) Dissolve 19.655 g BeS0 4 -4H 2 in water, add 5 ml HC1 (or HN0 3 ), and dilute to volume. (2) Dissolve 1.000 g Be in 25 ml 2 M HC1, then dilute to volume. Dissolve 1.000 g Bi in 8 ml of 10 M HNO s , boil gently to expel brown fumes, and dilute to volume. Dissolve 5.720 g fresh crystals of H 3 B0 3 and dilute to volume. Dissolve 1.489 g KBr (or 1.288 g NaBr) in water and dilute to volume. (1) Dissolve 1.000 g Cd in 10 ml of 2 M HC1; dilute to volume. (2) Dissolve 2.282 g 3CdS0 4 -8H 2 in water; dilute to volume. Place 2.4973 g CaC0 3 in volumetric flask with 300 ml water, carefully add 10 ml HO, after C0 2 is released by swirling, dilute to volume. (1) Dissolve 4.515 g (NH 4 ) 4 Ce(S0 4 ) 4 -2H 2 in 500 ml water to which 30 ml H 2 S0 4 had been added, cool, and dilute to volume. Advisable to standardize against As 2 3 . (2) Dissolve 3.913 g (NH 4 ) 2 Ce(N0 3 ) 6 in 10 ml H 2 S0 4 , stir 2min, cautiously introduce 15 ml water and again stir 2 min. Repeat addition of water and stirring until all the salt has dissolved, then dilute to volume. Dissolve 1.267 g CsCl and dilute to volume. Standardize: Pipette 25 ml of final solution to Pt dish, add 1 drop H 2 S0 4 , evaporate to dryness, and heat to constant weight at >800°C. Cs (in ^g/ml) = (40)(0.734)(wt of residue) Dissolve 1.648 g NaCl and dilute to volume. (1) Dissolve 2.829 g K 2 Cr 2 7 in water and dilute to volume. (2) Dissolve 1.000 g Cr in 10 ml HC1, and dilute to volume. Dissolve 1.000 g Co in 10 ml of 2 M HC1, and dilute to volume. (1) Dissolve 3.929 g fresh crystals of CuS0 4 5H 2 0, and dilute to volume. (2) Dissolve 1.000 g Cu in 10 ml HO plus 5 ml water to which HNO3 (or 30% H 2 2 ) is added dropwise until dissolution is complete. Boil to expel oxides of nitrogen and chlorine, then dilute to volume. Dissolve 1.1477 g Dy 2 3 in 50 ml of 2 M HC1; dilute to volume. Dissolve 1.1436 g Er 2 3 in 50 ml of 2 M HCI; dilute to volume. Dissolve 1.1579 g Eu 2 3 in 50 ml of 2 M HCI; dilute to volume. Dissolve 2.210 g NaF in water and dilute to volume. Dissolve 1.152 g Gd 2 3 in 50 ml of 2 M HCI; dilute to volume. Dissolve 1.000 g Ga in 50 ml of 2 M HCI; dilute to volume. Dissolve 1.4408 g Ge0 2 with 50 g oxalic acid in 100 ml of water; dilute to volume. 253 JOHN A. DEAN aND THEODORE C. RAINS Appendix 1: (Cont.) Element Procedure Gold Hafnium Holmium Indium Iodine Iridium Iron Lanthanum Lead Lithium Lutetium Magnesium Manganese Mercury Molybdenum Neodymium Nickel Niobium Dissolve 1.000 g Au in 10 ml of hot HN0 3 by dropwise addition of HCI, boil to expel oxides of nitrogen and chlorine, and dilute to volume. Store in amber container away from light. Transfer 1 .000 g Hf to Pt dish, add 10 ml of 9 M H 2 S0 4 , and then slowly add HF dropwise until dissolution is complete. Dilute to volume with 10% H 2 S0 4 . Dissolve 1.1455 g Ho 2 3 in 50 ml of 2 M HCI; dilute to volume. Dissolve 1.000 g In in 50 ml of 2 M HCI; dilute to volume. Dissolve 1.308 g KI in water and dilute to volume. (1) Dissolve 2.465 g Na 3 IrCI 6 in water and dilute to volume. (2) Transfer 1.000 g Ir sponge to a glass tube, add 20 ml of HCI and 1 ml of HCIO,. Seal the tube and place in an oven at 300°C for 24 hr. Cool, break open the tube, transfer the solution to a volu- metric flask, and dilute to volume. Consult (4) and observe all safety precautions in opening the glass tube. Dissolve 1.000 g Fe wire in 20 ml of 5 M HCI; dilute to volume. Dissolve 1.1717 g La 2 3 (dried at 110°C) in 50 ml of 5 M HCI, and dilute to volume. (1) Dissolve 1.5985 g Pb(N0 3 ) 2 in water plus 10 ml HN0 3 , and dilute to volume. (2) Dissolve 1.000 g Pb in 10 ml HN0 3 , and dilute to volume. Dissolve a slurry of 5.3228 g Li 2 C0 3 in 300 ml of water by addition of 15 ml HCI, after release of C0 2 by swirling, dilute to volume. Dissolve 1.6079 g LuCI 3 in water and dilute to volume. Dissolve 1.000 g Mg in 50 ml of 1 M HCI and dilute to volume. (1) Dissolve 1.000 g Mn in 10 ml HCI plus 1 ml HN0 3 , and dilute to volume. (2) Dissolve 3.0764 g MnS0 4 H 2 (dried at 105°C for 4 hr) in water and dilute to volume. (3) Dissolve 1.5824 g Mn0 2 in 10 HCI in a good hood, evaporate to gentle dryness, dissolve residue in water and dilute to volume. Dissolve 1.000 g Hg in 10 ml of 5 M HN0 3 and dilute to volume. (1) Dissolve 2.0425 g (NH,) 2 Mo0 4 in water and dilute to volume. (2) Dissolve 1.5003 g Mo0 3 in 100 ml of 2 M ammonia, and dilute to volume. Dissolve 1.7373 g NdCI 3 in 100 ml 1 M HCI and dilute to volume. Dissolve 1.000 g Ni in 10 ml hot HN0 3 , cool, and dilute to volume. Transfer 1 .000 g Nb (or 1.4305 g Nb 2 O s ) to Pt dish, add 20 ml HF, and heat gently to complete dissolution. Cool, add 40 ml H 2 S0 4) and evaporate to fumes of S0 3 . Cool and dilute to volume with 8A/H 2 S0 4 . 254 13 STANDARD SOLUTIONS FOR FLAME SPECTROMETRY Appendix 1: (Cont.) Element Procedure Osmium Dissolve 1 .3360 g Os0 4 in water and dilute to 100 ml. Prepare only as needed as solution loses strength on standing unless Os is reduced by S0 2 and water is replaced by 100 ml 0.1 M HC1. Palladium Dissolve 1.000 g Pd in 10 ml of HN0 3 by dropwise addition of HC1 to hot solution; dilute to volume. Phosphorus Dissolve 4.260 g (NH 4 ) 2 HP04 in water and dilute to volume. Platinum Dissolve 1.000 g Pt in 40 ml of hot aqua regia, evaporate to incipient dryness, add 10 ml HCI and again evaporate to moist residue. Add 10 ml HCI and dilute to volume. Potassium Dissolve 1.9067 g KCI (or 2.8415 g KN0 3 ) in water and dilute to volume. Praseodymium Dissolve 1.1703 g Pr 2 3 in 50 ml of 2 M HCI; dilute to volume. Rhenium Dissolve 1.000 g Re in 10 ml of 8 M HN0 3 in an ice bath until initial reaction subsides, then dilute to volume. Rhodium Dissolve 1.000 g Rh by the sealed-tube method described under iridium (4). Rubidium Dissolve 1.4148 g RbCl in water. Standardize as described under cesium. Rb (in ,Mg/ml) = (40)(0.320)(wt of residue). Ruthenium Dissolve 1 .317 g Ru0 2 in 15 ml of HO; dilute to volume. Samarium Dissolve 1.1596 g Sm 2 3 in 50 ml of 2 M HCI; dilute to volume. Scandium Dissolve 1.5338 g Sc 2 3 in 50 ml of 2 M HCI; dilute to volume. Selenium Dissolve 1 .4050 g Se0 2 in water and dilute to volume or dissolve 1.000 g Se in 5 ml, of HN0 3 , then dilute to volume. Silicon Fuse 2.1393 g Si0 2 with 4.60 g Na 2 C0 3 , maintaining melt for 15 min in Pt crucible. Cool, dissolve in warm water, and dilute to volume. Solution contains also 2000 ^g/ml sodium. Silver (1) Dissolve 1.5748 g AgN0 3 in water and dilute to volume. (2) Dissolve 1.000 g Ag in 10 ml of HNO a ; dilute to volume. Store in amber glass container away from light. Sodium Dissolve 2.5421 g NaCl in water and dilute to volume. Strontium Dissolve a slurry of 1.6849 g SrC0 3 in 300 ml of water by careful addition of 10 ml of HCI, after release of C0 2 by swirling, dilute to volume. Sulfur Dissolve 4.122 g (NH 4 ) 2 S0 4 in water and dilute to volume. Tantalum Transfer 1.000 g Ta (or 1 .2210 g Ta 2 O s ) to Pt dish, add 20 ml of HF, and heat gently to complete the dissolution. Cool, add 40 ml of H 2 S0 4 and evaporate to heavy fumes of S0 3 . Cool and dilute to volume with 50% H 2 S0 4 . Tellurium (1) Dissolve 1 .2508 g Te0 2 in 10 ml of HCI ; dilute to volume. (2) Dissolve 1.000 g Te in 10 ml of warm HCI with dropwise addition of HN0 3 , then dilute to volume. 255 JOHN A. DEAN AND THEODORE C. RAINS Appendix 1: (Cont.) Element Procedure Terbium Dissolve 1 .6692 g of TbCl 3 in water, add 1 ml of HC1, and dilute to volume. Thallium Dissolve 1 .3034 g T1N0 3 in water and dilute to volume. Thorium Dissolve 2.3794 g Th(N0 3 ) 4 -4H 2 in water, add 5 ml HN0 3) and dilute to volume. Thulium Dissolve 1.142 g Tm 2 3 in 50 ml of 2 M HC1; dilute to volume. Tin Dissolve 1.000 g Sn in 15 ml of warm HC1; dilute to volume. Titanium Dissolve 1.000 g Ti in 10 ml of H 2 S0 4 with dropwise addition of HN0 3 ; dilute to volume with 5% H 2 S0 4 . Tungsten Dissolve 1 .7941 g of Na 2 W0 4 -2H 2 in water and dilute to volume. Uranium Dissolve 2.1095 g U0 2 (N0 3 ) 2 6H 2 (or 1.7734 g uranyl acetate dihydrate) in water and dilute to volume. Vanadium Dissolve 2.2963 g NH 4 VO a in 100 ml of water plus 10 ml of HN0 3 ; dilute to volume. Ytterbium Dissolve 1.6147 g YbCl 3 in water and dilute to volume. Yttrium Dissolve 1 .2692 g Y 2 3 in 50 ml of 2 M HC1 and dilute to volume. Zinc Dissolve 1 .000 g Zn in 10 ml of HC1 ; dilute to volume. Zirconium Dissolve 3.533 g ZrOCl 2 -8H 2 in 50 ml of 2 M HO, and dilute to volume. Solution should be standardized. Appendix 2: Special Aqueous Stock Solutions Concentration Procedure Element Barium Calcium Lithium 0.0625 N 100 meq per liter 200 meq per liter (as Li) 2000 fig/ml (as Li 2 Q) Dissolve 6.506 g BaCl 2 (dried at 250°C for 2 hr), dilute to volume. One ml is equivalent to 1 .00 mg of sulfate sulfur. Dissolve a slurry of 5.006 g CaC0 3 (dried at 105°C) in 300 ml of water by careful addition of 10 ml HC1; after C0 2 is re- leased, dilute to volume. Dissolve a slurry of 7.392 g Li 2 C0 3 in 300 ml of water by addition of 15 ml HC1 ; after release of C0 2 , dilute to volume. Dissolve 4.945 g Li 2 C0 3 as above. 256 13 STANDARD SOLUTIONS FOR FLAME SPECTROMETRY Appendix 2: (Cont.) Concentration Procedure Magnesium 100 meq per liter (as Mg) Manganese 1000 jMg/ml (as Mn 2 3 ) Potassium 100 meq per liter (asK) lOOO^g/ml (as K 2 0) Sodium 100 meq per liter (as Na) 1000,ug/ml (as Na 2 0) Strontium 1000 /«g/ml (as SrO) Mixture Alkali- 1000,ug/ml each manganese Na 2 0, K 2 0, Mn 2 solution Lime-acid 6300 jMg/ml CaO in solution 1:9 HC1 Sea water Chlorinity 38% Dissolve 1 .21 56 g Mg in 50 ml of 1 M HC1 ; dilute to volume. Dissolve 2.1412 g MnS0 4 H 2 (dried at 105°C for 4 hr) and dilute to volume. Dissolve 7.456 g K.C1 and dilute to volume. Dissolve 1 .5830 g KG and dilute to volume. Dissolve 5.846 g NaCl and dilute to volume. Dissolve 1 .886 g NaCl and dilute to volume. Dissolve a slurry of 1 .6848 g SrC0 3 as described for calcium. Dissolve in water 1.886 g NaCl, 1.583 g KC1, and 2.1412 g MnS0 4 H 2 (dried at 105°C for 4 hr) and dilute to volume. Add 100 ml HCI to 11.244 g CaC0 3 suspended in 300 ml of water ; after C0 2 is released, dilute to volume. Dissolve 47.0 g NaCl, 21.3 g MgCl 2 -6H 2 0, 17.8 g Na 2 SO 4 10H 2 O, 2.20 g CaCl 2 , 1.46 g KC1, 0.38 g NaHC0 3 , and 0.015 g SrCl 2 ; and dilute to volume. Appendix 3: Standard Reference Materials in Lubricating Oil (Element of Interest, 500 [/.g/g) NBS No." Element of interest Compound Drying time, hr 6 Weight used,^ c 1075a Al 2-Ethylhexanoate 48 0.620 1051b Ba Cyclohexanebutyrate 24 0.174 1063a B Menthyl borate 2 2.08 1053 Cd Cyclohexanebutyrate 48 0.202 1074a Ca 2-Ethylhexanoate 48 0.373 257 JOHN A. DEAN AND THEODORE C. RAINS Appendix 3: (Cont.) Element of Drying Weight NBS No." interest Compound time, hr" used,g c 1078 Cr Tris( 1 -phenyl- 1 ,3-butanediono)- chromium(HI) 1 0.515 1055b Co Cyclohexanebutyrate 24 0.337 1080 Cu Bis(l -phenyl- 1 ,3-butanediono)- copper(II) 0.5 0.303 1079b Fe Tris(l -phenyl- 1 ,3-butanediono)- iron(lll) 1 0.485" 1059b Pb Cyclohexanebutyrate 48 0.136 1060a Li Cyclohexanebutyrate 24 1.22 1061b Mg Cyclohexanebutyrate 48 0.765 1062a Mn Manganese(II) cyclohexanebutyrate 48 0.362 1064 Hg Mercury(II) cyclohexanebutyrate 2 0.138 1065b Ni Cyclohexanebutyrate 48 0.360 1071a P Triphenyl phosphate 2 0.526 1076 K Erucate 2 0.495 1066 Si Octaphenylcyclotetrasiloxane 0.141' 1077a Ag 2-Ethylhexanoate 24 0.117 1069b Na Cyclohexanebutyrate 48 0.417 1070a Sr Cyclohexanebutyrate 24 0.242 1057b Sn Dibutyltin bis(2-ethylhexanoate) 2 0.218 1052b V Bis(l-phenyl-l ,3-butanediono)- oxovanadium(IV) 2 0.348 1073b Zn Cyclohexanebutyrate 48 0.299 " National Bureau of Standards (U.S.) standard reference material. " Dried over P 2 6 in desiccator. c Weight of dried material which contains 50 mg of element of interest. ' Dried in oven at 11 0°C. ' Weight of element, 20 mg, and gives a final concentration of 200 /xg mercury in 0.00LV HN0 3 . It is evident from this study that the anions of all acids do affect the release of mercury in varying degrees, especially iodide, bromide, sulfate, and phosphate. In the presence of minor quantities of cations, Mention of firm names or trade products does not imply endorsement or recommendation by the National Bureau of Standards over other firms or similar products not mentioned. 263 RAINS & MENIS: MERCURY IN STANDARD REFERENCE MATERIALS Table 3. Effect of extraneous anions on recovery of mercury" (reduction step) Anion Concn, N Rec, % N0 3 -' 0.06 95 0.6 88 3.0 88 6.0 81 9.0 86 ci-' 0.4 95 1.6 97 3.6 94 5.4 82 7.2 73 cio 4 -' 0.5 97 1.0 88 2.4 82 4.8 66 Br-'* 0.01 96 0.02 75 |-16 0.002 97 0.02 3 S0 4 - 2 0.4 99 0.8 95 3.6 78 7.2 50 C 2 C-4- J 0.2 92 0.4 90 0.6 77 PO4- 3 0.18 98 1.8 82 9.0 65 " Total mercury present = 0.1 fig. 6 Added as the sodium salt. no adverse effect was observed. However, when an excess of reducible ion was present, e.g., Cr207~ 2 , the mercury was not released, owing to an insufficient quantity of reducing agent. Contamination Problems. — Contamination from reagents, glassware, and the environment were encountered with mercury. To minimize this contamination, all glassware was rinsed with 10% nitric acid. Samples stored in glass for over one week showed that values for mercury increased by 50-100%. Therefore, polypropylene flasks were used for storage. To illustrate the degree of contamination of reagents by mercury, 3 bottles of the same lot of hydroxylamine hydrochloride were analyzed and were found to contain 0.008, 0.027, and 0.031 Mg Hg/g, respectively. In an- other lot, the mercury content was 0.001 ng Hg/g. Ultra high purity acids were also tested. Some lots of ACS reagent grade nitric acid con- tained as much as 30 ng Hg/ml and had to be redistilled by sub-boiling distillation from an all- quartz still. Table 4. Comparison of results obtained with different methods for the determination of mercury in NBS Standard Reference Materials" Rel. Std No. of Hg, Std Dev. t Determinations MS/g Dev. 6 % Method Orchard Leaves (NBS-SRM-1571) AAS C (single beam) 7 0.160 0.006 4.0 AAS (ratio) 6 0.160 0.006 4.0 Neutron activation 11 0.155 0.003 1.8 Isotopic dilution mass spectrom- etry 7 0.14 0.02 16 Liver (NBS-SRM-1577) AAS (single beam) Neutron activation 0.0165 0.0145 0.0008 0.0017 Coal (NBS-SRM-1630) AAS (single beam) Neutron activation 7 55 0.139 0.127 0.007 0.006 4.7 4.8 NBS Standard Reference Materials are available from the Office of Standard Reference Materials, Na- tional Bureau of Standards, Washington, D.C. 20234. b For a single determination. c AAS = atomic absorption spectrometry. Another possible source of contamination is the vacuum oven and/or desiccant. Samples of or- chard leaves, which were dried at different inter- vals over a period of 3 months in 2 different vacuum ovens at 90°C for 24 hr and then cooled in a desiccator over anhydrous calcium sulfate, produced mercury values which were 25-50% higher than the undried samples. The anhydrous calcium sulfate was tested and was found to con- tain traces of mercury. These studies reinforce the conclusion for the need of careful checks on all sources of possible contamination. Comparison of Single Beam and Double Beam (Ratio) Systems. — This study was designed to evaluate the accuracy of the flameless atomic absorption method for determining mercury. By comparing data in the literature by the flameless method with other analytical techniques, we ob- served that the flameless method produced higher values. The higher values may have resulted from improper background correction due to molecular absorption or from scattering of light. To correct these 2 sources of error, mercury was determined in orchard leaves by both the single beam and double beam (ratio) system. As shown in Table 4, the 2 methods gave the same values when the 264 JOURNAL OF THE AOAC (Vol. 55, No. 6, 1972) proposed method was used. However, when the temperature of the absorption cell was less than 145°C, the absorption values for mercury were erratic and higher in the single beam system. In this case, by taking a ratio of the Hg line at 2537 A to the Al line at 2652 A in the double beam system, a correction for light scattering and/or molecular absorption, due to water and other vapors, was made. Application of the Method. — The method was tested on 3 types of NBS Standard Reference Materials as shown in Table 4. This technique has also been used for the determination of nanogram quantities of mercury in samples of aluminum alloys, reactor cooling water, and water effluents. In summary, this study reveals that nameless atomic absorption analysis of mercury can be made to be precise and accurate in comparison with 2 independent techniques. Acknowledgments The authors would like to thank E. C. Kuehner, T. J. Murphy, and associates for making available the special purity reagents. References (1) Muller, K. (1930) Z. Physik 65, 739-758 (2) Woodson, T. T. (1939) Rev. Set. Inslr. 10, 308 (3) Manning, D. C. (1970). Al. Absorption News- letter 9, 97-99 (4) Hatch, W. R., & Ott, W. L. (1968) Anal. Chem. 40, 2085-2087 (5) Hillebrand, W. F., Lundell, G. E. F., Bright, H. A., & Hoffman, J. I. (1953) Applied Inor- ganic Analysis, 2nd Ed., Wiley, New York, pp. 347-351 (6) Toribara, T. Y., Shields, C. P., & Koval, L. (1970) Talanta 17, 1025-1028 (7) Munns, R. K., & Holland, D. C. (1971) J AOAC 54, 202-205 265 Extrait de la revue du GAMS, 1971, n" 3, p. 39. APPENDIX XVI DISCUSSION OF SOME EXPERIMENTAL AND FUNDAMENTAL CONDITIONS IN ANALYTICAL FLAME SPECTROSCOPY by R. MAVRODINEANU Institute for Materials Research, National Bureau of Standards, Washington, D.C. 20234, U.S.A. « Do not all fix'd Bodies, when heated beyond a certain degree, emit Light and shine; and is not this Emission perform'd by the vibrating motions of their parts? » Sir Isaac Newton. Opticks, 1730. The first part of this paper will be devoted to a more detailed discription of this measuring tool which was designed and built at NBS and is now used for analytical work and for studies of various flame parameters. In the second part the reader will find a general discussion of nonthermal excitation phenomena occurring in combustion flames. The particular case of chilled hydrogen flames will be examined and illustrated with two examples involving the exci- tation of sulfur and phosphorus contained in air as inorganic and organic molecules. Discussion de quelques conditions experimentales et fondamentales en spectroscopic analytique de flamme. La premiere partie de ce travail est consacree a une description detaillee des proced6s de mesure qui ont ete concus et realises au National Bureau of Standards, et maintenant utilises dans les travaux analytiques et pour l'etude des divers parametres de flamme. Dans la seconde partie, le lecteur trouvera une discussion gen£- rale des phenomenes d'excitation non thermique se produisant dans les flammes de combustion. Le cas particulier de la flamme hydrogene refroidie, est examine et illustre de deux exemples sur l'excitation du soufre et du phosphore contenus dans l'air, sous forme de molecules mindrales et organiques. Diskussion iiber einige experimentelle und grundlegende Bedingungen in der analytischen Flamm-Spektroskopie. Der erste Teil dieser Arbeiten wird einer ins Einzelne gehenden Beschreibung der Messverfahren gewidmet, welche im National Bureau of Standards ausgearbeitet und durchgefuhrt wurden, und jetzt fur die Analysen und die Studie der diversen Flamm- parameter zur Anwendung kommen. Im zweiten Teil findet der Leser eine allgemeine Beschreibung iiber nicht thermische Erregungsvorgange, die in der Verbren- nungsflamme auftreten. Der Sonderfall der gekuhlten Wasser- stofflamme wird gepruft und an zwei Beispielen iiber die Erre- gung des in der Luft enthaltenen Schwefels und L-Phosphors erlautert, in Form von mineralischen und organischen Mole- kiilen. Introduction. and are based on the phenomenon that matter, when properly energized, may emit or absorb the exciting energy as electromagnetic radiations having frequencies characteristic of the excited species and whose energy is proportional to the number of particles involved in the process. Thus the frequencies radiated or absorbed serve to identify the excited species, whereas the amount of radiation gives a measure of the quantity of matter involved in the phenomenon. These properties constitute the basis of qualitative and quantitative analytical flame spectroscopy. Instruments for making measurements in the emis- sion mode have been known for a long time and include an exciting flame with its analytical sample supplying system, a radiation-dispersing instrument, and a photosensitive detector. These basic compo- nents are common to all emission spectrophoto- meters and can be found in the earliest instruments. Figure 1 describes the arrangement used in 1860 by Kirchhoff and Bunsen for their spectroscopic investigations ( x ). The same basic elements were applied 13 years later by Champion", Pellet and Grenier who produced in 1873 the first quantitative emission spectrophotometer ( 2 ). The instrument, named by its developers the « spectronatrometre », was based on the principle enunciated by Janssen This discussion will cover certain aspects of the Fig. 1. — Components used by Bunsen and Kirchhoff in their spectroscopic investigations. At the right, the flame excitation source with its sample supplying system; in the middle, the spectroscope; the photosensitive device used to observe the radiations was, in this case, the human eye. experimental and fundamental nature of flame spectroscopy when used as a measuring tool in ana- lytical chemistry. The underlying principles of this measuring tool have been known for a long time, 266 in 1870 ( 3 ). The same principle and instrumental elements have been used by Lundegardh in 1929 and 1934 in his pioneering work on analytical flame spectroscopy, and are found today in every emission flame spectrophotometer ( 4 ). The instrumental means for performing measure- ments in the atomic absorption mode include, in addition to those mentioned for the emission mode of operation, a source emitting those radiations which originate from the chemical species to be determined. Again, as was true for the flame emission mode, the principle and the instrumental elements of flame atomic absorption have been known for a long time. The optical arrangement described in 1872 and 1873 by John Tyndall and shown in figure 2 ( 5 ) can be used to illustrate the point. One can see the emission source is an electric arc, followed by the absorbing flame with its sample-supplying sys- tem, and finally a means to observe the absorption phenomenon. The same arrangement was used in 1953 by Alan Walsh and described in his pioneering works on the application of the atomic absorption principle to chemical analysis ( 6 ). These same com- ponents are found today in all analytical instruments using the atomic absorption principle. Fig. 2. — Optical arrangement used by Tyndall to demons- trate the principle of atomic absorption. From left to right, emission source followed by the absorbing flame with its sample supplying system and the means to produce and observe the spectrum. The instrumentation which will be described now, contains the basic components described above. It was designed to utilize the fundamental property of spectral sources to emit or absorb simultaneously at all transitions originating from all species which are excited in the particular source considered. As a consequence of this property, the measuring system was designed and built to detect and measure simul- taneously a number of individual radiations emitted or absorbed in the wavelength region of about 2 000 A to 10.000 A. The radiations are detected and their intensities are measured in a conventional manner using appropriate photomultiplier tubes and integration of the optical signal over a predetermined time interval. The current developed in the photo- multipliers, which is related to the radiation inten- sity originating from the excitation source and to the concentration of the excited species, is displayed both in analog and digital form. The digital infor- mation is printed and punched simultaneously on appropriate tapes. The punched tape is compatible for direct use with a computer programmed to produce meaningful data to the analytical chemist. While this instrumentation does not include any original or new element, it is believed, nevertheless, that the combination of principles and instrumental elements will provide a final measuring tool showing improvements and advantages over the more conven- tional flame spectrometers of the past. The first part of this paper will be devoted to a more detailed description of this measuring tool which was designed and built at NBS and is now used for analytical work and for studies of various flame parameters. In the second part the reader will find a general discussion of nonthermal excitation phenomena occuring in combustion flames. The particular case of chilled hydrogen flames will be examined and illustrated with two examples involving the exci- tation of sulfur and phosphorus contained in air as inorganic and organic molecules. 2. — Instrumentation. 2.1. The Spectrometer. The spectrometer used to detect the radiations originating from the exciting flame is a conven- tional 1 m grating instrument using the Czerny- Turner optical arrangement illustrated in figure 3 ( 7 ). The bilateral variable entrance slit is seen at the left of the instrument. It is followed by a colli- mating concave mirror which projects a parallel radiation beam on an interchangeable flat grating (100 mm by 100 mm carrying 600 lines/mm or 1 200 lines /mm). The diffracted radiations are focused as mono- chromatic images of the entrance slit at the focal plane, contained in the rectangular light-tight enclo- sure at the right of the spectrometer. A vertical frame is located at the focal plane where it can be moved up and down through a mechanism operated from the outside. It is provided with a counter to indicate its position. The frame can accept interchangeably a bilateral variable exit slit or a cassette capable of holding a photographic plate or film 20 cm by 5 cm. An adapter for Polaroid 267 Fig. 3. — NBS flame spectrometer for simultaneous multielement emission and atomic absorption measurement left, console with the electronics used in the scanning mode with digital and analog data presentation center, 1 m Czerny-Turner type scanning monochromator-spectrograph-multielement instrument. At the righ bed type optical bench carrying the all-alumina sprayer and burner unit and the dual sample solution s system. The exponential water column manometer used to monitor the pressure inside the sprayer and bu: is seen at the extreme right ( 7 ). film is also available. The vertical frame is provided at its lower end with a longitudinal opening 20 cm by 2.5 cm which can accept a number of fixed exit slits or a metal plate into which can be cut a number of slits. When the variable exit slit is placed in the proper position in the vertical frame, the spectrometer operates as a scanning monochromator. The scanning is performed by rotating the grating through a synchronous electric motor provided with 12 scan- ning speeds (0.5 to 2 000A/mn). A wavelength counter indicates the wavelength directly in 0.1 A units. When the exit slit is exchanged for the cassette, the instrument can be used as a spectrograph. Finally, when the vertical frame is lifted in its upper position, the multislit arrangement is placed at the focal plane and permits the isolation of a number of individual monochromatic radiations. These radia- tions are received by individual photomultiplier tubes located on a horizontal support. In this case, the spectrometer operates as a multichannel ment; this latter arrangement is illustrc figure 4. In this case, six individual slits in the multislit plate at wavelengths selei isolate the radiations from Na (X 5 893 (X 7 667 A); Li (X 6 707 A); Ca (X 4 227 A): (X 6 230 A); and Mg (X 2 852 A). Behind ev is a side-on photomultiplier provided wi corresponding connections, and a seventh used for the scanning mode. As can be seen in figure 3, a precision la type optical bench 160 cm long is placed alt optical axis in front of the entrance slit of the s meter. The following equipment is placed bench using appropriate carriers : a quar which produces a demagnified image of the tical flame on the entrance slit, and the bun sprayer unit with its dual analytical sample J supplying system. A second quartz lens, a chopper and a dual hollow cathode arran 268 Fig. 4. — Light-tight enclosure located at the focal plane of the spectrometer. It contains seven photomultiplier tubes placed behind the exit slits carrying plate ('). Fig. 5. — The opposite extremity of the optical bench from figure 3 is seen here with the sprayer and burner unit together with the dual sample supplying system, and with the addi- tional equipment. This consists of a quartz lens, the mirror chopper, and the dual hollow cathode arrangement. At upper left, console containing the electronics used in conjunc- tion with the multichannel mode of operation. At the right, console carrying the equipment used to regulate and monitor the flow of oxidant and fuel gases supplied to the sprayer and burner unit ('). 269 follows and is illustrated in figure 5. The mirror chopper directs successive radiations from the two hollow cathodes through the optical axis of the instrument, and at the same time generates an ac signal. Such a dual hollow cathode system was proposed previously by other workers. The lenses, burner and sprayer unit, chopper, and hollow cathodes can be positioned along the optical axis in the x-y-z directions and can be tightly secured in the correct position. 2.2. — Electronic Circuitry. The electronic equipment used with the scanning mode is shown in figure 3. It consists of conventional solid state elements, such as a wide band pre- taneous flame emission and atomic absorption measurements uses a less conventional electronic arrangement which will be discussed briefly. This arrangement was designed and built in 1965 by P. Cath and G. Ziedens. Its principle is described in the block diagram shown in figure 6 ( 8 ). It consists of four major steps : 1) conversion of light intensity to electrical current; 2) measurement of this current; 3) conversion of the measurement from analog to digital form; and 4) print-out and simultaneous punch-out of the data. O O- SEftlAL CHANNEL SWITCH INO ANAL09 TO 0I9ITAL COHVtHTIB Fig. 6. — Block diagram of the electronic system used for the multichannel mode of operation ( 8 ). amplifier, providing a gain of 100, a selective ampli- fier providing a gain of 10, and a lock-in amplifier providing a gain of 10. The output from the lock-in amplifier is supplied to a 4 digit digital voltmeter and to an analog recorder. The power supply for the scanning photomultiplier is seen at the top of the console and it has a dc output of 2 100 volts and 30 milliamperes with a line and load regulation of 0.001 percent and a stability of 0.005 percent per hour. At the bottom of the same console, one can see the power supply for the hollow cathode tubes which has the following general characteris- tics : current output range 2.2 [xA to 220 mA ; voltage 730V; current regulation 0.1 percent; stability: 0.2 percent. The operation of the spectrometer in the scanning mode is conventional and does not require further discussion (since the electronic system uses a lock-in amplifier the lock-in chopper seen in figure 5 is moved between the flame source and the entrance slit of the spectrometer). The multichannel mode of operation for simul- The first step is achieved by a photomultiplier tube, and the system can accept up to eight such tubes. Only two tubes are described in figure 6 since the remaining six are of similar design. The measurement of the current generated by the photo- multiplier is performed by individual integrating capacitors where it is stored and then sequentially converted to measurement in digital form. This information is then printed and punched simulta- neously for all channels on appropriate tapes, and the punched tape is compatible for direct input to a computer. The eight channels can accept either ac or dc inputs in any combination. The ac (290 HL) mode is used in conjunction with the atomic absorp- tion measurements while the dc input is used for flame emission measurements. Each channel can be switched on or off and has a sensitivity adjustment and a dark current compensation capability. The integration time can be selected step-wise, as needed, for 10, 20, 40, 100, 200, 400 and 1 000 seconds. The measurements can be made with 1 to 4 repli- 270 Fig. 7. — Front panel of the console containing the electronic circuitry used in the multichannel mode of ope- ration ( 8 ). SAMPLE IDENTIFICATION NUMBER OF RUNS = 3 INTEGRATION TIME CHANNEL SENSI- SO SECONDS AVERAGE ESTIMATED NUMBER TIVITY CONCENTRATION STD DEV (PPM) 1 - MG 3 4.01939 .245 A - K 5 59.14011 4.480 6 - CA 6 15.26471 2.692 6 - NA 6 8.06193 2.040 SAMPLE IDENTIFICATION 10 NUMBER OF RUNS = 5 INTEGRATION TIME 20 SECONDS CHANNEL SENSI- AVERAGE ESTIMATED NUMBER TIVITY CONCENTRATION STD DEV (PPM) 1 - MG 3 3.98570 .153 4 - K 5 58.31750 3.900 6 - CA 6 14.81440 2.128 8 - NA 6 7.8924 1 1.560 SAMPLE IDENTIFICATION 11 NUMBER OF RUNS = 8 INTEGRATION TIME 20 SECONDS CHANNEL SENSI- AVERAGE ESTIMATED NUMBER TIVITY CONCENTRATION STD DEV (PPM) 1 - MG 3 3.89575 .134 4 - K 5 57.80517 3.910 6 - CA 6 14.88400 1 .950 8 - NA 6 7.99028 1 .302 END OF COMPUTATION CPU TIME: 27 SECS. TERMINAL TIME: 0:3! 38 PLEASE LOG IN: 05 Fig. 8. — ■ Digital and computer data presentation obtained from the unit from figure 6 and 7. At the right, printed tape; center, punched tape; at the left, computer data presentation. 271 cations or continuously, the .sequence in this case being interrupted manually. The console which contains the electronic circuitry is seen in figure 5, while the panel which carries the corresponding command switches is illustrated in figure 7. The simultaneous digital print-out and punch-out display contains the following information : channel number, sensitivity setting, integration time, and the arbi- trary counts corresponding to the charge of the individual capacitors. The arbitrary counts are converted to meaningful information through cali- bration with analytical reference solutions containing known amounts of elements. An actual example of data presentation is illustrated in figure 8 which shows a portion of the printed tape at right, followed by the corresponding punched-out tape and the computer information. The printed tape contains, from left to right, in the first column the channel number, the sensitivity setting, and the arbitrary counts corresponding to the charge of the capacitors. The first row lists an identifying number permitting the computer to identify the reference solution (0) and the unknown samples (1), followed by the inte- gration time which, in this case, is 20 s. The com- puter program which handles the analytical data obtained with the multichannel spectrometer was written by Dr. J. Wing. In the present work, the electronic system was set to accept four signals in the dc mode for the flame emission determination of sodium, potassium, calcium, with lithium as an internal standard and one signal in the ac mode for the determination of magnesium by flame atomic absorption. Since only one magnesium hollow cathode tube was needed, the chopper was placed on the optical axis in a perpendicular position and a conventional blade was substituted for the mirror blade illustrated in figure 5. Since the amplifier system used in conjunction with the scanning mode is the lock-in type, it was necessary to select a suitable chopping frequency which would be different from that of the ac line current. A major source of interference in the ope- ration of the lock-in amplifier could be the analytical flame itself and its supply system. A series of measure- ments was designed and performed to determine whether any periodicity originated in the flame. These measurements were performed by setting the monochromator on the sodium emission and then connecting the output from the scanning photo- multiplier to an oscilloscope. The time scan obtained with the flame alone (A), the flame supplied with distilled water (B), and the flame supplied with a sodium salt (C) is illustrated in figure 9. As can be seen, the unsupplied flame in A and B exhibits a random noise of low amplitude. When sodium was supplied to the flame through the pneumatic sprayer, > Fig. 9. — Illustration of premixed acetylene-air flame perio- dicity. A, oscilloscope tracing of the signal from the unsup- plied flame, B, same tracing from the flame supplied with distilled water. C, same tracing with the flame supplied with a sodium salt. In all three cases, the monochromator was set on the sodium emission at X 5 890 ( 7 ). The abscissa represents the time; the ordinate represents relative inten- sities. the resulting signal indicated a strong periodicity with a frequency of 25 H, as illustrated in C. This periodicity is probably due to pulsations generated in the pneumatic sprayer and their frequency should depend on the geometry and on the operating condi- tions of the particular unit used. As a result of these measurements, the chopping frequency in the ac mode was selected to avoid interferences with the fundamental frequency of the flame or any of its harmonics. The stability of the electronic circuitry, including the photomultiplier tubes, was measured by radia- tions provided by a fluorescent screen activated with 2Ci of tritium (12.5 years half-life) ( 9 ). This source, having an output of 56 y.\m, was placed at the focal plane preceding the photomultiplier tubes. An integration time of 10 s was selected for each channel, and a series of measurements was obtained for each photomultiplier tube. The standard devia- tion in response to this constant illumination was 0.1 to 0.3 percent. 272 2.3. — Burner and Sprayer Systems. The burner and sprayer system used with the spectrometer is illustrated in figure 10 ( 10 ). It consists of an indirect external pneumatic sprayer, a spraying vessel, and a burner body, all made of pure and Fig. 10. — Pure alumina (99 % A1 2 3 ) pneumatic sprayer and premixed burner system. At the right, spraying vessel with the external pneumatic sprayer at the bottom (only the sample supplying tube is visible). In the middle, burner body with the fuel nozzle made of tantalum. At the left, burner head with water cooled tantalum flame plate. The thin Teflon flash back safety membrane is seen at extre- mity of the horizontal tube ( 10 ). At the bottom, inter- changeable burner head. Courtesy Plenum Press. impervious alumina (99%AI 2 3 ); the interchan- geable burner plate, made of tantalum, is water cooled. This unit is corrosion and erosion proof and non-combustible. The gas mixtures which can be used to produce a premixed flame are air-acetylene or nitrous oxide-propane with the same burner plate and nitrous oxide-acetylene with an appropriate burner plate. Details of the unit are given in the isometric drawing from figure 11, and various para- meters are given in table I and II ( 10 ). The spraying vessel is connected to a polyethylene drain bottle through a glass hydraulic trap as illustrated in figure 12. The side tube at the top of the trap is connected to an exponential water manometer which indicates the pressure inside the burner and sprayer unit. When the air-acetylene or nitrous oxide- propane mixture is used, this pressure is of about 1 cm of water; with the nitrous oxide-acetylene the pressure should be 7 cm. Since this burner and sprayer unit has been described in previous work, no further details will be given here ( 10 ). The system for regulating and monitoring the flow of oxidant and fuel gases is illustrated in figure 5 ( 7 ). The oxidant gas is supplied to the panel first after passing through a two-stage pressure regulator provided with a stain- less steel check valve and a flash-back safety consis- ting of a porous metal plate. A second pressure regulator controls the oxidant flow and its pressure is measured with a two-turn 8 1/2 in-(21.59 cm) diameter precision gauge which is capable of measu- ring 0.1 (0.0068 atm) psi per division. The flow is controlled with a 250 mm long glass flowmeter; copper tubing and appropriate fittings are used with air, oxygen, and nitrous oxide. The fuel is supplied to the panel from a corresponding container through a two-stage pressure regulator provided with a stain- TABLE I Functioning Parameters of the Sprayer. Air to sprayer Vacuum at sprayer inlet, cm Hg Capillary, x 1/1 000 in. Analytical solution Sprayer efficiency, 0/ /o Pressure Flow, liter /mn Consumed by the sprayer g/mn Supplied to the burner Q„, air Qv, snl g/mn g/liter 20 30 40 9.8 12.5 15.0 30.0 42.0 38.0 9 6 9 12 14 9 5.0 1.7 5.7 11.0 14.0 5.5 0.25 0.21 0.32 0.36 0.37 0.33 0.026 0.017 0.026 0.029 0.030 0.022 5.0 12.4 5.6 3.3 2.6 6.0 2 000 7 353 2 200 1 140 890 2 730 273 Capillary, x 1 /l 000 in. 6 9 14 6 9 14 TABLE II Particle Size Distribution at 30 psi. Particle diameter, [x 0.5^1>1<2>2<4>4^6>6< 10 > 10 ^ 12 > 12 14 61.4 62.6 62.3 2.8 2.5 1.5 23.7 22.9 21.8 11.8 11.2 11.4 2.7 2.6 3.2 0.2 0.5 0.7 0.05 0.2 0.3 Total weight, % 8.5 33.8 30.6 7.4 4.3 29.4 31.2 8.5 17.9 3.7 11.6 17.7 22.9 13.6 10.5 0.05 0.2 0.2 6.7 11.4 18.1 POROUS STAINLESS STEEL 316 075 #17 OR 1 j t J-.I44 9/64" DB F.200 7/64" OR J_26l #47 DR j 369 #60 OR } 449 #70 OR f ALL DIMENSIONS ARE IN INCHES Fig. 11. Isometric drawing of the sprayer and burner unit shown in figure 10 ( 10 ). Courtesy Plenum Press. less steel check valve and flash-back safety consisting of a porous metal plate; the acetylene purifying cylinder is placed after this plate (see para- graph 2.4.1). The flow of fuel is further regulated at the panel and its pressure is monitored by, a 8 1/2 in (21.59 cm) diameter two-turn precision gauge capable of measuring 0.01 psi (0.000 68 atm) per division. The flow is controlled by a 250 mm long glass flow-meter. Stainless steel tubing and appro- priate fittings are used with acetylene. 274 RUBBER PAD Fig. 12. Hydraulic trap made of glass ( 10 ). Courtesy Plenum Press. 2.3.1. Operation of the Burner and Sprayer Unit. We have mentioned that the sprayer can be supplied through a two-way Teflon (*) stopcock in succession with the analytical solution or with the sprayer with the stopcock and the constant level flask on one side, and the analytical sample container on the other side. This dual arrangement simulates to some extent a double-beam operation. Indeed, if one assumes that the emission source and the flame, with its supply system, are stable during the interval required to change from the sample to the blank, the system is equivalent to a double beam operation. This interval of 15 s which is required to change from sample to blank is determined by the memory of the sprayer and burner units. Since we use 20 s integrations of the sample and blank signals, the total time required to perform a sample measurement is 35 s; a similar interval of 35 s is required for the blank. Hence, one sample-blank measurement cycle requires 70 s. 2.3.2. Analytical Data. The technique described was used to obtain the analytical data reported in table III and to establish the accuracy of the measurements. The elements selected in this case were Mg, K, Ca, and Na because of their significance in biochemistry. It is obvious that any combination of elements which can be determined by flame emission and atomic absorption TABLE III Accuracy of Analytical Resulls for Mg, K, Ca, Na Determined Simultaneously. Element determined, ji.g/ml Number of Mg (A) = = 4.00 K(*) = = 60.00 Ca(*) = = 15.00 Na(*) = 8.00 Runs Found % Difl Found % Difl Found % Difl Found % Difl 3 4.02 + 0.50 59.14 — 1.43 15.26 + 1.73 8.06 + 0.75 5 3.99 — 0.25 58.32 — 2.80 14.81 — 1.27 7.89 — 1.38 8 3.90 — 2.5 57.81 — 3.65 14.88 — 0.80 7.99 — 0.13 12 3.97 — 0.75 58.66 — 2.23 15.01 + 0.07 8.00 0.0 15 3.95 — 1.25 58.32 — 2.80 14.92 — 0.53 7.97 — 0.38 20 3.92 — 2.0 58.04 — 3.27 14.88 — 0.80 7.96 — 0.50 25 3.90 — 2.5 57.68 — 3.87 14.84 — 1.07 7.93 — 0.88 (A) By flame atomic absorption. (*) By flame emission. distilled water blank solution. This dual supplying system is illustrated in figure 5 ( 7 ). One can see the stopcock, the constant level polyethylene flask, and the polyethylene capillary tubing which connects the (*) The name of manufactured products are given for information only and do not constitute an endorsement of the product. techniques can be chosen. Magnesium was determined in the atomic absorption mode while K, Ca and Na were determined in the emission mode. The cali- bration was made on eight solutions containing all four elements of known concentrations; Li was used as an internal standard for emission measurements. Each standard solution was run in six replications. 275 each measurement consisting of the blank and sample run. A solution was then selected from these stan- dards and treated as the « unknown » sample. This solution was run in 3 to 25 replications as indicated in column one from table III, each run consisting of the blank and sample measurements. Columns 2 to 5 show the analytical results obtained for Mg, K, Ca, and Na as well as the percent difference between the concentration found and the concen- tration taken. In all atomic absorption measurements the computer was programmed to interpolate from a log versus concentration system, while the sample interpolation for the emission measurements was made from a sample and blank versus concentration system. In the case described in table III, the internal standard values obtained for lithium were not used to compute and correct the emission measurements. The analytical results obtained demonstrate the accuracy which can be expected from the multi- channel spectrometer when simultaneous measure- ments are made by emission and atomic absorption on four elements. 2.4. Increasing the Sensitivity of Measurements. Various means can be used to increase the sensi- tivity of measurements in analytical flame spectros- copy, and three possibilities will be discussed. 2.4.1. Decreasing the Flame Background. 2.4.1.1. It has been known for a long time that phosphine (H 3 P) is one of the impurities of acetylene which is difficult to eliminate. This compound imparts a noxious odor and toxic properties to acetylene. When introduced into the air-acetylene and espe- cially oxygen-acetylene flames, it radiates a strong visible continuous or non-resolved spectrum as illus- trated in figure 13 a and b. This emission can inter- feie with the determination of the elements, and it is essential to eliminate this contaminant. Since the purification of acetylene was discussed in detail in some of our earlier works ( u > 12 ). only the practical aspect will be discussed here. The H 3 P can be elimi- nated from acetylene to an extent that no radiation can be detected in the flames with the instrumen- tation described in this work. This is done by oxi- dation to H 3 P0 4 using a ferric chloride solution containing an oxidation accelerator, adsorbed on an inert support such as diatomaceous earth. The mate- rials are packed in a stainless steel cylinder 15 in (38.1 cm) long and 2 in (5.08 cm) in diameter which is provided with side tubes. The cylinder is placed between the acetylene tank and the gas monitoring console illustrated in figure 5. Acetylene thus obtained produces a faint blue flame when burned with air, as illustrated in figure 13b. Before puri- fication, the flame is seen in figure 13a and the corresponding spectrophotometer scannings are given in figure 14. 2.4.1.2. Impurities which can interfere with the measurements are those introduced into the flame RAD1AM0N FftOfl AH ACETYLENE-AIR FLME PRODUCED WITH CVKKIAL GRADE AND PURIFIED FUEL. IHI RADIATION ORIGINATES FROM HjP PRESENI III THE COWEfiCIAJ. GRAtt FUEL HAVtLLNGTH. A Fig. 14. — Monochromator scanning of the" radiations excited in the flame illustrated in figure 13 a and b. Kf CONTAMINATION OF LABORATORY AIR - MEASURED BY AN C 2 H 2 -AIR FLAME Fig. 15. — Monochromator scanning illustrating the emission of sodium present in the laboratory air surrounding the flame. In this case, the monochromator was set on the sodium emission and the scanning was done in time. 276 he surrounding air. Such impurities are sodium ilcium salts which are always present in the :ory atmosphere. The presence of sodium in boratory was determined by monitoring its ons, which were excited in an acetylene ;d) — air (fdtered) flame, using our spectro- set on the Na wavelength. The results are in figure 15 and show the recording of Na on intensities on the ordinate versus time on scissa. The full scale deflection corresponds )b Na. As can be seen, the Na content of the rounding the flame can be more than 1 ppb. element must be determined at the ppb level, bvious that the air entering the flame from mosphere must be free of sodium contami- This can be done by working in a clean )r by supplying purified air to the flame. ',. Increasing the Optical Pathlenglh Through lme. r as demonstrated by Gouy ( 13 ) and later by :gardh ( 4 ) that the emission intensity of the radiations from a flame can be increased by leng- thening the optical path; similarly, Walsh ( 14 ) has shown that the absorption of a flame can be aug- mented by increasing its length. This increase can be achieved by producing a longer flame or by increasing the optical path through the flame using multiple reflection arrangements. Both have been used in our work and will be described. 2.4.2.1. Long Path Flames. Figure 16 shows a long path burner head capable of producing a 30 cm long and 1 cm wide flame. Figure 17 gives details on the construction of this burner ( 7 ). It is made of stainless steel type 316, tubular sections assembled with appropriate fitting of the same metal welded in a hydrogen furnace. The premixed acetylene-air or propane-nitrous oxide, together with the fine spray of analytical sample, are supplied to the burner head from both extre- mities to insure the production of an even flame. The air is supplied to the external pneumatic sprayer under a pressure of 39 psi (2.65 atm) and at a rate . — Stainless steel long-path premixed acetylene-air burner head adapted to a Teflon spraying vessel provided h a tantalum external variable sprayer. 277 Fig. 13. — a. premixed acetylene-air flame produced with purified fuel, commercial grade acetylene. Exposure time same as that used for a. b, same flame produced with unpurified Fig. 25. — a, appearance of the dual hydrogen flame supplied with distilled water, sulfuric acid, c, same flame supplied with diluted phos-phoric acid. b, same flame supplied with diluted 278 type mirrors with an asymmetric biconvex quartz lens. The principle of the proposed system and the specific dimensions are shown in figures 18 and 19. Fig. 17. — Drawing of the unit shown in figure 16. The burner head carries 112 holes of 1 mm in diameter distributed over three parallel rows. The holes from the center row are staggered between those from the two lateral rows. Length of rows, 2.50 mm. Total width, 5 mm (center to center). Distance between rows, 2.5 mm (center to center). Distance between individual holes, 8 mm (center to center). The stainless steel tubing has an inside diameter of 13 mm and an outside diameter of 16 mm. of 2 200 1 /hr. The acetylene was supplied to the burner body through a tantalum nozzle to produce a stoichiometric flame. The consumption of analy- tical solution in the sprayer 25 ml/mn and the amount of analytical sample solution entering the flame was 1.3g/mn. The optical system used with this 30 cm flame consisted of two quartz lenses which produce an optical beam having an average cross section of about 10 mm. Under these circumstances, the analytical results obtained when Mg was deter- mined by atomic absorption in conjunction with the scanning electronics described in paragraph 2.2 showed an increase in sensitivity by a factor 3, as compared with a 10 cm path burner. The optical system using the quartz lenses could be improved by using a two Cassegrain mirror on- axis arrangement as described by Muller-Herget. This would eliminate the chromatic aberrations characteristic to all lens systems and produce a more adequate optical beam geometry ( 15 ). 2.4.2.2. Triple Pass On-Axis Optical Arrangement. This on-axis triple arrangement system (*) can be used for atomic absorption and flame emission measu- rements and is based on the use of two Mangin- (*) Designed by "W. Muller-Herget, Philips Laboratories, BriarclifT Manor, New York, 10510, U.S.A. Fig. 18. — Principle of the on-axis triple pass system. ALL DIMENSIONS SHOWN ARE IN INCHES Fig. 19. — Drawing of the triple-pass, center bore, Mangin mirrors assembly. A view of the actual arrangement in working condi- tions is illustrated in figure 20. The flame cavity is designed as a concentric triple- pass system with parallel radiations in the second pass. Because of heat and corrosion, it is advan- tageous to use rear surface mirrors made of fused quartz. The optical elements are Mangin-type mirrors with a center bore. The cavity has been designed to minimize spherical and primary color aberrations in the spectral range 240 nm to 436 nm with the design center at 330 nm. The object at P. is imaged through the cavity with unit magnification. The conjugated points are both outside, providing 6.48 mm clearance beyond the mirror's back surface. The minimum optical diameters are 6.70 mm for the bore and 35.0 mm for the outside diameter of the mirrors. The hollow cathode is imaged by an asymmetric biconvex quartz lens at the intermediate image position P,. It is designed for the same spectral range with the design center at 330 nm. The sphero- chromatic aberrations have been minimized but are not suppressed in a simple lens. The aberrations of the Mangin cell are only 2 percent of the transfer lens blur. The hollow cathode image is demagnified (1 : 0.635) and is inverse at position P. and upright at position P. 279 Fig. 20. — On-axis triple-pass system from figures 18 and 19 shown in actual use. If five passes through the flame are required, both Mangin mirrors may be exchanged for two others of corresponding design, while retaining all previous dimensions. The focal length, f, of the mirrors for the five-pass cell will be, in first order approxi- mation : f = 2.618 x d — 398.98 mm for required clearance of d = 152 .4 mm between the mirrors. The required diameter will be approximately 30 percent larger. The triple-pass Mangin mirror assembly is placed on the optical bench as illustrated in figure 20, with the analytical flame located symmetrically between the mirrors and at a height selected for best sensi- tivity. The system was tested in the atomic absorp- tion mode using Mg. An increased sensitivity of 2.7 was observed when measurements were compared with the sensitivity obtained under the same condi- tions, but without the triple-pass unit. The triple-pass unit was also tested in emission using sodium. When the rear Mangin mirror was blocked out, an increase in the emission signal of 3.3 percent was observed when compared with the signal obtained without the triple-pass unit. When the front Mangin mirror was blocked out, the increase was 43 percent; when both Mangin mirrors were unobstructed, the increase was 70 percent. Finally, when a flat mirror was placed at the opening of the rear Mangin mirror, the increase in the sodium signal was 158 percent. The increased sensitivity of flame measurements, using the on-axis Mangin triple-pass mirror system, recommends its use. This system requires careful optical alignment and is somewhat wavelength depen- dent. This dependence can be eliminated by using reflecting rather than refractive optics in place of the 2 lenses mentioned above. 3. — Discussion of some fundamental excitation conditions in combustion flames. 3.1. Introduction. As a consequence of exothermic combustion reac- tions, the gases which constitute a flame attain different temperatures, according to combustion conditions, flame structure, and the geometry of the 280 system. The thermal energy in a flame is distributed between particles activated to the various transla- tional energies, and rotational, vibrational, electronic and ionization states. Each state can be defined separately by its own temperature. When this value is identical for all states, the radiation from the flame is in thermal equilibrium. Boltzmann's well- known equation ( 11 ), which relates the intensity radiated by a given number of activated particles to the temperature of the excited source, can be applied only in this state of equilibrium. A further limitation to its use is presented by chemical reac- tions and ionization and self-absorption phenomena occuring in the excitation source. When the tempe- ratures of translation, rotation and vibration are not equal and differ from the electronic and ionization temperatures, the system is not in thermal equi- librium. In a flame, various reactions produce radia- tions which deviate from thermal equilibrium. We call this process overexcitation. The Boltzmann equation cannot be applied to interpret such emis- sions, and the notion of temperature loses its meaning. Thermal excitation and overexcitation are thus two widely different types of excitation mecha- nisms which are both found in flames used in ana- lytical flame spectroscopy, and a study of the tempe- rature distribution related to the flame structure will help to identify the origin of the resulting radia- tions. Temperature measurements performed in the outer cone of stoichiometric premixed combustion flames using various procedures and states of excitation (vibration, rotation, and electronic), have produced results in agreement among themselves within the experimental errors and agree also with the thermo- dynamically calculated value. When similar measure- ments were made in the inner cone of the same flames, the temperatures differed widely depending to the measuring conditions. From the results, it can be concluded that thermal equilibrium is achieved in the outer cone of premixed flames and the radia- tion from this zone is of thermal nature (perhaps with the exception of ionization). On the other hand, thermal equilibrium is not reached in the inner cone; the radiations of excited particles in this zone result from various reactions. It is necessary to mention, however, that whether the radiation of a chemical species excited in a flame is or is not in thermal equilibrium does not preclude its use for analytical purposes. The only condition to be satisfied is that the radiation must be constant and reproducible. The mechanism responsible for overexcitation was discussed in 1951 by Laidler and Shuler (see ref. 12) who studied the elementary reactions and collision processes involving excited electronic states and then applied the results to various reactions occuring in combustion flames. Some of the reactions listed by Shuler are : Collisional activation A + B -*A +B* I Energy transfer A* + B -> A + B* II True chemiluminescence A + B -> AB* III A+B + M -» AB* + M III' Induced chemiluminescence A + B + C -> AB + C* IV True chemiluminescence AB + C -» A + BC* V True chemiluminescence AB + C -> A* + BC VI True chemiluminescence AB + M -> A* + B + M VII Induced chemiluminescence AB + C -> A+B + C* VIII The A, -B, C, and M may be radicals or molecules as well as atoms. Equations I and II are energy- transfer processes; in the first, the excitation energy of one of the colliding particles comes from trans- lational energy of the other particle. Equations III, III', and IV are associated reactions; for III' and IV, a three-body collision occurs; in III' the third body, M, acts only in stabilizing the product, whereas in IV, it absorbs the energy released by the formation AB. Equations V and VI are meta- thetical reactions; part of a molecule or a radical is transferred to the colliding particle, and the released energy may be absorbed by the new species (V) or by the remaining part of the dissociated molecule (VI). Equations VII and VIII are disso- ciation reactions in which the colliding particle splits the molecule AB into two parts; the energy released by the process, if any, is absorbed by a part of the dissociated molecule (VII) or by the colliding par- ticle (VIII) — this latter reaction being very un- likely, due to energy requirements. Equations III, III', V, VI, and VII may be termed chemiluminescent since the excited product results from the reaction. Equations I and II enter the category of energy transfers. In equations IV and VIII, the excited partner is a product of the colli- 281 sion, but is not chemically modified; it is acting only to dissipate the energy made available by the process. Such processes may be termed « induced chemiluminescence ». Distinctions between true chemiluminescence, induced chemiluminescence, and energy transfer processes can be made only when the process is actually known. To distinguish nonthermal emis- sions from thermal emissions, but without empha- sizing the actual process, it is proposed to designate nonthermal emissions as « overemission » and their excitation as « overexcitation » ( 12 ). The preferred area of overexcitation is the primary reaction zone or blue cone of premixed flames. Spectroscopic studies have shown that a number of chemical species can be excited to emit specific radiations in this particular zone. This is illustrated in table IV where it can be seen that elements such as Sn, Cd, Ni, Sb, Be, As, Pt, Co, Au, Pd, C, Fe, Ir, Hg, In, Ru, Pb, Bi, Ga. Tl, Mg, are excited preferentially in the blue cone of the premixed acetylene-air or acetylene-oxygen flames with inten- sities varying from very, very weak (vvV) to strong (S) ( 16 ). Most of these emissions originate from an overexcitation reaction and can be used for ana- lytical measurements to detect the particular species and to determine its concentration ( 17 ). The phenomenon of overexcitation was observed many years ago. George-Louis Gouy was the first to describe the differences in radiations which origi- nate from the blue cone and the outer cone of a premixed city gas-air flame when various chemical species were supplied to the flame ( 13 ). Detailed studies of this phenomenon were carried out by De Watteville and Hemsalech who, at an early date, made a clear discrimination between thermal and nonthermal excitation of additives in combustion flames. In 1917 Hemsalech wrote : « The light radiations from vapors in outer mantles of the various flames behave as if governed by the temperatures of these flames. The spectrum emitted by metal vapours in the region of the blue cone of the air-coal gas flame cannot be accounted for by simple thermal actions, but is probably due to some special chemical process prevailing in the explosion region. In the case of iron, we established the existence of two distinct groups of lines. One group, some 200 lines, is emitted by the outer mantle of the air- coal gas flame, and the lines merely increase in intensity on passing to flames of higher temperature. We considered them as the fundamental lines of the spectrum, which are easily excited by thermal actions. The other group of over 500 lines, is emitted by the region of the blue cone only, and even in the oxy-acetylene flame we could detect only traces of some of the lines ( 18 ) ». TABLE IV Intensities of Spectrum Lines Rradiated by the Atoms of Various Elements Excited in the Inner Cone (IC) and Outer Cone (OC) of the Premixed Acetylene-Air and Premixed Acetylene-Oxygen Flames. Excita- Emission observed in tion Energy Wavelength (A) Element C 2 H 2 — 2 C 2 H 2 — Air (eV) IC OC IC OC 5.84 . . . 2 199.3 Sn vW 6.03 . . . 2 209.7 Sn vW 5.18 . . . 2 246.1 Sn wM 5.89 . . . 2 268.9 Sn wM 5.84 . . . 2 286.7 Sn vW 5.42 . . . 2 288.0 Cd vW 6.77 . . . 2 288.1 As W 5.41 .. . 2 290.0 Ni vvW 5.36 . . . 2 311.0 Ni W 5.36 . . . 2 311.5 Sb wM M 5.52 . . . 2 312.3 Ni W 5.63 . . . 2 314.0 Ni W 6.42 . . . 2 317.2 Sn vW 5.51 . . . 2 317.2 Ni W 5.34 . . . 2 320.0 Ni W 5.61 . . . 2 321.4 Ni W 5.49 . . . 2 325.8 Ni W 5.59 . . . 2 330.0 Ni vW 5.52 . . . 2 334.8 Sn wM 5.28 . . . 2 345.5 Ni vW 5.28 . . . 2 348.6 Be vvW 6.59 . . . 2 349.8 As wM 5.47 . . . 2 354.8 Sn W M 5.35 . . . 2 357.1 Pt W 5.15 . .. 2 407.3 Co vvW 5.24 . . . 2 411.6 Co vvW 5.31 ... 2 414.5 Co vvW 5.36 . . . 2 415.3 Co vvW 6.19 . . . 2 421.7 Sn wM 5.11 ... 2 424.9 Co vvW 5.10 . . . 2 428.0 Au vW vvW W 5.23 . . . 2 429.5 Sn W M 5.20 . . . 2 432.2 Co vvW 5.26 . . . 2 436.7 Co vvW 5.18 . . . 2 436.7 Pt vW 5.31 ... 2 439.0 Co vvW 5.08 . . . 2 440.1 Pt wM 5.06 . . . 2 447.9 Pd vvW vvW 5.02 . . . 2 467.4 Pt W 5.00 . . . 2 476.4 Pd vvW vvW 7.68 . . . 2 478.6 C M 4.99 . . . 2 483.3 Fe vvW vW vvW 5.42 . . . 2 483.4 Sn wM 4.98 . . . 2 487.2 Pt M 5.03 . . . 2 488.1 Fe vvW vW vvW 5.10 . . . 2 489.8 Fe vvW 5.08 . . . 2 490.1 Pt vW 5.06 . . . 2 490.6 Fe vW 6.28 . . . 2 492.9 As vvW 6.03 . . . 2 495.7 Sn W 5.78 . . . 2 495.8 Pt vvW 282 Excita- tion Energy (eV) Wavelength (A) Element 2 498.5 2 502.6 2 503.0 2 511.0 2 515.6 2 521.4 2 522.8 2 527.4 2 528.5 2 529.0 2 536.0 2 536.5 2 539.2 2 544.0 2 544.3 2 544.9 2 546.6 2 560.2 2 571.6 2 594.4 2 598.1 2 611.3 2 612.1 2 613.7 2 614.2 2 627.9 2 628.0 2 631.3 2 639.3 2 639.7 2 641.5 2 643.0 2 646.9 2 650.9 2 651.3 2 651.8 2 652.6 2 659.0 2 659.4 2 659.9 2 661.2 2 662.0 2 663.2 2 664.8 2 670.6 2 676.0 2 677.1 2 694.2 2 696.8 2 698.4 2 702.4 2 705.9 2 706.5 2 709.2 2 710.3 2 718.5 2 718.9 2 719.0 2 719.0 Pt Ir Ir Co Pt Co Fe Fe Sb Co Co Hg Pt Ir Co Co Sn In Sn Sn Sb Ir Ru Pb Pb Bi Pt Ru Pt Ir Au Ru Pt Pt Ru Ru Sb Rh Pt Ga Sn Ir Pb Ir Sb Au Pt Ir Bi Pt Pt Pt Sn Ru In Rh Sb Pt Fe Emission observed in C 2 H, IC OC vvW vvW I vW vvW vvW S v\V ! vvW W vvW v\V vW W M vvW vvW M vW wM mS W w w M C,H, Air IC W vw ] vW I vvW I W vW WW ; vvW vS vvW | vvW S vW w vvW vvW M vvW wM W vS vW vvW M M W mS vvW vW wM vvW vvW mS mS vvW vvW vvW vvW S vvW M vW W wM W W wM wM vvW W mS mS S W vW vvW W mS vW OC Excita- tion Energy Emission observed in Wavelength (A) Element CH, -o, C,H., — Air (eV) IC vvW OC IC OC 4.61 2 720.9 Fe w vv\Y 4.64 . . 2 723.6 Fe vvW v\V 5.35 . . 2 729.9 Pt v\Y 4.63 . . 2 734.0 Pt v\Y w mS 4.53 . . 2 735.7 Ru M 4.64 . . 2 737.3 Fe vW 4.61 . . 2 742.4 Fe w vvW 6.77 . . 2 745.0 As vW 5.65 . . 2 748.3 Au vvW S 4.50 . . 2 753.9 In vvW 5.76 . . 2 753.9 Pt vvW 5.75 .. 2 754.9 Pt w 4.91 . . 2 761.8 Sn vv\V 4.49 . . 2 763.1 Pd w vW 4.63 . . 2 763.4 Ru vvW 4.48 . . 2 767.9 Tl wM M wM vvW 5.70 . . 2 769.9 Sb W mS 4.57 .. 2 771.7 Pt vvW M 7.17 . . 2 779.8 Mg vvW 5.53 . . 2 779.8 Sn wM 6.77 . . 2 780.2 As wM 5.87 . . 2 780.5 Bi vW 5.52 . . 2 785.0 Sn W 5.30 . . 2 788.1 Fe vvW vvW i A medium which is suitable to the production and observation of overexcitation phenomena is hydrogen burning in air or oxygen under various conditions and especially with an excess of fuel. The overexcitation is greatly enhanced when the hydrogen flame is cooled by water. This observation was made more than 100 years ago by Salet who studied the spectra produced by nonmetals in a hydrogen-air flame chilled with water ( 19 ). Salet studied phos- phorus, sulfur, halogens, tin, selenium, tellurium and silicon. Sulfur and phosphorus seem to have retained his particular interest. The pioneering works of Salet were forgotten for a long time, and it was only during the last 15 years that spectroscopists have shown interest in the use of hydrogen flames to excite nonmetals. At the present time, the literature cites new instrumen- tation and many applications of this flame to solve particular analytical problems. A recent compre- hensive and authoritative review of these works, covering over 200 pages and containing 602 refe- rences, together with a complete discussion of the original works, was made by Gilbert ( 20 ). The non- metals examined in his review are H, B, C, SL N, P, As, 0, S, Se, Te, F, CI, Br, and I. We have selected to discuss phosphorus and sulfur due to the parti- 283 cular properties of the two elements. They are easily excited in chilled hydrogen flame and radiate beau- tiful characteristic molecular spectra; they are also important to environmental pollution problems. Since the excitation of P and S in various hydrogen flames was discussed by Gilbert in the above review ( 20 ), we will mention here the original works of Salet ( 19 ) together with only a very few essential and more recent contributions. A complete survey of the field is given in Gilbert's review. 3.2. Experimental Arrangements to Excite Phosphorus and Sulfur, and Discussion. The experimental arrangement used by Salet in 1869 and 1871 to excite P and S is illustrated in figure 21. It consists of a hydrogen flame which is JjffP Fig. 21. — Arrangement used by Salet to produce a water chilled hydrogen-air flame. produced at the extremity of a glass tube. The flame is chilled by directing it against a water curtain. The hydrogen is supplied with the vapor by passing the gas over the surface of a P or S compound contained in a closed vessel which can be warmed to produce the necessary vapor. Salet describes the blue and green radiations obtained from S and P compounds, and mentions the high sensitivities obtained in detecting these two elements. These observations were used in a patent involving the detection and determination of P and S com- pounds in gaseous matrices, by the Dragerwerk Company in 1961 ( 21 ). The hydrogen flame burns with an excess of fuel and is produced inside a quartz tube cooled with water. The excess hydrogen burns at the extremity of this tube where it comes in contact with the surrounding air. This hydrogen burner is illustrated in figure 22. The detection OUTER HYDROGEN FLAME NOZZLE RADIATION FROM P or S INNER HYDROGEN FLAME DRAIN SILICA TUBE RADIATION FROM P or S ]— ► WATER WATER CONDENSER ]«— WATER HYDROGEN 0.1-0.4 l/mln. AIR CONTAINING - Sor P VAPORS 0.3-0.6 l/mln. Fig. 22. — Drager burner designed to produce a dual hydro- gen-air flame. The inner diffusion flame is water chilled and burns with an excess of hydrogen; the outer flame burns by diffusion with the surrounding air ( M ). Courtesy Plenum Press. limits of P or S obtained with this burner were 10~ 8 g of the compound per liter of air (< 0.01 ppm). The water-cooled hydrogen flame of Salet was used in our exploratory work combining the Drager- werk burner with a sample supplying system similar to that described long ago by Gouy ( 13 ). The chilled hydrogen flame can be supplied with solution samples by pneumatic spraying. Argon was used for this purpose. The experimental arrangement is described in figure 23 and a corresponding drawing showing some construction details is illustrated in figure 24. The horizontal spraying vessel of Teflon is provided at left with an external concentric pneumatic sprayer made of tantalum. The sample solution enters the sprayer through the polyethylene capillary tube. Its supply rate can be varied between and 5 ml/mn through the needle valve and its graduated dial. The argon is introduced into the sprayer through the side tube. 284 Fig. 23. — Experimental arrangement used to produce a dual hydrogen flame supplied through pneumatic spraying of the analytical solution. The horizontal Teflon spraying vessel with the tantalum external variable pneumatic sprayer is located on the optical bench. The vertical all quartz burner enclosure is adapted to the spraying vessel through an air tight rubber seal. OUTER FLAME — kti— ■12 00-30 10-23 00-42 ID-38 INNER FLAME WATER IN DRAIN BLOW-OUT ^MEMBRANE SPRAYER Fig. 24. Drawing of the unit illustrated in figure 23; dimensions in mm. The sprayer vessel has a draining tube connected to a hydraulic trap (see fig. 12), and at the right side a rupture disk made of a thin sheet of Teflon. The polyethylene tube, seen at the right side of the spraying vessel, is used to provide oxygen to the argon-sample spray. It is adapted to the tantalum nozzle which extends inside the vessel through a fitting. The oxygen nozzle has a circular opening 0.5 mm in diameter at its extremity. On top of the spraying vessel is a slot-shaped tantalum tube through which the argon, oxygen, and sample spray flows. This tube is surrounded by an all-quartz cylindrical vessel open at both ends and provided with a water cooling jacket. This quartz vessel is adapted to the spraying vessel through a rubber washer which provides an air-tight assembly. The lower side tube at right is used to drain intermittently the water which condenses on the inside walls when the flame burns inside the quartz vessel. The side tube on the left is used to supply the hydrogen which is metered by the flowmeter at left. The entire unit is placed on the optical bench in front of the spectrometer described in Section 2 of this paper, and can be moved in the x-y-z direction as required and secured in the chosen position. Discussion of of similar works will be found in reference ( 20 ). The hydrogen-argon-oxygen flame is produced in the following manner : The cooling water is turned on and the argon is supplied first to the sprayer at a pressure of 40 psi and at a rate of 300 1/h. Hydrogen is then introduced into the cylindrical quartz vessel and ignited at the upper end of the vessel where it burns by diffusing with the surrounding air. Oxygen is added progres- sively until the upper hydrogen-air flame flashes back inside the cylindrical quartz vessel and the hydrogen burns with oxygen by diffusion. Sufficient hydrogen is added to supply and maintain both the inner flame and the outer flame. Under these conditions two flames are produced : an inner chilled H 2 — 2 — Ar diffusion flame which burns with an excess of fuel, hence a reducing atmosphere, and an outer diffusion hydrogen flame which burns with the surrounding air, and hence with an excess of oxidant. The sample solution is supplied as a fine spray to both flames through the pneumatic sprayer. Argon was chosen to operate the sprayer since it has a low quenching cross section. This property should emphasize overexcitation reactions. Sulphur and phosphorus were supplied to the flame as sulfuric and phosphoric acids, ammonium sulfate, and ammonium phosphate, as well as organic aqueous solutions of dimethyl sulfoxide and hex- methyl phosphoramide. The hydrogen flames produced under these cir- cumstances are illustrated in figure 25 a, b, and c. Figure 25a shows the appearance of the two flames 285 Fig. 26. WAVELENGTH, A Monochromator recording of the inner hydrogen flame supplied with diluted sulfuric acid. when only distilled water is supplied to the sprayer; figure 25b shows the flames supplied with the sulfur containing solution, while figure 25c shows the flames supplied with the phosphorus containing solu- tions. The same types of radiations are obtained with the inorganic or organic bound sulfur or phos- phorus. Concentrations of these elements was about 1 percent. The detection limits for both elements are below 0.1 ppm. The radiations from the inner chilled hydrogen flame supplied with sulphur and phosphorus com- pounds were examined with the spectrometer in the scanning mode using a grating with 1 200 lines/mm, a scanning speed of 200 A/mn, and a slit of 0.200 mm (1.6 A). The recordings of the spectra obtained are given in figure 26 for the sulphur flame and figure 27 for phosphorus flame. As can be recognized from data available in the literature ( 20 ), these radiations originate from the S 2 and HPO molecules. The types and intensities of the radiations excited in the chilled hydrogen flame with an excess of hydrogen depend strongly on the experimental conditions used to produce them. The burner system described in this work is only one of many possible arrangements which can be used to produce these kinds of excitation conditions. Complete and up-to- date information on this subject can be found in reference ( 20 ). 500 4750 5000 5250 5500 5750 60C WAVELENGTH. A Fig. 27. — Monochromator recordii ^ of the inner hydrogen flame supplied with diluted hosphoric acid. Both S 2 and HPO emissions are characteristic examples of the nonthermal excitation process which we have called overexcitation. Indeed, the thermal energy available in the chilled hydrogen flame is certainly not capable of exciting species such as HPO and S 2 which require about 2.4 and 3.9 eV respectively (*). Various potential reactions have been postulated to account for these non- thermal emissions. Thus, for HPO, Lam Thanh and Peyron considered both reaction ( 22 ). and H + PO + M H + PO* HPO* HPO*, M, (HI') (III) while Gilbert proposed the chemiluminescent reac- tion ( 20 ), PO + H 2 + OH HPO* + H 2 0. (Ill') For S 2 , Crone had suggested the chemiluminescent reaction ( 20 ), H + H + S 2 S 2 * + H 2 , (IV) which was also postulated by Peyron. Sugden and Demerdache proposed another chemiluminescent reaction ( 23 ), H + OH S 2 * + H 2 0. (IV) At this stage it seems that further work is needed to elucidate these flame reactions and to arrive at an accurate interpretation of the excitation pheno- mena occuring in chilled hydrogen flames burning in an excess of fuel. Acknowledgment. The author is particularly grateful to Dr. 0. Menis, Section Chief of the Coordination Chemistry Section, (*) Communication from Dr. H. Okate, Physical Chemistry Division, National Bureau of Standards. 286 Analytical Chemistry Division, for approving and encouraging this work, and to Mr. J. I. Shultz, Assistant Chief of the Coordination Chemistry Section, Analytical Chemistry Division, for critically reviewing the manuscript. BIBLIOGRAPHY (1) G. Kirchhoff and R. Bun-sen, J. Prakt. Chem., I860, 80, 449-477; Ann. Physik, 1860, 110, 161-189; ibid., 1861, 113, 337-381; Phil. Mag., 1860, 20, 89-109; ibid., 1861, 22, 329-349; Ann. Chim. el Phys., 1861, 113, 337- 381. (2) P. Champion, H. Pellett and M. Grenier, Compt. Rend., 1873, 76, 707-711. (3) J. Janssen, Compt. Rend., 1870, 71, 626-629. (4) H. Lundegardh, The Quantitative Spectral Analysis of the Elements, vol. 1 and 2, G. Fischer, Jena, 1929 and 1934. (5) J. Tyndall, Six Lectures on Light, 1872-1873, Appletoh, New York, 1883. (6) A. Walsh, Australian Patent, 163, 586, Application, 1953, Issued, 1955. (7) R. Mavrodineanu, National Bureau of Standards Tech- nical, Note 544, 1970, 58-70, O. Menis, J. Shultz, ed. (8) G. Ziedens, Electronic Readout System for Multichannel Flame Spectroscopy, Technical Report 228, Oct. 31, 1968, Philips Laboratories, BriarclifT Manor, New York, 10510. (9) R. Mavrodineanu and R. C. Hughes, Applied Optics, 1968, 7, 1281-1285. (10) R. Mavrodineanu, Developments in Applied Spectros- copy, 1970, 8, 18-32. (11) R. Mavrodineanu and H. Boiteux, L' Analyse Speclrale Quantitative par la Flamme, Masson, Paris, 1964. (12) R. Mavrodineanu and H. Boiteaux, Flame Spectros- copy, Wiley, New York, 1965. (13) G. L. Gouy, Ann. Chim. et Phys.. 1879, 18, 5-101. (14) B. J. Russell, J. P. Shelton and A. Walsh, Spectro- chim. Acta, 1957, 8, 317-328. (15) W. Muller Herget, Some Considerations on Optical Design and Selection of Spectroscopic Instruments in « Analytical Flame Spectroscopy, Selected Topics », R. Mavrodineanu, ed. McMillan, London, Springer, New York, 1970. (16) R. Mavrodineanu, Developments in Applied Spectros- copy, 1965, 5, 371-401. (17) P. T, Gilbert, Jr. Proc. Colloq. Spectros. Intern, 10th, Univ. Maryland, 1963, 1962, 171-215. (18) G. A. Hemsalech, Phil. Mag., 1917, 33, 1-18. (19) G. Salet, Compt. Rend., 1869, 68, 404-406; ibid., 1871, 73, 742-745; ibid., 1871, 73, 862-863; ibid., 1871, 73, 1056-1059; ibid., 1890, 110, 282-283; Bull. Soc. chim., France, 1869, 11, 302-304; ibid., 1890, 3, 328-329; Ann. Physic, 1869, 137, 171-174; Ann. Chim., et Phys. 1873, 28, 5-71. (20) P. T. Gilbert, Jr. Nonmetals, in « Analytical Flame Spectroscopy, Selected Topics », R. Mavrodineanu, ed. McMillan, London; Springer, New York, 1970. (21) Dragerwerk, German Pat., 1962, 1, 133, 918; Recorded, 1961; and U.S. Patent, 1965, 3, 213, 747 to C. E. van der Smissen. (22) M. Lam Thanh and M. Peyron, J. Chim. Phys., 1964, 61, 1531-1546. (23) T. ML Sugden and A. Demerdache, Mature, 1962, 195, 516. 287 ■&U.S. GOVERNMENT PRINTING OFFICE: 1977 O— 248-270 NBS-114A (REV. 7-73) U.S. DEPT. OF COMM. BIBLIOGRAPHIC DATA SHEET 1. PUBLICATION OR REPORT NO. NBS SP-492 2. Gov't Accession No. 3. Recipient's Accession No. 4. TITLE AND SUBTITLE PROCEDURES USED AT THE NATIONAL BUREAU OF STANDARDS TO DETERMINE SELECTED TRACE ELEMENTS IN BIOLOGICAL AND BOTANICAL MATERIALS 5. Publication Date November 1977 6. Performing Organization Code 7. AUTHOR(S) 8. Performing Organ. Report No. R. Mavrodineanu, Editor 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) Environmental Protection Agency, Washington, D.C. and partially supported by the Energy Research and Development Administration, Washington, P.O. 2Q545 20460 13. Type of Report & Period Covered 14. Sponsoring Agency Code 15. SUPPLEMENTARY NOTES Library of Congress Catalog Card Number: 77-608301 16. ABSTRACT (A 200-word or less factual summary of most significant information. If document includes a significant bibliography or literature survey, mention it here.) This volume consists of thirteen papers describing the analytical procedures selected at the National Bureau of Standards (NBS) for the determination of Ag, Al, As, Be, Bi, Ca, Cd, Cr, Cu, Fe, Hg, K, Mg, Mn, Mo, Na, Ni, Pb, Pt, Sb, Se, Te, Tl, V, and Zn in biological and botanical materials. These procedures are used at the present time for the certification of various substances issued by NBS as Standard Reference Materials, and belong to the following analytical disciplines: sample preparation, neutron activation analysis, spark source mass spectrometric isotope dilution, atomic absorption and flame emission spectrometry, molecular absorption spectrometry, fluorescence spectrometry, and polarography. Further details on the analytical methods including sample preparation, purity of reagents, and problems associated with blanks are given in sixteen additional papers which are reproduced in the Appendix to this volume. 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) Analytical disciplines; analytical procedures; atomic absorption spectrometry; biological materials; blanks; botanical materials; chemical species; flame emission spectrometry; fluorescence spectrometry; molecular absorption spectrometry; neutron activation analysis; polarography; pure reagents; sample preparation; spark source mass spectrometric isotope dilution; tissues. 18. AVAILABILITY [^ Unlimited j | For Official Distribution. Do Not Release to NTIS |X 1 Order From Sup. of Doc, U.S. Government Printine Office Washington, D.C. 20402, SD Cat. No. CI 3. 10:492 I 12 Order From National Technical Information Service (NTIS) Springfield, Virginia 22151 19. SECURITY CLASS (THIS REPORT) 21. NO. OF PAGES UNCLASSIFIED 295 20. SECURITY CLASS (THIS PAGE) 22. Price UNCLASSIFIED $7.50 USCOMM-DC 29042-P74 There's anew look 1 *-X? I / • • • the monthly / >^L ^^\, / >/ magazine of the Nation- ' .^V^^^^^^^^/ a ' Bureau Q f Standards. [ ^^^, / Still featured are special ar- ^^^^^^^T/ tides of general interest on ^^^^^^j / current topics such as consum- N^^v^ / er product safety and building *™^ technology. In addition, new sec- tions are designed to . . . PROVIDE SCIENTISTS with illustrated discussions of recent technical developments and work in progress . . . INFORM INDUSTRIAL MANAGERS of technology transfer activities in Federal and private labs. . . DESCRIBE TO MAN- UFACTURERS advances in the field of voluntary and mandatory standards. The new DIMENSIONS/NBS also carries complete listings of upcoming conferences to be held at NBS and reports on all the latest NBS publications, with information on how to order. Finally, each issue carries a page of News Briefs, aimed at keeping scientist and consum- alike up to date on major developments at the Nation's physi- cal sciences and measurement laboratory. (please detach here) SUBSCRIPTION- ORDER FORM Enter my Subscription To DIMENSIONS/NBS at $12.50. Add $3.15 for foreign mailing. No additional postage is required for mailing within the United States or its possessions. Domestic remittances should be made either by postal money order, express money order, or check. Foreign remittances should be made either by international money order, draft on an American bank, or by UNESCO coupons. Send Subscription to: □ Remittance Enclosed (Make checks payable to Superintendent of Documents) O Charge to my Deposit Account No. NAME-FIRST, LAST II I I I I I I I I I I I I I I COMPANY NAME OR ADDITIONAL ADDRESS LINE 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i i i i STREET ADDRESS 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ! 1 1 1 1 1 1 1 CITY 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 STATE 1 _LI ZIP CODE MM MAIL ORDER FORM TO: Superintendent of Documents Government Printing Office Washington, D.C. 20402 PLEASE PRINT TA6& 2-V9 (mo foL