. V^-^fA h- '."AW-3-^ NBS-GCR-ETIP 76-22 Development of Flame Retardants for Polyester/ Cotton Blends Robert H. Barker and Michael J. Drews Department of Textiles Clemson University Clemson, South Carolina 29631 Final Report September 1976 Prepared for Experimental Technology Incentives Program National Bureau of Standards Washington, D. C. 20234 f£TiP> EXPERIMENTAL TECHNOLOGY INCENTIVES PROGRAM (ETIP) The Experimental Technology Incentives Program was initiated in fiscal year 1973 as part of the President's program to learn how the Government could stimulate technological innovation. The objective of the program is to learn how the Federal Government can provide policies and incentives which will encourage greater technological innovation by the private sector. Broader applica- tion of innovative technology could lead to the amelioration of several national problems such as a low rate of increase in pro- ductivity, negative trade balances, environmental pollution, and public health and safety. The interrelation of the Government and private sector is complex and not enough is known to predict the effect on technological innovation of a change in government policy. Consequently, various hypotheses regarding possible federal policy are being tested with analyses and experiments. ETIP has identified four policy-related program areas for inves- tigation and experimentation. These areas relate to procurement practices at all three governmental levels, regulatory practices at the federal and state levels, federal practices for the funding of civilian R6D, and federal economic assistance practices. In each of these areas, new or modified governmental policies will be tested in cooperation with the agencies responsible for implementation . In addition, ETIP will be evaluating the impact of its experi- mentation on the cooperating agencies and on the commercial sector. These evaluations should result in a body of knowledge that encompasses the practices tested, any barriers encountered, effectiveness of use and potential for adoption. The accompanying report was prepared under contract as part of the ETIP program of the National Bureau of Standards. Statements contained in this document represent the views of the originating organization and do not necessarily reflect those of the National Bureau of Standards. Director Experimental Technology Incentives Program National Bureau of Standards U. S. Department of Commerce NBS-GCR-ETIP 76-22 DEVELOPMENT OF FLAME RETARDANTS FOR POLYESTER/ COTTON BLENDS Robert H. Barker and Michael J. Drews Department of Textiles Clemson University Clemson, South Carolina 29631 Final Report Contract 4-35963 September 1976 Prepared for Experimental Technology Incentives Program a <3 o a. a> Q -> • fc »r,u o« * U.S. DEPARTMENT OF COMMERCE, Elliot L. Richardson, Secretary Edward O. Vetter, Under Secretary Dr. Betsy Ancker-Johnson, Assistant Secretary for Science and Technology NATIONAL BUREAU OF STANDARDS, Ernest Ambler, Acting Director ACKNOWLEDGMENT The authors wish to acknowledge the contributions to this report which were made by all of the members of the project steering committee and by the entire staff of the Textile Department of Clemson Univer- sity. Special thanks are due to Dr. C. Jarvis for help with writing, proofreading and indexing, P. Marion for the art work, and to T. Lawler, J. Langley, J. Watson, B. Childress, S. Amujiogu, J. McCollum, W. Stuckey, D. Handlin and S. Sewell for much of the technical and editor- ial work. Of course, none of this would have produced anything without the tireless efforts of L. Cobb and B. Camas, the two secretaries who handled all of the paper work and kept the project running for two years. / TABLE OF CONTENTS PAGE INTRODUCTION 1 PROJECT OBJECTIVES 2 ORGANIZATION OF THE CONSORTIUM 3 PROJECT ADMINISTRATION 6 OPERATION OF THE CONSORTIUM 7 INTERACTION WITH OTHER RESEARCH GROUPS 23 BACKGROUND INVESTIGATIONS 26 RESEARCH STRATEGY 27 STATE OF THE ART AS OF JUNE, 1974 . 28 1. Theory of Flame Retardant Action 33 2. Development of Potentially Commercial Flame Retardants 65 2a. Blending of Fire Resistant Fibers 68 2b. Chemical Aftertreatment of PET/ Cotton Blend Fabrics 70 TESTING METHODOLOGY 81 1. Oxygen Index and 45° Angle Burning Tests 82 2. Calori metric Measurements 84 3. Correlation of Test Methods 104 POLYESTER 120 1. Basic Factors Affecting PET Flammability 120 la. Pyrolysis of Untreated PET 120 lb. Effect of Bromine on PET Flammability 127 1c. Effect of Phosphorus on PET Flammability 143 Id. Role of Condensed Phase Oxidation in PET Flammability 149 2. Flame Retardant Polyester 171 2a. Inherently Flame Retardant PET 171 2b. Grafting Studies 174 STUDIES ON POLYESTER/ COTTON BLENDS 214 FLAME RETARDANT SYSTEMS BASED ON PHOSPHORUS ALONE 215 1. AntiblazeW 19 215 2. N-methylol -3- (di phenyl phosphinyl ) propionamide 219 3. Other Organophosphorus Retardants 242 4. Mixtures and Precondensates of Phosphorus Retardants 246 FLAME RETARDANT SYSTEMS BASED ON BROMINE ALONE 304 1. Calori metric Evaluation-of FR P-44VV 304 2. Application of FR P-44vV with Durable Press Resins 312 3. Application of FR P-44vJ!/ with a Bromine- Containing Latex 312 4. Application of Other Bromine-Containing Flame Retardants 322 FLAME RETARDANT SYSTEMS BASED ON COMBINATIONS OF PHOSPHORUS AND BROMINE 328 1. Optimization of Phosphorus-Bromine Formulations 328 2. Interaction of Flame Retardant Fibers 335 3. Phosphorus-Bromine Systems Based on Phosphonium_Salts 356 3a. THPSv9/Urea/PVBr and Related Systems 356 3b. Determination of Flame Retardant Efficiencies of Bromine Compounds 366 3c. Utilization of Other Bromine Compounds with Phosphonium Salts 370 4. Systems Based on a Bromine-Containing Phosphazene 385 5. Systems Amenable to Fixation by Irradiation 393 SUMMARY AND CONCLUSIONS 426 BIBLIOGRAPHY 436 INDEX 440 li GLOSSARY OF CHEMICAL ABBREVIATIONS BABA 50 BDPOM DAP DAVP DBDPO DBPA DBPM DEVP DMAP DMVP DPA DVP H 6 or "Hexa" MDMP MD3P NDPA OBBP PCHDT PCPM P(DBPA) bis-acrylate of 2-hydroxyethyl ether of tetrabromobi sphenol -A bis (2,3-dibromopropyl )phosphonyl-2-oxyethyl methacrylate di ammonium phosphate dimethyl 1-acetoxyvinyl phosphonate decabromodi phenyl ene oxide 2,3-dibromopropyl acrylate 2,3-dibromopropyl methacrylate diethyl vinyl phosphonate dimethyl allyl phosphonate dimethyl 1 -methoxyvinyl phosphonate dimethyl phosphonomethyl acrylate dimethyl vinyl phosphonate l-(l,2,3,4,7,7-hexachlorobicylco[2,2,l]- 2-hepten-5-yl )-ethene N-methylol-3-(dimethyl phosphinyl )propionamide N-methylol-3-(di phenyl phosphinyl )propionamide N- (dimethyl phosphonomethyl )acryl amide octabromobi phenyl poly(l ,4-cyclohexylene dimethylene tere- phthalate) pentachlorophenyl methacrylate poly (2,3-dibromopropyl acrylate) in GLOSSARY OF CHEMICAL ABBREVIATIONS (con't) PVBr poly(vinyl bromide) P(VBr/VCl) vinyl bromide/vinyl chloride copolymers TBBPA tetrabromobisphenol A TBPA 2,4,6-tribromophenyl acrylate TBPM 2,4,6-tribromophenyl methacrylate TBPOEA 2,4,6-tribromophenoxy ethyl acrylate TBPOEMA 2,4,6-tribromophenoxy ethyl methacrylate TBPP tris(2,3-dibromopropyl ) phosphate THP trihydroxymethyl phosphine T23P tris(2,3-dibromopropyl ) phosphate TPP triphenylphosphate TPPO triphenyl phosphine oxide VBr^ vinyl idene bromide IV List of Tables Table No. Title Page No I Heat of Combustion of Polyester and Cotton Cellulose II Reaction Rate Constants and Activation Energies of the Thermal Degradation of PET III Comparison of Top and Bottom Oxygen Index Methods IV Comparison of Flammability Test Methods V Calorimetric Parameters of Various Treated Systems VI Isoperibol Results of Polyester/Cotton Blends VII Isoperibol Results of Diammonium Phosphate Treated 50/50 Polyester/Cotton VIII Calorimetric Data of Decabromodi phenyl Oxide Treated Fabrics with Fiber-Glass-Grid Support IX Char Length in cm as a Function of Mode of Ignition and Exposure Time. ETIP 50/50 Polyester/Cotton X Char Length as a Function of Exposure Time for ETIP 50/50 Polyester/Cotton with Several Precondensate Based Finishes XI Calorimetric and MAFT Data for Some Selected Fabrics XII Calorimetric and MAFT Data for F.R. Treated Fabrics XIII Char Length (FF-3) and 01 on Treated 100% Cotton Fabrics XIV Flammability of Fabrics Treated with Fyrol 76 U XV Flammability of Fabrics Treated with Pyroset^TK-115 XVI Flammability of Fabrics Treated with FR P-44^ XVII Flammability of Flame Retardant Treated Polyester Fabrics 118 XVIII Flammability Behavior of Ralvester Single Knits Treated with 15% TanatardCS)oN-2 119 XIX Polyester Samples Prepared for Flammability Studies 121 XX Thermogravimetric Weight Loss and Melting Point of Polyesters 124 XXI Oxygen Index of Polyesters 125 XXII Thermogravimetric Weight Loss Analysis and Oxygen Index of Copolymers and Homopolymer Mixtures 128 XXIII Summary of Linear Regression Analysis of Isoperibol Calorimetry Results 138 49 61 83 85 92 95 98 100 106 108 109 110 114 115 116 117 List of Tables (cont. ) Table No . Title Page No XXIV Hydrogen Bromide Release of Bromine Fire Retardants 140 XXV Thermal Analysis of Bromine Fire Retardants 144 XXVI P Containing Additives and/or Comonomers in PET 147 XXVII Flammability of Diammonium Phosphate Treated Polyester 148 XXVIII Comparison of Rate Constants, k, For Different Configurations of PET in Varying Oxygen Environments at 356 C 150 XXIX The Effect^of Oxygen on the Decomposition Rate, k, of Dacron© 54 Poly(Ethylene Terephthalate) 152 XXX Comparison of the decomposition rate constants, k, for Dacron(^)54 and deep-dyeing PET 154 XXXI Rate Constants for the Decomposition of DacronvJ 900F and the Comparison to Dacron^ 54 156 XXXII Decomposition Rate Constants for Phosphazene-containing PET (PFR) 158 XXXIII Comparison of TGA and ESR Data for PET Decomposition 168 XXXIV Cyclophosphazene Physical Properties 172 XXXV Flame Retardant Efficiency of VBr or PET as a Function of Location 192 XXXVI Monomers Grafted 193 XXXVII Physical Properties of Grafted PET 195 XXXVIII Tenacity and Elongation Vs Copolymer Graft 196 XXXIX Thermogravimetric Analysis of Grafted Polyester 200 XL Typical Data on Grafted PET Fibers 202 XLI Initial Grafting Results on PET Fabric 206 XLII Flame Retardant Efficiency as a Function of Thermal Stability 209 XLI I I Grafts with No Low Temperature Decomposition Point 210 XLIV Maximum Thermal Stability of Polymer Repeat Units 211 XLV Effect of Phosphonium Precondensates on PET Flammability 212 XLVI Flammability of Blends Treated with Antiblaze®19 216 XLVII Isoperibol Calorimetry Results from Antiblaze^l9 Treated 50/50 Polyester/Cotton Blend 218 VI List of Tables (cont.) Table No. Title Page No, XLVIII 100% Cotton Fabric Treated with MD3P (Unfixed) 226 XLIX 50/50 PET/Cotton Treated with MD3P (Unfixed) 227 L 100% Cotton Fabric Treated with MD3P (Fixed) 233 LI 50/50 PET/Cotton Treated with MD3P (Fixed) 234 LI I Isoperibol Results of DAP/Anti blaze® 19 Treated 50/50 Polyester/Cotton Blend 247 LIII Flammability of Blend Fabrics with MCC-1 00/200/300® 251 LIV Burn Data from Fabrics Treated with Phosphorous Combinations 256 31 LV P n.m.r. Chemical Shifts and Electronic Integrals 265 31 LVI p Chemical Shifts of Some Selected Phosphoramides 267 LVII Calorimetric Results From. ETIP 50/50 Blend Fabric Treated with THPC/Urea and (NH 4 ) 2 HP0 4 270 LVI I I Calorimetric Results From the ETIP 50/50 Blend Fabric Treated with Pyrovatex(J)3762 271 LIX Calorimetric Results From the ETIP 50/50 Blend Fabric Treated with Combination Flame Retajxiants THPC and Pyrovatex©CP, THPC and Pyrovatex®3762 272 LX Calorimetric Results From the ETIP 50/50 Blend Fabric Treated with THPC-MCC-1Q0® Oligomer 273 LXI Calorimetric Results From the ETIP 50/50 Blend Fabric Treated with THP0H-MCC-1 00© Oligomer 274 LXI I Calorimetric Results From the ETIP 50/50 Blend Fabric Treated with THPC-CH 3 NH 2 Oligomer 275 LXI 1 1 Calorimetric Results From DuPont 50/50 P/C and 50/50 900F©/Cotton Blend Fabrics with the THPC-CH 3 NH 2 Oligomer 276 LXIV Calorimetric Results->From the EIIP 50/50 Blend Fabric Treated with a THPSVJP (MCC-1 00© ) Urea Precondensate LXV Durability of MCC-1 00® /THPC/UREA Finish on 50/50 Blend 286 LXVI Durability of Oligomeric THPC/MCC-100^Finishes 287 LXVII Pad Bath THPC/MCC-1 00® (Prepared under Non-Oxidizing Atmosphere) 290 VII 283 List of Tables (cont.) Table No, Title Page No LXVIII Pad Rath THPS®/MCC-1 00® (Prepared Under Non- Oxidizing Atmosphere 291 LXIX Pad Bath THPS©/MCC-100® Neutralized 292 LXX Pad Bath THPSC/MCC-1 GOVS' Neutralized 293 LXXI Pad Bath THP0H/HCC-100v^ 294 LXXII Pad Bath THPC/MCC- 100/ 200/300® 295 LXXIII Pad Bath TRPC/MCC-100® /Guanidine Carbonate 296 LXXIV Pad Bath THPC/MCC-100® /Guanidine Carbonate 297 LXXV Pad Bath THPS©/MCC-1 00® /Guanidine Carbonate 298 LXXVI THPS0/MCC-1OQ® Aerotex®23 on ETIP 50/50 Polyester/Cotton 299 LXXVII THPS®/MCC-10Q/2Q0®0n ETIP 50/50 Polyester/Cotton 300 LXXVI 1 1 Pad Bath THPS® /Carbamate/MCC-200® 301 LXXIX Isoperibol Data of the Decabromodi phenyl Oxide (White Chemical) Treated Polyester/Cotton Blends 305 LXXX Thermal Analysis of FR P-44® Finished Fabrics 309 LXXXI Bromine Contents of Fabric Samples Treated With P-44® Retardant 310 LXXXII P-44^ Formulations with Permanent Press Treatments 313 LXXXI 1 1 Flammability of Permanent Press/P-44® Treatments 315 LXXXIV Characterization of DBPA Emulsion Polymer 318 LXXXV Physical Properties of 50/50 Blend Treated with FR P-44(2)and P(DBPA) 319 LXXXVI Pad Bath Formulation for P-44©with P(DBPA) Based on 91% Wet Pick-Up 320 LXXXVII Emulsifier Formulation 323 LXXXVIII Application of Citex®BT-93 to 50/50 Blend Fabric 324 LXXXIX Citex®BT-93 and Sb^ on 50/50 Blend Fabrics 325 XC Characterization of 50/50 Blend Fabrics Treated with Citex®BT-93 326 XCI Calorimetric Data of 50/50 Blend Treated with DAP/TBPP (With Fixed TBPP) 330 XCI I Calorimetric Data of Polyester/Cotton Blends—Com- parison of Calorimeters 337 viii List of Tables (cont.) Table No. Title Page No XCIII Calorimetric Data of Some Experimental Polyester/ Cellulose Blends 340 XCIV Isoperibol Data of 900F®/Cotton Blends With Various Fabric Weight and Construction 343 XCV Isoperibol Data of H 3 P0 4 Treated 900F/ Cotton Blends 344 XCVI Char Yields of H.PO, Treated 900F©/Cotton and Polyester/Cotton Blends^ 345 XCVI I Isoperibol Data of QOOF^ /Cotton Blends With Varying Bromine Content 352 XCVIII SRRC formulation 357 XCIX Isoperibol Data of PVBr/THPC and VBr/THPC Treated Blends 360 C Heat-balance in Flame Retardant Treated Cotton Systems 363 CI Linear Regression of Heat Release From P-44^Treated Fabrics 368 CII Linear Regression of Net Heat Reduction from P-44^ Treated Fabrics 369 CIII P-53©Treated Fabrics (w/o Grid Support) 371 CIV Efficiencies of P-44®and P-53 ©Treated Fabrics 373 CV Efficiencies of Bromine - Containing Retardants 375 CVI Formulation for THPS^VP(DBPA) Finish 379 CVII Test Data on THPC - Latex Treated Fabrics 380 CVI I I Pad Bath Formulation I for Phosphonium Condensate and P(DBPA) 382 CIX Pad Bath Formulation II for Phosphonium Condensate and P(DBPA) 383 CX Effect of Some Selected Additives on the Film Proper- ties of Sandoz 1030/190 with Resin and Accelerator 390 CXI Electron Beam Fixation of P and Br Containing Monomers 394 CXI I Flammability of Cotton Fabrics Grafted with Vinyl Bromide 398 CXI I I Electron Beam Grafting Results 401 CXIV Electron Beam Grafting Results - NDPA Copolymers 402 CXV Electron Beam Grafting of NDPA Copolymers without Preswelling 403 IX List of Tables (cont.) Table No. Title Page No, CXVI Effect of NDPA on Copolmerizability of Monomers 404 CXVII Effect of Preswelling on Grafting and Flame Retardant Efficiencies 408 CXVI 1 1 Electron Beam Grafting of Phosphorus Monomers with DBPA 409 CXIX Grafting Using Monomer Emulsions 410 CXX Emulsion NDPA/TBPM (50/50) PET/Cotton Fibers y-Radia- tion Vacuum 411 CXXI Durability of Fabrics Grafted Using Neat Monomer Mixtures 413 CXX I I Durability of Fabrics Grafted Under N 2 Using Monomer Emulsions 414 CXXI 1 1 Durability of Fabrics Grafted in Air Using Monomer Emulsions 415 CXXIV Effect of Grafting Parameters on Durability 416 CXXV Effect of Methyl amine Pretreatment on NDPA/DBPA Grafting and Flame Retardance 418 CXXVI Comparison of Grafting Sources 419 CXXVII Selected Formulations for e" Curtain Experiments 420 CXXVI 1 1 Durability of Samples Grafted in an e" Curtain 421 CXXIX Dyebath Application of Flame Ratardants to PET 423 CXXX Dyebath Application of FR Monomers to 50/50 Blends 424 CXXXI Durability of FR P-53^Using Grafted NDPA as a Binder 425 x Page No. 9 10 11 12. 13 14 15 16 17 18 List of Figures Figure No. Title 1 Start-up operations 2 American Enka Research Plan 3 Hooker Chemicals and Plastics Corp. Research Plan 4 Polytechnic New York Operations 5 Southern Regional Research Center Research Plan 6 Research Triangle Institute Research Plan 7 Research Plan for Clemson University and the University of Maryland 8 Commercialization Plan 9 Pert Program for Overall Project 10 Pert Program for Study of Existing FR Systems 11 Pert Program for Design & Synthesis of Prototype Treatments 19 12 Pert Program for Scale-up of Inherently FR Poly- ester Blends 21 13 Effect of moisture on 01 values for cotton/PET blends. 29 14 Temperature-01 relationships for fabrics contain- ing cotton and or PET. 31 15 DTA thermograms for textile fabrics. 32 16 Rate of heat release for cotton/PET blends. 34 17 Thermal analysis of phosphoric acid treated cellulose. 36 18 Estimated tar formation for treated cotton. 39 19 Flame retardant efficiencies of trivalent phos- phorus compounds. 41 20 Flame retardant efficiencies of pentavalent phosphorus compounds. 42 21 Heats of combustion of char. 45 22 Estimated fuel formation for treated PCHDT. 47 23 Treated heat releases of HgPO* PET/cotton blend fabrics. 55 24 Minimum FR content for self extinguishment in vertical test. 64 XI List of Figures (cont.) Figure No. Title Page No, 25 Effect of phosphorus flame retardants on the 01 of 65/35 PET/cotton blends. 66 26 Effect of phosphorus flame retardants on the 01 of 50/50 PET/cotton blends. 67 27 Temperature-OI relationships for fabrics (3.4 oz. /yd. 2 ) treated with THPC-APO. 72 28 Temperature-OI relationships for fabrics (3.4 oz. /yd. 2 ), treated with THPOH/amide. 73 29 Temperature-OI relationships for fabrics (3.4 oz. /yd. 2 ), treated with THP0H/NH 3 . 74 30 Diagram of polymer fire. 88 31 Calorimetric combustion scheme. 89 32 Heat release of PET/cotton blends burned with and without support. 96 33 Heat-release of diammonium phosphate treated ETIP blend burned with and without fiber-blass support. 99 34 Heat release of P-44 treated fabrics. 101 35 Net Heat Reduction of P-44 treated fabrics 103 36 Ignition Exposure Time Tester 105 37 MAFT heat transfer as a function of fabric heat release per unit area. Ill 38 MAFT heat transfer as a function of fabric heat release. 112 39 Effect of scan rate on the TGA of PET in air. 123 40 Log 01 as a function of Log ^^ + 2 for PET. 126 41 GC pyrolysis of bromine containing flame retardants. 132 42 GC pyrolysis of PET films containing bromine flame retardants. 133 43 The relationship between 01 values and weight % Br in PET fabrics treated with bromine fire retardants. 135 44 Heat reduction of burning PET fabrics treated with bromine fire retardants. 137 45 Normalized ESR signal versus percent weight loss for Dacron(*)54 at a decomposition temperature of 356°. 160 xii List of Figures (cont.) Figure No. Title s Page No 46 Normalized £SR signal versus percent weight loss for Dacron©54 decomposed at 387°C. 161 47 Normal ized^ESR signal versus percent weight loss for DacronQS) 900F decomposed at 356°C. 163 48 Normalized J3R signal versus percent weight loss for Dacron®900F decomposed at 387°. 164 49 Normalized ESR signal versus percent weight loss for phosphazene containing PET (PFR) decomposed at 356°C. 166 50 Normalized ESR signal versus percent weight loss for phosphazene containing PET (PFR) decomposed at 387°C. 167 51 01 as a function of PFR-1 in PET. 175 52 01 as a function of the wt. % PET residue at 800°C. 175 53 SEM photomicrograph showing a cross-section of embedded PET fibers. 185 54 PET fiber cross-section partially freed from embedded medium. 185 55 Energy spectrum for x-rays emitted from a PET sample solution coated with poly(vinyl bromide). 186 56 Secondary electron image of sample yielding energy spectrum in figure 55. 186 57 Bromine L x-ray map for sample shown in figure 56 using "window" indicated in figure 55 (white portion of spectrum) . 187 58 Energy spectrum from sample shown in figure 56 showing "window" on background map used for map in figure 59.187 59 Background x-ray map obtained from cross-section shown in figure 56 using window in figure 58. 188 60 Secondary electron image from a PET fiber to which vinyl bromide has been grafted. 188 61 Bromine L x-ray map for fiber cross-section depicted in figure 60. 189 62 Secondary electron image from a PET fiber to which diethyl vinyl phosphonate has been grafted. 189 63 Phosphorus K x-ray image of the cross-section de- picted in figure 62. 190 64 Enhanced version of figure 63. 190 Xlll List of Figures (cont.) Figure No. Title Page No. 65 Effect of grafting on PET m.p. as measured by DSC. 198 66 Analysis of grafted PET fiber thermogram. 199 67 01 as a function of % MD3P (unfixed). 220 68 Char formation as a function of MD3P (unfixed) on cotton. 221 69 Char formation as a function of MD3P (unfixed) on 50/50 PET cotton. 222 70 Comparison of the flame retardant efficiency of MD3P (unfixed) on cotton and 50/50 PET cotton. 224 71 AH ? /(AH°) F for MD3P (unfixed) on cotton compared to 50/50 PET/cotton. 225 72 01 of MD3P (fixed) on PET/cotton blend fabrics. 228 Effect of fixation on chars from MD3P treated cotton. 230 73 74 Effect of fixation on chars from MD3P treated 50/50 PET/cotton. 231 75 Effect of fixation of MD3P on its flame retardant efficiency. 232 76 TGA of MD3P (unfixed) on cellulose. 236 77 TGA of MD3P (unfixed) on 50/50 PET/cotton. 237 78 TGA of MD3P (fixed) on cotton. 238 79 TGA of MD3P (fixed) on 50/50 PET/cotton. 239 80 Heat release of DAP/Antiblaze^ 19 treated 50/50 PET/cotton blend fabrics. 248 81 Y/(l-X) as a function of log % P for PyrovatexvJ 3762 treated fabrics. ^ 250 o< 82 AH ? /(AH°) F as a function of log % P for MCC-100/20(5 /300© treated fabrics. 252 83 Variation of AHL/(AH r ) F with phosphorus content of (NH 4 ) 2 HP0 4 treated fabrics. 254 84 31 p spectrum of THPC. 260 85 31 p spectrum of THP0H. 261 86 31 p spectrum of THPC-CH 3 NH 2 oligomer. 262 87 31 p spectrum of THPC-MCC 100®oligomer. 263 88 31 p spectrum of THP0H-MCC 100® oligomer. 264 xiv List of Figures (cont.) Figure No. Title Page No. 31 89 Comparison of P spectra of phosphonium salt oligomers. 269 90 Generalized relation between AH, and %P for flame retardant 50/50 PET/cotton blends. 278 91 AH-|/(AH C ) F as a function of %P for flame retardant 50/50 PET/cotton blends. 279 92 Generalized function of AHL vs. %P for flame retard- ant 50/50 PET/cotton blends. 280 93 Generalized dependance of residue yield on initial P content of FR 50/50 PET/cotton blends 281 94 Variation of AH, with P content of THPC-CH 3 NH 2 treated blends. 284 95 Heat release of decabromodi phenyl oxide treated blends. 306 96 Net heat reduction from decabromodi phenyl oxide treated PET/cotton blends. 307 fin 97 Bromine contents of P-44 vv treated fabrics. 311 98 Heat release from DAP/T23P treated blend fabric with constant T23P content. 331 99 Net heat release of DAP/T23P treated 50/50 blends. 332 100 Calorimetric data for optimization of P/Br finishes. 333 101 Effect of fiberglass grid on AH, of PET/cotton blends. 338 102 Rates of heat release of PET/cotton blends. 339 103 Heat release of H 3 P0 4 treated 900F/cotton blends. 347 104 AH, of H^PO* treated 900F/cotton and PET/cotton 50/50 blend fabrics. 349 105 AH, of HoPO- treated 65/35 900F/cotton and PET /cotton Blend fabrics. 350 106 Heat release of various HgPQ^ treated 900F/cotton blends. 353 107 Heat release of various 900F/cotton and PET/cotton treated with H 3 P0 4 or PBVr/THPC. 354 108 Variation of AH, with %P of PVBr and P(VBr/VCl) treated PET/cotton blends. 362 109 Dependance of AH,/AH ? on Br content of PVBr and P(VBr/VCl) treated 50/50 blends. 364 xv List of Figures (cont.) Figure No. Title Page No 110 AH,/(AH5) E and AH ? /(AH°) F from PBVr and P(VBr/VCl) treated PET/cotton blendi. 365 111 Heat release of P-44®and P-53® treated fabrics. 372 112 Net heat reduction from P-44®and P-53® treated fabric. 374 113 Comparison of Sandoz 1030/190 film DSC results. 387 114 TGA of Sandoz 1030/190 plus resin and accelerator film. 388 115 Heat release from Sandoz 1030 as a function of Br content. 391 116 Heat release of Sandoz 1030 as a function of wt. %P. 392 117 Effect of NDPA:BDP0M copolymer grafts of 50/50 PET/cotton TGA. 405 118 TGA data on NDPA and BDP0M homopolymer grafts on 50/50 PET/cotton. 406 xvi INTRODUCTION PROJECT OBJECTIVES This project has had the dual objective of developing a commerci- ally practicable process for rendering cotton/polyester apparel fabrics flame resistant, and evaluating the feasibility of stimulating commer- cially important technological innovation through the organization and federal subsidization of a consortium of academic, industrial, govern- mental and private non-profit research laboratories. Since there are currently no accepted criteria for judging the flame resistance of general apparel fabrics, it was decided that the technical achievements of the project should be evaluated in terms of the existing standard for children's sleepwear sizes 7-14 (FF 5-74) and the proposed Catagory I classification as defined by the National Bureau of Stand- ards using the mushroom apparel flammability tester (MAFT)(1). The evaluation of the consortium approach to conducting the re- search and stimulating technological innovation will be made on the basis of the degree of success or failure made toward the achievement of the technical objective. ORGANIZATION OF THE CONSORTIUM Because of the broad scope and the complex nature of the problem it seemed highly unlikely that any single laboratory or company could succeed in developing an appropriate commercial solution within the two year time frame required by the needs of adequate consumer protec- tion as expressed in the Request for Proposals issued by the Department of Commerce. For this reason a consortium of interested and experi- enced research and development organizations was formed for the speci- fic task of conducting an integrated and coordinated search for such a solution. This consortium was composed of investigation teams from Clemson University, the Unviersity of Maryland, the Research Triangle Institute, the Polytechnic Institute of New York, the Southern Regional Research Center of the U. S. Department of Agriculture, the Hooker Chemicals and Plastics Corporation, the American Enka Company, and United Merchants and Manufacturers, Incorporated. The Dow-Corning Corporation joined the consortium in May, 1975. The technical administration and coordination of the various phases of the project have been the responsibility of the Principal Investi- gator. He has been assisted in this function by the team leaders from each member of the consortium, who sit on the project steering committ- ee. The committee members have been: Dr. Michael J. Drews Assistant Professor Textile Department Clemson University Clemson, South Carolina 29631 803-656-3177 Dr. Alan W. Meierhoefer (1974-75) Dr. Gerald W. McNeely (1975-76) Research Center American Enka Company Enka, North Carolina 28728 704-667-7283 Dr. Betty F. Smith Dr. Kwan-nan Yeh Department of Textiles & Consumer Economics University of Maryland College Park, Maryland 20742 301-454-2137 Mr. Richard Lyons United Merchants & Manufacturers P.O. Box 2148 Greenville, South Carolina 29602 803-233-4641 Dr. Raimond Liepins Research Triangle Institute Research Triangle Park, North Carolina 27709 919-549-8311 Dr. Eli M. Pearce Professor of Chemistry & Chemical Engineering Chemistry Department Polytechnic Institute of New York 333 Jay Street Brooklyn, New York 11201 212-643-5000 Mr. George L. Drake, Jr., Head Special Finishes Investigations Southern Regional Research Center P.O. Box 19687 New Orleans, Louisiana 70179 504-589-7061 Dr. James J. Duffy Hooker Chemicals and Plastics Corporation Hooker Research Center Niagara Falls, New York 14302 716-773-8520 Dr. John W. Ryan Manager, Fluids Research Corporate Center Dow-Corning Corporation Midland, Michigan 48640 517-496-4847 Mr. Richard T. Penn (ex -officio) Room A-740, Building 101 National Bureau of Standards Gaithersburg, Maryland 20760 301-921-3185 In addition, Mr. James H. Winger, Chief, Program for Fire Prevention- Products, National Bureau of Standards and Dr. Vivian T. Stannett, Vice-Provost, Dean of the Graduate School, and Professor of Chemical Engineering, North Carolina State University have served as advisory members of the steering committee. This committee, chaired by the Principal Investigator, met to review the progress of the individual teams and plan future investigations on a quarterly basis. All major decisions concerning the technical progress of the project were re- ferred to the steering committee for advice and consent. Each team leader has been responsible for supervising his team in a specific phase of the project. Although the exact work plan for each group was the responsibility of the team leader, the allocation of general areas of investigation among the teams was the responsibil- ity of the Principal Investigator in consultation with the steering committee. It was felt that in general no formal agreements were necessary for participation in the consortium; however, formal subcontracts were required for those groups receiving funds from the ETIP grant. Contracts were therefore drawn up and signed by Hooker Chemicals and Plastics Corporation, the Polytechnic Institute of New York, the Re- search Triangle Institute, and the University of Maryland. A memoran- dum of understanding was also developed to formalize the interaction with the Department of Agriculture. Provision was also made to assure maximum input from companies and laboratories not formally affiliated with the consortium. Such inter- actions were arranged on a bilateral basis with several of the consort- ium members. The conduct of these relationships was considered to be the sole responsibility of the individual research team involved but the results arrising from them were integrated into the total effort. PROJECT ADMINISTRATION OPERATION OF THE CONSORTIUM The project was divided into two phases. The first phase consist- ed of a determination of the basic factors controlling the flammability of cotton/polyester blends. A series of concurrent investigations was conducted into such factors as the effect of distribution of both vapor phase and condensed phase active flame retardants among the components at various blend levels. The effects of the chemical structure of selected types of phosphorus and halogen-containing retardants was also examined. These were evaluated on both cotton and polyester and their interactions on blends studied. This information was then utilized in attempts to systematically design optimum flame retardant formulations. In order to assure maximum utilization of the results of these mechanistic studies, a parallel and concurrent set of empirical invest- igations of more practical treatments was carried out. A number of currently existing semi -commercial and experimental processes were evaluated to determine their modes of action, types and causes of deficiencies and potential for modification to remove these deficienc- ies. Samples and experimental treatments were solicited from all seg- ments of the industry. In addition to developing new data, these ef- forts allowed the utilization of the results of the theoretical studies on a continuous basis as they were developed. Although the major emphasis of this work was directed toward blends containing normal commercial fibers, there seemed to be a good possibil- ity that special classes of polyester fibers could be developed contain- ing specific types and quantities of flame retardants specifically formulated for inclusion in blends which would require only minimal aftertreatment in fabric form. A portion of the project has, there- fore, dealt with the development of such fibers and the application of special flame retardant finishes to blends containing these intrins- ically flame resistant fibers. The second phase of the project involved the actual design and development of flame retardant treatments with commercial potential. These systems have been based on the results of the previous evaulations and theoretical investigations. Considerable emphasis was placed on such factors as treating methodology in both solution and emulsion pro- cesses and fixation methodology using both conventional pad-dry-cure techniques and radiation grafting. This work was planned with the ex- pectation that several systems would reach this point of development in pilot plant operations and that the most promising of these would then be selected for a full-scale mill trial to demonstrate its com- mercial acceptability. Thus the final results from the project was ex- pected to include one system of demonstrated commercial potential and several additional systems having considerable promise but not develop- ed to the point of acceptability for mill operation. The actual operation of the consortium and the division of labor among the consortium members is illustrated diagramatically in Figures 1-8. These flow charts are intended to show the general catagories of investigations undertaken by the various research teams and the interactions of these teams. In order to establish a schedule for the completion of the various research tasks, a series of PERT diagrams were prepared. These are presented in Figures 9-12. The numbers indicated refer to the mini- mum, most probable and maximum projected times in weeks to complete the various phases of the work. These diagrams were prepared by the Principal Investigator and Dr. M. J. Drews, in consultation with Dr. John J. Willard of A. D. Little & Co. Dr. Willard was asked to comment on the organization and adminstration of the project since he has considerable expertise in this area. Comments regarding the gen- eral organization and goals of the project were also solicited from Dr. Giui liana Tesoro. In most cases it was possible to keep fairly close to the sche- dules established in the PERT diagrams. The actual design and formu- lation of the candidates for commercialization required more time than originally projected but this was partially due to a greater than an- ticipated success with the prototype systems. This meant that there were more candidates to evaluate than had been expected. In all phases of the project the consortium seemed to operate very 8 c o co CO CD C •r~ +-> "+- 13 J_ S- CD CD +-> CO C CD *r~ >^ CD CD O CO Cl M- O CO CO •r" " -O CO C CD 3 -C O 4-> Q C E >> O CO U c , •r— CD CD X5 CD O CO E 4- c . O 03 1 — / samples fabrics on by ca TGA \ •f* * Prepare treated evaluat 101, DTA -0 O c JD 03 'r- S- co CD T3 Cl c: O CD CO r— •1 — JD O co co +-> u c CD CD 1— S- r— i_ O Z3 "O ^ c rtJ CO CD CD C ^— O Cl E 03 03 . CO CO CO k. TD +-> CD C U JQ QJ CD •r— r— 1 — M- JD r— O CC. C C_> Li_ '•"" (/) +-> 05 s- S- CO ID C5 z cc LU i 1 a. Z3 CJ J* Of I O K \J_j« e ^ * <+- O 4- ri IS) o-<~ tt, •!-«♦- •p" +J •M 3^ 1 e o> u_ o. «/> o OJ "g US X5 5 o 01 £ a: T> s*-a u. &- c o o Fa It £& 18 %. t 0) r- Q. Cl 3 <0 l/> CT»-S reeni cand film a 8 o cc c WU.1- fc .* 10 "S c en "r- V) «*- (U O T CO J- CO >> 0> •r- r— E CO #— O o> *^ c sz '^ o 4-» u E c ,:!,; >, q,q; 4rt CO c o •r«* +j 0J § c f— o fO e c i_ a: CD IX. +J £ H- O CO •«— a> ■»-> i— c aw ai E •- -M W «3 O co u Q- U (1) O M- O O CD co o i ?o -M E CO S- H- >> o o co <+- i. CO o> CD -f- c Q. CO •r— \ >»+J ■»-> i— CO CO rO »0— O C X O tTJ to ro a. -o c: ta to 'iio o CD o s- o o CO LjJ a: CJ3 11 >'1 c o +■■• s_ Q. LU CD o s u o Q- c fe ^ ^ o C_) ■*-> >+- 0) o O) "O « — - O (O E -^ c 4- UJ O — ' CO i- • i— > c r— >> o CO a. determination of effect of PET structure on flammability W ► •M CO >> 3 21 12 00 o C C co (O O >i i — ■>- CO to a: ■*-> 3 u_ o r— .— co o c •>- E 0> o Q.T- E 3 (O i— 00 co s- s- c 4-> v en a> t- en .0 TJ %~ r ™ O. E ^ c oj a> O -M r— •r- A3 flj +-* "O u a> -o 1— c u M- O C QJ O +-> -4-> -O u CD m •O c C ^ 1/1 0J U T> X 3 O ■M -♦-> o ai +■» IB «» r— 3 r— Q..— r— S^ IB > U E a» I ai +-» »— e e +j 3 a. ct> o ^~ E i- >— o i— a. a» I o © «9 a. 01 ■*-> r— C IO « c VUr-O o ■•-> l/l Q.-I- o O) (O LT> r— r— +J 3 Q.I— Oi— E ^o L- t 01 4-> ^-* C C u f=- 'r— o *-» l/l o.*» r— OI 3 Q.I— O i— § j| i— a. oi l/l oi ■»— ^~ 1 £ l/l O ■X* I •X; O *»■ i CM I OI ■»-> >— IB O «/> -^ "*= 00 4 >> C l/l i~ O O +» OI O "i- -r- c o ■M +5 +* t~ 0) o ai 4- S. u i— a> o •*-> 1 c -^ 01 I/) OI i/> co f— •^ 1— J- Q.TJ CF> OI 3 C -Q Q P •f— -r— u l/» l/J «♦- '50 I o CO I CO Oi o. "O >> «/> C t/> +-» ■»-> o OI IB *r- O C o ■*-> in +J a) l~~ Oi C Oi O E f— OT-C i- P Q.-i- •»-> Q. IB E to C Oi O 0) >»«*- t- U "O l/l O -M i 1— LU a. c ■g o $- OI C t- o o +J IB +-> <«- o OI IB en r- • 3 CSC Ql Q-f— U, E 41 IB O W > <*- o a. oi o I 00 * l/> 0) -r- l/l ■•-> (/> ^_ O OI 41 ^J O ■— .C E 5 >>*- i o w o l S l ,fm 4-> XJ) c 1/) o °9~ R3 4-> o X c i£> 0) 10 u 1 >i «o= ■p XI L. IB .3 J3 41 •P ft3 J- 1/) ««- ■»-> ^>r— — L "■0 m i o i Csl c o L. QJ •I— O o •u 4-1 i*- o 4* IB 00 p— > 3 QC a. IB u. »— o > «4- UJ 1 o 4* o ex. 1 ._-. 9€-fr2-0Z 17 o •r- S- c ■u o a; £■»-> *d- — J_ 3 Q. Oi — E r— u o CD CO 00 q: Li- en C to x LU a: >> zd -a >— 4-> LL. OO s_ o E >u S- CD o s- CD ^A to CD I ID C o •"- TJ ■M dJ f0 "M s- -O co s_ (O t— Q- 10 r— O •f— i- > .a t= O «4- •f— *r— U O x«*- o CM I CD I LO, QJ to ■M r— •r— > l£) a. C r— E OJ its ai a. "" (T3 a: <+- +3 CO o c o a» r— 4-* l/l F 10 r — a a> *r— o a» r— s- •f- i— o. 4- 18 J- c 4-» O O 0> 0) •■- »-■»-» E ••-» f— W 3 aor- §i— * IQ > s s CL S> o 4-> o i~ Cl. O ai $ CD •r— ^— E >> Q. C E Ot > -C Q. o 1/1 >, s +-> (/I jQ |Q IS >» E 4t e L- fm* SO g L. O -C to J— ■=■ ^ 5, 3 O a. 01 o. E E 4-1 V) o 41 <4- 4- a o -o o Q. o> A 00 <£> co (. at cm a> c > O 4J •r - •»— vo at ie +-> IT—' .-« K Q. »l~ s 4-> <© t. ai U -Q V£> ^— •r— ^r CNJ "- 8 o •1- (J 10 ai 4-» ••- +> e= ••- t- +■> U? at +-> c ■--■ 01 U UO.-0 M3 €—•<-•.- t. E -Q CL4-> *3 5 «a o.t- at ! o «*- «o 5 i- c O (/> •i- o !f-. at -M ••- ..-_.-, 4J at •<- a. c (MM 4-> >, IC o at at •*-> -o V£> i— JS o t. Ou4-> 4-> ia E c o +5 O >. 1- 0) O w> Q. L. Jf I \ / JP «*- u O 00 III v» at •*-> CNJ at -t- o. c O •♦J in >» «> "■'--.' nthe otot tard o >» t at u «/t a. i. n n • •— JT< o> c •r- IB at +3 x: c co ■*-» o u ociium V£> >> Q. 4J *-» C at o c io 4-» •<-> «0 *8"E°- E -M 4-> o <«- at o o c t- -o 19 ,,, a o a iu-j i— o KO Ot -r- <_ +j +-> a at «o s- r- 34- o.r— a > ae o at u. 4 « <**• CM u •r- 1. c <£> 4-> O r— a> a» t- vo +J E +j a» •<- * f— S- 3 Q. O ■— E r— (O O rO > o u at O o rj:i A vo i UD CM I « / c o •r- O «* Ol +J -r- I--P n> i. vo at o jq r— -f- (O Q.1— <«- E a. O Q- o U Ifl+J i M- C O o> •r- Wt (/I •!- CO at to >> ^~ "O Of r — i — 40 <£> x: i— (a -i- c •♦-> C "r- U <0 at >>-t-> i- "o r— CO c at l- o. at E E +-> o c o o at (J «3 O. O i- 1 CO on 20 a: oo -a c oo o; >> 'o a. +-> c O. I ai rO o oo o s- o s- Q- C CO O ■— •i- ro +-> O > S- -C a> a> Q- +-> +-> (U (_) o i— +-> Q. a> a: u a>u- a> on oj+j o. E o <_> >-. J- ro r— •i — O •r— Q. i — •r— -o F CD X 3 o S- o +-> o o.+-> 00 I c o 0J +J c +-> ra a> a; o E r- -r- a: -P CLi— u_ O c o > >> r~ r-~ s- o o CU Q_ Q_ +■> \ — _ to £= c C CU o o o >>■*-> +J -M 1— +-> +-> +J o o o O Q- C_) CJ C_3 fO -Q O "O O LO CNJ o LO o 3 Q. O LO s- o o E i. o opua LO LU => CD 32 a series of cotton/polyester blends. As shown in Figure 16, the rate of heat release was found to be a direct function of the amount of cotton in the blend with pure cotton exhibiting the highest rate. Pre- vious investigations by these same workers had shown a correlation be- tween the rate of heat release and the rate of flame propagation. 1. Theory of Flame Retardant Action Effective flame retardants may act either in the condensed phase or in the vapor phase above the decomposing polymer. Retardants which act in the gas phase exert their effect by functioning as either inert diluents or as free radical inhibitors which alter the oxidation pro- cesses and decrease the heat returned to the polymer surface. Those retardants which act in the condensed phase may operate by several mechanisms. They may inhibit the polymer pyrolysis so that it does not break down to produce the small volatile molecules necessary for flame propagation. More commonly, however, they act to alter rather than inhibit this thermal degradation reaction. The alteration is such that the mode of decomposition is changed and lesser quantities of flammable gas are produced. Finally, they may also exert their effect in a physical rather than chemical manner. In this case they act as a shield to prevent the transfer of heat from the flame back to the polymer surface. This reduces the rate of pyrolysis and fuel pro- duction is decreased. In order to evaluate a particular flame retardant system, it be- comes necessary to determine the mechanism of action of the various flame retardants on the substrate. This usually requires a knowledge of whether the flame retardant is gas phase or condensed phase active. Once this is known, an investigation may be begun to determine the actual chemical mechanism involved in the retardation process. A wide variety of phosphorus-containing flame retardants are known to be effective on cellulosic substrates. Of these, phosphoric acid is one of the simplest and most effective. For many years it has been postulated that this material acts completely in the condensed phase 33 c o 4-> ■M > — s O o <_> CD l/l o-i-; i — E U LO , — I (O u o en + • o r~- +i CM ii U0 II -o CD to 4J * — * ■TJ cc T o o o CO o o o c_> o o en C 0) c o ■l-> o <_> i. o en o •3- O I Q. 3: Q. VO CO z 3: <_> :r I— - — ' <] o o C\J X o r— a. o IT) O O 00 o o o o «3- o CXI c o +-> o o "O (fO 4-» S- O 4-> n3 +J ■a re £ CO UJ uo;^od w6/w6 X-L as 39 thermally stable sytems is therefore given by the effectiveness of the interaction between cellulose and the phosphorus oxides. The only possibility for their efficiency to be enhanced would be for them to be converted into some degradation product other than phosphorus oxides by preceding reactions with an additional reagent applied to the cellu- lose. This has been found to constitute the basis for the phosphorus- nitrogen synergistic effects observed by many workers (16). In this case, the phosphorus and nitrogen compounds react on heating to form a P-N polymer. The phosphorus atoms in this polymer have considerably enhances reactivity toward the cellulose and are much more efficient as flame retardants. This has been confirmed by model compound studies The situation becomes considerably more complex when the flame re- tardants are thermally stable up to the point of polymer degradation. In such systems the nature of the phosphorus-containing function group is important in determining the flame retardant efficiency of the sys- tem. In order to study this effect, compounds of the type P(0) 3 and ^(OHQ^k having n=0, 1, 2 and 3 have been synthesized and applied to cotton fabric. Calorimetric evaluation of these shows that the efficiency of the flame retardant system increases as the number of oxygen atoms bonded to the phosphorus increases (15) as shown in Figures 19 and 20. An exception to this behavior is observed with tri- phenyl phosphate which probably represents a special case because of its volatility. This would indicate that for thermally stable flame re- tardant additives the efficiency of the system can be limited by the chemical nature of the phosphorus compound. Substantiation for this proposal has recently been realized by a study of phosphoramides and related compounds (16, 17). Simple phos- phoramides such as (CH 3 NH) 3 P=0 and (cf>NH) 3 P=0 exhibited efficiency char- acteristics that were significantly different from those P-0 compounds studied earlier. In all cases, the P-N systems were more effective flame retardants. The volatility of the methyl compound was a major factor in determining its effectiveness but this could be altered by various fixation techniques. In these systems there apparently is a competition between volatilization and reaction with the cellulose. 40 o o 00 o o o o o o in o o o o LU CD X-L/A 41 o / / / / / p / o o in O o CO O CM / / / / / / / / / CO •a c rs o Q. E O O c/> 3 S- O .c a. to o 4-> c +-> a.; Q. CD o c: o a> / / / / CM O p -e- w a. <— » a. CO O CM ■»»• O <-» ^- — ■ O COO© Q- CO •©• Q. -©• CO -e- — - -e- •"^ 3: O D < O O d •a $- ro 4-> )- n and (c])0) P(0)(NHcj>)~ . n 3 — n n o — n All of these results, when viewed in the context of other exper- iments involving product analysis and model compound studies (18, 19), allow the postulation of chemical mechanisms to explain the interaction of phosphorus flame retardants with cellulose. The efficiency of such reagents appears to be related to the electron density at the phos- phorus atom. As the electron deficiency increases, the molecule be- comes more electrophilic and more acidic. This leads to enhances re- activity in phosphorylation reactions and increased its efficiency in catalyzing the degradation of the cellulose. Both of these reactions inhibit the formation of flammable volatiles from the cellulose and promote formation of char and incombustible volatiles. Although such correlations are fairly straightforward in most cases, the calorimetric technique is also capable of discerning unusual and unexpected effects. For example, it has recently been found that for structures of Type I the flame retardant efficiency of the phos- phorus compound is in the order indicated and depends strongly upon the chemical nature of the moiety attached to the other end of the carbon c ha i n ( 1 6 ) . Q (CH 3 0)^ P — CH 2 CH 2 R I R = C(0)NHCH 2 0H> C(0)NH 2 »CH> C(0)0CH 3 The reasons for these unusual structural effects are not obvious from the calorimetric data alone; but these results have been inter- preted in terms of an interaction between the phosphorus- and nitrogen- containing groups within the same molecule. This happens prior to in- teraction with the cellulose so that the species which reacts with the cellulose is most probably a phosphoramide rather than a phosphonate. Since the results of applying the calorimetric technique to cell- ulose were so useful and instructive, an attempt was made to apply the 43 method to polyester (12). Pure poly(ethylene terephthalate) (PET) fabric under normal atmospheric conditions was found to melt in the igniting flame. This resulted in dripping of the flaming polymer. Collection of the molten material resulted in quenching which did not allow for complete air combustion of the PET. Also a large volume of black smoke was liberated and this was not easily collected for further combustion. This study also indicated that while PET could not be investigated directly by static oxygen bomb calorimetry, a related polyester, poly- (1 s 4-cyclohexylene dimethylene terephthalate) (PCHDT) could be studied. The PCHDT has a higher melting point (290 C) which allowed the fabric integrity to be maintained long enough for ignition to be achieved. Since tris(2,3-dibromopropyl )phosphate (TBPP) has been used as a flame retardant on polyester, it was studied along with triphenyl phosphate (TPP) to contrast the efficiency of two phosphates as flame retardants, one of which contained halogen. The calorimetric heats of combustion of PCHDT fabric treated with TBPP decreased linearly with increasing amounts of TBPP due to its lower heat of combustion (1841 cal/g). However, with TPP the heats of combustion of treated fabric remained essentially the same since the (AH ) Tpp is 6799 cal/g which is almost the same as that of PCHDT (6878 cal/g). The conditions used in deter- mining these heats of combustion are such that total combustion of all the organic material is achieved. The situation is much different when the combustion are carried out under atmospheric conditions. Thus, samples of these treated fabrics were burned and weighed. With TPP the char yield increased only slightly as the phosphorus content in the fabric was increased. However, with TBPP there was a significant increase in the residue. Determination of the heats of combustion of these chars (Figure 21) indicated that the TPP and TBPP chars were quite different thermo- chemically, and apparently the chemical nature of the char is modified significantly when larger quantities of phosphorus are present. Also, while bromine is generally assumed to operate in the vapor phase when used as a flame retardant, the large differences in the heats of com- 44 7000 6000 - o o < 5000 4000 — % Phosphorus FIGURE 21. Heats of combustion of char. 45 bustion of the chars suggest that some of the bromine may be effecting the decomposition in the condensed phase (12). Using these data, corrections can be applied to the heats of com- bustion obtained for the fabrics and AH values corresponding to the heat liberation in the open air can be calculated. These are the values which are expected to parallel the flame retardancy observed for the two treatments. With TBPP the incorporation of increasing amounts of phosphorus into the PCHDT fabric not only reduced the overall heat of combustion, but also increased the char residue to better than 50% at a level of 1.6% phosphorus. At the same time the net heat evolution was reduced from 5073 cal/g with no phosphorus to 3796 cal/g at 0.7% phosphorus and approximately 2100 cal/g at slightly less than 1.7% phosphorus. In an analogous treatment, the fabric impregnated with TPP ex- hibited total heats of combustion which are essentially the same with increasing phosphorus content. In this case, the TPP was not effect- ive in increasing the residue formation significantly, and the net heat values only decreased from 4463 cal/g at 0»6% phosphorus to 3805 cal/g at 2.85% phosphorus. Thus, this material would not be expected to exhibit good flame retardant properties. In an attempt to increase the mechanistic significance of these data, Y values similar to those used in the discussion of the cellu- losic were calculated. The data for TPP treated samples (Figure 22) showed decreasing values for Y for increasing phosphorus content. But even with approximately 2.7% phosphorus, the fabric was still decom- posing in such a way that about thirty percent of the polyester was being converted into the flammable volatile compounds which serve as fuel for the flame. As was observed with the AH ? values, a linear relationship exists between Y/(l-X), the fraction of polyester con- verted to flammable gases, and the logarithm of the phosphorus content. As before, the mechanistic significance of the linearity is somewhat ambiguous. With TBPP, the fraction of treated polyester converted to flamm- able gases, Y/(l-X), increased more rapidly than with TPP when the 46 o CO o CM o o CD o o O O O O CM w6 x-j^ IQHDd *u35 ' a Z3 Q s. o DC .c O o. Q. OJ o J- — c^ o o c: CO o •r- d 4J ca r^ o O VO o -o a> 4-> A3 E •i— •»-> to UJ OJ LU a: i— i u. 47 lower levels of phosphorus contents in the fabrics were compared. This indicated that the polyester burns more easily with small amounts of TBPP as a flame retardants, and that in char formation the effects of scaffolding were greater than those of the inhibition reaction at the lower percentages of phosphorus from TBPP. After the TBPP and TPP lines cross, the inhibition reaction dominated, and the TBPP was clearly more efficient than TPP in reducing the fraction of treated polyester con- verted to flammable gases. Also, these results indicate that using a semi-durable flame retardant finish which could be removed or hydro- lyzed to leave a low percentage of phosphorus may create a polyester fabric that could present an increased hazard. This apparently greater hazard which has been indicated thermochemically confirms a visual observation that many investigators have made while conducting actual burn tests. The further utility of this method has been demonstrated in studies of blends containing polyester (12). It has been established that most synthetic polymers burn rather well when suspended by a carbonaceous grid formed from combustion of cellulosic or other nonmelting materials. If polyester is blended with cotton cellulose, the molten polymer is not permitted to drip away and the combined system is flammable. The calorimetric heats of combustion of cotton cellulose and a series of cotton/polyester blends were measured and are tabulated in Table I. In this case the presence of the nonthermoplastic cotton allowed the combustion experiments to be carried out at the 45° angle so that there was no appreciable amount of material lost from the burning fabric due to melting and dripping. These data show that there is a linear re- lationship between the percent of cotton and heat liberated. As the cotton cellulose was decreased from 100 percent to twenty percent in a cotton/polyester blend, the heat liberation increased from 3890 cal/g for pure cotton to 4746 cal/g for a fabric containing 80% polyester. This indicated that the combustion of the polyester portion is a sign- ificant factor in such blends. Of course, as previously noted, all of the heats and fuel blends Y/(l-X) obtained for polyesters and the cotton/ polyester blends con- 48 E en • o t— CT> o cm co in r^ co «vt- «=d- «3- «* «d- CO 1— o 1 _l E ZD en _l • _l t— UJ o >— .» en 2: r^ o C_> H- .^->. 1— o o O zc o & Q 1 Z - r— w^ «a- Li_ i o u. 1— o o * ID Ol r- (M ^ > in i— i— i— i— vo n cm cfi eft in cj VO ID ID ID N VO i— CT> in CM UD ID oo ro <* m "sj- «3 o o o o o o o o o o o o to oo m r-^ r»» os *$■ cm co in o in r— co in f"N. CT> r— >3- <3- «* «a- «sj- in o CD o s- CD +■> CO > o O. in 01 Q. c c X- o o o> +-> -p XJ +-> +J •r- o o U. C_) o in cm in i— cm in oo in in oo in CM 1 exp(-b 1 t) {3} where a, is a function of the initial conditions and . is a geometric factor which depends on the location of the temperature measuring devise in the calorimeter. When the T of the calorimeter is defined by equation {3} then the calorimeter is said to be in a rating period and the change in temperature is given by equation {4}. £=-b l( T-T„) (4} The only requirement then is that the calorimeter must reach a rating period before AT is evaluated. Experimentally Yeh found that the calorimeter entered the final rating period 150 seconds after ignition of a sample (20). The initial rating period at t=0, T=T is the period before ignition. From equation {4} and the slope of the response curve at two different times during the final rating b, and T^ cal- culated. From these data and the value of the integral in equation {2} from t=0 up to t > 150 sec. T is calculated. The total heat r — c release is then obtained from expression {1}. The value of E is determined in calibration experiments in which known amounts of propane are combusted in the calorimeter. In his calibration experiments Yeh (20) observed a precision of ±3%. However, he estimated that the overall accuracy of the calorimeter was really about ±7%, due to various systematic errors such as radiant energy loss from the bottom of the calorimeter and incomplete cooling of the exhaust gases. The rate of burning is measured from the slope of the response curve. In the work of Yeh and co-workers a least-squares re- 51 gression of the data yielded the expression: Rate = Q = 236.6S where S is the slope of the line in mV/sec. The standard deviation of the slope in their work was 5% with an estimated accuracy of 10%. Comparisons of the heat release values measured in the isoperibol calorimeter with those determined by the static oxygen bomb method were made. The oxygen bomb heats were higher. This is not unexpected since AH ? represents the maximum possible heat release which would be obtained only with the total combustion of all volatized materials. Except in those cases where the melt-drip properties of the fabrics made it impossible to collect residues representative of the fabric's ability to form chars AH,/AH 2 was found to be in the range of 80-90%. Cotton showed the largest percentage heat release indicating the val- idity of using the oxygen bomb calorimetry method for lightweight cotton fabrics. To extend this, Yeh has used isoperibol calorimetry to measure heat release and rates of burning of cotton fabrics treated with some of the same model flame retardants used in the earlier oxygen bomb ex- periments (21). As part of this experimental set-up, a combustible gas detector was attached to the exhaust of the calorimeter. This al- lowed the measurement of the heat contents of the uncombusted gases. From these data he was able to show that AH, + Al-L = AH 2 . Thus, the explanation that AhL < AH, due to incomplete combustion was verified by experiment. This also allows one to use this difference as a probe into the vapor phase behavior of these systems. Yeh also compared his isoperibol heat values with the AH 2 values obtained earlier (21). In all cases AH, was found to exhibit the same trends as AH ? . In addition, on the basis of heat values, the results gave a similar evaluation of the flame retardant efficiencies of the three retardants (diammonium phosphate > phosphoric acid > THPOH-am- monia). In each case, however, the actual differences among the 3 were quite small. Of particular significance are the ratios AH,/AH 2 and AH-/AH,. The fraction of heat released or the fraction of complete combustion in the gaseous phase is reflected in AH,/AH 2 . Within each 52 of the treated systems the fraction of heat released is essentially constant and independent of the flame retardant add-on. This is of significance not only in determining the completeness of combustion but also in establishing the mechanism of flame retardant action (21). Since the presence of the flame retardant does not seem to affect the completeness of the combustion of the volatilized materials, it can be assumed that the locus of action of all three retardants is predominant- ly in the condensed phase. This is in agreement with the results of other studies using other techniques. Although it has not yet been done, it would seem on this basis that the comparison of calorimetric values offers a powerful tool for the study of flame retardants and their location of action. It should be of particular interest in the study of those retardants presumed to be active in the vapor phase. The rates of heat release also constitute an important character- istic of flame retardant fabrics. These are easily obtained by the isoperibol technique. This is shown by the results of the work of Yeh and co-workers (21). An identical dependence of rate of phosphorus content was observed for all three systems and interpreted in terms of considerable mechanistic significance. The fuel values Y/(l-X) for all three treated systems had previously shown an identical dependence on the phosphorus content. Thus it was concluded that these three retard- ants operate through the same mechanism. Isoperibol calorimetry con- firms this. A direct correlation of this theoretical fuel yield with the rate of heat release was found. This correlation would seem to indicate that this is somewhat reflecting the kinetics of cellulose pyrolysis. However, due to the ambiguity of the heterogeneous kinetics involved, exact interpretation could not be made. A correlation was also found to exist between the isoperibol rates of heat evolution and the rates of flame propagation measured earlier by Hendrix et al_. On the basis of the relationship Q/Y = -AH6x Yeh and his co-workers were able to interpret these results. In this expression Q/Y gives the rate of heat release in cal/sec cm., -AH is the heat release in cal/g, 6 is the fabric weight, and x is the burning 53 rate in cm/sec. It is obvious from this that the linearity can only be followed up to a small range of retardant add-ons where incorpora- tion of the retardant into the fabric does not significantly alter — c 20 40 60 % Cotton 80 100 FIGURE 23. Treated heat releases of H^PCL PET/cotton blend fabrics 55 blends having only the polyester portion treated. Unfortunately, the calorimetric techniques which have proven to be such powerful mechanistic probes for cellulose have not been widely applied to thermoplastics. Thus much less is known about the action of flame retardants in polyester than in cellulose. To date, the only really significant advances in this area relate to the understanding of the reactions responsible to the production of flammable gases from thermalyzing polyester. A number of simple esters have been studied as models for poly(ethylene terephthalate). From these a number of the features of the polyester degradation have been determined. The thermal degradation of polyester itself has also been studied by several investigators (26-31) in recent years. Although Marshall and Todd (31) proposed that the rupture of a C-0 bond during the thermal degradation of PET would produce resonance stabilized radicals, none of the other investigators have found evidence to substantiate such a free radical mechanism. PET decomposes by random chain scission at the ester links (A scission), and the principal point of weakness in the polyester chain appears to be the beta-methyl ene group (32). The major thermal degradation reactions of PET are summarized as follows: (26-28) C 6 H 4 C0CH 2 CH 2 0CC 6 H 4 " \/ C 6 H 4 C0CH=CH 2 + H0C-C 6 H 4 ~- Vinyl Polymer Radical grafting formation of cross links C 6 H 4 C0H + CH v II I 3 f ^C.H.COCH-C-CH '6"4 '6"4 II 6' '4 C.H.C-O-C-C.H 6' '4 CH 3 CH0 polyenaldehyde 4. escapes XH=CH-CH=CH Polyene 56 This mechanism is very similar to the mechanisms proposed for the pyro- lysis of ethylene dibenzoate and vinyl benzoate. If 2-hydroxyethyl end groups are considered, the following reaction mechanism for the decomposition of PET is proposed (28): 1 ^^H.COCHoCHoOCC^H^ '6' '4 2""2 V ^6"4 » CH %C 6 H 4 C0CH=CH 2 + H0C-C 6 H 4 ^- f H 3 ^< 6 H 4 C0-CH-C-C^H^ L .... I fW J * II C 6 H 4 C0CH 2 -CH 2 0H M^C-O-CCgH^ CH 3 CH0 ^ 6 H 4 C0CH 2 CH 2 0CC 6 H 4 ^ CH 3 CH0 < 6 H 4 C0CH 2 CH 2 0H 64 22 64 64 nCH 3 CH0- ->CH 3 (— CH=CH— )— CHO + (n-l)H 2 n-1 polyenaldehyde 'vC^H.COCHoCHoOCC^H^ + H o 0==^^C r H,C0H + HOCH^CHoOCC^H^ , 6 m 4 wwm 2 um 2 vw 6 ., 4 - ,. 2 , '6"4 V ! 2""2 V ^6"4 I J ^ 6 H 4 C-0-C-C 6 H 4 ^ + H 2 0^ == ^'\£ 6 H 4 C0H + < 6 H 4 C0H Thus, as long as free hydroxy! end groups exist in the PET melt, the broken polymer chains will be reformed with production of an equivalent number of acetaldehyde molecules and carbonyl end groups. The mole- cular weight of the polymer will begin to decrease when most of the hydroxyl groups have been consumed. The detailed mechanism of the con- version of a PET unit, i.e. 57 M II II %C 6 H 4 C0CH 2 CH 2 0CC 6 H 4 ^—>< 6 H 4 C0CH=CH 2 + HOC-CgH/v is as yet only speculative, although the close agreement between the decomposition of poly(ethylene terephthalate) and ethylene dibenzoate suggests that it is probably similar to the mechanism for simple ester pyrolysis (28). When molten poly(ethylene terephthalate) is maintained under ni- trogen at ca_. 280 C, the polymer decomposes with (a) evolution of gas, (b) the formation of low molecular weight products of varying degrees of volatility, (c) the formation of different functional groupings in the polymer, and (d) the discoloration of the polymer. The following gaseous products (concentration in mole-percent) are formed (28): CH 3 CH0, 79.5; C0 2 8.5; CO, 8.0; C^, 2.0; H 2 0, 0.8; CH 4 , 0.4; CgHg, 0.4; 2-methyldioxolane, 0.4. The 2-methyldioxolane arises from the interaction of acetaldehyde and ethylene glycol. Although the con- centration of acetaldehyde varies greatly with the temperature of de- composition, it is always the major decomposition product. The major nongaseous degradation products are terephthalic acid and acidic oligomers (26, 28). An equilibrium seems to exist between the polymer and the cyclic oligomers (mainly the trimer) formed dur- ing the thermal degradation. Although the mechanism of oligomer form- ation is unknown, it is assumed to be transesterification involving end groups. A number of compounds, mainly derivatives of substituted benzoic acids and diphenyldi- and monocarboxylic acids, have been isolated (28) but these were found in very low concentration (0.005 - 0.09 weight %). Wall (23, 34) has studied the rate of volatilization of PET to determine the mechanism of degradation. Theoretically, random degrad- ation of linear chains is characterized by a maximum in the rate of volatilization which occurs at 26% conversion to volatile products (35) Poly(ethylene terephthalate) (33) fits the theoretical curves for ran- dom degradation except that the maximum is at a lower conversion and the volatilization tends to cease at about 80% conversion. The rate 58 of volatilization at the maximum (dc/dt) v , is independent of the max molecular weight of the polymer and is controlled by the expression kL (dc/dt) = •*-. 7 where k is the rate constant for rupture of the chain max c . / bonds and is related to a, the fraction of bonds broken at the time t, -kt by the relationship a = 1-e (35). The quantity L is the number of basic units in the smallest chain that does not evaporate without de- gradation. For PET, L=5 in the temperature range studied. These results are consistent with Goodings (28) and Buxbaum's (26) findings of acidic oligomers in the nongaseous products. As thermal degradation proceeds, the concentration of hydroxy! groups in the polymer decreases and the concentration of carboxyl groups increases. Anhydride groups are formed by the reaction of car- boxyl groups with vinyl end groups and by dehydration between two car- boxyl groups (28). The concentration of anhydride groups begins to increase when most of the hydroxyl end groups have been consumed. Goodings (28) found that PET which had been degraded for 65 hours under nitrogen at 306°C gave an IR spectrum which indicated that anhy- dride and carboxyl groups were present in the ratio 1:2:6. The color of PET changes from white, to yellow, to brown, and finally to black as the polymer degrades thermally. Pohl (29) reported on the basis of selective solubility that the chromphore was chemically bonded to the polymer chain. The chromphore is reported to be a highly unsaturated aliphatic molecule with a molecular weight of about 300-1000 (26, 28). Goodings (28) assumed that the route leading to color formation in PET was the formation of polyenealdehydes from acetaldehyde. Zimmerman, (30, 36) on the other hand, has shown that unsaturated colorforming molecules are produced from polyvinyl esters). Buxbaum (32) reports that both reactions contribute to color formation with the polymerization conditions determining the predominant reaction sequence. Investigations of the decomposition reaction kinetics of simple esters (37-39) have shown that the reactions are homogeneous, follow first order kinetics in both the vapor and liquid phases, and are un- affected by radical scavengers. On a theoretical basis, pure chain • 59 scission of the ester link (A scission) would require an activation energy of ca_. 60 kcal/mole. Since an activation energy of ca_. 40 kcal/ mole was found for simple ester pyrolysis, this lower observed value indicates a mechanism probably involving a cyclic transition state. The rate of thermal degradation of PET can be measured in terms of: (a) the products formed, (b) the rate of change of molecular weight of the polymer (as measured by the intrinsic viscosity or melt viscosity), or (c) the rate of change in the concentration of end groups (32). The reaction rate constants and activation energies found by the various investigators using one or more of the methods mentioned above are shown in Table II. It is significant that all the rate constants have the same order of magnitude. Thus, even though the pyrolysis reactions of PET have not been completely characterized, enough is known about them to make several predictions. For example, it would seem that acidic materials would enhance the thermolysis pro- cess and produce increased quantities of acetaldehyde, a combustible gas. Furhter speculation can also be made concerning the low proba- bility of finding chemical flame retardants which would exhibit con- ventional types of condensed phase activity. This latter speculation has been borne out in the one published case in which a polyester flame retardant system was subjected to a thorough mechanistic study (12, 40, 41). In this work an investigation was made of the effects of tri phenyl phosphine oxide (TPPO) on the thermal decomposition and combustion of polyester. Differential therm- al analysis and thermogravimetric analysis were used to monitor the thermolysis reactions. It was expected that phosphorus-containing flame retardant additives would exert their effects by alteration of the condensed phase pyrolysis reactions. The additives originally chosen were nylon 6 and TPPO. Prior laboratory flame tests had shown clearly that the PET/TPPO/ nylon system was the most effective in im- parting flame resistance (42). The results of thermal analysis in- dicated wery positively that TPPO did not operate as a flame retardant in the condensed phase since the main decomposition of PET was not affected by increasing the amount of TPPO present. These results im- 60 o •I— «3 ■o to s- o> 4- o O CM 00 tO CM r— rc 1 > sz •i— Smm* -l-> O i^ u_ 03 CC ■a c +-> o CO CM CO 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 (_> D I 1 1 1 I O I I I I I CP» 1 1 1 1 1 CM CO CM CM CM O f— I— I I to Q. =3 o s- CJ3 X> c CO +-> o >> +-> •1— to O a >^ to +-> •r— •r- > to o o o •I— to to •r- c > •r— s- +J -l-> r~ c CD CO 61 plied that any effect exerted by TPPO msut be exerted in the vapor phase. In order to substantiate this, oxygen and nitrous oxide indices were used as previously mentioned. A comparison of 01 and NOI for the PET/ nylon 6 system showed parallel lines. This indicated that any inhibition effects caused by the inclusion of nylon in the polyester occurred in the condensed phase. The 01 and NOI lines for PET/TPPO and PET/TPPO/ nylon were divergent. This intimated that TPPO poisoned the flame reactions (vapor phase) since there was inhibition in oxygen but not in nitrous oxide. Phosphorus analyses of the modified polyester samples indicated the reason for the lower flammability of the PET/TPPO/nylon 6 system. These analyses showed that whenever nylon was present in the polymer, an increased amount of phosphorus was held in the sample. Consequent- ly, the nylon increased the TPPO physically held in the polyester and slowed its sublimation. This has been confirmed in a recent mass spectrometric investigation (41). The observation has been made that a large amount of energy de- livered over an extended period of time will not propagate a flame while the same amount of energy delivered instantaneously will success- fully propagate the flame, due to a rapid rise in the temperature. This observation suggested that the hazard due to flammable fabrics is related to both the total amount of heat released and the rate of heat release. An isoperibol calorimeter was used to obtain the total heat re- leased. These studies suggested that the PET/TPPO system should ex- hi bit the best flame retardancy properties because of the rapid de- crease in the heats evolved. But a study of the flammability of these systems using the match test- had shown that the PET/TPPO/nylon 6 system was the superior flame retardant system. These results indicated that in vapor phase flame retardant systems the total heat released must be considered along with the other experimental data such as the flamm- abil ity test results. This apparent discrepancy was resolved when the rate of heat re- 62 lease was considered. These results showed clearly that the ternary system had the lowest rate of heat release. These data suggested that for vapor phase active additives the rate of heat release was a more important parameter for determing flammability than the total heat released during combustion. Because of the nature of the flame re- tardant activity, these conclusions should be as valid for cotton/ polyester as they are for 100% polyester. Although wery little mechanistic data is available concerning cotton/polyester blends, it appears that this is actually the case. Most of the flame retardant systems which show effectiveness on blends seem to act on cellulose in the condensed phase and polyester in the vapor phase. None of the studies to date have demonstrated this un- equivocally; but all of the data are compatible with such an inter- pretation. For example, Loss, Hofmann and Nachbur (43) have compared the action of a phosphonate and a phosphonium compound on cotton and blend fabrics. They found that both compounds had lower efficiencies on 65/35 polyester/cotton than on 100% cotton but the relative effic- iencies were different (Figure 24). The phosphonate exhibited higher efficiency than the phosphonium compound on 100% cotton whereas the reverse was found in the case of the blend. Although these workers did not attempt to explain their results, it seems likely that the phosphonate is more effective in reacting with the cellulose to pro- vide condensed phase retardancy. Similarly, the phosphonium salt should react more slowly with the cellulose thus allowing some of the material to volatilize and provide some vapor phase activity. Of course, this latter mode of action would be particularly important when the fabric contained significant portions of polyester. Similar conclusions can be reached on the basis of studies using phosphonates and phosphine oxides. Tesoro (44) has used oxygen index measurements to compare the activity of (CH 3 0) 2 P(0)CH 2 C(0)NHCH 2 0H and (CH 3 ) 3 P(0)CH 2 CH 2 C(0)NHCH 2 0H on 100% cotton and cotton/polyester blends In her studies the retardants were applied to the fabrics along with melamine resins and fixed by a pad-dry-cure process. Under these con- ditions she found the phosphine oxide to have considerably greater 63 CO CM 00 z + o. +-> 10 s- (1) > o CM E c o '•"^ m cr> 4-> 2 1 u .c •r— 10 i- •r— O -O ZJ o CM Ll_ C •r- ■!-> X CD M- o Ol O s- o M- 4-> c CD +-> E O o on u_ E E •C" c IE • CM LU en CJ3 64 effectiveness in both 65/35 (Figure 25) and 50/50 (Figure 26) blends. The reasons for this become clear from her findings that the efficiency of the phosphine oxide was the same on both 100% cotton and the 50/50 blend whereas that of the phosphonate was much lower on the blend than on the 100% cotton. The fact that the phosphine oxide was essentially substrate-independent is indicative of vapor phase activity while the finding that the efficiency of the phosphonate was directly related to the chemical nature of the substrate is strong evidence for con- densed phase activity. Similar results had begun to accrue from an investigation which was in progress at Clemson University (45). The efficiencies of triphenylphosphine oxide and (CH 3 0) 2 P(0)CH 2 CH 2 C(0)NHCH 2 0H were evaluated. It was found that the phosphonate had essentially the same effect regardless of whether or not it was chemically fixed in 100% cotton fabrics. However, with polyester/cotton blends the situation is quite different. Apparently the reactivity of the phosphonate toward cellulose is high enough to prevent its loss by volatilization when it is present in low concentration; but when it is overloaded as it would be in the cotton portion of a 50/50 or 65/35 blend some volatilization may occur. This could provide some vapor active species to retard ths polyester portion and account for the higher efficiency of the unfixed retardant on the blends. In support of this interpret- ation, a large difference was found in the effectiveness of the phos- phonate and triphenylphosphine oxide on 100% cotton fabric. But the two retardants seem to be almost equally effective on 65/35 blends. There was good reason to believe that the triphenylphosphine oxide exerts essentially all of its effect in the vapor phase. These results were therefore interpreted as demonstrating the importance of vola- tility of the flame retardants when polyester is included in the fabrics 2. Development of Potentially Commercial Flame Retardants Although the theory of flame retardant action on polyester/cotton blends had not been elucidated, a number of partially successful treat- ments were developed by a completely empirical approach. Most of 65 -e- o o O) -o IP M X o 0) -M O! TO c C •r- o -C JZ CL CL (/) o o © o L_L_i o CM ro o o O o o O o O o o o O o o cr> CO I s -. UD LO **• OO CM o CTi co n CM CM CM CM CM CM CM CM CM CM r— 00 c; c o +-> o u UJ LO CO LO LO o e: fO -o S- O) fO LO S- o Q. CO o xapui ue6/"xo LO CAJ 66 .310 .300 .290 .280 .270 .260 X | .250 I .240 ° .230 .220 .210 .200 190 180 ® Phosphonate - 100% Cotton O Phosphonate - 50/50 P/C A Phosphine Oxide - 100% Cotton D Phosphine Oxide - 50/50 P/C /v J L I i ±„ 1.0 2.0 3.0 % P FIGURE 26. Effect of phosphorus flame retardants on the 01 of 50/50 PET/ cotton blends. 67 these had rather severe deficiencies, but they could serve as good prototypes for further studies and were informative subjects for theoretical examinations. There are two general methods for obtaining flame-resistant polyester/cotton blend products. 2. a. Blending Fire Resistant Fibers: This approach is more long range than the other alternatives but it offers some special advantages as well as disadvantages. The flame retardants used in rendering the fibers inherently flame resistant should be the most efficient since they can be specifically designed for each individual polymer. Also, since the modifications are made prior to spinning in most cases there is greater opportunity to max- imize the durability of the retardant and minimize the undesirable effects of physical properties of the fibers. On the other hand, such specialty fibers frequently pose significant problems in such areas as cost and inventory managment. The production of such fibers can presumably be achieved by sev- eral routes such as copolymerization or grafting with flame retardant monomers or the inclusion of unreactive flame retardants in the spin- ning melt or solution. Copolymers containing flame retardant comonomers should be easily attained for polyesters as attested by the numerous instances of their use in polyester resins (46). However, the reason that these techniques are not readily applicable to fiber manufacture is basic. Polyester is a linear polymer which must be highly oriented after spinning. It is the basic morphology of these polymers that give the fibers the properties which have made them commercially ac- ceptable. The presence of foreign segments in the polymer backbone tends to disrupt the desired morphology, often to such an extent that the fiber attained has completely different and less desirable proper- ties. Several comonomers also introduce unwanted rigidity and brittle- ness into the fibers. Thus, if the copolymerization technique is to be successful, comonomers must be selected which either do not change 68 the properties of the parent fiber or which have new desirable proper- ties of their own. This has been partially achieved in at least one case and a semi -commercial polyester is now available containing a small amount (less than 10%) of a brominated bisphenol-A to impart marginal flame resistance. It was thought that this fiber might be suitable for blending with inherently flame resistant rayon or cotton. Flame resistant copolymers are, at least in principle, also at- tainable by grafting reactions. This route has had some success with cellulose but was practically untried on polyester, although there were reports of successful grafting of other types of comonomers on poly- ester (47). The dispersion of a flame retardant in the spinning solution or melt prior to extrusion often circumvents some of the property losses inherent in copolymerization techniques. This method has met with some degree of success as exemplified by FMC Corporation's PFR rayon (48, 49). In this method, halogen-free alkoxy-phosphazene mixes throughout the rayon fiber in small packets or sacs; and remains stable throughout processing. With proper processing they are also not harmful to properties of the parent fiber. This approach has also been tried on an experimental basis with numerous retardants in poly- ester. In this case there are stringent requirements on the thermal stability of the additives since they must remain inert in the poly- ester melt. Of course, for the production of flame retardant cotton the only possible approach is through chemical treatment of the yarn or loose stock. Although essentially all of the treatments developed for cotton fabrics could be adapted for such purposes, this has not yet been done on a commercial scale. In what may ultimately prove to be a more practical approach, several investigators have shown that a reduced level of flame retard- ancy is obtained by blending a flammable fiber with one having inher- ent flame resistance. The level of flame retardancy may be greater or less than predicted from the composition of the blend. Thus the flame resistance of PFR rayon blended with cotton is reduced less than 69 may be expected from consideration of the blend composition (50). This effect is ascribed to the fact that the flame retardancy in the rayon is equally good for cotton and rayon and to the fact that the flame retardant effect imparted to rayon or cotton required progressively greater amounts of fire retardant to achieve higher degrees of flame resistance. In another case, cotton fibers treated with the THPC-amide retard- ant to an oxygen index of 0.34 and blended 50/50 with untreated cotton with an 01 of 0.18 are reported (51) to produce a blend product having an 01 of 0.27, which is what one would expect from calculations. When the same fire retardant cotton fiber was blended with untreated poly- ester having an 01 of 0.20, a 50/50 blend product exhibited an 01 of 0.27. In another experiment, a predi eatable degree of fire retardancy was also obtained when polyester fibers were treated with tris(2,3-di- bromopropyl )phosphate to an 01 of 0.29 and then blended with untreated cotton having an 01 of 0.18. The resulting product had an 01 of about 0.24. The 01 values of these particular blends are essentially the numerical averages obtained from a consideration of the oxygen index values of the respective fibers and percentage composition of the blend. These results indicated that if sufficent fire retardant of the proper type could be put in either cotton or polyester, blends of these pro- ducts could exhibit good fire retardance. However, this has not been the usual observation made by various researchers. This method appears to work satisfactorily when the flame retardant fiber is present to about 65% or greater concentration. Many flame retardants for cotton are not as good for polyester/cotton blends as the THPC-amide and the tris(2,3-dibromopropyl )phosphate cited above. 2.b. Chemical Aftertreatment of Polyester/Cotton Blend Fabrics : With the present state of technology, treatment of blend fabrics was considered as the most feasible commercially, particularly when cotton and polyester are both present to an extent of over 25%. Two approaches for the treatment of blend fabrics had been studied - use of a single fire retardant and use of two retardants in a single process. 70 Use of single component treatments has been most effective for the flame retardation of blends containing a large proportion of a single fiber. Thus, flame retardants designed specifically for use on polyester have their maximum effect on blends containing 80% or more polyester. Similarly, a number of the flame retardants which have been designed specifically for use on cotton can produce accept- able degrees of fire resistance in blends containing 70% or more cotton. This latter case has been studied by Hendrix and co-workers (6, 7) using oxygen index and a variety of environmental temperatures to char- acterize the flammability of fabrics treated with THPC-APO, a fire re- tardant specifically designed for use on 100% cotton. As seen in Figure 27 the 01 values indicate that the retardant system is much more efficient on the 100% cotton than either of the two blends. The rea- son for this seemed to be connected to the fact that the flame retard- ant cotton component was less flammable than the ineffectively treated polyester, thus it was the polyester in this case that dictated the limits of flammability. The convergence of the 01 values for the flame retardant cotton with those obtained on the blend fabrics at higher environmental temperatures simply reflected the inefficiency of the flame retardant in maintaining its effectiveness at these higher environmental temperatures. This was further illustrated in studies using the THPOH/amide finish (Figure 28) or the THPOH/ammonia finish (Figure 29) (6, 7). One attempt to design a single compound capable of imparting flame resistance to the entire blend was reported by Tesoro and co- workers (51). They designed cellulose reactive phosphorus compounds containing sufficient bromine to reduce the flammability of both fiber components with a. single flame retarding chemical. The application of N-methylol-bis(2,3-dibromopropyl )phosphonopropionamide to 50/50 poly- ester/cotton with an amine hydrochloride catalyst from water/dioxane solution in a conventional pad/dry/cure/wash procedure was reported (51). Although the bromimated compound appeared to be insolubilized in high yield after such an application procedure and the 01 was sub- 71 c o c o I «3 -Q O c o I TO XJ jig cH5 5^ 00 00 o o LO CO o o • en r>. s- s- > >> o o C E C o o o +■> +J 4-> 4-> +-> +J O O O O O CJ flJD u 1 o D- I o Q_ o LO <3- o ID CO o ID C\J xapuj u96Axo o •1— ID 5 CM X> CD 4-> fO a> s- X— s ^-^ O CO o XJ O >» ID CD • p - 13 N O -»-> ^J- (G • S- CO CD ^""^ Q. 10 o Ij=~ s- -O >> M- CD O C S- o «4- LD E •r - JC co o Q. •r~ J= CO c O •r— •4-> O c LD »— 1 1 o 1 » in c: ^— ^ ^w^ o i j_ t. -o 4-> > >> -!-> 4-> 4-> +■> O o o O O o oo jo -Q o ro .a o I ■o •r— E «o o a. X o LO CO O IT) C\J o x: +■> •r— C\J -a .«—"■» •d O O J- 4-> **""" ft s- CM —J, ■a o 4-> >> LO A3 ">». r — s- N > 1/1 (J ■r- C s- E -O o -C f0 4- LO o CI Q. •r— XT 10 o •t— O ?> LO s- i 1 — 1 o • a) s- 4-> 03 s- cu a. cu xapuj ua6Axo 00 CM to 73 c c o o X> "O T3 T3 ro *o &§ ^5 eg ^f CM CM « «% o o oo id o o r^. Ln c > 'V .» o 1 S~ S- X3 CU O) •o +J +-> >n 00 CO QJ CD o^ >> >> *zt r— i— • CM o o Q_ Q_ c C C o O O 4-> +-> ■»-> +J 4-> -M o o o O O O A3 -Q Q (Di3 U CO o o ci- LO in: CXI 1— •i— XJ +-> ^""^ 4-> O o A **■*' CM O LO Q) XJ , — S~ >> 3 7 +■> N > S- (0 &- CD D. o IT) CO O ir> CNJ xapuj ua6Axo CU CM cc 74 stantially increased, it was shown by dioxane extractions that the compound was simply deposited in the fabric rather than reacting with the cellulosic hydroxyls as would be necessary for proper durability. The authors felt that this result, along with similar results obtained with other reagents of this type, reflected inherent difficulties with respect to solubility, distribution, and/or reactivity in finishing blends with reactive organophosphorus compounds of high bromine con- tent. In order to circumvent this problem an alternative procedure was used in which the cotton component was reacted with an unsaturated phosphonate, and the unsaturated groups then post-brominated (51). Thus fabric samples were reacted with N-methylol-3-(diallylphosphona)- propionamide in a heat-cure procedure using an acid catalyst and soaked in a solution containing a three-fold excess of bromine in chloroform. Low efficiency of the initial cellulose reaction with the unsaturated phosphonate (30-40%) was the limiting factor in this procedure al- though relatively high bromine contents could be obtained in the post- bromination step. Samples were reported containing up to 1.1% phos- phorus and approximately 5% bromine and having 01 values of 0.258. Phosphorus and bromine contents decreased about 25% in 25 launderings but the 01 was not seriously affected. These results were compared with those obtained using fabrics in which the polyester portion was specifically treated with tris-2,3(di- bromopropyl )phosphate and the cotton specifically treated with N-meth- ylol-3-(dimethylphosphono)propionamide. It appeared that such specific treatment produced a somewhat more effective flame retardancy since less reagent was needed to react a given 01 value. However, it was not clear whether this was a general phenomenon or one specific for these systems. A more successful single component treatment was that of Ciba-Geigy, A.G. based on the use of an oligomeric phosphonium salt prepared by the selfcondensation of THPC in anhydrous medium at ele- vated temperatures (43, 52, 53). Such oligomeric phosphonium salts applied together with a melamine resin by a pad/dry/cure technique to 75 polyester/cotton blends containing up to 70% polyester were claimed to result in washfast flame resistant finishes. A phosphorus content of 3.5% and 1.7-1.9% nitrogen in the fabric is recommended for durable flame retardancy. The efficiency of the phosphorus fixation was approx- imately 70%. Most of the finish was believed to be absorbed within the cotton fiber, but electron microscopy demonstrated that a portion of the finish was deposited on the surface of the polyester; but this portion showed good adhesion even after multiple launderings. Because of this surface deposition, the hand of the treated blends was firm; and some type of mechanical breaking was therefore recommended. Although these single compound systems were attractive for several reasons, there were some inherent problems associated with such treat- ments if only one fiber was affected. When all or nearly all of the retardant must be put in one fiber in order to provide retardancy for the total composition, it frequently happens that the fiber which is treated becomes overloaded and the physical properties are changed substantially. The fabric then becomes stiff or other aesthetic properties are changed adversely. Treatment of polyester/cotton blend fabrics with a single formula- tion containing two flame retardants appeared more practical. In this approach one flame retardant could be used which was especially durable and effective on cotton and a second could be used which was durable and effective on polyester. For such a system to be suitable, the retardants must be compatible in a single formulation and their cure or fixation processes must also be similar. The three processes des- cribed below represent significant progress in this area. Bromoform Adduct of Hexaallylphosphazine with THPC-amide This retardant was a one-bath formulation which could be applied to polyester/cotton blend fabrics with some success. The formulation was an emulsion which was applied by a pad/dry/cure process (54). The THPC-amide retardant penetrated the cotton fiber and during the heat cure formed an insoluble polymer and reacted to a modest degree with the 76 cellulose of the cotton. The bromoform adduct was applied to the blend fabric as a low molecular weight polymer but further polymeri- zation occurred during tire heat cure. Also, the polymer coated both the cotton and the polyester fibers. The bromoform adduct of the hexaallylphosphazine was prepared by introducing bromoform with hexaallylphosphazine during polymerization of hexaallylphosphazine using a peroxide catalyst and polyvinyl alcohol to emulsify the system. During the reaction, bromoform added to some of the allyl groups of the hexaallylphosphazine. Other allyl groups participated in vinyl polymerization and some allyl groups remained unreacted and were available for further reaction on the fabric during the heat cure process. The emulsion contained about 30% by weight of the brominated phosphazine. This product was in an aqueous emulsion which could be added directly to aqueous THPC-amide flame retardant solutions, with which it was compatible. This combined flame retardant exhibited fairly good properties and seemed to merit further develop- ment. Its primary deficiency was the durability of the adduct. Thus, rather high add-ons were needed on 50/50 blend products to withstand 50 laundry cycles and pass the FF-3-71 test. The amount of bromoform adduct needed was proportional to the amount of polyester in the blend fabric. The higher concentrations of the bromoform adduct were required for 50/50 polyester/cotton blends. Somewhat less of the adduct was suitable for fabrics containing 35% or less polyester. It was found that fabrics weighing about seven ounces per square yard and containing not more than 50% polyester should contain about 20% add-on of the combined retardant in order for the fabric to pass a vertical flame test. Treated fabrics sometimes exhibited slight stiff- ness. The stiffening was caused mostly by the bromoform adduct; ideally prepared emulsions could cause considerable stiffness. The THPC-amide portion of this retardant which penetrates the cotton fiber was exceedingly durable to both home and commercial -type laundering as well as to drycleaning. The adduct polymer had moderate durability. It coated both cotton and polyester fibers. Some of the adduct was lost 77 gradually during each successive laundering, but a substantial part remained through 50 laundry cycles. THPC-amide with tris(2,3-dibromopropyl )phosphate This retardant was another one-bath process applicable to blend fabrics containing up to about 50% polyester (55). The emulsified combined retardant was applied to fabric by a pad/dry/cure procedure. The fabrics were padded to about a 70-80% wet pick-up then dried at about 95 C and finally cured three minutes at 160°C. Fabrics impreg- nated with this retardant, cured and washed frequently yellowed when treated with sodium hypochlorite bleach solution. This was due to the presence of mel amine. Heavy fabrics had good aethetic properties. However, three and one-half ounce and lighter materials generally ex- hibited some stiffness. Both the stiffness and discoloration would be essentially eliminated by omitting the trimethylolmel amine and using instead di methyl oldihydroxyethyleneurea, dimethyl ol propyl eneurea, or dimethyl oluron. It took twice as much of these materials as of the trimethylolmelamine. This was due to the greater polymerization efficiency of the trimethylolmelamine and to the fact that melamine imparts a greater degree of fire retardancy per unit weight. These formulations were suitable for treatment of polyester/cotton blends in which polyester content does not appreciably exceed 50% of the fabric weight. The THPC part of the retardant penetrated the cotton and formed an insoluble polymer and was exceedingly durable. A small amount was deposited on the surface of the polyester fiber and was lost during laundering. The tris(2,3-dibromopropyl )phosphate penetrated both the cotton and the polyester "fiber and was equally good as a fire retardant on both fibers. However, the DBPP was readily removed from the cotton fiber during laundering but was durable in the polyester. Loss of the phosphate from the cotton during laundering meant that only about one half of the bromopropyl phosphate remained in the textile structure to contribute fire retardancy after about fifteen laundry cycles. 78 The bromopropyl phosphate was found to be just as effective on 100% cotton as it was on 100% polyester fabric (56). In both cases it raised the oxygen index about 0.05 units. This effect was accomplished with about 11.5% add-on and without the THPC-amide retardant. With 15% THPC-amide and without the bromopropyl phosphate, the oxygen index of cotton is about 0.27 and gradually drops as the percentage of polyester is increased. Addition of 11.5% of the bromopropyl phosphate to 13% THPC-amide raised the oxygen index of 100% cotton to about 0.35. Again the oxygen index gradually decreased as the polyester content was in- creased. This parallels the activity of the THPC-amide finish alone. THPC-urea plus polyvinyl bromide A fire retardant based upon a blend of THPC-urea with polyvinyl - bromide was found effective for fabrics containing various amounts of cotton and polyester fiber (57). This retardant was applied to fabrics by a pad/dry/cure process. The THPC-urea component of this formulation was developed primarily for cotton fabrics of various weights. It represented one of the least expensive fire retardants for cotton based on THPC. The polyvinyl bromide component was developed specifically for use with phosphorus-based fire retardants to extend their use to poly- ester/cotton blend fabrics. The polyvinyl bromide was produced in emulsion so that it was compatible with aqueous fire retardant formula- tions. The mole ratio of THPC to crystalline urea in the THPC-urea fire retardant had a significant influence upon durability to laundering. Since the polymerization functionality of THPC is three and that of urea is four, one would expect the greatest insolubility of the polymer and greatest durability of retardant would occur when the mole ratio of THPC to urea is about 1.3:1. This was determined to be approximately correct. During the course of the reaction of THPC with urea, one mole of formaldehyde was released which could combine with the urea. Thus a preferred mole ratio of about 1:1 was expected. This would be in the middle of the experimentally observed durable ratios. Some of PCHpOH groups from THPC and NCH ? 0H from methyl ol urea (formed in situ ) 79 react with cellulose. These reactions were deemed important in aiding durability to laundering, and crossl inking cellulose through polymer structure and thus imparting wrinkle resistance to treated cellulosic fabrics. With a ratio of 1:2 the add-on was very good but the retard- ant was not very durable to laundering. At a mole ratio of 1:1 the add-on was just as good as with the 1:2 ratio and durability was ex- cellent. The durability seemed to drop some when a ratio of 2:1 was used but this was partly due to the lower add-on obtained with this excess of THPC. Di sodium hydrogen phosphate was employed in this treating system for several purposes. The main value appeared to be maintenance of proper pH for reaction of PCHLOH groups with NH groups of urea. The buffering also helped to prevent acid degradation of the cellulose. Some of the phosphate esterified the cellulose. The polyvinyl bromide concentration was also an important factor. Four percent polyvinyl bromide appeared to be inadequate when used with 26% THPC-urea, assuming that the fabric must withstand 50 laundry cycles. However, 6.4% of the polymer appeared to be adequate since it gave a char length of 4.5 inches after 50 laundry cycles. With 30% THPC-urea in the formulation the amount of polyvinyl bromide could be reduced to as little as 3%. All of the formulations produced fabrics with essentially no loss in breaking strength but tearing strengths were reduced about 50%. Other factors such as drying time and temperature, curing time and temperature and use of softening agents were also important. These have been recently reviewed by Reeves and co-workers (50). All of this leads to the conclusions that a great deal was known about flame retardance of cotton and polyester and that some beginnings had been made to develop treatments for blends by June 1974. However, a great deal remained to be done before commercially feasible systems could be formulated. 80 TESTING METHODOLOGY In attempting to evaluate existing flame retardant treatments for 50/50 polyester-cotton blend fabrics a primary concern is that of appropriate testing methodology. This is also a basic factor in trying to set goals for the research work. The flammability or flame resistance of the samples must be judged in a manner that produces information relative to the ultimate utilization and marketability of the treated fabrics. Since it had been decided that the fabrics to be studied would be those which are encountered in general apparel uses, this would require that the test methods be as closely related as possible to those which will be used to evaluate general apparel in the marketplace. However 9 it must be recognized that these mar- kets are completely reliant on government regulations and without such regulations there would probably be only a wery small market for flame retardant apparel. Thus the tests which are appropriate are those prescribed by the U. S. Consumer Product Safety Commission; but at present CPSC has not indicated exactly which test or method it will prescribe. For this reason, it was decided that dual goals should be established for the ETIP research project and all candidate systems were, therefore, evaluated in terms of their ability to produce fa- brics which would pass the requirements of the childrens 1 sleepwear standard for sizes 7-14 (FF 5-74) and which could meet the criteria for classification as class I on the Mushroom Apparel Flammability Tester (MAFT) as described in the recommendations made by NBS to CPSC in March, 1976. 81 1. Oxygen Index and 45° Angle Burning Tests Unfortunately, neither of the standard test methods is capable of detecting subtle changes in flame retardant effectiveness and thus neither is adequate for the needs of flame retardant research. Thus other more sensitive tests must be used in the research work but these tests must be re! a table to the MAFT and FF 5-74. The most obvious candidate for this would seem to be the oxygen index (01) method. It has been utilized extensively in the early phases of the ETIP pro- ject but it was soon found that, in its standard top ignition mode; it was not always an accurate reflection of the behavior which the samples would exhibit in vertical upward burning in either the MAFT or FF-5. Other workers have apparently reached similar conclusions recently. A subjective burning test was thus developed using a 2" x 5" sample mounted on a pin frame supported at an angle of 45 from the vertical and ignited on the bottom edge using a wooden kitchen match. This seemed to give important information on the overall flammability of the samples and proved to be a valuable adjunct to the 01. In several cases pairs of samples were found which gave similar results on FF-5 or 01 but which exhibited grossly different burning charac- teristics on the 45° frame; both would burn the entire length but one sample would burn rapidly over the width of the sample with a large frame whereas the other would burn slowly over a narrow strip with a small frame. These fabrics would obviously pose significantly different levels of hazard to the consumer. Presumably the difference in behavior between the 01 and the 45° of FF-5 tests is due to a difference in geometry of burning. Thus an attempt was made to develop a reproducible technique for measuring the oxygen index using bottom ignition (BOI) in a manner similar to that previously reported in the literature (58). That this can be done in a reproducible manner is indicated by the data in Table III. These data also indicate that some flame retardants are more sensitive than others to changes in geometry. This is re- 82 TABLE III COMPARISON OF TOP AND BOTTOM OXYGEN INDEX METHODS % P 01 BOI Bor 65/35 DAP 4.0 28.8 20.5 20.5 2.8 27.0 20.0 ---- 2.1 19.2 19.5 1.7 18.5 1.3 23.5 18.5 18.2 50/50 DAP 4.5 31.5 21.8 21.5 3.3 28.8 20.0 2.4 20.0 19.8 2.1 20.0 19.5 aze®19 1.4 23.5 19.5 19.0 65/35 Antibl 3.0 25.0 19.5 2.2 23.0 18.2 18.2 1.5 22.2 17.8 aze®19 0.89 21.2 16.8 50/50 Antibl 3.2 25.5 23.0 ---- 2.2 23.8 20.0 19.8 1.6 19.0 18.8 1.2 22.5 18.5 18.2 0.9 _ 18.0 17.5 BOI - second determination at least 24 hours later. 83 lated to the mechanisms of action of the retardants and is discussed in detail in later sections. A number of reference compounds have now been studied in this fashion and the results of these studies are given in Table IV. Also included in this table are observations based on 45° burning tests and isoperibol calorimetry. On the basis of this limited data set, it appears that oxygen index using BOI is a more accurate indicator of the way a sample will perform in normal atmosphere burning tests. The limit for burning in the 45 angle test as it is performed at Clemson has been found to be between 19.5 and 20.5 for BOI as com- pared to 25.0 and 28.0 for normal oxygen index using top ignition. Because of the flow environment, isoperibol calorimetry appears to be an even more stringent test than the 45° angle burning. The limit for burning in the isoperibol is apparently indicated by a BOI be- tween 21.0 and 22.0 with the samples studied to date. These data indicate that there are two different types of flame retardant response as indicated by 01 and BOI. In the first category, represented by retardants such as dibromopropyl phosphate and DAP, the 01 increases much faster than the BOI as the add-on level in- creases. In the second category represented by compounds such as N-methyl ol -3- (di phenyl phos^hinyl) propinamide QJD3P) fixed with a melamine resin, PyrovatexvV3762 and AntiblazeQ9l9, the 01 and BOI increase at approximately the same rate. It is interesting to note that these latter compounds have been found to have a much smaller effect on the heat release values as measured by an isoperibol cal- orimeter than those which fall into the former category. 2. Calorimetric Measurements Both static oxygen bomb (SOB) and isoperibol calorimetry have been found in previous work to constitute powerful probes into both mechanisms and efficiency characteristics of flame retardant action (59). In an attempt to utilize these techniques more fully in the present work, efforts were made to develop a more complete interpre- 84 2 Q. O flu dS cq qq oo c c z z as z j_ j_ o Q —< Q 3 3 "8 OO S ■s CO "8 c zs a: c C Z Oi c c t. Q »— » s- S- Q ►—i s- s- 5 3 3 -O 3 .3 o in a * oo co — c c z z i. i. a a 2 E CM CO CO CO CO "8 CO CO Ct z z cc: z z Oi c z z Oi c Q Q Q Q 3 Q Q s. 3 I/O Q O mcoocow>oooOLf>oommom o lO .............. CQ i— lT>r— OCTiOOCTii — CTi CO O CO 00 r^ «— CM CM i — CM CM <— C\J i — i — i — CMi — i — c — co oo co to r^ CO i— <— r— CO CO >x> CO r*. *» CT1 lO LD oo O r^ i£> *3" CM CM CM CM CM CM CM CM CM OO CM CM CM I— UJ 00 —I in o co m OO «3- o CO r— l£> oo m CO in in oo 00 co in <*• CO CM ^J- CM CM <— — o o .— .— o o ■•-> o. in o o o UJ 5 >. O- O s- D. § J- — i s- i— i 3 3 o in -M o u s s I CO % S eg "8 ■s c c c 2: Ct c c z ae c C s- t- J-, Q *—t t- s- o »— « 1- U s 5 E J 3 -Q 3 J CO 8 8 CO 8 IS CQ ca S CD CO s 2: sz QC c z a: c c ^~ r^ C£ C Z z cc c O 3 a L. E O Q " o o ~ 5- 2 o CM 00 CM co cr> r>» o eo co «•» cm i— co >— ocnco CM CM r- f— CM CM I— I— t^l in CO CM CM r*. If) Ifi CM VO <£> in CO o m O O U ) O m o m O in m io in l£> in in CM IV. CM in {SI m o csi CM VO CM > o •r- C Q >- r- CM 86 tation of the calorimetric data in terms of the general principles of flame retardant action. It has generally been accepted that the burning of polymeric materials is a cyclic process. A simplified diagram of this cyclic burning process is schematically shown in Figure 30. The process starts with the initial heat input (ignition source) to the polymer substrate, which raises the temperature of the substrate and triggers its thermal decomposition. Combustible gases which are generated as the result of the degradation of the substrate combine with the proper amount of oxygen from the surrounding atmos- phere and, with proper ignition source, a rapid oxidation process (combustion) takes place. As the result of the combustion process, heat is given off, although a fraction of the combustible gases escape without being completely oxidized. Part of the heat generated from the combustion process is transferred to the surrounding environ- ment, while the rest is fed back to the substrate to further promote the pyrolysis process and thus completes the cycle. There are two distinct types of processes involved, those in the substrate which can be regarded as fuel generating, and those in the gaseous phase which can be regarded as fuel consuming. Thermodynamically, there are two distinct parameters associated with these two processes (Figure 31 ) : 1) the heat value of the combustible gases generated (AH«) which is associated with and is essentially controlled by the fuel -generating processes and, 2) the actual heat released (AH-j) which is mainly controlled by the combustion (heat consuming) pro- cesses. It is generally believed that there are two types of chemical flame retardants: 1) a condensed phase active retardant which is effective in the substrate and reduces the amount of combustible gases generated, presumably by retarding or altering the normal degradation process and, 2) vapor phase retardant which is active in the gaseous combustion process and reduces the actual amount of heat generated, presumably by retarding the oxidation reactions through free-radical termination and radical recombination processes. The effect of each of these two types of retardant on the entire burning process and particularly on the two parameters mentioned above, can be 87 ( FLAME ) N .COMBUSTIBLES (AH 3 ) COMBUSTION COMBUSTIBLES (AH 2 ) HE HEAT (AH^ viiii/uiii&uyiiwjujA PYROLYSIS FIGURE 30, Diagram of polymer fire. 88 POLYMER (s) cc I— < C0 2 (g), H 2 0(g), RESIDUE (s), COMBUSTIBLES(g), SMOKE (s) (AH°) p 2 , PRESSURE AH. (AH°) p = AH 2 + R (AH°) R AH = AH, + AH 2 1 3 C0 2 (g), H 2 0(g), OXIDES OF HETERO- ATOMS o O I CO CM CO CO r>» «3- o o o •vT CM co ro CO CM CO CO CO CO CM CO CO «5i- r-^ LO r^ lo ifO CO CO »* CM LO LO «cr 00 oo CO o CTi r-<» r>>. 00 r— j— CO CO CM CM CM CM CO CM LU _J CO < oo CO O o o o OO LU O CO t— I oc LU D_ O OO E (J I o CD CO 03 < C£ »— i OO LU cn .— CM CM *d- o LO CM LT) OO ■^ co r— LO CM «tf- LT) N N N lO U5 U1 <* <* LO «d- i— 00 CM LT) «3" LO LO OO CM CO CO LO f^ LO OO LO LO co r-^ r- «3- o o LO o LO o o o o o o CO LO CM LO o LO LO CO o CM o S- cn oo co 03 en i- -Q •o CO ■p s- o • Q. LO Q. CU 3 -C tf) o +J • cu •r~ cu CO 4-> N o CO Z3 03 +-> CU en cu •r- CO S- 2 CO 03 03 ^— r - CO 03 CJ> CU +-> &_ CO cu cu CU sz E X) +-> CO 4- c cu -c -C i- +-> -M f0 • (— •r— Q- 3: s sz •o XJ •p~ cu cu 4-> 4-> 00 s_ %. cu o o 3 Q. Q- F Q. Q. OJ 3 Z3 > OO OO i— CM CO 95 5000 4000 3000 E 2000 1000 O control (with support) # CONTROL O 0.5% D 1.0% A 1.5% O 2.0% © PET/GLASS BLEND J. 1 20 40 60 % Cotton 30 100 FIGURE 32. Heat release of PET/cotton blends burned with and without support. 96 The magnitude of the effect due to the grid appears to decrease with increasing levels of condensed phase active flame retardant on the cotton portion of the blend. This can be seen from the results of a series of investigations on fabrics treated with increasing le- vels of diammonium phosphate with and without the fiber glass support. These results are tabulated in Table VII, and the heat release results are shown in Figure 33. The presence of the glass support can be seen to increase the heat release by approximately 500 calories. However, this difference decreases with increasing levels of phos- phorus in the cotton and there is essentially no difference for sam- ples with more than 2% phosphorus content. It would appear that this is because there is enough char formed from cotton containing 2% or more phosphorus to support combustion of the polyester and prevent any melt flow. The additional support provided by the fiber glass does not seem to be of any significance. The effect of the grid upon the rate of heat release appears to be more complicated. Visual observa- tion indicates that the flame does not break through the glass grid and thus only the top half of the calorimeter directly receives heat from the burning sample. This could cause the recorded time-tempera- ture curve to shift up the time scale, and would result in a lower rate of heat release, as observed. Such behavior is different from that observed with the untreated blend when there was initial delay followed by the intense heat release from the burning sample breaking through the glass grid. In order to explore the nature of these effects more deeply, samples of 100% cotton and 50/50 polyester/cotton blend fabrics were treated with antimony oxide, decabromodi phenyl oxide and an acrylic binder (FR-P-44 from White Chemical Company) and evaluated in the iso- peribol calorimeter. The results obtained using the fiber glass support are given in Table VIII. These heat release values were plotted versus calculated % bromine as shown in Figure 34. Previous results without the support are also included for comparison. The effect of using the support was found to be an increase in heat re- lease in these two series. This is the opposite of that observed 97 E «3 O CX> r— 00 ^Q r— Q"> f— «3" *fr CO CO co cm cm c\j l^» r— 00 S M tO O O 00 CM CM r— < ) CM ^ ^D r^» O^ ** a> *t cvji CM CT> un- -.-O > =~ CTi CM r 3 o co CT> CO CM CM CM CM CM r— o cj> co UJ >- o o_ o LO O LO LU * s u 5 u cu yj u cu CM vo <«D Li";; CO o CO r— o CO ■J*; <£» r— LO I/) LO an <5? LT:> CO CM s. CJ> t/i in *rs en i t- i — r~ p— r— CM un ^- LO p^ CTi O >0 CO CM LO CM oo LO CM -a cu +-> i- o CL O. CO a> to o cu s- (O 1,0 ■I— ai -C +J c cu s- (O a. i— CM CM CM p— O O r». • — LO CO LO +J O r— r— CM CM c o o CO oo CU =3 (TJ * 98 4000 3000 1X3 U < 2000 1000 O No support A With Support 1.0 2.0 % Phosphorus 3.0 FIGURE 33. Heat-release of di ammonium phosphate treated ETIP blend burned with and without fiber-blass support. 99 a: o a. to I to « - . _) CD I E < O ea- rn < «— < UJ —I t— UJ QC —J CO X o 3: ,r.v. I — 1 O o £ o c£ CO O O o a: O at ,-<5 a; S- "S >» r- c to ■J— C d* •0 00 CM 00 0> IT) r** r». 00 10 eft 1— VO r— ■— a> «* »— cm o r^ o in 00 en 00 CO £ $ co cm »o in ai «*•«*■ 00 r>» «* cm CM CM CM O * 0> W r- to r— *? co m *»- 00 «*■ c* 10 CO CM CM f— r— CT> CM 00 OO 0O in r— 1-^ r^ m 10 CO 40 10 »— O 00 10 «? B (VJ 00 04 **■ v0 CM CO CM en m r»* CO 00 *»■ O CO VO f*. r— CM 00 ro 00 en ca 1 co 1 00 CM in O CM «* ** UTS r~ CM *« W «* «*• h* i- 00 en in CM m ••-> • • • • c in 40 C* CM *»■ r— ^~ o o J- CU +J a> JH "o o m o in % co CM O o 1^. 10 CO r- O i- ■•-> c o o 00 5t o 1— CM *»■ 100 3000 2000 o < I 1000 P-44 Treated w/o jjrid 100% Ct -50/50 P/C 65/35 P/C A o w/ grid % Bromine FIGURE 34. Heat release of P-44 treated fabrics. 101 with the di ammonium phosphate treatments. The DAP treated blend fabric showed a difference between the heats observed with and with- out support which decreased with increasing %? content in the fabric; there was essentially no difference with treatment levels above 2%P. However, current data on P-44 treated fabrics show a difference in heat release between the supported and unsupported samples which in- crease with increasing %Br content in the fabric. Similar results were also observed on P-44 treated fabrics of 100% cotton where addi- tional physical support from the grid is not expected to have signi- ficant effect on the heat release. These results tend to suggest that the presence of the grid may have caused premature loss of flame retardant by vaporization. The grid, acting as a thermal shield could channel the hot convective current upward ahead of the flame, thus increasing the pre-heating of the fabric and causing the vapor- ization of flame retardant ahead of the flame zone. The net heat reductions, AH-,-(AH-, )„„„.,. „ nl , of these series are II 'control shown in Figure 35. Previous data obtained without support showed that P-44 has identical efficiencies on cotton and the blend; the net heat reductions from both cotton and the blend showed identical dependence on the %Br content. However, in the current series with the support, P-44 seems to be more efficient on the 50/50 blend than on the 100% cotton. This could be due to the melting of the polyes- ter in the blend which might cause the fabric to sag and make contact with the grid. Although the contact would provide additional physi- cal support to the burning polyester and thus increase its combustion, it would also hinder any upward convective current between the fabric and the grid. There is essentially no char formed by these fabrics which is consistent with the earlier indications that P-44 is a completely vapor-phase active retardant. The effect due to the physical support provided by the grid is thus expected to be independent of the retard- ant content in the sample for a given type of fabric. For the blend, the difference in heats observed with and without the grid is con- stant up to about 6%Br. This is evidently due to the physical 102 2000 E 1000 o o P-44 Treated 100% Ct w/ o grid O w/ grid • 50/50 P/C A ▲ 65/35 P/C O /S O ' **/ % Bromine FIGURE 35, Net Heat Reduction of P-44 treated fabrics 103 support provided by the grid. However, for 100% cotton, this differ- ence consistently increases with additional bromine content in the fabric. 3. Correlation of Test Methods A major area of concern in the testing program has been that in- volving the lack of correlation between the calorimetric data and the results of burning tests such as a vertical test (FF-5), the MAFT, or the subjective 45° angle test. While those fabrics which would not burn in the isoperibol calorimeter always were found to pass FF-5, the re- verse is frequently not true. Many samples were found to perform well in FF-5 but burned completely in the isoperibol calorimeter with the evolution of relatively large amounts of heat. It would appear that these discrepancies result primarily from the very severe ignition conditions employed in the isoperibol technique. A study of the igni- tion behavior of flame retardant blend fabrics was therefore under- taken. A small ignition chamber was designed and constructed to study the effect of ignition time and the difference between edge and sur- face ignition. This is shown in Figure 36. The ignition source chosen for the chamber was a 1" propane flame extending from the tip of a 22 . gauge hypodermic needle. This design should result in ignition condi- tions somewhat similar to those proposed for use in the MAFT. To test the chamber design and the reproducibility of the results obtainable with the device, a sequence of 12 blend fabrics was tested. These represented treatments with diammonium phosphate at two levels, the vinyl chloridevinyl bromide copolymer P(VBr/VCl) at two levels, and Pyrovatex^CP (with no resin). All of the samples were exposed to edge and surface ignition for 1, 3, 5, 7 and 12 seconds. The re- sults are given in Table IX. As the data clearly show, there was little difference between edge and surface ignition within this small sampling of flame retardants. It does appear, however, that in the case of the totally condensed-phase active diammonium phosphate, sur- 104 (a) Sample Mounted for Edge Ignition (b) Sample Mounted for Surface Ignition, FIGURE 36. Ignition Exposure Time Tester. 105 H pq Eh W co o P-. Q is, O H Eh H o o w O o o O I— I EH O o Eh Eh O O K P EH co w >H hH o Ph o LA o LT\ Ph H EH CO S o M Eh O w o o CO OJ H o CO CO o CO LA O 0.1 CO oo CM VO O CM • • t- t— vo c\j -H- • tt m +1 +1 +1 -H- 00 o • • • Lf\ LA t— t- OJ CVJ LA +1 +i +i o, OJ t- • m ■ ro -=r LT\ LA OJ ON +1 +l +1 H o LA ■ * « on oo OO LA -H- _j- +1 +1 +1 CO 00 H H H H pq CO PQ VO LA -H- o o p o O o ^ fl Ph • • M O M •H «H OJ +1 +1 0) -p - — -. > o -H- CO c— cd w o H K U s OJ rH — • • •H CO (U !h O a O X! 1) P a? > O 106 OJ CO 00 +1 oo CO oo fO H pq oo oo o +1 vo _H- OJ OO on OJ vo LA VO • + 1 OJ oo +1 +1 -H- H CO > Ph Ph OA O H 00 -H- _h- 00 00 -H- 00 0O -H- _H- 00 oo +1 LA on LA 00 • • +1 +1 On CO oo OJ oo 00 o +1 o pq LA O O +1 CO 00 00 H CO pq co o o +1 vo H OJ Xi P W CO H CO !h •H -P Xt bO nd W a o -p p o pq p CO H CO Sh •H P CO OJ fl CO cd H Ph CO X •H CO H C! o •H P •H CO o o5 CO OJ face ignition is slightly more severe; whereas the reverse is true of the vaporphase active P(VBr/VCl). While the char length reproducibil- ity was found to be excellent for the DAP samples and the second set of VBr/VCl copolymer fabrics, some difficulties were encountered with those fabrics which tended to exhibit borderline burning characteris- tics. This was not surprising considering the complex nature of the ignition process. On the basis of these preliminary trials it would seem that once a large enough data base is accumulated this test should provide valua- ble information concerning the mechanism of flame retardation. As a further example of the type of information available, some additional ignition exposure data for several fabrics treated with selected THPC precondensate finishes at various levels are presented in Table X. In an attempt to establish direct correlations, calorimetric data and MAFT test results on some 40 fabrics have been obtained (mushroom data courtesy of NBS). These fabrics included both natural and synthetic fibers, with and without flame retardant treatments. Both the calorimetric and the MAFT data are tabulated in Table XI and XII. The calorimetric data are shown in terms of both the heat re- lease, AH, , and the heat release per unit area, -Q, . The latter para- meter takes in account the effect of the fabric weight. Both vaules, -Q, and -AH,, were plotted vs. mushroom results, as shown in Figures 37 and 38, respectively. Figure 37 shows that there is no direct correlation between -Q-j and the mushroom. Furthermore, there is no clear-cut value of -Q-, , 2 which corresponds to a passing margin in the MAFT, 0.10 cal/cm -sec. In addition, the data in Figure 37 indicate that MAFT results are relatively insensitive to difference in fabric weight, i.e., fabrics of same fiber type tend to have similar results, regardless of their differences in fabric weight. A plot of -AH-| vs MAFT results, as shown in Figure 38, also shows no direct correlation between the two values. However, a cut-off value of 1800 cal/gm for -AH-j can be chosen, where no fabric with -AH, less than this value has failed the MAFT. Although, there are a few fabrics with -AH, larger than 107 • TABLE X CHAR LENGTH AS A FUNCTION OF EXPOSURE TIME FOR ETIP 50/50 POLYESTER/COTTON WITH SEVERAL PRECONDENSATE BASED FINISHES Exposure time, bottom ignition Finish, wt. % 3 sec 5 sec 7 sec 12 sec THPC/MeNH 2 .188 .154 .129 THPOH/MCC-lOr" 1 .229 .179 .131 THPC Acrylic .190 .252 Pyrovatex 3762^ ^ >.250 THPC/MCC-lOOvV ' ' 2.8cm 10.0cm 8.6cm >.250 THPC/ Pyrovatex 4013^ 2.9cm. 6.9cm 8.2cm >.250 THPC/Diamine^ 1 ' 3.5cm 6.8cm 8.6cm >.250 (1) Pass FF-5 (initial) 2.5cm 4.5cm 7.4cm 8.1cm 3.0cm 5.2cm 8.3cm 8.6cm BEL 8.3cm 9.0cm 9.4cm 2.7cm 6.5cm 7.5cm 6.8cm BEL 10.1cm 10.6cm 11 .Ocm BEL BEL BEL BEL 4.0cm 8.8cm 8.6cm 8.4cm 4.4cm 7.3cm 7.5cm 7.3cm 108 GO o GO HH O X LU I— —I <£ CO Q O O x GO Q o DC o < c_> CO CMI O 03 a • 00 1 — O r-^ VO lo ^" VO r— O CvJ OO LO m CO 1 — CO CO C\J 00 OO 1— r— csj CM , — CO E § S- _C CO 3 £ GO (_) t— 1 cm cc E CQ O < Li_ A3 r-. O O ro ■"" VO ro vO C\J ro r«v O Q VO r— vo l-^ r^ CM 00 VO VO ^~ CM LO VO LU CO •^f CO ro CM LO vO CM CM CO LO ro CO t— A (_> 1 — cy E r0 u CM O E en u i_ .0 re •3" <3- *1- CM VO VO VO VO cr> CTi 00 00 co ro «d" *3" ** ^f O 00 00 00 r^ 1— 00 ro CO O CM LO «3- LO ro 1— cr> 1— LO CTi CM ro O^ CO *3- CM CO CT> LO cr> r»* c o +J +-> o c_> CU c c c o +-> +-> o (_) i- c o o CO o LO o LO C_) LO ro LO VO (_> LO ro LO VO o LO ro LO VO o o +J (T3 CU o < CJJ rO CU CU ro +-> CU (J c o 0J +-> (O +-> CU U < o CM O CO o 109 TABLE XII CALORIMETRIC AND MUSHROOM DATA OF F.R. TREATED FABRICS Mushroom Flame Retardant 2 wt. gm/cm -AH, , cal/gm 2 -Q, , cal/cm 2 cal/cm -sec COTTON Fyrol 76® 11.7 2170 27.6 .19 Fyrol 76® 11.7 FTB* .01 Fyrol 76® 27.4 1973 47.7 .23 THPOH-NH 3 11.7 1682 19.7 .01 THPOH-NH 3 11.7 1746 20.4 .01 35/65 PET/CT_ Fyrol 76® 13.2 2254 29.8 .15 Fyrol 76® 13.2 2222 29.3 .15 Fyrol 76® 13.2 2217 29.3 .01 THP0H-NH 3 13.2 2202 29.3 .18 THP0H-NH 3 13.2 1942 25.6 .01 THP0H-NH 3 13.2 1901 25.1 .01 50/50 PET/CI-. Fyrol 76® Fyrol 76® Fyrol 76® 23.9 2244 47.2 .17 23.9 2255 47.7 .13 23.9 2236 48.8 .11 THP0H-NH 3 23.9 2225 42.9 .13 TH PC /M&6J 00/200/ 300VlV(ETIP) 14.9 1911 28.5 .20 M 14.9 1842 27.4 .01 M 14.9 1778 25.4 .04 (.10) 50/50 PET/RAXDN Fyrol 76® 10.1 2339 26.3 .18 THP0H-NH 3 10.1 2168 22.5 .21 THP0H-NH 3 10.1 2268 22.9 .05 THP0H-NH 3 10.1 2098 21.2 .05 65/35 PET/CT Fyrol 76® 8.7 2189 20.5 .11 Fyrol 76® Fyrol 76® • 8.7 2338 22.1 .09 8.7 2261 19.7 .07 THP0H-NH 3 8.7 2134 19.2 .11 THP0H-NH 3 8.7 2070 19.1 .10 THP0H-NH 3 8.7 2036 17.7 .07 ♦Fail to burn 110 o 00 ■ to f0 o r»» c J- Oil PL o Q) *© m (0 0) OJ J- o re ir> o CO i CD -j id in laJ co i— ld CO o _l o CO en CO in , — ^3- CM • LU • = CM CQ 1 — ■ 1— r— C-J eg C\J i — i — eg LO CO yo m cr> cm i— CO CO CT\ uo ^f oo oo oo oo ^J- <3- CO LO eg co oo oo OO . CO ro ro co 0^ CO LU ^- *3" O O a. o. ck 3: dd DJ ^_3 DD DD DD DD DC dd E E cc CU CS CU CU CU ( QtTT ( QCJ co co co V / V_y 1 dd 1 dd 1 DD 1 1 IE i 3D 03 1 DD <0 1 DD E +-> o o o c_> c_> o o O o O O o O >>,>> dd dd a_ a. a. i— »— I— DD DD DD I— DD 1— DD DD 1— DD 114 -C -o 4-> c cz o d> •r— _l ■M ■r— S- ■o o c o c •r" o s © o a: >- S- i— fO CO -C -r- <_> >- X L CO o a: o CO c o I ■o < to CD >> o O u ft3 o C\J 00 C\J CM CNJ o C\J c\j CO O O c O O CO -t-> "^ -M O O o s- c o <_> c o o o LO m -t-> ~«^ ^■M o o o ir> r^. o 115 I -C 13 ■»-> <1J C7> c c o 0» ■ r— _J 4-> •r~ s» "O •-0 C JC o O O ■ c o o CO o Of T3 S- r— CO o >- * CO CSJ CO CsJ CT> CsJ LO OsJ X I ,u _J CO CO O N— I Of CO < o CO o c: o ■a ■a s- a» 4-> CO oj >> o Q. o 4-> O U CsJ lo LfJ CM CO Osl CO. OsJ CsJ CM O .a o LO o LO o LO o LO o CO o O co o 116 TABLE XVI FLAMMABILITY OF FABRICS TREATED WITH FR P-44^ % Char Char Length Fabric (cotton/polyester) Add-on 01 Yield Conditioned 50/50 --- 31.8 3% 1.9 in 50/50 --- 29.7 6% 2.1 in 50/50 30.8 29.0 5% 2.7 in 50/50 22.0 28.6 3% 4.2 in 70/30 --- 34.1 3% 1.9 in 70/30 — 31.8 5% 70/30 26.7 30.9 3% 2.2 in 70/30 21.3 28.6 4% 5.5 in 117 CO IS) c_> t— I CO LU h- oo LU >- _J o LU O c£ on LU - CO -C ^O 4-> cu en c c CD •p- __l +J • 1— s- •a 03 c -C C_> c o -a -a c c c c c c c 00 lt> c\j o co o r^ co CO LO 00 MD O O 1 — C\J 1 — 00 C\J CO CO CO CO CO 00 f— CO C\J CO ( © C O •r— X en E +-> © CT> 00 © O © © 3 E CD C CT> t— CNJ .a y—t 1 1 CNJ rO OH LU 2: _l -Q Ll_ Q_ Q S- 3 cd to C ■O -O ■a 4-> - r— 00 S- S~ s- 1/1 -M O 03 rO fC OJ • 1— S*. CD C7> \— +-> E c 4-> E S- 1 (T3 CD 3 C (T3 n3 O C S- fT3 O r— .c r— fO •r— CD O Ll_ O C£5 1— U_ CO 118 TABLE XVIII FLAMMABILITY BEHAVIOR OF POLYESTER SINGLF KNITS TREATFO WITH 15% TANATARD PN-2 Sample 01 Char Length, in. 150 Den. Textured PET TWILL 38.0 2.5 150 Den. Textured PET LACOSTE 35.0 2.6 150 Den. Textured PET CREPE 33.1 2.4 150 Den. Textured PET BROKEN RIB 33.5 1.8 150 Den. Textured PET CORD 35.6 2.6 150 Den. Textured PET ARGYLE 37.8 3.1 150 Den. Textured PET RIB 31.1 2.5 18/1 Spun PET JERSEY 33.5 2.1 18/1 Spun PET JERSEY 33.1 1.7 18/1 Spun PET RIB 31.4 1.0 150 Den. Textured PET PONTE DI ROMA 36.4 1.7 150 Den. Textured PET TWILL 35.9 1.9 119 POLYESTER 1. Basic Factors Affecting PET Flammability l.a. Pyrolysis of Untreated Polyester: In order to develop the type of fundamental information on the thermal reactions of polyester and treated polyester which seemed to be required for the design of flame retardants which would be effici- ent on polyester, and indepth fundamental study was initiated at the Polytechnic Institute of New York under the direction of Dr. Eli M. Pearce. The objective of the early studies was to investigate the re- lationship between PET structure parameters and its thermal degrada- tion and flammability. A series of specially designed polyesters was prepared by the research group at American Enka Company under the di- rection of Dr. Alan Meierhoefer. In all, 11 samples were made with systematic variations in molecular weight, carboxy end group concen- tration, and di ethylene glycol concentration (DEG) as shown in Table XIX. Complete characterization of the polymers was carried out by American Enka. TGA analysis of the polyester samples was conducted on a Mettler No. 13 thermal analyzer using a deep dish-shaped alum- inum crucible and weight losses were obtained in the range of 60-80%. This weight loss was considerably less than expected. In order to vertify the results, one sample was analyzed in a nitrogen atmosphere and the rate of heating was changed from 6° per minute to 15° per min- ute. There was no change in the observed weight loss. Subsequently, the other samples were analyzed on a DuPont 950 thermal analyzer using an open boat-shaped crucible and heated in circulating air. Losses under these conditions were considerably higher, and generally in the neighborhood of 90%. In view of these differences in weight loss, one polyester sample was analyzed on the Mettler equipment using a flat dish platium cruc- ible. The rate of heating varied from 6° to 20 per minute. Weight loss was found to be 87% indicating that more degradation had occur- 120 TABLE XIX POLYESTER SAMPLES PREPARED FOR FLAMMABILITY STUDIES Polymer Number I V COOH Meg /kg Z-422-1P .47 8 Z-423-1P .64 11 Z-424-1P .54 12 Z-425-2P .74 16 T-443-1P .57 42 T-444-1P .63 12 T-451-1P .64 28 T-453-1P .56 30 T-461-1P .71 16 T-461-1P-D .61 70 T-461-2P .64 23 COOCH. Meg/kg* Mn ppm Sb ppm 1.00 0.80 3 2.06 3 0.94 3 1.63 3 1.63 3 1.75 3 1.75 3 1.96 3 1.96 3 0.93 3 274 263 250 296 307 278 258 278 254 254 263 121 red. The anomalous result with the deep dish procedure and observation that char initially forms on the surface in this procedure may indicate diffusion controlled processes which relate to the presence of surface char. However , such an interpretation is made less attractive by a subsequent series of experiments in which PET was degraded under TGA conditions at various heating rates. Heating at 6° and 25° per minute produced weight losses of 89-91% while 50° per minute produced a weight loss of 82-88%. These results are summarized in Figure 39. In a concurrent set of thermal experiments, the polyester samples were analyzed in both TGA and DTA modes on Mettler thermal analyzer in an attempt to determine transition points. In these experiments, however, the glass transition points were not obtainable because of poor instrument sensitivity. Melting points were found to vary from 250° to 270° and are tabulated in Table XX. More systematic investi- gation with more sensitive equipment would be required before meaning- ful interpretation of these parameters could be accomplished. As a measure of polymer flammabil ity, the series of polyester samples was subjected to oxygen index evaluation in powder form. This was accomplished in an inverted porcelain crucible. When the sample was heated for ignition with a propane flame, the sample fused into a flat lump in the center of the lid; the top of the lump then melted and subsequently discolored. At this point, ignition occurred with the production of a sooty, but steady flame. After increasing the nitrogen gas pressure, the flame fluttered and then extinguished. At this point, an oxygen index was determined. Every sample was repli- cated 3 times and an average 01 value calculated (Table XXI). Corre- lation of the oxygen index with {C00H} 9 DEG content, intrinsic vis- cosity, metal content, and weight loss was attempted. Only {C00H} was found to have any relation to 01. The log 01 was observed to be linearly proportional to the reciprocal of log {C00H} (Figure 40). A log-log plot was necessary to remove the considerable scatter in the data. Although these results are ambiguous, they have been in- terpreted by the PINY group as indicating that the flammabil ity of the polymer sample tends to increase with increasing concentration of 122 o o 10 o o - 8 ' -S£ p— C <0 uj e <: c n} u O i- •i- J- i. ■»-> 0) 0) ro c-i *f >* i i > to oo t- s_ 0) O) U i/> E E , > O (— Q a. 0) O <_) ■mo o (O O vO Q£ CSJ c5 S- '•ST •I— m c O * r= O I— UJ O o a. ro 3 *-» «+- «3 o L. a. < § CD h- O 0J o -C C\» 4-> o o O 4-> ?0 O CO 4-> U 0) LU O o o 00 o vO o o CSJ (%) ^46l9m lenpisay CO CD 123 TABLE XX THERMOGRAVIMETRIC WEIGHT LOSS AND MELTING POINT OF POLYESTERS Weight Loss Melting Point** Polymer Number (at 500°C) °C Z-422-1P 82* 270 Z-423-1P 80* 260 Z-424-1P 76* 270 Z-425-1P 93° 255 T-443-1P 90° 250-270 T-444-1P 89° — T-451-1P 90° 252 T-453-1P 91° 255 T-461-1P 89° 260 T-461-1P-D 88.5° 252 T-461-2P 90° 249 *Run on Mettler Analyzer in deep dish crucible °Run on DuPont 950MyjGA in boat shaDe crucible G) **DTA on Mettler^^Thermo Analyzer at heatinq rate of 15°C/min -o. 124 TABLE XXI OXYGEN INDEX OF POLYESTERS Oxygen Index % Polymer Number I II III Average Z-422-1P 18.1 17.9 18.4 18.1 Z-423-1P 18.4 20.23 20.23 19.6 Z-424-1P 22.42 21.31 21.43 21.7 Z-425-2P 19.8 18.1 18.1 18.6 Z-443-1P 16.6 17.2 16.6 16.8 Z-444-1P 16.6 16.6 17.2 16.8 T-451-1P 16.4 16.6 16.6 16.5 T-453-1P 16.6 16.6 17.27 16.8 T-461-1P 16.6 16.6 17.2 16.8 T-461-1P-D 16.4 16.25 16.25 16.3 T-461-2P 16.6 17.27 17.27 17.0 125 G log 01 L °9 C00H + 2 FIGURE 40. Log 01 as a function of Log —^rr + 2 for PET. 126 carboxy end groups. This would be consistent with carboxy catalyzed degradation reactions such as ester hydrolysis. It has also been noted by several workers that the concentration of carboxy groups tends to increase during the process of PET thermolysis (60). Thus, the effect of carboxy on flammability could become increasingly important as the degradation continues. The dependency upon the initial carboxy content should be small , as indicated, because of the effect, and the occurr- ence of the random cyclic ester pyrolysis mechanism, which may become dominant during degradation, and which does not appear to be affected by the presence of other compositional features. I.b. Effect of Bromine on Polyester Flammability: In addition to the studies of the pyrolysis of pure PET, the PINY project had the added objective of determining the effect of mode of incorporation of flame retardants on the properties of PET. Both bromine and phosphorus based systems were studied. A series of polyester copolymers and homopolymer mixture had been prepared previously by Yoon, Liepins and Pearce at RTI. These polymer systems contained brominated flame retardants at a level sufficient to give incorporation of 4-8% bromine. The copolymers were prepared from PET and the bishydroxyethyl ether of tetrabromobisphenol-A. Homopoly- mer mixtures contained polyester and the tetrabromobisphenol-A as an additive. When these samples were analyzed thermogravimetrically up to 500°C in an air atmosphere, using a 15° per minute heating rate, the weight loss of the copolymers was found to decrease as the bromine content increased. The results with the polymer mixtures were more erratic in relation to the bromine content. They are summarized in Table XXII. These samples were also evaluated in terms of oxygen index. The copolymers containing 4% bromine melted completely and later started burning on the top, whereas samples containing 6-8% bromine melted partially before igniting. The 01 increased with increasing bromine content. All of the homopolymer mixtures partially melted before 127 o TABLE XXII THERMOGRAVIMETRIC WEIGHT LOSS ANALYSIS AND OXYGEN INDEX OF COPOLYMERS AND HOMOPOLYMER MXITURES Sample Sample Bromine Weight Loss* 01 Number Type Content % % (500°C) % C-4 Copolymer 4.03 81.25 15.7 C-6 Copolymer 6.28 79.07 17.3 C-8 Copolymer 7.62 73.72 18.8 M-4 Mixture 4.00 79.07 17.2 M-6 Mixture 6.00 74.04 22.2 M-8 Mixture 8.00 77.16 23.9 PET None 90.00 15.6 Mixture of Bis(2,3-dibromo- propyl ether) of TBBPA 4.00 8.00 76.00 75.00 21.5 21.7 *15°C/min. Heating Rate. 128 burning and the 01 values increased with increasing bromine content. The value of the 01 was comparitively higher with the mixtures than in the case of the copolymers. These results are also summarized in Table XXII. Since these experiments indicated that, in the case of bis(hydr- oxyethyl )tetrabromobisphenol-A, the incorporation of the flame re- tardant into the polymer backbone reduced the efficiency of the re- tardant relative to that which it exhibited when present in mixtures with PET homopolymer. To explain this, it was proposed that the bis- (hydroxyethyl )ester was capable of undergoing a facile pyrolysis with the liberation of ethylene oxide and the free bromo phenol functions. The bromo phenol could then undergo its well known reaction with the elimination of hydrogen bromide. However, when the mixture was in- corporated into the polymer backbone, this mode of decomposition was inhibited resulting in less efficient liberation of the bromine re- tardant. In an effort to substantiate this hyposthesis, a mixture of PET with bis(2,3-dibromopropyl )tetrabromobisphenol-A was prepared and evaluated by oxygen index. As shown in Table XXII, the 01 values ob- tained on these samples were higher than those obtained from any of the previously studied copolymers. However, a comparison of these results with those obtained on the bis(hydroxyethyl )tetrabromobisphen- on-A was considerably more confusing. The bromopropyl ether appears to be more efficient at a bromine content of 4% and less efficient at a content of 8%. More surprisingly, the effect of the bromopropyl ether appears to be essentially independent of the bromine content within this limited range. The reasons for this behavior are not at all clear. In order to better understand the nature of the activity exhibited by bromine compounds on polyester and blends containing polyester, a study of model organobromine reagents was initiated at Clemson. These studies were frustrated by repeated failures to achieve reasonable add- on levels of the model compounds. Only bromo phenols were found to give sufficient add-on when padded from a 20% solution. An application of 2,4,6-tribromophenol at a level of approximately 12% was found to raise the oxygen index of 100% polyester to 27.5. Similar applications 129 to 50/50 blends yielded less efficient retardancy. The tribromophenol produced an oxygen index of 21.5 at an add-on of 16.4% on the blend. The reason for this difference in efficiencies is not clear since the bromo compound presumably acts entirely as a vapor phase retardant and thus should be insensitive to the chemical nature of the substrate. It was felt, however, that this might be related to the problem of dis- tribution of the flame retardant among the cotton and polyester por- tions of the sample. In order to study the effect of distribution in greater detail, samples of 65/35 and 50/50 blend fabrics were treated with tris(dibromopropyl )phosphate as a model system. Following appli- cation and drying, portions of the treated fabric were subjected to oxygen index evaluation. The remainder of the samples were heated at temperatures above 130 to effect migration of the phosphate into the polyester in a manner analagous to thermosol treatment. Portions of these samples were then evaluated by oxygen index also. The results of these tests also indicated no difference in flame retardancy as a function of location of the bromophosphate in the polyester. The re- maining portion of the samples were extracted with perchlorethylene and evaluated for flammability. This, of course, reduced the total level of phosphorus and bromine in the system as it extracted the majority of the material from cellulose. Oxygen index values, however, were found to be essentially the same as those obtained on the previous samples containing corresponding levels of retardant. These data tend to indicate that the effectiveness of a retardant such as tris(dibromo- propyl )phosphate, which is presumed to act primarily in the vapor phase, is a function only of the amount of retardant present and not significantly dependent upon the location of the retardant in the various fibers. There is, however, some evidence that other organobromine deriv- ativies may have a small, but measureable, effect in the condensed phase in both cellulosic and polyester fibers. Thermal analysis has been carried out on polyester films containing approximately 20 weight percent of a series of bromine containing flame retardants. These included polyvinyl bromide, octabromobi phenyl and decabromod i phenyl - 130 ® oxide. Samples of DuPont's type 900F Dacron^xonta ining tetrabromo- bisphenol-A were also analyzed. It appears that all of the bromo com- pounds exhibit a small catalytic effect on the polyester degradation. However, the amount of residue remaining at 500 does not appear to be significantly altered by the presence of the retardants. Pyrolysis-gas chromatography (GC) studies indicated a similar small interaction. Samples were prepared as films containing 20% of the flame retardant. In addition, films of pure polyester and of DacronV^900F were cast for comparison purposes. Small specimens were cut from each of the films and weighed carefully. They were then pyrolyzed at 800°C for 200 msec into a gas chromatograph with the column temperature programed from 70 C to 190 C at a rate of 5° per minute. The column was then held at 190°C for ten minutes to complete the run. The results of these studies are shown in Figures 41 and 42. The development of a new peak in the flame retardant films is clearly demonstrated. This peak was el u ted from the column at a temperature of 156 C. The fact that this new peak did not result from pyrolysis of the flame retardants in clearly shown in Figure 41 since the peak was not observable in the pyrolysis chromatograms obtained on the pure retardants. Polyvinyl bromide has two similar peaks, but they were el u ted from the column before the FR/PET compound. It is interesting to note that Dacron^-^OOF, which contains a brominated comonomer, also has this newly detected pyrolysis product. Although the compound exists in differing amounts, it is present in all of the flame retard- ant polyesters studied. Attempts were made to identify this new pyrolysis product and to determine its significance in the overall process of flammability and flame retardation, but no conclusive results could be obtained. The elution time of the material suggests a brominated ring structure. The formation of this material, however, may be quite incidental as the amount of low molecular weight products given off during the pyro- lysis of the flame retardant samples was not significantly affected by any of the retardants. Polyvinyl bromide actually seemed to increase the amount of low molecular weight gases released. 131 J Q X O o o co < x CO o CO > Q- O CD a: CD 132 o o cr> a. i s- > Q. s- cu s- «T3 O'J c 'e o t. jQ D"» E +■> o o to D_ CO O S: O CD CM CD 133 Since all these studies indicated that the predominate flame re- tardant effects of organobromine compounds could be described in terms of their vapor phase activity, it seemed that one of the controlling factors for FR efficiency should be their thermal stabilities. Be- cause aromatic bromine compounds, in general, degrade at a higher tem- perature than aliphatic compounds, they should be more effective as fire retardants for PET. Hence a study was initiated to measure the efficiency of a series of aromatic and aliphatic bromine compounds on polyester fabric in an attempt to relate their effectiveness as fire retardants to their chemical structure. Of the compounds studied, PVBr and poly(2,3-dibromopropyl acry- late){ P(DBPA)} contained bromine atoms attached to aliphatic structures, while octabromobiphenyl (DBBP), decabromodi phenyl ene oxide (DBDPO), and TBBPA contained bromine attached to aromatic nuclei. All fabric samples were treated from aqueous emulsions of flame retardants, with the exception of TBBPA, which was padded from acetone solution. The OBBP and DBDPO were applied with an acrylic latex, Dur-O-Cryl^XWC, as a binder. The TBBPA treated samples were topped with the same acrylic latex in an attempt to insure that the test re- sults were not influenced by differences in physical behavior of the treated samples. Fabric flammability was measured by oxygen index and isoperibol calorimetry. The 01 values were determined for treated PET samples containing increasing amounts of each of the fire retardants and are shown in Figure 43. Linear regression analysis of the data showed that the 01 results from all of the samples fit on one line with a confidence level of r = 0.97. Thus the 01 values appear to be de- pendent only on the amount of bromine in the fabric, regardless of which brominated compound was applied. Contrary to the 01 results, the isoperibol calorimeter measured significant differences in flame retardant efficiencies for the dif- ferent bromine-containing compounds. The heat released by the treated samples when they burned in the calorimeter (AH-,) was subtracted from the heat released by the untreated PET control, (AH-, ) . This 134 - C\J o O o O < < o Q Q O I— O. Q_ < o o« CO <£> o •r- S- JD ^3 l|- h- UJ u Q. T m i- C -O i|M IQ U. S- c CO o 5^ t- GQ +J &« jc: CT> ■M >r « -C CD o 5 °r- CD 3 13 C 03 l/> CD 3 ^~ (—1 o C CD CD £ 4-> CD J3 Q. • •i- (/> .c +J in c C fO O XJ •r- s- +■> «j ro +-> i— cu cd s- s- t O s- a. O Q o (— o- a. +-> U ■o a> s- QJ n: o o o CVJ .o o o o o co o o <■£> O o o o CM «3- uiej6/saijoieo ( ^HV) - S ( W) uoi^onpey }e9H (X. 137 o T3 «— •,- (V 4- > C CD O —1 O X X CQ oo ►—1 oo OO >- I— —i _l - • — 1 Cxi 00 1— 00 LU LU 2: ce: *— 1 cc: LU Cxi _i - 00 C£ ►— « U i_ 3 cd -a Q. cu q: S- en 4-> *>v» 03 1 — CD ro 31 (J 4-> Q. O) O C c X3 CU S~ E ro 03 +-> r— a; li_ cc: CO v£> en CD en cr» cr> en O Ch o o .— ID en o CXI C\J en C\J 00 in o CNJ *3- CO o Q_ v ^ CO CO CO Q_ Q_ h- O O 138 very effective flame inhibitor, HBr, is known to be released when PVBr decomposes. However, this evolution occurs at 175 C, which should make the effectiveness of PVBr sensitive to burning configuration. In the 01, PVBr was as effective as any other bromine compound treated, but its relative efficiency was greatly reduced when burned in the isoper- ibol calorimeter. The low degradation temperature of PVBr is typical of many alipha- tic bromine compounds. However, the calorimetric results showed that PDBPA exhibited fire retardant efficiency similar to TBBPA. This in- dicated that in PET the thermal stability of aliphatic PDBPA was also similar to that of aromatic TBBPA. These results preclude any specific statements as to the effectiveness of aliphatic bromine compounds versus aromatic compounds. However, the data suggest a general trend, supporting the assumption that aromatic compounds are more effective, possibly because they release their inhibiting species at higher temp- eratures, closer to the advancing flame front. In an attempt to verify this interpretation an experimental pro- cedure was designed to measure HBr evolution from fire retardants during pyrolysis. TGA in conjunction with a Br" specific ion electrode was used to monitor HBr evolution. Dacron^900F and fabric samples treated with the organobromine compounds were heated at 10°C/min under nitrogen and air atmospheres to a maximum temperature of 500°C. The compounds flame retardant additives were pyrolyzed alone under the same condi- tions for comparative purposes. Pyrolysis gases from the TGA were passed through a beaker containing 50 ml of distilled water and the bromide ion concentration monitored. It was assumed that all Br" absorbed came from HBr evolved from the pyrolyzing samples. Table XXIV gives a summary of the total amount of bromide ion absorbed in the distilled water when the flame retardants and treated samples were pyrolyzed. Rr R is the ratio of bromine absorbed to bro- mine originally present for the pure fire retardant, and R FAB the same ratio for treated fabric samples. The values reported are an average of at least three TGA experiments. There was no apparent difference 139 TABLE XXIV HYDROGEN BROMIDE RELEASE OF BROMINE FIRE RETARDANTS Flame R R Retardant FR FAB PVBr 0.33 PDBPA 0.12 900F ---- TBBPA 0.033 DBDPO 0.002 OBBP 0.001 0.26 0.13 0.12 0.092 0.027 0.006 140 in bromide ion absorption or TGA curves when the pyrolysis atmosphere was changed from N« to air, indicating that oxygen was not involved in HBr evolution. These results show that PVBr released the most HBr of the com- pounds studied with approximately 30% of the bromine unitially present in the sample being absorbed by the distilled water. HBr evolution occurred over a temperature range of 175°-250 C. Since PVBr is known to release practically all of its bromine as HBr, it was taken as a reference to which the other compounds were compared. P(DBPA) pyrolysis resulted in 12% of its bromine being absorbed as HBr, with absoption occurring in the temperature range of 250-31 5°C. This represented only 30-40% as much HBr evolution as PVBr, indicating that significant amounts of bromine were released in other forms. The TGA of P(DBPA) suggest that HBr evolution is occurring simultaneously with polymer molecular fragmentation. This could result in the vol- atilization of bromine containing polymer fragments, resulting in a decrease in the relative amount of HBr produced. The three aromatic flame retardants behaved differently than PVBr and P(DBPA). Only TBBPA released any significant HBr, and it was only 10% of the amount released by PVBr. During the thermal analysis of the aromatic compounds, white powders were deposited on the glass TGA cover and the inside of the glass bubbler. The bubbler clogged com- pletely during two of the TBBPA tests. Small amounts of these powders were collected and infrared analysis showed the deposits to be un- decomposed flame retardants in all three cases. Apparently the aro- matic bromine fire retardants tested are entering the vapor, phase as intact, undecomposed molecules. TBBPA volatilized at 200-275°C, DBBP at 250-340°C, and DBDP0 at 275-370°C. The evolution of HBr resulting from the pyrolysis of treated fabric samples and Dacronv^QOF was also studied. Comparison of these thermograms shows good agreement between the amount of condensed phase reactivity detected in the pyrolysis-G.C. experiments and the amount of HBr released during the time of polymer thermal degradation. This supports the conjecture of reactions occurring between vinyl groups 141 present during PET degradation and escaping HBr. The treated fabric thermograms also showed that PVBr again pro- duced more HBr than any of the other fire retardants. Nearly all of the HBr produced by PVBr was released before PET degradation began. This would be consistent with the interpretation that in the upward burning configurations the majority of the HBr is released before the flame front arrives, greatly reducing the effectivessness of PVBr as a flame retardant. P(DBPA) in the presence of PET released the same relative amount of HBr as P(DBPA) pyrolyzed alone, indicating no significant inter- action between the decomposing polymers. In contrast, TBBPA treated polyester fabric produced relatively larger quantities of HBr than TBBPA alone. Dacron^'QOOF released a similar fraction of its bro- mine as HBr compared to the topically treated fabric samples. This indicated that the increase in HBr production might have been caused by TBBPA being physically held in the polymer system, allowing it to reach higher temperatures where decomposition could occur. Evidently a volatilization process is competing with the thermal decomposition processes. The polymer melt could have decreased the rate of TBBPA escape and allowed more of the compound to decompose and release HBr. At the elevated temperatures of the polymer melt, it is conceivable that TBBPA is reacting with PET through transesterif ication, which would result in TBBPA being held in the condensed phase. This would allow the topically treated samples to exhibit similar behavior to the Dacron^^OOF, where the TBBPA is part of the polymer backbone. These data indicate that bromine flame retardant efficiency cannot be explained simply in terms of HBr release. This conclusion was further demonstrated by the OBBP and DBDPO fabric samples. These samples were the most efficiently retarded, but they released very small amounts of HBr. OBBP produced practically no HBr, while DBDPO produced only a small amount. The TGA's suggest that OBBP and DBDPO were probably released into the flame as intact molecules by volatilizing out of the fabric. Compound decomposition must occur in the flame itself, releasing the bromine in the location 142 where it can have the greatest inhibitory effect. It is interesting to note that a proposed mechanism for bromine- antimony synergism calls for the reaction of HBr with antimony oxide in the condensed phase of the polymer system (60). Other work has shown this synergistic effect to be a real one by examining the White Chem- ical FR P-44^y (antimony oxide-DBDPO). However, the TGA studies have shown that DBDPO produces negligible HBr when heated to 500°C by itself, and only a small amount of HBr when heated to 500°C with polyester. The large increase in flame ret.ardant efficiency by the addition of antimony oxide thus cannot be attributed to its reacting with such a small amount of HBr. This suggest that reactions between DBDPO and antimony oxide are occurring by a yet unknown mechanism. The comparative thermal stabilities of the studied bromine fire retardants is given in Table XXV. As mentioned it was noted that TBBPA volatilizes at higher temperatures in the presence of PET. With this in mind, a good correlation can be drawn between the temperature of the maximum rate of weight loss of the fire retardants and their relative efficiency in PET. This indicates that, for the compounds in this study, the temperature at which the bromine-containing species leaves the substrate is more important than the chemical nature of either the volatilizing species or the compound present in the substrate. These results imply that in PET systems containing only bromine as the flame retardant, the most efficient system would be one that released its bromine-containing species at the same time that the poly- mer released its flammable gases. This should plctce the bromine di- rectly into the flame reaction zone where its resulting inhibition would be most efficient. I.e. Effect of Phosphorus on Polyester Flammability: In addition to the studies centering on bromine compounds, efforts were expended in the area of phosphorus chemistry. The team at PINY has synthesized a series of phosphorus containing materials to be used in a similar series of investigations. All of these are based on 143 <+- o QJ re S- Q£ +-> 3 .c o 4-> E oo 03 13 •r- S- E CD •> CD •i— 3 CO CL X CO E - _l <: _i <: en QJ C_) OO c ro ^ en CO CO QJ o S_ _J 3 4-> +-> ra -C i- a> •r- CD Li_ en LO o LO LO O o r^v CO OJ U OO CNJ 1 CO 1 o LO O LO o o CO o (T> r^. co » r— CvJ CNJ CM CNJ <: Q OO OO OO O- D_ \— o Q 144 phenyl phosphonyl dichloride, as commercially obtained from Stauffer Chemical Company and included phenyl phosphonyl di (methyl benzoate) and phenyl phosphonyl di (p-methyl oxybenzoate). The 1:1 molar copolymer was prepared by the reaction of bis(2- hydroxyethyl )tetephthalate and phenyl phosphonyl di (p-methyl oxybenzoate) by heating at 160° for 8 hours at 0.1 mm pressure. It contained 4.5 mole% phosphorus (theory for 1:1 copolymer - 4.8%). The copolymer con- tained 4.5 mole% phosphorus (calc. 4.8%), was pale brown and has a 0.78 inherent viscosity at 30°C. This sample could be spun manually and had a T of 42°C and a melting point of 191°C. The sample was evaluated by 01 and weight loss was evaluated by TGA. The 01 values was about 30.8 and the weight loss was around 84% at 500 C. The 1:1 molar copolymer obtained by the reaction of bis-(2-hydro- xyethyl )terephthalate and phenyl phosphonyl di(p-methyl benzoate) was prepared in a similar fashion heating for 8 hours at 200°C under 0.1 mm pressure. The sample contained about 5 mole% of phosphours, was dark brown and had a (n),- nh of 1.3. It could be spun manually. The polymer had a T of 42°C and a melting point of 172°C. The 01 value was 29.0 and the weight loss by TGA was 72%. These copolymer samples were compared to two additive systems. The phenyl phosphoyl di (p-methyl benzoate) monomer and tri phenyl phos- phine oxide, respectively, were added to PET as additives. The con- centration of phosphorus was 5 mole% in order to compare the flamm- ability effects directly with the phosphorus containing copolymers. Tri phenyl phosphine oxide had previously been shown to act in the vapor phase while reducing the flammability of PET (12). The monomer and triphenyl phosphine oxide (TPP0) samples were first studied alone by TGA to understand their potential behavior as addi- tives. The data indicate that tri phenyl phosptnne oxide essentially is degraded and volatilized 100% at 500°C. Some oxidation of TPP0 occurs and is indicated by an initial early small gain in weight, whereas phenyl phosphonyl di (p-methyl benzoate) degraded from the be- ginning and showed a weight loss of about 90% at 500°C. PET samples were melt blended with phenyl phosphonyl di (p-methyl 145 benzoate) monomer and TPPO, respectively, and then were studied by TGA and the oxygen index. PET melt blended with phenyl phosphonyl di(p- methyl benzoate) degraded after 220°C and decomposed after 290°C. The weight loss was 27% at 500°C and gave an 01 of 29. Ester interchange reactions are rapid and thus could account for the very similar results with the copolymer. PET melt blended with TPPO showed an early weight loss, with a final weight of 96% at 500 C as compared to PET itself which has a weight loss of 90% at 500°C. The 01 was 24.5, which is higher than the 17.0 value for unmodified polyester sample. These results suggest that TPPO apparently acts in vapor phase (12) while phenyl phosphonyl di(p-methyl benzoate) appears to operate primarily in the condensed phase. A simple powder blend of PET and the phenyl - phosphonyl di(p-methyl benzoate) also had an 01 of 26.5. A summary of the weight loss and oxygen index data is shown in Table XXVI. The contrast between the vapor phase action of tri phenyl phosphine oxide and the condensed phase action of phenyl phosphonyl di(p-methyl benzoate) was notable. It would also appear that a non- reactive phosphorus additive (TPPO) which operates in the vapor phase was a less efficient flame retardant for PET than those which are reactive or can be copolymerized and also may act in the condensed phase. A further indication that phosphorus may be capable of some con- densed phase activity on PET can be found in the results of a study at Clemson. As shown in Table XXVII, application of diammonium phos- phate to 100% PET fabric imparted significant flame resistance. Since DAP is essentially nonvolatile under the FF 5-74 test conditions, its action is presumed to be restricted to the condensed phase. However, the interpretation of these results is complicated by the effects of the increased levels of moisture expected to be present in the DAP treated fabrics. This is demonstrated by the observation that samples of the fabric containing 3.5% P burned readily when stitched with 3 fiberglass threads and oven dried before testing. A similar series of glass supported samples conditioned in the normal laboratory atmo- sphere exhibited borderline flame resistance with a failure rate of 146 TABLE XXVI P CONTAINING ADDITIVES AND/OR COMONOMERS IN PET (~5 mole% P) Structure PET PET + Type % Weight loss, (500°C) lomo polymer 90 copolymer 84 01 17.0 30.8 H Q C00C-(O\-0-P-0-<( ^-COOCH. 3 x^y I \^/ PET + H 3 C00C- 3 P=0 additive 96 24.5 147 TABLE XXVII Flammability of Di ammonium Phosphate Treated Polyester % Add-on % P (theoretical) FF 5-74 (Avg. Char Length) 26.3 4.78 4/4 (1.1 in.) 17.8 3.47 4/4 (1.6 in.)* * 4.7 seconds afterflame 148 50% in FF 5-74. l.d. Role of Condensed Phase Oxidation in PET Flammabil ity: In most of the previous studies it has been assumed that the effects of condensed phase oxidation were insignificant in the poly- ester combustion process. However, in view of recent work by Stuetz and others (61), it was felt that an attempt should be made to either confirm or disprove this experimentally. Samples of PET were therefore examined by TGA in both the pre- sence and absence of oxygen. The differential curves showed little reproducibility but in any event the curves found in inert environ- ments were different than those found in 20% oxygen. Since at higher heating rates the decomposition processes merge to form a single peak, it is not possible to determine the importance of heating rate on the processes taking place. On the basis of these results the temperature range chosen for the isothermal studies was 350°C to 420°C. The lower temperature was the point at which loss begins to take place. The upper limit was the approximate temperature at which the rate of weight loss was observed to be a maximum. To test the effect of sample surface area on the rate, the follow- ing three configurations of PET were examined: chip, fiber, and fiber Wiley milled to a 40 mesh. If the oxidation processes were strongly dependent on the surface area one would expect the rates to increase in order of powder > fiber > chip. The decompositions were carried out at 356°C in atmospheres of 0, 5, 12, and 20% oxygen, and the rate constants determined assuming a first-order rate dependence. The values as shown in Table XXVIII did not indicate any effect of the various configurations on this order or magnitude. For the purpose of homogeneity of samples the Wiley milled fiber was employed in the remainder of this work. For the remainder of the PET studies five temperatures were used: 334, 356, 369, 387, and 407°C. Table XXVIII contains the rate constants 149 QJ I O QC o X #* X OO h- LU 2E _J <: QQ i— «=c to 1— 7Z. O O CD >- X o o on <: O i—i o o o cc => o C_> o LT) 00 < CO o -Q c •I — E Q- t— 1 CVJ o CO 1 CO 1 CO CVJ I 1 X X X X *>J" r^ o 00 Lf> C\J • • . r-^ o CvJ 150 found at these temperatures with varying oxygen concentrations in the environment. The order was found to be pseudo first-orider, although the weight loss at the lowest temperature appeared to be zero order. At higher temperatures, however, the kinetics seemed to be approaching second order. A possible explanation for this may be that the reaction appeared to be zero order at the lower temperatures because the rate of weight loss was slow. Thus over the time span used, it could ap- pear linear. The onest of charing at higher temperatures could also cause an apparent deviation from first-order kinetics. Ideally, isothermal TGA can be used to discriminate between sev- eral processes that may result in weight loss. For PET it is difficult to completely differentiate between pyrolysis and oxidation, since the weight loss occurs over a relatively narrow temperature range and both processes appear to occur simultaneously. As the data in Table XXIX show, the value for the PET decomposition rate constant in the presence of a 20% oxygen environment is twice that in 100% nitrogen at 334 C. At higher temperatures less of an increase of rate in the presence of oxygen was observed. This possibly indicates that pyrolysis takes place faster than oxidation at higher temperatures where more energy is available. The dependence of activation energies on the amount of oxygen pre- sent during the degradation clearly shows that an increase in oxygen content catalyzes the weight loss. From the observed doubling of the rate constant it can be shown that the oxidative processes are actually occurring faster than the nonoxidative process. Theoretically, if the rate of decomposition is pseudo first-order the rate is given by the equation: Rate = k^PET) where k-. = rate constant for decomposition (PET) = weight of PET in the TGA experiment Experimentally in a nitrogen environment the rate is found to be ex- pressed given by: Rate N 2 = k py (PET) 151 o o cx> WD oo > Q CO to © o cr wj Q C > CD Q -t-> X X CO < DC to O O _J «t i— :n ex. LU o o WO lo oo >- x O LU LU o □ ** ro oo > o o O Q X X X -o oo oo o ** o oo ro CM o s« CM i- W (\J I I I I o o o o I— CM o o LO o WO r— C\J LO CO CO I I o o X X ^- I— o o "3- "3- en lo oo cr> o o C\J cm oo oo O CM CM CM CM I I I 1 o o o o LO r- CM CM OO I I O O i— X X WO CM i— OO OO CO CM I I I I O O O O WO CM «d- oo oo oo I I I I o o o o ex. .— r— I— CM O r— CM o o CM «— <— •— > O O 1 o o CD ^— f— ^™ r— Q X X X X • • CO "O WO o CM <■* CD 4-> o CM O o f — CO • • • • Q. o o o o E CO CM 1 7 1 1 CD CD ' — ■> o o o o S- ■"" ■ 1 1 1 -t-J c X X X X •r- +-> E CM LO WO oo CO .* o CM LO CD o r^ oo oo > CD Q ■o +-" OO CM O CT> c— .— i— M CM CM CM i I I I O O O O LO CM OO .— r— OO LO CM CM CM CM I I I I o o o o L»- o > WO WO oo WO CD CO 1^ co CM -o CM OO ** LO -o i- c CM O +-> O LO * — CM CO 152 where k = rate constant for pyrolysis. py Therefore the theoretical rate constant, k-, , is equal to the experi- mental rate constant k . When oxygen is introduced into the environ- ment, oxidation occurs primarily at the surface so the theoretical rate of decomposition will consist of contributions from both decom- position routes. It would then be described by the equation: Rate Mix " k lMix (PET > " V PET > + V PET >surface where k n = rate constant for decomposition by oxidation. u 2 If the rate constant doubles with the introduction of a 20% oxygen environment, k-iM.. = 2k , then the relation of k to k Q is given by: k py (PET) - yPET)^^ which can be rearranged to: < ^surface ° 2 Since the fraction (PET)/(PET) r * ace is clearly much larger than unity k Q is much larger than k . The results at 387°C for PET in the form of Dacron^>'54 were com- pared under the same environments to those from a deep dye PET from American Enka made by the introduction of an aliphatic diester comono- mer. Since the deep dye polyester has a more open structure it would seem reasonable to assume that it would Jiave the capacity to absorb more oxygen into its matrix than Dacron^54. If it does absorb more oxygen one would expect an increase in the rate constants that were found for the PET under comparable conditions. The results in Table XXX show that the rate constants are higher in the case of the deep dye polyester than with normal PET in all the situations studied. How- ever, the decomposition rate constants increased systematically by a factor of approximately 1.2 when the standard deviation from the mean was considered. This suggested that the increase in the rate constants was due to a feature present at the beginning of pyrolysis rather than 153 TABLE XXX Comparison of the decomposition rate constants, k, for m Dacron v o / 54 and deep-dyeing PET Dacron^y54 Deep-Dyeing PET % o 2 387° C 387° C Std. Dev. a k min" k min" 2.86 x 10" 2 3.53 x 10" 2 0.13 x 10' 2 5 3.76 x 10" 2 5.25 x 10" 2 0.43 x 10" 2 12 4.83 x 10" 2 5.51 x 10" 2 0.39 x 10" 2 20 5.25 x 10~ 2 5.60 x 10' 2 0.22 x 10" 2 a Standard deviation of the mean k of at least three samples. 154 an increase in adsorption of oxygen during pyrolysis. This phenomenon could be the result of an increase in adsorption of oxygen during pro- cessing or the result of uncharacterized surface effects. The effect of bromine on the condensed phase was examined by TGA using DuPont's Dacron^900F containing tetrabromobisphenol-A. 'The thermograms obtained at a heating rate of 20°C min" in both nitrogen and 20% oxygen in nitrogen show what appears to be two processes taking place. Oxygen appeared to have its most pronounced effect at lower temperatures. The correlation analysis of both first and second-order decompos- ition showed that an overall second-order process best approximated the experimental results and therefore the reaction rates were cal- culated assuming an order of two. The temperatures used were 356, 369, 387°C and these results are tabulated in Table XXXI. It is obvious from these data that bromine is affecting the condensed phase decompo- istion of PET. The effect, however, is to increase the rate at 356°C by 50% and to increase the rate at 387°C by about 250%. Although there seems to be condensed phase retardation of the oxidative processes causing weight loss, the overall decomposition rate is higher than that of the untreated PET. It must be emphasized, however, that it is the rate constants that are being compared and not the rates of the de- composition reactions. Since these rate constants were observed to reflect the overall thermal stability of the samples their comparison is well justified. A possible mechanism for the catalytic effect of bromine on PET is shown in equation { 5} 8 8 ^COCHpChLOC^ II ^COCHCH^Ccj^ {5} a^ >m>- + co 2 155 o o o> a. ir> © o 00 CO o C_> < O I/) •— i > Q ■o a> o o cr> .x o >— o X CTi X x © O _) O CQ CC I— <£ a o o oo o a. 2£ o > Q 00 E O h- O UJ -X Lx > an O i/o < »— oo o o o ID co 2 go en CM O 5^ oo co oo co C\J C\J CM C\J i i i i o o o o Cft Cft Ul Cfi O f— r- i— O O O O CM CM CM CM I I I I o o o o CO CO CM O r- cr» i— >jO co >x> r— CM I s -* CM o a-> o LTl ^r O C\J CM CM CM CM CM CM CM CM O O O O X X X X *!■ r^ UD en CM CM CM CM O o 1 o O X X X X CO f— CO en CO CTi u~> r- oo CM co (XI CM Csl CM CM O O O o X X X X O CM O O CM O CM CM CM CM I I I I O O O O CO CD a. 0) CL» V,. ai 4~ O a? ai CM O r— CM -M «3 •r" > CU -o T3 S» 00 15 6 The raljos of the rate constants for PET decomposition to those for Dacron^900F show that the effect of bromine in the condensed phase is limited mostly to the low temperatures in the presence of oxygen. This effect appears to diminish as the temperature increases, again suggesting- that at higher temperatures pyrolysis begins to be- come more significant. An alternate possibility is that the bromine in the Dacron^o^OOF was volatilized early in the decomposition at the higher temperatures so that the observed decomposition was essent- ially that of the PET alone. One phosphorus-containing flame retardant was also studied. Hexa- phenoxycyclotriphosphazene was added at a level of 10% of the melt be- fore spinning. TGA analysis showed that the presence of oxygen lowers the decomposition temperature of this fiber, but not as significantly as in the case of Dacron^-^900F samples. The differential TGA in- dicated that two processes are responsible for the weight loss in both the presence and absence of oxygen. The effect of oxygen, as illu- strated by the differential curves, is primarily on the second process which occurs at the higher temperature. This is in contrast to the Dacron^900F where the effect of oxygen was found to be on the lower temperature processes. A study of the hexaphenoxycyclotriphosphazene itself by TGA showed very little effect from the presence of oxygen; but when the differential curves were examined it became clear that the presence of oxygen either caused an additional process to occur or the rate of certain pyrolytic reactions to diminish. This suggests that a chemical reaction may take place between the hexaphenoxycyclo- triphosphazene and PET in the condensed phase. Statistical analysis of the data indicated that a second order expression gave the best description of the rate of the decomposition. Thus the rate constants given in Ta^le XXXII, were calculated on this basis. In contrast to the Dacron^o^OOF system, the phosphorus-contain- ing samples showed a detectable decrease in the rate at 356°C as the oxygen to nitrogen ratios increased. At 369°C the rate constants for all the oxygen to nitrogen ratios were essentially the same. As the temperature of decomposition increased to 387 C the rate in oxygen 157 0! CO ° o < o CO •—i a; Q -o +J CO C7> Q_ -XL > CD Q ■M E i °i — E c; -XL Q- > CD Q OO CT> <\J o 5^ CM «3- co o x cnj CO LO o LO CO ro CNJ IT) .— O r- CNj CO o O ro LO o ro o 00 CNJ i— r- CNJ CNJ I I I I o o o o LT) o en oo *3- o o O o o X X X X ro O o en LO o r- r— O r- LO cr. i-^ 00 LO LO CNJ CVJ CNJ OvJ CNJ 1 o o- o o X X X X CO o ro O ro O LO o CNJ CNJ CNJ CNJ 1 o 1 — o 1 o o X X X X CNJ CT> LO LO LO ro LO ro CNJ o l— CNJ CD CD CD CO CD O J*. CD ■o -o s- (O T3 c «o +J oo to 158 began to increase relative to the rate constants in 100% nitrogen. This can be misleading, however, because inspection of the ratios of the rate constants for the phosphorus-containing fiber to that of the Dacron^^54 indicates that the overall rate is increased on incorp- oration of the hexaphenoxycyclotriphosphazene. This means that al- though the flame retardant is affecting the decomposition processes in the condensed phase, it is not efficient enough to be effective in slowing the decomposition rate to a level below that of untreated PET. The decrease in the rate as 0~ content increases at lower temperatures may result from radical trapping by the hexaphenoxycyclotriphosphazene which becomes less significant at higher temperatures due to the dominance of pyrolysis reactions. The possibility also exists, how- ever, that the flame retardant will volatilize at a faster rate at higher temperatures causing a decrease in the condensed phase effici- ency. Since oxidation involves a free radical mechanism, an increase in the rate of decomposition due to oxidation should be accompanied by an increase in the radical concentrations. To confirm this, electron spin resonance (ESR) was used to monitor the relative concentration of radicals in the residues. The temperatures chosen for isothermal decomposition were 356°C and 387 C with oxygen/nitrogen ratios of zero to twenty percent. The radical concentrations of all the residues were measured at room temperature with the same modulation, frequency, and microwave power. Sample weight was kept close to 10 mg to mini- mize any dependence of the ESR signal on sample size. The observed signal consisted of a broad singlet; thus to establish a meaningful measure of the relative number of radicals the peak to peak height was divided by the residue weight. Figure 45 shows a plot of percent weight loss versus the normal - ized signal at 356 C for Dacron^-^54 . The concentration of radicals when the sample was degraded in 20% oxygen increased rapidly at a fifty percent weight loss; whereas decomposition in 100% nitrogen re- sulted in a significantly slower growth in radical concentration. At 387°C, as shown in Figure 46 an increase in the concentration of 159 240 220 200 180 C\J o 160 r ~ X c CD 140 ISI 120 (0 E s- o 100 80 60 40 20 A 20% 2 environment ■ 100% N« environment 1 30 40 50 60 Percent Weight Loss 70 80 FIGURE 45. Normal fted ESR signal versus percent weight loss for DacronVli/54 at a decomposition temperature of 356 . 160 240 220 - 200 180 CVJ 2 160 X J> 140 "8 N 120 o 100 — 80 60 — 40 20 A 20% 2 environment B 100% fi, environment 30 40 50 60 Percent Weight Loss 70 80 FIGURE 46. Normalized ESR signal versus percent weight loss for Dacron^54 decomposed at 387 C. 161 radicals appeared at a 50% weight loss in both atmospheres with the rate of increase being much slower in 100% nitrogen. In comparison, the oneset of radical formation in oxygen occurred eariler in the lower temperature experiments but there was little change in the curves for decomposition in nitrogen. This suggests that oxidation is more significant at the lower temperatures. This latter observation is in agreement with the earlier kinetic results obtained by thermogravi- metric analysis. The same experimental results were used to obtain an approximation of the rate of formation of radical species assuming a pseudo first- order decomposition. At 356°C the rate constants in 100% nitrogen and 2-1 2-1 20% oxygen were found to be 1.28 x 10 min and 7.21 x 10 min , respectively. The production of radicals in 20% oxygen was over five times that found in 100% nitrogen atmospheres suggesting that oxidative degradation is significant. At 387°C the rate constants were 1.03 x 10 3 min' 1 in 100% nitrogen and 4.56 x 10 3 min" 1 in. 20% oxygen. This reresents a four-fold increase in^rate. ESR measurements for Dacron^-^OOF were made in the same manner as for the Dacron^-^54 samples. Figure 47 shows that at 356°C radical formation commenced near a weight loss of thirty percent in both 100% nitrogen and 20% oxygen environments. The rate of radical formation, 3 assuming a pseudo first-order reaction, was found to be 3.07 x 10 -1 3 1 min for 100% nitrogen and 5.73 x 10 min for 20% oxygen. Since not all mechanisms will lead to radical species, it should not be considered inconsistent that the order established by ESR is first order while TGA studies showed the decomposition to be approx- imated by second order kinetics. Based on these rate constants the formation of radicals in 20% oxygen occurred 1.9 times faster than in 100% nitrogen. The total number of radicals increased significantly when the decomposition temperature was raised to 387°C. Figure 48 shows that the increase in radical species occurred near 20% weight loss in 20% oxygen and near 45% weight loss in 100% nitrogen. When 60% of the weight was lost radical formation increased significantly for both cases. The rate constants at this temperature were found 162 o c CD 240 - A 20% °? environment Ai 220 200 180 160 140 £ 120 o 100 80 60 40 20 100% N 2 environment 10 20 30 X _L -L 40 50 60 Percent Weight Loss 70 80 FIGURE 47. Normalised ESR signal versus percent weight loss for Dacron^900F decomposed at 356°C. 163 270 220 200 130 o 160 I— § 140 1/1 T3 £ 120 100 80 60 40 — 20 A 20% 0„ environment 100% N ? environment /"o/ 10 30 40 50 60 Percent Weight Loss 70 80 FIGURE 48. Normal y*«d ESR signal versus percent weight loss for Dacron ^900F decomposed at 387°. 164 to be 2.33 x 10 4 min' 1 in 20% oxygen and 7.4 x 10 3 min' 1 in 100% nit- rogen, representing a rate in oxygen which is three times faster than that in nitrogen. The ESR measurements of the phosphazene-containing fiber residues at 356°C are illustrated in Figure 49. The rise in the number of rad- ical species occurred at 25% weight loss in nitrogen and 35% weight loss in 20% oxygen indicating that the addition of hexaphenoxycyclo- triphosphazene inhibited the production of radicals in the presence of oxygen. The rate constants, assuming pseudo first-order kinetics, 2-1 2-1 were found to be 5.64 x 10 min in nitrogen and 2.54 x 10 min in 20% oxygen. At a temperature of 387°C the concentration of radicals began to increase rapidly at 55% weight loss in nitrogen and 65% weight loss in a 20% oxygen atmosphere (Figure 50). This is consistent with the results at the lower temperature in that the hexaphenoxycyclo- triphosphazene retarded the onset of radical formation. The rate con- 5-1 5-1 stants were 1.03 x 10 min and 1.33 x 10 min for nitrogen and 20% oxygen environments, respectively. Although there is a high degree of scatter, the ESR data are con- sistent and offer supporting evidence of the results found in the TGA studies. The rate constants established by ESR and by thermogravi- metric analysis are compared in Table XXXIII. Although these data point to the existance of condensed phase in- teractions between PET and the phosphorus and bromine compounds, it is difficult to elaborate on the actual mechanisms of the reactions. A chain transfer mechanisms would involve the donation of a hydrogen atom to terminate a chain branching reaction and form a stabilized free radical. This would result in a constant or decreasing concen- tration of radical species depending on the rate of recombination re- actions; or, if the free radical produced were extremely stable, it is possible that there could be an increase in radicals as oxidation and termination continue in the form of the production of stabilized rad- icals. At both 356°C and 387°C the concentration of radical species formed in Dacron^900F were higher than in Dacron^54 in both 165 CO o C en -o Nl <5 o 120 100 80 60 40 20 A = 20% 2 environment D = 100% N 2 environment 10 20 30 40 50 Percent weight loss 60 FIGURE 49. Normalized ESR signal versus percent weight loss for phosphazene containing PET (PFR) decomposed at 356 C. 166 185 100 80 o to en A 20% 0~ environment ■ 100% N ? environment M O 60 — 40 20 10 20 30 40 50 60 Percent Weight Loss 80 FIGURE 50. Normalized ESR signal versus percent weight loss for phosphazene containing PET (PFR)' decomposed at 387 C. 167 Q£ 1° 00 -|r- LU jh? CC U_ o_ fO ,_ J* 1 c c UJ °F= E <£ O o oo O O cc o i-i i— «o CM o CM ID in o o ;r. CM CM O CO I o ^1- CM o to to C%! cm oo oo oo o CO lA'-> CM CM o O 1 — ■ — X X OD CM CM O 00 <— CM CM r— r^~ CM I O CM IT) r— CO o o CM O «3- OO 00 CM 00 oo 00 o X oo o o CX> Cy> f— *f oo oo ■■ Ml CM CM 1 O O X X CO 00 o CT> oo o oo o •— l-~ X X oo o •X) r- <** CM CM I I o o CM IT) O CM O CM 168 nitrogen and 20% oxygen. The formation of radicals from the bromine- containing fiber in nitrogen was closer in Number and rate to ±he form- ation of radicals in 20% oxygen than observed with the Dacron^-^54. This suggested that the reaction of PET with bromine represented a lower energy pathway for decomposition than the reaction with oxygen. At lower temperatures there would be a competition between oxygen and bromine^ resulting in the smaller difference in the ESR curves from Dacron^900F in nitrogen and oxygen. At higher temperatures, where more energy is available, and with the additional oxidative pathway, there would be a net increase in the number of radical species in the presence of oxygen. The degree of formation of radicals in PET at 356°C and 20% oxygen is similar to the phosphorus-containing samples at the same temperature but in 100% nitrogen. This suggests that the hexaphenoxycyclotriphos- phazene is serving as an initiator in the absence of oxygen to decom- pose the PET in the same way as the bromine appears to do in the Dacron^^OOF. In the presence of oxygen, however, the additive serves as an antioxidant since the rate and concentration of radicals de- crease, indicating an interaction between oxygen and the hexaphenoxy- cyclotriphosphazene. Further decomposition of the flame retardant would then enable it to act in a manner similar to that proposed by Hastie (41). At 387 C a similar effect was observed and the phosphorus flame retardant delayed the oxidative production of radicals until 60% of the initial weight was lost, after which the concentration of radical species increased rapidly due to the volatilization of the retardant. Using a simple heat balance scheme, and assuming a heat of com- bustion for PET of 6,000 cal/gram, it can be shown that condensed phase oxidation represents a significant part of the heat experienced by the condensed phase. If 80% of a one gram sample undergo pyrolytic degradation in the condensed phase, the volatile products which under- go complete combustion to carbon dioxide and water will give off 4,800 cal/gram in the vapor phase. If the remaining 20% is half oxidized in the condensed phase and then completely oxidized in the 169 vapor phase, 600 calories will be produced in the condensed phase and 600 calories will be added to the vapor phase. The net result is 5,400 calories being produced in the vapor phase and 600 calories being produced in the condensed phase. Since calculations by Kanury (62) suggest that a minimum of 2% and a maximum of 20% of the heat is trans- ferred to the surface in the case of polypropylene rods, an average of 10% of the heat transferred to the surface from the vapor phase would result in a contribution of 540 calories compared to the 600 calories produced by oxidation. These values will, of course, vary depending upon the configuration of the burning sample, but they do show that the oxidative degradation that takes place in the condensed phase can be significant, even if the contribution to the overall scheme is small . 170 2. Flame Retardant Polyester 2. a. Inherently Flame Retardant Polyester: The development of an inherently flame retardant polyester has been undertaken as a proprietary project of the American Enka Company under the direction of Dr. Gerald W. McNeely. Although a portion of this work was carried out in cooperation with the ETIP consortium, all of the actual development work was conducted by American Enka at their own expense; thus the fiber is a proprietary product of American Enka. A variety of aryloxycyclotriphosphazenes were synthesized and screened as potential flame retardant additives for PET. Phosphoni- trilic chloride undergoes nucleophilic substitution reactions with nucleophiles such as alkoxides, thiol ates, primary and secondary amines, and ammonia. Thus a very large number of derivatives are possible. Unfortunately none of the allkloxy derivatives examined, with the ex- ception of the fluorinated derivatives, were found to have sufficient thermal stability to be incorporated into PET. This also was the case with the various amine and ammonia derivatives evaluated. As a result all further work was devoted to the aryloxycyclotriphosphazenes. These were synthesized by the reaction of (PNC1 2 )o with sodium phen- oxide in a faily high boiling solvent such as dioxane or toluene to give the best yield of completely substituted product. In Table XXXIV lists a number of compounds of this type along with some physical pro- perties. Although a large number of phosphazenes were evaluated, only hexaphenoxycyclotriphosphazene (PFR-1) has been extensively studied. Most of the other phosphazenes were ruled out as potential flame re- tardant candidates after a limited laboratory evaluation. Generally this evaluation consisted of first checking the thermal stability of the flame retardant and eliminating most of the compounds that de- composed below 280°C. Compounds with thermal stability greater than 280 C were mixed with PET, then the resultant polymers were either spun into yarns or pressed into films. The oxygen index values of 171 TABLE XXXIV CYCLOPHOSPHAZENE PHYSICAL PROPERTIES Compounds g.p.. °c T6A owl P. % {(C 6 H 5 N 2 PN} 3 {(C 6 H 5 0) 2 PN}„ {(p-BrC 6 H„0) 2 P=N} 3 {(C 6 H 5 ) 2 PN} 3 {(p-CH 3 C 6 H„0) 2 PN} 3 {(p-N0 2 C 6 H„0) 2 PN} 3 {(C 6 H 5 -C 6 H w 0) 2 PN} 3 {(C 6 H 5 CH 2 0) 2 PN} 3 {(C1C 6 H„0) 2 PN} 3 C s? /CH.-Cv / \CH 2 V 116 iih 174 153 120 269 277 49 155 340 290 314 350 275 310 320 280 225 310 275 13.4 13.4 8.0 11.8 12.0 9.6 8.1 11.95 10.3 21.0 * onset of weight loss, N 2 , 10 /min. 172 the yarns and films were determined and used as the initial measure- ment of flame retardancy. Based on 01 results, a few of these com- pounds were as effective as PFR-1 as flame retardants for polyester, but were eliminated as potential candidates because of other defici- encies such as the following: compound would not mix with PET; vapor pressure too high and compound volatilized during yarn preparation; compound too expensive to prepare; compound reacted with PET causing polymer degrad- ation. There are four possible times that PFR-1 can be added to the PET process: (1) before ester interchange; (2) before polymerization; (3) after polymerization; (4) during melt spinning. The purity of PFR-1 has a direct bearing on when it can be added to the process. In the laboratory, very pure PFR-1 (M.P. 1 1 2-1 1 4°C) has been added to the ester interchange reactants and retained through the entire process. However, experience with pilot plant quantities of PFR-1 has indicated that it will contain residual chloride which may cause polymer degrad- ation, depending upon the level present. In order to decrease the effect of chloride upon polymer properties, a minimum contact time between molten polymer and PFR-1 is suggested. Injection of the flame retardent into the molten polymer, followed by mixing, is the pre- ferred method, especially if the flame retardant contains more than 1% chloride. Although less than 1% chloride in the additive is preferred, it is not mandatory. The addition of 10 wt. % of impure PFR-1 to polyester via the pin mixer did not have a detrimental effect on the physical properties of the yarn. It was possible to obtain a yarn with 4 gpd, and elongation of 30%, and a Gardner b value of 4.2. The heat and light stability of the yarn was comparable to that of normal PET yarn. The flame retardant properties of polyester containing phospha- zenes were determined by 01 and the children's sleepwear test (FF 3-71), The 01 measurements were carried out on knit tubes and films. In order to prevent the molten polymer from dripping or melting away from the 173 flame front during burning, glass fibers were incorporated into the knit tubes and films. The relatioship between 01 and wt. % PFR-1 for both knit tubes and films are given in Figure 51 These results indicate that the effect of PFR-1 begins to level off between 13 and 15%. The 01 of PET contair^ng 10 wt. % PFR-1 is 0.255 compared to a value of 0.243 for Dacron^900F. Although 01 measurements are not directly related to any flamm- ability standards, the recent suggestion of van Krevelen (63) of a linear relationship between 01 and char residue of polymers has in- creased the utility of 01 measurements. One of the major ways to de- crease the flammability of polymers is to enhance the char residue formation. The relationship between 01 and char residue of PET sam- ples containing different amounts of PFR-1 are shown in Figure 52 When 100% PET fabrics are tested in the children's sleepwear test the burning polymer drops away form the sample. If PET fails the sleepwear test, it usually does so by the melt drip burning in the bottom of the cabinet. Polyester samples containing at least 8 wt. % PRF-1 have nto failed FF 3-71 in the unseamed state, however, fabrics seamed with untreated cotton thread and ignited at the seam fail the test. From these results it can be concluded that the flame retardancy of PET yarns and fibers can be improved by the addition of PFR-1 to the polymer. This material does not adversely affect the physical properties of PET, in fact, the addition of PFR-1 can improve the heat and light stability of PET as well as improving its dyeability with disperse dyes. When PET yarns or fibers containing PFR-1 is blended with other fibers such as modacrylics or rayon, the resultant blends also have improved flame retardancy compared to similar blends containing normal PET. Preliminary experiments also indicate that polyester/cotton blends in which the polyester portion contains 10% PFR-1 can be successfully rendered flame resistant by the application of existing phosphorus-based topical finishes. 2.b. Grafting Studies: 174 C\J en Q_ CO . -vt- CM CM LT> «^- CO CM CM CM CM CM CM 10 a. +-> c O u CtL u_ Q- 3CH 3 ~C-P<^ J 3 OCH CH 3 A - n NaHCO CH 3 3 iJ^OCH^ IWM "3 0CH 3 CH~ - C - P N J 178-2QQ°C v CH 9 =c' JOCH- + CH-,0H J OCH., ' c N p' 6 5 CH 3 8 0CH 3 (DMVP) Dimethyl l-(acetoxy)vinylphosphonate (DAVP) was prepared by the Go! born route (68). Its precusor, dimethyl acetyl phosphonate, was conveniently prepared in 81% yield from TMP and acetyl chloride at C, 11 O o c 8 fl,ocH 3 P(0CH 3 ) 3 + CH 3 CC1 u L > CH 3 C \ * + CH 3 C1 OCH '3 8^ CH 3 (DAVP) Four acetylating agent and catalyst combinations were evaluated for the enolacetylation of the diemthyl acetyl phosphonate. These were acetyl chloride with zinc borate and with pyridine, and acetic anhyd- ride with triethylamine and with pyridine. The combination of acetic anhydride with pyridine proved superior on the basis of color and con- version (about 85% in 17 hours at ambient temperature). High vacuum distillation proved unsuccessful for the isolation of 178 DAVP from reaction mixtures containing pyridine, acetic acid and acetic anhydride. An involved workup procedure was then used which resulted in significant loss of product due to both its alkaline hydrolysis and its water solubility. The best achieved yield of high purity material was 55%. That DAVP is a low melting solid was apparently not known to previous workers. Dimethyl phosphonomethyl acrylate (DPA) was prepared by acryloyl chloride esterification of dimethyl hydroxymethyl phosphonate. The latter was conveniently sythesized in good yield and purity by a new process recently developed at Hooker. The technique for preparation of dimethyl hydroxymethyl phospho- nate involves cautious , dropwise addition of triethylamine to a stirred and externally cooled mixture of aqueous formaldehyde and dimethyl phosphite. No exotherm was observed until sufficient triethylamine had been added to neutralize the acid constituents. Thereafter, large quantities of free triethylamine must be avoided in order to maintain control of the exothermic reaction. This is especially true early in the reaction where dilution of reactants by product is not extensive. The reaction is conveniently monitored by observing the disappear- ance of DMP using NMR. This is accomplished by observing the reson- ance located at S=0.88 which is one resonance of the doublet (J=697HZ) for the hydrogen attached to phosphorus in DMP. Triethylamine is added until this resonance is completely gone. The residue obtained on stripping under high vacuum proved distill able at 110°C under a pressure of 0.1-0.2 mm. The purity of the residue was adequate for the acryloyl chloride esterification in which triethylamine was used as an acid acceptor. Significant losses of DPA, which occurred during water washings to remove triethylamine hydrochloride, resulted in a crude yield of 65%. Distillation using hydroquinone inhibitor gave a 53% overall yield of DPA based on starting DMP. 179 N(C 2 H 5 } 3 H jj^OCH o* ?, [j/OCH CH 9 =CHCC1 + HOCH/' J ^-^ > CH 9 =CHCOCH 9 P_ J d L 0CH 3 "N(C 2 H 5 ) 3 - HC1 c C X 0CH 3 (DPA) Dimethyl vinyl phosphonate (DVP) was prepared by the Zenftman and Colder route (69). The Arbusov reaction of TMP with excess 1,2-dibromo- thane resulted in poor yields (15-30%) of the desired 2-bromethyl- phosphonate; the major reaction product was dimethyl methyl phosphonate formed by the methyl bromine catalyzed isomerization of TMP. The crude dimethyl 2-bromoethyl phosphonate was dehydrobrominated with tri- ethyl amine. o ^ 0CH 3 P(0CHJ, + BrCH 9 CH 9 Br 110-115 U C N BrCH 9 CH 9 P v J + ChLBr '3 BrCH CH P, ?l y 0CH 3 Q M^OCH 3 9 CH 9 P V ° *- N(C 9 Hj 9 25 u ^ CH 9 =CHP V ° + N(C 9 H c ) 9 -HBr 2 2 V 0CH 3 2 5 3 > 2 N 0CH 3 2 5 3 (DVP) Dimethyl ally! phosphonate (DMAP) was prepared conveniently by the procedure of Arbusov and Razumov (70). This reaction, which involed refluxing excess allylbromide with TMP, resulted in a 48% yield of dis- tilled DMAP. o "x 0CH 3 CH 9 =CHCH 9 Br + P(0CHj~ 76 U C v CH 9 =CHCH 9 P' J + CH^Br (DMAP) Unsuccessful attempts were made to prepare several other monomers containing both phosphorus and bromine using propritary technology from Hooker Chemicals and Plastics Corporation. Attention was then turned to radiation grafting as a technique for imparting durable flame resistance to PET fabrics. The methodo- logy for grafting flame retardants and related materials to 100% cellu- losic materials has been quite well worked out in previously published 180 studies by several groups of workers. However, little has been done with grafting onto polyester. For this reason, initial investigations at RTI under the direction of Dr. Raimond Liepins have dealt with de- vising FR systems amenable to radiation fixation on PET. The work with the pure polyester system was divided into two ssentially separate phases. In the first, efforts were made to deter- mine the effect of distribution of the flame retardant within the grafted polyester system. This was done with two model monomers, one based on phosphorus, and the other on bromine. Di ethyl vinyl phosphonate (DEVP) was tentatively selected for the phosphorus studies while vinyl- bromide (VBr) was used for the bromine investigations. The initial studies were conducted on single polyester yarns (150/96 semidull) supplied in filament form by American Enka Company. The vinylbromide was supplied by Ethyl Corporation and the diethyl vinyl phosphonate synthesized by Hooker Chemicals and Plastics Corporation. fin The grafting procedures have involved Co,, radiation to initiate the free radical reactions. A series of initial investigations showed no discernible differences in the percent weight gain obtained with scoured and unscoured yarns; thus all subsequent work has been carried out on unscoured samples. The samples were de-gased by means of three _5 freeze-thaw cycles to at least 10 torr. In most cases, the grafting was performed using a mutual irradiation technique in a small glass ampule at dose rates of from 0.01 to 0.1 Mrad/hr. Following irradia- tion, the fiber samples were extracted with solvents for the homo or copolymers, first at room temperature and then at elevated temperature. The extracted fibers were vacuum dried to constant weight and stored in a desiccator for subsequent evaluation. Data for weight gain achieved during grafting was obtained following this drying procedure. In general, the vinylbromide was found to graft with little dif- ficultly. The percent weight gain increased linearly with total dose (time) without any induction period when grafting from a DMS0 solution at room temperature. Preswelling of the fiber in ethylene dichloride produced a parallel dependence of weight gain on total dose, but dis- placed 1% higher on the weight gain axis. The addition of a small 181 amount of DMSO or hexamethylphosphoramide increased the rate of graft- ing markedly. This was followed by a gradual decrease with increasing amounts of either of the two materials. The largest increase in weight gain was realized at h to 1 molar concentration of either of the two liquids. A similar rate effect has been observed in eariler experi- ments on the grafting of styrene onto polyester. In general, the best approach to increased weight gain seemed to be through the use of an organohalogen compound which is capable of functioning as both a swelling agent and a radiation sensitizer. The use of solubility par- ameter correlations in the choice of non-halogen containing swelling agents led to no improvement in weight gain in these studies. On the other hand, the use of organohalogen compounds as swelling agents increased the weight gain by more than 100%. In summary, the evalua- tion of the various vinyl bromide grafts gave the following results: filament diameter increased uniformly with the percent weight gain from 14.4 for the 2.9% graft to 16.9 for the 24.4% graft. 01 increased uniformly with the percent weight gain from 20.8 for the 0.5% graft to 27.9 for the 24.4% graft. char yield the increase in char yield is not uniform with the percent weight gain, however, grafts with percent weight gains of more than 5 gave increased char yields of 1 to 6% above control. tensile strength showed only a slight, non-uniform variation (4.4 to 4.8 g/denier) for the various grafts and in no case did grafting impair the tensile strength of the fiber. elongation (reported as the ratio of the elongations of the sample to control x 100) the most spectactular change that the grafting produced in the fibers was in imparting to the greater elasticity; the increased elasticity (elongation at the yield point more than 240%) was obtained at the lower grafting levels (up to 5.5%) which then decreased slightly to about 190% as the grafting level incresed. The sole ex- ception was the 24.4% graft, which was the most elastic fiber encountered and had an elongation of 298% and a tensile strength of 4.6 g/denier. Grafting of VBr, as well as VBr 2 , and VC1 2 led to colored grafts. It was found that this problem could be elminated by co-grafting with other monomers, grafting in the presence of morpholine, or grafting at temperatures below -44°C. Grafting of VBr on PET fiber that did not contain Ti0 2 led to 182 clear grafts, however, drying of the extracted, grafted fiber in a vacuum oven at 50°C resulted in a colored graft. Thus, the absence of Ti0 2 does not eliminate the discoloration of VBr grafts. Grafting of VBr at 81 °C did not seem to offer any advantages over RT or below RT grafting as the add-on was only 17.7%. Grafting of vinyl bromide at -64°C led to a colorless graft (8.2% add-on at 2.1 Mrads) which discolored only slightly when dried in a vacuum over at 50°C/16 hrs. Since the samples in the grafting experiments were all in the form of individual yarns or fibers, it was necessary to develop a modified technique for measuring the oxygen index of these materials. After several attempts, a glass holder was developed which permitted one fiber or one yarn to be measured in a vertical position in the oxygen index tester. This material not only permitted the determina- tion of the 01 value, but also an estimation of the char yield as the material burned in the tester. A paper describing the details of this procedure, including data on PET fibers and comparison of the fiber samples, has been published. (64) The effect of the location of flame retardant grafting on the PET yarns was then investigated. Samples were prepared using VBr and DEVP in which it was expected that the retardants would be grafted either on the surface or uniformly throughout the yarns. This was accomplished by varying the amount of ethylene di chloride preswelling to which the samples were subjected. For comparison, samples were also prepared having the ungrafted flame retardant homopolymer de- posited on the suface of the yarns from solution. These samples were studied by 01 and scanning electron microscopy. The scanning electron micrographs were obtained from l-2y thick cross-sections. These were obtained by embedding the fibers in a Dow Epoxy resin and cutting the sections with a Reichert ultramicro- tome. A scanning electron micrograph showing the thickness of a section through the embedment is given in Figure 53 . The fiber cross-sections are seen in the top of the micrograph. It may be noted that all these micrographs were obtained on sections which were 183 deposited on carbon stubs with no further preparation. No carbon coat- ing was required. In some of the sections part of some fibers would break loose from the embedding material revealing the thickness of the fiber cross- section. This is shown in Figure 54 where the fiber cross-section is shown to be lu thick. The plot of intensity of fluorescent X-rays versus energy of the X-rays for PVBr coated fiber is given in Figure 55.c Location of the L and K emission lines for bromine are given by the vertical bars at the top of the figure. The secondary electron image of the cross-section from which this plot was obtained is given in Figure 56. The L emission line of the plot (the white part of the spectrum in Figure 55) was used for mapping, i.e. to indicate from what part of the sample the X-rays are emitted. Figure 57 shows a map indicating extremely good definition of the boundaries of the fiber seen in Figure 56. In order to ascertain that such a map actually indicated the site of bromine atoms and not especially favorable top- ography at the boundaries of the fiber for emissions of all X-rays, a map was obtained with a "window" corresponding to background (see white part of Figure 58 for the "window"). Such a map is presented in Figure 59 and is featureless. This procedure was routinely followed for subsequent maps. The secondary electron image and bromine L map for the fiber to which vinyl bromide had been grafted are given in Figures 60 and 61. It can be seen that in this case the graft is localized at the surface. The secondary electron image, plotted as intensity versus energy of emitted X-rays and the phosphorus K plot for the fiber to which diethyl vinyl phosphonate had been grafted are given in Figures 62 and 63. The graft is shown to be distributed evenly through the cross-section of the fiber. This interpretation can be made with con- fidence because of the excellent resolution demonstrated in the pre- vious maps. An enhanced version of Figure 63, in which the contrib- ution to the map from some of the background evident in Figure 63 has been subtracted, is shown in Figure 64. Once the location of the model flame retardant grafts had been 184 \0p FIGURE 53. SEM photomicrograph showing a cross-section of embedded PET fibers. I 1 5p FIGURE 54. PET fiber cross-section partially freed from embedded medium. 185 FIGURE 55. Energy spectrum for x-rays emitted from a PET sample solution coated with poly( vinyl bromide). 5^ FIGURE 56. Secondary electron image of sample yielding energy spectrum in figure 55. 186 5jJ FIGURE 57. Bromine L x-ray map for sample shown in figure 56 using "window" indicated in figure 55 (white portion of spectrum) FIGURE 58. Energy spectrum from sample shown in figure 56 showing "window" on background map used for map in figure 59. 187 5p FIGURE 59. Background x-ray map obtained from cross-section shown in figure 56 using window in figure 58. i 1 5p FIGURE 60. Secondary electron image from a PET fiber to which vinyl bromide has been grafted. 188 5.P FIGURE 61. Bromine L x-ray map for fiber cross-section depicted in figure 60. 5jj FIGURE 62. Secondary electron image from a PET fiber to which diethyl vinyl phosphonate has been grafted. 189 I 1 5p FIGURE 63. Phosphorus K x-ray image of the cross-section depicted in figure 62. -sT FIGURE 64. Enhanced version of figure 63 190 determined with certainty, it was possible to develop a correlation betweem the retardant location and its efficiency as measured by 01. These results are presented in Table XXXV. The placement of the PVBr in a uniform distribution throughout the fibers enhanced its flame retardant efficiency the most, followed by surface grafting and solu- tion coating. The coating method seemed to be the least efficient in terms of raising the 01. Uniform grafting also resulted in a more efficient utilization of the poly(DEVP) as compared to surface grafting Apparently those retardants such as PVBr and poly(DEVP) which do not posses a high degree of thermal stability become more efficient as they are incorporated deeper inside the filaments. In order to confirm this interpretation, an attempt was made to graft vinyl bromide in such a way that it would reside predominately in the core of the filaments. This involved preswelling the fibers in ethylene dichloride and VBr followed by exposing the treated samples to ambiant air before irradiation. It was hoped that in this way the VBr in the outer levels would be lost by evaporation leaving the majo- rity of the retardant in the core. This technique was only partially successful. Core grafting was achieved but only at a low add-on level. A core grafted sample was produced with a 1.2% add-on of grafted retardant and this sample ex- hibited an 01 value of 21.9. In order to reach this same 01 using homogeneously grafted VBr required an add-on of 2.5%. If these data can be considered as reliable indicators of flammability behavior at these low levels of treatment, this would tend to confirm the inter- pretation of the previous data. In a concurrent series of experiments, various phosphorus com- pounds deemed to have potential as commercial flame retardants for polyester and polyester/cotton blends were evaluated in radiation grafting experiments on PET yarns. Some of the compounds studied were specifically designed and synthesized by Dr. Howard Day of Hooker Chemicals and Plastics Corporation. Other were obtained from a var- iety of commercial suppliers. The monomers evaluated are tabulated in Table XXXVI. In several cases copolymerization and 191 TABLE XXXV FLAME RETARDANT EFFICIENCY OF VBr ON PET AS A FUNCTION OF LOCATION VBR GRAFTS 01 25.2 26.3 23.5 21.9 %F.R. 55 18.8 18.8 1.2 TGA, initial - 138° 150° - Substrate H 325° 353° 353° Location y $ wt. loss 3__ * Low Temp. Decomp. t — ^^^ ^— t \ end i Dec. pt. subst. •H w ^. -p > start — * ll 1 \ \l High Temp. Shoulder N * — end % Residue V High Temp Break 0% V_ i 50°C * 820 Temp FIGURE 66. Analysis of grafted PET fiber thermogram. 199 c «« o , °H i -* »— «o co .g L. CD CT CI CO c o> co >J a> «/> 01 cc c_> o fr« m ro r>» • o o o m • ■*-> o ■!-> CO 3 Q oo w c o co c o o co CM in 3 ■o Iff o> o vo CO • vo»— o o in co <_> o m CM Iff iff o o »* to l/» ©» • CM lf> o VO o <_> x o in co • o o m CO o o vo CO CO Iff o o o in • CM 00 o o o in o m o CO CO VO Iff <9 01 m w 3 O ** •— •a m in ••- . 01 Iff ••-> O 3 U tl« T3 O OC >>-i- in •—«—«/» O &« UJ C 01 in o% a. o oe *« ** CD 00 © o 1. o o vo 3- •^ (/> 01 QC <_> »« o cm © CM in •— o CM in o o o o CM Iff in O o «■ o • r^ ^- CM «— O o CO c o a> c o 200 the substrate decomposition point only a little (13°C), forms more char at 500°C (12%) and require muchJwgher temperatures (633°C) to pyrolyze away the formed char. Fyron-BB-VBr grafts decompose at temperatures near 100°C, depress the substrate decomposition point by at least 30°C, and form a 12% char at 500°C which is extremely thermal- ly stable. In one run there was still 1.9% of cha^Jeft at 820°C. Further studies were carried out on the Fyror-ftB system in com- bination witJk other monomers and the results are shown in Table XXXVII. If the FyroVfiB was used in conjunction with VBr, the results were much better when the preswelling of the fiber prior to grafting was eliminated. Preswelling with ethylene dichloride seemed to produce a more rigid char but lower 01 values. Presumably the differences observed were due to the fact that the fibers which were not preswol- len have the grafting predominantly on the surface of the fiber. How- ever, it is not clear why this should produce the observed effects. Grafts were also attempted using DMVP and DAVP. The products from these trials gave anomalously high 01 values (over 30 at add-ons as low as 4%). These values are probably not truly reflective of the flammability characteristics of the fibers since they were found to exhibit an extremely large and rapid shrinkage when exposed to the heat source. Thus, ignition in the oxygen index test was a major pro- blem and probably contributed significantly to the anomolously high 01 values. At present it is unclear whether a similar behavior would be observed if fabric rather than fiber samples were used and what im- plications this might have for the development of new FR treatments for 100% PET fabrics. A variety of other monomers have been studied, both by themselves and in various combinations. Selected results obtained with the homo- polymer grafts are shown in Table XL. From these data it is obvious that the efficiencies of the flame retardants vary quite widely de- pending upon the structure of the monomer. Also, a few of the monomers did not yield acceptable levels of homopolymerization and several techniques have been investigated in attempts to circumvent this pro- blem. It has been found that DMVP and DAVP can be homografted to 201 TABLE XL TYPICAL DATA ON GRAFTED PET FIBERS % Add-on 01 Char Filam. % Elong. Yield, % Diam. ,u at yield pt. Tenacity, g/den NDPA (postirradiation graftinq) 3.0 10.4 24.7 28.3 14.4 263 24.0 13.5 175 4.06 3.80 DMVP-VBr 1.8 4.7 26.3 33.1 1) 15.4 232 1) 14.7 265 4.13 4.73 DMVP-VBr 0.6 31.8 1) DMVP-VClo 8.3 26.7 1) 15.0 234 5.13 DMVP-VBr-VClo 4.2 29.4 1) 13.5 246 4.80 DMVP-VBr -VCl 2 1.7 34.1 1) 13.8 242 4.53 DMVP-Acrylami de 20.4 27.9 17.1 15.0 188 4.53 DAVP ■ 1.2 23.3 1) 15.0 249 4.13 DAVP-VBr 1.4 26.3 1) 248 4.93 DAVP-VBr 1.5 29.4 18.8 DAVP-VC1 19.6 24.3 22.2 15.0 170 4.00 DAVP-VBr-VCl 2.8 36.6 1) 13.9 226 4.47 202 DAVP-VBr -VCl ■2 31 c 2.8 .4 DAVP-Acrylami de 8.9 28 .3 DPA 32.9 27, ,1 DPA-VBr 14.4 27. .6 DPA-VBr 14.9 28. .6 DPA-VBr-VCU 27. 18.1 6 DPA-VBr -BCl TABLE XL (continued) 1) 15.0 261 4.67 21.5 13.7 224 4.73 23.6 16.5 138 3.47 26.3 15.3 219 4.73 16.1 14.1 214 4.73 18.4 15.0 223 5.00 21.4 28.6 18.0 15.0 213 4.80 BDPOM (postirradiation grafting) 10.3 29.7 1) 13.9 197 3.00 BDPOM- Fyrol^BB 10.5 29.4 22.3 15.0 170 5.33 BDPOM-Acryl amide 7.5 32.8 14.2 194 4.33 DBPA 9.1 27.6 17.1 13.7 152 4.67 DBPA-FyrolWBB 6.4 28.3 36.0 13.9 111 4.53 DBPA-Fyrol^BB-VCU 17.8 30.5 14.5 14.5 107 4.07 DBPA-Acryl amide 71.5 31.8 17.3 150 3.20 203 TABLE XL (continued) DBPM 5.6 26.7 13.3 215 4.53 DBPM-FyrolWBB 5.0 30.9 23.4 13.5 145 4.20 nRPM-Fvrnl^yBB-VCl, 14.9 213 4.40 11.5 13.9 223 4.20 1) 13.9 213 3.20 1) 228 4.40 1) 234 3.47 5.9 ■ ■ ■ ■ — L 31.8 DBPM-Acryl amide 10.6 30.1 "Hexa" 2.1 31.8 "Hexa"-VBr 1.8 33.5 l^BB "Hexa"-Fyro 1.5 33.5 l w BB-VBr "Hexa"-Fyro 13.7 33.7 22.0 13.3 TBPM 13.5 32.5 17.4 13.5 215 4.27 TBPM-Fyrol^BB 26.3 14.4 232 4.60 19.6 32.5 TBPM-NDPA 10.0 30.9 VBr 9.4 28.6 32.2 13.5 182 4.73 15.7 163 3.87 *1) The melt was extremely "liquid" and it was not possible to obtain easily reproducible char yield data. 204 adequate add-ons (~10%) if complexed with anhydrous ZnClp before irra- diation. Complexing of FyroS^BB with ZnCl 2 led to more than 40% homo- graft. No doubt both ionic and free radical grafting species are gen- erated in such a system and are responible for the high add-ons. High add-ons can be also achieved with monomers that do not graft easily if one utilized the gel effect. Use of allyl methacrylate, diallylchoromethyl phosphonate, and N-allyl acryl amide as the cross- linking agents in 1% concentration have been evaluated and it has been found that the methacrylate is the most efficient monomer for this purpose. Some exploratory grafting work with BABA-50 showed it to also be a^yery efficient gel-former. For example, at 1% it increased the FyroV-tfB add-on from less than 1% to more than 10%. The results obtained from grafting trials using combination of monomers are shown in Table XL. Again, a wide range of efficiencies were found. Particularly good results were obtained when DMVP or DAVP was used in conjunction with a bromine containing monomer. The DVPM and "Hexa" systems also showed promising effectiveness in render- ing the fibers flame resistant, as measured by oxygen index. Since all of these results were obtained on fiber and yarn samples, there is a definite possibility that some of the high 01 values and some of the variability observed may be artifacts of the experimental methods and thus not truly reflective of the flame retardant ability of the compounds. By using the best PET fiber grafting technique on PET fabric, graft samples as large as 3" x 22" with VBr were obtained with add-ons above 20%. Checking one other swelling agent (DMSO) confirmed the previous experience with fibers in that organohalogen compounds are superior swelling agents and radiation sensitizers and lead to the best add-ons. Grafting on DPA on PET fabric led to about 7% add-on. In this case apparently the best fiber grafting conditions were not the best for the fabrics. Some VBr and DPA 01 data on strips of the grafted fabrics are listed in Table XLI. The high 01 for the control is an artifact of 205 TABLE XLI Initial Grafting Results in PET Fabric Flame Retardant % Add-on 01. Char Yield Control 25.2 Vinyl bromide 26 30.1 24.9% Dimethyl phosphonomethyl acrylate 7 25.2 206 this particular type of fabric construction. The TGA evaluation of 26 chemically different grafts on PET fabric permitted a number of generalizations to be made. Some "struc- ture vs decomposition point" data are summarized below: (1) Grafts without a decomposition point below that of the substrate are the most efficient flame retardants. (2) Aromatic bromine grafts are the only bromine grafts without a de- composition point below that of the substrate and is the most efficient bromine containing flame retardant. (3) Two bromine atoms on the same carbon atom lead to a more thermally stable graft than two bromine atoms on adjacent carbon atoms. (4) One aliphatic bromine atom leads to a more thermally stable struc- ture than two bromine atoms on adjacent carbon atoms. (5) 8 * ■P I OCH -CNHCH o -P-0CH~ I 3 is more thermally stable than II * -C0CH 9 -P-0ChL 2 | 3 0CH 3 (6) -P-0CH 2 CH 2 C1 0CH 2 CH 2 C1 is mQre thermally stable than ? -P-0CH 2 CH 3 0CH 2 CH 3 (7) Two isolated chlorine atoms (as compared to perchlorocyclopenta- diene structure) on the same carbon atom lead to a thermally un- stable structure. (8) Perchlorocyclopentadiene structure is a relatively thermally stable structure and as a result is a very efficient flame retard- ant for PET fiber. In the bromine graft series the apparent thermal stability of the 207 graft and its flame retardant efficiency may be related to the ali- phatic hydrogen -bromine ratio (see Table XLII); the lower the ratio the higher the efficiency. The most efficient flame retardants appear to be those with no low decomposition point, see Table XLIII. The amide vs. ester and bromine graft structure-thermal stability correl- ations agree with pertinent: data in the literature, see Table XLIV. 2.c. Topical Treatments Utilizing Conventional Fixation Methodology: Although the problem of devising flame retardant treatments for 100% PET fabrics was itself outside of the scope of this project, a few preliminary experiments were carried out on 100% PET to provide a background for the subsequent work with the blends. In one study the feasibility of applying flame retardants to the polyester by dyeing techniques was evaluated. In an atlernate approach to the treatment of PET fabrics, heat cureable and ammonia cureable phosphonium derivatives were examined. Of these, onlv~±he precondensates of tetrakis-hydroxymethyl phosphonium sulfate (THPS^) produced finishes which appeared to have significant effect on the PET flammability. Of particular interest was the con- densation^product formed by heating THPS^with trimethylphosphoramide (MCC-100^) and trimethylolmel amine at reflux for 16-20 hours. When this material was applied to 100% PET with urea in a pad-dry-NH~ cure process under laboratory conditions, the results shown in Table XLV were obtained. The fact that all of the samples exhibited char length of less than 5 in. when burned without support whereas the highest level burned completely when stitched with three fiberglass threads may indicate that even though the treatments result in increased levels of char production the retardant is acting predominently by lowering the melt viscosity of the PET. This would not be unexpected in view of the known decompositon of phosphonium compounds to acid anhydrides. It would also be consistent with the results obtained with the DAP treated PET described earlier. It would also be consistent with an observation that small amounts of HUSO- will allow PET to 208 TABLE XLII Flame Retardant Efficiency as a Function of Thermal Stabilility Compound Aliphatic H/Br Dec. Pt. 0I/l%Br Br -S-o-- Br Br ll •C-0-CH 2 CH 2 Br 0-(Q-Br Br Br CH, II I 2 ■COCH,-C-CH.,-Br 2 l 2 CH, « I z Br Br I ■C- l Br C0CH o -C-Br 2 i BR ■CH.-C- 2i BR 2/3 303°C 300°C 220°C 204°C 1.5 1.2 1.1 1.0 2 i BR 150°C 0.5 2 i | 2 Br Br 2.5 132°C (0.2) ^8r Br d 127°C CH 3 ■CH.-C-CH-CH,- 2 i i 2 BrBr 122°C 209 TABLE XLIII GRAFTS WITH NO LOW TEMPERATURE DECOMPOSITION POINT AT Substrate Dec. Pt. Monomer Substrate Dec. Pt. Control PET 355° C TBPM 303° Hexa 325° NDPA Fyrol^L^B 325° 342° 52° C 30° 30° 13° 210 TABLE XLIV MAXIMUM STABILITY OF POLYMER REPEAT UNITS* -CH 2 -CH- T dec (N2) 6oo°c U80° 260° Vbr (138°C) -CH-CH- I I CI CI CI - CH 2"f- CI > 215-220 Vbr (110-127°; 20l+°) C1C1 -chJ-c-ch 2 - C1C1 320 •CH 2 -CH- 386 C -CH 2 -C- F 1*30 o -CH-CH- F F 365° * Reference 71. 211 TABLE XLV Effect of Phosphonium Precondensates on PET Flammability Average Initial Average Char Length After 50 Launderings 4.1 in. (7.0 sec.)' 3) 4.3 in. (6.4 sec.)' 3 ) 2.5 in. (4.2 sec.)' 3) % Add-on Char Length 22. 4' 1 ' 1.0 in. 10.2< 2 > 2.9 in. (6.4 sec.)' 3 ) 7.8< 2 > 3.2 in. (5.1 sec.)' 3 ^ 4.5 < 2 > 2.8 in. (4.9 sec.)' 3 ) (1) Prepared by Clemson (2) Prepared by United Merchants & Manufacturers (3) Average afterflame time; at each level, at least one sample exhibited an afterflame time of greater than 10 sec. 212 achieve very small char lengths. 213 STUDIES ON POLYESTER/COTTON BLENDS 214 In order that the results from all of the participating laborator- ies might be comparable, a single lot of fabric was chosen to be used as the standard for all of the ETIP work. This was a 4.2 oz. 50/50 poplin purchased from Dan River Mills and prepared by the Old Fort Finishing Plant of United Merchants. Unless otherwise specified, all results are reported on this standard ETIP fabric. FLAME RETARDANT SYSTEMS BASED ON PHOSPHORUS ALONE Based on the knowledge that phosphorus may exhibit flame retardant action in both the fuel generating and fuel consuming processes, it should be possible to design organophosphorus compounds capable of pro- ducing both effects on polyester/ cotton blends. ® 1. Antiblaze^^lQ One organophosphorus retardant which would appear to meet this criterion is Antiblazev-^19 produced by Mobil Chemical Company. The literature indicates that this material is a mixture of cyclic phos- phonates, probably a dimer and trimer, with closely related structures but different volatilies, and having a total phosphorus content of 21%. A series of 65/35 and 50/50 polyester/cotton blend fabrics was treated with this retardant and evaluated by the 45 angle subjective burn test. The results of these tests along with the corresponding oxygen index values are given in Table XLVI. As might be expected from the previous discussion of testing methodology, there seems to be little correlation between the observed oxygen index values and the behavior of the fabrics on the 45° angle burn test using edge ignition with a wooden kitchen match. The 45° angle test, as carried out, should constitute an extremely severe test. Those fabrics which are designated as "difficult to ignite" resist ignition for periods as long as 20 seconds, indicating that they should be more than satisfactory for passing most vertical burn tests. Because it seems likely that the most important factor in causing this discrepancy in results is one of geometry, these samples were also evaluated by bottom ignition oxygen 215 TABLE XLVI FLAMMABILITY OF BLENDS TREATED WITH ANTIBLAZE^19 SP (1) 01 45°Burn 3.0* 2 ' 25.0 Does not burn 2.2* 2 ' 23.0 Difficult to ignite 1.4 (2 * 22.2 Burns 3.3^ 25.5 Does not burn 2.2^ 23.8 Difficult to ignite 1.5^ 22.5 Burns Calculated assuming Antiblaze^ig 2UP. 2 On 65/35 polyester/cotton blend 3 On 50/50 polyester/cotton blend 216 index. The BOI values were much more consistent with the vertical test results than the 01 values. Thus it would appear that Antiblazev-yi9 should be particularly effective in vertical upward burning situations. Because of the rather unusual activity demonstrated by Anti- blaze^o^, fabrics containing this material have been evaluated more thoroughly than would normally be done with compounds of undesignated structure. A series^pf 50/50 cotton/polyester blends treated at five levels of Anti blazed 19 ranging in phosphorus content from 0.94 to 3.20 have been evaluated by isoperibol calorimetry and the results are tabu- lated in Table XLVII. The char yields, heat evolution and rates of heat evolution from these samples clearly indicate that this particular flame retardant does not exert its effect in any manner similar to those previously studied. For a phosphorus content of 3.20%, char yields were increased only from 6.6% to 17.4% and the heat release was reduced by about 500 calories from the control. This compares to char yields of approximately 18% and a heat release reduction of almost 800 cal/gm for a 50/50 blend treated with a phosphoric acid to a level of 1%P~«. It was also noted that the behavior of fabrics treated with Antiblaze^-^19 was very susceptible to changes in burning environment. In the rela- tively high air-flow of the isoperibol calorimeter, fabrics burned which would not burn in open air 45° angle test. This may indicate some type of physical effect rather than chemical retardation. Unfortunately the sources of these discrepancies have not been found. Further work with this material was abandoned after repeated attempts to fix the re- tardant to produce a durable finish on the 50/50 blend. 217 E X <] cr> co vo «3- co r^* o vo «* co cr> vo vo «* *t co cm r-. CM CM CM CM CM CM cx> © 00 _i CO o ac iS) Z3 CO UJ oc >~ OC CC O _l - _J o Q- o Lf> o lo Q J— UJ u OJ JO u cu ■♦-> oc CM CO CO «* CO 00 r • • # •• • CM ft ** O *3" r- +1 +1 +1 +1 +1 +1 ID CO r- CO O O CTi CO IT) lo CM CO oc r— CO CO CO O. O s. «* O o o O +-> 50/50 Dacron/Cotton □ 65/35 Dacron/Cotton + 100% Polyester 10 % MD3P 20 FIGURE 67. 01 as a function of % MD3P (unfixed). 220 16 ° For Static Oxygen Bomb • From Isoperibol 14 12 % Char 10 % P FIGURE 68. Char formation as a function of MD3P (unfixed) on cotton 221 50/50 Dacron/Cotton with MD3P Unfixed 18 16 14 12 % Char 10 O For Static Oxygen Bomb ► From Isoperibol % P FIGURE 69. Char formation as a function of MD3P (unfixed) on 50/50 PET cotton. 222 The results of the static oxygen bomb calorimetry on the fabrics containing the unfixed MD3P are shown in Figures 70 and 71. The high values obtained in both the Y/(l-X) and theAH 2 /^H°) treatments of the data are easily interpretable in terms of relatively minor contributions to the flame retardant activity by condensed phase reactions. Isoperibol calorimetry produced the results shown in Tables XLVIII and XLIX. The ratio of AH, toAhL, which is a measure of the incomplete combustion of the fuel gases liberated on pyrolysis of the fabric, in- dicates an increase in the unburned vapors with increasing amounts of unfixed MD3P. This suggests vapor phase activity. The MD3P was alsa studied in a fixed form. The retardant was ap- plied with an AerotexN-^23 resin and a phosphoric acid catalyst using a drying of five minutes at 60°C followed by curing at 150 C for two minutes. After fixing, the 01 values for the treated blends were increased , whereas those for the 100% cotton fabric were essentially the same as with the unfixed form (Figure 72). Based on the hand of the fabric, it would seem that the unfixed samples have significant amounts of surface treatments on the blends, whereas penetration is possible in the case of 100% cellulose. With the materials present in an unfixed form the flame retardant is probably released ahead of the flame front, decreasing its efficiency. Fixation reduces the volatility of the material and retards its volatilization so that more effective vapor phase action can be achieved. Such an interpretation would seem to be consistent with the results of the subjective 45 angle burning evalua- tions. In these tests, the flame retardant was found to be much more efficient than would be predicted on the basis of 01 values. As in the case of Antiblaze^-^19, it was necessary to resort to bottom burning oxygen index (B0I) to evaluate the flammability of these samples (see Table IV), Calorimetric evaluation of the fabrics has also been carried out. Samples of 50/50 polyester/cotton treated with the MD3P in a fixed form would not support combustion at levels of 25% or greater. The char for the static oxygen bomb was prepared by holding a match under the fabric until charring was completed. In the isoperibol, the sample containing 223 c o 4-> O a> o o c r- O 3 J- — o q> O O O i* in © -^. o o r— in o A co CM o in o in c O O ■o Qi X Q. CO Q 3E M- ^_ S O Cl X >> u •r— U X «♦- c c 0) => •1— Q. o in CO Q c c o o V) +■» •I- +J s- o CO 224 CO CM o o to ^>» c •— o 3 S. •— o r— «0 01 Q <_) O fc-S Lf) o ^. o o i— Lf> Q in c o +-> ■p o o Ul o. o in "■■>» o in o -a «o a. E o u o 4-> O O c o ■o 0) X c 3 O A co CO O J- o a* oo in o o < C\J < o o C\J a: C3 225 TABLE XLVIII 100% COTTON FABRIC TREATED WITH MD3P (UNFIXED) %MD3P -AH os cal/qm -AH,, cal/am AHt/AH 2 , cal/gm -AH,, cal/gm AH-J/AH2 23.6 3650 1992 .55 12.9 3395 2219 .65 7.9 3433 2447 .71 5.2 3456 2506 .73 226 TABLE XLIX 50/50 PET/COTTON TREATED WITH MD3P (UNFIXED) %MD3P -AH 2 , cal/gm -AH, , cal/gm AH 1 /AH 2 21.6 3848 2099 • .55 10.6 3862 2345 .61 5.9 3878 2466 .64 3.7 3964 2593 .65 227 LOI % 100% Cellulose > 50/50 Dacron/Cotton 26 — D 65/35 Dacron/Cotton 24 — ^^? &- 22 20 18 1 i 1 » 1 — 10 20 30 % MD3P FIGURE 72. 01 of MD3P (fixed) on PET/cotton blend fab ncs 228 26.6% MD3P did not burn completely and left a char residue of 50.1%. It should also be noted that 50/50 blend fabrics containing 19.6% MD3P were extremely difficult to ignite in open air burning, but would burn the entire length after ignition. They would undoubtedly pass any of the standard vertical tests using 3 or 12 second ignition. Samples of 100% cotton with 27.7% MD3P behaved in manner similar to that of the 50/50 blend with 19.6% MD3P There are other indications which support this suggestion that the entire mechanism of action may change when MD3P is fixed. Figures 73 and 74 show the char yields achieved as a function of the phosphorus content. In the case of the fixed retardant the chars increase signifi- cantly above 2%P, whereas the unfixed system shows a slight decrease in this area. As would be expected, the dependence of Y/(l-X) on log %P exhibited a similar effect (Figure 75). The unfixed reagent produced a low order of dependence whereas in the fixed form a steep slope similar to that found with typical effective condensed phase retardants was found. This behavior was also reflected in the values ofAH-AhL as shown in Tables L and LI. In an attempt to understand the origin of this anomalous behavior, elemental analysis has been carried out on the chars from the burned samples. With the unfixed retardant on both fabrics, the phosphorus retention was low and showed no apparent correlations with initial re- tardant add-on. On the 100% cellulose fabric, the nitrogen retention was only slightly higher than that of phosphorus; but with the 50/50 blend, the nitrogen was lh to 2 times that of the phosphorus retention. The fixed samples showed higher retention than the unfixed. The much higher retention observed in both cases for the samples with the highest weight add-on compares with the unusually large char yield and low Y/(l-X) values observed. For both fabrics, nitrogen retention was \H to 2 times that of phosphorus. The results of TGA studies in this system agree with the data ob- tained in the previous studies. The unfixed samples show \iery little residue at 500°C and the material seems to exhibit no pronounced cata- lytic effect on the polymer decomposition. These results are presented 229 % Char 32 28 Unfixed • S.O.B. A Isoperibol 24 — Fixed O S.O.B. A Isoperibol 20 16 /> 12 — 8 - 4 ■ I J » • • 1 . % p FIGURE 73. Effect of fixation on chars from MD3P treated cotton. 230 MD3P on 50/50 PET/Cotton % Char 40 - 36 - 32 - 28 - 24 - 20 - 16 - 12 - 8 - 4 _ % P FIGURE 74. Effect of fixation on chars from MD3P treated 50/50 PET/ cotton. 231 C\J o >> o c CD •r— U •r— «+- <4- CD ■o 03 4-> S- fO 00 o D_ CO Q 5: 4-> ro X o^ CO r^ <£> LO «* CO o o o o o o o x i IT) a: 232 TABLE L 100% COTTON FABRIC TREATED WITH MD3P (FIXED) XMD3P -AH 2 , cal/gm -AH,, cal/gm AH 1 /AH 2 27,74 2578 1920 .74 20.97 3173 2142 .68 14.95 3299 2225 .67 9.11 3371 2355 .70 233 TABLE LI 50/50 PET/COTTON TREATED WITH MD3P (FIXED) %MD3P -AH 2 , cal/gm -AH-j , cal/gm AH-j/A^ 25.90 2584 1999 .77 19.59 3107 2099 .68 12.09 3766 2246 .60 7.15 3928 2330 .60 234 in Figures 76 and 77. The weight loss which occurs between 140 and 210° corresponds with weight loss observed in the case of pure MD3P as shown in Figure 76. The weight loss for MD3P at about 85 occurs at the melting point and is probably due to solvent trapped during the re- crystallization process. This has also been indicated by NMR. The weight loss which occurs between 140° and 210 agrees with calculations for loss of formaldehyde. Analysis by TGA of 100% cellulosic material with the MD3P in fixed form (Figure 78) shows no significant modification except at the highest phosphorus add-on where a large change is observed in both decomposition temperature and char yield. The blend fabrics with the fixed reagents (Figure 79) showed this same large decrease in initiation temperature at the two highest levels of phosphorus incorporation. There is no apparent correlation between percent phosphorus and the amount of char residue with the blends; but the change in slope during decomposition, which is likely due to a change from cellulose to PET decomposition, occurs at a lower weight loss as the percent phosphorus increases. Differential thermal analysis of the cotton fabric containing the unfixed reagent shows a decrease in the endotherm nadir attributable to the cellulose decomposition from 375° to 340° as the phosphorus content increases. The 50/50 blend fabrics exhibit no observable change in the nadir of the endotherm. In these materials a shoulder appears 10° to 15° below the nadir which could be caused either by decomposition of the MD3P or by reaction with the cellulose. The DTA of neat MD3P has four endotherms with nadirs at 95°, 180°, 365° and 425°C. The first three endotherms are attributable to the melting point, the probable formalde- hyde loss and the large weight loss observed by TGA. The origin of the endotherm at 425 is not known. Fixed MD3P on 100% cotton fabric shows no significant changes in DTA except in the case of the samples containing 2.8% phosphorus. This is in keeping with the TGA results. In the case of the 50/50 blend fab- ric, the nadir of the decomposition endotherm decreases to approximately 300° for samples containing 2.0 or 2.6% phosphorus. Thus it would seem 235 >'< \\ r*/ 1 1 If ""* 1 u • • Q f0 Q. CXI p— o o s: O) o ST 1 1 i 1 i c: ■ 1 •—< 1 OsJ ro «* u-> ID 1 § := 6w 4 *}M U3 1*^ C3 236 ■»-> /v* 3 2^\ t^ \V\l\ 50/50 PET/Cotton \\v \\v MD3P Unfixed \\v 1- 2.2% P \\\\ 2- 1.08 \\V 3- 0.6 \\v 4- 0.4 \\\ Untreated Vv - — w w » 1 • 250 350 Temperature, C (Chromel/Alumel ) 450 500 FIGURE 77. TGA of MD3P (unfixed) on 50/50 PET/cotton. 237 en E -t-> 5 200 250 300 100% Cellulose MD3P Fixed 1- 2.81% P 2- 2.05 3- 1.53 4- 0.90 Untreated 350 400 450 500 Temperature, C (Chromel/Alumel ) FIGURE 78. TGA of MD3P (fixed) on cotton. r 238 en E 50/50 PET/Cotton MD3P Fixed 1- 2.65% P 2- 2.00 3- 1.26 4- 0.73 Untreated 150 200 250 300 350 400 450 Temperature, C (Chromel/Alumel ) 500 FIGURE 79. TGA of MD3P (fixed) on 50/50 PET/cotton. 239 that the efficiency of MD3P as a flame retardant for polyester/cotton blends could probably be enhanced by mixing it with other retardants which have a greater propensity for condensed phase activity. Experi- ments to combine MD3P with Pyrovatex CP have been inconclusive. Attempts at fixation from baihs containing 5-20% MD3P and 10-30% Pyrovatex CP along with Aerotex^>73 resin and a phosphoric acid catalyst gave varia- tions in add-on of only 4% on both 50/50 blend and 100% cellulose sam- ples. At the highest add-on levels, the fabric burned readily on the 45 pin frame. Another anomally of MD3P is the fact that add-ons are achieved that are approximately twice those obtained when other N-meth- ylol propionamide compounds, such as Pyrovatex^-^CP, are applied using the pad-dry-cure technique. To elucidate the nature of the high add-ons of MD3P, 100% cellulose and 50/50 blend fabric with various levels of MD3P were extracted by Soxhlet extraction with methanol. The level of compound remaining in the fabric after extraction was comparable to levels obtained with Pyrovatexv_yCP. Nuclear magnetic resonance spectra of the residue after evaporation of the methanol indicated the majority of the extracted material was 3-di phenyl phosphinyl propionamide. As previously noted, the TGA of MD3P, Figure 76, shows a weight loss cor- responding to the temperature used in the cure (150°C). The precusor to MD3P, which is present after formaldehyde loss, is not-water soluble and would not be removed by the afterwash. The Pyrovatex^CP precusor is water soluble, and the afterwash leaves only the chemically bound flame retardant. However, these data are consistent with earlier conclusions that fixation reduces volatility to improve condensed phase and vapor phase reactions. Condensed phase reactions decrease the flammable gases to require less phosphorus in the vapor phase. Alternately, efficient vapor phase action decreases the heat return to the fabric to enhance the opportunity of condensed phase reaction, i.e. less volatilization of the flame retardant. These results indicate that there is some validity to the proposal that a single organophosphorus retardant should be capable of effective action on 50/50 blends, but it will be very difficult to predict a priori the types of structures which will produce the needed balance 240 a vapor phase and condensed phase activity. However, this work with a single flame retardant substantiates the assumption that condensed and vapor phase retardants work together to give better results on a 50/50 blend than a flame retardant system which displays only one mode of retardance. 241 3. Other Organophosphorus Retardants In an attempt to find classes of organophosphorus compounds which might possess the proper balance of volatility and reactivity on 50/50 blends, several structural categories were evaluated. The methyl analog of MD3P, N-methylol-3-(dimethylphosphinyl )propionamide (MDMP) was prepared and studied on 100% cotton and 50/50 polyester/cotton blend fabrics. Using a double pad-dry-cure-rinse process with a bath containing 40% MD3? along with Aerotex^23 and H 3 P0 4 , the flame retardant levels were 15% on 100% cotton and 7.5% on a 50/50 blend fab- ric. These samples were easily ignited and burned completely when sub- jected to a 45° angle burn test. Other phosphorus structures prepared and screened included triaza- phosphaadamantane; 1 ,3,2-diazaphosphorinane PH ; and 2- oxo-2-methoxy-l,3,2-diazaphosphorinane. The triazaphosphaadamantane was obtained from SRRC as a crystalline solid and was applied to both cotton and 50/50 blend fabrics from a chloroform solution. Add-on levels of 2.6% and 2.1%P were achieved on the cotton and the blend respectively. o Both fabrics burned readily in the 45 angle pin frame and exhibited 01 's of approximately 23.5. Based on these results, work was suspended on this compound. A more promising series of phosphorus compounds has resulted from studies of cyclotriphosphazene derivatives. Hexakis(2,4,6-tribromo- phenoxy)cyclotriphosphazene, a previously unknown compound, was pre- pared in good yield by reaction of the tribromophenoxide with the par- ent hexachloride. The product was characterized by IR and NMR spetro- scopy. It showed good thermal stability with only a small degree of de- gradation apparent after heating at 250°C for 48 hours. Its 70.5% Br and 4.5% P content should make it an attractive candidate for either topical treatment of fabrics or melt blending during polyester manu- facture. Unfortunately, no satisfactory fabric treatment was achieved and thus its potential as a flame retardant remains unmeasured. As an alternate route to a bromine-containing phosphazene, the attention was focused on the bromoform adduct of hexaallyloxyphospha- 242 zene previously prepared at SRRC. The hexaallylcyclotriphosphazene was prepared in 80% yield by the reaction of sodium allylate with the tri- chloride. This material was characterized by infra-red, DTA, and NMR. The NMR of the crude reaction mixture shows approximately 95% conversion of the hexachloro compound to the hexaallyloxy. It exists as a therm- ally unstable, pale yellow oil which resisted attempts at crystalliza- tion. An emulsion of the ally! compound was prepared with an almost equal amount of bromoform in conjunction with PVA, sodium bicarbonate and potassium peroxydi sulfate and applied to fabric in a one dip, one nip process, dried for 10 minutes at 110 and then cured for 60 minutes at 140 . Oxygen index values were 24.5 for an 11.8% add-on to 65/35 blend fabric, and 25.0 for a 14.0% add-on for 50/50 fabric. In the 45° angle test the samples burned the entire length but showed considerable ignition resistance. At present it is unclear whether these disappoint- ing oxygen index values and flammability characteristics are due to the low weight add-ons used or whether they indicate generally poor effi- ciency of the retardant. The picture becomes considerably more promising when the hexaally- loxy phosphazene-bromoform adduct is used in conjunction with THPC. Fabrics of 65/35 blend treated to a 14.9% add-on with this material gave an oxygen index of 25.7, whereas 50/50 blend fabrics treated to a 17% add-on gave an 01 value of 27.2. In the 45 angle test, all of the samples showed considerable ignition resistance and the 50/50 blend treated to a 17% add-on failed to burn the entire length. All of the fabrics exhibited a tendency to discolor on heating above 140 C. Be- cause of this and because this system did not appear to offer any sig- nificant advantages in terms of flame retardant efficiency, it was not pursued further. A similar set of experiments was carried out using trial lylphos- phate in place of the allylphosphazene. Triallylphosphate was allowed to react with bromoform and telomerized in a fashion similar to that used with the phosphazene. The resulting white, stable emulsion was pad- ded onto blend fabrics and evaluated by 01. The adduct appeared to have adequate flame retardant efficiency but the resulting fabrics were stiff 243 and slightly yellow; thus this system was not examined further. Other phosphazene systems were also examined. A sample of 6,6- diaminotetrachlorocycletriphosphazene, a white powder, was isolated from the reaction of a concentrated methylene chloride solution of the hexachlorocyclotriphosphazene and aqueous ammonia. Upon standing, this powder evolved heat and HC1 and acquired the characteristics of a high- ly cross-linking between cyclotriphosphazene residues and is not the same polymer that arises from ammonolysis of the linear polydichloro- phosphazene. Such a polymer containing the six membered phosphorus nitrogen ring has not been previously reported. Several methods of fabric treatment were investigated in attempts to fix aminophosphazene derivatives on to blend fabric. In the first attempt, the hexchloride was padded on to the fabric and then treated with ammonia. These samples were then washed and dried at 100 for three minutes in an attempt to remove untreated starting material which tends to sublime at this temperature. Only low levels of add-on were achieved and are indicative that the reactions to produce a cross- linked polymeric ammonolysis did not take place. Treatment of fabrics with the linear dichlorophosphazene polymer with subsequent reaction with ammonia in organic solvent yielded add-ons of 5% and 10% after washing and drying. Assuming disubstitution by ammonia, this would correspond to approximately 1.6% to 3.2%P on the fabric. These samples exhibited 01 values ranging from 25.2 to 27.0. They also exhibited ig- nition resistance from 5 to 7 seconds with an add-on level of 5.4%. Comparison of these 01 values with those obtained using other phosphorus derivatives indicate that the efficiency of the chloro and amino deriv- atives are nearly twice those of other analogs. This would seem to in- dicate some promise for these materials as topical flame retardants if complete chlorine removal could be accomplished. The diaminotetrachloro derivatives have also been evaluated in com- bination with THPC. A sample of 65/35 polyester/cotton blend fabric treated at a level of 14% add-on exhibited an 01 value of 27.7 and these fabrics did not burn on the 45° angle tester. The hand of the fabrics was firm but they were not discolored by heating at 140 C for three 244 minutes. On this basis, this system would seen to have some potential for further development if a satisfactory technique could be developed for its application to fabric. In order to test the possibility of forming cross-linked methylo- 1 amine polymers, three aminocyclophosphazenes were treated with an ace- tone solution of THPC. Tetrachlorodiaminocyclotriphosphazene and THPC gave a crystalline adduct whose melting point and IR spectrum differed from those of the starting materials. Reaction of THPC with hexallyl- aminocyclotriphosphazene and hexamethylaminocyclotriphosphazene was indicated by the formation of syrupy, viscous products which resisted attempts at purification. Column chromatography of these products gave only a small amount of elutable material indicating formation of telo- meric or polymeric products. Samples of 65/35 polyester/cotton were treated with tetrachlorodiaminocyclotriphosphazene/THPC adduct and line dried to give fabrics with approximately 14% add-on. The efficiency of this flame retardant was quite high and the fabrics were not discolored by heating at 140° for three minutes. 245 4. Mixtures and Precondensates of Phosphorus Retardants Because it appeared that success in obtaining flame retardant ef- fectiveness with phosphorus systems on blends would require more than simple condensed phase activity by the phosphorus on the cellulosic com- ponent, it was decided to more clearly define the limits of this mode of interaction. A series of fabrics was prepared containing various levels of diammonium phosphate as a model non-volatile phosphorus compound at levels of phosphorus ranging from 0.7 to 3.1%. The heat release values for these were then compared with similar fabrics containing both diam- monium phosphate and a retardant such as Mobil's AntiblazeOl9 which is a potential source of volatile phosphorus. These results are tabulated in Table LII and Figure 80, Heat release values for the diammonium phosphate treated blend show the same general dependency on percent phosphorus as that of phosphoric acid when the values are adjusted to allow for differences in the fabric weight. Tj^e blend treated with a mixture of diammonium phosphate and Antiblaze^l9 gave a slightly high- er heat than that treated with diammonium phosphate alone. However, of perhaps greater significance, the two series of diammonium phosphate/ An ti blaze ^19 treated blends (one with constant diammonium phosphate content and the other with constant Antiblaze^-^19 content) fell on the same curve when plotted versustotal phosphorus content suggesting that phosphorus from the Antiblaze^is essentially as effective as that from the diammonium phosphate in the combined treatment. This is sur- prising since data obtained on 50/50 blends treated with Antiblaze^ 19 alone indicated that the latter compound is less effective than diammo- nium phosphate. This again indicates that the combined action is of a different type- rather than either condensed phase or vapor phase active systems alone. In an attempt to get more data on this type of effect several series of blend fabrics treated with Pyrovatex^CP in an unfixed form, resin fixed Pyrovatex^3762 and resin fixed MCC-lOd-^were studied by the oxygen bomb technique. The plots of Y/(l-XL^yersus log percent (r) (r) phosphorus for the Pyrovatex^CP and Pyrovatex w 3762 are shown in 246 CQ O O at to LU >- _i o o LO o lo LU C£ I— Cn CQ <: © _i CO O to I— _l to LU C£ O CQ i— i cc LU o CO E cn 03 LO O LO cn VO r— CO CO O O CM r— CO CO o CM 1 — CM CO «3" CM *3" "d" CO CM ^ CM CM CM CM CM CM CM CM E o <_> CD C_> CD * CQ I * < Q CT> Cn ! CO • • • • ■"* CM <3- CM + 1 IT) cn CM CO • • • • CX> r— CM CM lo LO LO LO LO O r— cn • • • • ld r^ co co i— i— i— CM cn LO CM LO ■ — LO CM lo • • • • o o r— ' CM — ' — '• cn LO CO LO * • • • o CM LO CM < Q T3 CD X cn LO «d" r-^ cn cn cn 00 • • • • o o o o — — — CM r_ o r-^ ■ • • • «3- *3- «3- CO © c c o u CO S- o -c a. to o 4-> 03 U CO •r- CO CD -C c CD s- «3 co C1J CO 247 3000 CT> u I 2000 Q< O^D O Fixed DAP(UP) D Fixed A-B(UP) 1.0 % Phosphorus 2.0 FIGURE 80. Heat release of DAP/AntiblazeQ9 treated 50/50 Pet/cotton blend fabrics. 248 ®, comparison with unfixed Pyrovatex^CP on 100% cotton in Figure 81 . As expected, the CP treatment was shown to be less effective on the blend than the 100% cellulosic. It is interesting to note, however, that the effectiveness of both treatments is about the same on the blend fabric. Previous results have shown that when the CP is fixed with a mel amine resin, its effectiveness decreases markedly on the blend. Experiments have also been carried out in which 65/35 and 50/50 polyester/cotton blends were treated with di ammonium phosphate. Two trends in the bomb data were observed. First, it was noted that all of the treatments were shown to be more efficient on the 50/50 blend than on the 65/35. Second, the order of effectiveness for the treatments was di ammonium phosphate>CP>3762on the same blend fabric. These results can be inter- preted as measuring the efficiency of the flame retardant interaction in the condensed phase presumably in the cellulosic part of the blend. A series of blend fabrics treated with diammonium phosphate has also been examined by oxygen index. Blends of cotton and PET treated with diammonium phosphate showed a strong dependence of oxygen index on the phosphorus content. This is a measure of the effect of the con- densed phase interaction with the cellulose component and in agreement with the static oxygen bomb results. -Jhe oxygen bomb calorimetric results from the Monsanto 100/200/ 300^finish on both 65/35 and 50/50 polyester/cotton blend fabrics are given in Table LIII. Also included in Table LI 1 1 are the oxygen index and evaluation by the subjective 45 angle burning data for this finish. Since some question has been raised concerning the validity of using the function Y(l-X) from static oxygen bomb studies on phosphoramide derivatives such as MC-100^! the data from these studies were treated in terms of the ratio of A H« (heat released as measured by static oxy- gen bomb calorimetry) to the total heat of combustion of the sample fcH°)^. This treatment is shown in Figure 82. These data can be in- terpreted in the same way as plots of Y/(l-X), and indicate that the high char forming condensed phase efficiency of the treatment on 100% cotton is not duplicated in a blend environment. Subjective evalua- tion in a 45 angle burning test is consistent with this. These re- 249 ° m o < ■ 1 ' o ■ o ■M O <_3 CL CJ CL O - ■ O o c o G- (_> O Lf) O IT) C o Q. o lo c o CM CO ©©© X X X QJ 0) O) +J 5-> +-> rc3 ITJ na > > > O O o s_ S_ S- >1 >> 5, CL CL CL <* ■ 1 1 1 1 1 o . m 10 O •r— &- o -Q sf «3 "O as o ■*■> re co CL) S- +J CM iO 1^. o CO CM X a* +j s© > o s- >> o o o c o o co o o o o o o LO o o o o CO o o CM 00 (X-U/A o I— I Li. 250 TABLE LI 1 1 FLAMMABILITY OF BLEND FABRICS WITH MCC-1 00/200/300 ® %P (1) 01 45°Burn -AH, 1.30 0.90 0.60 0.40 1.40 1.00 0.70 0.50 (3 ( 24.8 Difficult to ignite 3595 (2 * 23.6 Burns 3799 (2 ' 22.8 Burns 3852 (2 1 22.0 Burns 3929 (31 1 26.1 Difficult to ignite 3139 (3] 1 24.8 Difficult to ignite 3262 (3; 1 23.6 Burns 3345 (3! 1 22.5 Burns 3454 AH 2 /(AH°) C'F .74 .77 .78 .79 .66 .68 .70 .73 1 Estimated on total add-on of finish. 'On 65/35 polyester/cotton blend On 50/50 polyester/cotton blend, 251 o o CO o o o o O O Lf) o o ■sfr o o ro to o •I— s- 03 XJ aj +J 5- © O O CO o o CM O o I o o o a. CD O *+- o c o •1 — ■p o c M- .13 CI as o o < o c_> < CM a: < CM Z3Z <3 CM CO a: CD 252 suits are not entirely unexpected since it is quite likely that a large percentage of the retardant is deposited as a coating on the surface of the polyester, and therefore not available for condensed phase inter- action with the cellulose. Some evidence in support of this interpreta- tion is found in the change of slope observed in the plot of AhL/(AH°) F as one goes from 100% cotton to a 50/50 blend, and from a 50/50 blend to a 65/35 blend. On the other hand, diammonium phosphate, which is probably treated primarily in the cellulose component of the blend with little surface deposition on the polyester, shows no such change of slope as shown in Figure 83. Any phosphorus from the MCC-lOO^on the surface of the polyester would, of course, be available for other modes of inhibition, but its concentration would probably be too low at these add-ons to be truly effective for flame retardation. Self-extinguish- ing characteristics were observed for 50/50 blend fabrics over treated with MCC-100/200/300VVto add-ons of 35% and phosphorus levels of =2.5%. These results seemed to confirm previous indications that it should be possible to devise effective flame retardants for polyester/cotton blends based on phosphorus alone. Two fundamental criteria for such flame retardants were defined: (1) The FR agent should be relatively active in the condensed phase on the cotton portion of the blend; (2) Some significant amount of the phosphorus must volatilize in a suitable temperature range to act as a vapor phase inhibitor for the fuel produced by the decomposing polyester. These two constraints require that the flame retardant possess an appropriate combination of high thermal stability, low volatility be- low 300°C and higher volatility above 300°C. Also, since rather large amounts of phosphorus are usually required to pass tests such as that incorporated in FF 3-71, the system should have a phosphorus content approaching 20%. Although these criteria were later revised, as discussed at the end of this section, they served as effective guides for much of the developmental work during the project. There were several ways of po- tentially satisfying these requirements, the simplest of which was to 253 o +J ■>»_ o 0l C_) o &-S IT) o "•s,. o O <— LO c c o o 01 au +J • 4-> r «. C C o o § E E o a. a: C\J «3- o CO ',0 a> 4-> c o o S- o -C Q. CO O •r™ O C_3 < CM < o o o O O o un O o o o CO 4-> n3 O U < C\J a: < CO LU a: C3 254 mix known retardants of previously determined activity. Attention was, therefore .focused on mixtures of Pyrovatex^CP with THPC/urea, or Pyrovatex^3762 with THPC/urea. The reasoning behind the first formu- lation was based on the assumption that although THPC/urea acts pri- marily in the condensed phase in 100% cotton fabric, it could be made to volatilize from a blend where some of the retardant should not be in contact with cellulose. The tendency for volatilization should also be increased by using the THPC in combination with a flame retardant which is inherently more reactive toward the cellulose than is the THPC. Pre- vious research has shown that one flame retardant meeting this require- ment was Pyrovatex^CP. A slightly different line of thought was invoked in developing the Pyrovatex^3762/THPC/urea formulation. In this case the system was designed to take advantage of the inherent reactivity of the THPC in the condensed phase while maximizing its tendency for partial volatili- zation by forming the phosphonium oligomer which should not be capable of penetrating the cellulose and thus should reside on the surface of the fabric. This system would have the advantage of having all of the phosphorus in the same chemical form. Previous work has shown that (r) Pyrovatex v -^3762 is a more effective retardant for polyester-containing blends than the THPC itself but that it is compatible with these sys- tems. The validity of this approach is demonstrated by the data in Table LIV. These data were particularly encouraging since all of the results were obtained on 50/50 mixtures of the two retardants with no effort toward optimizing the formulations. On the basis of these re- sults it seemed that optimized formulations might be capable of impart- ing significant flame resistance to cotton/polyester blends at reason- able levels of retardant add-on. In all cases, the fabrics appeared to have fair strength and reasonable, but firm, hand. _ As clearly shown by the data in Table LIV, the Pyrovatex^-^3762/ THPC/urea combination is the most effective of the two formulations. This was not unexpected based on previous results with Pyrovatex^3762 alone on blend fabrics. In order to eliminate the problems associated with applying the thermal phosphonium salt precondensate while retain- 255 TABLE LIV BURN DATA FROM FABRICS TREATED WITH PHOSPHORUS COMBINATIONS X (1) 45°Burn FF-3 (2) 01 BO I THPC/UREA .234 DNB (3) (.558) (4) 2.7 28.8 21.2 .169 I.R. (5) (.426) BEL 27.0 21.0 .124 I.R. (.346) BEL 24.5 19.5 THPC/P.yrovatexv!9cP/UREA 266 DNB (.731) 2.7 28.8 21.5 206 DNB (.634) 2.9 27.0 21.2 154 I.R. (.451) BEL THPC/Pyrovate> 26.2 ^3762/UREA 21.8 216 DNB (.489) 2.9 28.0 22.0 160 I.R. (.293) 2.7 26.2 20.8 117 I.R. (.237) BEL 24.8 19.0 (1) Weight percent of FR finish (2) Average of three samples, char length in (3) Does not burn (4) Char weight percent. For samples which DNB charring done with external flame (5) Ignition resistant (>3 seconds with kitchen match) 256 ing its flame retardant effectiveness, attempts were made to prepare chemical precondensates of tetrakishydroxymethyl phosphonium salts. Initially it was decided to attempt to synthesize oligomeric struc- tures having two distinct types of linkages between the phosphorus moieties. This decision was based on the assumption that the chemical stability of these bonds would determine the overall efficiency of the oligomer in a degradation scheme such as that shown in reaction (6). FLAME INHIBITION P(oligomer) ^ — > P(oxides) + P(fragments) {6} * CELL-OH CHAR P&fwhii The simple condensation of MC-lOO^which had been shown by IR, NMR and mass spectra at the Southern Regional Research Center in 1971 to be primarily (CH-HCKPO] and an aqueous solution of methylamine with THPC were tried in an attempt to form the oligomeric structures shown in reactions {7} and {8}, respectively. +~,- \ / P(CH OH)/C1" + P(0)(NCH,), ^ P-CH o -N-CH 9 -P(0) {7} 3 3 '/ 2 | 2 v H CH 3 P(CH 2 0H) 4 + CL" + H 2 NCH 3 > P-CH 2 ~N-CH 2 -P {8} CH 3 In the case of the MCC-100, 100 ml. of the 70% aqueous solution of the phosphoramide was slowly added to 100 ml. of the 85% THPC solution 257 with stirring. The reaction mixture warmed to = 70 C on mixing and was then allowed to cool to room temperature. Under these conditions (ini- tial pH - 2.0 and the presence of free formaldehyde) in addition to the direct reaction between THPC and the phosphoramide, N-methylolation of the phosphoramide and the acid catalyzed hydrolysis of the phosphoramide to H-^PO, and methylamine would also accur. The methylamine produced might then react as in reaction {8}. To form the methylamine precondensate 100 ml. of a 40% aqueous solution of the amine was slowly added with stirring to 100 ml. of the 85% THPC solution. A third condensation reaction involved neutralization of the 100 ml. of the 80% THPC solution cold with 20% NaOH to a pH = 8.0. To this was added 100 ml. of the 70% MCC-lOQ^solution and 20 ml. of a 10% MgCl 2 ~ citric acid catalyst solution. The reaction mixture was heated with stirring to 70 C for 1 hour and then allowed to cool to room temperature. Although chemical changes in the reaction mixtures had obviously 1 31 13 occurred in all three cases H n.m.r., P n.m.r. and C n.m.r. were used to establish more clearly the nature of these interactions. The proton n.m.r. of MCC-100^in aqueous solution with DSS as the reference shows CH^, 2.58 ppm, doublet of doublets, J pNrH = 12.5H , J npuM = 5.5H and NH, 3.75 ppm, 6-1 line multiplet, JuruM = 5.5H z , J PNH " 9 - 0H z- fo When equal volumes of MCC-100^and 80% THPC were mixed as above, the NMR spectrum taken 10 minutes after mixing showed a pair of doublets at 2.55 ppm. and a weaker doublet at 2.85 ppm., J = 0.5 Hz. A very broad doublet was present at 4.3 ppm. When taken at 70°C thirty minutes after mixing, the original - CH 3 doublet for MCC-100®at 2.58 ppm. was much weaker while the broadened doublet at 2.8 ppm. increased as did the broad singlet in the 4.1-4.3 ppm. region. The normal doublet for THPC near 4.77 ppm. was replaced by a broad singlet. These spectral changes definitely indicate that reaction has occurred at both phosphorus _ (r) moieties but do not confirm the direct reaction between the MCC-lOO^ and THPC. Similar changes showing reaction at both components were observed 258 in the H spectra for the condensation reactions between THPC and methylamine and THPOH and MCC-100^. 31 In Figures 84 and 85 the P n.m.r. of 40% aqueous solutions of THPC and THPC (THPOH) neutralized to a pH - 8.0 with NaOH are shown 31 respectively. The P n.m.r.'s of_the condensation reactions products between THPC-CH 3 NH 2> THPC-MCC-100^ , and THP0H-MCC-100® are presented in Figures 86 through 88. The chemical shifts relative to phosphoric acid and the integrals for these spectra are tabulated in Table LV. As expected the THPC spectrum shows a singlet at -26.2 ppm. while THPC neutralized to a pH - 8.0 shows the residual THPC signal and the THP (trihydroxymethyl phosphine) multiplet in the region of +23.0 to +31.4 ppm. At this pH THPOH still contains approximately 40% of the phosphonium salt. The THP multiplet arises from the complex equilibrium which exists between the various phosphine moieties present in this solution. The reaction between THPC and CH 3 NH« was found to go completely to substituted phosphine derivatives as shown in Figure 86. There is no evidence of any phosphonium salt even though the initial mole ratio of THPC: CH~NH 9 was approximately 2:1. As in the case of the THPOH spectra the P of the precondensate shows the presence of several different phosphine adducts and the mixture is probably an equilibrium between several of them. The chemical shifts of the adduct are all upfield from those of THP alone even though the pH of the solution is similar. This upfield shift is as would be expected for the formation of P-CH 2 N-CH 2 -P bonding. In going from THP (+31.0 ppm.) to P{CH 2 N(CH 2 CH 3 ) 2 > 3 the chemi- cal shift was found to increase to +65.5, 31 The P n.m.r.'s of the THPC and THPOH oligomers are somewhat more complicated and the interpretations of the chemical shift data less cer- tain. However, tentative assignments can be made on the basis of the results observed above and some related chemical shift data from the literature. In the case of the THPC-MCC-100^ oligomer spectrum (Figure 87) it can be seen that in addition to phosphoric acid formed by the hydrolysis of the phosphoramide and the residual THPC resonance there are at least 259 o O + o cni O Q- i— Q. + -t-> cO «3 O O Q. ^ I— 4- O o E CM =3 i S- +J o a> Q- i/> Q. o ^~ 1 oo • *d- 00 UJ q: r3 CD 260 Q. Q. 1X3 C_> DC o D- O o CM E 3 S- 4-> O a; O- w Q_ o ,— 1 CO • IS) 00 LU OH ZD CJ3 261 o + o on + o CO + K Q- o r— +J + 4- •r- O) E 1 O CD CD .c CJ O 3Z CO re O m of THPC- o 13 CNJ S- +-> O CD Q- CO Q_ o CO 1 CO 00 q: CD 262 o + o + o C\J + o + o o. E o 01 © o 1 o 00 O O o o I o Q_ 3 S- +■> U CD Q. CO CO 00 on ZD C£5 263 o o o _ o E 03 O E O) cu E o en - o o O 1 — o o o 1 i :n o a. n: t— if- o o CNJ 1 E Z3 S- +-> 264 © o o o o X o a. Q. © O 2 CD O 1 a. ^_- o o O CO an CO i— o m <_J » A ar ■•-> CM 00 in co o X 1 "+- •«* ro ro CM CM o A o •^ i 1 ■ 1 I *s- a. .c 1 (/> r— l- co + hj o LO •— 2 I/) -J s s X o a. o a. a. P & ^- CM CO « tC lA O O O r^Or— f)r— ^-r— in CMOOOOOOO I/) 10 CM t— CM CO l/> «/» CM O O CM in CO CM CO n n n oo CO CM CM i— o o co in r-^ »— in CO CO CO ^ ^ + + + + + in co m O «» O O CM CO O *» o o o o o cm cm o i— in oo m cm *n cr» i— co CM CO CO CO *» «s- I + + + + + m *r co o o ^- oo co CO f— ^ o o d o cm o i— *»• **■ io co r^ t— «— CM CM CM CO CO I + + + + A o CM c a o a. o c o •I— o SSI o O a. co t— ■ a* i. a • u— >4- r ~ (II H CM M 0© O O C\ o o X p* fMB z 1 1 cr u o x o X k 5 o a. o a. o CO 'II CM I at ac a. f— CM 01 o X 01 2 5 265 3 other types of phosphorus. The phosphorus in the region +30.0 to +50.0 ppm. can be assigned as due to the presence of phosphine phospho- rus containing a P-CFL-N bond similar to that formed in the THPC-CHoNhL oligomer. Although the resonance at -23.1 was originally assigned as unreacted phosphoramide it now appears that this is due to the substi- tuted phosphoramide reaction product in the oligomer. This assignment was initially based on the expected insensitivity of the phosphoramide chemical shift with substitution as shown by the data in Table LVI and also the integrated areas. From the integrated areas it can be shown that at least some of the reacted^ohosphoramide structure must have chemical shifts similar to MCC-lOONw'. Confirmation of reaction at the phosphoramide nitrogens 13 and this chemical shift assignment was obtained from C n.m.r. data provided by Dr. Dan Scharf of Hooker Chemicals and Plastics Corporation. 13 The C n.m.r-^pectra of a reaction mixture with a 1:3 mole ratio of THPC: MCX-100VV showed three types of carbon. At this ratio of THPC: MCC-10(KVall of the THPC was consumed and the spectra consisted of a broad singlet at 37.6 ppm. s a doublet centered at 41.8 ppm. and a sharp singlet at 27.4 ppm. relative to dioxane. The doublet at 41.8 ppm. was assigned to the -rP-CHp- carbon from the phosphine, the broad sing- let to the reacted I -N-CHo nitrogen methyl carbon and the sharp singlet to the untreated -N-CH- nitrogen methyl carbon. These data thus con- firm the direct reaction between the THPC and the MCC-lOQV-^to form an oligomer with the general structure P-CH 9 N(CH Q )P(0) and the assign- 31 ment of the P chemical shift at -23.1. 31 The remaining resonances in the P spectrum are assigned to part- ially reacted phosphonium derivatives. These downfield shifts are in accord with literature data (72) for substituted THPS^ salts and would be expected to be present in this reaction mixture due to the slight excess of available THPC. The P spectrum of the THP0H-MCC-1 00^01 igomer is shown in Figure 88 to be as wouJd be anticipated based on the assignments given for the THPC-MCC-100woligomer. All of the major spectral shifts can be accounted for by analogy to the previous case. In addition the ratio 266 TABLE LVI 31 P Chemical Shifts of Some Selected Phosphoramides^ ' Phosphoramide P(NH 2 ) 3 P(NHCH 3 ) 3 P{N(CH 3 ) 2 > 3 P{N(CH 2 CH 3 ) 2 ) 3 OP CH. *N — ' CH 3 — Chemical Shift, ppm, -22.0 -24.0 -23.0 -23.5 -23.1 0) Reference 72. 267 of the relative concentration of substituted phosphonium salts to the substituted phosphine is much less than in the case of the THPC-MCC- 10CK-/ reaction. This reflects the much lower concentration of THPC present due to the neutralization to THPOH. It is also important to note that the less acidic conditions result in much less hydrolysis of 31 the phosphoramide to phosphoric acid. The P spectra of these three precondensation reactions are compared to each other in Figure 89 . With the formation of structurally different THPC precondensates firmly established by n.m.r., calorimetric evaluation of their flame retardant efficiency and mode of action were undertaken. Samples were sent to the University of Maryland for isoperibol calorimetry while the static oxygen bomb calorimetry on the treated fabric and isoperibol chars were performed at Clemson University, The results of these cal- orimetric investigations are presented in Table LVII through LXIII. The untreated ETIP 50/50 P/C blend was calculated to have a -AHL heat release of 4064 cal/gm based on a -(AH°) F of 4690 cal/grn, a -(AH ) r of 7540 cal/gm and a residue yield of 8.3%. Unfortunately, the interpretation of the calorimetric data from the phosphorus based non-phosphate or phosphonate FR-systems is not as straight forward as in the case of the halogen or the mixed halogen- model phosphorus compounds presented elsewhere in this report. These difficulties arise because it appears that the systems, as represented by the data in Tables LVII through LXIII, are neither simple cellulose active condensed phase nor classical vapor phase retardants. Their mode of inhibition seems to involve a complex interaction in the con- densed phase with both the cellulose and polyester. In addition, it may also involve as yet an undefined vapor phase activity. Another difficulty involved in interpreting this data is that no classical vapor phase only, phosphorus compounds were found during the course of this study. Thus, no model compound data on this type of flame inhib- ition for phosphorus is available. Because of these considerations any discussions of these data must, at the present time, be limited to mechanistic generalizations and comparisons of their relative effici- encies. 268 31 P n.m.r . P(CH ? OH) A CT/H ? NCH q THPOH/P(0)(NHCH 3 ) 3 fl)(HUCHj &* - -30 -20 -10 +10 +20 +30 +40 Chemical Shift ppm FIGURE 89. Comparison of 31 P spectra of phosphonium salt oligomers 269 IT a < 2 u E a: o o :r < CO 0£ CQ a; a. c ■er cr> oo o i-~. «* ro i— «* cm «» ro r— cri oo ro ro co cm cm cm vo co >— ro vo en in «* r-. ro cn ID IT) CO CM i— CO co co ro ro co cm CM CM O CO CO ^- CM O CM f— O Cfl vo vo v© m o s r>. O CO ro in #— O r- o*> r- cm O r«. in co vo vo m in in lO CO o s on «d- CM o «* VO VO >j0 LT> uo '* <■ ** <• «* UD cn !-» vo o ur> CT» 00 CM 00 00 CM «■ ^ «* CO CM CM ■o- "» **■ tj- ** xf CO 1-^ ^- i— r>» vo r-» vo r- Ol N VO CM i— i— i— i— ffi v£> LD 00 ro O VO o 00 O LD O LD Cn O cm cr> *r r— r— CM o «s- m r-~ cn o o CM CM CM i— <— io ro O un VO r— O .— r- Cn CM VO i— i— t— CM o Q. ZC CM ^-*. rmm 00 un 00 ) — «* «* ■jz ro •3- vo CTl r— ro s- Q cn S- C3 cn , — 00 r- , — 00 r— cn r— r-~ CM VO >, <* ro ro O o CO -O CM CM CM CM CM *-> a> cn cn . *— i_ 3 ai I ••-> V- c O > T3 S 1 •^cc ■o ITJ -C •■-> ^ => jO o oo -o a> 0> •r* JZ r— 4-> a. O.VI- o 1- i—i nj C LlJ u 1 u_ in s r— IfJ r- 10 m vo M3 lO i£> cm CO VO r- CM •— in in l© ^- CM CO CO CO «* *»■ <— o cm «* CM r— cm «* i— r-. UD UD l£> in s vo i^ oo vo oo o co oo oo cm «* ^ ^- «* a: □3 VO CO CO **• CM o oo oo CM r— i— CO o CTi CM r-» r-. CM co r-~ «*■ O CM p-» I s * cm •— .— ■— 00 m r- CM vo CM 1--. CM cm in o CM CM vo +1 +1 +1 +i T — <- o in co in in CM VO CM in o in co r- i— CVJ cm rs. t— >— oo co r- r— CM co CM CM o in V_> rx. ,— m CO r- CM o ,— **• CM VO r^ £ m CM >> ^£> oo «s- s* CO o © O O 3: < I E en o o < © » CO N N in in 10 oo ■ — i— cr> r-» in vo in in CM r— I— l£> co in r^ cm in in in in *»- «» *r «* rs Cft rx » •— i— co in «* CM r— o OO CO oo oo r^ cm in cm .— oo oo <* O CTl 00 m in in in in ce CO 0O 0O 00 CM Q oo * ffi o> •-< i— .— CTi in O CM CM r— i— CM cm (V) cm en CM oo CM CM m in io in CO CTl i — l£> i— i— CM CM «* oo **- ■* Ol N ^ o S- >> Q- CM oo T3 ro CL in 00 CM >* ID i— in r— i— f— CM CT. oo *r lO i — CM CM CM X o o OO ** l£> i- ■ * >> CO OO in 00 Q. > — < — l — CM ID px oo in r-» cr. \£> •— oo 0O CM O CO 10 00 CTl CTl in in «* in oo o in oo 7— cr. i~^ > o >> * a. ■o o c -t-> (O O O- Q- <_> © x cu •M > O a. © > O i. o 1/1 ■o o .— oo ■— 5 7 o ■o c Q. CM e o in i "8 -a +j fO 0) "8 •^ 5 T3 -O T3 AS 18 3 X TO o O LO 3 on u •r— S_ cy E < I u uo 00 eg s . CTl 5J- Co * G a 5 8 ^■■s o u X < t o o < E U vo o oo r— m o "* CO r— CO CO *C0 s co as s 00 r«» CM C-J r* » r^ LO LO LO Q o '-J (J r«~ o cm cr> cm i— -rsi r»« VO LO LO «g- CM r>. r»» in «a- cm f- r-» !■>•» © o CSJ en c i— CM CXl O CO O O I/) i— O 00 LD CM CM p— i— e— C-J 00 r— «* p^ CTl CO r-^ o CO o o r*. in CM LO co CM +i + 1 + 1 +1 +1 LO o CNJ LO LO CTl LO o LO lO m 10 © 10 0) lO CO cr> LO ^o oc I. r> o CM LO r™ uri Vt ■ — . u C7> CM CM CM CO n*> o. ©' co 00 IX) LO LO x "- — CVJ co «o- O o F™ -C o V o •r- z c 1 LL. o a. •£> lO LO CM LO s: t© CT> M 00 *r fr« P ?=- CM § CM U CO un LO o • • X >— 'VC od fsl CXI CM I— (■». 00 r~» © O o CM m a O o CM to « at r_ s_ o -Q c ■ o _l 4-> JC 5- c fO a> at -d lO c cu • o QC >> K . ^ 'aj s- *o (O > cx cu c •■- t. o ■»-> c 3 pi CD &» a. "\ -C o ♦J o; i/i «/ ■*— at o s C J- o S- r~ CO ai +j 1 L. JC «1 «_} 3 •M >> C_i (J 3 r— X ^^ o > 00 +-> * • 1_ lO o T3 •M O a. eo 3: ■o" T5 h- A3 c c <*. Q. 3 (O o •a C o >» C l/> •F" JQ •O 0) *» i— l- 5 "S Q. 0) •^ ■•-> JC ^ ^-> o ■•-> f— ' a. £ Q. •1- (/) «*. CV ITJ r— CM 273 © CO o CO CO CM CO *t in O r»» CM o ro e^ on ,-; > o s « '£> vo to in i£> CM «■ «* CO CM f— *> r*. r^ r>. r^ ** "=3- *t «» +l 00 CT> in cr\ VO If) m CM s '£1 vo © o o CM I CJ u o a. o 3; i— en CM i- I O i— I— CM I— f— r— o a U0 cm *r o o co CO 00 CM CM Lf> Lf) (_ o «4- J3 "8 a. a. O i— © CTv CM 00 CM O CO CO CO co CU c © o o o o 1 g o o CM o s- ' 1 £ Q. C_> 1 l/r>^i <-• 31 • r— V^V O 1— 5 • *c3 o O o r~ O 0) a. cm 1 4-> s- ■n i <_> 3 h- o <_) o o CJ s: a. 3: >. "S* • • i— 1- T3 JC <_> T3 +J ■O 4-> Q. it- c (O .i- IE O & 2 2 <*- C 1/1 o a> <♦- o •f— t. o •■- IC in O 3 -t-J ai •r- o O *J >> i~ ♦ J ^^ •i- fM J3 * >> +-> -r- c J- C (O i— "S o T3 S- ■f* ■»-) ■o -o O) +-> ^~ •r- •r- (O i— 3 a. c •— a. £S Q. o o CM o a: o u < 1 o u E en x < -u L- (T3 Ll 0; •(-> -3 1/1 ai oc a u QC +j OJ &« © O r— *S" CT> O CO CM LO LO LO CO CO on co 00 co o «* o CM LO LO CM LO LO LO ** 00 xf CO LO LO LO '.O LO LO co rs CO LO LO J — IS rs is rs "tf •3- ^ «* CO CO en LO LO <_> O- X CO 0-1 CM Is co o rs CM 28 1— co LD 1— LD LO OO LO O r— 1— CM O.' O SI S_ CD t- rs CTl CM LO C71 •* ■f— O O « — 1 — r— O CM X 00 00 © 00 is LO en o> is o «*■ lo 1— co co ■ — o cr> CO CO CO CO CM is O rs VO CTv LO r— cr> ct» LO Lf) LO Oi LO 0. rs CT> is 1— LO LO CO LO LO LO 1 — I — LO O CT» •3" CM CT> is O rs rs lo vo co Xt" **• «s- •* «* CM 00 rs LO *r CM co co IS o co LT) LO LO LO LO CM <_> O s: ■— rs O LO O'i CJ CM 00 LO IS OI o_ CM CM CM CM CM 3: 1— S. S_ CD E O LO LO o\ O rs O) ■1— 1 — r— CM 00 co O CM X CM CM LO LO LO *5- LO CO CM 0/ L. a. >» X) c • *— 1 — ■r* a. +J a. CJ < n d; 4. _ c O m +-■ s. ■ a C/l c. © cu E O cu O CI -0 O c CM r— t O cu CM 4- ?■ X Q. Z CM c_> X s^ • • >> O i_ Q. ■ L_> ? X +J a. r— c X ■0 as 1— «4- 10 a. © +-> O (0 1/1 •e- O CU O >, +-> CM S_ °i- -Q 03 1 +-> S- CJ c ia ■s O s- t/> 21 ■r- »l— T3 +J a> [— •r- -C •1— r— Q. .— -4-> C O a. O •>- cn at < CO 2 1—* r- CM 275 <-) •I— -O o CM uj ^a: —i a. z CD CO «t o I i— in o o o o a* CM < O U < E cn o o < < i a: a: CO ID If) O cm r— r^ o in «a- cm i— CO CO CO CO «* ^ O CM r>» in in r-» •3- cm .— co CO CO CO CM co oo 1X5 «* CM in KO ID 5 N i- 00 W *S" CM <— ID «»■ *J" *3" CO id in vo vo CO CM t— p-» .— cn o i— r^ m r^ vn *s- *r **■ ** m «d" oo in Sm in in m m in ** «* *r «* cr> <«t 1*^ CO r>» r— 1 — ■ — + i +i + 1 in *s- a> s cn cn 00 oo CO lO *s- F— r-^ m o oo 00 00 o oo o co f~^ r~ CO r~- m r^ ID CM CM a. o m CM <_> o a. O >* CM o CO in l-~ 00 «*■ CM ID o CO CM in oo £) CO co cn r-^. CM o f-v. l£> vr> vo id ^x ** m in in m in u. ^ o o o m o-. *-> •*-> c c o o a. a. ."3 3 a y o o o «3" r— 00 o o VO 00 CO in o CM m • • • CM 1 r-. cn ^ CO 1 «3- CO ^— CO 00 C_> ^— ^ ^— CM CM O CM CM CM o (J £ z: -•**. "-s^ s» (. V a; E o § o '£> CO , CO o 00 CM in r>* CD CD »r— o O o i — f— •^ o o i— i— • — O o CM CM :r X z z CO CO a: ^— 3: ^- *— * i o o 1 o £3, o *J CM f-» o ^— <_> +J r-». o p«. CO a. c • • • ex c • * 3: o f*. oo CM in X o p^ o CM ID t— u r— r— t— <_> f— r— >— 276 If AH, for all of the data in Tables LVII through LXIII were plotted versus the % P in the finish, with the possible exception of three treatments, all of the data points would fall within the bound- aries of the generalized response curve given in Figure 90 . However when the data are plotted vs AH-,/(AH )_ the distinction between these finishes becomes less pronounced (Figure 91). The three finishes which may he exceptions to this behavior are (NH.) ? HPCL and the two THPC-MCC- 100^-^oligomer finishes given in Table LXI. Two other generalized response curves for these finishes may be constructed from these data and are presented in Figures 92 and 93. . In Figure 92 AH ? is plotted as a function of % P and no exceptions to this response are observed for any of the data. In Figure 93 , R is plotted as a function of the % P in the finish and in this case the (NH-KHPCL data once again appears to deviate from the response ob- served for the other systems. The deviation of the (NH.LHPO- data in Figures 90 and 93 might be rationalized in terms of its activity in the condensed phase of cellu- lose. Since it can only interact with the cellulose substrate and is nonvolatile it would be expected to be the least efficient treatment in reducing AH-. from a blend. On the other hand, it should be the most efficient in terms of residue formation, especially at low P add-ons. However, this is not found to be the case. Even at only 1% P it is a significantly less efficient char former than the other treatments. This is not due to some of the (NH z ,) ? HP0 4 depositing on the surface of the polyester and hence not available to interact with the cellulose substrate. If this were the situation, then the oligomeric finishes should be even less efficient char formers since they are likely to have even a larger percentage of their phosphorus is on the surface of the polyester. There is the possibility that some of the other finishes increase the residue forming tendencies of the polyester but this can- not be ascertained with any certainty from the data. This would ex- plain the decreased efficiency of (NH 4 ) 2 HP0 4 which is the same as the other finishes as shown in Figure 93 . In line with this conclusion it can be seen by a careful comparison of the data in Tables LVII through 277 o o en <3J <_> a) I Q- Q. c Z I DC CD CD o 4 • ■ a3 Q t/1 C c o 4-> +J o o o o in i — •— i > « — • o CO -o S- rC +-> a> s- ai E >— — » i o I — •— I a. -o fl3 l ! — • — ' o CM I • 1 i * i a; -Q C o 'a; s- T3 at N 03 CD O o O O O o o o O O o o o eg 00 *»• o <£> «\J CO C\J CM . 1 .«* 1 — r— ■=> CD i— i 278 < ' Urea '200/300 ©© o o o o •^ 1— I— O 1 I Q_ O C_> 3Z <_> O CM 21 21 *"-> 1 1 re Q. Q_ z: zc :n ^_- H- i— ♦ B • < ♦ • • ♦ B OB a • Q O O ox ©a — • 1 1 1 t> a* 1 l Lav- O . • CO «* -o c O) r— .o c o +J 4-> O O \ 1— LU O. O O LO ro o LO +-> c «3 ■o s- 03 -!-> 0) S- Q- 0) ^ E ^5 ro ^— 4- o S- • o CM 4- Q. < r-» ^o IT) «a- CO CM o o o o J (oHV)/ L HV o o LU 279 1/1 c 8. I/) fM i. c 0) I • • -•— • I • 1 *— • • • ft — I •— • 1 • ft- 1 o o 00 o o o o o o O o O ^£> CSJ C*) ro o o oo CM o o «*■ CM o o o CM w6/ieo *2 HV -o c o 4-> 4-> O O O m o ID c 03 "O i. CO +-> (U S_ (V E fO O CO > CVi < c o o c 4- XJ C o Q. 1/1 a' Q£ «* o a. IN CM «* 'a j_ v. z ^ ■ c 0.' CD o O ir> o in o I • 1 +-> C o u •r- •I— c -•H ►— • 1 o CM t* • o ■o I— QJ ^ >» •r— to s- » — #- 1 o c -o containing blend series were observed to be significantly more ignition resistant than the corresponding add-ons for the Dacron^-^ blend. This indicates that the shape of tfcie response curve is in fact similar to that found with the H 3 P0 4 /900FViy series and shown by the dashed line in Figure 94 . In general, all of the oligomers studied so far exhibit a high degree of stability once formed and may be stored for several weeks without any apprecible changes in chemical or physical properties occurring. The product obtained in the oligomerization reaction is also dependent on the ratio of retardants employed, although the 282 TABLE LXIV CALORIMETRIC RESULTS FROM THE ETIP 50/50 BLEND FABRIC TREATED WITH A THPS^/MCC-IOO^/UREA PRECONDENSATE (1 * COMPARED TO THPC/UREA % Finish % P % R AH, Oligomer 8.9 1.46 19.1 ± 0.3 1936 ± 36 12.3 1.95 20.3 ± 0.3 1847 ± 116 16.7 2.40 25.4 ± 0.3 1826 ± 28 23.4 3.60 27.9 ± 0.7 1648 ± 59 THPC/Urea 6.7 0.82 17.1 2008 12.4 1.90 22.3 1868 21.5 3.30 31.7 1745 (^Solids ratio of THPS® MCC-100®in the precondensation reaction = 5:1. 283 o CM to -a c: a* .Q a* 4-> 03 a> CM CO O a. 1 C_) M ex a: o I— O +■> c- CD +-> E O o u> a. O r- o o o oo o o 8 o eg UI6/IPD • ^HV- 8 8 •r- a: CU 284 chemical structure does not vary. Only the THPC-CH 3 NH 2 oligomer was found to have an inherent aesthetic property which may be difficult to overcome. The residual ChLNhL proved to be yery difficult to remove from the reaction mixutre, probably due to the various equlibria in- volved, and consequently considereablediscoloration was observed after the cure. _^ During the initial investigation of the THPC-MCC-1 00^-^0 1 igomer an attempt was made to evaluate this finish, using urea, in an applic- ation similar to that which would be encountered in commerical produc- tion. A series of laboratory experiments were carried out at SRRC duplicating the conditions which would be encountered in a pilot plant. During these examinations an unacceptable lack of durability of the finish was observed in the process wash following curing. This is demonstrated by the data in Table LXV. In order to overcome this problem, the urea was replaced by^a mel- amine derivative in the formulation. Use of Monsanto' s MCC-20Q^>'and 300 system produced a reasonably durable system as shown by the data in Table LXV I. This system also shows good fire retardant effective- ness in both the 45° angle and vertical tests. Because of the promise shown by this system a small scale pilot plant trial was carried out at SRRC. Three ten yard samples of fabric were treated at levels of 1.9, 2.8, and 3.9% phosphorus. The highest add-on passed the vertical burn test initially but failed after five laundering cycles. Elemental analysis of these fabrics indicated a major loss of phosphorus during laundering. Since the oligomer itself is stable to these con- ditions, and since the oliogmer is water soluble, it is presumed that the loss of flame retardancy results from cleavage of bonds between the oligomer and the mel amine resins. Two^other difficulties which were encountered in using the THPC/ MCC-100W/200/300VVfinish that were not encountered in the corre- sponding urea system were a considerably harsher hand and a significant loss of tear strength, ibwe'ver, it was found that inclusion of Dow-, Coming's Emulsion 1111^-Aand modifier and resin extender into the formulation resulted in a distinct improvement in both hand and tear 285 TABLE LXV :-ioo®/ Durability of MCC-IOQV^/THPC/UREA Finish on 50/50 Blend %H conc. rllafter cure X,co1d wash X^hot wash X^hot wash — jt — jt 3.9 0.330 .263 .255 .065 9.7 0.328 .246 .256 .063 (1 ) weight percent (2) one minute (3) twenty minutes 286 strength. To more completely define the nature of the durability problem en- countered with the THPC-MCC-lOO^/adduct, samples of 50/ 5Q_poly ester/ cotton fabric were treated with variations of the MCC-10oL^THPC finish and^scoured at the boil for four hours in a solution of 4g/l Orvus ABv^and 2.5 g/1 soda ash. The scour baths were then neutralized with HC1 to pH ^ 6.0 and the water stripped off using a rotary evaporator. The results of this scour are shown in Table LXVI. The NMR spectra of the extraction residues were run in D^O. The spectrum from the THPC-MCC-100/200/300vL'sample contained two broad bands at 0.5 - 1.5 ppm indicating the presence of alkyl groups. It also contained a broad group of peaks at 2.5 and 3.0 ppm and a sharp peak at 4.1 ppm. These latter^peaks are very similar to those observed in the NMR spectra of MCC-10(KVand THPC respectively Two samples treated with the same THPC-MCC-100^-f urea system but cured under different conditions were also examined. The residue from Sample II cured at 160° for 2.5 minutes exhibited broad multiplets at 1.22 - 3.08 ppm and 3.54 - 4.0 ppm. The residue from the sample cured at 150° for 2.0 minutes vislded a spectrum which resembled that from the THPC-MCC-100/200/30CKV system except for the absence of the broad bands at 0.5 - 1.5 ppm. These data would seem to indicate that at least in one case the lack of durability might be a function of poor fixation rather than hydrolytic instability of the oligomer. _^ In an attempt to improve the fixation of the THPC/MCC-100vi/finish, MgCl^ an( * MgCl^-citric acid were tried as catalysts. Using equal molar concentrations in the pad bath the NlgCl^-citric acid complex appears to be the more efficient catalyst. The add-ons for the MgClp-citric acid bath ranged from 25.1 to 15.1 weight percent while those from the un- complexed MgCl 2 went from 21.0 to 10.8 weight percent with the same dilutions. The performance of these materials in the vertical and 45° burns was similar although the MgCl^-citric acid fabrics performed silghtly better. These fabrics showed a small improvement in durability but this was not pursued because of the development of more effective approaches. 287 TABLE LXVI © Fini Durability of Oligomeric THPC/MCC-100^^ Finishes Initial Wt. % % Finish Sample Finish Reagent Loss ,© I THPC/MCC-100/200/300 v ^ / 21.9 40.6 II THPC/MCC-IOOV^/UREA^ 27.5 79.3 III THPC/MCC-IOOW/UREA^ 31.3 42.2 (1) Four hours at boil, 4g/l Orvus AB, 2.5 g/1 soda ash (2) 2.5 minutes at 160° (3) 2.0 minutes at 150° 288 In one of these preparation of the oligomer was modified. Instead of adding the MCC-lOOvSvdropwise to the THPC, the solutions were mixed rapidly producing a different pH for the condensation reaction and allowing less time fcr both condensation and hydrolysis. A solution of MCC-200vMind MCC-300Wwas^then added to the resulting oligomer solution along with a Sapamine^-^softener. After curing at 16°C for 3 minutes this technique produced a fabric which exhibited good dur- ability (ca. 3-4% weight loss) in a four hour soap/soda boil. However, the high resin content resulted in a finish with a very harsh hand and relatively poor tear strength. The effect of the atmosphere over the precondensate reaction vessel, the pH of the precondensate baths, and the efficiency of several diff- erent resins were also investigated. The pad baths, add-ons, FF-5 results and some durability data from these experiments are summarized in Tables LXVII through LXXVII. On the basis of these results it would seem that the controlling factor still remains the % P in the finish and in this respect the resin employed has little effect on the efficiency of the retardant. While the pH must be important in terms of bath stability and resin efficiency, in these limited results, the performance of the neutralized precondensate baths showed no significant improvement over a non-neutral- ized formulation. Similarily no significant effect was discerned from blanketing the reaction vessel with argon during the formation of the precondensate. A initiaU4nvestigation was also carried out witk precondensates of THPC and THPSvLVith trimethylphosphoramide (MCC-10Ck5/from Monsanto) and guanidine carbonate. Similar materials have been reported in a recent patent to Toyobo (U. S. 3,855, 349, December, 1974). The phos- phorus content of these precojadensates seem to be equivalent to the initial THPC/MCC-100/200/30(Njyadducts made at Clemson. Hence the efficiency of these systems is quite high as shown in Tables LXXXIII, LXXIV, and LXXV. However, the durability of the modified formulations also seems to be much better than that obtained in the eariler work. Typical pad bath formulations and flame retardant results obtained with 289 TABLE LXVII Pad Bath THPC/MCC-lOO^f Prepared Under Nnn-Oxidizinq Atmnsnhprp ) 100 ml of an 80% THPC solution 100 ml of a 70% MCC-1 0QvL) S ol uti on stir together under Argon for 45 minutes cool in an ice bath add a solujkion prepared by dissolving 45 g MCC-2QoQ9in 45 ml of H o '2 solution ( 15 g SapamineVVAPN rritonvUx-100 iu wt ine®/ 200 ml H 2 1 Dip, 1 Nip on lab pad Dry 2 min at 150°C Cure 4 min at 150°C Wash 3 min in warm 2% Na« C0 3 Oxidize 3 min in warm solution of 20 ml 35 wt % H 2 2 6 g sodium silicate 4 ml 40 wt % NaOH 10 ml 10 wt % Triton^X-100 H«0 to 1 liter of solution Rinse in cold water Dry 2 min at 150°C Results (on ETIP 50/50) Approximate Add-on FF-5 (initial) Bath (1) 22% 7.9 in. 290 TABLE LXVIII Pad Bath THPS^/MCC-100^( Prepared Under Non-Oxidizing Atmosphere ) 100 ml of a 70% THPS ^solution 100 ml of a 70% MCC-lQQv9sGlution stir together under Argon for 45 min cool spontaneously add a solution prepared by dissolving 22 = 5 g MCC-200^in 22.5 ml H 2 add /( ~ 35 ml of Aerotex^23 Special 15 g of SapamineV^APN ^ 20 ml of a 10 wt % solution of TritonVVx-ioo 200 ml of H 2 Dry, Cure, etc. See Table LXVII Results (on ETIP 50/50) Approximate Add-on FF-5 (initial) Bath (1) 20% fail 291 TABLE LXIX PAD BATH THPS&7MCC-1 OCtMoTRALIZED 100 ml of a 70% THPSW^jlution 100 ml of a 70% MCC-100v9solution stir together under Argon for 45 min cool spontaneously place in an ice bath neutralize solution to pH 7.2 with ^50 ml of a 20% NaOH solution add to the-,above solution a solution prepared by dissolving 22.5 g MCC-200Win 22.5 ml H 2 add ^-^ 35 ml of Aerotex^23 special -^ 20 ml of a 10 wt % solution of TritonV^X-lOO 200 ml H 2 Dry, cure, etc. See Table LXVII Results (on ETIP 50/50) Approximate Wt % Add-on FF5(initia1 ) 4 Hr Scour Loss Char Bath (1) 25% fail 8% 50% 292 TABLE LXX PAD BATH THPS^VMCC-IOO^NEUTRALIZED 100 ml of a 70% THPS^splution 100 ml of a 70% MCC-100v9solution stir together under Argon for 45 min cool spontaneously place in ice bath neutralize solution to pH 7.2 with ^50 ml of a 20% NaOH solution add to the above solution a solution prepared b^L dissolving 22.5 g MCC-200^in 22.5 ml H 2 then add^^ 35 ml Lyofix^CHN 15 g SapamineQvAPN 20 ml of a 10% Tritonv£/X-100 solution 200 ml H 2 Dry, cure, etc. See Table LXVII Results (on ETIP 50/50) Approximate Wt % Add-on FF5(initia1 ) 4 Hr Scour Loss Char Bath (1) 25% fail 9% 50% 293 TABLE LXXI PAD BATH THPOH/MCC-100^ place in ice bath 100 ml of a 70% THPS^solution add 50 ml H 2 neutralize to pH 7.2 by slowly adding ^25 ml of a 20% NaOH solution add _ 100 ml of a 70% MCC-100^solution 20 ml of a 20% solution of MgCl 2 citric acid heat to 85°C under Argon for 1 hr strip on rotovap to dryness to the above .oligomer, add a solution prepared by dissolving 22.5 g of MCC-200^in 22.5 ml H 2 add GD 35 ml Lyofix^lCHN 15 g Sapamine^APN ^ 20 ml of a 10% solution of Triton^X-100 add hLO to bring total bath volume to 300 ml Dry, cure, etc. See Table LXVII Results (on ETIP 50/50) Approximate Add-on FF^SC initial ) 4 Hr Scour Loss Char Bath (1) 45% fail 25% 25% Bath (1) + 32% 150 ml H 2 0=Bath (2) fail Bath (2) + 22% fail 150 ml H 2 0=Bath (3) Bath (3) + '15% fail 200 ml H 2 0=Bath (4) 294 TABLE LXXII PAD BATH THPC/MCC-1 00/200/300^ slowly add to 100 ml of an 80% THPC solution 100 ml of a 70% MCC-lOOVVsolution to the^bove solution, add 50 MCC-200v9dissolved in 200 ml H 2 then add 8.7 g MCC-300® Dry : Group I fabrics dry at 85°C for 10 min Group II fabrics dry at 165°C for 2 min Cure at 165°C for 2.5 min Wash 3 min in warm 2% Na 2 C0~ Oxidize 3 min in warm solution of 20 ml 35 wt % H 2 2 6 g sodium silicate 4 ml 40 wt % NaOH ^ 10 ml 10 wt % Triton09x-100 hLO to 1 liter of solution Reoxidize 3 min. in 3.5% hL0 2 Rinse Group I fabrics in cold water Rinse Group II fabrics for 20 min in hot water Dry for 2 min at 165°C Results (on ETIP 50/50) Approximate Add-on FF5( initial ) 4 Hr Scour Loss Char GI Bath (1) 43% 4.7 in 1 Bath (1) + 28% 200 ml H 2 0=Bath (2) GII Bath (1) 41% Bath (1) + 28% 200 ml H 2 0=Bath (2) 295 \0J 5.9 in 8% 40% 3.2 in 14% 40% 4.3 in 10% 50% TABLE LXXIII PAD BATH THPC/MCC-IOO^/GUANIDINE CARBONATE Dissolve 3 g Guanidine carbonate in 72 ml HLO add 128 g an 80% solution of THPC 31.3 ml of a 70% solution of MCC-IOOVV reflux the above solution for 1 hr cool to 200 g of the above oligomer, add 20 g Urea and 1 g solid NaOH dissolved in 80 ml H 2 Dry, cure, etc. See Table LXXII. Group II conditions. Results (on ETIP 50/50) Approximate Add-on FF5(initial) FF5( After Scour) Bath (1) 37% pass pass Bath (1) + 22% h fail fail 100 ml H 2 0=Bath (2) 4 Hr Sc :our Lc iss 4. 5% 2. 5% 296 TABLE LXXIV PAD BATH THPC/MCC-IOO^/GUANIDINE CARBONATE Dissolve 3 g Guanidine carbonate in 72 ml H^O add 128 g of an 80% solution of THPC -^ 31.2 ml of a 70% solution of MCC-100^ reflux the above solution for 1 hr cool to 200 g of the above oligomer, add 10 g Urea ^ 14 g MCC-200^and 1 g NaOH dissolved in 80 ml H 2 Dry, cure, etc. See Table LXXII. Group II Conditions. Results (on ETIP 50/50) 4 Hr Approximate Add-on FF5(initial ) FF5 After Scour Scour Loss Bath (1) 31% pass pass 3% Bath (1) + 22% h fail fail 2% 100 ml H 2 0=Bath (2) 297 TABLE LXXV PAD BATH THPS^/MCC-IQO^/GUANIDINE CARBONATE Dissolve 3 g Guanidine carbonate in 54 ml H ? add 146.3 g of a 70% solution of THPS*o^~ 31.3 ml of a 70% solution of MCC-100^ reflux the above solution for 1 hr cool to 200 g of the above oligomer, add 20 g Urea 1 g NaOH dissolved in 80 ml H 2 Dry, cure, etc. See Table LXXII. Group II Conditions. Results (on ETIP 50/50) Approximate Add-on FF5(initia1 ) 4 Hr Scour Loss Char Bath (1) 27% pass 5% 40% Bath (1) + 19% fail 4% 25% 100 ml H 2 0=Bath (2) Bath (2) + 14% 3% 20% 100 ml H 2 0=Bath (3) Bath (3) + 10% 100 ml H 2 298 TABLE LXXVI THPS^/MCC-100®AerotexW23 ON ETIP 50/50 POLYESTER/COTTON Preparation of precondensat e Mix 146.3 g THPSvV(70% solution) 31.3 ml 70% MCCJOO R 11.7 g AerotexvS/23 special 54 g H 2 Reflux for 1 hour Bath : Dissolve 20 g Urea 1 g NaOH in 80 ml H 2 Add 200 g of the above precondensate Dry : 2 mi n at 165°C Cure : 4 min at 165°C Afterwash : 3 min with agitation in a warm 2% Na^COp solution Oxidize : 3 min with agitation in a warm solution of (per liter of soln) 20 ml 35% H 2 2 6 g sodium silicate 4 ml 40% NaOH 10 ml 10% Triton®X-100 Dry: 2 min at 150 C Test Results Add-on FF5 % Loss 4 Hr Scour % Char 27 3/4 (3.9 in) 10 30 18 1/3 (3.7 in) 7 20 299 TABLE LXXVII THPS^/MCC-100/20(/M)N ETIP 50/50 POLYESTER/COTTON Preparation of precondensate Dissolve 7 g MCC-200 in 70 m Add 146.3 g "MPS 1 in 70 ml H o ® 2 ,® 16.5 ml 70% MCC-100 Reflux for 1 hour Bath : 200 g of the above 50% precondensate Add 20 g Urea and 1 g NaOH Dissolved in 80 ml f-LO Dry, cure, etc. See Table LXXVI Test Results Add-on FF5 33% 4/4 (2.9 in) 25% 1/4 (3.7 in) 300 TABLE LXXVIII PAD BATH THPS^/CARBAMATE/MCC-200^ Dissolve ^^ 7 g MCC-200vVin 49 ml H 2 add ^^ 146.3 g of a 70% solution of THPS^ 49.5 ml of a 50% solution of Protorez^/CHD reflux the above solution for 4 hr cool to 200 g jaf the above oligomer, add 48 g MCC-200^dissolved in 60 g H 2 1 Dip, 1 Nip on lab. pad. Dry 2 min at 165°C Cure 4 min at 165 C Afterwash : post-oxidize as in Table LXXVI Results (on ETIP 50/50) Add-on FF5 Bath (1) 38% 3.31, 3.40, 4.80 ^ Bath (1) + 100 ml H 2 0=Bath (2) 28% 2.76, BEL, 6.77 ^ * 'After 50 home launderings ( 2 ^F5 Initial 301 these oligomers are given in Table LXXVI and LXXVII. These results show that the precondensate can be used to impart self-extinguishing pro- perties to 50/50 blend fabrics but that high add-ons will be required to achieve successful performances in FF-5. For this reason, attempts were made to find additional components which could be added to the precondensate formulations to improve their retardant efficiency and the inherent aesthetics of the fabrics. One approach to increasing the efficiency of a precondensate finish is to decrease the fuel content of the resin used in fixation. In practice this can be accomplished by the use of an ammonia rather than a resin cure. An attempt was made to ammonia cure several of the pre- condensates listed in Tables LXVII through LXXVIII using a wery crude ammoniating chamber at Clemson. In general a high phosphorus to nitro- gen ratio in the precondensate was necessary to obtain an effective ammonia cure. The nature of the nitrogen containing components of the precondensate did not appear to be \/ery important in terms of curing efficiency. Those precondensates which could be successfully ammonia cured required about 20% less of an add-on to pass an initial FF-5 on the ETIP 50/50 PET/Cotton blend fabric. These wery preliminary results indicated that ammonia curing could be a very effective means of applying these phosphorus containing precondensates. Another approach which was tried involved using a carbamate in the precondensation reaction. It was felt that the formation of a more linear oligomer would result in a considerably improvement in hand over the highly cross-linked three-dimensional oligomers which had been made. The oligomer reaction mixture, pad bath formulation; FF-5 and durability data for this precondensate is presented in Table LXXVIII. The vertical flame test performance of this precondensate was found to be comparable to that of the other oligomers of this type as shown by the data in Table LXXVII. In addition the anticipated improvement in hand and tear strength was subjectively observed. There are two important conclusions which can be drawn concerning oligomeric phosphorium structures based on the data presently available. The most important being that they have great potential for flame 302 retarding polyester/cotton blend fabrics but this potential has not yet been fully realized. Secondly, despite the fact that their mode of flame retardant action is fairly independent of the oligomer chemical structure, there are significant differences in both fabris aesthetics and phosphorus content which are dependent on the precondensate struc- ture. 303 FLAME RETARDANT SYSTEMS BASED ON BROMINE ALONE Unlike the phosphorus systems, relatively few treatments have been designed in which bromine is the only flame retardant species. To date the only finish of this type which has achieved any measure of success in textile applications is that based on decabromoidiphenylene oxide in conjunction with an antimony oxide synergist (FR P-44v_yfrom White Chemical Company). Although basically an efficient flame retardant, this system suffers from problems due to the flammability and aesthetic properties of the acrylic binder, a lack of durable press properties, the shade changes which occur when used in dark colors and certain processing problems such as build-up on the pad rolls. A series of investigations was, therefore, carried out in an attempt to character- ize the flame retardant action of P-44v-/and design modifications which might be able to circumvent some of its problems while preserving its high molar efficiency. 1. Calorimetric Evaluation of FR P-44 ® ® Three series of samples treated with P-44Wwere investigated calorimetrically. These included a 100% cotton and 50/50 and 65/35 polyester/cotton blends each treated at five levels of add-on ranging from 2.5 to 10.50% Br. The isoperibol results are tabulated in Table LXXIX. All these samples exhibited a substantial amount of after-glow. Residue yield (%R) indicated that the retardant action was predominently in the vapor phase. The heat-release values from the isoperibol calorimeter were plotted vs_ Br% content as shown in Figure 95. In all three series (including 100% cotton), the "thres- hold" was reached at approximately 8 to 10% Br. The most striking observation was probably the retardant effect exhibited by this retard- ant on 100% cotton. Its effectiveness on 100% cotton was essentially the same as on the two blends. This can be seen more clearly when the net heat reduction of each treated sample is plotted vs %Br as shown in Figure 96 . The net heat reduction is given by the difference 304 e W TJ i C 0) c— r^ 31 CQ <3 C o ••-> ••-> o o i. I. ai N» •»-> r— i/> n: O) o >> o ^— a. X < ■s I 4J r— X ia X a* u - 1 +j „ UJ f 00 ^_* n3 < c£ h- X) o r— □e >> c ,
&« «+- o Ifl ■»-» IO a S- 0) a. o i/> 1— 1 c o +-> *-> o a> o Q. 5^ E O - in in «3- oo «* oo oo in cm fn O CO O 00 CO 00 in i— oo "3- in O CO CM VD O CO CM <7> *J" CM i — U0 ID 00 O CM CM **■ i— co 00 id co in co vo ^- in in cri «* co oo «* «*• m O in CO CM CM CM CM CM r— CO CO +1 in ^J- O *J- *t 1 CM LO co co in o CO CM io «o io in m « ic N in io <■ cvi CM CO O i— 00 "nT L. O ID O CM CM i— •»-> C CM CO CO «t VD IN O o i- 00 O r— CTl CM C r- rs rsin n ia ia i— • t— t. c c i — co in in id u l/l £1 c 3 J3 (/I S- S e 5 4-> !- -e— »4- JD -f— to -C c X o 0) ♦-> tl <3J r— !_ CL «3 E > (Q (/> a> r— u ^— C o u ^1000 » 3: <3 500 - O 100% Cotton A 50/50 PET/Ctn □ 65/35 PET/Ctn Decabromodi phenyl Oxide (White Chem. ) % Bromine FIGURE 96. Net heat reduction from decabromodi phenyl oxide treated PET/cotton blends. 307 between the heat release of the treated samples and that of the control (AH 1 - AH i 'control^' Fi 9 ure 96 snows tnat tne net neat reductions of all treated samples are a single function of %Br content, indicating that the effectiveness on cotton is essentially the same as that on polyester. The decomposition temperature of decabromodi phenyl oxide is 310°C, which is wery close to that of cotton, assuming that there is no solid-phase action by the retardant (Br) on cotton. The absence of solid-phase action on cotton may be the key to the effectiveness of this retardant. However, this lack of condensed phase activity could not be verified by TGA as shown by the data in Table LXXX. Unfortunately, it has been found that it is difficult to draw generalized conclusions on the basis of specific data such as these. This is demonstrated by the results obtained from a second set of fabrics treated on a different occasion than those above and using a different lot of FR P-44v!9. Bromine contents of all the P-44® treated fabrics were analyzed using an x-ray technique and the data compiled in Table LXXXI. Finish contents calculated from wet pick-ups are also included in Table LXXXI. Decabromodi phenyl oxide (DBDPO) contents were calculated from the analyzed bromine content for each sample and are shown in Column 3 of Table LXXXI. The DBDPO content in the finish (% DBDPO/% Finish) is shown in Column 4. These results indicate that the DBDPO content in the P-44^->'finish can vary with the treatment as well as with the type of fabric used. In the first treatment, DBDPO contents in the finish are consistently about 66% for all fabrics treated, whereas, in the second treatment, DBDPO contents in the finish are 50% for 100% cotton and 46% for the 50/50 blend. However, the DBDPO contents in the finish for a given fabric in a given treatment are consistent. Figure 97 shows a plot of analyzed bromine contents vs % finish. Presumably the variation of DBDPO contents in the finish would affect the balance of bromine and antimony contents present; and this is generally regarded as the key to the efficiency of this type of retardant. An additional set of anomolous bromine analyses was encountered in experiments to determine the minimum levels of P-44^^appli cation 308 TABLE LXXX Thermal Analysis of FR P-44v^Finished Fabrics %Br (1) Ti f (2) WL (3) Ts (4) WLs (5) Tm i • Tm s (6) (7) 100% Cotton 3.6 5.5 6.0 6.6 9.1 10.4 312 304 307 313 291 299 .794 .735 .703 .704 .686 .660 337 .2.04 327 .172 327 .173 335 .191 315 .178 324 .204 50/50 Cotton/Polyester 3.1 328 5.8 312 6.1 318 8.4 9.1 303 306 .393 .360 .338 .366 .335 349 328 335 322 320 .422 .424 .394 .385 .419 345 424 329 395 337 403 320 385 323 378 156 ,189 229 239 148 65/35 Cotton/Polyester 2.6 4.7 5.6 7.4 8.2 (1 (2 (3 (4 (5 (6 (7 333 327 328 324 323 .250 .259 .234 .272 .242 352 344 345 342 338 539 352 433 538 346 427 493 346 427 529 343 424 515 341 416 .200 .165 .223 .171 .193 %Br estimated on the basis of weight % add-on of finish. Onset temperature of initial weight loss (cellulose fraction) % weight loss during initial weight loss Onset temperature of secondary weight loss (polyester fraction) % weight loss occurring during secondary weight loss Temperatures of derivative maxima for initial and secondary weight losses Residue at 500°C 309 TABLE LXXXI BROMINE CONTENTS OF FABRIC SAMPLES TREATED WITH P-44^1AeTARDANT <&, % Finish % Br Analyzed % DBDPO* calcd. % DBDPO/ % Finish Treatment #1 100% Cotton 4.32 2.01 6.58 3.68 7.91 3.04 7.14 4.26 10.87 6.23 12.48 7.19 50/50 PET/Cotton 3.71 1.89 6.90 4.00 7.34 4.15 10.08 5.91 10.91 6.27 65/35 PET/Cotton 3.08 1.67 6.73 3.73 8.90 5.22 9.88 5.68 Treatment #2 100% Cotton 5.80 6.35 9.54 12.24 13.67 14.57 2.64 2.70 4.18 4.27 4.94 6.54 2.41 4.42 3.65 5.11 7.48 8.63 2.27 4.80 4.98 7.08 7.52 2.00 4.48 6.26 6.82 3.17 3.24 5.02 5.13 5.93 7.85 0.56 0.67 0.46 0.72 0.69 0.69 0.61 0.70 0.68 0.70 0.69 0.65 0.65 0.69 0.69 Ave. 0.66 ± 0.07 0.55 0.51 0.53 0.42 0.43 0.54 Ave. 0.50 ± 0.06 50/50 PET/Cotton 5.98 9.96 13.20 16.41 18.04 2.38 3.79 4.97 6.34 6.88 Decabromodi phenyl Oxide 2.86 4.55 5.97 7.61 8.26 0.48 0.46 0.45 0.46 0.46 Ave. 0.46 ± 0.01 310 & e ai © ■3- c o ■•-> o o a. o o *e m o \ o o in «— in 10 o <] a o CM 4-> C a) c o o O) c •r- E O i- CQ en auiuioug % m a: CD 311 necessary to achieve satisfactory performance in FF 3-76 and on the MAFT^Samples of 50/50 polyester/cotton blend fabric were treated with P-44^at four different levels of application, and each laundered 20 and 50 times. 2. Application of FR P-44v9with Durable Press Resins Since the calorimetric evaluation of P-44v-^had shown a high level of efficiency, it seemed that attempts to improve its utility were in order. A series of studies were therefore undertaken in cooperation with Dr. Vladimir Mischutin of White Chemical Corporation. After some initial laboratory experimentation, the formulations given in Table LXXXII were developed using a glyoxal resin (Permafresh 113BV.J) either in the flame retardant bath or as a top finish. These formulations were evaluated in a pad-dry-cure operation using the pilot plant at SRRC. Three styles of polyester/cotton blends were used (two 50/50 blends and one 65/35 blend) along with one 50/50 polyester/rayon. The results are shown in Table LXXXIII. After exami- nation of the samples, the group at White Chemical Corporation con- cluded that treatments I, II, and III were over cured thus resulting in lowered durability due to decomposition of the resin. However, because of the apparently deleterious effect of the permanent press resin coupled with the poor hand of the fabrics and the build-up observed in the pad rolls during processing, this approach was abandoned. >-44Wwi 3. Application of FR P-44 v ^with a Bromine-containing Latex '-44® Since many of the problems encountered in the use of FR P- are related to the necessity for high add-ons, an effort was made to modify the application formulation to reduce the amount of chemical required. It was thought that part of the basis for needing so much flame retardant was the flammability of the acrylic binder used; thus it was reasoned that the use of a flame resistant acrylate would be beneficial . 312 TABLE LXXXII P-44®F0RMULATI0NS WITH PERMANENT PRESS TREATMENTS Treatment Component Wt. % in Pad Bath I Water -. Carbowax 400^ 40.0 2.0 P-50 softener ._ Rhoplfix TR-485® P-44® 3.0 10.0 45.0 Resin (1) II Water ^ Nyacol 1550®^ Carbowax 400^ 44.0 9.0 2.0 P-50 softener ^ Rhoplfix TR-485® P-44® 3.0 10.0 32.0 Resin (1) III Water Carbowax 400® 25.0 2.0 P-50 softener _^ Rhoplex TR-485® P-44® Permafresh 1138® 3.0 10.0 45.0 15.0 IV Water 27.0 Nyacol 1550V}*. 9.0 Carbowax 400^ 2.0 P-50 softener ^^ Rhoplfix TR-485® P_44vJy Permafresh 113B® 3.0 10.0 32.0 17.0 313 TABLE LXXXII (con't.) Treatment Component Wt. % in Pad Bath V Water Nyacol 1550v£) Carbowax 400W 39.0 9.0 2.0 P-50 softener ^ Rhoplex TR-485W P-53® 3.0 15.0 32.0 ^'.Top treated with 15% Permafresh 113Bs5/ 3% catalyst X-4^and 82% water. 314 TABLE LXXXIII FLAMMABILITY OF PERMANENT PRESS/P-44 V ^ / TREATMENTS M®- Average Char Length (FF 3-71) Treatment Blend Scoured Initial 50 washes I 50/50 PET/Cotton yes 3-1/8 i n fail no 3-1/4 i n fail 65/35 PET/Cotton yes 3-5/8 i n fail no 3-3/8 i n fail 50/50 PET/ Rayon yes 3-1/2 i n fail • no 3-3/4 i n fail II 50/50 PET/Cotton yes 3-3/4 i n fail no 3-1/2 i n fail 65/35 PET/Cotton yes 4-1/2 i n fail no 4-1/4 i n fail 50/50 PET/Rayon yes 3-3/8 i n 6-1/2" no 4 i n fail III 50/50 PET/Cotton yes 3-1/4 ■ n fail no 3-1/4 i n fail 65/35 PET/Cotton yes 3-3/4 i n fail no 4-1/4 i n fail 50/50 PET/Rayon yes 3-1/4 i n fail no 2-3/4 i n fail' IV 50/50 PET/Cotton yes 3-1/2 i n 4-5/8" no 3-1/2 i n 4" 65/35 PET/Cotton yes 4-1/2 i n fail no 3-5/8 • n fail 50/50 PET/Rayon yes 3-1/2 i n 5-1/8" no 3-5/8 i n 5-7/8" 315 TABLE LXXXIII (con't.) Treatment Blend Scoured Initial 50 washes V 50/50 PET/Cotton yes 3-1/2 in 5-3/4" no 4-1/2 in fail 65/35 PET/Cotton yes 3-3/4 in fail no 4-1/4 in 5-3/8" 50/50 PET/Rayon yes 3-3/8 in fail no 4-1/2 in fail 316 A series of experiments was therefore initiated to design and develop suitable flame retardant latexes. After screening a number of monomers, emulsion polymerizations of 2,3-dibromopropyl acylate (DBPA, Great Lakes Chemical AE-59) and 2,4,6-tribromophenoxyethyl acrylate (TBPOEA) were carried out. Of these, the DBPA polymer, P(DBPA), was adjudged the best commercial candidate and attention was focused on it. Through the cooperation of the Charles S. Tanner Division of Ciba-Geigy Corporation a commercially feasible preparation for a latex having the characteristics given in Table LHXIV was developed. This material has now been disignated as Dur-O-Cryv-^BL^l by Charles S. Tanner. Attempts to incorporate P(DBPA) into the P-44v~'formulation led to a marked in- crease in flame retardant efficiency but decreased durability to laun- dering. In order to improve the durability, a small amount of TMM was added to the formulation (10% on the weight of the latex). It is also found that this formulation could be made compatible with a glyoxal resin (Permafresh^AF) to impart both flame resistance and durable press characteristics at the same time. When applied in a normal pad-dry-cure process with a reactive silicone softener this system produces a durable flame retardant finish with good permanent press properties and hand. Strength losses are also minimized as shown by the data in Table LXXXV. The level of finish required to pass FF 3-71 is less than that previously recommended for P-44v_yeven though the finish now contains the durable press resin. An additional proc- essing advantage was also seen in the observation that there was no build-up of flame retardant on the pad rolls under laboratory appli- cation conditions. Many P-44\^)formulations do produce a build-up even under laboratory conditions. On the basis of these promising results, a mill trial was conducted at the Clearwater Finishing Plant of United Merchants & Manufacturers using the formulation in Table LXXXVI. The fabrics used included 150 yds of the standard 50/50 poplin with a normal preparation (scouring and bleaching), a 50 yd sample of the same fabric which had been re-soaped at Clearwater and a 50 yd sample which had been rescoured at SRRC using a desizing agent and an organic scouring agent. These modifications in fabric preparation were 317 TABLE LXXXIV CHARACTERIZATION OF DBPA EMULSION POLYMER ^ % Solids 40 Viscosity 50 cps Tg (by DSC) -10°C Ave. particle size 0.13 y % Br (theoretical) 21 ^ W-o-cryM^BL-1 Dur-o-cryl v -^BL-l from C. S. Tanner 318 -p a -p o o 0) fl •H g CO •H u O CU -p hi u ■3 H •H -P •H to bO •H u o cu s t-1 H cd •H -P •H H > « o i -=1- t— • • vo -d- CM -3- t- IA O CO VO VO 00 H H i — VO • vo CM H *-— * *«•— % <•••— X CO oo 00 PQ oo o o o o lf\ ^t ITS ^t CM CM CM OJ* OJ I oo i CM I CO £> H '-w-'' -=t Xi • d) -P vo H W) •H C « CO CU Ci ^ 00 d) +3 • EH CO t— CO £> H * — bO -P , t- a to vo •H 2 U CU X as ^ CU -P On Eh co t- VO IA X CM t— LA X OO VO a 0) to cu CM LTV o oo CM CO g 2 H H cu H ,o •p •H S 4 CM 00 319 TABLE LXXXVI PAD BATH FORMULATION FOR P-44®WITH P(DBPA) BASED ON 91% WET PICK-UP Reagent weight (lbs) volume (gal) FR p-44® (1) 135.2 8.44 Dur-o-cryl®BL-l ^ (40% solids) 44.6 4.31 Trycol®0P-407 ^ 2.6 0.28 Permafresh®LF 31.1 3.27 Catalyst X-4 ^/ 5.0 0.50 D-C 1111 Emulsion®^ 17<9 2 .14 D-C 182 A^ (5 ) 1.8 0.22 D-C 149®^ 1.8 0.22 TMM ^ 1.8 0.22 Acetic acid (glacial) 1.8 0.22 Water 144.7 17.34 * 'White Chemical Company (2) Chas. S. Tanner Company (3) v '40 mole ethoxylated octaphenol derivative, Emery Industries ( M0% Zn(N0 3 ) 2 from Sun Chemical Company (5) v 'Dow Corning Corporation 'Monsanto Chemical Company 320 36.94 gal included since work at SRRC had indicated a significajit improvement in both hand and fire resistance achieved with the THPCw- Urea - P(VBr/ VC1) finish after more extensive scouring. After this first 250 yds of fabric had been padded, approximately 5 gallons of water were added to the pad bath and an additional 150 yds of the standard fabric with the normal preparation were padded. During this operation no significant build-up of material was noted on the pad rolls, even though the range was stopped off twice with several minutes standing time in each instance. The pad was also found to clean easily with only running water after the run was completed. After padding, the fabric was passed through a 75 ft. tenter frame with temperatures of 233°F, 250°F and 230 F in the three boxes using a running speed of 50 yds per minute. The fabric was then cured for 3 minutes at 310°F in a loop oven, soaped in an open width soaper and dried. This produced a set of fabrics with good color, strength, hand and flame resistance and acceptable permanent press characteristics. These properties are demonstrated by the data in Table LXXXV. The only problem observed at this point was a dusting phenomenon which could be detected upon tearing the fabrics having the higher add-on levels. After removing samples of fabric for further testing, the remain- der of the material was printed using pigment colors and a flame re- tardant binder. The print was somewhat mealy, but it otherwise pro- cessed well. It v/as felt that this problem could be fairly easily overcome. More significant was the finding that the dusting phenomenon appartently represented an instability of the treatment toward mechani- cal agitation which was reflected in durability testing. Subjection of these fabrics to a 4 hour soap-soda boil produced no significant decrease in their flame resistance, indicating a high degree of hydro- lytic stability of the finish. However, when these fabrics were sub- jected to home laundry conditions, a significant loss of bromine occur- red (Table LXXXV) and the samples failed the vertical test. Experiments were carried out to determine the cause of this low durability. Changes in the resin, catalyst and cure conditions which appear to improve the 321 durability of this finish have been made in the laboratory and a second mill trial scheduled for early September, 1976. 4. Application of Other Bromine-containing Flame Retardants Although no other bromine systems have been developed to the same point as P-44w, there are several which have been screened to deter- mine their potential for further development. One of the most promising of these is an experimental material of unreported structure designated as Citex^BT-93 by Cities Service Company. This material contains 67% aromatic bromine, has good thermal and light stability, and low toxicity. It is supplied as a pale yellow powder (ca. 1 y particle size) and is reported by Cities Service to have a high flame retardant effici- ency. (R) Samples of Citex^BT-93 were dispersed in water using the formula- tion in Table LXXXVII and applied to 50/50 blend fabric using the pro- cess outlined in Table LXXXVIII. Attempts were also made to apply the Citex®BT-93 with the P(DBPA) but the pad bath was found to be unstable to the shear encountered during padding. Colloidal antimony pentoxide was also evaluated as a synergist for Citex^BT-93 using the formulation in Table LXXXIX. The results are tabulated in Table XC. All of the samples appeared to be covered with a fine yellow powder which could be removed mechanically although the finish was found to be durable to 50 home launderings. The level of finish required to pass FF 3-71 was comparable to that required of decabromodi phenyl ene oxide when both are used in combination with an antimony oxide. A comparison of the levels required to pass FF 3-71 without antimony was not possible since a sufficiently concentrated bath of neither compound could be made. The technique of combining a bromine/antimony oxide flame retardant with a bromine-containing latex binder appears to be applicable to many bromine systems. Preliminary results .indicate that pentabromodipheny- lene oxide (PBDPO, Cav-gard FRR 3-39 ^< Cavedon Chemjcal Company) can be used as the flame retardant either with the P-44^/P(DBPA) 322 00 CD 03 CO O Z5 CO r~ 4-> <1J +-> CD >> X o S- T3 N CO O CD £- 03 E -o O CO O CO I c o 03 2 o ^5 oo O 6^ o C\J CD o c 03 03 323 TABLE LXXXVIII APPLICATION OF CITEX^BT-93 TO 50/50 BLEND FABRIC Bath : TOO g 50% dispersion BT-93 1.68 g Trycol OP-407 Q9 (1) (40 mole octaphenol polyethylene oxide) 23.6 g Dur-o-cryl XWC '^ (acrylic emulsion)^ ' 1.2 g MCC-200 O (trimethylol melamine)( 3 ) 1 g Acetic acid 53 g H 2 18 g Permafresh LF V-' (unbuffered glyoxal resin)(4) 2.9 g 40% Zn (N0 3 ) 2 ^ Dry : 10 min. @ 85°C Cure: 4 min. @ 165°C Wash : 3 min. with agitation in hot 2% Na2C0^ Dry : 10 Weight % add-on : 22 Dry : 10 min. @ 85°C * 'Emery Industries (2) v 'Ciba-Geigy (3) v 'Monsanto Chemical Company (4) v 'Sun Chemical Company 324 TABLE LXXXIX ; I TEX® | CITEXV^BT-93 AND Sb 2 3 ON 50/50 BLEND FABRICS Bath : 100 g 50% dispersion BT-93 1.68 g Trycol OP-407 15/ I 1 '(40 mole octaphenol (polyethylene oxide) 23.6 g Rhoplex TR-485 (5/^' (self-cross-1 inking acrylic emulsion) 1.2 g MCC-200 v9 (trimethylol melamine)^ 1 g Acetic acid 55.5 g Nyacol A-1530 v!9 (30% colloid Sb^)^ 18 g Permafresh LF 'L^' (unbuffered gloxal resinr ' 2.9 g 40% Zn (N0 3 ) 2 ^ 20 g H 2 Dry: 10 min. @ 85°C Cure: 4 min. @1 55°C W ash : 3 min. with agitation in hot 2% Na-CO., Dry : 10 Weight % add-on : 26 Dry : 10 min. @ 85°C Note : Colloidal Sb ? 3 used is not compatible with Dur-o-cry XWC Emery Industries (2) v 'Rohm and Haas (3) v 'Monsanto Chemical Company (4) 'Nyacol Inc. (5) v 'Sun Chemical Company 325 Lf) co en en o c_> X CO OQ OQ o in o ir> o z o I— I t— a i— i UJ t— o s 3: ro O CM _Q I (— CO CO i i— CD CTi ro • »"■ O^ #« A r^ CM CO , o o • • • o CM VO m CM CM IT) CM Lf> Lf) O CTl c E o 1 u -o ^ "O sz r^ < ■*-> C U) o &« c r— a; r— 4-> h^ _i r— sz i •r- C7> CO i- M- •r— ITJ 326 formulation or with colloidal SbpCL and P(DBPA). The addition of a small amount of the PBDPO to the P-44^/P(DBPA) seems to be a viable way to increase the bromine content of the fabric without adversely affecting hand since PBDPO appears to thermosol into the fabric. When PBDPO is used alone it tends to give a somewhat greasy hand to the fabric but the presence of the P(DBPA) alleviates this problem. Similarily, a number of bromine-containing monomers studied by radiation grafting appear to be suitable candidates for forming bromine containing latex binders. Preliminary results show that a polymer formed from tribromophenoxyethyl methacrylate (TBPOEMA) is as efficient a flame retardant as P(DBPA). Unfortunately the P(TBPOEMA) was avail- able from Dow Chemical Company only as a powder at the time the ex- periments were performed. Since a good emulsion could not be made from this powder, this finish could not be adequately evaluated. As soon as the P(TBPOEMA) becomes available as a latex from emulsion polymer- ization, it should be subjected to more thorough characterization in these applications. 327 FLAME RETARDANT SYSTEMS BASED ON COMBINATIONS OF PHOSPHORUS AND BROMINE 1. Optimization of Phosphorus-bromine Formulations Once a sufficient understanding of the action of phosphorus and bromine had been established, a series of experiments were undertaken to determine on a more quantitative basis the effects and interactions of phosphorus and bromine retardants when they are used together. One system meeting this requirement is tris(dibromopropyl ) phosphate (TBPP). Because of durability considerations, this is not a practical candidate for commercial treatment of blends but it should be an instructive model for the study of this particular approach. For this reason, a series of blend fabrics was impregnated with varying levels of TBPP and their thermal properties studied. In an effort to determine the effect of the distribution of the flame retardant among the two fibers, a series of fabrics was padded with the retardant and heat treated to 150°C. Half of them were then washed in acetone to remove as much of the TBPP as possible from the cotton portion of the blend. Oxygen index values were found to reflect only the amount of retardant present in the sample and not the method by which level was achieved. This would indicate only minimal dependence of flammability on flame retardant distribution. Since it would also seem to be important to determine the rela- tive contribution of the two atomic species to the flame retardant action of TBPP, trial lylphosphate was prepared as a model containing only phosphorus. The triallylphosphate was applied to samples of 65/35 and 50/50 polyester/cotton blends at levels ranging from 10% to 30% add-on. This would give a range of phosphorus from approximately 1% to 4%. The 01 values of these fabrics were found to increase slowly with increasing phosphorus content up to about 2% phosphorus, beyond which the added phosphorus had essentially no effect on 01 values. In an effort to more completely characterize the bromine- phosphorus system, experiments were carried out utilizing variable 328 Br/P ratios. In this work diammonium phosphate (DAPO) was used as the model phosphorus compound and tris(dibromopropyl) phosphate (TBPP) as the model bromine compound. Bromine and phosphorus contents of the treated samples, together with heat release data are shown in Table XCI. Heat release values are also shown in Figure 98, plotted as a function of phosphorus content. It appears that data from all three treatments fall on the same curve, with slight variation in the values from that treatment in which the order of adding the retardants was reversed. This is probably due to a difference in TBPP content. As shown in Table XCI, bromine contents of these samples vary from 7 to 8%. Because of the variations in bromine content, -AH, of the control was not used in the calculation of the net heat reduction, AH, -(AH,) con trol . Instead, (AH-| ) con + ro j was calculated from the efficiency equation (Table XCI) for TBPP" for each sample as shown in Column 4 of Table XCI, and net heat reduction of each sample was obtained as the difference of the heat release, AH-j, and the calculated ( AH i) con t r or The plot of net heat reduction vs. %P is shown in Figure 99. Values from all three treatments are on the same curve, indicating that the order of treatment in this binary system does not seem to affect the efficiency of either retardant.- Values of heat reduction from 50/50 blend treated with DAP alone are included in Figure 99. There is a consistent difference of about 70 cal between the two curves, with the single treatment of DAP having the lower values. However, the standard experimental error estimated for AH, is ± 5%, or about 70 cal for these samples, the difference observed is not significant enough to warrant an argument of synergistic effects in DAP/TBPP systems. Since the order of treatment is apparently not important in determining the efficiency of this system, it should be possible to predict the effect of any particular formulation from the characteris- tics of each component alone. Experimental confirmation of this is shown in Figure 100. The efficiency curves produced by adding DAP to the fabric have the same shape regardless of whether addition is made to a fabric containing 0% or 7% Br. Similarly the curves produced by adding T23P are parallel starting at 0% and 1%P. Thus one can use 329 TABLE XCI CALORIMETRIC DATA OF 50/50 BLEND TREATED WITH DAP/TBPP (WITH FIXED TBPP) ah £il /au \Calc* cal AU ,. u , cal % Br % P ~ m ] [ gm ~ iAH l Control \ W AH T (AH 1 'control 'J FT 1876 2291 415 1641 2261 620 1487 2229 742 1462 2223 761 2255 2261 6 1398 2243 845 1356 2191 835 1277 2258 981 1228 2269 1041 l! ;t. Treatment 7. ,04 .16 7, ,36 .35 7 .70 .64 7, ,76 ,89 2nd. Treatment 7. ,36 7, ,55 .86 8 .11 1. .38 7. ,39 2, ,03 7. ,27 2 ,26 Reversed Treatment 7.56 0.54 1467 2242 775 7.40 0.95 1415 2257 842 7.07 1.27 1359 2288 929 6.85 1.73 1360 2309 949 ♦Calculated from the % Br and the regression equation for the efficiency of TBPP; AH, - (AH,) , = -7 + 93.5 {% Br) and -AH, of the untreated control II control i ] 0rder of treatment reversed; TBPP first, the DAP. 330 2000 ^ 1500 < i 1000 DAP/T23P Treated (with fixed T23P) 4 ,30 |_ \ ^ Ao • First Treatment O Second Treatment A Reversed Treatment 1.0 % Phosphorus 2.0 FIGURE 98. Heat release from DAP/T23P treated blend fabric with constant T23P content. 331 o o it c ■I— E C> S- c: > O to Cl. a. OO 00 - — - CNJ C\J Q.. h- I-- O I/O l/l m £ +-> Q- CO CM a. CD 332 I in m > <0 (— ( QC -* s- on ►- *< oc Lul S > ►— » I—* or o .* _j s_ •a: 2 s-s o o o o co <* o r- I- r- CM oo CM 00 +i +1 +1 + 1 f| +1 +1 t^. IX) LO CT> — LO LO ao LO r-. CO r^ LO cn LD LO lo i— ; o i— o i— +1 + 1 +1 +1 +1 + 1 r» r». O0 i— I CM CM CM CM CO co o px LO eg o LO o o 00 ("». LO LO CO CM 3 cr> i- c: o OJ cu L- E .c o +-> r> +-> ' • in CM to r— o» • c o ^ 00 *-> •r— «T5 -C (0 OJ o <*- S- "O 4-> o -O *i— »r- S ■»-> l/l r— ut o> s. LO T3 >t- p« »rt e g t. .Q L. o Q -C •f* O ■r- J- ■(-> <+■ Q. +J «♦- ai • <" •r- 1/) V c 3S 3 *-> "O •r- i — 2 TJ ■a 3 ■a a> 0.1 1/) t. a 4-> <-> CD « o t. s_ t- -o o o c: CO o„ CL r— ro •»-> a. Ql ^— +J T3 3 3 «-c LO a: LO OO CM CO >* LO 337 -i£ *~ :* IC s- s c *J Ol c i- 13 (. l/> 3 (tJ l/> 3: Z3 •t-j o c ■f— OJ > s- h- uS O uu Q. s« O o CO o OJ o s- to to cn i- 4-> c -m CD c F CD QJ F S_ CD 3 S- S_ d) S- S- 3 o. o o < o en O CO O O to o ID o o CO o CNJ c o +-> ■M o (_) o 00 o o VO o o o oas-uiD* Leo 'a^ey -o CD O +J O O Q. CD 00 A3 0) CD s- +-> (0 CD CD 4-> CNJ o LU ZD CU 339 TABLE XCIII CALORIMETRIC DATA OF SOME EXPERIMENTAL POLYESTER/CELLULOSE BLENDS (1) Blend % Residue -AH, cal/gm (2) Rate, cal/cm-sec (3 50/50 PET/HWM Rayon 25/75 PET/PFR 50/50 PET/PFR Rayon 35/65 900F/PFR Rayon 50/50 900F/Cotton 65/35 900F/Cotton 39.9 4.1 22.4 15.1 21.2 4.6 10.8 (4) 1436 3055 2245 2256 1936 2809 2443 (4) 50.5 ± 1.7 46.3 ± 0.4 (4) 60.9 ± 3.4 52.7 ± 3.0 44.8 ± 1.7 43.7 ± 0.9 44.2 ± 2.1 All results reported are averages of at least three runs S. D. = ±3.0% (estimated from calibration) Rate obtained from sample 5.1 cm wide and 12.7 cm long Supported with fiber glass gauze 340 appeared to be due to excessive melting and dripping which was complete- ly unexpected since similar blends of\ polyester and cotton scarcely melt or drip during burning. However, when the polyester/rayon blend was burned with a fiber glass gauze support, the residue yield decreased by 10-fold and heat evolution more than doubled. The rate of heat release data from this blend obtained on both supported and unsupported burning may be misleading since the rate is dependent upon the width of the specimen, or the width of the flame. Excessive melting and dripping would completely distort the shape of the specimen; thus no uniform width of flame front could be expected throughout the burning. The two polyester/PFR rayon blends gave almost identical amounts of heat when burned. Residue yield was much higher for the 25/75 blend than for the 50/50, as expected, since the char comes essentially from the cellulosic component. However, the exact levels of retardant incorporated in these two blends is not known and further comparison of the results is meaningless at the present time. The blend of the 900F polyester with PFR rayon seems ta be the most promising of this series. The availability of Dacron^900F as a flame retardant polyester has also led to its use in studying the effect of adding specific retardants to the cotton portion of a flame resistant blend. Blends using Dacronv-Aype 900F aftertreated with di ammonium phosphate produced good flame resistance with good aesthetic and physi- cal properties at relatively low add-on. These data suggested an approach consisting of a wery efficient treatment on the cellulose ^^^ combined with an inherently flame retardant polyester, such as Dacronv^ 900F. Based on this observation, an attempt was made to treat a 50/50 blend of cotton and DacronWtype 900F with the Monsanto MCC100/200/300 v^Jfinish. Under the conditions employed, only a low level of flame retardant add-on was achieved, and thus there was no significant improve- ment in the flame retardant characteristics of the blend. Further experiments in this vein should be fruitful but were discontinued in this work because of the lack of a source of supply for large quanti- ties of the blend fabric. 341 Instead efforts were concentrated on attempts to elucidate the basic characteristics of DacronV^900F in blends with cotton. Three fabrics of 50/50 blends and three of 65/35 of different weights and constructions were obtained from DuPont, along with 50/50 and 75/25 experimental 900F/cotton blends containing higher bromine contents. The results of the isoperibol calorimetric studies on the six blends containing the regular type 900F are shown in Table XCIV. These data seem to confirm that physical properties, such as fabric weight and con- struction, do not significantly affect results of the calorimetric studies. A similar effect was observed recently on a series of 100% cotton fabrics. This points out one of the important characteristics of the isoperibol calorimetric technique which enables it to effectively detect and evaluate chemical flame retardants on specific fibers in a variety of fabric types. Results from two of the regular 900F/cotton blends, one a 65/35 and one a 50/50 blend, show that the untreated material yielded slight increases in char formation and noticeable decreases in heat evolution as the contents of 900F in the fabrics increased. On this basis it was determined that these blends should have potential for effective after- treatment using systems capable of exerting condensed phase flame retardant activity on the cellulosic portion. These fabrics were therefore treated with phosphoric acid as a model system at varying levels of add-on ranging from l^%-6% (0.5%-2.0% P). The treated blends were then studies by isoperibol calorimetry and the results tabulated in Tables XCV and XCVI. The char yields obtained with these blends are compared with those obtained from normal polyester/cotton blends in Table XCVI. The ex- pected char yields from the cotton portion of the blends were calculated from previous data on phosphoric acid treated 100% cotton fabrics and are included in Table XCVI for comparison. The 50/50 900F/cotton blend produced char yields which were essentially the same as those calculated on the basis of the cotton content up to about 1^% phosphorus, However, the char yield from the fabric containing 1.9% phosphorus was more than double that expected on the basis of char formation from the 342 o X! w n CT> o <_) CD I E u c o cm >> IM O 2 -Q O CTi r^ r— r>» co CO r-» +i r»« to r»» C\J CM C\J CM , 10 , o o ro +1 *3- «a- «* CM CM CM LT) CM CO o LT) CTi + 1 CT> r->* r— *fr LD Lf> 00 CM IT) co O 9 CTi CO CM CO +1 «3- V0 i— co LO LO o (0 cr* CO o +i o o CO CD -M > > •<- o c 1 2 ^ 0) CO -M > > "I" CO CM CO CO o LO 00 CM o LT) O O o <_> © o o CT> O lt> o LT) o o LD CO ir> 343 TABLE XCV ISOPERIBOL DATA OF H 3 P0 4 TREATED 900fCS/c0TT0N BLENDS Samples %P %R Rate, cal/cm-sec -AH,, cal/gm 50/50 900FW/ control 4.6 ± 0.3 43.7 ± 0.9 2809 Cotton 0.58 15.7 ± 0.6 38.8 ± 2.6 2001 1.43 23.4 ± 5.2 24.4 ± 6.1 1658 1.90 55.1 ± 0.2 21.3 ± 0.5 1066 65/35 900F®/ control 10.8 ± 2.0 44.2 ± 2.0 2443 Cotton 0.65 15.9 ± 0.3 32.7 ± 2.6 1524 1.20 23.5 ± 2.0 15.2 ± 2.6 1441 1.57* 26.2 17 .3 1303 1.83* 40.1 11 .2 1162 *0nly one specimen run 344 TABLE XCVI CHAR YIELDS OF H 3 PQ 4 TREATED gOOF^S/COTTON AND POLYESTER/COTTON BLEND S XR Sample %P %P cotton^ 900fCB)/ cotton PET/cotton calc.( 2 ) 50/50 blend control -- 0.58 1.16 1.43 2.86 1.90 3.80 4.6 5.2 5.2 15.7 14.6 16.0 23.4 19.8 . 23.7 55.1 21.6 26.0 65/35 blend control -- 10.8 8.7 8.7 0.65 1.86 15.9 13.6 14.0 1.20 3.43 23.5 16.6 17.6 1.57 4.49 26.2 17.8 19.3 1.83 5.22 40.1 18.5 20.1 %P content based on mass of cotton in the blend 2, "Calculated from previous data on H 3 P0, treated 100% cotton using %P cotton 345 cotton portion only. Similar behavior was noted for the 65/35 900F/cot- ton blends. These fabrics gave slightly higher char yields than those calculated on the basis of the cotton portion. However, the largest difference again occurred at about 1.8% phosphorus content where the 900F/cotton blend gave twice the amount of char calculated on the basis of cotton content. Both blends (50/50 and 65/35) became very difficult to ignite at their highest level of treatment. On the basis of subjec- tive observations of the burning characteristics in the calorimeter, it is estimated that a phosphorus content of 2.0% should be sufficient to effectively inhibit burning and that these fabrics should pass FF 3-71. The flame retardant contribution of the bromine in the 900F Dacronw'is probably due to a vapor phase mechanism and is not signifi- cantly related to char formation. At the lower levels of add-on (up to 1.5% phosphorus for 50/50 and 1.0% for 65/35) where there is still sufficient heat generation from the cotton portion of the blend to sustain the combustion process, there is essentially no residue from the polyester remaining in the char. However, at higher phosphorus contents, (1.8% to 1.9%) the diminished heat generation from the cotton portion coupled with the probably vapor phase retardance from the 900F results in a large reduction in total heat generation and thus in the heat fed back to the substrate. This apparently results in a decrease of the heat flux at the fabric surface to a point where it is not sufficient to sustain complete degradation of either the cotton or the polyester. Therefore, a large quantity of residue is left. The dependence of the heat release for both the 900F/cotton blends and the normal PET/cotton blends is presented graphically in Figure 103. The 50/50 blends exhibit heat release values which are essentially the same for the two blends up to about 0.5% phosphorus. At higher phosphoric acid add-ons, the heat release from the PET/Cotton blend remained unchanged. A similar behavior was observed with the 65/35 blend, although the heat release of the untreated system contain- ing 900F was about 400 calories lower than that containing the normal polyester, indicating the effect of the bromine in the 900F. Similarly the heat release from the untreated 65/35 blend containing 900F was 346 3000 2000 ? 1000 65/35 PET/Cotton 50/50 PET/Cotton 50/50 900F/Cotton 1 1.0 % P 2.0 FIGURE 103. Heat release of H 3 P0 4 treated 900F/cotton blends. 347 also about 400 calories lower than that of the 50/50 900F cotton blends, This is presumably due to the larger amount of bromine in the 65/35 blend. The effect of the bromine can be seen more clearly if the heat release (AH-j) is shown as the sum of the heat release from the cotton and the polyester portions of the blend as follows: AH, = Q c + Q p {9} where Q and Q are respectively the heat release from the cotton and c p polyester portions of the blend. If one assumes that the combustion of cotton in the blend is the same as that in the form of 100% cotton fabric, the Q can be calculated from the previous data obtained on 100% cotton treated with phosphoric acid, and Q can be calculated from p equation {9}. The plots of AH-,, Q , and Q as a function of percent phosphorus are shown in Figure 104 for the 50/50 blends, and in Figure 105 for the 65/35 blends. Both Figures 104 and 105 show the drastic difference in heat release from the 900F and the regular polyester. For either blend, as Q decreases with increasing phosphoric acid con- tent, Q_ increases, while the total heat (AH,) remains essentially P i constant. However, Q in the case of 900F remains unchanged until the phosphoric acid content is increased to about 1.5% phosphorus. It then drops drastically to below 1000 calories at 1 .8%-! .9% phosphorus, while Q is less than 400 calories. These data indicate a distinct retardant effect as compared with those of the regular polyester. The heat release values are in agreement with the char yield data, and show that up to 1.5% phosphorus the heat release from 900F is essen- tially unaffected by the presence of the phosphoric acid. Such results were expected since phosphoric acid is a condensed phase retardant and has no known retardant effect on polyester. The drastic retardant effect observed on 900F cotton blends at a phosphorus content above 1.5% is probably due to the synergistic effect which results from the reduction in heat release from the cotton portion and vapor phase re- tardant effect from the bromine. This should produce an insufficient heat feedback to the substrate, resulting in less than complete degra- dation of both polymers. This, in turn, of course, reduces the amount of fuel gas available to the flame, and the effect is consider- 348 3000 2000 < I 1000 _L 1.0 % p PET/COTTON 13 2.0 FIGURE 104. AH, of H 3 P0 4 treated 900F/cotton and PET/cotton 50/50 blend fabrics. 349 3000 2000 E o 1000 PET/COTTON X 1.0 ', P 900F ^Q ;otton 2.0 FIGURE 105. AH ] of H 3 P0 4 treated 65/35 900F/cotton and PET/cotton blend fabrics. 350 ably greater than additive. In an effort to further evaluate the flame retardant action of the bromine, experimental samples of 900F containing 7.5% bromine in- stead of the normal 6.0% were obtained from DuPont in the form of 50/50 and 75/25 blends with cotton. These two fabrics were also treated with phosphoric acid at three different levels of add-on up to 1.8% phos- phorus. The isoperibol results on these blends in treated and untreat- ed forms are shown in Table XCVII, together with similar samples based on the normal 900F fiber. Heat release values were also plotted as a function of %P and are shown in Figure 1 06 • These four series of blends contain different levels of bromine, ranging from 3.0 to 5.6 due to the different ratios of 900F and cotton. Figure 106 shows that blends with higher bromine content tend to release less heat than the ones with the lower bromine content at the same phosphorus add-on. Two phosphoric acid treated 75/25 blends failed to burn in the calorimeter; these had phosphorus contents of 1.68% and 1.80%. However, the trend which shows decreasing heat release with increasing bromine content is clearly demonstrated. The significance of bromine in these blends can be shown if the heat release data is further reduced. Figure 107 shows a plot of heat release versus bromine content at several levels of phosphorus. This graph shows the linear decrease of AH., with increas- ing bromine content. Furthermore, since these data represent three different blends, the linear dependency of AH. on bromine content significantly demonstrates one of the important characteristics of vapor phase retardancy, i.e. the effectiveness of vapor phase retar- dants should be independent of the nature of the substrate. Values for the 65/35 blend are consistently off the line. This may be due to an error in the phosphorus content in the treated samples. The fact that all of the lines in Figure 107 seem to parallel each other indicates that the effectiveness of the bromine on these blends is identical irrespective of the amount of phosphorus present in the blend. This tends to suggest that bromine and phosphorus are working independently of each other and is evidence that there is little or no bromine/phosphorus interaction or synergism such as has 351 TABLE XCVII :® ISOPERIBOL DATA OF 900F^/C0TT0N BLENDS WITH VARYING BROMINE CONTFNT Fabric %Br %P %R Rate, cal/cm-sec -AH, cal/gm 50/50 900F^yct 3.0 control 4.6 43.7 2739 0.58 15.7 38.8 2001 1.43 23.4 24.4 1658 1.90 55.1 21.3 1066 50/50 900FV$ct 3.8 control 7.2 52.7 2546 0.47 17.7 35.1 1920 1.28 21.9 23.2 1443 1.67 35.5 20.1 1141 65/35 900F^Vct 3.9 control 10.8 44.2 2409 0.65 15.9 32.7 1524 1.20 23.5 15.2 1441 1.57 26.2 17.3 1303 1.83 40.1 11.2 1162 75/25 90oWct 5.6 control 11.1 39.3 2277 0.72 21.0 15.5 1653 1.28 37.6 ± 8.9* 1259 ± 208* 1.68 did not burn 1.80 did not burn ♦Average of five runs, large error probably due to uneven treatment. 352 3000 2000 — on < i A3 r— ai ■m 1000 — 03 O 50/50 900F/Cotton (3.0% Br) O 50/50 900F/Cotton (3.8% Br) A 65/35 900F/Cotton (4.2% Br) D 75/25 900F/Cotton (5.6% Br) ^SAMPLE DID NOT BURN 1.0 Percent Phosphorus Content 2.0 FIGURE 106. Heat release of various H 3 P0 4 treated 900F/cotton blends 353 3000 ^ 2000 < 1001 L CONTROL O 50/50 900F/Cotton (3.0% Br) O 50/50 900F/Cotton (3.8% Br) A 65/35 900F/Cotton (4.2% Br) Q 75/25 900F/Cotton (5.6% Br) _ PVBr/THPC Treated PET/Cotton (1.0% P) VBr/THPC Treated PET/Cotton (1.0% P) ^ L 1.5% P x j L 2.0 3.0 4.0 5.0 Percent Bromine Content 6.0 FIGURE 107. Heat release of various 900F/ cotton and PET/cotton treated with H 3 P0 4 or PBVr/THPC. 354 previously been suggested to exist in other systems 355 3. Phosphorus/Bromine Systems Based on Phosphonium Salts 3. a. THPS^/urea/PVBr and related systems : One of the first potentially commercial systems in which the ac- tion of both bromine and phosphorus was important was that based on the use of THPC/urea/polyvinyl bromide as developed at SRRC under the direc- tion of Mr. George L. Drake, Jr. several years ago. Because of the commercial potential of this system, considerable effort has been expended to attempt to modify the formulation to circumvent the problems of low reproducibility, color formation, and harsh hand which have been noted in several laboratories. Efforts were therefore undertaken by both Hooker Chemicals and Plastics Corporation, and SRRC to modify the application of the THPC portion of the system. At the same time, investigations were initiated at both Clemson University and the Uni- versity of Maryland to further elucidate the chemical nature of the processes responsible for both the retardancy and the deleterious effects. Ethyl Corporation, the major producer of the polyvinyl bromide used in this system, has actively participated and cooperated in these investigations. Ethyl Corporation provided samples of the polyvinyl- bromide in both pure and latex form, and also provided samples of selected copolymers of polyvinyl bromide with other vinyl materials - particularly vinyl chloride. Initial investigations with these systems indicated some effect by both constituents, although the relative contributions could not be determined. Studies conducted both at SRRC and Clemson demonstrated that the copolymers give much better performance during curing and produce onTy slight, if any, discoloration. Thus a program was undertaken at SRRC to further develop the THPC-urea-P(VBr/VCl ) system. After consider- able laboratory experimentation, the formulation given in Table CVIII was found to produce good results when applied in the SRRC pilot plant. Under these conditions a white fabric could be prepared having no dis- coloration, good flame resistance and good hand. It was also found that the hand of the samples produced using the standard ETIP 50/50 356 TABLE XCVIII THPC/UREA/P(VBr/VCl) FINISH FORMULATION % by weight THPC (80% aq.) Urea Na 2 HP0 4 NaOH (50% aq.) P(VBr/VCl) emulsion (40%) H 2 Triton Procedure: © 770 33 3 8 5 4 3 .0 16 .0 35. 2 Dissolve Na^HPO, in hLO with stirring Add THPC with stirring Add Urea Adjust pH to 6-6.5 with NaOH Add P(VBr/V£l) emulsion Add Triton®770 Application : Pad through above solution using ca. 6 tons pressure on squeeze rolls to get wet pick-up ca. 85%. Dry 2 min at 85°C Cure 1 min at 160°C Wash in 1% H^O^ 2 min followed by 1 hot and 1 cold rinse Dry Fabric Analysis : original 50X P 2.5% Br 3.2% CI 1.5% Char length (AATCC) 2.5 in 4.5 in 357 poplin could be further improved if the fabric was prescoured for \H hours using a warm scour bath containing 5% Mayquest^80 (EDTA), 5% PreChem^SN (a halogenated aromatic), 5% PreChem^70 (wetting agent) and 1% NaOH (all owf). Because these results were so promising, a 250 yd. mill run was made at the Clearwater Finishing Plant of UM & M. This used the form- ulation that had been the best under the pilot plant conditions (Table XCVIII). After padding and drying for 58 sec. in a frame with three heating zones (reading 182°C, 202°C, and 152°C, respectively) the fabric was found to be insufficiently dried. It was therefore redryed using the same temperatures with a residence time in the frame of 30 sec. The fabric was then cured for 2 min 15 sec at 154°C, washed 15 min in dilute Na^CO-, rinsed 5 min hot and 5 min cold, put through a scutcher and dried. The resulting fabric was very light purplish brown and somewhat stiffened. Presumably, at least part of this discoloration and stiffness was due to the extra drying that was required after padding. Three polyvinyl bromide lattices were applied to samples of 100% cotton, 100% polyester, and 50/50 cotton/polyester blend fabric. The three latices were polyvinvlbromide, a 50/50 copolymer of vinylbromide and vinyl chloride, and a 94.7 to 5.3 copolymer of vinylbromide and N-methylolacrylamide. These latices were applied in varying concentra- 9 tions and the resulting 01' s showed a linear relationship with the per- cent bromine present in the fabric. It was interesting that all three samples - cotton, polyester, and the blend - fell on the same line. It was also noticed that the chlorine in the polyvinyl chloride portion of the copolymer seemed to have no significant effect on the oxygen index values exhibited by the samples. A different set of samples was treated with the same polymer , lattices but with a small amount of antimony oxide added to the pad bath. The addition of the antimony oxide resulted in an increase in the 01 values on all the samples, with the largest effect being noticed on the polyester, and the smallest effect being noticed on the cotton. This would indicate that the antimony oxide was increasing the vapor phase 358 activity of the bromine present in the sample. The antimony oxide did not, however, increase the activity of the chlorine present, as again the chlorine and the polyvinylchloride did not affect the oxygen index observed for the samples. Samples of both sets of treated fabrics were then burned on the 45° angle pin-frame and their burning characteristics subjectively evaluated. The antimony oxide had little effect on the manner in which the samples burned. All the samples burned with a thick grey smoke which seemed to result from the flame retardant itself burning off the surface of the fabric. The blend fabrics as well as the cotton fabrics would burn the entire length of the specimen leaving a substantial char which would then be consumed by an afterglow. The cotton samples from both treatments had an oxygen index of 26 but would not burn in the subjective evaluation. The blend fabrics and the polyester fabrics had similar 01 values at the same bromine add-on, but they were observed to burn readily. This probably resulted because the cotton flame was not hot enough to burn off the flame retardant polyvinyl bromide latex. The flame of the blend fabric, however, was hot enough to burn off the flame retardant and the fabric would, therefore, burn. In another series of experiments, twenty (20) polyester/cotton blend fabric samples were prepared by Ethyl Corporation. These were first treated with THPC to a level of 1.0%P and then over-treated with polyvinylbromide (PVBr) or poly(vinylbromide/vinylchloride){P(VBr/VCl)} copolymer at various levels of add-on. Twelve of these fabrics, inclu- ding a control, were selected for study by isoperibol calorimetry. The results of this study are given in Table XCIX. Two additional fabrics, one a 65/35 polyester/cotton blend, and the other a 50/50 polyester/ rayon blend, treated with the THPC and PVBr at SRRC were also included in the study. The SRRC 65/35 polyester/cotton gave a lower char yield, a higher heat release, and a higher rate of heat release than the 50/50 blend, due mainly to the higher polyester content. Even though both blends were overtreated with THPC, a 3% bromine content was shown to be insufficient to achieve satisfactory flame retardance on either fabric. The previous study of phosphoric acid treated 50/50 and 65/35 359 u m cc m o Ul f==i) ^ 'jC. '^Sr CO o «3- CO cm CM r- CO CO CO CO «* CM CM CO +i +1 + 1 +1 +1 +i +1 +1 + i +1 +1 ^si- CO C\J \o CXI oo w> LD in CO CM IT) 450 3F 00 CO CM CO o in *3- in **• in oo Q CO Q UJ !_, CM =3r oo o f— CM on 3 lf= *t in GO m CO CO Cv' 00 Kf CO o *3- CO o in •— o CO co l£> «* en o\ oo in o CO o o. X H U s- CO OT CO %G r** CO CO CO r«* m in r— CM CM CM m r»» CO B*** oo CO CO r^ 00 CO «3- 9 CO o a a. o oo Q. CO a. in co in in SZ t- a. CO a. *— >> or o m o m o a. o in o in CO — Q. O in o in CO o CO CO co CM •—•—<_) CO co o CO CO r— CM O a. o in o in <-> o_ o in o in o_ o in o in c_> a. o in o in c_> o. o in o in CO 00 o CO in i— i— o o_ o in o in o o_ o in o in o a. o in o in CO r— «* in >P > r— CM co in r» P«> ^-» r~ 2E =**= =*= =*t =»t =*: =*t =*= =«*= =»*= CO a> o c fc- _l _l >> >> >» >> >» >> >> >> >» •i— zs cc cc JZ J= -C -C JZ JZ JZ JZ JZ s. o cc cc 4-> +-> •p ••-> +-» ♦J -4-> 4-> 4-» Q. oo oo OO UJ UJ UJ UJ UJ UJ UJ UJ UJ OO 360 900F cotton blends containing 3% bromine indicated that these blends should be self-extinguishing at about 2% phosphorus. Comparison of the THPC/PVBr treated polyester/cotton and the phosphoric acid treated 900F/cotton indicates that the bromine from the PVBr is probably some- what less efficient than that from the 900F polyester. This is probably due to the low decomposition point of PVBr; however, it must also be noted that the use of PVBr introduces additional fuel into the system. The results shown in Table XCIX and Figure 108 indicate that the retardant effect of PVBr is real and significant. The heat release of the THPC/PVBr treated 50/50 blends decreases consistently with increas- ing PVBr add-on. The rate of heat release also decreases as the amount of bromine increases. Similar results were obtained with the THPC/P(VBr/VCl) system, but the changes in rate and in total heat evol- ved were less, suggesting that the copolymer is less efficient than pure PVBr. The decreased efficiency may be due to the fact that the co- polymer contains less bromine than the pure PVBr and that the chlorine is somewhat less effective. Another factor which may very well affect the flame retardant efficiency of the copolymer at its lower decompo- sition point than the PVBr. In order to obtain a better evaluation of the importance of these effects, the PVBr and P(VBr/VCl) systems were compared by static oxygen bomb and isoperibol calorimetry. In the case of a strictly solid-phase system, previous data on treated 100% cotton had shown that both AHL and AH« decrease as a logarithmic function of %P content, and Table XCIX shows that the ratio AhL/AHL, is independent of the phosphorus content. Current data on PVBr and VBr/VCl treated polyester/cotton blends indi- cate that the ratio AH,/AH 2 for both systems decreases linearly with increasing halogen content as shown in Figure 109. Figure 110 shows that AH«/(AH ) F decreases as a function of increasing halogen content. Of perhaps greater significance, AH«/(AH°)p for both the PVBr and VBr/VCl treated systems fall on the same line, indicating that the fuel- generating process (pyrolysis) is the same and not affected by the retardant. However ,AH,/(AH°)p shows the difference in efficiency between the two systems, i.e. the PVBr treated system is slightly more efficient. 361 2000 E en < > 1000 O PVBr ▲ VBr/VCl J. 10.0 % Add-on 20.0 FIGURE 108. Variation of AH ] with %? of PVBr and P(VBr/VC1) treated PET/cotton blends. 362 TABLE C HEAT BALANCE IN FLAME RETARDANT TREATED COTTON SYSTEMS %P AHj AH 3 -A^*, cal/gm -AH 2 *, cal/gm ' n AH 2 ' -AH 3 , cal/gm v AH 2 3324 3688 90.1 436 11.8 H 3 P0 4 .56 2041 2218 92.0 301 13.6 .77 1823 1980 92.1 273 13.8 1.03 1623 1762 92.1 219 12.4 1.38 1422 1543 92.2 177 11.5 Ave. 92.1 Ave. 12.8 (NH 4 ) 2 HP0 4 .49 2042 2342 87.9 337 14.5 .81 1649 1867 88.3 277 14.8 .93 1541 1741 88.5 265 • 15.2 1.11 1403 1580 88.8 246 15.6 Ave. 88.4 Ave. 15.0 THPOH-NH 3 ( \ .15 2852 3197 89.2 394 12.3 .53 2132 2392 89.1 297 12.4 .98 1781 2000 89,0 244 12.2 2.23 1311 1475 88.9 174 11.8 Ave. 89.0 Ave. 12.3 * Smoothed values except for untreated cotton (0%P) 363 c o 00 c o m o IX) -a cu > O u s- CQ sr o C\J < 4- O a> u c O o LU -o 4-> «o +-> O CO c CO Q_ o 00 < CM < -o U_ OU 3: < Q3SV313cJ 1V3H JO NOUOVcJJ a: Z3 CD 365 Of considerable additional significance is the observation that char yields were not affected by the addition of either PVBr or the PVBr or the VBr/VCl copolymer. The char yields from the treated fabrics were identical to those of the controls; furthermore, oxygen bomb cal- orimetry on these chars revealed that the compositions of the chars were quite similar to that of the control, as would be expected for a vapor phase active retardant. Comparisons of the data obtained on the VBr systems with those from the 900F/cotton blends show two significant differences (Figure 107). First the heat release values for both 900F/cotton and the PVBr/ THPC treated polyester/cottons show linear dependency on the bromine content. However, the slope obtained with the 900F system is consider- ably steeper, indicating that the bromine from the 900F is, in fact, more effective. Secondly, the PVBr/THPC treated blends show no increase in char yield up to about 20% add-on while the 900F cotton system shows significant increase in char yield beyond that expected from phosphorus alone. These data indicate that bromine in the form of PVBr or its copolymers is probably active only in the vapor phase, whereas the aromatic bromine in the tetrabromobisphenol-A monomer of 900F could be active in both the PET and the cellulose. Alternatively, the increased residue could simply be the result of incomplete pyroly- sis of the polyester due to the reduced heat liberated from the flame in the presence of a more efficient vapor phase active retardant. In any case, the difference in efficiency between the bromine in Dacron® 900F and PVBr is real and significant. This would suggest that improved formulations of the THPC/urea/PVBr type should be possible if bromine latices could be found which have higher flame retardant efficiencies. 3.b. Determination of Flame Retardant Efficiencies of Bromine Compounds : In order to provide the necessary data on flame retardant ef- ficiencies of bromine compounds, a systematic evaluation by isoperibol calorimetry was undertaken. Since subjective evaluations had shown that decabromobisphenylene oxide in the presence of Sb 2 3 (FR P-44^0 was 366 a particularly effective system, it was the next one chosen for more detailed study. The heat release from fabrics treated with this for- mulation is shown in Figure 34 . Values of heat release from these fabrics with and without grid support, decrease linearly with increas- ing Br content. Linear regressions of -AH, vs % Br are tabulated in Table CI. The linear regression gave a slope of 228.2 cal/gm-%Br on the 50/50 blend. This is essentially identical to the regression slope obtained on the 100% cotton fabrics without support. This suggests that P-44 has identical efficiency on all three different fabrics when (5) burned without support. In other words, the efficiency of P-44 K ^ f is not substrate-dependent which can be taken as an indication of vapor- phase retardant action. With a grid support, -AHL of 100% cotton gave a regression slope of 193.6 cal/gm-%Br which is slightly lower, but not significantly different from the slope of the data obtained with- out support. However, data from the 50/50 blend gave significantly larger slopes when burned with support, 292.4 cal/gm-Br vs 228.2 cal/ gm-Br. Of course, the conclusions relating to optimum formulations which can be drawn from these experiments are highly dependent on the efficiency of the bromine-containing retardant. Thus it becomes nec- essary to characterize the flame retardant efficiencies of a variety of bromine-containing structures. Several series of fabrics treated with bromine retardants were therefore examined by isoperibol calorimetry. A plot of net heat reduction, AHj - (AH 1 ) CQntrol , vs Br content is shown in Figure 35. This plot is essentially the same as that in Figure 34, but with the differences in AH, of various control fabrics normalized. Linear regression of various groupings of these data were taken and the regression results are tabulated in Table CII. These results indicate that the best regression fit on the basis of intercept and the standard deviation of the slope is obtained when all fabrics with or without grid are fitted on the same line. This tends to suggest that the grid support has no effect on the efficiency of P-44 Viy . The scattering of data among the results is obvious, and is 367 TABLE CI LINEAR REGRESSION OF HFAT RELEASE FROM P-44 V ^T Heat Release: -AHj = A + B(%Br) .-44®! Fabric A* B, cal/gm-%Br r W/0 Grid 100% Cotton -3329(3334) 226.2 ±20.4 .9802 50/50 • -3004(2942) 258.2 ±36.0 .9624 65/35 -2743(2846) 193.4 ±19.0 .9813 50/50 and 65/35 -2872(2942,2846) 228.2 ±21.5 .9602 W/ Grid 100% Cotton -3485(3424) 193.6 ±29.3 .9571 50/50 -3485(3460) 292.4 ±14.7 .9962 * Values in parentheses are -AH, of the controls 368 TABLE CI I LINEAR REGRESSION OF NET HEAT REDUCTION FROM P-44^TREATED FABRICS Net Heat Reduction: AHj - (AH 1 ) control = A + B(%Br) Fabrics A B, cal/gm-%Br W/O Grid 11 228.7±14.7 .9666 All Fabrics W/ Grid 100% Cotton -61 193.6±29.3 .9571 50/50 -25 292.4±14.7 .9692 50/50 W/ and W/O Grid -40 272.U23.Q .9692 All Fabrics W/O Grid and 100% Cotton W/grid -13 222.6±15.2 .9503 All Fabrics W/ and W/O Grid -17 234.6±15.6 .9430 369 probably due to uneven treatment of P-44^on the fabrics. Two additional series of fabrics, 100% cotton and 50/50 blend, treated with the decabromobi phenyl ene oxide without the Sb^O, (FR P-53 v - y from White Chemical Company) were obtained and studied. Calor- imetric results from the fabrics treated with P-53^are tabulated in Table CIII. Heat release data are shown in Figure 111 plotted vs Br content. Data from previously obtained P-44 v - / treated fabrics are also included for comparison. Values of heat release of both the P-53^ treated 100% cotton and 50/50 blend are linearly dependent on % Br, with the variation in the data slightly larger for the 50/50 blend. Linear regression of both sets of data are shown in Table CIV Data from the P-53^ treated blend exhibit a slightly smaller slope (ft than from 100% cotton. However, both P-SS^ treated fabrics exhibited much smaller slopes than those treated with P-44^. Linear regressions were also obtained for net heat reduction. The best fit for P-53^ treated fabrics, based on the intercept, A, and the standard deviation of the slope, B, occurs when both sets of data are fitted on the same straight line with a slope of 108.6 cal/%Br. This is less than a half of the slope for the P-44^ treatment. The effect of antimony in P-44^is very profound. Plots of net heat reduction vs % Br for both the P-44^and P-53^ treated fabrics are shown in Figure 112 . Efficiencies of all bromine-containing retardants studied are compiled in Table CV. P-44 ^clearly is the most efficient system but, unfortunately, it is not amenable to use with THPC or other phos- phorus treatments. Phosphorus is known to exhibit an antagonism toward antimony/halogen systems. As expected, this antagonism was encountered when the P-44^was used to top fabrics treated with a THPS^/urea finish. 3.c. Utilization of Other Bromine Compounds with Phosphonium Salts : In an effort to develop more efficient bromine sources for use in lieu of PVBr or the VBr/VCl copolymers, a series of structures 370 -53© TABLE cm P- TREATED BLEND FABRICS (W/0 GRID SUPPORT) % FINISH ON % Br % Br (owf) -AH, , cal/gm AH, - AH, . - 1 1 , control 100% COTT Control — — 3334 -- 8.19 3.94 4.29 2864 470 14.94 7.42 8.72 2436 898 24.84 9.49 12.63 2168 1166 32.51 13.28 19.68 1832 1502 44.67 15.18 27.44 1562 1772 50/50 PET/COTTON Control — — 2942 -- 9.18 4.50 4.95 2348 594 15.91 7.34 8.73 2384 558 25.09 9.61 12.83 1936 1006 45.96 14.23 26.33 1549 1393 371 P-44 P-53 3000 100% ct £ O 50/50 P/cA A 65/35 P/C^ CT. < ! 2000 1000 5 10 % Bromine 15 FIGURE 111. Heat release o f p-44^and P-53VV and P-53 v -^ treated fabrics 372 TABLE CIV EFFICIENCIES OF P-44^AND P-53^TREATED FABRICS -AHj = A + B(%Br) Fabric A B, cal/gm-%Br r P-44 w Treated 100% Cotton -3329 226.2 ± 20.4 .9802 50/50 and 65/35 -2872 228.2 ± 21.5 .9602 p_53WTreated 100% Cotton -3313 115.3 ± 3.0 .9986 50/50 -2913 95.5 ± 12.3 .9758 AH 1 - < AH 1 'control " A + B < %Br > Fabric A b, cal/qm-%Br r P-44^/|"reated All Fabrics 11 228.7 ± 14.7 .9666 P-53^Treated 100% Cotton 21 115.3 ± 3.0 .9986 50/50 29 95.5 ± 12.3 .9758 All Fabrics 12 108.6 ± 7.4 .9796 373 2000 olOOO < i < 10 100% ct % O 50/50 P/cA A 65/35 ?/zW 15 % Bromine •-44® and P-53VV FIGURE 112. Net heat reduction from P-44W and P-53W treated fabric 374 > u w CQ < Eh co Q C£ c£ o o OC CO O CO O co &« CO + II o o < I s- I E cn CO u •I— s- LO o lo cn co r^. LO CO CX> cn in r^. CO CO r*. cn cn CO CT> cn cn CTi cn cn i^ O «* cn «it lo «a- ** r^ O co LO ■ — ' — +i +1 +i +i +i +i r-^ LO LO LO CM co co CM LO CNJ 8 co CTi (jO O LO CM ^~ i— «— co © O Li_ lo o ■^ o o cn LO LO CO o +J o o o o LO o LO LO CO LO LO LO CNJ LO CM -o 1 T3 C cu CO o_ o LO o LO c o +-> -)-> o cj> o o -o c O) CO C_) O. o LO o LO +-> CU or ■o c O. o LO o LO <~> n_ CO ^O- u CO «=i- O LO CO CO > 1 O 1 CM > Q_ Cn o_ 1— o_ Q. C O *f* +-> O 3 ■»-> -M c ITJ CU CU t/> •r- ^ ■•-> c •»~ ■M «4- zzz. c «/) r— cu- <3J .»-%. L. (. t— 1 cn S- 3: cu O < cc O n II 1 CO 375 related to dibromocyanoacetamide was prepared and evaluated. Dibromo- cyanoacetylurea has been synthesized and applied to 50/50 blend fabric in conjunction with THPC. An 18% add-on was achieved to yield a sample exhibiting an 01 value of 29.0 and a BOI 23.0. This material showed a significant weight loss in an oven at 110°C for thirty minutes due to possible steam distillation of the material. Attempts to fix the material by cross-linking with urea, Aerotex 23 V£y , and dimethylol dihydroxyethylene urea resins resulted in gross discoloration. There was also a marked loss in flame retardant characteristics following either drying or cross-linking. The treated fabric was noticeably hydroscopic in the absence of resin treatments. More promising results were obtained with dibromoacetamide which was prepared by hydrolyzing dibromoacetonitrile. A white crystalline product with a melting point of 152°-154°C was obtained. This material melted with no evidence of decomposition. A weight loss was observed in the region 130°-145°C by TGA, presumably due to sublimiation of the compound. The dibromoacetamide was padded on to 50/50 blend fabric to produce a sample which showed no discoloration upon heat treatment for three minutes at 150°C. However, this fabric exhibited very low flame retardant efficiency as either the free amide or its N-methylol derivative. Complexation of the amide with 2 moles of THPC produced a reagent with improved retardant characteristics but attempts to fix the retardant into a durable finish failed. Other attempts to produce reactive bromo compounds included the bromi nation of cyanoacetyl pi peri dine which failed to yield an i sol able product and the sny thesis of bromocyanoacetyl urethane which let to a product with a melting point of 78-80°C. This latter material showed considerable decomposition at 150°C on a 50/50 blend fabric. The compound itself decomposed slowly with evolution of HBr. Better results were obtained using dibromomalondiamide. This material was found by TGA to be stable to 175° and exhibited good flame retardance on blend fabric. N-methylol a ti on of the material was achieved readily and the resulting product appeared stable to heat. However no fix could be obtained with this material. 376 Since it would seem that the hydrolytic stability of the dibromo- malondiamide might be the subject of concern, an investigation was carried out to determine its stability toward base. After three hours in a water slurry no evidence of HBr evolution could be found. The pH of the slurry was then raised to approximately 10 using sodium hydroxide and the slurry allowed to stir overnight. During this time the pH dropped to approximately 8 indicating some loss in HBr from the material. A large quantity of the water-insoluble diamide remained however, in- dicating the retention of most of the bromine. On this basis an attempt was made to develop a formulation suitable for larger scale fabric evaluations based on the THPC adduct of the diamide but only limited success was achieved. The dibromomalondiamide (DBMA) was also evaluated as a co-reac- tant for phosphorus containing flame retardants. The compound was padded on to 50/50 fabric along with Fyrol 76^ to produce an excellent flame retardant fabric but no cure could be obtained with persulfate catalyst. Better results were obtained using the bromo compound in the form of its complex with 2 moles of THPC. This produced good flame resistance and after curing the finish exhibited limited dur- ability to a normal process wash. Attempts to improve the durability by fixation with a variety of nitrogenous resins led to extensive discoloration on curing. The solubility problems encountered with DBMA were overcome by N-methylolation to produce a water soluble species. This was applied to 50/50 blend fabric using a non-nitro- genous resin based on formaldehyde (Tetraset ST-l®from Riegel Textile) and cured for 30 seconds at 165°C. These samples discolored strongly. Experiments were also run using the N-methylol DBMA (1) by itself, (2) with Tetraset ST-1®, (3) with THPC and (4) with THPC and Tetraset S" 1 "" 1 • These were cured at 150° for 1, 2, and 3 minutes. All of the fabrics discolored to varying degrees with those containing THPC giving the most pronounced color formation. The brominated amides were therefore abandoned and attention turned to efforts to develop a bromine-containing latex more suitable than PVBr or P(VBr/VCl) for application in conjunction with phosphonium salt 377 finishes. As previously reported, techniques were devised for the emulsion polymerization of DBPA and TBPOEA. The DBPA latex contained approximately 40% solids as it was prepared and was suitable for direct application to fabrics. The same recipe was inappropriate for either VBr ? or TBPOEA. In both of these latter cases, polymerization was effected but no stable emulsion could be prepared. In one series of experiments, fabrics previously treated with Pyrovatex CP^and THPC/urea finishes were over-treated with poly (vinylidene bromide) or with the P(DBPA) latex. Based on 01 and vertical flame test results, these finishes were not as effective as those based on Fyrol 76^and the bromine compounds. The latex was also applied to the 50/50 ETIP fabric using the formulation in Table CVI. The resulting fabrics exhibited borderline flame retardancy and limited durability. When a slightly heavier fabric was used with the same formulation an initial char length of 5.1 in. was observed. This fabric exhibited a borderline failure after 50 launderings. The samples had a harsh, stiff hand and suffered from noticeable strength loss. At this point the materials were turned over to the Research Committee of the Palmetto Section, AATCC. This group, composed of representatives from United Merchants and Manufacturers, M. Lowenstein and Sons, the Graniteville Co,, C, S. Tanner Co., Emery Industries, C. H. Patrick and Co., and Clemson University, has undertaken a com- prehensive study of the P(DBPA) latex and is currently working to evaluate the feasibility of using this latex in conjunction with a THPS®-urea finish. A series of samples were treated with THPC-urea P(VBr/VCl) by Dr. John Holston of M. Lowenstein and Sons using the pro- cedure of Donaldson and co-workers (57) and compared to a similar series prepared in the same way except for the replacement of the P(VBr/VCl) by P(DBPA). The results are shown in Table CVII. Satisfactory flame resistance was not obtained with either finish when evaluated by the criteria of FF 3-71; however all of the samples were found to fall into Class I using the MAFT. On this basis, it would seem probable that satisfactory performance in FF 3-71 could be obtained with either form- ulation using a heavier weight fabric or a fabric containing softer 378 CD e > OO O CO CO 4-> O 4-> 4) < CQ Q o> CD D"> > u <: O O »— i I— o o o o < — ■» o o o o 00 CO CO •"""" r— "O •r— ^ i o o LT> 00 — ^ r^. co o «3- «nT Q_ N — •* 8 zn o_ .© e o o 00 X CD 4-> f0 © "0 o r— r^ r— ^ o V ^ o GO ( — <: o Q_ tO o o_ >> zn CQ c o •I— +-> 13 'o to DO o o oo +-> ZJ o CO CD 4-> 3 379 H H > u w < o> ••-» •« OT3 ^ c » X X X X X X X X X L. CO • V0 co r— • c^ co 10 to CM a> *J 1"" • • • • • • ■ >— w< ~— » CO cn CM CO CSJ CM CM CM Cv* to -O ^~ s: ^~ »„;-, -f— e l/l *> c k * «-> »— t/> 01 £ rt? o> '+■ CT »o t. i- &> a* *-» > <♦- < «j •f- cr > c •»- CD 2 u r- 01 •— < to Q. E CM r— o CO o Cn «* *> in o o • - «* O o i--=* £=. r«» X X X 10 oj • • •■a tO O • * «• CO o O m • • CM o o • r— • ^— co ** * • o 10 O - • • o oo CO 1 — o in o o •43 o p« * O * w o «o m f™ CO X X X r- r- m lO «* co •a. m co «d- co CM ■— •"*■ VO CO •**■ CO o ^ _ ^ ^^^ a o r— in *»■ CO r> *> s o cn ^ • • X *»■ CM o X X o CO o-l M c E i— 4- at =) E ©'! 380 yarns. In a concurrent series of experiments conducted by Mr. Charles Bailey of the Graniteville Company, a number of modified formulations for this finish were evaluated. Donaldson and co-workers (57) had previously recommended the incorporation of resins such as trimethylol- melamine or trimethylolglycourea to improve the fixation efficiencies of soft latex systems. Thus a series of samples was prepared using the same formulation except for the addition of 4.0 weight % (owb) of Aero- tex vjy Resin M-3 in with the THPC-urea-Na^HPO, bath component mix prior to addition to the P(DBPA) latex. The results from these experiments are also given in Table CVII. These data would seem to indicate that the achievement of higher fixed add-ons should lead to a satisfactory performance level. Thus attention was focused on efforts to modify the P(DBPA) formulations to increase the efficiency of fixation. Both pad- dry-heat cure and pad-dry -ammonia cure processes were evaluated. Three different phosphonium systems were studied in combination with five different resins and two different catalyst systems. Many of the formulations failed to give sufficient fixation to produce satisfactory flame resistance, but several were found which showed considerable promise. Representative results are shown in Tables CVII I and CIX. The melamine resin by itself appeared to give sufficiently effective fixation to allow the more easily finishable fabrics to pass FF 3-71; however, with the more finish-resistant poplin it was necessary to include both the melamine and urea to produce a durable treatment capable of passing the vertical test after 50 home launderings. Unfortunately, these finishes with the high resin contents yield fabrics which are significantly stiffened. Attempts to alleviate this problem with a variety of standard softeners have not proved successful to date. Better results appear to be obtained using reactive silicone softeners. Incorporation of Dow-Corning 1111 Emulsion^ into the formulation produced a significant improvement in the fabric hand but detracted somewhat from the flame resistance. It would also appear tht the hand of these fabrics could be im- 381 TABLE CVIII Pad Bath Formulation I for Phosphonium Condensate and P(DBPA) Phosphonium Condensate (1) 300 gm/1 TMM (2) 100 gm/1 Urea 100 gm/1 Amine Hydrochloride (3) 20 gm/1 P(DBPA) (4) 350 gm/1 Wetting Agent (5) 10 gm/1 50/50 popl in 50/50 denim 65/35 twill Wet Pick-up 113% 69% 71% Tensile (warp) 100 175 1.9 Tear (warp) 3 4.6 5.5 Wrinkle Recovery (w&f) 223 --- --- Char Length (in) initial warp 3.3 0.8 1.3 fill 5.6 0.7 0.4 50 HL, warp 2.6 0.8 0.9 fill 3.1 0.8 1.1 After Flame, sec initial warp 18 fill 19 50 HL, warp 16 fill 18 (1) Fireaway 2® (2) Valmel 40^ m Valcat AH0v9 (4) Dur-o-' cryl BL-1 (5) Valdet 4016® 382 PDOSpnoniUIII luimcnoo^ v ■ / TMM (2) 100 gm/1 Amine Hydrochloride (3) 20 gm/1 Wetting Agent (4) 10 gm/1 P(DBPA) (5) 350 gm/1 50/50 poplin 50/50 twill 65/35 twill % Wet Pick-up 114% 69% 70% Tensile (warp) 99 214 219 Tear (warp) 3.7 9.5 7.9 Wrinkle Recovery (w&f) 267 --- — Char Length (in) initial warp 3.5 1.0 1.2 fill 5.7 1.3 2.4 50 HL, warp BEL 1.4 1.9 fill BEL 1.0 1.3 After Flame, sec initial warp 25 1.0 1.0 fill 7 0. 12 50 HL, warp BEL 4 19 fill BEL 4 (1) Fireaway 2® (2) Valmel 40® (3) Valcat AHO® (4) Valdet 4016® (5) Dur-o-cryl BL-1® 383 proved by mechanical action. A scanning electron microscopy study showed significant fiber-fiber bonding resulting from the latex treat- ment. This type of bonding is frequently responsible for stiffening effects of surface finishes and can usually be markedly reduced by mechanical breaking. This and other factors affecting the commercial utility of the THPSvS)-urea-P(DBPA) finish will be the subject of a study to be initi ated in the near future at SRRC. 384 4. Systems Based on a Bromine-containing Phosphazene A somewhat different approach to incorporating phosphorus and bromine in one system has been taken by Sandoz Colors and Chemicals who have developed a flame retardant based on a bromine-containing phos- phazene. This material, when applied to polyester/cotton blends in conjunction with a resin such as hexamethoxymethylol melamine, has been found to produce an effective and durable treatment. In this form the treatment was found to impart a stiff hand to the fabric. Although this hand would be improved by mechanical action, a series of experiments was undertaken in cooperation with Sandoz in an attempt to find a non-mechanical solution to the hand problem. Electron micrographs of fabrics treated with the Sandoz 1030 finish with and without resin indicated that the hand problem is re- lated to the physical properties of the film it forms on the individual fibers during curing. There was no evidence in the micrographs of significant fiber-fiber bonding, even with the resin present. Conse- quently efforts were focused on modifying the physical properties of this film. The Dow-Corning silicone emulsions, which are also film formers, were found to offer some promise in solving these hand problems by plasticizing the flame-retardant films. It was found that a rather heavy loading of Dow-Corning llll^emulsion gave a good soft hand to fabrics prepared in the laboratory and line dried or tumble dried. However, unexpected difficulties were encountered with fabrics prepared in the laboratory but dried and cured under conditions similar to those which might be expected in the pilot plant. Fabrics dried under these conditions (150°C for 2 min. in the Dispatch oven) were essentially as stiff as those prepared without the silicone. This phenomenon is completely reversible and the stiff fabrics can be softened by wetting thoroughly followed by line drying or tumble drying. Similarly, the soft fabrics can be stiffened by subjection to heat in the form of oven drying or ironing (dry or steam using the durable press setting). Since it was thought that this problem might be the result of overdrying and loss of moisture which could act as a 385 plasticizer for the fiber coating. However, experiments in which treated fabric samples were dried down to different moisture contents at various temperatures indicated that this was not the case. In an attempt to define the origin of the fabric stiffness differential scanning calorimetry was performed on a series of cast films containing the Sandoz 1030/190; the Sandoz 130/190 plus resin and accelerator, the Sandoz 1030/190 plus resin and accelerator and Dow Coring lin^emulsion, and the Sandoz 1030/190 plus resin and accelera- nt Cr) tor and Dow Corning llll Viy emulsion and Trycol w QP-407 (an ethylene oxide-actapheno derivative). The films were cast from formulations, similar to those which might be used in an actual application. The films were dried overnight at 40 C in a forced air oven and then cured at 120 C for 4 hours. The thermograms for all of the above films were \/ery similar. This is demonstrated by Figure 113 where the DSC for the 1030/190 plus resin and accelerator is compared to that obtained from the film with the DC-im®and the Trycol®0P-407. All of the films were character- ized by a broad endotherm at = 60-80 C followed by the onset of the exothermic decomposition of the phosphazene at = 180°C. The assign- ment of these thermal transitions as melting and decomposition re- spectively was supported by thermogravimetric analysis. This is shown by Figure 114 where no weight loss is observed in the region 60-80°C whereas the major weight loss begins at = 180°C. Despite the fact that the thermal responses of the films showed no significant differences, their physical properties were quite dis- similar. The film containing the 1030/190 plus resin and accelerator was very hard, clear and brittle. When the DC-llll^was added to the formulation the film became opaque, was not quite as hard and was much easier to chip and powder. With both the DC-llll V£y and Trycol vjy OP-407 present, the film became rubbery and opaque. Not surprizingly when the formulation containing the DC-llll^and Trycol ^OP-407 was 2 applied to a 7.5 oz/yd brushed denim a significant improvement in hand was observed after the cure. However, the highly water soluble Trycol ^ OP-407 was lost in the afterwash resulting in a fabric with the same aesthetics as was obtained by using the DC-1 111 ^formulation 386 o o CM o •X'' S/i •J-S r™ 3 «/) o ,;.;, CM J- r— <_> O ( _j » CO t- Q o •X5 *= O o 00 o o -a c fO 00 «+- o c o •r— s- Q. E o o CO UJ Of CD 387 o i- O +-> - >- >- >- 1 1 3 Q <*- o to a, *r- +J J- 0) CL s- o o i. +J a. E «C x: +-> X! c o «o O c t/> °^ 1 > to ^J •^ x £n -o ■»-> ■o •^" CO S- XI 3 U T3 4- cu J* o V. X> ISt OJ -* J* -t-> -t-> -t-> r— ^~ 3 3 O Qt a. *»- <4- 4-> O 4-> 4J O o 00 t/i c c ai •M ■»-> •r- s- a> 0) X) >> "Zj *J >> J= -t-> t- X> en o> •r— C7» •r— a> Xt 3 3 u 3 t_ > 3 o o J3 O XI „ i- ■»-> ■M A +-> » 0J m * * QJ *• i- 3 s- i- S_ 3 J- T3 CT T3 f0 0) r— CL s ^~ ) — Q. ^-» U o O o o O o O o c a> (_ x o ai Qu a> t. c •I-) c n3 c a> o to o 4-> o c •1— 1T3 (J U o OJ to O •r* O •^ . +-> ^~ r— r— M- 3 "8 ■^ "S •r» O B to l^ oo ul ■M +J iO 0) fO a> <_> U c > i— > •f— •^ 0J •r— >> •r— C c CT ■M X ■M o o o • O o U •f— •^ t- 1 » I t- (V J- 1 E . x: >£> r^ O O CO LO o O i CM **• LO o 1 © a. 0( ^ one® one® o © r— fmm CM +-> •m X fM o o •r- »f« o a» i— <-> 00 u s- r-» C 1 >> 1 X) Xi >> £ £ 1— Q «o u. C •r- i- a> T3 i O o Lf> 390 o 00 0J 4-> c o o o • S- V£> CQ ■♦-> 4-> o c o o o (_> ^. 4-> **^ H- +J U_ LU O O Q. O O O U_ 1 co O CM — o o X r— a> N t £ o «* "•««. -o o C_) C Q. Q. 5 <4- O o o •1— CSJ 4-> U o o CM CO o o 00 CM o o CNJ o o o o o o o V£> CM CM r— ^~ w6/ieD « l HV 10 rO O CO O N o -o c «a oo CD to 'cu s- +-> n CD JO <& on CJ3 392 5. Systems Amenable to Fixation by Irradiation Some preliminary screening experiments were run at Hooker to determine which of the six previously described, specially synthesized monomers could be effectively homopolymerized on the 50 cotton/50 polyester ETIP standard fabric- Free radical polymerization of monomer was initiated in air on fabric by electron beam radiation. Also used were thermal and photochemical decomposition of 2-azo-bis-isobutyro- nitrile (AIBN) in both air and nitrogen atmospheres. Fixation of N-(dimethylphosphonomethyl)acryl amide (NDPA) and dimethyl phosphonomethyl acrylate (DPA) was observed in these experi- ments. The other monomers failed to homopolymerize under these condi- tions. Table CXI shows data obtained with NDPA and DPA using electron beam radiation and photochemical decomposition of AIBN in a nitrogen atmosphere. These data show that higher electron beam radiation doses result in higher fixation efficiencies for both NDPA and DPA; fixation of NDPA is more efficient than DPA at the 24 Mrad dose level; fixation of DPA is more efficient than NDPA with photochemical decomposition of AIBN in a nitrogen atmosphere; and wash durability is excellent for NDPA and DPA derived finishes which were fixed by electron beam radiation. It should be noted that untreated blend fabric after 50 washes in high phosphate detergent analyzes for about 0.4 ± 0.1% phosphorus using an X-ray method. This may be a result of inorganic phosphorus residues derived from the detergent, or it may result from interference by calcium that is present in mineral deposits which accumulate during washing in hard water, or possibly a combination of these two factors. Accordingly, it is probably justifiable to think of the phosphorus values at 50 washes as being inflated by this amount. Not shown in Table CXI are results from the other three conditions under which AIBN was decomposed. Its photochemical decomposition in air resulted in no fixation with any monomer. Thermal AIBN decom- position in air resulted in no apparent fixation of DPA but the highly polymerizable NDPA was fixed in an estimated efficiency of 22% based 393 a. < »t in L/i 10 1/1 O O O O O ' rH + + + + + I TJ ^^-Tt WU1N o o o o o o o + + + + + + + HHMNNHtx i O O O O O O O + + + + + + I fO f-1 LT1 T* «* »* (Nl I i— I o o o o o o + + + + + + I o I/) ul '.O CO Ol O 00 i c-1 O O O i—l i i—l Ul CJnO Ol CO o t CM O O O O <-4 CM LO tO O i-H ul *© i— I CM O O -H O O tO to u"> r-» o z o HHMt invoco cm o o o o o cm H l/l N Ifl K1 Ul N i CM O O O O O rg * 00 i-l © O © © © <-"J CJ 2 < H 2 c u (- CQ X Q u z < w -a a. < ft H C :.' c I— I f- < X CO c 3 o a, [/) Xi ft "0 a. < uionioocoo> 00 U1 H O) 't CO H N W M (N •<» fl t © > < Ol, Oh ft c ft > > < a. < oq s: < ft > s z z q a a a a o ft «J < Ol, 0- ft go, > > < ft, < °a z s z 2 Q O Q Q Q © © o u. «J< ft, ft, ft, r §ft > >s r ^2 Q Q Q Q Q £" ft cc o OS f- u w ,-J UJ !-, DJ 6 o c o e f-1 < a, CO a, CQ H 2 CO : H as ft CQ = H ■H ■iH < 2 cm: Z M X) a) X : : fti : w : : 394 c o o CQ 2 X o O i— I O O O O i o no t ^10 i— iro< v J< v Jcr>oo'*cri Or-IOOOOOO OrHOOOOOO t I I t I i I I Ki^o^Ta-^ooor^c-J oooooooo + OKHOHNlft l-4f-HOr-l.-lr-l.-IO OOOOrHi— lr-4i— I oi>»ooooo O i-t O O i-H c/3 os w z o 35 OOOOOr-^tvjmo NlN00MOlH>Oi/l OOtO^OoOMO N M N r- I .— I (SI a < OlUlOlHOOHWOC \OHNONl/HtN i— I CO C < c •H e o u CQ a> E o c o s Sh 0Q © mo a> < a, a> a.*^ c a. > > < o. < o Z ZQQQQQ>^ UU *> < a. a, Ca.>>s? ZzSqqqo^ a. gCu>> > < a, < a rz < a, > «g r z a q q a o x < a. m H •H < DO H ■H < a. « r 3S a, CQ H Z Z f« O u o • rH .H *-> >- co ca <—i u- a. e TJ O C 4-> CO •<-> O 01 a: 132° 210° 347° 1 cc c?^ 4-» 3 k FIGURE 118. TGA data on NDPA and BDPOM homopolymer grafts on 50/50 PET/cotton. 406 their flame retardant efficiencies (see Table CXVII). This was pre- sumably due to the deposition of much of the flame retardant on the .surface of the polyester where it would be more subject to rapid bro- mine loss during burning and thus have a tendency to lose much of the inhibitor species prior to the flame front. Although these effects were significant they did not seem to be of such a magnitude as to preclude utilization of direct application procedures without preswelling. Thus, a series of experiments using DBPA in combination with three different phosphorus monomers were carried out. The treated 50/50 blend fabrics were prepared at Clemson and irradiated with a dose of 1.5 Mrads using the electron beam equipment at Deering Mil liken Research Corporation. Evaluation by 01 gave the results listed in Table CXVIII. Although the data are quite limited, and there is considerable scatter in the results, it appears that the most efficient phosphorus source for this system is the NDPA. A parallel series of experiments was carried out at RTI using emulsions of the flame retardant monomers rather than solutions. After padding with the emulsions, the fabrics were subjected to heating at 170 C for two minutes and then to irradiation in an electron accelera- tor. The results of these experiments are shown in Table CXIX. In all cases the phosphorus compound used with NDPA in a 1:1 weight ratio of the bromine containing retardants. The data showed considerable scatter which was probably due to volatilization of some of the bro- mine-containing monomer from the fabric during drying, but in general, this technique was found to result in an adequate level of add-on and encouraging 01 values. Scanning electron - X-ray microprobe analysis of the graft from an emulsion of NDPA and TBPM on 50/50 PET/Cotton blend fabrics showed that it was possible to graft NDPA not only to cotton but also PET. Some of the data are recorded in Table CXX. More detailed analysis showed that there was a minimal, if any, copolymer formation between the NDPA and TBPM. Preliminary experiments indicated that the durability of the grafted samples prepared using the emulsions might be a function of both the application procedure and the grafting conditions. For this 407 CO UJ C_> LU t— < O <: Q cc\ Q UJ —I CD 00 X C_> ■2? c +-> r- 13 OJ • O 3 •— I JZ (/I . +J ♦ r— • 3 CD T3 O 3 -O -C a> ■r- s_ &? 5 Ql a> c •— o <— i o S- +-> CO «J o o s- »— Q IE CO LO r— CO CM lo O CM CM *d" CO CO CO CO C\J CM LO CO LO LO cr> CO CM «3- CO co r*. CO CO CO CO CO CM CM LO LO «3- CO CM LO CO o CT> CM CO cr> oo «3- CO CO 1^ CD O Q-l E O CO CD CD CM O LO CO LO CM CM CO CO CO LO *3" «3- LO CM CO o LO o o o LO o o o LO o o o •— O i— O r— ^ co - Q. X _l a o ^ LU - o o •r— U •r— o> a 4J c O o 0) CO (V "O CO «3 3 CO C +J o o •r" S- •^» U) o 4J C U o CD _J w CM CO CO 00 CO co LT) *£> VD 1— *a- -Q +-> •r- o Ll. o s- a> +-> CO >■> 411 reason, the durability has been studied in considerable detail. Most of the samples were laundered for 25 cycles. The samples were washed according to AATCC Test Method 124-1967 in a Sears Kenmore®600 washer, normal cycle, low water level, with a 3 lb dummy load and the water at o approximately 130 F. A 12-minute washing cycle was used with 90 grams of AATCC standard detergent in each cycle. The samples were dried in a closed dryer after the 10th and 25th cycles only and were then evaluated for oxygen index. One series of samples was prepared by applying the neat monomer mixtures to the fabrics prior to irradiation. The results of these are listed in Table CXXI. All of the samples except the NDPA/BABA 50 showed a significant problem with durability. In most cases, the first 10 cycles were effective in removing a significant portion of the flame retardant. Beyond this there appeared to be little change through 25 cycles. A second set of samples was prepared from a 50% aqueous monomer emulsion and irradiated in a nitrogen atmosphere. The results obtained with these samples are given in Table CXXI I. As can be seen, these samples exhibited significantly higher durability than those prepared using neat monomer application. Similar results were obtained on samples padded from emulsion but irradiated in an air atmosphere (Table CXXI II). This is particularly significant in terms of the applicability of these treatments in commercial processes. The effect of application and fixation conditions are compared in Table CXXIV. On the basis of these results there would seem to be little reason to expend extra effort to protect the fabric from air during irradiation. These results also indicate that application from aqueous emulsion should be easier to control although the appli- cation of the neat monomers should be feasible using printing methods or other techniques to moderate the add-ons. Other application parameters were also studied. Because of the reduced retardant efficiency observed when the methanol preswelling was omitted from the application procedure, attempts were made to find an acceptable alternate pre- treatment method which would result in 412 TABLE CXXI DURABILITY OF FABRICS GRAFTED USING NEAT MONOMER MIXTURES Composition/Total Dose % Add on 01 Before Washing 01 After 2 Cycles 01 After 10 Cycles 01 After 25 Cycles NDPA/H 6 /0.5 Mrads 8 22.7 28.6 21.4 1 16 24.3 30.1 22.7 2 17 27.1 30.5 23.5 NDPA/TBPM/0.5 23 28.6 29.0 22.2 22.2 1 43 30.5 30.1 23.3 22.7 2 49 31.4 31.4 24.7 24.3 NDPA/BABA/0.5 40 26.0 28.3 28.6 28.3 1 44 29.4 32.5 31.1 31.1 2 47 30.9 32.5 31.8 31.4 NDPA/FyrolCiyBB/0.5 49 30.5 24.7 25.5 25.2 1 63 32.1 26.7 26.7 26.3 2 (51) 32.5 28.6 28.3 27.6 NDPA/BDPOM/0.5 37 32.1 33.1 22.7 22.2 1 56 34.1 34.4 23.5 23.5 2 66 34.6 35.6 24.3 24.0 NDPA/DBPA/0.5 68 32.8 34.1 28.3 27.9 1 79 35.0 35.0 31.4 31.4 2 93 36.6 37.8 33.5 31.8 413 TABLE CXXII DURABILITY OF FABRICS GRAFTED UNDER N„ USING MONOMER EMULSIONS Composition/ Total Dose % Add-on 01 Before Washing 01 After 10 cycles 01 After 25 cycles NDPA/DBPA/1 • 28 29.0 29.0 28.3 60 40 2 26 29.7 29.0 27.9 4 41 29.7 30.1 30.1 NDPA/DBPA/1 28 27.1 27.9 27.1 40 60 2 28 27.9 29.0 28.3 4 25 28.3 29.7 28.6 NDPA/FyrolWi 3B/1 19 29.0 26.3 26.0 60 40 2 24 29.7 26.7 26.0 4 32 30.5 28.6 27.1 NDPA/FyrolvB/l 3B/1 18 28.3 26.3 25.2 40 60 2 13 28.6 26.3 24.7 4 ■ 15 28.6 26.3 24.3 NDPA/TBPM/1 25 28.6 28.6 27.6 60 40 2 24 29.0 29.0 27.6 4 25 29.0 29.4 27.9 NDPA/TBPM/1 18 27.6 27.1 26.3 40 60 2 18 27.9 27.9 26.3 4 21 28.3 27.9 26.3 NDPA (50%)/l 37 29.7 27.9 27.1 2 42 30.9 29.0 27.9 4 47 30.9 29.0 28.3 414 TABLE CXXIII DURABILITY OF FABRICS GRAFTED IN AIR USING MONOMER EM ULSIONS Composition/ Total Dose % Add-on 01 Before Washing 01 After 10 cycles After 25 cycles NDPA/DBPA/1 60 40 2 4 30 35 44 29.0 28.6 31.1 27.6 28.6 30.1 29.0 31.1 31.4 70-30/1 2 4 26 28 37 29.7 29.4 30.5 28.3 28.3 29.0 28.6 29.4 29.0 80-20/1 2 4 39 56 30.9 31.1 33.1 29.0 29.4 30.5 29.0 30.1 30.1 NDPA/Fyrol^«B/DBPA/l 50 25 25 2 4 34 31 31.8 31.8 32.5 30.1 29.7 29.4 29.4 29.4 29.0 60-20-20/1 2 4 38 30 35 28.3 28.6 30.5 27.9 28.6 29.7 27.9 29.7 29.4 70-15-15/1 2 4 42 31 32 31.4 31.8 31.8 29.4 29.4 30.1 30.1 30.1 29.7 NDPA/TBPM/1 60 40 2 4 32 40 (68) 29.7 30.1 30.1 29.0 29.0 30.1 29.7 29.4 29.4 70-30/1 2 4 36 56 28.6 29.7 30.9 29.4 30.1 31.1 29.0 30.5 30.9 NDPA (25%)/l 2 4 19 17 16 27.9 26.3 26.3 26.7 26.7 26.3 27.6 27.6 26.3 415 o GO X X at CO <: to 3 I— Li. (-1 U- O t— O o (/> OJ L. ^-» 1—* QJ o O-M >. "4- O , vt- <_> <: in V)- o «£ to CM CTl 0) C u •p* •— i O JC o«*. l/l 0) rtJ CO 3 c o a* i "O T3 < 0! l/> O Q , 2 o 1- ^■^ c o •f— ■»-> •r- t/> o a o i o i— ■«*■ cri i— p- CM CO CO I O VO >— I CM CO i— I MCMC1 CO CM r- I 00NO I CM CM CO I O f>« 1^ i cm cm cm I CM CM CM I CO VO •— cr> «* co fx I— I— I CM CO co • co o v© cm in vo CO CO CO CO CM CO I Q. O O CO UO "* Q < o. o o Q ID VO I CM CM CM I CM CM CM I 1-^ r— i— Ol O O CM CO CO CM O 00 CO «* VO I VO VO cn I VO O O CO CM CM CM CM CM I IT) *»■ If) I CM CM CM CM CM ^" CM CM CM VO U1 *t I OOOr- CM CO CO CO CO CM CM ■* "» VOvOls CM CM CM Orsifi I CM CM O I CM CM CO I r— CM CO CM CO VO I in vo i^ CM CM CM O CM CM CO CO CO CM CO i— I **• VO in I LT) CD CO CO ' •— CM *T too co m *»■ Q. OO q in vo e o * 4. O O >>m *»• a. o o o in vo r^ r^ vo CM CM CM K VO VO CM CM CM r-» f~» CO CM CM CM f-» CM CM CM O O CM CO CO co ■* «* a. Q 416 higher efficiency of the grafted flame retardant but still be compati- ble with commerical processing requirements. One of these which looks particularly promising is based on the use of liquid methyl amine. This seemed to result in both an increase in the flame retardant efficiency as measured by 01, and a slight increase in the flame re- tardant add-on. The effect is most pronounced in the case of 100% cotton, presumably because of methylamine's ability to de-crystallize cellulose. An example of this effect can be seen in the data presented in Table CXXV. The effect of varying the nature of the source was also studied. A significant reduction had been noted in changing from Co to the scanned electron beam (see Table CXXVI for typical data). This, how- ever, may be dependent to some degree on the structure of the monomers since Fyrol BB seems to exhibit essentially the same grafting efficiency with both radiation sources. More importantly, the NDPA systems seem to graft more efficiently in the lower energy sources. Such effects could be wery important, particularly if the flame retardant systems developed with one radiation source are to be applied using a different source. As an example, there is considerable interest in the textile industry in the El ectrocurtain^ processor developed by Energy Sciences, Inc. This equipment operates at lower energy than the scanned electron beam source. Thus attempts have been made to evaluate some of the co- polymer graft treatments using the Electrocurtain w . Typical results are presented in Tables CXXVI I and CXXVIII. These results are en- couraging but indicate the need for additional studies to optimize the application and fixation techniques. In an alternate approach to the application problem, experiments were carried out to determine the feasibility of applying the monomers by conventional drying techniques. Samples of NDPA, DBPA and TBPOEA (tribromophenoxyethyl acrylate) were applied to cotton fabric using procedures for the application of direct dyes but none of the three were found to exhibit any significant affinity and thus only wery low pick-ups were achieved. The TBPEOA was used in these studies in place of the TBPM since it contained similar bromine functionality but 417 X X <_> o Z Q e£ C3 <: o_ CO Q Q. OJ CD C O 8 c CD E 4-> fQ CD S- h- I OJ s- c o I -a at _i >- o I— cu to Q 10 r— 03 i— ro OJ CO CO ho co CO oo LO OJ r— OJ CO CO LO to r— to 00 !*>■ to to oo o to CO OO LO co cr> CO OO co «d- to r*-» en LO OvJ O LO O LO O OJ LO o LO O OJ LO o o O i— O r— OJ c o +J 4-> o cu o f~" Q- s-s E o fO o CO 1— s- a» CO cu o D_ O o c o +J 4J o o LO o LO 418 TABLE CXXVI COMPARISON OF GRAFTING SOURCES MONOMER COMPOSITION SOURCE DOSE(Mrads) %ADD-ON oi FyroV^BB 2% BABA-50 Y Y 1.95 4.3 6 7 - 7.95 9 - e~ beam 2 6 - e~ beam 4 6 - e~ beam 8 7.5 - DAVP/ZnCl 2 2% BABA-50 Y Y 1.95 4.3 8 9 — _Y 7.95 10 _ e_beam 2 _ e_beam 4 __ e'beam 8 0.4 - BABA-50 0.7 45 _ e'beam 1 3 - NDPA^ Y 1 20.4 Y 5 3 20.8 e~beam 1 19 27.9 e"beam 4 16 26.3 e'curtain 1 6 24.7 e"curtain 5 10 25.5 NDPA/FyroWBB 55 45 (2) Y Y 1 5 3 10 22.7 25.5 e"beam 1 19 29.0 e'beam 4 32 30.5 e~curtain 1 12 26.3 e'curtain 5 19 28.6 (1 ) Solution (2) Emulsion 419 TABLE CXXVII SELECTED FORMULATION FOR e" CURTAIN EXPERIMENTS Formulations NDPA - 33% H 2 solution NDPA/DBPA - 50% H 2 emulsion, 4% Triton X-10 d© 40 60 NDPA/DBPA/TBPM - 43% H 2 /DCE emulsion, 3% Triton X-10 40 30 30 3.3 1 M}R- 50% H^O emulsion. 4°/ Trit.nn X-100^ NDPA/DBPA/FyroN^BB- 50% H 2 emulsion, 4% Triton X-l 40 30 30 Sample Size - 18 in. x 72 in. Wet add on - 120 - 160% Predrying - 94°C (200°F), 7 mins., forced air NDPA/DBPA NDPA/DBPA/TBPM NDPA/DBPA/Fyrol^BB Total Dose - 1 Mrad and 5 Mrads Atmosphere - essentially ambient air 420 TABLE CXXVIII DURABILITY OF SAMPLES GRAFTED IN AN e" CURTAIN 01 01 Sample/Total Dose % Add on After 1 Washing After 10 Washinq Ungrafted - 1 18.5 - 5 — 19.2 NDPA - 1 6 24.7 24.3 - 5 10 25.5 25.5 NDPA/DBPA - 1 11 25.5 24.7 - 5 17 26.3 26.3 NDPA/DBPA/TBPM - 1 8 24.3 22.2 - 5 11 25.5 23.5 NDPA/DBPA/BB - 1 12 26.3 26.0 - 5 19 28.6 27.6 421 appeared to be more readily available on a commercial scale. Similarly, application of the retardants to polyester was attempted using disperse dye procedures with and without carriers. Considerably better results were obtained in these experiments as shown in Table CXXIX These samples were fixed by irradiation using the Electrocurtain va/ . Attempts to use this same approach for the treatment of 50/50 blends was less successful as shown by the data in Table CXXX. Only the TBPOEA appeared to be a good candidate for dyebath applica- tion. All dyeings were carried out in the presence of both carrier and NaCl in an attempt to maximum exhaustion into both fibers. Al- though the TBPOEA appeared to exhaust almost completely onto both 100% PET and the 50/50 blend, examination of the treated fabrics indicated that very little of the retardant had penetrated the fibers and most of it was present as a coating on the fabric. This caused significant problems upon attempts at further processing. Attempts to pad an NDPA solution onto the TBPOEA treated fabric followed by radiation curing led to a finish with poor flame retardant efficiency due to the inability of the NDPA to penetrate the TBPOEA coated fibers. Attempts to apply the NDPA prior to the TBPOEA dyeing failed because of the extraction of the NDPA from the fabric if previously unfixed and degradation during high pressure dyeing of the NDPA treated fabric when previously fixed. Efforts to apply the two monomers together in a padding operation also failed because of the difficulties encountered in trying to prepare a stable pad bath with a sufficiently high con- centration of TBPOEA. In still a different approach to the application and fixation problems the reactive bromine containing monomers were replaced by the essentially inert and insoluble decabromodi phenyl oxide in the form of the commercial FR P-53^ system. The NDPA was then cured using the scanned electron beam to give an apparently durable finish in which the NDPA acts as the binder for the P-53. The results of these preliminary experiments are given in Table CXXI. Unfortunately these data give little information about the effectiveness of the finish since the fixation was much better than expected and thus the add-on levels were extremely high. 422 TABLE CXXIX DYEBATH APPLICATION OF FLAME RETARDANTS TO PET Monomer Carrier % Add-on NDPA NDPA DBPA DBPA TBPOEA benzene none benzene none benzene 2.6 9.3 8.4 17.6 All samples dyed 2 hours @ 100°C, extracted, dried subjected to 5 Mrad irradiation, and washed. 423 QQ O LT> O m cc X LU X ^~ X o o 2H O LU ^r* _J CD C£ - Q o oo o C\J 00 CO s- (V - CD or 0) S- i— ' •— > >, LO C\J Oj J- r— •— i a» o o +-> >, ^ o <: o 01 a> c CT. 00 00 $_ -r- • • >— O -C ,— CM «+- «/> CO CO oo 43 See: MCC-100^ Phosphoric acid 33-42, 52, 54, 91-92, 242, 258, 282, 363 Pittsburg Plate Glass Ind. 23 Poly(l ,4-cyclohexylene dimethyl eneterephthal ate) (PCHDT) 44-47 Poly(2,3-dibromopropylacrylate) 317 Polytechnic Institute of New York 12, 120, 143 Polyvinyl bromide (PVBr) 79-80, 91, 130, 134, 356-366, 370, 375, 377, 378, 397, 398) Polyvinylbromide/polyvinylchloride copolymer P(VBr/VCl) 91, 93, 104-107, 356-366, 370, 375, 377-380 Precondensate See: oligomer PVBr See: polyvinyl bromide P(VBr/VCl) See: Polyvinyl bromide/poly vinyl chloride copolymer Pyrolysis 33, 35, 87-88, 131-133 Pyroset®TK-115 116 Pyroset^TKS 114 Pyrovatex®3762 75-76, 84, 86, 108, 246, 250, 255, 256, 271, 272 ■ ®, Pyrovatex^CP 104, 106, 240, 246, 249-250, 255-256, 272, 378 448 Rate of heat release 53-54, 63, 98, 100, 270-276 Research Triangle Institute 14, 24, 127, 177 Riegel Textile Corp. 23 Sandoz 1030 385-392 Sandoz Colors & Chemicals 23-24, 385 Scanning electron micrographs (SEM) 183-189, 385, 407 SEM See: scanning electron micrographs Silicone softeners, reactive See: Dow Corning Emulsion 1111 Sodium allylate 243 Southern Regional Res Center 13, 79-80, 243, 257, 270, 273, 312, 317, 356, 359-360, 384 Springs Mills, Inc. 23, 360 Static oxygen bomb 37, 44, 52, 84, 89-90, 223-225, 229, 232, 246, 249, 268, 270-276 Stauffer Chem 145 J. P. Stevens & Co. 23 Sun Chemical Co. 23 Tanatard®DN-2 118-120 Chas. Tanner 23, 25, 317, 378 TBBPA See: tetrabromobisphenol A TBPA See: 2,4,6-tribromophenylacrylate TBPM See: 2,4,6-tribromophenylmethacrylate 449 TBPOEA See : 2 ,4 , 6-tri bromophenoxyethy 1 acryl ate TBPOEMA See: 2, 4, 6-tri bromophenoxyethy Imethacry late TBPP See: tris(2,3-dibromopropyl ) phosphate Tennessee Eastman Co. 23 Terephthalic acid 58 Tetrabromobisphenol-A 69 Tetrabromobisphenol A bis(2,3-dibromopropyl ) 129-130 Tetrabromobisphenol A bis(hydroxyethyl )ether 127 Tetrakishydroxymethylphosphonium chloride See: THPC® Tetrakishydroxymethylphosphonium sulfate See: THPS® TGA See: thermogravimetric analysis Thermal stability of bromine flame retardants 208-209 Thermal stability of polymer units 211 Thermogravimetric analysis (TGA) 37,60, 120, 122-124, 139, 141, 149, 197, 229, 235-239, 376, 386, 388, 400, 405-406 THP See: trihydroxymethylphosphine THPC® 75, 243, 257-260, 266, 272, 289, 321, 356-366, 375-381 THPC *&7 amide 70, 76-79, 114 THPcQy-APO 71 _ 72 THPC ^/Diamine 108 (r) THPC^^/hexaallylaminocyclotriphosphazene 245 450 THPC^/hexamethylaminocyclotriphosphazene 245 THPC®/MCC-100® 108, 259, 263, 265-266, 268-269, 273, 277-279, 285-288, 290, 296 THPC®/MCC-100/200/300® 110, 273, 278-279, 285, 288, 295 THPC^ / -McNH 2 108, 259, 262, 265, 269, 275-276, 282, 284-285 THPC ® /Pyrovatex^ 401 3 1 08 THPC^/tetrachlorodiaminocyclotriphosphazene 245 THPC®-urea 79-80, 91, 93, 114, 255-256, 270, 272-273, 283 THPOH 261, 265, 268 THPOH/amide 71, 73 THPOH/ammonia 38-39, 52, 71, 74, 91-92, 110, 114, 363 THPOH/MCC-100® 108, 259, 264-266, 269, 274, 294 THPS® 208, 266, 289 THPS® /carbamate/200 301-302 THPS®/MCC-100® 283, 291, 293, 298-300 THPS®/urea 356, 378-379, 384 Toyobo, Inc. 23-24, 289 T23P See: tris(2,3-dibromopropyl ) phosphate TPP See: tri phenyl phosphate TPPO See: tri phenyl phosphineoxide Triallylphosphate 243, 328 Triazaphosphaadamantane 242 2,4,6-tribromophenol 130 2,4,6-tribromophenoxyethylacrylate (TBPOEA) 317, 378, 393-425 451 2,4,6-tribromophenoxyethylmethacrylate (TBPOEMA) 327 2,4,6-tribromophenylacrylate (TBPA) 394-425 2,4,6-tribromophenylmethacrylate (TBPM) 394-425 Trihydroxymethylphosphine (THP) 259 Trimethyl phosphorannde See: MCC-100® Tri phenyl phosphate (TPP) 44-48 Triphenylphosphine oxide (TPPO) 37, 39, 42, 60, 62, 65 Tris See: tris ( 2, 3-di bromopropyl ) phosphate Tris(2,3-dibromopropyl)phosphate (TBPP) (T23P) (Tris) 44-48, 70, 75, 78-79, 84-85, 91, 93, 328, 329, 334, 375 Union Carbide Corp. 23 United Merchants 25, 317, 358, 378 Univ. of Maryland 15, 24, 268, 356 VBr 2 See: vinyl idenebromide Vertical test 77, 81-82, 104, 113-120, 174, 208, 212, 253, 256, 285, 290-301, 312, 317, 378, 380-383, 393 Vinyl benzoate 57 Vinylidene bromide (VBrJ 378, 401 West Point Pepperell Co. 23 White Chemical Co. 24, 143, 19.3, 312 X-ray analysis 184, 187-189, 308, 310 452 NBS-114A (REV. 7-73) U.S. DEPT. OF COMM. BIBLIOGRAPHIC DATA SHEET 1. PUBLICATION OR REPORT NO. NBS-GCR-ETIP 76-22 2. Gov't Accession No. 3. Recipient's Accession No. 4. TITLE AND SUBTITLE DEVELOPMENT OF FLAME RETARDANTS FOR POLYESTER/COTTON BLENDS 5. Publication Date September 19 76 6. Performing Organization Code 7. AUTHOR(S) Robert H. Barker and Michael J. Drews 8. Performing Organ. Report No. 9. PERFORMING ORGANIZATION NAME AND ADDRESS Department of Textiles Clemson University Clemson, SC 2 9631 10. Project/Task/Work Unit No. 1150066 11. Contract/Grant No. 4-35963 12. Sponsoring Organization Name and Complete Address (Street, City, State, ZIP) Experimental Technology Incentives Program National Bureau of Standards Department of Commerce Washington, D.C. 20234 13. Type of Report & Period Covered Final 14. Sponsoring Agency Code 15. SUPPLEMENTARY NOTES 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.) Initial studies were carried out to determine the flame retardant characteristics of the individual polyester and cotton fibers. Since previous studies had dealt primarily with cotton and other cellulosic materials, emphasis was placed on the polyester. Structural and chemical factors affecting flammability were determined and evaluated. The inter- action of the fibers was studied in both the presence and absence of flame retardants. Several specific types of phosphorus and bromine-containing materials were evaluated to determine their relative retardant efficien- cies. Various methods of fixing flame retardants onto polyester were also studied with particular emphasis placed on radiation grafting tech- niques. Based on the results of these studies several series of model flame retardant treatments were prepared and evaluated on 50/50 blend fabrics. The results of these studies were then used as the basis for designing new flame retardant systems having a potential for commercial application. These systems include phosphonium oligomers, a bromine- containing phosphazene and brominated aromatics with a brominated aery late binder. 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) Antimony oxide; bromine; bromine flame retardants- calorimetry; cotton; cotton/polyester; ETIP; fabric flammability; flame retardants; flame retardant mechanisms; flame retardant monomers; hand modifiers; inherently flame retardant polyester; phosphazenes ; phosphorus flame retardants; polyester; precondensates ; r a di a tion g raf ti ng 18. AVAILABILITY HX 1 Unlimited I 1 For Official Distribution. Do Not Release to NTIS 1 1 Order From Sup. of Doc, U.S. Government Printing Office Washington, D.C. 20402, SD Cat. No. CH IX 1 Order From National Technical Information Service (NTIS) Springfield, Virginia 22151 19. SECURITY CLASS (THIS REPORT) UNCLASSIFIED 20. SECURITY CLASS (THIS PAGE) UNCLASSIFIED 21. NO. OF PAGES 472 22. Price $12.00 ■SCOMM-DC 29042-P74 <*J.S. GOVERNMENT PRINTING OFFICE: 1976 626-791/9 14 1-3 p Tn*ilWf,i?m*s'Tf LIBRARIES A °°0070^022=j 7