This preview has intentionally blurred parts. Sign up to view the full document

View Full Document

Unformatted Document Excerpt

NOVOLAC/EPOXY CRESOL NETWORKS: SYNTHESIS, PROPERTIES, AND PROCESSABILITY by Sheng Lin-Gibson Dissertation submitted to the faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Chemistry Judy S. Riffle, Chair John J. Lesko James E. McGrath Allan R. Shultz Thomas C. Ward 12 April 2001 Blacksburg, Virginia Keywords: Cresol novolac, controlled molecular weight, phenolic, epoxy, structureproperty relationships, flame retardance, composite, latent catalyst Copyright 2001, Sheng Lin-Gibson Cresol Novolac/Epoxy Networks: Synthesis, Properties and Processability Sheng Lin-Gibson Abstract Void-free phenolic networks have been prepared by the reaction of phenolic novolac resins with various diepoxides. The stoichiometric ratio can be adjusted to achieve networks with good mechanical properties while maintaining excellent flame retardance. A series of linear, controlled molecular weight, 2,6-dimethylphenol endcapped cresol novolac resins have been synthesized and characterized. The molecular weight control was achieved by adjusting the stoichiometric ratio of cresol to 2,6dimethylphenol and using an excess of formaldehyde. A dynamic equilibrium reaction was proposed to occur which allowed the targeted molecular weight to be obtained. A 2000 g/mol ortho-cresol novolac resin was crosslinked by a diepoxide oligomer and by an epoxidized phenolic oligomer in defined weight ratios and the structureproperty relationships were investigated. The networks comprised of 60 or 70 weight percent cresol novolac exhibited improved fracture toughness, high glass transition temperatures, low water uptake, and good flame retardance. The molecular weights between crosslinks were also determined for these networks. The stress relaxation moduli were measured as a function of temperature near the glass transition temperatures. Crosslink densities as well as the ability to hydrogen bond affect the glassy moduli of these networks. Rheological measurements indicated that cresol novolac/epoxy mixtures have an increased processing window compared to phenolic novolac/epoxy mixtures. Maleimide functionalities were incorporated into cresol novolac oligomers, and these were crosslinked with bisphenol-A epoxy. The processability of oligomers containing thermally labile maleimides were limited to lower temperatures. However, sufficiently high molecular weight oligomers were necessary to obtain good network ii mechanical properties. Networks prepared from 1250 g/mol cresol novolac containing maleimide functionilities and epoxy exhibited good network properties and could be processed easily. Latent triphenylphosphine catalysts which are inert at processing temperatures (~140C) but possess significant catalytic activity at cure temperatures 180-220C were necessary for efficient composite fabrication using phenolic novolac/epoxy matrix resins. Both sequestered catalyst particles and sizings were investigated for this purpose. Phenolic novolac/epoxy mixtures containing sequestered catalysts exhibited significantly longer processing time windows than those containing free catalysts. The resins also showed accelerated reaction rates in the presence of sequestered catalysts at cure temperatures. Trihexylamine salt of a poly(amic acid) was sized onto reinforcing carbon fibers and the composite properties indicated that fast phenolic novolac/epoxy cure could be achieved in its presence. iii Acknowledgements I would like to dedicate this dissertation to my family; especially my loving husband Ben, who gave me constant love, support, and encouragement, and my parents Yin-Nian and Qin who instilled in me the important value of continued education and a strong work ethic. I am truly blessed to have such wonderful parents who dedicated their lives to bettering the lives of my brother, Dave, and I. I am grateful and fortunate to have an incredible committee with a great wealth of knowledge in polymer science and an undying devotion to the field. I express sincere gratitude to, and the utter most respect for, my adviser and mentor, Dr. Judy S. Riffle who opened my eyes to the world of polymer science. She is truly an inspiration to all of her students as well as a role model for women in science. She has provided me with both technical and personal guidance throughout my undergraduate and graduate studies. I am also deeply indebted to Dr. James E. McGrath who constantly provided insight on various aspects of polymer chemistry, Dr. Allan R. Shultz for his invaluable suggestions and comments related to my research, Dr. John J. Lesko for his guidance on composite properties, and Dr. Thomas C. Ward for his suggestions on the physical chemistry aspect of my research. I would also like to thank the other CASS faculty and staff members, especially Dr. John Dillard and Dr. Jim Wightman, who gave me the opportunity for undergraduate research here at Virginia Tech. I would like to thank my fellow graduate students in the McGrath group, the Lesko group, and the Poly-P-Chem group, and especially the Riffle group, for their advice and critical suggestions. I am grateful to the Riffle girls and Brian Starr with whom I developed valuable friendships throughout my years at Virginia Tech. I would particularly like to thank Angie and Mark Flynn for their invaluable assistance. Lastly, I would like to acknowledge the summer undergraduate students who assisted me in my research, Michael Shane Thompson and Vince Baranauskas. iv Contents Abstract............................................................................................................................... ii Acknowledgements............................................................................................................ iv Contents...............................................................................................................................v List of Figures ................................................................................................................... ix List of Tables .....................................................................................................................xv 1. Introduction................................................................................................................... 1 2. Literature Review .......................................................................................................... 3 2.1. Introduction............................................................................................................ 3 2.2. Materials for the synthesis of novolac and resole phenolic oligomers .............. 4 2.2.1. Phenols.............................................................................................................. 4 2.2.2. Formaldehyde and formaldehyde sources ........................................................ 5 2.3. Novolac resins......................................................................................................... 7 2.3.1. Synthesis of novolac resins............................................................................... 8 2.3.2. High ortho novolac resins ............................................................................. 9 2.3.3. Model phenolic oligomer synthesis ................................................................ 11 2.3.4. Reaction conditions and copolymer effects .................................................... 12 2.3.5. Molecular weight and molecular weight distribution calculations ................. 15 2.3.6. Hydrogen bonding .......................................................................................... 19 2.3.7. Novolac crosslinking with Hexamethylene Tetramine (HMTA) ................... 21 2.3.7.1. Initial reactions of novolacs with HMTA ................................................ 22 2.3.7.2. Hydroxybenzylamine and Benzoxazine decompositions in novolac/HMTA cures............................................................................................ 27 2.4. Resole resins and networks ................................................................................. 32 2.4.1. Resole resin syntheses..................................................................................... 32 2.4.2. Crosslinking reactions of resole resins............................................................ 43 2.4.3. Resole characterization ................................................................................... 45 2.4.4. Resole network properties............................................................................... 50 2.4.5. Modified phenol-formaldehyde resins............................................................ 51 2.5. Epoxy/phenol networks ....................................................................................... 52 2.5.1. Mechanism of the epoxy/phenolic reaction .................................................... 53 2.5.2. Epoxy phenolic reaction kinetics .................................................................... 55 2.5.3. Epoxy/phenol network properties ................................................................... 58 2.6. Benzoxazines......................................................................................................... 62 2.7. Phenolic triazine (PT) resins ............................................................................... 65 2.8. Thermal and thermo-oxidative degradation ..................................................... 66 3. Controlled Molecular Weight Cresol-Formaldehyde Oligomers ............................... 75 v 3.1. Introduction......................................................................................................... 75 3.2. Experimental ........................................................................................................ 79 3.2.1. Materials ......................................................................................................... 79 3.2.2. Molecular Weight Calculations ...................................................................... 79 3.2.3. Synthesis of 2,6-Dimethylphenol Endcapped Cresol Novolac Resin............. 79 3.2.4. Sample Preparation for Viscosity Measurements ........................................... 80 3.2. Characterization .................................................................................................. 80 3.2.1. Nuclear Magnetic Resonance Spectroscopy................................................... 80 3.2.2. Gel Permeation Chromatography ................................................................... 80 3.2.3. Viscosity Determinations................................................................................ 81 3.3. Results and Discussion......................................................................................... 81 3.3.1. Introduction..................................................................................................... 81 3.3.2. Molecular Weight Control and Calculations .................................................. 83 3.3.3. Structure of Reaction Intermediates and Products.......................................... 86 3.3.4. Molecular Weight and Molecular Weight Distributions Determined via GPC ................................................................................................................................. 100 3.3.5. Dynamic Viscosities of Cresol Novolac Resins ........................................... 104 3.4. Conclusions......................................................................................................... 105 Chapter 4. Structure-Property Relationships of Cresol Novolac/Epoxy Networks ..... 107 4.1. Introduction........................................................................................................ 107 4.1.1. Crosslink density and its affects on network properties ............................... 107 4.1.2. Cooperativity................................................................................................. 112 4.1.3. Thermal and thermo-oxidative stability of novolac/epoxy networks ........... 116 4.2. Experimental ...................................................................................................... 120 4.2.1. Materials ....................................................................................................... 120 4.2.2. Methods......................................................................................................... 120 4.2.2.1. Preparation of ortho-cresol novolac networks cured with epoxies........ 120 4.2.2.2. Sample preparation for viscosity determinations................................... 121 4.2.2.3. Network formation of phenolic control ................................................. 121 4.2.3. Characterization ............................................................................................ 121 4.2.3.1. Resin glass transition temperatures........................................................ 121 4.2.3.2. Network glass transition temperatures................................................... 121 4.2.3.3. Critical stress intensity factor, KIC ......................................................... 122 4.2.3.4. Sol/gel fraction separation ..................................................................... 123 4.2.3.5. 1H NMR sol fraction characterization.................................................... 123 4.2.3.6. Room temperature density measurements ............................................. 124 4.2.3.7. Determination of coefficient of thermal expansion () ......................... 124 4.2.4.8. Rubbery moduli determination via creep tests....................................... 124 4.2.3.9. 10sec relaxation moduli determination via stress relaxation tests ......... 125 4.2.3.10. Flame retardance measured via cone calorimeter................................ 126 4.2.3.11. Thermal and thermo-oxidative degradation......................................... 126 4.2.3.12. Viscosity measurements....................................................................... 127 4.2.3.13. Equilibrium moisture uptake................................................................ 127 vi 4.2.3.14. Kinetic studies via DSC ....................................................................... 128 4.2.3.15. Flexural strength and moduli of composites........................................ 128 4.3. Results and Discussion....................................................................................... 129 4.3.1. Properties of ortho-cresol novolac/epoxy networks ..................................... 129 4.3.1.1. Network formation and characterization ............................................... 129 4.3.1.2. Master curves and cooperativity ............................................................ 137 4.3.1.3. Thermal and thermo-oxidative stability................................................. 142 4.3.1.4. Flame results .......................................................................................... 144 4.3.1.5. Water absorption and diffusion efficient ............................................... 146 4.3.1.6. Reaction kinetics.................................................................................... 148 4.3.1.7. Processability ......................................................................................... 151 4.3.2. Composites properties................................................................................... 156 4.3.3. Para-cresol based networks and their properties.......................................... 157 4.4. Conclusions......................................................................................................... 159 5. Maleimide Containing Cresol Novolac Networks and Their Properties ................. 161 5.1. Introduction........................................................................................................ 161 5.2. Experimental ...................................................................................................... 166 5.2.1. Reagents........................................................................................................ 166 5.2.2. Synthetic Methods ........................................................................................ 166 5.2.2.1. Synthesis of 4-Hydroxyphenylmaleimide (4-HPM).............................. 166 5.2.2.2. Synthesis of 2-hydroxy-5-methylphenylmaleimide............................... 167 5.2.2.3. Synthesis of 2,6-dimethylphenol endcapped o-cresol-co-HPM novolac oligomers............................................................................................................. 167 5.2.2.4. Synthesis of cresol novolacs with 2-Hydroxy-5-methylphenylmaleimide endgroups............................................................................................................ 169 5.2.3. Characterization ............................................................................................ 169 5.3. Results and Discussion....................................................................................... 169 5.3.1. 4-Hydroxyphenylmaleimide synthesis and characterization ........................ 169 5.3.2. Cresol-co-HPM novolac oligomers and their properties .............................. 172 5.3.3. Cresol-co-HPM novolac/epoxy network properties ..................................... 173 5.3.4. Characterization of 2-Hydroxy-5-methylphenylmaleimide.......................... 178 5.4. Conclusions......................................................................................................... 180 6. Latent Initiators for Novolac/Epoxy Cure Reactions .............................................. 181 6.1. Introduction........................................................................................................ 181 6.2. Experimental ...................................................................................................... 190 6.2.1. Materials ....................................................................................................... 190 6.2.2. Methods......................................................................................................... 196 6.2.2.1. Melt mixing of phenolic novolac/epoxy resins...................................... 196 6.2.2.2. Preparation of polymer/TPP sequestered catalysts ................................ 197 6.2.2.3. Synthesis of Poly(arylene ether phosphine oxide)................................. 197 6.2.2.4. Reduction of Poly(arylene ether phosphine oxide)................................ 199 6.2.2.5. Synthesis of Ultem type poly(amic acid)............................................... 199 vii 6.2.2.6. Preparation of Ultem type poly(amic acid) salt with TTMPP ............... 201 6.2.2.7. Synthesis of FDA/BPDA based poly(amic acid) salts........................... 201 6.2.3. Characterization ............................................................................................ 202 6.2.3.1. Differential scanning calorimetry (DSC)............................................... 202 6.2.3.2. Viscosity measurements......................................................................... 203 6.2.3.4. 1H NMR ................................................................................................. 203 6.2.3.5. 31P NMR ................................................................................................ 203 6.2.3.6. Scanning electron microscopy ............................................................... 204 6.2.3.7. Thermogravimetric analysis................................................................... 204 6.2.4. Composite preparation and testing methods ................................................. 204 6.2.4.1. Synthesis of Ultem type poly(amic acid) salt with trihexylamine......... 204 6.2.4.2. Sizing of carbon fiber............................................................................. 205 6.2.4.3. Hot-melt prepregging and composite fabrication .................................. 206 6.2.4.4. Composite fiber volume fraction ........................................................... 207 6.2.4.5. Kinetic studies of novolac/epoxy reaction with trihexylamine.............. 208 6.2.4.6. Flexural properties ................................................................................. 208 6.2.4.7. Tensile testing ........................................................................................ 209 6.2.4.8. Mode II Toughness (GIIC) ...................................................................... 210 6.3. Results and Discussion....................................................................................... 212 6.3.1. Miscible polyimide/TPP sequestered catalysts............................................. 213 6.3.1.1. Effect of TPP on the glass transition temperatures of the blends .......... 213 6.3.1.2. Particle formation and characterization ................................................. 214 6.3.1.3. Processing windows and cure times ...................................................... 215 6.3.1.4. Surface and cross-section morphologies of the catalyst particles.......... 220 6.3.2. Udel/TPP sequestered catalysts .................................................................. 222 6.3.2.1. Blend Composition ................................................................................ 222 6.3.2.2. Processing windows and cure times ...................................................... 223 6.3.2.3. SEM of Udel/TPP Sequestered catalysts ............................................... 224 6.3.3. Partially reduced poly(arylene ether phosphine oxide)s............................... 225 6.3.3.1. Reduction of P(AEPO) .......................................................................... 225 6.3.3.2. Processing windows and cure times ...................................................... 227 6.3.4. Poly(amic acid) salts ..................................................................................... 229 6.3.5. Processability of a lower molecular weight phenolic novolac mixed with epoxy....................................................................................................................... 235 6.3.6. Properties of poly(amic acid)/trihexylamine salt sized carbon fiber reinforced novolac/epoxy composites ...................................................................................... 237 6.3.6.1. Reaction Kinetics ................................................................................... 237 6.3.6.2. Flexural properties ................................................................................. 239 6.3.6.2. Mode II toughness.................................................................................. 240 6.3.6.3. Quasistatic tensile properties ................................................................. 241 6.4. Conclusions......................................................................................................... 242 7. Conclusions ................................................................................................................ 244 8. Recommendation for Future Work ........................................................................... 248 viii List of Figures Figure 2. 1. Preparation of phenol monomer ...................................................................... 5 Figure 2. 2. Synthesis of formaldehyde .............................................................................. 6 Figure 2. 3. Formation of hemiformals............................................................................... 6 Figure 2. 4. Depolymerization of aqueous polyoxymethylene glycol ................................ 7 Figure 2. 5. Synthesis of hexamethylenetetramine ............................................................. 7 Figure 2. 6. Mechanism of novolac synthesis via electrophilic aromatic substitution ....... 8 Figure 2. 7. Byproducts of novolac synthesis ..................................................................... 9 Figure 2. 8. High ortho novolacs ...................................................................................... 10 Figure 2. 9. Proposed chelate structures in the synthesis of high ortho novolac oligomers ....................................................................................................................... 10 Figure 2. 10. Intramolecular hydrogen bonding of high ortho novolacs ....................... 11 Figure 2. 11. Selective ortho coupling reaction using bromomagnesium salts.............. 11 Figure 2. 12. Synthesis of model phenolic compound ................................................... 12 Figure 2. 13. Initial reaction of novolac and HMTA via a hydrogen bonding mechanism ................................................................................................................... 23 Figure 2. 14. Decomposition of HMTA ......................................................................... 24 Figure 2. 15. Possible reaction intermediates for reaction of 2,4-xylenol with HTMA. 26 Figure 2. 16. Thermal decomposition of hydroxybenzylamine ..................................... 27 Figure 2. 17. Thermal decomposition of benzoxazine ................................................... 28 Figure 2. 18. Reaction of benzoxazines and 2,4-xylenol ............................................... 28 Figure 2. 19. Reaction pathways for formation of ortho-ortho, ortho-para, and parapara through the reaction of para-trishydroxybenzylamine and 2,4-xylenol........... 30 Figure 2. 20. Mechanism of resole synthesis ................................................................. 33 Figure 2. 21. Reaction pathways for phenol/formaldehyde reactions under alkaline conditions.................................................................................................................. 33 Figure 2. 22. Condensation of hydroxymethyl groups................................................... 34 Figure 2. 23. Dehydration of methylols or benzylic ethers to form quinone methides.. 35 Figure 2. 24. Resonance of quinone methides................................................................ 35 Figure 2. 25. Dimer and trimer structures of ortho quinone methides ........................... 36 Figure 2. 26. Quinoid resonance forms activating the para ring position...................... 37 Figure 2. 27. Preferential formation of para quinone methides ..................................... 40 Figure 2. 28. Reactions of a quinone methide with a hydroxymethyl substituted phenolate ................................................................................................................... 41 Figure 2. 29. Reaction mechanism of phenol and formaldehyde using base catalyst involving the formation of chelate............................................................................ 42 Figure 2. 30. Ethane and ethene linkages derived from quinone methide structures..... 44 Figure 2. 31. Reaction of hydroxymethylphenol and urea ............................................. 51 Figure 2. 32. Reaction of hydroxymethylphenol and melamine .................................... 52 Figure 2. 33. Reaction of phenol and epichlorohydrin to form epoxidized novolacs .... 53 Figure 2. 34. Mechanism for the triphenylphosphine catalyzed phenol/epoxy reaction 54 Figure 2. 35. Proposed mechanism for tertiary amine catalyzed phenol/epoxy reaction55 Figure 2. 36. Network formation of phenolic novolac and epoxy.................................. 58 ix Figure 2. 37. Diepoxide structures: (1) bisphenol-A based diepoxide, (2) brominated bisphenol-A based diepoxide, and (3) siloxane diepoxide ...................................... 59 Figure 2. 38. Synthesis of bisphenol-A based benzoxazines ......................................... 63 Figure 2. 39. Reaction of benzoxazines with free ortho positions on phenolic compounds ................................................................................................................ 64 Figure 2. 40. Synthesis of phenolic triazine resins......................................................... 65 Figure 2. 41. Dehydration of hydroxyl groups............................................................... 67 Figure 2. 42. Thermal crosslinking of phenolic hydroxyl and methylene linkages ....... 67 Figure 2. 43. Thermal bond rupture: a) fragmentation reaction b) oxidation degradation. ................................................................................................................... 68 Figure 2. 44. Oxidation degradation on methylene carbon ............................................ 69 Figure 2. 45. Formation of benzenoid species................................................................ 69 Figure 2. 46. Decomposition via phenoxy radical pathways ......................................... 70 Figure 2. 47. Condensation of ortho hydroxyl groups ................................................... 71 Figure 2. 48. Char formation .......................................................................................... 71 Figure 2. 49. Decomposition of tribenzylamine............................................................. 72 Figure 3. 1. Mechanism for the major process of phenolic novolac resin synthesis ...... 76 Figure 3. 2. Synthesis of 2,6-dimethylphenol endcapped para-cresol novolac resins... 82 Figure 3. 3. 13C NMR spectra monitoring a 2000g/mol ortho-cresol novolac resin synthesis as a function of reaction time. The product was reacted for 24 hours at 100C, then heated to 200C under mild vacuum to decompose the catalyst. ......... 89 Figure 3. 4. Condensation of ortho-hydroxymethyl substituent forming stable ortholinked dimethylene ethers ......................................................................................... 90 Figure 3. 5. Expanded 13C NMR spectra monitoring a 2000 g/mol ortho-cresol novolac resin synthesis as a function of reaction time ........................................................... 91 Figure 3. 6. Deconvolution of methyl carbon peaks ....................................................... 92 Figure 3. 7. Expanded 13C NMR spectra of a series of ortho-cresol novolac resins with various molecular weights: a) methyl carbons within the repeat units, b) methyl carbons on the endgroups.......................................................................................... 93 Figure 3. 8. 13C NMR spectra of a 2000g/mol para-cresol novolac resin synthesis monitored as a function of reaction time ................................................................. 95 Figure 3. 9. Expanded 13C NMR spectra monitoring the synthesis of a 2000g/mol paracresol novolac resin................................................................................................... 96 Figure 3. 10. 1H NMR spectra of a) ortho-cresol, and b) a 2000 g/mol ortho-cresol novolac ................................................................................................................... 98 Figure 3. 11. 1H NMR spectra of a) para-cresol, and b) a 2000 g/mol para-cresol novolac ................................................................................................................... 99 Figure 3. 12. GPC monitoring the synthesis of a 2000 g/mol ortho-cresol novolac resin as a function of reaction time.................................................................................. 100 Figure 3. 13. GPC of cresol novolac resins with various molecular weights: a)orthocresol novolac, b)para-cresol novolac.................................................................... 101 Figure 3. 14. Dynamic viscosity of cresol novolacs measured as a function of molecular weight a) ortho-cresol novolac resins, and b) para-cresol novolac resins ............. 105 x Figure 4. 1. Idealized phenolic/epoxy networks ........................................................... 111 Figure 4. 2. a) Stress-relaxation experiment, and b) creep experiments....................... 113 Figure 4. 3. Illustration of cooperativity domain size where z = 7 ............................... 114 Figure 4. 4. Schematic of a cone calorimeter................................................................ 118 Figure 4. 5. Experimental implementation of the eccentric axial load technique......... 122 Figure 4. 6. Crosslinking reaction of ortho-cresol novolac and epoxy (Epon 828 or D.E.N. 438) using triphenylphosphine as the catalyst ............................................ 131 Figure 4. 7. 1H NMR of the sol fraction of cresol novolac/Epon 828 networks............ 133 Figure 4. 8. 10s Relaxation moduli as functions of temperatures for cresol novolac/Epon 828 networks........................................................................................................... 136 Figure 4. 9. 10s Relaxation moduli as functions of temperatures for phenolic novolac/Epon 828 networks.................................................................................... 136 Figure 4. 10. 10s Stress relaxation moduli as functions of temperatures for cresol novolac crosslinked with D.E.N. 438 epoxy........................................................... 137 Figure 4. 11. Master curve constructions for a typical cresol novolac/epoxy network: a) stress relaxation moduli of a cresol novolac/epoxy network measured from Tg-60C to Tg+40C at 5C intervals, and b) the master curve............................................. 138 Figure 4. 12. The shift factor plot................................................................................. 139 Figure 4. 13. Cooperativity plots of cresol novolac/Epon 828 networks .................... 140 Figure 4. 14. Cooperativity plots of cresol novolac/D.E.N. 438 networks ................. 140 Figure 4. 15. Weight loss measured as a function of temperature for cresol novolac/Epon 828 networks A) in air, and B) in nitrogen...................................... 143 Figure 4. 16. Cone calorimetry results of A) cresol novolac/Epon 828 (70:30 wt:wt ratio), and B) cresol novolac/D.E.N. 438 (70:30 wt:wt ratio) ................................ 144 Figure 4. 17. Room temperature weight percent water uptake for cresol novolac/Epon 828 networks (70:30 wt:wt ratio)............................................................................ 146 Figure 4. 18. Water uptake results for cresol novolac networks at room temperature and 62C ................................................................................................................. 147 Figure 4. 19. Log heating rate versus 1/T for cresol novolac/epoxy mixture (70:30 wt:wt ratio) with 1 mole % TPP catalyst .......................................................................... 149 Figure 4. 20. Rate constant (k) versus temperature for a cresol novolac/epoxy mixture (70:30 wt:wt ratio) with 1 mole % TPP catalyst..................................................... 150 Figure 4. 21. Dynamic DSC scans of an untreated sample versus a heat treated sample .. ................................................................................................................. 151 Figure 4. 22. Complex viscosity of a 2000 g/mol neat cresol novolac resin measured as a function of temperature ........................................................................................ 152 Figure 4. 23. Complex viscosity of a phenolic novolac resin before and after heat treatment (2 hours at 160C)................................................................................... 153 Figure 4. 24. Viscosity measurements of cresol novolac/Epon 828 mixtures A) dynamic scans for various compositions, B) isothermal scan of the 70:30 composition at 145C, and C) isothermal scan of the 60:40 composition at 120C ....................... 154 Figure 4. 25. Isothermal viscosity measurements: A) 65:35 wt:wt phenolic novolac/Epon 828 mixture measured at 140C, and B) 70:30 wt:wt cresol novolac/Epon 828 mixture measured at 145C ...................................................... 155 Figure 4. 26. Viscosity measurements for cresol novolac/D.E.N. 438 mixtures: A) dynamic measurements, B) isothermal scan for the 60:40 composition at 160C . 156 xi Figure 4. 27. 2000 g/mol para-cresol novolac cured with Epon 828 ............................. 158 Figure 4. 28. Viscosity of a 2000g/mol para-cresol novolac resin (heat rate = 2.5C /min) ................................................................................................................. 158 Figure 5. 1. Preparation of bismaleimide from a diamine and maleic anhydride......... 161 Figure 5. 2. Reactions of bismaleimide in the presence of a diamine: A) chain extension due to an amine addition, and B) crosslinking obtained by maleimide homopolymerization reactions................................................................................ 163 Figure 5. 3. Synthesis of novolac resins containing maleimide functionalities............ 165 Figure 5. 4. Synthesis of 4-hydroxyphenylmaleimide .................................................. 167 Figure 5. 5. Synthesis of 2-hydroxyl-5-methylphenylmaleimide ................................. 167 Figure 5. 6. Synthesis of 2,6-dimethylphenol endcapped cresol-co-HMP novolac resin... ..................................................................................................................... 168 Figure 5. 7. Synthesis of 2-hydroxy-5-methylphenylmaleimide terminated cresol novolac resins.......................................................................................................... 169 Figure 5. 8. 1H NMR spectrum of 4-hydroxyphenylmaleimide monomer ................... 170 Figure 5. 9. Melting point of 4-HPM determined via DSC ......................................... 171 Figure 5. 10. Thermal stability of 4-HPM monomer measured via TGA (10C/min, N2) .. ................................................................................................................. 171 1 Figure 5. 11. H NMR of a typical cresol-co-HPM novolac resin ................................. 172 Figure 5. 12. Percent weight loss for cresol-co-HPM novolac/epoxy networks (80:20 wt:wt ratio) prepared with different oligomer molecular weights, monitored using thermogravimetric analysis..................................................................................... 175 Figure 5. 13. Heat release rate curves for cresol-co-HPM novolac/Epon 828 networks ... ................................................................................................................. 177 Figure 5. 14. 1H NMR of 2-hydroxy-4-methylphenylmaleimide................................. 178 Figure 5. 15. Successive dynamic DSC scans of 2-hydorxy-4-methylphenylmaleimide .. ................................................................................................................. 179 Figure 5. 16. TGA monitoring the weight loss of 2-hydroxy-4-methylphenylmaleimide monomer as a function of temperature (10C/min, N2).......................................... 180 Figure 6. 1. Mechanism of TPP catalyzed phenolic novolac/epoxy reaction ................ 182 Figure 6. 3. Diagram of pultrusion processing .............................................................. 183 Figure 6. 4. High temperature imidization of PAAS to release TTMPP catalyst .......... 185 Figure 6. 5. The chemical structure of N-benyzlpyrazinium hexafluoroantimonate ..... 186 Figure 6. 6. Decarboxylation reaction of salicylic acid salt to form phenolate ............ 188 Figure 6. 7. Preparation of phosphonium ylides ........................................................... 189 Figure 6. 8. Synthesis of poly(arylene ether phosphine oxide)..................................... 198 Figure 6. 9. Reduction of phosphine oxide to phosphine using phenylsilane............... 199 Figure 6. 10. Synthesis of Ultem type poly(amic acid) salt with TTMPP..................... 200 Figure 6. 11. Synthesis of biphenyl dianhydride and FDA based poly(amic acid) and poly(amic acid) salt................................................................................................. 202 Figure 6. 12. Preparation of Ultem type poly(amic acid) salt with trihexylamine....... 205 Figure 6. 13. Schematic of a sizing line ....................................................................... 206 Figure 6. 14. Schematic representation of the hot melt prepregging process .............. 207 xii Figure 6. 15. Composite ply lay up to form crossply or unidirectional specimen for tensile testing .......................................................................................................... 209 Figure 6. 16. Tensile test specimen with epoxy/glass fiber tabs .................................. 209 Figure 6. 17. Compliance determination of the uncracked sample .............................. 211 Figure 6. 18. Compliance determination of cracked samples ...................................... 211 Figure 6. 19. Glass transition temperature of polyimide/ TPP blend measured as a function of TPP content a) Ultem /TPP blend b) Matrimid /TPP blend............ 213 Figure 6. 20. Percent weight loss of Matrimid/TPP blend as a function of temperature ... ................................................................................................................. 214 Figure 6. 21. SEM of Matrimid/TPP particles a) before separation, b) fine particles that passed through the sieve, and c) larger particles that did not pass through the sieves . ................................................................................................................. 215 Figure 6. 22. Isothermal DSC at 135C for phenolic novolac/Epon 828 epoxy mixtures with no catalyst, with a Matrimid/TPP (50:50) sequestered catalyst, or with free triphenylphosphine catalyst (arbitrary vertical placements of curves) .................. 216 Figure 6. 23. Isothermal DSC at 200C for phenolic novolac/Epon 828 epoxy mixtures with no catalyst, with a Matrimid/TPP (50:50) sequestered catalyst, or with free triphenylphosphine catalyst .................................................................................... 217 Figure 6. 24. Isothermal DSC at 220C for phenolic novolac/Epon 828 epoxy mixtures with no catalyst, with a Matrimid/TPP (50:50) sequestered catalyst, or with free triphenylphosphine catalyst .................................................................................... 218 Figure 6. 25. Isothermal viscosity at 140C for phenolic novolac /Epon 828 epoxy mixtures with Matrimid/TPP sequestered catalysts (50:50), unwashed, acetone washed and methanol washed................................................................................. 219 Figure 6. 26. SEM of Matrimid/TPP particle surfaces................................................. 221 Figure 6. 27. SEM of a cross-section of a Matrimid/TPP particle ............................... 221 Figure 6. 28. 1H NMR of TPP, Udel, Udel/TPP, and methanol washed Udel/TPP ..... 222 Figure 6. 29. DSC scans of phenolic novolac/epoxy mixtures containing Udel/TPP catalyst ................................................................................................................. 223 Figure 6. 30. Isothermal viscosity determination of phenolic novolac/epoxy mixtures at 140C without catalyst, with 0.65 mol % catalyst, and with 1.6 mol % catalyst. .. 224 Figure 6. 31. SEM of a cross-section of an Udel/TPP particle .................................... 225 Figure 6. 32. Glass transition temperature vs. percent reduction of P(AEPO.............. 226 Figure 6. 33. Percent reduction of (P=O) as a function of reaction time for P=O:SiH3Ph (1:1.5 molar ratio) ................................................................................................... 226 Figure 6. 34. Isothermal DSC of phenolic novolac/Epon 828 with 1 mol % reduced P(AEPO) at 140C and at 220C ........................................................................... 228 Figure 6. 35. Isothermal viscosity (140C) of phenolic novolac/Epon 828 epoxy with reduced P(AEPO).................................................................................................... 228 Figure 6. 36. Poly(amic acid) salts 1) Ultem type PAAS/TTMPP, 2) FDA/BPDA based PAAS/imidazole, and 3) FDA/BPDA based PAAS/trihexylamine........................ 230 Figure 6. 37. Dynamic DSC scans of (1) Ultem PAAS/TTMPP, (2) FDA/BPDA based PAAS/imidazole, and (3) FDA/BPDA based PAAS/trihexylamine....................... 231 Figure 6. 38. Dynamic DSC scans of novolac/epoxy mixture with 2 mole % PAAS (1) Ultem PAAS/TTMPP, (2) FDA/BPDA PAAS/imidazole, and (3) FDA/BPDA PAAS/trihexylamine ............................................................................................... 232 xiii Figure 6. 39. Phenolic novolac/epoxy with 2 mol % Ultem type PAAS/TTMPP ....... 232 Figure 6. 40. Phenolic novolac/epoxy with 2 mol % PAAS (FDA/BPDA) imidazole...... ................................................................................................................ 233 Figure 6. 41. Isothermal DSC at 140oC for novolac/epoxy mixtures: no catalyst, with PAAS (FDP/BPDA/trihexylamine), and with free trihexylamine.......................... 234 Figure 6. 42. Isothermal DSC at 200oC of phenolic novolac/epoxy without catalyst, with PAAS (FDP/BPDA/trihexylamine), and with free trihexylamine.......................... 234 Figure 6. 43. Viscosity during heating and holding at 140C of phenolic novolac/epoxy with 2 mole % PAAS-3 (FDP/BPDA/trihexylamine), ........................................... 235 Figure 6. 44: Isothermal viscosities of phenolic novolac/epoxy mixtures, Novolac G (Georgia Pacific resin, f(OH) = 7) and Novolac O (Occidental resin, f(OH) = 4.4)... 236 Figure 6. 45: Isothermal viscosity of lower molecular weight phenolic novolac/epoxy mixtures with and without sequestered catalysts .................................................... 237 Figure 6. 46. Dynamic DSC scans of a novolac/epoxy/trihexylamine mixture measured at different heating rates. The peak shift due to the instrument response lag was corrected by measuring the indium melting point at these same heating rates....... 238 Figure 6. 47. a) log heating rate () versus 1/peak temperature of the exotherm, and b) rate constant versus temperature for a novolac/epoxy mixture containing 3 mole percent trihexylamine.............................................................................................. 239 Figure 6. 48. Stress vs. transverse strain for crossply PAAS/trihexylamine sized AS-4 carbon fiber reinforced phenolic novolac/epoxy composites ................................. 242 xiv List of Tables Table 2. 1. U.S. Phenolic production (in millions of pounds on a gross weight basis) .... 3 Table 2. 2. Relative reaction rates of various phenols with formaldehyde under basic conditions.................................................................................................................. 13 Table 2. 3. Peak assignments for 13C NMR chemical shifts of phenolic resins , ............ 18 Table 2. 4. Relative positional reaction rates in base catalyzed phenol-formaldehyde reaction...................................................................................................................... 37 Table 2. 5. Second order rate constants for reaction of phenolic monomers with formaldehyde ............................................................................................................ 38 Table 2. 6. % yield of methylene and ether linkages of 2-hydroxylmethyl-4,6dimethylphenol self-reaction, 1:1 with 2,4-xylenol, and 1:1 with 2,6-xylenol......... 44 Table 2. 7. FTIR absorption band assignment of resole resins ....................................... 48 Table 2. 8. Tg and KIC of phenolic novolac/epoxy networks .......................................... 60 Table 2. 9. Flame retardance of networks prepared form a phenolic novolac crosslinked with various epoxies ................................................................................................. 61 Table 3. 1. Molecular weight of ortho- and para-cresol novolac resins calculated using 13 C NMR. The molecular weights were controlled by adjusting NAA/NZA ratio.... 86 Table 3. 2. 13C NMR assignments for novolac resins and related reaction intermediates.. ....................................................................................................................... 87 Table 3. 3. Mole percent ortho-dimethylene ether linkages ........................................... 90 Table 3. 4. Percentage isomers formed in ortho-cresol novolac resins .......................... 93 Table 3. 5. Polydispersities and intrinsic viscosities of cresol novolac resins.............. 103 Table 3. 6. Tg of cresol novolac resins as a function of molecular weight.................... 103 Table 4. 1. Fracture toughness of phenolic novolac/epoxy networks ........................... 111 Table 4. 2. Cone calorimetry results on phenolic novolac/epoxy networks (65:35 wt:wt ratio) ..................................................................................................................... 119 Table 4. 3. Phenolic materials and their properties........................................................ 130 Table 4. 4. Network properties of ortho-cresol novolac/epoxy networks ................... 132 Table 4. 5. Crosslink densities of cresol novolac/epoxy networks ............................... 134 Table 4. 6. Fragility measuring the crosslink densities and degree of hydrogen boning interaction for cresol novolac/epoxy networks ....................................................... 141 Table 4. 7. Flame retardance of cresol novolac/epoxy networks................................. 145 Table 4. 8. Diffusion efficient of cresol novolac/epoxy networks................................ 148 Table 4. 9. Cure condition determination for ortho-cresol novolac/Epon 828 network (70:30 wt:wt %), no catalyst ................................................................................... 157 Table 4. 10. Flexural strength and moduli of composites............................................... 157 Table 4. 11. KIC and Tg of para-cresol novolac/Epon 828 networks.............................. 159 Table 5. 1. Tg of cresol-co-HMP oligomer as a function of Mn .................................... 173 Table 5. 2. Properties of ortho-cresol-co-HPM/Epon 828 networks ............................ 174 Table 5. 3. Thermal stability of cresol-co-HMP novolac/epoxy networks measured using thermogravimetric analysis..................................................................................... 175 xv Table 5. 4. Cone calorimetry measuring the peak heat release rate (PHRR) and the char yield of 1250 g/mol cresol-co-HPM/epoxy networks............................................. 176 Table 5. 5. Cone calorimetry results for 60:40 wt:wt cresol-co-HPM/epoxy networks prepared with different molecular weight oligomers.............................................. 177 Table 6. 2. Processing windows and cure times of phenolic novolac/epoxy and Matrimid sequestered catalysts ............................................................................................... 220 Table 6. 3. Particle compositions of unwashed and methanol washed Udel/TPP particles ..................................................................................................................... 223 Table 6. 4. Properties of partially reduced P(AEPO).................................................... 227 Table 6. 5. Resin and network properties of novolac resins of different molecular weights .................................................................................................................... 236 Table 6. 6. Transverse flexural strength and modulus of unidirectional AS-4 carbon fiber reinforced phenolic novolac/epoxy composites...................................................... 240 Table 6. 7. Mode II composite toughness of unidirectional AS-4 carbon fiber reinforced phenolic novolac/epoxy composites ....................................................................... 241 Table 6. 8. Static tensile properties of AS-4 carbon fiber reinforced phenolic novolac/epoxy composites ...................................................................................... 241 xvi 1. Introduction Fiber reinforced polymer matrix composites for structural applications, generally comprised of continuous fiber embedded in a polymer matrix, have high strength to weight ratios. Such composites also have superior oxidative resistance relative to steel and better freeze-thaw durability relative to concrete. However, the high combustibility of organic matrix materials limits their use in construction or transportation applications. Phenolics are widely used as adhesives, coatings, and in various electric, structural, and aerospace applications. The main advantages of phenolic resins, both novolacs and resoles, and their networks are excellent flame retardance and low cost. A significant amount of academic research, therefore, has been devoted to understanding phenolic resin synthesis and network formation both mechanistically and kinetically. The thermal and thermo-oxidative degradation pathways have also been extensively investigated. A survey on phenolic resin syntheses, network formations, and degradation pathways is included in a phenolic chemistry literature review (chapter 2). One major shortcoming of typical phenolic networks is their large void contents due to released volatiles in the cure reactions. This, and the lack of control over network crosslink density, gives rise to brittle networks. Therefore, we and others have investigated phenolic or phenolic network modifications in order to improve these mechanical properties while retaining high thermal stability and flame retardance. Previous work in our group focused on developing novolac/epoxy networks as composite matrix materials for structural applications.1 A relatively high molecular weight novolac (7 hydroxyl groups per chain) was reacted with various diepoxides where the phenolic was the major component. The network density was controlled by adjusting the ratio of phenol to epoxy. Fracture toughness of the networks having 3 to 5 phenols per epoxy exceeded that of an untoughened epoxy control (bisphenol-A stoichiometrically cured with 4,4-DDS) and far exceeded that of phenolic resoles. The flame retardance of the phenolic novolac/epoxy networks was significantly improved relative to the epoxy control. 1 C. S. Tyberg, Void-Free Flame Retardant Phenolic Networks: Properties and Processability, Dissertation, Virginia Tech, March 22, 2000. 1 Latent catalysts were subsequently developed to allow melt processing of the phenolic novolac/bisphenol-A epoxy mixtures. Catalysts were encapsulated onto the fiber, which eliminated resin/catalyst contact, and therefore prevented premature curing during processing. Results indicated that a poly(amic acid) salt of tris-(trimethoxyphenylphosphine) was effective in catalyzing the phenolic novolac/epoxy reaction. The ultimate goal of this research was to develop tough, flame retardant matrix resins which can be processed easily using typical composite fabrication methods. Attempts were also made to improve the specific drawbacks on existing phenolic novolac/epoxy systems such as high water uptake, and relatively short processing windows even in the absence of added catalysts. Following a literature review, the specific work of this dissertation is presented. This work has four major sections. 1) The first section focuses on the synthesis and characterization of controlled molecular weight cresol novolac resins. Specific reaction conditions and means to achieve molecular weight control are described. 2) The second section discusses in detail the network properties of a 2000 g/mol cresol novolac resin crosslinked with various epoxies at defined compositions. It investigates the network structure-property relationships, which allowed for understanding of the parameters that affect the network mechanical properties, flame retardance and processability. It also examines the molecular relaxation behaviors (cooperativity) and its relationships with network crosslink density and chemical structures. 3) The third section extends the results obtained in section 2 and assesses the effects of incorporating maleimide functionalities into cresol novolac/epoxy networks. The balance between network properties and processability is addressed. 4) The last section considers various approaches to sequester and encapsulate tertiary amines or phosphine catalysts which can be added directly to novolac/epoxy mixtures at melt processing temperatures. novolac/epoxy reactions. It also presents poly(amic acid)/trihexylamine as a more cost effective latent catalyst sizing for phenolic 2 2. Literature Review Chemistry and Properties of Phenolic Resins and Networks 2.1. Introduction Phenolic resins comprise a large family of oligomers and polymers (Table 2. 1), which are various products of phenols, reacted with formaldehyde. They are versatile synthetic materials with a large range of commercial applications. Plywood adhesives account for nearly half of all phenolic applications while wood binding and insulation materials also make up a significant portion.2 Other uses for phenolics include coatings, adhesives, binders for abrasives, automotive and electrical components, electronic packaging and as matrices for composites. Table 2. 1. U.S. Phenolic production (in millions of pounds on a gross weight basis)3 1998 3940 1997 3734 % Change 5.5 Phenolic oligomers are prepared by reacting phenol or substituted phenols with formaldehyde or other aldehydes. Depending on the reaction conditions (e.g., pH) and the ratio of phenol to formaldehyde, two types of phenolic resins are obtained. Novolacs are derived from an excess of phenol under neutral to acidic conditions, while reactions under basic conditions using an excess of formaldehyde result in resoles. Phenolic resins were discovered by Baeyer in 1872 through acid catalyzed reactions of phenols and acetaldehyde. Kleeberg found in 1891 that resinous products could also be formed by reacting phenol with formaldehyde. But it was Baekeland who was granted patents in 1909 describing both base catalyzed resoles (known as Bakelite resins) and acid catalyzed novolac products.4 This chapter emphasizes the recent mechanistic and kinetic findings on both phenolic oligomer syntheses and network formation. The synthesis and characterization 2 3 Society of Plastic Industries Facts and Figures, SPI, Washington, D.C. (1994). Society of Plastic Industries Facts and Figures, SPI, Washington D.C. (1999). 3 of both novolac and resole type phenolic resins and their resulting networks are described. Three types of networks, novolac/hexamethylene tetramine (HMTA), Since phenolic materials novolac/epoxies, and thermally cured resoles will be primarily discussed. Other phenolic based networks include benzoxazines and cyanate esters. degradation pathways will be included. references.4,5,6,7,8 possess excellent flame retardance, a discussion of the thermal and thermo-oxidative Detailed information on the chemistry, applications, and processing of phenolic materials can be found in a number of 2.2. Materials for the synthesis of novolac and resole phenolic oligomers 2.2.1. Phenols The most common precursor to phenolic resins is phenol. More than 95% of phenol is produced via the cumene process developed by Hock and Lang (Figure 2. 1). Cumene is obtained from the reaction of propylene and benzene through acid catalyzed alkylation. Oxidation of cumene in air gives rise to cumene hydroperoxide, which decomposes rapidly at elevated temperatures under acidic conditions to form phenol and acetone. A small amount of phenol is also derived from coal. 4 A. Knop and L. A. Pilato, Phenolic Resins--Chemistry, Applications and Performance, A. Knop, W and W. Scheib, Chemistry and Application of Phenolic Resins, SpringerS. R Sandler and W. Karo, Polymer Synthesis, 2nd editions, Academic Press, Boston, P. W. Kopf in J. I. Kroschulitz, ed., Encyclopedia of Chemical Technology, 4th Ed., Vol R.T. Conley, Thermal Stability of Polymers, Marcel Dekker, Inc., New York, 1970, pp. Springer-Verlag, Berlin, 1985. 5 Verlag, New York, 1979. 6 Vol. 2, 1992. 7 18, John Wiley & Sons, 1996, pp 603-644. 8 459-496. 4 OOH CH3 CH CH2 O2 catalyst H + OH O phosphoric acid + CH3 C CH3 Figure 2. 1. Preparation of phenol monomer Substituted phenols such as cresols, p-tert-butylphenol, p-phenylphenol, resorcinol, and cardanol (derived from cashew nut shells) have also been used as precursors for phenolic resins. Alkylphenols with at least three carbons in the substituent lead to more hydrophobic phenolic resins which are compatible with many oils, natural resins and rubbers.9 Such alkylphenolic resins are used as modifying and crosslinking agents for oil varnishes, as coatings and printing inks, and as antioxidants and stabilizers. Bisphenol-A (2,2-p-hydroxyphenylpropane), a precursor to a number of phenolic resins, is the reaction product of phenol and acetone under acidic conditions. An additional activating hydroxyl group on the phenolic ring allows resorcinol to react rapidly with formaldehyde even in the absence of catalysts.10 formaldehyde/resorcinol-formaldehyde resins. This provides a method for room temperature cure of resorcinol-formaldehyde resins or mixed phenolTrihydric phenols have not achieved commercial importance, probably due to their higher costs. 2.2.2. Formaldehyde and formaldehyde sources Formaldehyde, produced by dehydrogenation of methanol, is used almost exclusively in the synthesis of phenolic resins (Figure 2. 2). Iron oxide, molybdenum oxide or silver catalysts are typically used for preparing formaldehyde. Air is a safe source of oxygen for this oxidation process. 9 K. Hultzsch Recent Chemical and Technical Aspects on Alkylphenolic Resins, American Chemical Society, Division of Organic Coating & Plastic Chemistyr, Pap. 26(1), 121-128 (1966) 10 U.S. Patents 2,385,370 (1947) A. J. Norton; U.S Patents 2,385,372 (1946) P. H. Phodes. 5 CH3 OH + 1/2 O2 Figure 2. 2. Synthesis of formaldehyde catalyst O H C H + H2O Since formaldehyde is a colorless pungent irritating gas, it is generally marketed as a mixture of oligomers of polymethylene glycols either in aqueous solutions (formalin) or in more concentrated solid forms (paraformaldehyde). The concentration of formalin ranges between about 37 and 50 wt %. A 40 wt % aqueous formalin solution at 35C typically consists of methylene glycols with 1 to 10 repeat units. The molar concentration of methylene glycol with one repeat unit (HO-CH2-OH) is highest and the concentrations decrease with increasing numbers of repeat units.11 Paraformaldehyde, a white solid, contains mostly polymethylene glycols with 10 to 100 repeat units. It is prepared by distilling aqueous formaldehyde solutions and generally contains 1-7 wt % water. Methanol, the starting reagent for producing formaldehyde, stabilizes the formalin solution by forming acetal endgroups and is usually present in at least small amounts (Figure 2. 3). Methanol may also be formed by disproportionation during storage. The presence of methanol reduces the rate of phenol/formaldehyde reactions but does not affect the activation energies.12 It is generally removed by stripping at the end of the reaction. CH3 Figure 2. 3. OH + HO CH2 O n H CH3 O CH2 O n H + H2O Formation of hemiformals Water is necessary for decomposing paraformaldehyde to formaldehyde (Figure 2. 4). However, water can serve as an ion sink and water-phenol mixtures phase separate 11 H Diehm and A. Hit, Formaldehyde Ullmanns Encyclopadie der Techn. Chem., 4th C. M. Chen and S. L. Chen, Effects of Methanol on the Reactions of the Phenol- ed, Verlag Chemie, Weinheim, Vol.11, 1976. 12 Formaldehyde System, Forest Products Journal 38(5), 49-52 (1988). 6 as the water concentration increases. Therefore, large amounts of water reduce the rate of reaction between phenol and formaldehyde.13 HO CH2 Figure 2. 4. O n H + H2O HO CH2 O n-1 H + HO CH OH Depolymerization of aqueous polyoxymethylene glycol Hexamethylenetetramine (HMTA) used for crosslinking novolacs or catalyzing resole syntheses is prepared by reacting formaldehyde with ammonia (Figure 2. 5). The reaction is reversible at high temperatures, especially above 250C. HMTA can also be hydrolyzed in the presence of water. N 6 CH2O + 4 NH3 Figure 2. 5. N N N + 6 H2O Synthesis of hexamethylenetetramine 2.3. Novolac resins The most common precursors for preparing novolac oligomers and resins are phenol, formaldehyde sources and to a lesser extent, cresols. Three reactive sites for electrophilic aromatic substitution are available on phenol which give rise to three types of linkages between aromatic rings, i.e. ortho-ortho, ortho-para, and para-para. The complexity of the isomers leads to amorphous materials. For a novolac chain with ten phenol groups, 13,203 isomers14 can statistically form, making the separation of pure phenolic compounds from novolacs nearly impossible. 13 A. J. Rojas and R. J. J. Williams, Novolacs From Paraformaldehyde, Journal of N. J. L. Megson, Unsolved Problems in Phenol Resin Chemistry, Chem.-Ztg., 96(1- Applied Polymer Science 23, 2083-2088 (1979). 14 2), 15-19 (1972). 7 2.3.1. Synthesis of novolac resins Novolacs are prepared with an excess of phenol over formaldehyde under acidic conditions (Figure 2. 6). A methylene glycol is protonated by an acid from the reaction medium, which then releases water to form a hydroxymethylene cation (step 1 in Figure 2. 6). This ion hydroxyalkylates a phenol via electrophilic aromatic substitution. The rate determining step of the sequence occurs in step 2 where a pair of electrons from the phenol ring attacks the electrophile forming a carbocation intermediate. The methylol group of the hydroxymethylated phenol is unstable in the presence of acid and loses water readily to form a benzylic carbonium ion (step 3). This ion then reacts with another phenol to form a methylene bridge in another electrophilic aromatic substitution. This major process repeats until the formaldehyde is exhausted. + + 1) HO CH2 OH + OH H CH2 OH + H2O OH OH + + 2) OH CH2 OH slow + CH2 OH fast CH2 OH + H+ OH CH2 OH + +H 3) CH2 + + H2O OH 4) OH CH2 + OH + OH CH2 + H+ Figure 2. 6. Mechanism of novolac synthesis via electrophilic aromatic substitution The reaction between phenol and formaldehyde is exothermic. Therefore the temperature must be controlled to prevent the build-up of heat, particularly during the early stages of reaction.6 When formalin is used, water provides a medium for heat dissipation. Typical formaldehyde to phenol ratios in novolac syntheses range from about 0.7 to 0.85 to maintain oligomers with sufficiently low molecular weights and reasonable 8 melt viscosities. This is especially important since phenol is trifunctional and a gel fraction begins to form as conversion increases. As a result, the number average molecular weights of novolac resins are generally below 1000g/mol. The acidic catalysts used for these reactions include formic acid, HX (X=F, Cl, Br), oxalic acid, phosphoric acid, sulfuric acid, sulfamic acid, and p-toluenesulfonic acid.6 Oxalic acid is preferred since resins with low color can be obtained. Oxalic acid also decomposes at high temperatures (>180C) to CO2, CO and water, which facilitates the removal of this catalyst thermally. Typically, 1-6 wt % catalyst is used. Hydrochloric acid results in corrosive materials and reportedly releases carcinogenic chloromethyl ether by-products during resin synthesis.4 Approximately 4-6 wt % phenol can typically be recovered following novolac reactions. Free phenol can be removed by washing with water repeatedly. The recovered phenolic components may contain 1,3-benzodioxane, probably derived from benzyl hemiformals (Figure 2. 7).4 O O OH O OH 1,3-benzodioxane Figure 2. 7. Byproducts of novolac synthesis benzyl hemiformal 2.3.2. High ortho novolac resins High ortho-novolacs (Figure 2. 8) are sometimes more desirable since they cure more rapidly with HMTA. A number of oxides, hydroxide or organic salts of electropositive metals increase the reactivity of the ortho position during oligomer formation.15 These high ortho-novolacs are typically formed at pHs of 4 to 6 as opposed to the more common strongly acidic conditions. 15 U.S. Patents 2,464,207 and U.S. Patent 2,475,587 (1949) H. L. Bender, A.G Farnham and J. W. Guyer. 9 OH HO OH O OH OH n Figure 2. 8. High ortho novolacs m Metal hydroxides of first and second group elements can enhance ortho substitution, the degree of which depends on the strength of metal chelating effects linking the phenolic oxygen with the formaldehyde as it approaches the ortho position. Transition metal ions of elements such as Fe, Cu, Cr, Ni, Co, Mn and Zn as well as boric acid also direct ortho substitutions via chelating effects (Figure 2. 9). Ph M O O CH 2 H O OH Ph M O CH2 OH H O HO B O CH2 H OH M = transition metal Figure 2. 9. Proposed chelate structures in the synthesis of high ortho novolac oligomers Phenol-formaldehyde reactions catalyzed by zinc acetate as opposed to strong acids have been investigated, but this results in lower yields and requires longer reaction times. The reported ortho-ortho content yield was as high as 97%. Several divalent metal species such as calcium, barium, strontium, magnesium, zinc, cobalt and lead combined with an organic acid (such as sulfonic and/or fluoroboric acid) improved the reaction efficiencies.16 The importance of an acid catalyst was attributed to facilitated decomposition of any dibenzyl ether groups formed in the process. It was also found the reaction rates could be accelerated with continuous azeotropic removal of water. An interesting aspect of high ortho novolac oligomers is their so-called hyperacidity. The enhanced acidity of high ortho novolac resins, intermediate between 16 U.S. Pat. 4,113,700 (Sept. 12, 1978), W. Aubertson (to Monsanto Co.). 10 phenols and carboxylic acids, has been attributed to increased dissociation of the phenol protons due to strong intramolecular hydrogen bonding (Figure 2. 10). These materials are also reported to form strong complexes with di- and tri-valent metals and nonmetals.4 H H H O O H O H + O O H O H O O Figure 2. 10. Intramolecular hydrogen bonding of high ortho novolacs 2.3.3. Model phenolic oligomer synthesis Linear novolac oligomers containing only ortho linkages were prepared using bromomagnesium salts under dry conditions.17 The bromomagnesium salt of phenol coordinates with the incoming formaldehyde (Figure 2. 11A) or quinone methide (Figure 2. 11B) directing the reaction onto only ortho positions. O O CH2 O CH2 MgBr MgBr O CH2 H O O O CH2 A. CH2 MgBr O H O MgBr O MgBr - HO-MgBr OH MgBr OH O CH2 B. CH2 Figure 2. 11. Selective ortho coupling reaction using bromomagnesium salts 17 G. Casnati. A. Pochini, G. Sartori, and R. Ungaro, Template Catalysis via Non- Transition Metal Complexes-New Highly Selective Synthesis on Phenol Systems, Pure and Applied Chemistry 55(11), 1677-1688 (1983). 11 Solomon et al.18 prepared low molecular weight model novolac compounds containing 4 to 8 phenolic units utilizing the bromomagnesium salt methodology (Figure 2. 12). A para-para linked dimer was used as the starting material where tertbutyldimethylsilyl chloride was reacted with one phenol on a dimer to deactivate its ring against electrophilic reaction with formaldehyde. Selective ortho coupling formed bridges between the remaining phenol rings; then the tert-butyldimethylsilyl protecting groups were removed with fluoride ion. These compounds and all ortho linked model compounds prepared using bromomagnesium salts were subsequently used as molecular weight standards for calibrating gel permeation chromatography and to study model reactions with HMTA (see section 3.8). OH OH OH OH OH OH TBSCl imidazole DMF CH2O CH3CH2MgBr TBAF OH OTBS OTBS OTBS OH OH Figure 2. 12. Synthesis of model phenolic compound 2.3.4. Reaction conditions and copolymer effects Alkyl substituted phenols have different reactivities than phenol toward reaction with formaldehyde. Relative reactivities determined by monitoring the disappearance of formaldehyde in phenol/paraformaldehyde reactions (Table 2. 2) show that under basic conditions, meta-cresol reacts with formaldehyde approximately 3 times faster than phenol while ortho- and para-cresols react at approximately 1/3 the rate of phenol.19 18 P. J. de Bruyn, A. S. L. Lim, M. G. Looney, and D. H. Solomon, Strategic Synthesis M. M. Sprung, Reactivity of Phenol Toward Formaldehyde, Journal of Applied of Model Novolac Resins, Tetrahedron Letters 35(26), 4627-4630 (1994). 19 Polymer Science 63(2), 334-343 (1941). 12 Similar trends were observed for the reactivities of acid catalyzed phenolic monomers with formaldehyde. One comparison study of oxalic acid catalyzed reactions involving ortho- and meta-cresol mixtures demonstrated that meta-cresol was preferentially incorporated into the oligomers during the early stages of reaction.20 Given the same reaction conditions and time, higher ortho-cresol compositions (of the mixtures) resulted in decreased overall yields since there was insufficient time for ortho-cresol to fully react. Consequently, the molecular weights and Tgs were also lower in these partially reacted materials. As expected, the molecular weight increased if a larger amount of catalyst was used or if more time was allowed for reaction. Increased catalyst concentrations also broadened molecular weight distributions. Table 2. 2. Relative reaction rates of various phenols with formaldehyde under basic conditions5 Compound 2,6-xylenol ortho-cresol para-cresol 2,5-xylenol 3,4-xylenol phenol 2,3,5-trimethylphenol meta-cresol 3,5-xylenol Relative reactivity 0.16 0.26 0.35 0.71 0.83 1.00 1.49 2.88 7.75 Bogan conducted similar studies in which meta- and/or para-cresols were reacted with formaldehyde at 99C for 3 hours using oxalic acid dihydrate as the catalyst to form 20 St. Miloshev, P. Novakov, Vl. Dimitrov, and I. Gitsov, Synthesis of Novolac Resins. I. Influence of the Chemical Structure of the Monomers and Reaction Conditions on Some Properties of Novolac Oligomers, Chemtronics 4, 251-253 (1989). 13 novolac type structures.21 Using a relative reactivity of 0.090.03 for para-cresol with formaldehyde versus. meta-cresol with formaldehyde, a statistical model was employed to predict the amounts of unreacted cresols during the reactions, branching density, and m/p-cresol copolymer compositions. Good agreement was found between the predictions and experimental results. Since para-cresol reacted much slower that meta-cresol, it was to a first approximation considered an unreactive diluent. When meta- and para-cresol mixtures were reacted, oligomers consisting of mostly meta-cresol formed first, then when the meta-cresol content was depleted, para-cresol incorporation was observed (mostly at the chain ends). Full conversions were not achieved in these investigations, probably due to insufficient reaction times for para-cresol to react completely. Linear novolac resins prepared by reacting para-alkylphenols with paraformaldehyde are of interest for adhesive tackifiers. As expected for step-growth polymerization, the molecular weights and viscosities of such oligomers prepared in one exemplary study increased as the ratio of formaldehyde to para-nonylphenol was increased from 0.32 to 1.00.22 As is usually the case, however, these reactions were not carried out to full conversion and the measured Mn of an oligomer prepared with an equimolar phenol to formaldehyde ratio was 1400 g/mol. rheological behavior. Reaction media play an important role in meta-cresol/paraformaldehyde reactions.23 Higher molecular weight resins, especially those formed from near Plots of apparent shear viscosity vs. shear rate of these p-nonylphenol novolac resins showed non-Newtonian 21 L. E. Bogan, Jr., in P. N. Prasad, ed., Understanding the Novolac Synthesis Reaction, Frontiers of Polymers and Advanced Materials, Plenum Press, New York, 1994, 311318. 22 C. N. Cascaval, D. Rosu, and F. Mustata, Synthesis and Characterization of Some para-Nonylphenol Formaldehyde Resins, European Polymer Journal 30(3), 329-333 (1994). 23 St. Miloshev, P. Novakov, Vl. Dimitrov, and I. Gitsov, Synthesis of Novolac Resins: 2. Influence of the Reaction Medium on the Properties of the Novolac Oligomers, Polymer 32(16), 3067-3070 (1991). 14 equimolar meta-cresol to formaldehyde ratios, can be obtained by introducing a water miscible solvent such as ethanol, methanol, or dioxane to the reaction. Small amounts of solvent (0.5 moles solvent per mole cresol) increased reaction rates by reducing the viscosity and improving homogeneity. Further increases in solvent, however, diluted the reagent concentrations to an extent that decreased the rates of reaction. 2.3.5. Molecular weight and molecular weight distribution calculations The molecular weights and molecular weight distributions of phenolic oligomers have been evaluated using gel permeation chromatography (GPC),24,25 NMR spectroscopy,26 vapor-pressure osmometry,27 intrinsic viscosity,28 and more recently by matrix assisted laser desorption/ionization time of flight mass spectrometry (MALDI TOFMS). 29 24 T. Yoshikawa, K. Kimura and S. Fujimura, The Gel Permeation Chromatography of T. A. Yamagishi, M. Nomoto, S. Ito, S. Ishida, and Y. Nakamoto, Preparation and Phenolic Compound, Journal of Applied Polymer Science 15, 2513-2520 (1971). 25 Characterization of High Molecular Weight Novolac Resins, Polymer Bulletin 32, 501507 (1994). 26 L. E. Bogan, Jr., Determination of Cresol Novolac Copolymer and Branch Density M. G. Kim, W. L. Nieh, T. Sellers, Jr., W. W. Wilson, and J. W. Mays, Polymer Using C-13 NMR Spectroscopy, Macromolecules 24, 4807-4812 (1991). 27 Solution Properties of a Phenol-Formaldehyde Resol Resin by Gel Permeation Chromatography, Intrinsic-Viscosity, Static Light-Scattering, and Vapor Pressure Osmometric Methods, Industrial & Engineering Chemistry Research 31(3), 973-979 (1992). 28 F. L. Tobiason, C. Chandler, and F. E. Schwarz, Molecular Weight-Intrinsic Viscosity Relationships for Phenol-Formaldehyde Novolak Resins, Macromolecules 5(3), 321-325 (1972). 29 H. Mandal and A. S. Hay, M.A.L.D.I.-T.O.F. Mass Spectrometry Characterization of 4-Alkyl Substituted Phenol-Formaldehyde Novolac Type Resins, Polymer 38(26), 62676271 (1997). 15 The most widely used molecular weight characterization method has been GPC which separates compounds based on hydrodynamic volume. State-of-the-art GPC instruments are equipped with a concentration detector (e.g., differential refractometer, UV and/or IR) in combination with viscosity or light scattering. A viscosity detector provides in-line solution viscosity data at each elution volume, which in combination with a concentration measurement, can be converted to specific viscosity. Since the polymer concentration at each elution volume is quite dilute, the specific viscosity is considered a reasonable approximation for the dilute solution intrinsic viscosity. The plot of log[]M vs. elution volume (where [] is the intrinsic viscosity) provides a universal calibration curve where absolute molecular weights of a variety of polymers can be obtained. Unfortunately, many reported analyses for phenolic oligomers and resins are simply based on polystyrene standards and only provide relative molecular weights instead of absolute numbers. Dargaville et al.30 and Yoshikawa et al.24 recognized the difficulties in obtaining accurate GPC molecular weights of phenolic resins due to large amounts of isomers and their associated differences in hydrodynamic sizes. These workers generated GPC calibration curves using a series of low molecular weight model novolac compounds: (1) linear compounds with only ortho-ortho methylene linkages, (2) compounds with orthoortho methylene linked backbones and where each unit had a pendent para-para methylene linked unit, and (3) compounds with ortho-ortho methylene linked backbones and where each unit had a pendent para-ortho methylene linked unit.30 For a given molecular weight, the hydrodynamic volume of oligomers with only the ortho-ortho methylene links was smaller than the others. It was reasoned that the reduced hydrodynamic volume was caused by extra intramolecular hydrogen bonding in highortho novolacs, which was a similar argument to that suggested previously by Yoshikawa et al.24 Based on the GPC calibration curves of the model compounds and their known 30 T. R. Dargaville, F. N. Guerzoni, M. G. Looney, D. A. Shipp, D. H. Solomon, and X. Zhang, Determination of Molecular Weight Distribution of Novolac Resins by Gel Permeation Chromatography, Journal of Polymer Science. Part A: Polymer Chemistry 35(8), 1399-1407 (1997). 16 chemical structures, simulated calibration curves were generated for idealized 100% ortho-para methylene linked oligomers and for 100% para-para linked oligomers. GPC chromatograms for a series of commercial novolacs, including resins with statistical distributions of ortho and para linkages and high ortho novolac resins were measured. Carbon-13 NMR provided the relative compositions of o,o, o,p, and p,p linked methylene groups. Molecular weights from GPC were calculated by considering the fractions of each type of linkage multiplied by the MW calculated from each of the 3 o,o (experimental), o,p (simulated), and o,o (simulated) GPC calibration curves. Good agreement was found between the resin molecular weights measured from 1H NMR and the interpolated GPC numbers for oligomers up to an average of 4-5 units per chain, whereas more deviation was observed for higher molecular weights. This was attributed to complicated intramolecular hydrogen bonding in the higher molecular weight materials. Another factor may be that branching becomes significant in the higher molecular weight materials and the hydrodynamic volume effects of architecture are also complicated. 1 H NMR integrations of methylene and aromatic regions can be used to calculate the number average molecular weights of novolac resins.30 [CH2]/[Ar] = (2n-2)/(3n+2) (2. 1) where [CH2]/[Ar] is the ratio of methylene protons to aromatic protons and n is the number of phenolic units. The method is quite accurate for novolacs with less than 8 repeat units. Solution 13C NMR has been used extensively to examine the chemical structures of phenolic resins.26,31 By ratioing the integration of peaks, degree of polymerization, number average molecular weights, degrees of branching, numbers of free ortho and para positions, and isomer distributions have been evaluated. A typical 13C NMR spectrum of a novolac resin shows three regions (Table 2. 3): the methylene linkages resonate between 30 and 40 ppm; the peaks between 146-157 ppm are due to hydroxyl substituted 31 R. A. Pethrick and B. Thomson, 13C Nuclear Magnetic Resonance Studies of Phenol- Formaldehyde Resins 1-Model Compounds, British Polymer Journal 18(3), 171-180 (1986). 17 aromatic carbons; and peaks between 113 and 135 ppm represent the remainder of the aromatic carbons. Table 2. 3. Peak assignments for 13C NMR chemical shifts of phenolic resins 32,42 Chemical Shift Region (ppm) 150-156 127-135 121 116 85.9 81.4 71.1 68.2 40.8 35.5 31.5 Assignment Hydroxyl-substituted phenolic carbons Other phenolic carbons Para-unsubstituted phenolic carbons Ortho- unsubstituted phenolic carbons HO-CH2-O-CH2-OH HO-CH2-OH Para-linked dimethylene ether Ortho- linked dimethylene ether Para-para methylene linkages Para-ortho methylene linkages Ortho-ortho methylene linkages The number of remaining ortho reactive sites versus the number of para reactive sites can also be calculated using 13C NMR (Table 2. 3). Since the rates of novolac cure reactions differ with the amount of ortho versus para reactive sites available, it is of great interest to calculate these parameters. Degrees of polymerization can be calculated from quantitative 13C NMR data by considering the number of substituted (reacted) relative to unsubstituted (not yet reacted) ortho and para phenolic carbons where ([S]) is the sum of substituted ortho and para carbons and ([S]+[U]) is the total ortho and para carbons. The fraction of reacted ortho and para sites is denoted by fs (equation 2). Thus, the number average number of phenol units per chain (n) can be calculated using equation 3. This leads to a simple calculation of Mn = (n) 106 14. fs = [S]/([S] +[U]) (2. 2) (2. 3) n = 1/(1-1.5 fs) 18 FTIR and FT Raman spectroscopy have been used to characterize phenolic compounds. The lack of hydroxyl interference is a major advantage of using FT Raman spectroscopy as opposed to FTIR to characterize phenolic compounds. Two regions of interest in Raman spectra are between 2800 and 4000 cm-1 where phenyl C-H stretching and methylene bridges are observed and between 400 and 1800 cm-1.32 For a high ortho novolac resin, the phenyl C-H stretch and methylene bridge appear at 3060 and 2940 cm-1 respectively. In the fingerprint region, the main bands are 1430-1470 cm-1 representative of methylene linkages, and 600-950 cm-1 for out-of-plane phenyl C-H bonds. Phenol, mono-ortho, di- and tri-substituted phenolic rings can be monitored between 814-831 cm-1, 753-794 cm-1, 820-855 cm-1 and 912-917 cm-1 respectively. phenolic rings also absorb in the 820-855 cm-1 region. Hay et al.29 used MALDITOF mass spectrometry to determine the absolute molecular masses and endgroups of 4-phenylphenol novolac resins prepared in xylene or chlorobenzene. Peaks with a mass difference of 44 (the molecular weight of a xylene endgroup) suggested that reactions conducted in xylene included some incorporation of xylene onto the chain ends when a strong acid such as sulfuric acid was used to catalyze the reaction. By contrast, no xylene was reacted into the chain when a milder acid catalyst such as oxalic acid was used. No chlorobenzene was incorporated regardless of the catalyst used. Para substituted 2.3.6. Hydrogen bonding The abundant hydroxyl groups on phenolic resins causes these materials to form strong intra- and intermolecular hydrogen bonds. Intramolecular hydrogen bonding of phenolic resins gives rise to their hyperacidity while intermolecular hydrogen bonding facilitates miscibility with a number of materials containing electron donors such as carbonyl, amide, hydroxyl, ether and ester groups. Miscible polymer blends of novolac 32 B. Ottenbourgs, P. Adriaensens, R. Carleer, D. Vanderzande, and J. Gelan, Quantitative Carbon-13 Solid-State n.m.r. And FT-Raman Spectroscopy in Novolac Resins, Polymer 39(22), 5293-5300 (1998). 19 resins include those with some polyamides,33 poly(ethylene oxide),34 poly(hydroxyether)s,35 poly(vinyl alcohol),36 and poly(decamethylene adipate) and other poly(adipate ester)s37. The specific strength of hydrogen bonding is a function of the groups involved, e.g. hydroxyl-hydroxyl interactions are stronger than hydroxyl-ether interactions.38 The effects of intermolecular hydrogen bonding on neat novolac resins with compounds containing hydrogen acceptors, e.g., 1,4-diazabicyclo[2,2,2]octane (DABCO) and hexamethylene tetramine, were also investigated.39 Glass transition temperatures of neat resins and blends were measured using differential scanning calorimetry to assess 33 F. Y. Wang, C. C. M. Ma, and H. D. Wu, Hydrogen Bonding in Polyamide Toughened Novolac Type Phenolic Resin, Journal of Applied Polymer Science 74, 2283-2289 (1999). 34 P.P. Chu, H. D. Wu, and C. T. Lee, Thermodynamic Properties of Novolac-Type Phenolic Resin Blended with Poly(ethylene oxide), Journal of Polymer Science, Part B: Polymer Physics 36(10), 1647-1655 (1998). 35 H. D. Wu, C. C. M. Ma, and P. P. Chu, Hydrogen Bonding in the Novolac Type H. D. Wu, P. P. Chu, C.C. M. Ma, and F. C. Chang, Effects of Molecular Structure of Phenolic Resin Blended with Phenoxy Resin, Polymer 38(21), 5419-5429 (1997). 36 Modifiers on the Thermodynamics of Phenolic Blends: An Entropic Factor Complementing PCAM, Macromolecules 32(9), 3097-3105 (1999). 37 H. D. Wu, C. C. M. Ma, P.P. Chu, H. T. Tseng, and C. T. Lee, The Phase Behaviour of Novolac Type Phenolic Resin Blended with Poly(adipic ester), Polymer 39(13), 2856-2865 (1998). 38 C. C. M. Ma, H. D. Wu, and C. T. Lee, Strength of Hydrogen Bonding in the Novolac-Type Phenolic Resin Blends, Journal of Polymer Science. Part B: Polymer Physics 36(10), 1721-1729 (1998). 39 Z. Katovic and M. Stefanic, Intermolecular Hydrogen Bonding in Novolacs, Industrial & Engineering Chemistry Product Research And Development 24, 179-185 (1985). 20 the degrees of hydrogen bonding. Hydrogen-bonding interactions of novolac resins with electron donor sites such as oxygen, nitrogen, or chlorine atoms resulted in increased Tgs. The propensity for dry novolac resins to absorb water at room temperature under 100% humidity is another indication that strong hydrogen bonds form. Approximately 15 wt% water is absorbed after 4 days which corresponds to one water molecule per hydroxyl group.39 Holland et al.40 conducted dielectric measurements on novolac resins to evaluate the degrees of inter- and intramolecular hydrogen bonding. The frequency dependence of complex permittivity (*) within a relaxation region can be described with a Havriliak and Negami function (HN function, equation 4) = + (1 + (i ) ) S - (2. 4) where S and are the relaxed and unrelaxed dielectric constants, is the angular frequency, o is the relaxation time, and and are fitting parameters. The complex permittivity is comprised of permittivity () and dielectric loss (). Fitting parameters in the HN function are related to shape parameters, m and n, which describe the limiting behavior of dielectric loss () at low and high frequencies respectively. Intermolecular (characterized by m) and intramolecular (characterized by n) hydrogen bonding can be correlated with m and n values which range from 0 to 1 (where lower values correspond to stronger hydrogen bonding). For one novolac resin examined (Mn = 1526 determined via GPC using polystyrene standards, MWD = 2.6, Tg = 57C), m was 0.52 and n was 0.2. These results were considered indicative of strong intramolecular hydrogen bonding within the novolac structures. 2.3.7. Novolac crosslinking with Hexamethylene Tetramine (HMTA) The most common crosslinking agent for novolac resins is HMTA which provides a source of formaldehyde. Novolac resins prepared from a P/F ratio of 1/0.8 can be cured 40 C. Holland, W. Stark, and G. Hinrichsen, Dielectric Investigations on Novolac Phenol-Formaldehyde Resins, Acta Polymer 46, 64-67 (1995). 21 with 8-15 wt % HMTA, although it has been reported that 9-10 wt % results in networks with the best overall performance.5 2.3.7.1. Initial reactions of novolacs with HMTA The initial cure reactions of a novolac with HMTA were studied by heating reactants at 90C for 6 hours, then raising the temperature incrementally to a maximum of 205C. This leads to mostly hydroxybenzylamine and benzoxazine intermediates (Figure 2. 13).41,42 Hydroxybenzylamines form via repeated electrophilic aromatic substitutions of the active phenolic ring carbons on the methylenes of the HMTA (and derivatives of HMTA). Since novolac resins form strong intermolecular hydrogen bonds with electron donors, a plausible mechanism for the initial reaction between novolac and HMTA involves hydrogen bonding between phenolic hydroxyl groups and an HMTA nitrogen. Such hydrogen bonding can lead to proton transfer where a phenolate ion is generated. A negatively charged ortho or para carbon can attack the methylene carbon next to the positively charged nitrogen on HTMA which results in cleavage of a C-N bond. Benzoxazines form by nucleophilic attack of the phenolic oxygen on HMTAphenolic intermediates. Upon further reaction, methylene linkages form as the major product of both types of intermediates through various thermal decomposition pathways. 41 P. W. Kopf and E. R. Wagner, Formation and Cure of Novolacs-NMR Study of Transient Molecules, Journal of Polymer Science: Polymer Chemistry Edition 11(5), 939-960 (1973). 42 S. A. Sojka, R. A. Wolfe, and G. D. Guenther, Formation of Phenolic Resins: Mechanism and Time Dependence of the Reaction of Phenol and Hexamethylenetetramine as Studied by Carbon-13 Nuclear Magnetic Resonance and Fourier Transform Infrared Spectroscopy, Macromolecules 14,1539-1543 (1981). 22 N OH OH N N N OH O+ N HN N N n n OH OH N OH OH OH OH N n HN N OH NH n HN N N - HN NH - HN NH NH NH - H2O OH OH N OH OH O N OH n HO n benzoxazine tris-hydroxybenzylamine Figure 2. 13. Initial reaction of novolac and HMTA via a hydrogen bonding mechanism Since a small amount of water is always present in novolac resins, it has also been suggested that some decomposition of HMTA proceeds by hydrolysis, leading to the elimination of formaldehyde and amino-methylol compounds (Figure 2. 14).43 Phenols can react with the formaldehyde elimination product to extend the novolac chain or form methylene bridged crosslinks. Alternatively, phenol can react with amino-methylol 43 Y. Ogata and A. Kawasaki in J. Zabicky ed., Equilibrium Additions to Carbonyl Compounds, The Chemistry of the Carbonyl Group, Interscience, London, Vol. 2. 1970. 23 intermediates in combination with formaldehyde to produce ortho- or parahydroxybenzylamines (i.e., Mannich type reactions). + H2O -CH2O N HN N N N N N + H2O - NH(CH2OH)2 NH NH NH + H2O H2N CH2 NH2 HN + HO CH2 NH CH2 OH Figure 2. 14. Decomposition of HMTA Reaction pathways involved in the curing of novolacs with HMTA have been extensively investigated by Solomon and coworkers.44,45,46,47,48,49,50,51 In a series of model 44 T. R. Dargaville, P. J. De Bruyn, A. S. C. Lim, M. G. Looney, A. C. Potter, and D. H. Solomon, Chemistry of Novolac Resins. II, Reaction of Model Phenols with Hexamethylenetetramine, Journal of Polymer Science. Part A. 35, 1389-1398 (1997). 45 X. Zhang, M. G. Looney, D. H. Solomon, and A. K. Whittaker, The Chemistry of Novolac Resins: 3. 13C and 15N n.m.r. Studies of Curing with Hexamethylenetetramine, Polymer 38(23), 5835-5948 (1997). 46 X. Zhang, A. C. Potter, and D. H. Solomon, The Chemistry of Novolac Resins-V. X. Zhang and D. H. Solomon, The Chemistry of Novolac Resins-VI. Reactions Reactions of Benzoxazine Intermediates, Polymer 39(2), 399-404 (1998). 47 Between Benzoxazine Intermediates and Model Phenols, Polymer 39(2), 405-412 (1998). 48 X. Zhang, A. C. Potter, and D. H. Solomon, The Chemistry of Novolac Resins: Part 7. Reactions of para-Hydroxybenzylamine Intermediates, Polymer 39(10), 1957-1966 (1998). 49 X. Zhang, A. C. Potter, and D. H. Solomon, The Chemistry of Novolac Resins: Part 8. Reaction of para-Hydroxybenzylamines with Model Compounds, Polymer 39(10), 1967-1975 (1998). 50 X. Zhang and D. H. Solomon, The Chemistry of Novolac Resins: 9. Reaction Pathways Studied via Model Systems of ortho-Hydroxybenzylamine Intermediates and Phenols, Polymer 39(24), 6153-6162 (1998). 24 studies where 2,6-xylenol and/or 2,4-xylenol were reacted with HMTA, these workers found that the types of linkages formed were affected by the initial chemical structure of the novolac, i.e. amount of ortho vs. para reactive positions, the amount of HMTA, and the pH. Reaction intermediates for the cure were identified, mostly via FTIR, 13C NMR and 15N NMR. As previously described, the main intermediates generated from the initial reaction between ortho reactive sites on novolac resins and HMTA are hydroxybenzylamines and benzoxazines.45 Triazines, diamines, and in the presence of trace amounts of water, benzyl alcohols and ethers also form (Figure 2. 15). Similar intermediates, with the exception of benzoxazines, are also observed when para sites react with HMTA. The thermolysis rates to form methylene linkages depend on the stabilities of hydroxybenzylamine and benzoxazine intermediates. Comparatively, ortho-linked hydroxybenzylamine intermediates are more stable than para-linked structures because six-membered rings can form between the nitrogen and phenolic hydroxyl groups via intramolecular hydrogen bonding. For the same reason, benzoxazines are the most stable intermediates and decompose only at higher temperatures (185C).46 If a high orthonovolac resin is cured with HMTA, the reaction occurs at lower temperatures due to formation of relatively unstable intermediates and the amount of side products is low. If, however, a typical novolac is used, the reaction temperature must be higher to decompose the more thermally stable ortho intermediates, and the amount of nitrogen containing side products is significantly higher.47,48 51 A. S. C. Lim, D. H. Solomon, and X. Zhang, Chemistry of Novolac Resins. X. Polymerization Studies of HMTA and Strategically Synthesized Model Compounds, Journal of Polymer Science Part A: Polymer Chemistry 37, 1347-1355 (1999). 25 OH N N N N heat Most Stable Intermediates hydroxybenzylamine OH NH OH N N Less Stable Intermediates In the presence of water triazine OH N N OH benzyl alcohol OH OH 2 3 HO benzoxazines OH O N ether OH OH O diamine OH N HO N OH OH Figure 2. 15. Possible reaction intermediates for reaction of 2,4-xylenol with HTMA If only ortho sites are available for reaction, the amount of hydroxybenzylamine vs. benzoxazine generated is largely dependent on the novolac/HMTA ratio. Hydroxybenzylamine is favored when the HMTA content is low whereas more benzoxazine is formed at higher HMTA concentrations. This is expected since only one HMTA carbon is needed per reactive ortho position in the formation of hydroxybenzylamine, but the formation of benzoxazine requires three HMTA carbons per two reactive ortho positions. The HMTA concentration therefore is one key in determining the structure of the resulting networks. Lower HMTA contents leading to more hydroxybenzylamine intermediates means that lower temperatures can be used for decomposition into methylene bridges and correspondingly lower levels of side products form under such conditions. 26 2.3.7.2. Hydroxybenzylamine and Benzoxazine decompositions in novolac/HMTA cures Thermal Decomposition of Hydroxybenzylamines. Depending on the concentration of HMTA and mobility of the system, hydroxybenzylamine and benzoxazine intermediates react by a number of pathways to form crosslinked novolac networks. Trishydroxybenzylamines eliminate benzoquinone methide between 90-120C to form bishydroxybenzylamines, which decompose to methylene linkages with elimination of CH2=NH at higher temperatures (Figure 2. 16).47 OH N - O CH2 OH NH - CH2 NH OH OH 3 2 Figure 2. 16. Thermal decomposition of hydroxybenzylamine Thermal Decomposition of Benzoxazines. Thermal decomposition of benzoxazines does not occur substantially until the temperature reaches ~160C. This begins with proton transfer from a phenolic hydroxyl group to a nitrogen. Cleavage of the C-O bond with water generates a tertiary hydroxymethylamine which can eliminate formaldehyde, then CH2=NH, to form methylene linkages (Figure 2. 17A). Alternatively, C-N bond cleavage in the benzoxazine leads to elimination of a benzoquinone methide, which can react with phenols to primarily yield the product methylene bridged species (Figure 2. 17B).46 Further decomposition of benzoxazines can also lead to a variety of side products in small amounts. 27 OH O N OH N CH2 O O NH OH -HCHO OH NH OH A H2O OH - CH2=NH OH B H2C O OH OH + O NH Figure 2. 17. Thermal decomposition of benzoxazine Reactions of Benzoxazines with Phenols. In the presence of 2,4-xylenol, benzoxazine intermediates react at lower temperatures (~90C) to form hydroxybenzylamines (Figure 2. 18), which can then decompose to ortho-ortho methylene linkages (as described in Figure 2. 16).47 The reaction between benzoxazine and free ortho reactive positions on 2,4-xylenol occurs via electrophilic aromatic substitution facilitated by hydrogen bonding between benzoxazine oxygen and phenolic hydroxyl groups (Figure 2. 18). O H O N OH HO OH N OH Figure 2. 18. Reaction of benzoxazines and 2,4-xylenol 28 The reaction of benzoxazine in the presence of 2,6-xylenol does not occur until ~135C, presumably because the hydrogen bonded intermediate depicted for the 2,4xylenol reaction (Figure 2. 18) cannot occur. All three types of linkages are obtained in this case. Para-para methylene linked 2,6-xylenol dimers, obtained from reaction of 2,6xylenol with formaldehyde, formed in decomposition of the benzoxazine, (or with other by-products of that process) dominate. Possible side products from benzoxazine decomposition include formaldehyde and CH2=NH, either of which may provide the source of methylene linkages. The amount of ortho-para linkages, formed by reaction of 2,6-xylenol with benzoxazine is low. Ortho-ortho methylene linked products presumably form by a decomposition pathway from benzoxazine (as in Figure 2. 17). HMTA Crosslinking Reactions of Novolacs Containing Both Ortho and Para Reactive Sites. When both ortho and para positions on novolac materials are available for reaction with HMTA, ortho-ortho, ortho-para, and para-para methylene linkages form through several pathways. This section will address crosslinking reaction pathways where components which have been eliminated as by-products re-enter the reactions. In particular, reactions of quinone methides, formaldehyde and imine will be discussed. We will also describe exchange reactions between hydroxybenzylamine intermediates with phenolic methylol derivatives which lead to methylene bridged final products. Exchange reactions between two different hydroxybenzylamine intermediates, which lead to primarily ortho-ortho linked products, are also important. In one model reaction where tris(para-hydroxybenzyl)amine was heated to 205C in the presence of 2,4-xylenol (1:1 ratio), the ortho-ortho, ortho-para, and para-para methylene bridge ratio in the products was found to be 44%, 14%, and 38% respectively (Figure 2. 19).49 This model study demonstrated the importance of benzoquinone methide intermediates in the formation of various products in the novolac/HMTA curing reaction. Formaldehyde, CH2=NH, and water liberated during the cure reaction also affect the reaction pathways (pathways 3, 4 and 5). Approximately 4% of 1,2-bis(parahydroxyphenyl)ethane was also observed, presumably formed through dimerization of two quinone methides. 29 OH N HO + 3 para-para (44%) + ortho-para (14%) + ortho-ortho(38%) para-para formation: N HO 3 1. O HO NH 2 - CH2 - CH2 NH HO OH ortho-para formation: OH OH 2. CH2 O OH O or hydroxybenzylamine O HCH OH CH2OH -H C H 3. NH HO 2 OH or N HO 3 O OH -NH3 ortho-ortho formation: N HO 3 OH N HO CH2NH2 NH HO 2 OH N OH OH 4. CH2 NH OH OH OH NH OH OH NH OH OH mainly OH CH2OH OH OH O O OH OH -H C H 5. dimerize - H2O Figure 2. 19. Reaction pathways for formation of ortho-ortho, ortho-para, and para-para through the reaction of para-trishydroxybenzylamine and 2,4-xylenol 30 Para-para methylene linkages appeared first via hydroxybenzylamine decomposition at lower temperatures (pathway 1 in Figure 20). Ortho-para methylene linkages also formed at the lower reaction temperatures (pathway 2). Since the only source of an ortho- methylene linked phenol product was the 2,4-xylenol starting material, these mixed products must have formed by reaction of 2,4-xylenol with either a para-hydroxybenzylamine or with a quinone methide eliminated in pathway 1. Orthopara methylene linkages also formed at higher reaction temperatures, which were attributed to exchange reactions between a methylol derivative of 2,4-xylenol and a hydroxybenzylamine (pathway 3). Ortho-ortho methylene linkages formed only at higher temperatures via hydroxybenzylamine exchange and methylol dimerization reactions described in pathways 4 and 5. The reactions depicted in pathway 4 involved sequential exchanges between para- and ortho- substituted intermediates through nucleophilic substitutions on hydroxybenzylamines. Since the amount of ortho-para linked products was low, it was suggested that the major product of pathway 4 was the ortho-ortho linkage. This is reasonable since the equilibrium of these exchange reactions lies toward ortho-hydroxybenzylamines where hydrogen bonding provides stability. These more thermally stable hydroxybenzylamines then decompose at higher temperatures to form ortho-ortho linkages. Small amounts of various phenolic side products incorporating groups such as imines, amides, ethers and ethanes into the networks also form. A number of these side products undergo further reactions which eventually lead to methylene linkages. Some side products generally remain in the networks even after heating at 205C. Solomon et al. also investigated HMTA-phenolic reactions with somewhat larger model compounds (e.g., 2 and 4 ring compounds), and established that similar reaction pathways to those described previously occurred.51 For these model compounds (as opposed to 1-ring model compounds) that are more representative of typical oligomeric systems, increased molecular weight favored the formation of hydroxybenzylamines, but not benzoxazines. This was suggested to be a steric effect. Other crosslinking agents that provide sources of formaldehyde for methylene linkages include paraformaldehyde and trioxane, but these have only achieved limited importance. Quantitative 13C solid-state NMR and FT-Raman spectroscopy were used to 31 monitor the cure reactions of a high ortho-novolac resin using paraformaldehyde under different conditions.32 The weight percent paraformaldehyde needed to achieve the maximum crosslinking (1.5 moles formaldehyde per mole phenol) for the particular novolac examined (Mn=430 g/mol determined via 13C NMR) was calculated to be 17.76 wt %. Eleven weight percent formaldehyde (1.18 moles formaldehyde relative to phenol) was used in these studies so that phenol sites were in excess. The degree of conversion was assessed by comparing the formaldehyde to phenol ratio in the polymer to 1.18. As expected, higher temperatures and/or pressures lead to higher reaction conversions. However, none of these reaction conversions reached 100%, and this was attributed to a lack of mobility. 2.4. Resole resins and networks 2.4.1. Resole resin syntheses Resoles are prepared under alkaline conditions using an excess of formaldehyde over phenol (1:1 to 3:1) at typical temperatures of 60-80C. The basic catalysts commonly used are NaOH, Na2CO3, KOH, K2CO3, Ba(OH) 2, R4NOH, NH3, RNH2 and R2NH.6 In aqueous solutions, ammonia and HMTA are easily hydrolyzed to amines and also catalyze resole syntheses. Typical resole resins comprise a mixture of monomers, dimers, trimers, and small amounts of higher molecular weight oligomers with multiple methylol functional groups. Resole syntheses entail substitution of formaldehyde (or formaldehyde derivatives) on phenolic ortho and para positions (Figure 2. 20) followed by methylol condensation reactions which form dimers and oligomers (Figure 2. 22). Under basic conditions, phenolate rings are the reactive species for electrophilic aromatic substitution reactions. A simplified mechanism is generally used to depict the formaldehyde substitution on the phenol rings (Figure 2. 20). It should be noted that this mechanism does not account for pH effects, the type of catalyst, or the formation of hemiformals. Mixtures of mono-, di-, and tri-hydroxymethyl substituted phenols are produced. 32 OH + - OOH O O + H2O O + CH2 O O + CH2 O O CH2OH O OCH2OH O- H CH2OH CH2OH Figure 2. 20. Mechanism of resole synthesis Phenol reacts with formaldehyde in either the ortho or the para position to form mono-hydroxymethyl substituted phenols, which further react with formaldehyde to form di- and tri-hydroxymethyl substituted phenols (Figure 2. 21). OH CH2OH OH CH2OH CH2O OH CH2O CH2OH CH2O OH HOCH2 OH CH2OH CH2O OH CH2O HOCH2 CH 2OH CH2OH CH2O CH 2OH CH2O Figure 2. 21. Reaction pathways for phenol/formaldehyde reactions under alkaline conditions 33 Condensation reactions between two hydroxymethyl substituents eliminate water to form ether linkages (Figure 2. 22A) or eliminate both water and formaldehyde to form methylene linkages (Figure 2. 22B). Ether formation is favored under neutral or acidic conditions and up to ~130C above which formaldehyde departs and methylene linkages are generated. The methylene linkage formation reaction, which eliminates water and formaldehyde, is more prevalent under basic conditions. Condensation reactions between hydroxymethyl groups and reactive ortho or para ring positions also lead to methylene bridges between phenolic rings (Figure 2. 22C). Relative reactivities of hydroxymethyl substituted phenols with formaldehyde and with other hydroxymethyl substituted phenols appear to be strongly dependent on interactions between ortho-methylol groups and the phenolic hydroxyl position. Hydroxymethyl condensation reactions under basic conditions strongly favor the formation of para-para and ortho-para methylene linkages. A OH + CH2OH OH + CH2OH OH - H2O CH2OH OH HO CH2 O CH2 OH B - H2O - CH2O CH2OH HO CH2 OH C OH + CH2OH OH CH2OH OH - H2O OH CH2OH Figure 2. 22. Condensation of hydroxymethyl groups 34 Quinone methides are the key intermediates in both resole resin syntheses and crosslinking reactions. They form by dehydration of hydroxymethylphenols or dimethylether linkages (Figure 2. 23). OH CH2 OH -H2O O CH2 OH CH2 O CH2 OH -H2O O 2 CH2 Figure 2. 23. Dehydration of methylols or benzylic ethers to form quinone methides Resonance forms for quinone methides include both quinoid and benzoid structures (Figure 2. 24). The oligomerization or crosslinking reaction proceeds by nucleophilic attack on the quinone methide carbon. O- O CH2 OCH2+ O CH2 CH2+ Figure 2. 24. Resonance of quinone methides Ortho-quinone methides are difficult to isolate due to their high reactivity which leads to rapid Diels-Alder dimerization or trimerization (Figure 2. 25). At 150C, a partial retro-Diels-Alder reaction of the trimer can occur to form ortho-quinone methide and bis(2-hydroxy-3,5-dimethylphenyl) ethane (dimer).52 52 K. Lenghaus, G. G. Qiao, and D. H. Solomon, Model Studies of the Curing of Resole Phenol-Formaldehyde Resins Part 1. The Behavior of ortho Quinone Methide in a Curing Resin, Polymer 41, 1973-1979 (2000). 35 O O CH2 O O CH2 O O O 2 dimer 150 C trimer o Figure 2. 25. Dimer and trimer structures of ortho quinone methides Base catalyzed phenol-formaldehyde reactions exhibit second order kinetics (equation 2.5). Several alkylphenols such as cresols also follow this rate equation. rate = k[phenolate][formaldehyde] (2. 5) The rate constants for various hydroxymethylation steps (Figure 2. 21) have been evaluated by several groups (Table 2. 4)53,54,55 and more recently by Grenier-Loustalot et al.57 53 J. H. Freemann and C. W. Lewis, Alkaline Catalyzed Reaction of Formaldehyde and the Methylol of Phenol-A Kinetic Study, Journal of American Chemical Society 76(8), 2080-2087 (1954). 54 A. A. Zsavitsas and R. D. Beaulieu, Base Catalyzed Hydroxymethylation of Phenol by Aqueous Formaldehyde: Kinetic and Mechanism, Journal of Polymer Science AI 6, 2451 (1969). 55 K. C. Eapen and L. M. Yeddanapalli, Kinetics and Mechanism of Alkaline Catalyzed Addition of Formaldehyde to Phenol and Substituted Phenol, Makromolekulare Chemie 119, 4 (1968). 36 Table 2. 4. Relative positional reaction rates in base catalyzed phenol-formaldehyde reaction Relative reaction rates Freeman Zsavitsas Eapen et al.53 et al.54 et al.55 1.00 1.00 1.00 1.18 1.66 1.39 0.71 1.73 7.94 1.09 1.98 1.80 0.79 1.67 3.33 1.46 1.75 3.05 0.85 2.04 4.36 phenol phenol 2-hydroxymethylphenol 4-hydroxymethylphenol 2,6-dihydroxymethylphenol 2,4-dihydroxymethylphenol 2,4-dihydroxymethylphenol 2-hydroxymethylphenol 2-hydroxymethylphenol 4-hydroxymethylphenol 2,4-dihydroxymethylphenol 2,4,6-trihydroxymethylphenol 2,6-dihydroxymethylphenol 2,4,6-trihydroxymethylphenol Some consensus observations for reactions conducted at 30C indicate that the para reactive site on phenol is slightly more reactive than ortho reactive sites due to higher electron density on the para position. In addition, ortho hydroxymethyl substituents significantly activate the rings toward further electrophilic addition of formaldehyde. This is especially pronounced for 2,6-dihydroxymethylphenols. Orthohydroxymethyl substituents are proposed to stabilize the quinoid resonance form via hydrogen bonding between the phenolic hydroxyl and ortho-hydroxymethyl groups in basic aqueous media (Figure 2. 26). This intramolecular stabilization activates the para position by intensifying electron density on the para carbon. This reasoning, however, does not explain the reduced reactivity reported for 2,4-dihydroxymethylphenol. OH H OH OH O O O Figure 2. 26. Quinoid resonance forms activating the para ring position More recently, the reaction advancement of resole syntheses (pH=8 and 60C) was monitored using high performance liquid chromatography, 37 13 C NMR, and chemical assays.56,57 The disappearance of phenol and the appearances of various hydroxymethyl substituted phenolic monomers and dimers has been measured. By assessing residual monomer as a function of reaction time, this work also demonstrated the unusually high reactivity of 2,6-dihydroxymethylphenol. The rate constants for phenolic monomers towards formaldehyde substitution have been measured (Table 2. 5). Table 2. 5. Second order rate constants for reaction of phenolic monomers with formaldehyde57 Compound phenol 2-hydroxymethylphenol 4-hydroxymethylphenol 2,4-dihydroxymethylphenol 2,6-dihydroxymethylphenol k (mol-1h-1)x102 5.1 9.9 10.7 8.6 13.0 As the reactions proceed, the disappearance of phenol is delayed due to competitions for reaction with formaldehyde between phenol and faster-reacting hydroxymethyl substituted phenols. Since the limiting step for phenolic reactions is formaldehyde substitution on phenol, particularly on the ortho positions, reaction conditions should be oriented toward fast phenol/formaldehyde reaction during the initial stages of reaction. Competition also exists between formaldehyde substitution reactions and condensation reactions between rings. Condensation reactions between two orthohydroxymethyl substituents are the least favorable condensation pathway. Depending on the reaction conditions, substitutions occur predominately in the earlier stages of reaction and condensations become the major reactions in later stages.56 56 M. F. Grenier-Loustalot, S Larroque, P. Grenier, J. Leca, and K. Bedel, Phenolic Resins: 1. Mechanisms and Kinetics of Phenol and of the First Polycondensations Towards Formaldehyde in Solution, Polymer 35(14), 3046-3054 (1994). 57 M. F. Grenier-Loustalot, S. Larroque, P. Grenier, and D. Bedel, Phenolic Resins: 3. Study of the Reactivity of the Initial Monomers Towards Formaldehyde at Constant pH, Temperature and Catalyst type, Polymer 37(6), 939-953 (1996). 38 As described previously, condensation reactions of hydroxymethyl substituents strongly favor the formation of para-para and ortho-para linkages.58,59,60 Various hydroxymethyl substituted phenolic monomers were heated in the absence of formaldehyde (60C, pH=8.0) to investigate condensation reactions under typical resole synthesis conditions but without formaldehyde substitution.59 Only methylene linkages were observed under the particular experimental conditions. Highly substituted dimers were predominant in the product mixture since monomers with more hydroxymethyl substituents had higher probabilities for condensation. Ortho-hydroxymethyl groups only condensed with substituents in the para position, and therefore no ortho-ortho methylene linkages were observed. Para-hydroxymethyl substituents, on the other hand, reacted with either ortho or para hydroxymethyl substituents or reactive ring positions, but preferentially with para-hydroxymethyl groups. 13 C and 1 H NMR monitoring condensation reactions of resole resins comprised of two to five phenolic units showed that, with the exception of one trimer containing a dimethylene ether linkage, only parapara and ortho-para methylene linkages formed. Upon further reaction, especially at higher temperatures (70-100C), hydroxymethylated compounds reacted to form almost exclusively para-para and orthopara methylene linkages. Since the key intermediates for the condensation of hydroxymethylphenols are quinone methides, the formation of para-para and ortho-para methylene linkages is attributed to exclusive formation of a para-quinone methide intermediate (Figure 2. 27).7 This is attributed to intramolecular hydrogen bonding 58 B. Mechin, D. Hanton, J. Le Goff and J. P. Tanneur, HPLC and NMR Identification of the Main Polynuclear Constituents of Resol-Type Phenol-Formaldehyde Resins, European Polymer Journal 22(2), 115-124 (1986). 59 M. F. Grenier-Loustalot, S. Larroque, P. Grenier, and D. Bedel, Phenolic Resins: 4. Self-Condensation of Methylolphenols in Formaldehyde-Free Media, Polymer 37(6), 955-964 (1996). 60 L. Prokai, Separation and Identification of Phenol-Formaldehyde Condensates by Gas Chromatography-Mass Spectrometry. II. Base-Catalyzed Condensation Products, Journal of Chromatography 333(1), 91-98 (1985). 39 between both ortho hydroxymethyl substituents with quinone methide oxygen, which lead to stable para-quinone methide structures. The para-quinone methide intermediates then react with ortho or para reactive positions to form ortho-para and para-para methylene linkages; or the quinone methide reacts with hydroxymethyl groups to form ethers which further advance to methylene linkages. OH HOCH2 CH2OH HOCH2 O CH2OH HOCH2 O CH2 + CH2OH CH2 CH2OH Figure 2. 27. Preferential formation of para quinone methides The mechanisms for model condensation reactions of para-hydroxymethyl substituted phenol (and therefore para-quinone methide) with reactive ortho positions are described in Figure 2. 28. The phenolate derivatives react with para-quinone methide via a Michael type addition to form methylene linkages (Figure 2. 28 A). Hydroxyl groups on methylol can also attack methide carbons to form dibenzyl ether linkages which subsequently eliminate formaldehyde to form methylene links (Figure 2. 28 B). An ipso substitution where a nucleophilic ring carbon having a hydroxymethyl substituent attacks a quinone methide has also been postulated to generate methylene linkages (Figure 2. 28 C). 40 O H - OH CH2 HO O- O CH2 CH2OH CH2OH A CH2OH + O B CH2 HO CH2 O CH2 O- - CH2O - O O CH2 OH C - O CH2 CH2OH Figure 2. 28. Reactions of a quinone methide with a hydroxymethyl substituted phenolate The reaction conditions, formaldehyde to phenol ratio, and the concentration and type of catalyst govern the mechanisms and the kinetics of resole syntheses. Higher formaldehyde to phenol ratios accelerate the reaction rates. This is to be expected since phenol/formaldehyde reactions follow second order kinetics. Increased hydroxymethyl substitution on phenols due to higher formaldehyde compositions also leads to more condensation products.56 The amount of catalyst and the pH of reaction determine the extent of phenolate formation. Phenol/formaldehyde mixtures (F/P=1.5, 60C) did not react at pH=5.5 and reaction rate increased as the pH was increased to about 9.25.56 There was a linear relationship between the rate constant and the [NaOH]/[phenol] ratio (between pHs 5.5 and 9.25). It was suggested that a limiting pH of approximately 9 exists, above which an increase in pH does not enhance the rate of reaction due to saturation of phenolate anions. Considerable Canizarro side reactions occurred on formaldehyde at pH>10.56,61 The type of catalyst influences the rate and the mechanism of reactions. Reactions catalyzed with both monovalent and divalent metal hydroxides, KOH, NaOH, LiOH, and Ba(OH)2, Ca(OH)2 and Mg(OH)2, showed that both valence and ionic radius 61 R. A Haupt and T. Sellers, Characterization of Phenol-Formaldehyde Resol Resins, Industrial & Engineering Chemistry Research 33(3), 693-697 (1994). 41 of hydrated cations affect the formation rate and final concentrations of various reaction intermediates and products.62 For the same valence, a linear relationship was observed between formaldehyde disappearance rate and ionic radius of hydrated cations where larger cation radii gave rise to higher rate constants. In addition, irrespective of the ionic radii, divalent cations lead to faster formaldehyde disappearance rates than monovalent cations. For the proposed mechanism where an intermediate chelate participates in the reaction (Figure 2. 29), an increase in positive charge density in smaller cations was suggested to improve the stability of the chelate complex, and therefore, decrease the rate of the reaction. The radii and valence also affect the formation and disappearance of various hydroxymethylated phenolic compounds which dictate the composition of final products. Na+ O- +Na + CH2 O O O + CH O or O H Na+ H 2 CH3 OH O CH2O Na -+ O- +Na CH2OH OH CH2OH + O- +Na Figure 2. 29. Reaction mechanism of phenol and formaldehyde using base catalyst involving the formation of chelate Tetraalkylammonium hydroxides have slightly lower catalytic activities than NaOH in resole syntheses. Increased alkyl length on tetraalkylammonium ions (larger ionic radii) decreased the catalytic activity. Contrary to the chelating effect, the reduced activity observed with tetraalkylammonium hydroxides was attributed to screening effects of alkyl groups. Water solubility was limited to resole resins prepared with 62 M. F. Grenier-Loustalot, S. Larroque, D. Grande, P. Grenier, and D. Bedel, Phenolic Resins: 2. Influence of Catalyst Type on Reaction Mechanisms and Kinetics, Polymer 37(8), 1363-1369 (1996). 42 tetramethylammonium hydroxide and tetraethylammonium hydroxide. These catalysts also give rise to resins with longer gelation times. Resole syntheses catalyzed with various amounts of triethylamine (pH adjusted to 8 using NaOH) and various pHs (pH = 8.0, 8.23 and 8.36) were monitored.63 As expected, shorter condensation times, faster reaction rates, and higher advancement in polymerizations were reached with increased catalyst concentrations. The pH, on the other hand, did not affect these parameters significantly. The reaction mechanisms differed when NaOH was used to adjust the pH since the hydroxide formed phenolate ions which favored para addition reactions. In the absence of NaOH, free phenolic hydroxyl groups formed complexes with triethylamine to promote ortho substitution. 2.4.2. Crosslinking reactions of resole resins Resole resins are generally crosslinked under neutral conditions between 130 and 200C or in the presence of an acid catalyst such as hydrochloric acid, phosphoric acid, p-toluenesulfonic acid, and phenolsulfonic acid under ambient conditions.5 The mechanisms for crosslinking under acidic conditions are similar to acid catalyzed novolac formation. Quinone methides are the key reaction intermediates. Further condensation reactions in resole resin syntheses under basic conditions at elevated temperatures also lead to crosslinking. The self-condensation of ortho-hydroxymethyl substituents and the condensation between this substituent with ortho or para reactive sites were investigated under neutral conditions.52 2-Hydroxymethyl-4,6-dimethylphenol was reacted 1) alone, 2) in the presence of 2,4-xylenol, and 3) in the presence of 2,6-xylenol. The rates of methylene versus dimethylether formation between rings at 120C were monitored as a function of time and the percent yields after 5 hours were recorded (Table 2. 6). The ether linkage was more prevalent in the self-condensation of 2-hydroxymethyl-4,6-dimethylphenol. Possibly the 5% methylene bridged product formed via ipso substitution of an ortho 63 G. Astarloa-Aierbe, J.M. Echeverria, A. Vazquez, and I. Mondragon, Influence of the Amount of Catalyst and Initial pH on the Phenolic Resole Resin Formation, Polymer 41, 3311-3315 (2000). 43 quinone methide electrophile onto the methylene position of another ring. Essentially no differences in product composition were observed between the 2-hydroxymethyl-4,6dimethylphenol self-condensation and reaction of this compound in the presence of 2,6xylenol. The formation of methylene linkages proceeded much more favorably in the presence of 2,4-xylenol. Moreover, increases in 2,4-xylenol concentrations further This suggests vacant ortho positions are increased the methylene linkage yield. significantly more reactive than para reactive sites in reactions with ortho-quinone methide. These model reactions provide further evidence supporting quinone methides as the key reactive intermediates. Table 2. 6. % yield of methylene and ether linkages of 2-hydroxylmethyl-4,6dimethylphenol self-reaction, 1:1 with 2,4-xylenol, and 1:1 with 2,6-xylenol. Methylene % yield Self-reaction with 2,4-xylenol with 2,6-xylenol 5 38 5 Ether % yield 80 65 80 In addition to methylene and dimethylether linkages, cured networks contain ethane and ethene linkages (Figure 2. 30). These side products are proposed to form through quinone methide intermediates. OH OH OH OH Figure 2. 30. Ethane and ethene linkages derived from quinone methide structures 44 Crosslinking resoles in the presence of sodium carbonate or potassium carbonate lead to preferential formation of ortho-ortho methylene linkages.64 Resole networks crosslinked under basic conditions showed that crosslink density depends on the degree of hydroxymethyl substitution, which is affected by the formaldehyde to phenol ratio and the reaction time, the type and concentration of catalyst (uncatalyzed, with 2% NaOH, with 5% NaOH).65 As expected, NaOH accelerated the rates of both hydroxymethyl substitution and methylene ether formation. Significant rate increases were observed for ortho substitutions as the amount of NaOH increased. The para substitution, which does not occur in the absence of the catalyst, formed only in small amounts in the presence of NaOH. . 2.4.3. Resole characterization A number of analytical techniques such as Fourier transform infrared spectroscopy (FTIR),66,67 13 C NMR,68,69 solid-state 13 C NMR,70 gel permeation 64 B. D. Park and B. Riedl, C-13-NMR Study of Cure-Accelerated Phenol- Formaldehyde Resins with Carbonates, Journal of Applied Polymer Science 77(6), 1284-1293 (2000). 65 M. Grenier-Loustalot, S. Larroque and P. Grenier, Phenolic Resins: 5. Solid-State Physicochemical Study of Resoles with Variable F/P Ratios, Polymer 37(4), 639-650 (1996). 66 T. Holopainen, L. Alvila, J. Rainio, and T. T. Pakkanen, IR Analysis of Phenol- Formaldehyde Resole Resins, Journal of Applied Polymer Science 69(11), 2175-2185 (1998). 67 G. Carotenuto and L. Nicolais, Kinetic Study of Phenolic Resin Cured by IR T. Holopainen, L. Alvila, J. Rainio, and T. T. Pakkanen, Phenol-Formaldehyde Resol Spectroscopy, Journal of Applied Polymer Science 74(11), 2703-2715 (1999). 68 Resins Studied by C-13-NMR Spectroscopy, Gel Permeation Chromatography, and Differential Scanning Calorimetry, Journal of Applied Polymer Science 66(6), 11831193 (1997). 45 chromatography or size exclusion chromatography (GPC),68,69,71,72,73 high performance liquid chromatography (HPLC),74 mass spectrometric analysis,75 differential scanning calorimetry (DSC),68,76,77 and dynamic mechanical analysis (DMA)78,79 have been utilized 69 M. G. Kim, L.W. Amos, and E. E. Barnes, Study of the Reaction Rates and Structures of a Phenol Formaldehyde Resol Resin by C-13 NMR and Gel-Permeation Chromatography, Industrial & Engineering Chemistry Research 29(10), 2032-2037 (1990). 70 P. Luukko, L. Alvila, T. Holopainen, J. Rainio, and T. T. Pakkanen, Optimizing the 13 Conditions of Quantitative 71 C NMR Spectroscopy Analysis for Phenol-Formaldehyde Resole Resins, Journal of Applied Polymer Science 69, 1805-1812 (1998). T. Sellers and M. L. Prewitt Applications of Gel-Filtration Chromatography for Resole Phenolic Resins using Aqueous Sodium-Hydroxide and Solvent, Journal of Chromatography 513, 271-278 (1990). 72 G. Gobec, M. Dunky, T. Zich, and K. Lederer, Gel Permeation Chromatography and of Resolic Phenol-Formaldehyde Condensates, Angewandte Calibration 73 Makromolekulare Chemie 251, 171-179 (1997). M.G. Kim, W. L. Nieh, T. Sellers, W.W. Wilson, and J. W. Mays, Polymer-Solution Properties of a Phenol Formaldehyde Resol Resin by Gel-Permeation Chromatography, Intrinsic-Viscosity, Static Light Scattering, and Vapor-Pressure Osmometric Methods, Industrial & Engineering Chemistry Research 31(3), 973-979 (1992). 74 G. Astarloa-Aierbe, J. M. Echeverria, J. L. Egiburu, M, Ormaetxea, and I. Mondradon, Kinetics of Phenolic Resol Resin Formation by HPLC, Polymer 39(14), 3147-3153 (1998). 75 L. Prokai and W. J. Simonsick, Direct Mass-Spectrometric Analysis of Phenol Formaldehyde Oligocondensates- A Comparative Desorption Ionization Study, Macromolecules 25(24), 6532-6539 (1992). 76 J. M. Kenny, G. Pisaniello, F. Farina, and S. Puzziello, Calorimetric Analysis of the Polymerization Reaction of a Phenolic Resin, Thermochimica Acta 269, 201-211(1995). 46 to characterize resole syntheses and crosslinking reactions. Packed-column supercritical fluid chromatography with a negative-ion atmospheric-pressure chemical ionization mass spectrometric detector has also been used to separate and characterize resole resins.80 This section provides some examples of how these techniques are used in practical applications. Using FTIR spectroscopy, resole resin formation and cure reactions can be examined (Table 2. 7). FTIR can be used to monitor the appearance and disappearance of hydroxymethyl groups and/or methylene ether linkages, ortho reactive groups, and para reactive groups for resole resin syntheses. Other useful information deduced from FTIR are the type of hydrogen bonding, i.e. intra- vs. inter-molecular, the amount of free phenol present in the product, and the formaldehyde/phenol molar ratio. FTIR bands and patterns for various mono-, di-, and tri-substituted phenols have been identified using a series of model compounds. The kinetics of resole cure reactions via FTIR indicates that a diffusion mechanism dominates below 140C. The cure above 140C exhibits a homogeneous first order reaction rate. The activation energy of the cure reaction was ~ 49.6 KJ/mol.67 77 W. W. Focke, M. S. Smit, A. T. Tolmay, L. S. Vandervalt, and W. L. Vanwyk, Differential Scanning Calorimetry Analysis of Thermoset Cure Kinetics- Phenolic Resol Resin, Polymer Engineering and Science 31(23), 1665-1669 (1991). 78 M. G. Kim, W. L. S. Nieh, and R. M. Meacham, Study of the Curing of Phenol- Formaldehyde Resol Resins by Dynamic Mechanical Analysis, Industrial & Engineering Chemistry Research 30(4), 798-803 (1991). 79 R. A. Follensbee, J. A. Koutsky, A.W. Christiansen, G.E. Myers, and R. L. Geimer, Development of Dynamic Mechanical Methods to Characterize the Cure State of Phenolic Resole Resins, Journal of Applied Polymer Science 47(8), 1481-1496 (1993). 80 M. J. Carrott and G. Davidson, Separation of Characterization of PhenolFluid Formaldehyde (Resole) Prepolymers using Packed-Column Supercritical (1999). 47 Chromatography with APCI Mass Spectrometric Detection, Analyst 124(7), 993-997 13 C NMR has proven to be an extremely powerful technique for both monitoring 13 the phenolic resin synthesis and determining the product compositions and structures. Insoluble resole networks can be examined using solid state C NMR which characterizes substitutions on ortho and para positions, the formation and disappearance of hydroxymethyl groups, and the formation of para-para methylene linkages. Analyses using 13C NMR have shown good agreement with those obtained from FTIR.65 Table 2. 7. -1 FTIR absorption band assignment of resole resins66 Wave No. (cm ) 3350 3060 3020 2930 2860 1610 1500 1470 1450 1370 1240 1160 1100 1010 880 820 790 760 Assignment* v(CH) v (CH) v (CH) v ip(CH2) v op (CH2) v (C=C) v (C=C) d(CH2) v (C=C) d ip (OH) v ip (C-O) d ip (CH) d ip (CH) v (C-O) d op (CH) d op (CH) d op (CH) d op (CH) Nature Phenolic and methylol Aromatic Aromatic Aliphatic Aliphatic Benzene ring Benzene ring Aliphatic Benzene ring Phenolic Phenolic Aromatic Aromatic Methylol Isolated H Adjacent 2H, para substituted Adjacent 3H Adjacent 4H, ortho substituted 690 d op (CH) Adjacent 5H, phenol *v=stretching, d=deformation, ip=in plane, op=out of plane Various ionization methods were used to bombard phenol/formaldehyde oligomers in mass spectroscopic analysis. The molecular weights of resole resins were 48 calculated using field desorption mass spectroscopy of acetyl-derivatized samples.75 Peracetylation was used to enable quantitative characterization of all molecular fractions by increasing the molecular weight in increments of 42. Dynamic DSC scans of resole resins show two distinguishable reaction peaks, which correspond to formaldehyde addition and formation of ether and methylene bridges characterized by different activation energies. Kinetic parameters calculated using a regression analysis show good agreement with experimental values.76 DMA was used to determine the cure times and the onset of vitrification in resole cure reactions.78 The time at which two tangents to the storage modulus curve intersect (near the final storage modulus plateau) was suggested to correspond to the cure times. In addition, the time to reach the peak of the tan delta curve was suggested to correspond to the vitrification point. As expected, higher cure temperatures reduced the cure times. DMA was also used to measure the degree of cure achieved by resole resins subsequent to their exposure to combinations of reaction time, temperature and humidity.78 The ultimate moduli increased with longer reaction times and lower initial moisture contents. The area under the tan delta curve during isothermal experiments was suggested to be inversely proportional to the degree of cure developed in samples prior to the measurement. A NaOH catalyzed resole resin, acetylated or treated with an ion exchange resin (neutralized and free of sodium), was analyzed using GPC in THF solvent. 81 The The molecular weight of the ion exchange treated resin, calculated by GPC using polystyrene standards, was significantly lower than that estimated for the acetylated resin. agreed with the results from GPC. association in acetylated samples. molecular weight for the ion exchange treated resin calculated by 1H NMR and VPO The higher molecular weight observed for the acetylated resin was attributed to higher hydrodynamic volume and/or intermolecular 81 Y. Yazaki, P. J. Collins, M. J. Reilly, S. D. Terrill and T. Nikpour, Fast-Curing Phenol-Formaldehyde (PF) Resins, Holzforschung 48(1), 41-48 (1994). 49 2.4.4. Resole network properties Voids in resole networks detract from the mechanical properties. Irrespective of the curing conditions, all resole networks contain a significant amount of voids due to volatiles released during the cure reactions. The catalyst concentration in resole crosslinking reactions can lead to different pore microstructures, which influence the mechanical properties.82 Resole networks cured using p-toluenesulfonic acid between 40 and 80C showed that increased catalyst concentrations led to reduced average void diameters. Higher cure temperatures also resulted in reduced void diameters although the effect was not as substantial. The same study showed that while the catalyst concentration did not affect the network Tgs, higher cure and post-cure temperatures increased Tgs and reduced fracture strains. In another exemplary study, optical microscopy revealed that the void content of resole networks ranged from 0.13 to 0.21.83 Resole networks prepared from different F/P molar ratios showed comparable void distributions. A bimodal distribution was observed for all networks, which was attributed to thermodynamic phase separations of reaction volatile (free phenol, formaldehyde and water) and reaction kinetics. The glass transition temperatures were determined from the peaks of tan delta curves measured using dynamic mechanical analysis for a series of resole networks (prepared with F/P molar ratios ranging from 1 to 2.5).83 Networks obtained from resoles with high F/P molar ratios (>1.2) had fairly consistent Tgs (between 240 and 260C). The lowest Tg (190C) was observed for networks prepared with low hydroxymethyl substituted resoles (F/P=1) and this was attributed to the low network crosslink densities at this ratio. The highest Tg occurred at F/P=1.2 (~280C), but it superimposed the degradation temperatures. The width of the tan delta curves was used to assess the 82 J. Wolfrum and G. W. Ehrenstein, Interdependence Between the Curing, Structure, and the Mechanical Properties of Phenolic Resins, Journal of Applied Polymer Science 74, 3173-3185 (1999). 83 L. B. Manfredi, O. de la Osa, N. Galego Fernandez, and A. Vazquez, Structure- Properties Relationship for Resoles with Different Formaldehyde/Phenol Molar Ratio, Polymer 40, 3867-3875 (1999). 50 distributions of chain length as well as the crosslink densities. Networks cured with resole (F/P=1.3 and 1.4) exhibited the highest Tgs and therefore the highest crosslink densities. 2.4.5. Modified phenol-formaldehyde resins Phenol, formaldehyde and urea have been copolymerized to achieve resins and subsequent networks with improved flame retardance and lower cost relative to phenol/formaldehyde analogues. The condensation of a phenolic methylol group with urea (Figure 2. 31) is believed to be the primary reaction under the weakly acidic conditions normally used. OH CH2OH O + H2N C NH2 -H2O OH O CH2 NH C NH2 Figure 2. 31. Reaction of hydroxymethylphenol and urea Resins were prepared by co-condensing low molecular weight hydroxymethyl substituted resoles with urea. The rate of the urea-methylol reaction was greatly enhanced by increased acidity in the reaction medium.84 Para-methylol groups reacted faster than ortho methylol groups. The extent of urea incorporation depended on the F/P ratio and the resole/urea composition. Increased urea incorporation ensued at higher urea concentrations and/or in the presence of highly hydroxymethylated resole resins (prepared from larger F/P ratios). Since reactions were conducted under acidic conditions, the condensation of methylol and urea competes with methylol self condensation. Increased urea concentrations suppress the self-condensation reactions. 84 B. Tomita and C. Hse, Synthesis and Structual Analysis of Cocondensed Resins from Urea and Methylolphenols, Mokuzai Gakkaishi 39(11), 1276-1284 (1993). 51 The curing process of tri-hydroxymethylphenol reacted with urea was monitored using torsional braid analysis.85 Curing proceeded in two stages where the first stage occurred at lower temperatures and was attributed to the reaction of para-hydroxymethyl and urea groups. The second stage was due to the higher temperature reaction of orthohydroxymethyl and urea groups. Condensation reactions of hydroxymethyl groups on phenolic resoles and amines on melamine take place between pHs 5 and 6 (Figure 2. 32). Only self-condensations of hydroxymethyl substituents occur under strongly acidic or basic conditions. OH OH CH2OH H2N N N N NH2 NH2 -H2O CH2 NH N N N NH2 NH2 + Figure 2. 32. Reaction of hydroxymethylphenol and melamine 2.5. Epoxy/phenol networks Void free phenolic networks can be prepared by crosslinking novolacs with epoxies instead of HMTA. A variety of difunctional and multifunctional epoxy reagents can be used to generate networks with excellent dielectric properties.4 One example of epoxy reagents used in this manner is the epoxidized novolac (Figure 2. 33) derived from the reaction of novolac oligomers with an excess of epichlorohydrin. 85 B. Tomita, M. Ohyama, A. Itoh, K. Doi, and C. Hse Analysis of Curing Process and Thermal Properties of Phenol-Urea-Formaldehyde Cocondensation Resins, Mokuzai Gakkaishi 40(2), 170-175 (1994). 52 O OH O CH2 O CH2 CH CH2 + Cl CH2 CH CH2 CH2 Figure 2. 33. Reaction of phenol and epichlorohydrin to form epoxidized novolacs 2.5.1. Mechanism of the epoxy/phenolic reaction The reactions between phenolic hydroxyl groups and epoxides have been catalyzed by a variety of acid and base catalysts, group 5a compounds, and quaternary ammonium complexes,86 although they are typically catalyzed by tertiary amines or phosphines, with triphenylphosphine being the most commonly used reagent. The reaction mechanism (Figure 2. 34) involves triphenylphosphine attacking an epoxide which results in ring opening and produces a zwitterion. Rapid proton transfer occurs from the phenolic hydroxyl group to the zwitterions to form phenoxide anions and secondary alcohols. The phenoxide anion subsequently reacts with either an electrophilic carbon next to the phosphorus regenerating the triphenylphosphine (Figure 2. 34 A)87 or it can ring open an epoxy followed by proton transfer from another phenol to regenerate the phenoxide anion (Figure 2. 34 B). The phenolate anion is the reactive species for the crosslinking reaction. 86 R. W. Biernath and D. S. Soane in J. S. Salamone and J. S. Riffle, Ed., Cure Kinetics of Epoxy Cresol Novolac Encapsulant for Microelectronic Packaging, Contemporary Topic in Polymer Science, Advances in New Material, Plenum Press, New York, Vol. 7, 1992, pp. 103-160. 87 W. A. Romanchick, J. E. Sohn, and J. F. Geibel in R. S. Bauer, ed., Synthesis, Morphology, and Thermal Stability of Elastomer-Modified Epoxy Resin, ACS Symposium Series 221 - Epoxy Resin Chemistry II, American Chemical Society: Washington D.C. 1982, pp. 85-118. 53 PPh3 + O OR OPh3P+ R OH OH + Ph3P + R OH O - + Ph3P OH + R O + Ph3P - + R A O OO R + PPh3 O- + O R B R Figure 2. 34. Mechanism for the triphenylphosphine catalyzed phenol/epoxy reaction Melt reaction mechanisms of tertiary aliphatic amine catalyzed phenolic/epoxy reactions was proposed to begin with a trialkylamine abstracting a phenolic hydroxyl proton to form an ion pair (Figure 2. 35).88 The ion pair was suggested to complex with an epoxy ring which then dissociates to form a -hydroxyether and regenerate the trialkylamine. 88 D. Gagnebien, P. J. Madec, and E. Marechal, Synthesis of Poly(Sulphone-b Siloxane)s 1- Model Study of the Epoxy-Phenol Reaction in the Melt, European Polymer Journal 21(3), 273-283 (1985). 54 OH NR3 O - +NHR3 O- +NHR3 O R OR3N + R O H OR3N + OH R O O R H Figure 2. 35. Proposed mechanism for tertiary amine catalyzed phenol/epoxy reaction Side reactions involving branching through a secondary hydroxyl group can also occur. The extent of these side reactions should decrease as the ratio of epoxy to phenol decreases since phenolate anions are significantly more nucleophilic than aliphatic hydroxyl groups. 2.5.2. Epoxy phenolic reaction kinetics A review on epoxy/novolac reaction mechanisms and kinetics is provided by Soane.86 Depending on the structures of novolac and epoxies, reactions may proceed through an nth order mechanism or an autocatalytic mechanism.89 First order reaction kinetics where epoxy ring opening dictates the reaction rate has been found to fit well for a number of novolac/epoxy reactions. Reactions are autocatalytic if more catalyst is produced and the rate accelerates as the reaction proceeds. Autocatalytic kinetics appears to fit well if an active catalytic complex, C, must form initially to generate reaction products. Soane concluded that phenolic novolac and epoxidized cresol novolac cure reactions using triphenylphosphine as catalyst showed a short initiation regime wherein 89 W. G. Kim, J. Y. Lee and K. Y. Park, Curing Reaction of o-Cresol Novolac Epoxy Resin According to Hardener Change, Journal of Polymer Science, Part A: Polymer Chemistry 31, 633-639 (1993). 55 the concentration of phenolate ion increased followed by a (steady-state) propagation regime where the number of reactive phenolate species was constant.86 The epoxy ring opening reaction was reportedly first-order in the steady-state regime. d = k1 ( max ) dt (2. 6) where is the fraction of epoxy reacted, max is the maximum fraction of epoxy reacted at the given stoichiometry and temperature, and k1 is the first order kinetic rate constant. The isothermal reaction rate for autocatalytic cure kinetics is calculated using d = k ' m (1 )n dt where k is the kinetic rate constant, and m and n are the reaction orders. To describe the reaction rate where the initial rate is not zero, the following modification was made90 d = (k1 + k2 m )(1 ) n dt where k1 and k2 are kinetic rate constants. The mechanism for the tertiary amine catalyzed reaction between phenol and epoxy was proposed by Sorokin and Shode91 E+P+B E+P+I k1 k2 (2. 7) (2. 8) I Pr + I (2. 9) 90 M. R. Kamal, Thermoset Characterization for Moldability Analysis, Polymeric M. F. Sorokin and L. G. Shode, Reactions of -Oxides with Proton Donor Engineering Science 14(3), 231-239 (1974). 91 Compounds in the Presence of Tertiary Amines. 1. Reaction of Phenyl Glycidyl Ether with Phenol in the Presence of Teritary Amines Zhurnal Organicheskoi Khimii 2(8), 1463-1468 (1966). 56 where E represents epoxy groups, P the phenol, B the basic catalyst, I an intermediate complex, PR the product of the phenol and epoxide reaction, and k1 and k2 the kinetic constants. This mechanism suggests that the isothermal reaction rate is directly Thus, the catalyst concentration may be proportional to catalyst concentration. introduced into the rate expression. A diffusion effect was incorporated into equation 7 to improve the conversion vs. time prediction fit above 90 % conversion.92 The diffusion factor f() is based on free volume principles.93 d = (k1 '+ k2 ' m )(1 ) n [B ] f ( ) dt f ( ) = 1 1 + exp[C ( c )] (2. 10) (2. 11) where C is the material constant, c is the critical conversion, and k1 and k2 are normalized kinetic rate constants. For << c, in which the rate of diffusion is not reaction rate limiting, f() is essentially equal to unity. As approaches c, the diffusion factor decreases. The measured reaction conversion for triphenylphosphine catalyzed biphenyl epoxy/phenol novolac cure plotted vs. reaction time fits extremely well the conversion values calculated via the above expression for all catalyst concentrations over the conversion range. Ortho-cresol novolac epoxy oligomers were cured with a phenolic novolac or a phenolic novolac acetate resin catalyzed by 2-methylimidzole.94 While the phenolic 92 S. Han, H. G. Yoon, K. S. Suh, W. G. Kim and T. J. Moon, Curing Kinetics of Biphenyl Epoxy-Phenol Novolac Resin System Using Triphenylphosphine as Catalyst, Journal Polymeric Science, Part A: Polymeric Chemistry 37, 713-720 (1999). 93 C. S. Chern and G.W. Poehlein, A Kinetic Model for Curing Reaction of Epoxides X. W. Luo, Z. H. Ping, J. P. Ding, Y. D. Ding, and S. J. Li, Mechanism Studies on with Amines, Polymer Engineering & Science 27(11), 788-795 (1987). 94 Water Sorption and Permeation in Epoxy Resin by Impedance Spectroscopy. II. Cure 57 novolac acetate system clearly followed nth order kinetics by showing a linear plot of log(d/dt) vs. log(1-), the phenolic novolac cured system was better fitted with autocatalytic reaction kinetics. 2.5.3. Epoxy/phenol network properties Void-free phenolic/epoxy networks prepared from an excess of phenolic novolac resins to various diepoxides have been investigated by Riffle et al. (Figure 2. 36).95,96 The novolacs and diepoxide were cured at approximately 200C in the presence of triphenylphosphine and other phosphine derivatives. Network densities were controlled by stoichiometric offsets between phenol and epoxide groups. These networks contained high phenolic concentrations (up to ~80 wt %) to retain the high flame retardance of the phenolic materials while the mechanical properties were tailored by controlling the crosslink densities and molecular structures. OH CH2 OH CH2 OH OH HO OH O CH2 OH HO O R O OH O CH2 CH2 Triphenylphosphine CH2 5.3 Phenlic Novolac O O + R O O Epoxy crosslinked networks Figure 2. 36. Network formation of phenolic novolac and epoxy Kinetics of o-Cresol Novolac Resin with Esterified Phenol Novolac Resin, Pure Applied Chemistry A34(11), 2279-2291 (1997). 95 C. S. Tyberg, M. Sankarapandian, K. Bears, P. Shih, A. C. Loos, D. Dillard, J. E. McGrath, and J. S. Riffle, Tough, Void-Free, Flame Retardant Phenolic Matrix Materials, Construction and Building Materials 13, 343-353 (1999). 96 C. S. Tyberg, K. Bergeron, M. Sankarapandian, P. Shih, A. C. Loos, D. A. Dillard, J. E. McGrath, and J. S. Riffle, Structure Property Relationships of Void Free Phenolic Epoxy Matrix Materials, Polymer 41(13), 5053-5062 (2000). 58 Several diepoxides were utilized to determine structure property relationships of the networks and flame properties (Figure 2. 37). (1) O O Br Br O OH O Br Br O O (2) O O Br Br Si O Si O OH O Br O Br O (3) O O O O Figure 2. 37. Diepoxide structures: (1) bisphenol-A based diepoxide, (2) brominated bisphenol-A based diepoxide, and (3) siloxane diepoxide Network structure property relationships and flame properties were determined for a novolac cured with various diepoxides at defined compositions. The fracture toughness of the networks, determined by the plane-strain stress intensity factors (KIC), increased with increased stoichiometric offset to a maximum of approximately 1.0 MPa*m1/2 for the 5 phenol/1 epoxy equivalence ratio (Table 2. 8). All novolac/epoxy networks were significantly tougher than a thermally cured resole network (0.16 MPa*m1/2). Further stoichiometric offset (to 7/1) reduced the fracture toughness. This was undoubtedly related to the increase in dangling ends and unconnected phenolic chains at these very high phenol to epoxy ratios. Likewise, as expected, glass transition temperatures of fully cured networks decreased as the distances between crosslinks (Mx) increased. In general, network toughness increased with the average molecular weight between crosslinks up to some point beyond which the amount of unconnected phenolic chains began to detract from properties. increased. Likewise, as expected, glass transition temperatures of fully cured networks decreased as the distances between crosslinks (Mx) 59 Table 2. 8. Tg and KIC of phenolic novolac/epoxy networks Epoxy Phenol/Epoxy (wt/wt) Phenol/Epoxy (mol:mol) Tg (C) 127 -114 127 151 130 148 96 87 KIC (MPa*m ) 0.62 0.16 070 0.85 0.64 0.74 0.84 0.62 0.77 1/2 Mx (g/mol) Bisphenol-A Epoxy cured with 4,4- -DDS Phenolic control (thermally cured -resole) 80/20 Bisphenol-A Brominated Bisphenol-A Disiloxane epoxy 65/35 50/50 65/35 50/50 80/20 65/35 7:1 3:1 2:1 5.8:1 3.1:1 7.2:1 3:1 4539 1413 643 3511 1554 4051 1030 The flame retardance was measured using a cone calorimeter with a heat flux of 50 kW/m2 and 20.95 mol % O2 content (atmospheric oxygen). All phenolic novolac/epoxy networks with relatively high novolac compositions showed much lower peak heat release rates (PHRR) than a typical amine cured epoxy network (bisphenol-A epoxy stoichiometrically cured with p,p-diaminodiphenylsulfone) (Table 2. 9). Brominated epoxy reagents were also investigated since halogenated materials are well known for their roles in promoting flame retardance. Networks cured with the brominated diepoxide showed the lowest peak heat release rates, but the char yields of these networks were lower and the smoke toxicity (CO yield /CO2 yield) was increased. Incorporating siloxane moieties into networks reduced the peak heat release rates and smoke toxicities compared to the novolac networks cured with bisphenol-A diepoxides. 60 Table 2. 9. Flame retardance of networks prepared form a phenolic novolac crosslinked with various epoxies97 Epoxy Phenolic/Epoxy (wt/wt) PHRR (KW/m ) 1230 116 260 360 380 165 158 226 325 2 Char yield (%) 5 63 33 29 23 8 9 35 24 Smoke Toxicity* (x10-3) 44 27 34 36 189 175 15 27 Bisphenol-A Epoxy cured with 4,4-DDS Phenolic control (thermally cured resole) Bisphenol-A Bisphenol-A Bisphenol-A Brominated bisphenol-A Brominated bisphenol-A Disiloxane epoxy 80/20 65/35 50/50 65/35 50/50 80/20 Disiloxane epoxy 65/35 *smoke toxicity: CO/CO2 release A biphenol diglycidyl ether based epoxy resin was crosslinked with amine curing agents (4,4-diaminodiphenylmethane and aniline novolac) and phenol curing agents (phenol novolac and catechol novolac), and the thermo-mechanical properties were investigated.98 Unlike bisphenol-A epoxy based networks, distinct Tgs were not observed with the amine cured biphenol epoxy networks. This was hypothesized to be caused by the mesogenic nature of the biphenol groups which allowed the chains to be more closely packed. Thus, it was reasoned that the mobility did not decrease significantly with the transition into the rubbery region. The presence of a distinct Tg when phenols were used to cure biphenol based diepoxide depended on the phenolic structure. Whereas a distinct Tg was evident in the phenolic novolac cured systems, no definite Tgs were observed when catechol novolac was used. The higher moduli shown by the catechol novolac 97 98 Tests were conducted at Naval Surface Warfare Center, Carderock, Maryland. M. Ochi, N. Tsuyuno, K. Sakaga, Y. Nakanishi, and Y. Murata, Effect of Network Structure on Thermal and Mechanical Properties of Biphenyl-Type Epoxy Resins Cured with Phenols, Journal of Applied Polymer Science 56, 1161-1167 (1995). 61 cured networks were attributed to the orientation of mesogenic biphenyl groups which suppressed micro-Brownian chain motions. Network properties and microscopic structures of various epoxy resins crosslinked by phenolic novolacs were investigated by Suzuki et al.99 Positron annihilation spectroscopy (PAS) was utilized to characterize intermolecular-spacing of networks and the results were compared to bulk polymer properties. The lifetime (3) and intensity (I3) of the active species (positronium ions) correspond to volume and number of holes which constitute the free volume in the network. Networks cured with flexible epoxies had more holes throughout the temperature range, and were more affected by the temperature changes. Glass transition temperatures and thermal expansions () were calculated from plots of 3 (free volume) versus temperature. The Tgs and thermal expansion efficients obtained from PA were lower than these obtained from thermomechanical analysis. These differences were attributed to the microBrownian motions determined by PAS versus large-scale polymer properties determined by thermomechanical analysis. The differences in Tgs and thermal expansion efficients were more pronounced for the more highly crosslinked materials. The rate of moisture absorption is proportional to I3 of the network, therefore networks with larger free volume absorbed water at a faster rate. 2.6. Benzoxazines Benzoxazines are heterocyclic compounds obtained from Mannich reactions of phenols, primary amines, and formaldehyde (Figure 2. 38).100,101 As described previously, 99 T. Suzuki, Y. Oki, M. Numajiri, T. Miura, K. Kondo, Y. Shiomi, and Y. Ito, Novolac Epoxy Resins and Positron Annihilation, Journal of Applied Polymer Science 49, 19211929 (1993). 100 W. J. Burke, E. L. M. Glennie, and C. Weatherbee, Condensation of Halophenols X. Ning and H. Ishida, Phenolic Materials via Ring-Opening Polymerization: with Formaldehyde and Primary Amines, Journal Organic Chemistry 29, 909 (1964). 101 Synthesis and Characterization of Bisphenol-A Based Benzoxazines and Their 62 they are key reaction intermediates in the hexamethylenetetramine (HMTA) novolac cure reaction.41,44 Crosslinking benzoxazines at high temperatures give rise to void free networks with high Tgs, excellent heat resistance, good flame retardance, and low smoke toxicity.102 As in HMTA cured novolac networks, further structural rearrangement may occur at higher temperatures. NH2 HO OH +2 + 4 CH2O N O O N Figure 2. 38. Synthesis of bisphenol-A based benzoxazines A difunctional bisphenol-A epoxy based benzoxazine has been synthesized and characterized by GPC and 1H NMR.101 A small of amount of dimers and oligomers also formed. Thermal crosslinking of bisphenol-A benzoxazine containing dimers and oligomers resulted in networks with relatively high Tgs. Dynamic mechanical analysis of the network showed a peak of tan delta at approximately 185C. The kinetics of bisphenol-A benzoxazine crosslinking reactions was studied using differential scanning calorimetry.102 The activation energy, estimated from plots of conversion as a function of time for different isothermal cure temperatures, was between 102 and 116 KJ/mol. Phenolic compounds with free ortho positions were suggested to initiate the benzoxazine reaction (Figure 2. 39).103 Fast reactions between benzoxazines Polymers, Journal of Polymer Science. Part A: Polymer Chemistry 32, 1121-1129 (1994). 102 H. Ishida and Y. Rodriguez, Cure Kinetics of a New Benzoxazine-Based Phenolic G. Riess, J. M. Schwob, G. Guth, M. Roche, and B. Lande in B. M. Culbertson and J. Resin by Differential Scanning Calorimetry, Polymer 36(16), 3151-3158 (1995). 103 E. McGrath, eds, Ring Opening Polymerization of Benzoxazines-A New Route to 63 and free ortho phenolic positions, which formed hydroxybenzylamines, were facilitated by hydrogen bonding between the phenol hydroxyl and benzoxazine oxygen (as shown in Figure 2. 18). Subsequent thermal decompositions of these less stable hydroxybenzylamines lead to more rapid thermal crosslinking (as described for the HMTA/novolac cure). R2 OH + R1 O N OH N R2 n R1 R1 Figure 2. 39. Reaction of benzoxazines with free ortho positions on phenolic compounds The reaction of bisphenol-A benzoxazine under strong and weak acidic conditions was also investigated.104 The proposed mechanism for the benzoxazine ring opening reaction in the presence of a weak acid involves an initial tautomerization between the benzoxazine ring and chain forms. Electrophilic aromatic substitution reaction between a phenolic ring position and the chain tautomer, an iminium ion was suggested to follow. Strongly acidic conditions, high temperatures and the presence of water lead to various side reactions, including benzoxazine hydrolysis in a reverse Mannich reaction. Side reactions could also terminate reaction or lead to crosslinking. The oxazine ring in benzoxazine assumes a distorted semi-chair conformation.105 The ring strain and the strong basicity of the nitrogen and oxygen allow benzoxazines to Phenolic Resins, in Advances in Polymer Synthsis,. Plenum Press, New York, 1985, pp 27-50. 104 J. Dunkers and H. Ishida, Reaction of Benzoxazine-Based Phenolic Resins with Strong and Weak Carboxylic Acids and Phenol as Catalysts, Journal of Polymer Science. Part A: Polymer Chemistry 37(13), 1913-1921 (1999). 105 H. Ishida and D. J. Allen, Physical and Mechanical Characterization of Near Zero Shrinkage Polybenzoxazines, Journal Polymer Science. Polymer Physical Ed. 34(6), 1019-1031 (1996). 64 undergo cationic ring opening reactions. A number of catalysts and/or initiators such as PCl5, PCl3, POCl3, TiCl4, AlCl3 and MeOTf are effective in promoting benzoxazine polymerization at moderate temperatures (20-50C).106 Dynamic DSC studies revealed multiple exotherms in polymerization of benzoxazine, indicating a complex reaction mechanism. 2.7. Phenolic triazine (PT) resins Novolac hydroxyl groups reacted with cyanogen bromide under basic conditions to produce cyanate ester resins (Figure 2. 40).107,108 Cyanate esters can thermally crosslink to form void free networks, wherein at least some triazine rings form. The resultant networks possess high Tgs, high char yield at 900C and high decomposition temperatures.107 CH2 OH CH2 BrCN OCN CH2 Heat O N O CH2 N N O CH2 n n Figure 2. 40. 106 Synthesis of phenolic triazine resins Y. X. Wang and H. Ishida, Cationic Ring-Opening Polymerization of Benzoxazines, U.S. Pat. 4,831,086 (May 16, 1989), S. Das and D. C. Prevorsek, Cyanate group S. Das, Phenolic-Triazine (PT) Resin-A New Family of High Performance Polymer 40(16), 4563-4570 (1999). 107 containing phenolic resins, phenolic triazines derived therefrom (to Allied-Signal, Inc.). 108 Thermosets, Abstracts of papers of the American Chemical Society-PMSE 203, 259 (1992). 65 Novolac resins containing cardanol moieties have also been converted to cyanate ester resins.109 The thermal stability and char yield, however, was reduced when cardanol was incorporated into the networks. 2.8. Thermal and thermo-oxidative degradation Phenolic networks are well known for their excellent thermal and thermooxidative stabilities. The mechanisms for high temperature phenolic degradation include dehydration, thermal crosslinking, and oxidation which eventually lead to char. Thermal degradation below 300C in inert atmospheres produces only small amounts of gaseous products. These are mostly unreacted monomers or water, which are by-products eliminated from condensation reactions between hydroxymethyl groups and reactive ortho or para positions on phenolic rings. A small amount of oxidation may occur in air as some carbonyl peaks have been observed using 13C NMR.110 Degradation in inert atmospheres between 300 and 600C results in porous materials. Little shrinkage has been observed in this temperature range. Water, carbon monoxide, carbon dioxide, formaldehyde, methane, phenol, cresols and xylenols are released. According to various thermogravimetric analyses, the weight loss rate reaches a maximum during this temperature range. The elimination of water at this stage may also be caused by the phenolic hydroxyl condensations which give rise to biphenyl ether linkages (Figure 2. 41). 109 C. P. R. Nair, R. L. Bindu, and V. C. Joseph, Cyanate Esters Based on Cardanol Modified-Phenol-Formaldehyde Resins-Synthesis and Thermal Characterizations, Journal of Polymer Science. Part. A. Polymer Chemistry 33(4), 621-627 (1995). 110 C. A. Fyfe, M. S. McKinnon, A. Rudin, and W. J. Tchir, Investigation of the 13 Mechanisms of the Thermal Decomposition of Cured Phenolic Resins by High Resolution C CP/MAS Solid-State NMR Spectroscopy, Macromolecules 16, 12161219 (1983). 66 OH + HO -H2O O Figure 2. 41. Dehydration of hydroxyl groups Morterra and Low111,112 proposed that thermal crosslinking may occur between 300C and 500C where phenolic hydroxyl groups react with methylene linkages to eliminate water (Figure 2. 42). Evidence for this mechanism is provided by IR spectra which show decreased OH stretches and bending absorptions as well as increased complexity of the aliphatic CH stretch patterns in this temperature range. HO OH -H2O HO + OH OH Figure 2. 42. Thermal crosslinking of phenolic hydroxyl and methylene linkages At elevated temperatures, methylene carbons cleave from aromatic rings to form radicals (Figure 2. 43). Further fragmentation decomposes xylenol to cresols and Alternatively, auto-oxidation occurs (Figure 2. 43B). methane (Figure 2. 43A). Aldehydes and ketones are intermediates before decarboxylation or decarbonylation takes place to generate cresols and carbon dioxide. These oxidative reactions are possible even in inert atmospheres due to the presence of hydroxyl radicals and water.7 111 C. Morterra and M. I. D. Low, IR Studied of Carbons-VII. The Pyrolysis of a PhenolC. Morterra and M. I. D. Low, Infrared Studies of Carbon. 8. The Oxidation of Formaldehyde Resin, Carbon 23(5), 525-530 (1985). 112 Phenol-Formaldehyde Chars, Langmuir 1, 320-326 (1985). 67 OH OH OH CH2 + OH OH CH2 OH CH2 CH3 OH + CH3 CH2 CH4 A OH CH2 -CO2 OH CH2 OH OH CH2OH OH -H2O OH O CH OH OH O COH B Figure 2. 43. Thermal bond rupture: a) fragmentation reaction b) oxidation degradation Oxidative degradation begins at lower temperatures in air (<300 C) and oxidation occurs most readily on benzylic methylene carbons since they are the most vulnerable sites. This leads to dihydroxybenzohydrole and dihydroxybenzophenone derivatives (Figure 2. 44). Dihydroxybenzophenone may cleave and further oxidize to carboxylic acid before decomposing to form cresols, xylenols, CO and CO2. 68 OH OH OOH OH OH OH OH OH benzohydrole OH O + OH OH OH O C OH -CO benzophenone OH O CO H OH -CO2 Figure 2. 44. Oxidation degradation on methylene carbon Hydroxyl elimination is necessary for the formation of benzaldehyde and benzoic acid derivatives, and ultimately, benzene and toluene (Figure 2. 45).4 It is proposed that a cleavage between the hydroxyl group and aromatic ring leads to benzonoid species which undergo further cleavage coupled with oxidation to give various decomposition products. OH CH2 - CH2CH3 CH2 + OH CH2 CH2 - CH3 O HOCH2 HC O HOC -CO2 + OH Figure 2. 45. Formation of benzenoid species 69 Oxidative branching and crosslinking are prevalent degradation pathways in air (Figure 2. 46). Phenoxy radicals are formed via hydrogen abstraction. These relatively stable intermediates can couple with each other, and, depending on carbon-carbon or carbon-oxygen coupling, form ether linkages or ketones. Diphenolquinones derived from carbon-carbon dimerization further oxidize.4 OH O O O O O O O O O O + OH HO OH O O O O diphenoquinone quinol ether cyclohexadienone Figure 2. 46. Decomposition via phenoxy radical pathways Upon further heating above 600C, the density increases as shrinkage occurs at a high rate. High temperature degradation monitored via gas chromatography indicated that the formation of carbon char parallels carbon monoxide evolution. Along with the same by-products released at lower temperatures, pyrolysis-GC-MS identified the formation of other low volatility compounds including naphthalene, methylnaphthalenes, biphenyl, dibenzofuran, fluorene, phenanthrene, and anthracene. These products may 70 result from condensation of hydroxyl groups of adjacent ortho-ortho linked phenolic rings113 (Figure 2. 47). CH2 OH HO -H2O O Figure 2. 47. Condensation of ortho hydroxyl groups The reaction sequence for char formation was also proposed to occur through initial formation of quinone type linkages, which lead to polycyclic products7 (Figure 2. 48). The quinone functionalities were confirmed through IR studies. NMR spectra showed that at 400C almost all methylene linkages and carbonyl groups shift or disappear. Mostly aromatic species are present. This is consistent with the proposed char forming mechanism but is also compatible with the formation of biphenyls through direct elimination of CO.110 O O O HOC OH OH O OH + OH OH O OH + C O Figure 2. 48. 113 Char formation J. Hetper and M. Sobera, Thermal Degradation of Novolac Resins by Pyrolysis-Gas Chromatography with a Movable Reaction Zone, Journal of Chromatography A 833, 277-281 (1999). 71 Conley showed that the primary degradation route for resole networks is oxidation regardless of the atmosphere.8 By contrast, Morterra et al. found that autooxidation was not the major degradation pathway for novolac resins.111 The degradation is proposed to occur in two stages depending on the temperature. Fragmentation of the polymer chain, beginning at approximately 350C, does not affect the polymer integrity. Above 500C the network collapses as polyaromatic domains form. Novolac network degradation mechanisms vary from those of resole networks due to differences in crosslinking methods. Nitrogen containing linkages must also be considered when HMTA (or other crosslinking agent) was used to cure novolac networks. For example, tribenzylamines, formed in HMTA cured novolac networks, decompose to cresols and azomethines (Figure 2. 49). OH N OH N OH + OH 3 Figure 2. 49. Decomposition of tribenzylamine Novolacs and networks which had been slowly heated in an inert atmosphere were studied. 1-2 wt. % of water and phenol evolved between 100 and 350C.114 Above that temperature, small quantities of CO, CO2, CH4, and low molecular weight aromatics evolved leaving behind 7585% by weight of polymer. The degradation of these preexposed networks were then studied at temperatures up to ~700C in both air and nitrogen. The rate of weight loss in air suggested that degradation under these conditions is a multistage process. Evidence suggested that this multistage thermo-oxidative process was due to two types of linking units present, which pyrolyzed at 375-400C and 500550C respectively. By contrast, the degradation process in nitrogen was a single stage process. 114 Both processes formed a char representing 90-95% by weight of the pre- V. Jha, A. Banthia, and A. Paul, Thermal Analysis of Phenolic Resin Based Pyropolymers, Journal of Thermal Analysis 35, 1229-1235 (1989). 72 exposed materials. As expected, lower weight loss occurred with increased levels of the HMTA crosslinking reagent introduced into the network. Oxygen indices of a series of phenolic networks, i.e. heat cured resoles, novolacs cured with formaldehyde, trioxane, or terephthaloyl chloride were investigated.115 It has been previously shown that oxygen indices (OI) correlate linearly with char yields at 800C in nitrogen except for cases involving gas phase retarders such as halogenated aromatic materials (van Krevelen). Meta- and para-cresol formaldehyde networks exhibited slightly lower OI values than unsubstituted phenolics, presumably due to the presence of flammable methyl substituents. The halogenated phenolics showed higher OI values but lower char yields since halogenated materials undergo a gas phase retardation of burning and the heavy halogen substituents convert to gaseous products. The crosslinking agent also has an affect on OI. Networks with methylene linkages derived from formaldehyde crosslinking reagents give rise to higher OI values than these with ester linkages derived from terephthaloyl chloride. The formaldehyde cured novolacs also showed higher OI than trioxane cured materials since less stable ether linkages were formed in the trioxane crosslinked systems. The effects of structures and the levels of crosslink densities on thermal stabilities of novolac resins were investigated.116 Three types of novolacs were prepared by reacting phenol with formaldehyde, meta-cresol with formaldehyde, and para-cresol with formaldehyde in formaldehyde/phenol molar ratio of 0.95. Lightly crosslinked networks were obtained when the trifunctional phenol or meta-cresol was used. As expected, paracresol novolac was linear. The thermal degradation behaviors for the lightly crosslinked networks, the para-cresol novolac as well as typical novolac resins were examined. Under inert atmosphere, TGA results revealed that the crosslinked samples had lower 115 Y. Zaks, J. Lo, D. Raucher, and E.M. Pearce, Some Structural Property Relationships in Polymer Flammability: Studies of Phenolic-Derived Polymers, Journal Applied Polymer Science 27, 913-930 (1982). 116 L. Costa, L. Rossi di Montelera, G. Camino, E. D. Weil, and E. M. Pearce, Structure- Charring Relationship in Phenol-Formaldehyde Type Resins, Polymer Degradation and Stability 56, 27-35 (1997). 73 decomposition temperatures but the char yields were significantly higher. The greater stabilities exhibited at higher temperatures (500C+) for the crosslinked materials were attributed to lesser fragmentation, and therefore, lower volatiles. In air, the crosslinked materials were less stable throughout the whole temperature range with complete degradation occurring 50C below the low molecular weight resins. It was suggested that the crosslinked materials may be less susceptible to oxidation, and therefore, unable to form more thermally stable intermediates. Comparatively, the phenol-formaldehyde resin showed the highest thermal and thermal oxidative stability, followed by the meta-cresolformaldehyde resin. The para-cresol-formaldehyde showed the least stability. Since the reactivity of starting phenolic monomer was different, novolac oligomers prepared via the same approach resulted in different structures. The results obtained in this study therefore could not be compared quantitatively. A low molecular weight para-cresol novolac resin (Mn~560) prepared by Belbachir et al.117 showed high crystalline ratio according to X-ray diffraction. No weight loss was observed below 400C. However, the morphology changes to a semicrystalline state after repeat thermal heating/cooling cycles. 117 A. Hamou, C. Devallencourt, F. Burel, J. M. Saiter, and M. Belbachir, Thermal Stability of a para-Cresol Novolac Resin, Journal of Thermal Analysis 52, 697-703 (1998). 74 3. Controlled Molecular Weight Cresol-Formaldehyde Oligomers 3.1. Introduction Phenolic resins are among the oldest known and highest volume thermosetting materials produced in the United States.118 Among the numerous attractive properties of phenolic resins and their networks are low cost and excellent flame retardance.119,120 Therefore, we and others are investigating this class of materials as possible matrix resins for flame retardant structural composites. The most common phenolic prepolymers are derived from reacting phenol with formaldehyde or with formaldehyde derivatives. This reaction occurs most rapidly under extremely acidic or basic conditions. The pH of the reactions and the stoichiometric ratio of the monomers give rise to two classes of phenolic prepolymers known as novolacs and resoles. Novolac oligomers are prepared in acidic media using an excess of phenol over formaldehyde. The mechanism associated with this reaction has been described in four steps (Figure 3. 1). First a methylene glycol is protonated by an acid from the reaction medium, which then releases water to form a hydroxymethylene carbonium ion (step 1). This ion acts as a hydroxyalkylating agent by reacting with a phenol via electrophilic aromatic substitution. A pair of electrons from the benzene ring attacks the electrophile forming a carbocation intermediate followed by deprotonation and regain of aromaticity (step 2). The methylol group of the hydroxymethylated phenol is unstable under acidic conditions and loses water readily to form a benzylic carbonium ion (step 3). This ion 118 A. Knopp and L. A. Pilato, Phenolic Resins: Chemistry, Application and F. Y. Hsieh and H. D. Beeson, Flammability Testing of Flame Retarded Epoxy C. J. Hilado, A M. Machado, and D. P. Brauer, Effect of Char Yield and Chemical Performance-Future Directions; Springer-Verlag, New York, 2000. 119 Composites and Phenolic Composites, Fire and Materials 21, 41-49 (1997). 120 Structure on Toxicity of Pyrolysis Gases, Proc. West. Pharmacol. Soc. 22, 201-4 (1979). 75 then reacts with another phenol to form a methylene bridge in another electrophilic aromatic substitution. This major process repeats until the formaldehyde is exhausted.121 Typically 0.75 to 0.85 moles of formaldehyde are used for each mole of phenol in the synthesis of low molecular weight novolacs,118 and branched oligomers with phenol endgroups are formed since phenol is used in excess. These prepolymers are thermally stable and can be stored effectively. Novolac crosslinking is usually achieved by introducing a source of methylene groups to form additional methylene bridges between aromatic rings. Hexamethylenetetramine (HMTA) is the most widely used curing agent (source of formaldehyde) for these reactions. importance include paraformaldehyde and trioxane. + 1) HO CH2 OH OH Other curing agents with limited 118 H+ OH + CH2 OH + H2O OH 2) OH + + slow + CH2 OH CH2 OH fast CH2 OH + H+ OH CH2 OH + +H 3) CH2 + + H2O 4) OH CH2 + OH + OH CH2 OH + H+ Figure 3. 1. Mechanism for the major process of phenolic novolac resin synthesis Resoles are obtained by reacting an excess of formaldehyde with phenol under basic conditions. This produces resins with aromatic methylol groups derived from the excess of formaldehyde. Resoles are fairly stable at ambient temperatures, but react rapidly at elevated temperatures forming methylene linkages by eliminating water and 121 A. Knopp and W. Scheib, Chemistry and Application of Phenolic Resins, Springer- Verlag, New York, 1979. 76 other by-products. Since these materials can be self-crosslinked thermally, long-term storage is more difficult. Regardless of the curing method, either by introducing a crosslinking agent or by thermal self-condensation, the network forming process is accompanied by the generation of volatile by-products such as ammonia, water and formaldehyde. Volatiles often cause voids in the networks.122,123,124 This, along with a lack of control over crosslink density, results in brittle networks. Void-free networks can be prepared by reacting phenolic novolacs with epoxies in reactions where the phenolic hydroxyl groups react with the epoxy groups.125,126,127,128 Workers in our laboratories have previously demonstrated that phenolic-epoxy networks with high phenolic compositions, and with a relatively high phenol functionality per chain (~7), exhibit significantly improved toughness while retaining most of the flame- 122 C. M. Branco, J. M. Ferreira, and M. O. W. Richardson, A Comparative Study of the Fatigue Behaviour of GRP Hand Lay-up and Pultruded Phenolic Composites, Int. J. Fatigue 18(4), 255-2643 (1995). 123 J. Wolfrum and G. W. Ehrenstein, Interdependence Between the Curing, Structure, and the Mechanical Properties of Phenolic Resins, Journal of Applied Polymer Science 74, 3173-3185 (1999). 124 L. B. Manfredi, O. de la Osa, N. Galego Fernandez, and A. Vazquez, Structure- Properties Relationship for Resoles with Different Formaldehyde/Phenol Molar Ratio, Polymer 40, 3867-3875 (1999). 125 126 Potter, W. G, Epoxide Resins, Springer-Verlag, New York, 1970. A. Hale, C. W. Macosko, and H. E. Bair, DSC and C-13-NMR Studies of the Imidazole-Accelerated Reaction Between Epoxides and Phenols, Journal of Applied Polymer Science 38(7), 1253-1269 (1989). 127 M. Ogata, N. Kinjo, and T. Kawata, Effects of Crosslinking on Physical Properties of Phenol-Formaldehyde Novolac Cured Epoxy Resins, Journal of Applied Polymer Science 48, 583-601 (1993). 128 A. K. Banthia and J. E. McGrath,.Catalysts for Bisphenol-Diglycidyl Ether Linear Step-Growth Polymerization, ACS Polym. Prepr. Div. Polym. Chem. 20(2), 629 (1979). 77 retardant properties.129,130 These improved mechanical properties have been correlated with increased molecular weight between crosslinks achieved by leaving many phenolic hydroxyl groups unreacted. Linear, cresol novolac oligomers, obtained by reacting difunctional ortho- or para-cresol with formaldehyde, are widely used as coatings, adhesives, electronic insulation materials, and for automotive applications.118 Low molecular weight oligomers have also been reacted with epichlorohydrin to form epoxy resins.125 A paracresol novolac (Mn=580 g/mol) exhibited a semi-crystalline structure with good thermal stability.131 Ortho- and para-cresol have also been copolymerized with meta-cresol for use in the electronic industry as photoresists. Since ortho- and para-cresol have slower reaction rates with formaldehyde than meta-cresol, oligomers with meta-cresol blocks were obtained having primarily ortho- or para-cresol endgroups.132,133 Ortho- or para-cresol novolac formation, which involves exclusively difunctional monomers, are linear condensation polymerizations. The molecular weights of these 129 C. S. Tyberg, K. Bergeron, M. Sankarapandian, P. Shih, A. C. Loos, D. A. Dillard, J. E. McGrath, and J. S. Riffle; Structure Property Relationshiips of Void Free Phenolic Epoxy Matrix Materials, Polymer 41(13), 5053-5062 (2000). 130 H. Ghassemi, H. K. Shoba, M. Sankarapandian, A. Shultz, C. L. Sensenich, J. S. Riffle, J. J. Lesko, and J. E. McGrath, Volatile-Free Phenolic Networks for Infrastructure, in Fiber Composites in Infrastructure, H. Saasatamanesh, M. R. Ehsani, editors, 1, 14-22 (1998). 131 A. Hamou, C. Devallencourt, F. Burel, J. M. Saiter, and M. Belbachir, Thermal Stability of a para-Cresol Novolac Resin, Journal of Thermal Analysis and Calorimetry 52(3), 697-703 (1998). 132 L. E. Bogan, Jr., in P. N. Prasad, ed., Understanding the Novolac Synthesis Reaction, Frontiers of Polymers and Advanced Materials, Plenum Press, New York, 1994, 311-318. 133 St. Miloshev, P. Novakov, Vl. Dimitrov, and I. Gitsov, Synthesis of Novolac Resins. I. Influence of the Chemical Structure of the Monomers and Reaction Conditions on Some Properties of Novolac Oligomers, Chemtronics 4, 251-253 (1989). 78 cresol novolacs, therefore, should depend on the monomer feed ratio. However, to date, cresol resin syntheses follow typical phenolic novolac synthesis procedures where reactions are terminated at a pre-determined viscosity or reaction time. The difficulty in molecular weight control arises from side reactions, which offset the stoichiometric ratio necessary to obtain target molecular weights. This paper describes the synthesis of linear controlled molecular weight cresol novolacs. The degree of molecular weight control achievable and properties such as glass transition temperatures, molecular weight distributions, and melt viscosities will be discussed. These cresol novolacs have also been crosslinked with epoxies. The network formation reactions and their properties will be addressed in a separate paper. 3.2. Experimental 3.2.1. Materials Ortho-cresol (99+%), para-cresol (99%), 2,6-dimethylphenol (99%), paraformaldehyde (powder, 95%), formaldehyde (37 wt % solution in water), and oxalic acid dihydrate (99%) were obtained from Aldrich. A commercial phenolic resin was kindly provided by Georgia-Pacific (Product #GP-2073). All reagents were used as received. 3.2.2. Molecular Weight Calculations The following method was used to calculate the stoichiometric ratio of monomers required to obtain specified number average molecular weights. The molecular weight of two endcapping molecules, 2,6-dimethylphenol, and one methylene linkage, -CH2-, were subtracted from the total targeted molecular weight. The remaining weight was divided by the molecular weight of each repeat unit (120 g/mol) to obtain the number of repeat units within the chain (x). The stoichiometric ratio then consisted of two moles of 2,6dimethylphenol, x moles of cresol, and x+1 moles of formaldehyde. 3.2.3. Synthesis of 2,6-Dimethylphenol Endcapped Cresol Novolac Resin Ortho-cresol novolac and para-cresol novolac resins were prepared in the same manner. The following shows a sample reaction for preparing a 2000 g/mol ortho-cresol 79 novolac resin. In a resin kettle equipped with a stainless steel mechanical stirrer and a condenser connected to an outlet, ortho-cresol (303.5g, 2.81mol) and 2,6-dimethylphenol (47.2g, 0.39mol) and paraformaldehyde (94.9g, 3.0 mol) were added. This mixture, along with oxalic acid dihydrate (2.5 wt. % (2.14 mol %) based on the weight of cresol, 7.59g) was heated for approximately 6 hours at 100C, then an ~10mol % excess of formaldehyde (37 wt. % formaldehyde in water, 27 ml) was added to the reaction. The reaction was continued for an additional 18 hours. It was washed twice with boiling deionized water, then stripped under mild vacuum while being slowly heated to 215C. 3.2.4. Sample Preparation for Viscosity Measurements All cresol novolac resins and the control commercial phenolic resin were vacuum stripped (30 Hg) for 2 hours at 165C prior to any measurements. 3.2. Characterization 3.2.1. Nuclear Magnetic Resonance Spectroscopy 1 H NMR and 13 C NMR spectra were obtained on a Varian Unity 400 NMR The spectrometer. For 1H NMR, 5 mm diameter tubes containing approximately 20 mg samples dissolved in DMSO-d6 were analyzed under ambient conditions. experimental parameters included a 1.0 second relaxation delay, 23.6 degree pulse, and 6744.9 Hz spectral width. Thirty-two repetitions were performed for each sample. For 13 C NMR, samples of approximately 0.6 g were dissolved in ~ 2 ml acetone or DMSO. The samples were placed in 10 mm diameter tubes for analysis under ambient conditions. An inverse gated decoupling technique with a 90 degree pulse, a 6 second relaxation delay, a frequency of 100.578 MHz, and 1.2 seconds acquisition time were used to obtain quantitative 13C NMR data. Approximately 1000 repetitions were used for each sample. 3.2.2. Gel Permeation Chromatography GPC was conducted on a Waters GPC/ALC 150-C chromatograph equipped with a differential refractometer detector connected in parallel to a differential viscometer 80 detector Viscotek model 150R. The injection and column compartment, connecting line, and DV detector were individually controlled and maintained at the same temperature (60C). The signals from the RI and DV detectors permitted the calculation of intrinsic viscosity for universal calibration purposes by using Viscotek software Unical 4.04 assuming that the polymer concentration at the outlet of the SEC columns approached infinite dilution due to separation and column dispersion. The mobile phase was NMP (dried over phosphorus pentoxide, then vacuum distilled) with a flow rate of one ml/min. The columns were Styragel HT with pore sizes of 103 and 104 angstroms. The injection volume was 100 l. 3.2.3. Viscosity Determinations Complex viscosities were obtained from a Bohlin VOR Rheometer operating in continuous oscillation mode at a frequency of 1 Hz. Temperature control was accomplished with a Bohlin HTC. The auto-strain was set to maintain the torque at 25% of the maximum torque allowed. The maximum strain for the instrument was 0.25. Approximately 0.7g of cresol novolac pellets were placed between the preheated 25 mm diameter parallel plates of the rheometer. The gap was closed to approximately 1mm and the sides were scraped to remove excess sample before the run was started. The glass transition temperatures of neat resins were obtained with a PerkinElmer DSC-7 instrument. The DSC was calibrated with indium and zinc standards, and ice water was used as the coolant. Samples in aluminum pans were heated from 20C to 180C. The glass transition temperatures were calculated as the midpoints of the curves. 3.3. Results and Discussion 3.3.1. Introduction A series of linear, controlled molecular weight, 2,6-dimethylphenol endcapped cresol novolac resins have been synthesized via electrophilic aromatic substitution (Figure 3. 2). The use of ortho- or para-cresol as a monomer allows for preparing linear oligomers since the cresol ring has only two activated positions for formaldehyde 81 substitution. By contrast, phenol has three reactive sites (i.e. both ortho and the para positions) and therefore, branching is inevitable as higher molecular weight develops. For example, branching has been shown to occur significantly once the molecular weight reached 900-1000 g/mol.118 Addition of calculated amounts of 2,6-dimethylphenol endgroups to the linear ortho- or para-cresol-formaldehyde reactions allows for controlled molecular weight materials to be generated. OH + CH 3 H3C HO H3C OH H O CH 2 O H n CH3 + CH3 2.5 wt % oxalic acid dihydrate o 100 C OH CH2 CH2 m CH3 CH3 OH CH3 Figure 3. 2. Synthesis of 2,6-dimethylphenol endcapped para-cresol novolac resins The reactivity rates for phenol versus cresol formaldehyde substitutions are different. Phenol reacts with formaldehyde approximately three times faster than orthoor para-cresol.134 Water reduces the rate of reaction between phenol and formaldehyde if used in large amounts.121 In this work, cresol novolacs were prepared with paraformaldehyde, as opposed to aqueous formaldehyde, to achieve faster reaction rates. Paraformaldehyde contains only 1-9 wt. % water whereas the formaldehyde typically used in phenolic syntheses contains approximately 50 to 63 wt. % water. Oxalic acid dihydrate was used as the catalyst since it is a relatively strong acid. Oxalic acid dihydrate is preferred over other catalysts because resins with less color can 134 M. M. Sprung, Reactivity of Phenol Toward Formaldehyde, Journal of Applied Polymer Science 63(2), 334-343 (1941). 82 be obtained. Moreover, there is no need to remove the catalyst after the reaction since it can be thermally decomposed to CO, CO2, and water above approximately 180C.118 In these reactions, the initial viscosities were low and the solutions were miscible. However, as the reactions proceeded and molecular weights increased, the solutions phase separated. The low molecular weight oligomers formed a water-insoluble melt, while the acid catalyst and the formaldehyde probably remained predominantly in the aqueous phase. Slow reaction rates were observed which are probably attributable to the 2-phase nature. The formaldehyde added toward the end of the reactions was in an aqueous solution. Water was desirable in this stage to plasticize the reaction mixtures and lower the viscosities. This was particularly important in the syntheses of higher molecular weight cresol novolacs when the viscosities were high. 3.3.2. Molecular Weight Control and Calculations Typical phenolic novolac syntheses lack molecular weight control. The reactions are generally terminated after a certain reaction time or once a specified viscosity is reached.135 The molecular weights of the cresol novolac resins described herein were strategically controlled by the stoichiometric ratio of cresol to 2,6-dimethylphenol (Table 3. 1). The molecular weights of oligomers increased as the amount of endcapping reagent was decreased. The number average molecular weights in these cresol novolac syntheses was controlled by the cresol to endgroup molar ratio. However, in contrast to usual practice, it was necessary to add formaldehyde in excess to achieve full conversions of phenolic reactive ring positions. When the calculated amounts of formaldehyde were used, the molecular weights of products were always lower than the targeted molecular weights, and it was evident from 13 C NMR spectra that unreacted ring positions on cresols remained under such conditions. Formaldehyde was added in two portions to couple all of the reactive sites on cresol. Initially, the stoichiometrically calculated amount of formaldehyde was charged to the reactions with cresol and 2,6-dimethylphenol at 100C. 135 S. R Sandler and W. Karo, Polymer Synthesis, 2nd edition, Academic Press, Boston, Vol. 2, 1992, p49-86. 83 The early stages of reactions were exothermic and the reactions refluxed. After 6 hours, more formaldehyde (10 mol % of the calculated amount in the form of formalin) was added to ensure that sufficient formaldehyde was available to complete the reactions. Targeted molecular weights were consistently achieved using the approach of adding excess formaldehyde as described above. This suggests a reversible reaction between cresol or its derivatives and formaldehyde whereby substitution and elimination of formaldehyde occurs. This would allow for coupling regenerated ring positions and methylols to form methylene linkages and achieve the targeted molecular weights. It is also possible that some gaseous formaldehyde, formed by depolymerization of polyoxymethylene, escaped from the reactions during the initial exothermic stages. The required stoichiometries for controlling molecular weights were calculated using the Carothers approach. A step-growth polymerization was considered involving the reaction of monomers AWA, BYB, and AZ in which the functional groups A react with functional groups B. It was assumed that very high conversion was achieved and that stoichiometric amounts of A and B groups were in the reaction feed. This latter assumption can be expressed as N(BB) = N(AA) + N(A)/2 where N(BB) = moles of BYB N(AA) = moles of AWA N(A) = moles of AZ The reaction of N(AA) with N(BB) yields a statistically determined size distribution of N(A)/2 moles of product molecules (oligomeric and polymeric) plus byproduct molecules which can be represented schematically as follows: (3. 1) N(AA) + N(BB) + N(A) [ZA-BYB-(AWA-BYB-)x AZ] + by-products (3. 2) 84 A combined endgroup is defined as ZA-BYB- plus AZ and an internal repeat unit is defined as AWA-BYB-. Each mole of a given product molecule therefore has one mole of combined endgroup and x moles of internal repeat units. The number average molecular weight of the reaction product is then: Mn = total mass of product molecules moles of product molecules Mn = (m e + x m u ) N(A)/2 (3. 3) where me is the molar mass of the combined endgroup and mu is the molar mass of an internal unit. The summation is over all N(A)/2 moles of product molecules and all x values. This leads to M n = me + X n mu (3. 4) where Xn is the number average number of internal units in the product molecules. By inspection of the schematic molecular structure (3.2) and equation (3.4), one sees that xn = N(AA) N(AA) =2 N(A)/2 N(A) (3. 5) Arbitrarily choosing the number of moles of AWA, then substituting equation (3.5) into equation (3.4) and rearranging, one now has a complete description of the feed composition required to achieve a target Mn: moles of monomer AWA: moles of endcapper AZ: moles of monomer BYB: N(AA) 2 mu N(A) = N(AA) M n - me N(BB) = N(AA) + N(A)/2 (3. 6) In the present work AWA is cresol, AZ is 2,6-dimethylphenol, and BYB is formaldehyde. 85 Table 3. 1. 13 Molecular weight of ortho- and para-cresol novolac resins calculated using C NMR. The molecular weights were controlled by adjusting NAA/NZA ratio. Target Mn (g/mol) 500 1000 1500 2000 NAA (cresol) (mole) 1 1 1 1 NZA (2,6-DMP*) (moles) 0.984 0.323 0.193 0.138 Ortho Series Mn (g/mol) 490 930 1380 2250 Para Series Mn (g/mol) 510 1010 1460 2150 *2,6-DMP 2,6-dimethylphenol 3.3.3. Structure of Reaction Intermediates and Products 13 C NMR has been used extensively to characterize phenolic resins and their synthesis and crosslinking reactions.136,137,138,139,140 Carbon chemical shifts of typical 136 X. Zhang and D. H. Solomon, The Chemistry of Novolac Resins: 9. Reaction Pathways Studied via Model Systems of ortho-Hydroxybenzylamine Intermediates and Phenols, Polymer 39(24), 6153-6162 (1998). 137 X. Zhang, M. G. Looney, D. H. Solomon, and A. K. Whittaker, The Chemistry of Novolac Resins: 3. 13C and 15N n.m.r. Studies of Curing with Hexamethylenetetramine, Polymer 38(23), 5835-5948 (1997). 138 P. Luukko, L. Alvila, T. Holopainen, J. Rainio, and T. T. Pakkanen, Optimizing the 13 Conditions of Quantitative 139 C NMR Spectroscopy Analysis for Phenol-Formaldehyde Resole Resins, Journal of Applied Polymer Science 69, 1805-1812 (1998). M. G. Kim, L.W. Amos, and E. E. Barnes, Study of the Reaction Rates and Structures of a Phenol Formaldehyde Resol Resin by C-13 NMR and Gel-Permeation Chromatography, Industrial & Engineering Chemistry Research 29(10), 2032-2037 (1990). 140 P. W. Kopf and E. R. Wagner, Formation and Cure of Novolacs-NMR Study of Transient Molecules, Journal of Polymer Science: Polymer Chemistry Edition 11(5), 939-960 (1973). 86 phenolic resins and some related reaction intermediates are provided in Table 3. 2. Quantitative 13 C NMR was used in this study to monitor reaction progress and to determine the molecular weights of the final products. Acetone was the preferred solvent for the ortho-cresol novolacs since its carbon peak did not overlap with the sample peaks but para-cresol novolacs were not soluble in acetone. DMSO, which was used to analyze the para-cresol novolacs, resonates at 40 ppm and overlapped with the para-para methylene linkages. There were no significant differences in the chemical shifts in these two solvents. Table 3. 2. 13 C NMR assignments intermediates137 for novolac resins and related reaction Chemical Shift Region (ppm) 150-156 127-135 121 116 85.9 81.4 71.1 68.2 40.8 35.5 31.5 Assignment Hydroxyl-substituted phenolic carbons Other phenolic carbons Para-unsubstituted phenolic carbons Ortho-unsubstituted phenolic carbons HO-CH2-O-CH2-OH HO-CH2-OH Para-linked dimethylene ether Ortho- linked dimethylene ether Para-para methylene linkages Ortho-para methylene linkages Ortho-ortho methylene linkages 13 C NMR spectra provided an excellent means for structural characterization of the cresol novolac oligomers (Figure 3. 3). The chemical shifts for cresol novolac resins matched well with those observed for phenolic novolacs. The peaks between 150-156 ppm represent hydroxyl substituted aromatic carbons. Ortho-unsubstituted aromatic 13 carbons resonate at 118 ppm, and para-unsubstituted aromatic carbons are observed at 120 ppm. The rest of the aromatic carbon peaks resonate between 121 and 136 ppm. C NMR spectra can define three distinct types of methylene linkages between aromatic 87 rings, para-para (41 ppm), ortho-para (36 ppm), and ortho-ortho (31.5 ppm). The peaks between 15 and 18 ppm represent the methyl carbons on the aromatic rings. The formation of oligomers in bulk reactions at 100C with 2.5 wt. % oxalic acid catalyst was monitored by 13 C NMR (Figure 3. 3). The first spectrum represents the reaction mixture immediately after becoming homogeneous (~20 minutes). The reaction had clearly begun at this stage. This was evidenced by the downfield shift of hydroxylsubstituted carbons, shifts in aromatic regions, the appearance of acetone soluble oxymethylene peaks (83-93 ppm), the formation of para-linked dimethylene ethers (67.5 ppm) and methylols (64 ppm), and the formation of para-para and ortho-para methylene linkages. It should be noted that paraformaldehyde was insoluble in the acetone NMR solvent, but its derivatives were soluble. As the reactions progressed, the amount of methylol intermediates and ortho and para unsubstituted aromatic carbons decreased, and the peaks for methylene linkages increased. positions. As shown in the literature,134 para positions on phenolic compounds react faster than ortho positions. 13 13 C NMR spectra showed no ortho or para unsubstituted carbon peaks in products indicating full conversion of reactive ring C NMR spectra revealed that para-para methylene linkages formed most rapidly followed by ortho-para methylene linkages (Figure 3. 3). Orthoortho methylene linkages were observed in small amounts after 1 hour. 88 Aromatic carbons Hydroxyl Acetone Methylene linkages 80 60 40 20 carbons o-cresol 160 140 120 100 Methyl carbons PPM 2,6-dimethylphenol 160 140 120 100 80 60 40 20 PPM 20 minutes 160 140 120 100 80 60 40 20 PPM 1 hour 160 140 120 100 80 60 40 20 PPM 8 hours 160 140 120 100 80 60 40 20 PPM Product 160 140 120 100 80 60 40 20 PPM Figure 3. 3. 13 C NMR spectra monitoring a 2000g/mol ortho-cresol novolac resin synthesis as a function of reaction time. The product was reacted for 24 hours at 100C, then heated to 200C under mild vacuum to decompose the catalyst. 89 Hydroxymethyl condensation reactions, which eliminate water to form dimethylene ether linkages, are prevalent under acidic conditions. It has been suggested that dimethylene ether linkages decompose at elevated temperatures to form methylene bridges between rings.118 13 C NMR monitoring the reaction progress of these cresol novolac reactions confirmed the formation of both ortho (66.5 ppm) and para linked dimethylene ethers (67.5 ppm). Para-dimethylene ether linkages formed early and decomposed as the reaction proceeded (Figure 3. 3). Ortho-linked dimethylene ethers formed later and remained in the oligomer chain even after heating to 200C to decompose the catalyst. The high stability of ortho-linked dimethylene ethers was attributed to the formation of strong intramolecular hydrogen bonding (Figure 3. 4). OH OH CH2OH + HOCH2 OH O - H2O H O Figure 3. 4. Condensation of ortho-hydroxymethyl substituent forming stable ortholinked dimethylene ethers The residual dimethylene ether linkages can only account for a small fraction of the excess formaldehyde required in these reactions to achieve the targeted molecular weights (Table 3. 3). These dimethylene ether linkages do not change molecular weights significantly. Table 3. 3. Mole percent ortho-dimethylene ether linkages Mn (g/mol) 500 1000 1500 2000 Ortho series 1.3 1.5 1.6 1.7 Para series 2.8 3.1 3.6 3.5 90 13 C NMR peaks for methyl carbons were also used to monitor the cresol novolac reactions (Figure 3. 5). OH 16.8 CH3 OH 16.7 CH3 CH3 OH 16.6 CH3 OH OH 16.2 CH3 CH3 16.3 CH3 A B C D E ABC DE o-cresol 2,6dimethylphenol 20 minutes 1 7 1 6 1 5 1 hour 8 hours Product 17 Figure 3. 5. 16 15 Expanded 13C NMR spectra monitoring a 2000 g/mol ortho-cresol novolac resin synthesis as a function of reaction time The methyl groups on ortho-cresol (peak E) resonate at 16.13 ppm, and the methyl groups on 2,6-dimethylphenol (peak C) resonate at 16.60 ppm. The methyl carbon on both monomers shift downfield upon reaction of one site, then the cresol methyl shifts further downfield upon reaction of the second site. The endgroup methyls 91 are not well resolved with the methyl groups on internal units due to the similarities in their structures. A small peak at 15.4 ppm was attributed to methyl carbons on cresol units linked with dimethylene ethers. This corresponds well with ortho methyl carbon shift for 2-hydroxymethyl-4,6-dimethylphenol (15.53 ppm).141 The the product, further confirming quantitative conversion. The molecular weights of cresol novolac oligomers were calculated by comparing the peak intensities corresponding to methyl carbons on the endgroups versus the internal methyl carbons. Since the two peaks are not well resolved, a deconvolution technique was used to determine the peak integrations (Figure 3. 6). The peak area corresponding to the 2,6-dimethylphenol endgroups accounted for four methyl carbons per chain. The relative number of methyl groups within the repeat units was determined by ratioing the peak integrations of the interior methyl carbons versus the endgroup carbons, then multiplying by 4. 13 C NMR analysis showed that all the reactive positions on cresol and 2,6-dimethylphenol were reacted in CH 3 HO CH 3 CH 2 OH a CH 3 CH 2 n b CH 3 OH CH 3 a b 17.4 17.2 17.0 16.8 16.6 16.4 16.2 PPM Figure 3. 6. Deconvolution of methyl carbon peaks 13 The methyl regions of C NMR spectra for a series of ortho-cresol novolac oligomers with different molecular weights were compared (Figure 3. 7). The peak integration ratio of internal methyl carbons to those on the endgroups (peak a to b) 141 K. Lenghaus, G. G. Qiao, and D. H. Solomon, Model Studies of the Curing of Resole Phenol-Formaldehyde Resins Part 1. The Behavior of ortho Quinone Methide in a Curing Resin, Polymer 41, 1973-1979 (2000). 92 increased as the molecular weight increased. This was expected since more repeat units, relative to endgroups, were incorporated as higher molecular weights developed. b 500 g/mol a 1000 g/mol a 1500 g/mol a 2000 g/mol b a b b 5 1 7 .0 1 6 .5 1 6 .0 1 7 .0 1 6 .5 1 6 .0 5 1 7 .0 1 6 .5 16. 17.0 16.5 16.0 Figure 3. 7. Expanded 13C NMR spectra of a series of ortho-cresol novolac resins with various molecular weights: a) methyl carbons within the repeat units, b) methyl carbons on the endgroups The type of methylene linkages (ortho-ortho, ortho-para, and para-para) and the amount in which they form can be calculated using 13 C NMR. Since the endgroups formed only para methylene linkages, the number of para linked species was higher for low molecular weight oligomers (Table 3. 4). As the molecular weight was increased, the para-para, ortho-para, and ortho-ortho linked methylenes approached the expected 1:2:1 statistical distribution. Table 3. 4. Percentage isomers formed in ortho-cresol novolac resins Mn (g/mol) 500 1000 1500 2000 p-p (%) 45.2 31.2 30.0 29.2 o-p (%) 47.8 49.7 49.8 49.3 o-o (%) 7.0 19.1 20.1 21.5 93 Para-cresol novolacs syntheses, monitored by 13C NMR, showed similar reaction progress as the ortho-cresol novolac reactions (Figure 3. 8). Formations of para-cresol novolacs were accompanied by the downfield shift of hydroxyl substituted aromatic carbon peaks, increases in both the ortho-ortho (36.5 ppm) and the ortho-para (40.5 ppm) methylene carbon peaks, and upfield shifts of methyl carbon peaks. The DMSO solvent peak overlapped with the para-para methylene carbon peak. However, this was not as important for monitoring the syntheses of para-cresol novolac resins since parapara methylene linkages only formed when two 2,6-dimethylphenol units dimerized and this was assumed to form small in amounts (especially at low 2,6-dimethylphenol concentrations). 94 Hydroxyl Aromatic carbons carbons p-cresol DMSO Methyl carbons Meth ylene 2,6-dimethylphenol 2 hours 10 hours 16 hours Product 160 140 13 120 100 80 60 40 20 PP Figure 3. 8. C NMR spectra of a 2000g/mol para-cresol novolac resin synthesis monitored as a function of reaction time Two types of methyl groups were observed for the para-cresol novolacs (Figure 3. 9). The para methyl carbons within the repeat units resonate at 26.2 ppm, and the ortho methyl carbons on the endcapping reagent resonate at 22.4 ppm. The formation of ortho-para and ortho-ortho methylene linkages was also monitored. Several peaks in the 95 ortho-ortho methylene region were observed early in the reaction which were attributed to reaction intermediates. At the end of the reaction, only one sharp peak was present in this region. d CH3 OH CH3 OH CH3 + CH3 a OH b CH3 H O CH2 O H + n H3C HO H3C 2.5 wt % oxalic acid dihydrate 100oC e CH2 OH m CH c3 f CH2 p-cresol a b 2,6-dimethylphenol e f c 2 d 10 hours 16 hours produc 40 35 30 25 20 Figure 3. 9. Expanded 13C NMR spectra monitoring the synthesis of a 2000g/mol paracresol novolac resin 96 1 H NMR was used to analyze the monomers and the novolac oligomers to confirm their expected structures (Figure 3. 10 and Figure 3. 11). Six distinct sets of peaks were in the ortho-cresol spectrum (Figure 3. 10a). The 1H NMR spectrum of a 2000 g/mol ortho-cresol novolac resin (Figure 3. 10b) had broader peaks than the monomer due to the different isomer sequences present. The methyl protons resonated between 1.9 and 2.2 ppm and three sets of peaks between 3.5 and 3.9 ppm were from protons on the methylene linkages. These corresponded to ortho-ortho (3.5ppm), orthopara (3.6ppm), and para-para (3.7ppm) methylene units. Small amounts of dimethylene ether also formed which resonated at 4.8 ppm. Similar peaks were observed for para-cresol novolacs (Figure 3. 11). Fours sets of peaks were observed for the para-cresol monomer. The main difference between the chain structures of ortho-cresol versus para-cresol novolac resins was the type of methylene linkages. Only ortho-ortho linkages are possible between the repeat units for para-cresol novolac resins. Therefore, protons on the methylene linkages resonated closer together and were well resolved from any residual water. The hydroxyl proton peaks were also sharper for the para-cresol novolacs. 97 c c b b OH a CH3 a water DMSO 9 8 7 6 5 4 3 2 1 CH3 HO CH3 f CH2 h OH e CH3 CH2 CH3 OH n CH3 g a c d b DMSO 9 8 1 7 6 5 4 3 2 PPM Figure 3. 10. H NMR spectra of a) ortho-cresol, and b) a 2000 g/mol ortho-cresol novolac 98 OH d c b a d b c CH3 a wate DMSO 8 CH3 HO CH3 f CH2 CH3 e h OH g 6 CH3 CH2 n OH CH3 4 2 g e f h 10 9 8 7 6 5 4 3 2 PPM Figure 3. 11. 1 H NMR spectra of a) para-cresol, and b) a 2000 g/mol para-cresol novolac 99 3.3.4. Molecular Weight and Molecular Weight Distributions Determined via GPC The growth of molecular weight with time for both ortho- and para-cresol novolac synthesis was monitored using gel permeation chromatography (Figure 3. 12). In gel permeation chromatography, the higher molecular weight oligomers bypass the smaller pores in the packing column and elute faster, and therefore, appear at lower retention volumes. As the reactions progressed, the GPC peaks shifted to lower elution volumes and the area in the region where monomers and low molecular weight oligomers eluted (retention volume 32.636 ml) decreased. The final products were comprised of mostly higher molecular weight oligomers eluting between 27 and 32.6 ml. 45.0 1 hour 3 hours 8 hours 14 hours product RI Chromatogram 36.0 Response (mV) 27.0 18.0 9.0 0.0 25.0 27.4 29.8 32.2 34.6 37.0 Retention Volume (mL) Figure 3. 12. GPC monitoring the synthesis of a 2000 g/mol ortho-cresol novolac resin as a function of reaction time GPC traces of ortho- and para-cresol novolacs with approximate molecular weights of 500 g/mol, 1000 g/mol, 1500 g/mol, and 2000 g/mol were measured (Figure 3. 13). For both series, the peaks shifted toward lower elution volumes as the average molecular weights increased. 100 45.0 RI Chromatogram 500 g/mol 1000 g/mol 1500 g/mol 2000 g/mol 36.0 A Response (mV) 27.0 18.0 9.0 0.0 25.0 27.4 29.8 32.2 34.6 37.0 Retention Volume (mL) 55.0 g 500 g/mol 1000 g/mol 1500 g/mol 2000 g/mol 44.0 B Response (mV) 33.0 22.0 11.0 0.0 25.0 27.4 29.8 32.2 34.6 37.0 Retention Volume (mL) Figure 3. 13. GPC of cresol novolac resins with various molecular weights: a)ortho-cresol novolac, b)para-cresol novolac GPC was used to qualitatively compare the molecular weights. Absolute molecular weights could not be derived from GPC using the viscosity detector due to their low molecular weights and correspondingly low solution viscosities. Thus, only the polydispersities of these oligomers and their intrinsic viscosities were evaluated (Table 3. 5). Since cresol novolac resin syntheses are condensation polymerizations, the polydispersity should approach 2 as the molecular weight increases. Low molecular 101 weight oligomers (500-1000 g/mol) had low polydispersities, but the polydispersity increased as the molecular weight was increased. As expected, the intrinsic viscosities increased as the molecular weight increased. However, when comparing ortho-cresol novolacs to para-cresol novolacs with similar number average molecular weights, ortho-cresol novolacs consistently had higher intrinsic viscosities (Table 3. 5). This suggested that ortho-cresol novolacs have higher hydrodynamic volumes in the NMP chromatography solvent. It was previously reported that different novolac isomers have different elution behaviors.142,143 The difference in hydrodynamic volumes from isomer structures of ortho-ortho, ortho-para, and para-para was due to varying degrees of solvation. Oligomers with high compositions of orthoortho linkages (i.e. para-cresol novolac resins) may have a higher degree of intramolecular hydrogen bonding, resulting in less solvation and lower hydrodynamic volumes. randomly. The propensity for hydrogen bonding may be disrupted for ortho-cresol novolacs since three types of methylene linkages are available and were distributed 142 T. R. Dargaville, F. N. Guerzoni, M. G. Looney, D. A. Shipp, D. H. Solomon, and X. Zhang, Determination of Molecular Weight Distribution of Novolac Resins by Gel Permeation Chromatography, Journal of Polymer Science. Part A: Polymer Chemistry 35(8), 1399-1407 (1997). 143 T. Yoshikawa, K. Kimura and S. Fujimura, The Gel Permeation Chromatography of Phenolic Compound, Journal of Applied Polymer Science 15, 2513-2520 (1971). 102 Table 3. 5. Polydispersities and intrinsic viscosities of cresol novolac resins Cresol Mn (g/mol) 500 1000 1500 2000 500 1000 1500 2000 Polydispersity 1.41 1.33 1.73 1.61 1.25 1.62 1.62 2.08 Intrinsic Viscosity* (dL/g) 0.037 0.049 0.066 0.075 0.031 0.046 0.052 0.067 ortho para * in NMP solvent at 60C Glass transition temperatures for both the ortho- and para-cresol novolacs increased as the molecular weight increased (Table 3. 6). The Tgs ranged from approximately 40C for 500 g/mol resins to 105-110C for oligomers with number average molecular weights of 2000 g/mol. No significant differences were observed between the Tgs of ortho and para-cresol novolacs with similar molecular weights. Table 3. 6. Tg of cresol novolac resins as a function of molecular weight Ortho Series Mn* (g/mol) 490 930 1380 2250 Tg (C) 40 76 92 104 510 1010 1460 2150 Para Series Mn* (g/mol) Tg (C) 43 76 99 110 * Calculated using 13C NMR 103 3.3.5. Dynamic Viscosities of Cresol Novolac Resins Efficient melt composite fabrication procedures require low viscosity resins to wet out the fiber. The viscosity profiles of a typical phenolic novolac resin with a molecular weight of ~700 g/mol (Mn) and a series of cresol novolacs were examined as a function of temperature at a heating rate of 2.5C/minute (Figure 3. 14). All samples were vacuum stripped at 165C for 2 hours prior to the measurements to remove residual water. As molecular weights were increased, the temperatures required for the viscosity to fall to 10 Pa*s increased for both series of cresol novolac materials. Ortho-cresol novolacs (Figure 3. 14. a) had similar viscosities to the para-cresol novolacs (Figure 3. 14. b) with similar molecular weights at any given temperature. The viscosity of the 2000 g/mol para-cresol oligomer decreased with increased temperature until ~180C, then gradually increased upon further heating. The reason for this increase in viscosity at high temperatures is presumed to be attributable to degradative crosslinking, but this is as yet unclear. It is possible that the increase in viscosity is due to o,o-dimethylene ether oxidation to o,o-dimethylene ether hydroperoxide followed by degradative crosslinking. The viscosity of the phenolic control reached 10 Pa*s at ~165C, a higher temperature than that required by a 1500 g/mol cresol novolacs to reach the same viscosity. This suggested that the cresol oligomers may be significantly more amenable to melt processing relative to phenol derived oligomers due to a wider processing window. The phenolic novolac control material behaved similarly to the 2000 g/mol para-cresol novolac resin where the viscosity showed a minimum at ~180C. 104 20 Viscosity (Pa s) 15 10 5 0 80 100 120 140 160 o A 500g/mol 1000 g/mol 1500 g/mol 2000 g/mol control 180 200 Temperature ( C) 25 20 Viscosity (Pa s) 15 10 5 0 80 100 120 140 160 o B 500 g/mol 1000 g/mol 1500 g/mol 2000 g/mol 180 200 T emperature ( C) Figure 3. 14. Dynamic viscosity of cresol novolacs measured as a function of molecular weight a) ortho-cresol novolac resins, and b) para-cresol novolac resins 3.4. Conclusions A series of controlled molecular weight, 2,6-dimethylphenol endcapped cresol novolac resins have been synthesized. An excess of formaldehyde was required to achieve the targeted molecular weights. determined from 13 The number average molecular weights, C NMR spectra, showed good agreement with the targeted number 105 average molecular weights for both ortho- and para-cresol novolacs. The amount of ortho-ortho, ortho-para and para-para methylene linkages for the ortho-cresol novolac resins, also determined from 13 C NMR, approached the statistical distribution as the molecular weights were increased and the contributions from endgroups were less significant. The polydispersities obtained from GPC suggested that molecular weight distributions were reasonably narrow (< 2). The glass transition temperatures increased from ~40C to 110C as the molecular weights were increased from 500g/mol to 2000g/mol, but there were no significant differences between the ortho and para-cresol novolacs with similar molecular weights. In general, the viscosities of ortho- and para-cresol novolacs with similar molecular weights were almost identical, but were significantly lower than phenol based oligomers. These cresol novolac resins will be crosslinked with various epoxy resins to form void-free flame retardant networks. Network properties such as toughness, flame retardance, and water uptake, as well as processability will be investigated. 106 Chapter 4. Structure-Property Relationships of Cresol Novolac/Epoxy Networks 4.1. Introduction A 2000 g/mol linear ortho-cresol novolac resin was crosslinked with various epoxies at defined compositions for use as tough and flame retardant matrix resins. The cresol novolac/epoxy network mechanical and flame properties were compared to a phenolic control, an epoxy control, and a phenolic novolac/epoxy network. Undoubtedly the network properties depended on both the chemical and the network structures, thus this research focused on determining the influences of network structures on properties. The processability of cresol novolac/epoxy mixtures was also evaluated. 4.1.1. Crosslink density and its affects on network properties Crosslink density () or degree of crosslinking is a measure of the total links between chains in a given mass of materials. It is frequently represented in terms of the molecular weight between crosslinks (Mc) where higher crosslink densities correspond to lower Mc values and vice versa.144 The portion of crosslinked networks that remains soluble and has finite molecular weights is referred to as the sol fraction. The insoluble crosslinked portion is termed the gel fraction. Two general types of experimental methods widely used to determine the degree of crosslinking are based on swelling and mechanical testing. However, the observed molecular weight between crosslinks frequently deviates from the theoretically calculated Mc value due to over-simplified assumptions relating swelling or mechanical properties to crosslink densities. For example, trapped physical entanglements should improve the mechanical properties and therefore lead to higher apparent crosslink densities. The presence of dangling chain ends, on the other hand, does not contribute mechanically, and therefore is expected to reduce the apparent crosslink densities. It is also argued that not 144 Encyclopedia of Polymer Science and Engineering, Vol 4. J. I. Kroschwitz, Ed. John Wiley & Sons, New York, 1986. 107 all crosslink points are mechanically effective which may also introduce errors in the calculations. For a perfect network with no elastically inactive chains or dangling ends, the swelling by a good solvent at equilibrium is given by 1/ = M c = - V1 P 1/3 - /2 ln (1 - ) + + 1 2 (4. 1) where V1 = molar volume of the solvent P = density of the polymer = volume fraction of the polymer in the swollen state 1 = polymer solvent (Flory-Huggins) interaction parameter The elastically inactive chains in the networks may be accounted for by dividing the Mc values by a correction factor, 1-2Mc/Mn, where Mn is the number average molecular weight of the polymer before crosslinking. This procedure is most accurate for calculating crosslink densities of typical vulcanized rubbers (Mc ~ 5000 g/mol). Although the general trend of increasing Mc with increasing swelling still holds true for more highly crosslinked networks, this method is no longer effective at these lower Mc values.145 Crosslink densities can also be derived from mechanical testing such as stressstrain or stress-relaxation experiments for polymers at temperatures well above their Tgs. According to the theory of rubber elasticity, the shear modulus (G) of an ideal rubber is given by G = RT = RT/Mc where = density, R = gas constant, T = absolute temperature (K) 145 (4. 2) L. E. Nielsen, Crosslinking-Effects on Physical Properties of Polymers, Journal of Macromolecular Science-Reviews Macromolecular Chemistry C3(1), 69-103 (1969). 108 The correction factor described previously is sometimes incorporated equation (4.2) to account for dangling polymer chain ends G = RT/Mc (1-2Mc/Mn) The Youngs modulus (E) is approximated by E = 3G (4. 4) (4. 3) Four major assumptions are made in developing the statistical theory of rubber elasticity: 1) the internal energy of the system is independent of the conformation of the individual chains, 2) the network chains obey Gaussian statistics, 3) the total number of conformations of an isotropic network is the sum of the number of conformations of the individual network chains, and 4) crosslinked junctions in the network are fixed at their mean position; an affine transformation occurs upon deformation.146 Crosslinking and the crosslink density greatly influence material properties especially above Tg. Unlike linear polymer chains, which undergo viscous flow above Tg and are incapable of sustaining a constant load, crosslinked networks show elastic characteristics above their Tgs. Crosslinking generally improves physical properties especially above the glass transition temperature. For example, network glassy moduli generally vary slightly as a function of crosslink density, whereas the rubbery moduli increase significantly with increasing degrees of crosslinking. function of crosslink density has been approximated by Tg = A (4. 5) The glass transition temperatures increase with higher degrees of crosslinking. The extent of Tg increase as a 146 J. J. Aklonis and W. J. MacKnight, Introduction to Polymer Viscoelasticity, Second Edition, John Wiley and Sons, New York, 1983. 109 where is the moles of crosslinks per gram of polymer, and A is a constant on the order of 104-105 depending on the material.147 The dependence of Tg may be more pronounced in more highly crosslinked networks than in lightly crosslinked networks.144 Crosslinking generally decreases creep, compression set, and stress relaxation, and increases tensile strength; it increases refractive index, and lowers thermal expansion and heat capacity. Fracture toughness is also strongly dependent on the network crosslink density. Highly crosslinked networks are generally brittle and their toughness increases as the crosslink density decreases until the network becomes too loose. Tyberg et al.148,149 investigated the effects of crosslink density on network properties, including fracture toughness, for phenolic novolac/epoxy networks. The network crosslink density was adjusted by controlling the stoichiometric ratio of phenolic hydroxyl to epoxy groups. A phenolic novolac with a functionality of approximately 7 (determined via 1H NMR) was used. A schematic representing an idealized crosslinking reaction between phenolic novolac (f=7) and a diepoxide is depicted in Figure 4. 1. If equimolar amounts of phenolic hydroxyl and epoxy groups are used, and assuming 100 % conversion, highly crosslinked materials are expected to form. As the stoichiometric ratio is offset to increase the amount of phenolic hydroxyl groups (or epoxy groups), the network crosslink density decreases. This trend continues until the amount of epoxy (or novolac) present is insufficient to generate fully crosslinked networks. 147 T. G. Fox and S. Loshaek, Influence of Molecular Weight and Degree of Crosslink on the Specific Volume and Glass Temperature of Polymers, Journal of Polymer Science 15(80), 371-390 (1955). 148 C. S. Tyberg, M. Sankarapandian, K. Bears, P. Shih, A. C. Loos, D. Dillard, J. E. McGrath, and J. S. Riffle, Tough, Void-Free, Flame Retardant Phenolic Matrix Materials, Construction and Building Materials 13, 343-353 (1999). 149 C. S. Tyberg, K. Bergeron, M. Sankarapandian, P. Shih, A. C. Loos, D. A. Dillard, J. E. McGrath, and J. S. Riffle, Structure-Property Relationships of Void-Free PhenolicEpoxy Matrix Materials, Polymer 41(13), 5053-5062 (2000). 110 OH OH OH OH OH OH OH OH OH OH OH OH Decreasing Network Density = Increasing Molecular Wt. Between Crosslinks (Mc) Increasing concentration of unreacted phenolic hydroxyl groups Figure 4. 1. Idealized phenolic/epoxy networks The critical stress intensity factors, KIC, of phenolic novolac/bisphenol-A epoxy networks were measured as a function of network composition (Table 4. 1).148 The fracture toughness increased as the phenolic hydroxyl to epoxy ratio was offset from 1/1 to 5/1 where the fracture toughness appeared to reach a maximum. component necessary to form well-connected networks. Further stoichiometric imbalance reduced the fracture toughness due to a lack of epoxy Table 4. 1. Fracture toughness of phenolic novolac/epoxy networks150 Phenol/Epoxy (eq/eq) 1/1 2/1 3/1 5/1 7/1 150 Novolac/Epoxy (wt/wt) 36/64 53/47 63/37 74/26 80/20 KIC (MPa*m1/2) 0.57 0.64 0.87 1.03 0.70 C. S. Tyberg, Void-Free Flame Retardant Phenolic Network: Properties and Processability, Dissertation, Virginia Tech, 2000. 111 The toughnesses of neat and rubber toughened diglycidyl ether based epoxy networks were measured as a function of crosslink density.151 Slow speed fracture toughness (GIC) test and notched high-speed tensile toughness (Gh) tests revealed that increased toughness occurred with increased Mx for both rubber toughened and unmodified networks. reacted with increased Mxs. well Studies on polyurethane networks derived from triisocyanate defined diols,152 and DGEBA cured with 4,4-diamino- 3,3dimethyldicyclohexylmethane (3DCM)153 also correlated increased toughness with 4.1.2. Cooperativity Polymers exhibit viscoelastic behaviors in which the combined characteristics of elastic solids and viscous fluids are shown. Two simple transient viscoelastic experimental methods are stress relaxation and creep tests. A stress relaxation test is performed under constant strain and the stress rises initially and decays with time due to dissipation by its fluid like component (Figure 4. 2.a). A creep test involves applying a constant stress to samples and observing the strain increases with time (Figure 4. 2.b). 151 M.C. M. van der Sanden and H. E. H. Meijer, Deformation and Toughness of Polymeric Systems: 3. Influence of Crosslink Density, Polymer 34(24), 5063-5072 (1993). 152 H. L. Bos and J. J. H. Nusselder, Toughness of Model Polymeric Networks in the E. Espuche, J. Galy, F. F. Gerard, J. P. Pascault, and H. Sautereau, Influence of Glassy State: Effect of Crosslink Density, Polymer 35(13), 2793-2799 (1994). 153 Crosslink Density and Chain Flexibility on Mechanical Properties of Model Epoxy Networks, Macromolecular Symposia 93, 107-115 (1995). 112 Figure 4. 2. a) Stress-relaxation experiment, and b) creep experiments When an external force is applied to polymer chains in their equilibrium state, a relaxation process occurs to redistribute the chain conformations until equilibrium is reached. The relaxation of the smallest units in polymer chain requires the shortest relaxation time. Polymer chains composed of many relaxation units can cooperatively relax as a larger unit, which requires a longer relaxation time. A polymer relaxation process, which consists of several mechanisms, can be described using a distribution of relaxation times, or a relaxation spectrum G(t) = G i exp ( i t ) (4. 6) where G(t), the relaxation modulus at time t, is the sum of individual relaxing elements each with a relaxation time i and a relative intensity Gi. In a continuum relaxation process, the smallest unit should exhibit the shortest relaxation time. A deformation study on polypropylene showed that the rotation of main chain bonds, most likely between trans and gauche conformations, represents the smallest domain and therefore is the predominate mode of deformation.154 The rate of rotation from one stable conformation to another is controlled by the number of bonds exceeding 154 D. N. Theodorou and U. W. Suter, Shape of Unperturbed Linear Polymer: Polypropylene, Macromolecules 18(6), 1206-1214 (1985). 113 the highest energy state encountered in the rotation process. The classical rate constant (k) can be used to estimate the kinetics of relaxation k =1 dN = k * exp N dt kT (4. 7) where N is the number of those conformers in the ground state, is the energy difference between the activated and ground state, k* is the efficiency term that depends on the intensity of the liberation of the bonds in question, k is the Boltzmann constant, and T is the absolute temperature. The relaxation of a small unit crowded in a dense polymer domain requires its neighbors to move together. This simultaneous relaxation of one unit along with its neighbors is termed cooperativity. The probability of relaxation becomes smaller in a denser environment, and therefore the cooperative domain size increases. single segment (Figure 4. 3). The cooperative domain size (z) therefore affects the activation energy for the relaxation of a Figure 4. 3. Illustration of cooperativity domain size where z = 7 The temperature dependence of the transition region and non-equilibrium glassy state of a polymer can be investigated by determining its intermolecular cooperativity. Polymers with low cooperativity follow Arrhenius behaviors when they are cooled below their Tg and exhibit narrow relaxation time distributions. For example, in the absence of 114 neighbor interference, the following Arrhenius equation can be used to describe the relaxation time, 1 1 ln = ln * + RT RT * where = activation energy of conformer rotation (or change) R = gas constant T = absolute temperature T* = temperature at which the conformer can rotate freely without any neighbor interference * = relaxation time at which the conformer can rotate freely without any neighbor interference For cooperative relaxation, the domain size (z) must be introduced. described using a non-Arrhenius equation 1 z ln = ln * + RT RT * The above equation is non-Arrhenius since z is temperature dependent. Materials with Arrhenius behaviors are described as strong; on the other hand, materials with non-Arrhenius behaviors are referred to as fragile. Fragility therefore is a measure of the temperature dependence of the most probable segmental relaxation time near Tg where more fragile corresponds to higher temperature dependence and vice versa. Connolly and Karasz155 have correlated fragility to the degree of cooperativity. A fragility plot, or cooperativity plot, can be generated by plotting the time temperature shift factor versus (Tg/T). The slope of the line at T = Tg is termed fragility (m) and is proportional to the apparent activation energies normalized by Tg. 155 (4. 8) The relaxation time of polymers that consist of larger sterically hindered units must be (4. 9) M. Connolly and F. Karasz, Viscoelastic and Dielectric Relaxation Behavior of Substituted Poly(p-Phenylenes) , Macromolecules 28, 1872-1881 (1995). 115 m = E/2.303 R Tg (4. 10) A higher m value corresponds to a more fragile material with a greater degree of intermolecular cooperativity which therefore exhibits a higher temperature sensitivity. Hydrogen bonding plays an important role in cooperativity for both low molecular weight materials and polymers. The cooperativity of a series of disubstituted benzenes with various potentials for hydrogen bonding, investigated by Angell,156 showed that materials that can form stronger hydrogen bonding were more fragile. Bohmer et al.157 and Roland and Ngai158 observed the same trends. Phenolic novolac/bisphenol-A epoxy networks, showed that both the crosslink density and hydrogen bonding affected the glass formation processes.150 On the contrary, relatively non-polar polymers such as polyethylene and polytetrahydrofuran showed low fragility values which corresponded to weak intermolecular interactions. 4.1.3. Thermal and thermo-oxidative stability of novolac/epoxy networks Phenolic networks are well known for their excellent thermal and thermooxidative stabilities. A detailed review on the decomposition pathways can be found in chapter 1.8. The present work focuses on characterizing novolac/epoxy networks using cone calorimetry. The cone calorimetry, so called because of the geometric arrangement of the electric heater, measures the heat release rate, total heat released, ignitability, effective heat of combustion, specific extinction area (a measure of the smoke production), soot yield, mass loss rate, and the evolution of carbon monoxide, carbon dioxide, and other 156 C. A. Angell, Transport Processes, Relaxation, and Glass Formation in Hydrogen- Bonded Liquids, Hydrogen Bonded liquids, J. C. Dore and J. Teixeira, Eds., Vol. 329, pp59-79, 1991. 157 R. Bohmer, K. L. Ngai, C. A. Angel, and D. J. Plazek, Nonexponential Relaxations in Strong and Fragile Glass Formers, Journal of Chemistry and Physics 99(5), 42014209 (1993). 158 C. M. Roland and K. L. Ngai, Normalization of the Temperature Dependence of Segmental Relaxation Times, Macromolecules 25(21), 5765-5768 (1992). 116 combustion products. The results obtained using cone calorimetry, specifically ignitability, surface flame spread rate and available surface area, and heat release rate per unit area can be used to predict the behaviors of materials in a full-scale fire. Ignitability and flame spread effects have been shown to be insignificant for most materials. Heat release rate therefore is generally considered the most important material parameter in fire studies and is calculated based on the oxygen consumption principle. The principle uses the basic assumption that the heat of combustion per unit mass of oxygen consumed is essentially constant and has an average value of 13.1 KJ/kg for most organic materials. The heat rate, q, is related to the difference of the initial inflow and the total mass flow of oxygen in the combustion products,159 h q= c r o where m o , m o 2 2 ( ) (4. 11) q = heat rate (kW) hc = net heat of combustion (kJ/kg) ro = stoichiometric oxygen to fuel mass ratio mO2, = initial mass flow of oxygen (kg/s) mO2 = total mass flow of oxygen in combustion products (kg/s) A primary advantage of this principle is that the instrument does not need to be thermally insulated, allowing for design of an accurate yet simple instrument (Figure 4. 4). Samples can be tested in a horizontal or vertical orientation and can be irradiated with a heat flux from zero to over 100 kW/m2 with good uniformity. The only limitations are that the sample should not swell sufficiently to interfere with the spark plug operation or with the heater, and the specimens should not show explosive spalling or delamination.159 Samples that melt and flow upon heating must be tested in the horizontal orientation. The mass loss rate is measured using a load cell; the specific extinction area is 159 V. Babrauskas, Development of the Cone Calorimeter A Bench-Scale Heat Release Rate Apparatus Based on Oxygen Consumption, Fire and Materials 8(2), 81-95 (1984). 117 determined using a helium-neon laser; and the CO and CO2 yield during combustion is obtained using a CO/CO2 analyzer. Temperature and differential pressure measurements taken Laser extinction beam including temperature measurement Soot Sample Tube Location Exhaust Blower Gas Samples Taken Here Controlled Flow Rate Exhaust Cone Heater Spark Igniter Sample Soot Collection Load Cell Vertical Orientation Figure 4. 4. Schematic of a cone calorimeter Most materials exhibit a single peak heat release rate (PHRR) or maximum in the heat release rate curves. However, in some char forming materials, there may be an initial peak followed by a gradual increase to a second peak. The heat release rate drop after the first peak is attributed to the formation of an isolating char layer and the second peak is suggested to occur when the specimen was heated through to the rear face and no longer behaves as if thermally thick.159 An alternative explanation for the observed initial spike is the rapid burning of volatiles that build up prior to ignition. It is possible that both char layer and rapid burning of volatiles contribute to the initial spike. However, to date, there is no concrete evidence for the char layer while intense burning masks the burning of volatiles in non-char yielding materials. The flame retardance of phenolic novolac/epoxy networks cured with various diepoxides has been measured (Table 4. 2).148,149 The heat release rate curves for all 118 networks investigated exhibited an initial spike in addition to the peak heat release rate. The brominated bisphenol-A epoxy cured networks had the lowest peak heat release rates, but produced large amounts of toxic CO gas (high CO/CO2 ratio). In addition, bromine incorporation released HBr and other carcinogenic by-products during burning. Phenolic novolac cured with siloxane based epoxy yielded networks with low peak heat release rates and the lowest toxicity. Surprisingly, curing networks with bisphenol-F epoxy, fluorinated epoxy, or stilbene epoxy did not alter the peak heat release rates in comparison with the bisphenol-A epoxy cured networks. networks. Networks cured with bisphenol-F epoxy gave rise to higher smoke toxicity compared to the bisphenol-A cured Table 4. 2. Cone calorimetry results on phenolic novolac/epoxy networks (65:35 wt:wt ratio) Epoxy Phenolic Control (Resole) Epoxy Control Bisphenol-A Brominated bisphenol-A Bisphenol-F Disiloxane Fluorinated Stilbene PHRR (kW/m2) 116 1230 357 165 397 325 353 407 Char Yield (wt %) 65 5 29 8 17 22 --- Smoke Toxicity (CO/CO2) (x10-3) 11 44 34 189 76 24 --- As expected, compositions containing larger amounts of phenolic component showed lower peak heat release rates, higher char yield, and lower smoke toxicity. The same trend in char yield was observed using thermogravimetric analysis.148,149 Cone calorimetry was used to evaluate the flame properties of brominated epoxy composites and phenolic composites reinforced with fiberglass, aramid, or graphite fiber under controlled oxygen atmospheres.160 The time to ignition (TTI) was essentially 160 F. Hshieh and H. D. Beeson, Flammability Testing of Flame-Retarded Epoxy Composites and Phenolic Composites, Fire and Materials 21, 41-49 (1997). 119 identical for all of the epoxy composites under normal oxygen atmosphere. The epoxy/graphite composite showed higher TTI under oxygen-depleted environments, probably due to the higher thermal conductivity of the graphite fiber. The TTI of phenolic composites were proportional to their thermal conductivities at all oxygen levels. Using the TTI to PHRR ratio as an indication of the propensity to flashover, increased oxygen concentration reduced the flame resistance for both epoxy composites and phenolic composites. As expected, phenolic composites showed more resistance to flame and had much lower smoke productions than epoxy composites. 4.2. Experimental 4.2.1. Materials The 2000 g/mol cresol novolac resin was prepared according to procedures described in chapter 3. Triphenylphosphine (TPP) was purchased from Aldrich. The commercial phenolic novolac resin used in the control experiments was provided by Georgia Pacific (Product #GP-2073). Epon 828 epoxy resin was obtained from Shell Chemical. D.E.N. 438 epoxy was supplied by the Dow Chemicals Co. All reagents were used as received. 4.2.2. Methods 4.2.2.1. Preparation of ortho-cresol novolac networks cured with epoxies A 2000 g/mol ortho-cresol novolac resin was cured with a difunctional or a multifunctional epoxy using triphenylphosphine as the catalyst. To a three neck round bottom flask equipped with a vacuum tight mechanical stirrer and a vacuum adapter was added cresol novolac and ~85 weight % of the required epoxy. The flask was heated in an oil bath to 170C. When the novolac began to soften at ~170C, mechanical stirring was begun. Vacuum was applied incrementally to prevent the material from foaming into the vacuum line. Once full vacuum was achieved (2-5 Torr) the solution was stirred for about 10 minutes to degas the blend. During this time the remaining epoxy, with the 120 catalyst (0.1 mol % based on the total weight of epoxy) dissolved in it, was degassed in a vacuum oven at ~80C. The vacuum was temporarily released to add the remaining epoxy with catalyst. This was stirred for about 3 minutes to fully degas the samples. The melt was then poured into a mold and placed in a preheated oven. The samples were cured at 200C for 2 hours and then 220C for 2 hours. 4.2.2.2. Sample preparation for viscosity determinations Ortho-cresol novolac/epoxy mixtures were melt mixed at 165C. The exposure time to heat was maintained for less than 3 minutes to prevent premature curing. The mixed samples were quenched in dry ice/isopropanol chilled aluminum pans. samples were ground into powder prior to use. 4.2.2.3. Network formation of phenolic control A resole resin was cured thermally to form a typical phenolic network (phenolic control). The cure cycle consisted of 70C for 4 days, 130C for 24 hours, and then 200C for 24 hours. The 4.2.3. Characterization 4.2.3.1. Resin glass transition temperatures The glass transition temperatures of neat resins were obtained via a Perkin-Elmer differential scanning calorimetry (DSC-7 Instrument). The DSC was calibrated with indium and zinc standards, and ice water was used as the coolant. Samples in aluminum pans were heated from 25 to 180C at 10C/min. The glass transition temperatures were calculated as the midpoints of the curves obtained from the second temperature scan. 4.2.3.2. Network glass transition temperatures A Perkin-Elmer dynamic mechanical analyzer (model DMA-7), in a three-point bending mode, was used to determine the glass transition temperatures of cured networks. The Tgs were calculated from the peaks of the tan delta curves. The static force was set to 200 mN and the dynamic force was set to 175 mN. Samples were heated at 5C/min 121 from 25 to 200C. Two samples of each material were measured and the results were averaged. 4.2.3.3. Critical stress intensity factor, KIC The critical stress intensity factor, K1C, was used to evaluate the fracture toughness of the phenolic/epoxy networks. The K1C values were obtained from a threepoint bend test using an Instron instrument, according to ASTM standard D5045-91.161 The specimens had a thickness (b; ~3.1mm) and a width (w; ~6.3mm). The single edge notched bending method was used. An eccentric compressive load was utilized to aid the pre-cracking of the specimens.162 First, a sharp notch was created in the sample by sawing. The sample was placed in a vise where it was subjected to tension and compression (Figure 4. 5), a cold razor blade (which had been immersed in liquid nitrogen) was inserted into the notch and force was applied to initiate a natural crack. The depth of the crack (a) was between 40 and 60 percent of the width (w). The precracked notched specimen was loaded crack down into a three-point bend fixture and tested using an Instron model 4204 instrument. The single edge notched bending rig had moving rollers to avoid excessive plastic indentation. The three-point bend fixture was set up so that the line of action of applied load passed midway between the support roll centers within 1% of the distance between these centers. The crosshead speed was 1.27 mm/minute, and the testing was conducted at room temperature. Figure 4. 5. 161 Experimental implementation of the eccentric axial load technique ASTM D 5045-91 Standard test methods for plane-strain fracture toughness and D. A. Dillard, P. R. McDaniels, and J. A. Hinkley, The Use of an Eccentric strain energy release rate of plastic materials, 1991. 162 Compressive Load to Aid in Precracking Single Edge Notch Bend Specimens, Journal of Materials Science letters 12, 1258-1260 (1993). 122 The critical stress intensity factor, KIC, was calculated using the following equation. K IC = PS B W 3/2 3 (x)1/2 [1.99 - x (1 - x) (2.15 - 3.93x + 2.7x 2 )] 2 (1 + 2x) (1 - x) 3/2 (4. 12) where P = maximum load (kN) S = span (cm) B = specimen thickness (cm) W = specimen depth or width (cm) a = crack length x = the ratio of crack length to width of the specimen, (a/W) 4.2.3.4. Sol/gel fraction separation The sol/gel fractions of the networks were determined as follows. Samples of approximately 0.6g were submerged at room temperature in ~15 ml acetone for 3 days. The sol fractions were soluble and the gel fractions stayed intact. The gel fractions were separated from the sol fractions by filtering the solids from the solutions. The gel fractions were dried in a vacuum oven at 180C for 24 hours and compared to their initial weights. The acetone in the sol fractions was evaporated at room temperature in a vacuum oven. The sol fractions were also analyzed by 1H NMR in order to determine their chemical contents. 4.2.3.5. 1H NMR sol fraction characterization 1 H NMR spectra were obtained on a Varian Unity 400 NMR spectrometer. 5 mm tubes containing approximately 20 mg sample dissolved in DMSO-d6 or acetone-d6 were analyzed under ambient conditions. The experiments parameters include 1.000 second relaxation delay, 23.6 degrees pulse, and 6744.9 Hz widths. Thirty-two repetitions were performed for each sample. 123 4.2.3.6. Room temperature density measurements Room temperature density measurements were conducted using a Mettler-Toledo AG204 balance adapted with a Mettler-Toledo density determination kit for AT/AG and PG/PR balance. Samples with dimensions of approximately 19 mm x 6.4 mm x 3.2 mm were sanded, then polished, to prevent any trapping of air bubbles. Deionized water was degassed in a vacuum oven at room temperature for 30 min prior to use. Using this setup, the weight of the solid in air (A) and the weight of the solid in water (B) were measured. The temperature of the water was recorded to within 0.1C and the density of distilled water at that temperature ((H2O)) was obtained from a density table. Room temperature densities were calculated using the following equation = [A/(A-B)] (H2O) (4. 13) The densities at Tg+50C were calculated using the densities at room temperature and the coefficients of thermal expansion (CTE) below and above the Tg. 4.2.3.7. Determination of coefficient of thermal expansion () One dimension thermal expansion was measured using a thermomechanical analyzer equipped with a large quartz parallel plate. The specimen height was monitored as the sample was heated from 25C to 200C at 5C/min. The coefficients of thermal expansion (CTE) were determined by calculating the slope above and below the Tg of the material. 4.2.4.8. Rubbery moduli determination via creep tests The rubbery moduli (E) of networks at 50C above the glass transition temperatures were determined using a stress-strain test via a Dynastat instrument. Samples with dimensions of approximately 19mm x 6.3mm x 3.1mm were placed in a three-point bend fixture and heated to 50C above the Tg. A small load (0.01kg) was placed on the sample and the displacements at equilibrium were measured. Equilibrium was reached within 5 seconds. The load was increased by increments of 0.01 kg up to about 0.08 kg while the equilibrium displacement was recorded at each load. The 124 rubbery moduli were determined from the slopes of load versus displacement curves. The displacement values measured from the Dynastat are accurate to within 0.05 mm. 3 P L E =g 48 I (4. 14) where g = gravitational constant = 9.81 m/s2 P/ = slope of load versus displacement L = length between supports = 2.54 cm and I = (1/12)wb3 where w = width of the sample b = height of the sample According to the theory of rubber elasticity, the number average molecular weight between crosslinks (Mx) is inversely proportional to the rubbery modulus, Mx = 3RT/E where R = gas constant, T = experimental temperature (K) = density of the sample at the experimental temperature; calculated from the room temperature density and CTE values of the samples, E = elastic modulus obtained from the Dynastat measurements. 4.2.3.9. 10sec relaxation moduli determination via stress relaxation tests Stress relaxation experiments were also conducted on the Dynastat instrument in the three-point bend mode. The samples were initially heated to 70C below the Tg (Tg determined from the peak of the DMA tan delta curve). The upper beam was set to a position just above the sample with no contact or applied forces, and the instrument was equilibrated for at least 3 minutes. To begin testing, the upper beam was lowered by 0.1 (4. 16) (4. 15) 125 mm. The load was measured and recorded as a function of time as the sample relaxed for 1000 seconds. The displacement was then removed and the samples were allowed to recover for 2000 seconds. This test was repeated at each temperature interval from Tg70C to Tg+30C at 5C increments. In addition to the applied displacement, the displacement increased with temperature increases due to thermal expansion by the instrument and the sample. The CTE of instrument was approximately 0.0033 mm/C and the CTE of the samples was determined using TMA. The sample expansion can be ignored since it is insignificant compared to the instrument expansion and applied displacement. The modulus at each temperature for each displacement was calculated using equation (4.14). The modulus calculated after 10 seconds relaxation was plotted as a function of temperature. Master curves were constructed from these stress relaxation results according to the time-temperature superposition principle. After a reference spectrum was assigned (generally at the Tg of the network), other spectra, determined at various temperatures, were shifted horizontally until a single continuous curve was generated. 4.2.3.10. Flame retardance measured via cone calorimeter Cone calorimetry was used to measure the flame retardance of novolac/epoxy networks. Sample panels of approximately 6.3 mm thickness and surface dimensions of 10x10 cm were tested at a 50.0 kW/m2 incident heat flux in a controlled atmosphere cone calorimeter. The ignitability, peak heat release rate, the evolution of CO and CO2, and the time to sustained ignition are among some of the parameters measured. 4.2.3.11. Thermal and thermo-oxidative degradation A Perkin-Elmer TGA-7 thermogravimetric analyzer was used to determine the thermal and thermo-oxidative stabilities of cresol novolac/epoxy networks. Samples of approximately 5-8 mg were placed in a platinum sample pan and heated in a furnace at 10C/min from 30 to 900C. Air or nitrogen was used as the carrier gas. The sample weight loss was monitored as a function of temperature. 126 4.2.3.12. Viscosity measurements Dynamic complex viscosities were obtained from a Bohlin VOC Rheometer operating in continuous oscillation mode with a frequency of 1 Hz. Temperature control was accomplished with a Bohlin HTC. The auto-strain was set to maintain the torque at 25% of the maximum torque allowed. The maximum strain for the instrument was 0.25. Approximately 0.7g of dry powder sample were pressed into pellets, then placed between the preheated 25 mm diameter parallel plates of the rheometer. The gap was closed to approximately 1mm and the sides were scraped to remove excess sample before the run was started. A Brookfield DV-III Programmable Rheometer was used to determine the isothermal viscosities of novolac/epoxy mixtures. Approximately 10g of powder samples were placed in disposable sample tubes and heated to the test temperatures. A spindle, which was driven through a calibrated spring, was immersed in the test fluid. The viscous drag of the fluid against the spindle was measured by the spring deflection. The torque was used to calculate the viscosities. (cp) = (100/rpm) x Tk x SMC x torque where rpm = 10 Tk = a constant, equal to 1.0 in these experiments SMC = a constant which depends on the spindle diameter = 25 4.2.3.13. Equilibrium moisture uptake To investigate moisture uptake, samples with dimensions of approximately 19 mm x 6.35mm x 1 mm were dried in a vacuum oven for 12 hours at 150C. The initial weight of each sample was recorded. The samples were then placed in vials containing deionized water. The weights of each sample were measured as a function of time until a constant weight was reached. Water absorption at room temperature and at 62C was determined. t1/2/thickness. The sample thickness, the dimension in which water penetrates most rapidly, was normalized as the results were plotted as percent water uptake versus (4. 17) 127 4.2.3.14. Kinetic studies via DSC Arrhenius activation energies, pre-exponential factors, and first order rate constants for novolac/epoxy cure reactions were determined using differential scanning calorimetry according to ASTM E 69879.163 Samples were heated at various heating rates and the exothermic reaction peaks were recorded. The temperatures at which the exotherm peaks occurred were plotted as a function of their respective heating rates. Kinetic parameters were calculated from the slope of this plot. The time that was required to reach 50% conversion, or the half-life time, can be predicted. A sample was aged to reach 50% conversion according to the predicted time at the selected temperature and compared to an unaged sample to confirm the validity of this method. 4.2.3.15. Flexural strength and moduli of composites Transverse flexural tests were performed according to ASTM standard D 79098. Composite samples with dimensions of 127 mm x 12.7 mm x 2.4 mm were measured in a three-point bend set up. The rate of crosshead motion (R) used was 7.11 mm/min which was calculated based on the following equation, R = ZL2/6d where L = support span length (101.6 mm), d = depth of the beam (2.4 mm), Z = rate of straining of the outer fiber (0.01). Flexural strength (f) and modulus (E) were calculated as follows, f = 3PL/2bd2 E = L3m/4bd3 where P = load b = width of beam tested (12.7 mm) 163 (4. 18) (4. 19) (4. 20) ASTM E 698-79 (reapproved 1993) Standard test method for Arrhenius kinetic constant for thermally unstable materials, 1993. 128 L = support span length (101.6 mm), d = depth of the beam (2.4 mm), m = slope of the tangent to the initial straight-line portion of the load-deflection curve. 4.3. Results and Discussion 4.3.1. Properties of ortho-cresol novolac/epoxy networks 4.3.1.1. Network formation and characterization Network structures in crosslinked systems generally have an enormous influence over the mechanical properties. For example, high crosslink densities generally correlate with high strengths but low fracture toughnesses. Typical phenolic novolac networks crosslinked with hexamethylenetetramine (HMTA) are brittle because they are highly crosslinked and filled with voids. Tough, void-free, networks have been prepared by curing phenolic novolacs with epoxies. In addition, if the phenolic novolac was used in excess, the resulting networks exhibited excellent flame retardance properties.148 The fracture toughness of crosslinked networks is related to the molecular weight between crosslinks. Increased molecular weight between crosslinks (Mc), or decreased crosslink densities, generally gave rise to tougher networks up to some point where dangling chain ends begin to dominate properties. A primary objective of this research was to improve the fracture toughness of novolac/epoxy networks while retaining excellent flame retardance. The approach utilizes the known direct correlation between molecular weight between crosslinks and fracture toughness by using higher molecular weight oligomers in the network formation. However, increasing the molecular weight of the phenolic novolac is impractical since branching occurs once the molecular weight reaches 900-1000 g/mol.164 Branching may lead to higher viscosities and reduce the processability. Branching sites on oligomers 164 Knopp, A.; Pilato, L. A., Phenolic Resins: Chemistry, Application and Performance- Future Directions; Springer-Verlag, New York, 1985. 129 translate to crosslink points in the cured networks, which effectively increase the crosslink densities. Cresol novolac resins, on the other hand, are linear. Moreover, the molecular weight was strategically controlled with an endcapping reagent. Table 4. 3. Phenolic materials and their properties Sample Phenolic novolac Ortho-cresol novolac Mn (g/mol) ~ 700 ~ 2000 Tg (C) 78 104 In this study, a controlled molecular weight 2000g/mol ortho-cresol novolac resin was crosslinked with two epoxies, a difunctional bisphenol-A epoxy (Epon 828) and a multifunctional epoxidized novolac (D.E.N. 438) in systematically controlled weight ratios (Figure 4. 6). The used of Epon 828 epoxy was expected to offer better processabilities since it significantly reduced the melt viscosities of novolac/epoxy mixtures. Networks crosslinked with DEN 438 were anticipated to be more flame retardant. The properties of these cresol novolac networks were evaluated and compared to an epoxy control (Epon 828 crosslinked with a stoichiometrical amount of p,pdiaminodiphenylsulfone [4,4-DDS]), a phenolic control (heat cured resole network), and a phenolic novolac/Epon 828 network (65:35 wt:wt ratio). 130 H3C HO H3C CH2 OH CH3 CH3 OH m CH2 + epoxy CH3 Triphenylphosphine Crosslinked Network O O O O epoxy = Epon 828; f=2, EEW=187g/mol O O CH2 3.5 D.E.N. 438; f=3.5, EEW~200 g/mol Figure 4. 6. Crosslinking reaction of ortho-cresol novolac and epoxy (Epon 828 or D.E.N. 438) using triphenylphosphine as the catalyst The network properties including fracture toughness, Tg, and weight percentage of sol fractions were determined for Epon 828 and D.E.N. 438 epoxy cured cresol novolac networks (Table 4. 4). Network fracture toughness was determined by measuring the plane-strain stress intensity factor, KIC. Higher KIC values correspond to improved resistance to crack propagation or increased toughness. For both series, the 60:40 wt:wt cresol novolac/epoxy networks exhibited the highest fracture toughness and this remained relatively high as the novolac composition was increased to 70 wt. %. The toughness decreased drastically when the novolac contents were increased to 80 wt. %. The 70:30 and 60:40 wt:wt cresol novolac/epoxy networks had higher toughness values than the phenolic control, the 65:35 wt:wt phenolic novolac/Epon 828 network, and the epoxy 131 control. The 70:30 compositions had the highest Tgs and lowest sol fractions. The Tg and the sol fraction increased when the novolac composition was increased or decreased. Table 4. 4. Network properties of ortho-cresol novolac/epoxy networks Epoxy Epoxy Control Phenolic control Epon 828* Epon 828** Novolac Wt % --63 80 70 60 80 70 60 OH:epoxy --3.0:1 6.2:1 3.6:1 2.3:1 5.0:1 2.9:1 1.9:1 K1C (MPa/m1/2) 0.62 0.16 0.85 0.650.06 1.060.04 1.200.12 0.460.04 1.050.09 1.080.10 Tg ( C) 127 o Sol fraction (%) --12.5 16.7 5.4 7.1 15.6 4.3 11.7 -137 144 154 133 145 152 142 D.E.N. 438** * Phenolic novolac/epoxy networks ** Ortho-cresol novolac/epoxy networks To understand these unexpected network properties, network structures were investigated in an attempt to relate the structures with properties. A simple approach that determined the chemical compositions of the sol fractions is 1H NMR (Figure 4. 7). Since epoxides show five distinct sets of peaks, the presence of epoxy functionality can be detected with ease. 1H NMR of sol fractions showed traces of epoxy for the network compositions higher in novolac (80:20 and 70:30 wt:wt cresol novolac/epoxy networks). The sol fractions of the 60:40 wt:wt cresol novolac/epoxy network, on the other hand, consisted of large amounts of unreacted epoxies, even though these had a stoichiometric excess of phenol. This suggests that the networks at 60:40 compositions became too dense for further phenolic hydroxyl/epoxy reactions to occur since post-curing well above Tg did not promote any further reactions. This may also explain why the fracture toughness reached a certain maximum limit above which an increase in epoxy content did not increase the toughness. 132 Epon 80:20 sol 70:30 sol 60:40 sol 4.5 4.0 3.5 3.0 2.5 2.0 1.5 PPM Figure 4. 7. 1 H NMR of the sol fraction of cresol novolac/Epon 828 networks The effect of network structures on Tg was also explored (Table 4. 4). For both series, the Tg was highest for the 70:30 wt:wt cresol novolac/epoxy networks. This is consistent with the low sol fractions in these compositions suggesting that the chains in these networks are well connected. By contrast, the 60:40 wt:wt cresol novolac/epoxy networks which contained a significant amount of unreacted epoxy groups had lower Tgs. The 80:20 composition also exhibited lower Tgs due to their low crosslink densities and high sol fractions. Tg is a function of crosslink density and the amount of sol fraction/dangling ends in the networks. To further understand the network structures, the molecular weights between crosslinks were calculated (Table 4. 5). The theory of rubbery elasticity was used to 133 estimate the molecular weight between crosslinks for the cresol novolac/epoxy networks. According to this theory, network rubbery moduli are inversely proportional to the molecular weights between crosslinks. The rubbery moduli were measured using a stress-strain test in a three-point bend set up at Tg+50C, a temperature well above the material glass transition temperature. The network glass transition temperatures and densities at the measurement temperature (Tg+50C) were determined prior to calculation. It should be noted that the rubbery moduli should be measured for networks containing only the gel fraction. However, the rubbery moduli were measured on unextracted networks in these experiments. Table 4. 5. Crosslink densities of cresol novolac/epoxy networks Epoxy Epon828* Epon 828** Novolac Wt % 65 80 70 60 80 70 60 Tg (C) 127 144 154 133 145 152 142 (25C) 1.23 1.178 1.183 1.181 1.199 1.191 1.191 (Tg+50C) 1.18 1.163 1.168 1.168 1.184 1.173 1.174 Modulus Tg+50C 9.35 x 106 1.82 x 10 8.25 x 106 3.20 x 106 2.63 x 106 9.12 x 106 8.02 x 106 6 Mx (g/mol) 1413 6650 1510 3730 4700 1360 1510 D.E.N. 438** * Phenolic novolac/epoxy networks ** o-Cresol novolac/epoxy networks For cresol novolac/Epon 828 networks, the 70:30 composition exhibited the highest rubbery modulus and therefore the lowest apparent Mx. The calculated Mx (1510 g/mol) is a reasonable value for well connected networks prepared with a 2000 g/mol oligomer. The rubbery modulus decreased and the Mx increased significantly at the 60:40 composition. According to the 1H NMR analysis on sol fractions, a large amount of epoxy remained unreacted at this composition. The low molecular weight fraction, along with dangling chain ends, probably did not contribute mechanically, and therefore reduced the rubber modulus and increased the apparent molecular weight between crosslinks. When the cresol novolac content was increased to 80 weight percent, the 134 rubbery modulus again decreased. The high molecular weight between crosslinks was attributed to a looser network formed with less epoxy. The same trends were observed for the D.E.N. 438 epoxy cured cresol novolac networks. Surprisingly, these results indicated that the networks comprised of 80:20 wt:wt cresol novolac/epoxy (or 6.2 hydroxyl per 1 epoxy) did not form sufficiently well connected networks. The stoichiometric ratio was optimized at 70:30 wt:wt cresol novolac/epoxy compositions (or 3.6 hydroxyl per 1 epoxy) where all the epoxies were reacted into the networks and the apparent Mx was most reasonable. At this composition, networks with superior properties, i.e. high fracture toughness, high Tgs, low sol fractions, and sufficiently high crosslink densities were generated. At the 60:40 composition (2.3 hydroxyl per 1 epoxy), the networks became too dense. The lack of molecular mobility probably prevented further phenolic hydroxyl/epoxy reactions, thereby leaving unreacted epoxies in the sol fractions. Both the glassy and the rubbery moduli were evaluated from a plot of 10-second stress relaxation moduli verses temperature. As described earlier, these moduli were evaluated for unextreacted cresol novolac/epoxy networks. As expected, the rubbery moduli for cresol novolac/Epon 828 networks (Figure 4. 8), phenolic novolac/Epon 828 networks (Figure 4. 9), and cresol novolac/DEN 438 networks (Figure 4. 10) increased as the network crosslink density increased. For most materials, the same trends are observed for glassy moduli where higher crosslink densities lead to slightly higher glassy moduli. However, for Epon 828 cured cresol novolac and phenolic novolac networks, the glassy moduli for the 80:20 composition was significantly higher than for the 70:30 and 60:40 compositions. This increase in the glassy modulus for the 80:20 composition was most likely due to its ability to form strong hydrogen bonding between unreacted phenolic hydroxyl groups. Since the sol fractions were not removed prior to measurement, an antiplasticization effect could also led to the increased glassy moduli. The hydrogen bonding interactions did not play an as important role in the 70:30 and the 60:40 compositions since more hydroxyl groups were reacted with epoxies at these compositions. 135 10 80:20 Epon 828 70:30 Epon 828 60:40 Epon 828 log E' (Pa) 9 8 7 40 Figure 4. 8. 70 100 130 Temperature (oC) 160 190 10s Relaxation moduli as functions of temperatures for cresol novolac/Epon 828 networks 10 9.5 9 80/20 8.5 8 7.5 7 45 65 85 105 125 145 165 185 65/35 50/50 Temperature (C) Figure 4. 9. 10s Relaxation moduli as functions of temperatures for phenolic novolac/Epon 828 networks The hydrogen bonding interaction, which affects the glassy moduli, was not observed for the D.E.N. 438 cured cresol novolac networks. The glassy moduli for all compositions were comparable (Figure 4. 10). Both the rubbery and the glassy moduli follow the expected trends. 136 10 9 80:20 DEN 438 70:30 DEN 438 60:40 DEN 438 log E (Pa) 8 7 70 100 130 160 o Temperature ( C) 190 Figure 4. 10. 10s Stress relaxation moduli as functions of temperatures for cresol novolac crosslinked with D.E.N. 438 epoxy 4.3.1.2. Master curves and cooperativity Network crosslink densities and chemical compositions play important roles in the glass formation processes. The behaviors of cresol novolac/epoxy networks during cooling toward the glass transition region were investigated by the cooperativity theory. The temperature dependence of stress relaxation moduli over a 1000-second period were determined for cresol novolac/epoxy networks near the glass transition regions (Tg-70 to Tg+30C) (Figure 4. 11). The relaxation spectrum measured near the glass transition temperature was assigned as the reference. Using the time-temperature superposition principle, horizontal superposition was performed on the log time scale until a single continuous master curve was generated. The shifting involved multiplication of the original time by a temperature shift factor (aT). The stress relaxation behaviors over a wide range of time or frequency are depicted in these master curves. 137 9 log E (Pa) reference 8 7 0.7 1.2 1.7 2.2 2.7 3.2 log Time (s) 9 log E (Pa) reference 8 7 -8 -4 0 log Time (s) Figure 4. 11. Master curve constructions for a typical cresol novolac/epoxy network: a) stress relaxation moduli of a cresol novolac/epoxy network measured from Tg-60C to Tg+40C at 5C intervals, and b) the master curve 4 8 138 The log of the shift factors were plotted against T-Tg (Figure 4. 12). Both the master curve and the shift factor plot must be continuous and show a reasonable shape for this approach to be valid. 10 log aT 5 0 -5 -50 Figure 4. 12. -25 0 T-Tg 25 50 The shift factor plot The nature of segmental motions from glass to rubber transition was investigated using cooperativity plots (or fragility plots), which are generated by plotting log aT versus Tg/T (Figure 4. 13 and Figure 4. 14). Data were fitted to a 3rd degree polynomial curve. The slope of the curve where Tg/T equals 1 is described as its fragility (m). As a material is heated from its glassy state through the Tg into its rubbery region, the chains begin to relax. Due to crowding, the relaxation of a single chain requires its neighboring chains to relax simultaneously. The larger the volume of neighboring chains, the more cooperative the material becomes. The fragility (m) therefore depends on the local friction coefficient, inter- and intra-molecular hydrogen bonding, and when applicable, the stiffness of the chain, the morphology, and low molecular weight additives. 139 4 3 2 log aT 1 0 -1 -2 -3 0.96 80:20 Epon 828 70:30 Epon 828 60:40 Epon 828 Poly. (80:20 Epon 828) Poly. (70:30 Epon 828) Poly. (60:40 Epon 828) 0.98 1 Tg/T 1.02 1.04 Figure 4. 13. Cooperativity plots of cresol novolac/Epon 828 networks 4 3 2 80:20 DEN 438 70:30 DEN 438 60:40 DEN 438 Poly. (80:20 DEN 438) Poly. (70:30 DEN 438) Poly. (60:40 DEN 438) log aT 1 0 -1 -2 -3 0.96 0.98 Tg/T 1 1.02 1.04 Figure 4. 14. Cooperativity plots of cresol novolac/D.E.N. 438 networks 140 For cresol novolac/Epon 828 networks, the fragility was highest for 70:30 composition due to its high crosslink density (Table 4. 6). A higher fragility value was observed for the 80:20 composition although it had a lower crosslink density than the 60:40 composition. This was attributed to the presence of large amounts of free phenolic hydroxyl groups which formed strong inter- and intramolecular hydrogen bonding. The 60:40 composition exhibited the lowest fragility value due to its relatively low crosslink density and low propensity for hydrogen bonding. Cresol novolac/Epon 828 networks (70:30 composition) exhibited a significantly higher fragility value than the phenolic novolac/Epon 828 networks with similar phenolic compositions (65:35 composition). Since a pendent methyl substituent was present on each cresol novolac repeat unit, it was anticipated that these bulkier groups reduced the free volume and, therefore, increased the fragility. Table 4. 6. Fragility measuring the crosslink densities and degree of hydrogen boning interaction for cresol novolac/epoxy networks Epoxy Epon 828* Epon 828** Novolac/Epoxy (wt:wt) 65:35 80:20 70:30 60:40 80:20 70:30 60:40 Mx (g/mol) 1410 7260 1640 4110 5130 1480 1650 Fragility (m) 67 75 84 64 71 74 77 DEN 438** * Phenolic novolac/epoxy networks ** Ortho-cresol novolac/epoxy networks Fragility results indicated that inter- and intramolecular hydrogen bonding was not an important factor in cresol novolac/ D.E.N. 438 networks (Table 4. 6). The 80:20 composition had the lowest crosslink density and therefore the lowest fragility. The 70:30 and the 60:40 compositions had slightly higher fragility values due to higher crosslink densities. 141 4.3.1.3. Thermal and thermo-oxidative stability The thermal and thermo-oxidative stabilities of cresol novolac networks were investigated using a thermogravimetric analyzer; the weight loss profiles were recorded as functions of temperature. Samples analyzed in air measured thermo-oxidative stability (Figure 4. 15a). All cresol novolac/Epon 828 epoxy networks were stable up to 400C, above which the materials began to degrade. A two-step degradation was observed for all samples and the weight loss profiles were similar for all compositions. The 80:20 composition showed a slightly higher thermo-oxidative stability than the 70:30 and the 60:40 compositions. The networks completely degraded at approximately 630C in air as the weight loss approached 100 percent. The thermal stabilities were investigated by examining the same samples under nitrogen (Figure 4. 15b). The weight loss began at a lower temperature in nitrogen than in air. The weight loss occurred mainly between 400C and 500C where the weight dropped from greater than 90% to ~ 35%. The 80:20 composition again showed only slightly higher heat resistance and had a slightly higher char yield at 900C. Cresol novolac/D.E.N. 438 networks showed almost identical results as the cresol novolac/Epon 828 networks. Approximately 20 to 30 weight percent char remained for all cresol novolac networks examined in nitrogen. 142 A) Air 70:30 60:40 80:20 Temperature (C) B) Nitrogen 80:20 70:30 60:40 Temperature Figure 4. 15. Weight loss measured as a function of temperature for cresol novolac/Epon 828 networks A) in air, and B) in nitrogen 143 4.3.1.4. Flame results The flame retardance of these cresol novolac/epoxy networks were measured using a cone calorimeter with a heat flux of 50 kW/m2 and 20.9% O2 (atmospheric oxygen). The heat release rate curves, measured as a function of time, showed different burning behaviors for the Epon 828 than those of the D.E.N. 438 epoxy cured networks (Figure 4. 16). The shapes of the curves are representative of the type of epoxy used in the networks. The peak heat release rate (PHRR) occurred at the ignition for networks cured with Epon 828 epoxy. The heat release rate then decreased rapidly as a function of time. The networks cured with D.E.N. 438 epoxy showed an initial spike at ignition. The heat release rate then gradually increased to a maximum followed by a rapid decline. For these networks, the maximum for the bulk heat release rate was assigned as the peak heat release rate. The heat release rate curves of cresol novolac/D.E.N. 438 networks were more desirable since longer times were required to reach the peak heat release rate, and this was more representative of flame retardant char forming materials. HRR (kW/m2) Time (s) Figure 4. 16. Cone calorimetry results of A) cresol novolac/Epon 828 (70:30 wt:wt ratio), and B) cresol novolac/D.E.N. 438 (70:30 wt:wt ratio) As expected, networks containing higher novolac contents showed lower peak heat release rates since novolac contributes to the flame retardance (Table 4. 7). Lower 144 HRR (kW/m2) A B peak heat release rates were observed for networks cured with D.E.N. 438 epoxy probably because D.E.N. 438, an epoxidized novolac, resembles the novolac structure and therefore contributes to the flame retardance. The peak heat release rates of cresol novolac/Epon 828 epoxy networks were higher than for the phenolic novolac/Epon 828 epoxy networks at similar compositions. The extra methyl groups on the cresols may have contributed to the higher peak heat release rate. However, the peak heat release rates of all cresol novolac/epoxy networks (300-450 kW/m2) were significantly lower than that of the epoxy control (1230 kW/m2). Table 4. 7. Flame retardance of cresol novolac/epoxy networks Networks Epoxy control Phenolic control Phenolic Novo/Epon 828 Cresol Novolac/Epon 828 Phenolic/Epoxy (wt/wt) --35/65 70/30 80/20 60/40 70/30 80/20 PHRR (kW/m2) 1230 116 357 448 391 380 404 310 Char Yield (wt.%) 5 63 29 16 15 17 18 18 Cresol Novolac/D.E.N. 438 Higher char yields are desirable since char forms an isolation layer which generally improves flame retardance. According to cone calorimetric results, there were no significant differences in the char yield among the various cresol novolac/epoxy networks. Networks with different compositions prepared with either Epon 828 or D.E.N. 438 epoxy formed 15 to 18 percent char. The char yields were lower for cresol novolac/Epon 828 networks (~16 wt %) than for phenolic novolac/Epon 828 networks (~ 29 wt %) with similar network compositions, but significantly higher than for the epoxy control (~5 wt %). All novolac/epoxy networks exhibited reduced char formation compared to that of the phenolic control network (~63 wt %). 145 4.3.1.5. Water absorption and diffusion efficient Water has a Tg of 130C. Polymers that absorb water are also plasticized by water. Therefore, water generally has a negative effect on the Tgs of polymers and reduces the upper temperature limit for the polymers applications. Typical phenolic materials are fairly hydrophilic because each phenolic repeat unit contains a free hydroxyl group. By adding a hydrocarbon substituent, such as a methyl group, in close proximity of the hydroxyl groups, such as in the case of cresol novolac, the polymer chains should become less polar and absorb less water. Equilibrium moisture uptake was studied at both room temperature and 62C for a series of cresol novolac/epoxy networks. The 62C water uptake test was designed as an accelerated protocol, but these networks also absorb more water at higher temperatures. Three samples of each specimen were measured at room temperature and the results were extremely reproducible (Figure 4. 17). 3 2.5 2 % wt 1.5 1 0.5 0 0 20 40 60 t1/2/thickness Figure 4. 17. Room temperature weight percent water uptake for cresol novolac/Epon 828 networks (70:30 wt:wt ratio) sample 1 sample 2 sample 3 The water absorption levels for cresol novolac/epoxy networks were relatively unaffected by the structures of the epoxies (Figure 4. 18). The 80:20 cresol 146 novolac/epoxy network absorbed slightly more water than the 70:30 and 60:40 cresol novolac/epoxy networks, probably as a result of having more polar phenolic hydroxyl groups in the higher novolac compositions. All of the cresol novolac networks (1.8-2.2 wt %) showed low equilibrium water uptake comparable to that of the epoxy control (2 wt %). Cresol novolac networks absorb significantly lower amounts of water than do their phenolic novolac counterparts. For example, at room temperature the 65:35 wt:wt phenolic novolac/Epon 828 epoxy networks absorbed 3.5 wt. % water whereas the 70:30 wt:wt cresol novolac/Epon 828 networks absorbed only 1.9 wt. % water. 4.5 4 3.5 room temperature 62C % water uptake 3 2.5 2 1.5 1 0.5 0 -A 8 8 8 8 8 43 EN 60 :4 0 D EN ro 82 82 82 43 nt :B is on on on PN Ep Ep Ep y D EN co ox 5 0 0 Ep :3 0 0 :2 :3 65 :2 :4 80 80 70 Figure 4. 18. Water uptake results for cresol novolac networks at room temperature and 62C The diffusion coefficient (D) for cresol novolac/epoxy networks were calculated using the weight percent room temperature water uptake measured as a function of time according to the following equation. D = (sb/4M)2 147 (4. 21) 60 70 :3 0 D 43 8 l Where s is the initial slope of the plot wt % water uptake vs. time1/2 b is the sample thickness M is the equilibrium weight percent water uptake This calculation assumes that the sample thickness is significantly less than the sample width and height. The network crosslink densities and chemical structures as well as the interactions between the small molecular (water) and the polymer matrix affect the diffusion coefficient. Table 4. 8. Diffusion efficient of cresol novolac/epoxy networks Epoxy Cresol novolac/Epoxy (wt:wt) 80:20 70:30 60:40 80:20 70:30 60:40 D (cm2/sec) 1.02 x 10-9 1.27 x 10-9 1.71 x 10-9 1.23 x 10-9 1.94 x 10-9 1.56 x 10-9 Epon 828 DEN 438 4.3.1.6. Reaction kinetics A differential scanning calorimeter was used to determine the kinetic parameters for the cresol novolac/epoxy reactions. This procedure allowed for the calculation of reaction activation energy and the rate constants as a function temperature. The time that was required to achieve a full conversion at any given temperature thus can be predicted using these kinetic parameters. Samples were heated from 50C to 250C at several heating rates () and the temperature at which the exothermic reaction peak occurred (T) was recorded at each heating rate. The activation energy (E) was approximated from the slope of the plot of log versus 1/T (Figure 4. 19). E - 2.19 R [d log10 /d(1/T) ] (4. 22) 148 where R is the gas constant (8.3145 J K-1 mol-1). 1.4 1.1 log 0.8 0.5 0.0021 0.00215 0.0022 0.00225 0.0023 1/T (K-1) y = -4084.7x + 9.9934 R2 = 0.9993 Figure 4. 19. Log heating rate versus 1/T for cresol novolac/epoxy mixture (70:30 wt:wt ratio) with 1 mole % TPP catalyst The activation energy was refined according to the procedures described in ASTM E 698 until self-consistent. For cresol novolac/epoxy reactions, the activation energy was found to be ~69 KJ/mol, which is comparable to those cited in literature for epoxy ring opening reactions. The Arrhenius pre-exponential factor (Z) was calculated as follows Z= E e(E/RT) RT 2 (4. 23) where is a heating rate (taken from the middle of the range tested) and T is the temperature (K) in the middle of the range. The kinetic rate constant, k, was calculated using the activation energy of the reaction and the pre-exponential factor E k = Z exp RT (4. 24) 149 The rate constant was plotted as a function of temperature (Figure 4. 20). As expected, the rate constant increased with increased temperatures. 0.04 k (1/sec) 0.03 0.02 0.01 0 90 120 150 o k= Ze-E/RT 180 210 Temperature ( C) Figure 4. 20. Rate constant (k) versus temperature for a cresol novolac/epoxy mixture (70:30 wt:wt ratio) with 1 mole % TPP catalyst To confirm the rate constant by an isothermal test, a time (t) was calculated to treat the sample at a chosen temperature (150C) to achieve 50 % cure, t1/2 = 0.693/k (4. 25) The aged sample and an unaged sample were run in a dynamic scan and compared (Figure 4. 21). The reaction kinetics was confirmed if on an equal weight basis, the peak area or displacement from baselines of the aged sample is approximately one half of that of the unaged sample. 150 Heat treated at 150C for 5.2 minutes Untreated sample 80 Figure 4. 21. 140 180 Temperature (oC) 220 260 Dynamic DSC scans of an untreated sample versus a heat treated sample 4.3.1.7. Processability The novolac/epoxy networks studied in this research were evaluated for their potential use in tough, flame retardant composites. Regardless of the fabrication method, the novolac/epoxy resin mixtures must be heated to obtain sufficiently low viscosities for processing (2-10 Pa*s). However, novolacs react with epoxies at elevated temperatures even in the absence of an added catalyst. Even a small amount of reaction greatly increases the viscosity. Therefore, it was essential to determine the processing window in which the novolac/epoxy mixtures remain below a given viscosity at the processing temperatures. The complex viscosity, measured as a function of temperature, (Figure 4. 22) showed that relatively high temperatures (>185C) were required for a neat 2000 g/mol ortho-cresol novolac resin to reach ~2 Pa*s. 151 70 60 Viscosity (Pa s) 50 40 30 20 10 0 150 Viscosity Pas 160 170 180 190 200 Temperature (oC) Figure 4. 22. Complex viscosity of a 2000 g/mol neat cresol novolac resin measured as a function of temperature It was of interest to investigate the viscosity behaviors of the neat phenolic novolac resin used in the control experiments for comparison purposes. Since a significant amount of phenol and other low molecular weight volatile components were present in the typical novolac resin, the complex viscosities of the neat phenolic novolac resin (untreated) and the complex viscosity of this resin heated for 2 hours at 160C were examined. Phenol and other low molecular weight components clearly reduced the viscosities of the resin (Figure 4. 23). It is important to note that these resins were melt mixed under vacuum at > 140C, a condition that removed phenol, which should affect the viscosities of the novolac/epoxy mixtures during the prepreg process in composite fabrications. 152 150 Viscosity (Pa s) 120 90 60 30 0 120 Before After 140 160 180 Temperature (oC) Figure 4. 23. Complex viscosity of a phenolic novolac resin before and after heat treatment (2 hours at 160C) Dynamic viscosity measurements for cresol novolac/Epon 828 mixtures showed that increased epoxy content reduced the melt viscosities (Figure 4. 24 a), and therefore, the processability was enhanced as the amount of low molecular weight epoxy increased. From dynamic viscosity measurements, the temperature at which the viscosity reached approximately 2 Pa*s was chosen as the processing temperature. Higher processing temperatures were necessary for compositions with lower amounts of epoxies. The isothermal viscosities were then evaluated at the processing temperatures of each composition to determine the processing windows (Figure 4. 24 b and c). The 70:30 and 60:40 compositions both exhibited great stabilities at their processing temperatures (140C and 120C respectively) with very little viscosity increase over 150 minutes. 153 A 10 Viscosity (Pa s) 8 6 4 2 0 110 80:20 70:30 60:40 120 130 140 150 o 160 170 180 Temperature ( C) B Viscosity (Pa s) 10 8 6 4 2 0 0 20 40 60 80 100 120 Time (min) C Viscosity (Pa s) 10 8 6 4 2 0 0 30 60 90 120 150 Time (min) Figure 4. 24. Viscosity measurements of cresol novolac/Epon 828 mixtures A) dynamic scans for various compositions, B) isothermal scan of the 70:30 composition at 145C, and C) isothermal scan of the 60:40 composition at 120C 154 Isothermal viscosities were measured for the 65:35 wt:wt phenolic novolac/Epon 828 mixture (140C) and compared with those of the 70:30 cresol novolac/Epon 828 mixture (145C) (Figure 4. 25). The cresol novolac/epoxy mixture showed only a slight viscosity increase (from 2.5 to 3 Pa*s) over a 2-hour period, whereas the viscosity of the phenolic novolac/epoxy mixture increased substantially over the same period (3 to 12 Pa*s). Thus, the processing window for the cresol novolac/epoxy mixtures is significantly more desirable than for the phenolic novolac/epoxy mixtures. This increase in the processing window was attributed to the lower reactivity of the cresol novolac with epoxy groups. The methyl group ortho to the hydroxyl group probably caused extra steric hindrance to the phenolic hydroxyl/epoxy reactions. A Viscosity (Pa s) 15 10 5 0 0 30 60 90 Time (min) 120 B Viscosity (Pa s) 15 10 5 0 0 30 60 90 Time (min) 120 Figure 4. 25. Isothermal viscosity measurements: A) 65:35 wt:wt phenolic novolac/Epon 828 mixture measured at 140C, and B) 70:30 wt:wt cresol novolac/Epon 828 mixture measured at 145C D.E.N. 438 epoxy was not as effective as Epon 828 epoxy in reducing the melt viscosities of cresol novolac/epoxy mixtures (Figure 4. 26A). At the same compositions, higher temperatures were needed to obtain similar viscosities when D.E.N. 438 epoxy was used. All cresol novolac/D.E.N. 438 mixtures required at least 160C to reach the processable viscosities. At these temperatures, the inherent reaction between cresol novolac and epoxies became apparent as the viscosities increased from approximately 1.5 Pa*s to approximately 3 Pa*s over 100 minutes (Figure 4. 26B). 155 A Viscosity (Pa s) 10 8 6 4 2 0 130 80:20 70:30 60:40 140 150 160 o 170 180 190 Temperature ( C) B Viscosity (Pa s) 10 8 6 4 2 0 0 20 40 60 80 100 Time (min) Figure 4. 26. Viscosity measurements for cresol novolac/D.E.N. 438 mixtures: A) dynamic measurements, B) isothermal scan for the 60:40 composition at 160C 4.3.2. Composites properties Flexural properties were examined for cresol novolac/Epon 828 composites. The 70:30 composition was chosen since it had the best overall properties, i.e. high KIC and Tg, low sol fractions, good flame retardance, and excellent processability. The composites were reinforced with phenoxy-sized carbon fiber. No catalysts were used for the cresol novolac/epoxy cure reaction. To determine the cure conditions necessary to achieve a cured network, the glass transition temperature was measured as a function of cure time (Table 4. 9). The cresol novolac/epoxy reaction processed very slowly, even at 156 220C, in the absence of a catalyst. A sample cured at 220C for over 10 hours had a Tg that was significantly lower than that of a fully cured (TPP catalyzed) network. Table 4. 9. Cure condition determination for ortho-cresol novolac/Epon 828 network (70:30 wt:wt %), no catalyst sample 1 2 3 4 5 Cure conditions 1 hr @ 180C, 4 hrs @ 220C 1 hr @ 180C, 6.5 hrs @ 220C 1 hr @ 180C, 8.5 hrs @ 220C 1 hr @ 180C, 10.5 hrs @ 220C TPP cured sample* Tg (C) Soluble Partially soluble 120 135 154 * cured using 0.1 mole % TPP at 200C for 2 hours and 220C for 2 hours Composite properties were evaluated for the cresol novolac/Epon 828 networks cured under condition 4, i.e. the network was not fully cured. The flexural properties were superior to those of the control epoxy, but they were not as high as the fully cured phenolic novolac/Epon 828 composites (Table 4. 10). The properties of composites with fully cured cresol novolac/epoxy matrix resin should exhibit comparable to those properties of the phenolic novolac/epoxy based composites. Table 4. 10. Flexural strength and moduli of composites Sample Control epoxy Phenolic novolac/Epon 828 (65:35) Cresol novolac/Epon 828 (70:30) max (MPa) 29 66 46 Eb (GPa) 8.9 11.3 10.1 4.3.3. Para-cresol based networks and their properties The networks prepared using a 2000 g/mol para-cresol novolac and their properties were somewhat explored (Figure 4. 27). 157 OH HO CH2 + CH3 O O O OH O n O O CH2 OH n Triphenylphosphine Crosslinked Network Figure 4. 27. 2000 g/mol para-cresol novolac cured with Epon 828 The viscosity profile of a neat 2000 g/mol para-cresol novolac resin showed an unexpected increase at higher temperatures (>175C) (Figure 4. 28). Para-cresol novolac oligomers are highly stereo-regular, and therefore may form ordered structures at higher temperatures. In sample preparations, high temperatures were required to obtain low viscosities for proper degassing; however, increasing the temperature to ~ 170C increased the viscosity when para-cresol novolac resin was used. Therefore, sample preparations for these para-cresol novolac/epoxy networks became more challenging especially at high novolac compositions (80:20 wt:wt ratio). compositions were successfully prepared. The 70:30 and 60:40 40 35 30 25 20 15 10 160 Viscosity (Pa s) 180 Temperature ( C) o 200 Figure 4. 28. Viscosity of a 2000g/mol para-cresol novolac resin (heat rate = 2.5C /min) 158 Para-cresol novolac/Epon 828 epoxy network properties did not seem to follow any trend (Table 4. 11). The 70:30 composition showed a lower toughness but a higher Tg than the 60:40 composition. The 70:30 wt:wt para-cresol novolac/epoxy networks showed a similar Tg but a significantly lower KIC compared to the ortho-cresol novolac/epoxy networks with the same novolac content. These network properties were not only a result of the crosslink density but were complicated by the packing order of para-cresol novolac chains. At higher para-cresol novolac contents, a higher degree of sterioregularity may lead to the observed higher Tg yet lower KIC values. Table 4. 11. KIC and Tg of para-cresol novolac/Epon 828 networks Cresol ortho para para Sample 70:30 70:30 60:40 KIC (MPa*m1/2) 1.06 0.78 0.89 Tg ( C) o 154 155 140 4.4. Conclusions Cresol novolac/epoxy networks were prepared using a controlled molecular weight 2,6-dimethylphenol endcapped ortho-cresol novolac resin and the network properties were compared to those of an epoxy control, a phenolic control and a phenolic novolac/epoxy control. The 70:30 and 60:40 compositions exhibited relatively high toughness (> 1 MPa/m1/2) and high glass transition temperatures (>140C). These properties were superior to these of the epoxy control, the phenolic control and the phenolic novolac/epoxy networks (65:35 wt:wt ratio). The molecular weight between crosslinks (~1500 g/mol), determined using the theory of rubber elasticity, showed that the network density was optimized at the 70:30 composition. This Mx was most desirable for networks cured with 2000 g/mol oligomers. Compositions high in novolac content (80:20 wt:wt ratio) had reduced crosslink densities due to insufficient network connectivity; compositions high in epoxy content (60:40 wt:wt ratio) also had reduced 159 apparent crosslink densities because a lack of mobility prohibited further phenolic/epoxy reactions, which led to increased epoxy sol fractions. Master curves and cooperativity plots were generated for cresol novolac/epoxy networks. The fragility was used to investigate the temperature dependence of segmental relaxation through the transition region. Both crosslink density and hydrogen bonding affected the properties of Epon 828 epoxy cured cresol novolac networks. Crosslink density seemed to be the dominating factor in the D.E.N. 438 epoxy cured cresol novolac networks. Cresol novolac/epoxy networks exhibited comparable, relatively low, equilibrium water absorptions similar to that of the epoxy control (~ 2 wt % at room temperature). Their equilibrium water uptake were relatively unaffected by the network compositions. The methyl group on each novolac repeat unit probably reduced water absorption. The peak heat release rates of ortho-cresol novolac/epoxy networks were between 300-450 kW/m2. The presence of a methyl group on each repeat unit slightly increased the peak heat release rates and reduced the char yields of cresol novolac networks relative to those of phenol novolac networks. The flame retardance of all the novolac/epoxy networks were significantly superior to that of the epoxy control, but inferior to that of the phenolic control. The kinetic parameters were determined for the cresol novolac/epoxy reactions. The activation energy for these reactions was approximately 69 KJ/mol, which was comparable to the literature values for phenolic hydroxyl/epoxy ring opening reactions. The kinetic parameters allowed for prediction of the cure time required to achieve 99% conversion. Cresol novolac resins gave rise to longer processing window times when mixed with epoxies at elevated temperatures. This was again attributed to the methyl group in close proximity to the hydroxyl group which sterically hindered the phenolic hydroxyl/epoxy reaction. 160 5. Maleimide Containing Cresol Novolac Networks and Their Properties 5.1. Introduction Maleimides are a relatively new class of materials used in high performance structural composite and adhesive applications. Major advantages of maleimide networks include excellent stabilities at elevated temperatures and in hot-wet conditions, low moisture absorptions, and excellent chemical stabilities. The thermal and mechanical properties of maleimide networks are superior to those of most epoxies.165 Typical maleimide oligomers are difunctional (bismaleimides) obtained by reacting diamines with maleic anhydride (Figure 5. 1). O H2N R NH2 + 2 O O O N O O RN O Figure 5. 1. Preparation of bismaleimide from a diamine and maleic anhydride Bismaleimides can be crosslinked alone or in the presence of a curing agent. Curing bismaleimides alone generally requires high temperatures to initiate reactions and achieve high conversions. Introducing an initiator effectively reduces the cure 165 K. N. Ninan, K. Krishnan, and J. Mathew, Addition Polyimide: 1. Kinetics of Cure Reaction and Thermal Decomposition of Bismaleimide, Journal of Applied Polymer Science 32(7), 6033-6042 (1986). 161 temperatures. Adding peroxide initiators166 results in free-radical initiations while adding tertiary amines or imidazole167 leads to anionic initiations. The thermal cure of bismaleimides such as 4,4-bismaleimidodiphenylmethane in the absence of initiator was suggested to proceed via a free radical process.168 This was supported by the increased activation energies observed in the presence of a small amount of impurities, which was proposed to interfere with the free radical reaction. Thermally initiated polymerization of maleimide was also suggested to occur in a heterogeneous manner.169 Gelation was reached rapidly in the form of microgels, which were attributed to slow initiation rates but fast propagation rates. Macrogelation occurred much later in the reactions. Anionically initiated maleimide polymerizations proceeded via a more homogeneous mechanism.169 Both initiation and propagation occurred more rapidly. As expected, maleimide polymerization rates decreased significantly after vitrification due to slow diffusion. Bismaleimides have been cured with amines to increase the toughness by reducing the network crosslink densities. Curing bismaleimides with diamines involves both a lower temperature primary amine addition to maleimide double bonds (Figure 5. 2a) and a higher temperature homopolymerization of maleimide double bonds (Figure 5. 166 M. Acevedo, J. de Abajo, and J. G. de la Campa, Kinetic Study of the Cross-linking H. D. Stenzenberger, M. Herzog, W. Romer, R. Scherblich, S. Pierce, and M. Reaction of Flexible Bismaleimide, Polymer 31, 1955, (1990). 167 Canning, Compimides: A Family of High Performance Bismaleimide Resins, 30th Nat. SAMPL Symp. 30, 1568-1586 (1985). 168 I. M. Brown and T. C. Sandreczki, Crosslinking Reactions in Maleimide and A. Seris, M. Feve, F. Mechin, and J. P. Pascault, Thermally and Anionically Initiated Bis(maleimide) Polymers-an ESR Study, Macromolecules 23(1), 94-100, (1990). 169 Cure of Bismaleimide Monomers, Journal of Applied Polymer Science 48, 257-269 (1993). 162 2b).170 The cure reaction of 1,1-(methylenedi-4,1-phenylene) bismaleimide and 4,4methylenedianiline, monitored by FTIR and DSC, showed that the nucleophilic addition of amine to the maleimide double bond, which resulted in extension of network chains, occurred via a second-order reaction. amine addition. The homopolymerization, which led to crosslinking, proceeded at a rate at least two orders of magnitude lower than did the The homopolymerization involved thermal initiations and chain propagations. Lower cure temperatures or increased diamine contents therefore favored the chain extension reactions over the crosslinking reactions.171 A O N O + NH2 O N NH O B O N O + O N O O N O O N O Figure 5. 2. Reactions of bismaleimide in the presence of a diamine: A) chain extension due to an amine addition, and B) crosslinking obtained by maleimide homopolymerization reactions According to DSC measurements, the temperature of exothermic transition due to bismaleimide/amine reactions decreased as the basicity of the amines increased. It was suggested that faster reactions could be achieved if more basic amines were used in the 170 A. V. Tungare and G. C. Martin, Analysis of the Curing Behavior of Bismaleimide T. M. Donnellan and D. Roylance, Relationships in Bismaleimide Resin System. Part Resins, Journal of Applied Polymer Science 46, 1125-1135 (1992). 171 I: Cure Mechanisms, Polymer Engineering and Science 32(6), 409-414 (1992). 163 reactions. However, post-curing above 200C was necessary for all maleimide/amine reactions to achieve high conversions.172 The structures of maleimide and amine affect the network thermal stabilities and flame retardance. For example, higher char yields were obtained for bismaleimides cured with phosphorus based amines.172 Several mechanisms were proposed to be involved in the bismaleimide/diallylbisphenol cure. They include 1) homopolymerization of maleimide groups, 2) homopolymerization of allyl groups, 3) reactions between maleimide double bonds and allyl groups via a Diels-Alder reaction, 4) reactions of maleimide with the allyl component and/or another maleimide via a free radical site,173 and 5) crosslinking reactions via dehydration of the allylbisphenol hydroxyl groups (self-condensation).174 Cure reaction of 4,4-methylenebis-(maleimidobenzene) and 2,2-diallylbisphenol A, monitored using near-IR spectroscopy, showed that the principal reaction pathway involved alternating copolymerization of maleimide and allyl double bonds.175 Maleimide homopolymerization was found to be significant only during the initial stage of the reaction when heated above 200C. The self-condensation reactions (pathway 5), which released water as the by-product, were observed over the entire temperature range investigated (140C to 250C). 172 I.K. Varma and D.S. Varma Addition Polyimides. III. Thermal Behavior of Bismaleimide, Journal of Polymer Science: Polymer Chemistry Ed. 22, 1419-1483 (1984). 173 I. M. Brown and T. C. Sandreczki, Characterization of Bismaleimide Cure Reaction by Electron-Spin Resonance Techniques, Abstract Paper to American Chemical Society, S, Polymer Material Science and Engineering-128, 169 (1988). 174 R. J. Morgan, R. J. Jurek, A, Yen, and T Donnellan, Toughening Procedures, Processing and Performance of Bismaleimide Carbon Fiber Composites, Polymer 34(4), 835-843 (1993) 175 J. Mijovic and S. Andjelic Study of the Mechanism and Rate of Bismaleimide Cure by Remote in-situ Real Time Fiber Optic Near-Infrared Spectroscopy, Macromolecules 29, 239-246 (1996). 164 Addition curable novolac resins containing maleimide functionalities were prepared by co-reacting phenol and N-4-hydroxylphenylmaleimide with formaldehyde (Figure 5. 3). The oligomers were thermally crosslinked or cured with epoxy resins for structural adhesive applications.176 Network adhesion properties depended on the cure conditions and the test temperatures. Self-cured networks had low lap shear strength and low T-peel strength under ambient conditions, but these properties improved at higher temperatures. The enhanced adhesion at higher temperatures was attributed to thermally induced molecular relaxations in the tightly crosslinked networks. Maleimide-novolac oligomers cured with epoxies showed improved adhesions and higher thermal stabilities compared to novolac/epoxy networks without maleimide. Improved adhesions were attributed to secondary forces of attraction induced by the polar imide groups through their partial polymerization. OH O + O N O HCH CH2 x O N O OH OH OH CH2 y Figure 5. 3. Synthesis of novolac resins containing maleimide functionalities The goal of this part of the research was to incorporate the maleimide moiety into cresol novolac/epoxy networks and determine the network properties. Cresol novolac oligomers containing maleimide groups (cresol-co-HPM novolac) were first synthesized, followed by crosslinking with bisphenol-A epoxy to form networks. Maleimide groups were expected to behave as latent crosslink sites since the self-cure occurs only at higher temperatures. Adding maleimide groups into networks therefore should improve thermal stabilities and may enhance flame retardance. 176 C. Gouri, C. P. Reghunadhan Nair, and R. Ramaswamy, Adhesive and Thermal Characteristics of Maleimide-Functional Novolac Resins, Journal of Applied Polymer Science 73, 695-705 (1999). 165 5.2. Experimental 5.2.1. Reagents 2-Amino-p-cresol, 4-aminophenol, ortho-cresol, 2,6-dimethylphenol, formaldehyde (37 wt% solution in water), maleic anhydride, oxalic acid dihydrate, paraformaldehyde (powder, 95%), phosphorus pentoxide, and triphenylphosphine were purchased from Aldrich. N,N-dimethylformamide, hexane, and diethyl ether were obtained from EM Science. Isopropanol was purchased from Allied Signal, and sulfuric acid was obtained from Fisher Chemical. All reagents were used as received. 5.2.2. Synthetic Methods 5.2.2.1. Synthesis of 4-Hydroxyphenylmaleimide (4-HPM) To a 2000 ml 3-neck round bottom flask equipped with a mechanical stirrer and an addition funnel was added maleic anhydride (grounded into small pieces) (180g, 1.1mol) and DMF (250 ml) (Figure 5. 4). Once the maleic anhydride completely dissolved, 4-aminophenol (109g, 1mol) was added gradually to the reaction flask through a powder funnel. The mixture was allowed to react at room temperature for 2 hours. In a separate 1000 ml one-neck round bottom flask, P2O5 (57 g) and DMF (350 ml) were added. While stirring, H2SO4 (25 g) was added drop wise. The H2SO4/P2O5/DMF mixture was stirred until a homogenous solution was obtained; it was then added drop wise to the reaction through the addition funnel. The reaction mixture was heated to 70C and held at this temperature for 2 hours to achieve full imidization. To work up the reaction product, approximately 250 ml DMF was removed via vacuum distillation. The reaction mixture was precipitated in ice water, filtered through a Buchner funnel, and then rinsed with deionized water 5 times or until the product was neutral. 70C for 2 hours under vacuum prior to use. It was recrystallized in isopropanol, then washed with hexane. The 4-HPM product was dried at 166 O O O + OH H O N CH3 CH3 OH H2SO4 DMF/P2O5 NH OH O OH NH2 O O N O Figure 5. 4. Synthesis of 4-hydroxyphenylmaleimide 5.2.2.2. Synthesis of 2-hydroxy-5-methylphenylmaleimide The procedure used to prepare 4-HPM was also used to prepare 2-hydroxy-5methylphenylmaleimide. To work up the reaction, a vacuum distillation was used to remove ~ 250 ml DMF. The reaction mixture was precipitated in ice water and filtered through a Buchner funnel. The precipitated product was dissolved in an 85/15 vol/vol ether/isopropanol mixture and washed with deionized water in a separatory funnel until neutral. A rotovap apparatus was used to remove most of the organic solvent. The product was recrystallized in the remaining solvent with the addition of a small amount of hexane. It was washed with hexane and dried at 70C for 2 hours under vacuum. O O O OH H2N + CH3 DMF O O HN OH OH H2SO4/DMF/P2O5 N O O OH CH3 CH3 Figure 5. 5. Synthesis of 2-hydroxyl-5-methylphenylmaleimide 5.2.2.3. Synthesis of 2,6-dimethylphenol endcapped o-cresol-co-HPM novolac oligomers Controlled molecular weight, 2,6-dimethylphenol endcapped cresol-co-HMP oligomers were prepared via electrophilic aromatic substitution (Figure 5. 6). The predetermined amount of reagents, ortho-cresol (180.0g), 2,6-dimethylphenol (50.9g), 4HPM (55.6g), formaldehyde (71.9g), and oxalic acid dihydrate (7.4g) were added to a resin kettle and reacted at 80C for 48 hours. The product was dissolved in acetone, 167 precipitated into ice water, then filtered on a Buchner funnel. It was washed with deionized water until neutral and then dried drying at 60C under vacuum for 5 hours. OH OH + O OH + N O paraformaldehyde oxalic acid dihydrate OH HO CH2 CH2 x O N O OH CH2 y OH Figure 5. 6. Synthesis of 2,6-dimethylphenol endcapped cresol-co-HMP novolac resin A sample calculation for the amount of reagent needed to obtain a 1500g/mol oligomer with 15 wt% internal maleimide is as follows. The combined molecular weight of the repeat units per chain (Mu) is obtained by subtracting the combined end group molecular weight (Me) from the targeted molecular weight (Mn) Mu = 1500 Me = 1500 (121+121+14) = 1244 percent cresol (repeat unit x) and 15 weight percent 4-HPM (repeat unit y) (5. 1) A normalized molecular weight per repeat unit is calculated for oligomers containing 85 weight Normalized Mn = 0.85 * Mn(x) + 0.15 * Mn(y) = 0.85(120)+0.15(201) = 132.2 unit into Mu (x+y) = Mu/normalized Mn = 1244/132.2 = 9.41 (5. 3) (5. 2) The number of internal units was calculated by dividing the normalized Mn per repeat 168 The number of cresol containing repeat units is 85 percent of (x+y), and the number of 4HPM containing repeat units is 15 percent of (x+y). There are 2 moles of 2,6dimethylphenol per chain, and (x+y)+1 or 10.41 moles of formaldehyde. 5.2.2.4. Synthesis of cresol novolacs with 2-Hydroxy-5-methylphenylmaleimide endgroups 2-Hydroxy-5-methylphenylmaleimide was used as opposed to 2,6-dimethylphenol as the endcapper for the synthesis of controlled molecular weight cresol novolac oligomers (Figure 5. 7). The procedures for these reactions were the same as those described in section 5.2.2.3. OH O OH N O CH3 + OH CH3 O HCH oxalic acid dehydrate O CH3 N O OH CH2 CH3 CH2 HO O N n CH3 O Figure 5. 7. Synthesis of 2-hydroxy-5-methylphenylmaleimide terminated cresol novolac resins 5.2.3. Characterization The characterization methods used in these experiments, including 1H NMR, DSC, TGA, fracture toughness, gel fraction, and cone calorimetry are described in detail in Chapter 4.2.3. 5.3. Results and Discussion 5.3.1. 4-Hydroxyphenylmaleimide synthesis and characterization 4-HPM monomer was prepared by reacting 4-aminophenol with maleic anhydride.177 177 1 H NMR verified the chemical structures and the purity of the monomer J. O. Park and S H. Jang, Synthesis and Characterization of Bismaleimides from Epoxy Resins, Journal of Polymer Science. Part A: Polymer Chemistry 30, 723-729 (1992). 169 (Figure 5. 8). As expected, four sets of peaks were observed for 4-HPM. Protons on imide double bonds (a) resonate at 7.1 ppm; two types of aromatic protons (b and c) are observed at 6.8 and 7.06 ppm, respectively. Phenolic hydroxyl protons resonate at 9.8 ppm. Water and DMSO were found at 3.4 and 2.5 ppm respectively. A clean 1H NMR spectrum of 4-HMP showed that the monomer was relatively pure a c b OH d c b 7 .1 7 .0 6 .9 6 .8 6 .7 PPM O N O a d 10 9 8 7 6 5 4 3 2PPM Figure 5. 8. 1 H NMR spectrum of 4-hydroxyphenylmaleimide monomer The melting point of 4-HPM was measured using differential scanning calorimetry (Figure 5. 9). The observed melting point (191C) was comparable to that cited in the literature.177 Further heating above the melting point led to an exotherm which was attributed to the crosslinking reactions. 170 Tm = 191oC Figure 5. 9. Melting point of 4-HPM determined via DSC The thermal stability of 4-HMP was investigated using thermogravimetric analysis (Figure 5. 10). No weight loss was observed up to 160C. Upon further heating, thermal degradation may have led to the observed weight loss. Surprisingly, the onset of weight loss for 4-hydroxyphenylmaleimide occurred below its melting point. Figure 5. 10. Thermal stability of 4-HPM monomer measured via TGA (10C/min, N2) 171 5.3.2. Cresol-co-HPM novolac oligomers and their properties Maleimide functionalities were incorporated into cresol novolac oligomers via electrophilic aromatic substitution reactions of cresol and 4-HPM with formaldehyde. The hydroxyl group on 4-HPM should dominate in electron donating compared to the imide group, therefore the monomer was assumed to be difunctional in formaldehyde substitution reactions. The molecular weight of presumed linear cresol-co-HPM novolac oligomers was controlled by adjusting the stoichiometric ratio of monomers (cresol and 4-HPM) to endgroups (2,6-dimethylphenol) (as discussed in Chapter 2). Cresol-co-HPM novolac oligomers were prepared in three molecular weights. The 1500 and 1250 g/mol oligomers contained 15 mole % maleimide. At this composition, there was at least one maleimide group per chain statistically. A higher maleimide composition was necessary to maintain one maleimide site per chain (statistically) at lower oligomer molecular weights (such as 1000g/mol). The molecular weights of cresol-co-HMP novolac could not be calculated quantitatively using 1H NMR or 13 C NMR spectroscopy due to the complexity of the spectra and extensive peak overlapping. 1H NMR spectra confirmed the presence of maleimide groups which were clearly distinguishable from the other aromatic peaks (Figure 5. 11). Methylene linkages are also evident between 3.4 and 3.9 ppm. 7 .4 7 .2 7 .0 6 .8 6 .6 6 .4 6 .2 PPM 9 8 7 6 5 4 3 2 1P P M Figure 5. 11. 1 H NMR of a typical cresol-co-HPM novolac resin 172 The glass transition temperatures of cresol-co-HPM novolac oligomers were measured as a function of targeted molecular weight. As expected, increased molecular weights led to higher glass transition temperatures (Table 5. 1). The oligomer molecular weights were particularly important in terms of processability considerations. Higher molecular weight oligomers required more elevated processing temperatures. However, the presence of thermally labile maleimide groups limited the upper processing temperature range. Therefore, the number average molecular weights of cresol-co-HPM novolac oligomers were kept reasonably low. correlated with oligomer processability. The glass transition temperatures Table 5. 1. Tg of cresol-co-HMP oligomer as a function of Mn Target Mn (g/mol) 1500 1250 1000 4-HPM content* (Mole %) 15 15 20 Tg ( C) o 87 54 35 * Mole % of internal repeat units bearing maleimide groups 5.3.3. Cresol-co-HPM novolac/epoxy network properties The fracture toughness, Tg, and sol fractions of cresol-co-HPM novolac/Epon 828 networks cured at various compositions were measured (Table 5. 2). The network properties were affected by the molecular weight of cresol-co-HMP novolac as well as the novolac/epoxy composition. When higher molecular weight cresol-co-HPM novolac oligomers were used (1500g/mol and 1250 g/mol), the 70:30 and the 60:40 compositions showed relatively high Tgs and KICs and low sol fractions. As the oligomer molecular weight decreased, a larger amount of epoxy was necessary to form well-connected networks. Therefore, only the 60:40 composition exhibited good network properties when the 1000 g/mol oligomer was used. For the well connected networks, the network properties were comparable to those of a 2000g/mol ortho-cresol novolac/Epon 828 epoxy network (70:30 wt:wt ratio). 173 Table 5. 2. Properties of ortho-cresol-co-HPM/Epon 828 networks Mn (g/mol) 2000* Novolac/Epoxy (wt:wt) 70:30 80:20 Tg (C) 154 129 129 113 -131 156 121 125 KIC (MPa*m1/2) 1.060.04 0.47 1.180.08 0.900.10 -0.950.05 0.970.05 -0.240.03 0.870.10 Sol fraction (wt. %) 5 12 3 13 25 9 5 43 15 7 1500 70:30 60:40 80:20 1250 70:30 60:40 80:20 1000 70:30 60:40 122 * cresol novolac/epoxy networks The thermal stabilities of cresol-co-HPM novolac/epoxy networks were evaluated using thermogravimetric analysis in nitrogen (Table 5. 3). No weight loss occurred below 300C and all networks were relatively stable up to 400C. Interestingly, the char yield and the temperature for 5 percent weight loss seemed relatively unaffected by the network composition but was somewhat affected by the molecular weight of the cresolco-HPM novolac oligomer. In general, slightly higher char yields and temparatures for 5 percent weight loss were observed for networks prepared with higher molecular weight oligomers. For all network compositions, the char yield was between 21 and 27 weight percent and the temperature where 5 percent weight loss occurred ranged from 340C to 390C. 174 Table 5. 3. Thermal stability of cresol-co-HMP novolac/epoxy networks measured using thermogravimetric analysis Mn (g/mol) 1500 Novolac/Epoxy (wt:wt) 80:20 70:30 60:40 80:20 Char yield (Wt. %) 25 26 27 26 27 23 23 25 21 Temperature (5 wt% loss) 388 370 376 358 362 357 348 340 369 1250 70:30 60:40 80:20 1000 70:30 60:40 Cresol-co-HPM novolac/epoxy networks (80/20 composition) derived from higher molecular weight novolac oligomers showed slightly higher initial degradation temperatures (Figure 5. 12). The char yields at 800C were similar for these networks. 1500 g/mol 1250 g/mol 1000 g/mol Figure 5. 12. Percent weight loss for cresol-co-HPM novolac/epoxy networks (80:20 wt:wt ratio) prepared with different oligomer molecular weights, monitored using thermogravimetric analysis 175 Cone calorimetry was used to evaluate the flame retardance of cresol-co-HPM novolac epoxy networks. Samples were ignited at 50 kW/m2 radiant heat flux under atmospheric air conditions. The effects of network composition were investigated using networks prepared with 1250 g/mol cresol-co-HPM novolac oligomers (Table 5. 4). The 80:20 and 70:30 compositions showed comparable relatively low peak heat release rates (~400 kW/m2). The char yield for the 80:20 composition was slightly higher than for the 70:30 composition. As expected, the composition with the highest epoxy content (60:40 wt:wt ratio) exhibited the highest peak heat release rate and the lowest char yield. Table 5. 4. Cone calorimetry measuring the peak heat release rate (PHRR) and the char yield of 1250 g/mol cresol-co-HPM/epoxy networks Novolac:Epoxy (wt:wt) 80:20 70:30 60:40 PHRR (kW/m2) 407 396 572 Char yield (C) 16 15 12 The heat release rate curves of the 1250 g/mol cresol-co-HPM novolac/epoxy networks showed a two stage burning for the 80:20 and the 70:30 compositions (Figure 5. 13). The burning intensified to a single peak heat release for the 60:40 composition. Although the 80:20 and the 70:30 compositions had similar peak heat release rates, the heat released at both stages was significantly lower for the 80:20 composition. Therefore, the 80:20 composition showed superior flame resistance compared to the 70:30 composition. 176 50 novolac:epoxy (wt:wt) 40 30 20 100 0 60:40 70:30 80:20 HRR (kW/m2) 0 40 Time 80 120 Figure 5. 13. Heat release rate curves for cresol-co-HPM novolac/Epon 828 networks The effects of novolac oligomer molecular weight on the flame retardance were also explored (Table 5. 5). The peak heat release rates and the char yields were obtained for cresol-co-HPM novolac/epoxy networks (60:40 wt:wt ratio) prepared with different molecular weight novolacs . The networks prepared with the highest oligomer molecular weight (1500 g/mol) exhibited the lowest peak heat release rate and the highest char yield. The flame retardance decreased slightly as the oligomer molecular weight decreased. Networks containing higher molecular weight novolacs may improve flame retardance by favoring the phenolic type degradation pathways, which leads to higher thermal and thermo-oxidative stabilities. Table 5. 5. Cone calorimetry results for 60:40 wt:wt cresol-co-HPM/epoxy networks prepared with different molecular weight oligomers Mn (g/mol) 1500 1250 1000 PHRR (kW/m2) 514 572 543 Char yield (C) 15 12 10 177 5.3.4. Characterization of 2-Hydroxy-5-methylphenylmaleimide 2-Hydroxy-5-methylphenylmaleimide was prepared using the same method as the 4-HPM syntheses. This monomer was intended for use as an endcapper in the cresol novolac synthesis. The most reactive site for electrophilic aromatic substitution was expected to be the free ortho position with respect to the hydroxyl group. 1 H NMR indicated that 2-hydroxy-5-methylphenylmaleimide monomer was successfully prepared. The correct peak integrations in a clean NMR spectrum identified the pure compound (Figure 5. 14). b e O 7 .5 7 .0 6 .5 c b d O OH a N d CH3 c e f a water f DMSO 10 9 8 7 6 5 4 3 2 PPM Figure 5. 14. 1 H NMR of 2-hydroxy-4-methylphenylmaleimide Successive dynamic DSC scans were used to monitor the thermal crosslinking reactions of 2-hydroxy-4-methylphenylmaleimide (Figure 5. 15). The first dynamic DSC scan showed a melting point at 160.8C followed by an exotherm. Successive heatings 178 showed increased Tgs with each heating cycle which was indicative of increases in molecular weight. The amount of exotherm decreased with each heating cycle. 4th 3rd 2nd 1st Figure 5. 15. Successive dynamic DSC scans of 2-hydorxy-4-methylphenylmaleimide The thermal stabilities of 2-hydroxy-4-methylphenylmaleimide, investigated using TGA, indicated that the monomer was thermally unstable even below its melting point. The degradation initiated at approximately 110C (Figure 5. 16). Only 20 weight percent remained at 250C for 2-hydroxy-4-methylphenylmaleimide as opposed to 80 weight percent for 4-HPM (Figure 5. 10). The major difference between 2-hydroxy-4methylphenylmaleimide and 4-HPM is the position of the imide group relative to the hydroxyl group. The two functionalities are para to each other on 4-HPM, which The observed ease for 2eliminates the possibility for interactions. The imide and hydroxyl group are ortho to each other on the 2-hydroxy-4-methylphenylmaleimide. hydroxy-4-methylphenylmaleimide degradation may be facilitated by the interaction between the hydroxyl and carbonyl oxygen on the imide group. This degradation, which produced volatile materials, prohibited the use of 2-hydroxy-4-methylphenylmaleimide monomer in void free networks. 179 Figure 5. 16. TGA monitoring the weight loss of 2-hydroxy-4-methylphenylmaleimide monomer as a function of temperature (10C/min, N2) 5.4. Conclusions 4-hydroxyphenylmaleimide and 2-hydroxy-4-methylphenylmaleimde were synthesized and characterized. 4-HPM was significantly more thermally stable than 2hydroxy-4-methylphenylmaleimide which begin to degrade even at 110C. 4-HPM was incorporated into 2,6-dimethylphenol endcapped cresol novolac oligomers. Cresol-coHPM oligomers prepared in different molecular weights were subsequently crosslinked with Epon 828 epoxy. ratios. The network properties and processability depended on the molecular weight of the cresol-co-HPM novolac oligomers and the novolac to epoxy Lower molecular weight novolac oligomers favored processability, whereas higher molecular weight novolac oligomers led to superior network properties. Network composition also played an important role since compositions high in novolac exhibited enhanced flame retardance, but a sufficient amount of epoxy was required to form fully crosslinked networks with good physical integrity. The 1250 g/mol cresol-co-HPM novolac/Epon 828 network (70:30 wt:wt ratio) presented a good compromise by which networks with excellent properties could be processed easily. 180 6. Latent Initiators for Novolac/Epoxy Cure Reactions 6.1. Introduction Composites are becoming increasingly more important in replacing metals and concrete in structural applications due to their excellent oxidative stability and freeze thaw durability. One factor governing the feasibility of mass production is processability, which may comprise up to 80 percent of the final cost. Therefore, efficient processing is essential in producing cost effective composites. Novolac/epoxy networks are investigated in this research as matrix materials for tough, flame retardant composites. The mechanism for triphenylphosphine (TPP) catalyzed phenolic novolac/epoxy cure reactions was proposed to consist of two steps ( Figure 6. 1).178 The reaction initiates with a triphenylphosphine attack on an epoxy ring to form a zwitterion (A). In the presence of phenolic hydroxyl groups, the zwitterion undergoes rapid proton transfer to generate a phenolate anion and an alcohol (B). The propagation step consists of phenolate anion attacking another epoxy group (C) or an electrophilic carbon next to the triphenylphosphine to form a linkage and regenerate the catalyst. These steps are repeated until a crosslinked network is formed. 178 R. W. Biernath and D. S. Soane, Cure Kinetics of Epoxy Cresol Novolac Encapsulants for Microelectronics Packaging, Contemporary Topics in Polymer Science, J. C. Salamone and J. S. Riffle Eds., 7, 130-159 (1992). 181 A) O R O P O O R CH2 CH CH2 P B) OH CH2 OH CH2 n Proton Transfer OH CH2 OH CH2 n O O OH R CH2 CH CH2 P OH O O R CH2 CH CH2 P C) OH CH2 OH CH2 n O O R O O OH CH2 OH CH2 n OH CH2 OH OH CH2 OH CH2 n O CH2 CH CH2 R OH + CH2 O OH CH2 n O CH2 CH CH2 R O OH OH CH2 n O Figure 6. 1. Mechanism of TPP catalyzed phenolic novolac/epoxy reaction 182 One method for composite fabrication is pultrusion (Figure 6. 3) where spools of carbon fiber are pulled through a bath containing a low viscosity resin or the resin is poured onto the fibers. The fiber embedded in resin then enters a heated metal mold where rapid cure proceeds. Regardless of the fabrication method, a low viscosity resin (< 10 Pa*s) is a requisite for the wetting of the fibers. Once the resin reaches the heated metal mold, fast cure (99% conversion within 5 to 10 minutes) is ideal. Figure 6. 3. Diagram of pultrusion processing The major setback thus far with these phenolic novolac/epoxy systems has been optimizing the processing conditions such that long processing time windows and short cure times are obtained. Phenolic novolac/epoxy mixtures (65:35 wt:wt ratio) must be heated to approximately 140C to reduce the viscosity to the processable range (2-10 Pa*s). At this temperature, the resin mixture reacts slowly and remains processable for approximately 100 minutes. However, curing uninitiated phenolic novolac/epoxy mixtures requires one hour at 240C. The cure rates can be greatly accelerated at reduced temperatures with the addition of an initiator such as TPP. However, adding as little as 183 0.07 mole % TPP (based on moles of epoxy) reduces the processing window to approximately 10 minutes 140C while the cure still required 2 hours at 180C.179 One approach to overcome this problem is to place the catalyst directly onto the fiber so that the resin is not exposed to the catalyst until just prior to cure. An advantage for this approach is that high concentrations of catalyst can be loaded onto the fiber. A salt of Ultem type poly(amic acid) and tris(2,4,6-trimethoxylphenylphosphine) (TTMPP) has been used as sizing materials for carbon fiber (Figure 6. 4).180 TTMPP was used as opposed to TPP since the more basic TTMPP was necessary for the salt formation. The salt is stable under ambient conditions but imidizes rapidly at elevated temperatures to release the tertiary phosphine catalyst. Studies have shown that rapid cure rates can be achieved using this approach. However, the high cost associated with this process and raw materials limit its feasibility. Therefore, the key issue remains that a latent catalyst, which can be added directly into the resin mixtures, must be developed to efficiently process these composites. 179 C. S. Tyberg, K. Bergeron, M. Sankarapandian, P. Shih, A. C. Loos, D. A. Dillard, J. E. McGrath, and J. S. Riffle, Structure Property Relationships of Void Free Phenolic Epoxy Matrix Materials, Polymer 41(13), 5053-5062 (2000). 180 C. S. Tyberg, P. Shih, K. N. E. Verghese, A. C. Loos, J. J. Lesko, and J. S. Riffle, Latent Nucleophilic Initiator for Melt Processing Phenolic-Epoxy Matrix Composites, Polymer 41(26), 9033-9048 (2000). 184 O H N O OCH3 H3CO 3 O O O H N O n O PH H 3CO o OCH3 O 3 PH OCH3 OCH3 260 C ~3 minutes - H2O O N O O O O N + H3CO OCH3 P 3 OCH3 n O Figure 6. 4. High temperature imidization of PAAS to release TTMPP catalyst The use of latent catalysts or initiators have been somewhat explored. Latent catalysts, which form active species by external stimulation such as heating or photoirradiation to catalyze or initiate the polymerization or curing, enhance the pot life and the handling of the thermosetting resins, or increase the processability of the resins. Latent catalysts comprised of primary or secondary amine salts of a strong acid have been added to phenolic resole resins to extend storage stabilities while retaining rapid cure characteristics of typical acid catalyzed resole resins.181 These resins can be cured in the presence of a strong acid to enhance the reaction rate without heating to excessively high temperatures. However, the pot life of such compositions suffers in the presence of strong acid catalysts. Strong acids, especially sulfonic acids, were therefore mixed with amines to form latent catalysts which dissociate at elevated temperature to regenerate the acids. The cure rates in the presence of an amine salt of a strong acid were found to be comparable to those of the free strong acid catalyzed resole reactions; the pot life, on the other hand, was extended relative to that of the free strong acid catalyzed resoles. 181 U.S. Patent 5,344,909 to D. A. Jutching, T. M. McVay, and R. F. Pennock Latent Catalyzed Phenolic Resole Resin Composition, Sep. 6, 1994. 185 A cationic salt, N-benzylpyrazinium hexafluoroantimonate (BPH) (Figure 6. 5), was investigated as a latent thermal catalyst for epoxy/phenolic novolac reactions.182 BPH was considered as a latent catalyst since no appreciable amount of reaction occurred below 100C in its presence. An autocatalytic reaction kinetic mechanism was proposed to occur for the epoxy/phenolic novolac/BPH cure. Dynamic differential scanning calorimetry showed that an increased phenolic novolac content resulted in decreased initial cure temperatures and therefore reduced the cure times. CH2 N N Figure 6. 5. The chemical structure of N-benyzlpyrazinium hexafluoroantimonate SbF6 Microencapsulated reactant and/or platinum or tin catalysts were utilized in storage stable hydrosilation reactions.183 Mixtures of organohydrogensiloxane reactants and catalyst are only stable under ambient conditions for a few hours. Encapsulating the reactant or the catalyst in a thermoplastic organic polymer, which has a Tg higher than room temperature but lower than the cure temperature, effectively isolates the two under storage conditions. The reagents are liberated to allow rapid mixin...

SLGETD

End of Preview

Sign up now to access the rest of the document