"Forestry, Faculty of"@en . "DSpace"@en . "UBCV"@en . "Ellis, Simon Colin"@en . "2010-10-11T17:05:12Z"@en . "1989"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "Powdered adhesive resins based on polymers of phenol and formaldehyde are prevalent in the Canadian waferboard industry. In the development of faster curing resins, problems associated with low levels of resin flow and subsequent penetration into wood have been encountered. Little is known about how resin characteristics affect flow. This study involved the analysis of eleven samples of commercial resins and six produced in the laboratory. Molecular weight distributions and functional group characteristics were determined and related to flow characteristics. Two-ply parallel laminate boards were produced from Picea glauca (Moench.) Voss and Populus tremuloides Michx. using four resins of each type. Samples from these boards were tested in tension shear to produce strength and wood failure values. Resin penetration was evaluated using electron and light microscopy. The resins produced in the laboratory incorporated a raeta-bromophenol label. This allowed the distribution of the resin to be determined at the glueline using SEM/WDX.\r\nAll the resins were of the resol type and contained an average of 0.38 to 1.32 methylol groups per phenol ring. Resins containing greater proportions of low molecular weight species generally exhibited more flow which tended to occur at lower temperatures than in higher molecular weight resins. A resin sample which had been stored for over seven years exhibited greatly reduced flow compared to fresh samples of the same resin.\r\nFor the laboratory resins, the extent of penetration appeared to be the limiting factor in the quality of the bond formed only for the highest molecular weight resin. An optimum number average molecular weight was identified. Above this value, insufficient low molecular weight species were present to bring about good bond formation. Below this value excess low molecular weight species lengthened the time required for cure.\r\nThe quality of the bonds as determined from percentage wood failure values was most dependent on pressing time and wood species, less dependent on resin type, with the moisture content of the wood stock having the least effect.\r\nThe wood failure values obtained with the highest molecular weight laboratory resin increased with higher wood moisture contents. This effect was believed to be due, in part, to an increase in resin flow properties in addition to a more rapid heat transfer to the glueline.\r\nSpecies anatomy was observed to influence the extent to which the\r\nresins flowed down the lumens of the wood cells away from the glueline. In\r\naspen gluelines, resin was observed to have flowed in the longitudinal direction down the lumens of vessels elements to a distance of 300 \u00CE\u00BCm from the glueline on the cross-section. Flow down the lumens of longitudinal tracheids in spruce confined the resin to the three or four rows of cells immediately adjacent to the glueline (within approximately 100 of the glueline on the cross-section). The greater retention of resin at the glueline of the spruce bonds manifested itself as higher wood failure values at all pressing conditions.\r\nThis study indicated possible directions that designs of future resin systems might follow."@en . "https://circle.library.ubc.ca/rest/handle/2429/29095?expand=metadata"@en . "SOME FACTORS AFFECTING THE FLOW AND PENETRATION OF POWDERED PHENOLIC RESINS INTO WOOD By SIMON COLIN ELLIS B.Sc.(Hons), University of Wales (Bangor), 1983 M.Sc, University of British Columbia, 1986 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS OF THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Harvesting and Wood Science) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA July, 1989 \u00C2\u00A9 Simon Colin Ellis, 1989 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of HavNW^ Vcrya orjL U^QOOV Saidscg. The University of British Columbia Vancouver, Canada DE-6 (2/88) A B S T R A C T Powdered adhesive resins based on polymers of phenol and formaldehyde are prevalent i n the Canadian waferboard industry. In the development of faster curing resins, problems associated with low levels of resin flow and subsequent penetration into wood have been encountered. Li t t l e is known about how resin characteristics affect flow. This study involved the analysis of eleven samples of commercial resins and six produced i n the laboratory. Molecu la r weight distributions and functional group characteristics were determined and related to flow characteristics. Two-ply paral lel laminate boards were produced from Picea glauca (Moench.) Voss and Populus tremuloides Michx . using four resins of each type. Samples from these boards were tested i n tension shear to produce strength and wood failure values. Res in penetration was evaluated using electron and light microscopy. The resins produced i n the laboratory incorporated a raeta-bromophenol label. This allowed the distribution of the resin to be determined at the glueline using S E M / W D X . A l l the resins were of the resol type and contained an average of 0.38 to 1.32 methylol groups per phenol ring. Resins containing greater proportions of low molecular weight species generally exhibited more flow which tended to occur at lower temperatures than i n higher molecular weight resins. A resin sample which had been stored for over seven years exhibited greatly reduced flow compared to fresh samples of the same resin. F o r the laboratory resins, the extent of penetration appeared to be the l imit ing factor i n the quality of the bond formed only for the highest molecular weight resin. A n opt imum number average molecular weight was identified. A b o v e this value, insufficient low molecular weight species were present to bring about good bond formation. Be low this value excess low molecular weight species lengthened the time required for cure. The quality of the bonds as determined from percentage wood failure values was most dependent on pressing time and wood species, less dependent on resin type, with the moisture content of the wood stock having the least effect. The wood failure values obtained with the highest molecular weight laboratory resin increased with higher wood moisture contents. This effect was believed to be due, i n part, to an increase i n resin flow properties i n addit ion to a more rapid heat transfer to the glueline. Species anatomy was observed to influence the extent to which the resins flowed down the lumens of the wood cells away from the glueline. In t aspen gluelines, resin was observed to have flowed i n the longitudinal direction down the lumens of vessels elements to a distance of 300 [im from the glueline on the cross-section. F l o w down the lumens of longitudinal tracheids i n spruce confined the resin to the three or four rows of cells immediately adjacent to the glueline (within approximately 100 of the glueline on the cross-section). The greater retention of resin at the glueline of the spruce bonds manifested itself as higher wood failure values at a l l pressing conditions. This study indicated possible directions that designs of future resin systems might follow. i v TABLE OF CONTENTS A B S T R A C T i i T A B L E O F C O N T E N T S iv L I S T O F F I G U R E S v i i i L I S T O F T A B L E S xiv A C K N O W L E D G E M E N T S xvi A B B R E V I A T I O N S U S E D xvi i D E D I C A T I O N xvi i i Q U O T A T I O N xix 1. I N T R O D U C T I O N 1 2. L I T E R A T U R E R E V I E W 5 2.1. Res in Chemistry 5 2.1.1. Res in preparation 5 2.1.2. Res in curing.... 12 2.2. Waferboard Resins 17 2.3. Analy t ica l Techniques 19 2.3.1. Introduction 19 2.3.2. Infrared absorbance spectroscopy 21 2.3.3. Proton magnetic resonance spectroscopy 23 2.3.4. G e l permeation chromatography 24 2.3.5. Thermal mechanical analysis 26 2.3.6. Fus ion diameter 27 2.3.7. Stroke cure 27 2.4. Res in Chemistry and Adhesive Performance 28 2.5. Res in F l o w 34 V 2.6. Detect ion of Res in Penetration into W o o d 40 2.6.1. Scanning electron microscopy and analysis of x-rays 44 3. M E T H O D O L O G Y 46 3.1. Resins 46 3.1.1. Laboratory resins 46 3.1.1.1. Synthesis 46 3.1.1.2. Dry ing '. 48 3.1.2 Commerc ia l resins 49 3.2. Res in Analysis 50 3.2.1. Infrared absorbance spectroscopy 50 3.2.2. Proton magnetic resonance spectroscopy 50 3.2.2.1. Sample preparation 50 3.2.2.2. Calculations 51 3.2.3. G e l permeation chromatography...; 55 3.2.3.1. Instrumentation 55 3.2.3.2. Cal ibra t ion 56 3.2.3.3. Sample analysis 58 3.2.4. Addit ives 58 3.2.4.1. Hexamethylenetetramine 58 3.2.4.2. Inclusions 59 3.2.5. Thermal mechanical analysis 59 3.2.6. Fus ion diameter 60 3.2.7. Stroke cure 60 3.3. W o o d Bonding Study 60 3.3.1. Preparation of veneer 62 3.3.2. G l u i n g procedure 62 v i 3.3.3. Sample preparation and testing 63 3.4. Detect ion of Res in Penetration... 65 3.4.1. S E M / W D X 65 3.4.2. Photomicroscopy \u00E2\u0080\u00A2 67 4. R E S U L T S A N D D I S C U S S I O N 68 4.1. Res in Synthesis ; 68 4.2. Res in Properties 70 4.2.1. Infrared absorbance spectroscopy 70 4.2.2. Proton magnetic resonance spectroscopy 74 4.2.3. G e l permeation chromatography 78 4.2.4. Additives.. . . 95 4.2.4.1. Hexamethylenetetramine 95 4.2.4.2. Inclusions 97 4.2.5. Thermal mechanical analysis 99 4.2.5.1. Laboratory resins 99 4.2.5.2. Commerc ia l resins 103 4.2.6. Fus ion diameter 108 4.2.7. Stroke cure H O 4.2.8. Summary of resin properties 112 4.3. W o o d Bonding Study 115 4.3.1. W o o d failure values 117 4.3.1.1. Laboratory resins 119 4.3.1.2. Commerc ia l resins 123 4.3.1.3. Discussion 127 4.3.2. Tension shear strengths 138 4.4. Detect ion of Res in Penetration 140 4.4.1. E lec t ron microscopy 140 v i i 4.4.2. Photomicroscopy 173 4.5. Summary of Res in Penetration and W o o d Bonding 182 5. S U M M A R Y A N D C O N C L U S I O N S 191 5.1. Future Developments 193 6. L I T E R A T U R E C I T E D 196 A P P E N D I X . . . . 202 L I S T O F F I G U R E S Figure 1. Dependence of flow of novolak resins upon molecular weight 36 Figure 2. F l o w process of an activated complex 39 Figure 3. Diagram of apparatus used i n preparation of laboratory resins 47 Figure 4. A representative ^ - n u c l e a r magnetic resonance spectrum of a phenolic resin, showing peak assignments ...52 Figure 5. G e l permeation chromatography system 57 Figure 6. Dimensions of the shear specimens. 64 Figure 7. Temperature and viscosity plots against t ime for the synthesis of the laboratory resins 69 Figure 8. I R spectra of (a) Cook #5 and (b) C o o k #3 71 Figure 9. I R spectra of (a) Res in F 2 and (b) Res in D 2 72 Figure 10. I R spectra of (a) Hexamethylenetetramine and (b) Cook #5 containing 8% Hexamethylenetetramine 73 Figure 11. Cal ibra t ion curve for the G P C columns, using polystyrene standards 79 Figure 12. (a) Molecu la r weight distribution and (b) G P C detector response for Res in A 80 Figure 13. G P C detector responses for the first series of laboratory cooks 84 Figure 14. G P C detector responses for the second series of laboratory cooks 85 Figure 15. G P C detector responses for C o o k #5 (a) freeze-dried and (b) spray-dried 87 Figure 16. G P C detector responses for commercial resins; Res in A , Res in B and Res in C 89 Figure 17. G P C detector responses for commercial resins; Res in D I , Res in D 2 and Res in D3 . . . .90 Figure 18. G P C detector responses for commercial resins; Res in E , Res in F l and Res in F 2 91 Figure 19. G P C detector responses for commercial resins; Res in G and Res in H 92 i x Figure 20. Effective molecular diameters of phenolic resin molecules 94 Figure 21. I R spectrum of inclusions from Res in F 2 98 Figure 22. T M A profiles for the first series of laboratory cooks 100 Figure 23. T M A profiles for the second series of laboratory cooks 101 Figure 24. T M A profiles for commercial resins; Res in A Res in B and Res in C 104 Figure 25. T M A profiles for commercial resins; Res in D l , Res in D 2 and R e s i n D 3 105 Figure 26. T M A profiles for commercial resins; Res in E , Res in F l and Res in F 2 106 Figure 27. T M A profiles for commercial resins; Res in G and Res in H 107 Figure 28. Fus ion diameter against number average molecular weight.. 109 Figure 29. Fus ion diameter against weight average molecular we igh t . . . . I l l Figure 30. W o o d failure against moisture content for aspen bonded with laboratory resins. a) 75 seconds b) 90 seconds c) 120 seconds d) 240 seconds 120 Figure 31. W o o d failure against moisture content for spruce bonded with laboratory resins. a) 75 seconds b) 90 seconds c) 120 seconds d) 240 seconds 123 Figure 32. W o o d failure against moisture content for aspen bonded with commercial resin. a) 75 seconds b) 90 seconds c) 120 seconds d) 240 seconds '. 125 Figure 33. W o o d failure against moisture content for spruce bonded with commercial resin. a) 75 seconds b) 90 seconds c) 120 seconds d) 240 seconds 126 X Figure 34. Temperature rise at the gluelines of spruce boards pressed at different moisture contents 130 Figure 35. W o o d failure against moisture content for aspen bonded with laboratory resins. a) C o o k #4 b) C o o k #5 c) C o o k #6 d) Cook #3 134 Figure 36. W o o d failure against moisture content for spruce bonded with laboratory resins. a) C o o k #4 b) C o o k #5 c) C o o k #6 d) C o o k #3 135 Figure 37. W o o d failure against moisture content for aspen bonded with commercial resins. a) Res in B b) Res in D 2 c) Res in F 2 d) Res in G 136 Figure 38. W o o d failure against moisture content for spruce bonded with commercial resins. a) Res in B b) Res in D 2 c) Res in F 2 d) Res in G 137 Figure 39. Electronmicrograph, line scan and dot map for the molecular weight series, aspen. Species - A s p e n Moisture content - 6.7% Pressing time - 240 seconds Res in - C o o k #4 142 Figure 40. Electronmicrograph, l ine scan and dot map for the molecular weight series, aspen. Species - A s p e n Moisture content - 6.7% Pressing time - 240 seconds Res in - C o o k #5 144 Figure 41. Electronmicrograph, l ine scan and dot map for the molecular weight series, aspen. Species - A s p e n Moisture content - 6.7% Pressing time - 240 seconds Res in - Cook #6 146 x i Figure 42. Electronmicrograph, l ine scan and dot map for the molecular weight series, aspen. Species - A s p e n Mois ture content - 6.7% Pressing time - 240 seconds Res in - C o o k #3 148 Figure 43. Electronmicrographs and line scans for the molecular weight series, spruce. Species - Spruce Moisture content - 7.0% Pressing time - 240 seconds Res in - a) C o o k #4 b) C o o k #5 151 Figure 44. Electronmicrographs and line scans for the molecular weight series, spruce. Species - Spruce Moisture content - 7.0% Pressing time - 240 seconds Res in - a) C o o k #6 b) C o o k #3 153 Figure 45. Electronmicrographs and line scans for the low molecular weight resin, pressing time series. Species - Spruce Moisture content - 7.0% Res in - Cook #4 Pressing time - a) 90 seconds b) 120 seconds 156 Figure 46. Electronmicrograph and line scan for the low molecular weight resin, pressing time series. Species - Spruce Mois ture content - 7.0% Res in - C o o k #4 Pressing time - 240 seconds 158 Figure 47. Electronmicrographs and line scans for the high molecular weight resin, pressing time series. Species - A s p e n Mois ture content - 6.7% R e s i n - C o o k #3 Pressing time - a) 75 seconds b) 90seconds. 161 Figure 48. Electronmicrographs and line scans for the high molecular weight resin, pressing time series. Species - A s p e n Moisture content - 6.7% Res in - C o o k #3 Pressing time - a) 120 seconds b) 240 seconds 163 x i i Figure 49. Electronrnicrographs and line scans at high magnification, showing the extent of cel l wal l penetration. Species - A s p e n Mois ture content - 6.7% Pressing time - 240 seconds Res in - a) C o o k #4 b) C o o k #3 166 Figure 50. Molecu la r Weight Distributions for C o o k #4 and C o o k #3 167 Figure 51. Electronrnicrographs of types of bond failure. Species - Spruce Moisture content - 7.0% Pressing time - 240 seconds Res in - a) C o o k #3 b) C o o k #4 c) C o o k #3 d) C o o k #3 170 Figure 52. Electronrnicrographs for the commercial resin series. Species - Spruce Moisture content - - 7.0% Pressing time - 240 seconds Res in - a) R e s i n B b) R e s i n D 2 c) Res in F 2 d) Res in G 172 Figure 53. Photomicrographs of gluelines i n aspen bonded with laboratory resins (xl25). Species - A s p e n Moisture content - 6.7% Pressing time - 240 seconds Res in a) C o o k #4 b) C o o k #5 c) C o o k #6 d) C o o k #3 175 Figure 54. Photomicrographs of gluelines i n aspen bonded with commercial resins (xl25). Species - A s p e n Moisture content - 6.7% Pressing time - 240 seconds Res in a) Res in B b) Res in D 2 c) Res in F 2 d) Res in G 177 x i i i Figure 55. Photomicrographs of gluelines i n spruce bonded wi th laboratory resins (xl25). Species - Spruce Mois ture content - 7.0% Pressing time - 240 seconds Res in a) C o o k #4 b) C o o k #5 c) C o o k #6 d) Cook #3 179 Figure 56. Photomicrographs of gluelines i n spruce bonded with commercial resins (xl25). Species - Spruce Moisture content - 7.0% Pressing time - 240 seconds Res in a) R e s i n B b) R e s i n D 2 c) Res in F 2 d) Res in G 181 Figure 57. W o o d failure against number average molecular weight for both species bonded with laboratory resins tested i n the soaked condition, a} 75 seconds b) 90 seconds c) 120 seconds d) 240 seconds 184 Figure 58. W o o d failure against number average molecular weight for both species bonded with commercial resins tested i n the soaked condition. a) 75 seconds b) 90 seconds c) 120 seconds d) 240 seconds 186 x i v L I S T O F T A B L E S Table 1. Assignments for infrared absorption bands 22 Table 2. Summary of commercial resins 49 Table 3. Assignments for ^ - N M R peaks 53 Table 4. Percentage resin solids, final viscosities and pH's of laboratory resins 68 Table 5. Funct ional groups present i n laboratory resins 75 Table 6. Funct ional groups present i n commercial resins 77 Table 7. Theoretical molecular weights of acetylated resins 81 Table 8. Molecu la r weights of acetylated laboratory resins 83 Table 9. Molecu la r weights of acetylated commercial resins 88 Table 10. H M T A content of commercial resins 95 Table 11. p H values of commercial resins 96 Table 12. Corrected H M T A content of commercial resins 97 Table 13. Stroke cure times of the commercial resins 110 Table 14. Summary of laboratory resins' properties 113 Table 15. Summary of commercial resins' properties 114 Table 16. E q u i l i b r i u m moisture contents of spruce and aspen after conditioning i n different environments 116 Table 17. Analysis of variance tables for transformed wood failure values for both species tested in the soaked condit ion 118 Table 18. Analysis of variance tables for transformed wood failure values for both species bonded with laboratory resins tested i n the soaked condition 120 Table 19. Analysis of variance tables for transformed wood failure values for both species bonded with commercia l resins tested i n the soaked condition 124 Table 20. Average tension shear strenghts (calculated from data from both laboratory and commercial resins) 139 XV Table A - l . Shear strengths and percentage wood failure values for spruce specimens tested i n the dry condit ion (laboratory resins) 203 Table A - 2 . Shear strengths and percentage wood failure values for aspen specimens tested i n the dry condit ion (laboratory resins) 204 Table A - 3 . Shear strengths and percentage wood failure values for spruce specimens tested i n the dry condit ion (commercial resins) 205 Table A - 4 . Shear strengths and percentage wood failure values for aspen specimens tested i n the dry condit ion (commercial resins) 206 Table A - 5 . Shear strengths and percentage wood failure values for spruce specimens tested i n the soaked condit ion (laboratory resins) 207 Table A - 6 . Shear strengths and percentage wood failure values for aspen specimens tested i n the soaked condit ion (laboratory resins) 208 Table A - 7 . Shear strengths and percentage wood failure values for spruce specimens tested i n the soaked condit ion (commercial resin) 209 Table A - 8 . Shear strengths and percentage wood failure values for aspen specimens tested i n the soaked condit ion (commercial resins) 210 Table A - 9 . Shear strengths and percentage wood failure values for spruce specimens tested i n the boi led condit ion (laboratory resins) 211 Table A-10 . Shear strengths and percentage wood failure values for aspen specimens tested i n the boi led condit ion (laboratory resins) 212 Table A - l l . Shear strengths and percentage wood failure values for spruce specimens tested i n the boi led condit ion (commercial resins) 213 Table A-12 . Shear strengths and percentage wood failure values for aspen specimens tested i n the boi led condit ion (commercial resins) 214 x v i A C K N O W L E D G E M E N T S I would l ike to thank D r . Pau l Steiner, Faculty of Forestry, U B C , for his invaluable direction and advice throughout this project. I also wish to thank D r . Laszlo Paszner, Faculty of Forestry, U B C , for his support and guidance during my graduate studies at U B C . Thanks also go to D e a n Rober t Kennedy, Faculty of Forestry U B C , for his encouragement and suggestions. The assistance of Dermot McCar thy i n unravelling some of the mysteries of computers and the late G r e g Bohnenkamp for his many useful suggestions and practical help are very readily acknowledged. The electron microscopy was performed i n the Metal lurgy Department, U B C , and the photomicroscopy work was performed i n the Plant Sciences Department, U B C with the k ind permission and assistance of D r . M a r y Mager and D r . Rober t Copeman respectively. I would also l ike to thank Forintek Canada Corpora t ion for their financial support of my P h D studies and this project. The help of the staff i n the Composites Section at Forintek, especially A x e l Anderson, with the treatment of the shear specimens is also appreciated. Final ly , my greatest gratitude goes to my parents for their continuous support, encouragement and enthusiasm for my educational studies. x v i i ABBREVIATIONS USED A N O V A analysis of variance D S C differential scanning calorimetry E D X energy disperive x-ray analysis F : P formaldehyde:phenol G P C gel permeation chromatography H M T A hexamethylenetetramine I B internal bond I R infrared M n number average molecular weight M O R modulus of rupture M w weight average molecular weight N M R nuclear magnetic resonance O S B oriented strandboard P F phenol formaldehyde P M R proton magnetic resonance S E M scanning electron microscopy T E M transmission electron micrsocopy T g glass transition temperature T H F tetrahydrofuran T M A thermal mechanical analysis U V / V I S ultraviolet/visible W D X wavelength dispersive analysis x v i i i T o my parents, M a r y and C o l i n . x i x \"Examinations, sir, are pure humbug from beginning to end. If a man is a gentleman, he knows quite enough, and i f he is not a gentleman, whatever he knows is bad for him.\" Oscar W i l d e The Picture of D o r i a n Gray (1890). 1 1. I N T R O D U C T I O N The reaction between phenols and aldehydes to form resinous materials has been recognized since the end of the last century. V o n Bayer (1872) was the first to study the reaction between phenol and formaldehyde and around the turn of the century there was some evaluation of the materials produced as electrical insulators. The first successful commercial exploitation of phenol-formaldehyde products was by Baekeland (1907, 1909). Since 1910 there has been no one contribution to the chemistry of phenolic resins as great as that of Baekeland. Consequently the development of phenolic resins has tended to be marked more by steady and continuous progress than by outstanding milestones. Considering that phenolic resins were developed commercially as early as 1907 it might be expected that their chemical and physical properties would have been completely elucidated by now, but even today some of their characteristics are not fully understood. One reason for this apparent lack of knowledge is the variety of isomerides with different chain lengths that may be formed from the condensation of phenol and formaldehyde. Polyfunctional phenols may react wi th formaldehyde i n both the ortho- and para-positions to the hydroxyl group and their condensation products exist as numerous posit ional isomerides for any chain length. This makes the chemistry of the reactions very complex and study of the condensation reactions extremely difficult (P izz i 1983). It may be argued that a complete understanding of the chemistry of the resins is unnecessary since successful resins have been produced for over 70 years. However, a gradual increase i n the understanding of the reaction mechanisms and structures produced has helped i n the production of phenolic resins \"tailor-made\" for specific applications. Further knowledge of the chemistry and structure w i l l enable adhesive chemists to better 2 understand existing properties of resins available today and to p lan their strategies for producing the resins of the future. Waferboard and oriented strandboard are structural panels products manufactured from primari ly hardwood resources, e.g. aspen i n Canada and northern central U . S . A . . The 1987 production of these panels i n Canada was 1.6 m i l l i o n m 3 , valued at $227 mi l l ion (Statistics Canada 1988). Thir teen mills are currently producing these panels i n Canada. Phenol-formaldehyde is used as the binder i n these panels and i n Canada the resin tends to be used i n the powdered form. A recent trend i n the manufacture of powdered phenolic resins has been to develop resins with higher molecular weight ranges. Since the chemical structure of these resins is closer to the final cured, cross-linked structure than lower molecular weight formulations, faster cure times are possible. F r o m the economic viewpoint of a waferboard m i l l , these highly reactive resins are an attractive development since they allow an increase i n the hot-press throughput. However, these resin modifications may adversely affect certain aspects of the bond formation process. Precure of the resin on the press platens may occur and insufficient flow of the resin and subsequent reduced penetration of the wood particles may result. It is this phenomenon of the ability of the powdered resin to melt and flow that may be the major factor governing the range of molecular weights that a suitable adhesive may encompass. Thus it is important that the key factors involved i n the flow processes of powdered phenolic resins are identified. Al though adequate commercial adhesives are available now, advancements i n powdered resin technology may be hampered by the l imited knowledge of such factors. If controlling variables can be identified and their effects understood it may be possible to produce adhesive formulations which 3 w i l l achieve the desired goal of minimizing pressing time while at the same time ameliorating some of the problems which have arisen. W h i l e work has appeared i n the scientific and technical literature investigating the effects of resin characteristics on the performance of adhesives (Wi lson 1979, G o l l o b 1983, Peterson 1985) this work has dealt solely with l iqu id adhesives. This is primari ly a result of the shorter time for which powdered adhesives have been used. The measurement of the penetration of l iqu id adhesives into wood, both on the gross level and at the cel l wa l l level, has been attempted using a variety of techniques. These have met with various levels of success and certain limitations have been recognized. However, observations from l iquid systems cannot be directly extended to powdered resins due to the different physical phases present i n the two types of systems. W i t h powdered resins, the solid particles must first melt before any wetting of the wood substrate can take place. Powdered resin systems are capable of more rapid curing times than l iqu id systems since water does not have to be removed before cure can be achieved as with l iqu id systems. This study undertook as its primary objective to determine the factors which affect the flow properties of phenolic resins and establish how different flow levels affect bonding characteristics. W h i l e it is recognized that the resin structures needed to produce faster setting resins generally reduce flow, no quantitative approach to resin flow assessment has been attempted. Measurement of flow tends to be by crudely empir ical observations or purely subjectively. A more objective measurement of resin flow and the variables affecting it is necessary i f a greater understanding of how these powdered resin systems behave i n pressing situations is to be developed. It is only with reliable information describing how resin 4 attributes affect fundamental properties such as reactivity, flow and penetration of wood that resin producers can better approach the full potential of phenolic resin systems. The overall study objective was addressed by considering three sub-objectives. The first sub-objective of this study was to produce powdered phenolic resins i n the laboratory which would exhibit similar physical characteristics to commercially available resins. These laboratory resins were to incorporate a bromine label i n order to al low the distinction between resin and wood substrate to be made using S E M / W D X , to describe resin distribution at the gluelines. The resins were to be produced i n a series of molecular weights. The second sub-objective was to develop sufficiently detailed descriptions of the chemical structures and flow characteristics of these resins, together with a number of commercially available resins, to determine the effects of molecular weight distribution and reactivity of the resins on their flow properties. The third sub-objective was to produce bonds between wood veneers using these resins. Observation of these bonds under both the electron and light microscope would show how the flow properties of the resins manifested themselves as penetration into the wood substrate. Testing of the bonds using tension shear specimens would allow the evaluation of flow and penetration as important parameters i n the development of strength. 5 2. L I T E R A T U R E R E V I E W 2.1. Res in Chemistry 2.1.1. Resin preparat ion The conditions under which reactions between phenols and aldehydes are carried out (temperature, p H , molar ratio, catalyst and reaction time) have a profound influence on the character of the products obtained. Two prepolymer resin types are generally recognized - resols and novolaks. The following definitions of these have been published (Whitehouse et al 1967). \"Resol - A synthetic resin produced from a phenol and an aldehyde. The molecule contains reactive methylol (or substituted methylol groups); heating causes the reactive resol molecules to condense together to form larger molecules, a result achieved without the addition of a substance containing reactive methylene (or substituted methylene) groups.\" \"Novolak - A soluble fusible synthetic resin produced from a phenol and an aldehyde, having no reactive methylol (or substituted methylol) groups in the molecule and therefore incapable of condensing with other novolak molecules on . heating without the addition of hardening agents.\" Resol resins are obtained from the reaction of a phenol with an aldehyde under alkaline conditions where the aldehyde is in molar excess. Typical phenokformaldehyde (P:F) ratios used are 1:1.0 to 1:3.0. In alkaline solution the phenol will be present as the phenoxide ion: OH O Derealization of an unshared pair of electrons on the oxygen atom results in increased electron densities at the ortho- and /;ara-positions, notably increasing the electron density of the benzene ring. 6 The reaction of phenol and formaldehyde i n alkaline solut ion therefore results i n the formation oiortho- and /?ara-methylol groups. The resulting brtho- and /?ara-methylol phenols are sti l l reactive towards formaldehyde and rapidly undergo further substitution with the formation of d i - and tri-substituted methylol derivatives. T h e possible products are shown below (Saunders 1973): OH OH CH2OH CH2OH III v -7 F o r example, when a mixture of phenol, formalin and sodium hydroxide (molar ratios 1:3:1) is heated at 30\u00C2\u00B0C for 5 hours, the composit ion of the resulting mixture is approximately (Freeman and Lewis 1954); I Phenol 3 % II 2-Methylolphenol 12 % m 4-Methylolphenol 17 % I V 2,4-Dimethylolphenol 24 % V , 2,6-Dimethylolphenol 7 % V I 2,4,6-Trimethylolphenol 37 % The composit ion of this mixture w i l l vary according to the relative molar ratios, reaction conditions and time of reaction. Freeman and Lewis (1954) graphed the formation and disappearance of each phenol alcohol formed from the addit ion of one, two or three methylol groups to phenol . The effect of substitutes i n the phenolic ring plays an important role i n determining the nature and rate of further substitution into the ring, aside from the fact that they l imit the number of positions available for reaction. The substitutes may be ring activating or deactivating and may direct substitution to particular positions i n the ring. In general, an ortho-, para-directing group located meta- to the phenolic hydroxyl group enhances the reactivity of the phenol (Mar t in 1956). Thus raeta-cresol is nearly three times as reactive towards formaldehyde than is phenol, whilst 3,5-xylenol is nearly eight times as reactive i n the early stages of reaction (Sprung 1941). If an ortho-, para- directing group is introduced ortho- or para- to the phenol hydroxyl group, the activation is not as noticeable as i n the meta- posit ion and i n fact the substituted phenol may be less reactive than phenol itself. A methylene group substituted between two phenols, however, greatly increases their reactivities. A l l meta- directing groups tend to deactivate the ring. It has been proposed (F inn and Lewis 1951) that with phenol having a substituent i n the para- or meta- position, the addition of a methylol group at 8 one ortho- posit ion greatly facilitates the further addi t ion of a second methylol group at the other ortho- position. These factors combine to make 2,6-dimethylolphenol the most reactive species i n a phenokformaldehyde reaction mixture, as evidenced by its low concentration i n the reactions studied by Freeman and Lewis (1954). The enhanced reactivity of methylol phenols over phenol itself w i l l lead to a mixture including d i - and tri-methylol phenols even i f the formaldehyde:phenol ratio is not that large and a propor t ion of unreacted phenol w i l l remain. Fundamenta l condensation reactions occurring between the methylol phenols are: OH OH It has been proposed that pathway II is the most important reaction under strongly alkaline conditions. Condensations may involve two methylol groups or one methylol group and a hydrogen atom at an unsubstituted ortho- or /jara-position on the methylol phenol . In the first case, a methylol group is subsequently el iminated as formaldehyde and i n the second case, a proton is subsequently eliminated. The ratio of these reactions and the 9 nature of the condensation products w i l l depend upon the structure of the methylolphenol involved. Genera l ly para-hydrogen atoms and para-methylol groups are the predominant sites for the self-condensation of methylolphenols (Saunders 1973). The reactions described above which lead to dinuclear phenols as shown may be repeated so that trinuclear phenols are formed and so on. Thus a complex mixture of mononuclear and polynuclear phenols l inked by methylene bridges results. Hydrogen bonding is believed to play some part i n the observed tendency of coupling between methylol phenols to occur through para-groups rather than ortho- groups i f both are available. The intramolecular hydrogen bonding between the oraio-methylol group and the phenol hydroxyl group stabilizes structure I. II I Al though intermolecular hydrogen bonding with the para-substituted methylol phenol is possible (II), it is not as common as the intramolecular variety (Richards and Thompson 1947) and thus the para-substituted methylol group is less stable than its ortho- counterpart. Novo lak resins are normally prepared under acidic conditions where there is a molar excess of phenol over formaldehyde (F :P 0.75:1-0.85:1). A different mechanism from that described for resols takes place. T h e in i t ia l step is the protonation of the formaldehyde to form an effective electrophil ic species. 1 0 H C=0 H H -> JZ=OH H H C-OH H The phenol undergoes electrophilic substitution with the formation of ortho- and /?ara-methylol groups. + The substitution reaction occurs slowly. In the formation of novolaks the further reaction of these phenol alcohols to form the 2,4- and 2,6-dimethylol phenols and the 2,4,6-trimethylol phenol is unlikely due to the deficiency of formaldehyde and the rapid speed of condensation of the phenol alcohols with more phenol. Under acidic conditions, benzylic carbonium ions result which rapidly react with phenol yielding dihydroxydiphenylmethane. OH 11 At low pHs (below pH 2) methylene bridge formation is favoured at the para-position over the ort/io-position, whilst at higher pHs (pH 3-6) the reverse holds true (Saunders 1973). The novolak molecule is built up from the dihydroxydiphenylmethanes by further addition of formaldehyde, followed immediately by condensation of the alcohol group thus formed with another phenol or polynuclear phenol molecule. This type of reaction leads to a linear compound of general formula H fC^^^OH^C ^ h-CgH^H and occasionally branched polymers in which some of the benzene rings have three methylene bridges attached. These types of reactions continue until all the formaldehyde has been consumed and the final mixture consists of a complex variety of polynuclear phenols linked by ortho- and para-methylene bridges. As the number of phenolic nuclei per molecule increases, the problems encountered in determining and separating an individual compound become very great. With reaction at only the ortho- and para-positions of phenol it is possible to form three two-ring, seven three-ring, twenty-one four-ring and fifty-seven five-ring compounds (Martin 1956). An average of 5-6 phenyl rings per molecule with a range of 2-13 would be typical of a novolak resin. A representative novolak resin molecule might be as illustrated below. HO OH OH 12 The essential feature of novolaks is that they represent a completed reaction and by themselves have no ability to continue increasing i n average molecular weight. This characteristic is i n contrast to resols which have reactive methylol groups and are capable of cross-linking reactions upon heating. T o convert novolak resins to network polymers the addit ion of an auxiliary cross-linking agent is required. 2.1.2. Resin curing The curing of resols involves a change from the relatively low molecular weight prepolymers to a highly-branched, cross-linked, three-dimensional network. The curing process is extremely complex, involving a number of competing reactions. Since the cured product is infusible and insoluble it is not amenable to chemical investigations. Thus much of the information regarding the curing mechanisms has been obtained from studying the reactions of more simple systems containing the types of functional groups present i n the prepolymers. In the presence of acid, two types of primary reaction between the methylol phenols have been established. A methylol group may react with an ortho- or para-hydrogen atom to form a methylene linkage or it may react with another methylol group to form an ether linkage. It would appear that the methylene linkages w i l l be favoured when a proport ion of the ortho- and para-positions are unsubstituted whereas ether linkages w i l l be favoured by a high degree of methylolation (Saunders 1973). 13 The ether linkages are then broken down but there is less agreement concerning the mechanisms involved here. Z i n k e et al (1951) concluded that the methylene ether linkages break down to form methylene bridges l iberating formaldehyde. The formaldehyde may then react with any available positions on the phenolic nuclei, with the methylene bridges or with the phenol ic hydroxyl groups. The following structures may occur: Hul tzsch (1950) disagreed, proposing that the ma in reaction i n the breakdown of ether linkages involves the formation of quinone methides. 14 Quinone methides are highly reactive molecules and may lead to a variety of structures (P izz i 1983). The dark colour often associated with cured P F material is thought to be due to compounds derived f rom quinone methides. A t temperatures between 130-180\u00C2\u00B0C cross-linking reactions of the types occurring v ia carbonium ions are more important whereas quinone methide formation and subsequent condensation is probably more important at higher temperatures (Knop and Scheib 1979). A possible structure of a cured resol network may be represented as below. The relative amounts of the linkages shown are not intended to have any quantitative significance. The chains are extremely irregular and their geometry precludes a large propor t ion of the potential cross-linking (Megson 1948). This is partly due to extremely l imi ted rotation about the methylene bridge (Pritchett 1949). O H O H O H rJ 15 Since novolak resins contain no reactive methylol groups, a cross-l inking agent must be introduced to bring about the network formation. Novolaks may be cross-linked with addit ional formaldehyde or paraform but hexamethylenetetramine ( H M T A ) is usually the additive chosen. N i n e to ten percent by weight of H M T A seems to be the most commonly employed level ( K n o p and Scheib 1979). H e r e again, the mechanism of curing is not fully understood but work with model compounds has indicated possible routes. The small amount of water that is always present in novolak resins leads to the hydrolysis of H M T A . H 2 o NH_ CH 2 OH rVH CH, I I I I C H 2 + NH CH 2 NH + NH(CH 2 OH) 2 N H 2 C H 2 O H N H \u00E2\u0080\u0094 C H 2 Carbon ium ions are generated from these -amino alcohols, which then react with phenol to give d i - and tri-benzylamines. W h e n these benzylamines are heated a variety of structures may arise such as: OH OH OH OH CH=N-CH 2 Thus it would appear that a possible reaction scheme is one where the primary reaction of the phenol structures with H M T A leads to a complex structure containing secondary and tertiary amine linkages. O n further small proport ion of nitrogen left i n the structure. A figure of six percent has been given for the amount of chemically bound nitrogen remaining i n the hardened novolak resin (Knop and Scheib 1979). The possible structure of a hardened novolak resin network is shown below. He re again the relative amounts of the links illustrated have no quantitative significance. T h e novolak network polymer can be seen to have predominantly the same structure as the resol network polymer. heating many of these linkages break down to give methylene linkages with a OH OH OH CH 2NH-CH 2, CH=N-CH2 C H 2 17 2.2. Waferboard Resins Ei ther solid or l iqu id resins may be used i n the construction of waferboard and oriented strandboard ( O S B ) . Sol id powdered phenolic resins may be produced by more than one technique. Pulverized phenolic resols and novolaks were the earliest manufactured. He re resin first is produced i n a batch process as a l iquid . The condensation product is disti l led under reduced pressure to remove water and excess monomer and the resin is poured onto a cooling belt flaker or onto a cooling floor to solidify. The solid resin chunks or crusts are then crushed, pulverized and mixed with other desirable additives such as a releasing agent, a powdered flow promoter and a hardener (usually H M T A ) . A disadvantage of this process is that a high level of free phenol is required during the manufacture i n order to allow for the dehydration of the aqueous condensation product and to obtain the molten resin without gelling i n the reactor. This high free phenol level causes such resins to have a tendency to stick to caul plates. Releasing agents can be added to alleviate this problem but cost considerations become a l imitation. Spray-drying may also be used to produce phenolic powders. L i q u i d resins of suitable formulation, usually of the resol type, are produced and then spray-dried using established technology. These resins have lower free phenol than the pulverized resins and are non-sticking. The spray-dried resins also tend to be more uniform than their counterparts prepared by pulverization since they can be produced using continuous operations rather than batch processes. M o r e recently, a novel approach has been used to produce powders. Phenol-formaldehyde resins have been prepared by an aqueous suspension polymerization process (Regina-Mazzuca et al 1982). The term \"phenolic 18 thermospheres\" has been applied to the products (Brode et al 1982). Resins of very uniform composition can be isolated as a free-flowing powder directly without the need for a pulverizing or spray-drying step. Precise end-point control is possible and the full range of conventional thermosetting and thermoplastic phenolics may be produced by this process. W h i l e l iqu id phenolic resins can also be used i n waferboard production, resin distribution problems are encountered i f resin levels similar to those used for powdered resins (2.0-2.5%) are attempted. Acceptable board properties can only be achieved with higher l iqu id resin levels. Very low solids content l iquid resins have been used i n an attempt to improve distribution but this introduces significant amounts of water into the furnish. Improved spraying technology may also increase the usage of l iqu id adhesives i n future. Powdered waferboard resins of different formulations are produced for use i n different parts of the board. A distinction is made between the face layers of the board and the core layer. Dur ing hot-pressing the core temperature w i l l not reach that of the face layers, which are closer to the platens. Thus core resins must be more rapid-setting. General ly, these core resins are of greater molecular weight than their face counterparts, so that fewer cross-linking reactions are required to attain the infusible resin structure. Some resins are produced which may be used throughout the construction of the board but the emphasis has tended to be placed on the use of two resin systems for waferboard construction. 19 2.3. Ana ly t ica l Techniques 2.3.1. Introduction The analysis of a thermosetting resin such as a phenol formaldehyde resin presents a number of difficulties. In the ini t ia l stages of polymerization the resin consists of relatively low molecular weight species to which certain analytical techniques may be applied. A s the resin becomes polymerized to a molecular weight range of a useful adhesive, the state of the resin of most interest to the wood scientist, some of these techniques may not be applicable. This may be due to the physical state of the sample, problems associated with dissolution of the sample or poor resolution of a spectroscopic method (Steiner 1975). In the cured state, thermosetting polymers are generally insoluble, infusible materials which are difficult to examine by many analytical techniques. This state results from their cross-l inked structure, which i n turn is responsible for many of their desirable physical properties. A s far as species formed during the preparation of a phenolic resin are concerned, it would be desirable to determine the structure and concentration of each component formed when phenol, formaldehyde and their derivatives react under any set of conditions (Woodbrey et al 1965). The ini t ia l polymerization stage has been studied i n some detail especially i n terms of reaction kinetics (Freeman and Lewis 1954) but as the molecular weight and complexity of the species present increase, one must resort to determining the relative amounts of various functional groups present. Wet chemistry methods can be applied to the measurement of certain functional groups such as the hydroxymethyl and ether-bridge content of phenolic resin. However, these methods provide no information about the number and posit ion of these groups i n relation to the number of phenolic nuclei and the 20 positions of the phenolic hydroxyl groups. Infrared absorption spectroscopy has been used and can provide information about a number of functional groups. However , most of the results obtained are only qualitative. Chromatographic techniques such as paper chromatography and thin layer chromatography have been used to study the early sequences i n the phenol/formaldehyde condensation reaction. G e l permeation chromatography provides an excellent method for determining the molecular weights of the components of a resin formulation. Separation of individual methylol phenols of low molecular weight is possible and while high molecular weight species may not be able to be completely separated, data on the average molecular weight of the mixture can be obtained (Armonas 1970, Wagner and K o p f 1971). Nuclear magnetic resonance spectroscopy has proved a most useful tool i n identifying the relative proportions of different functional groups present in a resin and how these may vary during the preparation of a resin. Techniques which have been applied to P F resins i n the fully-cured state include x-ray spectroscopy, pyrolysis gas chromatography, solid state N M R , infrared spectroscopy and thermal analysis. Problems arise since the exact nature of the preparation of commercial resins are not released by the manufacturers for propriety reasons. Thus although the chemical nature of the bulk adhesive formulation can be determined by spectroscopic methods, minor components, which may be included, are not so readily detected. These additives may be included at the spray-drying stage to aid efficient particle formation, or to aid release of the panel from the press platens or to act as flow promoters or cure promoters. Thus some of the additives may mask or enhance certain characteristics of the adhesive 21 formulation, although i n some cases this is the deliberate objective of the manufacturer, such as with the addition of a flow promoter. 2,3.2. Infrared absorbance spectroscopy Several articles have appeared i n the literature concerning the identification of phenolic resins by means of infrared ( IR) absorption spectroscopy. There has not always been complete agreement on the assignments of some of the various absorption bands (especially those associated with the substitution patterns of the benzene rings) but the technique has been demonstrated to be useful i n the characterization and identification of a range of phenolic resins. Richards and Thompson (1947) made an extensive study of the I R spectra of various substituted phenols and related compounds together with some phenolic resins. This work helped i n assigning substitution patterns i n the 11-15 fim region (900-650cm- 1). Burke et al (1956, 1959, 1960) used I R spectroscopy i n their studies of linear phenolic polymers. L inea r polymers were produced with various combinations of ortho-ortho', ortho-para' and para-para' linkages. Infrared spectroscopy was used to confirm the existence of desired structures. Most of their work concentrated on novolak resins and was performed using nujol mulls. Grisenthwaite and Hunter (1956) studied the I R spectra of 2,2'-dihydroxy-5,5'-dimethylphenylmethane and a number of its homologues, two para-cresol novolak resins, a trinuclear resol and related compounds. Assignments were made for a number of bands observed i n the spectra. In 1965 Secrest produced a thorough treatment of the work on I R spectroscopy of P F resins up to that date. H e reviewed the peak assignments of a number of authors and produced spectra of various types of phenolic resins which he described. Has l am and Wi l l i s (1965) presented a number of I R spectra for a range of thermosetting resins. Assignments for peaks 22 associated with resol and novolak P F resins were made and the effects of cross-linking on the spectra were discussed. Pshenitsyna and Kotre lev (1969) studied the reaction between phenol and hexamethylenetetramine ( H M T A ) using I R spectroscopy. Makarevich et al (1973) used I R spectroscopy to characterize a number of commercial resins and the same workers (Sushka et al 1973) studied the changes undergone during heating for various periods of time. H u m m e l (1974) discussed the influence of the substitution patterns of the benzene ring on the out-of-plane vibrations of the ring hydrogen atoms in phenolic resins. M o r e recently, Raczniak et al (1983) used I R spectroscopy i n connection with 1 H - N M R spectra to characterize a number of phenolic resins, both resol and novolak. Table 1 gives the consensus of opinion regarding the assignments of the peaks observed i n the I R spectra of both resols and novolaks. Table 1. Assignments for infrared absorption bands Wavelength (cm\" 1) Assignment 3300 O - H stretch 2900,2800 ^ C - H stretch 1700-1900 weak absorption bands characteristic of the substitution type 1600,1500 C - C stretch 1480 1,2,6 ; 1,2,4,6 substitution 1450 C - H bend 1370 A r O ^ I bend 1270 - C H 2 0 - H bend 1230 A r - O H stretch 1150 1,2,4 ; 1,2,6 in-plane bending 1090 - C H 2 - 0 - C H 2 - stretch 1010 - C H o - O H stretch 960 1,2,4; 1,2,6 in-plane bending 930 1,2,4 out-of-plane bending 880 1,2,4 ; 1,2,4,6 840 1,4 ; 1,2,4 810 1,4; 1,2,4 780 1,2,4 ; C - H rock 760 1,2; 1,2,6 690 1 (free phenol) 23 2.3.3. Proton magnetic resonance spectroscopy Proton magnetic resonance ( P M R ) spectroscopy has proved to be an extremely useful technique i n the analysis of the structure of phenolic resins. Hi rs t et al (1965) were amongst the first to demonstrate the potential of P M R with regards to phenol-formaldehyde polymers when they provided spectral assignments of various functional groups of previously we l l characterized linear phenol-formaldehyde polymers. They were able to distinguish between ortho- and para- positions on the aromatic ring, through the use of methyl- and chloro-substituted phenols. Woodbrey et al (1965) also studied the reaction between phenol and formaldehyde i n its early stages. They produced comprehensive spectral assignments for both acetylated and non-acetylated resins and gave quantitative data on the average number of various functional groups per phenol unit. They confirmed that acetylation, considered to be the best method of preparing the material for analysis, did not alter the basic structure of the resin. Szymansk and Bluemle (1965) investigated the use of arsenic trichloride as a solvent for P M R since the conventional N M R solvents such as C D C 1 3 , C C 1 3 and C S 2 do not readily solubilize underivatized phenolic resins. Spectral assignments for model compounds were given and primarily novolak resins were investigated. He re again, they were able to distinguish between ortho-and para-linkages. K o p f and Wagner (1973) used P M R to study intermediates i n the formation of novolak resins and their subsequent reaction with hexamethylenetetramine. Steiner (1975) studied acetylated phenol-formaldehyde wood adhesives. Comparisons were made of a series of resins produced with different ini t ial F : P charged ratios and different catalysts. The degree of polymerization, methylol content and aromatic protons remaining were among the parameters determined. A n estimate of 24 the in i t ia l F : P ratio could also be calculated from the spectra. G o l l o b (1983) also used P M R to characterize resins prepared under different conditions. Numbers of organic protons, methylol groups, dibenzyl ether bridges and benzyl-type hemiformal units per phenol unit were determined, i n addition to a value for the charged F : P ratio. 2.3.4. Gel permeation chromatography One of the most important descriptive parameters of any adhesive resin is that of its molecular weight. Often the term \"average molecular weight\" (based on a number average, a weight average or a \"z\" average) is used to describe the molecular structure. However, this term can be misleading because polymers of identical average molecular weights may have different molecular weight distributions. G e l permeation chromatography ( G P C ) is a relatively simple, rapid procedure for determining the molecular weight distribution of a resin. Polymer molecules are separated according to their effective molecular size i n solution using a column packing through which larger molecules are eluted earlier than smaller ones. Us ing a suitable calibration technique, this molecular size distribution may then be converted into a molecular weight distribution. Armonas (1970) presented chromatograms of phenol-formaldehyde and urea-formaldehyde resins and demonstrated how the molecular weight distribution changed during a resin cook. Wagner and K o p f (1971) also studied the change i n the molecular weight distribution of a sodium hydroxide-catalyzed resol cook. They identified some of the species present and the nature of the changes undergone by means of proton N M R . Yosh ikawa et al (1971) separated a variety of phenolic compounds using polystyrene gel and tetrahydrofuran ( T H F ) as the solvent. The difference between the calculated molecular volume of the phenolic compounds and 25 that obtained by actual determination with G P C was ascribed to T H F solvation of the phenolic hydroxyl group. Furthermore, it was found that T H F solvation was affected by the steric hindrance of ort/jo-substituted phenol and by inactivation of the phenolic hydroxyl group resulting from internal hydrogen bonding. D u v a l et al (1972) studied a variety of resols using G P C . The influence of various reaction parameters (the nature of the catalysts, the proport ion of the starting materials, the treatments undergone by the resols) on the resols' composition was determined. A n insight into the progress of the polycondensation reaction as a function of time was afforded and some information concerning the reactivity of different groups and unblocked ring positions was gained. They also discussed the possible conformations obtained by the various phenols upon solvation with T H F . A n internal standard technique was used by Tsuge et al (1973) to determine the free phenol present i n a l iquid resol resin. K i n g et al (1974) also used G P C , i n conjunction with N M R , to determine the structure of some l iquid resols. Wel lons and G o l l o b (1980) employed a low-angle laser light scattering detection method i n their G P C apparatus, instead of the more commonly applied ultra-violet or refractive index equipment. The values they obtained for the molecular weights of plywood resins were much greater than any reported by previous authors. They attributed this partly to incomplete solubility achieved i n some other determinations and to the inadequacy of previously used detection and calibration methods. However, the detection system used did not register molecules with molecular weights of less than 1000 Daltons. They used hexafluoro-iso-propanol as the solvent. Takeuchi et al (1983) employed m i c r o - G P C , whereby a more narrow diameter column than normal is used, to obtain a high resolution i n the separation of novolak resins. B a i n and Wagner (1984) solubilized high caustic l iqu id resol resins 26 using trichloroacetic acid such that they could be used i n a conventional T H F / a - S t y r a g e l system. 2.3.5. Thermal mechanical analysis Thermomechanical analysis ( T M A ) involves the measurement of the changes i n dimension of a sample as a function of its temperature. It can be used to determine the coefficient of expansion, glass transition temperature and softening point of polymers. Li t t l e information has been published i n the literature regarding the softening and subsequent flow of uncured solid P F resin as it is subjected to a temperature gradient. The relationship between the softening temperature of cured resins and their durability has been determined for a variety of resins by Chow (1973). Rosenberg (1978) examined the use of thermal softening to determine the degree of cure of a P F resin. This work was performed on l iquid resin which had been hardened by curing rather than on unreacted solid resin. Katovic and Stefanic (1985) determined the dimensional changes undergone by powdered novolak resins over the temperature range 30-140\u00C2\u00B0C. Two distinct regions in the rate of penetration of the T M A probe vs. temperature were noticed at a low sensitivity setting of the instrument. The first one, around 95\u00C2\u00B0C, was ascribed to the visually observed melting range of the resin, whereby the aggregation of powdered resin particles becomes melted into a continuous f i lm. The nature of the second region around 115\u00C2\u00B0C was not proposed. Us ing a higher sensitivity setting, a glass transition temperature, T , was observed by a change in the linear expansion coefficients at around 75\u00C2\u00B0C. This figure agreed we l l with the T value found by differential scanning calorimetry. 27 2.3.6. Fus ion Diameter The purpose of the fusion diameter test is to determine the fluidity of a fusible heat-reactive resin when heated on a hot plate at a specified temperature and pressure. The fusion of a resin is a gradual process of changing from a hard to a soft state as heat is applied rather than the abrupt transition of organic crystals to a l iquid . The conditions under which the measurement is made must be rigidly defined. Two distinct processes are involved i n such fusion. Firstly, the temperature and pressure experienced w i l l increase the plasticity of the solid powdered resins resulting i n flow of the particles. However, cross-linking reactions during the curing of the resin at the elevated temperature also occur leading to a solid, thermoset resin. Thus it might be expected that resins of lower molecular weight w i l l produce larger fusion diameters for two reasons. Considering the model of flow proposed by Jones (1952) (see section 2.5.), lower molecular weight resin molecules are more l ikely to flow within the aggregate structure. Secondly, a lower molecular weight resin w i l l take longer at a given temperature to attain the cross-linked, infusible structure of the fully cured polymer and thus more time for flow of the resin is available. These considerations do not account for the role that any additives such as plasticizers or cure promoters may play i n the flow and fusion of the resins. 2.3.7. Stroke cure The stroke cure test is intended to provide an indication of the speed at which a resin cures or sets. It is an empir ical measurement of the time taken for a powdered resin to melt and then cure, through condensation reactions, to form an infusible solid (Anonymous 1985). 28 2.4. Resin Characteristics and Adhesive Performance It would seem logical that the chemical properties of a particular phenolic resin formulation w i l l affect its adhesive performance. N o w that analytical instrumentation exists that allows the elucidation of the structure of these resins, attempts have been made to correlate a range of molecular parameters with the bonding performances of the resins. Characteristics of phenolic resins which have been routinely determined i n the past are viscosity, specific gravity, free formaldehyde, free phenol, molecular weight and solids content. M o r e recently with the development of sophisticated chromatographic techniques such as gel permeation chromatography ( G P C ) a more detailed description of the profile of molecular weights present i n a resin is possible. Nuclear magnetic resonance ( N M R ) spectroscopy studies allow more detailed description of functional groups present i n the resin and thermal techniques (differential scanning calorimetry, differential thermal analysis and thermal mechanical analysis) allow chemical and physical changes occurring during the curing process to be observed. Adhes ion can be thought to be brought about by three consecutive processes:- wetting, adsorption and interdiffusion of the resin with respect to the wood substrate. Wett ing is the process by which resin flows or spreads and establishes molecular contact with the wood. Adsorp t ion is a surface phenomenon involving orientation and bonding of molecules at the resin-wood interface. Interdiffusion occurs as adhesive molecules diffuse through the interface and into the wood substrate. The ability of a resin to penetrate a wood substance relates i n part to the size or molecular weight of the molecules i n the resin. The distribution of molecular weights i n a resin depends upon its ini t ial components and method of preparation. The molecular weight distribution of a resin would appear to be a major factor 29 influencing adhesive performance and it has been one of the resin parameters most actively studied (Wi lson et al 1979). W i l s o n et al (1979) investigated possible correlations between wetting properties, molecular weight distribution, internal bond (IB) and modulus of rupture ( M O R ) strengths i n waferboard produced from oak using l iquid phenolic resins and a series of lignosulphonate phenolics. N o significant correlations between wetting variables and IB or M O R strengths were found. Wett ing is an important component of the bond formation process but is insufficient alone to produce a strong adhesive bond. A l imitat ion of their study was noted that wetting measurements were taken at ambient temperature whilst much higher temperatures are present i n industrial applications. A t higher temperatures the viscosity of the resin would decrease and result i n improved wetting. In addition, the pressure applied during bonding would tend to increase wetting phenomena. F o r the molecular weight studies, molecular weight distributions were obtained using G P C and the components were allotted to high, medium and low divisions based on arbitrary molecular weight values. It was found that molecular weight distribution and I B strengths correlated significantly but no correlation was found between molecular weight distribution and M O R strengths. Lower I B strengths were observed for resins containing a greater proport ion of small and medium sized molecules. Resins containing increasingly high molecular weight molecules were thought to retain a higher percentage of resin i n the glueline for bonding. It was noted that the relationships observed were only applicable to the particular resins and wood species used. W i t h less absorbent species than that used (oak), different results may be observed. 30 G o l l o b (1983,1985) prepared phenolic resins with a variety of formulations and evaluated them as adhesives both neat and with additives. Parallel- laminated panels of Douglas fir were prepared and breaking load and percentage wood failure were measured. Resins of high molecular weight lead to a low observed wood failure. This was attributed to the fact that they flowed poorly, allowing water to migrate away from the resin into the wood. These resins would tend not to be very thermoplastic compared to lower molecular weight resins and would therefore need more moisture to flow. This effect was exaggerated with longer assembly times. A relative branching index was calculated as the ratio log M w / l o g viscosity. Larger ratios indicate a greater relative degree of branching. L inea r resins tend to be more thermoplastic and melt and flow better at high temperature even in the absence of moisture. W h e n the resins were mixed with extenders and other additives the effects of these properties were obscured. H i g h molecular weight resins flowed better than low molecular weight resins since the extended mix retained moisture i n the glueline improving the thermoplasticity of the resin, whilst condensation was inhibi ted i n the low molecular weight resins. The molecular weight values determined for the resins prepared were much higher than typical values observed by other authors (Wi l son et al 1979, P i zz i 1983). This is most l ikely attributable to the detection method employed in the G P C analyses - low-angle laser light scattering. However, even i f absolute values are not i n agreement with others, the comparisons made between resins of high and low molecular weights are still val id . Peterson (1985) aimed to develop a medium density hardwood laminate ( O S B type) that met Amer ican Plywood Associa t ion performance standards for structural panels while having high strength retention and 31 resistance to delamination when weathered. Fourteen l iqu id resol resins were obtained directly from suppliers and then analyzed for chemical and molecular properties. G P C was used to obtain molecular weight distributions, D S C was used to study the curing behaviour of the various resins and to determine percent cure as a function of time. N M R was used to determine the F : P ratio of the resins. G P C analysis indicated that the highly durable resins were composed of primari ly low molecular weight fractions. Greater than 7 0 % of the molecular weight distribution of these resins was composed of molecules of less than 2200 molecular weight units. It was felt that the lower molecular weight molecules were able to penetrate deeply into the micropore structure of the wood resulting i n a mechanical anchorage effect. The medium and high molecular weight range resins d id not produce durable bonds, which was attributed i n part to insufficient penetration into the wood micropore system. D S C profiles agreed with others published. Percent cure results indicated that complete or near complete cure was obtained i n two minutes or less at a temperature of 160\u00C2\u00B0C. Complete cure could also be achieved at a lower temperature and a longer per iod of time. It was suggested that immediate hot stacking of panels after removal from the hot-press would allow shorter pressing cycles since complete resin cure would occur i n the hot stack as a result of the residual heat. A F : P ratio of 1.5:1 was found to produce the highest performance resin. This is i n agreement with the theoretical ideal combining proport ion for complete cross-linking during resin cure (Mar t i n 1956). Those resins with the highest buffering capacities were found to produce the most durable bonds. This characteristic relates to the ability of a resin system to tolerate changes i n p H brought about for example by migration of acidic extractives into the glueline during hot pressing. A c i d extractives such as hydrolyzable 32 and condensed tannins and phenolic acids may affect resin cure by altering the resin p H or by becoming involved in cross-linking reactions and causing premature gelation. A resin with a high buffering capacity w i l l reduce these effects. Stephens and Kutscha (1987) studied the effects of high and low molecular weight fragments of a l iqu id phenolic resin on the properties of aspen flakeboard. The resin was fractionated using diafiltration and then boards were produced using the two fragments plus various recombinations thereof for comparison with the neat resin. The neat resin fractionated as 2 7 % low molecular weight material and 7 3 % predominantly high molecular weight which also contained some low molecular weight material . W h e n boards were produced using the high molecular weight or low molecular weight fragments at the same total loading as neat resin it was found that the high molecular weight fraction performed as wel l as the unfractionated resin. The boards produced from the low molecular weight fraction exhibited much poorer performance than the unfractionated resin i n internal bond strength, thickness swelling and water absorption. Add i t iona l panels were prepared with a high molecular weight fraction loading proportionally reduced to give a loading equivalent to that applied with the neat resin. In this case the properties attained were not as good as with the neat resin. It appears that the low molecular weight components serve to l ink the larger high molecular weight polymers as curing proceeds. Penetration studies seemed to indicate that with low molecular weight material gross penetration of the glueline left insufficient adhesive for an adequate bond whilst high molecular weight material exhibited much less gross penetration. He re again a combination of low and high molecular weight materials i n the neat resin would appear to produce a formulation that provides sufficient mechanical interlocking at the 3 3 surface together with adequate cel l wal l penetration below the surface for satisfactory bonding to occur. F r o m the l imited number of investigations reported on the relationships between resin characteristics and adhesive performance it is apparent that an extremely complex situation exists. A number of resin parameters are inherently inter-related, such as F : P ratio and molecular weight distribution. It would be improper to attempt to ascribe a property of a resin to one molecular parameter alone. Comparisons between results are not always simple. F o r instance, i n the works described, different molecular weight values were used to determine the boundaries for the individual fractions. The low molecular weight fragment of W i l s o n et al (1979) was that below 1000 and the high fraction was that above 20000. Peterson (1985) recognized four fractions, 0-660, 660-2200, 2200-7400 and >7400. Stephens and Kutscha (1987) separated their resin samples into fractions above and below 1000. Peterson (1985) showed that a high proport ion of the low molecular weight fraction (<2200) was preferred. W i l s o n et al (1979) however, found resins with a higher proport ion of high molecular weight molecules gave greater I B strengths. Stephens and Kutscha (1987) found that although strong bonds were produced with high molecular weight resins, a combination of high and low molecular weights was opt imum. G o l l o b (1983, 1985) found that high molecular weights produced little wood failure due to poor flow. The consensus would appear to be that low molecular weight resin molecules are required i n order to achieve sufficient penetration into the wood substrate to produce a permanent bond. Higher molecular weight molecules are also required for good bond formation, however, because of 34 their better gap-filling properties, they tend to remain i n the gross glueline, bridging the adjacent wood surfaces. 2.5. Res in F low A number of authors have recognized that there are certain requirements i n the flow properties of a resin which are necessary i n order to al low bond formation to occur. If flow of the resin is poor then little or no diffusion into the wood substrate is observed and poor bonding results. If flow is excessive, starvation of the glueline may occur as much of the resin becomes absorbed by the wood. The primary resin characteristic which would appear related to flow is that of the molecular weight distribution. However , other molecular parameters, such as hydrogen bonding capacity and level of caustic i n the resin, must not be neglected. Whi ls t the published results have concentrated on l iquid resins, obtaining sufficient flow is just as important when the resin is applied i n the solid form. H e r e the plasticity of the solid powdered resins w i l l increase at elevated temperatures and greater flow w i l l result. Cur ing also takes place at the temperatures concerned, which results i n a thermoset resin. Thus a balance w i l l exist between the processes of flow and cure, whereby it is important that sufficient flow of resin has occurred to allow penetration into the substrate before the resin becomes fully cured. Mos t research into resin viscosity, flow and hydrogen bonding has been performed on resins i n solutions. Jones (1952) however, studied the relationships between molecular weight, flow and melt viscosity i n solid novolak resins. Novolaks were used rather than resols since the latter are heat sensitive and curing of the resins would occur under the conditions employed i n the study. F l o w was measured by the distance a pellet of resin flowed down an inclined glass plate at 150\u00C2\u00B0C. The smooth relationship found 35 between distance of flow and number average molecular weight is shown i n Figure 1. It was crit ical to remove a l l free phenol and water from the resins since the viscosity of solid novolaks is known to decrease almost five times for an increase of 1% i n water content (Dienes 1949). A s evidenced from the changes i n viscosity observed, flow was also temperature dependent. A clear description of viscous flow is given by Glasstone et al (1941). The movement of a molecule taking part i n a flow process is regarded as needing a certain quantity of energy, the activation energy for viscous flow, before flow takes place, i n much the same way as a molecule needs activation energy before chemical reaction may take place. A molecule which receives the necessary activation energy forms an activated complex which i n due course flows i n a flowing system. F l o w is imagined to take place by a series of jumps performed by a molecule from one equi l ibr ium posit ion to another adjacent vacant position. In a l iquid at rest, this process occurs i n a random direction but when a force is applied the process is helped along i n the direction of the force. The force reduces the potential energy barrier needed to be overcome by a molecule moving from one equi l ibr ium posit ion to another and hence increases the number of times a molecule moves i n that direction. Phenolic resins exist largely i n the associated state and the breakdown of these associated complexes w i l l increase as the temperature rises. This w i l l lead to a decrease i n viscosity since the energy required for a molecule to flow or move from one equi l ibr ium posit ion to the next becomes less. The association of resin molecules is through hydrogen bonds. One result of the molecular geometry of phenolic resins is the relative ease with which the structures may form hydrogen bonds ( D r u m m and L e B l a n c 1972). 36 100 80 1 60 o \u00C2\u00B0o \u00C2\u00AB 5 20 4 i\u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 200 400 600 M 800 1000 n Figure 1. Dependence of flow of novolak resins upon molecular weight (Jones 1952). 37 Three types of hydrogen bonds are thought to exist in novolaks:-intramolecular OH\u00E2\u0080\u0094-OH (I), intermolecular OH\u00E2\u0080\u0094-OH (II) and intramolecular OH-\u00E2\u0080\u0094benzene ring (III). Intramolecular bonds would form between adjacent hydroxyl groups on ort/io-bonded novolaks or through OH groups which come into proximity through molecular mobility and coiling. Intermolecular bonds could form as single OH\u00E2\u0080\u0094-OH bonds between molecules or as multiples which would lead to ring structures, as illustrated. The OH\u00E2\u0080\u0094-benzene ring hydrogen bond is the weakest of the three and is not of great importance. From infrared spectroscopy studies, Cairns and Eglinton (1962) concluded that the ring-like structures were predominant. It has been suggested that hydrogen bonding can effect the properties of a resin in a number of ways. Ionization of the phenolic OH can be influenced, the reactivity of an end position may be hindered by cyclization, inductive effects may be reinforced and physical 38 behaviour, i n melt or solution, can be affected by the presence of hydrogen bonds. Jones (1952) described the novolak structure as an ordered amorphous phase where the number of hydrogen bonded neighbouring molecules contribute to the apparent much greater size of molecule than that obtained from molecular weight measurements. A s temperatures rise, the hydrogen bonded structure is disrupted and molecules not associated within the aggregate w i l l flow (Figure 2). The application of energy at the point indicated is sufficient to break the hydrogen bonds marked \"x\" and move the resulting dimer to the vacant space to the right. This manifests itself as the glass transition point T observed by D S C and T M A . The T observed may however be caused by other mechanisms. Be low the T the polymer chains are locked i n a particular configuration. A t the T the thermal energy available is sufficient to overcome the rotational energy barriers i n the chain (Cowie 1973) and chain segments can move cooperatively. The T is not particularly marked in phenolic resins due to the amorphous arrangement of the polymer segments. A s the temperature rises and more energy is available to break hydrogen bonds, individual molecules become involved i n the flow process and discernible bulk flow of the resin occurs (Jones 1952). It has been proposed that the nature of the hydrogen bonds between the resin molecules changes during the curing process. The tendency is to form intermolecular hydrogen bonds as curing proceeds and this aids the formation of a highly cross-linked structure. F r o m Jones's model we can see how the formation of intermolecular hydrogen bonds would reduce the tendency of the resin to flow. 39 Figure 2. F l o w process of an activated complex (Jones 1952). 40 Go's work (1988) with powdered waferboard resins indicated that poor resin flow was associated with low I B strengths. However , it was suggested that resin flow is only of importance at borderline moisture contents, pressing times, or both. A t short pressing times and low moisture content, poor bonding may be a result of insufficient heat reaching the core of the panel to allow resin cure, rather than a lack of resin flow. 2.6. Detection of Resin Penetration into W o o d In developing an understanding of wood-adhesive interactions the need for a method to determine the depth of penetration of adhesives into wood has been recognized by many authors. Evidence exists that the strength of bonds increases with the depth of penetration, up to certain l imit . Starvation of the glueline can occur i f penetration into the wood substrate occurs to an excessive extent leaving insufficient resin at the glueline to form an adequate bond. However, the measurement of resin penetration is not a simple matter. A variety of approaches have been reported i n the literature. M a r i a n and Suchsland (1957) studied the gluelines between blocks of Scots pine bonded with a variety of adhesives. Incident light fluorescence microscopy i n conjunction with a variety of dyes was used. Resul t ing photographs illustrated various degrees of penetration of adhesive into the wood surfaces. Nea rn (1965, 1974) used a variety of techniques to investigate resin penetration into wood. These included white light microscopy (transmitted and reflected light), radiography uti l izing 1 4 C incorporated i n the resin, ultraviolet light microscopy (reflected and transmitted light), scanning electron microscopy ( S E M ) and transmission electron microscopy ( T E M ) . These had varying degrees of success but the author was able to demonstrate both gross penetration of resins into cel l cavities and penetration into cel l walls. Lehmann (1967) took colour photographs of 41 flakeboard gluelines using ultraviolet light microscopy. A dye was incorporated into the resins used. The photographs illustrated the distribution of resin along the gluelines and the extent of penetration into the flakes. T E M was used by Fengel and K u m a r (1970) to study pinewood scarf joints after testing. Penetration of the P F resin into the cells near the glueline was demonstrated. In his review of S E M applications Collet t (1970) incorporated lead oxide i n the caustic addition of a P F glue mix to enhance the contrast, with some l imited success. Ha re and Kutscha (1974) examined gluelines of eastern white spruce plywood using both ordinary bright field and scanning electron microscopy. Penetration of the glue into cells adjacent to the glueline was clearly demonstrated. Whi te et al (1977) developed a more quantitative technique for measuring resin penetration. Bromine was incorporated into a resorcinol formaldehyde resin and specimens from southern pine with microtomed surfaces were prepared. After curing, thin sections were removed from near the glueline with a microtome. Neut ron activation analysis was used to determine the concentration of bromine i n the different sections. A clear picture of bromine distribution i n the glueline was obtained. Bo l ton et al (1985) incorporated a sulphur label (as thio-urea) into a urea-formaldehyde resin during synthesis. They followed the penetration of the resin into the wood using scanning electron microscopy with energy dispersive analysis of x-rays ( S E M / E D A X ) . Quantitative results were possible for distribution of resin along the glueline and for penetration into cel l lumens and cel l walls. The majority of the resin was concentrated close to the glueline and was rarely found more than two cells away from the chip junction. Evidence of cel l wal l penetration was found but little resin was observed i n the cel l lumens. Brady and K a m k e (1988) studied the effects on resin penetration of hot-pressing parameters. A s p e n and Douglas fir flakes 42 were used together with a l iquid P F resin. Fluorescence microscopy and a manual digitization technique were used to evaluate penetration. They found that the pressing parameters of temperature, moisture content and time control the viscosity of the resin and hence influence the extent of resin penetration. However, the uniformity of the penetration of the resin was influenced more by the natural variability present i n the wood than by the pressing variables. The results obtained by these authors can be summarized as follows. Resins do not generally penetrate to a depth of more than two or three cells i n sound wood (with tangential surfaces greater penetration along ray elements may occur). The primary anatomical impediment to further gross penetration would appear to be the small diameter pits connecting cells (Smith 1971, Fengel and K u m a r 1970). Resins penetrate cel l walls as we l l as the lumens. Poor bond performance is associated with shallow penetration by adhesive. The viscosity of l iqu id adhesives appears to effect penetration -resins of greater viscosity penetrating less (White et al 1977). A number of problems with some of these techniques have been recognized. Incompatibility of dyes with adhesives used has been experienced. Poor contrast between wood and adhesive is often noted with S E M and T E M . Where lead oxide has been added to improve contrast a complicat ion may arise. L e a d ions are smaller than intermediate weight phenolic molecules and thus may be differentially absorbed by the cel l wal l . This would lead to an anomalous picture of resin distribution. Differential absorption may also occur with sodium hydroxide used i n the preparation of many resins. Sodium hydroxide causes the wood to fluoresce at the same wavelength as does P F resin and thus difficulty occurs i n determining i f observed fluorescence is due to resin or not. 4 3 In finding a technique which will allow a description of the resin penetration, one inherent problem arises that PF resins and wood have similar elemental composition. Simple visual inspection of bonded wood strips under the scanning electron microscope (SEM) can provide some qualitative information on the penetration but a more quantitative description is desired. A label is required in order that the glue can be differentiated from the surrounding substrate. Bromine is the label which has previously been applied to PF resins, in the form of raeta-bromophenol (Smith and Cote 1971, Ayla and Parameswaran 1980). The label must be chemically incorporated into the resin rather than just being physically mixed and must be introduced before the formation of the prepolymer for a number of reasons. If the label is not chemically incorporated into the resin there is no guarantee that it will follow identically the distribution of the resin itself. Factors such as differential diffusion, molecular sieving and differential adsorption may cause the separation of the resin and the label (Bolton et al 1985). The label must be incorporated whilst forming the prepolymer otherwise different molecular weight fractions may contain different proportions of the label. This is especially important in a study concerning the behaviour of resins having different molecular weight distributions. Thus a proportion of the phenol in the initial reactants in the resin kettle is replaced with raefa-bromophenol. Although a bromine atom in the 3-position will affect the reactivity of other ring positions to some extent it is ortho-, para-directing as is the hydroxyl group of the phenol and thus the same ring positions will undergo reaction as with pure phenol. Past studies have not found it necessary to replace all of an ingredient in a resin cook with its labelled counterpart (Bolton et al 1985, Smith 1971). 44 2.6.1. Scanning electron microscopy and analysis of x-rays In the scanning electron microscope an electron beam is focused on the object of interest. A number of possible interactions between the electron beam and the material take place which may be divided into two sorts:- elastic scattering and inelastic scattering. Elast ic scattering involves no transfer of energy from the beam to the sample but affects the trajectory of the beam. Inelastic scattering involves a transfer of energy from the beam to the sample. Inelastic scattering is responsible for producing the signals which are of most interest i n scanning electron microscopy including secondary electrons, Auger electrons, characteristic x-ray lines and continuum x-ray radiation. Cont inuum x-ray radiation is produced i f inelastic scattering occurs through interaction with the nuclei of the atoms. The moving electrons lose energy i n the Coulomb field of the nucleus and emit continuum or background x-ray radiation. Characteristic x-ray lines can be produced i f inelastic scattering causes the ejection of an inner shell electron. A s an electron from a higher energy level shell falls to f i l l the gap created, the accompanying change i n energy is released as a characteristic x-ray or as an ejected Auger electron. The energy of the characteristic x-ray is dependent on the difference between the energies of the level from which the electron was ejected and the level of the higher energy level shell. The characteristic x-ray lines are named according to the shells involved. Thus i f an electron from the L shell moves to replace one from the K shell, the transition is termed a K T_ water out Stirrer Figure 3. Apparatus used in preparation of laboratory resins. 48 methylolation step of the synthesis over the condensation step. Formaldehyde (the balance of the requirement) and sodium hydroxide (25% of the requirement) were added and the mixture heated to 85\u00C2\u00B0C. This temperature was held for 30 minutes. The balance of the sodium hydroxide requirement was then added and the reaction mixture held at 85\u00C2\u00B0C unt i l the desired viscosity was reached, at which point the reaction mixture was rapidly cooled. Throughout the cook, the temperature of the reaction mixture was monitored with a thermometer probe. The viscosity of the resin during the synthesis was measured by removing a sample, rapidly cooling it to room temperature and comparing its viscosity to Gardner -Hol t viscosity tubes. Resins having different molecular weight distributions were produced by varying the viscosity at which the reaction was ended by rapid cooling. A lower viscosity would indicate a lower degree of polymerization, for a given level of caustic. The viscosity of each resin cook was determined using a Brookf ie ld synchroelectronic viscometer. Measurements were taken at 25\u00C2\u00B0C using three different spindles. The percentage solids content of the resins was determined by heating a known weight of resin to a constant weight at a temperature of 120\u00C2\u00B0C. 3.1.1.2. Dry ing It had been hoped to spray-dry the laboratory-prepared resins so as to approach as closely as possible the mode of preparation used industrially. Unfortunately it was not possible to obtain access to a spray drier to produce the quantities of resins required for this study and thus the resins were made into powders by freeze-drying. The l iqu id resin was poured slowly into a tray of l iqu id nitrogen which was then placed i n a vacuum freeze dryer. Once dry, the resins were ground using a pestle and mortar and sieved to pass a 200 mesh screen (75//m). It was possible to spray dry a smaller quantity of one 49 bromine-labelled resin (Cook #5) and this allowed some comparisons to be made between the chemical and physical properties of the resin produced by the two drying techniques. 3.1.2. Commercial resins F o r convenience, the commercially-available resins are referred to by an alphanumeric code. Whether the resin is used i n the face layers of a board's construction, i n the core layer or throughout the board, together with the date of introduction of the resin, when known, is summarized i n Table 2. Table 2. Summary of commercial resins Res in Type Introduction A Face B * Face 1985 C F a c e / C o r e D I F a c e / C o r e 1977 D 2 * F a c e / C o r e 1977 D 3 F a c e / C o r e 1977 E Core 1987/8 F l Core 1981 F 2 * Core 1981 G * Core 1987 H Core 1985 Resins F l and F 2 were samples of the same resin obtained eighteen months apart. Resins D I , D 2 and D 3 were a l l samples of the same resin. Resins D I and D 2 were obtained eighteen months apart and R e s i n D 3 was a sample that was over seven years old. A l l of these resins were analyzed by the techniques to be described. Those resins marked with an asterisk (*) were used i n the construction of two-ply paral lel laminate boards. 50 3.2. Resin Analysis The analytical procedures used were applied to the six resins produced i n the laboratory and to the eleven commercial adhesives. 3.2.1. Infrared absorption spectroscopy The I R spectra were obtained using freeze-dried samples prepared i n potassium bromide discs i n the usual manner. Some of the spectra were run on a Pe rk in E l m e r SP-800 instrument and others on a Perk in -Elmer 1600 series Four ier transform instrument. The instruments were used i n the transmission mode. A purge of nitrogen gas was applied during the recording of the spectra. 3.2.2. Nuclear magnetic resonance spectroscopy 3.2.2.1. Sample preparation Acetyla t ion of the resin prior to analysis is thought to be beneficial for several reasons. Reactive methylol groups are converted into relatively unreactive ester groups stabilizing the storage life of the polymer. The sample material produced is soluble i n N M R solvents such as deuteriated chloroform ( C D C 1 3 ) which is less hygroscopic than more polar solvents such as d 6-acetone or deuteriated alcohols. Thus problems associated with extraneous protons from absorbed moisture are largely el iminated (Go l lob 1983). T o 400 mg of freeze-dried powder, 20 m L of anhydrous pyridine and 20 m L of freshly-distilled acetic anhydride were added whilst keeping the mixture i n an ice-bath unti l the powder had dissolved. After 72 hours at r o o m temperature the mixture was poured into 150 m L of ice-water i n a separatory funnel and extracted into 60 m L of dichloromethane. The dichloromethane was then rinsed three times with 50 m L of 2 % aqueous hydrochloric acid, 7% aqueous sodium bicarbonate and disti l led water. The 51 organic layer was then dried over magnesium sulphate overnight, filtered, rotary evaporated and placed i n a vacuum over phosphorus pentoxide for 24 hours. The acetylated resin was dissolved i n deuteriated chloroform ( C D C 1 3 ) for N M R analysis. The spectra were obtained on a V a r i a n X L - 3 0 0 instrument i n the Chemistry Department, U . B . C . , using tetramethylsilane as the internal standard. Two acetylated samples per resin were prepared and two integrations were performed for each spectrum. The average of these four integrations was used i n subsequent calculations. 3.2.2.2. Calculat ions A representative spectrum is shown i n Figure 4. Peak assignments were as follows:- 6.5-7.5 p p m for aromatic protons, 5.2-5.5 p p m for methylene protons at the acetoxy end of benzyl-type hemiformal units, 4.8-5.2 p p m for methylene protons of acetoxymethyl groups, 4.5-4.8 ppm for methylene protons at the benzyl end of benzyl-type hemiformal units and methylene protons of dibenzyl ether bridges, 3.5-4.2 p p m for methylene protons of diphenylmethane-type bridges, 2.1-2.5 p p m for methyl protons of acetoxyphenyl groups and 1.5-2.0 p p m for methyl protons of acetoxymethyl groups. These peak assignments are also shown i n Table 3. 52 Figure 4. A representative ^ - n u c l e a r magnetic resonance spectrum of a phenol ic resin, showing peak assignments. 53 Table 3. Assignments for * H - N M R peaks P P M A r e a Funct ional Groups 6.5 - 7.5 A 1 A r - H 5.2 - 5.5 A 2 A r - C H ^ O - C F L j - O A c 4 .8 -5 .2 A 3 A r - C H 2 - O A c 4 .5-4 .8 A 4 A r - C H ^ - 0 - C H 2 - O A c A r - C H 2 - O C H 2 - A r 3 .5-4 .2 A j A r - C H 2 - A r 2 .1-2 .5 A 6 A r - O - C O - C H ^ 1.5-2.0 A , A r - C H 2 - 0 - C O - C H 3 A r - C H 2 - 0 - C H 2 - 0 - C O - C H 3 The value of A 6 , three times the relative number of aromatic rings, can be calculated i n three ways. In low molecular weight resins it may be read directly from the integrated area of the peak at 2.1-2.5 p p m for acetoxyphenyl groups. However, the resolution between the acetoxyphenyl signal and that for the acetoxymethyl (1.5-2.0 ppm) becomes poorer with higher molecular weight resins and thus the value for the acetoxyphenyl groups is arrived at indirectly. Assuming that a l l the methylol groups are acetylated, A 6 can be calculated from the difference between the sum of the acetoxy signals and the methylol and hemiformal signals, as A 6 = ( A 6 + A , ) - 3 / 2 ( A 2 + A 3 ) It has been noted that acetylated resins may frequently contain small amounts of acetic acid as an impurity and this leads to an increase i n the value of A ? . In this case the equation to be used is A 6 = 3/5 CAj + (A3 + A 4 ) / 2 + 54 which accounts for a l l the linkages i n which the aromatic ring may be involved (Woodbrey et al 1965). It follows that; (Go l lob 1983) Number of aromatic protons = \ Number of phenol units = A 6 / 3 Number of benzyl-type hemiformal units = P^/2 Number of methylol units = A^/2 Number of dibenzyl ether bridge units = Number of methylene bridge units = A^/2 The following relationships hold; Average number of active (o,p) protons per phenol unit Number of aromatic protons - 2 A 1 - 2 3 A 1 - 2 Number of phenol units As/^ A Average number of methylol units per phenol unit Number of methylol groups 3Aj Number of phenol units AJ3 2 A ( 6 Average number of methylene bridge units to which each phenol unit is connected Number of methylene bridge units x 2 \/2 x 2 3 A 5 Number of phenol units \/3 \ Average number of benzyl-type hemiformal units per phenol unit Number of benzyl-type hemiformal units A^/2 ?>A^ Number of phenol units \/3 2 A ( 6 Average number of dibenzyl ether bridge units to which each phenol unit is connected Number of dibenzyl ether bridge units x 2 ( A 4 - A 2 ) / 4 x 2 Number of phenol units AJ3 5 5 = 3 ( A 4 - A , ) B o u n d formaldehyde:phenol ratio (F:P) To ta l bound formaldehyde Number of phenol units = 2(^/2) + A3 /2 + 2(A 4 - + A5/2 V 3 3(A2 + A 3 + A 4 + A 5 ) Woodbrey et al (1965) also provided formulae for the calculation of the average number of phenol rings per molecule, n, and the number average molecular weight of the acetylated resin, M n ( A c ) . \" l - 3 ( A 4 - A 2 ) - 3 A ~ n 4 A 6 2 A 6 -1 M n ( A c ) = 131 + A 1 + 1 0 3 ( 3 ^ + 73(3A3) +44(3(A 4 - AJ) + 14(3A 5 ) A 6 2 A 6 2 A 6 4 A 6 2 A 6 3.2.3. G e l permeation chromatography 3.2.3.1. Instrumentation The G P C system used an isocratic Spectra Physics 8810 pump that fed the solvent through a Rheodyne 7125 injector loop to the V a r i a n Mic ropak T S K exclusion column system. A G H 8 P guard column was used followed by four analytical columns i n series (1000H, 2500H, 3000H and 4000H). This system of columns should provide separation over a molecular range of 100-400 000 according to the V a r i a n literature. The T S K typeH columns contain spherical, cross-linked polystyrene/divinyl benzene particles of 8 to 10 jum 56 diameter, packed i n tetrahydrofuran. The analytical columns were 30 cm i n length. A Kratos Spectroflow 757 U V / V I S detector was used at a detection wavelength of 280 n m and the output signal was fed to a Spectra Physics 4290 integrator. The integrator was equipped with a memory module unit and a G P C - P l u s microchip which allowed the integrator to perform a variety of G P C calibrations and calculations. Printed chromatograms and molecular weight distributions were obtained on the integrator's X - Y plotter. R a w slice data from the integrator was also fed to an I B M - P C A T compatible computer for storage on a hard disc. This allowed data to be retransmitted to the integrator for recalculation and reporting at a later time. D a t a could also be transferred to the U B C - M T S mainframe computer system which allowed the detector responses to be plotted using a graphics package. This was necessary since the detector responses could not be generated again on the integrator's plotter once the run had been completed. The system is illustrated i n Figure 5. 3.2.3.2. Calibration Initially, the four analytical columns were connected i n the system individually i n order to assess the performance of each column. Polystyrene standards, matching those run by the manufacturer as closely as possible, were run through each column using T H F as the eluting solvent. The number of theoretical plates per column was calculated from these peaks. The four columns were then assembled i n series and calibrated using a number of polystyrene standards of different molecular weights. The number of theoretical plates for the four columns i n series was also calculated. The values based on a number of the peaks were averaged. 57 Solvent Bott le Pump Injector G u a r d Ana ly t i ca l Va lve C o l u m n Columns Detector Computer Integrator Waste Figure 5. G e l Permeat ion Chromatography System 58 3.2.3.3. Sample analysis R e s i n samples were acetylated prior to analysis since preliminary investigations showed that incomplete solubility was achieved for most resins when used i n the unacetylated condition. The acetylated resin samples prepared for the - \" U - N M R analyses were used. Solutions of 0.5% w / v of the acetylated resins i n T H F were used. A flow rate of 1.0 m L of T H F was used i n a l l the analyses. F ive replicate analyses of each sample were performed. 3.2.4. Addit ives 3.2.4.1. Hexamethylenetetramine N o definitive method for the determination of H M T A appears in the literature. The method used is based on an official hydrolytic method (Anonymous 1942). One gram of resin was dissolved i n 50 m L of water. For ty m L of 1.0 N sulphuric acid was added and the mixture was boi led on a hot plate for 5 hours or unti l the odour of formaldehyde had disappeared. Water was added at suitable intervals to maintain the original volume. After cooling, the mixture was titrated with 1.0 N sodium hydroxide using methyl red as the indicator, titrating to a yellow end-point. These determinations were performed i n duplicate. The H M T A content is calculated assuming that four moles of sulphuric acid are required to neutralize one mole of H M T A (see mechanism of H M T A breakdown to form carbonium ions, page 15). % H M T A = (blank - titer) x molecular weight H M T A x 100 1000 4 wt. of sample % H M T A = (blank - titer) x 3.5 wt. of sample 59 3.2.4.2. Inclusions Nine of the resins (C , D l , D 2 , D 3 , E , F l , F2 , G , H ) contained some small white particles. These particles were not mentioned i n any of the literature obtained from the manufacturers. Particles were separated from each of the resins and infrared spectra were obtained using potassium bromide discs. Proton N M R spectra could not be obtained since the particles were insoluble i n C D C 1 3 and d 6-acetone. 3.2.5. Thermal mechanical analysis The thermal mechanical analyses were performed on a Perk in -Elmer T M S - 2 instrument. The instrument measures dimensional changes i n a sample as a function of temperature. A quartz probe is placed i n contact with the sample, and a weight is applied to the probe. A s the dimensions of the sample change, the probe tracks the sample movement. A linear variable differential transformer is used to convert the probe movement into a signal which is displayed on a recorder chart. The resin (15 \u00C2\u00B1 2 mg) was weighed into a small a luminium pan. The pan was tapped lightly on the bench (holding it with a pair of tweezers) to ensure even packing of the resin. A n a luminium l i d was then placed on top. The assembly was then placed under the probe and the ini t ia l height of the resin calculated (the thickness of the pan and l i d were measured prior to the weighing of the resin). The temperature range scanned was 25-200\u00C2\u00B0C. The rate of temperature increase was 10\u00C2\u00B0C per minute. The weight placed on the probe was 300 g producing a pressure of 95.2 k P a (13.8 psi) over the area of the pan l id . Three replications were performed for each resin. The height of the resin under the probe was plotted against temperature automatically on an X - Y plotter. Fo r each sample, the height of the resin under the probe as a percentage of its 60 original value was calculated from the resulting plots and replotted using the U B C - M T S mainframe computer. 3.2.6. Fus ion diameter Us ing a pellet-making apparatus, 0.5 g of resin was made up into a cylindrical pellet of 12.0 m m diameter and 5.0 m m height. The densification of the resin in . the pellet was in . the order of 30-40%. The pellet was transferred to a preheated glass plate on top of a hot plate heated to 1 4 0 \u00C2\u00B1 2 \u00C2\u00B0 C . A second, preheated glass plate was placed on top of the pellet followed by a weight sufficient to deliver an in i t ia l load of 0.47 M P a (68 psi). A stop watch was started immediately and after three minutes the weight was removed and the diameter of the melted pellet was measured at four places to determine the average diameter. F o r each resin three replications were performed. 3.2.7. Stroke cure A hot plate was heated to 150\u00C2\u00B0C and a number of glass microscope slides were preheated. A 0.25 g sample of resin was placed on a glass slide and spread using a small steel spatula. A timer was started simultaneously. The resin was alternatively smoothed and patted and any changes occurring were observed. The resin was worked unt i l strings no longer formed when the spatula was momentarily raised and the resin felt dry. 3.3. W o o d Bond ing Study Wood-glue bonds were produced using wood veneers rather than wafers i n an attempt to minimize the variability present i n the gluelines. In a waferboard panel the numerous gluelines at different positions i n the profile of the panel experience different temperatures and pressures during hot-pressing. This makes it difficult to discern effects which are due to the resin alone rather than the physical conditions experienced. B y using veneers a 61 bond was produced with a single glueline, the conditions at which, it was hoped, would be consistent from board to board for a given set of conditions (i.e., moisture content, pressing time). Thus, i n essence, plywood-type bonds (except that the two veneers were arranged with their grain paral le l rather than perpendicular) were produced using these waferboard resins. This experiment was a completely randomized design with factorial arrangement of the 8 x 2 x 3 x 4 levels of experimental factors into 192 treatment combinations. Three replications of each set of combinations were performed. Exper imental Number of Levels Studied Factor levels R e s i n 8 Cook #3, C o o k #4, C o o k #5, C o o k #6 Res in B , R e s i n D 2 , Res in F 2 , R e s i n G Species 2 Whi te spruce, trembling aspen Mois ture Content 3 White spruce -5 .0 ,7 .0 ,9 .5% Trembl ing aspen - 4.6, 6.7, 9.2 % Pressing T i m e 4 75, 90,120, 240 seconds Eight resins were chosen for inclusion i n the wood bonding study. Not a l l the resins studied were used due to a l imit i n the wood stock available and a desire to keep the number of samples to be analyzed down to a manageable quantity. Four laboratory and four commercial resins were chosen. The three laboratory resins from the second series of cooks, that had G P C profiles and p H values closer to those of the commercial resins than the first series of cooks, were chosen. These resins provided a series of molecular weights. The highest molecular weight resin prepared i n the laboratory (Cook #3) was also chosen to provide an extreme molecular weight. O f the commercial resins chosen one was a face resin (Res in B ) , one was suitable for face/core use (Resin D 2 ) and two were core resins (Resins 62 F 2 and G ) . These four resins were chosen to provide information concerning each of the three types of resin available. 3.3.1. Preparat ion of veneer T w o wood species, white spruce {Picea glauca (Moench.) Voss.) and trembling aspen (Populus tremuloides Michx. ) were used i n this study. Veneers were produced from 50.8 m m x 230.2 m m (2 in . x 8 in.) (spruce) or 50.8 m m x 152.4 m m (2 in . x 6 in.) (aspen) predominantly flat sawn nominal stock. The wood was cut into lengths of 406.4 m m (16 in.) on a cross-cut saw and then cut into thicknesses of 12.7 m m (0.5 in.) using a band saw. The veneers were then planed on both sides to a thickness of just greater than 5.08 m m (0.2 in.). The veneers were stored i n three different relative humidity environments for a min imum of four weeks to reach an equi l ibr ium moisture content. They were then planed to their final thickness of 5.08 m m (0.2 in.) immediately prior to gluing and cut to a length of 203.2 m m (8 in.). This short per iod between final planing and assembly was designed to minimize any possible surface deactivation. 3.3.2. G l u i n g procedure The resin was applied to one veneer by means of a salt shaker with no addit ional additive. A resin loading of 0.23 g / c m 2 was used. A second veneer was then gently placed on top of the first veneer. The panel construction was pressed i n a 304.8 m m x 304.8 m m (12 in . x 12 in.) electrically heated hot press. The panels were pressed at a hot plate temperature of 200\u00C2\u00B0C to a thickness of 9.53 m m (0.375 in.) (a compression to 93.75% of the original thickness) using press stops. A pressure of 1.04 M P a (150 psi) was used. 63 3.3.3. Sample preparation and testing F r o m the resulting pressed panels twelve (spruce) or eight (aspen) specimens for shear testing (82.55 x 25.4 mm, 3.25 in . x 1 in.) were cut. A narrow section from the center of each board was retained for subsequent S E M observation. The dimensions of the shear specimens are shown i n Figure 6. The shear specimens from each board were randomly assigned so that equal numbers of specimens of each species were exposed to one of three treatments. The three treatments of the shear specimens were:-a) A conditioning per iod of a m i m m u m of two weeks i n a constant temperature and humidity environment (28.3 \u00C2\u00B1 1\u00C2\u00B0C, 5 0 \u00C2\u00B1 2 % R . H . ) . b) A vacuum/pressure cycle, where the specimens were submerged i n cold water and a vacuum of 710 m m drawn for 30 minutes, followed by a pressure of 0.52-0.55 M P a (75-80 psi) for 30 minutes. The samples were tested i n the wet condition. c) A b o i l treatment, where the specimens were boi led i n water for four hours, dried for 20 hours at 60\u00C2\u00B0C and then boi led again for four hours. The samples were cooled i n water prior to testing. The specimens were tested on an Instron universal strength testing machine using pneumatic jaws to grip the specimens. The specimens were pul led apart i n tension to give a tension shear test. A loading rate of 4.83-6.90 M P a (700-1,000 pounds) per minute was maintained for a l l specimens. The wood failure percentage values were estimated visually. A piece of plexiglass\u00C2\u00AE (2.54 m m x 2.54 mm, 1 in . x 1 in.) with twenty equal divisions marked on it (each 5 % of the total area) was used to aid i n the evaluations. Values for each specimen were estimated to the nearest 5%. The samples which had been tested i n the soaked and boi led condit ion were al lowed to dry before the wood failure was evaluated. 64 Figure 6. Dimensions of the tension shear specimens. 65 3.4. Detection of Res in Penetration 3.4.1. S E M / W D X The small pieces that had been saved from the boards, when preparing the shear specimens, were aspirated under water overnight. The surfaces to be examined were cut using fresh razor blades for each cut. The samples were then t r immed down to cubes of approximately 3 m m dimension. The samples were air-dried for at least two days before mounting on a luminium S E M stubs. Pr io r to observation i n the electron microscope the samples were coated with carbon, to aid the dissipation of any charge bu i ld up on the sample under the electron beam. The accelerating voltage of the S E M was 20.0 k V and the tilt of the samples relative to the detector was 45\u00C2\u00B0. The bromine standard used was potassium bromide. The characteristic x-ray line used by the detector was the L-e* line. The following series of samples were analyzed:-Series 1 - molecular weight, aspen Species A s p e n Moisture content 6.7% Press time 240 seconds Res in C o o k #4 C o o k #5 C o o k #6 C o o k #3 66 Series 2 - molecular weight, spruce Species Moisture content Press time Res in Spruce 7.0% 240 seconds Cook #4 C o o k #5 C o o k #6 C o o k #3 Series 3 - pressing time, low molecular weight Species Moisture content Res in Pressing time Spruce . 7.0% C o o k #4 75 seconds 90 seconds 120 seconds 240 seconds Series 4 - pressing time, high molecular weight Species Moisture content Res in Pressing time Series 5 - commercial resins Species Mois ture content Pressing time R e s i n A s p e n 6.7% C o o k #3 75 seconds 90 seconds 120 seconds 240 seconds Spruce 7.0% 240 seconds Res in B Res in D 2 Res in F 2 Res in G 67 The majority of the analyses were performed using the detector i n the line scan mode since this mode gave the best indication of bromine distribution across the glueline. Three samples of each set of conditions were analyzed at a number of points along the length of the glueline before a representative electron micrograph and line scan or dot map were recorded. 3.4.2. Photomicroscopy T h i n sections of gluelines were prepared using a sliding microtome and disposable blades. Samples were aspirated under water overnight prior to sectioning. The samples were 30 pcm in . thickness. Thinner sections could not be made without the glueline separating. The sections were observed using a Zeiss Universa l microscope. The images were recorded on Kodaco lor G o l d A S A 200 fi lm, using exposure times of 1/4 to 1/2 second. 68 4. R E S U L T S A N D D I S C U S S I O N .4.1. Res in Synthesis The temperature and viscosity profiles during the cooks of the two series of laboratory resins are shown i n Figure 7. The points at which the reactions were ended by rapid cooling is indicated by an arrow for each of the six cooks. The rise i n viscosity as measured by the Gardner -Hol t tubes was seen to occur sooner for the cooks made using the lower caustic level. A higher level of caustic i n a resin is believed to reduce its viscosity by decreasing the number of intermolecular hydrogen bonds present and also solubil izing the aromatic ring by phenoxide i on formation. Thus the cooks with a greater level of caustic required the condensation reaction to proceed for a greater length of time before viscosities equivalent to those i n resins with the lower caustic level were reached. The percentage resin solids, the final viscosity of the cooks at 25\u00C2\u00B0C and pH's as measured i n water are shown i n Table 4. Table 4. Percentage resin solids, final viscosities and pH's of laboratory resins C o o k # % Res in Solids Viscosity (Cps) p H 1 39.7 220 9.65 2 38.7 1680 9.92 3 38.6 3333 9.87 4 40.1 165 8.77 5 38.9 745 8.93 6 39.5 1975 8.89 The final viscosities of the resins as measured by the Brookf ie ld viscometer at 25\u00C2\u00B0C were seen to be higher than those indicated by the Gardner -Hol t tubes at the times the resins were cooled. This was due to the 69 T i m e (M inutes ) Figure 7. Temperature and viscosity plots against time for the synthesis of the laboratory resins. 70 condensation reactions continuing as the resins were cooling from the cooking temperature. 4.2. Resin Properties 4.2.1 Infrared absorbance spectroscopy The I R spectrum of C o o k #5 is shown i n Figure 8a. The spectrum was typical of a resol resin (Has lam and W i l l i s 1965, Secrest 1965). The peaks corresponding to the methylol groups (1010 cm\" 1 and 1270 cm\" 1) were quite distinct. L i t t l e difference was noted between the spectra for Cooks #4, #5 and #6. This would indicate that I R spectroscopy was not a sufficiently sensitive technique to discern differences of the order that existed between these resins. Some differences were observed between the spectra for the two series of laboratory cooks. Figure 8b illustrates the I R spectrum of C o o k #3, which was typical of the first series of cooks. The methylol-related peaks were seen to be smaller relative to the phenolic ring peaks (1500 cm\" 1 and 1600 cm\") i n C o o k #3 than i n C o o k #5. This was indicative of a higher molecular weight i n the first series of cooks since as the resins became more highly condensed, the absorption peaks tended to become less distinct. The I R spectra of Res in F 2 and Res in D 2 are shown i n Figure 9. These spectra were typical of the commercial resins. H e r e again the spectra are characteristic of resol resins. The spectra of some of the commercial resins indicated the presence of H M T A . Figure 10 shows the spectra of H M T A and C o o k #5 including 8% H M T A . Compar ing the spectrum for C o o k #5 and that for the resin with the addition of H M T A it was seen that the H M T A - r e l a t e d peaks that showed up in the resin mixture were those at 1235 cm\" 1 and 1005 cm\" 1. These peaks were observed i n Resins B , C , D I and D 2 . 71 4000 3500 3000 2500 2000 1500 1000 cm\"1 500 4000 3500 3000 2500 2000 1500 1000 cm\"1 500 Figure 8. I R spectra of (a) C o o k #5 and (b) C o o k #3. I I I I I 1 1 1\u00E2\u0080\u0094 4000 3500 3000 2500 2000 1500 1000 era\"1 500 4000 3500 3000 2500 2000 1500 1000 Figure 9. I R spectra of (a) Res in F 2 and (b) Res in D 2 . Figure 10. I R spectra of (a) Hexamethylenetetramine and (b) C o o k #5 containing 8% Hexamethylenetetramine. 74 4.2.2. Proton magnetic resonance spectroscopy The results of the calculations performed for the laboratory resins are shown i n Table 5. The number of unsubstituted reactive positions i n the phenol rings (ortho- plus para-) were similar wi thin each set of laboratory cooks (the first column i n Table 5). However, Cooks # l -#3 have more free hydrogen groups than Cooks #4-#6. E a c h set of cooks was made with a different formaldehyde solution. Al though both solutions were label led as containing 46.6% formaldehyde, it would appear that the formaldehyde solution used for Cooks #1-3 contained slightly less formaldehyde than that used for Cooks # 4-6. This was also supported by the higher values calculated for the amount of bound formaldehyde per phenol unit. The decrease i n the relative proport ion of methylol groups as the length of time of the cook proceeded and the corresponding increase i n the proport ion of methylene bridges was i n keeping with the accepted mechanisms for the condensation reactions. It was perhaps slightly surprising that small proportions of hemiformal groups were present but no methylene ether bridges were found. The bound F : P ratios found for Cooks #1-3 were i n very good agreement with the known charged ratios. Those for Cooks #4-6 were slightly higher than the believed charged ratios as indicated. The values for the average number of phenol units per molecule were seen to increase as the length of the time of the cook proceeded and thus also the calculated number average molecular weight. These were more indications of the increase i n molecular weight of the resins with longer cooking times. It was interesting to note the differences between the samples of Cook #5 that were freeze-dried and spray-dried. The spray-dried sample contained fewer methylol groups and hemiformal groups and a correspondingly greater proport ion of methylene linkages. This indicated Table 5. Functional groups present in laboratory resins Resin -H -CH20H -Chy CH 2-0-CH 2- -CH2-0-CH2OH F:P n _ * Cook #1 0.27 1.13 1.56 0.00 0.04 2.02 4.53 1044 Cook #2 0.26 1.00 1.69 0.00 0.04 2.00 6.45 1429 Cook #3 0.26 0.98 1.73 0.00 0.02 1.98 7.45 1618 Cook #4 0.09 1.27 1.54 0.00 0.08 2.21 4.36 1061 Cook #5 0.11 1.25 1.56 0.00 0.09 2.20 4.62 1121 Cook #6 0.09 1.11 1.70 0.00 0.09 2.15 6.62 1552 Cook #5 (spray dried) 0.12 1.12 1.71 0.00 0.05 2.07 6.87 1581 * acetylated form 76 that the resin had undergone a greater extent of condensation during the spray-drying procedure than during the freeze-drying procedure. The temperature used i n the spray-dryer reached 200\u00C2\u00B0C. However , the resin droplets sprayed into the drying cyclone would not attain this temperature due to the short residence time. It would appear though, that the temperature attained was sufficient to cause some condensation of the resin. The results of the calculations for the commercial resins are shown i n Table 6. L o o k i n g first at the value calculated for the bound F : P ratio it was seen that a l l but two of the resins had values close to or greater than 2.00. The bound formaldehyde :phenol ratio was not necessarily the same as the charged ratio in the cook since formaldehyde may have been lost, for example, through the Cannizzaro reaction. However, the bound ratio gave a good indication of the conditions used i n the preparation of the resins. This would indicate that they were clearly of the resol type. The two resins which d id not follow this pattern were Res in A and Res in D 3 . The F : P ratio of R e s i n A was only 1.32:1. Res in A appeared to be more intermediate i n nature between a resol and a novolak. It had a much higher propor t ion of unreacted hydrogen sites and a much lower level of free methylol groups through which condensation reactions would occur. Thus Res in A might be expected to be the slowest curing of the resins studied. Res in D 3 was the resin sample that was over seven years old. Its F : P ratio was 1.78:1. It was possible that during this period of time some formaldehyde was lost from the methylol groups. The possibility of the methylol groups undergoing reaction during this period was supported by Res in D 3 containing a greater proport ion of ether bridges than the other resins. W h i l e the formation of methylene ether linkages would not effect the calculation of the bound F : P Table 6. Functional groups present in commercial resins Resin -H -CH2OH -CH 2- -CH 2-0-CH 2- -CH2-0-CH2OH F:P n * M n A 0.93 0.38 1.51 0.18 0.00 1.32 6.45 1137 B 0.25 1.16 1.53 0.01 0.05 2.03 4.31 1001 C 0.41 0.97 1.50 0.07 0.05 1.90 4.65 1030 D1 0.32 1.01 1.53 0.08 0.06 2.00 5.13 1157 D2 0.33 1.07 1.51 0.04 0.05 1.97 4.43 1001 D3 0.51 0.72 1.48 0.24 0.05 1.78 7.26 1486 E 0.13 1.19 1.56 0.04 0.08 2.16 5.02 1193 F1 0.12 1.22 1.48 0.09 0.09 2.22 4.65 1134 F2 0.03 1,32 1.56 0.03 0.06 2.24 4.89 1199 G 0.05 1.25 1.64 0.00 0.06 2.19 5.52 1327 H 0.08 1.03 1.64 0.20 0.04 2.13 12.50 2854 * acetylated form 78 ratio it illustrated that the methylol groups of the resin were capable of reacting during the storage period. The resins which were generally considered to be suitable for use throughout the construction of a panel (Res in C , R e s i n D l and R e s i n D 2 ) had slightly more unreacted hydrogen sites, slightly fewer free methylol groups and slightly fewer cross-linking groups than the core designated resins. Together with the bound F : P ratio this would indicate that the core resins had a higher ini t ial formaldehyde rphenol charge i n the resin reactor and that the cooking procedure was allowed to proceed for a greater length of t ime to produce highly reactive, high molecular weight adhesives. The values for the average number of phenol rings per molecule and the number average molecular weights calculated gave a good indicat ion of the degree of polymerization of the resins. However, it was felt that the molecular weight values obtained from the G P C analyses were more reliable for comparative purposes. 4.2.3. Gel permeation chromatography F r o m the determinations using polystyrene standards each co lumn was found to have a similar number of theoretical plates to that demonstrated by the manufacturer. The value obtained for the four columns i n series was approximately 36,000 plates or based on a unit length, 30,000 theoretical plates per meter. These values compared favourably wi th those found i n the literature (Johnson and Stevenson 1978). A greater number of theoretical plates per unit length indicates a greater separation potential. The calibration curve obtained using the polystyrene standards is shown i n Figure 11. Figure 12 shows a typical detector output and molecular weight distribution for the separation of a commercial resin. In order to 79 CD ioo\u00C2\u00BBoo a> U Q> 10000-o 2 - i 1 1 1 r-18 22 26 30 34 38 Elution Volume (ml) 42 46 Figure 11. Cal ibra t ion curve for the G P C columns, using polystyrene standards. 80 Figure 12. (a) Molecular weight distribution and (b) G P C detector response for Resin A . 81 assign structures to the species indicated by the molecular weight distributions, Table 7 was constructed. The replacement of one hydroxyl group wi th an acetyl group leads to an increase of 42 molecular weight units. Thus phenolic monomers containing one, two and three methylol groups would increase by 84, 126 and 168 molecular weight units respectively i f fully acetylated. The assumptions made were: the resins were fully acetylated, only methylene bridges joined the phenolic nuclei together and the only branches present were protons or methylol groups. Whi ls t it was recognized that these assumptions were not completely val id they simplified the situation sufficiently to allow some deductions to be made about the resins' structures. Table 7. Theoretical molecular weights of acetylated resins number of number of methylol groups phenolic nuclei 0 1 2 3 4 5 6 1 136 194 252 310 2 284 342 400 458 516 3 432 490 548 606 664 722 4 580 638 696 754 812 871 928 5 728 786 844 902 960 1018 1076 6 876 934 992 1050 1108 1166 1224 Thus it was possible to determine the number of phenolic nuclei i n the molecules giving rise to the peaks observed i n the lower molecular weight range. Referr ing to Figure 12, the two quite large, distinct peaks at molecular weight values between 150 and 200 most l ikely corresponded to free phenol and mono substituted phenol. The next peak observed was around a 82 molecular weight of 220. This was attributed to a phenolic ring containing two methylol substituents. The next species observed had a molecular weight of around 260-280. This may have been due to either trisubstituted phenol or a phenolic dimer containing no methylol groups. A large peak at a molecular weight of around 400-430 was attributed to a disubstituted phenolic dimer. A t a molecular weight of 480-520 a large peak was found and attributed to a disubstituted phenolic trimer. A s the molecular weight increased the assignments made to the peaks became less definite since the number of possible species contributing to each peak also increased. It was quite probable that some trisubstituted phenolic dimer was also present i n the peaks between 400 and 520. Another large peak occurred at a value of 630-670. This was attributed to a trisubstituted phenolic trimer (plus possibly some tetrasubstituted phenolic trimer and monosubstituted phenolic tetramer). The final two peaks which could be discerned individually from the detector responses and molecular weight distributions were at values of 850 and 1000-1050. Assignments at these values became quite tenuous but tetrasubstituted phenolic tetramers together with pentasubstituted phenolic tetramers and disubstituted phenolic pentamers may have been responsible for the lower of these two peaks whilst pentasubstituted phenolic pentamers and trisubstituted phenolic hexamers may have made up the peak above 1000. Mos t of the possible species described were l ikely considering the values found for the average number of methylol groups per phenol ring as found by the ^ - N M R analyses. Mos t of the resins averaged around one methylol group per phenol ring. Therefore, the common species would have been expected to be mono-, d i - and tr i- substituted dimers, di- , t r i - and tetra-substituted trimers, etc. 83 The detector responses for the two sets of laboratory cooks are shown i n Figures 13 and 14. It was somewhat easier to discern differences between the cooks by examining the detector responses rather than the molecular weight distributions since the former could be replotted using the M T S mainframe graphics package T E L L A G R A F to allow more direct comparisons. The number average and weight average molecular weights calculated by the integrator for the laboratory resins are shown i n Table 8. It is stressed that the molecular weights given are for the resins i n the acetylated condition. Us ing the figures from the ^ - N M R results it was calculated that the molecular weight of the unacetylated resins would be i n the order of 60-70% of the acetylated values. Table 8. Molecu la r weights of acetylated laboratory resins Res in Mn M w i M n C o o k # l 1264 4853 3.84 Cook #2 1564 7577 4.85 Cook #3 1669 18668 11.19 Cook #4 1126 2484 2.21 C o o k #5 1160 4660 4.02 C o o k #6 1454 5187 3.57 A l l the number average molecular weights calculated from the G P C results were higher than those calculated from the ^ - N M R results except for C o o k #6. F r o m Figure 13 the distinct shift i n the molecular weight distribution of the resins towards the higher molecular weight end of the distribution as the cooking time increased was clearly evident. No t only d id the higher molecular weight peak shift to the left (greater molecular weight) but it also increased i n magnitude relative to the lower molecular weight peaks. 84 E L U T I O N V O L U M E (mL) Figure 13. G P C detector responses for the first series of laboratory cooks. 85 C o o k #4 1 C o o k #5 / ^ \ . A C o o k #6 l 1 1 1 i i r 15 20 25 30 35 40 45 E L U T I O N V O L U M E (mL) Figure 14. G P C detector responses for the second series of laboratory cooks. 86 The decrease i n the peaks at 33.0 m L and 34.5 m L , which corresponded to trimers, was clearly seen to progress as the cooking time increased. Figure 14 shows a similar trend for the resins cooked with a lower in i t ia l charge of alkal i . The effects of the lower concentration of a lka l i and shorter cooking times were readily apparent. The molecular weight distribution was much more narrow and the proport ion of lower molecular weight molecules was much greater than with the first series of cooks. He re again, the shift of the molecular weight distribution to higher values and the decrease i n the lower molecular weight species was evident as the length of t ime of cook proceeded. A s w i l l be seen following, the molecular weight distributions of the second series of cooks, i n which the p H values of the final resins were similar to those of the commercial resins, were much closer i n appearance to those of the commercial resins than the first series. Considering the viscosity results given i n section 4.1. it was observed that for a given viscosity, the resins produced with a higher level of a lkal i had a greater molecular weight than their counterparts made with the lower caustic level. Figure 15 shows the detector responses for the freeze-dried and spray-dried samples of C o o k #5. There was a distinct shift towards the higher molecular weight range i n the spray-dried sample. This confirmed the 1 H - N M R results which indicated a greater degree of polymerizat ion i n the spray-dried sample. The profile of the spray-dried sample was very similar to that of C o o k #6 and it would thus appear that the spray-drying procedure produced an amount of condensation equivalent to approximately 10-15 minutes of cooking time (the difference i n the length of the cooking time of Cooks #5 and #6). 87 E L U T I O N V O L U M E (mL) Figure 15. G P C detector responses for C o o k #5 (a) freeze-dried and (b) spray-dried. 88 Figures 16 -19 show the detector responses for the G P C separations of the commercial resins. The number average and weight average molecular weights calculated for the commercial resins are shown i n Table 9. Table 9. Molecu la r weights for acetylated commercia l resins Res in Mn M w M w / M n A 720 3497 4.86 B 905 2242 2.48 C 742 2710 3.65 D I 714 1964 2.75 D 2 1018 2507 2.46 D 3 1226 13386 10.92 E 1080 2945 2.73 F l 866 1877 2.16 F 2 963 1769 1.84 G 1418 7613 5.39 H 954 4179 4.38 The general profiles of the resins were fairly alike. There were only relatively small proportions of monomers and dimers, especially i n the core resins. The peaks which were most prominent were those at 33.0 m L and 34.5 m L elution volumes which corresponded to trimers. One of the face resins, Res in A , was seen to contain the greatest proport ion of monomers and dimers. Two of the general resins, R e s i n C and R e s i n D I , also contained some of these species. Considering the properties displayed by Res in A and the high number of monomer and dimer molecules that it contained, as indicated by the G P C analyses, it was perhaps surprising at first glance that the M w as calculated by G P C was the fourth highest of the eleven resins. However , on further consideration this apparent anomaly was explained by the spatial conformation assumed by the resin molecules. A s previously indicated, 89 E L U T I O N V O L U M E (mL) Figure 16. G P C detector responses for commercia l resins; Res in A , Res in B and Res in C . 90 1 1 1 1 1 1 r 15 20 25 30 35 40 45 E L U T I O N V O L U M E (mL) Figure 17. G P C detector responses for commercia l resins; Res in D l , Res in D 2 and Res in D 3 . 91 p 1 Res in E y i l 1 1 1 T i Res in F2 i \u00E2\u0080\u0094 , , i \u00E2\u0080\u0094 r\u00E2\u0080\u0094- 1 1\u00E2\u0080\u0094 1 H r 15 20 25 30 35 40 45 E L U T I O N V O L U M E (mL) Figure 18. GPC detector responses for commercial resins; Resin E, Resin F l and Resin F2. 92 Figure 19. G P C detector responses for commercia l resins; Res in G and Res in H . 93 Res in A appeared to be of an intermediate type between that of a resol and a novolak. Novolaks are known to form more linear molecules than resols. The effective molecular diameter of a long, narrow molecule is defined by a sphere with a diameter equal to its length (Figure 20). This diameter is instrumental i n determining the retention time of a molecule i n the G P C column since it dictates the size of pores i n the column packing which the molecule may enter. Thus the effective molecular diameter of a novolak resin molecule w i l l be greater than that of a resol resin molecule of equivalent molecular weight. Al though Res in A may contain molecules of lower molecular weight than the other resins, the linear orientation of a port ion of the molecules may bias the molecular weight calculations as performed i n the G P C analyses towards a high value. The profile of Res in G was interesting since the high molecular weight peak was shifted towards the high range, yet the peaks corresponding to trimers were still very much i n evidence. A s shown from the profiles of the laboratory cooks, as the cooking time lengthened the high molecular weight peak shifted to a higher value and increased i n size relative to the lower molecular weight peaks. The profile of Res in G would suggest that this resin was produced either by a blending process or that addit ional low molecular weight phenolics were added to the cook near the complet ion of the cooking schedule. The production of a blended resin would involve mixing one or more resin cooks in the l iqu id form, pr ior to spray-drying. One cook would be allowed to continue for a comparatively long time to produce the high molecular weight components. Another would be cooked for a shorter time to produce the low molecular species. Figure 17 shows the G P C profiles of three samples of the same resin. R e s i n D l and Res in D 2 were both fresh samples, analyzed eighteen months 94 O H O H , O H novolak Figure 20. Effective molecular diameters of phenolic resin molecules. 95 apart, and Res in D 3 was a sample that had been stored for over seven years. The profiles of Resins D l and D 2 were similar except that Res in D l contained a slightly higher proport ion of monomers and dimers than R e s i n D 2 . Assuming that the formulation of the resin had not changed greatly over time, the G P C profile of Res in D 3 showed that substantial changes were undergone by the resin during storage. Apar t from the two peaks at 38.0 m L and 39.0 m L i n the profile of Res in D 3 , the peaks were not as readily discernable as i n the profiles of the fresh samples. The profile was also shifted towards the high molecular weight region. These were indications that condensation reactions had occurred i n the resin during the storage per iod increasing the molecular weight. This was i n agreement with the * H - N M R results. It was not readily apparent why the two low molecular weight peaks were evident i n Res in D 3 . 4.2.4. Additives 4.2.4.1. Hexamethylenetetramine The values init ially calculated for the H M T A content of the resins are shown i n Table 10. Table 10. H M T A content of commercial resins Res in % H M T A A B C D l D 2 D 3 E F l F 2 G H .5.7 11.9 16.0 12.8 12.1 13.0 6.3 5.7 6.0 6.5 6.0 96 The determination of H M T A i n the resins was complicated by the fact that sodium hydroxide was most l ikely present i n a l l of the resins since it is used as a catalyst i n the resin preparation. This would lead to an increase i n the value for the H M T A content as calculated since the amount of H M T A present was calculated from the volume of acid consumed i n the neutralization of the resin. A n assumption was made that the six resins which had values calculated for H M T A content of around six per cent i n fact contained no H M T A and that the volume of acid consumed i n the titration was solely due to the sodium hydroxide present. Whi ls t it was unlikely that exactly the same amount of sodium hydroxide was present i n a l l the resins, this was assumed to be the case. The pH's of the resins as measured i n water are shown i n Table 11. Table 11. p H values of commercial resins Res in m A 8.16 B 8.56 C 9.02 D I 8.85 D 2 8.34 D 3 8.95 E 8.93 F l 9.15 F 2 8.82 G 8.94 H 9.38 These figures indicated that the resins were a l l quite highly alkaline and of comparable p H . Thus to calculate the H M T A content of the other five resins this figure of six per cent was then subtracted from the original figure calculated. The recalculated values of the H M T A content of the resins are shown i n Table 12. 97 Table 12. Corrected H M T A content of commercial resins Res in % H M T A A B C D l D 2 D 3 E F l F 2 G H 0.0 5.9 10.0 6.8 6.1 7.0 0.0 0.0 0.0 0.0 0.0 The figure for Res in C was very close to that given by the manufacturer and it was thus decided that the method used above to calculate the H M T A content was val id. These results confirmed the I R results which indicated the presence of H M T A i n Resins B , C , D l and D 2 . The presence of H M T A was not indicated by the I R spectrum for R e s i n D 3 . This was most l ikely due to the generally less distinct nature of this spectrum i n comparison to the others. This was the resin that had been stored for at least seven years and the condensation reactions which had probably occurred over this period led to the less distinct spectrum observed. 4.2.4.2. Inclusions The I R spectra of the white particles found i n the commercia l resins were a l l similar. Figure 21 shows the spectrum of those from R e s i n F 2 . It was unclear as to what purpose these additives served but it was thought that they might act as flow promoters. References have been found to the suggested nature of compounds that may act as flow promoters. Compounds suggested for use i n l iquid plywood resins have included polyhydroxy compounds, tricresyl phosphate, tributyl phosphate, ethylene glycol and 98 4000 3500 3000 2500 2000 1500 1000 CBH 500 Figure 21. I R spectrum of inclusions from R e s i n F 2 . 99 certain modified phenolics (Sellers 1985). W h i l e no positive identification was made from Figure 21, the peaks at 3500 cm\" 1 and 1100 cm\" 1 indicate the presence of hydroxyl groups. These flow promoters probably act by interfering with the hydrogen bonds through which individual resin molecules are associated into larger entities. A polyhydroxy compound might occupy a number of adjacent potential hydrogen bonding sites on a resin molecule so that the associations which tend to inhibit flow could not occur as readily. Alternat ively a flow promoter molecule with hydroxyl groups at each end might act by becoming involved with two resin molecules, pushing them apart to the extent that other potential sites d id not become involved i n hydrogen bonds. This would also reduce the number of hydrogen bonds between the resin molecules making them more l ikely to flow upon the application of heat and pressure. 4.2.5. Thermal mechanical analysis 4.2.5.1. Laboratory resins The T M A profiles of the two series of laboratory resins are shown i n Figures 22 and 23. The extent of the flow of the resins was seen to decline sharply from C o o k #1 to Cook. #3. This coincided with an increase i n the molecular weight of the resins. A s the molecular weight of the resins increased during the cook, the proport ion of the smaller molecules decreased. Considering the model for flow previously described, it would appear logical that as the amount of low molecular weight species decreased there became less potential for flow of the resin. A similar trend was observed for Cooks #4, #5 and #6. Al though the decrease i n flow with increasing molecular weight was not as dramatic as i n the first series (corresponding to the less vivid differences i n molecular weight distributions), it was still evident. The greater flow of the second 1 0 0 Figure 22. TMA profiles for the first series of laboratory cooks. 101 0 20 40 60 80 100 120 140 160 180 200 T E M P E R A T U R E (\u00C2\u00B0C) Figure 23. T M A profiles for the second series of laboratory cooks. 102 series of cooks i n comparison to the first was i n keeping with their greater content of low molecular weight species. In the second series of cooks, two different observations about the flow profiles were made. Firstly, the final level of the probe was lower for the resin of lowest molecular weight, showing that more flow had occurred, Secondly, i n the temperature range of 40-60\u00C2\u00B0C, where the majority of the flow occurred i n these resins, the rate of drop i n the level of the probe was more rapid for the lower molecular weight resin. O f course, it was highly l ikely that these two factors are related. A s the temperature increased, cross-l inking of the melted resin to an infusible solid would occur. Since the higher molecular weight resins flowed more slowly at first, it may have been the case that the resins set before they had achieved their full potential for flow. It had been hoped to determine the effect of the moisture content of the laboratory resins on their flow properties as measured by T M A . This property may be an important factor l imit ing or promoting flow depending upon its level. The moisture content of the powdered resins at the glueline would be expected to rise as hot pressing proceeds. A s steam is generated from the moisture present i n the wood furnish it migrates towards the center of the board where it would interact with the resin. However , no suitable method of introducing moisture to the resin i n the T M A system i n a manner comparable to that experienced i n practice was available. Resins were conditioned i n different relative humidity environments to impart various moisture contents. Resins of different moisture contents were obtained but a visual inspection indicated that other changes had occurred. A t the higher moisture contents the resins had darkened considerably. A few of the resin samples were acetylated and analyzed by G P C and a distinct shift to the higher range of the molecular weight distribution was apparent. This shift 103 was noted especially for the lower molecular weight cooks. A t the highest relative humidity employed (78%) Cook #4 deliquesced and C o o k #5 clearly lost its properties of a free flowing, dry powder. 4.2.5.2. Commerc ia l resins The T M A profiles of the commercial resins are shown i n Figures 24 -27. R e s i n A shows a profile different from a l l the other resins. The other resins exhibited a one-stage flow process whereby the probe fell (at varying rates to different levels) to one level where it remained. R e s i n A exhibited two distinct stages of flow. The first took place over the temperature range 55-70\u00C2\u00B0C and leveled off at around 100\u00C2\u00B0C. A t 110\u00C2\u00B0C another drop i n the level of the probes occurred, of equal magnitude to the first and was complete by 130\u00C2\u00B0C. Res in A was seen to contain a high proport ion of monomers, dimers and trimers relative to the other resins (from the molecular weight distributions) and this would help to explain the ini t ial rapid flow. A s indicated by the ^ - N M R results, the resin contained fewer methylol groups through which cross-linking might occur than did the other resins. Thus at the temperature of 110\u00C2\u00B0C the resin had not fully undergone cross-linking to an infusible state and was able to undergo the observed second stage of flow. These two stages of flow were observed i n novolak resins by Katov ic and Stefanic (1985). In their resin the first level of flow occurred at approximately 95\u00C2\u00B0C and was attributed to the visual melting of the adhesive. The second level of flow occurred at a temperature of approximately 110-115\u00C2\u00B0C. However, the authors proposed no mechanism for the nature of this second level of flow. Res in C also appeared to exhibit this two-stage type of flow but to a much lesser degree. F r o m the * H - N M R analyses this resin was seen to contain slightly fewer methylol groups than the other resins and might be expected to cure more slowly. In fact, flow was not complete for 1 0 4 Figure 24. TMA profiles for commercial resins; Resin A, Resin B and Resin C. 1 0 5 Figure 25. TMA profiles for commercial resins; Resin DI, Resin D2 and Resin D3. 1 0 6 Figure 26. TMA profiles for commercial resins; Resin E, Resin F l and Resin F2. 107 Figure 27. T M A profiles for commercial resins; Resin G and Resin H . 108 this resin unt i l approximately 140\u00C2\u00B0C. However, the H M T A content of the resin (10.0%) would tend to increase the rate of cure that might be expected. The Resins D I , D 2 , E , F l , F 2 and G a l l showed fairly similar flow profiles. The relative rates at which they flowed, the temperatures at which the most rapid flow occurred and the final level of the probe were a l l i n good agreement with the G P C results. The resins with a greater content of low molecular weight species tended to exhibit greater and more rapid flow. Resins D 3 and H both showed very little flow at a l l from the T M A profiles. This was to be expected considering the relatively small proportions of low molecular weight species that they contained. As ide from R e s i n A (which may have been an anomaly as explained earlier) these are the only resins for which the peaks at 33.0 m L and 34.5 m L on the G P C detector output were lower i n height than the high molecular weights peak (25.0 -30.0 m L ) . The flow of Res in D 3 was greatly reduced compared to the fresh samples of the same resin. Thus the condensation reactions that had occurred during storage manifested themselves i n the reduced flow observed. 4.2.6. Fus ion diameter The determination of the fusion diameter of a resin is an empir ical evaluation and as such is a much less sophisticated method for the measurement of flow than the T M A instrumentation. However , the determination of the fusion diameter is a relatively quick procedure which gives a good indication of the flow properties of the resins. In Figure 28 the results of the fusion diameter determination are plotted against the number average molecular weight ( M n ) of the resins. The decrease i n the fusion diameter as the molecular weight increased was i n agreement with the flow characteristics indicated by the G P C and T M A analyses. In general, for a 109 48-, 44- X 40H 36 X o aj UL T3 O O 5 100 76 60 26 4 6 8 10 M o i s t u r e C o n t e n t (%) c ) 120 S e c o n d s 100-C 76-Wood Failure cn o . . i . , , . i . . . o \u00E2\u0080\u00A2 O ^ ^ ^ ^ 3 n U i i i i | i i i | . i i | 8 8 10 M o i s t u r e C o n t e n t (%) 100 C 76 CD 3 CO LL T3 O O 5 60 26 d ) 2 4 0 S e c o n d s 4 6 8 10 M o i s t u r e C o n t e n t (%) Figure 30. Wood failure against moisture content for aspen bonded with laboratory resins tested in the soaked condition. a) 75 seconds b) 90 seconds c) 120 seconds d) 240- seconds o cook 13 \u00E2\u0080\u00A2 Cook \u00C2\u00BB4 \u00E2\u0080\u00A2 Cook \u00C2\u00AB5 \u00E2\u0080\u00A2 Cook #6 1 2 2 a) 75 s e c o n d s 100 ^ 76 a> X3 O O 5 50-26 4 6 8 10 M o i s t u r e C o n t e n t (%) b) 9 0 S e c o n d s 100 C 76 a> D LL XJ O O 5 60 26 4 6 8 II M o i s t u r e C o n t e n t (%) c ) 120 S e c o n d s 100-i 76-- 60H \u00E2\u0080\u00A2o o o 5 26 10 M o i s t u r e C o n t e n t (%) d ) 2 4 0 S e c o n d s 100 ^ 75 a) 3 O O 5 60 25 \u00E2\u0080\u0094I 10 M o i s t u r e C o n t e n t (%) Figure 31. Wood failure against moisture content for spruce bonded with laboratory resins tested in the soaked condition. _ \u00E2\u0080\u009E \ nc j \u00E2\u0080\u00A2 Cook #4 a) 75 seconds b) 90 seconds D c) 120 seconds \u00E2\u0080\u00A2 cook \u00C2\u00BBe d) 240 seconds o cook \u00C2\u00BB3 123 moisture content of the wood rose Cook #3 produced much greater wood failure. The observations from these analyses and Figures 30 and 31 can be summarized as follows: - greater wood failure was observed with longer pressing times, - greater wood failure was observed with spruce than wi th aspen for a given set of conditions, - greater wood failure tended to be observed at higher moisture contents, - the general decreasing order of wood failure values for the four laboratory resins was Cook #6, C o o k #5, C o o k #4 and C o o k #3, - the wood failure values for C o o k #3 appeared especially dependent on moisture content. 4.3.1.2. Commerc ia l resins Table 19 shows the A N O V A ' s performed on the wood failure values for the samples from the boards bonded with the commercial resins. Figures 32 and 33 show the untransformed wood failure values to aid i n the interpretation of the A N O V A results. After 75 seconds a l l of the sources of variat ion were significant at the 9 5 % confidence level except the S x R x M three-way interaction and the S x M two-way interaction. Figure 32a and 33a illustrate that the only resin to produce any wood failure was Res in G . Spruce produced greater wood failure than aspen. After 90 seconds a l l sources of variation were significant except for the S x M two-way interaction and the M main effect. Figure 32b and 33b illustrate that Res in G produced the greatest wood failure, followed by Res in F 2 with Res in D 2 and Res in B producing comparable values. 124 Table 19. Analysis of variance tables for transformed wood failure values for both species bonded with commercial resins tested i n the soaked condit ion Source of Sum of D F M e a n F -Ra t io P Var ia t ion Squares Square a) 75 seconds pressing time S 4160.7 1 4160.7 18.26 0.000 R 27848.0 3 9282.7 40.74 0.000 M 3212.7 2 1606.3 7.05 0.001 S x R 12482.0 3 4160.7 18.26 0.000 S x M 694.4 2 347.2 1.52 0.220 R x M 9638.1 6 1606.3 7.05 0.000 S x R x M 2083.1 6 347.2 1.52 0.172 E r r o r 43751.1 192 227.9 b) 90 seconds pressing time S 32659.0 1 32659.0 31.50 0.000 R 91401.6 3 30467.2 29.39 0.000 M 2148.5 2 1074.2 1.04 0.357 S x R 8361.6 3 2787.2 2.69 0.048 S x M 4527.4 2 2263.7 2.18 0.115 R x M 16124.0 6 2687.3 2.59 0.019 S x R x M 16951.8 6 2825.3 2.73 0.015 E r r o r 199059.1 192 1036.8 a) 120 seconds pressing time S 60267.0 1 60267.0 80.20 0.000 R 14257.4 3 4752.5 6.32 0.000 M 1810.8 2 905.4 1.21 0.302 S x R 14387.7 3 4795.9 6.38 0.000 S x M 1926.4 2 963.2 1.28 0.280 R x M 4941.0 6 823.5 1010 0.366 S x R x M 16040.7 6 2673.4 3.56 0.002 E r r o r 144285.1 192 751.5 b) 240 seconds pressing time S 1906.3 1 1906.3 6.51 0.012 R 2881.5 3 960.5 3.28 0.022 M 1251.9 2 626.0 2.14 0.121 S x R 566.3 3 188.8 0.64 0.587 S x M 2189.0 2 1094.5 3.74 0.026 R x M 8776.6 6 1462.8 4.99 0.000 S x R x M 12503.4 6 2083.9 7.11 0.000 E r r o r 56242.4 192 292.9 S - species, R - resin, M - moisture content. 1 2 5 \u00E2\u0080\u00A2o o o 5 a) 75 seconds 100 CD K\u00E2\u0080\u0094 3 60 26 4 6 B 10 Moisture Content (%) 100 C 75 td u. O O 5 b) 90 Seconds 4 e 8 10 Moisture Content (%) T3 o o 5 c) 120 Seconds 100 C 76 CD v. 3 \u00E2\u0080\u0094 60 26 4 6 8 10 Moisture Content (%) d)240 Seconds 100-C 76 a) 60 o O 26 5 4 6 8 10 M o i s t u r e C o n t e n t (%) c ) R e s i n F2 100-i ^ 76 o o 5 26 4 6 8 10 M o i s t u r e C o n t e n t (%) d) R e s i n G 100 76 ID 3 X) o o 5 60 26 4 6 M o i s t u r e C o n t e n t Figure 37. Wood failure against moisture content for aspen bonded with commercial resin tested in the soaked condition. _ . _ \u00E2\u0080\u00A2 76 \u00E2\u0080\u00A2\u00E2\u0080\u00A2cond\u00C2\u00AB Resin B D 90..cond. b) Resin D 2 \u00E2\u0080\u00A2 120 . . c o n d . c) Resin F 2 o 240 t\u00C2\u00ABconn. d) Resin G 137 100 C 75 50 \u00E2\u0080\u00A2o O O 5 25 a) Resin B T 4 6 e 10 Moisture Content (%) 100 C 76-a> 3 \u00E2\u0080\u00A2- 60-X) O O 5 26 b) Resin D2 4 e 8 10 Moisture Content (%) c) Resin F2 100 \u00C2\u00A3 76 a> i _ 3 = 60H M U. XI o O 26 5 4 e 8 10 Moisture Content (%) d) Resin G 100 75 - 60 OS u. X) o o 5 26-10 Moisture Content (%) 76 >ccond> Figure 38. W o o d failure against moisture content for spruce bonded with commercial resins tested i n the soaked condit ion. a) Res in B b) R e s i n D 2 c) R e s i n F 2 \u00C2\u00B0 2 4 0 \" c < \" \" \" d) R e s i n G D 90 n c o n d i 0 120 second* 138 The expected increase with longer pressing times was also evident for the commercial resins (Figures 37 and 38). Resins B and D 2 performed similarly producing generally lower wood failure than Resins G and F 2 . 4.3.2. Tension shear strengths Table 20 shows the average tension shear strengths for both species tested after the three sample treatments. The averages were calculated from the samples bonded with both the laboratory and the commercia l resins. The first observation was that the strength values decreased as the sample treatments became more severe. This was expected since the b o i l / d r y / b o i l cycle would be expected to weaken the wood to the greatest extent of the three treatments. The pressure/soak cycle would also weaken the wood compared to the dry conditioning per iod but to a lesser degree than boi l ing. The second observation was that at the two shorter pressing times (75 and 90 seconds) the spruce samples produced greater strengths than the aspen samples. However, at the two longer pressing times (120 and 240 seconds) the aspen samples produced the greater strengths. This is i n good agreement with the slower buildup of wood failure observed for aspen compared to spruce. In the green condition the shear value of aspen is approximately 5 - 1 0 % greater than that of spruce (Jessome 1977). The third observation was that for the spruce samples tested i n the dry condit ion the greatest strength was observed after 120 seconds pressing time whereas there was no significant difference between the strength values after 120 and 240 seconds for the samples tested after the soak and the bo i l treatments, Figure 34 showed that after 120 seconds the temperature of the glueline was 140\u00C2\u00B0C and after 240 seconds it had risen to 170\u00C2\u00B0C. A t these temperatures some thermal breakdown of the wood might be expected. Thus 139 Table 20. Average tension shear strengths (calculated from data from both laboratory and commercial resins) Sample Treatment Pressing T ime (seconds) Average Tensior ( M l Spruce L Shear Strength Pa) A s p e n D r y 75 90 120 240 0.85 C 1.83 B 2.08 A 1.98 B 0.90 C 2.21 B 2.73 A 2.66 A Soaked 75 90 120 240 0.71 C 1.66 B 1.78 A 1.68 A B 0.45 C 1.35 B 1.89 A 1.99 A B o i l e d 75 90 120 240 0.65 C 1.49 B 1.67 A 1.62 A 0.47 C 1.34 B 1.74 A 1.86 A F o r each species and sample condition, means with the same letter are not significantly different at the 95% confidence level. 140 i n the time from 120 to 240 seconds pressing time the wood became sufficiently degraded to produce the lower strengths observed. The temperatures of the gluelines i n industrial pressing situations are unlikely to attain values at high as these. The center glueline i n plywood, that is normally tested, would not reach these values since it is a greater distance from the platens than the glueline here. In waferboard, pressing is complete before core temperatures reach values this high. This phenomenon was not observed with aspen which would again appear to be a result of the slower bu i ld up of strength and wood failure. 4.4. Detection of Resin Penetration into W o o d 4.4.1. Electron microscopy Figures 39 to 42 show the electronrnicrographs, l ine scans and dot maps taken of the samples observed from series 1 (molecular weight, aspen). In Figure 39 the flow of C o o k #4 was seen to be such that the resin was found i n the lumens of vessel elements over 120 / i m away from the glueline. The cells at the glueline appeared to be crushed to a greater degree than with the higher molecular weight resins but a good intermingling of the resin with the wood had occurred. F r o m the line scan, i f the peaks are assumed to represent the bromine concentration i n the cel l lumens and the troughs that i n the cel l walls, it was observed that along the line scanned five to seven cells are intimately associated with the resin. A s the molecular weight of the resins increased, less resin was found at any distance from the glueline. In addition, the extent of penetration into the cells immediately adjacent to the glueline became less with increasing molecular weight. W i t h the resin of greatest molecular weight (Cook #3) it appeared that no flow of the resin had occurred down any of the cel l lumens and that no penetration of the cel l walls at the glueline had occurred. 141 Figure 39. Electronmicrograph and dot map for the molecular weight series, aspen. Species - Aspen Mois ture content - 6.7% Pressing time - 240 seconds Res in - Cook #4 1 4 2 143 Figure 40. Electronmicrograph and dot map for the molecular weight series, aspen. Species - A s p e n Mois ture content - 6.7% Pressing time - 240 seconds R e s i n - Cook #5 144 1671 261 i g 0 u M 145 Figure 41. Electronmicrograph, l ine scan and dot map for the molecular weight series, aspen. Species - A s p e n Mois ture content - 6.7% Pressing time - 240 seconds Res in - C o o k #6 146 147 Figure 42. Electronmicrograph, line scan and dot map for the molecular weight series, aspen. Species - A s p e n Moisture content - 6.7% Pressing time - 240 seconds R e s i n - C o o k #3 148 0U685 20KV 149 Referr ing to Figures 39 and 40 the appearance of resin i n the lumens of vessel elements some distance from the glueline was most l ikely due to flow down the vessel rather than from the glueline through the cel l walls of the adjacent cells. Since it was highly unlikely that the longitudinal axis of the vessels was paral lel to the glueline due to the natural variat ion i n the grain angle of the wood, the vessels i n effect would act as channels along which the resin could flow away from the glueline. The possibility existed i n this situation that i f the ability of the resin to flow was great enough then excessive amounts of resin would flow away from the glueline down the vessels, effectively starving the glueline of sufficient adhesive to form a bond. This effect would be more pronounced i n the production of waferboard or O S B where the longitudinal axis of the vessels would be more randomly arranged than i n the wood used i n this study and thus greater access to the channels of flow would be available to the resin. F o r the resins observed the flow of resin away from the glueline d id not occur to the extent that starvation of the glueline occurred since sufficient resin was observed at the gluelines. However , these observations indicated that i n resins of high flow a l l the resin applied may not be contributing to the bonding process. If a means of al lowing the flow of resin into the cells and cel l walls of the cells adjacent to the glueline but preventing the removal of resin away from the glueline down the vessel lumens could be developed, then the resin content would be used more efficiently and it might allow a lower level of resin loading to be used while still attaining satisfactory bonds. Figures 43 and 44 show the electron micrographs and l ine scans for series 2 (molecular weight, spruce). A similar trend was observed to that i n aspen. A s the molecular weight of the resin increased, flow to the cells 150 Figure 43. Electronmicrographs and line scans for the molecular weight series, spruce. Species - Spruce Mois ture content - 7.0% Pressing time - 240 seconds Res in - a) Cook #4 b) C o o k #5 151 152 Figure 44. Electronrnicrographs and line scans for the molecular weight series, spruce. Spruce 7.0% 240 seconds a) C o o k #6 b) C o o k #3 Species Moisture content Pressing time R e s i n 153 154 adjacent to the glueline decreased unti l with the highest molecular weight resin min imal penetration was observed. However , no resin was observed i n the lumens of cells any distance from the glueline as i n aspen. This was attributed to the relative sizes of the conducting elements i n each species. The lumens of the longitudinal tracheids of the spruce were 20-30 jum i n tangential diameter. The tangential diameters of the lumens of the vessel elements and l ibr i form fibers of the aspen were 45-60 fim and 10-25 jum, respectively. The presence of resin was detected i n vessel elements some distance from the glueline but none was detected i n the lumens of either the longitudinal tracheids or l ibr i form fibers. It appeared that a cri t ical diameter of conducting element was required before flow of the resin would occur. Vessel elements exceeded this size but longitudinal tracheids and l ibr i form fibers d id not. It has been observed ( G o 1988) that the m i n i m u m resin content required for the successful production of waferboard or O S B is species dependent, with resinous species requiring less synthetic resin. This may be attributed to the resinous wood extractives playing some positive role i n the bonding process but some differences between hardwood and softwood synthetic resin requirements may be due to the relative size of their conducting elements which may provide means for the resin to migrate from the glueline. Figures 45 and 46 show the electronrnicrographs for series 3 (pressing time, low molecular weight). N o information was available for the shortest pressing time (75 seconds) since the samples fell apart during preparation. F r o m the line scans it appeared that the majority of resin penetration had occurred after 90 seconds but that it continued to a lesser extent for the remainder of the pressing time. This was supported i n part by the increasing 155 Figure 45. Electronmicrographs and resin, pressing time series. Species Mois ture content Res in Pressing time scans for the low molecular weight Spruce 7.0% C o o k #4 a) 90 seconds b) 120 seconds 156 157 Figure 46. Electronmicrograph and resin, pressing time series. Species Mois ture content R e s i n Pressing time scan for the low molecular weight Spruce 7.0% C o o k #4 240 seconds 158 159 wood failure values with pressing time observed for this series. A t pressing times of 90,120 and 240 seconds the percentage wood failure values were 82, 95 and 98 respectively. This increase could also be attributed to the greater cure of the resin with longer pressing times. Figure 34 showed that after 75 seconds the temperature of the glueline i n spruce pressed at 7.0% moisture content was approximately 122\u00C2\u00B0C. F r o m Figure 23 the flow of C o o k #4 was seen to occur pr imari ly over the temperature range of 40-60\u00C2\u00B0C and to have effectively ended by 120\u00C2\u00B0C. This would indicate that after 75 seconds sufficient flow of the resin had occurred to allow bond formation but that insufficient cross-linking of the resin molecules had occurred to form a permanent bond. F r o m the flow profile, little additional flow would have been expected after 90 seconds but the extra time would have allowed the cross-linking reactions to proceed far enough to allow bond formation to occur. In making comparisons between the temperature levels experienced i n the boards and those experienced i n the T M A it should be noted that the rate of temperature increase i n the T M A was 10\u00C2\u00B0C /min while Figure 34 shows the temperature of the board to rise by approximately 90\u00C2\u00B0C during the first minute of pressing. Figures 47 and 48 show the electronmicrographs and line scans for series 4 (pressing time, high molecular weight). N o apparent penetration of the high molecular weight resin into the wood occurred throughout the 240 seconds pressing time. The T M A profile i n Figure 22 showed that C o o k #3 exhibited very little flow over the whole temperature range scanned but that what little flow did exist continued throughout the temperature range. The percentage wood failure values after 75, 90, 120 and 240 seconds were 0, 12, 5 and 27 respectively. He re again, this general increase with longer pressing 160 Figure 47. Electronmicrographs and line scans for the high molecular weight resin, pressing time series. Species Mois ture content Res in Pressing time A s p e n 6.7% Cook #3 a) 75 seconds b) 90 seconds 161 162 Figure 48. Electronmicrographs and line scans for the high molecular weight resin, pressing time series. Species - A s p e n Mois ture content - 6.7% Res in - C o o k #3 Pressing time - a) 120 seconds b) 240 seconds 163 (b) 164 time may have been due to the increase i n temperature with time leading to a greater extent of cross-linking. However, the observation that the flow of this resin had not ceased at temperatures above 100\u00C2\u00B0C might indicate that as the pressing time increased additional flow of the resin d id occur contributing to the greater wood failure values recorded. Figure 49 shows the electronrnicrographs and line scans for the cel l walls of l ibr i form fibers i n aspen. In each electronmicrograph the \"x\" indicates the visually observed boundary between the cel l wa l l and the resin. The fall i n the concentration of bromine across the cel l wa l l i n the case of the low molecular weight resin (Figure 49a) was much less sharp than that for the high molecular weight resin (Figure 49b). This indicated that some molecules of the low molecular weight resin had penetrated the cel l wa l l of the l ibr i form fiber. Ve ry little, i f any, penetration of the cel l wa l l had occurred with the high molecular weight resin. This was i n agreement with its G P C profile which indicated a very low proport ion of low molecular weight species. Tarkow et al (1966) found that the crit ical molecular weight below which polyethylene glycol could penetrate the cel l wal l of green Si tka spruce was 3000. W h i l e this value was likely to differ for polymers of different densities and for different species it was still of interest to compare the relative proportions of each of the two resins observed that were below this crit ical value. Figure 50 shows the molecular weight distributions of C o o k #4 and Cook #3. F r o m the cumulative plot it was found that about 80% of the molecules of Cook #4 were smaller than 3000 while only about 4 0 % of the molecules of C o o k #3 were below this value. Thus C o o k #4 would be expected to exhibit more penetration of the cel l wa l l than C o o k #3, as was observed. If a lower crit ical level were chosen, say 1000 (at which much 165 Figure 49. E lec t ron micrographs and l ine scans at high magnification, showing the extent of cell wal l penetration. Species Moisture content Pressing time Res in A s p e n 6.7% 240 seconds a) C o o k #4 b) Cook #3 166 1 6 7 Figure 50. Molecular weight distributions for (a) Cook #4 and (b) Cook #3. 168 greater flow would occur than at 3000), it was observed that about 4 5 % of the molecules of C o o k #4 but only about 2 0 % of the molecules of C o o k #3 were below this value. Figure 51 shows the electronmicrographs for the different types of failure observed i n the shear specimens. The ideal type of failure is that shown i n Figure 51b which illustrates 100% wood failure. Here , the wood failed a number of cells away from the glueline rather than the glueline itself. Figure 51a illustrates a type of wood failure referred to as \"fine fiber\" failure. Here , failure occurred at the glueline or i n the row of cells immediately adjacent to the glueline. The resin is believed to have penetrated the cel l walls of the cells adjacent to the glueline only minimal ly i n this case so that only very shallow failure occurred. The fine fiber failure is shown at higher magnification i n Figure 51c and d. It appeared almost fluffy i n nature to the naked eye. Figure 52 shows the electronmicrographs for series 5 (commercial resins). The extent of penetration was seen to decrease i n the order R e s i n B , R e s i n D 2 , Res in F 2 and Res in G . Res in B was one of the face resins and was seen to exhibit rapid flow at a temperature lower than the other resins (Figure 24). Thus the amount of penetration was to be expected. Res in D 2 had a similar molecular weight profile to that of R e s i n B and exhibited flow at a temperature approximately 10\u00C2\u00B0C higher than Res in B . Thus it also would be expected to exhibit the relative amount of penetration observed. Res in F 2 contained a greater proport ion of trimers than did Res in D 2 but the T M A profile demonstrated less flow. This may have been due to the higher methylol content of Res in F 2 (1.32 methylol groups per phenol ring) compared to Res in D 2 (1.07 methylol groups per phenol ring) which would tend to cause Res in F 2 to cure more rapidly than Res in D 2 . However , the 169 Figure 51. E lec t ron micrographs of types of bond failure. Species Mois ture Content Pressing T i m e Res in Spruce 7.0% 240 seconds a) C o o k #3 b) C o o k #4 c) C o o k #3 d) C o o k #3 170 171 Figure 52. Electronrnicrographs for the commercial resin series. Species - Spruce Mois ture content - 7.0% Pressing time - 240 seconds Res in - a) Res in B b) Res in D 2 c) Res in F 2 d) Res in G 172 173 presence of H M T A (6.1%) i n Res in D 2 would increase its rate of cure. Res in G exhibited the lowest amount of penetration of the four commercial resins. This resin was the one that appeared to be a blend of cooks. It contained some distinct low molecular weight peaks but also a pronounced high molecular weight peak. Its T M A profile indicated the lowest level of flow of the four which was i n agreement with its G P C profile and relatively high methylol group content (1.25 methylol groups per phenol ring). 4.4.2, Photomicroscopy W h i l e the white light photomicrographs d id not give information about the penetration of the laboratory and commercial resins at the same resolution as the electronmicrographs, they did provide a good indicat ion of the level of gross penetration of the resin into the cells close to the glueline. Figure 53 shows the photomicrographs for aspen bonded with the laboratory resins. A s was observed under the S E M , i n the lower molecular weight resins the resin was seen to have flowed down the lumens of the vessel to the extent that it was observed as far away as 300 /um from the glueline (Figure 53a). A s was expected, as the molecular weight qf the resins increased the penetration into the cell walls of the cells immediately adjacent to the glueline and down the vessel lumens decreased. Cook #3 was observed to have undergone very little flow and remained as an immobi le mass between the two wood surfaces. Figure 54 shows the photomicrographs for aspen bonded with the commercial resins. Resins B , D 2 and F 2 exhibited a high degree of flow with apparent penetration of the cel l walls at the glueline occurring, i n addit ion to flow down the vessel lumens. R e s i n G exhibited little flow but d id not appear to have been as immobi le as C o o k #3. Figure 55 shows the photomicrographs of spruce bonded with the laboratory resins. Once again the penetration of the resins into the cell walls 174 Figure 53. Photomicrograph of gluelines i n aspen bonded with laboratory resins (xl25) . Species - A s p e n Moisture content - 6.7% Pressing time - 240 seconds R e s i n - a) C o o k #4 b) Cook #5 c) C o o k #6 d) Cook #3 175 176 Figure 54. Photomicrograph of gluelines i n aspen bonded with commercial resins (xl25). Species - A s p e n Mois ture content - 6.7% Pressing time R e s i n 240 seconds a) Res in B b) Res in D 2 c) Res in F 2 d) Res in G 177 178 Figure 55. Photomicrograph of gluelines i n spruce bonded with laboratory resins (xl25). Species - Spruce Moisture content - 7.0% Pressing time R e s i n 240 seconds a) Cook #4 b) C o o k #5 c) Cook #6 d) C o o k #3 179 180 Figure 56. Photomicrograph of gluelines i n spruce bonded with commercia l resins (xl25). Species - Spruce Mois ture content - 7.0% Pressing time Res in 240 seconds a) Res in B b) Res in D 2 c) Res in F 2 d) Res in G 181 182 adjacent to the glueline was seen to decrease with an increase i n molecular weight. The flow of resin down the lumens of cells away from the glueline was not observed, confirming the S E M observations. Figure 56 shows the photomicrographs of spruce bonded with the commercial resins. Penetration decreased i n the order Res in B , Res in D 2 , Res in F 2 and R e s i n G . 4.5. Summary of Resin Penetration and Wood Bonding It is desirable to be able to use a single resin characteristic to describe or predict the resin's penetration and subsequent development of wood failure. This might enable a rapid evaluation of the bonding potential of a resin to be performed by means of a single measurement or observation. However , the interelated nature of many of the characteristics of a resin make this a im difficult to achieve. The various potential measures such as F : P ratio, methylol content, M n , M w , molecular weight distribution, fusion diameter, stroke cure and T M A profile are a l l intimately associated. It is almost impossible to predict the adhesive potential of a resin from a single factor from amongst these. Fo r example, a resin might produce a small fusion diameter due to the number of large molecules present which might indicate a poor bonding potential. However, i f sufficient small molecules are present adequate penetration might still be achieved. Alternat ively a resin might have a low M w or M n perhaps indicating that it w i l l flow sufficiently but i f the resin contained a very high proport ion of methylol groups (i.e., it was very reactive), rapid condensation reactions might occur before adequate flow had occurred. The penetration studies indicated that resins containing greater proportions of low molecular weight species produced greater penetration and high wood failure. Thus the best indicator of bonding potential as measured by wood failure may be one concerning the relative proport ion of 183 low molecular weight species. Whi l e it would be possible to obtain values for the proport ion of molecules present below a certain molecular weight value, from figures such as Figure 50, the value chosen would be arbitrary. Thus it was decided to plot wood failure against M n . O f the molecular weight descriptors, M n is the one that most emphasizes the smaller molecular weight species present. Figure 57 shows the wood failure values plotted against the number average molecular weight of the four laboratory resins for both species at each of the four pressing times. It was observed that C o o k #6 ( M n = 1454) generally produced high wood failure values at shorter pressing times than the other resins. C o o k #5 ( M n = 1160) was slightly more rapid than C o o k #4 ( M n = 1129) i n developing high wood failure values. The wood failure values for C o o k #3 ( M n = 1669) were seen to be lower than the other resins once cure had been fully achieved. He re the relationship between rate of cure and extent of flow becomes important. The observations of resin penetration indicated that C o o k #6 contained enough small molecules to bring about sufficient penetration to cause high wood failure values. C o o k #5 contained a greater proport ion of these molecules and Cook #4 even more. The key issue appears to be that Cook #6 contained a sufficient proport ion of these molecules. A t the longest pressing time, although Cooks #4 and #5 did contain more of these species, maximum wood failure values possible were attained with a lower proport ion of small molecules and thus Cooks #4 and #5 contained an excess over that which was required. Greater penetration was observed with Cooks #4 and #5 but this penetration was above the min imum level required to produce good bonds. 1 8 4 a) 75 seconds 100 \u00C2\u00A3 : 76 X) O O 5 26 0 -} \u00E2\u0080\u00A2 a 1100 1300 _ 1500 M\u00E2\u0080\u009E c) 120 Seconds 100 1100 1300 _ 1500 1700 d)240 Seconds 100 C 7 6 -

(sec) (MPa) (%) (%) (sec) (MPa) (%) Cook #4 5.0 75 0.09 0 Cook #5 5.0 75 1.00 41 90 1.62 45 90 2.10 55 120 2.21 85 120 1.61 99 240 1.58 83 240 1.76 96 7.0 75 0.47 24 7.0 75 0.73 33 90 1.66 75 90 1.77 78 120 1.90 93 120 1.85 98 240 1.47 99 240 1.64 100 9.5 75 0.20 0 9.5 75 1.23 34 90 1.87 83 90 1.85 94 120 1.73 81 120 1.57 99 240 1.59 96 240 1.94 88 Cook #6 5.0 75 0.99 20 Cook #3 5.0 75 0.88 0 90 1.81 73 90 0.88 0 120 1.50 96 120 0.94 0 240 1.44 94 240 1.26 38 7.0 75 1.38 53 7.0 75 1.35 34 90 1.61 100 90 1.52 53 120 1.37 93 120 2.11 50 240 1.62 90 240 1.64 57 9.5 75 1.79 27 9.5 75 1.37 50 90 1.99 71 90 1.59 40 120 1.73 65 120 1.59 17 240 2.02 79 240 1.79 79 Table A-10. Shear strengths and percentage wood fa i lure values for aspen samples tested boiled Resin Moisture Press Shear Wood Resin Moisture Press Shear Wood Content Time Strength Failure Content Time Strength Failure (%) (sec) (MPa) (%) (%) (sec) (MPa) (%) Cook #4 4.6 75 0.00 0 Cook #5 4.6 75 1.26 12 90 1.08 16 90 1.92 30 120 1.99 48 120 1.68 63 240 2.49 96 240 2.20 100 6.7 . 75 0.12 0 6.7 75 1.19 2 90 0.94 35 90 1.82 84 120 1.73 93 120 1.99 81 240 1.46 100 240 2.11 98 9.2 75 0.00 0 9.2 75 1.19 25 90 1.43 10 90 1.82 60 120 1.87 99 120 1.96 83 240 1.46 100 240 1.99 97 Cook #6 4.6 75 0.37 0 Cook #3 4.6 75 0.26 0 90 1.61 100 90 0.32 0 120 2.23 78 120 0.67 0 240 1.43 90 240 0.85 0 6.7 75 0.91 18 6.7 75 1.73 43 90 1.76 94 90 1.39 47 120 2.17 93 120 1.57 40 240 1.59 95 240 1.24 100 9.2 75 1.11 20 9.2 75 1.39 32 90 1.61 48 90 1.76 2 120 1.59 62 120 1.37 29 240 1.35 61 240 1.91 77 Table A-11. Shear strengths and percentage wood fa i lure values for spruce samples tested boiled Resin Moisture Press Shear Uood Resin Moisture Press Shear Wood Content Time Strength Failure Content Time Strength Fai lure (%) (sec) (MPa) (%) (%) (sec) (MPa) (%) Resin B 5.0 75 0.00 0 Resin D2 5.0 75 0.00 0 90 1.29 71 90 0.62 17 120 1.92 94 120 2.03 83 240 1.18 100 240 1.67 100 7.0 75 0.00 0 7.0 75 0.00 0 90 0.56 4 90 0.90 33 120 1.50 93 120 1.82 94 240 1.62 100 240 1.38 100 9.5 75 0.00 0 9.5 75 0.00 0 90 1.23 46 90 1.14 36 120 1.62 100 120 1.35 100 240 1.85 100 240 1.21 100 Resin F2 5.0 75 0.00 0 Resin G 5.0 75 0.94 28 90 0.79 38 90 1.77 73 120 1.68 91 120 1.58 78 240 1.85 86 240 1.85 87 7.0 75 0.00 0 7.0 75 1.68 69 90 1.35 68 90 1.68 74 120 1.67 94 120 1.70 90 240 1.67 93 240 1.70 89 9.5 75 0.00 0 9.5 75 1.46 60 90 2.14 55 90 2.00 77 120 1.39 100 120 1.82 81 240 1.62 81 240 1.62 88 Table A-12. Shear strengths and percentage wood fa i lure values for aspen samples tested boiled Resin Moisture Press Shear Wood Resin Moisture Press Shear Wood Content Time Strength Failure Content Time Strength Failure (%) (sec) (MPa) (%) (%) (sec) (MPa) (%) Resin B 4.6 75 0.00 ' 0 Resin D2 4.6 75 0.00 0 90 0.97 19 90 0.81 6 120 1.73 63 120 2.12 93 240 2.05 100 240 2.38 70 6.7 75 0.00 0 6.7 75 0.00 0 90 0.47 11 90 0.35 0 120 1.23 37 120 0.38 13 240 2.34 99 240 1.99 100 9.2 75 0.00 0 9.2 75 0.00 0 90 0.85 7 90 0.66 27 120 1.08 22 120 1.37 59 240 1.67 100 240 1.81 98 Resin F2 4.6 75 0.00 0 Resin G 4.6 s 75 0.00 0 90 1.39 20 90 2.43 55 120 2.05 72 120 1.73 100 240 2.08 84 240 1.94 89 6.7 75 0.00 0 6.7 75 0.43 16 90 1.24 25 90 1.73 100 120 2.06 77 120 1.99 83 240 2.11 94 240 2.37 79 9.2 75 0.00 0 9.2 75 1.23 14 90 1.88 40 90 1.81 21 120 2.64 78 120 2.44 71 240 2.30 95 240 1.49 48 "@en . "Thesis/Dissertation"@en . "10.14288/1.0098287"@en . "eng"@en . "Forestry"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Some factors affecting the flow and penetration of powdered phenolic resins into wood"@en . "Text"@en . "http://hdl.handle.net/2429/29095"@en .