«CURE KINETICS OF WOOD PHENOL-FORMALDEHYDE SYSTEMS By JINWU WANG A dissertation submitted in partial fulfillment of the requirements for the degree of ...»
GPC has shown that the core PF has a higher molecular weight than that of the face resin.
Table 2.10 13C-NMR chemical shifts of PF resins in D2O solvent.
*the numbers are corresponding to that in Figure 2.
Molecular weights of the neat PF resins by 13C-NMR The 13C-NMR spectra of the neat resins were recorded under the parameters, which meet the requirement for quantitative analysis as discussed for the acetylated resins. Rmb and Rme were directly obtained from respective integrals of signals by setting integral value of phenolic carbon region as unity (Rmb = 1+2+3, Rme = 6+7 in Table 2.10). Hence, Eq. (4) was used to calculate the molecular weights as presented in Table 2.11.
Table 2.11 Summary of characteristic structures and molecular weights of two neat PF resins based on 13C-NMR spectra in D2O.
Mn: number average molecular weight; p.p.u.: per phenolic unit.
Comparison of different methods Each method of obtaining molecular weights with NMR techniques yielded consistent results: the core PF had a higher content of methylene bridge per phenolic unit, a higher degree of polymerization and molecular weights, but a lower content of methylol per phenolic unit. Table 2.12 summarizes the molecular weights of the neat resins obtained from the 1H- and 13C-NMR spectra, as well as the molecular weights of the neat resins, which were calculated from the spectra of the acetylated resins by removing the acetoxy groups.
For the neat resin, there was no significant difference between the molecular weights obtained from the 1H- and 13C-NMR spectra. However, the molecular weight of the neat resin derived from 1H-NMR spectra of the acetylated resins was larger than that derived from 13C-NMR spectra of the acetylated resin. There are better agreements of the molecular weights from 1H-NMR spectra than those from 13C-NMR.
It seems that the molecular weights derived from 13C-NMR spectra of the acetylated resins were incorrect, as shown in Table 2.12. The molecular weights obtained from H-NMR spectra of the acetylated resins were larger than those from the neat resins.
Acetylation increased the molecular weight slightly (Wellons and Gollob 1980;
Yazaki et al. 1994). In this sense, 1H-NMR had a better quantitative analysis for the resin structure than 13C-NMR despite every effort to obtain quantitative 13C-NMR spectra.
By comparing the molecular weight from 1H-NMR and 13C-NMR (Table 2.13), it can be seen that for the lower molecular weight resins, the GPC and NMR techniques yield similar results. Larger discrepancies are found for the higher molecular weight resin. There were reports that GPC analysis of the acetylated higher molecular weight resin tended to over-estimate the molecular weight ((Wellons and Gollob 1980; Yazaki et al. 1994). However, Steiner (1975) found that 1H- NMR was relatively insensitive when the molecular weight of the resins was large. For example, the assumption of linear structure for Woodbrey’s formula may not hold true at high molecular weight. Hence, the molecular weight calculated by NMR technique may also have a loophole. Therefore, for the core PF resin, it is unclear which technique is more credible. Perhaps the real value of molecular weight of the core PF lies between those obtained by GPC and NMR.
Table 2.12 Comparison of number average molecular weight (Mn) and degree of polymerization (n) of the neat resins by 1H- and 13C-NMR.
*Mn and n of the neat resins were calculated from spectra of the acetylated resins by taking out of acetoxy groups.
Table 2.13 Comparison of number average molecular weight (Mn) and degree of polymerization (n) of the acetylated PF resins by 1H- and 13C-NMR and GPC.
CONCLUSION Both 1H-NMR and 13C-NMR spectroscopy proved to be valuable techniques in a detailed analysis of acetylated and neat resins. Chemical shifts of PF resins could be assigned according to the literature. With 13C-NMR for the neat resins in D2O, the phenolic carbon region from 150-160 ppm was an informative spectrum for analyzing the varieties of its environments. The broader distribution of the core PF resin in this region implied its advanced structures as compared with the face PF resin. 13C-NMR probed the spectrum of o-o, o-p, and p-p methylene bridges in the region of 20-40 ppm separately for the acetylated resin. The core PF resin clearly presented the signals of o-o methylene bridges at 30-32 ppm while the signal for the face PF resin was weak at this region. Inverse-gated decoupling and addition of relaxation reagents of Chromium (III) acetylacetonate and Gadolinium (III) chloride hexahydrate into NMR solvents were effective for use in quantitative analysis with 13C-NMR for the resin chemical structures. The quantitative analyses with 1H- and 13C-NMR have demonstrated that the core PF resin has a higher methylene bridge content per phenolic unit, degree of polymerization, and molecular weights but lower methylol content per phenolic unit than the face PF resins.
Although 13C-NMR had a higher spectrum resolution and gave more detailed information for the PF chemical function groups, quantitative analysis with 13C-NMR was not more effective than that with 1H-NMR. GPC chromatograms clearly showed that both resins have three fractions of molecular weights and the core PF is more advanced than the face PF resin. For the face PF resin with a low molecular weight, GPC and NMR gave values of molecular weight at the same order, while GPC gave a much higher molecular weight for the core PF resin than NMR. This work suggests that both GPC and 1H- and 13C-NMR are useful qualitative tools for differentiating the resins. However, the accuracy of these methods for determining the molecular weight distribution of PF resins should be investigated further.
Achnowledgement: GPC test was conducted at the Forest Products Department at University of Idaho under assistance of Dr. J. Andy Soria.
REFERENCES Bouajila, J., Raffin, G., Waton, H., Sanglar, C., Paisse, J. O., Grenier-Loustalot, M. F.
Phenolic resins - characterizations and kinetic studies of different resols prepared with different catalysts and formaldehyde/phenol ratios (I). Polymers & Polymer Composites,(2002) 10 (5), 341-359.
Dargaville, T. R.; Guerzoni, F. N.; Looney, M. G.; Shipp, D. A.; Solomon, D. H.;
Zhang, X. Determination of molecular weight distributions of novolac resins by gel permeation chromatography. Journal of Polymer Science A: Polymer Chemistry (1997), 35, 1399–1407.
Holopainen, T.; Alvila, L.; Rainio, J.; Pakkanen, T. T.; Tuula, T. Phenol-formaldehyde resol resins studied by 13C-NMR spectroscopy, gel permeation chromatography, and differential scanning calorimetry. Journal of Applied Polymer Science (1997), 66 (6), 1183-1193.
Luukko, P. A.; Holopainen, T.; Rainio, J.; Pakkanen, T.T. Optimizing the conditions of quantitative 13C-NMR spectroscopy analysis for phenol-formaldehyde resol resins.
Journal of Applied Polymer Science (1998), 69 (9), 1805-1812.
McGraw, G. W.; Landucci, L. L.; Ohara, S.; Hemingway, R. W. Proton and carbon-13 NMR studies on phenol-formaldehyde prepolymers for tannin-based adhesives.
Journal of Wood Chemistry and Technology (1989), 9 (2), 201-217.
Nieh, W. L. S.; Sellers, T., Jr.. Performance of flakeboard bonded with three phenol-formaldehyde resins of different mole ratios and molecular weights. Forest Products Journal (1991), 41(6), 49-53.
Riedl, B.; Calve, L.; Blanchette, L. Size - exclusion chromatography of spray-dried phenol -formaldehyde resins on different columns and solvent systems.
Holzforschung (1988), 42 (5), 315-318.
Rego, R., Adriaensens, P. J., Carleer, R. A., & Gelan, Jan M. (2004). Fully quantitative carbon-13NMR characterization of resol phenol-formaldehyde pre-polymer resins.
Polymer (2004), 45 (1), 33-38.
Steiner, P. R. Phenol – formaldehyde wood adhesive characterization by proton magnetic resonance spectroscopy. Journal of Applied Polymer Science (1975), 19 (1), 215-225.
Wellons, J. D.; Gollob, L. GPC and light scattering of phenolic resins - problems in determining molecular weights. Wood Science (1980), 13 (2), 68-74.
Wilson, J. B.; Jay, G. L.; Krahmer, R. L. Using resin properties to predict bond strength of oak particleboard. Adhesives Age (1979), 22 (6), 26-30.
Woodbrey, J. C.; Higginbottom, H. P.; Culbertson, H. M. Proton magnetic resonance study on the structure of phenol-formaldehyde resins. Journal of Polymer Science, Part A: General Papers (1965), 3(3), 1079-1106.
Yazaki, Y.; Collins, P. J.; Reilly, M. J.; Terrill, S. D.; Nikpour, T. Fast - curing phenol formaldehyde (PF) resins. Part 1 Molecular weight distribution of PF resins.
Holzforschung (1994), 48 (1), 41-48.
Chapter 3 Comparison of Model-fitting Kinetics for Predicting the Cure Behavior of Commercial Phenol-formaldehyde Resins ABSTRACT Phenol-formaldehyde (PF) resins have been the subject of many model-fitting cure kinetic studies, yet the best model for predicting PF dynamic and isothermal cure has not been established. The objective of this research is to compare and contrast several commonly used kinetic models for predicting degree of cure and cure rate of PF resins. Toward this objective, the nth-order Borchardt-Daniels (nth-BD), ASTM E698 (E698), autocatalytic Borchardt-Daniels (Auto-BD), and modified autocatalytic methods (M-Auto) are evaluated on two commercial PF resins containing different molecular weight distributions and thus cure behaviors. The nth-BD, E698 and M-Auto methods all produce comparable values of activation energies while Auto-BD method yields aberrant values. For dynamic cure prediction, all models fail to predict reaction rate, while degree of cure is reasonably well predicted with all three methods.
As a whole, the nth-BD method best predicts degree of cure for both resins as assessed by mean squared error of prediction.
Key words: resins, activation energy, modeling, kinetics (polym.), differential scanning calorimetry (DSC)
INTRODUCTIONSeveral hot-pressing models of engineered wood-based composites have been developed to predict properties such as moisture content and internal pressure during mat-solidification in the past decades ((Bolton and Humphrey 1988; Thoemen and Humphrey 2003; Dai and Yu 2004; Zombori et al. 2004). Such models are important to design and optimize hot-pressing parameters during the manufacture of engineered wood-based composites. During panel consolidation, the heat of resin polymerization plays an important role. Yet hot-pressing models have either used an arbitrary kinetic model or have not incorporated the resin cure kinetics (Bolton and Humphrey 1988;
Thoemen and Humphrey 2003; Dai and Yu 2004; Zombori et al. 2004), hence limiting their application. In order to improve the accuracy of hot-pressing models, cure kinetics needs to be incorporated. During a DSC temperature scan phenol-formaldehyde (PF) resoles typically exhibit two exotherms (Holopainen et al.
1997). Although a subject of controversy the first exotherm is often ascribed to hydroxymethylphenols formation and condensation while the second exotherm is attributed to dimethylene ether linkages decomposition into methylene linkages between phenolic moieties (Holopainen et al. 1997). To model resin cure kinetics, model-fitting (MF) (Harper et al. 2001) and model-free kinetics (Wang et al. 2005) can be used in combination with differential scanning calorimetry (DSC) (Prime 1997). For commercial PF resins, model-free kinetics has recently demonstrated excellent modeling and prediction abilities for both degree of cure and reaction rate during dynamic and isothermal cure (Wang et al. 2005). However model-free kinetics involves complex computations that may not be easily implemented in a comprehensive hot-pressing model. Indeed, hot-pressing models require solving simultaneously two governing partial differential equations, one on heat transfer and one on mass transfer (Zomborie et al. 2004). As a result, an explicit cure kinetic model can be more easily incorporated into the solving process. In contrast, MF methods assume a definite reaction model facilitating simple computations with kinetic parameters such as activation energy, reaction order and pre-exponential factor.
As such, they remain of interest when an approximate prediction of cure development is needed, and will be easily incorporated into a hot-pressing model. In fact, MF kinetics has long been used to characterize and compare the cure kinetics of different PF resins (Kay and Westwood 1975; Rials 1992; Vazquez et al. 2002; Park et al.
2002). In particular, the nth order model with the Borchardt-Daniels (ASTM E 2041) and the ASTM E 698 methods have been widely utilized. Yet different kinetic methods often generate different kinetic parameters (Alonso et al. 2004). For instance, the nth order with the Borchardt-Daniels method was reported to yield activation energy values that are 30% higher than those obtained with the Ozawa or Kissinger equations used in ASTM E698 (Alonso et al. 2004). These observations raise a concern about which MF method is best suited to model the cure kinetics of different commercial resins including the PF varieties studied in this research. More importantly, the prediction ability of MF methods for phenolic resin cure has not been established. The choice of an MF method to predict PF cure kinetics for incorporating into hot-pressing models is, therefore, not evident. In this perspective, the objective of this study is to determine and compare the suitability of four MF kinetic methods to model and predict the cure kinetics of PF resins. The specific models studied include the nth order with Borchardt-Daniels, autocatalytic model with Borchardt-Daniels, ASTM E698 and modified autocatalytic methods (Harper et al. 2001).