«CURE KINETICS OF WOOD PHENOL-FORMALDEHYDE SYSTEMS By JINWU WANG A dissertation submitted in partial fulfillment of the requirements for the degree of ...»
Sample preparation effects on Tan δ In practice, the time between sample preparation and DMA testing varied from sample to sample. As mentioned above, the curing event of vitrification is consistently evident and gelation point does not appear in the tan δ curve for the un-wrapped wood joints. The prominence and temperature of the peak 1 (Figure 7.1) varied with open and closed assembling time during the sample preparations. Figure 7.3 shows the effects of closed assembly time at different temperature on this peak for the PF-low bonded wood joints. With a minimal closed assembly time, the peak appears at 20 °C.
With increasing closed assembly time or pre-conditioning temperature, the temperature of peak 1 increases. DSC scanning has shown that the PF-low resin begins cure at around 70 °C (Wang et al. 2005). This behavior appears consistent with the findings using DMA here. When the sample was pre-conditioned at 70 °C for 8 hours, it appears to undergo substantial pre-cure. Accordingly, the first peak disappears and the intensity of the vitrification peak decreases (Figure 7.3). For PF-high bonded wood joints, the peak 1 appears at around 40-70 °C, which is consistent with its higher molecular weight.
When using a glass fiber braid for characterizing the polyurethane cure process, Toffey and Glasser (1997) defined this first peak as the glass transition temperature of uncured resins. Due to the fact that the peak temperature changes with water content in our research, it is reasonable to assume this peak may, in fact, be a glass transition temperature of the uncured PF resins. With DSC, Park and Wang (2005) reported that the glass transition temperature of an un-cured powder OSB resin was ca. 50 °C. This is consistent with Menard (1999) who defined the initial peak as the softening point of the un-cured resin, especially when the resin in its dehydrated state. However, we note that the melting point of phenol and hydroxymethelated phenol is in this same region (Merck 2001). Finally, it is also worthwhile to mention that the glass transition temperature of plasticized lignin is located in this temperature region (Kelly et al. 1987). In summary, Peak 1 is likely to be a glass transition of un-cured PF resin but other explanations might be possible. Regardless of the cause of this peak, the specimen preparation substantially influences its appearance and temperature while the effects on the further curing events appear negligible.
0.6 0.5 3
Figure 7.3 The effect of preparations on cure development of PF-low bonded wood joints scanned at 2 °C/min: 1.
DMA scan beginning at room temperature immediately followed sample preparation; 2. DMA scanning immediately followed sample preparation from low temperature; 3. closed assembly at 30 °C for16 hours; 4. closed assembly at 50 °C for 16 hours; 5. closed assembly at 60 °C for 16 hours; 6. closed assembly at 70 °C for 8 hour.
Mechanical cure development Using DMA, the degree of mechanical cure (β) can be defined (Vazquez et al.
where E’(t) is the storage modulus at time t. The effects of foil-wrapping on the mechanical cure development of PF-high bonded wood joints cured under linear heating and isothermal regimes are shown in Figure 7.4 and Figure 7.5; respectively.
Assuming that the primary influence of the foil-wrap is to limit moisture loss in the
specimen, the cure development appears to be determined by at least two processes:
moisture loss and resin crosslinking. Therefore, it is the superposition of these two processes that dictate mechanical cure development, especially during early stages for un-wrapped samples. Under linear heating regimes, the foil-wrapping reduces delays the beginning of cure as well as the peak temperature. The long inital tail in the storage modulus curves of un-wrapped PF-high bonded samples likely results from both moisture loss and resin curing (Figure 7.4). It is assumed that the abrupt inflection represents a shift to a process dominated by resin curing. Under isothermal temperature, moisture loss promoted a fast initial mechanical cure as compared with foil-wrapped PF-high bonded joints. However, these specimens experienced a plateau in the rising E’ which was likely associated with the shifting mechanisms.
In Figure 7.6, it is shown that under the linear heating regime the un-wrapped PF-high bonded wood joints begin curing at a lower temperature than PF-low bonded joints. These specimens were also observed to cure faster at the same isothermal temperature. It seems that the effect of moisture loss on cure development of PF-high bonded joints is larger than that on PF-low bonded joints. However, we note that the interaction between water and the PF-high resin is weaker than that between water and PF-low since PF-high is more advanced and has less hydroxyl groups available.
Thus the moisture in the PF-high bonded adhesive layer is less restricted and may evaporated at lower temperature and faster than in the PF-low bonded adhesive layer.
This potential moisture influence simply adds to the demonstrated increased cure rate of the PF-high resin when comparted to the PF-low (Wang et al. 2005). Despite the difference in initial cure rate between the PF-high and PF-low bonded joints, they appear to reach full mechanical cure at same temperature under linear heating regime.
This observation is also consistent with the results characterized by DSC (Wang et al.
2007). In summary, PF-low bonded wood joints cured more slowly than PF-high bonded joints, while foil-wrapped samples of PF-high bonded joints reduced the cure rate to the similar extent comparable to that of PF-low bonded joints.
Figure 7.4 The effects of foil-wrapping on the mechanical cure development at linear heating rate for the PF-high bonded wood joints.
Figure 7.6 Comparison of the mechanical cure development at linear heating rate between the PF-high and PF-low bonded wood joints.
Model-fitting kinetics It was observed that the E’ development under a isothermal cure regime followed a sigmoidal shape as illustrated in Figure 7.7 for a PF-low bonded wood joint. The autocatalytic (Eq. (55)), Prout-Tompkins (Eq. (56)), and Avrami-Erofeev (Eq. (57)) were used to fit the curve at each isothermal temperature (Galwey & Brown 1998).
where dβ/dt is rate of mechanical cure; k is rate constant; m and n are model constants, or reaction order. The data from the isothermal experiments were analyzed to yield values for the rate constant ki at each temperature Ti (subscript i refers to different isothermal temperature). The rate constants, ki, were found to increase with temperature. Using a linear regression on the relationship between ln ki versus 1/Ti (Arrhenius plot) yielded values for Ea/R and lnA (ASTM E2070).
Figure 7.7 An example of cure development under isothermal temperatures for PF-low bonded wood joints.
The advantage of the Prout-Tompkins and Avrami-Erofeev models over the general autocatalytic model is that it can obtain explicit equations for the degree of cure at specific cure times (Eqs (58) and (59)). Therefore, Eqs (58) and (59) can be used to directly fit the experimental data under isothermal cure regimes.
where T is temperature in Celsius degree; T0 and kT was two fitting constants. The variables required to conduct an isothermal kinetic analyses include β, t,and dβ/dt under isothermal conditions, and β, T, and dβ/dt under a linear heating regime. These data are then tested for fitting accuracy to three models. For the autocatalytic model, multiple linear regression analyses were used to extract necessary constants. For the Prout-Tomkins and Avrami-Erofeev models, a non-linear regression with Levenberg-Marquardt algorithm was used to extract constants using the explicit equation forms found in Eqs. (58) and (59). For each individual temperature program, all three models fit the data very well with R2 0.99.
For PF-low bonded wood joints, not all extracted constants for rate equations (Eqs. (55)-(57)) are independent of temperature. The autocatalytic model parameter, n, did not show any consistent pattern and was treated as a constant with an average value of n = 1.34 and a standard deviation of 0.18; m displays a bell-shape relationship with isothermal temperature (Figure 7.8) and was fitted with a three parameter Gaussian function (61) with R2 = 0.99.
where T is temperature in Celsius degree. As shown in Figure 7.8, m approaches zero at high temperature, where the rate of storage modulus development appears to follow an nth order model, indicating some mechanisms change. A high isothermal temperature may be beyond the glass transition temperature of fully cured resins.
The Avrami-Erofeev model constant n (Figure 7.9) provides information similar to that of the autocatalytic model parameter, m. With low value of n, the storage modulus development follows nth order. Like m, the Avrami-Erofeev constant n also
0.6 0.4 0.2 0.0 Figure 7.9 Kinetic parameter n of Avrami-Erofeev model changes with isothermal temperature for PF-low bonded wood joints.
The Prout-Tompkins constant t0 decreases exponentially with temperature, and can be fit linearly with the ln(t0) and temperature (63) with R2 = 0.99.
The three models and extracted constants under the isothermal regime for the PF-low bonded wood joints are summarized in Table 7.3. As an example, the constants of the Prout-Tomkins model for the PF-low, PF-high, and foil-wrapped PF-high bonded wood joints are summarized in Table 7.4 for both the isothermal and linear heating regimes. With these models and extracted parameters, the mechanical cure development can be readily described for both the isothermal and linear regimes.
Table 7.3 Summary of models and constants for PF-low bonded wood joints under isothermal temperature
All regression R20.96; T: temperature in Celsius degree; ϕ: heating rate in °C/min The activation energies obtained by these models were summarized on Table
7.5. The values listed as Peak time and Peak temperature were derived from vitrification peaks under the isothermal and linear heating regime (from Table 7.1 and Table 7.2), respectively, and listed here for comparison. The activation energies derived with the autocatalytic, Prout-Tompkins, Avrami-Erofeev, and Peak time methods utilized the isothermal data and were more similar than the Peak time method.
This approach utilizeds the Kissinger equation to analyze the linear heating data and is slightly larger than those obtained from isothermal data. An ANOVA analysis has indicated there is significant difference for activation energy between PF-low and PF-high, but no significant difference between PF-high and PF-high aluminum foil-wrapped. These activation energies from DMA mechanical cure development are smaller than those obtained from the DSC data for same neat resins (Wang et al 2006).
The activation energy obtained for PF-high at ramp mode was at same order with that for PF/wood mixture at a wood content 35 percent obtained by DSC (Wang et al.
2007a). The activation energy in presence of wood has decreased.
Table 7.5 Summary of activation energy (kJ/mol) with different methods
R2 0.99; Peak temp: calculated from vitrification temperature by Kissinger equation; all others are calculated from isothermal data.
CONCLUSION The transition temperatures of the curing process and cure development could be clearly assigned to storage modulus changes or to tan δ. The reproducibility among samples for recording curing process was good although the variations of sample preparations affected the glass transition temperature of un-cured wood-adhesive systems. Vitrification was probed in all samples, while gelation point was only detected for foil-wrapped wood joints under linear heating regime. It is assumed that moisture loss in un-wrapped joints muffled the gelation points. The activation energy of foil-wrapped PF-high joints for gelation and vitrification were 40 and 48 kJ/mol;
repectively. DMA mechanical cure development showed that the PF-low bonded wood joints cured slower than PF-high bonded wood joints. Foil-wrapping retarded moisture loss and delayed mechanical degree of cure for PF-high bonded wood joints.
The model-fitting kinetics, which generates single value characteristic parameters, especially activation energy, represents an important established method of reporting and comparing kinetic data. Model-fitting kinetics of autocatalytic, Prout-Tompkins, and Avrami-Erofeev models was selected to model mechanical cure development since the E’ development followed a sigmoid. The activation energies by three model-fitting models were closer to each other and in agreement with that from time events of vitrification with isothermal data. This is evidence that three models are comparable and capable to describing cure development. The activation energies obtained from ramp data by the Kissinger equation were a bit larger than those from isothermal data. Generally, the activation energy obtained from these methods under both linear heating and isothermal regime are around 50-70 kJ/mol. They were less than those obtained from neat resins by DSC (85-100 kJ/mol) (Wang et al. 2007) and in agreement with those of PF/wood mixtures obtained by DSC (Wang et al. 2007b).
The results imply that the activation energy of cure processes decreased in the presence of wood.
REFERENCES ASTM E2070-03 Standard test method for kinetic parameters by differential scanning calorimetry using isothermal methods. West Conshohocken, PA, USA.
Chi, F. T.; Hui, K. H. An alternative method to the curing study of polymeric die attach adhesives using dynamic mechanical analysis. Thermochimica Acta (2001), 367-368, 169-175.
Christiansen, A. W.; Follensbee, R. A.; Geimer, R. L.; Koutsky, J. A.; Myers, G. E.
Phenol-formaldehyde resin curing and bonding in steam-injection pressing. Part II.
Differences between rates of chemical and mechanical responses to resin cure.