# «CURE KINETICS OF WOOD PHENOL-FORMALDEHYDE SYSTEMS By JINWU WANG A dissertation submitted in partial fulfillment of the requirements for the degree of ...»

Specimen preparations Planed basswood strips (Midwest Products, Inc.) with nominal dimensions of 50x12x1 mm were oven-dried at 103 °C and stored in a desiccator over anhydrous calcium sulfate until use. Sandwich-type specimens (Figure 6.1) were produced using a layer of PF resin between two wood strips. Care was taken to match the grain, thickness, and weight of the adherend pairs within the specimen to maintain a balanced composite design. The bonding surfaces were lightly hand sanded along the grain with 220-grit sandpaper and cleaned with a paper towel immediately prior to resin application. The resin was uniformly applied to the prepared surface of both wood strips using a small airbrush (BADGER Model 350). The amount of resin solid applied to each surface was set at ca. 50 g/m2, which equates to ca. 12% of dried wood mass.

Maintaining a consistent resin content was deemed important to repeated cure analysis. He & Yan (2005b) demonstrated that the degree of resin loading can influence the cure development. They concluded that this influenced occurred primarily through water absorption and evaporation during the DMA test. Therefore, other measures to maintain moisture content during the tests were investigated. These include (1) short open and closed assembly times in producing the specimens and (2) foil wrapping of the specimens for the DMA analysis.

DMA and rheology DMA measurements were conducted on the sandwich specimens in three point bending mode using either a Tritec 2000 instrument (Triton Technology) (span 25 mm) or a Rhemetric RSA II DMA (span 48 mm). The frequency was fixed at 1 Hz. Strain sweep tests have been conducted to establish the linear viscoelastic ranges at working temperature. Oscillation displacement amplitude of 0.03 mm was thus chosen for Tritec DMA and a strain of 10-4 for RSA II DMA. DMA was performed isothermally at 90, 100, 110, 115, 120, and 130 °C. In each test, the DMA oven was preheated to predetermined isothermal temperature, and then the specimen was installed quickly and held at the cure temperature until both modulus and damping approached a constant value signifying the completion of detectable cure. The specimen then cooled down to room temperature, and re-scan at 2 °C/min. In addition, ramp experiments were performed at heating rates of 2, 3, 4, and 5 °C/min from room temperature to 250 °C. Low heating rates were selected to make sure that the effect of thermal lag was minimal.

Rheological experiments of the uncured samples were conducted on a Rhemetric RDA III rheometer using the 25 mm parallel plates. A strain of 1% and linear heating rate at 3 °C/min from 25 to 200 °C was used.

** Figure 6.1 The three point bending sandwich beam, the gray adhesive layer between two wood adherends.**

## RESULTS AND DISCUTION

In situ shear modulus development of adhesives An idealized sandwich specimen geometry for the three-point bending test is shown in Figure 6.1. Under the forced oscillation test used by most DMA instruments, the load (P) and mid-span deflection (∆) of the beam are out of phase by some angle δ.The storage component of the sandwich beam stiffness is given as C ' = P ′ / ∆ where P ′ = P cos δ. The beam stiffness can be related to the material properties and geometric variables of the adherends and adhesive using an analytical solution analog to the static mechanic solution as following (Adams and Weinstein 1975). In this solution, we consider the dynamic stiffness properties (K) for the components and

**total laminate:**

These beam stiffness equations consider the storage modulus (E’) and moment of inertia values (I) of both the adhesive and adherend as represented by subscripts a and f; respectively. Separate components, the I for the adhesive and adherend layers can be

Where: γ and M represents variables used to combined terms and simplify the expression without specific physical meaning. The units for γ are the reciprocal of length while M is dimensionless. Further, M will be bound by 0 ≤ M ≤ 3 depending on the shear modulus of the adhesive layer as will be discussed later. Note that these equations assume that the adhesive layer is isotropic when relating the adhesive E and

Equations (44) through (46) provide a means to compute the G’a from the experimentally determined C’. Modern DMA instruments such as the Tritec 2000 and Rheometric RSA II allow storage stiffness C’ to be directly output as an option. The change in C’ during curing process was shown in Figure 6.2 for a typical test at a linear heating rate. In addition, a number of material properties must be assumed. In our case, a constant adhesive Poisson’s ratio of νa= 0.35 and wood flexural storage modulus Ef = 9000 MPa at 12% MC (Wood handbook) were used. For any specific specimen, the geometric variables (L, b, h, and t) are measured. With these known

interpolation process implemented to avoid iterations. First, an assumed vector of Ga′ was created (e.g. Ga′ = [1, 100, 200,…, 108] Pa). Then, the correspond vectors γ, M, and C ′ were computed by sequential substitution into Eq. (45), (46), and (44),

' corresponding Ga value for each measured C’ from a scan was determined using the interpolation function within Matlab and the previously established Ga′ and C ′

MPa during curing process for a foil-wrapping wood joint bonded with PF-high resin.

This range is in general agreement with experimental data collected with a parallel-plate rheometer using the same linear heating rate, however differences exist (Figure 6.3). Before the onset of the curing process, both techniques determined that

was much more pronounced than that determined using the rheometer. Recall that the beam solutions assume that the adherend modulus is constant throughout the test. The softening that occurs in the wood substrate is, therefore, combined with the resin softening. In contrast, the rheometer showed difficulty in determining a consistent ' value for Ga for temperatures following vitrification. The latter difficulty is consistent with observations by others (Mekernd 1998) Both Laza et al. (2002), and

changes from 10-6 to 1 MPa for epoxy resins during curing at 150 °C (lager than its fully cured Tg). These reported values are of the same order with the calculated storage shear modulus of PF resin here and lend credibility to the results.

value than that of the PF-high bonded wood joint. The bulk shear modulus of the fully cured PF resin was reported to be 209.9 MPa at room temperature (Lee and Wu 2003), which is an order of magnitude higher than that calculated here by DMA. The discrepancy might be explained by the fact that the DMA measures the in situ shear modulus under the effects of the elevated temperature and somewhere near or in the rubbery state. Using torsion tests on fully cured expoxy resin, Dean et al. 2005 found that the shear modulus decreased from 1800 to 5 MPa while passing through the glass to rubbery transition. He et al. (2001) found that the shear modulus calculated with the Adams and Weinstein’s Eq. (44) was in agreement with the bulk shear modulus obtained using torsion when the resin is in rubber state. However, the negligible contribution of the adhesive layer to the beam stiffness in the glass state (i.e.

' dC '/ dGa is very small) places doubt on the calculation (He et al. 2001).

** Figure 6.2 A typical of DMA output.**

Effective storage modulus (E’) and storage stiffness (C’) changes with temperature during curing at 3 °C/min for a foil wrapped PF-high bonded wood

The accuracy of shear modulus calculation depends on the accuracy of estimating material properties and the sensitivity of the analysis to geometric variables.

For instance, the wood modulus is affected by changes in moisture content and temperature during thermal scanning. In addition, the Poisson ratio of the PF resin is expected to change during curing process where it experiences the sol, gelation, and vitrification stages. Finally, there are uncertainties in measuring the adhesive because the layer undergoes severe physical and morphological changes. Hence, the calculated shear modulus as mentioned above is merely an estimate of the interfacial shear modulus in the specimen.

During the curing process, the shear modulus changes from a minimum to a maximum as it passes from a sol to vitrified state. Let us investigate the two extreme cases of the resin in the lowest and highest shear modulus states. When the resin softens to a minimum viscosity before curing, the sandwich beam behaves as three

this case, the variable M in Eq. (44) approaches 3 if thickness of the adhesive layer is small (Figure 6.5). This point with minimum shear modulus is referred with subscript

0. Hence, Eq. (44) can be simplified as following:

In contrast, when the shear modulus of the fully cured approaches a large value (i.e. comparable to that of the adherend) then M approaches zero (Figure 6.5) and the shear deformation of the bonded layer becomes negligible. The point of minimum shear deformaton is referenced with a subscript ∞. In this case, Eq. (44) can

**be simplified as following:**

adhesive layer is in a state of pure shear and the contribution of adhesive bending to

**the total beam stiffness is negligible. Hence, Eqs. (47) and (48) become:**

Note that Eq. (49) indicates that when the adhesive shear modulus is very low and the h/t is large, the sandwich beam can be treated as two separate homogenous beams bending about their own axes. Consequently, the total flexural rigidity of sandwich beam can be reasonably approximated by the sum of flexural rigidities for the two adherends. Note also that Eq. (50) indicates that when the adhesive modulus is comparable to that of the adherend, the total flexural rigidity of the beam converges to the pure bending rigidity of the beam treating two adherends as a homogeneous beam, since the shear deformation of the bonded layer becomes negligible.

For metals and other composites whose modulus is not influenced by moisture and temperature, E 'f 0 = E 'f ∞. That is saying that the flexural storage modulus should not change during the curing process. However, DMA uses the simple beam theory treating the three-layer sandwich structure as a solid homogeneous beam without including the contribution of the shear deformation in the adhesive. The result is an

Comparing Eqs (49) and (51) we can deduce that the calculated effective shear modulus of the sandwich beam when the resin is at its lowest shear modulus is one

conclusion that the ratio of the un-cured to cured modulus ( R = Et′=0 / E t′=∞ ) would be the most effective parameter to monitor for cure because this variable would eliminate variability in the adherend properties. However, this assumption is only valid if the storage modulus of the adherends does not change during curing process.

** Figure 6.5 The effects of thickness of the adhesive layer on the item M.**

Optimizating DMA derived parameters for directly evaluating a wood-adhesive systems In Figure 6.6a, two sandwich cure scan are depicted to show the ideal case where R ≈ 4 for both a PF-high and –low resin system. For cases where the bond formation is deficient, 1 R 4. Experimentally determined values for R of 129 specimens in three categories (PF-low, PF-high and foil wrapped PF-high bonded wood joints) are shown in Figure 6.6c. For these specimens 72 percent fell into the range of 2.5 R 4, for 11 percent samples, R 2.5, and for the remaining 17 percent R 4.

We speculate that some specimens may produce an R 4 because the assumption that adherend properties remain constant throughout the test is violated.

To investigate this hypothesis, a sample of overlapping wood adherends conditioned to 12% MC and lacking in PF-resin was scanned with the results shown in Figure 6.6a.

Note that the modulus increased for temperatures less than 120°C and then decreased somewhat for higher temperatures. It is likely that moisture evaporation resulted in the initial stiffening while thermal softening prevailed after the wood had dried. Whatever the exact mechanism, it is clear that the adherend stiffness changes during the test and that it results in a higher value at completion. Assuming that the wood completely dries during the test, the ration of the modulus in the dry and wet states could reach

1.5. Following this reasoning, the flexural rigidity in the cured state could be 8EfdIf and compared to 2EfwIf before curing; resulting in an R ≤ 6 (8Efd/2Efw = 6). It is likely that even with specimens where R 4 the results may be biased by a portion of modulus increment resulting from moisture loss and not simply adhesion effects.

** Figure 6.6 Summary of effective storage modulus (E’) development with all three kinds of samples together.**

(a) Typical E’ development for PF-low and PF-high bonded