3.1. ABC Methodology
The titration curves for different mixing times have very similar shapes, excluding the first few measurement points corresponding to 1–10 milliliters of titrant added. After reaching the boundary of pH 4, the curves are almost linear (
Figure 2). A small displacement was caused by the different pH at the titration starting point. The data in
Figure 2 also show the effect of the different latency time (suspension mixing prior to titration) that renders slight shifts in the curve position and, subsequently, may provide slightly different results—the final volume of the titrant added ranged between 70 mL and 82 mL.
The ABC values can be predicted by two models, β~(β
>4) and β~(β
>4, β
4–3), with a coefficient of determination as high as R
2 = 0.99. The RMSE values of 1.27 and 1.32 show differences between the predicted and observed values. The higher the RMSE value, the less accurate the model is. Thus, the β~(β
>4, β
4–3) model seems to be slightly inferior, though both can be considered correct [
18]. These models confirm that the buffering capacity of chip suspension under the boundary of pH 4 becomes irrelevant for the study. Moreover, the “simplified” β~(β
>4) model exhibited a slightly smaller coefficient of variation equal to 1.60 (
Table 1). Hence, titration to pH 4 instead of pH 3 seems sufficient for a proper determination of the ABC of a woody material.
As shown in
Figure 3 and
Table 2, particle size had a strong effect on the titration curves. Fractions of 0.15–1.0 mm provided an almost linear response, while the curves for 1.0–3.0 mm and 3.0–6.0 mm fractions became linear after some amount of time as the system required more time for equilibration. This indicates that the effective surface area of particles has a strong influence on the ABC measurement, as the equilibration rate is the highest for the smallest chips, as well as on the resultant ABC value. A high surface area in fine chips can be associated with an increased ABC. Thus, the data in
Figure 3 indicate that it is not plausible to compare the ABCs found for different particle sizes. Surprisingly, the observation is contrary to reports by Pedieu et al. who found that particle size did not impact the buffering capacity [
11]; however, it is coherent with the findings of Elias and Irle who recognized that particle size had an effect on wood acidity [
1]. However, it is noteworthy that the authors used a classic method based on the analysis of aqueous extracts as described by Johns and Niazi [
9].
The data in
Table 2 clearly indicate that the observed acid buffering capacity decreases as the particle size increases. The initial pH of the suspension—after 60 min—reveals that a slower penetration of water into the particles’ structure causes a slower diffusion of the extractives into water, hindering acidification. The pH curves for the chip suspensions shown in
Figure 4 are quite similar in shape. After some amount of time all three become almost stable and linear. The changes in pH are dynamic for the first 10 min and come from the equilibration in the system. This suggests that starting the titration of the suspension—when mixing for at least 30 min is avoided—leads to serious errors caused by a significant change in the initial pH which, subsequently, provides an incorrect ΔpH.
According to previous investigations [
14,
16] and results shown in this work, it can be concluded that measuring the ABC for woody materials is a process susceptible to a number of factors (e.g., wood presence, chip size, and latency time) that affect the response of the system. None of the factors can be neglected as they have a significant influence on the measured value. Therefore, it is impossible to compare the ABC results from different methods as, for instance, different particle sizes give different ABC values even if the same determination method was used. Moreover, the presented models indicate that there is no need to titrate the chip suspension to below pH 4 as the difference in the final result is statistically insignificant. What must be considered, too, is the suspension mixing time as it is critical for the initial pH; therefore, the latency time to titration has been defined as at least 30 min.
3.2. Developed Protocol for ABC Determination
Regarding all the above, a protocol for ABC determination based on an original methodology including the presence of woody material (chips) in the titrated system has been defined. The developed protocol comprises four steps: (1) preparation of material, (2) chip suspension mixing until stabilized pH, (3) pH-metric titration of chip suspension, and (4) calculations. The presence of chips in the system and the avoidance of boiling are what makes the difference between the new method and the classic one involving aqueous extracts [
9].
Thus, the procedure can be delineated as follows: 20 g chips of known moisture content and defined size range as used in a given particleboard production process are added to 200 g distilled water. (The amount of these can be changed as long as the weight ratio of 1:10 is kept.) The suspension needs to be mixed (160 rpm) at room temperature for 30 min or until a stabilized pH is reached. Do not filter the chips prior to titration. Then titrate the suspension with 0.025 N HCl or H
2SO
4 (1.0 mL titrant per minute) to pH 4. Use Equation (3) for the calculation of the acid buffering capacity per gram:
where
βpg is the acid buffering capacity per gram,
β is the acid buffering capacity,
m is the dry chip mass (g), Δ
n is the amount of added acid, Δ
pH is the resulting change in pH, Δ
V is the titrant amount in 1.0 liter, and
n is the mmol titrant per liter.
After the substitution of the fixed values,
m = 20 g,
n = 50 mmol/L, Equation (4) can be calculated:
3.3. ABC Effect
The tensile strength perpendicular to the surface of a board is considered a measure of the internal bond (IB) of the particleboard that is strongly correlated with the strength of the adhesive bond between the chips in a panel [
18,
19,
20,
21,
22]. Therefore, the IB was involved in building a correlation with the ABC of the chips used as the raw material in the particleboard preparation. Boards made of pine and oak chips, with different amounts of hardener, were prepared. In the case of the pine panels (
Table 3), the thickness of the P0 sample was apparently higher, while the structure on the cross-section was rough and loose. In contrast to P0, the structure in the cross-sections of P1 and P2 was more homogeneous and compact. In P3, where the hardener amount was beyond the optimum, a loose structure in the cross-section also occurred. This phenomenon results from the rapid curing of the adhesive with an excess of acidic hardener and subsequent deterioration of the bond line by acid upon storage [
1,
5]. In the case of the oak panels (
Table 4), it was difficult to observe any significant differences in the structure or thickness of the three boards, regardless of the hardener amount. However, the IB results shown in
Figure 5 clearly indicate that the maximum internal bonding is achieved with a hardener addition as small as 0.75 pbw. This is in accordance with the observations reported by Roffael and co-workers who found that highly acidic oak chips can be bonded with UF resins without any hardener at all [
23].
As indicated in
Figure 5, different amounts of hardener result in different IBs for particleboards. For both pine and oak particleboards, a second-degree linear correlation (R
2 > 0.99) between the IB and the hardener amount were found. Thus, the hardener amount for the highest IB could be predicted. The correlation of
IB =
f(hardener amount) calculated for the pine and oak particleboards can be used to predict the optimal amount of hardener for a material with a totally different chemical nature. Hence, the optimal hardener amount was calculated for the ammonia-treated oak (
Table 5), taking into account the three different ABC measuring methods (M1–M3).
Predicted amounts of hardener were used to produce panels from ammonia-treated oak chips (
Table 5). In the case of ammonia-treated oak, the A4 variant, where the amount of hardener was far beyond the typical values, disintegrated during removal from the press (
Table 6). It is worth noting that the thickness of specimens A1 and A2, as well as the quality of the chip bonding in the core, were clearly better than those in A3. The obtained results were used to calculate the optimal amount of hardener using a linear correlation of the second degree (
Figure 6). It can be seen in
Table 5 that the most accurate prediction was obtained from Method M3, i.e., 3.0 g hardener per 50 g resin against 3.2 g hardener per 50 g resin calculated from the model based on the empirical IBs. The results show that Methods M1 and M2 provided erroneous data, which proved them to be inapplicable in the described approach.
Regardless of the extremely low IBs observed for the ammonia-treated oak boards (0.012–0.028 N/mm2) that make such material inapplicable for particleboards at all, the results obtained in this study proved that the developed procedure for ABC determination based on pH drift in chip suspension (M3) is reliable and applicable for woody material. It was shown that the methods based on either titration of the chip suspension (M1) or pre-acidification of the chip suspension (M2) provided less useful outcomes and, thus, were considered ineffective. It was possible to develop an efficient model based on the ABC of chips that allows the prediction of the amount of hardener for the maximized IB of a particleboard.
Keeping in mind that the conditions of raw material storage render changes in the raw material’s chemical properties (e.g., acidity and extractive content) [
1,
24,
25,
26], a proper ABC determination may help in order to adjust the amount of hardener in the adhesive system to the acidity of the raw material (chips, fibers), so that an excessive acidification of the bond line is avoided and the mechanical properties of the boards are not lowered upon service.