3.1. Surface properties
3.1.1. Wettability – Wetting Angles and Surface Energy
The type of machining process influenced the wetting of the oak wood surface, which was determined by wetting angle measurements listed in Table 3. The wetting angles of formamide generally show the lowest wetting angles for all machining treatments compared to other used liquids.
Table 3
Average values of wetting angles measurements of differently machined oak-wood surfaces; Surface energy of tested liquids according to Ström et al. (1987) (Ström et al. 1987)
Machining | Wetting Angle |
Water | Diiodomethane | Formamide |
Face Milling | 48.67 (5.90) | 25.00 (2.83) | 15.43 (3.93) |
Planing | 57.53 (2.80) | 26.00 (1.07) | 15.60 (1.60) |
Sanding | 30.63 (4.30) | 6.00 (1.77) | 0.00 (0.00) |
Surface Energy | 72.8 | 50.8 | 58.0 |
Polar Component | 51.0 | 0 | 19.0 |
Dispersive Component | 21.8 | 50.8 | 39.0 |
The average value is based on 45 measurements (15 measurements per specimen). The standard deviation is in parentheses.
Furthermore, a one-way ANOVA was conducted for each liquid to compare the effect of three different types of wood surface machining on the wetting angle (Table 4). Various types of machining resulted in statistically significant different wetting angles for water (F (2,132) = 415; p = 1.18*10− 57), diiodomethane (F (2,132) = 1395; p = 1.64*10− 89), and formamide (F (2,132) = 602; p = 4.66*10− 67). Furthermore, the effects sizes of water (ω2 = .86), diiodomethane (ω2 = 0.95) and formamide (ω2 = 0.89) indicate that the machining process had a large effect on the wetting angles of all three liquids, as per Fields's (2013) conventions (Field 2013). With additional statistical analysis, the pairwise comparison test showed that there is a statistically significant difference between all three treatments in terms of water wetting angles (p = 0.000). As for wetting angles with the other two liquids, only sanding was statistically different from other machining's (p = 0.000). Therefore, there was no statistically significant difference between face milling and planing in the wetting angle of formamide (p = 0.942) and diiodomethane (p = 0.053).
In the case of all liquids, the sanded surface results in the best wettability, especially when it comes to formamide, where it thoroughly wets the surface after a few seconds. Therefore, planed and faced milled surfaces show approximately equal wettability, except in the water wetting angle, where planed surface results in the lowest wettability, i.e. the highest wetting angles.
Table 4. Statistical summary of the effect of three different types of wood surface treatment on the wetting angle of liquids.
Liquid
|
ANOVA Summary
|
Pairwise Comparison**
|
Source
of Variation
|
Deg. of Freedom
|
Sum
of Sq
|
Mean Sq
|
F
|
p
|
Effect Size*
|
Planing
vs.
F. Milling
|
Sanding
vs.
F. Milling
|
Sanding
vs.
Planing
|
ω2
|
CI 95 %
|
Water
|
Between
|
2
|
16913
|
8457
|
415
|
1.18*10-57
|
0.86
|
[0.81-0.90]
|
0.000
|
0.000
|
0.000
|
Within
|
132
|
2690
|
20
|
|
|
Diiodomethane
|
Between
|
2
|
11430
|
5715
|
1395
|
1.64*10-89
|
0.95
|
[0.93-0.97]
|
0.053
|
0.000
|
0.000
|
Within
|
132
|
542
|
4
|
|
|
Formamide
|
Between
|
2
|
7222
|
3611.1
|
602
|
4.66*10-67
|
0.89
|
[0.86-0.93]
|
0.942
|
0.000
|
0.000
|
Within
|
132
|
792
|
6
|
|
|
*Effect Size: < 0.01 – very small, 0.01 ≤ ω2 < 0.06 – small, 0.06 ≤ ω2 < 0.14 – medium, ≥ 0.14 – large.
**Pairwise Comparison: p-values.
Calculating surface energy was another way of characterising the surface's attraction to other substances (Mihulja G and Bogner A 2005). The highest total surface energy was observed on sanded surfaces. This can be attributed to the changes in surface morphology and structure of chemical components on the wood surface (Qin et al. 2015). Thus, sanding leads to fibrillation and dust accumulation (Gurau et al. 2005) and exposure of active functional groups (e.g. hydroxyl groups), which makes the surface more hydrophilic (Qin et al. 2015). Similar results were obtained by Liptáková et al. (1995) after testing oak wood (Liptáková et al. 1995). Comparably, Leggate et al. (2020) found that planing resulted in poor wettability for southern pine and spotted gum, whereas face milling and sanding treatments improved wettability after planing (Leggate et al. 2020a).
The amount of surface energy computed with both methods for different types of machining shows an expected correspondence with the results of wetting angles. Using different types of surface machining, the surface energy of oak wood changes its values of polar and dispersive components. It is evident that the polar component is affected most; the dispersive component is affected very little or not at all. Therefore, the sanded surface had the highest surface energy, while planed and face-milled surfaces had approximately equal surface energies.
For all machining methods, the dispersive component of surface free energy is dominant and approximately constant (Table 5). According to Qin et al. (2014), the dispersive component's high value results from the dispersive component's high interaction ability of available carbon-oxygen and carbon-carbon bonds within the wood. On the other hand, the polar component refers to the interaction between hydroxyl groups of wood and functional groups of adhesive by forming the hydrogen bond (Qin et al. 2014). This is a typical feature characteristic for polymers of which the wood is composed (Mohan et al. 2011). It can be concluded that the increase in surface energy of each machining treatment is associated with a substantial increase in the polar component (Table 5). For example, according to the OWRK method, a face-milled surface has slightly higher surface energy than a planed surface (60.53 versus 57.53 m/Nm). However, the polar components of the two characters differ more significantly (16.48 versus 11.30 m/Nm). So, the manufacturing process obviously affects the polar component of surface energy, as Dillingham (2020) noted (Dillingham 2020). This suggests that the surfaces contain an increased number of hydrophilic polar groups after machining. Wood surfaces' high polarity and good wettability is given mainly by the presence of hydroxyl, carbonyl, and carboxyl groups in the lignin-polysaccharide matrix of the cell walls. This results in the formation of strong physical bonds with various polar adhesives. Thus, wood surfaces with higher polarities are more wettable with water-based adhesives (Iždinský et al. 2020). Consequently, wood species with a high concentration of non-polar extracts are more difficult to bond (Kamke and Lee 2007).
Table 5
The surface energy with components of differently machined oak wood surfaces calculated according to OWRK and Wu's principles.
Machining | WU | OWRK |
Surface Energy Total (m/Nm) | Dispersive Component (m/Nm) | Polar Component (m/Nm) | Surface Energy Total (m/Nm) | Dispersive Component (m/Nm) | Polar Component (m/Nm) |
Face Milling | 63.86 (4.44) | 42.54 (1.33) | 21.32 (5.68) | 60.53 (4.41) | 44.05 (2.14) | 16.48 (6.48) |
Planing | 60.86 (1.59) | 43.58 (1.17) | 17.28 (2.59) | 57.53 (1.87) | 46.24 (1.66) | 11.30 (3.53) |
Sanding | 71.83 (0.69) | 43.59 (0.18) | 28.24 (0.55) | 68.32 (0.67) | 44.42 (0.16) | 23.90 (0.57) |
The standard deviation is in parentheses.
Finally, it is interesting to compare the wetting angles between liquids of different polarities and the surface energy components of differently machined surfaces. The surface was more wettable with liquids with a dispersive and polar component ratio similar to ratio of components in wood surface energy of the machined surface. Of course, this can only happen when the surface tension of the liquid is equal to or less than the surface energy of wood, e.g. surface energy components. This ratio allows for a prediction of the adhesion between wood and adhesive phases. The closer the ratios match the more interactions are possible between the phases and the higher the adhesion is to be expected (see Fig. 3).
3.1.2. Roughness
Roughness was evaluated through 2 roughness parameters Ra and Rz. Welch's Anova F-test was conducted to examine the differences in surface roughness of oak wood according to the machining treatments. Significant differences were found among the different machining treatments for both observed roughness parameters: Ra (Fw (2, 11.13) = 26.70, p = 5.67*10− 5 < 0.05) and Rz (Fw (2, 11.27) = 9.28, p = 4.15*10− 3 < 0.05. The effect size of Ra (ω2 = 0.78) and Rz (ω2 = 0.54) indicates that the effect of machining process on both roughness parameters was large, as per Fields's (2013) conventions (Field 2013). Further statistical analysis found that there was statistically significant differences between planed and sanded surfaces in Ra (p = 4.55*10− 5 < 0.05) and Rz (p = 9.32*10− 3 < 0.05) parameters. The planing appears to have significantly lower values of Ra and Rz roughness parameters than the sanding and face milling (Fig. 4).
However, due to the large scatter of data for face-milled surfaces, this significance was statistically proven only in comparison with sanded surfaces. Therefore, it can be concluded that at the sample level, planed surfaces have a lower roughness than face-milled or sanded surfaces. While at the population level, planed surfaces only have a lower roughness than sanded surfaces. The obtained results of roughness parameters is in agreement with those of Hiziroglu et al. (2014), Ugulino and Hernández (2015) and Smajic et al. (2020) (Hiziroglu et al. 2014; Ugulino and Hernández 2015; Smajic et al. 2020). Obviously, commonly used ways of mechanical treatment of oak wood surface (face milling, sanding, planing) affects its structure, morphology andchemical composition, resulting in surfaces with different wettability characteristics. The above agrees with the statements of Gardner et al. (2014), Kúdela et al. (2016), Liptáková et al. (1995); Santoni and Pizzo (2011); Stehr et al. (2001); Ugulino and Hernández (2015) (Liptáková et al. 1995; Stehr et al. 2001; Santoni and Pizzo 2011; Ugulino and Hernández 2015; Kúdela et al. 2016).
3.2. Bonding quality
Table 6 presents a comprehensive statistical analysis of joint shear strengths in both dry and AA states, categorized by machining methods. Additionally, the table displays the joint shear strength loss (e.g., difference) for each adhesive and machining method matrix after AA, which effectively indicates joint durability.
Table 6
The comparison of shear strengths (N/mm2) of adhesives for each type of machining and both test states, based on statistical analysis with letters of significance and percentage difference.
Adhesive | Face Milling | Planing | Sanding |
Dry | AA | Diff. (%) | Dry | AA | Diff. (%) | Dry | AA | Diff. (%) |
ER | 11.07 (1.74) | b | 11.19 (1.09) | a | 1.08 | 10.33 (1.39) | a | 10.06 (1.83) | b | -2.61 | 12.51 (1.46) | b | 12.42 (1.49) | b | -0.72 |
PUR1 | 14.68 (1.34) | a | 12.65 (2.35) | a | -13.83 | 13.41 (1.05) | b | 13.30 (1.08) | a | -0.82 | 14.76 (1.24) | a | 14.01 (1.28) | a | -5.08 |
PUR2 | 13.79 (1.32) | a | 12.36 (2.26) | a | -10.37 | 13.34 (0.63) | b | 13.25 (0.89) | a | -0.67 | 13.82 (1.28) | a | 13.65 (1.31) | a | -1.23 |
PUR3 | 14.25 (1.30) | a | 14.39 (1.48) | a | 0.98 | 13.63 (1.04) | b | 13.35 (2.05) | a | -2.05 | 14.01 (1.90) | a | 13.91 (1.43) | a | -0.71 |
PVAc | 10.45 (1.06) | b | 7.07 (2.17) | b | -32.34 | 14.02 (1.15) | b | 13.77 (0.93) | a | -1.78 | 14.07 (1.47) | a | 12.42 (1.09) | b | -11.73 |
Letters indicate belonging to a particular group as well as significant differences between groups at 5% significance level.
Therefore, regardless of machining method, joints glued with ER showed statistically significant lower shear strengths in dry state than other adhesives (from 10.33 to 12.51 N/mm2). Nevertheless, the results obtained in the dry state are quite good, especially knowing that only in combination with all machining methods, ER, in addition to PUR3, which implied excellent durability of such joints, i.e. without a significant reduction in shear strength or only with a minor reduction after AA. In face-milled joints, except ER (11.07 N/mm2), PVAc also (10.45 N/mm2) showed statistically significant lower shear strength in dry state than other adhesives. After AA, only the joints glued with PVAc adhesive demonstrated statistically significant lower shear strength than the other adhesives. Although the shear strengths of PUR adhesive types and PVAc after AA are similar, there are variations in their durability. PUR 3 exhibited excellent durability and no loss of shear strength, whereas PUR 1 and PUR 2 had a slightly more noticeable loss, and PVAc had the most significant loss. As a result, PVAc's durability was found to be extremely poor when combined with face milling. When wood surface had been planed, joints glued with ER showed statistically significant lower shear strengths in both, dry and AA, state. It interesting to emphasise that shear strengths of all adhesives after AA were slightly decreased. So, glued joints with planed wood surface showed very good durability. When it comes to sanded surfaces, ER and PVAc show statistically significant lower shear strength in both conditions. However, in contrast to ER and other adhesives, PVAc showed a significant loss in shear strength after AA, dropping from 14.07 to 12.42 N/mm2. It suggests that PVAc is less durable when used in a sanded matrix, but it is still better than using it in a face-milled matrix.
Generally, based on the results of the shear strength tests, PUR types of adhesives have proven to be the most suitable for bonding moisture-resistant face-milled, planed or sanded joints. Further, PVAc was shown to be unsuitable for gluing face-milled surfaces. In several studies, face milling has generally been shown to improve the bond performance of wood surface compared to planing and sanding, although the effectiveness depends on the type of adhesive and type of wood being used (Kläusler et al. 2014; Knorz et al. 2015; Leggate et al. 2020c). It is worth noting that all the tested adhesives showed the highest level of shear strength on sanded surfaces. On the other hand, only in the planed ones, none of the adhesives showed a significant loss of strength after AA, which indicates a good durability of such treated joints. Further, PVAc was shown to be unsuitable for gluing face-milled surfaces.
For a better understanding of the findings, it is necessary to study in more detail the effect of the machining method and AA on the adhesive, as well as on its durability. Therefore, the adhesives were divided into groups based on a statistical comparison of their dry and AA shear strengths, and the above results were confirmed by average percentage of main failure patterns and SEM micrographs that were compared with reference micrographs shown in Fig. 5. It was proven that some adhesive type joints are affected only by different machining. the others are affected by different machining and/or by AA and some are not affected by different machining neither by weather exposure.
3.2.1. Adhesive type joints affected only by different machining
Joints glued with PVAc showed a significant influence of the machining method on shear strength values in dry (F (2, 57.08) = 97.78, p = 3.65*10− 19 < 0.05, ω2 = 0.76) and AA state (F (2, 22) = 45.68, p = 1.47*10− 8 < .05, ω2 = 0.78). The effect sizes (ω2 ≥ 0.14 – large) indicates that the effect of machining method in both states was large, as per Fields's (2013) conventions (Field 2013). Face-milled wood surface glued with PVAc performed had statistically significant lower shear strength values than planed or sanded, which was especially pronounced after AA (Fig. 6). Also, it is important to emphasise a 32.34% reduction (µdry = 10.45 N/mm2 > µAA = 7.07 N/mm2) in the shear strength of face-milled joints and 11.73% reduction in shear strength of sanded joints after AA. SEM micrographs of adhesive penetration and adhesive-line integrity in an PVAc glued joint face-milled before gluing is shown in Fig. 7. Evidently, PVAc has penetrated adequately (not over-penetrate) and the bonding line has integrated well, which is also in agreement with previous researches. According to Frihart (2016) and Hunt et. al. (2018), once applied, water-based glue loses water in the wood, which increases the glue's viscosity, reducing penetration into the wood (Frihart 2016; Hunt et al. 2018). An example where an adhesive usually over-penetrate is ER (see subchapter 3.2.2). Additionally, the authors have elaborated that apart from viscosity, the proper penetration of PVAc also depends on the capability of small polar molecules to permeate cell walls, despite PVAc being a prepolymerized adhesive.
Face milling of oak surfaces before gluing with PVAc proved to be non-effective surface preparation method, especially after AA. Also, even though the PVAc adhesive is water-borne and very polar it was not conclusive that the shear strength depends on the increase in the polar component of the machined surface. Increasing the polarity of the polymers can increase adhesion and cohesion in the dry condition, while greater polarity has little effect in wet and boiling water conditions. Most of the original PVAc emulsions could pass only the dry test since the secondary forces among polymers and between polymer and wood are very weak at high temperature or in water (Qiao and Easteal 2001).
The joint surface failure patterns' mean percentage values clearly demonstrate that PVAc-glued joints primarily exhibit wood failure for all machining methods in both states. (Fig. 5.). This result confirms the obtained shear strengths of the joint, which is essential since most evaluations of wood bonding quality aim to achieve joints that are at least as strong as the wood itself. Although all machining methods showed satisfactory failure patterns values, some adhesive failure appeared in planed joints. Planing is often characterised by deforming wood cells and plasticising the surface instead of properly cutting, as evidenced by the roughness results (Fig. 4), where planed oak surfaces have the lowest roughness. During the wetting cycles, such wooden cells swell and tend to return to their original, undeformed state. This is usually destructive for the glue line. It creates internal stresses in glue line that lead to cohesive failure of the adhesive because it cannot follow the wood deformations. However, PVAc adhesive, which is prepolymerised type of adhesive and based on elastomers, can adapt to the wood's dimensional changes. Moreover, the acetyl groups can be partially hydrolysed to hydroxy groups latively easily with water or/in a high-humidity environment. When applied to wood, the acetyl and hydroxy groups do not form covalent links to wood components: the interaction is through secondary forces. The performance of polyvinyl acetate and its modification as adhesives for wood can be attributed to adsorption and specific adhesion (Qiao and Easteal 2001; Iždinský et al. 2020).
Although sanded surfaces exhibit higher surface energies and polar components than planed surfaces, there is no significant difference in the shear strengths of planed and sanded joints. The prevalence of the dispersive component and non-polar substances in oak makes raising its polarity for waterborne adhesives challenging. Mechanical processing cannot increase the polarity of oak sufficiently to make it more compatible with water-born adhesives because oak is dominantly non-polar. The assumption is that PVAc dominantly creates mechanical adhesion, followed by diffusion due to its anatomical structure and very ineffective adsorption adhesion due to its non-polarity. However, mechanical anchoring can be highly challenging after planing because it results in smooth plasticised surface. The significance of secondary bonds, such as hydrogen bonds, present within the adhesive polymer as well as in the wood adhesive interface, is clear from the ability to restore bond strength after AA. According to Kläusler et al., this is also characteristic of 1k polyurethane-based adhesives (Kläusler et al. 2014), which is also used in this research.
3.2.2. Adhesive type joints affected by different machining and/or by AA
In the second group of joints, the type of machining had a statistically significant influence on shear strength values in the dry state, but this effect was reduced after AA. Such behaviour has been statistically demonstrated for joints glued with ER and PUR1 adhesives (Fig. 8). So, there was a statistically significant difference between machining methods in the dry state of ER (F (2, 85) = 15.28, p = 2.14*10− 6 < 0.05, ω2 = 0.25) and PUR1 (F (2, 85) = 11.29, p = 4.49*10− 5 < 0.05, ω2 = 0.19). In accordance with Fields' (2013) recommendations, the effect of machining method in dry state was large for each adhesive (ω2 ≥ .14 – large). After AA, difference in shear strength between machining methods was reduced for PUR1 (F (2, 22) = 1.57, p = 0.23 > 0.05, ω2 = 0.04) (Fig. 8a) and was not present at all with ER (F (2, 23) = 6.00, p = 8.02 *10− 3 < 0.05, ω2 = 0.28) (Fig. 8b).
In addition, the effect size (ω2 = 0.69 ≥ 0.14 – large) indicates that the effect of machining method in recondition state for ER adhesive was large, but for PUR1 that effect was small (0.01 ≤ ω2 = 0.04 < 0.06 – small), as per Fields's (2013) conventions (Field 2013).
Upon conducting additional post-hoc analysis of the shear strengths of ER joints, it was discovered that a sanded wood surface exhibited statistically significant higher shear strength in dry state compared to other machining methods. Nevertheless, this disparity decreased after AA, and sanded joints only demonstrated statistically significant higher shear strength than planed joints. Sanding proved to be the best surface preparation prior to ER bonding due to higher shear strength and predominantly wood failure percentage than the other two machining methods. Although the percentage of wood failure decreased after AA in favour of adhesive and interface failures, the results were still acceptable. It can be assumed that the use of such joins is possible under conditions of decreased or changing relative air humidity. Based on the data in Fig. 3, sanding has the highest roughness level compared to the other machining methods. As a result, such a surface has a greater specific surface for adequate mechanical anchoring and adhesion, which ultimately improves bonding.
Regarding face-milled surfaces, ER displayed a noticeable lack of flexibility and, interestingly, demonstrated a lower cohesive strength than expected because, in the dry state, the failure pattern was predominantly cohesion failure of the adhesive, amounting to 75%. After AA, this value decreased to 10%. In joints with planed surfaces, adhesive failure was also present but to a lesser degree, at approximately 20%, in each test state. Hunt et al. (2018) stated that starved adhesive lines typically show cohesive failure in the bulk adhesive or in the adhesive interphase (Hunt et al. 2018).
SEM micrographs of adhesive penetration and adhesive-line integrity in an ER-glued joint is shown in Fig. 9. It appears evident that there is a concern with the adhesive-line integrity in face-milled joints (Fig. 9a), which is very pronounced in planed joints (Fig. 9b). However, the shows a good penetration of the adhesive for all surface machining methods. To understand the mentioned phenomena, it is necessary to consider in more detail the flow and penetration of the glue, which largely depended on the characteristics of the glue, the surface treatment and the anatomical features of the oak.
The ER adhesive used in the study had a significantly lower viscosity than other adhesives, causing it to squeeze out of the joint during the long phase under pressure or overpenetrates the surface. The end result of both options is an inconsistent adhesive film that leads to cohesion failure of the adhesive and also reduces the joint's shear strength, which relies only on the film fragments that hold elements. In addition, oak ring-porous anatomy of vessels can result in areas with high lumen and others with low lumens. There is uncertainty regarding the ability of the high-penetration regions (vessels) to make up for low-penetration areas (cell-wall). A potential issue is that if the adhesive penetrates deeply into the wood's lumens, it may draw the adhesive away from the adhesive line, resulting in inadequate bonding between the two surfaces, starved adhesive line and a lack of adhesion. According to literature a starved adhesive line is often caused by the type of wood, the consistency of glue, and the amount of pressure used to bond the wood (Mihulja G and Bogner A 2005; Mihulja and Bogner 2007).
Another useful fact is the in-situ/prepolymerized categorisation of adhesives according to which ER adhesives belong to the group of in-situ polymerised adhesives. in situ polymerised adhesives use both vessels filling and cell wall infiltration to provide permanent bonds, while prepolymerized adhesives are limited to vessels filling. Such adhesives creates a gradual transition from rigid adhesive to solid wood in terms of mechanical properties. Therefore, it is necessary to observe the penetration of the glue through the vessels of oak earlywood and into the cell wall separately.
Based on micrographs 9a-c, lighter areas can be observed around the adhesive line, similar to the wood-adhesive composite layer depicted in Fig. 7c. It appears that the epoxy adhesive has effectively penetrated the cell wall, regardless of used machining method, likely due to its in-situ properties. So, regardless of bad planing influence and its result of surface cells crushing, ER should adhere well, since it will impregnate the damaged interface layer of the joint. This supports the findings made by other authors infiltrating in situ polymerised adhesives have been shown to strengthen weak, mechanically damaged cells adjacent to the adhesive. (Konnerth and Gindl 2006; Frihart 2016). However, upon observing the glue penetration through the pores of the early wood, it was noticed that the joints that were planed and face-milled before gluing had hardened glue (similar to Fig. 7a-b) visible at greater depths than sanded. The vessels filled with wood dust after sanding, preventing ER from overpenetration and nascency of starved glue line as well as the impregnated layer of wood also creates a transition from rigid glue to viscoelastic wood preventing high inner stress and crack propagation through a solid interface layer.
In contrast, face milling and planing typically occurs suitable levels of cell-wall fibrillation (De Moura et al. 2010; Kuljich et al. 2013). ER overpenetrates such surface, thereby diminishing the integrity of the glue line, i.e. it becomes starved (Fig. 9a-b). Therefore, crushed wood cells during machining, ER's low viscosity compared to other adhesives, and its properties related to in-situ polymerisation contribute to failure was prevalent in the adhesive. A critical review paper by Hunt et al. (2018), also states that excessive or insufficient penetration can cause joint weakness and failure, especially when the joint is exposed to water. Weakened adhesive lines usually indicate an adhesive or wood-adhesive interface failure (Hunt et al. 2018).
Despite all of the above, the joints glued with epoxy glue proved to be very durable, as their shear strength did not decrease after AA, regardless of the machining method. Because of in situ polymerization, such highly rigid adhesives have been found to penetrate the cell wall effectively and minimize swelling even in moist environments. Reduced swelling in the interphase of the wood leads to less interphase stress, and thus to a joint more resistant to moisture. Prepolymerized adhesives, on the other hand, are usually not as stiff and can therefore stretch to accommodate the swelling of the wood, thus avoiding stress concentrations (Frihart 2016; Hunt et al. 2018). Hovewer, in instances where in situ adhesives feature a component that is unable to penetrate the cell wall (usually due to high molecular weight), they tend to perform poorly when subjected to accelerated water exposure tests. In epoxy adhesives, this component is usually bisphenol A (Hunt et al. 2018). Therefore, although mechanical interlocking is most likely important for dry-joint wood joint strength, it is not always sufficient for good wet-joint strength.
Furthermore, as a result of a post-hoc test of PUR1 glued joints, planed joints had statistically significant lower shear strength in the dry state, but not after AA. As with PVAc, most joints glued with PUR1 adhesive showed cohesive wood failures regardless of the machining method. A certain percentage of cohesion adhesive failures did appear in planed joints in both states and face-milled joints in dry state. However, after AA, that adhesive failure percentage was reduced for planed joints and not present for face-milled joints. SEM micrographs depicting a joint that had deep penetration, but excellent integrity of adhesive-line which indicate good bonding properties of PUR1 (Fig. 9). (Fig. 9). The prepolymerized adhesives, such as PUR and PVAc, are not as brittle, so they can stretch in order to adapt wood swelling. For this reason, PUR1 can also easily adapt to changes in moisture levels, as previously described for PVAc.
3.2.3. Adhesive type joints affected neither by different machining nor by AA
Figure 7 illustrates that the machining method does not affect the shear strength of joints glued with PUR3 (F (2, 50) = 2.51, p = .09 > .05, ω2 = .05 ) or PUR2 (F (2, 55) = 2.07, p = .14 > .05, ω2 = .04 ) adhesives, tested in dry state. The effect size (.01 ≤ ω2 < .06 – small) indicates that the effect of machining method in dry state for both adhesives was small, as per Fields's (2013) conventions (Field 2013). Moreover, statistically significant difference was not confirmed neither in the AA state of joints glued with, PUR_fibers (F (2, 21) = 1.38, p = .27 > .05, ω2 = .03) or PUR2 (F (2, 24) = 0 .68, p = 0.51 > 0.05, ω2 = − .02) adhesives. The effect size (.01 ≤ ω2 < .06 – small)) indicates that the effect of machining method in recondition state for PUR3 adhesive was small, and for PUR2 that effect was very small (ω2 < .01 = .04 – very small), according to Fields's (2013) conventions (Field 2013). The shear strength of joints glued with PUR2 increased from 11.51 to 13.64 % depending on the machning method. While the joints glued with PUR3 had approximately the same shear strength for face-milled joints and about 11 % increased shear strenth for planed and sanded joints. Based on the examination of the surface failures, it was determined that the PUR2 adhesive produced a quality joint for all machining methods, it both states (Fig. 7b). In the case of PUR3 glued joints, the proportion of adhesive failure is dominant for all machining methods in the dry state (Fig. 7a).
However, after AA, adhesive failures decreased primarily in favour of interface or wood failures. The fibre-reinforced 1k PUR adhesive is a moisture-curing adhesive, which most likely accounts for the ratio of surface failures observed in a dry state. Consequently, the adhesive was not able to bond properly until it was subjected to increased humidity. So, despite the poor cohesion of the glue and thus poor adhesion in dry state, PUR3 form joints of adequate quality after the wetting cycle.