Analysis of Wood Bonding Failures that Initiated Before Adhesive Solidification: Air Fingers and Cavitation

Bond line openings while an adhesive is still viscous (liquid or paste-like) leave characteristic marks. Air fingers and cavitation develop on the surface of the adhesive when joint parts are fully or partially separated before solidifying of the adhesive. Therefore, the observation of air fingers or cavitation provides important clues on the root cause of the failure of bond lines. There is still limited knowledge about the factors that lead to the formation of either air fingers or cavitation. Additionally, a resoftening of thermoplastic adhesives by high temperatures may be confused with air fingers or cavitation that developed before the initial curing. To improve the understanding of the adhesive structures, we assemble some experiments on this phenomenon. Air fingers dominate when air easily penetrates the adhesive, the adhesive thickness is high, and the adhesive has a low viscosity (early in the process). Cavitation dominates when the adhesive layer is thin, when the viscosity is high (late in the process) or when the air ingress is restricted.


Introduction
Significant deviations from ideal processing conditions (especially in combination) may significantly weaken adhesive joints. If not detected by subsequent control measures, these weak joints may lead to costly esthetic or technical problems and sometimes to injuries and deaths. An evidence-based analysis of such bonding failures is required to address whether the materials used, the processing conditions, the improper use conditions, or combinations of the three factors are the root cause.
The root cause of adhesive problems is recognizable by the appearance of the failed glue line, the location of defects, and the frequency of its occurrence [1,2]. Light microscopy and scanning electron microscopy are frequently used techniques for failure analysis of wood bonds. The type of adhesive is determined by infrared spectroscopy. Staining methods may also identify the type of adhesive (e.g., staining of PVAc with iodine) and improve the contrast of the adhesive to the wood [1,3]. Figure 1 shows schematic microscopic illustrations of cross sections from a perfect bond line and some typical failures of wood and wood-based materials. The solidified adhesive may separate within the glue line (cohesion failure), in the interface of wood and adhesive (adhesion failure) or within the wood (wood failure). An overly low pressure or overly high viscosity of the adhesive leads to thick bond lines with low adhesive penetration in the wood pores [4]. The advantage of a good penetration of low molecular weight thermosetting adhesives (melamine-, urea-, and phenol-formaldehyde) has been demonstrated [5,6]. These adhesives also penetrate into the wood cell walls and stabilize the weak boundary layer of the wood surface. A positive influence of a good penetration is also observed for not cell wall penetrating adhesives like polyurethane, polyvinyl-acetate and epoxy-resins [1]. A too high pressure or a too low viscosity of the adhesive may lead to a starved bond line with overly high penetration and almost no adhesive remaining in the glue line. There is no perfect level of penetration. For a particular mill under a particular set of conditions the penetration correlates to bond performance, but a general answer to the question of ideal level of penetration does not exist [5].
Separations that occur while the adhesive is still in a viscous, liquid or paste-like, state are regularly observed in failure analysis [1]. The two pieces may have never contacted, or they may have contacted initially but separated before the adhesive solidified (Fig. 2). In the latter case, either cavitation (a honeycomb-like structure) or air fingers (dendritic structures) develop. The separation of the glue line may be complete. Often, the two pieces are still fixed by small remaining contact points, and final separation occurs during use.
The usefulness of these air fingers and cavitation for failure analysis is easily recognizable. The observation of these structures in failing adhesive joints allows a profound statement concerning the viscosity of the adhesive at separation. However, there are limitations to the interpretation of these patterns. For thermoplastic adhesives, resoftening may be promoted by elevated temperatures or chemical influences during use. Additionally, the structures might be confused with foaming polyurethane adhesives; however, the bubbles in polyurethanes may stack, while cavitation appears in this work as a monolayer. Soft adhesives that are not commonly used in the wood industry might develop similar patters by creep.
A resoftening of thermoplastic adhesives has thus far not been identified in failure analysis by the Fraunhofer Institute for wood research (WKI); nevertheless, the possibility of its occurrence cannot be excluded. We have observed this pattern several times in blisters of dark coatings with overly temperature-sensitive base coats. These patterns have been observed in renovation coatings when linseed oil residues in the wood are present. The oil penetrates and resoftens the new coat.
Several authors have investigated the principles of the formation of fingers and cavitation in detail. Depending on the separation speed, the viscosity and the thickness characteristics of the fluid, air fingers or cavitation may occur [7][8][9]. Cavitation originates from unavoidable preexisting bubble nuclei in the range of a few micrometers (lm) that enlarge when pulling forces are applied [9]. Further studies have concentrated on the transition from the growth of individual cavities to the collective growth of a population of cavities with elongating walls [10]. Several studies have improved the understanding and mathematical modeling of the formation of fingers [11][12][13]. The experiments described in the literature have been performed with water-free  substances, such as silicone oils, and air-impermeable substrates, such as steel and glass, however, most adhesives in the wood-based products industry have water contents between 30 and 60% and wood quickly absorbs water and thereby reduces the viscosity of an adhesive and wood contains air that might change the formation of the structures. The factors that influence the shapes of the structures and the occurrence of the finger-type vs. the cavitationtype, the research presented here investigates on the influences of the separation speed, substrate, thickness and viscosity characteristics of the adhesive.

Material and Methods
A commercial PVAc (polyvinyl acetate) adhesive (solid matter 50% and viscosity 12.000 mPaÁs) was stained with rhodamine B to increase contrast for the video-documentation. 0,1 ml of the stained adhesive was applied on the center of a 60 mm 9 60 mm glass plate (Fig. 3). A 50 mm 9 50 mm block was pressed on the glass for 30 s; a pressure of 10 N was the standard, and lower or higher loadings than the standard was applied to deviate the film thickness or viscosity. Subsequently, the block was pulled off according to EN 319 with a self-aligning ball-andsocket joint in a tensile testing machine Zwick Z010 (Fig. 3). The pulling speed was 2 mm/min as a standard. The formation processes of fingers and cavitation were video documented through the glass plate. Different viscosities of the adhesive were attained through reduction in the water content by drying. To prevent both skin formation and air intrusion, the adhesive was kneaded manually during drying. The viscosity of the adhesive was measured with a rotational viscometer CV 120 Bohlin according to EN 12092 [14]. We used three different surfaces for the experiments: the flat white melamine-resin top side of a high-pressure laminate (HPL), the reverse side of an HPL consisting of paper impregnated with phenol resin and pronounced unidirectional sanding marks, and birch plywood. Four replicas of each variant were tested. The sequence starts with the development of small air fingers at the rim of the adhesive (Image B). As the fingers grow, some bubbles appear in the center of the adhesive (Images C and D). These bubbles are attributed to the expansion of preexisting microvoids [9]. Immediately before the final separation, the adhesive concentrates in one area (Image E). In this area, the previously formed fingers decrease, followed by the formation of cavitation. Image F shows the dried adhesive on the glass plate. Obviously, some finer structures are lost during the drying process; however, the structures of fingers and some cavitation are still easily recognizable. With a low separation speed (2 mm/min), the lasting structure of the adhesive consists of only fingers. With increasing separation speed (20 or 200 mm/min), the amount of cavitation in the center of the adhesive increases (Fig. 5).
The thickness of the structure is adjusted by using different levels of initial pressure. The higher the initial pressure is, the larger the area spread by the adhesive. The adhesive is applied in a known volume, and the resulting thickness of the adhesive is calculated from the area covered by the adhesive. With increasing adhesive thickness, only air fingers with decreasing numbers of fingers and branches develop. (Fig. 6).
The increase in the viscosity of the adhesive thins the branches of the fingers and increases the amount of cavitation in the center (Fig. 7).
The use of different substrates has a pronounced influence on the structures of the adhesive (Fig. 8). Only fingers form on the smooth top side and the rough reverse side of the HPL. On the reverse side of the HPL, the preferential direction of the fine fingers corresponds to the direction of the sanding marks. The patterns on the plywood are a Fig. 4 Time resolved development of air fingers and cavitation, photographed through a glass plate while pulling the overlaying block with an HPL surface. The viscosity of the adhesive was 12,000 mPaÁs, initial pressure was 10 N for 30 s, pull-off speed was 2 mm/min, thickness of the wet adhesive film was 75 lm, and Fmax was 57 N. Image (F) shows the glass plate after the adhesive dried Fig. 5 Patterns of the adhesive resulting from different pulling speeds. The viscosity was 12,000 mPaÁs, and the initial pressure was 10 N for 30 s. The wet adhesive had a 70-lm thickness Fig. 6 Patterns of the adhesive resulting from the increasing thickness. The pull-off speed was 2 mm/min Fig. 7 Patterns of the adhesive resulting from the increasing viscosity and initial pressure. The pull-off speed was 2 mm/ min mixture of air fingers and cavitation. Wood absorbs water. Therefore, the initial viscosity of the adhesive immediately increases upon contact with wood, and the resulting structures are typical for a high-viscosity adhesive.
In the tests described thus far, one side of the assembly is glass. To gain better insight into real-life assemblies, we have tested the combination plywood-plywood and reverse side HPL-reverse side HPL. We have performed three setups: 1. Separation of the blocks after 30 s. 2. Separation after 5 min of holding time. 3. Separation after several days. Before separation, these blocks are heated to 90°C for 24 h. The hot blocks are torn apart immediately after being removed from the oven.
The separation after 30 s does not reveal substantially changed structures relative to the previous experiment with a glass plate on one side. However, after 5 min of holding time, the structures differ significantly. Within 5 min, the viscosity of the adhesive obviously increases significantly due to the water uptake of the substrate. This phenomenon results in small cavitation in different shapes, sometimes creating irregular structures where the cavitation is not easily detected (Fig. 9).
The resulting structure of the samples heated after the full curing of the adhesive always consists solely of very fine cavitation that requires microscopic images for documentation. The structures match with the patterns of the separation after 5 min of holding time and with the patterns observed in open blisters of solar-heated coatings with an overly temperature-sensitive base coat (Fig. 10).
In all experiments, at least the very rim of the adhesive showed air fingers. Airtight substrates only form fingers through air intrusions from the edges. If no intruding air is present, only cavitation develops. The formation of cavitation requires energy for the flow of the adhesive and additional energy to enlarge the preexisting microvoids [9]. The birch plywood that we use for some of the experiments consists of approximately 50% air. In theory, this air easily penetrates the adhesive and therefore promotes the formation of air fingers because no vacuum has to be built. Nevertheless, in experiments and practical failure analysis, we have often observed cavitation on wooden substrates. This phenomenon may be explained by the fast increase in viscosity due to water absorption into the wood. We speculate that a gradient of viscosity within the bond line might be present. The outer zones of the adhesive with high viscosity may prevent air ingress, which only allows the formation of cavitation in the low-viscosity center of the bond line. Figure 11 shows a scheme of the possible pathways of air fingers and/or cavitation.
In a failure analysis, some of the major influencing factors for air fingers/cavitation can usually be easily identified. The adhesive type is either known or identifiable (for example, by infrared spectroscopy), and the thickness of the adhesive is measurable. Furthermore, it is possible to assume or determine whether air intrusion from the substrate is possible. It is feasible to estimate the initial viscosity of the adhesive for a specific application, leaving the separation speed and the viscosity of the adhesive during the separation as factors determining the presence of either finger structures or cavitation in the surface of the adhesive. To date, the interpretation of these structures cannot be regarded as an exact science, but it offers an opportunity to rule out at least some scenarios of failure.

Conclusions
The experiments have confirmed that factors facilitating the flow of the adhesive during separation lead to large finger structures. Conversely, factors hindering the flow of the adhesive lead to small finger or honeycomb structures (cavitation). The flow of an adhesive is promoted by a low internal friction (viscosity), a slow separation speed and a high adhesive thickness [9].
For thermoplastic adhesives, the experiments do not fully reveal a clear indication of how to distinguish between separation during production and during use. If fingers are present, a separation during production can be assumed while the viscosity of the adhesive is very low. However, for a high-viscosity separation, the patterns may be similar for initial separation and separation during use at elevated temperatures. Therefore, we must leave the possible differentiation of these failure sequences of thermoplastic adhesives to future research. Fortunately, failures caused by the softening of adhesives through elevated temperatures during use are presumably rare and can usually be ruled out by the assumed use conditions. The effect has a higher relevance for the detachment of overly temperature-sensitive coatings.
Funding Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.