|Abstract or Summary
- There is no doubt that adhesive penetration plays an important role in wood bondline joint performance and durability; yet, to date there is no direct experimental evidence linking penetration depth with bond performance. This is, in part, because adhesive penetration is commonly assessed with various 2D microscopy techniques that are destructive to the specimen, and are incapable of detecting the 3D penetration path followed during bond formation. Micro x-ray computed tomography (XCT) is a non-destructive imaging technique, capable of providing 3D bondline data. A significant challenge, however, has been generating sufficient x-ray attenuation contrast between cured adhesive polymers and wood cell walls for quantitative material segmentation. In this work, three separate wood-adhesive types, phenol formaldehyde, polymeric diphenylmethane diisocyanate, and a hybrid polyvinyl acetate, were uniformly tagged with iodine to overcome these challenges. Laminate bondlines, prepared with these
adhesives and three different wood species, Douglas-fir, loblolly pine, and hybrid poplar, were analyzed with synchrotron-based, micro XCT. Resulting bondline reconstructions had approximately 1.5 μm voxel dimensions, and were segmented into wood and adhesive material phases with simple gray-scale image histogram threshold operations.
Elemental analyses and fluorescence microscopy were used to confirm iodine tags remained associated with adhesive polymers in the liquid resins, and throughout the bonding process. Both iodine concentration and cured adhesive density were demonstrated to significantly impact x-ray attenuation behavior. Furthermore, XCT acquisition parameters were compared and optimized for the highest absorption contrast in these wood-bondline specimens. However, phase-contrast edge enhancement artifacts were also discussed, and the use of a quantitative phase retrieval reconstruction method was demonstrated with one wood-bondline specimen.
Volumetric penetration behavior was quantified and compared between six XCT replicates from nine different adhesive/species treatment combinations, using two different calculations. Results showed penetration variability between different treatment combinations, between individual replicates within a particular treatment, along the length of a single specimen, and from one side of the bondline to the other. Penetration behavior appears to be correlated with adhesive viscosity, but is also strongly dependent on the wood anatomy in the vicinity of the bondline. Cell wall penetration behavior was also observed around adhesive-filled lumens for some of the treatment types; though, distinguishing between the pure material signals inside the wood cell walls was below the resolution limits in this study. XCT penetration results showed good agreement with
results from fluorescence micrographs of specimens excised from the same laminates. However, the 3D adhesive-phase sub-volumes provided novel views of the cured adhesive networks and morphology, previously indistinguishable with conventional 2D surface microscopy techniques.
Undamaged XCT specimens, 3D penetration results, and digital, segmented bondline data from this work are being used in a broader, ongoing project to develop a micro-mechanics numerical model to assess wood-bondline joint performance, while quantitatively accounting for the role of adhesive penetration.