Method for evaluating the bond strength development of pMDI adhesive using ABES

The primary adhesive used was I BOND PM 4300 from Huntsman (Tienen, Belgium), a standard polymeric diphenyl methane diisocyanate adhesive suitable for OSB production. It has a viscosity of 220 mPa·s (PU/VIS-1 method) at 25 °C and an NCO value of 30.75 with a specific gravity at 25 °C of 1.23 (Huntsman 2016). In addition to pMDI, a phenol-resorcinol-formaldehyde (PRF) adhesive, Aerodux 185 from Dynea AS (Krems, Austria), was used as a reference to determine the influence of pressing pressure and to determine the ultimate bond strength achievable with the veneers used. The PRF adhesive has a solid content of 55–61%, a viscosity at 25 °C of 400–1500 mPa·s (Brookfield RVT, spindle 4 at 20 rpm) and a specific gravity at 25 °C of 1.15 (Dynea 2014). It was prepared with a 100:20 weight ratio of adhesive to hardener.

Measurements were done using an ABES device (Adhesive Evaluation System, Inc., Oregon, USA). Several preliminary tests were conducted to identify the factors that influence the curing of pMDI during ABES measurements. The identified factors were:

To investigate the influence of water, one series of measurements was carried out with and one series without additional water. In the series with additional water, 30 g/m² of water was applied followed by 30 g/m² of adhesive. Water and adhesive were applied to both veneers, resulting in a total application of 60 g/m² for water and adhesive, respectively. The overlap area was 5 × 20 mm² for all measurements in this study, as prescribed by the ASTM-D7998-15 (2015). Water and adhesive were applied in a very uniform manner using a 4 mm wide flat paintbrush with synthetic bristles. Adhesive and water were applied with individual paintbrushes. The desired amount of application was controlled by weighing the samples to an accuracy of ± 0.001 g. Any excess water was removed by pressing the tip of the sample against the paper towel. The water was applied before the adhesive because pMDI does not mix with water and when applied first it forms a repellent film on the surface, preventing water from penetrating into the wood. Additionally, this better simulates the industrial process sequence, where dry strands or particles are typically moistened with water to the desired moisture content prior to adhesive application. Samples were prepared for a maximum of 90 s, no longer to avoid evaporation of the applied water. Only a few samples at the time were taken from the climate room and were kept in a sealed plastic bag to prevent drying. As pMDI also bonds to steel, a 0.08 mm thick polytetrafluoroethylene (PTFE) sheet was placed between the sample and the press platens. Baking paper can be used too, but PTFE was found to be more durable. An influence on the temperature development was not noticeable. The pressing temperature was set at 110 °C. This is approximately the plateau temperature which is typically reached in the core of a panel during hot pressing when sufficient moisture and pressure are present (Solt et al. 2019a). The specific pressing pressure was set at 2.5 MPa. Measurements were done without cooling the samples before pulling them apart. After testing, the percentage of wood failure was visually examined and the actual overlap area was measured using a calliper. According to ASTM-D7998-15 (2015), a minimum of eight repetitions should be done at each selected pressing time. Instead of the proposed eight replicates, we only did one repetition per selected pressing time, but the increment between the measurements was small. This allows a better visualization of the bond strength development and a better fitting of a curve to the measured data. At the beginning, when the curing was fastest, more measurements were done so that more data points were available to define the slope of the curve, which is important in determining the curing rate. As curing slowed down, the intervals between measurements were increased.

To examine how long the applied water actually stays in the bond line before it dries out, a series of measurements was carried out with only water applied and no adhesive. As with all the other measurements, 30 g/m² of water was applied to each veneer. The veneers were then placed in the device and pressed together. After a certain pressing time, the samples were taken out and the overlapping area was cut away with a paper cutter, which allowed a precise and quick cut. For each sample pair, two cut-offs were obtained and the moisture content was determined by oven dry method. The cut-offs were weighed to an accuracy of ± 0.0001 g. A schematic description of this procedure is shown in Fig. 1.

The effect of pressing pressure was investigated by conducting measurements at three different specific pressing pressures, namely 1.5, 2.0 and 2.5 MPa. In this setup, only the pressure was varied, all other parameters remained the same as described above. Measurements at different pressing pressures were also performed with PRF adhesive as a reference. In these measurements, the adhesive application was 300 g/m² (one-sided) and the pressing time was 100 s aiming for full cure of the adhesive. All other parameters remained the same as for the other measurements.

Furthermore, to analyse the influence of adhesive application, measurements were conducted with both one-sided and two-sided adhesive applications. In the one-sided application, 30 g/m² of water was applied to the first sample, followed by 60 g/m² of adhesive. Then, 30 g/m² of water was applied to the second sample without adhesive. For the two-sided application, 30 g/m² of water was applied first on one sample, followed by 30 g/m² of adhesive application, and then the process was repeated for the second sample. The pressing parameters were the same as in the previous measurements, with a pressing temperature of 110 °C, a specific pressing pressure of 2.5 MPa, and a pressing time of 800 s.

Overexposure to MDI is known to be harmful to the respiratory system and may lead to asthma (Allport et al. 2003). The authors also report that MDI can be irritating to the eyes and skin, and in some cases skin sensitisation may occur. Especially at higher temperatures, the vapour pressure of MDI increases, presumably leading to higher emissions. This is particularly relevant during hot pressing at elevated temperatures. To determine whether MDI emissions are also emitted during ABES measurements, they were measured using the portable ISoSense Sampling Unit from DOD Technologies, Inc., Cary, USA. It consists of a small air pump that pumps air through the tube at a constant flow rate. At the beginning of the tube is a holder for chemically impregnated filter paper, which produces a coloured stain when it comes into contact with isocyanates. From the intensity of the stain, a concentration can be determined by visual comparison with the intensity reference concentration card. The device is calibrated for monomeric MDI, so it serves only as a rough estimation when using polymeric MDI (DOD Technologies Inc). Measurements were done in a closed and switched off fume hood measuring approximately 145 cm x 80 cm x 145 cm, giving a volume of about 1.7 m³. The first measurement was taken with the sample holder positioned just about 1 cm above the press, and the second measurement was taken approximately 60 cm away, which is roughly the distance of an operator during testing. The setup was the same as described above, with a total water and adhesive (pMDI) application of 60 g/m² each (two-sided), a pressing temperature of 110 °C and a specific pressing pressure of 2.5 MPa. The pressing time was set to 100 s and five ABES measurements were conducted during the 10-minute emission measurement. The primary goal of the measurements was not to measure the exact emission concentration, but rather to determine if any detectable emissions were present.

The statistical analysis and curve fitting were carried out using the Python programming language. Libraries used for this study included Pandas for data manipulation, NumPy for numerical operations, SciPy and Statsmodels for statistical analysis, and Matplotlib and Seaborn for data visualization. The Hill function (1) (Hill 1913) was used for curve fitting to the measured data, as it showed the best fit among the tested models, including the hyperbolic tangent function, which was previously shown to be suitable for amino resins (Costa et al. 2013; Martins et al. 2013), and the Gompertz function. The Hill function is a three-parameter model. In the context of our study, τ_max represents the upper asymptote, t₅₀ is the half-time required to reach that asymptote, and n is the Hill coefficient. The curve_fit function from the SciPy package, which uses the Levenberg–Marquardt algorithm for nonlinear least squares fitting, was employed. Initial parameter estimates were set as follows: τ_max as the maximum measured bond strength, t₅₀ as 100, and n as 1. All data sets began at 0 s pressing time and 0 MPa bond strength.

A t-test was used to determine whether there was a significant difference between two groups. ANOVA (analysis of variance) was used to compare the means of several groups. The α-value for all tests was 0.05.

The influence of water on curing of pMDI was investigated by performing ABES measurements with and without application of additional water. The measurements are shown in Fig. 2, where each point represents one measurement and the circles indicate the percentage of wood failure. Figure 3 shows representative samples with different percentages of wood failure, which were visually assessed after testing. When measurements were conducted without water, low strength values and almost no wood failure were obtained compared to when water was applied to veneers (Fig. 2). To analyse the presence of water during pressing, the moisture content of the overlapping area of the samples with applied additional water was determined. It was found that most of the water was gone after a very short time, about 100 s. This is shown by the orange curve in the Fig. 2. At the time when there is still some water in the sample, the curing process is also the fastest visible as the steepest part of the blue curve in Fig. 2. The interpretation of the moisture content is in our case a bit simplified, as the calculation considers the water already present in wood as well as the one applied to the surface. Moreover, these measurements were done without adhesive, which would also consume water during curing. This aspect has been neglected for simplification. Additionally, the moisture content in the overlap area may change immediately upon opening the press due to water evaporation, introducing a potential source of systematic error. However, since all samples were measured in the same way, the trend of decreasing moisture content over time can still be observed reliably. Anyhow, the initial moisture content of the samples was around 20%. In the industrial production of wood-based panels, the upper limit of mat moisture content is 10–15% when using pMDI (Frazier 2003). Yet it is possible to produce lab-scale OSB boards with moisture content of the strands as low as 3% (Direske et al. 2017). However, panel production is a closed system where most moisture evaporation could occur mainly at the edges, which represent a comparable small share of surface especially when considering industrial dimensions.

The pressing times used to achieve maximum strength when additional water was applied were quite long as it took approximately 800 s to reach 100% wood failure (Fig. 2). A similar study using ABES with 0.84 mm thick beech veneers found that it took about 180 s to achieve maximum strength with UF adhesive at 110 °C and 300 s with PF adhesive at a pressing temperature of 130 °C (Wibowo et al. 2024). Another research by Hogger et al. (2020) used 1.5 mm thick birch veneers and reported that the final tensile shear strengths for UF and PF at 110 °C were achieved after 300 s. The longer pressing time in our study may be attributed to the specific wood species used, as preliminary tests indicated that beech was more challenging to bond compared to pine, oak, and birch. The species dependence of pMDI adhesion has been previously documented (Das et al. 2007; He and Yan 2007). Beech was used because it is specified by ASTM-D7998-15 (2015), although hard maple (Acer saccharum spp.) can also be used.

Based on the measurements performed, it can be seen that the application of water has a major influence on the curing of pMDI. This effect was also found by Smith (2004) in ABES measurements performed with 0.69 mm thick aspen veneers, although comparable strengths were not achieved; the highest strength recorded approximately 1.4 MPa at a moisture content of 23%. Similarly, Solt et al. (2019b) observed that increased moisture content enhanced the curing rate of pMDI in lap shear tests, but their samples were 1.5 mm thick with an overlapping area of 10 × 30 mm, which is inconsistent with standard ABES samples.

Beside importance of water, the influence of pressing pressure was also investigated by performing ABES measurements at three different specific pressing pressures as shown in Fig. 4. It can be seen that the final strength achieved within the observed pressing times increases with higher pressing pressures, while the slopes of the curves are similar at the initial stage. Only at a pressure of 2.5 MPa, 100% wood failure was consistently achieved. This indicates that the adhesive strength was higher than the wood strength. This increase could be due to the local densification of the wood in the overlapping area, leading to locally higher shear strength of the wood. But when measurements were done with PRF adhesive as a reference at the same specific pressing pressures, no significant differences were found (p-value 0.99). The average tensile shear strengths for samples bonded with PRF adhesive were 9.25 ± 0.94 MPa, 9.22 ± 1.08 MPa, and 9.24 ± 1.48 MPa, corresponding to specific pressing pressures of 1.5 MPa, 2.0 MPa, and 2.5 MPa, respectively. pMDI is also used for rigid wood fibre insulation panels (Krug et al. 2023), which are produced at low pressing pressures. For example, in a study where wood fibre insulation panels were produced on a laboratory scale using emulsified MDI, the specific pressing pressure was 0.49 MPa (Lee et al. 2019). This raises the question of why higher pressures are beneficial. In a study by Gavrilović-Grmuša et al. (2016), using UF adhesive and performing tensile shear tests using solid wood lamellas as substrate according to EN 205, they observed that the shear strength increased with increasing pressure, but then decreased again. The pressures used were 0.5, 1.0 and 1.5 MPa, showing no clear trend for increased strength with increasing pressure. Furthermore, they observed that penetration increased with increasing pressure up to a certain level and then levelled off. Gruver and Brown (2006) also found a rather poor correlation (r² = 0.59) between penetration depth and shear strength when using pMDI adhesive and conducting compression shear block tests. pMDI is known to perform well even when it overpenetrates, which contradicts traditional views on wood adhesion (Frazier 2003). In addition, microscopic images of ABES samples pressed at different pressing times were also taken in our study (not shown here), but no apparent difference in penetration depth was observed. However, a plastic deformation of the wood structure could be observed resulting in reduced porosity. This results in more contact area in the interface and more opportunities for secondary forces to form. Yet this does not explain the high tensile shear strengths (8.1–9.7 MPa) obtained by Sonnenschein et al. (2005), where specimens were bonded with pMDI by pressing them only with binder clips and therefore at comparatively low pressure. It should be noted that their specimens were made of pine and oak with a thickness of 3.175 mm (1/8 inch) and an overlapping area of 6.35 × 12.7 mm. Curing was carried out in an oven at 150 °C for one hour. Thus, the reason may be similar to that proposed by Solt et al. (2019b), that thicker veneers contain more water for curing. The reason for achieving higher strength at higher pressing pressure in our study might be due to different heat and mass transfer during pressing. When wood is more densified, water vapour convection is reduced, which may result in slower water transport and consequently slower drying of the veneers, leaving more water available for reaction. On the other hand, less porous wood has better thermal conductivity. Nevertheless, our findings show that, in addition to the application of additional water, high pressure also plays a crucial role in the curing process during ABES measurements.

The third parameter investigated was the adhesive application, where ABES measurements were done with one-sided and two-sided adhesive application. When the adhesive was applied two-sided, higher shear strengths were obtained than when applied one-sided (Fig. 5). The difference was found to be significant (p-value = 0.0141). This could be attributed to the greater symmetry in the bond line, where the adhesive penetrates similarly deep into both adherends and has intimate contact and equal access to moisture from both adherends. Two-sided application resulted in a very thin film rather than the thick film produced by applying twice the amount of adhesive to one side only. This gives the functional NCO groups of the pMDI more opportunities to react (similar to mixing pMDI with water, only some of the pMDI will react with the water at the interface, the rest remains uncured). Regarding the type of wood failure, no notable differences were observed between the two adhesive application methods in terms of a higher concentration of wood failure occurring in one adherend over the other. Some studies have claimed that pMDI performs better when it remains on the surface (Sonnenschein et al. 2005, 2009), while other studies have claimed that deep penetration into the wood structure resulted in better performance (Smith 2004). The findings of our study do not agree with those of Smith (2005), who found that the tensile shear strengths using ABES were higher when the adhesive was applied on one side only. The adhesive was applied by a modified flexographic printing technique (Smith 2003). Additionally it was found that it was better if the resin application was low (application of 2% weight gain was better than 5% and 10%) (Smith 2005). In another similar research, Smith (2004) also concluded that the highest strength was achieved by samples with the lowest resin content (1% per dry weight of wood) combined with the highest moisture content tested (23%). Humphrey (2005) concluded as well that low resin application rates are preferable when using pMDI. Smith (2005) attributed the higher strength with lower resin application to the available water. If there is excessive adhesive in the bond line, it is likely that some of it cannot be fully cured due to lack of water and the rest remains uncured. Lower strengths with two-sided application are explained by a possible lack of adhesive in the interface (Smith 2005). In our case, the adhesive application rate was higher than reported by Smith (2005), but additional water was applied, which was assumed to allow the adhesive to fully cure. In our preliminary tests, adhesive was also applied at a rate of 100 g/m² (two-sided), which gave high strengths and 100% wood failure. Therefore, it cannot be confirmed that the applying lower resin rates will give better results. We assume that it is a matter of available water to achieve complete curing of the applied adhesive. However, lower adhesive application rates in the form of dots, as presented by Smith (2003), better represent the industrial conditions of OSB production. The adhesive application rate in our study was also reduced from 100 g/m² to 60 g/m² to more closely reflect industrial conditions, where resin application rates are known to be lower when using pMDI compared to formaldehyde-based polycondensation adhesives. Correspondingly, the amount of additional water was also reduced to the lowest possible level while still ensuring effective curing. Additionally, once 100% wood failure is reached, the strength cannot be increased any further, so increasing the adhesive application rate will not result in any further improvement of tensile shear strength. It can therefore be concluded that the selected amounts of water and adhesive are appropriate and that two-sided application is preferable.

To assess potential operator exposure, MDI emission measurements were conducted during ABES testing. The results showed that some emissions were detected (13 ppb) when the sample holder was positioned about 1 cm above the ABES press. The typical short-term (15 min) occupational exposure limit for MDI is 20 ppb (Allport et al. 2003). No emissions were measured when the sample holder was 60 cm away from the press, reflecting a realistic operator distance. Although not necessary according to the measurements, we decided to use a point extraction system above the press to eliminate any possible emissions and to ensure the safety of the operator.