Time-resolved laser spectroscopy for the in situ characterization of methacrylate monomer flow within spruce

Flow and sorption within wood are non-trivial processes as evidenced by numerous investigations spanning more than four decades (Devi et al. 2004; Siau 1984; Bramhall 1971; Bratasz et al. 2012; Gething et al. 2013; Ahmed et al. 2013; Zhang et al. 2014; Fackler and Thygesen 2013). Most of these studies have focused on the sorption of moisture by wood, which has implications for living plants (Tyree and Ewers 1991) and for timber as a construction material (Bratasz et al. 2012; Xie et al. 2011; Telkki et al. 2013). Although the majority of studies focus on moisture movement during wetting or drying (Eitelberger et al. 2011; Hill et al. 2010), there have also been investigations into water and solute migration within saturated wood for biorefining purposes (Jacobson et al. 2006). The movement of supercritical CO₂ within wood is an area that has also recently gained interest as supercritical CO₂ has the ability to act as an environmentally benign carrier for wood preservatives (Gething et al. 2013; Acda et al. 2001; Muin and Tsunoda 2004). With the exception of supercritical CO₂, however, few recent studies exist investigating the flow of non-aqueous solutions through wood. Herein, the movement of monomer, specifically, is of interest. While investigation of the final distribution (topochemistry) of non-aqueous treatments within wood is well known (Mahnert et al. 2013), as are studies into wettability and sorption with non-aqueous media (Moghaddam et al. 2014), the kinetics and flow processes have not typically been characterized.

Changes in liquid viscosity, surface tension and ability to wet the solid surface have a dramatic effect on the flow behavior within a material, for example, the rise of fluid within a capillary (Reed and Wilson 1993; Zhang et al. 2007). The results gained from the study of moisture within wood cannot be directly applied to monomer movement through wood as many of the physical properties of the liquid have changed. New methods to monitor the movement of non-aqueous solutions are, thus, required.

The treatment of wood with preservatives has long been carried out to extend the in-service lifetime of a timber section. Although some species of wood are regarded as durable and resistant to biological or environmental degradation, other species of wood such as the spruce investigated here are more susceptible to attack (Moore 2011). Other treatments have also been carried out in order to increase the dimensional stability or hardness of the wood under investigation. Preservatives, which have in the past been utilized extensively, include pentachlorophenol and copper chrome arsenate (CCA) (Schultz et al. 2008). Chemical modification of the wood itself through acetylation or furfurylation and monomer treatment have also been carried out (Jebrane and Sebe 2008; Chang and Chang 2003; Evans et al. 2000; Hill et al. 2005; Goldstein and Dreher 1960; Zhang et al. 2006; Ajji 2006). Some of these treatments rely on aqueous or gaseous transport through wood, but others depend on non-aqueous flow processes.

In order to measure the flow of a liquid through wood in situ, nondestructive methods are essential so that a single area can be measured repeatedly over time. Nondestructive wood characterization methods such as gamma- or X-ray tomography (Steppe et al. 2004), thermography, as well as microwave, ultrasonic and NMR spectroscopy (Bucur 2003) are powerful tools for mapping changes in density and, hence, anatomical characteristics within wood (Lindgren et al. 1992). NMR spectroscopy has even been used in the investigation of the pathways of moisture drainage from softwoods (Almeida et al. 2008). Other than NMR, however, many nondestructive characterization techniques are unable to chemically identify the treatment material but rather detect changes in the dielectric or density profile of the wood (Bucur 2003). Additionally, the detection of treatment solutions may require staining to improve contrast, particularly with X-ray tomography. Moreover, the high-energy nature of X-ray and gamma-ray treatment renders them unsuitable probes for in situ characterization of the monomer movement investigated here as they can initiate polymerization.

Near-infrared spectroscopy (NIRS) allows for fast, nondestructive, cost-effective materials characterization with minimal sample preparation (Chen et al. 2010; Foley et al. 1998). For these reasons, NIRS is a favored analytical technique within many industries including agriculture (Morra et al. 1991), food (Cen and He 2007), textiles (Barton II et al. 2002) and biomedicine (Maki et al. 1995; Durduran et al. 2010). The use of the so-called therapeutic, or optical, window of wavelengths in the range of 600–1300 nm is found to be nondestructive for biological tissue and allows for a depth penetration in some cases of as much as 20–70 mm (Villringer and Chance 1997). NIRS has also been applied to the characterization of wood and pulps (Tsuchikawa and Schwanninger 2013; Hans et al. 2015), and more specifically, the analysis of transgenic trees (Yamada et al. 2006), the assessment of lignin in maritime pine (Alves et al. 2006) and quantification of linseed oil uptake (Eriksson et al. 2011; Geladi et al. 2014). Lande et al. (2010) also used NIRS correlated with thermogravimetric analysis to determine treatment levels in furfuryl-treated Scots pine wood flour, but the distribution of polymer through the treated samples was not characterized (Lande et al. 2010).

Time-resolved diffuse optical spectroscopy (TRS) using NIR lasers is a sophisticated NIRS technique providing spectra of a sample’s absorption and scattering, thus allowing for quantification of different chemical components within the material (Patterson et al. 1989; Cubeddu et al. 1999). TRS allows for more detailed nondestructive characterization, without the need for multivariate analysis, and has been applied as an oxygen monitor in human tissues (Suzuki 1999) and also for characterization of fruits (Nicolai et al. 2008; Cubeddu et al. 2001). TRS has previously been used in the characterization of water within wood (D’Andrea et al. 2009) and has been shown to quantitatively detect monomer (Farina et al. 2014). Here, TRS is applied to the characterization of monomer distribution and flow within wood, exploring the utility of this technique for the in situ characterization of timber during modification.

All spruce samples were cut from 2 m lengths of \(38\,\hbox {mm}\times 89\,\hbox {mm} \,\,(2''\times 4'')\) taken from the same pallet of kiln-dried wood. The selected 2 m lumber was largely defect-free, the grain roughly parallel to the axial direction and the grain spacing homogeneous. The \(10\,\hbox {mm} \times 10\,\hbox {mm} \times 100\,\hbox {mm}\) sample size was cut by hand from 200 mm lengths, which had been sectioned using a bench saw. Other than kiln-drying by the supplier, samples were not dried further prior to impregnation but rather allowed to reach the ambient moisture content in the lab (\(35.8 \, \pm \,5\%\) RH, measured over 50 days).

TRS can be used to nondestructively quantify monomer within a sample of treated spruce (Farina et al. 2014). Here, the technique was extended to characterize the distribution of monomer at specific positions along the longitudinal length of Mix-treated samples, as shown in Fig. 2. TRS measurements were taken at 10 mm intervals along the longitudinal length of 8 different samples post-impregnation. The position 10 mm from the longitudinal end of the sample is position 1, and subsequent positions are numbered accordingly. A step size of 10 mm was chosen as the lateral penetration sampled by the laser was found to be on the order of 5 mm within this sample type. Boundary effects were detected when measurements were taken closer than 5 mm to the edge of the sample.

All samples in Fig. 2, with the exception of the data shown in circles, display a similar profile with higher Mix concentration at the end of the sample relative to the center, a distribution typical for wood impregnation (Kolavali and Theliander 2013; Ahmed et al. 2013). There was a significant variation in measured Mix mass percentage between the different samples investigated. This variation in Mix mass percentage measured relative to the total mass of the sample is thought to be indicative of differences in permeability and microstructure between samples, a result of natural variation. The different distribution pattern of the data in circles, as shown in Fig. 2, compared with other samples is thought to be an indication of significantly increased permeability in this specific sample, perhaps arising from micro-cracks invisible by eye. Alternatively, variation in the angle of the grain relative to the samples’ longitudinal axis or changes in the sapwood content of the sample are possible explanations for the variation in Mix mass distribution. The scattering data of the sample shown in circles did display a different scattering coefficient when compared to the other samples tested, supporting the suggestion that changes in grain orientation could be the cause.

Polymer distribution was then investigated visually and microscopically to confirm correlation with Mix mass distribution as determined by TRS. Dye molecules were doped into Mix and the arrangement, post-polymerization, of dyed polymer within the wood visualized on the macroscopic level, as shown in Fig. 3a. It was evident that at the left-hand end of the sample, as shown in Fig. 3a, which was in direct contact with Mix, there was more polymer than toward the center. Figure 3b illustrates the co-localization of polymer and dye as the cells with polymer-filled lumen are also dyed pink and there was no evidence of pink areas without the presence of polymer. Scanning electron micrographs (SEM) taken from sections at the end and middle of the treated wood also indicated higher polymer content at the end of the sample where many lumen were filled compared to the center of the sample where empty cell lumen predominate. In both cases, destructive characterization of polymer distribution was in agreement with the monomer distribution prior to polymerization as determined by TRS, demonstrating the suitability of this technique for nondestructive assessment of Mix distribution. It also suggests that no significant movement of Mix occurred at the elevated temperatures during polymerization as the distribution of polymer resembles that of the monomer. The limited mass change before and after polymerization is in agreement with this conclusion.

TRS can be carried out at atmospheric pressure, under vacuum and when submerged as the optical fibers used were resistant to variations in pressure and chemical attack. This allowed for in situ TRS measurements of Mix concentration within the spruce samples during impregnation. In situ monitoring enabled a profile of Mix transport to be determined through the wood with both spatial and temporal resolution.

In order to minimize microstructure differences, and allow for meaningful comparison of rates of increase in Mix mass percentage measured relative to the total mass of the sample at different positions, a single sample was measured at four different positions during the impregnation process. The measurements indicated no significant increase in Mix mass percentage relative to the total mass of the sample within the wood during the applied vacuum, first 55–65 min, as shown in Fig. 4. This indicates that application of vacuum resulted in the removal of air from the wood and not Mix uptake, as illustrated schematically in Fig. 1, left. Once vacuum was removed, the Mix mass percentage within the wood rapidly increased at all positions, as shown in Fig. 4, indicating Mix infiltration of the wood.

Figure 4 shows Mix concentration measured at positions 1, 2, 3 and 4 along a single spruce sample measured during impregnation treatment. The rate of Mix mass percentage measured relative to the total mass of the sample increased fastest at position 1 and slowest at position 4, which is consistent with the final Mix distribution determined by TRS, as shown in Fig. 2. Initial results, as shown in Figure S1, measured at a selection of positions on different samples suggest that variation in microstructure between the wood samples had greater influence on the rate at which Mix was able to flow through the wood than the position at which measurements were taken. Observation of the positional dependance on the rate of Mix uptake was, therefore, only possible when comparing different positions along a single sample.

Under an applied external force, Darcy’s law is commonly used to describe flow through porous media. Examination of flow by Darcy’s law has found application in describing extraction of petrochemicals from geological formation, water movement through bedrock, and chromatography, though there are well-known regimes where experiments deviate from the model (Hlushkou and Tallarek 2006). It has also been extensively applied to the characterization of various types of flow within wood (Gronli and Melaaen 2000; Fredeen and Sage 1999; Perre and Turner 2001). Darcy’s law for liquids in its derivative form can be written as follows (Siau 1984):where \({\mathrm {d}V=v{_\mathrm{a}}A\mathrm {d}x}, K =\) specific permeability (m³ m⁻¹), \(x =\) penetration depth (m), \(\eta =\) dynamic viscosity (Pa s), \(A =\) cross-sectional area perpendicular to flow (m²), \(\Delta \hbox {P} =\) pressure differential (Pa) and \(v_\mathrm{a} =\) porosity or void volume fraction of wood. Once rearranged:and the integration performed:Equation 4 demonstrates that x, the penetration depth of the liquid, should have a \(\sqrt{t}\) dependance. The TRS experiments performed did not, however, measure the extent of permeation with time but rather variation in Mix mass percentage at a certain location with time. The plots derived from these experiments are plots of density-weighted Mix percentage retention against time. Based on calculations shown by Siau assuming total filling of the voids from the longitudinal end of the sample to the liquid–gas interface (Siau 1984):where \(F_\mathrm{vl}\) is the fractional volumetric retention and L is the longitudinal length of the sample. Substituting the equation for x from Darcy’s law then gives:According to Eq. 6, \(F_\mathrm{vl}\) over the whole sample has a \(\sqrt{t}\) dependance. Figure 5 confirms a broadly linear relationship with \(\sqrt{t}\) at each position measured. The gradient of the Mix mass percentage versus \(\sqrt{t}\) plot, however, varies with the position at which the data were measured. Position 1 has the steepest gradient and position 4 the shallowest, as shown in Fig. 5. This suggests that Darcy’s law does not accurately describe the flow of monomer through the spruce wood investigated as the model cannot account for the position dependance of monomer accumulation. Moreover, the penetration of the sample with Mix is not proportional to \(\sqrt{t}\) as predicted by Eq. 4, as shown in Fig. 4, nor does it proceed as a single liquid–gas interface. As shown in Fig. 4, Mix is detected by TRS measurements at position 4 after only a short period of time, and Mix continues to accumulate at all positions throughout the duration of the measurement period. This behavior is not in accordance with Darcy’s law and suggests that multiple gas–liquid interfaces are moving at different rates through the wood, rather than a single front.

An apparent decreasing sample permeability with increasing length has previously been observed and modeled by Bramhall (1971) who found that for some less permeable wood species, such as spruce, axial impregnation was not well described by Darcy’s law. In this model, it was proposed that the random blockage of tracheids by pit aspiration, insoluble extractives, etc. leads to the exponential decrease in the number of conductive tracheids with increasing penetration depth (Bramhall 1971). This behavior then leads to a decrease in the apparent permeability of the wood with increasing sample length and incomplete filling of voids within the sample.

The assumption of total filling of the voids made during data analysis using Darcy’s law was not valid for the samples under consideration, as SEM analysis clearly shows voids within the impregnated and polymerized samples, as shown in Fig. 6. Mix is thought to shrink upon polymerization and, therefore, could contribute to some of the voids shown in Fig. 6; however, this cannot explain the deviation in monomer flow from Darcy’s law. Bramhall’s model, and specifically incorporation of incomplete filling into the model, was thus investigated. Bramhall’s modified version of Darcy’s law is given by:where \(K_{0}\) is the permeability at zero length and b is the exponential coefficient describing the decrease in number of conducting fiber tracheids with increasing length. The related fractional volumetric retention accounting for penetration from both ends of the sample as shown by Siau (1984) is given by:where c is the combination of constants describing the relationship between x and \(\sqrt{t}\), as given in Eq. 4. This indicates that \(F_\mathrm{vl}\) over the whole sample has an exponential dependance on \(\sqrt{t}\) rather than the linear dependance predicted by Darcy’s law. Summation of Mix mass percentage over all points measured during a certain time period can give an indication of the \(F_\mathrm{vl}\) in the half of the sample measured by TRS. If the sample is assumed to be symmetric, then doubling the sum of Mix mass percentage measured in one half of the sample allows for an estimation of the total Mix mass percentage in the whole sample and comparison to Bramhall’s model. It must be noted that this method to estimate the total Mix mass percentage in the whole sample does not account for the Mix uptake in the 5 mm at each end of the sample nor the central 10 mm as these areas where not measured by TRS; however, it is assumed that this does not significantly effect the overall profile of the plot. When plotted against \(\sqrt{t}\) the sum of Mix mass percentage, doubled to estimate the mass in the whole sample, shows a relatively linear profile, except for at short times, as shown in Fig. 7. This was then fitted according to equations in the form of Bramhall’s and Darcy’s models.Equation 9 describes the linear dependance of \(F_\mathrm{vl}\) on \(\sqrt{t}\) as predicted by Darcy’s law. Bramhall’s model in contrast predicts \(F_\mathrm{vl}\) dependance on \(\sqrt{t}\) to have the form of Eq. 10.The results shown in Fig. 7 demonstrate that the mass increase over the whole sample is better described by Bramhall’s model than by Darcy’s as there is better agreement between Eq. 10 and the experimental data than with the linear fit according to Eq. 9. Moreover, the decreasing number of conducting tracheids with increasing sample length as predicted by Bramhall can explain the decreasing rate of monomer accumulation when moving from position 1 to position 4 as the effective permeability at position 4 is decreased compared to position 1.

Comparing Eq. 10 with Eq. 8, the constants A and B in Eq. 10 should both be positive; however, the fitting shown in Fig. 7 arising from Eq. 10 results in small but negative values for A and B (−0.61 and \(-9.81 \times 10^{-3}\) , respectively). This suggests that b the exponential coefficient describing the decrease in number of conducting tracheids with increasing length in Eq. 8 must also be negative. Bramhall’s model, however, relies on a positive value of b. Thus, faster measurements in the time immediately after the vacuum is released and also measurements at much longer time periods would be required to more conclusively determine whether the deviation from Darcy’s law shown in Fig. 7 is real and whether Bramhall’s model fitted over a longer time period would result in a positive value of b. With the data shown, relatively small errors in the first five data points after the vacuum was released would have a significant influence on the fitting of the data.

To further examine the utility of the TRS technique for monomer treatment characterization and demonstrate that the technique is not limited to Mix alone, TRS was used to monitor the impregnation of spruce with the each of the individual monomers MMA, GMA and EDMA separately, as shown in Fig. 8. Impregnation of porous media is generally viscosity dependant as depicted by Darcy’s law and other models of fluid dynamics. The rate of monomer accumulation should also display a viscosity dependance.

MMA and GMA treatments were performed separately on two end-matched samples. The natural permeability of the sections was, therefore, assumed to be comparable. The slower increase in GMA concentration with time when compared to MMA suggests that the more viscous GMA moved more slowly within the wood in accordance with the expected behavior. Additionally, EDMA which has a higher viscosity again moved even slower. It must be noted that EDMA treatment was carried out on a separate section of spruce; therefore, variation in sample permeability will have an effect. The movement of EDMA was slower than that seen for any of the Mix-treated samples, which does suggest that the slower kinetics are viscosity related rather than due to different microstructure. In each case, the distribution of monomer through the samples displayed the same profile as the Mix-treated samples, as shown in Fig. 8b, indicating that although the kinetics are slowed by increasing viscosity, the distribution profile was unaffected.

TRS allowed for nondestructive, quantitative analysis of monomer distribution within impregnated wood. The distribution profile determined by TRS was consistent with previous literature and destructive characterization of polymer distribution. This indicates great potential for the use of TRS in impregnated wood composite materials as a quantitative method for nondestructive characterization of treatment distribution in the field.

Moreover, TRS was able to demonstrate that the impregnation of even relatively small spruce samples with monomer cannot be accurately described by Darcy’s law as it does not describe accurately the local changes in Mix concentration. Bramhall’s description of fluid flow within wood was also found to be an inadequate model for the monomer accumulation data as measured by TRS. Although equations in the form of Bramhall’s model provided a better fit of the experimental data, the results of the fitting did not lead to physically meaningful results. Additionally, spruce impregnation with monomer of varying viscosity monitored by TRS displayed decreasing monomer flow kinetics with increasing viscosity, again as predicted by fluid mechanics.

Taken together these results demonstrate that TRS is a powerful, nondestructive characterization technique for monomer treated wood. Moreover, TRS allows for post-impregnation and in situ measurements of treatment accumulation providing access to data on the kinetics and flow of monomer through wood.