Wettability and swelling of acetylated and furfurylated wood analyzed by multicycle Wilhelmy plate method

Introduction

In outdoor applications, wood is sensitive to UV irradiation and has limited dimensional stability (Rowell 2010). Wood with high moisture content is prone to decay by fungi (Rowell 2010; Fredriksson and Lindgren 2013). Wood modification is aimed at improving decay resistance, dimensional stability and weathering performance, and at reducing water sorption (Hill 2006). One promising approach in this regard is treatment with chemicals, which form chemical bonds to the cell wall polymers via their hydroxyl groups. Such chemical reactions may lead to critical alternation of the chemical and microstructural characteristics of the modified wood. A substitution of hydrophilic hydroxyl groups by hydrophobic reagents reduces the wettability and moisture sorption of the wood and results in better dimensional stability under varying moisture conditions (Rowell et al. 1994). Acetylation via formation of ester bonds between acetic anhydride and wood OH groups (Rowell et al. 1985, 1994), cross-linking of cellulose molecules with formaldehyde (Stevens et al. 1979) and furfurylation (Schneider 1995; Lande et al. 2008a,b) have been developed. Ester bonds form between acetic anhydride and hydroxyl groups in the wood cell walls during acetylation (Rowell et al. 1994). In the course of furfurylation, furfuryl alcohol polymerizes inside the cell wall, but there are some uncertainties about the existence of chemical bonds between poly(furfuryl alcohol) and the cell wall components. Based on NMR studies with lignin model compound, formation of chemical bonds between lignin and furfuryl alcohol seems to be possible (Lande et al. 2004; Nordstierna et al. 2008).

Liquid uptake in wood is controlled by the chemistry and morphology of the wood tissue (Sedighi-Gilani et al. 2012), although this process is heavily influenced by the anisotropy and porosity of the wood. Because wood is a porous material and has a rough surface, capillary actions are involved in the wetting phenomena (Lande et al. 2008a,b). Wetting properties are also different in various tissues of the same wood. Latewood is more permeable for capillary liquid uptake than earlywood, and it is thus the supposed main pathway of liquid transport (Sedighi-Gilani et al. 2012). Water as swelling liquid is adsorbed within the cell wall and leads to dimensional changes, whereas non-swelling liquids, such as octane, only fill the pores and the voids. This difference can be used to distinguish between the simple capillary transport and swelling-inducing part of the liquid uptake. Although the wettability and dimensional stability of acetylated and furfurylated wood is well investigated (Stamm 1977; Ramsden et al. 1997; Baysal et al. 2004; Obataya et al. 2007; Lande et al. 2008a,b; Bryne and Wålinder 2010; Thygesen et al. 2010; Wålinder et al. 2013), little is known about the dynamic wetting behaviour and dimensional stability of chemically modified wood.

The present work aimed to provide more details on the wetting, sorption and dimensional stability of modified wood by the Wilhelmy plate method. We focused on the dynamic contact angle (CA) and dynamic wetting behaviour of wood following the methodology of Sedighi Moghaddam et al. (2013), who developed the multicycle Wilhelmy plate method to study these dynamic properties. In this approach, the sample is immersed in and withdrawn from a liquid several times. Complete withdrawal of the veneer from the probe liquid in each cycle allows the determination of liquid uptake and monitoring of the corresponding dimensional changes in the sample due to swelling. We report the dynamic wetting and swelling of the modified wood samples, which had been acetylated and furfurylated in a wide range of conditions. The results are discussed in view of capillary liquid uptake and swelling rate of the modified wood samples.

Materials and methods

Southern yellow pine (SYP) (Pinus palustris) boards (40×14×2.5 cm³) (L×T×R) were acetylated in a pilot plant at SP Wood Technology in Borås, Sweden, in a microwave-heated reaction vessel according to Rowell et al. (1985) to 15.9% (SYP_acet15.9) and 22.2% (SYP_acet22.2) acetyl contents. Two unmodified matched control samples were prepared from the same boards (SYP_ctrl15.9 and SYP_ctrl22.2, respectively).

Furfurylated samples were prepared in an autoclave by a vacuum and pressure process, and subsequently cured and dried in a vacuum drying kiln according to Lande et al. (2004) at the Kebony’s pilot plant in Porsgrunn, Norway. The weight percentage gain (WPG) of the furfurylated samples of maple (Acer platanoides) (Maple_furf32) and beech (Fagus sylvatica L.) (Beech_furf22) were ca. 32% and ca. 22%, respectively. Unmodified maple (Maple_ctrl) and beech (Beech_ctrl) were collected from the same board. Furfurylation of the SYP samples (SYP_furf28 and SYP_furf45) was performed to WPG levels of ca. 28% and 45%, respectively.

To prevent end-grain sorption, a cross section of the shorter wood blocks (modified and unmodified) was sealed by a polyurethane lacquer. The wood blocks were cut using a veneer saw to smaller specimens from the same annual rings with a dimension of ca. 30×7×50 mm (L×R×T). Veneers with a dimension of 30×7×1 mm³ (L×R×T) were obtained by splitting these smaller samples with a chisel along the direction of the fibre and perpendicular to the annual rings (see also Sedighi Moghaddam et al. 2013 and Sedighi Moghaddam et al. 2014). The control specimens were prepared analogously from the same annual rings. The SYP_ctrl15.9 veneers had 3–4 annual rings, whereas the SYP_ctrl22.2 veneers in most cases had only one annual ring. The width of growth rings ranged from 0.4 to 1.3 mm for SYP_ctrl15.9 and from 1 to 1.8 mm for SYP_ctrl22.2. The latewood portion was 61±9% and 37±9% for SYP_ctrl15.9 and SYP_ctrl22.2, respectively.

Freshly cut veneers were dried and thermally treated at 104°C for 1 h in an oven. The water contact angles (CA_water) were determined directly after cutting in a series of samples. Before drying, the unmodified and modified veneers had different moisture contents (MC), ranging from 1.5% to 9% (based on oven-dry weight) (Table 1).

Multicycle Wilhelmy plate method was performed on a Sigma 70 tensiometer (KSV Instruments, Helsinki, Finland) in four replicates. Veneers were immersed for 20 cycles in ultrapure water (resistivity >18 MΩ cm) and for 10 cycles in n-octane (≥99%; Alfa Aesar, Karlsruhe, Germany) as swelling and non-swelling liquids, respectively, under a test velocity of 12 mm min^-1. The values for surface tension measured in the present study and for surface energy components of water and n-octane found in the literature are presented in Table 1. Measurements were performed at 22°C–23°C and 35%±5% RH.

The basic principle of the multicycle Wilhelmy plate method is that a plate is immersed into and withdrawn from a probe liquid repeatedly for several cycles (Sedighi Moghaddam et al. 2013). Figure 1 shows the first four cycles of a typical multicycle experiment in water, and illustrates the quantitative methods applied. The Wilhelmy equation for porous and hygroscopic materials such as wood is as follows (Sedighi Moghaddam et al. 2014):

Figure 1:

A typical multicycle Wilhelmy plate test on wood veneer in water. The curves demonstrate the force (F) as a function of immersion depth (h) during a 4-cycle measurement. Receding force (F_R), advancing force (F_A) and final force (F_f) are shown for the first cycle.

where F(h,t) is the detected force, m_i is the initial mass of the plate (veneer), g is the gravitational constant, P is the wetted perimeter of the plate, γ is the surface tension of the liquid, θ is the liquid-solid-air contact angle, ρ is the density of the probe liquid, A is the cross-sectional area of the plate, h is the immersion depth and F_w(t) is the force due to wicking and sorption of the liquid and vapour at time t. To quantify the effect of vapour sorption in this study, some measurements were done by leaving the veneer just above the water surface and the force was recorded as a function of time. The mass uptake due to vapour sorption for 50 min, which is the time of a 20 cycle measurement, corresponds to about 4% of the liquid uptake during a 20 cycle measurement. For the first cycle, the mass uptake due to vapour sorption amounts to lower than 2% of the mass change in the first cycle. To determine the veneer perimeter (P), a single cycle immersion in a wetting-out liquid (n-octane) can be done (for details, see Sedighi Moghaddam et al. 2014). Prior to measurement, the balance is set to zero, and the force due to the initial mass of the veneer is subtracted from the detected force and the instrument records the value of (F-m_ig). From such measurements and some supplementary tests, several properties can be determined. First, advancing and receding contact angles are obtained by linear regression of the advancing and receding curves to zero depth for each cycle (Sedighi Moghaddam et al. 2014) (Figure 1). Second, sorption and swelling are determined by linear regression of the final force, F_f (which remains after each cycle due to sorption), to zero depth for each cycle (Figure 1). The percentage of change in liquid mass uptake at cycle n, l_n, of the wood veneer after cycle n is calculated as:

where F_f,n is the final force after cycle n.

Octane as a non-swelling liquid fills only the voids of the wood structure, whereas water also enters into the cell wall by diffusion thereby causing swelling. The swelling part of the water uptake can be calculated by subtracting the octane uptake from the water uptake:

where l_w,20 is the water uptake at cycle 20; l_oct,10 is the final octane uptake at cycle 10; and ρ_w and ρ_oct are the density of water and octane, respectively. In octane experiments, it was observed that the specimens were close to full saturation after 10 cycles (Figure 2). Thus, measurements with octane were limited to 10 cycles.

Figure 2:

Water and octane uptake of the acetylated and furfurylated samples as a function of cycle number.

In evaluating the modification efficiency in terms of reduced liquid uptake, the liquid uptake reduction (LUR) is defined as:

where l_n,c and l_n,t are the liquid mass uptake levels at cycle n of the untreated (control) and treated wood samples, respectively. This parameter can be calculated for both water (n=20) and octane (n=10).

The veneer perimeter was measured before and directly after the multicycle treatment with n-octane immersion. It was assumed that the dimension of the sample does not change within 3 min, which is the time between retrieving from water and immersing in octane (Son and Gardner 2004). The dynamic perimeter change was calculated according to Equation 5 (Sedighi Moghaddam et al. 2013):

where P_n and P_n-1 are the veneer perimeter after cycle n and n-1, respectively; F_f,n and F_f,n-1 are the final force for cycle n and n-1, respectively; ΔP is the total perimeter change during the measurement; and ΔF_f is the difference in the final force between the last and first cycles. Compared with that of the previous one (Sedighi Moghaddam et al. 2013), this model was modified by assuming that void filling completely dominates the sorption process during the first cycle, whereas swelling is prevalent during the following cycles. Figure 3 shows the results of the two models. The final perimeter change (Δp_n) can also be calculated as follows:

Figure 3:

Comparison of apparent perimeter changes calculated from the previous model (▉) (Sedighi Moghaddam et al. 2013) and the new model (○) developed in this paper.

where P₀ is the initial veneer perimeter measured by octane immersion.

In the present work, we defined the perimeter change reduction (PCR) to quantify the modification efficiency in terms of dimensional stability as:

where Δp_n,c and Δp_n,t are the final perimeter changes of the untreated and treated samples, respectively. PCR is similar to the anti-shrinkage efficiency (ASE), which is a common parameter for evaluating the dimensional stability of wood samples, but it differs in that the PCR is evaluated before the wood samples are fully saturated with liquid.

The estimation of the surface tension of the water before and after performing a multicycle Wilhelmy test reflects the migration of the extractives to the air-water interface. This estimate is quantified by measuring (Δγ_meas) and calculating the apparent surface tension reduction (Δγ_app) (Sedighi Moghaddam et al. 2013):

γ_app is calculated as follows:

where γ is the surface tension of pure water; γ_f is the water surface tension after performing the multicycle Wilhelmy experiment; and F_R and F_f are the receding and final forces of the last cycle, respectively. It is worth mentioning that the apparent surface tension reduction is a manifestation of the accumulation of surface active components, mainly extractives, and that no additional surface tension measurement is required.

Results and discussion

Sorption and dimensional stability

The liquid sorption of wood was greatly affected by acetylation and furfurylation (Figures 2 and 4a). The data were obtained by extrapolating the final forces (F_f) to zero depth (Equation 2). The shape of all water uptake curves is rather similar, and they can be separated into two regions that are dominated by different transport mechanisms, i.e., a relatively fast capillary-driven transport of liquid water (the first 2–3 cycles) and a slower water uptake process associated with swelling of the cell walls. It is worth noting that the first fast-transport process started with the liquid spreading on the wood surface (decreasing CA to 0°) followed by capillary transport (void filling). The extent of water uptake due to swelling is given by Δl₂₀ (Equation 3) (Table 3), and is discussed below.

Figure 4:

(a) Liquid mass uptake and (b) apparent perimeter change of furfurylated SYP at two modification levels as a function of cycle number: weight percent gain (WPG)=28% and 45%.

Table 3:

Sorption and dimensional stability results extracted from multicycle Wilhelmy plate measurements for the acetylated and furfurylated samples.

Sample name	Water uptake, Equation 2 (Ff20) (%)	LUR (water) (%)	Octane uptake Equation 2 (Ff10) (%)	LUR (octane) (%)	Δp20 Equation 6 (%)	PCR (%)	Δl20 Equation 3 (%)
SYPacet15.9	34.1±1.6	37.2	21.0±1.5	25.0	3.7±0.5	69.7	13.1±2.2
SYPctrl15.9	54.3±1.5	–	28.0±2.9	–	12.2±5.6	–	26.3±3.3
SYPacet22.2	41.9±3.7	40.7	34.8±6.2	4.1	2.5±0.3	74.7	7.1±7.2
SYPctrl22.2	70.6±4.2	–	36.3±3.2	–	9.9±1.2	–	34.3±5.3
Maplefurf32	13.9±2.4	76.5	9.6±1.4	68.0	1.9±1.2	70.8	4.3±2.3
Maplectrl	59.2±3.4	–	30.0±2.6	–	6.5±3.3	–	29.2±4.3
Beechfurf22	14.9±7.8	70.2	10.3±1.8	63.2	2.0±1.8	75.6	4.6±8.0
Beechctrl	50.0±7.7	–	28.0±0.7	–	8.2±3.4	–	22.0±7.7
SYPfurf28	21.3±4.5	60.8	11.8±0.6	57.9	7.0±4.1	42.6	9.5±4.5
SYPfurf45	12.7±1.5	76.6	8.5±0.7	69.9	0.8±0.7	93.4	4.2±1.7

LUR, Liquid Uptake Reduction as defined by Equation 4.

PCR, Perimeter Change Reduction as defined by Equation 7.

The acetylated samples showed similar (SYP_acet22.2) or slightly lower (SYP_acet15.9) octane uptake than the control samples (Figure 2 and Table 3). In contrast, the water uptake decreased significantly after acetylation. Acetylation swells the cell wall and the whole dimension of wood, i.e., the porosity should remain fairly unchanged, but in high-level furfurylated wood, the furan polymer fills some voids and decreases the porosity and hence the amount of capillary liquid in wood. Thus, our data can be interpreted to mean that in acetylated wood less hydroxyl groups are accessible, resulting in strongly decreased swelling, whereas the effect on the capillary uptake (void filling) is minor.

The acetylated and unmodified SYP samples with higher earlywood proportion (SYP_acet22.2 and SYP_ctrl22.2) showed higher liquid (especially water) uptake than those with less earlywood (SYP_acet15.9 and SYP_ctrl15.9) (Figure 2). This result is consistent with the finding of Wålinder et al. (2013), who reported that latewood in acetylated SYP had a significant lower liquid uptake than the corresponding earlywood in this type of liquid immersion experiments. According to Sedighi-Gilani et al. (2012), LW is the preferred pathway in the process of liquid uptake in the L and T directions, whereas the latewood region of the growth rings acts as a boundary layer that slows down liquid transport in the R direction. Sedighi-Gilani et al. (2012) reported that latewood exhibited a faster liquid uptake than earlywood. Thus, in the present study, we have demonstrated that the total amount of absorbable liquid in latewood is lower.

The reduced water uptake of furfurylated wood is also obvious (Figures 2 and 4a, Table 3). For instance, the final water uptake of Maple_furf32 and its Δl₂₀ decreased by 77% and 85%, respectively, compared with that of Maple_ctrl. The main reason for this decrease is that poly(furfuryl alcohol) is fixed in cell lumens and large pores of the cell walls, thereby causing a permanently swollen cell wall (Lande et al. 2008a,b) and restricting the access for water uptake. However, in contrast to acetylation, furfurylation also significantly reduced the octane uptake as manifested in porosity decrement.

The SYP_acet15.9 sample showed a liquid uptake reduction (LUR) of 37% (water) and 25% (octane) compared with that of the control (Table 3). The corresponding LUR values for SYP_furf45 were significantly larger (77% for water and 70% for octane). The LUR decreases with decreasing degree of furfurylation because of the favoured cross-linking reactions, resulting in a highly branched and cross-linked polymer (Lande et al. 2008a,b; Wålinder et al. 2009). Moreover, the results presented herein confirm the filling of voids by the cross-linked poly(furfuryl alcohol) and restricting of water uptake. In summary, higher LUR data can be achieved at higher levels of furfurylation for the SYP samples than for the other samples investigated in the present study.

Figure 3 is a comparison of the recent and former parameters before model modification (Sedighi Moghaddam et al. 2013; note that in the new model there is, by definition, no perimeter change in the first cycle). In Table 3 and Figures 4b and 5, the results are presented in terms of perimeter change. The results in Figures 4b and 5 were obtained using Equation 5, with the exception of the initial and final perimeters that were obtained from octane immersion. Pronounced differences can be seen concerning the dimensional stability between the unmodified and modified wood. The perimeter change reduction (PCR) in modified wood is due to the formation of chemical bonds between the modifying chemicals and the wood cell wall that makes it less prone to swelling. For example, the acetylated samples SYP_acet15.9 and SYP_acet22.2 exhibited 70% and 75% PCR, respectively, compared with the corresponding untreated samples (see Table 3), whereas the PCR for the furfurylated samples varied from 43% for SYP_furf28 to 93% for SYP_furf45. Accordingly, furfurylation at high level (WPG 45%) led to higher dimensional stability than acetylation, which correlates with lower water uptake (Table 3). This observation is in line with that of other studies. Lande et al. (2008a,b) found that at WPG 32 and 47 the ASE is close to 50% and 70%, respectively. Baysal et al. (2004) furfurylated Japanese cedar and Scots pine wood in the presence of different catalysts to various WPG degrees, and reported ASE values ranging from 70% to 88%, as also seen in Table 3. Epmeier et al. (2004) pointed out that ASE does not improve above a certain level of furfurylation.

Correction added after online publication 11 September 2015: The originally published value of the last SYP_furf in the fifth sentence of the previous paragraph was updated from SYP_furf28 to SYP_furf45.

Figure 5:

Apparent perimeter change of (a) acetylated SYP and (b) furfurylated beech and maple during multicycle Wilhelmy experiments. All data were calculated using the perimeter model (Equation 6) with the exception of the final perimeter (P20), which was determined by a single immersion in octane after 20 cycles in water.

Dissolution of extractives

The measured (Δγ_meas) and apparent (Δγ_app) surface tension reduction after 20 cycles were evaluated (Figure 6). The unmodified samples generally showed more reduction except for SYP_furf28, for which the surface tension reduction was relatively high compared with that of the other modified samples. One reason for this observation could be the removal of a significant fraction of extractives during the modification process. It is also possible that extractives are also modified and become less mobile in the hydrophobic and thicker cell walls. Blocking of voids may also play a role in this context.

Figure 6:

Surface tension reduction of the air-water interface after multicycle Wilhelmy experiments with different modified wood veneers.

Conclusions

A dynamic wettability technique based on a multicycle Wilhelmy plate method was developed to investigate acetylated and furfurylated wood veneers in terms of liquid penetration, sorption and dimensional stability. The results showed lower CA for freshly cut veneers than for thermally treated ones. It was demonstrated that acetylation had made the wood surface more hydrophobic (for fresh cut veneers), whereas no significant change in CA was observed for the furfurylated samples. Based on the results of wettability tests, the furfurylated SYP with a WPG of 45% had lower water and octane uptake, lower swelling and higher dimensional stability than the other samples. Moreover, the acetylated samples with a higher proportion of latewood had lower liquid uptake and swelling than those with a lower proportion. Combining the results of water and octane uptake with multicycle Wilhelmy plate measurements makes it possible to distinguish between capillary liquid uptake and swelling. Acetylation mainly reduces swelling by water, whereas furfurylation reduces both capillary liquid uptake and swelling.