Introduction

The construction sector is responsible for a substantial part of global carbon dioxide emissions. Building with wood can lower the carbon footprint of the building sector as well as store carbon in the built environment (Churkina et al. 2020; Mishra et al. 2022). In moist outdoor environments, long-term storage of carbon in wood is challenging because of fungal degradation of the material. A modern way of ensuring a long service life in such environments is wood modification, which does not have the adverse environmental effects of traditional biocide impregnation (Sandberg et al. 2017; Zelinka 2022). One such modification is acetylation which has been widely used in research and is commercially available. In this process, wood is impregnated with acetic anhydride and cured at elevated temperature. Acetylation decreases the cell wall moisture content (Beck et al. 2018b; Čermák 2022; Digaitis et al. 2021; Himmel and Mai 2015; Passarini et al. 2017), affects the interaction between capillary water and the wood (Beck et al. 2018b; Digaitis et al. 2021) and substantially increases the durability (Beck et al. 2018a; Mantanis et al. 2020; Mohebby and Militz 2010). The underlying mechanism of the increased durability for modified wood is not resolved (Ringman et al. 2019; Zelinka et al. 2020), but an important factor is thought to be changes in wood–water interactions (Brischke and Alfredsen 2020; Thybring 2013). Water is an essential requirement for fungal degradation, although the exact role of water is not yet fully understood (Brischke and Alfredsen 2020). Fungal hyphae need the presence of capillary water in wood (Rayner and Boddy 1988), but sufficient space for water within cell walls also appear to be a requirement (Thybring 2013). The latter possibly relates to that a threshold cell wall moisture content should be met to enable diffusion of the low molecular weight substances that are excreted by the fungi (Zelinka et al. 2015). It is thus likely, that the fungal degradation of wood depends not only the absolute value of the moisture content, but also the location of this water within the wood structure.

Further studies of the role of location of water on fungal degradation require that it is possible to control where in the wood structure water is present. A way to achieve this is by controlling the spatial distribution of a wood modification known to affect the moisture sorption properties. Such a targeted modification strategy would enable to create specimens with a range of wood–water interactions, not only in terms of moisture content, but also in terms of location of water (within or outside of the cell wall) and state of water (interaction with the wood material). In a previous paper, we showed that the wood–water interactions can be altered in different parts of the wood structure by targeted acetylation (Digaitis et al. 2021). Protocols were developed to achieve both a uniform acetylation of the whole cell wall and an interface acetylation acting only at the cell wall-lumen interface. The uniform acetylation changed the interaction between the wood and the cell wall water as well as the capillary water, while acetylation of only the cell wall-lumen interface changed the interaction with capillary water without significantly affecting water within cell walls. These results were obtained using water saturated samples and in the hygroscopic moisture range. The water-saturated state is, however, not a realistic moisture state for fungal degradation due to lack of oxygen, and fungal degradation does generally not occur in the hygroscopic moisture range. Typically, the lower limit of wood degrading fungi is around − 44 bar (Boddy 1983; Griffin 1977) which corresponds to a water potential of about − 4400 J/kg and a relative humidity of 96.8%. Controlling the moisture states at such high RH is more challenging than in the hygroscopic moisture range, but precise control can be achieved by pressure plate instrumentation (Fredriksson 2019; Thybring and Fredriksson 2023). Therefore, by combining a targeted modification approach with precise conditioning at moisture states relevant for fungal degradation using the pressure plate technique, it could thus be possible to achieve a range of wood–water interactions in terms of not only amount of water, but also location and state of water.

The purpose of the present study was to investigate the effect of targeted acetylation on capillary water and cell wall water, respectively, at high non-saturated moisture states relevant for fungal degradation. The study was performed by combining pressure plate conditioning with low field nuclear magnetic resonance (LFNMR) and X-ray computed tomography. LFNMR enabled separation of water in different locations within the wood structure while X-ray computed tomography was used to visualise location of capillary water in selected samples.

Materials and methods

Modification procedure

Mature sapwood of Norway spruce (Picea abies L. Karst.) from an experimental forest in Vindeln, Sweden was used. The average density was 410 kg/m3, for more details see Fredriksson et al. (2016). Specimens of 5 × 5 × 10 mm3 were cut, where 10 mm was in the longitudinal direction. Targeted acetylation of these specimens was then performed using two modification strategies: one that uniformly modified the cell wall, and one that modified the interface between the cell lumina and the cell wall. This targeted acetylation approach was achieved by tuning the reaction conditions as described in Digaitis et al. (2021):

  1. 1.

    Uniform acetylation was achieved by using a 1:4 (v/v) mixture of acetic anhydride (99.3%, VWR Chemicals, Radnor, PA, USA) and pyridine (Sigma Aldrich, Merck KGaA, Darmstadt, Germany). The reaction was carried out at 80 °C and three different reaction times (20 min, 1 h and 3 h) were used to achieve three degrees of acetylation.

  2. 2.

    Interface acetylation was achieved by using pure acetic anhydride. The reaction was carried out at 75 °C and two reaction times (5 h and 24 h) were used to achieve 2° of interface acetylation.

In addition, control specimens were heated in pure pyridine at 80 °C for 3 h. Further details on the modification procedure are found in Digaitis et al. (2021). The relative mass changes obtained by the different modifications are shown in Table 1.

Table 1 The average change in mass, Rmod, after the different modifications
Full size table

LFNMR to separate water in different locations

Moisture conditioning by pressure plate

The moisture conditioning was performed in the custom-built pressure plate system at Lund University described by Fredriksson and Thybring (2019). Forty-eight specimens of each modification type and untreated controls, as well as 16 pyridine controls were included in the study and conditioned at eight moisture levels in both desorption and absorption. A total of 304 specimens was thus included in the study.

Initially, half of the specimens were vacuum water-saturated with deionised water and half of the specimens were dried at 60 °C under vacuum. A total of 16 ceramic plates for the pressure plate system were also vacuum water-saturated with deionised water and placed in the pressure plate cells. Three water-saturated specimens of each modification type as well as one pyridine reference were then placed in each of the eight desorption cells. Similarly, three dried specimens of each modification type and one pyridine reference were placed in each of the eight absorption cells. The specimens in the absorption cells were placed on a stainless steel grid which was elevated from the ceramic plate. This was done to avoid wetting of the dry specimen with liquid water by touching the water saturated plate.

Eight different pressure levels were applied across the pressure plate cells. At each pressure level, one absorption and one desorption cell were connected to the same pressure regulator to ensure the same pressure in both cells. When the specimens had been placed in the cells and these had been closed, the pressures were applied and the peristaltic pumps enabling water supply for the absorption cells were started. The applied pressure and the corresponding RH levels and water potentials are shown in Table 2.

Table 2 Pressure levels used in the pressure plate conditioning system and the corresponding relative humidity (RH) and water potential, Ψ
Full size table

After 52 days, the conditioning of the first specimens were ended, and they were transferred to LFNMR-tubes. To prevent drying of the specimens during handling, this transfer was made using the following procedure. One pressure plate cell was detached from the pressure regulator and because of the quick connect used, the applied pressure in the cell remained; see Fredriksson and Thybring (2019) for details on the design of the system. The cell was placed inside a glove box which was connected to a moisture generator (2500 Humidity Generator, Thunder Scientific Corporation, Albuquerque, New Mexico, USA) that generated a humidity > 97% RH. In addition, moist cloths were placed in the glove box to keep the RH as high as possible. In this way, the difference in relative humidity between the air in the glove box and the pressure plate cell was small which would limit drying of the specimens when opening the cell. Pre-weighed LFNMR-tubes as well as Teflon sticks to seal the tubes were already in the glove box. The pressure in the cell was then released, one specimen was transferred to one of the LFNMR-tubes and sealed with a Teflon stick. This procedure was then repeated for all specimens in that pressure plate cell. The mass of all sealed tubes with the specimens inside was then determined. The LFNMR-tubes with the specimens were transported in an insulated box to the University of Copenhagen for LFNMR measurements. This procedure was repeated for all pressure plate cells and took about 3 weeks to finish.

LFNMR measurements and evaluation

The sealed tubes with a specimen inside was placed one at a time in an LFNMR instrument (Bruker mq20-Minispec, Bruker, Billerica, MA, USA) with a 0.47 Tesla permanent magnet. The probe temperature of the instrument was set to 21 ± 0.5 °C and was controlled by a water-cooling system. Before starting the measurement, the specimen was kept in the instrument for 5 min to equilibrate to the instrument temperature. The spin–spin relaxation (T2) was then measured using the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence with a pulse separation of 0.1 ms, 16,000 echoes, 32 scans and a recycle delay of 30 s. The gain was set individually for each specimen and ranged between 84 and 102 dB.

The T2 relaxation curve was evaluated for each specimen by multi-exponential decay analysis which fits a large number of exponential functions (Istratov and Vyvenko 1999). Hereby, characteristic times (T2 relaxation times) and pre-exponential coefficients are obtained that describe the specific relaxation curve. In this study, 200 exponentials were fitted to the data with fixed, logarithmically spaced T2 relaxation times in the range from the first to last time point (0.2 ms to 3.3 s). This provided a spectrum of T2 relaxation times with clusters of adjacent non-zero pre-exponential coefficients (peaks) interrupted by clusters of coefficients of zero. The amount of information that can be extracted from LFNMR data by multi-exponential decay analysis is, however, limited by the noise of the data. Thus, the method cannot resolve peaks that are spaced closer than a resolution limit defined by the signal-to-noise ratio (Bertero et al. 1982). Therefore, the analysis method reduced the obtained peaks to a limited number (range 2–6) of significant exponentials describing each T2 relaxation curve. Each exponential is defined by one T2 relaxation time for each resolvable peak and a weight calculated as the sum of non-zero pre-exponential coefficients describing this peak. Further details about the analysis are given in (Digaitis et al. 2021).

The influence of modification and moisture conditions on the location of water within the wood structure was evaluated from the determined T2 relaxation times. The T2 is related to both chemical environment and physical environment (surface-to-volume ratio of pores) (Mitchell et al. 2005). It is therefore possible to separate water in differently sized voids within the wood structure (Almeida et al. 2007; Araujo et al. 1992; Fredriksson and Thygesen 2017; Labbé et al. 2006; Thygesen and Elder 2008). The moisture content represented by each significant exponential, ωi, was evaluated as (Telkki et al. 2013):

$${\omega }_{i}={\omega }_{{\text{tot}}}\frac{{S}_{i}}{{S}_{{\text{tot}}}}$$
(1)

where Si (−) is the weight of the exponential i, Stot (−) is the sum of the weights of all exponentials, and ωtot (g g−1) is the total moisture content of the wood specimen determined as:

$${\omega }_{{\text{tot}}}=\frac{{m}_{{\text{eq}}}-{m}_{{\text{dry}}}}{{m}_{{\text{dry}},0}}$$
(2)

where meq (g) is the equilibrium mass after modification, mdry (g) is the dry mass after modification and mdry,0 (g) is the dry mass before modification which can be calculated from:

$${m}_{{\text{dry}},0}=\frac{{m}_{{\text{dry}}}}{1+{R}_{{\text{mod}}}}$$
(3)

Rmod is the relative change in mass caused by the modification procedure which is shown in Table 1. For the uniform modification, the average values for each modification intensity were used for the moisture content correction. For the interface modification, values for individual specimens were used in Eq. 3. Note that no correction was made for the pyridine controls where a mass loss was seen.

The water population with the shortest T2 represents water in cell walls while the peaks with longer T2 represent water outside of the cell wall in differently sized voids (Almeida et al. 2007; Fredriksson and Thygesen 2017). By using this peak assignment, the moisture content representing cell wall water was determined by Eq. 1 using S1, i.e. the weight of the first exponential. The moisture content representing capillary water was similarly determined by adding the exponentials for all peaks representing capillary water, i.e. all peaks except the cell wall water peak. The number of peaks representing capillary water was different for the different moisture states.

X-ray computed tomography to visualise location of capillary water

One untreated specimen and one specimen of interface acetylated wood (Rmod = 0.099, see Table 1) were vacuum saturated with deionised water. Cylinders with a diameter of about 2 mm and a height of 10 mm where then cut from each of these specimens. The specimens were then mounted in two custom-built pressure plate cells (Engqvist et al. 2019). The design of the pressure plate cell was similar as the one described by Fredriksson and Thybring (2019), but with smaller dimensions. The cells used for the tomography measurements were design around a 0.625″ porous ceramic plate (5 Bar standard, Soilmoisture Equipment Corp., Santa Barbara, CA, USA). The outer diameter of the cells were 16 mm and the internal volume approximately 8 ml. The two water saturated wood specimens were mounted in two cells. To ensure that the wood did not dry, the whole pressure plate chamber was initially filled with deionised water which was let out before the first scan was performed. Each of the specimens was then imaged in water saturated state using a Zeiss Xradia Versa XRM520 at the 4D Imaging Lab, Lund University. For each 3D volume, 2001 radiographs with a pixel side length of 1.5 µm were acquired with an X-ray tube voltage of 120 kV, a power of 10 W and an exposure time of 2.8 s. The radiographs were collected over 360° rotation. The field of view was focused on the centre part of the samples, 3 mm from the top surface. After acquiring the radiographs, the volume reconstruction was performed using the Zeiss Xradia Reconstructor software.

A target pressure of 1.08 bar was then applied in the two chambers. This pressure was chosen since it was similar to one of the pressure levels used in the LFNMR-experiment (see Table 2). The temperature in the tomograph chamber is, however, 28 °C and it is thus not possible to achieve the exact same moisture state as in the pressure plate conditioning performed at 20 °C. Each specimen was imaged after 3 h, 24 h, 1 week and 2 weeks. In addition, the specimens were imaged after drying at 60 °C for 24 h in a sorption balance with a resolution of 0.1 µg (DVS Advantage, Surface Measurement Systems, UK). Unfortunately, there was a leakage sometime between 1 and 2 weeks and the image after 2 weeks showed that the specimens had dried (no capillary water present). For this reason, the images after 1 week were used in the evaluation. Although the specimens were perhaps not fully equilibrated after one week, it was possible to compare the moisture states of the two specimens under the same conditions which was the purpose.

Differential scanning calorimetry in water saturated state

Differential scanning calorimetry (DSC) was used to determine the cell wall and capillary moisture content in water saturated state. Measurements in water saturated state on the same sample types were done by Digaitis et al. (2021) using LFNMR, but since LFNMR at temperatures above 0 °C is known to underestimate the cell wall moisture content at saturation (Beck et al. 2018b; Thybring et al. 2020a), DSC measurements were also performed to get a better comparison between the water saturated state and the different high moisture states achieved by pressure plate conditioning.

Six specimens of each sample type were vacuum saturated with deionised water and cut to fit DSC-pans of 40 µl (diameter 5 mm, height 1.4 mm). The fibre direction was in the plane of the small discs. The DSC-pans with the samples as well as seven pans with deionised water, were then hermetically sealed and placed in the sample robot of the DSC (DSC 3+ , Mettler Toledo AG, Schwerzenbach, Switzerland). The method of Fredriksson and Thybring (2019) was applied: − 20 °C for 5 min, a ramp from − 20 to 20 °C with 2 °C min−1, and finally a temperature stabilisation period of 5 min at 20 °C. After the measurements, the lids of the DSC-pans were pierced with a needle and placed in a vacuum oven at 60 °C for 24 h. The dry masses of the specimens were determined after temperature stabilisation in a desiccator with molecular sieves. The total moisture content, ωtot, of each specimen was evaluated by:

$${\omega }_{{\text{tot}}}=\frac{{m}_{{\text{sat}}}-{m}_{{\text{dry}}}}{{m}_{{\text{dry}}}}$$
(4)

where msat (g) is the vacuum water-saturated mass of each specimen and mdry (g) is the dry mass of each specimen. The moisture content related to capillary water ωcap (g g−1) was determined as the amount of freezable water by evaluating the heat of melting Q (J) by integration of the melting peak. The integration was performed in the software STARe Evaluation Software (version 17.00, Mettler Toledo AG, Schwerzenbach, Switzerland). The moisture content corresponding to capillary water and cell wall water were then determined by:

$${\omega }_{{\text{cap}}}=\frac{\left(\frac{{\text{Q}}\cdot {m}_{{\text{sat}}}}{{H}_{{\text{f}}}}\right)}{{m}_{{\text{dry}}}}$$
(5)
$${\omega }_{{\text{cw}}}={\omega }_{{\text{tot}}}-{\omega }_{{\text{cap}}}=\frac{{m}_{{\text{sat}}}-{m}_{{\text{dry}}}-\left(\frac{{\text{Q}}\cdot {m}_{{\text{sat}}}}{{H}_{{\text{f}}}}\right)}{{m}_{{\text{dry}}}}$$
(6)

where Hf (J g−1) is the enthalpy of fusion of water of 333.7 J g−1. Calibration of the DSC for enthalpy of fusion was done with deionised water (melting point 0 °C, enthalpy of fusion 333.7 J g−1).

Results

Amount and location of water

In absorption, two water populations could be distinguished from the LFNMR-spectra for all specimen types (Figs. S1 and S2); one representing cell wall water and one representing capillary water. Capillary water was present at all moisture levels in absorption, i.e. also at the lowest moisture level at − 487 J kg−1 (99.64% RH), but the amounts were generally small; for example, for untreated wood the amount of capillary water corresponded to a moisture content of about 0.025 g g−1 (Fig. 1). In desorption, the two peaks that represent capillary water in water saturated state (Digaitis et al. 2021) was often divided into several peaks at the moisture states included in the present study (Figs. S1 and S2).

Fig. 1
figure 1

Sorption isotherms for capillary water (upper) and cell wall water (lower) for untreated wood, pyridine controls and interface modifications. Filled symbols show desorption data and open symbols show absorption data. Data for water saturated state (obtained by DSC) (*) and are arbitrarily shown at − 1 J kg−1. No data are available for absorption at 2.97 bar since there was a leakage in this pressure plate cell

Full size image

Total sorption isotherms for all specimen types are shown in Figure S3. In Figs. 1 and 2, these total sorption isotherms are divided into cell wall water and capillary water respectively based on the data from the LFNMR and DSC measurements. As expected, the large decrease in total moisture content seen in desorption (Fig. S3) was mainly caused by a decrease in amount of capillary water (Figs. 1 and 2). Sorption hysteresis was observed not only in total sorption isotherms (Fig. S3), but generally seen also in cell wall water and capillary water respectively for all specimen types (Figs. 1 and 2).

Fig. 2
figure 2

Sorption isotherms for capillary water (upper) and cell wall water (lower) for untreated wood, pyridine controls and uniform modifications. Filled symbols show desorption data and open symbols show absorption data. Data for water saturated state (obtained by DSC) (*) and are arbitrarily shown at − 1 J kg−1. No data are available for absorption at 2.97 bar since there was a leakage in this pressure plate cell

Full size image

For the interface modifications, the amount of capillary water increased with increasing modification intensity at all moisture states except − 26 J kg−1 (Fig. 3). Additionally, at the lowest moisture state (− 487 J kg−1), a significant amount of capillary water was still present in the interface modified wood, but not in the untreated wood. The results from the LFNMR measurements thus showed that the interface modified wood held water in the cell lumina to a higher extent than in untreated wood. This was also observed from the images obtained by X-ray computed tomography where the interface modified specimen contained more capillary water than the untreated specimen (Fig. 4). When comparing the location of water between the two specimen types, the difference was primarily in the latewood; in the interface modified specimen, the latewood was filled to a higher extent than in the untreated specimen (Fig. 4). In the untreated wood, both latewood and earlywood cells had both filled and empty cell lumina. The ray cells were mostly filled in both specimen types, but some emptied ray cells were seen in the untreated wood. Figure S7 shows the X-ray images for all scan times.

Fig. 3
figure 3

The capillary moisture content ωcap as a function of the intensity of the modification Rmod at the different moisture states achieved by pressure plate conditioning. Results for uniform modifications (blue) and interface modifications (green) are shown together with data for untreated wood (at Rmod = 0). Solid lines show data after conditioning in desorption and dashed line show data after conditioning in absorption. Pyridine controls are shown as grey filled (desorption) and open (absorption) circles at Rmod = − 0.023. No data are available for absorption at −  298 J kg−1 since there was a leakage in this pressure plate cell

Full size image
Fig. 4
figure 4

Results from X-ray computed tomography. Imaging was performed for one untreated and one interface modified specimen (high, Rmod = 0.099). Both specimens were imaged after 1 week of conditioning in the pressure plate cells

Full size image

In contrast, in the uniformly acetylated wood, there was no clear trend regarding amounts of capillary water as function of modification intensity. The amount of capillary water in the uniformly modified wood with highest Rmod was, however, often higher than in the untreated wood (Fig. 3).

Figure 5 shows the cell wall moisture content as a function of Rmod for the interface modifications and uniform modifications, respectively. For the uniform modifications, there were a decrease in cell wall moisture content with increasing modification intensity in both absorption and desorption (Fig. 5). For the interface modification of the lowest intensity, the cell wall moisture content was generally similar as for the untreated controls (Fig. 5). However, for the interface modification with higher intensity, a decrease in cell wall moisture content was seen at all moisture states in both desorption and absorption (Fig. 5).

Fig. 5
figure 5

The cell wall moisture content ωcw as a function of the intensity of the modification Rmod at the different moisture states achieved by pressure plate conditioning. Results for uniform modifications (blue) and interface modifications (green) are shown together with data for untreated wood (at Rmod = 0). Solid lines show data after conditioning in desorption and dashed lines show data after conditioning in absorption. Pyridine controls are shown as grey filled (desorption) and open (absorption) circles at Rmod = −  0.023. No data are available for absorption at − 298 J kg−1 since there was a leakage in this pressure plate cell

Full size image

State of water in terms of relaxation time

The T2 for cell wall water generally increased with increasing modification intensity in both absorption and desorption, primarily for the uniform modification (Fig. 6). However, in absorption, the T2 for the cell wall water for the interface modification with the lowest intensity, was sometimes similar to that of untreated wood (Fig. 6). For the single capillary water peak found in absorption, there was no clear relationship between the T2 and the modification intensity (Fig. S4). In desorption, these data were more difficult to interpret since there were different numbers of capillary water peaks at the different moisture states. However, from Figs. S1 and S2 it is seen that the T2 of the peak representing water in cell lumina (the peak with the longest T2) generally increased with increasing modification intensity for uniform as well as interface modified wood at the different moisture states.

Fig. 6
figure 6

The T2 for cell wall water as a function of Rmod for uniformly (green) and interface (blue) modified wood at different moisture states. The data at Rmod = 0 are for untreated controls and pyridine controls are shown as grey filled (desorption) and open (absorption) circles at Rmod = −  0.023

Full size image

Discussion

Influence of modifications on capillary water

The results from the LFNMR measurements showed that in desorption, capillary water was present at all moisture states studied (Figs. S1 and S2). The T2 of the peak representing water in cell lumina increased with increasing degree of modification for both the interface modified and uniformly modified specimens (Figs. S1 and S2). This is similar to what was seen in water saturated state for the same material (Digaitis et al. 2021) as well as in other studies on acetylated wood (Beck et al. 2018b; Thygesen and Elder 2008; Yang et al. 2020). These results reflect that the T2 is affected by the change in chemical environment caused by the acetylation of the lumen-cell wall interface.

In absorption, only one peak for capillary water was present (Figs. S1 and S2). The mechanism of uptake of capillary water is capillary condensation and the lowest RH used (99.64%) corresponds to a pore radius of about 2.8 µm for cylindrical pores and 1.4 µm for slit-shaped pores (Thybring and Fredriksson 2023). This, together with that only small amounts of capillary water were present in absorption, makes it is reasonable to assign this single capillary water peak in absorption to water in small macro voids such as pit chambers and the pointed ends of tracheid lumina. The T2 of this single capillary water peak was, however, not affected by the degree of modification. This is thus different from the results obtained in desorption and the results at water saturated state from Digaitis et al. (2021).

For the interface modified wood, the amount of capillary water in desorption increased with increasing modification intensity (Fig. 3) except at one moisture state, − 26 J kg−1, i.e. the moisture state closest to saturation. In addition, the results from the X-ray imaging showed that more water was present in the interface modified specimen than in the untreated wood. The results thus indicate that the interface acetylated wood held more capillary water during desorption from water-saturation than untreated wood. This might appear surprising, especially since this effect was not seen to the same extent for the uniformly modified wood. A possible explanation could be that the large gradients in acetylation across the cell walls for interface modified wood (Digaitis et al. 2021) results in higher modification intensity at the cell wall-lumen interface than in uniformly acetylated wood, even though the overall Rmod is lower. Acetylation affects the contact angles between water and wood surfaces which is seen from studies of acetylated wood and lignocellulosic fibres. These show that the contact angle generally is slightly higher for acetylated wood than for untreated controls (Adebawo et al. 2016; Bryne and Wålinder 2010; El Boustani et al. 2015; Englund et al. 2009; Moghaddam et al. 2015; Wålinder et al. 2013; Yin et al. 2021) and increases with increasing degree of modification (Adebawo et al. 2016; El Boustani et al. 2015). That is, if the local Rmod at the interface was higher in the interface modified material, this could be the reason why the effect of acetylation on the capillary water was seemingly more pronounced in the interface modified wood than in the uniformly modified wood. However, Engelund et al. (2010) calculated the influence of the contact angle on the amount of capillary water at high relative humidity levels and found that the largest amount of capillary water should be present with a contact angle of zero and the amount of water decreases as the contact angle increases, i.e. higher contact angle from acetylation should result in smaller amounts of capillary water. However, the calculations of Engelund et al. (2010) show only small differences in moisture content of about 0.0002 g g−1 at 99.5% RH and 0.0025 g g−1 at 99.9% RH between a contact angle of 0 and 80°. Since the difference in contact angle between acetylated wood and untreated wood is generally in the range of 14–30° (Adebawo et al. 2016; Bryne and Wålinder 2010; El Boustani et al. 2015; Englund et al. 2009; Moghaddam et al. 2015; Wålinder et al. 2013), the effect calculated by Engelund et al. (2010) would be too small to be detected in the present study. Moreover, these calculations should primarily be valid for absorption since they were based on the capillary condensation mechanism. For these reasons, it is unlikely that a change in contact angle can explain these results. In desorption, the decisive mechanism for when capillary water is emptied is the ink-bottle effect (Fortin 1979). That is, the large cell lumina are emptied at the relative humidity that correspond to the size of the opening of the bordered pits which acts as ink-bottle necks. Therefore, a second possible explanation to why the interface modified wood would hold more water in desorption could be that the cell wall bulking caused by the intense acetylation at the interfaces might shrink the size of the bordered pits which means that a lower water potential/relative humidity is needed to empty the cell lumina in acetylated wood. To conclude, there is thus no clear explanation to why the amounts of capillary water in the interface modified wood generally are higher than in untreated wood at the same moisture state.

Influence of modifications on water within cell walls

For the uniform modifications, there were a decrease in cell wall moisture content with increasing modification intensity in both absorption and desorption (Fig. 5). This is in line with previous studies in hygroscopic and water saturated states (Beck et al. 2018b; Digaitis et al. 2021; Hill et al. 2005; Papadopoulos and Hill 2003; Popescu et al. 2014; Thybring et al. 2020b) and the reason is that acetylation bulks the cell walls as well as reduces the hydroxyl accessibility (Thybring et al. 2020b). In addition, a small hysteresis in cell wall moisture content was observed (Figs. 1, 2 and 5). This is a well-documented phenomenon for untreated as well as acetylated wood in the hygroscopic range (Digaitis et al. 2021; Fredriksson and Thybring 2018; Popescu et al. 2014), and has also been shown for untreated Douglas fir in the over-hygroscopic range (Fredriksson and Thybring 2019), but have not previously been shown for modified wood in the over-hygroscopic moisture range.

For the interface modification of the lowest intensity, the cell wall moisture content was generally similar as for the untreated controls in absorption as well as in desorption (Fig. 1). However, for the interface modification with higher intensity, a decrease in cell wall moisture content was seen at all moisture states in both desorption and absorption (Fig. 1). These results are in line with previous results in water saturated state (Digaitis et al. 2021). The T2 of the cell wall water was higher in the uniformly modified wood than in the untreated wood (Fig. 6) in line with previous studies (Beck et al. 2018b; Thygesen and Elder 2008, 2009; Yang et al. 2020) even though the amount of water within cell walls was lower. Typically, the T2 decreases with a lower cell wall moisture content (Cox et al. 2010) as the water induced cell wall porosity, and presumably the size of the cell wall nano-pores, decrease (Penttilä et al. 2020). The increased T2 in the uniformly modified wood with a lower cell wall moisture content thus points to weakened wood–water interactions from acetylation. Similarly, the T2 of cell wall water slightly increased in the interface modified wood with increasing degree of modification, but significantly less than in uniformly modified wood (Fig. 6). This again points to weakened wood–water interactions, since the cell wall moisture content decreased in the interface modified wood with the highest Rmod (Fig. 1). Compared to the study in water saturated state by Digaitis et al. (2021), the results for the interface modification with the lowest intensity was slightly different. In water saturated state, no shift in T2 or change in moisture content was seen for the lowest degree of interface modification (Digaitis et al. 2021). However, in the non-saturated states in the present study, the interface modification of the lowest intensity gave a slight shift in T2 although the cell wall moisture content was similar as for the untreated material.

Methodological aspects

Figures S5 and S6 show the moisture content split into cell wall water and capillary from the present study together with data for water saturated state from Digaitis et al. (2021). In the water-saturated state, a slightly higher cell wall moisture content was found with DSC (the present study) than with LFNMR (Digaitis et al. 2021). Moreover, the cell wall moisture contents in the water-saturated state in Digaitis et al. (2021) are lower than those found in non-saturated states of the present study (Figs. 1 and 2); both using LFNMR. Conversely, the saturated cell wall moisture content evaluated with DSC appears reasonable when comparing the trends in the cell wall moisture contents in the non-saturated states (Figs. 1, 2, S5, S6). However, the lower cell wall moisture content in the water-saturated state obtained by LFNMR is most probably an artefact related to the LFNMR method, which is known to underestimate the cell wall moisture content in untreated and modified wood in the water saturated state. For instance, the same pattern has been identified in previous studies where LFNMR measurements were performed during drying of specimens. Among these, Menon et al. (1987) observed a lower cell wall moisture content at a total moisture content of 216% than after drying to 92%. Likewise, Thygesen and Elder (2008) observed lower cell wall moisture contents in untreated and acetylated Norway spruce at water-saturation than after drying to 120% and 74% moisture content, respectively. In addition, data from Gezici-Koç et al. (2017) show a higher moisture content after conditioning to a high RH-level under hygroscopic conditions than the cell wall moisture content in water saturated state evaluated from LFNMR data for teak and oak. Furthermore, LFNMR at room temperature has been shown to underestimate the cell wall moisture content in water saturated state compared to other methods (Beck et al. 2018b; Thybring et al. 2020a). The lower cell wall moisture contents measured with LFNMR in water saturated state at room temperature is probably caused by an exchange of water molecules between different environments during the measurement (Thybring et al. 2020a). Since the populations of cell wall water and capillary water have a larger contact area at water-saturation than at moisture states with less capillary water present, the exchange of water molecules is likely slower in non-saturated states. Because of these method related uncertainties, comparison between cell wall moisture contents/capillary moisture contents at different moisture levels should be done with care. In the present study, we have therefore focused on comparing different specimen types at the same moisture state.

Despite the above-mentioned drawback with the LFNMR method it is worth noting that it has the advantage that high samples (10 mm) can be used. Since the specimen height in the longitudinal direction is larger than the length of one tracheid cell, the ink-bottle effect, which is the main mechanism for hysteresis in the over-hygroscopic moisture range, is remained. For DSC-measurements, thinner samples are generally used because of the size of the DSC-pans, with the drawback that the amount of capillary water might be underestimated (Fredriksson and Thybring 2019). In Figures S5 and S6 it is seen that the capillary moisture content is sometimes lower for the DSC-data than for the LFNMR-data in the saturated state which is probably an effect of this. This effect was, however, minimised by cutting the DSC-samples so that the longitudinal direction was in the plane of the disc, i.e. the length in this direction was up to 4–5 mm instead of about 2 mm which would have been the case if the specimens had been cut with the longitudinal direction in the thickness of the disc.

Future outlook

The results from the present study show that combining pressure plate conditioning and targeted modification makes it possible to create a range of moisture states and distributions of water within the wood structure. Here, conditioning a targeted modified wood specimen to a particular water potential resulted in very different amounts of capillary water and cell wall water depending on the conditioning history (absorption or desorption) and choice of modification. The exact role of water in the fungal degradation of native as well as modified wood is not yet clarified (Brischke and Alfredsen 2020; Thybring et al. 2018; Zelinka et al. 2016). A lot of focus has historically been on total amount of water (moisture content), but the location of water is presumably also important for the fungi. Extensive research has been performed to determine moisture thresholds required for fungal growth on wood (Brischke and Alfredsen 2020), but it is challenging to control the moisture conditions in these experiments. Although degradation of wood often is studied in a controlled climate, the fungi generally has access to liquid water outside of the wood specimen, and additionally, an average moisture content of a whole piece of wood is generally used as a measure (Brischke and Alfredsen 2020). These experiments are generally designed to compare the performance of different materials, and designing experiments to investigate the influence of location of water on fungal degradation is challenging. Such experiments, with more precise control of the moisture conditions, are, however, needed in order to gain a detailed understanding about the role of water in degradation processes (Alfredsen et al. 2015). The approach of combining precise conditioning with targeted wood modification, as outlined in the present study, aids to create well-defined moisture states and serves as a mean in studies aiming to understand the role of amount, location and state of water in fungal degradation processes.

Conclusions

By combining pressure plate conditioning and targeted modification using acetic anhydride, it was possible to vary the amount, location and state of moisture in wood. Capillary water was present in both absorption and desorption for all the included moisture states for all specimen types, but the amounts of capillary water in absorption were generally small. Increasing degree of interface modification increased the amount of capillary water in desorption to non-saturated states compared to untreated wood. The reason for this effect is not clear and the same trend was not seen for the uniformly modified wood, although the amount of capillary water in the uniformly modified wood with highest degree of modification often was higher than in the untreated wood. The amount of cell wall water was decreased by uniform modification; increased degree of modification gave lower cell wall moisture content. The interface modification of the lowest intensity did generally not affect the moisture content, but a slight effect was seen for the interface modification with higher intensity. Hysteresis was observed in both cell wall water and capillary water, especially in the latter. The wood–water interactions were weakened by acetylation, but only slightly for the interface modified material. The results show that it is possible to tune the amount, location and state of water in wood at the high moisture states relevant for fungal degradation by combining pressure plate conditioning and targeted wood modification. This opens new possibilities for designing materials and moisture states for fungal degradation experiments.