Swelling of beech wood (Fagus sylvatica L.) during gaseous ammonia treatment as a function of pressure

Modifying the properties of indigenous wood species is a way of producing materials with improved or adapted properties from a regional sustainable resource. Mechanical densification of wood leads to an increase in bulk density and enables an increase in the strength and stiffness of the material. To reduce structural damages and pressing forces, the wood must be plasticised before densification. After densification, the compacted state must again be fixed, as the remaining internal stresses can cause the compacted wood to deform back to almost its original state when exposed to moisture and temperature (Blomberg et al. 2006; Buchelt et al. 2014; Kutnar and Kamke 2012). Ammonia treatment combines these properties with very good plasticisation as well as by fixing the compacted state of the wood (Hackenberg et al. 2022a, 2022b).

Indeed, the modification with gaseous ammonia was at least in a related procedure already an industrially applied process (Stojčev 1979). However, the understanding of the process at that time was only empirical and a renewed application requires a comprehensive understanding of the processes during the modification.

Ammonia interacts with wood in a similar way to water, as it is a strong dipole of similar size and also forms hydrogen bonds to the wood substance. Therefore, as in the process used here, it also leads to a swelling of the wood (Bariska 1975). Table 1 lists the swelling values of several authors and wood species. The first studies on the swelling behaviour of wood in liquid and gaseous ammonia were carried out in the middle of the twentieth century. Stamm (1955) described that the swelling in liquid ammonia of selected North American softwoods is increased in tangential direction by approximately 50% compared to water swelling. The radial direction is not affected except for Douglas fir. Bariska (1974) published precise swelling values of beech impregnated in liquid ammonia for 24 h. The swelling of the tangential direction is increased from 14 to 19% when comparing the swelling in water and in liquid ammonia. In the radial direction, however, there is a strong reduction from 6 to 3%. Even more surprising is the reduction in the longitudinal swelling from 1.4% to a negative value of − 1%.

Unlike water, ammonia is able to swell the crystalline areas of cellulose (Yamashita et al. 2018). It is reasonable to assume that the changed swelling behaviour is in relation with the swelling of crystalline cellulose, as cellulose makes up the largest amount of the cell mass.

The aim of this paper at this background was to find out how the wood behaves during the gaseous ammonia treatment and what effects can be observed. For this purpose, the non-steady swelling behaviour was recorded by means of a special method.

The experimental setup was to be kept simple, and thus, an optical measurement of the dimensional change was used to determine the swelling behaviour. Since the expected absolute dimensional change for common laboratory specimen geometries would be very small in the longitudinal direction, only the radial and tangential anatomical orientations were investigated. Beech wood (Fagus sylvatica L.) specimens were taken from defectless wood with a dimension of 20 × 20 × 20 mm³ (L × T × R). Gaseous ammonia treatment was conducted 24 h long at 20 °C and at 9 different pressure levels. The lowest pressure level was slightly higher than ambient pressure, because of the automatic pressure control. The highest pressure level was almost saturated steam pressure in order to prevent condensation effects. The pressure stages are listed in Table 2 and are given in the ratio of partial pressure to saturated steam pressure.

When cutting the samples, 10 samples each were made from a wooden bar (in longitudinal direction) and assigned to the 9 pressure levels in ascending order. A tenth sample was used as a reference for swelling in water. A total of 2 sets of specimens were made from wooden bars originally placed next to each other (matched samples). One set of specimens was oven dried (103 °C) and then stored over dry silica gel. The other set of samples was also oven dried and afterwards stored at 65% humidity and 20 °C. After drying and conditioning, the mass and dimensions were determined.

Figure 1 shows the schematic setup of the experiment. The experiment was carried out in a small autoclave (5), which was closed on the upper side with a pressure resistant glass pane (3). The inner pressure chamber had a diameter of 5 cm and a height of 3 cm. The autoclave used was a 1 ½ inch end cap with a female thread, as used in pipe construction. This had several advantages: it is pressure resistant, inexpensive and required only minor adjustments. In addition, due to the internal thread, it also had a very large surface on the inside, which improved the temperature control of the pressure chamber. The autoclave was held together with the help of two flanges (2) and (10) and screws (1). Above and below the glass pane were seals (4), the lower one to seal the autoclave and the upper one to compensate mechanical stresses on the glass. Evacuation and gas supply was done through a PTFE tube (12). The entire autoclave was placed in a water bath (14) filled up to the lower edge of the glass pane. A laboratory thermostat was connected to a spiral pipe (13) that was positioned as a heat exchanger in the water bath around the autoclave and a small flow pump (11) ensured circulation inside of the bath. The water bath was set to a temperature of 20 °C. The specimen (7) was positioned centrally in the autoclave on the longitudinal surface and was held in position by a small spiral spring (6) located between the glass pane and the sample. Below the sample was a plastic plate (9) made of polyethylene, which was grained in the area of the sample in such a way that gas could also penetrate the sample from below. The colour of the plastic plate was green to create an easily recognisable optical contrast to the measuring marks. The measuring marks (8) were made out of white polyethylene and were attached to the four radial and tangential surfaces of the sample. For this purpose, the measuring marks had a groove in which a rubber band ran that enclosed the sample and thus fixed all four measuring marks. The contact surface of the measuring mark was 8 × 8 mm and the distance measured from the contact surface to the measuring edge was 4 mm. The reference surface was half the height of the measuring mark, so that when the measuring mark was positioned centrally on the respective radial or tangential sample surface, the reference surface was also half the surface of the sample. The reference surface was the plane on which the optical focus was later set, so that the outer edge of the measuring mark was in focus. The optical distance between two opposite outer edges of the measuring mark was measured.

A digital camera (Olympus E-M10-3) and a lens with a short working distance (Olympus Zuiko Digital 14–42 mm f:3.5–5.6 2 II R) were used to take the pictures, so that the front of the lens was 150 mm from the focal plane away. An LED ring light was attached to the front end of the lens for illumination. The top view on the autoclave is shown in Fig. 2 left.

During the experiment, images were taken automatically at a time interval of 1 min. The resulting raw images were analysed with the image processing software ImageJ/Fiji. With the help of a macro, the images of each test could be evaluated automatically. The raw image had a dimension of 3456 × 3456 pixels and was first cropped to a section of 2000 × 2000 pixels (Fig. 2 middle). Depending on the necessity, the desired image area was aligned appropriate to the image middle axes by rotation and shifting. This was important to guarantee the stability of the algorithm and also to reduce the distortion caused by the optical aberration. The distortion was evaluated with the software ImageJ/Fiji and as the distortion is particularly low in the areas of the image middle axes, the distortion could not be detected for this experimental setup. After cropping, an area was defined around each measuring mark in which the image area was divided into its red, green, and blue colour channels and then, reassembled into a black and white image with the help of a threshold function. The values of the threshold function were initially set for each set of images. They were optimised for maximum contrast based on slight changes in lighting. The background now appeared white and the measurement marks black (Fig. 2 right). Subsequently, the change from white to black was determined centrally at each measuring mark over a length of 200 pixels (approx. 3.3 mm). In order not to take into account partial image errors in the area of the evaluated measuring mark edge and thus, to make the measuring result stable against deviations, the median was used as the output value of the measuring points. The determined positions were given as a table at the end of the programme. The actual distance between the measuring mark edges was approximately 28 mm at the beginning of the test. Respectively, the number of pixels between these measuring mark edges in the image was 1700, resulting in a resolution of 60.7 pixels/mm. The pixel length of 0.016 mm thus represented the maximum measurement accuracy due to the quantisation of the measurement distances. Further non-quantifiable measurement errors resulted from blurring at the measuring mark edge, image noise and possibly changing lighting conditions. Furthermore, the sample moved minimally due to strong swelling in the autoclave, which also caused the sample to move in the evaluated image area. The macro used for the evaluation is included in the additional data.

Before each experiment, the sample had to be precisely positioned in the centre of the autoclave and then, the autoclave was assembled. At the beginning of a gassing, a lot of condensate was formed. The condensate is formed by the sorption of the ammonia on the surface of the wood. On the one hand, the ammonia fumigation leads to a strong heating of the sample, which causes it to dry and the water to condense on the surface of the autoclave. On the other hand, the ammonia replaces the water on the surface of the wood due to a higher binding energy. The resulting condensate will always be a water-ammonia solution, as both substances are very soluble in each other. To ensure that this condensate condensed on the autoclave wall and not on the glass pane, the glass pane was tempered to 50 °C before each experiment (laboratory oven). Preliminary heating was not necessary, as large amounts of condensate only were formed at the beginning. The glass pane already had a temperature of 25 °C at the end of the first 10 min. After the period of 24 h, 1440 pictures were taken for each experiment. The autoclave could then be evacuated, opened and prepared for the next experiment. Tangential and radial swelling was calculated according to Eq. 1:with d₀ as the dimension before ammonia treatment and d₁ as the swollen dimension.

The swelling process of specimens which were oven-dry prior to ammonia gas treatment is shown in Fig. 3 for the tangential direction and in Fig. 4 for the radial direction. The curves are labelled with the saturated steam pressure ratio of the test and the maximum swelling of a sample after water storage is shown as a reference.

The swelling process of specimens which were stored in humid condition at 65% RH/20 °C prior to ammonia gas treatment are shown in Fig. 5 for the tangential direction and in Fig. 6 for the radial direction. The reference dimension is the dimension before ammonia treatment, which is therefore the humid dimension.

As a summary to the results already presented, Fig. 7 shows the maximum swelling values of all tests. The reference values for water swelling are also included. The test series in which the samples were previously conditioned in humid air is shown in two variants: One with reference to the humid dimensions before ammonia treatment and the other with reference to the original oven-dry dimensions. This makes it easier to compare the absolute swelling values. This allows the comparison of the absolute swelling values, as otherwise the reference dimensions are different. Prior swelling of the beech samples ranged from 3.5 to 3.7% in the tangential direction and 1.7 to 1.9% in the radial direction. The average oven-dry density of all beech specimens was 660 kg/m³.

The swelling of oven-dry beech (Fig. 3) wood samples increases almost constantly in tangential direction depending on the pressure level. At low pressures, maximum swelling is already reached after 2 h. In the medium pressure range, swelling is strongly delayed, indicating an additional reaction. Especially at p_i/p_s = 0.6, it is possible that the sample was still not in equilibrium after 24 h. In the upper pressure range (p_i/p_s ≥ 0.7), the swelling is accelerated again with a simultaneous increase in the maximum swelling up to 23%. The test series p_i/p_s = 0.6 is of particular interest for two reasons. The pressure levels above this led to stronger swelling than is possible when stored in water (tangential water swelling: 16.6%). This over-swelling is consistent with the observation from previous work, as shown in Table 1. In addition, in the pressure range of p_i/p_s = 0.5–0.7, ammonia begins to swell the crystalline regions of the cellulose (Bariska 1974; Bariska et al. 1969). Ammonia starts to infiltrate the cellulose I crystals and forms ammonia-cellulose I. It is likely that the swelling of the crystalline areas of the cellulose is responsible for the over-swelling in the tangential direction. In relation to the test p_i/p_s = 0.6, the beginning of swelling the crystalline areas becomes visible by the strongly delayed swelling process (Bariska and Popper 1971).

In the radial direction of dry beech wood (Fig. 4), the swelling rate increases continuously up to the pressure level p_i/p_s = 0.6. The higher pressure levels do not lead to any further change in the radial dimension, only the swelling rate at the beginning is significantly increased. It becomes clear that the swelling dimension of the reference water sample is not exceeded (radial water swelling: 5.4%). Only Stamm (1955) reported a difference in the case of Douglas fir, which had a similar swelling dimension in the radial direction after swelling in ammonia as in the tangential direction, which was not observed by the authors in other wood species either. However, the question arises, why does the wood behave like this? Since the wood rays are oriented in the radial direction, it is possible that they prevent over-swelling by resisting the internal swelling pressure like a reinforcement. For swelling in ammonia, the same limit value applies as for swelling by water.

The curves of the samples that were humid before ammonia treatment are clearly different from those of the dry samples. In the tangential direction (Fig. 5), swelling initially increases with the pressure level, but already reaches the maximum value at p_i/p_s = 0.6. The curves of the higher pressure levels are then almost identical. Swelling of the crystalline areas of the cellulose also occurs under humid conditions, starting at p_i/p_s = 0.4 (unpublished data). But in contrast to the dry samples, no delayed swelling is observed here. The moisture thus has an accelerating effect, which can be justified, for instance, by better accessibility.

In the radial direction (Fig. 6), the curves at low pressure show similar curves to those observed previously. From p_i/p_s = 0.3, however, the swelling is initially particularly high and then, decreases again. This leads to the fact that the final swelling measures of the high pressure levels are even lower than those of the lower pressure levels. Figure 7 shows the final swelling dimensions of all test series together. In addition, the test series with moist samples is shown with reference to the dry dimensions. This makes the actual change in dimensions visible and can be compared with the dry values as well as the water values. From p_i/p_s = 0.3, the tangential swelling of the moist samples exceeds the value of the water sample, and the structure is thus over-swollen. At the same time, the reduction behaviour in the radial direction starts. There is thus the possibility that the wood is subjected to a transverse contraction. The further expansion in the tangential direction leads to a reduction in the radial direction.

The investigations carried out give a first impression of the swelling behaviour of wood during treatment with gaseous ammonia. The swelling of the crystalline areas of the cellulose and the formation of ammonium cellulose I can be made visible by observing the swelling processes. In this way, ammonia leads to strong swelling in the tangential direction of the wood. The radial direction of the wood, however, remains unchanged. Pressure level and initial moisture content have a strong influence on this swelling behaviour.

For a further understanding of the gaseous ammonia treatment of wood, investigations of the temperature and mass behaviour during the treatment are necessary.