The effect of ionic liquid and superbase pre-treatment on the spring-back, set-recovery and Brinell hardness of surface-densified Scots pine

Specimen preparation

The study was carried out on knot-free sapwood specimens of Scots pine (Pinus sylvestris L.) with dimensions of 124 mm (L)×25 mm (T)×18.5 mm (R). The annual ring orientation was within ±15° parallel to the tangential wood surface to be densified. The specimens were kept in a climate chamber at a temperature of 20°C and 65% relative humidity (RH) until they reached the equilibrium moisture content (EMC). In order to obtain specimen groups with roughly the same average density between the groups, the specimens were sorted into seven categories according to their density, and the specimens belonging to either the highest or the lowest density category were discarded. The remaining five density categories ranged from 400 kg m⁻³ to 525 kg m⁻³, and to make the final grouping, the specimens were then selected so that each specimen group (Table 1) had the same number of specimens from each density category.

Preparation of chemicals

The specimens were treated with various ILs or organic superbases, diluted with ethanol to different concentrations (Table 1). The IL 1-butyl-3-methylimidazolium chloride [Bmim(Cl)], and the organic superbases 1,5-diazabicyclo(4.3.0)non-5-ene (DBN), 1,8-diazabicyclo(5.4.0)undec-7-ene (DBU) and 1,1,3,3-tetramethyl-guanidine (TMG) were purchased from Acros Organics, Geel, Belgium (purity: 98+%). Another IL was made by mixing DBN with propionic acid in an equimolar ratio, resulting in DBN propionate [DBN(EtCO₂H)].

Chemical pre-treatment

All specimen groups to be chemically treated were oven dried at 103°C to an MC of 0% prior to the pre-treatment. The chemical solutions were first applied with a pipette onto the tangential barkside-oriented wood surface to be densified. In order to keep the amount of solution consistent, the content of one pipette filling (3.5 g) was distributed over the surfaces of two specimens. Within 1 min, the solutions were completely soaked up by the wood. This procedure was repeated 3 more times for each specimen. In order to encourage the penetration of the chemicals into the wood, the surfaces were punctured with 3–4-mm deep holes with a diameter of about 0.8 mm and distance between the holes of 5 mm in the tangential direction and 10 mm in the longitudinal direction. This was done manually with a needle, which was marked with tape to ensure a constant depth of the holes. Only group 250NDBU (Table 1) was left without holes. After the pre-treatment, the specimens were placed in an oven at a temperature of 60°C for 14 h to evaporate the ethanol. Omitting this step would lead to the destruction of the wood during the densification stage due to too high a steam pressure.

Densification process

The specimens were densified in the radial direction in a press with one of the platens heated and equipped with a water-cooling system. Two or three specimens were densified at the same time. For the first 10 s of the process, only a low pressure sufficient to maintain proper contact between the wood surfaces and the heated press platen was applied. The pressure was then increased to a level that would lead only to densification if the wood cells were sufficiently plasticised. The correct amount of pressure to achieve this behaviour was determined during a pre-study by calibrating the pressure levels for different densities of the raw material. Depending on the parameters and the treatment, the target thickness of 14.98 mm (set by metal stops in the press) was achieved after a time between 90 s and 240 s (Table 1). Regardless of the time taken to reach the target compression, the specimens were kept at the set temperature for a total of 240 s, after which the cooling system was started. The press was opened when the press platen temperature reached 60°C, and the SB was then measured (see section elastic SB), after which the specimens were kept in an oven at 60°C before the SR determinations were carried out (see section SR). No SR occurred during this stage.

Specimen grouping

Table 1 lists the specimen groups with their respective treatments and densification parameters. Group C150 is the reference group, densified with parameters similar to those commonly used in surface densification procedures. The purpose of groups R270-200 was to study the effect of thermal modification on the SB and SR and to separate it from the effect of the chemical pre-treatment. After being densified with the same parameters as group C150, these four groups were put back in the press for 240 s at temperatures between 270°C and 200°C, corresponding to the densification temperatures of the chemically treated specimens.

Elastic spring-back (SB)

The SB was determined from measurements made with a digital calliper with a precision of ±0.03 mm (according to the manufacturer). SB is defined as:

where B₀ is the original dimension of the specimen in the densification direction before densification, B_c is the target dimension of the specimen after densification (in this case 14.98 mm) and B′c is the actual dimension of the specimen after densification. The SB can thus vary between 0% and 100%, where 0% indicates no SB and 100% indicates total SB.

Set-recovery (SR)

The SR measurements were carried out on specimens with dimensions of 20 mm (L)×25 mm (T)×≈15 mm (R), cut from the densified specimens. Before the SR procedure the samples were kept in an oven at 60°C in order to prevent re-moistening. The specimens were measured with a digital calliper after one and two cycles of soaking in water for at least 14 h at room temperature, followed by oven-drying at 60°C for 24 h. The SR is defined as:

where R₀ is the oven-dry dimension of the specimen in the densification direction before densification, R_c is the dimension of the specimen after densification and R′c is the dimension of the densified specimen after the wet-dry cycling. The SR can vary between 0% and 100%, where 0% indicates that there is no SR, i.e. that the compression-set is completely fixed, and 100% indicates that the recovery is total. After the first wet-dry cycle, the specimens exhibited cupping of the densified surface, and for this reason the average value of the maximum and minimum dimension over the surface of each specimen was used for R′c.

There was a problem in this way of calculating the SR because the initial, uncompressed oven-dry dimensions were not known for the specimens densified at an MC of 14.8% or 5.0%. In this case, the oven-dry dimensions of the specimens were assumed to be equal to the average dimensions of all the specimens densified at an MC of 0%. As the standard deviation in the dimensions of these specimens was only 0.07 mm, this assumption appears to be sufficiently accurate.

Brinell hardness (HB)

The HB was measured in accordance with the standard EN 1534, but with several modifications (CEN 2010). A Zwick Roell ZwickiLine 2.5 TS universal testing machine equipped with a 2.5 kN load cell and a steel ball with a diameter (D) of 10 mm was used. The force was increased at a rate of 4 kN per minute to a nominal force (F) of 1 kN, which was held for 25 s before being released.

According to the standard, the HB is calculated from the diameter of the indentation in the specimen, but in this study the HB was calculated from the depth of the indentation. The reason for this modification of the test procedure was that it was difficult to measure the diameter of the indentation in wood. In contrast to steel or other metals, wood does not leave an indentation with a clear circular border because of its anisotropic properties. The depth of the indentation (h) can, however, be measured with the hardness testing equipment with high accuracy and precision. The HB values in this study were used only for comparison between the test groups. Niemz and Stübi (2000) and Laine et al. (2013a) previously implemented a similar approach. Thus, the HB is here defined as:

A total of 30 hardness measurements were made on each specimen group, 15 on the densified surface and 15 on the opposite, undensified surface as control measurements. The 30 measurements per group were distributed over five specimens with six measurements per specimen.

Data analysis

All the collected data were processed in Microsoft Excel, after which a principal component analysis (PCA) screening was done using the software package SIMCA provided by Umetrics (Umeå, Sweden). With the help of PCA, a multivariate data analysis method, it is possible to find correlations in large datasets that are difficult to detect with univariate data analysis methods.

Elastic SB

It has been reported that the SB of untreated surface-densified Scots pine is about 50% without a cooling stage after the densification stage (Neyses et al. 2017). With a cooling stage, the SB can be reduced to 10–15% (Laine et al. 2013b; Neyses et al. 2017), and this was confirmed in the present study by group C150, which had an average SB of 11%. For all other groups, the dimension of the specimen in the densification direction was, within the measurement accuracy, the same as the target compression, and the SB can thereby be considered negligible.

For untreated and surface-densified wood, Laine et al. (2013c) have reported a reduction in SB with increasing densification temperature, when the densification stage is followed by a cooling stage. In this case, however, the low degree of SB seems to be at least partly due to an effective plasticisation of the wood modified by the ILs or superbases, in combination with a very low deformation of the unmodified wood volume beneath the modified wood during the densification process.

Set-recovery (SR)

Figure 2 shows the average SR of the untreated specimen groups after two wet-dry cycles. For all the chemically treated groups, there was no significant difference in SR between the first and second wet-dry cycles, i.e. the SR did not increase after the first wet-dry cycle. For the untreated groups, there was a slight increase between the first and second wet-dry cycle.

Figure 2:

Overview of the SR.

Dark grey bars are untreated specimen groups and light grey bars are chemically treated specimen groups according to Table 1. The error bars show ±1 standard deviation.

The reduction in SR can be attributed both to the press-platen temperature and to the chemical treatment, but the SR of the R200-270 groups compared to that of the modified wood at the corresponding temperatures suggests that the effect of heat, i.e. thermal degradation, was much less than the effect of the chemical treatment. It was possible with all the tested chemicals to reduce the SR to 20% or even lower. For the DBU treatment, an increase in press-platen temperature (200DBU, 230DBU, 250DBU, 270DBU), or an increase in chemical concentration (250DBU, 250LDBU, 250HDBU) reduced the SR. An increase in MC of the specimen before densification (250WDBU) or the absence of punctured holes in the surface to increase the penetration of chemicals (250NDBU) increased the SR. It was not possible to see any influence on the SR due to the variation in annual ring orientation in the cross-sections of the specimens.

Figure 3 shows that the treatment with Bmim(Cl) or DBN(EtCO₂H), followed by densification at 270°C led to an almost black surface, whereas the superbase treatment resulted in a dark brown colour. This indicates that there are different mechanisms underlying the two types of chemical treatment. Our hypothesis of the formation of an in-situ IL after treatment with a superbase, may not be correct, and instead there may be a different mechanism.

Figure 3:

Surface colours after the different chemical treatments at a press-platen temperature of 270°C.

Figure 4 shows the relationship between the average SR and the densification temperature. The light curve (squares) corresponds to specimens without chemical treatment (C150 and R200-270 groups), and the dark curve to the 270-200DBU groups. The reduction in SR increases with increasing temperature and, in the case of the specimens treated with DBU, the relationship appears to be almost linear.

Figure 4:

Average SR as a function of the densification temperature.

The dark curve corresponds to groups C150 and R200-270, and the light curve to groups 200-270DBU.

Figure 5 (untreated specimens) and Figure 6 (specimens treated with DBU) show the colour changes in the specimens in relation to the pressing temperature. With increasing pressing temperature, the untreated specimens gradually became darker, but there were only small differences in colour between 200DBU and 230DBU, and between 250DBU and 270DBU, with a relatively large difference between 230DBU and 250DBU. However, there is no correlation between this observation and the reduction in SR. We suggest that even higher temperatures should be investigated and also lower temperatures in the case of the chemically treated specimens, in order to obtain a more complete picture of the relationship between the SR of the chemically modified specimens and the temperature.

Figure 5:

Surface colour changes due to the effect of heat; untreated specimens.

Figure 6:

Surface colour changes due to the effect of heat; specimens treated with DBU.

A comparison of the 250DBU, 250HDBU and 250LDBU groups with, respectively, 20%, 30%, and 10% concentration of DBU in ethanol, indicates that an increase in concentration leads to a reduction in SR, but further studies are needed to determine the precise nature of this relationship.

The SR of group 250WDBU (initial MC=5%) suggests that moisture is detrimental to the efficacy of the pre-treatment with DBU, as was expected on the basis of published studies on the dissolution of cellulose in ILs (Mäki-Arvela et al. 2010). Zavrel et al. (2009) reported that water molecules create hydrodynamic shells around the IL molecules and that this inhibits direct interactions between the IL and the cellulose. Nevertheless, the decrease in SR was substantial with a reduction from 81% to only 37%.

The surface of the specimens of almost all groups was punctured with small holes in order to facilitate penetration of the chemical solutions into the wood. Surprisingly, the SR of group 250NDBU without holes was only slightly higher than the SR of group 250DBU – which was otherwise treated and densified in the same way. Depending on possible applications of surface-densified wood and their requirements, the puncturing of the surface may thus not be necessary.

Most groups that were exposed to a temperature of at least 200°C show a high intra-group variation in SR (Figure 2), primarily due to outliers with a much higher SR than the other specimens. Based on colour changes in the wood surface caused by the high press-platen temperature, we suspect that an uneven temperature distribution within the press platen was responsible for this. Even though it was easier to notice in reality, it can be seen in Figures 3–6 that the densified surfaces are slightly darker at one end than at the other (the darker ends are facing the viewer). A PCA screening did not show a systematic relation between the uptake of the chemical solutions and the resulting SR.

Brinell hardness (HB)

Figure 7 shows the HB of undensified and densified specimen surfaces. There was no significant difference in the average HB of the undensified surfaces. The densification process resulted in an increase in the hardness of the densified surfaces. Densification without chemical treatment (groups C150 and R270-200) resulted in a two-fold increase in hardness from 13.4 N mm⁻² to 26.2 N mm⁻², and there were no differences between the groups. This increase in hardness is in line with results from previous studies (Gong et al. 2010; Laine et al. 2013a; Neyses et al. 2017). The chemically treated specimens showed a increase in HB after densification by a factor 2.7, i.e. from 12.9 N mm⁻² to 34.4 N mm⁻², which is a significant increase over that of the untreated specimens. There were no significant differences in HB between the different chemically treated groups.

Figure 7:

Average HB before (left-handed bar) and after (right-handed bar) densification.

The error bars show ±1 standard deviation.

It appears that the reason for the higher HB of the chemically treated specimens is a more thorough and targeted plasticisation than that of the untreated specimens. Figure 8a shows that, for the chemically treated specimens, the region with the highest level of densification is located at the very surface of the specimen, and in this respect the specimen shown is representative of all the chemically treated specimens. Directly beneath the surface there appears to be a layer approximately 1.2 mm in thickness with a three-fold increase in density. The density then gradually decreases towards the core of the specimen. In Figure 8b, which shows an untreated and densified specimen, it appears that the density peak of these specimens is not located at the very surface, but is somewhat below the surface. This could be due to the heated press platen drying the specimen surface, and thus inhibiting the plasticisation of this region. For the chemically treated specimens, this phenomenon is not relevant because their plasticisation does not rely on moisture. Another observation supporting our hypothesis is the bulging of the side surfaces in the densified area (Figure 8): the chemically treated specimens bulge out at the very surface, while the untreated ones bulge out slightly below the surface.

Figure 8:

Densification close to the surface of (a) a chemically treated specimen and (b) an untreated specimen.

The bars show the distance between annual rings.

As the HB method is a volume-based measurement, it appears not to be representative of the perceived surface hardness of the specimens. In theory, the observed three-fold increase in density should result in a threefold increase in hardness, which was not confirmed by the calculated HB values. Perhaps another type of hardness or scratch resistance measurement method would provide more meaningful results – at least, with respect to the planned use of the densified wood. For this reason, we think it could be worthwhile to study the relationship between different hardness and scratch resistance measurement methods and surface-densified wood treated in different ways. This is especially important with regard to possible applications of surface-densified wood products.

The chemically treated specimens showed a higher intra- and inter-group variation (Figure 7) than the untreated specimens. We can only speculate on the reasons for this variation. A PCA screening revealed a positive correlation between weight gain due to the chemical treatment and HB. The weight gain varied significantly between the specimen groups and was particularly high for the 270DBNIL, 250HDBU and 250WDBU groups. Unfortunately, the data are not strong enough to explain intra-group variations in HB. The apparently uneven temperature of the heated press platen does not explain this variation either, as there was no correlation between press-platen temperature and HB.