1 Introduction
Commercial wood modification with thermosetting resins relies on an impregnation step, during which the liquid resin is forced into the bulk structure of the wood. The sapwood of Radiata pine (
Pinus radiata) is currently the preferred species of solid wood modified in commercial processes requiring impregnation, such as
Accoya and
Kebony (Jones et al.
2019). Despite its availability in clear (knot-free) lengths and fast growth, the main reason for use is the excellent treatability of its sapwood, which enables a high and uniform resin uptake. This overreliance on Radiata pine and the growing demand of the wood modifying sector for such a highly permeable material have contributed to a considerable price increase in recent years.
Liquid flow during impregnation is influenced significantly by wood anatomy. Limiting factors in both softwoods and hardwoods are the pit diameter and the degree of pit aspiration (Siau
1984). Sufficient permeability is prerequisite for any wood species used for resin modification. However, permeability of solid wood can be improved by various techniques (Tarmian et al.
2020; Nath et al.
2022). In other applications, like the impregnation of veneers, permeability ceases to be a limiting factor as the dimensions through which fluid is required to flow have been radically reduced. This led to the adoption of plywood-based systems by Millett et al. (
1943) in Compreg and to more recent advances in this re-emerging field (Fleckenstein et al.
2018; Bliem et al.
2020; Wascher et al.
2020). Aspects other than the permeability and availability of timber are rarely considered in the choice of wood species, and the literature lacks a comparative study that investigates how wood species with different chemical compositions and cell structures behave in the same treatment. The current study addresses this topic by focussing on the wood species as the main variable.
How can the quality of resin modification be ensured? This question must be answered on multiple levels. In the following, a list with relevant criteria is provided and subsequently discussed in more detail. Quality criteria for the impregnation modification process with thermosetting resin include:
-
Gross uptake due to resin-treatment, evaluated by:
-
Uniformity of resin distribution throughout the sample
-
at macroscopic level, evaluated by:
-
at microscopic level, evaluated by:
-
microscopy (light-, electron-, spectroscopy-) to detect distribution and morphology of resin in wood cells (Behr et al.
2018)
-
at submicroscopic level, evaluated by:
-
spectroscopy (NMR, FTIR, or pyrolysis-GC–MS) to detect molecular incorporation and interactions of resin molecules in the cell wall polymers (He and Riedl
2004; Nishida et al.
2019; Kurkowiak et al.
2022)
-
Effectiveness of the resin within the wood
-
degree of cell wall penetration, evaluated by:
-
physical measurement of bulking coefficient (Tanaka et al.
2015)
-
inverse relation between safranin stain-uptake and cell wall penetration using light microscopy (Biziks et al.
2019)
-
spectroscopy-microscopic information that relates to the resin content in cell wall and lumen (Huang et al.
2014)
-
reduction of water-induced swelling (Hill
2006), evaluated by:
-
Longevity and retention of resin during service life, evaluated by:
It is widely accepted that resin should be distributed uniformly throughout the sample, primarily as it has a swelling effect on the wood, so stresses would develop if areas remained unmodified, while adjacent tissue was modified. This distribution can be considered at different levels of scale. Macroscopic effects such as poor impregnation or migration of resin during drying stages can be detected visually to some extent, and in commercial processes, visual inspection of a cross-section is used to monitor the location of the liquid front after impregnation, or areas of poor uptake, such as heartwood (Pitman
2020). At laboratory scale, this is often not necessary because all regions of the specimens are impregnated equally. However, small-scale effects are frequently studied in laboratory specimens such as the distribution of the resin in cell lumen or detection of resin within cell walls (Rapp et al.
1999; Behr et al.
2018; Biziks et al.
2019). Metrics such as weight percent gain (WPG) describe the overall gravimetric uptake of resin but make almost no statement about the quality of modification in terms of location or its effectiveness within the wood. It is well known that resin in the cell walls is more effective than in the cell lumen (Furuno et al.
2004). The bulking coefficient (BU) is an indicator for the amount of resin located in the cell walls, quantified in terms of the swelling caused by resin access into pore spaces usually accessed by water on wetting (Tanaka et al.
2015). However, some agents that bulk the cell wall initially are not fixed and may leach upon water exposure (Meints et al.
2018). Similarly, some resins polymerise but certain fractions, which remain unreacted due to steric hindrances in the cell wall, do leach over multiple cycles (Laborie et al.
2006). For this reason, it is useful to include longer term cyclic testing.
In terms of performance, improved dimensional stability is a defining characteristic of many modified woods (Ellis and Rowell
1982; Ormondroyd et al.
2015). A commonly adopted test for this is the anti-swelling efficiency (ASE) method during which small wood samples are subjected to cyclic water soaking and oven drying (Hill
2006). During an ASE test, the swelling of a modified group of samples is compared against unmodified controls over several wetting and drying cycles. The modified and unmodified sets should be matched where possible—i.e., from the same plank of timber or at least a plank of similar growth rate and the same species. The inability to use matched samples, for example where systems move out of the laboratory onto production scale, can cause substantial errors within the calculated results (Sargent
2022). Both the control and the modified set are soaked with deionized water for several days and then dried again. This procedure can be repeated multiple times and the swelling and water uptake of both groups are compared. Some authors reject the data of the first swelling cycle, because it may generate artificially high ASE values (Hill
2006). Information about the nature of leaching substances can also be obtained from an ASE test, by analysis of the soaking solution. Typically, the solid residue in the soaking solution is measured by simply evaporating the water from an aliquot (Grinins et al.
2019). However, chemical analytical techniques like ultraviolet spectroscopy (UV), and liquid chromatography mass spectroscopy (LC–MS) can provide more insight into the nature of leaching substances (Kurkowiak et al.
2022).
Using the ASE method, many authors have studied the dimensional stability of different wood species modified with thermosetting resins (Millett and Stamm
1954; Epmeier et al.
2004; Furuno et al.
2004; Grinins et al.
2019; Altgen et al.
2020). However, it is often difficult to compare results due to variations in resin type (melamine formaldehyde, phenol formaldehyde, urea formaldehyde etc.), resin properties (molecular weight, solids content, pH) or variations in the method (sample dimensions, soaking time, use of high relative humidity instead of soaking). Consequently, the effect that species differences have in the treatment cannot be derived from the existing literature. The present study examines the behaviour of seven hardwoods and four softwoods in this modification process. Species were chosen on the basis of different anatomy, chemistry, and density. A comparison of selected studies is provided in Table
1. Two studies in this table investigated multiple species (Shams et al.
2006; Millett and Stamm
1954). Shams et al. (
2006) compared intermediate sized specimens of eight hardwoods and softwoods that were impregnated with PF-resin and subsequently densified. It was noted that the swelling capacity of the cell wall, resulting from PF-modification, differed between species. Since this study focussed on densification the authors did not conduct ASE tests. Millett and Stamm (
1954) compared plank sized sapwood specimens of nine, mainly North American, wood species. Due to the specimen dimensions, the study investigated the treatability of different wood species rather than the influence of their anatomy and chemistry on the treatment. Accordingly, the quality of modification was assessed through the uniformity of the treatment comparing the swelling coefficient, density, and surface hardness in different locations along the plank. Yellow birch (
Betula alleghaniensis), sweetgum (
Liquidambar styraciflua), and tulipwood
(Liriodendron tulipifera) showed uniform swelling coefficients across the whole length of the plank. Tulipwood also displayed a uniform density after treatment and consistent ball indentation hardness across the whole length.
Table 1
Summary of ASE studies comparing different timbers modified with thermosetting resins: s.c. = solids content of resin
Acer saccharum | 7 × 7 × 180 | 20 (PF) | – | – | Pressure imp. 14 bar and 5 h | Diffusion 2 days at ambient temp. drying 14 days and < 70 °C, | heat cure with hot plate and 155 °C | 15.7 | – | 37–53 | 1 |
Betula alleghaniensis | 23.8 | – | 42–59 |
Liquidambar styraciflua | 23.2 | – | 63.5–70 |
Liriodendron tulipifera | 30.8 | – | 52–61 |
Populus section Aigeiros | 31.5 | – | 57.5–63.5 |
Pseudotsuga menziesii | 11 | – | 13.5–50.5 |
Pinus ponderosa | 26.6 | – | 37.5–57 |
Picea sitchensis | 19.5 | – | 31–65 |
Anthocephalus cadamba Miq | 1 × 0.5 × 8 | 30 (PF) | – | 9.5 | Vacuum imp | – | Heat cure 100 °C | 33.7 | 14.45 | 68 | 2 |
Cryptomeria japonica | 3 × 3 × 0.5 | 30 (PF) | 309–335 | 6.7–10.3 | Vacuum imp. 1 h | Diffusion submerged in resin 1 day | Heat cure in oven and gradual from 60 to 180 °C | – | – | 60 (any pH) | 3 |
30 (PF) | 820 | 10 | – | – | 10 (since higher MW) |
Paraserianthes falkata | 10 × 6 × 0.6 | 20 (PF) | 300 | 5.5 | Vacuum imp. 12 h repeat 7× | Diffusion 2 days and ambient, drying 50 °C vacuum | Subsequent densification and cure | 48.9 | 5.2 | – | 4 |
Shorea sp. | 72.4 | 7.8 | – |
Cryptomeria japonica | 43 | 8 | – |
Picea abies | 47 | 7.8 | – |
Pseudotsuga douglasii | 34.7 | 8.9 | – |
Ulmus sp. | 42.3 | 13.5 | – |
Fagus crenata | 33.7 | 12.4 | – |
Betula ma×imowicziana | 28.1 | 16.3 | – |
Pinus sylvestris | 2.5 × 2.5 × 1 | 30 (PF) | 400 | 9.6 | Vacuum pressure imp. 12 bar and 2 h | – | Drying and cure gradual from 20 to 103 °C | 50 | ~ 13 | – | 5 |
Betula pendula | 2 × 2 × 2 | 5–15 (PF) | 460 | – | Vacuum imp. 1 h | Diffusion submerged in resin 2 h; drying gradual | Heat cure 140 °C | 6 (low conc)–14 (high conc) | 2.2 (low conc)–5.2 (high conc) | 30 (low conc)–40 (high conc) | 6 |
Fagus sylvatica | 2.5 × 2.5 × 1 | 9–27 (PF) | 297–854 | – | Vacuum imp. 45 min | Diffusion submerged in resin 1 h; drying gradual 25–103 °C | Heat cure 140 °C | 8 (any MW, low conc)–23 (any MW, high conc) | 6 (high MW, low conc)–14 (low MW, high conc) | – | 7 |
Pinus massoniana | 2 × 2 × 2 | 15–30 (PF) | 300 | – | Vacuum pressure imp. 8 bar and 2 h | Drying at ambient | Heat cure 130 °C 2 h | 20 (low conc)–38 (high conc) | – | 26 (low conc)–54 (high conc) | 8 |
Pinus sylvestris | 2.5 × 2.5 × 1 | 10–25 (MF) | 840 | 10.1 | Vacuum imp. 1 h | Diffusion submerged in resin 1 h | Drying and heat cure gradual 25–103 °C | 14 (low conc)–30 (high conc) | 2.5 (low conc)–5 (high conc) | – | 9 |
2 Material and methods
2.1 Wood samples and PUF-resin
Seven hardwoods and four softwoods were selected for this study. The hardwoods were European beech (Fagus sylvatica), silver birch (Betula pendula), sycamore (Acer pseudoplatanus), willow (Salix alba), common poplar (Populus tremula), European lime (Tilia x europaea) and tulipwood (Liriodendron tulipifera). The softwoods included Southern yellow pine or SYP (this is a commercial mix of four pine species: Pinus palustris, P. elliottii, P. taeda and P. echinata), Scots pine (Pinus sylvestris), Radiata pine (Pinus radiata) and Pinus taeda. The Pinus taeda timber was procured from a single species plantation in Brazil for comparison with the mixed species SYP from North America. All timbers were sourced commercially, with requirement for low heartwood content. Visual examination on receipt and use of heartwood indicator solutions where relevant confirmed that sapwood was used in this study. Sapwood specimens from each species were cut with good alignment of the grain and growth rings to the edges of the specimen, the dimensions were 20 (r) × 20 (t) × 5 (l) mm. These were stored at ambient conditions until further processing. Every unmodified and modified test group in this study consisted of 10 specimens from the same board.
Commercial PUF-resin (Prefere 5K600M) was provided by Prefere GmbH, Germany. Prior to impregnation, the resin was diluted with deionized water to a solids content of 30% and pH 9.2. The molecular weight information as reported by Prefere is Mn = 406 (g/mol) and Mw 484 (g/mol). Potassium hydroxide (KOH) is present as an alkaline catalyst.
2.2 Resin treatment
The sample mass was measured with a 4 d.p. balance (Ohaus Explorer Analytical) and the surface area was measured with a digital calliper (2 d.p.) at the radial, tangential, and longitudinal edge of the specimen. The oven dry mass and dimensions of unmodified test blocks were determined after drying at 105 °C for 24 h (OD 0). Subsequently, samples were conditioned at ambient moisture and room temperature for 24 h prior to use. For the resin impregnation, samples were placed in a 250 ml beaker in a desiccator attached to a vacuum line. A vacuum was drawn for 20 min prior to impregnation. Then, the resin was injected through a dropping funnel until all samples were fully submerged. Ballast was used above the samples to ensure they remained below the level of the fluid throughout the procedure. A volume of 100 ml resin was used for each impregnation. Still under vacuum, the samples were immersed for 20 min, before the vacuum was released. Subsequently, the treatment solution was drained off and samples were removed from the beaker, excess resin was blotted with tissue and the mass and dimensions were noted (Imp). The samples were now carefully dried at 50 °C for 16 h to reduce the moisture content before cure. Heat curing took place at 150 °C for a duration of 8 h. The mass and dimensions after cure were noted (OD 1).
2.3 Cyclic swelling and drying
Sets of 10 modified and 10 unmodified control specimens were subjected to cyclic water soaking and oven drying. Oven dry mass and dimensions of control (OD 0) and modified groups (OD 1) were determined, as described previously. The water soaking was carried out by vacuum impregnating each test group with 100 ml deionized water, using a similar procedure to that described for resin treatment, within a beaker in a desiccator. The soaked samples remained submerged in water for 5 days at room temperature. Subsequently, excess water was removed with a tissue before mass and dimensions were noted (WS 1). Water-soaked samples were then dried at 50 °C for 16 h to eliminate moisture without initiating stresses in the samples, and then at 105 °C for 24 h. The mass and dimensions were noted (OD 2) and the procedure was repeated until three wetting and drying cycles were completed.
2.4 ASE soaking solution analysis
After each soaking cycle, the amount of leached substances was estimated by measuring the solid residue in solution by evaporation. The nature of leached substances was investigated by measuring the pH and UV absorbance at 272 nm of the leachate.
The soaking solution was transferred to a volumetric flask and diluted to a volume of 1 l, to ensure the concentration and UV data could be used for quantitative comparison. The high level of dilution was necessary to obtain meaningful results in the UV measurements within the range of the UV spectrometer used. Approximately 20 ml of the diluted solution was poured in a glass vial of known dry mass for evaporation at 105 °C. The solid residue
\(SR \left[\%\right]\) was determined by dividing the dry mass with the liquid mass. The solid residue in percent was then converted to a volumetric concentration
\(SR \left[\frac{g}{l}\right]\):
$$SR \left[\%\right] = \frac{{m}_{solid}}{{m}_{liquid}} *100\%$$
$$SR \left[\frac{g}{l}\right]= SR \left[\%\right]*10 \left[\frac{g}{l}\right]$$
A small aliquot of the diluted soaking solution was transferred to a quartz cuvette for UV absorption measurement at 272 nm. This wavelength is characteristic for aromatic compounds and displays the maximum absorption peak for phenolics (Pretsch et al.
2000). Approximately 50 ml diluted soaking solution was transferred to a small beaker for pH measurement. The pH meter was calibrated with standard buffer solution at pH 4 and 7 prior to use. The procedure was performed for every test and control group of every investigated species.
2.5 Calculations of resin treatment and cyclic swelling results
The liquid resin uptake describes the mass change between the oven dry and the freshly impregnated state. The WPG describes the dry weight gain achieved after cure. The
BU describes the dry volume gain after cure. They are calculated as follows:
$$Liquid\, resin\, uptake = \frac{{m}_{Imp}-{m}_{OD 0 }}{{m}_{OD 0 }}*100\%$$
$$WPG= \frac{{m}_{OD 1}-{m}_{OD 0 }}{{m}_{OD 0 }}*100\%$$
$$BU= \frac{{(r*t)}_{OD 1}-{(r*t)}_{OD 0 }}{{(r*t)}_{OD 0 }}*100\%$$
The water-induced swelling coefficient of unmodified control groups
\({S}_{control}\) relates to the swelling of oven dry unmodified data (
OD 0) and the swelling coefficient of the modified groups
\({S}_{mod}\) uses the oven dry treated data (
OD 1). Within this study, the swelling coefficient was calculated on an area basis, to minimise error arising from longitudinal dimension data. They are calculated as follows (Ohmae et al.
2002; Hill
2006):
$${S}_{control}= \frac{{(r*t)}_{WS 1}-{(r*t)}_{OD 0}}{{(r*t)}_{OD 0}}*100\%$$
$${S}_{mod}= \frac{{(r*t)}_{WS 1}-{(r*t)}_{OD 1}}{{(r*t)}_{OD 1}}*100\%$$
$$ASE= \frac{{mean(S}_{Control})-{mean(S}_{mod})}{mean({S}_{Control})}*100\%$$
The resin-induced swelling coefficient
\({S}_{resin}\) of modified groups described the volume gain during impregnation and relates to
OD 0. It is calculated as follows:
$${S}_{resin}= \frac{{(r*t)}_{Imp}-{(r*t)}_{OD 0}}{{(r*t)}_{OD 0}}*100\%$$
The total water-induced swelling coefficient
\(TS\) describes the volume change from the state
OD 0 to
WS 1 for both modified and unmodified groups. In modified groups, this coefficient includes the permanent volume change by bulking. In unmodified control groups
\(TS\) is identical with
\({S}_{control}\).
$$TS= \frac{{(r*t)}_{WS 1}-{(r*t)}_{OD 0}}{{(r*t)}_{OD 0}}*100\%$$
The relative volume
\({V}_{rel}\) in the cyclic swelling experiments refers to
OD 0 in the control groups and to
OD 1 in the modified groups. As mentioned previously, the changes in longitudinal direction are negligible in the displayed calculation of
\({V}_{rel}\). Therefore, the relative area calculation is assumed to yield values equivalent to the relative volume, while excluding the longitudinal direction as a source of error. It is calculated as follows:
$${V}_{rel i}= \frac{{\left(r*t\right)}_{i}}{{\left(r*t\right)}_{OD 0 or 1}} i=O{D}_{0}, O{D}_{1}, W{S}_{1}, O{D}_{2}, W{S}_{2}, O{D}_{3}, W{S}_{3}$$
The water uptake
\({WU}_{i}\) describes the mass gain from each oven dry state
OD i to the following water-soaked state
WS i as a percentage. It is calculated as follows:
$${WU}_{i}= \frac{{m}_{WS i}-{m}_{OD i }}{{m}_{OD i }}*100\%\; i=0, 1, 2, 3$$
The mass loss during ASE tests
\({ML}_{total}\) compares the oven dry mass of the last cycle with first one. In modified groups, the first cycle starts at
OD 0 and in modified groups it starts at
OD 1. It is calculated as follows:
$${ML}_{total}= \frac{{m}_{ OD 3}-{m}_{OD 0 or 1 }}{{m}_{OD 0 or 1 }}*100\%$$
2.6 Statistical analysis
The experimental set-up consists of two variables. Variable one describes whether a test group is modified or unmodified. Variable two is the wood species. Every wood species was tested in a modified and unmodified state giving a total of 22 test groups each containing 10 specimens.
All analyses were performed using R Statistical Software (v4.1.2; R Core Team 2021). Boxplots used in several figures show the median value, standard deviation, as well as maximum and minimum values in each test group. Linear regression models have been calculated using the R functions stat_regline_equation() and stat_cor(). Two-way analysis of variance (ANOVA) was carried out to discern significant differences at a 95% level of confidence between swelling, total swelling, and total mass loss of modified and unmodified wood. Extreme outliers have been removed from the data sets prior to ANOVA analysis using the R function identify_outliers(). ANOVA was performed using the R functions aov(). A Tukey Honest Significance Difference (HSD) test was performed on the same data to identify significant differences within each wood species. The R function TukeyHSD() was used.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.