Chemical Composition of Larch Oleoresin before and during Thermal Modification

Abstract

Larch is a strong timber, which grows rapidly in the UK climate, but can contain abundant resin pockets. To address the resin exudation issue, a mild thermal modification process has been developed, promoting the curing of the resin. This paper reports a series of studies which characterised the chemical profile of larch oleoresin before and after the mild thermal treatment, explaining the changes which occur when resin is dried. Further experiments were used to simulate specific points in time during the mild treatment process. The non-polar components of the fresh (untreated) and treated larch oleoresin were profiled using gas chromatography mass spectrometry (GC-MS). Fresh larch oleoresin was also subjected to isothermal experiments at different temperatures in a thermogravimetric analyser–differential scanning calorimeter (TGA/DSC), followed by re-analysing the resin composition. This demonstrated the loss of monoterpenes at temperatures of 120 °C and above, with complete loss by isothermal conditions of 150 °C and 60 min. The partial loss of sesquiterpene alkanes and alkenes were also observed at all temperatures, although completeness of this loss was achieved at isothermal temperatures of 150 °C and above. The diterpene composition was seen to change for isothermal experiments conducted at 150 °C and above, with a dehydration of terpenols to form the equivalent terpene alkenes. The observed physical changes in the TGA/DSC experiment were in good agreement with observations of the oleoresin sampled from thermally modified larch planks.

1. Introduction

Japanese larch (Larix kaempferi Lamb.) is a strong durable timber, which grows rapidly in the UK climate, yet has a higher density, and higher tensile and bending strength than Sitka spruce and Scots pine. While slow-grown European larch and Siberian larch from Alpine or Scandinavian forests is imported to the UK for use in cladding and joinery, the faster-grown material from the UK has often occupied the garden fencing and rough-sawn timber market due to perceived properties relating to its machinability and fast growth rate. In addition, fast-grown larch is prone to resin pockets, which may lead to issues in processing. The outbreak of Phytopthera ramorum, which has led to the widespread felling of larch in the UK is also related to the high resin content in the timber [1].

Thermal modification of UK-grown larch timber has shown potential to reduce issues in machining [2], and to ‘cure’ the resin, leaving dry deposits on the timber surface or within resin pockets, which are relatively easily machined and leave less residue on the planer blades [3]. The mild thermal modification conditions employed mean that only a limited onset of chemical changes to the structural wood components (cellulose, hemicellulose, and lignin) are observed [4]. In this paper, the chemical changes which give rise to the ‘drying’ process that the larch oleoresin undergoes is considered.

It is well known that the kiln-drying of softwoods, and therefore also the thermal modification of softwoods, results in the liberation of terpenes and other volatile organic compounds (VOCs) at elevated temperatures [5,6,7,8,9]. It is increasingly recognised that the control of terpene emissions from timber-drying kilns and other drying operations requires consideration [10]. Studies by Bannerjee and co-workers have shown that the evolution of monoterpenes versus sesquiterpenes is influenced by the moisture content of and the quantity of moisture being driven off from the wood [11,12]. During the thermal modification of wood, various extractive components of wood are redistributed, and the terpene content of softwoods following thermal modification, such as in the Thermowood process, is reduced while other volatile compounds may have increased [8,13,14,15]. In thermal modification at high temperatures (usually 180 °C and above), such as in the Thermowood process, degradation of hemicellulose contributes to the formation of new volatile compounds, and the study of this process is frequently reported [4,8,15,16]. The mild modification system described above uses temperatures closer to 170 °C, so it does not result in the same degree of conversion of hemicellulose [3,16], and negligible change in lignin would be expected at this temperature [4]. The mild modification may therefore demonstrate changes in terpene profile, which are intermediate between those reported for high-temperature kiln-drying and full thermal modification processes. As a result, the chemical composition of oleoresin and extractives in mild thermally modified wood are likely to reveal interesting trends with potential relevance to kiln-drying and modification systems.

Oleoresin is present in several species of coniferous timber, and is principally distributed in resin canals that run longitudinally and radially throughout the wood. Its production can be enhanced as a result of injury, disease or insect attack, and in these cases, resin pockets may increase in prominence. Conifer resins are known to contain a range of terpenes [17] and have been widely used in the art industry for many years, both as turpentine and as colophony or rosin, the higher-molecular-weight solid which remains after the distillation of the oleoresin. As a result, some of the data relating to pine or larch resin composition have been determined to allow for the identification of rosin in lacquers and other artefacts [18,19]. Venice turpentine was formed from European larch oleoresin, while many rosins were sourced from pines. The term ‘rosin’ will be used in this paper to refer to dried resin, from which the more volatile compounds have evaporated, while resin will be used to refer to fresh or pliable oleoresin from wood.

Monoterpenes, such as the pinenes, comprise two isoprene units and a structure containing ten carbon atoms, while sesquiterpenes contain three isoprene units and so are based on a fifteen-carbon skeleton (Figure 1). Many diterpenes, for example, are based on a pimarane skeleton and 20 carbons, with three six carbon rings and various substituents. Migration of the single double bonds around the ring structures during formation permits the biosynthesis of many closely related but different molecules. Terpenoid molecules may have a large number of skeletal forms—at least 38 are known for the monoterpenes, increasing to over 200 for the sesquiterpenes [20]. The terpenes and terpenoids may have alkane, alkene, aldehyde, ester, alcohol or carboxylic acid functionalities. Two examples of monoterpenes, sesquiterpenes, and diterpenoids are shown in Figure 1. The commonly occurring α- and β-pinenes are monoterpenes, with the formula C₁₀H₁₆ due to the two-ringed structure and alkene functionality.

The natural larch oleoresin contains various mono-, sesqui-, and di-terpenoids, as well as some alkanes and fatty acids [17,21]. The larch monoterpenes tend to be alkene in structure, e.g., α- and β-pinenes and Δ-3-carene; similarly, the sesquiterpenes present tend to be alkenes, e.g., germacrene D and longifolene. The diterpenes in larch resin include alkenes, alcohols, and aldehyde functional groups, as well as the resin acids such as abietic acid, dehydroabietic acid, pimaric acid, and palustric acid [17,22]. The larch diterpenes also contain the characteristic larixyl acetate and its free diol larixol, or torulosol acetate and epi-torulosol, depending on species [18,23]. The larixol and its acetate occur in Larix decidua and Larix gmelini, while epi-torulosol and its acetate are detected in the species where larixol is absent, such as Larix kaempferi. Hybrid larch (Larix X. eurolepis Henry) contains components of both types as an intermediate between its parent species. Larix kaempferi is reported to have a similar composition to Larix laricina ((DuRoi) K.Koch.) (tamarak) and the other North American species, but a higher quantity of thunbergol [18].

This study reports a set of chemical profiling experiments that compare the larch resin before and after thermal modification under different conditions. The aim of this study was to explain the changes which occur when resin dries during the thermal modification process. The composition of oleoresin from fresh and thermally modified larch timber was observed using gas chromatography–mass spectrometry (GC-MS). In a second study, DSC experiments were conducted to expose fresh oleoresin to different isothermal temperature sequences in a thermogravimetric analyser–differential scanning calorimeter (TGA/DSC). The composition of the material generated by TGA/DSC experiments was also evaluated.

2. Materials and Methods

2.1. Thermal Modification

Japanese larch (Larix kaempferi) harvested in Wales, UK, was thermally treated in a pilot-scale kiln by Coed Cymru according to a three-stage thermal treatment procedure. This used a ramp phase, a high-temperature hold phase, and a conditioning phase during which the temperature of the kiln decreases. A pre-treatment step was used to initiate drying—this is referred to as Day 1 in later sections—and had a target temperature of 120 °C. During the high-temperature phase, the maximum kiln temperature was 190 °C, but the internal temperature of the timber lagged below this. Thermocouple data confirmed the wood core to have been above 170 °C for between 2 h and 4.5 h [2]. The mild thermal treatment was achieved with 2–3 h above 170 °C, while the moderate thermal treatment required longer duration with the wood core at a temperature of 170 °C or above. The treatment was developed to ensure a process which enhanced machinability, rather than targeting dimensional stability or durability aspects, as are commonly the fixed objective of commercial systems; the treatment details therefore differ from the Thermowood or Retiwood processes, which are discussed elsewhere [2,24].

The treated planks were 110 mm by 30 mm in cross-section, and 90 cm in length, cut from longer planks which had been air-dried outdoors under cover. The moisture content of the planks prior to treatment was 18.04%, determined by the removal of a 2 cm moisture content block from within the length of each longer plank while preparing the 90 cm length planks for treatment (average of 30 samples). A further three moisture content samples were taken from selected planks and used for moisture content determination after treatment. These moisture content samples were also used to evaluate the weight change due to the thermal treatment process, based on calculated oven dry weight per plank before and after treatment. The apparent weight loss was 2.2 to 2.6%.

Planks were stacked in a pilot-scale kiln, in a stack that was seven planks wide by seven planks tall, and good air circulation was forced by the use of a baffle above the stack to direct the heated air through the stack. Thermocouples were drilled into five planks within the stack at the centre of the top and bottom layers of planks, and three in the central layer (left, centre, and right). Two additional thermocouples were stapled onto the surface of the planks on the left and right of the kiln stack, providing paired internal and surface temperature data for the planks nearest to and furthest from the hot air infeed on the right-hand side of the oven.

2.2. Gas Chromatography–Mass Spectrometry of Oleoresin

Resin was sampled from resin pockets identified in untreated and treated planks. These resin pockets were exposed in the sawn face of the untreated planks, and as the planks were recently cut, the resin was a viscous liquid which could be sampled using a spatula for the analysis. In the case of thermally modified planks, resin pockets were exposed by cross-cutting for sampling.

Approx. 10 mg of resin was dissolved in the selected solvent for gas chromatography–mass spectrometry analysis (GC-MS). Heptane was used as solvent for the first run, as this is suitable for a broad range of non-polar molecules present in larch resin. The dissolution was repeated using dichloromethane and with methanol to increase the polar fractions represented in the chromatograms; however, these changes were relatively small.

GC-MS was carried out on a Perkin Elmer Clarus 600 chromatograph fitted with a VF5 column. The initial oven temperature was 60 °C, held for one minute prior to increasing the oven temperature at a rate of 6 °C/min to 300 °C, where the temperature was held for 10 min. Mass spectra were analysed using TurboMass 6.10 software and compared with reference spectra from the NIST 2011 and Adams spectral libraries and selected relevant publications (e.g., [25]) to verify the identification of the dominant components within the chromatograph.

2.3. Fourier Transform Infrared (FTIR) Spectroscopy

Resin from the untreated and treated planks was also observed using FTIR spectroscopy in attenuated total reflectance (ATR) mode with a Gladi-ATR (PIKE Technologies, Madison, WI, USA) and Thermo 8000 FTIR spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). The oleoresin was smeared directly on the diamond crystal, and the ATR surface was cleaned using the solvent between each sample of resin. Wood and dried rosin were sampled from resin pockets in the treated wood using a sharp blade to provide a flat surface, which was pressed directly against the diamond crystal, allowing the ATR spectrum to be recorded.

2.4. Temperature Simulations Using TGA/DSC

Small samples of untreated resin were heated under controlled conditions using a Mettler Toledo TGA/DSC 1 STAR^e System, under constant flow of nitrogen gas (50 cm³/min). The resin samples were placed in open 40 μL aluminium crucibles to allow for evaporation to occur during heating. A ramp rate of 10 °C/min was used to heat the resin to the selected isothermal temperature, which was held for a 60-minute hold period before cooling. Four different profiles were investigated, with isothermal temperatures at 120 °C, 150 °C, 170 °C, and 190 °C. These isotherms were selected to reflect significant temperature steps within the thermal modification process. Resin samples from these experiments were also subjected to GC-MS analysis.

3. Results

3.1. FTIR Spectra

Samples of fresh larch resin were observed, and the typical spectrum is presented in Figure 2, alongside the spectrum for the larch earlywood. The oleoresin is a mixture of many terpenes, but it is clear that the resin spectrum contains compounds containing a hydrocarbon structure with some sp³ CH units in addition to alkane and alkene structures (absorptions at 2843 to 2925 cm⁻¹). There are also carboxylic acids and alcohols present, as evidenced by strong absorptions at 1690 and 1638 cm⁻¹ [26]; however, these occur within a broad peak (1730–1550 cm⁻¹) where many absorptions overlap, possibly relating to other carbonyl absorptions such as aldehydes or alcohol absorptions, which occur within a similar range.

Many traits within the spectrum for the oleoresin were seen to correlate with the spectra for α- and β-pinenes, such as the small absorption at 3078 cm⁻¹, the dominant 2929 cm⁻¹ peak, and the pair of alkane bend absorptions at 1450 and 1373 cm⁻¹; while other aspects, e.g., the 1690 cm⁻¹ and 1638 cm⁻¹ carboxylic acid absorptions and 1448, 1366, 1262, 1240 and 1149 cm⁻¹ fingerprint absorptions, correlated with the spectra for diterpene-based plant rosins such as pine rosin, sandarac, and copal [19].

Rosin was sampled from within dried resin pockets found within treated planks for FTIR-ATR spectral determination (Figure 3). This revealed a reduction in several strong absorptions previously present in the fresh resin. For example, the strong CH stretches at 2925 and 2863 cm⁻¹ were reduced, and peaks within the carbonyl and alcohol region decreased (1750–1550 cm⁻¹), indicating the loss or conversion of resin components. The relative proportion of hydroxyl appeared to increase, as the broad absorption centred on 3300 cm⁻¹ relating to OH stretches became more prominent. This would be consistent with the loss of alkene dominated structures such as the monoterpenes, which are expected to evaporate in large quantities from the oleoresin.

3.2. GC-MS for Oleoresins from Untreated and Thermally Treated Wood

The composition of fresh larch resin, and samples of larch resin and rosin collected from planks subjected to mild thermal modification, are presented in Figure 4 and Figure 5, as well as Table 1.

The components observed in the chromatogram for the unmodified larch resin (Figure 4) were broadly in agreement with the profiles observed by previous authors [18]. The main three monoterpenes identified were α-pinene (retention time = 5.13 min), β-pinene (rt = 6.04 min), and Δ-3-carene (rt = 7.16 min), all of which were seen before the 10 min elution time. The main sesquiterpenes were longifolene, γ-elemene, germacrene D, β-cadinene, and germacrene B. These were seen between 15 and 19 min, and all were present in relatively low abundance. The diterpenes were identified at higher retention times, above 25 min; this region contained a mixture of many compounds, including diterpenes (thunbergene), diterpene alcohols (e.g., epimanool, isopimarol, torulosol), acetates of some diterpene alcohols (e.g., torulosol acetate), and diterpene resin acids (palustric acid, abietic acid, neoabietic acid).

Table 1. Oleoresin composition for freshly cut larch and for larch after mild thermal modification process.

Compound	Larch Oleoresin	Oleoresin from Mild Thermally Modified Wood
α-pinene	dominant	medium
β-pinene	large	small
camphene	minor	absent
Δ-3-carene	medium	absent
d-limonene	minor	absent
l-α-terpineol	minor	minor
α-terpineol acetate	minor	absent
longifolene	minor	absent
γ-elemene	minor	minor
one of the germacrenes	minor	absent
germacrene D	small	absent
β-cadinene (or δ-cadinene)	minor	absent
germacrene B	minor	absent
thunbergol	medium	absent
epi-manool	medium	present—strong
isopimara 7,15-dien-3-one	small	minor
isopimara 7,15-dienol?	medium	minor
a compound peak mixture	medium	one component still present
larixol	medium	absent
torulosol	small	strong
dehydroabietol or neoabietinol	small	minor
palustric acid	large	absent
torulosol acetate	small	medium—large
isopimaric acid (?)	small	minor
abietic acid	small	minor
neoabietic acid	small	absent
thunbergene	not present	present—medium
pimaradiene	not present	present—small
benzenedicarboxylic acid, ester	not present	(33.76)
unidentified	not present	(35.29)
squalene (C30H50)	not present	present—dominant peak (37.08)

Table 1. Oleoresin composition for freshly cut larch and for larch after mild thermal modification process.

Compound	Larch Oleoresin	Oleoresin from Mild Thermally Modified Wood
α-pinene	dominant	medium
β-pinene	large	small
camphene	minor	absent
Δ-3-carene	medium	absent
d-limonene	minor	absent
l-α-terpineol	minor	minor
α-terpineol acetate	minor	absent
longifolene	minor	absent
γ-elemene	minor	minor
one of the germacrenes	minor	absent
germacrene D	small	absent
β-cadinene (or δ-cadinene)	minor	absent
germacrene B	minor	absent
thunbergol	medium	absent
epi-manool	medium	present—strong
isopimara 7,15-dien-3-one	small	minor
isopimara 7,15-dienol?	medium	minor
a compound peak mixture	medium	one component still present
larixol	medium	absent
torulosol	small	strong
dehydroabietol or neoabietinol	small	minor
palustric acid	large	absent
torulosol acetate	small	medium—large
isopimaric acid (?)	small	minor
abietic acid	small	minor
neoabietic acid	small	absent
thunbergene	not present	present—medium
pimaradiene	not present	present—small
benzenedicarboxylic acid, ester	not present	(33.76)
unidentified	not present	(35.29)
squalene (C30H50)	not present	present—dominant peak (37.08)

The typical profile of terpenes changed in the thermally treated material, with α-pinene content falling to 6%–11% from 21% in untreated material, and β-pinene content falling to 0.6%–1.1% from 7% (Table 2, Figure 5). The proportion of sesquiterpenes also decreased, while the proportion of all diterpene compounds increased, as these form the greatest component of the residue. These changes would reflect the expected evaporation of mono- and sesquiterpenes due to low boiling points. Differences were also observed between the dry rosin and the liquid resin from further inside the same treated plank; however, the reduction in α- and β-pinene content was very small (Figure 5 middle and lower traces, Table 2).

Figure 5. Gas chromatograph for untreated resin (upper trace), and resin sampled from low-temperature thermally modified wood (middle and lower traces). Middle trace is dry rosin from the surface of the treated wood and lower trace is liquid resin samples from within the resin pocket of the same plank (TM CC moderate, Table 2). The upper trace was run on a slightly longer column resulting in an approximately 12 s offset compared to the treated wood. The epimanool peak in the middle trace is at 27.33, and in upper trace is at 27.21.

Figure 5. Gas chromatograph for untreated resin (upper trace), and resin sampled from low-temperature thermally modified wood (middle and lower traces). Middle trace is dry rosin from the surface of the treated wood and lower trace is liquid resin samples from within the resin pocket of the same plank (TM CC moderate, Table 2). The upper trace was run on a slightly longer column resulting in an approximately 12 s offset compared to the treated wood. The epimanool peak in the middle trace is at 27.33, and in upper trace is at 27.21.

Table 2. Proportion of α- and β-pinenes relative to total terpene content. Sesquiterpenes and diterpenes are presented as an aggregate figure due to variability of relative proportions of compounds within these groups.

	α-Pinene	β-Pinene	Sum of Sesquiterpenes	Sum of Diterpenes
Untreated CC	20.67%	6.66%	2.37%	70.30%
TM CC mild	6.41%	1.14%	2.51%	89.9%
TM CC moderate, liquid resin	10.71%	0.67%	0.66%	87.96%
TM CC moderate, dry resin	9.48%	0.60%	0.98%	88.94%
Untreated CM	35.19%	2.94%	3.04%	58.84%
TM high CMT	31.7%	3.64%	4.52%	60.14%

Table 2. Proportion of α- and β-pinenes relative to total terpene content. Sesquiterpenes and diterpenes are presented as an aggregate figure due to variability of relative proportions of compounds within these groups.

	α-Pinene	β-Pinene	Sum of Sesquiterpenes	Sum of Diterpenes
Untreated CC	20.67%	6.66%	2.37%	70.30%
TM CC mild	6.41%	1.14%	2.51%	89.9%
TM CC moderate, liquid resin	10.71%	0.67%	0.66%	87.96%
TM CC moderate, dry resin	9.48%	0.60%	0.98%	88.94%
Untreated CM	35.19%	2.94%	3.04%	58.84%
TM high CMT	31.7%	3.64%	4.52%	60.14%

The components present in the diterpene fraction of the resin were not static and appear to have been altered during heating. There was evidence of a thermal conversion of the thunbergol and isopimaradienol into thunbergene and pimaradiene, with a reduction in the peaks at 27.18 and 30.02 min, and the creation of new peaks at 25.27 and 25.50 in the oleoresin sampled from treated wood (Figure 5). This dehydration reaction has been observed in laboratory studies on pure terpenes [27] and in high-temperature reactions during the distillation of tall oil rosins [22], but has not been previously reported in wood modification. The yield of thunbergene (25.27 min) was greater than that of pimaradiene (25.50 min) in the liquid fraction of the sampled material (lower trace of Figure 5), which corresponds with the near disappearance of the thunbergol peak (expected at 27.18 min) from the chromatograph, while the neighbouring epimanool peak (27.33 min) was still present. The newly formed pimaradiene is believed to have been formed from the isopimaradienol; there was a strong reduction in this peak, at 29.90 min in the upper trace, and minor in the lower trace (30.02 min).

The identification of torulosol and torulosol acetate was not possible using the NIST database; however, the structure of torulosol was inferred from the fragmentation pattern of an unidentifiable component (eluted at 31.53 min) with reference to [28]. Additional samples of larch from other planks used by the project team from other sources revealed that the 31.53 min peak was dominant in some, while the larixol peak (31.46 min) was dominant in others. It has been reported that the dominance of torulosol relates to Japanese larch, which is the focus of this study, but that in hybrid larch (which is also grown in Wales) both torulosol and larixol may be identified [18]. Rosin from hybrid larch which had been thermally modified was also evaluated using GC-MS; however, for simplicity, the results will not be presented here. The majority of changes seen in the oleoresin of Japanese larch after thermal modification were also observed in oleoresin from hybrid larch when exposed to the same thermal treatment process.

3.3. TGA/DSC Temperature Simulations

The weight change of the resin samples at each stage in the thermal profile was calculated from the TGA/DSC outputs. The experiments were run on four different untreated resins from different larch planks, and consistent results were seen for all resin sources. In all cases, weight was lost during the ramp stage, and this increased as the isotherm temperature increased—relating to the additional ramp time. The quantity of resin weight lost during the isotherm period increased with an increasing set point, as would be expected (

Table 3. Weight loss on the ramp and the isotherm within DSC runs (1 h at 120 °C, 150 °C, 170 °C or 190 °C); --- indicates experiment at this isotherm was not conducted for this resin sample.

		120 °C Isotherm	150 °C Isotherm	170 °C Isotherm	190 °C Isotherm
Untreated Resin A	ramp	-	−2.3270%	−3.6629%	-
	isotherm	-	−12.5246%	−23.1080%	-
Untreated resin CC	ramp	−1.0444%	−2.0191%	−7.1936%	−10.5407%
	isotherm	−9.5621%	−14.8054%	−31.0611%	−48.5842%
Untreated resin D1	ramp	−3.0689%	−3.1069%	−7.3155%	−11.9764%
	isotherm	−8.2138%	−15.5338%	−27.6076%	−60.5482%
Untreated resin D2	ramp	−2.6315%	−4.8903%	−6.4693%	-
	isotherm	−8.1963%	−11.3445%	−28.0898%	-

). The total weight loss doubled on increasing the isotherm temperature from 150 °C to 170 °C (16% and 38% weight loss, respectively), with a further increase in loss seen on increasing the isotherm temperature to 190 °C (59% weight loss). None of the isotherms reached a stable weight during the 60-minute hold period, although the 120 °C isotherm showed the greatest sign of approaching a plateau at an 11% weight loss.

It was observed that weight loss followed a gradient, relating to temperature and duration, or ramp rate (Figure 6), with a corresponding intake of energy, but no distinct endotherms relating to single components were seen. This implies that evaporation of the resin components occurs as a mixture rather than by an evolution of distinct components or fractions at their boiling points during the temperature profile. Neither was there much indication of a separate evolution of water vapour at 100 °C. The strong interaction between monoterpenes and moisture is known [11,12], and an azeotropic behaviour of mixtures of terpenes with water can be expected.

The residue which remained in the crucibles at the end of each TGA/DSC run was prepared for GC-MS analysis (Figure 7). In the case of the 170 °C isotherm, which will be presented later, the weight of residue remaining was very small, and peaks were identified by elution time, and not mass spectrum. Further reduction in residue weight for the 190 °C sample led to a chromatograph in which the compounds were approaching the detection limit, which prevented analysis.

When larch resin A was heated to 120 °C using the TGA/DSC to simulate a Day 1 heating cycle, the GC/MS trace showed a clear reduction in monoterpenes (Figure 7). The β-pinene and Δ-3-carene peaks (6.04 and 7.16 min) were not present, and only a very small α-pinene peak (5.13 min) remained. Similarly, the sesquiterpene peaks normally observed at 10 to 21 min were greatly reduced, although trace amounts of the main sesquiterpenes were observed. The diterpenes were still present; for example, both thunbergol and epimanool (27.06 min, 27.21 min) are visible, and palustric acid appears the dominant component at 32.17 min. The great reduction in α-pinene peak and disappearance of the β-pinene and Δ-3-carene peaks during the 120 °C isotherm experiment clearly supports the loss of these monoterpenes well below their boiling points (at 155, 163, and 170 °C, respectively) probably within an azeotropic mixture with water vapour.

Samples of the same resin were exposed to isothermal temperatures of 150, 170, and 190 °C (Figure 8). For these samples, both the monoterpenes and sesquiterpenes were completely absent. In the 150 °C sample, there was a change in the pair of peaks relating to thunbergol and epimanool, with the thunbergol (27.16 min) being greatly reduced, while epi-manool remained a strong peak. This is in line with the reduction in thunbergol reported for thermally modified wood oleoresin (Table 1). Small new peaks had appeared at 25.25 and 25.46 min, relating to thunbergene and pimaradiene; these had also been seen in the samples from thermal modification experiments (Figure 5). Other minor peaks in the sesquiterpene region were also observed, including 25.80, 26.00, and 26.27 min. These were too weak to allow for identification; however, the mass spectra contained characteristic fragments for pimaratriene or androstandiene structures. This would be consistent with their development due to thermally activated dehydration reactions of the equivalent diterpene alcohol. Within the same region, there was also a somewhat stronger peak at 26.82 min. The mass spectrum of this compound revealed a phenyl ester structure, possibly characteristic of lignin degradation or of lignans within the resin. Callus resins (formed in response to injury) are typically rich in lignans and lignin esters [21], so a trace of lignin may have been present. Another peak at 27.63 min (with no good library matches) appeared to have a furan structure, and may relate to the degradation of polysaccharide (if present as contamination, e.g., traces of wood dust) occurring at elevated temperatures.

In the samples from an isothermal temperature of 170 °C, similar sesquiterpenes were seen at the equivalent elution times, but in lower relative abundance, indicating a greater degree of evaporation at this higher temperature. A new sesquiterpene was observed at 25.72 min, and again, the structure indicated that this could be a triene with a pimarane skeleton, for example, norpimaratriene. This could be consistent with further dehydration or degradation of the original diterpene components. For the 190 °C isotherm, the sample contained only the 25.71 min norpimaratriene, the 27.21 min epimanool, and the 27.52 min unidentified furan derivative.

It is clear that heat has two modes of operation on the diterpenes within larch resin—firstly, dehydration reactions, converting diterpene alcohols to di- or trienes; and secondly, slow evaporative loss occurring over long periods at higher temperatures. Within a 60 min period at 190 °C, the loss of diterpenes was significant, in addition to the previous loss of mono- and sesquiterpenes seen at lower temperatures.

4. Discussion

The GC-MS analysis of pure oleoresin from TGA/DSC studies demonstrated that the monoterpenes in pure larch resin samples were almost completely evaporated by isothermal treatment at 120 °C, and completely removed in higher isothermal treatments at 150, 170, and 190 °C. This is significant because the boiling point of pure α-pinene is 155 °C, and β-pinene, 163 °C. Both pinenes have relatively low enthalpies of vaporisation (37.8 and 38.6 kJ/mol, respectively), which may lead to the volatilisation of individual molecules below the boiling point. The vapour pressure of α- and β-pinenes is 3.5 and 2.4 mmHg, while water is 17.5 mmHg. Water appears to be closely linked to the volatility of terpenes; for example, Banerjee suggests three stages in the evaporation of monoterpenes from sawdust during high-temperature drying, relating to the moisture content of the material and the stage in the drying process [12].

In addition, this study has also shown that sesquiterpenes are driven off readily at temperatures from 150 °C upwards, despite the boiling point of these compounds being higher, e.g., 204 °C for longifolene and 208 °C for germacrene. This is consistent with the work reported by Granström for sesquiterpenes in high-temperature (170 to 200 °C) kiln-drying of softwoods, where up to 20% sesquiterpenes were identified in the kiln emissions alongside monoterpenes [29]. In the TGA output from the TGA/DSC runs, it is obvious that the weight change due to the loss of terpenes is not sudden, as might be expected for the evaporation of a pure compound, accompanied by a sharp endotherm in the DSC trace. Neither was it a series of such evaporation events, as would indicate a sequence of components from a separable mixture. Instead, the TGA records a gradual weight loss over the full 60-minute isotherm. This implies a strong interaction between the various components, which is also likely, given the azeotropic tendencies of the various monoterpenes as discussed above and as indicated in the studies by Banerjee et al. [11,12].

Relatively little has been written regarding the diterpene composition of larch oleoresin. Holmbom et al. [21] derivatised the oleoresin and analysed the trimethyl silyl (TMS) derivatives of the resin acids; as a result, the authors observed a profile dominated by isopimaric acid (17.4%) and abietic acid (11.3%), with various other resin acids. However, no alkene diterpenes were listed in the identified compounds for the oleoresin studied. Sato and co-workers [23] analysed diterpenes from branch bark without derivatisation, isolating the diterpenes in pure form using dichloromethane. Similarly, in this present study, no derivatisation step was performed to maximise the knowledge gained about the raw oleoresin compounds before and after the thermal treatment or the thermal simulation. The absence of TMS derivatisation is likely to have suppressed the quantity of carboxylic acid diterpenoids observed in this experiment. However, the direct study of mono- and sesquiterpene profiles was effective, and the losses of both components during heating were clearly observed.

Within this study, the most noticeable change in the diterpene region of the 150 °C isotherm sample was the formation of two new peaks at 25.25 and 25.48 min. The untreated resin did not contain these peaks. These were identified as a thunbergene and a pimaradiene, which likely formed by the dehydration of thunbergol and pimaradienol during heating. The dehydration of thunbergol to thunbergene is widely known in museum and fine art curation [18,30], where Venice turpentine (larch rosin from European larch) was widely used in oil paintings of old masters. The dehydration of pimaradienol to piamaradiene is less widely reported; however, the mass spectrum of the second new peak observed at 25.48 min contained the characteristic mass fragments for pimara 8,15-diene.

The temperatures used in the DSC experiments and the thermal modification kiln were lower than is common in conventional thermal modification systems (typically 180 to 230 °C), where chemical changes in the wood-cell-wall components are sought. But this mild and moderate thermal modification process was conducted at a sufficiently high temperature to mobilise and volatilise a significant proportion of the larch resins. The temperatures studied (120 to 190 °C) were higher than those investigated by Grüll and co-workers [31], where 70 °C with and without partial vacuum was tested for European larch to reduce resin bleed in coated larch for exterior applications. They were also higher than the work by Kačik and co-workers [32], who used 60 °C and 120 °C to reduce the monoterpene content of fir to limit attack by wood-boring insects. DSC results indicate that 70 °C would be insufficient to mobilise the liquid component of the larch resin.

5. Conclusions

This study reports chemical profiling experiments to explain the changes which occur when larch resin dries during the mild thermal modification process. This indicates potential for the mild modification to reduce the dulling of blades in paling or other secondary machining processes, increasing efficiency and product quality. After mild thermal modification, the chemical composition of resin had changed as expected, showing a reduction in monoterpenes such as α- and β-pinenes, which have a boiling point below the treatment temperature. The loss of short-chain alkanes and fatty acids is believed to explain the solidification or ‘curing’ of larch resin, which occurs during thermal modification; however, relatively few short-chain alkanes were detected even in the GC-MS analysis of fresh resin. The chief differences observed were the evaporation of mono- and sesquiterpenes. Additionally, changes in the diterpene composition were observed, relating to chemical reactions occurring to the terpene alcohols under the action of heat.

The use of TGA/DSC combined with GC-MS has allowed for the evaporation processes seen in mild thermal modification to be observed independently of the chemical composition changes which occur within wood. A near-complete removal of monoterpenes occurred in the 120 °C isotherm schedule. Both mono- and sesquiterpenes were removed in the schedules at 150 °C and above. The profile of terpenes present in the dried oleoresin from 150 °C isothermal runs was found to be similar to that harvested from mildly thermally treated larch timber, while the 120 °C isothermal run resembled resin from the Day 1 pre-treatment step. This experiment revealed that the monoterpenes and sesquiterpenes are driven off at temperatures which are lower than their boiling points, in a process which is believed to be related to the evaporation of water from within the resin. This corresponds with previously reported observations of terpene evaporation from solid wood during high-temperature drying processes.