Oak wood drying: precipitation of crystalline ellagic acid leads to discoloration

Abstract

Oak heartwood usually darkens during and after drying. This darkening can be heterogeneous, leaving noncolored areas in the wood board. These light discolorations have been linked to heterogeneous distribution of tannins, but compelling evidence on the microscale is lacking. In this study Raman and fluorescence microscopy revealed precipitations of crystalline ellagic acid, especially in the ray cells but also in lumina, cell corners and cell walls in the non-colored areas (NCA), which also had higher density. In these denser areas free water is longer present during drying and leads to accumulation of hydrolyzed tannins. When eventually falling dry, these tannins precipitate irreversible as non-colored ellagic acid and are not available for chemical reactions leading to darkening of the wood. Therefore, pronounced density fluctuations in wood boards require adjusting the drying and processing parameters so that water domains and ellagic acid precipitations are avoided during drying.

1. Introduction

Oak heartwood is one of the most esthetic European hardwoods, appreciated for numerous applications. For most of these, the native water must be removed by seasoning and kiln drying to allow further processing and protection (Ramage et al. 2017). Kiln drying is a process whereby wood boards are set to different temperatures and moisture profiles for a certain duration. Generally, the drying can be divided into two phases: a water removing phase, during which free water is removed from the lumen and a drying phase where the bound water is removed from the cell wall (Perré 2007). Oak timber is known to darken during and after drying and in this context, non-darkening areas, visible as white discolorations, are an esthetic problem, as they decrease the visual quality and result in financial losses for industry (Koch and Skarvelis 2007; Wassipaul and Fellner 1992).

A wealth of studies unveiled that the white discoloration is caused by an irregular accumulation of extractives (tannins), which results from the water transport to the outer regions (Kisseloff 1993; Wassipaul and Fellner 1992; Wassipaul et al. 1987; Wegener and Fengel. 1988). It is known that the darkening process is induced by hydrolysis and oxidative reactions of tannins (e.g., ellagitannins) during kiln drying and that the drying conditions play a crucial role (Charrier et al. 1995). Initially, it was suggested that characteristic conditions, such as 30–60% moisture content (MC), a drying temperature above 30°C and a relative humidity around 70% favor the white stains, which start from the surface and move inwards (Fortuin et al. 1988a,b; Koch and Skarvelis 2007; Wassipaul et al. 1987). Furthermore, evidence has been found that quick vacuum drying, drying in nitrogen atmosphere, super-slow airdrying or low temperature pre-drying (from green to 25% MC) prior to kiln drying can prevent these discolorations (Brunner 1999, Charrier et al. 1992a,b, Fortuin et al. 1988a,b; Wassipaul et al. 1992, Welling et al. 1995). So far, these techniques have only been tried on a laboratory scale and have not yet found an application in industry (Koch and Skarvelis 2007). Besides physiological reactions in the living/freshly felled tree, it has been suggested that the discoloration is essentially based on chemical reactions of extractives (tannins) and cell wall components (lignin and hemicellulose) (Koch and Skarvelis 2007).

Tannins are water soluble, non-toxic, and highly reactive polyphenols, which are classified into condensed (polyflavonoids), complex and hydrolyzable tannins (e.g., ellagitannins in oak). Condensed tannins are stable and rarely subjected to hydrolysis (Pizzi 1980) and represent more than 90% of the most commercially used tannins worldwide. Although oak heartwood contains only a very small amount of condensed tannins (Miranda et al. 2017), several authors have reported that under hot acidic conditions the inter-flavanoid bonds break and red colored anthocyanidins are released (Higuchi 2012; Puech et al. 1999). Complex tannins are formed by a C-glycosidic linkage of ellagitannins and flavan-3-ol (Ferreira and Slade 2002; Okuda 2005), but are not detectable in oak heartwood (Quideau et al. 2003).

Hydrolyzable tannins such as ellagitannins are the most important extractives found in oak heartwood and can account for 10% of the dry weight (Puech et al. 1999). Sapwood and heartwood contain ellagitannins, the latter at higher amounts (Miranda et al. 2017; Viriot et al. 1993). The basic structure of ellagitannins contains a polyhydric alcohol (D-glucose) where the hydroxyl groups are esterified by gallic acid or 3,3’,4,4’,5,5’-hexahydroxydiphenic acid (HHDP) (Puech et al. 1999; Viriot et al. 1993). In oak, the main ellagitannins are vescalagin and castalagin, which can be easily hydrolyzed (enzymatically or in acid/ base conditions) to vescalin and castalin, respectively, with concomitant formation of free gallic or HHDP acid, where the latter can spontaneously lactonize to ellagic acid (Mayer et al. 1967; Puech et al. 1999).

Large quantities of ellagitannins have been observed in woody tissue containing a high proportion of parenchyma cells (Puech et al. 1999). The parenchyma cells (radial and axial) play a major role in the biosynthesis of the ellagitannins, especially during the sapwood/heartwood transition. As the ellagitannins are responsible for natural durability, different grades are achieved depending on species, age, and position within the stem. During the natural aging and seasoning of oak heartwood, the ellagitannins content decreases regularly and free ellagic acid and gallic acid increase towards the pith (Charrier et al. 1992a,b; Puech et al. 1999; Viriot et al. 1994). However, hydrolysis alone does not explain this decrease of ellagitannins, as more than 80% of the loss of ellagitannins during aging results from the polymerization into larger, dark, and insoluble polymers. These compounds are thought to be covalently linked via the HHDP unit to the wood structure, causing the darkening of oak heartwood with age (Klumpers et al. 1994). Most studies on discoloration and extractives in oak are based on wet chemical methods and do not show the distribution and changes within the wood microstructure (Charrier et al. 1995; Koch and Skarvelis 2007). Therefore, there is a lack of direct visualization of these discoloration events on the micro scale.

In recent years, Raman microspectroscopy has been recognized as viable method to track different wood components on the micro level with a high lateral resolution (~300 nm). Particularly, the Raman imaging approach has demonstrated an unparalleled accuracy of extractive detection due to the highly specific, “fingerprint” – like, spatial Raman signatures (Belt et al. 2017; Felhofer et al. 2018; Gierlinger 2018). However, the detection of different phenolic extractives remains limited due to different mechanisms strongly affecting the individual Raman intensities (Bock and Gierlinger 2019; Bock et al. 2020). Thus, only a combination of several methods allows to answer the question, how the discolorations evolve on the microlevel.

In this study, a combination of Raman microscopy, FT-IR spectroscopy, HPLC, UV-spectroscopy, fluorescence and light microscopy was used to track chemical changes in oak boards with severe discoloration after kiln drying. In particular, the non-colored areas (NCA) were compared with the darker colored regions (CA) of the board.

2. Materials and methods

2.1. Wood material

Oak boards (Quercus robur L.) of 5 cm thickness and 180 cm length were used, the width of different boards varying between 20 and 40 cm. In total, 27 boards were dried for this experiment; detailed information is listed in the Supplementary Material. Briefly, the first step was outdoor pre-drying until a moisture content (MC) of around 16%, followed by drying experiments performed in a laboratory convection kiln dryer. Only boards with severe discoloration were considered for microscopic analysis. For the density and porosity calculations, blocks (1 cm³; n = 14) were cut with a circular saw from board cross sections containing non- and colored areas. First, the blocks were kiln dried at 105 °C and then the bulk density was determined. From these data, the porosity was calculated after Plötze and Niemz (2011).

2.2. Microsectioning

Blocks from non-colored areas (NCA) and colored areas (CA) were split with a chisel from the oak boards after drying. Blocks containing both areas were also cut out to capture the difference at once. Before sectioning, the blocks were immersed in D₂O under vacuum, to reduce the overlap of the OH and CH stretching bands during the Raman measurements. The blocks were tightly clamped into a rotary microtome (RM2235, Leica Biosystems Nussloch GmbH, Wetzlar, Germany) with an orientation perpendicular to the main fiber axis. Disposable microtome blades (N35HR Blade 35°, Feather, Osaka, Japan) were used for cutting 10–15 μm thick transverse sections. The thin sections were placed on a standard glass slide with a drop of D₂O, covered with a glass coverslip (0.17 mm thick) and sealed with nail polish. For the extraction, the micro sections were washed several times with 80% EtOH and subsequently embedded in D₂O.

2.3. Confocal Raman microscopy (CRM) and IR spectroscopy

Raman measurements of the microsections were performed with a confocal Raman microscope (Alpha300RA, WITec GmbH, Germany). For exciting the sample, a polarized, coherent compass sapphire VIS laser (λ_ex = 532 nm, WITec GmbH, Germany) was used. The radiation was focused through a 100× oil immersion objective (numerical aperture = 1.4, coverslip correction 0.17 mm) (Carl Zeiss, Germany) onto the sample. Measurements at 785 nm were conducted the same way, but used a linear polarized XTRA II laser (785 nm, Toptica Photonics, Germany). The Raman scattering directed through optical multifiber (50/100 μm diameter) to a spectrometer (UHTS 300 WITec, Germany) (600 g.mm^-1 grating) and finally to the CCD camera (Andor DU401 BV and Andor DU401 DD, Belfast, North Ireland). The lateral resolution was about 0.3 μm one full wavenumber spectrum with an integration time of 0.08 s and a laser power of 35/190 mW (532/785 nm) was obtained from every image pixel. The Control Four (WITec GmbH, Germany) software was used for acquisition of the Raman measurements. Reference spectra of ellagic acid and tannic acid, purchased from Sigma-Aldrich (Vienna, Austria), were measured on the same system.

IR measurements were conducted on an FT-IR ATR spectrometer (Vertex 70, Bruker, Billerica, USA) with 32 scans per measurements. Five measurements were averaged using OPUS 7.5 software from Bruker (Billerica, USA).

2.4. Data analysis

Raman data analysis was performed using the WITec project FOUR 4.1 (Germany) software. Prior to image generation, pre-processing was performed as follows: the scans were cropped, and a cosmic ray removal filter was applied. Based on the integrated images, average spectra of defined sample areas were obtained by manually selecting an area or using an intensity threshold to select areas. Detailed spectra analysis was performed using OPUS 7.5 software (Bruker, Billerica, USA).

2.5. FCA staining

Sections were immersed in FCA staining solution (0.1 mg/mL of new Fuchsin, 0.143 mg/mL Chrysoidine, 1.25 mg/mL Astra blue and acetic acid [v:v = 1:50]) at room temperature for 10 min. Sections were iteratively washed with Milli-Q water, ethanol (30%, 70%) and Isopropanol. Finally, sections were embedded in Euparal and photographed under a Labophot-2 microscope (Nikon Corporation, Tokyo, Japan).

2.6. Confocal laser scanning microscopy

Confocal fluorescence imaging was performed using a Leica TCS SP8 confocal system with a Zeiss 60 × /0.9 numerical aperture (NA) water objective. Emission spectra of pure ellagic acid were acquired using the “lambda scan” mode with 405 nm excitation. For staining of lignin, Basic Fuchsin was used according to published protocols with modification (Ursache et al. 2018). Sections were stained in Basic Fuchsin staining solution for 1.0 h, and then washed several times with Milli-Q water and embedded with deuterium. Basic Fuchsin was excited at 561 nm (fluorescence emission: 600–650 nm), Ellagic acid at 405 nm.

2.7. SEM

The Thermo Scientific Apreo VS (Figure 2) was used to scan different samples. A field emission gun (FEG) with a voltage of 2 kV and a beam current of 0.1 nA was used. For sensitive samples the beam current was reduced to 13 pA/6.3 pA, respectively. The Apreo creates a high vacuum (1.0 × 10^-4 Pa) and variable pressures (ranging from 10–500 Pa) were used in the working chamber.

Figure 2. Raman microscopy of colored and non-colored areas reveals ellagic acid in rays.

(a) Raman measurements performed on an oak block including non-colored (NCA, left) and colored area (CA, right). First image shows a higher fluorescence in NCA. Second image shows the distribution of ellagic acid. Note that signal intensity of ellagic acid coincides with Raman fluorescence, the highest intensities being found in the radial rays. (b) Brightfield image of a ray in the non-colored area. (c) Fluorescence image and ellagic acid distribution (1706 cm^-1). (d) Average spectrum of the deposits in the radial ray compared with ellagic acid reference spectrum (measured with 532 nm laser). (e–f) Detailed Raman images of cells in the NCA and CA. (g–h) Average spectra from the cell corner (CC)/middle lamella (ML) and the cell wall (CW). In the CA the CC/ML and outer cell wall show a higher fluorescence compared to the NCA. Insert shows the increase of the fluorescence in the CA compared to the NCA for different tissues.

2.8. HPLC

Samples were ground and dissolved in 10 mM NaOH and shaken for 24 h under nitrogen atmosphere. Standards of ellagic acid were prepared the same way immediately before measurement. Samples were measured on an Agilent 1100 HPLC system (Agilent Technologies, Santa Clara, US) equipped with a diode array detector. Acetonitril (ACN) and water were used as eluents. The following gradients were used: 0–15 min 45/55 ACN/H2O isocratic; 15–25 min 10/90 ACN/H2O.

2.9. UV–vis spectroscopy

Reference spectra were recorded on a Hitachi U-2900 Spectrophotometer (Hitachi High Technologies Corporation, Japan). Samples were dissolved in ethanol.

3. Results and discussion

Ellagitannins are the most important hydrolyzable tannins found in oak (Puech et al. 1999). In the context of wood drying and discoloration, water movement within the wood tissue is important because a large portion of ellagitannins is highly water soluble (Vivas et al. 2020). Water movement depends on the continuity of the main water pathways, pore size and the distance to the surface (Figure 1a and Supplementary Figure S1a). On the macroscopic scale, oak wood can be divided into the outer lighter sapwood and the inner darker heartwood (Fromm et al. 2001), this is also shown in Figure 1a. The darker color of the heartwood relates to a higher content of ellagitannins (vescalagin and castalagin) and colored components (Klumpers et al. 1994). However, aging (Viriot et al. 1994; Vivas et al. 2020 and references therein), seasoning, and drying can cause strong color variety and inhomogeneities in heartwood (Fortuin et al. 1988; Koch and Skarvelis 2007; Wassipaul and Fellner 1992), which could also be seen in the measurements as brownish colored areas (CA) and non-colored areas (NCA) (Figure 1a and b). On the micro-scale, earlywood and latewood differ in their function due to lumen and cell wall size. Additionally, fibers with thick and tracheids with thin cell walls, vessels and axial/radial parenchyma cells represent different water pathways (Segmehl et al. 2018). The different cell wall thickness could already be observed under the normal light microscope (Figure 1b) but were better visible when the section was stained (Figure 1c). From that it could also be deduced that the color inhomogeneities do not stem from different cell wall thickness.

Figure 1. Overview of a kiln dried oak wood sample with colored and non-colored areas.

(a) Photograph of the transverse face of the block showing the outer sapwood (SW) and heartwood (HW). Boxed region includes colored areas (CA) and non-colored areas (NCA). (b) Zoom-into heartwood showing earlywood (EW), latewood (LW) and radial rays in a transition zone from CA to NCA. (c) FCA staining shows higher intensity in the CA (left image) compared to the NCA (right image). (d) Boxplots of the density of the non-colored and colored area. (e) Boxplots of the porosity of the non-colored and colored area. The specific density (ρ_s) is taken from Plötze and Niemz (2011). Statistical significance was tested by Student’s t-test (**p < 0.05). (f) Average FT-IR ATR spectra from sapwood and heartwood (CA and NCA). (g) FT-IR ATR spectra of hot water extracts from NCA and CA as well as reference spectra of vescalagin and ellagic acid.

FCA staining of the transition zone between colored and non-colored area showed more reddish tissues implying higher phenolic content in the colored area (Figure 1c). The results are in keeping with Charrier et al. 1995, who observed that the concentration of high molecular weight phenolic compounds was five times greater in colored areas than in non-colored areas. The density of non-colored areas was significantly higher when compared to colored areas (Figure 1d), which implied a lower porosity (Figure 1e). Lower porosity in turn would mean that such regions stay moist for a prolonged time during drying. Previous research shows that remaining liquid water during drying can be preferably found in the lumina of the least accessible fibers (Hernández and Cáceres 2010).

Infrared spectra of sapwood and heartwood as well as of colored and non-colored areas were almost identical (Figure 1f). The interpretation was that the chemical compounds involved in the discoloration appear only in small amounts, too small to be resolved against the large amount of lignin and (hemi)cellulose signal. Also, spectra of the freeze-dried water extract from colored- and non-colored areas were very similar but matched well with reference spectra of ellagitannins (Figure 1g, compared to vescalagin). As the IR-measurements were only showing the existence of ellagitannins but gave no hint on their identity or amount, also HPLC-analysis of the Water/ NaOH-extract was performed. Results showed a higher eluent peak for the non-colored extract and this peak was assigned to vescalagin/castalagin (see Supplementary Figure S2). This was interpreted that the hydrolyzable fraction of ellagitannins played an important role in discoloration in the samples used. Based on tentatively assigned spectra, also the gallotannin content seemed to differ. Almost no difference was found for ellagic acid. This is explained by its poor solubility in water (Alfei et al. 2019), which could also explain the occurrence of deposits of precipitated ellagic acid deposits observed in the Raman images (Figure 2a–c). Deposits were mainly observed in the non-colored area and were less abundant or totally absent in the colored area, as also observed by fluorescence microscopy (Figure 3). Depending on the excitation wavelength, these deposits showed high fluorescence in the Raman spectrum, but could be unambiguously assigned to pure ellagic acid (Figure 2c, d and Figure 4).

Figure 3. Fluorescence microscopy images show additional phenolic compounds in the lumen, pits and cell corners.

(a) Bright field image of a micro section taken from a block including non-colored (NCA) and colored areas (CA). NCA are bordered with red lines. (b) Scanning electron microscope (SEM) images of the two areas. (c) Auto fluorescence images for additional phenolic compounds ellagic acid (EA), gallic acid (GA), 3,3’,4,4’,5,5’-hexahydroxydiphenic acid (HHDP) and tannic acid (TA). The NCA shows deposits in cell corners and cell walls, whereas the CA shows no signal. (d) Corresponding emission spectrum for additional phenolic deposits in wood compared with pure ellagic acid as a standard. (e) Merged images of lignin (basic fuchsin) and additional phenolic compounds. The images show that the deposits are in the cell corners, lumen/vessels and pits. (f) Schematic representation of the distribution of these deposits within the wood structure.

Figure 4. Deposits analyzed by Raman micro spectroscopy.

(a) Merged Raman images of lignin (integration of the peak 1600 cm^-1) with ellagic acid deposit (integration of the peak 445 cm^-1). (b) Comparison of the extracted average spectrum from the deposit in the lumen (red) with pure ellagic acid as a reference (measured at 785 nm).

Interestingly, in the cell walls of the colored area, Raman spectroscopy revealed a higher fluorescence towards the compound middle lamella (CML) (Figure 2e–h). This fluorescence could be caused by the same compounds also being responsible for the darker color, because extended π-conjugation of colored compounds shifts the absorption in the visible range, accessible to the laser. Unfortunately, high fluorescence did not allow for characterization by Raman microscopy. However, previous research suggests that hexahydroxybiphenic acid (HHDP), which is a hydrolysis product of ellagitannins as well, forms insoluble, dark compounds (Klumpers et al. 1994; Peng et al. 1991). Hexahydroxybiphenic acid constitutes of two rings joined over a coannular bond, therefore satisfying the condition of a conjugated π-electron system. In addition, a recent study confirmed that HHDP can form colored complexes with polysaccharides (Vivas et al. 2020). Finally, the extraction might have negatively affected the fluorescence, as Raman images of colored areas showed more fluorescence after extraction with ethanol. Recent research shows that polyflavonoid tannins (condensed tannins) are not extractable with ethanol, but with ionic liquids (Donaldson et al. 2019; Hejazifar et al. 2016; Sukor et al. 2020). This leaves space for further research fine-tuning the extraction parameters.

The location of the precipitated ellagic acids, however, allowed some conclusions about the water transport during drying and the fate of ellagitannins and their hydrolysis products, because deposits found in cells and pits point to the importance of these locations in the water transport process leading to discoloration. To investigate the role of water flow, the density of colored and non-colored areas was measured and porosity determined (Figure 1d and e). It turned out that non-colored areas had a higher density and lower porosity than colored areas. It therefore seemed likely that water retention time was a determining factor for wood discoloration in the experiments, because hydrolysis of ellagitannins depends on time (Viriot et al. 1994). Furthermore, in denser, less accessible regions, free water remains longer below the fiber saturation point (FSP) (Hernández and Cáceres 2010). This is also true for pit membranes (Menon et al. 1987). It can be speculated that during drying, ellagitannins and their hydrolysis products concentrate in denser regions where free water remains even below the FSP and are eventually precipitated as ellagic acid. This process has been shown experimentally for vescalagin, which, at a critical concentration, reacts with itself and precipitates as ellagic acid (Viriot et al. 1994).

In summary, regions with a higher density than their surroundings seem to favor longer water retention times, which in turn increased ellagitannins in solution and concentrated hydrolysis products in such areas. This is because the equilibrium moisture content depends on the size distribution of the capillaries (Almeide and Hernández 2006; Passarini et al. 2014), therefore depends on wood anatomy. This corresponds to the results of measured density and porosity (Figure 1d and e), where a high density related to non-colored areas. High concentrations and/ or sudden desiccation would lead to the previously described self-reaction of ellagitannins and subsequent precipitation of ellagic acid, predominantly in those regions where the last spots of free water remain, as pits or rays, exactly where deposits of ellagic acid could be found by Raman and fluorescence microscopy (see Figures 2 and 3). Since ellagic acid does not dissolve in water, the precipitation process described in this paper is irreversible and in such regions, no brown coloration process as described in Vivas et al. (2020) can set in.

4. Conclusion

This study suggests that the susceptibility of oak boards to heterogeneous darkening is caused by the existence of regions, which have a higher density, leading to higher retention times of free, unbound water than in the surrounding, porous regions. Drying regime and speed affect the movement of water on a micro level and cause denser tissue to contain free water even below the fiber saturation point. Hydrolysis products of ellagitannins seem to concentrate within these last moisture domains and precipitate as ellagic acid when these areas eventually fall dry. Therefore, tannins precipitated this way cannot take part in coloring reactions leading to a darkening of the boards. Since ellagic acid is water insoluble, the precipitation process is irreversible and must be prevented at all cost. This suggests that adjusting the drying parameters is the most critical factor to prevent inhomogeneous discoloration.