The Potential of Size-Exclusion Chromatography for Evaluating the Suitability of Hydrophilic Extracts in Wood Preservation

Abstract

European beech wood has low natural resistance to microbial attacks especially when it is exposed outdoors. We looked at ways of improving this by applying three hydrophilic extracts from other species known for their ability to inhibit fungal growth. We prepared oak heartwood and black locust bark extracts by accelerated-solvent extraction with aqueous methanol and freeze-drying, and obtained black wattle bark hot-water extract commercially. The molecular size of the phenolic components and associated saccharides in the extracts were determined by size-exclusion chromatography (SEC). We found that two extracts improved beech wood durability even at low concentrations (5 wt.% solution); the most effective extract was black wattle extract. The worst performance, by black locust bark extract, was attributed to the presence of small-molecule phenolics. The total phenolic content was up to 9× lower than that reported for fresh extracts. Even though the extracts were not stored specifically to preserve the original phenolic content, we found that two were still effective as fungal inhibitors.

1. Introduction

Beech wood (Fagus sylvatica) is a traditional and available wood species in Europe. Due to its poor rot resistance, untreated beech wood is suitable only for interior applications where the relative humidity is controlled and water exposure minimised. The vulnerability of wood is due to the sparse presence of extractives [1] which led to beech being classified as perishable/non-durable (durability class DC 5) by EN 350 [2]. In contrast, the higher concentrations of extractives in many tropical wood species put them into the DC 1 (very durable) or DC 2 (durable) categories [3]. Extractive content may differ not only within the tree [4,5] but also with its place of origin [5]. For instance, the durability of black locust (Robinia pseudoacacia) heartwood can range from DC 1 to 4 [5,6]. European oak species are more consistent: Quercus robur and Q. petraea are classified as DC 2-4 [2], while Q. pyrenaica wood is DC 1 [7]. The age of the tree affects its resistance to white rot (Trametes versicolor). Heartwood from younger black locust trees (up to 20 years old) has a sparse extractive content and is classed as DC 2-4, while mature heartwood falls into the category DC 1 [6].

Several natural botanical compounds, also called secondary metabolites, have been assessed as replacements for traditional wood preservatives, some of which have been shown to be harmful to the environment or human health and even banned from use [8]. Examples include terpenes, found in conifers, and tannins and flavonoids from broadleaf species where they provide defence against wood-destroying organisms [9].

Tannins are the third most abundant group in vascular plants after carbohydrates and lignin [10]. Based on their main molecular unit, tannins are divided into two groups: (i) hydrolysable tannins (HTs) or gallo- and ellagitannins; and (ii) condensed tannins (CTs), also called proanthocyanidins [11]. The phenolic structure of tannins allows for the development of new bio-based polymers and a variety of degradable materials with several functions [10,12,13]. Plant polyphenols are capable of precipitating proteins and thus affecting signalling paths within cells [14]. They also have biological properties—for example, antioxidant, anticancer and antimicrobial activities [14]. CTs are in general more effective than HTs against microorganisms, although some are still able to grow on tannins, especially on simple (and small) phenolic molecular units of both types (gallic acid in HTs and catechin in CTs) at low concentrations [15,16].

The main industrial application of tannins is in centuries-old leather production methods [17,18]. Tannins for commercial use are taken from the bark and heartwood of broadleaf trees known to be rich in polyphenols—the bark of Acacia mearnsii (black wattle, also called mimosa), the heartwood of Schinopsis spp. (quebracho) and Castanea spp. (chestnut), and both the wood and bark of Quercus spp. (oak) [10,19]. The highly abundant sources of CTs are Acacia and Schinopsis spp. and bark from Quercus spp. [16,20,21,22], while the wood of Castanea spp. and Quercus spp. has high levels of HTs [16,23,24].

Besides the source of the tannin, the following parameters have a significant effect on its properties and thus effectiveness: extraction method, solvent, temperature, number of extracting cycles, and its concentration in the solvent. The conventional extraction method is to use heated reflux (with a Soxhlet apparatus); a more advanced alternative (fexIKA) saves time with computer-controlled parameters, allowing four extractions to be made simultaneously; another very simple extraction method employs ultra-sound [25]. Accelerated solvent extraction (ASE) is done at a pressure that is 100 times greater than atmospheric pressure and takes minutes, whereas ultra-sound or Soxhlet would take hours [26]. It also offers the possibility of sequentially extracting different compounds that are soluble in different solvents—the hydrophilic phenolics and unwanted fractions are collected separately, making the clean-up process more efficient. Aqueous methanol is widely used for phenolic extraction from European wood species [27,28], while hot water is preferred for tropical wood to achieve higher phenolic yields, and thus antioxidant capacity [29].

Climate change makes the need to promote sustainable forest planting and restoration more urgent [30,31,32], calling for forest reorganisation to achieve a more climatically suitable species distribution in Europe [32]. New climate conditions support the spread of invasive species—for example, Acacia spp. in Mediterranean areas, Quercus rubra (red oak) in temperate forests and Robinia pseudoacacia in mixed forests [33,34,35]. The arrival of such species can impact biodiversity and forestry economics [35]. When strictly controlled, some non-native species offer the possibility of yielding high-value chemicals, such as polyphenols, a benefit that may mitigate their invasive impact [35]. Some such chemicals have potential uses in bioenergy, feed, paper products, biorefinery and health products [33,35]. Polyphenols may have a positive environmental impact—for example: (i) by reducing the use of more harmful substances—for instance, phenol in phenol-formaldehyde adhesives [36]; (ii) by replacing plastics in food packaging [37]; (iii) by reducing the amount of polluting wastewater derived from the use of hazardous chemicals [38]; and (iv) in nutrition and medicine [12,39]. Comprehensive policy-making to control invasive tree populations is complicated by the different behaviour of different species and by the presence of hybrids [40].

Research has been reported on the potential of several botanical extracts but only by highlighting those with the highest phenolic, antioxidant, etc., content. In contrast, our aim was to take older bark and heartwood (stored indoors for 4–5 years) and measure the effect of their extractives on beech wood durability against the common white-rot fungus Trametes versicolor. The relationship between wood durability and the molecular size of extractives is rarely mentioned in the literature [24,41], so we measured the molecular size of the following extracts by size-exclusion chromatography (SEC): two bark extracts containing CTs (Robinia pseudoacacia and commercial extract from Acacia mearnsii) and one heartwood extract containing HTs (Quercus spp.).

2. Materials and Methods

2.1. Extraction Process

Black locust (Robinia pseudoacacia L.) bark and European oak (Quercus spp.) heartwood originated from the South Moravian region (Brno’s surroundings). We obtained hydrophilic extracts from them with 50 vol.% methanol (Penta Chemicals, Prague, Czech Republic) by accelerated-solvent extraction with a Dionex ASE 350 (Thermo Fisher Scientific, Waltham, MA, USA). Powdered samples were put into 100 mL of extraction cells and processed in two extraction cycles (5 min each) at 80 °C. Extraction was repeated several times with further samples. The extracts of each type were left in the dark at room temperature (22 ± 1 °C) for the methanol to evaporate. They were then frozen at −20 °C and freeze-dried (LaboGene CoolSafe, Allerød, Denmark) at −55 °C until dryness and labelled for further use as Robinia and Quercus. For the third test sample, Acacia mearnsii De Wild. hot-water bark extract was acquired in powdered form as “Weibull AQ” from TANAC S.A. (Montenegro, RS, Brazil) and labelled Acacia.

2.2. Chemical Analysis of Hydrophilic Extractives

2.2.1. Total Phenolic Content (TPC)

The determination of the total phenolic content of the three extracts was based on a reaction with Folin–Ciocâlteu’s phenol reagent [42] and performed according to Výbohová and Oberle [43]. Briefly, 200 µL of Folin–Ciocâlteu’s reagent (Sigma-Aldrich, Prague, Czech Republic) was added to each diluted extract, then the pH was increased by adding 1 mL of a 20 wt.% solution of sodium carbonate (Penta Chemicals, Czech Republic). After colour stabilisation (30 min), the absorbance of the mixture was measured at 700 nm. Calibration was done with gallic acid (Sigma-Aldrich, Czech Republic) after thorough dissolution in demineralised water at six concentration points (R² = 0.9998). The phenolic content present in the extracts was expressed in milligrams of gallic acid equivalents per gram of powdered extract (mg GAE/g E) and taken as an average value of three replicates.

2.2.2. Size-Exclusion Chromatography (SEC)

The molecular size distribution of the three hydrophilic extracts was determined using an HPLC Agilent 1260 Infinity chromatograph (Santa Clara, CA, USA). A separation was done with an Asahipak GS-320 HQ multi-mode analytical column (Shodex, Tokyo, Japan) of dimensions 300 × 7.5 mm (particle size, 6 µm; pore size, 400 Å). We kept the column temperature at 30 °C. As a mobile phase we used deionised water with an isocratic flow rate of 0.5 mL/min. For the detection, we coupled two detectors in series: a diode-array detector (DAD), set up at wavelengths 254 and 280 nm, and an evaporative light-scattering detector (ELSD); both were controlled using the OpenLab chromatographic software (Agilent, Santa Clara, CA, USA).

Before taking the measurements, we optimised the ELSD (N₂ evaporation and nebuliser temperature, flow, column temperature and injection volume) with three pullulan standards, 342, 6600 and 22,000 Da (Polymer Standards Service, Mainz, Germany), while observing their retention time and changes in peak areas. Based on the results from the optimisation of the ELSD, we ran the test under the following conditions, similarly to a study by Muñoz-Almagro et al. [44]: evaporation temperature, 73 °C; nebuliser temperature, 63 °C; and N₂ flow, 1 SLPM.

Calibration was done with glucose (Sigma-Aldrich, Czech Republic) having 180 Da and six pullulan standards in the range of 342–50,000 Da (Polymer Standards Service, Germany). The standards were dissolved at the recommended concentration in demineralised water, as suggested by Held and Kilz [45], and filtered through a 0.45 µm PP-syringe filter prior to injection. Aqueous solutions of the two lyophilised extracts (Quercus and Robinia) and one commercial extract (Acacia) were prepared just before the analysis with a concentration of 5 wt.% (the same as was used for wood impregnation). Each solution was diluted 25 times and filtered through a 0.45 µm PP-syringe filter. Finally, we injected a volume of 10 µL by an autosampler into the HPLC system. The determination of the molecular size of each extract was based on the calibration fit model from the standards. Data were processed and visualised using Python 3.14 and the matplotlib Python library.

2.3. Wood Impregnation

We obtained European beech (Fagus sylvatica L.) wood from the Training Forest Enterprise of Mendel University in Brno (Křtiny, Czech Republic). Samples without visible cracks or discolouration were cut into dimensions of 30 × 10 × 5 mm (longitudinal × radial × tangential), making a total of 76 samples. Impregnation solutions of 5 wt.% were prepared by dissolving the powders from commercial Acacia bark extract, lyophilised Quercus heartwood extract and lyophilised Robinia bark extract in demineralised water. Each group consisted of 12 samples. One group (impregnated with demineralised water alone) served as a reference (REF), while the remainder served as a control and correction samples during wood durability testing. The samples were vacuum-impregnated while fully immersed in the solutions in a desiccator at ambient temperature (22 ± 1 °C). First, air was evacuated for 11 min, then the samples were kept for 20 min under the established vacuum [46]. They were then kept in solution at atmospheric pressure for 1 h, then taken out and had excessive liquid removed. To avoid any unwanted thermal decomposition of the phenolics in the wood structure, samples were gently dried at room temperature (22 ± 1 °C) for two weeks and rotated regularly to ensure uniform drying. During the first four days a xylene atmosphere in a closed box served to prevent mould growth due to the high moisture content of the samples. The dry mass was corrected separately by drying three impregnated samples of each group at 103 °C until they achieved a constant weight.

2.4. Wood Durability Testing Against White-Rot Fungus

We performed a test for the assessment of biocidal efficacy of wood preservatives according to EN 113 [47] with reduced sample dimensions and therefore exposed them to the fungi for a shorter time. Petri dishes with malt extract agar (Himedia) were inoculated with the white-rot fungus Trametes versicolor (strain BAM 116). When the substrate was fully covered by the fungus, steam-sterilised wood samples (three treated samples of the same group and one reference sample) were placed in the dish under sterile conditions. The dishes were kept at 20 °C and 72% RH for eight weeks in an atmosphere-conditioned chamber. After this time, we wiped the mycelium from the samples, weighed them, oven-dried them at 103 °C to zero moisture content and weighed them again. The mass loss $\left(ML\right)$ of the samples was calculated using Equation (1):

$$ML \left(\%\right)=\left({\displaystyle \frac{m_i - m_d}{m_i}}\right)\times 100,$$

(1)

where $m_i$ is the mass of the wood sample in grammes before inoculation with fungi (i.e., the mass of a conditioned impregnated wood sample which was corrected by the elimination of moisture content and weight percentage gain gained from the correction samples that were oven-dried after the impregnation process), and $m_d$ is the dry mass in grammes of the decayed wood sample after exposure to fungi. The results were evaluated using Statistica 14 (TIBCO Software Inc., Boston, MA, USA) by a non-parametric one-way Kruskal–Wallis analysis of variance (ANOVA), followed by multiple comparisons of mean ranks for all groups. Differences at p ≤ 0.05 were considered significant.

3. Results and Discussion

3.1. Results of Chemical Analysis of Hydrophilic Extractives

3.1.1. Total Phenolic Content of Extracts

The total phenolic content of the hydrophilic extracts from Acacia, Quercus and Robinia is shown in Figure 1. The values were significantly dependent on the source: Acacia extract had the most, Quercus fell in the middle, and Robinia had the least.

The Acacia tannins of commercial origin were reported to contain between 589 (laboratory extraction) and 630–680 mg GAE/g dried extract (commercial extraction) [48,49,50]. Besides the differences in methodology among laboratories, manufacturer handling, extracting and sources (bark or heartwood), the biggest difference might be due to the storage conditions (time and temperature). Our recorded values for Acacia bark extract, being a fifth of those reported elsewhere [48,49,50], may have been influenced by our method of storage—in the dark, though without temperature control—so the ambient temperature may have occasionally reached 35 °C, a level shown by Kim et al. to reduce polyphenol levels from around 330 to 230 mg GAE/g within only two months in matcha (Camellia sinensis), while catechin levels remained stable only for 3 h at 24 °C in an aqueous solution [51]. Longer periods of time might have also affected the structure. Salazar-Orbea et al. showed that diverse polyphenolic compounds present in apple and strawberry puree decreased over one year at 24 °C; the time in storage and exposure to elevated temperatures during processing (freezing, thermal treatment, and high pressure) all had an effect [52]. After a year in storage, apple purees lost up to 30% of their proanthocyanidin and 20% of their hydrocinnamic content, while strawberry purees lost even more: 40% of total phenols, 30% of proanthocyanidins, 60% of ellagitannins and 99% of anthocyanins [52]. Although the content and stability of polyphenols and diverse flavonoids are important for nutrition and medicine, other sources of natural polyphenols, for instance, for engineering purposes, could have similar thermal and chemical limitations. Studies of stability are missing from the engineering literature so direct comparisons cannot be made.

Our two other tannin samples (Quercus and Robinia) were stored at ambient temperature in dark, closed bottles for over four years prior to extraction. Enzymes present in those samples could have caused oxidation and degradation of the polyphenols, leading to their diminished concentration [22]. The TPC value for our Robinia bark extract was 8× lower than the TPC found by Sillero et al. [53]. Q. rubra and Q. robur bark revealed 4× and 9× higher content [53] than our Quercus spp. heartwood extract. Interestingly, the highest content has been found in the bark of rather younger trees (with Robinia and Q. robur being 13 and 17 years old compared to 60-year-old Q. rubra) [53].

Besides the species, age and environmental conditions of the tree, the extraction method (solvent, temperature and pressure) also affects the TPC value. The heartwood of European oak has been reported by Keržič et al. to contain 50 mg GAE/g dw of wood (theoretically calculated based on the extraction gain yields as 515 mg GAE/g of extract) after extraction at 110 °C with Soxhlet [27]. In contrast, Baar et al., who extracted at 40 °C (ultra-sound), have reported a TPC content of around 25 mg GAE/g dw of wood and in subfossil oak, only 8 mg GAE/g dw of wood as a result of hundreds years of exposure to outdoor conditions (i.e., extractives leaching into water) [24].

3.1.2. SEC Calibration

Knowing the molecular size distribution is useful for several purposes, including in wood impregnation (where pores are of finite dimensions), surface treatments and synthesis. We investigated the molecular weight distribution of water-soluble oligo-/polymers within phenolic compounds and saccharides obtained from durable wood species. We selected a multi-mode size-exclusion chromatography column compatible with water as the mobile phase and water-soluble standard pullulan for calibration. Since pullulan is a saccharide without chromophores, we coupled ELSD (which is destructive) in series after DAD. In saccharide analyses a refractive index (RI) detector is usually used because it does not destroy the analyte, allowing it to be applied in further analyses. However, ELSD is much more sensitive than RI, has better resolution and has a detection limit of 10—up to 100 times less than RI [44,54].

The calibration for the qualitative range of molar masses was performed using molar masses at the peak maximum $\left(M_p\right)$ and by selecting the fit function. With an increasing polynomial degree, the deviation reduced (

Table 1. HPLC-SEC calibration models: degree, regression coefficients (R²), deviation from the real values of $M_p$ (sum of residua squares) and band width for the first derivative.

Calibration Fit Model (Degree)	R2	Sum of Residua Squares	Band Width for the 1st Derivative (Slope)
Linear	0.997096	0.017062	-
Quadratic (2)	0.997105	0.017011	0.006090
Cubic (3)	0.998327	0.009827	0.088387
Polynomial (4)	0.998473	0.008974	0.136622
Polynomial (5)	0.999876	0.000729	0.402127
Polynomial (6)	0.999980	0.000116	0.476142
Polynomial (7)	0.999998	0.000013	0.446019

). However, Held and Kilz have stated that (i) in physics, it is not meaningful to use the function with the highest degree, even though this will always generate the lowest average deviation, and (ii) it is more important that the shape of the calibration is in agreement with the separation mechanism (i.e., the slope of the calibration curve) [45]. The quadratic (second-order polynomial) function model was chosen to calculate the $M_p$ present in the samples because: (i) the first derivative was monotone as the only one among other models, and (ii) the band width of the 1st derivative was the narrowest (and had a slightly higher regression coefficient than in the linear model).

As we did, Cesprini et al., Kron et al. and Pascotto et al. [49,55,56] applied two detectors in series (DAD and RI) for the simultaneous detection of phenolic and non-phenolic compounds. Others, notably Pascotto et al., Watrelot et al. and Campbell et al., have reported the direct use of phenolic standards, such as gallic acid, (epi)catechin, castalagin and ellagic acid, with a UV detector at a wavelength of 280 nm [56,57,58,59]. Yet other studies have relied on standards typical for SEC, and the further fraction determination of analytes with chromophores has been reported using RP-HPLC, often coupled to a mass spectrometer (MS) [60]. The biggest advantage of SEC over phenolic standards for molecular weight determination is their exact structure and size distribution, which are maintained by synthesis. In contrast to SEC standards, phenolic compounds are highly variable in their structure and molecular size (due to substitutions and linkages) and lack proper chemical stability, thus are not available to cover the chosen molecular weight range with appropriate analytical reliability.

3.1.3. Molecular Weight Distribution of Extracts

The molecular weights of individual peaks were estimated according to the calibration model, using the position of the peak maximum separately for phenolic (Figure 2) and saccharidic (Figure 3) compounds present in the extracts. The peak areas of four molecular weight regions were calculated to allow for quantitative comparisons of phenolics present in the extracts (Table 2).

(A) Phenolics

The phenolics detected in the extracts are visualised in Figure 2. The chromatograms were divided into four fractions to allow for a better explanation of the differences between the extracts. The relative distribution within the extracts is described in Table 2. While the Acacia bark extract showed the greatest relative abundance in fraction II (50,000–2500 Da) with almost 60%, the Quercus heartwood extract delivered two similar fractions with almost 42% and 40% for I (over 50,000 Da) and II, respectively. In the Robinia bark extract, the most abundant (53%) was II, followed by III (2500–150 Da) with almost 32%. The Robinia extract gave a more concentrated result than Acacia and Quercus (i.e., it had a higher peak area; see Table 2), which could mean that Acacia was less soluble or bigger fractions were filtered out prior to the analysis.

Figure 2. Chromatograms from size-exclusion chromatography of extracts seen in UV at 280 nm; extracts from (a) Acacia, (b) Quercus and (c) Robinia. The numbers above the curve represent the assigned molar mass at the peak maximum within the calibrated area (blue line, fractions II and III).

Figure 2. Chromatograms from size-exclusion chromatography of extracts seen in UV at 280 nm; extracts from (a) Acacia, (b) Quercus and (c) Robinia. The numbers above the curve represent the assigned molar mass at the peak maximum within the calibrated area (blue line, fractions II and III).

Table 2. HPLC-SEC molecular weight distribution of phenolics detected at 280 nm and obtained from hydrophilic extracts from three sources (Acacia, Quercus and Robinia) after longer indoor storage.

Source of Extractives	Fractions of Phenolics								Total Peak Area: I + II + III + IV (mAU × s)
	I *		II		III		IV *
	Over 50,000 Da *		50,000–2500 Da		2500–150 Da		Below 150 Da *
	Peak Area (mAU × s)	Peak Area (%)	Peak Area (mAU × s)	Peak Area (%)	Peak Area (mAU × s)	Peak Area (%)	Peak Area (mAU × s)	Peak Area (%)
Acacia	104.6 *	15.85 *	394.8	59.82	74.1	11.23	86.5 *	13.10 *	659.9
Quercus	963.8 *	41.65 *	932.9	40.31	260.8	11.27	156.6 *	6.77 *	2314.1
Robinia	142.1 *	4.01 *	1880.2	53.07	1128.7	31.86	391.7 *	11.06 *	3542.6

* The chromatogram fraction is out of exact calibration.

Table 2. HPLC-SEC molecular weight distribution of phenolics detected at 280 nm and obtained from hydrophilic extracts from three sources (Acacia, Quercus and Robinia) after longer indoor storage.

Source of Extractives	Fractions of Phenolics								Total Peak Area: I + II + III + IV (mAU × s)
	I *		II		III		IV *
	Over 50,000 Da *		50,000–2500 Da		2500–150 Da		Below 150 Da *
	Peak Area (mAU × s)	Peak Area (%)	Peak Area (mAU × s)	Peak Area (%)	Peak Area (mAU × s)	Peak Area (%)	Peak Area (mAU × s)	Peak Area (%)
Acacia	104.6 *	15.85 *	394.8	59.82	74.1	11.23	86.5 *	13.10 *	659.9
Quercus	963.8 *	41.65 *	932.9	40.31	260.8	11.27	156.6 *	6.77 *	2314.1
Robinia	142.1 *	4.01 *	1880.2	53.07	1128.7	31.86	391.7 *	11.06 *	3542.6

* The chromatogram fraction is out of exact calibration.

The following peaks were identified as phenolics in the Acacia extract: 42,006, 36,260, 33,090, 14,734, 5101 and 1889 (Figure 2a). A fraction over 50,000 Da was also found. Smaller phenolic components were of very low concentration. The molecular weights of CTs typically range from around 500 to more than 20,000 Da [61]. It has been reported by Cesprini et al. using aqueous SEC for mimosa polyphenols that $M_w$ covered the range below 20,000 Da [49]. Polymeric proanthocyanidins with 20,000 Da and around 40 flavan-3-ol units have been found to remain soluble [17]. Acacia bark tannins are characterised as prorobinetinidins [21,62,63]. The spatial structure of Acacia tannins is highly branched, in contrast to those in quebracho, which form a linear structure composed of profisetinidins [63]. The key difference between the main units fisetinidol (quebracho) and robinetinidol (Acacia) lies in the amount of aromatic OH groups—Acacia has one more than quebracho, giving the extract better water solubility [20]. Some studies [64,65] have determined the molecular weight of tannins with an RI detector and tetrahydrofuran (THF), a polar solvent with aprotic properties traditionally used in gel permeation chromatography (GPC) [66]. In many studies tannins have undergone chemical treatment, such as acid hydrolysis [59], acetylation to increase their solubility in THF [67,68,69] or reaction with H₂O₂ [70,71], prior to GPC analysis. Comparisons between laboratories are difficult to make since the choice of solvent, standard solubility, detector type, the slope of the calibration (selection of the suitable function to fit the calibration points), varying particle size of the same column type and other settings can influence molecular weight distributions.

Oak phenolics were mainly represented by large-molecule compounds: 53,198, 38,714 and 32,114 Da; the smaller ones of 260 and 169 Da became almost unrecognisable (Figure 2b). Differences within oak heartwood (Q. petraea and Q. robur) might come from a differentiation between the outer and inner part [72]. Outer heartwood contains phenols of low $M_w$, such as gallic acid, castalin and vescalin; monomeric ellagitannins (castalagin and vescalagin); and dimeric ellagitannins [72]. On the other hand, inner heartwood is based on low-polar ellagic acid and mainly polymeric ellagitannins that cause more retention on a separation column [72]. The following phenolic compounds belonging to ellagitannins have been found in oak wood of different origin (American, French, Hungarian, Romanian and Russian) by MS detection: vescalagin ($M_w$ = 934.63), castalagin (934.63), grandinin (1066.748) and roburins (1851.251) [73]. It should be noted that these compounds were detected as building blocks present in the oak wood, most likely in oligomeric or even polymeric form. This implies that they were not necessarily naturally present as individual monomers.

SEC highlighted the difference between Robinia bark and other extracts in the intensity of eluted peaks, as well as by the presence of highly abundant small-molecule phenolic compounds visible only in the UV region and not present in the other two extracts (Figure 2c). These peaks were identified as 52,011, 26,988, 24,140, 17,338, 1197, 360 and 242 Da. The current literature focuses on lesser-used Robinia heartwood but not so much on its bark extracts. Moore and Mann have worked with acetone and found out that Robinia bark extract had the same chemicals as its wood extract, however in higher concentrations [74]. Conversely, flavonol robinetin (302 Da) and flavanonol dihydrorobinetin (304 Da) have been reported as more abundant in heartwood than in the bark extract [75]. These phenolic compounds, as well as robinetinidin (323 Da), were also recovered from methanol–water extracts in similar amounts as from acetone [76]. However, none of these compounds could be clearly assigned to our identified peaks. With the exception of fraction III, the molecular distribution of the phenolic compounds seems to be more similar to the Acacia bark extract.

(B) Saccharides

SEC chromatograms of saccharides present in the extracts are shown in Figure 3. Since the response of ELSD is quantitatively not linear [77], peak areas of saccharides were not calculated, but the molecular weights were assigned to the most intense peaks within the calibration range.

The Acacia extract showed the presence of polysaccharides of greater molecular weights reaching more than 50,000 Da (RT below 10 min was outside the calibration range), and around 32,755 Da additional oligosaccharides with 2373 Da were also observed (Figure 3a). Low-molecular saccharides with 382 and 248 Da could be identified as dimeric and monomeric saccharide moieties. This is in agreement with another study of wattle extract by Roux, in which simple sugars, such as sucrose, glucose and fructose, and a gum fraction consisting of carbohydrate polymers with an average molecular weight of up to 92,000 Da were detected [21,78]. Compared to black wattle hydrocolloids, the “stinking wattle” Acacia cambagei differed in that it contained gum of a rather lower $M_w$, between 15,000 and 27,000 Da [79]. The SEC of mimosa (i.e., black wattle) tannin extract revealed two saccharidic fractions in the range of 20,000–1000 Da, while the more concentrated fraction was eluted later (with a lower molecular weight of around 3000 Da) [49]. While this study [49] worked with the same commercial extract as we did, the differences from our results can be explained by longer indoor storage that might have shifted the $M_w$ distribution upwards over time and increased the concentration of small monosaccharides and disaccharides. Interaction between any present aromatics and saccharides could be strong (covalent bonding) [49]; thus, the polysaccharide shift to a higher $M_w$ could be led by the polycondensation of phenolics whose presence could be mistakenly understood as “pure” polysaccharides. The SEC column works as a physical sieve without interacting or cleaving covalent bonds for different compound groups, so the occurrence of both types (saccharides in ELSD and phenolics in DAD) when covalently bonded would be seen at one eluting point (more likely at a higher RT).

In the Quercus extract, the saccharidic distribution was similar to Acacia: over 50,000, 30,742, 11,603, 2338, 378 and a highly concentrated fraction with 165 Da (Figure 3b). Four of these were similar to those found by Kainuma et al. in freeze-dried extracts from French oak (Q. petraea)—71,000, 10,400, 1060, 573, 356 and 156 Da [80]. The smallest could be attributed to cis-oak and trans-oak lactones, which have approximately 156 Da [81].

The Robinia extract differed slightly from the other two in showing more large-molecule $M_w$ polysaccharides with 42,137, 25,099 and 16,390 (Figure 3c). Smaller fractions with 380 and 225 Da had similar weights to Acacia. Since in the Robinia bark extract quite large amounts of simple phenolic molecules were found, the extract cannot be considered suitable for wood protection without further purification. On the other hand, the Robinia extract could be used as a source for green synthesis due to the presence of low $M_w$ phenolic compounds.

Figure 3. Chromatograms from size-exclusion chromatography of extracts seen in ELSD; extracts from (a) Acacia, (b) Quercus and (c) Robinia. The numbers above the curve represent the assigned molar mass at the peak maximum within the calibrated area (blue line).

Figure 3. Chromatograms from size-exclusion chromatography of extracts seen in ELSD; extracts from (a) Acacia, (b) Quercus and (c) Robinia. The numbers above the curve represent the assigned molar mass at the peak maximum within the calibrated area (blue line).

3.2. Beech Wood Durability Against White-Rot Fungus

3.2.1. Selection of Temperature for Extraction and Wood Drying

We set the temperature for Quercus and Robinia accelerated solvent extractions at 80 °C to reduce thermal degradation that might reduce the phenolic content [82]. Thermal degradation and reduced amounts of robinetin and dihydrorobinetin in black locust extract have been observed after thermal treatment at 105–110 °C in other studies [76,83]. Vergara-Salinas et al. came to similar conclusions using HPLC analysis after observing a reduction in polyphenol diversity at temperatures above 100 °C and longer exposure times [84]. In the case of chestnut, CTs and HTs underwent degradation at extracting temperatures above 120 °C, with an increasing monomeric polyphenol content [85]. The spectrophotometric assay and standard used should be taken into account when comparing studies because each has a different specificity. Polyphenols from different sources have a broad spectrum of phenolics, some even being labile below 60 °C, which affects the total phenolic and total flavonoid content and antioxidant properties, though in different ways [82]. The resulting effect is dependent on the type of polyphenol, and desirable properties may disappear, though some assays promise high yields. The second reason for the lower temperature setting was that accelerated solvent extraction works at 100 times atmospheric pressure, and thus higher gains may be obtained without the need for a high working temperature.

Again, to avoid thermal degradation of polyphenols, the drying temperature of beech wood specimens after impregnation with extractives was set to ambient conditions. A hardening temperature of 103 °C after implementation of the same concentration of Acacia tannins has been found to lead to cracks and delamination of the solidified tannins in the vessels of beech wood, which then easily leach out [46]. Ambient conditioning of beech wood after impregnation with mimosa tannins has been performed by other researchers [86,87], who also applied other industrial tannin extracts (valonia, chestnut and tara) to Scots pine [88]. Using the same impregnation procedure with beech wood and wattle tannin extract, wet uptake reached 645 ± 34 kg/m³ (not published data). Weight percentage gain after impregnation can reach 4.75% (based on dry mass) when applying a 5 wt.% wattle tannin solution [46].

3.2.2. Effects of Extracts on Beech Wood Durability

The white-rot fungus Trametes versicolor, which is able to digest phenolic wood components, was used to evaluate whether these extracts (Acacia, Quercus and Robinia) were suitable for beech wood protection. T. versicolor causes non-selective white-rot decay in wood, which means that all wood components (lignin, cellulose, and hemicelluloses) are degraded almost simultaneously. The mass loss (ML) from untreated beech wood reached the 20% test validity threshold for fungus virulence [47]. It has to be noted that we conducted our tests on smaller samples and shorter exposition times than specified by EN 113 [47]. The application of two of the three extracts (from Acacia and Quercus) led to a decrease in mass loss and thus to an overall durability improvement; however, this was not the case with the Robinia extract (Figure 4).

Acacia proved to be the most efficacious. The average mass loss of beech wood treated with the Acacia bark extract was 18.5% compared with 34.5% for the untreated reference. The main reason for this is that Acacia bark is known to be a prolific source of condensed tannins (proanthocyanidins) [18,21,63], which are more resistant to microbial attacks than hydrolysable tannins [16]. Acacia extract is claimed to contain up to 80% CTs, followed by hydrocolloid gums, saccharides and smaller molecules [89]. The complexity and more oxygenated form of CT units (gallo)catechin and robinetinidol present in Acacia mearnsii bark might be the reason for its powerful defence effects [21].

Generally speaking, the higher the concentration of (condensed) tannins, the better the resistance. Although in our study the concentration was quite low (5 wt.%), the beech wood durability improvement was significant. Acacia tannin was reported as effective against another white rot (Pycnoporus sanguineus) in the same concentration impregnated into Acacia wood [90]. Acacia tannin was impregnated at different concentration levels into oriental beech (Fagus orientalis L.) and tested against T. versicolor (at 3, 6, 9 and 12 wt.%) [86] and against termites (at 6 and 12 wt.%) [87]. The extracts worked well against both, and a reduction in wood mass loss and increased termite mortality was evident in the case of the highest concentration, where the mass loss was reduced to approximately half of the control beech wood [86,87]. In a different study, where an Acacia mearnsii tannin solution was tested against T. versicolor directly in Petri dishes rather than impregnated into wood, a 20% solution delivered full resistance, while a 10% solution worked for only 20 days, after which the fungus reappeared [91]. Although several other studies have looked at A. mearnsii tannins as a wood protection agent via impregnation, they were combined with hardeners to reduce leaching [92,93,94]; thus, a direct comparison with our study is not feasible.

Quercus extract produced a slight improvement in beech wood durability (ML, 28.6%). Its inferior performance (compared with Acacia) can be explained by the presence of hydrolysable tannins in Quercus heartwood extract. Trametes spp. grow better on hydrolysable tannins than on the (condensed) tannins found in wattle [15]. Since the concentration of agents plays an important role in microbial defence, the concentration used (5% solids) affected the final resistance less than a higher tannin concentration might have done. Significantly diminished oak resistance against the same white-rot fungus has been shown to be caused in waterlogged oak by low concentrations of ellagitannins (typical HTs) due to leaching over hundreds years in water [24]. In contrast, bark extract from Quercus laeta at a concentration of 1 g/L has been reported to work only mildly against white rot exposed to treated Alnus acuminata wood by losing only 19.4% of its mass (compared with a reference ML of 75.7% without any extract treatment) [95]. The better durability in a study by Gálvez-Martínez et al. [95] might be connected with the greater content of CTs in the bark than in the heartwood, which has been reported in various wood species [96,97], whereas tree bark is generally known to be the major source of CTs [10]. A similar limitation in antifungal effects against the same white-rot fungus was observed with the sapwood and heartwood of African mahogany (Khaya ivorensis), while the bark containing more CTs showed a positive effect [97].

Wood durability after treatment with Robinia extract showed the worst performance against white-rot fungus—even worse than the untreated reference beech wood (34.5% vs. 44.7% ML). This is in accordance with the finding of Vek et al. [98], who did not observe any growth inhibition of Trametes versicolor in the presence of a hydrophilic extract of black locust heartwood at a similar concentration (5% w/v heartwood extract, as opposed to the 5% w/w bark extract we used for the impregnation solution). Moreover, neither robinetin nor dihydrorobinetin—the main compounds in the extract—showed any inhibitive action on white rot [98]. On the contrary, Robinia bark extract has been shown to possess antioxidant properties [4,99]. In Robinia bark extract there is a greater variety of chemicals than in heartwood extract [74], while differences in extractive content are observable between the inner and outer bark, as well as according to the height on the tree from which the sample is taken [100,101]. Sablík et al. [102] used black locust bark extract on beech wood and observed a mass loss of between 18.7 and 37.5% (29.1% on average), compared with our finding of 44.7%. They found that the extract of black locust heartwood was even more effective—with a 12.7% mass loss on average [102]. This ambivalent response of Robinia bark and heartwood hydrophilic extract might derive from the presence of antifungal compounds that are unaffected by the extraction conditions. Robinia contains 4.1% hydrophilic extractives in mature heartwood (close to the pit) and 8.3% in juvenile heartwood, with bark having 5.4% [75]. However, these percentages do not include non-extractable bonded flavonoids [75]. Their presence can be visualised in the wood structure by UV microspectrophotometry [6], but they cannot be quantified in the same way as extracts, since chromatographical techniques rely on the dissolution of the compounds in a solvent.

As white-rot fungus possesses a ligninolytic capacity that allows it to digest phenolic compounds [9], the small phenolic molecules present in Robinia extract might have served as an easily accessible substrate. According to Williams, who studied Sclerotinia fructigena (the fungus that causes brown-rotting of apples) in the presence of tea phenolics, phenolics with a molecular weight of at least 500 Da showed inhibitory activity [103]. Further study confirmed this, as polyflavanoids of less than ca. 800 Da might not have an affinity to protein (such as collagen in leather-tanning) [104]. Microorganism proteins may have the same property, which de facto would mean tannins with a similar molecular size would have limited defence properties. Small-molecule compounds, such as gallic acid, catechin and simple sugars, are considered “non-tannins” and are often industrially removed from the extract [105].

4. Conclusions

We found that tannins extracted from Acacia mearnsii (black wattle) bark and Quercus spp. (European oak) heartwood improved the resistance of beech wood to white-rot fungus, but tannins from Robinia pseudoacacia (black locust) bark did not. Despite the improvement we observed, our results do not point to a viable biological solution for the preservation of wood, as a mass loss of more than 5% is not deemed acceptable in the timber industry. The extent of durability enhancement depends on the type of tannins applied, their molecular dimensions and the amount of co-extracted saccharides that are often covalently bonded to tannins. Size-exclusion chromatography with diode-array and evaporative light-scattering detectors allowed us to compare three extracts for their molecular characteristics within both phenolic and saccharide moieties. Through such an analysis the potential effectiveness of plant extracts can be predicted and time saved on practical wood durability tests. Potential extracts with smaller phenolic or saccharide molecules delivering unsatisfactory protection properties (e.g., Robinia bark extract) can be filtered out in this way and potentially be considered for different purposes, for instance, green-chemistry synthesis, where smaller molecules are favoured.