FTIR Analysis of Changes in Chipboard Properties after Pretreatment with Pleurotus ostreatus (Jacq.) P. Kumm

Abstract

A commercial three-layer particleboard served as model furniture for testing pretreatment with the oyster mushroom (Pleurotus ostreatus (Jacq.) P. Kumm.) over 9-, 12-, 16-, and 20-week periods based on the effects of reducing the enzymatic resistance of component cellulose. The effects of pretreatment were assessed based on Fourier-transform infrared spectroscopy (FTIR) of the treated particleboards, wherein indexes (peaks and peak ratios) connected with parameters influencing enzymatic cellulose hydrolysis were analysed. The data were analysed in two ways: the measurement of peak heights in both primary spectra and deconvoluted spectra. The peak heights for the determination of the total crystallinity index (TCI) were measured according to narrow and broad baselines. Time and how indexes are calculated were found to be the main factors significantly influencing the values of indexes of pretreatment in most cases. Until week 9, P. ostreatus pretreatment seems to be advantageous for biofuel production, which was illustrated by decreases in the intensity of the 1735 and 1505 cm⁻¹ peaks and A₁₅₀₅/A₁₇₃₅, A₁₅₀₅/A₁₃₇₅, A₁₅₀₅/A₁₁₅₈, and A₁₅₀₅/A₈₉₆ ratios in addition to a reduction in crystallinity.

1. Introduction

The issue of waste-particleboard valorisation originates from substantial development of the furniture industry in Poland. The demand for construction and furniture materials thus determines particleboard production, which reached 4.9 mln m³ in 2018. For other similar boards from wood and wood-related materials, total production of the same year was nearly 5.9 mln m³. Taking import into account, the amount of utilised materials of this group was close to 7 mln m³, and production has since been increasing year by year [1].

Products of the furniture industry, including chipboard, have a limited lifespan, and after a certain period, some of them can be utilised in valorisation. Since particleboards are made from a lignocellulosic material—namely wood chips, which are glued with resin, usually urea–formaldehyde [2]—they can be utilised by conversion to biofuels [3].

Lignocellulosic materials offer the prospect of use as a second generation feedstock for commercial bioethanol production [4]. However, the production of bioethanol from lignocellulosic materials is hampered by enzyme resistance of this feedstock. It is thus advantageous to pretreat such materials prior to their conversion to bioethanol. Pretreatment of lignocellulosic material is an important step in the biological processing of cellulose. It aims to prepare the raw material for enzymatic decomposition. This is achieved by separating lignin and hemicelluloses from cellulose, as well as increasing the raw material porosity and decreasing the cellulose crystallinity [4]. The removal of lignin from loblolly pine wood (Pinus taeda) by enzymatic hydrolysis results in the physicochemical modification of the cell-wall structure, which in turn facilitates enzymatic access to cellulose [5]. In contrast, the crystallinity of cellulose has a negative effect on its enzymatic hydrolysis dynamics, and the latter are linearly related to the amount of cellulose.

Based on the results from the enzymatic hydrolysis of Miscanthus sinensis, it was found that reducing the crystallinity of the material also increased the hydrolysis efficiency during the initial stage. The initial hydrolysis efficiency is related to the access of cellulose hydrolysing enzymes to the substrate, and this in turn depends on the particle size of the biomass [6]. Substrate delignification and xylanase supplementation allow complete hydrolysis of hemicelluloses to monosaccharides in the <150 μm fraction of Miscanthus sinensis, and glucan to glucose conversion can reach 91%. However, cellulase activity can still be suppressed by residual lignin in the material following delignification [6].

After a 15-day biological treatment of rapeseed straw with Phanerochaete chrysosporium, the hydrolysis ratio increased from 2.2% to 6.6% despite an increase in crystallinity from 33.17% to 39.47%. This increase in crystallinity was associated with the removal of lignin and hemicelluloses. Biological pretreatment reduced the enzymatic hydrolysis rate. A linear relationship was found between the hydrolysis ratio and lignin distribution. However, both lignin loss (6.18%) and the hydrolysis ratio were low [7].

In the catalytic hydrogenolysis of biomass, delignification level and glucose yield are shown to be correlated. Lignin removal during hydrogenolysis is correlated with glucose yield, and the inclusion of lignin depolymerisation catalysts significantly improves lignin removal and enzymatic hydrolysis efficiency [8].

In another experiment, no selective loss of lignin was observed, yet the degree of glucan–glucose conversion increased. This was for conversion of the glucose content of the hydrolysate and the anhydroglucose content of the biomass. Pretreatment of pine wood with ionic liquids did not change the lignin content, and the degree of glucan conversion to glucose increased from 5% to as much as 84%. It is presumed that the increased conversion may be due to changes in the spatial distribution of lignin. The increase in glucan conversion efficiency has also been linked to an increase in the porosity of the test material (pores with diameters of 5−202 nm) and an increase in the accessible surface area of the material due to the ionic liquid treatment [9].

The yield of glucose extraction during enzymatic hydrolysis is influenced not only by the lignin itself but also by the proportions of different lignin-building monomers, i.e., syringyl, S, and guaiacyl, G. In poplar Populus trichocarpa wood, both native and hot-water pretreated, a higher S/G ratio value was found to be associated with a higher glucose yield from the raw material [10]. Similarly, a higher sugar yield was found to be positively correlated with the S/G ratio in different variants of willow Salix viminalis L. after their high-throughput pretreatments [11]. In the analysis of different eucalyptus samples, no statistically significant correlation was found between glucose and xylose yield and S/G ratio, regardless of whether pretreatment had taken place [12]. G is shown to have a greater capacity for enzyme adsorption than S [13], and the amount of S alone does not appear to positively influence the conversion rate of cellulose to glucose [14].

Lignin monomers S and G have different susceptibility for pretreatment methods. The higher lignin content caused by hemicellulose removal during pretreatment seems to be more important than lignin removal during the creation of pores allowing access of cellulases [15]. In poplar and eucalyptus wood, increases in S/G ratios are associated with a considerable reduction in lignin content [16]. Substrates richer in lignin usually have a higher S/G ratio and are more vulnerable for delignification, because S forms less stable bonds than G [17]. However, the relationship between the S/G value and the release of sugars has not been clearly explained.

The removal of hemicelluloses also increases the enzymatic decomposition of cellulose. It was observed that, after carrying out the delignification of poplar with NaClO₂ at an elevated temperature, the cellulose conversion rate was 68.2% after 34 h of enzymatic hydrolysis. After delignification, the removal of hemicellulose with diluted NaOH allowed for the enzymatic degradation of 75.2% its constituent cellulose [18].

Factors influencing enzymatic hydrolysis have been analysed using various techniques and technologies. One of the cheapest and most environmentally friendly pretreatment technologies involves the utilisation of white rot fungi. As an example, corn straw was pretreated with six white rot species, and higher lignin removal (38.3%) was achieved after a 25-day pretreatment with Pleurotus sajor-caju. The hydrolysis yield reached 71.2% in 60 h. The ethanol yield from cellulose with Candida utilis yeast fermentation was 42.1% of the theoretical limit. The ethanol yield was lower than with Saccharomyces cerevisiae, which is usually around 50% of the theoretical value; however, after ethanol distillation, a byproduct with a higher protein content was obtained [19]. Compared with physicochemical methods, biological pretreatment with white rot fungi is a long and less efficient process. For example, 20 min FeCl₃ pretreatment allowed a 98% hydrolysis yield. Physicochemical methods are, however, expensive and put a burden on the environment [19] as well, in addition to causing the formation of larger amounts of byproducts that inhibit yeast growth, such as furfural, 5-hydroxymetylofurfural, and phenol [20].

Some researchers have suggested combining biological and physicochemical pretreatment [19]. Two stage pretreatment of Chinese white poplar Populus tomentosa with fungus Lenzites betulina (L.) Fr. and hot water increased the glucose yield by 2.25 compared with sole biological pretreatment [19]. Pretreatment of California pine Pinus radiata with steam explosion pretreatment, followed by 12 weeks with Trametes versicolor (L.) Lloyd, removed 13% of lignin and increased the glucose yield 6-fold [21].

Gravimetric analysis, Fourier-transform infrared spectroscopy (FTIR), and nuclear magnetic resonance spectroscopy were used in one study for compound identification and to observe intramolecular changes and changes in the chemical composition of wood during fungal pretreatment [22]. FTIR is a useful technique because it allows the analysis of small amounts of wood samples. Most published articles describe qualitive changes in FTIR spectra during long time periods. Information about changes in spectra during a shorter fungal decay of particleboard are scant.

In the available literature, data about the biological pretreatment of particleboard waste for ethanol production are lacking.

The aim of our study was to assess the utilisation of the oyster mushroom, Pleurotus ostreatus (Jacq.) P. Kumm., a white rot fungus, in the pretreatment of commercial particleboards. Pretreated material was analysed by FTIR to assess changes in parameters connected with the ethanol yields of lignocellulosic materials.

The choice of P. ostreatus for waste particleboard pretreatment was inspired by the results of switchgrass (Panicum virgatum var. Kanlow) pretreatment with this mushroom [23]. Moreover, P. ostreatus is an easily accessible and versatile mushroom as it grows on all continents except Antarctica and is the second most cultivated edible mushroom after Agaricus bisporus [24]. P. ostreatus has the potential to be utilised in many ways, such as the mycoremediation of landfill leachate [25] and soil [26], the decomposition of textile dying effluents [27], enzymes production [28] and forage valorisation [29].

The results of particleboard pretreatment with P. ostreatus were estimated based on the total crystallinity index, TCI; lateral order index, LOI; relative lignin content, LC; delignification against polysaccharides; and changes in absorbance for characteristic peaks of lignin/carbohydrates.

The relative intensities of peaks characteristic for lignin/carbohydrates were interpreted based on information from the literature. The absorbance values of wavenumbers characteristic for chemical compounds are shown in

Table 1. Wavenumbers and their assignments.

Wavenumber cm−1	Assignment	Reference
2900	C–H and CH2 stretching	[30]
1735	unconjugated C=O in xylans (hemicellulose) [22]; unconjugated C–O in acetyl group (xylan in hardwoods, glucomannan in softwoods) [31]	[22,31]
1505	aromatic skeletal in lignin [22]; aromatic skeletal vibration C=C [31]; stretching modes of benzene ring in lignin [32]	[22,31,32]
1420	CH2 scissoring motion [30]; CH2 scissor vibration in cellulose and CH bonds in methoxyl groups in lignin [32]; C–H deformation in lignin and carbohydrates [22]; C–H in plane deformation with aromatic ring stretching, in lignin [31]	[22,31,32]
1375	CH bending modes in cellulose and hemicellulose [32]; C–H bending mode [30]; C–H deformation in cellulose and hemicellulose [31]; CH stretching in CH2 in urea–formaldehyde resin [33]	[30,31,32,33]
1158	C–O–C vibration in cellulose and hemicellulose	[31]
1120	aromatic skeletal and C–O stretching	[34]
1078	C–O stretch in cellulose and hemicellulose	[34]
896	C–H deformation in cellulose [22]; vibrational mode involving C1 and the four atoms attached to it [32]	[22,32]

.

2. Materials and Methods

2.1. Sample Preparation and Pretreatment

The three-layer particleboard produced by Swiss Krono (Żary, Poland) was made of wood shavings blending mainly from tree species: Pinus (78%), Picea (11%), Betula (4.5%), Populus (2%), Alnus (1.5%) and Larix (1%). The remaining species, Tilia, Acer and Pseudotsuga, constituted a total of 2%. The particleboard was cut perpendicular to the surface in 50 mm × 16 mm × 16 mm samples. Next, samples were sterilised in a polypropylene bag for 30 min at 121 °C and under 1 bar of pressure.

Malt extract agar medium was poured into 0.5 L glass containers with lids and a cotton gauze/cellulose wool seal to enable air diffusion. Containers were sterilised in the abovementioned conditions.

P. ostreatus inoculum, not older than 2 weeks and in the active growth phrase, was transferred from slants into containers. The containers were placed in a bacteriological incubator (Incudigit 80, JP Selecta, Barcelona, Spain) at 24 °C until mycelium covered the surface of the agar medium. An additional dish with water was placed into the incubator to maintain air humidity at 80–90% RH for fungal growth.

In each container, four sterile particleboard samples were put onto the mycelium surface according to the cut plane. This sample orientation was chosen to more readily facilitate fungal growth. Samples in closed, gauze/wool sealed containers were incubated in an incubator at 24 ± 1 °C. Air humidity was maintained using a dish with water, as mentioned above. Samples were taken after 9, 12, 16, and 20 weeks; oven dried at 105 °C; and left in zipped bags for further analysis.

2.2. Research Procedure

Half of the samples from each time interval were milled in a laboratory knife mill (LNM100, Testchem Ltd., Warsaw, Poland) operated at a 2870 rpm rotational speed, equipped with 3 high speed steel knives and a 0.425 mm sieve. Milled samples were put back into bags with unmilled material.

Milled samples were analysed with an IR spectrophotometer (IRAffinity1S, Kyoto, Japan) using an ATR accessory. Each sample was scanned in 3 spots. Each spot was scanned 32 times with 4 cm⁻¹ spectral resolution in the 4000−400 cm⁻¹ range. Spectra were processed with OMNIC^TM 8 software (Thermo Fisher Scientific, Waltham, MA, USA). Before further analysis, spectra were smoothed with the 9-point Savitzky–Golay algorithm.

Before deconvolution, each spectrum was divided in 3 ranges. Range I—around the peak characteristic for CH vibrations, about 2900 cm⁻¹; range II—the whole region between 1750 and 1200 cm⁻¹; and range III—a region between 1200 and 840 cm⁻¹ (Figure 1). Spectra were divided because it was impossible to deconvolute them as a whole while also maintaining the appropriate algorithmic sensitivity. Ranges were chosen to avoid the influence of the division of peaks after deconvolution.

Spectra were deconvoluted using the Fletcher–Powell–McCormick algorithm with high sensitivity. Peaks of Gaussian shape and full width at half maximum of 5 cm⁻¹ were assumed. The threshold for noise was individually determined for each part of the spectra.

Analysed indexes were calculated based on absorbance peak heights in deconvoluted spectra, where individual peaks were chosen based on the authors’ experience as the nearest to those representing the actual chemical bonds of interest in the literature.

2.3. Indexes for Evaluation of Pretreatment

Particleboard pretreatment was evaluated according to the following indexes:

(1) Changes in lignin content—evaluated by changes in absorbance for aromatic rings in lignin (A₁₅₀₅).

(2) Delignification selectivity—calculated as a change in the absorbance of wavenumber characteristic for lignin against the absorbance of wavenumbers characteristic for polysaccharides: hemicelluloses, holocellulose, and cellulose. Relative lignin content is shown by Equation (1), which is a modified version of an equation from another study [31].

$$L{C}_r=\frac{A_{1505}}{A_{CH}}$$

(1)

where LC_r is the relative lignin content; A₁₅₀₅ is the absorbance of wavenumber characteristic for lignin 1505 cm⁻¹; A_CH is the absorbance of wavenumbers characteristic for cellulose at 896 cm⁻¹, holocellulose at 1158 and 1375 cm⁻¹, and acetyl groups in hemicelluloses at 1734 cm⁻¹.

The peak positions were chosen as for hardwoods in [22], because this article describes the decay of sole wood, not wood-high density polyethylene composition, as in [31].

(3.1) Lateral order index, LOI, introduced by O’Connor [35], calculated as a A₁₄₂₀ to A₈₉₆ ratio.

$$LOI=\frac{A_{1420}}{A_{896}}$$

(2)

where LOI is the lateral order index; A₁₄₂₀ is the absorbance of the peak for wavenumber 1420 cm⁻¹, characteristic for CH₂ scissoring vibration assigned to crystalline fraction (amount of crystalline cellulose is connected with “crystallinity peak”, around 1420–1430 cm⁻¹); A₈₉₆ is the absorbance of the peak at wavenumber 896 cm⁻¹, which is characteristic for C–O–C β-(1,4)-glycosidic stretching vibrations assigned to the amorphous fraction (peak around 893–898 cm⁻¹ is referred to as the amorphous peak) [36].

(3.2) Total crystallinity index, TCI, defined by Nelson and O’Connor [30] based on the ratio of absorbances of peaks at 1372 and 2900 cm⁻¹.

$$TCI=\frac{A_{1375}}{A_{2900}}$$

(3)

where TCI is the total crystallinity index, A₂₉₀₀ is the absorbance of the peak at wavenumber 2900 cm⁻¹, characteristic for CH₂ stretching vibrations, assigned to the amorphous fraction; A₁₃₇₅ is the absorbance of the peak at wavenumber 1375 cm⁻¹, characteristic for CH bending vibrations.

Differently than other indexes, TCI was computed from the heights of respective peaks with individually chosen baselines (named narrow baselines) and heights of respective peaks obtained from spectra deconvolution, and an additional ratio was computed on the basis of the height of the 1375 cm⁻¹ peak with a baseline common for a group of peaks, similarly to some research for pure cellulose [30], e.g., between 1400 and 1290 cm⁻¹ (named broad baseline). A₂₉₀₀, a denominator in TCI, was the same for A₁₃₇₅ with narrow and broad baselines.

(3.3) Crystallinity R—calculated as a ratio of absorbances of peaks 1120–1078 cm⁻¹; when the ratio R increases, crystallinity decreases [37].

$$R=\frac{A_{1120}}{A_{1078}}$$

(4)

where R is the crystallinity, A₁₁₂₀ is the absorbance of the peak for wavenumber 1120 cm⁻¹ characteristic for CH bending, and A₁₀₇₈ is the absorbance of the peak for wavenumber 1078 cm⁻¹ characteristic for C–O–C stretching in a pyranose ring.

Each of the indexes was calculated based on the height of absorbance peaks and on spectral deconvolution. No A₁₃₂₇ and A₁₂₇₂ peaks were observed, so the S/G ratio was not calculated.

2.4. Selection of Indexes

Indexes measured during the pretreatment of particleboard with P. ostreatus were selected by a feature correlation method, and strong correlated indexes were eliminated. The aim of this was to reduce the number of indexes to the suggested 7 ± 2 [38] and thus reduce the amount of laborious manual information processing required.

The selection of indexes was initiated by first preparing a correlation coefficient matrix R for these indexes and later converting this to an inverted matrix R⁻¹, R⁻¹ = R^T/det(R), where R^T is a transpose of matrix R, and det(R) is a determinant of matrix R (det(R) ≠ 0 is required). When an index was strongly correlated with others, elements of R⁻¹ exceeded 10, and numerical determination of the matrix was wrong. An index was removed from the primary set of indexes if the diagonal value was ≥10. After the removal of one index, the procedure was repeated again and again until the diagonal value was <10. Results were checked by calculating identity matrix I; I = R·R⁻¹. Correlation coefficient matrix R was calculated in Statistica v.13.3 (StatSoft Ltd., Cracow, Poland) and det(R) and I in Excel.

2.5. Statistical Analysis

For each of the indexes of pretreatment, the normality of distribution was checked with the following tests: Kolmogorov–Smirnov (K–S) with Lilliefors correction (K–S–L) and Shapiro–Wilk (S–W). The homogeneity of variance for the values of pretreatment indexes measured with peak height and deconvolution methods was checked with Levene and Brown–Forsythe tests. Less favourable results were obtained for K–S–L and S–W tests, although only in 8 of 145 cases. K–S test results did not allow rejection of the H₀ hypothesis that pretreatment index values were normally distributed. For most of the indexes, the H₀ hypothesis, namely homogeneity of variance, could not be rejected on the basis of the Levene and Brown–Forsyth tests. Absorbances for wavenumber 1158 cm⁻¹ and TCI p-values indicated that the results were not significant at the level of 0.05 (0.007 and 0.003, respectively)—any conclusion based on ANOVA results should thus be taken with caution.

The influences of pretreatment time and methods for calculating indexes (height and deconvolution), along with differentiation of indexes, for particleboard pretreatment with P. ostreatus were analysed using ANOVA and F (Fisher–Snedecor) tests. Detailed analyses of average indexes were tested conducted using the Turkey test. Correlation coefficients were also subjected to analysis based on a Pearson matrix. For correlation coefficient values, descriptor synonyms were used [39]. The significance of both coefficients’ influences on indexes and the differences between average indexes were analysed for critical value p = 0.05. The statistical analysis of the results of the research was conducted with the Statistica v.13.3 software (StatSoft Poland Ltd., Cracow, Poland).

3. Results and Discussion

The results of the analysis of variance and the mean values and standard deviations of the analysed indexes are shown in

Table 2. Results of the analysis of variance and the mean values and standard deviations (SDs) of indexes of particleboard pretreatment with P. ostreatus.

F **	A1505	A1735	A1505/A1735	A1505/A1375	A1505/A1158	A1505/A896	A1375b	A1375n	A1158	A896	LOI	A1420	A2900	TCIb	TCIn	A1120/A1078
p
t	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.083	0.028	0.015	0.021	0.118	0.006	0.003	0.307	0.005
M	<0.001	<0.001	0.738	<0.001	<0.001	0.468	0.425	ND	<0.001	<0.001	0.003	0.003	0.004	<0.001	ND	ND
t × M	<0.001	<0.001	0.774	0.032	0.015	0.115	<0.001	ND	0.792	0.714	0.094	0.183	0.594	0.013	ND	ND
	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD	A ± SD
t, w
0	0.016 c,* ± 0.002	0.006 b ± 0.001	2.54 b ± 0.3	2.41 c ± 0.4	1.49 c,d ± 0.6	2.37 c ± 0.56	0.007 a,b ± 0.001	0.006 a ± 0.001	0.012 a,b ± 0.005	0.007 a,b ± 0.001	1.22 a,b ± 0.34	0.008 a ± 0.001	0.005 a ± 0.001	1.45 a,b ± 0.3	0.99 a ± 0.05	0.75 b ± 0.01
9	0.010 c ± 0.002	0.006 b ± 0.002	1.75 a ± 0.12	1.42 a,b ± 0.24	1.14 b,c ± 0.3	1.65 b ± 0.16	0.007 a,b ± 0.001	0.005 a ± 0.001	0.009 a ± 0.004	0.006 a ± 0.002	1.07 a,b ± 0.1	0.006 a ± 0.001	0.005 a ± 0.001	1.49 a,b ± 0.29	1.05 a ± 0.08	0.61 a,b ± 0.14
12	0.016 b ± 0.004	0.009 c ± 0.002	1.73 a ± 0.1	1.73 b ± 0.09	1.65 d ± 0.31	2.62 c ± 0.31	0.009 c ± 0.002	0.006 a ± 0.001	0.010 a,b ± 0.004	0.006 a ± 0.002	1.31 b ± 0.12	0.008 a ± 0.001	0.006 a,b ± 0.001	1.65 b ± 0.54	1.00 a ± 0.08	0.51 a ± 0.03
16	0.011 b ± 0.004	0.006 b ± 0.002	1.96 a ± 0.49	1.43 a,b ± 0.69	0.79 a,b ± 0.13	1.36 a,b ± 0.35	0.008 b,c ± 0.002	0.007 a ± 0.001	0.014 b ± 0.005	0.008 b ± 0.002	1.01 a ± 0.22	0.008 a ± 0.002	0.007 b ± 0.001	1.21 a ± 0.2	0.94 a ± 0.19	0.50 a ± 0.04
20	0.006 a ± 0.002	0.004 a ± 0.001	1.46 a ± 0.32	1.13 a ± 0.59	0.55 a ± 0.15	0.84 a ± 0.19	0.005 a ± 0.001	0.005 a ± 0.001	0.011 a,b ± 0.005	0.007 a,b ± 0.003	1.03 a,b ± 0.13	0.007 a ± 0.002	0.005 a ± 0.002	1.23 a ± 0.17	0.85 a ± 0.13	0.53 a ± 0.11
M
h	0.010 a ± 0.004	0.005 a ± 0.002	1.91 a ± 0.45	1.44 a ± 0.65	1.35 b ± 0.62	1.87 a ± 0.82	0.007 a ± 0.002	0.006 ± 0.001	0.008 a ± 0.002	0.005 a ± 0.001	1.24 b ± 0.2	0.007 a ± 0.001	0.006 b ± 0.001	1.20 a ± 0.12	0.97 ± 0.12	ND
d	0.014 b ± 0.005	0.007 b ± 0.002	1.95 a ± 0.53	1.91 b ± 0.53	0.95 a ± 0.37	1.74 a ± 0.67	0.007 a ± 0.002	ND	0.015 b ± 0.004	0.008 b ± 0.001	1.02 a ± 0.22	0.008 b ± 0.002	0.005 a ± 0.001	1.61 b ± 0.37	ND	0.59 ± 0.12

* Means with the different letters in the same column are significantly different (p ≤ 0.05) by Tukey’s multiple range test. ND: not determined. ** F—factor; p—p-value; t—time; M—method; w—week; A—average value; SD—standard deviation; h—height; d—deconvolution; b—broad base; n—narrow base; A₈₉₆, A₁₁₅₈, A₁₃₇₅, A₁₄₂₀, A₁₅₀₅, A₁₇₃₅, and A₂₉₀₀—absorbance for wavenumbers characteristic for cellulose 896 cm⁻¹, holocellulose 1158 cm⁻¹ and 1375 cm⁻¹, crystalline fraction 1420 cm⁻¹, lignin 1505 cm⁻¹, hemicelluloses (acetyl groups) 1735 cm⁻¹, and amorphous fraction 2900 cm⁻¹, respectively; A₁₅₀₅/A₈₉₆, A₁₅₀₅/A₁₁₅₈, A₁₅₀₅/A₁₃₇₅, and A₁₅₀₅/A₁₇₃₅—relative lignin content against cellulose A₈₉₆, holocellulose A₁₁₅₈ and A₁₃₇₅, and hemicelluloses A₁₇₃₅, respectively; LOI—lateral order index; TCI—total crystallinity index.

; the corresponding matrix of correlation coefficients is in

Table 3. Pearson matrix of correlation coefficients for indexes of particleboard pretreatment with P. ostreatus.

Parameter	A1505	A1735	A1505/A1735	A1505/A1375	A1375b	A1505/A1158	A1158	A1505/A896	A896	LOI	A1420	TCIb	A2900
A1505	1.000
A1735	0.803 a	1.000
A1505/A1735	0.615 a	0.042	1.000
A1505/A1375	0.755 a	0.356 a	0.825 a	1.000
A1375b	0.473 a	0.664 a	−0.085	−0.188	1.000
A1505/A1158	0.502 a	0.365 a	0.378 a	0.390 a	0.195	1.000
A1158	0.425 a	0.373 a	0.221	0.393 a	0.132	−0.498 a	1.000
A1505/A896	0.769 a	0.597 a	0.506 a	0.586 a	0.326	0.912 a	−0.190	1.000
A896	0.307	0.322	0.108	0.257	0.163	−0.528 a	0.931 a	−0.315	1.000
LOI	0.288	0.186	0.227	0.148	0.206	0.772 a	−0.524 a	0.699 a	−0.579 a	1.000
A1420	0.618 a	0.537 a	0.330	0.403 a	0.425 a	0.011	0.624 a	0.217	0.641 a	0.225	1.000
TCIb	0.522 a	0.584 a	0.116	0.303	0.377 a	0.103	0.288	0.342	0.193	0.033	0.333	1.000
A2900	−0.002	0.112	−0.169	−0.436 a	0.618 a	0.083	−0.120	0.010	−0.010	0.133	0.085	−0.476 a	1.000

^a—Statistically significant value of correlation coefficient for p = 0.05. A₈₉₆, A₁₁₅₈, A₁₃₇₅, A₁₄₂₀, A₁₅₀₅, A₁₇₃₅, and A₂₉₀₀—absorbance for wavenumbers characteristic for: cellulose 896 cm⁻¹, holocellulose 1158 cm⁻¹ and 1375 cm⁻¹, crystalline fraction 1420 cm⁻¹, lignin 1505 cm⁻¹, hemicelluloses (acetyl group) 1735 cm⁻¹, and amorphous fraction 2900 cm⁻¹, respectively; A₁₅₀₅/A₁₁₅₈, A₁₅₀₅/A₁₃₇₅, and A₁₅₀₅/A₁₇₃₅—relative lignin content against cellulose A₈₉₆, holocellulose A₁₁₅₈ and A₁₃₇₅, and hemicelluloses A₁₇₃₅, respectively; LOI—lateral order index; TCI—total crystallinity index; b—broad base.

.

The results of the analysis of variance and the mean values and standard deviations of analysed indexes are shown in Table 2; the corresponding matrix of correlation coefficients is in Table 3.

The main factors, e.g., time and methods for the calculation of indexes, were, in most cases, statistically significant variables describing the variability of indexes during particleboard pretreatment with P. ostreatus. Statistically insignificant against the time of pretreatment were variability in absorbance at wavenumbers 1375 cm⁻¹ (p = 0.083) and 1420 cm⁻¹ (p = 0.118) and TCI value (p = 0.305), and against method of calculation were the variability of A₁₅₀₅/A₁₇₃₅, A₁₅₀₅/A₈₉₆ ratios and absorbance for wavenumber 1375 cm⁻¹ (for board base), with p-values of 0.738, 0.468 and 0.425, respectively. The interaction of the main factors significantly influenced the variability in absorbances for wavenumbers 505, 1735 and 1375 cm⁻¹ (for broad base) and A₁₅₀₅/A₁₃₇₅ and A₁₅₀₅/A₁₁₅₈ ratios and the TCIb (for broad base) ratio.

Changes in the described indexes of particleboard pretreatment with P. ostreatus are presented in the following subsections.

3.1. Delignification

In week 9, the value of A₁₅₀₅ (Figure 2), caused by vibrations in aromatic rings in lignin and vibrations in formaldehyde, decreased (on average from 0.016 to 0.010), which indicates lignin degradation by P. ostreatus (Figure 3). In week 12, A₁₅₀₅ increased, perhaps as a result of formaldehyde production during the degradation of methoxyl groups in lignin by the fungus [40], and this assumption is supported by an increase in A₁₇₃₅ (Figure 3), because peak 1735 cm⁻¹ is present in the formaldehyde spectrum [41]. Formaldehyde production may compensate for a decrease in A₁₅₀₅ caused by lignin degradation.

In weeks 16 and 20, the expected successive decrease in A₁₅₀₅ resulting from lignin degradation appears (Figure 2); however, changes in A₁₇₃₅, A₁₅₀₅, and A₈₉₆ during the whole experiment do not point to an exclusive role of formaldehyde, as changes in the absorbances of these peaks do not present the same trend. This is clearly visible from the lack of correlation between A₁₇₃₅ and A₈₉₆ and between A₁₅₀₅ and A₈₉₆, whereas A₁₇₃₅ and A₁₅₀₅ are highly correlated (r = 0.803), Table 3). The sources of changes in values for these wavenumbers should be investigated in further research.

The change in the A₁₅₀₅ value resulting from deconvolution is in line with that of peak height but with a higher mean value, amounting to 0.014 and 0.010, respectively (Table 2). Deconvolution of spectra allowed for excluding changes in absorbance caused by changes in the heights of neighbour peaks.

3.2. Delignification Based on Comparison with Polysaccharides

In week 9, the A₁₅₀₅/A₁₇₃₅, A₁₅₀₅/A₁₃₇₅, A₁₅₀₅/A₁₁₅₈, and A₁₅₀₅/A₈₉₆ ratios decreased (Figure 3 and Figure 4), which indicates selective degradation of lignin over hemicelluloses, holocellulose, and cellulose. From week 12, the tendency of changes in the A₁₅₀₅/A₁₇₃₅ ratio differed from those of other ratios. The A₁₅₀₅/A₁₃₇₅, A₁₅₀₅/A₁₁₅₈, and A₁₅₀₅/A₈₉₆ ratios increased between weeks 9 and 12 and later decreased until week 20.

3.2.1. Delignification Based on Comparison with Hemicelluloses

In week 9, A₁₅₀₅/A₁₇₃₅ decreased and later plateaued (Figure 4). The initial decrease was caused mainly by a decrease in A₁₇₃₅, which might be a result of hemicellulose degradation by P. ostreatus. The further plateau was a result of simultaneous changes in ratio A₁₅₀₅/A₁₇₃₅, and the absolute values are presented in Figure 3.

3.2.2. Delignification Based on Comparison with Holocellulose

The A₁₅₀₅/A₁₃₇₅ ratio, related to the amount of lignin compared with hemicelluloses, decreased during the experiment, which may indicate selective delignification in preference to degradation of hemicelluloses (Figure 4).

The decrease in A₁₅₀₅/A₁₃₇₅, for both peak heights and deconvolution, was overestimated due to an increase in the value of A₁₃₇₅ in week 16 (Figure 5), and these changes could be caused by the appearance of metabolites and further removed in week 20. Peak A₁₃₇₅ (Figure 2) also originated from C–H w CH₂ vibrations in urea–formaldehyde resin [33], so resin degradation might cause an increase in the A₁₅₀₅/A₁₃₇₅ ratio.

The A₁₅₀₅/A₁₁₅₈ ratio (Figure 6), similar to A₁₅₀₅/A₁₃₇₅, indicates the amount of lignin compared with holocellulose and presents the same tendency.

Absorbance for peak A₁₁₅₈ decreased in week 9 (on average from 0.012 to 0.009, Table 2), successively increased until week 16 (up to 0.014), and finally decreased (to 0.011, Figure 7) in week 20. During the experiments, values from deconvolution had similar oscillations as heights, although absolute values were almost twice those of 0.015 and 0.008, respectively (Table 2).

3.2.3. Delignification Based on Comparison with Cellulose

The A₁₅₀₅/A₈₉₆ ratio indicates the amount of lignin compared with cellulose. Values of this ratio, estimated from peak heights, changed similarly to A₁₅₀₅/A₁₃₇₅, presenting a decreasing trend, except from weeks 9 to 12 (Figure 4), and the correlation coefficient between these ratios was 0.586 (Table 2). Similarly, the A₁₅₀₅/A₈₉₆ ratio increase was caused by the aforementioned increase in the value of A₁₅₀₅ (Figure 3), which is demonstrated by the high correlation r = 0.769 (Table 2). A₁₅₀₅/A₈₉₆ ratio decreased from weeks 0 to 9, both for the numerator and denominator (Figs. 3 and 7). Later, in week 12, because of the decrease in the peak A₁₅₀₅ value (Figure 3), A₁₅₀₅/A₈₉₆ increased, and the A₈₉₆ value (Figure 7) presented a minor change between weeks 9 and 12.

3.3. Changes in Indexes of Crystallinity

Values of LOI ratios calculated from peak heights and from deconvoluted spectra both decreased in week 9 (from 1.22 to 1.07, Table 2), increased to 1.31 in week 12, and decreased in week 16, plateauing until the end of experiment (Figure 8). Changes in LOI indicate increases in crystallinity to week 12 after its initial decrease. Next, from week 12, crystallinity decreased and later stabilised. The decrease in LOI in week 9 can be explained by lignin decomposition, as the value of A₁₄₂₀ is connected with the presence of methoxyl groups in lignin [32]. The influence of lignin on the decrease in LOI is supported by the simultaneous decrease in A₁₅₀₅ (Figure 3).

Figure 9 shows changes in A₁₄₂₀, a numerator of the LOI ratio. The value of A₁₄₂₀ decreased in week 9 from 0.008 to 0.006; it decreased to 0.008 in week 16; and a small decrease was observed in week 20. A decrease in A₁₄₂₀ in week 9 (Figure 2) reveals a reduction in crystallinity, and the degradation of hemicelluloses because A₁₄₂₀ originates, for example, from C–H vibrations in methoxyl groups in hemicelluloses [22]. Additionally, a decrease in this peak, which originates from –OCH₃ vibrations in lignin [32] or even whole fragments of lignin, is suggestive of lignin demethoxylation, as there is a simultaneous decrease in the A₁₅₀₅ value (Figure 3). The correlation between A₁₄₂₀ and A₁₅₀₅ is high, 0.618. Later changes were not consistent with changes in lignin and hemicellulose content, and additional research is required to explain this observation.

Indexes of Crystallinity

The TCI for broad baseline increased until week 12, from 1.45 to 1.65, and later decreased to 1.21, stabilizing to a near-plateau (Figure 10). TCI for a narrow baseline changed in a similar way (r = 0.954, almost full correlation, Table 3), but the average value was 19% lower than for the broad baseline. TCI calculated on the basis of deconvolution increased in week 9 and decreased irregularly until the end of experiment. The course of TCI suggests no changes in cellulose crystallinity in particleboard pretreated by P. ostreatus.

The value of A₂₉₀₀ (Figure 9), the denominator in TCI, was chosen to be the same for TCI calculated on the basis of A₁₃₇₅ with broad and narrow baselines [30].

Generally, in the definition of TCI, A₂₉₀₀ was utilised as an internal standard because this peak is relatively unchanged in cellulose. For the internal standard, the influence of random scattering of radiation on particles in the sample and a different amount of analysed material in KBr pellets had to be minimised [30].

During the activity of P. ostreatus, changes in chemical bonds measured at A₂₉₀₀ cannot be excluded, so the utilisation of A₂₉₀₀ as an internal standard is doubtful. Moreover, particleboards with UF resin are not the same as those with pure cellulose, and changes in A₂₉₀₀ might depend on changes in other components of the particleboards.

As indexes of crystallinity, LOI and TCI were elaborated on the basis of research conducted on sole cellulose [30], so these indexes do not include the influences of other wood components, e.g., hemicelluloses and lignin. The influence of these components makes changes in crystallinity based on FTIR spectra difficult to interpret. It is recommended to extend the scope of this research by using other methods, which will allow more conclusive interpretation of the effects of particleboard pretreatment by P. ostreatus.

The values of R = A₁₁₂₀/A₁₀₇₈, another ratio connected with crystallinity, noticeably decreased in week 9 (from 0.75 to 0.61, Table 2), and similarly in week 12, and were thereafter almost unchanged until the end of the experiment (Figure 11).

3.4. Results of Feature Correlation Method

As high values of correlation coefficients between indexes suggested their redundance, some indexes were excluded: the amount of lignin compared with the amount of cellulose, A₁₅₀₅/A₈₉₆ (almost full correlation with A₁₅₀₅/A₁₁₅₈, r = 0.912); absorbance at A₁₇₃₅ (very high correlation with A₁₅₀₅, r = 0.803); absorbance at A₈₉₆ (almost complete correlation with A₁₁₅₈, r = 0.931); the amount of lignin compared with the amount of holocellulose A₁₅₀₅/A₁₃₇₅ (very high correlation with A₁₅₀₅/A₁₇₃₅, r = 0.825); absorbance at A₂₉₀₀ (high correlation with A_1375b, r = 0.618); and absorbance at A₁₁₅₈ (high correlation with A₁₄₂₀, r = 0.624). Finally, seven indexes were used for pretreatment estimation: LOI; TCIb (for broad baseline); amount of lignin compared with hemicellulose, A₁₅₀₅/A₁₇₃₅; amount of lignin compared with holocellulose, A₁₅₀₅/A₁₁₅₈; and absorbances A₁₅₀₅, A₁₄₂₀, and A₁₃₇₅ for wavenumbers characteristic of lignin (1505 cm⁻¹), crystalline fraction (1420 cm⁻¹), and holocellulose (1375 cm⁻¹), respectively. This selection of criterial parameters, with the help of the feature correlation method [38], allows one to increase the perception of information processing and reveals important indexes for characterizing particleboard pretreatment by P. ostreatus at a sufficient level.

4. Conclusions

How to sufficiently utilise furniture industry wastes is an increasing problem, and a possible solution is to convert them into biofuels such as ethanol. However, this process requires pretreatment, which was conducted here.

The main factors of time and how to calculate indexes were found, in most cases, to be statistically important variables influencing the differentiation of indexes of particleboard pretreatment with P. ostreatus. The interaction time and the influence of indexes were determined to be statistically significant based on the absorbance of peaks at 1505, 1735, and 1375 cm⁻¹ for the broad baseline along with the influences on the ratios A₁₅₀₅/A₁₃₇₅ and A₁₅₀₅/A₁₁₅₈ and TCI for the broad baseline.

Up to week 9, pretreatment seemed beneficial for biofuel production, e.g., decreases in values measured at 1735 and 1505 cm⁻¹ values and A₁₅₀₅/A₁₃₇₅, A₁₅₀₅/A₁₇₃₅, A₁₅₀₅/A₁₃₇₅, A₁₅₀₅/A₁₁₅₈, and A₁₅₀₅/A₈₉₆ ratios were indicative of lignin and hemicellulose decomposition in addition to selective delignification by P. ostreatus. Similarly, positive changes were noticed in the case of crystallinity indexes (LOI, TCI, R). The breakdown of the direction of changes after week 16 might have been caused by the fungal defence being overcome, which occurred between weeks 16 and 20. Changes that occurred after week 9 of pretreatment should be analysed using different analytical methods, as FTIR does not allow one to formulate unequivocal conclusions.