Introduction

The continuous advance of industrialization demanded a great need for energy, mainly from fossil fuels, which resulted in environmental pollution and higher costs in the energy supply of homes and industries [1]. According to estimates, only 19% of global energy demand is met by renewable sources, of which biomass contributes 9% of the total, while 10% consists of wind, geothermal, solar, and biofuels [2]. The use of renewable energies is an important strategy to increase independence from fossil fuels, providing a reduction in environmental problems, the decentralization, and sustainability of the countries’ energy matrix [3, 4]. Forest plantations, including in degraded areas, have been encouraged as a solution to the energy problems of modern society [2].

Forest biomass has been used for several energy purposes, directly as firewood and chips, or transformed into solid fuels, such as charcoal for domestic use (e.g., barbecue) and to supply the demand of the steel making process [5]. Charcoal production in Brazil represents 11% of the total world production, and its main application is in the steel industry for the reduction of iron ore and the production of pig iron [6] to produce the so-called “green steel.” The increase of charcoal production in the industrial plants is a major challenge of a sustainable character, in view of the great dependence on coal and its impacts on the environment [7].

The genera Eucalyptus stands out in the energy production due to its rapid growth, high volumetric productivity, and wood properties suitable for cogeneration and conversion into steel plant charcoal [7,8,9,10]. Currently, the forest plantations in Brazil cover an area of 7.83 million ha, in which 5.67 million (72.4%) consist of Eucalyptus crops [6]. Therefore, it is required to select and indicate the genetic materials most suitable for the production of charcoal with high productivity and adequate quality as reducing agent in blast furnaces of companies.

For the charcoal production, it is essential to consider some qualitative criteria of wood, such as high values of basic density (≥ 500 kg m−3) and content of lignin (≥ 28%), to ensure high yield and satisfactory quality of the bioreducer [11,12,13]. The wood chemical composition, specifically the carbohydrates and lignin contents, and the anatomical features, such as fibers of thicker cell walls with a smaller width, are related to the wood performance during pyrolysis and can directly influence the mass balance of the thermochemical conversion process of wood into charcoal [14, 15].

In addition to the productivity of forests destined for charcoal production and wood quality [11], several studies have highlighted the importance of evaluating the chemical composition of Eucalyptus wood, especially the lignin macromolecule [16, 17]. Woods with high lignin contents and low syringyl/guaiacyl (S/G) ratios are prone to increase the charcoal yield [14, 16]. Araújo et al. [18] have investigated this issue on Eucalyptus wood and charcoal, but reported no correlation between the S/G ratio and the carbonization process yield for Eucalyptus clones. On the other hand, Castro et al. [16] reported a significant negative correlation between S/G ratio and charcoal yield for Eucalyptus clones. In short, few studies were conducted to better understand the relationship between wood chemical constituents and gravimetric yielded of wood/charcoal from fast-growing Eucalyptus plantations. Thus, the few results reported in the literature suggest that the S/G ratio is not the key factor to control carbonization mass balance [18].

Here, we hypothesize that phenols derived from lignin, such as the guaiacyl (G) and syringyl (S) mass present in the wood need to be considered for the correct classification of superior genetic materials for the bioreducer production. Although the S/G ratio is used to classify Eucalyptus clones for bioenergy [11, 19], research on the lignin quality specifying the G and S content per wood mass is incipient in the literature. Thus, the aim of this research was to analyze the effect of Eucalyptus spp. in the carbonization mass balances and; thus, to understand how the wood quality can influence the yields of the products and by-products of pyrolysis. The detailed analysis of the carbonization mass balances and the wood and charcoal properties will allow a better understanding of this thermochemical process and may assist in the choice of the best genetic materials for charcoal production, a relevant industrial input in Brazil.

Material and Methods

Wood Sampling

In this study, 14 genetic materials of Eucalyptus spp. from a clonal test installed at 3 m × 3 m spacing by the Plantar group, located in the Curvelo town, Minas Gerais state (18° 45′ 06.44″ S; 44° 33′ 39.16″ O and 690 m altitude) were analyzed at 81 months of age (Table 1).

Table 1 Eucalyptus clones evaluated in the study
Full size table

Vwb commercial volume of wood, with bark, DBHwb diameter at breast height, with bark, BD wood basic density. Mean ± standard deviation. Source: Protásio et al. [11, 20]

Three trees of medium diameter were sampled by genetic material. The diameter at breast height (1.30 m from the soil) and the commercial height of 42 trees were measured. Disks with 2.5 cm thick were removed in five longitudinal positions, at 2, 10, 30, 50, and 75% of the tree commercial height, considered up to a minimum diameter of 4.0 cm with bark, as recommended by Downes et al. [21]. The disks removed from the trees were cut into four wedges passing through the pith. Two opposite wedges from all the longitudinal positions were used to determine the chemical composition of the wood, carbonizations on a laboratory scale, and ultimate analysis of the charcoal produced.

Chemical composition of the wood

One sample composed of the disks taken from all the longitudinal sampling positions was considered for the wood chemical characterization. Samples ground, sieved, and retained between 200 mesh (0.074 mm) and 270 mesh (0.053 mm) sieves were used for ultimate analysis. The fraction that was retained between the 40 (0.420 mm) and 60 mesh (0.250 mm) sieves was used in the other wood analyses.

The insoluble lignin content (Klason) was obtained according to the methodology of Gomide and Demuner [22], and the soluble lignin content (Klason) in sulfuric acid was determined according to the method of Goldschimid [23]. The total lignin was obtained by the sum of insoluble and soluble lignin.

The lignin degradation was carried out by the alkaline oxidation method of the wood with nitrobenzene, followed by high-performance liquid chromatography (HPLC), for the quantification of its derivatives, according to the methodology of Chen [24], with some adaptations described in Araújo et al. [18] and Protásio et al. [20]. HPLC analyses were carried out in a chromatograph Shimadzu® model CBM-20A (Kyoto-Japan), equipped with an LC-18 column and UV-SPD-20A detector, operating at 280-nm wavelength. The mobile phase was composed of acetonitrile:water (1:6 v/v), pH adjusted to 2.6 with trifluoroacetic acid, and injected at a flow rate of 1.0 mL min−1. The oven temperature was adjusted to 40 °C. The injection volume of the lignin samples was 20 μL and the lignin oxidation products were quantified using the vanillin and syringaldehyde standards. Subsequently, the mass of syringyl and guaiacyl units per kilogram of dry wood and the S/G ratio were determined. It is worth mentioning that the masses of the syringyl and guaiacyl units were not determined directly, but by quantifying the mass of syringaldehyde and vanillin, respectively, which are the main products derived from the oxidation of these constituent units of the lignin macromolecule.

The extractives soluble in acetone and ash contents were determined according to the procedures described in the standards T 280 pm-99 [25] and D1762-84 [26], respectively. The carbon (C), hydrogen (H), nitrogen (N), and sulfur (S) contents were obtained by ultimate analysis, performed in an Elementar® analyzer Vario Micro Cube (Langenselbold-Germany). The oxygen content was obtained by difference [O (%) = 100 − C (%) − H (%) − N (%) − S (%) − Ashes (%)]. All elements were quantified in relation to the dry wood mass. The data regarding the chemical composition of the analyzed woods are available in the study of Protásio et al. [20].

Charcoal production and process mass balance

The carbonizations in the laboratory scale were performed for each tree in an electric oven (muffle) with a metal reactor connected to a water-cooled condenser, which is coupled to a condensable gas collecting bottle. Approximately 500 g of wood of the opposite wedges from the disks removed at different heights in the trees were used in each carbonization. The wood samples were previously oven dried at 103 ± 2 °C. The initial temperature of carbonization was 100 °C and the final temperature was 450 °C. The electric oven remained stabilized at 450 °C for 30 min. The heating rate of the electric oven was 100 °C h−1 and the total time of carbonization was 4 h.

The gravimetric yield in charcoal (GYC) of each sample from the different Eucalyptus clones was calculated on a dry basis after carbonization (Equation 1).

$$\mathrm{GYC}=\left(\frac{\mathrm{DMC}}{\mathrm{DMW}}\right)\times 100$$
(1)

where, GYC is the gravimetric yield in charcoal (%, dry basis); DMC, dry mass of charcoal (g); and DMW, dry mass of wood (g).

The same procedures adopted for wood were followed for the ultimate analysis and ash determination of the charcoal (see item “Chemical composition of the wood”). For the determination of C, O, and H balances, the contents of these elements present in charcoal and wood and the GYC were considered (Equations 2, 3, and 4). The percentages of C, O, and H retained in the charcoal structure, based on the dry mass of each chemical element present in the wood, were considered in the calculations. Finally, the percentages of C, O, and H volatilized were obtained by difference, which is, subtracting from 100% the percentage of each element retained in charcoal.

$$\mathrm{RC}=\mathrm{GYC}\times \left(\frac{\mathrm{CCC}}{\mathrm{CCW}}\right)$$
(2)

where, RC is the retained carbon in charcoal (%); GYC, gravimetric yield in charcoal (%, dry basis); CCC, carbon content in charcoal (%); and CCW, carbon content in wood (%).

$$\mathrm{RO}=\mathrm{GYC}\times \left(\frac{\mathrm{OCC}}{\mathrm{OCW}}\right)$$
(3)

where, RO is the retained oxygen in charcoal (%), GYC, gravimetric yield in charcoal (%, dry basis); OCC, oxygen content in charcoal (%); and OCW, oxygen content in wood (%).

$$\mathrm{RH}=\mathrm{GYC}\times \left(\frac{\mathrm{HCC}}{\mathrm{HCW}}\right)$$
(4)

where, RH is the retained hydrogen in charcoal (%); GYC, gravimetric yield in charcoal (%, dry basis); HCC, hydrogen content in charcoal (%); and HCW, hydrogen content in wood (%).

Statistical analysis

The data were submitted to the Bartlett, Durbin-Watson, and Shapiro-Wilk tests, at the level of 5% of significance, to verify the homogeneity of the variances, autocorrelation of the residuals, and normality, respectively. Subsequently, univariate analyses of variance were performed, in which a completely randomized design was adopted, aiming to evaluate the effect of clones on the wood chemical characteristics and the carbonization process, considering three repetitions. The following variables were analyzed in the wood: total lignin content, S/G ratio, grams of guaiacyl kg−1 of wood, grams of syringyl kg−1 of wood, content of extractives soluble in acetone, and ultimate analysis composition. The following variables were analyzed in the charcoal: contents of carbon (C), hydrogen (H), nitrogen (N), and sulfur (S). In the pyrolysis process were analyzed the gravimetric yield in charcoal, retained C, volatilized C, retained H, volatilized H, retained O, and volatilized O. The Scott-Knott test was used at levels of 5% and 10% of significance, for the multiple comparisons of means and univariate grouping of Eucalyptus clones.

Multivariate statistical techniques are more efficient in grouping and selecting Eucalyptus clones compared to univariate procedures [19]. Thus, the clones were grouped by principal component analysis (PCA), specifically by dispersion analysis of scores. Variables with the greatest contributions to the principal components (PC) were identified based on the eigenvectors.

Canonical correlation analysis was applied to verify the linear correlations between the pyrolysis mass balance and wood chemical characteristics. Each variable was represented by 3 replicates and 14 Eucalyptus clones, totaling 42 observations. This multivariate statistical analysis was performed between the group formed by the wood characteristics (independent variables - X) and the group formed by the variables of the pyrolysis process (dependent variables - Y). Group X was represented by the following variables: total lignin, extractives soluble in acetone, S/G ratio, and mass of guaiacyl and syringyl units per kilogram of dry wood. On the other hand, group Y was composed of the following variables: gravimetric yield in charcoal and the percentages of C, H, and O retained in the charcoal. Thus, it was possible to determine four canonical functions or four pairs of canonical statistical variables, interpreted by the canonical charges and the crossed canonical charges. The level of explained variance, that is, the percentage of variance in the dependent canonical statistical variable that can be explained by the independent canonical statistical variable, was determined by squaring the canonical correlation (canonical R2). The same procedure was adopted by Protásio et al. [27] in the canonical correlation analysis between the characteristics of charcoal.

All statistical analyses were performed in R software, version 3.4.3 [28], using the packages car [29], lmtest [30], CCA [31], CCP [32], and yacca [33].

Results and discussion

Chemical composition of the wood and charcoal

Clone effect was verified by the F (Table 2) and Scott-Knott tests in the contents of total extractives, total lignin, S/G ratio, mass of guaiacyl, and syringyl per kg of wood (Table 3). For the content of extractives soluble in acetone, there was the formation of four groups of clones, in which clones 1039 (E. grandis hybrid), 1037 (Eucalyptus spp.), and 1036 (E. urophylla) had higher mean values (2.48%). The choice of these genetic materials can result in an increase in gravimetric yield in charcoal (GYC) due to the higher content of extractives. These chemical constituents may have a direct correlation with the charcoal production, depending on their chemical constitution and resistance to thermal degradation [16, 17]. Extractives are complex macromolecules that can increase the energy value of biomass [34] and positively influence the energy yield and the mass balance of pyrolysis [14]. The extractives present different form and composition, which may explain the variability (0.7 to 2.6%) and the formation of four groups among the studied clones. The extractives soluble in acetone represent chemical components as fatty acids, resin acids, sterols, waxes, and non-volatile hydrocarbons [25]. These substances can present high molecular mass and thermal degradation range between 250 and 505 °C [35], consequently, contributing to increase the charcoal production. Despite the lower proportion of extractives compared to lignin macromolecule, these secondary components should not be neglected in the classification of the best Eucalyptus clones for the production of bioreducer.

Table 2 Variance analyses for the wood chemical composition of the Eucalyptus spp. clones
Full size table
Table 3 Wood chemical composition of the Eucalyptus spp. clones
Full size table

Regarding the total lignin content, clones 1039, 1025, 1009, 1033, 1006, 1005, and 1008 were statistically similar and formed the group with the highest mean (31%). This value is considered suitable, based on studies available in the literature for the genera Eucalyptus, which reported lignin contents ranging from 28 to 32% [7, 19]. In our study, low variability of the clones was verified for the total lignin content in the wood. This is probably related to genetics pre-selections made by forestry companies in Brazil in which materials with lower percentages of lignin were disregarded. Lignin is a macromolecule with a complex chemical structure considered in the literature as the main responsible for the charcoal production due to its high thermal stability [15]. Thus, clones 1039 (E. grandis hybrid) and 1025 (E. camaldulensis hybrid) stood out due to the higher proportion of lignin and extractives soluble in acetone. Considering the sum of the percentages of total lignin and extractives soluble in acetone, the clones with the greatest possibility of presenting better results in the carbonization C balance are in order: 1039 (1st), 1025 (2nd), and 1033 (3rd).

Considering the S/G ratio, 3 groups of clones were formed, in which 64% of the evaluated genetic materials are part of the group of lowest mean. Clones 1039 and 1025 are in this group and numerically presented the lowest S/G ratios. Lower S/G ratio is required for the classification and selection of genetic materials for charcoal production, as well as for reducing raw material costs [16, 17]. It is expected that wood with a lower S/G ratio will provide greater efficiency in the carbonization process, higher productivity of the charcoal plants, and lower gas emissions resulting from the thermal decomposition of the wood. This is related to the chemical structure of guaiacyl (G) monomeric units being more condensed and, consequently, more thermally stable [13, 15]. According to Wang et al. [15], the reactivity of the lignin linkages is influenced by the substituted functional groups (phenolic hydroxyl group and methoxyl group), and, consequently, the content of methoxyl groups in lignin is correlated with the formation of lignin pyrolysis charcoal. Thus, wood with lignins with high methoxyl group contents (higher S/G ratio) results in less charcoal during the pyrolysis process.

The lowest S/G ratios and the highest percentages of lignin and extractives in wood were observed for the clones 1039 and 1025. Santos et al. [17] verified that the S/G ratio may be inversely correlated to the wood lignin content. This inverse relationship is beneficial for carbonization since wood with higher lignin content and lower S/G ratio could simultaneously show greater efficiency in converting wood into charcoal and better quality of this energy input.

The mass of G and S per kg of dry wood observed in the evaluated clones varied between 19.6 g (clone 1024) and 27.4 g (clone 1025) and between 66.6 g (clone 1008) and 98.7 g (clone 1023), respectively. These values are consistent with those found by Araújo et al. [18]. The selection of clones for charcoal production must prioritize a higher proportion of G in lignin because it can result in an increase in GYC due to the higher thermal stability [20]. Araújo et al. [18] observed a significant and inversely proportional effect of the mass of G on the thermal decomposition of wood between temperatures of 315 °C and 390 °C; however, this behavior was not verified at 450 °C (final carbonization temperature). Therefore, research aimed at investigating the effect of oxidation derivatives of the G and S units of lignin on the carbonization mass balance of Eucalyptus clonal materials is necessary to clarify these associations.

Analysis of variance indicates that there is no clonal effect for the ultimate analysis of wood (Tables 4 and 5). For charcoal, there was a statistical difference between genetic materials for the contents of C, H, O, and N (Table 4), with the result attributed to differences in the structural chemical composition (lignin quality) and non-structural (extractives soluble in acetone) from the cell wall (see Table 3). The ultimate analysis of wood and charcoal is fundamental for the classification of the best Eucalyptus clones for energy purposes (Table 5), since knowing the percentages of C, H, and O, the energy released in combustion can be estimated. In addition, the management of the firing equipment can be better performed. The C, H, and O contents in the wood may be correlated with the GYC [13].

Table 4 Variance analyses summary for ultimate analysis of wood and charcoal from Eucalyptus clones
Full size table
Table 5 Ultimate analysis of wood and charcoal from Eucalyptus clones
Full size table

The clones showed contents of C in the wood varying from 47.5 to 48.9% and from 79.8 to 82.3% for charcoal. The O content in wood varied from 44.1 to 45.7% and in charcoal between 11.9 and 15.0%. Due to the pyrolysis process, the O content reduced approximately 70.4%. This is a satisfactory result due to the negative relationship between this chemical component and the energy performance of charcoal [36, 37]. This phenomenon occurs due to the thermal decomposition process in the absence or controlled presence of O that promotes the highest concentration of C and volatilization of O present, mainly in the carbohydrates of the woody cell wall [7]. In addition, the variation in the O proportion in charcoal was higher than that observed for wood. On average, the O content in charcoal varied by 26%, while in wood, it was found a considerably lower variation among the Eucalyptus clones (3.6%).

The charcoals of clones 1006 and 1024 had the lowest and highest O content, respectively. This result is possibly related to the differences in the chemical composition of the cell wall of the fibers, especially the composition of the lignin macromolecule [15, 38]. The wood of clone 1006 is part of the group with the lowest S/G ratios and the highest proportions of G units in lignin (see Table 3). The opposite was observed for wood from clone 1024, high S/G ratio, and lower proportion of G units. The S/G ratio of clone 1024 was 36% higher than that observed for clone 1006. Soares et al. [38] reported that the amounts of C and O present in the Eucalyptus charcoal are more related to the type of chemical structure present in the cell wall and not necessarily to the proportion of these chemical elements in the wood, corroborating with the results described in our research.

Analyzing the percentage formulas of the G (C66.7%H6.7%O26.7%) and S (C62.9%H6.7%O30.5%) units, it can be seen that the G unit presents proportionally more C than O in the molecule [38]. The presence of a greater amount of G units can improve the resistance to thermal degradation of wood due to the lower number of methoxyl groups (–O–CH3) present in the molecule and, consequently, in the amount of active sites in the phenylpropane unit [20]. Wang et al. [15] reported that the lignins with high methoxyl group contents (bigger S/G ratio) produce less charcoal during the pyrolysis process, that is, less resistant to thermal degradation. Therefore, the lowest S/G ratio and the highest mass of G units observed for wood from clone 1006 may be associated with improved resistance to wood thermal degradation and, consequently, the highest proportion between the C and O levels on charcoal.

In relation to sulfur, insignificant levels were detected for wood (0.00–0.07%) and charcoal (0.00–0.06%) of Eucalyptus clones, that is, only traits that do not compromise the bioenergy use. The nitrogen content stood out due to the mean observed for charcoal (1.4%), being 2.8 times higher than that of wood (0.5%). This means that this element did not volatilize during carbonization, and its percentage in charcoal increased due to the thermal decomposition of wood, corroborating the study by Leite et al. [39]. The nitrogen and sulfur are relevant to indicate the polluting potential of biomass and charcoal, due to the release of sulfur and nitrogen oxides during burning [40]. Therefore, the results show the advantage of using energy from Eucalyptus wood and charcoal, compared to coal widely used in the steel industry, with sulfur contents of up to 2.07% [41].

Gravimetric Yield and Mass Balance of the Carbonization Process

There was statistical difference between the clones for the GYC (Fig. 1 and Table 6). Clones 1025 (E. camaldulensis hybrid), 1039 (E. grandis hybrid), 1024 (E. urophylla hybrid), 1033 (E. urophylla hybrid), and 1031 (Eucalyptus spp.) showed the highest GYC. These clones were part of a single group with mean of 35.2% for GYC, a difference of 3.1% in relation to the second group. Disregarding clone 1024, the others were characterized by the highest number of G units per dry mass of wood and the lowest S/G ratio. Genetic materials 1025 and 1039 showed high contents of lignin and extractives soluble in acetone in wood and the highest percentages of C. The literature has shown the influence of low S/G ratio [13, 14] and higher levels of lignin [15, 42], extractives soluble in acetone [14], and C [13] in GYC, corroborating the reported findings. As previously reported, the lignin present in the cell wall of Eucalyptus fibers is formed by different proportions of syringyl (S) and guaiacyl (G) basic units. The main difference between G and S units is the amount of methoxyl groups (–O–CH3) in aromatic rings (see topic “Chemical composition of the wood and charcoal”). This directly influences the number of active sites in the phenylpropane unit and the thermal stability of lignin [13]. The relationship between lignin chemical structure and pyrolysis behaviors of the biomass is meaningful. As the S unit has only two active sites associated with carbon atoms, the number of ether linkages (C–O–C) is greater, for example, β-aryl-ether bonds (β–O–4) [15, 43]. In most cases, this type of bond is easily cleaved [15], and; therefore, lignins with a higher proportion of S units are less resistant to thermal degradation during pyrolysis. On the other hand, G unit has three active sites associated with carbon atoms and, consequently, has a lower proportion of ether linkages (C–O–C) and a greater amount of carbon-carbon bonds (e.g., 5-5 and β-5 linkages). Wang et al. [15] reported that carbon-carbon linkages such as β-5 present relatively low reactivity and affirmed that the cleavage of linkages combining the carbon atoms in different aromatic rings, such as 5-5, is difficult. Therefore, wood with a greater proportion of G units present a higher condensation degree and is more resistant to thermal degradation [15, 44, 45]. These observations, combined with the results of the wood chemical composition (see Table 3), explain the results of gravimetric yield in charcoal for Eucalyptus clones: 1025, 1039, and 1033 (Fig. 1).

Fig. 1
figure 1

Gravimetric yield in charcoal (GYC) and sum of lignin (Lig) and extractives soluble in acetone (Ext) of the studied Eucalyptus clones. Columns followed by the same letter do not show difference according to the Scott-Knott test (p > 0.05). The error bars for both properties represent the standard deviation

Full size image
Table 6 Variance analyses for gravimetric yield and mass balance of the carbonization process
Full size table

Despite the higher S/G ratio, clone 1024 showed more than 31% of the dry wood mass represented by lignin and extractives (Lig + Ext), which may justify the higher GYC (35.1%). Santos et al. [17] stated that high lignin content in wood should be prioritized when evaluating Eucalyptus clones for carbonization. The lignin macromolecule has greater thermal stability and, therefore, is directly related to the production of charcoal. Due to the complexity of chemical bonds (e.g., carbon-carbon linkages such as β-1, β-5, and 5-5), the thermal degradation of lignin shows a broader peak compared to cellulose and hemicellulose due to the successive cleavage of linkages as there action temperature increases [15].

On the other hand, Araújo et al. [18] and Castro et al. [16] have reported no correlation between the S/G ratio with the GYC for Eucalyptus clones at different ages. Araújo et al. [18] suggested that this divergence of results shows that the greater number of G units or low S/G ratio does not necessarily imply the occurrence of more thermally stable bonds, justifying the greater GYC obtained for the wood carbonization of clone 1024. Liu et al. [43] analyzed two lignin samples with similar functional groups and S/G ratio, but with differences in side-chain linkages. Therefore, the results suggest that it is possible to obtain higher productivity of charcoal from raw materials with a high S/G ratio, as long as the wood presents a higher percentage of lignin and extractives soluble in acetone (≥ 31%). In addition, it is important to verify the type of linkages between the S and G monomers, mainly the frequency of β-aryl-ether bonds (β-O-4) and carbon-carbon bonds (e.g., 5-5 and β-5 linkages), because this may influence in the proportion of C and O in the charcoal. Guo et al. [45] and Zhang et al. [46] extracted lignin from woody biomass and non-woody biomass and then characterized the chemical structures and the interunitary linkages of the lignin using nuclear magnetic resonance spectrometry (NMR). These authors observed various interunitary linkages in lignin assigned in the side-chain region of NMR spectra, for example, β-O-4, β-5, β-β, 5-5, and β-1. The results obtained by Guo et al. [45] suggest that the percentage of β-O-4 linkages in the lignin of the biomass is an indicative of the condensation degree. Overall, lignins with a lower percentage of β-O-4 linkages are more resistant to pyrolysis due to the higher content of condensed linkages. Therefore, future studies should consider this technique for analyzing the interunitary linkages of the lignin in Eucalyptus wood, simultaneously with the S/G ratio and total lignin content.

By the F test of the variance analyses (Table 6), there was no clonal effect for the C and H balances, only for the O balance (Fig. 2). The percentages of retained C and volatilized C ranged from 55.5 to 59.8% and 40.2 to 44.5%, respectively. The percentages of retained H and volatilized H ranged from 19.2 to 20.7% and 79.3 to 80.8%, respectively. The low variation found for the C and H balance of the carbonization may be related to the lignin content of Eucalyptus woods. There was low variability in the lignin content (28.03 to 32.64%), and, consequently, the products derived from the pyrolysis of this macromolecule can be very similar between the clones. The different lignin content, S/G ratio, and demethoxylation degree under determine the distributions of p-hydroxyphenyl phenols, guaiacyl phenols, and syringyl phenols [43].

Fig. 2
figure 2

Mean values of carbon, hydrogen, and oxygen balance of carbonization of wood from Eucalyptus clones. Means followed by the same letter do not differ according to the Scott-Knott test (p > 0.05)

Full size image

For the content of retained and volatilized O, the results showed amplitude of 9.0 to 11.6% and 88.4 to 91.0%, respectively. The carbonization of wood from clones 1006 and 1024 resulted in the smallest and largest amount of retained O in the charcoal. As previously reported, this result may be associated with the S/G ratio of wood lignin in these Eucalyptus clones. The wood from clone 1006 showed S/G ratio of 3.0 whereas the wood from clone 1024 showed S/G ratio of 4.0 (see Table 3). The lower S/G ratio and the higher proportion of G units of the clone 1006 wood may explain the lower amount of retained O in charcoal since the G units have a higher C/O ratio (2.50) in their chemical composition comparatively the S units that presented a C/O ratio of 2.06 [38]. In addition, the wood of these clones had lignin content close to 30%, indicating that the O balance is more related to the composition than to the lignin macromolecule content.

Grouping of Eucalyptus Clones by Principal Component Analysis

The first three principal components had the highest eigenvalues and explained 73.79% of the total variance of the data (Table 7). Therefore, these three latent variables must be considered for the classification and grouping of Eucalyptus clones for charcoal production. Considering the eigenvectors, it is noted that the first three principal components bring together the wood and process characteristics more relevant for the Eucalyptus species selection for charcoal production.

Table 7 Eigenvectors (ê) and contribution of the original variables (cont) in the first three principal components (PC)
Full size table

Researchers have used the principal component analysis for the classification and grouping of biomasses considering their energy characteristics, especially in situations where many variables are measured [47], as in this research with Eucalyptus clones. The study of eigenvectors can contribute by highlighting the most important variables in the formation of the principal components, allowing the Eucalyptus clones grouping and the dispersion of the properties evaluated in the wood and in the pyrolysis process of charcoal in the Cartesian plane [48].

Analyzing the first principal component (PC1), it is observed that the largest positive eigenvectors are related to the total lignin content, GYC, C, and H retained in charcoal. The most significant negative eigenvectors refer to the S/G ratio, C, and H volatilized during pyrolysis. The signs (+ or −) of the eigenvectors indicate the correlations between the original variables; thus, the highest contents of lignin and the lowest S/G ratio are directly related to the highest GYC and the best C and H balance during thermal decomposition of Eucalyptus wood, reinforcing the trends previously detected (see items “Chemical composition of the wood and charcoal” and “Gravimetric yield and mass balance of the carbonization process”).

Principal component 1 (PC1) can be interpreted as a general index for the classification of Eucalyptus clones for charcoal production because it presents the most relevant wood characteristics and the pyrolysis process. Therefore, the higher the values of this component (scores), the more propitious the clone for charcoal production will be. Based on the PC1 scores (Fig. 3), the order of the most suitable clones for charcoal production is: 1025 (E. camaldulensis hybrid), 1039 (E. grandis hybrid), 1006 (E. urophylla), and 1033 (E. urophylla hybrid). The least suitable clonal materials for the production of bioreducer were 1023 (E. urophylla hybrid) and 1005 (E. urophylla).

Fig. 3
figure 3

Scores scattering of the principal components (PC1, PC2, and PC3) and grouping of Eucalyptus clones according to the chemical characteristics of the wood and the pyrolysis process

Full size image

Principal component 2 (PC2) is basically related to the O balance retained in charcoal and volatilized during pyrolysis. The eigenvector associated with the percentage of retained oxygen in charcoal presented a negative contribution in the second principal component. Therefore, the lower scores of this principal component indicate a higher proportion of retained oxygen in charcoal and a lower amount of volatilized oxygen during pyrolysis. Among the 14 clones evaluated, 1024 was the one with the lowest score for PC2, a result attributed to the higher O content retained and, consequently, the lower proportion of volatilized O during the thermal decomposition of the wood. We attribute this result to the high proportion of S units, compared to G units, in the wood lignin of this clone.

Principal component 3 (PC3) showed positive and significant eigenvectors for the S/G ratio and mass of S units per kg of dry wood. The most relevant negative eigenvectors are associated with the mass of G units per kg of wood and the content of extractives soluble in acetone. The higher the scores of this principal component, the less propitious the clone will be for charcoal production. The high scores in the third principal component indicate a high S/G ratio, a higher number of S units, and a lower proportion of G units and extractives soluble in acetone in the Eucalyptus wood.

The highest PC3 scores were observed for clones 1023 and 1024, indicating that they are less recommended for charcoal production. These clones are considered the most dissimilar in relation to the others. The results found for PC3 corroborate the previous discussions and reinforce the potential of clonal materials 1025 and 1039 for the composition of specific energy forests for the production of charcoal. These two Eucalyptus clones formed a single group and are considered similar because they have more suitable wood quality and characteristics of the pyrolysis process.

Clones 1006, 1008, 1009, 1031, and 1033 formed the group with the highest number of similar genetic materials, in which the most relevant characteristics in this grouping were GYC, retained C, retained H, total lignin, and grams of G per kg of dry wood. These genetic materials were ranked as intermediates for the production of steel bioreducer.

Clones 1004, 1015, and 1037 formed a single group due to the similarity of GYC (Fig. 1) and are the most similar genetic materials (Fig. 3). Clones 1005 and 1036 showed similarities for scores of PC1, PC2, and PC3; therefore, they formed a single group. These materials had the highest percentages of C and H volatilized during pyrolysis. For PC1, negative scores attributed to these clones were found, simultaneously indicating unfavorable characteristics of the wood and the pyrolysis process.

The genetic materials 1023 (E. urophylla hybrid) and 1039 (E. grandis hybrid) are the most dissimilar or divergent, due to the wide variability or inverse relationship between the characteristics of the wood and the pyrolysis process. Thus, clones with these characteristics are indicated for crossing when the objective is to increase the variability and expression of the heterotic effect [11]. The PCA scores reinforce the relevance of simultaneously considering the characteristics of the wood and the process for the classification of genetic materials for charcoal production. The applied technique, combined with the researcher’s experience, allows minimizing the chances of failure in the management of the Eucalyptus energy forests and allows greater control of the productivity of charcoal from the wood pyrolysis with similar chemical properties.

Canonical Correlations Between the Characteristics of Wood and Charcoal

Based on the canonical correlation analysis, it is observed that the first canonical function was statistically significant by the Hotelling multivariate test (p value = 0.0017), with an approximation of the F distribution. The canonical correlation coefficient can be considered moderate (0.68) and indicates that approximately 46% of the variance between the groups of characteristics of wood and charcoal was explained by the first pair of canonical variables. Through canonical correlation analysis, the interdependence of the characteristics of groups X and Y were evidenced, that is, the pyrolysis mass balance depends on the wood chemical characteristics (Fig. 4). This result contributes to the more precise identification of the effect of the wood chemical composition in the bioreducer production and, consequently, genetic improvement of the Eucalyptus clones currently used in Brazil.

Fig. 4
figure 4

Scores of significant canonical functions by Hotelling’s test

Full size image

According to Hair Junior et al. [49], canonical loads measure linear correlations between the original variables and their respective canonical statistical variables, while the crossed canonical loads represent the correlation between an original variable in a certain group and the canonical statistical variable in the other group. The analysis of these correlations involves examining the signal and its magnitude (Table 8). In the dependent group (Y), the largest canonical charge, in module, is related to the GYC. In the independent canonical statistical variable (X), the content and composition of the lignin macromolecule were more important. The same trend was observed for crossed canonical charges.

Table 8 Canonical loads (CL) and crossed canonical loads (CCL) of the first canonical function
Full size table

Analyzing the sign of canonical charges, it is noticed that there is a tendency to increase GYC, and the percentages of C and H retained in charcoal with the increase of lignin, extractives, and the mass of G units in wood. On the other hand, the increase in the S/G ratio and the proportion of S units in lignin results in a decrease in GYC and the percentages of C and H retained in charcoal. The high S/G ratio of the lignin macromolecule is directly related to the higher percentage of O retained in charcoal due to a higher proportion of O in the S unit [38].

In addition, lower scores attributed to the canonical statistical variables X and Y indicate the best clones for the charcoal production, as they present the best wood chemical characteristics, associated with the highest GYC. In this sense, clonal materials 1039, 1025, 1006, and 1033 were better classified. These results confirm the observations previously made about the effect of the wood chemical composition on the pyrolysis mass balance and reinforce the trends obtained in the principal component analysis.

Previous research has reported that the lower S/G ratio, or higher proportion of G units, contributes to a higher GYC [13, 14, 17], similar to what was observed in our study in the results of multivariate statistical analyses. Unit G has one methoxyl group less in carbon five (C5) of phenylpropane, and this allows for more stable chemical bonds between aromatic rings and side chains of the lignin macromolecule. Consequently, it is expected to improve the resistance to thermal degradation of wood, which presents higher levels of lignin and higher proportions of G units [13, 15, 38]. Therefore, lignins with fewer methoxyl groups, or with a higher proportion of G units, provide an improvement in the C and H balance of pyrolysis and minimization of the production of by-products during the thermal decomposition of wood.

The findings of this research provide subsidies for further studies for understanding better the drivers of the wood pyrolysis process in order to maximize the sustainable production of charcoal. These studies will be important for the rational expansion of Eucalyptus forests dedicated to the production of bioreducers, an important raw material for the steel industry in Brazil.

Conclusions

Woods with high lignin and extractives soluble in acetone contents, low S/G ratio, and a higher proportion of guaiacyl units are prone to promote high gravimetric yield in charcoal, and better carbon and hydrogen balance during the pyrolysis. Eucalyptus woods with low S/G ratio, high mass of guaiacyl units, and a higher amount of extractives soluble in acetone produced charcoal with higher proportion of carbon and hydrogen retained. These characteristics can be considered the main quality indexes for young Eucalyptus wood intended for charcoal production and must be simultaneously analyzed. The results observed indicate that the O balance is more related to the composition than to the lignin macromolecule content.

Eucalyptus camaldulensis (represented by clone 1025) and Eucalyptus grandis (represented by clone 1039) hybrids are the most suitable vegetal materials for charcoal production because they presented most adequate wood quality and better performance in the pyrolysis mass balance.