The influence of solvent type on organosolv fractionation of sorghum bagasse
Sorghum bagasse was fractionated into three primary components important in biorefinery processes: cellulose, hemicellulose, and lignin. In this study, fractionation was evaluated by comparing pretreatment using five different organic solvents (ethanol, 1-propanol, 2-propanol, 1-butanol, and 1-pentanol) at a low concentration (12.5 %). Pretreatment with no addition of solvent was used as a control.
After treatment at 180 °C for 45 min, samples were centrifuged (Fig. 1). Interestingly, three fractions (solid, liquid, and black liquor) were obtained with 1-butanol or 1-pentanol as the solvent, but only two fractions (solid and liquid) were obtained using ethanol, 1-propanol, or 2-propanol as the solvent and when no solvent was used (control). These results suggest that differences in the physiochemical properties of the solvents affect fractionation. The greater hydrophobicity of 1-butanol and 1-pentanol (partition coefficients: logP
ow = 0.88 and 1.51, respectively) compared with ethanol, 1-propanol, and 2-propanol (logP
ow = −0.31, 0.25, and 0.05, respectively) resulted in clearer fractionation when using the former solvents. The control condition was equal to dilute acid pretreatment [16]. Thus, hydrophobic 1-butanol and 1-pentanol separated the black liquor fraction from the acid solution even at the low concentration of 12.5 %.
Cellulose-enriched solid fraction
To clarify the effect of solvent type on raw sorghum bagasse, the solid fraction obtained after pretreatment was characterized. Compared with the control, the dry weight of the solid fraction decreased when 1-butanol or 1-pentanol was used as the solvent, but ethanol, 1-propanol, and 2-propanol had no effect on solid fraction dry weight (Fig. 2a). Cellulose is reportedly enriched in the solid fraction of samples pretreated using acid [9], and an increase in the cellulose content in all solid fractions was observed compared with the raw biomass (Fig. 2b). In particular, the cellulose content was higher in the solid fraction when 1-butanol or 1-pentanol was used as the solvent (59.1 and 62.2 %, respectively) compared with ethanol, 1-propanol, 2-propanol, or no solvent (control) (51.6–55.6 %). Cellulose recovery in the solid fraction ranged from 84.4 to 91.1 % when ethanol, 1-propanol, 2-propanol, or 1-butanol was used as the solvent and when no solvent was used (control) (Additional file 1). Cellulose recovery in the solid fraction was 71.9 % when 1-pentanol was used as the solvent. Similar to the trend with dry weight (Fig. 2a), the acid-insoluble lignin content was lower in the solid fraction when 1-butanol or 1-pentanol was used as the solvent compared with other solvents (Fig. 2b). Accordingly, the greatest increase in glucose yield compared with the control was observed with 1-butanol or 1-pentanol as the solvent, but increased glucose yield was also observed with ethanol, 1-propanol, and 2-propanol (Fig. 2c).
Lignin removal, as calculated using the equation [(lignin in bagasse)—(lignin in solid fraction)]/(lignin in bagasse), was higher when 1-butanol or 1-pentanol was used as the solvent (64.7–74.3 %), compared with the use of no solvent, ethanol, 1-propanol, or 2-propanol (44.1–49.7 %) (Fig. 2a, b). Lignin, which is degraded by sulfuric acid, is reportedly liberated from the cell wall and forms droplets that attach to cellulose molecules [17]. Hydrophobic alcohols would dissolve these lignin droplets and prevent their attachment to cellulose. The removal of lignin from cellulose reportedly increases the accessibility to hydrolytic enzymes and reduces the degree of irreversible adsorption of the enzyme to lignin [18]. Thus, organosolv pretreatment using 1-butanol or 1-pentanol efficiently removes acid-insoluble lignin, resulting in enhanced enzymatic hydrolysis of cellulose.
Liquid fraction containing hemicellulose-derived sugars
It was expected that hemicellulose-derived sugars, mainly xylose, would be contained in the liquid fraction [2]. The amount of xylose contained in the liquid fraction of samples pretreated with ethanol, 1-propanol, 2-propanol, 1-butanol, 1-pentanol, or with no solvent (control) was similar (Fig. 3a). The xylose recovery in the liquid fraction was higher (90.6–97.4 %) when using ethanol, 1-propanol, 1-pentanol, or no solvent, compared with 2-propanol and 1-butanol (85.5 and 79.2 %, respectively), as calculated using the following equation: (xylose amount in liquid fraction)/[(total xylose)—(xylose amount in solid fraction)]. The reason for the decreased xylose recovery when using 2-propanol or 1-butanol as the solvent is unclear.
Compared with the control, the amount of glucose in the liquid fraction was lower in the solvent-pretreated samples (Fig. 3a). Furfural and 5-hydroxymethylfurfural (5-HMF) were contained in the liquid fraction (Fig. 3b). The harsh conditions of pretreatment resulted in the production of similar amounts of acetic and formic acids in the liquid fraction (Fig. 3c). In addition, solvents were blended in the liquid fraction (data not shown). The presence of acetic and formic acids, furfural, 5-HMF, and the pretreatment solvent would inhibit subsequent fermentation. These fermentation inhibitors could be removed, and the xylose and glucose in the liquid fraction could be concentrated, by subsequent application of membrane separation nanofiltration [19, 20].
Lignin contained in the black liquor fraction
A black liquor fraction was obtained when 12.5 % 1-butanol or 1-pentanol was used as the solvent (Fig. 1). Lignin was precipitated from the black liquor fraction by dilution with water [2]. Lignin recovery, as calculated using the equation (lignin in black liquor)/[(lignin in raw biomass)—(lignin in solid fraction)], was 12.5 and 25.6 % using 1-butanol or 1-pentanol as the solvent, respectively, corresponding to 8.1 and 19.1 % lignin content in the raw sorghum bagasse, respectively (Fig. 4). The lignin recovery in the present study was similar to that of a previous study, which reported 5.1–17.2 % lignin recovery from wheat straw pretreated with 50–60 % ethanol at 190–210 °C [5]. However, our lignin recovery was lower than that from rye straw pretreated at 190 °C for 3 h at a 1.3 % acid concentration (35 %) [21]. Therefore, process optimization of lignin recovery with respect to solvent concentration, temperature, time, and acid concentration will be necessary in the future. It is also possible that low lignin recovery was due to the loss of water-soluble products from lignin degradation [21].
Next, the composition and structure of the lignins in the black liquor fraction were analyzed using 2D 1H-13C HSQC NMR [15]. The syringyl, guaiacyl, and p-hydroxyphenyl lignin units are important aromatic elements of the plant cell wall [22]. Strong signals related to these lignin components (regions of interest [ROIs] 7–13) were detected in the black liquor, compared with raw sorghum bagasse (Fig. 5 and Additional file 2). Other strong signals related to aromatic components, such as p-coumarate (ROIs 1–6) and ferulate (ROIs 15 and 16), were detected in the black liquor. The only aromatic region signal that was weaker in the black liquor was that related to cinnamyl alcohol (ROI 14). Accordingly, signals related to methoxyl (ROIs 17 and 18) groups were stronger, as methoxyl groups are found in the side chains of syringyl, guaiacyl, and ferulate [23].
In contrast, the signal related to a major interunit structure, β–O–4 (ROI 24), was weaker in the black liquor than in raw sorghum bagasse. This was probably due to partial cleavage of the thermally labile β–O–4 unit in raw sorghum bagasse under heat treatment [24, 25]. Another signal (ROI 25) related to the β–O–4 unit was stronger in the black liquor, but the reason for this difference is unclear. Signals related to other minor interunit structures in raw sorghum bagasse were either stronger or weaker in the black liquor. These results suggest that most of the lignin aromatic components were concentrated in the black liquor in samples pretreated using 1-butanol or 1-pentanol as the solvent; however, the major linkage structure was lost.
Which lignin constituent is most affected by 1-butanol or 1-pentanol was also investigated by comparing 2D NMR signals of solid fractions obtained by pretreatment with no solvent (control), 1-butanol, or 1-pentanol (Fig. 6 and Additional file 3). Compared with the solid fraction obtained using no solvent (control), signals related to p-coumarate (ROIs 2–6) were weaker in solid fractions obtained using 1-butanol or 1-pentanol as the solvent. In addition, the signals related to guaiacyl (ROI 10), syringyl (ROIs 11 and 12), and minor β–O–4 interunits (ROIs 21–23, 26, and 27) in solid fractions obtained using 1-butanol or 1-pentanol as the solvent were weaker compared with the control. Accordingly, because the guaiacyl, syringyl, and β–O–4 units contain a methoxyl side chain, methoxyl-related signals (ROIs 17 and 18) were weaker in solid fractions obtained using 1-butanol or 1-pentanol [23]. The removal of p-coumarate, syringyl, and guaiacyl from the solid fraction using 1-butanol or 1-pentanol corresponded with the results of increased aromatic lignin regions in the black liquor fraction and decreased acid-insoluble content in the solid fraction.
Ethanol fermentation using the solid fraction
The solid fraction obtained after organosolv pretreatment can be utilized as a sugar source for microbial fermentation. The solid fractions obtained after organosolv pretreatment using 1-butanol or 1-pentanol as the solvent were subjected to simultaneous saccharification and fermentation by S. cerevisiae. Ethanol production using each of these solid fractions was compared with that obtained with control solid fraction (Fig. 7). As expected, the ethanol production rates obtained with solid fractions of samples pretreated with 1-butanol or 1-pentanol (2.6 and 2.8 g/L/h, respectively) were about 1.6 times higher in the first 9 h than those obtained with the solid fraction of samples pretreated with no solvent (1.6 g/L/h). After 96 h of fermentation, ethanol production from the 1-butanol and 1-pentanol solid fractions was about twice that of the control, reaching 43.1 and 47.2 g/L, respectively. In addition, the theoretical ethanol yield from the 1-butanol and 1-pentanol solid fractions was 59.6 and 62.1 %, respectively, compared with 32.4 % for the control. As suggested previously [26], organosolv delignification and the resultant supply of readily hydrolyzable cellulose substrates results in an increase in the concentration of microbial fermentation products.