Composition analysis of the PHL
During the pre-hydrolysis stage used for kraft-based dissolving pulp production, the removal extents of hemicellulose and lignin depend upon both treatments conditions and woody feedstock [5,25]. In consideration of these variables, it is necessary to obtain a detailed chemical composition for the mixed hardwood PHL used in this work. The chemical compositions of PHL are shown in Table 1. It can be seen that xylose (86.3 g/L) and xylooligosaccharide (32.3 g/L) are the major sugars in this PHL. In addition, a certain amount of glucose (7.5 g/L), arabinose (6.1 g/L) and sugar-derived by-products (0.9 g/L of furfural and 0.4 g/L of HMF) were also quantified. The high concentration of xylooligosaccharide in PHL confirms that this stream can be further transformed into xylose-rich solution by hydrolyzing xylooligosaccharides into xylose. Comparing to the reported work, it can be seen that the sugars in PHL was significantly higher than those in the PHL (40-60 g/L) used in the work of Shi et al., (2012) [29]. This is due to the used mixed hardwood PHL was obtained after it being concentrated at concentration workshop section by steam. In addition, it is important to note the high concentration (42.1 g/L) of soluble lignin in PHL. These solutes have the potential to significantly affect efficient biological utilization of PHL [30]. Therefore, it is necessary to remove lignin from PHL that leaves most of the xylose preserved in order to ensure an efficient process for producing ethanol or XA.
To understand the structural information of the chemical compositions in PHL, 2D-HSQC NMR analysis was carried and the obtained spectra are shown in Fig. 1(a). Peak assignments within the HSQC spectra are assigned according to recent works [5,31,32].
The internal xylan units (X) were clear and obvious signals within the spectra. Specifically, the C2/H2, C3/H3, C4/H4, and C5/H5 positions showed the correlation signals at 72.3/3.03, 73.9/3.22, 75.3/3.48, and 62.9/3.15, 3.85 ppm, respectively. In addition, xylan with reducing-end units (XR) were also identified from the C2/H2 and C4/H4 correlation signals at 74.4/2.89 and 75.3/3.48 ppm, respectively. The presence of XR indicates that the soluble xylan (xylo-oligosaccharide) in the PHL possess the relatively low degree of polymerization, which makes it a promising resource for xylose production [33].
In addition, the spectra of PHL showed common lignin substructures of β-O-4 (A), β−β (B), and β-5 (C), which can be identified by Ca-Ha signals at 71.8/4.86, 84.9/4.69, and 86.8/5.49 ppm, respectively. Correlation signals for syringyl units (S) and guaiacyl units (G) were also noted from their C2-H2 and C2,6-H2,6 at 111.0/7.01 and 104.1/6.74 ppm, respectively. Hence, it can verify the lignin fraction was existed in PHL, which is in accordance to the results in composition analysis.
Acid hydrolysis of the PHL
In the previous work, it reported that even PS-DVB resin show low affinity for carbohydrate, while a portion of high molecular weight sugars can be still adsorbed by the resin [5,34]. In order to reduce the loss of sugars in PHL after being treated by PS-DVB resin, the oligosaccharide mono-sugars in PHL was intended to be hydrolyzed into mono-saccharide before purification. Acid hydrolysis was carried out with sulfuric acid catalyst at different reaction conditions (acid concentration, time, and temperature) to identify a recipe which maximizes xylooligosaccharide conversion into xylose.
The increase to xylose concentration in the PHL under different acid concentrations (1-5%) at 100 °C for 60 min is shown in Fig. 2a. First, increasing xylose concentrations were found with increasing acid catalyst concentrations. For example, the concentration of xylose in PHL increased from 86.1 g/mL to 95.4 g/L when acid concentration increased from 1% to 4%. However, a maximum was noted, as xylose concentration decreased to 88.9 g/L when the acid concentration was further increased to 5%. This can be explained by the mechanism that xylose is easily dehydrated into furfural at the higher acid concentration (and temperatures too). As observed by Yang et al., (2012) [35], increasing the acid catalyst (formic acid) concentration from 5-10 g/L could effectively dehydrate xylose into furfural with yields rising from 15.8 % to 74%. Depending on the maximum xylose concentration in PHL, it can be speculated that sulfuric acid with 5% concentration was the optimum condition to produce the xylose from PHL.
According to the aforementioned results, the effects of different reaction temperature (90-130 oC) with 4% acid and 60 min of acid hydrolysis on increases to xylose concentration in PHL were investigated and shown in Fig. 2b. The maximum concentration of xylose in PHL was 95.4 g/L at 110 oC. Increasing reaction temperature from 100 oC to 130 oC significantly reduced xylose concentrations from 95.4 g/L to 75.1 g/L. This was because increasing reaction temperature further accelerated dehydration of xylose into furfural given the high acid dosage applied in this series of experiments [35]. In the work of Chen et al., (2018) [33], they also found that the xylose in PHL could be further degraded under reaction temperatures from 120 oC to 135 oC with 4% aqueous sulfuric acid for 30 min. Based on the maximum xylose concentration in PHL, it can be speculated that acid hydrolysis with 4% acid concentration at 100 oC were the optimum conditions to produce the xylose from PHL.
The effect of the reaction time (20-100 min) of acid hydrolysis with 4% acid at 100 oC on xylose concentration in PHL is illustrated in Fig. 2c. It can be observed that the xylose concentration in the PHL increased at first and then decreased with increasing reaction time from 20 min to 100 min. The optimal reaction temperatures was 80 min, where a maximum xylose concentration of 101.1 g/L was obtained. Therefore, a highest xylose concentration in PHL with a 17% increase (relative to original PHL) can be obtained by acid hydrolysis at 100 oC for 80 min with 4% sulfuric acid dosage.
Although a higher xylose concentration can be obtained in PHL by applying acid hydrolysis, this method is also shown In this work to unavoidable lead to the dehydration of sugars (xylose and glucose) into fermentation inhibitors (furfural and HMF). As can be seen in Table 1, the acid-treated PHL at the aforementioned ideal condition (A-PHL) contained a higher xylose concentration than the original PHL. However, concentrations of the fermentation inhibitors furfural and HMF were 7.8 g/L and 1.9 g/L. These quantities are significantly higher than those in PHL (0.9 g/L and 0.4 g/L). In the reported work of Bellido et al (2011) [36], it was found that the pentose-fermenting yeast P. stipitis had poor tolerance towards furfural and HMF, with cell growth almost completely inhibited when the fermentation media contained at least 2 g/L furfural. In addition, various phenolic compounds formed from the degradation of lignin can also inhibit the efficiency of ethanol fermentation. Zhou et al., (2017) [37] found a significant inhibition effect from furfural occurred using a G. oxydans fermentation system with prehydrolyzate containing 6.5 g/L furfural, revealing the toxicity of furfural for XA production from G. oxydans. In addition, the concentration of water soluble lignin (44.9 g/L) was slightly increased in A-PHL relative to PHL. The high concentration (>2 g/L ) of lignin fractions in the sugar solutions can also cause the inhibitory effect for fermentation. Thus, removing the fermentation inhibitors in A-PHL not only provide a favorable environment for pentose-fermenting yeast but also removes a critical hurdle towards fermentation of xylose to XA by G. oxydans.
Purification of PHL by PS-DVB resin
Lignin and sugar-derived byproducts in the PHL are difficult to separate from the sugars due to their similar molecular weight and water solubility [27]. In this work, we used a lignin-selective adsorptive resin to separate the lignin and sugars by-products (furfural and HMF) from PHL while retaining most of the sugars in the PHL. A diagram for the process used is shown in Fig. 3. As can be seen, the lignin and sugar by-products in PHL can be adsorbed in the resin, while the purified sugars solution proceed to the next unit operation. The ethanol used to regenerate the resin by desorbing the adsorbates can also be recycled to prepare the lignin-based materials. Hence, the present concept for purification of PHL by PS-DVB resin is inherently environmentally friendly and sustainable.
As can be seen in Table 1, concentrations of soluble lignin, HMF, and furfural in PS-DVB treated A-PHL (P-A-PHL) were 1.1 g/L, 0.04 g/L, and 0.61 g/L. Each of these quantified values are significant lower than those in original A-PHL solution (46.9 g/L, 1.9 g/L, and 7.8 g/L, respectively). These results indicate that not only 97% of the lignin can be removed, but also 92-97% of sugars by-products can also be separated from the xylose solution. In addition, a high proportion (96%) of xylose was retained in P-A-PHL. These results indicate that the this process can simultaneously separate and recover the lignin and sugars in PHL with yields > 90%, which was more effective than the reported technologies of acidification, nanofiltration or microfiltration, applying cationic polymers (e.g. polydiallyldimethylammonium chloride), activated carbon, or lime mud [1,19,25]. Overall, the low concentration of fermentation inhibitors in P-A-PHL suggest that the fermentability of PHL should be sufficient for producing ethanol and XA. However, it remains to be seen if this removal was sufficient enough to achieve efficient fermentation results.
To further understand the structural changes of the carbohydrate and lignin in A-PHL after being treated by PS-DVB resin, the 2D-HSQC spectra of A-PHL and P-A-PHL were obtained and shown in Fig. 1b and Fig. 1c, respectively. It can be seen that A-PHL showed the similar cross-signals for various substructures of lignin and carbohydrate in the 2D-HSQC spectra comparing to those in original PHL spectra (Fig. 1a). However, the signals intensity of β-O-4 substructures in A-PHL’s spectra were significant lower than those in PHL spectra. This might be due to the cleavage of this ether bond under the acidic hydrothermal conditions of acid hydrolysis [29]. For the 2D-HSQC spectra of P-A-PHL, signals observed for carbohydrates remained in similar positions compared to those noted in the A-PHL spectra (Fig. 1c). Importantly, the correlations of lignin substructures and units were absent in P-A-PHL’s spectra. This again indicates the successful removal of the lignin from A-PHL, which is attribute to the adsorption ability of PS-DVB resin.
Fermentation of A-PHL and P-A-PHL by P. stipitis to produce ethanol
To understand the fermentability improvement of the A-PHL after being treated by PS-DVB resin, both A-PHL and P-A-PHL (with same initial xylose concentration of 30 g/L and 50 g/L) were used as the fermentation stocks to produce ethanol by P. stipitis. The concentration of xylose was proposed based on the general tolerance ability of P. stipitis for xylose. The fermentation inhibitor concentrations in either sample are listed in Table 2. Xylose consumption and ethanol production of A-PHL and P-A-PHL are shown in Fig. 4a and Fig. 4b, respectively. The ethanol yields, which is defined as the percentage of the total amount of ethanol that could be produced from consumed xylose, are shown in Fig. 4c.
Results in Fig. 4a revealed that A-PHL has low ethanol fermentation productivity by P. stipitis. Specifically, xylose in A-PHL was slowly consumed during the fermentation process. After 36 h of fermentation , 20.2 g/L and 41.6 g/L of xylose remained unfermented in the A-PHL with initial xylose concentration of 30 g/L and 50 g/L. These concentrations correlated to a paltry 32.6% and 16.8% conversion, respectively. Ethanol yield (Fig. 4c) obtained from the A-PHL with 50 g/L xylose was only 12.0%, which was lower than that of A-PHL with 30 g/L xylose (23.5%). The low fermentability of A-PHL for producing the ethanol by P. stipitis was therefore due to its high concentration of inhibitors [38,39]. As these samples contain 0.57 g/L of HMF and 2.36 g/L of furfural in A-PHL with 30 g/L xylose and 0.82 g/L of HMF and 3.93 g/L of furfural in A-PHL with 50 g/L xylose. As pointed out by Díaz et al. (2009) [38], an inhibitory effect in ethanol fermentation by P. stipitis can be observed with the furfural concentrations of 1-2 g/L, and almost no sugars can be consumed when there is greater than 4 g/L furfural in fermentation media.
From Fig. 4b, it can be seen that PS-DVB resin treatment remarkably improved the fermentability of A-PHL for ethanol production. 80% and 47% of xylose in P-A-PHL with initial 30 g/L xylose and 50 g/L xylose were consumed after 36 h of fermentation, resulting in ethanol yields of 61.8% and 45.9%, respectively. Compared to the results of untreated A-PHL with 30 g/L and 50 g/L xylose, ethanol yields improved by 162% and 282%, respectively. The improved fermentability of P-A-PHL therefore must be due to the removal of carbohydrate-degradation products (HMF and furfural) and lignin-degradation products. As can be seen in Table 3, the concentration of HMF, furfural, and lignin-degradation products in P-A-PHL with 30 g/L and 50 g/L xylose were 0 g/L, 0.21-0.30 g/L, and 0.75-1.01 g/L, significant lower than those in untreated A-PHL. In view the other technologies have been done to improve the fermentability of sugar solution with inhibitors, Lai et al., (2016) [39] reported that the ethanol production increased by 45.5% and 42.8% could be achieved for the prehydrolyzate treated by cetyltrimethylammonium- and benzyltrimethylammonium-modified bentonites, respectively. Zhu et al., (2011) [40] reported that a trialkylamine extraction technology to remove 45.7% of HMF and 100% of furfural could be removed from prehydrolyzate, which could improve its fermentability by P. stipitis with a degree of 89.6%. Based on the aforementioned results, it can be concluded that PS-DVB resin treatment was a valid detoxification method to improve PHL fermentability for ethanol production.
Fermentation of A-PHL and P-A-PHL for producing XA by G. oxydans
The fermentability of A-PHL treated by PS-DVB resin with different xylose concentration (40, 90, and 150 g/L) were investigated and shown in Fig. 5. Fig.5a shows that the xylose in A-PHL with 40 g/L xylose could be completely consumed within 48 h, resulting in a XA production of 44.4 g/L and a corresponding yield of 96.6 % (Fig.5c). When increasing the A-PHL’s xylose concentration to 90 g/L, even a better XA production with 58.8 g/L could be achieved at 48 h. However, at the aforementioned conditions, only 49.7% of the xylose was consumed by G. oxydans to produce XA with a yield of 60.6%. When further increasing the A-PHL’s xylose concentration to 150 g/L, only 12.1% of the xylose could be consumed to produce the XA with a yield of 12.1% ( 4.7 g/L) at 48 h. The bioconversion of xylose in A-PHL was strongly inhibited by the increased concentration xylose was due to the corresponding increased concentration of furfural in A-PHL. As can be seen in Table S1, A-PHL with 40 g/L and 90 g/L xylose contained 3.16 g/L and 6.96 g/L of furfural, which were within the tolerance of G. oxydans to produce XA. Meanwhile, there was 10.8 g/L of furfural in the A-PHL with 40 150 g/L xylose, which is beyond the tolerance threshold for bioconversion. As pointed out by Zhou et al., (2018) [37], the bioconversion ability of G. oxydans can be reduced by 60% when the furfural concentration was increases from ~10 to 15 g/L.
Since high concentrations of furfural causes severe inhibition of bioconversion ability of G. oxydans [41], we also tried to use the PS-DVB resin treated A-PHL (P-A-PHL) to produce the XA by G. oxydans. As can be seen in Fig.5b, a predictably enhanced productivity of XA with 71.6 % and 26.1% yields were achieved from P-A-PHL with 90 g/L and 150 g/L xylose, which were improved by 18% and 828% comparing to those of A-PHL with 90 g/L and 150 g/L xylose, respectively. The tremendous improvement of the A-PHL was due to its low concentration of furfural (as shown in Table S1), which was achieved by PS-DVB treatment. According to the research of Chai et al.(2013) [42], the inhibition by furfural toward whole-cell catalysis of G. oxydans can be avoided by implementation of furfural removal. Our observations support this previous finding. In all, it can be concluded that PS-DVB resin treatment was also a potential detoxification method for PHL to improve its fermentability for XA production, especially from the PHL with high concentrations of xylose.