Optimization of sodium hydroxide pretreatment of WSP
It was reported that the fixed pretreatment conditions had different effects on cellulose recovery, hemicellulose recovery and lignin removal [22]. Therefore, in our opinion, it was difficult to determine the optimal pretreatment conditions exactly using cellulose recovery, hemicellulose recovery and lignin removal as responses, respectively. In our opinion, reducing sugar and xylose yields were more suitable to be adopted as responses directly during optimization of pretreatment, as the main aim of pretreatment was to enhance sugars yields. As shown in Table 1, yields of reducing sugar and xylose in trials differed from each other, which indicated that the responses could be influenced by the adopted variables. Results in Table 2 indicated that variables including sodium hydroxide concentration, pretreatment temperature and time had significant and positive effects on responses. The three significant variables were also found to play positive and significant roles during pretreatment of sugarcane bagasse [3] and poplar (Populus deltoides) biomass [23], respectively. Whereas, it was reported that pretreatment time had insignificant effect during microwave assisted sodium hydroxide pretreatment of wheat straw [1] and sodium hydroxide concentration had insignificant effect during microwave assisted sodium hydroxide pretreatment of cassava stem [24], respectively. In this work, variables including solid loading and particle size had insignificant effects during pretreatment. However, it was also reported that particle size had significant effect during aqueous ammonia pretreatment of sugarcane bagasse [25] and solid loading had significant effect during microwave assisted sodium hydroxide pretreatment of cassava stem [24]. In our opinion, discrepancy of significant factors for pretreatment in different reports was mainly related with differences of pretreatment methods and lignocellulosic substrates. In this work, particle size had negative effect during pretreatment, as smaller particle could increase surface area of biomass for pretreatment and the following enzymatic hydrolysis [25]. Solid loading had positive effect during pretreatment, as too lower levels of solid loading could lead to too higher sodium hydroxide concentration, which could intensify degradation of cellulose and hemicellulose simultaneously [26]. Therefore, WSP with particle size 0.3 mm was adopted and pretreated at 25% (w/v) of solid loading in the following experiments. As to the three significant variables, the steepest ascent method was used to determine the optimal regions of them (Table 3). Results in Table 3 indicated that reducing sugar and xylose yields reached the plateau while sodium hydroxide concentration, pretreatment temperature and time were 1.8% (w/v), 95.0 °C and 45.0 min, respectively. After the plateau, sugars yields decreased as too severe conditions may result in more lose of cellulose and hemicellulose.
Based on results of the steepest ascent method, CCD (Table 4) was adopted to determine the optimal values of the significant factors. Statistical analysis of CCD (Table 5) indicated that both linear terms (P = 0.000, P = 0.000) and square terms (P = 0.000, P = 0.000) had significant effects on reducing sugar and xylose yields, respectively. P values of the models (P = 0.000, P = 0.000 ) and lack of fit (P = 0.104, P = 0.122) indicated that the models were also adequate to predict the optimal pretreatment conditions. High values of R2 (99.6%, 99.5%) and adjusted R2 (99.2%, 99.1%) also indicated the accuracy of the models. According to canonical analysis, maximal reducing sugar yield (290.99 mg/g) could be obtained after pretreatment using 1.84% (w/v) sodium hydroxide solution at 94.1 °C for 47.0 min. Whereas, maximum xylose yield (52.77 mg/g) could be obtained after pretreatment using 1.76% (w/v) sodium hydroxide solution at 93.8 °C for 43.9 min.
The corresponding regression models for reducing sugar and xylose yields during optimization of pretreatment conditions were given below in Eq. (9) and Eq. (10):
Y1 = 289.737 + 6.537x1 – 3.445x2 + 6.472x3 – 24.540x12 – 20.085x22 – 16.468x32 – 0.103x1x2 + 1.365x1x3 + 0.180x2x3 (9)
Y2 = 52.5066 – 1.6063x1 – 1.0720x2 – 1.4076x3 – 5.6265x12 – 4.6030x22 – 6.2700x32 + 0.3262x1x2 – 0.1138x1x3 – 0.1013x2x3 (10)
in which, Y1 and Y2 were predicted reducing sugar and xylose yields, x1, x2 and x3 were codes of sodium hydroxide concentration, pretreatment temperature and time, respectively.
To validate the predicted conditions, after adjustment, WSP was pretreated using 1.8% (w/v) sodium hydroxide solution at 94.0 °C for 46.0 min. After enzymatic hydrolysis, reducing sugar (291.91 mg/g) and xylose (53.20 mg/g) (average of three replicates) were obtained, which were in close proximity with the predicted values of models. Furthermore, cellulosic compositions of raw and pretreated WSP were determined and the results were shown in Table 6. After calculation, values of solid recovery (74.2%), cellulose recovery (90.9%), hemicellulose recovery (54.6%) and lignin removal (72.7%) were obtained, respectively. Comparisons of the above parameters in this work with those in some previous literature about pretreatment of wheat straw and rice straw were shown in Table 7.
Among different values of cellulose recovery in Table 7, cellulose recovery (90.9%) in this work was relatively lower than 97.9% described by Qiu et al. [27] and 96.0% described by Jaisamut et al. [28], whereas relatively higher levels of lignin removal (72.7%), hemicellulose recovery (54.6%) and solid loading (25.0%, w/v) in this work compared favorable. Though cellulose recovery (92.59%) and lignin removal (95.2%) described by Li et al. [29] were also higher than those (90.9%, 72.7%) in this work, relatively higher levels of hemicellulose recovery (54.6%), shorter pretreatment time (46.0 min) and easier operation in this work were still competitive. As to hemicellulose recovery, only 64.8% described by Qi et al [30] was higher than that (54.6%) in this work. Whereas, relatively lower levels of cellulose recovery (74.7%) and lignin removal (65.54%) still existed in that work [30]. Of course, values of hemicellulose recovery (53.38%) and lignin removal (73.17%) described by Tsegaye et al. [31] were approximate with those (54.6%, 72.7%) in this work, whereas higher cellulose recovery (90.9%) in this work was also competitive. In addition, it was obvious that sodium hydroxide pretreatment of wheat straw was also performed by Tsegayea et al. [1] and the adopted lower sodium hydroxide concentration and shorter pretreatment time were more competitive than those in this work. However, disadvantages including lower solid loading, too higher temperature, relatively lower levels of recovery of cellulose and hemicellulose and lower lignin removal still existed in that work [1]. In our opinion, the optimized pretreatment conditions in this work could simultaneously guarantee satisfactory levels of cellulose recovery, hemicellulose recovery and lignin removal by adopting moderate levels of pretreatment temperature and pretreatment time with higher solid loading and uncomplicated operation. Therefore, the pretreated WSP could be applied to the following optimization of enzymatic hydrolysis.
Statistical optimization of enzymatic hydrolysis conditions
Tween-80, one type of surfactant, could prevent enzymes from being absorbed to lignin and allow more cellulases to catalyze enzymatic hydrolysis of substrates more effectively [32]. Therefore, Tween-80 was directly applied to enzymatic hydrolysis in this work. Similar operation existed in enzymatic hydrolysis described by Jin et al. [33]. As shown in Table 8, yields of reducing sugar and xylose varied from each other, which indicated that the adopted variables could influence the responses. Results in Table 9 indicated that variables including enzyme loading, biomass loading and reaction time have positive and significant effects on reducing sugar and xylose yields. Similar results were also found during hydrolysis of paddy straw [34], sugarcane tops [35] and rice straw [36]. In addition, enzyme loading and reaction time were also found to have significant effects during hydrolysis of wheat straw described by Singh and Bishnoi [37]. It was also reported that enzyme loading and biomass loading had significant effects on sugars production during hydrolysis of sweet sorghum bagasse described by Saini et al. [2]. In this work, reaction temperature, pH and Tween-80 concentration had insignificant effects on sugars yields. Whereas, reaction temperature was reported to have significant effect on sugar yields during hydrolysis of corn cob described by Gaiai et al. [12]. As to Tween-80 concentration, it was also reported to have insignificant effect on sugars production during hydrolysis of wheat straw described by Singh and Bishnoi [37]. However, it was mentioned that Tween-80 concentration had significant effect on sugars production during hydrolysis of pine foliage [38] and oil palm empty fruit bunch [39], respectively. In addition, reaction pH was also found to have significant effect on hydrolysis of cotton stalk [40]. In our opinion, variance of significant variables during enzymatic hydrolysis among different reports was related with lignocellulosic substrates types and hydrolytic enzymes sources. Therefore, the optimal regions of the three significant variables were investigated using the steepest ascent method by increasing the levels of them. The corresponding hydrolysis was carried out at 50 °C, pH 4.8 with 0.2% (w/v) of Tween-80, respectively. As shown in Table 10, sugars yields could be improved obviously by increasing the levels of the three significant parameters and yields of reducing sugar and xylose reached the plateau while enzyme loading, biomass loading and reaction time were 8.0 FPU/g, 7.5% (w/v) and 45.0 h, respectively. It is well known that enzymatic hydrolysis should need enough enzyme loading and reaction time to reach the maximum sugars yields. Furthermore, increment of solid loading could also promote enzymatic hydrolysis by improving enzymes accessibility to substrates due to a fixed number of active sites in enzymes to bind the substrates [39]. Whereas, sugars yields decreased after the plateau, which was probably related with feedback inhibition caused by end product, poor stirring caused by too higher biomass loading and attachment of enzymes on amorphous regions of cellulose [38, 41].
Subsequently, the optimal values of the significant factors were further investigated using BBD (Table 11). Statistical analysis of BBD in Table 12 indicated that linear terms and square terms had significant effects on sugars yields. According to P values of the models (P = 0.000, P = 0.000) and lack of fit (0.118, 0.109) along with high values of R2 (99.4%, 98.7%) and adjusted R2 (98.3%, 96.3%), it was obvious that the models were adequate to predict reducing sugar and xylose yields, respectively.
According to canonical analysis, maximal yield of reducing sugar (633.13 mg/g) could be obtained while enzyme loading, solid loading and reaction time were 8.45 FPU/g, 7.26% (w/v) and 45.9 h, respectively. Whereas, maximal xylose yield (150.29 mg/g) could be obtained while enzyme loading, solid loading and reaction time were 7.76 FPU/g, 6.85% (w/v) and 43.6 h, respectively. The corresponding regression models for reducing sugar and xylose yields were given below in Eq. (11) and Eq. (12):
Y3 = 630.583 + 8.521X1 – 12.395X2 + 11.434X3 – 18.507X12 – 68.759X22 – 30.477X32 – 3.413X1X2 – 3.285X1X3 – 0.167X2X3 (11)
Y4 = 148.617 – 3.217X1 – 5.786X2 – 5.101X3 – 16.270X12 – 10.522X22 – 8.702X32 + 0.455X1X2 + 2.010X1X3 – 1.337X2X3 (12)
in which, Y3 and Y4 were predicted reducing sugar and xylose yields, X1, X2 and X3 were codes of enzyme loading, biomass loading and reaction time, respectively.
In order to determine the accuracy of the models and verify the optimization results, experiments were repeated three times under the adjusted optimized conditions, i.e., enzyme loading 8.1 FPU/g, solid loading 7.1% (w/v) and reaction time 44.8 h. Yields of reducing sugar (632.92 mg/g) and xylose (149.83 mg/g) could be obtained, which were very in close with the predicted values. Compared with the initial yields of reducing sugar (291.91 mg/g) and xylose (53.20 mg/g) under unoptimized conditions, optimization lead to 1.17-fold for reducing sugar yield and 1.82-fold for xylose yield, respectively.
Comparisons of hydrolysis conditions and sugars yields in this work with those in other previous reports were shown in Table 13. Though reducing sugar yield in this work (632.92 mg/g) was lower than 778.30 mg/g described by Annamalai et al. [19] and 772.72 mg/g descirbed by Gupta and Parkhey [36], relatively higher biomass loading (7.1%, w/v) and shorter reaction time (44.8 h) used in this work compared favorable. Though xylose yield (149.83 mg/g) in this work was lower than 156.91 mg/g described by He et al. [42], adoption of higher biomass loading (7.1%, w/v), shorter reaction time (44.8 h) and higher reducing sugar yield (632.92 mg/g) in this work was still competitive. Of course, as shown in Table 13, sugars yields in this work were also higher than those in the other three reports described by Xie et al. [7], Singh and Bishnoi [37] and Patel et al. [43], respectively. In addition, comparisons of enzyme loading among the different reports were not available, as enzymes assay conditions differed from each other. Even so, enzyme loading (8.1 FPU/g) in this work was at the moderate level. Therefore, in our opinion, the optimized hydrolysis conditions could result in considerable sugars yields by adopting moderate levels of enzyme loading, shorter reaction time and higher biomass loading.
Comparisons of pretreatment conditions among different reports were also shown in Table 13. It was obvious that pretreatment temperature (94.0 °C) in this work was higher than that (85.0 °C) during sodium hydroxide pretreatment of rice straw descirbed by Annamalai et al. [19]. However, relatively shorter pretreatment time (46.0 min) and lower sodium hydroxide concentration (1.8%, w/v) in this work were still competitive. Of course, values of pretreatment time (20.0 min, 22 min and 20 min) described by Gupta and Parkhey [36], Singh and Bishnoi [37] and Patel ea al. [43] were shorter than that (46.0 min) in this work. Whereas, we believe that adoption of microwave in the above three reports [36, 37, 43] could enhance equipment input cost for pretreatment and restrict large-scale application of the pretreatment conditions. Furthermore, adoption of lower pretreatment temperature, relatively easier operation and higher solid loading during pretreatment in this work compared favorable to those during pretreatment of wheat straw described by Xie et al. [7]. In addition, compared with pretreatment of corn stover powder described by He et al. [42], levels of sodium hydroxide concentration and temperature in this work were relatively higher. However, relatively higher solid loading and shorter pretreatment time in this work were still advantageous. In general, the optimized pretreatment conditions in this work had some advantages such as relatively higher solid loading, easier operation and lower equipment requirement.