NC State
Chi, S., Yu, G., Zhang, X., Zhang, Y., Liu, C., Li, Z., Li, B., and Cui, Q. (2019). "Response surface optimization of ammonium sulfite pretreatment for fermentable sugar production from wheat straw," BioRes. 14(2), 4603-4622.


The development of a clean and sustainable pretreatment is of great importance for the production of fermentable sugars. In this study, an ammonium sulfite (AS) pretreatment of wheat straw was optimized based on response surface methodology with a three-level, three-factor Box-Behnken design. The investigated factors were AS dosage, pretreatment time, and pretreatment temperature. The effectiveness of the AS pretreatment was evaluated using the standard enzymatic hydrolysis procedure. A second-order polynomial fit was performed to fit the experimental data, and the model analysis showed that the effect of the AS dosage on the final total sugar yields was much more significant than that of the other two factors. Under the optimum pretreatment conditions (27% of the AS dosage (based on the dry wheat straw) at 160 ºC for 63 min), the final total sugar yield achieved was 74.4% after saccharification, which was in agreement with the predicted value (76.5%). Furthermore, it was found that pre-impregnation with acetic acid before AS pretreatment or the post-mechanical refining after AS pretreatment could further increase the fermentable sugar yields to approximately 77%. In addition, the spent liquor containing nitrogen could be used for the production of lignin-based fertilizer, thus making the whole process clean and sustainable.

Download PDF

Full Article

Response Surface Optimization of Ammonium Sulfite Pretreatment for Fermentable Sugar Production from Wheat Straw

Shanshan Chi,a,b,1 Guang Yu,b,c,1 Xihui Zhang, a,b Yuedong Zhang,b Chao Liu,b Zhenqiu Li,a,* Bin Li,b,* and Qiu Cui b

The development of a clean and sustainable pretreatment is of great importance for the production of fermentable sugars. In this study, an ammonium sulfite (AS) pretreatment of wheat straw was optimized based on response surface methodology with a three-level, three-factor Box-Behnken design. The investigated factors were AS dosage, pretreatment time, and pretreatment temperature. The effectiveness of the AS pretreatment was evaluated using the standard enzymatic hydrolysis procedure. A second-order polynomial fit was performed to fit the experimental data, and the model analysis showed that the effect of the AS dosage on the final total sugar yields was much more significant than that of the other two factors. Under the optimum pretreatment conditions (27% of the AS dosage (based on the dry wheat straw) at 160 ºC for 63 min), the final total sugar yield achieved was 74.4% after saccharification, which was in agreement with the predicted value (76.5%). Furthermore, it was found that pre-impregnation with acetic acid before AS pretreatment or the post-mechanical refining after AS pretreatment could further increase the fermentable sugar yields to approximately 77%. In addition, the spent liquor containing nitrogen could be used for the production of lignin-based fertilizer, thus making the whole process clean and sustainable.

Keywords: Wheat straw; Ammonium sulfite pretreatment; Response surface methodology; Fermentable sugars; Biorefinery

Contact information: a: College of Marine Science and Biological Engineering, Qingdao University of Science & Technology, Qingdao 266400, China; b: Chinese Academy of Sciences (CAS) Key Laboratory of Biofuels, Dalian National Laboratory for Clean Energy, Qingdao Institute of Bioenergy and Bioprocess Technology, CAS, Qingdao, Shandong, 266101, China; c: University of Chinese Academy of Sciences, Beijing 100049, China; *Corresponding authors:;

1 Both authors contributed equally to this work.


Due to the increasing consumption of fossil fuels, the development of environmentally friendly and renewable energy has received special attention (Rastogi and Shrivastava 2017). Lignocellulose is the most abundant renewable material with an annual global production of approximately 200 billion metric tons (Juhász et al. 2005), and its utilization is not in competition with food. Thus, converting lignocellulosic biomass into liquid fuel, valuable chemicals, and materials has been the research hotspot in recent years (Moreira et al. 2011; Kumar et al. 2014).

Wheat straw, as one of the most abundant renewable lignocellulose resources (the global annual output of wheat straw was over 600 million tons next to rice straw and corn stover), usually contains 33% to 40% cellulose and 20% to 25% hemicellulose (Prasad et al. 2007), which can be converted to fermentable sugars (i.e., sugar platform) for further conversion. However, most of wheat straw has been burned directly in the field in China for a long time, causing serious environmental issues such as the increase of fine particulate matter (particulate matter (PM) 2.5, particles smaller than 2.5 μm) in the air (Li et al. 2018). Currently, the burning of agricultural waste is strictly prohibited by the Chinese government. Therefore, to make better use of agricultural wastes (e.g., wheat straw), the production of fermentable sugars for the manufacturing of bioethanol has become an urgent concern both for environmental protection and the development of alternative energy sources for vehicles and other forms of transportation (Xie et al. 2018).

Lignocellulosic biomass is mainly composed of cellulose, hemicellulose, and lignin, which are strongly interacted and connected by non-covalent forces, covalent cross-linkages, or hydrogen bonding, resulting in a strong resistance to degradation against outside forces (Ding et al. 2012). Therefore, pretreatment is required to loosen the recalcitrant structure of natural lignocelluloses for saccharification and fermentation for biofuel production (Xu et al. 2016).

Over the past decades, extensive efforts have been made and various pretreatment approaches have been employed, such as biological (e.g., white-rot fungi), physical (e.g., grinding, irradiation), chemical (e.g., alkali, acids, solvents, or supercritical fluids), and a combination of these methods (Galbe and Zacchi 2007; Lee et al. 2009; Fatriasari et al. 2014, 2015). However, different pretreatments have different advantages and disadvantages, such as low efficiency, strict operating conditions, or serious pollution risk, which may dramatically increase the process cost and hinder the feasibility of commercialization (Li et al. 2017a,b).

Furthermore, a high lignin content of lignocellulosic biomass could restrict the enzymatic hydrolysis, because of the nonproductive binding between lignin and enzyme, causing the inactivation of cellulase (Paixão et al. 2016). Therefore, the efficient removal of lignin from lignocelluloses is the most commonly used strategy for biomass pretreatment (Xu et al. 2016). Up to now, several kinds of pretreatment approaches have been developed for lignin removal with varied successes, including alkali-based pretreatments (Li et al. 2017b), organosolv pretreatment (Mou et al. 2014), deep eutectic solvents pretreatment (Satlewal et al. 2018), and sulfite pretreatment (Zhu et al. 2009; Yang et al. 2013). However, the process’ cost or capital cost for chemical or solvent recovery is quite high, particularly for the small-scale mills using agricultural waste (e.g., wheat straw) as the feedstock, due to the limitation of the material collection radius (Xie et al. 2018). Thus, the development of an economical and eco-friendly pretreatment process for efficient delignification of lignocellulose is of crucial importance for the production of fermentable sugars via enzymatic saccharification, particularly for industrial applications (Wu et al. 2018).

In the case of sulfite pretreatment, the degradation and sulfonation of lignin promotes lignin removal, thus increasing the digestibility of the pretreated substrate (Li et al. 2014; Qi et al., 2018). Additionally, the hydrophilicity of the residual lignin could be increased due to the sulfonation, leading to the reduction of non-productive hydrophobic adsorption of enzymes by lignin (Gong et al. 2015). Usually, sodium sulfite is used for sulfite pretreatment (Zhu et al. 2009; Yang et al. 2013), and the chemical recovery has to be well handled (Jin et al. 2015). In the authors’ previous work, ammonium sulfite (AS) pretreatment was used to treat tobacco stalk (Wang et al. 2018) and Eulaliopsis binata (Wu et al. 2018) for efficient delignification, and the spent liquor of AS pretreatment can be used to produce fertilizer due to the high nitrogen content. Qi et al. (2018) studied the impact of AS pretreatment on the improvement of enzymatic hydrolysis of wheat straw by single factor experiments, and verified that lignin could be efficiently removed by sulfonation and ammonolysis during AS pretreatment. But they didn’t investigate the combined effect of the AS pretreatment factors, and the AS pretreatment process was not fully optimized. In addition, the safety of the production of fertilizer using the spent liquor from AS pretreatment was not evaluated.

Therefore, in the present work, AS pretreatment was used to treat wheat straw for the production of fermentable sugars, and the effectiveness of AS pretreatment was evaluated through enzymatic hydrolysis following the standard National Renewable Energy Laboratory (NREL) procedure (Sluiter et al. 2008). Based on the authors’ previous study (single factor experiment), the response surface methodology (RSM) with a three-level, three-factor Box-Behnken design was adopted for the optimization of AS pretreatment and the investigation of the combined impact of the key factors of AS pretreatment on the downstream saccharification. Three key variables of AS dosage, pretreatment temperature, and pretreatment time were investigated for the RSM experiments, and the final total sugar yield after AS pretreatment and saccharification was selected as the response variable. In addition, the mass balance under the optimized conditions was discussed, and the wheat straw before and after the AS pretreatment was comprehensively characterized using elemental analysis, scanning electron microscopy (SEM), Fourier transform infrared (FTIR) spectroscopy, and X-ray diffraction (XRD) to investigate the impact of pretreatment on the property changes of wheat straw. Most importantly, the safety of the spent liquor derived from AS pretreatment for the production of lignin-based fertilizer was evaluated. The manufacture of lignin-based fertilizer using spent liquor could make the whole process of AS pretreatment clean and sustainable. In a word, the clean and effective production of fermentable sugars is of crucial importance for the sustainable conversion of lignocelluloses to bioenergy or chemicals (e.g., furfural, glycol).



Wheat straw was collected from Dingyuan, Anhui province, China. The air-dried straw was ground and milled with a twin-screw extrusion (self-designed; Manufactured by Tianzheng Screening Pulping Equipment Co., Ltd., Hebei, China) system (the maximum throughput was 200 kg/h) before the experiments (Liu et al. 2013), and the milled wheat straw (with the size of about 1 cm) was stored in sealed plastic bags at 4 ºC for moisture balance before pretreatment. According to the analysis based on the NREL procedure, the wheat straw contained 34.6% ± 0.3% glucan, 16.2% ± 0.1% xylan, 1.0% ± 0.0% arabinan, 0.7% ± 0.0% acetyl groups, 22.6% ± 0.2% lignin, 14.4% ± 0.2% extractives, and 10.6% ± 0.0% ash. All analytical-grade chemical reagents, including ammonium sulfite (AS), sodium citrate, citric acid, ethanol, acetic acid, and sulfuric acid, were obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) and directly used without further purification. The cellulase (from Trichoderma reesei) used for enzymatic hydrolysis was gifted by Qingdao Vland Biotech Inc. (Qingdao, China), and the activity of cellulase was 85 FPU/mL, which was determined by the standard procedure (Ghose 1987).

Pretreatment of wheat straw

The AS pretreatments were completed in a Parr Hastelloy reactor (Series 4560; Moline, IL, USA) that consisted of a heater, stirrer, stainless steel vessel (300 mL), and controllers for the adjustment of stirring speed and pretreatment temperature. For each pretreatment, 10 g of wheat straw (dry weight basis) was used with a certain AS dosage and a solid/liquid ratio of 1:8 (w/w). The acquired amount of AS was dissolved in water to make AS solution before use, and then wheat straw was added in AS solution. After sufficient mixing, the mixtures were heated from room temperature to the designated temperature (150, 160, and 170 °C, respectively) with a stirring speed of 200 rpm (optimized). The key factors of pretreatment time, pretreatment temperature, and AS dosage were investigated for the optimization of AS pretreatment. After pretreatment, the heater was turned off and the cooking pot was rapidly cooled for 10 min with cool water. Then, the pretreated material was removed and separated into solid (pretreated wheat straw) and liquid (spent liquor) using a filtration with non-woven fabric bag (self-made) with 500 mesh. The pretreated stock was washed with de-ionized water until neutrality. The washed stock and spent liquor samples were stored at 4 °C for further tests and analysis. All of the experiments were conducted in triplicate in this study and the average value was reported.

In addition, the scale-up experiment of AS pretreatment under the optimized conditions was also conducted in a cooking digester (VRD-42SD-A; China Pulp and Paper Research Institute, Beijing, China) with a total sample amount of 200 g (dry weight straw).

Additionally, a higher content of extractives and ash in the substrate could hinder the downstream enzymatic hydrolysis (Yuan et al. 2018), and post mechanical refining could increase the specific surface area of substrate to facilitate the enzymatic hydrolysis (Xu et al. 2014). Hence, a pre-impregnation stage (for the removal of extractives and ash) before AS pretreatment and post-mechanical refining (for the increase of the specific surface of substrate) after AS pretreatment was employed to further improve the enzymatic digestibility of the pretreated wheat straw. According to the authors’ previous work, pre-impregnation was performed at 40 °C for 3 h with 3 wt% of acetic acid (AA) dosage (based on the dry weight of straw) and a solid/liquid ratio of 1:10 (w/w). Upon completion of pre-impregnation, the solid stock was washed with de-ionized water until neutrality, and then the washed stock was stored at 4 °C for further treatment and analysis.

As for post-mechanical refining, the pretreated and washed wheat straw was refined by a pulp refining instrument (PFI) mill (PL11-00; Xianyang TEST Equipment Co., Ltd., Xianyang, China), which was conducted with 10 wt% of solid consistency, 4000 revolution numbers, and refining clearance of 0.24 mm, based on the authors’ previous work (Xu et al. 2014). Afterwards, the refined pulp was collected and stored at 4 °C for the further evaluation of enzymatic hydrolysis.

Experimental Design

To examine the individual and combined effects of the key pretreatment variables on the response factor, the Box-Behnken design with three key variables was employed to optimize the AS pretreatment conditions. The selected independent variables were pretreatment temperature (°C), AS dosage (wt%), and pretreatment time (min). The total sugar yield (%) was selected as a response for analysis. The levels of factors were chosen based on preliminary trials to ensure a proper range. A set of 17 trials that ran with five replicates at central points were designed using Design-Expert 8.0.6 software (Stat-Ease Inc., Minneapolis, MN, USA). Each of the trial points was replicated three times.

The multifactor experiment was designed by the interaction between the pretreatment temperature, pretreatment time, and AS dosage, and the response of the total sugar yield to the varied conditions were modeled via RSM to obtain the optimal conditions. The ranges of factors were chosen based on preliminary trials to describe the reaction space. The independent variables were transformed to the range between -1 and 1 for the appraisals of factors and the corresponding levels for the three variables are summarized in Table 1. The data were fitted with a regression model that was described using a quadratic polynomial equation, and the fit of the models were evaluated through comparing R2 and the adjusted R2. The validations under the optimized conditions and critical conditions were performed, and the statistical analysis of the data was performed using an analysis of variance (ANOVA).

Table 1. Ranges of Each Independent Variable in the Box-Behnken Design


Enzymatic hydrolysis

The enzymatic hydrolysis of pretreated wheat straw was performed in 50-mL bottles with the addition of 0.4 g pretreated biomass (dry weight) and cellulase (25 FPU/g-substrate, or 30 FPU/g-glucan). The pH was adjusted to 4.8 with a sodium citrate buffer, and 0.02% sodium azide was added to prevent microbial contamination, resulting in a final substrate consistency of 2% (w/v). The mixture was then incubated in an incubator shaker at 50 °C with 120 rpm for 48 h. At the end of the process, the mixture was immediately cooled to 4 °C to stop saccharification, and then the supernatants were filtered through a membrane (pore size of 0.22 μm) and analyzed for monomeric sugars using high performance liquid chromatography (HPLC).

Composition analysis

The compositional analysis of the native and pretreated wheat straw was determined using the two-step sulfuric acid hydrolysis method following the NREL analytical procedure (Sluiter et al. 2008). The amount of monomeric sugars in the liquid fractions were measured directly by an HPLC system (Model 1200; Agilent Technologies, Santa Clara, CA, USA) after neutralization and filtration through 0.22-μm nylon filters. The HPLC system was equipped with a Bio-Rad Aminex HPX-87H column (Bio-Rad Laboratories, Waters, USA) at 55 °C and a refractive index detector (Chengdu GELAI Co., Ltd., Chengdu, China). The analysis was completed using sulfuric acid (0.005 M) as the mobile phase at a flow rate of 0.6 mL/min for 20 min. The evaluation of AS pretreatment and enzymatic hydrolysis was calculated using the following equations,

where M refers to the mass of the corresponding substance (g), C refers to the content of the corresponding component in biomass (wt%), and Rsolid (%) is the percentage of solid recovery after pretreatment. Rglucan (%) and Rxylan (%) are the recovery rates of glucan and xylan, respectively, and Dlignin (%) is the delignification rate. The total sugar yield (Ytotal sugar, %) was calculated as the percentage of the total sugar (glucose + xylose) in enzymatic hydrolysate divided by the total sugar in the corresponding raw material. All of the experiments were performed in triplicate, and the average data were reported.


The FTIR analysis was conducted to detect the changes in the functional groups of the raw and pretreated wheat straw. Sample tablets were prepared by mixing each sample with potassium bromide (the ratio of sample to KBr was 1:100) before analysis. The spectra were recorded within the frequency range of 400 cm-1 to 4000 cm-1 using a FTIR spectrometer (Nicolet 6700; Thermo Fisher Scientific Inc., Waltham, MA, USA) with a detector at 4 cm-1 resolution and 32 scans per sample.

The surface morphologies and characteristics of the untreated and pretreated wheat straw were analyzed using SEM (S4800; Hitachi, Tokyo, Japan) and energy dispersive spectroscopy (EDS) (D8 Advance; Bruker, Karlsruhe, Germany) with the accelerating voltage of 5.0 kV and the backscatter electron detector. The images were taken with the magnification of 2000, and the working distance was 8.1-8.6 mm for different images. The freeze-dried samples were sputter coated (vacuum spraying for 90 s) with a thin gold layer of 2.5-2.8 nm in thickness (Hitachi, Tokyo, Japan) prior to analysis.

Inductively coupled plasma mass spectrometry (ICP-MS) (Agilent ICP-OES 730; Agilent Technologies, Santa Clara, CA, USA) as a highly sensitive mass spectrometry was used on the spent liquor samples to assess the presence of several trace elements.

The crystallinities of the lyophilized samples were measured using an X-ray diffractometer (D8 ADVANCE; Bruker Co., Karlsruhe, Germany). The scattering angle (2θ) ranged from 10º to 50º with a scanning speed of 0.5º per min and the Ni-filtered Cu Kα radiation was generated at 80 mA and 40 kV. The crystallinity index (CrI) was calculated according to the empirical method (Segal et al. 1959),

where I002 is the maximum of the diffraction intensity (the 002 plane) and Iam is the minimum diffraction intensity (between 101 peaks and 002 peaks).


Model Fitting and Statistical Analysis

A summary of the experimental design is listed in Table 2. The final total sugar yields (Ytotal sugar) for the 17 trials were in the range of 42.4% to 75.8%, while the corresponding delignification rate ranged from 44.7% to 78.0%. Severe pretreatment conditions could lead to a higher lignin removal. For instance, for Trial 1 with 10 wt% AS dosage at 160 °C for 40 min, the Dlignin was 57.6%, which was lower compared to Trial 2 (Dlignin of 66.2%) that had a higher AS dosage (30 wt%). However, excessive reaction severity could cause the over-degradation of carbohydrates. To obtain a high final total sugar yield, a high delignification rate was needed for the improvement of cellulase accessibility, and simultaneously, a higher sugar recovery was also required after pretreatment (Mathew et al. 2011; Kim and Han 2012). Therefore, Trial 14 with the Dlignin of 67% and solid recovery of 61% had the highest Ytotal sugar (75.8%), but Trial 8 with a relatively higher Dlignin of 78% had a lower Ytotal sugar (66.6%) because the corresponding Rsolid was only 47%, which indicated a severe sugar loss during pretreatment.

Table 2. Summary of the Box-Behnken Design Used for RSM Analysis and the Corresponding Responses

According to the regression analysis of each response, quadratic models were selected, as verified by the software based on the effect of the combined independent variables on the responses. The Ytotal sugar for the trials was set as the responsive variable for modelling to reach the highest final total sugar yield. Thus, the final quadratic polynomial model in terms of actual factors is shown by Eq. 7,

Ytotal sugar (%) = -2382.3755 + 12.3636X+ 3.1409X+ 27.5134X3 – 2.5 × 10-5X1X2 – 0.0582X1X3 – 0.0133X2X3 – 0.0579X12 – 8.0819 × 10-3X22 – 0.0787X3(7)

where, X1, X2, and X3 are the AS dosage (wt%), pretreatment time (min), and pretreatment temperature (°C), respectively.

In the above equation, positive signs of the regression coefficients showed a synergistic effect, while negative signs indicated an antagonistic effect. In addition, statistical analysis of the full quadratic models was performed using an analysis of variance (ANOVA) to evaluate the effects of the variables on Ytotal sugar and the possible interaction between the key variables. The fit of the models were evaluated by comparing the R2 and the adjusted R2. The statistical significance was determined by an F-test and P-value. It has been suggested that the R2 value should be no less than 0.80, and the P-value should be less than 0.05, for a good fit of the developed model (Ruangmee and Sangwichien 2013; Auxenfans et al. 2014).

The ANOVA results of the developed model are presented in Table 3. The model fit the data with an R2 of 0.9692 for Ytotal sugar, which implied that only 3.08% of the total variation could not be explained by the model. This regression model was significant at the 99.98% confidence level indicating that the model equation adequately described the response. The low P-value (0.0002) of the model also confirmed that the fit was good and the predictive power of the adjusted model was rather high. Additionally, the lack of fit (LOF) value represented the misfit probability of the predicted and adjusted coefficient, and no significant LOF indicated a good fitting of the data (Sindhu et al. 2011). The F-value of 4.68 implied the LOF was not significant relative to the pure error, and there was only an 8.50% chance that this large (F) could occur because of noise.

Table 3. ANOVA for Response Surface Quadratic Model for Total Sugar Yield

Moreover, as shown in Table 3, the AS dosage (X1) and pretreatment temperature (X3) (P < 0.05) were significant in affecting the total sugar yield, while the impact of pretreatment time (X2) was not significant (P > 0.05). According to the P-value, the impact of independent variables followed the order X1 > X3 > X2, and the quadratic terms of X12, X22, and X32 were all significant with a P-value less than 0.05. Additionally, the interactions of X1X3 were significant, while the interactions of X1X2 and X2X3 were not significant. All statistical data showed that the developed model could be used to analyze and predict the relationship between the pretreatment variables and Ytotal sugar.

Impact of Variables on Response

Response surfaces for the impacts of variables and their mutual effects on Ytotal sugar are shown in Fig. 1 (A to C), and the corresponding contour plots are presented in Fig. 1 (a to c), respectively. Two variables were varied in the certain ranges, while the other one was fixed at the central value. The slope of response surface indicated the extent of the response with pretreatment conditions. For example, the steep curved surface meant that the response value was very sensitive to the pretreatment conditions. At the same time, the shape of the contour plot can reflect the intensity of the interaction effects. In general, the oval reveals a strong interaction of the two factors, which was opposite to the circle (Lu et al. 2017).

Fig. 1. Response surfaces and contour plots for the impacts of variables and their mutual effects on total sugar yield: (A, a) AS dosage and time (temperature = 160 ºC); (B, b) time and temperature (AS dosage = 20%); and (C, c) AS dosage and temperature (time = 60 min)

Figure 1A shows the effects of pretreatment time and AS dosage on the final total sugar yield. The Ytotal sugar increased with increased pretreatment time when the AS dosage was in the range of 10 wt% to 30 wt%, but when the AS dosage was fixed, the impact of the pretreatment time on Ytotal sugar was relatively weak. As shown in Fig. 1B, with the prolonging of pretreatment time, the Ytotal sugar first increased and then gradually decreased, which indicated that the excessive pretreatment time could lead to the severe degradation of carbohydrates (Li et al. 2014), reducing the sugar recovery after pretreatment. Therefore, to obtain a higher Ytotal sugar, the pretreatment time should be properly shortened. In addition, the combined effect of pretreatment temperature and AS dosage on Ytotal sugar is displayed in Fig. 1C, which exhibits that when the pretreatment time was fixed at 60 min, both the increase of AS dosage and pretreatment temperature were beneficial for the improvement of Ytotal sugar; however, excessively severe pretreatment conditions were not needed. An excessively high AS dosage (e.g., 30 wt%) and pretreatment temperature (170 ºC) could increase the process’ cost and cause over-degradation of carbohydrates, resulting in a lower final total sugar yield. Glucan was more resistant to harsh conditions when compared with xylan (Xu et al. 2016), but excessively high pretreatment temperatures could lead to a degradation of sugars, thus gaining a lower solid recovery after pretreatment (Table 2).

In addition, Fig. 1a and Fig. 1b clearly show that the change of Ytotal sugar was not obvious with the increase of pretreatment time under the constant AS dosage and pretreatment temperature, indicating the interactions of the AS dosage vs. pretreatment time and pretreatment temperature vs. pretreatment time were not significant. However, the interaction between the AS dosage and pretreatment temperature was strong, as presented in Fig. 1c. The results were in line with the ANOVA analysis (Table 3).

Validation of the Model

Based on the above model analysis, the optimized AS pretreatment conditions were an AS dosage of 27 wt%, a pretreatment time of 63 min, and a pretreatment temperature of 160 ºC. Under these optimized conditions, the predicted Ytotal sugar was 76.5%. To validate the adequacy and appropriateness of the model Eq. 5, additional AS pretreatment experiments under the optimum conditions were repeatedly conducted, and the experiment result was 75.8% ± 0.6%, which was in good agreement with the predicted value. The authors also did a scale-up of the AS pretreatment experiment with a larger sample size of 200 g (dry wheat straw), and it verified that the scale-up result (74.4% ± 0.5%) also mostly matched the predicted value (76.5%) by the developed model with the accuracy rate of 97.3%. The relatively lower result of the scale-up experiment was probably due to the different reactors with different mixing approaches (i.e., small size reactor with mechanical stirring, while large size reactor with the blending of the rotational reactor (Chi et al. 2019)). The mixing technique could affect the mass transfer during pretreatment, thus influencing the result. The substrate obtained from the scale-up experiment was used for saccharification and the evaluation of mass balance. In addition, the accuracy rate of the predicted value under the critical conditions was larger than 95% as well.

Enzymatic Hydrolysis of the Pretreated Wheat Straw

After AS pretreatment under the optimum conditions, the pretreated wheat straw was subjected to enzymatic hydrolysis. The sugar conversion and the sugar yields as the function of saccharification time are displayed in Figs. 2a and 2b, respectively. The sugar conversion was calculated as the percent of the glucose/xylose in hydrolysate divided by the corresponding glucose/xylose in the substrates (i.e., pretreated wheat straw). As shown, the enzymolysis ran rapidly in the first 4 h and then gradually slowed down. This trend was attributed to the increased concentration of fermentable sugars in the hydrolysate, which inhibited the enzymatic hydrolysis of the substrate (Xu et al. 2015). Similar results were also reported in previous studies (Liu et al. 2013; Gong et al. 2015). Figure 2a exhibits that the glucan conversion was up to 99%, indicating that almost all glucan in the pretreated wheat straw was hydrolyzed to glucose. In other words, the wheat straw had a very good digestibility of glucan after the AS pretreatment under the optimum conditions. Yet, approximately 89% of xylan conversion was obtained, which was probably because the xylanase content in the enzyme was not high enough. Furthermore, Fig. 2b presents that the glucose yield and xylose yield were approximately 80% and 60%, respectively, although the corresponding sugar conversion was relatively high in saccharification. These results were due to the degradation of carbohydrates during AS pretreatment, particularly for xylan. To reduce the sugar loss in pretreatment, the addition of suitable carbohydrate stabilizers (e.g., anthraquinone or magnesium salts) could be a possible solution (Xu et al. 2015). In addition, the Ytotal sugar was 74.4% after 48 h saccharification, which was 3.8 times higher in comparison with the raw wheat straw (19.6%).

Fig. 2. Sugar conversion (a) and sugar yields (b) as the function of the enzymatic hydrolysis time

Effect of Pre-impregnation and Mechanical Refining on Enzymatic Hydrolysis

It has been demonstrated that the high content of extractive and ash in a substrate can impede the enzymatic hydrolysis (Mou et al. 2013; Yuan et al. 2018), and post-mechanical refining after chemical pretreatment can increase the porosity of the substrate and decrease the crystallinity of cellulose, thus improving enzymatic saccharification (Xu et al. 2014). It was reported that about 10 to 20% improvement of biomass digestibility could be achieved by post-mechanical refining (Chen et al. 2013), but the improvement was highly dependent upon the severity of chemical treatment (Jones et al. 2014; Xu et al. 2016). Therefore, to further ameliorate the final total sugar yield, pre-impregnation before AS pretreatment and post-mechanical refining after AS pretreatment were completed. Figure 3 shows the constituent changes of wheat straw before and after different treatments. In Fig. 3, both glucan and xylan contents in wheat straw increased after pre-impregnation, which was mainly due to the removal of extractives and ash (just 24% of lignin was removed during pre-impregnation). The removal rates of extractives and ash were 57% and 43%, respectively, after pre-impregnation. Compared to the raw material, the glucan content increased from 34.6% to 48.9% with the AS pretreatment alone. This phenomenon was due to the efficient delignification (74%), as well as the removal of extractives and ash, hence leading to high sugar conversion in saccharification (Fig. 2a). In addition, the glucan and xylan contents were further increased after pre-impregnation combined with AS pretreatment (Fig. 3), and a larger removal of extractives and ash could lead to a higher accessibility of enzymes to polysaccharides (Yuan et al. 2018), thus further raising the final total sugar yield from 74.4% ± 0.5% (without pre-impregnation) to 77% ± 0.6%.

In contrast, as expected, post-mechanical refining after AS pretreatment increased the Ytotal sugar to 76.9% ± 1.4%. These results indicated that we could either do pre-impregnation or do post mechanical refining to further improve enzymatic hydrolysis. Additionally, by combining pre-impregnation, AS pretreatment, and post-mechanical refining, the Ytotal sugar could reach to 78% ± 0.8%, which was comparable with the reported alkali-based pretreatments (Gong et al. 2015; Xu et al. 2015). However, the full combination of pre-impregnation and post mechanical refining for AS pretreatment was not feasible with only about 1% increase of Ytotal sugar, due to the increase of process cost.

Fig. 3. Constituent of raw material and the samples after different treatments (AS pretreatment was run under the optimum conditions)

Physicochemical Properties of the Pretreated Wheat Straw

To further evaluate the effectiveness of AS pretreatment, the changes of the physicochemical properties of the pretreated wheat straw were comprehensively characterized. Figure 4 presents the morphology changes of wheat straw before and after pretreatment. As shown in Fig. 4a, the untreated wheat straw appeared as a smooth, tight, and rigid surface with a highly ordered surface structure, which hindered the accessibility of cellulase to cellulose.