Ethanol production from sugarcane bagasse by fed‐batch simultaneous saccharification and fermentation at high solids loading

To make the bioethanol conversion process economically viable, improving solids loading (>15% [w/v]) in different unit operations is crucial. In this study, alkali pretreatment of sugarcane bagasse was carried out at high solids loading. Temperature and pH were optimised via batch simultaneous saccharification and fermentation (SSF) experiments using Saccharomyces cerevisiae Y‐2034 or Kluyveromyces marxianus NCYC‐587. The effects of enzyme addition mode and solids loading on ethanol production were studied in fed‐batch SSF experiments. It was demonstrated that the optimised pH and temperature for both S. cerevisiae Y‐2034 and K. marxianus NCYC‐587 were 5.2 and 37°C. The mode of enzyme addition “one‐time dose” improved ethanol productivity better than the “batch addition” mode. With fresh substrates added, solids loading strongly increased from 19% to 33% (w/v), corresponded to a small decline in the ethanol yield. At the final solids loading of 33% (w/v), the highest ethanol concentration of 75.57 ± 2.14 g/L was obtained, with ethanol productivity of 0.63 ± 0.02 g/(L h) and a yield of 66.17 ± 1.87%.

bioethanol, which is considered a suitable substitute of fossil fuels with benefits for the economy and environment. 2 In the past decades, bioethanol manufacture from lignocellulosic materials received much research attention. 3 Many researchers have investigated the influences of high solids content (>15% solids, w/w) on different unit operations in order to make the process of conversing lignocellulosic biomass to ethanol economically viable. 4 It is expected that performing this process at high solids loading-particularly with inclusion of high-solids pretreatment following up high-solids hydrolysis and fermentation-is a great means to improve the process efficiency. 5 Less water is required for the pretreatment and enzymatic hydrolysis. Capital costs are decreased due to the shrinking sizes of reactors and equipment for even saccharides and ethanol concentrations. Energy consumption during heating, cooling, agitating, and ethanol distillation is decreased. 6 Technical challenges are also caused by high-solids loading, for example, enhanced difficulties with stirring and mixing, limited heat and mass transfer and thus a prolonged process. 7 To decrease these negative impacts and simultaneously reduce the inhibitory effect of high glucose concentration on cellulase, a simultaneous saccharification and fermentation (SSF) process combined with a fed-batch measure have been regarded as a feasible approach. 8 During saccharification, the substrate is degraded and converted to glucose, which will be quickly transformed into ethanol by a fermenting microorganism. 9 More free water will be liberated. The system viscosity in the slurry is decreased, and diffusion and mixing limitations are minimised. The liquefaction could thereby be beneficial to add a fresh substrate and mitigate the issue of high viscosity.
Strategies which contribute to the fermenting efficiency of SSF mainly include increasing substrate loading, choosing a favorable yeast strain, and optimizing temperature and pH value. 10 It was proposed that the pH value between 4.5 and 5.0 was favorable for saccharification of lignocellulosic materials, while the most suitable pH range for yeast fermentation was 5.0-6.0. 11 The conflict of optimal pH value during enzymatic hydrolysis and fermentation processes decreased the ethanol concentration produced by SSF process. The optimal enzyme activity of cellulases is shown in the range of 45-55°C, while most microorganisms have good ethanol productivity and yield in the range of 28-35°C. 12 To improve the effectiveness of SSF process, variations between the optimal temperatures for the cellulase and microorganisms may as well be alleviated. Many researchers have focused on carry out SSF of cellulosic substrates via thermotolerant strains such as Kluyveromyces marxianus 13 and Kluyveromyces fragilis. 14 The comparison of fermenting capacities was necessary between normal temperature strain-Saccharomyces cerevisiae and thermotolerant strain-K. marxianus under optimized conditions for SSF.
There had been many reports on combining fed-batch additions of substrate with SSF, which showed a good potential for increasing ethanol concentrations. A research group 10 reported that a high ethanol concentration of 84.7 g/L can be achieved using corncob pretreated by combinatorial methods at a solids loading of 25% (w/v). However, both dilute sulfuric acid and sodium hydroxide pre-treatments were involved, which obviously increased the energy consumption and equipment investment. Another group 15 employed diluted NaOH-pre-treated SCB as a feedstock and obtained an ethanol concentration of 68.05 g/L and an ethanol productivity of 0.57 g/(L h) at 30% (w/v) solids loading. The pretreatment process was conducted at low solids loading of 5% (w/v), which led to huge water consumption. Some researchers 16 yielded only 40 g/L ethanol using spruce pretreated by SO 2 and steam explosion at a solids loading of 20% (w/v). To make the process of conversing lignocellulosic biomass to ethanol economically viable on an industrial scale, selecting a substrate pretreated by a low-energy-consuming way at high-solids loading and optimisation of the fed-batch SSF process for high bioethanol production might be necessary but lack of systematic studies.
Alkali pretreatments may increase the specific surface area and carbohydrate accessibility to cellulase. They are effective at dissolving lignin while preserving most of cellulose and hemicellulose. 17 As one of the most ordinary means applying to the industrial production for pretreating a cellulosic biomass, alkali pretreatment is considered promising for high-solids pretreatments. 18,19 In the present research, crude SCB was pretreated with diluted NaOH at a solid to liquid ratio of 1:6. Temperature and pH were optimised in batch SSF processes with S. cerevisiae Y-2034 or K. marxianus NCYC-587, which represent conventional yeast and thermotolerant yeast. To increase the final ethanol concentration and productivity, different enzyme addition modes and solids loading levels were evaluated in the fed-batch SSF processes. The findings of this investigation will help in understanding bioethanol production at high-solids loading during all unit operations as an important utilization means of SCBs.

| Materials
Sugarcane bagasse was supplied by Guangxi Feng Hao Group Company Limited. (Pingxiang, China). After milling, the segments between mesh sizes 20 and 80 were screened and employed for experiments. Cellic CTec2 was used as cellulase provided by Novozymes A/S (Bagsaevrd, Denmark); the activity of 310 FPU/mL was tested on the basis of the method of the IUPAC. 20 Two fermenting microorganisms, K. marxianus NCYC-587 and S. cerevisiae Y-2034, were used in the batch SSF processes, which were supplied by the National Collection of Yeast Cultures (Norwich, UK) and National Center for Agricultural Utilization Research (Peoria, IL), respectively. Both inoculums were prepared in 150 mL Erlenmeyer flasks with 50 mL growth medium consisted of 20 g/L glucose, 20 g/L peptone, and 10 g/L yeast extract. K. marxianus NCYC-587 was cultivated at 42°C for 12 hours at 150 rpm on a shaker, whereas S. cerevisiae Y-2034 was grown at 30°C for 12 hours. Each inoculum was employed in all SSF experiments at 0.1 g/L (dry weight/volume).

| Alkali pretreatment
For removing lignin from the complex cellulosic material and to improve the subsequent processes, SCB was pretreated with a 1.8% (w/v) NaOH solution for 60 minutes at 110°C according to the method published elsewhere. 18,19 The solid residues were washed with tap water to neutral and then dried in a forced-air oven at 50°C. It was kept in a desiccator for the subsequent SSF experiments.

| Batch SSF experiments
For batch SSF experiments with K. marxianus NCYC-587 or S. cerevisiae Y-2034, optimal pH and temperature were investigated. A fermentation medium that consisted of 2 g/L yeast extract, 5 g/L KH 2 PO 4 , 2 g/L (NH 4 ) 2 SO 4 , 0.4 g/L MgSO 4 ·7H 2 O, 12% (w/v) alkali-pre-treated SCB was loaded in 250 mL Erlenmeyer flasks with 100 mL of 0.2 mol/L sodium acetate buffer. Cellulase loading was 10 FPU/(g substrate). Firstly, a series of batch SSF experiments was conducted at the initial pH of 4.8, 5.2, 5.6, or 6.0. Our 12% (w/v) SCB was prehydrolysed for 12 hours at 50°C and 150 rpm. As for S. cerevisiae Y-2034 and K. marxianus NCYC-587, the temperatures were turned down to 30-42°C; and the inocula were added into the flasks for the SSF experiments. After ethanol yield detection, the optimal pH for SSF was determined. As for S. cerevisiae Y-2034, after prehydrolysis, the temperatures were set to 30, 34, 37, or 40°C to conduct the SSF process. For K. marxianus NCYC-587, the temperatures were set to 37, 40, 42, or 45°C to conduct the SSF experiments. The overall SSF duration including prehydrolysis and fermentation was 96 hours. Rubber plugs were used to seal the Erlenmeyer flasks. Time points for collecting samples were set at 0, 12, 24, 48, 72, and 96 hours.

| Fed-batch SSF experiments
The optimal pH and temperature determined in the batch SSF were taken for the initial condition for the fed-batch SSF experiments. The fed-batch SSF experiments were initiated with 12% (w/v) solids loading. To confirm two modes of enzyme, adding of enzyme-"one-time dose" and "batch addition", fed-batch processes "A" and "B"-were executed. The fresh substrates (7% [w/v]) were fed after 6 hours to get the final solids loading of 19% (w/v), and the enzyme loading of 10 FPU/(g substrate) was used for both experiments. For process A, cellulosic enzyme was added into the flask at the beginning of hydrolysis in accordance with 19% (w/v) dry solids loading. For process B, enzyme was fed companying with the corresponding amount of the substrate. To further increase the ethanol concentration and survey the effect of solids loading on ethanol production, more fresh solids were fed. After prehydrolysis for 12 hours, 4% (v/v) S. cerevisiae Y-2034 was added into flasks, and simultaneously, the temperature was set to the optimal one. After that, 7% (w/v) of fresh solids was fed consecutively at 6 and 12 hours to attain the final solid loading of 26% (w/v) and one more at 24 hours to achieve the final solids loading of 33% (w/v). Time points for collecting samples were set at 0, 12, 24, 48, 72, 96, and 120 hours to detecting the ethanol and sugar concentrations. All the experiments for each operation condition were carried out in duplicate.

| Analytical methods
The proportions of cellulose, hemicellulose, and lignin were estimated using the standardised protocols established by the National Renewable Energy Laboratory (NREL, Golden, CO). 21 Concentrations of glucose, xylose, and cellobiose were analysed on a high-performance liquid chromatography system (Waters 2695, Milford, MA, USA). Ethanol concentrations were measured by means of a gas chromatograph system (Agilent HP 6820, Santa Clara, CA, USA). The detailed detecting conditions were described in previous literature. 12

| Calculations
The following formula was used to calculate the ethanol yield Ethanol (g/L) is the concentration of ethanol in fermentation broth, Solid (g/L) is the solid loadings, Percentage (Glucan) is the percentage of glucan in the pretreated SCB, and the factor of 0.568 is the theoretical factor for the conversion of cellulose to ethanol. 22

| Alkali pretreatment of SCB
Pretreatment is one important step for production of ethanol from lignocellulosic materials and has a significant effect on the process cost by concluding the extent of sugar recovery, fermentation toxicity, enzymatic hydrolysis rates, enzyme loading, and other process variables. 23 In the present work, dry SCB was pretreated with 1.8% (w/w) NaOH at the S/L ratio of 1:6, and both untreated SCB and pretreated SCB were assayed. The concentrations of glucan, xylan, and lignin in the raw material were 41.95 ± 0.51%, 21.7 ± 0.12%, and 24.11 ± 1.21%. After alkali pre-treatment, the glucan and xylan concentrations increased to 61.01 ± 1.31% and 26.94 ± 0.68%, respectively. Simultaneously, the lignin concentrations decreased obviously from 24.11 ± 1.21% to 8.95 ± 0.33% after structural disruption caused by NaOH. After the swelling of NaOH and the removal of lignin, the compact structure in raw SCB was broken severely, and some small fragments with porous morphology were formed. All these properties should positively affect the interaction between the carbohydrates and enzyme and increase fermentation efficiency. 24 Many studies involving alkali pretreatment of SCB have been reported. Some researchers 25 pretreated SCB at 80°C for 120 minutes at an S/L ratio of 1:10 and produced substrates with glucan, hemicellulose, and lignin concentrations of 63.80 ± 2.19%, 24.30 ± 0.82%, and 7.80 ± 0.79%. This approach may increase the consumption of water and energy and also the cost of equipment dealing with equivalent biomass materials. A pretreatment performed at high solids loading might be more efficient (than that at low solids loading), since a greater amount of biomass available in the reaction would produce higher sugar concentrations and thus ethanol concentrations. 26

| The effect of pH on the SSF process
Four initial pH values were chosen to determine the optimal pH of the SSF process with S. cerevisiae Y-2034 or K. marxianus NCYC-587. As presented in Figure 1, the ethanol concentration varied depending on the initial pH value. When K. marxianus NCYC-587 was added into the system at initial pH 4.8, no ethanol was detected. It may be due to that a lower pH value was harmful to the yeast and blocked the metabolic activity. 18,19 Then, with the initial pH increasing, the ethanol concentration decreased, and the maximum ethanol concentration of 26.29 ± 0.31 g/L was obtained in the system at pH 5.2 at the time point of 72 hours with productivity of 0.37 g/ (L h). No obvious difference in terms of ethanol production was observed between the systems at pH 5.2 and 5.6. The enzymatic hydrolysis process was severely hindered with the increase in pH, and thus ethanol concentration produced was the lowest at pH 6.0. Saccharomyces cerevisiae Y-2034 was added into the system at initial pH 4.8, it had a short lag phase. Less than 10 g/L ethanol was produced and sugar concentration had a small increase at 24 hours. When pH was 5.2, the highest F I G U R E 2 Effects of temperature variation on the ethanol and sugar concentrations during the simultaneous saccharification and fermentation (SSF) process; A, the SSF process with Saccharomyces cerevisiae Y-2034, B, the SSF process with Kluyveromyces marxianus NCYC-587 ethanol concentration of 28.41 ± 1.82 g/L corresponding to a productivity of 0.39 ± 0.02 g/(L h) was detected for S. cerevisiae Y-2034 after 72 hours of SSF. When the initial pH increased to 6.0, the glucose concentration of enzymatic hydrolysis produced after 12 hours was the lowest, which meant that the enzymatic hydrolysis process was restrained because the enzymatic activity was affected by high pH. The concentration of glucose was detected to be >10 g/L in the slurry after 72 hours, indicating that the fermentation process was blocked. At initial pH of 6.0, the maximum ethanol concentration for K. marxianus NCYC-587 was only 13.12 ± 0.66 g/L corresponding to a productivity of 0.18 ± 0.01 g/(L h) and for S. cerevisiae Y-2034, the maximum ethanol concentration was 18.93 ± 0.95 g/L corresponding to a productivity of 0.26 ± 0.01 g/(L h). Therefore, pH 5.2 was chosen as optimal for the SSF process with either S. cerevisiae Y-2034 or K. marxianus NCYC-587.

SSF process
Effects of temperature variation on the ethanol and sugar concentrations during the SSF process with S. cerevisiae Y-2034 were depicted in Figure 2A. After prehydrolysis for 12 hours at 50°C, the rate of hydrolysis was high, and nearly 40 g/L glucose was released. S. cerevisiae Y-2034 inocula were added to the hydrolysates, and system temperatures were adjusted to 30, 34, 37, or 40°C. As for the former three temperature conditions, the fermentation rates could be faster than the rate of enzymatic hydrolysis. The glucose concentration sharply decreased as glucose was converted to ethanol. The glucose build-up was small and vanished after 36 hours. With the process temperature rising from 30 to 37°C, the highest ethanol concentration increased from 28.41 ± 1.82 to 31.25 ± 1.56 g/L after 72 hours of the SSF process. At 37°C, due to the higher temperature stress on yeast, it needed more time to adapt to the environment and yield the lowest ethanol concentration between hours 12 and 24. In contrast, the increase in temperature enhanced the enzymatic activity, which may have promoted enzymatic hydrolysis and eventually yielded the highest ethanol concentration. Therefore, 37°C has been chosen as the optimal temperature of the SSF process with S. cerevisiae Y-2034. Temperature at 40°C has been tested during the fermentation of S. cerevisiae Y-2034 and no ethanol was yielded owing to the limited thermotolerance of this strain. Effects of temperature variation on the ethanol and glucose concentrations during the SSF process with K. marxianus NCYC-587 are presented in Figure 2B. After prehydrolysis for 12 hours at 50°C, K. marxianus NCYC-587 inocula were added to hydrolysates, and system temperatures were adjusted to 37, 40, 42, or 45°C, respectively. It was found that ethanol concentrations of the SSF processes showed no big difference between 37 and 40°C, which yielded the highest concentration of approximately 29.20 ± 1.46 g/L at 96 hours. According to our previous experiment, K. marxianus shows an excellent ethanol yield when utilising pure glucose at 50 g/L as the sole carbon source at 42°C. Incontrast, with the process temperature rising from 37 to 42°C, the remaining glucose in three fermenting liquors increased from 0 to 5.58 ± 0.28 g/L, which meant that the fermentation was the rate-limiting step in this SSF process. This result might be related to the death of yeast under high temperature. 26 In addition, 45°C was tested as fermentation temperature of K. marxianus and yielded no ethanol owing to the limited thermotolerance of this strain.
Moreover, high-concentration ethanol as the endproduct severely blocked the glucose transport and metabolic system and thus suppressed the downstream metabolic pathway. 28 The ethanol tolerance of strains should be taken into consideration to reach high-solids loading and ethanol concentration in fed-batch SSF experiments. The influence of the initial ethanol concentration on the growth of S. cerevisiae Y-2034 and K. marxianus NCYC-587 was studied next. The cell growth was remarkably decreased with the exogenous ethanol concentration increasing, especially for K. marxianus NCYC-587, whose tolerable ethanol concentration was below 31.57 g/L, while the cell growth of S. cerevisiae Y-2034 could still proceed at the ethanol concentration of over 63.15 g/L. According to all the above experiments, the performance of S. cerevisiae Y-2034 was superior to that of K. marxianus NCYC-587, and thus S. cerevisiae Y-2034 was chosen as the fermenting strain for the subsequent fed-batch SSF experiments. The effect of enzyme addition mode on the ethanol production was presented in Figure 3. For mode "A", all the necessary enzymes were dosed at the beginning. This approach accelerated the enzymatic hydrolysis of SCB owing to the existence of more cellulase. The amount of insoluble solids and the viscosity of slurry decreased.
Mixing and mass transfer were facilitated. Much more sugars were released in process A than in process B during the first 12 hours. After S. cerevisiae Y-2034 was introduced, more ethanol was produced in process A than process B until the final stage of 72 hours. It was demonstrated that the mode of enzyme addition "onetime dose" improved ethanol productivity better than the "batch addition" mode. Considering the result above and the simplified procedures of process A, all the cellulase was completely fed at the beginning in the subsequent experiments.

| Ethanol production at high solids loading
In the previous study, e29 an optimised fed-batch mode was determined in an enzymatic hydrolysis system. All the feeding steps can be completed within only 24 hours at the final solids loading of 33% (w/v). The highest concentrations of cellobiose, glucose, and xylose obtained by that enzymatic hydrolysis system were 9.376, 129.50, and 56.03 g/L. These preferable conditions of substrate addition, including the addition interval and added amount during fed-batch saccharification and the aforementioned optimal conditions of SSF (including pH, temperature, and the yeast strain) were referred to in order to accomplish high solids loading during SSF. The variations of ethanol and sugar concentrations during fed-batch SSF processes at solids loading 26% and 33% (w/v) are depicted in Figure 4. After prehydrolysis for 12 hours, S. cerevisiae Y-2034 was added to system "a" and to system "b". During the initial fermentation phase of 12 hours, ethanol concentration was below 10 g/L, together with low consumption of glucose. After that, significant increases of approximately 50 g/L emerged during the next 24 hours in the two groups. Because glucose, which was accumulated in the prehydrolysis stage, was running out and owing to a limited release of fresh monosaccharide, the increases in ethanol concentration became slow after 48 hours. As for group with the 26% loading (w/v), the maximum ethanol concentration of 61.07 ± 3.05 g/L was obtained at 96 hours, and nearly no further increase was observed after extension to 120 hours. As for the solids loading of 33% (w/v), the maximum ethanol concentration of 75.57 ± 2.14 g/L was obtained when the residue time was prolonged to 120 hours. With the prolongation, the ethanol concentration continued to increase, and a small quantity of glucose remained in the final liquor. This finding means that extra 24 or 48 hours may be beneficial for a further increase in the ethanol concentration. Park et al 30 utilised alkali-pretreated palm empty fruit bunches as a feedstock and obtained 62.5 g/L ethanol concentration at 30% solids loading. Li et al 31 took phosphoric acid-acetone-pretreated reed as feedstock and attained 69.3 g/L ethanol concentration at 36% solids loading. Zhang and Zhu 32 used SCB pretreated by Fenton reaction combined with NaOH as substrate, ethanol concentrations of 70.45 g/L were obtained in fed-batch mode at a solid loading of 30% (w/v). Compared with these reports at high solids loading, a higher ethanol concentration (75.57 ± 2.14 g/L) was implemented in this study. From the standpoint of economically viable ethanol production from lignocellulosic biomass on an industrial scale, these values meet the requirements, and the fed-batch SSF process performed at high-solids loading showed satisfactory performance.
To identify the effect of solids loading on ethanol production, fermentation differences among three levels of solids loading in fields of the ethanol productivity and yield were presented in Figure 5. Throughout the fed-batch SSF process, the solids loading increased, and much more time was needed to achieve maximum ethanol concentration. The biggest ethanol productivity for loadings 19% (w/v), 26% (w/v), and 33% (w/v) was achieved at 48 hours corresponding to 0.88 ± 0.04, 1.04 ± 0.05, and 1.15 ± 0.06 g/(L h). The ethanol productivity began to decrease with time. At the solids loading of 19% (w/v), the maximum ethanol concentration of 46.13 ± 2.14 g/L was obtained in 72 hours, which corresponded to ethanol productivity of 0.64 ± 0.03 g/ (L h). At the solids loading of 26% (w/v), the maximum ethanol concentration of 61.07 ± 3.05 g/L was obtained in 96 hours, which corresponded to ethanol productivity of 0.64 ± 0.03 g/(L h). At the solids loading of 33% (w/v), the maximum ethanol concentration of 75.57 ± 2.14 g/L was achieved in 120 hours, which corresponded to ethanol productivity of 0.63 ± 0.03 g/(L h). During the SSF process, ethanol yields for the three levels of solids loading all gradually decreased. At the final stage of 120 hours, ethanol yields reached 70.06 ± 3.25%, 67.78 ± 3.39%, and 66.08 ± 1.87%. No obvious differences were observed among the three levels of solids loading in fields of the final ethanol productivity and yield. It was demonstrated that this fed-batch SSF condition was effective at increasing ethanol concentration with a small decrease in productivity (from 0.64 ± 0.03 g/[L h] to 0.63 ± 0.03 g/[L h]) and yield (from 70.06 ± 3.25% to 66.08 ± 1.87%).

| CONCLUSIONS
Compared to K. marxianus NCYC-587, S. cerevisiae Y-2034 had higher fermenting ability and ethanol tolerance, which was beneficial to be applied in high concentration ethanol production with SSF process. Alkali pretreatment was identified as an efficient method to increase SCB digestibility at high-solids loading. Throughout the fed-batch process, solids loading could be efficiently increased, and the maximum ethanol concentration increased remarkably to 75.57 g/L. It can be concluded that a combination of highsolids NaOH pretreatment and a fed-batch SSF process was effective at achieving a high ethanol concentration as well as high ethanol productivity. This study could provide useful information for ethanol production at high solids loading from lignocellulosic materials of the theoretical yield.