Yield-determining components in high-solid integrated first and second generation bioethanol production from cassava residues, furfual residues and corn

Abstract

Industrial wastes that are well pre-treated during related upstream industrial processes, such as cassava residuals and furfural residuals, are becoming promising feedstocks for producing biofuels. To better use these wastes, an approach must be flexible in their use during biologically based ethanol production. In the current study, three scenarios, including the cellulosic ethanol process, the starchy ethanol process, and the integration of 1 G and 2 G ethanol, were compared for ethanol production from CR, FR, and corn. Compared with the starch ethanol process, the cellulosic ethanol process and integration of 1 G and 2 G ethanol produced ethanol at a higher ethanol concentration and yield. Of the three processes, the integrated process utilized over 75% of polysaccharides when using multiple materials as feedstocks. Protein, cellulose, and starch were yield-determining components in the high-solid integration process. Protein provided important nutrients for yeast cells. The influences of cellulose and starch were associated with product inhibition of cellulases and viscosity. Lignin had a low or negligible influence. For an integration process at a 20% (w/w) substrate loading, the optimum concentrations of protein, cellulose, and starch included >1%, from 4% to 5%, and from 8% to 11%, respectively. At the optimum concentrations, the integration process obtained a final ethanol concentration of about 70 g L⁻¹ and an ethanol yield of about 80%.

---

Introduction

Over the past few decades, as a potential substitution for traditional fossil fuels, bioethanol has been attracting worldwide attention due to its potential sustainability and environment-friendly superiority. The superiority of bioethanol varies with the types of raw biomass. First generation bioethanol (1 G ethanol) is produced from sugar or food crops, and has been industrialized due to the high accessibility to enzymes and excellent fermentability of these substrates. Considering the increased land use for producing 1 G ethanol instead of food and animal feed has resulted in a growing ethical concern for food security, which slows down the expansion of 1 G ethanol production.¹ As an alternative to food crops, lignocellulosic biomass is being developed to produce second generation bioethanol (2 G ethanol). 2 G ethanol is supposed to reduce annual greenhouse gas (GHG) emissions by displacing fossil fuel. However, in order to produce more 2 G ethanol, more land in undisturbed ecosystems needs to be converted to biofuel production, which eventually increases GHG emissions. This is because converting native habitats to cropland rapidly releases CO₂ by burning or microbial decomposition of organic carbon stored in plant biomass and soils. Studies suggest that whether 2 G ethanol could offer carbon savings compared to gasoline production depends on the sources of lignocelluloses.^2,3 Therefore, the development of other bioenergy systems, such as waste biomass or biomass grown on degraded and abandoned agricultural lands, provides feedstocks for 2 G ethanol with less negative impacts on soil organic carbon, GHG emissions, soil erosion, food security and biodiversity than from food crops or from the removal of corn residue.^3,4

Ethanol production from cellulosic wastes, especially industrial wastes, could reduce environment pollution, and it is a way of reusing these wastes to produce a valuable product. Every year, furfural industries in China produce more than 20 million tons of solid wastes, called furfural residue (FR). During furfural production, hemicelluloses in cobs are converted into furfural at acidic conditions and high temperatures (170–185 °C), and the cellulose with lignin are left over to form FR. Because FRs are rich in cellulose (about 50%), FR has been investigated as feedstock for 2 G ethanol.⁵ Cassava residue (CR) is another solid waste and the starch industry in China annually produces about 30 000 tons of CR. During the extraction of starch from cassava roots, half of starch, which is entrapped in the matrix of cellulose and hemicelluloses, goes unextracted and is left over to form CR. Several studies have illustrated that the cassava-based bioethanol has high accessibility and intense energy, indicating that the cellulo-starch waste may be a desired feedstock for 2 G ethanol.^6,7 By using different enzymes, different ethanol production processes are further investigated to produce ethanol from CR and FR, including Simultaneous Saccharification and Fermentation (SSF) and Separate Hydrolysis and Fermentation (SHF).^8,9 Like other lignocellulosic materials, in SSF of FR, it is difficult to achieve good ethanol yields above substrate loading of around 10%, and the final ethanol concentration is limited to be below 4%.¹⁰ The trapped starch in CR has a poor accessibility to amylolytic enzymes which results in a low ethanol concentration and ethanol yield in SHF. Researchers combined different hydrolysis technologies to enhance the release of glucose in CR, including amylolytic enzyme hydrolysis, acid hydrolysis and cellulolytic enzyme hydrolysis.^8,9 However, more research should be devoted to enhancing ethanol production from FR and CR, especially increasing the final ethanol concentration.

Co-utilization of multiple materials by combining different technologies could improve the ethanol concentration during ethanol production from FR without decreasing ethanol yield. An obstacle limiting the final ethanol concentration in SSF of FR is the decreased hydrolysis rates of a cellulose molecule as the “easier” parts are hydrolyzed over time. Pretreatment such as strong phosphoric acid pretreatment is needed to break the crystalline structure, achieving a high cellulose conversion and hydrolysis rate.¹¹ Another obstacle is lignin accumulation that takes place as the ratio of cellulose decrease with the hydrolysis of celluloses. Lignin accumulation reduced the possibility of increasing FR substrate loading in SSF and made more cellulolytic enzymes “inactive”, thus limiting the final ethanol concentration.¹² To partially eliminate the negative effect of lignin accumulation, alkaline peroxide delignification was investigated to remove lignin in FR.¹³ The high consumption of chemicals likely offsets the benefits of delignification.¹³ Given the concern, an alternative approach of delignification is by co-utilizing FR and corn, as shown in our previous study.¹² Co-utilization of FR and corn by combining 1 G and 2 G ethanol technologies increased ethanol concentration in fermentation slurry to over 70 g L⁻¹ with a high ethanol yield (over 80%), helping decrease the energy consumption of downstream distillation which is the dominant unit operation for separating ethanol from fermentation slurry.^12,14 Other works demonstrated the economic feasibility of the integration process, and its flexibility in realizing the production of different downstream end products based on the market demands.^15–17 Few studies are done on improving ethanol production from CR by the integration process.

The integration process likely improves ethanol production from CR because cellulolytic enzymes could increase the accessibility of the tapped starch in CR to amylolytic enzymes by hydrolyzing cellulose. Much work is still needed to figure out which part CR should be considered as, cellulosic materials or starchy materials, in the integration process. Because CR is rich in both starch and cellulose, it can be regarded as starchy (cellulosic) materials and mixed with other cellulosic (starchy) materials in the integration process. Compared to corn and wheat meal, CR has lower starch content (50% vs. 75%) and lower protein content (4% vs. 7%). On one hand, it might be anticipated that the low protein content and starch content would prove problematic in the integration process that uses CR as starchy materials. On another hand, partial saccharification instead of complete saccharification was always performed to hydrolyze starch in starchy materials to prevent end-product inhibition of the enzymes and osmotic stress to the yeast cells. Therefore, Co-using CR as cellulosic materials may result in a high glucose concentration at the beginning of the integration process.

Factors that influence the production of 2 G/1 G ethanol have already been well evaluated, but few studies have focused on factors determining the yield of the high-solid integration process. Osmotic stress affects the yeast cell when the glucose in the solution is over 150 g L⁻¹, which limits the substrate loading as well as the saccharification degrees in 1 G ethanol.¹⁸ Kristensen et al., investigated the yield-determining factors in the high-solids enzymatic hydrolysis of lignocelluloses, and found that the inhibition of cellulose adsorption to cellulose at increasing solids concentration instead of lignin content was causing the decrease in yield.¹⁹ Although we previously investigated the effect of process parameters such as material ratio and carbon : nitrogen ratio on the integration process,²⁰ the relative contributions of chemical components (including cellulose, starch, protein and lignin) at different concentrations to ethanol concentration (ethanol yield) in the integration process is still elusive.

This paper will compare different enzymes for releasing glucose from CR, and compare the related ethanol processes using these enzymes, including cellulosic ethanol process (scenario A), starch ethanol process (scenario B), and the integration of 1 G and 2 G ethanol (scenario C). We tried to figure out which part CR should be considered as in the integration process, and what are yield-determined chemical components of the high-solid integration process. To understand the contributions of protein and starch at different concentrations to ethanol concentration (ethanol yield) in the integration process, corn is also used in the current study because corn contains a high content of protein and starch. Results here provide flexible options for the efficient utilization of CR and FR, and show how to optimize the integration of 1 G and 2 G ethanol that uses ternary or more mixtures.

Materials and methods

Double enzyme hydrolysis and cellulolytic enzyme hydrolysis were compared for biodegradation of cassava residue (CR), furfural residue (FR), and/or corn. Three scenarios, including SSF (scenario A), starchy ethanol process (scenario B), and the integration process (scenario C), were compared for ethanol production from CR, FR, and corn.

Raw materials

CR was kindly provided by GuangXi Key Laboratory of Chemistry and Engineering of Forest Products (Nanning, China). CR was milled and the fraction that passes through the 40 mesh screen was collected as experiment samples. The average contents of starch, cellulose, lignin and protein in CR samples were 47.2%, 25.0%, 6.75% and 3.72% (dry weight basis), respectively. Partial starch in CR (33.3% among 47.2% of starch) could be hydrolysed by cellulolytic enzymes. Raw FR was kindly provided by Chunlei furfural Company (Xingtai, China). In raw FR, there were inhibitors, such as inorganic acid, furfural, and 5-hydroxymethylfurfural (5-HMF). Before being used, raw FR was water-rinsed until a neutral pH to remove these inhibitors, and dried at 60 °C for 12 hours. The washed and dried FR with 7% moisture was then milled, and the fraction passing through the 40 mesh screen was collected as experiment samples. The average contents of glucan, lignin, and ash in FR samples were 48.17%, 43.29% and 6.42% (dry weight basis), respectively. Corn kernels were provided by COFCO Corporation (Beijing, China), and comprised of 75.2% starch, 11.9% non-starch glucan, and 7.46% protein. All other reagents and chemicals used were of analytical grade unless otherwise noted.

Enzymes

Amylolytic enzymes used for double enzyme hydrolysis were thermostable Alpha-amylase (4 KU g⁻¹) and glucoamylase (100 KU g⁻¹) (AoboxingUniverseen Bio-Tech Company Ltd, Beijing, China). The loading of thermostable Alpha-amylase was 150 U per gram substrate and the loading of glucoamylase was 20 U per gram substrate. Cellulolytic enzymes were cellulase (Celluclast1.5L; 75 filter paper units per mL) and β-glucosidase (Novozyme 188; 174 IU mL⁻¹) (both Novozymes A/S, Bagsvaerd, Denmark). The loading of celluclast1.5L for CR, corn and FR were 7.5 FPU per g cellulose, 10 FPU per g cellulose, and 15 FPU per g cellulose, respectively. The β-glucosidase loading was 8.5 IU per g cellulose for CR, 11.3 IU per g cellulose for corn, and 15 IU per g cellulose for FR, respectively.

Double enzyme hydrolysis

Double enzyme hydrolysis was performed in a 500 ml flask with a total weight of 350 g and 20% or 15% dry matter. CR, corn, and water were loaded to the flask to reach the set concentration, following by adding a certain amount of Alpha-amylase. CR and corn were liquefied by Alpha-amylase at 85 °C for 2 hours in a water bath. Subsequently, glucoamylase (20 U per gram substrate) was added to the flask and saccharification was performed at 60 °C for 25 hours (pH 4.0). Throughout the duration of saccharification, duplicate samples with a volume of about 0.5 mL were withdrawn at the set time points. After being withdrawn, the samples were centrifuged at 10 000g for 5 min. The supernatant liquor was filtered (0.22 μm pore), and the filtrate was diluted by about 200 times and stored at −18 °C awaiting glucose analysis.

Cellulolytic enzyme hydrolysis

Cellulolytic enzyme hydrolysis of CR with (without) FR was performed in a 100 mL conical flask with a total working weight of 60 gram. The concentrations ranged from 1% to 20% for CR, and from 0% to 7% for FR. After CR, FR and sodium citrate buffer (50 mM, pH 4.8) were loaded to the conical flask to reach the set concentrations, cellulolytic enzymes at set loading, including celluclast 1.5L and β-glucosidase, were added to the flask. Afterward, the flask was put in a shaking air-bath at 120 rpm and 50 °C to start hydrolysis. Throughout 24 hours of hydrolysis, duplicate samples with a volume of about 0.5 mL were withdrawn at 4, 6, 8, 12 and 24 hours. After being withdrawn, the samples were centrifuged at 10 000g for 5 min. The supernatant liquor was filtered (0.22 μm pore), and the filtrate was diluted by 10–200 times and stored at −18 °C awaiting glucose analysis.

SSF (scenario A)

In scenario A, SSF was performed under nonsterile conditions in a 100 ml conical flask. Before being loaded to the flask, FR, CR and fermentation medium were separately sterilized (121 °C for 20 minutes). CR, FR, and medium (pH 5.5) were then added to the conical flask to reach the set concentrations and a total weight of 60 g. The concentrations ranged from 13% to 20% for CR, and from 0% to 7% for FR. Afterward, cellulolytic enzymes and yeast (3.3 g dry matter per liter) were added to the flask. Subsequently, the flask was sealed by a loop trap containing sterile glycerol. Immediately, the sealed flask was put in a shaking air-bath at 120 rpm and 38 °C for 120 hours of SSF. During SSF, sampling conditions were the same as mentioned above in scenario A. Most of SSFs used water as fermentation medium. In one experiment, SSF used a fully supplemented medium (yeast extract 1 g L⁻¹, (NH₄)₂HPO₄ 0.05 g L⁻¹, MgSO₄ 7H₂O 0.05 g L⁻¹) to investigate the potential of CR as nitrogen and mineral sources.

Starchy ethanol process (scenario B)

In scenario B, CR with/without corn was subjected to double enzyme hydrolysis as described above. After the pH of saccharification liquid was adjusted to 5.5, a 60 g of cooled saccharification liquid was added to a 100 mL conical flask under non-sterile conditions. Subsequently, the activated yeast with an initial concentration of 3.3 g dry matter per liter was added to the flask. Afterward, the conical flask was sealed and put in a shaking air-bath at 120 rpm and 38 °C for 120 hours of ethanol fermentation. Duplicate samples with a volume of about 0.5 mL were withdrawn at the set time points. After withdrawn, the samples were centrifuged at 10 000g for 5 min, and the supernatant liquor was filtered (0.22 μm pore). The filtrate was diluted by 100 times and stored at −18 °C awaiting carbohydrate and ethanol analysis.

The integration process (scenario C)

The integration process was performed in a 100 mL conical flask equipped the same loop trap. CR and corn were firstly subjected to double enzyme hydrolysis as mentioned above. After the pH of saccharification liquid was adjusted to 5.5, a certain amount of saccharification liquid (Table 1, S1†) was loaded to a 100 mL conical flask under non-sterile conditions. Subsequently, sterile FR and water were added to the same conical flask to reach set concentrations (Tables 1 and S1†) and a total weight of 60 g. Afterward, cellulolytic enzymes and yeast (3.3 g dry matter per liter) were added to the flask. The flask was then sealed and placed in a shaking air-bath (120 rpm, 38 °C) to start fermentation. During 120 hours of fermentation, sampling conditions were the same as scenario A. To investigate the effect of protein concentration on ethanol yield at relatively low substrate loading of 12%, FR loading ranged from 2% to 10% (2%, 4%, 8% and 10%) while CR loading ranged from 10% to 2% (10%, 8%, 4%, 2%).

Microorganism

The microorganism for fermentation was S. cerevisiae in the form of dry yeast (Angel Yeast Company Ltd, Yichang, China). Before being used, the yeast was activated in 2% glucose solution at 36 °C for 15 minutes, then at 34 °C for 1 hour.

Analysis method

Glucan and acid-insoluble lignin of CR, FR and corn were analyzed according to the National Renewable Energy Laboratory (NREL) structural carbohydrates and lignin analysis procedure for biomass.²¹ Starch in CR and corn was completely converted to glucose by double enzymes method, and the glucose was then quantified by HPLC to determine the starch content.²² The nitrogen content of CR and corn was quantified by Kjeldahl nitrogen determination.²³ Then protein content was obtained by multiplying the nitrogen content by the universal factor of 6.25. The glucose concentration in the filtrate was detected by HPLC (Waters 2695e, USA) equipped with an AminexHPX-87P (300 × 7.8 mm, Bio-Rad, USA) at 85 °C and a refractive index detection detector at 30 °C. The eluent was ultrapure water at a flow rate of 0.6 mL min⁻¹. The ethanol concentration in the filtrate was determined in the same HPLC system, but an Aminex HPX-87H (300 × 7.8 mm, Bio-Rad, USA) was used to separate ethanol. The eluent used for ethanol separation was sulfuric acid at a flow rate of 0.6 mL min⁻¹. For HPLC test, the injection volumes were 10 μL.

Calculations of the amount of glucose, ethanol yield and polysaccharide utilization ratio could be found in ESI.† Experiments were performed in triplicate. Mean values as well standard deviations from the mean were presented. SPSS Statistics 17.0, Sufer 8.0 and matlab 7.0.1 were used to analyze the data.

Results

Double enzyme hydrolysis

Double enzyme hydrolysis, including liquefaction and subsequent saccharification, is normally used to hydrolyze starch to glucose. During this process, saccharification time plays an important role in the degree of releasing glucose monomers in starch. Erdei et al., investigated the starch hydrolysis profiles of wheat meal, and found that starch was completely hydrolyzed to glucose monomers (referred to complete saccharification) when saccharification time over 24 h.¹⁸ Erdei et al., obtained about 310 g L⁻¹ glucose after complet saccharification of wheat meal. Complete saccharification of 20% corn gave an initial glucose concentration of about 183 g L⁻¹ at 25 h (Fig. 1A). The main problem of using complete saccharification for further ethanol fermentation was a high initial concentration of glucose (>150 g L⁻¹) which led to a high osmotic stress to the yeast. Given this concern, partial saccharification was suggested to hydrolyze starch in corn and wheat meal.^12,18 Compared to corn and wheat meal, complete saccharification of 20% CR obtained a lower glucose concentration of 117 g L⁻¹ at 25 h, corresponding to a glucose yield of 99.5% (Fig. 1A). This is because CR has lower starch content than corn and wheat meal. The relatively low glucose concentration indicated that complete saccharification of CR can be used in starch ethanol process. When the mixtures of CR and corn with varying ratio were saccharified at a total substrate loading of 20%, the glucose concentration at 25 h ranged from 117 g L⁻¹ to 183 g L⁻¹ (Fig. 1A). The glucose concentration linearly decreased with the amount of CR fraction (y = 182 − 340x, R² = 0.99; y is the glucose concentration, x is the loading of CR fraction (w/w)). To keep the glucose concentration in saccharification liquid below 150 g L⁻¹, the loading of CR should be over 10% (calculated according to the equation above) otherwise partial saccharification should be performed.

Fig. 1 Comparison of double enzyme hydrolysis (A) and cellulolytic enzyme hydrolysis (B). Open symbols with double enzyme hydrolysis; closed symbols with cellulolytic enzyme hydrolysis. In most of the cases, no error bars are shown because the differences between the replicate measurements were too small to show on this scale.

Cellulolytic enzyme hydrolysis

In cellulolytic enzyme hydrolysis, the glucose concentration at 24 h increased from 6.5 g L⁻¹ to 92 g L⁻¹ with increasing the CR loading from 1% to 20% (Fig. 1B). However, the glucose yield at 24 h decreased from 100% to 55%. Like other lignocelluloses, this decreasing trend is supposed to be due to the strong inhibitory effect of high concentration glucose and cellobiose on cellulolytic enzymes.³⁶ Assuming that cellulose in CR is completely hydrolysed to glucose, the glucose concentration in cellulolytic enzyme hydrolysis of 1% CR is calculated to be 2.8 g L⁻¹ based on the cellulose content of CR (25%). In practice, the glucose concentration reached 6.5 g L⁻¹, higher than 2.8 g L⁻¹. This was because that Celluclast1.5L and β-glucosidase were also capable of hydrolyzing starch to glucose as mentioned by Divya Nair et al.³⁷ Our results showed that cellulolytic enzymes hydrolyzed all CR cellulose (25.0% of starting material) and about 70% CR starch (33.3% of starting material) to glucose in the case of 1% CR.

Comparison of double enzyme hydrolysis and cellulolytic enzyme hydrolysis

Nair et al., compared the efficiency of different enzymes by the index of “the amount of glucose released from CSFR (CR) by a certain enzyme at set time point”.³⁷ They found that the amount of glucose released by combining Multifect XL, Optimash XL and Accellerase at 24 h ranged from 133 to 323.5 (g per 1 kg CSFR). This study employed this index to compare double enzyme hydrolysis and cellulolytic enzyme hydrolysis. In complete saccharification of 20% CR, the amount of glucose released was 467 (g per 1 kg CR) at 25 h. The amount of glucose released by double enzyme hydrolysis did not change with CR loading and the value was 466 (g per 1 kg CR) in complete saccharification of 15% CR. Unlike double enzyme hydrolysis, the amount of glucose released by cellulolytic enzyme hydrolysis at 24 h increased from 356 (g per 1 kg CR) to 674 (g per 1 kg CR) with decreasing CR loading from 20% to 1%. This is because decreasing CR loading results in an increased glucose yield as mentioned above (Fig. 1B). The results indicated that at a high CR loading of 20%, double enzyme hydrolysis produced glucose faster than cellulolytic enzyme hydrolysis (467 g per 1 kg CR vs. 356 g per 1 kg CR). However, if product inhibition was reduced by decreasing CR loading or using SSF, cellulolytic enzyme hydrolysis released more glucose from CR than double enzyme hydrolysis (674 g per 1 kg CR vs. 467 g per 1 kg CR).

SSF

After comparing double enzyme hydrolysis and cellulolytic enzyme hydrolysis, we evaluated the performance of these hydrolysis methods in different scenarios of ethanol production. In scenario A and B, cellulolytic enzyme hydrolysis and double enzyme hydrolysis were used, respectively. The combination of cellulolytic enzyme hydrolysis and double enzyme hydrolysis were used in scenario C. In scenario A, SSF of 20% CR with fully supplemented medium was compared to that with water. A 77.6 g L⁻¹ of ethanol concentration in SSF with water at 120 h was on par with an 80.8 g L⁻¹ of ethanol concentration in that with the fully supplemented medium (Fig. 2A). Our earlier work suggested that the addition of corn hydrolysates lowered the amount of nitrogen source needed for cellulosic ethanol.¹² The current work showed that CR hydrolysates were similar to corn hydrolysates and the mixed fermentations could also be done without the addition of minerals. CR is, therefore, desired material for cellulosic ethanol. It is interesting to see if nutrients in CR are enough to support SSF that co-uses CR and FR. When we added some FR in scenario A and keep the total substrate loading at 20%, both ethanol concentration and ethanol yield at 120 h decreased with increasing the amount of FR. Increasing the amount of FR from 0% to 7% decreased the ethanol yield from 92.1% to 73.9% (Fig. 2B), and ethanol concentration from 77.6 g L⁻¹ to 58.7 g L⁻¹ (Fig. 2A). The increased lignin with increasing of the amount of FR is likely the major reason for the decreasing trend of ethanol yield. Work with various kinds of lignocellulosic materials has identified the negative effect of lignin content on both of enzymatic conversion and SSF.^13,38 Another reason is supposed as the addition of FR makes the concentration of protein and polysaccharides diluted. The result indicated that the low protein content of CR was problematic in the integration process that uses CR as starchy materials. This result also indicated that less consumption of chemicals as nutrients is not a stability characteristic of scenario C, depending on the protein structure and protein content of starch materials used.

Fig. 2 Concentration (A) and yield (B) of ethanol during simultaneous saccharification and fermentation (SSF) of cassava residue (CR) and furfural residue (FR). Closed symbols contained fully supplemented medium; open symbols contained water.

Starchy ethanol process (scenario B)

A stuck fermentation took place in scenario B of 20% CR and liquefaction did not happen even after 72 h. In this case, there was about 50 g L⁻¹ of cellulose in the system. The cellulose cannot be hydrolysed by amylolytic enzymes and would lead to high viscosity, thus resulting in stuck fermentation.¹⁰ Amylolytic enzymes hydrolysis could be improved by combining other hydrolysis technology. Akaracharanya et al., combined saccharification by amylolytic enzymes and dilute acid hydrolysis to enhance the release of glucose in CR. Fermentation of these hydrolysates obtained 0.429 g ethanol per g CR (ethanol concentration is less than 25 g L⁻¹), equal to an ethanol yield of 86%.⁹ Wang et al., found that cellulase, but not pectinase, plays a major role in viscosity reduction in enzymatic hydrolysis of during enzymatic hydrolysis of high-gravity SPRs which is a material containing cellulose and starch.²⁴ When the loading of CR was decreased to 15%, liquefaction happened in 24 h and the ethanol concentration at 72 h reached 32.9 g L⁻¹ (Fig. 3A), corresponding to an ethanol yield of 90.7% (Fig. 3B). Because cellulolytic enzymes could decrease the viscosity attributed to cellulose, adding cellulolytic enzymes in scenario B, which is referred to scenario C, is another way of improving the hydrolysis of CR. Compared to acid, there is no need for detoxification of hydrolysate if cellulolytic enzymes were used.

Fig. 3 Comparison of scenario B (closed symbols) and scenario C (open symbols) of CR and corn. (A) The concentration profiles of ethanol (solid lines) and glucose (dashed lines); (B) the ethanol yield at 72 h.

Integration process of 1 G and 2 G ethanol

Although using the mixture of amylolytic enzymes and cellulolytic enzymes is a way of improving the performance of amylolytic enzymes, it showed a low efficiency. Zhu et al., used the combination of amylolytic enzymes and cellulolytic enzymes in SHF and SSF of 20% CR, and obtained the ethanol concentration of 23.5 and 34.7 g L⁻¹, respectively.⁸ In this study, a sequential hydrolysis was used for ethanol production in scenario C. Scenario C was firstly compared to scenario B. In the early stage of fermentation (before 20 h), the fermentation rate of scenario C of 15% CR was higher than that of scenario B of 15% CR (Fig. 3A). Scenario C obtained lower ethanol yield than scenario B, but higher polysaccharide utilization ratio (Table 1). Scenario C was then compared to scenario A. Scenario A of 20% CR had a higher ethanol yield and polysaccharide utilization ratio compared to scenario C of 20% CR (Table 1). Of these three scenarios, scenario A is the most efficient in using single CR, suggesting that it is favourable to use CR as cellulosic materials. More basic experiments of scenario C were then conducted by using CR as cellulosic materials and using corn kernel as starchy materials. Fig. 3A showed the effect of the proportion of corn on scenario C. Ethanol concentration increased with increasing the proportion of corn fraction. However, ethanol yield firstly increased with increasing the proportion of corn fraction, but further increasing corn proportion decreased ethanol yield (Fig. 3B). Overall, using single CR, scenarios A was prone to obtain the highest ethanol yield among the three scenarios. However, using multiple materials as the substrate, scenario C gave the highest polysaccharide utilization ratio (Table 1). This study obtained a higher concentration of ethanol than 34.7 g L⁻¹, suggesting that sequential hydrolysis is more efficient than single hydrolysis using enzyme cocktail.

Table 1 Comparison of ethanol production after 72 h of fermentation in different processes

	Fraction concentration for each basic material (%, w/w)			Ethanol concentration (g L^−1)	Polysaccharide utilization ratio based on total sugars (%)	Ethanol yield based on total available sugars (%)
	CR	FR	Corn
Scenario A	20	0	0	71.4 ± 2.3	76.2 ± 2.4	86.4 ± 2.7
	16.5	3.5	0	65.3 ± 0.1	64.4 ± 0.1	79.3 ± 0.1
	13	7	0	57.5 ± 0.6	60.8 ± 0.6	72.2 ± 0.7
Scenario B	15	0	0	32.7 ± 0.2	59.3 ± 0.4	90.0 ± 0.5
Scenario C	15	0	0	37.6 ± 0.5	64.8 ± 0.1	64.8 ± 0.1
	20	0	0	67.7 ± 1.9	72.2 ± 2.0	72.2 ± 2.0
	15	0	5	72.2 ± 1.2	75.7 ± 1.2	75.7 ± 1.2
	10	0	10	74.6 ± 0.3	75.2 ± 0.3	75.2 ± 0.3
	3	4	13	71.3 ± 0.1	83.9 ± 0.1	83.9 ± 0.1
	2	5	13	66.0 ± 2.0	79.6 ± 2.4	79.6 ± 2.4
	1.1	6	12.9	66.0 ± 0.4	81.8 ± 0.5	81.8 ± 0.5

---

The yield-determining components in scenario C

We further investigated the yield-determining components in scenario C. Due to the absence of hemicelluloses in FR, CR, and corn kernel, there were four significant polymers in scenario C, including cellulose, starch, lignin, and protein. The influences of these polymers on ethanol yield highly depended on the substrate loading. At a low substrate loading of 12%, a strong correlation (0.97/0.77) was evident between lignin/protein concentration and ethanol yield (P < 0.05/P < 0.01). Fig. 4 showed the linear relationship between protein concentration and ethanol yield at 12% substrate loading. However, these correlations were not repeated when the total substrate loading was increased from 12% to 20% (Fig. 4). Although lignin was regarded as a major obstacle restricting the enzymatic hydrolysis of lignocelluloses, lignin content was not responsible for the decrease in yields of high-solid enzymatic hydrolysis of lignocelluloses, according to Kristensen'work.¹⁹ This conclusion was applicable to high-solid scenario C. For scenario C at 20% substrate loading, the Pearson's correlation coefficient between ethanol yield and protein obtained a higher value (0.396) followed by starch (0.249) and cellulose (−0.278) and, to a lesser extent, lignin (0.104). The results suggested that protein, cellulose, and starch were yield-determining components in scenario C.

Fig. 4 Modeling the relationship between ethanol yield (72 h) and protein concentration in scenario C. Closed symbols with substrate loading of 12%; open symbols with substrate loading of 20%.

Fig. 4 suggested that the optimized concentration of protein was over 1%. This figure also indicated that increasing protein concentration did not distinctly improve ethanol yield when it over 1% (Fig. 4). The results further proved that less consumption of chemicals as nutrients is not a stability characteristic of scenario C, depending on the protein structure and protein content of starch materials used. After we got the optimum concentration of protein, the concentration of cellulose and starch was optimized as follow.

The concentration ranges of cellulose and starch

Due to the limited material number, there should be limited concentration ranges of cellulose and starch in scenario C. To better assess the effect of cellulose concentration and starch concentration, we firstly investigated the available concentration ranges of cellulose and starch. The available concentration ranges depend on the substrate loading and material number. By keeping the substrate loading constant, the concentration range of cellulose and starch in scenario C that uses sole material, two materials or three materials was described by a point (Fig. 5A and B), line (Fig. 5C–E) or plane (Fig. 5F) in a rectangular coordinate system. For example, using CR and FR at 20% of total substrate loading, the concentration range of cellulose and starch is the red line in Fig. 5C . In the red line, starch concentration (C_S) got a maximum value of 9.44 when CR loading and FR loading were 20% and 0%, respectively; cellulose concentration (C_C) got a maximum value of 6.61 when CR loading and FR loading were 13% and 7%, respectively. By changing total substrate loading from 20% to 10%, the concentration range of cellulose and starch changed to the black line (Fig. 5C). Of these material groups, the group using three materials could obtain the largest available concentration ranges of cellulose and starch (Fig. 5F).

Fig. 5 The concentration range of cellulose and starch in scenario C. The detailed description of this figure can be found in the ESI.†

Influence of the concentrations of cellulose and starch

After we got the available concentration ranges of cellulose and starch in Fig. 5F, more experiments (Table 1S†) were conducted to model the effect of varying the concentration of cellulose and starch in the available range on ethanol concentration and yield. It should be noted that ethanol yield is equal to polysaccharide utilization ratio in scenario C. The effect of starch and cellulose concentration on ethanol yield appeared to be fairly complex (Fig. 6A) while that of ethanol concentration was relatively simple (Fig. 6B). According to Fig. 6A and B, the optimum cellulose concentration includes from about 4% to 5% and the optimum starch concentration includes from 8% to 11%. Under these optimum conditions, scenario C could simultaneously obtain a final ethanol concentration of about 70 g L⁻¹ and an ethanol yield of over 80%.

Fig. 6 Three-dimensional surface area plot showing the effect of cellulose concentration and starch concentration on ethanol production. (A) In term of ethanol yield; (B) in term of ethanol concentration.

Achieving the optimum concentrations of yield-determining components

The equations showing the relationships between the ratio of each material and the concentrations of each chemical component were in supplementary information 1.2. Related to eqn (2)–(4), Fig. 7A–C presented three-dimensional surface area plots related to cellulose concentration, starch concentration and protein concentration and their two-dimensional contour graphs. As mentioned above, the optimum concentration range is >1% for protein, 4–5% for cellulose, and 8–11% for starch. According to Fig. 7A, one can get the contour lines of 1% protein and 1.4% protein (the maximum protein concentration is 1.4% in scenario C). Similarly, one can get the contour lines of 4% and 5% cellulose, and that of 8% and 11% starch. By presenting the six contour lines in a coordinate system, one could get a dark gray area in Fig. 7D. The dark gray area showed the optimum range of FR loading and CR loading. Once the FR loading and CR loading were fixed, the corn loading was calculated according to the equation: L_Corn = 20 − L_CR − L_FR.

Fig. 7 Optimizing substrate fractions to realize optimum starch concentration and cellulose concentration. (A) Cellulose concentration as a function of CR loading and FR loading; (B) starch concentration as a function of CR loading and FR loading; (C) protein concentration as a function of CR loading and FR loading; (D) the optimum range of substrate fractions closed by the contours related to optimum cellulose, starch and protein concentrations.

To verify the optimum concentrations of yield-determining components, three substrate dosages in the dark gray area in Fig. 7D, including (L_FR, L_CR, L_corn) = (4%, 3%, 13%), (L_FR, L_CR, L_corn) = (5%, 2%, 13%) and (L_FR, L_CR, L_corn) = (6%, 1.1%, 12.9%), were used in scenario C. At each substrate dosage, the protein concentration was about 1.1%. After 72 h of fermentation, scenario C using the three dosages obtained an ethanol concentration of 71 g L⁻¹, 66 g L⁻¹ and 66 g L⁻¹, respectively, corresponding to an 84%, 80%, and 82% of ethanol yield.

Discussion

As one of the yield-determining components, protein provides important nutrients for yeast cells. Compared to FR fraction, the fermentation promoters in CR fraction are nitrogen, helping meet the nutrient requirements of yeast for the growth and fermentation. Our data support the hypothesis that the lower protein content of cassava residue compared to corn would prove problematic in the integration process that uses CR as starchy materials. It was risky to decrease the proportion of CR fraction in scenario C. The use of other materials especially protein-rich materials (protein content > 7%) likely eliminates this risk. The protein fraction of non-food biomass is between 3% and 13%.²⁵ For some protein-rich materials, such as yeast and soybean meal, the protein contents of these materials reach 48.1% and 49.7%. These materials at a loading of about 2% may provide enough nitrogen for scenario C. In practice, however, treatments, such catalytic hydrolysis and protease hydrolysis, is necessary for releasing amino acid of those materials before they are used in scenario C. These treatments resulted in that these protein-rich materials were comparable to beef extract and yeast extract.^20,26,27 The main disadvantages of these protein-rich materials include the absence of abundant carbohydrates and the potentially adverse effect of non-protein components.

Cellulose is the main carbon source in scenario A while both cellulose and starch provide carbon sources in scenario C. During the three processes, the influence of cellulose is in part associated with viscosity and the nature of cellulose itself, such as the accessibility of cellulases to cellulose and its crystallinity.^11,28 As mentioned previously, the high viscosity due to ∼5% of cellulose plus other insoluble components likely led to a stuck fermentation in scenario B, which partly explained that the optimized cellulose concentration of scenario C ranged from 4% to 5%. To decrease viscosity, increase the substrate loading, and finally, improve final ethanol concentration in cellulosic ethanol, prehydrolysis and fed-batch approach have been so far well evaluated.^10,29 Scenario C could be regarded as a prehydrolysis. However, it has a different mechanism of increasing final ethanol concentration compared to the prehydrolysis of lignocelluloses. In contrast to using cellulolytic enzymes at a temperature below 60 °C to prehydrolyze cellulose in cellulosic ethanol, scenario C pre-hydrolyzed starch by amylolytic enzymes at an elevated temperature of 85 °C to mainly provide a high concentration of fermentative sugars.²⁹ Enzyme cost, product inhibitory effect and enzyme deactivation limit the economical feasibility of prehydrolysis using cellulolytic enzymes.³⁰ By contrast, amylolytic enzymes are much cheaper, more thermostable and efficient than cellulolytic enzymes. After the prehydrolysis by amylolytic enzymes, cellulolytic enzymes was added in a fed-batch way and began to break down cellulose. The addition of cellulase helps to decrease the viscosity leaded by cellulose. As shown by Wang et al., even at a low loading of 4 FPU per g CR, cellulase decreased the viscosity of hydrolysis system from about 7 × 10⁴ cp to 2 × 10⁴ cp after 2 hours of incubation.²⁴ In this study, relatively higher loading of cellulase (7.5 FPU per g cellulose) was used in scenario C, explaining its high efficiency.

Starch is the main carbon source in scenario B. Compared to scenario B, lower optimized starch concentration is observed in scenario C, 8–11% vs. about 13.5%. In scenario B, the influence of starch concentration on ethanol fermentation was associated with the osmotic stress to yeast cell. When the glucose concentration was over 150 g L⁻¹, osmotic stress affected the yeast cell.³¹ In scenario B, 13.5% of starch concentration is equal to about 150 g L⁻¹ of glucose concentration. In scenario C, the influence of starch concentration is related to not only yeast osmotic stress but also the inhibitory effect of glucoses and cellobiose on cellulolytic enzymes. The inhibitory effects of sugars on cellulolytic enzymes include product inhibition and adsorption inhibition. Xiao et al., stated that a glucose concentration ranging from 0 to 100 g L⁻¹ led to a high degree of inhibition of cellulase activity.³² The inhibitory effects of sugar on cellulolytic enzymes are responsible for the decreased optimized concentration of starch in scenario C. The potential promotion of starch concentration in scenario C is likely achieved by improving cellulase CBDs or using cellulases with a high tolerance to high concentration glucose. However, it is possible to envision both good and bad outcomes. We previously observed that high sugar concentrations (∼120 g L⁻¹) obviously promote yeast reproduction.¹² The difficulty of reusing yeast cells in lignin-rich solid residue indicated that more yeast reproduction means more loss of sugars and protein, thus offsetting the advantages of further increasing starch concentration. Meanwhile, the abundant existence of yeast cells will hinder the reuse of lignin.

It is beneficial to reuse lignin in the lignin-rich solid residue. As mentioned in a recent review about lignin processing, the presence of proteins and microorganisms likely limits suitability of lignin for direct material applications.³³ Most studies using enzymes to deconstruct plant polysaccharides focus on the lignin properties contributing to biomass recalcitrance, ease of extraction of a lignin stream, and the use of lignin as a high-value coproduct.³³ By contrast, very little is known about lignin properties associated with the separation of microorganisms and related protein from lignin-rich solid residue. Despite its apparent advantages of recovering and reusing cellulolytic enzymes in lignin-rich residues if technologies become technically available, the dominated protein likely from microorganisms and starch materials instead of enzymes (>3 g L⁻¹ vs. <1 g L⁻¹) may limit the practical operability of recycling and reusing enzymes. Recovering the protein-rich residue from the saccharified starchy materials before mixing them with cellulosic materials could avoid the problem above by reducing the amount of protein in later lignin-rich streams. This process could provide direct economic benefits, such as DDGS sold as animal feed.¹⁵ However, it may decrease the capacity of starch materials as nitrogen sources in downstream fermentation. Alternatively, another way is likely to extract lignin prior to enzyme hydrolysis and fermentation by pretreatments that specifically target lignin extraction, which can facilitate the recovery of non-catalysis protein and enable new uses for protein-rich solid residues, including animal fodder, nutrients in bioconversion processes, and commodity chemicals.

The integration process is a green technology that is capable of producing bioethanol, biologically based chemicals, and multiple products, including bioethanol and biologically derived platform chemicals, thus better balancing fuel and chemical needs with the needs of the environment.^{12,18,20,34,35}

Conclusions

Cellulolytic enzyme hydrolysis released more glucose from CR than double enzyme hydrolysis. Compared to starch ethanol process, SSF and the integration of 1 G and 2 G ethanol produced ethanol from CR at a high concentration and yield. It is favourable to use CR as cellulosic materials in the integration of 1 G and 2 G ethanol.

Of the three ethanol processes, the integration process obtained the highest utilization ratio of polysaccharides when multiple materials were used. Protein, cellulose, and starch were yield-determining components in scenario C. By contrast, the impact of lignin became insignificant. As the major factors affecting the fermentation efficiency of the high-solid integration process, the influence of starch concentration and cellulose concentration are respectively associated with product inhibition to cellulases and the viscosity. The optimization of the concentrations of yield-determining components further improved the ethanol concentration and yield of the integration process.

More work is needed to evaluate the utilization of protein rich materials, stronger microorganism, and cellulases of higher efficiency in the integration process, further improving the viability of the integration process to produce not only bioethanol but also biologically derived chemicals.

Acknowledgements

The Guangxi Key laboratory of Chemistry and Engineering of forest Products (GXFC12-07), China Ministry of Science and Technology (2014DFG32550) are gratefully acknowledged for their financial support of this research project.

Notes and references

1. H. Youngs and C. Somerville, Science, 2014, 344, 1095–1096.
2. J. Fargione, J. Hill, D. Tilman, S. Polasky and P. Hawthorne, Science, 2008, 319, 1235–1238.
3. A. J. Liska, H. S. Yang, M. Milner, S. Goddard, H. Blanco-Canqui, M. P. Pelton, X. X. Fang, H. T. Zhu and A. E. Suyker, Nat. Clim. Change, 2014, 4, 398–401.
4. J. P. Scharlemann and W. F. Laurance, Science, 2008, 319, 43–44.
5. H. Yu, Y. Xing, F. Lei, Z. Liu, Z. Liu and J. Jiang, Bioresour. Technol., 2014, 167, 46–52.
6. H. Yang, L. Chen, Z. C. Yan and H. L. Wang, Biomass Bioenergy, 2011, 35, 581–589.
7. L. V. Rasmussen, K. Rasmussen, T. Birch-Thomsen, S. B. P. Kristensen and O. Traore, Energy Policy, 2012, 41, 575–583.
8. M. Zhu, P. Li, X. Gong and J. F. Wang, Biosci., Biotechnol., Biochem., 2012, 76, 671–678.
9. A. Akaracharanya, J. Kesornsit, N. Leepipatpiboon, T. Srinorakutara, V. Kitpreechavanich and V. Tolieng, Ann. Microbiol., 2011, 61, 431–436.
10. M. Cui, Y. M. Zhang, R. L. Huang, R. X. Su, W. Qi and Z. M. He, RSC Adv., 2014, 4, 44659–44665.
11. K. Olofsson, M. Bertilsson and G. Liden, Biotechnol. Biofuels, 2008, 1, 7.
12. Y. Tang, D. Q. Zhao, C. Cristhian and J. X. Jiang, Biotechnol. Biofuels, 2011, 4, 22.
13. L. X. Bu, Y. Tang, Y. Xing, W. M. Zhang, X. H. Shang and J. X. Jiang, Biosci., Biotechnol., Biochem., 2014, 78, 1435–1443.
14. A. J. Ragauskas, C. K. Williams, B. H. Davison, G. Britovsek, J. Cairney, C. A. Eckert, W. J. Frederick, J. P. Hallett, D. J. Leak, C. L. Liotta, J. R. Mielenz, R. Murphy, R. Templer and T. Tschaplinski, Science, 2006, 311, 484–489.
15. B. Erdei, D. Hancz, M. Galbe and G. Zacchi, Biotechnol. Biofuels, 2013, 6, 169.
16. S. Macrelli, M. Galbe and O. Wallberg, Biotechnol. Biofuels, 2014, 7, 26.
17. S. Macrelli, J. Mogensen and G. Zacchi, Biotechnol. Biofuels, 2012, 5, 22.
18. B. Erdei, Z. Barta, B. Sipos, K. Reczey, M. Galbe and G. Zacchi, Biotechnol. Biofuels, 2010, 3, 16.
19. J. B. Kristensen, C. Felby and H. Jorgensen, Biotechnol. Biofuels, 2009, 2, 11.
20. Y. Tang, L. X. Bu, J. He and J. X. Jiang, Eur. Food Res. Technol., 2013, 236, 365–371.
21. Determination of structural carbohydrates and lignin in biomass, ed. A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter and D. Templeton, N. R. E. Laboratory, Golden, CO, USA, 2004.
22. B. V. Mccleary, V. Solah and T. S. Gibson, J. Cereal Sci., 1994, 20, 51–58.
23. J. Kjeldahl, Z. Anal. Chem., 1883, 22, 366–382.
24. F. Z. Wang, Y. Jiang, W. Guo, K. Niu, R. Q. Zhang, S. L. Hou, M. Y. Wang, Y. Yong, C. X. Zhu, C. J. Jia and X. Fang, Biotechnol. Biofuels, 2016, 9, 50.
25. B. E. Dale, M. S. Allen, M. Laser and L. R. Lynd, Biofuels, Bioprod. Biorefin., 2009, 3, 219–230.
26. M. T. Gao, M. Hirata, E. Toorisaka and T. Hano, Biochem. Eng. J., 2006, 28, 87–91.
27. K. Q. Chen, J. A. Li, J. F. Ma, M. Jiang, P. Wei, Z. M. Liu and H. J. Ying, Bioresour. Technol., 2011, 102, 1704–1708.
28. J. G. Hu, V. Arantes, A. Pribowo, K. Gourlay and J. N. Saddler, Energy Environ. Sci., 2014, 7, 2308–2315.
29. K. Hoyer, M. Galbe and G. Zacchi, Process Biochem., 2013, 48, 289–293.
30. K. Ohgren, J. Vehmaanpera, M. Siika-Aho, M. Galbe, L. Viikari and G. Zacchi, Enzyme Microb. Technol., 2007, 40, 607–613.
31. S. Nikolic, L. Mojovic, M. Rakin, D. Pejin and V. Nedovic, J. Chem. Technol. Biotechnol., 2009, 4, 497–503.
32. Z. Xiao, X. Zhang, D. Gregg and J. N. Saddler, Appl. Biochem. Biotechnol., 2004, 1115–1126.
33. A. J. Ragauskas, G. T. Beckham, M. J. Biddy, R. Chandra, F. Chen, M. F. Davis, B. H. Davison, R. A. Dixon, P. Gilna, M. Keller, P. Langan, A. K. Naskar, J. N. Saddler, T. J. Tschaplinski, G. A. Tuskan and C. E. Wyman, Science, 2014, 344, 1246843.
34. Y. Tang, L. Zhu, W. Zhang, X. Shang and J. Jiang, Appl. Microbiol. Biotechnol., 2013, 97, 1923–1932.
35. J. H. Clark, V. Budarin, F. E. I. Deswarte, J. J. E. Hardy, F. M. Kerton, A. J. Hunt, R. Luque, D. J. Macquarrie, K. Milkowski, A. Rodriguez, O. Samuel, S. J. Tavener, R. J. White and A. J. Wilson, Green Chem., 2006, 8, 853–860.
36. W. D. Murray, Biotechnol. Bioeng., 1987, 29, 1151–1154.
37. M. P. D. Nair, G. Padmaja and S. N. Moorthy, Biomass Bioenergy, 2011, 35, 1211–1218.
38. F. Masarin, D. B. Gurpilhares, D. C. Baffa, M. H. Barbosa, W. Carvalho, A. Ferraz and A. M. Milagres, Biotechnol. Biofuels, 2011, 4, 55.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra08036g

---