Introduction

Sugarcane represents a preferred crop for bioenergy (ethanol) production due to high biomass yields and high fermentable sugar content [1]. However, the fibrous residue (bagasse) generated after sugar juice (mainly sucrose) extraction for ethanol production may also be used to increase the ethanol yield per ton of harvested cane. However, converting sugarcane bagasse (SB) that is recalcitrant to enzymatic hydrolysis into fermentable sugars requires a costly pretreatment process to make its structural carbohydrates more accessible [2, 3]. One of the strategies of reducing pretreatment cost is to improve feedstock quality (high structural carbohydrates content and high convertibility) through crop development and selection with the view of maximizing ethanol output from both sugar juice and bagasse per unit land.

The feedstock quality of sugarcane varieties can be improved through plant breeding by classical or genetic engineering, and both of these have shown the possibility to produce sugarcane lines which are less recalcitrant to bioconversion without affecting plant performance in controlled environmental conditions [4, 5]. In order to identify sugarcane varieties with improved potential for combined ethanol production from both sugar juice and bagasse, samples of varieties in the breeding program at South Africa Sugarcane Research Institute (SASRI) were screened. Of this collection, 100 varieties from classical breeding were selected on the basis of high biomass yields, while an additional 15 varieties from precision breeding (genetic engineering) aimed at increasing soluble sugar content were also included. These 115 varieties were screened in terms of potential ethanol yields per hectare from both sugar juice and bagasse, as reported previously [6]. After the screening, the next step is pretreatment optimization to fully demonstrate the advantage of variety selection on fermentable sugar yield from the bagasse.

Dilute sulfuric acid (DSA) pretreatment represents the most widely researched technology on different types of feedstocks ranging from agricultural residues to woody and herbaceous crops [2, 7–9]. In this method, the soaked material is hold at elevated temperature for a specific period of time. Hemicellulose is hydrolyzed into the liquid fraction, leaving the solid material porous, enriched of cellulose and lignin [10]. The removal of hemicellulose weakens the carbohydrates–lignin matrix structure, thus increasing cellulose accessibility. Nevertheless, at severe conditions, pentose and hexose sugars may become destructed into non-sugar compounds [9]. Therefore, optimization of pretreatment conditions is important for efficient conversion of lignocellulose material into fermentable sugars.

Optimization of pretreatment conditions for either xylose recovery or cellulose digestibility has been actively researched [11–16]. Working on corncob, Cai et al. [17] found that the condition for the highest xylose yield was less severe compared to that for the maximum glucose yield after enzymatic saccharification. However, the condition for the highest glucose yield also resulted into high sugar degradation. Other researchers have proposed a two-step process to minimize sugar degradation [10, 18]. The first step is performed at low temperature to target xylose and the other is conducted at high temperature for high glucose yield. However, such suggestion is question of economics and the energy costs.

Another approach is to find the pretreatment conditions that could maximize the combined sugar yield (CSY) (the total pentose and hexose sugars released after the combined pretreatment and enzymatic hydrolysis), while keeping the by-products formation as low as possible. Lloyd and Wyman [19], working with the corn stover showed that the optimization of xylose yield did not lead to maximum CSY. Similarly, the maximum glucose yield from enzymatic hydrolysis also failed to release the highest CSY. The conditions that provide maximum sugar yield are therefore a compromise between those for maximum xylose recovery and those for maximum glucose yields. However, to the best of these authors' knowledge, none these research studies have considered the optimization of the CSY of bagasse coming from different sugarcane varieties.

The objective of this study was to investigate the effects of the DSA pretreatment conditions (temperature, acid concentration, and reaction time) on the pretreatment and the enzymatic hydrolysis responses of the bagasse from the six selected varieties of sugarcane and one sample was with industrial origin. The purpose was to identify varieties with reduced pretreatment requirements. To maximize CSY, the optimization was performed according to a central composite design (CCD) under response surface methodology (RSM) as a statistical method.

Materials and Methods

Raw Materials and Samples Preparation

The bagasse samples from six varieties sugarcane used in the present study were supplied by SASRI. The varieties were developed through classical and precision breeding technologies. The precision breeding varieties were developed by downregulating expression of an endogenous enzyme UDP glucose dehydrogenase as described elsewhere [20]. The feedstocks were sampled from mature sugarcane (12 months old) in an experimental field located at Mount Edgecombe (29.7000° S and 31.0333° E), KwaZulu-Natal in November 2009. The genotypes were first planted field trial in 2006. This means that the bagasse evaluated in this study were from third ratoon crops. The varieties 99 F200455, 00 F088470, and 01G166274 were derived from classical breeding and 05TG004101, 05TG008104, and 05TG018114 were derived from precision breeding. The superscripts 55, 70 74, 101, 104, and 114 will be used to describe and discuss the genotypes further in the manuscript. The detailed on how these substrates were sampled, prepared, and analyzed for chemical compositions shown in Table 1 is reported elsewhere [6].

Table 1 Chemical composition of bagasse from different varieties of sugarcane on a dry weight basis
Full size table

The industrial SB (labeled 120) was provided by TSB Sugar Mill in Malelane, Mpumalanga, South Africa. The sample was washed four times and each wash was collected and measured for residual sugar content. The washed bagasse was oven-dried at 40 °C for 72 h, followed by milling them in a laboratory ultra-centrifugal mill model ZM200 basic (Resch GmbH, Germany). The milled sample had a moisture content of 5 %. Prior to its use, the milled samples were sieved in a vibratory sieve shaker model AS200 basic (Resch GmbH, Germany) to obtain a representative particle size suitable for the raw material composition analysis and for the pretreatment studies. The particles retained between 425 and 825 μm were packed in plastic bags and then stored in a temperature- and moisture-controlled room set at 20 °C and relative humidity of 65 % until needed.

DSA Pretreatment

DSA pretreatment was carried out in small tubular batch reactors, according to Yang and Wyman [21]. Dry material (DM, 1.5 g) was soaked in 30 ml of DSA solution for 12 h. Soaked samples were concentrated through filtering to a solid loading of 30 % (w/v). The obtained wet biomass was loaded into the reactor and compressed by a metal rod to ensure uniform heat and mass transfer. The reactor was first submerged into a heating-up fluidized sand bath set at 30 °C above the target temperature. The reactor was heated until the target temperature was reached (approximately within 120 s), after which it was transferred into the second fluidized sand bath set at the target reaction temperature. After the reaction time was completed, the reactor was quenched by submerging into cold water bath. After cooling, the whole slurry was mixed with 100 ml of distilled water and vacuum-filtered into a solid and a liquid fraction. One part of liquid fraction was analyzed for major monomeric sugars (xylose, glucose, and arabinose), sugar degradation (furfural and HMF), and acetic acid formation, and the other part was used to determine the total sugars in the pretreated liquor (monomers and oligomers) by post-hydrolysis as described below. The solid fraction was further washed in three washes (each wash with 100 ml) to raise the pH up to 5 prior to enzymatic hydrolysis, and is subsequently referred to as water insoluble solids (WIS).

Experimental Design and Optimization

A CCD under RSM was selected to determine the relationship between temperature, acid concentration, and time as the main pretreatment parameters. The experimental conditions were designed by Design Expert, version 8.0.2 (State Ease Inc., Minneapolis, MN, USA). Two-level, three-factor CCD was used to determine conditions leading to maximize the CSY as dependent response variable. The number of experiments was six at axial points, six replicates at center point, and eight at factorial points leading to 20 runs (23 + 2 × 3 + 6 = 20). The independent variables in real values are shown in Table 2. The range and levels of independent variables were selected based on the results obtained after a preliminary study on all samples (data not shown). Variables in coded values were calculated by Design Expert. The values for the axial points, factorial points, and center point were (−1.682 at the lowest point and +1.682 at the highest point), (−1 and +1), and (0), respectively. The second-order quadratic model with interactions in coded form was used to predict the optimal pretreatment condition maximum CSY as expressed in Eq. 1.

Table 2 Recovery of glucose, Xylose, and acid insoluble lignin in the WIS after dilute acid pretreatment of different sugarcane bagasse samples (55, 70, 74, 101, 104, 114, and 120) expressed as percentage of theoretical (content in the raw material)
Full size table
$$ Y={\beta}_0+{\beta}_1{X}_1+{\beta}_2{X}_2+{\beta}_3{X}_3+{\beta}_{11}{X}_1^2+{\beta}_{22}{X}_2^2+{\beta}_{33}{X}_3^2+{\beta}_{12}{X}_1{X}_2+{\beta}_{13}{X}_1{X}_3+{\beta}_{23}{X}_2{X}_3 $$
(1)

Where Y, X 1, X 2, and X 3 stand for CSY, temperature, acid concentration, and reaction time, respectively, in coded form: β 0, β 1, β 2, β 3, β 11, β 22, β 33, β 12, β 13, and β 23 are regression coefficients estimated from the experimental data.

Enzymatic Hydrolysis

The WIS fraction was subjected to enzymatic hydrolysis to evaluate the effect of the pretreatment on the enzyme accessibility. These experiments were conducted in 24 ml glass tubes. The tubes were loaded with 200 mg (dry weight) of WIS and 10 ml of 0.05 M citrate buffer (pH 4.8) with the enzyme solution. Sodium azide was added at a concentration of 0.02 % (w/v) to prevent microbial contamination. Two commercial enzymes preparations were used: Spezyme CP (Genencor-Danisco, Denmark) with protein concentration of 140 mg/ml (cellulase activity of 65 FPU/ml) and Novozym 188 (Novozymes A/S, Denmark) with protein concentration of 95 mg/ml (β-glucosidase activity of 700 IU/ml). Protein concentration and enzyme activities of both undiluted enzymes were determined by applying analysis protocol described elsewhere [22]. Cellulase loading of 32.31 mg protein/g WIS (corresponding to 15 FPU/g WIS) of Spezyme CP supplemented with β-glucosidase of 2.02 mg protein/g WIS (equivalent to 15 IU/g WIS) was applied in all the experiments. Tubes loaded with the mixtures were placed in water bath shaker maintained at 50 °C with shaking at 90 rpm. Samples were withdrawn after 72 h, prepared as described below, and analyzed for sugars by HPLC.

Post-Hydrolysis

After the pretreatment, a 5-ml sample of the pretreatment liquor was taken to perform post-hydrolysis by using 72 % sulfuric acid, according to NREL procedure [23], to determine the content of oligomeric carbohydrates. Finally, the sample was prepared for sugars detection by HPLC as described below.

Chemical Composition Analysis Methods

The NREL procedure described by Sluiter et al. [24–26] was used for the chemical composition analysis after being consecutively extracted with water and with 95 % ethanol for 48 h in total in a Soxhlet apparatus. For the pretreated material, the same procedure was used, except that no water or ethanol extraction was carried out because of pretreatment remove most of extractives. The acid soluble lignin of the pretreated material was not measured.

The concentration of glucose, xylose, arabinose, and acetic acid from the liquid fractions resulting from untreated and pretreated materials compositional analysis, pretreated liquor, post-hydrolysis, and enzymatic hydrolysis were quantified by HPLC on an Aminex HPX-87H Column equipped with a Cation-H Micro-Guard Cartridge (Bio-Rad, Johannesburg, South Africa). The column was set to a temperature of 65 °C with a mobile phase of 5 mM sulfuric acid and a flow rate of 0.6 ml/min. Sugars and acetic acid concentrations were measured with a RI detector (Shodex, RI-101) operated at 45 °C. Under these conditions, the column does not resolve xylose, galactose, and mannose. The presence of mannose and galactose in untreated samples were checked on an XbridgeTM Amide column (4.6 × 250 mm, 3.5 μm particle size) equipped with an XbridgeTM Amide precolumn (Waters) at 30 °C, eluted at a rate of 0.7 ml/min with 0.05 % ammonium hydroxide in water (A) and 0.05 % ammonium hydroxide in 90 % acetonitrile (B). Sugars were detected by a Varian 380-LC evaporative light scattering detector. No mannose and galactose peaks were detected. Therefore, the xylose quantification obtained on an Aminex HPX-87H Column was accurate. The glucan, xylan, arabinan, and o-acetyl group contents were calculated as (0.95 × cellobiose + 0.9 × glucose), 0.88 × xylose, 0.88 × arabinose, and 0.683 × acetic acid, respectively [25].

The concentration of HMF and furfural in the pretreated liquor were analyzed on a Phenomenex Luna C18(2) reversed phase column equipped with a Phenomenex Luna C18(2) precolumn (Separations, Johannesburg, South Africa) with column temperature set to 25 °C and a flow rate of 0.7 ml/min. The mobile phases used for elution were 5 mM trifluoroacetic acid in water (A) and 5 mM trifluoroacetic acid in acetonitrile (B). Separation was carried out by gradient elution from 5 % mobile phase B, increasing to 11 % B over 14 min and then increasing to 40 % B over 3 min. The mobile phase composition was then kept constant at 40 % for 2 min, followed by a decrease to 5 % B over 5 min and ending with a final step of constant composition at 5 % B for 4 min in order to equilibrate. HMF and furfural concentrations were measured with a Dionex Ultimate 3000 diode array detector at 215 and 285 nm.

Data and Statistical Analysis

The composition of raw material was performed in four replicates. The average values and standard deviation (average ± standard deviation) was used to present the results. For the pretreatment and enzymatic hydrolysis, the experiments were performed in duplicate and the average values were presented in the Tables 2, 3 and 4.

Table 3 Xylose yield, xylose recovery, and combined furfural and MMF yields in the hydrolysate liquor after dilute acid pretreatment sugarcane bagasse samples (55, 70, 74, 101, 104, 114, and 120) at various conditions
Full size table
Table 4 Yields of glucose after enzymatic hydrolysis and combined sugar (sum of glucose, xylose, and arabinose) after dilute acid pretreatment and enzymatic hydrolysis, and combined sugar recovery (combined sugar yield divided by maximum sugar content in percentage) from sugarcane bagasse samples (55, 70, 74, 101, 104, 114, and 120)
Full size table

The Design Expert, version 8.0 (State Ease Inc., Minneapolis, MN, USA), was applied for the regression analyses and was also used to find the optimum values of CSY. The fitness of the second-order polynomial model obtained from the regression analysis was evaluated by coefficient of determination R 2, and its statistical significance was checked by F test at a probability (p < 0.05). The Student's t test was also performed to determine the statistical significance of the regression coefficients. The STATISTICA software, version 10 (Statsoft Inc., Tulsa, USA), was employed to generate the response surface plot.

The comparison of the responses of the SB samples was facilitated by employing factorial ANOVA. The significant differences in sugar yields among the samples were confirmed by Bonferroni's post hoc test. The hypothesis was accepted or rejected at 95 % confidence interval. Likewise, the correlation coefficients were calculated using STATISTCA (software, version 10).

Results

Feedstocks Chemical Composition

The chemical composition of the bagasse samples obtained from classical breeding varieties (55, 70, and 74) and precision breeding varieties (101, 104, and 114) are summarized in Table 1. The industrial bagasse (120), obtained from sugar mill, was used as reference material. The samples showed considerable variations in chemical components. The values for glucan, xylan, arabinan, lignin, acetyl group, extractives, and ash in dry weight basis ranged from 34.1 to 40.7, 19.5 to 27.2, 1.3 to 2.7, 14.4 to 22.4, 2.8 to 3.2, 3.8 to 6.2, and 0.8 to 2.0 %, respectively. The sum of all components measured varied between 90 and 96.5 %. This could be attributed to components that were not quantified (i.e., methyl glucuronic acid) and some degradation of the sugars occurring during the acid hydrolysis. The sum of glucan, xylan, and arabinan makes the measured total structural carbohydrate vary from 60.5 to 69.2 %. This means that upon hydrolysis the potential sugar released as monomeric (glucose, xylose, and arabinose) ranges between 67.7 and 77.6 g/100 g DM, which make these materials promising feedstocks for ethanol production.

Effect of Pretreatment Conditions on WIS Composition

Table 2 shows the composition of the WIS after DSA pretreatment expressed as percentage of theoretical values. Acid soluble lignin was not measured after pretreatment because its amount was not significant. The recovered WIS was between 50.1 and 76.5 g/100 g raw material (RM) (data not shown), with the lowest value being at the harshest condition (190 °C. 0.85 % (w/w) for 15 min). The sum of glucan, acid insoluble lignin, and xylan accounted for 79.3 to 96.8 % of the WIS recovery. Glucan was less hydrolyzed in most of the pretreatment conditions. The glucan solubilization ranged from 4 to 16 % theoretical. Similarly, no significant differences in the acid insoluble lignin before and after pretreatment at many instances were observed, except at severe conditions where the acid insoluble lignin was higher than the initial values (up to 108 % theoretical). This could be related to the lignin condensation phenomena when severe condition is applied as previously reported elsewhere [18]. The Bonferroni's post hoc test revealed that variety 101 presented the highest ratio of glucan/acid insoluble lignin, whereas the industrial SB (120) had the lowest ratio (not shown). This suggests that variety 101 could be more digestible than others.

Hemicellulose is the prime target for acid hydrolysis. Since xylan is a largest component of hemicellulose (81.2 − 84.5 %), it was used to describe hemicellulose hydrolysis. Xylan remained in the WIS was decreasing exponentially as the pretreatment severity (temperature, acid concentration, and reaction time) was increasing (Table 2). Up to 98.4 % of theoretical xylan was hydrolyzed. Industrial bagasse showed slightly lower xylan solubilization than the rest.

Effect of Pretreatment Conditions on Xylose Hydrolysate Fractions

The composition hydrolysate liquor after pretreatment across conditions is presented in Table 3. Other components measured were glucose, arabinose, and acetic acid (not shown). However, xylose and by-products (furfural and HMF) were selected to analyze the effects of the pretreatment because xylose is the major sugar (56 to 88.7 %), while furfural and HMF are important sugar degradation products.

Xylose yield as a sum of monomeric and oligomeric sugar after applying various different conditions of DSA pretreatment is summarized in Table 3. Xylose yields ranged from 6.9 for variety 55 (170 °C, 0.45 %, 5 min) to 20.2 g/100 g RM for variety 101 (180 °C, 0.65 %, 10). Figure 1 depicts how temperature, acid concentration, and reaction time determined xylose yields. A clear trend was observed of increasing temperature, acid concentration, and reaction time with increasing xylose yields (Fig. 1a). Nevertheless, severe conditions resulted into a significant reduction in xylose yield (Fig. 1d). The decrease in xylose yield at severe conditions suggests rapid destruction of xylose [27]. The highest yields were observed at 180 °C, 0.65 %, 10 min for varieties 55, 70, 74, and 101, and 170 °C, 0.85 %, 15 min for varieties 104 and 114. These conditions yielded 19.4, 19.4, 19.3, 20.2, 19.9, and 20 g/100 g RM, respectively, were corresponding to 69.4, 70.3, 70.8, 67.6, 68.7, and 64.7 % of xylose in native material (Tables 1 and 3). Alternatively, the highest yield from the industrial bagasse was found at 180 °C, 0.99 %, 10 min (16.9 g/100 g RM corresponding to 76.3 % of theoretical) but it was significantly lower than the highest values obtained from the varieties (Table 3). Nevertheless, the precision breeding varieties (101, 104, and 114) exhibited improved xylose yield per gram of biomass than classical breeding (Fig. 2).

Fig. 1
figure 1

The effects of temperature, acid concentration, and reaction time on xylose yield from different sugarcane bagasse samples (55, 70, 74, 101, 104, 114, and 120). a 0.45 (% w/w) for 5 min; b 0.85 (% w/w) for 5 min; c 0.45 (% w/w) for 15 min; d 0.85 (% w/w) for 15 min. The error bars denote 95 % confidence intervals

Full size image
Fig. 2
figure 2

Comparison of xylose yields from different sugarcane bagasse samples (55, 70, 74, 101, 104, 114, and 120) after the dilute sulfuric acid pretreatment. The yields are based mean values of all factorial points. The columns with similar inserted letters do not differ between each other at a significance level of 0.05

Full size image

As expected, furfural and HMF production increased with increasing in pretreatment severity (Table 3). The highest furfural and HMF yields were 4.37 and 0.49 g/100 g RM, respectively. HMF formation was directly correlated with glucose measured in the hydrolysate liquor (not shown). In the case of furfural, no correlation was observed with xylose lost during pretreatment. The lack of correlation suggests that xylose degraded into other compounds rather than furfural alone [11, 28]. In general, varieties 70 and 114 showed higher HMF formation than others whereas samples 55, 104, and 120 showed the lowest. Additionally, the formation of furfural was much faster in variety 101 than the rest while SB 120 showed the least.

Effect of Pretreatment Conditions on Enzymatic Digestibility of WIS

The effectiveness of DSA pretreatment on cellulose digestibility was evaluated in terms of glucose yield (EH glucose) after enzymatic hydrolysis of the WIS and the results are presented in Table 4. Increasing temperature, acid concentration, and reaction time significantly enhanced EH glucose yields (Fig. 3). For example, 71.3 % improvement in glucose yield for variety 70 was obtained when pretreatment temperature increased from 170 to 190 °C (Fig. 3a). However, temperature was less important for glucose yield for variety 104 in many instances. As such, no significant difference in EH glucose yields was observed when the temperature was increased from 170 to 190 °C (Fig. 3b–d).

Fig. 3
figure 3

Glucose yields as the function of temperature after enzymatic hydrolysis of different sugarcane bagasse (55, 70, 74, 101, 104, 114, and 120) for a 0.45 % w/w sulfuric acid and 5 min; b 0.85 % w/w and 5 min; c 0.45 % w/w and 15 min; d 0.85 % w/w and 15 min applied in the pretreatment. The error bars denote 95 % confidence intervals

Full size image

Comparative analysis of EH glucose yields between varieties revealed considerably variations. Up to 86.8 % differences in glucose yield was observed at the mild condition (170 °C, 0.45 % for 5 min). Remarkably, under this condition (170 °C, 0.45 % for 5 min), glucose yield achieved by the best performing variety (101) of 31.2 g/100 g RM was statistically comparable to the maximum yield obtained by the poor performing varieties (70 and 74) of 31.9–32.3 g/100 g RM or by the control (120) of 33.7 g/100 g RM. Ranking the varieties based mean EH glucose yields across experiments 1 to 8, the data clearly demonstrates how specific variety outperforms others (Fig. 4). The yields differences between the varieties were in the order of 101 > 1114 > 55 > 104 > others (70, 74).

Fig. 4
figure 4

Comparison of glucose yields from different sugarcane bagasse (55, 70, 74, 101, 104, 114, and 120) after enzymatic hydrolysis of dilute sulfuric pretreated materials. The yields are based on mean values of all factorial points. The vertical error bars denote 95 % confidence intervals

Full size image

Figure 5 systematically compares glucose recovery (calculated as EH glucose yield divided by potential glucose in the WIS expressed as percentage) from best variety 101 vs. medium performing (55) vs. industrial bagasse (control) across selected conditions. DSA pretreatment increased considerably the glucan conversion giving values from 43.2–72 % for the less severe conditions to 100 % for the harshest conditions. Varieties 55 and 101 seemed to be more digestible than the industrial bagasse. At mild conditions, the differences in glucose recovery between the samples were bigger but decreased at severe conditions.

Fig. 5
figure 5

Bioconversion of different bagasse samples (55, 101, and 120) across selected dilute acid pretreatment conditions of increasing severities. EH glucose recovery was calculated as percentage of potential glucose in the WIS. Inserted letters (a, b, c) means values with different alphabet differed significantly at p < 0.05

Full size image

Statistical Modeling of CSY and Validation

The experimental results on CSY summarized in Table 4 were fitted into the quadratic model (Eq. 1) to quantitatively estimate the effect of each independent variable on the CSY. The statistical significant of each factor was determined by ANOVA, which revealed that all process parameters influenced the CSY. Table 5 shows the model coefficients and significance term for each sample. All linear increased the CSY, except the acid concentration for variety 70, which showed the negative effect. The rest of the models coefficients impacted CSY in a negative manner. This suggests that the increases of temperature, acid concentration, and reaction time do not always translate into high CSY. Furthermore, the p value of the model was lower than 0.005, indicating that the model was significant (not shown). The lack of fit was not significant, which imply that the models reasonably predict the experimental data (not shown). Likewise, the determination coefficient (R 2) of the model was also high (0.91–0.98), showing that more than 90 % of the result variability was attributed to the process variables.

Table 5 Regression coefficients of the mathematical models of combined sugar yield from sugarcane bagasse samples (55, 70, 74, 101, 104, 114, and 120, probability from the ANOVA table and determination coefficient
Full size table

The models were plotted in a three-dimensional surface response to understand the effects of the independent variables on CSY (Fig. 6). The plots represent the interactions of the two independent variables, while the third variable is held constant. The analysis of the surface responses for samples 55, 70, 74, and 120 showed that the increase of temperature and reaction time improved the CSY to some extent. However, further increase of these two variables resulted in the considerable reduction of the CSY. Similar plots on CSY were observed for varieties 101, 104, and 114, when acid concentration and reaction time were varied while temperature was kept constant at the center point (180 °C).

Fig. 6
figure 6

The response surface plot showing (a) the influence of temperature and reaction time on the combined sugar yield from sugarcane bagasse with identification 55, 70, 74, and 120 and (b) the influence of acid concentration and reaction time on the combined sugar yield from sugarcane bagasse with identification 101, 104, and 114

Full size image

The numerical optimization process was conducted to determine the optimal condition leading to maximum CSY. The optimization criteria were set according to maximization of EH glucose and CSY, and to minimization of by-products formation. The maximization of sugars was set higher level of importance compared to furfural and HMF formation. After the optimization procedures, the optimal pretreatment condition and the maximum CSY were identified (Fig. 7). The predicted maximum CSY was validated by performing extra experiments in triplicates at the predicted best pretreatment conditions and results are also depicted in Fig. 7. There was a good agreement between the model prediction and the experimental data. The total sugar recoveries obtained after optimization were 84.9, 79.3, 78.8, 87.1, 82.4, 79.0, and 74.5 % of theoretical for samples 55, 70, 74, 101, 104, 114, and 120, respectively. This show an increase of up to 4.6 % compared the highest average values obtained before optimization (Table 4).

Fig. 7
figure 7

The predicted condition for the model optimization, predicted values (M) and the validation experimental values (E) of combined sugar yields from different sugarcane bagasse. Xylose Pr is xylose yield after pretreatment and Glucose EH is glucose yield after enzymatic hydrolysis. Others stands for the sum of glucose, arabinose (after pretreatment), and xylose after enzymatic hydrolysis

Full size image

The optimal condition for each variety was substituted into all models (Table 5) and the values obtained were compared to the maximum CSY (Fig. 7). The optimal conditions for varieties 70 and 74 underpredicted the CSY for varieties 55, 101, and 104 up to 10.7 %. Conversely, the optimal conditions for varieties 55, 101, 104, and 114 accurately estimated the CSY for all varieties. The estimated yields were within the experimental errors (0.1 − 2.4 %).

The Effect of Chemical Composition on Xylose and EH Glucose Yields

The influence of chemical compositions on xylose and EH glucose yields was estimated by calculating correlation coefficients across pretreatment conditions (Table 6). Lignin and ash content showed negative correlation with xylose and EH glucose yields. Xylan, arabinan, and acetyl content positively correlated with xylose and EH glucose yields.

Table 6 Correlation coefficient (r) between chemical components and xylose, or EH glucose yields across pretreatment conditions
Full size table

Mass Balance

Figure 8 depicts the overall mass balance of the solids and liquid fractions of the best performing variety (101) and the control (industrial bagasse) after DSA pretreatment at the center point conditions (180 °C, 0.65 %, for 10 min). This condition was taken as an example to evaluate the material balance because it was replicated six times (experiments 15 to 20) and results were statistically reproducible (Table 4, Fig. 6). In addition, CSY obtained by this condition was very close to the maximum values (Fig. 7). The industrial bagasse showed higher WIS recovery (68.1 %) than that of variety 101 (59.2 %) due to higher acid insoluble lignin. The overall mass losses were 10.8 % for variety 101 and 9.3 % for industrial bagasse. The possible reason for these lost mass could be the decomposition and degradation of xylan into other compounds than furfural, which were not measured by HPLC. In addition, this study did not quantify the acid soluble lignin in the liquid fractions after pretreatment. Other components such as ash and extractives were also not measured after the pretreatment. All this could contribute to the above losses. Despite the overall mass balance losses, the overall material balances of glucan, arabinan, and acid insoluble lignin of both samples were above 93 %, showing that the analytical methods used were accurate.

Fig. 8
figure 8

The overall mass balance of liquid and solid fractions obtained after dilute sulfuric acid pretreatment at center point conditions (180 °C, 0.65 % acid for 10 min): (a) best performer variety (101) and (b) industrial bagasse (120). AS and AI refer to as acid soluble and acid insoluble, respectively

Full size image

Discussion

A Combination of Pretreatment Optimization and Feedstock Selection for Increases Sugar Yields

Xylose yields after pretreatment (Table 3), EH glucose yields after enzymatic hydrolysis (Table 4), and combined sugar yields after pretreatment-hydrolysis (Table 4) varied significantly, in part depending on the pretreatment condition and the chemical composition, with the latter distinguished the varieties at the same pretreatment condition. Increasing pretreatment severity (temperature, acid concentration, and reaction time) led to improved sugar yield from the varieties (Figs. 1, 3 and 6). However, severe conditions reduced the xylose yield as well as CSY due to excessive degradation of xylose to furfural (Table 3). The exponential accumulation of furfural with the increase of pretreatment severity has been reported elsewhere [29]. Conversely, severe conditions, particularly high temperatures, promoted to glucose yield/recoveries by producing highly digestible solids [18, 27]. The effectiveness of DSA pretreatment method is primarily based on solubilization of hemicellulose thereby increases the available surface area, making cellulose more accessible to enzymatic hydrolysis [2, 10]. In addition, high temperature assists in weakening the structure of the biomass, which further increases digestibility [27].

Furthermore, varieties with lower lignin, lower ash, and higher structural carbohydrates content showed higher yields of xylose and EH glucose as well as CSY (Figs. 2, 4 and 7; Tables 1 and 6). This favored most of the precision breeding varieties over the majority of the classical breeding varieties. As such, most of the precision breeding varieties showed improved xylose yields (Fig. 2), glucose yields (Fig. 4), as well as CSY (Fig. 7). Nevertheless, this is the first study to observe the influence of chemical composition on xylose yield [30–32]. Most of these correlations reported in literature were based on a single pretreatment condition. Similar conclusion could also be drawn in the current study by utilizing single pretreatment condition that was not leading to strong correlation. The weaker relationship between chemical composition and xylose yields could be related to sugar degradation.

To date, CSY after DSA pretreatment-hydrolysis of SB from South Africa mill is not been higher than 49.5 g/100 g RM [33], which was similar to 50.4 g/100 g RM for industrial bagasse obtained in the present study. According to our results, the CSY could be increased up to 34.1 % by selecting the best performing variety (Fig. 7). These results clearly demonstrate the importance of pretreatment optimization and feedstock selection for increasing fermentable sugar yield per gram of biomass, with the latter depending on feedstock quality, i.e. high structural carbohydrates content, reduced lignin content, and improved digestibility.

Feedstock Quality Determines Bioconversion Efficiency of Biomass

Bioconversion trend observed for the recovery of glucose across pretreatment conditions was attributed to chemical composition differences among the varieties. The glucose recovery was higher for the samples with improved quality, i.e., high ratio of carbohydrates/lignin and low ash content (Fig. 5; Tables 1 and 6). It was further observed that the differences in the recovery between the samples were greater at mild conditions than at severe conditions. This observation is in agreement with a most recent work on maize genotypes [34], which demonstrated that digestibility of corn stover at sup-optimal pretreatment was largely determined by the chemical composition features but its influence was less evident when increasing pretreatment severity.

Furthermore, from an economics point of view, the feasibility of cellulosic ethanol production ethanol at the industrial scale is conditioned to the efficient release of sugar (currently glucose) from lignocellulose biomass [35, 36]. This favors severe conditions for maximum cellulose conversion (Fig. 5). However, high pretreatment severity implies high energy or chemical demands. In addition, severe conditions results into lower fiber recovery in the process and xylose degradation (Fig. 1, Table 3). Biorefinery industry is currently seeking for a new solution that is able to reduce the production costs. Our results have demonstrated that high bioconversion efficiency could be obtained by applying less severe conditions (Fig. 5). More recently, diverse studies on feedstock quality have proven that some of the intrinsic properties of the crop are heritable [37–39]. Therefore, the crop properties that caused the best performing variety to have reduced recalcitrance to pretreatment and enzymatic hydrolysis needs to be further investigated. Through this way, the processing costs might be reduced while increasing ethanol yield per gram of feedstock input.

Assessment of Common Optimal Pretreatment Conditions

One of the purposes of this study was to establish common optimal pretreatment conditions that could be applied during screening of sugarcane varieties. Previous optimization of DSA pretreatment conditions determined that pretreatment severity around 3.5 was the best compromise between xylose recovery and cellulose digestibility [40–42]. This fact also remained true for the optimal conditions for varieties 70 and 74. However, these conditions were slightly severe compared to best conditions (severity of 3.3 − 3.4) for other varieties; consequently, they underestimated maximum CSY for varieties 55, 101, 104, and 114, up to 10 %. In contrast, the remaining four conditions (179 °C, 0.54 %, 12 min; 177 °C, 0.7 %, 10 min; 176 °C, 0.77 %, 12 min; and 181 °C, 0.65 %, 10 min) were appropriate for simulating maximum CSY from each variety. Therefore, any of these conditions can be applied for varieties screening, provided that experiments are conducted in a similar manner as that was used in the present study.

Conclusions

A combination of feedstock selection and pretreatment optimization has the potential to improve conversion efficiency of the biomass and reduce pretreatment. In the present study, conversion efficiencies of bagasse from six varieties of sugarcane after DSA pretreatment and enzymatic hydrolysis were investigated. The results revealed considerable differences in sugar yields/recoveries among the varieties. The maximum CSY ranged from 55.1 g/100 g RM (78.8 % of theoretical) to 67.6 g/100 g RM (87.1 % of theoretical). The maximum CSY from industrial bagasse was only 50.4 g/100 g RM (74.5 % of theoretical). Generally, high ratio of carbohydrate/lignin and reduced ash content favored sugar yields from the samples. However, at severe condition, the bioconversion efficiency of the varieties was largely determined by the severity of the pretreatment. As such, the differences in glucose yield/recovery between varieties were smaller at severe condition than at mild condition. At mild conditions, glucose yield/recovery was higher to those varieties with higher substitution of structural carbohydrate, lower lignin, and lower ash content. It was further established that different sugarcane varieties had a common optimal pretreatment conditions for maximum CSY. Identification of such conditions is of high contribution to biofuel industry as single pretreatment condition can be applied during screening of sugarcane varieties.