Synthesis of DESs
Betaine chloride (BaCl) and ethylamine chloride (EaCl) possess similar structure as ChCl, they were explored as HBA in the synthesis of DESs. As shown in Scheme 1, EaCl is smaller than ChCl, while BaCl is similar to ChCl. All of them contain quaternary nitrogen cation, which is favourable to DES formation. There is a free hydroxy group in ChCl, a carboxyl group in BaCl and a free aliphatic terminal in EaCl. Lactic acid (LAC), ethyl glycol (EG), glycerol (GLY) and urea (UR), which have been commonly used and proved to be effective in forming DESs with ChCl, were introduced as HBDs. Generally, DESs were empirically synthesized by combination of various kinds and ratios of HBD and HBA. However, most of the combinations were hard to be synthesized. Considering these unpredictable combination patterns, both experimental synthesis employing methods for ChCl based DESs and DFT calculations with different functions were performed to explore the feasibility of rational design of DESs.
For ChCl and EaCl based DESs, all of the four combinations, ChCl:LAC, ChCl:EG, ChCl:GLY and ChCl:UR, EaCl:LAC, EaCl:EG, EaCl:GLY and EaCl:UR respectively, were successfully obtained. However, with regard to BaCl as HBA, only BaCl:EG and BaCl:GLY could be facilely synthesized in clear and homogenous liquid. It should be mentioned that further optimization of reaction conditions such as ratios of HBD to HBA and temperature might also produce BaCl:LAC and BaCl:UR. Furthermore, ΔGrxn of the reaction (ΔGrxn=GDES–GHBA–n×GHBD) was calculated employing three mostly common used DFT including B3LYP, M062X and ωB97X and basis set of 6-311+G**. Negative values of ΔGrxn can be used to indicate the thermodynamically feasibility. Previously, B3LYP and M062X have been used in the simulation of ChCl based DESs [35–38]. In Table 1, ΔGrxn values of easily obtained DESs were lower than 0, according to the ΔGrxn results of B3LYP (ΔGB3LYP rxn). Especially, the ΔGB3LYP rxn values of ChCl:LAC and EaCl:LAC were the lowest. With regard to BaCl:LAC and BaCl:UR, the ΔGB3LYP rxn values were 3.59 and 1.03 kcal·mol–1 respectively. For the results using M062X, there is no definite patterns. For example, the ΔGM062X rxn value of BaCl:UR was –10.8 kcal·mol–1, ranking the lowest, whereas, ΔGB3LYP rxn of BaCl:UR was the highest. All the ΔGωB97XD rxn values were negative, which were hard to be correlated with the reactivity. As a result, calculation method of B3LYP/6-311+G** is more favorable in predicting the potential of DES synthesis, and might be used to elucidate the reactivity and mechanism of DES mediated systems in pretreatment of lignocellulosic biomass.
Table 1 Deep eutectic solvents synthesized in this study.
DES
|
Ratio of
HBA to HBD
|
ΔGB3LYP rxn [kcal·mol–1]
|
ΔGM062X rxn [kcal·mol–1]
|
ΔGωB97XD rxn [kcal·mol–1]
|
ChCl:UR
|
1:2
|
–2.54
|
–6.39
|
–6.50
|
ChCl:EG
|
1:2
|
–2.84
|
–7.15
|
–6.55
|
ChCl:GLY
|
1:2
|
–2.27
|
–2.86
|
–2.58
|
ChCl:LAC
|
1:1
|
–8.96
|
–10.3
|
–9.95
|
BaCl:UR
|
1:2
|
3.59
|
–10.8
|
–0.54
|
BaCl:EG
|
1:2
|
–1.42
|
–11.2
|
–7.15
|
BaCl:GLY
|
1:2
|
–1.21
|
–9.88
|
–1.73
|
BaCl:LAC
|
1:1
|
1.03
|
–4.92
|
–3.23
|
EaCl:UR
|
1:2
|
–2.25
|
–4.55
|
–3.80
|
EaCl:EG
|
1:2
|
–1.62
|
–4.21
|
–1.73
|
EaCl:GLY
|
1:2
|
–2.38
|
–4.19
|
–3.31
|
EaCl:LAC
|
1:1
|
–4.67
|
–7.94
|
–8.23
|
Note: ChCl: choline chloride, BaCl: betaine chloride, EaCl: ethylamine chloride, UR: urea, EG: ethylene glycol, GLY: glycine, LAC: lactic acid.
The optimized geometries of EaCl:LAC, ChCl:LAC and BaCl:LAC were obtained from DFT calculations. Distance and interactions among HBD and HBA were also analyzed. Three potential hydrogen bonds could formed between EaCl and LAC, which are favorable for the formation of EaCl:LAC (Fig. 1A). In addition, due to the small size of EaCl, the strong electrostatic interaction between nitrogen cation of ethylamine and carboxy group of lactic acid could also contribute to the stabilization of EaCl:LAC. In ChCl:LAC, two hydrogen bonds could be formed between ChCl and LAC (Fig. 1B). However, the distance from nitrogen cation and carboxyl group is too large to form stable interaction. With regard to BaCl:LAC, few interactions were found between BaCl and LAC (Fig. 1C), which might account for the high ΔGrxn value and also the difficulties in preparation of BaCl:LAC. It is well known that electrostatic forces of the ions in ILs or DESs could provide or receive H-bond or localize charges on the chemical bonds of the matrix of lignocellulosic biomass. As a result, EaCl:LAC is presumably to display high efficacy in pretreatment of lignocellulosic biomass.
To further evaluate the effects of EaCl:LAC in reducing the recalcitrance of biomass, component analysis was conducted. Contents of cellulose, hemicellulose and lignin were determined and shown in Table 2. For raw biomass, the cellulose content of corncob was 30.0%, which was higher than that of pod (21.8%) whereas much lower than 38.1% of wheat straw, 35.0% of rice husk, 32.0% bagasse and 31.7% of rice straw. Remarkably, the cellulose content of corncob was increased to 70% after pretreatment with EaCl:LAC. In fact, cellulose contents of all other tested biomass were increased to some extent (11–34%), indicating the effectiveness of EaCl:LAC in reducing recalcitrance of biomass. The cellulose yield of corncob was as high as 98.0%, much higher than other biomass. The hemicellulose contents of corncob, rice straw, pod, wheat straw, bagasse and rice husk were 14.6%, 10.0%, 10.9%, 8.4%, 11.6% and 6.7% respectively. After EaCl:LAC pretreatment, the hemicellulose removal of 87.9%, 81.1%, 75.3%, 69.9%, 83.1% and 62.7% were achieved for corncob, rice straw, pod, wheat straw, bagasse and rice husk. With regard to lignin including acid-soluble and acid-insoluble, their content in corncob was 26.5%, while in rice straw, pod, wheat straw and rice husk were higher than 30%. The lignin removal of corncob and wheat straw was 71.5% and 67.0% respectively, much higher than 61.3%, 42.7%, 57.2% and 62.7% of rice straw, pod, bagasse and rice husk. The solid recovery rate of all the tested biomass fell into a range of 40–58%. It should be noted that other components including pigments, proteins and fatty acids etc accounted for 9.7–44.5% of raw biomass (Table 2). Most of them could also be efficiently removed by EaCl:LAC pretreatment (Table 2). The results suggest that EaCl:LAC could effectively reduce the stubborn resistance of lignocellulose and lignin in corncob and enhance the cellulose accessibility to cellulase. The excellent performance of EaCl:LAC in lignin removal is consistent with the high potential of DES in the extraction of lignin and metals [32]. In comparison with other DESs in the pretreatment of corncob, EaCl:LAC displayed higher efficiency in removing both hemicellulose and lignin, which is usually difficult to be achieve by choline chloride or betaine based DESs (Additional file 1) [42]. Generally, combinatorial pretreatment method using alkaline or oxidant should be introduced to achieve higher glucose yield. As a result, the newly developed EaCl:LAC is an efficient pretreatment reagent for lignocellulosic biomass.
Table 2 Component analysis of various lignocellulosic biomass before and after treatment with EaCl:LAC.
Component (%)
|
Corncorb
|
Rice straw
|
Pod
|
Wheat straw
|
Bagasse
|
Rice husk
|
Raw
|
Treated
|
Raw
|
Treated
|
Raw
|
Treated
|
Raw
|
Treated
|
Raw
|
Treated
|
Raw
|
Treated
|
Cellulose
|
30.0±0.8
|
70.0±3.1
|
31.7±2.2
|
51.0±2.4
|
21.8±0.1
|
42.0±0.4
|
38.1±0.8
|
63.8±1.4
|
32.0±1.1
|
66.0±2.7
|
35.0±1.7
|
44.0±1.7
|
Cellulose yield
|
‒
|
98.0
|
‒
|
70.7
|
‒
|
92.4
|
‒
|
92.3
|
‒
|
82.5
|
‒
|
72.9
|
Hemicellulose
|
14.6±0.1
|
4.2±0.2
|
10.0±1.2
|
4.3±0.4
|
10.9±0.7
|
5.6±0.7
|
8.4±0.8
|
4.6±0.3
|
11.6±1.4
|
4.9±0.2
|
6.7±1.1
|
4.3±0.4
|
HC removala
|
‒
|
87.9
|
‒
|
81.1
|
‒
|
75.3
|
‒
|
69.9
|
‒
|
83.1
|
‒
|
62.7
|
AS ligninb
|
2.3±0.1
|
0.9±0.1
|
1.5±0.2
|
0.7±0.0
|
1.5±0.2
|
0.7±0.1
|
0.9±0.1
|
0.7±0.0
|
1.5±0.0
|
0.9±0.0
|
0.8±0.1
|
0.5±0.0
|
AIS ligninc
|
24.2±1.4
|
17.1±1.9
|
31.4±1.2
|
28.2±0.2
|
31.1±0.5
|
38.2±0.5
|
37.1±1.0
|
21.9±1.0
|
9.8±0.6
|
11.2±1.4
|
36.1±1.3
|
29.6±1.2
|
Lignin removal
|
‒
|
71.5
|
‒
|
61.3
|
‒
|
42.7
|
‒
|
67.0
|
‒
|
57.2
|
‒
|
52.7
|
Solid yield
|
‒
|
42.0
|
‒
|
44.0
|
‒
|
48.0
|
‒
|
55.0
|
‒
|
40.0
|
‒
|
58.0
|
Ash
|
0.4±0.0
|
0.7±0.0
|
6.6±0.0
|
11.7±0.0
|
0.5±0.1
|
0.4±0.1
|
1.0±0.0
|
1.9±0.0
|
0.6±0.0
|
1.2±0.0
|
11.7±0.2
|
17.0±1.0
|
Others
|
28.9±2.3
|
7.1±0.7
|
18.8±3.6
|
4.1±0.3
|
34.2±1.1
|
13.1±0.7
|
14.5±1.4
|
7.1±2.0
|
44.5±3.0
|
18.8±1.9
|
9.7±1.0
|
4.6±1.7
|
Note: a HC removal: hemicellulose removal; b AS Lignin: acid-soluble lignin; c AIS lignin: acid-insoluble lignin.
Physical characterization of corncob pretreated by EaCl:LAC
In corncob, lignin and hemicellulose form a tight network structure wrapping around the outer layer of cellulose, which seriously hinders the accessibility of cellulose by cellulase [13]. SEM analysis was implemented to monitor the surface structure of untreated and pretreated corncobs (Additional file 2). In untreated corncob, a smooth and compact surface with strong rigid structure was observed. However, an entirely different landscape was detected in the pretreated corncob. The surface of pretreated corncob became loose and rough with obvious fracture delamination, revealing destroyed lignin and hemicellulose around cellulose, which was favorable for improved cellulose accessibility in corncob. Moreover, the observed changes in corncob surface are consistent with the high lignin and hemicellulose removal after EaCl:LAC pretreatment.
Furthermore, XRD assay was conducted to explore changes of the crystallinity index (CrI) of untreated and pretreated corncobs. According to the overlapped XRD spectrum (Additional file 3), no new peak appeared in the pretreated corncob, indicating no structural change after pretreatment. The diffraction peaks at 16° and 21° represent the typical crystalline structures of cellulose I, and could be used to calculate CrI [11]. The CrI value could be regarded as an indicator for the exposure and accessibility of cellulose. Above two characteristic absorption peaks of pretreated corncob were much higher than those of raw corncob, largely due to the increased cellulose content after removal of lignin and hemicellulose. The CrI values of raw and pretreated corncob were 31.0% and 42.8%, respectively. The increased CrI value of pretreated corncob agrees with the increased cellulose content (Table 2), and indicates the successful removal of certain stubborn components wrapped around the cellulose. It is favorable for increasing the accessibility of cellulase to cellulose in lignocellulosic biomass [43].
FTIR spectrum of untreated and pretreated corncobs was obtained (Addition file 4). The absorption peaks at 830 and 1166 cm–1 refers to the vibration of C-C bond in lignin, indicating the lignin in corncob is SGH lignin (Syringyl-guaiacyl-p-hydroxyphenyl) [44]. In comparison with untreated corncob, the characteristic absorption peaks of lignin in pretreated corncob were significantly reduced, revealing that a large amount of lignin was removed. The absorption peak at 1638 cm–1 is attributed to the stretching vibration of γ-lactone, and the decrease value means that the lignin was largely removed after pretreatment [28,45]. The increased absorption peak at 895 cm–1, relating to β-glycosidic bond in cellulose, indicates the removal of hemicellulose and exposure of more cellulose. Furthermore, the absorption peak at 1383 cm–1 is caused by the stretching vibration of C-H bond in cellulose, and the increased value shows that the amorphous cellulose was removed after EaCl:LAC pretreatment. The absorption peak at 1736 cm–1 represents the vibration of carboxyl group in hemicellulose, and the decreased adsorption peak of pretreated corncob reveals the removal of hemicellulose in comparison with raw corncob [44]. In summary, the FTIR result agrees with the composition analysis. After pretreatment with EaCl:LAC, a large amount of lignin and hemicellulose in corncob were removed, and the relative content of cellulose was significantly increased to 70.0%, resulting in enhanced cellulose accessibility.
Development of fed-batch pretreatment process
To establish an efficient and economic corncob pretreatment process, various factors were optimized. Firstly, conditions including temperature, incubation time and solid-liquid ratios were systematically investigated, and the resultant corncobs pretreated by EaCl:LAC were subjected to enzymatic hydrolysis for determination of total sugars (Additional file 5). At 90°C and 110°C, elongated pretreatment time from 0.5 h to 3.0 h resulted in higher total sugars. However, when the temperature increased to 130°C and 150°C, different profiles were observed. At over 130℃, longer incubation time led to decreased total sugars, which might be attributed to the destruction of cellulose structure. As a result, either high temperature for short time or low temperature for long time is beneficial to the performance of EaCl:LAC. Under the optimum pretreatment conditions of 150°C for 0.5 h and solid-liquid ratio of 1:15, the highest total sugars concentration of about 55.6 g·L–1 were obtained from the pretreated corncob (Additional file 5).
Furthermore, factors including cellulase dosage, hydrolysis time, solid to liquid ratio and supplementation of Tween80, which might influence the enzymatic hydrolysis process, were investigated. Different amounts of cellulase, ranging from 10 to 70 FPU·g–1 pretreated corncob, were loaded, and the released total sugars were monitored as illustrated in Fig. 3A. Along with the increase of hydrolysis time, the total sugars elevated rapidly during the initial 24 h, and then slowly increased until 72 h. Although longer hydrolysis time could lead to higher concentrations of total sugars, it also results in compromised space-time yield. At 50 FPU·g–1 cellullase, total sugars of 57.0 g·L–1 was obtained at 24 h, merely 4.5 g·L–1 lower than that of 70 FPU·g–1 cellullase. Considering the relative lower loading of cellulase and higher efficiency, hydrolysis with 50 FPU·g–1 cellullase for 24 h was selected as the suitable condition. Influence of solid to liquid ratios at 1:8, 1:10, 1:12 and 1:15 on releasing of total sugars were also investigated at 50 FPU·g–1 cellullase (Fig. 3B). Increased liquid ratios represent lower addition of biomass. Along with the increase of solid to liquid ratios, the total sugars decreased from 64 g·L–1 to 44 g·L–1 after 24 h of hydrolysis. The total sugar yield per pretreated corncob increased from 513 g·kg–1 to 661 g·kg–1. Although enhanced total sugar yield could be achieved at a relatively high liquid ratio at the expense of cellulase, however, lower titer of total sugars could be resulted, which is disadvantageous for biobutanol fermentation since extra energy-consuming concentration step is required to increase the sugar concentration. In the view of better mass transferring and relatively higher total sugars, solid to liquid ratio of 1:12 is considered as optimum, at which the total sugars of 50 g·L–1 could be achieved after 24 h of hydrolysis.
Although most of the lignin and hemicellulose could be removed from corncob after EaCl:LAC pretreatment, residual lignin could competitively adsorb free cellulase, which might result in losing of cellulase and impairing hydrolysis efficiency. Supplementation of bovine serum albumin (BSA) or Tween80 has been proved to be effective solutions for reducing inefficient adsorption of cellulase on lignin and deactivation of absorbed cellulase by enzyme-substrate interaction [46,47]. Herein, addition of Tween80 was also attempted (Fig. 3C). In comparison with the control (without Tween80), supplementation of 0.1%–1.0% (v/v) of Tween80 resulted in increased total sugars. At 1.0% Tween80, total sugars concentration of as high as 55.1 g·L–1 was attained, 10.2% higher than 50.0 g·L–1 of control. Excessive addition of Tween80 could however complicate the compositions and affect the biocompatibility of hydrolysates in biobutanol fermentation. At 0.5% Tween80, the total sugar reached 53.8 g·L–1, which was adequate for butanol fermentation [38]. As a result, addition of 0.5% Tween80 is selected for the hydrolysis of pretreated corncob into fermentable sugars.
To further reduce the enzyme loading, cellulases absorbed on residual corncobs were recycled and reused in the consecutive batches. Herein, two processes with and without addition of 0.5% Tween80 were evaluated. At the end of each batch, the residual solids which might absorb cellulases as previous reported [3], were collected and reloaded into the next batch. The loadings of cellulase were decreased by 5 FPU·g–1 for the following batches. The absorbed cellulases were recycled for five times, and sugars including glucose, xylose and arabinose were determined and illustrated in Fig. 4. The total sugars increased rapidly within the initial 6 h, and the addition of cellulase attached to corncob did not result in a decrease in enzymatic efficiency since it could lead to compromised mass transfer compared with the first batch (Cycle I). In the process with Tween80 (Fig. 4B), total sugars concentration of Cycle I was 52.9 g·L–1, while it was 49.5 g·L–1 in control (without Tween80) (Fig. 4A). Addition of Tween80 was favorable for the enzymatic hydrolysis, exhibiting 7–14% increase in total sugars at each batch. In the sixth batch (Cycle IV), only 25 FPU·g–1 of fresh cellulase was supplemented. The total sugars reached 58.8 g·L–1 and 54.9 g·L–1 for processes with and without Tween80 respectively, which were 706 and 659 g·kg–1 corncob pretreated by EaCl:LAC. The total sugars increased by about 11% than that of Cycle I. It is presumed that Tween80 might reduce the inactivation of cellulase caused by interaction between cellulase and substrate. Thus, the cellulases adsorbed on corncob displayed stable and even improved enzymatic activity in the next cycle, which is consistent with previous study [47]. The total sugars concentrations of each batch were enough as carbon source for the butanol fermentation with C. saccharobutylicum. It should be noted that about 50% of cellulases could be saved through this newly developed recycling process.
Biobutanol fermentation with corncob hydrolysates by C. saccharobutylicum DSM13864
Application of hydrolysates from EaCl:LAC-pretreated corncob was evaluated in biobutanol fermentation. C. saccharobutylicum DSM13864 could utilize pentoses, such as xylose, as carbon sources and is regarded as one of the most promising bacteria for biobutanol fermentation. Hydrolysates of the sixth batch were collected and designated as Cycle VITween80 and Cycle VI for with and without Tween80 addition respectively. The total sugars concentrations of Cycle VITweeen80 and Cycle VI were determined to be 58.8 g·L–1 and 54.9 g·L–1. Control experiments were also carried out with glucose as carbon source instead of hydrolysates. The glucose concentrations of the control groups were kept at the same level with the total sugars of the hydrolysates from Cycle VITween80 and Cycle VI. Consumption of reducing sugars and production of acetone, butanol and ethanol (ABE) were monitored and illustrated in Fig. 5 and Table 3. During the initial 48 h, C. saccharobutylicum grew quickly with high sugar consumption and ABE production rates. After 48 h, ABE production decreased, along with a slower sugar consumption rate. After 72 h, butanol titers of 10.2 and 10.4 g·L–1 were reached for Cycle VI (Fig. 5A) and Cycle VITween80 (Fig. 5C) respectively, slightly lower than the corresponding glucose controls of 11.2 (Fig. 5B) and 11.4 g·L–1 (Fig. 5D). Although the concentrations of total sugars in hydrolysates are at same level with glucose controls, they are mixture of arabinose, xylose and glucose. The lower ABE titers of hydrolysates might be attributed to the lower glucose concentrations and the complicated metabolite fluxes of hydrolysates. It should be noted that the butanol yield and productivity of Cycle VITween80 and Cycle VI hydrolysates were 194 g∙kg‒1 total sugar and 0.15 g∙L‒1∙h‒1, and 206 g∙g‒1 total sugar, and 0.14 g∙L‒1∙h‒1 respectively, which are at similar level with those of glucose controls (Table 3). The specific yields of butanol of Cycle VITween80 and Cycle VI per pretreated corncob were 137 and 136 g∙kg–1. With regard to total solvents of ABE, the titers of Cycle VITween80 and Cycle VI were 15.8 and 15.6 g·L–1, with calculated yields per total sugars of 295 and 315 g∙kg‒1 total sugar, and calculated yields per pretreated corncob of 208 and 207 g∙kg–1 respectively. As a result, corncob hydrolysates from Cycle VI could be efficiently utilized by C. saccharobutylicum as carbon source for biobutanol fermentation. Moreover, the corncob hydrolysates did not display obvious inhibitory effect on the growth and biobutanol production of C. saccharobutylicum.
Table 3 Biobutanol fermentation with corncob hydrolysates by C. saccharobutylicum DSM13864.
Carbon source
|
Butanol
|
Acetone-butanol-ethanol (ABE)
|
Titer
[g∙L‒1]
|
Yielda
[g∙kg–1]
|
Prod.b
[g∙L‒1∙h‒1]
|
Titer
[g∙L‒1]
|
Yielda
[g∙kg–1]
|
Prod.
[g∙L‒1∙h‒1]
|
Cycle VI
|
10.2
|
206
(136)
|
0.14
|
15.6
|
315
(207)
|
0.21
|
Control Ic
|
11.2
|
210
|
0.15
|
16.3
|
320
|
0.22
|
Cycle VITween80
|
10.4
|
194
(137)
|
0.15
|
15.8
|
295
(208)
|
0.22
|
Control IId
|
11.4
|
206
|
0.16
|
16.9
|
306
|
0.23
|
Note: a Numbers outside the brackets represent specific yields per total sugars or glucose, numbers inside the brackets refer to specific yields per pretreated corncob. b Prod.: productivity. c Control I: 54 g·L–1 glucose. d Control II: 59 g·L–1 glucose.
This study provides a simple and biocompatible process for the facile conversion of corncob into biobutanol. Compared with other established processes, EaCl:LAC is a low-price, environmental friendly and biocompatible reagent. The specific ABE yields per pretreated and raw biomass of this process were calculated to be 208 and 87.4 g·kg–1, second only to that of corncob pretreated by 0.5 M NaOH [48]. In view of its low price and high biocompatibility, EaCl:LAC is more efficient and promising than traditional ionic liquids such as [Bmim][Cl], and DESs such as ChCl:FA and ChCl:FA:AA. There is no need to add other reagents which are commonly used in combinatorial pretreatments, such as NaOH or Na2CO3 [3,41]. Moreover, EaCl:LAC could also be facilely recycled by filtration. The hydrolysates with high titers of reducing sugars could presumably be applied in the biosynthesis of other biofuels, amino acids and high-valued natural products. Consequently, this study provides a promising reagent for significantly reducing the recalcitrance of lignocellulosic biomass, and an economic cellulase recycling process for biobutanol production.