Activation of lignocellulosic biomass for higher sugar yields using aqueous ionic liquid at low severity process conditions

Background Concerns around greenhouse gas emissions necessitate the development of sustainable processes for the production of chemicals, materials, and fuels from alternative renewable sources. The lignocellulosic plant cell walls are one of the most abundant sources of carbon for renewable bioenergy production. Certain ionic liquids (ILs) are very effective at disrupting the plant cell walls of lignocellulose, and generate a substrate that is effectively hydrolyzed into fermentable sugars. Conventional ILs are relatively expensive in terms of purchase price, and the most effective imidazolium-based ILs also require energy intensive processing conditions (>140 °C, 3 h) to release >90 % fermentable sugar yields after saccharification. Results We have developed a highly effective pretreatment technology utilizing the relatively inexpensive IL comprised tetrabutylammonium [TBA]+ and hydroxide [OH]− ions that generate high glucose yields (~95 %) after pretreatment at very mild processing conditions (50 °C). The efficiency of [TBA][OH] pretreatment of lignocellulose was further studied by analyzing chemical composition, powder X-ray diffraction for cellulose structure, NMR and SEC for lignin dissolution/depolymerization, and glycome profiling for cell wall modifications. Glycome profiling experiments and computational results indicate that removal of the noncellulosic polysaccharides occurs due to the ionic mobility of [TBA][OH] and is the key factor in determining pretreatment efficiency. Process modeling and energy demand analysis suggests that this [TBA][OH] pretreatment could potentially reduce the energy required in the pretreatment unit operation by more than 75 %. Conclusions By leveraging the benefits of ILs that are effective at very mild processing conditions, such as [TBA][OH], lignocellulosic biomass can be pretreated at similar efficiency as top performing conventional ILs, such as 1-ethyl-3-methylimidazolium acetate [C2C1Im][OAc], but at much lower temperatures, and with less than half the IL normally required to be effective. [TBA][OH] IL is more reactive in terms of ionic mobility which extends removal of lignin and noncellulosic components of biomass at the lower temperature pretreatment. This approach to biomass pretreatment at lower temperatures could be transformative in the affordability and energy efficiency of lignocellulosic biorefineries. Electronic supplementary material The online version of this article (doi:10.1186/s13068-016-0561-7) contains supplementary material, which is available to authorized users.


Background
Lignocellulosic biomass is an abundant renewable resource primarily composed of cellulose, lignin, and hemicellulose that form a complex composite structure [1][2][3][4]. The recalcitrance of this complex composite poses a significant barrier to economical, chemical, or biological conversion technologies that can convert the fermentable sugars present in lignocellulose into advanced biofuels and renewable chemicals [5]. Several physical and/or chemical pretreatment processes have been implemented to reduce the recalcitrance of lignocellulosic materials and improve their utilization [6][7][8].
Conventional pretreatments, such as those that use concentrated or dilute acids and bases, are only effective in producing a substrate capable of generating high fermentable sugar yields using severe process conditions (~120-200 °C). After pretreatment, the recovered substrate is saccharified using enzyme mixtures at 40-80 °C. There is an intricate interplay between the type of pretreatment and fermentable sugar yields achieved (Fig. 1) [6,[9][10][11][12][13][14][15]. These higher temperature process technologies increase the energy required, and thereby increase production costs [7]. Generally, temperature of pretreatment process has been set around the range of the glass transition temperature of lignin, thereby impacting the physicochemical properties of lignin and cellulose [16], hemicellulose hydrolysis [17], and cellulose digestion [18].
Certain ionic liquids (ILs) are able to dissolve either lignocellulosic materials or one of its main constituents, such as cellulose, hemicellulose, or lignin [19]. The IL 1-ethyl-3-methylimidazolium acetate ([C 2 C 1 Im][OAc]) based pretreatment process typically requires temperatures above 140 °C for 3 h reaction time to be effective and typically use pure IL as the pretreatment solvent [10,12,[20][21][22]. There have been recent efforts focused on the discovery and demonstration of ILs for the pretreatment/fractionation of lignocellulosic materials at less severe process conditions [23][24][25]. [OH] based ILs with [C 2 C 1 Im] were used for biochemical synthesis, such as sugars [26], biodiesel [27] and 5-hydroxymethylfurfural [26,28]. Studies on quaternary ammonium cations, such as tetrabutylammonium fluoride trihydrate ([TBA]F) [29], tetraethylammonium chloride ([TEA]Cl) [30], and tetrabutylammonium hydroxides ([TBA][OH]) [31] with co-solvents, were reported to dissolve cellulose rapidly at low temperatures. This rapid dissolution of cellulose at low temperatures has been hypothesized to occur due to the strong proton accepting capacity of the anion, even in the presence of water that weakens the association of the hydrogen bonding network and destabilizing the cellulose microstructure.
Recent studies using tetrabutylammonium acetate ([TBA][OAc]) with dimethyl sulfoxide (DMSO) and crown ether (18-crown-6) demonstrated the feasibility of 8 wt% cellulose dissolution within 5 min at 40 °C [32]. Tetrabutylphosphonium hydroxides ([TBP][OH]) containing 30-50 wt% water can dissolve cellulose at 25 °C [33], and Ohno and co-workers recently reported the rapid (~5 min) dissolution of 15 [35]. Unfortunately, from a biorefinery perspective, all of these methods require multi-step treatments, use of co-solvents, water washes and have not been proven on a wide range of "real world" lignocellulosic biomass substrates, such as switchgrass, pine and eucalyptus. The possibility of IL pretreatment at lower operating temperatures may facilitate the development of more affordable and practical pretreatment processes with seamless biomass integrated conversion processes [3,13,36]. We report here that the relatively inexpensive [37] [TBA][OH] processing of lignocellulose can pretreat biomass to similar efficiency as top performing conventional ILs, such as [C 2 C 1 Im][OAc], but at much lower temperatures, and with less than half the IL normally required to be effective.

Compositional analysis and lignin fractionation
The compositional analysis of switchgrass before and after pretreatment is summarized in Table 1. Solid recovery refers to the mass percentage of biomass (dry weight) recovered from the original biomass load. Three of the major plant cell wall components of switchgrass (i.e., glucan, xylan, and acid-insoluble lignin), were monitored before and after the pretreatment. Untreated dry switchgrass contained 31.9 % glucan, 20.2 % xylan and 20.7 % acid-insoluble lignin. The pretreatment experiments were conducted at different conditions (i.e., 25 °C for 0.5, 1 and 3 h; 50 °C for 0.5, 1 and 3 h). The solid recovery was obviously decreased with increasing temperature and time. Using the conditions of 50 °C for 3 h, approximately 48 wt% of the biomass was recovered, of which 62 % was glucan, 12.7 % was xylan and 10.5 % was lignin. Based on the compositional change, the mass loss is caused by significant removal of lignin, xylan, and/or other soluble extractives. While the loss of glucan was approximately 6.5 wt%, the removal of xylan was significantly higher (~69.8 wt%). Also, the lignin removal during pretreatment process (~75.7 wt%) was comparable with our previous results for switchgrass with 49-87 % lignin removal after pretreatment with different types ILs at high temperature (140 °C) [21].
Cellulose crystallinity X-ray diffraction (XRD) studies were conducted to determine the changes in the crystalline vs. noncrystalline components (i.e., amorphous cellulose, hemicellulose and lignin) found in the switchgrass sample, and to monitor the structural changes in these polymers that occur during [TBA][OH] pretreatment. Commercial Avicel was used as cellulose standard to validate the results. Further, components isolated from the pretreatment condition (50 °C for 3 h) were utilized for cellulose crystallinity and lignin characterization studies.
Additional file 1: Fig. S1 shows the X-ray diffractograms of the untreated and pretreated switchgrass after processing at 50 °C for 3 h. The diffractogram obtained from the untreated switchgrass has two major diffraction peaks at 22.5° and 15.7° 2θ, characteristic of the cellulose I polymorph that corresponds to [002] and combined [101] + [10 1 ] lattice planes, respectively. The third small peak at 34.5° ([040] lattice plane) corresponds to 1/4 of the length of one cellobiose unit and arises from ordering along the fiber direction [38][39][40]. Obtained crystallinity index from the XRD patterns of the pretreated switch grass indicates that the low temperature pretreatment based on 40 [21,41]. This also reflected in the increase in the crystallinity index (CrI) of switchgrass from 67 to 76 % after pretreatment.

Sugar yields
The sugar yields are calculated based on the glucan or xylan present in the original biomass (converting pretreated biomass to original using the solid recovery data in Table 1). As shown in Table 2 [35]. Using the conditions (e.g., 50 °C, vacuum degree 0.1 MPa), it is expected that after lignin filtration, [TBA][OH] could be generated and reused for next run.

Lignin characterization
To examine the effect of [TBA][OH] pretreatment process on lignin, we carried out detailed lignin characterization studies using size exclusion chromatography (SEC) and 2D 13  isolated by adjusting the pH to 2-3. The isolated lignin was compared with enzymatic mild acid lignin (EMAL) extracted from switchgrass, as it is commonly believed to be a close representation of the 'native' switchgrass lignin. The EMAL lignin from switchgrass was isolated based on the procedure reported by Wu and Argyropoulos [42]. The elution profiles acquired by monitoring UV absorbance (λ = 280 nm) from SEC measurements of EMAL and the lignin isolated from the liquid stream (L 1 ) are depicted in Additional file 1: Fig. S2. Although the main elution peaks (M w = 1.0-10.0 kDa) for both EMAL and L 1 are comparable, a through comparison in the higher molecular weight region (M w > 10.0 kDa) region shows that L 1 has slightly lower molecular weight than EMAL, indicating small reduction of size. Apart from that L 1 shows greater low molecular weight tails (M w > 1.0 kDa) along with one new intense low molecular weight peak (M w = 322 Da), both these observations indicate more abundant lower molecular weight lignin fractions as compared to EMAL.
To understand the structural changes that occur in lignin during the pretreatment process, the isolated lignin was compared with EMAL and raw switchgrass using 2D 13 C-1 H heteronuclear single quantum coherence (HSQC) nuclear magnetic resonance (NMR) (Fig. 2). The cross peaks were assigned according to standards reported in the literature [43][44][45][46][47][48][49]. The structures in these HSQC spectra correspond to the color-coded structures depicted in Additional file 1: The HSQC spectrum of the cell wall of the untreated switchgrass shows that the β-aryl ether interunit linkages (A α , A β(H/G) , A β(S) , substructure A) are the predominant linkages in the lignin with small contributions of phenylcoumaran (β-5, substructure B), resinol (β-β, substructure C), and dibenzodioxocin (substructure D) linkages. The aliphatic region also exhibits two distinct peaks of 2-O-Ac-β-d-Xylp(R) (X2 2 ) and 3-O-Ac-β-d-Xylp(R) (X3 3 ) that represent major acetylated components of hemicelluloses. The aromatic region of the cell wall of the raw switchgrass indicates that the lignin is a S/G type lignin with a minor amount of H-type lignin containing p-coumarates (pCA) and ferulates (FA), which is consistent with previous literature reports [43,46,47,49]. The HSQC spectrum of the untreated switchgrass also shows the presence of tricin moieties (substructure T).
The HSQC spectrum of L 1 is shown in Fig. 2g-i. The weaker signal intensity of the A α interunit linkages suggest chemical changes in the β-aryl ethers during the [TBA][OH] pretreatment. The absence of dibenzodioxocin signal (δC/δH: 83.3/4.81 ppm) indicates that the lignin isolated from the liquid stream is more linear as compared to the branched lignin in the untreated switchgrass due to removal of the points of branching [50]. Additionally, the absence of X2 2 and X3 3 correlations suggest deacetylation of hemicellulose occurred more readily at C 2 /H 2 position. The anomeric regions of the untreated biomass and L 1 demonstrates a noticeable decrease of α-d-Glcp(R)/α-d-Xlyp(R), may be due to glycosidic bond cleavage and reduction in the degree of polymerization (DP) of hemicellulose during [TBA][OH] pretreatment. When compared with the switchgrass EMAL ( Fig. 2d-f ), L 1 has similar interunit traits in both aliphatic and aromatic regions of the HSQC spectra. The relative abundance of different interunit linkages in EMAL and L 1 is shown in Additional file 1: Fig. S4. In L 1 , the β-aryl ether interunit linkages decrease from 59 to 43 % as compared to the EMAL, with a relatively smaller decrease in both phenylcoumaran and resinol substructures. This result is in agreement with the SEC results, confirming reduction of the lignin size due to depolymerization during [TBA] [OH] pretreatment process. The absence of detectable dibenzodioxocin substructure suggests relatively linear lignin structure of both types of lignins. Tricin substructures were also detected in both EMAL and L 1 [51]. The SEC and 2D NMR studies suggested that L 1 has similar structure traits as switchgrass EMAL with a relatively smaller size. These results indicate that [TBA][OH] can very efficiently solubilize and partially depolymerize the lignin present in switchgrass.

Glycome profiling
Glycome profiling of untreated and [TBA][OH] pretreated switchgrass biomass was conducted to facilitate a  [TBA][OH] pretreatment process. Interestingly, prominent variations were observed in the abundances of noncellulosic glycan epitopes in oxalate and carbonate extracts from pretreated biomass samples in comparison to the untreated biomass. For instance, a considerably higher abundance of xylan epitopes (as indicated by the higher binding of xylan-3 through xylan-7 groups of mAbs that recognize both unsubstituted and substituted xylans) was observed in oxalate and carbonate extracts from [TBA][OH] pretreated switchgrass at 50 °C that indicates enhanced xylan extractability. As previous studies have noted, such enhanced extractability of xylan epitopes is indicative of structural changes in the cell walls that results in reduced recalcitrance [52]. Substantial differences in patterns of extractabilities of epitopes of pectic components were also observed between the untreated and [TBA][OH] pretreated switchgrass. For example, oxalate extracts from [TBA][OH] pretreated samples showed significantly higher abundance of pectic backbone epitopes (as indicated by the higher binding of homogalacturonan backbone-1 and rhamnogalacturonan-I backbone groups of mAbs). Pectic arabinogalactan and arabinogalactan epitopes (as indicated by the binding of RG-I/AG and AG groups of mAbs) were highly abundant in oxalate and carbonate extracts from untreated switchgrass, however, this abundance decreased in [TBA][OH] pretreated switchgrass samples (potentially due to enhanced glycan fragmentation). The chlorite extraction step employed in sequential extraction breaks up and removes lignin to release any lignin-associated polysaccharides into the extract. Chlorite extracts from pretreated switchgrass under [TBA][OH] at 50 °C conditions contained higher abundance of xylan epitopes as compared to untreated samples indicating ILs perturbing effect on potential lignin-xylan associations. Overall, glycome profiling studies revealed the major structural modifications induced by [TBA][OH] at the mild pretreatment condition on switchgrass biomass resulting in efficient hydrolysis of cellulose to its constituent sugars.

Computational modeling
To understand the molecular level forces on the biomass dissolution at low temperature using [OH] that the distinct separation of positive (blue) and negative regions (red) play a dominant role by influencing strong ionic interactions, electrostatics, and hydrophobic interactions with biomass components [53].
We evaluated the influence of the anion and cation interactions on biomass dissolution by performing quantum chemical calculations of [OH] − , [TBA] + and [TBA] [OH] interacting with a model dilignol and cellobiose compounds (Fig. 5). From the calculated IEs, it was found that [OH] − interacts with dilignol and cellobiose more strongly than [TBA] + cation, and the IE of [OH] − with cellobiose is slightly higher than that of its interactions with dilignol. In the case of [TBA] + interactions, our calculations show a slightly higher IE for cellobiose than for dilignol, which are most probably due to hydrophobic interactions. Interestingly, IE strength of ion pair complexes with biomass components are more favored toward dilignol than cellobiose.  [7]. It has been reported that alkaline ions could instigate the progression of following steps involving (i) cellulose swelling; (ii) internal surface area enhancement; (iii) changes in cellulose crystallinity; (iv) hemicellulose removal; (v) reducing the lignin and carbohydrate association; and (vi) disrupting the lignin structure by breaking its glycosidic ether bond. Hence, lignin cannot further act as a protective shield to the cellulose after lignin dissolution, consequently making cellulose more susceptible for degradation.
We then sought to determine what properties and interactions are dominant at these lower temperatures. Ionic conductivity, ion mobility, and viscosity of ILs are the key factors and vary based on ion type, size, charge, and temperature [54]. The higher pH (~14) of [TBA] [OH] significantly influences the lignin removal [55]. The low temperature mechanism of biomass dissolution is particularly dependent on the ion exchange and contribution from ionic interactions between the biomass and ILs. Optimized geometries of lignin with [C 2 C 1 Im][OAc]   [57]. Overall, four scenarios were constructed using these two ILs and two different biomass loadings (i.e., 10 and 20 %). Energy demand calculations revealed more than 75 % reduction in steam requirement during the low temperature pretreatment process to liberate carbohydrates with reduced energy input (Fig. 6) [36]. As shown in the Fig. 6 [OH], this study opens up an avenue for novel process designs that could significantly enhance the energy efficiency and affordability of the biorefinery by overcoming the temperature mismatch of pretreatment and saccharification unit operations.

Biomass pretreatment
A 10 % (w/w) biomass solution was prepared by combining 1 g of switchgrass with 9 g of IL in a 25 mL tube reactor. The reactor was heated in an oil bath to the desired temperature and stirred at 150 rpm with a magnetic stir bar for 3 h. All pretreatment reactions were conducted in duplicate. Following pretreatment, 30 mL of deionized (DI) water was slowly added to the biomass/IL slurry with continued stirring. The mixture was transferred to 25 mL Falcon tubes and centrifuged at high speed (14,000 rpm) to separate solids. The pretreated biomass was further washed with 4 × 30 mL of DI water to remove any residual IL. The solids were lyophilized and stored at 4 °C for analysis.

Compositional analysis
Compositional analysis of switchgrass before and after pretreatment was performed using NREL acidolysis protocols (LAP) LAP-002 and LAP-005. Briefly, 200 mg of biomass and 2 mL 72 % H 2 SO 4 were incubated at 30 °C while shaking at 300 rpm for 1 h. The solution was diluted to 4 % H 2 SO 4 with 56 mL of DI water and autoclaved for 1 h at 121 °C. The reaction was quenched by placing samples into an ice bath before removing the biomass by filtration. The filtrate was neutralized with CaCO 3 and monomeric sugars were determined from the filtrate by Agilent HPLC 1200 Series equipped with a Bio-Rad Aminex HPX-87P column and a refractive index detector (aqueous mobile phase, 0.6 mL/min, column temperature 85 °C). The injection volume was 10 µL with a run time of 25 min. Acid-insoluble lignin was quantified gravimetrically from the solid after heating overnight at 105 °C (the weight of acid-insoluble lignin + ash) and then 575 °C for at least 6 h (the weight of ash).

Enzymatic saccharification
Enzymatic saccharification of pretreated and untreated biomass was carried out using commercially available enzymes, Cellic ® CTec2 and HTec2 from Novozymes, at 50 °C, pH 5.5, and rotation speed of 150 rpm in a rotary incubator (Enviro-Genie, Scientific Industries, Inc.). All reactions were conducted at 10 % biomass loading by placing 500 mg of biomass (dry weight) in a 25 mL centrifuge tube. The pH of the mixture was adjusted to 5.5 with 50 mM sodium citrate buffer (pH 4.8) supplemented with 0.02 % NaN3 to prevent microbial contamination. The total reaction volume (5 mL) included a total protein content of 10 mg protein/g biomass (before pretreatment). The ratio of CTec2:HTec2 mixtures were held constant at 9:1 for all reactions. Reactions were monitored by centrifuging 50 µL aliquots of supernatant (5 min, 14,000 rpm) at specific time intervals and measuring monomeric sugar concentrations by HPLC as described previously.

X-ray diffraction (XRD)
The raw and pretreated biomass/Avicel were dried and characterized with powder X-ray diffraction (PXRD). The XRD analysis were performed on a PANalytical Empyrean X-ray diffractometer equipped with a PIXcel 3D detector and operated at 45 kV and 40 kA using Cu Kα radiation (λ = 1.5418 Å). The patterns are collected in the 2θ range from 5° to 60° with a step size of 0.039° and the exposure time of 300 s. A reflection-transmission spinner was used as a sample holder and the spinning rate was set at 8 rpm throughout the experiment. Crystallinity index (CrI) was determined by Segal's method [58].

Size exclusion chromatography (SEC)
The molecular weight distribution of lignin was investigated using a gel permeation chromatography (GPC).
The lignin was acetylated with pyridine and acetic anhydride following a previously published procedure [59]. The acetylated lignin was dissolved in tetrahydrofuran (THF) with a concentration of 1 g/L. GPC analysis was performed using a Tosoh Ecosec HLC-8320 GPC equipped with a refractive index (RI) and diode array detector (DAD) detector. Separation was achieved with an Agilent PLgel 5 μm Mixed-D column at 35 °C using a mobile phase of THF at a flow rate of 1.0 mL/min. The GPC standards, which contained polystyrene ranging from 162 to 29,150 g/mol, were purchased from Agilent and used for calibration. Absorbance of materials eluting from the column was detected at 280 nm (UV). The enzymatic mild acidolysis lignin (EMAL) process was used to extract lignin from switchgrass and it was used as a control.

Computational details
The geometry optimizations of [TBA] cations and hydroxide anions, cellobiose, lignin dimer model (dilignol with β-O-4 linkage between two arene rings) were performed using density functional theory (DFT) with the M06-2X hybrid exchange-correlation functional and the 6-311 ++G(d, p) basis set. Frequency calculations were carried out to verify that the computed structures corresponded to energy minima. The most stable isolated cation/anion and their IL complexes obtained from our calculations are herein described. Several complexes of anions and cations interacting with cellobiose and dilignol (guided by the electrostatic potentials) were constructed and optimized at M06-2X/6-31G (d, p) basis set. The most stable complexes of cation and anion with cellobiose and dilignol were used to calculate interaction energies (IEs) at M06-2X/6-311 ++G(d, p) level using the supermolecular approach, where E Complex refer to the energies of cation and anion pair (for IL), anion or cation with biomass components, anion and cation with biomass complexes, respectively, and E i refer to the energies of monomers. The results were corrected for basis set superposition error (BSSE) following the procedure adopted by Boys and Bernardi [60]. All quantum chemical calculations were performed using the Gaussian 09 suite of programs [Frisch et al.
[Alcohol Insoluble Residues (AIR)], sequential extractions of AIR were carried out as previously described [62,63]. Plant cell wall glycan-directed monoclonal antibodies (mAbs) were from laboratory stocks (CCRC, JIM and MAC series) at the Complex Carbohydrate Research Center (available through CarboSource Services; http:// www.carbosource.net) or were obtained from BioSupplies (Australia) (BG1, LAMP). Supporting information on mAbs [64] used in this study can be found in the Supplementary Information Table S1, including the link to WallMabDB (http://www.wallmabdb.net) that provides detailed information for each antibody. : tetrabutylphosphonium hydroxide; XRD: X-ray diffraction; EMAL: enzymatic mild acid lignin; HSQC: heteronuclear single quantum coherence; NMR: nuclear magnetic resonance; SEC: size exclusion chromatography; GPC: gel permeation chromatography; BSSE: basis set superposition error.