Abstract
High biomass loading is a key technique to reduce the pretreatment cost of lignocellulosic biomass. In this work, various biomass species such as bagasse, erianthus, cedar, and eucalyptus were pretreated using an ionic liquid, 1-ethyl-3-methylimidazolium acetate, at different biomass loadings, particularly focusing on a high loading region. Cellulose structural changes in pretreated biomass were investigated via X-ray scattering and 13C solid-state nuclear magnetic resonance (SSNMR) spectroscopy. The structural behaviors roughly fell into two categories, corresponding to either grassy (bagasse and erianthus) or woody (cedar and hardwood) biomass. The grassy biomass gradually transformed from cellulose-I to cellulose-II in a monotonic manner against the biomass loading. In contrast, the transformation in the woody biomass occurred abruptly as solids was decreased within the high loadings range (50 wt% to 33 wt%). Below 33 wt%, a reformation of cellulose-I from cellulose-II proceeded readily. In terms of cellulose crystallinity, erianthus as well as bagasse showed a minimum value at 25 wt% loading, whereas the crystallinity for the woody biomass did not possess such a clear minimum. Acid hydrolysis of these pretreated biomass was also conducted and the relationship between the reactivity and the cellulose structural changes was discussed.
Download PDF
Full Article
Cellulose Structural Change in Various Biomass Species Pretreated by Ionic Liquid at Different Biomass Loadings
Takatsugu Endo,a Shunsuke Fujii,b Ei Mon Aung,b, c Kosuke Kuroda,b Takayuki Tsukegi,d Kazuaki Ninomiya,e and Kenji Takahashi b,*
High biomass loading is a key technique to reduce the pretreatment cost of lignocellulosic biomass. In this work, various biomass species such as bagasse, erianthus, cedar, and eucalyptus were pretreated using an ionic liquid, 1-ethyl-3-methylimidazolium acetate, at different biomass loadings, particularly focusing on a high loading region. Cellulose structural changes in pretreated biomass were investigated via X-ray scattering and 13C solid-state nuclear magnetic resonance (SSNMR) spectroscopy. The structural behaviors roughly fell into two categories, corresponding to either grassy (bagasse and erianthus) or woody (cedar and hardwood) biomass. The grassy biomass gradually transformed from cellulose-I to cellulose-II in a monotonic manner against the biomass loading. In contrast, the transformation in the woody biomass occurred abruptly as solids was decreased within the high loadings range (50 wt% to 33 wt%). Below 33 wt%, a reformation of cellulose-I from cellulose-II proceeded readily. In terms of cellulose crystallinity, erianthus as well as bagasse showed a minimum value at 25 wt% loading, whereas the crystallinity for the woody biomass did not possess such a clear minimum. Acid hydrolysis of these pretreated biomass was also conducted and the relationship between the reactivity and the cellulose structural changes was discussed.
Keywords: Ionic liquid; Pretreatment; High loading; Bagasse; Erianthus; Cedar; Eucalyptus; X-ray; NMR; Acid hydrolysis
Contact information: a: Department of Molecular Chemistry and Biochemistry, Faculty of Science and Engineering, Doshisha University, 1-3 Tatara Miyakodani, Kyotanabe-Shi, Kyoto-fu 610-0394, Japan; b: Faculty of Natural System, Institute of Science and Engineering, Kanazawa University, Kakuma-Machi, Kanazawa 920-1192, Japan; c: Department of Chemical Engineering, Mandalay Technological University, Patheingyi Township, Mandalay, Myanmar; d: Innovative Composite Materials Research and Development Center (ICC), Kanazawa Institute of Technology, 2-2 Yatsukaho, Haku-San 924-0838, Japan; e: Institute for Frontier Science Initiatives, Kanazawa University, Kakuma-Machi, Kanazawa 920-1192, Japan; *Corresponding author: ktkenji@staff.kanazawa-u.ac.jp
INTRODUCTION
Ionic liquid (IL) pretreatment is a newly emerged and efficient technique to deconstruct the recalcitrance of lignocellulosic biomass (Wang et al. 2012; Brandt et al. 2013). The ILs are salts that exist in liquid state at room temperature. Some specifically designed ILs are feasible to dissolve cellulose (Swatloski et al. 2002) and even lignocellulose (Kilpeläinen et al. 2007). After removing IL from the lignocellulose solution, the regenerated lignocellulose becomes much less recalcitrant. Ionic liquid pretreatments have attracted much attention because of their high efficiency as well as their negligible vapor pressure and flammability. However, the high costs of IL compared to conventional organic solvents is one of the main impediments for their use (Reddy 2015; Baral and Shah 2016).
A high biomass loading during IL pretreatment is a direct and feasible solution. Normally, ca. 5 wt% of biomass loading is employed (Brandt et al. 2013), which makes the lignocellulose/IL mixture a viscous solution. Much effort has been devoted to establish higher lignocellulose loadings (Wu et al. 2011; Cruz et al. 2013; da Silva et al. 2013; Haykir and Bakir 2013; Li et al. 2013; Ninomiya et al. 2013; Qiu et al. 2014; Uju et al. 2013; Zhang et al. 2014; Wang et al. 2015; Ren et al. 2016) compared to the normal loading amount. In these reports, biomass loading amounts higher than 5 wt% were successfully and consistently used, e.g. 30 wt%, which did not decrease the subsequent hydrolysis efficiency. Compared to 5 wt%, a concentration of 30 wt% means that loading times more than eight times higher is feasible when the same amount of IL is used, providing a robust cost reduction. Among these reports, only two studies intriguingly stated that when grassy biomass was used with a certain IL pretreatment condition, it lowered the crystallinity of regenerated cellulose at a higher biomass loading to some extent (Cruz et al. 2013; Zhang et al. 2014). Because the crystallinity of cellulose is considered a major contributor to the recalcitrance effect, the hydrolysis reactivity could be enhanced at higher loading; however, such a phenomenon was not observed.
Recently, the authors addressed this inconsistency using microcrystalline cellulose as a model compound (Endo et al. 2017). It was observed that compared to ca. 5 wt% loading, the crystallinity of cellulose declined when 25 wt% to 40 wt% loading was used. A subsequent hydrolysis catalyzed by an enzyme, the common hydrolysis method employed by many other research groups, was not enhanced at the high loadings. In contrast, another hydrolysis route, viz. acid hydrolysis, proceeded 1.6 times more efficiently in accordance with the crystallinity change. The structural and morphological investigations revealed that the difference between these catalysts could be attributed to their size. At higher loadings, the pores that formed in the regenerated cellulose were smaller, and hence the access of the enzyme with large molecular size into inner cellulose was restricted. The concept that higher cellulose loading in an IL pretreatment causes lower crystallinity of cellulose and that this consequently enhances the acid hydrolysis efficiency was successfully applied to bagasse (up to 33 wt%) as lignocellulosic biomass (Aung et al. 2018), where the cellulose is encapsulated by the other biopolymers.
To expand this concept to other types of biomass, especially woody biomass, in this work the authors focused on the investigations of cellulose structural changes in pretreated biomass in different biomass species. Bagasse is an agricultural residue and a grassy biomass. The three lignocellulosic materials examined in this study were erianthus (grass, but not an agricultural residue), cedar (softwood), and eucalyptus (hardwood). The cellulosic structural changes in biomass, including cellulose crystallinity, a major factor that governs cellulose recalcitrance, were investigated in detail viawide angle X-ray scattering (WAXS) and 13C solid-state nuclear magnetic resonance (SSNMR) spectroscopy against various biomass loadings ranging from 5 wt% to 50 wt%. Acid hydrolysis was also performed and the relationship between glucose yield and cellulose structural changes was discussed. It should be noted that the data for bagasse were taken from the authors’ previous paper (Aung et al.2018), whereas the WAXS peak analyses and the experiments at 25 wt% loading of bagasse were from the current study.
EXPERIMENTAL
Materials
IL-pretreatment
An IL, 1-ethyl-3-methylimidazolium acetate ([Emim][OAc]) that is the most representative IL for lignocellulose pretreatments, was purchased from IoLiTec (> 95% purity; Ionic Liquid Technologies GmbH, Heilbronn, Germany). The IL was vacuum-dried overnight at 50 °C before use. The biomass materials, bagasse, cedar, and eucalyptus were purchased from the Toyota Motor Corporation (Miyoshi, Japan), while the National Agriculture and Food Research Organization (Ibaraki, Japan) kindly supplied the erianthus sample. Each lignocellulosic sample was first ground by a mill and then sieved to obtain a powder with a particle diameter of 75 m to 125 m. The powdered sample was vacuum-dried at 70 °C for 3 h to eliminate adsorbed moisture. The chemical composition of the lignocellulosic sample was then analyzed according to the National Renewable Energy Laboratory (NREL) method (Sluiter et al. 2012) as described in the previous paper (Aung et al. 2018). The measurements were recorded in triplicate to display the experimental error as one standard deviation, and the data are shown in Fig. 1.
Fig. 1. Chemical composition of biomass samples
In the IL-pretreatment process, 0.1 g of lignocellulose and various amounts of [Emim][OAc] were mixed using a spatula for approximately 5 min at room temperature. The mixture was then heated at 130 °C for 3 h, as an optimized condition for bagasse (Aunget al. 2018). After the mixture was cooled down to room temperature, the IL was removed by washing with distilled water five times. Various lignocellulose-to-IL ratios of 1:1, 1:2, 1:3, 1:5, 1:10, and 1:20 (w/w) correspond to lignocellulosic concentrations of 50, 33, 25, 17, 9, and 5 wt%, respectively.
Methods
Structural assessments
The WAXS measurements were carried out using a NANO Viewer (IP system; Rigaku Corporation, Tokyo, Japan). The CuK radiation source ( = 0.154 nm) was employed at an applied voltage of 40 kV and a filament current of 30 mA. The lignocellulose samples were freeze-dried and then first packed into a Teflon cell (1-mm-thick) and sandwiched between two sheets of Kapton film. The samples irradiation times were 1 h to 4 h. The crystallinity index (CrI) was obtained using the Segal method (Segal et al. 1959) as shown in Eq. 1,
(1)
where Imax and Iam represent the X-ray intensities at the maximum (crystalline) and 2θ = 18.5° (amorphous), respectively. To evaluate peak position, a Gauss function was fitted to observed peaks.
The SSNMR measurements were carried out using a JNM-ECX 500II spectrometer (11.75 T; JEOL, Tokyo, Japan). The freeze-dried lignocellulose was placed in a 3.2-mm zirconia rotor (49 µL) to perform 13C cross-polarization/magic-angle spinning (CP/MAS) measurements with an MAS speed of 10 kHz. These spectra were collected using a contact time of 2 ms and a recycle delay of 3 s. Each 13C spectrum was the accumulation of 4096 to 16384 scans. The 13C chemical shifts were determined by reference to adamantane (at δ29.5 ppm for CH). The crystallinity index from the SSNMR measurements (CrINMR) was estimated based on the fraction of order (at δ 86.8 ppm to 92.4 ppm) to the total area of order and disorder (at δ 80.4 ppm to 86.8 ppm) for the C4 peak of cellulose. The fractions for the tg (trans-gauche), gt (gauche-trans), and gg (gauche-gauche) conformers of the C6 carbon site of cellulose were estimated in the regions of δ 64.2 ppm to 66.0 ppm, 61.7 ppm to 64.2 ppm, and 59.1 ppm to 61.7 ppm, respectively.
Acid hydrolysis
The procedure for acid hydrolysis of lignocellulose was followed from the authors’ previous study (Endo et al. 2017). Approximately 3 mL of 0.5 M H2SO4 solution was placed in a sealed cell with 0.1 g lignocellulose and heated at 130 °C with a roller oven (RDV-TM2; SAN-AI Kagaku, Aichi, Japan). After heating for 4 h, the reaction was halted by placing the sample in an ice bath followed by centrifuging the hydrolysate for further analysis. The released glucose in the hydrolysate was detected by high pressure liquid chromatography (HPLC; Shimadzu Corporation, Kyoto, Japan) equipped with a refractive index detector and a CARBOSep CHO-682 column (Tokyo Chemical Industry Co., Tokyo, Japan). The analysis conditions in HPLC were as follows: the volume of the injected sample was 20 µL, the column was operated at 70 °C, and 0.0085 N H2SO4 solution was used in the mobile phase with a flow rate set at 0.4 mL/min. The hydrolysate measurements were taken in triplicate to display experimental error as one standard deviation.
Thermogravimetric analysis
Thermogravimetric analyses were performed with DTG-60AH (Shimadzu Corporation, Kyoto, Japan). Dried untreated biomass (4 to 5 mg) was placed in a platinum pan. The pan was heated to 1000 °C with the scan rate of 10 °C/min.
RESULTS AND DISCUSSION
WAXS Measurements
Figure 2(a) shows the WAXS patterns of untreated lignocellulose in addition to theoretical reference patterns of cellulose-I and cellulose-II (Nishiyama et al. 2002; French 2014). The cellulose-I and cellulose-II possess three intense peaks at 2θ values 14.9° (10), 16.7° (110), 23.0° (200), and 12.2° (10), 19.9° (110), and 22.1° (020), respectively. Native cellulose takes the cellulose-I form, while it tends to transform to the more thermodynamically stable cellulose-II structure after IL pretreatments (Cheng et al. 2011; Samayam et al.2011). All four biomass samples exhibited similar WAXS patterns that resemble cellulose-I. However, the peak for the (200) plane was shifted to a lower angle probably because of the presence of hemicellulose and lignin (Sun et al. 2014). It was noted that small sharp peaks sometimes appeared (e.g., one at 27° for bagasse) that were assigned to some inorganic crystallized components contained in lignocellulose (Konsolakis et al. 2015).
The WAXS patterns of the lignocellulose in different loading amounts are displayed in Figs. 2(b through d). The behaviors with respect to loading amount in the grassy biomass (bagasse and erianthus) samples differed distinctively from the woody biomass (cedar and eucalyptus). The patterns for the grassy biomass gradually broadened with decreasing loading amount from 50 wt% to 5 wt%, and at the most dilute condition only one broad peak appeared centered at ca. 21°. In contrast, the WAXS patterns for the woody biomass abruptly changed between the 50 wt% and 33 wt% loading amounts. Even at the lower loading amounts, unlike the grassy biomass, a shoulder peak around 14° remained. This peak position was lower than the original (see Fig. 3(b)) sample, indicating the presence of the cellulose-II structure. Before proceeding with further discussion, it should be noted that the chemical composition of IL-pretreated biomass is not notably altered by changing the loading amount (Wu et al. 2011; da Silva et al. 2013; Ninomiya et al. 2013; Aung et al.2018). Hence, the changes in WAXS pattern were attributed to the cellulose crystalline states.
Fig. 2. (a) WAXS patterns of various untreated biomass and theoretical powder patterns of cellulose I and cellulose II (French 2014; Nishiyama et al. 2002). The colors used are consistent with the following figures, i.e., black for bagasse, red for erianthus, blue for cedar, and green for eucalyptus; (b through e) WAXS patterns of untreated (UT) and pretreated biomass at various loading amounts.
To gain further insights for the peak change behaviors, peak analyses were performed. The peak position changes of the strong peak around 21° are shown in Fig. 3(a), and those of the shoulder peak for cedar and eucalyptus are shown in Fig. 3(b). These results highlighted the difference between the grassy and woody biomass samples. The intense peak was shifted gradually to low 2θ values from 50 wt% to 5 wt% loadings in the grassy biomass. These results were similar to those previously observed (Cheng et al. 2011), and were mainly attributable to the gradual change from cellulose-I to cellulose-II. The behavior for the woody biomass was more complex. After the abrupt change between 50 wt% and 33 wt% loadings, the peak position was shifted to a large angle with decreased biomass loadings. Because a large change was also observed for the shoulder peak from 50 wt% to 33 wt% (Fig. 3(b)), it indicated the transformation into the cellulose-II structure. This result strongly suggested that a large portion of cellulose-I was transformed to cellulose-II at the 33 wt% loading. At lower loadings, the peak position was gradually returned to higher angle (Fig. 3(a)), which was more prominent in eucalyptus than cedar. This behavior is rather unique, and a plausible explanation was the reformation of the cellulose-I structure at the lower loadings. Normally thermodynamically stable cellulose-II is formed first from cellulose-I via an amorphous state in IL pretreatments (Cheng et al.2011; Samayam et al. 2011). However, there was a high energy barrier in this transformation due to the large difference in cellulose chain alignment; parallel for cellulose-I and antiparallel for cellulose-II. Then, occasionally the amorphous domains of cellulose dissolved in an IL return to cellulose-I after the IL is removed by washing (Lucas et al. 2011). It was noted that the large peak position shifting observed in eucalyptus may not have been explained only by the reformation of cellulose-I. The lattice expansion and shrinking of cellulose crystal can be involved in the peak shifting. The authors will discuss the crystal structural changes again in the SSNMR section, which support the abovementioned explanations.
Fig. 3. Shifts of (a) the strong WAXS peak and (b) the weak shoulder peak. The inset shows the second derivative of the WAXS pattern of the 5 wt% eucalyptus sample after pattern smoothing operations (box smoothing), displayed as an example. The negative peak, indicated by an arrow, demonstrates the existence of the shoulder peak. (c) CrI estimated from the WAXS measurements with the Segal method.
The CrI of cellulose is considered to govern the cellulose accessibility, which is a factor of concern in this study. The CrIWAXSvalues were estimated from the Segal method and are displayed in Fig. 3(c). As the authors previously reported, the CrI of pretreated bagasse was lower at the higher loadings of 17 to 33 wt% than the conventional 5 wt% loading (Aung et al. 2018). The cellulose crystalline structure was mostly deconstructed even at the high loadings, which means that the majority of cellulose exists in the amorphous state even in the highly lignocellulose-loaded mixture (Endo et al. 2016). When the IL was removed from the mixture by washing, the recrystallization to cellulose-II was inhibited because of the restricted movements of the cellulose chain by a highly viscous environment (Aung et al. 2018; Endo et al. 2017). The CrIWAXSvalues for erianthus as the grassy biomass also followed the same trend, i.e., the lowest CrIWAXS appeared at 25 wt% biomass loading. This also may be the case for the woody biomass, but it was ambiguous for cedar. The result for eucalyptus needed more caution as the Segal method was employed which simply uses two peak intensities to estimate the CrIWAXS. The peak position shifts affect the CrIWAXS values, particularly for eucalyptus. Lowering the strongest peak made Imax large which caused an underestimation of the CrIWAXS values.
SSNMR Measurements
To ensure the changes in the CrI behaviors, 13C SSNMR experiments were performed. From the NMR spectra, the crystallinity of cellulose was estimated at the molecular level. These values would be less sensitive to the crystalline states (e.g., allomorphs and lattice size changes) as well as the presence of hemicellulose and lignin.
Fig. 4. 13C CP/MAS spectra of untreated (UT) and pretreated (a) bagasse, (b) erianthus, (c) cedar, and (d) eucalyptus at different loading amounts
Figure 4 shows the 13C cross-polarization magic angle spinning (CP/MAS) spectra of the untreated and pretreated lignocellulose sample. The strong peaks found at δ 60 to 110 ppm were mainly from cellulose. The C-4 carbon in the glucose unit splits into two regions, as assigned to ordered and disordered (i.e., in amorphous or surface domains) conformations (Horii et al. 1982; Larsson et al. 1997). The ordered component relative to the total of the ordered and disordered components was used as the CrI estimated from NMR (CrINMR). The C-6 site also splits into three, which in turn provides useful information at the molecular level. The different hydroxymethyl group conformations of cellulose (tg, gt, and gg) divide the C-6 carbon peak into three regions (Horii et al. 1983; Fernandes et al.2011). The tg conformer is mainly present in the crystalline core of cellulose-I (Nishiyama et al. 2002, 2003; Wang et al. 2016), whereas cellulose-II has the gt conformation in its core (Langan et al. 2001). The gt and gg conformers are both present on the crystal surfaces (Horii et al. 1983; Viëtor et al. 2002; Newman and Davidson 2004), but the gg conformer exists predominantly in the amorphous domain (Mazeau and Heux 2003; Newman and Davidson 2004; Idström et al.2016).