Investigation of Xylose Reversion Reactions That Can Occur during

Oct 16, 2013 - ABSTRACT: Xylose reversion reactions to form xylooligomers represent a potentially important mechanism of sugar loss during dilute acid...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/EF

Investigation of Xylose Reversion Reactions That Can Occur during Dilute Acid Pretreatment Heidi M. Pilath,*,† William E. Michener,*,† Rui Katahira,† Ashutosh Mittal,‡ Jared M. Clark,† Michael E. Himmel,‡ Mark R. Nimlos,† and David K. Johnson‡ †

National Bioenergy Center, National Renewable Energy Laboratory, Golden, Colorado, 80401 United States Biosciences Center, National Renewable Energy Laboratory, Golden, Colorado, 80401 United States



S Supporting Information *

ABSTRACT: Xylose reversion reactions to form xylooligomers represent a potentially important mechanism of sugar loss during dilute acid pretreatment of biomass. We have conducted a study to identify the products that result from these reactions and to determine the kinetics of their formation. A major obstacle is that there are few commercial standards available for xylose disaccharides, which are essential for the identification and quantification of the xylose reversion products formed during these reactions. To overcome this obstacle, we have used GC/MS and NMR analysis of xylose disaccharides isolated by preparative HPLC to identify the reaction products. At the xylose concentration we used (300 g L−1), only xylose disaccharides were observed. As with glucose reversion reactions [Pilath, H. M.; et al. J. Agric. Food Chem. 2010, 58, 6131], the disaccharides contained linkages that involved the anomeric carbon atom of one of the sugar monomers. Eight out of the nine possible disaccharides, including alpha and beta anomers, were observed. Whereas the GC/MS allowed for the identification of the linkages, NMR was needed to distinguish between the α and β isomers of the disaccharides. The kinetics of combined xylose disaccharide formation was measured using HPLC. Arrhenius parameters for the rates of disaccharide formation were calculated by fitting the data to a simple model.



INTRODUCTION

reversion reaction is bimolecular, one could expect a secondorder reaction in terms of xylose concentration. As a result, this reaction becomes more important as the process is conducted at higher substrate loadings. Because it is more economical to conduct biomass conversion at high solids loading, the concentrations of sugar solutions produced during pretreatment can be high. At solid loadings greater than 20% (200 g L−1), a concentration of about 45.5 g L−1 of xylose could be achieved. With glucose, these sugar concentrations can lead to reversion product yields >12%.1 The yield of reversion products from xylose has not been well characterized, but should be similar to glucose, which has been studied more frequently and in greater detail.1,5

Pretreatment is an essential step in the conversion of biomass to renewable fuels, leading to increased enzyme accessibility for the subsequent hydrolysis of the cellulose. By treating biomass in solutions of hot dilute acid one can effectively disassemble hemicelluloses (consisting of xylan, glucuronoxylan, arabinoxylan, and/or glucomannan) into soluble, fermentable sugars. In corn stover, an economically attractive feedstock, xylans alone comprise about 20 ± 5% of the biomass2 and thus are an important potential source of fermentable sugars. However, under acidic conditions pentose sugars tend to degrade to products that are nonfermentable and even inhibitory to fermentation. The conversion of pentoses to furfural in acidic solutions is an important loss mechanism that has been well documented;3 however, reversion reactions have not been well studied.4 These acid catalyzed self-reactions lead to the loss of monomeric sugars from formation of di- and oligosaccharides1 that are difficult to ferment. Reversion products are formed after protonation of the anomeric carbon’s hydroxyl group (C-1), which is susceptible to acid attack. Reactions 1 and 2 show the mechanism for the formation of the 1,4-linked disaccharide from the acid catalyzed reversion of xylose. Elimination of water results in the formation of a carbocation (reaction 1), which can then react with the hydroxyl group of another sugar molecule to form the disaccharide (reaction 2). In this instance, the O4 on the second sugar adds β to the C1 carbon of the carbocation forming 1,4-O-β-D-xylopyranosyl-D-xylose (1,4-β-xylobiose). The second xylose can also add α to this carbon. Addition of the other hydroxyl groups to the carbocation can form the α and β isomers of 1,3-, 1,2-, and 1,1-linked disaccharides for a total of nine different xylose disaccharides. Because this © 2013 American Chemical Society

The only significant research on xylose reversion was published in 1958 by Ball and co-workers4a and describes the isolation of three disaccharides from β-D-xylose using paper chromatography, namely 1,4-β-xylobiose, 1,1-O-α-D-xylopyranosylα-D-xylopyranoside, and 1,3-O-α-D-xylopyranosyl-D-xylose. His Received: May 13, 2013 Revised: October 1, 2013 Published: October 16, 2013 7389

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels

Article

experiment. At the end of the reaction, the reactor cooled to room temperature within 30 s. To prepare reversion products for linkage analysis, reactions were conducted at 140 °C with a reaction time of 3 min and a high xylose loading (300 g L−1). The solution was filtered through a 0.45 μm nylon membrane syringe filter (Pall Corp., East Hills, NY) prior to use. These conditions produced the highest levels of disaccharide products and showed low levels of furfural production as measured by HPLC. Additional reactions were conducted at various times and temperatures to measure the reaction kinetics. UPLC and Fractionation. An ultra performance liquid chromatography (UPLC) system (Waters Co., Milford, MA) was used to fractionate the reversion product mixture in 2 μL aliquots using a Fraction Collector III (FC). The samples were kept at 4 °C within the Acquity Autosampler. The chromatography system utilized a Shodex Sugar SZ5532 column (6 × 150 mm, 6 μm particle size, Showa Denko K.K., Japan) to separate the disaccharide mixture for fractionation. The column was maintained at 60 °C, and water (A) and acetonitrile (B) made up the mobile phase. Separation was performed using a linear gradient program of (A) = 12% and (B) = 88% at time t = 0; (A) = 20% and (B) = 80% at t = 15 min; (A) = 35% and (B) = 65% at t = 25 min; (A) = 12% and (B) = 88% at t = 28 min. The flow rate was held constant at 0.8 mL min−1 and each sample was split 8:1 between the FC and an evaporative light scattering detector (ELSD). Figure 1

work showed evidence for two additional disaccharides, 1,2-O-α-Dxylopyranosyl-D-xylose and 1,4-O-α-D-xylopyranosyl-D-xylose, but they could not be separated, and the results were thus inconclusive. Minor6 touched on this subject as an extension of his glucose reversion studies, noting its importance when working with xylan-rich biomass as a substrate for conversion into monomeric sugars and subsequent fermentation. Minor’s study consisted of xylose solutions, at 100, 200, and 300 g L−1 loading in 1 wt % (0.102 mol L−1) H2SO4, reacting in an oil bath for 10 min at 170 °C and demonstrated that reversion took place at 300 g L−1 xylose. In addition, there was little evidence for the formation of trisaccharides from glucose.6 No follow-up study could be found. One of the barriers that have prevented a detailed study of the reversion reactions of xylose is the lack of commercially available xylose disaccharide standards. For glucose, the relevant disaccharides are commercially available, which has allowed detailed studies of these reactions.1,5b,6,7 However, the kinetics and product yields for xylose reversion reactions can be anticipated by examining the reactions of glucose. As mentioned above, yields of reversion products have been measured as high as 12%1 from glucose under process relevant reaction conditions, 140 °C and 200 g L−1 sugar loading. For glucose, the 1,6-linked disaccharides are the most prominent reversion products7a,b,8 due to reduced steric hindrance between the bulky sugar rings when there is an additional methylene group in the linkage.7b Because these products are not possible with xylose, it seems likely that there would be a lower overall yield of reversion products. In the present study, xylose reversion reactions were studied under mildly acidic conditions and the xylose disaccharides formed were isolated and identified. Experiments were conducted in a microwave-heated reactor system, where sugar solutions were reacted at conditions severe enough to produce high levels of disaccharides while minimizing degradation products. Product solutions were separated using liquid chromatography (LC) and molecular species were identified using linkage analysis with gas chromatography/mass spectrometry (GC/MS) and proton nuclear magnetic resonance (1H NMR) spectroscopy. Only disaccharides were identified in the product mixture, and the kinetics of their formation was calculated.



Figure 1. Chromatogram of the xylose reversion product mixture separated by LC/ELSD. All labeled peaks are xylose disaccharides as confirmed by linkage analysis and 1H NMR. Xylose monomer elutes prior to fraction 1.

EXPERIMENTAL SECTION

Reversion Reactions. Batch reactions with xylose were performed using a microwave reactor system with a 48-position autosampler (Explorer, CEM Corp., Matthews, NC).1 Microwave radiation of 230 MHz with a maximum, pulsed radiation power of 200 W was used to heat the samples. Reactions were carried out in sealed 10 mL reactor tubes containing 2 mL of aqueous xylose solution containing sulfuric acid (1.2 wt %, 0.122 mol L−1). Solutions were continuously stirred with Teflon-coated stir bars during the experiments so as to ensure uniform heating. An infrared sensor monitored the temperature inside the reactor, and the pressure was measured with a transducer attached to the top of the glass reactor tube. A computer-controlled temperature and pressure feedback system was used to regulate the microwave power for rapid heating and constant temperature. Steam table pressures were used to validate target temperatures. Samples were cooled via a stream of compressed air that was blown on the reactor tubes at the end of each run. For each temperature studied, control methods for the microwave reactor were developed to minimize heat-up time, reduce overshoot, and provide a steady target temperature. Temperature, pressure, and microwave power profiles were acquired using reactor control software (Synergy, CEM Corp.). The reactor was rapidly heated (within 2 min) to the target temperature, and was maintained to within 1 °C throughout each

shows a plot of a typical chromatogram obtained under these conditions. Based on our knowledge of the retention time of each analyte (each peak in the chromatogram), the flow rate, and time delay in the FC line due to volume, a fraction time method was calculated. The exact split conditions between the FC and ELSD detectors was estimated so the majority of sample was diverted to the FC while still being able to detect each analyte on the ELSD. The volume of the line going to the FC from the split point was measured with a gastight syringe, and was calculated to contain 300 μL of liquid. The method was programmed into the FC as follows: F1 (fraction 1), retention time (RT) in minutes = 12.50−13.41 min; F2, RT = 13.47−14.17 min; F3, RT = 14.23−15.05 min; F4, RT = 15.17−15.56 min; F5, RT = 16.11−16.50 min; F6, RT = 17.08−17.41 min; F7, RT = 17.53− 18.26 min; F8, RT = 18.29−19.14 min. The ELSD method for detection of each analyte was set as follows: The drift tube temperature of the light-scattering detector was set to 76 °C and the gain to 10. The gas pressure was set to 28 psi, while the nebulizer mode, heater-cooler, and data channel were disabled. Figure 2 shows chromatograms of the individual fractions run on the LC/ELSD using this collection program. In order to collect sufficient 7390

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels

Article

concentrations of each disaccharide for analysis, a total of 150 runs were performed. The xylose disaccharide solids obtained from each fraction weighed from 0.5 to 1.2 mg. Linkage Analysis. Linkage analysis of complex carbohydrates provides information on the position of the glycosidic linkages between sugar residues. The products are converted to a mixture of partially methylated alditol acetate derivatives (PMAAs), which are then readily analyzed by GC/MS.9 Preparation of Partially Methylated Alditol Acetates. The procedure was made up of the following steps: (A) permethylation of free hydroxyl groups in the disaccharide (reaction 3), (B) hydrolysis of glycosidic linkages (reaction 4), (C) reduction of the sugar residues with NaBD4 (reactions 5 and 6), and (D) acetylation of the remaining free hydroxyl groups to produce stable alditol acetate derivatives (reactions 7 and 8). The derivatives were identified by their characteristic mass spectral fragmentation patterns.9c,10 In this study, all of the disaccharides necessarily contain a linkage to the C1 anomeric carbon of one xylose, which limited the number of PMAAs that could be formed. It should be emphasized that this analysis cannot determine the anomeric linkage of the disaccharide. Figure 2. LC/ELSD chromatogram showing the fractionation efficiency of the xylose reversion products. Each trace has been normalized to the height of the largest peak. Fraction 9 not shown. was added and the headspace purged with nitrogen. The reaction mixture was stirred for an additional 40 min. The methylation reaction was quenched by adding deionized water (2 mL) and then mixed well. Excess methyl iodide was bubbled off with nitrogen until all of the tubes were colorless. Methylene chloride (2 mL) was added to the solution and mixed to extract the methylated carbohydrates. The solution was centrifuged and the aqueous top layer discarded. The organic layer was washed three times with water (2 mL × 3), and then the organic layer was transferred to another tube and dried down with nitrogen. At this stage, an internal standard (myo-inositol, 5 μg) was added. The permethylated sample was hydrolyzed (reaction 4) by adding 2 M trifluoroacetic acid (20 drops), capping tightly, and incubating on a heating block at 121 °C for 2 h. The sample was allowed to cool and then dried down with nitrogen. The procedure for alditol formation was as follows,9c,15 with modification: Sodium borodeuteride solution (10 g L−1 in 1 M ammonium hydroxide; 10−15 drops) was added to each sample and incubated overnight (reactions 5 and 6). The solution was neutralized with glacial acetic acid (3 to 5 drops). To remove the borate, methanol (5 drops) was added and the methyl borate was evaporated with nitrogen. A 9:1 (v/v) methanol−acetic acid solution (10 drops) was added and the solid was again dried down; this step was repeated once more. Methanol (10 drops) was then added and the solid was dried down; this was repeated until a crusty white precipitate formed. The remaining free hydroxyl groups were acetylated by addition of acetic anhydride (250 μL) to each tube, which was vortexed until the sample dissolved. Concentrated trifluoroacetic acid (230 μL) was added to this mixture. The tubes were incubated at 50 °C for 10 min and then allowed to cool. Isopropanol (2 mL) was added, and the sample was evaporated to dryness; this step was repeated once more. Then 0.2 M sodium carbonate (2 mL) was added and the liquid vortexed. Methylene chloride (2 mL) was added and gently mixed to allow partitioning of the PMAAs into the organic layer. The solution was centrifuged and the aqueous top layer discarded. Additional water was added, the solution was centrifuged, and again the aqueous layer discarded. This step was repeated 3 more times (2 mL × 3). Finally, the organic layer was dried with nitrogen, and the derivatives redissolved in methylene chloride (15 drops). GC/MS Analysis. GC/MS was employed to characterize the PMAA derivatives. The samples were analyzed on an Agilent HP 6890 GC and a 5973 MS (Agilent Technologies, Palo Alto, CA). Analyte separation was achieved with a SP-2330 capillary column (30 m × 0.25 mm × 0.2 μm film thickness, Supelco Analytical, Bellefonte, PA). For the acquisition of the electron ionization (EI) mass spectra, we used an MS source temperature of 230 °C, MS Quad temperature of

The methylation procedure using dimsyl reagent (methyl sulfanyl anion) as base catalyst9d,11 led to incomplete methylation.12 The methylation procedure was modified by replacing the dimsyl reagent with a sodium hydroxide (NaOH)−DMSO suspension.9g Permethylation (reaction 3), was conducted as follows,9g,13 with modification:14 Approximately 500 μg of carbohydrate was added to a clean, dry (13 × 120 mm) screw top vial with stir bar. Dry DMSO (200 μL) was added to the sample. The tube was purged with dry nitrogen and the sample stirred for 24 h until fully dissolved. NaOH base solution (400 μL), as prepared by Anumala et al.,13 was added to this mixture then stirred for 10 min. Finally, methyl iodide (100 μL) 7391

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels

Article

150 °C, and electron energy of 70 eV. Each sample was loaded onto an Agilent autosampler, and 1 uL was injected. The inlet temperature was set to 240 °C and the MS transfer line to 240 °C, and He carrier gas flow was maintained at 1 mL min−1. The initial temperature of 80 °C was held for 2 min and then ramped at 30 °C min−1 to a temperature of 170 °C. This was followed by a temperature increase of 4 °C min−1 until a final temperature of 240 °C was reached. This temperature was maintained until the end of the 42.5 min run. The mass spectrum was collected over a range of 35−450 m/z. The total ion chromatogram (TIC) and MS fragmentation pattern for each sample was collected and analyzed on Agilent Environmental ChemStation G1701DA D.00.00.38 software (Agilent Technologies, Palo Alto, CA). High Performance Liquid Chromatography (HPLC) Analysis of Mono- and Oligosaccharides. The product solutions from the microwave reactor were analyzed using a HPLC equipped with refractive index and photodiode array detectors (Agilent 1100, Agilent Technologies, Palo Alto, CA). A Rezex RFQ Fast Acids column (100 × 7.8 mm, 8 μm particle size, Phenomenex, Torrance, CA) and Cation H+ guard column (BioRad Laboratories, Hercules, CA) operated at 85 °C were used to separate sugar monomers, total oligomers, and degradation products present in the reaction solutions. The eluent was 0.02 M H2SO4 with a flow rate of 1.0 mL min−1. Samples and standards were filtered through a 0.45 μm nylon membrane syringe filter (Pall Corp., East Hills, NY) prior to injection (2.5 μL) onto the column. Calibration standards were prepared in-house. Triplicates were run at various reaction times within each experiment. The HPLC was controlled and data was analyzed using Agilent ChemStation software (Rev.B.03.02). Anion-Exchange Chromatography. To confirm peak purity, anion-exchange chromatography with pulsed amperometric detection (HPAE-PAD) was used (Dionex ICS 3000, Thermo Scientific, San Jose, CA). Individual fractions were analyzed using a CarboPac PA-20 column (3 × 150 mm, Dionex Corp., Sunnyvale, CA) operated at 35 °C. The column was used with an electrochemically generated KOH eluent gradient set to run from 50 to 100 mM over 20 min. It was held an additional 5 min for regeneration of the column and then equilibrated at the starting eluent composition for 5 min at the end of each run. Sample concentrations were measured with an ED50 electrochemical detector operated with waveform A. The chromatograph was controlled and the data analyzed using Dionex Chromeleon software (versions 6.6 and 6.8). Samples and standards were diluted to a maximum concentration of 4 mg L−1 with fucose (6 mg L−1) as an internal standard and sodium azide (0.04%) as a preservative. Liquid-State NMR Analysis. Each carbohydrate fraction was airdried and dissolved in D2O (0.5 mL). 3-(trimethylsilyl)-1-propanesulfonic acid sodium salt was used as an internal standard. All NMR experiments were performed at 25 °C on a 600 MHz spectrometer (Ultrashield Plus, Bruker) with a 5 mm PABBO probe. 1H NMR spectral width was 0 to 15 ppm; 90° pulse; 1 s pulse delay (d1); 2.656 s acquisition time; and 256 scans. For 13C NMR experiments, the spectral width was 0 to 210 ppm; 90° pulse width; 2 s pulse delay (d1); 0.908 s acquisition time; and 100 000 scans. 1H−1H 2D-NMR spectra−correlation spectroscopy (COSY), and the total correlation spectroscopy (TOCSY), were recorded. The parameters for the TOCSY experiment were as follows: spectral width of 9014.4 Hz, relaxation time 2.0 s, mixing time 80 ms, mixing sequence mlev-17, and 96 scans. 13C−1H heteronuclear single quantum correlations (HSQC) were also performed to assign each signal and determine the stereochemistry of each isolated fraction. In the HSQC experiment, spectral widths were 6000 and 31 500 Hz for 1H and 13 C dimensions, respectively. The number of collected complex points was 1024 for 1H dimension, with a recycle delay of 1.5 s. The number of transients was 64 and 256 time increments were always recorded in 13C dimension. Assignments of the NMR spectra will be discussed below.

disaccharides and tentatively identify the ninth. Furthermore, as will be shown, the disaccharides account for all of the major peaks in the chromatograms, with the exception of two peaks that appear to contain unknown analytes. The chromatograms in Figures 1 and 2 show that the LC peaks were efficiently isolated in the xylose reversion mixture. Nine separate peaks were initially labeled as fractions 1 through 9 (F1−F9). The reversion mixture was fractionated, consolidated, concentrated, and evaporated to dryness. Each fraction was analyzed separately by LC/ELSD and HPAE/PAD to verify peak purity. The fractions were reasonably clean and showed minimal carryover. Linkage analysis with subsequent GC/MS, and 1H NMR spectroscopy were used to identify the xylose disaccharides. The individual fractions, as well as the reversion mixture (containing all fractions), were analyzed using HPAE/PAD, which allowed further confirmation of each peak. Figure 3

Figure 3. Typical anion-exchange chromatogram of products from the reversion reaction of xylose (200 g L−1) in 1.2 wt % H2SO4 at 140 °C for 3 min.

shows a typical anion chromatogram of the xylose reversion mixture. The peaks in this spectrum were correlated with the LC/ELSD in Figures 1 and 2, by injecting small quantities of each fraction into the anion-exchange chromatograph. In the chromatogram shown in Figure 3, nearly all of the possible xylose disaccharide peaks were resolved with the exception of F1 and F3 which coeluted. Further analysis of each individual fraction by HPAE/PAD showed one additional unknown peak in fractions F4 and F5, which was not evident in the ELSD chromatograms. In order to explain these discrepancies, further analysis was conducted by 1H NMR. To determine the glycosyl linkage of the sugars, the individual fractions were converted to complex mixtures of PMAAs using the procedure described above, and then analyzed using GC/MS. Conversion to alditol acetates produce two peaks per sample in the GC chromatogram. For example, the 1,4-linked disaccharide shown in reactions 3−8 produces two PMAAs and shows two dominant peaks in the gas chromatogram, one for a 1-linked and another for the 4-linked sugar. The corresponding mass spectrum for each PMAA had distinctive masses associated with deuteration, methylation, and acetylation at specific positions in each possible PMMA, making identification relatively straightforward.



RESULTS AND DISCUSSION Product Identification. Using the analyses described above, we were able to positively identify eight of the nine possible xylose 7392

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels

Article

Figure 4. GC/MS profiles of each PMAA derivative recorded over the range m/z 80 to 250 and corresponding fragmentation patterns: (a) 1-linked xylose disaccharide, (b) 2-linked, (c) 3-linked, (d) 4-linked. OMe, O-methylation; OAc, O-acetylation; D, deuterium atom.

glucose disaccharide. Table 1 shows the 1H NMR chemical shifts of the anomeric protons of carbohydrates in each fraction. F1 and F3 were both identified as 1,3-linked xylose disaccharides by the GC/MS fragmentation patterns of the resulting PMAAs (Figure 4c). In support of the discussion of NMR spectra, Figure 5 shows the sequential carbon positions in 1,4-β-D-xylobiose. Figure 6 shows the 1H NMR spectra of anomeric protons in F1 and F3 compared to the glucodisaccharides, 1,3-O-β-D-glucopyranosyl-D-glucose (laminaribiose), and 1,3O-α-D-glucopyranosyl-D-glucose (nigerose). As shown in Figure 6a, the xylose disaccharide in F1 has two β-anomeric proton peaks [β-B(α)1 and β-B(β)1] at 4.65 and 4.62 ppm, respectively, in the nonreducing unit (B) connected to the α- and β-anomeric configuration of the reducing unit (A). It also possesses a similar signal pattern of anomeric protons in the reducing and nonreducing units of laminaribiose (Figure 6b). The coupling constants (J) of the β-anomeric protons [β-B(α)1 and β-B(β)1] in the nonreducing unit in F1 are J = 7.80 and 7.80, respectively, are also similar to those in laminaribiose, J = 7.80 and 7.98, respectively. Note that F1 does not contain α-anomeric proton

There are four possible linkages for xylose disaccharides, each possessing a unique, identifiable group of primary ions associated with the PMAAs from that linkage. Figure 4a−d shows the mass spectra obtained from these PMAAs and each spectrum is identified as a monomer derived from a 1-, 2-, 3-, or 4-linked disaccharide. The inserts in each spectrum show the possible primary fragment ions for each PMAA and through a process of elimination the linkages could be assigned. For instance, in Figure 4a the peak at m/z 161 could only arise from a PMAA with a linkage at position 1 or 2; the peak at m/z 162 is associated with a linkage at position 1 or 4; the peak at m/z 117 with a linkage at 1, 2, or 3; and the peak at 118 with a linkage at 1, 3, or 4. The primary ions that were used for identification are shown in bold in the inserts. Secondary ions, such as m/z 101, 102, 129, and 130, do not assist in the initial identification of the linkages but can be used to further strengthen identification. 1 H NMR spectroscopy was used to confirm the identity of the disaccharides and determine the stereochemistry of the anomeric linkage. Identification with NMR spectroscopy was aided by comparison with the spectra of the corresponding 7393

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels

Article

Table 1. 1H Chemical Shifts of the Anomeric Protons (H-1) of Each Xylose Disaccharide Fraction chemical shift (ppm) reducing unit (Aa) fraction no. F1 F3 F4 F5

type of linkage β1→3 α1→3 β1→2 α1→2 β1→4

F6

α1→4

F7

α1→ β1

F8

α1→ α1

α-A1b 5.17 (d, J 5.18 (d, J 5.32 (d, J 5.37 (d, J 5.18 (d, J 5.18 (d, J

= 3.60) = 3.60) = 3.66) = 3.42) = 3.60) = 3.66)

β-A1b 4.58 (d, J 4.59 (d, J 4.55 (d, J 4.72 (d, J 4.58 (d, J 4.57 (d, J

nonreducing unit (Ba) α-B(α)1c

α-B(β)1c

α-B1e

= 7.92) = 7.92)

5.33 (d, J = 3.84)

β-B(β)1d

4.65 (d, J = 7.80)

4.62 (d, J = 7.80)

4.45 (d, J = 7.86)

4.45 (d, J = 7.86)

β-B1e

5.312 (d, J = 3.90)

= 7.74) = 7.86)

β-B(α)1d

5.04 (d, J = 3.78)

5.33 (d, J = 3.74)

5.11 (d, J = 3.68)

5.11 (d, J = 3.84)

= 7.92) = 7.92)

5.17 (d, J = 3.72) 5.11 (d, J = 3.84)

4.57 (d, J = 7.68)

A and B represent the anomeric configuration of the reducing and nonreducing units. See Figure 5. bα-A and β-A represent the α-anomeric and βanomeric configuration of the reducing unit A. cα-B(α) and α-B(β) represent the α-anomeric configuration of the nonreducing unit B connected to the α- and β-anomer of residue A. dβ-B(α) and β-B(β) represent the β-anomeric configuration of the nonreducing unit B connected to the α- and βanomer of residue A. eα-B and β-B represent the α-anomeric and β-anomeric configuration of the nonreducing unit B. a

those in sophorose, consisting of two doublet peaks at 5.32 and 4.55 ppm, respectively. Conversely, F4 shows no similarity with the signal pattern of 1,2-O-α-D-glucopyranosyl-D-glucose (kojibiose) nor with any minor peaks in the region (4.50−5.60 ppm) of the anomeric proton. Also, there are large multiplet proton signals in the region of 4.00 to 4.50 ppm that are not found in any of the other 1H NMR spectra. It should be noted that some of the minor peaks arise from F3 and F5 contamination in F4 due to limitations in the separation efficiency of the column used during fractionation. These data indicate that F4 contains not only 1,2-β-xylose disaccharide, but also unknown products. GC/MS analysis of the PMAAs from F5 showed the presence of two disaccharides: 1,2- and 1,4-linked xylose disaccharides. 1H NMR spectroscopy confirmed the presence of 1,4-β-xylobiose; as well as a 1,2-α-xylose disaccharide by comparison to 1,4-β-xylobiose and kojibiose, respectively, as shown in Figure S2 in the Supporting Information. The 1H NMR spectrum of F5 has the same proton signals as 1,4-βxylobiose, and similar signal patterns to 1,4-O-β-D-glucopyranosyl-D-glucose (cellobiose). 1H−1H TOCSY spectrum shows correlation between all protons into groups or coupling networks, and is useful for dividing the proton signals of ring protons in each C5 unit. Four different correlation patterns were detected on the TOCSY spectrum of F5; two of them arose from the reducing and nonreducing unit of the 1,4-linked xylose disaccharide. The other two correlation patterns appeared to be derived from another disaccharide. Table 1 shows the chemical shifts of H1-α associated with a 1,2-linkage; 5.37 ppm (α-A1) and 4.72 ppm (β-A1) in the reducing unit and 5.04 [α-B(α)1] and 5.33 ppm [α-B(β)1] in the nonreducing unit. F6 contained a 1,4-linked disaccharide as determined by GC/MS, which was identified as 1,4-α-D-xylopyranosyl-D-xylose based upon comparison of the 1H NMR spectrum to 1,4-O-αD-glucopyranosyl-D-glucose (maltose), as shown in Figure S3 in the Supporting Information. In the 1H NMR spectrum of F6, the anomeric signal patterns in both reducing and nonreducing

Figure 5. Chemical structure of 1,4-β-D-xylobiose. A refers to the reducing unit and B to the nonreducing unit.

peaks in the nonreducing unit. These data indicate that this fraction contains the 1,3-β-xylose disaccharide. Conversely, the disaccharide in F3 (Figure 6c) has two αanomeric proton peaks [α-B(α)1 and α-B(β)1] at 5.33 and 5.31 ppm, respectively, in the nonreducing unit connected to the α- and β-anomeric configuration of the reducing unit (A). It also has similar chemical shift patterns to the anomeric protons in nigerose (Figure 6d). The coupling constants of the two β-anomeric protons (α-B(α)1 and α-B(β)1) in the nonreducing unit in F3 (J = 3.84 and 3.90) are similar to those in nigerose (J = 3.54 and 3.66). In F3, no β-anomeric proton peaks in the nonreducing unit were detected. Based on these 1H NMR data, the molecule in F3 was assigned as the 1,3-α-xylose disaccharide. The α and β-anomeric protons in the reducing unit [α-A1 and β-A1] of the disaccharides in both F1 and F3 appeared at nearly the same chemical shifts of 5.17 and 4.58 ppm, respectively. F4 was identified as a 1,2-linked xylose disaccharide using GC/MS of the resulting PMAAs. The assignment of this species as a β-isomer was based upon comparison to the chemical shifts of 1,2-O-β-D-glucopyranosyl-D-glucose (sophorose) as shown in Figure S1 in the Supporting Information. The β-anomeric proton peaks in the nonreducing unit [βB(α)1 and β-B(β)1] in F4 appear at the same chemical shift of 4.55 ppm. F4 also has similar signal patterns of the α and βanomeric protons in the reducing unit (α-A1 and β-A1) to 7394

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels

Article

Figure 6. 1H NMR spectra of (a) fraction 1, (b) laminaribiose, 1,3-β-D-glucose, (c) fraction 3, and (d) nigerose, 1,3-α-D-glucose. Fractions 1 and 3 have the same spectral patterns as laminaribiose and nigerose, respectively.

analyzed the reversion reaction mixture for the presence of 1,4-linked xylooligomers, which are commercially available up to xylohexaose. Figure S5 in the Supporting Information shows a typical chromatogram from the reaction mixture of xylose compared to a standard containing xylose and 1,4-linked xylooligomers from xylobiose to xylohexaose. Xylose and xylobiose were clearly identified in the chromatogram, but none of the larger oligomers were observed. If xylotrisaccharides or higher were formed from reversion reactions, one would expect to see the 1,4-xylotrisaccharide in this mixture. This was not the case. Reaction Kinetics. The kinetics of xylose reversion reactions are important for optimizing biomass pretreatment processes. In analyses with the Rezex HPLC column all of the oligomer products elute in one peak at 2.30 min and xylose at 2.65 min, as shown in Figure 7. In the kinetic analysis, this oligomer peak (which has been shown to be comprised solely of disaccharides) was used to estimate the total concentration of the reversion products by calibrating with xylobiose, for which a standard was available. Conducting kinetic analysis using the oligomer peak as representative of the global formation of xylose disaccharides is legitimate, since the product identification indicates that disaccharides account for the overwhelming majority of the products. Calibration with xylobiose provides a reasonable estimate of the total disaccharide concentrations, considering that we have found that the response factors of all the glucose disaccharides were similar.1 The concentrations of the sum of the reversion disaccharides determined in this manner was used to extract the reaction kinetics. Figure 8 shows the percentage yields of disaccharides from xylose as a function of reaction time at different xylose loadings (10−300 g L−1) at a reaction temperature of 120 °C. As can be seen, the yields of disaccharides increase with reaction time until 10 min at which point the disaccharides reach the equilibrium concentrations, and further increases in reaction

units of the disaccharide were similar to those of maltose. The signal order of the α-anomeric proton (5.11 ppm) in the nonreducing unit of F6 did not match that of maltose; however, the integration of the doublet at 5.11 ppm is nearly the same as the total integration of two doublets at 4.57 and 5.18 ppm, suggesting that the doublet at 5.11 ppm corresponded to one proton. Based on this proton signal information, the doublet at 5.11 ppm was identified as the α-anomeric proton in the nonreducing unit of F6. Fractions 7 and 8 were identified as containing only 1,1linked disaccharides by GC/MS analysis. Comparing the 1H NMR spectra of 1,1-α-D-glucopyranosyl-β-D-glucopyranoside (neotrehalose), 1,1-β-D-glucopyranosyl-β-D-glucopyranoside (isotrehalose), and 1,1-α-D-glucopyranosyl-α-D-glucopyranoside (trehalose), F7 and F8 have similar anomeric proton signal patterns to those in neotrehalose and trehalose, respectively, as shown in Figure S4 in the Supporting Information. F7 also contains a multiplet signal at 5.11 ppm, which is contamination from F6 and F8. There is a small peak at 4.80 ppm in the 1H NMR spectrum of F7 overlapping with the H2O signal, which has a similar chemical shift to that of isotrehalose. No other fractions have a signal at the same chemical shift thus indicating that the small peak may be assigned to the anomeric proton of 1,1-β-D-xylopyranosyl-β-D-xylopyranoside. Preparation and subsequent GC/MS analysis of PMAA from fractions F2 and F9 indicated that neither fraction possessed any fragment ions associated with xylose disaccharides. The fragmentation patterns were not consistent with any xylose disaccharide spectra, and also owing to the fact that both peaks in the LC chromatogram were of extremely low concentration, as seen in Figure 1, no further analysis was conducted on these fractions. As additional evidence that only disaccharides were formed to an appreciable extent in our reversion reactions, we also 7395

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels

Article

controlled by this equilibrium constant at reaction times above 10 min for an acid concentration of 1.2 wt % (0.122 mol L−1). We have found that this equilibrium constant is independent of acid concentration, but the reaction time at which equilibrium is reached decreases with increasing acid concentration. As will be shown below, this time also decreases with temperature. In order to estimate the global kinetics of the reversion reaction of xylose, we have fit the total observed concentration of xylose disaccharides to a simple kinetic model as a function of reaction time (s) and temperature (K). For this fit we have assumed the reaction mechanism shown below, where k1 is the bimolecular forward rate constant governing the formation of the observed family of disaccharides, k−1 is the rate constant representing the lumped disaccharide hydrolysis rate constant, and k2 is the overall rate constant for the conversion of xylose to furfural.

Figure 7. Typical HPLC chromatogram with the Rezex column of products from the reversion reaction of xylose (200 g L−1) in 1.2 wt % H2SO4 at 120 °C. The peak containing all of the oligomers, B, increases with reaction time, whereas C is the peak for xylose. Fucose, peak A, was the internal standard.

2 xylose → xylose disaccharides

k1

xylose dissacharides → 2 xylose

k −1

xylose → furfural

k2

The latter reaction is important as it represents a net loss mechanism for xylose. Since the reversion of xylose is bimolecular, loss of xylose via k2 will reduce the extent of disaccharide formation. Arrhenius temperature dependences were assumed for all of the rate constants. The pre-exponential factors, A1, A−1, and A2, and activation energies, Ea1, Ea‑1, and Ea2, were then obtained by solving the system of ordinary differential equations (proposed model) using Isoda from the Fortran library odepack as implemented in Python. The Arrhenius parameters were optimized via the Levenberg−Marquardt algorithm through least-squared fit of the model to the data. A plot of the fit of the data is presented in Figure 9. The parameters used to obtain the

Figure 8. Plot of the formation of disaccharides from xylose at different xylose loadings (10 to 300 g L−1) and reaction time. Temperature was 120 °C and acid concentration 1.2 wt %.

time do not increase the concentrations of the disaccharides. This type of equilibrium behavior was also seen with disaccharides formed from reversion reactions of glucose.1 Note that at high loadings of xylose, 300 g L−1, the yields of xylose disaccharides reach slightly above 10%. This represents a significant loss of sugars, showing that reversion reactions should be considered when developing biomass pretreatment conditions. The equilibrium concentration of the disaccharides plotted as a function of starting xylose concentration, Figure S6 in Supporting Information, shows that the disaccharide concentration is dependent upon the xylose concentration squared, which is consistent with the bimolecular mechanism for disaccharide formation shown in reactions 1 and 2. The fit of this data is obtained assuming an equilibrium constant of Keq = [xylose disaccharides]/[xylose]2 = 0.259 ± 0.008 L g−1. At 120 °C, the amount of xylose disaccharides formed is

Figure 9. Dependence of xylose disaccharide formation on reaction temperature (100−140 °C) and time. Xylose loading was 10 g L−1 and acid concentration 1.2 wt %. Points are experimental data and lines are calculated using the Arrhenius kinetic model.

fits found in Figure 9 are A1 = 3.3 × 106 L mol−1 s−1, A−1 = 4.7 × 1013 L mol−1 s−1, A2 = 1.8 × 109 s−1, Ea1 = 117 kJ mol−1, Ea‑1 = 123 kJ mol−1, and Ea2 = 112 kJ mol−1. These Arrhenius parameters are similar to those obtained from the reversion reactions of glucose, where Ea values in the range of 117−163 kcal mol−1 were obtained.1 7396

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397

Energy & Fuels



Article

M.; Stevenson, T. T.; Albersheim, P. In Methods in Enzymology; Ginsburg, V., Ed.; Academic Press: New York, 1985; Vol. 118. (d) Lindberg, B. In Methods in Enzymology; Ginsburg, V., Ed.; Academic Press: New York, 1972; Vol. 28. (e) Laine, R. A.; Esselman, W. J.; Sweeley, C. C. In Methods in Enzymology; Ginsburg, V., Ed.; Academic Press: New York, 1972; Vol. 28. (f) Ciucanu, I.; Kerek, F. Carbohydr. Res. 1984, 131, 209. (g) Ciucanu, I.; Costello, C. E. J. Am. Chem. Soc. 2003, 125, 16213. (10) Price, N. P. J. Appl. Biochem. Biotechnol. 2008, 148, 271. (11) (a) Hakomori, S. J. Biochem. 1964, 55, 205. (b) Corey, E. J.; Chaykovsky, M. J. Am. Chem. Soc. 1965, 87, 1345. (c) Carpita, N. C.; Shea, E. M. In Analysis of carbohydrates by GLC and MS; Biermann, C. J., McGinnis, G. D., Ed.; CRC Press: Boca Raton, FL, 1989. (d) Needs, P. W.; Selvendran, R. R. Carbohydr. Res. 1993, 245, 1. (12) (a) Conner, A. H.; Wood, B. F.; Hill, C. G.; Harris, J. F. J. Wood Chem. Technol. 1985, 5, 461. (b) Zaehringer, U.; Rietschel, E. T. Carbohydr. Res. 1986, 152, 81. (13) Anumula, K. R.; Taylor, P. B. Anal. Biochem. 1992, 203, 101. (14) Azadi, P. Complex Carbohydrate Research Center: Athens, GA, 2010. (15) (a) Gunner, S. W.; Jones, J. K. N.; Perry, M. B. Can. J. Chem. 1961, 39, 1892. (b) Biermann, C. J. In Analysis of carbohydrates by GLC and MS; Biermann, C. J., McGinnis, G. D., Eds.; CRC Press: Boca Raton, 1989.

CONCLUSIONS Dilute acid hydrolysis of xylose can be expected to produce both α and β forms of 1−1, 1−2, 1−3, and 1−4-linked disaccharides. Considering that the only standard commercially available for xylooligomers is the β-1,4 linked disaccharide (xylobiose), it is difficult to identify xylose reversion products in product streams. Through careful isolation and analysis with GC/MS and 1H NMR spectroscopy, we have shown that the only reversion products formed from xylose are disaccharides. Furthermore, our measurements of the NMR constants for these disaccharides should allow future identification of all of these species in pretreatment broths. Finally, the measured kinetics presented here will aid in the prediction of the concentrations of reversion products.



ASSOCIATED CONTENT

* Supporting Information S

Additional material as discussed in the text including Figures S1−S6. This material is available free of charge via the Internet at http:// pubs.acs.org.



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by the U.S. Department of Energy through the Bioenergy Technologies Office. REFERENCES

(1) Pilath, H. M.; Nimlos, M. R.; Mittal, A.; Himmel, M. E.; Johnson, D. K. J. Agric. Food Chem. 2010, 58, 6131. (2) Wiselogel, A.; Tyson, S.; Johnson, D. Biomass Feedstock Resources and Composition. In Handbook on Bioethanol: Production and Utilization; Wyman, C. E., Ed; Taylor and Francis: Washington, DC, 1996; pp 105−118. (3) (a) Grethlein, H. E. Biotechnol. Bioeng. 1978, 20, 503. (c) Dunlop, A. P. Ind. Eng. Chem. 1948, 40, 204. (d) Hurd, C. D.; Isenhour, L. L. J. Am. Chem. Soc. 1932, 54, 317. (e) Harris, D. W.; Feather, M. S. Carbohydr. Res. 1973, 30, 359. (f) Feather, M. S. Tetrahedron Lett. 1970, 4143. (g) Feather, M. S.; Harris, D. W.; Nichols, S. B. J. Org. Chem. 1972, 37, 1606. (h) Antal, M. J.; Leesomboon, T.; Mok, W. S.; Richards, G. N. Carbohydr. Res. 1991, 217, 71. (i) Root, D. F.; Saeman, J. F.; Harris, J. F.; Neill, W. K. For. Prod. J. 1959, 158. (b) Saeman, J. Ind. Eng. Chem. 1945, 43. (4) (a) Ball, D. H.; Jones, J. K. N. J. Chem. Soc. 1958, 33. (b) Jacobsen, S. E.; Wyman, C. E. Appl. Biochem. Biotechnol. 2000, 84− 6, 81. (c) Mosier, N.; Wyman, C.; Dale, B.; Elander, R.; Lee, Y. Y.; Holtzapple, M.; Ladisch, M. Bioresour. Technol. 2005, 96, 673. (5) (a) Helm, R. F. University of Wisconsin: Madison, WI, 1987. (b) Helm, R. F.; Young, R. A.; Conner, A. H. Carbohydr. Res. 1989, 185, 249. (6) Minor, J. L. J. Appl. Polym. Sci., Appl. Polym. Symp. 1983, 37, 617. (7) (a) Peat, S.; Whelan, J.; Edwards, T. E.; Owen, O. J. Chem. Soc. 1958, 586. (b) Thompson, A.; Anno, K.; Wolfrum, M.; Inatome, M. J. Am. Chem. Soc. 1954, 76, 1309. (c) van Dam, H. E.; Kieboom, A. P. G.; van Bekkum, H. Starch 1986, 38, 95. (d) Wolfrom, M. L.; Thompson, A.; Timberlake, C. E. Cereal Chem. 1963, 40, 82. (8) Silberman, H. J. Org. Chem. 1961, 26, 1967. (9) (a) Sassaki, G. L.; Iacomini, M.; Gorin, P. A. J. Anais Acad. Br. Cienc. 2005, 77, 223. (b) Doares, S. H.; Albersheim, P.; Darvill, A. G. Carbohy. Res. 1991, 210, 311. (c) York, W. S.; Darvill, A. G.; McNeil, 7397

dx.doi.org/10.1021/ef400889u | Energy Fuels 2013, 27, 7389−7397