The kinetics of Brønsted acid-catalyzed hydrolysis of hemicellulose dissolved in 1-ethyl-3-methylimidazolium chloride

Abstract

The conversion of plant biomass provides a sustainable pathway towards the production of renewable fuels. Hemicellulose, a readily available form of biomass, can be catalytically converted to provide a range of fuel molecules, from furans and sugar alcohols to alkanes and aromatics. Using ionic liquids as solvent and Brønsted acid catalysts for biomass deconstruction, we investigated the kinetics of hemicellulose (xylan) hydrolysis and the subsequent dehydration/degradation reactions. These findings were compared to those found for similar reactions involving cellulose. In 1-ethyl-3-methylimidazolium chloride ([Emim][Cl]) at 80 °C, we report that hemicellulose can be hydrolyzed to xylose in 90% yield, with 5 wt % dehydration products and 4 wt % humins, when water is added stepwise. This chemical process presents a viable pathway for producing sugars capable of being chemically (via dehydration/hydrogenation) or biologically (via fermentation) upgraded to potential fuel molecules.

---

Introduction

Lignocellulosic biomass is an attractive feedstock for the production of fuels and chemicals in a sustainable manner.^1–3 The challenge is to deconstruct biomass (such as Miscanthus x giganteus) into its component parts—cellulose (25–50 wt%), hemicellulose (20–40 wt%), and lignin (10–30 wt%)⁴—so that they can then be converted into final products by either biochemical^5–7 or chemical means.^6–11 Recent studies have shown that ionic liquids, salts with melting points below 100 °C, are capable of complete biomass dissolution,¹² thereby facilitating the further processing of the components.^13,14 It has also been shown that once dissolved in ionic liquids, cellulose and hemicellulose can be hydrolyzed to their component sugars using acid catalysts.^14–17 The hydrolysis of both cellulosic components is, however, accompanied by the dehydration of the released sugars to form furanics, such as 5-hydroxymethyl furfural (HMF) and furfural, and by the condensation of the furanic products with sugars to form humic substances or humins.^16–18 For these reasons there is an interest in understanding the kinetics of the acid-catalyzed cellulose and hemicellulose hydrolysis occurring in ionic liquids, so that reaction conditions can be chosen to minimize the loss of organic matter to humins and to furfural and its derivatives. Investigations of cellulose hydrolysis in ionic liquids has shown that the rate is first order in the concentrations of β-1,4 glycosidic bonds and acid, but zero order in the concentration of water.¹⁶ It has also been demonstrated that the progressive addition of water during the course of cellulose hydrolysis reduces the dehydration of glucose to HMF and the formation of humins.^9,18 By contrast, the kinetics of hemicellulose dissolution under comparable conditions have not been investigated and, hence, are a subject of interest.

The focus of the present study is on understanding the mechanisms and kinetics of hemicellulose hydrolysis to xylose (and other monomeric sugars) occurring in ionic liquids with minimal formation of side products such as furfural and humins. The results of this investigation are compared to similar kinetics for cellulose hydrolysis and glucose dehydration. The ability of water to suppress the kinetics of sugar dehydration was also explored.

Experimental

Materials

The ionic liquid 1-ethyl-3-methyl imidazolium chloride ([Emim][Cl], 95%), xylan (from Birchwood, ≥90% xylose residues by HPAE), D-xylose (99%), microcrystalline cellulose (avicel ph-101), L-arabinose (99%), 2-furaldehyde (furfural, 99%), 5-hydroxymethal furfural (5-HMF, 99%), sulfuric acid (H₂SO₄, 98%), methanesulfonic acid (CH₃SO₃H, 99%), trifluoroacetic acid (CF₃COOH, 99%), phosphoric acid (H₃PO₄, 99%), acetic acid (CH₃COOH, glacial), and 1,4-dioxane (anhydrous, 99.8%) were purchased from Sigma-Aldrich. Hydrochloric acid (HCl, 37%) was purchased from Fischer Scientific. Xylobiose (XB, >95%), xylotriose (XTr, >95%), and xylotetraose (XTe, >95%) were purchased from MegaZyme. All ionic liquids were vacuum dried at 100 °C for 24 h prior to use to ensure minimal water content (measured via Karl-Fischer titration to be approximately 2 wt%). All other materials were used as purchased, without further purification or modification.

Experimental approach

All experiments were performed using a Symx core module deck to maintain a constant reaction temperature and stirring rate. A representative procedure for xylan hydrolysis: xylan (27 mg, 200 μmol sugar residues) is allowed to dissolve in [Emim][Cl] (500 mg, 480 μl) in a 5 ml threaded cap vial at 80 °C and a 200 rpm stirring rate for two hours, or until a clear solution is formed. De-ionized water (1 mmol) and H₂SO₄ (100 μmol) are then added and the reactor vial is sealed. Upon completion of the reaction, the sample is removed and quenched in an ice bath. An internal standard (1 ml of 10 mg ml⁻¹ 1,6-hexanediol in water) is added and the sample is centrifuged to remove all water insoluble particulates. A portion of the reaction mixture (500 μl) is removed and mixed with H⁺/OH⁻ ion exchange resin to remove the bulk of [Emim][Cl], after which 200 μl is taken for analysis. For reactions involving an additional organic phase, 1,4-dioxane (500 μl) is added prior to the addition of H₂SO₄, and is removed via centrifuging prior to addition of the internal standard. A different internal standard (1 ml of 5 mg ml⁻¹ guaiacol in 1,4-dioxane) is added to this organic phase prior to analysis.

Catalyst screening

A range of Brønsted-acids were screened for effectiveness in hydrolyzing xylan. Consistent with a similar study on cellobiose hydrolysis¹⁷ and correlations between pK_a values of acids in water and acids in ionic liquids,¹⁵ the relative order of acid strengths in [Emim][Cl] is the same as that in water. The curve generated from this experiment (see ESI,† Fig. A) closely resembles the extent of acid dissociation in water and is similar to what is seen for the acid hydrolysis of cellulose in ionic liquids.¹⁸ Weak acids (pK_a, aqueous >2.1, specific catalysis region) showed virtually no reactivity, medium strength acids (−2< pKa, aqueous <2.1) showed moderate reactivity, and strong acids (pK_a, aqueous <−2, general catalysis region) showed the greatest reaction rates. Strong acid reactivity was uniform in acids of aqueous pK_a less than −2.

Product analysis

A Shimadzu HPLC equipped with a Biorad Aminex HPX-87H column (300 × 7.8 mm; 0.01 N H₂SO₄; 0.6 ml min⁻¹; 50 °C) and a refractive index detector (RID) was used to analyze all aqueous/ionic liquid-phase samples. Product quantities were determined by converting the integrated HPLC peak areas into concentrations using a 7-point calibration curve generated from purchased standards.

A Varian CP-3800 gas chromatograph equipped with a FactorFour capillary column (UF-5 ms 30 m, 0.25 mm, 0.25 μm, P/N CP8944) connected to a Varian quadrupole-mass spectrometer (MS) and flame ionization detector (FID) was used to analyze all organic phase samples. After product identification by mass spectrometry, product concentrations were determined from the integrated FID peak areas using a 6-point calibration curve generated from purchased standards.

Insoluble particulates were quantified by weighing after washing and drying. Upon separation of the insoluble particulates from the liquid phase(s), the solids were washed three times with de-ionized water and dried under a 1.3 × 10⁻⁵ Pa vacuum for 24 h before being weighed.

Xylan was used on a wet basis. Before each use, the xylan was first tested for water content (Karl-Fischer), and then that water content was subtracted from the xylan mass to determine the true xylan mass. The water content for xylan was typically 8 wt %. Product yields were determined on a molar basis relative to the number of product residues present in xylan. For example, the xylose yield was calculated by dividing the moles of xylose produced by the initial number of moles of xylose residues present in xylan. The initial moles of xylose residues in xylan were calculated by dividing 90% of the weighed xylan (the xylose residue containing a portion of xylan as determined by the HPAE analysis of xylan performed by Sigma-Aldrich) by the molecular weight of a xylose residue (150 g mol⁻¹). Humins yields are reported as the mass of recovered humins over the total mass of initial xylan after correction for water content. All reported yields were typically reproducible to within a ±5% relative error (based upon the calculation of one standard deviation).

Results and discussions

Xylan hydrolysis and xylose dehydration

The temporal evolution of the products formed during the H₂SO₄-catalyzed hydrolysis of 5 wt% xylan dissolved in [Emim][Cl] is shown in Fig. 1 (the overall reaction pathway can be seen in Scheme 1). Over the course of the first 20 min, xylobiose and xylose were observed as the primary products. Both product yields increased until xylobiose reached its maximum at 14 wt% after 30 min and xylose reached its maximum at 47 wt% after 60 min. Furfural was produced from the dehydration of xylose after 20 min and reached a level of 18 wt% after 180 min. Glucose and arabinose were also identified by HPLC analysis; however, their yields were relatively small and never exceeded 4 wt% each. The yields of these sugars were consistent with the makeup of xylan, which contains approximately 10 wt% arabinose and glucose residues. The dehydration product of glucose, 5-HMF, was seen at long reaction times, but never in quantities greater than 1 wt%. After 45 min, a black precipitate (humins) began to form and continued to accumulate throughout the duration of the experiment, reaching a level of 10 wt% at the end of the reaction (180 min). Recent work has suggested that this humic material is formed from the coupling loss reaction between xylose and furfural.^17,18 Oligomers larger than xylobiose (xylotriose and xylotetraose) were not detected, which is consistent with work done under similar conditions in water.¹⁹ This would suggest that the hydrolysis of β-1,4 bonds near the xylan chain ends is preferential to inner bond scission. After 60 min, at which point xylose yields were maximal, all xylan had completely reacted. This was evidenced by the observation that with excessive water dilution of the reaction mixture after 60 min, xylan did not precipitate. In total, seven major products were identified in this reaction. Additional soluble products, most likely resulting from the condensation of xylose and furfural or the degradation of these products, accounted for 40 wt% of the original mass of the dissolved xylan. The presence of these unidentified degradation products was indicated by the change in solution color over the course of the reaction, from clear yellow to dark amber.

Fig. 1 Hydrolysis of xylan at 80 °C catalyzed by H₂SO₄ (200 mM) with H₂O (1.8 M). 27 mg xylan in 500 μl [Emim][Cl].

Scheme 1 A representative reaction pathway of a hemicellulose, homoxylan. Major products shown include xylobiose and xylose from the hydrolysis of hemicellulose, furfural from the dehydration of xylose, degradation products from xylose, and humins from the reaction between xylose and furfural.

Fig. 2 shows the H₂SO₄-catalyzed reaction of xylose dissolved in [Emim][Cl]. In the presence of H₂SO₄ after 90 min, 80% of the starting xylose is converted. Furfural is produced via the dehydration of xylose at a yield of 20%, while the other 60% of reacted xylose is converted into unidentified degradation products, both soluble and insoluble (humins, 20 wt%). This reaction illustrates that the rate of xylose degradation is greater than that of xylose dehydration.

Fig. 2 Degradation of xylose at 80 °C catalyzed by H₂SO₄ (200 mM) with H₂O (1.8 M). 27 mg xylose in 500 μl [Emim][Cl].

Kinetics of xylan hydrolysis

The kinetics for the initial rate of xylan hydrolysis can be represented by the following expression:

r₀ = k_app[H₂O]^α[H⁺]^β[β-1,4]^γ (1)

where r₀ is the initial rate of hydrolysis, k_app is the apparent rate coefficient, [H₂O] is the concentration of water, [β-1,4] is the concentration of β-1,4 glycosidic linkages in xylan, [H⁺] is the concentration of free protons (assuming that the acid catalyst is fully dissociated), and α, β, and γ are the orders in H₂O, H⁺, and β-1,4 glycosidic linkages, respectively. It was determined (see Fig. 3) that the initial rate dependence is zero-order with respect to water, by independently varying the water, free proton, and β-1,4 glycosidic linkage concentrations, and first-order with respect to the concentrations of both free protons and β-1,4 glycosidic linkages, yielding the following modified initial rate expression:

r₀ = k_app[H⁺][β-1,4] (2)

Fig. 3 The rate law data for xylan hydrolysis: r₀ = k*[H₂O]^x[H⁺]^y[β-1,4 linkage]^z, where reaction orders were found from the ln–ln plot to be x = 0.011, y = 0.944, and z = 1.0062. Reaction conditions: 353 K, 1.8 M H₂O, 200 mM H₂SO₄, and 400 mM xylan unless otherwise varied to obtain the respective order of magnitude (H₂O was varied from 400 to 5700 mM, H₂SO₄ was varied from 100 to 1000 mM, and the concentration of β-1,4 linkages were varied from 75 to 570 mM).

The form of this equation can be rationalized on the basis of the mechanism presented in Scheme 2. It is envisioned that the first step in the sequence is the reversible protonation of β-1,4 glycosidic linkages, which is then followed by rapid formation of an oxonium cation. The latter species undergoes rapid hydration to form xylose. It should be noted that since this step follows the rate-limiting step, it is kinetically irrelevant, and for this reason the concentration of water does not influence the initial rate of xylan hydrolysis. From Scheme 2, it can be concluded that k_app can be presented as:

k_app = K^‡₁k^‡₂ (3)

where K^‡₁ is the equilibrium constant for the protonation of the β-1,4 glycosidic linkages, reaction 1, and k^‡₂ is the rate coefficient for the formation of the oxonium cation, reaction 2.

Scheme 2 The proposed mechanism for the hydrolysis of xylan, initiated by the reversible protonation of a β-1,4 glycosidic linkage (where K^‡₁ is the associated equilibrium constant), followed by the rapid formation of an oxonium cation (with rate constant k^‡₂) and subsequent rapid hydration to xylose.

The apparent rate coefficient k_app can be further represented in its Arrhenius form as k_app = Aexp(−E_a/RT), where A is the apparent pre-exponential factor, E_a is the apparent activation energy, and T is the temperature. The apparent activation energy was found by performing the hydrolysis of xylan at temperatures ranging from 353 K to 373 K. To avoid the effects of secondary processes, all measurements were made for a xylan conversion of less than 10 wt%. Fig. 4 shows an Arrhenius plot of the initial rate of xylan hydrolysis versus the inverse temperature. The apparent activation energy calculated from this plot is 60 kJ mol⁻¹. This apparent activation energy is considerably lower than that found specifically from the H₂SO₄-catalyzed hydrolysis of xylan (from Birchwood) in water (127 kJ mol⁻¹ over the range 100–130 °C).⁷ Similarly, the activation energy of xylose degradation was calculated to be 72 kJ mol⁻¹ and that of furfural formation from the dehydration of xylose was found to be 114 kJ mol⁻¹.

Fig. 4 Arrhenius plots for xylan hydrolysis (▼, E_a = 60 kJ mol⁻¹; ln (r_o) = −7245.9 × T⁻¹ + 17.739), xylose degradation (○, E_a = 72 kJ mol⁻¹; ln (r_o) = −8823.3 × T⁻¹ + 22.845), and the dehydration of xylose to furfural (●, E_a = 114 kJ mol⁻¹; ln (r_o) = −13 670 × T⁻¹ + 33.819) in [Emim][Cl].

Comparison of the kinetics of xylan and cellulose hydrolysis

The form of eqn (2) is identical to that recently reported for the hydrolysis of cellulose (Avicel) in [Bmim][Cl],¹⁸ and can also be rationalized on the basis of the mechanism presented in Scheme 2. Values of A, E_a, and k_app evaluated at 353, 363, and 373 K are presented in Table 1 for the hydrolysis of xylan and are compared with values of these parameters determined for the initial hydrolysis of cellulose (Avicel) under identical conditions. The data presented in Table 1 show that the activation energy for the hydrolysis of xylan is considerably smaller than that for cellulose hydrolysis, 60 kJ mol⁻¹versus 96 kJ mol⁻¹. Although the pre-exponential factor for xylan hydrolysis is much lower than that for the hydrolysis of cellulose, 5.71 × 10² mL μmol⁻¹ s⁻¹versus 1.31 × 10⁷ mL μmol⁻¹ s⁻¹, the apparent rate coefficient for xylan hydrolysis is about an order of magnitude higher than that for the hydrolysis of cellulose.

Table 1 Comparing the initial rates and kinetic parameters of hemicellulose and cellulose hydrolyses at 353, 363, and 373 K. Both hydrolyses were performed in [Emim][Cl] with initial reactant and H₂SO₄ concentrations of 400 mM and 200 mM, respectively.

	Xylan (from Birchwood)			Cellulose (Avicel)
Ea, kJ mol^−1	60			93
T/K	353	363	373	353	363	373
r0, μmol mL^−1 s^−1	6.21 × 10^−2	1.09 × 10^−1	1.87 × 10^−1	1.84 × 10^−2	3.75 × 10^−2	8.53 × 10^−2
A, mL μmol^−1 s^−1	5.70 × 10^2	5.71 × 10^2	5.72 × 10^2	1.12 × 10^7	1.12 × 10^7	1.11 × 10^7
k, mL μmol^−1 s^−1	7.60 × 10^−7	1.34 × 10^−6	2.28 × 10^−6	2.31 × 10^−7	4.70 × 10^−7	1.07 × 10^−6

---

The significant differences in the rate parameters for the hydrolysis of xylan and Avicel can be attributed to the differences in the structures of the two carbohydrates. Glucose, being a hexose sugar, has an additional CHO group that extends equatorially from the C5 position in the sugar, a feature not present in xylose. This hydroxymethyl group terminates in a primary alcohol that can act as a protecting group in cellulose. The rearrangement of this group (shown in Scheme 3) can result in intramolecular hydrogen bonding between the primary hydrogen of the CHO group and the oxygen of the β-1,4 glycosidic bond.²⁰ This interaction would form a low energy, stable chair conformation that could hinder reaction at the β-1,4 oxygen in a couple of ways. Firstly, the chair conformation would sterically limit the approach space within which the proton can manoeuvre before it is able to react at a glycosidic bond site. Electronically, the hydrogen bonding of the methoxy group to the glycosidic bond oxygen would effectively reduce the number of available electron donor sites at the β-1,4 glycosidic bond for proton attack. Both these steric and electronic hindrances are expected to raise the energy barrier for hydrolysis, providing a potential explanation for the higher observed activation energy for cellulose hydrolysis over that of hemicellulose hydrolysis.

Scheme 3 Rearrangement chemistry of the C5 hydroxymethyl group of cellulose, showing the potential Van der Waals interaction between the active hydrogen of a hydroxymethyl group with a β-1,4 glycosidic bond.

The observed initial rate constant of xylan hydrolysis is approximately 8 times higher on average than that of cellulose under similar reaction conditions over the explored temperature range of 353–373 K. Again, the presence of the additional hydroxymethyl groups within cellulose likely contributed to a lowered initial rate. Unlike the secondary alcohols present in glucose and xylose, the primary alcohol in the methoxy group of glucose has the ability to abstract catalytic protons from solution through an equilibrium protonation–deprotonation reaction,²⁰ lowering the concentration of the catalyst available for hydrolyzing the β-1,4 ether bond in cellulose (see Scheme 4). Thus at the same initial catalyst loadings, fewer catalytic protons are available for cellulose hydrolysis, resulting in a lower initial rate when compared with hemicellulose hydrolysis.

Scheme 4 Equilibrium protonation reaction of the C5 hydroxymethyl oxygen of cellulose in the presence of a strong acid.

Comparing the pre-exponential factor of hemicellulose and cellulose hydrolyses shows that the factor for cellulose is four orders of magnitude larger than that for hemicellulose. This can be interpreted entropically by applying transition state theory to our system using the Erying equation (modified to include E_a):

where the preexponential factor is now defined as:

Solving for ΔS^‡ for both cellulose and hemicellulose shows that the change in entropy for cellulose hydrolysis is approximately 2.8 times higher than for hemicellulose hydrolysis.

Kinetics of xylose dehydration and degradation reactions

Two principal mechanisms have been proposed for the mechanism of xylose dehydration to furfural.^21–25 The first pathway (see Scheme 5) envisions that xylose undergoes a ring transformation in which the C2 hydroxyl group is protonated and leaves the ring as water. The resulting carbocation forms a bond with the ring oxygen, thereby breaking the bond between the ring oxygen and the anomeric carbon. The reactive intermediate, 2,5-anhydroxylose, then further dehydrates to form furfural. Quantum chemical studies have shown that protonation of the C2 hydroxyl exhibits the lowest energy barrier for initiating the dehydration of xylose to furfural compared to all other protonation sites.²¹ Similar results were found for the initiation of glucose dehydration to 5-HMF.²² The second proposed pathway (see Scheme 6) begins with xylose isomerization to its acyclic form and subsequent enolization²³ or direct conversion to xylulose through hydride transfer. Recent NMR labeling studies have shown the hydride transfer mechanism to be the correct one for xylose dehydration occurring in ionic liquids.²⁴

Scheme 5 The proposed reaction pathway for the dehydration of xylose to furfural, initiated by the dehydration of xylose at position 2.

Scheme 6 The proposed reaction pathway for the dehydration of xylose to furfural via an acyclic pathway that features xylulose as a reactive intermediate.

Although the dehydration of xylose to furfural is the most easily characterized reaction stemming from the xylose disappearance in an acidic medium, other side reactions are known to occur. As seen from Fig. 2, approximately 60 wt% of the original xylose is converted to soluble and insoluble products (humins). These degradation products can be attributed to the following reactions: (1) xylose coupling with xylose-to-furfural intermediates; (2) furfural resinification (self-coupling); (3) reaction between furfural and either xylose or xylose-to-furfural intermediates.²⁵ This set of xylose-derived reactions is shown in Scheme 7 and can be modeled using the following ordinary differential equations:

where [X], [I], [F], [H], [D], and [R] are the xylose, intermediate product, furfural, humins, degradation product, and resinification product concentrations, respectively. These equations can be simplified by considering experimental results. Xylose-to-furfural intermediates were not observed in the present study during product analysis, nor have they been clearly identified in the literature. Thus, the steady-state assumption can be invoked in the modeling of the intermediates such that d[I]/dt = k₁[X] − k₄[I] ≈ 0, and hence k₁[X] ≈ k₄[I]. Furthermore, we can assume that resinification does not occur to a significant extent based on the recent observation that furfural dissolved in [Bmim][Cl] is relatively stable in the presence of an acid catalyst and in the absence of xylose.²¹ These assumptions lead to a modified differential equation representing change in furfural concentration with respect to time:

Scheme 7 The proposed xylose dehydration/degradation pathway (including secondary reactions) used in kinetic modeling.

Values for k₁, k₂, and k₃ were determined by least squares minimization of the residuals between the predicted concentrations of X, H, D, and F, obtained by solving eqn (6), (9), (10), and (12), and the experimental data shown in Fig. 2. A good fit to the data could be obtained with the rate coefficients listed in Table 2, with r² fit values of 0.98 and 0.99 for xylose and furfural, respectively (see Fig. D in the ESI†). These rate coefficients reveal that degradation products are formed 20% faster than humins.

Table 2 The reaction rate constants were determined by least squares minimization of residuals between predictions of kinetic models and the experimental data from Fig. 2 (for xylose dehydration) and Fig. 1 (for xylan hydrolysis) using the lsqcurvefit and ode45 routines within MATLAB (R2010b).

Xylose Dehydration
k 1/10^−4, [L mol^−1s^−1]	k 2/10^−3, [L^2 mol^−2s^−1]	k 3/10^−3, [L^2 mol^−2s^−1]
5.38	3.18	4.40
Xylan Hydrolysis
k x1/10^−4, [L mol^−1s^−1]	k x2/10^−4, [L mol^−1s^−1]		k xb/10^−3, [L mol^−1s^−1]
4.76	3.73		5.34

---

Similarly, the rate coefficients for the parent reactions involved in the hydrolysis of xylan to xylobiose and xylose (shown in Scheme 8) were found using the previously determined constants (k₁, k₂ and k₃) and the following set of equations:

where [β-1,4] is the concentration of β-1,4 glycosidic linkages in xylan and [XB] is the xylobiose concentration. A satisfactory fit to the data presented in Fig. 1 could be obtained with these rate coefficients (listed in Table 2), with r² fit values of 0.91, 0.93 and 0.94 for xylobiose, xylose, and furfural, respectively (see Fig. E in the ESI†). These rate constants show that xylan hydrolysis to xylose is 28% faster than xylan hydrolysis to xylobiose, while xylobiose hydrolysis is an order of magnitude faster than xylan hydrolysis. These trends are consistent with those found for the hydrolysis of xylan in water.⁷ We also find that xylose dehydrates to furfural 13% faster than xylan hydrolyzes to xylose. Finally, it is observed that the sum of k_x1 and k_x2 approximates the value of k_app reported in Table 1.

Scheme 8 The xylan hydrolysis scheme used in kinetic modeling.

Role of water in xylose dehydration and degradation reactions

While the initial rate of xylan hydrolysis was found to have a zero-order dependence on water, the reactions that occur beyond hydrolysis have an observable dependence on the concentration of water. Fig. 5 shows the temporal evolution of xylose from xylan for a range starting from 400 mM (1 mole of water per mole of β-1,4 glycosidic bonds) to 6 M (15 moles of water per mole of β-1,4 glycosidic bonds). As expected, at short reaction times the concentration of water has virtually no effect on xylose yield, as hydrolysis is the principal reaction. However, over longer periods of time, higher water concentrations enhance the production of xylose, increasing maximum yields from 37% with 400 mM of water to 57% with 6 M of water. Over this range of water concentrations, the yields of furfural (not shown) decreased from 15% to 8%, and the formation of humins decreased (as judged by the visual appearance of the reactions solutions) with increasing initial water concentration.

Fig. 5 The effect of the starting water concentration on the acid-catalyzed (200 mM H₂SO₄) dehydration of xylan at 80 °C in [Emim][Cl].

Various hypotheses have been offered to explain the ability of water to inhibit the extent of xylose dehydration and degradation reactions. It has been suggested that Le Chatelier's principle can explain the shift in equilibrium away from dehydration products in favor of xylose in the presence of higher concentrations of water.⁹ However, this seems unlikely, as it has been reported that starting with excess furfural and performing the xylose dehydration reaction does not impede the complete conversion of xylose.¹⁷ Other authors have suggested that water is a stronger base than any of the hydroxyl groups of xylose, and thus would be preferentially protonated.²¹ In support of this suggestion is an ab initio molecular dynamics study that examined the affinity of free protons for xylose in the presence of differing numbers of water molecules.²⁶ The authors of this work show that water molecules compete with the hydroxyl groups on xylose for available protons, and that, in the presence of enough water molecules, solvent protonation is preferred over xylose protonation, resulting in the termination of the xylose dehydration pathway. Our observations of an inverse relationship between initial water concentration and furfural/humins yields are consistent with these findings.

The results of these experiments do not, however, suggest that xylan hydrolysis experiments should be initiated with a high concentration of water in order to achieve optimal yields of xylose. Xylan, like cellulose, is not water-soluble and will precipitate completely from an ionic liquid–water solution containing >20 M water, resulting in the quenching of xylose hydrolysis. Recent studies have addressed this issue by employing a multistage water addition strategy to maximize glucose yields in the depolymerization of cellulose in ionic liquid.^9,18 In this strategy, water was added at varying time intervals such that as the hydrolysis of cellulose proceeded, shorter polymer chains with higher water tolerances would evolve. Building on these results, we tested various intervals and quantities of water addition to develop an optimal water addition strategy for xylan hydrolysis in [Emim][Cl]. Table 3 details these results. Using the method that provided the best overall xylose yield, we found that we could double the xylose yields obtained when water was only introduced at the start of the reaction. Gradually introducing water from an initial 9 wt% to 48 wt% over 60 min (entry 3 of Table 3), the xylose yield increased from 45% to 90%, while furfural yields were reduced to 5% from 15% over the course of 120 min. Additionally, humin content was measured to be only 4% at reaction termination. To compare under similar conditions where an optimized water addition strategy was used, one study found that cellulose (Avicel) could be hydrolyzed to 76 wt% glucose, 10 wt % 5-HMF, and 0 wt % humins after 90 min at 105 °C.¹⁸

Table 3 The effect of multistage water addition on the hydrolysis of xylan. Reaction conditions: 80 °C, 200 mM H₂SO₄, 27 mg xylan in 500 μl [Emim][Cl]

Water Content, wt %^a
a Represents the total water present in weight percentage at the given time.
Entry	0 min	15 min	30 min	60 min	Time, minutes		Xylose Yield (%)	Furfural Yield (%)
1	3%	—	—	—	15		23%	2%
					30		42%	5%
					60		44%	11%
					90		33%	16%
2	7%	—	—	—	15		23%	2%
					30		36%	3%
					60		49%	9%
					90		44%	14%
3	9%	—	—	—	15		19%	0%
					30		40%	3%
					60		56%	7%
					90		48%	12%
4	7%	12%	26%	39%	15		19%	0%
					30		33%	0%
					60		61%	0%
					90		68%	6%
					120		60%	10%
5	9%	17%	34%	48%	15		15%	0%
					30		35%	0%
					60		57%	0%
					90		79%	0%
					120		90%	4%
6	9%	20%	40%	58%	15		13%	0%
					30		28%	0%
					60		57%	0%
					90		56%	0%
					120		61%	0%

---

Conclusions

We have found that hemicellulose (xylan) can be easily hydrolyzed in [Emim][Cl] to its primary sugar, xylose, in high yields with minimal dehydration and degradation products (furfural and humins). Desirable product yields showed marked increases when a multi-stage water addition strategy was employed. A kinetic model defining all essential reactions involved in, and stemming from, xylan hydrolysis was developed and discussed. Xylan hydrolysis was found to exhibit faster initial reaction rates than cellulose hydrolysis under similar conditions in ionic liquid. Both xylan hydrolysis and xylose degradation showed lower activation energies over cellulose hydrolysis and glucose degradation, respectively. These observations were attributed to the inherent structural differences between xylan and cellulose.

Acknowledgements

This work was supported by the Energy Biosciences Institute.

References

1. G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006, 106(9), 4044–4098.
2. (a) M. Kleinert and T. Berth, Energy Fuels, 2008, 22(2), 1371–1379; (b) D. S. Argyropoulos and S. Menachem, Adv. Biochem. Eng./Biotech., 1997, 57, 127–158.
3. (a) T. Werpy, G. Peterson, A. Aden, J. Holladay, J. White and A. Manheim, Top Value-Added Chemicals from Biomass Volume I—Results of Screening for Potential Candidates from Sugars and Synthesis Gas, Official PNNL and NREL Report, 2004, 1, 1–77; (b) J. E. Holladay, J. J. Bozell, J. F. White and D. Johnson, Top Value-Added Chemicals from Biomass Volume II—Results of Screening for Potential Candidates from Biorefinery Lignin, Official PNNL and NREL Report, 2007, 2, 1–87.
4. M. B. Jones and M. Walsh, Miscanthus for Energy and Fibre, James & James, London, 2001, pp. 109–154.
5. B. C. Saha, J. Ind. Microbiol. Biotechnol., 2003, 30(5), 279–291.
6. (a) C. E. Wyman, S. R. Decker, M. E. Himmel, J. W. Brady, C. E. Skopec and L. Viikari, Hydrolysis of Cellulose and Hemicellulose, Polysaccharides: Structural Diversity and Functional Versatility, ed. Severian Dumitriu, Marcel Dekker, New York, 2nd edn, 2005, pp. 995-1034; (b) S. E. Jacobsen and C. E. Wyman, Appl. Biochem. Biotechnol., 2000, 84–86(1–9), 81–96.
7. P. Mäki-Arvela, T. Salmi, B. Holmbom, S. Willfor and D. Y. Murzin, Synthesis of Sugars by Hydrolysis of Hemicelluloses—A Review, Chem. Rev., 2011, 111(9), 5638–5666.
8. B. P. Lavarackm, G. J. Griffin and D. Rodman, Biomass Bioenergy, 2002, 23, 367–380.
9. J. B. Binder and R. T. Raines, Proc. Natl. Acad. Sci. U. S. A., 2010, 107(10), 4516–4521.
10. (a) J. B. Binder and R. T. Raines, J. Am. Chem. Soc., 2009, 131(5), 1979–1985; (b) J. B. Binder, M. Gray, J. White and Z. Zhang, Biomass Bioenergy, 2009, 33, 1122–1130.
11. J. N. Chheda, Y. Roman-Leshkov and J. A. Dumesic, Green Chem., 2007, 9(4), 342–350.
12. R. Swatloski, S. Spear, J. Holbrey and R. Rogers, J. Am. Chem. Soc., 2002, 124, 4974–4975.
13. O. A. El Seoud, A. Koschella, L. C. Fidale, S. Dorn and T. Heinze, Biomacromolecules, 2007, 8(9), 2629–2647.
14. H. Olivier-Bourbigou, L. Magna and D. Morvan, Appl. Catal., A, 2010, 373, 1–56.
15. D. MacFarlane and S. Forsyth, Acids and Bases in Ionic Liquids, Ionic Liquids as Green Solvents: Progress and Prospects, ed. R. D. Rogers and K. R. Seddon, Oxford University Press, ACS Symposium Series 856, Washington, D.C., 2003, pp. 264–276.
16. L. Vanoye, M. Fanselow, J. D. Holbrey, M. P. Atkins and K. R. Seddon, Green Chem., 2009, 11(3), 390–396.
17. C. Sievers, I. Musin, T. Marzialetti, M. B. Valenzuela Olarte, P. K. Agrawal and C. W. Jones, ChemSusChem, 2009, 2(7), 665–671.
18. S. Dee and A. T. Bell, ChemSusChem, 2011, 4(8), 1166–1173.
19. O. Akpinar, K. Erdogan, U. Bakir and L. Yilmaz, LWT–Food Sci. Technol., 2010, 43, 119–125.
20. T. W. G. Solomons and C. B. Fryle, Organic Chemistry, ed. D. Brennan, John Wiley & Sons, New Jersey, 8th edn, 2004.
21. M. R. Nimlos, X. Qian, M. Davis, M. E. Himmel and D. K. Johnson, J. Phys. Chem. A, 2006, 110, 11824–11838.
22. X. H. Qian, M. R. Nimlos, M. Davis, D. K. Johnson and M. E. Himmel, Carbohydr. Res., 2005, 340, 2319–2327.
23. K. J. Zeitsch, Sugar Series, 2000, 13, 1–353.
24. J. B. Binder, J.J. Blank, A. V. Cefali and R. T. Raines, ChemSusChem, 2010, 3(11), 1268–1272.
25. G. Marcotullio and W. de Jong, Carbohydr. Res., 2011, 346(11), 1291–1293.
26. X. H. Qian, M. R. Nimlos, D. K. Johnson and M. E. Himmel, Appl. Biochem. Biotechnol., 2005, 124, 989–997.

---

Footnote

† Electronic Supplementary Information (ESI) available. See DOI: 10.1039/c2ra21650g

---