In Situ Kinetic Study on Hydrothermal ... - ACS Publications

Nov 9, 2011 - Hiroshi Kimura†, Masaru Nakahara†, and Nobuyuki Matubayasi*†‡. Institute for Chemical Research, Kyoto University, Uji, Kyoto 611...
0 downloads 0 Views 1MB Size
ARTICLE pubs.acs.org/JPCA

In Situ Kinetic Study on Hydrothermal Transformation of D-Glucose into 5-Hydroxymethylfurfural through D-Fructose with 13C NMR Hiroshi Kimura,† Masaru Nakahara,† and Nobuyuki Matubayasi*,†,‡ † ‡

Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan Japan Science and Technology Agency (JST), CREST, Kawaguchi, Saitama 332-0012, Japan ABSTRACT: Kinetics of hydrothermal reaction of D-glucose was investigated at 0.02 M over a temperature range of 120160 °C by applying in situ 13C NMR spectroscopy. D-Glucose was found to be reversibly transformed first into D-fructose (intermediate) and successively into 5-hydroxymethylfurfural (5-HMF) through dehydration. The carbon mass balance has been kept within the detection limit, and no other reaction pathways are present. The hydrothermal reaction of D-glucose is thus understood as that of D-fructose in the sense that the D-glucose reaction proceeds only through D-fructose. All the isomers of D-glucose and D-fructose were detected by the in situ 13C NMR in D2O: they are the open chains and the pyranoses and furanoses of α- and β-types. The β-forms are the most stable due to the hydration. For both D-glucose and D-fructose, the isomers are in a rapid equilibrium for each monosaccharide, and they are treated collectively in the kinetic analysis of the slower hydrothermal reactions. The reactions are of the first order with respect to the concentrations of D-glucose and D-fructose, and D-glucose converts to 5-HMF on the order of hours. The kinetic parameters were determined by the in situ method.

1. INTRODUCTION The conversion of biomass molecules (cellulose and starch) into smaller molecules that can be used as renewable fuels or starting compounds for drug synthesis has recently attracted much attention from the viewpoint of green chemistry.19 D-Glucose is the building block of biomass molecules. It is thus of considerable interest to elucidate the reaction kinetics and pathways for the transformation from D-glucose into useful and renewable organics. In view of the increasing demand of earth-benign processes, development of hydrothermal treatment of D-glucose is of importance. There have been a variety of promising applications of D-glucose to produce liquid and gas fuels1020 and to synthesize nanostructured carbonaceous materials used for device components.2129 However, studies on the kinetics and pathways of D-glucose (or saccharides) are still in an early stage of progress, probably because of the complexities due to the presence of a variety of ring and open chain isomers of sugars. For developing further applications of the biomass, it is required to clarify the reaction pathways and elucidate the kinetics for the D-glucose hydrothermolysis. In previous studies on the hydrothermal reactions of D-glucose in supercritical water1015 above 374 °C and at lower temperatures in subcritical water,1020 many kinds of products have been reported on the basis of the product analysis by quenching method. A limitation of the quenching method is that transient species cannot be detected quantitatively or often even qualitatively. In this work, we perform quantitative, kinetic analysis of Dglucose hydrothermolysis to disclose its reaction pathways, using the time-resolved spectroscopic information on all the reactive species. We discuss the following reaction pathways 3H2 O

f 5-HMF fD-glucose isomersg h fD-fructose isomersg s ð1Þ r 2011 American Chemical Society

where the isomers are collectively expressed by the curly braces. The pyranose and furanose isomers are shown to be in a rapid equilibrium with each other by means of in situ NMR spectroscopy. The hydrothermal decomposition of D-glucose has been studied with the quenching method in many papers,1015,3033 but it has never been clarified whether D-glucose is reversibly transformed into D-fructose or not. We demonstrate that Dglucose is reversibly transformed into D-fructose as a first step and that D-fructose is successively dehydrated to form 5-hydroxymethylfurfural (5-HMF). There are no byproducts as confirmed by the mass balance. Here D-fructose is found to be a stable reaction intermediate according to the in situ spectroscopic analysis. The presence of a variety of isomers makes it difficult to analyze the reactions of saccharides. To overcome the difficulty, we have taken advantage of in situ 13C NMR spectroscopy; by this we can distinguish the isomers of the reactant and intermediate and the final product. At the same time, we can check the total mass balance that is essential for reliable kinetic analysis. As another merit, 13C NMR spectroscopy can be combined with a site-selective labeling technique that enables us to identify competing isomers. In the present kinetic study, we have examined a relatively low temperature range 120160 °C. Then, we can slow down the reactions and improve time resolution of spectroscopy so that all the reactive species generated through the reactions can be quantified as a function of time. The experimental procedure is described in the following section 2. In subsection 3.1, we establish the reaction pathways for the D-glucose hydrothermolysis on the basis of the in situ Received: July 6, 2011 Revised: October 21, 2011 Published: November 09, 2011 14013

dx.doi.org/10.1021/jp206355e | J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A analysis for all the reactive species including the isomers. In subsection 3.2, the time evolutions of all the reactive species are examined, and then the kinetic parameters are determined. Conclusions are given in section 4.

2. EXPERIMENTAL SECTION D-Glucose (Nacalai, 99.9%), (ISOTEC, 99 atom % 13C), D-[2,5-13C]-glucose (ISOTEC, 99 atom % 13C), D-[6-13C]-glucose (ISOTEC, 99 atom % 13C), D-[1,2,3,4,5,6-13C]-glucose (ISOTEC, 99 atom % 13C), 13 13 13 D-[1- C]-fructose (ISOTEC, 99 atom % C), D-[2- C]-fructose 13 (ISOTEC, 99 atom % C), and deuterated water, D2O (ISOTEC, 99.95 atom % D), were used without further purification. In the kinetic study, D-[1-13C]-glucose was used as a reactant and weighed to be 0.02 M (mol dm3) in D2O; for determining the reaction order with respect to the concentration of D-glucose or D-fructose, D-glucose solution prepared at 0.1 M was also examined. The sample was then loaded into a Pyrex NMR tube (SHIGEMI; 7.0-mm i.d. and 8.0-mm o.d.). The sample tube was sealed using a gas burner after the air in the reactor was replaced by argon. The reactor tube was 6 cm in height. The filling factor, which was defined as the ratio of the solution volume to the vessel volume at room temperature, was fixed at 0.75 (75%). For the full determination of saccharide isomers, saturated solution of nonlabeled D-glucose was prepared at 100 °C (∼30 M). When the temperature is lowered to 30 °C, it remained supersaturated. For peak assignments of D-glucose and D-fructose isomers in the 13C NMR spectra, the labeled compounds listed above were used as an authentic sample. 2.2. Apparatus and Procedures. A sealed sample tube was loaded on a NMR probe of 600-MHz spectrometer (ECA, JEOL) for the in situ measurements.34,35 The temperature was set at 120, 130, 140, 150, and 160 °C and was controlled within (2 °C. The temperature calibration was performed with an alumelchromel thermocouple using ice water as a reference. In our temperature range, the reaction system is in the liquid branch of the gas-liquid coexistence curve, and the pressure and the water density are 0.20.6 MPa and 0.9080.943 g cm3, respectively. The in situ 13C NMR measurement was performed without proton irradiation; 1H was not decoupled to keep the spectral intensity proportional to the number of carbon atoms. Thus, the number of carbon atoms can be determined, leading to the accurate mass balance analysis. We quantitatively determined the concentrations of the reactant and reactive species on the basis of their integrated spectral intensities averaged for 32128 scans. It took 520 min for one measurement, which is short enough compared to the time scale of the concentration variations associated with the hydrothermal reactions discussed below; the effect of the measurement time on the reaction progress thus needs not to be taken into account. The maximum (relative) error of a minor peak signal with low signal-to-noise ratio (S/N) was ∼5%, whereas a major peak signal with high S/N had less than 0.5% error in the peak integration. When more than one peak component is present in the system, the overall error, Δoverall, is typically much smaller than the component Δi because of (Δoverall)2 = Σfi2Δi2; here fi denotes the fraction of species i. The overall error like that involved in the mass balance is dominated by the main species with large fi and small Δi. The number of the major species is more than one, and the component errors are expected to cancel each other. Thus, the precision (relative error) of the following

2.1. Sample Preparation.

13 D-[1- C]-glucose

ARTICLE

mass balance data is better than of the data for each reactive species. For each of the ring and chain isomers of D-glucose and D-fructose shown in Figure 1, we performed ab initio MO calculations in vacuum and in the PCM (continuum) water using GAUSSIAN 09 program.36 The geometry was optimized using the hybrid density functional B3LYP with the correlation consistent polarized valence triple-ζ (cc-pVDZ) basis set. At the optimized geometry, a single-point energy calculation was carried out at the B3LYP level of theory with the augmented cc-pVDZ (aug-cc-pVDZ) basis set.

3. RESULTS AND DISCUSSION For understanding the reaction pathways and clarifying the kinetics of the hydrothermolysis of D-glucose, it is required to detect and quantitatively analyze all the reactive species generated through the reaction. To this end, the high-resolution in situ 13 C NMR measurements are carried out as a function of time by taking advantage of the site-selectively labeled D-[1-13C]-glucose. In subsection 3.1, we demonstrate the reaction scheme for the hydrothermal conversion of D-glucose via D-fructose into 5-hydroxymethylfurfural (5-HMF). In 3.1.1, we validate the reaction scheme on the basis of the spectral evidence. In 3.1.2, we describe the temperature dependence of the isomer fractions. In subsection 3.2, we perform quantitative kinetic analysis. In 3.2.1, we show the time evolutions of all the reactive species (including isomers) with the total mass balance confirmed. In 3.2.2, we establish the rate laws and determine the in situ kinetic parameters. 3.1. Reaction Pathways. We will prove the reaction pathways for the hydrothermal transformation of D-glucose shown in Figure 1. In the reaction scheme, D-glucose, a six-carbon aldose,37 is reversibly (the rate constants are denoted as k+1 and k1, respectively) and rather slowly transformed into D-fructose, a sixcarbon ketose,37 as the first step via the open-chain isomer and successively (k2) into 5-hydroxymethylfurfural (5-HMF). To be precise, various isomers of D-glucose or D-fructose in Figure 1 are collectively treated as a single unit in the kinetic analysis, and the kinetic parameters k+1, k1, and k2 are introduced into the rate equation formulated as such. The application of the in situ 13C NMR spectroscopy has enabled us to detect all of the reactive species (see 3.1.1) and demonstrate that the six- and five-membered ring as well as open chain isomers of D-glucose are in a rapid equilibrium with each other (see thick arrows in Figure 1) and that the situation is the same as that for the case of the six- and five-membered ring isomers of the intermediate, D-fructose (see 3.2.2). Previous works proposed a similar reaction network for the hydrothermolysis of D-glucose and D-fructose.10,13,15,38 In the present study, the reactions in Figure 1 are more elementary. Their kinetic parameters will be addressed below at high precision, and need not be treated through multivariate fitting to the complicated reaction network. In the present kinetic analysis, the saccharide isomers in a rapid equilibrium are treated collectively. In Figure 1, the pyranoses and furanoses of α- and β-types are denoted as 6-α, 6-β, 5-α, and 5-β, respectively, for both D-glucose and D-fructose. In 3.1.1, the reaction scheme is validated by using the time-resolved in situ 13 C NMR for the reactant D-glucose enriched at C-1. 3.1.1. Spectral Evidence for Reactive Species and Isomers. To establish the rate laws corresponding to the reaction scheme given in Figure 1, it is essential to identify and quantify all the 14014

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A

ARTICLE

Figure 1. Proposed reaction pathways for the hydrothermolysis of D-glucose. Curly braces mean that all the isomers of each sugar are collectively treated in the kinetic analysis. For D-fructose, the configurations of 6-α and 5-α forms are omitted here for brevity. For the kinetic analysis, k+1, k1, and k2 are defined as the rate constants for the forward and backward transformation from D-glucose into D-fructose and the conversion of D-fructose into 5-HMF, respectively.

reactive species including the isomers shown in Figure 1 as functions of time. When we do the kinetic analysis relying upon eq 1 we need to confirm how the rapid enough equilibrium populations of D-glucose and D-fructose isomers depend on the concentration, the temperature, and the reaction time. These effects are discussed below by means of the time-resolved in situ 13C NMR kinetic analysis. Figure 2 shows the in situ 13C NMR spectrum of saturated D2O solution (∼30 M) of nonlabeled D-glucose at 100 °C after a long accumulation time of 20 h (a), and those of site-selectively

labeled D-[1-13C]-glucose solution (0.02 M) for the reaction times 15 (b) and 405 (c) min at 160 °C. As seen in panel a, we have succeeded in detecting all of the isomers, such as chains and fiveand six-membered rings of α- and β-types in saturated (∼30 M) solution of D-glucose. The peaks were determined by employing all and site-selectively labeled D-glucose and D-fructose. The low-field portion of the spectrum is expanded in the left inset. The open chain forms of D-glucose and D-fructose are observed at 205.0 and 213.1 ppm, respectively, for the carbonyl carbons. 14015

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A

ARTICLE

indicates that the decomposition reactions are extremely slow at 100 °C. The singlet C-1 peaks (64.8 and 65.6 ppm) in the expanded spectrum in the right inset, which are assigned to the 5-β and 6-β forms of D-fructose, respectively, appear to be weaker than those of the corresponding C-2 peaks (not shown here). This is caused by the 13C2H coupling due to the partial hydrogen/deuterium exchange occurring via the ring-opening; more detailed discussion is given below on the site-selectively labeled C-1 peaks (Figure 2b). The fraction of the chain form of D-glucose is 0.04% (∼12 mM) against the overall D-glucose isomers. For D-fructose, the chain isomer is 5% (∼10 mM) of D-fructose generated. When the temperature is lowered to 30 °C with the solution supercooled, the open chain isomers are scarcely detected in comparison to the broadened peak for 5-HMF.39 At 100 °C, the fractions of the chain and ring isomers for D-glucose are in the following order 6  βð54%Þ > 6  αð44%Þ . 5  βð1%Þ > 5  αð0:5%Þ > chainð0:04%Þ

ð2Þ where the numbers in the parentheses are the fraction (%) of interest normalized by the overall D-glucose isomers.40 For Dfructose, the isomers are in the following order 6  βð48%Þ > 5  βð36%Þ > 5  αð8%Þ > chainð5%Þ > 6  αð4%Þ

ð3Þ

where similarly the fractions are determined against the overall generated. The following tendencies are observed: (i) in D-glucose, the pyranose forms, 6-β and 6-α, are dominant with the sum close to 100%; (ii) in D-fructose, the β types, 6-β and 5-β, are predominant; (iii) the fraction of the chain form of D-glucose is smaller by a factor of ∼100 than that of D-fructose; and (iv) the furanose forms, 5-β and 5-α, are more preferable in D-fructose than in D-glucose. These tendencies experimentally observed can be interpreted by focusing on the role played by the hydration of the multiple OH groups; see the molecular structures in Figure 1. When the OH group is in the equatorial direction in the pyranose or in the pseudoequatorial in the furanose more hydration waters can be accommodated. Hence, the stability of the ring isomers is expected to be correlated with the difference in the number between the equatorial or pseudoequatorial and the axial or pseudoaxial in the presence of the strong hydration interaction. The difference values for the ring isomers of D-glucose are in the following order D-fructose

6  βð4Þ > 6  αð2Þ . 5  βð  1Þ > 5  αð  3Þ Figure 2. (a) 13C NMR spectrum of saturated (∼30 M) solution of nonlabeled D-glucose at 100 °C with a long accumulation time of 20 h, until which the reaction proceeds slightly. The peak assignments are determined by using all and site-selectively labeled D-glucose and Dfructose. The parenthesized numbers are equal to the site number of carbon atom shown in Figure 1. Though all the isomers of D-glucose and D-fructose are detected, the weak peaks derived from the minor isomers are not all shown here for clarity. (b, c) In situ 13C NMR spectra of 13 D-[1- C]-glucose at 0.02 M after reacting for 15 and 405 min at 160 °C, respectively. The C-1 chemical shifts of the chain and ring isomers of Dglucose and D-fructose and 5-HMF are summarized as follows: D-glucose isomers (chain, 205.0 ppm; 6-α, 92.5 ppm; 6-β, 97.4 ppm; 5-α, 97.7 ppm; 5-β, 102.8 ppm), D-fructose isomers (chain, 213.1 ppm; 6-α, 66.2 ppm; 6-β, 65.6 ppm; 5-α, 64.4 ppm; 5-β, 64.8 ppm), and 5-HMF (180.9 ppm).

All of the ring isomers are also detected at the high magnetic field. The intermediate and final products are weakly observed, which

ð4Þ

which is in excellent agreement with eq 2. For D-fructose, the difference values for the ring isomers are in the following order 5  βð3Þ > 6  βð2Þ > 5  αð1Þ > 6  αð0Þ

ð5Þ

which agrees with eq 3 except for the first two isomers. However the agreement comes up at the lower concentration of 0.02 M as described below. Thus, we can establish an empirical rule that the stability of the isomers depends on the difference between the number of equatorial or pseudoequatorial OH groups and that of axial or pseudoaxial OH groups. In the dilute aqueous solution (0.02 M), as shown in Figure 2b, not all but most of the isomers are confirmed. Only D-glucose and D-fructose are observed at the early stage of the reaction (15 min), cf. Figure 2c. Here let us compare the isomer fractions of D-glucose and D-fructose in the dilute (0.02 M) solution with those in the saturated solution (∼30 M) at 100 °C to understand 14016

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A

ARTICLE

Figure 3. Temperature dependence of the isomer populations of Dglucose (red) and D-fructose (blue) in equilibrium.

the concentration effect on the isomer stability; in the case of 0.02 M solution, the isomer fractions were determined through extrapolation of the higher-temperature data in Figure 3. The isomer fraction for D-glucose in 0.02 M solution thus determined is given in the following order 6  βð58%Þ > 6  αð40%Þ . 5  βð2%Þ > 5  αð0:5%Þ

ð6Þ

which corresponds to eqs 2 and 4. Thus, the concentration effect on the isomer fractions for D-glucose is negligibly small. For Dfructose, the isomer fractions are given as 5  βð48%Þ > 6  βð44%Þ > 5  αð8%Þ

ð7Þ

which is consistent not with eq 3 but with eq 5. This is due to the concentration effect on the hydration. The more the hydration waters increase, the stronger the hydration effects are. Note that the chain form does not appear in eqs 6 and 7 since its concentration is too low for NMR detection in the 0.02 M solution. The hydration effect experimentally observed on the isomer stability can be understood from quantum-chemical calculations of each of the ring and chain isomers for D-glucose and D-fructose in vacuum and in the PCM water. For D-glucose, the stability orders of the isomers are the following: 5  β > 6  α > 5  α > 6  β > open chain ðvacuumÞ

ð8Þ

6  α > 6  β > 5  α > 5  β > open chain ðPCM waterÞ

ð9Þ

The stability sequence is reversed by the hydration according to the continuum model (PCM water). The computed stability order in eq 9 is in fairly good agreement with the experimental results, eq 6. For D-fructose isomers, the stability sequences are as follows: 6  α > 5  β > 5  α > 6  β > open chain ðvacuumÞ

ð10Þ

5  β > 6  β > 6  α > 5  α > open chain ðPCM waterÞ

ð11Þ

Similarly, the computed result in eq 11 is comparable with the experimental results in eq 7. Is there any evidence for the ketoenol tautomerism involved in the transformation between the chain and ring isomers? Note that the C-1 peaks (64.4, 64.8, and 65.6 ppm) derived from the ring isomers of D-fructose are split into a triplet, as illustrated in Figure 2b. This is attributed to the 13C2H coupling at C-1 as briefly discussed in the corresponding Figure 2b. If the mechanism is to follow the ketoenol tautomerism, hydrogen/deuterium

Figure 4. Time evolutions of the concentrations of the reactant and reactive species for the hydrothermal reaction of 0.02 M D-glucose at 120 and 160 °C, respectively. The vertical axis on the left shows the concentration normalized by the initial concentration of D-glucose, and that on the right shows the 13C-based mass balance (M.B.).

exchange can occur at C-1 due to deuterated water molecules as solvent. The triplet 13C signal, therefore, demonstrates the presence of tautomerism. It is of interest that the hydrothermal conversion of D-glucose into D-fructose takes place without base catalyst. The absence of base catalyst differentiates the hydrothermal transformation from the Lobryde Bruynvan Ekenstein transformation,41 the classical base-catalyzed rearrangement reaction between aldoses and ketoses in ambient conditions. In our previous study on the hydrothermal reactions of formaldehyde in sub- and supercritical water, it is found that disproportionation reactions proceed without base catalyst and are sharply different from the base-catalyzed Cannizzaro reaction. The keto enol tautomerization under hydrothermal conditions also occurs without base catalyst in common with the hydrothermal disproportionation.42,43 3.1.2. Temperature Dependence of Isomer Fraction. Here let us examine how the distribution of the ring isomers varies with temperature. In Figure 3, we illustrate the plots of the isomer fractions of D-glucose and D-fructose as functions of temperature. For D-glucose, the 6-β decreases very slightly with increasing temperature, and the other isomers increase correspondingly. Thus, the temperature effect on the isomer populations is considerably small. For D-fructose, the same tendency can be said of the isomer fraction, but the isomers are more subjected to the temperature effect compared with D-glucose. The temperature effect is related to the hydration; to be precise, at higher temperatures the hydrogen-bond network between adjacent water molecules is more broken down and the hydrogen bonds between solute and water molecules are more formed. Therefore, the tendencies observed can be interpreted from the viewpoint of the empirical rule mentioned above. For D-glucose, the 6-β and 6-α are considered to be stabilized to almost the maximum extent by the hydration in comparison with the other ring isomers (see eq 4), whereas for D-fructose the stabilities of the ring isomers are expected to be close to each other (see eq 5). Thus, D-fructose can more strongly reflect the hydration effect compared with D-glucose. 3.2. Kinetic Analysis. On the basis of the time evolution of all the chemical species detected by in situ 13C NMR spectroscopy, we have established the rate laws, solved the rate equations, and determined the kinetic parameters defined in the reaction scheme in Figure 1. 14017

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A

ARTICLE

Here, t is the time, k+1, k1, and k2 are the rate constants defined in each equation as well as in Figure 1, and the square and curly braces denote the concentration and the set of all the isomers of each sugar, respectively. The stationary state of D-fructose enables us to solve the rate laws. Eq 14 is simply integrated with time as follows: ½5  HMF ¼ k2 ½fD  fructoseg  t

Figure 5. Populations of D-glucose and D-fructose isomers after reaction for 45, 195, and 405 min at 160 °C.

3.2.1. Time Evolution. It is shown in Figure 4 how the reactant is depleted and reactive species are evolved with time at the lowest and highest reaction temperatures of 120 and 160 °C. As D-glucose is exponentially decreased from the initial concentration of 0.02 M, only D-fructose rises at the early stage until ∼160 and ∼15 min, respectively, at 120 and 160 °C. D-Fructose reaches a stationary state (plateau), and 5-HMF is gradually generated all over the reaction time covered here.44 5-HMF continues to increase with D-glucose consumed. The stationary concentration of the intermediate D-fructose increases with temperature; the concentration at 160 °C is larger by a factor of 1.3 than that at 120 °C. The carbon mass balance is kept within the limit of NMR detection during the reaction time; no side-reactions are present, and only the reaction pathways of Figure 1 are to be kinetically analyzed. 3.2.2. Rate and Equilibrium Constants. First we show the evidence that all of the isomers can be treated collectively for the kinetic analysis before going to discuss the rate laws for the hydrothermal transformation of D-glucose into 5-HMF through D-fructose. Looking at Figure 5 one can find that the isomer fractions of D-glucose and D-fructose are constant irrespective of the presence of the final product 5-HMF all over the reaction time investigated here. This clearly indicates that most of the isomerization rate constants are significantly larger than those not only for the forward and backward transformation from D-glucose into D-fructose (k+1 and k1) but also for the formation of 5-HMF (k2). All of the isomers are, therefore, treated collectively in the kinetic analysis done below. By varying the initial concentration of D-glucose from 0.02 to 0.1 M, we have confirmed that there have been no differences in the reactive species fractions after the reaction for 195 min at 160 °C: Dglucose (84%), D-fructose (10%), and 5-HMF (5%). Since the hydrothermal reactions of D-glucose are independent of the initial concentration of D-glucose, the reactions represented in Figure 1 are all the first order with respect to the concentrations of D-glucose and D-fructose. Thus, the following rate equations can be established: d½fD  glucoseg ¼  kþ1 ½fD  glucoseg þ k1 ½fD  fructoseg dt

ð12Þ d½fD  fructoseg ¼ kþ1 ½fD  glucoseg  k1 ½fD  fructoseg dt ð13Þ  k2 ½fD  fructoseg

d½5  HMF ¼ k2 ½fD  fructoseg dt

ð14Þ

ð15Þ

The value of k2 is obtained from the slope of the linear fitting of the 5-HMF concentration. To obtain the values of k+1 and k1, we fit the collective concentration of D-glucose to a quadratic function because the D-glucose concentration varies slowly. Substituting a quadratic equation into the left-hand side of eq 12, we have numerically solved the differential equation to determine k+1 and k1. The total time region examined for the present kinetic analysis is divided into 7 segments with equal intervals of 30120 min depending on the reaction temperature. In each time region we have simultaneously determined k+1 and k1 using eq 12. The values of k+1 and k1 thus obtained at 160 °C, for example, are between 1.66  104 and 1.86  104 s1 and between 1.30  103 and 1.33  103 s1, respectively. Thus, the numerical determination of the kinetic parameters is sufficiently selfconsistent. Following the procedure mentioned above, the rate and equilibrium constants at the different temperatures are determined, as summarized in Table 1. The value of k+1 is 1 order of magnitude smaller than that of k1. The equilibrium constant defined as K = k+1/k1 is ∼0.1, and the equilibrium lies in favor of D-glucose. The reason why the backward transformation predominantly occurs can be interpreted from the viewpoint of the above-mentioned empirical rule; D-glucose can be stabilized more strongly than D-fructose due to stronger hydration, and the barrier for the forward process is higher than that of the reverse one. In previous kinetic studies,11,3133 the overall decomposition of D-glucose was treated by a rate equation of d½fD  glucoseg ¼  kapp ½fD  glucoseg dt

ð16Þ

where kapp is the apparent rate constant. According to eqs 1214, when the steady-state approximation is valid for D-fructose, kapp is given by  kapp ¼

 k2 kþ1 k1 þ k2

ð17Þ

As seen from Table 1, k1 is 12 orders of magnitude larger than k2. As a result, the multistep quantity kapp is 12 orders of magnitude smaller than the single-step quantity k+1. When compared with k+1, the kapp values in the literatures are smaller by a factor of 24.11,3133 This is consistent with eq 17, given the difference in the kinetic analysis (in situ vs quenching). 3.2.3. Activation Energies and Enthalpy Change. Here we have examined the temperature effects on the rate and equilibrium constants. First let us see the Arrhenius plots of the rate constants, k+1, k1, and k2 as shown in Figure 6. The pressure of the water system in the relatively low temperature range 120160 °C is not very far away from the ambient. The activation energy (Ea) is thus defined as   ∂ln k Ea ¼  R ð18Þ ∂ð1=TÞ P 14018

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A

ARTICLE

Table 1. Rate (k+1, k1, and k2) and Equilibrium (K) Constants for the Hydrothermal Reaction of D-Glucose at the Temperature Range 120160 °C rate constant/s1 equilibrium constant k+1

120

(3.1 ( 0.4)  106

130 140

6

150

k2

K (= k+1/k1)

2.9  105

(2.0 ( 0.3)  106

0.11 ( 0.01

(9.5 ( 0.3)  10 (3.3 ( 0.4)  105

5

7.7  10 2.7  104

(4.8 ( 0.7)  106 (8.3 ( 0.7)  106

0.12 ( 0.01 0.12 ( 0.02

(6.9 ( 0.5)  105

5.1  104

(2.3 ( 1.0)  105

0.14 ( 0.01

4

3

(3.1 ( 0.2)  105

0.14 ( 0.01

(1.8 ( 0.2)  10

160 a

k1a

temp/°C

1.3  10

The error of each value is smaller than 0.1 in the corresponding order of magnitude.

Figure 6. Arrhenius plots of the rate constants, k+1, k1, and k2 as functions of 1/T. They are linearly fitted to the solid line. The slopes of the lines give the activation energies Ea for k+1, k1, and k2: Ea,+1 = 143 kJ mol1, Ea,1 = 134 kJ mol1, and Ea,2 = 100 kJ mol1, respectively.

Figure 7. van’t Hoff plots of the equilibrium constant K (= k+1/k1) as a function of 1/T. The solid line is the linear fitting of the experimental data. According to the line slope, the enthalpy change (ΔH) of the transformation of D-glucose into D-fructose is estimated as 9 kJ mol1.

According to the slope of linear fitting lines, the values of Ea for k+1, k1, and k2 are determined as Ea,+1 = 143, Ea,1 = 134, and Ea,2 = 100 kJ mol1, respectively. It is found that the activation energy for the forward transformation of D-glucose is slightly larger (9 kJ mol1) than that for the backward transformation. The activation energy for the dehydration of D-fructose is quite small (∼40 kJ mol1) in comparison with those for the forward and backward transformations of D-glucose. k+1 and k1 are dependent strongly on temperature compared with k2. In consequence, a temperature increase makes the bulbs of the reactions for k+1 and k1 more open than that for k2. This leads us to explain why the concentration of the intermediate D-fructose increases with increasing temperature as observed.45 The van’t Hoff plot for the equilibrium constant K (= k+1/k1) between D-glucose and D-fructose is illustrated in Figure 7. The enthalpy change (ΔH) is denoted as follows:

4. CONCLUSIONS In situ 13C NMR spectroscopy has been employed to elucidate the kinetics and pathways of the hydrothermal decomposition of D-glucose at a low concentration of 0.02 M over a relatively low temperature range of 120160 °C. For each species involved in the D-glucose transformation the time evolutions have been explored in detail by combining the time-resolved in situ 13C NMR spectroscopy and the site-selective 13C labeling technique. As a result, the following reaction pathways and rate laws have been established with concrete spectroscopic evidences



∂ln K ΔH ¼  R ∂ð1=TÞ

 ð19Þ P

According to the van’t Hoff plot, the value of ΔH is obtained as 9 kJ mol1. This means that the hydrothermal transformation from D-glucose into D-fructose is endothermic. The intermediate D-fructose goes up as the temperature increases.

kþ1

k2

fD  glucose isomersg hfD  fructose isomersg sf 5  HMF k1

where the curly braces denote a set of all the isomers of D-glucose or D-fructose, 5-HMF indicates 5-hydroxymethylfurfural, and the k’s are the first-order rate constants for the corresponding reaction paths shown by the arrows. No other reaction paths are involved because of the carbon mass balance confirmed spectroscopically. For both monosaccharides there have been detected all of the isomers so far considered, such as the open chains and the furanoses and pyranoses of α- and β-types. These isomers have been treated as pre-equilibria in the kinetic analysis of the k’s. Spectroscopic evidence for this has been successfully given here for the first time. The concentrations of {D-fructose} 14019

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A have been found to reach a plateau after an early rise, so that {D-fructose} can be treated as the stationary state all over the reaction time. When compared with the value of k+1, k1 is 1 order of magnitude larger and k2 is on the same order. According to the Arrhenius activation energy, we have found that the energy barrier for k2 is remarkably smaller than those for k+1 and k1. The D-fructose formation plays a key role in controlling the pathways of the hydrothermal conversion of D-glucose to 5-HMF.

’ AUTHOR INFORMATION Corresponding Author

*Phone/fax: +81-774-38-3071. E-mail: [email protected].

’ ACKNOWLEDGMENT This work is supported by the Grants-in-Aid for Scientific Research (Nos. 21300111 and 23651202) from the Japan Society for the Promotion of Science, and by the Grant-in-Aid for Scientific Research on Priority Areas (No. 20038034), the Grant-in-Aid for Scientific Research on Innovative Areas (No. 20118002), and the Next-Generation Supercomputing Project, Nanoscience Program from the Ministry of Education, Culture, Sports, Science, and Technology. N.M. is also grateful for the grants from the Association of the Progress of New Chemistry and the Suntory Institute of Bioorganic Research. M.N. acknowledges the support for the Water Chemistry Energy Laboratory (AGC) from Asahi Glass Co., Ltd. ’ REFERENCES (1) Onda, A.; Ochi, T.; Yanagisawa, K. Green Chem. 2008, 10, 1033–1037. (2) Jollet, V.; Chambon, F.; Rataboul, F.; Cabiac, A.; Pinel, C.; Guillon, E.; Essayem, N. Green Chem. 2009, 11, 2052–2060. (3) Huber, G. W.; Iborra, H.; Corma, A. Chem. Rev. 2006, 106, 4044–4098. (4) Klemm, D.; Heublein, B.; Fink, H. P.; Bohn, A. Angew. Chem., Int. Ed. 2005, 44, 3358–3393. (5) Sasaki, M.; Fang, Z.; Fukushima, Y.; Adschiri, T.; Arai, K. Ind. Eng. Chem. Res. 2000, 39, 2883–2890. (6) Sasaki, M.; Fang, Z.; Fukushima, Y.; Adschiri, T.; Arai, K. AIChE J. 2004, 50, 192–202. (7) Resende, F. L. P.; Savage, P. E. AIChE J. 2010, 56, 2412–2420. (8) Resende, F. L. P.; Savage, P. E. Energy Fuels 2009, 23, 6213–6221. (9) Yoshida, T.; Oshima, Y.; Matsumura, Y. Biomass Bioenergy 2004, 26, 71–78. (10) Sasaki, M.; Goto, K.; Tajima, K.; Adschiri, T.; Arai, K. Green Chem. 2002, 4, 285–287. (11) Matsumura, Y.; Yanachi, S.; Yoshida, T. Ind. Eng. Chem. Res. 2006, 45, 1875–1879. (12) Williams, P. T.; Onwudili, J. Ind. Eng. Chem. Res. 2005, 44, 8739–8749. (13) Aida, T. M.; Sato, Y.; Watanabe, M.; Tajima, K.; Nonaka, T.; Hattori, H.; Arai, K. J. Supercrit. Fluids 2007, 40, 381–388. (14) Watanabe, M.; Aizawa, Y.; Iida, T.; Levy, C.; Aida, T. M.; Inomata, H. Carbohydr. Res. 2005, 340, 1931–1939. (15) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Ind. Eng. Chem. Res. 1999, 38, 2888–2895. (16) Cortright, R. D.; Davda, R. R.; Dumesic, J. A. Nature 2002, 418, 964–967. (17) Srokol, Z.; Bouch, A. G.; van Estrik, A.; Strik, R. C. J.; Maschmeyer, T; Peter, J. A. Carbohydr. Res. 2004, 339, 1717–1726. (18) Roman-Leshkov, Y.; Barrett, C. J.; Liu, Z. Y.; Dumesic, J. A. Nature 2007, 447, 982–986.

ARTICLE

(19) Jin, F.; Yun, J.; Li, G.; Kishita, A.; Tohji, K.; Enomoto, H. Green Chem. 2008, 10, 612–615. (20) Watanabe, M.; Aizawa, Y.; Iida, T.; Aida, T. M.; Levy, C.; Sue, K.; Inomata, H. Carbohydr. Res. 2005, 340, 1925–1930. (21) Titirici, M. M.; Antonietti, M.; Baccile, N. Green Chem. 2008, 10, 1204–1212. (22) Nata, F. I.; Lee, C. K. Green Chem. 2010, 12, 1454–1459. (23) Yao, C.; Shin, Y.; Wang, L. Q.; Windisch, C. F., Jr.; Samuels, W. D.; Arey, B. W.; Wang, C.; Risen, W. M., Jr.; Exarhos, G. J. J. Phys. Chem. C 2007, 111, 15141–15145. (24) Shin, Y.; Wang, L. Q.; Bae, I. T.; Arey, B. W.; Exarhos, G. J. J. Phys. Chem. C 2008, 112, 14236–14240. (25) Baccile, N.; Laurent, G.; Babonneau, F.; Fayon, F.; Titirici, M. M.; Antonietti, M. J. Phys. Chem. C 2009, 113, 9644–9654. (26) Murugan, V.; Muraliganth, T.; Manthiram, A. J. Phys. Chem. C 2008, 112, 14665–14671. (27) Sun, X.; Zheng, C.; Zhang, F.; Yang, Y.; Wu, G.; Yu, A.; Guan, N. J. Phys. Chem. C 2009, 113, 16002–16008. (28) Chen, J. S.; Cheah, Y. L.; Chen, Y. T.; Jayaprakash, N.; Madhavi, S.; Yang, Y. H.; Lou, X. W. J. Phys. Chem. C 2009, 113, 20504–20508. (29) Lin, Y. S.; Duh, J. G.; Hung, M. H. J. Phys. Chem. C 2010, 114, 13136–13141. (30) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Ind. Eng. Chem. Res. 1997, 36, 1552–1558. (31) Bobleter, O.; Page, G. Monatsh. Chem. 1968, 99, 1560–1567. (32) Amin, S.; Reid, R. C.; Modell, M. Intersociety Conference on Environmental Engineers (ASME): New York, 1975; ASME Pap., No. 75-ENAs-21. (33) Knezevic, D.; van Swaaij, W. P. M.; Kersten, S. R. A. Ind. Eng. Chem. Res. 2009, 48, 4731–4743. (34) Matubayasi, N.; Wakai, C.; Nakahara, M. Phys. Rev. Lett. 1997, 78, 2573–2576. (35) Matubayasi, N.; Wakai, C.; Nakahara, M. J. Chem. Phys. 1997, 107, 9133–9140. (36) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. GAUSSIAN 09, Revision A.02; Gaussian, Inc.: Wallingford CT, 2009. (37) McMurry, , J. E. Organic Chemistry, 7th ed.; Brooks Cole : Pacific Grove, CA, 2007; Chapter 25. (38) Aida, T. M.; Tajima, K.; Watanabe, M.; Saito, Y.; Kuroda, K. J. Supercrit. Fluids 2007, 42, 110–119. (39) Although the chain form of D-glucose cannot be seen in the inset, an extremely weak peak is detected in a more expanded spectrum at the same position as that at 100 °C with a population of ∼0.002% in agreement with the reported value.37 (40) The fraction (%) of the isomer is determined by Ii/∑iIi, where Ii means the integrated intensity of interest. According to the definition, the sum of the fraction over species i is 100%. The two quantities of the Ii/∑iIi value and the error of Ii should not be confused with each other. Although both are expressed in % in the main text, they refer to different quantities. Even when Ii/∑iIi is small (cf. eqs 2 and 6), it is not masked by the error of ∑iIi when the corresponding signal is distinctively observed in the NMR spectrum. (41) Lobry de Bruyn, C. A.; Alberda van Ekenstein, W. Recl. Trav. Chim. Pays-Bas 1895, 14, 203–216. 14020

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021

The Journal of Physical Chemistry A

ARTICLE

(42) Morooka, S.; Wakai, C.; Matubayasi, N.; Nakahara, M. J. Phys. Chem. A 2005, 109, 6610–6619. (43) Morooka, S.; Wakai, C.; Matubayasi, N.; Nakahara, M. J. Phys. Chem. A 2007, 111, 2697–2705. (44) In the long-time region (>∼600 min) at 160 °C, we have observed that formic acid begins to be produced through a minor fragmentation path of the intermediate D-fructose. (45) The stationary concentration of D-fructose normalized is 0.09, 0.10, 0.10, 0.11, and 0.12 at 120, 130, 140, 150, and 160 °C, respectively.

14021

dx.doi.org/10.1021/jp206355e |J. Phys. Chem. A 2011, 115, 14013–14021