A High-Level Quantum Chemical Study - American Chemical Society

Dec 20, 2011 - Industrial level conversion of glucose to useful chemicals, such as furfural, ...... Research Center funded by the Office of Science an...
0 downloads 0 Views 608KB Size
Article pubs.acs.org/EF

Comparison of Sugar Molecule Decomposition through Glucose and Fructose: A High-Level Quantum Chemical Study Rajeev S. Assary*,†,‡ and Larry A. Curtiss*,†,§ †

Materials Science Division, Argonne National Laboratory, Argonne, Illinois 60439, United States Chemical and Biological Engineering, Northwestern University, Evanston, Illinois 60208, United States § Center for Nanoscale Materials, Argonne National Laboratory, Argonne, Illinois 60439, United States ‡

S Supporting Information *

ABSTRACT: Efficient chemical conversion of biomass is essential to produce sustainable energy and industrial chemicals. Industrial level conversion of glucose to useful chemicals, such as furfural, hydroxymethylfurfural, and levulinic acid, is a major step in the biomass conversion but is difficult because of the formation of undesired products and side reactions. To understand the molecular level reaction mechanisms involved in the decomposition of glucose and fructose, we have carried out high-level quantum chemical calculations [Gaussian-4 (G4) theory]. Selective 1,2-dehydration, keto−enol tautomerization, isomerization, retro-aldol condensation, and hydride shifts of glucose and fructose molecules were investigated. Detailed kinetic and thermodynamic analyses indicate that, for acyclic glucose and fructose molecules, the dehydration and isomerization require larger activation barriers compared to the retro-aldol reaction at 298 K in neutral medium. The retro-aldol reaction results in the formation of C2 and C4 species from glucose and C3 species from fructose. The formation of the most stable C3 species, dihydroxyacetone from fructose, is thermodynamically downhill. The 1,3-hydride shift leads to the cleavage of the C−C bond in the acyclic species; however, the enthalpy of activation is significantly higher (50−55 kcal/mol) than that of the retro-aldol reaction (38 kcal/mol) mainly because of the sterically hindered distorted four-membered transition state compared to the hexamembered transition state in the retro-aldol reaction. Both tautomerization and dehydration are catalyzed by a water molecule in aqueous medium; however, water has little effect on the retro-aldol reaction. Isomerization of glucose to fructose and glyceraldehyde to dihydroxyacetone proceeds through hydride shifts that require an activation enthalpy of about 40 kcal/mol at 298 K in water medium. This investigation maps out accurate energetics of the decomposition of glucose and fructose molecules that is needed to help find more efficient catalyts for the conversion of hexose to useful chemicals.

1. INTRODUCTION Conversion of biomass to useful chemicals for alternative fuels and industrial chemicals is important for sustainability and increasing energy demands. Catalytic conversion of monosaccharides to useful industrial chemicals, such as hydroxymethylfurfural (HMF), levulinic acid (LA), γ-valerolactone (GVL), was the subject of various recent experimental studies.1−8 An accurate molecular level understanding of reaction mechanisms and intermediates of chemical transformation of sugar molecules is needed to find more efficient catalysts and prevent undesired reactions. Recent experimental and theoretical studies show that both glucose and fructose undergo decomposition to form HMF or LA by Brønsted acid catalysts in aqueous media, while catalysis promoted by Lewis acids in non-aqueous media and supercritical conditions lead to the formation of decomposition products, such as glycolaldehyde and glyceraldehyde.9−13 Accurate molecular structures, conformational analysis, and interaction energies of carbohydrates and reaction energetics were the subjects of various previous studies.6−15 In a recent study,16 we presented high-level quantum chemical predictions for thermochemical data of intermediates in a likely decomposition mechanism of glucose to LA and assessment of the performance of density functional methods. The computed gas-phase reaction enthalpies indicated that the first two steps involving water elimination from fructose are highly © 2011 American Chemical Society

endothermic (by up to 22 kcal/mol) and rehydration of HMF to LA is highly exothermic (32 kcal/mol).16 We have also reported detailed mechanistic studies for fructose dehydration to HMF in neutral and acidic environments.17 The computed enthalpy of activation for decomposition through dehydration of fructose to HMF is 60 kcal/mol for an uncatalyzed reaction in neutral media and 38 kcal/mol in acidic media; this clearly suggests that fructose dehydration is an acid-catalyzed reaction at experimental conditions. Previous experimental studies reported that the apparent enthalpy of activation for this reaction is in the range of 31−34 kcal/mol in the presence of acids or ionic liquids.17,18 The yield of HMF from glucose through conventional mineral acid dehydration is very small compared to that of fructose because of their competing multiple protonation sites, while only one protonation site leads to the precursor of HMF. In contrast, cyclic fructose is very reactive, and protonation predominantly occurs on the C1-hydroxyl position because of the relatively large proton affinity of the tertiary hydroxyl group.18 In addition to desired dehydration of glucose to HMF, many side reactions have been reported during pyrolysis and acid-catalyzed reactions, and the majority of these reactions Received: October 25, 2011 Revised: December 16, 2011 Published: December 20, 2011 1344

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels

Article

results in the decomposition of hexoses to C2−C4 species3,4 and/or condensation to form polymers.19 These products are predominantly formed from the decomposition of glucose through dehydration, aldol condensation, and retro-aldol reactions. The catalytic conversion of glucose to more reactive fructose in aqueous solution to facilitate the selective dehydration to HMF is a key issue and has been the subject of various recent studies because of the abundance of glucose over fructose.9,31 In this paper, we report on an investigation of detailed energetics of key decomposition reactions of glucose and fructose at a high-level theory, which is an important aid for the designing of efficient catalysts for the conversion of these molecules to useful chemicals and a fundamental understanding of sugar decomposition reactions. Reactions that we have investigated in this paper are schematically shown in Figure 1.

species involved in the reaction mechanisms. This would require a detailed study because of the large number of conformations possible for species, such as glucose and fructose. Instead, we have selected conformers based on previous investigations.13,14,16 To account for the effects of an aqueous environment, calculations were also performed in a water dielectric using the SMD solvation model at the B3LYP/ 6-31G(2df,p) level of theory.25,26 An explicit water molecule is also included for modeling the ring-opening, dehydration, and tautomerization reactions in addition to the SMD model. Free energies (ΔGrxn) and enthalpies of activation (ΔH†) of reactions reported in this paper are at the G4 level of theory (298 K) and include the effects from the aqueous dielectric using the solvation model. An accurate description of free energies of activation for the reactions with significant entropy change (e.g., association and dissociation reactions) in solution is difficult based on gas-phase calculations; therefore, instead of activation free energies, the enthalpies of activation are computed and discussed throughout. The non-electrostatic free energy contribution of solvation energy is excluded for the calculation of enthalpy of activation in solution. The calculations for this investigation were performed using Gaussian 09.27

3. RESULTS AND DISCUSSION 3.1. Glucose to Fructose. The conversion of glucose to fructose through ring-opening and isomerization reactions is shown in Figure 2. The ring-opening reaction of the glucose and isomerization reaction were modeled with and without the presence of an explicit solvent (water) molecule. In the absence of an explicit solvent molecule, both ring-opening and tautomerization reactions proceed through a four-membered transition state rather than a six-membered transition state when an explicit water molecule is included.28,29 The formation of acyclic glucose (2) from cyclic glucose (1) requires an enthalpy of activation of 46.1 kcal/mol, where the hydrogen from the hydroxyl group adjacent to the pyranose oxygen is transferred to the oxygen in the ring. This process is computed to be thermodynamically uphill by 10.3 kcal/mol in an aqueous dielectric. Inclusion of an explicit water molecule in the calculation reduces this activation barrier significantly to 28.1 kcal/mol, indicating that the ringopening reaction is a solvent-assisted process (Scheme 2). The formation of acyclic fructose (4) and 1,2-enediol (5) are thermodynamically favorable from the acyclic glucose. The computed free energy of the reactions are −2.0 and −4.2 kcal/mol for the formation of compounds 4 and 5, respectively. The thermodynamic stability of acyclic fructose over acyclic glucose is predominantly due to the internal hydrogen bonding between the keto group and the neighboring hydroxyl groups compared to the ones in their aldo form. The thermodynamic stability of enol (5) is mainly due to the relatively large solvation energy in aqueous solution compared to acyclic glucose. The PCHS to form acyclic fructose (4) from acyclic glucose (2) requires an enthalpy of activation of 40.9 kcal/mol. This concerted process was modeled without participation of any explicit solvent molecule. In the absence of an explicit water molecule, the formation of enol (5) from acyclic glucose (2) requires a large activation barrier (73.2 kcal/mol) compared to 40.1 kcal/mol when an explicit water molecule is incorporated in the transition state. The tautomerization of enol to acyclic fructose (4) is thermodynamically uphill by +2.2 kcal/mol, and the enthalpy of activation required is 49 kcal/mol when an explicit water molecule is included (75 kcal/mol without the explicit water molecule). Similar to the glucose ring-opening reaction, the ring-opening reaction of the fructose molecule (3) occurs through a hydrogen shift from the tertiary hydroxyl group

Figure 1. Schematic diagram of various reactions considered in this study.

Detailed energetics and reaction barriers were computed for selective dehydration, tautomerization, and isomerization reactions through proton-coupled hydride shift (PCHS), retro-aldol reactions, and 1,3-hydride shift (HS) associated with glucose and fructose decomposition using the highly accurate Gaussian-4 (G4) level of theory.13,14 Details of the computational methods employed in this investigation are presented in the next section. Results and discussions are presented for various decomposition routes for glucose and fructose to C5−C1 species.

2. COMPUTATIONAL METHODS The G4 theory20 was developed with the goal of calculating molecular energies within chemical accuracy. The G4 theory has an average absolute deviation from the experiment of 0.83 kcal/mol from an assessment on the 454 energies in the G3/05 test set composed of enthalpies of formation, ionization energies, electron affinities, proton affinities, and hydrogen-bond energies.21 Also, several recent studies have shown that 1 kcal/mol accuracies can be obtained using G4 methods for the prediction of transition-state barriers and energetics compared to very accurate large basis set coupled cluster calculations.17,22,23 We have employed the G4 level of theory to calculate the thermochemical data and barriers for the reactions explained in the paper. For some selected calculations, we used the less expensive G4MP224 level of theory. The geometries and zero-point energies (scaled by 0.9854) for the G4 methods were calculated at the B3LYP/ 6-31G(2df,p) level of theory. Frequency calculations were performed to verify the nature of all of the stationary points as either minima or transition states and to provide zero-point energy corrections. We have not carried out an accurate conformational analysis of the molecular 1345

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels

Article

Figure 2. Schematic representation of the structure and energetics of ring-opening, isomerization, and tautomerization reactions of glucose and fructose. The computed enthalpies of activation (ΔH†) and free energies of the reactions (ΔGrxn) at the G4 level of theory at 298 K are also given. The ring-opening and tautomerization reactions were catalyzed by a single water molecule, and the solvation effects from the aqueous medium were included using the SMD model. All energies are reported in kcal/mol. The schematic representation of transition states is shown in Figure S1 of the Supporting Information.

Scheme 1. Dehydration Patterns Considered for Cyclic Glycopyranose

to the furanose oxygen. This reaction (uncatalyzed) requires an enthalpy of activation of 44.5 kcal/mol and is thermodynamically uphill by 5 kcal/mol. In the presence of an explicit water molecule, the activation enthalpy is 24.3 kcal/mol, indicating the solvent participation in the ring-opening process of the fructose molecule. Our calculations clearly suggest that the effect of an explicit solvent molecule is significant in both ring-opening/ closing and in keto−enol tautomerization reactions.

Similar barriers (enthalpies of activation) for the formation of acyclic fructose (4) and enediol (5) from acyclic glucose (2) suggest that both products will be formed in aqueous conditions. Acyclic fructose could form the cyclic fructose or undergo further decomposition reactions (explained later). Tautomerization of acyclic glucose to acyclic fructose requires a hydride transfer, and this is often catalyzed by bases, ionic liquids, or Lewis acids, such as Sn-β or Ti-β zeolite.30,31 1346

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels

Article

Scheme 2. (Left) Schematic Representation of the Transition-State Structure for the Ring-Opening Reaction of Glycopyranose (1) Catalyzed by One Water Molecule and (Right) Schematic Representation of 1,2-Dehydration of Glucopyranose Assisted by One Water Molecule

3.2. Decomposition of Cyclic Glucose through Initial Dehydration. Initially, we assessed the energetics of primary dehydration patterns of cyclic glucose, followed by the decomposition of acyclic glucose. Dehydration of cyclic and acyclic glucose is catalyzed by acids, bases, or solvents. It has been reported that initial dehydration of cyclic glucose in acidic conditions requires 30 kcal/mol enthalpy of activation and further dehydration patterns proceed through protonated intermediates.32 In neutral-aqueous media, we have computed the energetics and reaction barriers for dehydration of all of the hydroxyl groups of glucopyranose through solvent-assisted 1,2dehydration (for example, see Scheme 2). The dehydration reactions considered in this study are shown in Scheme 1. Computed enthalpies of activation (ΔH†) and free energies

Recently, experimental studies reported that dehydration of glucose at 600 K using high-temperature steam results in the formation of 1,6-anhydroglucopyranose (B) as the major component, and it is believed that the formation is through intramolecular dehydration.33 The computed free energy of this reaction is −3.6 and −6.6 kcal/mol in the gas phase and in aqueous media, respectively. Because the computed barriers are relatively large, prediction of kinetic feasibility in neutral solution is difficult. However, the computed enthalpy of activation for transition states for species C (60.5 kcal/mol) and G (63.2 kcal/mol) is marginally lower than the rest, which indicates that they are more likely to occur than the rest (provided sufficient activation energy) and both species could rearrange to species B (anhydrosugar). This rearrangement occurs via the formation of a carbon−oxygen bond by releasing a proton. We note that high temperatures are required for these reactions, which will result in the ionization of water to protons and hydroxyl groups. The effect of these ions in the dehydration energetics of glucose requires a detailed investigation. Details of decomposition of levoglucosan (species B) through dehydration were subjected to a detailed computational study.15 Dehydration reactions that lead to the formation of species D are also thermodynamically favorable in both the gas phase and in aqueous solutions. This process requires an activation barrier of 64.4 kcal/mol in neutral conditions, and this intermediate is believed to be the precursor for the formation of HMF.34 This could be one of the reasons for the relatively small yield of HMF during the decomposition of glucose, and the majority of HMF formation is from the fructofuranose, which is in thermodynamic equilibrium with glucopyrannose in aqueous solution. Other dehydration reactions resulting in the formation of species E and F are also found to be thermodynamically downhill (−3.5 kcal/mol) in aqueous solutions and require activation enthalpies of 63.4 and 63.8 kcal/mol, respectively. Overall, because of large activation barriers (∼60 kcal/mol), these dehydration reactions may only occur at high temperatures in aqueous media. 3.3. Decomposition of Acyclic Glucose. Figure 3 presents the chemical transformations of acyclic glucose through 1,3-HS, retro-aldol, and dehydration reactions. Also shown in Figure 3 are the computed free energies and the enthalpies of activation of reactions at 298 K in water dielectric media (ε = 78). The 1,3-HS reaction results in the formation of the enol form of pentose (6′) and formaldehyde, and subsequently, enol tautomerizes to stable aldose (6). The retro-aldol reaction results in the formation of the enol form of glycolaldehyde (20′) and erythrose (10). The selective dehydration of acyclic glucose (2) shown in Figure 3 results in the formation of 2,3-enol (7). Both the 1,3-HS and

Table 1. Computed Free Energies (ΔGrxn), Enthalpies of Reactions (ΔHrxn), and Enthalpies of Activation (ΔH†) for Various Dehydration Reactions of Glucose Shown in Scheme 1a ΔGrxn

ΔHrxn

species

gas

water

gas

water

ΔH†

A B C D E F G

1.2 −3.6 7.8 −0.2 2.5 2.6 −1.2

−3.9 −6.6 4.0 −4.4 −3.5 −3.5 −4.1

+13.2 6.1 +19.9 +12.1 14.7 14.7 7.2

6.8 3.5 14.0 7.9 7.5 6.5 5.3

67.4 60.5 64.4 63.4 63.8 63.2

a

The free energies of the reactions and the enthalpies of activation were computed at the G4 and G4MP2 levels of theory at 298 K, respectively. The solvation energy contributions from the aqueous medium were evaluated using the SMD model and are included in the energetics. The dehydration reactions are catalyzed by an explicit water molecule (see Figure S2 of the Supporting Information). All energies are reported in kcal/mol.

(ΔGrxn) of the dehydration reactions shown in Scheme 1 are tabulated in Table 1. The formation of all species, except B, are modeled through 1,2-dehydration mediated by an explicit water molecule (in water dielectric) and, hence, through a six-membered transition state (see Figure S2 of the Supporting Information). Thermodynamically, the species B, D, and G formed by 1,2dehydration (condensation OH from compound 1 and H from compound 2) are favorable in the gas phase. All reactions, except the formation of species C, are thermodynamically downhill when including the effects of the water dielectric at 298 K. 1347

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels

Article

Figure 3. Schematic representation of structures and energetics of decomposition of acyclic glucose through 1,3-HS, selective dehydration, and retroaldol reactions. Computed enthalpies of activation (ΔH†) and free energies of reactions (ΔGrxn) at the G4 level of theory at 298 K are also given. The dehydration and tautomerization reactions were catalyzed by a single water molecule, and the solvation effects from the aqueous medium were included using the SMD model. All energies are reported in kcal/mol.

thermodynamically uphill (+12.1.0 kcal/mol) at 298 K. The enol form of glycolaldehyde (20′) tautomerizes to thermodynamically stable aldehyde (20) (by 9.3 kcal/mol). This reaction is catalyzed by a single water molecule and requires an activation enthalpy of 30.8 kcal/mol. Overall, the conversion of cyclic glucose to erythrose and glycolaldehyde at 298 K is thermodynamically (2.8 kcal/mol) uphill in aqueous media. In addition to the retro-aldol reaction of glucose, the resulting erythrose (10) molecule could undergo carbon−carbon bond cleavage through a subsequent retro-aldol reaction to produce the enol (20′) and keto form of the glycolaldehyde molecule. This reaction is also thermodynamically uphill (12.1 kcal/mol) and requires an activation enthalpy of 37.9 kcal/mol. Thus, the overall outcome of the glucose decomposition to three glycolaldehydes is thermodynamically uphill by 6.3 kcal/mol at 298 K in aqueous media. Also shown in Figure 3 is a selective dehydration of acyclic glucose (2) to form 2,3-enol (7), which is computed as thermodynamically downhill (−8.3 kcal/mol) and requires an enthalpy of activation of 41.5 kcal/mol in a water-mediated dehydration. It has been reported that HMF can also be formed from 3-deoxyglucosone (8) through a dehydration rather than the dehydration of fructofuranose.38 The 2,3-enol (7) tautomerizes favorably to 3-deoxyglucozone thermodynamically (by 3.6 kcal/mol). This tautomerization is a hydrogen shift, also catalyzed by an explicit water, and requires an enthalpy of activation of 37.0 kcal/mol. Removal of two water molecules from 3-deoxyglucosone results in the formation of HMF; however, the mechanism of dehydration is complex and is limited by the cyclization of 3-deoxyglucosone to its thermodynamically stable (by 9.7 kcal/mol) cyclic form (8′). The enthalpies of activation and reaction free energies of the retro-aldol and dehydration reactions of glucose indicate that the retro-aldol reaction is kinetically preferred over the

retro-aldol reactions result in a similar outcome for glucose and fructose molecules, i.e., cleaving of the carbon−carbon bond.10,35 The computed energetics suggests that both the 1,3-HS and the retro-aldol reactions resulting in the C−C bond cleavage are thermodynamically uphill, while the dehydration reaction is thermodynamically downhill. In the 1,3-HS reaction, the hydrogen attached to the C3 carbon is transferred to the C1 carbon via a four-membered transition state, and this is associated with the bond breaking of the C1−C2 bond of the acyclic glucose molecule. The enthalpy of the activation barrier for this process is 54.2 kcal/mol, and this reaction is thermodynamically uphill by 14.5 kcal/mol. The tautomerization of the enol (6′) to the aldo (6) form is thermodynamically downhill by −6.8 kcal/mol. This tautomerization process requires an enthalpy of activation of 38.4 kcal/mol when catalyzed by a single water molecule in aqueous media. Overall, the C−C bond cleavage through a 1,3-HS reaction is less likely to proceed because of the requirement of a large enthalpy of activation and lower thermodynamic feasibility. The high barrier is mainly due to the formation of a sterically hindered, distorted four-membered transition state. Recently, retro-aldol reactions have been reported for the carbon−carbon bond cleavage of sugar molecules during hydrothermal decomposition and heterogeneous catalysis.10,36,37 The retro-aldol reaction of glucose is modeled here as a hydrogen transfer from the hydroxyl group at C3 to the oxygen atom of the aldehyde group, resulting in the breaking of the C2−C3 bond. This results in the formation of the enol form of glycolaldehyde (20′) and erythrose (10). The computed activation barrier for this reaction is +38.5 kcal/mol and occurs through a hexa-membered transition state without the direct participation of a solvent molecule (see Figure S3 of the Supporting Information). The retro-aldol reaction is highly endothermic (+26.2 kcal/mol) and 1348

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels

Article

Figure 4. Schematic representation of the structure and energetics of decomposition of acyclic fructose through 1,3-HS, tautomerization, and retroaldol condensation reactions. Computed enthalpies of activation (ΔH†) and free energies of reactions (ΔGrxn) at the G4 level of theory at 298 K are also given. All energies are reported in kcal/mol. The aqueous solvation energy contributions were evaluated using the SMD model. A schematic representation of 1,3-HS and PCHS reactions is shown in Figure S4 of the Supporting Information.

tautomerization reactions. This is similar to that found for glucose. The possibility of a further 1,3-HS reaction from 3-keto fructose to form glyceraldehyde and its enol form was also considered. This process requires an enthalpy of activation of 49.2 kcal/mol, and the reaction is thermodynamically uphill (9.9 kcal/mol). Both the enol (14′) and keto (14) forms can isomerize to thermodynamically stable dihydroxyacetone (15). However, kinetically, the initial 1,3-HS reaction is not feasible at temperatures even as high as 500 K because of the large barrier. Therefore, the most likely decomposition patterns of fructose are dehydration and retro-aldol reactions, similar to that of glucose. The enthalpy of activation required for the retro-aldol reaction of acyclic fructose is 34.3 kcal/mol (see Figure S4 of the Supporting Information for the retro-aldol transition-state geometry of acyclic fructose). This results in the formation of glyceraldehyde and its enol form. This reaction is thermodynamically uphill by 11.3 kcal/mol. However, the tautomerization reaction from the enol to the keto form (from 14′ to 14) is downhill by 5.8 kcal/mol. This process is catalyzed by a single water molecule and requires an enthalpy of activation of 36.5 kcal/mol at 298 K. Further rearrangement of compound 14 leads to the formation of thermodynamically stable dihydroxyacetone (15). This process is thermodynamically downhill by 5.5 kcal/mol and requires an enthalpy of activation of 41.6 kcal/mol.39 Overall, the formation of two dihydroxyacetones (15) from acyclic fructose is thermodynamically favorable (ΔGrxn = −3.7 kcal/mol) at 298 K in aqueous media. Because of the small free energy difference between 3-keto fructose and acyclic fructose, it is possible that both forms of these isomers may exist in thermodynamic equilibrium in an aqueous environment. The 3-keto fructose can undergo two types of retro-aldol reactions depending upon the hydroxyl groups from where the hydrogen atom is transferred to the keto oxygen. The retro-aldol reaction 1, shown in Figure 4, results in the formation of the enol form of erythrose and glycolaldehyde. This requires an enthalpy of activation of 35.5 kcal/mol, and

dehydration reaction at 298 K in aqueous solution. It is also noteworthy that the retro-aldol transition state does not require any active participation of a solvent molecule. Meanwhile, inclusion of an explicit water molecule in the reaction (in the transition state) significantly reduces the activation enthalpies for dehydration (by 30 kcal/mol). The dehydration is thermodynamically more favored, and previously, we have shown that dehydration becomes thermodynamically more downhill as the temperature increases for a glucose molecule.16 Therefore, we conclude that, at low temperature and in nonaqueous solvents, the retro-aldol reaction is dominant over the dehydration reaction for a glucose molecule and that the activation enthalpy with respect to acyclic glucose is +38.5 kcal/mol. 3.4. Decomposition of Fructose. Previously, we have reported detailed kinetic and thermodynamic studies of dehydration of fructofuranose to HMF in both neutral and acidic media.16,17 Because of the tertiary hydroxyl group, fructose is more reactive to mineral acids and this results in a likely acid-catalyzed pathway for dehydration through cationic intermediates. Here, we have discussed decomposition of fructose through the retro-aldol and 1,3-HS reactions. Shown in Figure 4 are the reactions that we have considered for the decomposition of acyclic fructose. Also shown in Figure 4 are the computed free energies and the activation enthalpies of reactions at 298 K in the water dielectric. Similar to the glucose decomposition, we have computed the energetics for the transformation of acyclic fructose through 1,3-HS and isomerization reactions (via PCHS). The former results in the formation of the enol form of erythrose (10′) and glycoldehyde, and the latter results in the formation of 3-keto fructose (9). The enthalpies of activation required for these rearrangements are 51.7 and 40.3 kcal/mol, respectively, and they are thermodynamically uphill by 10.7 and 2.4 kcal/mol, respectively. The large activation barrier for the 1,3-HS reaction suggests that this reaction is highly unlikely considering the other possible reactions, such as the retro-aldol, dehydration, or 1349

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels

Article

Figure 5. Schematic representation of the chemical transformations of erythrose and glyceraldehyde to form methylvinylglycolate and methyl lactate (or lactic acid). Computed enthalpies of activation (ΔH†) and free energies of the reaction (ΔGrxn) at the G4 level of theory at 298 K are also given. The dehydration and tautomerization reactions were catalyzed by a single water molecule, and the solvation effects from the aqueous medium were included using the SMD model. All energies are reported in kcal/mol.

dominates over dehydration in non-aqueous solutions in neutral conditions. 3.5. Reactions of Glyceraldehyde and Erythrose. Recently, experimental studies by Holm et al.10 reported the formation of methylvinylglycolate (13) in the presence of the methanol solvent as one of the dominant products of glucose decomposition. The formation of compound 13 is presumably through intermediates 11 and 12 by dehydration and methanol addition, followed by a 1,2-HS reaction from erythrose (10) (Figure 5). Holm et al.10 also reported the formation of lactic acid derivatives as the major decomposition products from glucose and fructose catalyzed by Sn-β zeolite in methanol. The lactic acid derivatives were believed to be formed from the reaction between pyruvaldehyde (16) and methanol. The detailed energetics involved in the formation of methyl lactate (19) from the C3 species are shown in Figure 5. The molecular species 16 is formed from the glyceraldehyde (14) through dehydration and subsequent tautomerization reactions. The 1,2-dehydration reaction of glyceraldehyde to form the enol (16′) is thermodynamically downhill (8.9 kcal/mol) and requires an enthalpy of activation of 38.5 kcal/mol. The tautomerization reaction of the enol (16′) to the aldo form is thermodynamically downhill (8.2 kcal/mol) and requires a relatively small enthalpy of activation (33.1 kcal/mol) compared to the dehydration. The formation of intermediates 18 and 19 by the addition of methanol and the subsequent 1,2-HS reaction from pyruvaldehyde are thermodynamically favorable processes. The possible 1,2-HS reaction proposed to form methyl lactate (19) from intermediate 18 requires an enthalpy of activation of 40.3 kcal/mol, and this reaction is exothermic in nature. The addition of a water molecule to pyruvaldehyde to form lactic acid is thermodynamically a much more downhill process (9.1 kcal/mol) than the addition of methanol.

the reaction is thermodynamically uphill by 13.5 kcal/mol. The enol form of erythrose (10′) tautomerizes to the thermodynamically stable keto form (10) and could undergo a further retro-aldol reaction to result in the formation of glycolaldehyde (20). The retro-aldol reaction 2 results in the formation of formaldehyde and the enol form of the pentose molecule (6′). This requires an enthalpy of activation of 32.0 kcal/mol, and the reaction is thermodynamically uphill by 15.7 kcal/mol. On the basis of the reaction barrier of the two different retro-aldol reactions of 3-keto fructose, we can see that the pentose species is marginally preferred over the formation of erythrose. The enol form of pentose tautomerizes to its keto form, and a further retro-aldol reaction results in the formation of glyceraldehyde and glycolaldehyde (ΔH† = 35 kcal/mol). Overall, the formation of formaldehyde, glyceraldehyde, and glycoldehyde from 3-keto fructose is thermodynamically uphill (11.4 kcal/mol), indicating that the retro-aldol reactions are consistently unfavorable in terms of Gibbs free energy. Experimentally, the enthalpy of activation for the decomposition of fructofuranose to HMF through dehydration is in the range of 31−34 kcal/mol40,41 at 350 K in dilute mineral acids, and the computed value at the G4 level of theory is 38 kcal/molat 498 K using protonated sugar fructose in a solvent model.17 Thermodynamically, this reaction (fructose to HMF) is significantly downhill even at room temperature (−20 kcal/mol). The computed enthalpy of activation for the ring-opening reaction of cyclic fructose (catalyzed by an explicit water molecule) is 24.3 kcal, and the computed enthalpy of activation for the retro-aldol reaction of acyclic fructose is 34.3 kcal/mol at 298 K. The water-assisted catalysis is well-studied elsewhere.42,43 On the basis of the computed enthalpies of activation for the dehydration and retro-aldol reactions of fructose, it can be concluded that, kinetically, the retro-aldol reaction is preferred over dehydration. Because no explicit water molecule is essential for the retro-aldol reaction, this reaction probably 1350

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels



4. CONCLUSION In this paper, we report on a high-level, CCSD(T)-based (G4) quantum chemical investigation of the energetics and enthalpies of activation of various dehydrations, keto−enol tautomerizations, isomerizations, retro-aldol reactions, and HS reactions for glucose and fructose molecules. The following conclusions can be drawn from this investigation: (1) The ring-opening reaction of glucose requires an activation of 28.1 kcal/mol and is thermodynamically uphill because of the stability of the ring structure in aqueous solution. Detailed kinetic and thermodynamic analyses show that the enthalpy of activation for the glucose−fructose isomerization (40.1 kcal/mol) and retro-aldol (38.3 kcal/mol) reactions are very similar in magnitude. The enthalpy of activation for 2,3dehydration reaction of acyclic glucose is 41.5 kcal/mol, and this reaction is thermodynamically downhill. The 1,2-dehydration reaction of glycopyranose requires an enthalpy of activation in the range of 55−70 kcal/mol, even though these reactions are thermodynamically favorable. (2) Ring-opening (24.3 kcal/mol) and retro-aldol (34.3 kcal/mol) reactions of fructose require relatively low activation barriers compared to the corresponding reactions of glucose. The formation of thermodynamically favorable pyruvaldehyde and dihydroxyacetone from acyclic fructose requires enthalpies of activation of 38.5 and 41.6 kcal/mol, respectively. Apart from the 1,2-dehydration reaction of cyclic compounds, isomerization of glucose to fructose and glyceraldehyde to dihydroxyacetone requires the highest activation barrier (41 kcal/mol). The 1,2-dehydration reaction through cyclic intermediates is less likely to occur in neutral aqueous or non-aqueous solutions because of the requirement of large activation enthalpies, and this opens the possibility of retro-aldol reactions. These retro-aldol reactions are not thermodynamically favorable, but they require relatively smaller activation enthalpies compared to the 1,2-dehydration reactions in neutral conditions. Additionally, solvents with higher proton affinity than water or aprotic solvents would suppress the mineral-acid-catalyzed dehydration of sugar molecules and catalyze the retro-aldol reactions. Therefore, catalysts that promote isomerization reactions (for example, Sn-β in methanol) in the absence of mineral acids would also lead to the possibility of retro-aldol products from hexoses. A molecular level understanding of these processes should be useful for catalyst design for biomass conversion of sugar molecules to precursors of alternative fuels and industrial chemicals. For example, catalysts that promote the isomerization reaction (glucose−fructose) can be coupled with mineral acids for a “one pot” conversion of glucose to HMF,44 while in the absence of mineral acids, hexoses yield retro-aldol products (glycoldehyde, glyceraldehyde, and erythrose). The choice of solvents, such as alcohol, is important to stabilize the retro-aldol products to form their corresponding esters or ethers.



ACKNOWLEDGMENTS This work was supported by the U.S. Department of Energy under Contract DE-AC0206CH11357. This material is based on work supported as part of the Institute for Atom-Efficient Chemical Transformations (IACT), an Energy Frontier Research Center funded by the Office of Science and Office of Basic Energy Sciences of the U.S. Department of Energy. We gratefully acknowledge grants of computer time from the Argonne National Laboratory (ANL) Laboratory Computing Resource Center (LCRC) and the ANL Center for Nanoscale Materials. This research used resources of the National Energy Research Scientific Computing Center, which is supported by the Office of Science of the U.S. Department of Energy under Contract DE-AC02-05CH11231



REFERENCES

(1) Chheda, J. N.; Huber, G. W.; Dumesic, J. A. Angew. Chem., Int. Ed. 2007, 46, 7164−7183. (2) Huber, G. W.; Chheda, J. N.; Barrett, C. J.; Dumesic, J. A. Science 2005, 308, 1446−1450. (3) Kunkes, E. L.; Simonetti, D. A.; West, R. M.; Serrano-Ruiz, J. C.; Gartner, C. A.; Dumesic, J. A. Science 2008, 322, 417−421. (4) Huber, G. W.; Iborra, S.; Corma, A. Chem. Rev. 2006, 106, 4044− 4098. (5) Zhao, H.; Holladay, J. E.; Brown, H.; Zhang, Z. C. Science 2007, 316, 1597−1600. (6) Van de Vyver, S.; Geboers, J.; Jacobs, P. A.; Sels, B. F. ChemCatChem 2011, 3, 82−94. (7) Climent, M. J.; Corma, A.; Iborra, S. Green Chem. 2011, 13, 520− 540. (8) Rackemann, D. W.; Doherty, W. O. S. Biofuels, Bioprod. Biorefin. 2011, 5, 198−214. (9) Moliner, M.; Roman-Leshkov, Y.; Davis, M. E. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 6164−6168. (10) Holm, M. S.; Saravanamurugan, S.; Taarning, E. Science 2010, 328, 602−605. (11) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Ind. Eng. Chem. Res. 1999, 38, 2888−2895. (12) Knezevic, D.; van Swaaij, W. P. M.; Kersten, S. R. A. Ind. Eng. Chem. Res. 2009, 48, 4731−4743. (13) Aida, T. M.; Tajima, K.; Watanabe, M.; Saito, Y.; Kuroda, K.; Nonaka, T.; Hattori, H.; Smith, R. L. Jr.; Arai, K. J. Supercrit. Fluids 2007, 42, 110−119. (14) Vasiliu, M.; Guynn, K.; Dixon, D. A. J. Phys. Chem. C 2011, 115, 15686−15702. (15) Assary, R. S.; Curtiss, L. A. ChemCatChem 2011, DOI: 10.1002/ cctc.201100280. (16) Assary, R. S.; Redfern, P. C.; Hammond, J. R.; Greeley, J.; Curtiss, L. A. J. Phys. Chem. B 2010, 114, 9002−9009. (17) Assary, R. S.; Redfern, P. C.; Greeley, J.; Curtiss, L. A. J. Phys. Chem. B 2011, 115, 4341−4349. (18) Qian, X.; Nimlos, M. R.; Davis, M.; Johnson, D. K.; Himmel, M. E. Carbohydr. Res. 2005, 340, 2319−2327. (19) Patil, S. K. R.; Lund, C. R. F. Energy Fuels 2011, 25, 4745−4755. (20) Curtiss, L. A.; Redfern, P. C.; Raghavachari, K. J. Chem. Phys. 2007, 126, 084108. (21) Curtiss, L. A.; Redfern, P. C.; Raghavachari, K. J. Chem. Phys. 2005, 123, 124107−124114. (22) Wheeler, S. E.; Ess, D. H.; Houk, K. N. J. Phys. Chem. A 2008, 112, 1798−1807. (23) Curtiss, L. A.; Redfern, P. C.; Raghavachari, K. Chem. Phys. Lett. 2010, 499, 168−172. (24) Curtiss, L. A.; Redfern, P. C.; Raghavachari, K. J. Chem. Phys. 2007, 127, 124105−124112. (25) Cramer, C. J.; Truhlar, D. G. Acc. Chem. Res. 2009, 42, 493−497. (26) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2009, 113, 4538−4543.

ASSOCIATED CONTENT

S Supporting Information *

Complete citation of ref 27, selected geometries of transition states (Figures S1−S4), and G4 energies. This material is available free of charge via the Internet at http://pubs.acs.org.



Article

AUTHOR INFORMATION

Corresponding Author

*Telephone: 630-252-7020 (R.S.A.); 630-252-7380 (L.A.C.). E-mail: [email protected] (R.S.A.); [email protected] (L.A.C.). 1351

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352

Energy & Fuels

Article

(27) Frisch, M. J.; et al. Gaussian 09; Gaussian, Inc.: Wallingford, CT, 2009 (see the Supporting Information for the complete citation). (28) Cucinotta, C. S.; Ruini, A.; Catellani, A.; Stirling, A. ChemPhysChem 2006, 7, 1229−1234. (29) Lewis, B. E.; Choytun, N.; Schramm, V. L.; Bennet, A. J. J. Am. Chem. Soc. 2006, 128, 5049−5058. (30) Pidko, E. A.; Degirmenci, V.; van Santen, R. A.; Hensen, E. J. M. Angew. Chem., Int. Ed. 2010, 49, 2530−2534. (31) Román-Leshkov, Y.; Moliner, M.; Labinger, J. A.; Davis, M. E. Angew. Chem., Int. Ed. 2010, 49, 8954−8957. (32) Liu, D.; Nimlos, M. R.; Johnson, D. K.; Himmel, M. E.; Qian, X. J. Phys. Chem. A 2010, 114, 12936−12944. (33) Sasaki, M.; Takahashi, K.; Haneda, Y.; Satoh, H.; Sasaki, A.; Narumi, A.; Satoh, T.; Kakuchi, T.; Kaga, H. Carbohydr. Res. 2008, 343, 848−854. (34) Qian, X. H.; Nimlos, M. R.; Davis, M.; Johnson, D. K.; Himmel, M. E. Carbohydr. Res. 2005, 340, 2319−2327. (35) Shanks, B. H. Ind. Eng. Chem. Res. 2010, 49, 10212−10217. (36) Srokol, Z.; Bouche, A.-G.; van Estrik, A.; Strik, R. C. J.; Maschmeyer, T.; Peters, J. A. Carbohydr. Res. 2004, 339, 1717−1726. (37) Wang, K.; Hawley, M. C.; Furney, T. D. Ind. Eng. Chem. Res. 1995, 34, 3766−3770. (38) Perez Locas, C.; Yaylayan, V. A. J. Agric. Food Chem. 2008, 56, 6717−6723. (39) Assary, R. S.; Curtiss, L. A. J. Phys. Chem. A 2011, 115, 8754− 8760. (40) Li, Y.; Lu, X.; Yuan, L.; Liu, X. Biomass Bioenergy 2009, 33, 1182−1187. (41) Moreau, C.; Finiels, A.; Vanoye, L. J. Mol. Catal. A: Chem. 2006, 253, 165−169. (42) Nguyen, M. T.; Matus, M. H.; Jackson, V. E.; Ngan, V. T.; Rustad, J. R.; Dixon, D. A. J. Phys. Chem. A 2008, 112, 10386−10398. (43) Komornicki, A.; Taylor, P.; Dixon, D. A. J. Chem. Phys. 1992, 96, 2920−2925. (44) Nikolla, E.; Roman-Leshkov, Y.; Moliner, M.; Davis, M. E. ACS Catal. 2011, 1, 408−410.

1352

dx.doi.org/10.1021/ef201654s | Energy Fuels 2012, 26, 1344−1352