Insights into the interplay of lewis and Br?nsted acid catalysts in glucose and fructose conversion to 5-(hydroxymethyl)furfural and levulinic acid in aqueous media

Article abstract of DOI:10.1021/ja3122763

5-(Hydroxymethyl)furfural (HMF) and levulinic acid production from glucose in a cascade of reactions using a Lewis acid (CrCl₃) catalyst together with a Br?nsted acid (HCl) catalyst in aqueous media is investigated. It is shown that CrCl₃ is an active Lewis acid catalyst in glucose isomerization to fructose, and the combined Lewis and Br?nsted acid catalysts perform the isomerization and dehydration/rehydration reactions. A CrCl₃ speciation model in conjunction with kinetics results indicates that the hydrolyzed Cr(III) complex [Cr(H₂O)₅OH]²⁺ is the most active Cr species in glucose isomerization and probably acts as a Lewis acid-Br?nsted base bifunctional site. Extended X-ray absorption fine structure spectroscopy and Car-Parrinello molecular dynamics simulations indicate a strong interaction between the Cr cation and the glucose molecule whereby some water molecules are displaced from the first coordination sphere of Cr by the glucose to enable ring-opening and isomerization of glucose. Additionally, complex interactions between the two catalysts are revealed: Br?nsted acidity retards aldose-to-ketose isomerization by decreasing the equilibrium concentration of [Cr(H₂O)₅OH]²⁺. In contrast, Lewis acidity increases the overall rate of consumption of fructose and HMF compared to Br?nsted acid catalysis by promoting side reactions. Even in the absence of HCl, hydrolysis of Cr(III) decreases the solution pH, and this intrinsic Br?nsted acidity drives the dehydration and rehydration reactions. Yields of 46% levulinic acid in a single phase and 59% HMF in a biphasic system have been achieved at moderate temperatures by combining CrCl₃ and HCl.

Full text of DOI:10.1021/ja3122763

Article
pubs.acs.org/JACS
Insights into the Interplay of Lewis and Brønsted Acid Catalysts in
Glucose and Fructose Conversion to 5‑(Hydroxymethyl)furfural and
Levulinic Acid in Aqueous Media
Vinit Choudhary,† Samir H. Mushrif,‡ Christopher Ho,† Andrzej Anderko,§ Vladimiros Nikolakis,†
Nebojsa S. Marinkovic,† Anatoly I. Frenkel,⊥ Stanley I. Sandler,*,† and Dionisios G. Vlachos*,†
†Center for Catalytic Science and Technology and Catalysis Center for Energy Innovation, Department of Chemical & Biomolecular
Engineering, University of Delaware, Newark, Delaware 19716, United States
‡School of Chemical and Biomedical Engineering, Nanyang Technological University, 62 Nanyang Drive, Singapore 637459
§OLI Systems Inc., 108 The American Road, Morris Plains, New Jersey 07950, United States
⊥Physics Department, Yeshiva University, 245 Lexington Avenue, New York, New York 10016, United States
S
* Supporting Information
ABSTRACT: 5-(Hydroxymethyl)furfural (HMF) and levulinic
acid production from glucose in a cascade of reactions using a
Lewis acid (CrCl3) catalyst together with a Brønsted acid (HCl)
catalyst in aqueous media is investigated. It is shown that CrCl3 is
an active Lewis acid catalyst in glucose isomerization to fructose,
and the combined Lewis and Brønsted acid catalysts perform the
isomerization and dehydration/rehydration reactions. A CrCl3
speciation model in conjunction with kinetics results indicates
that the hydrolyzed Cr(III) complex [Cr(H2O)5OH]2+ is the most active Cr species in glucose isomerization and probably acts as
a Lewis acid−Brønsted base bifunctional site. Extended X-ray absorption fine structure spectroscopy and Car−Parrinello
molecular dynamics simulations indicate a strong interaction between the Cr cation and the glucose molecule whereby some
water molecules are displaced from the first coordination sphere of Cr by the glucose to enable ring-opening and isomerization of
glucose. Additionally, complex interactions between the two catalysts are revealed: Brønsted acidity retards aldose-to-ketose
isomerization by decreasing the equilibrium concentration of [Cr(H2O)5OH]2+. In contrast, Lewis acidity increases the overall
rate of consumption of fructose and HMF compared to Brønsted acid catalysis by promoting side reactions. Even in the absence
of HCl, hydrolysis of Cr(III) decreases the solution pH, and this intrinsic Brønsted acidity drives the dehydration and
rehydration reactions. Yields of 46% levulinic acid in a single phase and 59% HMF in a biphasic system have been achieved at
moderate temperatures by combining CrCl3 and HCl.
and other polymers containing an aromatic moiety.25 HMF
hydrolysis produces levulinic acid, which can be used to
INTRODUCTION
■
The diminishing availability of petroleum reserves and growing
produce a variety of products, e.g., acrylate polymers and fuel
environmental concerns over greenhouse gas emissions have
additives such as γ-valerolactone, 2-methyl-tetrahydrofuran, and
spurred research into utilizing biomass as a feedstock for the
ethyl levulinate.12
production of chemicals and transportation fuels.1−5 A major
High yields of HMF from fructose have been reported in
challenge is developing economic, efficient, and environ-
mentally benign technologies to transform lignocellulosic
aqueous media,14,26−28 organic solvents,29−33 and ionic
liquids.22,34,35 However, fructose is not an ideal feedstock for
biomass into fuels and chemicals. Consequently, cellulosic
HMF production due to its high cost.36 The cost of glucose is
biomass conversion to platform chemicals, such as 5-
about half that of fructose; as a result, glucose isomerization to
(hydroxymethyl)furfural (HMF) and levulinic acid, has recently
gained significant attention.6−12
fructose followed by its subsequent dehydration to HMF in a
one-pot synthesis has recently gained attention. High yields of
HMF from glucose have been reported in ionic liquids22,37−39
and organic solvents,40−43 particularly using various metal salts
as catalysts for the glucose (aldose)-to-fructose (ketose)
isomerization. However, the efficient downstream separation
HMF is produced from the dehydration of hexose sugars,
such as glucose and fructose, and is envisaged as a potential
platform chemical for the biofuel, biochemical, and biopolymer
industries.3,8,9,13−22 HMF derivatives, such as 2,5-dimethylfuran
and 5-(ethoxymethyl)furfural, have been reported as promising
biofuel components.23,24 2,5-Furandicarboxylic acid, produced
by the oxidation of HMF, can substitute for terephthalic acid in
the production of polyesters such as polyethylene-furanoate
Received: December 16, 2012
Published: February 22, 2013
© 2013 American Chemical Society
3997
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
detector and a photodiode array detector. Sugars were separated using
two Biorad HPX87C (300 × 7.8) columns in series at a column
temperature of 348 K with HPLC-grade water as mobile phase at a
flow rate of 0.50 mL/min. Glucose, mannose, and fructose eluted at
23.3, 27.2, and 31.1 min, respectively. HMF, levulinic acid, and formic
acid were separated from the reaction mixture using a Biorad HPX87H
(300 × 7.8) column, with water (0.005 M sulfuric acid) as the mobile
phase at a flow rate of 0.65 mL/min and a column temperature of 338
K. Formic acid, levulinic acid, and HMF eluted at 13.1, 14.7, and 28.1
min, respectively. During the HPLC analysis, the typical relative
standard error was ∼1% for multiple injections from the same sample
and ∼5−8% for replicate samples. Conversion of the reactant, yield,
and selectivity of the products were calculated as follows:
of HMF from high-boiling organic solvents or ionic liquids, and
the recycling of the reaction media, pose challenges to the
commercialization of these processes. Consequently, water
appears to be a better choice for HMF production. Recently,
the Sn-β zeolite has been demonstrated to be an efficient
catalyst for the aldose-to-ketose isomerization reaction in
aqueous media.44−46 An HMF yield of ∼57% with a glucose
conversion of ∼80% at 453 K using a combination of Sn-β and
HCl as catalysts has been reported.47
Typically, heterogeneous catalysis is preferred over homoge-
neous catalysis because of the ease of catalyst separation and its
reusability. However, the formation of unwanted side products,
such as soluble polymers and humins, which can deposit in the
catalyst pores, requires frequent catalyst regeneration. This
suggests a need to explore homogeneous catalytic processes for
HMF production from glucose in aqueous media.
In the past, various metal salts have been tested, and some of
them, e.g., salts of Cr(III), Al(III), Zn(II), and Sn(IV), have
been found to be active in HMF synthesis from glucose in
aqueous media.48−53 Recently a HMF yield of ∼60% was
reported at 443 K using AlCl3 and HCl as catalysts in aqueous
media.54 However, obtaining a high HMF yield from glucose at
low to moderate reaction temperatures in aqueous media
remains a challenge. Moreover, there is a need to establish a
fundamental understanding of the role of metal salts in these
reactions.
In this work we investigate the Lewis acid (CrCl3)-catalyzed
glucose transformation in aqueous media combined with a
Brønsted acid (HCl) to produce HMF and levulinic acid. The
glucose consumption kinetics in both single-phase and biphasic
systems are reported, and the effect of intrinsic and extrinsic
Brønsted acidity on the product of isomerization is elucidated.
The effect of CrCl3 on Brønsted acid-catalyzed fructose
dehydration and HMF hydrolysis is also investigated. The
CrCl3 speciation in water is explored, and the glucose-to-
fructose (or aldose-to-ketose) isomerization is correlated with
various Cr species for the first time. Further, we report the first
results of an extended X-ray absorption fine structure (EXAFS)
study and Car−Parrinello molecular dynamics (CPMD)
simulations to investigate the interaction of the key chromium
ion, indicated from the kinetics experiments and speciation
model, with the glucose molecule.
(Creactant,t=0 − Creactant
)
% conversion of reactant =
% yield of product i =
× 100
Creactant,t=0
Ci
× 100
Creactant,t=0
Ci
% selectivity of product i =
× 100
(Creactant,t=0 − Creactant
)
Here, Ci is the molar concentration of species i.
The Cr K-edge EXAFS analyses of aqueous CrCl3 solutions with
and without glucose were carried out at beamline X19A of National
Synchrotron Light Source, Brookhaven National Laboratory. Samples
were kept in a Kapton foil pouch, and the fluorescence signal was
detected by a passivated implanted planar silicon detector (Canberra,
USA). The monochromator was calibrated with a Cr foil, and the
EXAFS data were analyzed using the IFEFFIT software package.55,56
Theoretical EXAFS data for analyzing the nearest coordination
environment around the central Cr atom in CrCl3 solutions were
constructed using FEFF6 code.57 The atomic coordinates used in
theoretical modeling were based on the structure of solid Cr2O3. It is
assumed that the CrCl3·6H2O salt, since it is crystallized from a
solution in which there are strong solute−solvent interactions, has
Cr−O nearest-neighbor bonds in the local envinroment around Cr,
thus resembling that in Cr oxide. Hence, the Cr2O3 compound can be
used as the model structure for the local environment of the Cr(III)
cations in solution. In the Cr2O3 structure, chromium is coordinated to
six oxygen atoms in the first shell, in which three oxygen atoms in the
first subshell are slightly closer than the three oxygen atoms from the
second subshell (1.965 vs 2.016 Å). From the fit to the experimental
data of bulk Cr2O3 oxide powders, the parameter S2o was obtained, and
its value (0.71) was used in further fittings of the solution data. It was
also assumed that each Cr(III) ion remains coordinated to six water
molecules with their oxygen end oriented toward the cation. Finally, to
reduce the number of adjustable parameters, it was further assumed
that the spread of Cr−O distances was small, and only one type of
Cr−O bond, weighted by a coordination number, was used for the fit.
Dynamic light scattering (DLS) experiments on freshly prepared
and heat-treated CrCl3 solutions were carried out using a Brookhaven
Instruments ZETAPALS instrument. The freshly prepared CrCl3
solution was filtered using a syringe filter (200 μm), and then a
portion of it was heated at 413 K for 3 h. DLS measurements were
carried out both before heating and after heating of these samples
without any further treatment. The autocorrelation functions were
analyzed using the Multimodal Size Distribution algorithm of the
accompanying software, assuming that the viscosity of the solutions
was equal to that of water.
METHODS
■
Experimental Details. The reactions were conducted using 10 mL
thick-walled glass vials (Sigma-Aldrich) heated in a temperature-
controlled oil bath on a digital stirring hot plate (Fisher Scientific).
Glucose, fructose, mannose, HMF, levulinic acid, formic acid,
CrCl3·6H2O, and CrCl2 were purchased from Sigma-Aldrich and
were used as received. In a typical experiment, 2 mL of an aqueous
solution of reactant (10 wt %) was added to the reactor that was then
sealed. A Cr-to-reactant molar ratio of 3:100, which corresponds to a
CrCl3 concentration of ∼18.6 mM for 10 wt % glucose solution, was
used in the kinetics measurements. The HCl concentration was fixed
at 0.1 M when used with CrCl3. Tetrahydrofuran (THF) was used as
an extracting medium with an organic-to-aqueous volume ratio of 2:1
when biphasic systems were studied. An extra vial filled with heating
oil was placed along with the reaction vials to record temperature
during the reaction. Magnetic stirrers were used for mixing during
reactions. The kinetics experiments were carried out using multiple vial
reactors, which were taken out from the oil bath at specified times and
immediately quenched to room temperature in water.
Speciation Model. CrCl3 speciation in aqueous media was
generated using the mixed-solvent electrolyte model of Wang et
al.,58−61 as implemented in the OLI Systems’ Stream Analyzer software
(OLI Systems, 2012). The details of these calculations are summarized
in the Supporting Information.
Car−Parrinello Molecular Dynamics Simulations. The first-
principles CPMD calculations were performed using the planewave−
pseudopotential implementation of the density functional theory.62
The simulation system consisting of Cr(OH)2+, glucose molecule,
The liquid samples were analyzed with high-performance liquid
chromatography (HPLC) using a Waters Alliance System instrument
(e2695). The chromatograph is equipped with a refractive index
3998
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
Figure 1. (a) Glucose conversion using CrCl3 as a catalyst in an aqueous medium. (b) Arrhenius plot based on initial rate of glucose conversion. The
estimated apparent activation energy (E) is 15.3 kcal/mol. Reaction conditions: initial reactant ∼10 wt %, Cr to glucose molar ratio of 3:100, 413 K.
chloride ions, and water molecules was equilibrated for 0.5 ps at 350 K.
Further technical details of CPMD simulations are included in the
Supporting Information.
hydroxyhexanal, an intermediate produced from HMF, is
primarily responsible for the growth of humins.63,64
An apparent activation energy barrier of 15.3 kcal/mol is
estimated for glucose isomerization from the initial kinetics, as
shown in Figure 1b. This value agrees with our recently
reported activation barrier (15.5 kcal/mol) for the isomer-
ization of xylose, a pentose, using CrCl3 as a catalyst in aqueous
media;65 it can be inferred that the activation barrier for CrCl3-
catalyzed aldose-to-ketose isomerization is nearly unaffected by
the size of the sugar.
Effect of Lewis Acidity on Fructose Dehydration and
HMF Rehydration. The chromium chlorides appear to be
efficient catalysts for glucose isomerization; however, they alone
do not appear to be very selective in the subsequent formation
of HMF and levulinic acid. Therefore, a systematic study was
conducted to understand the effect of CrCl3 on the Brønsted
acid-catalyzed fructose dehydration to HMF and HMF
rehydration to levulinic acid.
RESULTS AND DISCUSSION
■
Glucose Conversion Using a Lewis Acid Catalyst.
Figure 1a shows the time evolution of glucose consumption
using a Lewis acid (CrCl3) as a catalyst in aqueous media.
CrCl3 catalyzes glucose isomerization and forms fructose as the
primary product together with a small amount of mannose.
With time, the fructose concentration passes through a
maximum, and HMF, levulinic acid, and formic acid are
observed. HMF is produced by fructose dehydration and
subsequently hydrolyzes to form levulinic and formic acids, as
shown in Scheme 1. Both of these reactions are typically
Scheme 1. Glucose Isomerization to Fructose and Its
Transformation to HMF and Levulinic Acid
The results from starting with fructose as a reactant using
CrCl3 as a catalyst are shown in Figure 2a. The initial fructose
disappearance rate (0.043 min−1) is faster than that of glucose
(0.036 min−1), as shown in Table 1. It should be noted that
fructose not only isomerizes to glucose and mannose but also
dehydrates to HMF in a parallel path. Moreover, CrCl3 also
catalyzes undesired side reactions involving fructose; this is
further discussed below. The maximum HMF yield is higher
(∼21%) when starting with fructose than with glucose (∼16%).
Note that the glucose and mannose formation rates are very
similar to each other when starting with fructose. This can be
explained with our earlier proposed mechanism for aldose-to-
ketose isomerization that suggests that the formation of glucose
and mannose from fructose follows similar reaction pathways,
and the only difference is which hydrogen atom on the C1-
carbon of fructose is transferred to the C2 position in the
isomerization pathway.46
Fructose dehydration was also carried out using HCl only,
and HCl with CrCl3, and the results from the two kinetics
experiments are shown in Figure 2b and c, respectively. In these
experiments, the pH was about 1. The presence of CrCl3 does
not alter the pH of the solution at these low values. It is evident
(see Table 1) that the initial rate of fructose disappearance is
faster in the presence of CrCl3 with HCl (0.073 min−1)
compared to the only HCl case (0.035 min−1). However, the
maximum yield to HMF and levulinic acid drops from 39% to
28% and 60% to 47%, respectively. This indicates that CrCl3
catalyzed by Brønsted acids. In order to understand why these
products form, the pH of the solution was measured. The pH
of the reaction solution was ∼2.9 at the beginning of the
reaction and ∼1.6 at the end (measured by quenching the
reaction sample to room temperature). It is clear that typical
metal salts used for isomerization as Lewis acids render the
solution fairly acidic. As a result, we hypothesize that the actual
chemistry in these systems happens in a combined Lewis and
Brønsted environment; the implications of this are discussed
below. The source of this (intrinsic) Brønsted acidity when
using CrCl3 as a catalyst in water is investigated by modeling
the CrCl3 speciation in water, as discussed below in detail.
As typical in these reactions, significant carbon loss to
unknown products was noticed. Most of the carbon loss during
HMF and levulinic acid synthesis from hexoses is attributed to
the formation of soluble polymers and insoluble humins, dark-
colored tarry solids. Understanding and controlling these side
reactions is a big challenge. Recently, some efforts have been
made to analyze humins using infrared spectroscopy. It was
suggested that aldol addition or condensation of 2,5-dioxo-6-
3999
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
Figure 2. Fructose transformation to HMF and levulinic acid using (a) CrCl3, (b) HCl, and (c) CrCl3 with HCl as catalysts. Reaction conditions:
initial reactant ∼10 wt %, HCl (0.1 M, when used), Cr (when used)-to-fructose molar ratio of 3:100, and 413 K.
the only HCl case (Figure 3a) to ∼72% in the case of HCl with
CrCl3 (Figure 3b).
Table 1. Reaction Rate Constants Associated with the
Disappearance of Reactants in Various Reactions Carried
Out at 413 K
a
Our data underscore that the intrinsic Brønsted acidity
generated by CrCl3 drives sufficient dehydration and
rehydration reactions without external addition of Brønsted
acid, and this needs to be accounted for in understanding the
chemistry of sugars in water. This is consistent with the remarks
made above in glucose chemistry. The above experiments
indicate that the addition of CrCl3 adversely affects both
fructose conversion to HMF and HMF rehydration to levulinic
and formic acids. These observations partly agree with an
earlier reported observation that Lewis acidity in zeolites
adversely affects xylose dehydration to furfural.66 However, it
should be noticed that not all Lewis acid sites are active for the
aldose-to-ketose isomerization, and in that case, the Lewis
acidity primarily catalyzes the undesired side reactions and
adversely affects the formation of furfural and HMF from xylose
and glucose, respectively.
In contrast, in our measurements, CrCl3 is very active for
aldose-to-ketose isomerization, as discussed above. Thus, using
CrCl3 in the aldose (glucose or xylose) conversion to its
respective furan exhibits a trade-off between accelerating the
rate of aldose-to-ketose isomerization and increasing the rate of
formation of the undesired products. It should also be noticed
that the selectivity to HMF increases in fructose dehydration
with increasing temperature, indicating that the apparent
activation energy associated with the HMF formation from
fructose is higher compared with the side reactions.19,27,67 An
earlier study reported apparent activation barriers of 38.4, 27.5,
reactant
catalyst
CrCl3
k (min−1
)
corresponding concentration profile
glucose
glucose
glucose
glucose
fructose
fructose
fructose
HMF
0.036
0.024
0.014
0.007
0.043
0.035
0.073
0.012
0.016
Figure 1a
Figure 4a
Figure S1b
Figure S1b
Figure 2a
Figure 2b
Figure 2c
Figure 3a
Figure 3b
CrCl3, HCl
AlCl3
AlCl3, HCl
CrCl3
HCl
CrCl3, HCl
HCl
HMF
CrCl3, HCl
a
The rate constants were obtained by fitting the initial concentration
profiles (up to 15 min) of the reactant to a first-order reaction rate
expression.
also catalyzes side reactions involving fructose, and hence
adversely affects the yields of the desired products. In the case
of only CrCl3 (Figure 2a), the rate of fructose disappearance
(0.043 min−1) is faster than the only HCl case (Figure 2b) but
slower compared to the combined CrCl3 and HCl case (Figure
2c).
In order to understand the effect of CrCl3 on HMF
hydrolysis to levulinic and formic acids, a similar set of
experiments was conducted starting with HMF, as shown in
Figure 3. The addition of CrCl3 accelerates the rate of HMF
consumption compared to the HCl only case (see Table 1),
and the maximum levulinic acid yield decreases from 82% in
4000
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
Figure 3. HMF transformation to levulinic and formic acids using (a) HCl and (b) HCl with CrCl3 (Cr-to-HMF molar ratio of 3:100). Reaction
conditions: initial reactant ∼1 wt %, HCl (0.1 M), and 413 K.
Figure 4. Glucose conversion and yield to various products using CrCl3 and HCl in a single aqueous phase (a) and in a biphasic system (b).
Reaction conditions: initial reactant ∼10 wt %, HCl (0.1 M), Cr-to-glucose molar ratio of 3:100, and 413 K. In the biphasic system, THF was used as
the organic solvent with the organic-to-aqueous phase volume ratio of 2:1, and 20 wt % NaCl in the aqueous media.
and 22.9 kcal/mol for HMF formation from fructose, levulinic
acid formation from HMF, and HMF decomposition to the
unknown side products, respectively.68 Thus, it appears that
HMF selectivity can be improved by either lowering the
solution pH (increasing Brønsted acidity) or increasing the
reaction temperature. We provide data in the next section that
support this hypothesis.
Effect of Extrinsic Brønsted Acid on Lewis Acid-
Catalyzed Conversion of Glucose to HMF. Next, we used
HCl together with CrCl3 in the glucose transformation to
accelerate the fructose dehydration to HMF. Figure 4a shows
the time evolution of the reactant and products in a single-
phase aqueous medium. A yield of ∼46% to levulinic acid is
achieved in 360 min at 413 K in a single-phase aqueous system.
This is the highest reported yield so far of levulinic acid from
glucose at moderate reaction temperature in aqueous media.
This shows that the use of a Lewis acid (in this case CrCl3,
which is active for aldose-to-ketose isomerization) in
combination with a Brønsted acid can significantly improve
yields of dehydration or rehydration products at the expense of
side reactions driven by CrCl3.
partitioning between the organic and aqueous phases.69 The
glucose transformation using HCl and CrCl3 as catalysts in the
biphasic system with THF as the extracting medium and 20 wt
% of NaCl in the aqueous phase is shown in Figure 4b. The
glucose consumption rate in the biphasic system is similar to
that in the single-phase reaction (e.g., ∼41% glucose conversion
(biphasic) compared to ∼44% (single phase) in 30 min). The
HMF partition coefficients (the ratio of the HMF concen-
trations in the organic media to the aqueous media) for various
reaction samples were estimated to be about 7 from samples
taken from the organic and aqueous phases. Overall, a high
HMF yield of ∼59% along with a ∼7% yield of levulinic acid
from glucose have been achieved in 180 min at a moderate
reaction temperature of ∼413 K in this biphasic system. When
using a biphasic system, the aqueous phase can be recycled, so
that the catalyst can be reused.
CrCl3 Speciation in Aqueous Media: Insights into the
Combined Lewis and Brønsted Acid-Catalyzed Reac-
tions. To gain further insight into CrCl3-catalyzed glucose-to-
fructose isomerization and, more generally, into aldose-to-
ketose isomerization (e.g., xylose to xylulose), we next
considered the CrCl3 speciation in water. In aqueous media,
CrCl3 dissociates to form ions. These ions are solvated by water
and form complex ions, such as [Cr(H2O)6]3+ (see next section
about the first coordination sphere of the Cr ion as determined
from EXAFS experiments). These ions can be further
To further maximize the yield of HMF from glucose, we used
an extractive-reaction process, where an organic solvent is used
to extract HMF continuously from the aqueous media in a
biphasic system.8,14 It has also been reported earlier that
addition of NaCl in the aqueous phase improves HMF
4001
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
hydrolyzed, releasing H+, resulting in a drop in the solution pH
(for simplicity only a key reaction is shown here).70
concentrations of various species, such as oxides, hydroxides,
and chlorides of the chromium ion, are estimated; the
thermodynamic distribution of these species strongly depends
on the initial CrCl3 concentration and the system temperature.
The complete speciation results are tabulated in the Supporting
Information. It is found that [Cr(H2O)6]3+, [Cr(H2O)5Cl]2+,
and [Cr(H2O)5OH]2+ account for most of the Cr (>99%) as
shown in Figure 6, where these ions are represented as Cr3+,
[Cr(H2O)6]3+ + H2O ↔ [Cr(H2O)5OH]2+ + H3O+
(1)
The generated protons catalyze the fructose dehydration to
HMF, and its subsequent hydrolysis to levulinic and formic
acids, even when only CrCl3 is used as a catalyst. Equation 1
provides, for the first time, the rationalization as to how Lewis
acid catalysts (various salts) used for isomerization of aldoses to
ketoses48,54,65 drive Brønsted acid-catalyzed dehydration of
ketoses to their corresponding furans. Since CrCl3 dissolution
results in the formation of various complex ions, understanding
the speciation of CrCl3 in aqueous media is required. Figure 5
Figure 6. Effect of HCl addition on the species distribution in the
CrCl3 speciation (calculated using the OLI software) in aqueous media
at 413 K.
CrCl2+, and CrOH2+, respectively. It is predicted that the
concentration of CrCl2+ and CrOH2+ will increase with an
increase in the solution temperature, except at very low CrCl3
concentration (<3 mM), and is compensated by a decrease in
the Cr3+ concentration. The dominant complex ions of
chromium in solution are shown in Scheme 2. It should be
Figure 5. Comparison of the measured pH of aqueous solutions
(symbols) of varying CrCl3 concentrations with the pH values
calculated using OLI software at 295 and 413 K.
compares the calculated pH with the measured pH as a
function of CrCl3 concentration at room temperature (295 K).
The pH was measured after stirring the solution for 3 h at room
temperature. The calculations are in good agreement with the
measurements; the deviations in the pH values are slightly
higher at high CrCl3 concentrations. A notable observation is
that, when the CrCl3 solution is heated at 413 K for 1 h and
then cooled to room temperature, the solution pH drops
significantly and then remains unchanged, even after 192 h.
Moreover, these pH values compare very well with the
calculated values at 413 K, particularly at lower CrCl3
concentrations, as shown in Figure 5. These observations
suggest that the decrease in pH of the CrCl3 solution at higher
temperatures is largely an irreversible effect and can be
attributed to the precipitation of Cr(OH)3 as predicted by
the CrCl3 speciation calculations (see Supporting Information).
We were not able to observe any Cr(OH)3 particles with the
naked eye. However, DLS measurements of heat-treated CrCl3
solution revealed the presence of suspended particles. Since it
was not possible to identify such particles in a freshly prepared
sample, we can infer that the particle population observed in
the heat-treated sample could be the Cr(OH)3 predicted by the
model. Importantly, our experiments and calculations predict
that, as the CrCl3 concentration increases, the intrinsic
Brønsted acidity increases (eq 1), contributing to dehydration
(of fructose) and rehydration (of HMF) reactions.
Scheme 2. Schematic Representation of the Key Cr(III)
Cations Generated from the Dissolution of CrCl3 in
Aqueous Media
noticed that the chromium is present in the 3+ oxidation state
in the aqueous solutions, which is a nontoxic state, and is in fact
considered as an essential nutrient.71
The effect of HCl addition on the CrCl3 speciation is shown
in Figure 6. With increasing HCl concentration, the CrCl2+
concentration increases monotonically, and that of CrOH2+
decreases monotonically, whereas that of Cr3+ first rises and
then drops. The change in CrCl3 speciation can be explained on
the basis of the shifts in the thermodynamic equilibrium among
various Cr species with increasing the concentration of H+ and
Cl−, associated with an increase in HCl concentration. For
example, an increase in H+ concentration would shift the
thermodynamic equilibrium toward the left (Le Chatelier’s
principle) in eq 1, and suppress the formation of CrOH2+. In
the kinetic experiments, the rate of glucose consumption drops
significantly in the presence of CrCl3 with the addition of HCl
The CrCl3 speciation was predicted at varying CrCl3
concentrations (0.01−30 mM) at 295 and 413 K. The
4002
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
Figure 7. Correlation of initial rate of glucose consumption vs the CrCl3 concentration and vs the associated concentrations of various Cr ions
predicted from the speciation model using the OLI software. Reaction conditions: initial reactant ∼10 wt % and 413 K.
(∼24% conversion in 10 min; see Figure 4a) compared to the
CrCl3 only case (∼32% conversion in 10 min; see Figure 1a).
That is, Brønsted acidity inhibits the Lewis acid-catalyzed
isomerization of aldoses. This experimental observation
combined with the above-discussed trends in the CrCl3
speciation in the presence of HCl suggests that CrOH2+
could be the most active species for the isomerization, as it is
the only Cr species whose formation is suppressed in the
presence of HCl (in the range 0−0.1 M).
To further correlate the rate of glucose consumption with the
various Cr species, a series of kinetics experiments starting with
glucose and varying CrCl3 concentrations was conducted. The
initial rates (at t = 0) of glucose consumption were obtained by
fitting the glucose concentration profiles at short times with a
first-order rate expression. These rates are plotted in Figure 7 as
a function of the concentrations of various Cr species calculated
using the OLI software at the CrCl3 concentration and
temperature of the reaction. It is apparent that the rate of
glucose consumption scales linearly with the CrOH2+
concentration (Figure 7d), whereas nonlinear trends are
found with the concentration of the other Cr species (Figure
7a−c). The linear scaling of glucose consumption rate with the
concentration of CrOH2+ further corroborates that CrOH2+ is
the most active Cr species for glucose isomerization.
to the trends observed in the CrCl3 speciation. Also, the
addition of HCl to AlCl3 adversely affects the glucose
disappearance rate (see Figure S1b), similar to the CrCl3-
catalyzed glucose transformation, as shown in Table 1. Overall,
these observations clearly indicate that Al(OH)2+ is the most
active species for the AlCl3-catalyzed isomerization, analogous
to Cr(OH)2+ in the case of CrCl3. Further, it is observed that
the rate of glucose disappearance is slower in the case of AlCl3
(0.014 min−1) compared to CrCl3 (0.036 min−1) under similar
reaction conditions and a metal-to-glucose molar ratio of 3:100
(see Table 1). Here, it should also be noticed that the predicted
concentration of Al(OH)2+ is lower than that of Cr(OH)2+
under similar operating conditions. This could also explain the
use of a higher reaction temperature (443 K) in an earlier
reported study of AlCl3 (with HCl)-catalyzed HMF production
from glucose.54
We expect that the −OH on the metal center can assist in
the deprotonation step in the aldose-to-ketose isomerization
reaction, similar to the Sn-BEA zeolite case.72,73 A plausible
reaction mechanism of Lewis acid-catalyzed glucose (aldose)-
to-fructose (ketose) isomerization is included in the Supporting
Information (see Scheme S1). Deprotonation at the O2
position activates the glucose molecule to form a complex
between glucose and the Lewis acid site, followed by hydride
transfer from the C2 to C1 carbon and back proton transfer to
O1 to complete the catalytic cycle. This Lewis acid−Brønsted
base bifunctional behavior is reminiscent of the open SnOH
site in Sn-β zeolite72,74 and could explain the higher activity of
CrOH2+ compared with the other Cr species. The speciation
calculations also show that the CrOH2+ concentration increases
We also investigated AlCl3-catalyzed glucose transformation
to assess whether the observations here can be extended to
other salts (Lewis acids) that are active for the glucose
isomerization. It was found that Al3+ and Al(OH)2+ are the
dominant species in the AlCl3 speciation and the Al(OH)2+
concentration drops upon adding HCl (see Figure S1a), similar
4003
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
Figure 8. EXAFS data showing the effect of glucose addition on the Cr−O distance in water at two different CrCl3 concentrations at room
temperature.
with increasing temperature (Supporting Information). This
suggests that the reaction temperature plays a dual role in the
isomerization reaction: it not only helps in overcoming the
activation energy barrier but also generates more catalytically
active species that in turn accelerate the isomerization reaction.
Also, a recent study showed that tungsten acid dissolves at
higher temperature and acts as a homogeneous catalyst in
ethylene glycol production from cellulose that precipitates upon
cooling.75 In these regards, speciation models, such as the one
employed here, are critical in our ability to quantitatively
describe reaction rates and species concentrations in biomass
processing.
molecule and water molecules was calculated using CPMD
simulations. It has to be noted that the Cr(OH)2+−glucose
configuration for the CPMD simulations was based on our
preliminary calculations with the unhydrolyzed Cr cation. Since
two metal centers are required for the enzyme-catalyzed
glucose-to-fructose reaction,76 we initially placed two unhy-
drolyzed Cr cations coordinated with a glucose molecule
(solvated in water). Upon relaxing the system with CPMD
simulations at 350 K, we observed that only one Cr remained
coordinated to the oxygens (of C1 and C2, as shown in Figure
9a) of the closed-chain glucose molecule. Hence, the
Glucose Coordination with the Chromium Ion. To help
elucidate the interaction of the glucose molecule with the metal
center, two sets of CrCl3 solutions of concentrations 0.05 and
0.1 M were analyzed using EXAFS spectroscopy, at varying
glucose concentrations. When CrCl3 is dissolved in pure water,
the Cr(III) cation is surrounded on average by six water
molecules, as obtained by the EXAFS analysis. Coordination
between the negatively charged oxygen atom of water and the
Cr(III) cation is expected at all the times during reaction. For
homogeneous catalysis to occur, it is also reasonable to expect
that the Cr cation should be in close contact with a glucose
molecule, with the glucose molecule replacing water molecules
in the first coordination shell. Similar to what occurs in water,
oxygen atoms of the glucose hydroxyl groups will coordinate
with the metal cation. Since EXAFS cannot distinguish between
the oxygen atoms of water and glucose, it is not possible to
determine the number of water molecules that have been
replaced by glucose. However, since the effective radius (the
Stokes radius) of glucose, at 3.6 Å, is larger than that of water
(1.35 Å), it is expected that the steric effect from the glucose
molecule disturbs the coordination sphere around the
chromium ion and increases the average Cr(III) cation−
oxygen atom (Cr−O) distance. Such expansion can be, in
principle, detected by the EXAFS analysis, and it is evident in
our results. Figure 8 shows the effect of glucose addition on the
average Cr−O distance obtained from the EXAFS analysis. The
Cr−O distance increases with increasing glucose concentration
(in the studied range of 0−2.6 M glucose). This could be
explained by an increasing number of water molecules being
displaced by glucose, indicating the presence of a glucose
molecule in the first coordination shell of the Cr cation.
In addition to the EXAFS analysis presented here, the
coordination of Cr cation [Cr(OH)2+] with the glucose
Figure 9. (a) Molecular snapshot from the simulation showing the
coordination of the Cr(OH)2+ cation, the glucose molecule, and water
molecules (background waters are not shown for clarity). (b) Radial
pair distribution function, G(R), of the Cr(OH)2+ cation with oxygen
atoms and number integral of G(R). Only the portion of the G(R)
showing the first coordination shell is shown here. The number
integral of G(R) gives the coordination number of Cr with oxygens. It
has to be noted that the radial pair distribution function is calculated
on the basis of a short CPMD trajectory of 1 ps.
configuration with one Cr(OH)2+ group placed near the C1
and C2 hydroxyl groups of the glucose molecule, solvated in
water, was chosen to investigate the coordination of the
hydrolyzed Cr with the glucose molecule and neighboring
waters. Details of CPMD simulations are provided in the
Supporting Information. The radial pair distribution function
and the coordination number between the metal center and the
oxygen atoms in the system are shown in Figure 9b. Overall,
the metal center is coordinated with six oxygen atoms
consistent with the EXAFS datain an octahedral structure:
4004
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
one O from the hydroxyl group, two from the glucose
molecule, and the three from water molecules (cf. Figure 9a).
The Cr−O distance for the hydroxyl group is significantly
shorter than the Cr−O distances of glucose or water since the
interaction of the metal cation is much stronger with a
negatively charged hydroxyl group than with the oxygen atoms
of neutral water or glucose molecules. The average Cr−O(H)
distance is 1.89 0.07 Å, whereas the average Cr−O distances
for Cr and water oxygens and for Cr and glucose oxygens are
speciation, as well as reactions and a plausible reaction
mechanism for the Lewis acid-catalyzed glucose isomerization.
This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
vlachos@udel.edu; sandler@udel.edu
■
Notes
2.06
0.06 and 2.11
0.07 Å, respectively. This is in a
The authors declare no competing financial interest.
qualitative agreement with the EXAFS data reported above.
Our simulation results indicate that the presence of glucose
displaces water from the active Cr ion as an important step in
ring-opening and subsequent isomerization.
ACKNOWLEDGMENTS
■
This research is based on work financially supported as part of
the Catalysis Center for Energy Innovation, an Energy Frontier
Research Center funded by the U.S. Department of Energy,
Office of Science, Office of Basic Energy Sciences under award
no. DE-SC0001004.
CONCLUSIONS
■
We have used kinetics experiments, speciation modeling,
EXAFS analysis, and CPMD simulation to gain insights into
the combined Lewis (CrCl3) and Brønsted (HCl) acid catalysis
of aldoses (glucose) to ketoses (fructose) to furans (HMF) to
levulinic and formic acids. CrCl3 is found to be an efficient
catalyst for glucose isomerization to fructose in aqueous media.
The hydrolysis of the Cr(III) ion releases H+ and decreases the
solution pH. This Lewis acid-derived (intrinsic) Brønsted
acidity is primarily responsible for the fructose dehydration to
HMF and its subsequent hydrolysis to levulinic acid at low
temperatures when no external Brønsted acids are added. It is
also found that the Lewis acid catalyst promotes side reactions
and thus increases the overall rate of disappearance of ketoses
(fructose) and furans (HMF). As a result, while CrCl3 can
isomerize glucose to fructose and drive the Brønsted acid-
catalyzed dehydration/rehydration reactions, it is not selective
in the HMF and levulinic acid production when used by itself.
For the first time, we propose that, among the various Cr
ions generated in aqueous media, CrOH2+ is the most active for
the aldose-to-ketose isomerization and possibly exhibits a Lewis
acid−Brønsted base bifunctional behavior. We also show for
the first time that Brønsted acid catalysts inhibit the Lewis acid-
catalyzed isomerization by decreasing the concentration of the
active Cr species CrOH2+ due to a shift in the chemical
equilibrium of hydrolysis of the metal cation. A strong
interaction between the Cr cation and the glucose molecule
in the first coordination sphere of the metal ion is indicated
from EXAFS analysis and CPMD simulation, whereby some of
the water is displaced in order for ring-opening and
isomerization to proceed.
REFERENCES
■
(1) Schmidt, L. D.; Dauenhauer, P. J. Nature 2007, 447, 914.
(2) Alonso, D. M.; Bond, J. Q.; Dumesic, J. A. Green Chem. 2010, 12,
1493.
(3) Climent, M. J.; Corma, A.; Iborra, S. Green Chem. 2011, 13, 520.
(4) Vlachos, D. G.; Caratzoulas, S. Chem. Eng. Sci. 2010, 65, 18.
(5) Stocker, M. Angew. Chem., Int. Ed. 2008, 47, 9200.
(6) Lewkowski, J. Arkivoc 2001, 2.
(7) James, O. O.; Maity, S.; Usman, L. A.; Ajanaku, K. O.; Ajani, O.
O.; Siyanbola, T. O.; Sahu, S.; Chaubey, R. Energy Environ. Sci. 2010, 3,
1833.
(8) Kuster, B. F. M. Starch-Starke 1990, 42, 314.
(9) Zakrzewska, M. E.; Bogel-Lukasik, E.; Bogel-Lukasik, R. Chem.
Rev. 2011, 111, 397.
(10) Rackemann, D. W.; Doherty, W. O. S. Biofuels, Bioprod. Biorefin.
2011, 5, 198.
(11) Peng, L.; Lin, L.; Li, H. Prog. Chem. 2012, 24, 801.
(12) Serrano-Ruiz, J. C.; West, R. M.; Durnesic, J. A. In Annual
Review of Chemical and Biomolecular Engineering; Annual Reviews: Palo
Alto, CA, Prausnitz, J. M., Doherty, M. F., Segalman, M. A., Eds.;
2010; Vol. 1, p 79.
(13) Vandam, H. E.; Kieboom, A. P. G.; Vanbekkum, H. Starch-
Starke 1986, 38, 95.
(14) Roman-Leshkov, Y.; Chheda, J. N.; Dumesic, J. A. Science 2006,
312, 1933.
(15) Rigal, L.; Gaset, A.; Gorrichon, J. P. Ind. Eng. Chem. Prod. Res.
Dev. 1981, 20, 719.
(16) Chuntanapum, A.; Yong, T. L. K.; Miyake, S.; Matsumura, Y.
Ind. Eng. Chem. Res. 2008, 47, 2956.
(17) Bicker, M.; Hirth, J.; Vogel, H. Green Chem. 2003, 5, 280.
(18) Brown, D. W.; Floyd, A. J.; Kinsman, R. G.; Roshanali, Y. J.
Chem. Technol. Biotechnol. 1982, 32, 920.
(19) Moreau, C.; Durand, R.; Razigade, S.; Duhamet, J.; Faugeras, P.;
Rivalier, P.; Ros, P.; Avignon, G. Appl. Catal., A 1996, 145, 211.
(20) Kruger, J. S.; Nikolakis, V.; Vlachos, D. G. Curr. Opin. Chem.
Eng. 2012, 1, 312.
(21) Choudhary, V.; Burnett, R. I.; Vlachos, D. G.; Sandler, S. I. J.
Phys. Chem. C 2012, 116, 5116.
(22) Zhao, H.; Holladay, J. E.; Brown, H.; Zhang, Z. C. Science 2007,
316, 1597.
(23) Mascal, M.; Nikitin, E. B. Chemsuschem 2009, 2, 423.
(24) Lew, C. M.; Rajabbeigi, N.; Tsapatsis, M. Ind. Eng. Chem. Res.
2012, 51, 5364.
(25) Boisen, A.; Christensen, T. B.; Fu, W.; Gorbanev, Y. Y.; Hansen,
T. S.; Jensen, J. S.; Klitgaard, S. K.; Pedersen, S.; Riisager, A.; Stahlberg,
T.; Woodley, J. M. Chem. Eng. Res. Des. 2009, 87, 1318.
(26) Chheda, J. N.; Roman-Leshkov, Y.; Dumesic, J. A. Green Chem.
2007, 9, 342.
The interplay between the acids indicates that, from a
practical standpoint, optimizing the concentrations of Lewis
and Brønsted acids in the cascade reactions to maximize the
desired products yield is feasible and advisable. For example,
high yields of glucose to levulinic acid in a single aqueous phase
(46%) and to HMF in a biphasic system (59%) were achieved
at a moderate reaction temperature (413 K) by combining
CrCl3 (Lewis acid) with HCl (Brønsted acid). Further
optimization should be possible by varying additional operating
conditions and amounts of catalysts.
ASSOCIATED CONTENT
* Supporting Information
■
S
Comparison between CrCl2- and CrCl3-catalyzed glucose
conversion, detailed information about the CrCl3 speciation
calculation using the OLI software, complete CrCl3 speciation
results, details of the CPMD simulations, results for AlCl3
(27) Asghari, F. S.; Yoshida, H. Ind. Eng. Chem. Res. 2006, 45, 2163.
4005
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006

Journal of the American Chemical Society
Article
(28) Bicker, M.; Kaiser, D.; Ott, L.; Vogel, H. J. Supercrit. Fluids 2005,
36, 118.
(29) Shimizu, K.; Uozumi, R.; Satsuma, A. Catal. Commun. 2009, 10,
(65) Choudhary, V.; Sandler, S. I.; Vlachos, D. G. ACS Catal. 2012, 2,
2022.
(66) Weingarten, R.; Tompsett, G. A.; Conner, W. C., Jr.; Huber, G.
W. J. Catal. 2011, 279, 174.
1849.
(67) Nikbin, N.; Caratzoulas, S.; Vlachos, D. G. Chemcatchem 2012,
4, 504.
(30) Qi, X. H.; Watanabe, M.; Aida, T. M.; Smith, R. L. Ind. Eng.
Chem. Res. 2008, 47, 9234.
(68) Asghari, F. S.; Yoshida, H. Ind. Eng. Chem. Res. 2007, 46, 7703.
(69) Roman-Leshkov, Y.; Dumesic, J. A. Top. Catal. 2009, 52, 297.
(70) Bailar, J. C., Emeleus, H. J., Nyholm, R, Trotman-Dickenson, A.
́
F., Eds. Comprehensive Inorganic Chemistry; Pergamon Press: Oxford,
1973; Vol. 3.
(31) Qi, X. H.; Watanabe, M.; Aida, T. M.; Smith, R. L. Catal.
Commun. 2009, 10, 1771.
(32) Wang, F. F.; Shi, A. W.; Qin, X. X.; Liu, C. L.; Dong, W. S.
Carbohydr. Res. 2011, 346, 982.
(33) Zhang, Z. H.; Liu, B.; Zhao, Z. B. Carbohydr. Polym. 2012, 88,
(71) Pechova, A.; Pavlata, L. Vet. Med. (Prague, Czech Repub.) 2007,
52, 1.
(72) Bermejo-Deval, R.; Assary, R. S.; Nikolla, E.; Moliner, M.;
Roman-Leshkov, Y.; Hwang, S.-J.; Palsdottir, A.; Silverman, D.; Lobo,
R. F.; Curtiss, L. A.; Davis, M. E. Proc. Natl. Acad. Sci. U.S.A. 2012, 109,
9727.
(73) Choudhary, V.; Caratzoulas, S.; Vlachos, D. G. Carbohydr. Res.
2013, 368, 89.
(74) Boronat, M.; Concepcion, P.; Corma, A.; Renz, M.; Valencia, S.
J. Catal. 2005, 234, 111.
891.
(34) Shi, C.; Zhao, Y.; Xin, J.; Wang, J.; Lu, X.; Zhang, X.; Zhang, S.
Chem. Commun. 2012, 48, 4103.
(35) Jadhav, A. H.; Kim, H.; Hwang, I. T. Catal. Commun. 2012, 21,
96.
(36) Torres, A. I.; Tsapatsis, M.; Daoutidis, P. Comput. Chem. Eng.
2012, 42, 130.
(37) Hu, S.; Zhang, Z.; Song, J.; Zhou, Y.; Han, B. Green Chem. 2009,
11, 1746.
(38) Yong, G.; Zhang, Y. G.; Ying, J. Y. Angew. Chem., Int. Ed. 2008,
47, 9345.
(39) Sievers, C.; Musin, I.; Marzialetti, T.; Olarte, M. B. V.; Agrawal,
P. K.; Jones, C. W. Chemsuschem 2009, 2, 665.
(40) Binder, J. B.; Raines, R. T. J. Am. Chem. Soc. 2009, 131, 1979.
(41) Ohara, M.; Takagaki, A.; Nishimura, S.; Ebitani, K. Appl. Catal.,
A 2010, 383, 149.
(75) Tai, Z.; Zhang, J.; Wang, A.; Zheng, M.; Zhang, T. Chem.
Commun. 2012, 48, 7052.
(76) Bennett, B. C.; Yeager, M. Structure 2010, 18, 657.
(42) Rasrendra, C.; Soetedjo, J.; Makertihartha, I.; Adisasmito, S.;
Heeres, H. Top. Catal. 2012, 55, 543.
(43) Dutta, S.; De, S.; Alam, M. I.; Abu-Omar, M. M.; Saha, B. J.
Catal. 2012, 288, 8.
(44) Moliner, M.; Roman-Leshkov, Y.; Davis, M. E. Proc. Natl. Acad.
Sci. U.S.A. 2010, 107, 6164.
(45) Roman-Leshkov, Y.; Moliner, M.; Labinger, J. A.; Davis, M. E.
Angew. Chem., Int. Ed. 2010, 49, 8954.
(46) Choudhary, V.; Pinar, A. B.; Sandler, S. I.; Vlachos, D. G.; Lobo,
R. F. ACS Catal. 2011, 1, 1724.
(47) Nikolla, E. N. E.; Roman-Leshkov, Y.; Moliner, M.; Davis, M. E.
ACS Catal. 2011, 1, 408.
(48) Rasrendra, C. B.; Makertihartha, I. G. B. N.; Adisasmito, S.;
Heeres, H. J. Top. Catal. 2010, 53, 1241.
(49) Efremov, A.; Pervyshina, G.; Kuznetsov, B. Chem. Nat. Compd.
1998, 34, 182.
(50) Peng, L.; Lin, L.; Zhang, J.; Zhuang, J.; Zhang, B.; Gong, Y.
Molecules 2010, 15, 5258.
(51) Tyrlik, S. K.; Szerszen, D.; Kurzak, B.; Bal, K. Starch-Starke
1995, 47, 171.
(52) Tyrlik, S. K.; Szerszen, D.; Olejnik, M.; Danikiewicz, W. J. Mol.
Catal. A: Chem. 1996, 106, 223.
(53) De, S.; Dutta, S.; Saha, B. Green Chem. 2011, 13, 2859.
(54) Pagan-Torres, Y. J.; Wang, T.; Gallo, J. M. R.; Shanks, B. H.;
Dumesic, J. A. ACS Catal. 2012, 2, 930.
(55) Newville, M. J. Synchrotron Radiat. 2001, 8, 322.
(56) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537.
(57) Zabinsky, S. I.; Rehr, J. J.; Ankudinov, A.; Albers, R. C.; Eller, M.
J. Phys. Rev. B 1995, 52, 2995.
(58) Wang, P. M.; Anderko, A.; Young, R. D. Fluid Phase Equilib.
2002, 203, 141.
(59) Wang, P.; Springer, R. D.; Anderko, A.; Young, R. D. Fluid Phase
Equilib. 2004, 222, 11.
(60) Wang, P.; Anderko, A.; Springer, R. D.; Young, R. D. J. Mol. Liq.
2006, 125, 37.
(61) Kosinski, J. J.; Wang, P.; Springer, R. D.; Anderko, A. Fluid
Phase Equilib. 2007, 256, 34.
(62) Laio, A.; Gervasio, F. L. Rep. Prog. Phys. 2008, 71.
(63) Patil, S. K. R.; Heltzel, J.; Lund, C. R. F. Energy Fuels 2012, 26,
5281.
(64) Horvat, J.; Klaic, B.; Metelko, B.; Sunjic, V. Croat. Chem. Acta
1986, 59, 429.
4006
dx.doi.org/10.1021/ja3122763 | J. Am. Chem. Soc. 2013, 135, 3997−4006