Comparison of homogeneous and heterogeneous catalysts for glucose-to-fructose isomerization in aqueous media

Article abstract of DOI:10.1002/cssc.201300328

Herein, the first comparison of the mechanisms of glucose-to-fructose isomerization in aqueous media enabled by homogeneous (CrCl₃ and AlCl₃) and heterogeneous catalysts (Sn-beta) by using isotopic-labeling studies is reported. A pronounced kinetic isotope effect (KIE) was observed if the deuterium label was at the C2 position, thus suggesting that a hydrogen shift from the C2 to C1 positions was the rate-limiting step with the three catalysts. ¹³C and ¹HNMR spectroscopic investigations confirmed that an intra-hydride-transfer reaction pathway was the predominant reaction channel for all three catalysts in aqueous media. Furthermore, the deuterium atom in the labeled glucose could be mapped onto hydroxymethylfurfural and formic acid through reactions that followed the isomerization step in the presence of Br?nsted acids. In all three catalysts, the active site appeared to be a bifunctional Lewis-acidic/ Br?nsted-basic site, based on a speciation model and first-principles calculations. For the first time, a mechanistic similarities between the homogeneous and heterogeneous catalysis of aldose-to-ketose isomerization is established and it is suggested that learning from homogeneous catalysis could assist in the development of improved heterogeneous catalysts. Different strokes: Isotopic-labeling experiments reveal mechanistic similarities between homogeneous and heterogeneous catalysts for the aldose-to-ketose isomerization reaction in aqueous media. An intra-hydride transfer is the dominant reaction channel for the isomerization and a Lewis acid-Br?nsted base bifunctional site is the most active site for all three catalysts. Copyright

Full text of DOI:10.1002/cssc.201300328

CHEMSUSCHEM
FULL PAPERS
DOI: 10.1002/cssc.201300328
Comparison of Homogeneous and Heterogeneous
Catalysts for Glucose-to-Fructose Isomerization in
Aqueous Media
[
a]
Vinit Choudhary, Ana B. Pinar, Raul F. Lobo,* Dionisios G. Vlachos,* and Stanley I. Sandler
Herein, the first comparison of the mechanisms of glucose-to-
fructose isomerization in aqueous media enabled by homoge-
neous (CrCl and AlCl ) and heterogeneous catalysts (Sn-beta)
beled glucose could be mapped onto hydroxymethylfurfural
and formic acid through reactions that followed the isomeriza-
tion step in the presence of Brønsted acids. In all three cata-
lysts, the active site appeared to be a bifunctional Lewis-acidic/
Brønsted-basic site, based on a speciation model and first-prin-
ciples calculations. For the first time, a mechanistic similarities
between the homogeneous and heterogeneous catalysis of
aldose-to-ketose isomerization is established and it is suggest-
ed that learning from homogeneous catalysis could assist in
the development of improved heterogeneous catalysts.
3
3
by using isotopic-labeling studies is reported. A pronounced
kinetic isotope effect (KIE) was observed if the deuterium label
was at the C2 position, thus suggesting that a hydrogen shift
from the C2 to C1 positions was the rate-limiting step with the
13
1
three catalysts. C and H NMR spectroscopic investigations
confirmed that an intra-hydride-transfer reaction pathway was
the predominant reaction channel for all three catalysts in
aqueous media. Furthermore, the deuterium atom in the la-
Introduction
Furan derivatives obtained from biomass hold great potential
as substitutes for petroleum-derived chemical intermediates in
relatively high temperatures (>413 K). The aldose-to-ketose
isomerization is used in large industrial applications, which use
[
1]
[5]
the manufacture of polymers and fuels. Typically, cellulosic
and hemicellulosic biomass is first transformed into furan mol-
ecules, such as 5-hydroxymethylfurfural (HMF) and furfural,
which are subsequently processed to form methylfuran, dime-
thylfuran, ethoxymethylfurfural, and furandicarboxylic acid, as
well as other molecules, such as valerolactone, levulinic acid,
enzymatic or base-catalyzed reactions; however, neither of
these processes is compatible with the dehydration reaction
when using a single reactor. Thus, the development of cata-
lysts for this reaction in aqueous media is an important mile-
stone in our ability to utilize biomass.
The conversion of sugar into HMF/furfural has been reported
[
2]
[6]
and long-chain alkanes. The most-common approach in the
synthesis of HMF from biomass involves: 1) the hydrolysis of
cellulosic biomass into simple sugars (primarily glucose), 2) glu-
cose (aldose)-to-fructose (ketose) isomerization, and 3) fructose
dehydration to HMF. The aldose-to-ketose isomerization reac-
tion is a key step in this process, and there are ongoing efforts
in various reaction media, such as ionic liquids, organic sol-
[7]
[3b,8]
vents, and aqueous solutions;
aqueous solution (inexpen-
sive and nontoxic) is the natural choice as solvent for these re-
actions. Several metal salts can catalyze the aldose-to-ketose
[
3b,c]
[6,9]
isomerization in aqueous solution,
in ionic liquids, and in
[7a]
organic solvents. Among the various metal salts, CrCl and
3
[
3]
to develop efficient catalysts for it. The aldose-to-ketose iso-
AlCl have been reported to be the most-active homogeneous
3
[
4]
[3b,c,10]
merization is an equilibrium-limited reaction and is beneficial
for performing steps 2 and 3 in a single reactor, in which the
ketose-dehydration step drives the equilibrium-limited aldose-
to-ketose isomerization step. Thus, the isomerization catalyst
should be compatible with the subsequent dehydration reac-
tion, which is performed by using a Brønsted-acid catalyst at
catalysts for isomerization in aqueous media.
On the
other hand, Sn-beta zeolite is an efficient heterogeneous cata-
[3a,11]
lyst for this isomerization reaction in aqueous media.
Ketose formation from aldose occurs in three steps:
1) aldose ring-opening, 2) aldose-to-ketose isomerization, and
3) ketose ring-closure, as shown in Scheme 1. Formally, step 1
involves a hydrogen transfer from the O1 atom to the O5
atom of the aldose and step 3 involves a hydrogen shift from
the O5 atom to the O2 atom on closure of the ketose ring.
Step 2 involves a hydrogen shift from the O2 atom to the O1
atom and another hydrogen migration from the C2 atom to
the C1 atom. This latter hydrogen transfer is considered to be
the rate-limiting step in the isomerization reaction. Mechanisti-
cally, hydrogen transfers between oxygen atoms are, in fact,
proton transfers, whereas the hydrogen shift from the C2 posi-
tion to the C1 position can either be a proton transfer or an in-
[a] Dr. V. Choudhary, Dr. A. B. Pinar, Prof. R. F. Lobo, Prof. D. G. Vlachos,
Prof. S. I. Sandler
Catalysis Center for Energy Innovation
Department of Chemical and Biomolecular Engineering
University of Delaware
Newark, DE 19716 (USA)
Fax: (+1)302-831-1048
E-mail: lobo@udel.edu
vlachos@udel.edu
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201300328.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
1
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
lacking. Herein, we investigate the reaction pathway
in the isomerization of glucose into fructose by using
CrCl and AlCl3 as catalysts in aqueous media and
3
compare the results to investigations of Sn-beta-zeo-
lite-catalyzed isomerization reactions. The kinetic iso-
tope effect (KIE) was studied for the first two cata-
1
13
lysts and H and C NMR spectroscopy was used to
follow the fate of the isotopic labels for all three cata-
lysts. In addition, the deuterium label in D2 glucose
was tracked through the CrCl -catalyzed isomeriza-
3
tion reaction and the subsequent Brønsted-acid-cata-
lyzed dehydration and rehydration reactions by using
2
H NMR spectroscopy to further substantiate the find-
Scheme 1. Reaction pathway for the formation of fructose and mannose from glucose.
1
13
ings from H and C NMR spectroscopy. Overall, cata-
lytic and mechanistic similarities between the CrCl3-,
tramolecular hydride transfer. Proton transfer through an eno-
late intermediate is considered to be the main reaction path-
way in the base-catalyzed Lobry de Bruyn–Alberda van Eken-
stein (LdB–AvE) transformation, which includes aldose-to-
ketose interconversion and epimerization (e.g., glucose-to-
AlCl - (both homogeneous catalysts), and Sn-beta-catalyzed re-
3
actions (heterogeneous catalyst) are indicated in the aldose-to-
ketose isomerization reaction. Our results support the idea that
it could be possible to apply knowledge from homogeneous
catalysts, which are easier to study and do not invoke synthe-
sis challenges, to the identification and improvement of heter-
ogeneous catalysts.
[
12]
mannose) reactions.
The proton-transfer mechanism can
become competitive with intramolecular hydride transfer in
aldose-to-ketose isomerization (glyceraldehyde-to-dihydroxya-
[
13]
cetone) in dilute alkaline aqueous solution. The role of the
cation in the base-catalyzed aldose-to-ketose isomerization is Results and Discussion
2
+
not well-understood. However, Ni complexes with diamine
Kinetic measurements
[
14]
co-catalysts, as well as various alkaline-metal and rare-earth-
[
15]
metal-ion-based
complexes, catalyze the glucose-to-man-
The time evolution of the concentration of the reactant and
the key products in the isomerization of (unlabeled) glucose
by using CrCl or AlCl as a catalyst in aqueous media is shown
nose epimerization reaction through a different reaction path-
way, which involves a 1,2-carbon shift (the bond to the C3
atom migrates from the C2 atom to the C1 atom), analogous
3
3
in Figure 1. In the case of CrCl , the initial rate of glucose con-
3
[
16]
to the Bilik reaction.
sumption was faster than that with AlCl by a factor of 4.6.
3
Several mechanistic studies of aldose-to-ketose isomerization
reactions on various metal-based catalysts have been reported
and many different reaction pathways have been suggested.
In the enzymatic reaction (xylose isomerase), the function of
the metal center of the enzyme is to stabilize the open-chain
form and to catalyze the aldose-to-ketose isomerization reac-
tion; however, the exact mechanism for the movement of the
hydrogen atom from the C2 position to the C1 position is not
Fructose and mannose were the main and secondary isomeri-
zation products, respectively, with both catalysts. An unidenti-
fied compound, presumably another hexose (consistent with
the retention time and by ruling out other possibilities), was
eluted as a shoulder of the mannose peak in the HPLC analysis.
We quantified this combined peak on the basis of the re-
sponse factor of the dominant mannose and refer to it as man-
nose in the rest of the discussion. Moreover, small amounts
(<2% in the case of CrCl and <1% in the case of AlCl ) of
[
17]
yet understood.
For metal-salt-catalyzed aldose-to-ketose
3
3
isomerization reactions in organic media, the intramolecular
HMF, levulinic acid, and formic acid were observed. Their for-
mation is discussed below. The initial pH values of the solu-
tions of CrCl and AlCl in these experiments were 3.4 and 3.9,
[
18]
hydride-transfer pathway is well-accepted; however, multiple
pathways have been identified in aqueous media. For the iso-
merization of glyceraldehyde (a three-carbon sugar) into dihy-
3
3
respectively, which were fairly high and, thus, the dehydration
of fructose into HMF was slow under these conditions. In addi-
droxyacetone by using ZnCl in aqueous solution, an intramo-
2
2+
lecular hydride transfer, as well as a Zn -catalyzed termolecu-
lar proton transfer (through a combined action with a Brønsted
tion, trace amounts (<0.5% yield) of C sugars (glyceraldehyde
3
and dihydroxyacetone) and lactic acid were detected, possibly
[
18e]
base), has been reported.
In the case of Sn-beta in aqueous
because of retro-aldol fragmentation of the C sugars.
6
media, the framework Sn sites catalyze the intramolecular hy-
dride transfer, whereas a proton-transfer mechanism domi-
Deuterium-isotope-labeling experiments with glucose la-
beled at the C2 position (D2 glucose) revealed a clear KIE be-
cause D2 glucose was consumed at a slower rate with both
CrCl and AlCl (Figure 1). These observations were consistent
[
19]
nates if extra-framework Sn is involved.
Although heterogeneous catalysts are preferred, the identifi-
cation and synthesis of suitable microporous materials with de-
sirable substitution in the framework is often far from trivial.
In contrast, homogeneous catalysts are easier to study, but the
mechanistic insights for heterogeneous catalysts are currently
3
3
with C2!C1 hydrogen transfer as the rate-determining step.
The KIE, k /k , was estimated to be 1.77ꢀ0.11 for CrCl and
H
D
3
1.71ꢀ0.10 for AlCl at 413 K (Table 1). The rate constants were
3
calculated by fitting a first-order rate expression to the glu-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
2
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Table 1. Kinetic isotope effects (KIEs) for labeled glucose (deuterated at
the C2 position) in the glucose-isomerization reactions catalyzed by CrCl
AlCl , and Sn-beta in water.
3
,
3
Entry
Catalyst
KIE (k
H D
/k )
T [K]
1
2
3
CrCl
AlCl
Sn-beta zeolite
3
1.77ꢀ0.11
1.71ꢀ0.10
1.98
413
413
383
3
[
19a]
the high reactivity of Sn-beta (50% conversion within 10 min,
Sn/glucose molar ratio 1:100, 413 K), not enough data points
could be obtained at low conversions to calculate the KIE for
Sn-beta at 413 K by using the procedure described above.
Based on a complete scissoring vibration of the ꢁCꢁH bond,
ꢁ
1
which had a value of 1500 cm , we estimated that the KIE
would increase by about 0.1 units by increasing the tempera-
ture from 383 to 413 K. Furthermore, a complete comparison
of homogeneous and heterogeneous catalysts would require
classic diffusion-limitation studies by changing the particle
size. As expected, the rate of fructose formation also slowed
with D2 glucose compared with unlabeled glucose for both
CrCl (by almost one half) and AlCl (by almost one third), as
3
3
shown in Figure 1b. On the other hand, the rate of formation
of mannose from D2 glucose was similar to the rate from unla-
beled glucose (Figure 1c). This result suggests that either the
hydrogen shift is not the rate-limiting step in the formation of
mannose from glucose or that the dominant reaction pathway
for the formation of mannose from glucose does not involve
C2!C1 hydrogen transfer; one possible alternative pathway
involves a 1,2-carbon shift, as discussed in the introduction.
Herein, we focus on the glucose-to-fructose isomerization reac-
tion, which is of prime interest in the production of HMF from
glucose.
Mechanism of the glucose-to-fructose isomerization
The isomerization of D2 glucose (20 wt%) was performed by
using CrCl , AlCl , and zeolite Sn-beta as catalysts, and the
3
3
sugar fractions (mixture of the unreacted glucose and the fruc-
13
1
tose product) were analyzed by using C and H NMR spec-
troscopy. D2 glucose was used as the reactant, and the deute-
rium label was tracked by using NMR spectroscopy. Because
the catalytic activities of the three catalysts for the isomeriza-
tion reaction were different, the experimental conditions for
the three catalysts, in particular the amount of catalyst and the
reaction time, were adjusted to maximize the yield of fructose
to obtain clear peaks in the NMR analysis. The resonances at
Figure 1. Evolution of the reactant and products in the isomerization of un-
labeled and labeled glucose (D2 glucose) by using CrCl and AlCl as cata-
3 3
lysts in aqueous media. Reaction conditions: glucose (1 wt% in water),
metal/glucose molar ratio 3:100, 413 K.
13
d=74.1 and 71.3 ppm in the C spectrum of unlabeled glu-
cose appeared as low-intensity triplets (see the Supporting In-
formation) in the spectrum of D2 glucose; this result is attrib-
uted to the presence of deuterium at the C2 positions of a-
glucopyranose and b-glucopyranose, which are the two domi-
nant configurations of glucose in solution. Broad-band proton
cose-concentration profiles at low conversion (<15%); contri-
butions to the rate of glucose disappearance from the reverse
reactions were neglected because of the very low concentra-
tions of the isomerization products. These KIE values for the
CrCl - and AlCl -catalyzed reactions are comparable to the pre-
13
decoupling is used to observe C nuclei and a nuclear Over-
hauser enhancement (NOE) effect of up to 200% in the reso-
3
3
viously reported KIE of 1.98 (at 383 K) for the Sn-beta-zeolite-
[
19a]
13
1
catalyzed glucose-isomerization reaction (Table 1).
Owing to
nance intensities is observed for directly bonded Cꢁ H
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
3
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
[
20]
pairs. Evidently, this decoupling is not effective for a carbon
atom bonded to a deuterium atom and amplification of the
resonance is not observed. This result explains the diminished
intensity of the C2 resonances in D2 glucose compared to un-
labeled glucose.
ing the presence of deuterium at the C1 position in the fruc-
tose product.
These observations point to the operation of an intramolec-
ular hydride-transfer mechanism in the glucose-to-fructose iso-
1
merization reaction for all three catalysts. The H NMR spectra
13
Figure 2 compares the C NMR spectra of sugar fractions
obtained from the CrCl -, AlCl -, and Sn-beta-catalyzed reac-
of the sugar fractions (Figure 3) corroborated this finding. The
1
H NMR spectra of glucose before and after the reaction were
3
3
tions, as well as that of a control sample prepared by mixing
similar in the CrCl -, AlCl -, and Sn-beta-catalyzed D2 glucose-
3
3
1
D2 glucose and unlabeled fructose in D O. The spectra re-
isomerization reactions, as shown in Figure 3b–d. The H NMR
2
vealed that D2 glucose remained unchanged post-reaction in
1
3
1
Figure 2. Top : C NMR spectra of (a) the control sample (D2 glucose and
fructose in D O), (b) the sugar fraction obtained from the Sn-beta-catalyzed
D2 glucose-isomerization reaction (20 min, Sn/glucose molar ratio 1:100),
c) the sugar fraction obtained from the CrCl -catalyzed D2 glucose-isomeri-
zation reaction (20 min, Cr/glucose molar ratio 3:100), and (d) the sugar frac-
tion obtained from the AlCl -catalyzed D2 glucose-isomerization reaction
30 min, Al/glucose molar ratio 3:100). Reaction conditions: D2 glucose
20 wt% in water), 413 K. Bottom: Magnification of the spectra within the
Figure 3. Top : H NMR spectra of (a) the control sample (D2 glucose and
2
fructose in D O), (b) the sugar fraction obtained from the Sn-beta-catalyzed
D2 glucose-isomerization reaction (20 min, Sn/glucose molar ratio 1:100),
(c) the sugar fraction obtained from the CrCl -catalyzed D2 glucose-isomeri-
zation reaction (20 min, Cr/glucose molar ratio 3:100), and (d) the sugar frac-
tion obtained from the AlCl -catalyzed D2 glucose-isomerization reaction
(30 min, Al/glucose molar ratio 3:100). Reaction conditions: D2 glucose
(20 wt% in water), 413 K. Bottom: Magnification of the spectra within the
2
(
3
3
3
3
(
(
region d=62–64 ppm.
region d=3.0–3.5 ppm.
all three cases, that is, there was no significant deuterium/hy-
drogen exchange during the reactions. This result is further
spectra of the fructose products from the CrCl -, AlCl -, and Sn-
3
3
beta-catalyzed D2 glucose-isomerization reactions also resem-
bled each other very closely; however, they were different
from that of unlabeled fructose. In particular, the resonance at
d=3.45 ppm was absent from the spectrum of the produced
fructose because the proton at this position had been substi-
tuted by deuterium in these samples. These results clearly
show that the deuterium atom at the C2 position of D2 glu-
cose is transferred onto the C1 position of the fructose product
in the CrCl -, AlCl -, and Sn-beta-catalyzed isomerization reac-
2
discussed in the H NMR analysis. The spectra of fructose ob-
tained from the CrCl -, AlCl -, and Sn-beta-catalyzed reactions
3
3
closely resembled each other, but were different from the
spectrum of the unlabeled-fructose standard. Specifically, the
resonances at d=63.8 and 62.6 ppm, which were assigned to
the C1 positions of the b-fructopyranose and b-fructofuranose
configurations in the unlabeled fructose standard, were ob-
served as very-low-intensity triplets in the fructose product.
This decrease in the peak area of the C1 atom was consistent
with the absence of a NOE effect, as noted earlier, thus imply-
3
3
tions, in agreement with an intramolecular hydride-transfer
13
1
mechanism. The C and H NMR spectra of the Sn-beta-cata-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
4
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
lyzed glucose-isomerization reaction were consistent with an
earlier NMR study, in which framework Sn was the active cata-
parison of the intrinsic strengths of different Lewis acid centers
will require future work, in which one would account for the
actual number of active species in the measured apparent rate
constants.
[
19a]
lytic site.
If metal salts are dissolved in water, they dissociate and gen-
erate various ionic species. In the case of CrCl and AlCl , the
3
3
3+
3+
dominant ionic species are [Cr(H O) ] and [Al(H O) ] , respec-
2
6
2
6
2
H NMR investigations of the CrCl -catalyzed glucose trans-
3
tively; by using a speciation model in conjunction with kinetics
measurements, we recently showed that the hydrolyzed ions,
formation
2
+
2+
[
Cr(H O) OH] and [Al(H O) OH] , were the most-active metal
We observed that the CrCl catalyst produced HMF, levulinic
2
5
2
5
3
[
10c]
species in the isomerization of glucose.
Notably,
acid, and formic acid as the reaction progresses. There were
a few additional unidentified products, although most of the
carbon loss was attributed to the formation of humins and
soluble polymers. HMF is produced by the dehydration of fruc-
tose and its rehydration gives levulinic and formic acids, as
shown in Scheme 4. The dehydration and rehydration reactions
are typically catalyzed by a Brønsted acid. Notably, the hydroly-
sis of Cr (or Al) cations releases protons that can catalyze dehy-
dration and rehydration reactions, according to Equa-
2+
2+
[
Cr(H O) OH] and [Al(H O) OH] closely resemble the partial-
2
5
2
5
ly hydrolyzed Sn site (ꢁSnOH), as shown in Scheme 2, which
[10c]
tion (1).
3þ
2þ
þ
½
CrðH OÞ ꢂ þH O $ ½CrðH OÞ OHꢂ þH O
ð1Þ
2
6
2
2
5
3
Scheme 2. Schematic representation of the most-active sites in aldose-to-
ketose isomerization by using (a) CrCl
lysts in aqueous media.
3 3
, (b) AlCl , and (c) Sn-beta zeolite cata-
Presumably, if the deuterium atom is transferred from D2
glucose onto fructose in the CrCl -catalyzed reaction, it can be
3
subsequently passed on to the dehydration and rehydration
products. We used isotope-labeling experiments to track the
has been proposed as the active catalytic site in Sn-beta-cata-
[
21]
lyzed aldose-to-ketose isomerization reactions.
In all three
fate of deuterium in the CrCl -catalyzed conversion of D2 glu-
3
2
cases, the active metal center (Lewis acid) coordinates to
cose by using H NMR spectroscopy. Figure 4a shows the evo-
oxygen atoms in the first shell and has a covalently bonded ꢁ lution of the conversion of D2 glucose and the corresponding
2
OH (Brønsted base) group. In addition, it has been shown that
H NMR spectra of samples collected after cooling the reaction
to room temperature (Figure 4b). The initial pH value of the
2+
the glucose molecule binds to [Cr(H O) OH] by replacing
2
5
water molecules in the first shell of the metal center with its
OH groups; the metal center has an octahedral structure in the
CrCl reaction solution at room temperature was about 2.9.
3
At t=0, two peaks were observed in the spectra, at d=3.04
and 3.33 ppm, which corresponded to b-glucopyranose and a-
glucopyranose, respectively, in a ratio of 64:36, consistent with
[
10c]
complex.
for the ꢁSnOH/glucose complex in the Sn-beta-catalyzed
aldose-to-ketose isomerization reaction. Overall, CrCl , AlCl ,
A similar octahedral structure has been reported
[
21]
[22]
3
3
earlier studies.
and Sn-beta operate through similar catalytic mechanisms. It is
likely that a bifunctional Lewis-acidic/Brønsted-basic catalytic
site is required for aldose-to-ketose isomerization in aqueous
media, in which the Brønsted base (ꢁOH) facilitates an initial
proton transfer, which activates the subsequent Lewis-acid
New NMR peaks appeared as the reaction progressed, which
were assigned to the various products from the isomerization,
dehydration, and rehydration reactions (see the Supporting In-
formation). The peak at d=3.51 ppm was assigned to the deu-
terium atom at the C1 position of b-fructopyranose, whereas
a shoulder at around d=3.40 ppm was observed on the a-pyr-
anose peak of glucose, which was assigned to the deuterium
atom at the C1 position of b-fructofuranose. A peak at d=
4.71 ppm was attributed to deuterium in water; we thought
that the deuterium atoms in the various species exchanged
(
metal center)-catalyzed C2!C1 hydrogen shift (Scheme 3) in
the isomerization reaction, as shown in recent theoretical in-
vestigations of Sn-beta-catalyzed aldose-to-ketose isomeriza-
[
21]
tion reactions. Our results clearly show remarkable qualita-
tive similarities between these catalysts. A quantitative com-
Scheme 3. Schematic representation of the proposed reaction pathway in the metal-salt-catalyzed aldose-to-ketose isomerization in water. “M” represents the
metal center [i.e., Cr, Al, and Sn (in Sn-beta)]. In the case of Sn-beta, the metal center is coordinated to oxygen atoms in the framework instead of water
molecules.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
5
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
tion of the fructose from D2 glu-
cose is then transferred onto the
C1 position of HMF in the fruc-
tose-dehydration reaction and to
formic acid in the subsequent re-
hydration of HMF. Notably, the
deuterium atom from HMF was
Scheme 4. Schematic representation of the deuterium-atom transfer during the transformation of D2 glucose into
levulinic acid by using CrCl
3
as a catalyst in aqueous media.
transferred onto formic acid and
not onto levulinic acid, which in-
dicated that the rehydration of
HMF involved 1,2-CꢁC bond breaking to form formic and levu-
[23]
linic acids, as also indicated in a recent study. As demonstrat-
ed herein, the mapping of hydrogen atoms from the reactants
to the products or intermediates can assist in the development
of mechanisms for the dehydration and rehydration reactions
in biomass-derived molecules.
Conclusions
We have confirmed that CrCl and AlCl are effective catalysts
3
3
for the glucose (aldose)-to-fructose (ketose) isomerization in
aqueous media (without ionic liquids) and that CrCl is signifi-
3
cantly more active than AlCl . A pronounced kinetic isotope
3
effect (KIE) is observed if deuterated glucose (D2 glucose) is
used for both CrCl and AlCl , analogous to the Sn-beta-cata-
3
3
lyzed isomerization reaction. The KIE is consistent with the
conclusion that hydrogen transfer from the C2 position to the
C1 position is the rate-limiting step.
13
1
C and H NMR studies confirmed that intra-hydride transfer
(
C2 to C1) was the dominant reaction pathway in the CrCl -,
3
2
AlCl -, and Sn-beta-catalyzed isomerization reactions. A H NMR
3
13
1
study verified the findings of the C and H NMR studies and
revealed that the deuterium atom initially located at the C2
position of glucose ended up at the C1 position of fructose
and, upon subsequent Brønsted-acid-catalyzed dehydration
and rehydration, at the C1 positions of HMF and formic acid.
In all three catalysts, the active catalytic site for the aldose-to-
ketose isomerization reaction was a bifunctional Lewis-acidic/
Brønsted-basic site. The Brønsted base (ꢁOH) lowered the
energy barrier for the initial deprotonation step, which activat-
ed the subsequent Lewis-acid-catalyzed intramolecular hydride
shift. These observations support the view that the CrCl3-,
3
Figure 4. Transformation of D2 glucose by using CrCl as a catalyst in aque-
2
ous media: (a) progress of the reaction and (b) H NMR spectra at various
times. The peak labels correspond to the deuterium-labeling scheme in
Scheme 4. Reaction conditions: D2 glucose (20 wt% in water), Cr/glucose
molar ratio 3:100, 413 K.
AlCl -, and Sn-beta-catalyzed aldose-to-ketose isomerization re-
3
actions follow a similar mechanism and that a bifunctional
Lewis-acidic/Brønsted-basic active site is required for the iso-
merization reaction in aqueous media. This mechanistic similar-
ity, which has been established herein for the first time, sug-
gests that the knowledge gained from homogeneous Lewis-
acid catalysts could be valuable in the development of im-
proved heterogeneous Lewis-acid catalysts for isomerization
reactions.
with water or could also come from water released during the
dehydration of fructose. About 3% and 13.7% of the total deu-
terium content (based on the D2 glucose area at t=0) was
found in the water after 20 and 60 min reaction time, respec-
1
3
1
tively. Notably, the C and H NMR spectra were measured
after a reaction time of 20 min, during which time no signifi-
cant deuterium exchange was observed. In addition, two
peaks at d=9.29 and 8.07 ppm were detected. The intensities
of these peaks increased with reaction time. These peaks were
assigned to deuterium atoms at the C1 positions of HMF and
formic acid, respectively (Scheme 4). Deuterium transfer from
D2 glucose in the proposed cascade reaction is shown in
Scheme 4. The deuterium atom transferred onto the C1 posi-
Experimental Section
The Sn-beta zeolite was synthesized according to a procedure re-
[24]
ported by Corma and co-workers; the details of the synthesis are
provided in the Supporting Information. All of the chemicals were
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
6
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
obtained from Sigma Aldrich, except for the deuterium-labeled
glucose (D2 glucose, 98%), which was obtained from Cambridge
Isotope Laboratories, Inc. The metal salts were purchased in their
hydrated forms, that is, CrCl ·6H O and AlCl ·6H O, and were used
C. B. Rasrendra, H. J. Heeres, J. G. de Vries, Chem. Rev. 2013, 113, 1499–
1
597.
2] a) L. Hu, G. Zhao, W. Hao, X. Tang, Y. Sun, L. Lin, S. Liu, RSC Adv. 2012, 2,
1184–11206; b) O. O. James, S. Maity, L. A. Usman, K. O. Ajanaku, O. O.
Ajani, T. O. Siyanbola, S. Sahu, R. Chaubey, Energy Environ. Sci. 2010, 3,
833–1850; c) J. Lewkowski, ARKIVOC 2001, 2, 17–54.
3] a) M. Moliner, Y. Roman-Leshkov, M. E. Davis, Proc. Natl. Acad. Sci. USA
010, 107, 6164–6168; b) V. Choudhary, S. I. Sandler, D. G. Vlachos, ACS
[
1
3
2
3
2
as received. The kinetics measurements were performed in 10 mL
glass vials (Wheaton) in a temperature-controlled oil bath on
a magnetic hotplate stirrer (Chemglass). In a typical experiment, an
aqueous solution of the reactant (2 mL) was added into the vial re-
actors, which were then sealed with crimp caps. The metal (cata-
lyst)-to-reactant molar ratio was fixed at 3:100 for the kinetics
measurements. Multiple vial reactors were used for each reaction,
and vials were removed one at a time from the oil bath at different
times and quenched with water at RT to monitor the progress of
the reaction. The error bars in Figure 1 denote the standard devia-
tion, as obtained from the repetition (one) of the unlabeled-glu-
cose experiments. The lines represent a smoothed interpolation of
the experimental data (symbols).
1
[
2
Catal. 2012, 2, 2022–2028; c) Y. J. Pagꢂn-Torres, T. Wang, J. M. R. Gallo,
B. H. Shanks, J. A. Dumesic, ACS Catal. 2012, 2, 930–934; d) Y. Yang, C.-
W. Hu, M. M. Abu-Omar, ChemSusChem 2012, 5, 405–410.
[
4] V. Choudhary, R. I. Burnett, D. G. Vlachos, S. I. Sandler, J. Phys. Chem. C
2
012, 116, 5116–5120.
[
[
5] T. D. Fenn, D. Ringe, G. A. Petsko, Biochemistry 2004, 43, 6464–6474.
6] a) H. Zhao, J. E. Holladay, H. Brown, Z. C. Zhang, Science 2007, 316,
1597–1600; b) M. E. Zakrzewska, E. Bogel-Lukasik, R. Bogel-Lukasik,
Chem. Rev. 2011, 111, 397–417.
[
7] a) J. B. Binder, R. T. Raines, J. Am. Chem. Soc. 2009, 131, 1979–1985;
b) J. B. Binder, J. J. Blank, A. V. Cefali, R. T. Raines, Chemsuschem 2010, 3,
1268–1272; c) C. Rasrendra, J. Soetedjo, I. Makertihartha, S. Adisasmito,
H. Heeres, Top. Catal. 2012, 55, 543–549.
The liquid samples were analyzed by HPLC on a Waters Alliance
System (e2695) equipped with a refractive index (RI) detector and
a photodiode array (PDA) detector. Two Biorad HPX87C (300ꢁ7.8)
columns connected in series (column temperature: 348 K, mobile
[8] a) Y. Roman-Leshkov, J. N. Chheda, J. A. Dumesic, Science 2006, 312,
933–1937; b) M. J. Climent, A. Corma, S. Iborra, Green Chem. 2011, 13,
1
ꢁ1
520–540; c) E. N. E. Nikolla, Y. Roman-Leshkov, M. Moliner, M. E. Davis,
ACS Catal. 2011, 1, 408–410.
9] E. A. Pidko, V. Degirmenci, R. A. van Santen, E. J. M. Hensen, Angew.
Chem. 2010, 122, 2584–2588; Angew. Chem. Int. Ed. 2010, 49, 2530–
phase: HPLC-grade water, flow rate: 0.50 mLmin ) were used to
separate the sugars. All other products were separated from the re-
action mixture by using a Biorad HPX87H (300ꢁ7.8) column
[
(
column temperature: 338 K, mobile phase: 5 mm aqueous sulfuric
2534.
ꢁ1
acid, flow rate: 0.65 mLmin ). The conversions of the reactants
and the yields of the products were calculated according to Equa-
[
10] a) C. B. Rasrendra, I. G. B. N. Makertihartha, S. Adisasmito, H. J. Heeres,
Top. Catal. 2010, 53, 1241–1247; b) Y. Yang, C. W. Hu, M. M. Abu-Omar,
Bioresour. Technol. 2012, 116, 190–194; c) V. Choudhary, S. H. Mushrif, C.
Ho, A. Anderko, V. Nikolakis, N. S. Marinkovi c´ , A. I. Frenkel, S. I. Sandler,
D. G. Vlachos, J. Am. Chem. Soc. 2013, 135, 3997–4006.
11] a) V. Choudhary, A. B. Pinar, S. I. Sandler, D. G. Vlachos, R. F. Lobo, ACS
Catal. 2011, 1, 1724–1728; b) C. C. Chang, Z. P. Wang, P. Dornath, H. J.
Cho, W. Fan, RSC Adv. 2012, 2, 10475–10477.
12] S. J. Angyal in Glyoscience: Epimerization, Isomerization, and Rearrange-
ment Reactions of Carbohydrates; Topics in Current Chemistry, Vol. 215
(Ed.: A. E. Stꢃtz) 2001, Springer, Berlin/New York, pp 1–14.
13] R. W. Nagorski, J. P. Richard, J. Am. Chem. Soc. 1996, 118, 7432–7433.
14] T. Tanase, F. Shimizu, M. Kuse, S. Yano, M. Hidai, S. Yoshikawa, Inorg.
Chem. 1988, 27, 4085–4094.
15] a) S. J. Angyal, Adv. Carbohydr. Chem. Biochem. 1989, 47, 1–43; b) S. J.
Angyal, Carbohydr. Res. 1997, 300, 279–281.
[16] V. Bilik, Chem. Zvesti 1972, 26, 183.
[17] a) I. A. Rose, E. L. Oconnell, R. P. Mortlock, Biochim. Biophys. Acta Enzy-
mol. 1969, 178, 376–379; b) K. J. Schray, I. A. Rose, Biochemistry 1971,
10, 1058–1062; c) K. N. Allen, A. Lavie, G. K. Farber, A. Glasfeld, G. A.
Petsko, D. Ringe, Biochemistry 1994, 33, 1481–1487; d) A. Y. Kovalevsky,
L. Hanson, S. Z. Fisher, M. Mustyakimov, S. A. Mason, V. T. Forsyth, M. P.
Blakeley, D. A. Keen, T. Wagner, H. L. Carrell, A. K. Katz, J. P. Glusker, P.
Langan, Structure 2010, 18, 688–699.
tions (2) and (3), in which C is the molar concentration of species i.
i
ðC
ꢁCreactantÞ
reactant t¼0
Conversion ½%ꢂ ¼
ꢃ 100 %
ð2Þ
ð3Þ
[
C
reactant t¼0
Ci
Yield of product i ½%ꢂ ¼ C
ꢃ 100 %
[
reactant t¼0
1
13
2
H, C, and H NMR spectra were recorded on a 400 MHz NMR
[
[
1
2
spectrometer (Bruker Biospin AG). H and H chemical shifts are re-
ported relative to residual H O and D O (d=4.7 ppm), respectively.
2
2
1
3
C chemical shifts are reported relative to the C1 carbon atom of
[
a-glucopyranose (d=92 ppm). After the reactions, the sugar frac-
tions (a mixture of glucose and fructose) were collected by using
a Waters Fraction Collector III attached to the HPLC. These fractions
were concentrated on a rotary evaporator and diluted with D O to
2
1
13
measure their H and C NMR spectra. The reaction samples were
filtered through 200 mm syringe filters and were used without fur-
2
ther treatment to record the H NMR spectra.
[
18] a) C. I. F. Watt, Adv. Phys. Org. Chem. 1988, 24, 57–112; b) G. Casiraghi,
G. Casnati, G. Sartori, G. T. Zanafredi, J. Chem. Soc. Perkin Trans. 2 1980,
Acknowledgements
407–411; c) E. W. Warnhoff, P. Reynoldswarnhoff, M. Y. H. Wong, J. Am.
Chem. Soc. 1980, 102, 5956–5957; d) I. Watt, S. N. Whittleton, S. M.
Whitworth, Tetrahedron 1986, 42, 1047–1062; e) R. W. Nagorski, J. P. Ri-
chard, J. Am. Chem. Soc. 2001, 123, 794–802; f) J. B. Binder, A. V. Cefali,
J. J. Blank, R. T. Raines, Energy Environ. Sci. 2010, 3, 765–771; g) V. Degir-
menci, E. A. Pidko, P. C. M. M. Magusin, E. J. M. Hensen, ChemCatChem
We thank Dr. Stavros Caratzoulas for his useful discussions on
the isomerization and epimerization mechanisms. This research
was financially supported as part of the Catalysis Center for
Energy Innovation, an Energy Frontier Research Center funded by
the US Department of Energy, Office of Science, Office of Basic
Energy Sciences (Award No. DE-SC0001004).
2
011, 3, 969–972; h) E. A. Pidko, V. Degirmenci, E. J. M. Hensen, Chem-
CatChem 2012, 4, 1263–1271.
[19] a) Y. Romꢂn-Leshkov, M. Moliner, J. A. Labinger, M. E. Davis, Angew.
Chem. 2010, 122, 9138–9141; Angew. Chem. Int. Ed. 2010, 49, 8954–
8957; b) R. Bermejo-Deval, R. Gounder, M. E. Davis, ACS Catal. 2012, 2,
Keywords: biomass · homogeneous catalysis · isomerization ·
isotopic labeling · zeolites
2705–2713.
[
20] David Neuhaus, M. P. Williamson, The Nuclear Overhauser Effect in Struc-
tural and Conformational Analysis, Wiley-VCH, Weinheim, 1989.
21] a) R. Bermejo-Deval, R. S. Assary, E. Nikolla, M. Moliner, Y. Roman-Lesh-
kov, S.-J. Hwang, A. Palsdottir, D. Silverman, R. F. Lobo, L. A. Curtiss,
M. E. Davis, Proc. Natl. Acad. Sci. USA 2012, 109, 9727–9732; b) V.
[
[1] a) S. Bali, M. A. Tofanelli, R. D. Ernst, E. M. Eyring, Biomass Bioenergy
2012, 42, 224–227; b) R.-J. van Putten, J. C. van der Waal, E. de Jong,
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
7
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Choudhary, S. Caratzoulas, D. G. Vlachos, Carbohydr. Res. 2013, 368, 89–
[24] S. Valencia, A. Corma (UOP LCC), US 5 968 473A, 1999.
9
5.
22] M. U. Roshind, P. Tahtinen, M. Niemitz, R. Sjhohn, Carbohydr. Res. 2008,
43, 101–112.
23] J. Zhang, E. Weitz, ACS Catal. 2012, 2, 1211–1218.
[
[
Received: April 11, 2013
Revised: June 22, 2013
Published online on && &&, 0000
3
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
8
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
V. Choudhary, A. B. Pinar, R. F. Lobo,*
D. G. Vlachos,* S. I. Sandler
&& – &&
Comparison of Homogeneous and
Heterogeneous Catalysts for Glucose-
to-Fructose Isomerization in Aqueous
Media
Different strokes: Isotopic-labeling ex-
periments reveal mechanistic similarities
between homogeneous and heteroge-
neous catalysts for the aldose-to-ketose
isomerization reaction in aqueous
media. An intra-hydride transfer is the
dominant reaction channel for the iso-
merization and a Lewis acid–Brønsted
base bifunctional site is the most active
site for all three catalysts.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
&
9
&
ÞÞ
These are not the final page numbers!