Distinctive aldose isomerization characteristics and the coordination chemistry of metal chlorides in 1-butyl-3-methylimidazolium chloride

Article abstract of DOI:10.1021/cs5012684

The catalytic isomerization of aldoses to ketones is an important fundamental step for the transformation of cellulosic biomass to biobased chemicals and liquid fuels. The results of this work reveal for the first time the distinctive coordination chemistry features of four classes of metal chlorides, CrCl₃, VCl₃, FeCl₃, and PtCl₂ in 1-butyl-3-methylimidazolium chloride ([BMIM]Cl), that are well correlated to the drastically different catalytic performances of the metal chlorides in the isomerization of glucose. The relative bond strengths and the number of ligands to which the metal ions are coordinated by oxygen atoms of di fferent sources and by chloride were studied by probing model compounds with in situ far-infrared (FIR) and by reaction studies. The superior performance of CrCl₃ for this reaction is now distinguished from that of other metal chlorides, on the basis of its selective Cr(III) ene-diol coordination chemistry. We also offer new insights into the mechanism involved in the conversion of glucose to 5-hydroxymethylfurfural (5-HMF). In situ FIR is established as a powerful tool in the study of the coordination chemistry of metal complexes in ionic liquids. (Chemical Equation Presented).

Full text of DOI:10.1021/cs5012684

Research Article
pubs.acs.org/acscatalysis
Distinctive Aldose Isomerization Characteristics and the
Coordination Chemistry of Metal Chlorides in 1‑Butyl-3-
methylimidazolium Chloride
†
,‡
†
†,‡
†
†
†
†
Huixiang Li, Wenjuan Xu, Tingyu Huang, Songyan Jia, Zhanwei Xu, Peifang Yan, Xiumei Liu,
,†
and Z. Conrad Zhang*
†State Key Laboratory of Catalysis, Dalian National Laboratory for Clean Energy, Dalian Institute of Chemical Physics, Chinese
Academy of Sciences, 457 Zhongshan Road, Dalian 116023, Liaoning, People’s Republic of China
‡University of Chinese Academy of Sciences, Bejing 10049, People’s Republic of China
*
S Supporting Information
ABSTRACT: The catalytic isomerization of aldoses to ketones is an
important fundamental step for the transformation of cellulosic biomass to
biobased chemicals and liquid fuels. The results of this work reveal for the
first time the distinctive coordination chemistry features of four classes of
metal chlorides, CrCl , VCl , FeCl , and PtCl in 1-butyl-3-methylimida-
3
3
3
2
zolium chloride ([BMIM]Cl), that are well correlated to the drastically
different catalytic performances of the metal chlorides in the isomerization
of glucose. The relative bond strengths and the number of ligands to which
the metal ions are coordinated by oxygen atoms of different sources and by
chloride were studied by probing model compounds with in situ far-
infrared (FIR) and by reaction studies. The superior performance of CrCl3
for this reaction is now distinguished from that of other metal chlorides, on
the basis of its selective Cr(III) ene-diol coordination chemistry. We also
offer new insights into the mechanism involved in the conversion of glucose to 5-hydroxymethylfurfural (5-HMF). In situ FIR is
established as a powerful tool in the study of the coordination chemistry of metal complexes in ionic liquids.
KEYWORDS: in situ far-infrared, coordination chemistry, 5-hydroxymethylfurfural, metal chlorides, ionic liquids
1
8−20
INTRODUCTION
ides.
Importantly, chromium chlorides remain the top-
■
performing catalysts in ionic liquids and in other solvents,
including the aqueous phase.
that ionic liquids that dissolve cellulose provide a highly
desirable medium for the conversion of cellulose in a single step
to HMF in the presence of CrCl as a component of paired
In view of the growing consumption of fossil fuels,
lignocellulosic biomass represents the most abundant hydro-
carbon source in nature to meet the ultimate sustainable
societal need for chemical products and liquid fuels. A growing
economy based on renewable hydrocarbons will also benefit the
environment with carbon-neutral greenhouse gas emissions.
Glucose is the most abundant biomolecule and is the
fundamental building block of cellulose and starch. One of
the most notable advances toward biorefineries in recent years
is the discovery of new catalytic systems that enable the
conversion of glucose to the potential platform chemical 5-
8
,21−24
It was further demonstrated
2
25
metal chlorides. Despite the enormous research progress
made in this important process,
25−32
a key question remains
unanswered regarding the fundamental characteristics of
chromium chloride catalysts responsible for their superior
catalytic performance over many other metal chlorides for the
isomerization of glucose to fructose. In this work, we employed
FIR as an in situ tool, which as been demonstrated to be
uniquely suited to distinguish the fundamental differences in
the coordination chemistry of metal chlorides in [BMIM]Cl
ionic liquid. In situ FIR was used to follow the progress of the
coordination chemistry changes of four classes of metal
chlorides, CrCl , VCl , FeCl , and PtCl , during the reaction
1
HMF. Since the discovery of chromium(II,III) chlorides as the
most effective catalysts in 1-alkyl-3-methylimidazolum
[EMIM]Cl) for the isomerization of glucose to fructose,
(
2
which is readily dehydrated to form 5-HMF in high yield, a
considerably large number of publications have appeared that
report results from a combination of a wide variety of ionic
3
3
3
2
3
,4
5
6−8
involving glucose, model compounds of different oxygen
sources which include cyclohexanone, n-butyl alcohol, glyco-
liquids,
other solvents, other catalysts,
and process
9
conditions. In addition, earlier publications and a limited few
recent ones studied the coordination state catalysts involved in
catalytic reactions.
1
0−17
Hensen et al. recently studied the
Received: August 26, 2014
Revised: October 23, 2014
mechanism of glucose conversion to 5-HMF based on DFT
calculated energy profiles involving several metal chlor-
©
XXXX American Chemical Society
4446
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
laldehyde, and deionized water, and a combination of both
glucose and one of the model compounds in [BMIM]Cl. The
effect of competitive oxygen atom coordination to the metal
ions by the model compounds on glucose conversion was
determined by reaction studies, in which the products were
quantitatively measured by HPLC. The results enabled us to
correlate the coordination chemistry of each metal chloride to
its catalytic performance in glucose conversion.
Table 1. Stretching Vibration Absorption Wavenumbers of
Different Metal Chloride Complexes
metal−Cl stretching vibration abs/cm−
1
Cr−Cl
Pt−Cl
Fe−Cl
V−Cl
−
([VCl ]3−)
(
[CrCl ]3−)
([PtCl ]2−)
([FeCl ] )
6
4
4
6
this work
302
311
313, 147, 165
381
287
3
4−38
lit.
315, 199
378, 136
355, 305
RESULTS AND DISCUSSION
Extending our original discovery of the strikingly effective
catalytic conversion of glucose to 5-HMF in [EMIM]Cl, we
also measured the glucose conversions and the HMF yields of
different metal halide/glucose/[BMIM]Cl systems, as the basis
of this work. Figure 1 shows the representative results of
■
on the basis of the results of in situ measurements. In this work,
the far-infrared instrument was specifically modified for an in
situ study of the catalysts under a protected inert atmosphere.
First, we measured the FIR absorption data of pure metal
chlorides in [BMIM]Cl. We then assigned the characteristic
absorption peaks of metal chlorides in the far-infrared region
2
(
Figure S1, Supporting Information) in reference to previous
34−38
studies in the literature.
Significant differences in FIR
absorption bands were observed for the metal chlorides in the
ionic liquids in comparison to those reported in other solvent
systems (Table 1). This difference is expected, as the solvent
and the cations in the system would cause a shift in the peak
absorption wavenumber by the anionic metal chloride
3
4,39
complexes.
2
We did not consider absorption peaks under
00 cm− because they are vulnerable to the interference of air
in this region.
1
4
0
Because CrCl is a highly efficient catalyst for the conversion
3
of glucose to 5-HMF, we first employed in situ far-infrared
spectroscopy to follow the transition from solid CrCl (Figure
3
S2, Supporting Information) to that dissolved in [BMIM]Cl
Figure 1. Catalytic characteristics of metal chlorides for glucose
conversion to 5-HMF in [BMIM]Cl (reaction at 96 °C).
(
Figure S1, Supporting Information). The dissolved CrCl was
3
found to form new Cr(III) complexes in the CrCl /[BMIM]-
3
Cl/glucose reaction system under typical reaction conditions, as
shown in Figure 2a. An intense absorption band at 302 cm−
1
catalytic conversion of glucose by various metal chlorides.
CrCl , CrCl , and VCl presented a group with high glucose
for the CrCl /[BMIM]Cl system is ascribed to a strong
3
2
3
3
3
6
−
34
conversion and 5-HMF yield. In contrast, PtCl2 showed
inefficient glucose conversion, and similarly to other tested
catalysts, ZrCl , MoCl , NbCl , CuCl , and FeCl , all showed
stretching vibration of Cr−Cl bonds in the anionic CrCl
With increasing time, the absorbance of the Cr−Cl band at 302
cm decreased gradually at the beginning (Figure 2) and then
showed a limited restoration (Figure S3a, Supporting
Information) after an extended period of reaction, due to the
consumption of glucose. Meanwhile, the peak intensity at 497
cm increased gradually (Figure 2a) and then decreased slowly
(Figure S3b, Supporting Information). We infer that the peak
at 497 cm− is due to the absorption of a Cr−O (from glucose)
coordination bond in this spectral region. It was verified that
the absorption peaks of glucose, fructose, and 5-HMF in
[BMIM]Cl as well as of the CrCl /[BMIM]Cl and CrCl /
.
−
1
4
5
5
2
3
poor 5-HMF selectivity. In order to understand the distinctively
different characteristics of the catalysts, the coordination
chemistry and the effect of various oxygen sources on the
catalytic aldose isomerization reactions were studied in this
work. We chose four representative metal chlorides, CrCl3,
VCl , PtCl , and FeCl , for detailed in situ FIR and a probe for
−1
1
3
2
3
reaction studies. Our objective is to identify the most critical
properties of CrCl3 underlying this most efficient aldose
isomerization catalyst.
3
3
Far infrared spectroscopy refers to the absorption of a
[BMIM]Cl/5-HMF systems (Figure S4, Supporting Informa-
tion) do not lie at 497 cm . Though the coordination between
−1
−1
substance in the 50−650 cm spectral region. We verified that
far-infrared spectroscopy is particularly suited for the
quantitative analysis of metal complexes in this reaction system
fructose and chromium is similar to that between glucose and
chromium, the amount of fructose during the reaction is known
3
3
20
according to the Bouguer−Lambert−Beer law. Metal
chlorides dissolved in [BMIM]Cl are dominated by metal−Cl
bonds, and the FIR absorption wavenumbers of different metal
chlorides are shown in Table 1. Upon addition of glucose or
other carbohydrate model compounds, some metal−Cl bonds
are replaced by metal−O bonds, as evidenced by FIR
spectroscopy. Therefore, FIR spectroscopy is a highly
informative tool to follow the changes in the coordination
chemistry of the metal complexes during the reaction. Because
the changes in the coordination chemistry and the catalytic
performance of the metal ion complexes in response to the
oxygen sources are dynamic over the course of the study at the
specified reaction temperature, it is essential to correlate them
to be very low. An isosbestic point was observed in Figure 2a,
which further reflects the transition from Cr−Cl coordination
to Cr−O
coordination with the progression of glucose
coordination. It is noted, however, that the new bands
glucose
associated with Cr−O
are much broader than that of
glucose
Cr−Cl.
It is important to understand which oxygen atoms and the
relative bond strength of the different oxygen atoms of the
glucose molecule that are coordinated to the metal ion centers.
Because the reactant glucose and the product 5-HMF have both
hydroxyl and carbonyl groups, we first studied model
compounds with either a carbonyl or a hydroxyl group, but
not both, to clarify their coordination strength to the metal
4
447
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
Figure 2. Far-infrared spectra of the CrCl /[BMIM]Cl/glucose reaction system and CrCl /[BMIM]Cl in the presence of different probing model
3
3
compounds: (a) CrCl /[BMIM]Cl/glucose system; (b) Cr−Cl stretching vibration in the CrCl /[BMIMCl]/cyclohexanone system; (c) Cr−Cl
3
3
stretching vibration in the CrCl /[BMIMCl]/n-butanol system; (d) Cr−Cl stretching vibration in the CrCl /[BMIM]Cl/water system. The far-
3
3
infrared spectra in (a)−(d) were recorded at 100 °C immediately after adding the model compounds. The background spectra of the CrCl /
3
[
BMIM]Cl system were taken before addition of model compounds. The arrows in (b)−(d) represent the recovery of the Cr−Cl coordination bond
during evaporation of the model compounds.
ions. Cyclohexanone and n-butanol were selected for having a
carbonyl group and a hydroxy group, respectively, in the study.
When these two model compounds were individually added
Scheme 1. Conversion of Glycoaldehyde in [BMIM]Cl
Containing CrCl3
to the CrCl /[BMIM]Cl system, the Cr−Cl absorbance band
3
−
1
at 302 cm decreased in response to the added model
compounds, but it was then recovered when the model
compounds were evaporated (as indicated by the upward
arrows in Figures 2b,c). The interaction between Cr(III) and
water was also studied in the same way. Figure 2d showed that
the absorbance of Cr−Cl bonds was restored gradually with the
vaporization of water at 100 °C. The results suggest that
Cr(III) forms a coordination bond with the oxygen atoms of
carbonyl, alcohol, and water. However, these bonds are so weak
that their absorption peaks are outside the far-infrared spectral
region (Figure S5, Supporting Information).
Figure 3a. When cyclohexanone and n-butyl alcohol were
separately added to the reaction system, the glucose conversion
and 5-HMF yield showed a negligible change in comparison
with the results of the CrCl /[BMIM]Cl/glucose system. The
3
results indicate that cyclohexanone and n-butyl alcohol are not
competitive against glucose in the coordination with Cr(III)
ion. However, adding glycolaldehyde dramatically suppressed
the glucose conversion and 5-HMF yield. Therefore, the
catalysis results as shown in Figure 3a and the FIR
spectroscopic results are consistent by showing that both
carbonyl and hydroxyl are weakly coordinating to the Cr(III)
ion, but glycolaldehyde competes strongly with glucose for
coordination with the Cr(III) ion. In addition, a previous study
has established that addition of glycerol to the reaction system
did not affect the glucose isomerization by chromium chloride
catalysts, suggesting that glycerol is a weaker ligand to the
Cr(III) catalyst than glucose. In a separate experiment, we
verified that ethylene glycol also did not inhibit the glucose
conversion. It can therefore be unambiguously concluded that
Glycerolaldehyde has been found to inhibit the conversion of
2
glucose to 5-HMF by CrCl . However, because glycerolalde-
2
hyde can also be converted by the same catalyst to
corresponding products that would complicate the interpreta-
41
tion of the infrared spectra, glycolaldehyde was chosen as a
model compound to probe the coordination chemistry of metal
chlorides in this study. A hydrogen transfer due to the catalysis
of CrCl in the [BMIM]Cl solvent would only be expected to
3
flip the molecule without the complication of new product
formation, as displayed in Scheme 1.
Cyclohexanone, n-butyl alcohol, and glycolaldehyde were
the coordination of glucose with CrCl responsible for the
3
evaluated as additives to study their effect on the performance
desired catalysis is mainly via the “glycolaldehyde” end group,
of CrCl -catalyzed glucose conversion to 5-HMF. Though
with a favorable ene-diol form of dioxygen coordination to the
3
18
oxygen atoms of carbonyl and alcohol compounds coordinate
with Cr(III) ions, their effects on glucose conversion are
distinctively different from that of glycolaldehyde, as seen in
metal ion.
We then used glycolaldehyde as a model compound to
3−
quantify how many Cr−Cl bonds of the CrCl6 anion in
4
448
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
Figure 3. (a) Effect of different probing model compounds on glucose conversion: (1) none; (2) glycolaldehyde; (3) n-butanol; (4) cyclohexanone.
(
b) FIR spectra of the Cr−Cl stretching vibration in the CrCl /[BMIM]Cl/glycolaldehyde system with excess glycolaldehyde. The spectra were
3
recorded at 80 °C. (c) FIR spectra of the Cr−Cl stretching vibration in the CrCl /[BMIM]Cl/glucose system with excess glucose. The spectra were
3
recorded at 100 °C. The background spectra in (b) and (c) for the CrCl /[BMIM]Cl system were recorded before addition of glycolaldehyde or
3
glucose.
Figure 4. (a) Far-infrared spectra of the V−Cl stretching vibration in the VCl /[BMIM]Cl/cyclohexanone system. (b) Far-infrared spectra of the
3
V−Cl stretching vibration in the VCl /[BMIM]Cl/glucose system. The spectra in (a) and (b) were recorded at 100 °C. (c) Effect of cyclohexanone
3
on glucose conversion and 5-HMF yield in theCrCl /[BMIM]Cl/glucose and VCl /[BMIM]Cl/glucose systems: (1) CrCl /[BMIM]Cl/glucose;
3
3
3
(
2) CrCl /[BMIM]Cl/glucose/cyclohexanone; (3) VCl /[BMIM]Cl/glucose; (4) VCl /[BMIM]Cl/glucose/cyclohexanone.
3
3
3
2
0
absorbance at about 302 cm−1 was reduced by nearly half from
0.096 to 0.047 (Figure 3b) and the same phenomenon was also
[
BMIM]Cl were replaced. When excess glycolaldehyde was
added to the CrCl /[BMIM]Cl system, the Cr−Cl bond
3
4
449
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
observed in the CrCl /[BMIM]Cl/glucose system, in which the
Scheme 2. (a) Proposed Coordination Structures between
Vanadium and the Oxygen Atoms, (b) Overall Possible
Conversion Pathways to Humins, and (c) Possible
3
Cr−Cl bond absorbance was reduced by nearly half from 0.12
a
to 0.068 (Figure 3c). Roughly half of the Cr−Cl bonds in the
[
BMIM]Cl/CrCl system were replaced due to the formation
Reaction of 5-HMF with Cyclohexanone in [BMIM]Cl
3
b
of Cr−O (coming from glucose or glycolaldehyde) coordina-
Containing VCl3
20
tion bonds and Cr−Cl−Cr bridge bonds. There exists a small
blue shift, reproducible in multiple repeated measurements, of
the Cr−Cl bond absorption wavenumber, which appears
20
consistent with the EXAFS analysis results, in which the
length of the Cr−Cl bond was shortened, due to the trans
effect. In addition, an isosbestic point was observed in Figure
3b, which is less pronounced in Figure 3c. The stronger
isosebestic point due to increased coordination by glycolalde-
hyde to Cr(III) ions, as shown in Figure 3b, occurs when the
−1
Cr−Cl absorption band at 302 cm is decreased by half of the
original Cr−Cl band intensity. The less pronounced isosbestic
point in Figure 3c appears consistent with a slightly less than
half decrease in the intensity of the original Cr−Cl absorption
band. This spectral difference in isosbestic point between the
glycolaldehyde and glucose systems may be due to the
unhindered coordination by glycolaldehyde to Cr(III) ion,
but the coordination by more than one glycolaldehyde is
evidently less favored than that by monoglycolaldehyde
coordination, as supported by the limited small decrease in
the intensity of the Cr−Cl absorption band after the
appearance of an isosbestic point. For glucose, the C2 is
directly linked to C3 of the glucose molecule; the coordination
by an additional glucose is likely restricted by steric hindrance.
To understand the differences of other metal chloride
catalysts from CrCl in glucose conversion in [BMIM]Cl, we
3
further investigated the coordination chemistries of other metal
chloride/[BMIM]Cl systems during glucose conversion and in
the presence of model probing molecules.
In the VCl /[BMIM]Cl/cyclohexanone system, the V−Cl
3
bond absorbance showed a small but noticeable decrease
(
Figure 4a), due to the formation of a V−O bond between
V(III) and the carbonyl oxygen of cyclohexanone. It is
important, however, to note that the V−O bond persisted
under conditions when cyclohexanone was evaporated. Differ-
ent from the case for the weak Cr−O bond in the CrCl /
3
[
BMIM]Cl/cyclohexanone system (Figure 2b), the bond
a
b
between V(III) ion and carbonyl oxygen is very strong once
it is formed. The coordination bonds between V(III) ion and
the oxygen atoms in n-butanol and water are relatively weak,
however, as the V−Cl FIR absorbance was restored during the
evaporation of the alcohol and water (Figure S6, Supporting
Information). In addition, in situ far infrared spectra of the
VCl /[BMIM]Cl/glucose system (Figure 4b) indicate that the
V−Cl bond absorbance declined much more than the Cr−Cl
bond absorbance in the CrCl /[BMIM]Cl/glucose system with
The complex compounds are only shown in simplified form. Only
simplified complex forms of the metal ions with the compounds are
shown.
also consistent with lower side reactions that would contribute
to humin generation.
3
In order to compare the relative rates of the side reactions
leading to humins by CrCl and VCl , we determined their r −
3
3
1
3
r paths in greater detail by measuring the conversions of
4
time, with a concomitant change in V−O bond FIR absorbance.
The strong coordination between vanadium(III) and the
carbonyl oxygen and the deep decrease in the V−Cl FIR
absorbance intensity suggest that the vanadium ion coordinates
with more than one glucose molecule and can coordinate with
the oxygen of a carbonyl group and a glycolaldehyde structure
at the same time (Scheme 2a), resulting in increased side
reactions dominated by humins (the processes represented by
r , r , and r in Scheme 2b). For the CrCl catalyst, the weak
glycolaldehyde and 5-HMF. First, the stabilities of HMF
(represented by r ) in pure [BMIM]Cl and in [BMIM]Cl
3
containing CrCl and VCl were measured. The results in
3
3
Figure S7a (Supporting Information) show that 5-HMF is
rather stable in [BMIM]Cl with and without the metal
chlorides (CrCl or VCl ). We further chose glycolaldehyde
3
3
as a model compound of glucose and fructose to study the r1
and r processes. The self-condensation reaction of compounds
2
containing the glycolaldehyde structure is followed by
measuring the consumption of glycolaldehyde with time for
the r and r processes. The results (Figure S7b, Supporting
1
2
4
3
coordination between the chromium(III) and carbonyl oxygen
and the limited decrease in the Cr−Cl absorbance intensity
suggest that the coordination structures in Scheme 2a are not
favored for CrCl . This mechanistic view of the CrCl catalyst is
1
2
Information) show that VCl condensed the glycolaldehyde
3
much more rapidly than did CrCl . It is therefore evident that
3
3
3
4
450
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
VCl favors the condensation of glucose or fructose to humins
cyclohexanone peak (Figure 6a). Evidently the strong Fe−O
bonds contribute to the nonselective catalytic performance of
Fe(III) catalyst due to the dominant formation of humins, as
supported by the catalysis study (Figure 1). In addition, more
3
especially at the first stage of the reaction, possibly through the
intermediate structures (Scheme 2a), which was supported by
the FIR spectroscopic results as discussed above. The
degradation of 5-HMF became significant only in the presence
of glycolaldehyde containing CrCl or VCl in the [BMIM]Cl−
energy is needed to generate the open glucose−FeCl complex
3
18
and 1,2-enediol−FeCl complex, which would result in less
glucose conversion in the desired reaction (Figure 1 and Figure
S9 (Supporting Information)).
3
3
3
r process (Scheme 2a). It was again observed that VCl
4
3
degraded 5-HMF more rapidly than CrCl in the presence of
3
the glycolaldehyde structure (Figure S7c, Supporting Informa-
tion). On the basis of the analysis above, we conclude that VCl3
prevails in inducing side reactions in comparison to CrCl3.
Furthermore, when nearly the same amounts of cyclo-
In addition, there is a fatal feature in the process of FeCl3
catalysis in comparison with that of CrCl catalysis. It is known
3
that 3 molar equiv of water is produced as a byproduct of
glucose conversion to 5-HMF. A good catalyst, such as CrCl3,
must be able to tolerate a sufficient amount of water. It was
found that the Fe−Cl bond FIR absorbance decreased sharply
and could not be recovered due to the strong interaction
between Fe(III) and the oxygen atom from water (Figure 6c).
Thus, the coordination sites of Fe(III) are fully blocked by the
presence of water. However, the coordination bond between
water and the Cr(III) ion is rather weak, and the Cr−Cl could
be recovered when water was evaporated (Figure 2d). Most
importantly, Figure 6d reveals that when about 30 mg of water
was added to the CrCl /[BMIM]Cl/glucose and FeCl /
hexanone were added to the CrCl /[BMIM]Cl/glucose and
3
VCl /[BMIM]Cl/glucose systems, respectively, the yield of 5-
3
HMF was suppressed in the latter system, but the 5-HMF yield
in the former system was little affected (Figure 4c). Because
both 5-HMF and cyclohexanone have a carbonyl group and the
5-HMF alone was stable in the [BMIM]Cl system, the results
in Figure S8a (Supporting Information) show that the aldol
condensation reaction of 5-HMF with cyclohexanone (Scheme
2
c) possibly occurred at a common V(III) site via the multiple
V−O coordination bonds (Figure S8b, Supporting Informa-
tion), contributing to the sharply reduced 5-HMF yield in the
3
3
[BMIM]Cl/glucose systems, respectively, the glucose con-
version and 5-HMF yield were substantially unaffected in the
former system, while the glucose conversion in the FeCl3/
[BMIM]Cl/glucose system was completely suppressed. There-
VCl /[BMIM]Cl/glucose system.
3
The PtCl catalyst represents another class of metal chlorides
2
in the glucose conversion. The spectra in Figure 5 show the
fore, water has little impact on the coordination of CrCl with
3
the hydroxyaldehyde moiety of the glucose molecule. In
contrast, FeCl has little capacity to tolerate a sufficient amount
3
of water, contributing to the lowered catalytic activity in glucose
conversion in the presence of water.
CONCLUSION
■
By using in situ far-infrared spectroscopy and a combination of
well-selected model compounds of different oxygen sources,
cyclohexanone, n-butyl alcohol, glycolaldehyde, and deionized
water, the distinctively different coordination structures of four
classes of representative catalysts, CrCl , PtCl , FeCl , and
3
2
3
VCl , with oxygen atoms of different sources were found to be
3
well correlated to the catalytic performances of the metal
chlorides. The results are summarized in Figure 7a. New
mechanistic insights were further established by correlating the
coordination chemistries of the metal chlorides with their
drastically different catalytic characteristics in glucose con-
Figure 5. Far-infrared spectra of the PtCl /[BMIM]Cl/glucose
system.
2
version. The superior performance of CrCl catalyst for the
3
FIR features of the PtCl /[BMIM]Cl/glucose system. Both the
formation of 5-HMF from glucose among the studied metal
chlorides can be ascribed to preferential coordination of CrCl3
with the glycolaldehyde group of glucose. Water generated
2
−1
glucose absorption peak at 554 cm and the Pt−Cl stretching
−1
vibration band near 310 cm showed a less pronounced
change in 80 min in comparison to that in CrCl /[BMIM]Cl/
during the reaction makes little difference in the CrCl catalytic
3
3
glucose (Figure 2a). Evidently, replacement of the Pt−Cl bond
activity due to the reversible interaction between chromium ion
and water. In addition, the relatively weaker interactions of
Cr(III) with individual oxygen atoms in the hydroxy group of
alcohols and in the carbonyl group of ketones with respect to
by a Pt−O bond is less favored, as indicated by the FIR spectra.
As a result, PtCl displays rather low catalytic activity for
2
glucose conversion (Figure 1 and Figure S9 (Supporting
Information)).
that of the glycolaldehyde group made CrCl uniquely superior
3
With the FeCl catalyst, it was found that the Fe−O bonds
for selective aldose isomerization. Isolated hydroxy and
carbonyl in 5-HMF do not inhibit the catalytic activity of
3
between the Fe(III) ion and the oxygen atoms of cyclo-
hexanone and n-butyl alcohol, the model ketone and alcohol
compounds, are very strong (Figure 6a,b). Once Fe−Cl bonds
are replaced by Fe−O bonds in the presence of n-butyl alcohol
or cyclohexanone, evaporation of n-butyl alcohol and cyclo-
hexanone does not restore the lost Fe−Cl bonds even after an
extended period of time. The removal of cyclohexanone
through evaporation is indicated by the down arrow of the
CrCl for the reaction, which also explains the stability of 5-
3
HMF in the chromium chloride/ionic liquid systems. There-
fore, CrCl showed higher catalytic selectivity in the r process
3
5
and lower catalytic selectivity in the r , r , and r processes
1
2
4
(Figure 7a). The strong coordination between vanadium and
the carbonyl oxygen and the coexisting multiple V−O
coordination bonds involving the oxygen atoms from the
4
451
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
Figure 6. (a) Far-infrared spectra of the FeCl /[BMIMCl]/cyclohexanone system. (b) Far-infrared spectra of the Fe−Cl stretching vibration in the
3
FeCl /[BMIMCl]/n-butyl alcohol system. (c) Far-infrared spectra of the Fe−Cl stretching vibration in FeCl /[BMIM]Cl, before and after the
3
3
addition of 1 drop of water. The spectra in (a)−(c) were recorded at 100 °C immediately after the addition of the model compounds. (d) Effect of
water on glucose conversion and HMF yield in the CrCl /[BMIM]Cl/glucose and FeCl /[BMIM]Cl/glucose systems: (1) CrCl /[BMIM]Cl/
3
3
3
glucose; (2) CrCl /[BMIM]Cl/H O/glucose; (3) FeCl /[BMIM]Cl/glucose; (4) FeCl /[BMIM]Cl/glucose in the presence of 30 mg of added
3
2
3
3
water.
carbonyl groups of glucose and 5-HMF molecules contribute to
Technologies was used for the infrared spectroscopy measure-
ments, and the FIR instrument model is Thermo Scientific
Nicolet iS50. The accessory is equipped with a resistance wire
for heating. In addition, a steady flow of nitrogen and a high-
temperature vacuum-grease-sealed glass lid were used above the
sample to prevent air and moisture from leaking in. As soon as
the dissolved metal halide was mixed with glucose or a model
compound, a drop of the sample was placed on the diamond
plate and then the in situ far-infrared spectrum was recorded
with a DTGS/polyethylene detector, which had 16 cm−
resolution. The background and sample were scanned for 128
times at one sitting. It was verified that the absorbance of a non-
reaction system stays unchanged even when the thickness of
the sample was changed.
the pronounced humin formation in VCl -catalyzed systems.
3
That is, VCl catalyst dominantly converts carbonyl compounds
3
to humins involved in the r , r , and r processes in Figure 7b.
1
2
4
The PtCl catalyst hardly coordinates with glucose, due to its
2
weaker coordination, and therefore PtCl is not an effective
2
catalyst for converting glucosethe r process. Because Fe(III)
1
indiscriminately forms strong and irreversible coordination
bonds with any of the oxygen sources of the selected classes of
model compounds, a process that is particularly detrimental
1
with water (Figure 7a), FeCl is a poor catalyst for selectively
3
converting glucose (Figure 7b). Solving the structures of the
oxides when the metal ions are strongly bonded by oxygens of
different sources is outside the scope of this work. Therefore,
HPLC Analysis. HPLC analysis was performed on an
Agilent 1260 series with a refractive index detector and an
Agilent Hi-Plex H column (300 × 7.7 mm, 8 μm). Diluted
CrCl presents the most favorable coordination chemistry that
3
contributes to the desired transformation of aldose to ketose
with the CrCl catalyst. Importantly, this work provides new
3
H SO solution (0.005 M) at a flow rate of 0.6 mL/min was
insights into the mechanism of the coordination properties of
some metal chlorides that are responsible for humin formation.
The knowledge obtained from this work may help guide further
research in improved aldose conversions with minimized humin
formation and for an understanding of the humin structure,
which is a long-term objective for the clean utilization of
biomass.
2
4
used as the mobile phase. The column and detector
temperatures were 65 and 50 °C, respectively. Standard
compounds were used to identify the retention times. Glycerol
was added as an internal standard for the quantitative
calculations. Glucose conversion and product yield are defined
as in eqs 1 and 2. During the course of the operation, when the
same amount (15 mol % of 50 mg glucose) of metal chloride
were dissolved in dried [BMIM]Cl, glucose was added and then
reacted at specified conditions. At the end of the reaction, cold
deionized water was used for stopping the reaction. The sample
was then diluted and analyzed by HPLC with internal standard
method. Glycerol was chosen as internal standard substance. In
EXPERIMENTAL SECTION
■
Fourier Transformation Infrared Spectroscopic Anal-
ysis. A leveled attenuated total reflectance (ATR) accessory
with a 3 mm diameter diamond plate purchased from Pike
4
452
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
not expected to promote its dissolution. In addition, exposure
to moist air was carefully prevented during our far-infrared
measurement. Also, chromium hydrate formation would also
interfere with the far-infrared measurements. Therefore, all
samples were weighed and heated in a glovebox filled with
nitrogen. After the metal chlorides were dissolved homoge-
neously, a specified amount of glucose was loaded. FIR
spectroscopic measurements or reactions were typically carried
out immediately. For reaction studies, the liquid inside the vials
was stirred at 500 rpm, at 100 °C for 1 h before HPLC analysis.
Typically, the weights of CrCl , PtCl , FeCl , and VCl were
3
2
3
3
6
.5, 11.2, 6.8, and 6.6 mg (15 mol % with respect to glucose),
respectively. The masses of [BMIM]Cl and glucose were 500
and 50 mg, respectively. In addition, the amount of model
compounds was approximately equimolar with respect to 50
mg of glucose.
ASSOCIATED CONTENT
■
*
S
Supporting Information
The following file is available free of charge on the ACS
Publications website at DOI: 10.1021/cs5012684.
Detailed sample information and background spectra for
all reaction systems (PDF)
AUTHOR INFORMATION
■
Corresponding Author
E-mail for Z.C.Z.: zczhang@yahoo.com.
Figure 7. (a) Distinctively different coordination chemistry of metal
ion with oxygen atoms of different sources. (b) Overall glucose
conversion pathways in the presence of metal chloride catalyst. The
complexes of the metal ions with the compounds are shown in
simplified form.
*
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by Chinese Government “Thousand
Talent” program funding, the National Natural Science
Foundation of China (21306186), and the China Postdoctoral
Science Foundation (2013M530952, 2013M540236).
glucose conversion (wt %)
⎛
⎞
⎟
quantity remaining of glucose by HPLC
quantity remaining of initially added glucose ⎠
=
⎜1 −
⎝
×
100%
(1)
(2)
REFERENCES
■
(
1) van Putten, R. J.; van der Waal, J. C.; de Jong, E.; Rasrendra, C.
B.; Heeres, H. J.; de Vries, J. G. Chem. Rev. 2013, 113, 1499−1597.
yield (wt %)
amount of product detected by HPLC
ideal amount of product from added feedstock
100%
(
2
2) Zhao, H. B.; Holladay, J. E.; Brown, H.; Zhang, Z. C. Science
007, 316, 1597−1600.
(3) Stahlberg, T.; Fu, W.; Woodley, J. M.; Riisager, A. ChemSusChem.
011, 4, 451−458.
(4) Zhang, Z. C. WIRES Energy Environ. 2013, 2, 655−672.
5) Choudhary, V.; Mushrif, S. H.; Ho, C.; Anderko, A.; Nikolakis,
=
̊
2
×
(
V.; Marinkovic, N. S.; Frenkel, A. I.; Sandler, S. I.; Vlachos, D. G. J.
Am. Chem. Soc. 2013, 135, 3997−4006.
addition, the amount of the glycolaldehyde, cyclohexanone and
n-butyl alcohol was added in equivalent mol to 50 mg glucose.
Sample Preparation Procedure. For the dehydration of
(
6) Stah
2, 321−325.
7) Hu, S. Q.; Zhang, Z. F.; Song, J. L.; Zhou, Y. X.; Han, B. X. Green
̊
lberg, T.; Sorensen, M. G.; Riisager, A. Green Chem. 2010,
1
(
[
BMIM]Cl, specific amounts of fresh [BMIM]Cl were weighed
into a round-bottom flask; the material was heated at 90 °C for
at least 8 h under vacuum. The flask was then filled with
nitrogen and transferred to a glovebox in preparation for the
experiments.
Specified amounts of metal chlorides and [BMIM]Cl were
weighed, and each metal chloride was dissolved in [BMIM]Cl
with agitation using a magnetic stirrer bar. Dissolution of
Chem. 2009, 11, 1746−1749.
(8) Kim, B.; Jeong, J.; Lee, D.; Kim, S.; Yoon, H. J.; Leeb, Y. S.; Cho,
J. K. Green Chem. 2011, 13, 1503−1506.
(9) Nikolla, E.; Yuriy, R. L.; Moliner, M.; Davis, M. E. ACS Catal.
2
(
011, 1, 408−410.
10) Chauvin, Y.; Di Marco-van Tiggelen, F.; Olivier, H. J. Chem.
Soc., Dalton Trans. 1993, 7, 1009−1011.
(
11) Hanz, K. R.; Riechel, T. L. Inorg. Chem. 1997, 36, 4024−4028.
anhydrous CrCl in the [BMIM]Cl was carried out immediately
3
(12) Abbott, A. P.; Capper, G.; Davies, D. L.; Rasheed, R. Inorg.
under an inert atmosphere, and the mixture was heated to 150
Chem. 2004, 43, 3447−3452.
13) Hartley, J. M.; Ip, C.-M.; Forrest, G. C. H.; Singh, K.; Gurman,
°
C and stirred at 1000 rpm for about 3 h, in order to achieve
homogeneous dissolution. It should be noted that anhydrous
(
S. J.; Ryder, K. S.; Abbott, A. P.; Frisch, G. Inorg. Chem. 2014, 53,
CrCl is insoluble in water; therefore, the presence of water is
6280−6288.
3
4
453
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454

ACS Catalysis
Research Article
(
14) Sitze, M. S.; Schreiter, E. R.; Patterson, E. V.; Freeman, R. G.
Inorg. Chem. 2001, 40, 2298−2304.
15) Hasan, M.; Kozhevnikov, I. V.; Siddiqui, M.; Rafiq, H.; Femoni,
C.; Steiner, A.; Winterton, N. Inorg. Chem. 2001, 40, 795−800.
16) Varga, Z.; Kolonits, M.; Hargitta, M. Inorg. Chem. 2010, 49,
039−1045.
17) Abbott, A. P.; Frisch, G.; Ryder, K. S. Annu. Rep. Prog. Chem.,
Sect. A 2008, 104, 21−45.
18) Guan, J.; Cao, Q.; Guo, X. C.; Mu, X. D. Comput. Theor. Chem.
011, 963, 453−462.
19) Pidko, E. A.; Degirmenci, V.; Hensen, E. J. M. ChemCatChem
012, 4, 1263−1271.
20) Zhang, Y. M.; Pidko, E. V.; Hensen, E. J. M. Chem. Eur. J. 2011,
7, 5281−5288.
21) Binder, J. B.; Raines, R. T. J. Am. Chem. Soc. 2009, 131, 1979−
985.
22) Yong, G.; Zhang, Y.; Ying, J. Y. Angew. Chem., Int. Ed. 2008, 47,
345−9348.
23) Li, C.; Zhang, Z.; Zhao, Z. K. Tetrahedron Lett. 2009, 50, 5403−
405.
24) Yuan, Z. S.; Xu, C. C.; Cheng, S.; Leitch, M. Carbohydr. Res.
011, 346, 2019−2023.
25) Su, Y.; Brown, H. M.; Huang, X. W.; Zhou, X. D.; Amonette, J.
E.; Zhang, Z. C. Appl. Catal. A: Gen. 2009, 361, 117−122.
26) Su, Y.; Brown, H. M.; Li, G. S.; Zhou, X. D.; Amonette, J. E.;
Fulton, J. L.; Camaioni, D. M.; Zhang, Z. C. Appl. Catal. A: Gen. 2011,
91, 436−442.
27) Pidko, E. A.; Degirmenci, V.; Van, S.; Rutger, A.; Hensen, E. J.
(
(
1
(
(
2
(
2
(
1
(
1
(
9
(
5
(
2
(
(
3
(
M. Angew. Chem., Int. Ed. 2010, 49, 2530−2534.
(28) Zhang, J.; Weitz, E. ACS Catal. 2012, 2, 1211−1218.
(29) Ranoux, A.; Djanashvili, K.; Isabel W. C. E, A.; Hanefeld, U.
ACS Catal. 2013, 3, 760−763.
(
30) Binder, J. B.; Cefali, A. V.; Blankc, J. J.; Raines, R. T. Energy
Environ. Sci. 2010, 3, 765−771.
(31) Zhang, Z. C. WIREs Energy Environ. 2013, 2, 655−672.
(32) Rasrendra, C. B.; Soetedlo, J. N. M.; Makertihartha, I. G. B. N.;
Adisaamito, S.; Heeres, H. J. Top. Catal. 2012, 55, 543−549.
33) Griffiths, P. R.; de Haseth, J. A.; Fourier Transform Infrared
Spectrometry, 2nd ed.; Wiley: Hoboken, NJ, 2007; p198.
(
(
(
34) Leipzig, J. A. B. Z. Anorg. allg. Chem. 1972, 390, 210−216.
35) Avery, J. S.; Burbridge, C. D.; Goodgame, D. M. L. Spectrochim.
Acta, Part A: Mol. Spectrosc. 1968, 24, 1721−1726.
36) Walton, R. A.; Brisdon, B. J. Spectrochim. Acta, Part A: Mol.
Spectrosc. 1967, 23, 2222−2223.
37) Goggin, P. L.; Mink, J. J. Chem. Soc., Dalton Trans. 1974, 14,
479−1483.
38) Nakamoto, K. Infrared and Raman Spectra of Inorganic and
Coordination Compounds, 3rd ed.; Wiley: New York, 1978; pp 107−
80.
39) Fumino, K.; Fossog, V.; Stange, P.; Wittler, K.; Polet, W.;
Hempelmann, R.; Ludwig, R. ChemPhysChem 2014, 15, 2604−2609.
40) Griffiths, P. R.; de Haseth, J. A. Fourier Transform Infrared
Spectrometry, 2nd ed.; Wiley: Hoboken, NJ, 2007; p 146.
41) Holladay, J. E.; Brown, H. M.; Appel, A. M.; Zhang, Z. C. In
(
(
1
(
1
(
(
(
Catalysis of Organic Reactions: Twenty-second Conference; Prunier, M. L.,
Ed.; CRC Press: Boca Raton, FL, 2008; Chemical Industries Series.
4
454
dx.doi.org/10.1021/cs5012684 | ACS Catal. 2014, 4, 4446−4454