Dehydration of different ketoses and aldoses to 5-hydroxymethylfurfural

Article abstract of DOI:10.1002/cssc.201300345

5-Hydroxymethylfurfural (HMF) is considered an important building block for future bio-based chemicals. Here, we present an experimental study using different ketoses (fructose, sorbose, tagatose) and aldoses (glucose, mannose, galactose) under aqueous acidic conditions (65 g L^-1 substrate, 100-160 °C, 33-300 mM H₂SO₄) to gain insights into reaction pathways for hexose dehydration to HMF. Both reaction rates and HMF selectivities were significantly higher for ketoses than for aldoses, which is in line with literature. Screening and kinetic experiments showed that the reactivity of the different ketoses is a function of the hydroxyl group orientation at the C3 and C4 positions. These results, in combination with DFT calculations, point to a dehydration mechanism involving cyclic intermediates. For aldoses, no influence of the hydroxyl group orientation was observed, indicating a different rate-determining step. The combination of the knowledge from the literature and the findings in this work indicates that aldoses require an isomerization to ketose prior to dehydration to obtain high HMF yields. Ketose dries out: Hexose dehydration to 5-hydroxymethylfurfural is studied using different ketoses and aldoses. The reactivity of the different ketoses is found to be a function of the hydroxyl group orientation at the C3 and C4 positions. The results point to a dehydration mechanism involving cyclic intermediates. For aldoses, no influence of the hydroxyl group orientation is observed, indicating a different rate-determining step. Copyright

Full text of DOI:10.1002/cssc.201300345

CHEMSUSCHEM
FULL PAPERS
DOI: 10.1002/cssc.201300345
Dehydration of Different Ketoses and Aldoses to 5-
Hydroxymethylfurfural
Robert-Jan van Putten,[a, b] Jenny N. M. Soetedjo,[b] Evgeny A. Pidko,[c] Jan C. van der Waal,[a]
Emiel J. M. Hensen,[c] Ed de Jong,*[a] and Hero J. Heeres*[b]
5-Hydroxymethylfurfural (HMF) is considered an important
building block for future bio-based chemicals. Here, we pres-
ent an experimental study using different ketoses (fructose,
sorbose, tagatose) and aldoses (glucose, mannose, galactose)
under aqueous acidic conditions (65 gLꢀ1 substrate, 100–
1608C, 33–300 mm H2SO4) to gain insights into reaction path-
ways for hexose dehydration to HMF. Both reaction rates and
HMF selectivities were significantly higher for ketoses than for
aldoses, which is in line with literature. Screening and kinetic
experiments showed that the reactivity of the different ketoses
is a function of the hydroxyl group orientation at the C3 and
C4 positions. These results, in combination with DFT calcula-
tions, point to a dehydration mechanism involving cyclic inter-
mediates. For aldoses, no influence of the hydroxyl group ori-
entation was observed, indicating a different rate-determining
step. The combination of the knowledge from the literature
and the findings in this work indicates that aldoses require an
isomerization to ketose prior to dehydration to obtain high
HMF yields.
Introduction
The replacement of fossil feedstocks with sustainable resources
for energy generation, transportation fuels, bulk and fine
chemicals, and materials is currently considered as a pivotal
challenge, receiving increasing social and scientific interest.
To achieve this, alternative sources of organic carbon need to
be found. Biomass is an attractive option as it is the most
abundant non-fossil source of organic carbon. Biomass mainly
comprises carbohydrates, lignin, fatty acids, lipids, and pro-
teins. Carbohydrates represent the largest fraction of biomass,
predominantly present in polymeric form (cellulose, hemi-cellu-
lose, starch, inulin) and are built up from hexoses (glucose,
fructose, mannose, galactose) and pentoses (arabinose, xylose).
The acid-catalyzed dehydration of pentoses[1,2] and hexoses[3]
leads to the formation of furfural and 5-hydroxymethylfurfural
(HMF), respectively, along with many by-products. Both mole-
cules, and derivatives thereof, are in Bozell’s ‘Top 10+4’ list of
bio-based chemicals and are considered to be key components
in the development of a bio-based economy.[4] This has led to
an enormous increase in research published on acid-catalyzed
dehydration of sugars over the last decade.[2,3]
Both furfural and 5-hydroxymethylfurfural (HMF) can be
used in different application areas. Furfural has high potential
in fuel and solvent applications. HMF is considered a promising
platform chemical due to its high derivatization potential.
It can be converted to a wide range of interesting bulk and
fine chemicals, for instance, as a monomer for new bio-based
polymers. Avantium is currently developing a process for the
production of polyethylenefurandicarboxylate (PEF) from C6
sugars as a next-generation replacement material for polyethy-
leneterephtalate (PET), having improved barrier properties.[5]
Attention to the development of highly efficient routes to
HMF has strongly increased in recent years.[3] Glucose or glu-
cose-based oligomers and polymers, especially those derived
from lignocellulosic sources, are favored feedstocks due to
their availability and presence in agricultural side streams and
other waste.[6] The vast majority of experimental studies, how-
ever, show that fructose, a ketose, is much more efficiently de-
hydrated to HMF than glucose, an aldose. Under aqueous
acidic conditions, fructose yields a maximum of around 50%
HMF at best because of the formation of polymeric material,
known as humins, and hydration of HMF to levulinic and
formic acids (Scheme 1).[3,7,8] For glucose, the maximum HMF
yield is only around 5%. Higher HMF yields from fructose
(>80%) have been reported in other solvent systems, especial-
ly in ionic liquids and aprotic polar solvents such as DMSO.[3]
Work on glucose dehydration with heterogeneous base/acid
bi-catalytic systems,[9] chromium-catalyzed glucose dehydration
in ionic liquids[10,11] and organic solvents,[12] and other cata-
lysts,[13] indicate that, apart from an acid, an additional catalyst
[a] R.-J. van Putten, Dr. J. C. van der Waal, Dr. E. de Jong
Avantium Chemicals
Zekeringstraat 29, 1014 BV Amsterdam (The Netherlands)
E-mail: ed.dejong@avantium.com
[b] R.-J. van Putten, J. N. M. Soetedjo, Prof. Dr. H. J. Heeres
Department of Chemical Engineering
University of Groningen
Nijenborgh 4, 9747 AG Groningen (The Netherlands)
E-mail: h.j.heeres@rug.nl
[c] Dr. E. A. Pidko, Prof. Dr. E. J. M. Hensen
Laboratory of Inorganic Materials Chemistry
Schuit Institute of Catalysis
Eindhoven University of Technology
P.O. Box 513, 5600 MB Eindhoven (The Netherlands)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201300345.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 1681 – 1687 1681

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Scheme 1. Hydration of HMF to levulinic acid and formic acid.
Scheme 4. Ketoses used in this study.
is required to efficiently dehydrate glucose to HMF. This addi-
tional catalyst is generally believed to facilitate the isomeriza-
tion of glucose to fructose prior to dehydration to HMF.[3]
To improve the yields, significant steps must be made in the
development of catalysts, preferably heterogeneous in nature.
To do so, a detailed knowledge of the reaction mechanism of
the main and side reactions is required. A number of reaction
mechanisms have been proposed for the dehydration reaction
in water, though no definitive evidence has yet been found to
confirm these.[3] The postulated mechanisms can be divided
into mechanisms with cyclic (Scheme 2) or acyclic intermedi-
ates (Scheme 3).
Scheme 5. Aldoses used in this study.
First a high-throughput screening study was performed at
a range of temperatures, acid concentrations, and reaction
times to identify trends in reactivity for various ketoses and
aldoses. Subsequently, a detailed
kinetic study at relevant process
conditions, established in the
screening, was performed to de-
termine the kinetic constants
and activation energies for the
rate of reaction of the various
Scheme 2. Proposed dehydration mechanism with cyclic intermediates.
hexose feeds. Finally, DFT calcu-
lations were performed, the re-
sults of which were compared
with the experimental data.
Results and Discussion
High-throughput screening
To determine trends in the reac-
tivity of ketoses (fructose, taga-
tose, and sorbose),
a
high-
throughput screening in a batch
mode was performed. Experi-
ments were performed at 100,
120, and 1408C in water contain-
Scheme 3. Proposed dehydration mechanism with acyclic intermediates.
ing 33, 100, or 300 mm sulfuric
acid at 30, 45, and 60 min reac-
Knowledge about the effect of the relative orientation of the
hydroxyl groups of different ketoses and aldoses on the dehy-
dration rate and selectivity to HMF is scarcely available in the
literature. Seri et al. reported a lower HMF yield from sorbose
than from fructose in DMSO, though no conversion data were
provided.[14] Determination of the influence of the orientation
of hydroxyl groups in hexoses could very well provide new in-
sights into the mechanisms of ketose and aldose dehydration.
For this reason the acid-catalyzed dehydration of fructose, sor-
bose, and tagatose (all ketoses, Scheme 4), as well as glucose,
mannose, and galactose (all aldoses, Scheme 5) in an aqueous
environment was studied by using sulfuric acid as the catalyst.
A complementary, integrated research strategy was applied.
tion time and a fixed initial hexose concentration of 65 gLꢀ1
(0.36 m). The results for the experiments at 1208C with
100 mm sulfuric acid are given in Figure 1 and Figure 2.
In Figure 1, the conversion of the three ketoses is plotted
against the reaction time. After 30 min, over 60% of the taga-
tose is converted, compared to around 40% conversion for
fructose and sorbose. After 60 min, over 90% of the tagatose
is converted, whereas around 70% of fructose and less than
60% of sorbose is converted. This shows a clear difference in
reaction rates between the ketoses, with tagatose by far the
most reactive. This trend was observed for the entire dataset,
see the Supporting Information for more details (Figure S1–S3,
the Supporting Information).
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 1681 – 1687 1682

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Figure 3. HMF yield against conversion for fructose (~), tagatose (&), and
sorbose (*). Experiments were performed at 100, 120 and 1408C with 33,
100, or 300 mm H2SO4 at 30, 45, and 60 min reaction time. Duplicate experi-
ments are shown.
Figure 1. Conversion of fructose (~), tagatose (&), and sorbose (*)
(65 gLꢀ1) against time at 1208C with 100 mm aqueous sulfuric acid. Dupli-
cate experiments are shown.
ic and formic acid also occurs to a larger extent, leading to
lower HMF yields.[7,8]
Figure 3 provides an overview of all experimental data ob-
tained in the high-throughput screening. The experimental
error in some cases is rather large because of the fact that this
was a screening with the purpose of identifying general trends
for detailed kinetic experiments. The maximum attainable yield
of HMF is about the same for all three ketoses within the
margin of error, and is around 40–45% at 90% conversion.
This is in the lower range of the general trends reported in the
literature for acid-catalyzed fructose dehydration in water.[3]
Figure 2. HMF yield from fructose (~), tagatose (&), and sorbose (*)
(65 gLꢀ1) against time at 1208C with 100 mm aqueous sulfuric acid. Dupli-
cate experiments are shown.
The HMF yield over time at 1208C with 100 mm sulfuric acid
is given in Figure 2. The graph shows that, under these condi-
tions, the HMF yield is consistently the highest for tagatose.
The HMF yield increases in time up to about 40% at a conver-
sion of around 85% (Figure 1) and then remains constant as
a result of the subsequent decomposition of HMF to levulinic
acid and formic acid. The HMF yields from fructose and sor-
bose are equal after 30 min (around 22%) and increase to 35%
and 28%, respectively, after 60 min reaction time.
Figure 4. Levulinic acid yield against conversion for fructose (~), tagatose
(&), and sorbose (*). Experiments were performed at 100, 120 and 1408C
with 33, 100 or 300 mm H2SO4 at 30, 45 and 60 min reaction time. Duplicate
experiments are shown.
The experimental results for the high-throughput screening
of the ketoses, exemplified by Figure 1 and Figure 2, indicate
that tagatose possesses the highest reactivity for the acid-cata-
lyzed dehydration in the entire process window. The difference
in reaction rate between fructose and sorbose is very small
and within the error of experimentation. The key difference be-
tween the ketoses tested is the orientation of hydroxyl groups
on the C3 and C4 position. This is a strong indication that the
stereochemistry at C3 and C4 has a significant influence on the
reactivity.
Figure 4 shows a strong increase in levulinic acid yield at
high conversions. Together with the decrease in HMF yield at
high conversions (Figure 3), this confirms that, at high conver-
sion, the HMF yield is reduced by a consecutive reaction to lev-
ulinic acid and formic acid, as described in the literature.[3,7,8]
Apart from the ketoses discussed above, three aldoses were
also screened in batch mode at similar conditions as for the
ketoses. Glucose, mannose, and galactose were reacted at 120,
140, and 1608C with 33, 100, and 300 mm sulfuric acid for 30–
60 min. A higher temperature range was selected because glu-
cose is known to be less reactive than fructose.[3] The observed
maximum HMF yield was around 5%, at conversions of 30–
70%, which is in line with the majority of the literature.[3]
Under reaction conditions in which hexose conversion is
complete, the HMF yield from tagatose is consistently lower
than the HMF yield from fructose and sorbose under the same
conditions (Figure S4, the Supporting Information). This can be
explained by the higher reaction rate of tagatose. As a result,
the subsequent HMF decomposition to, among others, levulin-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 1681 – 1687 1683

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Furthermore, no clear differences in aldose conversion rates or
selectivities to HMF were observed for the three aldoses. This
indicates that the orientation of the hydroxyl groups does not
affect the dehydration rate of the aldoses, which suggests that
the acid-catalyzed aldose dehydration possesses a different
rate-determining step than the acid-catalyzed ketose de-
hydration.
Table 1. Kinetic data for the acid-catalyzed dehydration of different
ketoses.
Ketose
k
ref [Lmolꢀ1 minꢀ1
]
Ea [kJmolꢀ1
]
R2
sorbose
fructose
tagatose
0.9ꢃ0.2
0.9ꢃ0.1
2.0ꢃ0.6
138ꢃ41
124ꢃ22
89ꢃ15
0.976
0.989
0.990
ketose. This is clearly illustrated by the significantly higher k1ref
value and significantly lower Ea value for acid-catalyzed dehy-
dration of tagatose than for the acid-catalyzed dehydration of
fructose and sorbose. The k1ref and Ea values for sorbose and
fructose dehydration are essentially equal within the experi-
mental error.
Kinetic study
The screening results indicated clear differences in the reactivi-
ty of the three ketoses. In the next step, kinetic experiments
were performed to quantify the differences in reactivity of the
ketoses at 1378C with 33, 100, and 300 mm sulfuric acid in
a time frame of 0–90 min. The experiments were performed in
a batch mode using glass ampoules. At t=0 min, the am-
poules were placed in an oven at 1378C and allowed to react
for a predetermined time. The experimental data were mod-
eled using a simple kinetic expression assuming first order in
sugar concentration and acid concentration [Eqs. (1) and (2)].
Figure 6 shows the yield against the conversion for all kinet-
ic experiments. For fructose and tagatose, the general trend is
the same as for the experiments in the high-throughput
r ¼ ꢀk½Sugarꢁ½H2SO4ꢁ
ð1Þ
ð2Þ
ð3Þ
Ea
1
1
ꢂ
ꢀT
k ¼ krefeꢀ
ðR Þ ðT
Þ
real
ref
Treal ¼ TrefꢀðTrefꢀTiÞeðꢀhtÞ
Here, kref describes the reaction rate constant at 1378C (Tref)
and Ti is the initial temperature (258C). At the initial stage of
the reaction, the temperature is not constant as it takes typi-
cally 5–10 min to reach 1378C. This effect was compensated
for by extending the model with an energy balance. After inte-
gration, Equation (3) is obtained, which, combined with the
mass balances in batch for the hexoses, allows calculation of
the concentration and the temperature as a function of the
Figure 6. The HMF yield plotted against conversion for the kinetic studies
using the three ketoses at 1378C with 33, 100, and 300 mm acid at 0–90 min
[fructose (~), tagatose (&), and sorbose (*)].
screening, with approximately 55% selectivity up to around
60% conversion and a maximum HMF yield of just over 40%
at around 85% conversion. The trend for sorbose, however, is
different. At conversions higher than 30%, the HMF yield line
for sorbose is below that of the other two ketoses. This indi-
cates that additional reactions are involved in sorbose dehy-
dration. This effect was not observed in the screening, presum-
ably due to analytical issues (e.g., overlapping product peaks).
Further research is in progress to identify additional reaction
products.
batch time. The value for the parameter h [0.2135 minꢀ1
,
Eq. (3)] was determined independently.[7]
In Figure 5 the experimental data points (1378C, 33 mm sul-
furic acid) and the model line for the three ketoses are provid-
ed. Agreement between model and experimental data is very
satisfactory (c.f. R2 values and the error in k1ref as seen in
Table 1). The results from the kinetic study are in line with the
screening experiments, with tagatose being the most reactive
DFT calculations
To gain insights into the origin of the experimentally observed
differences in reactivity between the three ketoses, model DFT
calculations were performed. It has previously been proposed
that the reactivity and selectivity of the acid-catalyzed hexose
conversion is determined by the regeoselectivity of the initial
protonation and the accompanying dehydration step.[15]
A characteristic aspect of fructofuranose dehydration by Brøn-
sted acids is the preferred protonation of the OH group at the
anomeric C2 carbon (Figure 7a) that initiates a sequence of
fast reactions towards HMF. In the case of glucose conversion,
the initial protonation is much more difficult.[15] Here, the pre-
ferred protonation site is the anomeric hydroxyl at C1. Howev-
Figure 5. The concentration of sugar in time at 1378C with 33 mm sulfuric
acid, with the points describing the experimental results [fructose (~), taga-
tose (&), and sorbose (*)] and the lines describing the model.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 1681 – 1687 1684

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
trations are consistent and fit well with the experimental re-
sults. A possible explanation for the differences in reactivity of
the ketoses is the relative orientation of their hydroxyl groups
at the C3 and C4 position.
DFT calculations on acid-catalyzed furanose dehydration in-
dicate that the carbocation formed by dehydration at C2 is
most stable for tagatose, followed by fructose and sorbose.
This is also the order of decreasing reactivity as found in the ki-
netic study, suggesting a relationship between the determined
activation energies and the calculated DG from the DFT calcu-
lations (Figure 8).
The studies provided above imply that the differences in re-
action rate of the different ketoses are due to the orientation
of the hydroxyl groups on C3 and C4. This suggests that the
reaction follows
a mechanism with cyclic intermediates
Figure 7. (a) Furanose dehydration initiated by the preferred protonation of
the anomeric hydroxyl group and (b) the optimised structures of the cation-
ic dehydration products originating from psico-, sorbo-, fructo-, and tagato-
furanose and the respective DFT-computed average Gibbs free energies of
dehydration of alpha- and beta-anomers of the sugars (kJmolꢀ1). The arrows
depict the direction of the dipole moment in the cations.
er, its activation does not lead to the desired products but
rather opens pathways towards humin precursors. In both
cases, the reactivity is determined by the stability of cationic
intermediates resulting from the dehydration of the initial pro-
tonated complex.
Figure 8. The observed Ea plotted against calculated DG8298K of the single
dehydrated ketose carbocations.
Reaction Gibbs free energies (DG8298K) of the dehydration of
alpha- and beta-anomers of fructo-, sorbo-, psico-, and tagato-
furanose in water resulting from protonation of the anomeric
hydroxyl group were calculated at the B3LYP/6-311+G(d,p)
level of theory using the Gaussian 09 program.[16] Bulk solvent
effects were approximated using the polarized continuum sol-
vent (PCM) model within the conductor reaction field (COSMO)
(Scheme 2). For a dehydration mechanism with acyclic inter-
mediates (Scheme 3), it is more difficult to explain the signifi-
cant differences in reactivity of the different ketoses. The
mechanisms are based on a series of ß-dehydrations. Accord-
ing to literature, intermediates are generally not observed in
water-based acid-catalyzed fructose dehydration, which implies
that the formation of the 1,2-enediol would be the rate-deter-
mining step in the case of a mechanism with acyclic intermedi-
ates, as shown in Scheme 6. This shows that the orientation of
the hydroxyl group on the chiral C3 of the 1,2-enediol relative
to the double bond between C1 and C2 is not relevant, as C2
and C1 are both achiral. The same can be said for C4 in 3-de-
oxyhexosulose. This means that any differences in reactivity
between the ketoses can only be explained on the basis of the
interactions between groups on C3 with those on C4 and C5,
which are connected with free-rotating carbon–carbon bonds,
which in turn makes large differences in reaction rates less
likely.
+
approach.[17] The Brønsted acid was modeled as a H5O2
cation.[15]
The optimized structures of the cationic sugar dehydration
products and the respective DG8298K values are summarized in
Figure 7b. The most favorable reaction is predicted for tagato-
furanose. The free energies of fructofuranose and sorbofura-
nose dehydration are very similar and are substantially higher
than that computed for psicofuranose, which was not included
in the experimental study. Interestingly, it appears that the
computed DG8298K values correlate with the direction of the
dipole moment in the dehydration products. The most stable
cationic species is formed when the dipole moment is oriented
parallel to the furanose ring (tagatofuranose). Devia-
tions from such an orientation lead to the destabiliza-
tion of the dehydrated cation (Figure 7b).
Mechanistic aspect
The experimental data from both the screening and
kinetic experiments clearly show differences in reac-
tivity of the different ketoses. The kinetic models ob-
tained for the three ketoses at different acid concen- Scheme 6. Formation of HMF from hexose through acyclic intermediates.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 1681 – 1687 1685

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
dration under acidic conditions is the isomerization to
a ketose. The so-called Lobry de Bruyn–Alberda Van Eekensteijn
transformation for aldose–ketose isomerization is base-cata-
lyzed.[20] It is thus not surprising that the HMF yield in dehydra-
tions from aldoses under acidic conditions is so low. This infor-
mation furthermore suggests a cyclic dehydration mechanism
starting from the ketofuranose form, as 1,2-enediol formation
appears to be favored in the presence of basic catalysts.
Scheme 7. Possible tautomers of ketohexoses.
Conclusions
In this study, the combination of high throughput screening
and detailed kinetic studies was successfully applied to deter-
mine the differences in reactivity between hexoses in the acid-
catalyzed dehydration to HMF. The screening showed that dif-
ferent ketoses have different reactivities in acid-catalyzed dehy-
dration to HMF, which must be caused by the different orienta-
tion of the hydroxyl groups on C3 and C4. Tagatose showed
significantly higher conversion rates than fructose and sorbose.
Differences between fructose and sorbose were much smaller.
For these ketoses (sorbose, fructose, and tagatose) k and Ea
values were successfully determined from kinetic experiments.
Between the aldoses tested (glucose, mannose, and galac-
tose), no differences in reactivity were observed. This indicates
that HMF formation from ketoses and aldoses involves differ-
ent rate-determining steps.
Since a cyclic reaction mechanism requires the ketose to
react through its furanose form, it is important to take the tau-
tomeric distribution into account. The different tautomeric
forms of ketohexoses are shown in Scheme 7. The tautomeric
distribution of the ketohexoses in water was studied by Que
and Gray (Table 2),[18] showing that, of the ketoses addressed in
Table 2. Proportions of pyranose and furanose forms of ketoses at equi-
librium in aqueous solutions at 308C.[18]
Ketose
a-Pyranose
b-Pyranose
a-Furanose
b-Furanose
fructose
tagatose
sorbose
psicose
0
72
15
0
5
5
5
38
23
9
0
71
95
26
21
15
Initial screening experiments using UPLC analyses indicated
a maximum HMF yield of around 40% at around 85% conver-
sion. The kinetic experiments in combination with HPLC analy-
ses, however, showed that the maximum HMF yield from sor-
bose was only 30%, which leads to the conclusion that a differ-
ent parallel pathway is available for sorbose. This aspect is still
under investigation.
this research, fructose has the largest fraction present in the
furanose forms (28%) at equilibrium at 308C, followed by taga-
tose (14%) and sorbose (5%). There is no relationship between
these values and the differences in reaction rate observed in
our study. A study on galactopyranose tautomerization in
water showed that at slightly acidic conditions (pH 4.3) equilib-
rium can be reached after about 1 h at room temperature.[19]
At increasing temperatures and acid concentrations, this rate
will increase. Therefore, it can be safely assumed that the tau-
tomerization rate is much higher than the dehydration rate
under the conditions studied in this research. This, in combina-
tion with the data in Table 2, excludes the tautomeric distribu-
tion as an important factor in the dehydration rate.
DFT calculations on the ketoses indicate that tagatose forms
the most stable furanose carbocation upon single dehydration
at the anomeric carbon (C2), followed by fructose and sorbose.
As this is also observed to be the order of decreasing reactivity,
the calculations provide a plausible explanation for the ob-
tained results; a dehydration mechanism with cyclic inter-
mediates.
The difference in HMF yield between sorbose and the other
two ketoses is difficult to explain. The DFT calculations did indi-
cate that the proposed carbocation intermediate for sorbose is
less stable than for fructose and, especially, for tagatose. This
could be favorable for a parallel reaction pathway that results
in the formation of other products. This phenomenon is cur-
rently under investigation.
The results of the experimental data in combination with
DFT calculations are best explained by the formation of HMF
through a ketose dehydration mechanism with cyclic inter-
mediates. This means that for aldose conversion to HMF, an
isomerization to a ketose would be required prior to dehydra-
tion. The fact that this is typically base-catalyzed explains why
the acid-catalyzed dehydration of aldoses is much less efficient
than the acid-catalyzed dehydration of ketoses.
Another important observation in this study is that the ori-
entation of the hydroxyl groups on the aldoses has no influ-
ence on their dehydration rate, indicating that this orientation
has no effect on the rate-determining step. It suggests that
aldose dehydration to HMF involves a different rate-determin-
ing step than ketose dehydration to HMF, which likely de-
mands a different type of catalyst. As a number of researchers
have found evidence that suggests that an isomerization co-
catalyst is required in the glucose dehydration to HMF,[3,9,10]
this could mean that the rate-determining step in aldose dehy-
Experimental Section
Screening: Glucose (99%), fructose (99%), sulfuric acid (96%), 1,4-
dioxane (99.8%), and saccharine (98%) were purchased from
Sigma–Aldrich, galactose (99%), and mannose (99%) were pur-
chased from Acros and sorbose (99.6%) and tagatose (99.8%) were
purchased from Carbosynth. Milli-Q quality water was used for all
experiments and sample preparations. All experiments were per-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 1681 – 1687 1686

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Wiley-VCH, Weinheim, Germany, 2010, pp. 167–186; b) J.-P. Lange, E.
van der Heide, J. van Buijtenen, R. Price, ChemSusChem 2012, 5, 150–
166.
formed at a 1 mL scale under 20 bar N2 in batch on an Avantium
Quick Catalyst Screening system with a substrate concentration of
65 gLꢀ1 (0.36m) in water. Sulfuric acid concentrations of 33, 100,
and 300 mm were tested. The ketoses were reacted at 100, 120,
and 1408C and the aldoses were reacted at 120, 140, and 1608C.
After the appropriate reaction time, the reactor blocks were cooled
in an ice bath. The analysis of sugars and furans was performed on
a Waters Acquity UPLC with an Acquity ULPC BEH C18 2.1ꢁ
5.0 mm, 1.7 mm column (the Supporting Information, Table S1). The
sugars were detected using an evaporative light scattering detec-
tor and HMF was detected on a photodiode array detector at
230 nm. Saccharine was used as external standard. The analysis of
levulinic acid was performed on an Interscience TraceGC with an
Agilent J&W FactorFour VF-WAXms, 30 mꢁ0.25 mm, 0.25 mm
column with a flame ionization detector and 1,4-dioxane as the ex-
ternal standard. The standard mixture was added to the reaction
mixture upon opening of the reactor. This was followed by the ap-
propriate dilutions for UPLC and GC analyses.
[2] R.-J. van Putten, A. S. Dias, E. de Jong in Catalytic Process Development
for Renewable Materials, 1st ed. (Eds.: P. Imhof, J. C. van der Waal), Wiley-
VCH, Weinheim, 2013, pp. 81–117.
[3] R.-J. van Putten, J. C. van der Waal, E. de Jong, C. B. Rasrendra, H. J.
Heeres, J. G. de Vries, Chem. Rev. 2013, 113, 1499–1597.
[4] J. J. Bozell, G. R. Petersen, Green Chem. 2010, 12, 539–554.
[5] a) G. J. M. Gruter, F. Dautzenberg (Avantium International BV), WO
2007104514, 2007; b) C. Munoz de Diego, W. P. Schammel, M. A. Dam,
G. J. M. Gruter (Furanix Technologies BV), WO2011043660, 2011; c) L.
Sipos (Furanix Technologies BV), WO2010077133, 2010; d) E. De Jong,
M. A. Dam, L. Sipos, G. J. M. Gruter, in ACS Symposium Series (Eds.: P. B.
Smith, R. Gross), 2012, pp. 1–13. DOI: 10.1021/bk-2012-1105.ch001.
[6] a) A. Bauen, G. Berndes, M. Junginger, M. Londo, F. Vuille, R. Ball, T. Bole,
C. Chudziak, A. Faaij, H. Mozaffarian, Bioenergy: A Sustainable and Reli-
able Energy Source. A Review of Status and Prospects, http://www.iea-
bioenergy.com/LibItem.aspx?id=6479 2009, p. 108; b) J. C. Serrano-
Ruiz, J. A. Dumesic, Energy Environ. Sci. 2011, 4, 83–99.
[7] B. Girisuta, L. P. B. M. Janssen, H. J. Heeres, Green Chem. 2006, 8, 701–
709.
[8] B. F. M. Kuster, Starch/Staerke 1990, 42, 314–321.
Kinetic experiments: Fructose (99%), sorbose (98%), tagatose
(98%) and sulfuric acid (96%) were purchased from Sigma–Aldrich.
Milli-Q quality water was used for all experiments and sample
preparations. All experiments were performed at a 0.5 mL scale in
sealed glass ampoules which were heated in an oven at 1378C. A
substrate concentration of 65 gLꢀ1 (0.36m) in water was used. Sul-
furic acid concentrations of 33, 100, and 300 mm were used. The
ampoules were cooled in cold water after the appropriate reaction
time. The reaction mixture was then filtered over a 0.45 mm PTFE
syringe filter and diluted 7–9 times in water for analysis on an Agi-
lent 1200 HPLC with a Bio-rad Aminex HPX-87H column. 5 mm sul-
[9] a) M. Ohara, A. Takagaki, S. Nishimura, K. Ebitani, Appl. Catal. A 2010,
383, 149–155; b) E. Nikolla, Y. Romꢂn-Leshkov, M. Moliner, M. E. Davis,
ACS Catal. 2011, 1, 408–410; c) P. M. Grande, C. Bergs, P. Dominguez de
Maria, ChemSusChem 2012, 5, 1203–1206; d) A. Chareonlimkun, V.
Champreda, A. Shotipruk, N. Laosiripojana, Fuel 2010, 89, 2873–2880.
[10] H. Zhao, J. E. Holladay, H. Brown, Z. C. Zhang, Science 2007, 316, 1597.
[11] 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–
2534.
[12] J. B. Binder, R. T. Raines, J. Am. Chem. Soc. 2009, 131, 1979–1985.
[13] T. Stꢃhlberg, S. Rodriguez-Rodriguez, P. Fristrup, A. Riisager, Chem. Eur. J.
2011, 17, 1456–1464; S. Hu, Z. Zhang, J. Song, Y. Zhou, B. Han, Green
Chem. 2009, 11, 1746–1749.
furic acid was used as the eluent with a flow rate of 0.55 mLminꢀ1
Refractive index and UV (210 nm) detection were used.
.
Determination of the kinetic parameters: The kinetic parameters
were determined using a maximum likelihood approach, which is
based on minimization of errors between the experimental data
and the kinetic model. For each hexose, the datasets obtained at
the three different acid concentrations were solved simultaneously.
Error minimization was performed using the MATLAB toolbox
lsqnonlin.
[14] K. Seri, Y. Inoue, H. Ishida, Chem. Lett. 2000, 22–23.
[15] G. Yang, E. A. Pidko, E. J. M. Hensen, J. Catal. 2012, 295, 122.
[16] Gaussian X, Revision A.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E.
Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Men-
nucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian,
A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara,
K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O.
Kitao, H. Nakai, T. Vreven, J. J. A. Montgomery, J. E. Peralta, F. Ogliaro, M.
Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobaya-
shi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J.
Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V.
Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Moro-
kuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dap-
prich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski,
D. J. Fox, Gaussian, Inc., Wallingford CT, 2009.
Acknowledgements
The majority of the work was financed by Avantium Chemicals.
Part of the work was performed with financial support from the
European Community under the seventh framework programme
(Project SPLASH, contractnr. 311956).Support in kinetic calcula-
tions was provided by Henk van de Bovenkamp at the Depart-
ment of Chemical Engineering at the University of Groningen.
[17] V. Barone, M. Cossi, J. Phys. Chem. A 1998, 102, 1995; F. Eckert, A. Klamt,
AIChE J. 2002, 48, 369.
[18] L. Que, Jr., G. R. Gray, Biochemistry 1974, 13, 146–153.
[19] P. W. Wertz, J. C. Garver, L. Anderson, J. Am. Chem. Soc. 1981, 103,
3916–3922.
[20] J. C. Speck, Jr., Adv. Carbohydr. Chem. 1958, 13, 63–103.
Keywords: 5-hydroxymethylfurfural
ketose · sugar dehydration
· aldose · biomass ·
Received: April 16, 2013
[1] a) A. S. Dias, S. Lima, M. Pillinger, A. A. Valente in Ideas in Chemistry and
Molecular Sciences: Advances in Synthetic Chemistry (Ed.: B. Pignataro),
Revised: July 5, 2013
Published online on August 23, 2013
Part of a Special Issue on ”CatchBio“. To view the complete issue, visit:
http://onlinelibrary.wiley.com/doi/10.1002/cssc.v6.9/issuetoc
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 1681 – 1687 1687