Fructose dehydration to 5-hydroxymethylfurfural over solid acid catalysts in a biphasic system

Article abstract of DOI:10.1002/cssc.201200072

Different acidic heterogeneous catalysts like alumina, aluminosilicate, zirconium phosphate, niobic acid, ion-exchange resin Amberlyst-15, and zeolite MOR have been studied in fructose dehydration to 5-hydroxymethylfurfural (HMF). The acidity of these materials was characterized using temperature-programmed desorption of NH₃ and IR spectroscopy of adsorbed pyridine. The nature and strength of acid sites was shown to play a crucial role in the selectivity towards HMF. Bronsted acid sites in the case of zeolites and ion-exchange resin led to high selectivities in the dehydration of fructose with an increase in selectivity with the addition of an organic phase. Lewis acidity in the case of phosphate and oxides resulted in the intensive production of humins from fructose at the initial stages of the process, whereas organic phase addition did not affect selectivity. Copyright

Full text of DOI:10.1002/cssc.201200072

DOI: 10.1002/cssc.201200072
Fructose Dehydration to 5-Hydroxymethylfurfural over
Solid Acid Catalysts in a Biphasic System
Vitaly V. Ordomsky, John van der Schaaf, Jaap C. Schouten, and T. Alexander Nijhuis*[a]
Different acidic heterogeneous catalysts like alumina, alumino-
silicate, zirconium phosphate, niobic acid, ion-exchange resin
Amberlyst-15, and zeolite MOR have been studied in fructose
dehydration to 5-hydroxymethylfurfural (HMF). The acidity of
these materials was characterized using temperature-pro-
grammed desorption of NH3 and IR spectroscopy of adsorbed
pyridine. The nature and strength of acid sites was shown to
play a crucial role in the selectivity towards HMF. Brønsted acid
sites in the case of zeolites and ion-exchange resin led to high
selectivities in the dehydration of fructose with an increase in
selectivity with the addition of an organic phase. Lewis acidity
in the case of phosphate and oxides resulted in the intensive
production of humins from fructose at the initial stages of
the process, whereas organic phase addition did not affect
selectivity.
Introduction
Over the last few years the dehydration of carbohydrates to 5-
hydroxymethylfurfural (HMF) has attracted increasing attention
due to its possible application as a substitute for petroleum-
based building blocks.[1] HMF and its derivatives can be applied
as platform chemicals, precursors for polymers, fuels, or sol-
vents.[2]
water/THF by the group of Davis.[19] However, explanations re-
garding the effect of organic phase addition to catalytic sys-
tems were not presented.
In our previous study the effect of organic phase addition
during fructose dehydration over zeolites was due to the pres-
ence of strong Brønsted acid sites interacting with MIBK.[20]
Herein, we present the results of a comprehensive study of dif-
ferent acidic heterogeneous systems in the dehydration of
fructose with and without the presence of an organic phase.
Different heterogeneous and homogeneous catalysts have
been applied for this reaction, such as mineral acids,[3,4] ion ex-
change resins,[5–7] oxides,[8,9] phosphates,[10,11] and zeolites.[12,13]
It was shown that water as a reaction medium promotes the
rehydration of HMF resulting in the formation of acids and
polymeric products.[14] Dumesic and Roman-Leshkov[15] investi-
gated the organic phase effect on the dehydration of fructose
in a biphasic system. Addition of solvents like 2-butanol or
methyl isobutyl ketone (MIBK), which extracted HMF from the
reaction mixture, was shown to increase HMF selectivity. It was
reported that 80% selectivity could be achieved over HCl in
the presence of an organic phase.[15]
Results and Discussion
Physicochemical properties of catalysts
The acidity of solid catalysts was characterized by IR analysis of
adsorbed pyridine (Py) and temperature-programmed desorp-
tion (TPD) of NH3. IR spectra recorded before and after the ad-
sorption of Py in the region of OH groups and the Py region
are presented in Figure 1.
Zirconium phosphate exhibits an intense band at 3665 cmÀ1
and a weak band at 3760 cmÀ1, which is attributed to POÀH
and ZrOÀH vibrations.[22] Py adsorption results in the disap-
pearance of the 3760 cmÀ1 band and a significant decrease of
the band at 3665 cmÀ1. Chemically adsorbed Py is usually re-
vealed by bands at 1545 and 1636 cmÀ1 assigned to the pyridi-
nium ion (PyH+), two bands at 1454 and 1622 cmÀ1 related to
coordinately adsorbed Py, and the superposition of signals of
species adsorbed on Lewis and Brønsted acid sites at
Heterogeneous catalysts offer several advantages over dis-
solved homogeneous acid catalysts, (e.g., easy separation of
product, reusability of catalyst, and no corrosion of equip-
ment), which makes them more suitable for industrial applica-
tions. Currently, several heterogeneous catalytic systems are
described in the literature for sugar dehydration with the addi-
tion of an organic phase. Moreau et al. studied the dehydration
of fructose in the presence of zeolites at 438 K in MIBK/water
(5:1 v/v).[12] The highest selectivity (90%) is observed over de-
aluminated mordenite and is attributed to its 2D structure and
absence of cavities.[13] Chheda and Dumesic[16] showed that
good yields of HMF from fructose were achieved over Dia-
ion PK216 in MIBK/water by modification of the aqueous
phase with DMSO or N-methyl-2-pyrrolidone (NMP). Yang et al.
studied sugar dehydration in the biphasic system water/buta-
nol over niobium and tantalum hydroxides.[17,18] Synthesis of
HMF from carbohydrates over tin-beta zeolite was reported in
[a] Dr. V. V. Ordomsky, Dr. J. van der Schaaf, Prof. Dr. J. C. Schouten,
Dr. T. A. Nijhuis
Laboratory of Chemical Reactor Engineering
Eindhoven University of Technology
P.O. Box 513, 5600 MB Eindhoven (The Netherlands)
Fax: (+31)40-2446653
E-mail: t.a.nijhuis@tue.nl
ChemSusChem 2012, 5, 1 – 9
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
ÞÞ
These are not the final page numbers!

T. A. Nijhuis et al.
Brønsted acidity. An intense band at 1622 cmÀ1 and
a broad band at 1450 cmÀ1 indicate Lewis acid sites
with different acid strengths on the surface of the
catalyst.[27]
The spectrum of activated amorphous aluminosili-
cate (Figure 1) shows an intense line of SiOH groups
at 3740 cmÀ1 and a broad band of hydrogen-bonded
hydroxyl groups at 3600 cmÀ1. However, there is no
band attributed to isolated bridging Al(OH)Si groups.
Py adsorption leads to a small decrease of the band
of silanol groups with the appearance of the weak
band of PyH+. This was explained earlier by an induc-
tive effect of the Al atom in the structure of amor-
phous aluminosilicate leading to an increase in the
acidity of silanol groups.[28] Lewis acidity in the
sample stems from octahedral and tetrahedral Al3+
sites.
Mordenite (MOR) zeolite demonstrates five differ-
ent bands of hydroxyl group vibrations (Figure 1).
Bands at 3745 and 3730 cmÀ1 correspond to external
and internal silanol groups, respectively.[29] The band
at 3606 cmÀ1 is due to bridging hydroxyl groups.
Figure 1. FTIR spectra of dehydrated catalysts observed before and after adsorption of
pyridine.
1491 cmÀ1. Scheme 1 depicts the presence of a reasonable
amount of geminal P(OH)2 groups which are responsible for
Brønsted acidity in the sample.[23]
A broad shoulder at about 3690 cmÀ1 is usually ascribed to OH
groups attached to extraframework Al species.[30] A very broad
band at about 3510 cmÀ1 is assigned to SiOH groups interact-
ing through hydrogen bonds with bridging hydroxyl groups or
other silanol groups.[30] Adsorption of Py results in the disap-
pearance of the band of bridging hydroxyl groups and the ap-
pearance of the intense bands assigned to the pyridinium ion.
Lewis acidity in MOR zeolite appears due to the partial destruc-
tion of the zeolite structure.
Zirconium atoms in the structure
of zirconium phosphate form
Lewis acid sites for Py adsorption
(Figure 1).
Free niobic acid has the compo-
Scheme 1. Acid sites of zirco-
sition H8Nb6O19 with eight protons
above eight faces of the octahe-
dron formed by six Nb atoms.[24]
nium phosphate.
The concentration of Brønsted (B) and Lewis (L) acid sites
are calculated based on published extinction coefficients for
Heating niobic acid at 573 K leads to partial dehydration of the
catalyst with the removal of the intense bands at 3400 and
1600 cmÀ1 assigned to water
the bands at 1545 and 1454 cmÀ1 [31]
.
The results given in
Table 1 suggest that the contribution of Brønsted acidity de-
molecules (Figure 1).[24] After
Table 1. Properties of catalysts.
evacuation at 573 K niobic acid
gives only weak band at
a
Sample
Surface area
Pore
size [ꢁ]
TPD NH3
Py adsorption
Adsorption
[mmolgÀ1
3710 cmÀ1, which might be as-
signed to surface hydroxyl
groups on partially decomposed
niobic acid.[25] Py adsorption
leads to neutralization of the
band at 3710 cmÀ1 with the ap-
pearance of the broad band of
the pyridinium ion at 1545 cmÀ1
and an intense band at
1454 cmÀ1. Tanabe[24] showed
[m2 gÀ1
]
[mmolgÀ1
]
[mmolgÀ1
]
]
[a]
B
L
L/B
Fructose
HMF
SiO2–Al2O3
Al2O3
ZrPO4
327
262
93
95
98
85
80
5–7.5
300[b]
12
72
111
242
1100
–
28
–
45
11
56
135
92
27
42
–
2
–
2
2.4
0.2
–
0.052
0.44
0.17
0.58
0.008
0.013
0.100
0.012
0.010
0.005
0.05
Nb2O5
MOR
Amberlyst-15
108
420
53[b]
229
4700[b]
0.19
[a] B: Brønsted acidity, L: Lewis acidity. [b] Parameters are taken from the technical data of the product.
that the Brønsted and Lewis acid bands turn into each other
during dehydration or saturation of the sample with water.[24]
Dehydration of alumina leads to the appearance of the
usual set of bands at 3770, 3752, 3730, and 3675 cmÀ1
(Figure 1). According to the literature these hydroxyl bands are
associated with hydroxyl groups of Al atoms with different co-
ordination numbers.[26] Py adsorption does not significantly
change the intensity of hydroxyl peaks due to their weak
creases in the order: MOR>ZrPO4 ꢀSiO2–Al2O3 >Nb2O5 >Al2O3.
Amberlyst-15 should have the highest contribution of Brønsted
acidity due to the absence of Lewis acid sites in sulfated poly-
styrene.
The strength of acid sites might be determined by analysis
of spectra of NH3 TPD (Figure 2). MOR zeolite has the highest
strength of acid sites with desorption of NH3 up to 1000 K.
Results of Py adsorption indicate that desorption proceeds
&2
&
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
ÝÝ
These are not the final page numbers!

Fructose Dehydration over Solid Acid Catalysts
Figure 2. TPD NH3 profiles for catalysts.
Figure 3. a) Fructose conversion versus time; b) selectivity to HMF versus
fructose conversion over catalysts. Reaction conditions: catalyst (4 g), fruc-
tose (20 g), water (300 mL), 408 K.
from the Brønsted acid sites. The amount of desorbed NH3
(Table 1) and the strength of acid sites (Figure 2) decrease for
oxides and phosphate in the order: Nb2O5 >ZrPO4 >Al2O3 >
SiO2–Al2O3. This order corresponds to the decrease of electro-
negativity of the metals and of the acidity of the oxides.[32] The
strength of Brønsted and Lewis acidity should simultaneously
decrease in this order.
The chosen catalysts possess different acidic properties,
a fact which allows us to study the influence of acidity on fruc-
tose dehydration.
Dehydration of fructose
Figure 3 and 4 show the conversion of fructose and the selec-
tivity for HMF and acids versus fructose conversion during the
dehydration of fructose in an aqueous solution over different
heterogeneous catalysts. The only products detected in all ex-
periments were formic acid, levulinic acid, and furfural. The se-
lectivity towards furfural did not exceed 1 mol% and therefore
we did not investigate its formation further. The appearance of
furfural was explained previously by a fast reverse aldol cleav-
age of carbohydrates.[33] Formic and levulinic acids were the
products of HMF rehydration over acidic catalysts.[34] The main
byproducts in the reaction were polymeric humins, which
were not observed in HPLC analyses.
Figure 4. Selectivity for acids versus fructose conversion over catalysts. Reac-
tion conditions: catalyst (4 g), fructose (20 g), water (300 mL), 408 K.
conversion, in spite of the high amount of acid sites (Figure 1,
Table 1). This fact may be explained by serious diffusion limita-
tions for fructose inside the zeolite’s narrow pores (Table 1).
Zeolite activity was high at 438 K.[12] All other catalytic systems
have pores larger than 8 nm.
Catalytic activity in the dehydration of fructose (over hetero-
geneous catalysts) at 408 K decreases in the order: Nb2O5 >
ZrPO4 >Amberlyst-15>Al2O3 >SiO2–Al2O3 >MOR. The high ac-
tivity over catalysts having mainly Lewis acid sites (Figure 1) in-
dicates that the reaction proceeds over Brønsted and Lewis
acid sites of these heterogeneous catalysts. Notably, activity in
the transformation of fructose correlates with the strength of
the Lewis acid sites of the heterogeneous catalysts as follows:
Nb2O5 >ZrPO4 >Al2O3 >SiO2–Al2O3 (Figure 2). The activity of
Amberlyst-15 is comparable with the activity of Lewis acid cat-
alysts, whereas MOR zeolite demonstrates the lowest fructose
HMF selectivity (about 60%) is highest for catalysts with
Brønsted acid sites such as Amberlyst-15 and MOR zeolite.
Catalytic systems containing both Lewis and Brønsted acid
sites like aluminosilicate, niobic acid, and zirconium phosphate
show selectivity in the range 25–30%. The lowest selectivity
(2–17%) was observed over alumina. Results indicate a de-
crease in selectivity towards HMF with an increase in Lewis
acidity contribution and high selectivity over catalysts with
Brønsted acidity. The strength of Brønsted acid sites also plays
an important role in the dehydration reaction. Thus, the use of
amorphous aluminosilicate with weak Brønsted acidity does
ChemSusChem 0000, 00, 1 – 9
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
&
3
&
ÞÞ
These are not the final page numbers!

T. A. Nijhuis et al.
not lead to high selectivities for HMF (Figure 3). Testing of zeo-
lites with different strengths of acid sites also showed that the
highest selectivity was observed over MOR zeolite with acid
sites of the highest strength.[20] Those acid sites which are able
to interact with Py influence HMF selectivity. Figure 5 shows
showed similar results (low selectivity at the beginning of the
reaction).[32] This was explained by a xylose reaction with furfu-
ral, which gave humins. Accordingly, fructose might undergo
a dehydration over an acid site to form HMF (Figure 6, reac-
tion 2), or alternatively undergo condensation with HMF or
Figure 6. Fructose dehydration reaction scheme.
fructose to form humins (Figure 6, reaction 1). HMF could also
desorb from the acid site and be extracted in an organic sol-
vent (in a biphasic system) (Figure 6, reaction 3) or it may react
with itself to form humins, or rehydrate into acids (Figure 6, re-
action 4). To investigate the conversion of fructose into by-
products, the HMF reaction was studied over different
catalysts.
Figure 5. Selectivity towards HMF at fructose conversion of 10% versus
Brønsted/Lewis ratio determined by Py adsorption.
a strong correlation between selectivity for HMF at the same
levels of fructose conversion and Brønsted/Lewis ratios (deter-
mined by Py adsorption).
Formation of formic and levulinic acids as byproducts was
observed mainly over Amberlyst-15 and MOR zeolite (Figure 4).
In those reactions, the selectivity for acids increased with an in-
crease in fructose conversion. Catalysts with a high contribu-
tion of Lewis acidity formed small amounts of acids. Thus, fruc-
tose mainly transformed into humins over those catalysts. The
important role of Lewis acid sites in the conversion of xylose
into humins was also shown by Weingarten et al.[35]
HMF transformation
Consumption of HMF over Amberlyst-15 is much faster relative
to niobic acid and alumina. Conversion of HMF over both
these catalysts was not higher than 10% in 6 h. The main
products over Amberlyst-15 were formic and levulinic acid
with a selectivity of about 80%. Alumina and niobic acid
showed a selectivity of 50 and 30%, respectively (Figure 7).
These results indicated that catalysts with Lewis acid sites were
not highly efficient. The low adsorption of HMF on these cata-
lysts correlated with catalytic results (Table 1).
The selectivity of all studied heterogeneous catalysts is low
at the beginning of the reaction with a subsequent increase to
17–60%. The low initial selectivity might be explained by
a strong adsorption of HMF on the surface of the catalyst at
the beginning of the reaction. A study of HMF adsorption over
the catalysts (Table 1) shows that significant adsorption of HMF
is observed only in the case of aluminosilicate and Amberlyst-
15. It indicates an important role of Brønsted acid sites in the
interaction with HMF. Brønsted acid sites should strongly inter-
act with the carbonyl group of HMF.[36] Low adsorption of HMF
over MOR zeolite may be the result of diffusion limitations.
The presence of strong Lewis acidity in the case of zirconium
phosphate, niobic acid, and alumina does not lead to HMF ad-
sorption on the surface of the catalyst. As a result, aluminosili-
cate and Amberlyst-15 show an initial increase in selectivity for
HMF at low conversions due to a gradual desorption of HMF
from the surface. The low linear increase in HMF selectivity
with time, in the case of other catalysts, can be explained by
the deactivation of strong Lewis acid sites with the increase in
selectivity for HMF (Figure 3).
Thus, low selectivity during fructose transformation over alu-
mina, aluminosilicate, niobic acid, and zirconium phosphate
cannot be explained by secondary HMF reactions (Figure 3).
A more plausible explanation is the fast oligomerization of
fructose over the surface of these catalysts without an inter-
mediate desorption of HMF. Fructose adsorption experiments
support this hypothesis. Catalysts with strong Lewis acid sites
like niobic acid and alumina show a strong adsorption of fruc-
tose. Conversely, Amberlyst-15 shows significant adsorption of
HMF, but not fructose. This difference might be explained by
the presence of an aromatic ring on HMF. Interaction of the
carbonyl group on HMF with a Lewis acid site leads to migra-
tion of the positive charge to the aromatic ring with repulsion
of HMF from a neighboring Lewis acid site. Lewis acidity is
more effective in the adsorption of fructose by multiple surface
interactions of carbonyl and hydroxyl groups with Lewis acid
sites. This results in surface condensation of fructose molecules
and in turn the formation of humins.
Fructose dehydration characteristics indicate a parallel trans-
formation of fructose over the catalyst to polymeric byprod-
ucts or a fast secondary reaction of adsorbed HMF. An earlier
study of xylose dehydration using heterogeneous catalysts
Although Amberlyst-15 shows a high activity in the secon-
dary transformation of HMF with a high selectivity for acids
&4
&
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
ÝÝ
These are not the final page numbers!

Fructose Dehydration over Solid Acid Catalysts
Figure 7. a) HMF conversion; b) selectivity towards acids over Al2O2, Nb2O5,
and Amberlyst-15. MIBK was added (MIBK/water=3:1 v/v in the case of Am-
berlyst-15) after 5 h. Reaction conditions: catalyst (4 g), fructose (20 g), water
(300 mL), 408 K.
Figure 8. a) Influence of addition of MIBK on fructose conversion versus
time; b) selectivity for HMF versus fructose conversion over HCl. Reaction
conditions: HCl (1.6 g, 36%), fructose (20 g), water (300 mL), MIBK (900 mL),
~
~
: MIBK/water=3.
&
&
408 K.
,
: water;
,
(Figure 7), the dehydration of fructose proceeds with the for-
mation of HMF in a maximum yield of 60%, with a further de-
crease in HMF selectivity due to its secondary reaction. This in-
dicates that direct fructose or intermediate product conversion
into humins (Figure 6, reaction 1) might also take place over
catalysts with Brønsted acid sites. A comparison with fructose
dehydration over homogeneous HCl supports this hypothesis
(Figure 8). The selectivity for HMF over HCl at low conversions
is 75% with the formation of 5% acids. During the reaction it
decreases to 45% with an increase in selectivity for acids to
40%. Thus, acids are the main products of the HMF secondary
reaction over HCl. This means that 80% is the maximum HMF
selectivity over Brønsted acid sites before the secondary con-
sumption of HMF. The lower selectivity of HMF formation over
Amberyst-15 is the result of partial direct formation of humins
(Figure 6, reaction 1).
The effect of MIBK addition to the aqueous solution of fruc-
tose with niobic acid as a heterogeneous catalyst is shown in
Figure 9. Results suggest that organic phase addition does not
result in any significant changes in the conversion of fructose
or selectivity for HMF. Similarly, solvent addition had no signifi-
Effect of organic solvent addition
The addition of a water-immiscible solvent may suppress sec-
ondary transformation of HMF.[15] Indeed, Figure 8 shows that
the addition of excess MIBK (MIBK/water=3:1 v/v) to an aque-
ous solution of HCl leads to a stabilization of selectivity for
HMF at 75%. The addition of MIBK does not lead to a signifi-
cant change in catalytic activity because HCl remains in the
aqueous phase. The amount of HMF extracted by MIBK ex-
ceeds the amount of HMF in water more than three-fold,
therefore significantly inhibiting HMF rehydration over HCl.
The decrease in HMF selectivity only starts at higher conver-
sions of fructose (Figure 8) and agrees with the initial thought
that the addition of organic solvent would decrease the rate of
HMF transformation by extraction of HMF by the solvent.[15]
Figure 9. a) Influence of addition of MIBK on fructose conversion versus
time; b) selectivity for HMF versus fructose conversion over Nb2O5. Reaction
conditions: catalyst (4 g), fructose (20 g), water (300 mL), MIBK (900 mL),
&
&
408 K. : water; : MIBK/water=3.
ChemSusChem 0000, 00, 1 – 9
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
&
5
&
ÞÞ
These are not the final page numbers!

T. A. Nijhuis et al.
cant effect for zirconium phosphate, alumina, and amorphous
aluminosilicate catalysts (not shown). This agrees with the pre-
vious assumption regarding the transformation of fructose on
the surface of the catalyst directly to humins without inter-
mediate HMF formation (Figure 6). The addition of MIBK leads
just to a redistribution of HMF between the aqueous and the
organic phase without influencing selectivity for HMF
(Figure 9). The partition ratio for HMF between water and MIBK
at the reaction temperature is about 0.9. This means that at
this MIBK/water ratio (3:1 v/v), the amount of HMF present in
the organic phase is 2.7 times higher than the amount in
water. The preservation of the activity of the catalysts after the
addition of MIBK means that the acid sites of the catalyst
during the reaction do not have contact with the organic
phase (Figure 9). An aqueous layer coats the acidic oxides and
zirconium phosphate even when the particles are in the organ-
ic solvent.
the curve in the initial time period. It suggests an absence of
any interaction of the catalyst with MIBK as it was observed in
the case of zeolites.[17] Thus, the enhanced selectivity in the
presence of MIBK for Amberlyst-15 might only be explained by
the extraction of HMF by water-immiscible solvent.
HMF transformation over this resin shows that the addition
of an organic phase leads to a suppression of HMF conversion
(Figure 7). The main products of HMF transformation over Am-
berlyst-15 are formic and levulinic acid (Figure 7), and addition
of MIBK during fructose dehydration results in a decrease in
the rate of acid formation (Figure 10). The selectivity for acids
without MIBK is 13%, whereas after addition of MIBK (MIBK/
water=5:1 v/v) only 2% of converted fructose is in the form
of acids. However, it is difficult to explain the effect of organic
solvent addition only by the suppression of secondary HMF
consumption. HMF selectivity in this case should not decrease
(remaining at around 60%) similarly to the HCl reaction
(Figure 8). An increase in selectivity is due to the decrease in
the amount of humins formed after the addition of MIBK.
There are several possible reasons for this: the formation of
humins might be accompanied by its depolymerization over
acid sites. In this case, the addition of organic solvent at a high
conversion of fructose should lead to a fast increase in selectiv-
ity due to the shift of equilibrium in the direction of humin de-
polymerization. However, no increase in selectivity was ob-
served in this experiment (not shown). The other possible ex-
planation might be a decrease in the rate of formation of
humins directly from fructose and HMF (Figure 6, reaction 1).
To check this possible reaction pathway, the dehydration of
fructose over Amberlyst-15 was studied in the presence of
HMF. Figure 11 shows a comparison of fructose conversion and
Figure 10 shows the effect of MIBK addition during fructose
dehydration over Amberlyst-15. Relative to a catalyst contain-
Figure 10. a) Influence of MIBK/water ratio (v/v) on fructose conversion
versus time; b) selectivity for HMF versus fructose conversion over Amber-
lyst-15. Reaction conditions: catalyst (4 g), fructose (20 g), water (300 mL),
~
&
&
*
408 K. : water; : MIBK/water=1; : MIBK/water=3; : MIBK/water=5.
ing Lewis acid sites, the addition of a water-immiscible solvent
causes an increase in selectivity for HMF. The higher the
amount of added MIBK, the higher the selectivity. Addition of
MIBK to water (1:1 v/v) results in an increase in selectivity for
HMF on the order of 53–66% with fructose conversion at 35%.
At a MIBK/water ratio of 3:1 v/v, the selectivity is already 74%,
but a further increase in the amount of organic phase does
not significantly influence selectivity (77% at MIBK/water=5:1
v/v). Furthermore, the addition of MIBK does not lead to any
considerable change in the conversion of fructose. A high
amount of added organic solvent results only in a decrease of
Figure 11. a) Comparison of fructose conversion versus time; b) selectivity
for humins versus fructose conversion over Amberlyst-15 with fructose ex-
clusively and with addition of HMF. Reaction conditions: catalyst (4 g), fruc-
&
&
tose (20 g), HMF (2.5 g), water (300 mL), 408 K. : fructose; : fructo-
se+HMF.
&6
&
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
ÝÝ
These are not the final page numbers!

Fructose Dehydration over Solid Acid Catalysts
humin selectivity for a standard experiment and one in which
HMF (20 mmol) is added besides fructose. The presence of
HMF in the reaction mixture leads to a significant increase in
fructose conversion (Figure 11). This might be due to a reaction
with HMF. The decrease in fructose conversion after addition
of MIBK might be explained by a decrease in the rate of the
fructose reaction with HMF towards humins due to HMF ex-
traction (Figure 10). The selectivity for humins due to the pri-
mary fructose transformation (Figure 6, reaction 1) was calcu-
lated by taking into account the difference between the
amount of fructose and HMF before and after the reaction by
subtraction of formed acids multiplied by a coefficient of 1.25.
This coefficient reflects the formation of humins in addition to
acids in the secondary transformation of HMF (Figure 7).
As a result, Figure 11 shows the selectivity towards humins
from the reaction of fructose with other fructose molecules or
HMF. The selectivity for humins without addition of HMF slight-
ly decreases from 40 to 30% with an increase in fructose con-
version. Addition of HMF leads to an increase in humin selec-
tivity to 70%. Thus, the results show that fructose condensa-
tion with HMF takes place during fructose dehydration over
Amberlyst-15. Addition of MIBK leads to a reaction-rate de-
crease by HMF extraction.
Experimental Section
Catalysts
g-Alumina (Al2O3) was purchased from Aldrich. Amorphous alumi-
nosilicate (SiO2–Al2O3) was obtained from BASF with a Si/Al ratio of
11:1. MOR Zeolite with a Si/Al ratio of 12:1 was purchased from
Zeolyst. Niobic acid (Nb2O5) was obtained from CBMM (Brazil).
Amberlyst-15 (wet) was purchased from Rohm and Haas. Zirconi-
um phosphate (ZrPO4) was prepared in a similar manner as report-
ed in the literature[21] by precipitation of ZrOCl2·8H2O and H3PO4 at
a P/Zr molar ratio equal to 2:1. The precipitate was filtered,
washed with water, and dried at 373 K. The catalyst was calcined
at 673 K in air. All catalysts were dried at 373 K and pulverized
before reaction.
Characterization
Acidic properties were studied by means of NH3 TPD using an AU-
TOCHEM II (Micromeritics). Prior to adsorption, samples were cal-
cined in situ in a flow of dry air at 573 K for 1 h, and subsequently
in a flow of dry He for 1 h, and cooled to ambient temperature.
For NH3 adsorption, a sample was subjected to a flow of diluted
NH3 for 30 min at 373 K. Physisorbed NH3 was removed in a flow
of dry He at 373 K for 1 h. Typical TPD experiments were carried
out in the temperature range 373–1100 K in a flow of dry He. The
Extraction of HMF by MIBK in the reaction medium results in
an increase in HMF selectivity from 60 to almost 80% at a fruc-
tose conversion of 20% (Figure 10). Thus, approximately 20%
of humins are formed by fructose condensation, which is
equivalent to MIBK addition. At the same time, the selectivity
towards humins during fructose dehydration is 30%, which
means that about 10% of humins are produced by fructose
condensation with HMF. Addition of MIBK leads to the sup-
pression of this reaction.
rate of heating was 9 KminÀ1
.
IR spectra were recorded with a Nicolet Protꢂgꢂ 460 FTIR spec-
trometer at an optical resolution of 4 cmÀ1. Prior to measurements,
catalysts were pressed in self-supporting discs and activated in the
IR cell attached to a vacuum line at 573 K for 4 h. Adsorption of Py
was performed at 423 K for 30 min. Excess Py was further evacuat-
ed at 423 K for 1 h. Adsorption–evacuation was repeated several
times until no changes in spectra were observed.
Adsorption of HMF or fructose over the catalysts was studied by
addition of catalyst (0.5 g) to a solution of HMF or fructose
(3 mmol) in water (25 mL). The mixture was stirred for 1 h at 298 K.
The amount of adsorbed HMF or fructose was determined chroma-
tographically by determining the concentration of fructose remain-
ing in water.
Conclusions
A series of well-characterized acid catalysts were tested in the
aqueous-phase dehydration of fructose to HMF. The concentra-
tion of Brønsted and Lewis acid sites was determined by Py ad-
sorption. The contribution of Brønsted acidity of the samples
decreases in the order: Amberlyst-15>MOR>ZrPO4 >SiO2–
Al2O3 >Nb2O5 >Al2O3.
Catalysis
Experiments were carried out in a 2 L stirred autoclave working in
a batch mode and equipped with two valves for sampling liquid
from the aqueous and organic phases. The procedure for testing
the catalysts was as follows: heterogeneous catalyst (4 g) or HCl
(1.6 g of 36%) and water (250 mL) were poured into the autoclave.
MIBK was added to the autoclave for experiments with organic sol-
vents. The autoclave was purged with nitrogen. Fructose (20 g,
0.37m) or HMF (5 g, 0.13m) or fructose with HMF (20 g fructose,
2.5 g HMF) dissolved in 50 mL of water was poured into the auto-
clave after the temperature had been increased to 408 K, at which
point the catalytic experiment was started. The agitation speed
was 500 rpm.
Periodically liquid samples were taken from the autoclave, which
were analyzed using HPLC (Shimadzu) equipped with refractive
index and UV-Vis detectors with a BIO-RAD Aminex HPX-87H
column.
Reactant conversion (mol%) and product selectivity (mol%) were
defined as follows:
The HMF selectivity correlates with the contribution of
Brønsted acid sites with the highest selectivity over Amber-
slyst-15 and MOR. Lewis acidity is responsible for the decrease
in HMF selectivity due to the fast initial condensation of fruc-
tose into humins over Lewis acid sites.
The addition of water-immiscible MIBK to the reaction mix-
ture with oxides and phosphates does not have any effect on
selectivity. Amberlyst-15 has an improved selectivity for HMF
with the addition of organic solvents. This effect was ascribed
to extraction of HMF by MIBK with suppression of the rehydra-
tion reaction of formed HMF and primary fructose condensa-
tion with HMF.
Conversion (mol%)=(moles of fructose reacted)/(moles of initial
fructose)ꢃ100%
ChemSusChem 0000, 00, 1 – 9
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
&
7
&
ÞÞ
These are not the final page numbers!

T. A. Nijhuis et al.
[13] C. Moreau, R. Durand, S. Razigade, J. Duhamet, P. Faugeras, P. Rivalier, P.
Ross, G. Avignon, Appl. Catal. A 1996, 145, 211.
Selectivity (%)=(moles of HMF or formic acid produced)/(moles of
fructose reacted)ꢃ100%.
[14] C. Moreau, Agro Food Ind. Hi-Tech 2002, 13, 17.
[15] Y. Romꢄn-Leshkov, J. A. Dumesic, Top. Catal. 2009, 52, 297.
[16] J. N. Chheda, J. A. Dumesic, Catal. Today 2007, 123, 59.
[17] F. Yang, Q. Liu, X. Bai, Y. Du, Bioresour. Technol. 2011, 102, 3424.
[18] F. Yang, Q. Liu, M. Yue, X. Bai, Y. Du, Chem. Commun. 2011, 47, 4469.
[19] E. Nikolla, Y. Romꢄn-Leshkov, M. Moliner, M. E. Davis, ACS Catal. 2011, 1,
408.
Selectivity for humins in the experiment with HMF addition was
calculated as follows:
[(moles of HMF+moles of fructose)before reactionÀ(moles of HMF+
moles of fructose)after reactionÀformic acid producedꢃ1.25]/(moles of
fructose reacted)ꢃ100%.
[20] V. Ordomsky, J. van der Schaaf, J. C. Schouten, T. A. Nijhuis, J. Catal.
2012, 287, 68.
[21] Y. Kamiya, S. Sakata, Y. Yoshinaga, R. Ohnishi, T. Okuhara, Catal. Lett.
2004, 94, 45.
[22] M. Ziyad, M. Rouimi, J.-L. Portefaix, Appl. Catal. A 1999, 183, 93.
[23] A. Corma, Chem. Rev. 1995, 95, 559.
[24] K. Tanabe, Mater. Chem. Phys. 1987, 17, 217.
[25] I. Ahmad, T. J. Dines, J. A. Anderson, C. H. Rochester, Spectrochim. Acta,
Part A 1999, 55, 397.
[26] C. Morterra, G. Magnacca, Catal. Today 1996, 27, 497.
[27] X. Liu, J. Phys. Chem. C 2008, 112, 5066.
Acknowledgements
This research has been performed within the framework of the
CatchBio program (053.70.001). The authors gratefully acknowl-
edge support through the Smart Mix Program of the Netherlands
Ministry of Economic Affairs and the Netherlands Ministry of Edu-
cation, Culture and Science.
[28] V. L. Zholobenko, D. Plant, A. J. Evans, S. M. Holmes, Microporous Meso-
porous Mater. 2001, 44–45, 793.
[29] N. S. Nesterenko, F. Thibault-Starzyk, V. Montouillout, V. V. Yuschenko, C.
Fernandez, J.-P. Gilson, F. Fajula, I. I. Ivanova, Microporous Mesoporous
Mater. 2004, 71, 157.
Keywords: acidity
catalysis · renewable resources · spectroscopy
·
biphasic catalysis
·
heterogeneous
[1] X. Tong, Y. Ma, Y. Li, Appl. Catal. A 2010, 385, 1.
[2] A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411.
[3] B. F. M. Kuster, Starch/Staerke 1990, 42, 314.
[4] K. Hamada, H. Yoshihara, G. Suzukamo, Chem. Lett. 1982, 617.
[5] T. El Hajj, A. Masroua, J.-C. Martin, G. Descotes, Bull. Soc. Chim. Fr. 1987,
85.
[6] P. Vinke, H. van Bekkum, Starch/Staerke 1992, 44, 90.
[7] L. Rigal, A. Gaset, Biomass 1985, 8, 267.
[8] M. Watanabe, Y. Aizawa, T. Iida, R. Nishimura, H. Inomata, Appl. Catal. A
2005, 295, 150.
[30] M. Maache, A. Janin, J. C. Lavalley, E. Benazzi, Zeolites 1995, 15, 507.
[31] C. A. Emeis, J. Catal. 1993, 141, 347.
[32] N. C. Jeong, J. S. Lee, E. L. Tae, Y. J. Lee, K. B. Yoon, Angew. Chem. 2008,
120, 10282; Angew. Chem. Int. Ed. 2008, 47, 10128.
[33] D. A. Nelson, P. M. Molton, J. A. Russell, R. T. Hallon, Ind. Eng. Chem. Prod.
Res. Dev. 1984, 23, 471.
[34] B. F. M. Kuster, Carbohydr. Res. 1977, 54, 177.
[35] R. Weingarten, G. A. Tompsett, W. C. Conner, Jr., G. W. Huber, J. Catal.
2011, 279, 174.
[9] H. Yan, Y. Yang, D. Tong, X. Xiang, C. Hu, Catal. Commun. 2009, 10,
1558.
[36] H. Fang, A. Zheng, Y. Chu, F. Deng, J. Phys. Chem. C 2010, 114, 12711.
[10] F. Benvenuti, C. Carlini, P. Patrono, A. M. R. Galletti, G. Sbrana, M. A. Mas-
succi, P. Galli, Appl. Catal. A 2000, 193, 147.
[11] P. Carniti, A. Gervasini, S. Biella, A. Auroux, Catal. Today 2006, 118, 373.
[12] C. Moreau, R. Durand, C. Pourcheron, S. Razigade, Ind. Crops Prod. 1994,
3, 85.
Received: January 28, 2012
Published online on && &&, 0000
&8
&
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 9
ÝÝ
These are not the final page numbers!

FULL PAPERS
V. V. Ordomsky, J. van der Schaaf,
J. C. Schouten, T. A. Nijhuis*
&& – &&
Fructose Dehydration to 5-
Hydroxymethylfurfural over Solid Acid
Catalysts in a Biphasic System
Fructose dehydration over heteroge-
neous acidic catalysts is studied. Lewis
acidity leads to direct fructose conden-
sation and humins are produced.
Organic solvent addition increases selec-
tivity exclusively over Brønsted acid cat-
alysts due to suppression of the rehy-
dration reaction of HMF and primary
fructose condensation with HMF (see
figure).
Brønsted acidity results in selective 5-
hydroxymethylfurfural (HMF) formation
with subsequent rehydration into acids.
ChemSusChem 0000, 00, 1 – 9
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
&
9
&
ÞÞ
These are not the final page numbers!