Dehydration of fructose to 5-hydroxymethylfurfural in sub-critical water over heterogeneous zirconium phosphate catalysts

Article abstract of DOI:10.1016/j.carres.2006.06.025

Dehydration of fructose to 5-hydroxymethylfurfural in a batch-type process with sub-critical water (sub-CW) was performed in the presence of different laboratory-made zirconium phosphate solid acids at 240 °C. A direct relation was found between increasing the crystallinity and decreasing the surface area of solid acids. However, irrespective of the different surface areas, similar catalytic behaviors were observed. Meanwhile, calcination of the samples showed no improvement in the activity of the solid acids. In the presence of the amorphous form of zirconium phosphate, about 80% of fructose was decomposed in sub-critical water at 240 °C after 120 s, and the selectivity of the dehydration reaction of fructose to 5-hydroxymethylfurfural rose to 61%. No rehydration products were identified. Soluble polymers and furaldehyde were the only major and minor side products found, respectively. Finally it was found that zirconium phosphate solid acids were stable under sub-CW conditions, and they can easily be recovered without changing their catalytic properties.

Full text of DOI:10.1016/j.carres.2006.06.025

Carbohydrate Research 341 (2006) 2379–2387
Dehydration of fructose to 5-hydroxymethylfurfural in
sub-critical water over heterogeneous zirconium phosphate catalysts
Feridoun Salak Asghari and Hiroyuki Yoshida*
Department of Chemical Engineering, College of Engineering, Osaka Prefecture University, 1-1 Gakuen-cho,
Sakai-shi, Osaka 599-8531, Japan
Received 11 May 2006; received in revised form 22 June 2006; accepted 30 June 2006
Available online 25 July 2006
Abstract—Dehydration of fructose to 5-hydroxymethylfurfural in a batch-type process with sub-critical water (sub-CW) was per-
formed in the presence of different laboratory-made zirconium phosphate solid acids at 240 ꢀC. A direct relation was found between
increasing the crystallinity and decreasing the surface area of solid acids. However, irrespective of the different surface areas, similar
catalytic behaviors were observed. Meanwhile, calcination of the samples showed no improvement in the activity of the solid acids.
In the presence of the amorphous form of zirconium phosphate, about 80% of fructose was decomposed in sub-critical water at
2
40 ꢀC after 120 s, and the selectivity of the dehydration reaction of fructose to 5-hydroxymethylfurfural rose to 61%. No rehydra-
tion products were identified. Soluble polymers and furaldehyde were the only major and minor side products found, respectively.
Finally it was found that zirconium phosphate solid acids were stable under sub-CW conditions, and they can easily be recovered
without changing their catalytic properties.
ꢁ
2006 Elsevier Ltd. All rights reserved.
Keywords: Sub-critical water; Solid acid; Zirconium phosphate; Fructose; Dehydration; 5-Hydroxymethylfurfural
1
. Introduction
tion of saccharides in water and/or organic solvents
mixture (Scheme 1). Since D u¨ ll and Kiermeyer indepen-
dently introduced a method of synthesis of HMF in
In recent years, attention has been focused on the
production of various chemical compounds by using
sub- and supercritical water systems as an alternative
technique to conventional methods.
framework, sub-critical water (hereafter called sub-
CW) is used for the conversion of monosaccharides
8
1895, many researchers reported different preparation
methods along with application of homogeneous and
heterogeneous acids both in aqueous and non-aqueous
media. Some examples are the following: mineral
1
,2
Within this
9
–11
12–16
acids
(H SO , HCl, and H PO ), organic acids
2 4 3 4
(
and also oligo- and polysaccharides that can yield hex-
(oxalic, levulinic, p-toluenesulfonic, and maleic acids),
oses on hydrolysis) as major renewable raw materials for
industrial chemicals and related materials. In particular
this process is used for the conversion to the petrochem-
Lewis acids (ZnCl , AlCl , and BF ), salts (NH SO /
2
3
3
3
4
4
ꢀ
SO , pyridine=PO , pyridine/HCl, and also Al, Th,
3
4
1
7–20
6,21–24
Zr, lanthanide ions),
and solid catalysts
(ion-
3
–5
ical-like compound, 5-hydroxymethylfurfural
(here-
exchange resins, zeolites, and metal(IV) phosphates),
which were all commonly used in the production of
HMF.
after called HMF). HMF is a valuable chemical
feedstock that has numerous industrial applications,
6
7
especially in polymer chemistry. Seri and Ishida have
correctly labeled it as a ‘very expensive chemical’. This
compound is synthesized via the acid-catalyzed dehydra-
+
H
Saccharides
HOH2C
O
CHO
-
3H O
2
*
Corresponding author. Tel./fax: +81 254 9298; e-mail: Yoshida@
chemeng.osakafu-u.ac.jp
Scheme 1. Dehydration of saccharides to HMF.
0
008-6215/$ - see front matter ꢁ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2006.06.025

2380
F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
3
3
Application of homogenous acids, however, are typi-
ers studied the comparison of the homogeneous acid
and base (H SO and NaOH) and two metal oxides as
an example of a heterogeneous acid and base (TiO2,
cally associated with the problems of high toxicity, cor-
rosion, catalyst waste, use of large amounts of catalyst,
and the difficulty of separation and recovery. Hetero-
geneous acids, on the other hand, overcome most dis-
advantages of homogeneous acids as they offer the
advantages of easy separation from reaction products,
recycling, and higher selectivities than homogeneous
2
4
ZrO ) for glucose reactions under sub-CW conditions
2
at 200 ꢀC toward isomerization and dehydration of glu-
cose. They observed isomerization in the reaction of glu-
cose by NaOH and ZrO by the basic catalysts, and a
2
dehydration reaction by H SO and TiO by the acidic
2
4
2
2
5
ones.
catalysts.
From another point of view, it is obvious that aque-
ous-phase reactions have interesting advantages: Water
is the cheapest, non-toxic, non-flammable, clean solvent,
which increases the economic feasibility of the process.
However, there are quite a few numbers of solid acids
that are acceptable in terms of their activity, stability,
and insolubility for the reactions in aqueous media and
particularly in water at high temperature and pressure.
In fact most solid acids lose their catalytic activities in
The well-matched properties of ZrP solids as catalysts
with aqueous-phase reactions make these compounds
attractive for use under sub-CW conditions. In the pres-
ent work, we studied the applicability of ZrP solid acids
under sub-CW conditions as alternatives to homo-
geneous acids for dehydration of fructose to HMF.
2. Experimental
2.1. Materials
2
5
aqueous solutions. Zirconium phosphate (hereafter
called ZrP) solid acids possess properties that make them
compatible, not only with the reactions in which water is
generated, but also in all reactions with aqueous media.
Zirconium chloride (Chameleon reagent, 98%) was used
as the zirconium source. All other reagents, phosphoric
acid (85%), HCl (35%), HF (46–48%), MeOH (HPLC
grade), D-fructose (99%), D-glucose (99%), 5-hydroxy-
methylfurfural (99%), furaldehyde (99%), levulinic (95%),
and formic (99%) acids were purchased from Wako Pure
Chemical Industries (Japan). All reagents were used
without purification.
2
6
ZrP solids have been primarily used as ion exchangers.
However, the use of these solids as catalysts in various
reactions such as dehydration, isomerization, polymeri-
zation, and alkylation has received considerable atten-
2
7–29
tion during the last few years,
as attested to by the
numbers of papers and patents that have recently been
published. In the presence of different ZrP solids, the
3
0
dehydration of alcohols, esterification of acetic acid
3
1
32
with ethanol, oxidation of propene, and dehydration
2.2. Solid acid sample preparation
2
8
of cyclohexanol have been widely studied.
For reactions under sub-CW conditions, and particu-
larly for reactions of saccharides, the application of only
a few homogeneous and heterogeneous catalysts has
Schematic preparation procedure is shown in Figure 1.
The catalyst was prepared by modifying the procedure
given by Clearfield and Stynes. The best results were
3
5
3
–5,33
16
been reported.
In previous work, we studied the
obtained as follows: ZrCl (0.5 g, 2.15 mmol) as a zirco-
4
effect of various homogeneous acids in the range
of pH 1.5–5 on dehydration of fructose to HMF under
sub-CW conditions. The application of ZrP solids (or
generally metal(IV) phosphates salts) in the reaction of
saccharides is limited to a few available academic and
nium precursor salt was dissolved in HCl (3 mL, 2 N).
H PO (3 mL, 7 N) was then added while the solution
3
4
was stirred at room temperature. The gel that was
produced was then refluxed in about 10 N H PO until
3
4
2
2
patent reports. Nakamura in a patent used normal
temperature and pressure for the preparation of HMF
from hexoses in DMSO with ZrP catalyst and got a
maximum yield of 81% for HMF after 3 h of reaction
at 140 ꢀC. A catalytic activity equal to a turnover
zirconium chloride
ZrCl4
dissolve in 3 mL of 2 N HCl
ꢀ
1
ꢀ1
frequency (TOF) of 0.23 g g
h
was obtained.
3 4
dropwise excess of H PO
3
4
Benvenuti et al. studied the application of various
zirconium and titanium phosphate salts in dehydration
of fructose and inulin to HMF in aqueous media under
normal temperature and pressure conditions in batch
and continuous systems at 110 ꢀC. They obtained TOF
centrifuge off the gel and wash until
free of phosphate and chloride ions
zirconium phosphate gel
reflux with concd H
3 4
PO
dry to
(from 30 min to 100 h)
constant weight
ꢀ
1
ꢀ1
of about 1.62 g g
h
and almost complete selectivity
centrifuge off the solids and wash until
free of phosphate and chloride ions
toward HMF. In titanium phosphates under the same
calcination (3 h)
ꢀ
1
ꢀ1
conditions, the TOF of about 1.08 g g
h , and good
selectivity to HMF was reported. Inomata and co-work-
Figure 1. Schematic representation of the preparation method.

F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
2381
crystalline zirconium phosphate (ZrP) was produced.
The crystalline ZrP solid can be made with any degree
modules with a CSPAK narrow-bore column C18
(2.1 · 100 mm) coupled with a pulsed amperometric
(PDA) detector (Varian PDA 330 Detector), which
was set to 284 nm. Gradient elution program at
0.1 mL/min flow rate was used as below:
36
of crystallinity by varying the refluxing period. The
precipitated solid was washed with 2 N H PO and then
3
4
with distilled water until the pH reached a constant
value (pH ꢁ 4) and until free of phosphate and chloride
ions. Thereafter the sample was dried at 70 ꢀC for 72 h
and finely crushed. Finally, for calcination to occur,
the samples were heated, followed by thermal treatment
at 300 and 600 ꢀC for 3 h.
Time % of Mobile phase A
% of Mobile
(min) (1% aq solution of HOAc) phase B (MeOH)
0
1
1
5
5
7
85
0:00 74
15
26
98
98
15
15
5:00
3:00
2
2
2
.3. Characterization of zirconium phosphates
5:00 85
3:00 85
Surface areas were measured by adsorption of nitrogen
gas at ꢀ196 ꢀC (Belsorp 18plus-T-SP, Japan). Before
measurements, all samples were evacuated at 70 ꢀC for
In order to identify the presence of organic acids, an
HPLC instrument using Shimadzu LC-10AD VP
pumps with two serial ion-exclusion chromatography
columns (Shim-pack SCR-102H, 8 · 300 mm) and
post-column pH-buffered electroconductivity detection
9
h. The surface areas were determined using the BET
equation. The mercury porosimetry technique (Pascal
40 and 240, Thermo Finnigan) was used to measure
1
the pore size distribution of ZrP solids.
Elemental analysis of phosphate and zirconium was
carried out using an induced coupled plasma (ICP)
atomic emission spectrophotometer (SPS-7800 Plasma
spectrometer, Seiko, Japan). The composition of solids
was determined as follows: about 5–7 mg of the sample
was dissolved in 2 mL of 0.2 N of HF in a PVC flask. It
was then diluted to 50 mL, and zirconium and phos-
phate contents were determined by ICP.
(Shimadzu CDD-6A) was used. The mobile phase
was 5 mM of p-toluenesulfonic acid solution at a flow
rate of 0.8 mL/min. A mixture of 5 mM of p-toluene-
sulfonic acid and 100 mM of EDTA was used as the
post-column reagent at a flow rate of 0.8 mL/min.
The column temperature was kept at 40 ꢀC. Two serial
size-exclusion chromatography columns (Shodex-sugar
KS 804 and KS801, 8 · 300 mm) in an HPLC using a
Jasco PU-2080 plus a pump coupled with a refractive
index detector (RI 2031 plus) were used for quantita-
tive analysis of products that were undetectable by
the UV detector. The HPLC was operated at an oven
temperature of 30 ꢀC and 0.4 mL/min flow rate of
water as the mobile phase. The total organic carbon
2
.4. Catalytic dehydration reaction of fructose
A stainless steel tube (SUS 316, 8 mm i.d. and total vol-
ume of 7.7 mL) and a Swagelok fitting (purchased as
from Swagelok AG, Switzerland) were used for batch-
type sub-CW experiments at 240 ꢀC. The typical proce-
dure was as follows: fructose (0.05 g), catalyst (0.01–
(TOC) concentration of samples was measured using
a TOC analyzer (Shimadzu TOC-500). To heat the
samples, a furnace thermoregulated within ± 1 ꢀC was
used. A Horiba f-23 digital pH meter was used for
pH measurements.
The conversion, selectivity, absolute yield, and turn-
over frequency (TOF) were calculated by Eqs. 1–4,
respectively:
0
.05 g), and water (5.5 g) were poured into the reactor.
After the air in the reactor was replaced with argon, it
was capped tightly and immersed in a salt bath (Thomas
Kagaku Co. Ltd.) preheated to 240 ꢀC. The reaction was
allowed to proceed for a given reaction time. The reac-
tor was shaken before and also during immersion in
the salt bath in order to avoid settling of the catalyst
particles to the bottom, which inhibits accessibility for
the dissolved reactant. The reactor was then removed
from the salt bath and quickly quenched in a water bath
at ambient temperature. The pressure inside the reactor
C
substrateðiniÞ ꢀ Csubstrate
conversion ¼
ð1Þ
ð2Þ
ð3Þ
C
substrateðiniÞ
Cproduct
(
3.35 MPa at 240 ꢀC) was estimated from a steam
selectivity ¼ C
substrateðiniÞ ꢀ Csubstrate
37
table.
The sample after reaction was diluted to
5
0 mL with distilled water and analyzed for HMF,
C
product ꢂ MWsubstrate
absolute yield ¼ C
remaining fructose, and other byproducts.
substrateðiniÞ
ꢂ MWproduct
Turnover frequency ðTOFÞ
2
.5. Analysis
ðg ofÞ HMF
The products of the sub-CW reaction were analyzed by
HPLC using two Varian ProStar210 solvent delivery
¼
ð4Þ
ðg ofÞ catalyst ꢂ residence time ðhÞ

2382
F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
3
. Results and discussion
A3 to F3 in Table 1), respectively. These are ascribed
to the disappearance of crystalline water and condensa-
tion of hydroxyl groups. It seems at 300 ꢀC crystalline
water was lost from molecules, and at 600 ꢀC crystalline
water-loss and condensation reactions of hydroxyl
3
.1. Properties of solid acids
Table 1 shows the surface area, weight loss by heating,
and phosphate-to-zirconium molar ratio for a series of
ZrP solids that were prepared in this work. The initially
prepared solid acid, ZrP (0:0), had an amorphous struc-
ture. It has been previously shown that amorphous
forms of ZrP solids could change to semi-crystalline
and crystalline forms by refluxing in concentrated phos-
3
8
groups occurred.
The phosphate-to-zirconium ratio was close to 2 (1.8–
2) in all samples (see Table 1), which is relevant to its
basic formula Zr(O POH) ÆnH O, where n depends on
3
1
3
2
2
2
6
the drying method and degree of crystallinity. As
shown in Table 1, a small deviation from the ratio 2
was obtained for these samples. However, in general,
the crystals obtained after refluxing have approximately
the same composition as the amorphous sample.
2
6
phoric acid, while increasing with refluxing time.
Therefore, in Table 1 the crystallinity increased from
entry A1 toward F1. All the prepared samples, including
the amorphous to crystalline forms, were calcined at two
temperatures, 300 and 600 ꢀC for 3 h. The data for the
calcined samples are also shown in Table 1 (A2–3 to
F2–3).
Based on the IUPAC classification on adsorption iso-
therms, from Figure 2 it can be found that the prepared
3
9
samples were non-porous or macroporous structures.
The experimental results showed that the surface area
decreased almost linearly by refluxing time and/or by
heat treatment (Table 1). These results are quite similar
ZrP (0:0)
ZrP (100:600)
1
00
50
0
2
9
to that already reported by Thakur and Clearfield. The
effect of reflux was greater than heat treatment on the
surface area of the solid samples. By increasing the crys-
tallinity, the surface area decreased. Therefore, with re-
spect to the data of surface areas, it can be realized that
both refluxing and calcination may increase the crystal-
linity of solids toward higher crystalline forms. On the
other hand, the results of Table 1 showed that calcining
has a more noticeable effect on the amorphous forms
(
i.e., on A1 and A3 in Table 1) than on the higher crys-
0
0.2
0.4
0.6
0.8
1
P/P0
talline forms (i.e., on F1 and F3 in Table 1).
Average weight losses by the calcination at 300 and
00 ꢀC were 7.9 ± 1% and 11.6 ± 0.2% (A2 to F2 and
Figure 2. Typical adsorption isotherm of N2 on ZrP salts,
T = ꢀ196 ꢀC.
6
Table 1. Properties of various prepared zirconium phosphate solids
2
Entry
Solid acid
Surface area, m /g (BET)
wt% loss
P/Zr
molar ratio)
(
a
b
A
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
ZrP (0 :0 )
ZrP (0:300)
ZrP (0:600)
ZrP (0.5:0)
ZrP (0.5:300)
ZrP (0.5:600)
ZrP (1:0)
63.6
54.4
43.8
38.1
32.4
26.2
35.7
24.6
23.4
19.8
18.7
16.5
13.3
12.5
9.7
—
1.8
1.8
1.8
1.8
1.8
1.8
1.9
1.9
1.9
1.9
1.9
1.9
1.9
1.9
1.9
1.9
2.0
2.0
7.9
11.7
—
7.9
11.2
—
8.0
11.8
—
7.9
11.9
—
8.1
11.6
—
7.8
11.5
B
C
D
E
F
ZrP (1:300)
ZrP (1:600)
ZrP (3:0)
ZrP (3:300)
ZrP (3:600)
ZrP (10:0)
ZrP (10:300)
ZrP (10:600)
ZrP (100:0)
ZrP (100:300)
ZrP (100:600)
8.8
6.8
6.7
a
Reflux time of the zirconium phosphate gel (h).
Calcination temperature [ꢀC] (0 means no calcination).
b

F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
2383
Furthermore the mercury porosimetry technique has
been used in order to understand the size of the pore
of the samples (in the case of macroporous). The amor-
phous and crystalline forms of ZrP (0:0) and ZrP
3.2.1. Control experiments for catalytic activity.
Primary experiments were done with and without the
catalyst in order to demonstrate the existence of
catalytic activity of these solid acids under the sub-CW
conditions. Figure 4 shows the time course for the
production yields of HMF from fructose with or with-
out the solid acid catalyst [ZrP (0:0)] under the sub-
CW conditions at 240 ꢀC. Fructose was dehydrated to
HMF, even in the absence of catalyst. With increasing
sub-CW temperature, the production yield of HMF
increased to 120 s. Afterward the yield of HMF
decreased. This may be caused by decomposition and
polymerization of HMF. It is clear that ZrP (0:0) acted
as a catalyst because the yield increased about three
times compared to the non-catalytic condition.
(
(
100:600) showed different pore sizes and distribution
see Fig. 3a and b), respectively. Specific pore volume
decreased with increasing crystallinity, and the crystal-
line form had a narrow range of pore distribution com-
pared to the amorphous one. However, it was found
that for all solid samples, their volumes of macroporous
were small.
In this work we investigated the possibility of using
these solid acids as catalysts under sub-CW conditions.
Extensive work has been done to elucidate the nature,
crystallinity, and physicochemical properties of these
4
0,41
solids in detail elsewhere.
3.2.2. Effect of properties of various solid acids on
3.2. Catalytic activities of prepared solid acids under
sub-CW conditions
dehydration reaction. In general, the catalytic proper-
ties of the solid acids depend on the method of precipi-
tation, degree of crystallinity, calcination temperature,
and other physicochemical parameters. Thus, in order
to test the effect of the structure of various ZrP solids
made in this work in the sub-CW reaction, a series of
reactions were carried out for comparison by applica-
tion of the solid acids under the same reaction condi-
tions (residence time of 120 s, temperature of 240 ꢀC,
and catalyst-to-fructose ratio of 1:2). The experimental
results, along with the results of the controlled reaction
From the catalytic dehydration reaction of fructose, we
obtained some information on compatibility, activity,
selectivity, and other relevant information of ZrP solid
acids under the sub-CW conditions.
8
6
4
2
00
00
00
00
0
15
(
a) ZrP (0:0)
relative pore volume
specific volume
(reaction without catalyst), are summarized in Table 2.
The results are given by conversion of fructose, selectiv-
ity to HMF, furaldehyde, and soluble polymers, and
absolute yield of HMF. Soluble polymers could not be
quantitated; therefore, their selectivities were calculated
by difference, which attributed to the lost amounts in the
decomposition reaction of fructose.
1
5
0
In the absence of catalysts, fructose conversion, and
selectivity to HMF were lower than those with ZrP solid
1
00
101
102
103
104
pore radius, R (nM)
with catalyst
without catalyst
8
6
4
2
00
00
00
00
0
15
6
4
2
0
(
b) ZrP (100:600)
relative pore volume
specific volume
1
5
0
0
30
60
90
120
150
residence time (s)
1
00
101
102
103
104
Figure 4. HMF production from dehydration of 1% solution of
fructose under sub-CW conditions at 240 ꢀC and 3.35 MPa, as
function of time, in the presence and absence of amorphous form of
zirconium phosphate.
pore radius, R (nM)
Figure 3. Typical pore-size distribution curves determined by mercury
intrusion porosimetry on (a) ZrP (0:0) and (b) ZrP (100:600) solids.

2384
F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
Table 2. Catalytic activities of various zirconium phosphates in the dehydration of fructose (1 wt % aq solution) at 240 ꢀC, 3.35 MPa, 120 s, with an
initial catalyst-to-fructose ratio of 1:2 by weight
Entry
Catalyst
Conversion (%)
Selectivity to (%)
Absolute yield of HMF (%)
a
b
HMF
FA
Soluble polymers
G
H
I
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
ZrP (0:0)
80.6
78.4
77.1
80.9
74.7
72.2
79.1
76.2
73.7
79.2
72.5
70.9
79.7
68.7
67.2
81.3
67.6
65.6
59.1
61.3
56.2
52.1
62.1
56.8
51.4
59.3
56.6
50.5
62.3
57.4
51.7
62.0
53.9
47.2
60.9
54.2
45.3
31.8
2.2
1.9
1.9
2.4
2.2
1.9
2.2
2.2
1.9
2.3
2.2
1.8
2.4
2.2
1.6
2.1
1.9
1.5
1.1
36.5
41.9
46.0
35.5
41.0
46.7
38.5
41.2
47.6
35.4
40.4
46.5
35.6
43.9
51.2
37.0
43.9
53.2
67.1
49.4
44.1
40.7
50.2
42.5
37.2
46.9
43.2
37.2
49.4
41.7
36.6
49.4
37.0
31.7
49.6
36.7
29.7
18.6
ZrP (0:300)
ZrP (0:600)
ZrP (0.5:0)
ZrP (0.5:300)
ZrP (0.5:600)
ZrP (1:0)
ZrP (1:300)
ZrP (1:600)
ZrP (3:0)
J
ZrP (3:300)
ZrP (3:600)
ZrP (10:0)
K
L
M
ZrP (10:300)
ZrP (10:600)
ZrP (100:0)
ZrP (100:300)
ZrP (100:600)
c
Control
a
b
c
Furaldehyde.
Amounts of soluble polymers calculated by difference.
Reaction carried out in the absence of catalyst.
acids under the same experimental conditions (Table 2).
With application of the amorphous form of catalyst,
maximum conversion of fructose, about 80%, was ob-
served, and the best selectivity of HMF was obtained
at about 61%. Only 2.2% of selectivity was found for
the furaldehyde (see Table 2, G1). The refluxing time
and consequent degree of crystallinity, as well as the sur-
face area of solid acids, showed no increasing effects on
conversion of fructose and selectivity of HMF. (See Ta-
ble 1 for surface areas and Table 2, G–L.) It has already
been reported that there is a close relationship between
surface area, which corresponds to hydroxyl groups of
that surface hydroxyl groups were not solely responsible
for the catalytic activities, and the Lewis acid sites could
play a relevant role in the catalytic reactions. Another
possibility may be attributed to the existence of new
phases not previously reported.
Table 2 (G2–3 to L2–3) shows that the calcined sam-
ples had no similar influence on the decomposition reac-
tion of fructose. With the increase of calcination
temperature, catalytic activity decreased a little. Conver-
sion of fructose decreased about 2–16%, and conse-
quently the selectivity and absolute yield of HMF
decreased about 3–16% and 4–20%, respectively. On
the other hand, there was a relation between the surface
areas of calcined samples and their catalytic activities.
For example, the catalyst ZrP (100:600) with a surface
2
8
surface, and catalytic activity of ZrP solids. On the
other hand, under the normal reaction conditions, it
has been thought that the decrease of catalytic activity
is attributed to the decrease in surface area. Therefore,
it has been reported that the surface areas of the ZrP
solid acids were of prime importance in assessing their
2
area of 6.7 m /g rendered about 11% and 6.8% lower
yield and selectivity, respectively, than ZrP (0:600) with
2
a surface area of 43.8 m /g. If we assume that most of
2
7,28
behavior under catalysis.
In the present work, an
the hydroxyl groups of the surface acids condense at
higher temperatures, then again it can be concluded that
Lewis acid sites play a major role as the catalytic sites.
However, the decreasing yield of the former samples
may be attributed to less accessibility of Lewis acid sites
for the reaction process (Table 2, G3 and L3). Another
possibility is the regeneration of some hydroxyl groups
originating from the hydrolysis of P–O–P (condensed)
interesting result was obtained from the comparison of
surface areas and catalytic activities. Despite the differ-
ent surface areas, nearly similar performances were
observed for decomposition reactions of fructose. In
other words, there was no relationship between the
surface area and activity of the catalysts under sub-
CW conditions. Usually two different unique sites of
2
8
42
ZrP solids may be responsible for catalytic activities.
bonds caused by the water at high temperature. How-
One of the sites belong to hydroxyl groups (Brønsted
sites), while the nature of the second type of site is not
clearly understood as of yet, but it is potentially a Lewis
acid type. Based on what we obtained from different
surface areas and their activities, it can be concluded
ever, less activity compared to non-calcined samples
may be ascribed to the number of released hydroxyl
groups.
Contrary to what was previously reported for the
decomposition reaction of fructose under the normal

F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
2385
4
3
(
temperature and pressure) conditions, it was evident
In all experiments, the results of total organic carbon
analysis did not show formation of any solid polymers
(Humin).
that in the presence of the ZrP solid acids, there were
not any side-reaction products, including rehydration
of HMF to levulinic and formic acids and/or isomeriza-
tion to glucose. Byproducts identified were limited to
soluble polymers and furaldehyde. One of the well-
known reaction pathways under the normal conditions
for the rehydration reaction is the presence of hydro-
nium-ion species within the macroporous of solid
acids, which may catalyze HMF to levulinic and for-
mic acids. From Figures 2 and 3 it was found that the
ZrP solids may have only a very small proportion of
macroporous structures. Therefore, the non-appearance
of any rehydration products (levulinic and formic acid)
may be ascribed to the low availability of macroporous
structures, which as reported before, have a main role in
the dehydration reaction. The decomposition pathways
of fructose and byproducts, from another point of view,
can be explained by the surface acidity of catalysts and
or the acidity of the reaction media. It has already been
3.3. Solid acid catalyst reactions: study of the reaction
conditions
From the evaluation of the performances of different
ZrP solids, the amorphous form [ZrP (0:0)] was selected
for further studies, mainly because of the higher cata-
lytic activity than that of either the semi-crystalline or
crystalline ones. A series of experiments were done at
a variety of residence times and catalyst-to-substrate
ratios (1:1, 1:2, and 1:4) under sub-CW conditions at
240 ꢀC in order to determine better reaction parameters
(Table 3). Fructose and glucose, which are an example
of a ketohexose and an aldohexose, respectively, were
used as substrates. From Table 3 it can be seen that
the conversion of substrate and selectivity were con-
trolled by the residence time and the amount of catalyst.
As mentioned before, the selectivity of soluble polymers
was calculated by difference (Table 3), which was attrib-
uted to the lost amounts in decomposition reaction of
fructose and glucose.
4
4
4
5
reported that it is possible to form levulinic and formic
acids from the decomposition of fructose at pH <2.7
under the normal reaction conditions, but at pH >3.9,
no formation of HMF is observed. It seems that under
sub-CW conditions, the catalysts tested showed moder-
ate acidity, so that no rehydration products were identi-
fied in the sub-CW decomposition reactions.
For both substrates, with increasing residence time up
to 240 s with any catalyst-to-substrate ratio, the conver-
sion amount increased. The conversion of glucose was
Table 3. Effect of residence time and catalyst/substrate ratio on dehydration reaction of fructose and glucose under sub-CW conditions at 240 ꢀC,
3.35 MPa and amorphous ZrP [ZrP (0:0)] as acid catalyst
Entry
Substrate
Catalyst/fructose
w/w)
Residence
time (s)
Conversion
(%)
Selectivity (%) to
Absolute yield
of HMF (%)
(
a
b
HMF
FA
Soluble polymers
N
O
P
1
Fructose 1 wt%
1:1
1:2
1:4
60
120
180
240
60
120
180
240
60
45.1
88.5
96.6
98.6
41.7
79.6
94.6
97.5
37.1
78.4
94.0
97.0
56.1
56.2
55.4
44.9
50.2
61.8
55.7
44.6
54.6
56.1
53.4
42.5
1.8
2.8
3.3
3.2
2.2
2.3
2.9
3.1
1.8
2.2
2.5
2.8
42.1
41.0
41.3
51.9
47.6
35.9
41.7
47.7
43.6
41.7
44.1
54.7
25.3
49.7
53.5
44.3
20.9
49.1
52.8
43.5
20.2
44.0
50.2
41.2
2
3
4
1
2
3
4
1
2
3
4
120
180
240
Q
R
S
1
2
3
4
1
2
3
4
1
2
3
4
Glucose 1 wt%
1:1
1:2
1:4
60
120
180
240
60
120
180
240
60
21.2
40.8
53.1
72.3
19.3
38.7
48.9
63.9
17.3
36.9
46.6
60.9
21.1
34.9
39.0
32.6
18.6
31.7
37.4
31.5
18.3
32.9
36.4
30.5
0.4
0.8
1.5
1.6
0.5
0.7
5.0
1.4
0.5
0.7
0.8
1.3
78.4
64.3
59.5
65.8
80.9
67.6
62.1
67.1
81.2
66.4
62.8
68.2
4.5
14.2
20.7
23.5
3.6
12.3
18.3
20.1
3.2
120
180
240
12.1
17.0
18.6
a
Furaldehyde.
b
Amounts of soluble polymers calculated by difference.

2386
F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
ꢀ
1
ꢀ1
less than that of fructose under identical conditions.
This can be attributed to less reactivity of aldohexoses
than ketohexoses in the dehydration and/or decomposi-
tion reactions. Similar behavior was obtained in the
presence of phosphoric acid as catalyst in our previous
TOF of 26.77 g g
h
was obtained for 80% conver-
sion and for best selectivity to HMF under sub-CW con-
ditions (Table 3, O2). This value is much higher than
those reported for the conversion of fructose by using
ZrP solid acids with classical methods. The maximum
1
6
28,34
ꢀ1 ꢀ1
work. In addition the conversion of the substrate
was a function of its ratio with the catalyst.
TOF reported
was less than 1.80 g g
h . The
TOF can be decreased by decreasing selectivity, and in-
creased by shortening the residence time. In sub-CW
reactions, because of the short residence time compared
to what was already reported for conventional methods
(30–60 min), the TOF showed a quite higher value.
Usually it is difficult to get higher selectivity of HMF
in aqueous reactions due to formation of soluble poly-
mers from polymerization of the substrate and/or
HMF, cleavage of HMF to levulinic and formic acid,
and other decomposition products. Generally the selec-
tivity of the reaction varies with the reaction media, ini-
tial concentration of substrate, reaction temperature,
residence time, catalyst type, pH of solution, and other
relevant factors. The experimental results showed the
best selectivity of nearly 61% from about 80% of conver-
sion of fructose after 120 s residence time and with a cat-
alyst-to-fructose ratio of 1:2 (see Table 3, O2). At longer
residence times, the selectivity did not change in spite of
increasing the conversion of fructose. It can be con-
cluded that with increasing residence time, in addition
to formation of HMF, the formation of polymers as
the main side-reaction increased. Therefore, in spite of
the increasing conversion of fructose, as shown in Table
3.4. Regeneration of the ZrP solid acids
After using ZrP solid acids in the sub-CW reaction,
regeneration is necessary. The results showed that these
solids could be regenerated after 30 min shaking in a
solution of 2 N of phosphoric acid. After drying, the cat-
alysts can be recycled back to a fresh batch reactor
about 6–7 times without losing any catalytic properties
such as selectivity and activity. However, after using
these 6–7 times, their catalytic properties decrease some-
what. Therefore these results showed an important
improvement with respect to the sub-CW reactions
using mineral acids.
3
, the selectivity did not increase more than about 61%.
In the case of glucose as substrate, the best selectivity
3.5. Effect of ZrP solid acids on the corrosion of sub-CW
reactors
for HMF, about 39%, was obtained from conversion of
nearly 53% after 180 s residence time, with a catalyst-to-
substrate ratio of 1:1.
One of the advantages of heterogeneous catalysts is low-
er corrosion of reactors compared to that experienced
with homogeneous acids. For comparison, two series
of experiments were done for the dehydration reaction
For the comparison of two substrates, in the case of
glucose, longer residence time and larger amounts of
catalyst were necessary to obtain higher selectivity than
that of fructose. Table 3 shows that in the case of fruc-
tose as substrate, with increasing residence time, the
yield of HMF initially increased and then decreased
due to polymerization reactions. In the case of glucose,
there is a linear increase in the yield of HMF with resi-
dence time. Due to lower reactivity of aldohexoses com-
pared to ketohexoses, it seems that even with longer
residence times, although soluble polymers are pro-
duced, there is still an adequate amount of substrate
for the reaction process. Therefore increasing the
amount of HMF by residence time can be observed.
Under all conditions the production and/or decompo-
sition amount of furaldehyde was almost similar to that
of HMF. And also contrary to what is observed in the
dehydration of fructose, the isomerization of glucose
to fructose was identified in the dehydration reaction,
starting with glucose as substrate.
1
6
of fructose by using ZrP (0:0) and phosphoric acid at
pH 2 as heterogeneous and homogeneous acid catalysts,
respectively. The contents of the reactors after reaction
at 240 ꢀC for 120 s were analyzed for iron, which may
correspond to the amount of corrosion in the reactors.
With ZrP, solid acid and iron ions were not identified.
However, an iron content of about 10 ppm was detected
in the case of phosphoric acid catalyst. In spite of the
fact that the sub-CW reactor showed corrosion in the
presence of heterogeneous acid catalyst, there was no
detected corrosion when using ZrP solid acid.
4. Conclusions
We have found that ZrP solids have catalytic activities
under sub-CW conditions for the dehydration of fruc-
tose and glucose to HMF. Various kinds of ZrP solids,
from amorphous to crystalline forms, were prepared in
order to study their catalytic behavior under sub-CW
conditions.
The turnover frequency (TOF) is usually used for
expression of the effect of solid catalysts on the reaction,
which contains the parameters of selectivity, conversion,
catalyst amount, and residence time (see Eq. 4). In com-
parison of sub-critical water with conventional methods
for dehydration of fructose, based on calculations, a
The prepared catalysts had no micro and ultra-micro
pores. However, very small volumes of macroporous

F. S. Asghari, H. Yoshida / Carbohydrate Research 341 (2006) 2379–2387
2387
regions were observed. Under sub-CW conditions, no
relationship was observed between catalytic activity
and the surface area of the catalysts. It seems that not
only Brønsted sites, but also Lewis sites of the ZrP sol-
ids, are responsible for their catalytic activities. The ZrP
solids in amorphous forms were found to have the same
or higher activities than the other crystalline forms.
By application of ZrP solids as catalysts under sub-
CW conditions, with amorphous forms of the catalyst,
about 80% of fructose was decomposed at 240 ꢀC after
12. Szmant, H. H.; Chundury, D. D. J. Chem. Technol.
Biotechnol. 1981, 31, 135–145.
1
1
3. Mendnick, M. L. J. Org. Chem. 1962, 27, 398–403.
4. Chen, J.; Kuster, B. F. M.; Der Wiele, K. V. Biomass
Bioenergy 1991, 1, 217–223.
15. Jow, J.; Rorrer, G. L.; Hawley, M. C. Biomass 1987, 14,
185–194.
1
6. Salak Asghari, F.; Yoshida, H. Ind. Eng. Chem. Res. 2006,
5, 2163–2173.
4
1
1
7. Kuster, B. F. M. Starch 1990, 42, 314–321.
8. Seri, K.; Inoue, Y.; Ishida, H. Bull. Chem. Soc. Jpn. 2001,
74, 1145–1150.
1
20 s, and the selectivity of the dehydration reaction of
19. Garber, J. D.; Jones, R. E. U. S. Patent 3,483,228, 1969.
2
2
0. Szmant, H. H.; Chundury, D. D. J. Chem. Technol.
Biotechnol. 1981, 31, 135–145.
1. Mercadier, D.; Rigal, L.; Gaset, A.; Gorrichon, J. P. J.
Chem. Technol. Biotechnol. 1981, 31, 489–496.
fructose to HMF rose to 61%. Under the same experi-
mental conditions, starting with glucose as substrate,
3
9% of selectivity was obtained. The only identified
byproducts were soluble polymers and furaldehyde.
These solid acids showed important improvements
with respect to using mineral acids. They can be easily
separated from reaction media and regenerated and
used for several runs. Meanwhile, no corrosion occurred
under sub-CW conditions with ZrP solid acids.
Finally, we believe that, based on the type of reactions
and substrates, optimization, and modification of these
solid acids will increase the performance of catalysts
under sub-CW conditions.
22. Nakamura, Y. Jpn. Kokai Tokkyo Koho, Jpn. Pat.
55013243, 1980.
2
3. Armaroli, T.; Busca, G.; Carlini, C.; Giuttari, M.; Galletti,
A. M. R.; Sbeana, G. J. Mol. Catal. A: Chem. 2000, 151,
2
33–243.
2
4. Moreau, G.; Durand, R.; Razigade, S.; Duhamet, J.;
Faugeras, P.; Rivalier, P.; Ros, P.; Avignon, G. Appl.
Catal., A 1996, 145, 211–224.
2
2
5. Okuhara, T. Chem. Rev. 2002, 102, 3641–3666.
6. Bogdanov, S. G.; Valiev, E. Z.; Dorofeev, Yu A.; Pirogov,
A. N.; Sharygin, L. M.; Moisseev, V. E.; Galkin, V. M. J.
Phys.: Condens. Matter 1997, 9, 4031–4039.
2
2
2
3
3
7. Segawa, K.; Nakajima, Y. J. Catal. 1986, 101, 81–89.
8. Clearfield, A.; Thakur, D. J. Catal. 1980, 65, 185–194.
9. Thakur, D.; Clearfield, A. J. Catal. 1981, 69, 230–233.
0. Moffat, J. B. Catal. Rev. Sci. Eng. 1978, 18, 18–25.
1. Segawa, K.; Ozawa, T. J. Mol. Catal. A: Chem. 1999, 141,
Acknowledgements
A part of this work was supported by the ministry of
Education, Culture, Sports, Science and Technology of
Japan in the form of 21st Century COE program
E19, Science and Engineering for Water Assisted Evo-
lution of Valuable Resources and Energy from Organic
Wastes) which is gratefully acknowledged.
2
49–255.
32. Iwamoto, M.; Nomura, Y.; Kagawa, S. J. Catal. 1981, 69,
234–237.
3
3. Watanabe, M.; Aizawa, Y.; Iida, T.; Aida, T. M.; Levy,
(
C.; Sue, K.; Inomata, H. Carbohydr. Res. 2005, 340, 1925–
1
930.
3
4. Benvenuti, F.; Carlini, C.; Patrono, P.; Raspolli, G. A. M.;
Sbrana, G.; Antonietta, M. M.; Galli, P. Appl. Catal., A
2
000, 193, 147–153.
35. Clearfield, A.; Stynes, J. A. J. Inorg. Nucl. Chem. 1964, 26,
17–129.
References
1
3
3
6. Clearfield, A.; Thakur, D. Appl. Catal. 1986, 26, 1–26.
7. Yoshida, H.; Tavakoli, O. J. Chem. Eng. Jpn. 2004, 37,
1
2
. Fukushima, Y. R&D Rev. Toyota CRDL 2000, 35, 1–9.
. Jennings, J. M.; Bryson, T. A.; Gibson, J. M. Green Chem.
2
53–260.
2
000, 2, 87–88.
3
3
4
4
4
8. Alberti, G.; Costantino, U.; Millini, R.; Perego, G.;
Vivani, R. J. Solid State Chem. 1994, 113, 289–295.
9. Barton, T. J.; Bull, L. M.; Klempere, W. G.; Loy, D. A.
Chem. Mater. 1999, 11, 2633–2656.
0. Hattori, T.; Ishiguro, A.; Murakami, Y. J. Inorg. Nucl.
Chem. 1978, 40, 1107–1111.
1. Busca, G.; Lorenzelli, V.; Galli, P.; Ginestra, A.; Patrono,
P. J. Chem. Soc., Faraday Trans. 1987, 83, 853–864.
2. Ginestra, A. L.; Patrono, P.; Berardelli, M. L.; Galli, P.;
Ferragina, C.; Massucci, M. A. J. Catal. 1987, 103, 346–
3
4
5
. Antal, M. J.; Mok, W. S. Res. Thermochem. Biomass
Convers. 1988, 464–472.
. Antal, M. J.; Mok, W. S. Carbohydr. Res. 1990, 199, 91–
1
09.
. Bicker, M.; Hirth, J.; Vogel, H. Green Chem. 2003, 5, 280–
84.
2
6
7
. Fleche, G.; Toulouse, A. G. U. S. Patent 4,339,387, 1982.
. Seri, K.; Ishida, H. J. Mol. Catal. A: Chem. 1996, 112,
L163–L165.
8
9
. Lewkowski, J. Arkivoc 2001, 2, 17–54.
. Kuster, B. F. M.; Der Baan, H. V. Carbohydr. Res. 1977,
3
56.
4
4
3. Kuster, B. F. M. Carbohydr. Res. 1977, 54, 177–183.
4. Moreau, C.; Durand, R.; Pourcheron, C.; Razigade, S.
Ind. Crops Prod. 1994, 3, 85–90.
5
4, 165–176.
0. Harris, D. W.; Feather, M. S. J. Org. Chem. 1974, 39, 724–
25.
1. Moye, C. J.; Goldsck, R. J. J. Appl. Chem. 1966, 16, 206–
08.
1
1
7
4
5. Kuster, B. F. M.; Temmink, H. M. G. Carbohydr. Res.
1
977, 54, 185–191.
2