SnCl4-catalyzed isomerization/dehydration of xylose and glucose to furanics in water

Article abstract of DOI:10.1039/c5cy00077g

A number of Lewis acid catalysts were screened for their effectiveness in converting both xylose and glucose in aqueous media to furfural and 5-HMF, respectively. While other catalysts were found to be more active, SnCl₄ was identified as the most selective Lewis acid. Hydrolysis of SnCl₄ was observed at various concentrations and temperatures resulting in the production of Bronsted acidic protons in a 3.5:1 ratio to Sn⁴⁺ at all SnCl₄ concentrations above 60°C. As a consequence, there was no need to add a Bronsted acid in order to promote the dehydration of either xylose or glucose. SnCl₄-promoted isomerization/dehydration of xylose and glucose at 140°C in water resulted in conversions of 55% and 33%, respectively, after 2 h of reaction, and furfural and 5-HMF selectivities of up to 58% and 27%, respectively. Significant conversion of sugars to humins was observed in both cases, and in the case of glucose, degradation of 5-HMF to levulinic and formic acids was also noted. The effects of secondary reactions could be greatly suppressed by extraction of the furanic product as it was produced. Using n-butanol as the extracting agent, xylose and glucose conversions of 90% and 75%, respectively, were observed after 5 h of reaction, and the selectivities to furfural and 5-HMF increased to 85% and 69%, respectively. Small additional increases in the furfural and 5-HMF selectivities were obtained by adding LiCl to the aqueous phase without much effect on the conversion of either sugar. In this case, the selectivities to furfural and 5-HMF were 88% and 72%, respectively, after 5 h of reaction at 140°C.

Full text of DOI:10.1039/c5cy00077g

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
SnCl -catalyzed isomerization/dehydration of
4
xylose and glucose to furanics in water†
Cite this: DOI: 10.1039/c5cy00077g
ab
Kristopher R. Enslow and Alexis T. Bell*ab
A number of Lewis acid catalysts were screened for their effectiveness in converting both xylose and
glucose in aqueous media to furfural and 5-HMF, respectively. While other catalysts were found to be more
4 4
active, SnCl was identified as the most selective Lewis acid. Hydrolysis of SnCl was observed at various
concentrations and temperatures resulting in the production of Brønsted acidic protons in a 3.5 : 1 ratio to
4
+
Sn at all SnCl
4
concentrations above 60 °C. As a consequence, there was no need to add a Brønsted acid
-promoted isomerization/dehydra-
in order to promote the dehydration of either xylose or glucose. SnCl
4
tion of xylose and glucose at 140 °C in water resulted in conversions of 55% and 33%, respectively, after 2
h of reaction, and furfural and 5-HMF selectivities of up to 58% and 27%, respectively. Significant conver-
sion of sugars to humins was observed in both cases, and in the case of glucose, degradation of 5-HMF to
levulinic and formic acids was also noted. The effects of secondary reactions could be greatly suppressed
by extraction of the furanic product as it was produced. Using n-butanol as the extracting agent, xylose
and glucose conversions of 90% and 75%, respectively, were observed after 5 h of reaction, and the selec-
tivities to furfural and 5-HMF increased to 85% and 69%, respectively. Small additional increases in the fur-
fural and 5-HMF selectivities were obtained by adding LiCl to the aqueous phase without much effect on
the conversion of either sugar. In this case, the selectivities to furfural and 5-HMF were 88% and 72%,
respectively, after 5 h of reaction at 140 °C.
Received 19th January 2015,
Accepted 7th March 2015
DOI: 10.1039/c5cy00077g
www.rsc.org/catalysis
4
is thought to occur via enolization. Xylulose can then dehy-
drate with the aid of a Brønsted acid to form furfural. Scheme
1
. Introduction
Furfural and 5-hydroxymethyl furfural (5-HMF) are of great
interest as platform molecules for the production of specialty
chemicals and fuels. Both products can be obtained from
2 shows an analogous set of reaction pathways for the conver-
sion of glucose to 5-HMF in the presence of Brønsted and
Lewis acids. It should be noted that furfural and 5-HMF can
1
the carbohydrate portion of lignocellulosic biomass. Furfural
is produced by dehydration of xylose, obtained by hydrolysis
of hemicellulose, and 5-HMF is produced by dehydration of
glucose, obtained by hydrolysis of cellulose. The primary reac-
tions occurring during both dehydration processes are shown
2
,3
in Schemes 1 and 2, with the pathways shown in black cata-
lyzed by Brønsted acids and those shown in blue catalyzed by
Lewis acids. The dehydration of xylose can occur directly to
furfural catalyzed by a Brønsted acid. However, intermediates
involved in this process can react with either xylose or furfu-
ral to condense and form humins, undesired soluble and
insoluble degradation products. In the presence of a Lewis
acid, xylose can epimerize to lyxose, and either carbohydrate
can isomerize to xylulose. This aldose to ketose isomerization
a
Energy Biosciences Institute, Energy Biosciences Building, 2151 Berkeley Way,
Berkeley, CA 94704-1011, USA. E-mail: bell@cchem.berkeley.edu
b
Department of Chemical and Biomolecular Engineering, University of California,
Berkeley, CA 94720-1462, USA
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
Scheme 1 The reaction network for xylose dehydration catalyzed by a
c5cy00077g
Brønsted acid (shown in black) and a Lewis acid (shown in blue).
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Scheme 2 The reaction network for glucose catalyzed by a Brønsted acid (shown in black) and a Lewis acid (shown in blue).
both undergo Brønsted acid-catalyzed secondary reactions. In
the case of furfural, self-condensation or resinification leads
to humins, and in the case of 5-HMF, hydrolysis of the pri-
mary product leads to levulinic and formic acids.
levulinic/formic acid production, as 5-HMF was found to
hydrolyze to the acid products in all cases. For both sugars,
all Lewis acids increased the rate of sugar conversion and the
total product yield when compared to the results observed for
Brønsted acid-catalyzed reactions. Of the metal triflates,
While high yields of furfural and 5-HMF have been
5
–7
reported in organic solvents,
similar yields have not been
InIJOTf) converted both sugars at the highest rate (in excess
3
achieved for dehydration occurring in water at moderate reac-
tion temperatures. Higher conversions and yields of furanics
can be achieved using a mixture of Brønsted and Lewis acids
2
of 93% after 2 h), however, SnIJOTf) was the most selective to
2
,3,7
than is possible using only a Brønsted acid catalyst.
The
Table 1 The conversion of xylose (750 mM) and the yield of furfural at
40 °C catalyzed by various Lewis acidic metal chlorides and triflates (25
higher conversion of xylose and glucose is a consequence of
the lower apparent activation energy barrier for Brønsted
acid-catalyzed dehydration of xylulose and fructose, and the
higher selectivity to furanics is a consequence of the lower
susceptibility of the furanose isomers to participate in degra-
dation reactions compared to the corresponding pyranose
1
mM) in a single aqueous phase measured after 1 and 2 h of reaction. Data
for the HCl-catalyzed (25 mM) dehydration under the same conditions
are shown for comparison
Time Xylose conversion, Furfural yield, Selectivity,
Catalyst
HCl
(h)
mol%
mol%
mol%
7
isomers. Additional improvements in the yields of furanics
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
4
8
1.3
2.4
31
29
4
32
30
41
32
14
20
33
30
46
39
27
25
30
31
48
48
65
58
25
25
29
25
35
41
30
31
can be attained by in situ extraction of the product from the
1
,7–9
AlCl
3
× 6H
2
O
75
91
28
35
75
92
70
92
18
30
24
40
29
42
35
55
89
96
48
64
32
39
32
45
aqueous phase into an immiscible organic phase.
In this work we investigate a number of Lewis acid cata-
lysts and evaluate their ability to selectively dehydrate both
xylose and glucose to their corresponding furanics, furfural
and 5-HMF, in water at moderate temperatures. Screening
studies revealed that the highest yields of furfural and 5-HMF
CeCl
3
× 7H
2
O
7
CrCl
CrCl
FeCl
FeCl
2
3
2
3
25
28
32
36
5
× 6H
× 4H
× 6H
2
O
O
O
4
could be achieved using SnCl and hence all further work
2
2
was carried out with this catalyst. Minimization of the effects
of secondary reactions by means of in situ solvent extraction
of the furanic products was also investigated.
8
7
12
14
20
23
32
22
24
14
16
11
16
9
SnCl
2
4
× 2H
× 5H
2
O
SnCl
2
O
2
. Results and discussion
InIJOTf)
3
2
.1 Xylose and glucose isomerization/dehydration in water
A number of metal chlorides and triflates were screened in
order to establish their effectiveness as Lewis acid catalysts
in converting xylose and glucose to furfural and 5-HMF,
respectively. The results are summarized in Tables 1 and 2.
Glucose selectivity is tabulated to reflect 5-HMF and
ScIJOTf)3
SnIJOTf)
2
YbIJOTf)
3
14
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
Table 2 The conversion of glucose (750 mM) and the yields of 5-HMF,
levulinic acid (LA), and formic acid (FA) at 140 °C catalyzed by various
Lewis acidic metal chlorides and triflates (25 mM) in a single aqueous
phase measured after 1 and 2 h of reaction. Data for HCl-catalyzed (25
mM) dehydration of glucose under the same conditions are shown for
comparison. Product selectivity is the sum of glucose selectivity toward
5-HMF and formic acid
Glucose
Time conversion, yield,
5-HMF LA/FA
yield,
Product
selectivity,
Catalyst
HCl
(h)
mol%
mol%
mol% mol%
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
2
3
—
0.6
10
18
3
4
9
10
11
13
2
3
4
6
4
5
5
4
11
9
—
—
4
—
20
21
35
24
24
23
22
24
26
22
27
36
39
60
63
69
70
57
59
46
45
65
70
61
61
AlCl
3
× 6H
2
O
68
80
21
25
70
89
68
90
9
15
11
18
22
26
26
33
87
93
59
66
49
58
21
31
10
2
2
7
10
5
CeCl
3
× 7H
2
O
Fig. 1 The conversion of xylose (750 mM) and the yield of furfural at
CrCl
CrCl
FeCl
FeCl
2
4
140 °C catalyzed by 25 mM SnCl in a single aqueous phase.
3
× 6H
× 4H
× 6H
2
O
O
O
10
0
1
0
1
9
to form α-angelica lactone (found to be present in <5%). A
proposed pathway for this side reaction is given in Scheme 3,
wherein the cyclization is shown to be initiated via either car-
bonyl or hydroxyl group attack of the ketone carbon, followed
by dehydration.
2
3
2
2
SnCl
2
4
× 2H
2
O
9
11
13
19
39
46
22
25
20
24
9
SnCl
× 5H
2
O
InIJOTf)
ScIJOTf)
SnIJOTf)
YbIJOTf)
3
2.2 Lewis acid hydrolysis in water
The production of any dehydration product from xylose and
glucose is an indication of the presence of a Brønsted acid,
as protons in solution are necessary for hydroxyl group
removal during all dehydration steps leading to furfural and
3
5
5
12
17
4
2
3
9
7
12
5-HMF. Furthermore, levulinic and formic acid formation
from 5-HMF and α-angelica lactone formation from levulinic
9
acid can also only occur in the presence of a Brønsted acid.
dehydration products (41% for xylose and 70% for glucose).
ChromiumIJII) and (III) chloride exhibited the highest activities
for sugar conversion of all metal chlorides (in excess of 90%
Therefore it is important to evaluate the extent to which
SnCl generates Brønsted acidity in water.
4
4
+
When SnCl is added to water, it dissociates into Sn and
Cl , followed immediately by the complexation of Sn ions
with 6 water molecules to form the octahedrally-coordinated
4
−
4+
after 2 h); however, SnCl
4
was the most selective to dehydra-
tion products (58% for xylose and 70% for glucose). For this
(4−x)+
reason, SnCl
study.
4
was used exclusively for the remainder of this
aquachlorotinIJIV), ijSnCl IJH O) ]
6−x
, and hexaaquatinIJIV),
x
2
The temporal evolution of products for SnCl -catalyzed
4
dehydration of xylose and glucose is shown in Fig. 1 and 2,
respectively. For both sugars, the isomeric products lyxose,
xylulose, mannose, and fructose were either not observed or
were present in negligible quantities. Fig. 1 shows xylose con-
version reaching 55% over 2 h, with furfural production
increasing toward 32% over the same period. Any unac-
counted mass is attributed to the formation of soluble degra-
dation products and humins. Fig. 2 shows that glucose is
converted to 67% of its initial concentration in 2 h, while
5
-HMF is produced in small quantities and goes through a
maximum of 6% after 1 h of reaction. The low concentration
of 5-HMF is due to its hydrolysis to levulinic and formic
acids, shown to reach yields of 12% and 19%, respectively,
after 2 h of reaction. Produced in a 1 : 1 molar ratio from
Fig. 2 The conversion of glucose (750 mM) and the yields of 5-HMF,
5
-HMF, the two acid yields are not found in the expected stoi-
4
levulinic acid, and formic acid at 140 °C catalyzed by 25 mM SnCl in a
chiometric ratio because levulinic acid undergoes cyclization
single aqueous phase.
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Scheme 3 A proposed reaction scheme for the Brønsted acid-catalyzed cyclization of levulinic acid to an equilibrium mixture of α- and
β-angelica lactone.
(
4−y)+
10–12
ijSnIJOH)
y
(H
2
O)6−y
]
, complexes.
However, at SnCl
4
We note that regardless of the exact distribution of SnIJIV)
complexes in solution, SnCl4 provides both Lewis- and
Brønsted-acid centers for aqueous phase reactions. Previous
work has shown that even with 200 mM of HCl, xylose con-
version only reached 45% after 2 h using purely Brønsted-
acid centers at 140 °C. When the isomerization/dehydration
4
of xylose is carried out with 25 mM SnCl in the absence of
added Brønsted acid, 55% of the xylose was converted over 2
concentrations of 100 mM and lower, only a single peak is
observed when such solutions are analyzed via
This peak has been exclusively attributed to hexaaquatinIJIV)
ions, indicating that any chloride-containing complexes are
1
19
Sn NMR.
1
0
15
completely dissociated at low SnCl
4
concentrations. The
complex ions can undergo hydrolysis,
releasing a proton for each stage of hydrolysis, for example:
(
4−y)+
ijSnIJOH)
y
IJH
2
O)6−y
]
h even though tin hydrolysis results in less Brønsted acidity
by comparison. Therefore, SnO and ijSnIJOH) IJH O)6−y]
2 y 2
4
+
3+
+
(4−y)+
[
Sn(H
2
O)
6
]
+ H
2
O ⇆ [Sn(OH)(H
2
O)
5
]
+ H
3
O
(1)
complexes evolved from SnCl hydrolysis are active as Lewis
4
It has also been reported that SnCl
water-insoluble SnO releasing
4
can hydrolyze to form
protons per oxide
acids and providers of Brønsted acidity.
2
,
4
Lastly, the fact that little to no xylulose and fructose were
observed can be explained by the hydrolysis of the tin
increasing the Brønsted acidity of our aqueous solutions such
that the rate of ketose carbohydrate dehydration became
competitive with the isomerization of xylose and glucose
under these conditions.
1
1,13
formed.
observed after allowing aqueous SnCl
tions 10, 20, 40, and 60 mM to sit at room temperature over
4 h, which is likely an indication that the metal oxide forms
Small quantities of white precipitate were
4
solutions of concentra-
2
under the conditions used in the present study.
The exact distribution of hexaaquatinIJIV) complexes and
SnO2 existing in sub-100 mM SnCl4 solutions is undefined
and consequently an assessment of aggregate Lewis acidity
cannot be made. However, the distinction does not appear to
be important. A recent study has found that chitosan could
be converted to 27.4% levulinic acid via 5-HMF when either
2
.3 Effect of Lewis acid concentration and reaction
temperature on selectivity
The initial rates of xylose/glucose conversion and furfural/5-
HMF production were evaluated as a function of SnCl initial
4
4 2
SnCl or the equivalent of SnO and HCl were used, indicat-
ing either complete oxidation of tin in the first case or a rela-
tively similar Lewis acidity exists for hexaaquatinIJIV) cationic
1
4
complexes and SnO
the overall extent of hydrolysis by measuring solution pH as
a function of SnCl concentration in water. Fig. 3 shows that
proton concentration increases with both SnCl concentra-
tion and temperature over the range of 20–80 °C. For a given
SnCl concentration, increasing temperature appears to have
2
species. In either case, we investigated
4
4
4
a diminishing effect on proton concentration, such that at
temperatures in excess of 80 °C, the proton concentration
varies little with increasing temperature. This conclusion is
reinforced by the observation that at 140 °C, a 25 mM SnCl
4
+
solution has a pH of 1.1 ([H ] = 80 mM). Fig. 3 also shows
+
that for all conditions tested, the H : Sn ratio is always less
than 4, an indication that SnCl
4
does not hydrolyze
+
Fig.
3
The effect of SnCl
4
concentration on
H
concentration
(
4−y)+
completely to SnO nor do ijSnIJOH) IJH O)
]
complexes
2
y
2
6−y
observed for temperatures between 20 °C and 80 °C. Measurements
hydrolyze to form neutral or negative charged Sn species.
were taken after solutions were allowed to equilibrate for 12 h.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
−
1
concentration and temperature. The results are presented in
furfural are reported to be 32, 15.5, and 23.1 kcal mol ,
2
Table 3. As the SnCl concentration increases from 25 to 75
respectively (where CrCl was used for isomerization). Simi-
4
3
mM at either 140 °C or 160 °C, the rates of both xylose and
glucose conversion increase, as do those of furfural and
larly, the activation energies for direct dehydration of glucose
to 5-HMF, isomerization to fructose, dehydration of fructose
to 5-HMF, and hydrolysis of 5-HMF to levulinic and formic
5
-HMF production. The selectivity toward furfural does not
−
1
change with Lewis acid concentration, however, 5-HMF selec-
acids have been found to be 36.4, 20, 29.4, 23.1 kcal mol ,
7
tivity increases with SnCl concentration. As previously noted,
respectively. Reactions with higher activation energy barriers
4
4
+
Sn hydrolysis increases linearly with SnCl concentration at
are more sensitive to increases in temperature. For example,
by taking the ratio of Arrhenius rate constants at two temper-
atures, one finds that the rate of the direct dehydration of
xylose increases by 6.2-fold whereas the rate of xylose isomer-
ization only increases by 2.4-fold when the temperature is
adjusted from 140 °C to 160 °C (similarly, 7.8-fold compared
to 3-fold for glucose). Therefore, the direct dehydration path-
way for both sugars becomes increasingly competitive with
the isomerization/dehydration pathway as the temperature
increases. The overall selectivity to furanics decreases as a
result of the increased production of highly reactive interme-
diates involved in the direct dehydration process, which are
proposed to be the primary participants in degradation path-
4
temperatures above 40–60 °C, such that proton concentration
increases proportionally with the amount of metal cations
4
+
dissolved (about 3.5 protons per Sn cation). Since xylose
selectivity to furfural is unaffected by SnCl concentration,
4
Brønsted acid- and Lewis acid-catalyzed reaction pathways
shown in Scheme 1 must have the same rate-order dependen-
4
+
cies on proton and Sn concentrations, respectively. With
glucose selectivity toward 5-HMF increasing slightly with
SnCl concentrations, however, Lewis acid-catalyzed reaction
4
pathways shown in Scheme 2 may have higher rate-order
4
+
dependencies on Sn concentration than the Brønsted acid-
catalyzed reaction pathways have on proton concentration.
Comparing the results from 140 °C and 160 °C for a given
catalyst concentration, the rates of carbohydrate conversion
and furanic production increase with temperature, however,
selectivities toward furanics decrease for both furfural and
1
5
ways leading to humins. Maintaining lower reaction tem-
peratures, therefore, results in a shift in the ratio of direct
dehydration to isomerization pathways in favor of isomeriza-
tion and higher furanic selectivity.
5
-HMF. This is due to differences in the activation energies
for the direct sugar dehydration pathways versus the sugar
isomerization-initiated pathways (see Schemes 1 and 2). For
xylose, the activation energies for direct dehydration to furfu-
ral, isomerization to xylulose, and dehydration of xylulose to
2
.4 Combination of SnCl and a Brønsted acid
4
4
The use of SnCl produces very little quantifiable xylulose or
fructose since the hydrolysis of the Lewis acid catalyst pro-
vides ample Brønsted acidity to catalyze the dehydration of
the ketose carbohydrates at a rate equal to or superior than
the isomerization of the aldose carbohydrates. Thus, the fur-
ther addition of a Brønsted acid would only accelerate the
competing direct carbohydrate dehydration and furanic deg-
Table 3 The initial rates of reaction of A) xylose (750 mM) and B) glucose
(
750 mM) and the initial rates of furfural and 5-HMF formation for differ-
ent SnCl concentrations (25–75 mM) and temperatures of 140 and 160
C. Selectivities toward furanic products are also included. Measurements
4
radation reactions. However, a comparison of the SnCl -cata-
4
°
were taken within the first 10 min of reaction
lyzed isomerization/dehydration of either xylose or glucose
with and without the addition of a small amount of HCl (50
mM) shows that Brønsted acid addition has little impact on
the chemistry (Fig. S1 in ESI†). This is so because when low
A.
Xylose initial Furfural initial
rate of
rate of
Furfural
selectivity,
mol%
4
concentrations are added to solutions of SnCl under 100
Temperature, [SnCl
mM
4
], reaction,
formation,
μmol L
+
−
1
−1
−1 −1
°
C
μmol L
s
s
mM, hydrolytic exchange occurs between the positive H O
3
and the anionic Sn–OH species, evidenced by a broadening
2
50
7
2
50
5
94
56
60
59
61
42
41
44
1
19
(4−y)+ 10
of the
Sn NMR peak attributed to ijSnIJOH)
y
IJH
2
O)6−y
]
.
1
1
40
60
182
279
197
385
585
108
170
83
159
257
5
5
This exchange shifts the equilibrium from hydrolyzed vari-
4+
4+
ants of Sn toward ijSnIJH
2
O)
6
]
while the solution pH
remains approximately the same. As a consequence, the
extent to which this hydrolysis occurs is governed by the solu-
7
5
4
+
B.
tion pH, such that for pH values <2, solvated Sn exists pri-
1
2
Glucose initial 5-HMF initial
rate of rate of
], reaction, μmol formation,
marily as cationic hexaaquatinIJIV) species.
5-HMF
selectivity,
mol%
Temperature, [SnCl
mM
4
−1
−1
−1 −1
2.5 Xylose and glucose isomerization/dehydration in a
°C
L
s
μmol L
s
biphasic system
2
50
7
2
50
5
87
95
103
199
215
230
18
28
32
38
55
61
20
30
31
19
26
27
1
1
40
60
The dehydration of xylose and glucose in the presence of
Brønsted acidity, whether added extrinsically or derived via
Lewis acid cation hydrolysis, is necessary for furanic produc-
tion but also detrimental because it promotes furanic-
5
5
7
5
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
consuming side reactions that lead to the formation of
humins. Furanic consumption can be avoided by removing
furfural and 5-HMF immediately as they are produced. One
way to accomplish this is through the employment of a
biphasic system that utilizes an appropriate solvent capable
of extracting furfural and 5-HMF from the reactive aqueous
phase in situ. Such systems have been observed to be success-
ful in curtailing degradation and improving overall furanics
8
yields. However, identifying a solvent that is serviceable for
the removal of both furfural and 5-HMF is complicated due
to the structural differences in the two molecules affecting
solvation. Instead of performing an exhaustive screening of
organic solvent candidates by physically performing biphasic
extraction experiments, we used the UNIFAC (UNIversal
quasichemical Functional-group Activity Coefficients) method
to rapidly determine activity coefficients for both furfural and
1
6,17
5
-HMF in water and various organic solvents.
This
method has been found to accurately predict activity coeffi-
cients for small molecule solvents and solutes at tempera-
17
tures below 200 °C. After acquiring these activity coeffi-
cients, we calculated the partition coefficient for each solute
between water and a given organic solvent as:
solute
organic





log

K

 log


(2)
solute
water


where K is the partition coefficient and γ is the activity coeffi-
cient of the solute in either the water or organic phase. Parti-
tion coefficient values >1 calculated in such a fashion indi-
cated a preference for the organic phase. After qualitative
comparison of solvent K values, n-butanol was found to strike
the best balance of partition coefficients for both furanics of
interest. The UNIFAC results for both furfural and 5-HMF in
some optimal biphasic systems at 140 °C are shown in Fig.
S2 in the ESI.† For the low mole fractions of furanics
expected under the typical reaction conditions (<0.02), the
partition coefficients for furfural and 5-HMF between
n-butanol and water were found to be ≥ 11.3 and ≥ 3.8,
respectively.
Fig. 4 The SnCl
4
-catalyzed conversion of A) xylose and B) glucose in
a 1 : 2 water : butanol IJv/v) biphasic system. Conditions: 750 mM initial
sugar concentration, 25 mM SnCl
upon the aqueous phase.
4
, 140 °C. All concentrations based
5-HMF degradation to these products as compared to the
reaction in water alone (19% and 12%, respectively, after only
2 h). Including formic acid yields with those of 5-HMF, selec-
tivity to desirable products from glucose increases to 83% in
a biphasic system. Levulinic acid degradation was also lower
in the two-phase system, with only 5% degradation products
(mainly angelica lactone) produced when n-butanol was used
compared to 7% after 2 h with water alone.
Using n-butanol as an extracting agent, we performed the
SnCl -catalyzed isomerization/dehydration of xylose and glu-
4
cose, separately, in a biphasic system consisting of a 1 : 2
ratio of water : butanol (by volume). The results for this exper-
iment are shown in Fig. 4A and B for xylose and glucose,
respectively. Xylose conversion proceeds at a rate similar to
that observed for its reaction in water alone, reaching a con-
version of 90% after 5 h. The furfural selectivity for the
biphasic system was nearly constant at 85% throughout the
reaction, resulting in a yield of 77% after 5 h of reaction. The
initial conversion of glucose in the biphasic system was lower
than in the single-phase system, but reached a conversion of
Despite the employment of a biphasic system, selectivities
to furfural and 5-HMF never reach 100%, consistent with pre-
vious observation. A review of Schemes 1 and 2 reveals that
7
reaction pathways to degradation products not involving
furanics exist, which can consume carbohydrates in the aque-
ous phase regardless of the extraction of furfural/5-HMF and
result in the reduction of apparent product selectivity. These
reactions can be mitigated by controlling the Brønsted acidity
of the aqueous phase, though, this can be difficult to accom-
plish when using hydrolyzable Lewis acid catalysts as in this
study. Also, while n-butanol is a good extracting agent, the
partitioning of either furanic does not reach 100%, resulting
in an observable amount of furanics remaining within the
aqueous phase capable of participating in Brønsted acid-
7
5% after 5 h. The selectivity of 5-HMF tracked the conver-
sion of its glucose and remained nearly constant at 69%,
resulting in a 5-HMF yield of 52% after 5 h. Furthermore,
formic acid (10%) and levulinic acid (5%) production was
reduced in the biphasic system after 5 h, indicating less
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
catalyzed side reactions. The latter issue can be moderated
through the use of aqueous phase modifiers, which is
addressed in the next section.
furfural and to 72% for 5-HMF. The slight increase in carbo-
hydrate conversion is likely due to dissolved LiCl assisting in
the dehydrations of xylulose and fructose, while selectivity
increases toward furanics can be attributed to a minor
enhancement in partitioning from the reactive aqueous
phase to the n-butanol phase. A similar enhancement in the
selectivity to furanics has been reported by Abu-Omar, et al.
2
.6 Effect of LiCl on xylose and glucose isomerization/
dehydration in a biphasic system
The addition of metal halides to aqueous solutions has been
observed to improve the thermodynamics of solute extraction
from aqueous phases into organic phases, with higher metal
who observed that the addition of NaCl to the AlCl -catalyzed
3
dehydration of glucose in a H
improved the selectivity of HMF from 52% to 61%.
The present observations of glucose dehydration in a
biphasic system catalyzed by SnCl in the aqueous phase are
2
O/THF biphasic system
8
a
1
8,19
halide concentrations found to be more effective.
How-
ever, previous work has also shown that, while the addition
of metal halides to the Brønsted acid-catalyzed dehydration
of xylose increases the rate of xylose consumption and furfu-
ral production, product selectivity decreases with increasing
salt concentrations because the metal halide can also pro-
4
similar to those found for glucose dehydration in a water/
6
THF biphasic system catalyzed by Sn-BEA zeolite and HCl.
However, the biphasic Lewis acid-catalyzed xylose conversion
results exceed those typically reported in aqueous media (for
1
5
mote the undesired formation of humins. Therefore, it is
important to select an appropriate metal halide concentra-
tion such that additives can improve the partitioning efficacy
of a biphasic system by reducing water activity and increasing
the activity coefficients of furanics in water without interfer-
ing in the dehydration chemistry. The concentration at which
the metal halide activity coefficient in water is lowest is also
the point at which it is at its lowest chemical potential and
least likely to interact with other solutes, namely carbohy-
drates. LiCl was selected based on earlier work showing this
salt to be an optimal additive for the dehydration of xylose in
example, selectivities of ~80% were found when CrCl was
3
2
used as catalyst).
3
. Conclusion
SnCl4 was investigated as a selective Lewis acid catalyst for
the production of both furfural and 5-HMF from xylose and
glucose, respectively, in aqueous media. At moderate concen-
trations and reaction temperature, SnCl provides the Lewis
4
acidity necessary for isomerization of both xylose to xylulose
and glucose to fructose, as well as the Brønsted acidity neces-
sary to dehydrate both xylulose to furfural and fructose to
5-HMF. The significant Brønsted acidity resulting from
1
9
a biphasic system. The activity coefficient minimum in
water was found to be 0.734 with a water activity of 0.983 for
2
0
a concentration of 500 mM LiCl.
4
hydrolysis of SnCl also catalyzes secondary reactions that
The biphasic experiment described previously was
repeated with the addition of 500 mM LiCl to the aqueous
phase. These results are presented in Fig. 5. Xylose and glu-
cose conversion rates increased slightly, such that 95% of the
xylose and 81% of the glucose were consumed after 5 h. Simi-
larly, the furfural yield reached 84% and the 5-HMF yield
reached 58%, raising the furanic selectivities to 88% for
consume furfural and 5-HMF to produce humins and hydro-
lyze 5-HMF to produce levulinic and formic acid, both pro-
cesses limiting the selectivities to the desired furanic product.
n-Butanol was found to be an appropriate solvent for removing
both furanics from water. Working in a 1 : 1 water : butanol
4
biphasic system with SnCl in the aqueous phase, conver-
sions of both carbohydrates in excess of 75% with selectiv-
ities of 85% toward furfural and 69% toward 5-HMF could be
achieved at 140 °C in 5 h. The addition of LiCl to the aque-
ous phase increased the conversion of xylose to 95% and that
of glucose to 81%, and increased the selectivity to furfural to
88% and that to 5-HMF to 72%.
4. Experimental
4.1 Materials
D-Xylose (99%, Sigma-Aldrich), D-glucose (≥99.5%, Sigma),
D-xylulose (≥98%, Santa Cruz Biotechnology), D-fructose
(
(
5
≥99% Sigma), D-lyxose (99%, Sigma-Aldrich), D-mannose
≥99% Sigma), 2-furaldehyde (furfural, 99%, Sigma-Aldrich),
-hydroxymethylfurfural (5-HMF, 98%, Acros Organics),
levulinic acid (≥97%, Aldrich), formic acid (≥99.5%, Fisher
Chemical), and α-angelica lactone (98%, Sigma-Aldrich) were
used as reagents and standards (for quantification).
Fig. 5 The effects of 500 mM LiCl added to the aqueous phase on the
SnCl
IJv/v) biphasic system. Conditions: 750 mM initial sugar concentration,
5 mM SnCl , 140 °C.
4
-catalyzed conversion of xylose and glucose in a 1 : 2 water : butanol
Aluminum chloride hexahydrate (AlCl3 × 6H O, 99.0%,
2
2
4
crystalline), ceriumIJIII) chloride heptahydrate (CeCl × 7H O,
3
2
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
9
9
9.999%, trace metals basis), chromiumIJII) chloride (CrCl2,
5%, water soluble), chromiumIJIII) chloride hexahydrate
μm, P/N CP8944) connected to a Varian quadrupole-mass
spectrometer (MS) and flame ionization detector (FID) was
used to analyze all organic phase samples. After product
identification by mass spectrometer, product concentrations
were determined from integrated FID peak areas using a
6-point calibration curve generated from purchased
standards: furfural, 5-HMF, levulinic acid, and formic acid.
Reagent and product yields are reported as molar percent-
ages relative to initial molar concentrations of carbohydrate
(i.e. furfural yields = moles furfural/initial moles xylose). All
reported yields were typically reproducible to within a ±5%
relative error (based upon the calculation of one standard
deviation).
(CrCl
3
× 6H
2 2
O, 98%), ironIJII) chloride tetrahydrate (FeCl ×
4
H O, 99.0%), ironIJIII) chloride hexahydrate (FeCl × 6H O,
2
3
2
ACS
trifluoromethanesulfonate IJInIJOTf)
LiCl, BioXtra, ≥99.0% (titration)), scandiumIJIII) triflate IJSc-
IJOTf) , 99%), tinIJII) chloride dihydrate (SnCl × 2H O,
reagent,
98.0–102%,
crystalline),
indiumIJIII)
3
, 99%), lithium chloride
(
3
2
2
9
9.995%, trace metals basis), tinIJIV) chloride pentahydrate
(SnCl4 × 5H O, 98%), tinIJII) trifluoromethanesulfonate IJSn-
2
IJOTf) , 98%), and ytterbiumIJIII) trifluoromethanesulfonate
IJYbIJOTf) , 99.99%) were purchased from Sigma-Aldrich.
3
2
Hydrochloric acid (HCl, 37% v/v, Fisher Scientific) was used
as the Brønsted acid catalyst. n-Butanol (ACS reagent, ≥99%,
Fluka) was used as the organic extracting agent in biphasic
experiments. All materials were used as purchased, without
further purification or modification.
Acknowledgements
This work was supported by the Energy Biosciences Institute
funded by BP. Special thanks for his assistance in conducting
the reported research is given to Daniel Tjandra.
4
.2 Experimental approach
All experiments were performed in 10 ml glass vials (from
Sigma-Aldrich via supplier Supelco) sealed with 20 mm
aluminum-crimped PTFE septa and heated using a silicon oil
bath to maintain constant reaction temperature and stirring
rate. In a representative glucose dehydration experiment, glu-
cose was dissolved in nanopure water, to which SnCl4 and
References
1
(a) M. J. Climent, A. Corma and S. Iborra, Green Chem.,
011, 13, 520–540; (b) G. A. Tompsett, N. Li and G. W.
2
Huber, Catalytic Conversions of Sugars to Fuels, in
Thermochemical Processing of Biomass: Conversion into Fuels,
Chemicals and Power, ed. R. C. Brown, John Wiley & Sons,
Chichester, 2011, pp. 223–279; (c) J. N. Chheda, G. W. Huber
and J. A. Dumesic, Angew. Chem., Int. Ed., 2007, 46(38),
HCl were added to create a solution (750 mM glucose, 25
+
mM SnCl
4
, 50 mM H ). A 4 ml aliquot of this solution was
sealed into a 10 ml glass vial. The vial was then placed in a
silicone oil bath heated to 140 °C and stirred at 600 rpm.
Upon completion of the reaction, the sample was removed
and quenched in an ice bath. An internal standard (1 ml of
7
164–7183; (d) T. Werpy, G. Petersen, A. Aden, J. Bozell, J.
Holladay, J. White and A. Manheim, Report No. NREL/TP-
10–35523. Top Value Added Chemicals from Biomass: Vol. 1
Results of Screening for Potential Candidates from Sugars
5
—
−
1
7
5 mg ml 1,6-hexanediol in water) was added and the sam-
ple centrifuged to remove all water insoluble particulates. A
portion of the reaction mixture (500 μl) was diluted in a 1 : 1
ratio with nanopure water and taken for HPLC analysis. For
reactions involving an additional organic phase, the aqueous
phase volume was reduced to 1 ml and butanol (4 ml) was
added prior to sealing the reaction vial. At reaction end, the
organic phase was separated via centrifuging. The aqueous
phase was treated as above and analyzed via HPLC while a
and Synthesis Gas, National Renewable Energy Laboratory,
Golden, CO, 2004, pp. 1–77; (e) K. J. Zeitsch, Sugar Ser.,
2
000, 13, 1–353.
(a) V. Choudhary, S. Sandler and D. Vlachos, ACS Catal.,
012, 2(9), 2022–2028; (b) V. Choudhary, S. Caratzoulas and
2
3
2
D. Vlachos, Carbohydr. Res., 2013, 368, 89–95.
V. Choudhary, S. H. Mushrif, C. Ho, A. Anderko, V.
Nikolakis, N. S. Marinkovic, A. I. Frenkel, S. I. Sandler and
D. G. Vlachos, J. Am. Chem. Soc., 2013, 135, 3997–4006.
J. B. Binder, J. J. Blank, A. V. Cefali and R. T. Raines,
ChemSusChem, 2010, 3(11), 1268–1272.
(a) R. Weingarten, A. Rodriguez-Beuerman, F. Cao, J. S.
Luterbacher, D. M. Alonso, J. A. Dumesic and G. W. Huber,
ChemCatChem, 2014, 6(8), 2229–2234; (b) E. I. Gürbüz,
J. M. R. Gallo, D. M. Alonso, S. G. Wettstein, W. Y. Lim and
J. A. Dumesic, Angew. Chem., Int. Ed., 2012, 52(4),
1270–1274.
−
1
different internal standard (1 ml of 5 mg ml guaiacol in tol-
uene) was added to the organic phase prior to GC/MS
analysis.
4
5
4
.3 Product analysis
A Shimadzu HPLC equipped with a Phenomenex Rezex RFQ-
+
2 4
Fast Acid H column (100 × 7.8 mm; 0.01 N H SO ; 1.0 ml
−
1
min ; 55 °C) and a refractive index detector (RID) was used
to analyze all aqueous samples. Product quantities were
determined by converting integrated HPLC peak areas into
concentrations using a 7-point calibration curve generated
from purchased standards: xylose, glucose, xylulose, fructose,
lyxose, mannose, furfural, 5-HMF, levulinic acid, and formic acid.
A Varian CP-3800 Gas Chromatograph equipped with a
FactorFour Capillary Column (UF-5 ms 30 m, 0.25 mm, 0.25
6 E. Nikolla, Y. Roman-Leshkov, M. Moliner and M. E. Davis,
ACS Catal., 2011, 1(4), 408–410.
7 (a) R. van Putten, J. C. van der Waal, E. de Jong, C. B.
Rasrendra, H. J. Heeres and J. G. de Vries, Chem. Rev.,
2013, 113, 1499–1597; (b) P. Gallezot, Chem. Soc. Rev.,
2012, 41, 1538–1558.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Catalysis Science & Technology
Paper
8
(a) Y. Yang, C. Hu and M. Abu-Omar, Green Chem.,
012, 14(2), 509–513; (b) Y. J. Pagan-Torres, T. Wang, J. M. R.
Gallo, B. H. Shanks and J. A. Dumesic, ACS Catal.,
012, 2(6), 930–934; (c) Y. Roman-Leshkov and J. A.
14 K. W. Omari, J. E. Besaw and F. M. Kerton, Green Chem.,
2012, 14, 1480–1487.
15 K. R. Enslow and A. T. Bell, ChemCatChem, 2015, 7(3),
479–489.
2
2
Dumesic, Top. Catal., 2009, 52, 297–303.
G. W. Huber, S. Iborra and A. Corma, Chem. Rev.,
16 (a) A. Fredenslund, R. L. Jones and J. M. Prausnitz, AIChE J.,
1975, 21, 1086–1099; (b) T. Magnussen, T. Rasmussen and A.
Fredenslund, Ind. Eng. Chem. Process Des. Dev., 1980, 20,
1269–1277.
17 R. Wittig, J. Lohmann and J. Gmehling, Ind. Eng. Chem. Res.,
2003, 42, 183–188.
18 (a) K. Pitzer and J. Kim, J. Am. Chem. Soc., 1974, 96(18),
5701–5707; (b) F. A. Long and W. F. McDevit, Chem. Rev.,
1952, 51(1), 119–169; (c) R. G. Curtis and H. H. Hatt, Aust. J.
Chem., 1948, 1(2), 213–235; (d) F. Hofmeister, Arch. Exp.
Pathol. Pharmakol., 1888, 24, 247–260.
9
2
006, 106(9), 4044–4098.
0 M. J. Taylor and J. M. Coddington, Polyhedron, 1992, 11(12),
531–1544.
1
1
1
1 K. M. O. Jensen, M. Christensen, P. Juhas, C. Tyrsted, E. D.
Bojesen, N. Lock, S. J. L. Billinge and B. B. Iversen, J. Am.
Chem. Soc., 2012, 134, 6785–6792.
1
1
2 F. Fringuelli, F. Pizzo and L. Vaccaro, J. Org. Chem., 2001, 66,
4
719–4722.
3 (a) N. Wiberg and A. F. Holleman, Inorganic Chemistry,
Academic Press, San Diego, CA, 34th edn, 2001; (b) Z. M.
Jarzevski and J. P. Marton, J. Electrochem. Soc., 1976, 123(7),
19 E. O. Eisen and J. Joffe, J. Chem. Eng. Data, 1966, 11(4),
480–484.
1
99–205; (c) M. Nagasawa, S. Shionoya and S. Makishima,
20 W. J. Hamer and Y. C. Wu, J. Phys. Chem. Ref. Data, 1972, 1,
1047–1100.
Jpn. J. Appl. Phys., 1965, 4(3), 195–202.
This journal is © The Royal Society of Chemistry 2015
Catal. Sci. Technol.