Solvent basicity controlled deformylation for the formation of furfural from glucose and fructose

Article abstract of DOI:10.1039/c9gc02600b

The conversion of glucose and fructose to furfural can improve the productivity of furfural synthesis from biomass. We report that sulfolane, a weakly basic polar aprotic solvent, promoted the formation of furfural from hexoses with a maximum yield of 50% obtained from fructose. Addition of Sn/SBA-15 that acted as a Lewis acid catalyst enabled the conversion of glucose to furfural with 36% yield. Analysis of products obtained from isotopically labelled glucose showed that furfural is produced by elimination of the C6 carbon atom as a formaldehyde molecule. DFT calculations revealed that this elimination reaction is plausible in the presence of weakly basic solvents that are unable to abstract the proton from the C-H bond in the last step of the reaction, which would otherwise lead to formation of 5-HMF. The furfural yield was correlated with the basicity of solvents, calculated as proton affinity (E_pa), confirming the hypothesis that the basicity of solvent determines the selectivity for the formation of furfural or 5-HMF. Hence furfural formation from hexoses can be achieved by acid catalysed reaction of hexoses in the presence of low basicity polar aprotic solvents.

Full text of DOI:10.1039/c9gc02600b

Green Chemistry
View Article Online
View Journal | View Issue
PAPER
Solvent basicity controlled deformylation for the
formation of furfural from glucose and fructose†
Cite this: Green Chem., 2019, 21,
146
6
a,b
a
a
Miyuki Asakawa, Abhijit Shrotri, * Hirokazu Kobayashi
and
a
Atsushi Fukuoka
*
The conversion of glucose and fructose to furfural can improve the productivity of furfural synthesis from
biomass. We report that sulfolane, a weakly basic polar aprotic solvent, promoted the formation of furfural
from hexoses with a maximum yield of 50% obtained from fructose. Addition of Sn/SBA-15 that acted as a
Lewis acid catalyst enabled the conversion of glucose to furfural with 36% yield. Analysis of products
obtained from isotopically labelled glucose showed that furfural is produced by elimination of the C6
carbon atom as a formaldehyde molecule. DFT calculations revealed that this elimination reaction is
plausible in the presence of weakly basic solvents that are unable to abstract the proton from the C–H
bond in the last step of the reaction, which would otherwise lead to formation of 5-HMF. The furfural
yield was correlated with the basicity of solvents, calculated as proton affinity (Epa), confirming the
hypothesis that the basicity of solvent determines the selectivity for the formation of furfural or 5-HMF.
Hence furfural formation from hexoses can be achieved by acid catalysed reaction of hexoses in the pres-
ence of low basicity polar aprotic solvents.
Received 26th July 2019,
Accepted 18th October 2019
DOI: 10.1039/c9gc02600b
rsc.li/greenchem
furfural from hexoses can overcome this problem by providing
a single product stream from the total carbohydrate content of
Introduction
Lignocellulosic biomass, a renewable carbon resource, is pri- biomass, which accounts for 50–80% of biomass by weight.
marily composed of carbohydrates such as cellulose and This approach is attractive as it would maximize the yield of
1
–3
hemicellulose. Conversion of these carbohydrates to furanic furfural and simplify downstream processes. Furthermore,
compounds is an important step in producing value added hexoses are abundantly available in the form of starch and
4
–8
chemicals such as polymers, fuels, resins, and solvents.
In molasses, which can be utilized for furfural synthesis without
this step, monomeric pentoses are converted to furfural and relying on the hemicellulose fraction.
hexoses are converted to 5-hydroxymethylfurfural (5-HMF), by
The acid catalysed reaction of glucose and fructose to
a combination of isomerization and dehydration reactions 5-HMF in water and organic solvents frequently produces fur-
9
,10
19,20
(Fig. 1).
Furfural is currently produced industrially by de- fural as a minor by-product.
Pyrolysis of glucose and fruc-
hydration of pentose (mainly xylose), obtained from hemi- tose in the absence of any solvent also yields furfural in
1
1
21,22
cellulose, in the presence of a homogeneous acid catalyst.
appreciable amounts.
amounts even during the hydrolysis of cellulose to glucose.
SAPO-44, and beta-zeolite, have been reported These results suggest that formation of furfural from hexoses
to be active for conversion of xylose to furfural. However, com- and its polymers is an inevitable side reaction, which can be
Furfural is obtained in very small
2
−
12
13
23
Heterogeneous catalysts such as SO4 /SnO2,
Nafion,
1
4,15
16
17
2 5
Nb O ,
mercial application of these catalysts is not yet viable.
promoted to obtain furfural as the primary product.
One of the drawbacks of furfural synthesis is the low abun-
Higher yields of furfural have been achieved in the presence
2
4,25
dance of xylose in biomass leading to low yield of furfural of polar aprotic solvents.
The presence of an acid catalyst
1
8
(
4–12%) per unit weight of biomass processed. Synthesis of such as zeolite was essential for the furfural formation from
2
6
hexoses during these reactions. Catalysts with Lewis acid
sites (for example Sn-beta zeolite) were required to catalyse the
a
25
Institute for Catalysis, Hokkaido University, Kita 21 Nishi 10, Kita-ku, Sapporo,
isomerization of glucose to fructose. Sn-Beta zeolite is a well-
Hokkaido 001-0021, Japan. E-mail: ashrotri@cat.hokudai.ac.jp,
fukuoka@cat.hokudai.ac.jp
known catalyst for the isomerization of glucose to fructose and
its subsequent dehydration to 5-HMF. However, its contri-
bution towards formation of furfural beyond isomerization is
not clearly understood. The choice of solvent was also crucial
in all previous studies. Polar aprotic solvents such as
27
b
Graduate School of Chemical Sciences and Engineering, Hokkaido University,
Kita 13 Nishi 8, Kita-ku, Sapporo, Hokkaido 060-8628, Japan
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
c9gc02600b
6146 | Green Chem., 2019, 21, 6146–6153
This journal is © The Royal Society of Chemistry 2019

View Article Online
Green Chemistry
Paper
Fig. 1 Established reactions of glucose and xylose to 5-HMF and furfural, respectively (solid arrows) and the proposed reaction for producing fur-
fural from hexoses to increase overall furfural productivity from biomass (dashed arrow).
γ-valerolactone and γ-butyrolactone were especially active for
furfural synthesis. These solvents were predicted to enhance
Materials and methods
Catalyst preparation
2
8
the formation of acyclic fructose that would easily diffuse
2
9
through the pores of zeolite. However, the formation of fur- SBA-15 was prepared according to method reported by Stucky
3
4
fural in the presence of catalysts other than zeolites does not et al. During synthesis, 450 mL of 1.6 M hydrochloric acid
2
4
fit this hypothesis.
was added to 12 g of poly(ethylene glycol)-block-poly(propylene
The reaction pathway for furfural formation from hexoses glycol)-block-poly(ethylene glycol) (P123) and stirred at room
and the associated mechanism is not yet understood. Earlier temperature until the P123 was hydrolysed. The solution was
studies focusing on 5-HMF synthesis from fructose have con- stirred rapidly at 35 °C and 25.5 g of tetraethyl orthosilicate
cluded that furfural is not formed via 5-HMF and it was the was added dropwise over a span of 30 minutes. The solution
3
0,31
result of a separate pathway.
Recently, it was predicted that was kept standing at 35 °C for 24 h and then kept standing at
pentose sugars such as xylose or arabinose are formed via 100 °C for another 24 h. The obtained solid was filtered and
retro-aldol cleavage of the C–C bonds of an acyclic fructose washed with water and ethanol until the presence of chloride
2
6,29
molecule.
Subsequent dehydration of pentoses on the ions was not detected in the filtrate. The solid was dried
zeolitic acid sites was proposed as the pathway for furfural at 110 °C overnight and then calcined at 560 °C for 16 h to
formation. However, retro-aldol reaction of fructose yields obtain SBA-15.
3
2
glyceraldehyde and dihydroxyacetone, and the formation
Sn/SBA-15 catalyst was prepared by impregnation method.
·5H O, corresponding to 1 wt% Sn in
of xylose is not viable through this pathway. Another study Desired amount of SnCl
4
2
experimentally deduced that the C6 carbon atom on the the catalyst, was dissolved in 5 mL of distilled water and then
fructose molecule was eliminated when furfural was formed 500 mg of SBA-15 was added to the solution. The mixture was
3
3
as a byproduct.
However, no attempts were made to sonicated for 5 minutes and then stirred continuously over a
deduce the C–C cleavage mechanism. Consequently, the hot plate maintained at 105 °C until a powder was obtained.
mechanism of furfural formation from fructose remains After further drying at 110 °C for 2 h, the powder was calcined
unclear. The elucidation of mechanism for this reaction is at 500 °C for 2 h.
necessary to design catalysts and processes for furfural
synthesis from hexoses.
Bulk SnO catalyst for comparison was prepared by dissol-
2
4 2
ving 3.5 g SnCl ·5H O in 100 mL water and then adding 28%
In this study, we explore the mechanism of furfural for- aqueous NH
3
solution until the pH was neutral. The precipi-
mation from glucose and fructose in detail and examine the tated solid was washed with water and then ethanol. The
role of solvents and catalytically active species. An Sn/SBA-15 washed solid was dried at 110 °C and then calcined at 500 °C
catalyst was used as a Lewis acid catalyst in the presence of for 2 h.
polar aprotic solvents. We establish the reaction pathway for
1
3
Catalyst characterization
formation of furfural by using C labelled sugar molecules
1
3
and evaluating the products by C NMR and mass spec- X-ray diffraction (XRD) was measured with Rigaku MiniFlex
trometry. Density functional theory (DFT) calculations were using CuKα X-ray (λ = 1.54 Å). UV visible diffuse reflectance
used to determine the influence of solvent in the reaction spectroscopy (UV-vis) measurement was done using Jasco
mechanism. Based on these experimental and theoretical data V-650 spectrophotometer. Transmission electron microscope
we propose a mechanism for furfural formation.
(TEM) images were taken with a JEM-2100F field emission
This journal is © The Royal Society of Chemistry 2019
Green Chem., 2019, 21, 6146–6153 | 6147

View Article Online
Paper
Green Chemistry
+
+
transmission electron microscope at an accelerating voltage of defined as Epa = −ΔH° = [H°(M − H ) − H°(M) − H°(H )].
2
00 kV. The sample was prepared by dispersing a small Standard enthalpy, H° = Eelectronic + Evib + Etrans + Erot + p°V, of
amount of catalyst in ethanol and then dropping it on a each structure was determined with the harmonic vibration
+
carbon-coated copper grid. N
2
adsorption isotherms were calculations except for proton. H°(H ) at 298 K was approxi-
measured at −196 °C using a Belsorp mini analyser. NH3 mated to 5/2RT = 6.19 kJ mol−
temperature experiments were performed in at BELCAT-A
instrument.
1
.
Results and discussion
Catalytic reaction
Catalytic reactions were performed in a pressure resistant glass Several polar aprotic solvents were first tested for conversion of
reaction vessel (Ace Glass). In a typical reaction, 0.22 mmol of fructose to furfural at 160 °C. The formation of furfural or
substrate (glucose or fructose), 5 mL of solvent and catalyst 5-HMF from fructose was highly dependent on the solvent
were added to the reaction vessel and stirred in oil bath at used. A reaction of fructose in the presence of sulfolane
1
60 °C for 2 h. After the reaction, the solution was cooled to without the addition of any catalyst produced 50% yield of fur-
room temperature and 5 mL of distilled water was added. The fural and 7.2% of 5-HMF (Table 1). In contrast, when dimethyl
mixture was then analysed using high performance liquid sulfoxide (DMSO) was used as the solvent, 5-HMF was the
chromatography (HPLC) (Shodex SH-1011 column and Rezex primary product with a yield of 69% under the same reaction
RPM monosaccharide Pb++ column) equipped with a refractive conditions. Moreover, furfural yield was only 2.9% in the pres-
index detector. Product yields were calculated by using the fol- ence of DMSO. Tetramethylene-sulfoxide, also produced
lowing equation.
5-HMF as the major product (56%) with low yield of furfural
(
0.4%). These results indicated that the nature of solvent
moles of product obtained
moles of substrate added
Yield of products ¼
ꢀ 100
altered the selectivity of products formed during the reaction.
Other solvents such as γ-valerolactone, and γ-butyrolactone,
previously reported to be good for furfural synthesis, produced
furfural but with low yields of 3.2% and 8.7%, respectively.
Addition of a small amount of phosphoric acid (1.5 mM in
the reaction mixture) as a Brønsted acid catalyst enhanced the
rate of furfural formation in sulfolane (Fig. S1†). Final furfural
yield of 50% was obtained in the presence of acid catalyst in
about 30 minutes (Table 1, entry 6). The addition of acid cata-
lyst also led to increased amount of furfural formation in
γ-valerolactone (7.6%) and γ-butyrolactone (18%). It was
Gram scale experiment was performed using 8 g of fructose
in 50 mL of sulfolane containing 1.5 mM H PO . The reaction
3
4
was performed at 160 °C for 45 minutes. After the reaction,
00 mL of water was added to the solution and the mixture
was distilled at 100 °C and 500 hPa. The resulting aqueous
solution was extracted twice using dichloromethane, which
was subsequently evaporated to obtain the product.
2
Isotopic tracer study
1
3
Glucose and fructose labelled with C at different position evident that these solvents can also produce furfural as
were used for the isotopic tracer study. After reaction with demonstrated in previous reports when a suitable acid catalyst
the isotopically labelled samples, products were analysed is present. The formation of both furfural and 5-HMF requires
using C NMR and gas chromatography-mass spectrometry removal of hydroxyl groups via the dehydration reaction, which
GC-MS). NMR analysis was performed using a JNM-ECX600, will be accelerated by the addition of a Brønsted acid catalyst.
1
3
(
JEOL instrument operating at magnetic strength of 600 MHz. The acceleration of furfural formation suggests that a common
Prior to NMR analysis, 0.5 mL of product solution was added intermediate may be present for both the reactions, which is
to equal amount of D O containing 4,4-dimethyl-4-silapen- formed by the dehydration of fructose. The possibility of fur-
2
3
8
tane-1-sulfonic acid (DSS) as reference and the resulting fural formation from 5-HMF was discarded because a reac-
mixture was filtered. GC-MS analysis was done using a tion with 5-HMF as the substrate did not produce any detect-
SHIMADZU GCMS-QP2010 equipped with an ULBON HR-20M able amount of furfural in sulfolane (Table 1, entry 14).
column, 0.25 mm × 25 m.
The conversion of glucose to furfural and 5-HMF required
the presence of a Lewis acid catalyst. We synthesized SBA-15
supported Sn catalyst for this reaction. The impregnation of Sn
DFT calculation
All the DFT calculations were performed with the Gaussian 09 on the SBA-15 did not alter the structure of support as
program or Gaussian 16 with the G09Defaults option. measured by N
2
adsorption isotherm (Fig. S2-a†) and the
2
−1
2
Chemical structure of molecules (M) and their protonated surface area reduced only slightly from 876 m g to 851 m
+
−1
forms (M − H ) were optimized using the hybrid functional
g
. XRD pattern of the prepared catalyst showed only one
3
5
36
B3LYP with the split-valence triple-ζ basis set 6-311G(d,p).
In the reaction analysis, solvent effect was included by self-con- phous silica in SBA-15 (Fig. 2a). Diffraction for SnO
broad peak centred at 25 degree 2θ corresponding to amor-
crystals
2
sistent reaction field (SCRF) method with a polarized conti- were not observed suggesting high dispersion of Sn species.
nuum model (PCM) and dielectric constant (32.74), radius The high resolution TEM image of the catalyst did not show
−
1
(
3.35 Å), and density (1.179 g cm ) for sulfolane at a tempera- any Sn particles further confirming high dispersion of the tin
3
7
ture similar to that in our reaction (125 °C). Proton affinity is species (Fig. S3†). The absorption edge for the Sn/SBA-15 cata-
6148 | Green Chem., 2019, 21, 6146–6153
This journal is © The Royal Society of Chemistry 2019

View Article Online
Green Chemistry
Paper
a
Table 1 Yields of furfural and 5-HMF from hexoses in the presence of various polar aprotic solvents
Product yield (%)
Entry
Substrate
Fructose
Solvent
Catalyst
Time (min)
Furfural
5-HMF
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
Sulfolane
DMSO
—
—
—
—
120
120
120
120
120
30
30
30
30
30
30
30
45
120
120
120
50
2.9
0.4
3.2
8.7
50
11
23
7.6
18
2.1
0.4
22
0
7.2
69
56
Tetramethylene-sulfoxide
γ-Valerolactone
γ-Butyrolactone
Sulfolane
Diethyl sulfone
3-Methyl sulfolane
γ-Valerolactone
γ-Butyrolactone
DMSO
Tetramethylene-sulfoxide
Sulfolane
Sulfolane
Sulfolane
DMSO
c
—
—
0.5
16
14
11
b
3
H PO
3
H PO
3
H PO
3
H PO
3
H PO
3
H PO
3
H PO
3
H PO
4
4
4
4
4
4
4
4
b
b
b
b
b
b
b
c
—
1
1
1
1
1
1
1
15
77
73
21
—
8.2
11
d
5-HMF
Glucose
—
e
e
Sn/SBA-15
Sn/SBA-15
36
1.2
a
b
c
Reaction conditions: 0.22 mmol substrate in 5 mL of solvent, 160 °C, atmospheric pressure. 1.5 mM H
PO
3 4
in solvent. 5-HMF yield could not
d
be determined due to overlapping peak of γ-valerolactone in the HPLC chromatograms. Large-scale experiment with high fructose concen-
tration: 8 g fructose in 50 mL of sulfolane. 20 mg of 1 wt% Sn/SBA-15 was added during the reaction.
e
same conditions produced 36% furfural from glucose in sulfo-
lane (Table 1, entry 15). However, fructose was not directly
observed as an intermediate in the HPLC chromatogram
owing to its low concentration and multiple overlapping
peaks. In our study, the formation of fructose as an intermedi-
ate for furfural synthesis from glucose was confirmed by using
1
3
13
C-1 and C-6 labelled glucose and analysing the product
1
3
solution by C NMR analysis (Fig. S4 and S5†). The NMR spec-
trum after 5 min of reaction showed peaks that were assigned
1
3
to the furanic form of fructose at 68.2 ppm in C-1 NMR spec-
13
trum and at 66.9–67.3 ppm in the C-6 NMR spectrum. The
activity of Sn/SBA-15 catalyst for isomerization of glucose to
fructose in sulfolane was further confirmed by performing the
reaction at lower temperature (120 °C) at which detectable
amount of fructose was present (8.8%, 5 min). Furthermore,
the Sn/SBA-15 catalyst showed excellent activity for the isomeri-
Fig. 2 (a) XRD spectra of SBA-15 support, Sn/SBA-15 catalyst and bulk
SnO . (b) UV-Vis diffuse reflectance spectra for Sn/SBA-15 catalyst and
bulk SnO
2
2
.
lyst in the UV-Vis diffuse reflectance spectra was at 210 and zation of glucose to fructose in ethanol with 85% selectivity
33 nm indicating the presence of tetrahedrally coordinated and 66% yield (90 °C 2 h) attesting its activity as an isomeriza-
Sn species and its hydrated analogue (Fig. 2b). These Sn tion catalyst.
species are known to exhibit Lewis acidity that is capable of To understand the mechanism of furfural formation from
catalysing the isomerization of glucose. Bulk SnO showed fructose, or from glucose via fructose, it is essential to first elu-
absorbance band at 280 nm which corresponds to octahedrally cidate the pathway responsible for the elimination of one
2
2
1
3
coordinated Sn species that are not active as Lewis acid sites. carbon atom. We used isotopically labelled C-1-glucose and
1
3
NH
the catalyst. SBA-15 support had very low acid density as fate of carbon atoms. The product solution obtained after reac-
shown by the absence of NH desorption peaks (Fig. S2-b†). tion was analysed using GC-MS and the mass spectra of fur-
3
TPD experiment was performed to measure the acidity of
C-6-glucose as substrate for furfural synthesis to track the
3
Presence of Sn on the catalyst surface resulted in incorporation fural were compared (Fig. 3). Furfural with an additional
of 0.141 mmol g− of acid sites. These acid sites were associ- atomic mass (m/z = 97) was observed when C-1 labelled
1
13
ated with both Brønsted and Lewis acidic species formed on glucose was used. The furan fragment appeared in the mass
1
3
the catalyst.
spectrum at an m/z = 67 indicating that the C carbon atom
Reaction of glucose in sulfolane without any catalyst did was not present within the furan ring. This result confirmed
1
3
not yield furfural or 5-HMF and produced levoglucosan and that the C carbon atom was present at the C1 position of the
anhydroglucofuranose, resulting from dehydration of glucose, furfural molecule. Normal furfural with an m/z = 96 was
1
3
as the main products. Addition of Sn/SBA-15 catalyst under the obtained when
C-6-glucose was used as the substrate
This journal is © The Royal Society of Chemistry 2019
Green Chem., 2019, 21, 6146–6153 | 6149

View Article Online
Paper
Green Chemistry
Based on the stoichiometry of the reaction the elimination
of C6 carbon atom from fructose is expected to produce a for-
maldehyde molecule. Detection of formaldehyde by chromato-
graphy was difficult due to its tendency to react at high tem-
peratures. However, when the product of reaction with
1
3
13
C-6 glucose was analysed by C NMR, a small distinct peak
appeared at 84.1 ppm for methylene glycol, the hydrated form
of formaldehyde, confirming its formation (Fig. S5,† Inset of
3
9
30 min NMR spectrum).
To elucidate the mechanism of furfural formation we need
to first examine the formation of 5-HMF from hexose, which is
a well-studied reaction. The mechanism of its formation has
been described in detail using experimental and compu-
4
0,41
tational methods.
In DMSO, 5-HMF is formed by sequen-
tial dehydration of fructose catalysed by an acid catalyst start-
ing from the C2 carbon atom of fructose. After removal of –OH
groups from the C2 and C3 positions the intermediate com-
pound 2 is formed (Fig. 4). The penultimate step in 5-HMF for-
mation is the removal of the water molecule to produce com-
pound 3 by protonation of hydroxyl group at the C4 position.
Finally, the compound 3 loses the proton attached to the C5
carbon atom to DMSO and the formation of furan ring is com-
pleted. The formation of the five membered π conjugated
furan ring drives the last step of the reaction.
Furfural formation is also expected to follow the same path
initially as the addition of a small amount of acid accelerated
the furfural formation. Furthermore, addition of a small
3
amount of base (20 mM NaHCO ) completely blocked the
forward reaction of fructose in both sulfolane and DMSO.
Therefore, we predict that the furfural formation would also
proceed via compound 3. Based on the isotope tracer experi-
ments, we expect that the removal of hydroxymethyl group at
Fig. 3 GC-MS spectra of furfural obtained after reaction using (a)
13
13
C-1-glucose and (b) C-6-glucose as substrate. Reaction conditions: C6 position from the compound 3 would produce furfural and
Substrate 0.22 mmol, Sn/SBA-15 catalyst 20 mg, 5 mL sulfolane, 160 °C, formaldehyde. Thus, the selectivity of furfural/5-HMF is deter-
2
h.
mined by competition of the deformylation at C6 position and
the proton abstraction from C–H at C5 position. Here, unlike
DMSO, sulfolane lacks the basicity required to dissociate C–H
group at the C5 position. Consequently, the deformylation
(Fig. 3b). Therefore, the C6 carbon atom was eliminated in the reaction is dominant (Fig. 4). Elimination of the hydroxy-
synthesis of furfural from glucose/fructose. This observation methyl group leads to the formation of a stable furan ring that
contradicts the mechanism proposed in previous studies that promotes the forward reaction.
claim the elimination of C1 carbon atom by retro-aldol elimin-
ation reaction of acyclic fructose to produce xylose.
To theoretically confirm if the reaction can take place by
this mechanism, we evaluated the energy profile for the defor-
2
5,29
Fig. 4 Proposed mechanism for furfural formation in comparison with mechanism for formation of 5-HMF.
6150 | Green Chem., 2019, 21, 6146–6153
This journal is © The Royal Society of Chemistry 2019

View Article Online
Green Chemistry
Paper
Fig. 6 Yield of furfural from fructose compared with the basicity of sol-
vents calculated as the proton affinity (Epa). Reaction condition: Fructose
Fig. 5 DFT calculations for the deformylation of the compound 3 at the
B3LYP/6-311G(d,p) level. E indicates sum of electronic and zero-point
vibration energy.
0
3 4
.22 mg, solvent 5 mL containing 1.5 mM H PO , 160 °C, 30 min.
Tetrahydrothiophene-1-oxide, which has a similar structure to
sulfolane albite with a sulfoxide group instead of the sulfonyl
group, was even more basic (Epa = 924 kJ mol− ) than DMSO
and produced almost no furfural as expected.
Based on the above mechanism sulfolane acts just as a
solvent without any catalytic influence. To test this phenom-
enon, we performed experiment with solvent mixture contain-
ing different concentration of DMSO in sulfolane. Furfural was
obtained in high yield (50%) only in the absence of DMSO
mylation by the DFT calculations at the B3LYP/6-311G(d,p)
level of theory (Fig. 5). Compound 3 hydrogen bonded with a
sulfolane molecule was set as the initial structure for the calcu-
lation. In the transition state, the C5–C6 bond was elongated
owing to the charge transfer from oxygen atom in the hydroxy-
methyl group to the oxocarbenium in the five-membered ring.
The activation energy for this step was calculated to be a low
1
−
1
value of 21 kJ mol . An imaginary vibration frequency (304i
−
1
cm ) and smooth connection between substrate and product
in intrinsic reaction coordinate (IRC) were confirmed. This
reaction produced formaldehyde in a protonated form, and
finally the proton was subtracted by sulfolane due to very weak
basicity of the solvent molecule. Accordingly, the DFT calcu-
lations indicated that this reaction is plausible.
(
Fig. 7). 5-HMF was obtained in 66% yield with 12% yield of
furfural even when the amount of DMSO was 5 vol%. Further
increase in DMSO concentration reduced furfural yield. These
results show that the formation of 5-HMF was catalysed by the
This mechanism suggests that polar aprotic solvents with
very low basicity can promote the formation of furfural and
moderately basic solvents will promote the formation of
5
-HMF. To validate this assumption, we compared the gas
phase proton affinity of the solvent molecules determined by
the DFT calculations and the experimental furfural yield
obtained from fructose in the presence of small amount of
H PO (Fig. 6). Prior to this, we verified that the calculations
3
4
for proton affinity are accurate using several molecules for
which experimental values are available, e.g., (Epa/kJ mol−
DMSO calc. 888, exp. 884; γ-butyrolactone calc. 839, exp. 840.
1
)
4
2
Consequently we found that the yield of furfural was inversely
correlated to the proton affinity of the solvents. Sulfolane
showed the lowest proton affinity among all solvents used in
−1
this study (Epa = 835 kJ mol ), which corresponds to its
highest furfural yield. The highest proton affinity for which
furfural was the main product was that of γ-valerolactone
(E
pa = 853 kJ mol− ). DMSO (Epa = 888 kJ mol ) was more
1
−1
Fig. 7 Yield of products after reaction of fructose in a solvent mixture
with varying concentration of DMSO in sulfolane. Reaction condition:
basic and produced only a small amount of furfural with high
5
-HMF yield, indicating a transition in the reaction selectivity. Fructose 0.22 mg, solvent 5 mL, 160 °C, 2 h.
This journal is © The Royal Society of Chemistry 2019
Green Chem., 2019, 21, 6146–6153 | 6151

View Article Online
Paper
Green Chemistry
presence of small amount of DMSO in sulfolane and furfural
was formed only in the presence of a low basicity polar aprotic
solvent.
To establish the industrial viability of this process we per-
formed a large-scale reaction with high concentration of fruc-
tose (8 g in 50 mL sulfolane). The furfural yield reduced to
References
1 D. L. Klass, Biomass for renewable energy, fuels, and chemi-
cals, Academic Press, San Diego, 1998.
2 N. Mosier, C. Wyman, B. Dale, R. Elander, Y. Y. Lee,
M. Holtzapple and M. Ladisch, Bioresour. Technol., 2005,
96, 673–686.
2
2% and the 5-HMF yield increased to 21% (Table 1, entry 13).
The reduction in furfural yield was caused by accumulation of
water released during the dehydration step of the reaction,
which favoured the formation of 5-HMF over furfural. Work up
of the reaction mixture gave 0.83 g of product which was ana-
lysed by H and C NMR (Fig. S6†). The resulting product con-
tained 92% furfural along with sulfolane (6.9%). Therefore, we
established that furfural can be produced and separated from
hexose sugars in sulfolane with relatively good purity by
simple distillation and extraction. Furthermore, separation of
product by simple steps along with use of heterogeneous cata-
lyst will enable the design of continuous process for commer-
cial application.
3 G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006,
106, 4044–4098.
4 A. Corma, S. Iborra and A. Velty, Chem. Rev., 2007, 107,
2411–2502.
5 J. B. Binder and R. T. Raines, J. Am. Chem. Soc., 2009, 131,
1979–1985.
6 D. M. Alonso, J. Q. Bond and J. A. Dumesic, Green Chem.,
2010, 12, 1493–1513.
7 P. Gallezot, Chem. Soc. Rev., 2012, 41, 1538–1558.
8 H. Kobayashi, M. Yabushita, T. Komanoya, K. Hara,
I. Fujita and A. Fukuoka, ACS Catal., 2013, 3, 581–587.
9 J. P. Lange, E. Van Der Heide, J. Van Buijtenen and
R. Price, ChemSusChem, 2012, 5, 150–166.
1
13
1
0 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.
Conclusion
1
1 R. Mariscal, P. Maireles-Torres, M. Ojeda, I. Sádaba and
M. López Granados, Energy Environ. Sci., 2016, 9, 1144–
Furfural and 5-HMF yields were found to be influenced by the
type of polar aprotic solvent used. The highest yield of furfural
1189.
(50%) was achieved from fructose in the presence of sulfolane.
1
1
2 T. Suzuki, T. Yokoi, R. Otomo, J. N. Kondo and T. Tatsumi,
Appl. Catal., A, 2011, 408, 117–124.
3 E. Lam, E. Majid, A. C. W. Leung, J. H. Chong,
K. A. Mahmoud and J. H. T. Luong, ChemSusChem, 2011, 4,
In contrast, reaction in DMSO produced 5-HMF as the primary
product with low furfural yield. Sn/SBA-15, a Lewis acid cata-
lyst, was necessary to produce furfural from glucose with a
maximum yield of 36%. Mechanistic study revealed that the
formation of furfural is similar to that of 5-HMF until the final
step leading to furan ring formation. Furfural was produced by
the elimination of the hydroxymethyl groups as formaldehyde.
The formation of furfural was preferred in the presence of
polar aprotic solvents with very low basicity as determined by
calculating the gas phase proton affinity of solvents. Sulfolane,
with the lowest proton affinity of 835 kJ mol , afforded the
highest furfural yield. Our results demonstrate the crucial role
of solvent in the formation of furfural from hexose sugars and
also indicates the importance of using basic additives during
the synthesis of 5-HMF.
535–541.
1
4 C. García-Sancho, I. Sádaba, R. Moreno-Tost,
J. Mérida-Robles, J. Santamaría-González, M. López-
Granados and P. Maireles-Torres, ChemSusChem, 2013,
6
, 635–642.
5 N. K. Gupta, A. Fukuoka and K. Nakajima, ACS Catal.,
017, 7, 2430–2436.
6 P. Bhaumik and P. L. Dhepe, RSC Adv., 2014, 4, 26215–
6221.
7 R. Otomo, T. Tatsumi and T. Yokoi, Catal. Sci. Technol.,
015, 5, 4001–4007.
1
1
1
2
−1
2
2
1
1
8 D. Tin Win, AU J. Technol., 2005, 8, 185–190.
9 G. C. A. Luijkx, F. van Rantwijk, H. van Bekkum and
M. J. Antal, Carbohydr. Res., 1995, 272, 191–202.
2
2
2
0 F. S. Asghari and H. Yoshida, Ind. Eng. Chem. Res., 2006,
Conflicts of interest
45, 2163–2173.
1 J. B. Paine, Y. B. Pithawalla and J. D. Naworal, J. Anal. Appl.
Pyrolysis, 2008, 83, 37–63.
2 D. Carnevali, O. Guévremont, M. G. Rigamonti, M. Stucchi,
F. Cavani and G. S. Patience, ACS Sustainable Chem. Eng.,
There are no conflicts to declare.
2
018, 6, 5580–5587.
Acknowledgements
2
3 A. Shrotri, H. Kobayashi and A. Fukuoka, ChemCatChem,
The authors would like to thank Dr Sho Yamaguchi and
2016, 8, 1059–1064.
Dr Shinji Inagaki from Toyota Central R&D Laboratories, Japan 24 E. I. Gürbüz, J. M. R. Gallo, D. M. Alonso, S. G. Wettstein,
for fruitful discussion on the mechanistic aspects of this
research.
W. Y. Lim and J. A. Dumesic, Angew. Chem., Int. Ed., 2013,
52, 1270–1274.
6152 | Green Chem., 2019, 21, 6146–6153
This journal is © The Royal Society of Chemistry 2019

View Article Online
Green Chemistry
Paper
2
2
2
2
2
3
5 L. Zhang, G. Xi, Z. Chen, D. Jiang, H. Yu and X. Wang, 33 J. Zhang and E. Weitz, ACS Catal., 2012, 2, 1211–1218.
Chem. Eng. J., 2017, 307, 868–876.
6 J. Cui, J. Tan, T. Deng, X. Cui, Y. Zhu and Y. Li, Green
Chem., 2016, 18, 1619–1624.
34 D. Zhao, Q. Huo, J. Feng, B. F. Chmelka and G. D. Stucky,
J. Am. Chem. Soc., 1998, 120, 6024–6036.
35 A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
7 E. Nikolla, Y. Román-Leshkov, M. Moliner and M. E. Davis, 36 W. J. Hehre, R. Ditchfield and J. A. Pople, J. Chem. Phys.,
ACS Catal., 2011, 1, 408–410. 1972, 56, 2257–2261.
8 L. Zhang, G. Xi, Z. Chen, D. Jiang, H. Yu and X. Wang, 37 J. F. Casteel and P. G. Sears, J. Chem. Eng. Data, 1974, 19,
Chem. Eng. J., 2017, 307, 868–876. 196–200.
9 L. Wang, H. Guo, Q. Xie, J. Wang, B. Hou, L. Jia, J. Cui and 38 F. Jin and H. Enomoto, Energy Environ. Sci., 2011, 4, 382–
D. Li, Appl. Catal., A, 2018, 572, 51–60. 397.
0 R. K. M. R. Kallury, C. Ambidge, T. T. Tidwell, 39 I. Hahnenstein, H. Hasse, C. G. Kreiter and G. Maurer, Ind.
D. G. B. Boocock, F. A. Agblevor and D. J. Stewart,
Carbohydr. Res., 1986, 158, 253–261.
1 T. M. Aida, Y. Sato, M. Watanabe, K. Tajima, T. Nonaka,
Eng. Chem. Res., 1994, 33, 1022–1029.
40 A. S. Amarasekara, L. T. D. Williams and C. C. Ebede,
Carbohydr. Res., 2008, 343, 3021–3024.
3
H. Hattori and K. Arai, J. Supercrit. Fluids, 2007, 40, 381– 41 G. Yang, E. A. Pidko and E. J. M. Hensen, J. Catal., 2012,
88. 295, 122–132.
2 M. Orazov and M. E. Davis, Proc. Natl. Acad. Sci. U. S. A., 42 R. W. Taft, in Progress in Physical Organic Chemistry, John
015, 112, 11777–11782. Wiley & Sons Ltd., 1983, vol. 14, pp. 247–350.
3
3
2
This journal is © The Royal Society of Chemistry 2019
Green Chem., 2019, 21, 6146–6153 | 6153