Mediating acid-catalyzed conversion of levoglucosan into platform chemicals with various solvents

Article abstract of DOI:10.1039/c2gc35961h

Acid-catalyzed conversions of levoglucosan have been investigated in mono-alcohols, poly-alcohols, water, chloroform, toluene, acetone, N,N-dimethyl formamide, dimethyl sulfoxide and some mixed solvents, aiming to mediate conversion of sugars into platform chemicals with solvents. The mono-alcohols can stabilize soluble polymers and thus suppress formation of insoluble polymers. Water does not have such an effect, leading to lower yields of levulinic acid. Chloroform cannot effectively dissolve levoglucosan, leading to dissolving of levoglucosan in the catalyst and the consequent rapid polymerization. Acetone reacted with sugars, forming substantial amounts of polymer. N,N-Dimethyl formamide poisoned the acid resin catalyst, leading to negligible conversion of levoglucosan. Dimethyl sulfoxide (DMSO) mainly catalyzed the conversion of levoglucosan into 5-(hydroxymethyl)furfural (HMF), 2,5-furandicarboxaldehyde, and the sulfur ether of HMF. DMSO has a low ability to transfer protons, which helps to avoid further contact of HMF with catalytic sites and stabilizes HMF.

Full text of DOI:10.1039/c2gc35961h

View Online / Journal Homepage
Dynamic Article Links
Green Chemistry
Cite this: DOI: 10.1039/c2gc35961h
www.rsc.org/greenchem
PAPER
Mediating acid-catalyzed conversion of levoglucosan into platform chemicals
with various solvents†
Xun Hu, Liping Wu, Yi Wang, Daniel Mourant, Caroline Lievens, Richard Gunawan and Chun-Zhu Li*
Received 23rd June 2012, Accepted 16th August 2012
DOI: 10.1039/c2gc35961h
Acid-catalyzed conversions of levoglucosan have been investigated in mono-alcohols, poly-alcohols,
water, chloroform, toluene, acetone, N,N-dimethyl formamide, dimethyl sulfoxide and some mixed
solvents, aiming to mediate conversion of sugars into platform chemicals with solvents. The mono-
alcohols can stabilize soluble polymers and thus suppress formation of insoluble polymers. Water does
not have such an effect, leading to lower yields of levulinic acid. Chloroform cannot effectively dissolve
levoglucosan, leading to “dissolving” of levoglucosan in the catalyst and the consequent rapid
polymerization. Acetone reacted with sugars, forming substantial amounts of polymer. N,N-Dimethyl
formamide poisoned the acid resin catalyst, leading to negligible conversion of levoglucosan. Dimethyl
sulfoxide (DMSO) mainly catalyzed the conversion of levoglucosan into 5-(hydroxymethyl)furfural
(HMF), 2,5-furandicarboxaldehyde, and the sulfur ether of HMF. DMSO has a low ability to transfer
protons, which helps to avoid further contact of HMF with catalytic sites and stabilizes HMF.
with yields up to 60% or the pyrolysis of wood with yields up to
Introduction
27%,35,36 in various solvents to mediate the production of the
Cellulose is an abundantly available bio-resource, the acid-treat-
ment of which produces 5-(hydroxymethyl)furfural (HMF) and/
or levulinic acid, the building block for value-added chemicals
or fuels.1–4 Significant efforts have been devoted to this research
area in recent years with the aim to achieve future energy
security.5–18 However, one of the main challenges associated
with the acid-catalyzed conversion of cellulose or sugars, is the
effective suppression of polymerization reactions in order to
achieve high yields of value-added chemicals.19,20 Using sol-
vents like methanol could significantly suppress polymerization,
as methanol could stabilize the reactive intermediates.21 The pro-
duction of HMF from fructose has also been investigated in
various solvents like ionic liquids,22,23 glycerol,24 betaine hydro-
chloride-based media,25 acetone/water,26 dimethyl sulfoxide,27
sulfolane,28 imidazolium chloride-based solvents,29–31 N,N-
dimethyl formamide,32 and biphasic solvents33,34 to suppress the
side-reactions. These solvents significantly modified the behav-
ior of fructose and further determined HMF yields. Apparently,
solvents not only physically dissolve but may also chemically
interact with the sugars, mediating their conversion. This study
is focused on the acid-catalyzed conversion of levoglucosan, an
anhydro sugar of glucose obtained from pyrolysis of cellulose
targeted products.
Acid-treatment of levoglucosan in methanol produces methyl
levulinate.21,37 Thus, other levulinic esters are expected to be
produced from the acid-catalyzed conversion of levoglucosan in
various alcohols. The levulinic esters such as ethyl levulinate is
a fuel and fragrance additive,38–40 while butyl levulinate has an
even higher heating value and better miscibility ability with
diesel fuel.41 However, ethanol and butanol as well as other alco-
hols, such as ethylene glycol and glycerin, are very different
from methanol in terms of molecular structures. Thus, do these
alcohols possess similar capacities to methanol for suppressing
polymerization in the acid-catalyzed conversion of levogluco-
san? Investigation of this could give us a general idea about the
correlations between molecular structures of the alcohols and
their abilities to suppress the polymerization reactions. Under-
standing of this also helps to assess the possibility of producing
higher levulinic esters directly via alcoholysis of sugars in the
corresponding alcohols.
In addition to alcohols, the behavior of levoglucosan in water,
acetone, chloroform, toluene, N,N-dimethyl formamide (DMF),
dimethyl sulfoxide (DMSO), and some mixed solvents
were investigated. These solvents have quite different
functional groups and properties. Do they, like alcohols,
react with sugars/intermediates and change the reaction path-
ways? Moreover, are they able to suppress the polymerization
reactions? This study attempts to address the above questions by
focusing on the formation of soluble polymers and insoluble
humins in various solvents. Moreover, the product distribution
and tendency towards polymerization were correlated with the
Fuels and Energy Technology Institute, Curtin University of Technology,
GPO Box U1987, Perth, WA 6845, Australia.
E-mail: chun-zhu.li@curtin.edu.au; Fax: (+) 61 8 9266 1138;
Tel: (+) 61 8 9266 1131
†Electronic supplementary information (ESI) available. See DOI:
10.1039/c2gc35961h
This journal is © The Royal Society of Chemistry 2012
Green Chem.

View Online
physical and chemical interactions of levoglucosan with the
solvents.
Results and discussion
Behavior of levoglucosan in acid-catalyzed reactions in alcohols
and water
Tables 1–4 summarize the conversion of levoglucosan and the
yields of HMF, levulinic acid and levulinic esters as a function
of time and solvent employed. Ethylene glycol and glycerin
(Table 1, entries 6 and 7) have some difficulty to dissolve levo-
glucosan even when some water was added. More importantly,
the conversion of levoglucosan was slowed down in these poly-
alcohols. The high viscosity of the poly-alcohols probably leads
to a diffusion limited transfer of levoglucosan to catalytic sites.
In comparison, the conversion of levoglucosan in the mono-alco-
hols like ethanol was very efficient, producing ethyl glucopyra-
nosides (Fig. S3 in ESI†) as the main products and HMF as the
minor initial products (Table 2, entry 1). The low yield of HMF
is partly due to ethanolysis of HMF to 5-(ethoxymethyl)-2-
furancarboxadhyde and acetal of HMF (not shown). However,
HMF yield in isopropanol is exceptionally high (Table 2, entry
3), which was followed by high yields of levulinic acid (Table 3,
entry 3). The secondary hydroxyl group in isopropanol creates
some steric hindrance and thus lowers the reactivity of isopropa-
nol in alcoholysis reactions. 2-Octanol is a much higher molar
mass alcohol and showed even lower activity in the alcoholysis
reactions. Therefore in this solvent, levoglucosan mainly under-
went dehydration to HMF (Table 2, entry 5) and then to levulinic
acid rapidly (Table 3, entry 5). The formation of 2-octyl levuli-
nate was relatively slow (Table 4, entry 5). In ethanol, ethyl levu-
linate and levulinic acid were formed in parallel and the yield of
ethyl levulinate (Table 4, entry 1) is much higher than that of
levulinic acid.
Experimental
Acid-treatments of levoglucosan were performed in a 130 ml
Hastalloy batch autoclave reactor in various solvents with
Amberlyst 70, an acidic resin, as the catalyst. The reaction mix-
tures were purged with nitrogen and then heated up to 170 °C in
20 min under autonomous pressure. The samples were taken at
an interval of 20 min at 170 °C for 180 min. After finishing the
experiments, the residual solution and catalyst was collected and
separated via filtration. The catalyst, together with solid particles,
was dried at 105 °C for 4 h to a constant weight. The stability of
this resin catalyst was checked by evaluating the catalytic activity
for the acid-catalyzed conversion of levoglucosan in water for
two cycles. Based on the yields of levulinic acid, the catalyst
maintains ca. 70–80% of initial activity in the second cycle
(Fig. S1† for the details).
Agilent GC-MS (6890 series GC with a 5973 MS detector)
with a capillary column (HP-INNOWax) was used to analyze the
products. Identification of the products was achieved with a
spectral library (NIST MS Search 2005), which was further
confirmed with standards if available. Standard solutions were
used to obtain the calibration curves to calculate concentrations
of the compounds of interest. It is impossible to calculate the
concentration of the compounds without the standards available.
In this case, the peak area in the MS spectrum was normalized
and used to measure the abundance changes of the compounds
without standards. Some typical compounds were identified with
the spectral library, but without standards, confirming the pres-
ence of 1,6-anhydro-β-D-glucofuranose, 1,4:3,6-dianhydro-α-D-
glucopyranose, and levoglucosenone (their mass spectra are pre-
sented in Schemes S1, S2, and S3,† respectively).
Formation of soluble polymers in mono-alcohols and water
The total yields of levulinic acid and ethyl levulinate in ethanol
(ca. 66.6%) are clearly higher than that of levulinic acid in water
(ca. 47.6%), which is due to the significant polymerization of
levoglucosan in water (Fig. 1). The polymerization also occurred
in alcohols but the difference from ethanol to butanol is not that
large and the amount of insoluble polymers formed is much
lower than that in water. Apparently, the bigger alcohols, such as
ethanol and butanol, show similar suppressive effects on
polymer formation. However, 2-octanol is not as efficient as the
smaller alcohols, which is due to its lower activity in alcoholysis
reactions, resulting from steric hindrance.
Soluble polymer is the precursor of insoluble polymer. Under-
standing the correlation between product distribution and soluble
polymer formation can help to understand when and how the
polymerization starts. The soluble polymer is difficult to quantify
but it is possible to characterize their relative abundance.
Polymerization of sugars involves dehydration and condensation
of sugars themselves or together with HMF.42,43 During conden-
sation, polymers with extended π-bonds increase their molecular
weight to soluble and eventually to insoluble polymers. Thus, it
is possible to characterize abundance of the conjugated π-bonds
in the soluble polymer with UV-fluorescence spectroscopy.
Fig. 2 shows the excitation intensity of the soluble polymers in
water and in ethanol. The wavelength indicates size of the
The yield of the products is defined as follows:
Yield ðmol%Þ ¼ ðmoles of the compound in products=
theoretical moles of the compound from the levoglucosan loadedÞ
ꢀ100%
UV-fluorescence spectra of the soluble polymers were
measured with a Perkin-Elmer LS50B spectrometer. The syn-
chronous spectra were recorded using a constant energy differ-
ence of −2800 cm−1 and a scan speed of 200 nm min−1. In
general, the concentrations of the samples were diluted to
800 ppm in methanol, which is in the linear range of response of
the UV-fluorescence spectra with the same concentration.
However, in Table 5 (entry 12), with acetone as the solvent, con-
centrations of the samples were diluted to 200 ppm as the con-
centration of 800 ppm was outside the range of linear response.
The degradation products from levoglucosan, such as furfural
and levulinic acid, did not show strong excitation spectra, as is
shown in Fig. S2.† The FT-IR spectra of the catalyst, together
with the insoluble polymers, were characterized with a Perkin-
Elmer Spectrum GX FT-IR/Raman Spectrometer with a spectral
resolution of 4 cm−1. The spectrum represents the average of 10
scans.
Green Chem.
This journal is © The Royal Society of Chemistry 2012

View Online
Table 1 Levoglucosan concentrations in different alcohols/water mixtures versus reaction timea
Reaction time/concentration (wt%)
Entry
Solvent
RTb
0 minb
20 min
60 min
120 min
180 min
1
2
3
4
5
6
7
8
Ethanol/water
9.83
8.68
9.63
8.61
8.44
6.84
4.82
9.48
0.88
0.98
1.29
0.88
0.42
2.31
3.33
2.43
0.41
0.31
0.58
0.40
0.39
0.70
1.22
0.88
0.24
0.15
0.10
0.12
0.03
0.44
0.76
0.56
0.07
0.04
0
0
0
0.39
0.43
0.22
0
0.02
0
0
0
0.22
0.32
0.06
1-Propanol/water
Isopropanol/water
1-Butanol/water
2-Octanol/water
Ethylene glycol/water
Glycerin/water
Water
a Experimental conditions: levoglucosan loaded: 9.5 g; catalyst loaded: 9.5 g; solvents: alcohol/water mixture: 90 g (75 g of alcohol and 15 g of
water); reaction temperature: 170 °C; reaction time at 170 °C: 180 min; stirring rate: 500 rpm. b “RT” means at room temperature. “0 min” means the
reaction temperature just reached 170 °C.
Table 2 The yields of HMF versus reaction time
Reaction time/yield (%)
Entry
Solvent
RT
0 min
20 min
60 min
120 min
180 min
1
2
3
4
5
6
7
8
Ethanol/water
0
0
0
0
0
0
0
0
1.35
1.77
3.95
3.46
5.67
1.52
0.42
4.23
2.82
3.92
10.36
3.46
1.77
0.72
0.36
5.98
3.38
2.81
5.21
3.52
0.20
0.46
0.18
3.34
1.17
0.63
0.76
0.66
0.01
0.17
0.12
1.13
0.28
0.10
0.15
0.66
0
0.12
0.10
0.53
1-Propanol/water
Isopropanol/water
1-Butanol/water
2-Octanol/water
Ethylene glycol/water
Glycerin/water
Water
Table 3 The yields of levulinic acid versus reaction time
Reaction time/yield (%)
Entry
Solvent
RT
0 min
20 min
60 min
120 min
180 min
1
2
3
4
5
6
7
8
Ethanol/water
0
0
0
0
0
0
0
0
0
0.97
3.38
4.34
3.49
27.17
2.79
3.66
8.34
7.85
12.47
21.73
11.34
28.86
10.81
11.48
31.41
15.25
19.17
28.48
15.37
26.71
17.49
16.08
47.14
21.15
21.02
37.92
16.77
25.82
26.47
17.78
47.56
1-Propanol/water
Isopropanol/water
1-Butanol/water
2-Octanol/water
Ethylene glycol/water
Glycerin/water
Water
0.25
0.37
0.38
4.50
0.11
0
0.43
Table 4 The yields or relative abundance of levulinic esters versus reaction time
Abundance to that of levulinic acid at 180 mina
Entry
1
Solvent
Levulinic ester
Ethyl levulinate
0 min
20 min
60 min
120 min
180 min
Ethanol/water
Yields (%)
0.01
0.22
0.02
0.01
0.06
0.02
0.02
0.20
3.87
0.70
0.23
1.06
0.75
0.55
1.36
25.02
2.12
1.07
3.21
2.40
1.76
2.28
39.78
3.01
1.20
3.91
3.89
2.55
2.43
45.41
3.44
1.42
4.89
3.91
3.10
2
3
4
5
6
1-Propanol/water
Isopropanol/water
1-Butanol/water
Propyl levulinate
Isopropyl levulinate
Butyl levulinate
2-Octanol/water
Ethylene glycol/water
2-Octyl levulinate
2-Hydroxyethyl levulinate
a Due to the difficulty of obtaining standards, the ratio of levulinic ester abundance to the abundance of levulinic acid at 180 min was used to show the
relative change of levulinic ester concentration versus reaction time.
This journal is © The Royal Society of Chemistry 2012
Green Chem.

View Online
conjugated π-bonds (the longer the wavelength, the bigger the
conjugated π-bonds), while the excitation intensity indicates the
abundance.44
then again be protonated to form another carbon cation. Simi-
larly, the carbon cation could attack any other dehydration
In water, the excitation intensity of the soluble polymer was
weak, indicating the small abundance of soluble polymer. In
comparison, excitation intensity of the soluble polymer in
ethanol was much stronger, indicating abundant soluble polymer
had formed in ethanol. This result contradicts that of Fig. 1
(where the amount of insoluble polymer formed is much higher
in water than in ethanol), as soluble polymers are precursors of
insoluble ones. The formation of more soluble polymer in
ethanol should consequently form more insoluble polymer.
However, this is not the case. Actually, this seemingly controver-
sial result suggests that different mechanisms act in the polymer-
ization of sugar in water and in ethanol. Scheme 1 shows the
proposed polymerization pathways of a typical fragment of
sugar. In water with the aid of acid, the fragment could dehydrate
to form a compound with a π bond. A carbon cation is also
formed as an intermediate. The carbon cation could react with
the dehydration products such as HMF via electrophilic addition
giving condensation products. The dehydration products could
Fig. 1 Formation of insoluble polymer in acid-treatment of levogluco-
Scheme 1 Proposed reaction pathways for polymerization of a typical
san in various solvents.
sugar fragment in water and in ethanol.
Fig. 2 Excitation fluorescence spectra for the soluble polymers in water and in ethanol. Samples were diluted in methanol to 800 ppm.
Green Chem. This journal is © The Royal Society of Chemistry 2012

View Online
product with π bonds via direct electrophilic addition. HMF
itself could also form a carbon cation that could further react
with others to give polymers.
In alcohol, sugars also undergo dehydration with carbon
cations formed as intermediates. However, the subsequent steps
could be quite different. If one carbon cation was formed, it
would react with alcohols to form the stabilized products. One
example is the formation of methyl glucosides from levogluco-
san in methanol.21 The dehydration products, such as HMF,
could also further react with alcohols to be stabilized. An
example is alcoholysis of HMF to form its acetals or ethers,45–47
which stabilizes the reactive functional groups and suppresses
their potential polymerization.21 Alcohols, like the inhibitor in
chain reactions, stabilize the carbon cations and terminates
polymerization, forming the soluble polymer with limited a size
in this medium. Moreover, both the soluble polymer and its pre-
cursor are in one phase, and hence there may be an equilibrium
between them, which would help to suppress the formation of
more soluble polymer. This case does not exist in water, as the
solubility of the polymer in water is negligible. It is therefore
conceivable that the polymerization to the insoluble polymer in
water proceeded very quickly, forming the soluble polymers as
the minor intermediates.
Correlations between polymer formation and product
distribution
Fig. 2b also revealed intrinsic correlation between polymeriz-
ation and product distribution. In Fig. 2b, from 0 min to 20 min
at 170 °C, the excitation shows one main peak centered at
ca. 320 nm. At this stage, the main reaction is the conversion of
levoglucosan to ethyl glucosides (Fig. S2†). The significant
dehydration of levoglucosan or ethyl glucosides to HMF
(Table 2, entry 1) or levulinic acid/ester (Table 4, entry 1) has
not started yet. From 20 to 60 or 120 min, significant amounts of
levulinic acid and ethyl levulinate were formed. Simultaneously,
strong excitation at 360 and 390 nm appeared (Fig. 2b). In
Fig. 2a, in water, the increase of the excitation at ca. 380 nm
also was accompanied by the formation of a large amount of
levulinic acid (Table 3, entry 8). Thus, the excitation at
ca. 320 nm should correspond to the polymers formed from the
dehydration of sugars, while that at ca. 390 nm corresponds to
the polymers formed from the decomposition of HMF to levuli-
nic acid/ester.
Fig. 3 Excitation fluorescence spectra for the soluble polymers in
mono-alcohols and in poly-alcohols. Samples were diluted in methanol
to 800 ppm.
Fig. 3a shows the excitation spectra of the soluble polymers in
the mono-alcohols after the residence time of 180 min at 170 °C.
The spectra from ethanol and butanol were very similar,
however, that of isopropanol has a stronger excitation. The acid-
treatment of levoglucosan in isopropanol produced higher
amounts of HMF (Table 2, entry 3) and levulinic acid (Table 2,
entry 3), which is the main reason for the stronger excitation at
the short and the long wavelengths. The excitation of 2-octanol
is exceptionally weaker than that of other alcohols. The long
carbon chain of 2-octanol makes it more like an alkane and less
like an alcohol. The low activity of 2-octanol in alcoholysis reac-
tions lowered its ability to stabilize the precursors of the poly-
mers, leading to the formation of more insoluble polymer
(Fig. 1).
Fig. 3b shows that excitation of the soluble polymer in ethyl-
ene glycol, which is mainly located at the short wavelengths,
and is somewhat different from that of ethanol. Levoglucosan
was converted much slower in ethylene glycol (Table 1, entry 6)
than in ethanol. Moreover, ethylene glycol is more polar than
ethanol and has a higher ability to react with the carbonyl group
to terminate growth of the conjugated π bond systems. As is
shown in Scheme 2, ethylene glycol can even react with
2-hydroxyethyl levulinate to form ethyl 3-(2-methyl-1,3-dioxo-
lan-2-yl)propanoate. Thus, the precursor of the polymer might
be stabilized with ethylene glycol more easily, forming the
soluble polymer with smaller conjugated π bonds. However, the
This journal is © The Royal Society of Chemistry 2012
Green Chem.

View Online
Scheme 2 The typical reactions of ethylene glycol in the presence of acid catalyst.
C–H bonds was found in ethanol and glycerin but not in water,
but it has not yet been clear about the reason for this difference.
Behavior of levoglucosan in acid-catalyzed reactions in CHCl3,
CHCl3/water, and toluene
Tables 5–7 summarize the conversion of levoglucosan and
yields of HMF and levulinic acid as a function of time and
solvent employed. Levoglucosan negligibly dissolved in chloro-
form (Table 5, entry 9) due to their quite different polarities.
However, the Amberlyst 70 catalyst contains a high concen-
tration of polar –SO3H groups and a large amount of water (ca.
57%) on its local surface. Thus, levoglucosan has high affinity
with the catalyst and might “dissolve” in the catalyst, as is
shown in Scheme 3. To prove this, levoglucosan and chloroform
were mixed with and without Amberlyst 70 at room temperature,
and the resulting mixtures are shown in Fig S4.† Levoglucosan
disappeared in the catalyst in the bottle containing Amberlyst 70,
while it was still discernible as an undissolved solid in the bottle
without Amberlyst 70. The “dissolving” of levoglucosan in the
catalyst must facilitate its conversion due to close contact with
the hydrogen ion in/on the local surface of the catalyst. As is
shown in Table 7 (entry 9), the yields of levulinic acid in chloro-
form reached ca. 24% when the reaction temperature reached
170 °C, which is close to the equilibrium value of ca. 30%.
However, the dissolving of levoglucosan in the catalyst facili-
tated not only conversion of levoglucosan, but also the formation
of the solid polymer (Fig. 1). It is known that the increase of
sugar loadings would favor polymerization. The dissolving of
levoglucosan in catalyst has similar effects.
Not only levoglucosan, but also polar dehydration products
such as HMF, could be trapped in the polar catalyst (Scheme 3).
Thus, sugar and its dehydration intermediates would have more
chance to condense together to polymerize within the catalyst
matrix. Furthermore, the condensation reactions could also be
facilitated by close contact with the hydrogen ions in/on the
local surface of the catalyst. In comparison, in the polar water,
levoglucosan and the dehydration products are mainly dispersed
in the aqueous phase. Although a longer time (ca. 120 min) was
needed to reach the equilibrium yields of levulinic acid (ca.
47%) (Table 3, entry 8), the amount of polymer formed is less
and the yield of levulinic acid is much higher (47% versus
30%). Apparently, the dispersion of sugar and the dehydration
products in the polar water lower their chance to collide and
Fig. 4 FT-IR spectrum of the insoluble polymers formed in the typical
solvents.
acid-treatment of levoglucosan in ethylene glycol formed more
solid polymer (Fig. 1), which is closely related to the multiple
hydroxyl groups in ethylene glycol. Ethylene glycol could con-
tinuously increase its size via etherification to triethylene glycol,
pentaethylene glycol or bigger ethers/alcohols without losing the
reactive centers, as is shown in Scheme 2. The big ethers/
alcohols could react not only with the degradation products from
levoglucosan, such as acetic acid, but also the soluble polymers,
forming insoluble polymers. This also explained the formation
of the significant amount of polymers from glycerin. In addition,
there is a shift of the excitation spectrum to a shorter wavelength
from ethanol to glycerin (Fig. 3c), which is probably related to
the higher ability of the poly-alcohols to terminate growth of the
conjugated π bonds.
The insoluble polymers formed in ethanol, glycerin, and water
were characterized with FT-IR, as is shown in Fig. 4. Compared
with that of the fresh catalyst, the insoluble polymer formed in
alcohols or water showed a characteristic adsorption of a carbo-
nyl group. The dehydration of levoglucosan produces HMF,
while the cross-polymerization of HMF with sugars could
produce insoluble polymers probably containing the carbonyl
group. In addition, the adsorption of the unsaturated aliphatic
Green Chem.
This journal is © The Royal Society of Chemistry 2012

View Online
Table 5 Levoglucosan concentrations in different solvents versus reaction timea
Reaction time/concentration (wt%)
Entry
Solvent
RT
0 min
20 min
60 min
120 min
180 min
9
Chloroform
Chloroform/water
Toluene
Acetone
Acetone/chloroform
DMF
0.05
0.02
0.82
0.05
0.98
0.05
9.84
7.17
0.01
0.33
0.04
0.27
0.01
9.39
2.88
0
0.16
0
0.02
0
9.95
0.70
0
0.02
0
0
0
9.63
0.13
0
0
0
0
0
9.66
0.02
10
11
12
13
14
15
4.39
0.02
3.40
0.57
10.05
8.80
DMSO
a Experimental conditions: levoglucosan loaded: 9.5 g; catalyst loaded: 9.5 g; the solvent loaded: 90 ml, 45 ml each for the two mixed solvents;
reaction temperature: 170 °C; reaction time at 170 °C: 180 min; stirring rate: 500 rpm; P = 20 bar.
Table 6 The yields of HMF versus reaction time
Reaction time/yield (%)
Entry
Solvent
RT
0 min
20 min
60 min
120 min
180 min
9
Chloroform
Chloroform/water
Toluene
Acetone
Acetone/chloroform
DMF
0
0
0
0
0
0
0
0.23
2.41
2.36
2.04
2.63
0.21
2.25
0.05
3.53
0.48
4.73
1.56
0.25
16.16
0.01
2.54
0.03
1.18
0.25
0.14
34.54
0
0.11
0
0.04
0
0.04
38.63
0
0
0
0
0
0
10
11
12
13
14
15
DMSO
34.42
Table 7 The yields of levulinic acid versus reaction time
Reaction time/yield (%)
Entry
Solvent
RT
0 min
20 min
60 min
120 min
180 min
9
Chloroform
Chloroform/water
Toluene
Acetone
Acetone/chloroform
DMF
0
0
0
0
0
0
0
24.17
0.48
8.14
5.21
8.85
0
24.81
9.38
16.81
13.71
29.36
0
27.85
38.26
19.38
17.01
27.97
0
30.94
49.77
22.10
17.41
28.73
0
30.25
52.76
24.33
16.12
28.12
0
10
11
12
13
14
15
DMSO
0
0.22
1.39
2.56
2.77
Behavior of levoglucosan in acid-catalyzed reactions in acetone
condense, i.e., the higher yield of levulinic acid under aqueous
conditions is a concentration effect.
The chloroform/water dual phase medium promoted the pro-
duction of levulinic acid and suppressed polymerization.
However, this is mainly due to the beneficial effect of water.
Toluene acted similarly to that of chloroform. Fig. 5 shows the
UV spectra in chloroform, toluene, and chloroform/water. The
excitation intensity of the soluble polymers are very small com-
pared with those in the alcohols, indicating that the soluble
polymer formed was quickly converted into the insoluble one.
The spectrum for chloroform/water is similar to that of water,
showing an adsorption at ca. 390 nm, suggesting that the behav-
ior of levoglucosan in chloroform/water was mainly governed by
water. The FT-IR spectrum of the insoluble polymer obtained in
chloroform (Fig. 4) was very similar to that from the reaction in
water, showing the characteristic adsorption of the carbonyl
group but a very weak adsorption of the unsaturated C–H bonds.
The solubility of levoglucosan in acetone is much higher than
that in chloroform. However, the levulinic acid yield is very
small (Table 7, entry 12), which is due to the large extent of
polymerization (Fig. 1). Acetone contains a carbonyl group and
an active α-H, making it very reactive in condensation reactions.
As is shown in Scheme 4, various kinds of products can be pro-
duced by acetone itself via aldol condensation. Some of these
condensation products still contain the active functional groups,
and they could further increase their size via either adol conden-
sation or electrophilic substitution reaction to eventually form
polymer. Acetone also reacted with sugars, producing 1,2:5,6-
bis-O-(1-methylethylidene)-α-D-glucofuranose
and
another
product with a similar structure (Scheme 5). The abundance of
these two products decreased rapidly with the progress of the
reaction (Fig. S5†). The formation of methyl glucosides in
This journal is © The Royal Society of Chemistry 2012
Green Chem.

View Online
producing furfurylideneacetone. Furfurylideneacetone contains
large conjugated π bonds and can react further to form bigger
condensation products (Fig. S6†). In addition, in acetone, levo-
glucosan could react with acetic acid to form 2,4-o-diacetyl-6-
anhydrohexopyranose, which was not found in any other
solvent. The mixed solvent acetone/chloroform did not change
the reaction network much. A substantial amount of polymer
was still formed. Fig. 6 shows the UV spectrum in acetone. The
adsorption increased almost linearly with reaction time and did
not reach the equilibrium as that in water or ethanol. The for-
mation of levulinic acid in acetone reached the equilibrium in
60 min, while the adsorption of the UV fluorescence continu-
ously increased to a significant extent. Clearly, the polymeriz-
ation reactions in acetone continued after complete conversion of
the sugars.
Behavior of levoglucosan in acid-catalyzed reactions in DMF
and DMSO
The conversion of levoglucosan in N,N-dimethyl formamide
(DMF) was small (Table 5, entry 14) and only a small amount of
HMF was formed (Table 6, entry 14). Apparently, Amberlyst 70
did not exert its capacity to convert levoglucosan. DMF is alka-
line and probably poisoned the acidic catalyst. The used Amber-
lyst 70 in DMF did not change its original brown color, but in
other solvents it changed to black, which is due to the formation
of the polymer. However, Fig. 1 shows that some “polymer” was
formed, which actually is probably due to the formation of salt
from the acid–base reaction between DMF and Amberlyst 70.
The local pH of the used catalyst in DMF, which was determined
by contacting the catalyst with the electrode, was ca. 7.48. This
value is much higher than that of the fresh catalyst (ca. 2.07).
Clearly, the catalyst was poisoned with DMF. The FT-IR charac-
terization shows that the used catalyst showed the characteristic
adsorption of the methyl group and the amide group (Fig. 4),
which further confirms the neutralization of the catalyst with
DMF. DMF slowly decomposes to CO and dimethylamine at its
boiling point. This process is catalyzed by acid and can be
dangerous. Base could catalyze the isomerization of glucose to
fructose to promote the production of HMF.48 However, the alka-
linity of DMF is not strong enough to effectively catalyze the
conversion of levoglucosan.
Scheme 3 The different dispersion model of levoglucosan and its
dehydration products in chloroform and water.
Conversely, in DMSO, levoglucosan was efficiently converted
at 170 °C (Table 5, entry 15). Interestingly, HMF was the major
product (Table 6, entry 15) while levulinic acid was the minor
one (Table 7, entry 15), which is quite different from that in
other solvents. The acid-catalyzed conversion of fructose also
produce HMF as the main product.49 Further conversion of HMF
into levulinic acid and formic acid needs the involvement of
water. However, the absence of water in DMSO was not the
main reason for the formation of the small amount of levulinic
acid. As is shown in Scheme 6, the overall reaction of levogluco-
san to levulinic acid produced and consumed the same amount
of water if the selectivity was 100%. Moreover, although other
non-aqueous solvents, such as acetone and chloroform, do not
contain water, levulinic acid was still the major product while
HMF was the minor intermediate. In fact, after mixing with
Amberlyst 70, the water content in DMSO increased to ca.
Fig. 5 Excitation fluorescence spectrum of the soluble polymers in
acid-treatments of levoglucosan in toluene, chloroform, and chloroform/
water. Samples were diluted in methanol to 800 ppm. The samples were
with a residence time of 180 min at 170 °C.
methanol protect the C1 hydroxyl group and consequently sup-
presses the polymerization.21 However, the protection of the
hydroxyl groups with acetone does not have such an effect.
Acetone forms tautomeric enols that exhibit some properties
of carbanions (Scheme 5), which could attack the carbon cations
formed in the dehydration of sugars. In addition to acetone, the
condensation products from acetone also could react with sugars
to form polymers. Moreover, acetone can also react with furfural,
Green Chem.
This journal is © The Royal Society of Chemistry 2012

View Online
Scheme 4 The intramolecular aldol condensation of acetone in the presence of an acid catalyst.
Scheme 5 The condensation reactions between levoglucosan and acetone.
This journal is © The Royal Society of Chemistry 2012
Green Chem.

View Online
5.8%. Evidently, some water in Amberlyst 70 was released into
DMSO via solvent effects. The water was not consumed but
increased to ca. 6.8% due to the dehydration of levoglucosan
into HMF.
In addition to HMF, levoglucosan also underwent complex
conversion to produce other anhydro sugars, levoglucosenone,
and acetic acid in DMSO (Scheme 7). Acetic acid could further
react with HMF to form 5-acetoxymethyl-2-furaldehyde. HMF
could be transformed into 2,5-furandicarboxaldehyde, which is
also a platform molecule for value-added chemicals.51 DMSO is
a mild oxidant, which oxidized the hydroxyl group of HMF, pro-
ducing 2,5-furandicarboxaldehyde. The polysulfides from the
decomposition of DMSO could further react with HMF to
DMSO is a polar but aprotic solvent. It cannot transfer
protons efficiently. Fig. 7 shows the spatial proton distribution in
DMSO and water solutions containing Amberlyst 70. In water, a
small amount of hydrogen ions could disperse in the whole solu-
tion, which was not the case in DMSO. In DMSO, the local pH
of the catalyst is also much higher than in water. Thus, in DMSO
the HMF produced on the local surface of Amberlyst 70 can be
extracted into the DMSO phase and avoid continuous contact
with the acidic sites in the catalyst, as in the other dual phase
reactions.33 Furfural is very reactive towards polymerization50 no
matter whether it is in water, alcohol, or acetone (Table 8).
However, in DMSO, furfural was quite stable. It was believed
that, like HMF, the interaction with DMSO and the avoidance of
further contact with the acidic catalytic sites are the main reasons
for its stabilization.
The UV-spectrum in DMSO shows a main adsorption peak at
a short wave length, which corresponds to the polymer formed
during the dehydration of levoglucosan to HMF. Interestingly,
the UV-spectrum decreased significantly when the production of
HMF reached a plateau at 60 min. The soluble polymer was
probably further converted to the insoluble polymer or reacted
with the polysulfide formed from the decomposition of DMSO
to eliminate the π-bond system. The insoluble polymer formed in
DMSO is similar to that in ethanol (Fig. 1). Clearly, DMSO also
shows the suppressive effects on the polymerization reactions.
produce
5-[(methylthio)methyl]-2-furancarboxaldehyde
(MMFC). MMFC was identified by analysis of its MS spectrum
(Schemes S4 and S5†), which is a major derivative of HMF. The
introduction of sulfur may make MMFC a precursor for the syn-
thesis of some pharmaceutical intermediates.
Conclusions
The nature of the solvent strongly impacts the product distri-
bution and polymer formation in acid-catalyzed conversions of
levoglucosan. Alcohols could inhibit the growth of soluble poly-
mers into insoluble polymers via stabilization of the reactive
intermediates, while water cannot. However, poly-alcohols could
polymerize themselves via etherification. Levoglucosan, in the
less polar solvents, such as chloroform, polymerized rapidly as
sugars, and its dehydration product “dissolved” in the catalyst,
creating more chances to condense. Acetone reacted with sugars
via aldol condensations, leading to serious polymerization and
Fig. 7 Acidity distribution in the DMSO solution (40 ml) containing
5 g Amberlyst 70 and the water solution (40 ml) with the same amount
of catalyst. A vial with a height of 10 cm was used to contain the solu-
tions. A digital pH meter was used to measure the pH value. “0 cm” in
the x-axis means direct contact of the catalyst with pH electrode.
Fig. 6 Excitation fluorescence spectrum of the soluble polymers
versus reaction time in acid-treatments of levoglucosan in acetone.
Samples were diluted in methanol to 200 ppm.
Scheme 6 Theoretical conversion of levoglucosan to levulinic acid and formic acid.
Green Chem.
This journal is © The Royal Society of Chemistry 2012

View Online
Table 8 The yields of furfural versus reaction time
Reaction time/yield (%)
Entry
Solvent
RT
0 min
20 min
60 min
120 min
180 min
1
8
12
15
Ethanol/water
Water
Acetone
DMSO
0
0
0
0
0.10
0.94
0.46
0.14
0.45
1.05
1.12
0.60
1.68
0.86
0.29
1.39
1.28
0.40
0.01
1.77
0.68
0.21
0
1.91
Scheme 7 The reaction pathways of levoglucosan in DMSO. Definition of the abbreviation: DAGP: 1,4:3,6-dianhydro-α-D-glucopyranose; HMF: 5-
(hydroxymethyl)furfural; MMFC: 5-[(methylthio)methyl]-2-furancarboxaldehyde.
producing minimum yields of levulinic acid. DMF poisoned the
acid catalyst, preventing the occurrence of acid-catalyzed reac-
tions. DMSO could not effectively transfer hydrogen ions and
hence could effectively convert levoglucosan into HMF and
5-[(methylthio)methyl]-2-furancarboxaldehyde and stabilize
them. The solvents determine the conversion efficiency of the
sugars via interaction with the sugars or their degradation pro-
ducts, both physically and chemically. In the acid-catalyzed con-
version of sugar oligomers or cellulose/biomass, the effects of
solvent on polymerization reactions needs to be taken into con-
sideration as it determines the efficiency of the whole process.
The solvents with lower boiling points, such as the low boiling-
point alcohols, could effectively suppress the polymerization and
make the separation of products from the solvents much easier.
Government of Western Australia via the Centre for Research
into Energy for Sustainable Transport (CREST).
References
1 A. Corma, O. Torre and M. Renz, Energy Environ. Sci., 2012, 5, 6328–
6344.
2 A. M. R. Galletti, C. Antonetti, V. D. Luise and M. Martinelli, Green
Chem., 2012, 14, 688–694.
3 Y.-C. Lin and G. W. Huber, Energy Environ. Sci., 2009, 2, 68–80.
4 R. Karinen, K. Vilonen and M. Niemel, ChemSusChem, 2011, 4, 1002–
1016.
5 J. C. Serrano-Ruiz and J. A. Dumesic, Energy Environ. Sci., 2011, 4, 83–
99.
6 R. Xing, W. Qi and G. W. Huber, Energy Environ. Sci., 2011, 4, 2193–
2205.
7 J. N. Chheda, G. W. Huber and J. A. Dumesic, Angew. Chem., Int. Ed.,
2007, 46, 7164–7183.
8 X.-L. Du, Q.-Y. Bi, Y.-M. Liu, Y. Cao and K.-N. Fan, ChemSusChem,
2011, 4, 1838–1843.
Acknowledgements
9 G. R. Akien, L. Qi and I. T. Horváth, Chem. Commun., 2012, 48, 5850–
5852.
10 M. Mascal and E. B. Nikitin, Angew. Chem., Int. Ed., 2008, 47, 7924–
Australian Government funding through the Second Generation
Biofuels Research and Development Grant Program supports
this project. The study also received support from the
7926.
This journal is © The Royal Society of Chemistry 2012
Green Chem.

View Online
11 E. I. Gürbüz, S. G. Wettstein and J. A. Dumesic, ChemSusChem, 2012, 5,
383–387.
12 S. Saravanamurugan, O. N. V. Buu and A. Riisager, ChemSusChem,
2011, 4, 723–726.
31 J. Young, G. Chan and Y. Zhang, ChemSusChem, 2009, 2, 731–734.
32 G. Yong, Y. Zhang and J. Y. Ying, Angew. Chem., 2008, 120, 9485–9488.
33 Y. Román-Leshkov, J. N. Chheda and J. A. Dumesic, Science, 2006, 312,
1933–1937.
13 K. W. Omari, J. E. Besaw and F. M. Kerton, Green Chem., 2012, 14,
1480–1487.
34 J. N. Chheda, Y. Román-Leshkov and J. A. Dumesic, Green Chem.,
2007, 9, 342–350.
14 M. G. Al-Shaal, W. R. H. Wright and R. Palkovits, Green Chem., 2012,
14, 1260–1263.
15 X.-L. Du, Q.-Y. Bi, Y.-M. Liu, Y. Cao, H.-Y. He and K.-N. Fan, Green
Chem., 2012, 14, 935–939.
16 E. J. García-Suárez, A. M. Balu, M. Tristany, A. B. García, K. Philippot
and R. Luque, Green Chem., 2012, 14, 1434–1439.
17 W. Deng, M. Liu, Q. Zhang, X. Tan and Y. Wang, Chem. Commun.,
2010, 46, 2668–2670.
18 J. B. Binder and R. T. Raines, J. Am. Chem. Soc., 2009, 131, 1979–1985.
19 J. J. Bozell and G. R. Petersen, Green Chem., 2010, 12, 539–554.
20 F. S. Asghari and H. Yoshida, Ind. Eng. Chem. Res., 2007, 46, 7703–
7710.
21 X. Hu and C.-Z. Li, Green Chem., 2011, 13, 1676–1679.
22 X. Tong and Y. Li, ChemSusChem, 2010, 3, 350–355.
23 X. Qi, M. Watanabe, T. M. Aida and R. L. Smith Jr., ChemSusChem,
2009, 2, 944–946.
24 M. Benoit, Y. Brissonnet, E. Guélou, K. D. O. Vigier, J. Barrault and
F. Jérôme, ChemSusChem, 2010, 3, 1304–1309.
25 K. D. O. Vigier, A. Benguerba, J. Barrault and F. Jérôme, Green Chem.,
2012, 14, 285–289.
35 F. Shafizadeh, J. Anal. Appl. Pyrolysis, 1982, 3, 283–305.
36 G. Dobele, G. Rossinskaja, T. Dizhbite, G. Telysheva, D. Meier and
O. Faix, J. Anal. Appl. Pyrolysis, 2005, 74, 401–405.
37 X. Hu, Y. Wang, D. Mourant, R. Gunawan, C. Lievens, W. Chaiwat,
M. Gholizadeh, L. Wu, X. Li and C.-Z. Li, AIChE J., 2012, DOI:
10.1002/aic.13857.
38 J.-P. Lange, R. Price, P. M. Ayoub, J. Louis, L. Petrus, L. Clarke and
H. Gosselink, Angew. Chem., Int. Ed., 2010, 49, 4479–4483.
39 J. P. Lange, W. D. van de Graaf and R. J. Haan, ChemSusChem, 2009, 2,
437–441.
40 M. Mascal and E. B. Nikitin, Green Chem., 2010, 12, 370–373.
41 D. J. Hayes, Catal. Today, 2009, 145, 138–151.
42 A. Corma, S. Iborra and A. Velty, Chem. Rev., 2007, 107, 2411–2502.
43 A. Chuntanapum and Y. Matsumura, Ind. Eng. Chem. Res., 2009, 48,
9837–9846.
44 Y. Wang, X. Li, D. Mourant, R. Gunawan, S. Zhang and C.-Z. Li, Energy
Fuels, 2012, 26, 241–247.
45 G. A. Kraus and T. Guney, Green Chem., 2012, 14, 1593.
46 M. Balakrishnan, E. R. Sacia and A. T. Bell, Green Chem., 2012, 14,
1626.
26 X. Qi, M. Watanabe, T. M. Aida and R. L. Smith Jr., Green Chem., 2008,
10, 799–805.
47 X. Hu, C. Lievens, A. Larcher and C.-Z. Li, Bioresour. Technol., 2011,
102, 10104–10113.
27 J. Wang, W. Xu, J. Ren, X. Liu, G. Lu and Y. Wang, Green Chem., 2011,
13, 2678–2681.
48 A. Takagaki, M. Ohara, S. Nishimura and K. Ebitani, Chem. Commun.,
2009, 6276–6278.
28 B. R. Caes and R. T. Raines, ChemSusChem, 2011, 4, 353–356.
29 C. Moreau, A. Finiels and L. Vanoye, J. Mol. Catal. A: Chem., 2006,
253, 165–169.
30 H. Zhao, J. E. Holladay, H. Brown and Z. C. Zhang, Science, 2007, 316,
1597–1600.
49 G. A. Halliday, R. J. Young and V. V. Grushin, Org. Lett., 2003, 5, 2003–
2005.
50 X. Hu, C. Lievens and C.-Z. Li, ChemSusChem, 2012, 5, 1427.
51 M. J. Climent, A. Corma and S. Iborra, Green Chem., 2011, 13, 520–
540.
Green Chem.
This journal is © The Royal Society of Chemistry 2012