Molecular mapping of the acid catalysed dehydration of fructose

Article abstract of DOI:10.1039/c2cc31689g

Several intermediates and different reaction paths were identified for the acid catalysed conversion of fructose to 5-(hydroxymethyl)-2-furaldehyde (HMF) in different solvents. The structural information combined with results of isotopic-labelling experiments allowed the determination of the irreversibility of the three steps from the fructofuranosyl oxocarbenium ion to HMF as well as the analogous pyranose route.

Full text of DOI:10.1039/c2cc31689g

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
ChemComm
Cite this: Chem. Commun., 2012, 48, 5850–5852
www.rsc.org/chemcomm
COMMUNICATION
Molecular mapping of the acid catalysed dehydration of fructosew
´
´
Geoffrey R. Akien,z Long Qi and Istvan T. Horvath*
Received 7th March 2012, Accepted 19th April 2012
DOI: 10.1039/c2cc31689g
Several intermediates and different reaction paths were identified
for the acid catalysed conversion of fructose to 5-(hydroxymethyl)-
2-furaldehyde (HMF) in different solvents. The structural information
combined with results of isotopic-labelling experiments allowed
the determination of the irreversibility of the three steps from
the fructofuranosyl oxocarbenium ion to HMF as well as the
analogous pyranose route.
The identification of the ‘‘half of the NMR signals’’ of (4S,5R)-
4-hydroxy-5-hydroxymethyl-4,5-dihydrofuran-2-carbaldehyde
(5 in Scheme 1) supported the intermediacy of 1b and 1c.17
We report the first in situ characterisation of four intermediates
in the acid catalysed dehydration of fructose, and the identifi-
cation of the different reaction paths to HMF and some of the
side products. In addition, the facile and reversible formation of
2,6-anhydro-b-D-fructofuranose (3) was observed. This might
result in an intrinsic mechanistic limitation of selective HMF
synthesis by providing an alternative connection between the
preferred path to HMF (via furanoses 1b and 1c) and that of the
formation of unwanted side-products (via pyranoses 1d and 1e).
DMSO has been reported to be one of the better solvents for
the conversion of carbohydrates to HMF.18,19 We have therefore
first investigated the dehydration of fructose in DMSO (and/or
DMSO-d6) in the absence and presence of sulfuric acid as a
catalyst. All intermediates were characterised by in situ NMR.
For species with particularly low concentrations, it was essen-
tial to use singly 13C-labelled fructoses in conjunction with 2D
techniques to assign the resonances (Tables S1–S7, ESIw).
While fructose can be readily dissolved in DMSO at room
temperature (Fig. S1a, ESIw), it took several days to reach
equilibria between the five isomers (1a: D-fructoketose 2.4%,
Acid catalysed conversion of biomass to carbon based chemicals1,2
including liquid fuels3,4 will be important for the sustainability of
the chemical and pharmaceutical industry,5 and to a lesser extent
the energy industry.6 The largest part of biomass is composed of
mono-, di-, oligo-, and polysaccharides,7 which are key compo-
nents of sugar,8 starch,8 lignocellulose,8 and even some algae.9
Depolymerisation of the glycosidic bonds by hydrolysis can
produce monosaccharides for subsequent conversions.10
A frequently underestimated complication in the conversion
of fructose is that it has three types of structural isomers in
equilibria,8 with each of the two cyclic types having two
diastereoisomers (1a–e, Scheme 1).
Their concentrations in solution depend on the solvent,11
which can in turn be affected by the presence of water.12
The dehydration of carbohydrates to 5-(hydroxymethyl)-2-
furaldehyde has been studied for decades, and many strategies
have been tested to increase the yields.13 One of the key issues has
been the formation of a variety of side-products referred to as
‘‘humins’’,14 a mixture of strongly coloured, soluble and insoluble
oligomers and polymers which have so far eluded detailed
structural characterisation.15 Consequently, the prevention of
their formation has been practically limited to trial and error
approaches, with little molecular insight. Another important issue
was to understand which isomers of fructose (1a–e) can lead to
the formation of HMF. It was proposed that fructose is a
necessary intermediate in the predominant route to convert
glucose to HMF, involving the fructofuranoses 1b and 1c.16,17
Department of Biology and Chemistry, City University of Hong Kong,
83 Tat Chee Avenue, Kowloon, Hong Kong SAR.
E-mail: istvan.t.horvath@cityu.edu.hk; Fax: +852 3442-7402;
Tel: +852 3442-0522
w Electronic supplementary information (ESI) available: Selected
spectra (Fig. S1–S7) and tabulated NMR data for all the intermediates
(Tables S1–S7). Schemes S1–S4 clarify mechanistic points related to
reversibility and alternate pathways to HMF. See DOI: 10.1039/
c2cc31689g
Scheme 1 Molecular isomers of fructose (1a–e) and their acid catalyzed
z Present address: Center for Environmentally Beneficial Catalysis,
1501 Wakarusa Drive, Lawrence, Kansas 66047, USA.
dehydration to HMF and side products.
c
5850 Chem. Commun., 2012, 48, 5850–5852
This journal is The Royal Society of Chemistry 2012

View Article Online
The three steps from 2 leading to HMF are irreversible. This
was demonstrated by the stereospecificity of the intermediates,
and the lack of deuterium incorporation in experiments
where the fructose hydroxyls were previously replaced with
deuterium, or, with the addition of 1.7–22.4 equivalents of
D2O (Scheme S1, ESIw).
The same deuterium labelling experiments also allowed us
to rule out the formation of HMF from the linear isomer 1a.
This requires the formation of the intermediate 3-deoxy-D-
erythro-hexos-2-ulose (12, Scheme S4, ESIw),22 which in turn
requires the acquisition of a proton from solution to form the
CH2 at C-3. Less than 1% deuterium was measured at the C-3
position of HMF.
Fig. 1 Selected 13C NMR spectra for the heating of 2-13C-fructose in
DMSO-d6 solution at 150 1C.
1b: a-D-fructofuranose, 19.6%, 1c b-D-fructofuranose: 46.1%,
1d: a-D-fructopyranose 4.3%, and 1e: b-D-fructopyranose,
27.4%) as shown by 13C NMR of 2-13C-fructose (Fig. S1b,
ESIw). These equilibria can be established much faster at
higher temperatures,19 for example, after heating at 150 1C
for 5 minutes and rapidly cooling the sample, the composition
was 1a: 2.4%, 1b: 18.5%, 1c: 42.9%, 1d: 4.1%, and 1e: 32.2%
(Fig. S2, ESIw). After 5 minutes, 49% of fructose was con-
verted, and the largest new peaks in the 13C NMR spectrum
were at 108.3 and 151.6 ppm (Fig. 1), due to 2,6-anhydro-b-
D-fructofuranose (3, 12.4%) (Fig. S4 and Table S1, ESIw) and
HMF (Table S2, ESIw), respectively. The new resonance at
139.2 ppm was assigned to (2R,3S,4S)-2-(hydroxymethyl)-
5-(hydroxyl-methylene)tetrahydrofuran-3,4-diol (4, 0.3%)
(Fig. S5 and Table S3, ESIw), formed by the deprotonation
of 2.20 The peak at 156.5 ppm was due to 5 (0.6%) (Fig. S6 and
Table S4, ESIw) which was partially characterised previously,17
and was formed by the loss of water from 4. Upon heating for
an additional 20 minutes, the conversion of fructose increased
to 86.2%, while 4 was barely detectable and the concentration
of 5 increased to 0.9%. Furthermore, the concentration of 3
decreased to 1.4%, while the yield of HMF increased to 54.5%. In
addition, we also observed the transient formation of (3S,4R,5R)-
2-(hydroxymethylene)-tetrahydro-2H-pyran-3,4,5-triol (7)
(Table S5, ESIw) and (3R,4S)-3,4-dihydroxy-3,4-dihydro-2H-
pyran-6-carbaldehyde (8) (Fig. S7 and Table S6, ESIw), two
key intermediates for the pyranose route (Fig. 1). The maximum
yield of 7 was 1.9% after 30 minutes, which was gradually
converted into 8 and then to side-products such as humins, since
unlike 5, 8 cannot readily aromatize to form a stable species.
Peaks due to at least six difructose dianhydrides were also
observed (DFAs, 9a–f, up to 12.5%) (Fig. S3, ESIw).21 The
chemical shifts of one of the DFAs, a-D-fructofuranose
b-D-fructofuranose 1,20:2,30-dianhydride (9a) (Scheme 1), were
identified by comparison with an authentic sample (Table S7,
ESIw). After 4 hours, the major detectable species were 9a–f
(4.3%) and HMF (79%). Although heating the reaction
mixture for a total of 8 hours resulted in the complete
disappearance of 9a–f, the HMF yield also decreased. The
relative reactivity of DFAs compared to fructose was studied
by heating a sample of 9a in DMSO. After 4 hours the HMF
yield was only 8.7%, although this increased to 56.7% after
16 hours heating.
Although some mechanisms have shown that 4 must first
tautomerise to form a 2,5-anhydro-D-aldofuranose,16 we could
detect only trace quantities of it in the later stages of the
reaction. An authentic sample was found to be more robust
than fructose, requiring 16 h to be completely converted and
reach a maximum HMF yield.
No scrambling of any of the carbon atoms of all the possible
singly 13C-labelled fructoses was observed, indicating that no
C–C bond cleavage occurred during the reaction. Once the
reaction mixture had equilibrated, the relative proportions
of the isomers did not change significantly, indicating that
isomerisation was faster than any of the fructose-consuming
reactions.
Trace amounts of formic acid (FA),23 2-furyl hydroxy-
methyl ketone24 and furfural23 were also detected. Of the
intermediates described, it was only possible to isolate crude
3 and HMF (4 : 1 CH2Cl2 : MeOH), and a mixture of DFAs
(after acetylation) by column chromatography due to their
higher stability in solution and in the presence of silica.25
Since DMSO could decompose to various products including
CH3SO3H and H2SO4 at higher temperatures,26 the in situ
formed acids served as the catalysts for the dehydration.27 This
was confirmed by heating a mixture of fructose, DMSO-d6
and NaHCO3 at 150 1C for 2 hours, after which only trace
quantities of HMF were detected. Further evidence was
provided by studying the reaction in the absence and presence
of added H2SO4 at 120 1C. The HMF yield started to increase
significantly when the concentration of H2SO4 was higher
than 10ꢀ6 mol Lꢀ1 and reached a maximum value of 80% at
10ꢀ2 mol Lꢀ1 H2SO4 (Fig. 2). Higher acid concentrations
resulted in the formation of levulinic acid (LA) and FA.
A wide range of solvents have been tested for the dehydration
of fructose, so our study was broadened to include other polar
aprotic systems (dimethylacetamide (DMA)/LiCl,28 1-butyl-
3-imazolium chloride (BMIMCl), g-valerolactone (GVL)/
BMIMCl mixtures and an aqueous acidic (D2O/D2SO4) system.
In DMA/LiCl, the product distribution is superficially similar
to that in DMSO/DMSO-d6, except that none of the DFAs
were present. This is an interesting result, which suggests that
the chloride anion plays a critical role in either breaking up the
fructose oligomers formed in situ or preventing their formation.
When using unlabelled fructose in BMIMCl or GVL/BMIMCl
mixtures 5 and 7 were easily detectable.
If the reaction was monitored at one minute intervals, 3 and
4 were easily observable as the first two new species (Fig. 1),
followed by the formation of 5, 7, 8 and HMF, as expected.
By contrast in D2O in the presence of 0.18–0.47 mol Lꢀ1
D2SO4 at 95 1C, 3, none of the DFAs, and none of the
intermediates 4, 5, 7, and 8 were observed. It is plausible that
c
This journal is The Royal Society of Chemistry 2012
Chem. Commun., 2012, 48, 5850–5852 5851

View Article Online
fructose to HMF. The structural information combined with
isotopic-labelling allowed the determination of the irreversibility
of the three steps from the fructofuranosyl oxocarbenium ions to
HMF as well as the analogous pyranose route.
The work described in this paper was supported by a grant
from the City University of Hong Kong (Project No. 9380047).
Notes and references
1 J. J. Bozell and G. R. Petersen, Green Chem., 2010, 12, 539–554.
2 T. Werpy and G. Petersen, Top Value Added Chemicals
from Biomass, Vol. 1, NREL/TP-510-35523, 2004.
3 D. M. Alonso, J. Q. Bond and J. A. Dumesic, Green Chem., 2010,
12, 1493–1513.
4 G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006, 106,
4044–4098.
5 H. Clark, F. E. I. Deswarte and T. J. Farmer, Biofuels, Bioprod.
Biorefin., 2009, 3, 72–90.
6 S. Zinoviev, F. Muller-Langer, P. Das, N. Bertero, P. Fornasiero,
¨
Fig. 2 Product yields from fructose in DMSO as a function of H2SO4
concentration.
M. Kaltschmitt, G. Centi and S. Miertus, ChemSusChem, 2010, 3,
1106–1133.
7 F. W. Lichtenthaler, Acc. Chem. Res., 2002, 35, 728–737.
8 J. F. Robyt, Essentials of Carbohydrate Chemistry, Springer-Verlag,
New York, 1st edn, 1998.
in the presence of large quantities of water, the fructosyl
oxocarbenium can easily react with water to reform fructose.
Thus, maintaining a higher concentration of free fructose
increases the rate of side product formation. No deuterium
was incorporated into HMF, consistent with previous work,29
and consistent with the mechanism being largely solvent-
independent. HMF was also hydrated to form levulinic and
formic acids, the formic acid being almost exclusively derived
from the fructose C-1, as expected.30
9 E. W. Becker, Biotechnol. Adv., 2007, 25, 207–210.
10 R. Rinaldi and F. Schuth, ChemSusChem, 2009, 2, 1096–1107;
¨
S. Van de Vyver, J. Geboers, P. A. Jacobs and B. F. Sels,
ChemCatChem, 2011, 3, 82–94.
11 F. Franks, Pure Appl. Chem., 1987, 59, 1189–1202; S. J. Angyal,
Adv. Carbohydr. Chem. Biochem., 1991, 49, 19–35.
12 F. W. Lichtenthaler and S. Ronninger, J. Chem. Soc., Perkin
Trans. 2, 1990, 1489–1497.
13 B. F. M. Kuster, Starch/Staerke, 1990, 42, 314–321; A. D. Kulkarni,
H. M. Modak, S. J. Jadhav and R. Khan, J. Sci. Ind. Res., 1988, 47,
335–339.
Based on our results, we can draw a detailed reaction map
to describe the acid-catalysed conversion of fructose. Its
protonation and dehydration lead to fructosyl oxocarbenium
ions (Scheme 1), among which, the formation of 2 is probably
more energetically favoured than 6.30 Reversible intramolecular
nucleophilic attack of 2 by the terminal OH-6, followed by
deprotonation could yield 3. Alternatively, intermolecular nucleo-
philic attack by other fructose molecules could form DFAs, which
in turn can act as a reversible fructose ‘‘reservoir’’. In the later
stages of the reaction the water concentration is also high enough
to compete for 2 by hydrating it to re-form 1b and 1c.
14 A. Chuntanapum and Y. Matsumura, Ind. Eng. Chem. Res., 2009,
48, 9837–9846.
15 A. Chuntanapum and Y. Matsumura, Ind. Eng. Chem. Res., 2010,
49, 4055–4062; N. Baccile, G. Laurent, F. Babonneau, F. Fayon,
M. M. Titirici and M. Antonietti, J. Phys. Chem. C, 2009, 113,
9644–9654; J. Herzfeld, D. Rand, Y. Matsuki, E. Daviso, M. Mak-
Jurkauskas and I. Mamajanov, J. Phys. Chem. B, 2011, 115, 5741.
16 E. F. L. J. Anet and L. W. Melville, Adv. Carbohydr. Chem., 1964,
19, 181–218; M. J. Antal, W. S. L. Mok and G. N. Richards,
Carbohydr. Res., 1990, 199, 91–109; E. A. Pidko, V. Degirmenci,
R. A. van Santen and E. J. M. Hensen, Angew. Chem., Int. Ed.,
2010, 49, 2530–2534; J. B. Binder, A. V. Cefali, J. J. Blank and
R. T. Raines, Energy Environ. Sci., 2010, 3, 765–771.
17 A. S. Amarasekara, L. D. Williams and C. C. Ebede, Carbohydr.
Res., 2008, 343, 3021–3024.
Alternatively, 2 can be deprotonated to form 4, and then
readily lose water to yield 5, which can then dehydrate to form
HMF. Similar arguments can be made for the conversion of 6
to 7 and 8, but in this case 8 presumably decomposes to side
products such as humins. In the early stages of the reaction,
the concentration of 3 was high, and that of 4, 5 and HMF was
low, showing that the loss of a proton from 2 at O-6 by
breaking the O–H bond is more energetically favoured than
from C-1 by C–H bond cleavage. In the later stages, the
concentrations of both 3 and 4 were low, coinciding with an
increase in the water concentration promoting the reverse
reaction of 2 to 1b or 1c. Experiments with varying water
concentrations did not show significant changes in the rate of
formation of HMF, leading us to the conclusion that under
both anhydrous and ‘‘hydrous’’ conditions the deprotonation
of 2 to form 4 is always the rate limiting step.
18 H. H. Szmant and D. D. Chundury, J. Chem. Technol. Biotechnol.,
1981, 31, 135–145; R. M. Musau and R. M. Munavu, Biomass,
1987, 13, 67–74.
19 B. Schneider, F. W. Lichtenthaler, G. Steinle and H. Schiweck,
Liebigs Ann. Chem., 1985, 2443–2453.
20 P. Koll, E. Steinweg, B. Meyer and J. Metzger, Liebigs Ann. Chem.,
¨
1982, 1063–1067.
21 M. Manley-Harris, G. N. Richards and H. Derek, Adv. Carbohydr.
Chem. Biochem., 1997, 52, 207–266.
22 H. Weenen and S. B. Tjan, ACS Symp. Ser., 1992, 490, 217–231.
23 J. Joseph, C. Baker, S. Mukkamala, S. H. Beis, M. C. Wheeler,
W. J. DeSisto, B. L. Jensen and B. G. Frederick, Energy Fuels,
2010, 24, 5153–5162.
24 R. M. Moriarty, B. A. Berglund and R. Penmasta, Tetrahedron
Lett., 1992, 33, 6065–6068.
25 E. Suarez-Pereira, E. M. Rubio, S. Pilard, C. O. Mellet and J. M. G.
Fernandez, J. Agric. Food Chem., 2010, 58, 1777–1787.
26 T. M. Santosusso and D. Swern, J. Org. Chem., 1976, 41, 2762.
27 W. Moody and G. N. Richards, Carbohydr. Res., 1982, 108, 13–22.
28 J. B. Binder and R. T. Raines, J. Am. Chem. Soc., 2009, 131, 1979–1985.
29 M. S. Feather, J. F. Harris, R. S. Tipson and H. Derek, Adv.
Carbohydr. Chem. Biochem., 1973, 28, 161–224.
The concentrations of intermediates 4, 5, 7 and 8 were too
low to be detected in the presence of added acid, indicating
that their dehydration is even more sensitive to acid catalysis
than the rate-limiting deprotonations of 2 and 6.
In conclusion, several intermediates and different reaction
paths were identified for the acid catalysed conversion of
30 J. W. Larsen and S. Ewing, J. Am. Chem. Soc., 1971, 93, 5107–5111.
c
5852 Chem. Commun., 2012, 48, 5850–5852
This journal is The Royal Society of Chemistry 2012