Effect of water on hydrolytic cleavage of non-terminal α-glycosidic bonds in cyclodextrins to generate monosaccharides and their derivatives in a dimethyl sulfoxide-water mixture

Article abstract of DOI:10.1021/jp412628y

Hydrolytic cleavage of the non-terminal α-1,4-glycosidic bonds in α-, β-, and γ-cyclodextrins and the anomeric-terminal one in d-maltose was investigated to examine how the cleavage rate for α-, β-, and γ-cyclodextrins is slower than that for d-maltose. Effects of water and temperature were studied by applying in situ ¹³C NMR spectroscopy and using a dimethyl sulfoxide (DMSO)-water mixture over a wide range of water mole fraction, x_w = 0.004-1, at temperatures of 120-180 °C. The cleavage rate constant for the non-anomeric glycosidic bond was smaller by a factor of 6-10 than that of the anomeric-terminal one. The glycosidic-bond cleavage is significantly accelerated through the keto-enol tautomerization of the anomeric-terminal D-glucose unit into the d-fructose one. The smaller the size of the cyclodextrin, the easier the bond cleavage due to the ring strain. The remarkable enhancement in the cleavage rate with decreasing water content was observed for the cyclodextrins and D-maltose as well as D-cellobiose. This shows the important effect of the solitary water whose hydrogen bonding to other water molecules is prohibited by the presence of the organic dipolar aprotic solvent, DMSO, and which has more naked partial charges and higher reactivity. A high 5-hydroxymethyl-2-furaldehyde (5-HMF) yield of 64% was attained in a non-catalytic conversion by tuning the water content to x_w = 0.30, at which the undesired polymerization by-paths can be most effectively suppressed. This study provides a step toward designing a new optimal, earth-benign generation process of 5-HMF starting from biomass.

Full text of DOI:10.1021/jp412628y

Article
pubs.acs.org/JPCA
Effect of Water on Hydrolytic Cleavage of Non-Terminal α‑Glycosidic
Bonds in Cyclodextrins To Generate Monosaccharides and Their
Derivatives in a Dimethyl Sulfoxide−Water Mixture
Hiroshi Kimura,† Masaki Hirayama,† Ken Yoshida,*,† Yasuhiro Uosaki,† and Masaru Nakahara‡
†Department of Chemical Science and Technology, Faculty of Engineering, University of Tokushima, 2-1 Minamijosanjima-cho,
Tokushima 770-8506, Japan
‡Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan
S
* Supporting Information
ABSTRACT: Hydrolytic cleavage of the non-terminal α-1,4-
glycosidic bonds in α-, β-, and γ-cyclodextrins and the anomeric-
terminal one in D-maltose was investigated to examine how the
cleavage rate for α-, β-, and γ-cyclodextrins is slower than that for D-
maltose. Effects of water and temperature were studied by applying
in situ 13C NMR spectroscopy and using a dimethyl sulfoxide
(DMSO)−water mixture over a wide range of water mole fraction,
xw = 0.004−1, at temperatures of 120−180 °C. The cleavage rate
constant for the non-anomeric glycosidic bond was smaller by a
factor of 6−10 than that of the anomeric-terminal one. The
glycosidic-bond cleavage is significantly accelerated through the
keto−enol tautomerization of the anomeric-terminal D-glucose unit
into the D-fructose one. The smaller the size of the cyclodextrin, the
easier the bond cleavage due to the ring strain. The remarkable enhancement in the cleavage rate with decreasing water content
was observed for the cyclodextrins and D-maltose as well as D-cellobiose. This shows the important effect of the solitary water
whose hydrogen bonding to other water molecules is prohibited by the presence of the organic dipolar aprotic solvent, DMSO,
and which has more naked partial charges and higher reactivity. A high 5-hydroxymethyl-2-furaldehyde (5-HMF) yield of 64%
was attained in a non-catalytic conversion by tuning the water content to xw = 0.30, at which the undesired polymerization by-
paths can be most effectively suppressed. This study provides a step toward designing a new optimal, earth-benign generation
process of 5-HMF starting from biomass.
why it is. For the purpose, we focus on cyclodextrins in the
present study as the model substrate before going to
substantially longer polymer chains. β-Cyclodextrin is com-
posed of the seven-membered closed ring of D-glucose units
linked by α-1,4-glycosidic bond and therefore contains only the
non-terminal glycosidic bonds. Owing to its highly symmetric
structure, the closed-ring and the open-ring heptamer produced
by one glycosidic-bond cleavage can be distinghished from each
other by 13C NMR spectroscopy. This enables us to distinguish
the cleavage position of glycosidic bond and to analyze the
difference in the kinetics between the non-terminal and the
anomeric-terminal glycosidic-bond cleavages, which has not
been quantitatively discussed so far in the polysaccharide
decomposition.8−17 The α-1,4-glycosidic bond is the one
involved in starch and malto-oligosaccharides, and thus the
present study will lead to an insightful step toward the practical
1. INTRODUCTION
Hydrolytic cleavage of the α- and β-glycosidic bonds is essential
for the conversion of biomass-derived polysaccharides into
valuable chemicals through the depolymerization. In our
previous studies,1−4 we have developed a non-catalytic method
for the hydrothermal decomposition of oligosaccharides (di-,
tri-, and tetraose) into a monosaccharide D-glucose and its
dehydrated derivatives through the elimination of the
anomeric-terminal D-glucose unit by monitoring the observable
elementary processes using the in situ NMR. Also it has been
found that the glycosidic-bond cleavage of D-cellobiose is
remarkablly sensitive to the water content in DMSO−water
mixtures.4 In particular, a key role is played by the “solitary
water” that is not hydrogen-bonded to other water molecules as
in supercritical water clusters with the molecular partial charges
more naked.5−7 Toward the full understandings of the
decomposition of biopolysaccharides whose glycosidic bonds
are mostly non-terminal, light should also be shed on the
cleavage of the non-terminal glycosidic bond as well as the
anomeric-terminal one. We should ask to what extent the non-
terminal cleavage is more difficult than the terminal one and
Received: December 25, 2013
Revised: February 3, 2014
Published: February 17, 2014
© 2014 American Chemical Society
1309
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
application of the hydrothermal method to the conversion of
carbohydrate-rich biomass into value-added chemicals.
D-maltose monohydrate (>98%) and D-cellobiose (>98%) from
Wako, and deuterated dimethyl sulfoxide (DMSO, 99.8 atom %
D) and water (99.95 atom % D) from ISOTEC. These were
used without further purification. The concentrations of these
oligosaccharides were prepared at 1.0 mol dm−3 (M) in a
DMSO−water mixture in terms of the monomeric D-glucose
unit unless otherwise stated: α-cyclodextrin (0.167 M), β-
cyclodextrin (0.143 M), γ-cyclodextrin (0.125 M), D-maltose
monohydrate (0.5 M), and D-cellobiose (0.5 M). The water
content is denoted in terms of the mole fraction of water, xw.
The xw value was varied in the range 0.004−1. Here the
impurity water contents in the substrates were determined by
use of 1H NMR as follows: 0.4 wt % (α-cyclodextrin), 0.7 wt %
(β-cyclodextrin), 0.5 wt % (γ-cyclodextrin), 0.4 wt % (D-
maltose monohydrate), and 0.4 wt % (D-cellobiose), corre-
sponding to the xw values of 0.003, 0.004, 0.003, 0.003, and
0.003, respectively. The solvent DMSO also contained impurity
water of 0.02 wt %, corresponding to the xw value of 0.001. The
sample with the lowest xw value included only water present
originally in the substrate oligosaccharide and the solvent
DMSO. D-Maltose was offered as monohydrate and the hydrate
water amounts to 5 wt %, which corresponds to the xw value of
0.05. The xw values are denoted by the initial values at the start
of the reaction; during the reaction, the water content varies
with time due to the hydrolysis of the glycosidic bond and the
dehydration into 5-HMF and anhydrosugars, and this increases
the xw value by 0.09 at most.
Water is required for the hydrolytic cleavage of the glycosidic
bonds into monocarbohydrates, C6H12O6. An interesting
question is in which state water is more reactive as a reactant,
in the bulk or in the solitary. The solitary water is more reactive
in the glycosidic-bond cleavage in a disaccharide D-cellobiose
because of the extensive breakdown of the hydrogen bonds due
to the presence of the dipolar aprotic solvent.4 Moreover, for
the purpose of the high-yield production of such target
chemicals as 5-hydroxymethyl-2-furaldehyde (5-HMF) and
anhydrous saccharides, the DMSO−water mixture, in which
the water−water hydrogen bonds are dramatically broken by
the dipolar aprotic solvent, was found to be a promising binary
mixture solvent. 5-HMF is a center of focus today as a biomass-
derived valuable that can be converted into biofuels, fine
chemicals, and polymers.18−26 In fact, the hydrothermal
dehydration (partial coking) of carbohydrates can largely
enhance the combustion energy density per unit mass.27−29
We have established that the highest 5-HMF yield is obtained
in the water mole fraction range 0.20−0.30 because this region
gives the optimal condition for the overall effects of the
competing water-dependent factors, such as the reactivity of the
water molecule as a reactant of the hydrolysis, the
tautomerization of the anomeric-terminal unit from the glucose
type into the fructose one, and the undesired by-paths into
polymeric species via anhydrous saccharides. A question raised
here is whether or not the optimization strategy for a
disaccharide can be directly applicable to oligosaccharides
with non-terminal carbons. Because the degradation of the
larger oligomers into monomers involves the slow cleavage of
non-terminal glycosidic bonds, for the larger oligomers the
more attentions are needed to the weights of the pathways in
competition during the glycosidic-bond cleavage. By taking
advantage of the in situ 13C NMR method,1−4 we quantitatively
monitor all the reactant and products as a function of the
reaction time.
One of the most important features of the role of water in
the DMSO−water mixture disclosed previously4 is the high
reactivity of the water at a low water content. When the water is
almost isolated at the water mole fraction xw of 0.007 as in the
state of “solitary water”, the cleavege of the glycosidic bond in
D-cellobiose is 4−5 times faster than that in pure water. It has
been known that the α-1,4-glycosidic bond is broken 3−4 times
faster than the β-1,4-glycosidic bond in pure water.2 The
difference in the reaction rate in pure water may arise from the
difference between the geometries of the α-1,4- and β-1,4-
glycosidic bonds that determines the hydrogen-bonding
configuration in the vicinity of the glycosidic bond, and
accordingly, the hydrogen-bonding state of the reactant water
in contact with the glycosidic bond is dependent on the
glycosidic-bond geometry (α-1,4 or β-1,4 type). To further
scrutinize the hydrolysis rate in relation to the state of water, it
is of great interest to examine the hydrolysis kinetics as a
function of the water content in the DMSO−water mixture.
Here we examine whether or not the difference in the cleavage
rate for α-1,4 or β-1,4 types of disaccarides can be observed also
when the water content is lowered and the state of water
becomses more solitary.
The apparatus and the experimental procedures are
described elsewhere.4 The solution prepared was loaded into
a Pyrex NMR tube (SHIGEMI, 5.0 or 10.0 mm o.d.). The
sample tube was flame-sealed after the tube inside was purged
1
with argon. The reaction was monitored by the in situ H and
13C NMR measurements. The temperature was set to 120−180
°C in the NMR probe and controlled within 1 °C during the
course of the reaction. The temperature increment up to the
target temperature from 20 °C below takes less than 1 min,
which is short enough and does not affect the discussion
concerning the reaction kinetics; the heat-up period at each
temperature can be neglected because the reaction rate totally
differs by a factor of 3−4 with a temperature difference of 20
°C. In the temperature range, the pressure of the reaction
system is not very far away from the ambient, so there is no
pressure effect on the reaction rates; in fact, the vapor pressure
for pure DMSO (bp 189 °C) is below 0.1 MPa and that for
pure water is 0.2−1.0 MPa.
3. RESULTS AND DISCUSSION
First, we describe the product distribution for the cyclodextrin
and D-maltose reactions in DMSO−water mixtures by taking
advantage of the high-resolution in situ NMR measurements
and establish the reaction scheme. Second, we discuss the
difference between the non-terminal and the anomeric-terminal
glycosidic bonds in the kinetics for the hydrolytic bond
breakage. We further show how the effect of water content
varies with the types of glycosidic bonds. Finally, we present the
optimal strategy for the production of the target compound 5-
HMF by tuning the water content.
3.1. 13C Spectral Analysis. Let us start with the description
of how the product species can be quantitatively analyzed by
using the in situ NMR method as the basis of the kinetic
analysis of the glycosidic-bond cleavage rate. The 13C NMR
chemical shift is so sensitive to the subtle differences in the
molecular structure that the carbon atomic sites in the
2. EXPERIMENTAL SECTION
A series of cyclodextrins, α-cyclodextrin (99%), β-cyclodextrin
(99%), and γ-cyclodextrin (99%) were purchased from Nacalai,
1310
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
cyclodextrins (ring oligomers), linear oligomers with different
D-glucose units, and D-glucose monomer can be distinguished
from each other.
3.1.1. Ring-Opening of Cyclodextrin. One of our main
purposes is to determine the cleavage rate of the non-terminal
glycosidic bond by monitoring the ring-opening of the
cyclodextrins and their subsequent conversions into linear
oligosaccharide intermediates. The molecular structure and the
in situ 13C NMR spectra of β-cyclodextrin are shown in Figures
1 and 2, respectively. In Figure 2a, we show the 13C spectra in
Figure 2. In situ 13C spectra for β-cyclodextrin in the DMSO−water
mixture at 140 °C with xw = 0.30 at the reaction time of (a) 0 h and
(b) 3 h. In panel b, the vertical axis is enlarged by a factor of 20.
oligosaccharides. This is the key advantage of the use of the
cyclodextrin for the spectral and kinetic analysis. The clear
distinguishability allows us to determine the cleavage rate of the
non-terminal glycosidic bonds.
For a better understanding of the oligosaccharide fragmenta-
tion scheme, let us explain the notations of the D-glucose units
in the malto-oligosaccharides introduced in Figure 1; the
notations are common to those in ref 2. The monomer unit of
D-glucose is simply expressed as G. The D-glucose unit is
numbered in order from the (rightmost) anomeric terminal to
the opposite non-anomeric terminal. The location of the ith
unit from the anomeric terminal in the malto-oligosaccharides
composed of n units is expressed as i/n in the superscript; see
Figure 1b. The anomeric-terminal unit can take α- and β-types
through isomerization via the ring-opening of a D-glucose unit
and its isomeric form is specified by the subscript; for example,
the pyranose (six-membered ring) of α-type is expressed as 6-α.
In the present study, the fractions of the furanoses (five-
membered ring) are below the detection limit. For the units
other than the anomeric terminal, the notation of the isomeric
form is omitted for brevity because the unit at any position is of
the 6-α type.2
We indicate the positions of the D-glucose units and the
glycosidic bonds also in terms. The correspondence between
the terms and the numbering is given by taking the heptamer as
an example as follows. For the positions of the D-glucose units,
“anomeric” or “anomeric-terminal”, indicates 1/7G, “non-
anomeric terminal” indicates 7/7G, which is the opposite end
of the anomeric terminal, and “non-terminal” indicates the
internal units of 2/7G to 6/7G. The glycosidic bonds are
categorized according to whether or not the bond is connected
to the anomeric-terminal D-glucose unit. The bond connects
the anomeric-terminal unit with the next one (2/7G−1/7G) and
is called an “anomeric-terminal” glycosidic bond, and the other
bonds, 7/7G−6/7G to 3/7G−2/7G, are called “non-anomeric
terminal” or “internal” bonds.
Figure 1. Notations for the D-glucose (G) units in the malto-
oligosaccharides: (a) β-cyclodextrin (β-CD) and (b) D-maltoheptaose.
For β-CD, seven D-glucose units are chemically and magnetically
equivalent to each other, and the position of the D-glucose unit is not
distinguishable. For D-maltoheptaose, the position of the D-glucose
unit is denoted in the superscripts as 1/7G−7/7G. The numbering starts
from the (rightmost) anomeric-terminal unit that can be isomerized or
tautomerized with the ring-opening of the D-glucose unit. The carbon
atomic sites in the D-glucose unit are numbered as shown in the figure.
the chemical-shift region of the C1 carbons of saccharides; the
numbering of the carbon atoms of the D-glucose unit is
indicated in Figure 1. The initial spectrum in Figure 2a was
taken at 0 h, i.e., immediately after the sample was heated to the
reaction temperature of 140 °C. There is observed a single peak
for the C1 carbon of β-cyclodextrin due to the high symmetry.
The spectrum at 3 h is shown in Figure 2b. At this reaction
time, 18% of β-cyclodextrin remained uncleaved and the
fragmentation of D-maltoheptaose into smaller malto-oligosac-
charides and D-glucose only partially proceeded; the C1 peaks
for the monomer D-glucose and those for linear malto-
oligosaccharides in addition to the remaining reactant of β-
cyclodextrin are simultaneously observed. Here it should be
noted that the C1 peaks for the non-anomeric units of the
linear malto-oligosaccharides are completely separated from
that for β-cyclodextrin, reflecting the subtle differences in the
intramolecular moieties between the linear and ring malto-
3.1.2. Fragmentation of Linear Malto-oligosaccharides.
After the ring-opening of β-cyclodextrin, the fragmentation of
1311
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
D-maltoheptaose into smaller oligomers proceeds. Let us take a
close look at the 13C NMR spectra for the dimer (D-maltose) to
the heptamer (D-maltoheptaose) used for the quantitative
analysis of the fragmentation. Figure 3 shows the enlarged view
fragmentation of the oligosaccharides. As can be seen from the
13C spectra at different reaction times shown in Figure 4, the
Figure 4. Time evolution of the spectrum of the C1 carbons of the
non-anomeric terminal and the non-terminal during the course of the
fragmentation of β-cyclodextrin into smaller oligosaccharides at 120
°C with xw = 0.30. The spectrum at 5 h was obtained at 30 °C after
quenching the reaction so that the peaks with high signal-to-noise ratio
can be detected (see Figure SI-1 in the Supporting Information for the
corresponding in situ spectrum at 120 °C). The initial concentration
of β-cyclodextrin is 0.286 M (corresponding to 2.0 M of the
monomeric D-glucose unit). The peaks at 101.0 and 100.6−100.9 ppm
are respectively assigned to the non-anomeric terminal (n/nG) and
non-terminal D-glucose units (2/nG to (n‑1)/nG; n = 3−7) for the
mixture of the dimer to the heptamer. The reaction times and the
concentrations of the non-anomeric D-glucose units are indicated in
the figure. The spectral intensities are normalized so that the integral
of the peaks at 100.6−101.0 ppm at each reaction time is the same.
Figure 3. Enlarged view of the 13C spectra for the C1 carbons of the
non-anomeric terminal and the non-terminal D-glucose units for a
series of malto-oligosaccharides as the authentic species: (a) D-maltose,
(b) D-maltotriose, (c) D-maltotetraose, and (d) D-maltoheptaose.
Spectra a, b, and c are obtained at 100 °C in pure water, taken from ref
2. Spectrum d is obtained at 80 °C in the DMSO−water mixture with
xw = 0.30. These spectral intensities are normalized so that the integral
of the peaks at 100.6−101.0 ppm at each reaction time is the same.
The details for the peak assignment have been described in ref 2.
relative intensity of the non-anomeric terminal unit at 101.0
ppm increases with the reaction time; the ratio of the non-
anomeric terminal peak to the non-terminal peak is 0.16, 0.20,
0.25, and 0.32, respectively, at 5, 20, 60, and 115 h, respectively.
The time evolution of the fraction θ will be further discussed on
the basis of the rate equations in section 3.2.2.
of the region of the non-anomeric C1 carbon of the malto-
oligosaccharides. The sharp peak at 101.0 ppm, observed in all
of the dimer to the heptamer, is assigned to the C1 carbon of
the non-anomeric terminal (n/nG; n = 2−7). In the dimer
spectrum in panel a, there is only one peak for the non-
anomeric terminal (2/2G), and in the trimer spectrum in panel
b, there are observed two clearly separated peaks for the non-
anomeric terminal (3/3G) and the non-terminal (2/3G). In the
spectra c and d for the tetramer and the heptamer, respectively,
the peaks for the non-terminal D-glucose units (2/4G and 3/4G
for the tetramer and 2/7G to 6/7G for the heptamer) are
observed in the range 100.6−100.9 ppm and they are separated
from the peak for the non-anomeric terminal unit (4/4G and
7/7G) at 101.0 ppm. Thus, for all the oligomers from dimer to
heptamer, the non-anomeric terminal unit can be distinguished
from the non-terminal units. The fraction of the non-anomeric
terminal unit can be quantified from the relative intensity of the
peak at 101.0 ppm in the entire range of C1 carbons of 100.6−
101.0 ppm. Here, the fraction is denoted as θ, and θ is
expressed in terms of the D-glucose unit notations introduced
3.2. Glycosidic-Bond Cleavage. 3.2.1. Cleavage Rates of
Non-Terminal Glycosidic Bonds of Cyclodextrins. In Figure
5a,b, we show the time evolutions of the concentrations of the
reactant and products of the β-cyclodextrin reactions at 120 and
180 °C, respectively, with xw = 0.30. At the water content of xw
= 0.30, the 5-HMF production yield from D-cellobiose has been
optimized;4 the effect of the water content will be discussed in
section 3.3.2. The lower temperature of 120 °C was employed
to obtain the time evolution with high time resolution with the
sufficient number of NMR signal acquisitions. The progress of
the reaction in Figure 5a corresponds to that in the early time
region up to ∼2 h at the higher temperature of 180 °C in
Figure 5b. As seen in Figure 5a, the amount of β-cyclodextrin
decreases monotonically due to the ring-opening caused by the
glycosidic-bond cleavage. It is noteworthy that D-glucose is
observed in nearly an equal molar quantity to the linear malto-
oligosaccharides. The early ratio of 1:1 corresponds to the D-
glucose and the residual oligomeric fragments. The simple ratio
is maintained up to ∼60 h at 120 °C (Figure 5a) because the
glycosidic-bond cleavage of the anomeric-terminal unit is by far
faster compared to the non-terminal and non-anomeric
terminal ones due to the tautomerization of the glucose-type
above as θ = ∑7n=2 n/nG]/{∑7n=2(∑ni=2[i/nG])}. Because the
[
fraction θ of the non-anomeric terminal unit increases with
decreasing molecular size of the oligomers, θ can be utilized to
quantitatively evaluate the time evolution of the hydrolysis
1312
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
opening of β-cyclodextrin occurs when any of the seven
glycosidic bonds is cleaved. In other words, the hydrolysis
cleavage rate of a single glycosidic bond in β-cyclodextrin is 1/7
of the decreasing rate of the β-cyclodextrin concentration when
we make a comparison with that in D-maltose. This is expressed
in terms of the rate equations for the decomposition of the
cyclodextrins and D-maltose as follows:
d[ξ‐CD]
= −nkξ‐CD[ξ‐CD]
(1)
dt
d[{D‐maltose}]
= −kmal[{D‐maltose}]
(2)
dt
where kξ‑CD and kmal are the first-order rate constants for a
cyclodextrin (CD) and D-maltose, n is the number of the D-
glucose units in the cyclodextrin, which is 6, 7, and 8 when ξ is
α, β, and γ, respectively, t is the reaction time, and the square
brackets express the concentration. The curly brackets in eq 2
indicate that all the possible isomers are treated collectively
because the mutual isomerizations of D-maltose are rapid
enough to be treated collectively in the pre-equilibrium states as
confirmed for other sugars in the previous studies.1−4 By this
definition, the effect of the water concentration is implicitly
included in the rate constants. The rate constants are
determined by using the data within the time region where
neither the acidic byproducts nor the mass-balance loss due to
the polymerization need to be taken into account, typically up
to 2−12 h, depending on the reaction temperature and the
water content.
The rate constants, kβ‑CD and kmal, were determined at the
water mole fraction of xw = 0.30 in the temperature range 120−
180 °C, as summarized in Table 1. The values of kmal are larger
Table 1. Rate Constants for Hydrolytic Glycosidic-Bond
Cleavage of β-Cyclodextrin (kβ‑CD) and D-Maltose (kmal) in a
DMSO−Water Mixture with xw = 0.30 at Temperatures of
120−180 °C
T/°C
k
β‑CD/s−1
kmal/s−1
120
130
140
150
180
(5.8 0.2) × 10−7
(1.2 0.1) × 10−7
(4.4 0.5) × 10−6
(7.0 0.4) × 10−6
(2.5 0.3) × 10−5
(6.0 0.2) × 10−6
(1.1 0.1) × 10−6
(4.1 0.3) × 10−5
(5.1 0.7) × 10−5
(1.5 0.3) × 10−4
Figure 5. (a, b) Time evolutions of the concentrations of the reactant
and products for the reactions of β-cyclodextrin at 120 and 180 °C,
respectively, with xw = 0.30. (c) Time evolutions of the concentrations
of the reactant and products for the reaction of D-maltose at 180 °C
with xw = 0.30. The vertical axis on the left shows the normalized
concentration, and that on the right shows the mass balance. The
normalized concentration denotes the concentration of the compound
of interest divided by the initial substrate concentration. The mass
balance denotes the ratio of the sum of the amount of carbon atoms in
the reactant and the products detected by 13C NMR at each reaction
time divided by the carbon amount in the reactant in the beginning of
the reaction.
by a factor of 6−10 than those of kβ‑CD over the temperature
range examined. The cleavage of the anomeric-terminal
glycosidic bond in D-maltose proceeds more rapidly than that
of the non-terminal one in β-cyclodextrin. This is in harmony
with the previous kinetic results for the hydrothermal
decompositions of the linear malto- and cello-oligosaccharides
into the monomers via the keto−enol tautomerization of the
anomeric-terminal D-glucose unit into the D-fructose one.2
To see the effect of the ring size on the glycosidic-bond
cleavage rate, we made a comparison of the reactions of α-, β-,
and γ-cyclodextrins (six-, seven-, and eight-membered,
respectively) at 180 °C with xw = 0.30; the time evolutions
are given in Figure SI-2 in the Supporting Information. The
ring-size dependence was found to be rather weak. The
determined rate constants per glycosidic bond for α-, β-, and γ-
cyclodextrins are (2.9 0.4) × 10−5, (2.5 0.3) × 10−5, and
(2.1 0.3) × 10−5 s−1, respectively. A relatively small effect of
the ring size can be understood to be due to the ring strain,
anomeric-terminal unit into the more easily cleavable fructose-
type one.2,4
There exists a large difference in the cleavage rate between a
non-terminal glycosidic bond and an anomeric-terminal one.
This is clearly seen in the comparison of the hydrolysis rate of
β-cyclodextrin with that of D-maltose in Figure 5c. The decrease
in the concentrations of β-cyclodextrin (seven non-terminal
bonds) and D-maltose (one anomeric-terminal bond) occurs on
a similar time scale. This means that the cleavage of the
anomeric-terminal glycosidic bond in D-maltose is faster than
that in β-cyclodextrin. Here it is important to note that the ring-
1313
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
which is larger for the smaller cyclodextrin. More importantly,
the difference between the non-terminal and the anomeric-
terminal glycosidic bonds is 1 order of magnitude larger than
the effect caused by the ring strain. The tautomerization at the
anomeric-terminal unit plays the essential role in controlling
the glycosidic-bond cleavage.
3.2.2. Cleavage Rates of Non-Terminal Glycosidic Bonds
in Linear Malto-Oligosaccharides. Let us move on to the
kinetics of the hydrolytic fragmentation of the linear D-
maltoheptaose generated from β-cyclodextrin by the ring-
opening. The degree of the fragmentation can be evaluated by
the quantity of the fraction of the terminal unit in the total units
other than the anomeric-terminal one; in terms of the notations
introduced in section 3.1.1, the fraction is equal to the ratio of
the non-anomeric terminal D-glucose (n/nG) to the non-
anomeric D-glucose units (2/nG to n/nG; n = 2−7). This
quantity can be experimentally determined from the 13C NMR
spectrum for the C1 carbon of the non-anomeric units in terms
Figure 7. Time dependence of the fraction θ of the non-anomeric
terminal D-glucose (n/nG) among the non-anomeric D-glucose units
(
2/nG to (n−1)/nG; n = 2−7) for the mixture of the dimer to the
heptamer during the course of the decomposition of β-cyclodextrin
into smaller oligomers. The open symbols are the experimental results,
and the solid curve is the fitting result. The broken curve denoted as
internal (blue) is the plot of the θ value estimated with kat = 0
representing the contribution of the internal bond cleavage, and that
denoted as anomeric terminal (green) is the result with knon‑at = 0
representing the contribution of the anomeric-terminal bond cleavage.
of the fraction θ = ∑7n=2 n/nG]/{∑n7=2(∑in=2[i/nG])} introduced
[
in section 3.1.2. The quantity θ measures the extent of the
fragmentation into smaller oligomers. The θ values thus
calculated (θcal) for the dimer to the heptamer are in the
range 0.17−1. The θ values independently observed by 13C
NMR (θobs) for the authentic samples of pure dimer to
heptamer are in good agreement with θcal, as shown in Figure 6.
kβ‑CD for β-cyclodextrin are treated as the fixed parameter. For
the k1/n value from the trimer to the heptamer, we employed
the rate constant for D-maltose, kmal (=k1/2). The rate constants
kmal and kβ‑CD were determined independently in the reactions
of D-maltose and β-cyclodextrin, respectively, as the starting
materials. The knon‑at value was determined by numerically
integrating the simultaneous rate equations for all the oligomers
to obtain the time evolution of the fraction θ in Figure 7, where
the mass balance is kept constant over the reaction time, as
shown in Figure 5a. These formulations are completely
satisfactory for the present purpose to discuss the relative
rates of the cleavage of the internal and anomeric-terminal
glycosidic bonds; the validations based on experimental results
are described in the Appendix.
The knon‑at value thus determined at 120 °C is (6.4 0.2) ×
10−7 s−1. This value of knon‑at is close to that of kβ‑CD of (5.8
0.2) × 10−7 s−1 at 120 °C and is 1 order of magnitude smaller
than that of kmal, (6.0 0.2) × 10−6 s−1. This indicates that the
kinetics for the cleavage of the non-terminal glycosidic bond in
the linear oligomers is similar to that in the cyclodextrins, which
supports our kinetic insights that the anomeric-terminal
glycosidic bond is broken more rapidly than the non-anomeric
one. It is to be noted, on the other hand, that the slower
internal-bond cleavage is of significance for the fragmentation
of D-maltoheptaose. In Figure 7, the contributions of knon‑at and
kat to the change in θ are shown as the broken curves; the
contribution of knon‑at (denoted as internal in Figure 7) was
evaluated by setting kat to 0 with knon‑at kept at the value of the
fitting result as determined above, and the contribution of kat
(denoted as anomeric terminal) evaluated vice versa. In the
earlier time region, the time evolution of θ is mainly
determined by the contribution of knon‑at. This can be of
more significance for larger polysaccharides like starch and
cellulose.
Figure 6. Plots of the θ value observed by NMR (θobs) against the
calculated one (θcal). The θobs value was determined by the 13C NMR
integral intensities of the signals of the non-anomeric terminal and
internal D-glucose units. The θcal value was calculated using eq 16 in
the Appendix for each pure malto-oligosaccharide. The number for
each plot in the figure is that of the D-glucose units in the
oligosaccharide and the solid line represents θobs = θcal.
This verifies the quantitative validity of the present analysis for
the fragmentation kinetics of linear oligosaccharides described
below. In Figure 7, the fraction θ for the decomposition of β-
cyclodextrin examined at 120 °C with xw = 0.30 is plotted
against the reaction time. The θ value monotonically increases
with the time from 0.17 in the beginning, and it finally reaches
0.33 after 115 h, at which 67% of β-cyclodextrin is decomposed.
Let us discuss the cleavage rates of the internal and
anomeric-terminal glycosidic bonds that determine the time
evolution of the fraction θ. Here we briefly summarize the
procedure to determine the cleavage rate constant for the non-
anomeric glycosidic bonds; the details are described in the
Appendix. The curve fitting of θ was carried out with only a
single variable parameter of the cleavage rate constant knon‑at for
the non-anomeric glycosidic bond. In the fitting of θ in Figure
7, the cleavage rate constants k1/n for the anomeric terminal and
On the basis of the time evolution analysis, the reaction
scheme for the hydrolytic decomposition of β-cyclodextrin can
be summarized as in Figure 8. First, β-cyclodextrin is
hydrolyzed into D-maltoheptaose through the non-terminal
glycosidic-bond cleavage. Subsequently, the anomeric-terminal
1314
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
D-glucose unit is selectively eliminated from the malto-
oligosaccharides after the tautomerization into the D-fructose
one. The cleavage of the internal bonds proceeds in parallel
with the cleavage of the anomeric-terminal bond, with the
reaction rate per bond 1 order of magnitude slower than that of
the anomeric-terminal one. Though slower, the internal-bond
cleavage plays an important role in particular in the early time
region because of the presence of the larger number of the
internal bonds than the anomeric-terminal ones; the logic is the
same as that for the similar decomposition rates observed for D-
maltose and β-cyclodextrin. The number of anomeric-terminal
D-glucose units increases with the fragmentation via the
internal-bond cleavage, and thus the contribution of the faster
anomeric-bond cleavage becomes more dominant as the
reaction proceeds.
The reaction steps following the monomerization into D-
glucose, corresponding to the time region after ∼3 h in Figure
5b, can be understood essentially in terms of the scheme and
pathways that have been established for D-cellobiose as
discussed elsewhere.4 In these steps, D-glucose is reversibly
transformed into D-fructose via keto−enol tautomerization and
D-fructose is dehydrated into 5-HMF through the precursors of
5-HMF″ (with no double bond) and 5-HMF′ (with one double
bond).3 There are also reversible by-paths into the
anhydromonosaccharides of 1,6-anhydro-β-D-glucopyranose
(levoglucosan) and 1,6-anhydro-β-D-glucofuranose (AGF).
The polymerization of these monosaccharide derivatives into
NMR-undetectable solid-state species competes with the 5-
HMF production. The quantitative analysis of the yields and
rates of the valuable monosaccharide derivatives are described
in section 3.3.
3.2.3. Effect of Water on Cleavage of Non-Terminal and
Anomeric-Terminal Glycosidic Bonds. Panels a and b of Figure
9 show respectively the rate constants for the hydrolytic
cleavage of the non-terminal glycosidic bonds in β-cyclodextrin
(kβ‑CD) and that of the anomeric-terminal glycosidic bond in D-
maltose (kmal) as a function of xw; in panel b, the data for β-1,4-
linked D-cellobiose (kcel) taken from ref 4 are shown for
comparison. As seen in Figure 9a, the cleavage of the non-
terminal glycosidic bond in β-cyclodextrin gets much faster
with decreasing water content. The large reaction rate at the
lower water contents is common to the case of D-cellobiose, as
shown in Figure 9b. Such enhancement of the cleavage rate is
due to the high reactivity of the “solitary water”. The solitary
Figure 8. Non-catalytic conversion scheme from β-cyclodextrin in the
DMSO−water mixture into 5-HMF. The curly brackets denote a set of
all isomers (pyranoses and furanoses of α- and β-types and open-chain
form) of each saccharide. The unit F stands for the monomer unit of
D-fructose. Three arrows represent the repeat of the hydrolytic
cleavage at the anomeric-terminal glycosidic bond after the keto−enol
tautomerization from the original glucose-type oligomer into the
fructose-type one.
Figure 9. (a) Rate constant for the cleavage of the glycosidic bond in β-cyclodextrin (kβ‑CD) at 140 °C plotted against xw. (b) Rate constants for the
cleavage of the glycosidic bonds in D-maltose (kmal) and D-cellobiose (kcel) at 120 °C plotted against xw. The solid and dashed curves are guides for
eyes.
1315
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
water is in the state of water in which the water is isolated from
other water molecules and its partial charges are more naked
because of the absence of the hydrogen bonding to water
molecules nearby.5−7 In the low xw region below ∼0.05, the
contact probability between water molecules is negligibly small,
as evaluated on the basis of the simple estimation of the
coordination number.4
3.2.4. Comparison of Water-Content Dependencies
between α- and β-1,4-Glycosidic Bonds. Here we discuss
the difference in the water-content dependence of the
glycosidic-bond breakage between the α- and β-1,4-linked
disaccharides; the former and the latter have, respectively, the
glycosidic-bond geometry involved in starch and cellulose. As
seen in Figure 9b, the variation of the kcel values for D-cellobiose
with xw is larger than that of the kmal values for D-maltose. The
difference in the water-content dependence can be ascribed to
the differences in the bond orientation and in the bond length
of the C1−O−C4 moiety between the α- and β-1,4-glycosidic
bonds. These structural factors can affect the contact
probability and hydration dynamics of the reactant water.
The α-1,4-glycosidic bond is in the axial direction against the
plane of the sugar ring (hydrophobic), whereas the β-1,4-
glycosidic bond is in the equatorial. The β-1,4-glycosidic bond
length calculated by using ab initio MO method is shorter by
∼0.04 Å than that of the α-1,4-glycosidic bond.2 On the basis of
the sugar-ring orientation and the bond length, we suggest that
the hydrophobic moiety (or the sugar ring) of the β-1,4-linked
D-cellobiose can be located closer to the glycosidic bond than
that of the α-1,4-linked D-maltose. The sugar ring provides a
more or less hydrophobic moiety for the water molecule
approaching the glycosidic bond as a reactant of the hydrolysis.
In pure water (xw = 1), the population of water molecules in
the C1−O−C4 moiety of D-cellobiose might be more or less
smaller than that in D-maltose. This can be one of the reasons
for the slower hydrolytic breakage of glycosidic bond at higher
water contents. The solitary water existing at the lower water
contents, on the other hand, can be distributed more
preferentially in the vicinity of the C1−O−C4 moiety,
regardless of the type of glycosidic bond, α-1,4 or β-1,4 type.
In this situation, there is absent the competing hydrogen bonds
that can disturb the attractive interactions between the solitary
water and the glycosidic-bond oxygen. In other words, the high
reactivity of the solitary water overwhelms the effect of the
glycosidic-bond geometry on the cleavage rate. At lower water
contents, the glycosidic-bond cleavage of D-cellobiose is
therefore markedly enhanced and its rate becomes almost the
same as that for D-maltose.
Figure 10. Water-content dependence of the population ratio, [D-
fructose]/[D-glucose], in the reaction of D-maltose examined at 120
°C. The xw values are indicated in the figure. The data in pure water
(xw = 1) are taken from ref 2.
transformation of D-fructose into D-glucose and the successive
consumption of D-fructose through the conversion into 5-HMF,
which are much faster in DMSO than in water.4 The difference
of the F/G ratio for D-maltose on the water content xw is
almost exactly the same as that for D-cellobiose, as shown in
Figure SI-3 in the Supporting Information. This means that the
tautomerization kinetics and equilibrium of the anomeric unit
between the glucose type and the fructose type is essentially
independent of the geometries of the anomeric-terminal
glycosidic bond.
3.2.5. Temperature Effect. The Arrhenius plots for the rate
constants kβ‑CD and kmal, defined respectively by eqs 1 and 2, are
shown in Figure 11. The Ea value (94 5 kJ mol−1) for the
Figure 11. Arrhenius plots of the rate constants for the cleavage of the
●
○
)
glycosidic bonds in β-cyclodextrin (kβ‑CD
,
) and D-maltose (kmal
,
at xw = 0.30. The activation energies for β-cyclodextrin and D-maltose
are 94 5 and 78 6 kJ mol−1, respectively.
Let us then see the water-content effect on the
tautomerization of the anomeric D-glucose residue in the α-
1,4-linked disaccharide, D-maltose, into the D-fructose one. As
illustrated in Figure 10, the higher the water content, the larger
the D-fructose/D-glucose (F/G) ratio. At the high water
contents of xw = 0.90−1, the F/G ratio in the beginning of
the reaction is over the equilibrium value of 0.11 in pure water
at 120 °C,1 and it decreases with the reaction time and finally
approaches the equilibrium. The large F/G ratio arises from the
glycosidic-bond breakage after the formation of the reactive
intermediate of the fructose-type D-maltose, whose anomeric D-
glucose residue is transformed into the D-fructose one. At the
low water contents of xw = 0.05−0.50, in contrast, the F/G
ratio is as low as 0.02−0.05. In this case, the F/G ratio attains
equilibrium within 1 h. The small F/G ratio at xw = 0.05−0.50
is due to the competing reaction steps, such as the back-
non-terminal glycosidic-bond breakage of β-cyclodextrin is
larger by 16 kJ mol−1 compared with that (78 6 kJ mol−1) for
the anomeric-terminal glycosidic-bond breakage of D-maltose.
This is in harmony with the conclusion stated above that the
transformation of the anomeric-terminal D-glucose unit into the
D-fructose significantly lowers the reaction barrier of the
glycosidic-bond cleavage. Furthermore, the Ea value for D-
maltose linked by α-1,4-glycosidic bond is smaller by 11 kJ
mol−1 when compared to that (89
4 kJ mol−1) for D-
cellobiose linked by β-1,4-glycosidic bond. This is in parallel to
the observation that the rate for the hydrolysis of D-maltose is
faster than that of D-cellobiose.
3.3. Dehydration of D-Glucose and D-Fructose.
3.3.1. Monosaccharide Derivatives. Subsequently to the
1316
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
glycosidic-bond cleavage of β-cyclodextrin and D-maltose, as
seen in the time evolutions shown in Figure 5b,c, respectively,
D-glucose generated is transformed into D-fructose and
successively into 5-HMF via the precursor with no double
bond (5-HMF″) and that with one double bond (5-HMF′) as
in the case of D-fructose transformation into 5-HMF in pure
DMSO.3 After ∼2 h, the anhydromonosaccharides of
levoglucosan and AGF are generated through other dehy-
dration path from D-glucose. These anhydrosugars are
reversibly hydrated into D-glucose or polymerized into NMR-
undetectable species. When the water content is changed, the
product species are common and only their populations are
varied. At the lower water contents, the anhydromonosacchar-
ides of levoglucosan and AGF are produced more, and the
production of the precursors of 5-HMF″ and 5-HMF′ becomes
more favorable. These observations are common to those for D-
cellobiose in the DMSO−water mixture.4 For 5-HMF, the yield
is also dependent on the water content, as described in detail in
the following section.
3.3.2. Water-Content Dependence of 5-HMF Yield. So far,
much attention has been paid to the generation of 5-HMF (5-
hydroxymethyl-2-furaldehyde) from biomass-derived carbohy-
drates.20−26 This is because 5-HMF is regarded as a potential
platform chemical with a wide application profile.18,19 Once we
succeed in obtaining the reactive monomer from such biomass
as starch (edible) and cellulose (inedible), this can meet the
increasing demand for green and sustainable chemistry. In
many studies, rare earth metals and acids were used as catalysts,
and the 5-HMF yields from starch and cellulose achieved by
using such non-green methods are at most 53% and 48%,
respectively.20 It is to be noted that even when a simple
monosaccharide, D-glucose, was used as the reactant, the yield
of 76−80% was achieved.20 Recently, we have found that the
yield of 5-HMF produced from the disaccharide D-cellobiose
can be controlled by tuning the water content in DMSO−water
mixtures and that the yield can be maximized up to ∼70% at
the low water content of xw = 0.20−0.30.4 Here we focus on
the 5-HMF yield from the higher-oligomeric cyclodextrins. As
shown in Figures 5 and SI-2 in the Supporting Information, the
time dependence of the 5-HMF yield reaches a maximum. The
reaction time longer than the maximum time of the yield of 5-
HMF leads to the yield loss into the undesired polymerization
path as revealed previously.4 For the purpose of determining
the optimal water content, we have examined the maximum
yield as a function of the xw value. In Figure 12, the maximum
yield in the reaction of β-cyclodextrin at 180 °C is plotted
against xw. The highest yield of 64% is attained with xw = 0.30
(at 16 h). This yield is essentially the same as the maximum
yield of 66% obtained from the disaccharide D-cellobiose with
xw = 0.30 at the same temperature.4 The yield of 64% is
significantly higher than that of 46% obtained from D-glucose in
hot pure water.4 The loss of the 5-HMF yield at higher water
contents comes mainly from the polymerization enhanced by
water. At the lower water contents, the polymerization of the
anhydromonosaccharides of levoglucosan and AGF lowers the
5-HMF yield. These polymerization paths were found to be
enhanced by a temperature increase in the previous study on
the D-cellobiose decomposition in the DMSO−water mixture.4
A higher maximum yield of 5-HMF could be attained by taking
a lower reaction temperature if the longer reaction time
required accordingly would be acceptable. The maximum yields
of 5-HMF are almost the same in the reactions of the
cyclodextrins and D-maltose. In other words, the maximum
yield is almost independent of the degree of oligomerization up
to the heptamer. When 5-HMF is produced by starting from D-
fructose, the yield is as high as 95% with much less yield loss
into undesirable polymerization by-paths shown in Figure 8.
The glycosidic-bond cleavages of D-cellobiose and D-maltose are
not the rate-determining step of the 5-HMF production
because the time scale is sufficiently smaller than that of the
conversion from D-glucose to 5-HMF. Thus the conversion of
D-glucose into D-fructose is considered to be the rate-
determining step. In the case of the biomass utilization, it is
to be noted that the situation can be dramatically changed by
the slow rate of the cleavage of the non-terminal glycosidic
bonds.
4. CONCLUSIONS
We have investigated the conversions of α-, β-, and γ-
cyclodextrins with the non-terminal glycosidic bond and D-
maltose with the anomeric-terminal one into valuable 5-HMF
in the DMSO−water mixture. By taking advantage of the in situ
13C NMR spectroscopy to determine the product time
evolution with the mass balance confirmed, we have
quantitatively evaluated the difference in the cleavage kinetics
between the non-terminal and the anomeric-terminal glycosidic
bonds and showed how to control the pathways and kinetics by
tuning the water content. The cleavage rate constants for the
non-terminal glycosidic bonds of both cyclodextrins and linear
malto-oligosaccharides have been found to be significantly small
compared to that for the anomeric-terminal glycosidic bond
because the tautomerization of the D-glucose unit into the more
reactive, easily cleavable D-fructose unit is impossible for the
monomer units in the cyclodextrins. The glycosidic-bond
cleavage is faster at the lower water content, because of the high
reactivity of solitary water molecules with the large partial
charges more naked. The dependency on the water content is
larger for the β-1,4-glycosidic bond in D-cellobiose than for the
α-1,4-glycosidic bond in D-maltose. Such a difference arising
from the type of glycosidic bond (α or β) can be interpreted as
the hydration states of water controlled by the atomic-site
geometries of bond orientation and bond length of the C1−O−
C4 moiety. The 5-HMF yield of 64% obtained from β-
cyclodextrin was as high as those of the disaccharides, D-maltose
and D-cellobiose. The highest 5-HMF yield for both β-
cyclodextrin and D-maltose was achieved at the relatively
poor water content of xw = 0.30 as in the case of D-cellobiose, at
which the polymerization by-paths via 5-HMF and anhydro-
monosaccharide were effectively suppressed. The optimal
Figure 12. Maximum yield of 5-HMF in the reaction of β-cyclodextrin
at 180 °C plotted against xw.
1317
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
condition was achieved at a moderate temperature of 180 °C
without using any catalysts. This study provides a promising
answer to design the biomass conversion process to develop the
green chemistry.
are treated to be independent of the oligomer sizes and the
positions and are collectively denoted as knon‑at. Although the
sugar-chain flexibilities for different oligomer sizes could more
or less affect the cleavage rate of the internal glycosidic bond,
the magnitude of such effects should be at most 10−20% as in
the case of the effect of the ring strain of different sizes of
cyclodextrins, as described in section 3.2.1. On the basis of
these assumptions, eqs 4−9 are rewritten as follows:
APPENDIX
■
Determination of Cleavage Rate Constants for
Non-Anomeric Glycosidic Bonds in Linear
Malto-Oligosaccharides
d[G7]
dt
The first-order rate equations for the decomposition of the
malto-oligosaccharides of the dimer D-maltose (G2) to the
heptamer D-maltoheptaose (G7) are expressed as
= 7kβ‐CD[β‐CD] − (5knon‐at + kat)[G7]
(10)
d[G6]
dt
d[β‐CD]
= (knon‐at + kat)[G7] − (4knon‐at + kat)[G6]
= −7kβ‐CD[β‐CD]
(11)
(3)
dt
d[G7]
dt
d[G5]
dt
= 7kβ‐CD[β‐CD] − (k6/7 + k5/7 + k4/7 + k3/7
= 2knon‐at[G7] + (knon‐at + kat)[G6]
+ k2/7 + k1/7)[G7]
− (3knon‐at + kat)[G5]
(4)
(12)
d[G6]
dt
d[G4]
dt
= (k6/7 + k1/7)[G7] − (k5/6 + k4/6 + k3/6 + k2/6
= 2knon‐at[G7] + 2knon‐at[G6] + (knon‐at + kat)[G5]
+ k1/6)[G6]
(5)
− (2knon‐at + kat)[G4]
(13)
d[G5]
dt
= (k5/7 + k2/7)[G7] + (k5/6 + k1/6)[G6]
d[G3]
dt
= 2knon‐at[G7] + 2knon‐at[G6] + 2knon‐at[G5]
− (k4/5 + k3/5 + k2/5 + k1/5)[G5]
(6)
+ (knon‐at + kat)[G4] − (knon‐at + kat)[G3]
d[G4]
dt
(14)
= (k4/7 + k3/7)[G7] + (k4/6 + k2/6)[G6]
d[G2]
dt
+ (k4/5 + k1/5)[G5] − (k3/4 + k2/4 + k1/4
)
= 2knon‐at[G7] + 2knon‐at[G6] + 2knon‐at[G5]
[G4]
(7)
+ 2knon‐at[G4] + (knon‐at + kat)[G3] − kat[G2]
(15)
d[G3]
dt
= (k4/7 + k3/7)[G7] + 2k3/6[G6] + (k3/5 + k2/5
)
Here the cleavage rate constant for D-maltose, kmal, is
substituted into kat. The rate constants, kmal and kβ‑CD, are
determined independently in the reactions of β-cyclodextrin
and D-maltose, respectively, as the starting material as shown in
Figure 5b,c. When the integration of these differential equations
is carried out, kmal and kβ‑CD are treated as the fixed parameter
with only a single variable parameter of the cleavage rate
constant, knon‑at, for the non-anomeric glycosidic bond. The
knon‑at value is thus determined by numerically integrating the
simultaneous rate equations to fit the time evolution of the
fraction θ in Figure 7.
[G5] + (k3/4 + k1/4)[G4] − (k2/3 + k1/3)[G3]
(8)
d[G2]
dt
= (k5/7 + k2/7)[G7] + (k4/6 + k2/6)[G6]
+ (k3/5 + k2/5)[G5] + 2k2/4[G4]
+ (k2/3 + k1/3)[G3] − k1/2[G2]
(9)
where ki/n is the rate constant for the cleavage of the glycosidic-
bond at the ith unit from the anomeric-terminal one (i.e., the
bond between (i+1)/nG and i/nG) in the linear oligomers
composed of n units. For the reliable curve fitting for the time
evolution of the fraction θ in Figure 7 to obtain the physically
reasonable rate constants, in our approach we minimized the
number of variable parameters as below. First, the cleavage rate
constants k1/n (n = 2−7) for the anomeric terminal, that is, the
glycosidic bond between the anomeric-terminal unit and the
next one (bond between 1/nG and 2/nG), are treated to be
independent of the oligomer size and are denoted collectively
as kat. The effect of the oligomer size on the cleavage rate of the
anomeric-terminal unit has been examined for D-maltose, D-
maltotriose, and D-maltotetraose, and the differences among
these oligomers are only 5−10%.2 Then, the cleavage rates for
all the other internal, non-anomeric terminal glycosidic bonds
Once the concentrations of the dimer to the heptamer are
obtained as a function of the reaction time, the fraction θ of the
non-anomeric terminal units among the non-anomeric ones is
obtained as
7
∑
n=2 [Gn]
θ =
7
∑
n=2 (n − 1)[Gn]
[G7] + [G6] + [G5] + [G4] + [G3] + [G2]
6[G7] + 5[G6] + 4[G5] + 3[G4] + 2[G3] + [G2]
=
(16)
and this expression is used for the optimization of the variable
parameter knon‑at in the curve fitting for θ discussed in section
3.2.1.
1318
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319

The Journal of Physical Chemistry A
Article
(12) Lin, Y.-C.; Cho, J.; Tompsett, G. A.; Westmoreland, P. R.;
Huber, G. W. Kinetics and Mechanism of Cellulose Pyrolysis. J. Phys.
Chem. C 2009, 113, 20097−20107.
ASSOCIATED CONTENT
■
S
* Supporting Information
The in situ 13C spectrum for β-cyclodextrin at 120 °C with xw =
0.30 at 5 h, the time evolutions of the concentrations of the
reactant and products for the reactions of α- and γ-
cyclodextrins at 180 °C with xw = 0.30, and the comparison
of the D-fructose/D-glucose ratio between the D-maltose and the
D-cellobiose reactions. This material is available free of charge
via the Internet at http://pubs.acs.org.
(13) Liu, X.-X.; Yu, L.; Xie, F.-W.; Li, M.; Chen, L.; Li, X.-X. Kinetics
and Mechanism of Thermal Decomposition of Cornstarches with
Different Amylose/Amylopectin Ratios. Starch/Starke 2010, 62, 139−
̈
146.
(14) Zheng, M.-Y.; Wang, A.-Q.; Ji, N.; Pang, J.-F.; Wang, X.-D.;
Zhang, T. Transition Metal-Tungsten Bimetallic Catalysts for the
Conversion of Cellulose into Ethylene Glycol. ChemSusChem. 2010, 3,
63−66.
(15) Zhao, S.; Cheng, M.-X.; Li, J.-Z.; Tian, J.-A.; Wang, X.-H. One
Pot Production of 5-Hydroxymethylfurfural with High Yield from
Cellulose by a Brønsted-Lewis-Surfactant-Combined Heteropolyacid
Catalyst. Chem. Commun. 2011, 47, 2176−2178.
(16) Mayes, H. B.; Broadbelt, L. J. Unraveling the Reactions that
Unravel Cellulose. J. Phys. Chem. A 2012, 116, 7098−7106.
(17) Abdullah, R.; Ueda, K.; Saka, S. Decomposition Behaviors of
Various Crystalline Celluloses as Treated by Semi-Flow Hot-
Compressed Water. Cellulose 2013, 20, 2321−2333.
AUTHOR INFORMATION
■
Corresponding Author
*K. Yoshida: phone, +81-88-656-7669; fax, +81-88-655-7025;
e-mail, yoshida@chem.tokushima-u.ac.jp.
Notes
The authors declare no competing financial interest.
(18) Chheda, J. N.; Huber, G. W.; Dumesic, J. A. Liquid-Phase
Catalytic Processing of Biomass-Derived Oxygenated Hydrocarbons to
Fuels and Chemicals. Angew. Chem,. Int. Ed. 2007, 46, 7164−7183.
(19) Rosatella, A. A.; Simeonov, S. P.; Frade, R. F. M.; Afonso, C. A.
M. 5-Hydroxymethylfurfural (HMF) as a Building Block Platform:
Biological Properties, Synthesis and Synthetic Applications. Green
Chem. 2011, 13, 754−793.
(20) van Putten, R.-J.; van der Waal, J. C.; de Jong, E.; Rasrendra, C.
B.; Heeres, H. J.; de Vries, J. G. Hydroxymethylfurfural, A Versatile
Platform Chemical Made From Renewable Resources. Chem. Rev.
2013, 113, 1499−1597.
ACKNOWLEDGMENTS
■
This work is supported by the Grants-in-Aid for Scientific
Research (No. 25410019) from the Japan Society for the
Promotion of Science. K.Y. is grateful for the donations from
the Takahashi Industrial and Economic Research Foundation,
the Suzuki Foundation, the Salt Science Research Foundation,
No. 1114, JGC-S Scholarship Foundation, and SEI group CSR
foundation. M.N. is grateful to AGC, Limited for the financial
support.
(21) van Putten, R.-J.; Soetedjo, J. N. M.; Pidko, E. A.; van der Waal,
J. C.; Hensen, E. J. M.; de Jong, E.; Heeres, H. J. Dehydration of
Different Ketoses and Aldoses to 5-Hydroxymethylfurfural. Chem-
SusChem. 2013, 6, 1681−1687.
(22) Choudhary, V.; Mushrif, S. H.; Ho, C.; Anderko, A.; Nikolakis,
V.; Marinkovic, N. S.; Frenkel, A. I.; Sandler, S. I.; Vlachos, D. G.
Insights into the Interplay of Lewis and Brønsted Acid Catalysts in
Glucose and Fructose Conversion to 5-(Hydroxymethyl)furfural and
Levulinic Acid in Aqueous Media. J. Am. Chem. Soc. 2013, 135, 3997−
4006.
(23) Shi, N.; Liu, Q.-Y.; Zhang, Q.; Wang, T.-J.; Ma, L.-L. High Yield
Production of 5-Hydroxymethylfurfural from Cellulose by High
Concentration of Sulfates in Biphasic System. Green Chem. 2013, 15,
1967−1974.
(24) Dutta, S.; De, S.; Saha, B. Advances in Biomass Transformation
to 5-Hydroxymethylfurfural and Mechanistic Aspects. Biomass &
Bioenergy 2013, 55, 355−369.
(25) Gallo, J. M. R.; Alonso, D. M.; Mellmer, M. A.; Dumesic, J. A.
Production and Upgrading of 5-Hydroxymethylfurfural Using
Heterogeneous Catalysts and Biomass-Derived Solvents. Green
Chem. 2013, 15, 85−90.
(26) Caes, B. R.; Palte, M. J.; Raines, R. T. Organocatalytic
Conversion of Cellulose into a Platform. Chem. Sci. 2013, 4, 196−199.
(27) The standard specific combustion energy for D-glucose is 15.6 kJ
g−1 (ref 28), whereas that for 5-HMF is 22.1 kJ g−1 (ref 29). The
specific combustion energy density is thus enhanced by the formation
of the unsaturated bonds due to the dehydration.
(28) Kabo, G. J.; Voitkevich, O. V.; Blokhin, A. V.; Kohut, S. V.;
Stepurko, E. N.; Paulechka, Y. U. Thermodynamic Properties of Starch
and Glucose. J. Chem. Thermodynamics 2013, 59, 87−93.
(29) Verevkin, S. P.; Emel’yanenko, V. N.; Stepurko, E. N.; Ralys, R.
V.; Zaitsau, D. H.; Stark, A. Biomass-derived Platform Chemicals:
Thermodynamic Studies on the Conversion of 5-Hydroxymethylfur-
fural into Bulk Intermediates. Ind. Eng. Chem. Res. 2009, 48, 10087−
10093.
REFERENCES
■
(1) Kimura, H.; Nakahara, M.; Matubayasi, N. In Situ Kinetic Study
on Hydrothermal Transformation of D-Glucose into 5-Hydroxyme-
thylfurfural through D-Fructose with 13C NMR. J. Phys. Chem. A 2011,
115, 14013−14021.
(2) Kimura, H.; Nakahara, M.; Matubayasi, N. Noncatalytic
Hydrothermal Elimination of the Terminal D-Glucose Unit from
Malto- and Cello-Oligosaccharides through Transformation to D-
Fructose. J. Phys. Chem. A 2012, 116, 10039−10049.
(3) Kimura, H.; Nakahara, M.; Matubayasi, N. Solvent Effect on
Pathways and Mechanisms for D-Fructose Conversion to 5-
Hydroxymethyl-2-furaldehyde: In Situ 13C NMR Study. J. Phys.
Chem. A 2013, 117, 2102−2113.
(4) Kimura, H.; Yoshida, K.; Uosaki, Y.; Nakahara, M. Effect of Water
Content on Conversion of D-Cellobiose into 5-Hydroxymethyl-2-
furaldehyde in a Dimethylsulfoxide−Water Mixture. J. Phys. Chem. A
2013, 117, 10987−10996.
(5) Nakahara, M.; Wakai, C. Monomeric and Cluster States of Water
Molecules in Organic Solvent. Chem. Lett. 1992, 21, 809−812.
(6) Nakahara, M.; Wakai, C. Effect of Solvent, Temperature, and
Pressure on Hydrogen Bonding and Reorientation of Water
Molecules. J. Mol. Liq. 1995, 65/66, 149−155.
(7) Wakai, C.; Nakahara, M. Attractive Potential Effect on the Self-
Diffusion Coefficients of A Solitary Water Molecule in Organic
Solvents. J. Chem. Phys. 1997, 106, 7512−7518.
(8) Sasaki, M.; Kabyemela, B.; Malaluan, R.; Hirose, S.; Takeda, N.;
Adschiri, T.; Arai, K. Cellulose Hydrolysis in Subcritical and
Supercritical Water. J. Supercrit. Fluids 1998, 13, 261−268.
(9) Sasaki, M.; Fang, Z.; Fukushima, Y.; Adschiri, T.; Arai, K.
Dissolution and Hydrolysis of Cellulose in Subcritical and Supercritical
Water. Ind. Eng. Chem. Res. 2000, 39, 2883−2890.
(10) Nagamori, M.; Funazukuri, T. Glucose Production by
Hydrolysis of Starch under Hydrothermal Conditions. J. Chem.
Technol. Biotechnol. 2004, 79, 229−233.
́
(11) Chheda, J. N.; Roman-Leshkov, Y.; Dumesic, J. A. Production of
5-Hydroxymethylfurfural and Furfural by Dehydration of Biomass-
Derived Mono- and Poly-Saccharides. Green Chem. 2007, 9, 342−350.
1319
dx.doi.org/10.1021/jp412628y | J. Phys. Chem. A 2014, 118, 1309−1319