Free energy landscape for glucose condensation and dehydration reactions in dimethyl sulfoxide and the effects of solvent

Article abstract of DOI:10.1016/j.carres.2014.02.010

The mechanisms and free energy surfaces (FES) for the initial critical steps during proton-catalyzed glucose condensation and dehydration reactions were elucidated in dimethyl sulfoxide (DMSO) using Car-Parrinello molecular dynamics (CPMD) coupled with me

Full text of DOI:10.1016/j.carres.2014.02.010

Carbohydrate Research 388 (2014) 50–60
Contents lists available at ScienceDirect
Carbohydrate Research
journal homepage: www.elsevier.com/locate/carres
Free energy landscape for glucose condensation and dehydration
reactions in dimethyl sulfoxide and the effects of solvent
Xianghong Qian , Dajiang Liu b
a,
⇑
a
Department of Chemical Engineering, University of Arkansas, Fayetteville, AR 72701, United States
Department of Mechanical Engineering, Colorado State University, Ft Collins, CO 80521, United States
b
a r t i c l e i n f o
a b s t r a c t
Article history:
The mechanisms and free energy surfaces (FES) for the initial critical steps during proton-catalyzed
glucose condensation and dehydration reactions were elucidated in dimethyl sulfoxide (DMSO) using
Car–Parrinello molecular dynamics (CPMD) coupled with metadynamics (MTD) simulations. Glucose
condensation reaction is initiated by protonation of C1AOH whereas dehydration reaction is initiated
by protonation of C2AOH. The mechanisms in DMSO are similar to those in aqueous solution. The DMSO
molecules closest to the C1AOH or C2AOH on glucose are directly involved in the reactions and act as
proton acceptors during the process. However, the energy barriers are strongly solvent dependent. More-
over, polarization from the long-range electrostatic interaction affects the mechanisms and energetics of
glucose reactions. Experimental measurements conducted in various DMSO/Water mixtures also show
that energy barriers are solvent dependent in agreement with our theoretical results.
Received 15 November 2013
Received in revised form 30 January 2014
Accepted 7 February 2014
Available online 18 February 2014
Keywords:
Glucose
DMSO
Free energy surface
Condensation
Dehydration
Ó 2014 Elsevier Ltd. All rights reserved.
1
. Introduction
Glucose is the most important and abundant monomeric sugar
in aqueous solutions. Here the critical initial steps during glucose
condensation and dehydration reactions in dimethyl sulfoxide
(DMSO) are investigated.
1
5
16
on earth. Elucidating glucose transformations and reactions is crit-
ical to our understanding and manipulation of this vital molecule.
Our earlier theoretical studies1–9 and experimental results1,10–13
show that proton catalyzed glucose reactions are generally not
selective due to multiple protonation sites on the glucose molecule
leading to multiple reaction pathways. Protonation of the O5 on
Car-Parrinello based ab initio molecular dynamics (CPMD)
1
7
coupled with metadynamics (MTD) simulations have been suc-
cessful in elucidating the mechanisms, the rate-limiting steps,
and associated barriers as well as free energy surfaces (FES) for glu-
6
,7,9,14
cose reactions in aqueous solutions.
For example, excellent
agreement was obtained between the calculated and experimental
1
4
5,6
the glucose ring leads to the mutarotation between
a- and b-glu-
barriers for glucose mutarotation, condensation,
isomeriza-
9
7
cose. Protonation of the C1AOH on glucose leads to the formation
of an oxocarbenium carbocation and the eventual 1, x (x = 2, 3, 4, 6)
linked oligosaccharides from condensation reactions. Protonation
of C2AOH on glucose leads to the formation of 5-hydroxylmethyl-
furfural (HMF) from dehydration reaction as well as isomerization
reaction to fructose. In addition, our earlier results1 indicate that
protonation of the ring O or the hydroxyl groups on the glucose
molecule and the subsequent breakage of the CAO bond is the
rate-limiting step. Moreover, it was found that glucose reactions
are strongly solvent dependent due to the competition for proton
from the solvent molecules. The reaction barriers are largely sol-
vent induced. Our earlier studies focused on the glucose mutarota-
tion , and dehydration reactions. Here CPMD–MTD simulations
for glucose condensation and dehydration reactions in DMSO sol-
vent medium are conducted to gain insights into the effects of sol-
vent on the mechanisms and barriers for glucose reactions.
Moreover, atomic charges of the glucose molecule in the gas phase
–9
2
as well as in H O and DMSO solvents were calculated using both
1
8
Gaussian09 and CPMD in order to gain deep insight into the sol-
vent effects on glucose reactivity.
In order to validate our theoretical results, experiments were
also conducted for glucose reactions in pure DMSO and several
DMSO/Water mixtures at temperatures ranging from 120 °C
(393 K) to 140 °C (413 K). The concentrations of glucose, major
reaction products, and their time dependence were determined
using Nuclear Magnetic Resonance (NMR). Moreover, activation
energy barriers in the solvent mixtures for glucose degradation,
tion, condensation, isomerization9 and dehydration7 reactions
1
4
6
⇑
Corresponding author. Tel.: +1 479 575 8401.
E-mail address: xqian@uark.edu (X. Qian).
its dehydration to 1,6-anhydro-b-
and HMF were determined and compared with theoretical results.
D-glucopyranose (levoglucosan)
http://dx.doi.org/10.1016/j.carres.2014.02.010
008-6215/Ó 2014 Elsevier Ltd. All rights reserved.
0

X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
51
2
. Computational details
investigated. However, it does not exclude the possible existence
of an alternative reaction pathway in DMSO. Similar to aqueous
solution, three CVs are adopted for proton-catalyzed glucose dehy-
dration to HMF in DMSO. These three CVs comprise protonation of
C2AOH (CV3), the breakage of the C2AO2 bond (CV1), and the for-
mation of C2AO5 bond (CV2).
CPMD–MTD allows for efficient and accelerated sampling of
chemical and biological processes for free energy calculations, par-
ticularly for chemical reactions involving the bond-breaking and
bond-forming processes in the time scale not accessible by the
conventional methods. The acceleration of the sampling process
is achieved by filling the reactant and product wells with repulsive
5
–9,20
Our earlier studies
showed that CVs using coordination
numbers (CN) are effective for exploring sugar reaction processes.
1
7,19
19
bias potentials
to facilitate barrier crossing. Once the reactant
The equation of CN is given by
well is filled with potentials close enough to the reaction barrier,
the system overcomes the barrier and moves to the product well.
When the product well is also filled and the FES becomes flat,
the system is able to sample the reactant and product states ran-
domly without any barrier. The original FES of the system is subse-
quently reconstructed based on the amount of bias potentials
added to reach the flat FES state. This method assumes that several
collective variables (CV), which distinguish the initial state from
the final state, are able to characterize the slow, rate-limiting steps.
The left panel in Figure 1 shows the CVs for the critical steps
during glucose condensation (left panel) and dehydration (right
ꢀ
ꢁ
p
d
ij
d
0
d
ij
d
0
1 ꢀ
1 ꢀ
CNði; jÞ ¼
;
ꢀ
ꢁ
q
where dij is the distance between atoms i and j, d0 is the cutoff
distance, and p and q are high-power integers used to distinguish
between the coordinated and non-coordinated states. The values
p = 6 and q = 12 are typically chosen for calculating the CNs.
The choice for the cutoff distance d0 depends on the specific
bond. For CꢀO and OꢀH bonds, values of 2.0 and 1.5 Å are usually
6
panel) reactions. Our previous results for glucose condensation
chosen for d respectively as is done here and previously in our
0
5
–7,9
reaction in aqueous solution show that the rate-limiting step in-
volves the protonation of C1AOH and the breakage of the C1AO1
bond. The C1 carbocation formed and subsequently the more sta-
ble oxocarbenium ion are the critical intermediates for various 1,
x (x = 2,3,4,6) linked disaccharides. For glucose condensation reac-
tion in DMSO, a similar mechanism is assumed. As a result, CVs for
the first steps in glucose condensation reaction comprise proton-
ation of C1AOH (CV2) and breakage of C1AO1 bond (CV1). In addi-
tion, our earlier studies5 in aqueous solution also show that
partial dehydration due to the migration of the hydronium ion to
the neighborhood of the sugar molecule also contribute substan-
tially to the barrier for condensation reaction.
Three CVs are needed for the critical steps during glucose dehy-
dration to form a cyclic HMF intermediate in aqueous solution as
shown in the right panel of Figure 1. Glucose dehydration reaction
is initiated by the protonation of C2AOH, the breakage of the
C2AO2 bond, and the formation of C2AO5 bond leading to the for-
work.
The dynamics of the CVs are controlled by the force constant k
and fictitious mass m. The values of k = 2.0 au and m = 100 amu
were used for all the CVs here. The bias potential chosen is a com-
monly used Gaussian functional. The height and the width of the
Gaussian bias potential were chosen to be 0.001 and 0.100 au
respectively for all the simulations. The bias potentials were added
whenever the displacements in the CVs were larger than 1.5 times
the width, but no shorter than 100 MD steps. Studies have shown
that this choice of parameters is efficient with uncertainty in the
,6
2
1
range of 1–2 kcal/mol. More details on the method applied to su-
5
–9,20
gar reactions could be found in our earlier work.
The simulations were conducted using the Becke, Lee, Yang,
22
2
3
and Parr (BLYP)
functional for the valence and semi-core
2
4
electrons. Goedecker pseudopotential was used for the core elec-
trons. The energy cut-off of 80 Ry was used for the plane wave ba-
sis set. The combinations of these parameters have found to yield
7
mation of a five-member aldehyde ring intermediate. However,
our previous results show that proton partial dehydration does
excellent structural properties as well as energetics and reactivity
7
25,26
for sugar molecules.
The simulations were conducted under
not contribute to the barrier for glucose dehydration reaction in
aqueous solution since the barrier of 30–35 kcal/mol is dominated
by protonation followed by the breakage of the C2AO2 bond only.
Once a C2-carbocation is formed, the formation of C2AO5 bond ap-
pears to be spontaneous without any barrier. In this study, an anal-
ogous mechanism for glucose dehydration to HMF in DMSO was
constant volume and constant temperature (NVT) with a Nosé–
2
7,28
Hoover chain thermostat.
The temperature was kept at 27 °C
(300 K) for the condensation reaction initiated by the protonation
of C1AOH and at 227 °C (500 K)for the dehydration reaction initi-
ated by the protonation of C2AOH. To effectively separate the fast
motions of the electrons from the slow movement of the nuclei, a
Figure 1. The collective variables (CVs) for critical steps during glucose condensation reaction are shown in the left panel with CV1 representing C1AO1 bond and CV2
representing O1AH bond. CVs for critical steps during glucose dehydration reaction are shown in the right panel with CV1 representing C2AO2 bond, CV2 representing
C2AO5 bond and CV3 representing O2AH bond.

5
2
X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
fictitious mass of 800 amu and a time step of 0.125 femtosecond
4. Results and discussion
(
fs) were used. The system contains one glucose molecule, 40
ꢀ
DMSO molecules, one proton, and one Cl counter ion. The initial
unit cell containing only one glucose and 40 DMSO molecules
was equilibrated for over 10 picoseconds (ps) using CPMD. After
the initial equilibration, one H and one Cl ion were then inserted
in the system to mimic the acidic environment. The unit cell has a
4.1. The mechanism and free energy surface for glucose
condensation reaction in DMSO
+
ꢀ
Figure 2 shows the mechanism for the critical steps during glu-
cose condensation reaction involving protonation of C1AOH and
the formation of an oxocarbenium ion from the CPMD–MTD simu-
lations. Only two DMSO molecules involved directly in the reaction
are shown in Figure 2. A total of over 800 MTD simulation steps
were conducted for the reaction. The initial state (A) includes a
proton attached to the S@O group on the closest DMSO molecule.
During the subsequent simulations, this proton and the other pro-
ton initially bonded to O1 transfer back and forth between the S@O
groups on the two neighboring DMSO molecules and C1AOH on
the glucose ring as shown in Figure 2 (B and C). Since the two
protons are identical, both of the two closest DMSO molecules
are directly involved in the proton transfer process. Finally, the
3
3
dimension of 18.5 ꢁ 18.5 ꢁ 18.5 Å with a density of 0.88 g/cm .
Periodic boundary conditions (PBC) were applied. Ewald summa-
2
9
tion was used to integrate the long-range electrostatic interac-
tion energies.
The atomic charges derived from the electrostatic potentials
ESP) based on a method developed by Hirshfeld31 were deter-
3
0
(
mined using CPMD for the glucose molecule in the gas phase as
well as solvated by explicit DMSO or H O molecules. Atomic
charges were determined both with and without periodic bound-
ary conditions. plane-wave cut-off of 100 Ry with BLYP
2
A
functional was used. For comparison, ESP charges were also calcu-
lated using Gaussian09 for the glucose molecule in the gas phase as
protonated OH group (i.e. H O) departs from the glucose ring lead-
2
well as when solvated implicitly by DMSO or H
2
O. The hybrid
ing to the formation of C1-carbocation and the more stable
oxocarbenium ion (D).
⁄⁄
B3LYP potential with 6-311++G basis set coupled with implicit
3
2,33
CPCM solvation model
tions using Gaussian09.
were used for the ESP charge calcula-
Even though the mechanism for glucose condensation reaction
in DMSO is similar to the corresponding reaction in the aqueous
solution, the role of solvent appears to be rather different. In the
simulations conducted in aqueous solution, the water molecules
form extensive hydrogen bonding network. The excess proton
can transport rapidly within the water cluster without any barrier
via this extensive hydrogen-bonding network. The high mobility of
the proton in the water cluster indicates its stability in aqueous
solution due to the significant contribution from entropy. Further
the small positively charged hydronium ion has a very large hydra-
3
3
. Experimental procedures
.1. Materials and methods
NMR spectra were recorded either by Varian Inova 300 (FT
00 MHz) or Varian Inova 400 spectrometer. Chemical shifts for
H spectra are reported as parts per million (ppm) relative to tetra-
methylsilane (TMS). Concentrations were determined based on the
ratios of peak areas for the products to those of biphenyl (internal
3
1
3
4,35
tion free energy of about ꢀ264 kcal/mol
further suggesting the
stability of proton in water. As a result, a barrier of about 25 kcal
kcal/mol from both experiments and theoretical calculations
1
3
6,36
reference). Typical NMR spectra of glucose, 1,6-anhydro-
D
-b-gluco-
pyranose (levoglucosan, AHG), fructose, and HMF were listed in the
Supporting information (SI).
was estimated for glucose condensation reaction in aqueous solu-
tion. The contribution to this barrier comes from partial proton
dehydration due to its migration from the bulk solvent to the
neighborhood of the glucose molecule as well as the protonation
of the C1AOH and the formation of the stable intermediate
oxocarbenium ion. For glucose condensation reaction in DMSO, a
proton can only attach to its closest DMSO molecule thus only
two nearest neighboring DMSO molecules are involved directly
in the reaction process. Proton solvation free energy in DMSO is
estimated to be ꢀ273.3 kcal/mol, slightly larger than the corre-
sponding value in water. Since the proton affinity and proton sol-
vation free energy in DMSO are different from those in water, the
reaction barrier for protonation of C1AOH in DMSO is expected
to be different as is confirmed from our CPMD–MTD results dis-
cussed in more detail below.
D
-Glucose (Fisher Chemical),
nyl (Aldrich), 5-hydromethylfurfural (HMF) (Aldrich) were pur-
2
chased and used without further purification. DMSO-d and D O
D-fructose (Mallinckrodt), biphe-
6
were used for all the NMR experiments. Reactions were con-
ducted in the oil-bathed NMR tubes. Reaction solutions were pre-
pared beforehand to ensure their homogeneity. Different ratios of
2
DMSO/D O solvent mixtures with pure DMSO, DMSO/D
2
O = 95/5
3
6
(
v/v), DMSO/D
2
O = 90/10 (v/v), DMSO/D O = 80/20 (v/v) were
2
used for the reactions. Small amount of hydrochloric acid (HCl)
was added as a catalyst. Biphenyl was used as an internal
reference.
3
.2. Representative procedure for glucose conversion in DMSO/
Water
Figure 3 shows the fluctuations of the two CVs during the
course of almost 900 CPMD–MTD simulation steps. The blue line
representing CV1 (C1AO1) describes the bond breakage and for-
mation between C1 and O1 whereas the red line representing
CV2 (O1AH) describes the protonation of C1AOH process. It can
be seen that proton initially bonded to the O on DMSO was rapidly
transferred to the C1AOH in less than 20 MTD steps. However, the
C1AO1 bond did not break to form a C1-carbocation until about
about 50 MTD steps. The sampling of the C1-carbocation is rather
brief as the proton was seen to transfer back to the neighboring
DMSO molecules and the neutral glucose molecule reforms. The
C1AO1 bond is again broken at around 300 MTD steps after the
proton transfers again to the C1AOH. The protonated hydroxyl
Glucose reactions and subsequent NMR experiments were con-
ducted in a closed lid NMR tube due to the volatility of HCl acid
catalyst. Solutions of glucose (56 mM) in various ratios of DMSO/
Water mixtures with HCl (5.6 mM) and biphenyl (2 mM) were
prepared in advance. Each NMR tube evenly charged with 0.6 mL
solution was then placed into the pre-equilibrated oil-bath at
pre-determined temperatures. Reactions were quenched by rap-
idly inserting the tubes into ice water. Subsequent NMR H exper-
iments were performed. The H NMR peaks used for quantification:
biphenyl d 7.66 (m, 1H, H3), 7.46 (m, 1H, H2), 7.36 (m, 1H, H1);
glucose d 4.94 (d, 1H, J = 3.6 Hz, H1-a), 4.20 (d, 1H, J = 7.5 Hz,
1
1
H1-b); for levoglucosan d 5.15 (s, 2H, H1-b), 4.76 (dd, 1H, H3);
2
group (i.e. H O) departs from the glucose ring to form an oxocarbe-
for HMF, d 9.54 (s, 1H, H1), 7.49 (d, 1H, J = 3.6 Hz, H2), 6.60 (d,
nium ion.The oxocarbenium cation was sampled during the subse-
quent 225 MTD steps. At around 490 MTD steps, the proton again
1
H, J = 3.6 Hz, H3), 4.50 (s, 2H, H6).

X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
53
Figure 2. The mechanism for protonation of C1AOH on glucose and the subsequent formation of oxocarbenium ion during glucose condensation reaction in DMSO (O (red), C
cyan), S (yellow), and H (white)). Only two participating DMSO molecules are shown. The reactive sites are circled. (For interpretation of the references to color in this figure
(
legend, the reader is referred to the web version of this article.)
Based on the amount of Gaussians potentials filled, the FES was
reconstructed. Figure 4 shows the projected two dimensional (2D)
free energy contour plot between CV1 (C1AO1) and CV2 (O1AH).
The general feature of FES for protonation of C1AOH on glucose
and the breakage of C1AO1 bond in DMSO is similar to the reaction
carried out in water. The stable or metastable structures during the
reaction process include the initial neutral glucose molecule when
Figure 3. The variation of the CV1 (C1AO1, blue) and CV2 (O1AH, red) during the
CPMD–MTD simulations for glucose condensation reaction in DMSO initiated by
protonation of C1AOH. (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of this article.)
transfers back to the neighboring DMSO molecule and the C1AO1
bond reforms. Subsequently the proton was seen transferring back
and forth between the C1AOH and the S@O on the neighboring
DMSO molecules. At 825 MTD steps, both protons transfer to the
neighboring DMSO molecules forming a different chemical species.
This completes the sampling process for the protonation of C1AOH
and the formation of the oxocarbenium ion. The FES was recon-
stucted based on the 825 MTD simulations.
Figure 4. The projected free energy contour plot between CV1 (C1AO1) and CV2
(
O1AH) for protonation of C1AOH and breakage of C1AO1 bond during glucose
condensation reaction in DMSO.

5
4
X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
Figure 5. The mechanism for glucose dehydration reaction to form a HMF intermediate via a direct cyclic pathway initiated by protonation of C2AOH, breakage of the C2AO2
bond, and the formation of the C2AO5 bond. Only the DMSO molecules directly participating in the reaction are shown. The reactive sites are shown in circles.
the proton is attached to the neighboring DMSO molecule, the final
oxocarbenium ion after the breakage of the C1AO1 bond. and the
intermediate when the proton is bridged between the C1AOH
and S@O groups. The distance between the two O atoms in the
bridged state is only around 2.56 Å. These are so-called low-barrier
(500 K). The simulations were successful only after increasing the
temperature to 227 °C. This indicates that this dehydration mecha-
nism in DMSO has either a high barrier or is energetically not a
favorable process. The schematics of the mechanism for glucose
dehydration reaction to form a HMF intermediate via a direct cyclic
pathway from CPMD–MTD simulations at 227 °C (500 K) are shown
in Figure 5. The reaction is initiated by protonation of the C2AOH on
the glucose molecule (E). Since the two protons bonded to the O2
atom are indistinguishable, both protons are observed during the
sampling process transferring back and forth between the O2 atom
3
7
hydrogen bonds (LBHB). They are much stronger than the ordin-
ary hydrogen bond, in particular the bridged proton is able to move
between the two O atoms freely without any barrier. The
formation of LBHB has also been observed during xylose and glu-
cose condensation reactions earlier.5 Based on the reconstructed
FES, the barrier associated with protonation of C1AOH is only
about 16 kcal/mol and the breakage of C1AO1 bond is about
,6
6
kcal/mol. The overall barrier for protonation of C1AOH and
breakage of C1AO1 bond is about 20 kcal/mol. The reaction barrier
of 20 kcal/mol is smaller than the corresponding reaction in water,
which has a barrier of about 25 kcal/mol. However, the oxocarbe-
nium ion is seen less stable than the initial glucose molecule
contrary to the case in aqueous solution indicating proton
catalyzed glucose condendation reaction in DMSO is not an
energetically favorable process. Our results indicate that solvent
could play a critical role in proton catalyzed glucose or other gen-
eral organic reactions in solution since proton solvation free energy
and solvent molecule’s proton affinity are typically different in
different solvent.
4
.2. The mechanism and free energy surface for glucose
dehydration reaction in DMSO
The CPMD–MTD simulations for glucose dehydration to the cyc-
lic HMF intermediate were not successful when conducted at the
room temperature, or for several other temperatures below 227 °C
Figure 6. The variation of the CV1 (C2AO2, blue), CV2 (C2AO5, red) and CV3
(
O2AH, black) during the CPMD–MTD simulations for glucose dehydration reaction
to form HMF in DMSO initiated by protonation of C2AOH.

X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
55
on glucose and the O atoms of their nearest DMSO molecules. The
proton can either be completely transferred or partially transferred
via the formation of LBHB to the S@O on a DMSO molecule (F). After
sufficient sampling, the C2AO2 bond starts to break (G). With addi-
tional sampling, the C2AO5 bond forms (H) and the C1AO5 bond
energy contour plots based on the amount of repulsive potentials
added during the 1134 MTD sampling steps. The left panel on
Figure 8 shows the 2D free energy contour plot between CV1 and
CV3, whereas the right panel on Figure 8 shows the contour plot
and FES between CV2 and CV3. It appears that there is only about
5 kcal/mol of barrier associated with the protonation of the
C2AOH. The breakage of C2AO2 bond entails another 15 kcal/
mol. The overall barrier for protonation of C2AOH and breakage
of C2AO2 bond is only about 20 kcal/mol. This barrier is signifi-
cantly smaller than the 35 kcal/mol barrier observed for proton-
ation of C2AOH and breakage of C2AO2 bond in aqueous
breaks leading to the formation of a five-member ring intermediate
+
(
I). The subsequently formed C1 AOH carbocation is located outside
the five-member ring structure. Finally, the O atom on the neighbor-
ing DMSO molecule takes away the proton from the O1 leading to
the formation of an aldehyde C1H@O intermediate (J). The closest
two DMSO molecules to the glucose C2AOH act as proton acceptors
during the reaction process.
7
solution as shown in our earlier study. This reduction in barrier
Figure 6 shows the variation of three CVs during the course of
over 1100 MTD simulation steps. The proton is initially attached
to the C2AOH, but rapidly transfers to the neighboring S@O on
the DMSO molecules. During the subsequent sampling process,
the proton was seen transferring back and forth between C2AOH
and the DMSO molecules. The C2AO2 bond breaks at around 250
MTD steps, but rapidly reforms. The C2AO2 bond was again seen
to break at around 450 MTD steps and forms again quickly. The
C2AO2 bond is finally broken at 580 MTD steps and remains bro-
ken till the end of simulations. On the other hand, the C2AO5 bond
only starts to form at around 525 MTD steps and becomes com-
pletely bonded at 580 MTD steps. This appears to be different from
our CPMD-MTD simulations for glucose dehydration reaction in
aqueous solution. The breakage of the C2AO2 bond and the forma-
tion of the C2AO5 bond almost occur simultaneously in water. This
indicates that the breakage of the C2AO5 bond is not the rate-lim-
iting step during glucose dehydration reaction in aqueous solution.
However, in DMSO solvent, breakage of the C2AO5 bond appears
to be more difficult and becomes likely the rate-limiting step.
Our reconstructed FES confirms that the barrier for C2AO5 break-
age is higher than that for C2AO2 breakage, which will be dis-
cussed in more detail later. Our continued sampling to about
height is consistent with our understanding of the effects of sol-
vent on sugar reactions. The higher the solvent’s affinity for the
proton , the more difficult it is for the hydroxyl groups on glucose
to compete for the proton and therefore a higher barrier is ex-
pected. Since the proton is highly stabilized by the water cluster
due to the extensive hydrogen bonding network formed in aqueous
solution, the barrier for glucose dehydration reaction is high and
largely solvent induced. The glucose dehydration reaction in DMSO
is expected to have a different barrier as DMSO’s affinity for the
proton is different from that of water. Since DMSO cannot form
hydrogen bonds with each other, proton transport in DMSO is
more localized and likely diffusion limited. As a result, it will be
easier for the localized proton to transfer to the neighboring glu-
cose molecule. Thus a lower barrier for protonation of C2AOH in
DMSO than in water is expected.
On the hand, the cyclic mechanism for glucose transformation
to HMF intermediate involves not only protonation of C2AOH
and the breakage of C2AO2 bond, but also the formation of the
C2AO5 bond and the breakage of the C1AO5 bond. In our previous
study for glucose dehydration to HMF in aqueous solution,7 the
barrier for the formation of C2AO5 bond is only about 20 kcal/
mol indicating that protonation of C2AOH and breakage of
C2AO2 bond is the rate-limiting step. However, in DMSO solvent,
the role appears to be reversed. The formation of C2AO5 bond in
DMSO becomes energetically more difficult as shown from the
right panel in Figure 8. The barrier for the formation of C2AO5
bond is about 55 kcal/mol in DMSO. This is significantly higher
than the reaction in aqueous solution. This is perhaps the reason
why our CPMD-MTD simulations for glucose dehydration reaction
are successful only when conducted at elevated temperature of
500 K. At a lower temperature, different reaction products were
observed.
1
100 MTD steps did not lead to the reformation of the glucose
reactant as is necessary for obtaining accurate FES due probably
to the relative stability of the five member ring intermediate. How-
ever, the barrier for the reaction process is generally accurate and
reliable as was discussed in our earlier studies.7
,9
Figure 7 shows the sampling trajectories between CV1 (C2AO2)
and CV3 (O2AH) (left panel), and between CV2 (O2AO5) and CV3
(
O2AH) (right panel). It can be seen that all three reaction coordi-
nates have extensive sampling indicating the quality of the free en-
ergy calculations. Figure 8 exhibits the reconstructed 2D free
Figure 7. The sampling trajectories between CV1 (C2AO2) and CV3 (O2AH) (left panel), between CV2 (C2AO5) and CV3 (O2AH) (right panel) during CPMD–MTD sampling
for glucose dehydration reaction to HMF in DMSO initiated by protonation of C2AOH.

5
6
X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
Figure 8. The 2D free energy contour plots between CV1 (C2AO2) and CV3 (O2AH) (left panel), between CV2 (C2AO5) and CV3 (O2AH) (right panel) during the CPMD–MTD
sampling for glucose dehydration reaction to HMF in DMSO initiated by protonation of C2AOH.
Figure 9. The 1H NMR spectra for glucose conversion in mixed DMSO/Water = 95/5 (v/v) solvent at various times at 120 °C. The peaks for glucose, levoglucosan, and HMF are
marked.
4
.3. Experimental results on glucose conversion in DMSO and
were conducted at temperatures ranging from 110 °C to 140 °C. Be-
DMSO/Water mixtures
sides HMF, a glucose dehydration product by losing three water
molecules, significant amount of 1,6-anhydro-D-b-glucopyranose
NMR Experiments were conducted to investigate the effects of
solvent on glucose dehydration to HMF in pure DMSO and DMSO/
Water mixed solvents. Investigated DMSO/Water ratios for the sol-
vent mixtures comprise 100/0, 95/5, 90/10, and 80/20 (v/v). Sol-
vents with higher water contents were not investigated due to
the evaporation of water during the experiments. The reactions
(levoglucosan), a glucose dehydration product by losing one water
molecule was also detected in agreement with earlier studies.38
Figure 9 shows the NMR spectra for glucose, levoglucosan, and
HMF and their respective characteristic peaks for quantitative mea-
surements for glucose in DMSO/Water = 95/5 (v/v) mixed solvent at
the beginning and after 0.5, 1, and 4 h of reaction time. It can be

X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
57
Glucose
Levoglucosan
HMF
A
Glucose
Levoglucosan
HMF
12
B
3.5
3.0
2
5
7
0
6
5
4
3
2
1
0
0
0
0
0
0
0
3
2
.0
.5
1
10°C
1
10°C
20°C
130°C
140°C
150°C
1
0
8
6
4
2
0
1
60
50
130°C 20
140°C
50°C
2
2
1
.5
.0
.5
1
2.0
1.5
1.0
0.5
0.0
1
1
5
0
5
0
4
3
2
0
0
0
1.0
0.5
0.0
1
1
1
1
10°C
110°C
115°C
120°C
125°C
130°C
120°C
1
20°C
15°C
20°C
25°C
125°C
130°C
125°C
130°C
135°C
140°C
10
0
1
35°C
40°C
1
1
30°C
0
2
4
6
8 101214
0
2
4
6
8
10
0
2
4
6
8
10
0
2
4
6
8
10
0
1
2
3
4
0
1
2
3
4
Time (h)
Time (h)
C
12
10
8
D
Glucose
Levoglucosan
HMF
Glucose
9
8
7
6
5
4
3
2
1
0
Levoglucosan
HMF
7
0
70
1
0
8
7
6
5
4
3
2
1
0
65
60
55
50
45
40
35
30
120°C
1
30°C
40°C
60
1
8
6
4
2
0
50
6
4
0
0
4
3
1
20°C
125°C
130°C
2
120°C
130°C
135°C
140°C
120°C
130°C
135°C
140°C
1
35°C
40°C
20
1
0
0
2
4
6
8
10 12
0
2
4
6
8
10 12
0
2
4
6
8
10 12
0
5
10 15 20 25 30
0
2
4
6
8 101214
0 2 4 6 8 101214
Time (h)
Time (h)
Figure 10. Concentrations of glucose, levoglucosan, and HMF as a function of reaction time are shown for glucose reaction in DMSO (A), and DMSO/Water = 95/5 (B), DMSO/
Water = 90/10 (C), and DMSO/Water = 80/20 (D) at various temperatures.
DMSO
-1.0
-1.5
-2.0
-2.5
-3.0
-3.5
-4.0
-4.5
-2.0
DMSO/Water 95/5
DMSO/Water 90/10
DMSO/Water 80/20
-
-
-
-
-
1
2
3
4
5
-2.5
-3.0
-3.5
-4.0
-4.5
-5.0
Glucose
Levoglucosan
HMF
-6
0
.0024
0.0026
0.0025
0.0026
0.0025
0.0026
-1
1/T (K )
Figure 11. The Arrhenius plots for glucose depletion, levoglucosan, and HMF formation in DMSO and DMSO/Water mixtures.
seen that dehydration product peaks from levoglucosan and HMF
keep increasing whereas glucose peaks keep decreasing as the reac-
tion continues. Figure 10 plots the concentrations of glucose, levo-
glucosan, and HMF as a function of reaction time at various reaction
temperatures in DMSO (A), DMSO/Water = 95/5 (B), DMSO/
Water = 90/10 (C) and DMSO/Water = 80/20 (D) solvents. It can be

5
8
X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
activation barriers for the reactions can be determined. These en-
ergy barriers are shown in Figure 12. It can be seen that energy bar-
riers are different in different solvents. For HMF production, the
barriers are around 40, 24, 28, and 34 kcal/mol in 100/0, 95/5,
9
0/10, and 80/20 DMSO/Water ratios respectively. Even though
the calculated theoretical barrier is somewhat higher than the
experimental value for glucose to HMF conversion in pure DMSO
solvent, the overall trend obtained from CPMD–MTD simulations
is correct in that HMF production in pure DMSO has a higher bar-
rier than in aqueous solution. On the other hand, levoglucosan ap-
pears to have the lowest barrier (19 kcal/mol) in DMSO and the
barrier increases with more water content in the mixed solvents
(
25–27 kcal/mol). This is understandable as water is a product of
glucose dehydration. Since glucose conversion to levoglucosan is
reversible, the presence of water will slow down or reverse the
levoglucosan production.
4
.4. The effects of solvent on atomic charges
Figure 12. Experimental energy barriers for glucose depletion, levoglucosan, and
HMF production in DMSO and DMSO/Water mixtures.
It is relatively easy to understand why the energy barriers in
seen that the rate of HMF production increases and the rate of levo-
glucosan production decreases in solvents as the water content in
the solvent mixture increases. The HMF concentration almost
reaches a plateau after 4 h of reaction in DMSO solvent whereas it
continues to increase linearly in solvents with larger water content
with DMSO/Water ratio at 90/10 and 80/20 (v/v). Furthermore,
levoglucosan production is much faster in pure DMSO solvent and
it slows down significantly when the water content increases. The
slow down in HMF production in pure DMSO after 2 h of reaction
can be partly caused by the rapid consumption of glucose due to
higher rate of levoglucosan production. Nevertheless, it is clear
from the data that the reaction rates are strongly solvent
dependent.
In order to determine quantitatively the reaction barriers for
HMF production in different DMSO/Water solvent mixtures, first-
order kinetics are assumed for glucose depletion as well as for
HMF and levoglucosan production for the reactions. Figure 11
shows the Arrhenius plots for glucose conversion, HMF, and
levoglucosan production at various reaction temperatures in dif-
ferent DMSO/Water mixtures. Based on these Arrenhius plots, the
water and DMSO solvents associated with protonation of C2AOH
and the breakage of C2AO2 bond during glucose dehydration reac-
tions are different. However, it is not obvious why the reactivity for
the subsequent bonding of C2AO5 is different in the two solvents.
Since water is more polar than DMSO, the C2-carbocation is ex-
pected to be more stable in water than in DMSO. It seems that
bonding between C2AO5 should have a higher energy barrier in
water than in DMSO. On the contrary, our calculations show that
the formation of C2AO5 bond in aqueous solution has a barrier
7
of about 20 kcal/mol whereas it has significantly a higher barrier
of over 50 kcal/mol in DMSO solvent. As a result, the formation
of C2AO5 bond becomes the slow and rate-limiting step during
glucose dehydration reaction in DMSO. It is known that charge dis-
tribution and electrostatic potential play a critical role in chemical
reactions in a polar solvent. In order to get a better understanding
of the effects of solvent on glucose reactivity, atomic charges based
3
0
on the electrostatic potential (ESP) were calculated using both
Gaussian09 and CPMD codes. The ESP atomic charges of glucose
atoms in the gas phase as well as in DMSO or H O using implicit
2
3
2,33,35
CPCM solvation model
were calculated with Gaussian09.
Table 1
2
Calculated ESP atomic charges for glucose molecule in the gas phase and in H O and DMSO solvents
Atom
Type
G09 Gas Phase w/o
PBC
CPMD Gas Phase w/o
PBC
G09 CPCM w/o
PBC
CPMD H
PBC
2
O
CPMD H
PBC
2
O w/o
CPMD DMSO
PBC
CPMD DMSO w/o
PBC
C1
O1
H1(C)
C2
O2
H2(C)
C3
O3
H3(C)
C4
O4
H4(C)
C5
O5
H5(C)
C6
O6
0.362
ꢀ0.654
0.033
0.486
ꢀ0.598
0.006
0.594
ꢀ0.723
0.016
ꢀ0.107
ꢀ0.709
0.129
ꢀ0.504
ꢀ0.063
0.502
0.086
ꢀ0.373
0.019
1.424
ꢀ1.05
0.347
ꢀ0.642
0.027
ꢀ0.227
0.120
0.123
ꢀ1.727
0.316
0.317
0.435
ꢀ0.693
ꢀ0.599
0.328
ꢀ0.451
ꢀ0.811
ꢀ0.247
1.308
ꢀ0.521
ꢀ0.438
0.559
ꢀ1.176
ꢀ0.362
1.544
ꢀ1.214
ꢀ0.403
0.682
ꢀ0.681
ꢀ0.009
0.042
0.113
0.005
0.734
0.111
0.220
0.538
0.381
ꢀ0.134
ꢀ0.38
0.468
ꢀ0.919
ꢀ0.167
0.55
ꢀ0.074
ꢀ0.05
0.397
ꢀ1.067
ꢀ0.266
0.408
0.231
ꢀ0.699
ꢀ0.569
ꢀ0.028
0.096
ꢀ0.724
ꢀ0.634
ꢀ0.015
0.026
ꢀ0.465
0.080
0.094
0.061
0.013
ꢀ0.144
ꢀ0.691
0.134
0.392
ꢀ0.694
ꢀ0.617
ꢀ0.539
ꢀ0.738
0.117
0.030
0.093
0.005
0.138
0.305
0.169
0.373
0.086
ꢀ0.471
ꢀ0.619
ꢀ0.491
ꢀ0.449
ꢀ0.478
0.083
0.023
0.076
0.021
0.065
0.263
0.240
0.330
0.106
0.216
ꢀ0.684
ꢀ0.006
0.102
ꢀ0.614
ꢀ0.767
ꢀ0.003
0.084
ꢀ0.366
ꢀ0.941
ꢀ0.251
ꢀ0.234
ꢀ0.647
ꢀ0.004
0.021
H6(C,1)
H6(C,2)
0.001
0.013
0.026
0.437
0.082
⁄
⁄
Calculations were also carried out using Gaussian09 with B3LYP functional and 6-311++G basis set in the gas phase and in DMSO, H
2
O solvents with implicit solvation
model (CPCM). ESP charges were also calculated for glucose in the gas phase and when the glucose is solvated by explicit H
functional both with and without periodic boundary conditions (PBC).
2
O or DMSO molecules using CPMD with BLYP

X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
59
Figure 13. The ESP atomic charges of C and O on glucose in the gas phase, in aqueous solution, and in DMSO solvent at various computation conditions.
Variations in calculated charges are negligible using other implicit
solvation methods. The calculated ESP atomic charges for the C, O,
and H (C) atoms on glucose are shown in Table 1. To ensure accu-
racy, the ESP charges for glucose in the gas phase, and when glu-
cose is solvated by the explicit solvent molecules both with and
without PBC were also determined using CPMD. The differences
in calculated charges using different exchange-correlation func-
tional and plane wave cut-offs on ESP atomic charges are small.
However, PBC is found to have a significant influence on the ESP
charges. The partial charges for H(O) are not shown in Table 1 as
the charges are all between 0.3 and 0.5 for all the AOH groups
on glucose from different calculations. The charges on C and O
atoms are plotted in Figure 13 in order to illustrate more clearly
solvated by H
81% in dipole moment upon solvation.
2
O is highly polarizable with an increase of over
As shown in Figure 13, the atomic charges on glucose C and O in
2
O change in the opposite direction to those in DMSO with PBC.
H
This indicates that besides their different proton affinity, DMSO
and H O solvents affect the charge distribution on glucose rather
2
differently. Because of these differences, the reactivity of glucose
in these two solvents is very different. However, it does not explain
why the barrier for C2AO5 bond formation is significantly higher
2
in DMSO than in H O. Besides the neutral molecule, the charge dis-
tribution for the C2-carbocation intermediate is likely to be af-
fected by the solvent as well. Moreover, the stability of the initial
glucose structure, the C2-carbocation intermediate and the final
cyclic 5-member ring are likely strongly affected by the solvent.
All these factors will affect the energetics of the glucose reactions
in solution.
2
large charge fluctuations in H O and DMSO solvents.
It can be seen that the corresponding atomic charges in the gas
phase are very similar to each other calculated by either Gauss-
ian09 or CPMD. Under solution using Gaussian09 with implicit sol-
vation model, the calculated charges on O atoms become slightly
more negative. The charges on C2 and C4 also become slightly neg-
ative in the solvent whereas charges on C1, C3, and C6 become
slightly more positive. The differences in calculated charges using
5
. Conclusion
The mechanisms and the energy barriers for the critical steps
during proton catalyzed glucose condensation and dehydration
reactions in DMSO were determined. Glucose condensation reac-
tion is initiated by protonation of C1AOH on glucose and subse-
quent breakage of the C1AO1 bond whereas glucose dehydration
reaction is initiated by protonation of C2AOH and the breakage
of C2AO2 bond followed by the formation of the C2AO5 bond.
The mechanisms are similar to those in the aqueous solutions.
However, energy barriers and critical steps are strongly influenced
by the solvent molecules. The DMSO molecules are found to partic-
ipate directly in these reactions as proton acceptors during the
reactions. Moreover, the long-range electrostatic interactions ap-
pear to influence the reactivity of the glucose molecule substan-
tially through charge redistribution from polarization. Due to
strong polarization from the long-range electrostatic interactions,
theoretical calculations conducted in the gas phase or using impli-
cit solvent models are generally not accurate for these complex
reactions involving carbohydrates. Our experimental results
2
either H O or DMSO solvent in the implicit solvent model are neg-
ligible. The ESP atomic charges calculated by CPMD with explicit
solvent model without PBC are only slightly different from the
charges calculated in the gas phase. However, when PBC is applied,
the atomic charges of C, O, and H(C) on glucose change dramati-
cally compared to the systems without PBC. The C atoms on glu-
cose become rather negative in H
positive in DMSO. The O atoms on glucose become less negative
in H O whereas they become more negative in DMSO. In addition,
the charges on glucose H(C) become positive in H O and negative
2
O whereas they become highly
2
2
in DMSO. This indicates that long range electrostatic interaction
has a profound impact on the charge distribution of the glucose
molecule. The C, O, and H(C) atoms on glucose are highly polariz-
able and can be strongly affected by their electrostatic environ-
ment. However, the H(O) charges are not affected by the specific
environment. Earlier studies39 also indicate that DMSO molecule

6
0
X. Qian, D. Liu / Carbohydrate Research 388 (2014) 50–60
support the solvent effects on the energy barrier for glucose dehy-
dration to HMF.
15. Car, R.; Parrinello, M. Phys. Rev. Lett. 1985, 55, 2471–2474.
16. CPMD 3.13, copyrighted jointly by IBM Corp and by Max-Planck Institute,
Stuttgart.
17. Laio, A.; Parrinello, M. Proc. Natl. Acad. Sci. 2002, 99, 12562–12566.
Acknowledgements
18. Frisch, M. J. T.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.;
Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.;
Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.;
Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven,
T., ; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.;
Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.;
Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.;
Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.;
Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.;
Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.;
Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas,
Ö.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09 Revision A.1;
Gaussian: Wallingford CT, 2009.
Funding from NSF CAREER (CBET 1137795) is gratefully
acknowledged. Initial structural generation for the calculations
by Dr. Hongbo Du is also appreciated. Calculations were carried
out on Razor from the High Performance Computing Center
(
HPC) at the University of Arkansas and Teragrid.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.carres.2014.02.
1
2
9. Iannuzzi, M.; Laio, A.; Parrinello, M. Phys. Rev. Lett. 2003, 90, 238302.
0. Dong, H.; Qian, X. Ab Initio Molecular Dynamics Investigation of Xylan
Hydrolysis In Computational Modeling in Lignocellulosic Biofuel Production;
Nimlos, M. R., Crowley, M. F., Eds.; ACS Symposium Series, 2011; pp 1–15.
1. Ensing, B.; De Vivo, M.; Liu, Z.; Moore, P.; Klein, M. L. Acc. Chem. Res. 2006, 39,
0
10.
2
2
References
73–81.
2. Becke, A. D. Phys. Rev. A 1988, 38, 3098–3100.
1
2
3
4
5
6
.
.
.
.
.
.
Qian, X. H.; Nimlos, M. R.; Davis, M.; Johnson, D. K.; Himmel, M. E. Carbohydr.
Res. 2005, 340, 2319–2327.
Qian, X. H.; Nimlos, M. R.; Johnson, D. K.; Himmel, M. E. Appl. Biochem.
Biotechnol. 2005, 121, 989–997.
23. Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789.
24. Goedecker, S.; Teter, M.; Hutter, J. Phys. Rev. B 1996, 54, 1703–1710.
25. Qian, X. H. Mol. Simul. 2008, 34, 183–191.
26. Qian, X. H.; Ding, S. Y.; Nimlos, M. R.; Johnson, D. K.; Himmel, M. E.
Macromolecules 2005, 38, 10580–10589.
27. Nose, S. J. Chem. Phys. 1984, 81, 511–519.
28. Hoover, W. G. Phys. Rev. A 1985, 31, 1695.
29. Ewald, P. Ann. Phys. 1921, 64, 253–287.
30. Singh, U. C.; Kollman, P. A. J. Comput. Chem. 1984, 5, 129–145.
31. Hirshfeld, F. L. Theor. Chim. Acta 1977, 44, 129–138.
32. Liptak, M. D.; Shields, G. C. Int. J. Quantum Chem. 2001, 85, 727–741.
33. Toth, A. M.; Liptak, M. D.; Phillips, D. L.; Shields, G. C. J. Chem. Phys. 2001, 114,
4595–4606.
34. Tissandier, M. D.; Cowen, K. A.; Feng, W. Y.; Gundlach, E.; Cohen, M. H.; Earhart,
A. D.; Coe, J. V.; Tuttle, T. R. J. Phys. Chem. A 1998, 102, 7787–7794.
35. Dong, H.; Du, H.; Qian, X. J. Phys. Chem. A 2008, 112, 12687–12694.
36. Kelly, C. P.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2006, 111, 408–422.
37. Cleland, W. W.; Kreevoy, M. M. Science 1994, 264, 1887–1890.
38. Ohara, M.; Takagaki, A.; Nishimura, S.; Ebitani, K. Appl. Catal. A: Gen. 2010, 383,
149–155.
Qian, X. H.; Johnson, D. K.; Himmel, M. E.; Nimlos, M. R. Carbohydr. Res. 2010,
345, 1945–1951.
Nimlos, M. R.; Qian, X. H.; Davis, M.; Himmel, M. E.; Johnson, D. K. J. Phys. Chem.
A 2006, 110, 11824–11838.
Dong, H.; Nimlos, M. R.; Himmel, M. E.; Johnson, D. K.; Qian, X. J. Phys. Chem. A
2
009, 113, 8577–8585.
Liu, D. J.; Nimlos, M. R.; Johnson, D. K.; Himmel, M. E.; Qian, X. H. J. Phys. Chem. A
010, 114, 12936–12944.
2
7
8
9
.
.
.
Qian, X. H. J. Phys. Chem. A 2011, 115, 11740–11748.
Qian, X. H. Top. Catal. 2012, 55, 218–226.
Qian, X. H.; Wei, X. F. J. Phys. Chem. B 2012, 116, 10898–10904.
1
1
1
0. Antal, M. J.; Mok, W. S. L.; Richards, G. N. Carbohydr. Res. 1990, 199, 111–115.
1. Antal, M. J.; Mok, W. S. L.; Richards, G. N. Carbohydr. Res. 1990, 199, 91–109.
2. Antal, M. J.; Leesomboon, T.; Mok, W. S.; Richards, G. N. Carbohydr. Res. 1991,
217, 71–85.
1
1
3. Pilath, H. M.; Nimlos, M. R.; Mittal, A.; Himmel, M. E.; Johnson, D. K. J. Agric.
Food Chem. 2010, 58, 6131–6140.
4. Qian, X. J. Phys. Chem. B 2013, 117, 11460–11465.
39. Kirchner, B.; Hutter, J. J. Chem. Phys. 2004, 121, 5133–5142.