Successive C1-C2 bond cleavage: The mechanism of vanadium(v)-catalyzed aerobic oxidation of d-glucose to formic acid in aqueous solution

Article abstract of DOI:10.1039/c8cp02352b

Vanadium(v)-catalyzed aerobic oxidation in aqueous solution shows high selectivity in the field of C-C bond cleavage of carbohydrates for chemicals with less carbon atoms. However, the pathway of C-C bond cleavage from carbohydrates and the conversion mechanism are unclear. In this work, we studied the pathway and the mechanism of d-glucose oxidation to formic acid (FA) in NaVO₃-H₂SO₄ aqueous solution using isotope-labeled glucoses as substrates. d-Glucose is first transformed to FA and d-arabinose via C1-C2 bond cleavage. d-Arabinose undergoes similar C1-C2 bond cleavage to form FA and the corresponding d-erythrose, which can be further degraded by C1-C2 bond cleavage. Dimerization and aldol condensation between carbohydrates can also proceed to make the reaction a much more complicated mixture. However, the fundamental reaction, C1-C2 bond cleavage, can drive all the intermediates to form the common product FA. Based on the detected intermediates, isotope-labelling experiments, the kinetic isotope effect study and kinetic analysis, this mechanism is proposed. d-Glucose first reacts with a vanadium(v) species to form a five-membered-ring complex. Then, electron transfer occurs and the C1-C2 bond weakens, followed by C1-C2 bond cleavage (with no C-H bond cleavage), to generate the H₃COO-vanadium(iv) complex and d-arabinose. FA is generated from H₃COO that is oxidized by another vanadium(v) species. The reduced vanadium species is oxidized by O₂ to regenerate to its oxidation state. This finding will provide a deeper insight into the process of C-C bond cleavage of carbohydrates for chemical synthesis and provide guidance for screening and synthesizing new highly-efficient catalyst systems for FA production.

Full text of DOI:10.1039/c8cp02352b

PCCP
View Article Online
View Journal
PAPER
Successive C1–C2 bond cleavage: the mechanism
of vanadium(V)-catalyzed aerobic oxidation
of D-glucose to formic acid in aqueous solution†
Cite this:DOI: 10.1039/c8cp02352b
c
Muge Niu,a Yucui Hou,b Weize Wu, *a Shuhang Rena and Ru Yang
Vanadium(V)-catalyzed aerobic oxidation in aqueous solution shows high selectivity in the field of C–C
bond cleavage of carbohydrates for chemicals with less carbon atoms. However, the pathway of C–C
bond cleavage from carbohydrates and the conversion mechanism are unclear. In this work, we studied
the pathway and the mechanism of D-glucose oxidation to formic acid (FA) in NaVO3–H2SO4 aqueous
solution using isotope-labeled glucoses as substrates. D-Glucose is first transformed to FA and D-arabinose
via C1–C2 bond cleavage. D-Arabinose undergoes similar C1–C2 bond cleavage to form FA and the
corresponding D-erythrose, which can be further degraded by C1–C2 bond cleavage. Dimerization and
aldol condensation between carbohydrates can also proceed to make the reaction a much more compli-
cated mixture. However, the fundamental reaction, C1–C2 bond cleavage, can drive all the intermediates
to form the common product FA. Based on the detected intermediates, isotope-labelling experiments, the
kinetic isotope effect study and kinetic analysis, this mechanism is proposed. D-Glucose first reacts with
a vanadium(V) species to form a five-membered-ring complex. Then, electron transfer occurs and the
C1–C2 bond weakens, followed by C1–C2 bond cleavage (with no C–H bond cleavage), to generate
the H3COOꢀ–vanadium(IV) complex and D-arabinose. FA is generated from H3COOꢀ that is oxidized
by another vanadium(V) species. The reduced vanadium species is oxidized by O2 to regenerate to its
oxidation state. This finding will provide a deeper insight into the process of C–C bond cleavage of carbo-
hydrates for chemical synthesis and provide guidance for screening and synthesizing new highly-efficient
catalyst systems for FA production.
Received 13th April 2018,
Accepted 1st June 2018
DOI: 10.1039/c8cp02352b
rsc.li/pccp
C–H bond activation to yield the corresponding sugar acids and
sugar alcohols.
1. Introduction
Selective conversion of sufficient renewable bio-derived carbo-
Notably, for value-added products with less carbon atoms
hydrates into chemicals shows great potential as an alternative from carbohydrates, selective C–C bond cleavage is required.
strategy for solving problems caused by fossil fuel utilization. However, the C–C bond cleavage of carbohydrates often exhibits
Compared with widely used biochemical processes, catalysis various byproducts or requires harsh conditions, even in some
often provides more options for the tunable design of products.1–3 current effective catalytic processes. For example, 3–11%
Highly selective catalytic routes for chemicals from carbohydrates byproduct formation in methyl lactate production from glucose
have been developed for decades. Representative successful and sucrose occurs in methanol catalyzed by Lewis acidic
examples are Lewis acid-catalyzed isomerization/dehydration of zeolites.4 Various kinds of alcohols are generated in ethylene
D-glucose to yield 5-hydroxymethyl furfural, and metal-catalyzed glycol production from cellulose and glucose via tungsten
carbide/oxide-catalyzed hydrogenation.5–7 High temperature
and an excess amount of alkali were required while producing
a State Key Laboratory of Chemical Resource Engineering, College of Chemical
C1–C3 products with high yields via oxidation.8–11
Engineering, Beijing University of Chemical Technology, Beijing 100029,
The low efficiency of C–C bond cleavage of carbohydrates
is possibly attributed to two reasons. First, the high steric
constraints derived from each carbon atom bonded to a
hydrogen atom and a hydroxyl group make the catalyst hard
to approach. Second, various reactions easily occur in carbo-
hydrate solution, including tautomerism, dehydration, rearrange-
ment acetalization and aldol/retro-aldol condensation,12
P. R. China. E-mail: wzwu@mail.buct.edu.cn; Fax: +86 10 64427603;
Tel: +86 10 64427603
b Department of Chemistry, Taiyuan Normal University, Taiyuan 030031,
P. R. China
c College of Materials Science and Engineering, Beijing University of Chemical
Technology, 100029, Beijing, P. R. China
† Electronic supplementary information (ESI) available: Schemes S1 and S2, and
Fig. S1–S12. See DOI: 10.1039/c8cp02352b
This journal is ©the Owner Societies 2018
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
leading to competition with or interference in catalytic C–C
bond cleavage.
The isotope-labelling technique is a powerful tool for the
detection of the mechanism of an unknown chemical process.
Stoichiometric inorganic high-valence oxidants (Cr(VI),13 Transformation of the 13C-labeled vanadium-catalyzed reaction
I(VII),14,15 Mn(IV),16 and Pb(IV)17) have long been used as reagents has already shown a clear contribution of each carbon atom in
for C–C bond cleavage of compounds with vicinal diols to glucose to the final products.26 In this work, we carried out a
produce the corresponding carbonyl compounds. The high further more in-depth investigation towards the mechanism of
selectivity of C–C bond cleavage was due to the strong inter- vanadium(V)-catalyzed oxidative C–C bond cleavage of carbo-
action with the adjacent hydroxyl groups to counteract the hydrates by converting 13C-/D-labeled D-glucose in NaVO3–H2SO4
effect of steric constraints and side reactions. However, these aqueous solution with O2 as an oxidant. The NaVO3–H2SO4
extremely strong oxidants can hardly be recovered (by oxida- system was employed because it produces the highest yield of
+
tion) after the reaction, leading to excess consumption and FA in aqueous media, and the aqueous form of NaVO3 (VO2
severe waste pollution.
species) was much more simple and representative than other
In this context, vanadium(V) species exhibit a similar manner vanadium(V) systems. Intermediates were detected by NMR and
to the stoichiometric oxidants with a milder oxidation capacity, LC-MS techniques, and the mechanism of C–C bond cleavage
and thus can be used in catalytic amounts with oxygen as an was reasonably proposed. Remarkably, this study suggests a
oxidant.18 Therefore, a series of homogeneous vanadium(V)- novel, successive C1–C2 bond cleavage to generate FA and the
catalyzed aerobic oxidations (H5PV2Mo10O40,19–26 H4PVMo11O40,27 corresponding carbohydrate with one carbon atom less. This new
30–32
PVMo11O404À-based ionic liquid,28 VOSO429 and NaVO3–H2SO4
)
finding will provide deeper insight into the selective C–C bond
were developed and significantly enhanced the selectivity of cleavage of carbohydrates and provide guidance for screening
C–C bond cleavage of carbohydrates. The solvent is water and and synthesizing new highly-efficient catalyst systems for FA
the product in the aqueous phase is solely formic acid (FA), production.
with yields up to B68% at 70–160 1C (the other product is CO2).
FA is an important chemical in industry, especially for very
recent applications in hydrogen storage33–37 and fuel cells.38–40
Therefore, this brand new highly selective C–C bond cleavage
is considered as a promising method for the production of
value-added small molecular chemicals from bio-derived
carbohydrates.
2. Experimental section
2.1 Chemicals
D-Glucose-1-13C (98%), D-glucose-2-13C (99%), D-glucose-3-13
C
(99%), D-glucose-1-D (98%), D-glucose-2-D (98%) and D-arabinose-
1-13C (99%) were purchased from Cambridge Isotope Laboratories
(USA). Anhydrous D-(+)-glucose (99%) was purchased from
Sinopharm Chemical Reagent Co., Ltd (Beijing, China). D-(À)-
Arabinose (99%), D-erythrose (75% syrup), D-glyceraldehyde
(99%), glycolaldehyde (dimer, 98%) and sodium metavanadate
(NaVO3, AR) were purchased from Aladdin Reagent Inc.
(Shanghai, China). Sulfuric acid (H2SO4, 98%) was purchased
from Beijing Modern Oriental Fine Chemistry Co., Ltd (Beijing,
China). Oxygen (O2, 99.995%) and nitrogen (N2, 99.999%) were
supplied by Beijing Haipu Gases Co., Ltd (Beijing, China).
All reagents were of analytical grade and used without further
purification.
Fruitful results have been achieved with respect to the
mechanism. Fu et al.23 and Wu et al.30 selected possible inter-
mediates as substrates, and proposed a pathway from D-glucose
(representative carbohydrate) to FA initiated with C2–C3 and/or
C3–C4 bond cleavage via retro-aldol condensation, followed by
continued C–C bond cleavage to form FA. Wasserscheid et al.19
and Wang et al.29 abstracted representative structures from
D-glucose to study the transformation of each functional group.
Neumann et al.26 tested the product distributions of each
carbon atom in D-glucose, and showed that C1 and C5 have the
highest selectivity to FA and CO2, respectively. Tomishige et al.41
proposed a mechanism between ketol and vanadium(V) ligands
under O2 oxidation with a ketol–vanadium(V)–biradical O2
complex as the key intermediate before C–C bond cleavage.
2.2 Conversion of substrates
However, the mechanism still remains problematic. For the The conversion of D-glucose and D-arabinose for 13C NMR
pathway, no direct detectable evidence supports the proposed analysis was carried out in a 25 cm3 batch reactor of Hastelloy
retro-aldol condensation of D-glucose at the initial step. In fact, alloy (HC 276) with a magnetic stirrer. The conversion of the
the retro-aldol condensation is more likely to be a base- isotope-labeled substrates was carried out in a 1.5 cm3 capped
catalyzed reaction,10 and is kinetically unfavorable in the acidic glass vial, which was equipped with a porous cap and a
vanadium(V) aqueous system. The inactive retro-aldol conden- magnetic stirrer, fixed in the above-mentioned batch reactor.
sation is in conflict with the generally high rate of FA formation. In a typical procedure, a certain amount of substrate and
Therefore, there must be another pathway to replace the retro- NaVO3–H2SO4 (H2SO4, NaVO3, or NaVO3–D2SO4) aqueous solution
aldol condensation. In addition, the information on the inter- were loaded into the reactor (or the glass vial). Then the reactor
action between D-glucose and the vanadium(V) catalyst and bond was sealed and purged with O2 (or N2). After that, O2 (or N2) was
cleavage is insufficient, except for an observation of a valence charged into the reactor to a desired pressure. Next, the reactor
change in the vanadium atom (VV " VIV interconversion) was put into a heating furnace and stirred at a speed of 500 rpm.
followed by a brief explanation relying on a pre-existing well- The pressure and temperature of the reactor were measured by a
established electron transfer–oxygen transfer mechanism.42
pressure transducer (Æ0.025 MPa) and a thermocouple (Æ0.5 1C),
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2018

View Article Online
PCCP
Paper
respectively. The reaction time was recorded when the desired
reaction temperature was reached (B25 min after being put into
the furnace). After the reaction, the reactor was quenched by cold
water. When the reactor reached room temperature, the gas was
released. The liquid sample was analyzed by NMR and LC-MS.
The kinetics experiments were carried out in a 50 cm3
reactor of Hastelloy alloy (HC 276) with a magnetic stirrer
and a liquid outlet. In a typical procedure, a certain amount
of substrate and 30 cm3 of NaVO3–H2SO4 aqueous solution
were loaded into the reactor. The operation was just the same
as that of the batch reactor, except for a higher stirring speed of
900 rpm and a shorter heat-up time of 5 min. The reaction time
was recorded when the desired reaction temperature was reached.
The liquid sample from the outlet was collected every 30 min. The
liquid sample was analyzed by liquid chromatography.
2.3 Analysis of the products
The yields of the products in the liquid sample were calculated
using
a high-performance liquid chromatograph (HPLC,
Waters 2695, USA) with a Shodex SH 1011 column (Shodex,
Japan). A differential refractive index detector (Waters 4110,
USA) was employed to analyze the products. The column oven
temperature was 55 1C. The mobile phase was diluted H2SO4
aqueous solution at a concentration of 0.01 mol dmÀ3 and a
flow rate of 0.5 cm3 minÀ1. All the yields of the products were
calculated on the carbon base. The intermediates formed
during the conversion were detected using LC-MS and NMR
techniques. The LC part of LC-MS was a high-pressure liquid
Fig. 1 13C NMR spectra of the reaction mixture with D-glucose as the
chromatography instrument (Agilent 1290 Infinity, USA) substrate at different reaction times. (A) 2 h, (B) 4 h, (C) 8 h, (D) 12 h, and
(E) the region of 55–100 ppm of (A), showing signals of D-arabinose (a).
equipped with a Zorbax NH2 column (Agilent, USA). The mobile
phase was water/acetonitrile with gradient elution. The MS part
was a micrOTOF QII mass spectrometer (Bruker Daltonics,
Reaction conditions: D-glucose, 1 g; NaVO3, 0.35 wt%; H2SO4, 1 wt%; H2O,
6 cm3; initial O2, 3 MPa; temperature, 100 1C.
Germany) using an electrospray ion (ESI) source in the positive
mode. The 13C NMR and 1H NMR spectra of the liquid mixture
spectrum (Fig. S1, ESI†) and the detection of C5H10O5Na+ (m/z:
were detected on a Bruker AV400 MHz spectrometer with
173, Fig. S2, ESI†) in LC-MS of the reaction mixture provide
deuterium oxide as the reaction medium. The FA formed in
further evidence for D-arabinose formation. Simultaneously, FA
the conversion of glucose in H218O was extracted with a twofold
was formed and accumulated (165.80 ppm, Fig. 1).
greater volume of ethyl ether (AR). The GC-MS measurements
D-Arabinose is a pentose with one carbon less than glucose.
of the extracted FA were recorded using an ISQ MS detector and
We observed that D-arabinose shares the same carbon skeleton
a TRACE 1300 gas chromatograph equipped with a TG-WaxMS
from C2 to C5 with D-glucose (on both linkage and chirality).
column (30 m  0.25 mm  0.25 mm). The carrier gas was
Therefore, we speculate that C1–C2 bond cleavage occurs in
helium.
glucose oxidation to generate FA and D-arabinose (Scheme 1).
A series of 13C-labeled experiments were then carried out for
further confirmation of this assumption. The 13C-labeled FA
3. Results and discussion
3.1 Initial C1–C2 bond cleavage to FA and arabinose
All the reactions of D-glucose oxidation by O2 in NaVO3–H2SO4
aqueous solution in this work were performed at 100 1C, lower
than the applied temperatures for FA production (usually
4100 1C). Under these relatively mild conditions, the D-glucose
conversion rate was slowed down and more intermediates were
possibly formed and preserved in the reaction mixture available
for detection. The 13C NMR spectra for the reaction mixtures
indicate that an intermediate, D-arabinose, was formed and then
diminished with a prolonged reaction time (Fig. 1). The 1H NMR Scheme 1 D-Arabinose and FA formation from D-glucose.
This journal is ©the Owner Societies 2018
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
products at one specific reaction time. Considering that
D-arabinose can soon be converted as formed, we conducted
the reaction (D-glucose-1-13C as the substrate) in a very short
time to prevent further degradation of the formed D-arabinose
as much as possible. The reaction mixture was analyzed by
1H NMR (Fig. 3). In the downfield of the 1H NMR spectrum, the
two symmetric peaks (7.85 ppm and 8.40 ppm) are assigned to
aldehydic H in 13C–FA due to splitting of the H signal by the
coupling effect between 1H and 13C (1JH,C = 219.2 Hz). The
weaker signal between the split two peaks is assigned to
aldehydic H in unlabeled FA. The integration of the peaks of
the H signals in 13C–FA and in H1 of D-arabinose in the
a-conformation (4.35 ppm) gives the ratio of the formed
13C–FA to D-arabinose B0.9 : 1. This result clearly shows that
C1–C2 bond cleavage to generate FA is the main reaction at the
initial step of the whole conversion.45
Fig. 2 13C NMR spectra of the reaction mixture with (A) D-glucose-1-13
C
and (B) D-glucose-2-13C as the substrate. Reaction conditions: D-glucose-
1-13C or -2-13C, 0.005 g; NaVO3, 0.2 wt%; D2SO4, 1 wt%; D2O, 0.5 cm3;
initial O2, 3 MPa; temperature, 100 1C; reaction time, 10 min.
In contrast, experiments with no NaVO3–H2SO4 or NaVO3
under the same conditions show no formation of any products,
suggesting a specific interaction between the catalyst and
D-glucose.
was largely formed from the short-time reaction of D-glucose-
1-13C, as detected by 13C NMR (Fig. 2A, the 13C assignment in
13C NMR is listed in Table S1, ESI†). Unlabeled D-arabinose
should be formed, but the intensity was too low to be observed
3.2 Parallel side-reaction: epimerization of D-glucose
to D-mannose with a C1–C2 shift
compared with the labeled products. When D-glucose-2-13
C
Interestingly, when a larger amount of NaVO3–H2SO4 aqueous
solution was employed, a side reaction appeared with the
detection of a new labeled carbohydrate, D-mannose-2-13C in
D-glucose-1-13C transformation (Fig. S3, ESI†). The reaction
with D-glucose-2-13C under identical conditions gave a labeled
D-mannose-1-13C (Fig. S4, ESI†). With the absence of O2
supply (replacement with N2), unlabeled D-glucose produces a
considerable amount of D-mannose, as clearly detected from
the 13C NMR spectrum (in order to give an observable result,
the reaction was carried out in one batch reactor three times by
adding 0.35 wt% NaVO3 each time). This result reveals that
NaVO3–H2SO4 catalyzes the epimerization of glucose to mannose
via a C1–C2 shift (Scheme 3). This epimerization possibly involves
an interaction of the catalyst and C3 of D-glucose, C2–C3 bond
cleavage and C1–C3 bond formation.12,46,47
was employed as the substrate under identical conditions,
D-arabinose-1-13C was detected in the 13C NMR spectrum with
a weaker signal of 13C-labeled FA (Fig. 2B). Notably, no other
labeled products were found in both spectra. These results
indicate that D-glucose firstly undergoes C1–C2 bond cleavage
with the C1 segment to form FA and the C2–C6 segment to form
D-arabinose. Continuous C1–C2 bond cleavage can proceed
(Scheme 2), as evidenced by the delayed formation of 13C–FA
in D-glucose-2-13C conversion.
In aqueous medium, carbohydrates can undergo retro-aldol
condensation, requiring cleavage of the C2–C3 bond43,44 or the
C3–C4 bond.9 Therefore, FA generation from glucose has
usually been considered to involve C2–C3 bond cleavage and/
or C3–C4 bond cleavage. In the NaVO3–H2SO4 catalytic system,
even the 13C-labeled products with two or three carbons were
not detected, and these two kinds of cleavages for FA formation
cannot be entirely excluded. It is also possible that two- or
three-carbon labeled products were formed followed by very
quick degradation to form 13C–FA. One simple way to rule out
this possibility is to measure and compare the amounts of the
Data based on liquid chromatography have revealed that the
D-mannose formation rate on D-glucose conversion is lower
than the glucose formation rate on mannose conversion
(Fig. S5, ESI†). Furthermore, the FA formation rate on glucose
conversion is faster than that on mannose conversion. Notably,
in the whole scope of the reaction time, while D-mannose-2-13
was formed in D-glucose-1-13C conversion, no D-arabinose-1-13
C
C
was observed (Fig. S3, ESI†). These phenomena suggest that,
even though D-mannose was formed, FA was still generated
primarily from D-glucose.
3.3 Further reactions: dimerization, deep C1–C2 bond
cleavage and C–C bond reformation
We used the intermediate D-arabinose as the starting material
to test the pathway of further degradation of D-glucose (Fig. S6,
ESI†). D-Arabinose was converted much faster than D-glucose,
as it was found to be diminished in 2 h by 13C NMR (D-glucose
in more than 8 h, as shown in Fig. 1). This can explain the low
Scheme 2 Successive C1–C2 bond cleavage of D-glucose.
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2018

View Article Online
PCCP
Paper
Fig. 3 1H NMR spectra of the reaction mixture with D-glucose-1-13C. (A) Full spectrum, and (B) the region of 7.5–8.6 ppm of (A). The central signal is
attributed to H in FA, and the split signal is attributed to H in 13C–FA (1JH,C = 219.2 Hz). Reaction conditions: D-glucose-1-13C, 0.005 g; NaVO3, 0.35 wt%;
D2SO4, 1 wt%; D2O, 0.5 cm3; initial O2, 3 MPa; temperature, 100 1C; reaction time, 0 min (the reaction was quenched as the reactor just reached the
desired temperature).
Scheme 3 Epimerization of D-glucose to D-mannose in the NaVO3–
H2SO4 aqueous solution with a C1–C2 shift.
accumulation of D-arabinose in the whole time scope of the
D-glucose conversion. The potential products of C1–C2 bond
Fig. 4 13C NMR spectra of the reaction mixture with D-glucose-1-13
C
cleavage of D-arabinose were indistinct in the 13C NMR spectra.
Valuable information on further reactions can be acquired
according to the 13C-labelling experiments. After a longer
reaction of D-glucose-1-13C, weak signals appeared between
97 and 107 ppm in the 13C NMR spectrum (Fig. 4). These can
be attributed to the C1 of disaccharides. The detection of
as the substrate after a prolonged reaction time. (A) Full spectrum.
(B) The region of 85–110 ppm of (A). Reaction conditions: D-glucose-
1-13C, 0.005 g; NaVO3, 0.2 wt%; D2SO4, 1 wt%; D2O, 0.5 cm3; initial O2,
3 MPa; temperature, 100 1C; reaction time, 2 h.
C
12H22O11Na+ (m/z: 365) in LC-MS analysis (with unlabeled an additional signal attributed to arabinonic acid-1-13
C
D-glucose as the starting material) further gives evidence of (176.12 ppm) in the 13C NMR spectrum (Fig. 5). The 13C NMR
disaccharide formation via the dimerization of D-glucose spectrum of the reaction with D-arabinose-1-13C also provides
(Fig. S7, ESI†). These results indicate that C1–C2 bond cleavage the same signal (176.18 ppm) assigned to arabinonic acid-1-13
C
proceeds predominantly in D-glucose degradation, accompanied (Fig. 6). These indicate that besides C–C bond cleavage, C–H
by a low level of dimerization.
bond cleavage in C1 of D-arabinose has occurred. In the region
Besides the detection of D-arabinose-1-13C and 13C-FA, the of 68–78 ppm, weak signals of C2 in disaccharides were also
reaction of D-glucose-2-13C under identical conditions provides observed.
This journal is ©the Owner Societies 2018
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
Fig. 5 13C NMR spectra of the reaction mixture with D-glucose-2-13C as
the substrate after a prolonged reaction time. (A) Full spectrum, showing
the signals of D-arabinose-1-13C (a) and D-arabinonic acid-1-13C (b).
(B) The region of 65–85 ppm of (A). Reaction conditions: D-glucose-2-13C,
0.005 g; NaVO3, 0.2 wt%; D2SO4, 1 wt%; D2O, 0.5 cm3; initial O2, 3 MPa;
temperature, 100 1C; reaction time, 2 h.
Fig. 7 13C NMR spectra of the reaction mixture with D-glucose-3-13C as
the substrate after a prolonged reaction time. (A) Full spectrum, showing
the signals of D-arabinose-2-13C (a). (B) The region of 60–110 ppm of (A),
showing the signals of D-arabinonic acid-2-13C (b) and D-erythrose-1-13
C
(c). Reaction conditions: D-glucose-3-13C, 0.005 g; NaVO3, 0.2 wt%;
D2SO4, 1 wt%; D2O, 0.5 cm3; initial O2, 3 MPa; temperature, 100 1C;
reaction time, 1 h.
Fig. 6 13C NMR spectra of the reaction mixture with D-arabinose-1-13C as
the substrate after a prolonged reaction time. (A) Full spectrum, showing
the signal of D-arabinonic acid-1-13C (a). (B) The region of 85–110 ppm of
(A). Reaction conditions: D-arabinose-1-13C, 0.005 g; NaVO3, 0.2 wt%;
D2SO4, 1 wt%; D2O, 0.5 cm3; initial O2, 3 MPa; temperature, 100 1C;
reaction time, 1 h.
Fig. 8 13C NMR spectra of the reaction mixture with D-glucose-6-13C as
the substrate. (A) Full spectrum, showing the signals of D-arabinose-5-13
C
(a). (B) The region of 55–105 ppm of (A), showing the signals of
D-erythrose-4-13C (b), D-glyceraldehyde-3-13C (c), glycolaldehyde-2-13
C
(d) and glycolic acid-2-13C (e). Reaction conditions: D-glucose-3-13C,
0.005 g; NaVO3, 0.2 wt%; D2SO4, 1 wt%; D2O, 0.5 cm3; initial O2, 3 MPa;
temperature, 100 1C; reaction time, 1 h.
The 13C NMR spectrum of the reaction with D-glucose-3-13C
shows the signals assigned to D-arabinose-2-13C (Fig. 7). The
signal at 73.79 ppm possibly refers to C2 in arabinonic acid, D-erythrose-1-13C was also identified in the 13C NMR spectrum
further confirming the oxidation of D-arabinose on C1.
of the reaction with D-glucose-3-13C (101.02 ppm and
More information on further reactions can be obtained using 95.87 ppm, Fig. 7). According to the experiments for either C1
D-glucose-6-13C as the substrate. C1–C2 bond cleavage leads to or C2 labeled D-glucose, no C2–C3 bond cleavage products were
D-arabinose-5-13C generation, as detected in the 13C NMR spectrum formed. Thus, D-erythrose-4-13C can only be possibly generated
(66.36 ppm and 62.44 ppm, Fig. 8). The signals of C6 in disac- via C1–C2 bond cleavage of the formed D-arabinose-5-13
C
charides (B61 ppm) and C5 in arabinonic acid should be highly (Scheme 4). In addition, the triose, D-glyceraldehyde-3-13C,
overlapped by the strong response of labeled C6 in D-glucose-6-13C. was also detected (62.31 ppm, Fig. 8). No significant signals
Notably, even though the intensity was weak, a tetrose, of C2 or C3 products were observed in the reaction with
D-erythrose-4-13C, was detected (71.47 ppm and 71.11 ppm). D-arabinose-1-13C, suggesting that D-glyceraldehyde-3-13C is
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2018

View Article Online
PCCP
Paper
68–103 ppm (Fig. S9, ESI†), indicating that the carbohydrate
intermediates actually undergo acid-catalyzed aldol condensation
to form longer-carbon-chain carbohydrates. Notably, the signals
between 85 and 103 ppm (Fig. 8) represent the anomeric carbons
of carbohydrates (that is, C1 of aldoses). This detection indicates
that some regenerated 2-13C- or 3-13C-labeled carbohydrates were
further degraded via C1–C2 bond cleavage, leading to the for-
mation of 1-13C-labeled carbohydrates. This can be evidenced by
the detection of C7H12O7 in LC-MS, which is possibly generated
from C8H14O8 via C1–C2 bond cleavage (Fig. S8, ESI†).
Scheme 4 C1–C2 bond cleavage of D-arabinose to D-erythrose.
In summary, even though various intermediates were
formed in glucose degradation (via C1–C2 bond cleavage,
dimerization or aldol condensation), they largely share similar
characteristics in terms of their chemical structure: most of
them were assigned to carbohydrates. Given the fact that the
final liquid product was solely FA, we can reasonably conclude
that these carbohydrates should share exactly the same funda-
mental mechanism to generate FA: C1–C2 bond cleavage.
Therefore, a detailed discussion about this representative
C1–C2 bond cleavage of D-glucose has been made in the
following section.
3.4 Mechanism of C1–C2 bond cleavage of glucose and
FA formation
Scheme 5 C1–C2 bond cleavage of D-erythrose to D-glyceraldehyde.
Before the mechanism was discussed, we focused on an inter-
mainly derived from D-erythrose-4-13C via C1–C2 bond cleavage mediate, C6H10O6, which was undetected in 13C NMR but
(Scheme 5). The observation of glycolaldehyde-2-13C in the 13C showed signals in LC-MS (m/z: 201, C6H10O6Na+, Fig. S7, ESI†).
NMR spectrum (64.47 ppm, Fig. 8) indicates C1–C2 bond This compound was further identified as 2-D-glucosone (C2
cleavage of D-glyceraldehyde-3-13C. In addition, the glycolic of D-glucose was oxidized by hydrogen abstraction) by the
acid-2-13C detection (59.25 ppm, Fig. 8) suggests the oxidation observation of glucosone–quinoxaline (m/z: 251, C12H15O4N2 ,
+
of the aldehyde group in glycolaldehyde-2-13C. The dimeriza- Scheme S1, ESI†) after adding o-phenylenediamine to the
tion can also be found between D-glucose and D-glyceraldehyde/ reaction mixture. It might be plausible that 2-glucosone is the
glycolaldehyde (or between D-arabinose and D-glyceraldehyde), intermediate from D-glucose to D-arabinose, via hydrogen
as evidenced by the detection of C9H16O8 and C8H14O7 by abstraction on C2 followed by hydrolysis to form FA and
LC-MS (Fig. S8, ESI†).
D-arabinose (a similar reaction has occurred in the degradation
It is also worth noting that in the D-glucose-6-13C reaction, of 2-D-glucosone in phosphate buffer48). We carried out
13C-labeled signals appeared in the region of 68–103 ppm D-glucose oxidation in D2O with NaVO3–D2SO4 and found that
(Fig. 8). The signals between 68 and 85 ppm, which share a no D-labeled D-arabinose was formed (LC-MS showed only the
similar region of the non-terminal carbons in the aldose signal of C5H10O5Na+, representing D-arabinose, Scheme S2A,
structure, suggest a shift of the 13C-labeled carbon from the ESI†). This result indicates that the hydrogen in C1 of
terminal position (in D-glucose) to the non-terminal position as D-arabinose is not from aqueous solution (neither D2SO4 nor
a form of some new types of carbohydrates. These carbohy- D2O), thus excluding 2-glucosone as the precursor.
drates can most likely be generated by C–C bond reformation
Additionally, no exchange between H2 and any other H atom
between the existing terminal 13C-labeled products in the reaction of D-glucose occurs during C1–C2 bond cleavage as proved
mixture, probably by aldol condensation, which commonly occurs respectively by using D-glucose-1-D and D-glucose-2-D as
in carbohydrate aqueous solution.
substrates for LC-MS analysis (Scheme S2B and C, ESI†). No
Detection of substances with larger m/z values (larger than D-arabinose-1-D was detected in the reaction of D-glucose-1-D,
203, which is assigned to D-glucose (C6H12O6Na+)) in LC-MS and no D-arabinose was detected in the reaction of D-glucose-2-D.
confirms the existence of aldol condensation (Fig. S8, ESI†). The kinetic isotope effect (KIE) is not obvious at H2 of glucose
C8H14O8 can possibly be generated via aldol condensation (kH2/kD2 = 1.11), suggesting a low level of C2–H bond cleavage
between C6H10O6 (detected as C6H10O6Na+, m/z: 201) and (Fig. 9). All these results indicate that no C2–H bond cleavage
glycolaldehyde (C2H4O2). The reactions of three sugars (glycol- occurs in D-arabinose formation.
aldehyde, D-glyceraldehyde and D-erythrose) and their mixtures
The exchange between C2 and any other C atom of the
were carried out respectively in H2SO4 solution in the absence D-glucose bond cannot occur likewise, as already evidenced by
of NaVO3 and O2 (replaced by N2 with the same initial pressure). 13C NMR analysis for the experiments with D-glucose-1-13C in
The 13C NMR spectrum for each shows similar signals at Section 3.1. A C1–C2 shift can occur in D-glucose epimerization
This journal is ©the Owner Societies 2018
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
D-glucose (see below). Actually, the fact that O2 was not
involved in FA formation can be precisely evidenced by the
experiment with D-glucose-1-13C in the absence of O2. The
13C NMR spectrum clearly shows no difference with that of the
experiment with O2 (FA–13C was still detected in the spectrum,
as shown in Fig. S10, ESI†). In other words, the effect of O2 can
only participate in the oxidation of reduced vanadium(IV) to
vanadium(V).
Our previous work has confirmed that the coexistence of
NaVO3 and H2SO4 enhances the selectivity of FA generation
from glucose by forming a high-reducibility VO2+ species under
acidic aqueous solution.30 The vanadium(V) species is reduced
by the substrate to vanadium(IV) and then is re-oxidized by O2.
Similar mechanisms were proposed in H5PV2Mo2O40 and
VOSO4 aqueous systems.19,23,29 NaVO3 in the acidic aqueous
Fig. 9 The comparison of the dependence of conversion of D-glucose,
D-glucose-1-D and D-glucose-2-D on the reaction time. Reaction condi-
tions: D-glucose/D-glucose-1-D/D-glucose-2-D, 0.005 g; NaVO3, 0.35 wt%;
H2SO4, 1 wt%; H2O, 0.5 cm3; initial O2, 3 MPa; temperature, 100 1C.
+
solution (pH o 1) predominantly exists as a VO2 species at
room temperature.49 This species is usually hydrolyzed in water
as its hydrated form, VO2(H2O)n+, which lacks hydroxyl groups
to D-mannose at a very low level only with an excess amount of bonded to the vanadium atom.50 At a higher temperature (near
+
NaVO3–H2SO4, and no C–C bond cleavage was found after the the reaction temperature), the VO2 species is probably in
shift (as discussed in Section 3.2).
the form of an analogue of phosphoric acid, VO(OH)3, which
After C1–C2 bond cleavage, FA was formed simultaneously has three hydroxyl groups bonded to one vanadium atom.51
with D-arabinose. At the initial step of the reaction of D-glucose- It is easily speculated that one possible interaction between
1-13C, the obvious split signal of H1 in the 13C NMR spectrum is the catalyst and D-glucose is initialized via the esterification
due to the spin-coupling of the adjacent 13C atom (as discussed of VO(OH)3 and D-glucose by dehydration from both hydroxyl
in Section 3.1). This shows that the C1–H bond is reserved in groups.
FA after C1–C2 bond cleavage. The KIE of H1 is also not
remarkable enough (kH1/kD1 = 1.30) to confirm the existence hardly provide accurate information about the complex of a
of C1–H bond cleavage (Fig. 9). vanadium species and D-glucose. The undetectability of the
Various techniques including 13C/51V NMR and LC-MS can
FA generation from D-glucose requires an extra oxygen complex is probably due to the readily-hydrolyzed characteristics
atom. We conducted the oxidation in 18O-labeled water of the vanadate in aqueous solution. However, the conclusion that
(H218O) to seek the source of oxygen in FA. Two oxygen atoms C1–C2 bond cleavage occurs at the initial stage of the reaction
were inserted into molecular FA (m/z: 50, CH218O2, found by (in Section 1) strongly suggests that the interaction site is around
GC-MS, Fig. 10). No unlabeled FA (m/z: 46, CH2O2) was C1 and C2 of D-glucose.
detected. The result suggests that the oxygen atom in FA
The oxidation of D-glucose to FA is a two-electron reaction,
was potentially derived from water, probably via hydrated which indicates that two equiv. of vanadium atoms are reduced
from V to IV when D-glucose is oxidized to 1 equiv. of FA. Thus,
we further carried out kinetic studies of glucose oxidation
with the NaVO3–H2SO4 system. In accordance with some
kinetic studies of vanadium(V)–carbohydrate systems (that is,
with no O2),52,53 the rate of glucose oxidation exhibits a first-
order dependence on both NaVO3 concentration and D-glucose
concentration (Fig. 11, detailed information is shown in
Fig. S11 and S12, ESI†). These data are consistent with a
mechanism in which D-glucose is first reacted with 1 equiv. of
vanadium(V) species and then the intermediate is further
reacted with another equiv. of vanadium(V) species to 1 equiv.
of FA and 2 equiv. of vanadium(IV) species.
Based on the above analysis, a mechanism of C1–C2 bond
cleavage of D-glucose and oxidation to FA by O2 under NaVO3–
H2SO4 aqueous solution was proposed (Scheme 6). D-Glucose
is first reversibly transformed to its hydrated form with two
hydroxyl groups bonded to C1. Esterification then occurs
Fig. 10 Mass spectrum of FA extracted by ethyl ether after reaction under
between VO(OH)3 and two hydroxyl groups in C1 and C2 of
D-glucose to form a five-membered ring. Electron transfer and
C1–C2 bond weakening consequently proceed. Afterwards, C1–C2
the NaVO3–H2SO4 system in H218O. Reaction conditions: D-glucose,
0.003 g; NaVO3, 0.35 wt%; H2SO4, 1 wt%; H218O, 0.2 cm3; initial O2,
3 MPa; temperature, 100 1C, reaction time, 1 h.
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2018

View Article Online
PCCP
Paper
Fig. 11 Dependence of the initial rate of D-glucose conversion on (A) NaVO3 concentration and (B) D-glucose concentration.
4. Conclusions
In this work, we studied the pathway from D-glucose to FA and
the mechanism in NaVO3–H2SO4 aqueous solution using
isotope-labeled glucoses as substrates. D-Glucose is first trans-
formed to FA and D-arabinose via C1–C2 bond cleavage.
D-Arabinose undergoes a similar C1–C2 bond cleavage to form
FA and the corresponding one-carbon-less D-erythrose, which
can be further degraded by C1–C2 bond cleavage. The epimeri-
zation from D-glucose to D-mannose can slightly occur via a
C1–C2 shift. Dimerization and aldol condensation between
carbohydrates (either the starting material or the intermediates)
can also proceed to make the reaction a much more complicated
mixture. However, the fundamental reaction, C1–C2 bond cleav-
age, can drive all the intermediates to form the common product
FA. Based on the detected intermediates, isotope-labelling
experiments, KIE study and kinetic analysis, this mechanism is
proposed. D-Glucose is reacted with a vanadium(V) species
(VO(OH)3) to form a complex with a five-membered ring. Then,
electron transfer occurs and the C1–C2 bond weakens, followed by
C1–C2 bond cleavage (with no C–H bond cleavage) to generate the
H3COOꢀ–vanadium(IV) complex and arabinose. The complex is
then hydrolyzed to liberate the vanadium(IV) species and free
radical H3COOꢀ, which is rapidly oxidized by another VO(OH)3
molecule to generate FA and vanadium(IV) species. The reduced
vanadium species is oxidized by O2 to recover to its oxidation
state.
Conflicts of interest
Scheme 6 The proposed mechanism of C1–C2 bond cleavage of D-
glucose oxidation under NaVO3–H2SO4 aqueous solution.
There are no conflicts of interest to declare.
bond cleavage occurs, without C1/C2–H bond cleavage or a carbon
skeleton rearrangement, to generate the H3COOꢀ–vanadium(IV)
Acknowledgements
complex and arabinose. The complex is then hydrolyzed to The authors thank Professors Zhenyu Liu and Qingya Liu for
liberate the vanadium(IV) species and free radical H3COOꢀ, which their help. This work was financially supported by the National
is rapidly oxidized by another VO(OH)3 molecule to generate FA Natural Science Foundation of China (21706007, 21676019 and
and vanadium(IV) species. The two vanadium(IV) species are then 51372012), the Fundamental Research Funds for the Central
oxidized by O2 to vanadium(V) species. The one-carbon-less Universities (ZY1601), and the long-term subsidy mechanism
D-arabinose and the further formed carbohydrates are degraded from the Ministry of Finance and the Ministry of Education of
following a similar mechanism.
PRC (BUCT).
This journal is ©the Owner Societies 2018
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
29 Z. Tang, W. Deng, Y. Wang, E. Zhu, X. Wan, Q. Zhang and
Y. Wang, ChemSusChem, 2014, 7, 1557–1567.
30 W. H. Wang, M. G. Niu, Y. C. Hou, W. Z. Wu, Z. Y. Liu,
Q. Y. Liu, S. H. Ren and K. N. Marsh, Green Chem., 2014, 16,
2614–2618.
31 M. G. Niu, Y. C. Hou, S. H. Ren, W. H. Wang, Q. T. Zheng
and W. Z. Wu, Green Chem., 2015, 17, 335–342.
32 M. G. Niu, Y. C. Hou, S. H. Ren, W. Z. Wu and K. N. Marsh,
Green Chem., 2015, 17, 453–459.
References
1 A. Corma, S. Iborra and A. Velty, Chem. Rev., 2007, 107,
2411–2502.
2 P. Gallezot, Chem. Soc. Rev., 2012, 41, 1538–1558.
3 M. l. Besson, P. Gallezot and C. Pinel, Chem. Rev., 2014, 114,
1827–1870.
4 M. S. Holm, S. Saravanamurugan and E. Taarning, Science,
2010, 328, 602–605.
5 A. Wang and T. Zhang, Acc. Chem. Res., 2013, 46, 1377–1386.
6 Y. Liu, C. Luo and H. Liu, Angew. Chem., Int. Ed., 2012, 51,
3249–3253.
7 G. Zhao, M. Zheng, J. Zhang, A. Wang and T. Zhang, Ind.
Eng. Chem. Res., 2013, 52, 9566–9572.
8 X. Y. Yan, F. M. Jin, K. Tohji, T. Moriya and H. Enomoto,
J. Mater. Sci., 2007, 42, 9995–9999.
9 F. Jin, Z. Zhou, T. Moriya, H. Kishida, H. Higashijima and
H. Enomoto, Environ. Sci. Technol., 2005, 39, 1893–1902.
10 F. M. Jin and H. Enomoto, Energy Environ. Sci., 2011, 4,
382–397.
11 F. M. Jin, J. Yun, G. M. Li, A. Kishita, K. Tohji and
H. Enomoto, Green Chem., 2008, 10, 612–615.
12 S. Caratzoulas, M. E. Davis, R. J. Gorte, R. Gounder,
R. F. Lobo, V. Nikolakis, S. I. Sandler, M. A. Snyder,
M. Tsapatsis and D. G. Vlachos, J. Phys. Chem. C, 2014,
118, 22815–22833.
13 J. Rocek and F. H. Westheimer, J. Am. Chem. Soc., 1962, 84,
2241–2246.
14 L. Malaprade, Bull. Soc. Chim. Fr., 1928, 43, 683–690.
15 J. M. Bobbitt, Adv. Carbohydr. Chem., 1956, 11, 1–41.
16 G. Ohloff and W. Giersch, Angew. Chem., Int. Ed., 1973, 12,
401–402.
33 A. Boddien, D. Mellmann, F. Gartner, R. Jackstell, H. Jungle,
P. J. Dyson, G. Laurenczy, R. Ludwig and M. Beller, Science,
2011, 333, 1733–1736.
34 Y. Himeda, Green Chem., 2009, 11, 2018–2022.
35 Z. L. Wang, J. M. Yan, Y. Ping, H. L. Wang, W. T. Zheng and
Q. Jiang, Angew. Chem., Int. Ed., 2013, 52, 4406–4409.
36 T. Zell, B. Butschke, Y. Ben-David and D. Milstein, Chem. –
Eur. J., 2013, 19, 8068–8072.
37 D. Mellmann, P. Sponholz, H. Junge and M. Beller, Chem.
Soc. Rev., 2016, 45, 3954–3988.
38 C. Rice, S. Ha, R. I. Masel, P. Waszczuk, A. Wieckowski and
T. Barnard, J. Power Sources, 2002, 111, 83–89.
39 M. Weber, J. T. Wang, S. Wasmus and R. F. Savinell,
J. Electrochem. Soc., 1996, 143, 158–160.
40 N. V. Rees and R. G. Compton, J. Solid State Electrochem.,
2011, 15, 2095–2100.
41 Y. Nakagawa, D. Sekine, N. Obara, M. Tamura and
K. Tomishige, ChemCatChem, 2017, 9, 3412–3419.
42 A. M. Khenkin and R. Neumann, J. Am. Chem. Soc., 2008,
130, 14474–14476.
43 M. Sasaki, K. Goto, K. Tajima, T. Adschiric and K. Araic,
Green Chem., 2002, 4, 285–287.
44 J. Zhang, B. Hou, A. Wang, Z. Li, H. Wang and T. Zhang,
AIChE J., 2014, 60, 3804–3813.
17 E. Baer, J. M. Grosheintz and H. O. L. Fischer, J. Am. Chem.
Soc., 1939, 61, 2607–2609.
18 E. Amadio, R. D. Lorenzo, C. Zonta and G. Licini, Coord.
Chem. Rev., 2015, 301–302, 147–162.
45 Similar C1–C2 bond cleavage of D-glucose can be realized
previously, but C1 comes to CO2 via sequential oxida-
tion reactions to yield a-keto-D-gluconic acid followed by
decarboxylation: J. A. Stapley and J. N. BeMiller, The Ruff
degradation: a review of previously proposed mechanisms
with evidence that the reaction proceeds by a Hofer–Moest-
type reaction, Carbohydr. Res., 2007, 342, 407–418.
46 I. Delidovich and R. Palkovits, ChemSusChem, 2016, 9,
547–561.
¨
¨
19 R. Wolfel, N. Taccardi, A. Bosmann and P. Wasserscheid,
Green Chem., 2011, 13, 2759–2763.
¨
20 J. Albert, R. Wolfel, A. Bosmann and P. Wasserscheid,
Energy Environ. Sci., 2012, 5, 7956–7962.
21 J. Reichert, B. Brunner, A. Jess, P. Wasserscheid and
J. Albert, Energy Environ. Sci., 2015, 8, 2985–2990.
22 J. Albert and P. Wasserscheid, Green Chem., 2015, 17,
5164–5171.
47 M. L. Hayes, N. J. Pennings, A. S. Serianni and R. Barker,
J. Am. Chem. Soc., 1982, 104, 6764–6769.
23 J. Li, D. J. Ding, L. Deng, Q. X. Guo and Y. Fu, ChemSusChem,
2012, 5, 1313–1318.
48 W. Zhang and A. S. Serianni, J. Am. Chem. Soc., 2012, 134,
11511–11524.
¨
24 J. Albert, D. Lu¨ders, A. Bosmann, D. M. Guldi and
49 D. C. Crans, J. J. Smee, E. Gaidamauskas and L. Yang, Chem.
Rev., 2004, 104, 849–902.
50 O. Bortolini and V. Conte, J. Inorg. Biochem., 2005, 99,
1549–1557.
P. Wasserscheid, Green Chem., 2014, 16, 226–237.
25 T. Lu, M. G. Niu, Y. C. Hou, W. Z. Wu, S. H. Ren and F. Yang,
Green Chem., 2016, 18, 4725–4732.
26 B. B. Sarma and R. Neumann, Nat. Commun., 2014, 5, 4621.
27 J. Z. Zhang, M. Sun, X. Liu and Y. Han, Catal. Today, 2014,
233, 77–82.
51 A. Sadoc, S. Messaoudi, E. Furet, R. Gautier, E. L. Fur,
`
L. l. Polles and J.-Y. Pivan, Inorg. Chem., 2007, 46, 4835–4843.
52 K. K. S. Gupta and S. Basu, Carbohydr. Res., 1979, 72, 139–149.
28 J. L. Xu, H. Y. Zhang, Y. F. Zhao, Z. Z. Yang, B. Yu, H. J. Xu 53 S. C. Pati and M. Panda, Int. J. Chem. Kinet., 1979, 11,
and Z. M. Liu, Green Chem., 2014, 16, 4931–4935.
731–740.
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2018