Sustainable Synthesis of Oxalic and Succinic Acid through Aerobic Oxidation of C6 Polyols Under Mild Conditions

Article abstract of DOI:10.1002/cssc.201702347

The sustainable chemical industry encompasses a shift from the use of fossil carbon to renewable carbon. The synthesis of chemicals from nonedible biomass (cellulosic or oil) represents one of the key steps for “greening” the chemical industry. In this paper, we report the aerobic oxidative cleavage of C6 polyols (5-HMF, glucose, fructose and sucrose) to oxalic acid (OA) and succinic acid (SA) in water under mild conditions using M@CNT and M@NCNT (M=Fe, V; CNT=carbon nanotubes; NCNT=N-doped CNT), which, under suitable conditions, were recoverable and reusable without any loss of efficiency. The influence of the temperature, O₂ pressure (P (Formula presented.)), reaction time and stirring rate are discussed and the best reaction conditions are determined for an almost complete conversion of the starting material and a good OA yield of 48 %. SA and formic acid were the only co-products. The former could be further converted into OA by oxidation in the presence of formic acid, resulting in an overall OA yield of >62 %. This process was clean and did not produce organic waste nor gas emissions.

Full text of DOI:10.1002/cssc.201702347

Accepted Article
Title: Sustainable synthesis of oxalic (and succinic) acid via aerobic
oxidation of C6 polyols by using M@CNT/NCNT (M=Fe, V) based
catalysts in mild conditions
Authors: Angela Dibenedetto, Maria Ventura, David Williamson,
Francesco Lobefaro, Matthew D. Jones, Davide Mattia,
Francesco Nocito, and Michele Aresta
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemSusChem 10.1002/cssc.201702347
Link to VoR: http://dx.doi.org/10.1002/cssc.201702347
A Journal of
www.chemsuschem.org

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
Sustainable synthesis of oxalic (and succinic) acid via aerobic
oxidation of C6 polyols by using M@CNT/NCNT (M=Fe, V) based
catalysts in mild conditions
[
a]
[c]
[d]
[e]
[c]*
Maria Ventura, David Williamson, Francesco Lobefaro, Matthew D. Jones, Davide Mattia,
[
b]
[d]
[a,b]*
Francesco Nocito, Michele Aresta, and Angela Dibenedetto
3
Abstract: Sustainable chemical industry encompasses the shift from
fossil carbon to renewable carbon. The synthesis of chemicals from
non-edible biomass (cellulosic or oily) represents one of the key steps
for “greening” the chemical industry. In this paper we report the
aerobic oxidation-cleavage of C6 polyols (5-HMF, glucose, fructose,
sucrose) in water to oxalic acid-OA (and succinic acid-SA) under mild
conditions using M@CNT/NCNT (M=Fe, V; CNT=carbon nanotubes;
NCNT= N-doped CNT), which, under suitable conditions, are
recoverable and reusable without any loss of efficiency. The influence
of temperature, PO2, reaction time, stirring rate are discussed and the
best reaction conditions are emphasized for an almost complete
conversion of the starting material, with a good yield of OA equal to
and new bio-based materials such as bioplastics. Recently, even
CO has been taken into serious consideration as building block
2
for chemicals or source of carbon for fuels.2b Among renewable
feedstock, ligno-cellulosic biomass has attracted considerable
attention due to its potential as a source of a wide range of
4
platform chemicals such as C6-polyols, 5-hydroxymethylfurfural
5
6
7
(5-HMF), levulinic acid (LA) or formic acid (FA). Among such
chemicals, 5-HMF, synthesized by dehydration of fructose, or
even directly from glucose,8 is considered one of the most
important intermediates. Its furan structure bearing an aldehyde
and a hydroxyl moiety can be transformed into numerous high value
chemicals capable of replacing their analogues synthesized from
fossil fuels (Scheme 1).
48%. SA and formic acid are the only co-products. The former can be
further converted into OA by oxidation in presence of formic acid
allowing to reach an overall yield of OA >62%. This process is clean
and does not produce organic waste nor gas emissions.
The oxidation-cleavage of glucose (or fructose), has been
considered as a candidate for the biomass-based production of
monocarboxylic and dicarboxylic acids, a technology that has not
yet reached a “sustainable” level today. Recently, efforts have
9
been made to convert C6 into lactic acid, succinic acid or oxalic
acid.
Introduction
Oxalic acid (OA) and its derivatives have widespread application
in industrial sectors such as textiles, tanning, oil refining, catalyst
preparation, pharmaceuticals, dyes, explosives, straw bleaching,
printing, marble polishing, and metal and cloth cleaning.10 It is also
The petrochemical industry has contributed to worldwide
economic development for the past 150 years or so, but now
serious environmental problems are associated with it. The need
to establish environmentally friendly chemical processes is an
imperative and requires the development of novel and cost-
a very important chemical in petroleum refinery, rare-earth and
inks manufacturing, rust and corrosion prevention, and dental
11
adhesive processing.
1
effective methods for pollution prevention. In recent decades, the
Currently, the methods for producing OA are classified according
to the starting raw material into six groups: (i) fusion of sawdust
with caustic soda, (ii) oxidation of olefins and glycols, (iii) radiation
processing of carbonate solutions and molasses, (iv) fermentation of
carbohydrates, (v) oxidation of carbohydrates with nitric and sulphuric
acid, and (vi) formates decomposition. The last three are the most
used technologies. Fermentation routes to OA have been studied
with renewed interest in the last decade, yielding mainly citric acid
and/or OA from sucrose or lactose, depending on the
substitution of non-renewable fossil resources such as crude oil,
coal, and natural gas with renewable-carbon such as biomass,
including lignocellulose and vegetal oils, as a sustainable
feedstock has been extensively investigated for the manufacture
of biofuels, commodity chemicals,2a high added-value products
[
[
[
a]
b]
c]
Prof A. Dibenedetto, Dr M. Ventura
CIRCC, Via Celso Ulpiani, 27, 70126 Bari, Italy
E-mail: angela.dibenedetto@uniba.it
Prof A. Dibenedetto, Dr F. Nocito
Department of Chemistry, University of Bari, Campus Universitario,
fermentation conditions. OA production at the level of >200 mM
has been achieved under strict control of pH.12 The oxidation of
C6 with nitric acid is currently the most widely used industrial
70126 Bari, Italy
Prof D. Mattia, Dr D. Williamson
process to OA. A strong acid solution composed of HNO
3
and
Centre for Advanced Separations Engineering and Department of
Chemical Engineering, University of Bath, Claverton Down, Bath
BA2 7AY, United Kingdom
H
SO together with V as the catalyst is used to achieve a
2 5
O
2
4
conversion of 99% of the substrate with yields of OA from 2.9%
to 54%, producing large volumes of waste.13 Coupling of formates
is an attractive route11,14 which, nevertheless, still uses strong
acids and produces waste with high environmental burden. From
an environmental point of view, none of the above routes follows
the 12 principles of green chemistry.
E-mail: D.Mattia@bath.ac.uk
[
[
d]
e]
Prof. M. Aresta, BSc F. Lobefaro
2
IC R, Viale Einaudi 25, 70125 Bari, Italy
Prof. M. D. Jones
Department of Chemistry, University of Bath, Bath BA27AY, UK.
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
(SA)
(
(
FA)
OA)
Scheme 1. Representative processes for cellulosic biomass conversion into chemicals
We have recently developed environmentally friendly
Results and Discussion
technologies based on the use of water as solvent and oxygen or
even air as oxidant for the conversion of 5-HMF into DFF, FFCA,
HMFCA, FDCA,2,15 targeting high conversion and selectivity.
Aerobic oxidation of 5-HMF with M@CNT/NCNT
(
Scheme 1, left side) Earth abundant metal catalysts in the form of
The oxidation of 5-HMF was investigated using the three catalysts,
Fe@CNT, Fe@NCNT and V-Fe@CNT in water using O as
2
mixed oxides were used with selective oxidation of the ring pending
functionalities without touching the integrity of the furan ring. We have
now tested in such oxidation metals supported on carbon nanotubes
oxidant. The oxidation of 5-HMF can occur following two
alternative routes according to the temperature (Scheme 1):
either the furan ring substituents are oxidized (left part) at 383 K,
or oxalic acid and succinic acid are formed (right part) at >403 K
via oxidation-cleavage of the furan ring. No organic solvents were
used and the product distribution could be governed by controlling
the reaction conditions (vide infra), minimizing the waste
generated in the process.
(M@CNT) and have found a completely different behavior with
respect to mixed oxides in almost the same operative conditions (vide
infra). CNTs are a new kind of carbon materials that offer interesting
possibilities as support for metal particles, due to the sp carbon
2
structure, the excellent electron transport performance and the
electronic interaction of active nanoparticles with the CNTs walls.16 In
the field of biomass transformation, it has been proved that M@CNT
are active in a wide range of conversions. It has been reported that
Au-Pdnanoparticles deposited on CNTs show higher catalytic activity
than on alumina or silica for the base-free aerobic oxidation of 5-HMF
towards FDCA.17 Therefore, CNTs can be considered promising
catalyst supports for the selective oxidation of 5-HMF.
Accordingly, we present in this paper the use of M@CNTs (M=Fe,
V),18 or M@NCNTs (where NCNT is a nitrogen-doped CNT) in the
aerobic oxidation of either 5-HMF, or fructose or glucose or even
sucrose in water. We show that the oxidation cleavage of the C6
polyol takes place affording interesting yields of oxalic acid, and
succinic acid as co-product. To the best of our knowledge this is
the first report on a sustainable synthesis of OA from a C6 polyol
with a best potential yield of ca. 60%, working under quite mild
conditions.
An aspect of the reaction that has attracted our attention was the
random conversion of 5-HMF into fructose: the reverse of the
formation of 5-HMF, never observed in our previous works in
which mixed oxides were used as catalysts2
a,15
of oxidation of 5-
HMF. The results of the catalytic runs are listed in Table 1, where
the activity of Fe@CNT, V-Fe@CNT and Fe@NCNT are
compared in different conditions of temperature and stirring rate.
The last parameter resulted to be quite important in driving the
conversion and selectivity, as we discuss below.
However, according to our habit,2a,15 after the first catalytic tests,
we have recovered the catalyst and verified whether any
modification of its structure and/or composition had occurred. The
firstcheckwedidwasonthestructureofthecatalyst(Entry1-2inTable
1
(
) used at 1000 rpm: leaching of Fe from CNTs was observed by EDX
Energy Dispersive X-ray Spectroscopy) analysis and the breaking of
CNTs was detected by TEM (Figure 1a and 1b).
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
Table 1. Oxidation-cleavage of 5-HMF with Fe@CNT (A), Fe@NCNT (B) and V,Fe@CNT (C) as catalysts
Cat
A
rpm
Time (h)
3
T (K)
383
Conv. (%)
23.7
% DFF
FFCA
89.6
FDCA
0
FA
5.6
OA
0
SA
0
Fructose
0
1
2
3
4
5
6
7
8
9
1
1000
4.41
A
A
A
A
B
B
C
C
C
1000
500
250
250
250
250
250
250
250
3
403
403
403
413
413
413
413
413
413
99
0.39
4.6
3.2
0
18.8
2.35
4.8
0
22.2
8.0
0
8.5
22.0
30.6
37.8
48.4
39.8
39.9
19.9
6.3
8.6
29.0
21.3
17.7
9.2
21.8
4.1
0
3
99
15.3
18.5
19.4
16.5
27.0
10.8
25.1
85.3
16.3
17.6
7.8
3
99
1.5
1.5
3
99
0
47.8
90.2
62.1
99
11.7
4.9
3.3
0
0
0
10.0
16.2
1.4
0
0
1.5
3
61.8
65.1
<1
0
0
3.5
0
0
6
99
0
0
14.0
0
0
Reaction Conditions: [5-HMF]= 0.16 M, 0.025 g of catalyst; Acronyms are reported in Scheme 1 and in the text.
2
a
b
c
as sp hybridized carbon, such as nitrogen in Fe@NCNTs, which
causes an increase in D peak intensity due to the formation of
lattice vacancies. The G peak at ca.1581 cm-1 is often attributed
to graphitic structure and order in a carbon nanotube sample,
while the G’ peak at ca. 2693 cm-1 is an indication of long-range
order. Thus, the ratio of the D and G peaks (I
an indication of the overall order and integrity of a carbon
nanotube sample. Here, the increase in the I /I ratio observed
D G
/I ) can be used as
5
0 nm
50 nm
5
00 nm
D
G
Figure 1. TEM micrographs of catalyst “A” after a run (Fig. 1a and 1b, Entry 1
and 2, Table 1 at 1000 rpm), indicating tube degradation after reaction, with
metal nanoparticles partially removed from the tube wall and the tube walls
themselves snapped or deteriorated; TEM of “A” after a catalytic run at 250 rpm
(Table 2) indicates that damage occurs to the CNT structure
during reaction. An increase of temperature at 1000 rpm
(
compare Entry 1 and 2 in Table 1) results in a further deeper
damage to the CNT structure.
(Entries 4 and 5) shows little visible damage to the CNT structure (Fig. 1c).
As such, we have reduced the stirring speed and modified the
geometry of the stirring bar (from a linear bar to a round bottom
magnet with a topping star for better mixing at lower speed) to
improve the stability of the catalyst (using a bar at 250 rpm still
causes damage to the catalyst).
Raman spectroscopy also confirmed the deterioration of the
catalyst at 1000 rpm (Figure 2). The D peak at ca. 1367 cm-1 is
commonly attributed to defects or disorder in the carbon nanotube
lattice that may occur through damage to the tube structure or
incorporation of dopants that do not fit into the lattice as perfectly
When the rpm was decreased from 1000 to 500 with the new
stirrer, the selectivity towards OA increased to 30.6 %, Entry 3.
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
increase of FA, Entry 3 (C6 species such as FDCA and fructose
were also present).
D
G
G'
Table 2. The calculated ratios of the D and G peaks of Raman spectra
D G
detailed in Figure 2. An increase in I /I indicates increased disorder in the
sample, caused by defects or damage in the CNT support structure
Sample
D G
I /I
0
500
1000
Wave Numbers (cm )
Fe@CNTs (Calcined) MA 110
1500
2000
2500
3000
3500
-
1
Fe@CNT Calcined
Entry 1
0.19
0.59
0.82
0.65
MA 130
MA3
Entry 2
Figure 2. Raman spectra of Fe@CNT catalysts pre- and post-reaction under
Entry 3-4
-
1
varying conditions. The X-axis unit is cm . MA110=Entry 1; MA130= Entry 2;
MA3= Entry 3, 4, Table 1.
The effect of the oxygen pressure was also investigated, and the
results are shown in Table 3. Noteworthy, either increasing the
pressure from 10 to 20 bar or decreasing it to 5 bar has a negative
effect on the selectivity of OA, which anyway was decreased. The
lower pressure caused incomplete cleavage, and decrease of the
OA selectivity Entry 1 - Table 3 (fructose was also present in the
reaction mixture and other C6 species).
Decreasing even more the rpm, the selectivity towards OA further
increased to 37.8-48.4%, while the stability of the catalyst
increased as shown by TEM (Fig. 1c) and EDX (no leaching of
Fe). Therefore, the stirring speed and stirrer geometry influence
the stability of the catalyst and, consequently, the selectivity of the
process. As a matter of fact at 250 rpm the catalysts was not
deteriorated and could be recovered and recycled several times
without reducing the conversion and selectivity (See Experimental
Section, Catalytic tests).
Table 3. Oxidative-cleavage of 5-HMF with Fe@CNT at different oxygen
pressures
To investigate the role of the temperature with 250 rpm as optimal
rotational speed, an experiment was performed at 413 K, Entry 5,
Table 1. The selectivity towards OA increased from 37.8% to
Selectivity (%)
Rpm
Time
h)
PO
Bar
2
Conv. DFF FFCA FA
(%)
LA
OA
48.4 % while the selectivity towards FA remained almost constant
(
(
18.5% vs 19.4%). Further increasing the temperature above 413
K (data not shown) strongly decreases the OA in favor of FA. We
then compared the activity of Fe@CNT to that of Fe@NCNT
1
2
250
250
6
5
65.2
91
0
0
41.7
5.3
4.1
25.1
38.4
1.5
10
8.0 4.1 19.8
(
Entries 6 and 7) and V-Fe@CNT (Entries 8-10), Table 1.
3
250
1.5
20
99
1.6
0
40.6
0
35.9
Comparing Fe@CNT (Entry 5) with Fe@NCNT (Entry 6), one can
see that there is a decrease by 9 points in the selectivity towards
OA in the same reaction conditions, but the global conversion of
Reaction conditions: [5-HMF]I = 0.16 M; 0.025 g of catalyst; T = 403 K.
5
-HMF is much lower when Fe@NCNT is used. This can be a
consequence of the increased value of the ratio strong acid
sites/strong basic sites, n /n , from 3.6 for Fe@CNT to 4.6 for
Fe@NCNT. Noteworthy, in Entries 2-6 a significant rehydration of
-HMF to fructose is observed. Considerably, in the best
The optimal pressure is thus 10 bar that minimizes the formation
of FA (Entry 2). Most likely at such pressure the optimal [O ] is
built-up in solution, for a controlled oxidation cleavage to 3 C . We
a
b
2
5
2
have carried out a study to gather information on the sequence of
formation of the products trying to propose a plausible reaction
pathway and understand whether OA is originated from SA and
then degrades to FA as the reaction time increases. The study
was performed using Fe@CNT at two temperatures: 403 K, and
operative conditions (413 K, 250 rpm, 1.5 h) comparison of
Entries 5 and 6 in Table 1 shows that modifying the nature of the
support (from CNT to NCNT) keeping constant the catalyst (Fe),
the selectivity towards OA is lowered by 9 points, while the
conversion is halved. Conversely, changing the catalyst from Fe
413 K. It is worth noting that the conversion of 5-HMF was almost
(
Entry 5) to V-Fe (Entry 8) the product distribution is strongly
quantitative in both reactions. However, the selectivity towards
OA reaches its maximum value within 3 h of reaction at 403 K and
influenced. FFCA is formed at short reaction times (1.5 h), while
FA becomes the prevalent product for long reaction times (Entries
1.5 h at 413 K. For longer times (>6 h), OA degrades to FA that
8-10).
increases up to ca. 30%. Obviously, the conversion of 5-HMF is
lower at shorter reaction times: from 99% after 3 h to 9.5 % after
It is clear the influence of V in reducing the formation of C4 and
C2 species. Noteworthy, V
in processes on stream.
2 5 3 2 4
O is the catalyst used in HNO -H SO
30 min. V, Fe@CNT is less active than Fe@CNT (Entries 5, 6
Table 1) at 413 K, 1.5 h. For longer reaction time an increase of
FA is observed (Entry 7). In conclusion, the best conditions for
On the other hand, when 20 bar were used the conversion
increased with a slight decrease of selectivity towards OA and
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
maximizing the conversion of 5-HMF and the formation of OA are
can be converted into OA and there is a stepwise cleavage of the
starting C6-skeleton to C4 (SA) + C2 (OA) and further to C1 (FA)
1.5-3 h at 413 or 403 K. In both cases we have not found a clear
evidence that OA is derived from SA.
2
compounds. We have, thus, reacted SA with O with and without
When V-Fe@CNT was used, significant changes in the product
distribution were observed: after 3h, (Entry 9, Table 1) FFCA was
the major product of the reaction. When the reaction was
performed for longer reaction times, 6 h Entry 10, FA was
obtained as the major product (85.3%), making such reaction an
interesting route to FA from C6 polyols. Vanadium increases the
oxidation power of the catalyst, producing a reduction of the
oxidation temperature (Fig. 3e, Experimental Section). Therefore,
the oxidation-cleavage is easier and FA, that is the pre-ultimate
product in the oxidative-cleavage chain of the C6 unit that
FA and checked the formation of OA. Table 4 shows the results
when fructose or SA was used as substrate. The reactions were
performed using Fe@CNT as catalyst. In the reaction performed
with fructose at 413 K, in 1 h (Entry 1, Table 4) a low conversion
(1.2%) was reached and a mixture of SA+FA was obtained.
After 6h, the overall conversion of fructose was increased to
30.5%, OA being the major product and SA present at similar
concentration than FA (Entry 2, Table 4).
For longer reaction times, (12 h, Entry 3, Table 4) fructose was
completely converted and FA+OA were the major products.
Working at 413 K for 12 h with V-Fe @CNT, a complete
conversion was observed with a 22.1% selectivity towards OA
2
terminates with CO formation, is predominantly obtained.
(
Entry 4) and high formation of FA.
Using SA as starting material, OA is formed very selectively, but
the conversion of SA remains around a few percent even if the
reaction is left going for a long time (Entries 5-6). Conversely, if
FA is added to SA (1:5 w/w), Entry 7, SA is converted into OA at
a higher rate. This suggests that FA somehow assists the
cleavage of SA into OA. For longer reaction times than 12 h, OA
was degraded affording FA.
Aerobic oxidation of fructose and the role of SA
Two aspects of the results reported above attracted our attention:
the hydration of 5-HMF to fructose and the presence of succinic
acid. However, we decided to: i) use fructose as substrate instead
of 5-HMF and ii) check even in this case whether succinic acid
Table 4. Aerobic oxidation-cleavage reaction of fructose or SA with Fe@CNT (R = residual starting reagents).
Selectivity (%)
Cat
Time (h)
1
T (K)
413
Substrate
Fruct
PO
2
Bar
Conv. (%)
1.2
FA
OA
-
SA
1
2
3
4
5
6
7
Fe@CNT
10
21.8
79.3
Fe@CNT
Fe@CNT
Fe-V@CNT
Fe@CNT
Fe@CNT
Fe@CNT
6
413
413
413
413
413
413
Fruct
Fruct
Fruct
SA
10
10
10
10
10
10
30.5
99
27.3
31
78.6
0
46.5
46. 8
22.1
98
26.3
21
0
12
12
6
98
2.4
2.5
12.5
R
12
12
SA
0
98
R
SA+FA
R
99
R
Table 5. Aerobic oxidation-cleavage reaction of glucose as substrate with Fe@CNT/NCNT
Selectivity (%)
Catalyst
Time (h) T (K)
PO
20
20
20
20
20
20
2
bar
[Glucose]
0.2
Conv (%)
0
FFCA
FA
OA
SA
Fructose
1
2
3
4
5
6
Fe@CNT
24
6
403
403
403
423
403
423
-
-
-
-
-
Fe@NCNT
Fe@NCNT
Fe@NCNT
V-Fe@CNT
V-Fe@CNT
0.2
10.3
22.0
41.8
36.8
96.6
0
0
0
32.9
41.7
16.4
50.1
31.8
19.4
27.0
37.1
16.6
47.9
24.2
28.3
17.3
17.7
8.01
8.4
0
12
6
0.2
0.2
29.2
0
12
12
0.2
12.9
11.7
0.2
0
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
Aerobic oxidation of glucose (and sucrose) in water
HMF is the easiest substrate for producing OA, glucose the most
hard to convert. As a matter of fact, as Scheme 1 shows, glucose
must be first isomerized to fructose (basic catalysis) which is then
dehydrated to 5-HMF (acid catalysis), the best source of OA. As
shown in Table 6 Fe@CNT, V-Fe@CNT and Fe@NCNT have
As fructose is usually produced by base-catalyzed isomerization
of glucose, the catalysts M@CNT/NCNT were also tested using
glucose as substrate. Table 5 shows the results when the reaction
was carried out at 403-423 K. A higher temperature was used with
respect to 5-HMF and fructose, as preliminary tests did show that
at 403 and 413 K the reaction rate was quite low. When Fe@CNT
was used as catalyst, Entry 1, no reaction was observed even for
long reaction times at 403 K.
a b
different strong n /n indexes, Fe@NCNT being the most acid and
V-Fe@NCNT the most basic. It is not surprising, thus, that V-
Fe@CNT is the most effective catalyst for glucose oxidation to
OA, most likely because is the one that facilitates the
isomerisation to fructose and further oxidation cleavage.
However, when Fe@NCNT was used as catalyst at the same
temperature (Entry 2, Table 5) glucose was converted (10.3%)
The conversion of the various substrates and the yield of OA are
not too different from the best figures obtained with technologies
into a mixture composed by OA, FA and SA with selectivity of 32.9, on stream, but the reaction conditions are much milder and safer,
9.4 and 24.2%, respectively. This shows that the support plays
than the harsh conditions used (HNO -H SO -V ) in processes
1
3
2
4
2 5
O
a role in the reaction. For longer reaction times (Entry 3, Table 5)
the conversion was increased from 10.3 to 22 %, as increased
was the amount of FA, OA and SA. At 423 K, after 6 h (Entry 4,
Table 5), the conversion increased compared to Entry 2 along
with the selectivity towards OA. Using V-Fe@CNT, Entries 5-6,
after 12 h the conversion of glucose and the yield of OA increased
with the temperature, but significant amounts of FA and other
compounds were formed. Moreover, fructose, the product of
isomerization of glucose, is also randomly present in the reactions. It
is mainly formed when Fe@NCNT is used. Interestingly, at 423 K
after 12 h using V-Fe@CNT a significant conversion of glucose was
observed (96.6%) with an interesting selectivity towards OA equal to
on stream, with no air born emission and limited waste
represented by FA and SA that can be separated.
We have also shown that succinic acid can be converted into
oxalic acid in presence of formic acid. However, if SA is converted
into OA, a putative total yield of oxalic acid equal to ca. 55-60%
can be targeted starting from 5-HMF (that can be still optimized)
or nearly equal starting from fructose and glucose in the suitable
reaction conditions. The catalytic conversion of C6 presented in
this paper responds to the criteria of sustainability as it occurs in
2
mild conditions, using recoverable and reusable catalysts, O as
oxidant and water as solvent, without generating any air borne
emission: further investigation is going on in our laboratory aimed
at up-scaling.
47.9%. It is worth noting that V-Fe@CNT is more basic than
Fe@NCT (Table 6, Entries 1 and 3, column 6). This can favor the
conversion of glucose into fructose, more reactive than glucose.
The above data show that glucose is (directly or indirectly)
catalytically converted into OA with ca. 97% conversion and 48%
selectivity, under much milder conditions than using technologies
today on stream and with a zero air-emission and a reduced
formation of waste. We have also used sucrose as starting
material, and yields and selectivity were intermediate between
fructose and glucose. Moving from 5-HMF to glucose, the rate of
oxidation increases in the order: glucose<fructose<5-HMF. In any
case an almost quantitative conversion of the substrate was
Experimental Section
Materials and analytical methods
Oxides for the synthesis of the catalysts, and standard: 2,5-
furandicarboxaldehyde ≥99%, 5-formyl-2-furoic acid >99%,
furandicarboxylic acid >99%, D-glucose, D-fructose, oxalic acid, formic
acid, succinic acid and levulinic acid were purchased from Sigma Aldrich.
2,5-
5-(hydroxymethyl)-furfural was prepared as reported in ref. 5. All the
reaction products were analyzed by using a Jasco HPLC equipped with an
RI detector and a Phenomenex Rezex RHM Monosaccharide H+(8%)
observed with
a selectivity around 48%, but in different
temperature conditions and using different catalysts.
300x7.8mm at 343 K. A 0.005 N solution of sulphuric acid was used as the
mobile phase. The flow rate was between 0.5-0.9 mL/min. The
concentration of the substrate and of reaction products were determined
using an RI detector.
The surface characterization of the catalysts was carried out by using the
Pulse ChemiSorb 2750 Micromeritics instrument. Analyses of the
Conclusions
M@CNT/NCNT catalyzes the aerobic oxidation of C6 polyols (5-
HMF, fructose, glucose, and even the dimeric sucrose) with O in
2
water at moderate temperature (403-423 K). Although we started
using 5-HMF, being this study a continuation of previous
investigation on the selective oxidation of side chains of 5-HMF,
we moved to fructose from which 5-HMF originates. In the most
3 2
acid/basic sites were carried out using NH or CO , respectively, as probe-
gas using 100 mg of catalyst. The samples were pretreated under N
mL min ) flow at 673 K. The Pulse Chemisorb was performed with NH
2
(30
or
-1
3
CO gas using He as carrier gas (30 mL min-1).
2
Brunauer Emmett Teller (BET) surface area was determined using N
as carrier gas at 273 K followed by heating up to 923 K.
2
/He
Temperature Programmed Desorption (TPD) was performed under He
2
favorable conditions (413 K, 1.5 h and 10 atm of O ), using
-
1
flow at 30 mL min . H
/Ar as carrier gas (30 mL min ) at room temperature, the sample was
heated up to 1273.15 K. Elemental analysis was carried out by using EDX
Energy-dispersive X-ray spectroscopy).
2
-TPR analysis was carried out using a mixture of
Fe@CNT, a conversion of 5-HMF equal to 99% was observed
with 48.4 % yield of OA and 7.8 % SA. Starting from fructose a
conversion equal to 99 % was observed after 12 h, with 46.8 %
selectivity towards OA using Fe@CNT at 413 K (Entry 3, Table
-
1
H
2
(
4), while V-Fe@CNT produces mainly FA (Entry 4, Table 4).
Glucose requires higher temperature: at 423 K the conversion
was 96.6 % with a selectivity towards OA equal to 47.9 %, using
V-Fe@CNT. The data shown above indicate that different
substrates demand different catalysts and reaction conditions. 5-
Catalysts preparation
Fe@CNTs and Fe@NCNTs were prepared via a floating catalyst CVD
method in a quartz tube (25 mm ID, 28 mm OD x 122 cm Length), which
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
was placed into a tube furnace to control the reaction temperature. During
the reaction, a precursor solution (45 mL) was loaded into a 50 mL syringe,
which was then placed onto a syringe pump for automated injection once
the reactor had reached the desired reaction temperature.18 To synthesize
Fe@CNTs, the precursor solution consisted of 1.0 g ferrocene (FcH)
dissolved in 50 mL toluene (20 mg/mL). To synthesize Fe@NCNTs,
toluene was replaced with acetonitrile in the precursor solution to act as a
nitrogen source. During synthesis, the quartz tube was purged with 50 mL
Ar as the tube furnace ramped up to the desired reaction temperature of
Figure 3a displays the H
, Fig. 3c for Fe@CNT, Fig. 3d for Fe@NCNT and Fig. 3e for V-
Fe@CNT.
Generally, the reduction of pure Fe
by the following consecutive transitions: Fe
are identified in the profile shown in Fig. 3a, with = 370 °C= 619 °C=
07 °C.19,20 (Note: The equipment does not allow to express the
temperature as K, therefore we discuss the properties of the catalysts
using the Celsius scale of temperature). As displayed in Fig. 3b, H -TPR
profiles for the sample Fe@CNT show two reductive peaks centered at
2 2 3
consumption profile for Fe O , Fig. 3b that of
2 5
V O
O
2 3
to metallic iron can be expressed
> FeO > Fe19 that
> Fe
O
2 3
3 4
O
7
1
063 K at a rate of 5 °/min. When the reaction temperature was reached,
the desired precursor solution was injected at a rate of 10 mL/h for 4 hours
under a flow of 50 mL H and 400 mL Ar. The quartz tube was then left to
2
2
o
o
about 420 C and 800 C, respectively.
cool to room temperature before the fresh catalyst was collected as a black,
powdery material. This typically provided a yield of ca. 1.4 g catalyst per
synthesis.
2
To achieve 0.5 wt. % vanadium doping, 0.013 g VO(acac) was dissolved
in 15 mL methanol. 0.5 g fresh catalyst was then added to this solution to
produce a dark slurry, which was stirred in a beaker at room temperature
under a hood for 24 hours. After stirring, the methanol had typically
evaporated completely and the dry, doped catalyst could be easily
collected.
Previous studies have shown that, immediately after synthesis, the iron
nanoparticles embedded in the walls of the Fe@CNTs are obscured by a
graphitic carbon layer.18 This prevents them from participating as catalytic
sites. To expose these iron particles, the graphitic carbon layer must be
removed via thermal oxidation in air, without damaging the CNT support
structure. This was achieved by packing 0.45 g catalyst into a ¼” x 12.5
cm stainless steel tube, which was loosely packed with quartz wool at the
bottom to allow for air flow while preventing the catalyst from escaping.
For Fe@CNT based catalysts, the catalyst was heated in a muffle oven at
2 2 3
Figure 3a. H -TPR profile for Fe O . = 370 C, = 619 C, and = 707 C
843 K for 40 minutes, ramping up at a rate of 10 K/min. For Fe@NCNTs,
the catalyst was activated at 673 K for 1 hour, due to the decreased
stability of the NCNT support structure after nitrogen doping. This thermal
activation step was always carried out following the vanadium doping step.
Characterization of the catalysts
2
Figure 3b. H -TPR profile for Fe@CNT
Table 6 presents the BET surface area and acid-base properties of the
different metal-nanotube catalysts used in the production of OA. The total
and strong basic and acid sites were calculated from the area under the
peaks in the TPD experiments, and are expressed as the volume of CO
and NH up-taken and released, respectively. The ratio strong acid/basic
sites (n /n ) is also reported.
2
3
a
b
A comparison between Entries 1 and 2 in Table 6 reveals that when CNTs
were doped with nitrogen (NCNT), a non-equal drop in acid or basic sites
was observed resulting in an increase in the ratio na/nb. A different
behavior was observed when vanadium was included in the catalysts: a
huge increase in both acid and basic sites was observed, but with a
resulting lower na/nb ratio. Therefore, both doping CNT with N and adding
V as metal, change the properties of the catalyst. Concerning the BET
surface area, no major differences were observed for the three catalysts.
2
The redox behavior was investigated via H -TPR analysis.
Table 6. BET surface area and basicity/acidity strength of different Fe@CNT.
Solid
Total Strong Total
CO CO NH
ads ads ads
mL/g) (mL/g) (mL/g)
Strong
NH
ads
BET
Strong
na/nb
2
2
3
3
surface
area
2
(
(mL/g)
(m /g)
1
2
3
. Fe@CNT
0.55 0.55 2.25
1.987
1.847
3.613
3.6
4.6
2.5
78
81
81
. Fe@NCNT 0.40
. V-Fe@CNT 1.59
0.39
1.46
1.94
3.85
2
Figure 3c. H -TPR profile for Fe@NCNT and its magnified region from 0 to 500
⁰C
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
a
Figure 3d. H
2
-TPR profile for V
O
2 5
25 µm
b
5
0 nm
Figure 3e. H
2
-TPR profile for V-Fe@CNT
Figure 4. a) SEM micrograph of Fe@CNT; b) TEM micrograph of V-Fe@CNT.
Both are representative of other catalysts
The low temperature peak is likely due to the reduction of Fe3+ species to
metallic Fe. The high temperature peak can be assigned to the reduction
of carbon species on the surfaces of CNTs.16,21 It can be noted that in the
Fe@CNT sample, the three peaks assigned to the three transitions of Fe
species, collapse to only one reduction peak. It is possible to identify
o
another reduction peak at low temperature ~185 C. These facts imply that
the confinement of Fe on the CNTs results in an easier reduction. The
sample Fe@NCNT shows three peaks in the magnified region at 186, 338
and 400 oC, located at lower temperature than in pure Fe
evidence of the different interaction of Fe with NCNTs than with CNTs.
-TPR profile for pure V , Fig. 3d, shows three reduction peaks, a =
62 °C, b = 703 °C, c = 764 °C associated to the transitions: V >V
> V
.[22] In an attempt to attribute the relevant transitions to Fe
2
3
O . This is an
H
6
>
2
2 5
O
O
2 5
O
6 13
0
500
1000
1500
2000
2500
3000
3500
V
2
O
4
O
2 3
Wave Numbers (cm-1)
or V in V-Fe@CNT (Fig.3e), the peak at lowest reduction temperature
centered at 386 °C is tentatively attributed to the transitions belonging to
Fe and the other two to the transitions of V. Both values are present at
lower temperatures than in pure oxides.
Fe@CNTs
Fe@NCNTs
V-Fe@CNTs
Figure 5. Raman spectra of activated Fe@CNTs, Fe@NCNTs, and V-
Fe@CNTs
Catalysts characterization by using surface techniques
SEM of the as produced Fe@CNT (Fig. 4a) show the formation of aligned
bundles of nanotubes. The micrograph is representative for all catalysts
produced. TEM micrographs of the V-Fe@CNT catalyst after activation
Table 7. The calculated ID/IG ratios of Fe@CNTs, Fe@NCNTs, and
V-Fe@CNTs following activation to remove the graphitic carbon layer
(
Fig. 4b) show the presence of a large number of metal catalyst particles
Sample
D G
I /I
on the surface of the nanotubes without the graphitic coating, making them
active for catalysis. The small D and large G and G’ peaks observed in
Fe@CNTs (Fig. 5) indicate good purity and long-range order in the CNT
Fe@CNTs
V-Fe@CNTs
Fe@NCNTs
0.19
0.69
0.91
support structure. Conversely, the increased I
D
/I
G
ratio and suppressed G’
2
peak observed in Fe@NCNTs indicate the alteration of the sp carbon
structure as a result of nitrogen doping. V-Fe@CNTs also show increased
disorder compared to the base material, as a result of the vanadium doping
process, though a degree of long-range order is maintained.
-
1
This is confirmed by the presence of the G’ peak at ca. 2788 cm . The
/I ratios per each catalyst are reported in Table 7.
I
D G
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
XRD analysis (Fig. 6) of Fe@CNTs, Fe@NCNTs, and V-Fe@CNTs was
conducted prior to reaction. All materials displayed a strong peak at 26.2°,
which is indicative of the CNT support structure. Peaks at 30.4°, 35.8°,
Acknowledgements
AD thanks LIDIA Project (CTN01_0063_255060) for partial
support. VALBIOR Project (Apulian Region) is gratefully thanked
for the use of equipment.
4
2 3
3.2°, 54.3°, 57.8°, and 62.8° all indicate the expected presence of Fe O
23
in all samples, though they are less pronounced in the Fe@NCNTs. This
is likely due to the lower activation temperature employed to remove the
graphitic layer from the catalytic nanoparticles (673 K compared to 843 K),
resulting in less complete oxidation of the iron.
Keywords: oxalic acid from biomass • aerobic oxidation • glucose
•
5-HMF • M@CNT/NCNT (M=Fe, V).
[
[
1]
R. D. Perlack, L. L. Wright, A. F. Turhollow, R. L. Graham, B. J. Stokes,
D. C. Erbach, 2005, www1.eere.energy.gov/bioenergy/pdfs/final_
billionton_vision_report2.pdf.
2]
a) M. Ventura, A. Dibenedetto, M. Aresta, Inorg Chim Acta, 2018, 470,
1-21; b) M. Aresta, A. Dibenedetto, E. Quaranta, "Reaction Mechanism
1
in Carbon Dioxide Conversion, Springer Verlag, 2016.
[
[
3]
4]
Z. Zhang, K. Deng, ACS Catalysis 2015, 5, 6529-6544.
Q. Xiang,Y. Y. Lee, R. W. Torget, in Proceedings of the Twenty-Fifth
Symposium on Biotechnology for Fuels and Chemicals Held May 4–7,
Figure 6. XRD spectra of activated Fe@CNTs, Fe@NCNTs and V-Fe@CNTs.
Spectra indicate the presence of the CNT support (+), Fe O (x), Fe O (∆) and
2 3 3 4
iron carbide species (□).
2
003, in Breckenridge, CO (Eds.: M. Finkelstein, J. D. McMillan, B. H.
Davison, B. Evans), Humana Press, Totowa, NJ, 2004, pp. 1127-1138.
A. Dibenedetto, M. Aresta, L. di Bitonto, C. Pastore, ChemSusChem
[
5]
2016, 9, 118-125.
[
[
6]
7]
F. D. Pileidis, M. –M. Titirici, ChemSusChem 2016, 9, 562-582.
J. Yun, G. D. Yao, F. M. Jin, H. Zhong, A. Kishita, K. Tohji, H. Enomoto,
L. Wang, Aiche Journal 2016, 62, 3657-3663.
Peaks at 35.8°, 43.2°, 57.8°, and 64.0° all indicate the presence of Fe
3 4
O
in all samples.23 Overlapping peaks at 43.2°, 44.0°, and 45.0° with a small
[8]
a) A. A. Rosatella, S. P. Simeonov, R. F. M. Frade, C. A. M. Afonso,
Green Chem. 2011, 13, 754-793; b) X. L. Tong, Y. Ma, Y. D. Li, Appl.
Catal. A-Gen. 2010, 385, 1-13; c) A. Dibenedetto, M. Aresta, C. Pastore,
L. di Bitonto, A. Angelini, E. Quaranta, RSC Adv. 2015, 5, 26941-26948.
L. Li, F. Shen, R. L. Smith, X. Qi, Green Chemistry 2017, 19, 76-81.
shoulder peak at 41.2° indicate the presence of iron carbides such as Fe
C being the most likely to occur.23 These peaks appear
and Fe , with Fe
3
C
C
5 2
3
more defined in the Fe@NCNTs once again due to the lower activation
temperature used, resulting in less oxidation of the iron, which has been
shown elsewhere to exist mostly as metallic iron, iron carbides and iron
nitrides prior to activation in air.18,23 The harsher activation conditions used
on Fe@CNT and V-Fe@CNTs thus convert more of the initial carbide
phase into the oxide.
[
[
9]
10] Intratec Solutions Enterprise, Economics of Oxalic acid from sugar, Vol.
E251110A0, 2017.
[
[
11] Encyclopedia of Chemical technology, Wiley: Kirk-Othmer, Vol. 9, 1991.
12] a) S. K. Mandal, P. C. Banerjee, Process Biochem. 2005, 40, 1605–
1610; b) H. Strasser, W. Burgstaller, F. Schinner, FEMS Microbiol Lett
1994, 119, 365–370; c) G. J. G. Ruijter, P. J. I. van de Vondervoort, J.
Visser, Microbiology 1999, 2569–2576.
Catalytic tests
[
13] a) W. Guo, H. Jiang, Y. N. Huaxue, J. CA. 1997, 45 (4), 240-241; b) L.
S. Richmond, Vol. US3720591 A, USA, 1973; c) E. Yonemitsu, T. Isshiki,
T. Suzuki, Y. Yashima, in U.S. Patent, Vol. 3,678,107 (Ed.: U. S. Patent),
U.S, 1969; d) C. L. Mehltretter, C. E. Rist, J. Agr. Food Chem. 1953, 1,
The conversion of the C6 substrate (5-HMF, fructose, glucose or sucrose)
at a fixed temperature was studied in a 50 mL stainless-steel reactor
equipped with a withdrawal valve and an electrical heating jacket. 1 mmol
of the desired substrate (5-HMF: 0.126 g; D-glucose: 0.180 g, D-fructose:
7
79-783.
14] G. M. M.-I. Radulescu, L. Cornos, G. Vapler, Vol. RO106878 B1 30: 4
Ed.: Rom.), 1993.
15] a) M. Aresta, A. Dibenedetto, M. Ventura, Italian patent
02016000014339, Italy, 2017; b) M. Ventura, M. Aresta, A. Dibenedetto,
[
[
0.182 g, sucrose 0.340 g; succinic acid: 0.12 g) was dissolved in 5 mL of
(
distilled water in a glass reactor, in which 0.02 g of the catalyst under study
and a magnetic stirrer was placed. The glass-reactor was then transferred
into the autoclave that was closed and purged three times with O . It was
2
1
ChemSusChem 2016, 9, 1096 – 1100.
charged with the appropriate pressure of oxygen and heated to the
reaction temperature as specified in Results and Discussion. At fixed
intervals of time, stirring was stopped, a sample of the liquid (0.1 mL) was
withdrawn and analysed by HPLC following the disappearance of the
substrate. When the concentration of the latter dropped to a constant value
[
[
[
[
[
16] X. Guo, J. Guan, B. Li, X. Wang, X. Mu, H. Liu, Scientific Reports Nature
2
015, 5, 16451, doi: 10.1038/srep16451
17] C. Zhou, W. Deng, X. Wan, Q. Zhang, Y. Yang, Y. Wang, ChemCatChem
015, 7, 2853-2863.
2
18] O'Byrne, J. P.; Owen, R. E.; Minett, D. R.; Pascu, S. I.; Plucinski, P.;
Jones, M. D.; Mattia, D. Catal. Sci. Technol 2013, 3, 1202-1207.
(or even zero), the reaction was stopped.
Recycling tests were carried out using Fe@CNT, working at 250 rpm. After
a catalytic run (typically Entry 4, Table 1) the catalyst was recovered by
centrifugation and reused in four more consecutive runs in the conditions
given in Table 1. The conversion remained at 99%, while the selectivity
towards OA was 48.0 (run 2), 48.1 (run 3) , 47.9 % (run 4), and 48.0 %
19] J. Zielinski, I. Zglinicka, L. Znak, Z. Kaszkur, Appl. Catal. A: Gen. 2010,
381, 191-196.
20] W. K. Jozwiak, E. Kaczmarek, T. P. Maniecki, W. Ignaczak, W.
Maniukiewicz, Appl. Catal. A: Gen. 2007, 326, 17-27.
[
[
21] H. E. A. Yang, J. Mol. Catal. A Chem 2010, 323, 33-39.
22] M. M. Koranne, J. G. Goodwin, G. Marcelin, J. Catal. 1994, 148, 369-
(
run 5), demonstrating an excellent stability. No Fe was leached.
In all experiments the selectivity toward the ith species is expressed as the
ratio of the nith (mol of the ith species as measured in the reaction mixture)
to the ni,t (number of mol of the ith species if the starting material would
have been converted 100% into such species). In this way, the sum of
377.
[
23] D. R. Minett, J. P. O'Byrne, S. I. Pascu, P. K. Plucinski, R. E. Owen, M.
D. Jones, D. Mattia, Catal. Sci. Technol. 2014, 3351-3358
selectivities is also the carbon balance, in fact,  s = 100 When the sum
i
is less than 100, it means that there are species that were not quantified.
In general, all experiments gave a carbon balance at 99-100%.
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201702347
FULL PAPER
Entry for the Table of Contents
FULL PAPER
Maria Ventura,[a] David Williamson,
[c]
[
d]
Francesco Lobefaro, Matthew D.
Jones,[ Davide Mattia,[c]* Francesco
Nocito,[b] Michele Aresta,[d] and Angela
Dibenedetto[a,b]*
e]
Page No. – Page No.
In this paper we report the aerobic oxidation-cleavage of C6 polyols (5-HMF,
glucose, fructose, sucrose) in water to oxalic acid-OA (overall yield 48-62%) (and
succinic acid-SA) under mild conditions using M@CNT/NCNT (M=Fe, V;
CNT=carbon nanotubes; NCNT= N-doped CNT), which are recoverable and
reusable without any loss of efficiency.
Sustainable synthesis of oxalic (and
succinic) acid via aerobic oxidation of
C6 polyols by using M@CNT/NCNT
(
M=Fe, V) based catalysts in mild
conditions
This article is protected by copyright. All rights reserved.