Levulinic acid upgrade to succinic acid with hydrogen peroxide

Article abstract of DOI:10.1016/j.apcata.2018.06.034

Levulinic acid is produced from the acidic aqueous degradation of 5-hydroxymethylfurfural, with potential applications in bio-value added chemicals synthesis. Here, we report for the first time, the Baeyer-Villiger oxidation of levulinic acid to succinic acid, with hydrogen peroxide and tungstic acid at mild conditions and without any organic solvent. We investigated the effects of time, amount of reagent-to-catalyst molar ratio and H₂O₂-to-levulinic acid molar ratio. The maximum succinic acid selectivity was 75% with a levulinic acid conversion as high as 48%, after 6 h at 90 °C. We propose a reaction mechanism based on results obtained from the reactivity of the intermediates. The catalyst interacts with the substrate, forming a cyclic species that enhances the formation of succinic acid versus 3-hydroxypropanoic acid.

Full text of DOI:10.1016/j.apcata.2018.06.034

Accepted Manuscript
Title: Levulinic acid upgrade to succinic acid with hydrogen
peroxide
Authors: Davide Carnevali, Marco G. Rigamonti, Tommaso
Tabanelli, Gregory S. Patience, Fabrizio Cavani
PII:
S0926-860X(18)30312-0
DOI:
Reference:
https://doi.org/10.1016/j.apcata.2018.06.034
APCATA 16722
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
16-4-2018
6-6-2018
25-6-2018
Please cite this article as: Carnevali D, Rigamonti MG, Tabanelli T, Patience GS, Cavani
F, Levulinic acid upgrade to succinic acid with hydrogen peroxide, Applied Catalysis
A, General (2018), https://doi.org/10.1016/j.apcata.2018.06.034
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Levulinic acid upgrade to succinic acid with hydrogen peroxide
a,b
a
b
a
Davide Carnevali , Marco G. Rigamonti , Tommaso Tabanelli , Gregory S. Patience , Fabrizio
Cavanib,
∗
aPolytechnique Montreal, Chemical Engineering department, 2900 Edouard Montpetit Blvd,
Montreal, H31 1J4, QC, Canada
bDipartimento di Chimica Industriale “Toso Montanari”, Alma Mater Studiorum Universitá di
Bologna, Viale del Risorgimento, 4, 40136, Bologna, Italy
Graphical abstarct
Highlights
-
-
Solvent-free conversion of levulinic acid to succinic acid with hydrogen peroxide and
WO
H
2
4
Formation of an adduct between the catalyst and the substrate, which enhances the
1

regioselectivity
-
-
Maximum conversion of 48 % with 75% selectivity to succinic acid, after 6 h
A new green process for synthesize succinic acid from bio-sourced levulinic acid
Abstract
Levulinic acid is produced from the acidic aqueous degradation of 5-hydroxymethylfurfural,
with potential applications in bio-value added chemicals synthesis. Here, we report for the first
time, the Baeyer-Villiger oxidation of levulinic acid to succinic acid, with hydrogen peroxide
and tungstic acid at mild conditions and without any organic solvent. We investigated the effects
of time, amount of reagent-to-catalyst molar ratio and H
2
2
O -to-levulinic acid molar ratio. The
maximum succinic acid selectivity was 75 % with a levulinic acid conversion s high as 48 %,
after 6 h at 90°C. We propose a reaction mechanism based on results obtained from the reactivity
of the intermediates. The catalyst interacts with the substrate, forming a cyclic species that
enhances the formation of succinic acid versus 3-hydroxypropanoic acid.
Keywords: levulinic acid, succinic acid, hydrogen peroxide, Baeyer-Villiger, oxidation, tungstic
acid, H
2
WO
4
1
. Introduction
Biomass is a primary source of energy and a potential feedstock for bio-chemicals. Compared to
fossil fuels, it has an elevated oxygen-to-carbon ratio, which makes it more suitable for the
synthesis of highly functionalized value-added molecules, like to 5-hydroxymethylfurfural
(HMF) by fructose dehydration [1,2]; a further oxidation of the intermediate forms 2,5-
2

furandicarboxylic acid (FDCA) [3,4]. The dehydration of fructose occurs with high temperature
or by the presence of an acidic Brønsted catalyst. The main synthesis hurdle is that HMF
degrades in aqueous acidic condition, and forms insoluble polymers (humins), or cleaves to
make levulinic acid (LEV) and formic acid. Most works in this area apply multi-phasic reactors
[
4], membrane reactors [5] or ionic liquids [6] to reduce humins.
LEV is an important bio-building block for valeric acid [7], methyltetrahydrofuran [8], alkyl
levulinate [9], diphenolic acid [10], δ-amino levulinic acid, γ-valerolactone [11,12], α-
methylene-γ-valerolactone [13] and succinic acid. Derivatives applications regard mainly the
production of rubber, plasticizers, pharmaceuticals and textiles [14]. Biomass is the preferred
starting material to produce LEV [15–17]. LEV and succinic acid are in the top 10 value added
chemicals from biomass [18,19]. The succinic acid market value reached 400 billion USD per
year [20]. Industrially, the favored route to succinic anhydride is the hydrogenation of maleic
anhydride, derived from fossil n-butane [21]. On the contrary, the main bio-pathway is the sugars
fermentation.
LEV reacts with basic, acidic and metal oxide catalysts to form succinic acid; most studies add
solvent to increase product yield (Table 1). One promising reaction is the Baeyer-Villiger
oxidation with hydrogen peroxide as an oxidizing agent. Adding catalyst increases the reaction
rate and improves the selectivity to the desired products. Examples of catalysts include alumina
[
[
22], tungstic acid [23], tin [24], titanium silicalite-1 (TS-1) [25], selenium [26] and bio-enzymes
27]. The mechanism comprises the nucleophilic addition of a hydroperoxide anion to the
carbonyl carbon [28]. Environmental conditions modulate the intermediate rearrangement; acidic
catalysts react with the substrate to form the methyl ester, while in basic conditions it produces
the acetate through an intramolecular addition. Starting from humic acids, hydrogen peroxide
3

achieved 93 % malonic acid or 62 % succinic acid [29].
Table 1: Review of the conversion with reaction parameters and yields. LEV: levulinic acid; M-
LEV: methyl levulinate; SUCC: succinic acid; M-SUCC: methyl succinate; DM-SUCC:
dimethyl succinate; MA: maleic anhydride.
Reagent
T, °C
Catalyst
Solvent
Time, h
Products
SUCC
SUCC
SUCC
Yield, % Ref.
LEV 300 – 400
V
2
O
5
[30]
[31]
LEV
LEV
40 – 60
V
2
O
5
+ HNO
Ru-based
magnetic nano
3
1 - 4
6
150
H
2
O
78
[32]
particles + O
2
LEV
M-LEV
LEV
90
80
Mn-based
catalyst
AcOAc
10
6
SUCC, MA
M-SUCC
DM-SUCC
SUCC
9, 26
41
[33]
[34]
[35]
[36]
[37]
Amberlyst-15 + Methanol
H
2
O
2
200
80
K
2
CO
3
Dimethyl
carbonate
6
21
Furfural
LEV
Amberlyst-15 +
H
2
O
24
1
74
H
2
O
2
25
I
2
t-BuOH
SUCC
87
Wu et al. [38] described the basic oxidation of LEV with H
2
2
O to 3-hydroxypropionic acid
(HPA), a precursor of acrylic acid. On the contrary, hydrogen peroxide, combined with sulfuric
acid, achieved a maximum of 48 % succinic acid selectivity. Replacing H
2
SO
4
with
trifluoroacetic acid, increased selectivity to 62 %, at 90 °C after 2h [39].
Here for the first time, we describe the acidic, organic solvent-free Baeyer-Villiger oxidation of
LEV to succinic acid with H and H WO . We investigated the effects of time,
reagent:catalyst ratio and H :LEV ratio on LEV conversion and on the main products
2
O
2
2
4
2
O
2
selectivity. Furthermore, we investigated the reaction pathway, testing the intermediates
4

reactivity.
2
. Experimental
2
.1. Materials
We used all reagents without further purification. The purity of LEV was greater than ≥ 97 %,
succinic acid > 99 %, formic acid purity varied between 95 % to 97 %, propionic acid > 99.5 %,
α-angelica lactone purity was 98 %, malonic acid was 99 %, and stabilized hydrogen peroxide
2
7 % w/w in H
2
O), sulfuric acid (ACS reagent, 95 % to 98 %) and potassium permanganate
(97 %) from Sigma Aldrich. Glacial acetic acid (ACS Reag. Ph Eur) was acquired from Carlo
Erba, while oxalic acid (98 % anhydrous) from Acros Organics. We purchased the 3-
hydroxypropionic acid sodium salt from Toronto Research Chemicals. The tungstic acid (99 %)
catalyst was purchased from Sigma Aldrich.
2
.2. Catalytic experiment
We designed a full factorial set of experiments with 3 factors and 3 levels each (27 tests) with the
statistical software Minitab®. The reactions time were at 1 h, 3 h, 6 h, with a LEV:H WO ratio
of 100:1, 50:1, 10:1 and H :LEV ratio of 1.5:1, 3:1, 5:1. An oil bath kept the temperature
2
4
2
O
2
constant at 90 °C. We derived the sample standard deviation, s (Table 2), by repeating
experiments at three different conditions.
5

Table 2. Repetition of three tests at various conditions. * H
2
O
2
:LEV molar ratio, ** LEV:
H
2
WO molar ratio. Conversion (X) and Selectivities (S) are in %. Succ: succinic acid; acet:
4
acetic acid; mal: malonic acid; oxal: oxalic acid, form: formic acid, AL: α-angelica lactone;
HPA: 3-hydroxypropionic acid; gly+glyox: glycolic acid and glyoxylic acid; prop: propionic
acid.
Conv
S
succ
S
acet
S
mal
S
oxal
S
form
S
AL
S
HPA
S
gly+glyox
S
prop
Sample standard deviation s
2
2
0.5 0.3 0.3 0.2
1
0.4
0
0.3
We investigated the reaction pathway, evaluating the reactivity of the intermediates at 1 h and 6
h, with a LEV:H WO molar ratio of 100 and H 2:LEV molar ratio of 5.
In a standard reaction procedure, we added 5 mmol (0.58 g) of LEV into an Erlenmeyer flask,
with the specified amount of catalyst (H WO , i.e. 0.124 g for a LEV:H WO molar ratio of 10)
and H solution (aqueous solution 27 % w/w, i.e. 0.93 g for a H 2:LEV molar ratio of 1.5).
2
4
2
O
2
4
2
4
2
O
2
2
O
The mixture was homogenized for few minutes, while an oil bath heated it to the desired
temperature (90°C). The flask was equipped with a reflux condenser to avoid an excessive
evaporation of the solution. By means of gas-chromatography we qualitatively analyzed the non-
condensable gases inside the flask, which turned out to contain CO . At the end of the test, the
2
mixture was cooled rapidly from 90 °C down to room temperature, diluted with distilled water
and filtered using a disposable syringe equipped with 0.20 µm PTFE filter to preserve the HPLC
column.
Tungstic acid is insoluble in cold water (only slightly soluble in hot water), however in the
presence of H
working at high H
recovery by filtration [40].
For catalyst recycle tests, we put the mixture in the rotavapor, in order to completely evaporate
2
O
2
the catalyst is soluble in the state of pertungstic acid (H
2
WO ). This means that
5
2
O
2
conversion would favor the precipitation of the catalyst facilitating its
water and traces of residual H
2
O
2
. In order to remove the adsorbed organic compounds, the
6

catalyst was washed with acetone, and dried in the oven at 80 °C overnight. This procedure
allowed to recover the 70% of the initial H WO amount, in the form of a yellow solid.
.3 Analytical procedure
2
4
2
A Varian CP-3380 gas chromatography equipped with a packed Carbosieve SII column (2m
length, stationary phase of active carbons of 80-100 mesh) was used to analyze the non-
condensable gases. A TCD monitored the concentration of CO, CO
2
, O
2
and N , with He as the
2
carrier gas. The oven was maintained at 40°C for 7 min, then the temperature was ramped up
until 220°C with a rate of 30°C/min, and finally left isothermal for 10 min.
An HPLC Agilent Technologies 1260 Infinity, equipped with an Aminex HPX-87H Ion
Exclusion column (300 mm x 7.8 mm), maintained at 30 °C, measured the concentration of the
crude products. The autosampler injected 10 µL of the solution, while the eluent was a 5 mM
−
1
H
2
SO solution, flowing at 0.4 mL min . The instrument monitored the response of the samples
4
at 8 wavelengths (from 210 nm to 284 nm). The LEV conversion (Eq. 1) and the products
selectivity based on carbon (Eq. 2) were calculated as follows:
푖
푓푖
ꢀꢁꢂ
푛
ꢁꢂ− 푛
ꢀ
푋퐿퐸푉
=
· 100
(1)
(2)
푖
푛
ꢀ
ꢁꢂ
푓
푛
휗ꢃꢄꢅꢆ
ꢃꢄꢅꢆ
푆푃푅푂퐷 = 푛푖
·
· 100
휗ꢀꢁꢂ
ꢀꢁꢂ− 푛푓
ꢀꢁꢂ
where niLEV and nfLEV are the initial moles and the final LEV moles in the system, nfPROD the final
moles of product analyzed by HPLC. ϑLEV and ϑPROD are the number of carbon atoms of LEV
and products.
We evaluated the amount of unconverted hydrogen peroxide with a 0.3 N potassium
permanganate solution and a diluted 1:4 aqueous sulfuric acid solution. The titration proceeded
with the addition of distilled water and 10 mL of acidic solution to the sample. We titrated the
7

sample using a calibrated burette until the solution turned permanent pink [41]. The mass
percentage of H by (Eq. 3):
퐻 ꢇ , % g g
2
O
2
푚퐿퐾푀ꢉꢅ4 · 푁퐾푀ꢉꢅ4· ꢊ.ꢊꢈ7ꢊ ·ꢈꢊꢊꢊ
−
ꢈ
=
(3)
2
2
푔푠푎ꢋ푝푙푒
where N is the normality of the KMnO
4
solution, mLKMnO4 is the volume of permanganate used
for the titration and 0.0170 is the mass per milliequivalent of H
2
2
O .
3
. Results and discussion
3
.1. LEV reactivity and selective oxidation
Reactivity screening tests at three H
2
O
2
concentration (H
2
O2:LEV molar ratio of 1:1, 1.5:1 and
3
:1), three temperatures (50 °C, 70 °C and 90 °C) and constant H
2
WO :LEV molar ratio of 100
4
demonstrated a very low LEV conversion, reaching a minimum of 5 % at the lowest temperature
and peroxide content (Test 31, Table 3). Increasing the temperature led to an increased
conversion of LEV. Therefore, all the further experiments were carried out at 90 °C.
Time, catalyst mass and H
Figure 1, Table 3). The maximum variation of conversion was recorded at H
and LEV:H WO ratio of 50; the value increased from 28 % after 1 h to 46 % after 6 h reaction
time (Table 3, entries 13 and 15). Addition of H and H WO amplified the oxidative power
2
O
2
content affect succinic acid selectivity and LEV conversion
(
2
O
2
:LEV ratio of 3
2
4
O
2 2
2
4
and the acidity of the reaction media, enhancing the reactivity of the substrate. The highest LEV
conversion was shown at 6 h, reaching 57 % at the highest hydrogen peroxide concentration and
catalyst loading (Table 3, entry 27). Furthermore, at these conditions, we recorded the highest
conversion after 1 h of reaction (Figure 1).
Succinic acid selectivity increased from 10 % (Table 3, entry 1) to 49 % (entry 9) with increasing
H
2
O
2
and time. Increasing the H
2
WO
4
, its behavior changed; at 50:1 LEV:H
2
WO molar ratio,
4
8

the selectivity increased with peroxide concentration, reaching a maximum after 3 h, but then
decreased (entries 10-18). This trend was accentuated at higher catalyst loading, where the
highest selectivity was registered after 1 h, and then decreased with time (entries 19-27). Figure
2
summarizes the effect of the two main reaction parameters on selectivity to succinic acid at 3 h.
Figure 1 Effects of various amount of H
2
O
2
and H
2
WO on the conversion at 1 h, 3 h and 6 h.
4
LEV:H WO molar ratio: 100 (left figure), 50 (middle), 10 (right).
2
4
We measured acetic acid, malonic acid, oxalic acid, formic acid, α-angelica lactone, 3-
hydroxypropionic acid, propionic acid, glycolic acid and glyoxylic acid concentration by means
of HPLC (Table 3). Some intermediates reacted further to CO and volatile compounds [42],
2
which were undetected. Acetic acid was the main by-product and it formed in all tests. Many
compounds decomposed to this acid, which accounts for its high selectivity, and increased with
H
2
WO
4
and H
2
O
2
concentration and longer time. Selectivity to acetic acid approached 10 % after
1
h, and rose to 20 % after 6 h. The highest selectivity was 23 % (Table 3, entry 23). We detected
low concentrations of oxalic acid in all the reactions, but it never exceeded 5 %, and mainly
remained below 1 %. Oxalic acid easily oxidizes in the presence of acidic oxidizing
environment, releasing CO
2
[43]. For LEV:H
2
WO catalytic ratio higher than 50, selectivity
4
9

hardly were higher than 1 %, while at ratio of 10, values increased from 2 % to 5 %.
Figure 2 Succinic acid selectivity reached a maximum with the highest concentration of H
and LEV:H WO ratio of 50, after 3 h.
2
O
2
2
4
Glycolic and glyoxylic acids showed low selectivity, which remained almost always lower than
%. Formic acid presented a similar trend: selectivity varied between 0 % and 5 %, with a
general increase with time and H WO loading. Results at low H concentration showed that
the absence of reactive oxygen stopped the Baeyer-Villiger oxidation, favouring the dehydration
1
2
4
2
O
2
of LEV to α-angelica lactone. In acidic conditions and low H
2
O
2
concentrations (LEV:H
2
O
2
2
molar ratio 1:1), selectivity to α-angelica lactone was 50 % from LEV. At higher H
2
O
concentration, α-angelica lactone degraded to acetic acid and malonic acid (Table 4). α-Angelica
lactone selectivity decreased with reaction time and was the greatest with low tungstic acid
loading (Table 3, entries 1, 4 and 7): at 1 h, 3 h, and 6 h, selectivity dropped from 38 % down to
2
1 %, and 14 %, respectively. In experiments at a 1.5:1 H
2
O
2
:LEV ratio and 100:1 LEV:H
2
WO
4
10

ratio, after 1 h the lactone selectivity was 38 %, but dropped to 2 % after 6 h (entries 1-3). This
behavior could be attributed to the degradation of the lactone, which is highly unstable in the
presence of oxidizing agents and acids. Selectivity to the lactone was very high in experiments
carried out with a H
The selectivity to 3-hydroxypropionic acid was always lower than 4 % with the lowest H
concentration. However, the selectivity increased with H concentration and reached 25 % at a
:LEV ratio of 5:1 (entries 8 and 9 in Table 3). Generally, the selectivity reached a maximum
after 3 h, and then slowly decreased. Furthermore, higher H WO loading boosted the reactivity
2
O
2
:LEV ratio of 0.5 (entries 28-30).
2
O
2
2
O
2
H
2
O
2
2
4
of 3-hydroxypropionic acid, resulting in selectivity values lower than 1 %.
Succinic acid may decarboxylate to propionic acid. At the highest amount of H
2
WO (low
4
LEV:H
2
WO
4
ratio) almost no propionic acid was detected, due to its degradation to acetic acid
and CO
2
. The highest values of propionic acid selectivity were shown with the lowest catalyst
loading and highest hydrogen peroxide concentration, reaching a selectivity of 6 % (Table 3,
entries 8 and 9).
11

Table 3 Effects of time, H
2
WO
4
and H
2
O
2
content on conversion and selectivity. * H
2
O
2
:LEV molar ratio, ** LEV:H
2
WO molar
4
ratio. Succ: succinic acid; acet: acetic acid; mal: malonic acid; oxal: oxalic acid, form: formic acid, AL: α-angelica lactone; HPA: 3-
hydroxypropionic acid; gly+glyox: glycolic acid and glyoxylic acid; prop: propionic acid.
*
** Test 31 at 50 °C, adding one drop of H
2
2
O every 10 seconds, and at the end leaving the reaction run for 20 min more
Exp
H
2
O
2
*
H
2
WO
4
**
Time, h
Conv, %
Ssucc, %
Sacet, %
S
mal, %
S
oxal, %
S
form, %
SAL, %
S
HPA, %
S
gly+glyox, %
S
prop, %
1
2
3
4
5
6
7
8
9
1.5
1.5
1.5
3
100
100
100
100
100
100
100
100
100
50
1
3
6
1
3
6
1
3
6
1
3
6
1
3
6
1
3
6
1
3
6
1
3
18
24
28
29
32
35
36
44
47
19
25
32
28
36
46
37
48
49
25
29
43
29
43
10
17
19
10
47
48
2
14
18
20
8
5
7
7
6
6
6
3
4
4
5
8
6
3
7
5
2
4
4
7
8
4
9
7
0.7
0.7
2
2
3
38
4
1
3
4
0.6
2
0.1
0.2
2
3
2
4
0.3
2
0.3
0.6
0.8
0.2
0.4
0.5
0.8
3
0.6
0.9
1
21
0.2
0.3
14
0.6
2
4
3
13
17
3
14
11
0.6
25
25
1
0.5
0.1
0.5
0.1
0
3
3
3
5
0.4
1
4
5
48
49
12
19
18
32
41
29
23
75
71
35
18
11
62
32
8
6
5
12
11
17
18
4
1
6
1
1
1
1
1
1
1
1
1
1
2
2
2
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
1.5
1.5
1.5
3
3
29
3
0.2
0.2
0.2
0.1
0.2
0.2
0
0.5
0.3
0.1
2
50
3
3
50
2
3
2
2
50
0.3
2
0.2
2
9
0
3
50
18
22
1
1
11
3
3
3
50
2
2
4
0.5
2
5
50
0.1
0.7
0.9
5
1
10
3
0
5
50
9
0.6
0.8
3
15
12
1
0.1
0.1
2
5
5
50
17
13
19
13
9
0
3
1.5
1.5
1.5
3
10
4
0.2
0
10
3
4
3
1
0.2
0.2
0
10
2
3
0
1
0
10
10
5
3
6
2
0.9
0.6
3
23
5
4
3
1
0.2
12

2
2
2
2
2
2
3
4
5
6
7
8
9
0
3
5
10
10
6
1
32
43
54
57
15
16
17
5
22
58
39
35
7
20
12
21
22
5
6
5
5
4
3
4
3
1
3
4
5
2
3
3
1
2
2
0
1
4
1
1
0.1
1
0
2
5
10
3
0.3
4
1
1
1
0.3
0
5
10
6
1
1
0
0.5
0.5
0.5
3
100
100
100
100
1
0.3
0.6
0.4
0
53
50
49
24
0.7
0.8
0.7
0
0.4
0.2
0.3
0
0.2
0
3
10
7
11
8
6
0.1
0
3
1***
0.35
6
2
13

In the three experiments at H
2
O
2
:LEV ratio 0.5:1 and LEV:H
2
WO ratio 100:1 (Table 3, entries
4
2
8-30), H conversion approached 100 % after 1 h reaction time; in fact, LEV conversion
2
O
2
showed a moderate growth, passing from 15 % at the first hour to 17 % after 6 h. Besides the
usual by-products, we also detected the formation of acrylic acid (not shown in Table 3); its
selectivity was 8 % after 1 h, slightly increasing to 13 % after 3 h and then decreasing down to
1
0 % after 6 h.
In order to check the stability and recyclability of H
2
WO , the catalyst was recovered after
4
reaction (see Experimental), and recycled a few times. Differences of catalytic performance
between each experiment were within the standard deviation; therefore, the behavior of the
recycled catalyst can be considered to be the same as that one of the fresh catalyst (Table S1 in
the Electronic Supplementary Information).
3
.2. Selectivity versus time
Most of the by-products selectivity were invariant with time (Figure 3): 3-HPA selectivity was
%, malonic acid 8 %, formic acid 4 %, oxalic acid about 2 %, whereas -angelica lactone
1
varied from 0 % to 5 %. Succinic acid selectivity decreased from 35 % to 11 % after 6 h. Acetic
acid, the main by-product, reached a maximum of 19 % at 3 h after which it dropped to 13 %.
LEV conversion was 25 % after 1 h and increased to 43 % at 6 h. We did not detect the
formation of propionic acid.
The total carbon selectivity decreased from 75% after 1 h down to 40-45% after 6 h reaction
time. This drop is due to the product overoxidation occurring along with the increasing LEV
conversion. We did not close the carbon mass balance due to the formation of volatile molecules
(
mainly CO ) that left the reactor and the formation of insoluble polymers, which slightly
2
14

changed the color of the solution during the reaction. For short reaction times the solution was
colourless.
We calculated the H
peroxide was the limiting reagent (H
complete (98 %). When the oxidant was in slight excess (ratio 1.5:1), after 1 h reaction time
conversion was 86 % (overall selectivity to oxidised products, calculated with respect to
converted H , was 56%; therefore, 44% of converted H decomposed), which increased up
to 92 % after 3 h (selectivity 54%). Working in large oxidant excess (ratio 5:1) and the highest
catalyst loading, the H conversion dropped to 64 %.
2
O
2
remaining in the reaction mixture by titration. When the hydrogen
2
O
2
:LEV ratio 0.5:1), the oxidant conversion was almost
H
2
O
2
2
O
2
2
O
2
2
O
2
Figure 3 Variation of the selectivity of the products with time. LEV:H
2
WO
4
=10 ; H
2
2
O :LEV
=1.5; T 90 °C.
3
.3. The reaction network
The oxidation pathway proceeds with the formation of various by-products and intermediates.
The Baeyer-Villiger mechanism consists of the addition of an oxygen atom to convert a ketone
15

to an ester. Oxygen from a peroxide compound attacks the molecule with two competitive
mechanisms to form methyl succinate or 3-acetoxypropanoic acid [37]. Moreover, in the
presence of an acidic environment or high temperature, LEV dehydrates to α-angelica lactone
[
44] (Figure 4).
Wu et al. [38] found that the regioselectivity of the Baeyer-Villiger oxidation with hydrogen
peroxide is pH controlled: under alkaline conditions the insertion of oxygen is favored on the
secondary carbon, leading to the 3-acetoxypropanoic acid intermediate and subsequently to 3-
HPA and acetic acid, with 50 % yield. In strong acidic conditions - trifluoroacetic acid as solvent
-
the migration of the less substituted carbon is favored. This mechanism produces methyl
succinate as the intermediate, with subsequent acidic cleavage of the ether to methanol and
succinic acid, with 60 % yield. As a side reaction, 10 % 3-HPA is also recovered [39].
Figure 4. Hydrogen peroxide adds an oxygen to LEV with two competitive mechanisms. The
first leads to 3-acetoxypropanoic acid and consequently to 3-HPA, while the second produces
methyl succinate, which is in equilibrium with succinic acid.
16

Methanol can further react with H
and finally CO , but MeOH and CH
compounds oxidize to CO in the presence of hydrogen peroxide.
-HPA, formed by hydrolysis of 3-acetoxypropanoic acid, can dehydrate to acrylic acid in the
2
O
2
to form oxidized derivatives: formaldehyde, formic acid
2
2
O were absent in the HPLC chromatogram traces. These
2
3
absence of oxidizing agents [45], or can follow the retro-aldolic reaction to acetic acid and
formaldehyde. The latter, in an oxidizing environment reacts to formic acid [39]. In an
alternative pathway, oxidation of the hydroxy group of 3-HPA gives malonic acid [46].
In order to confirm that the reaction network was similar also with our catalyst and reaction
conditions, we carried out experiments by testing the reactivity of both succinic acid and main
by-products/intermediates, using the lowest level of H
h reaction time (Table 4).
2
WO
4
and the highest of H
2
2
O , at 1 h and 6
The experiments with succinic acid as substrate demonstrated an increase in the conversion with
time, from 1 % registered after 1 h reaction time to 7 % after 6 h. Those low values provided
evidence for the high stability of the product in the reaction mixture. Decarboxylation of succinic
acid led to propionic acid, which further can lose a molecule of CO producing acetic acid.
2
Propionic acid selectivity reached 10 % after the first hour of reaction, and consequently dropped
to 1 % at 6 h (Table 4).
In tests carried out starting from propionic acid, the acid conversion slightly increased from 18 %
to 25 % along with the increase of reaction time. The main products were glycolic and glyoxylic
acid and acetic acid. The firsts remained almost constant after 6 h at 5%, while the acetic acid
increased from 0.6 % to 3 %. It is worth noting that in the experiments with succinic acid we did
not detect the presence of glycolic/glyoxylic acid.
Experiments carried out starting from 3-HPA confirmed the oxidising pathways proposed in the
17

literature; at the end of the reaction, we detected acetic acid, malonic acid, oxalic acid and formic
acid. Formic acid selectivity attested at 7 %, while acetic acid was around 22 % and malonic acid
around 6 %.
Table 4 Oxidation of the intermediates and the by-products at T 90 °C, reactant:H
2
WO molar
4
ratio 100 and H :reactant ratio 5. Succ: succinic acid; acet: acetic acid; mal: malonic acid;
2
O
2
oxal: oxalic acid, form: formic acid, AL: α-angelica lactone; HPA: 3-hydroxypropionic acid;
gly+glyox: glycolic acid and glyoxylic acid; prop: propionic acid
Reagent
t, h Conv
S
succ
S
acet
S
mal
0
4
-
S
oxal
S
form
S
AL
S
HPA
0
0
0
0
-
S
gly+glyox
S
prop
Succinic acid
Succinic acid
Malonic acid
Malonic acid
1
6
1
6
1
6
1
6
1
1
-
0
1
0
0
0
-
0
0
10
1
7
-
3
6
1
72
70
22
20
77
74
18
-
1
17
12
3
2
6
0.4
0
-
4
-
3
-
21
0
3
3
– HPA
– HPA
-
22
25
0
6
7
-
7
-
0
-
0.1
0
8
-
-
0.1
-
0.5
-
Glycolic acid
Glycolic acid
-
0
-
-
-
0.1
0.6
-
0
6
-
-
-
-
Propionic
acid
-
0
0
0
-
0
6
-
Propionic
acid
6
1
6
25
47
60
-
0
4
3
0
0
0
4
4
-
-
-
0
0
0
4
1
-
9
0
α-angelica
lactone
52
58
27
23
10
10
α-angelica
0.3
lactone
The tests on malonic acid degradation showed that at high H
2
2
O concentration, the molecule
presented high reactivity; in this case the conversion attested around 70 %. Malonic acid
oxidized to acetic acid, which continued the pathway to glycolic acid and oxalic acid; the
analysis showed an increase of around 15 % to glycolic acid and glyoxylic acid from 1 h to 6 h.
18

The two experiments with glycolic acid as starting material demonstrated an high reactivity of
the substrate when in contact with H (conversion around 75 %). However, we were unable to
detect any key products, due to the complete degradation to volatile compounds.
Three experiments of LEV oxidation at low loading of H (entries 28-30, Table 3) showed that
2
O
2
2
O
2
under conditions of oxidant starvation LEV preferably converted to α-angelica lactone instead of
being oxidised to succinic acid or 3-HPA. The lactone is in equilibrium with the starting material
[
47]. In experiments carried out starting from the lactone (Table 4), the latter was converted
mainly to acetic acid and malonic acid.
3
.4. The regioselectivity in Baeyer-Villiger oxidation of LEV
Results shown in Table 3 demonstrate that the H WO controls the regioselectivity of the two
2
4
competitive pathways from LEV to either succinic acid or 3-HPA. In fact, by modulating the
catalyst loading we obtained a selectivity ratio succinic acid:3-HPA of > 98 %, after 6 h and at
1
0 % molar ratio of catalyst (Table 3, entry 27). Each increment of H
2
2
O also augmented the
selectivity ratio of the two-competitive kinetics, enhancing the role of catalyst in a proportional
fashion (Figure 5).
19

Figure 5 Direct correlation between the H
HPA.
2
WO
4
catalyst and the selectivity ratio succinic acid:3-
in an organic solvent free environment arranges
Based on these results, we propose that H
2
WO
4
with the carboxylic group of LEV to form an octagonal cyclic adduct by hydrogen bonding
Figure 6).
(
Figure 6 The H
2
WO catalyst interacts with LEV, forming an octagonal cyclic adduct, which due
4
to sterical hindrance, reduced the migration of the most substituted carbon, and consequently
enhances the formation of the methyl succinate
20

This conformation locks the molecule and impedes the Baeyer-Villiger migration of the
secondary carbon by steric hindrance. Indeed, we also found that this interaction locks LEV and
impedes the dehydration and the consequent cyclization to α-angelica lactone. In fact, selectivity
to the lactone decreased with increasing the catalyst loading. The major example is the reduction
from 38 % to 4 % lactone selectivity after 1 h reaction time, indicating the interaction of the
H
2
WO with LEV (Table 3, entries 1 and 19).
4
4
. Conclusions
Tungstic acid catalysed the conversion of levulinic acid to succinic acid with hydrogen peroxide
and no organic solvent. We investigated the effects of time, ratio between H and both H WO
2
O
2
2
4
and the substrate. The results confirmed an interaction mechanism of the tungstic acid with the
substrate, which indicates the regioselectivity of the catalyst and the decrease of LEV
dehydration to α-angelica lactone. Controlling the catalyst loading modulated the selectivity of
the two-competitive kinetics, achieving more than 97 % ratio between selectivities of succinic
acid and 3-HPA. We achieved a maximum conversion of 48 % and 75 % selectivity to succinic
acid with LEV:H
pathway, we ran experiments with increasing time testing each intermediate's reactivity, with
lowest amount of catalyst and a H :LEV molar ratio of 5. We demonstrated the decomposition
2
WO
4
molar ratio of 50 and H
2
2
O :LEV of 5. To investigate the decomposition
2
O
2
pathway of succinic acid, which formed acetic acid as main by-product, followed by malonic
acid and formic acid.
References
21

[
[
[
1] C. Antonetti, M. Melloni, D. Licursi, S. Fulignati, E. Ribechini, S. Rivas, J.C. Parajó, F.
Cavani, A.M. Raspolli Galletti, Appl. Catal. B Environ. 206 (2017) 364–377.
2] Z. Bin, C. Xueshan, Z. Cunshan, Y. Xiaojie, M. Haile, Z. Jing, B. Xinjie, Can. J. Chem.
Eng. (2017).
3] T. Pasini, M. Piccinini, M. Blosi, R. Bonelli, S. Albonetti, N. Dimitratos, J.A. Lopez-
Sanchez, M. Sankar, Q. He, C.J. Kiely, G.J. Hutchings, F. Cavani, Green Chem. 13 (2011)
2
091.
[
[
[
[
4] Y. Román-Leshkov, J.N. Chheda, J.A. Dumesic, Science (80-. ). 312 (2006) 1933–1937.
5] M. Kroger, U. Pruße, K.-D. Vorlop, Top. Catal. 13 (2000) 237–242.
6] H. Xie, Z.K. Zhao, Q. Wang, ChemSusChem 5 (2012) 901–905.
7] V.V. Kumar, G. Naresh, S. Deepa, P.G. Bhavani, M. Nagaraju, M. Sudhakar, K.V.R.
Chary, J. Tardio, S.K. Bhargava, A. Venugopal, "Applied Catal. A, Gen. 531 (2017) 169–
1
76.
[
8] P.P. Upare, J.M. Lee, Y.K. Hwang, D.W. Hwang, J.H. Lee, S.B. Halligudi, J.S. Hwang,
J.S. Chang, ChemSusChem 4 (2011) 1749–1752.
[
[
[
[
[
[
[
[
9] K. Manikandan, K.K. Cheralathan, Appl. Catal. A Gen. 547 (2017) 237–247.
10] Y. Guo, K. Lib, J.H. Clark, Green Chem. (2007) 839–841.
11] J. Tan, J. Cui, G. Ding, T. Deng, Y. Zhu, Y. Li, Catal. Sci. Technol. 6 (2016) 1469–1475.
12] D. Sun, Y. Takahashi, Y. Yamada, S. Sato, Appl. Catal. A Gen. 526 (2016) 62–69.
13] L.E. Manzer, Appl. Catal. A Gen. 272 (2004) 249–256.
14] J. Lewkowski, Arkivoc 2001 (2005) 17–54.
15] Á. Szabolcs, M. Molnár, G. Dibó, L.T. Mika, Green Chem. 15 (2013) 439–445.
16] C. Antonetti, D. Licursi, S. Fulignati, G. Valentini, A. Raspolli Galletti, Catalysts 6 (2016)
22

1
96.
[
[
17] F.D. Pileidis, M.M. Titirici, ChemSusChem 9 (2016) 562–582.
18] T. Werpy, G. Petersen, Top Value Added Chemicals from Biomass: Volume I -- Results
of Screening for Potential Candidates from Sugars and Synthesis Gas, 2004.
19] J.J. Bozell, G.R. Petersen, Green Chem. 12 (2010) 539.
[
[
[
[
20] J.G. Zeikus, M.K. Jain, P. Elankovan, Appl Microbiol Biotechnol 51 (1999) 545–552.
21] J. Li, W.P. Tian, X. Wang, L. Shi, Chem. Eng. J. 175 (2011) 417–422.
22] D. Mandelli, M.C.A. Van Vliet, R.A. Sheldon, U. Schuchardt, Appl. Catal. A Gen. 219
(2001) 209–213.
[
[
23] G. Chatel, C. Monnier, N. Kardos, C. Voiron, B. Andrioletti, M. Draye, Appl. Catal. A
Gen. 478 (2014) 157–164.
24] X. Yang, Y. Jiang, Y. Li, X. Xu, D. Li, K. Lin, Microporous Mesoporous Mater. 253
(2017) 40–48.
[
[
[
25] F. Cavani, K. Raabova, F. Bigi, C. Quarantelli, Chem. - A Eur. J. 16 (2010) 12962–12969.
26] L. Yu, F. Chen, Y. Ding, ChemCatChem 8 (2016) 1033–1037.
27] M. Markiton, S. Boncel, D. Janas, A. Chrobok, ACS Sustain. Chem. Eng. 5 (2017) 1685–
1
691.
28] E. Rozhko, S. Solmi, F. Cavani, A. Albini, P. Righi, D. Ravelli, J. Org. Chem. 80 (2015)
425–6431.
[
6
[
[
[
29] L. Doskočil, L. Grasset, D. Válková, M. Pekař, Fuel 134 (2014) 406–413.
30] A. Dunlop, S. Shelbert, Preparation of Succinic Acid, US2676186, 1952.
31] F. Van Der Klis, D.S. Van ES, J. Van Haveren, Succinic Acid from Biomass,
WO2012044168 A1, 2012.
23

[
32] I. Podolean, V. Kuncser, N. Gheorghe, D. Macovei, V.I. Parvulescu, S.M. Coman, Green
Chem. 15 (2013) 3077.
[
[
[
[
[
33] F. Xia, Z. Du, J. Liu, Y. Ma, J. Xu, RSC Adv. 6 (2016) 72744–72749.
34] Y. Wang, F. Vogelgsang, Y. Román-Leshkov, ChemCatChem 7 (2015) 916–920.
35] A. Caretto, A. Perosa, ACS Sustain. Chem. Eng. 1 (2013) 989–994.
36] H. Choudhary, S. Nishimura, K. Ebitani, Appl. Catal. A Gen. 458 (2013) 55–62.
37] R. Kawasumi, S. Narita, K. Miyamoto, K. Tominaga, R. Takita, M. Uchiyama, Sci. Rep. 7
(2017) 17967.
[
[
[
[
[
[
38] L. Wu, S. Dutta, M. Mascal, ChemSusChem 8 (2015) 1167–1169.
39] S. Dutta, L. Wu, M. Mascal, Green Chem. 17 (2015) 2335–2338.
40] F. Weiss, Pierre-Benite, A.Isard, No Title, US3231329, 1966.
41] USP Technologies, www.h2o2.com (2018).
42] J. Feng, X. Zhang, J. Fu, H. Chen, Catal. Commun. 110 (2018) 28–32.
43] A. Ignaczak, E. Santos, W. Schmickler, T.F. da Costa, J. Electroanal. Chem. 819 (2017)
4
10–416.
[
[
[
[
44] K.R. Enslow, A.T. Bell, Catal. Sci. Technol. 5 (2015) 2839–2847.
45] T. Dishisha, S.H. Pyo, R. Hatti-Kaul, Microb. Cell Fact. 14 (2015) 1–11.
46] R. V. Shende, J. Levee, Ind. Eng. Chem. Res. 38 (1999) 2557–2563.
47] A. Chatzidimitriou, J.Q. Bond, Green Chem. 17 (2015) 4367–4376.
24