Aerobic oxidation of glycerol catalyzed by M salts of PMo12O403-(M = K+, Zn2+, Cu2+, Al3+, Cr3+, Fe3+)

Article abstract of DOI:10.1016/j.apcata.2019.04.014

Glycerol is a co-product in the manufacture of biodiesel from vegetable oil and animal fat. A significant increase in biodiesel production results in adramatic overproduction of glycerol. The oxidation of glycerol to lactic acid (LA)by O₂ is considered to be a promising technology to increase the value of glycerol. Studies on the fundamental reaction mechanism vary for different catalytic systems. In this work we have chosen Keggin polyoxometalates (POMs)as an exemplary family of water-soluble hydrolytically and oxidatively-stable, fully-inorganic complexes with transition-metal and non-transition-metal counter-cations (M in the formula, MPMo₁₂O₄₀, where M = H⁺, K⁺, Zn²⁺, Cu²⁺, Al³⁺, Cr³⁺, Fe³⁺ in appropriate number to counterbalance the 3- charge of PMo₁₂O₄₀^3?). Reactions involving hydrogen peroxide, an intermediate during glycerol oxidation, produce free radicals. A kinetic analysis reveals that the radical chain length under typical conditions is about 20–30. After minimal optimization, the yield of LA was 88% at glycerol conversion > 97% (1.0 M aqueous solution of glycerol, 4.0 mM of the Al salt of PMo₁₂O₄₀^3?, henceforth “AlPMo”, 10 bar O₂ at 60 °C, reaction time 6.0 h).

Full text of DOI:10.1016/j.apcata.2019.04.014

Accepted Manuscript
Title: Aerobic oxidation of glycerol catalyzed by M salts of
3
−
PMo O
40
+ 2+ 2+ 3+ 3+ 3+
(M = K , Zn , Cu , Al , Cr , Fe )
1
2
Authors: Meilin Tao, Yiming Li, Yurii V. Geletii, Craig L.
Hill, Xiaohong Wang
PII:
S0926-860X(19)30166-8
DOI:
Reference:
https://doi.org/10.1016/j.apcata.2019.04.014
APCATA 17046
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
19 February 2019
1 April 2019
15 April 2019
Pleasecitethisarticleas:TaoM, LiY, GeletiiYV, HillCL, WangX, Aerobicoxidationof
3
−
glycerol catalyzed by M salts of PMo O
40
+ 2+ 2+ 3+ 3+ 3+
(M = K , Zn , Cu , Al , Cr , Fe ),
1
2
Applied Catalysis A, General (2019), https://doi.org/10.1016/j.apcata.2019.04.014
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.

3-
+
2+
2+
3+
Aerobic oxidation of glycerol catalyzed by M salts of PMo12
O
40 (M = K , Zn , Cu , Al ,
3
+
3+
Cr , Fe )
a,b
a
b*
b*
a*
Meilin Tao, Yiming Li, Yurii V. Geletii, Craig L Hill, and Xiaohong Wang
aKey Lab of Polyoxometalate Science of Ministry of Education, Northeast Normal University,
Changchun 130024, P. R. China, E-mail: wangxh665@nenu.edu.cn, Fax: 0086-431-85099759
bDepartment of Chemistry, Emory University, 1515 Dickey Dr., Atlanta, 30322 GA, USA
Graphical Abstract
Highlights



The mechanism of glycerol oxidation to lactic acid catalyzed by polyoxometalates (POMs)
with different cations was studied.
The reaction is a free radical chain process involving initiation, propagation and
termination.
After minimal optimization, the yield of LA was 88 % at glycerol conversion > 97 %
within 6 h.
Abstract
Glycerol is a co-product in the manufacture of biodiesel from vegetable oil and animal fat. A
significant increase in biodiesel production results in adramatic overproduction of glycerol. The
oxidation of glycerol to lactic acid (LA) by O is considered to be a promising technology to
2
increase the value of glycerol. Studies on the fundamental reaction mechanism vary for different
catalytic systems. In this work we have chosen Keggin polyoxometalates (POMs) as an exemplary
family of water-soluble hydrolytically and oxidatively-stable, fully-inorganic complexes with
1

transition-metal and non-transition-metal counter-cations (M in the formula, MPMo12O40, where
+
+
2+
2+
3+
3+
3+
M = H , K , Zn , Cu , Al , Cr , Fe in appropriate number to counterbalance the 3- charge of
403-). Reactions involving hydrogen peroxide, an intermediate during glycerol oxidation,
PMo12O
produce free radicals. A kinetic analysis reveals that the radical chain length under typical
conditions is about 20-30. After minimal optimization, the yield of LA was 88 % at glycerol
conversion > 97% (1.0 M aqueous solution of glycerol, 4.0 mM of the Al salt of PMo12
henceforth “AlPMo”, 10 bar O at 60 °C, reaction time 6.0 h).
Keywords: Glycerol oxidation; lactic acid; reaction mechanism; free radicals.
. Introduction
Global energy consumption is quickly increasing because of population and economic growth.
Fossil energy resources are sufficient to cover most of the increasing demand, but lead to
increased atmospheric CO levels. Fuel derived from biomass, or biofuel, is an appealing
O
403-,
2
1
2
renewable alternative to fossil fuels. The two most common types of biofuels are ethanol and
biodiesel [1, 2]. The production of 100 kg of biodiesel from vegetable oils and animal fats creates
about 10 kg of crude glycerol. In 2013 the domestic production of biodiesel was 9.0 million
4
barrels per day with a projected 13 million barrels per day by 2020 [3]. Thus, about 10 tons of
glycerol is produced daily creating a large surplus of this by-product. This huge amount entering
into the market significantly impacts its commodity price. Crude glycerol from biodiesel plants
has low commercial value, and its disposal and utilization has become a serious financial problem
and environmental liability for the biodiesel industry [4, 5]. Thus, biodiesel producers are seeking
new, value-added uses for voluminous triol. One of the most attractive alternatives for glycerol
utilization is as a feedstock for production added-value compounds such as bioplastic, platform
chemicals, and fuel [4]. From chemical point of view, this by-product is a highly versatile
molecule with two primary OH groups and a secondary OH group, which offer a range of
possibilities for synthetic modification. The high-added-value oxygenated products derived from
glycerol oxidation include dihydroxyacetone, hydroxypyruvic acid, glyceric acid, tartaric acid,
oxalic acid, mesoxalic acid, glyoxylic acid, glyceraldehydes, and glycolic acid [4]. Numerous
catalytic systems for O -dependent oxidation are reported in the literature [6]. Surprisingly, the
2
fundamental reaction mechanism was not a focus of the glycerol-based publications. Commonly,
o
the oxidation of organic materials by O
2
at elevated temperature (60-120 C) proceeds through
2

radical chain processes. The oxidation mechanism can be different for different catalytic systems.
Recently, radical chain mechanism was suggested for the oxidation of propylene glycol to
hydroxyacetone over the UiO-66 metal-organic framework [7]. Prati and Rossi demonstrated that
gold could catalyze the oxidation of alcohols to acid in basic solutions [8-10] through the
reduction-oxidation mechanism. High selectivity toward carboxylic acids is not a rare
phenomenon even for radical chain oxidation. A homogeneous catalytic mechanism of para-xylene
to terephthalic acid based on free radical generation being catalyzed through cobalt catalyst was
studied by Fadzil group [11]. Glycerol is the most likely oxidized through this mechanism, but no
evidence has been reported. Here, we report that glycerol oxidation also proceeds through a
radical chain mechanism.
2. Experimental
2.1. Instruments
Elemental analysis was carried out using a Leeman Plasma Spec (I) ICP-ES. IR spectra
-1
(
4000-500 cm ) were recorded in KBr discs on a Nicolet Magna 560 IR spectrometer. X-ray
diffraction (XRD) patterns of the sample were collected on a Japan Rigaku Dmax 2000 X-ray
diffractometer with Cu K radiation (λ= 0.154178 nm). The measurements were obtained in steps
α
o
o 31
of 0.04 with an account time of 0.5 s and 2θ values ranging 5-90 . P MAS NMR spectra were
obtained using a Bruker AM500 spectrometer at 202.5 MHz. UV−vis spectra were recorded with
an Agilent 8453 spectrophotometer equipped with a diode array detector using a 1.0 cm
optical-path-length quartz cuvette.
2.2. Synthesis and characterization of polyoxometalate (POM) derivatives
The MPMo catalysts were synthesized by an ion-exchange method, eqs 1 and 2, following a
procedure described previously [12]. Initially, 9.1 g (5.0 mmol) of H PMo12O40 (HPMo) was
3
dissolved in 10 mL of vigorous stirred deionized water at room temperature. Then, the appropriate
amount of metal salt was added (chlorides for K, Cr and Zn; sulfates for Al, Cu and Fe). The
solutions were stirred for 2 hours at room temperature (for CuPMo, the color was light green, and
o
for CrPMo the color was dark green). The solutions were placed in the refrigerator (-4 C) for 2
days, and the yields of crystalline material were ≈ 76 %.
3
+
+
Ⅲ
Ⅲ
Ⅲ
M + H
3
PMo12
O
40 ⟶ 3H + M1.0PMo12
O
40 (M = Fe , Al , Cr )
(1)
(2)
2
+
+
II
II
1
.5 M + H
3
PMo12
O
40 ⟶ 3H + M1.5PMo12
O
40 (M =Cu , Zn )
3

3
1
The resulting complexes were characterized by ICP-ES, FTIR, XRD, and P NMR (see Table
S1 and Fig S1-2 for details).
2.3. General Procedure
Glycerol oxidation was performed in a high-pressure stainless-steel autoclave with a
o
polytetrafluoroethylene insert (10 mL) at a constant temperature of 60 C. The autoclave was
connected to the O
stirred magnetically. Typically, 5.0 mL of 1.0 M glycerol in water was oxidized in the
presence of 4.0 mM catalyst at 60 °C at aconstant pressure of 10 bars of O . After desired
2
supply system, which kept the pressure constant. The solution was
2
time, the reactor was quickly cooled down, depressurized and the catalyst was removed by
extraction with diethyl ether. The stability of the catalysts after reaction was tested by
FTIR and XRD, which did not change compared with the fresh ones [Fig. S3]. The
remaining solution was diluted 10 times with distilled water and analyzed by high
performance liquid chromatography (HPLC) using a Shimadzu LC10A-VP chromatograph
equipped with SPB-10A UV and RID-10A R.I. detectors, and a Prevail TM C18 (4.6 mm ×
2
50 mm) column. A solution of H
2
SO
4
(0.1 % w/w) in H O/acetonitrile (1/2 v/v) was used
2
-
1
as the eluent at a flow rateof 1.0 mL min at 50 °C. The glycerol conversion, α, and the
selectivity for lactic acid (LA), SLA, were calculated using eqs 3-4:
[
Gly]
0
– [Gly]
× 100 %
α =
(
(
3)
4)
[
Gly]
0
[
LA]
S
LA
=
× 100 %
[
Gly] – [Gly]
0
3. Results and discussion
3.1. Main results
The main glycerol oxidation reaction is presented on Scheme 1 [13]. The primary products of
oxidation are dihydroxyacetone (DHA) and glyceraldehyde (GCA), which can isomerically
interconvert. The dehydration of DHA produces pyruvaldehyde (PRA), which then reacts with
water to form lactic acid (LA).
Scheme 1
The catalytic activity and product distribution using different MPOMs are summarized in Table
4

1. The total amount of the four products is close to 100 % based on consumed glycerol. The
conversion of glycerol and the product distribution are strongly dependent on counter cation of the
POM (“M” in MPOM). The activity of MPMo with non-transition-metal counter cations
correlates with a charge of these cations, with AlPMo, the complex with the highest charged
3+
counter cation, Al , having the highest conversion and selectivity to LA. KPMo has the lowest
2
+
activity. The transition metal cations definitively show their own activity. Cu completely inhibits
the reaction, i.e. there is no reaction when CuPMo is used.
Table 1
No reaction occurs in the absence of any catalyst, confirming that the ability of O
of glycerol is limited. In an essential control reaction, free radicals are generated by 2, 2’-azobis
2-amidinopropane) dihydrochloride under otherwise the same reaction conditions; in this case,
2
in oxidation
(
the reaction becomes unselective and 40 % side products are generated. This supports the idea that
the POM catalyst is involved in free radical production.
Since AlPMo showed the best results, we studied the effects of O
concentration on the product yields in glycerol oxidation (Fig. 1), as well as the reactivity of the
individual oxidized organic intermediates under N (Table 2). In the overall production of LA,
2
pressure and AlPMo
2
only the first step involves oxidation; isomerization and dehydration/hydration reactions complete
the overall transformation (Scheme 1). In order to better clarify key relative rates of processes
during complete conversion of glycerol to LA, we examined, individually, the reactions of all
three principal intermediates, DHA, GCA and PRA, under the same conditions excepting the O
2
being instead by N
Table 2
2
in the same pressure (Table 2).
Fig. 1
As seen from Tables 1-2, PRA quickly undergoes hydration to form LA. GCA isomerizes to
DHA with a moderate rate, but GCA remains in low concentration as an intermediate to high
conversions of glycerol. These data indicate that DHA is the main intermediate in glycerol
oxidation to LA. The degree of the glycerol conversion after 2 hours (Fig. 1) represents the
average reaction rate. The reaction rate increases linearly with the O
plateau. A similar trend was also observed while varying the catalyst concentration.
.2. Reaction mechanism
2
pressure, but then reaches a
3
5

It is commonly assumed that POM-catalyzed air-based oxidation processes proceed through a
general two-step reaction mechanism: oxidation of the substrate by the oxidized POM followed by
the reoxidation of the reduced POM by O
2
[14], while O only oxidizes reduced POM to fulfil one
2
cycle. In this case, the reaction between a substrate and oxidized POM under Ar should be the
same or faster than the overall catalytic process. We used UV-Vis spectroscopy to follow the
o
reduction of 4.0 mM of AlPMo by 1.0 M glycerol in water at 60 C under anaerobic conditions.
The reaction rate was very slow (only a few percent conversion after 15 hours), which is not
consistent with the two-step reaction mechanism. Alternatively, the reaction may proceed through
a radical chain mechanism with a low initiation rate but a long chain length. To confirm the radical
producer is H
2
O
2
not O
2
, the generation of H
2
O was determined by reaction with potassium
2
iodide in the presence of starch (S1). It could be seen that the solution of KI and starch changed
from white to blue gradually as in the mixture of 4.0 mM of AlPMo by 1.0 M glycerol in water at
o
6
0 C with oxygen, which determined H
2
O was generated during this reaction acting as the true
2
active oxidant.
3.2.1 Overview of alcohol oxidation via a radical chain mechanism
Two types of peroxyl radicals are involved in the chain oxidation of alcohols,
α-hydroperoxyl, >COO•(OH), and hydroperoxyl, HO
C(OH)OO• ⇄ >C=O + HO
2
• [15], which are in equilibrium, eq 5.
>
2
•
(5)
We considered two extreme cases of equilibrium, namely one completely shifted either to the
right and one shifted to the left of eq 5. The data best fit equilibrium is shifted to the left. Glycerol
has two primary and one secondary hydroxyl groups, and two different peroxyl radicals. Since the
intermediate concentrations of GCA are lower, the oxidation of the secondary alcohol is the
dominant step.
For glycerol of extra high purity, a primary radical, e.g.COO•(OH), can form very slowly via a
different pathway. Then, the C-H bond alpha to COO•(OH) cleaves forming a carbon centered
radical >C•(OH), which reacts with O
radicals are well documented to react with O
we assume that the >CHOO•(OH) radical is the chain carrying species, eq 7.
2
regenerating>COO•(OH), eqs 6-7. Most carbon-based
2
at essentially diffusion-controlled rates [16]. Here,
>
>
C•(OH) + O
2
⇄ >COO•(OH)
(6)
(7)
COO•(OH) + >CH(OH) ⟶ >C(OOH)(OH) + >C•(OH)
6

Since the rate of glycerol oxidation depends on O pressure, the termination step is probably eq
2
8.
>
COO•(OH) + >C•(OH) ⟶ >C(OOH)(OH) + >C=O
The hydroperoxide intermediate >C(OOH)(OH) is in equilibrium with H
assume that for glycerol the equilibrium in eq 9 is shifted to the right,
C(OOH)(OH) ⇄ >C=O +
During a storage of glycerol under air or in the beginning of reaction, the
peroxides, >C(OOH)(OH) and H , accumulate in the solution and become a main source of free
(8)
2
O
2
. For simplicity, we
>
H
2
O
2
(9)
2
O
2
radicals [17]. As a result, glycerol undergoes autoxidation through the branching radical-chain
o
mechanism [15]. However, at moderately elevated temperature, 50-100 C, autoxidation is
commonly slow. Transition metal complexes tend to catalytically decompose peroxides via
intermediate formation of free radicals resulting in an increase of overall oxidation rate [18, 19].
Thus, glycerol oxidation via radical chain mechanism is not an unexpected process.
We applied several tests to confirm the presence of the radical chain pathway and then showed
2
+
that this pathway is dominant. The efficient inhibition of glycerol oxidation by Cu ions (6.0 mM
in 4.0 mM CuPMo, Table 1, entry 5) through the reactions in eqs 10-11 confirm the radical chain
nature of process [20].
2
+
+
+
7
-1 -1
>
C•(OH) + Cu ⟶ >C=O + Cu + H
k
10≈5×10 M s at RT
(10)
(11)
Cu+
+
¼ O
+ H ⟶ Cu
+
2+
+
½ H O
2
2
We also examined the effect of the water-soluble inhibitor, sodium phenolsulfonate, on this
POM-initiated O -based glycerol oxidation. The kinetics of this oxidation catalyzed by AlPMo in
2
the presence of 5.0 mM inhibitor is shown on Fig. 2. An induction period of 1 h is clearly seen. In
such cases, the initiation rate is usually calculated as 2×[inhibitor]/(induction period). Under the
-6
-1
conditions in Fig. 2, the calculated initiation reaction rate is around 3×10 Ms . The maximum
-5
-1
rate of glycerol consumption in Fig. 2 is about 7×10 Ms , indicating that the radical chain
length L = (reaction rate)/(initiation rate) is about 20.
Fig. 2
3.2.2 Radical formation initiated by MPMo
To check the hypothesis that MPMo can produce free radicals, we used Pyrogallol Red (PGR)
as a model radical scavenger; the UV-Vis spectrum of this indicator dye changes considerably
7

upon oxidation (Fig. 3). Earlier, PGR was used as a probe molecule to assess the antioxidant
activity of peroxynitrite (O=NO-OH) to generate free HO• and NO • radicals via homolytic
2
cleavage of the O-O bond [21, 22]. Here, we note that peroxynitrite is a very active oxygen atom
donor, a typical property of peroxides. However, PGR does not react with peroxynitrite as an
oxygen atom donor. In addition, we confirmed that each PGR molecule scavenges two radicals
using ABTS• as a model radical (Fig S4).
Fig. 3
Free radicals are formed when H
2
O , AlPMo and glycerol are all present in the solution. The
2
solution of 50 µM PGR in slightly acidified water (pH ~ 2) was placed in a thermostat cell of a
o
spectrophotometer and the kinetics of PGR consumption were measured at 60 C (Fig. 3). PGR
4
-1
-1
absorbs light at 468 nm with ɛ468 = 2×10 M cm (at pH > 7, λmax = 548 nm). No changes were
observed for the first 10 min. Then, when 5.0 µL of H (30% w/w, 10.0 M in molar
2
O
2
concentration) was injected into the cell and the spectra were recorded every 30 s. Again, no
significant spectral changes were observed. However, after addition of 5.0 µL of aqueous AlPMo
solution (150 mM), the absorbance at 468 nm and 280 nm decreased with time, while the
absorbance at 320 nm increased (Fig. S5). The presence of an isosbestic point indicates the
quantitative conversion of PGR to a single product with an absorbance in the range 350- 410 nm.
PGR was completely consumed after 7-8 min (Fig. 4), confirming that the mixture of AlPMo and
H
2
O
2
in the presence of glycerol generates free radicals.
Fig. 4
In order to determine the rate of radical generation, the PGR spectral changes were recorded
after addition of the desired quantity of AAPH (see Table 1 and Fig. 4). The decrease of
absorbance at 468 nm with time is shown in Fig. 4. The initial rate of PGR consumption in the
-7
-1 -1
presence of 10 mM AAPH is about 1×10 M s . AAPH undergoes thermal decomposition to
-6
-1
o
free radicals with the rate constant k ~ 3×10 s at 60 C [23]. Under conditions in Fig. 4, the
i
-7
-1 -1
expected rate of radical generation is 1×10 M s , which is in excellent agreement with the
experimental number.
3
+
2+
+
The rate of radical generation depends on the nature of cations, Al > Zn > K , as well as
3
+
3+
2+
3+
3+
Fe > Cr > Cu (Fig. 5). For the transition metal cations, the order of activity, Fe > Cr , is
directly related to their catalytic H
2
O
2
decomposition activity [24]. Copper ions are well known to
8

be efficient inhibitors of alcohol oxidation by a radical chain mechanism [18]. In control
experiments, we looked at the kinetics of PGR consumption in the presence of Al (SO
2
)
4 3
or ZnSO
4
3
+
2+
and 10 mM H
2
O but without POM. No acceleration of PGR consumption by Al or Zn was
2
observed. So far, the role of non-transition metal cations remains unclear. If PRG and glycerol are
both present, the rate of PGR consumption drops by about 20 %. In this case, about 20 % of all
generated radicals react with glycerol, implying that these radicals react with glycerol about 5
orders of magnitude slower than with PGR. For the less selective HO
2
• radical, the ratio of the
-1 -1
5
-1 -1
4
reaction rate constants for 2-propanol (~3 M s ) [15] and Trolox (10 M s ) is ≈ 3×10 [25].
Fig. 5
3.2.3 General discussion
In order to confirm that the radical chain process is dominant, we built a simplified kinetic
model for the catalytic oxidation of glycerol by O
2
. First, we assumed that the main chain carrier
is HO •. However, this model was not consistent with the experimental data. Therefore,
2
the >COO•(OH) radical was considered as the main chain carrier.
The rate constant for the reaction in eq 7 was taken to be, similar to that of 2-propanol: k ≈ 1
7
-1 -1
, i.e. k
, usually approaches 109
M s [25]. Again, the reaction of carbon centered radicals with O
2
6
-1 -1
9
-1 -1
M s . The same rate constant is assigned for the termination step, k
8
= 10 M s . POMs are
known to react with hydrogen peroxide to form peroxy-species [26]. For simplicity we ignore the
3-
speciation of PMo12
O
40 at pH 1.6-2.0, and write the reaction in eq 12:
O + PMo12
39(OO) , a two-electron oxidizing agent, is expected to react with glycerol, eq 13
3
-
39(OO)3-
H
2
O
2
+ PMo12
PMo12
PMo12 39(OO) + >CH(OH) ⟶ PMo12
We assume that a small fraction of PMo12
O
40 ⟶ H
2
O
(12)
3
-
O
3
-
3-
O
O
40 + >C=O + H
2
O
(13)
3
-
O
39(OO) reacts through a one-electron mechanism
oxidizing two glycerol molecules to two radicals, eqs 14-15, thus initiating the radical chain
process
3
-
+
2-
PMo12
O
O
39(OO) + >CH(OH) + H ⟶ PMo12
O
39(O) + >C•(OH) + H
2
O
(14)
(15)
2-
40 + >C•(OH) + H+
3-
PMo12
39(O) + >CH(OH) ⟶ PMo12
O
The reaction rate constants k13 and k14 were the variable parameters, and k15 was assumed to be
6
-1 -1
3+
fast, k15 > 10 M s . The dehydration reaction is known to be catalyzed by Al [13] and is written
for simplicity as eq 16
9

3+
3+
>
C=O + Al ⟶ PRA + Al
(16)
Similar reactions should be written for oxidation of the primary alcohol groups in glycerol. To
avoid the complexity, we simply considered that the reaction product GCA is in equilibrium
with >C=O (DHA), eq 17
GCA = >C=O
(17)
The transformation of PRA to LA was described by a simple pseudo-first-order reaction in eq
18
d[LA]/dt = - k18 [PRA]
(18)
The reaction rate constants in eqs 16-18 were the variable parameters. The concentration of O
2
o
at 60 C and at pressure10 bar is 8 mM [27]. Since the initiation of radicals involves the reaction
product H
concentration. The theoretical curves of glycerol consumption barely exhibit an induction period
at 1 mM [H . In our calculations, we used [H = 1 mM, which is a reasonable value for
2
O
2
, the shape of kinetic curves in the beginning of the reaction depends on initial H
2
O
2
2
O
]
2 o
2
O
]
2 o
not freshly purified glycerol [17]. In order to choose the best set of parameters, the data in Fig. 1,
403-], were also
the dependences of glycerol conversion and product yields on [O ] and [PMo12O
2
taken into account. This kinetic model is over-parameterized and does not give a single set of
variable parameters. We were mostly interested in whether the radical chain process is feasible at a
low rate of radical initiation. The radical chain length, L, is the ratio of the propagation and
initiation rates, and was estimated earlier to be ≈ 20. Therefore, only the sets of parameters with L
in the range 20-30 were considered. In addition, we calculated the product yields after 2 hours at
different concentrations of O and catalyst, and compared them with experimental data (Fig. 1).
2
The fitting of AlPMo data with the best sets of variable parameters are presented in Fig. 6. The
kinetic modelling confirmed that all available experimental data can be described by a classical
branching radical chain mechanism, in which the peroxide intermediate generates free radicals in
the presence of a transition metal complex.
Fig. 6
Contrary to a common belief that radical reactions are not selective, in this work a selectivity
of 92 % for lactic acid formation was achieved at 95 % glycerol conversion. This can be explained
by several factors. Aldehyde, PRA, forms LA quickly and selectively (entry 3 in the Table 2), and
PRA concentration remains low in the course of the reaction. Carboxylic acid, including LA, has a
10

very low oxidizability. The first step, the oxidation of glycerol is not selective, it gives two
products, ketone (DHA) and aldehyde (GCA). Hypothetically, the primary alcohol groups in DHA
could be further oxidized to give a broad spectrum of products. At the same time, GCA reversibly
isomerizes to DHA faster than it is further oxidized (entry 2 in the Table 2). Formally, the pathway
GLY ⟶ GCA ⟶ DHA makes an unselective oxidation of glycerol to a single selective GLY ⟶
DHA pathway. The isomerisation GCA ⟶ DHA is the key reaction for the high selectivity. The
3
+
DHA ⟶ PRA reaction is known to be catalysed by Al cation [13]. This is likely a key function of
that counter cation in our case. Thus, the efficient generation of free radicals from H by LPMo
O
2 2
combined with a relatively high efficiency GCA ⟶ DHA isomerization make the overall process
fast and selective.
4. Conclusions
+
+
2+
2+
A series of homogeneous phosphomolybdate catalysts MPMo12O40 (M =H , K , Zn , Cu ,
3
+
3+
3+
Al , Cr , Fe ) have been synthesized and characterized by several techniques. The POM
complexes with different counter cations show different catalytic activities. Of the 7 complexes
reported here, AlPMo had the highest efficiency: 92 % for LA at 95 % glycerol conversion within
2
+
6
h. Inhibition of the reaction by AAPH and by Cu (from CuPMo) and the generation of free
radicals from H catalyzed by MPMo confirm the branching-radical-chain mechanism of the
2
O
2
process. Despite the involvement of free radicals, the reaction is selective due to isomerization of
GCA to DHA, followed by catalytic dehydration of DHA to PRA. This work demonstrates that a
classical radical chain oxidation can result in the selective oxidation of glycerol to lactic acid.
Acknowledgements
We thank the National Natural Science Foundation of China (51578119) and Jilin Provincial
Science and Technology Department (20180414069GH), and the Department of Energy, Office of
Basic Energy Sciences, Solar Photochemistry program, grant number DE-FG02-07ER15906 for
support of this research.
Notes and references
[1] E.F. Aransiola, T.V. Ojumu, O.O. Oyekola, T.F. Madzimbamuto, D.I.O. Ikhu-Omoregbe,
Biomass and Bioenergy.61 (2014) 276-297.
[
2] M.H. Hassan, M.A. Kalam, Procedia Eng.56 (2013) 39-53.
3] http://www.eia.gov/forecasts/aeo/pdf/0383(2015).pdf.
[
11

[4] N. Vivek, R. Sindhu, A. Madhavan, A.J. Anju, E. Castro, V. Faraco, A. Pandey, P. Binod,
Bioresour Technol.239 (2017) 507-517.
[
[
[
5] F. Yang, M.A. Hanna, R. Sun, Biotechnol Biofuels.5(2012) 13.
6] J.C. Beltran-Prieto, K. Kolomaznik, J. Pecha, Aust J Chem.66 (2013) 511-521.
7] V.V. Torbina, N.S. Nedoseykina, I.D. Ivanchikova, O.A. Kholdeeva, O.V. Vodyankina, Catal
Today. (2018) DOI: 10.1016/j.cattod.2018.11.063.
[
[
[
[
[
8] G.J. Hutchings, M. Haruta, ApplCatal A: Gen. 291 (2005) 2-5.
9] L. Prati, M. Rossi, J Catal. 176 (1998) 552-560.
10] F. Porta, L. Prati, M. Rossi, S. Coluccia, G. Martra, Catal Today. 61 (2000) 165-172.
11]N.A.M. Fadzil, M.H.A. Rahim, G.P. Maniam, Chin J Catal. 35 (2014) 1641-1652.
12] M.L. Tao, D. Zhang, X. Deng, X.Y. Li, J.Y. Shi, X.H. Wang, ChemCommun.52 (2016)
3
332-3335.
13] C.B. Rasrendra, B.A. Fachri, I.G. Makertihartha, S. Adisasmito, H.J. Heeres, Chemsuschem.
(2011) 768-777.
[
4
[
[
[
[
[
[
[
[
14] S.S. Wang, G.Y. Yang, Chem. Rev. 115 (2015) 4893-4962.
15] E.T. Denisov, T.G. Denisova, Petroleum Chemistry. 46 (2006) 305-313.
16] D.A. Pratt, K.A. Tallman, N.A. Porter, AccChem Res. 44 (2011) 458-467.
17] S.R. Kumar, J.A. Imlay, J Bacteriol.195 (2013) 4569-4579.
18] K.R. Naqvi, J. Marsh, V. Chechik, Dalton Trans.43 (2014) 4745-4751.
19] A. Phaniendra, D.B. Jestadi, L. Periyasamy, Indian J ClinBiochem.30 (2015) 11-26.
20] B.G.V. Buxton, J.C. Green, J. Chem. Soc. Faraday Trans. 1.74 (1978) 697-714.
21] E. Hugo, J. Reyes, E. Montupil, R. Bridi, E. Lissi, A. Denicola, M.A. Rubio, C.
Lopez-Alarcon, Molecules.20 (2015) 10582-10593.
22] E. Atala, G. Velasquez, C. Vergara, C. Mardones, J. Reyes, R.A. Tapia, F. Quina, M.A.
[
Mendes, H. Speisky, E. Lissi, M.S. Ureta-Zanartu, A. Aspee,C. Lopez-Alarcon, J PhysChem
B.117 (2013) 4870-4879.
[
23] A.V. Ivanova, E.L. Gerasimova, E.R. Gazizullina, A.N. Kozitsina, A.I. Matern,Russ Chem
Bull.65 (2016) 419-424.
24] F. Magalhães, M.C. Pereira, S.E.C. Botrel, J.D. Fabris, W.A. Macedo, R. Mendonça, R.M.
Lago, L.C.A. Oliveira, Appl Catal A: Gen. 332 (2007) 115-123.
[
12

[
[
[
25] B.H. J.Bielski, D.E. Cabelli, R.L. Arudi, A.B. Ross, J Phys Chem. 14 (1985) 1041-1100.
26] V. Conte, B. Floris, Dalton Trans. 40 (2011) 1419-1436.
27] D. Tromans, Hydrometallurgy. 48 (1998) 327-342.
Fig. 1. The effect of O
2
pressure and initial [AlPMo] on glycerol consumption and product yields
; conditions: glycerol (1 M) and AlPMo (4 mM). (b)
(5 bar). Code:
o
after 2 hours at 60 C. (a) Dependence on O
2
Dependence on catalyst (AlPMo) concentration; conditions: glycerol (1 M) and O
2
glycerol, black; DHA, red; GCA, blue; PRA, green; and LA, pink. The open circles are
experimental and solid circles are calculated based on the proposed kinetic model.
13

Fig. 2. Study of phenolsulfonate inhibition during the glycerol oxidation process. Glycerol
consumption, black; yields of DHA, red; GCA, blue; PRA, green; LA, pink. Conditions: 1 M
o
glycerol, 4 mM of AlPMo, 10 bar O
2
, 60 C plus sodium phenolsulfonate (5 mM) as an inhibitor.
14

Fig. 3. UV-visible spectra of PGR under varying conditions: initial spectra of 50 µM PGR, black;
after 10 min, red; 10 min after 10 mM H addition, blue; after 0.25 mM AlPMo addition for 5
2
O
2
min, green; 10 min, pink; and 15 min, brown. Conditions: 50 mM sodium sulfate buffer (pH 1.6)
o
at 60 C.
15

Fig. 4. Oxidation of 50 µM PGR by 10 mM H
2
2
O in the presence of 1.0 M glycerol, 0.125 mM
FePMo, black; 0.125 mM AlPMo, red; 0.125 mM ZnPMo, blue; and in the presence of 0.5 mM
AAPH, green ; 10 mM AAPH, pink. Reaction conditions: 50 mM sodium sulfate buffer pH 1.6,
o
6
0 C.
16

Fig. 5. Oxidation of 50 µM PGR by 10 mM H
2
2
O in the presence of 0.125 mM FePMo, black;
AlPMo, red; and CrPMo, blue. Reaction conditions: 50 mM sodium sulfate buffer (pH=1.6), 60
oC.
17

o
Fig. 6. The kinetics of 1.0 M glycerol oxidation by O (10 bar) at 60 C catalyzed by 4 mM
2
AlPMo. Open circles are experimental data and solid lines are the theoretical curves. Code:
glycerol, black; DHA, red; GCA, blue; PRA, green; LA, pink.
Scheme 1. Glycerol oxidation products.
18

Table 1. Oxidation of glycerol by O in the presence of homogeneous MPMo catalysts.
2
Yields, %
DHA
14.5
13.9
8.6
Mass
Entry Catalyst
α, %
GCA
4.1
4.9
5.9
4.5
0
PRA
6.2
4.5
3.0
1.6
0
LA
25
balance (%)
95.8
1
2
3
4
5
6
7
AlPMo
FePMo
CrPMo
ZnPMo
CuPMo
KPMo
52
45
33
16
0
19.8 95.8
12.9 92.1
5.4
3.6
0
94.4
0
12
44
4.8
4.2
14.9
0.7
8.4
1.4
4.5
92.5
99.5
HPMo
16.0
Scheme 2. Structures of ABTS and PGR
19

8
9
none
2
0
8
0
4
0
0
0
0
AAPH a
20
60.0
Reaction conditions: 5.0 mL of 1.0 M aqueous solution of glycerol, 4.0 mM of POM, 10
bar O at 60 °C, reaction time 2.0 h.
2
aIn the presence of a radical initiator, 4.0mM of 2,2’-azobis(2-amidinopropane)
dihydrochloride (AAPH).
Table 2. Transformation of the main reaction intermediates catalyzed by AlPMo
Yields %
Substrates
α , %
DHA GCA
PRA
25
9
LA
15
5
DHA
GCA
PRA
44
41
91
--
1
25
--
--
--
--
87
Reaction conditions: 5.0 mL of 1.0 M aqueous solution of substrate, 4.0 mM of
POM, 10 bar N at 60 °C, reaction time 0.5 h.
2
20