Silica-supported chromia-titania catalysts for selective formation of lactic acid from a triose in water

Article abstract of DOI:10.1016/j.apcata.2018.11.018

A variety of silica-supported metal oxide catalysts were prepared by the incipient wetness impregnation method and were used for the conversion of dihydroxyacetone to lactic acid. A titanium oxide catalyst with Br?nsted acid sites was selective to an intermediate, pyruvaldehyde and a chromium oxide catalyst with Lewis acid sites was selective to lactic acid. The co-impregnation of chromium- and titanium oxides with different ratios accelerated the reaction rate and improved the lactic acid yield up to 80% at 130 °C. Pyridine-adsorbed Fourier-transform infrared spectroscopy indicated that the silica-supported mixed oxides had both Br?nsted acid and Lewis acid sites and the trend of the Lewis/Br?nsted ratio was close to that of selectivity to lactic acid. Diffuse reflectance UV–vis spectroscopy showed that the silica-supported chromia-titania catalyst composed of isolated Cr and Ti species in tetrahedral coordination. Kinetic analysis revealed that the two critical rate constants, pyruvaldehyde formation and lactic acid formation, for the chromia-titania catalyst were much higher than those of the titania and chromia catalysts.

Full text of DOI:10.1016/j.apcata.2018.11.018

AppliedCatalysisA,General570(2019)200–208
Contents lists available at ScienceDirect
Applied Catalysis A, General
journal homepage: www.elsevier.com/locate/apcata
Silica-supported chromia-titania catalysts for selective formation of lactic
acid from a triose in water
Atsushi Takagaki⁎,1, Hiroshi Goto, Ryuji Kikuchi, S. Ted Oyama
Department of Chemical System Engineering, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-8656, Japan
A R T I C L E I N F O
A B S T R A C T
Keywords:
A variety of silica-supported metal oxide catalysts were prepared by the incipient wetness impregnation method
and were used for the conversion of dihydroxyacetone to lactic acid. A titanium oxide catalyst with Brønsted acid
sites was selective to an intermediate, pyruvaldehyde and a chromium oxide catalyst with Lewis acid sites was
selective to lactic acid. The co-impregnation of chromium- and titanium oxides with different ratios accelerated
the reaction rate and improved the lactic acid yield up to 80% at 130 °C. Pyridine-adsorbed Fourier-transform
infrared spectroscopy indicated that the silica-supported mixed oxides had both Brønsted acid and Lewis acid
sites and the trend of the Lewis/Brønsted ratio was close to that of selectivity to lactic acid. Diffuse reflectance
UV–vis spectroscopy showed that the silica-supported chromia-titania catalyst composed of isolated Cr and Ti
species in tetrahedral coordination. Kinetic analysis revealed that the two critical rate constants, pyruvaldehyde
formation and lactic acid formation, for the chromia-titania catalyst were much higher than those of the titania
and chromia catalysts.
Lactic acid
Dihydroxyacetone
Brønsted acid
Lewis acid
Supported metal oxide catalyst
1. Introduction
A large amount of research for the catalytic production of lactic acid
or alkyl lactate by both homogeneous and heterogeneous catalysts has
Chemical production from biomass had attracted much attention
because of potential to replace existing petroleum-based industries and
to reduce carbon dioxide emissions. Lactic acid is one representative of
sugar-derived chemicals and is used as a raw material for biodegradable
plastics and solvents. The current industrial production method of lactic
acid is an anaerobic fermentation of glucose and sucrose using micro-
organisms. This process, however, has several drawbacks such as low
production rate and the use of a stoichiometric amount of base for
neutralization to keep the preferable pH level for microorganism.
Hence, the development of catalytic routes for conversion of sugars to
lactic acid is needed [1].
been conducted during recent years. As homogeneous catalysts, Al3+
and Cr3+ Lewis acids were found to be active in DHA transformation
into LA in aqueous media [6]. As heterogeneous catalysts, re-
presentative catalysts are zeolites [7–9], especially Sn incorporating
zeolites [3,10]. The Sn-β zeolite exhibited almost 100% methyl lactate
yield for DHA transformation and 40% yield for glucose transformation
in methanol media. There are several other Sn incorporating silicate
catalysts, using mesoporous silica [11] or a hybrid material composed
of Sn-Si-MCM-41 and graphite-like carbon [12]. These catalysts have
both Lewis acid sites and Brønsted acid sites. The combination of these
two kinds of acid sites was also responsible for the high activity of bulk-
type catalysts such as tin phosphate [13,14] and a niobium oxide with a
deformed orthorhombic structure [15].
One of the essential reaction steps for lactic acid (LA) production is
the conversion of triose sugars, dihydroxyacetone (DHA) and glycer-
aldehyde (GLA) to LA (Scheme 1). Both trioses can be obtained by retro-
aldol condensation of glucose [2] which is derived from cellulose, or
oxidation of glycerol which is the main byproduct of biodiesel from
triglycerides [3]. This reaction step proceeds through dehydration of
DHA or GLA to pyruvaldehyde (PA), an intermediate, followed by hy-
dration and hydride shift. The first step, DHA or GLA dehydration fol-
lowed by rearrangement, is catalyzed by both Brønsted and Lewis acid
[4–6]. PA is further transformed into LA catalyzed by Lewis acids.
In this study, a variety of silica-supported metal oxide catalysts were
prepared and were used for the transformation of DHA to LA. The
synthesis of the supported catalysts was facile and atom-efficient which
contrasts with those of Sn-β zeolite and bulk materials because the
former needs long synthesis time (days) and the latter can function only
on the surface. It was found that the supported titanium oxide and
chromium oxide were active for the transformation of DHA. The com-
bination of the two metal oxides which have both Brønsted acid and
⁎ Corresponding author.
E-mail address: atakagak@cstf.kyushu-u.ac.jp (A. Takagaki).
1 Present address: Department of Applied Chemistry, Faculty of Engineering, Kyushu University, 744 Motooka, Nishi-ku, Fukuoka, 819-0395, Japan.
https://doi.org/10.1016/j.apcata.2018.11.018
Received 16 October 2018; Received in revised form 20 November 2018; Accepted 23 November 2018
Availableonline24November2018
0926-860X/©2018ElsevierB.V.Allrightsreserved.

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Scheme 1. Reaction pathways for lactic acid formation.
Lewis acid sites accelerated both the rates of DHA conversion and LA
formation resulting in LA yield up to 80%.
2.22 cm μmol−1 for Lewis acid sites. [16]
2.4. Catalytic tests
2. Experimental
A quantity of 0.55 mmol (50 mg) of dihydroxyacetone and 50 mg of
silica-supported metal oxide catalysts was put into 3 mL water in a
reactor vessel. The reactor vessel was heated at 130 °C for 3 h in an oil
bath under stirring. After the reaction, reactant mixture was analyzed
by high-performance liquid chromatography (HPLC; LC-2000 plus,
JASCO) equipped with a differential refractive index detector (RI-2031
2.1. Chemicals
Dihydroxyacetone dimer (98%, Wako), DL-glyceraldehyde (97%,
Wako), pyruvaldehyde (40% solution in water, Sigma-Aldrich) and
lactic acid (85–92%, Kanto Chemical) were used for the reactions and
the analysis. A fumed silica, Cab-o-Sil EH5 (Cabot) was used as inert
support. Metal salts were purchased from Wako. Used as precursors
were Al(NO3)3·9H2O, Ti(SO4)2 30% solution in water, Cr(NO3)3·9H2O,
Mn(NO3)2·6H2O, Fe(NO3)3·9H2O, Ni(NO3)2·6H2O, Zn(NO3)2·6H2O, Ga
(NO3)3·nH2O, In(NO3)3·3H2O, SnCl4, WCl4, and Pb(NO3)2.
plus, JASCO) with Aminex HPX-87H column (flow rate: 0.5 mL min−1
eluent: 10 mM H SO ).
,
2
4
3. Results and discussions
3.1. Screening of silica-supported metal oxide catalysts
2.2. Catalysts preparation
The catalytic activities of a variety of supported metal oxides for
lactic acid formation were surveyed. Twelve metal oxides, Al, Ti, Cr,
Mn, Fe, Ni, Zn, Ga, In, Sn, W, and Pb oxides were chosen for the
screening test because these metal salts were reported to be effective
homogeneous catalysts to produce lactic acid from cellulose in aqueous
The silica-supported metal oxide catalysts were prepared by in-
cipient wetness impregnation and subsequent calcination. Typically, a
quantity of 1 mmol of metal salts dissolved in distilled water was im-
pregnated dropwise on 1 g of a fumed silica (Cab-o-sil, EH5) followed
by calcination in air at 500 °C for 3 h. The catalysts used were named
with their metal loading amount. For example, Cr(1.0)/SiO2 indicates
that 1.0 mmol of chromium was impregnated on 1.0 g of silica support.
media [17]. The amount of loading of metals was set to 1.0 mmol g−1
.
Fig. 1 shows the XRD patterns of silica-supported metal oxide catalysts.
A broad peak around 20° was observed for all samples, which was at-
tributed to amorphous silica. The samples of Cr, Mn, Fe, Ni, In, Sn, and
W oxides had diffraction peaks attributable to corresponding metal
oxides for each species. In contrast, the samples of Al, Ti, Zn, Ga, and Pb
oxides showed no apparent peaks, indicating that these metal species
were highly dispersed on silica or incorporated into the silica network
because these elements are known as additives of glass materials [18].
Fig. 2 shows the result of lactic acid formation from dihydrox-
yacetone using a variety of silica-supported metal oxides. The products
detected were lactic acid (LA), pyruvaldehyde (PA) and glyceraldehyde
(GLA). Among the silica-supported catalysts tested, the chromium oxide
catalyst gave the highest yield of LA of 46% with a high conversion of
dihydroxyacetone of 72% and a selectivity to LA of 64%. The titanium
oxide catalyst gave a moderate yield of LA of 29%, and the gallium
oxide catalyst of 27%. The titanium oxide catalyst also afforded an
intermediate, PA with high selectivity of 43%. The lead oxide catalyst
gave undetected products likely due to polymerization. Other metal
oxides such as Mn, Zn, and W oxides showed high conversion, but no or
little yields of LA. After the reaction using these unselective catalysts,
the solutions were colored dark brown, which is indicative of deep
polymerization. Table 1 summaries the results of crystal structure,
surface area, and the catalytic activity of a variety of silica-supported
2.3. Characterization
The prepared catalysts were characterized by X-ray diffraction
(XRD; RINT-2700, Rigaku), and diffuse reflectance UV–vis spectroscopy
(DR UV–vis; V-670, JASCO). The types of acid sites were distinguished
using pyridine adsorbed Fourier-transform infrared spectroscopy (FTIR;
FT/IR 6100, JASCO) equipped with a mercury cadmium telluride
(MCT) detector with a resolution of 4 cm−1. The sample was pressed
into disks with a radius of 1.0 cm at 15 MPa and a weight of 30 mg and
then pretreated in the cell at 150 °C in a vacuum for 1 h. The sample
was cooled to 100 °C, and a spectrum was measured. Pyridine gas was
introduced to the cell, and the sample was contacted with pyridine for
30 min at 5 Torr. After pyridine was pumped off for 30 min, a spectrum
of pyridine adsorbed sample was recorded. A difference spectrum was
obtained by subtracting the spectrum of the dehydrated sample from
the spectrum of the adsorbed sample. The amounts of Brønsted acid
sites and Lewis acid sites were determined on the basis of the integral
absorbance of the characteristic bands at 1545 cm−1 for Brønsted acid
sites and 1450 cm−1 for Lewis acid sites by using integrated molar
extinction coefficients, 1.67 cm μmol−1 for Brønsted acid sites and
201

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Fig. 2. Dihydroxyacetone conversion using a variety of silica-supported metal
oxide catalysts. Reaction conditions: Dihydroxyacetone (0.55 mmol), catalyst
(50 mg, metal loading; 1 mmol/g-support), water (3 mL), 130 °C, 3 h.
Table 1
Transformation of DHA using a variety of silica-supported metal oxide catalysts
a
.
Catalyst
Crystal phase
SBET /m2 g−1
Conv. /%
Selectivity /%
GLA
PA
LA
Al(1.0)/SiO2
Ti(1.0)/SiO2
Cr(1.0)/SiO2
Mn(1.0)/SiO2
Fe(1.0)/SiO2
Ni(1.0)/SiO2
Zn(1.0)/SiO2
Ga(1.0)/SiO2
In(1.0)/SiO2
Sn(1.0)/SiO2
W(1.0)/SiO2
Pb(1.0)/SiO2
SiO2
Amorphous
Amorphous
Cr2O3
273
292
280
314
306
318
265
237
265
304
270
186
323
54
86
72
90
31
30
89
81
43
48
90
93
21
10
5
17
43
9
33
34
64
0
2
MnO2
0
1
Fe2O3
8
41
36
0
0
NiO
12
0
3
Amorphous
Amorphous
In2O3
6
5
4
34
35
15
0
6
30
49
6
SnO2
9
WO3
6
Amorphous
0
0
0
0
60
0
a
Reaction conditions: Dihydroxyacetone (0.55 mmol), catalyst (50 mg,
metal loading: 1 mmol/g-support), water (3 mL), 130 °C, 3 h.
than the chromium oxide, resulting in a moderate yield of LA.
In contrast, the chromium oxide catalyst showed a lower conversion
rate of DHA than the titanium oxide but afforded higher selectivity to
LA. At ca. 50% conversion, the selectivity to LA was 46% for the
chromium oxide catalyst (2 h reaction), compared to 18% for the tita-
nium oxide catalyst (0.5 h reaction). After 9 h, the chromium oxide
catalyst showed a high LA selectivity of 68% at a conversion of 95%.
The titanium oxide catalyst afforded an LA selectivity of 47% at full
conversion. These results indicated that the titanium oxide catalyst
could quickly convert DHA to PA, and the chromium oxide catalyst
could smoothly transform PA to LA. Therefore, the nature of the active
sites was deduced to be different between these two catalysts.
Fig. 1. XRD patterns of silica-supported metal oxide catalysts.
metal oxide catalysts. All silica-supported metal oxides showed lower
BET surface areas than the silica support. The samples of amorphous
oxides such as Pb, Ga and Zn had much lower surface areas, indicating
that these metal species were incorporated into the silica network.
There was no correlation between the surface area and conversion of
DHA. From these results, the catalytic activity and textural properties of
silica-supported chromium oxide and titanium oxide catalysts were
further investigated.
Solid acid properties of silica-supported titanium oxide and chro-
mium oxide were evaluated by using FTIR of adsorbed pyridine. The
results are shown in Fig. 4. Two characteristic peaks attributed to
3.2. Silica-supported titanium oxide and chromium oxide catalysts
Brønsted acid sites (ca. 1545 cm−1
) and Lewis acid sites (ca.
Fig. 3 shows the time course of transformation of dihydroxyacetone
using silica-supported titanium oxide and chromium oxide catalysts. In
the initial stage of the reaction, GLA and PA were formed. With increase
of reaction time, yields of these intermediates decreased, and LA was
successively formed as a final product. The titanium oxide catalyst
exhibited a higher conversion rate of DHA with a higher concentration
of PA than the chromium oxide catalyst. However, the titanium oxide
showed a lower consumption rate of PA and lower formation rate of LA
1455 cm−1) were observed for both samples. For Ti(1.0)/SiO2, the
amount of Brønsted acid sites was 0.32 mmol g−1, and those of Lewis
acid sites was 0.24 mmol g−1 with the low ratio of Lewis/Brønsted acid
being 0.75. For Cr(1.0)/SiO2, the amount of Brønsted acid sites was
0.16 mmol g−1, and that of Lewis acid sites was 0.48 mmol g−1. The
ratio of Lewis acid sites to Brønsted acid sites was 3. Although both
samples had both Brønsted acid sites and Lewis acid sites, the titanium
202

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Fig. 3. Time course of dihydroxyacetone transformation using silica-supported titanium oxide catalyst and silica-supported chromium oxide catalyst. Reaction
conditions: Dihydroxyacetone (0.55 mmol), catalyst (50 mg, metal loading; 1 mmol/g-support), water (3 mL), 130 °C.
Fig. 5. XRD patterns of silica-supported titanium oxide catalysts.
Fig. 4. Pyridine-adsorbed FTIR spectra for silica-supported chromium oxide
and silica-supported titanium oxide. Self-supporting disk of 30 mg. 100 °C.
oxide catalyst had more Brønsted acid sites, and the chromium oxide
catalyst had more Lewis acid sites. This difference could explain the
reason for the contrast in the reaction results.
3.3. Effect of loading amount of titanium oxide and chromium oxide
The two metal oxide catalysts were further studied. The effect of
loading amount on the state of impregnated metal species and their
activities was investigated.
Fig. 5 shows XRD patterns of Ti(0.5)/SiO2, Ti(1.0)/SiO2, and Ti
(2.0)/SiO2 catalysts. No apparent peaks were observed for all samples,
indicating no formation of crystalline titanium oxide such as anatase.
Fig. 6 shows DR UV–vis spectra of Ti(0.5)/SiO2, Ti(1.0)/SiO2, and Ti
(2.0)/SiO2 catalysts. A band centered at 200 nm was observed for Ti
(0.5)/SiO2 and Ti(1.0)/SiO2 samples, which was attributed to tetra-
hedral titanium species. Additionally, broad bands centered at 250 nm
and 300 nm were seen for both samples, which were assigned to octa-
hedral titanium species and [TiO2]n clusters, respectively [19,20]. A
band centered at 330 nm, corresponding to anatase TiO2 was not ob-
served, in good agreement with the XRD patterns [21]. A sharp band
centered at 200 nm was not observed for Ti(2.0)/SiO2, suggesting that
most of the titanium species were in octahedral coordination.
Table 2 shows the results of the transformation of DHA over Ti/
SiO2. Here, the molar ratio of DHA to metal was set as 11.1 by varying
the weight of catalyst used. With increase of the loading amount of
titanium oxide, the conversion of DHA decreased. To achieve ca. 50%
conversion, Ti(0.5)/SiO2 required 0.5 h whereas Ti(2.0)/SiO2 needed
Fig. 6. DR UV–vis spectra of silica-supported titanium oxide catalysts.
1.5 h. At the same conversion, the product selectivity was nearly the
same with all the samples giving a high amount of PA. Table 2 also lists
the number of Brønsted acid and Lewis acid sites on the catalysts used
for the reaction. The number of Brønsted acid sites decreased with the
increase in loading amount. This trend was consistent with the decrease
in the conversion, suggesting that Brønsted acid sites mostly catalyzed
the conversion of DHA to PA on titanium oxide.
Fig. 7 shows XRD patterns of Cr(0.5)/SiO2, Cr(1.0)/SiO2, and Cr
(2.0)/SiO2 catalysts. A clear difference among samples was observed.
203

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Table 2
a
DHA transformation over silica-supported titanium oxide catalysts
.
Catalyst
Amount of catalyst used /mg
Brønsted acid /μmol g−1
Lewis acid /μmol g−1
Reaction time /h
Conv. /%
Selectivity /%
GLA
PA
LA
Ti(0.5)/SiO2
Ti(1.0)/SiO2
100
50
140
240
110
320
0.5
0.5
1.5
1.5
52
46
62
55
12
15
11
9
69
75
62
74
19
18
23
17
Ti(2.0)/SiO2
a
25
80
120
Reaction conditions: Dihydroxyacetone (0.55 mmol), catalyst (25–100 mg, DHA/metal = 11.1), water (3 mL), 130 °C, 0.5 or 1.5 h.
amount of Lewis acid sites with the highest Lewis/Brønsted ratio, and
afforded high selectivity to LA with the highest selectivity to PA. The
sample of Cr(2.0)/SiO2 also showed high selectivity to LA even at low
conversion. With increase of the loading amount of Cr, the rate of DHA
conversion decreased.
3.4. Co-impregnation of chromium oxide and titanium oxide
It was found that the supported titanium oxide catalyst could
quickly convert DHA to PA, and the chromium oxide catalyst could
smoothly transform PA to LA. Thus, chromium and titanium were co-
impregnated into silica support to combine their properties. Fig. 9
shows DR UV–vis spectra for the silica-supported co-impregnated oxide
catalysts. Two bands at 285 and 385 nm which were ascribed to
monochromate species in tetrahedral coordination were observed for
all catalysts. A band centered at ca. 200 nm attributed to tetrahedral
titanium species was also observed.
Fig. 7. XRD patterns of silica-supported chromium oxide catalysts.
Table 4 shows amounts of Brønsted and Lewis acid sites of silica-
supported co-impregnated oxide catalysts. Fig. 10 shows the results of
DHA transformation over the silica-supported co-impregnated oxide
catalysts. The increase of Cr contents in the co-impregnated oxide
catalysts increased both conversion of DHA and selectivity to LA with a
decrease of selectivity to PA and GLA. The Cr(0.5)-Ti(0.5)/SiO2 catalyst
gave the highest selectivity to LA (84%) at a conversion of 85% for 1.5 h
of reaction. Further increase of Cr content decreased both conversion
and selectivity to LA. Fig. 10 also plots the Lewis acid/Brønsted acid
ratio as a function of Cr content. The co-impregnated catalysts showed
high L/B ratio from 3.4 to 4.0. The trend of the L/B ratio was close to
that of selectivity to lactic acid. The Cr(0.25)-Ti(0.75)/SiO2 and Cr
(0.5)-Ti(0.5)/SiO2 catalysts had high L/B ratio which gave high se-
lectivity to lactic acid at high DHA conversion.
The samples of high loading amount, Cr(1.0)/SiO2 and Cr(2.0)/SiO2
had several peaks which could be assigned to a Cr2O3 crystal phase. In
contrast, Cr(0.5)/SiO2 showed no peaks, indicating that chromium was
highly dispersed on the silica support.
Fig. 8 shows DR UV–vis spectra of silica-supported chromium oxide
catalysts. Two distinct bands centered at 270 and 380 nm with a
shoulder around 465 nm were observed for Cr(0.5)/SiO2. These bands
were attributed to monochromate species in tetrahedral coordination
[22,23]. The additional band centered at 465 nm was ascribed to
polychromate species. At loadings higher than 1.0 mmol/g, a minor
band at approximately 600 nm appeared. The band was assginable to
crystalline chromium oxide, α-Cr2O3, which was consistent with the
XRD results.
Table 3 summaries the results of DHA transformation over the silica-
supported chromium oxide catalysts. All samples gave high selectivity
to LA, ca. 50%, which was distinctly different from the supported ti-
tanium oxide catalysts. The sample of Cr(1.0)/SiO2 had the highest
Fig. 11 shows the time course of transformation of DHA using the Cr
(0.5)-Ti(0.5)/SiO2 catalyst. As a comparison, a physical mixture of Cr
(0.5)/SiO2 and Ti(0.5)/SiO2 was used. The co-impregnated catalyst, Cr
(0.5)-Ti(0.5)/SiO2 exhibited higher conversion with higher selectivity
to LA than a physical mixture, Cr(0.5)/SiO2 + Ti(0.5)/SiO2. The con-
version reached almost 100% at 5 h with a high selectivity (or yield) to
LA of 80%, which was the highest yield in this study. Although the
physical mixture also showed high conversion and selectivity to LA, the
selectivity to LA at full conversion was 72%, which was lower than that
of the co-impregnated catalyst.
3.5. Kinetic analysis of DHA transformation using silica-supported
chromium oxide, titanium oxide, and the co-impregnated oxide catalysts
A kinetic analysis assessed the reaction pathways to LA using the co-
impregnated catalyst. Fig. 12 shows the time course of DHA conversion
and product selectivity using the Cr(1.0)/SiO2 catalyst. In the case of
DHA (Fig. 12a) both GLA and PA were primary products because their
selectivity started high and decreased. LA was a final product because
its selectivity increased steadily with time. In the case of GLA (Fig. 12b)
DHA was a primary product because its selectivity started high and
monotonically decreased. PA was a secondary product because its se-
lectivity started low, kept a constant, and then decreased. LA was a final
Fig. 8. DR UV–vis spectra of silica-supported chromium oxide catalysts.
204

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Table 3
a
DHA transformation over silica-supported chromium oxide catalysts
.
Catalyst
Amount of catalyst used /mg
Brønsted acid /μmol g−1
Lewis acid /μmol g−1
L/B ratiob
Conv. /%
Selectivity /%
GLA
PA
LA
Cr(0.5)/SiO2
Cr(1.0)/SiO2
Cr(2.0)/SiO2
100
50
70
180
480
800
2.6
3.0
1.7
53
45
28
9
18
26
22
51
50
47
160
480
11
9
25
a
b
Reaction conditions: Dihydroxyacetone (0.55 mmol), catalyst (25–100 mg, DHA/metal = 11.1), water (3 mL), 130 °C, 1.5 h. Lewis acid/Brønsted acid.
product because its selectivity increased with time. From these results,
the reaction pathway is proposed as shown in Scheme 2 [24]. The
finding that GLA was a primary product from DHA and that conversely
DHA was a primary product from GLA indicates that their inter-
conversion is reversible. Other pathways are added for formation of
undetected byproducts which were likely oligomers, the so-called hu-
mins.
Fig. 13 shows the time course of triose transformation using Cr
(1.0)/SiO2, Ti(1.0)/SiO2 and Cr(0.5)-Ti(0.5)/SiO2 catalysts. The points
are experimental results and the curves are calculated fits. Although the
use of kinetic models such as Langmuir-Hinshelwood mechanism and
rake mechanism [25] is desirable, for simplicity, the pseudo-first-order
model was used. Calculations were conducted with Polymath 6.10
using the RKF 45 method to solve the following differential equations
which were based on the literature [24].
Fig. 9. DR UV–vis spectra of silica-supported co-impregnated chromium-tita-
d[DHA]
nium oxide catalysts.
=
kP [DHA] kG [DHA] + kG [GLY] kGO [DHA][GLY]
dt
Table 4
d[GLY]
dt
= kG [DHA] kG [GLY] kGO [DHA][GLY]
Amounts of Brønsted and Lewis acid sites of silica-supported chromium-tita-
nium mixed oxide catalystsa.
d [PA]
dt
Catalyst
Brønsted acid /μmol
g−1
Lewis acid /μmol
g−1
L/B ratioa
= kP [DHA] kL [PA] kPO [PA]2
d [LA]
dt
Ti(1.0)/SiO2
240
220
100
100
160
180
320
880
400
320
480
320
1.3
4.0
3.8
3.4
3.0
1.8
= kL [PA]
Cr(0.25)-Ti(0.75)/SiO2
Cr(0.5)-Ti(0.5)/SiO2
Cr(0.75)-Ti(0.25) /SiO2
Cr(1.0)/SiO2
d[Oligomers]
= kGO [DHA][GLY] + kPO [PA]2
dt
Cr(0.5)/SiO2 + Ti(0.5)/
SiO2
Table 5 lists the reaction rate constants used for the fitting. The
constant kP describes the conversion of DHA to PA, and the constant kL
the transformation of PA to LA. The constant kG describes the
a
Lewis acid/Brønsted acid.
Fig. 10. Conversion of DHA, selectivity of products, and Lewis acid/Brønsted acid ratios as a function of chromium contents in silica-supported chromium-titanium
mixed oxides. Reaction conditions: Dihydroxyacetone (0.55 mmol), catalyst (50 mg), water (3 mL), 130 °C, 1.5 h.
205

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Fig. 11. Time course of dihydroxyacetone transformation using silica-supported chromium-titanium mixed oxide catalysts and a physical mixture of silica-supported
chromium oxide and titanium oxide catalysts. Reaction conditions: Dihydroxyacetone (0.55 mmol), catalyst (50 mg), water (3 mL), 130 °C.
−
isomerization of DHA to GLA, and the constant kG its back reaction.
The constants kGO and kPO describe the formation of oligomers from
GLA and PA, respectively. Good fits were obtained for all species.
It should be noted that the results of DHA and GLA transformation
on Cr(1.0)/SiO2 were well fitting using the same rate constants re-
gardless of the reactants, which supports the accuracy of the proposed
reaction pathway. The rate of DHA consumption for the Cr catalyst was
moderate because the constant for formation of PA, kP was small and
conversion rate with higher selectivity to LA. The constant of PA for-
mation kP for the Cr-Ti catalyst was three times larger than that for the
Cr catalyst, and twice of that for the Ti catalyst. The constant of LA
formation, kL for the Cr-Ti catalyst was 1.7 times larger than that for the
Cr catalyst and 12.5 times larger than that for the Ti catalyst.
4. Conclusions
−
the constant for the back reaction from GLA, kG was considerably
The silica-supported metal oxide catalysts were used for the trans-
formation of DHA to LA. Silica- supported chromium oxide and tita-
nium oxide were active for the reaction in water. The titanium oxide
catalyst was selective to PA and the chromium oxide catalyst to LA. The
Brønsted acid sites of the titanium oxide and the Lewis acid sites of the
chromium oxide were attributed to the conversion of DHA to PA, and
successive transformation to LA, respectively. Co-impregnated chro-
mium- and titanium oxides on the silica support resulted in isolated
chromium and titanium species in tetrahedral coordination and ac-
celerated the reaction rate with improved the LA yield of up to 80%.
larger. The large constant for the back reaction was in good agreement
with the result of the reaction using GLA as a reactant. In contrast, the
rate of DHA consumption for the Ti catalyst was high because of the
large rate constant for the formation of PA, kP as well as the small in-
fluence of the isomerization between DHA and GLA. The constant for
production of LA, kL was small, resulting in lower selectivity to LA than
the Cr catalyst.
It is apparent that the combination of Cr and Ti was effective to
increase the two critical constants, kP and kL, resulting in the higher
Fig. 12. Time course of dihydroxyacetone and glyceraldehyde transformation using silica-supported chromium oxide catalyst. Reaction conditions: a)
Dihydroxyacetone or b) glyceraldehyde (0.55 mmol), catalyst (50 mg), water (3 mL), 130 °C. Conversions are indicated by ×.
206

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Scheme 2. Proposed reaction pathway for dihydroxyacetone transformation.
Fig. 13. Time course of triose transformation using Cr(1.0)/SiO2, Ti(1.0)/SiO2 and Cr(0.5)-Ti(0.5)/SiO2 catalysts. Reaction conditions: Dihydroxyacetone or gly-
ceraldehyde (0.55 mmol), catalyst (50 mg), water (3 mL), 130 °C. The points are experimental results and the curves are calculated fits.
207

A. Takagaki et al.
AppliedCatalysisA,General570(2019)200–208
Table 5
[6] C.B. Rasrendra, B.A. Fachri, I.G.B.N. Makertihartha, S. Adisasmito, H.J. Heeres,
ChemSusChem 4 (2011) 768–777.
Rate constants for DHA transformation using silica-supported chromium oxide,
titanium oxide, and chromium-titanium oxide catalysts (units in h−1).
[7] R.M. West, M.S. Holm, S. Saravanamurugan, J. Xiong, Z. Beversdorf, E. Taarning,
C.H. Christensen, J. Catal. 269 (2010) 122–130.
−
[8] P.Y. Dapsens, C. Mondelli, J.P. Ramírez, ChemSusChem 6 (2013) 831–839.
[9] P.Y. Dapsens, M.J. Menart, C. Mondelli, J.P. Ramírez, Green Chem. 16 (2014)
589–593.
kP
kL
kG
kG
kGO
kPO
Cr(1.0)/SiO2
0.36
0.6
1.8
0.24
3
0.3
3.8
1.1
1.2
4.8
6
36
1.2
30
[10] Q. Guo, F. Fan, E.A. Pidko, W.N.P.V. van der Graaff, Z. Feng, C. Li, E.J.M. Hensen,
ChemSusChem 6 (2013) 1352–1356.
Ti(1.0)/SiO2
0.18
0.12
Cr(0.5)-Ti(0.5)/SiO2
1.2
3
[11] L. Wang, J. Zhang, X. Wang, B. Zhang, W. Ji, X. Meng, J. Li, D.S. Su, X. Bao, F.-
S. Xiao, J. Mater. Chem. A 2 (2014) 3725–3729.
[12] F. de Clippel, M. Dusselier, R. van Rompaey, P. Vanelderen, J. Dijkmans,
E. Makshina, L. Giebeler, S. Oswald, G.V. Baron, J.F.M. Denayer, P.P. Pescarmona,
P.A. Jacobs, B.F. Sels, J. Am. Chem. Soc. 134 (2012) 10089–10101.
[13] X. Wang, F. Liang, C. Huang, Y. Li, B. Chen, Catal. Sci. Technol. 5 (2015)
4410–4421.
Kinetic analysis revealed that the rate constant for PA formation for the
co-impregnated catalyst was 2–3 times higher and that of LA formation
was 1.7–12.5 times higher than those for the Cr catalyst and the Ti
catalyst.
[14] X. Wang, F. Liang, C. Huang, Y. Li, B. Chen, Catal. Sci. Technol. 6 (2016)
6551–6560.
[15] K. Nakajima, J. Hirata, M. Kim, N.K. Gupta, T. Murayama, A. Yoshida, N. Hiyoshi,
A. Fukuoka, W. Ueda, ACS Catal. 8 (2018) 283–290.
Acknowledgments
[16] C.A. Emeis, J. Catal. 141 (1993) 347–354.
[17] Y. Wang, W. Deng, B. Wang, Q. Zhang, X. Wan, Z. Tang, Y. Wang, C. Zhu, Z. Cao,
G. Wang, H. Wan, Nat. Commun. 4 (2013) 2141.
The authors thank Dr. Takayuki Iida for the UV–vis measurement.
This work was supported by JSPS KAKENHI Grant Number JP
25709077.
[18] A.K. Yadav, P. Singh, RSC Adv. 5 (2015) 67583–67609.
[19] S. Sakugawa, K. Wada, M. Inoue, J. Catal. 275 (2010) 280–287.
[20] T.A. Zepedaa, A.I. Molina, J.N.D. Leon, R.O. Estrella, S. Fuentes, G.A. Nunez,
B. Pawelec, J. Mol. Catal. A Chem. 397 (2015) 26–35.
References
[21] K.-M. Choi, T. Yokoi, T. Tatsumi, K. Kuroda, J. Mater. Chem. A 1 (2013)
2485–2494.
[22] Y. Wang, Y. Ohnishi, T. Shishido, Q. Zhang, W. Yang, Q. Guo, H. Wan, K. Takehira,
J. Catal. 220 (2003) 347–357.
[1] P. Mäki-Arvela, I.L. Simakova, T. Salmi, D.Y. Murzin, Chem. Rev. 114 (2014)
1909–1971.
[23] S. Dzwigaj, T. Shishido, J. Phys. Chem. C 112 (2008) 5803–5809.
[24] D. Jolimaitre, D. Delcroix, N. Essayem, C. Pinel, M. Besson, Catal. Sci. Technol. 8
(2018) 1349–1356.
[2] M.S. Holm, S. Saravanamurugen, E. Taarning, Science 328 (2010) 602–605.
[3] N. Razali, A.Z. Abdullah, Appl. Catal. A Gen. 543 (2017) 234–246.
[4] L. Li, C. Stroobants, K. Lin, P.A. Jacobs, B.F. Sels, P.P. Pescarmona, Green Chem. 13
(2011) 1175–1181.
[25] A. Cho, H. Kim, A. Iino, A. Takagaki, S.T. Oyama, J. Catal. 318 (2014) 151–161.
[5] Y. Hayashi, Y. Sasaki, Chem. Commun. (2005) 2716–2718.
208