Mesoporous siliconiobium phosphate as a pure bransted acid catalyst with excellent performance for the dehydration of glycerol to acrolein

Article abstract of DOI:10.1002/cssc.201200587

The development of solid acid catalysts that contain a high density of Bransted acid sites with suitable acidity, as well as a long lifetime, is one of great challenges for the efficient dehydration of glycerol to acrolein. Herein, we report on a mesoporous siliconiobium phosphate (NbPSi-0.5) composite, which is a promising solid Bransted acid that is a potential candidate for such a high-performance catalyst. A variety of characterization results confirm that NbPSi-0.5 contains nearly pure Bransted acid sites and has well-defined large mesopores. In addition, NbPSi-0.5 contains a similar amount of acid sites and exhibits weaker acidity than that of the highly acidic niobium phosphate and HZSM-5 zeolite. NbPSi-0.5 is quite stable and has a high activity for the dehydration of glycerol. The stability of NbPSi-0.5 is about three times higher than that of the reported catalyst. The significantly enhanced catalytic performance of NbPSi-0.5 can be attributed to 1) nearly pure Bransted acidity, which suppresses side reactions that lead to coke formation; 2) a significant reduction of pore blocking due to the mesopores; and 3) a decrease in the amount and oxidation temperature of coke.

Full text of DOI:10.1002/cssc.201200587

DOI: 10.1002/cssc.201200587
Mesoporous Siliconiobium Phosphate as a Pure Brønsted
Acid Catalyst with Excellent Performance for the
Dehydration of Glycerol to Acrolein
Youngbo Choi, Dae Sung Park, Hyeong Jin Yun, Jayeon Baek, Danim Yun, and
Jongheop Yi*[a]
The development of solid acid catalysts that contain a high
density of Brønsted acid sites with suitable acidity, as well as
a long lifetime, is one of great challenges for the efficient de-
hydration of glycerol to acrolein. Herein, we report on a meso-
porous siliconiobium phosphate (NbPSi-0.5) composite, which
is a promising solid Brønsted acid that is a potential candidate
for such a high-performance catalyst. A variety of characteriza-
tion results confirm that NbPSi-0.5 contains nearly pure Brøn-
sted acid sites and has well-defined large mesopores. In addi-
tion, NbPSi-0.5 contains a similar amount of acid sites and ex-
hibits weaker acidity than that of the highly acidic niobium
phosphate and HZSM-5 zeolite. NbPSi-0.5 is quite stable and
has a high activity for the dehydration of glycerol. The stability
of NbPSi-0.5 is about three times higher than that of the re-
ported catalyst. The significantly enhanced catalytic perfor-
mance of NbPSi-0.5 can be attributed to 1) nearly pure Brøn-
sted acidity, which suppresses side reactions that lead to coke
formation; 2) a significant reduction of pore blocking due to
the mesopores; and 3) a decrease in the amount and oxidation
temperature of coke.
Introduction
Solid acids play key roles in the petrochemical industry by cat-
alyzing important conversion reactions of hydrocarbons. The
recent increase in efforts to utilize biomass as a renewable re-
source are expanding the importance and utility of solid acid
catalysts, since acid-catalyzed reactions, such as dehydration
and hydrogenolysis, can provide efficient pathways for the
conversion of biomass into valuable products.[1,2] However, the
oxygen content of biobased chemicals is typically higher than
that of the corresponding hydrocarbons.[3] This places new re-
quirements on the use of solid acid catalysts for biomass con-
version.[4,5]
dration reactions of other carbohydrates that are similar in
structure to glycerol. Recently, Weingarten et al. reported that
acid catalysts with a high ratio of Brønsted to Lewis acid sites
were much more selective for the preparation of furfural by
the dehydration of xylose.[5]
Second, catalysts with medium acidity (À8.0<H0 <3.0) per-
form well in the selective conversion of glycerol into acrolein.[8]
However, some contradictory results have appeared, reporting
that strong[14] or weak[22] acid catalysts show the best selectivi-
ty for acrolein, thus making the selection of catalysts with the
appropriate acidity somewhat controversial.
The dehydration of glycerol, which is a major byproduct in
the manufacture of biodiesel, is one of the intensively studied
examples in which solid acid catalysts can be utilized for the
transformation of biobased feedstock. The acid-catalyzed dehy-
dration of glycerol leads to the production of acrolein, which is
a versatile chemical in the production of acrylic acid and poly-
esters as well as in the agrochemical industry.[6–8]
Third, overcoming the low stability of catalysts would be
a significant breakthrough for practical applications. Although
many of the existing acid catalysts show high catalytic activi-
ties in the early stages of reactions, they frequently become
deactivated due to the deposition of coke.[6] Various ap-
proaches have been proposed to increase the stability of such
catalysts. These include suppressing the rate of coking by
adding O2 or H2 to the feed flow[14,21,23,24] and modifying the
catalysts with promoters to change the acid properties of the
catalysts.[16] However, these methods have not been altogether
successful and clearly additional improvements are needed.
For this reaction, numerous heterogeneous acid catalysts, in-
cluding
zeolites,[9–13] heteropolyacids,[14–18] and metal
oxides,[19–21] have been investigated. Previous studies reported
important implications of this, including 1) the types of acid
sites, 2) the acid strength, and 3) the stability of catalysts,
which are important issues in the preparation of efficient acid
catalysts.
[a] Y. Choi, D. S. Park, Dr. H. J. Yun, J. Baek, D. Yun, Prof. J. Yi
World Class University Program of Chemical Convergence
for Energy & Environment, Institute of Chemical Processes
School of Chemical and Biological Engineering
Seoul National University, Seoul 151-742 (Republic of Korea)
Fax: (+82)2-885-6670
It is generally accepted that Brønsted acid sites are superior
to Lewis acid sites for the selective conversion of glycerol into
acrolein.[8,10,14] Furthermore, it has been reported that a de-
crease in the number of Lewis acid sites has a significant effect
on the stability of catalysts.[19] The beneficial effect of Brønsted
acidity on catalytic efficiency has been also reported in dehy-
E-mail: jyi@snu.ac.kr
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201200587.
2460
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2012, 5, 2460 – 2468

Mesoporous Siliconiobium Phosphate
Considering these issues, there-
fore, the design of acid catalysts
with a higher density of Brøn-
sted acid sites with suitable acid-
ity, as well as high resistance to
deactivation, would be highly
desirable.
Herein, we report on the syn-
thesis and characterization of
a
mesoporous siliconiobium
phosphate (NbPSi-0.5) compo-
site, and applications as a promis-
ing solid Brønsted acid catalyst
for the efficient and stable dehy-
dration of glycerol into acrolein.
Niobium phosphate (NbP) can
be a good candidate as a base
material for the catalyst, as it has
advantages of high acidity and
good water tolerance.[25,26] These
properties can be beneficial to
the dehydration of glycerol,
which generally involves a high
concentration of water. However,
Figure 1. a) N2 adsorption–desorption isotherms and b) Barrett–Joyner–Halenda (BJH) pore size distribution curves
!
~
&
*
for NbPSi-0.5 ( ), NbPSi-1 ( ), NbP ( ), and HZSM-5 ( ). The isotherms for NbPSi-0.5, NbPSi-1, and NbP have been
offset by 240, 130, and 70 cm3 gÀ1, respectively, and the pore size distribution curves for NbPSi-0.5, NbPSi-1, and
NbP have been offset by 0.25, 0.19, and 0.04 cm3 gÀ1 nmÀ1 along the vertical axis, respectively.
NbP, as is the case for most solid acids, retains both Lewis and Results and Discussion
Brønsted acidity.[27,28] This problem was approached by com-
Structural characterization
bining NbP with silicate to form a homogeneous composite.
Generally, Lewis acid sites of NbP originate from coordinatively
unsaturated Nb5+ ions, and these are probably related to the
NbO4 tetrahedral structure.[29,30] It is assumed that the added Si
atoms, which form tetra-coordinate bonds, replace Nb atoms
in lower coordination environments, leading to a decrease in
the density of Lewis acid sites.
The N2 adsorption–desorption isotherms and pore size distri-
bution of NbPSi-0.5 (atomic ratio of Nb/P=0.5; atomic ratio of
Si/P=0.5), NbP (atomic ratio of Nb/P=1), and HZSM-5 are
shown in Figure 1. For comparison, the isotherm and pore size
distribution of NbPSi-1 (atomic ratio of Nb/P=1; atomic ratio
of Si/P=0.5) are also included. NbPSi-1 was prepared by proce-
dures identical to those of NbPSi-0.5 except for the Nb/P ratio.
The isotherm of NbPSi-0.5 was similar to a type IV isotherm
and exhibited an H1-type hysteresis loop at relative pressures
(P/P0) ranging from 0.6–0.8. This indicates that NbPSi-0.5 has
a well-developed mesoporous structure with large pores. The
average pore size of NbPSi-0.5, calculated from the BJH ad-
sorption branch, was about 7.2 nm and its surface area was
about 238 m2 gÀ1 (Table 1). On the other hand, NbP and HZSM-
5 displayed isotherms similar to type II with no significant hys-
teresis loops, consistent with the microporous structure of the
reference catalysts.[33] The average pore sizes of NbP and
Importantly, the synthesis of NbP-based materials with mes-
ostructures is highly challenging because the mesostructures
have beneficial effects on catalytic performance.[31,32] In this
study, mesoporous NbPSi-0.5 with large pores was successfully
prepared by means of a solvothermal method combined with
solvent evaporation. In catalytic tests for the dehydration of
glycerol to acrolein, NbPSi-0.5 showed excellent activity and
substantially enhanced stability in comparison with NbP, com-
mercial zeolite (HZSM-5), and the reported long-life catalyst.
Rationales for the high catalytic performance of NbPSi-0.5 are
also discussed.
Table 1. Textural properties, atomic concentration of elements, and uptake of NH3 for NbPSi-0.5, NbPSi-1, NbP, and HZSM-5.
[a]
[b]
[c]
[c]
[d]
Solid acid
SBET
[m2 gÀ1
Dav
[nm]
Smicro
[m2 gÀ1
Vmicro
[cm3 gÀ1
Vtotal
[cm3 gÀ1
Bulk atomic ratios[e]
Surface atomic ratios[f]
NH3 uptake[g]
[mmolgÀ1
]
]
]
]
]
Nb/P
Si/P
Nb/P
Si/P
NbPSi-0.5
NbPSi-1
NbP
238 (56)[h]
205 (36)[h]
370 (29)[h]
441 (20)[h]
7.2 (12.3)[h]
– (À)[h]
– (0.001)[h] 0.545 (0.172)[h] 0.53
– (0.002)[h] 0.329 (0.079)[h] 0.98
0.51
0.47
–
0.48
1.04
0.97
–
0.54
0.45
–
41.5
43.7
46.2
45.6
4.1 (14.2)[h]
<1.5 (19)[h]
<1.5 (6.4)[h]
– (À)[h]
246 (10)[h] 0.092 (0.004)[h] 0.301 (0.054)[h] 1.05
350 (6)[h]
HZSM-5
0.164 (0.003)[h] 0.273 (0.038)[h]
–
–
–
[a] The Brunauer–Emmett–Teller (BET) surface area (SBET) was calculated from the N2 adsorption branch in the relative pressure range from 0.05 to 0.25.
[b] The average pore size (Dav) was determined by the BJH method. [c] The micropore area (Smicro) and volume (Vmicro) were obtained by the t-plot method.
[d] Total pore volume (Vtotal) was evaluated at a relative pressure of 0.99. [e] Determined by inductively coupled plasma atomic emission spectroscopy (ICP-
AES). [f] Determined by X-ray photoelectron spectroscopy (XPS). [g] Estimated from NH3 pulse chemisorption. [h] Values in parentheses were measured
after reaction for 8 h at a gas hourly space velocity (GHSV) of 14940 mLgcatalystÀ1 hÀ1
.
ChemSusChem 2012, 5, 2460 – 2468
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
2461

J. Yi et al.
HZSM-5 were less than 1.5 nm. In the case of NbPSi-1, the hys-
teresis loop was similar to that of NbPSi-0.5. However, the hys-
teresis loop of NbPSi-1 was broader and its lower end occurred
at a P/P0 value of about 0.4, indicating that NbPSi-1 has smaller
pore sizes with less uniform distribution.
terials.[34,35] In NbP, the terminal PÀOH and NbÀOH groups
function as Brønsted acids and the PÀOH group is a stronger
Brønsted acid than the NbÀOH group.[26] On the other hand,
Lewis acidity is attributed to coordinatively unsaturated Nb5+
sites.[29] These coordinatively unsaturated Nb5+ sites are related
to NbO4 units in view of the fact that in Nb2O5, NbO4 tetrahe-
dra with Nb=O groups are Lewis acid sites.[30,36]
The pore structures were further investigated by TEM obser-
vations (Figure 2). The images clearly show that large meso-
pores were formed in NbPSi-0.5, and NbPSi-1 contained
Based on the above, we hypothesize that the partial substi-
tution of Nb atoms with Si atoms in NbP would lead
to an increase in the ratio of Brønsted to Lewis acid
sites. This is because Si occupies a tetrahedral envi-
ronment and the added Si atoms would be expected
to preferentially replace Nb atoms in tetrahedral envi-
ronments, leading to suppression in the formation of
Lewis acid sites. In addition, the properties of silica,
which generates Brønsted acidity by the reaction
with phosphate,[37] would also be useful in terms of
increasing the number of Brønsted acid sites.
NbPSi-0.5 and NbPSi-1 have an amorphous struc-
ture similar to NbP, as shown by X-ray diffraction pat-
terns (see Figure S1 in the Supporting Information),
so structural FTIR analyses were used to further inves-
tigate the structure of NbPSi-0.5 and NbPSi-1. The re-
sults shown in Figure 3 provide some clues in sup-
port of our speculations on modifying the Lewis acid
sites. The FTIR spectra for NbPSi-0.5, NbPSi-1, and
NbP revealed vibrational bands at around 1220, 1020,
and 510 cmÀ1 that are attributable to asymmetric and
symmetric stretching of P=O, and the harmonic vi-
bration of OÀPÀO, respectively.[38,39] The spectra of
NbPSi-0.5 and NbPSi-1 also provided information re-
garding the bonding environments of Si atoms.
Modes related to vibrations of asymmetric SiÀO, sym-
metric SiÀO, and OÀSiÀO appeared at around 1101,
800, and 468 cmÀ1, respectively.[40] With regard to the
bonding structure of Nb atoms, NbP showed two bands at
around 650 and 926 cmÀ1, which could be assigned to the
stretching of NbÀO and Nb=O groups, respectively.[29,38]
However, for NbPSi-0.5, a significant peak corresponding to
Figure 2. TEM images of a) NbPSi-0.5, b) NbPSi-1, c) NbP, and d) HZSM-5.
medium-sized mesopores. The TEM images also confirmed the
microporous nature of NbP and HZSM-5. The average pore
sizes, as determined from the TEM images, were in good
agreement with BET results.
The concentration of elements in the bulk and surface phase
of NbPSi-0.5, NbPSi-1, and NbP were measured by ICP-AES and
XPS, respectively. The results of elemental analyses are shown
in Table 1. The atomic ratio of Nb/P in NbPSi-0.5 was about
half that for NbP. In addition, the content of Nb and Si atoms
in NbPSi-0.5 was nearly the same. This indicates that approxi-
mately half of the Nb ions in NbPSi-0.5 are successfully substi-
tuted by Si ions, as expected. The atomic ratios of the bulk
phase were very similar to those of the surface phase, suggest-
ing the homogeneous distribution of elements in NbPSi-0.5.
In the case of NbPSi-1, the atomic ratio of Nb/P was close to
that for NbP. This implies that added Si atoms have a prefer-
ence for forming separate bonds rather than substituting Nb
atoms in NbPSi-1, and that NbPSi-1 has a much lower degree
of substitution of Nb atoms than NbPSi-0.5.
The structure of NbP is similar to that of Nb2O5.[25] It is
mainly composed of distorted NbO6 octahedra and PO4 tetra-
hedra,[26] and NbO4 tetrahedra are also found in NbP-based ma-
Figure 3. Structural FTIR spectra for a) NbPSi-0.5, b) NbPSi-1, and c) NbP.
2462
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2012, 5, 2460 – 2468

Mesoporous Siliconiobium Phosphate
Nb=O groups was not observed
at a similar position. This result
suggests that NbPSi-0.5 contains
very small amount of Nb=O
groups, which are related to
Lewis acid sites. In the case of
NbPSi-1, a small shoulder band
corresponding to Nb=O groups
appeared at around 928 cmÀ1
,
similar to NbP. This suggests that
Nb atoms in NbPSi-1 are not suf-
ficiently substituted by Si atoms,
and is in good agreement with
the elemental analysis results.
Characterization of acid
properties
In situ FTIR measurements with
ammonia were performed to dis-
tinguish between the Brønsted
and Lewis acid sites in the sam-
ples. Ammonia was used as
a probe molecule instead of pyri-
Figure 4. In situ FTIR spectra of a) NbPSi-0.5, b) NbPSi-1, c) NbP, and d) HZSM-5 for NH3 adsorption at room tem-
perature and desorption at different temperatures.
The amount of acid sites present in the samples was mea-
dine as the small kinetic diameter of ammonia allows easy
access inside the small pores of the reference samples.[41]
Figure 4 shows the FTIR spectra of ammonia adsorbed on cata-
lysts at room temperature and subsequently desorbed at in-
creasing temperatures.
sured by means of a NH3 pulse chemisorption method. To min-
imize the physisorption of NH3, chemisorption was performed
at 1008C. As shown in Table 1, the number of acid sites in
NbPSi-0.5 is comparable to those in NbP, HZSM-5, and NbPSi-1,
even though large numbers of Nb atoms, which are associated
with Lewis acidity, are substituted by Si atoms.
In the spectra, the peaks in the regions of 1440–1480 and
1670–1720 cmÀ1 are assigned to the symmetric and asymmet-
ric NÀH deformation of NH4+, which are formed by the interac-
tion of NH3 with the Brønsted acid sites. In addition, the peak
between 1600 and 1630 cmÀ1 originates from the asymmetric
NÀH deformation of NH3 coordinatively bonded to Lewis acid
sites.[42] The spectra of NbP and HZSM-5 clearly demonstrated
that these reference catalysts contained both Brønsted and
Lewis acid sites. This is in good agreement with previous re-
sults reported in the literature.[43,44] For NbPSi-1, a peak corre-
sponding to a Lewis acid site clearly appeared at 1610 cmÀ1, al-
though the size of the Lewis acid peak for NbPSi-1 was smaller
than those for NbP and HZSM-5. However, in the case of
NbPSi-0.5, no distinguishable peak for ammonia interacting
with Lewis acid sites was observed. Thus, it can be concluded
that most of the acid sites in NbPSi-0.5 are Brønsted acid sites,
and these findings are entirely consistent with the structural
FTIR results.
Figure 5 reveals NH3 temperature programmed desorption
(TPD) profiles for the catalysts. To determine the distribution of
acid strength, the NH3-TPD curves were deconvoluted by the
Gaussian method, in which three peaks were used to indicate
weak, medium, and strong acidity, respectively. The deconvolu-
tion results are listed in Table 2.
It should be noted that the IR peaks for HZSM-5 appeared at
a higher frequency than those for NbPSi-0.5, NbPSi-1, and NbP.
As the strength of acid sites increase, the peaks move toward
higher frequencies. Therefore, these results suggest that
HZSM-5 contains the strongest acid sites. Furthermore, on
evacuation at 2808C, the IR peaks for adsorbed ammonia were
substantially decreased for NbPSi-0.5, whereas they still re-
mained for NbP, NbPSi-1, and HZSM-5, indicating that ammo-
nia was less strongly adsorbed to NbPSi-0.5 than to other
catalysts.
Figure 5. TPD profiles of adsorbed NH3 on a) NbPSi-0.5, b) NbPSi-1, c) NbP,
and d) HZSM-5.
ChemSusChem 2012, 5, 2460 – 2468
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
2463

J. Yi et al.
reaction, indicating the much
higher stability and activity of
NbPSi-0.5.
Table 2. Deconvolution results of NH3-TPD profiles for the catalysts.
[a]
Solid acid
Atotal
Weak acidity
Medium acidity
Tm[b] [8C] X[c] [%]
Strong acidity
Tm[b] [8C] X[c] [%]
[a.u.]
Tm[b] [8C]
X[c] [%]
Furthermore, NbPSi-0.5 also
showed quite a difference in the
byproduct distribution com-
pared to other catalysts. Data on
product distribution can be
found in Table 3. For NbPSi-0.5,
hydroxyacetone was the main
NbPSi-0.5
NbPSi-1
NbP
62.5
65.8
69.6
67.2
154
155
155
167
24.1
23.4
23.3
45.2
218
216
220
292
37.5
34.1
25.5
22.4
302
300
301
366
38.4
42.5
51.2
32.4
HZSM-5
[a] Sum of the area of deconvoluted peaks. [b] Peak maximum temperature. [c] Peak fraction.
The peak maximum of NbPSi-0.5 appeared at a lower tem-
perature than that of the other catalysts, indicating that the
acidity of NbPSi-0.5 is weaker than that of other catalysts.
In addition, among the NbP-based catalysts, NbPSi-0.5 showed
a minimum density for strong acidity. These results are consis-
tent with the in situ FTIR results, implying that the acidic
center of NbPSi-0.5 is related to weak–medium acid sites.
Moreover, no significant differences between the total peak
areas of the samples were found. This result is again in agree-
ment with the NH3 pulse chemisorption results, indicating that
the number of acid sites on NbPSi-0.5 is similar to those for
NbP and HZSM-5.
Table 3. Product selectivity for NbPSi-0.5, NbPSi-1, NbP, and HZSM-5.[a]
Product
Selectivity [%]
NbPSi-0.5
NbPSi-1
NbP
HZSM-5
acrolein
76.3
15.8
0.4
0.8
1.7
70.5
7.3
3.2
3.8
3.4
65.6
6.8
5
4.7
5.2
1.1
11.6
76.6
5
hydroxyacetone (acetol)
acetaldehyde
acetone
propionaldehyde
propionic acid
others[b]
4.7
4.2
0.4
2.4
6.7
0.2
4.8
1.5
10.3
[a] Data are the mean values in the initial reaction period for 4 h. Reaction
conditions: 2508C and GHSV=14940 mLgcatalystÀ1 hÀ1. [b] Others include
unidentified products and missing carbon.
Catalytic performance
The catalytic performance of
NbPSi-0.5 for the dehydration of
glycerol was investigated at
2508C. First, the catalytic behav-
ior of NbPSi-0.5 was compared
with the corresponding values
for NbP, NbPSi-1, and HZSM-5,
and the results are displayed in
Figure 6a and b. All of the cata-
lysts achieved nearly complete
conversion of glycerol at the ini-
tial stage of the reaction. The se-
lectivity for acrolein (ca. 76%)
over NbPSi-0.5 was similar to
that for HZSM-5, whereas the
acrolein selectivity over NbP and
NbPSi-1 was about 12 and 6%
lower, respectively, than those of
others.
From Figure 6a, it can be
clearly seen that the conversion
over HZSM-5 and NbP steeply
decreased to 31 and 51% after
8 h. Although NbPSi-1 showed
longer stability than NbP and
HZSM-5, the glycerol conversion
over NbPSi-1 also decreased
after 5 h. However, NbPSi-0.5
~
*
Figure 6. Time course for a) glycerol conversion and b) selectivity for acrolein over NbPSi-0.5 ( ), NbPSi-1 ( ), NbP
catalyzed the complete conver-
sion of glycerol with a high se-
lectivity for acrolein during the
!
&
&
(
), and HZSM-5 ( ) at GHSV=14940 mLgcatalystÀ1 hÀ1. c) Stability results for glycerol conversion ( ) and selectivity
for acrolein ( ) and hydroxyacetone ( ) over fresh and regenerated NbPSi-0.5 at GHSV=12390 mLgcatalystÀ1 hÀ1
~
*
.
2464
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2012, 5, 2460 – 2468

Mesoporous Siliconiobium Phosphate
byproduct, and the selectivity for other byproducts did not
exceed 1% (except for propionaldehyde). On the other hand,
HZSM-5, NbP, and NbPSi-1 exhibited a lower selectivity for hy-
droxyacetone. Instead, they favored the formation of other by-
products, especially C2 compounds, such as acetaldehyde.
Acrolein and hydroxyacetone are both produced during the
dehydration of glycerol, while the formation of other byprod-
ucts requires secondary reactions.[6] Therefore, the product dis-
tribution over NbPSi-0.5 suggests that, over NbPSi-0.5, the
probabilities of the occurrence of secondary reactions are low.
To compare the stability of NbPSi-0.5 with that of the previ-
ously reported long-life catalyst more precisely, a long-term
stability test was performed under the same conditions as
those described in the literature.[16] As shown in Figure 6c,
during a 70 h reaction, the complete conversion of glycerol
over NbPSi-0.5 remained almost unchanged, and the selectivity
for acrolein remained over 74%. This stability is about three
times higher than that of the catalyst in the reference, which
indicates a glycerol conversion of 78% after 24–25 h.[16] More-
over, it should be noted that NbPSi-0.5 produces a higher yield
of acrolein (74%) after 24 h, although the acrolein selectivity of
NbPSi-0.5 is slightly lower.
Rationales for catalytic performance
The significantly enhanced catalytic performance of NbPSi-0.5
can be attributed to its favorable acidic and textural properties.
The reaction network proposed by Corma et al. explains the
formation of coke and its precursors by consecutive decompo-
sition reactions starting from glycerol, acrolein, and hydroxya-
cetone.[9] Based on this mechanism and other literature data, it
appears that dehydrogenation and hydrogenation comprise
a large portion of those decomposition reactions and result in
the production of various byproducts and coke precursors.[9,46]
On the other hand, it has been reported that Brønsted acid
sites are very inefficient in catalyzing the dehydrogenation of
alcohols.[47] As shown above, NbPSi-0.5 possesses almost pure
Brønsted acidity. This explains why NbPSi-0.5 facilitates the for-
mation of dehydration products, but suppresses side reactions
that can lead to the production of coke precursors. The prod-
uct distribution over NbPSi-0.5 (Table 2) confirms this specula-
tion. Over NbPSi-0.5, the total selectivity for acrolein and hy-
droxyacetone was more than 92%, and the concentrations of
minor byproducts were very small.
Large mesopores of NbPSi-0.5 also play a role in enhancing
the stability. As can be seen from the results given in Table 1,
for HZSM-5 and NbP, the total pore volume and BET surface
area decreased by more than 82 and 92%, and microporosity
almost disappeared after the reaction. However, in the case of
NbPSi-0.5, the total pore volume and BET surface area de-
creased by only 68 and 76%. NbPSi-1 also showed less de-
crease in the total pore volume and BET surface area than
those of HZSM-5 and NbP. This difference is related to the fact
that NbPSi-0.5 and NbPSi-1 have larger mesopores than those
of HZSM-5 and NbP. The large mesoporosity may confer an ad-
vantage to NbPSi-0.5 in reducing pore blocking by carbon
deposition, and as a result, the acid sites located inside the
pores will still be available for long reactions. Similar beneficial
effects of mesopores in deactivation can be found in the litera-
ture.[13,31]
Although NbPSi-0.5 has a high stability and activity for this
reaction, the regeneration and reuse of NbPSi-0.5 are preferred
options for practical applications. Therefore, the possibility of
regenerating the used NbPSi-0.5 was investigated by heat
treatment under air conditions, a process that is intended to
oxidize the coke formed on the catalysts. After the long-term
stability test, the used NbPSi-0.5 was heated to 4508C under
an air flow (30 mLminÀ1) for 4 h.
In the N2 adsorption–desorption test (see Figure S2 in the
Supporting Information), regenerated NbPSi-0.5 revealed an
H1-type hysteresis loop, similar to fresh NbPSi-0.5. However,
the hysteresis loop for regenerated NbPSi-0.5 was broader, and
the pore size and BET surface area were slightly decreased
after regeneration. A decrease in pore size was also observed
in the TEM image (Figure S3 in the Supporting Information).
These may be due to the partial collapse of pores during the
regeneration process.
In addition, NbPSi-0.5 showed different behaviors in terms
of the properties of the carbonaceous deposits. The results of
thermogravimetric (TG) and differential thermal analysis (DTA)
for used NbPSi-0.5, NbPSi-1, NbP, and HZSM-5 are shown in
Figure 7. Since the activities of the catalysts studied differ sig-
nificantly with time, the amount of coke produced during an
8 h reaction is expressed as mg of coke per mmol of glycerol
reacted. NbPSi-0.5 produced the smallest amount of coke
among the used catalysts. It is well known that the strong
acidity of catalyst is responsible for the formation of
coke,[15,16,18] and the acidity of NbPSi-0.5 is slightly weaker than
that of the others. Therefore, the weaker acid sites of NbPSi-0.5
can account for the lower coking tendency of NbPSi-0.5.
The DTA curves revealed two exothermic peaks in the tem-
perature range of 250–6008C, indicating the presence of two
different types of coke in the used catalysts. The first peak,
with a maximum at around 340–4308C, can be assigned to
easily oxidizable coke, whereas the second peak, with a maxi-
mum at above 5008C, is related to the formation of more
stable coke.[10,48] Compared with NbP, NbPSi-1, and HZSM-5, in
NH3-TPD analyses were also performed to examine the acid
properties of the regenerated NbPSi-0.5 (Figure S4 in the Sup-
porting Information). The peak shape and maximum tempera-
ture of regenerated NbPSi-0.5 were consistent with those of
fresh NbPSi-0.5, except for a small decrease in peak area. This
indicates that the regeneration process does not change the
acid strength and its distribution of NbPSi-0.5, but induces
a slight decrease in the amount of acid sites.
In the catalytic test, regenerated NbPSi-0.5 also showed
a high performance. Compared to fresh NbPSi-0.5, the regener-
ated NbPSi-0.5 exhibited almost the same selectivity for acrole-
in and the same glycerol conversion for about 40 h, although
slightly faster deactivation was observed after 45 h. Katryniok
et al. have recently reported similar results, indicating that re-
generated catalysts had lower stability than fresh catalysts.[45]
This can be attributed to heavy carbonaceous species, which
are not oxidized under regeneration conditions.
ChemSusChem 2012, 5, 2460 – 2468
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
2465

J. Yi et al.
area. Catalytic results showed that NbPSi-0.5 had a higher ac-
tivity and stability than those of NbP and HZSM-5. Moreover,
in long-term tests, the stability of NbPSi-0.5 was about three
times as high as that of the reported long-life catalyst. The sig-
nificantly enhanced catalytic performance of NbPSi-0.5 could
be attributed to the following reasons: 1) the nearly pure
Brønsted acidity promoted the production of main dehydra-
tion products, but suppressed side reactions leading to coke
formation; 2) large mesopores significantly reduced pore
blocking by coke deposition; and 3) the amount and oxidation
temperature of produced coke decreased. These findings indi-
cated that NbPSi-0.5 was a promising catalyst that could accel-
erate the practical viability of this reaction. We believe that
NbPSi-0.5 can also be applicable to other acid-catalyzed reac-
tions involving xylose, glucose, and so forth, as they are struc-
turally analogous to glycerol.
Experimental Section
Preparation of catalysts: The mesoporous NbPSi-0.5 composite was
synthesized by a solvothermal method combined with solvent
evaporation. Solvothermal techniques are widely used for the syn-
thesis of multicomponent materials,[50] and the solvent evaporation
methods can be very useful for preparing mesostructured inorgan-
ic minerals.[51] Hence, to synthesize a mesostructured homogene-
ous composite of NbP and silicate, a combination of solvothermal
treatment and solvent evaporation was developed. In addition,
a small amount of diethylphosphatoethyltriethoxysilane, which
contains both phosphate and silicate groups, was used to aid in
producing a homogenous combination of components. NbCl5,
triethyl phosphate (TEP), ethanol, H3PO4 (85%) and tetraethyl or-
thosilicate (TEOS) were purchased from Sigma–Aldrich; diethyl-
phosphatoethyltriethoxysilane (DEPETES, 95%) was from Gelest;
and Pluronic P123 was from BASF. All chemicals were used as
received.
In a typical synthesis, NbCl5 (0.015 mol), TEP (0.0135 mol), and
DEPETES (0.0015 mol) were dissolved in EtOH (85 mol). The dissolu-
tion of NbCl5 in EtOH was performed in a glove box to avoid mois-
ture contamination. A stirred solution of P123 (0.0006 mol) and
H3PO4 (0.015 mol) in EtOH (115 mol) was then added. After stirring
the mixture at 408C for 3 h, TEOS (0.015 mol) was added and the
solution was stirred at 408C for 24 h. The resulting solution was
then loaded into an autoclave and heated with continuous stirring
at 1508C for 24 h. The as-prepared pale white sol was transferred
into a dish. The ethanol solvent was evaporated without control-
ling the humidity at 808C for 24 h. The resulting solids were cal-
cined at 5508C for 4 h.
Figure 7. a) Amount of coke and b) DTA curves for used NbPSi-0.5, NbPSi-1,
NbP, and HZSM-5.
the DTA curve for NbPSi-0.5, the maximum for the first peak
occurred at a much lower temperature and the relative area
ratio of the first to second peaks increased in the order of
NbPSi-0.5>NbPSi-1>NbP>HZSM-5. These findings indicate
that the coke produced over NbPSi-0.5 is made up of the more
oxidizable form than those over other catalysts.
Analysis of the atomic H/C ratio in coke over the catalysts
gives results consistent with the DTA data. The atomic H/C
ratio in coke, as measured by CHNS elemental analysis, was
found to be in the following order: NbPSi-0.5 (1.21)>NbP-
Si-1 (1.17)>NbP (1.08)>HZSM-5 (0.99). The higher H/C ratio of
coked NbPSi-0.5 implies that the carbon deposits over NbP-
Si-0.5 have a less condensed and aromatized structure,[49] and
therefore, are more readily oxidized.
For comparison with NbPSi-0.5, NbPSi-1 (atomic ratio Nb/P=1)
was prepared by processes identical to those of NbBSi-0.5, except
for the content of NbCl5. To obtain the atomic ratio of Nb/P=1,
0.03 mol of NbCl5 was used.
Pure NbP was also prepared according to a previously reported
method.[33] HZSM-5 was prepared through the calcination of com-
mercial zeolite NH4ZSM-5 (zeolyst, SiO2/Al2O3 =50) at 5008C for 6 h.
Conclusions
This study reported a method for preparing a mesoporous sili-
coniobium phosphate (NbPSi-0.5) composite with a high con-
tent of Brønsted acidity and a demonstration of its potential
for the efficient and stable dehydration of glycerol into acrole-
in. Importantly, NbPSi-0.5 was comprised of nearly pure Brøn-
sted acid sites. In addition, NbPSi-0.5 contained comparable
amounts of acid sites and exhibited weaker acidity than that of
highly acidic NbP and the commercial zeolite HZSM-5. The
findings also confirmed that NbPSi-0.5 had a well-developed
mesoporous structure with large pores and a high surface
Characterization: The N2 adsorption–desorption isotherms were re-
corded on a Micrometrics ASAP-2010 system. The pore size was
calculated from the adsorption branches of isotherms by using BJH
methods. The TEM and energy-dispersive X-ray spectroscopy (EDS)
observations were carried out by using a JEOL JEM-3010 micro-
scope operating at 300 kV. The elemental analysis of the bulk
phase was performed by ICP-AES (Shimadzu, JP/ICPS-7500). The
2466
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2012, 5, 2460 – 2468

Mesoporous Siliconiobium Phosphate
XPS spectra were measured with a Kratos AXIS-His electron spec-
trometer equipped with a MgKa X-ray source and a hemispherical
electron energy analyzer. The structural FTIR spectra of samples
were obtained by using a Nicolet 6700 FT-IR spectrophotometer.
In situ FTIR spectra of NH3 adsorption were measured with a Midac
spectrometer (model 2100). A self-supported pellet containing
about 40 mg of the sample was placed in an in situ IR cell, de-
signed by Moon et al.[52] The pellet was pretreated at 1508C under
N2 flow for 1 h. After cooling to room temperature, the pellet was
exposed to NH3 gas at 2.67 kPa for 3 min. The desorption of NH3
was performed by evacuation at different temperatures (30, 100,
200, and 2808C). The spectrum was recorded after desorption at
each temperature. The NH3-pulse chemisorption was carried out
by using a Micrometrics Autochem II chemisorption analyzer. Prior
to analysis, 0.1 g of sample was pretreated under He flow at 5008C
for 1 h. After cooling to 1008C, 10.2% NH3/He gas was periodically
injected onto the sample by using an injection loop. The amount
of NH3 uptake was monitored by a thermal conductivity detector
(TCD) until successive peaks showed the same area. The NH3-TPD
spectra were obtained by using the same analyzer and pretreat-
ment procedure as that used for NH3-pulse chemisorption. The
samples were saturated with 10.2% NH3/He gas at 508C. Physi-
sorbed NH3 was eliminated by flushing with a stream of He at
1008C. After again cooling to 508C, the temperature was increased
to 8008C at a rate of 108CminÀ1 under a stream of He. TG and
DTA were performed simultaneously in a TA instruments SDT Q600
analyzer. The experiments were carried out from room temperature
to 8008C with a heating rate of 108CminÀ1 under air. The atomic
H/C ratio of coke over the used catalysts was determined by CHNS
elemental analysis by using a LECO CHNS-932 instrument.
funded by the Ministry of Education, Science, and Technology
through the National Research Foundation of Korea (R31-10013).
Keywords: Brønsted acids · dehydration · heterogeneous
catalysis · mesoporous materials · renewable resources
[1] R. Rinaldi, F. Schꢁth, Energy Environ. Sci. 2009, 2, 610–626.
[2] M. Hara, Energy Environ. Sci. 2010, 3, 601–607.
[3] J. N. Chheda, G. W. Huber, J. A. Dumesic, Angew. Chem. 2007, 119,
7298–7318; Angew. Chem. Int. Ed. 2007, 46, 7164–7183.
[4] E. Taarning, C. M. Osmundsen, X. Yang, B. Voss, S. I. Andersen, C. H.
Christensen, Energy Environ. Sci. 2011, 4, 793–804.
[5] R. Weingarten, G. A. Tompsett, W. C. Conner, Jr., G. W. Huber, J. Catal.
2011, 279, 174–182.
[6] B. Katryniok, S. Paul, V. Belliere-Baca, P. Rey, F. Dumeignil, Green Chem.
2010, 12, 2079–2098.
[7] B. Katryniok, S. Paul, M. Capron, F. Dumeignil, ChemSusChem 2009, 2,
719–730.
[8] S. H. Chai, H. P. Wang, Y. Liang, B. Q. Xu, Green Chem. 2007, 9, 1130–
1136.
[9] A. Corma, G. W. Huber, L. Sauvanaud, P. O’Connor, J. Catal. 2008, 257,
163–171.
[10] C. J. Jia, Y. Liu, W. Schmidt, A. H. Lu, F. Schꢁth, J. Catal. 2010, 269, 71–
79.
[11] Y. T. Kim, K. D. Jung, E. D. Park, Microporous Mesoporous Mater. 2010,
131, 28–36.
[12] K. Pathak, K. M. Reddy, N. N. Bakhshi, A. K. Dalai, Appl. Catal. A 2010,
372, 224–238.
[13] C. J. Zhou, C. J. Huang, W. G. Zhang, H. S. Zhai, H. L. Wu, Z. S. Chao,
Stud. Surf. Sci. Catal. 2007, 165, 527–530.
Catalytic reaction: The dehydration of glycerol was performed
under atmospheric pressure at 2508C. A catalyst sample of 0.3 g
was packed into a quartz reactor (8 mm inner diameter), and the
reactor was then placed in an electric furnace. The temperature of
the catalyst bed was monitored by a K-type thermocouple and
controlled by a PID controller. Before the reaction, the catalyst was
[14] A. Alhanash, E. F. Kozhevnikova, I. V. Kozhevnikov, Appl. Catal. A 2010,
378, 11–18.
[15] S. H. Chai, H. P. Wang, Y. Liang, B. Q. Xu, Green Chem. 2008, 10, 1087–
1093.
[16] B. Katryniok, S. Paul, M. Capron, C. Lancelot, V. Belliꢂre-Baca, P. Rey, F.
Dumeignil, Green Chem. 2010, 12, 1922–1925.
[17] E. Kraleva, R. Palcheva, L. Dimitrov, U. Armbruster, A. Brꢁckner, A. Spoja-
kina, J. Mater. Sci. 2011, 46, 7160–7168.
[18] E. Tsukuda, S. Sato, R. Takahashi, T. Sodesawa, Catal. Commun. 2007, 8,
1349–1353.
[19] P. Lauriol-Garbay, J. M. M. Millet, S. Loridant, V. Belliꢂre-Baca, P. Rey, J.
Catal. 2011, 280, 68–76.
[20] L. Z. Tao, S. H. Chai, Y. Zuo, W. T. Zheng, Y. Liang, B. Q. Xu, Catal. Today
2010, 158, 310–316.
preheated to the reaction temperature under
a N2 flow
(30 mLminÀ1) for 1 h. A 10 wt% aqueous solution of glycerol was
injected into the vaporizer by using a syringe pump. The reactant
vapor was diluted with a stream of N2 (30 mLminÀ1) in the vaporiz-
er and then fed into the catalyst bed. The aqueous solution of
glycerol was injected at a rate of 2.1 mLhÀ1 to obtain a molar com-
position of glycerol/H2O/N2 =1.3/53.6/40.2 and
a
GHSV of
14940 mLgcatalystÀ1 hÀ1. In the case of the long-term reaction, the
solution of glycerol was fed at a rate of 1.5 mLhÀ1, giving a molar
composition of glycerol/H2O/N2 =1.1:50.5:48.4 and a GHSV of
[21] A. Ulgen, W. Hoelderich, Catal. Lett. 2009, 131, 122–128.
[22] Q. Liu, Z. Zhang, Y. Du, J. Li, X. Yang, Catal. Lett. 2009, 127, 419–428.
[23] F. Wang, J. L. Dubois, W. Ueda, J. Catal. 2009, 268, 260–267.
[24] F. Wang, J. Xu, J. L. Dubois, W. Ueda, ChemSusChem 2010, 3, 1383–
1389.
12390 mLgcatalystÀ1 hÀ1
. The reaction product was collected in
a cold trap and analyzed by a gas chromatograph (Younglin ACME
6100 model) equipped with a FID detector and a HP-Innowax ca-
pillary column. To ensure the identification of products, GC-MS
analyses were also carried out by using an Agilent 7890A GC
system equipped with an Agilent 5975C MS detector and a HP-
5ms capillary column.
[25] I. Nowak, M. Ziolek, Chem. Rev. 1999, 99, 3603–3624.
[26] M. Ziolek, Catal. Today 2003, 78, 47–64.
[27] S. Okazaki, N. Wada, Catal. Today 1993, 16, 349–359.
[28] P. Carniti, A. Gervasini, S. Biella, A. Auroux, Catal. Today 2006, 118, 373–
378.
[29] T. Armaroli, G. Busca, C. Carlini, M. Giuttari, A. M. R. Galletti, G. Sbrana, J.
Mol. Catal. A 2000, 151, 233–243.
[30] K. Nakajima, Y. Baba, R. Noma, M. Kitano, J. N. Kondo, S. Hayashi, M.
Hara, J. Am. Chem. Soc. 2011, 133, 4224–4227.
[31] L. G. A. van de Water, J. C. van der Waal, J. C. Jansen, T. Maschmeyer, J.
Catal. 2004, 223, 170–178.
[32] H. Zhu, Z. Liu, D. Kong, Y. Wang, Z. Xie, J. Phys. Chem. C 2008, 112,
17257–17264.
Acknowledgements
[33] N. K. Mal, M. Fujiwara, Chem. Commun. 2002, 2702–2703.
[34] J. Zhu, Y. Huang, Inorg. Chem. 2009, 48, 10186–10192.
[35] S. V. Karpov, E. V. Kolobkova, Fiz. Khim. Stekla 1991, 17, 425–434.
[36] R. M. Pittman, A. T. Bell, J. Phys. Chem. 1993, 97, 12178–12185.
[37] G. Connell, J. A. Dumesic, J. Catal. 1987, 105, 285–298.
This work was financially supported by a grant from the Industri-
al Source Technology Development Programs (10033352) of the
Ministry of Knowledge Economy (MKE) of Korea. This research
was also supported by the World Class University program
ChemSusChem 2012, 5, 2460 – 2468
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
2467

J. Yi et al.
[38] M. H. C. de La Cruz, A. S. Rocha, E. R. Lachter, A. M. S. Forrester, M. C.
Reis, R. A. S. S. Gil, S. Caldarelli, A. D. Farias, W. A. Gonzalez, Appl. Catal.
A 2010, 386, 60–64.
[39] C. Dayanand, G. Bhikshamaiah, V. J. Tyagaraju, M. Salagram, A. S. R. K.
Murthy, J. Mater. Sci. 1996, 31, 1945–1967.
[40] H. El Rassy, A. C. Pierre, J. Non-Cryst. Solids 2005, 351, 1603–1610.
[41] G. V. A. Martins, G. Berlier, C. Bisio, S. Coluccia, H. O. Pastore, L. Mar-
chese, J. Phys. Chem. C 2008, 112, 7193–7200.
[47] C. P. Bezouhanova, M. A. Al-Zihari, Catal. Lett. 1991, 11, 245–248.
[48] H. Atia, U. Armbruster, A. Martin, J. Catal. 2008, 258, 71–82.
[49] M. Guisnet in Handbook of Heterogeneous Catalysis, Vol. 2 (Eds.: G. Ertl,
H. Knçzinger, J. Weitkamp), Wiley-VCH, Weinheim, 1997, pp. 626–632.
[50] S. Feng, L. Guanghua in Modern Inorganic Synthetic Chemistry (Eds.: R.
Xu, W. Pang, Q. Huo), Elsevier, Amsterdam, 2011, pp. 63–95.
[51] S. W. Boettcher, J. Fan, C. K. Tsung, Q. Shi, G. D. Stucky, Acc. Chem. Res.
2007, 40, 784–792.
[42] C. C. Hwang, C. Y. Mou, J. Phys. Chem. C 2009, 113, 5212–5221.
[43] Q. Sun, A. Auroux, J. Shen, J. Catal. 2006, 244, 1–9.
[52] S. H. Moon, H. Windawi, J. R. Katzer, Ind. Eng. Chem. Fundam. 1981, 20,
396–399.
[44] F. Lꢃnyi, J. Valyon, Microporous Mesoporous Mater. 2001, 47, 293–301.
[45] B. Katryniok, S. Paul, M. Capron, V. Belliꢂre-Baca, P. Rey, F. Dumeignil,
ChemSusChem 2012, 5, 1298–1306.
Received: August 10, 2012
[46] P. Lauriol-Garbey, G. Postole, S. Loridant, A. Auroux, V. Belliere-Baca, P.
Rey, J. M. M. Millet, Appl. Catal. B 2011, 106, 94–102.
Published online on November 6, 2012
2468
www.chemsuschem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2012, 5, 2460 – 2468