Conversion of 2,3-Butanediol over Phosphate Catalysts

Article abstract of DOI:10.1002/cctc.201501399

2,3-Butanediol (BDO) is an excellent feedstock for expanding the network of bio-based chemicals through catalytic technologies. The dehydration of BDO provides an alternative green route to important chemicals such as methyl ethyl ketone (MEK) and 1,3-butadiene (BD). In this contribution, we report on the catalytic performance of boron (BP), aluminum (AlP), titanium (TiP), zirconium (ZrP), and niobium (NbP) phosphates in BDO dehydration. The kinetic study points to three reaction pathways operating over phosphate catalysts, leading to MEK, 2-methyl-propanal (MP), and BD via intermediate 3-butene-2-ol (3B2OL) formation. The major reaction pathway is shown to involve pinacol rearrangement to MEK. The reactivity of phosphates is found to increase in the order: BP

Full text of DOI:10.1002/cctc.201501399

DOI: 10.1002/cctc.201501399
Full Papers
Conversion of 2,3-Butanediol over Phosphate Catalysts
Maria A. Nikitina[a, b] and Irina I. Ivanova*[a, c]
2,3-Butanediol (BDO) is an excellent feedstock for expanding
the network of bio-based chemicals through catalytic technol-
ogies. The dehydration of BDO provides an alternative green
route to important chemicals such as methyl ethyl ketone
(MEK) and 1,3-butadiene (BD). In this contribution, we report
on the catalytic performance of boron (BP), aluminum (AlP), ti-
tanium (TiP), zirconium (ZrP), and niobium (NbP) phosphates in
BDO dehydration. The kinetic study points to three reaction
pathways operating over phosphate catalysts, leading to MEK,
2-methyl-propanal (MP), and BD via intermediate 3-butene-2-ol
(3B2OL) formation. The major reaction pathway is shown to in-
volve pinacol rearrangement to MEK. The reactivity of phos-
phates is found to increase in the order: BP<TiP <ZrP=NbP<
AlP. The best catalytic performance is achieved over AlP, which
shows the highest selectivity towards MEK (78%) at 100% 2,3-
butanediol conversion.
Introduction
In the few last years the production of bio-based chemicals
from renewable resources has received growing attention, es-
pecially as the availability of petrochemical feedstocks is dimin-
ishing. 2,3-Butanediol (BDO) can be produced by fermentation
of sugars with various bacteria[1] as well as by fermentation of
syngas by using acetogenic organisms[2] and can be considered
as a potential feedstock for a number of useful chemicals. It
can be used as an antifreeze agent, solvent, and precursor for
plastics manufacturing. Its ester derivatives are precursors of
polyurethane and can be applied in cosmetic, pharmaceutical,
and other industries.[1,3] But the most important applications
are connected with BDO dehydration into butan-2-one (methyl
ethyl ketone, MEK), 3-buten-2-ol (3B2OL), and 1,3-butadiene
(BD).[1b,4–6]
opment of the ethylene industry are such that ethylene pro-
ducers are switching to more efficient ethylene production
technologies. This leads to a lack of butadiene on the market
and therefore requires the development of alternative technol-
ogies leading to butadiene. Therefore, the synthesis of MEK,
BD, and 3B2OL from BDO appears to be a challenging goal.
Synthesis of MEK from BDO has been reported over strong
solid catalysts such as H-Nafion,[10] heteropoly acids,[11] alumino-
silicates,[12] and zeolites.[1f,13,14] Bucsi et al.[10] reported high con-
versions of BDO into MEK over H-Nafion catalysts at tempera-
tures as low as 1508C; the acetalization of MEK with BDO
being the main side reaction. MEK and its acetals accounted
for 94% of the products over these catalysts. Other strong
acids such as heteropoly acids have also demonstrated high
conversion into MEK in the temperature range 150–1808C.[11]
The selectivity to MEK varied from 50 to 95%, the main by-
products were the corresponding dioxolanes and 2-methyl-
propanal (MP). Zeolites were found to be active catalysts for
transforming BDO into MEK in the temperature range 230–
3508C.[1f,13–15] Different zeolite structures, including FAU, MFI,
MOR, and BEA, have also been tested. Whereas medium-pore
zeolites favored MEK formation, the large-pore zeolites yielded
acetals and ketals.[13]
MEK is a widely used organic solvent as well as a precursor
for MEK peroxide. Nowadays, it is mainly produced from n-
butane feedstocks by consecutive dehydrogenation–hydra-
tion–dehydrogenation steps.[7] 3B2OL finds applications in
medicine and agrochemicals production.[8] Whereas, BD is
a bulk chemical that has an enormous application scope. It is
an important monomer for the production of such polymers as
polybutadiene, styrene–butadiene rubber, styrene–butadiene
latex, acrylonitrile–butadiene–styrene polymer, nitrile rubber,
and some others.[9] At present, about 95% of butadiene is pro-
duced by isolation from naphtha steam cracker fractions, gen-
erated during ethylene production. But the trends in the devel-
The selective synthesis of 3B2OL and BD from BDO has been
achieved over less acidic solids and at higher temperatur-
es.[8b,16,17] The first attempts for 3B2OL and BD synthesis go
back to the 1940s. Winfield[16] studied a large number of metal
oxides and salts and found that the best catalyst for 3B2OL
and BD production is ThO2. They disclosed that over ThO2,
3B2OL can be obtained with a yield of 70% at 3508C, whereas
62% yield of BD can be reached at 5008C.[16]
[a] M. A. Nikitina, Prof. I. I. Ivanova
Department of Chemistry
Moscow State University
Leninske Gory 1, bld. 3, 119991 Moscow (Russia)
E-mail: iiivanova@phys.chem.msu.ru
Significant contribution to the synthesis of 3B2OL and BD
from BDO has been made by the group of Sato et al.[8b,17,18]
during the few last years. These authors have investigated
BDO dehydration over a large range of rare-earth-oxide cata-
lysts. They disclosed that Sc2O3 calcined at 8008C shows excel-
lent catalytic activity at 3258C with 85% selectivity to 3B2OL at
[b] M. A. Nikitina
“UNISIT” LLC
Leninskye Gory 1, bld. 75V, 119991 Moscow (Russia)
[c] Prof. I. I. Ivanova
A.V. Topchiev Institute of Petrochemical Synthesis
Russian Academy of Science
Leninskiy prospect 29, 119991 Moscow (Russia)
ChemCatChem 2016, 8, 1346 – 1353
1346
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
100% BDO conversion.[18] The highest BD yield of 88.3% was
achieved over Sc2O3 at 4118C under H2 carrier gas flow.[18] In
addition, it was found that the yield of BD can be increased up
to 94% at 3188C in a double-bed reactor containing Sc2O3 in
an upper layer and Al2O3 in a lower layer.
Table 1. Characteristics of the phosphate catalysts.
Symbol M precursor M/P
SBET
Vpore
[cm3 gÀ1
dpore
[nm]
a0
[mol/mol] [m2 gÀ1
]
]
[mmolgÀ1
]
BP
AlP
TiP
ZrP
NbP
B(iPrO)3
AlCl3
Ti(OBu)4
ZrOCl2·8H2O 0.6
Nb2O5·nH2O 2.5
1
1.5
0.5
73
88
87
85
88
0.5
0.6
0.3
0.1
0.4
10–80 430
20–90 930
20–100 410
5–100 840
20–80 640
Analysis of the literature data suggests that BDO dehydra-
tion has been extensively studied over oxides, zeolites, and
heteropoly acids. However, as far as we know, no information
is available for phosphates, which are known to be efficient
solid catalysts for a number of acid-catalyzed reactions,[19,20] in-
cluding different pinacol-type rearrangements, ketalization of
ketones, isomerization of unsaturated compounds, dehydration
of alcohols and sugars,[19a,b] isoprene synthesis from formalde-
hyde and isobutene,[20] and so on. For example, Maurin[21] dis-
closed that phosphate catalysts (Li3PO4) can be used for iso-
prene synthesis from corresponding diol.
which points to some contribution from the corresponding
metal oxide phases.
To obtain TiP and BP catalysts with similar specific surface
areas, organic precursors were used. Boron phosphate is usual-
ly synthesized through interaction of boron acid with the
aqueous solution of H3PO4.[24,29] However, this method leads to
the formation of a crystalline material with a specific surface
area of ꢀ7 m2 gÀ1 and very low content of acidic sites. The ap-
plication of the organic precursor tri-iso-propylborate allowed
us to increase the specific surface area up to ꢀ73 m2 gÀ1
(Table 1), but also yielded crystalline material after calcina-
tion.[26] Attempts to obtain amorphous BP found no success.
Conversely, in the case of TiP, the use of titanium butoxide as
the precursor resulted in the formation of amorphous phos-
phate with high surface area (Table 1).
In this contribution, we report on the catalytic performance
of metal (III–V) phosphate catalysts in the BDO dehydration.
Boron (BP), aluminum (AlP), titanium (TiP), zirconium (ZrP), and
niobium (NbP) phosphates with high surface areas of 80 m2 gÀ1
were prepared and characterized by elemental analysis, X-ray
diffraction, low-temperature nitrogen adsorption, temperature-
programmed desorption (TPD) of ammonia, and FTIR of ad-
sorbed pyridine. The catalytic activity was investigated at
a wide range of temperatures (250–4008C) and weight hourly
space velocities (WHSV) of 1.5–500 hÀ1. The results point to the
high activity of phosphates in the BDO conversion to MEK.
The main characteristics of the phosphate catalysts prepared
are shown in Table 1 and Figures 1–4. According to the XRD
data, all samples except boron phosphate are amorphous ma-
terials (Figure 1). The BP catalyst shows a XRD pattern typical
for crystalline phosphate with a cristobalite structure.[30]
Results and discussion
Catalysts preparation and characterization
The literature data contains a lot of information on different
amorphous and crystalline phosphate materials and the meth-
ods of their preparation.[19a,22–26] Crystalline phosphates with a,
g, z structures are usually not thermally stable and convert
into other crystalline forms during calcination over 3008C.
Thermal treatment of such materials is accompanied by the
decrease of their acidity and specific surface area.[19,27] There-
fore, in this study we have concentrated on amorphous solids
with high surface areas and acid site availability. Metal (III, IV,
V) phosphates, namely, BP, AlP, TiP, ZrP, and NbP were selected
owing to their high acidity and wide scope of their application
as catalysts.[19b,20–22,28]
Figure 1. XRD patterns of the metal phosphates.
Synthesis of the ZrP, AlP, and NbP materials was carried out
according to the standard procedures based on the treatment
of the corresponding salts or oxides with orthophosphoric
acid, which leads to amorphous phosphates with high acidity
and surface areas.[24–27] The M/P ratio was adjusted to have
a range of catalysts with similar specific surface areas (Table 1).
The highest content of phosphorus was required for ZrP and
TiP. Attempts to decrease the phosphorous content resulted in
a significant decrease in the specific surface area and the
amount of acid sites. Quite the opposite situation was ob-
served in the case of AlP and NbP materials, which have the
highest metal content in the series of samples synthesized. In
these materials, the M/P ratio reaches 1.5 and 2.5, respectively,
Nitrogen adsorption–desorption measurements show that
all the samples studied have comparable specific surface areas
in the range 73–88 m2 gÀ1. According to Barrett–Joyner–Halen-
da (BJH) analysis, the pore size distribution is very broad
(Table 1), which is typical for amorphous materials.
The FTIR spectra of the calcined samples are shown in
Figure 2. A broad band at ꢀ3400 cmÀ1 and bands in the range
1635–1945 cmÀ1 correspond to hydroxyl groups and to ad-
sorbed water molecules.[31] The bands in the region 850–
1160 cmÀ1 are due to asymmetric and symmetric stretching vi-
brations of the P=O and PÀOH groups.[31b,32] Additionally, the
small bands in the range 1300–1400 cmÀ1 and 700–900 cmÀ1
ChemCatChem 2016, 8, 1346 – 1353
www.chemcatchem.org
1347
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 3. FTIR spectra of pyridine adsorbed on the phosphate catalysts.
Figure 2. FTIR spectra of the metal phosphate materials.
3À
can be attributed to free PO4 and the PÀOÀP link, respective-
sorbed water. Water molecules interact with unsaturated metal
sites (Lewis sites), leading to the formation of OH groups.[20]
Dehydroxylation of the samples leads to the transformation of
Brønsted sites into Lewis sites and, vice versa, the addition of
water may restore the content of Brønsted sites. The intensity
of the band assigned to pyridine adsorbed on Brønsted acid
sites increases in the order: AlP<ZrP<TiP <BP<NbP, whereas
the intensity of the band related to Lewis sites increases in the
order: ZrP<AlP<TiP <NbP<BP (Figure 3).
Comparison of the spectra in the region 1445–1465 cmÀ1 re-
veals a significant difference for the boron phosphate sample.
Whereas for most of the samples, the band attributed to pyri-
dine adsorbed on Lewis sites is observed at ꢀ1449 cmÀ1, in
the case of BP catalyst it is shifted to 1461 cmÀ1. A similar shift
was observed by Travert et al.[36b] in a series of metal oxide ma-
terials. In particular, the band of pyridine adsorbed on Lewis
sites was shifting from 1449 to 1462 cmÀ1 in the following
range of materials: MgO/ZrO2/Al2O3/SiO2/B2O3. The authors as-
cribed their observation to the increase in polarizing power of
the cation in the above range of oxides. We can speculate that
a similar effect can be observed for phosphate materials.
Indeed, a small shift to 1454 cmÀ1 is also seen in the case of
the AlP sample. Another explanation could be related to the
crystalline structure of the BP sample, which results in stronger
Lewis acidity for this sample.
ly.[31b,33]
The spectrum of boron phosphate shows significant differ-
ences with respect to the other spectra owing to the crystal-
line structure of this sample. The bands at approximately 555,
620, and 929 cmÀ1, observed only in the case of BP, can be as-
signed to bending vibrations in the [PO4]3À tetrahedron and to
pseudo-lattice translations of BÀO, respectively.[34]
In the case the NbP and AlP samples, a broad band in the
range 500–700 cmÀ1 is observed along with the bands attribut-
ed to the phosphates. This band corresponds to the stretching
vibrations of bridging MÀOÀM bonds in the oxide phases[35]
and indicates that these samples are a mixture of metal oxide
and phosphate phases.[35a] The possibility of metal oxide phase
formation depends on the elemental composition of the mate-
rials. In the case of AlP and NbP samples, the metal content is
the highest (Table 1), which is in line with the formation of the
corresponding oxide phases.
Acidic properties
The acidic properties of the prepared catalysts were studied by
using TPD NH3 measurements and IR spectroscopy of adsorbed
pyridine. Whereas the former was used for the determination
of the amount and strength of the acid sites, the latter was ap-
plied for the investigation of their nature.
The TPD NH3 profiles obtained for the phosphate materials
are presented in Figure 4. All the samples show only one peak
in the TPD profiles, which is typical for amorphous materials.[37]
However, the acid site distribution is very different. AlP and TiP
contain only weak and medium sites desorbing NH3 in the
range 100–4508C. Conversely, ZrP and NbP show very broad
TPD profiles (100–6008C) with a high contribution of strong
acid sites, whereas boron phosphate demonstrates intermedi-
ate behavior. It should be mentioned that the AlP sample
shows the narrowest acid site distribution and the highest con-
tent of sites, which is typical for AlPO4 zeotypes.[38] However,
the XRD data do not show any reflections corresponding to
the crystalline structure of AlPO4 materials (Figure 1). The total
amount of acid sites calculated from TPD NH3 data increases in
the order: TiP<BP<NbP<ZrP<AlP.
The FTIR spectra of adsorbed pyridine are presented in
Figure 3. Four main bands are observed in the region of pyri-
dine vibrations. In accordance with ref. [36], the bands at ap-
proximately 1592 and 1490 cmÀ1 are assigned to hydrogen-
bonded pyridine; the band at 1544 cmÀ1 is attributed to pyri-
dine protonated on Brønsted acid sites; whereas the band in
the region 1449–1461 cmÀ1 is assigned to pyridine adsorbed
on Lewis acid sites.
The results suggest that all the catalysts studied contain
both Brønsted and Lewis sites. The origin of the Brønsted acid-
ity of the samples is due to the terminal PÀOH groups along
with some contribution from MÀOH groups. The Lewis acidity
can be attributed to the coordinatively unsaturated Mn+
sites.[36a] The amount of Lewis and Brønsted sites in metal
phosphate catalysts usually depends on the amount of chemi-
ChemCatChem 2016, 8, 1346 – 1353
www.chemcatchem.org
1348
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ucts. Further transformations of these products may include
their condensation with BDO, leading to formation of cyclic
ethers, mainly dioxolane. Indeed, the kinetic curves of cyclic
ethers are typical for primary+secondary products formed as
a result of the interaction of the reactant with primary prod-
ucts (Figure 5).[39] Similar products were observed by Tçrçk
et al.[11] during BDO dehydration over heteropoly acids. The au-
thors suggested that MEK and MP are the intermediates in the
formation cyclic ethers.
The precise analysis of the kinetic curves at low conversion
levels allows the third primary product to be identified—
3B2OL, which shows a maximum at ꢀ40% conversion. The
origin of this maximum is most probably due to further trans-
formation of 3B2OL into BD, which behaves as a secondary
stable product (see inset in Figure 5). Similar conclusions about
the intermediate product in BD formation were reached by
Duan et al.,[18] who have investigated transformations of 3B2OL
and MEK over oxide catalysts and pointed out that the only
product of 3B2OL conversion is BD, whereas MEK does lead to
BD and gives mainly butene isomers, propylene, and ethylene.
The reaction network presented in Scheme 1 rationalizes fur-
ther the above observations. The elimination of the first water
molecule leads to the 3-butenium-2-ol ion, which can further
evolve by three possible routes:
Figure 4. TPD NH3 curves for the metal phosphates materials.
Dehydration of 2,3-butanediol
The dehydration reaction was studied at 200–4008C in the
WHSV range 1.5–500 ggÀ1 hÀ1. All the catalysts showed stable
catalytic performance with time on stream: the decrease in
conversion was less than 0.5% over 5 h of the experiment. It
should be mentioned that the high stability of the catalytic ac-
tivity demonstrated by phosphate catalysts is a significant ad-
vantage with respect to other solid catalysts, such as zeolites,
which show 15–20% decline in diol conversion within 3–4 h
on stream.
1) a hydride shift results in the formation of MEK;
2) a methyl shift leads to MP;
3) the elimination of a proton gives 3B2OL, which is further
converted into BD through elimination of a second water
molecule.
Reaction products and reaction network
The main reaction products observed over the phosphate cata-
lysts studied included MEK, MP, and BD. In addition, small
amounts of 3B2OL and C8–C12 heavy products including mainly
dioxolanes were observed. To determine the primary and sec-
ondary reaction products and to establish the reaction net-
work, a kinetic study was performed over the ZrP catalyst at
2508C in the WHSV range 1–400 hÀ1. The yields of different
products are plotted versus the conversion of BDO in Figure 5.
The kinetic analysis of these curves was performed on the
basis of the models calculated in ref. [39] for primary or secon-
dary, stable or unstable reaction products.
The results show that MEK and MP appear at very low con-
version levels and can be attributed to primary unstable prod-
Figure 5. Yields of reaction products versus BDO conversion with the ZrP
catalyst at 2508C.
Scheme 1. Reaction network.
ChemCatChem 2016, 8, 1346 – 1353
www.chemcatchem.org
1349
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Condensation of the primary products further leads to the
formation of heavy byproducts, including different cyclic
ethers.
Table 3. Product distribution at the same conversion level (2508C).
ZrP
11.2
BP
11.3
NbP
11.2
TiP
12.1
AlP
12.1
The reaction network presented in Scheme 1 is in line with
the results obtained over other types of catalysts: oxi-
des,[8b,17,18] zeolites,[13–15] and heteropoly acids.[11]
BDO conversion [%]
WHSV [hÀ1
]
400
300
400
300
500
Product selectivity [wt%]
MEK
2MP
3B2OL
BD
C8–C12
65.8
23.5
3.1
3.5
4.1
60.0
14.0
6.4
5.4
14.2
11.8
60.4
30.8
3.8
1.5
3.5
67.4
64.7
23.8
6.0
1.6
1.6
7.6
21.7
3.8
1.4
8.3
5.2
Effect of catalyst type
To compare the activity of different phosphate catalysts in the
BDO dehydration, the initial rates of BDO conversion have
been calculated from the initials slopes of the kinetic curves
obtained for each catalyst at 2508C. The WHSV values were ad-
justed for each catalyst to go down to conversions below
10%.
BD+3B2OL
6.6
5.3
ꢀ30 wt% is observed for the NbP sample, whereas the lowest
selectivity of ꢀ14 wt% is found in the case of BP.
The initial rates obtained can be compared in Table 2. The
results show that the activity of the phosphate catalysts in-
creases in the order: BP<TiP <ZrP=NbP<AlP. The highest ini-
tial rate of 1.9”10À4 mol(gs)À1 is observed for AlP, whereas the
The reaction pathway leading to BD (Scheme 1) gives only
a marginal contribution to the overall reaction products for
phosphate catalysts. The highest selectivity to 3B2OL and BD
combined reaches 12 wt% for the BP catalyst. Duan, Sato
et al.[8b,17,18] have reported that highly crystalline materials are
required to achieve high yields of BD with rare-earth-oxide cat-
alysts. We can speculate that similar requirements hold for the
phosphate catalysts as the BP sample is the only crystalline
material in the series of phosphates studied.
BP sample shows the lowest initial rate of 1.0”10À4 mol(gs)À1
.
The comparison of the catalytic properties of metal phos-
phates with other solid acids suggests that their activity in the
BDO dehydration is higher than in the case of oxides[8b] and
comparable with those of zeolites[14] and heteropoly acids.[11]
Indeed, high conversions of 90–100% can be reached over
phosphates already at 2508C, whereas rare-earth oxides re-
quire 325–3758C to reach 100% conversion.[8b] In the case of
zeolites catalysts,[14] high BDO conversions can be achieved at
2508C, whereas heteropoly acids, such as H3[PW12O40], show
high conversions already at 2008C.
The product distribution observed over various phosphate
catalysts is shown in Table 3 for the conversion level of 11–
12%. The same products, namely, MEK, MP, 3B2OL, BD, and
cyclic ethers are observed. The product selectivity is rather sim-
ilar for all the catalysts studied. The major reaction pathways
are pinacol-type rearrangements yielding MEK through a hy-
dride shift and MP through a methyl shift. Such product distri-
bution is typical for strong solid acids, such as zeolites and het-
eropoly acids, which catalyze dehydration at moderate temper-
atures.[11,14]
In addition to the typical dehydration products, a rather
high contribution of heavy byproducts resulting from the con-
densation of dehydration products with BDO (Scheme 1) is ob-
served with the phosphate catalysts. Boron and aluminum
phosphate samples show the highest selectivity to these prod-
ucts and, at the same time, these two samples give the lowest
contribution of MP. Conversely, NbP shows the highest selectiv-
ity to MP and the lowest content of C8–C12 heavy products.
One can suppose that C8–C12 heavy products result mainly
form the condensation of MP with BDO over phosphate cata-
lysts. This supposition is in agreement with Tçrçk et al.[11] who
have demonstrated that MP mainly contributes to dioxolane
formation with heteropoly acids.
Effect of reaction temperature
The content of MEK is between 60–67 wt% for all the phos-
phates, whereas the contribution of MP differs significantly for
the catalysts studied (Table 3). The highest selectivity of
The effect of temperature has been studied with AlP, which is
found to be the most active among the phosphate catalysts.
The temperature was varied between 200–4008C,
with WHSV being equal to 1.5 hÀ1 (Table 4). The re-
Table 2. Initial and regional rates of BDO conversion over phosphate catalysts
(2508C).
sults point to the fact that increasing the tempera-
ture from 200 to 2508C has a dramatic effect on BDO
conversion. Whereas at 2008C the conversion is only
20.7%, at 2508C it reaches 100%.
MEK is the main reaction product in the tempera-
ture range 250–4008C. At lower temperatures its con-
tent decreases owing to the significant contribution
of dioxolanes. A similar tendency is observed for MP,
the content of which is decreased by a factor of four
with a temperature drop from 250 to 2008C. These
observations point to a high contribution of conden-
AlP
1.9
ZrP
1.4
NbP
1.4
TiP
1.1
BP
Initial reaction rate
1.0
[10À4 mol(gs)À1
]
E1–E2
[kJmolÀ1
Regional amount of
acid sites [mmolgÀ1
Regional rates
[10À1 mol(mols)À1
]
]
]
80–120
120–155
155–220
755
164
12
221
417
203
229
316
117
200
171
41
170
209
59
1.2Æ0.2
6.6Æ0.9
À8.0Æ1.8
ChemCatChem 2016, 8, 1346 – 1353
www.chemcatchem.org
1350
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the catalysts in the selected series. The phosphate catalysts
used in this study satisfy all the above requirements.
According to the method of “regional analysis”,[40] the acidic
spectrum of a catalyst can be subdivided into several regions
with similar acidic strengths and, therefore, with the same cat-
alytic activity. With that in mind, the experimentally deter-
mined reaction rate for the i-th catalyst (vi) can be described
by Equation (1):
Table 4. Effect of temperature on BDO conversion over AlP catalyst
(WHSV=1.5 hÀ1).
T [8C]
BDO conversion [%]
200
20.7
250
100
280
100
400
100
Product selectivity [wt%]
MEK
2MP
3B2OL
BD
C8–C12
BD+3B2OL
45.1
3.0
3.7
1.6
46.6
5.3
78.1
13.4
0.0
6.9
1.5
72.0
16.9
0.0
9.4
1.7
69.2
13.8
0.0
13.3
1.8
X
p
*
vi ¼
ð
j¼1Þv0j sij
ð1Þ
6.9
9.4
13.3
where sij is the number of acid sites in the j-th region of the i-
th catalyst, v0j is the regional rate of the acid site in the j-th
region, and p is the number of regions.[40] As vi and sij can be
experimentally determined for a set of catalysts, the regional
rates can be easily estimated by means of the method of least
squares, when the number of catalysts employed is greater
than the number of regions, p. The regional rates determined
in such a way will give information about the activity of the
acid sites with different strengths.
sation reactions at low reaction temperatures, which results in
the transformation of MEK and MP into dioxolanes. It should
be mentioned that high yields of dioxolanes in the tempera-
ture range of 150–2008C were observed previously by Tçrçk
et al.[11] with heteropoly acids.
The increase in reaction temperature results in persistent
growth in BD selectivity, which is accompanied by a decrease
in MEK content. This result is in line with thermodynamic cal-
culations performed by Makshina et al.[5] and shows that the
probability of BD formation increases with temperature. The
highest BD selectivity of 13 wt% is reached at 4008C, whereas
the highest selectivity to MEK (78 wt%) is achieved at 2508C.
3B2OL is observed only at 2008C, when the conversion was
only 20%. At higher conversion levels, 3B2OL is most probably
converted into BD. The same tendency was observed by Duan
et al.[18] over rare-earth metal oxides.
To apply this approach to the phosphate catalysts, their TPD
NH3 spectra were subdivided into three regions, as shown in
Figure 4. The first region corresponds to weak acid sites with
activation energies for NH3 desorption of 80–120 kJmolÀ1, the
second one includes acid sites of medium strength (120–
155 kJmolÀ1), whereas the third one corresponds to strong
acid sites (155–220 kJmolÀ1). The activation energies of NH3
desorption were calculated as proposed in ref. [41]. The re-
gional contents of the acid sites determined for each catalyst
are presented in Table 2 along with the initial rates of BDO de-
hydration over these catalysts. By using the method of least
squares of the difference between the calculated (vic) and ex-
perimental (vie) initial rates, the set of regional rates was calcu-
lated. The results are shown in Table 2. To estimate the statisti-
cal significance of analysis, a Fisher test was conducted; it
showed that the results are significant at the 5% level. The cor-
relation coefficient between vic and vie was found to be 0.97,
which suggests that the model sufficiently describes the exper-
imental data.
Correlation with acidity
Comparison of the catalytic activity of the various phosphate
catalysts (Table 2) with their acidic properties (Figures 3 and 4,
and Table 1) does not show any direct correlation, neither with
the amount of Brønsted and Lewis sites (Figure 3), nor with
the total amount of sites determined by TPD NH3 (Table 1).
Indeed, the ZrP and NbP catalysts show the same initial reac-
tion rates (Table 2), whereas the number of acid sites is quite
different for these catalysts (Table 1). Likewise, the BP and TiP
samples contain the same amount of sites (Table 1), whereas
the initial rates found for these catalysts are quite different
(Table 2).
The analysis of the calculated regional rates suggest that
weak (80–120 kJmolÀ1) and medium (120–155 kJmolÀ1) acid
sites are preferable for 2,3-butanediol dehydration, the latter
being more active than the former by a factor of five. Surpris-
ingly, the strongest acid sites show a negative value of regional
rate. One of the possible reasons for this observation could be
due to fast deactivation of these sites and, therefore, the nega-
tive influence on the overall catalytic activity.
Analysis of the literature data suggests that both Brønsted
and Lewis acid sites can participate in BDO dehydration.[5,15]
Furthermore, it appears from the literature survey that the
nature of the acid sites is not as important as their
strength.[5,11,14] To verify the effect of acid sites with different
strengths on the BDO dehydration with phosphate catalysts,
the method of “regional analysis” proposed by Yoneda[40] has
been used. This method is based on the assumption that the
reaction rate over an acidic site of a catalyst is determined
solely by the acid strength of this site. The main requirements
for the application of this method involve i) similar types of
acid sites in the selected series of catalysts; ii) a sufficient
number of catalysts; and iii) different acid site distribution for
Conclusions
Zirconium, niobium, aluminum, boron, and titanium phos-
phates are shown to be active catalysts for 2,3-butanediol de-
hydration in the temperature range 200–4008C. The initial
rates of butane formation are found to be within 1.0”10À4
–
1.9”10À4 mol(gs)À1 at 2508C, aluminum phosphate being the
most active catalyst. The activity of phosphates is found to be
ChemCatChem 2016, 8, 1346 – 1353
www.chemcatchem.org
1351
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
24 h at 1108C. The product was calcined in a flow of dry air at
5008C for 3 h.
comparable with those of zeolite catalysts. However, in terms
of the stability of catalytic activity with time on stream, these
catalytic systems are shown to be superior with respect to
other solid catalysts.
TiP: Titanium phosphate was synthesized according to the proce-
dure described in ref. [25]. Unlike the other samples, an organic
compound, Ti(OBu)4, was used as the Ti precursor. Ti(OBu)4 (5 g)
was diluted with anhydrous EtOH (99%) to prevent TiO2 formation.
The resultant mixture was added to a H3PO4 solution (300 mL,
0.1m, at 258C, under stirring). The mixture was stirred at ambient
temperature for 1 h and at 808C for 24 h. The white precipitate
was filtered and washed with distilled water. The product was
dried at 110 8C and calcined in a flow of dry air at 5008C for 3 h.
The kinetic study pointed to three main reaction pathways
operating over the phosphate catalysts: 1) pinacol rearrange-
ment through a hydride shift, yielding methyl ethyl ketone;
2) pinacol rearrangement through a methyl shift, resulting in
2-methyl-propanal; 3) 1,2-elimination, leading to 3-butene-2-ol,
which is further dehydrated into 1,3-butadiene by elimination
of a second water molecule. Methyl ethyl ketone and 2-
methyl-propanal may further interact with 2,3-butanediol, re-
sulting in the formation of heavy byproducts, including differ-
ent cyclic ethers.
BP: Boron phosphate was synthesized by using an organic precur-
sor, tri-iso-propylborate (B(iPrO)3). B(iPrO)3 was mixed with ortho-
phosphoric acid (85%). The B/P molar ratio was equal to 1. The re-
sultant mixture was heated to 1208C until the complete evapora-
tion of water and iPrOH was achieved. The white solid was dried
under vacuum at 110 8C for 3 h and calcined in a flow of dry air at
5008C for 3 h.
The major reaction pathway with the phosphates involves
pinacol rearrangement to methyl ethyl ketone, the selectivity
for which being 60–78 wt% with the catalysts studied. The
highest yield of 78 wt% methyl ethyl ketone is obtained with
AlP at 2508C and WHSV=1.5 hÀ1. The 1,2-elimination gives
a minor contribution to the overall reaction pathway. The max-
imum selectivity to 3-butene-2-ol and 1,3-butadiene at 2508C
reaches 11 wt% with boron phosphate, which is the only crys-
talline material among the catalysts studied.
Catalysts characterization
X-ray fluorescence analysis was used to determine the chemical
composition of the samples. Measurements were performed with
the X-ray fluorescence scanning crystal diffraction spectrometer
“Spectroscan MaksGF2E” “Spectron”. The spectrometer was
equipped with a sharply focused X-ray tube with a copper anode.
The correlation obtained between the catalytic activity for
the dehydration of 2,3-butanediol and the acid strength distri-
bution over the phosphate catalysts points to the fact that
acid sites with medium strength (120–155 kJmolÀ1) are the
The structure and phase compositions of the samples were deter-
mined by X-ray analysis. XRD patterns were recorded with a diffrac-
tometer “Bruker D2 Phaser” with CuKa in the 2q range 5–508 with
steps of 0.18.
most active for dehydration; weaker sites (80–120 kJmolÀ1
)
show lower activity; whereas very strong sites (155–
220 kJmolÀ1) have a negative effect on the activity.
Low-temperature N2 adsorption–desorption measurements were
performed with an automated porosimeter Micrometrics ASAP
2000. Before the measurements, all samples were degassed at
2008C for 2 h. Isotherms were measured at 77 K.
IR spectra were recorded with a Nicolet ProtØgØ 380 FTIR spec-
trometer in transmission mode at 4 cmÀ1 optical resolution. Prior
to the measurements, the catalysts (20 mg) were pressed in self-
supporting discs and activated in the IR cell attached to a vacuum
line at 4008C for 4 h. The adsorption of pyridine (Py) was per-
formed at 1508C for 40 min. The excess pyridine was further evac-
uated at 1508C for 40 min.
Experimental Section
Catalyst preparation
Metal phosphate catalysts were prepared through the interaction
of different metal-containing precursors with phosphoric acid, fol-
lowing the procedures described in the literature.[22–26]
ZrP: Zirconium phosphate was prepared by the treatment of
ZrOCl2·8H2O aqueous solution (1m) with 0.5m solution of H3PO4.
The Zr/P molar ratio was equal to 0.5. The mixture was stirred for
30 min and the resultant gel was filtered and washed with distilled
water. Afterwards, the sample was dried at 110 8C and calcined at
4008C in a flow of dry air for 3 h.
Temperature-programmed desorption (TPD) of NH3 was performed
with a USGA-101 (“UNISIT”). Prior to NH3 adsorption, the samples
were calcined in a flow of dry air at 4008C for 1 h and subsequent-
ly in a flow of dry nitrogen for 1 h and cooled to ambient tempera-
ture. The adsorption was carried out for 30 min in a flow of NH3 di-
luted with N2 (1:1). The physisorbed ammonia was removed in
a flow of dry He at 1008C for 1 h. Typical TPD experiments were
carried out in the temperature range 50–8008C in a flow of dry He
NbP: Niobium phosphate was synthesized from niobium oxide
Nb2O5·nH2O (5 g), which was treated with diluted orthophosphoric
acid (0.5m) at 808C. The resultant mixture was stirred for 7 h. After-
wards, it was cooled down to ambient temperature, filtered, and
washed with distilled water until a pH value of 5 was obtained.
The product was dried and calcined at 4008C in a flow of dry air
for 3 h.[23]
(30 mLminÀ1). The rate of heating was 78CminÀ1
.
Catalyst evaluation
The dehydration of BDO was carried out in a quartz fixed-bed con-
tinuous-flow reactor. The catalysts were preheated at 2508C in
a flow of nitrogen before the reaction. The reaction was carried
out under atmospheric pressure in the temperature range 200–
4008C in a flow of nitrogen as a carrier gas. The WHSV of BDO was
varied from 1.5 to 500 ggÀ1 h. The products were analyzed by gas
chromatography with a 40 m FFAP capillary column. Methane and
AlP: Aluminum phosphate was prepared by the addition of aque-
ous ammonia (25 wt%) under continuous stirring to an aqueous
solution containing equimolar quantities of aluminum chloride and
orthophosphoric acid (1m).[24] The ammonia was added until pH=
7.0 was obtained. The resultant precipitate was aged (18 h, 258C),
filtered, washed with distilled water several times, and dried for
ChemCatChem 2016, 8, 1346 – 1353
www.chemcatchem.org
1352
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[13] J. Lee, J. B. Grutzner, W. E. Walters, W. N. Delgass, Stud. Surf. Sci. Catal.
2000, 130, 2603–2608.
[14] W. Zhang, D. Yu, X. Jia, H. Huang, Green Chem. 2012, 14, 3441–3450.
[15] A. Molnµr, I. Bucsi, M. Bartok, Stud. Surf. Sci. Catal. 1988, 41, 203–210.
[16] a) M. E. Winfield, J. C. S. I. R. (Aust.) 1945, 18, 412–423; b) M. Winfield,
Aust. J. Chem. 1950, 3, 290–305.
[17] a) S. Sato, F. Sato, H. Gotoh, Y. Yamada, ACS Catal. 2013, 3, 721–734;
b) S. Sato, R. Takahashi, T. Sodesawa, N. Honda, H. Shimizu, Catal.
Commun. 2003, 4, 77–81.
[18] H. Duan, Y. Yamada, S. Sato, Appl. Catal. A 2015, 491, 163–169.
[19] a) A. Clearfield, D. S. Thakur, Appl. Catal. 1986, 26, 1–26; b) S. Sebtia, M.
Zahouily, H. B. Lazrek, J. A. Mayoral, D. J. Macquarrie, Curr. Org. Chem.
2008, 12, 203–232; c) A. Tada, Phosphorus Res. Bull. 2004, 15, 100–109;
d) G. Aguilar-Armenta, M. E. Patino-Iglesias, J. Jimenez-Jimenez, E. Rodrı-
guez-Castellon, A. Jimenez-Lopez, Langmuir 2006, 22, 1260–1267.
[20] V. L. Sushkevich, V. V. Ordomsky, I. I. Ivanova, Appl. Catal. A 2012, 441–
442, 21–29.
dioxane were used as internal standards for the analysis of the gas
and liquid products, respectively. The products were identified by
using a GC-mass spectrometer “Thermo” equipped with quadru-
pole mass detector “DSQ II” and gas chromatograph “Trace GC”
with capillary column “HP-5 ms”. The data were collected after the
reaction reached a steady state.
Selectivity was defined as weight fraction of product in all prod-
ucts (not including water). Initial reaction rates were determined
from the initial slopes of the kinetic curves (BDO conversion versus
contact time) and defined as the number of moles of BDO convert-
ed into products per hour and per gram of the catalyst.
Acknowledgements
[21] J. Maurin (Compagnie Francaise de Raffinage) US Patent 3758612,
1973.
[22] V. F. D. lvaro, R. A. W. Johnstone, J. Mol. Catal. A 2008, 280, 131–141.
[23] S. Okazaki, A. Kurosaki, Catal. Today 1990, 8, 113–122.
[24] G. J. Hutchings, I. D. Hudson, D. Bethell, D. G. Timms, J. Catal. 1999, 188,
291–299.
The authors thank the Russian Science Foundation for financial
support (Grant no. 14-23-00094).
Keywords: 1,3-butadiene
·
2,3-butanediol dehydration
·
methyl ethyl ketone · phosphate catalysts
[25] H. Fei, X. Zhou, H. Zhou, Z. Shen, P. Sun, Z. Yuan, T. Chen, Microporous
Mesoporous Mater. 2007, 100, 139–145.
[1] a) M. Kçpke, C. Mihalcea, F. Liew, B. Al-Sinawi, S. D. Simpson, J. H. Tizard,
M. S. Ali, J. J. Conolly, Appl. Environ. Microbiol. 2011, 77, 5467–5475;
b) M. Voloch, N. B. Jansen, M. R. Ladisch, G. T. Tsao, R. Narayan, V. W.
Rodwell in Comprehensive Biotechnology, Vol. 45 (Eds.: M. Moo-Young,
H. W. Blanch, S. Drew, D. I. C. Wang), Pergamon Press, Oxford, 1985,
pp. 933–947; c) C.-F. Huang, Y.-F. Jiang, G.-L. Guo, W.-S. Hwang, Biore-
sour. Technol. 2013, 135, 446–453; d) M. Motwani, R. Seth, H. F. Dagina-
wala, R. Khanna, Bioresour. Technol. 1993, 44, 187–195; e) L.-Q. Jiang, Z.
Fang, F. Guo, L. Yang, Bioresour. Technol. 2012, 107, 405–410; f) A.
Multer, N. McGraw, K. Hohn, P. Vadlani, Ind. Eng. Chem. Res. 2013, 52,
56–60.
[26] J. B. Moffat, J. F. Neeleman, J. Catal. 1973, 31, 274–277.
[27] K. Segawa, Mater. Chem. Phys. 1987, 17, 181–200.
[28] A. La Ginestra, Mater. Chem. Phys. 1987, 17, 161–179.
[29] a) R. Vetrivel, B. Viswanathan, Surf. Technol. 1984, 22, 1–8; b) N. N. Gon-
chakova, N. G. Ivanova, L. Y. Kitaev, L. A. Novikova, A. A. Kubasov, V. Z.
Sharf, K. V. Topchieva, Petrol. Chem. USSR 1981, 21, 133–142.
[30] International Centre for Diffraction Data, ICDD PDF2 03-0332.
[31] a) L. Liang, Y. Xu, L. Zhang, Y. Sheng, J. Opt. Soc. Am. B 2007, 24, 1066–
1074; b) M. R. Ahsan, M. A. Uddin, M. G. Mortuza, Indian J. Pure Appl.
Phys. 2005, 43, 89–99; c) S. W. Lee, R. A. Condrate Sr., J. Mater. Sci.
1988, 23, 2951–2959.
[2] J. Daniell, M. Kçpke, S. D. Simpson, Energies 2012, 5, 5372–5417.
[3] M. Anvari, M. R. S. Motlagh, J. Biomed. Biotechnol. 2011, 2011, 1–7.
[4] J. van Haveren, E. L. Scott, J. Sanders, Biofuels Bioprod. Biorefin 2008, 2,
41–57.
[5] E. V. Makshina, M. Dusselier, W. Janssens, J. Degreve, P. A. Jacobs, B. F.
Sels, Chem. Soc. Rev. 2014, 43, 7917–7953.
[32] C. Dayanand, G. Bhikshamaiah, V. J. Tyagaraju, M. Salagram, A. S. R.
Krishna Murthy, J. Mater. Sci. 1996, 31, 1945–1967.
[33] J. C. Yu, L. Zhang, Z. Zheng, J. Zhao, Chem. Mater. 2003, 15, 2280–2286.
[34] a) S. Chen, M. Ye, H. Chen, X. Yang, J. Zhao, J. Inorg. Organomet. Polym.
2009, 19, 139–142; b) A. Baykal, M. Kizilyalli, M. Toprak, R. Kniep, Turk. J.
Chem. 2001, 25, 425–432.
[6] P. Lanzafame, G. Centi, S. Perathoner, Chem. Soc. Rev. 2014, 43, 7562–
7580.
[35] a) T. Armaroli, G. Busc, C. Carlini, M. Giuttari, A. M. Raspolli Galletti, G.
Sbrana, J. Mol. Catal. A 2000, 151, 233–243; b) Z. Ghezelbash, D.
Ashouri, S. Mousavian, A. H. Ghandi, Y. Rahnama, Bull. Mater. Sci. 2012,
35, 925–931.
[36] a) G. Busca, Phys. Chem. Chem. Phys. 1999, 1, 723–736; b) A. Travert, A.
Vimont, A. Sahibed-Dine, M. Daturi, J. Lavalley, Appl. Catal. A 2006, 307,
98–107; c) C. A. Emeis, J. Catal. 1993, 141, 347–354.
[37] a) G. Bagnasco, P. Ciambelli, A. La Ginestra, M. Turco, Thermochim. Acta
1990, 162, 91–97; b) M. Turco, P. Ciambelli, G. Bagnasco, A. La Ginestra,
J. Catal. 1989, 117, 355–361.
[38] S. G. Hedge, P. Ratnasamy, L. M. Kustov, V. B. Kazansky, Zeolites 1988, 8,
137–141.
[7] a) Z. Liu, W. Huo, H. Ma, K. Qiao, Chin. J. Chem. Eng. 2006, 14, 676–684;
b) L. H. Slaugh, G. W. Schoenthal, J. D. Richardson (Shell Oil Company
Houston, TX) US Patent 4453015, 1984; c) J. R. Black, J. Yang, J. L. Bue-
chele (Shell Oil Company, HoustonTX), US patent 7282613 B2, 2007;
d) S. M. Aldoshin, P. E. Matkovsky, Z. P. Pai, V. N. Parmon, L. N. Russiyan,
V. N. Troitsky, WO 2007/055608A2, 2007.
[8] a) M. Magnin-Lachaux, Z. Tan, B. Liang, E. Negishi, Org. Lett. 2004, 6,
1425–1427; b) H. Duan, D. Sun, Y. Yamada, S. Sato, Catal. Commun.
2014, 48, 1–4.
[9] a) P. K. Frolich, C. E. Morrell, Chem. Eng. News 1943, 21, 1138–1145;
b) N. L. Morrow, Environ. Health Perspect. 1990, 86, 7–8; c) Wm. C.
White, Chem.-Biol. Interact. 2007, 166, 10–14; d) M. Arakawa, H. Yoshio-
ka, K. Nakazawa (Japan Synthetic Rubber Co.Ltd, Tokyo) US Patent
4504692, 1985.
[39] A. N. Ko, B. W. Wojciechowski, Prog. React. Kinet. 1983, 12, 201–262.
[40] Y. Yoneda, J. Catal. 1967, 9, 51–56.
[41] V. V. Yushchenko, Zh. Fiz. Khim. 1997, 71 (4), 628–632.
[10] I. Bucsi, A. Molnar, M. Bartok, Tetrahedron 1994, 50, 8195–8202.
[11] B. Tçrçk, I. Bucsi, T. Beregszaszi, I. Kapocsi, A. Molnar, J. Mol. Catal. A
1996, 107, 305–311.
Received: December 23, 2015
[12] K. Ohara, S. Furuhashi, Nippon Nogei Kagaku Kaishi 1942, 18, 143–150.
Published online on February 23, 2016
ChemCatChem 2016, 8, 1346 – 1353
www.chemcatchem.org
1353
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim