Synthesis of isoprene from formaldehyde and isobutene over phosphate catalysts

Article abstract of DOI:10.1016/j.apcata.2012.06.034

Vapor phase Prins condensation of isobutene with formaldehyde has been studied over boron (BP), aluminum (AlP), titanium (TiP), zirconium (ZrP) and niobium (NbP) phosphates. The catalysts were characterized by elemental analysis, X-ray diffraction, low-te

Full text of DOI:10.1016/j.apcata.2012.06.034

Applied Catalysis A: General 441–442 (2012) 21–29
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Synthesis of isoprene from formaldehyde and isobutene over phosphate catalysts
V.L. Sushkevicha, V.V. Ordomskya,b, I.I. Ivanovaa,c,∗
a Department of Chemistry Lomonosov, Moscow State University, Leninskye Gory 1, bld. 3, 119991 Moscow, Russia
b “UNISIT” LLC, Leninskye Gory 1, bld. 75-V, 119991 Moscow, Russia
c A.V. Topchiev Institute of Petrochemical Synthesis, Russian Academy of Science, Leninskiy Prospect 29, 119991 Moscow, Russia
a r t i c l e i n f o
a b s t r a c t
Article history:
Vapor phase Prins condensation of isobutene with formaldehyde has been studied over boron (BP),
aluminum (AlP), titanium (TiP), zirconium (ZrP) and niobium (NbP) phosphates. The catalysts were
characterized by elemental analysis, X-ray diffraction, low-temperature adsorption of nitrogen, TPD of
ammonia and FTIR of adsorbed pyridine. The conversion of isobutene and formaldehyde over phos-
phates was found to increase in the following order: BP ꢀ AlP < TiP < ZrP < NbP, while the selectivity to
isoprene showed the reverse tendency. The main side reaction was formaldehyde decomposition into
carbon monoxide and dihydrogen. It is proposed that isoprene is formed on the Brønsted acid sites of
the catalyst, while M␦+ O␦− ion pairs are responsible for competitive carbon monoxide formation. The
high resistance of ZrP and NbP catalysts to deactivation is accounted for by the in situ restoration of the
Brønsted acid sites in the presence of water directly during the time course of the reaction.
© 2012 Elsevier B.V. All rights reserved.
Received 13 March 2012
Received in revised form 19 June 2012
Accepted 20 June 2012
Available online 16 July 2012
Keywords:
Prins condensation
Isoprene
Phosphates
Deactivation
1. Introduction
(4) in many cases, these by-products do not find application and
have to be just burned; (5) the first step of the process uses sul-
Condensation of isobutene with formaldehyde is among the
most important reactions of Prins addition [1–4], since it leads to
isoprene, which is a strategically important monomer for the pro-
duction of substitutes for natural rubbers. The increasing demand
in the synthetic rubber attracts great attention to this reaction. To
date, the main route for isoprene synthesis from formaldehyde and
isobutene is based on the two-step industrial process [5–8]. The
first step involves a liquid-phase condensation of isobutylene with
formaldehyde catalyzed by aqueous sulfuric acid and the second
-vapor-phase decomposition of 4,4-dimethyldioxane-1,3 (DMD)
into isoprene over solid phosphoric acid-catalysts:
furic acid as a catalyst, which has significant drawbacks related to
the environmental and corrosion problems. Therefore the future
challenge in this area will require the development of the one-step
selective process based on heterogeneous catalysts. The hetero-
geneous catalysts studied so far in formaldehyde and isobutene
condensation include zeolites, phosphates, sulfates and oxide cat-
alysts [9–14].
Dumitriu et al. investigated condensation of isobutene with
formaldehyde over zeolites in the gas phase [9,10]. The authors
pointed to the crucial role of the acid sites strength in the selec-
tive synthesis of isoprene. In this respect, weak Brønsted acid sites
were found to be the most efficient and the highest selectivity of
99–100% was obtained over boralites and ferrisilicates with MFI
structure. Unfortunately, all the measurements were performed in
a pulse catalytic reactor and the results are difficult to compare
with those obtained in continuous flow systems.
CH3
CH3
2 CH2O
C
CH2
O
CH3
CH3
CH3
O
CH2O
H2O
O
CH3
O
CH3
Krzywicki et al. [12] studied isobutene condensation with
formaldehyde over H3PO4/Al2O3 catalysts with different contents
of phosphoric acid. The authors reported the yields of isoprene
of 22% over the catalysts with high content of H3PO4 (>140 mmol
H3PO4/100 g Al2O3).
This process has many drawbacks: (1) two steps are econom-
ically and technologically unfavorable; (2) the selectivity of the
condensation as well as DMD decomposition is very low and the
selectivity of the overall process does not exceed 50%; (3) the high
amount of by-products requires complicated separation processes;
Much better results in terms of both conversion and selectivity
were obtained over sulfate and phosphate catalysts. Dang et al. [13]
suggested that doping of the strong acid catalysts such as CuSO4
with the basic oxides decreases their acidity and results in the for-
mation of new more selective in isoprene synthesis active sites.
The authors reported the formaldehyde conversion of 87% and the
∗
Corresponding author at: Department of Chemistry Lomonosov, Moscow State
University, Leninskye Gory 1, bld. 3, 119991 Moscow, Russia. Tel.: +7 495 939 3570;
fax: +7 495 939 3570.
E-mail address: iiivanova@phys.chem.msu.ru (I.I. Ivanova).
0926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.apcata.2012.06.034

22
V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
selectivity to isoprene of 73% over CuSO4/SiO2 catalyst doped with
magnesia. Ai [14] investigated formaldehyde condensation with
tert-butyl alcohol over vanadium, molybdenum and tungsten phos-
phates. Vanadium-based catalyst demonstrated the highest activity
and showed the yield of isoprene up to 50%.
Although sulfate and phosphate based catalytic systems showed
rather high catalytic activity and selectivity in the direct synthe-
sis of isoprene, they demonstrated very short life times and rapid
deactivation due to fast coke formation.
until pH 5 was attained. The slightly yellow precipitate was dried
and calcined at 573 K in a flow of air for 3 h [19,20].
2.1.6. 0.2Na/NbP and 0.5Na/NbP
0.2Na/NbP and 0.5Na/NbP were prepared by threefold ion
exchange of NbP at 323 K with 1 M solutions of sodium acetate and
sodium carbonate, respectively.
This paper aims to clarify the reasons of the deactivation of phos-
phate catalysts and to find the ways to prolong their lifetime. For
this purpose, the activity and stability of boron, aluminum, tita-
nium, zirconium and niobium phosphates have been studied in
formaldehyde condensation with isobutene and the active sites
responsible for the stable catalyst performance have been eluci-
dated by IR spectroscopy.
2.2. Catalyst characterization
The chemical composition of the samples was deter-
mined by inductively coupled plasma (ICP) spectroscopy.
Sorption–desorption isotherms of nitrogen were measured at
77 K using an automated porosimeter (Micrometrics ASAP 2000).
The XRD patterns were recorded with a DRON-3 M diffractometer,
applying Cu K␣ radiation.
Temperature programmed desorption of ammonia (TPD-NH3)
was performed in a homemade set-up equipped with TC detector.
Prior to NH3 adsorption, the samples were calcined in situ in a flow
of dry air at 673 K for 1 h and, subsequently, in a flow of dry nitrogen
for 1 h and cooled down to ambient temperature. The adsorption
was carried out for 30 min, at ambient temperature in a flow of
NH3 diluted with He (1/1). Subsequently, the physically adsorbed
ammonia was removed in a flow of dry He at 373 K for 1 h. Typ-
ical TPD experiment was carried out in the temperature range of
295–900 K in a flow of dry He (30 mL/min). The rate of heating was
7 K/min.
IR spectra were recorded on a Nicolet Protégé 380 FT-IR spec-
trometer at 4 cm−1 optical resolution. Prior to the measurements,
20 mg of a catalyst were pressed in self-supporting disc and acti-
vated in the IR cell attached to a vacuum line at 673 K for 4 h. The
adsorption of pyridine (Py) was performed at 423 K for 30 min. The
excess of probe molecules was further evacuated at 423 K for 0.5 h.
The adsorption–evacuation was repeated several times until no
changes in the spectra were observed. In a part of the experiments
the adsorption of water was carried out directly after adsorption
and evacuation of Py at 423 K. The excess of water was evacuated
at 373 K for 1 h.
2. Experimental
2.1. Catalysts preparation
Boron, aluminum, titanium, zirconium and niobium phosphates
were prepared using different procedures proposed in the litera-
ture [15–19].
2.1.1. BP
Boron phosphate was prepared by heating of phosphoric acid
(93 mL, 85 wt.%) with boric acid (100 g) at 333 K for 1 h. Water
(100 mL) was then added and the mixture was refluxed for 5 h, dried
(383 K, 16 h) and calcined in air (623 K, 4 h) [15].
2.1.2. AlP
Aluminum phosphate was prepared by the slow addition of
aqueous ammonia (25 wt.%, 278 K), with continuous stirring, to
an aqueous solution containing equimolar quantities of aluminum
chloride and orthophosphoric acid (1 M, 278 K). The ammonia addi-
tion was continued until pH 7.0 was attained. The white precipitate
was aged (18 h, 293 K) and collected by filtration, washed several
times with water and dried (24 h, 373 K). The solid was calcined in
air at 773 K for 3 h [16].
TG experiments over used catalysts were performed on “TA SDT
Q600” instrument. Temperature programmed oxidation was car-
ried out in a flow of dry air (200 mL/min) in the temperature range
of 293–1073 K with the rate of heating of 10 K/min.
2.1.3. TiP
Titanium phosphate was prepared by fast addition of an aque-
ous solution of TiCl4 (7.5 g) in 1 M HCl (20 mL) to orthophosphoric
acid (5.92 g in 20 mL water) at the intensive stirring. The resulting
precipitate was aged for 2 h at ambient temperature and then fil-
tered and washed with large amount of water. The solid was dried
(373 K) and calcined at 673 K for 3 h in a flow of air [17].
2.3. Catalyst evaluation
Condensation of formaldehyde with isobutene was studied in a
microreactor system equipped with quartz fixed bed tubular reac-
tor. In a typical experiment, 2 g of catalyst (particle size 0.5–1 mm)
was loaded into the reactor. Before the experiment, the catalysts
were heated in situ in a flow of nitrogen at 673 K. Then, the temper-
ature was decreased to the reaction temperature.
Formalin containing 37 wt.% of formaldehyde, 3 wt.% of
methanol, 60 wt.% of water was used as formaldehyde source. Liq-
uid formalin was delivered by a syringe pump, while isobutene
was supplied by a mass flow meter (Bronkhorst). The reaction was
carried out under atmospheric pressure, isobutene/formaldehyde
molar ratio was 7, the reaction temperature was within 523–623 K
and weight hourly space velocity (WHSV) of formaldehyde and
isobutene was varied in the range of 0.3–3.0 g/(g h). The products
were analyzed by gas chromatography. Porapak Q column (2 m)
was used for the analysis of CO and CO2 and capillary column with
SE-40 (40 m) – for the analysis of the other products. Methane was
used as an internal standard. Formaldehyde content was deter-
mined by the titration with Na2SO3. The selectivity toward different
2.1.4. ZrP
Phosphoric acid (23.3 g) in water (470 mL) was added quickly
to a rapidly stirred solution of zirconyl chloride (22.5 g) in water
(140 mL) at ambient temperature. The resulting gel was washed
with distilled water until the supernatant liquid reached pH 4. The
final gel was washed five times with ethanol (5× 300 mL). Then the
suspension was filtered and dried under vacuum (373 K/10 Torr)
for 1 h to give ZrP. The solid was calcined at 673 K for 3 h in a flow
of air [18].
2.1.5. NbP
Niobium phosphate was synthesized by the treatment of
hydrated niobium oxide (5 g) with diluted orthophosphoric acid
(9 g in 160 mL of H2O) at 343 K for 7 h. After the treatment, the
reaction mixture was cooled down, filtered and washed with water

V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
23
Table 1
Catalysts characteristics.
Sample
Elemental analysis Surface
Acidity (TPD NH3)
area, m2/g
M/P
Na/P
␮mol/g
␮mol/m2
BP
AlP
TiP
ZrP
NbP
0.2Na/NbP
0.5Na/NbP
1.0
0
0
0
0
0
0.18
0.55
5
140
80
210
70
56
1097
762
1725
736
-
11.2
7.8
9.5
8.2
10.5
-
1.11
0.87
0.60
2.44
2.44
2.44
70
70
products formed from formaldehyde (SjF ) and isobutene (SjI) were
calculated as follows:
kF(mj/Mrj) × 100%
j
Fig. 2. Pore size distribution in the phosphate catalysts.
SjF
=
,
nreacted
formaldehyde
kI(mj/Mrj) × 100%
preparation. They are most probably composed of the mixture of
phosphates and hydrophosphates with different M/P ratios.
The ZrP catalyst has the highest content of phosphorus, which
can be accounted for by the formation of double-dimension layered
structure, typical for zirconium phosphates. In such materials each
layer consists of the planes of Zr atoms bridged through phosphate
groups, which alternate above and below Zr atom planes. The dis-
tinct feature of the layered materials is that the layers are connected
by van der Waals forces and electrostatic potentials [22,23].
The lowest content of phosphorus has NbP sample due to
the incomplete transformation of niobium oxide into phosphate
[19,20]. All the attempts to increase the phosphorus content in the
samples by using higher concentration of orthophosphoric acid or
higher temperature lead to the drastic loss of the surface area of
the catalyst.
j
SjI =
,
nreacted
isobutene
where mj is the weight of product j in the reaction mixture, Mrj is
the molecular weight of product j, kjF is the amount of formaldehyde
molecules required for product j formation, kjI amount of isobutene
molecules required for product j formation.
Yield of isoprene was calculated as follows:
Y = XF × SF
isoprene
where XF is a conversion of formaldehyde.
3. Results and discussion
3.1. Catalysts structure and composition
The main characteristics of the catalysts are presented in Table 1
and Figs. 1–3. BP material has B/P molar ratio close to 1 and very
low surface area, which is typical for crystalline phosphate struc-
tures [21]. XRD pattern points to the formation of the only phase
corresponding to crystobalite structure (Fig. 1).
All the other catalysts show XRD patterns typical for amorphous
materials (Fig. 1). Nitrogen adsorption–desorption data (Table 1,
Fig. 2) point that amorphous phosphates have high surface area
˚
and are characterized by rather large pores (>100 A) with very
broad distribution of pore sizes. The exception is TiP material,
which has rather narrow pore size distribution in the range of
˚
50–120 A (Fig. 2). The amorphous phosphates have quite different
M/P molar ratios, which is apparently due to different ways of their
Fig. 3. FTIR spectra in the region of structural stretching vibrations of phosphate
Fig. 1. XRD patterns of the phosphate catalysts.
catalysts.

24
V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
The ion exchange of NbP sample does not lead to any changes the
Nb/P molar ratio. Therefore we can assume that the ion exchanged
materials preserve the structure of parent niobium phosphate. The
content of sodium introduced is given in Table 1. The results show
that ion exchange with sodium carbonate leads to the deep sub-
stitution of the acidic protons with sodium cations, whereas the
treatment with CH3COONa substitutes only part of the available
positions.
The IR spectra of phosphates are shown in Fig. 3. A broad band in
the range of 1000–1200 cm−1 can be assigned to O
P O asymmet-
ric stretching vibrations of phosphate or polyphosphate species.
This band gradually shifts from 1184 cm−1 for BP to 1028 cm−1 for
NbP. This effect is in line with the decreasing electronegativity of
the metals and the simultaneous increasing of the ionic character
of the bond between metal and phosphate species in the following
order: B < Al < Ti < Zr < Nb.
For AlP and NbP samples, a broad band in the range of
500–700 cm−1 is observed along with the band attributed to phos-
phates. This band corresponds to the stretching vibrations of
bridging M
O M bonds in oxide phases and indicates that these
samples are the mixture of metal oxide and phosphate phases
[24,25]. This observation is line with the chemical analysis of AlP
and NbP samples (Table 1).
Fig. 5. FTIR spectra of pyridine adsorbed over phosphate catalysts after evacuation
at 423 K.
3.2. Acid properties of phosphates
The amount and strength of the acid sites of phosphate catalysts
have been determined by the temperature-programmed desorp-
tion of ammonia (Fig. 4). The TPD-NH3 profiles show one peak in
the temperature range of 400–800 K for all the samples studied. For
NbP and AlP, this peak is rather narrow and is shifted to lower tem-
perature, which points to the acid sites with medium strength. On
the contrary, TiP and ZrP desorb ammonia in the wide region of tem-
peratures from 400 to 800 K, which corresponds to the medium and
the strong acid sites. The total amount of the acid sites determined
by TPD-NH3 correlates with the surface area of samples (Table 1).
The nature of the acid sites was studied by IR spectroscopy of
adsorbed pyridine, and the IR spectra are shown in Fig. 5. The main
bands observed over all the samples are assigned according to the
literature data [26,27] as follows:
The origin of Brønsted acidity of the samples is due to the termi-
nal P OH groups along with some contribution of M OH groups.
The Lewis acidity can be assigned to the coordinatively unsatu-
rated Mn+ sites [24,28]. The amount of Lewis and Brønsted sites
in metal phosphate catalysts usually depends on the amount of
chemisorbed water. The reaction with water involves its coor-
dination to the unsaturated metal site (Lewis site) and/or its
dissociative adsorption leading to the formation of OH groups
(Scheme 1).
Dehydroxylation of the samples leads to the transformation
of Brønsted sites into Lewis sites and visa versa the addition
of water may restore the content of Brønsted sites. The inten-
sity of the band assigned to pyridine adsorbed on Brønsted acid
sites increases drastically in the following order of catalysts:
BP ꢀAlP < TiP ≈ ZrP < NbP (Fig. 5). BP catalyst shows very weak
bands (not shown in Fig. 5) due to very low surface area. The con-
tent of Brønsted sites observed over phosphate catalysts studied
correlates with the ionic character of the bond between metal and
phosphate species: BP < AlP < TiP < ZrP < NbP (Fig. 3), suggesting that
the equilibrium shown on the Scheme 1 is shifted to the right in the
case of niobium phosphate and to the left in the case of AlP and BP
catalysts.
- the bands at 1592 and 1490 cm−1 are due to H-bonded pyridine;
- the band at 1544 cm−1 is attributed to pyridine protonated on
Brønsted sites;
- the band at 1449 cm−1 is assigned to pyridine adsorbed on Lewis
acid sites.
In summary, while the total amount of acid sites deter-
mined by TPD-NH3 is found to increase in the following order:
BP ꢀ NbP < TiP < AlP < ZrP, the content of Brønsted sites estimated
from IR spectroscopy of adsorbed pyridine gives quite a different
order of catalysts: BP ꢀ AlP < TiP ≈ ZrP < NbP, which is in line with
the increase of the ionic character of the bond between metal and
phosphate species.
Fig. 4. TPD–NH3 profiles.
Scheme 1.

V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
25
Fig. 6. Reaction network.
3.3. Condensation of formaldehyde with isobutene over
phosphate catalysts
with isobutylene leads to 3,4,4-trimethylcyclohexene formation
(Fig. 6).
Besides interacting with isobutene and isoprene, formaldehyde
can also decompose into hydrogen and carbon monoxide, which is
the main reaction by-product. According to [29] this reaction may
take place on acid–base sites of phosphate catalysts.
Isobutene can undergo oligomerization on the acidic sites lead-
ing to 2,2,4-trimethylpentenes. Dimerization of isoprene results in
various terpenes and teprenols.
3.3.1. Reaction products
On all the catalysts studied, the main reaction prod-
uct is isoprene, while the major by-products include: carbon
monoxide, 3-methylbutanal, methylcyclopentadiene, methyldehy-
dropyranes, 3,4,4-trimethylcyclohexene, 3-methyl-1,3-butandiol,
2,2,4-trimethylpentenes and terpenes.
The reaction network is shown in Fig. 6. The main reaction
pathway involves condensation of formaldehyde with isobutene
leading to 3-methylbuten-1-ols, which can be further converted
via 4 reaction routes. Dehydration of 3-methylbuten-1-ols leads to
the target reaction product – isoprene. 3-Methyl-3-buten-1-ol can
react with formaldehyde via the following reaction pathway giving
methylcyclopentadiene:
3.3.2. Effect of temperature
The experimental data obtained over NbP catalyst in the temper-
ature range of 523–623 K are presented in Table 2. The increase of
the temperature has a dramatic effect on both the activity and the
selectivity of the catalysts. Thus, the conversion of formaldehyde
increases from 66% to 100% in the above mentioned temperature
range, while the selectivity to isoprene shows maximum at 573 K.
At 523 K the reaction is accompanied by the formation of large
amounts of 3-methyl-1,3-butandiol and methyldehydropyranes. At
623 K carbon monoxide becomes the main reaction product due to
complete decomposition of formaldehyde. The highest selectivity
and the yield (57%) of isoprene were obtained at 573 K; therefore,
this temperature was selected for the further experiments.
3-Methyl-2-buten-1-ol can undergo double bond isomer-
ization into 3-methyl-1-buten-1-ol, the latter converts into
3-metylbutanal due to keto–enol tautomerism:
Besides that, 3-methylbuten-1-ols can be hydrated into 3-
methyl-1,3-butandiol, which is also observed in small amounts.
The secondary reaction routes include interaction of isoprene
with formaldehyde or isobutene. Thus, the electrophilic addition
of formaldehyde to isoprene on the acid sites of the catalysts
yields 4-methylhydropiranes. The similar reaction of isoprene
3.3.3. Effect of the catalyst type
The results of testing of different phosphate catalysts in
formaldehyde condensation with isobutene at 573 K are presented
in Table 3. The conversion of both formaldehyde and isobutene, as
well as the yield of isoprene increases in the following order of

26
V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
Table 2
growth of CO content in the products (Table 4). The latter observa-
tion could hardly be accounted for by the increase of Brønsted acid
sites concentration. Therefore, there should be other active sites
responsible for carbon monoxide formation, which are competing
with Brønsted acid sites for the formaldehyde conversion.
It is proposed that formaldehyde decomposition over phosphate
catalysts may occur on acid–base M␦+ O␦− ion pairs as it was sug-
gested previously for oxide catalysts [29]. The mechanism proposed
involves heterolytic interaction of formaldehyde with M␦+ O␦− ion
pair via a positively and a negatively charged hydrogen atoms, evo-
lution of CO and the recombination of dihydrogen molecule over
M␦+ O␦− ion pair (Scheme 2).
The reactivity of such sites depends on the basicity of oxygen and
on the ionic character of the bond between metal and phosphate
species and therefore should increase in the following order of
catalysts: BP < AlP < TiP < ZrP < NbP. Consequently, the same active
species are responsible for formaldehyde decomposition as shown
in (Scheme 2) and for the formation of Brønsted sites (Scheme 1),
which catalyze formaldehyde condensation. These considerations
explain the increase of the selectivity to carbon monoxide on the
expense of isoprene formation with the increase of ionic character
of M O bond and the basicity of oxygen in phosphate catalysts.
To verify this hypothesis, the NbP catalyst was subjected to the
ion exchange with sodium acetate and sodium carbonate. The ion
exchange with sodium acetate leads to the partial replacement of
acidic protons with sodium cations (Table 1) and to the decrease
of the amount of Brønsted acid sites in the sample (Fig. 7 and
Scheme 3).
Effect of temperature on the reaction of isobutene with formaldehyde over NbP
catalyst (WHSV = 3 g/(g h), iC4H8/CH2O = 7, TOS = 200 min).
Temperature
523 K
573 K
623 K
Conversion, %
Formaldehyde
Isobutylene
Selectivity to the products formed from formaldehyde, mol.%
Isoprene
Methylcyclopentadiene
3-Methylbutanal
Methyldehydropyranes
3,4,4-Trimethylcyclohexene
3-Methyl-1,3-butandiol
CO
Other products
Selectivity to the products formed from isobutene, mol.%
Isoprene
Methylcyclopentadiene
3-Methylbutanal
Methyldehydropyranes
3,4,4-Trimethylcyclohexene
3-Methyl-1,3-butandiol
2,2,4-Trimethylpentenes
Other productsa
66.2
7.9
91.1
10.1
100.0
1.0
43.9
3.9
1.9
8.0
2.3
11.0
15.9
13.1
62.6
2.2
1.0
2.5
1.0
0.0
29.7
1.0
2.2
0.8
1.0
2.2
1.0
0.0
90.2
2.6
67.9
2.1
2.0
4.2
1.5
12.0
0.2
10.1
86.6
2.5
1.2
1.2
0.6
0.0
2.5
5.4
71.1
1.0
1.2
1.1
0.6
0.0
12.2
12.8
a
Mainly terpenes and terpenols.
catalysts: BP < AlP < TiP < ZrP < NbP. The results observed do not
show any correlation with the total amount of acid sites determined
by TPD-NH3 neither normalized to the gram of the catalyst, nor to
the surface area (Table 1, Fig. 4). On the contrary, they give satisfac-
tory correlation with the amount of Brønsted acid sites determined
by FTIR spectroscopy of adsorbed pyridine (Fig. 5). This observation
suggests that formaldehyde condensation with isobutene is most
probably catalyzed by the Brønsted acid sites of phosphates, which
is line with the earlier observations [9–12].
To compare the catalysts selectivity, the same conversion levels
were achieved over different catalysts by the variation of WHSV
(Table 4). The results show that the selectivity to isoprene and
the secondary by-products formed from isoprene decreases in the
following order of catalysts: BP > AlP > TiP > ZrP > NbP, which is the
opposite to the increase of Brønsted acid sites content. The decrease
of the selectivity to isoprene is accompanied by the increase of the
selectivity to isobutene oligomers, isoprene oligomers and by the
The ion exchange with sodium carbonate results in the com-
plete substitution of protons (Table 1) followed by the complete
disappearance of Brønsted sites and the significant decrease of the
amount Lewis sites (Fig. 7 and Scheme 4).
The results on the catalytic activity of ion-exchanged samples
are presented in Table 4. Partial ion exchange does not lead to the
decrease of formaldehyde conversion but results in the increase of
carbon monoxide selectivity on the expense of isoprene formation,
which is due to the increase of the basicity of the oxygen atom
in M
O P bridges. The deep ion exchange shifts completely the
reaction pathway toward formaldehyde decomposition.
These results confirm our hypothesis on the active sites of
formaldehyde decomposition and condensation reactions and on
the changes of activity and selectivity in a series of phosphate
catalysts studied. It can be concluded that the decrease of the elec-
tronegativity of metal and the increase of the ionic character of
Table 3
Effect of the catalyst type on formaldehyde reaction with isobutene (573 K,
WHSV = 3 g/(g h), iC4H8/CH2O = 7, TOS = 200 min).
BP
AlP
TiP
ZrP
NbP
Conversion, %
Formaldehyde
Isobutylene
6.0
1.2
59.1
8.8
70.2
8.9
85.6
9.2
91.1
10.1
Selectivity to the products formed from formaldehyde, mol.%
Isoprene
83.1
1.0
1.8
5.0
2.6
4.3
2.2
81.8
0.8
0.6
5.0
2.9
6.9
2.0
74.1
1.6
1.5
4.5
2.8
65.9
1.8
0.4
3.7
0.6
62.6
2.2
1.0
2.5
1.0
Methylcyclopentadiene
3-Methylbutanal
Methyldehydropyranes
3,4,4-Trimethylcyclohexene
CO
Scheme 2.
Scheme 3.
Scheme 4.
13.9
1.6
27.3
0.3
29.7
1.0
Other products
Selectivity to the products formed from isobutene, mol.%
Isoprene
Methylcyclopentadiene
3-Methylbutanal
Methyldehydropyranes
3,4,4-Trimethylcyclohexene
2,2,4-Trimethylpentenes
Other productsa
91.7
0.6
2.0
2.6
1.4
1.2
0.5
6
91.9
0.4
0.7
2.7
1.6
1.3
1.4
37
90.3
0.8
1.6
2.4
1.5
1.4
2.0
30
88.9
2.0
0.6
1.6
0.3
2.2
4.4
44
86.6
2.5
1.2
1.2
0.6
2.5
5.4
37
Coke content, wt.%
Coke content, mg/m2
12.8
4.2
5.4
3.7
8.4
a
Mainly terpenes and terpenols.

V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
27
Table 4
Comparison of the selectivity of phosphate catalysts at the similar conversion level (573 K, iC4H8/CH2O = 7, TOS = 200 min).
BP
0.3
AlP
2.1
TiP
2.5
ZrP
3.0
NbP
3.0
0.2Na/NbP
2.5
0.5Na/NbP
WHSV, g/(g h)
Conversion, %
Formaldehyde
Isobutylene
2.3
82.8
8.4
86.1
8.8
84.9
9.0
85.6
9.2
91.1
10.1
86.2
9.8
86.7
0.2
Selectivity to the products formed from formaldehyde, mol.%
Isoprene
71.0
0.9
2.0
5.2
3.0
69.7
0.7
0.5
4.9
2.4
67.1
1.6
1.4
4.0
1.5
65.9
1.8
0.4
3.7
0.6
62.6
2.2
1.0
2.5
1.0
55.1
2.0
1.0
3.6
1.3
0.7
0.2
0.1
1.1
0.3
Methylcyclopentadiene
3-Methylbutanal
Methyldehydropyranes
3,4,4-Trimethylcyclohexene
CO
16.2
1.7
19.6
2.2
22.6
1.8
27.3
0.3
29.7
1.0
34.0
3.0
91.5
6.1
Other products
Selectivity to the products formed from isobutene, mol.%
Isoprene
Methylcyclopentadiene
3-Methylbutanal
Methyldehydropyranes
3,4,4-Trimethylcyclohexene
2,2,4-Trimethylpentenes
Other productsa
90.9
1.1
2.3
2.5
1.5
1.5
0.2
91.9
1.0
0.7
2.3
1.2
1.3
1.6
90.8
1.9
1.8
1.9
0.8
1.3
1.5
88.9
2.0
0.6
1.6
0.3
2.2
4.4
86.6
2.5
1.2
1.2
0.6
2.5
5.4
87.5
2.2
1.3
1.8
0.9
2.0
4.3
90.9
0.3
0.2
0.6
0.1
2.7
5.2
a
Mainly terpenes and terpenols.
the M O bond in phosphate catalysts lead to an increase of cat-
alytic activity both in formaldehyde decomposition, which occurs
on M␦+ O␦− ion pairs, and in formaldehyde condensation with
isobutene, which takes place over Brønsted acid sites. The selectiv-
ity toward isoprene decreases on the expense of carbon monoxide
formation due to the competitive action of the two types of active
sites in the target and side reactions.
reaction time to 30 h does not lead to any significant changes in the
conversion and the selectivity over these catalysts (Fig. 8).
According to Hutchings et al. [30], the deactivation of boron and
aluminum phosphates could be due to two reasons: (i) the loss of
surface phosphorus and (ii) coke deposition. Coke deposits can be
removed from the catalyst surface already by the reactivation of the
catalysts at 500 ◦C in an air atmosphere, while surface phosphorus
content can be restored only after high temperature calcinations
at 800 ◦C [30]. Calcination of our catalysts at 500 ◦C in an air flow
resulted in complete regeneration of the catalytic activity. Besides
that, verification of the amount of phosphorus on the AlP catalyst
before and after catalytic run has demonstrated that no phosphorus
loss occurs in our case. Therefore the deactivation was attributed
to coke deposition.
To study the effect of coke on the catalysts deactivation, the
carbonaceous deposits accumulated on the catalysts during 5 h on
stream were investigated in TG experiments. The DTG curves of
coke oxidation are presented at Fig. 9 and the amount of coke accu-
mulated over different catalysts is given in Table 3. All the catalysts
show DTG peaks in the temperature range of 600–800 K, which sug-
gests that the nature of coke deposits is similar over all the catalysts
studied. BP has the lowest coke content (<5 wt.%), which correlates
with the low surface area and the low acid sites content over this
catalyst (Table 1). On the contrary, this sample is characterized with
the highest amount of coke per surface square-meter (Table 3).
All the other catalysts show coke contents within 30–45 wt.%, the
3.3.4. Deactivation of the phosphate catalysts
The curves of isoprene yield versus time on stream observed
over different phosphate catalysts are presented in Fig. 8. All the
catalysts except BP show the initial increase of the yield, which is
due to the increase of the overall conversion at the initial steps of
the reaction. Such behavior could be explained by the modification
of the catalyst surface by the active carbonaceous species, respon-
sible for isoprene formation. Further behavior of the catalysts with
time on stream depends on the catalyst type. Boron, aluminum and
titanium phosphates show steady decline of the yield with time
on stream due to the steady decrease of the conversion. The rate
of the deactivation decreases in the following order of catalysts:
BP > TiP > AlP. On the contrary, ZrP and NbP catalysts show stable
catalytic activity after some initial drop down. The increase of the
Fig. 7. FTIR spectra of pyridine adsorbed over niobium phosphate catalysts with
different degrees of ion exchange with sodium.
Fig. 8. The yields of isoprene versus time on stream over phosphate catalysts.

28
V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
Scheme 5.
Fig. 9. DTG curves of temperature programmed oxidation of the deactivated cata-
lysts.
highest value being observed over ZrP. The analysis of the results
presented in Figs. 8, 9 and Table 3 show that the amount of coke
deposits calculated per gram of catalyst or per square-meter of cat-
alyst surface does not correlate with the rate of the deactivation
of different catalysts. Despite the highest amount of coke accumu-
lated on the surface of ZrP and NbP, these catalysts show the highest
catalytic stability with time on stream.
Fig. 11. The yields of isoprene versus time on stream over Na-exchanged niobium
phosphate catalysts.
Partial ion exchange of NbP catalyst with sodium acetate leads
to the decrease of the total amount of Brønsted acid sites (Fig. 7).
However, it does not prevent the catalyst from the generation of
Brønsted sites in the presence of water (Fig. 10 and Scheme 5)
As a result, 0.2Na/NbP shows rather high activity and extremely
high stability of catalytic activity with time on stream (Fig. 11).
On the contrary, full ion exchange of NbP catalyst with sodium
carbonate depicted in (Scheme 4) blocks completely the existing
Brønsted sites (Fig. 7) and prevents the generation of the novel ones
as confirmed by FTIR spectra of adsorbed pyridine (Fig. 10). As a
result, 0.5Na/NbP catalyst does not show any activity in isoprene
synthesis (Fig. 11).
We suppose that the regenerable acid sites shown in (Scheme 1)
are responsible for the stable catalytic performance of NbP and ZrP
catalysts in formaldehyde condensation with isobutene. Due to the
reversibility of the process (1), these sites can be eliminated and
restored on the catalyst surface directly during the catalytic run,
which provides the required amount of fresh active sites for stable
catalyst performance. The presence of water on the catalyst surface
plays a key role in this process.
To understand the reasons of high resistance of ZrP and NbP
catalysts to deactivation, the effect of water has been studied. Since
the reaction takes place in the presence of significant amount of
water, which is introduced in the reaction zone with formalin or
formed as the reaction product, it may have significant effect on
the catalysts active sites, their reactivity and their resistance to the
deactivation. To study the effect of water on the catalysts active
sites, FTIR spectroscopy of adsorbed pyridine has been used.
The results of the subtraction of IR spectra obtained before and
after exposure the catalysts to water are shown in Fig. 10. The
intensity of the band at 1544 cm−1 attributed to Brønsted acid sites
increases at the expense of the band at 1449 cm−1, correspond-
ing to Lewis acid sites. This observation is due to the dissociative
adsorption of water depicted in Scheme 1. Thus, Brønsted acid sites
can be generated in the presence of water directly during catalytic
run. This effect increases dramatically in the following range of
catalysts: AlP < TiP < ZrP < NbP, which can account for the higher
stability of catalytic activity over ZrP and NbP.
4. Conclusions
Phosphates of boron, aluminum, titanium, zirconium and nio-
bium are shown to be efficient catalysts of formaldehyde reaction
with isobutene. The main reaction pathway involves condensation
of formaldehyde with isobutene leading to isoprene as a major
product. Small amounts of 3-methylbutanal, methylcyclopen-
tadiene, methyldehydropyranes, 3,4,4-trimethylcyclohexene and
3-methyl-1,3-butandiol are also formed via this pathway. The com-
petitive reaction route includes formaldehyde decomposition into
carbon monoxide and dihydrogen.
It is proposed that isoprene is formed on the Brønsted acid sites
of the catalyst, which are due to the POH and MOH groups, while
M␦+ O␦− ion pairs are responsible for carbon monoxide formation.
The results obtained by FTIR spectroscopy of adsorbed pyridine
point that M␦+ O␦− ion pairs can be transformed into Brønsted acid
sites in the presence of water. On the contrary, dehydroxylation of
Brønsted acid sites may restore M␦+ O␦− ion pairs. The existence
of such equilibrium can account for the competitive action of these
sites for the target and side reactions.
The conversion of isobutene and formaldehyde over phosphates
increases in the following order: BP ꢀ AlP < TiP < ZrP < NbP, which
Fig. 10. The result of subtraction of FTIR spectra of adsorbed pyridine obtained
before and after the exposure the catalysts to water.

V.L. Sushkevich et al. / Applied Catalysis A: General 441–442 (2012) 21–29
29
can be explained by the decrease of electronegativity of the metal
and therefore by the increase of the ionic character of M O bond
in corresponding phosphate.
[7] K. Halcour, P. Losaker, W. Schwerdtel, W. Swodenk, DE 2347841 Bauer AG
(1975).
[8] K. Halcour, P. Losaker, W. Schwerdtel, W. Swodenk, DE2044623 Bauer AG
(1970).
The stability of the catalytic activity with time on stream
is found to increase in the following range of catalysts:
BP ꢀ AlP < TiP ꢀ ZrP ≈ NbP. The high resistance of ZrP and NbP cat-
alysts to deactivation is explained by the in situ restoration of the
Brønsted acid sites directly during the time course of the reaction.
Therefore the presence of water in the reaction zone may play a key
role in the stability of the catalytic activity of phosphate catalysts.
The best catalyst performance is observed over NbP catalyst,
which shows stable yields of isoprene of 57% within more than
30 h of stream.
[9] E. Dumitriu, V. Hulea, I. Fechete, A. Auroux, J.-F. Lacaze, C. Guimon, Appl. Catal.
A: Gen. 181 (1999) 15–28.
[10] E. Dumitriu, D. Trong On, S. Kaliaguine, J. Catal. 170 (1997) 150–160.
[11] G. Li, Y. Gu, Y. Ding, H. Zhang, J. Wang, Q. Gao, L. Yan, J. Suo, J. Mol. Catal.: A 218
(2004) 147–152.
[12] A. Krzywicki, T. Wilanowicz, S. Malinowski, React. Kinet. Catal. Lett. 11 (1979)
399–403.
[13] Z. Dang, J. Gu, L. Yu, C. Zhang, React. Kinet. Catal. Lett. 43 (1991) 495–500.
[14] M. Ai, J. Catal. 106 (1987) 280–286.
[15] G. Hutchings, I. Hudson, D. Timms, J. Chem. Soc. Chem. Commun. (1994)
2717–2718.
[16] G. Hutchings, I. Hudson, D. Timm, Catal. Lett. 61 (1999) 219–224.
[17] S.M. Patel, U.V. Chudasama, P.A. Ganeshpure, Green Chem. 3 (2001) 143–145.
[18] V.F.D. Alvaro, R.A.W. Johnstone, J. Mol. Catal. A: Chem. 280 (2008) 131–141.
[19] S. Okazaki, A. Kurosaki, Catal. Today 9 (1990) 113–122.
[20] S. Okazaki, N. Wada, Catal. Today 16 (1993) 349–359.
[21] J. Haber, U. Szybalska, Faraday Discuss. Chem. Soc. 72 (1981) 263–282.
[22] K. Segawa, S. Nakata, S. Asaoka, Mater. Chem. Phys. 17 (1987) 181–200.
[23] A. Clearfield, J. Mol. Catal. 27 (1984) 251–262.
Acknowledgement
The authors thank Federal Principal Program of Russian Ministry
of Education and Science “Research and Educational Specialists of
Innovative Russia” for the financial support.
[24] T. Armaroli, G. Busca, C. Carlini, M. Giuttari, A.R. Galletti, G. Sbrana, J. Mol. Catal.
A: Chem. 151 (2000) 233–243.
[25] X. Li, W. Zhang, G. Liu, L. Jiang, X. Zhu, C. Pan, D. Jiang, A. Tang, React. Kinet.
Catal. Lett. 79 (2003) 365–371.
References
[26] M.C. Kung, H.H. Kung, Catal. Rev. Sci. Eng. 27 (1985) 425–428.
[27] B.A. Morrow, I.A. Cody, J. Phys. Chem. 80 (1976) 1995–1998.
[28] M.H.C. de la Cruz, M.A. Abdel-Rehim, A.S. Rocha, J.F.C. da Silva, A.C. Faro Jr., E.R.
Lachter, Catal. Commun. 8 (2007) 1650–1654.
[29] M. Watanabe, M. Osada, H. Inomata, K. Arai, A. Kruse, J. Appl. Catal. A: Gen. 245
(2003) 333–341.
[1] O. Kriewitz, J. Chem. Soc. 76 (1899) 298–300.
[2] H. Prins, J. Chem Weekbl. 14 (1917) 933–935.
[3] H. Prins, J. Chem. Weekbl. 16 (1917) 1072–1075.
[4] H. Prins, J. Chem. Soc. 114 (1918) 261–264.
[5] Y. Hamamoto, A. Mitsutani, US3221075 Kurashiki Rayon Company Ltd.
(1965).
[30] G. Hutchings, I. Hudson, D. Bethell, D. Timms, J. Catal. 188 (1999) 291–299.
[6] A. Mitsutani, US3284533 Kurashiki Rayon Company Ltd. (1966).