Conversion of n-pentane and of n-butane catalyzed by platinum-containing WO(x)/TiO2

Article abstract of DOI:10.1039/b000054j

Tungstated titania, with and without platinum, was used to catalyze the conversion of n-butane and of n-pentane at atmospheric pressure and temperatures in the range of 423-573 K, both in the presence and in the absence of H₂ in the feed stream to a flow reactor. The catalysts were active for isomerization and cracking, and the products included alkenes, in contrast to those observed in catalysis by tungstated zirconia. The catalytic activity of tungstated titania is much less than that of sulfated zirconia and similar to that of tungstated zirconia. The results characterizing tungstated titania indicate a bifunctional reaction network involving metal and acidic sites responsible for hydrogenation/dehydrogenation and carbenium ion reactions, respectively. Platinum in the catalyst and H₂ in the feed strongly suppress the otherwise rapid catalyst deactivation associated with coke deposition. The catalyst was found to be stable after a period of initial deactivation, and kinetics data are reported for partially deactivated catalysts. High platinum contents lead to high selectivities for cracking in the presence of H₂ and high selectivities for unsaturated products in the absence of H₂. Increasing tungsten loadings raise catalyst acidity and favor isomerization. The performance of tungstated titania containing platinum resembles that of the zirconia-based catalysts less than it resembles the performance of platinum supported on (chlorided) alumina, a well-known naphtha reforming catalyst.

Full text of DOI:10.1039/b000054j

View Article Online / Journal Homepage / Table of Contents for this issue
Conversion of n-pentane and of n-butane catalyzed by
platinum-containing WO /TiO
x
2
S. Eibl,a R. E. Jentoft,b B. C. Gatesab and H. Knozingera
a Institut fur Physikalische Chemie, L MU Munchen, Butenandtstrasse 5-13 (Haus E), 81377,
Munchen, Germany
b Department of Chemical Engineering and Materials Science, University of California, Davis,
CA 95616, USA
Received 5th January 2000, Accepted 6th April 2000
Published on the Web 11th May 2000
Tungstated titania, with and without platinum, was used to catalyze the conversion of n-butane and of
n-pentane at atmospheric pressure and temperatures in the range of 423È573 K, both in the presence and in
the absence of H in the feed stream to a Ñow reactor. The catalysts were active for isomerization and
2
cracking, and the products included alkenes, in contrast to those observed in catalysis by tungstated zirconia.
The catalytic activity of tungstated titania is much less than that of sulfated zirconia and similar to that of
tungstated zirconia. The results characterizing tungstated titania indicate a bifunctional reaction network
involving metal and acidic sites responsible for hydrogenation/dehydrogenation and carbenium ion reactions,
respectively. Platinum in the catalyst and H in the feed strongly suppress the otherwise rapid catalyst
2
deactivation associated with coke deposition. The catalyst was found to be stable after a period of initial
deactivation, and kinetics data are reported for partially deactivated catalysts. High platinum contents lead to
high selectivities for cracking in the presence of H and high selectivities for unsaturated products in the
2
absence of H . Increasing tungsten loadings raise catalyst acidity and favor isomerization. The performance of
2
tungstated titania containing platinum resembles that of the zirconia-based catalysts less than it resembles the
performance of platinum supported on (chlorided) alumina, a well-known naphtha reforming catalyst.
compare its catalytic performance for conversion of n-pentane
and of n-butane with that of promoted tungstated zirconia.
Introduction
Branched alkanes are clean-burning automotive fuel com-
ponents with high octane numbers. They are favored thermo-
dynamically over straight-chain alkanes at low temperatures,
and consequently processes for isomerization of straight-chain
alkanes beneÐt from highly active catalysts that allow oper-
ation at low temperatures. Most such catalysts used currently
are either (a) strongly acidic halides, which are corrosive and
difficult to dispose of safely, and which also catalyze cracking,
and (b) metal-containing zeolites, which are bifunctional
Experimental methods
Catalyst preparation
Tungstated titania catalysts were prepared by suspension of
hydrous titanium oxide [precipitated from titanium isopro-
poxide (Heraeus, 99%) by the addition of water] in aqueous
solutions containing amounts of ammonium metatungstate
(Fluka, [85% WO ) sufficient to yield calcined materials with
(operating in the presence of H ) and require high tem-
peratures because their activities are relatively low. Alterna-
3
2
loadings between 10 and 20 wt.% WO . Water was removed
3
from the suspensions by evaporation, and the samples were
tively, catalysts such as (promoted) sulfated zirconia are
beginning to Ðnd applications as alkane isomerization cata-
lysts,1 being highly active, noncorrosive, and easy to dispose
of properly. However, these nonhalide solid acids undergo
rapid deactivation,2 in part by coke formation and possibly by
loss of sulfur.3 Tungstated zirconia o†ers an alternative of less
active but more stable catalysts than sulfated zirconia. Pro-
motion of the sulfated and tungstated catalysts with platinum
increases activity and selectivity for isomerization and sup-
dried at 383 K for 12 h and calcined at 923 K for 2 h in
stagnant air. Platinum was subsequently added by the incipi-
ent wetness method with aqueous solutions of
Pt(NH ) (NO ) (Fluka, [99.9%), and the samples were cal-
3 4
3 2
cined at 723 K for 2 h and reduced in situ in Ñowing H [20
2
ml(NTP) min~1] at 523 K. The samples contained 0.3 to 3
wt.% Pt. The sample nomenclature is as follows: the sup-
ported catalysts are referred to as Pt W , where x is the wt.%
x
y
Pt and y the wt.% WO , with the remainder being TiO .
presses deactivation;4h6 H in the feed prolongs catalyst life.7
3
2
2
A tungstated zirconia catalyst containing 19 wt.% WO
Oxides that can be sulfated to give catalysts with high
3
and 0.3 wt.% Pt was prepared analogously by impregnation
of hydrous zirconium oxide (MEL Chemicals), as described
previously.10
activities for alkane isomerization include, besides zirconia,
titania and iron oxide.8 Consequently, one might expect that
tungstated oxides other than zirconia would be of interest as
alkane isomerization catalysts. For example, tungstated
titania catalysts have been found to exhibit n-pentane isomer-
ization activity,9 but the activity is low, and no further investi-
gations have been reported. Our goal was to prepare and
characterize platinum-promoted tungstated titania and to
Surface area measurements
The BET surface areas of the catalysts were measured with a
Sorptomatic 1800 instrument (Carlo Erba) after the samples
had been dried at 473 K in dynamic vacuum for 1 h.
DOI: 10.1039/b000054j
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573
This journal is ( The Owner Societies 2000
2565

View Article Online
Temperature-programmed reduction
by a lower CO : Pt ratio. Therefore, the data representing
adsorption of CO on catalysts with high platinum contents
are expected to result in too-high dispersion values. It is rec-
ognized that the pulse Ñow method is limited in its ability to
give a quantitative measure of the dispersion; however, the
technique is useful for the identiÐcation of correlations
between catalyst properties and platinum surface area. The
results (Table 1) indicate that high platinum or tungsten con-
tents correspond to low platinum dispersions. At high tung-
sten loadings, exceeding those sufficient for monolayer
Temperature-programmed reduction (TPR) was carried out
with a mixture of 5 vol.% H in N at a Ñow rate of 12 ml
2
2
min~1. The temperature was ramped from 298 to 1073 K at
10 K min~1; the catalyst sample mass was 0.250 g. Hydrogen
consumption was measured with a thermal conductivity (TC)
detector.
Platinum dispersion
coverage of the support (ca. 10% WO ),13 we infer that the
Platinum dispersions (Pt
pulse Ñow method whereby pulses of CO (75 ll) were injected
/Pt ) were measured with a
3
surface total
tungsten oxide phase was sufficient to cover a substantial part
of the platinum. Thus, the relatively low platinum dispersions
indicated by the data for the samples with high tungsten load-
ings are not likely to be an indication of relatively large (low-
area) platinum particles, but instead to be an indication of a
limitation of the method for determining true dispersions of
platinum in such samples.
into a H stream [25 ml(NTP) min~1] and passed through a
2
bed containing 0.250 g of catalyst particles. CO adsorption on
the platinum surface was measured by monitoring the CO
concentration leaving the bed with a TC detector. Subsequent
pulses were injected until the amount of CO taken up reached
zero. Dispersions were calculated by integration of the
amount of CO taken up.
Temperature-programmed reduction
Low-temperature infrared spectroscopy with adsorbed CO
The reduction proÐles representing the uncalcined platinum-
containing samples treated in Ñowing H (Fig. 1, curves b, d,
and e) show intense signals in the range of 400È600 K. These
Infrared spectra were recorded with a Bruker IFS-66 FTIR
2
spectrometer equipped with a liquid N -cooled MCT detector.
2
signals were not observed for the platinum-free samples (Fig.
1, curves a and c), and they are therefore attributed to
reduction of Pt2` to Pt0. Weak reduction signals at 640È740
Self-supporting wafers of catalyst (ca. 15È20 mg cm~2) were
pressed at ca. 20 000 kPa and pretreated in dry O [50
2
ml(NTP)/min~1] at 673 K for 1 h and reduced in dry H for 2
2
h at 523 K. The pretreated samples were cooled to liquid N
2
K with H consumptions corresponding to 0.2 electrons per
2
W atom have been reported for WO /ZrO .14,15 Similarly,
temperature under vacuum (\10~4 kPa) and exposed to CO
x
2
for our samples, this Ðrst reduction step of tungsten (with a
at increasing equilibrium pressures in the range of 0.01È4 kPa.
The cell is described elsewhere.11
formal stoichiometry of WO ] WO ) is followed by a
3
2.9
reduction to WO , with signals observed at temperatures
2
above 900 K (Fig. 1, curve c); a further reduction to W0,
which is not depicted in Fig. 1, occurs only at temperatures
above 1200 K.
Catalysis experiments
Before each catalysis experiment, the catalyst in a stainless-
steel Ñow microreactor12 was heated in dry air [20 ml(NTP)
min~1; heating rate 10 K min~1] to 673 K and held at 673 K
The addition of platinum to the catalysts led to shifts in the
tungsten reduction signals to lower temperatures, by about
for 1 h followed by reduction in H [20 ml(NTP) min~1] at
100 K for Pt
W
and by about 50 K for Pt
W
, relative
2
0.3 10
0.3 20
523 K for 1 h. The catalyst particles (0.20 to 0.50 g) were
to the values for the platinum-free samples with the same
tungsten contents.
mixed with inert a-Al O particles to give a mass of particles
2
3
of 1.5 g. Reactions of n-pentane or of n-butane were conducted
at 101 kPa in the temperature-controlled microreactor. The
Acid site characterization
alkane was mixed with N (Puritan Bennett) and sometimes
2
The results of acid site characterization of several samples by
low-temperature infrared spectroscopy of adsorbed CO are
presented in Fig. 2. The spectra show bands at 2143 cm~1,
characteristic of physisorbed CO, as well as at 2170 cm~1,
indicative of CO adsorbed on BrÔnsted acid sites. Samples
Pt W and Pt W show bands at 2194 and 2201 cm~1,
also with H (99.999%, generated by electrolysis of water in a
2
Balston hydrogen generator). The partial pressure of n-
pentane or of n-butane in the reactant stream was varied from
0.17 to 1.0 kPa. The space velocity (WHSV) was varied from
0.036 to 0.440 g of alkane/(g of catalyst ] h). The reactor feed,
for example, was 1 vol.% alkane in N [20 ml(NTP) min~1]
x
10
x
20
2
respectively, which are indicative of CO on Lewis acid sites.
Bands characterizing samples W , W , and W at 2194,
or 1 : 1 mixtures by volume of H and 1 vol.% alkane in N
2
2
[40 ml min~1]. Products were analyzed by on-line gas chro-
matography (Hewlett Packard 5890) using a 0.52 mm ] 30 m
alumina PLOT column.
10
15
20
Table 1 Platinum dispersions in catalyst samples with various tung-
sten and platinum contentsa
Results
Catalyst composition (wt.%)
Platinum
dispersion
(%)
Catalyst surface areas
Pt
WO
3
The samples had BET surface areas in the range of 65 to 80
m2 g~1, with slightly higher values for higher tungsten load-
ings.
0.3b
0.3b
0.3b
3b
3b
3b
3c
3c
3c
10
15
20
10
15
20
10
15
20
10
15
15
17
12
11
13
3
Platinum dispersions
Calculation of the dispersion from CO uptake data collected
using the pulse Ñow method requires the assumption of an
adsorption stoichiometry. The stoichiometry assumed for our
calculations was 1 : 1 CO : Pt; however, the adsorption of
bridge-bonded CO, which has been observed by infrared spec-
troscopy (vide infra) in those samples with high platinum load-
ings (Pt W , Pt W ) causes the adsorption to be characterized
3
a Samples were calcined at 923 K before the addition of platinum and
at 723 K after the addition of platinum. b Reduced at 473 K.
c Reduced at 1073 K.
1
y
3
y
2566
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573

View Article Online
The Lewis acid strengths are slightly less (CO vibration fre-
quency about 5 cm~1 lower) for samples with high tungsten
and platinum loadings than for the platinum-free samples.
The bands at 2090 and 1865 cm~1 are attributed to CO
adsorbed on platinum in linear and bridged conÐgurations,
respectively.16 Band positions and intensities of CO on
BrÔnsted acid sites are indistinguishable from each other in
the spectra representing the platinum-free and platinum-
containing catalysts. Shifts in the OH bands characterizing
platinum-free samples (at approximately 3660 cm~1) induced
by CO adsorption were found to be 215, 230, and 240 cm~1
for W , W , and W
(not shown), respectively. OH
10
15
20
stretching bands were not detected in the black platinum-
containing samples because of low signal-to-noise ratios at
frequencies [3500 cm~1; however, the addition of platinum
to the WO /TiO samples is not expected to have had a
3
2
strong inÑuence on the BrÔnsted acidity.17 The data indicate
strong BrÔnsted acid sites18 that increase in strength with
increasing tungsten content.
Performance of catalysts without platinum
Catalysts without platinum and with a tungsten loading
(about 25% WO ) that was found to give nearly the maximum
3
activity catalyzed the conversion of n-pentane to give yields,
deÐned as 100 ] (concentration of product in the effluent
stream)/(concentration of alkane in the feed), of about 1.2%,
with selectivities for isopentane, deÐned as (concentration of
isopentane in the effluent stream)/(total product concentration
in the effluent stream), of about 85% for a feed of 1% n-
pentane in 1 : 1 H /N at 573 K. The yields were initially rela-
2
2
tively high at a WHSV of 0.072 g of alkane/(g of catalyst ] h)
and declined rapidly, becoming nearly constant at about 0.6%
after several hours on stream. As the temperature was raised,
the initial conversion of n-pentane increased from 1.2% at 573
K to 6.0% at 698 K. This increase in conversion was accom-
panied by a decrease in selectivity for the isomerization
product (isopentane), from 85% at 573 K to 24.4% at 698 K,
and an increase in the selectivity for cracking products
(particularly methane), from 15% at 573 K to 75.6% at 698 K.
Fig. 1 Temperature-programmed reduction proÐles: (a) W ; (b)
10
Pt
W
; (c) W ; (d): Pt
W
; (e) Pt W (spectra normalized and
0.3 10
20
0.3 20
3 20
adjusted vertically).
2200, and 2206 cm~1 (not shown), respectively, show an
increasing energy with increasing tungsten content. These
bands at ca. 2200 cm~1 are attributed to Lewis acid sites with
strongly overlapping bands for CO on Ti4` and W6` sites.
Without H in the feed there was no measurable conversion of
2
a feed of 1% n-pentane in N [20 ml min~1] at temperatures
2
up to 673 K with 0.5 g of WO /TiO in the reactor. Further
results are presented in a thesis.17
3
2
Performance of platinum-containing catalysts
n-Pentane and n-butane feeds without H . At 523 K the
2
major product of reaction of n-pentane was initially the isom-
erization product isopentane, with smaller amounts of the
cracking products isobutane, n-butane, and C ÈC alkanes. At
1
3
a WHSV of 0.18 g of alkane/(g of catalyst ] h), these saturat-
ed products are characterized by initially high yields (e.g., ca.
50% to isopentane) at 5 min on stream followed by rapid
catalyst deactivation within the Ðrst 2 h on stream. In addi-
tion to saturated products, there were signiÐcant amounts of
several C alkenes, which were characterized by initially low
5
yields followed by increasing yields to maxima of about 3.1 to
4.4% at 1 to 4 h on stream, followed by gradually decreasing
yields of these products. Traces of C and higher hydrocar-
6
bons were also present in the product stream (Table 2).
When the reactant was n-butane, the major product at 523
K was initially isobutane, with smaller amounts of C ÈC
1
3
alkanes (Fig. 3). At a WHSV of 0.14 g of alkane/(g of
catalyst ] h) these saturated products were characterized by
initially high yields (e.g., ca. 12% to isobutane) at 5 min on
stream followed by rapid deactivation within the Ðrst 2 h on
stream. In addition to these saturated products there were sig-
niÐcant amounts of butenes, including trans-but-2-ene, cis-
but-2-ene, but-1-ene, and isobutylene, which were
characterized by initially low yields followed by increasing
Fig. 2 Low-temperature (88 K) infrared spectra of catalysts in the
presence of 4 kPa of CO: (a) W ; (b) Pt
W
; (c) Pt W ; (d)
10
0.3 10
3
10
Pt W
.
3
20
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573
2567

View Article Online
Table 2 Catalyst performance data characterizing conversion of n-pentane (at 523 K) and of n-butane (at 573 K)a
Catalyst
composition
(wt.%)
Maximum yield (%)
Approx.
carbon
loss to
catalyst
(%)
At TOS \ 5 min
At TOS B 1È4 h
TOS
/h
1@2
Reactant
Pt
WO
Isopentane
C ÈC
Pentenes
C
3
1
4
6`
n-pentane
0.3
0.3
1
10
20
20
20
0.22
0.28
0.35
0.56
22
17
51
60
2.7
3.1
6.3
4.9
3.1(after D1.0 h)
2.2(after D1.8 h)
4.2(after D2.5 h)
4.4(after D3.7 h)
3.1
2.2
4.2
4.5
12
25
12
11
3
Isobutane
C ÈC
Butenes
C
1
3
5`
n-butane
0.3
3
20
20
0.19
0.20
12
21
10.0
14.2
8.1(after D2.6 h)
12.0(after D2.8 h)
4.3
2.8
a The feed partial pressures were the following: alkane, 1 kPa; N , 100 kPa; the feed Ñow rate was 20 ml(NTP) min~1; and the catalyst mass
2
was 0.5 g.
values to maximum total yields of unsaturated products of
about 10% at 2 to 5 h on stream followed by gradually
decreasing yields of these products (Table 2). Thus, the
occurrence of isomerization and of cracking is common to n-
pentane and n-butane.
The yield of isobutylene from n-butane passed through a
maximum with time on stream (Fig. 3). Thus, an induction
period is evident, and it is similar to the induction periods
representing the other butenes. The decline in yield of iso-
butylene following the induction period parallels the decline in
yield of isobutane. Similar statements pertain to n-pentane
conversion, but we were not able to distinguish between at
a single reaction. The formation of alkenes shows that the
catalyst is a dehydrogenation catalyst. The decrease in yield of
branched products with time on stream (TOS) suggests a
decreasing isomerization activity of the catalyst with TOS.
A carbon balance on the reactant and product streams
(Table 2) shows an initially high and rapidly decreasing loss of
carbon from the reactant stream to the catalyst. Thus, the
catalyst deactivation is attributed (at least in part) to coke for-
mation; coke formation is also an indication of the dehydro-
genation activity of the catalyst. The proÐle of carbon loss
with TOS is similar in shape to the proÐle of yield of iso-
pentane from n-pentane (or of isobutane from n-butane) with
TOS, in contrast to the proÐles of yields of alkenes from
alkanes with TOS.
least four of the unsaturated C compounds. Thus, for each
5
reactant, n-pentane and n-butane, we infer that the yield of the
isoalkene proceeds via a combination of reactions rather than
n-Pentane feeds without H : e†ect of temperature. In the
2
n-pentane isomerization catalyzed by Pt/WO /TiO , the tem-
3
2
perature inÑuenced the conversion, selectivity, and rate of
catalyst deactivation. To quantify the rate of deactivation, we
use the parameter time on stream corresponding to a conver-
sion equal to half of the initial conversion (TOS ). The
1@2
reported values of TOS were found by extrapolation of the
1@2
conversion data to zero TOS and interpolation to half of the
initial value by using an exponential Ðt to the conversion vs.
TOS data.19
The trends in the yields of di†erent products formed from
n-pentane in the temperature range of 423È598 K at 5 min
TOS are as follows: Higher temperatures gave higher yields of
C ÈC saturated products and isobutane. With increasing
1
4
temperature, the yield of isopentane passed through
a
maximum at 523 K (Table 3). The trends in the selectivities for
Fig. 3 Conversion of n-butane in a Ñow reactor containing Pt
W
0.3 20
catalyst. Feed, n-butane (1 kPa) in N (100 kPa); temperature, 573 K.
C ÈC saturated products and isobutane parallel those rep-
2
1
4
Table 3 Yields, selectivities, times on stream for half conversions (TOS ) and carbon losses for TOS \ 5 min in n-pentane conversion catalyzed
1@2
by Pt W
a
1
20
Yield and selectivity (%)
At TOS \ 5 min
At TOS \ 2 h
unsat.c
Approx.
carbon
loss to
catalyst
%
C ÈC sat.b
Isobutane
Isopentane
Yield
C
C
1
4
5
6`
Reaction
temp./K
TOS /h
Yield
Sel.
Yield
Sel.
1.5
7.41
19.3
Sel.
Yield
Sel.
Yield
Sel.
1@2
423
523
598
0.98
0.41
0.35
0.05
1.71
16.6
0.5
2.96
24.5
0.2
5.5
12.5
11.2
51.1
32.0
98.1
84.2
49.2
0.13
4.39
18.24
21.0
96.7
89.5
0.43
2.30
3.70
1.3
0.9
1.8
0
12
25
a The feed partial pressures were the following: n-pentane, 1 kPa; N , 100 kPa; the feed Ñow rate was 20 ml(NTP) min~1; the catalyst mass was
2
0.5 g. b Saturated. c Unsaturated.
2568
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573

View Article Online
resenting the yields of these products, with the selectivities
increasing with increasing temperature. In contrast, selectivity
for isopentane decreased with increasing temperature.
Increasing the temperature led to increasing production of
unsaturated and oligomerized compounds, which likely
include coke precursors. The increasing loss of carbon from
the gas stream to the catalyst with increasing temperature is
consistent with the observed deactivation and the formation
of coke on the catalyst surface.
At 598 K, initial yields of isomerization products are
expected to have been higher than those at lower tem-
peratures (Table 3), but the opposite is apparently indicated
by the data at TOS \ 5 min. The apparent anomaly is
explained by rapid deactivation during the Ðrst 5 min on
stream and conversion of the n-pentane to secondary products
after 5 min TOS. At this relatively high temperature, we were
unable to extrapolate the data to measure the initial yield
accurately.
Fig. 4 Conversion of n-pentane in
a
Ñow reactor containing
Pt
W
catalyst. Feed, n-pentane (1 kPa) with H (50 kPa) in N (50
0.3 20
2
2
kPa); temperature, 523 K.
product was propane, with lesser amounts of methane, ethane,
n-Pentane and n-butane feeds without H : e†ects of plati-
2
num and WO contents. In the conversion of n-pentane at 523
3
and isobutane. The H markedly reduced the rate of catalyst
2
deactivation. In contrast to the reaction carried out with
K, increasing the catalyst platinum loading from 0.3 to 3.0%
increased the yield of isopentane and retarded the deactiva-
platinum-containing catalysts without H , when H was
2
2
present in the feed there was no measurable loss of carbon to
the catalyst.
tion for conversion to isopentane (TOS increased from 0.28
1@2
to 0.56 h). In contrast, increasing the tungsten content of a
catalyst containing 0.3% Pt from 10 to 20% WO led to only
a slight decrease in the yield of isopentane and a slight
3
n-Pentane and n-butane feeds with H : e†ects of platinum
2
and WO contents. The selectivities for the reactions of n-
3
increase in TOS (Table 2). In the isomerization of n-butane
1@2
pentane and of n-butane, at 523 and 573 K, respectively, are
at 573 K, a higher platinum content (3% vs. 0.3%) in a 20%
presented in Table 4. In the conversion of n-pentane at a
WHSV of 0.18 g of alkane/(g of catalyst ] h), catalysts con-
taining low loadings of platinum (0.3% Pt) with 10 to 20%
WO catalyst led to increases in both the yield of isobutane
3
and the yield of C ÈC products (Table 2). A carbon balance
1
3
on the reactant and product streams showed essentially no
dependence on the platinum or tungsten content of the cata-
lyst.
WO gave conversions less than 7%, with nearly equimolar
3
amounts of ethane and propane, on the one hand, and of
methane and n-butane, on the other. Increasing the WO con-
3
tents of these samples led to increased yields of the isomer-
n-Pentane and n-butane feeds with H . When n-pentane in a
ization product, isopentane, and to slight decreases in the
2
1 : 1 mixture of H and N was fed to a reactor containing a
Pt
yields of cracking products (C ÈC alkanes). Catalysts with
2
2
1
4
W
catalyst at 523 K, the major product was initially
higher platinum loadings (3 wt.%) containing 20% WO gave
0.3 20
3
propane, with lesser amounts of isopentane, n-butane, ethane,
and methane (Fig. 4). The yields of the cracking products
decreased substantially during the Ðrst 4 h on stream, whereas
the yield of isopentane underwent a brief induction period
during the Ðrst hour on stream. Yields of both isomerization
and cracking products became nearly stable after 6 h on
stream and to the end of the run (up to 12 h on stream).
conversions up to 38% and higher selectivities for methane
than for n-butane and higher selectivities for ethane than for
propane (Table 4). Increasing the platinum loading in these
samples reduced the selectivity for isopentane. The Pt
W
0.3 20
sample is characterized by the highest selectivity for iso-
pentane and the Pt W sample by the highest yield of iso-
1
20
pentane.
When n-butane in a 1 : 1 mixture of H and N was fed to a
The trends for n-butane conversions at 573 K are similar to
those for n-pentane conversion at 523 K. Higher tungsten
2 2
catalyst at 573 K, the major
reactor containing Pt
W
0.3 20
Table 4 Yields in conversion of n-pentane (at 523 K) and of n-butane (at 573 K) and selectivities [normalized (norm.) and molar, see text] after
12 h on streama
Catalyst
composition
(wt%)
Selectivity (%)
Methane
Yield (%)
Ethane
Propane
Isobutane
n-Butane
Isopentane
Reactant
Pt
WO
C ÈC
Isopentane Norm. Molar Norm. Molar Norm. Molar Norm. Molar Norm. Molar Norm. Molar
3
1
4
n-pentane 0.3
10
15
20
20
20
4.9
4.5
4.3
12.0
35.2
0.57
0.87
2.34
4.49
2.74
8.4
7.4
5.6
7.8
11.3
21.4
19.6
17.1
20.9
26.1
22.4
20.9
16.0
21.6
29.3
28.7
27.8
23.4
29.0
33.8
33.6
31.8
24.6
27.3
35.3
28.7
28.2
23.9
24.5
27.1
0.2
0.2
0.3
0.4
0.4
0.1
0.1
0.3
0.3
0.2
24.9
23.5
18.2
15.6
16.5
15.9
15.6
13.8
10.6
9.5
10.5
16.3
35.2
27.3
7.2
5.3
8.7
21.5
14.8
3.3
0.3
0.3
1
3
Methane
Ethane
Propane
Isobutane
C ÈC
Isobutane
Norm. Molar Norm. Molar Norm. Molar Norm. Molar
1
3
n-butane
0.3
0.3
3
10
20
20
11.9
6.8
97.5
0.35
0.52
0.65
24.5
21.3
27.4
44.2
39.6
45.1
38.0
25.4
57.4
34.1
24.5
46.7
34.6
46.1
14.5
20.4
33.5
7.9
2.9
7.3
0.7
1.3
2.5
0.3
a The partial pressures: alkane, 1 kPa; H , 50 kPa; N , 50 kPa; feed Ñow rate, 40 ml(NTP) min~1; catalyst mass, 0.25 g.
2
2
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573
2569

View Article Online
loadings gave increased yield of (and selectivities for) the isom-
erization product, isobutane. Increasing platinum content
resulted in increased yields of cracking products (methane,
ethane, and propane) and decreased selectivity for isobutane.
n-Pentane and n-butane feeds with H : e†ect of space veloc-
2
ity There was no signiÐcant e†ect on selectivity of variation of
the space velocity from 0.036 to 0.440 g of alkane/(g of
catalyst ] h) for catalysts containing 0.3 wt.% Pt. However,
there was a signiÐcant e†ect when the catalysts contained
more than 0.3 wt.% Pt (Pt W and Pt W ). In the conver-
1
20
3
20
sion of n-pentane in the presence of Pt W at 473 K, the
3
20
selectivity for the isomerization product (isopentane) increased
with increasing space velocity, and the selectivity for the
cracking products (methane, ethane, propane, and butane)
decreased.
Fig. 6 E†ect of temperature on conversion after 12 h TOS for reac-
tion catalyzed by Pt
W
. Feed, n-pentane (1 kPa) with H (50 kPa)
0.3 20
2
in N (50 kPa).
2
Measurements after 15 h on stream are characterized by
stable, nearly constant conversions (at values lower than the
maximum) and linear correlations of conversions with inverse
space velocity w/F (catalyst mass/alkane molar Ñow rate) (Fig.
5). The slopes of the straight lines Ðtting the data were used to
calculate reaction rates (Table 5). The calculated rates under
steady-state conditions represent isomerization and cracking
reactions.
Pt
W
, Pt
W
, and Pt
W
(Fig. 6). At temperatures
0.3 10
0.3 15
0.3 20
higher than 623 K, deactivation was not negligible, and its
onset correlated with the appearance of traces of unsaturated
compounds in the product stream. Pt W and Pt W gave
1
20
3 20
predominantly cracking products, with increasing concentra-
tions of small saturated hydrocarbons at higher reaction tem-
peratures. Trends in changes in yield and selectivity resulting
from changes in the reaction temperature were found to be
similar for n-pentane and for n-butane, with generally higher
yields of cracking products observed with n-butane feeds.
Apparent activation energies (calculated from Arrhenius
plots of reaction rates, estimated from nearly di†erential con-
versions, e.g., Fig. 7) for n-pentane conversion are about the
same (roughly 200 kJ mol~1) for samples with di†erent tung-
sten loadings (Table 6). An e†ect of an increase in tungsten
loading was a slight reduction in the apparent activation
energy for cracking and a slight increase in the apparent acti-
n-Pentane and n-butane feeds with H : e†ect of tem-
2
perature. The catalyst samples containing 0.3% Pt reached
nearly steady-state conversion when run for 12 h on stream,
i.e., the rate of deactivation became almost negligible after this
period at temperatures up to 623 K. Thus, these partially
deactivated and nearly stable catalysts presented the
opportunity for measurement of the temperature dependence
of the rates of the catalytic reactions.
Isomerization selectivity and yield increased with increasing
temperature in the range of 473 to 623 K for samples
Fig. 5 Demonstration of di†erential conversion of n-pentane cata-
lyzed by Pt W at 473 K. Feed, n-pentane (1 kPa) with H (50 kPa)
Fig. 7 Arrhenius plot: formation of products from n-pentane cata-
lyzed by Pt
(50 kPa).
W
. Feed, n-pentane (1 kPa) with H (50 kPa) in N
3
in N (50 kPa).
2
20
2
0.3 20
2
2
Table 5 Reaction rates for conversion of n-pentane at 473 K and of n-butane at 573 K, determined from di†erential conversions
Catalyst
composition
(wt.%)
Rate of formation of product/mol (g of catalyst)~1 s~1
Reactant
Pt
WO
Methane
Ethane
Propane
Isobutane
n-Butane
Isopentane
3
n-pentane
0.3
0.3
3
10
20
20
1.21 ] 10~8
1.58 ] 10~8
7.07 ] 10~8
2.26 ] 10~8
3.92 ] 10~8
2.00 ] 10~7
3.73 ] 10~8
5.09 ] 10~8
3.00 ] 10~7
0
0
3.77 ] 10~9
4.98 ] 10~8
2.00 ] 10~7
1.45 ] 10~8
7.79 ] 10~8
6.74 ] 10~8
1.26 ] 10~9
n-butane
0.3
0.3
10
20
1.94 ] 10~6
7.92 ] 10~7
3.20 ] 10~6
1.03 ] 10~6
2.95 ] 10~6
1.83 ] 10~6
2.44 ] 10~7
2.95 ] 10~7
2570
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573

View Article Online
Table 6 Apparent activation energiesa for conversion of n-butane and of n-pentaneb
Catalyst
composition
(wt.%)
Apparent activation energy/kJ mol~1
Reactant
Pt
WO
Methane
Ethane
Propane
n-Butane
Isobutane
Isopentane
3
n-pentane
0.3
0.3
0.3
1
10
15
20
20
20
199
196
180
227
254
218
211
196
241
272
217
213
204
238
267
206
203
198
238
248
*c
*
*
310
298
179
189
204
238
300
3
n-butane
0.3
0.3
3
10
20
20
278
233
255
296
295
272
277
288
245
È
È
È
321
301
260
È
È
È
a Standard deviation from linear Ðt, approximately 5 kJ mol~1. b The feed partial pressures were the following: alkane, 1 kPa; H , 50 kPa; N ,
2
2
50 kPa; data were determined from rates measured at temperatures in the range of 423È573 K; feed Ñow rate, 40 ml(NTP) min~1; catalyst mass,
0.25 g. c * Conversion too low to allow determination.
vation energy for isomerization. Increasing the platinum
content from 0.3 to 3 wt.% increased the apparent activation
energies for both cracking (by ca. 60 kJ mol~1) and isomer-
ization (by ca. 100 kJ mol~1). n-Butane conversion is charac-
terized in general by higher apparent activation energies than
n-pentane conversion, but the same trends relating the appar-
ent activation energy and tungsten and platinum loading were
observed for n-butane as for n-pentane, the only exception
being that, for n-butane, high tungsten and platinum loadings
correspond to reduced apparent activation energies for isom-
erization.
selectivity and the decrease in isomerization selectivity with
increasing H partial pressure were found to be greater for
2
catalyst samples with high tungsten loadings (Table 7). Isom-
erization is favored at low H partial pressures on catalysts
2
with high tungsten loadings.
Reaction orders for a simple power law rate equation
r \ kpn
pm
(1)
alkane H2
were determined from the slopes of lines Ðtting rate vs. reac-
tant partial pressure on logarithmic coordinates (Fig. 8). Mea-
surements of the reaction orders in alkane (n) were made with
Whereas isomerization selectivities were found to decrease
with increasing platinum loading, changes in tungsten loading
(which is correlated with acidity) had only a little inÑuence on
catalyst performance. High cracking selectivities were gener-
ally related to the formation of small saturated hydrocarbons.
These trends correlate with the generally higher apparent acti-
vation energies for isomerization reactions rather than for
cracking reactions for catalysts with high platinum loadings.
Catalysts with only 0.3% platinum, on the other hand, were
characterized by lower apparent activation energies for isom-
erization rather than for cracking. These apparent activation
energies account for the increase of cracking selectivity
(especially for catalysts with high platinum contents) with
H
partial pressures high enough to largely prevent catalyst
2
deactivation.
Catalyst samples with high platinum loadings do not show
a signiÐcant dependence of isomerization selectivity on H
2
increasing temperature, independent of the presence of H in
2
the feed.
Kinetics of reactions of n-pentane and of n-butane. Doubling
the H partial pressure from 40 to 80 kPa (total feed Ñow
2
rate: 240 ml min~1) with a hydrocarbon partial pressure of
0.17 kPa led to a halving of the total conversion, higher crack-
ing selectivity, and a lower isomerization selectivity for both
n-butane and n-pentane conversions. The increase in cracking
Fig. 8 Determination of orders of reaction in H . Data shown for
2
various products of n-pentane conversion catalyzed by Pt
W
at
0.3 20
573 K; data were determined for a partially deactivated catalyst.
Table 7 E†ect of variation of H partial pressure in the feed (at 41.7 or 75.0 kPa) on conversion of n-pentane at 523 K and of n-butane at
2
573 Ka
Catalyst
composition
(wt.%)
Selectivity for
isomerization (%)
1010 ] isomerization rate
/mol g~1 s~1
Total conversion (%)
Reactant
Pt
WO
41.7 kPa
75.0 kPa
41.7 kPa
75.0 kPa
41.7 kPa
75.0 kPa
3
n-pentane
0.3
0.3
0.3
0.3
3
10
15
20
20
20
9.1
13.5
24.2
7.2
7.2
9.6
16.6
5.5
5.0
4.7
5.4
4.3
42.2
4.2
1.15
4.5
3.5
15.6
1.60
1.87
4.90
1.18
1.70
0.91
4.0
2.65
0.48
1.10
n-butane
0.9
1.1
a The partial pressures of the other feed components were the following: alkane, 1 kPa; N , balance to give a total pressure of 101 kPa; feed
2
Ñow rate, 120 ml(NTP) min~1; catalyst mass, 0.25 g.
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573
2571

View Article Online
partial pressure. Increasing n-pentane partial pressures in the
range of 1 to 5 kPa reduced total conversion and increased
isomerization selectivity (Table 8). These e†ects are strongest
for catalysts with high tungsten loadings. Reaction orders (n
and m) for cracking products of n-pentane are in the range of
[0.4 ^ 0.2 and do not vary signiÐcantly with tungsten and
The activity and selectivity of WTi for isomerization are
similar to those of tungstated zirconia (WZr) (Table 9), but
WTi catalysts are di†erent from WZr in producing much
higher yields of alkenes, and, likely as a consequence of the
presence of the alkenes, they are deactivated more rapidly
than WZr. For example, deactivation of WTi is too fast to
allow detection of isomerization activity at 523 K with the
platinum loadings. Reaction orders in H (m) for isomer-
2
ization are [0.9 ^ 0.2 and in alkane (n) are [0.1 ^ 0.2. The
reactant in the presence of N alone, whereas deactivation of
2
latter values are more negative for butane (\[1).
WZr at 523 K begins after an induction period at 1.5 h on
stream. However, the addition of H to the feed suppresses
2
alkene formation catalyzed by WTi and allows measurement
of some isomerization activity. Addition of platinum to WTi
improves its performance for alkane isomerization, much as it
improves the performance of WZr for this reaction.
Because WTi is not nearly as active an alkane isomerization
catalyst as sulfated zirconia and not nearly as selective an
alkane isomerization catalyst as WZr, it is not expected to be
practically useful for this reaction.
Discussion
Comparison of performance of WO /TiO and related alkane
isomerization catalysts
x
2
The data conÐrm that WO /TiO (WTi) is catalytically active
x
2
for alkane isomerization, but its activity is markedly lower
than that of sulfated zirconia, some form of which is active
enough to have found application as a commercial catalyst for
isomerization of C and C alkanes.1 For example, WTi cata-
lyzes n-pentane isomerization measurably at 573 K, whereas
sulfated zirconia catalyzes this reaction measurably at tem-
peratures as low as about 320 K.20
Bifunctional catalysis by platinum-containing WTi
5
6
Characteristics of the performance of the PtW catalyst, includ-
ing its rapid deactivation (presumably by coke formation) at
Table 8 E†ect of variation of alkane partial pressure (at 0.9 or 5.3 kPa) in conversion of n-pentane (at 523 K) and of n-butane (at 573 K)
Catalyst
composition
(wt.%)
Selectivity for
isomerization (%)
1010 ] isomerization rate
/mol g~1 s~1
Total conversion (%)
Reactant
Pt
WO
0.9 kPa
5.3 kPa
0.9 kPa
5.3 kPa
0.9 kPa
5.3 kPa
3
n-pentane
0.3
0.3
0.3
0.3
0.3
3
10
15
20
10
20
20
8.9
11.5
18.5
3.0
6.1
1.0
12.8
17.2
36.3
3.9
8.0
0.7
1.03
1.98
3.62
1.20
0.99
4.30
2.00
1.31
4.15
0.72
0.53
7.39
4.9
6.9
7.5
9.5
6.8
2.0
2.2
2.3
3.4
1.8
n-butane
35.1
22.4
Table 9 n-Pentane isomerization in the presence of various catalysts
WHSV/
g of
alkane/(g
of catalyst
] h)
Total n-
pentane
Total n-
pentane
conversion
Approx.
isomerization
selectivity
(%)
Reactant partial
pressures/
kPa
conversion
rate/
Temperature
/K
Catalysta
Notes
mol g~1 s~1 (%)
8.4 ] 10~9
0
Reference
this work
this work
W
Ti
n-pentane: 1
very slow
deactivation
20
T
\ 923 K 573
0.072
0.072
N
H
: 50
: 50
3
0
75
È
C
2
2
W
Ti
n-pentane: 1
no activity observed
at temperatures up to
673 K,
20
T
\ 923 K 523
N : 100
C
2
deactivation too fast
W
Zr
n-pentane: 1
: 100
n-pentane: 1
1.1 ] 10~9
1.5 ] 10~8
2 ] 10~8
19
T
\ 923 K 523
0.072
0.18
N
rapid deactivation
0.8
2.1
80
35
23
C
2
Pt
Pt
Pt
W
Ti
523
523
479
645
N
H
: 50
: 50
nearly constant
activity
this work
0.3 20
2
2
n-pentane: 2
induction period
followed by
deactivation
nearly constant
activity
W
Zr
0.036
0.347
0.19
N
: 50
7.1
24.8
17.9
95
92
94
this work
0.3 19
2
H : 50
2
n-pentane: 7.2
S
Zr
H
: 223
6.7 ] 10~7
1.3 ] 10~7
16
24
0.3 1.8
2
He: 869.8
n-pentane: 460
standard naphtha
reforming catalyst
Pt Al
H : 640
0.3
2
a Note: T is calcination temperature.
C
2572
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573

View Article Online
low H partial pressures and high temperatures, its increas-
ingly high cracking selectivity at increasing temperatures, and
the kinetics was not inÑuenced signiÐcantly by the acidity of
the catalysts and because the unsaturated intermediates play
an important role in the chemistry, we infer that the dehydro-
genation (metal) function was more important in inÑuencing
the catalyst performance than the acidic functionÈbut we are
aware of the impreciseness of this statement.
As the reaction network is complex and the primary and
secondary reactions not resolved, the kinetics results cannot
be regarded as more than just empirical; they represent
overall reactions and not a resolution of the individual reac-
tions.
2
the suppression of its deactivation in the presence of high H
partial pressures, are typical of the performance of bifunction-
2
al catalysts of the type used for naphtha reforming,21 exempli-
Ðed by platinum supported on chlorided Al O . The classical
2
3
bifunctional catalysts activate an alkane by Ðrst dehydrogen-
ating it (e.g., on Pt), and the resultant alkene is transferred to
an acidic site on the Al O (which is made a stronger proton
2
3
donor by the presence of chloride), where it is protonated to
give a carbenium ion, which isomerizes. The isomerized carbe-
nium ion gives a proton back to the catalyst to form an
alkene, which is transferred to a hydrogenation site to form
the product alkane.21 Alternatively, the carbenium ion inter-
mediate can undergo b-scission (cracking) or be involved in
CÈC bond formation reactions.22
Acknowledgements
The work done in Munich was supported Ðnancially by the
Deutsche Forschungsgemeinschaft (SFB 338). The coopera-
tion with the University of California, Davis, was supported
by the Alexander von Humboldt-Stiftung and the BMBF
(Max Planck-Research Award to HK). BCG thanks the Alex-
ander von Humboldt-Stiftung for support in Munich.
The unsaturated intermediates are responsible for the loss
of isomerization activity, e.g., by formation of higher-
molecular-weight species, including coke, which blocks cata-
lytic sites. Platinum in such catalysts, including ours, helps to
minimize deactivation resulting from coke formation by cata-
lyzing hydrogenation of unsaturated species that are coke pre-
cursors.
In our PtWTi samples, the platinum dispersions were about
the same for the catalysts containing 0.3 and 3% Pt, and thus
there was a much higher platinum surface area in the latter.
Consequently, a longer time was required to cover the plati-
num sites (with coke or related deposits), and thus a longer
time was required to suppress the dehydrogenation function
of the catalyst containing 3% Pt.
The fact that the dehydrogenation function remained active
over the whole period of operation is demonstrated by the
data showing that at long times on stream exclusively unsatu-
rated products were observed in the product stream. Under
these conditions the catalysts were still active for dehydrogen-
ation, but the acidic sites had been rendered largely inactive
for isomerization, evidently having undergone faster deactiva-
tion than the platinum dehydrogenation sites. We infer that at
short times on stream, when almost no unsaturated products
desorbed from the catalyst surfaces, the acidic sites e†ectively
removed almost all the alkenes formed on the platinum sites.
The data show that the catalyst performance, including the
kinetics, was nearly independent of the di†erences in acid
strength from one sample to another, which correlates with
the tungsten loading (Table 2).
By inference from this observation, we might speculate that
the di†erences in catalytic performance between tungstated
zirconia and tungstated titania are largely unrelated to di†er-
ences between the acid strengths of the zirconia and titania
supports. Thus, we speculate instead that the di†erences are
related primarily to the role of platinum, perhaps inÑuenced
by the structures of platinum and/or platinumÏs role in spill-
over of hydrogen. Data are lacking to provide insight into
these issues.
References
1
C. Gosling, R. Rosin, P. Bullen, T. Shimizu and T. Imai, Petrol.
T echnol. Q., 1997/98, Winter, 55.
2
3
4
5
6
J. C. Yori, J. C. Lui and J. M. Parera, Catal. T oday, 1989, 5, 493.
F. T. T. Ng and N. Horvat, Appl. Catal. A, 1995, 123, 195.
E. Ebitani, J. Konishi and H. Hattori, J. Catal., 1991, 130, 257.
G. Larsen and L. M. Petkovic, Appl. Catal. A, 1996, 148, 155.
S. R. Vaudagna, R. A. Comelli and N. S. F•goli, Appl. Catal. A,
1997, 164, 265.
7
8
9
F. Garin, D. Andriamasinoro, A. Abdulsamad and J. Sommer, J.
Catal., 1991, 131, 199.
K. Tanabe, H. Hattori and T. Yamaguchi, Crit. Rev. Surf. Chem.,
1999, 1, 1.
M. Hino and K. Arata, Chem. L ett., 1979, 1259.
10 M. Scheithauer, T.-K. Cheung, R. E. Jentoft, R. K. Grasselli, B.
C. Gates and H. Knozinger, J. Catal., 1998, 180, 1.
11 G. Kunzmann, Doctoral Dissertation, Universitat Munchen,
1987.
12 T.-K. Cheung, J. L. dÏItri and B. C. Gates, J. Catal., 1995, 151,
464.
13 G. C. Bond, S. Flamerz and L. Van Wijk, Catal. T oday, 1987, 1,
229.
14 E. Iglesia, D. G. Barton, S. L. Soled, S. Miseo, J. E. Baumgartner,
W. E. Gates, G. A. Fuentes and G. D. Meitzner, Stud. Surf. Sci.
Catal., 1996, 101, 533.
15 D. C. Vermaire and P. C. van Berge, J. Catal., 1989, 116, 309.
16 J. A. Anderson, F. K. Chong and C. H. Rochester, J. Mol. Catal.
A: Chem., 1999, 140, 65.
17 S. Eibl, Doctoral Dissertation, Universitat Munchen, 1999.
18 H. Knozinger, in Proc. Int. Symp. AcidÈBase Catalysis, Sapporo,
ed. K. Tanabe, H. Hattori, T. Yamaguchi and T. Tanaka, Else-
vier, Amsterdam, 1988, p. 147.
19 G. F. Froment and K. B. Bischo†, Chemical Reactor Analysis and
Design, Wiley, New York, 2nd edn., 1989, p. 93.
The occurrence of a bifunctional reaction network involving
acid- and metal-catalyzed reactions is supported by the
observed reaction kinetics characterizing partially deactivated
20 K. Arata, Adv. Catal., 1990, 37, 165.
21 B. C. Gates, J. R. Katzer and G. C. A. Schuit, Chemistry of Cata-
lytic Processes, McGraw-Hill, New York, 1979, p. 190.
22 F. C. Jentoft and B. C. Gates, T op. Catal., 1997, 4, 1.
23 G. Larsen, E. Lotero, S. Raghavan, R. D. Parra and C. A.
Querini, Appl. Catal., A, 1996, 139, 201.
catalysts. Reaction orders in H for n-butane and n-pentane
2
conversion catalyzed by PtWTi are negative (roughly [1), as
has been observed before for this kind of network.19 Because
24 J. H. Sinfelt, H. Hurwitz and J. C. Rohrer, J. Catal., 1962, 1, 481.
Phys. Chem. Chem. Phys., 2000, 2, 2565È2573
2573