Overcoming the deleterious effect of hafnium in tungsten-zirconia catalysts: The use of doping and thermal treatments

Article abstract of DOI:10.1016/j.jcat.2008.12.010

The effect of hafnium and aluminum addition to Ni/WO₃/ZrO₂ was examined as a function of calcination temperature for the isomerization of a 5 wt% cyclohexane in n-hexane feed. Hafnium, a common and difficult to separate impurity in zirconia, was found to decrease catalytic activity, primarily through the generation of monoclinic and cubic phases that are considered inactive for acid-catalyzed reactions. A hafnium induced monoclinic phase is formed at relatively low calcination temperatures and is fundamentally different from the thermodynamically stable monoclinic phase generated at higher temperatures. The addition of aluminum as well as higher calcination temperatures limited the formation of this monoclinic phase resulting in very active catalysts without the use of more expensive, hafnium-free catalyst precursors. Catalysts with tetragonal zirconia contents less than 40 wt% had initial isomerization activities that scaled linearly with tetragonal zirconia content. Catalysts with a tetragonal zirconia contents greater than 40% had initial n-hexane isomerization activities that were unrelated to zirconia morphology; a limitation that was a result of the hydrogenation function. The lack of correlation between zirconia crystal phases and catalytic activity for high tetragonal zirconia content catalysts implies that the measured isomerization activity may better represent the hydrogen spillover rate of nickel and not necessarily the intrinsic acid-catalyzed rate. Improved control over hydrogenation activity would allow for the preparation of more active catalysts.

Full text of DOI:10.1016/j.jcat.2008.12.010

Journal of Catalysis 262 (2009) 127–133
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Overcoming the deleterious effect of hafnium in tungsten–zirconia catalysts:
The use of doping and thermal treatments
David Simon a, Bradley Taylor b,∗
a
Analytical Services, ConocoPhillips, Bartlesville Technology Center, 234 Petroleum Laboratory, Bartlesville, OK 74004, USA
Advanced Hydrocarbon Fuels, ConocoPhillips, Bartlesville Technology Center, 333-1 Petroleum Laboratory, Bartlesville, OK 74004, USA
b
a r t i c l e i n f o
a b s t r a c t
Article history:
The effect of hafnium and aluminum addition to Ni/WO3/ZrO2 was examined as a function of calcination
Received 6 October 2008
Revised 1 December 2008
Accepted 16 December 2008
Available online 14 January 2009
temperature for the isomerization of
a 5 wt% cyclohexane in n-hexane feed. Hafnium, a common
and difficult to separate impurity in zirconia, was found to decrease catalytic activity, primarily
through the generation of monoclinic and cubic phases that are considered inactive for acid-catalyzed
reactions. A hafnium induced monoclinic phase is formed at relatively low calcination temperatures
and is fundamentally different from the thermodynamically stable monoclinic phase generated at
higher temperatures. The addition of aluminum as well as higher calcination temperatures limited the
formation of this monoclinic phase resulting in very active catalysts without the use of more expensive,
hafnium-free catalyst precursors. Catalysts with tetragonal zirconia contents less than 40 wt% had initial
isomerization activities that scaled linearly with tetragonal zirconia content. Catalysts with a tetragonal
zirconia contents greater than 40% had initial n-hexane isomerization activities that were unrelated to
zirconia morphology; a limitation that was a result of the hydrogenation function. The lack of correlation
between zirconia crystal phases and catalytic activity for high tetragonal zirconia content catalysts implies
that the measured isomerization activity may better represent the hydrogen spillover rate of nickel and
not necessarily the intrinsic acid-catalyzed rate. Improved control over hydrogenation activity would
allow for the preparation of more active catalysts.
Keywords:
Isomerization
Tungsten–zirconia
WO3/ZrO2
Hafnium
Nickel
Hexane isomerization
Cyclohexane isomerization
© 2008 Elsevier Inc. All rights reserved.
1. Introduction
was rekindled in 1987 with the discovery of Hammett acidity in
excess of that of sulfuric acid for tungsten oxides supported on
As part of a continuing effort to regulate the production of
cleaner automotive fuels, the U.S. Environmental Protection Agency
(EPA) has adopted new rules concerning the composition of gaso-
line that would further reduce the maximum benzene content to
0.62 vol% by 2011 [1]. This reduction effectively halves the content
of benzene within gasoline sold in the United States and results
in a significant loss of octane within the gasoline blending pool.
The isomerization of n-hexane represents one method of providing
high octane, paraffinic compounds for blending into gasoline.
Hexane isomerization over Pt/chlorided alumina is a mature
and widely commercialized technology. However, these materials
require replenishment of corrosive chlorine as well as the use of an
expensive hydrogenation metal. Tungsten–zirconia represents one
class of materials that could find use in future industrial applica-
tions.
zirconia [3]. Though Hammett acidity is considered an inaccurate
measure of acidity in solid materials, tungsten oxides supported on
zirconia were subsequently shown to be active in a number of acid
catalyzed reactions such as xylene isomerization [4–6], 2-butanol
dehydration [7,8], paraffin isomerization [9–11] and paraffin crack-
ing [12]; often at reaction temperatures lower than for acidic ze-
olites. Additionally, in contrast to zeolites, which provide acidity
within the confines of a pore system, tungsten–zirconia materi-
als provide the high acidity of H-mordenite [13] without any mass
transfer limitations or shape selectivity introduced by microporos-
ity.
Hafnium is a common impurity in some zirconia precursors
that is difficult and expensive to separate because of its chemical
similarity to zirconium. This work focuses on the effect of hafnium
on the catalyst morphology and ultimately the isomerization activ-
ity of Ni/WO3/ZrO2. Thermal treatments and aluminum doping are
presented as an efficient means of mitigating the adverse effects of
hafnium. Special care has been taken to track catalyst morphology
as a function of metals content and calcination temperature. While
the materials presented in this work do not possess all of the prop-
Despite research into the catalytic properties of tungsten ox-
ides dating back to the early 1900’s [2], interest in these materials
Corresponding author. Fax: +1 918 662 1097.
E-mail address: brad.m.taylor@conocophillips.com (B. Taylor).
*
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.12.010

128
D. Simon, B. Taylor / Journal of Catalysis 262 (2009) 127–133
Table 1
Calcination temperature and elemental composition of prepared catalysts.
Catalyst
Nickel loading
(wt%)
Tungsten surface
loading (W/nm2)
Tungsten
loading (wt%)
Aluminum
doping (wt%)
Hafnium impurity
(wt%)
Calcination
temperature ( C)
Surface area
◦
(m2/g)
◦
◦
◦
◦
◦
Ni/WO3/ZrO2-600
Ni/WO3/ZrO2-650
Ni/WO3/ZrO2-700
Ni/WO3/ZrO2-800
Ni/WO3/ZrO2-900
C
C
C
C
C
3.32
2.78
2.54
2.78
2.33
2.74
3.18
3.04
2.26
2.86
2.93
3.14
2.59
2.85
2.37
5.85
10.6
9.83
10.9
17.1
4.46
6.52
7.98
13.2
4.77
7.30
10.8
17.8
17.9
18.6
18.2
16.2
16.4
16.2
16.9
17.0
16.4
16.2
16.7
15.7
16.5
17.4
16.6
26.8
27.1
–
–
–
–
–
–
–
–
–
–
–
–
600
650
700
800
900
500
600
700
800
600
750
825
900
750
750
102
50.2
54.7
48.9
32.4
125
82.5
66.6
41.6
108
74.1
52.9
30.6
49.0
47.8
–
◦
◦
◦
◦
Ni/WO3/ZrO2(Hf)-500
Ni/WO3/ZrO2(Hf)-600
Ni/WO3/ZrO2(Hf)-700
Ni/WO3/ZrO2(Hf)-800
C
C
C
C
1.41
1.38
1.41
1.41
1.43
1.41
1.40
1.35
–
–
◦
◦
◦
◦
Ni/WO3/ZrO2(1AI,Hf)-600
Ni/WO3/ZrO2(1AI,Hf)-750
Ni/WO3/ZrO2(1AI,Hf)-825
Ni/WO3/ZrO2(1AI,Hf)-900
C
C
C
C
1.03
1.13
1.08
1.07
–
◦
Ni/WO3/ZrO2-750 C-3h
◦
Ni/WO3/ZrO2-750 C-24h
–
–
erties necessary for a commercialized isomerization catalyst, the
results show that relatively inexpensive and impure catalyst pre-
cursors, in combination with base metals, can be engineered to
produce considerable hydroisomerization activity.
zirconyl chloride solution prior to precipitation with ammonium
hydroxide. In this work, aluminum content ranged from 0–3 wt%
aluminum in the final catalyst. The results of the 1 wt% aluminum
catalysts are presented for brevity.
A summary of the prepared catalysts can be found in Table 1,
where the catalyst names are derived from the approximate alu-
minum content, hafnium content and calcination temperature such
2. Experimental
◦
Three variations of Ni/WO3/ZrO2 catalysts were prepared using
zirconia precursors with and without a hafnium impurity. One of
the hafnium-containing series was doped with aluminum. Nickel
and tungsten content were kept essentially constant on a per
weight basis.
that Ni/WO3/ZrO2(1Al,Hf)-600 C denotes a Ni/WO3/ZrO2 catalyst
containing approximately 1 wt% aluminum, a hafnium impurity
◦
and calcined at 600 C. Nickel and tungsten loading is not incorpo-
rated into the naming scheme as they were kept roughly constant
for all of the materials. Hafnium content was either 0 or 1.4 wt%.
2.1. Catalyst preparation
2.2. Catalytic measurements
Catalyst preparation began with the precipitation of amorphous
Zr(OH)4. Sufficient concentrated aqueous ammonium hydroxide
was added drop-wise to a 0.25 molar aqueous solution of zirconyl
chloride (Aldrich, Reagent Grade, 98%, ∼0.5% Hf or Aldrich, 99.99%)
under vigorous stirring to obtain a final pH of 10.5–11. The re-
sulting slurry was allowed to age for 1 h under vigorous stirring
before being filtered and washed with approximately 3 times its
volume in distilled water. The filter cake was dried in a vacuum
Catalytic measurements were taken using
a typical ¾-inch
outer diameter down flow packed bed reactor. Liquid feed was
metered to the system using an ISCO Model 500 D syringe pump
while gases were supplied by Brooks 5850E mass flow controllers.
The temperature at the center of the catalyst bed was measured
using a type K thermocouple. Pressure was maintained using a
Moore regulator. Analysis of the reactor effluent was completed
using a slipstream line to an online gas chromatograph. The online
effluent sampling allowed for the quantification of light cracking
products, however, the conditions in this study were sufficiently
mild that cracking is not a significant reaction path. The online
analysis was performed using an HP 6890 gas chromatograph out-
fitted with a 60 m × 0.25 mm CP-Sil 5 Low Bleed/MS column
(1 μm film) plumbed to the split/splitless inlet and a flame ioniza-
tion detector.
◦
oven for 2 days at 120 C and approximately 0.5 bar. Once dry, the
Zr(OH)4 was washed a second time in approximately 3 times its
volume of distilled water to ensure all of the residual chloride ions
were rinsed from the solid. The Zr(OH)4 was dried overnight in a
◦
vacuum oven at 120 C and approximately 0.5 bar. Tungsten was
deposited on the zirconium hydroxide via incipient wetness im-
pregnation using an aqueous solution of ammonium metatungstate
((NH4)6H2W12O40·xH2O, Aldrich, ꢀ66.5% W) added drop-wise to
Zr(OH)4 taken directly from the vacuum oven. The ammonium
metatungstate solution concentration was adjusted so as to pro-
duce a final material containing approximately 15 wt% W. The
The n-hexane isomerization activity was determined using a
5 wt% mixture of cyclohexane (Aldrich, CHROMASOLV®, ꢀ99.7%)
in n-hexane (Aldrich, CHROMASOLV® Plus, ꢀ95%). Reactor pressure
was 15 bar with a liquid weight hourly space velocity (LWHSV)
◦
−1
wetted support was dried overnight in a vacuum oven at 120 C
of 17 h
and hydrogen to hydrocarbon ratio (H2/HC) of 0.7. The
and approximately 0.5 bar. The dried material was calcined in air
for three hours at temperatures ranging from 500 to 900 C re-
typical charge of catalyst in the reactor was 1 g, sieved to 40–
100 mesh and mixed with sufficient alundum to produce a 13 mL
catalyst bed. The catalyst was pretreated in situ using a 120 sccm
◦
sulting in materials with different crystal phase compositions and
surface tungsten loadings (W/nm2). Nickel was subsequently added
by incipient wetness impregnation to calcined tungsten–zirconia
◦
stream of hydrogen at 370 C and 15 bar for one hour to fully re-
duce the hydrogenation metal and eliminate any water intercalated
in the catalyst pores. All gases were industrial grade and supplied
by Best Welders Supply, Tulsa, Oklahoma.
◦
materials that had been dried overnight in a vacuum oven (120 C,
∼0.5 bar) using aqueous solutions of nickel(II) nitrate (Aldrich,
Reagent Grade) at a concentration sufficient to produce a 3 wt%
Ni material. The catalyst was again dried overnight in a vacuum
2.3. Catalyst characterization
◦
◦
oven (120 C, ∼0.5 bar) before being calcined at 500 C for 3 h in
air.
Catalyst composition was determined by X-ray fluorescence
(Philips 1480 or 2440 Wavelength Spectrometer), surface area by
nitrogen adsorption (QuantaChrome AS-3 or AS-6) and crystal
Aluminum was incorporated into the zirconia phase through
the addition of aluminum nitrate (Aldrich, Reagent Grade) to the

D. Simon, B. Taylor / Journal of Catalysis 262 (2009) 127–133
129
phase composition by refinement of powder X-ray diffraction pat-
terns (Pan Analytical Xpert Pro, CuK ). Tungsten surface loading,
α
expressed in atoms per square nanometer, was calculated using
X-ray fluorescence data and BET surface area. X-ray refinements
◦
were performed on all materials following calcination at 500 C for
three hours and prior to activity testing.
The primary features of the X-ray diffraction patterns were
modeled using tetragonal zirconia and tungsten oxide. Monoclinic
zirconia, cubic zirconia and nickel tungstate were added to im-
prove the fit to the major features and account for peaks unas-
sociated with tetragonal zirconia. There was some difficultly in de-
termining the location of a predominance of the deposited nickel.
Since the nickel loading was known from X-ray fluorescence data,
an attempt was made to begin the refining of the complete pattern
containing nickel tungstate.
A significant amount of tungsten was not accounted for in the
tungsten oxide phase or the nickel tungstate phase. The exami-
nation of tungsten–zirconia materials (results not shown) showed
that in the absence of nickel, a zirconium tungstate phase was
present. The inclusion of this phase allowed for the accounting
of the deposited metal. In essence, the zirconium tungstate phase
modeled the interface between tungsten-containing phases and
the zirconia support. Similarly to the modeling of nickel phase,
X-ray fluorescence data provided starting compositions for the
structural refinements of the tungsten phases.
Fig. 1. Crystallographic phase content and initial n-hexane isomerization activity ver-
◦
◦
C
sus calcination temperature for catalysts Ni/WO3/ZrO2(Hf)-500 C through 800
◦
(288 C, 15 bar, LWHSV = 17 h−1, H2/HC = 0.7).
mote the formation of the more thermodynamically stable mon-
oclinic phase [4,5,7–12,14,15]. These literature accounts, however,
Despite a good overall fit to the experimental data, the three-
zirconia-phase model (tetragonal, monoclinic and cubic) did not
◦
generally refer to calcination temperatures above 800 C. The gen-
eration of tetragonal zirconia with increased calcination temper-
ature may also be an effect of tungsten surface loading. Sur-
face tungsten has been shown to stabilize the tetragonal zirconia
phase [14] and studies of hafnia/zirconia mixtures in the absence
of tungsten implied no regeneration of tetragonal zirconia follow-
ing formation of the monoclinic phase [16]. The elimination of
zirconia surface area due to sintering at high temperature may
result in a tungsten surface loading that ultimately stabilizes, in
this case, the tetragonal and cubic zirconia phases. The stabilizing
effect of tungsten is manifested in two respects. First, these materi-
als required higher calcination temperatures to produce crystalline
materials as compared to the tungsten-free case presented in refer-
ence [16]. Second, tungsten addition generated tetragonal zirconia
at the expense of monoclinic, but only with the aid of thermal
treatments.
◦
perfectly reproduce some of the small peaks (35–50 2θ) and
◦
shoulders (∼60 2θ) in the X-ray diffraction data. These minor dis-
crepancies led to the inclusion of an orthorhombic zirconia phase
and, in the case of amorphous materials, an additional monoclinic
phase. The data presented in this report utilizes X-ray fluores-
cence data coupled with X-ray diffraction data to assign nickel
tungstate (Space Group P12/c1) and zirconium tungstate (P213)
contents and includes up to five zirconia phases to fit the fea-
tures of the diffraction pattern (monoclinic = P121/c1, tetragonal =
P42/nm, orthorhombic = Pbc21 and cubic = P213). In an effort to
prevent over specification of the system, only the tetragonal phase
was modeled with Williamson–Hall strain.
3. Results and discussion
The orthorhombic phase appears as an intermediate phase dur-
ing the transition from monoclinic to tetragonal phase. Increases
in calcination temperature reduce the orthorhombic content in
favor of the catalytically active tetragonal phase. It is not clear
there is any real trend with calcination temperature in the cubic
phase content. Given the similar tungsten and nickel loadings for
each catalyst, the nickel tungstate and zirconium tungstate con-
centrations were constant. The appearance of tungsten oxide cor-
responded to calcination temperatures that resulted in a tungsten
surface loading of approximately two monolayers.
Catalytic activity increases with increasing tetragonal zirconia
content, a relationship previously asserted in the open literature
[4,5,7–12]. The initial isomerization rate appears to saturate at
tetragonal contents greater than approximately 40 wt%. Despite
being crystallographically distinct, the tetragonal and orthorhom-
bic phases are very similar. Their symmetry is essentially identical
except for a minor lengthening in one lattice parameter. A more
linear relationship between catalyst composition and activity can
be reproduced when initial isomerization activity is compared to
the sum of the tetragonal and orthorhombic phases. This makes a
case for the over specification of the model, given that orthorhom-
bic zirconia has not been identified as catalytically active in the
literature and activity has been shown previously to scale with
tetragonal content.
A number of Ni/WO3/ZrO2 catalysts, some containing hafnium
and/or aluminum and calcined at temperatures ranging from 500
to 900 C were prepared and are presented in Table 1. These cat-
alysts were prepared at essentially a constant weight of tungsten
with the surface tungsten concentration being set by the sintering
of the support at various calcination temperatures.
◦
3.1. The effect of hafnium
◦
◦
Catalysts Ni/WO3/ZrO2(Hf)-500 C through 800 C were pre-
pared from the comparatively inexpensive, hafnium-containing zir-
conia precursor. Shown graphically in Fig. 1, increasing calcination
temperature results in an increase in tetragonal zirconia content at
◦
the expense of the monoclinic. Ni/WO3/ZrO2(Hf)-500 C was best
fit using a short-range-ordered and a long-range-ordered mono-
clinic phase. If viewed as two separate monoclinic phases, the
long-range-ordered phase comprised 19.5% and the short-range-
ordered phase comprised 45.8% of the sample by weight. The
short-range-ordered monoclinic phase can be considered an amor-
phous phase given the low crystal domain size and its identifica-
tion only at the lowest calcination temperature. The elimination
of the monoclinic phase with thermal treatments is somewhat
contrary to literature accounts which show high temperatures pro-

130
D. Simon, B. Taylor / Journal of Catalysis 262 (2009) 127–133
Fig. 3. Crystallographic phase content and initial n-hexane isomerization activity
Fig. 2. Crystallographic phase content and initial n-hexane isomerization activity
◦
versus calcination temperature for catalysts Ni/WO3/ZrO2(1Al,Hf)-600
C through
◦
◦
C
versus calcination temperature for catalysts Ni/WO3/ZrO2-600
C
through 900
◦
◦
900 C (288 C, 15 bar, LWHSV = 17 h−1, H2/HC = 0.7).
◦
(288 C, 15 bar, LWHSV = 17 h−1, H2/HC = 0.7).
when the monoclinic phase is again produced at high tempera-
tures. The monoclinic phase present, especially at low calcination
temperatures, in the hafnium-containing material must be a phase
uniquely the result of hafnium. The increase in monoclinic zirconia
in the hafnium-free materials between a calcination temperature
◦
◦
Catalysts Ni/WO3/ZrO2-600 C through 900 C were prepared
using a hafnium-free zirconia precursor (Fig. 2). This group of
materials has essentially no trend in crystallographic phase with
calcination temperature, though the hafnium-free catalysts have
lower monoclinic and higher tetragonal zirconia contents than the
hafnium-containing materials. Whereas the hafnium-containing
materials showed a systematic transition to the tetragonal phase
at the expense of the monoclinic and orthorhombic phases, the
hafnium-free materials, at best, can be viewed as having a roughly
constant composition of tetragonal and orthorhombic phases. The
inclusion of a cubic zirconia phase did not improve the model fit.
Catalytic activity best tracks with the tetragonal zirconia con-
tent which is consistent with the assertion in literature that tetrag-
onal zirconia content determines the acid-character and thus cat-
alytic activity of these materials. The trends in activity for the
hafnium-free materials clearly do not track with the sum of the
orthorhombic and tetragonal phases, nor does it appear to track
with any of the other phases not pictured in Fig. 2. While not sub-
stantially more active than the hafnium-containing catalysts, the
activity of these materials is higher in rough proportion to the in-
creased tetragonal content.
◦
of 800 and 900 C is the generation of the polymorph generally
discussed in the literature.
There is no discussion in the literature of the presence, let alone
effect of, cubic and orthorhombic phases in these materials. Ulti-
mately, the scattering from both phases is similar to that of the
tetragonal phase making identification difficult. Monoclinic phases
are easily discernable from tetragonal phases by the splitting of a
strong diffraction peak.
3.2. The effect of aluminum doping
Hafnium-containing catalysts with an aluminum concentration
of approximately 1 wt% and calcined at temperatures ranging from
◦
◦
600 C to 900 C were prepared. The catalytic activity of these ma-
terials as a function of calcination temperature is presented in
Fig. 3.
The incorporation of roughly 1 wt% aluminum, a doping level
suggested in the patent literature [9], resulted in a material with a
tetragonal zirconia content commensurate with the hafnium-free
materials. There was no evidence of monoclinic zirconia, while
the cubic zirconia content was similar to hafnium-containing, un-
doped catalysts. Though appearing at lower concentrations than
the aluminum-free, hafnium-containing catalysts, the orthorhom-
bic zirconia phase shows similar behavior in that high temperature
treatments promote its transition to other phases. While the or-
thorhombic phase appears to be a transitional phase between the
monoclinic and tetragonal phase in the aluminum-free, hafnium-
containing materials, this is not as clear cut in the 1 wt% aluminum
samples. Given the lack of observed monoclinic phase, these ma-
terials provide no evidence that the orthorhombic structure is a
transitional phase formed only upon the conversion of monoclinic
phases to tetragonal. In addition, the elimination of the orthorhom-
bic phase via heat treatments did not result solely in the genera-
tion of the tetragonal phase (as was seen in the un-doped sam-
ples); the cubic phase was also formed.
In contrast to the hafnium-containing series of materials, mon-
oclinic zirconia content does increase with temperature, though
Ni/WO3/ZrO2-900 C was calcined at a higher temperature than
◦
any other catalysts. Whereas high temperature calcination of
hafnium-containing catalysts has appeared to form tetragonal zir-
conia at the expense of monoclinic, cubic and orthorhombic, high
temperature treatment of the hafnium-free catalysts produced
monoclinic zirconia at the expense of orthorhombic.
The predominance of orthorhombic and tetragonal phases in
the hafnium-free catalysts implies that the hafnium impurity may
catalyze, and thus be located predominantly in, the monoclinic
phase. Hafnium has been shown to be extremely miscible in mon-
oclinic zirconia phases [17,18], adding very little structural stress
to the resulting solid solution [19] and has long been known
to catalyze the formation of monoclinic phases [16]. The mono-
clinic phase found at low calcination temperatures in the hafnium-
containing catalyst, however, cannot be the thermodynamically
stable monoclinic phase described in the literature. A thermody-
namically stable phase would not convert under heat treatments
to the thermodynamically unstable tetragonal phase; particularly
The addition of trivalent ions (Al3+
and Ga3+ [9], Ln3+ [20,21])
has been shown to increase isomerization activity and the addition

D. Simon, B. Taylor / Journal of Catalysis 262 (2009) 127–133
131
Table 2
Crystallographic composition of prepared catalysts.
Zirconia (wt%)
Catalyst
Tungsten
oxide (wt%)
Zirconium
tungstate (wt%)
Nickel tungstate
(wt%)
Tetragonal
Monoclinic
Orthorhombic
Cubic
◦
Ni/WO3/ZrO2-600
Ni/WO3/ZrO2-650
Ni/WO3/ZrO2-700
Ni/WO3/ZrO2-800
Ni/WO3/ZrO2-900
C
C
C
C
C
50.3
50.3
57.1
50.6
56.9
8.2
23.7
38.2
46.6
60.0
56.7
58.0
70.6
43.5
44.2
3.8
–
16.6
24.0
17.4
22.4
6.6
–
24.1
21
13.8
9.4
6.5
5.2
2.4
–
–
–
–
–
–
11.5
3.1
6.9
6
11.8
10.2
5
–
11.7
11.5
8.1
11.7
9.2
13.9
10.6
12.1
6.5
9.6
8.4
3.1
5.1
17.6
14.1
13.3
14.0
12.0
12.6
16.7
17.7
14.8
15.0
15.0
16.6
11.2
14.3
10.4
◦
◦
◦
◦
0.2
3.5
0.3
8.1
–
0.5
1.1
7.3
65.3
13.4
7.6
6.4
–
–
–
–
–
◦
◦
◦
◦
Ni/WO3/ZrO2(Hf)-500
Ni/WO3/ZrO2(Hf)-600
Ni/WO3/ZrO2(Hf)-700
Ni/WO3/ZrO2(Hf)-800
C
C
C
C
–
0.3
4.9
–
1.7
7.0
5.7
9.1
11.7
◦
◦
◦
◦
Ni/WO3/ZrO2(1AI,Hf)-600
Ni/WO3/ZrO2(1AI,Hf)-750
Ni/WO3/ZrO2(1AI,Hf)-825
Ni/WO3/ZrO2(1AI,Hf)-900
C
C
C
C
◦
Ni/WO3/ZrO2-750 C-3h
16.6
15.8
–
–
16.5
17.9
◦
Ni/WO3/ZrO2-750 C-24h
–
of aluminum was found to stabilize amorphous zirconia and delay
the transition from tetragonal to monoclinic phases [22]. Support-
ing zirconia on an alumina support has shown similar proper-
ties [23,24]. Aluminum, in this case, serves a purpose beyond just
stabilizing the tetragonal phase in that it prevents the generation
of the low temperature monoclinic phase, allowing for the prepa-
ration of WO3/ZrO2 catalysts of high tetragonal zirconia content
using more benign thermal treatments.
Despite the high tetragonal zirconia content of these catalysts,
the isomerization activity was less than that of the hafnium-free
materials. The relatively constant isomerization activity across a
range of calcination temperatures does not track with tetrago-
nal zirconia content. The tungsten surface loading over this range
of calcination temperatures ranges from 5 to 18 W/nm2 which
spans the optimum loading (5–10 W/nm2) literature has shown
to produce the most active catalysts [4–8,14]. The relatively con-
stant activity across a range of surface tungsten loadings suggests
a relatively unchanging number of active sites and implies a rela-
tively small number of active sites relative to tungsten atoms. The
relatively small number of active acidic sites has been noted pre-
viously [6,8]. Only the nickel tungstate phase was essentially con-
stant with calcination temperature. While activity does not track
with nickel or nickel tungstate content, an isomerization activ-
ity inconsistent with the tetragonal zirconia content and invariant
with changes in tungsten surface loading reinforces the possibil-
ity that activity is determined by hydrogenation/dehydrogenation
activity rather than the number of acid sites.
the external surface of zirconia reinforces the previous assertion
that the density of active acidic sites appears to be small relative
to the available surface area.
3.4. The effect of calcination time
◦
◦
Catalysts Ni/WO3/ZrO2-750 C-3h and Ni/WO3/ZrO2-750 C-24h
were prepared from the same tungsten–zirconia starting material
◦
except that Ni/WO3/ZrO2-750 C-24h was calcined for 24 h rather
than the standard 3 h (Table 1). Given that the crystallographic
composition for the two catalysts is nearly identical despite the
difference in calcination time (Table 2), the short calcination time
◦
is sufficient at 750 C to produce a consistent composition. The
stability of the orthorhombic phase over the course of 24 h cal-
cination period, while not eliminating it as a transitional phase
between monoclinic and tetragonal zirconia, at least shows that it
is a reasonably stable phase in an of itself. The similarity in phase
composition between these materials and the hafnium-free series
suggests that none of the trends are a result of slight variations
in calcination time. With an initial n-hexane conversion of 22.7%,
◦
Ni/WO3/ZrO2-750 C-24h has an activity that fits within the range
already observed for hafnium-free materials.
3.5. Catalytic activity of orthorhombic zirconia
The apparent assignment of catalytic activity to the orthorhom-
bic phase in the aluminum-free, hafnium-containing materials is
likely either an artifact of the structural model or a coincidence.
When the aluminum-free, hafnium-containing catalysts are viewed
side-by-side with the hafnium-free materials (Fig. 4), the relation-
ship between the orthorhombic content and temperature appears
to be a result of drawing a trend from insufficient information. The
addition of hafnium certainly retards the formation of tetragonal
zirconia at a given calcination temperature as well as promotes the
formation of a cubic phase. In fact, the calcination of a hafnium-
Though not presented for brevity, catalysts were also prepared
with an aluminum content of 0.55 wt% and 2.5 wt%. The crystallo-
graphic trends mirrored those of the 1 wt% case.
3.3. The effect of tungsten loading
◦
Catalyst Ni/WO3/ZrO2-750 C-3h was prepared similarly to cat-
◦
◦
alysts Ni/WO3/ZrO2-600 C through 900 C except for a near dou-
bling of the tungsten loading. The resulting catalyst had a tung-
sten surface loading greater than Ni/WO3/ZrO2-900 C and a sur-
◦
◦
◦
free material at 600 C produced a similar phase composition as
face area similar to Ni/WO3/ZrO2-800 C. When adjusted for the
◦
a hafnium-containing material calcined at 800 C. Though an arti-
changes in tungsten composition, the tetragonal zirconia content is
consistent with the entirety of the hafnium-free series of materi-
als. Deposited tungsten in excess of the surface capacity resulted
ficial construct based on this observation, Fig. 4 suggests that the
activity and phase composition of the hafnium-free materials are
simply extensions of the trends seen in the hafnium-containing
catalysts. It is in this context that assigning catalytic significance
to the orthorhombic phase becomes untenable. On average, the or-
thorhombic content of the aluminum-free materials is 20 wt% and
essentially invariant with changes in calcination temperature. As a
result, the tetragonal content of the material should be more im-
portant for predicting catalytic activity.
◦
in a tungsten oxide loading similar to Ni/WO3/ZrO2-900 C. With
an initial n-hexane conversion of 28.8%, the activity of this cat-
alyst also fell within the range already reported and appears to
show that while sufficiently high tungsten loadings to generate
bulk tungsten oxide are not a detriment to activity, the high load-
ing constitutes a waste of material. The absence of an impact on
isomerization activity with the deposition of 10 wt% of material on

132
D. Simon, B. Taylor / Journal of Catalysis 262 (2009) 127–133
of nearly 3 wt% nickel, there was no indication of significant hy-
drogenolysis activity under these conditions. The selectivity to hex-
ane isomers was essentially constant despite the loss of 40% of the
isomerization activity over the course of 5.5 h. The steady selectiv-
ity to dimethylbutanes suggests that their formation is a result of a
single adsorption event of n-hexane. The generation of dimethylbu-
tanes from methylpentanes would have been expected to suffer a
decrease in selectivity as acid sites were eliminated with time on
stream since two adsorption steps involving two different acid sites
would be required. Taking into account the methylpentane impuri-
ties in the n-hexane feed, their conversion would have to be over
60% to account for all of the dimethylbutanes produced and would
still not account for the essentially constant selectivity.
3.7. Limitation of activity by nickel
The use of initial conversion data in Figs. 1–3 was a necessity
because of the significant deactivation of these catalysts within just
hours on stream. Pre-reduction in hydrogen was assumed to charge
the surface with ample surface hydrogen making the initial iso-
merization the best measure of acid-catalyzed activity. This surface
hydrogen was apparently insufficient for samples of high tetrago-
nal zirconia content, since activity was not proportional to zirconia
morphology. Catalysts Ni/WO3/ZrO2(1Al,Hf)-600 C through 825 C
represent the best case for catalytic activity limitations as a re-
sult of nickel. Despite higher tetragonal zirconia contents, these
three catalysts had lower initial isomerization activities than the
un-doped materials. Given that the three catalysts have only mi-
nor differences in phase composition, a range of surface tungsten
loadings (4.77–10.8 W/nm2) and similar nickel loadings, a differ-
ence in activity would have been expected as the tungsten surface
loading reached the literature asserted maximum in activity be-
tween 5 and 10 W/nm2.
Fig. 4. Side-by-side comparison of the crystallographic composition and initial
◦
◦
n-hexane conversion for catalysts Ni/WO3/ZrO2(Hf)-500
Ni/WO3/ZrO2-600 C through 900 C (288 C, 15 bar, H2/HC = 0.7, LWHSV = 17 h
C
through 800
C
and
,
◦
◦
◦
−1
Feed = 5 wt% cyclohexane in n-hexane).
◦
◦
Generally, as acid catalyzed isomerization activity increased, the
rate of deactivation increased. Since the stability of these catalysts
is dependant on the ability of the hydrogenation metal to deliver
sufficient surface hydrogen to hydrogenate adsorbed isomerization
intermediates [11], the deactivation of these catalysts should give
some measure of the ability of nickel to supply this hydrogen. As
shown in Fig. 6, the stability of the un-doped hafnium-containing
materials decreases with increasing activity, a sign that the hydro-
genation function of the catalyst is increasingly incapable of pro-
viding sufficient surface hydrogen to aid in the desorption of un-
saturated isomerization surface intermediates. The high variability
in catalyst stability for the Ni/WO3/ZrO2 and Ni/WO3/ZrO2(1Al,Hf)
series of catalysts is also indicative of an accumulation of car-
bonaceous material on the catalysts surface. Given that there is
no indication of a change in acid-site strength or concentration
for these materials, this deactivation must be attributable to the
hydrogenation function provided by nickel. Despite having similar
nickel loadings for all of the examined catalysts, nickel is not pro-
viding spillover hydrogen to the same degree in each material. This
inability to control the hydrogen spillover rate from nickel leads to
the variability in catalyst stability and, in the case of the 1 wt%
aluminum doped samples, the inability to achieve the high activity
seen in materials with similar zirconia phase content.
Fig. 5. Conversion and selectivity versus time on stream for catalysts Ni/WO3/ZrO2-
650 C (288 C, 15 bar, LWHSV = 17 h−1, H2/HC = 0.7).
◦
◦
3.6. Activity and selectivity
◦
Conversion and selectivity data for catalyst Ni/WO3/ZrO2-650 C
versus time on stream are presented in Fig. 5. Initial deactivation is
rapid and although the deactivation rate slows after 60 min, there
is no steady state isomerization rate during the 5.5 h the catalyst
was examined. The higher conversion of cyclohexane relative to
n-hexane is expected given that higher boiling compounds adsorb
more strongly to the catalyst surface. The greater stability of cyclo-
hexane/methylcyclopentane carbo-cations relative to n-hexane is
also expected to increase the coverage of naphthenes and enhance
the rate of their conversion [25]. The reaction of cyclohexane re-
sulted almost entirely in methylcyclopentane. Despite the presence
4. Conclusion
These materials show that tungsten–zirconia-based catalysts are
far more crystallographically complex than has previously been ad-
dressed in the open literature. The presence of a hafnium impurity
is detrimental to the n-hexane isomerization activity of nickel-
tungsten–zirconia due to the formation of catalytically inactive
monoclinic and cubic zirconia. This effect can be mitigated us-
ing high temperature thermal treatments which can convert some

D. Simon, B. Taylor / Journal of Catalysis 262 (2009) 127–133
133
ble n-hexane isomerization activity when the acid and hydrogena-
tion functionalities are appropriately managed. This work outlined
the management of the acid functionality through the control of
tetragonal zirconia content. Though intended as a means to use
inexpensive catalyst precursors, an extension of this work could
use the combination of hafnium and aluminum incorporation and
thermal treatments as a means of controlling acid site density
on tungsten zirconia materials. Unfortunately, the same level of
control could not be exercised over the hydrogenation function-
ality.
References
[1] U.S. Environmental Protection Agency, Control of Hazardous Air Pollutants from
Mobile Sources, Docket No. EPA-HQ-OAR-2005-0036, published February 26th,
2007.
[2] C.H. Kline, V. Kollonitsch, Ind. Eng. Chem. 57 (1965) 53.
[3] M. Hino, K. Arata, J. Chem. Soc. Chem. Commun. (1987) 1259.
[4] D.G. Barton, S.L. Soled, G.D. Meitzner, G.A. Fuentes, E. Iglesia, J. Catal. 181
(1999) 57.
[5] R.D. Wilson, D.G. Barton, C.D. Baertsch, E. Iglesia, J. Catal. 194 (2000) 175.
[6] C.D. Baertsch, S.L. Soled, E. Iglesia, J. Phys. Chem. B 105 (2001) 1320.
[7] J. Macht, C.D. Baertsch, M. May-Lozano, S.L. Soled, Y. Wang, E. Iglesia, J. Catal.
227 (2004) 479.
[8] C.D. Baertsch, K.T. Komala, Y.-H. Chua, E. Iglesia, J. Catal. 205 (2002) 44.
[9] J.Y. Ying, J. Xu, US Patent 6,767,859 (2004).
[10] T.N. Vu, J. van Gestel, J.P. Gilson, C. Collet, J.P. Dath, J.C. Duchet, J. Catal. 231
(2005) 468.
[11] J.G. Santiesteban, D.C. Calabro, C.D. Chang, J.C. Vartuli, T.J. Fiedbig, R.D. Bastian,
J. Catal. 202 (2001) 25.
[12] V.M. Benitez, J.C. Yori, J.M. Grau, C.L. Pieck, C.R. Vera, Energy Fuels 20 (2006)
422.
Fig. 6. Percentage loss in n-hexane activity after 5 h on stream versus initial n-hex-
◦
ane isomerization activity for the catalysts presented in Table 1 (288 C, 15 bar,
[13] G. Larsen, E. Lotero, L.M. Pekovic, D.S. Shobe, J. Catal. 169 (1997) 67.
[14] J.R. Sohn, M.Y. Park, Langmuir 14 (1998) 6140.
[15] D.G. Barton, S.L. Soled, E. Iglesia, Top. Catal. 6 (1998) 87.
[16] L.-M. Tau, R. Srinivasan, R.J. DeAngelis, T. Pinder, B.H. Davis, J. Mater. Res. 3
(1988) 561.
[17] M.A. Krebs, R.A. Condrate, J. Am. Ceram. Soc. 65 (1982) 144.
[18] R. Ruh, H.J. Garret, R.F. Domagala, N.M. Tallan, J. Am. Ceram. Soc. 51 (1968) 23.
[19] R. Ruh, P.W.R. Corfield, J. Am. Ceram. Soc. 53 (1970) 126.
[20] R.D. Gillespie, US Patent 6,818,589 (2004).
[21] R.D. Gillespie, US Patent 6,977,322 (2005).
[22] B.M. Reddy, P.M. Sreekanth, Y. Yamada, T. Kobayashi, J. Mol. Catal. A Chem. 227
(2005) 81.
[23] T.N. Vu, J. van Gestel, J.P. Gilson, C. Collet, J.P. Dath, J.C. Duchet, J. Catal. 231
(2005) 453.
[24] V.M. Benitez, C.R. Vera, C.L. Pieck, F.G. Lacamoire, J.C. Yori, J.M. Grau, J.M. Parera,
Catal. Today 107–108 (2005) 651.
LWHSV = 17 h−1, H2/HC = 0.7). The lines are meant to guide the eye and represent
neither a least squares fit or an average value.
of the hafnium-induced monoclinic phase into catalytically active
tetragonal zirconia. Catalytic activity is determined by tetragonal
zirconia content up to 40 wt%. At higher tetragonal zirconia con-
tents, catalytic activity is limited by the hydrogenation properties
of nickel. Aluminum incorporation is capable of preventing the for-
mation of the hafnium-induced monoclinic phase and promotes
the formation of tetragonal zirconia at calcination temperatures
less severe than in its absence. Aluminum-doping had no effect
of the formation of cubic zirconia.
This collection of catalysts shows that the use of base metals
and impure catalyst precursors can produce materials of apprecia-
[25] A. Hollo, J. Hancsok, D. Kallo, Stud. Surf. Sci. Catal. 135 (2001) 4336.