Relating n-pentane isomerization activity to the tungsten surface density of WOx/ZrO2

Article abstract of DOI:10.1021/ja105519y

Zirconia-supported tungsten oxide (WO_x/ZrO₂) is considered an important supported metal oxide model acid catalyst, for which structure-property relationships have been studied for numerous acid-catalyzed reactions. The catalytic activity for xylene isomerization, alcohol dehydration, and aromatic acylation follows a volcano-shape dependence on tungsten surface density. However, WO_x/ZrO₂ has not been studied for more acid-demanding reactions, like n-pentane isomerization, with regard to surface density dependence. In this work, WO_x/ZrO₂ was synthesized using commercially available amorphous ZrO_x(OH)_4-2x and model crystalline ZrO₂ as support precursors. They were analyzed for n-pentane isomerization activity and selectivity as a function of tungsten surface density, catalyst support type, and calcination temperature. Amorphous ZrO_x(OH)_4-2x led to WO_x/ZrO₂ (WZrOH) that exhibited maximum isomerization activity at ～5.2 W?nm ^-2, and the crystalline ZrO₂ led to a material (WZrO ₂) nearly inactive at all surface densities. Increasing the calcination temperature from 773 to 973 K increased the formation of 0.8-1 nm Zr-WO_x clusters detected through direct imaging on an aberration-corrected high-resolution scanning transmission electron microscope (STEM). Calcination temperature further increased catalytic activity by at least two times. Bronsted acidity was not affected but Lewis acidity decreased in number, as quantified via pyridine adsorption infrared spectroscopy. WO _x/ZrO₂ exhibited isomerization activity that peaked within the first 2 h time-on-stream, which may be due to Zr-WO_x clusters undergoing an activation process.

Full text of DOI:10.1021/ja105519y

Published on Web 09/03/2010
Relating n-Pentane Isomerization Activity to the Tungsten
Surface Density of WO /ZrO2
x
†
‡
§
|
Nikolaos Soultanidis, Wu Zhou, Antonis C. Psarras, Alejandro J. Gonzalez,
§
‡
⊥
Eleni F. Iliopoulou, Christopher J. Kiely, Israel E. Wachs, and
,
†,#
Michael S. Wong*
Department of Chemical and Biomolecular Engineering, Rice UniVersity, Houston, Texas 77005,
Department of Materials Science and Engineering, Lehigh UniVersity, Bethlehem, PennsylVania
1
8015, Laboratory of EnVironmental Fuels and Hydrocarbons, Chemical Process Engineering
Research Institute/Center of Research and Technology Hellas, Thessaloniki, Greece, Research
Department, DCG Partnership, Pearland, Texas 77581, Operando Molecular Spectroscopy and
Catalysis Laboratory, Department of Chemical Engineering, Lehigh UniVersity, Bethlehem,
PennsylVania 18015, and Department of Chemistry, Rice UniVersity, Houston, Texas 77005
Received June 23, 2010; E-mail: mswong@rice.edu
x 2
Abstract: Zirconia-supported tungsten oxide (WO /ZrO ) is considered an important supported metal oxide
model acid catalyst, for which structure-property relationships have been studied for numerous acid-
catalyzed reactions. The catalytic activity for xylene isomerization, alcohol dehydration, and aromatic
acylation follows a volcano-shape dependence on tungsten surface density. However, WO
been studied for more acid-demanding reactions, like n-pentane isomerization, with regard to surface density
dependence. In this work, WO /ZrO was synthesized using commercially available amorphous ZrO
and model crystalline ZrO as support precursors. They were analyzed for n-pentane isomerization activity
and selectivity as a function of tungsten surface density, catalyst support type, and calcination temperature.
Amorphous ZrO (OH)4-2x led to WO /ZrO
(WZrOH) that exhibited maximum isomerization activity at ∼5.2
W·nm , and the crystalline ZrO led to a material (WZrO ) nearly inactive at all surface densities. Increasing
the calcination temperature from 773 to 973 K increased the formation of 0.8-1 nm Zr-WO clusters detected
through direct imaging on an aberration-corrected high-resolution scanning transmission electron microscope
STEM). Calcination temperature further increased catalytic activity by at least two times. Brønsted acidity
was not affected but Lewis acidity decreased in number, as quantified via pyridine adsorption infrared
spectroscopy. WO /ZrO exhibited isomerization activity that peaked within the first 2 h time-on-stream,
which may be due to Zr-WO clusters undergoing an activation process.
x 2
/ZrO has not
x
2
x
(OH)4-2x
2
x
x
2
-2
2
2
x
(
x
2
x
1
. Introduction
Supported metal oxide catalysts comprise an important class
of industrial catalytic materials, with tungstated zirconia (WO
ZrO ) representing an important model for an acid catalytic
material.
than sulfated zirconia and chlorided Pt/Al
S and HCl during reaction and regeneration conditions.
/ZrO have been
co-workers attributed the balance between in situ strong
Brønsted and Lewis sites to the high n-pentane isomerization
8
-10
activity observed at intermediate tungsten oxide loadings.
x
/
4
+
Kn o¨ zinger and co-workers proposed that Zr -exposed WO
polytungstates generated strong Brønsted acidity for structures
similar to that of heteropolyacids.
x
2
1
-5
It is strongly acidic and structurally more stable
, which can release
1
1-13
o-Xylene isomerization
2
O
3
5
-7
and 2-butanol dehydration activity were investigated by Iglesia
and co-workers, who proposed that the slight reduction of
surface polytungstate species formed in situ Brønsted acid sites,
H
2
Structure-activity correlations for WO
x
2
studied actively by several research groups. Santiesteban and
-2
5,14-18
of which maximum activity was found at ∼7-8 W·nm .
†
Department of Chemical and Biomolecular Engineering, Rice University.
Department of Materials Science & Engineering, Lehigh University.
Chemical Process Engineering Research Institute/Center of Research
Most recently, Ross-Medgaarden et al. proposed a model in
‡
§
(8) Santiesteban, J. G.; Vartuli, J. C.; Han, S.; Bastian, R. D.; Chang,
C. D. J. Catal. 1997, 168, 431–441.
and Technology Hellas.
|
DCG Partnership.
Lehigh University.
Rice University.
(9) Vartuli, J. C.; Santiesteban, J. G.; Traverso, P.; Cardona-Martin e´ z,
N.; Chang, C. D.; Stevenson, S. A. J. Catal. 1999, 187, 131–138.
(10) Calabro, D. C.; Vartuli, J. C.; Santiesteban, J. G. Top. Catal. 2002,
18, 231–242.
⊥
#
(
(
(
1) Hino, M.; Arata, K. J. Chem. Soc., Chem. Commun. 1988, 18, 1259.
2) Hino, M.; Arata, K. Chem. Lett. 1989, 6, 971.
(11) Kuba, S.; Lukinskas, P.; Ahmad, R.; Jentoft, F. C.; Grasselli, R. K.;
Gates, B. C.; Kn o¨ zinger, H. J. Catal. 2003, 219, 376–388.
(12) Scheithauer, M.; Cheung, T. K.; Jentoft, R. E.; Grasselli, R. K.; Gates,
B. C.; Kn o¨ zinger, H. J. Catal. 1998, 180, 1–13.
3) Barton, D. G.; Soled, S. L.; Iglesia, E. Top. Catal. 1998, 6, 87–99.
(
4) Ross-Medgaarden, E. I.; Knowles, W. V.; Kim, T.; Wong, M. S.; Zhou,
W.; Kiely, C. J.; Wachs, I. E. J. Catal. 2008, 256, 108–125.
(
5) Barton, D. G.; Soled, S. L.; Meitzner, G. D.; Fuentes, G. A.; Iglesia,
E. J. Catal. 1999, 181, 57–72.
(13) Scheithauer, M.; Grasselli, R. K.; Knozinger, H. Langmuir 1998, 14,
3019–3029.
(
(
6) Thomas, J. M. Sci. Am. 1992, 266, 112–118.
7) Corma, A. Catal. Lett. 1993, 22, 33–52.
(14) Wilson, R. D.; Barton, D. G.; Baertsch, C. D.; Iglesia, E. J. Catal.
2000, 194, 175–187.
1
3462 9 J. AM. CHEM. SOC. 2010, 132, 13462–13471
10.1021/ja105519y  2010 American Chemical Society

Relating Isomerization Activity to W Surface Density
A R T I C L E S
which maximum methanol dehydration activity observed at
a fine powder with a catalyst particle size of 150 µm or less (-170
mesh). This powder was then pelletized, crushed, and sieved into
the 300-600 µm range for all catalytic reaction experiments
(Supporting Information).
surface densities of 6-7 W·nm- was attributed to high
2
concentrations of 0.8-1.0 nm Zr-containing WO
x
three-
4
dimensional (“Zr-WO
x
”) clusters, which was later confirmed
The following sample notations are employed in this paper.
by Zhou et al. via direct imaging of all the surface WO
x
species
F
surf-WZrOH(z,T) refers to AMT-impregnated amorphous
ZrO (OH)4-2x, where Fsurf is the surface density calculated using
the surface area of the catalyst after calcination (W ·nm ),
is the tungsten oxide weight loading (wt% of WO ), and T is the
calcination temperature (K). For the samples supported on model
crystalline ZrO (z,T). A
using aberration-corrected STEM high-angle annular dark-field
1
9
x
(
HAADF) imaging. The surface acid sites active for this
-
2
4,22,23
z
reaction are presumably weaker than those required for more
acid-demanding reactions like alkane isomerization and crack-
3
2
0,21
ing.
x 2
However, systematic investigation of WO /ZrO cata-
2
the nomenclature used was Fsurf-WZrO
2
lysts for n-pentane isomerization, a more acid-demanding
simplified notation is used when referring to a specific series of
samples, namely WZrOH(T) and WZrO (T). Bulk WO powder
reaction, as a function of WO
reported before.
x
surface density has not been
2
3
(Sigma) was used without further purification as a control sample.
2.2. Catalyst Characterization. Nitrogen physisorption studies
were performed on Micromeritics ASAP 2010 using Matheson
ultrahigh purity (UHP) nitrogen. All synthesized samples with the
exception of the amorphous support were evacuated for more than
In this work, we report the catalytic properties as a function
of tungsten surface density and deduce the structure-activity
x 2
relationship in supported WO /ZrO solid acid catalysts for
n-pentane isomerization. Amorphous and crystalline zirconia
materials were used as support to prepare the WO /ZrO
-
3
4
h at 523 K until the degas rate was less than 4 × 10
x
2
-1
mmHg ·min .
catalysts through incipient wetness impregnation, which allowed
us to investigate the effect of the support material on the nature
X-ray diffraction (XRD) patterns were acquired on a Rigaku
4
D/Max-2100PC using a continuous scanning mode with a 0.02°
of the active sites. The atomic structure of various surface WO
x
-1
step size and a scan rate of 2.5 s ·step .
species was characterized by aberration-corrected STEM-
HAADF imaging. We assessed surface acidity through pyridine
adsorption FTIR studies.
Bright field (BF) images, selective area diffraction (SAD), and
X-ray energy dispersive spectroscopy (XEDS) of the samples were
obtained using a JEOL 2000FX TEM operating at 200 kV. High
resolution TEM (HRTEM) imaging and high-angle annular dark
2
4
2
. Experimental Methods
.1. Catalyst Preparation. All catalysts were synthesized by
field (HAADF) imaging were performed on a 200 kV JEOL
200FS (S)TEM equipped with a CEOS probe C -corrector at
2
s
2
Lehigh University. The HAADF images presented have been low-
pass filtered to reduce background noise. The catalyst samples were
also characterized by secondary electron (SE) imaging and back-
scattered electron (BSE) imaging on a Hitachi 4300LV scanning
electron microscope (SEM). Samples suitable for SEM analysis
were made by directly dispersing the catalyst powder onto carbon
tape and coated with iridium (Ir) to mitigate charging effects.
incipient wetness impregnation of an aqueous solution of am-
monium metatungstate ((NH O, AMT) into (1)
amorphous zirconium oxyhydroxide (ZrO (OH)4-2x, MEI XZO 880/
1) and (2) model crystalline zirconium oxide (ZrO , Degussa)
4
)
6
H
2
W
12
O
40 ·5H
2
x
0
2
supports. These two supports were initially sieved (-170 mesh)
and mixed overnight using an automated VWR rocking platform.
x
The amorphous ZrO (OH)4-2x was found to have a specific surface
2
-1
3
-1
Qualitative and quantitative acid site measurements were per-
formed on a Nicolet 5700 FTIR spectrometer using an MCT-A
area (SSA) of 330 m ·g and pore volume of 0.33 cm ·g , while
the crystalline ZrO support was found to have a specific surface
2
2
-1
3
-1
detector and a homemade stainless steel, vacuum cell, with CaF
2
area of 58 m ·g and pore volume of 0.15 cm ·g , as determined
from nitrogen physisorption analysis of three different batches of
each support type.
25
windows. Lewis (L) and Brønsted (B) site concentrations were
calculated according to the Beer-Lambert law corrected with the
normalized weight of the wafers, with a radius of 0.405 cm and
thickness of ∼1 mm. The molar extinction coefficients of 1.67 and
Prior to impregnation, the support was degassed in a vacuum
oven overnight at a moderate temperature (343 K) in order to
remove the excess moisture without causing any significant
structural changes. Aqueous solutions of AMT (Aldrich), with
different tungsten oxide loadings, were impregnated up to 95% of
the pore volume of the support. A correction to the calculated
aqueous AMT solution volume was applied prior to impregnation
2
.22 cm ·µmol- for the L and B sites, respectively, were used.
1
26
The weak and moderate acid sites were quantified by the amount
of pyridine desorbed in the ranges of 423-523 K and 523-723
K; the amount of undesorbed pyridine quantified the amount of
strong acid sites. For each temperature, the sample was cooled down
and spectra were collected at 423 K to avoid inconsistencies caused
by band broadening and intensity amplification at elevated tem-
peratures. L and B acid site concentrations were expressed in two
ways: (1) in micromoles of chemisorbed Py per gram of catalyst
22
according to previously reported observations to ensure accuracy.
After impregnation, all samples were hand mixed and dried at 343
K overnight in static air. Samples were then crushed, sieved, and
-1
finally heated up at a ramp rate of 3.0 K ·min under flowing air
9
,14
3
-1
and (2) in sites per W atom.
.3. Catalytic Studies. The catalytic studies were performed on
an isothermal downflow reactor at 523 K, with an internal diameter
ID) of 6.26 mm, packed with ∼0.33 g of catalyst (particle size in
the 300-600 µm range). The catalyst bed length was fixed to be
.22 cm long by adjusting the catalyst loading (0.02 g in order to
ensure a constant gas-hourly space velocity (GHSV ) 68). GHSV
(
100 cm ·min ) and calcined at the desired calcination temperature
2
for 3 h. Crushing and sieving were repeated once more to acquire
(
(
(
(
(
(
15) Baertsch, C. D.; Komala, K. T.; Chua, Y.-H.; Iglesia, E. J. Catal.
002, 205, 44–57.
16) Baertsch, C. D.; Soled, S. L.; Iglesia, E. J. Phys. Chem. B 2001, 105,
320–1330.
17) Baertsch, C. D.; Wilson, R. D.; Barton, D. G.; Soled, S. L.; Iglesia,
E. Stud. Surf. Sci. Catal. 2000, 130D, 3225–3230.
18) Barton, D. G.; Shtein, M.; Wilson, R. D.; Soled, S. L.; Iglesia, E. J.
Phys. Chem. B 1999, 103, 630–640.
19) Zhou, W.; Ross-Medgaarden, E. I.; Knowles, W. V.; Wong, M. S.;
Wachs, I. E.; Kiely, C. J. Nat. Chem. 2009, 1, 722–728.
20) Corma, A. Chem. ReV. 1995, 95, 559–614.
2
2
1
0 0
is equal to u /V, where u
) 0.78 mL ·min- is the volumetric flow
1
(23) Soultanidis, N.; Knowles, W. V.; Ross-Medgaarden, E. I.; Wachs, I. E.;
Wong, M. S. In preparation.
(24) Nellist, P. D.; Pennycook, S. J. Science 1996, 274, 413–415.
(25) Psarras, A. C.; Iliopoulou, E. F.; Kostaras, K.; Lappas, A. A.; Pouwels,
C. Microporous Mesoporous Mater. 2009, 120, 141–146.
(26) Emeis, C. A. J. Catal. 1993, 141, 347–354.
(27) Macht, J.; Baertsch, C. D.; May-Lozano, M.; Soled, S. L.; Wang, Y.;
Iglesia, E. J. Catal. 2004, 227, 479–491.
(
(
(
21) Corma, A. Curr. Opin. Solid State Mater. Sci. 1997, 2, 63–75.
22) Knowles, W. V.; Nutt, M. O.; Wong, M. S. Supported Metal Oxides
and Surface Density Metric; Taylor and Francis: Boca Raton, FL, 2006;
Chapter 11.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010 13463

A R T I C L E S
Soultanidis et al.
Table 1. BET Surface Area, Pore Volume, and Calculated
rate measured at standard temperature and pressure (STP) and V
Tungsten Surface Density Values of Various Supported WO
Catalysts
x
/ZrO
2
)
0.68 mL is the volume of the catalytic bed.
Prior to each reaction run, samples were pretreated in situ at
3
-1
6
73 K under a continuous flow (100 cm ·min ) of ultrahigh purity
Catalyst sample
(WO wt %, calcination
temperature K)
BET
SSA
[m ·g
Pore
volume
[cm ·g
Surface density Fsurf
(
UHP) air for 1 h. Then, the reactor was cooled down to 523 K
3
2
-1
3
-1
-2
-2
3
-1
]
]
[W·nm
]
[W· nmsupp ]
under flowing UHP He (100 cm ·min ) in order to remove any
physisorbed oxygen. The reactor feed gas was a blend of 1%
n-pentane and 1% argon in helium (prepared by gravimetric
blending).
The duration of all runs presented in this paper was 10 h, during
which a chromatogram was collected every 23 min. For calculating
the partial n-pentane conversion to the various products, a carbon
mass balance approach was used (eq 1) similar to the one presented
2.5-WZrOH (7.0, 973)
72.0
0.185
0.183
0.199
0.210
0.204
0.168
2.5
3.6
5.2
6.0
8.5
2.8
5.2
7.6
8.8
12.0
13.0
3
5
6
8
1
.6-WZrOH (13.2, 973)
.2-WZrOH (18.5, 973)
.0-WZrOH (21.7, 973)
.5-WZrOH (30.0, 973)
1.0-WZrOH (32.4, 973)
92.3
93.0
95.5
92.0
76.0
11.0
2
.2-WZrOH (9.2, 773)
109.0
138.0
128.0
98
0.230
0.218
0.206
0.185
2.2
3.5
4.4
8.0
2.9
6.0
7.1
9.5
1
1
3.5-WZrOH (18.5, 773)
4
8
by Kuba et al.
.4-WZrOH (21.7, 773)
.0-WZrOH (30.0, 773)
(
i/5)Ci
2
.2-WZrO
2
(4.8, 773)
56.0
53.0
52.0
52.0
0.195
0.165
0.163
0.147
2.2
3.5
4.4
6.1
2.2
3.2
4.0
5.5
xi )
(1)
nC5in
3.5-WZrO (7.0, 773)
4
6
2
.4-WZrO
.1-WZrO
2
(8.8, 773)
(12.2, 773)
2
where i ) 1-6 is the number of carbon atoms of each product
from methane to C olefins and hexane, C is the moles of each
product, and nC5in is the moles of n-pentane (nC ) fed to the reactor.
After normalizing with respect to the internal standard concentration,
the total conversion X [(mol of n-pentane converted) ·(mol of
n-pentane fed) ] was calculated by using eq 2.
6
i
5
and i > 5, respectively. Steady-state TORs and selectivities were
collected at t ) 10 h.
-
1
3. Results and Discussion
6
3
.1. Catalyst Structure. The SSA and V for the catalyst
p
∑(
i/5)Ci
samples are summarized in Table 1. Surface tungsten oxide
density values were calculated in two ways: (1) using the
measured SSA of the catalysts after calcination, to give units
of W ·nm , and (2) using the SSA of the support materials
(ZrO (OH)
W·nmsupp (Table 1). These calculations gave similar values
for low tungsten oxide content and differed significantly at high
tungsten oxide content.
The SSA of WZrOH samples increased to a maximum before
decreasing with tungsten oxide loading, which is consistent with
i)1
X )
(2)
nC5in
-
2
Selectivity for each product is defined as S
i
[(conversion of i
-1
or ZrO ) after calcination, to give units of
product) ·(total conversion) ]
x
4-2x
2
-
2
xi
X
Si )
(3)
4,22
Since the total conversion remained low (e3%) for all runs, the
conditions were considered to be differential, and therefore the total
, RnC5 [(mol of nC
catalyst ·s) ], could be expressed as
x
the reported thermal stabilizing role of WO on the amorphous
conversion rate of nC
5
5
converted) ·(g of
13,16,23
ZrO
model WZrO
the calcination.
x
(OH)4-2x support during calcination.
The SSA of the
-1
2
materials, however, was not affected much by
FnC5X
The XRD patterns for the WZrOH(973) material are presented
in Figure 1. All samples calcined at 973 K were crystalline with
RnC5
)
(4)
Wcat
both tetragonal (“t-ZrO
phases. Monoclinic WO
at higher Fsurf. The observed trends were in agreement with
published reports that indicated the ability of WO to retard the
phase transformation of t-ZrO to the thermodynamically more
XRD peaks for the m-WO crystal
phase were seen to emerge at a surface density between 3.6
2
”) and monoclinic (“m-ZrO
2
”) zirconia
-
1
where FnC5 [(mol of nC
of nC and Wcat [g] is the amount of catalyst used for each run.
In comparing the activity for the catalyst samples, the nC
5
)·s ] is the steady-state molar flow rate
3
(“m-WO ”) crystals were also detected
3
5
5
x
-1
consumption turnover rate TOR [s ] and the isomerization turnover
-1
2
rate TORiC5 [s ] were used, which respectively represented the
number of nC
4,5,13
stable m-ZrO
2
phase.
3
5
molecules reacted per tungsten atom ·per second
eq 5) and the number nC molecules isomerized into isopentane
iC ) per tungsten atom ·per second (eq 6).
(
(
5
-2
-2
and 5.2 W ·nm , which was lower than 7-8 W·nm observed
5
4
,18
by others.
3
Bulk WO formed at a lower surface density than
(
RnC5)(NA)
expected, perhaps due to incomplete spreading of the meta-
TOR )
(5)
(6)
x
tungstate salt solution over the ZrO (OH)4-2x support during the
1
8
(
N )(SSA)(10 )
23
s
impregnation process.
In contrast to the observations made for the WZrOH samples
calcined at 973 K, the WZrOH samples calcined at 773 K
TORiC5 ) SiC5TOR
appeared to be mostly amorphous at low Fsurf, and no m-ZrO
2
where N
area, and N
was assumed to be numerically equivalent to Fsurf. However, it
should be noted that TOR would not represent a true turnover
frequency above WO monolayer (ML) coverage, i.e., Fsurf > 4.5
because the tungsten oxide content would not be
00% accessible for reaction. For the rest of the products, cracking
and oligomerization selectivities were calculated using i ) 1-4
A
is Avogadro’s number, SSA is the BET specific surface
was formed in samples with W surface densities below 8.0
-2
4,22,23
-
2
s
[sites·nm ] is the active site surface density.
W·nm ; WO
crystals were not detected in any of the
species in these samples were
3
N
s
WZrOH(773) catalysts. The WO
x
therefore expected to be monotungstates and polytungstates
x
23,28-30
species.
-
2 4,8,18,27
W·nm ,
1
(28) Wachs, I. E.; Kim, T.; Ross, E. I. Catal. Today 2006, 116, 162–168.
1
3464 J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010

Relating Isomerization Activity to W Surface Density
A R T I C L E S
x 3 3 2
Figure 1. Powder XRD patterns of WZrOH and Zr (OH)4-2x calcined at 973 K and bulk WO . Crystalline phases marked as (1) m-WO , (b) m-ZrO , and
(
2
9) t-ZrO .
2 3 2 2
Figure 2. Powder XRD patterns of WZrO (773) and WZrOH(773). Crystalline phases marked as (1) m-WO , (b) m-ZrO , and (9) t-ZrO .
The crystal structure of the ZrO
catalysts remained similar at all of the investigated WO
densities, and a small fraction of WO crystals were observed
in the 6.1WZrO (12.2, 773) catalyst (Figure 2).
.2. Electron Microscopy Analysis. 3.2.1. Catalyst with Low
WO Loading below Monolayer Coverage. Bright field (BF) TEM
2
support in the WZrO
2
x
surface
3
2
3
x
imaging was used to characterize the morphology of the catalyst
samples. As shown in Figure 3a, two distinct morphologies were
observed in the 2.5-WZrOH (7.0, 973) catalyst: agglomerates
of smaller (5-12 nm) ZrO
ates of larger (15-40 nm) ZrO
HRTEM image of a typical larger ZrO
shows clear lattice fringes extending right out to the surface of
the grain, indicating a loading below the surface WO monolayer
coverage. In contrast, some dark speckles can be seen (Figure
c) at the boundaries and surface of the aggregates of the smaller
ZrO particles, which can be caused by either amorphous
interfacial films or clusters on the surface.
The structure and distribution of WO surface species were
2
particles (labeled X) and agglomer-
particles (labeled Y). The
particle (Figure 3b)
2
2
x
3
2
x
investigated using HAADF-STEM imaging, which provides
Z-contrast information. In Figure 3d, the heavier W atoms show
2
up as very bright spots while the ZrO crystals show a fainter
lattice fringe contrast. Features corresponding to surface mono-
tungstate (i.e., isolated W atoms as circled in blue) and surface
Figure 3. Representative (a) TEM BF, (b, c) HRTEM, and (d) HAADF-
STEM images of the supported 2.5-WZrOH (7.0, 973) catalyst. Blue
circles: surface monotungstate species; Green circles: surface polytungstate
x
species; Red circles: sub-nm Zr-WO clusters.
(
(
29) Ross-Medgaarden, E. I.; Wachs, I. E. J. Phys. Chem. C 2007, 111,
5089–15099.
30) Kim, D. S.; Ostromecki, M.; Wachs, I. E. J. Mol. Catal A: Chem.
996, 106, 93–102.
polytungstate (i.e., interconnected two-dimensional WO
with W atoms linked by oxygen bridging bonds as circled in
green) are visible on both ZrO morphologies. This observed
x
species
1
1
2
J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010 13465

A R T I C L E S
Soultanidis et al.
Figure 5. Representative (a) SEM BSE images of 5.2-WZrOH (18.5, 973)
3
and (b) 8.5-WZrOH (30.0, 973). WO crystals a few hundred nanometers
in size are circled.
2
Sub-nm dark flecks can be seen in both ZrO morphologies
in HRTEM images (Figure 4b). However, an accurate evaluation
of the size and number density of these clusters is only possible
from a lower magnification HAADF-STEM image (Figure 4c)
which shows a high number density of 0.8-1 nm clusters. The
high resolution HAADF image (Figure 4d) shows the WO
x
Figure 4. Representative (a) TEM B, (b) HRTEM, and (c, d) HAADF-
STEM images of the supported 5.2-WZrOH (18.5, 973) catalyst. Blue
circles: surface monotungstate species; Green circles: surface polytungstate
clusters having a considerably higher image contrast as com-
pared with the monotungstate and polytungstate species, con-
firming that these WO clusters are three-dimensional in nature
x
with a thickness of 2-3 atomic layers. A rough estimate, based
on the size of the clusters, suggests that each cluster should
x
species; Red circles: sub-nm Zr-WO clusters.
coexistence of surface mono- and polytungstate species at this
-2
low surface density (2.5 W ·nm ) sample suggests that poly-
tungstate species begin to emerge at a WO coverage lower than
.5 W ·nm , consistent with previous UV-vis spectroscopy
contain between 10 and 20 WO
x
structural units (as compared
x
to 2-6 WO structural units in the polytungstate species). Zhou
x
-2
2
et al. explained that subtle contrast variations observed within
a single cluster in HAADF images were most likely caused by
-
2
4
-2
results (1.7 W·nm ) and counter to the ∼4.0 W·nm value
1
8
concluded by others. The surface W atoms were found to sit
preferentially above the Zr atom columns; this phenomenon
the intermixing of a small amount of ZrO
WO clusters to form the catalytically active Zr-WO
The grain boundary grooves between intersecting ZrO
again served as preferential sites for WO clustering, but the
overall increase in WO loading saturated the ZrO surface and
formed a high number density of these Zr-WO clusters. The
Zr-WO clusters found in this more highly loaded sample had
x
species within these
19
x
x
clusters.
particles
2
becomes clearer when the ZrO crystal is oriented along a major
2
zone axis as in the upper right-hand side particle in Figure 3d.
x
1
9
This preferential location of W atoms has been proposed to
be a consequence of the strong interaction between WO species
and surface defect sites on the ZrO support. Although the
nominal WO surface coverage of this catalyst was below the
x
2
x
x
2
x
x
a larger average domain size (Figure 4d) compared to those
found in the lower loaded 2.5-WZrOH (7.0, 973) sample.
The much larger particles (labeled Z in Figure 4a) were
identified using EDS and electron diffraction techniques to be
WO crystals, which were identified to be the monoclinic phase
3
via XRD (Figure 1). Backscattered electron (BSE) imaging in
ML, occasional sub-nm clusters were still found (circled in red
in Figure 3d) especially at the intersections of adjacent small
ZrO
formation of such WO
presumably due to capillary effects, while the edge of surface
pits provides a large number of step edge sites for trapping WO
2
support particles and at the edge of surface pits. The
x
clusters at particle boundaries is
x
SEM also contains Z-contrast information and can be used to
species. It is important to note that these sub-nm clusters were
exclusively found to be associated with the agglomerates of
smaller ZrO particles (i.e., X morphology) with a very low
2
number density. Thus, the larger particles (i.e., Y morphology)
were possibly formed as a consequence of local inhomogeneities
locate the WO
higher atomic number. SEM BSE images from the 5.2-WZrOH
18.5, 973) catalyst sample (Figure 5a) clearly reveal the
distribution of 200-600 nm WO crystals in the sample. These
bulk WO crystals are known to possess very low activity for
methanol dehydration.
.2.3. Catalyst with WO
3 x 2
crystals in the WO /ZrO catalysts due to their
(
3
3
4
,31
in the WO
.2.2. Catalyst with WO
age. Three different morphologies were found in 5.2-WZrOH
18.5, 973) (Figure 4a). The first morphology (labeled X)
consists of agglomerates of small (5-12 nm) ZrO support
particles; the second morphology (labeled Y) consists of clumps
of larger (15-50 nm) ZrO support particles; the third distinct
x
distribution and calcination conditions.
3
x
Loading Close to Monolayer Cover-
3
x
Loading above Monolayer Cover-
age. Similar to the 5.2-WZrOH (18.5, 973) sample, the three
distinct morphologies (X, Y, and Z) were also observed in the
(
2
8
(
.5-WZrOH (30.0, 973) catalyst. The two ZrO
X and Y) are shown in Figure 6a. The larger Y-type ZrO
particles appeared even less frequently when compared with
the two samples discussed previously. The increase in WO
loading tended to inhibit the sintering of the ZrO support
particles and stabilized the smaller metastable tetragonal ZrO
polymorphs, as indicated by the peak broadening and increase
in signals for the tetragonal ZrO phase in the XRD spectrum.
A high number density of sub-nm Zr-WO clusters were also
observed in this sample using both HRTEM and HAADF-STEM
2
morphologies
2
2
morphology (labeled Z) is comprised of large isolated single
crystalline particles up to a few hundred nanometers in size.
The X and Y morphologies are similar to those found in the
low loading catalyst, but with a smaller volume fraction of
morphology Y, which is consistent with the slight peak
broadening noted in the corresponding XRD spectrum. Evidence
of internal voids and/or surface pits due to the loss of water
x
2
2
2
x
from the ZrO
the support could be seen in both ZrO
again serve as preferential sites for WO
x
(OH)4-2x precursor material during calcination of
2
morphologies, which
cluster nucleation.
(
31) Kim, T.; Burrows, A.; Kiely, C. J.; Wachs, I. E. J. Catal. 2007, 246,
x
370–381.
1
3466 J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010

Relating Isomerization Activity to W Surface Density
A R T I C L E S
/ZrO
Table 2. Brønsted and Lewis Acidity of Supported WO
Catalysts Determined by Pyridine FTIR
x
2
Brønsted Lewis Brønsted Lewis
cat-
B:L
ratio
1
-1
Catalyst Sample
[µmol·g
]
[sites·W
]
ZrOH (0, 973)
-
29.4
-
-
-
2
3
5
6
8
1
.5-WZrOH (7.0, 973)
.6-WZrOH (13.2, 973)
.2-WZrOH (18.5, 973)
.0-WZrOH (21.7, 973)
.5-WZrOH (30.0, 973)
1.0-WZrOH (32.4, 973)
10.3
20.1
23.7
23.7
24.7
28.8
54.7 0.035 0.183 0.19
68.0 0.035 0.117 0.29
64.2 0.027 0.080 0.34
69.5 0.025 0.073 0.34
72.9 0.019 0.056 0.34
52.2 0.019 0.033 0.58
a
Spent 5.2-WZrOH (18.5, 973)
12.3
61.0 0.016 0.080 0.20
4
8
.4-WZrOH (21.7, 773)
.0-WZrOH (30.0, 773)
29.5 168.3 0.031 0.175 0.18
30.3 147.7 0.023 0.113 0.21
a
Postreaction sample collected after 10 h. Reaction conditions: 523
K, 1.04 atm, 1% nC
5
in He. Overall nC
5
conversion <3%.
a very small fraction of the ZrO
2
particles are crystalline in
nature as revealed by the occasional localized lattice fringes.
HAADF-STEM images were taken from both the amorphous
Figure 6. Representative (a) TEM BF, (b) HRTEM, and (c, d) HAADF-
STEM images of the supported 8.5-WZrOH (30.0, 973) catalyst. Blue
circles: surface monotungstate species; Green circles: surface polytungstate
(Figure 7c) and crystalline (Figure 7d) regions of the ZrO
support, respectively. Although the speckle contrast from the
amorphous ZrO made it slightly more difficult to definitively
locate the W atom positions, the WO species were seen in
Figure 7c to be highly dispersed on the amorphous ZrO support
surface as isolated WO units and probably as some polytung-
state species. In contrast, surface polytungstate was the dominant
species found on the crystalline portion of ZrO (Figure 7d),
with occasional sub-nm Zr-WO clusters also found.
and also in the present study, the W
2
x
species; Red circles: sub-nm Zr-WO clusters.
2
x
2
x
2
x
1
9,32
As noted previously
atoms have a tendency to sit directly above Zr sites on the ZrO
surface. Thus, the lack of structural order in the amorphous ZrO
2
2
support reduced the ability of adjacent W atoms to form a
4,19,32
x
polytungstate network or Zr-WO clusters. Previous reports
have suggested that Zr-WO clusters and polytungstate species
x
are much more catalytically active than monotungstate species.
Therefore, the predominance of the highly dispersed mono-
tungstate WO species on the amorphous ZrO surface could
x 2
be the underlying reason for the extremely low catalytic activity
exhibited by this sample. A more crystallized sample with a
similar W surface density was found to be more active as a
x
result of the formation of Zr-WO clusters and/or polytungstates.
3
.2.5. Surface Acidity As a Function of Tungsten Oxide
Figure 7. Representative (a) TEM BF, (b) HRTEM, and (c, d) HAADF-
STEM images of the supported 4.4-WZrOH (21.7, 773) catalyst. Blue
circles: surface monotungstate species; Green circles: surface polytungstate
Surface Density. Pyridine FTIR experiments were performed under
ultrahigh vacuum conditions and the acquired results represented
surface acidity, i.e., the acidic state of the catalyst assessed under
nonreaction conditions. The total acidity of the WZrOH(773) and
WZrOH(973) series are presented in Table 2.
x
species; Red circles: sub-nm Zr-WO clusters.
imaging (Figure 6b, c, and d). The BSE-SEM image (Figure
b) shows a higher volume fraction of bulk WO crystals in
this catalyst sample as compared with 5.2-WZrOH (18.5, 973)
catalyst. The average size of the WO particles also increased
as the nominal tungsten surface density increased from 5.2 to
5
3
The surface Brønsted acidity of the WZrOH(973) catalysts,
on a per-gram-catalyst basis, increased with Fsurf, correlating to
x
the increasing WO content. However, on a per-W-atom basis,
3
-
1
the number of B sites was constant at ∼0.035 sites ·W below
-2
8
.5 W·nm . The significant intensity increase of the m-WO
XRD peaks indicated an increased amount of WO crystals.
.2.4. Catalyst Calcined at Lower Temperature with WO
Loading Close to Monolayer Coverage. One of the WO /ZrO
3
-
2 4,8,18,27
ML coverage (Fsurf ≈ 4.5 W ·nm )
gradually above ML coverage; this apparent decrease in acidity
resulted from the presence of WO crystals above ML coverage.
and decreased
3
3
x
3
x
2
The B sites to which the pyridine molecule chemisorbed were
catalysts calcined at a lower temperature of 773 K was also
characterized for comparative purposes. The major morphology
found in the 4.4-WZrOH (21.7, 773) catalyst was aggregates
likely hydroxyl groups (W-O-W-OH or Zr-O-W-OH)
associated with W6 and W atoms
+
5+
33,34
depending on the
of small (5-15 nm) ZrO
the catalysts calcined at 973 K, in that the ZrO
amorphous, according to electron diffraction (Figure 7a) and
X-ray diffraction (Figure 2). The ZrO support structure is more
clearly shown in the HRTEM image in Figure 7b, where only
2
particles. However, they differed from
2
was largely
(32) Zhou, W.; Ross-Medgaarden, E.; Wachs, I. E.; Kiely, C. J. Microscopy
and Microanalysis 2008, 14, 1350–1351.
(
(
33) Di Gregorio, F.; Keller, V. J. Catal. 2004, 225, 45–55.
34) Kuba, S.; Heydorn, P. C.; Grasselli, R. K.; Gates, B. C.; Che, M.;
Kn o¨ zinger, H. Phys. Chem. Chem. Phys. 2001, 3, 146.
2
J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010 13467

A R T I C L E S
Soultanidis et al.
Figure 8. (a) Steady-state nC
5
consumption turnover rates (TOR) and (b) steady-state nC
5
isomerization turnover rates as a function of tungsten surface
cd oe nn vs iet ry s. i oS na m< p3 l%e .series include ([) WZrOH(973), (9) WZrOH(773), and (×) WZrO
2 5 5
(773). Reaction conditions: 523 K, 1.04 atm, 1% nC in He. Overall nC
3
5
12,13,15,16
individual cluster size. In fact, the formation of B sites has
been directly correlated to the existence of surface polytungstate
species.
reports.
Whereas B site content decreased slowly with
-2
increasing W surface density above 3.6 W ·nm , L site content
decreased more rapidly. Considered in a different way, per W
atom, the number of acidic hydroxyl groups decreases more
than the number of open Zr sites does, as W content increases.
The B:L ratios were measured to be less than 1 for all samples
5
,8,12-18
Calcination at a lower temperature (773 K)
slightly increased the number of B acid sites for comparable
surface densities in the WZrOH catalysts.
Our observations of B site density compared favorably with
results reported by others. Scheithauer et al.
acid site amount and strength using low-temperature CO-IR
spectroscopy. Based on the carbonyl stretching frequency, B
site density reached a maximum above ∼6.0 W ·nm (which
we calculated using the SSA and the WO loading values
reported for each sample) and remained unchanged by further
increasing Fsurf. Baertsch et al. studied the acidity of WO
ZrO catalysts using NH -IR spectroscopy combined with
1
2,13
-2
studied the B
with the ratio remaining identical between 5.2 and 8.5 W ·nm .
For comparison, Santiesteban et al. reported a “strong” B:L
8
ratio of 1:1 for coprecipitated WO
x
/ZrO
et al. reported an in situ B:L ratio of 1.75:1 for impregnated
WO /ZrO materials. The difference in B:L ratio could be due
2
catalysts and Baertsch
-2
15
3
x
2
(1) to the difference in catalyst surface environments under
reaction or nonreaction conditions; (2) to the difference in
material preparation, in which coprecipitated samples have a
1
6
x
/
2
3
8
temperature-programmed desorption TPD measurements and
concluded that the maximum B site density per W atom was
higher B site density than impregnated ones; (3) to the stronger
basicity of 2,6-di-tert-butyl-Py compared to Py in the gas
-2
36,37
found at an intermediate Fsurf of ∼5.5 W ·nm , which differs
phase;
and (4) to the in situ transformation of L to B sites
16
from the maximum o-xylene isomerization activity observed at
x 2
as shown by Baertsch et al. for WO /ZrO . In addition, during
-2
15
F
surf ≈ 10.0 W ·nm . The same group also investigated the
in situ acidity of WO /ZrO by studying its kinetics during
-butanol dehydration at reaction temperature of 373 K in the
presence of pyridine (Py) or 2,6-di-tert-butyl-Py. The maximum
in situ characterization the reactant molecule is cofed with the
probe molecule, which leads to a competitive chemisorption
mechanism that may underestimate the exact number of B and/
or L sites.
3.2.6. n-Pentane Isomerization Catalytic Activity. 3.2.6.1. De-
pendence on Surface Density. Steady-state turnover rates (TOR)
for all materials are presented in Figure 8. All the WZrOH
catalysts demonstrated cracking, isomerization, and oligomer-
x
2
2
-1
site density was estimated to be ∼0.04 B sites ·W and ∼0.04
L sites·W- at Fsurf ≈ 6.0 W·nm . These results agreed well
1
-2
-1
with the in situ B site density (∼0.033 sites ·W ) measured by
Santiesteban et al. during nC
5
isomerization in the presence of
8
-2
Py or 2,6-di-tert-butyl-Py but were much lower than those
ization activity with a TOR maxima at Fsurf ≈ 5.2 W·nm . In
-1
reported by Baertsch et al. using NH
3
-IR, i.e., 0.2 B sites·W
2
contrast, even though the WZrO samples calcined at 773 K
-
1
-2 16
4
and ∼0.08 L sites ·W at 5.5 W ·nm . The maximum
were reported to possess mild activity for methanol dehydration,
2
-butanol dehydration rate was similarly observed at Fsurf ≈ 9.0
they were found to be almost inactive for nC
due to the more demanding acid nature of the nC
reaction. Similarly, bulk WO
methanol dehydration were also found to be inactive for nC
isomerization.
5
isomerization
-2
W·nm , which again did not correspond to the maximum B
5
isomerization
site density.
3
crystals that are active for
4
The Lewis acidity of WZrOH(973), normalized per gram
catalyst, increased significantly with Fsurf up to 8.5 W ·nm and
dropped by 37% at 11.0 W ·nm . This follows a pattern similar
to that for the specific surface area of the catalysts (Table 1).
This can be well explained by the fact that the sintering of the
5
-2
-2
For the WZrOH(973) series, both the overall activity (Figure
8a) and isomerization activity (Figure 8b) reached a maximum
-
2
at 5.2 W ·nm , correlating well to the large population of Zr-
parent ZrO
x
(OH)4-2x support structure was hindered by a high
WO clusters in this sample as discussed in previous sections.
x
+
WO
3
content, yielding more coordinatively unsaturated Zr4
The stronger dependence of isomerization rates on surface
cations that are accessible as L acid surface sites. A lower
calcination temperature also led to higher L acid content due
(35) Cort e´ s-J a´ come, M. A.; Angeles-Chavez, C.; L o´ pez-Salinas, E.; Na-
5
,27
varrete, J.; Toribio, P.; Toledo, J. A. Appl. Catal. A: General 2007,
to reduced sintering of the support.
When normalized to the WO surface density, L acid content
decreased with increasing Fsurf, which is consistent with previous
3
18, 178–189.
x
(
36) Hosmane, R.; Liebman, J. Struct. Chem. 2009, 20, 693–697.
(37) Hunter, E. P. L.; Lias, S. G. J. Phys. Chem. Ref. Data 1998, 27, 413–656.
1
3468 J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010

Relating Isomerization Activity to W Surface Density
A R T I C L E S
Figure 9. Steady-state product distribution of (a) WZrOH(973) and (b) WZrOH(773). Reaction conditions: 523 K, 1.04 atm, 1% nC
conversion <3%.
5 5
in He. Overall nC
density suggested that these clusters favored the isomerization
of n-pentane over other acid-catalyzed pathways.
The WZrOH catalysts calcined at 773 K were found to be
isomerization could also occur through a bimolecular reaction
pathway. C
6
+
can be produced from the combination of methyl
41
cation (CH
3
5 5
, generated from D3 ꢀ-scission of nC ) with nC .
less active (Figure 8a) and less selective for isopentane (iC
formation (Figure 8b). While calcination temperature was not
found to impact the activity of WO /ZrO for methanol
dehydration, 2-butanol dehydration, and o-xylene isomeriza-
5
)
It can also come from a C10 surface intermediate formed from
the coupling of pentane and pentene molecules, which then
x
2
cracks into C
for SO /ZrO .
4 2
6
’s and C
4
’s, similar to what has been proposed
The same C10 intermediate could undergo
skeletal isomerization and cracking to generate iC . nC isomer-
ization is likely to occur via a bimolecular reaction mechanism
for WO /ZrO , but this point has not been established in
4
15
46-49
5
tion, it affected WO
For example, Scheithauer et al. reported that WO
isomerization than
x
/ZrO
2
catalytic activity for other reactions.
5
5
x
/ZrO calcined
2
at 923 K was ∼5 times more active for nC
5
x
2
12
those calcined at 1098 K. L o´ pez et al. reported up to ∼5 times
literature yet. We are currently studying this in more detail by
analyzing the effect of adding alkenes as cofeed.
higher acetic acid esterification activity for samples calcined at
3
8
2
1
073 K in the range 673-1173 K. With our results, such
Selectivity to iC
to a maximum of 46-48% for Fsurf between 5.2 and 8.5 W ·nm
5
increased from 38% for Fsurf ) 2.5 W ·nm-
-2
observations show a clear calcination temperature effect, for
which the optimum temperature promotes the formation of
for the WZrOH catalysts calcined at 973 K (Figure 9a), where
4
,19
highly active Zr-WO
The selectivity profiles at different surface densities indicate
cracking (C -C ), isomerization (iC ), and oligomerization
>C ) hydrocarbon products (Figure 9). Carbon mass balances
x
clusters.
the number density of Zr-WO clusters also reached a maximum.
x
The isobutane (iC
at the surface densities studied. Propane/propylene (C ) percent-
4
) percentage remained constant (∼27%)
1
4
5
3
(
6
ages were relatively low at all surface densities except at 2.5
-
2
closed at 95-98%, with the remaining carbon mass in the forms
of coke precursor (which are more volatile and therefore more
easily removable) and hard coke (which are harder to remove,
after being formed from further dehydrogenation of the coke
, where their relative concentration increased
and 3.6 W ·nm
to 14-20%. C was detected in low amounts at all surface
6
-2
densities except at 11.0 W ·nm , where its percentage increased
to 14%. Small amounts of C and C were detected, with
1
2
3
9
precursors). No H
released from cracking reactions was consumed in situ.
A monomolecular mechanism was previously proposed for
nC isomerization over WO /ZrO catalysts under similar
reaction conditions. The reaction is initiated via the formation
2
was detected, indicating that any H
2
n-butane (nC ) detected in trace amounts (not shown).
4
The WZrOH samples calcined at 773 K were found to greatly
favor cracking over isomerization of nC
remaining below 20% between 2.2 and 8.0 W ·nm (Figure
b). iC and C were the main cracking byproducts with their
5 5
, with iC selectivities
-2
5
x
2
1
1
9
4
3
+
40
of a carbenium (C
ization or ꢀ-scission to produce iC
similar to what has been reported for zeolite catalysts. The
detection of ethane and ethylene (C ) and traces of methane
) in this study was evidence for a monomolecular cracking
5
) cation followed by its skeletal isomer-
relative concentrations remaining unchanged (∼30% and ∼25%
4
1
5
or C
1
4
-C respectively,
respectively) at all surface densities. C
detected at all surface densities.
6
products were also
2
WZrOH samples calcined at 973 K were at least twice as
good as samples calcined at 773 K as nC isomerization
(C
1
4
2
5
mechanism.
Hexane and hexenes (C
catalysts. We suggest that the surface monotungstate speciess
found in large amounts in the 773-K-calcined WZrOH samples
6
) and traces of higher molecular-
weight hydrocarbons were detected in this study also, in
accordance with similar studies performed on paraffin cracking
(
Figure 7) but only in small amounts in the 973-K-calcined
samples near or above WO monolayer coverage promotes
x
41-45
9
46,47
overzeolites,
The detection of species larger than pentane indicates intermo-
lecular transformations were occurring, suggesting that nC
x 2 4 2
WO /ZrO , andsulfatedzirconia(SO /ZrO ).
(
(
43) Jolly, S.; Saussey, J.; Bettahar, M. M.; Lavalley, J. C.; Benazzi, E.
5
Appl. Catal. A: General 1997, 156, 71–96.
44) Chu, H. Y.; Rosynek, M. P.; Lunsford, J. H. J. Catal. 1998, 178, 352–362.
(
38) L o´ pez, D. E.; Suwannakarn, K.; Bruce, D. A.; Goodwin, J. G., Jr. J.
(45) Corma, A.; Miguel, P. J.; Orchilles, A. V. J. Catal. 1994, 145, 58–64.
(46) Lohitharn, N.; Lotero, E.; Goodwin, J. G., Jr. J. Catal. 2006, 241,
328–341.
Catal. 2007, 247, 43–50.
(
39) Wang, B.; Manos, G. J. Catal. 2007, 250, 121–127.
(40) Boronat, M.; Viruela, P.; Corma, A. J. Phys. Chem. 1996, 100, 16514–
(47) Tabora, J. E.; Davis, R. J. J. Am. Chem. Soc. 1996, 118, 12240–12241.
(48) Kuba, S.; Heydorn, P. C.; Grasselli, R. K.; Gates, B. C.; Che, M.;
Kn o¨ zinger, H. Phys. Chem. Chem. Phys. 2001, 146–154.
(49) Kuba, S.; Lukinskas, P.; Grasselli, R. K.; Gates, B. C.; Kn o¨ zinger, H.
J. Catal. 2003, 216, 353–361.
1
6521.
41) Buchanan, J. S.; Santiesteban, J. G.; Haag, W. O. J. Catal. 1996, 158,
79–287.
42) Jentoft, F. C.; Gates, B. C. Top. Catal. 1997, 4, 1–13.
(
(
2
J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010 13469

A R T I C L E S
Soultanidis et al.
Figure 10. (a) nC
5
transient turnover rates (TOR), (b) transient isomerization turnover rates (TORiC5) and product distributions for (c) 5.2-WZrOH (18.5,
9
5 5
73), (d) 2.5-WZrOH (7.0, 973), and (e) 11.0-WZrOH (32.4, 973). Reaction conditions: 523 K, 1.04 atm, 1% nC in He. Overall nC conversion <3%.
monomolecular cracking reaction (Figures 4 and 6). We further
suggest that the 0.8-1 nm Zr-WO clusters found in the 973-
K-calcined samples, with the highest presence near W mono-
layer coverage, are responsible for bimolecular isomerization
activity.
and a small decrease in strong B acidity were noticed in this
range, with moderate sites remaining unchanged. High Fsurf
decreased moderate and strong B acidity, which corresponded
to higher molecular-weight reaction products (Figure 9a). The
total B site content of WZrOH(773) material was in line with
that of WZrOH(973), on a W surface density basis. No pattern
could be discerned from the different B acid strength concentra-
tions (Figure S1a,b) to explain how WZrOH(773) were at least
two times less active. As noted earlier, the 773-K-calcined
x
3
.2.6.2. Correlation to Surface Acidity. The concentrations
of weak, moderate, and strong B sites were comparable for the
WZrOH(973) series up to 5.2 W ·nm- (Figure S1a in the
Supporting Information). A minor increase in weak B acidity
2
1
3470 J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010

Relating Isomerization Activity to W Surface Density
A R T I C L E S
samples contained more L sites per W than 973 K samples
more specifically, the bimolecular mechanism model for C
isomerization.
5
-2
-2
(
Table 2, cf. 3.6 and 4.0 W ·nm , and 8.5 and 8.0 W ·nm ).
No pattern could be discerned from the different L acid strength
concentrations either (Figure S1c,d), to explain the differences
in catalytic activity.
We speculate that the 5.2-WZrOH(18.5, 973) catalyst, which
contains comparable amounts of polytungstate and Zr-WO
clusters, catalyzes the cracking and isomerization of nC
Whereas the polytungstates deactivate immediately with time
on stream, the Zr-WO clusters undergo an activation process
through partial reduction
These sites catalyze the bimolecular pathway, leading to the
coformation of the observed C ’s, and deactivate with time as
x
5
.
Recognizing that coke deposition occurred during the reaction
that could impact the acid site amount and strengths, we assessed
the surface acidity of 5.2-WZrOH (18.5, 973) after 10 h under
reaction conditions. Pyridine adsorption FTIR results indicated
x
4
8,52
(
5
) to form the iC -forming B sites.
-1
that the total B acidity decreased from 0.028 to 0.016 sites ·W ;
6
the amounts of strong, medium, and weak acid sites proportion-
the result of C10 deposition or oligomerization.
ally decreased; and the total L acidity remained the same at
For 2.5-WZrOH and 11.0-WZrOH, the TOR did not
0
.08 sites ·W- (Figure S2). These observations pointed to the
1
increase to the same extent, almost reaching a maximum of 2.0
direct participation of B sites during nC
in agreement with previous reports,
sites did not participate as active sites.
5
,15
isomerization reaction,
and suggested that L
-7 -1
×
10 s . This may be due to the smaller populations of
8
polytungstates and Zr-WO
.2-WZrOH; the 2.5-WZrOH sample had relatively more
monotungstates, and the 11.0-WZrOH had more WO crystals.
x
clusters compared to that of
5
That the measured B and L site concentrations did not
correlate with the observed catalytic activity trends suggested
that the catalytically active sites are most likely to form in situ
and that the surface tungstate species sites likely generate these
active sites during the reaction. This latter point has been
considered by several research groups in the context of an
3
Whereas the 11.0-WZrOH reached maximum TOR in the same
time as 5.2-WZrOH, the lower surface density sample took
twice as long to reach maximum TOR (Figure 10a). This
observation suggested that the Zr-WO
become activated, perhaps due to their being located at the
interstices of adjacent small ZrO particles and defect sites
Figure 3d).
x
clusters took longer to
observed induction period during flow reactor studies. For WO
ZrO , the induction period was proposed to result from the in
situ generation of B sites from L sites.
material SO /ZrO , the induction period found in n-butane
skeletal isomerization was attributed to the formation of an
x
/
2
2
(
5
,14-16
For the related
4
2
4. Conclusions
46-49
oligomeric intermediate via a bimolecular reaction pathway.
.2.6.3. Time-Dependent Catalytic Behavior. Reaction rates
and product concentrations were quantified and monitored over
0 h for WO /ZrO samples at different surface densities
2.5-WZrOH (7.0, 973), 5.2-WZrOH (18.5, 973), and
The pentane isomerization activity of WO /ZrO is strongly
x
2
3
affected by the nature of the support, calcination temperature,
and tungsten oxide surface density. WZrOH samples demon-
strated a volcano-shape dependence on tungsten surface density
1
x
2
-
2
(
with maximum activity at 5.2 W ·nm , above ML coverage
1
1.0-WZrOH (32.4, 973)). In agreement with other re-
and at the onset of the WO crystallization, in contrast to model
3
1
2,49-51
ports,
an induction period was observed in all cases,
WZrO that were inactive. The calcination temperature of 973
2
during which the TOR reached a maximum (Figure 10a,b).
K, not 773 K, favored the formation of sub-nm Zr-WO clusters
x
5
1
1
.2-WZrOH (18.5, 973) showed the highest TOR at ∼5.5 ×
and in the overall activity of WZrOH, without promoting their
-
-
7
7
-1
-1
0
0
s
s
at 1 h, which decreased to a stable value of ∼2.0 ×
surface acidic properties. The induction period during catalysis
is critical for the activation of the clusters, which results in the
increased isomerization activity and selectivity seen most
pronouncedly at intermediate Fsurf. A bimolecular isomerization
mechanism, which plays a significant role and requires further
investigation, appears to be promoted by these in situ activated
x
after 7 h. The rapid drop in TOR corresponded with
the rapid decrease in iC
presumably due to the deactivation of the most active acid sites
4 3
and C formation (Figure 10c),
5
1
which favored cracking.
monomolecular mechanism as indicated by the absence of C
in the product stream. On the other hand iC can be formed via
both a monomolecular and a bimolecular mechanism. The
initially high concentration of iC that declines similarly to that
of C suggests they share a common monomolecular cracking
pathway. The fact that the C concentration drops to the
minimum faster (within ∼1 h) in contrast to iC that gradually
decreases within ∼5 h implies a secondary reaction pathway
contributing to the formation of iC . This bimolecular mecha-
nism generates other iC , reaching a steady state concentration
3
C is exclusively produced via a
7
4
Zr-WO sites.
Acknowledgment. We acknowledge the National Science
Foundation (Nanoscale Interdisciplinary Research Team (NIRT),
CBET-0609018), SABIC Americas, and 3M (NTF Award) for
funding.
4
3
3
4
Supporting Information Available: Catalyst particle selection;
Catalyst characterization; Catalytic studies; Table S1 (Selection
of catalyst particle size); Table S2 (Coke content determined
by TGA); Figure S1 (Brønsted and Lewis acid site strength as
a function of WO surface density expressed as sites per W
x
atom); Figure S2 (Amounts of Brønsted and Lewis acid sites
of varying acid strengths for 5.2-WZrOH (18.5, 973) before
5
and after the running nC isomerization. This material is
available free of charge via the Internet at http://pubs.acs.org.
4
4
within the same time the concentration of C
6
maximizes.
During the 10-h period, TORiC5 dropped to ∼1.0 × 10 s-1
-
7
5
(Figure 10b) but the iC selectivity increased from 14% to 46%
for the most active 5.2-WZrOH(18.5, 973) catalyst (Figure
0c), indicating that the acid sites responsible for isomerization
were deactivated to a lesser extent. C followed the same trend
as iC (Figure 10c), supporting the idea of a common reaction
pathway (via a C10 intermediate) for these two products and,
1
6
5
JA105519Y
(
(
50) De Rossi, S.; Ferraris, G.; Valigi, M.; Gazzoli, D. Appl. Catal. A:
General 2002, 231, 173–184.
51) Kuba, S.; Kn o¨ zinger, H. J. Raman Spectrosc. 2002, 33, 325–332.
(52) Di Gregorio, F.; Keller, N.; Keller, V. J. Catal. 2008, 256, 159–171.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 38, 2010 13471