Synergy between tungsten and palladium supported on titania for the catalytic total oxidation of propane

Article abstract of DOI:10.1016/j.jcat.2011.09.019

Titania-supported palladium catalysts modified by tungsten have been tested for the total oxidation of propane. The addition of tungsten significantly enhanced the catalytic activity. Highly active catalysts were prepared containing a low loading of 0.5 wt.% palladium, and activity increased as the tungsten loading was increased up to 6 wt.%. Catalysts were characterised using a variety of techniques, including powder X-ray diffraction, laser Raman spectroscopy, X-ray photoelectron spectroscopy, temperature-programmed reduction and aberration-corrected scanning transmission electron microscopy. Highly dispersed palladium nanoparticles were present on the catalyst with and without the addition of WO_x. However, the addition of WO_x slightly increases the average palladium particle size, and there was some evidence for the Pd forming epitaxial islands on the support in the tungsten-doped samples. Surface analysis identified a combination of Pd⁰ and Pd²⁺ on a Pd/TiO₂ catalyst, whereas all of the Pd loading was found in the form of Pd²⁺ with the addition of tungsten into the catalysts. At low tungsten loadings, isolated monotungstate and some polytungstate species were highly dispersed over the titania support. The concentration of polytungstate species increased as the loading was increased, and it was also promoted by the presence of palladium. The coverage of the highly dispersed tungstate species over the titania also increased as the tungsten loading increased. Some tungstate species were also found to be associated with the palladium oxide particles, and there was an enrichment of oxidised tungsten species at the peripheral interface of the palladium oxide nanoparticles and the titania. Sub-ambient temperature-programmed reduction experiments identified an increased concentration of highly reactive species on catalysts with palladium and tungsten present together, and we propose that the new WO _x-decorated interface between PdO_x and TiO₂ particles may be responsible for the enhanced catalytic activity in the co-impregnated catalysts.

Full text of DOI:10.1016/j.jcat.2011.09.019

Journal of Catalysis 285 (2012) 103–114
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Synergy between tungsten and palladium supported on titania for the catalytic
total oxidation of propane
a
b
c,
⇑
d
a
b
Marie N. Taylor , Wu Zhou , Tomas Garcia , Benjamin Solsona , Albert F. Carley , Christopher J. Kiely ,
Stuart H. Taylor
a
a,
⇑
Cardiff Catalysis Institute, School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff CF10 3AT, UK
Department of Materials Science and Engineering, Lehigh University, 5 East Packer Avenue, Bethlehem, PA 18015-3195, USA
Instituto de Carboquímica (ICB-CSIC), C/Miguel Luesma, 50018 Zaragoza, Spain
b
c
d
Departament d’Enginyeria Química, Universitat de València, C/Dr. Moliner 50, 46100 Burjassot, Valencia, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 7 June 2011
Revised 13 September 2011
Accepted 16 September 2011
Available online 20 October 2011
Titania-supported palladium catalysts modified by tungsten have been tested for the total oxidation of
propane. The addition of tungsten significantly enhanced the catalytic activity. Highly active catalysts
were prepared containing a low loading of 0.5 wt.% palladium, and activity increased as the tungsten
loading was increased up to 6 wt.%. Catalysts were characterised using a variety of techniques, including
powder X-ray diffraction, laser Raman spectroscopy, X-ray photoelectron spectroscopy, temperature-
programmed reduction and aberration-corrected scanning transmission electron microscopy. Highly dis-
Keywords:
Catalytic oxidation
Propane
Tungsten
Palladium
Palladium oxide
Titania
persed palladium nanoparticles were present on the catalyst with and without the addition of WO
ever, the addition of WO slightly increases the average palladium particle size, and there was some
evidence for the Pd forming epitaxial islands on the support in the tungsten-doped samples. Surface anal-
x
. How-
x
0
2+
2
ysis identified a combination of Pd and Pd on a Pd/TiO catalyst, whereas all of the Pd loading was
2
+
found in the form of Pd with the addition of tungsten into the catalysts. At low tungsten loadings, iso-
lated monotungstate and some polytungstate species were highly dispersed over the titania support. The
concentration of polytungstate species increased as the loading was increased, and it was also promoted
by the presence of palladium. The coverage of the highly dispersed tungstate species over the titania also
increased as the tungsten loading increased. Some tungstate species were also found to be associated
with the palladium oxide particles, and there was an enrichment of oxidised tungsten species at the
peripheral interface of the palladium oxide nanoparticles and the titania. Sub-ambient temperature–
programmed reduction experiments identified an increased concentration of highly reactive species on
Aberration-corrected electron microscopy
catalysts with palladium and tungsten present together, and we propose that the new WO
interface between PdO and TiO particles may be responsible for the enhanced catalytic activity in the
co-impregnated catalysts.
x
-decorated
x
2
Ó 2011 Elsevier Inc. All rights reserved.
1
. Introduction
Catalytic total oxidation has been widely used for the abatement
Many different catalysts, as bulk metal oxides or perovskites
2,3], have been studied for the total oxidation of VOCs. However,
[
it is generally accepted that supported noble metal catalysts, con-
taining palladium or platinum, are the most active catalysts for the
combustion of hydrocarbons [4,5]. Moreover, palladium-supported
catalysts have been demonstrated to be more active for the total
oxidation of short chain alkanes [6]. The behaviour of these sup-
ported palladium catalysts are strongly affected by the nature of
the support. Although c-Al O has been extensively used as a sup-
2 3
port, mainly due to its low manufacturing cost [7], various studies
[8–10] have shown that the use of other metal oxide supports, such
of dilute volatile organic compounds (VOCs), which are present in a
wide variety of gaseous emissions, such as those arising from petro-
chemical processes, manufacturing plants and the treatment of solid
and liquid wastes. Catalytic oxidation offers significant advantages
over alternative VOC abatement technologies. For example, the
operating temperature is lower than for thermal combustion, and
hence, operating costs are lower as less supplementary fuel is re-
quired to achieve VOC removal. Furthermore, catalytic oxidation
has the ability to treat VOCs in very dilute effluent streams [1].
as TiO
for the oxidation of hydrocarbons. In addition, other positive ef-
fects have also been reported [11]; for example, TiO is known to
be more resistant to sulphur poisoning than Al . It is generally
2 2 2 2
, ZrO , SiO and SnO , has resulted in more active catalysts
⇑
Corresponding authors.
E-mail addresses: tomas@icb.csic.es (T. Garcia), taylorsh@cardiff.ac.uk (S.H. Taylor).
2
2 3
O
0
021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.09.019

1
04
M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
accepted that the catalytic total oxidation of hydrocarbons over
palladium-supported catalysts proceeds by the reduction and oxi-
dation cycle of palladium with the oxygen species originating from
PdO species [12–14]. Thus, it is likely that the performance of the
catalyst is related to the nature of the support through the disper-
sion of Pd and the Pd-support interaction [15].
A Renishaw system 1000 dispersive laser Raman microscope
was used for recording Raman spectra. The excitation source used
was an argon ion laser (514.5 nm) operated at a power of 20 mW.
The laser was focused on powdered samples placed on a micro-
scope slide to produce a spot size ca. 3 lm in diameter. A backscat-
tering geometry with an angle of 180° between illuminating and
The activity of supported catalysts may be modified by the pres-
ence of a second active component. Modifiers are generally added
to promote activity and enhance resistance to deactivation. Vana-
dium [8,16], tungsten [17,18], niobium [19], cobalt [20], cerium
collected radiation was used for recording data.
Temperature-programmed reduction (TPR) was performed
using a micromeritics Autochem 2910 apparatus with a thermal
conductivity detector. The reducing gas used was 10% H
2
in argon
À1
[
21] and lanthanum [22] have all been used, and they are known
with a total flow rate of 50 ml min . The temperature range
to promote the activity of supported metal catalysts for the oxida-
tion of various VOCs. The positive promoting role of the different
components mentioned above, for the oxidation of short chain
alkanes, has generally been attributed to a range of different fac-
tors, such as controlling the noble metal particle size, catalyst
reducibility and metal oxidation state.
explored was from room temperature to 900 °C with a heating rate
of 10 °C min . Sub-ambient TPR was carried out on 100 mg of cat-
À1
alyst, which was pretreated by heating from room temperature to
À1
120 °C at a ramp rate of 20 °C min
in flowing 10% O
2
/Ar
/Ar
À1
(50 ml min ). The sample was maintained at 120 °C under O
2
for 1 h, before being cooled under Ar to À100 °C. TPR was per-
In this work, we have extended previous studies of high-activity
titania-supported palladium catalysts modified by vanadium [8]
and niobium [23], by probing the effect of modifying the supported
palladium catalyst with tungsten. It has been reported previously
formed from À100 °C to 120 °C under 10% H /Ar at a flow of
2
À1
À1
50 ml min at a ramp rate of 20 °C min
.
Palladium particle sizes for the unpromoted catalysts were
determined by pulsed CO chemisorption at 35 °C using an Ar flow
6
+
À1
that doping W cations into the crystalline structure of the TiO
2
of 20 ml min and pulses of 0.2 ml of 10% CO in Ar. Prior to CO up-
support, prior to palladium impregnation, has a positive effect on
the activity of Pd catalysts for benzene and ethyl acetate total oxi-
dation [18]. However, we report here the first data available on
take determination, all samples were treated under flowing hydro-
À1
À1
gen (50 ml min ) at 400 °C and then flushed by Ar (20 ml min
)
for 60 min. In order to calculate the Pd particle size, an adsorption
stoichiometry of Pd/CO = 1 was assumed, as this value has been
widely used, and we are most interested in determining the trend
of dispersion. Spherical particles were also assumed.
X-ray photoelectron spectroscopy (XPS) measurements were
taken using a Kratos Axis Ultra DLD spectrometer using monochro-
2
co-impregnated tungsten-containing Pd/TiO catalysts for the total
oxidation of a short chain alkane. We have focused on probing the
activity for the complete oxidation of propane, because short chain
alkanes are emitted from a wide range of sources and they are
recognised as some of the most difficult VOCs to oxidise [12]. In
particular, we have explored the role of tungsten and palladium
loading in the performance of these catalysts and develop a struc-
tural catalyst model to rationalise catalyst performance for pro-
pane total oxidation.
matised AlKa radiation, and analyser pass energies of 160 eV (sur-
vey scans) or 40 eV (detailed scans). Binding energies are
referenced to the C(1s) peak from adventitious carbonaceous con-
tamination, assumed to have a binding energy of 284.7 eV.
Ammonia temperature-programmed desorption (TPD) profiles
were obtained using a fixed bed flow system equipped with a ther-
mal conductivity detector. One-hundred milligrams of each cata-
lyst was pretreated in flowing Ar at 500 °C for 6 h and then
cooled down to 125 °C under vacuum. The samples were treated
2
. Experimental
2.1. Catalyst preparation
for 4 h with a flow of 1.3 kPa NH
3
/Ar, and the weakly adsorbed
Catalysts with different chemical compositions of Pd
x
/W
y
/TiO
2
NH was removed by evacuation at 125 °C for 1 h. The NH
3
3
-TPD
À1
were prepared by a wet impregnation technique. A known amount
of palladium(II) chloride (Aldrich, 99%) was dissolved in a mini-
mum amount of hot de-ionised water (ꢀ10 ml/g). The solution
was then heated to 80 °C, with continuous stirring, followed by
the addition of an appropriate amount of tungstic acid. Titanium
profiles were obtained by heating at 5 °C min up to 600 °C under
À1
a 30 ml min flow of helium.
2 x 2
The microstructures of the supported Pd/TiO and Pd/WO /TiO
catalysts were studied using high-angle annular dark-field
(HAADF) imaging in an aberration-corrected scanning transmis-
sion electron microscope (STEM), which is also known as Z-con-
trast imaging. Samples suitable for STEM imaging were prepared
by dipping a 300-mesh copper TEM grid, coated with a lacy carbon
film, into the dry catalyst powder and then shaking off any loosely
bound residue. STEM–HAADF imaging was performed on a 200 kV
2
À1
dioxide (ꢀ5 g, Degussa P25. SABET = 50 m g ) was added to the
heated solution and stirred continuously at 80 °C to form a paste,
which was dried in an oven at 110 °C for 24 h. The resulting solid
was ground in a pestle and mortar and calcined in static air at
5
50 °C for 6 h.
s
JEOL 2200FS (S)TEM equipped with a CEOS probe C -corrector.
Typically, a ꢀ1 Å (FWHM) coherent electron beam with ꢀ30 pA
2.2. Catalyst characterisation
probe current was used for imaging, and dwell times between
48 ls and 60 ls per pixel were applied. The HAADF images pre-
Catalyst surface areas were determined by multi-point N
2
sented have been low-pass filtered to reduce background noise.
Simultaneous HAADF and STEM–XEDS (X-ray energy dispersive
spectroscopy) spectrum imaging experiments were performed on
the same instrument with a ꢀ100 pA electron probe and 2 s per
pixel dwell time.
adsorption at 77 K, and the data were treated in accordance with
the BET method. Powder X-ray diffraction (XRD) was used to iden-
tify the crystalline phases present in the catalysts. An Enraf Nonius
FR590 sealed tube diffractometer was used, employing CuK
a1
X-rays (40 kV and 30 mA). XRD patterns were calibrated against
a silicon standard and phases were identified by matching experi-
mental patterns to the JCPDS powder diffraction file. The average
crystallite size was determined from the diffraction peak broaden-
ing, relative to a highly crystalline silicon standard, by using the
Scherrer equation [24].
2.3. Catalyst activity
Catalytic activity was measured using a fixed bed laboratory
micro-reactor. For each experiment, 50 mg of powdered catalyst
was placed in a 1/4 in. o.d. stainless steel reactor tube. The reactor

M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
105
feed contained 5000 vppm propane in air with a total flow rate of
there is a synergistic effect between TiO
and tungsten for the catalytic total oxidation of propane.
Fig. 2 compares the catalytic performance of Pd/TiO
2
-supported palladium
À1
5
0 ml min . Catalysts were packed to a constant volume to give a
À1
gas hourly space velocity (GHSV) of 45,000 h . The reactants and
products were analysed by online gas chromatography with a ther-
mal conductivity and flame ionisation detector. The 100–400 °C
temperature range was explored, and the reaction temperature
was measured using a thermocouple placed in the centre of the
catalyst bed. During reaction, the temperature measured within
the catalyst bed was consistent with the temperature of the exter-
nal heating furnace, indicating that there was no significant mea-
sureable exotherm within the catalyst bed. The differences
between the inlet and outlet concentrations were used to calculate
conversion data, and all carbon balances were in the range
2
and
tungsten-promoted Pd/TiO
constant tungsten content of 3 wt.%. The catalytic activity for the
tungsten-promoted Pd/TiO catalysts increased as the palladium
2
, for various palladium loadings, at a
2
content was increased. In the case of the W-free catalysts, the
activity was very similar regardless of the Pd content, indicating
that the activity per palladium content decreases when the Pd
loading increases.
A summary of catalytic activity is presented in Table 1, where
T
10, T50 and T90, the reaction temperatures for alkane conversions
of 10%, 50% and 90%, respectively, are shown. The most active cat-
alyst was 0.5Pd–6.0W/TiO , which had a much higher activity than
the tungsten-free 0.5Pd/TiO catalyst. It is clear that the simple one
step co-impregnation method of introducing palladium and tung-
1
00 ± 10%. Analyses were made at each temperature until steady-
2
state activity was attained, and three analyses were taken and data
averaged. Blank experiments were conducted in an empty reactor
and showed negligible activity over the temperature range used in
this study.
2
2
sten onto the TiO support significantly increases the activity of
the catalysts.
3.2. Catalyst characterisation
3
. Results and discussion
A summary of the physical characteristics of the catalysts pre-
pared is shown in Table 2. The surface area of the titania support
2
À1
3.1. Catalyst performance
was 50 m g , and this was reduced slightly by the addition of pal-
ladium and tungsten as expected. This can be attributed to the fill-
ing of the pores of the titania support as the impregnated species
were added. CO chemisorption was employed to determine the
palladium particle size on the surface of the unpromoted catalysts.
As expected, larger average particle sizes of palladium were formed
when the Pd content was increased (Table 2).
The only reaction product observed for palladium-containing
2 2 2
catalysts was CO . In the case of W/TiO catalysts, CO was the
major reaction product, although propylene was also detected,
but only at temperatures where conversion was relatively low
and selectivity to propylene did not exceed 10%.
Fig. 1 depicts propane conversion as a function of reaction tem-
The XRD patterns obtained from Pd–WO /TiO catalysts with
x
2
perature for tungsten-free and tungsten-containing 0.5Pd/TiO
2
different tungsten and palladium loadings are shown in Figs. 3
and 4. For comparative purposes, Fig. 4 also presents the XRD pat-
tern for Pd/TiO2 catalysts having different Pd loadings. The main
crystalline phases identified correspond to the anatase and rutile
phases of TiO2 with anatase being predominant. The relative pro-
portions of the crystalline titania phases were not modified by
the addition of palladium and tungsten. Furthermore, there was
no modification of titania crystallite size or change of microstrain.
No diffraction peaks corresponding to any tungsten-containing
crystalline phases were identified. This could be attributed to its
presence at very low concentrations and the nature and high dis-
persion of the tungsten species. However, diffraction peaks from
PdO with a characteristic reflection at 33.9° (JCPDS: 85-0624) were
identified in the palladium-containing catalysts. The intensity of
catalysts. The activities of the catalysts containing tungsten were
higher than those for the tungsten-free catalyst. Moreover, the
activity increased as the tungsten loading of the catalyst increased.
Tungsten loadings were limited to 6 wt.% as this is close to the the-
oretical monolayer, and as discussed later, it is known that above
monolayer coverage, less active three-dimensional crystallites of
WO
catalyst greatly increased the catalytic combustion of propane at
all temperatures. In fact, the Pd–W/TiO catalysts all showed com-
plete propane conversion to CO at ca. 375 °C. This enhancement of
activity is not the result of the additive effect of tungsten and pal-
ladium, since the activity of W/TiO catalysts is very low, as they
are almost inactive at 400 °C. Thus, these data clearly show that
3 2
are formed. Clearly, the presence of tungsten in the Pd/TiO
2
2
2
100
1
00
0
1
.5Pd
Pd
8
6
4
2
0
0
0
0
0
8
0
2Pd
0
1
2
3
.5Pd3W
Pd3W
Pd3W
W
0.5Pd
0
0
0
.5Pd1W
.5Pd3W
.5Pd6W
60
4
2
0
0
0
225
250
275
300
325
350
375
400
425
2
50
300
350
400
ReactionTemperature / ºC
Reaction temperature / ºC
Fig. 1. Propane conversion at different temperatures for W-promoted Pd/TiO
catalysts with varying tungsten loading and constant palladium loading. (h) 0.5Pd/
2
Fig. 2. Propane conversion at different temperatures for W-promoted Pd/TiO
catalysts, with constant tungsten loading and varying palladium loading. (h) 0.5Pd/
TiO , (s) 3W/TiO , (N) 0.5Pd–3W/TiO , (.) 1Pd–3W/TiO , (d) 2Pd–3W/TiO
2
TiO
TiO
2
, (j) 0.5Pd–1W/TiO
2
, (+) 0.5Pd–2W/TiO
2
, (N) 0.5Pd–3W/TiO
2
, (Â) 0.5Pd–6W/
2
.
2
2
2
2
2
.

1
06
M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
Table 1
2
Catalytic performance of Pd-, W- and Pd/W–TiO catalysts for propane total oxidation,
expressed in terms of the temperatures required for 10%, 50% and 90% conversion.
¡
Catalyst
T10 (°C)
T50 (°C)
T90 (°C)
0
.5%Pd/6%W/TiO2
0
1
2
3
6
0
0
0
0
1
2
.5Pd/TiO
2
280
280
280
430
425
270
270
265
250
260
255
350
345
350
–
395
395
390
–
Pd/TiO
Pd/TiO
2
2
0.5%Pd/3%W/TiO2
W/TiO
W/TiO
2
2
0.5%Pd/2%W/TiO2
0.5%Pd/1%W/TiO2
TiO2
.5Pd/1W/TiO
.5Pd/2W/TiO
.5Pd/3W/TiO
.5Pd/6W/TiO
.0Pd/3W/TiO
.0Pd/3W/TiO
2
2
2
2
2
2
330
325
315
290
305
295
365
360
355
340
350
345
20
40
60
80
2
θ
the PdO peaks was significantly higher in tungsten-promoted cat-
alysts when compared to those materials without tungsten. The
intensity of the PdO peak also increased with both increasing pal-
ladium and tungsten content (Figs. 3 and 4). The Scherrer equation
was used to estimate the average particle sizes of the PdO nano-
Fig. 3. Powder X-ray diffraction patterns for tungsten-promoted Pd/TiO
2
catalysts
at different W loadings.
2
crystals in some of the tungsten-promoted Pd/TiO catalysts. These
data are reported in Table 2. It was observed that the average pal-
ladium oxide particles in tungsten-free catalysts were smaller than
2.0%Pd/3%W/TiO
2
those in the Pd–WO
x
/TiO
2
catalysts, as values of ꢁ3–6 nm were
2
.0%Pd/TiO2
measured in the former as compared to ꢁ6–10 nm in the latter.
It is worth highlighting that when increasing the tungsten loading
at 0.5 wt.% palladium, the reflection corresponding to the Pd–O
nanocrystals can be much more clearly identified. This is associ-
ated with a lower dispersion of the PdO nanocrystals, since the to-
tal oxidation of the palladium species to PdO is observed for all the
1
.0%Pd/3%W/TiO2
1
.0%Pd/TiO2
0
.5%Pd/3%W/TiO2
Pd–WO
average size increases with the palladium content for Pd–WO
TiO catalysts. Finally, although the existence of metallic palladium
x 2
/TiO catalysts (qv post). In addition, the palladium oxide
x
/
0.5%Pd/TiO2
2
is very likely in the tungsten-free samples, at least at the surface as
shown by the XPS data, (qv post), diffraction peaks corresponding
to bulk metallic palladium were not detected.
26
28 30 32 34 36 38 40 42 44 46 48 50
2θ
Fig. 5A presents a series of Raman spectra recorded from 300 to
À1
Fig. 4. Comparison of powder X-ray diffraction patterns for tungsten-promoted and
unpromoted Pd/TiO catalysts at different Pd loadings.
1
050 cm for the tungsten/palladium catalysts. The main peaks
2
À1
were observed at ca. 650, 500 and 400 cm , and these are typical
À1
of the anatase phase of TiO
2
. Shoulders at ca. 450 and 790 cm are
and position of the anatase TiO bands, indicating that the anatase
2
À1
also from anatase, and a shoulder at ca. 605 cm can be ascribed
support was not affected by the deposition of tungsten or palla-
to rutile [25]. Table 2 reports the Raman shifts and full width at
half-maximum (FWHM) values for the 650 cm anatase Raman
band. The addition of either tungsten or palladium alone to the
titania support did not modify significantly the relative intensities
dium. However, as detailed in Fig. 5B, the co-addition of tungsten
and palladium to the titania displaces the anatase bands to lower
frequencies by ca. 5 cm and changes their shape. Both the shift
À1
À1
in the position and the width of the TiO Raman bands have been
2
Table 2
Physical characteristics of the catalysts studied in the present work.
Catalyst
Pd (wt.%)
W (wt.%)
Surface area
Pd particle sizea,b
(nm)
Raman peak (FWHM)c
2
À1
À1
(
m
g
)
(cm
)
TiO
2
–
0.5
1.0
–
–
–
50
46
44
–
3.4
5.2
639 (29.5)
640 (29.4)
638 (29.7)
b
0
.5Pd/TiO
2
b
1
Pd/TiO
2
2
3
6
Pd/TiO
2
2.0
–
–
–
3.0
6.0
45
48
49
6.1b
–
–
639 (29.7)
640 (29.4)
639 (28.9)
W/TiO
W/TiO
2
2
a
0
0
0
0
1
2
.5Pd–1W/TiO
.5Pd–2W/TiO
.5Pd–3W/TiO
.5Pd–6W/TiO
2
2
2
2
0.5
0.5
0.5
0.5
1.0
2.0
1.0
2.0
3.0
6.0
3.0
3.0
46
47
45
48
48
49
n.d.
635 (35.2)
636 (32.4)
636 (31.9)
635 (32.3)
635 (32.9)
636 (34.6)
a
n.q.
a
6
a
8
a
Pd–3W/TiO
Pd–3W/TiO
2
9
a
2
10
n.d.: not detected, n.q.: not quantified.
a
By XRD.
By CO chemisorption.
b
c
À1
Raman frequency and full width at half-maximum (FWHM) for the Eg band at 640 cm
.

M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
107
A
B
C
i
h
i
i
hh
hh
gg
g
gg
f
e
d
f
f
e
e
c
b
a
d
d
c
cc
b
a
bb
a
1000
800
600
400
660
640
620
1000 900
800
Raman shift / cm-1
Raman shift / cm-1
Raman shift / cm-1
Fig. 5. Laser Raman spectra for Pd/W/TiO
2
2 2 2 2 2 2
catalysts. Catalysts: (a) TiO , (b) 0.5Pd/TiO , (c) 0.5Pd–1W/TiO , (d) 0.5Pd–2W/TiO , (e) 0.5Pd–3W/TiO , (f) 0.5Pd–6W/TiO , (g)
1
2 2 2
Pd–3/TiO , (h) 2Pd–3W/TiO and (i) 6W/TiO .
studied previously. Interpretation of the variation of Raman peak
shapes have been attributed to a range of factors, such as (i) a pres-
sure effect due to the surface tension of grains [26], (ii) a decrease
in the Ti–O bond strength, (iii) non-stoichiometry as induced by
oxygen deficiency or disorder induced by some minor phases in
the sample [27], and (iv) phonon confinement effects with the
grain size variation [28]. Since the difference in crystallite sizes
and microstrain for Pd–WO
x
/TiO
2
catalysts is not significant, and
0.5%Pd/6%W/TiO
2
2
the same phases are detected for all the catalysts, we can tenta-
tively conclude that the frequency shift and the FWHM increase
0.5%Pd/3%W/TiO
À1
of the 650 cm anatase band, as the tungsten and palladium load-
ing are varied, are primarily associated with the presence of more
0
.5%Pd/2%W/TiO
2
oxygen defects in the TiO
density takes place, resulting in reduction of titanium and oxida-
tion of palladium and/or tungsten in the Pd/W/TiO catalysts.
2
support. Consequently, a shift of charge
0.5%Pd/1%W/TiO
2
2
Vibrations related to tungsten species can also be observed in
the Raman spectra (Fig. 5C), although these bands exhibit lower
1
00
200
300
400
500
600
700
800
o
Temperature / C
intensity than those presented by the TiO
2
. A broad band at around
À1
9
80 cm , which can be assigned to terminal W@O stretching
Fig. 6. Temperature-programmed reduction profiles for Pd/W/TiO
Profiles for 0.5Pd–1W/TiO
2
catalysts.
modes corresponding to highly dispersed WO
detected [29]. As the tungsten loading increased, the band at ca.
80 also increased in intensity, indicating a progressive build-up
of WO surface species. Accordingly, the shift to a higher wave
x
species, was
2
, 0.5Pd–2W/TiO
2
, 0.5Pd–3W/TiO
2
2
and 0.5Pd–6W/TiO .
9
x
number of this Raman band with higher tungsten surface density
was attributed by Eibl et al. [30] to an increased polymerisation
of tungsten species. Comparing the Raman spectra of the 6 wt.%
2
W/TiO catalysts with and without palladium (spectra f and i,
respectively), it can be observed that the incorporation of palla-
dium causes both a shift in the position of the vibrational band
from 978 cm to 996 cm , and this is accompanied by a slight in-
crease in intensity. Thus, the presence of palladium acts to increase
2.0%Pd/3%W/TiO2
À1
À1
x
the polymerisation of WO species.
Temperature-programmed reduction profiles up to 900 °C are
shown in Figs. 6 and 7. These data mainly provide information
on the influence of the palladium sites on the reducibility of the
tungsten species. All the tungsten-containing catalysts showed a
TPR peak at ca. 700 °C (Fig. 6), and there was a general increase
in the total hydrogen consumption with the increase in tungsten
loading. This suggests an increase in the amount of potential
reducible species as the tungsten content was increased. The
hydrogen reduction peak has been related to the reduction of
0
.5%Pd/3%W/TiO2
0
200
400
600
800
o
Temperature / C
Fig. 7. Temperature-programmed reduction profiles for Pd–WO
Profiles for 0.5Pd–3W/TiO and 2Pd–3W/TiO
x 2
/TiO catalysts.
2
2
.
6
+
4+
W
to W in the surface of the WO
to note that the TPR profiles are similar to those previously
reported for WO /TiO catalysts [31], although a reduction peak
x
domains [31]. It is interesting
at ca. 550 °C was also observed for the palladium-containing cata-
lyst with the highest tungsten loading. The lower temperature
reduction peak exhibited by the most active 6.0% tungsten Pd/
x
2

1
08
M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
TiO
how the reduction of WO
loading. The main WO reduction feature was shifted to a slightly
2
sample was absent in the other catalysts. Fig. 7 demonstrates
experiments which indicate that a complete PdO ? Pd° transition
does not occur in tungsten-free catalysts. However, if tungsten is
x
species is influenced by the palladium
x
added at any concentration, the XP spectra show the complete ab-
0
lower temperature when the concentration of palladium was in-
creased. The profile also shows a negative peak below 100 °C,
which is distinctive for the decomposition of palladium hydride.
It is generally accepted that the main reduction peaks of
palladium occur at sub-ambient temperatures [32]. Fig. 8 shows
sub-ambient TPR profiles for a representative subset of catalysts
in order to probe the reducibility and nature of palladium sites in
more detail. The peaks observed at ca. 0 °C for tungsten-free Pd/
sence of metallic Pd , the palladium being present purely as PdO
(Fig. 9b–d).
2
The XP W(4f) spectra for the Pd–W/TiO catalysts overlap
strongly with the Ti(3p) signal from the titania support, and thus,
the Ti(3p) spectrum from the tungsten-free catalyst must be sub-
tracted from the composite signal. The methodology is shown in
Fig. 10: the Ti(3p) signal from titania (curve a) is scaled appropri-
ately and subtracted from the composite spectrum from the palla-
dium–tungsten catalyst (curve b) to give the ‘pure’ W(4f) spectrum
(curve c) which can then be quantified. The resulting W(4f) differ-
ence spectra indicate the presence of only 6+ species in the various
palladium–tungsten and pure tungsten catalysts (Fig. 11).
TiO
palladium. Table 3 shows the H
during the sub-ambient TPR experiments. For the tungsten-free
Pd/TiO catalysts, the H consumption represents around 80% of
2
catalysts can be assigned to the reduction of PdO
x
to metallic
2
consumption for these catalysts
2
2
0
the total theoretical reduction of PdO to Pd and is in agreement
with XPS data presented later. For tungsten-containing Pd cata-
lysts, a peak, which can be deconvoluted into two peaks, appears
at ca. 20 °C. The hydrogen consumption represents between 110%
and 150% of the theoretical consumption for the reduction of
Fig. 12 shows the ratio of the integrated W(4f) and Ti(2p) inten-
sities against nominal tungsten loading; the excellent linear corre-
lation is consistent with a high degree of dispersion of the W
species for all loadings, including the palladium-free W/TiO
2
cata-
lyst. These data also demonstrate that the titania support gets pro-
0
PdO to Pd . Therefore, hydrogen consumption at ca. 20 °C must also
gressively more covered by surface WO
loading is increased.
x
species as the tungsten
-TPD, and re-
be related to the reduction of new highly reducible species, which
are only present on the catalyst when palladium and tungsten are
present simultaneously.
The acidity of the catalysts was probed using NH
3
sults are shown in Fig. 13. Generally, the profile had a broad band
Surface-sensitive XPS was used to investigate the oxidation
state of the Pd species and any possible correlations with catalytic
with several maxima between 100 and 450 °C. The peaks
correspond to different strengths of acid sites, weak acid sites (
at ca. 150 °C, medium acid sites (b) at ca. 250 °C and strong acid
sites ( ) at ca. 400 °C. It is apparent that the relative amount of
a)
activity. For W-promoted Pd/TiO
the oxidation state of the palladium species is strongly influenced
by the addition of tungsten to the Pd/TiO catalysts (Fig. 9). For the
pure 0.5 wt.% Pd/TiO sample (Fig. 9a), the Pd exists as both
metallic species and PdO particles, the intensity ratio indicating a
2
catalysts, it was observed that
c
2
the medium and strong acid sites is dependent on the palladium
and tungsten loading. Thus, the number of b sites increases with
the palladium loading and decreases with the tungsten loading.
2
0
2+
composition of 40% Pd and 60% Pd [23]. This agrees with the
hydrogen consumption obtained in the sub-ambient TPR
On the other hand, the number of
c sites, most likely related to
polymeric tungsten species, increases with both the palladium
2.0%Pd/3.0%W/TiO2
2.0%Pd/TiO2
0
.5%Pd/6.0%W/TiO2
.5%Pd/3.0%W/TiO2
.5%Pd/TiO2
0
0
-
60
-40
-20
0
20
40
60
Temperature / ºC
Fig. 8. Sub-ambient temperature-programmed reduction profiles for representa-
tive Pd–WO /TiO catalysts.
x 2
Fig. 9. Pd(3d) spectra for the W-doped 0.5 wt.% Pd/TiO
a) 0 wt.%, (b) 1 wt.%, (c) 3 wt.% and (d) 6 wt.%.
2
catalysts; W loadings are
(
Table 3
2
Hydrogen consumption during the sub-ambient temperature-programmed reduction experiments (lmol H /g).
À1
a
Catalyst
Tmax (°C)
H
2
consumption (lmol H
2
g
)
2 2
H consumed/H (PdO ? Pd°) ratio
0
2
0
0
2
.5Pd/TiO
2
3
40
143
59
72
215
83
73
123
150
111
Pd/TiO
2
À4
31
8
.5Pd–3W/TiO
.5Pd–6W/TiO
.0Pd–3W/TiO
2
2
2
28
a
consumed in the PdO ? Pd transition, in%. Theoretical values in the case of complete reduction of Pd2+ to Pd0 for catalysts containing 0.5 and 2.0 wt.% Pd
/g, respectively.
H
2
consumed/H
2
are 48 and 194 mol H
l
2

M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
109
β
A
α
γ
2.0%Pd/3.0%W/TiO2
0
3
.5%Pd/3.0%W/TiO2
.0%W/TiO2
100
150
200
250
300
350
400
450
500
Temperature / ºC
α
γ
B
β
0.5%Pd/6.0%W/TiO
Fig. 10. Ti(3p) + W(4f) spectra for (a) the 0.5 wt.%Pd/TiO
2
catalyst, (b) the
0
.5 wt.%Pd–3 wt.%W/TiO catalysts and (c) the results of subtracting the Ti(3p)
2
contribution from spectrum (b).
0.5%Pd/3.0%W/TiO
0
0
.5%Pd/1.0%W/TiO
.5%Pd/TiO
100
150
200
250
300
350
400
450
500
Temperature / ºC
Fig. 13. Ammonia temperature-programmed desorption profiles for the different
catalysts; (A) effect of palladium loading and (B) effect of tungsten loading.
Table 4
Total quantities of ammonia desorbed from temperature programmed desorption
experiments.
Catalyst
Quantity of NH
3
desorbed
STP cm3
g
À1
)
(
0
3
0
0
0
2
.5Pd/TiO
2
3.85
3.92
3.86
3.93
3.99
5.18
W/TiO
2
Fig. 11. W(4f) difference spectra for the W-doped Pd/TiO
loadings as indicated.
2
catalysts, with W
.5Pd–1W/TiO
.5Pd–3W/TiO
.5Pd–6W/TiO
.0Pd–3W/TiO
2
2
2
2
studies on WO
the total acidity of WO
x
/TiO
2
catalysts [33,34]. These authors suggest that
/TiO catalysts increase with the tungsten
x
2
oxide surface density at a surface coverage lower than the mono-
À2
layer (<5 W nm ). Therefore, the NH
3
-TPD data are in agreement
with the results from other characterisation techniques, since we
have observed an increasing amount of polymeric species due to
increasing the WO loading and to the presence of palladium. There
x
is no direct relationship between catalytic activity and the total
acidity, since the catalyst with the highest acidity is not the most
active. In addition, the marginal increase in the acidity of the bi-
component catalyst with increasing tungsten loading cannot alone
account for the marked improvement of catalyst performance.
Nevertheless, strong acidity could still be important, since the most
active catalysts are those with the largest peaks at 400 °C in the
NH
In order to probe the catalyst structure more fully, detailed elec-
tron microscopy studies have focussed primarily on the Pd/TiO
material and the most active 0.5Pd/6W/TiO catalyst. Representa-
tive STEM–HAADF images from the simpler 0.5Pd/TiO catalyst
3
-TPD profiles.
Fig. 12. Plot of the ratio of the intensities of the W(4f) and Ti(2p) spectra against
nominal W loadings (wt.%) for the Pd–W catalysts, and the W-only sample.
2
2
and the tungsten loading. In addition, the total number of catalyst
2
acid sites, quantified from the total amount of NH
creases with both the tungsten and the palladium loading. (Table
). These data are in agreement with the findings from other
3
desorbed, in-
are shown in Fig. 14. In such Z-contrast images, the particles and
atoms/atomic columns display bright contrast, with the contrast
ꢀ1.7
4
level proportional to Z
. Due to the higher atomic number of

1
10
M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
Fig. 14. Representative STEM–HAADF images for the 0.5% Pd/TiO
at the junction of TiO support materials. The coloured circles in (B) indicate different PdO
purple – surface polymeric PdO species; blue – two-dimensional surface PdO rafts with dimensions of about 0.6- to 1.5-nm; yellow – 1- to 2-nm disordered PdO
particles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
2
catalyst taken at different magnifications. The orange circle in (A) highlights the presence of PdO
x
particles
2
x
surface species found on TiO : red – isolated PdO monomer or single Pd atoms;
2
x
x
x
x
nanoparticles; green – 2- to 4-nm crystalline PdO
article.)
x
Pd, compared with Ti and O, PdO
via the positions of Pd atoms/atomic columns, which show higher
image contrast against the TiO support. As shown in the low mag-
nification image (Fig. 14A), small PdO particles can be found on
the TiO surface at very low number density, and these small PdO
particles tend to reside at the junctions of the TiO
x
species can be clearly identified
two metal oxide components, and WO
ity for TiO surface sites [35,36]. The WO
observed in this co-impregnated sample are very similar to those re-
ported for a supportedWO /TiO catalystwitha similar WO loading
[30,31], which suggests that the additional presence of PdO does
surface
x
tends to have a strong affin-
2
x
structure and dispersion
2
x
3
2
3
2
x
x
2
support parti-
not significantly affect the bonding between WO
sites.
x 2
and TiO
cles as highlighted in Fig. 14A. However, high-resolution HAADF
images revealed the presence of at least five different PdO species
in the sample. As shown in Fig. 14B: (i) isolated PdO monomer or
single Pd atoms, (ii) polymeric PdO surface species formed by a
few PdO units, (iii) two-dimensional surface PdO rafts with a size
of about 0.6- to 1.5-nm, (iv) 1- to 2-nm disordered PdO nanopar-
ticles and (v) 2- to 4-nm crystalline PdO particles can all be
observed on the TiO support surface, and these are highlighted
by different coloured circles. Larger sized PdO particles were not
x
The addition of WO
of PdO on the TiO surface and results in a very noticeable increase
in the average PdO particle size. Crystalline PdO particles, 8–
15 nm in size, were observed in this sample, whilst smaller PdO
particles ꢀ2–4 nm in size were also occasionally found. Smaller
surface PdO species, i.e. isolated PdO and polymeric PdO species
as highlighted in Fig. 14, may also coexist with surface WO spe-
cies. However, the number density of these smaller PdO species
would be expected to be much lower than in the sample without
WO , as a large fraction of the palladium loading would be con-
sumed by the larger PdO particles. Interestingly, the surface of
the 2- to 4-nm PdO particles and the peripheral interface between
these small PdO particles and the TiO support are often found to
be decorated with WO species, as highlighted in Fig. 15C.
The larger PdO particles found in the co-impregnated sample
often display some epitaxial orientation relationship with respect
to the TiO support (Fig. 15D), which were not observed in the
sample without WO addition (Fig. 14). The formation of these
new epitaxial structures suggests that WO may well be present
at the interface between PdO and TiO particles and change the
x
, however, tends to decrease the dispersion
x
x
2
x
x
x
x
x
x
x
x
x
x
x
2
x
x
x
observed in this sample, which is consistent with XRD and CO
chemisorption results (Table 2). Although oxygen atoms associated
x
with the surface PdO
HAADF images, XPS results (Fig. 9) suggest that these PdO
are not fully oxidised, and some of them consist of metallic Pd.
Even though PdO nanoparticles were formed in this sample, it is
important to note that a large fraction of the TiO support surface
remains uncovered by the PdO species. In contrast, we have re-
cently shown that when WO alone is dispersed onto the surface
of TiO , most of the TiO surface sites would be expected to be cov-
ered by atomically dispersed surface WO species, before any
three-dimensional WO nanoparticles could be formed [35,36].
The coexistence of large areas on the bare TiO surface and PdO
particles in the tungsten-free catalysts suggests that PdO has a rel-
atively low affinity for the TiO support compared with WO
Moreover, the PdO nanoparticles found in the Pd/TiO sample
support grains,
x
species cannot be directly observed from the
x
x
species
x
x
2
x
x
2
x
x
x
2
2
2
x
x
x
3
x
2
2
x
interaction between these two oxide components. In order to fur-
ther test this hypothesis, STEM–XEDS mapping was performed on
this sample (Fig. 16). The sum spectrum extracted from the
x
2
x
.
x
2
x x 2
spectrum image indicates the presence of PdO , WO and TiO
are randomly oriented with respect to the TiO
and no epitaxial structures were observed.
2
(the C and Cu signals come from the carbon-coated Cu TEM grid).
Background subtraction was performed when extracting elemental
maps, in order to remove the artefacts due to varying background
signals associated with different sample thicknesses. The elemen-
tal maps obtained from the spectrum image are shown in
Fig. 16C–E, which reveal the spatial distribution of the three oxide
components. From the Pd map (Fig. 16C), it is clear that the Pd
loading is mainly congregated into the resulting PdO nanoparti-
x
cles which displayed the highest image contrast in the HAADF
image (Fig. 16A). By comparing the W and Ti maps (Fig. 16D and
Representative STEM–HAADF images from the 0.5%Pd/6%WO
TiO catalyst are shown in Fig. 15. In this sample, the WO species
were found to be highly dispersed on the TiO surface, mainly in
the form of monotungstate (i.e. isolated WO units) and two-dimen-
sional surface polytungstate species (highlighted in Fig. 15B and C),
and this is consistent with the Raman data. Some small ꢀ1 nm WO
clusters were also observed on the TiO surface. Previous studies
x
/
2
x
2
x
x
2
suggested that the surface structure of supported metal oxide cata-
lysts is strongly influenced by the wetting interaction between the
x
E), one can notice that WO species are highly distributed over

M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
111
Fig. 15. Representative STEM–HAADF images for the 0.5Pd–6WO
isolated WO (monotungstate); dark red – polymeric WO
species (polytungstate); dark green – ꢀ1 nm disordered WO
orange – 8- to 15-nm crystalline PdO particles. Inset: fast Fourier transform (FFT) of (D). The two diffraction spots correspond to tetragonal PdO (101) and rutile TiO
respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
x
/TiO
2
catalyst. The various coloured circles indicate different surface species present on TiO
2
: light blue –
x
x
x
clusters; green – 2- to 4-nm crystalline PdO
x
particles;
(110),
x
2
the entire surface of the TiO
2
particles, with some enhancement on
dispersed palladium oxide was much less active than larger crys-
talline particles supported on alumina. In the present study, it
was evident from chemisorption and X-ray line broadening data
that the average palladium particle size was increased when tung-
sten was incorporated into the catalyst. This was confirmed by
STEM analysis, which also showed that the average particle sizes
measured by XRD and chemisorption are an over-simplification,
as at least five palladium containing species were found to coexist,
and these spanned a range of dimensions. Aberration-corrected
electron microscopy data of this type, which show such a heteroge-
neous distribution of palladium, perhaps highlight why there is
such disagreement over the relationship between particle size
and activity. Although we observed that the more active catalysts
containing tungsten have larger palladium particles, we do not
consider this to be the predominant factor for controlling catalyst
the edge of the TiO particles. By colour-coding the three elemental
2
maps, we have reconstructed the relative spatial distribution of
these three oxide components (Fig. 16F). Significantly, a noticeable
increase in the W signal was observed at the interface/periphery of
the PdO
Fig. 16F) as well as between adjacent PdO
observation confirms that WO species are present at the junction
between PdO and TiO particles and may be promoting the forma-
tion of larger epitaxial structures between PdO and the TiO sup-
x
and TiO
2
particles (as indicated by the white arrows in
x
nanoparticles. This
x
x
2
x
2
port particles. High-resolution imaging and chemical mapping
have been performed on supported metal catalysts; however, data
of the resolution presented here are rare for supported metal oxide
systems, and the state-of-the-art technique employed provides new
insight to the structure of the catalysts.
2
activity. For example, in W-free Pd/TiO catalysts, the activity per
3
.3. Comments on the origin of improved catalytic activity
palladium site decreases with increasing PdO crystallite size.
XPS was used to investigate the oxidation state of the near-sur-
face palladium species. The palladium catalyst without tungsten
A degree of uncertainty persists around the effect of the noble
0
2+
metal particle size on hydrocarbon catalytic total oxidation. Some
studies have not observed a correlation between activity and par-
ticle size [37], whilst others have identified a strong correlation
had Pd and Pd species present, with ꢀ40% in the metallic state
and ꢀ60% in the more oxidised state. Once tungsten was incorpo-
rated into the catalyst, the surface oxidation state of palladium was
2
+
[
38]. Yazawa et al. [9] suggested that propane oxidation was af-
modified significantly, and only Pd was identified in the fresh
catalyst. It has been suggested that the palladium oxidation state
is an important factor in determining hydrocarbon total oxidation
activity and that partially oxidised palladium results in a more
fected by both palladium dispersion and oxidation state; however
on a series of supports, the oxidation state was designated to be the
dominant factor. Conversely, Hicks et al. [38] showed that highly

1
12
M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
Fig. 16. STEM–XEDS spectrum imaging of the 0.5Pd–6WO
was performed. (B) XEDS sum spectrum extracted from the spectrum image. (C) Background-subtracted Pd map. (D) Background-subtracted W map. (E) Background-
subtracted Ti map. (F) Colour coded reconstruction map with Pd in green, W in blue and Ti in red. The white arrows in F indicate the peripheral interface between PdO and
TiO particles, where an increase in W signal was observed. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
x 2
/TiO catalyst. (A) STEM–HAADF survey image. The green square indicates the area where the spectrum imaging
x
2
article.)
0
2+
active catalyst [9]. This is clearly not the case for these fresh Pd–
WO /TiO catalysts, as the Pd/TiO material, which contained
x 2 2
controlling activity, the Pd /Pd ratio would be expected to change
as the amount of tungsten was increased, as this resulted in an
increase in activity. These data and discussion are related to fresh
unused catalysts, and it would be informative to consider the
relative proportions of reduced and oxidised palladium on a used
catalyst. Hence, XPS was carried out on a used sample of the most
partially oxidised palladium, was less active than the tungsten-
containing catalyst with fully oxidised palladium species. Further-
2
+
more, all the fresh catalysts containing tungsten only had Pd
present, and if the presence of partially oxidised palladium was

M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
113
Smaller
PdOx
nanoparticles
Metallic Pd
Pd/TiO
2
Anatase (rutile)
Larger
Smaller
PdOx
nanoparticles
PdOx
Isolated
WOx
nanoparticles
Pd-WOx/TiO2
Low W-loadings
Anatase (rutile)
WOx on
the PdOx Larger
surface
Small
PdOx
Smaller
Isolated Polymeric
WOx enrichment
nanoparticles
PdOx
WOx
WOx WOx
at the interface
clusters
nanoparticles
Pd-WOx/TiO
2
High W-loadings
Anatase (rutile)
Thin WOx
layer between
2
TiO and PdOx
x 2 x x
Fig. 17. Structural model of the Pd/WO /TiO catalyst without addition of WO and with increasing WO loading.
0
2+
active 0.5Pd/6W/TiO
5, showing a higher degree of reduced palladium.
Sub-ambient TPR data of the Pd–WO /TiO catalysts show that
2
catalyst. After use, the Pd /Pd ratio was 45/
shift to higher frequencies with increasing tungsten loading at a
fixed palladium content. Hence, these results provide evidence
for increasing polymerisation of the surface tungstate as the tung-
sten loading increases. However, the dispersion of tungsten, as cor-
roborated by XPS, is excellent. The STEM data are in agreement as
these indicate mainly the presence of isolated surface monotung-
state and polytungstate species. The presence of palladium in
palladium–tungsten catalysts led to a marginal improvement of
the reducibility of the oxidised tungsten species that are not asso-
ciated with palladium nanoparticles. However, we consider that
this particular modification is not directly responsible for the im-
5
x
2
the reduction of the palladium species takes place at lower temper-
atures with increasing both the tungsten and palladium loadings,
and this correlates with the catalytic activity. However, there is
no simple relationship between ease of palladium reduction and
the activity of the catalysts, as the Pd/TiO catalyst showed palla-
2
dium reduction at a lower temperature than the tungsten-contain-
ing catalysts, but it has lower activity. The more facile reduction of
0
the Pd/TiO
which are absent in Pd–WO
tion and dissociation of H . There does appear to be a correlation
between the activity of the Pd–WO /TiO catalysts and the magni-
2
catalyst could be due to the presence of Pd sites,
x
/TiO
2
, and these facilitate chemisorp-
x
proved catalytic performance, since WO , regardless of the way it
2
is structured, presents a very low reactivity for propane total oxi-
dation. Moreover, according to the TPR experiments, the most
reducible isolated tungsten species in W/Pd catalysts exhibit
reduction bands at 450–500 °C, which are temperatures much
higher than those required for the complete conversion of propane,
and indicate that these oxygen species are not particularly reactive.
The presence of tungsten also modified the structure of the pal-
ladium particles, as electron microscopy showed evidence for for-
mation of larger epitaxial palladium oxide particles, which were
not present without tungsten. The structural modification of the
palladium oxide particles could have some influence on increasing
activity, but the potential influence of the epitaxial orientation is
unclear. For supported metal particles, it has been reported that
epitaxial formation of particles, induced by interaction with a sup-
port, can result in strain between the oxidised metal particle and
metal oxide support [39]. It is also recognised that strained metal
nanoparticle surfaces have modified adsorption energies and acti-
vation energies for reaction that differ from unstrained particles
[40]. Although these studies were focussed on supported metal
nanoparticles, it is not unreasonable to expect that the activity of
x
2
tude of the low-temperature reduction feature (Table 3). The com-
bination of palladium and tungsten species gives rise to a larger
number of reducible sites than can be accounted for based on the
maximum quantity from palladium alone. Hence, a range of new
highly reducible sites are generated, and these have facile oxygen
species associated with them. We consider that these facile oxygen
species are responsible for the increased oxidation activity and
postulate that they are created at the peripheral interface between
the WO
x
species and the oxidised palladium nanoparticles on the
TiO support.
2
Just as the presence of tungsten modifies the structure of palla-
dium, it has also been observed that the presence of palladium af-
fects the nature of the tungsten species. According to Raman
spectroscopy data, more polymeric tungsten species are present
on a palladium–tungsten catalyst than in a tungsten only catalyst
with the same tungsten content. The characteristic vibrational
À1
band in the frequency range between 950 and ca. 1020 cm is as-
signed to W@O stretching vibration. The positions of these bands

1
14
M.N. Taylor et al. / Journal of Catalysis 285 (2012) 103–114
epitaxial palladium oxide particles may differ from particles that
do not show epitaxy. However, not all the palladium oxide parti-
cles of the tungsten modified catalyst showed epitaxial orientation
and there were many that did not demonstrate epitaxy, accord-
ingly it is not possible to deconvolute the relative activity from
each particle type.
Acknowledgments
Authors wish to thank the Ministry of Science and Innovation
(Spain) and the Generalitat Valenciana through Projects CTQ2009-
14495 and GVPRE2008/352 for funding. C.J.K. is grateful to the NSF
for funding provided under the Materials World Network Program
(Grant # DMR-0709887).
However, we consider that the most crucial factor for the
enhanced activity is associated with the interface between the sup-
ported WO
x
species and the palladium particles. By correlating the
References
structure observed from aberration-corrected STEM characterisa-
tion with catalytic activity data, we propose that the new
WO -decorated interface between PdO and TiO particles may
x x 2
[
[
1] R.E. Heyes, S.T. Kolazkowski, Introduction to Catalytic Combustion, Gordon and
Breach, Amsterdam, 1997.
2] Q. Liu, L.-C. Wan, M. Chen, Y. Cao, H.-Y. He, K.-N. Fan, J. Catal. 263 (2009) 104–
113.
3] N.A. Merino, B.P. Barbero, P. Ruiz, L.E. Cadus, J. Catal. 240 (2006) 245–257.
4] R.B. Anderson, K.C. Stein, J.J. Feenan, L.J.E. Hofer, Ind. Eng. Chem. 53 (1961)
be responsible for the enhanced catalytic activity in the co-impreg-
nated catalysts. Supporting evidence for this is supplied from sub-
ambient TPR studies, which identified an increased concentration
of highly reactive oxygen species on the catalyst. Some further evi-
dence for these species is also apparent from the XPS analysis of
the used palladium–tungsten catalyst, as after steady-state opera-
tion there was an increase in reduced palladium species on the cat-
alyst surface, indicating their lability to react with propane. The
reactive oxygen species are thought to be important for the in-
creased activity as palladium-based catalysts for hydrocarbon total
oxidation are known to operate via a Mars-van Krevelen-type
mechanism [12].
[
[
809–812.
[5] S. Specchia, E. Finocchio, G. Busca, P. Palmisano, V. Specchia, J. Catal. 263
(2009) 134–145.
[
[
[
6] T. Garcia, B. Solsona, S.H. Taylor, Catal. Lett. 97 (1–2) (2004) 99–103.
7] A. Brayner, D.S. Cunha, F. Bozon-Verduraz, Catal. Today 78 (2003) 419–432.
8] T. Garcia, B. Solsona, D.M. Murphy, K.L. Antcliff, S.H. Taylor, J. Catal. 229 (2005)
1–11.
9] Y. Yazawa, H. Yoshida, N. Takagi, S. Komai, A. Satsuma, T. Hattori, J. Catal. 187
(
[
1999) 15–23.
10] K. Eguchi, H.i. Arai, Appl. Catal. A 222 (2001) 359–367.
[11] H. Bosch, F. Janssen, Catal. Today 2 (1998) 369.
12] T.V. Choudhary, S. Banerjee, V.R. Choudhary, Appl. Catal. A 234 (2002) 1–23.
13] K.-I. Fujimoto, F.H. Riberio, M. Avalos-Borja, E. Iglesia, J. Catal. 179 (1998) 431–
[
[
[
Based on the catalytic data and the detailed characterisation
studies, we have been able to propose a model to describe the
structure of the catalyst (Fig. 17). With and without tungsten, pal-
ladium oxide nanoparticles were highly dispersed, although they
were on average slightly larger and had some epitaxial orientation
when tungsten was present. At lower tungsten loadings, isolated
tungstate and some polytungstate species were highly dispersed
over the titania support. The concentration of polytungstate in-
creased as the loading was increased, as did the coverage over
the titania. Finally, there were some tungstate species present on
the palladium oxide particles, and there was an increased concen-
tration of oxidised tungsten species at the peripheral interface of
the palladium oxide nanoparticles and the titania.
4
42.
[14] C.A. Müller, M. Maciejewski, R.A. Koeppel, R. Tschan, A. Baiker, J. Phys. Chem.
00 (1996) 20006–20014.
[
[
1
15] T.E. Hoost, K. Otto, Appl. Catal. A 92 (1992) 39–58.
16] M. Vassileva, A. Andreev, S. Dancheva, N. Kotsev, Appl. Catal. A 49 (1989) 125–
141.
17] P. Papaefthimiou, T. Ioannides, X.E. Verykios, Appl. Catal. B 15 (1998) 75–92.
18] P. Papaefthimiou, T. Ioannides, X.E. Verykios, Appl. Therm. Eng. 18 (1998)
[
[
1005–1012.
[19] F.B. Noronha, D.A.G. Aranda, A.P. Ordine, M. Schmal, Catal. Today 57 (2000)
275–282.
[
[
[
[
20] A. Törncrona, M. Skoglundh, P. Thormählen, E. Fridell, E. Jobson, Appl. Catal. B
4 (1997) 131–145.
1
21] P.O. Thevenin, A. Alcalde, L.J. Pettersson, S.G. Järås, J.L.G. Fierro, J. Catal. 215
(2003) 78–86.
22] G. Pecchi, P. Reyes, T. López, R. Gómez, J. Non-Cryst. Solids 345–346 (2004)
624–627.
23] M. Taylor, E. Ntainjua N., T. Garcia, B. Solsona, A.F. Carley, S.H. Taylor, Appl.
Catal. A 350 (2008) 63–70.
[
[
24] H.P. Klug, L.E. Alexander, X-Ray Diffraction Procedures, Wiley, New York, 1967.
25] J. Engweiler, J. Harr, A. Baiker, J. Catal. 159 (1996) 259–269.
4
. Conclusions
2
Pd/WO /TiO catalysts show high activity for the total oxidation
[26] M.H. Lee, B.C. Choi, J. Am. Ceram. Soc. 74 (1991) 2309.
[
[
[
[
27] J.C. Parker, R.W. Siegel, J. Mater. Res. 5 (1990) 1246.
28] H.C. Choi, Y.M. Jung, S.B. Kim, Vib. Spectrosc. 37 (2005) 33.
29] G.C. Bond, S. Flamerz, L. van Wijk, Catal. Today 1 (1987) 229–243.
30] S. Eibl, B.C. Gates, H. Knözinger, Langmuir 117 (2001) 107.
x
of propane, and the activity is significantly greater than equivalent
catalysts that do not contain tungsten. This enhanced activity is not
simply the result of the additive effect of active sites on the palla-
[31] D. Vermaire, P. Van Berge, J. Catal. 116 (1989) 309–317.
[
[
[
32] G. Chen-Bin, L. Hung-Kuan, H. Chi-Man, J. Mol. Catal. A 180 (2002) 285–291.
33] T. Kim, A. Burrows, C.J. Kiely, I.E. Wachs, J. Catal. 246 (2007) 370–381.
34] X.-L. Yang, W.-L. Dai, Guo, H. Chen, Y. Cao, H. Li, H. He, K. Fan, J. Catal. 234
(2005) 438–450.
35] W. Zhou, E.I. Ross-Medgaarden, W.V. Knowles, M.S. Wong, I.E. Wachs, C.J.
Kiely, Nat. Chem. 1 (2009) 722–728.
36] W. Zhou, I.E. Wachs, C.J. Kiely, Curr. Opin. Solid State Mater. Sci. (2011),
doi:10.1016/j.cossms.2011.06.00.
37] T.R. Baldwin, R. Burch, Appl. Catal. A 66 (1990) 337–358.
38] R.F. Hicks, H. Qi, M.L. Young, R.G. Lee, J. Catal. 122 (1990) 295–306.
39] N. Barrabés, K. Föttinger, A. Dafinov, F. Medina, G. Rupprechter, J. Llorca, J.E.
Sueiras, Appl. Catal. B 87 (2009) 84–91.
x
dium and tungsten species, as WO presents very low activity for
the total oxidation of propane. Hence, there is a synergistic rela-
tionship between palladium and tungsten, and it is proposed that
this is due to highly active oxidation species that are formed at
[
[
2 x
the peripheral interface between TiO , WO and PdO nanoparticles.
[
[
[
It is also interesting that the presence of tungsten induces epitaxial
orientation in palladium nanoparticles. At present, a relationship
between this epitaxial orientation and the high catalytic activity
can be neither demonstrated nor excluded, and the effect is worthy
of further study.
[
40] M. Mavrikakis, B. Hammer, J.K. Nørskov, Phys. Rev. Lett. 81 (1998) 2819.