L. Wang et al. / Catalysis Today 149 (2010) 163–166
165
yield, further confirming that a bi-functional catalyst comprising
palladium and an acidic support is favorable for the selective
formation of acetic acid [4]. The catalytic activity (or yield of acetic
acid) clearly increased as a function of Pd loading, with 2.0 wt.% Pd/
3 2
WO –ZrO exhibiting the highest ethylene conversion. Although
Pd dispersion also varied with the Pd loading (Table 1), the
difference in catalytic activity cannot be explained by this
difference, suggesting that Pd dispersion is not responsible for
2
the catalytic activity. Combining the H -TPR data and the results
for catalytic activity, it is noteworthy that the ethylene conversion
increased with the Pd content, in agreement with the polytung-
state species trend at 613 K in Fig. 1. Based on this finding, we
speculate that the presence of polytungstate species at 613 K plays
an important role in increasing the activity of Pd-containing WO
ZrO catalysts. On the other hand, the CO selectivity showed no
obvious change when the Pd loading was increased from 0.5 to
.0 wt.%, perhaps because of the similar Pd state. Only one
3
–
2
x
2
significant difference in product distribution was observed: lower
acetic acid and higher acetaldehyde selectivity for 0.5 wt.% Pd/
3 2
WO –ZrO . The reason for this could be the lower Pd loading,
which may not have been sufficient to oxidize the intermediate
acetaldehyde to acetic acid.
3.2. Effect of Pd precursor
The above results indicate that the catalyst with 1.0 wt.% Pd
loading exhibited satisfactory catalytic activity and acetic acid
yield. Thus, to minimize the amount of Pd used for economic
reasons, subsequent investigations focused on the 1.0 wt.% Pd/
2
Fig. 2. H -TPR spectra of Pd/WO
3
–ZrO
2
catalysts prepared with different Pd
WO
3
–ZrO
2
catalyst. The BET surface area and Pd dispersion
chemisorption results for this catalyst prepared
precursors: (a) PdCl
2
; (b) Pd(NH Cl ; (c) Pd(CH COO) .
3
)
2
2 3 2
calculated from H
2
from different Pd precursors are summarized in Table 2. No
appreciable change in BET surface area was observed, indicating
that the Pd precursor has almost no influence on the textural
precursors exhibited a very sharp negative peak at 360 K. This
suggests that PdO agglomeration occurred, leading to the
formation of large particles that were easily reduced to metallic
Pd, for which the characteristic negative TPR peak (360 K) is much
structure of the catalysts. An interesting observation from the H
2
chemisorption results is that a change in palladium precursor
yielded different Pd dispersion. Higher dispersion was obtained
larger than that for the sample prepared from PdCl
2
. In particular,
the sample prepared from Pd(CH COO) exhibited two negative
3
2
using PdCl
2
as the palladium source compared to Pd(NH
3
)
2
Cl
2
and
peaks, which may be due to the formation of considerably larger Pd
particles. This result is in good agreement with the dispersion and
particle size of palladium in Table 2. On the other hand, it is
important to note that the Pd precursor has an effect on the state of
Pd(CH COO)
3
2
precursors, clearly indicating that differences in
dispersion are related to the nature of the Pd precursor. It has been
reported that precursors with organic ligands yield lower
dispersion than the precursors containing ammonia [13]. The
reason is that localized heating due to organic ligand oxidation
during calcination may be sufficient to result in sintering of novel
metal particles. In addition, metal–support interaction is an
important factor that contributes to the overall dispersion on
tungsten, especially polytungstate species, which exhibit
reduction peak at approximately 613 K with intensity decreasing
in the order PdCl > Pd(CH COO) > Pd(NH Cl
Comparison of the catalytic performance of 1.0 wt.% Pd/WO
ZrO catalysts for direct oxidation of ethylene reveals a significant
differencefordifferentPdprecursors(Table 2).Thecatalystprepared
from PdCl was the most active and selective, with catalytic activity
decreasing in the order PdCl > Pd(CH COO) > Pd(NH Cl , which
a
2
3
2
3
)
2
2
.
3
–
2
catalysts. For PdCl
2
and Pd(NH
3
)
2
Cl
2
it is known that Pd exists as
2
À
2+
PdCl
4
and Pd(NH
3
)
4
, respectively. Anion exchange reactions
À
2
2
À
between PdCl
4
and OH groups on the support result in strong
2
3
2
)
3 2
2
interaction between Pd species and the support [14,15], which
could inhibit metal agglomeration on the surface and result in
highly dispersed Pd, as confirmed by particle size data.
is consistent with the order for the reduction peak at 613 K assigned
to polytungstate species (Fig. 2). This result further confirmsthat the
catalytic activity of Pd/WO
is strongly dependent on the amount of polytungstate species. In
addition, the sample prepared from PdCl , with the lowest-intensity
3 2
–ZrO for selective oxidation of ethylene
2
H -TPR measurements (Fig. 2) clearly demonstrate that the Pd
precursor has an important influence on the state of palladium
species. In comparison to the TPR profile for the sample prepared
2
negative peak at 360 K (Fig. 2) clearly indicating the presence of
well-dispersed Pd particles, showed the highest acetic acid
2 3 2 2 3 2
using PdCl , the samples for Pd(NH ) Cl and Pd(CH COO)
Table 2
Physical properties of 1.0 wt.% Pd/WO
3
–ZrO
2
catalysts prepared with different Pd precursors.
a
Pd precursors
Surface
area (m
Pd dispersion
(%)
Pd particle
Ethylene
Acetic acid yield
Selectivity (%)
Acetic acid
2
À1
a
À1 À1
g
)
size (nm)
conversion (%)
(g h
L
cat)
Acetaldehyde
CO
x
PdCl
Pd(NH
2
39
32
37
12.0
6.4
9.3
17.5
4.2
1.8
2.1
117
47
75
65
69
11
19
9
14
16
22
3
)
2
Cl
COO)
2
Pd(CH
3
2
0.6
192.5
59
a
2
Calculated from H chemisorption.