Deactivation and regeneration studies of a PdSb/TiO2 catalyst used in the gas-phase acetoxylation of toluene

Article abstract of DOI:10.1016/j.jcat.2011.06.005

A PdSb/TiO₂ catalyst was subjected to selective gas-phase acetoxylation of toluene (Tol) to benzyl acetate (BA). The catalyst showed low initial activity, which increased with time and displayed maximum activity after 6 h on-stream. The catalys

Full text of DOI:10.1016/j.jcat.2011.06.005

Journal of Catalysis 282 (2011) 103–111
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Deactivation and regeneration studies of a PdSb/TiO2 catalyst used in the gas-phase
acetoxylation of toluene
⇑
N. Madaan, S. Gatla, V.N. Kalevaru, J. Radnik, B. Lücke, A. Brückner, A. Martin
Leibniz-Institut für Katalyse e.V. an der Universität Rostock, Albert-Einstein-Str. 29a, D-18059 Rostock, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
A PdSb/TiO2 catalyst was subjected to selective gas-phase acetoxylation of toluene (Tol) to benzyl acetate
(BA). The catalyst showed low initial activity, which increased with time and displayed maximum activity
after 6 h on-stream. The catalyst exhibited stable performance for the next 25 h on-stream and then
started to deactivate rapidly. Carbon analysis pointed to coke deposition as a main reason for the loss
of catalytic activity. Regeneration at 250 °C could restore the catalyst activity but showed rapid deactiva-
tion again in a subsequent run. Regenerative treatment at 300 °C was much more effective not only in
restoring the performance but also in maintaining relatively good stability of the catalyst. However, treat-
ment at 350 °C and higher was found to be ineffective in restoring catalytic activity. This surprising
behavior of the catalyst was studied in detail with various techniques such as XRD, TEM, and XPS anal-
ysis. XRD data give evidence on shifting of Pd reflection to lower 2h values and the splitting into two parts
pointing to different types of Pd species. The surface ratio of Pd0/PdO of 1 seems to be essential for better
performance as revealed by XPS analysis. Surprisingly, this ratio is significantly changed in the deacti-
vated catalyst. Moreover, only metallic Pd species were found on the catalyst surface of this deactivated
solid. In addition, regeneration at 300 °C could restore PdO concentration to a greater extent, while the
one regenerated at 350 °C could not restore the PdO proportion. This behavior is also believed to be
one of the reasons for the unproductive regeneration of deactivated catalyst at 350 °C. Furthermore,
Pd0, PdO, and Sb synergy is needed for stable and high performance of the catalyst.
Received 12 April 2011
Revised 1 June 2011
Accepted 2 June 2011
Available online 5 July 2011
Keywords:
Acetoxylation
Gas-phase reaction
Deactivation
Coking
Regeneration
Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction
in chemical industry notably as a solvent for cellulose acetate [4].
The global demand for benzyl acetate ranges from 5000 to
The catalytic partial oxidation of various hydrocarbons is one of
the most important topics in the field of catalysis due to various
challenges involved. The major problem in oxidation reactions is
to prevent consecutive oxidation of the desired products into un-
wanted by-products. This difficulty can be overcome by carrying
out controlled oxidation in the presence of efficient catalysts
and/or using auxiliary agents. In acetoxylation, oxidation takes
place in the presence of acetic acid, and the acetyl group formed
prevents the oxygenate from further oxidation. One such good
example is the acetoxylation of ethylene [1] in the presence of a
Pd catalyst resulting in vinyl acetate, based on an early observation
of Moiseev and co-workers [2]; vinyl acetate is one of the big-scale
industrial chemicals [3]. A similar reaction with respect to aromat-
ics is the acetoxylation of toluene (Tol) to benzyl acetate (BA) via
gas-phase oxidation. BA is naturally found in plants such as
jasmine, hyacinth, gardenias, and azaleas. It is used chiefly in the
perfume and food industry due to its fruity aroma. It is also used
10,000 tonnes per annum. In addition, it might serve as an inter-
mediate for the manufacture of benzyl alcohol that is presently
obtained by rather harmful chlorination of toluene.
It is widely known that Pd-based catalysts are active and selec-
tive for different acetoxylation reactions [5–9], in general. Majority
of work in the acetoxylation of toluene to BA is carried out in liquid
phase [10–12]. However, there has not been much success in the
gas-phase acetoxylation of toluene in the past; very low yields of
acetoxylated products were claimed [13–15], and only a maximum
yield of BA around 7% could be achieved. Recently, our group [16–
20] reported that a titania (anatase)-supported Pd–Sb catalyst gave
a high conversion of toluene (X = 90%) with high selectivity of BA
(S = 85%). However, such catalyst required a long formation period
of around 10 h to exhibit the maximum performance. After acquir-
ing its best activity, the catalyst displayed rapid deactivation after
some hours on-stream. Inspired by these limitations of the cata-
lyst, we have recently investigated [21] the effects of calcination
atmosphere (air, He, and 10%H2/He) and temperature (300–
600 °C) on the surface properties of the ready-to-use catalyst. It
was found that the catalyst treated in helium at 600 °C showed a
significant reduction in the formation period from 10 to 6 h. Addi-
⇑
Corresponding author. Fax: +49 (0)381 1281 51246.
E-mail address: andreas.martin@catalysis.de (A. Martin).
0021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.06.005

104
N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
tionally, the catalyst revealed good long-term stability up to 26 h
on-stream. However, this catalyst also suffered from deactivation,
which presently seems to be inevitable.
The original state of a catalyst changes during deactivation,
which could be due to physical and/or chemical changes leading
to a loss in activity and/or selectivity [22,23]. Albers et al. [24] have
studied the deactivation of Pd catalysts in detail; they identified
particle growth, coke deposition transformation, alterations in
the support material, and modifications at the Pd surface such as
valency changes as main reasons.
2.1.2. Characterization
Thermogravimetric analysis and differential scanning calorime-
try (TG–DTA) were done with a TGA 92 (Setaram). The heating rate
was 10 K/min till 700 °C in air flow.
The elemental composition (Ti, Pd, Sb, and Na) was determined
by ICP-OES using a Varian 715-ES ICP-emission spectrometer. The
data analysis was performed using the ICP Expert software. Carbon,
hydrogen, nitrogen, and sulfur analysis was done using the CHNS
microanalysator TruSpec (Leco).
The surface area (BET) and pore size distribution of the catalysts
were determined using a Nova 4200e device (Quantachrome
Instruments). Before the measurement, the catalyst was evacuated
for 2 h at 200 °C to remove physisorbed water.
In general, coking is among the most common reasons for cata-
lyst deactivation. Many different forms of carbons have been ob-
served in different reactions [25,26]. Some are deposited on the
surface of the catalyst, some are transformed from one to another
form of carbon, and some are just generated during the reaction
and have adverse effect on the catalytic activity. For example, Kar-
ge et al. [27] have studied and confirmed the formation of different
types of coke on zeolite catalysts. They identified different carbon
forms depending upon the temperature of the coke-forming reac-
tion, i.e., at reaction temperature of less than 246 °C and other at
temperature higher than 300 °C. These forms could be discrimi-
nated by observing different IR bands. It is widely known that Pd
catalysts undergo deactivation due to deposited carbon species,
either by their deposition on the surface or by their interaction
with the surface, e.g., PdC formation. In some cases, the restoration
of the catalyst performance is easily and successfully obtained by
just burning off such deposits [28]. However, in some deactivated
catalysts, regeneration by burning off carbon was not helpful to re-
gain the catalyst initial behavior [29]. In such cases, it cannot just
be attributed to the carbon accumulation on the surface, but also to
its interaction with active sites of palladium on the surface and the
strength of such bonding. Further, some authors have reported [30]
that it is difficult to regenerate Pd catalysts because not only the
carbon removal is important but also other aspects play a role
for restoring activity, e.g., changes in the structure, morphology,
and/or valency of the active component.
X-ray diffraction (XRD) powder patterns at ambient conditions
were recorded in transmission geometry with Cu K
a radiation
(k = 1.5406 Å) in the 2h range of 10–55° (step width = 0.25°) on a
Stoe STADI P diffractometer, equipped with a linear position sensi-
tive detector (PSD). Time per step was 25 s for the samples without
Si as standard and 160 s with Si as standard. Processing and assign-
ment of the powder patterns was done using the software Win
Xpow (Stoe) and the Powder Diffraction File (PDF) database of
the International Centre of Diffraction Data (ICDD).
Transmission Electron Microscopy (TEM) investigations were
carried out at 120 keV using an EM-420T microscope (Philips).
Samples were prepared by depositing the catalysts on a copper
grid.
X-ray photoelectron spectra (XPS) were recorded on a VG ESCA-
LAB 220iXL instrument with Al K
a radiation (E = 1486.6 eV). The
samples were fixed with a double adhesive carbon tape on a
stainless steel sample holder. The peaks were fitted by Gaussian–
Lorentzian curves after Shirley background subtraction. The elec-
tron binding energy was referenced to the Ti 2p3/2 peak of TiO2
at 458.8 eV. For the quantitative analysis of the near-surface
region, the peak areas were determined and divided by the
element-specific Scofield factor and the analysator-depending
transmission function.
In the present study, we focus mainly on gaining insights into
the deactivation of a Pd–Sb catalyst supported on titania (anatase)
and calcined in helium at 600 °C for the acetoxylation of toluene.
Moreover, the research was extended further to obtain the best
route to regenerate the deactivated catalyst and to restore original
activity and selectivity.
2.2. Catalyst testing
Heterogeneously catalyzed acetoxylation runs in gas phase
were performed in a microcatalytic fixed-bed, vertical, and tubular
HastelloyÒ C reactor (length 250 mm; i.d. 9.4 mm). The tests were
carried out at a reaction temperature of 210 °C and at a reaction
pressure of 2 bar. The molar ratio of toluene/acetic acid/oxygen/
nitrogen was set to 1:4:3:16 with a gas hourly space velocity
2. Experimental
(GHSV) of 2688 hÀ1 and a residence time (
s) of 1.34 s. The reaction
gases, like O2 and 5%CH4/N2 (used as a diluent gas), were supplied
from commercially available compressed gas cylinders and applied
without further purification. Methane was used as an internal vol-
ume standard. The flow rates of these gases were measured with
mass flow controllers. The organic feed mixture (molar ratio of
1:4) of toluene (>99.9%, Roth, Germany) and acetic acid (>99.9%,
Walter CMP, Germany) was pumped to the reactor with a HPLC
pump (Shimadzu LC 9A). The liquid reactant mixture was vapor-
ized before it entered the reactor in a preheating zone provided
on the top of the reactor. Before, the reactor was heated to the de-
sired reaction temperature under inert gas. Afterward, the reactant
gases and liquid feed were introduced, and the reaction was car-
ried out. The product stream was analyzed online by gas chromato-
graph (Shimadzu GC-2010) using a WCOT-fused silica capillary
column (60 m, 0.32 mm) and a FID detector with methanizer.
Highly purified (>99.99) gases were used for thermal treatment
and for catalytic tests.
2.1. Catalyst preparation and characterization
2.1.1. Preparation
A 10 wt.% Pd and 16 wt.% Sb on TiO2 catalyst were prepared in
two steps. In the first step, Sb2O3 (>99.9% Sigma–Aldrich) powder
was suspended in 20 ml distilled water and stirred for 30 min.
Then, TiO2 (anatase, Kronos) was added to Sb2O3 slurry and was
stirred for 1 h at ambient temperature; 25% ammonia solution
was added dropwise to adjust pH = 7. The solution was heated at
70 °C for 1 h, followed by solvent removal in a rotary evaporator.
The solid obtained was dried at 120 °C for 16 h and calcined in
air at 400 °C for 3 h. In the second step, PdCl2 (>99.9% Alfa) was dis-
solved in diluted HCl. The solution obtained was cooled to room
temperature and 1 N Na2CO3 (>99% Alfa) solution was added to ad-
just the pH = 4. The solid prepared in step 1 was added to the above
solution and stirred for 1 h, followed by solvent evaporation in a
rotary evaporator. The resulting solid was dried at 120 °C for
16 h and then calcined for 4 h at 600 °C in He flow. Details can
be seen elsewhere [21,31].
Calculations of conversion of toluene (X-Tol) and acetic acid (X-
AcOH) are based on their molar streams at the inlet and the outlet
of the reactor. The yield of formed products (Yi) and their selectiv-

N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
105
ities (Si) were calculated from the molar streams too. Besides some
100
80
60
40
20
0
7
6
5
4
3
2
1
0
carbon oxides from total oxidation, and BA as the main product,
minor amounts of benzaldehyde (BAL) were seen. The carbon bal-
ance was evaluated from non-converted reactants and yields of
identified and calibrated products.
Regeneration of the deactivated catalyst was carried out in situ
in the used setup at four different temperatures (250, 300, 350, and
400 °C) for 2 h in air.
Several catalyst samples were taken out from the reactor after
various reaction stages, deactivation and regeneration. The list of
these catalysts studied and their notation used in this paper is gi-
ven in Table 1.
X-Tol
X-ACOH
Y-BA
Y-CO2
3. Results and discussion
3.1. Catalyst testing
1
3
6
9
12 15 18 21 24 27 30 33 36 39 42
Time (h)
Fig. 1 clearly demonstrates the catalytic performance of PdSb-f
catalyst with time-on-stream (tos). It displays a very low initial
activity (X-Tol = ca. 2%), which increased up to ca. 6 h (X-
Tol = >50%), remained stable for almost 30 more hours, and then
decreased with further increase in reaction time (ca. 18% after
42 h). The yields of BA and BAL as by-products were also observed
to change in a similar way as that of conversion of toluene. In addi-
tion, CO2 was also formed to a minor extent. This behavior is due to
the fact that catalytic activity is closely connected to Pd particle
size and particle composition [17–19]. The catalyst needs a certain
amount of time for the growth of Pd particles to a critical size and
composition [17]. However, the catalyst showed deactivation
above 32 h on-stream. This is due to a predominant coke deposi-
tion during the course of the reaction and thereby eventual cover-
age of the active sites. Table 2 summarizes the amount of carbon
deposits at different stages of the reaction. It can be clearly seen
that the carbon content increased steadily up to 3 wt.% in the deac-
tivated sample (PdSb-u42).
Fig. 1. Catalyst performance of the He calcined 10Pd16Sb/TiO2 with time-on-
stream.
Table 2
Carbon content and BET surface area of the 10Pd16Sb/TiO2 catalyst at different stages
of the reaction.
Catalyst at different stages
C (wt.%)
BET surface area (m2/g)
PdSb-f
PdSb-u6
PdSb-u42
PdSb-u42-r300
PdSb-u42-r350
PdSb-u42-r300-u7
PdSb-u42-r350-u7
0
0.65
3.0
0.38
0.26
1.78
1.63
28.6
40.6
34.5
31.3
32.2
16.2
18.1
At first, to find out the suitable temperature region for the effec-
tive removal of the carbonaceous species, TG–DTA analysis was
done until 700 °C in the presence of air. Fig. 2 shows the corre-
sponding plots of the deactivated sample (PdSb-u42). An endother-
mic broad peak at low temperature (70–100 °C) corresponds to the
desorption of physically adsorbed water. Maximum weight loss
due to the decomposition of carbon is in the range of 250–
400 °C. A weight loss of around 3 wt.% was observed in this tem-
perature range. Hence, temperatures 250, 300, 350, and 400 °C
were chosen for the regeneration of the catalyst.
Therefore, the deactivated catalyst was oxidatively treated at
four different temperatures with the aim to burn off carbon depos-
its and also to find out the optimum temperature for effective
regeneration. As mentioned earlier, after 42 h on-stream, the con-
version of toluene and the yield of BA were decreased from 54% to
18% and 48% to 15%, respectively. Thus, after testing for 42 h, the
reaction feed was stopped deliberately and the catalyst was regen-
erated at the above-mentioned temperatures for 2 h in air. The cat-
alytic performance of the solids after these regenerations is shown
in Fig. 3.
For example, PdSb-u42 was regenerated in air at 250 °C for 2 h
and then the reaction was continued under identical reaction con-
ditions in order to compare its performance with the fresh cata-
lyst. It is clear from the Fig. 3a that this catalyst (PdSb-u42-
r250) could immediately restore its maximum activity. It looks
like regeneration was effective at this temperature; however, this
sample again undergoes rapid deactivation with time-on-stream,
i.e., it is deactivated in this second acetoxylation run at much
shorter time-on-stream compared with the first run. This fact
suggests that the regeneration temperature (250 °C) was not suf-
ficient enough to effectively burn off coke and restore the activity.
3.2. Regeneration of 10Pd16Sb/TiO2 catalyst
As discussed above, a better understanding of deactivation and
regeneration processes is essential for improving and optimizing
the process conditions. As mentioned above, the major reason for
deactivation in the present case seems to be the formation of car-
bonaceous deposits on the catalyst surface. In addition to carbon
deposition, some other modifications of Pd surface such as valency
changes or formation of some stable molecular surface species are
also speculated.
Table 1
List of catalysts and their denotation used in the acetoxylation of toluene to benzyl
acetate and various characterizations.
No. Catalyst at different stages
Denotation
1
2
3
4
5
10Pd16Sb/TiO2-fresh
PdSb-f
10Pd16Sb/TiO2-(used-6 h)
10Pd16Sb/TiO2-(used-32 h)
10Pd16Sb/TiO2-(used-42 h)
PdSb-u6
PdSb-u32
PdSb-u42
10Pd16Sb/TiO2-(used-42 h)-Regeneration@250 °C/2 h/ PdSb-u42-r250
air
6
7
8
9
10Pd16Sb/TiO2-(used-42 h)-Regeneration@300 °C/2 h/ PdSb-u42-r300
air
10Pd16Sb/TiO2-(used-42 h)-Regeneration@350 °C/2 h/ PdSb-u42-r350
air
10Pd16Sb/TiO2-(used-42 h)-Regeneration@400 °C/2 h/ PdSb-u42-r400
air
10Pd16Sb/TiO2-(used-42 h)-Regeneration@300 °C/2 h/ PdSb-u42-
air(used-7 h)
10 10Pd16Sb/TiO2-(used-42 h)-Regeneration@350 °C/2 h/ PdSb-u42-
air(used-7 h) r350-u7
r300-u7

106
N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
ingly, it was found that this treatment completely failed to
restore activity. The sample showed poor performance (X-Tol
ꢀ18%, Y-BA ꢀ15%) as depicted in Fig. 3c. To confirm this result
further, we treated another deactivated catalyst even at higher
temperature of 400 °C, which led to poor performance as well
(Fig. 3d).
From the above observations, it can be concluded that catalyst
deactivation is reversible only if the catalyst is treated at the tem-
peratures of 250 and 300 °C. Obviously, higher temperature of
350 °C and above is not suitable to recover the catalyst perfor-
mance. The regeneration temperature of 250 °C can restore the
activity but not for a prolonged period of time. Among all the four
temperatures investigated, it is clear that 300 °C regeneration tem-
perature is most effective in restoring activity and also giving sta-
ble performance in a further catalytic run.
3.3. Characterization of fresh, deactivated, and regenerated 10Pd16Sb/
TiO2 catalysts
Fig. 2. TG–DTA curves of the deactivated 10Pd16Sb/TiO2 catalyst (PdSb-u42).
3.3.1. Carbon analysis
Therefore, another catalyst sample deactivated in an identical
manner was treated in a similar way but at higher temperature
of 300 °C (PdSb-u42-r300). In this case also, activity and selectiv-
ity of the catalyst were completely restored (Fig. 3b). Carbon anal-
ysis proved that the deposits were more or less completely
burned off (see Table 2). In addition, it has to be noticed that
the performance of the sample regenerated at 300 °C was much
more effective and stable compared with the catalyst regenerated
at 250 °C.
After acetoxylation runs on the catalyst PdSb-f, we observed
that the catalyst changed its color from brown to black, indicating
that the catalyst was obviously covered by coke. Therefore, carbon
content measurements were taken, and the relationship between
catalyst performance and the amount of coke present at different
stages of the reaction is shown in Fig. 4. It can be noticed that dur-
ing the formation period of 6 h (i.e., up to maximum activity), car-
bon already starts to deposit (about 0.6 wt.%, see Table 2).
However, such small amount of carbon in this stage is not large en-
ough to block the active sites and to decrease the activity. The car-
bon content is progressively increased further to about 2 wt.% after
32 h on-stream, but still the activity is maintained. Interestingly,
In order to further improve the effectiveness of regeneration
compared with those regenerated at 250 and 300 °C, another deac-
tivated sample was treated at 350 °C again for 2 h in air. Surpris-
Fig. 3. Catalyst performance and regeneration of the 10Pd16Sb/TiO2 at different temperatures for 2 h in air: (a) 250 °C, (b) 300 °C, (c) 350 °C, and (d) 400 °C.

N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
107
about 2 wt.% coke has only a marginal influence on the perfor-
mance. However, when the coke amounts further increase beyond
2%, the catalyst undeniably undergoes deactivation. As a result, a
sharp decrease in the conversion of toluene and yield of BA was
observed.
Carbon content was also determined for the four samples col-
lected after regeneration at 250, 300, 350, and 400 °C. Only the
data of the samples regenerated at 300 °C (PdSb-u42-r300) and
350 °C (PdSb-u42-r350) are given (see Table 2) because carbon
contents of the other samples are quite similar. Obviously, carbon
deposits can be removed from the catalyst at both the tempera-
tures by more than 80–90%. In contrast, XPS reveals that there is
more surface carbon in the sample regenerated at 350 °C than
one at 300 °C. This seems to be the reason for ineffective regener-
ation of deactivated solid at 350 °C. Then, second acetoxylation
runs were also carried out on the regenerated samples and the car-
bon content was analyzed again, which was found to increase con-
siderably in 6–7 h on-stream. The poor performance of the
catalysts regenerated at 350 °C and higher suggests that certain
structural changes or some other alterations such as valency
changes are probably taking place during higher temperature
treatment.
Fig. 5. Pore size distribution of 10Pd16Sb/TiO2 samples at different stages of
reaction.
3.3.3. X-ray diffraction
XRD patterns of the fresh (PdSb-f), maximum active (PdSb-u6),
and deactivated (PdSb-u42) catalysts along with the solids regen-
erated at 300 °C (PdSb-u42-r300) and 350 °C (PdSb-u42-r350) are
shown in Fig. 6a. The XRD patterns of all samples show reflections
of the anatase support, NaCl (which is formed by Na2CO3 and PdCl2
used during the preparation). In addition, a shift and a splitting of
metallic Pd reflections were observed compared with those in the
fresh catalyst during the course of reaction, which are shown in
Fig. 6b. It is known that pure metallic Pd reflections (Pd(1 1 1)) ap-
pear at 2h = 40.01° [PDF. No. 88-233] and are related to the cubic
structure with cell parameter of 3.905 Å. But in all diffractograms
except for that of the fresh sample, this peak is shifted to lower
2h values. In addition, XRD reflections of PdSb-u6 and the regener-
ated samples PdSb-u42-r300 and PdSb-u42-r350 show a splitting
of the Pd peak (see Fig. 6b). Such shift and splitting of the Pd peak
are due to the incorporation or formation of solid solution between
Pd and Sb or C, which was, however, explained earlier [21]. It is
known that such incorporated carbon can be removed with heating
in air below 300 °C [32]. The regenerated samples showed lattice
expansion with two separate reflections at 2h = 39.8 and 39.5°. This
clearly points to the incorporation of Sb in the Pd lattice. It must be
noted that the diffractograms of the most active (PdSb-u6) and the
regenerated samples are very similar. In contrast, the completely
deactivated PdSb-u42 showed a broad asymmetric reflection be-
low 2h = 40°. A deconvolution of this broad peak is not possible
due to the presence of various Pd species. The differences between
the Pd reflections in the spent sample (PdSb-u42) and both the
regenerated samples are mainly caused by the removal of carbon
during the regeneration. However, XRD provides no hint for the
poor performance of the samples regenerated at higher tempera-
ture (350 and 400 °C) in the subsequent catalytic run. This surpris-
ing result is clearly explained by other techniques as described
below.
3.3.2. BET surface areas and pore size distribution
The BET surface areas of fresh (PdSb-f), maximum active (PdSb-
u6), deactivated (PdSb-u42), regenerated at 300 °C (PdSb-u42-
r300) and 350 °C (PdSb-u42-r350), and used (second reaction
run) samples (PdSb-u42-r300-u7 and PdSb-u42-r350-u7) are pre-
sented in Table 2. A significant increase was observed from fresh
to maximum active catalyst, i.e., from 28.6 to 40.6 m2/g, respec-
tively. The surface area also changed to some extent during the
course of the reaction and even after regeneration, which remained
in the range of 40–32 m2/g. However, a dramatic decrease in sur-
face area from nearly 32 m2/g to 18–16 m2/g was observed in the
solids studied after the second reaction run. This suggests that cok-
ing of catalysts after regeneration step is much easier and/or other
structural alterations took place.
The pore size distribution of the deactivated, regenerated, and
used samples in the second reaction run after regeneration is quite
similar to that of the fresh catalyst (Fig. 5). However, the sample
regenerated at 350 °C shows a significant increase in pore volume;
it is nearly doubled for larger pores compared with the sample
regenerated at 300 °C. This might be a possible reason for the loss
in surface area.
Fig. 7a and b displays the diffractograms of the catalysts used in
a second acetoxylation run after regeneration at 300 and 350 °C
(PdSb-u42-r300-u7 and PdSb-u42-r350-u7). From these reflec-
tions, it can be seen that the shifted splitting of the Pd peak at
40.01° is significantly observed only in the sample treated at
300 °C for 2 h in air, whereas in the case of 350 °C, additional
reflections at 2h = 40.2° and 40.6° are related to a definite interme-
tallic compound (i.e., Pd20Sb7 phase [PDF. No. 31-102]) appeared
after second run. It is known that the formation of this alloy is
detrimental to the activity [21]. As such, from the above observa-
tions, it can be stated that bulk compositions are not determining
the performance of the catalyst. Near-surface region properties
Fig. 4. Correlation between the catalyst performance and the coke formation with
time-on-stream.

108
N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
Fig. 6. (a) XRD patterns of samples of fresh (PdSb-f), maximum active (PdSb-u6), deactivated (PdSb-u42), and regenerated 10Pd16Sb/TiO2 catalysts (PdSb-u42-r300 and
PdSb-u42-r350). (b) XRD patterns in the 2h range of the Pd(1 1 1) reflection (38–41°).
Fig. 7. (a) XRD patterns of 10Pd16Sb/TiO2 samples used in a second reaction run after regeneration at 300 and 350 °C (PdSb-u42-r300-u7 and PdSb-u42-r350-u7,
respectively). (b) XRD patterns in the 2h-range of the Pd(1 1 1) reflection (38–41°).
probably play an important role and are monitoring the activity of
the catalysts.
a more important role in determining the fate of the regenerated
catalysts in the acetoxylation reaction.
3.3.5. X-ray photoelectron spectroscopy
3.3.4. Transmission electron microscopy
The XP spectra of the fresh sample, after 6 h on-stream (PdSb-
u6), the deactivated sample (PdSb-u42), and two regenerated sam-
ples (i.e., one at 300 °C (PdSb-u42-r300) and the other at 350 °C
(PdSb-u42-r350)) are displayed in Fig. 9. The Pd binding energy
of the fresh sample was found to be 335.8 eV. This value corre-
sponds to that of oxidic Pd [34]. During the reaction, generation
of other Pd species was also observed (around 335 eV), which cor-
responds to metallic Pd [35] and PdO (336.5–337.5 eV) [34] Both
Pd0 and PdO phases were present until the catalyst showed stable
performance from 6 h up to 32 h on-stream (X-Tol = 54% and Y-
BA = 48%), while the PdO peak vanished in the deactivated catalyst
(where X-Tol = 18%, Y-BA = 15%; PdSb-u42).
In addition, spectra of catalysts used in a second reaction run for
additional 7 h after regeneration at above two temperatures (PdSb-
u42-r300-u7 and PdSb-u42-r350-u7) are shown. After 6 h on-
stream, in which the catalyst reaches its maximum performance,
peaks corresponding to both the metallic Pd0 and PdO were ob-
served. This means that during the first few hours of reaction,
reduction is occurring to a certain extent. As the reaction proceeds,
a further gradual reduction of PdO species was seen. After 42 h on-
stream, the catalyst is deactivated and PdO species completely
vanished from the surface, while only the metallic Pd0 species re-
mained. From these observations, it can be concluded that the
Fig. 8a–d shows electron micrographs of deactivated (PdSb-
u42) and regenerated catalyst samples. The deactivated samples
display spherical-shaped particles with the size varying in the
range of 50–100 nm. TEM images of the samples regenerated at
different temperatures also reveal that the shape and size of most
of the particles remain unchanged throughout irrespective of ther-
mal treatment given, i.e., no effect on the size and morphology was
observed. However, restructuring of particle with tri or hexagonal
array was observed only in the sample after regeneration at 400 °C.
From previous studies [19], it is known that one of the reasons
for high catalytic performance is an optimum particle size of Pd. It
was found recently [17,33] that it needs to attain some critical size
(50–100 nm) to show the best performance with good long-term
stability. However, in the present series, all regenerated catalysts
have almost the same particle size as that of the catalyst during
the reaction (50–100 nm). These observations suggest that the size
of the particles cannot be the reason for the observed differences in
their behavior after regenerations. Even regeneration treatments of
300–400 °C for 2 h in air revealed no influence on the particle size.
Therefore, we can say that the poor performance of the catalysts
regenerated at higher temperature has nothing to do with the par-
ticle size of the Pd of the catalyst. Surface properties probably play

N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
109
Fig. 8. Transmission electron micrographs of 10Pd16Sb/TiO2 samples after (a) deactivation (PdSb-u42) and regeneration at (b) 300 °C (PdSb-u42-r300), (c) 350 °C (PdSb-u42-
r350), and (d) 400 °C (PdSb-u42-r400) for 2 h in air.
Fig. 10. Relative surface concentration of Pd0 and PdO in 10Pd16Sb/TiO2 samples at
different stages of reaction (PdSb-u6, PdSb-u42, PdSb-u42-r300, and PdSb-u42-
r350) analyzed by XPS.
Fig. 9. XP spectra of the Pd 3d peak of 10Pd16Sb/TiO2 samples at different stages of
the reaction (PdSb-u6, PdSb-u42, PdSb-u42-r300, and PdSb-u42-r350) and these
samples used in a second reaction run after regeneration PdSb-u42-r300-u7 and
PdSb-u42-r350-u7).
treatment is not suitable to regenerate the initial concentration
of PdO on the surface. Additionally, we also estimated the surface
concentration of Pd0 and PdO species in these samples (Fig. 10),
which again gave clear hints that the presence of PdO in addition
to coke removal is essential to get back good performance. It is
clear from this estimation that the sample regenerated at 350 °C
only shows a Pd0/PdO surface ratio of 9:1, while the sample regen-
erated at 300 °C reveals a ratio of 6:4. On the other hand, the max-
imum active sample (PdSb-u6) shows a ratio of 4.5:5.5. It is very
interesting to note that even though some PdO species were found
on the surface of the catalyst treated at 350 °C, the catalyst activity
did not return to its initial level. This suggests that not just the
presence of both the Pd0 and PdO species on the surface is vital for
the best performance of the catalyst. Especially for the spent cata-
lysts, certain changes in the peak shape is noticed, i.e., significant
broader peaks were observed than for the regenerated samples,
indicating a higher chemical heterogeneity in the surrounding of
the Pd atoms.
XP spectra of samples after regeneration at 300 and 350 °C for
2 h in air (PdSb-u42-r300 and PdSb-u42-r350) again showed PdO
species, which were lost in the deactivated sample. However, the
spectrum of the catalyst regenerated at 350 °C shows that this

110
N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
Fig. 11. (a) Surface Sb/Ti atomic ratio in 10Pd16Sb/TiO2 samples at different stages of the reaction (PdSb-f, PdSb-u6, PdSb-u42, PdSb-u42-r300, and PdSb-u42-r350) analyzed
by XPS. (b) Surface Pd/Ti atomic ratio in 10Pd16Sb/TiO2 samples at different stages of the reaction (PdSb-f, PdSb-u6, PdSb-u42, PdSb-u42-r300, and PdSb-u42-r350) analyzed
by XPS. Surface atomic ratios were calculated from the areas of the XPS Sb3d3/2, Pd3d, and Ti2p peaks.
existence of PdO and Pd0 species but their optimum ratio seems to
be essential for the high performance of the catalyst.
mum combination of these three parameters is essential for main-
taining the good and stable performance of the catalyst.
The C1s spectra of all spent and regenerated catalysts consist
mainly of two peaks at 284.8 eV and 289.5 eV. The former one is
originated from carbon atoms bonded to other C atoms or H atoms,
while the latter one is correlated with carboxylic groups. As ex-
pected, the highest carbon amount was found for the deactivated
sample after 42 h on-stream. The regeneration at 300 °C has shown
a significant decrease in the carbon amount in the near-surface re-
gion, but surprisingly, it is not that effective for the one regener-
ated at 350 °C in air. Possible explanation for this unexpected
behavior at such higher temperature treatment is the formation
of stable carbon surface species (e.g., Pd carbides) or a possible
transformation of Pd surface structures toward those stabilizing
adsorbed carbon. This high amount of deposited carbon explains
the low activity of this sample regenerated at 350 °C.
Sb concentration on the surface of fresh (PdSb-f), maximum ac-
tive (PdSb-u6), deactivated (PdSb-u42), and regenerated solids
(PdSb-u42-r300 and PdSb-u42-r350) were also studied using
XPS. The spectra revealed that Sb is present in oxidized form, evi-
denced by a Sb 3d3/2 peak with a binding energy of 540 eV, being
characteristic for Sb2O3 [36]. From the surface Sb/Ti atomic ratios
(Fig. 11a), it is evident that the maximum active sample contains
the highest Sb proportion at the surface compared with deacti-
vated and regenerated samples. Interestingly, in the sample regen-
erated at 350 °C, the Sb concentration in the near-surface region
could not be restored considerably. In other words, the amount
of Sb regained after regeneration at 350 °C is only about 40% of
the surface Sb present in the sample regenerated at 300 °C. Proba-
bly, this deficit of surface Sb would be related to the small propor-
tion of PdO in the sample regenerated at 350 °C (PdSb-u42-r350).
A similar trend was also observed in the case of Pd/Ti surface ra-
tio (Fig. 11b). The surface Pd amount was found to be dramatically
reduced in the deactivated catalyst (PdSb-u42) compared with the
one at maximum activity (PdSb-u6). This loss of surface Pd could
not be restored adequately after regeneration. This might be a fur-
ther reason for faster deactivation and surface area loss after sec-
ond catalytic run. Here, one should note that elemental analysis
by ICP-OES of the spent catalysts at different stages, deactivated
and regenerated catalyst did not show any loss of Pd or Sb from
the catalyst. Therefore, the possibility of leaching of Pd or Sb from
the catalyst can be ruled out. In summary, the catalytic perfor-
mance is related to the surface concentration of Pd and Sb and
the amount of deposited coke. On the whole, it can be said the opti-
4. Conclusions
10Pd16Sb/TiO2 catalyst was used for the gas-phase acetoxyla-
tion of toluene showing best performance of X-Tol = 54% and Y-
BA = 48% after a formation period of 6 h. Despite this, after nearly
30 h, on-stream catalyst loses performance and reveals typical
deactivation behavior. Regeneration was carried out in air for 2 h
at four different temperatures (250–400 °C). Treatment at 250
and 300 °C was found to be suitable to regain catalyst performance.
Higher temperature treatment (350 and 400 °C) was not suitable
for regeneration, and hence, the catalyst did not regain the activity
lost during the deactivation. Carbon analysis yielded a more or less
effective removal of coke deposits after regeneration. However, the
samples regenerated at higher temperatures remained inactive in a
second catalytic run. XRD patterns revealed a splitting and shifting
of Pd(1 1 1) to lower 2h values. This behavior seems to be caused
by the interaction of Pd with surrounding Sb and/or Ti of the car-
rier. However, no real reason for the regeneration properties can
be seen. TEM studies revealed that there is no change in Pd particle
size after reaching the optimum of 50–100 nm during formation
period. Detailed XPS characterizations showed that an optimum
amount of PdO and Pd0 on the surface is necessary for the better
performance of the catalyst. The PdO deficit in the sample regener-
ated at 350 °C seems to be mainly responsible for the poor perfor-
mance after regeneration. This deficit is due to the migration of Pd
to deeper layers as discovered by XPS; in addition, the surface Sb
proportion also seems to be lowered in those regenerated catalysts.
Thus, the existence of Pd0, PdO, and Sb synergy on the catalyst sur-
face is needed to maintain a stable and high performance of the
catalyst.
Acknowledgments
The authors like to thank Leibniz-Institut für Katalyse e.V. and
Deutsche Forschungsgemeinschaft (Grant No. BR1380/13-2) for
the financial support. Authors also thank Mr. Samudra Sengupta,
Pennsylvania State University, USA, for carrying out TEM analysis
and our analytical staff, especially Dr. M. Schneider, at LIKAT for
CHN and XRD analyses.

N. Madaan et al. / Journal of Catalysis 282 (2011) 103–111
111
[19] J. Radnik, A. Benhmid, V.N. Kalevaru, M.-M. Pohl, A. Martin, B. Lücke, U.
Dingerdissen, Angew. Chem. Int. Ed. 44 (2005) 6771.
References
[20] V.N. Kalevaru, A. Benhmid, J. Radnik, B. Lücke, A. Martin, J. Catal. 243 (2006) 25.
[21] S. Gatla, N. Madaan, J. Radnik, V.N. Kalevaru, M.M. Pohl, B. Lücke, A. Martin, A.
Brückner, Appl. Catal. A: Gen. 398 (2011) 104.
[22] C.H. Bartholomew, Appl. Catal. A: Gen. 212 (2001) 17.
[23] U. Lassi, PhD Thesis, Department of Process and Environmental Engineering,
University of Oulu, 2003.
[24] P. Albers, J. Pietsch, S.F. Parker, J. Mol. Catal. A: Chem. 173 (2001) 275.
[25] C. Guéret, F. Billaud, B. Fixari, P. Le Perchec, Carbon 33 (1995) 159.
[26] J.S. Conroy, R.S. Slysh, D.B. Murphy, C.R. Kinney, in: Third Proc. Conf. Carbon,
1959, p. 395.
[27] H.G. Karge, W. Nießen, H. Bludau, Appl. Catal. A: Gen. 146 (1996) 339.
[28] M.A. Aramendía, V. Boráu, I.M. García, C. Jiménez, A. Marinas, J.M. Marinas, F.J.
Urbano, J. Mol. Catal. A: Chem. 151 (2000) 261.
[1] K. Weissermel, H. Arpe, Industrial Organic Chemistry, fourth ed., Wiley-VCH,
Weinheim, 2003.
[2] J.J. Moiseev, M.N. Vargaftik, J.K. Sykrin, Doklady Akad. Nauk SSSR 133 (1960)
21.
[3] G. Roscher, Ullmann’s Encyclopedia of Industrial Chemistry, Vinyl Esters, sixth
ed., vol. 38, Wiley-VCH, Weinheim, 2003.
[4] K. Bauer, D. Garbe, H. Surburg, Ullmann’s Encyclopaedia of Industrial
Chemistry, vol. A 11, VCH, Weinheim, 1988.
[5] W.D. Provine, P.L. Mills, J.J. Lerou, Joe W. Hightower, W.N. Delgass, E. Iglesia,
A.T. Bell, Stud. Surf. Sci. Catal. 101 (1996) 191.
[6] D.R. Bryant, J.E. McKeon, B.C. Ream, Tetrahedron Lett. 9 (1968) 3371.
[7] D.R. Bryant, J.E. McKeon, B.C. Ream, J. Org. Chem. 33 (1968) 4123.
[8] H. Shinohara, Appl. Catal. 10 (1984) 27.
[9] J.M. Davidson, C. Triggs, J. Chem. Soc. A: Inorg. Phys. Theor. (1968) 1331.
[10] S.K. Tanielyan, R.L. Augustine, J. Mol. Catal. 90 (1994) 267.
[11] S.K. Tanielyan, R.L. Augustine, J. Mol. Catal. 87 (1994) 311.
[12] E. Benazzi, C.J. Cameron, H. Mimoun, J. Mol. Catal. 69 (1991) 299.
[13] L. Eberson, E. Jonsson, Acta Chem. Scand. B 28 (1974) 771.
[14] L. Eberson, L.G. Gonzalez, Acta Chem. Scand. 27 (1973) 1249.
[15] T. Komatsu, K. Inaba, T. Uezono, A. Onda, T. Yashima, Appl. Catal. A: Gen. 251
(2003) 315.
[29] S.D. Jackson, I.J. Huntingdon, N.A. Hussain, S.R. Watson, Stud. Surf. Sci. Catal.
126 (1999) 453.
[30] P. Albers, K. Seibold, G. Prescher, H. Müller, Appl. Catal. A: Gen. 176 (1999) 135.
[31] G. Suresh, J. Radnik, V.N. Kalevaru, M.-M. Pohl, M. Schneider, B. Luecke, A.
Martin, N. Madaan, A. Brueckner, Phys. Chem. Chem. Phys. 12 (2010) 4833.
[32] M. Maciejewski, A. Baker, Pure Appl. Chem. 67 (1995) 1879.
[33] V.N. Kalevaru, A. Benhmid, J. Radnik, M.M. Pohl, B. Lücke, A. Martin, Catal.
Today 141 (2009) 317.
[34] NIST Standard Reference Database 20, V. 3.5, <http://srdata.nist.gov/xps>.
[35] H. Gabasch, W. Unterberger, K. Hayek, B. Klotzer, E. Kleimenov, D. Teschner, S.
Zafeiratos, M. Havecker, A.K. Gericke, R. Schlogl, J. Han, F.H. Ribeiro, B.A. Kiss, T.
Curtin, D. Zemlyanov, Surf. Sci. 600 (2006) 2980.
[16] A. Benhmid, K.V. Narayana, A. Martin, B. Lücke, S. Bischoff, DE. Pat., 10 2004
002 262 A1, 2005.
[17] A. Benhmid, K.V. Narayana, A. Martin, B. Lücke, S. Bischoff, M.M. Pohl, J. Radnik,
M. Schneider, J. Catal. 230 (2005) 420.
[18] A. Benhmid, K.V. Narayana, A. Martin, B. Lücke, M.M. Pohl, Catal. Today 112
(2006) 192.
[36] R. Delobel, H. Baussart, J.M. Leroy, J. Grimblot, L. Gengembre, J. Chem. Soc.
Faraday Trans. 179 (1983) 879.