Palladium-modified catalytic membranes for the direct synthesis of H 2O2: Preparation and performance in aqueous solution

Article abstract of DOI:10.1016/j.jcat.2005.07.024

A series of new tubular catalytic membranes (TCMs) have been prepared and tested in the direct synthesis of H₂O₂. Such TCMs are carbon-coated asymmetric α-alumina mesoporous membranes supported on macroporous α-alumina. Pd was introduced by deposition-precipitation to obtain an even Pd particle distribution inside the membrane pore network. This type of membrane was active in the direct synthesis of H₂O ₂. Catalytic tests were carried out in a semi-batch recirculating reactor under very mild conditions. Concentrations as high as 250-300 ppm H ₂O₂ were commonly achieved after 6-7 h on stream, with a particularly high decomposition rate in the presence of H₂. Important features are optimization of the metal deposition procedure and preactivation. To slow down the decomposition and favor the synthesis of H₂O ₂, a smooth metal particle surface is needed.

Full text of DOI:10.1016/j.jcat.2005.07.024

Journal of Catalysis 235 (2005) 241–248
www.elsevier.com/locate/jcat
Palladium-modified catalytic membranes for the direct synthesis of H2O2:
preparation and performance in aqueous solution
S. Melada a,∗, F. Pinna a, G. Strukul a, S. Perathoner b, G. Centi b
a
Dipartimento di Chimica, University Ca’ Foscari, INSTM Consortium R.U. of Venice, Dorsoduro 2137, 30123 Venice, Italy
b
Dipartimento di Chimica Industriale ed Ingegneria dei Materiali, University of Messina, INSTM Consortium R.U. of Messina, Salita Sperone 31,
98166 Messina, Italy
Received 20 May 2005; revised 20 July 2005; accepted 28 July 2005
Abstract
A series of new tubular catalytic membranes (TCMs) have been prepared and tested in the direct synthesis of H O . Such TCMs are
2
2
carbon-coated asymmetric α-alumina mesoporous membranes supported on macroporous α-alumina. Pd was introduced by deposition–
precipitation to obtain an even Pd particle distribution inside the membrane pore network. This type of membrane was active in the direct
synthesis of H O . Catalytic tests were carried out in a semi-batch recirculating reactor under very mild conditions. Concentrations as high
2
2
as 250–300 ppm H O were commonly achieved after 6–7 h on stream, with a particularly high decomposition rate in the presence of H .
2
2
2
Important features are optimization of the metal deposition procedure and preactivation. To slow down the decomposition and favor the
synthesis of H O , a smooth metal particle surface is needed.
2
2
2005 Elsevier Inc. All rights reserved.
Keywords: Hydrogen peroxide; Direct synthesis; Membrane; Catalytic membrane reactor; Palladium; Heterogeneous catalysis
1. Introduction
Au, Ag, Ni) supported catalysts [3–9], homogeneous cata-
lysts [10], electrochemical devices [11], and, only recently,
Pd-based monoliths [12], membranes [13,14], and fast flow
mixers [15,16].
Only a few related scientific works have been published
in the open literature. Most of these describe the use of sup-
ported Pd or Pd alloy catalysts [17–22], colloidal Pd clusters
[23–26], fuel cells [27], and Au catalysts [28,29] for the di-
rect synthesis of hydrogen peroxide. Among these works,
only one paper deals with membranes [30].
Despite the intense patent activity of major chemical
companies, no successful industrial application has been
claimed. This is because two severe drawbacks must be over-
come for the direct synthesis: (1) the formation of explosive
H2/O2 mixtures that should be avoided and (2) the selec-
tivity of the reaction (whose main product is water) that is
still unacceptably low. In principle, the use of specially de-
signed catalytic membranes could overcome both problems
[13,14,30].
Hydrogen peroxide is a commodity that today is almost
exclusively produced by the anthraquinone process, devel-
oped in Germany in the 1930s [1]. This process is eco-
nomically feasible only on large-scale plants, and the price
of H2O2 produced is influenced by the complexity of the
process and by the costly separation and concentration steps
that are needed. Direct synthesis, which could be a more eco-
nomic and environmentally acceptable alternative, has been
known since the beginning of the 20th century [2], but no
industrial application has been found, even given some suc-
cessful attempts at DuPont during the 1980s [3].
Numerous patents have been filed over the past 30 years
dealing with Pd metal or Pd metal alloy (e.g., Pd/X, X = Pt,
*
Corresponding author.
E-mail address: stefano.melada@unive.it (S. Melada).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.07.024
2005 Elsevier Inc. All rights reserved.

242
S. Melada et al. / Journal of Catalysis 235 (2005) 241–248
2. Experimental
analysis. CO+ fragments (m/z = 28) were used for quantita-
tive measurements. A 1/2 CO/Pd chemisorption stoichiom-
etry was assumed for calculation [34]. Calibrations were
carried out after each measurement by injecting a known
amount of CO from a calibrated loop. A specially shaped
sample holder was used to analyze the TCM as a whole. Be-
cause CO is strongly chemisorbed on Pd, all samples were
characterized after the catalytic tests. A methanation test
showed that CO is removed only above 300 ◦C. After the
methanation step, Pd particles sintered extensively. Before
each measurement, Pd was reduced in situ by passing a 5%
H2/He mixture at 25 ◦C, then thoroughly evacuated with He.
Pd loading was determined by atomic absorption spec-
troscopy (AAS) from the Pd(II) solutions used for metal
deposition on the membranes. Differences between Pd con-
centration before and after impregnation gave the amount
of loaded Pd. The reliability of this analytical method was
checked as follows. Pd was extracted from membranes by
soaking in boiling HNO3. AAS determination on dissolved
Pd(II) gave the loaded Pd amount. Differences of <3% were
obtained by comparing the two analytical methods. Thus
differential measurements on Pd solutions enabled determi-
nation of the Pd loading without destroying the membrane.
Membranes were 10-cm-long asymmetric α-Al2O3 tubu-
lar supports, externally coated with a synthetic carbon layer.
The carbon loading was in the range of 50–100 mg for each
tube. The coating was performed by MAST Carbon Ltd.,
Guildford, UK, and the alumina support was supplied by
Hermsdorfer Institut für Technische Keramik, Hermsdorf,
Germany.
2.1. Sample preparation
Carbon-coated membranes (CAMs) were activated in
CO2 at 850 ◦C and then impregnated by a deposition–
precipitation method (a classical method to obtain eggshell-
type catalysts) optimized during this work [31–33]. This
technique consisted of two steps: (1) basification of the
membrane surface by soaking in a NaOH solution (0.1 mol)
and (2) deposition of Pd(OH)2 by precipitation from an
2−
acidic PdCl4 solution. The pH of the starting Pd(II) solu-
tion changed from 0.7 to 2.7. During impregnation, Pd(OH)2
was deposited into the pore network of the external carbon
layer. After Pd deposition, membranes were dried at room
temperature, reduced at room temperature in H2 flow, and
washed with distilled water to remove chloride ions. Pd load-
ing was between 1 and 2 wt% with respect to carbon loading.
2.3. Catalytic tests
These were carried out in a semi-batch recirculation
reactor (Fig. 1), with the membrane sealed in a tubular
holder. From the inner side, H2 was fed at constant pres-
sure (2–5 bar) while an oxygen-saturated acidic solution was
continuously circulated on the outer side of the membrane
(where Pd was deposited) by means of a peristaltic pump
(25 ml/min) equipped with special Tygon® MH tubing. The
circulating solution was 100 ml of 0.03 mol H2SO4 contain-
ing 6 ppm of NaBr. H2O2 concentration was determined by
permanganometric titration.
2.2. Characterization
Membranes were thoroughly characterized by several
techniques. Transmission electron microscopy (TEM) im-
ages were taken with a Jeol 3010, operating at 300 kV,
equipped with a Gatan slow-scan CCD camera (model
794) and an Oxford Instrument EDS microanalysis detec-
tor (model 6636). External carbon layers have been scraped
away with a knife to give a fine carbon powder. This was
suspended in isopropyl alcohol, ultrasonicated for 5–10 min
so that the particles were well dispersed and then deposited
on a holey carbon film.
Scanning electron microscopy (SEM) images were taken
with a Jeol JSM 5600 LV (low-vacuum) microscope. The
membrane was gently broken in several chips (10–20 mm)
that were subsequently attached to a support with a conduc-
tive glue and put in the microscope chamber.
N2 physisorption isotherms were determined on small
chips of the CAMs. Samples were degassed in vacuo
(10−2 Torr) and heated at 150 ◦C for 2 h. N2 physisorption
isotherms were measured at −196 ◦C on a Micromeritics
ASAP 2000 instrument. Specific surface areas were deter-
mined by the BET equation in the 0.05–0.2 p/p0 range. Pore
size distributions were calculated with the BJH method.
CO chemisorption measurements were carried out with a
pulse technique on a home-made apparatus equipped with a
thermostatted reactor and an ESS Genesys quadrupole mass
spectrometer interfaced to a computer for data collection and
Fig. 1. Scheme of the experimental apparatus used for the catalytic tests.

S. Melada et al. / Journal of Catalysis 235 (2005) 241–248
243
3. Results and discussion
Some of the SEM images of the carbon coated asym-
metric α-alumina catalytic membranes used in this work are
shown in Figs. 2a–c. These images show the asymmetric
structure of the membranes in details. Carbon coating does
not form an external overlayer, but penetrates the meso-
porous external α-alumina layer, giving rise to a microp-
orous network inside the voids left by the Al2O3 particles.
The presence of both mesoporosity and microporosity
is clearly demonstrated by physisorption measurements. In
fact, the N2 adsorption–desorption isotherms (two typical
samples of which are given in Fig. 3) show that the nature
of the carbon coating is mainly microporous, but the whole
TCM exhibited a significant contribution from mesoporos-
ity and macroporosity, reflecting the porous structure of the
original Al2O3 particles. Consequently, a bimodal pore size
distribution is observed, associated with both the micro-
porous/mesoporous outside carbon coating structure (pore
size <10 nm) and the coated macroporous alumina support
skeleton (pore size >10 nm).
(a)
The starting α-alumina supports have a specific surface
area of <1 m2/g and a mean pore size of 0.3 µm (mea-
sured by mercury intrusion porosimetry), whereas the exter-
nal membrane layer has a mean pore size of 100 nm. This is a
consequence of the nature of the starting material, α-Al2O3,
which is almost nonporous, so that the main pore network
of the tubular support is generated by the interparticle voids.
The low surface area of the starting material is then negli-
gible with respect to the final carbon-coated TCMs’ specific
surface areas, so that the values obtained by BET analysis of
the N2 adsorption isotherms pertain almost exclusively to the
carbon coating. The surface area of the whole TCMs ranged
from 600 to about 800 m2/g for all measured samples (see
Table 1). This high surface area explains the high metal dis-
persion obtained with the deposition technique used in this
work.
(b)
Deposition–precipitation was chosen as the preferred
metal deposition method because of all the tested meth-
ods (including wet impregnation, homogeneous deposition–
precipitation, and deposition–reduction), it gives the best
results in term of Pd loading and reproducibility. It also has
other useful features; it is simple and reliable, cheap, ap-
plicable to both inside and outside coated membranes, and,
last but not least, easy to scale up.
Optimization of the deposition parameters [namely Pd(II)
concentration and pH] was also carried out. pH was varied in
the range 0.7–2.7 [Pd(OH)2 starts precipitating at pH 3], and
Pd(II) concentration was varied in the range 40–400 ppm.
The procedure was carried out at room temperature. Re-
producibility in metal deposition was also reflected in the
reproducibility of kinetic runs; that is, samples with similar
metal loading and dispersion gave very similar H2O2 pro-
ductivity (see below).
(c)
Fig. 2. SEM picture of a TCM sections (a) details of the asymmetric
α-alumina meso–macroporous structure (low magnification); (b) details of
the structure of the mesoporous α-alumina layers (intermediate magnifi-
cation); (c) details of the presence of the carbon loading on the external
α-alumina layer. The carbon layer covers the α-alumina particles giving
rise to a micropores network (high magnification).
The best compromise between Pd loading and dispersion
was reached at a Pd(II) concentration of 40 ppm and a pH

244
S. Melada et al. / Journal of Catalysis 235 (2005) 241–248
Fig. 3. N adsorption isotherms and BJH pore size distributions for two chosen TCM’s.
2
Table 1
Surface features, impregnation parameters and CO chemisorption data
Sample
BET SSA
2
Impregnation procedure
CO chemisorption
2
II
II
(m /g
)
carb
Pd conc. (ppm)
Pd pH
Pd
loaded
(mg)
A
(m /g
)
D (nm)
Pd
Disp. (%)
Pd
Pd
Pd
Pd
TCM1
TCM2
TCM3
TCM4
TCM5
TCM6
TCM7
TCM8
TCM9
TCM10
TCM11
748
–
584
–
845
–
634
–
400
40
1.7
4.1
4.9
1.9
1.7
1.1
1.0
1.0
1.6
0.9
1.4
1.1
15
27
16
46
26
50
49
48
65
78
91
28
16
26
4
7
4
12
7
13
13
13
17
21
24
2.7
1.7
9.0
15.5
8.3
8.5
8.6
6.4
5.4
4.6
0.7
674
–
636
of 0.7. Higher Pd concentrations led to the formation of an
external thin Pd film that is possibly scraped away during
washing and reduction steps, whereas a higher pH led to
samples with lower catalytic activity. These observations are
summarized in Fig. 4. The maximum Pd loading obtained by
this technique was about 2 wt% with respect to the weight of
the carbon coating. The optimized deposition–precipitation
technique is described in the Experimental section. Surface
features and impregnation results are reported in Table 1.
CO chemisorption measurements were carried out to
measure the metal dispersion (see Table 1) obtained by the
impregnation procedure. The effect of the Pd particle diame-
ter on the catalytic activity can be ascertained by comparing
the metal particle size of the different samples with the con-
centration of H2O2 obtained by the same samples in the
catalytic experiments. As shown in Fig. 5, 8-nm Pd parti-
cles have the best catalytic performance.
Either smaller (4–5 nm) or larger (15–16 nm) Pd particles
show lower overall catalytic activity. In the case of small
particles, this behavior may be ascribed to the presence of
highly energetic sites that can readily chemisorb O2 disso-
ciatively. This is supposed to be the first step in producing
water [35]. In fact, atomically adsorbed oxygen could react
with dissolved H2 or H+ to give hydroxyl species, which
eventually produce water. Moreover, highly energetic sites
are likely to decompose H2O2. Hydrogen peroxide can dis-
sociatively chemisorb onto these sites through cleavage of
the HO–OH bond. The produced surface HO species should
then lead to water [36]. In the case of large particles, the low
dispersion (lower number of active sites) can easily explain
the low catalytic activity.
Some images showing Pd particles of small, medium, and
large size are collected in Figs. 6a–c. Fig. 6a shows a small
Pd particle, which is an oval-shaped monodomain Pd crys-
tallite. About the same features were found for the 8-nm

S. Melada et al. / Journal of Catalysis 235 (2005) 241–248
245
Fig. 4. Pd loading as a function of the Pd(II) concentration in the starting
solution at pH 0.7.
Fig. 5. Effect of the Pd particles diameter on the catalytic activity.
particles, but in this case the particle shape was more round
(Fig. 6b). A large and irregular particle is shown in Fig. 6c;
this is actually a cluster of small Pd crystallites with super-
imposed diffraction patterns.
All samples were tested for H2O2 synthesis in acidic
aqueous solution. H2 conversion rates are not reported be-
cause of the difficulties in measuring the actual H2 con-
sumption in an open system like the present one. No reliable
values for H2 mass balance could be determined.
The effect of the mean particle diameter on H2O2 synthe-
sis and H2O2 decomposition is shown in Fig. 7. This image
gives the catalytic profiles of three representative samples
prepared from similar membranes but with differing parti-
cle diameters. The sample with the mean pore diameter in
the optimum range (8 nm) shows the highest catalytic activ-
ity and the lowest H2O2 decomposition rate. Samples with
Pd particles either smaller or larger than the optimum di-
ameter have a decreased ability to synthesize H2O2, mainly
because they have much higher decomposition rates. Thus

246
S. Melada et al. / Journal of Catalysis 235 (2005) 241–248
H2O2 concentration was as high as 200 ppm. Thus preoxi-
dation was effective in improving the catalytic performance
of the Pd particles.
This pretreatment did not lead to a bulk metal oxidation,
because under the experimental conditions, only the particle
surface should be covered with a monolayer of chemisorbed
oxygen. This seems to be a necessary condition for improv-
ing the catalytic activity. Centi et al. [37], based on very
preliminary results, stated that a reduced Pd surface was nec-
essary for valuable catalytic activity. Indeed, the contradic-
tion with the earlier reported results is apparent, because the
active sites for H2O2 formation are Pd0 centers in particles
with high oxygen coverage, also making the less energetic
sites available for reaction.
In a third test, the used membrane was reoxidized in the
same way as before, and then a new catalytic run was started.
After 1 h the oxygen stream was stopped, so that the mem-
brane remained in contact only with H2. As it can be seen
in Fig. 8 (represented by squares), in the absence of O2 flow
there was residual catalytic activity due to the residual dis-
solved oxygen that led to a small increase in H2O2 concen-
tration. As soon as O2 saturation was restored, the catalytic
activity was regained, but the initial H2O2 rate was not re-
covered. The expected trend is indicated by a dotted line in
Fig. 8. This seems to indicate an irreversible change in the
state of the Pd particles due to exposure to H2 in the absence
of O2. If after 420 min the catalyst was surface-reoxidized
following the foregoing procedure, then the initial activity
was fully recovered. Reproducibility tests demonstrated that
on reoxidation, the membranes could be used for several cat-
alytic cycles without loss of activity.
Fig. 7. Effect of the particle diameters on the catalytic performances:
Q, sample with mean particle diameter of 8 nm (TCM6); ", 5 nm (TCM11);
2, 16 nm (TCM5).
In earlier work [30] we suggested a mechanism in which
molecularly adsorbed O2 on a nondefective Pd surface is
responsible for H2O2 formation. The effect of metal parti-
cle size reported earlier in the present paper supports this
view. The preoxidation effect reported herein could be inter-
preted in the same way, because it is reasonable that surface
oxidation will be directed not only on the most energetic
(defective) sites where chemisorption occurs dissociatively,
but also on nondefective sites. This allows, at least initially,
better production of H2O2. Whether or not this effect is per-
manent cannot be determined without selectivity data, which
cannot be reliably measured in the present system.
A long-duration test was also carried out on the TCM8
sample. As can be seen from Fig. 9, the TCM was kept in
the reactor for 32 h, and a maximum H2O2 concentration of
410 ppm was achieved after 27 h, after which the concen-
tration began to slowly decrease. Again, after a reoxidation
treatment, the catalyst could be reused several times without
any significant loss of its original activity.
Fig. 8. Effect of the metal pre-oxidation on the catalytic activity of TCM7:
Q, pre-oxidized; !, non-oxidized; 2, O flow stopped for 2 h; dashed line,
2
expected trend after O flow restored.
2
the presence of highly energetic sites, found abundantly on
very small and large particles, is responsible for the fast
decomposition rate, which decreases the overall ability to
produce H2O2.
An important point concerns sample pretreatment. Be-
fore kinetic tests, samples were preoxidized by flowing an
O2-saturated solution over the catalyst while at the same
time replacing H2 with an inert gas (N2) from inside the
membrane. This preoxidation step took 4 h; then the mem-
brane was dried in N2 flow, and a kinetic run (H2 from
inside) was carried out. The effect of the (surface) Pd ox-
idation state on the catalytic activity is shown in Fig. 8.
Non-preoxidized samples show moderate catalytic activity
(empty circles) reaching about 130 ppm H2O2 after 7 h on
stream. With preoxidized samples (represented by the trian-
gles in Fig. 8), catalytic activity was improved, and the final
To study the effect of H2 partial pressure, a diluted H2/N2
gas mixture was used during the kinetic tests. The results of
kinetic tests with a 5% H2/N2 mixture are shown in Fig. 10.
As can be seen, with a very dilute H2 gas phase, only a
few ppm of H2O2 were produced, whereas after switching
to pure H2, the catalytic activity increased to much better

S. Melada et al. / Journal of Catalysis 235 (2005) 241–248
247
layer, which in fact should be placed very close to the exter-
nal membrane surface.
4. Conclusions
From our results, we can conclude that carbon-coated
tubular ceramic membranes with palladium deposits are
promising materials for the synthesis of H2O2 starting from
hydrogen and oxygen. They are intrinsically safe, because
hydrogen and oxygen are in contact only in close proxim-
ity of the palladium surface, and they can be prepared in a
reliable and reproducible manner by a palladium deposition–
precipitation technique. We have also demonstrated that
choosing appropriate preparation parameters allows control
of both metal loading and metal particle size.
Fig. 9. Long duration test for TCM8.
Productivities up to 25–30 mmol H2O2/(m2TCM h) were
obtained at room temperature and under very mild condi-
tions. Productivity is strongly dependent on metal particle
size and is the result of two opposing catalytic processes,
hydrogen peroxide formation and hydrogen peroxide de-
struction. Pd particles about 8 nm in size proved to be the
best compromise, producing both better activity in hydrogen
peroxide formation and lower hydrogen peroxide decompo-
sition.
An important issue in improved H2O2 productivity is the
existence of an oxidized Pd surface during the catalytic ex-
periment. This is related to the mechanism through which
hydrogen peroxide is formed on the Pd surface [30].
All tests were carried out at 1 bar external pressure with a
few bars of inside/outside differential pressure. Under these
conditions, the solubility of oxygen in water is quite modest.
It is conceivable that an increase of the external oxygen pres-
sure up to 50–60 bar significantly improves productivity, so
that the system may have practical synthetic utility.
Fig. 10. Effect of the H partial pressure on the catalytic activity for TCM8.
2
values. Fig. 10 also indicates that an increase in differential
gas pressure (from 2 to 3 bar) inside the membrane was inef-
fective in improving catalytic performance. This means that
a high H2 partial pressure is needed for good catalytic activ-
ity toward H2O2 production.
Acknowledgments
Financial support from EU (contract No. G5RD-CT2002-
00678) is gratefully acknowledged. The authors are indebted
to Dr. H. Richter (HITK e.V.) and Dr. S. Tennison (Mast Car-
bon Ltd.) for kindly providing respectively the α-alumina
tubular membranes and the carbon coating.
Some kinetic runs were carried out by feeding pure H2 at
different total pressures. The effect of an increase in the total
gas pressure inside the membrane should shift the gas–liquid
interface closer to the external membrane layer until a com-
plete blowout of the pores occurs (bubble point). According
to the Young–Laplace equation, this value is theoretically
several 10’s of bars high, but defects in the pore structure
cause it to drop (in our case, no more than 6–7 bar of dif-
ferential pressure as determined experimentally, using water
as the solvent). A <10% improvement in catalytic perfor-
mance was observed with the TCM8 sample by increasing
the internal H2 pressure from 2 to 3 bar, and no detectable
improvement was observed after further increasing the dif-
ferential pressure up to 5 bar. The explored range was limited
by the pore structure defects. Nonetheless, this observation
provides useful information on the position of the catalytic
References
[1] H.-J. Riedl, G. Pfleiderer, US Patent No. 2,215,883 (1940).
[2] H. Henkel, W. Weber, US Patent No. 1,108,752 (1914), to Henkel &
Cie.
[3] L.W. Gosser, US Patent 4,681,751 (1987), to Du Pont.
[4] L. Kim, G.W. Schoenthal, US Patent No.4,007,256 (1977), to Shell
Oil.
[5] Y. Izumi, H. Miyazaki, S. Kawahara, US Patent No. 4,009,252 (1977),
to Tokuyama Soda.
[6] J. Van Weynbergh, J.-P. Schoebrechts, J.-C. Colery, PCT Patent No.
WO92/15520 (1992), to Solvay Interox.

248
S. Melada et al. / Journal of Catalysis 235 (2005) 241–248
[7] G. Paparatto, R. D’Aloisio, G. De Alberti, R. Buzzoni, European
Patent No. EP 1,160,196A1 (2001), to Eni.
[20] V.R. Choudhary, C. Samanta, A.G. Gaikwad, Chem. Commun. (2004)
2054.
[21] V.R. Choudhary, C. Samanta, Catal. Lett. 99 (2005) 79.
[22] R. Burch, P.R. Ellis, Appl. Catal. B 42 (2003) 203.
[23] J.H. Lunsford, J. Catal. 216 (2003) 455.
[24] S. Chinta, J.H. Lunsford, J. Catal. 225 (2004) 249.
[25] Y.-F. Han, J.H. Lunsford, J. Catal. 230 (2005) 313.
[26] Y.-F. Han, J.H. Lunsford, Catal. Lett. 99 (2005) 13.
[27] I. Yamanaka, T. Onizawa, S. Takenaka, K. Otsuka, Angew. Chem. Int.
Ed. 42 (2003) 3653.
[28] M. Okumura, Y. Kitagawa, K. Yamaguchi, T. Akita, S. Tsubota, M.
Haruta, Chem. Lett. 32 (2003) 822.
[29] P. Landon, P.J. Collier, A.J. Papworth, C.J. Kiely, G.J. Hutchings,
Chem. Commun. (2002) 2058.
[30] V.R. Choudary, A.G. Gaiwad, S.D. Sansare, Angew. Chem. Int. Ed. 40
(2001) 1776.
[31] L.A.M. Hermans, J.W. Geus, Stud. Surf. Sci. Catal. 3 (1979) 113.
[32] A.J. van Dillen, J.W. Geus, L.A.M. Hermans, J.v.d. Meijden, in: G.C.
Bond, et al. (Eds.), Proceedings of the Sixth International Congress
Conference on Catalysis, London, 1976, p. 677.
[33] K.P. de Jong, Stud. Surf. Sci. Catal. 63 (1991) 19.
[34] G. Fagherazzi, P. Canton, P. Riello, N. Perticone, F. Pinna, M.
Battagliarin, Langmuir 16 (2000) 4539.
[8] B. Zhou, L.-K. Lee, US Patent No. 6,168,775B1 (2001), to Hydrocar-
bon Technologies.
[9] B. Bertsch-Frank, I. Hemme, L. Von Hoppel, S. Katusic, J. Rollmann,
US Patent No. 6,387,346B1 (2002), to Degussa-Huls.
[10] F. Moseley, P.N. Dyer, US Patent No. 4,336,240 (1982), to Air Product.
[11] J.H. White, M. Schwartz, A.F. Sammells, US Patent No. 5,645,700
(1997), to Eltron Research.
[12] F. Klemens, K. Gerd, S. Achim, F. Martin, H. Wolfgang, Q. Stefan,
M. Klemens, PCT Patent No. WO98/16463 (1998), to BASF.
[13] J.A. McIntyre, S.P. Webb, PCT Patent No. WO95/30474 (1995), to
Dow Chemical.
[14] V.R. Choudary, S.D. Sansare, A.G. Gaiwad, US Patent No.
6,448,199B1 (2002), to Council of Scientific & Industrial Research
– India.
[15] H.A. Hutchins, US Patent No. 5,641,467 (1997), to Princeton Ad-
vanced technology Inc.
[16] K.M. Vanden Bussche, S.F. Abdo, A.R. Oroskar, US Patent No.
6,713,036 (2004), to UOP.
[17] V.V. Krishnan, A.G. Dokoutchaev, M.E. Thompson, J. Catal. 196
(2000) 366.
[18] A.G. Gaiwad, S.D. Sansare, V.R. Choudhary, J. Mol. Catal. A 181
(2002) 143.
[35] P. Paredes Olivera, E.M. Patrito, H. Sellers, Surf. Sci. 313 (1994) 25.
[36] S. Abate, G. Centi, S. Melada, S. Perathoner, F. Pinna, G. Strukul,
Catal. Today 104 (2005) 325.
[37] G. Centi, R. Dittmeyer, S. Perathoner, M. Reif, Catal. Today 79–80
(2003) 139.
[19] P. Landon, P.J. Collier, A.F. Carley, D. Chadwick, A.J. Papworth, A.
Burrows, C.J. Kiely, G.J. Hutchings, Phys. Chem. Chem. Phys. 5
(2003) 1917.