Catalysis Today 117 (2006) 174–179

www.elsevier.com/locate/cattod

Non-steady catalytic performance as a tool for the identification

of the active surface in VPO, catalyst for n-butane
oxidation to maleic anhydride
N. Ballarini a, F. Cavani a,*, C. Cortelli a, M. Ricotta a,
F. Rodeghiero a, F. Trifirò a, C. Fumagalli b, G. Mazzoni b
a
Dipartimento di Chimica Industriale e dei Materiali, Viale Risorgimento 4, 40136 Bologna, Italy. INSTM,
Research Unit of Bologna. A Partner of Concorde CA (6th FP of the EU)
b
Lonza SpA, Via Fermi 51, 24020 Scanzorosciate (BG), Italy

Available online 11 July 2006

Abstract
The reactivity of equilibrated VPO catalysts with different P/V ratio was investigated under steady and unsteady conditions, with the aim of
gaining information on the status of the active surface in reaction conditions. Specific treatments were applied, and then the unsteady performance
was followed along with the reaction time, until the steady behaviour was recovered. These transient states made possible to infer that the nature of
the true active surface is a function of the VPO characteristics and of the reaction conditions. With the catalyst having P/V ratio >1.0, the hydrolysis
of the VPP surface under working conditions at temperatures higher than 340–350 8C, followed by the oxidation of vanadium, generates an active
layer made of vanadium oxide and of P2O5/H2O. When instead the P/V ratio is close to 1.0, the VPP surface is oxidized to VOPO4 already at mild
temperature under working conditions, with development of an active layer which is very active but unselective to maleic anhydride at 380 8C.
Higher temperatures then lead to the hydrolysis of VOPO4, with generation of an active layer which is similar, and behaves similarly, to that one
which develops with the catalyst having P/V ratio >1.0.
# 2006 Elsevier B.V. All rights reserved.

Keywords: Vanadyl pyrophosphate; Maleic anhydride; n-Butane oxidation; Vanadium phosphate

1. Introduction reactive gases, may substantially alter the surface arrangement of

atoms as compared to the bulk. However, due to the difficulties in
The industrial catalyst for n-butane oxidation is a vanadium/ getting information under reaction conditions on the catalytically
phosphorus mixed oxide (VPO), in which bulk vanadyl active surface, the alternative hypothesis either indicate the
pyrophosphate (VPP), (VO)2P2O7, is the main component. development of surface amorphous layers which play a direct
The nature of the active surface in VPP has been studied by role in the reaction [10–14], or are based on the crystallographic
several authors, often with the use of in situ techniques [1–5]. models of the VPP, assuming that specific planes contribute to the
Several articles discuss the role of different crystalline and/or reaction pattern [2,15–21], and that the redox process occurs
amorphous VPO compounds, in the complex transformation of reversibly between VPP and VOPO4 [3,22].
the alkane to MA [6–9]. Different hypothesis were formulated; An excellent discussion of the scientific literature dealing with
while bulk VPP is in all cases assumed to constitute the core of the working state of the active surface in VPP has been recently
the active phase, differences concern the nature of the first atomic published by Schlögl and co-workers [23]. The discussion is
layers, those that are in direct contact with the gas phase. It is well supported by in situ measurements of the structure of the active
known that surface reconstruction, especially in the presence of compound and of the composition of the surface [24,25].
HRTEM images of basal and prism edges of the VPP made
possible the identification of non-crystalline adlayer, about 1 nm
* Corresponding author. thick, on the surface of equilibrated VPO catalysts [23]. By
E-mail address: fabrizio.cavani@unibo.it (F. Cavani). application of in situ X-ray absorption spectroscopy (XAS) and
0920-5861/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2006.05.017
 N. Ballarini et al. / Catalysis Today 117 (2006) 174–179 175

in situ XPS, it was concluded that the working surface is a two-

dimensional structure containing more O than that present in the
VPP; it is originated in situ by hydrolysis of the VPP, which is
metastable under reaction conditions [26]. The authors viewed
this surface layer as being constituted of a binary VOy, the growth
of which is hindered by phosphate groups [23–25]. This is
different from the development of the thick layers of amorphous
phases or from segregation of VOPO4, detrimental for catalytic
performance, which are more typical of fresh or non-equilibrated
catalysts. In contrast with this hypothesis are the results reported
by Guliants et al. [27]; the authors found by means of in situ
Raman and XRD techniques that a disordered layer about 2 nm
thick, covering the surface (1 0 0) planes of fresh VPP,
disappeared during equilibration yielding a crystalline solid
with steady and optimal catalytic performance.
Therefore, it is evident that despite the many in situ techniques
employed for the investigation of VPP, it is not possible to draw a
general picture concerning the nature of the active phase in VPP
under working conditions. One possible explanation, which was
the starting point for our research activity, is that the nature of the
active phase that develops is a function of reaction conditions,
and of the main features of VPO catalysts. Amongst the latter, the
P/V atomic ratio is known to play an important role. With this in
mind, we prepared three different catalysts having P/V ratio of
0.99, 1.03 and 1.15, and investigated the catalytic performance
after catalyst equilibration had been reached.

2. Experimental

Catalysts were prepared by the ‘‘organic procedure’’ for the

synthesis of VOHPO40.5H2O, the precursor of VPP. This
preparation consists in the suspension of the desired amounts of Fig. 1. Conversion of n-butane (top) and selectivity to maleic anhydride
V2O5 and H3PO4 in isobutanol. The precipitate finally obtained (bottom) as functions of the reaction temperature for catalysts having P/
is then thermally treated according to the following procedure: V = 0.99 (~), 1.03 (&) and 1.15 (^).
(a) drying at 120 8C for 12 h, in static air; (b) pre-calcination
step in flowing air, with temperature gradient from room having P/V atomic ratio equal to 0.99, 1.03 and 1.15. It is worth
temperature up to 300 8C; then isothermal step at 300 8C in air noting that preliminarily to these tests, catalysts were
for 6 h; (c) calcination step in flowing N2, with temperature maintained for 150 h at 400 8C with the reaction stream, and
gradient from 300 to 550 8C, and final isothermal step at the that after this period they can be considered as ‘‘equilibrated’’
latter temperature for 6 h. Three samples were prepared, having or almost equilibrated ones. In fact, they did not exhibit any
P/V ratio equal to 0.99, 1.03 and 1.15; the atomic P/V ratio in further variation of catalytic performance along with the
precursors was confirmed by means of SEM-EDX. increasing reaction time. It is also useful to remind that for these
Catalytic tests were carried out in a laboratory glass flow- equilibrated catalysts, the performance shown in Fig. 1 was
reactor, loading 0.8 g of catalyst. Products were analyzed by completely reversible, that is, the curves were fully reproduced
means of gas chromatography (on-line sampling of exit when the temperature was raised and lowered in sequence
stream), equipped with a HP-1 column for the separation of several times. This is the starting point of our investigation;
C4 hydrocarbons, formaldehyde, acetic acid, acrylic acid, trends shown in Fig. 1 were not due to irreversible
maleic anhydride. A Carbosieve SII column was used for the modifications of the catalysts, and did not occur as a
separation of oxygen, carbon monoxide and carbon dioxide. consequence of the ongoing equilibration of the samples.
Results in Fig. 1 demonstrate that the P/V ratio greatly
3. Results and discussion affected the catalytic performance. While the sample having
excess P had the expected trend for both n-butane conversion
3.1. Steady-state catalytic performance of equilibrated and selectivity to maleic anhydride, the one having P/V 0.99
VPO catalysts gave a jump of activity at 340–350 8C, which was then
recovered at temperatures higher than 370–380 8C. The sample
Fig. 1 plots the effect of temperature on n-butane conversion, with P/V = 1.03 showed a catalytic behaviour which was
and on the selectivity to maleic anhydride, for three catalysts intermediate between those of samples with P/V = 0.99 and
176 N. Ballarini et al. / Catalysis Today 117 (2006) 174–179

1.15. Therefore, while the three catalysts had comparable

activity at low and high temperature, they differed considerably
in the intermediate temperature range. The same occurred with
the selectivity to maleic anhydride: the catalyst having P/
V = 0.99 showed a fall of selectivity in correspondence of the
activity increase; this fall was then recovered above 360 8C. It is
worth noting that the decrease of selectivity was not simply due
to the increase of conversion; in fact, ‘‘good’’ VPO catalysts
giving 50% n-butane conversion at 380 8C, produce maleic
anhydride with selectivity higher than 50% [7].
The experimental effects are attributable to reversible
modifications in the nature of the active surface, the occurrence
of which was a function of the reaction temperature; these
modifications were particularly evident in the sample having P/
V = 0.99. We assumed that these phenomena can be studied
through the analysis of these reversible transitions, and that the Fig. 2. Conversion of n-butane (^), selectivity to maleic anhydride (&), to
carbon oxides (~), and to formaldehyde plus acetic acid plus acrylic acid (*)
latter can be forced by inducing transient changes, through the as functions of the reaction time during Test 3a of Table 1.
application of treatments of oxidation and hydrolysis aimed to
the modification of the active surface characteristics.
we chose three representative temperatures: 320–340, 380 and
3.2. Non-steady catalytic performance of equilibrated VPO 440 8C, and at each temperature level we induced modifications
catalysts in the stable active surface by applying either oxidizing
treatments (feed of air), or hydrolyzing treatments (feed of 4%
We forced transient states, by applying conditions different steam in He). Then, we fed again the reaction mixture, and
from those the catalyst underwent under reaction. Specifically, followed the variation of catalytic performance that occurred

Table 1
Description of tests run with catalyst having P/V = 0.99
Test no Conditions for tests, T and feed a Initial performance Final performance Information gained
in step cb in step cb (time)c
1a (a)–, (b) 200 ! 320 8C, in He, X 9, 1 72 X 9, 1 72 Heating in He does not
(c) 320 8C, mix alter the active surface at 320 8C.
2a (a) 320 8C, mix, (b) 320 ! 380 8C, X 51, 1 31 X 51, 1 31 The active surface is the same at
in He, (c) 380 8C, mix 320 8C and at 380 8C.
3a (a) 440 8C, mix, (b) 440 8C ! 380 8C, X 43, 1 57 X 51, 1 32, (8–10 h) The active surface which forms at 440 8C
in He, (c) 380 8C, mix is less active but much more selective
than that one which forms at 380 8C.
4a (a) 320 8C, mix, (b) 320 8C, air, 2 h, X 10, 1 72 X 10, 1 72 The active surface at 320 8C is not altered
(c) 320 8C, mix by a treatment in air; therefore, the active
surface is in the higher oxidation state
for these conditions.
5a (a) 380 8C, mix, (b) 380 8C, air, 1 h, X 53, 1 32 X 51, 1 32 The active surface at 380 8C is not altered
(c) 380 8C, mix by a treatment in air; therefore, the active
surface is in the higher oxidation state for
these conditions.
6a (a) 380 8C, mix, (b) 380 8C, X 44, 1 41 X 53, 1 29, (2 h) At 380 8C, the hydrolysis renders the active
4% H2O in He, 1 h, (c) 380 8C, mix surface less active and more selective.
7a (a) 380 8C, mix, (b) 380 8C, X 55, 1 38 X 56, 1 30, (1.5 h) The hydrolysis (test 6a) renders the catalyst
4% H2O in air, 1 h, (c) 380 8C, mix oxidizable, and this makes the active surface
slightly more selective.
8a (a) 440 8C, mix, (b) 440 8C, air, 1 h, X 69, 1 47 X 67, 1 50 The active surface at 440 8C is not substantially
(c) 440 8C, mix altered by an oxidative treatment at 440 8C.
Therefore, the active surface is in the higher
oxidation state for these conditions.
9a (a) 440 8C, mix, (b) 440 8C, X 67, 1 46 X 65, 1 48 The active surface at 440 8C is not substantially
4% H2O in He, 1 h, (c) 440 8C, mix altered by an hydrolysis treatment at 440 8C.
Therefore, the active surface is already
hydrolysed under these conditions.
a
(a): Steady performance with the reaction mixture (mix); (b): Treatment; (c) unsteady performance with the reaction mixture (mix).
b
X = n-butane conversion, %; 1 = selectivity to maleic anhydride, %.
c
Time necessary to reach the steady final performance in step c, hours. If not indicated, the performance was stable all along step c.
 N. Ballarini et al. / Catalysis Today 117 (2006) 174–179 177

along with time-on-stream, during the transition of the catalyst raised, it gives high conversion but very poor selectivity. At
from the forced situation to that one stable under those T > 380 8C, however, the active surface which develops is
conditions. The steady state finally corresponds to that one moderately active (less active than that one which forms at
which is reported, for each temperature level, in Fig. 1. 380 8C), but much more selective. The transformation of the
In most cases, variations were slow and could be followed latter active surface into the former one generates the progre-
along reasonable periods of time, e.g., of a few hours. When the ssive variation of catalytic performance shown in Fig. 2.
treatment induced no variation in catalytic performance, we (2) The active surface that develops under reaction conditions at
assumed that either the active surface was stable towards that low temperature is not in a fully hydrolysed state, while it is
treatment, or that it was insensitive to it, because it was already likely oxidized. This picture suggests it to be closer to VOPO4
in the state we were meant to induce by that specific treatment. than to (VO)2P2O7 or to a VOx/polyphosphoric acid mixture.
Other tests were made by changing the reaction temperature in The active surface that develops at high temperature, instead,
He, in order to check how the active surface which had is hydrolysed and oxidised. This indicates that above 380 8C
developed at the starting temperature, adapted to the new one. VOPO4 is hydrolysed to generate a new active surface that is
As an example, Fig. 2 reports the variation of catalytic more selective to maleic anhydride.
performance along with the reaction time for sample with P/
V = 0.99, after the equilibrated catalyst had first been operating Table 2 reports some tests carried out with the catalyst
under the reaction mixture at 440 8C for a few hours, then was having P/V = 1.03. The main indications inferred from these
cooled down to 380 8C in He, and finally was run with the experiments are the following:
reaction mixture at 380 8C (Test 3a in Table 1). The progressive
slight increase of conversion was accompanied by a more (1) The active surface that develops at 320–340 8C is different
considerable decrease of selectivity to maleic anhydride. After from that one which forms at 380 8C; specifically, the latter is
a few hours at 380 8C, the catalyst reached the steady hydrolysed and partially oxidized. Moreover, it is more
performance. This test indicates that in sample with P/ active, but equally selective, than the former one. At 340 8C,
V = 0.99, the active surface at 440 8C is not the same as that temperature is too low to initiate the hydrolysis of the VPP,
one that is stable at 380 8C, and that the former is intrinsically while when the temperature is progressively increased,
less active but considerably more selective than the latter. conditions become more favourable for its hydrolysis, also
Table 1 summarizes the tests run with the catalyst having P/ because of the higher amount of water produced by the
V = 0.99, and the indications emerged from each test. The most reaction. Furthermore, the active phase at 380 8C is not fully
relevant information gained from these experiments is the oxidized, and the forced development of an higher oxidation
following: state generates an unstable, more selective active surface.
(2) At 440 8C, the active surface is similar to that one which
(1) The active surface that develops under reaction conditions, at develops at 380 8C.
320 8C is similar to that one which forms at 350–380 8C,
while it changes when the reaction is carried out above Therefore, differences at 380 8C between the catalysts
380 8C. The former active surface performs selectively at having P/V = 0.99 and 1.03 concern the degree of hydrolysis
moderate temperature, but as soon as the temperature is and of oxidation of the active surface. Data suggest that the VPP

Table 2
Description of tests run with catalyst having P/V = 1.03
Test No Conditions for tests, T and feed a Initial performance Final performance Information
in step cb in step cb (time)c gained
1b (a) 380 8C, mix, (b) 380 ! 340 8C, X 24, 1 73 X 10, 1 72, (4 h) The active surface at 340 8C is different
in He, (c) 340 8C, mix from that one which forms at 380 8C; the
latter is more active but equally selective
than the former one.
2b (a) –, (b) 440 8C, in He, 12 h, X 75, 1 53 X 76, 1 53 The permanence of the active surface in He
(c) 440 8C, mix at 440 8C does not alter its characteristics.
3b (a) 380 8C, mix, (b) 380 8C, air, X 42, 1 60 X 40, 1 53, (0.5 h) The oxidation at 380 8C improves the
1 h, (c) 380 8C, mix selectivity; therefore, under these conditions
the active surface is not fully oxidized.
4b (a) 380 8C, mix, (b) 380 8C, X 39, 1 50 X 40, 1 52 The active surface at 380 8C is not altered
4% H2O in He, 1 h, (c) 380 8C, mix by a treatment with steam; therefore, at
380 8C the active surface is already hydrolysed.
5b (a) 440 8C, mix, (b) 440 8C, X 84, 1 59 X 76, 1 52, (0.5 h) At 440 8C, the oxidation improves the
air, 1 h, (c) 440 8C, mix performance; therefore, at 440 8C the active
surface is not fully oxidized.
a
(a): Steady performance with the reaction mixture (mix); (b): treatment; (c) unsteady performance with the reaction mixture (mix).
b
X = n-butane conversion, %; 1 = selectivity to maleic anhydride, %.
c
Time necessary to reach the steady final performance in step c, h. If not indicated, the performance was stable all along step c.
178 N. Ballarini et al. / Catalysis Today 117 (2006) 174–179

in the sample with P/V = 1.03 is less prone to oxidation than the
sample having P/V = 0.99, but it undergoes hydrolysis already
for low temperatures. This information is in agreement with
indications from literature, that excess P stabilizes VPP towards
oxidation to VOPO4 [28]. Once the in situ hydrolysis begins,
the oxidation of V generates a mixed VOx/polyphosphoric acid
phase, which performs much better than the active surface
which develops under the same conditions with the sample
having P/V = 0.99, for temperatures between 340 and 400 8C.
The latter is essentially made of VOPO4. Therefore, it can be
inferred that the hydrolysis of the VPP is the preliminary
condition to develop an active surface that performs better than
VOPO4. An analogous mechanism of VPP hydrolysis, followed
by V oxidation, was proposed years ago by Xue and Schrader
[26], who were the first to characterize the VPO catalyst by in
situ Raman spectroscopy.
Finally, at high temperature differences between samples
become smaller, because the hydrolysis of VOPO4 in sample P/V
0.99, and the further increase of the oxidation level of Vin sample
P/V 1.03 (under more oxidizing conditions), make the Fig. 3. Scheme summarizing the composition of the active surface for catalysts
having P/V ratio equal to 0.99 and 1.03, and for different reaction temperatures.
corresponding active surfaces to become similar in the two cases.
Circled areas represent approximate compositions.
Data also indicate that the active surface which performs at
best is that one made of V2O5 + P2O5/H2O (the latter being a
mixture of polyphosphoric species having different condensa- 4. Conclusions
tion degree). This is achieved already at low temperature with
catalysts having P/V ratios definitely higher than 1.0; the excess The study of the steady and transient performance of
P avoids the formation of VOPO4, while it does favour the equilibrated VPO catalysts having different P/V ratio allowed
hydrolysis of VPP. In sample having P/V = 1.15, this generates us to gain information on the nature of the active surface under
an active surface layer with the optimal characteristics in the reaction conditions, during n-butane oxidation to maleic
entire range of temperature, in terms of both hydrolysis and anhydride. Specific treatments were applied, and then the
oxidation degree, and hence the best selectivity to maleic transient performance was followed along with the reaction
anhydride. When instead a P/V ratio close to 1.0 is used, the time, until the steady performance was recovered. With the
VPP is easily oxidized to VOPO4 already at low temperature; catalyst having P/V = 1.03, the hydrolysis of the VPP surface
the latter is hydrolysed to the selective surface layer only at under working conditions at temperatures higher than 340–
temperatures higher than 380–400 8C. 350 8C, followed by the oxidation of V, generated an active
The results achieved are summarized in Fig. 3, which surface made of vanadium oxide and polyphosphoric acids.
indicates the approximate range of composition for the active When instead the P/V ratio was equal to 0.99, the catalyst
surface. This model also justifies discrepancies in literature surface was oxidized to VOPO4 already at mild temperature,
concerning the in situ characterization of VPO; in fact, with development of a layer that was very active but unselective
depending on reaction temperature and on catalyst P/V ratio, at 360–380 8C. At T > 380 8C the hydrolysis of VOPO4 led to
either an active surface very close to the VPP is identified, or one the development of an active layer similar to that one which
which is close to VOPO4, or an hydrolysed and oxidised one. forms with the catalyst having P/V > 1.0.
One open question concerns the mechanisms that establish The results obtained explain discrepancies in literature
at the catalyst surface, leading to the reversible formation of the concerning the nature of the active layer in VPO as evaluated by
different types of active layers, when the reaction temperature in situ and ex situ measurements; the latter is not only a function
is varied (Fig. 1). Variations in the degree of hydrolysis and in of the VPO characteristics (specifically, of the P/V ratio), but
the oxidation state of vanadium are likely a function of the gas- also of the reaction conditions.
phase composition, and of the degree of n-butane conversion.
For instance, when the hydrocarbon is the limiting reactant, References
higher oxygen-to-hydrocarbon local ratios at high temperature
may favour the development of more oxidised surfaces. This [1] L.M. Cornaglia, E.A. Lombardo, Appl. Catal A 127 (1995) 125.
also implies that the nature and extent of transformations [2] I.E. Wachs, J.M. Jehng, G. Deo, B.M. Weckhuysen, V.V. Guliants, J.B.
illustrated in Fig. 3 are function of the feed composition. Benziger, Catal. Today 32 (1996) 47.
Another question concerns the reasons why a VPP that [3] G. Koyano, T. Okuhara, M. Misono, J. Am. Chem. Soc. 120 (1998) 767.
[4] F. Richter, H. Papp, T. Gotze, G.U. Wolf, B. Kubias, Surf. Inteface Anal.
develops a hydrolysed and oxidised surface gives better 26 (1998) 736.
selectivity to maleic anhydride, at intermediate and high [5] P. Delichere, K.E. Bere, M. Abon, Appl. Catal A 172 (1998) 295.
temperatures, than the intact VPP or VOPO4. [6] E. Bordes, C. R. Acad. Sci. Paris, Série IIc 3 (2000) 725.
 N. Ballarini et al. / Catalysis Today 117 (2006) 174–179 179

[7] F. Cavani, F. Trifirò, Catalysis 11 (1994) 246. [20] V.V. Guliants, J.B. Benziger, S. Sundaresan, N. Yao, I.E. Wachs, Catal.
[8] J.C. Vedrine, Topics Catal. 11 (2000) 147. Lett. 32 (1995) 379.
[9] J.C. Volta, Catal. Today 32 (1996) 29. [21] V.V. Guliants, J.B. Benziger, S. Sundaresan, I.E. Wachs, J.M. Jehng, J.E.
[10] H. Berndt, K. Buker, A. Martin, A. Brückner, B. Lücke, J. Chem. Soc., Roberts, Catal. Today 28 (1996) 275.
Faraday Trans. 91 (1995) 725. [22] G. Koyano, T. Saito, M. Misono, J. Mol. Catal. A 155 (2000) 31.
[11] A. Brückner, A. Martin, N. Steinfeldt, G.U. Wolf, B. Lücke, J. Chem. Soc. [23] H. Bluhm, M. Hävecker, E. Kleimenov, A. Knop-Gericke, A. Liskowski,
Faraday Trans. 92 (1996) 4257. R. Schlögl, D.S. Su, Topics Catal. 23 (2003) 99.
[12] A. Brückner, B. Kubias, B. Lücke, Catal. Today 32 (1996) 215. [24] M. Hävecker, A. Knop-Gericke, H. Bluhm, E. Kleimenov, R.W. Mayer, M.
[13] M. Ruitenbeck, A.J. van Dillen, A. Barbon, E.E. van Fassen, D.C. Fait, R. Schlögl, Appl. Surf. Sci. 230 (2004) 272.
Koningsberger, J.W. Geus, Catal. Lett. 55 (1998) 133. [25] M. Hävecker, R.W. Mayer, A. Knop-Gericke, H. Bluhm, E. Kleimenov, A.
[14] A. Brückner, A. Martin, B. Kubias, B. Lücke, J. Chem. Soc. Faraday Liskowski, D.S. Su, R. Follath, F.G. Requejo, D.F. Ogletree, M. Salmeron,
Trans. 94 (1998) 2221. J.A. Lopez-Sanchez, J.K. Bartley, G.J. Hutchings, R. Schlögl, J. Phys.
[15] E. Bordes, Topics Catal. 15 (2001) 131. Chem. B 107 (2003) 4587.
[16] J. Ziolkowski, E. Bordes, P. Courtine, J. Catal. 122 (1990) 126. [26] Z.Y. Xue, G.L. Schrader, J. Phys. Chem. B 103 (1999) 9459.
[17] J. Ziolkowski, E. Bordes, J. Mol. Catal. 84 (1993) 307. [27] V.V. Guliants, S.A. Holmes, J.B. Benziger, P. Heaney, D. Yates, I.E.
[18] P.A. Agaskar, L. DeCaul, R.K. Grasselli, Catal. Lett. 23 (1994) 339. Wachs, J. Mol. Catal. 172 (2001) 265.
[19] R. Ebner, M.R. Thompson, Catal. Today 16 (1993) 51. [28] F. Cavani, G. Centi, F. Trifirò, Appl. Catal. 15 (1985) 151.