Influence of supported-metal characteristics on deNOx catalytic activity over Pt/CeO2

Article abstract of DOI:10.1016/j.molcata.2009.02.007

Fine Pt particles were loaded on the CeO₂ supports with various specific surface areas by an incipient wetness method or a solvothermal one, and their catalytic properties for the NO_x reduction by hydrogen were tested. The Pt/CeO

Full text of DOI:10.1016/j.molcata.2009.02.007

Journal of Molecular Catalysis A: Chemical 304 (2009) 159–165
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Influence of supported-metal characteristics on deNO catalytic
x
activity over Pt/CeO2
a
a
a
a
Masahiro Itoh , Makoto Saito , Masahiko Takehara , Koji Motoki ,
b
a,∗
Jun Iwamoto , Ken-ichi Machida
Center for Advanced Science and Innovation, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan
a
b
Honda R&D Co., Ltd., Automobile R&D Center, 4630 Shimotakanezawa, Haga-machi, Haga-gun, Tochigi 321-3393, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Fine Pt particles were loaded on the CeO2 supports with various specific surface areas by an incipient
wetness method or a solvothermal one, and their catalytic properties for the NOx reduction by hydrogen
were tested. The Pt/CeO2 catalysts prepared by the solvothermal method provided higher activity than
those by the impregnation one when using the CeO2 support with high surface area. Each CeO2 supports
and Pt particles prepared by the solvothermal method did not show the good catalytic activity singly.
However the catalysts obtained by physically mixing Pt particles with the CeO2 supports gave excellent
NOx conversion of ca. 80% with relative high N2 selectivity (75%).
Received 15 October 2008
Received in revised form 4 February 2009
Accepted 4 February 2009
Available online 13 February 2009
Keywords:
H2-SCR
deNOx
©
2009 Elsevier B.V. All rights reserved.
Solvothermal
Ceria supported Pt catalyst
Metal–support interaction
1
. Introduction
the N2 selectivity and NOx conversion, because rare earth oxides
change their acidity/basicity by decreasing their ionic radii accom-
panied with lanthanide contraction [11]. Although almost all the
catalysts did not provide the good catalytic activity, only Pt/CeO2
NO and NO2 (referred to as NOx) emitted from mobile and sta-
tionary internal combustion is one of the most closely monitored
categories of pollutants, since NOx leads photochemical smog and
acid rain. To meet their strict emission regulations, selective cat-
alytic reduction (SCR) for NOx by using NH (including urea system),
CO, hydrocarbon, or H2 as a reducing agent has been extensively
showedthe highNOx conversionataround 373 K. NH temperature-
3
programmed desorption (TPD) profiles for the Pt/rare earth oxide
catalysts revealed that their solid acidity is not drastically differ-
ent, suggesting that the specific interaction between Pt and CeO2
influences the NOx reduction activity, rather than solid acidity.
In this study, two different Pt loading techniques such as an
incipient wetness method and a solvothermal one, where the latter
is effective to prepare the Pt metal with high crystallinity confirmed
from our previous work [12], were applied to the CeO2 supports
with different specific surface area to modify the metal/support
interface circumstance. Their catalytic properties for NOx reduction
were discussed from the viewpoint of metal–support interaction.
3
investigated [1–8]. Among them, the H -SCR has an advantage
2
on high conversion of NOx reduction at relatively low temper-
ature under excess oxygen condition such as lean fuel/air ratio
combustion. Additionally, hydrogen can be generated on board by
controlling engine emission or using Reforming and exhaust gas
recirculation (REGR) [9]. So the H -SCR system has been consid-
ered as a candidate of NOx control technology in place of the present
2
NH3 (or urea)-SCR method. Such H -SCR for NOx, however, tends
2
to produce more N O causing greenhouse worming rather than
2
N2 during the reaction. The N2 selectivity cannot be improved by
changing precious metal loadings or H2 concentration in the feed
gas, but is affected by acidity/basicity character of the oxides used
2. Experimental
Two types of CeO2 supports having high and low specific sur-
face area were used in this study. For the incipient wetness method,
as a support [10]. In our recent work, H -SCR activity for NOx over
2
the platinum supported rare earth oxide catalysts has been inves-
tigated to elucidate the influence of the solid acidity/basicity on
respective water absorption amounts for these CeO supports were
2
measured before the impregnation procedure. Then the CeO pow-
2
ders were added to the corresponding volume of Pt(NO ) aqueous
3
2
solutions with setting Pt loading of 1 wt% and mixed thoroughly to
get slurries. The resulting slurries were dried at 393 K overnight and
then calcined at 873 K for 2 h in air. For the solvothermal method,
∗ Corresponding author. Tel.: +81 6 6879 4209; fax: +81 6 6879 4209.
E-mail address: machida@casi.osaka-u.ac.jp (K.-i. Machida).
1381-1169/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2009.02.007

160
M. Itoh et al. / Journal of Molecular Catalysis A: Chemical 304 (2009) 159–165
appropriate amounts of the CeO2 powders and the Pt(NO3)2 aque-
ous solution were added to 80 ml of isopropyl alcohol in a test tube,
and the test tube was placed in a 100 ml autoclave vessel. Subse-
quently the sealed vessel was kept at 573 K for 12 h to conduct a Pt
deposition reaction completely. In fact, the solution color changed
from yellow to transparent after the reaction, indicating the above
deposition almost fully proceeded. After cooling to room temper-
ature, the products were collected by a centrifuge and repeatedly
washed with acetone. The resultant powders were dried at 323 K
overnight in air, and then calcined at 873 K for 2 h in air. For con-
venient, the catalysts are labeled as follows: the catalysts prepared
by the incipient wetness method with using low and high specific
surface area CeO2 are represented as IW-Pt/L-CeO2 and IW-Pt/H-
with flowing a reactant gas stream (0.4 vol% H and balanced He
2
−
1
with a total flow rate of 30 mL min ) at 323–623 K. Further phys-
ical properties were characterized by X-ray diffraction with Cu-K␣
radiation (RIGAKU, RINT2200), transmission electron microscopy
(JEOL, JEM-2010F), and scanning electron microscopy (HITACHI, S-
3000H). Catalytic testing was performed at atmospheric pressure
in a conventional fixed-bed quartz reactor. Gas mixtures of 0.1 vol%
NO, 0.4 vol% H , 10 vol% O , and balanced with N were fed to the
2
2
2
−
1
catalyst sample with 0.15 g (W/F = 0.3 g s mL ) placed in the reac-
tor tube. The effluent from the reactor was analyzed by an online
FT-IR spectrometer with a gas cell (PerkinElmer, Spectrum 2000).
Spectra were collected by scanning five times from 1000 cm to
4000 cm at resolution of 0.5 cm . Concentrations of NO, NO2,
and N O were determined from their calibration curves by inte-
grating the peak area at respective 1911, 1586, and 2214 cm . No
ammonia formation was observed in this study. The N2 produc-
tion was evaluated by subtracting NO, NO , and N O concentrations
−
1
−
1
−1
CeO , respectively. In a similar way, the catalysts prepared by the
2
2
−
1
solvothermal method are represented as ST-Pt/L-CeO and ST-Pt/H-
2
CeO2.
Nitrogen adsorption isotherm was measured at 77 K to evalu-
ate the specific surface area using Brunauer, Emmett, and Teller’s
2
2
of outlet gas from the inlet NO concentration. Diffuse reflectance
infrared Fourier transform spectroscopy (DRIFTs) was performed
by using the above spectrometer with a temperature controllable
diffuse reflectance reaction cell to observe adsorption chemical
species in situ.
(
BET) equation. CO chemisorption was carried out by a pulse tech-
nique at room temperature to determine the Pt metal dispersion
13]. CO2 TPD experiments were carried out to estimate the solid
[
basicity from the desorption temperature by using a gas chromato-
graph equipped with TCD. The sample surface area contacting CO2
roughly set to 10 m2 for all specimens by adjusting their weight
to enhance a signal/noise ratio. Temperature-programmed reaction
3. Results and discussion
(TPR) experiments were performed in order to investigate the reac-
x
2
Fig. 1 shows the NO conversion, N2 selectivity, and N O selec-
tion between hydrogen and NOx species adsorbed on a surface of
the catalysts. Before starting the TPR experiments, NO (0.1 vol%)
and O2 (10 vol%) mixed gases balanced with He (total flow rate of
tivity on the Pt/CeO2 catalysts prepared in this study as a function
of the reaction temperatures. ST-Pt/H-CeO2 showed the good NOx
conversion with the N2 selectivity of ca. 40%. On the contrary, the
NOx reduction activity for IW-Pt/H-CeO2 was poor as reported by
Machida et al. [14] The BET surface area values on the respective
−1
30 ml min ) are passed at 473 K. The effluent gas was analyzed
by an on-line mass spectrometer (Pfeiffer Vacuum, PrismaPlus)
Fig. 1. NOx conversion and selectivity as a function of temperature observed on (a) ST-Pt/H-CeO2, (b) IW-Pt/H-CeO2, (c) ST-Pt/L-CeO2, and (d) IW-Pt/L-CeO2. Grey and black
bars represent the N2O and N2 selectivity, respectively.

M. Itoh et al. / Journal of Molecular Catalysis A: Chemical 304 (2009) 159–165
161
Table 1
Characteristics of the Pt/CeO2 catalysts.
2
Dispersion [%]a
Catalysts
Surface area [m /g]
CO adsorption [␮mol/g]
Pt particle size [nm]
ST-Pt/H-CeO2
IW-Pt/H-CeO2
ST-Pt/L-CeO2
IW-Pt/L-CeO2
119
135
5
34
38
1.4
8.3
67
74
3
1.7
1.5
41
9
16
6.9
a
Calculated as CO:Pt = 1:1.
2
ST-Pt/H-CeO and IW-Pt/H-CeO were 119 and 135 m /g as summa-
2
2
rized in Table 1. The Pt metal dispersion for IW-Pt/H-CeO (74%) was
2
slightly higher than that for ST-Pt/H-CeO (67%). The Pt particle size
2
directly observed by the TEM dark field images for ST-Pt/H-CeO2
and IW-Pt/H-CeO2 were well consistent with the CO chemisorp-
tion results, and their morphology were analogously spherical as
shown in Fig. 2. On the other hand, the NOx conversion activity
on IW-Pt/L-CeO was considerably improved by decreasing surface
2
area of the CeO2 support as shown in Fig. 1(d). In using the CeO2
support with low surface area, the NOx reduction properties were
almost same for each catalyst prepared by the solvothermal and
incipient wetness methods, though their Pt metal dispersion greatly
varied as listed in Table 1. Fig. 3 shows the CO -TPD profiles obtained
2
on the CeO2 supports with high and low surface area. Both cata-
lysts had peaks around 500 K and 700 K, which were derived from
weak and moderate base sites, respectively [15], suggesting that
they had the almost same solid base strength regardless of their
surface area values. The results of CO chemisorption, TEM observa-
tion, and CO -TPD indicate that the NOx reduction activity over the
2
Fig. 3. CO2 temperature-programmed desorption profile on the CeO2 support with
high surface area (solid line) and low surface area (dashed line).
Pt/CeO2 catalysts was significantly affected by the metal–support
interaction. The IW-Pt/H-CeO catalysts, which had the largest con-
2
tact area between Pt and CeO2 in the catalysts prepared in this
study, resulted in the low NOx conversion. The suitable interaction
is necessary to induce the good catalytic activity. The solvothermal
method is probably effective to obtain such weak metal–support
contact force, because nucleus growth rather than crystal nucle-
ation is dominant during the solvothermal Pt deposition, resulting
in decrease of Pt-CeO2 bond leakage in the interface. In fact, the
obvious diffraction peaks derived from face centered cubic sym-
metry of Pt metal in addition to the peaks from the CeO2 phase
were detected only on the XRD pattern of ST-Pt/L-CeO2 as shown
Fig. 2. TEM dark field images observed on (a) ST-Pt/H-CeO2 and (b) IW-Pt/H-CeO2.
Fig. 4. XRD pattern obtained on ST-Pt/L-CeO2 with Cu-K␣ radiation.

162
M. Itoh et al. / Journal of Molecular Catalysis A: Chemical 304 (2009) 159–165
Fig. 5. NOx conversion and selectivity as a function of temperature observed on (a) the Pt metal particles prepared by the solvothermal method, (b) the sole CeO2 with high
surface area, (c) the catalyst prepared by physically mixing (a) and (b), and (d) the catalyst prepared by mixing the Pt metal particles (reflux method) with (b). Grey and black
bars represent the N2O and N2 selectivity, respectively.
in Fig. 4. Formation of the Pt particles with a several dozen nm was
also confirmed from the CO chemisorption measurement (Table 1)
and the direct TEM observation (not shown).
Figs. 6(a) and 7 (curve a). The Pt metal catalyst possessed the
high NOx conversion only at room temperature, in addition with
high N O selectivity. The NOx reduction activity became less with
2
To investigate the reaction mechanism on the present Pt/CeO2
catalysts, individual NOx reduction activity on the sole Pt metal
or CeO2 support with high surface area was tested as shown in
Fig. 5. The Pt metal particles were prepared by the solvothermal
method in a similar manner mentioned above without adding the
CeO2 support. In the case of the support absence, well-crystallized
Pt particles with fcc symmetry was obtained and they formed
in a spherical shape with a diameter of ca. 100 nm as shown in
increasing the reaction temperature, and most NO was converted to
NO2 above 373 K (not indicating NO2 selectivity in Fig. 5). Fig. 8(a)
shows the DRIFT spectrum on the Pt metal catalyst at 373 K. No
adsorbed species related nitrogen oxide was observed and symmet-
rical transmission peaks assigned to gaseous H O were observed
2
−
1
at around 1300–1800 cm . H2 wastes by reacting with excess O2
and cannot effectively work as a reducer for NOx. As expected,
the CeO2 support did not show the catalytic activity against NOx
Fig. 6. SEM images of the Pt metal particles prepared by (a) solvothermal method and (b) reflux one.

M. Itoh et al. / Journal of Molecular Catalysis A: Chemical 304 (2009) 159–165
163
in Fig. 8(b) [17]. The peak at 1430 cm 1 are also derived from nitrite
−
−
1
ion and the formation of nitrates was detected at 1300 cm . Such
strong adsorption of NOx species retards further adsorption of other
chemical reactants such as H2 and O2 and the poor NOx conversion
is resultantly attained.
In contrast, the catalyst prepared by physically mixing the above
Pt metal particles (1 wt%) and CeO support provided the good NOx
2
reduction activity with relative high N selectivity of 50% at 373 K as
2
shown in Fig. 5(c). Although deterioration of catalytic activity due
to agglomeration should be considered for such mixed type cata-
lysts, the deNOx activity on the second test was almost same and
the XRD patterns hardly changed before and after the first activity
test as shown in Fig. 7 curves b and c. The serious agglomeration
did not occur under the present reaction condition below 573 K. In
addition, the NOx conversion gave two maxima at 373 K and 473 K,
suggesting there includes the different mechanism for NOx reduc-
tion over this catalyst [18,19]. To our best knowledge, this is the
first observation of such two conversion maxima on Pt-based cata-
lysts having high affinity against H2 and O . In the present catalyst
2
system, Pt and CeO2 have an important role each other; Pt adsorbs
and dissociates chemical species such as H , O , NO, etc. and CeO
2
2
2
accumulates the NOx species on its surface by proper solid basicity.
It is considered that the NOx reduction reaction proceeds on the
Fig. 7. XRD pattern obtained on the Pt metal particles prepared by the solvothermal
method (curve a) as well as the Pt/CeO2 mixed catalyst before (curve b) and after
interface between Pt and CeO . The H -TPR experiments bears out
2
2
(
curve c) the catalytic test.
the above supposition as shown in Fig. 9. The NOx species adsorbed
on the sole Pt metal particle catalyst weakly reacts with hydrogen
to generate trace amounts of N2O (m/z = 44). Meanwhile, various
NOx-reduced species such as N2O, N2 (m/z = 28), and NH3 (m/z = 15
as NH ) were detected for the case of the Pt/CeO2 mixed catalyst,
suggesting that the catalytic sites are Pt metal particles attached
reduction singly as shown in Fig. 5(b). It is well known that CeO2
acts as a NOx storage component besides one for oxygen [16]. The
DRIFT spectrum for the CeO2 support showed the formation of a
+
−1
strong transmission peak at 1180 cm assigned to nitrite as seen
Fig. 8. DRIFT spectra observed on (a) the Pt metal particles prepared by the solvothermal method, (b) the sole CeO2 with high surface area, (c) the catalyst prepared by
physically mixing (a) and (b), and (d) ST-Pt/H-CeO2.

164
M. Itoh et al. / Journal of Molecular Catalysis A: Chemical 304 (2009) 159–165
solid basicity related to the NOx adsorption is in the following order:
CeO > ZrO > Al O > TiO [21,22]. It is considered that the adsorp-
2
2
2
3
2
tion site with certain degree of basicity contributes to enhance the
−
N selectivity by condensing the intermediate NO3 species. More-
2
over, the influence of the Pt metal particles on the NOx reducing
properties was investigated. The Pt metal particles were precip-
itated by refluxing 200 ml of a Pt(NO3)2 ethylene glycol solution
(1 mM) at 470 K for 3 h. The solution color changed from yellow to
black during the reaction, and the fine black powders were sepa-
rately obtained by centrifuging the resultant product. The Pt metal
particles prepared from the reflux method formed in a rectangular
shape as shown in Fig. 6(b). Fig. 5(d) shows the NOx conversion on
a catalyst by mixing the Pt particles (reflux method) with the CeO2
support (high surface area). The highest N2 selectivity of 75% was
observed with relative high NOx conversion (ca. 80%). This result
suggests that nature of the metal particles more influences on the
NOx reduction activity than the support. Although both XRD pat-
terns for the Pt particles prepared by the solvothermal and the
reflux methods were assigned to a fcc Pt metal phase, their particle
morphology was distinct as shown in Fig. 6. Balint et al. reported
that the facet effect on Pt metal plays an important role in NO dis-
sociation and controls the reaction selectivity to N2 and N O in
2
the NO–C H –O reaction system [23,24]. The rectangular shape
3
6
2
observed on the Pt particles prepared by the solvothermal method
indicates the anisotropic crystal growth along a specific plane. Such
specific facet exposed on the Pt particles may be responsible for the
high N2 selectivity.
4. Conclusions
The NOx reduction activity on the Pt supported ceria catalysts
was examined. The NOx reduction activity is significantly affected
by the physicochemical characteristic of the metal–support inter-
face and the best catalytic activity has been, consequently, obtained
on the Pt/CeO2 catalyst prepared by physically mixing the Pt metal
particles and the CeO2 support where the contact force, namely,
metal–support interaction has been moderately reduced. The DRIFT
Fig. 9. H2 temperature-programmed reaction profiles on (a) the sole Pt metal parti-
cles and (b) the Pt/CeO2 mixed catalyst. Before starting the reaction, the NOx species
were adsorbed on the catalyst surface by passing the NO-O2 mixed gases at 473 K.
spectra indicate that the relevant intermediate adsorption species is
−
NO3 and fast hydrogenation to it may leads the high N selectivity.
2
+
+
+
+
Respective m/z = 15, 28, 30, 44, and 46 are corresponded to NH , N2 , NO , N2O , and
The high performance deNOx catalyst can be designed by selecting
the supports with suitable basicity and the metal with high affinity
+
NO2
.
against NO, H , and O2 dissociation.
2
to the CeO2 support which can serve as a NOx reservoir. Fig. 8 (c)
shows the DRIFT spectrum for the Pt/CeO2 mixed catalyst. No peak
arose from nitrate ion could be detected. Strong transmission peak
Acknowledgements
−
1
at 1200 cm
might be originated from nitrite ion and shoulder
This work was partly supported by a Grant-in-Aid for Scientific
Research on Priority Area A of “Panoscopic Assembling and High
Ordered Functions for Rare Earth Materials” from the Ministry of
Education, Culture, Sports, Science, and Technology.
−1
−1
peaks at 1120 cm and 1380 cm is assigned to hyponitrite ion
2
−
(
N O
) [16]. This spectrum profile is somewhat different from
2
2
the DRIFT spectrum reported previously in the NO-H -O2 reaction
2
system [20]. It has been proposed that NOx transforms into N2 by
−
hydrogen attacking to NO3 ion species as observed on the DRIFT
References
spectrum for ST-Pt/H-CeO2 as shown in Fig. 8(d) [7]. Disappear-
−
ance of NO peak implies that the reaction between intermediate
[1] M. Iwamoto, H. Yahiro, Y. Ta-u, S. Sundo, N. Mizuno, Shokubai 32 (1990)
430–433.
3
−
NO3 and hydrogen is too fast to detect by the DRIFT spectroscopy
[
[
2] R. Burch, P.J. Millington, A.P. Walker, Appl. Catal. B: Environ. 4 (1994) 65–94.
3] K. Sakurai, Y. Okamoto, T. Imanaka, S. Teranishi, Bull. Chem. Soc. Jpn. 49 (1976)
1732–1735.
and this behavior supposedly contributes to the N2 selectivity. Fur-
ther investigation is needed to elucidate the reaction mechanism
in detail.
[4] M. Markvart, V.L. Pour, J. Catal. 7 (1967) 279–281.
[
[
5] M. Koebel, M. Elsener, T. Marti, Combust. Sci. Technol. 121 (1996) 85–102.
6] K. Shimizu, A. Satsuma, Appl. Catal. B: Environ. 77 (2007) 202–205.
Next, effect of metal oxide supports on the NOx conversion
2
and N selectivity was examined. The Pt/Al O (123 m /g), Pt/ZrO2
[7] M. Machida, S. Ikeda, J. Catal. 227 (2004) 53–59.
[8] C.N. Costa, P.G. Savva, J.L.G. Fierro, A.M. Efstathiou, Appl. Catal. B: Environ. 75
2
2
2
3
2
(
94 m /g), and Pt/TiO (73 m /g) catalysts were prepared by phys-
2
(
2007) 147–156.
9] A. Abu-Jrai, A. Tsolakis, A. Megaritis, Int. J. Hydrogen Energy 32 (2007)
565–3571.
[10] M. Machida, S. Ikeda, D. Kurogi, T. Kijima, Appl. Catal. B: Environ. 52 (2004)
81–286.
11] M. Itoh, M. Saito, K. Motoki, J. Iwamoto, K. Machida, in preparation.
ically mixing the Pt metal particles (1 wt%) with each support
supplied from the Catalysis Society of Japan as JRC-ALO-3, JRC-
TIO-1, and JRC-ZRO-3, respectively. These catalysts resulted in the
almost same catalytic activity with the NOx conversion of 70–80%
and N2 selectivity of 20–35%. The Pt/CeO2 mixed catalyst provided
the superior NOx reduction ability compared to other catalysts. The
[
3
2
[
[12] M. Itoh, N. Tajima, M. Saito, J. Iwamoto, K. Machida, in preparation.
[13] R.F. Hicks, Q.J. Yen, A. Bell, J. Catal. 89 (1984) 498–510.

M. Itoh et al. / Journal of Molecular Catalysis A: Chemical 304 (2009) 159–165
165
[
[
[
[
14] M. Machida, S. Ikeda, D. Kurogi, T. Kijima, Appl. Catal. B: Environ. 35 (2001)
[18] A. Ueda, T. Nakao, M. Azuma, T. Kobayashi, Catal. Today 45 (1998) 135–138.
[19] N. Macleod, R.M. Lambert, Appl. Catal. B: Environ. 35 (2002) 269–279.
[20] M. Machida, M. Uto, D. Kurogi, T. Kijima, J. Mater. Chem. 11 (2001) 900–904.
[21] A.L. Petre, A. Auroux, P. Gélin, M. Caldararu, N.I. Ionescu, Thermochim. Acta 379
(2001) 177–185.
[22] D. Martin, D. Duprez, J. Phys. Chem. 100 (1996) 9429–9438.
[23] I. Balint, A. Miyazaki, K. Aika, Phys. Chem. Chem. Phys. 6 (2004) 2000–2002.
[24] B. Ioan, A. Miyazaki, K. Aika, Appl. Catal. B: Environ. 59 (2005) 71–80.
107–116.
15] W. Khaodee, B. Jongsomjit, S. Assabumrungrat, P. Praserthdam, S. Goto, Catal.
Commun. 8 (2007) 548–556.
16] S. Philipp, A. Drochner, J. Kunert, H. Vogel, J. Theis, E.S. Lox, Top. Catal. 30 (31)
(
2004) 235–238.
17] A. Martinez-Arias, J. Soria, J.C. Conesa, X.L. Seoane, A. Arcoya, R. Cataluna, J.
Chem. Soc., Faraday Trans. 91 (1995) 1679–1688.