Isothermal reduction behaviour of undoped and Ca-, K- and P-doped NiMoO4 phases used for selective propane oxydehydrogenation

Article abstract of DOI:10.1023/A:1010165210030

Isothermal reduction under hydrogen of polymorphic NiMoO₄ system has been studied at different temperatures using thermogravimetry. Evidences found by TG (oxygen depletion rate) have shown a dependence on lattice oxygen mobility and source due to the crystal structure and temperature. Relative to the undoped NiMoO₄ phases, it has been found that the presence of K₂O, CaO and P₂O₅ promoters simultaneously affects both the reduction rate and the oxygen source. Monolayer lattice oxygen mobility of undoped and doped NiMoO₄ material was investigated by oxydehydrogenation of light alkanes performed by using a periodic flow micro-reactor. The alkane activity has been found to be linked to the reduction behaviour of the different undoped and doped NiMoO₄ phases.

Full text of DOI:10.1023/A:1010165210030

Journal of Thermal Analysis and Calorimetry, Vol. 63 (2001) 267–277
ISOTHERMAL REDUCTION BEHAVIOUR
OF UNDOPED AND Ca-, K- AND P-DOPED NiMoO4
PHASES USED FOR SELECTIVE PROPANE
OXYDEHYDROGENATION
A. Kaddouri*, R. Del Rosso, C. Mazzocchia and D. Fumagalli
Dipartimento di Chimica Industriale ed Ingegneria Chimica,‘G. Natta’ Politecnico di Milano,
P.za Leonardo Da Vinci, 32 – 20133 Milan, Italy
(Received September 18, 2000)
Abstract
Isothermal reduction under hydrogen of polymorphic NiMoO4 system has been studied at different
temperatures using thermogravimetry. Evidences found by TG (oxygen depletion rate) have shown
a dependence on lattice oxygen mobility and source due to the crystal structure and temperature.
Relative to the undoped NiMoO4 phases, it has been found that the presence of K2O, CaO and
P2O5 promoters simultaneously affects both the reduction rate and the oxygen source.
Monolayer lattice oxygen mobility of undoped and doped NiMoO4 material was investigated
by oxydehydrogenation of light alkanes performed by using a periodic flow micro-reactor. The
alkane activity has been found to be linked to the reduction behaviour of the different undoped and
doped NiMoO4 phases.
Keywords: isothermal reduction, lattice oxygen, polymorphic nickel molybdate, red-ox properties
Introduction
Inorganic metal-oxygen materials are of great interest in heterogeneous catalysis and
ceramic chemistry. One important aspect of the chemistry of these compounds is
their thermal behaviour, lattice oxygen mobility and structure defects [1, 2].
In previous works, unpromoted and promoted molybdate catalysts were pre-
pared by different techniques [3–5] but, to our knowledge, experimental data are
lacking on isothermal reduction of α and β phases of stoichiometric NiMoO4 doped
catalysts. The NiMoO4 compound is interesting as active catalyst for the oxidation of
light alkanes to their corresponding alkenes [6, 7]. In more recent works, we verified
and completed the study of the behaviour of all NiMoO4 phases working in a red-ox
cyclic mode [8, 9].
*
Phone: (+39) 02 23 99 32 47, Fax: (+39) 02 70 63 81 73, E-mail: akim.kaddouri@polimi.it
1418–2874/2001/ $ 5.00
Akadémiai Kiadó, Budapest
© 2001 Akadémiai Kiadó, Budapest
Kluwer Academic Publishers, Dordrecht

268
KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
Hereafter, we report the observed oxygen depletion rate during isothermal re-
duction under hydrogen of all undoped and K2O–, CaO– and P2O5–NiMoO4 phases.
The structural, superficial acid-base properties and activity of each phase are com-
pared.
Experimental
NiMoO4 preparation
A Mettler RC1 Reaction Calorimeter was used allowing a careful control of the main
experimental parameters (T, stirring speed, pH, etc.) which determine the structure
and surface characteristics of the active catalyst [3]. The stoichiometric catalyst
NiMoO4 was prepared by co-precipitation from an equimolar molybdic acid and
nickel nitrate solution (0.25 mol l–1), at 85°C with the pH adjusted at 5.25 by addition
of ammonia. The resulting precipitate was hot filtered, dried at 120°C for 15 h and
then heated at 550°C for 2 h leading to α-NiMoO4 formation. The β-NiMoO4 phase
was obtained in situ by thermal treatment of the α-phase up to 720°C which was then
cooled to reaction temperatures. The α’-phase is obtained by heating beyond 720°C
and complete cooling to ambient temperature. The catalyst particle size was found to
be in the range of 200–325 mesh.
Promoted catalysts
Incipient wetness impregnation technique was used to prepare K2O, CaO and P2O5
promoted NiMoO4 catalysts. The impregnation was made to start from stoichiometric
NiMoO4 activated (at T=550°C) catalyst and KOH, Ca(NO3)2, and (NH4)2HPO4 salts.
A further treatment at 550°C for two hours allowed to obtain the final catalyst. The
dopant percentages indicate the mass of the doping element with respect to the total
mass of the catalyst.
Thermal studies of the compounds were carried out on a Seiko TG-DTA instru-
ment. Weighed samples (ca. 20 mg) were placed in alumina crucibles and reduction
tests were carried out isothermally under a hydrogen atmosphere (gas feed 3 l h–1)
while recording the TG and DTG curves. For α- and α’-phases the samples were pre-
heated (10°C min–1) in air up to the reduction temperature (Tr) and were then washed
(20 min) with helium. For β-phase the sample was heated (10°C min–1) up to 720°C
and cooled (10°C min–1) to the reduction temperature (Tr) under air and was then
washed (20 min) with helium prior to the introduction of hydrogen.
Results and discussion
The difficulty of removing oxygen from the lattice of a solid is often related to the
strength of the metal-oxygen bond. The weaker the bond is, the higher the lattice oxy-
gen mobility is and vice versa. However, in selective oxidation reactions, it is gener-
ally observed that the activity and selectivity of an oxide are related to its reducibility.
J. Therm. Anal. Cal., 63, 2001

KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
269
High resistance to reducibility leads to a lower activity, while easier reducibility im-
plies a higher activity but also a lower selectivity.
At atmospheric pressure the system studied, stoichiometric nickel molybdate
presents two polymorphic phases: the low-temperature α-phase, with molybdenum in
a distorted octahedral environment [10, 11], and the high-temperature β-phase, with
molybdenum in a distorted tetrahedral site [12]. The β-phase is formed after heating
the α-phase up to 720°C and undergoes reverse transition at low temperature on cool-
ing at ca. 200°C. The complete cooling to ambient temperature leads to the formation
of the α’-NiMoO4 phase which possesses the same structure of the α-phase, with mo-
lybdenum in a distorted octahedral environment. The α’-phase differs only in its spe-
cific surface area which is lower (13 vs. 33 m2 g–1 for the α-phase) due to the thermal
treatment at high temperature (720 vs. 550°C). The α→β transition occurs without
mass loss. A progressive MoO3 sublimation from NiMoO4 may be observed at tem-
perature higher than 720°C. In Fig. 1 the DTA cycle of phase transitions of the stoi-
chiometric NiMoO4 system is shown.
Fig. 1 DTA cycle of stoichiometric NiMoO4 phase transitions
The isothermal reduction of NiMoO4 based systems was performed under hy-
drogen at different temperatures. In Fig. 2 the effect of temperature on reduction be-
haviour of pure α, β and α’-NiMoO4 phases with a constant H2 concentration is
shown.
Noticeable differences in terms of reduction rate vs. temperature for each phase
can be evidenced. These variations are due to the increase of oxygen mobility with
temperature. A comparison of the reduction trends of the three phases, for example at
460°C (Fig. 3), shows that the depletion rate evolve in the following order
β-NiMoO4>α’-NiMoO4>α-NiMoO4. These behaviours may be attributed to the
weaker strength of M–O bonds in the corresponding structures.
Regarding the general reduction trend, it can be seen that NiMoO4 presents two
steps of reduction with different slopes which indicate a consecutive depletion of dif-
ferent types of oxygen. In order to better investigate the observed behaviours we have
reported in Fig. 4 the DTG vs. TG curves profile of the pure phases i.e. the reduction
rate vs. reduction degree.
J. Therm. Anal. Cal., 63, 2001

270
KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
Fig. 2 Effect of temperature on reduction behaviour of NiMoO4
Fig. 3 TG mass loss vs. time for α-, α’- and β-NiMoO4 under hydrogen at 420°C
J. Therm. Anal. Cal., 63, 2001

KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
271
Fig. 4 Reduction rate vs. reduction extent (i.e. DTG vs. TG curves) of NiMoO4 pure
phases at 460°C
These results lead to hypothesise that in all cases there are at least two oxygen
types which could be extracted consecutively during reduction. These two oxygen
bonds could be attributed to Ni–O and Mo–O metal oxides present in the structure of
NiMoO4 since the oxygen coordination of the two metals is different. Nickel is sur-
rounded by four oxygen atoms in a tetrahedral symmetry while Molybdenum coordi-
nation is six in an octahedral symmetry.
In the case of the β-phase where both nickel and molybdenum are in a tetrahe-
dral symmetry, these two peaks are partially overlapped and inverted.
The products composition obtained from α-NiMoO4 reduction have not been es-
tablished unambiguously. In the literature some mechanisms are proposed for the re-
duction of the Ni–Mo–O system, but the intermediates as well as the final products of
reduction differ from author to author [13–17]. In order to obtain a deeper under-
standing of our systems, a study of both NiMoO4 phases reduction was performed
with the main objective of identifying the reduction products. In this sense, XRD
analyses were performed after isothermal reduction runs at selected temperatures
420–480°C. The XRD spectra showed the presence of three broad peaks characteris-
tic of metallic nickel and/or of the intermetallic Ni4Mo (JCPDS cards 4–850 and
3–1036, respectively). The formation of the intermetallic Ni4Mo (d=2.052Å) was
also reported by Brito et al. [13], whereas other authors [14] reported that such a com-
pound can be a reduction product of anhydrous nickel molybdate.
A deep reduction performed at higher temperatures leads to peaks characteristic
of MoO2 (JCPDS card 5-452). This phase was reported in almost all work regarding
NiMoO4 reduction, however, some authors found amorphous material. Kipnis et al.
[15] proposed that the reduction of α-nickel molybdate by hydrogen at 400–500°C
leads to the formation of a Ni–Mo alloy and a finely dispersed lower Mo oxide phase
J. Therm. Anal. Cal., 63, 2001

272
KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
that crystallises as MoO2 when heated at 600°C. According to Brito et al. [13] reduc-
tion of MoO3 to MoO2 under dynamic conditions of TPR proceeds with the formation
of highly amorphous oxide. These facts explain the broadness of the peaks in the
X-ray diffractograms and the absence of the MoO2 characteristic lines.
Since the molybdenum in α-NiMoO4 and MoO2 is in an octahedral co-ordina-
tion, it is most probable that the MoO2 phase formed by low temperature reduction of
α-NiMoO4 is built up of MoO6 octahedra connected irregularly by both vertices and
edges. On the other hand, it is known that the formation of a solid solution of molyb-
denum in nickel leads to an increase in the lattice constants, manifested in a shift of
the nickel lines on the X-rays diffractogram towards smaller angles. On the basis of
our data, when the α-NiMoO4 is subjected to reduction (H2) and heating (420<T<
480°C), the following reduction scheme can be proposed: NiMoO4→Ni+Ni4Mo+
amorphous MoO2. At temperatures lower than 500°C the reduction of α-NiMoO4 has
led to the formation of an unordered nickel–molybdenum alloy with a concentration
dishomogeneity in the distribution of Mo and Ni. Referring to the β-NiMoO4 phase,
the observed reduction behaviour was found to be different from the α-NiMoO4 one;
both reduction trend and rate are different (Figs 2 and 4). The reduction scheme re-
ported above may be still valid for the β-phase but with a different reduction products
concentration.
In the light of these findings and in order to get a deeper understanding of the de-
pletion mechanism of NiMoO4 system, different basic promoters, such as CaO and
K2O (usually used for alkanes oxydehydrogenation reaction) were introduced in
NiMoO4 systems by impregnation technique. After calcination of the impregnated
samples at 550°C, the effect of these doping element on the depletion rate was studied
by TG-DTG analyses. As an example, CaO and K2O doped α-NiMoO4 systems are
reported in Fig. 5.
Fig. 5 TG-DTG of undoped and CaO or K2O doped α-NiMoO4 system
J. Therm. Anal. Cal., 63, 2001

KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
273
It can be noticed that, following the addition of the promoters, the reduction rate
decreases indicating that the mobility of the oxygen has become less important. With
respect to the easier oxygen depletion rate of NiMoO4 pure phases, the reduction rate
is restrained by the introduction of basic promoters.
In the case of calcium doped catalyst a higher percent (ca. 10 fold increase) of
the dopant was necessary to obtain the same basicity as that of K2O doped catalysts
[9]. It should also be observed that although the basicity is the same, the reduction
rates of NiMoO4/0.1% K and α-NiMoO4/1% Ca are different. CaO load has a more
significant effect on the reduction rate than K2O load.
In order to obtain a better understanding of the relationship between acid-base prop-
erties and reducibilty of the catalysts, it was decided to introduce a second acidic pro-
moter, P2O5, which is often used in ODH reaction for increasing alkanes conversion to
counterbalance the effect of the cited basic promoters. The reduction behaviour of the
Ca–P/α-NiMoO4 and K–P/α-NiMoO4 prepared systems was studied. If compared to the
CaO and K2O-doped catalysts, the reduction profile of Ca–P/α-NiMoO4 and K–P/
α-NiMoO4 showed that the depletion rate was subsequently lowered in presence of P2O5
(Fig. 6).
Fig. 6 TG-DTG of undoped and P, Ca–P and K–P doped α-NiMoO4 system
Referring to the reduction rate vs. reduction degree curves (Figs 4–6), it is also im-
portant to notice that while no significant change was observed in the general trend of
DTG vs. TG curves when the pure phases were doped with Ca or K (Fig. 5), the addition
of phosphorus on pure phases (P/α-NiMoO4) and on doped NiMoO4 (Ca–P/α-NiMoO4
and K–P/α-NiMoO4) leads to an inhibition of the second peak formation (Fig. 6). This is
probably due to a rearrangement of the oxygen network so that only one oxygen type is
available for reduction. A possible explanation can be that a reaction between metallic
nickel, formed during reduction, and phosphorus oxide occurs yielding a new crystalline
phase Ni2P2O7 (pyrochlore compounds) not detectable by XRD due the low amounts
formed.
J. Therm. Anal. Cal., 63, 2001

274
KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
These results suggest that the introduction of P2O5 promoter decreases the reduction
rate of the Ca/α-NiMoO4 and K/α-NiMoO4. This reduction rate does not reach the same
level as for P2O5/α-NiMoO4 catalyst as the promoters basicity (Ca and K) decreases the
interaction between P2O5 and NiMoO4. Consequently, the curves in Fig. 6 are shifted to-
ward the P2O5/α-NiMoO4 curve, but above those of Ca/α-NiMoO4 and K/α-NiMoO4
(Fig. 5). The same reduction experiments under hydrogen carried out with α-NiMoO4/
0.1% K, α-NiMoO4/1% Ca, Ca–P/α-NiMoO4 and K–P/α-NiMoO4 were performed with
the corresponding doped β-phase systems. The data obtained (not reported) showed that
depletion rate was drastically lowered and in particular with Ca–P and K–P doped cata-
lysts.
The catalytic effect of phosphorus was studied by preparing a catalyst where
only P2O5 is present (P/NiMoO4). The effects of the acidic promoter P2O5 on both the
α→β phase transition under air, and the reducibility of the β-NiMoO4 phase under
hydrogen were studied by using the thermogravimetric technique.
Effect on the a→b phase transition
In the case of P/α-NiMoO4 (i.e activated under air at T≤550°C) the phosphorus main-
tains its initial structure P2O5 while if this system is treated in order to form the
β-phase (i.e. activated under air at T≥720°C) the phosphorus oxide would appear to
be present in another crystalline form. The phosphorus oxide seems to accelerate the
sublimation of the MoO3 contained in the NiMoO4 network (competition in terms of
acidity between MoO3 and P2O5). At ca. 590°C an abrupt mass loss of ca. 5% is re-
corded (Fig. 7). This mass loss was not observed for the undoped catalysts or for the
catalysts doped with both phosphorus and CaO or K2O. This fact can be explained in
terms of the acid-base affinity between basic promoters (CaO or K2O) and acidic
phosphorus which could limit the interaction between phosphorus and NiMoO4 at
high temperature. The α→β transition occurs in both doped and undoped NiMoO4
systems by a rearrangement of the oxygen co-ordination around the molybdenum.
During the transition both Ni–O and Mo–O bonds in the network are weakened.
Effect on the reducibility of b-NiMoO4
After cooling the P/β-NiMoO4 (prepared activating P/α-NiMoO4 under air at
T≥720°C) to the reduction temperature, tests carried out under hydrogen show that
P2O5 in P/α-NiMoO4 (Fig. 6) and P/β-NiMoO4 (Fig. 7), lowers the depletion rate with
respect to the undoped catalysts, particularly in the case of the β-phase as for the cata-
lysts twice doped (Ca–P/NiMoO4 and K–P/NiMoO4).
These results evidence a strong interaction between the P2O5 and the NiMoO4
system although the reduction of phosphorus to its metallic state is highly unlikely.
The formation of a stable P–O–Mo and/or P–O–Ni surface compound during the ini-
tial catalyst reduction would account for these findings. This is also consistent with
the observation that only relatively small additions of phosphorus to the CaO or K2O
doped catalysts are needed to produce the maximum effect on propene synthesis.
J. Therm. Anal. Cal., 63, 2001

KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
275
Fig. 7 Mass loss of NiMoO4 and P/NiMoO4, during α-β phase transformation under air
and during reduction under hydrogen
Catalytic behaviour of the catalysts under propane atmosphere
Catalytic tests performed in a periodic-flow micro-reactor [8] have shown (Table 1)
that although the presence of P2O5 did not increase the overall reduction rate of the
catalysts (Ca–P and K–P doped NiMoO4) the propane conversion is enhanced with
respect to Ca and K doped catalysts. These observations can be supported by the fact
that P2O5 is a typical acidic element (with no oxidising power) which induce an in-
crease of the NiMoO4 acidity leading to an increase of propane adsorption on the
NiMoO4 surface. As a consequence a higher propane conversion is obtained after re-
action with monolayer oxygen of the catalyst. The possible contribution of the new
Ni2P2O7 phase formed could also be taken into account since in oxidative dehydro-
genation of isobutane into isobutene higher yields are found [18, 19].
Table 1 Activity and selectivity of undoped and doped α and β-NiMoO4
Catalyst
T=500°C
Conversion/% Selectivity/%
C3H6 yield
(mmol min–1gcat)
Pulse period/s
C3H8
3.70
1.61
3.73
2.64
3.77
1.70
2.78
2.85
C3H6
74.62
89.35
84.50
81.15
76.42
94.40
94.01
87.86
α-NiMoO4
40
40
40
40
40
40
40
40
0.074
0.038
0.084
0.058
0.080
0.042
0.070
0.066
α-NiMoO4/0.1%K
α-NiMoO4/0.1%K-2%P
α-NiMoO4/1%Ca
α-NiMoO4/1%Ca-2%P
β-NiMoO4
β-NiMoO4/0.1%K-2%P
β-NiMoO4/1%Ca-2%P
Contact time : 0.3 g L–1 h–1 , helium washing gas feed : 3.6 L h–1
J. Therm. Anal. Cal., 63, 2001

276
KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
Referring to the apparent activation energy the data obtained for the solid reduc-
tion compared with those obtained for propane oxydehydrogenation using only lat-
tice oxygen (pulse reactor) are quite close and are in the range of 58.5–79.4 kJ mol–1.
This interesting fact implies that some of M–O bonds are broken selectively and
the surface oxygen vacancies created in the solid reduction step are related partly to
the type of oxygen involved in the propane oxidation which can be removed and re-
placed during the reduction-oxidation steps.
As the catalytic results evidenced the longer the period (reduction degree) is, the
best the alkene yield is; it is, therefore, possible to correlate the reduction behaviour
of NiMoO4 based system with the catalytic activity. It can also be noted that an eluci-
dation of the structure of the amorphous molybdenum oxide phase together with a hy-
pothetical presence of a newly formed Ni2P2O7 phase can be of great interest, since
their catalytic properties are important.
Conclusions
Source and mobility of polymorphic NiMoO4 lattice oxygen has been studied by iso-
thermal reduction under hydrogen at different temperatures where the oxygen deple-
tion rate has been evidenced to be dependent of the NiMoO4 phases. The presence of
K2O, CaO and P2O5 promoters simultaneously affect the reduction rate and oxygen
source. Monolayer lattice oxygen mobility of undoped and doped NiMoO4 materials
has also been investigated by propane oxydehydrogenation. The catalytic activity
measured with a periodic flow micro-reactor was found to be related to the reducibil-
ity of the catalysts, which may be explained in terms of their corresponding Mo and
Ni oxygen co-ordinations. The presence of P2O5, typical element, increases the sur-
face acidity, decreases subsequently the overall reduction rate of the pure and doped
catalysts (Ca–P and K–P doped NiMoO4) and enhances the propane conversion.
* * *
Financial support of CNR and of Polytechnic of Milano through ‘assegno di ricerca’ to A. Kaddouri
is gratefully acknowledged.
References
1 P. Mars and D. W. Van Krevelen, Chem. Eng. Sci. (Spec. Suppl.), 3 (1954) 41.
2 C. N. R. Rao and B. Raveau, in ‘Transitions Metal Oxides; Structures, Properties and Synthesis
of Ceramic Oxides’ 2nd edition, John Wiley & Sons, USA 1998.
3 C. Mazzocchia, R. Anouchinsky, A. Kaddouri, M. Sautel and G. Thomas, J. Thermal Anal.,
40 (1993) 1253.
4 R. Anouchinsky, A. Kaddouri and C. Mazzocchia, J. Thermal Anal., 47 (1996) 299.
5 C. Mazzocchia, A. Kaddouri, R. Anouchinsky, M. Sautel and G. Thomas, J. Solid. Stat. Ionics,
63–65 (1993) 731.
6 C. Mazzocchia, E. Tempesti and Ch. Aboumrad, Eur. Patent, 90-400137, 1990 and US Patent
5086032, 1992.
J. Therm. Anal. Cal., 63, 2001

KADDOURI et al.: SELECTIVE PROPANE OXYDEHYDROGENATION
277
7 C. Mazzocchia, Ch. Aboumrad, C. Diagne, E. Tempesti, J. Herrmann and G. Thomas, Catal.
Lett., 10 (1991) 181.
8 R. Del Rosso, A. Kaddouri, R. Anouchinsky, C. Mazzocchia, P. Gronchi and P. Centola,
J. Mol. Cat., 135 (1998) 181.
9 A. Kaddouri, R. Del Rosso, C. Mazzocchia, P. Gronchi and P. Centola, Catal. Lett., 63 (1999)
65.
10 G. W. Smith, Acta. Cryst., 15 (1962) 1054.
11 G. W. Smith and J. A. Ibers, Acta. Cryst., 19 (1965) 269.
12 G. W. Smith and B. L. Chamberland, Inorg. Chem., 7 (1968) 1672.
13 J. L. Brito, J. L. Laine and K. C. Pratt, J. Mater. Sci., 24 (1989) 425.
14 M. A. Tsurov, P. V. Afanasiev and V. V. Luniv, Appl. Catal. A, 105 (1993) 205.
15 M. A. Kipnis, D. A. Agievskii, Kin. Katal., 22 (1981) 1252.
16 A. I. Vagin, N. V. Burmistrova and V. I. Erofeev, React. Kinet. Catal. Lett., 28 (1985) 47.
17 T. S. Ismailov, G. Sh. Talipov and N. Sh. Inoyatov, Catalytic Processing of Hydrocarbon
Feedstocks, No. 5, Tashkent, 1971, p. 91.
18 Y. Takita, K. Kurosaki, Y. Misuhara and T. Ishihara, Chem. Lett., (1993) 335.
19 Y. Takita, K. Sano, K. Kurosaki, N. Kawata, H. Nishiguchi, M. Ito and T. Ishihara, Appl.
Catal. A, 167 (1998) 49.
J. Therm. Anal. Cal., 63, 2001