Catalysis Communications 10 (2009) 459–463
Contents lists available at ScienceDirect
Catalysis Communications journal homepage: www.elsevier.com/locate/catcom
Characterization and reactivity of Nb2O5 supported Ru catalysts Komandur V.R. Chary *, Chakravartula S. Srikanth, Vattikonda Venkat Rao Catalysis Division, Indian Institute of Chemical Technology, Hyderabad, India
a r t i c l e
i n f o
Article history: Received 29 March 2008 Received in revised form 19 September 2008 Accepted 9 October 2008 Available online 18 October 2008 Keywords: Hydrodechlorination 1,2,4-Trichlorobenzene Ruthenium Niobium oxide Dispersion
a b s t r a c t A series of Ru/Nb2O5 catalysts with Ru loadings from 0.5–7.18 wt.% were prepared and characterized by X-ray diffraction (XRD), temperature programmed reduction (TPR), temperature programmed desorption (TPD) of hydrogen and CO chemisorption measurements. XRD patterns reveal the presence of RuO2 from 3.0 wt.% Ru loading. TPR proﬁles show only one peak due to reduction of RuO2 to Ru0. The H2-TPD exhibits two peaks, one due to desorption of hydrogen from Ru metal and another due to spill over from metal to the support. The catalytic properties were evaluated for the vapor-phase hydrodechlorination of 1,2,4trichlorobenzene and related to Ru dispersion. Ó 2008 Published by Elsevier B.V.
1. Introduction Aromatic chlorides have been commonly used as solvents, insecticides, fungicides, and intermediates in many organic reactions. Due to their strong resistance to biodegradation , and toxic nature they are causing environmental and health concerns. These compounds are highly accumulative and environmentally persistent causing carcinogenic and mutagenic activity . Disposal of aromatic chlorides by incineration is a energy demanding process and moreover it is difﬁcult to decompose them completely to CO2, H2O and HCl. Incomplete combustion may lead to release of toxic furan/dioxin  products, which are not readily bio-degradable . Alternatively, catalytic hydrodechlorination is emerging as an alternative low energy and non-destructive process for recovery of the valuable feed-stocks . It is also proved to be a significant process for the abatement of chlorinated organic wastes under mild conditions. Advantages of HDC process include the operation at low temperatures and pressures without producing any harmful side products and also more selective towards chlorine removal with high conversion. Catalytic hydrodechlorination has been extensively studied with various chloro-organics over metals like Pd [5,6], Pt [7,8], Rh , and Ni [10,11] supported on various supports. The factors like support material, metal particle size, and the reaction conditions are found to inﬂuence both the activity and selectivity of hydrodechlorination reactions. Ruthenium is used as the most promising catalyst  in partial hydrogenation of benzene to cyclohexene. It is also well known as * Corresponding author. Tel.: +91 40 27193162; fax: +91 40 27160921. E-mail address: [email protected]
(K.V.R. Chary). 1566-7367/$ - see front matter Ó 2008 Published by Elsevier B.V. doi:10.1016/j.catcom.2008.10.006
very active catalyst for the hydrogenation of CO to hydrocarbons  and ammonia synthesis . Ruthenium is very selective in hydrogenation of [email protected]
group in the vicinity of conjugated or isolated double bonds or an aromatic ring . In addition to a variety of its applications in hydrogenation reactions, ruthenium is also used in hydrodechlorination reactions [16,17]. Niobium based material have been employed recently as catalysts in numerous catalytic applications [10,18–23]. There are various functions of niobium compounds in heterogeneous catalysis . Niobia can be used as a support, promoter and solid acid and is easily reducible over a wide temperature range. Niobia is also known as a typical strong metal-support interacting (SMSI) oxide . In the present investigation, we report for the ﬁrst time a systematic study of Ru/Nb2O5 catalysts for the hydrodechlorination of 1,2,4trichlorobenzene and the characterization of catalysts using various techniques like X-ray diffraction (XRD), temperature programmed reduction (TPR), temperature programmed desorption (TPD) of H2 and pulse CO-chemisorption.
2. Experimental A series of ruthenium catalysts with Ru loadings in the range of 0.5–7.18 wt.% supported on niobium oxide were prepared by wet impregnation with an aqueous solution containing RuCl3 3H2O. Prior to ruthenium impregnation, the Nb2O5 was prepared by calcination of hydrated niobia (CBMM, Brazil HY-340, surface area 55 m2/g) at 773 K for 5 h. After impregnation, the samples were dried at 383 K for 16 h and subsequently calcined at 773 K for 5 h in air.
K.V.R. Chary et al. / Catalysis Communications 10 (2009) 459–463
X-ray diffraction patterns were obtained on a Rigaku miniﬂex diffractometer using graphite ﬁltered Cu Ka radiation. The speciﬁc surface area of the catalysts was measured by N2 adsorption at 77 K by BET method taking 0.0162 nm2 as its crosssectional area on a Pulse Chemisorb 2700 (Micromeritics, USA). Temperature programmed reduction (TPR) experiments were carried out on an Auto Chem 2910 (Micromeritics, USA) instrument. In a typical experiment, ca. 150 mg of calcined catalysts sample was taken in a U-shaped quartz sample tube. Prior to TPR studies the catalyst was pretreated with an inert gas (He, 50 ml/ min) at 473 K and was cooled to ambient temperature. A stream of carrier gas consists of 5% H2–Ar (50 ml/min) was allowed to pass over the sample while heating from ambient to 873 K at the rate of 10 K/min. H2-temperature programmed desorption (TPD) experiments were carried out using Autochem 2910 (Micromeritics, USA). In a typical TPD experiment, ca. 200 mg of the calcined sample was taken in a U-shaped quartz sample tube supported on quartz wool bed. The sample was reduced in a ﬂow of pure hydrogen (50 ml/ min) at a temperature of 673 K for 2 h. Subsequently it was ﬂushed with Ar (50 ml/min) at the same temperature for an hour and cooled to ambient temperature in same gas ﬂow. The sample was saturated by 5% H2–Ar ﬂow (50 ml/min) for an hour at room temperature and ﬂushed with Ar (50 ml/min) at the same temperature for 2 h to remove all the physisorbed hydrogen. The sample was again heated from ambient temperature to 873 K in Ar (50 ml/min) ﬂow at a heating rate of 10 K/min monitoring hydrogen concentration in the efﬂuent stream with the thermal conductivity detector. CO chemisorption measurements were also carried out Auto Chem 2910 instrument. Prior to adsorption measurements, ca. 150 mg of the sample was reduced in a ﬂow of hydrogen (50 ml/ min) at 673 K for 2 h and ﬂushed subsequently in a pure argon ﬂow for an hour at 673 K and cooled to ambient temperature in the same gas ﬂow. CO uptake was determined by injecting pulses of 9.96% CO–He from a calibrated on-line sampling valve into helium stream passing over reduced samples at 673 K. Ruthenium metal surface area, dispersion and average particle size were calculated assuming the stoichiometric factor (CO/Rus) = 1. Adsorption was considered as complete after three successive pulses showed similar peak areas. Vapour phase hydrodechlorination of 1,2,4-trichlorobenzene (P99% Aldrich chemicals) was carried out in a vertical down-ﬂow glass reactor under normal atmospheric pressure. ca. 0.8 g of the catalyst, diluted with an equal amount of quartz grains was packed between the layers of quartz wool. Prior to the reaction, the catalyst was reduced in a ﬂow of hydrogen (50 ml/min) at 673 K for 3 h. The reactor was fed with 1,2,4-trichlorobenzene (P99% pure Aldrich) in the temperature ranges from 503–573 K. The product stream was allowed to pass thorough (0.1 M) NaOH solution prior to the collection of the products in the cold trap, where the HCl in the product stream is converted to NaCl. Thus making the products free from HCl formed prior to the GC analysis. The reaction products were analyzed by a HP-6890 gas chromatograph equipped with a HP-5 capillary column with a ﬂame-ionization detector (FID).
3. Results and discussion X-ray diffraction patterns of pure niobia, RuO2 and Ru supported on niobia with varying Ru contents are presented in Fig. 1. It can be seen from the Fig. 1 that all the samples (except RuO2) show XRD peaks due to low temperature form of niobia at 2h = 22.5°, 28.4°, 36.6°, 46° and 55.3°, as reported by Ko and Weissman . The XRD peaks corresponding to RuO2 are observed from
Ru Wt% RuO2
7.18% 5.0% 3.0% 1.5% 0.5% Pure Niobia
2 Theta Fig. 1. XRD patterns of pure Nb2O5, RuO2 and various RuO2/Nb2O5 catalysts.
3.0 wt.% catalysts in addition to the intense peaks of niobia. The intensity of XRD peaks due to RuO2 at 2h = 28.2°, 35.3° and 54.6° (d = 3.17, 2.554, 1.686, respectively)  (JCPDS 21-1172) increases with ruthenium loading in the catalyst (shown in Fig. 1, with dark circle). However XRD peak of RuO2 at 2h =28.2° (d = 3.175) is not well resolved as it is overlapped with the peak of Nb2O5. However, an increase in the intensity of Nb2O5 reﬂection at 28.4° is noticed. The absence of crystallographic peaks of RuO2 below 3.0 wt.% Ru catalyst might be due to the presence of ruthenium oxide in a highly dispersed amorphous state on the niobia support. The reducibility of Ru supported on niobia was investigated by TPR experiments, and the proﬁles along with pure Nb2O5 are shown in Fig. 2 and the results are summarized in Table 1. From the TPR proﬁles it is clear that the pure niobia was not shown any sign of reduction up to 873 K. However, Ru/ Nb2O5 catalyst samples show a sharp reduction peak between 466 and 498 K. RuO2 has only one reduction peak i.e., Ru in RuO2 changed directly from Ru4+ to Ru0  without forming intermediate valence states such as Ru3+ or Ru2+. As bulk ruthenium displays only one peak in the thermogram located at about 490 K [27,28] it can be assumed that the peak observed is due to the reduction of bulk RuO2. The Tmax of 0.5 wt.% Ru/Nb2O5 sample appeared slightly at low temperature (466 K) compared to other catalysts. This is due to smaller crystallite size (4 nm or less) of RuO2 at lower loading as evidenced from XRD and CO chemisorption methods. From the Fig. 2 it can be seen that the intensity of the reduction peak increases with ruthenium loading. The amount of hydrogen required for the reduction was calculated on the basis of the following theoretical hydrogen consumption.
RuO2 þ 2H2 ! Ru0 þ 2H2 O The amount of hydrogen consumed during the reduction of RuO2 is higher than the theoretical value and it could be probably due to spill over of hydrogen from the metal to the support. Temperature programmed desorption of hydrogen was performed on all the catalysts to obtain the information on the surface structure of the catalyst. The mechanism of adsorption-desorption of hydrogen is extremely complex especially over supported metal catalysts, because this is a phenomenon related to the interaction between the active phase and support. Fig. 3 shows the TPD proﬁles of H2 desorbed from ruthenium catalysts supported on niobia where the H2 adsorptions were carried out at ambient temperature. The TPD proﬁles in Fig. 3 show two desorption peaks with a ﬁrst peak maxima appeared at 350–360 K, and the area of this peak
K.V.R. Chary et al. / Catalysis Communications 10 (2009) 459–463
Fig. 2. TPR proﬁles of Nb2O5 and various RuO2/Nb2O5 catalysts.
Table 1 Results of temperature programmed reduction for various Ru/Nb2O5 catalysts. Ru Wt%
T max (K)
H2 aConsumption lmol/g
H2 bConsumption lmol/g
0.5 1.5 3.0 5.0 7.18
466.8 492.5 498.9 497.3 495.2
103.4 309.6 597.8 1016.3 1493.7
75.1 225.4 455.3 751.4 1075.0
H2 consumption from TPR. H2 consumption obtained theoretically.
corresponds to the H2 desorbed from Ru metal . The second peak in the TPD proﬁles appeared in a wide range of temperatures depending on the extent of metal-support interaction between Ru and niobia. The Tmax of second peak shifts to lower temperature at 1.5 wt.% Ru and merges at higher loadings due to decrease in the metal-support interaction with Ru loadings. Similar kind of results was also observed for Ru/TiO2 . The second peak in TPD again appeared at higher loadings of Ru, due to agglomerization of the metal that leaves more of the support surface available for H2 adsorption. The area under the second peak corresponds to the hydrogen spillover from Ru metal to support. During the desorption process, these hydrogen atoms on the support would migrate back to metal, then recombine with hydrogen molecules for desorption. Hydrogen spillover is related to the metal-support interaction and the perimeter of the interface between metal and support. Thus, the desorption peak appeared at high temperature was due to the hydrogen spill over from the metal to the support and the rate of hydrogen spillover decreased with metal dispersion.
Fig. 3. H2-TPD proﬁles of various RuO2/Nb2O5 catalysts.
The BET surface areas of various Ru/Nb2O5 catalysts are reported in Table 2. The speciﬁc surface area of pure Nb2O5 was found to 55 m2/g. The decrease in the BET surface area of Ru/ Nb2O5 catalysts with the increase in ruthenium loading is presumably as a result of pore blocking by the crystallites of RuO2 as shown by the results of X-ray diffraction. CO chemisorption is used to ﬁnd the dispersion, metal area and crystallite size of Ru supported on niobia and the results are presented in Table 2 along with BET surface area and turn over frequencies (TOF). The TOF’s are calculated form chemisorption data as the number of molecules of 1,2,4-trichlorobenzene converted by one surface Ru atom per second using following equation:
Rate ¼ ðVolume of reactant fed Fractional conversionÞ=weight of the catalyst TOF ¼ Rate=CO-uptake The conversion at 4th hour is taken for the sake of comparison. From the results of Table 2 it is clear that the dispersion and metal area (Ru) are decreasing with increase of Ru loading. The crystallite size of ruthenium increases with Ru loading on the support due to the formation of crystalline RuO2. The crystallite size of Ru measured from CO chemisorption is in good agreement with XRD results. Hydrodechlorination of 1,2,4-trichlorobenzene was carried out at 523 K on various Ru catalysts and the results are presented in Table 3. The results suggest that the conversion of 1,2,4-trichlorobenzene, and selectivity towards benzene increase with increasing Ru loading on niobia. However, the selectivity towards chlorobenzene decreases with Ru loading where as the selectivity for dichlo-
K.V.R. Chary et al. / Catalysis Communications 10 (2009) 459–463
Table 2 CO chemisorption and TOF values for various Ru/Nb2O5 catalysts. Wt.% of Ru
BET surface area (m2 g1)
CO uptake (lmol g1)
Metal area (m2/ g(cat))
Metal area (m2/ g(Ru)
Crystallite size (nm)
TOF 102 s1
0.5 1.5 3.0 5.0 7.18
53.1 49.8 46.3 41.2 38.1
12.1 26.8 43.3 49.0 52.0
25.7 19.2 15.5 10.4 7.8
0.59 1.33 2.13 2.40 2.55
119 88 71 48 36
4.0 5.5 6.8 10.1 13.6
0.57 0.70 0.99 1.09 0.99
Table 3 Effect of ruthenium loading on conversion/selectivity for hydrodechlorination of 1,2,4-trichlorobenzene. Ru wt.%
0.5 1.5 3.0 5.0 7.18
4 11 25 32 34
(%) Selectivity Benzene
0.5 3 49.9 63.0 53.5
92.5 84.5 44.6 36.0 46.0
7 12.5 5.5 1.0 0.5
robenzene decreases marginally at higher loadings. Reaction products like cyclohexane, chlorocyclohexane or meta-dichlorobenzene were not detected during hydrodechlorination of 1,2,4-trichlorobenzene. The fact that meta-dichlorobenzene was not observed and more ortho-dichlorobenzene was formed than para-dichlorobenzenes might be attributed to the inductive and steric effects, which form the C6H5–Cl bonds as proposed by Coq et al.  derived from the rate equations used for Pd catalysts. The conversion of 1,2,4-trichlorobenzene depends on the Ru content. Shin et al.  reported that conversions depend not only on the surface metal area but also on the amount of spill over hydrogen. The selectivity of the reaction is directly related to the conversion for the each catalyst as expected from the formal reaction Scheme 1. When the conversion increases with Ru loading the selectivity towards benzene is also increasing. The appearance of chlorinated intermediates for the catalysts could be related to the consecutive reaction mechanism as shown in the Scheme 1. The results presented in the Table 4 shows the effect of temperature on the conversion and selectivity for the hydrodechlorination of 1,2,4-trichlorobenzene for 5 wt.% Ru/Nb2O5 catalysts. For comparison we have considered the activity results at 4th hour of time on stream. The conversion of 1,2,4-trichlorobenzene and the selectivity towards benzene is increased up to 523 K and then decreased with reaction temperature. The selectivity towards benzene formation is also high at 523 K. At higher temperatures there is a decrease in conversion. The decrease in the conversion might be due to chlorine poisoning which enhances carbon deposition [30,33], and lowers the hydrodechlorination activity.
The characterization of the 5 wt.% Ru/Nb2O5 spent catalysts was carried out 573 K. XPS analysis of the spent catalysts reveals the presence of surface chlorine ions Cl 2p at B.E. 198.3 eV. XRD of the spent catalysts did not provide much information about the agglomerization of the ruthenium particles, as the amount of ruthenium used was very less. However, the BET surface area has shown a considerable decrease in the surface area from 41.2 m2/ g (calcined catalyst) to 34.8 m2/g (spent catalyst). This clearly suggests that the decrease in the surface are is probably due to the formation of carbon deposition when reaction was carried out at higher temperatures. The dependence of catalytic properties with reaction time (time on stream) was carried out for 5 wt.% Ru/Nb2O5 catalyst and the results are shown in Fig. 4. The reaction was carried out at 523 K was stable for 6.5 h and 573 K remained constant for 5h. A gradual deactivation occurs due to formation of HCl, carbon deposition (due to increase in surface acidity) and also poisoning of the active sites [32,33]. The deposition of carbon and chlorine was due to homolytic cleavage of C–Cl bond on the catalyst surface. Thus, a relatively hydrogen deﬁcient-environment is generated which is suggested to be an important reason for the deactivation in terms of 1,2,4-hydrodechlorination. But the selectivity’s towards benzene formation decreased with time due to HCl formation, which inhibits the total hydrodechlorination . Fig. 5 shows the dependence of TOF (calculated at 4th hour of time on stream at 523 K), crystallite size and number of active sites obtained from CO chemisorption on Ru/Nb2O5 catalysts. From the Table 2 it is clear that the crystallite size of Ru increases with loading. The CO uptake values also suggest that there is no signiﬁcant increase in the number of active sites beyond 5 wt.% Ru loading. The H2-TPD results also suggest that after 5 wt.% loading there is an agglomerization of metal particles leaving more of the support surface available for hydrogen adsorption. There is a signiﬁcant increase in the TOF (hydrodechlorination of 1,2,4-trichlorobenzene) as the crystallite size increases up to 10 nm beyond that it leveled off. From Fig. 5 we can also observe that the number of active sites increased up to certain crystallite size beyond that it also started to level off. The present results suggest that hydrodechlorination is structure sensitive up to particle size around (6–10 nm) due to availability of active sites, beyond that it is structure insensitive.
H2 Ru catalysts
1,4-dichlorobenzene Scheme 1. Reaction scheme for the hydrodechlorination of 1,2,4-trichlorobenzene.
K.V.R. Chary et al. / Catalysis Communications 10 (2009) 459–463 Table 4 Effect of temperature on conversion/selectivity for hydrodechlorination of 1,2,4trichlorobenzene. Temperature (K)
503 523 548 573
15 32 14 10
Reaction conditions: 2.77 103 ml/s.
(%) Selectivity Benzene
60 64 45 28
38.3 35 53.5 70.8
0.7 0.5 1.5 1.2
4. Conclusions Ru/Nb2O5 catalysts are found to be highly active for the hydrodechlorination of 1,2,4-trichlorobenzene. XRD results show the formation of RuO2 from 3.0 wt.% of catalyst, which is further supported by CO-chemisorption measurements. H2-TPD reveals a spill over of H2 from metal to support, which is hydrogenolytic in promoting the hydrodechlorination reaction. Spill over hydrogen species not only promote the hydrodechlorination but also favor the removal of chlorine ions from the surface of the catalyst. The hydrodechlorination of 1,2,4-trichlorobenzene on Ru/Nb2O5 catalysts is found to be a structure sensitive reaction up to 5 wt.% Ru and insensitive at higher Ru loadings. The catalytic activity in terms of TOF is directly related to the CO chemisorption sites.
The authors thank Director, IICT, Hyderabad for his encouragement. Ch.S.S Thanks CSIR, New Delhi for the award of Senior Research Fellowship.
30 25 20
10 5 0 0
Time (h) Fig. 4. Dependence of hydrodechlorination activity on the reaction time for 5 wt.% Ru/Nb2O5 catalyst, Reaction conditions: Feed rate = 2.77 103 ml/s, weight of the catalyst = 0.8 g.
tTOF x 10-2 s-1
0. 8 2.0
TOF No. of active sites
No. of active sites x1019/m2
1.0 0. 4
Crystallite size (nm) Fig. 5. Dependence of crystallite size on TOF and number of active sites.
This is because overloading of metal on the support leads to formation of agglomerates of metallic species. However, it does not increase the number of active sites.
 P.J. Morris, W.W. Mohn, J.F. Quenson, J.M. Tiedje, S.A. Boyd, Appl. Environ. Microbiol. 58 (1992) 3088.  W.J. Hayes, E.R. Laws, Handbook of Pesticide Toxicology, Academic Press, San Diego, 1991.  Y. Liu, M. Luo, Z. Wei, Q. Xin, P. Ying, C. Li, Appl. Catal. B 29 (2001) 61.  M.A. Keane, Environ. Bus. Mag. 80 (2001) 12.  L. Calvo, M.A. Gilarranz, J.A. Casas, A.F. Mohedano, J.J. Rodrıguez, Appl. Catal. B: Environ. 67 (2006) 68.  C. Amorim, G. Yuan, P.M. Patterson, MA. Keane, J. Catal. 234 (2005) 268.  J.W. Bae, J.S. Lee, K.H. Lee, Appl. Catal. A: Gen. 334 (2008) 156.  A.G.F. de Souza, A.M.P. Bentes Jr., A.C.C. Rodrigues, L.E.P. Borges, J.L. Fontes, Catal. Today 107–108 (2005) 493.  J.L. Benitez, G. Angel, React. Kinet. Catal. Lett. 70 (2000) 67.  K.V.R. Chary, K.S. Lakshmi, M.R.V.S. Murthy, K.S.R. Rao, M. Papadaki, Catal. Commun. 4 (2003) 531.  K.V.R. Chary, P.V. Ramana Rao, V. Vishwanathan, Catal. Commun. 7 (2006) 974.  M. Schoenmaker-Stolk, J. Verwijs, J. Scholten, Appl. Catal. 30 (1987) 339.  L. Guczi, R. Sundararajan, Zs. Koppany, Z. Zsoldos, Z. Schay, F. Mizukami, S. Niwa, J. Catal. 167 (1997) 482.  S.R. Tennison, in: J.R. Jennings (Ed.), Catalytic Ammonia Synthesis, Plenum Press, New York, 1991, p. 303.  P. Kluson, L. Cerveny, Appl. Catal. A 128 (1995) 13.  V. Felis, C. De Bellefon, P. Fouilloux, D. Schweich, Appl. Catal. B: Environ. 20 (1999) 91.  T. Mori, T. Yasuoka, Y. Morikawa, Catal. Today 88 (2004) 111.  I. Nowak, M. Ziolek, Chem. Rev. 99 (1999) 3603.  K.V.R. Chary, G. Kishan, T. Bhaskar, JCS Chem. Commun. (1999) 1399.  K.V.R. Chary, T. Bhaskar, G. Kishan, K.R. Reddy, J. Phys. Chem. 105 (2001) 4392.  K.V.R. Chary, G. Kishan, C.P. Kumar, G.V. Sagar, J.W. Niemantsverdriet, Appl. Catal. A: Gen. 245 (2003) 303.  K.V.R. Chary, K.K. Seela, G.V. Sagar, B. Sreedhar, J. Phys. Chem. 108 (2004) 658.  K.V.R. Chary, K. Rajender Reddy, Ch. Praveen Kumar, D. Naresh, V. Venkat Rao, G. Mestl, J. Mol. Catal. A: Chem. 223 (2004) 363.  E.I. Ko, J.M. Hupp, N.J. Wanger, J. Catal. 84 (1983) 85.  E.I. Ko, J.G. Weissman, Catal. Today 8 (1990) 27.  M. Nurunnabi, K. Murata, K. Okabe, M. Inaba, I. Takahara, Catal. Commun. 8 (2007) 1531.  B. Bachiller-Baeza, A. Guerrero-Ruiz, I. Rodriguez-Ramos, J. Catal. 229 (2005) 439.  P.G.J. Koopman, A.P.G. Kieboom, H. van Bekkum, React. Kinet. Catal. Lett. 8 (1978) 389.  H.Y. Lin, Y.W. Chen, Theromochimica acta 419 (2004) 283.  B. Coq, F. Figueras, P. Bodnariuk, in: F. Cossio, O. Bermudez, G. del Angel, R. Gomez (Eds.), Proceedings of the XI Symposium. Iberoamericano de catalisis, Guanajuato Mexico, June 1988, p. 1145.  E.J. Shin, A. Spiller, G. Tavoularis, M.A. Keane, PCCP 1 (1999) 3173.  E. Lopez, S. Ordonez, F.V. Diez, Appl. Catal. B: Environ. 62 (2006) 57.  B. Coq, J.M. Cognion, F. Figueras, D. Trournigant, J. Catal. 141 (1993) 21.