3rd World Congress on Oxidation Catalysis R.K. Grasselli, S.T. Oyama, A.M. Gaffney and J.E. Lyons (Editors) 9 1997 Elsevier Science B.V. All rights reserved.
Oxidative d e h y d r o g e n a t i o n o f propane on C e N i x O v (0 _< x _< 1) m i x e d oxides h y d r o g e n acceptors L. Jalowiecki-Duhamel a, A. Ponchel ~, and Y. Barbaux b aLaboratoire de Catalyse H6t6rog6ne et Homog6ne, U.R.A.C.N.R.S. D04020, Brit. C3, Universit6 des Sciences et Technologies de Lille, 59655 Villeneuve d'Ascq Cedex, France bUniversit6 d'Artois, SP 18, rue J. Souvraz, 62307 Lens Cedex, France
The oxidative dehydrogenation of propane has been studied on CeNixOy (0 < x < 1) mixed oxides previously reduced under hydrogen. The products obtained are propene and 202. No CO is obtained, whatever the temperature. An optimum propene yield of 6.9 % is obtained at 648 K on previously, at 473 K under H2, in-situ reduced CeNi0.5. Whereas without any pretreatment, a propene yield of 5.35 % is obtained. Therefore, a beneficial effect of the pretreatment under H2 is shown. The reducibility under HE of some CeNixOv mixed oxides has been previously studied and, in particular, it has been reported that these solids are able to store high quantities of hydrogen H* related to the existence of anionic vacancies. These H* species have been proposed to be one half H" species located in the anionic vacancies and the second half IT species, forming with the O 2" species of the solid, OH" groups. As dehydrogenation requires the abstraction of hydrogen from the molecule which could be performed by a lacunar phase, and by analogy to the dissociation of HE, a mechanism of the dehydrogenation step of the alkane is proposed, involving a heterolytic abstraction of a H" species by an anionic vacancy and of a IT species by an 02- species of the solid forming an OH" group.
1. INTRODUCTION Selective oxidation is one of the promising routes of utilizing alkanes which are relatively abundant in natural gas or in liquefied petroleum gas. Therefore, this research field is of great interest and of growing importance from both the industrial and the fundamental point of view. Recent studies emphasize on the importance of surface Broensted and Lewis acidity in selective oxidation of light alkanes on metal oxides [ 1,2], but the detailed mechanism and the characteristics of the active centers is still unclear. Nevertheless, it appears that it is generally admitted that i) the breakage of the C-H bond from the alkane is the rate determining step of the selective oxidation of propane  and ii) the reaction mechanism is of the Mars van Krevelen type (redox) .
384 A large variety of oxide catalysts have been claimed as beeing effective in the oxidative dehydrogenation (ODH) of propane [5-18]. Mainly vanadium based catalysts such as VPO, VMgO solids have been developed and some of them have been extensively studied in order to identify the active vanadate phases [7, 9-14]. Recently, a vanadia on precipitated silica catalyst has been found to exhibit high yield in the ODH of propane [ 17]. However, utilization of rare earth catalysts in oxidation reactions seems also to be attractive [ 18], because solids composed of oxides of Ce, Sm, Nd or Y and CeF3 are able to preserve high selectivity at high conversion . It is o~en agreed that the oxidation of hydrocarbons over oxide catalysts involves surface oxygen/oxygen vacancy participation [ 19, 20], and the oxygen mobility of metal oxide catalysts has something to do with catalytic activity. The fluorite type oxides, such as ceria, zirconia and thoria, have face-centered-cubic crystal structure in which each tetravalent metal ion is surrounded by eight equivalent nearest 02- ions forming the vertices of a cube. Oxygen vacancies are created when a fluorite oxide is doped by divalent or trivalent impurity ions. Thus the fluorite oxides have been extensively studied as oxygen-ion-conducting materials due to their high oxygen vacancy concentration and mobility properties. A long time ago, a redox mechanism involving lattice oxygen/oxygen vacancy participation was proposed for carbon monoxide oxidation on cerium oxide . In our laboratory, the mechanism of reduction of CeMxOv (M = Ni or Cu) mixed oxides has been studied . It has been found that the insertion of Ni 2§ or Cu 2§ in ceria leads to a decrease of the reduction temperature of the host oxide and a mechanism of reduction based on the formation of anionic vacancies in ceria, facilitated by the incorporation of transition-metal cations, the heterolytic dissociation of H2 and redox reactions between Ce 4+ and the transition element, has been proposed. These rare earth based mixed oxides have been reported to be able to accept and store large quantities of hydrogen . In order to elucidate further the ODH mechanism, and since dehydrogenation requires the abstraction of hydrogen from the alkane, we have studied the transformation of propane to propene on these CeNixOy compounds.
2. EXPERIMENTAL The mixed oxides CeNixOv were prepared by coprecipitation of hydroxides from mixtures of cerium and nickel nitrate using triethylamine (TEA) as precipitating agent, drying at 323 K and calcination in air at 773 K . The metal loading has been verified by microanalysis. The solids will be called CeNix. The catalytic oxidation of C3H8 was performed under atmospheric pressure in a fixedbed stainless-steel tubular reactor (lenght 300 mm, internal diameter 15 mm) by co-feeding the nitrogen-diluted reaction gases (C3Hs/O2/N2 = 5/15/80). The total flow rate was 100 ml.min"~ (down flow), the reaction temperature was in the range 473-673 K, and the catalyst mass was of 0.10 g. Propene and CO2 were the only products detected and no CO was observed. The experimental details have been published previously . When the oxidative dehydrogenation of C3Hs has been performed over a solid previously reduced <
385 temperatures, the catalyst has been purged under He for more than 2hrs after the reduction step. The dynamic method of titration of the hydrogen species (noted H* as the aim of the study is not their exact charge) stored by a solid has also been published previously and applied to various catalytic systems . The pretreatment and catalytic experiments were carried out in-situ at atmospheric pressure in an all glass, grease free flow apparatus. The solid (65 mg) was treated first under a purified hydrogen flow at various temperatures, then after elimination of molecular hydrogen, at 423 K, under isoprene+helium flow (2 l/h) the hydrogenation activity was followed as a function of time. This hydrogenation activity involves the participation of reactive hydrogen species H* from the solid which hydrogenate isoprene and are consumed by a diffusion process. As a function of time on stream, the isoprene conversion decreases and by integrating the curve obtained, the extractable H* content that the solid is able to store has been determined. The thermogravimetric experiments were performed under purified hydrogen flow on a Sartorius balance.
3.1. Oxidative dehydrogenation of propane. The conversions of propane to propene have been studied on the CeNix mixed oxides as a function of temperature. Appropriate blank runs showed that, under our experimental conditions, the contribution of the gas phase reaction is negligible. On Figures 1 and 2, the evolutions of the conversion and selectivity as a function of temperature obtained on CeNi0.2 and CeNi0.5 at the stationnary state are presented as examples. For the sake of comparison the catalytic activity of CeO2 was also evaluated. Propene and CO2 were the only products detected. No CO was observed whatever the temperature. On CeO2, at 373 K a propane conversion of 3% is observed with a selectivity to propene of 1.6%. As a function of temperature the conversion and selectivity increase and at 673 K, a propane conversion of 10% is obtained with a propene selectivity of 6%. On the CeNix compounds, for reaction temperatures higher than 523 K, the conversion increases while the propene selectivity decreases. On CeNi0.2 (Figure 1), propene selectivity becomes almost stable at about 10% for temperatures higher than 575 K, the yield increases up to 4.5 % for 673 K because the conversion increases. For CeNi0.5 (Figure 2), the selectivity decreases from 50 to 30 % in the temperature range 525-575 K, and an optimum yield of 5.35% is obtained at 648 K. As shown in Figure 3, when increasing the Ni content in CeNix up to x = 1, no better results are observed. One must only remark that for CeNix with x _>0.7 and temperatures higher than 675 K, CI-I4 is also observed among the products obtained. Clearly, much better performances have already been obtained on cerium based catalysts [ 16, 18] which was not the aim of the present study.
20 0 450
500 550 600 650 TEMPERATURE (K)
0 "" " 450
500 550 600 650 TEMPERATURE (K)
Figure 2 9 Propane conversion (O), and propene yield (o) and selectivity (I), as a function of temperature on CeNi0.5.
Figure 1 9 Propane conversion (O), and propene yield (0) and selectivity (I) as a function of temperature on CeNi0.2.
J uJ m
.......... . ...................
~ . . .~176176 ..... :- ..........
sso 6oo TEMPERATURE
Figure 3 ' Propene yield as a function of temperature on CeNix with x = 0.2 (0), x = 0.5 (o), x = 0.7 ( 0 ) , x - 1 (D), and on CeO2 (o).
of the solid to accept
The reducibility of the CeNi0.5 mixed oxide has already been studied under H2  and it has been reported that this solid is able to store high quantities of hydrogen (noted H*) in its reduced state. Taking into account that dehydrogenation requires the abstraction of hydrogen
387 species from the hydrocarbon, the ability of the solid to accept hydrogen has been studied. Therefore, the hydrogen content that the solid is able to store has been investigated by a dynamic method. Under isoprene+helium flow, at 423 K, isoprene hydrogenation activity is measured on CeNix previously treated under H2 at a given temperature. As an example on Figure 4 are presented the results obtained on reduced CeNi0.2 at 573 K under H2. The relative hydrogenation activity (HYD=t) corresponds to the ratio between the activity at time t and the activity at time zero under isoprene+helium flow. As a function of time on stream, the hydrogenation activity decreases and becomes nil. The integration of the curve obtained permits to determine the concentration of the hydrogen species H* of the solid which have reacted. Figure 5 shows the evolutions of the H* species that the solids CeNi0.2 and CeNi0.5 are able to store as a function of the treatment temperature under H2. The ability of the solid to store hydrogen depends on the treatment temperature under Hz which corresponds to various reduced states of the catalyst.
~ O . . .
a >0.4 "I-
~b ,.. 10
~(rrirt) Figure 4 : Relative hydrogenation activity at 423 K under isoprene + helium flow versus time of CeNio.2 reduced at 573 K under H2.
Figure 5 9 Hydrogen H* content as a function of treatment temperature under H2 of (o) CeNi0.2 and (11) CeNi0.5.
Besides, for each treatment temperature the relative hydrogenation activity can be reported as a function of the relative hydrogen H* content of the solid, and no proportionality is obtained (Figure 6). In fact, the H* consumption kinetic by the hydrocarbon is a complex phenomenon, in particular, the diffusion of the H* species from the <~bulk >>to the ~) of the solid has to be taken into account [24, 25]. For treatment temperatures higher than about 447 K, anionic vacancies are created in the CeNix catalyst by the loss of H20 (OH groups), as it is shown by thermogravimetry (Figure 3 1 , 7). After a treatment under H2 at 473 K the solid CeNi0.~ contains 17.5 10- mol.g of H
388 species, and this hydrogen storage has been correlated to the creation of anionic vacancies in the solid .
>- 0.4 -r
Figure 6 : Relative hydrogenation activity at 423 K under isoprene + helium flow versus the hydrogen H* species concentration of CeNi0.5 reduced at 423 K under H2.
Figure 7 9 Thermal treatment under H2 of CeNi0.5 (a) and CeNio.2 (b) followed by thermogravimetry.
A great analogy exists between the results presented in this study and those obtained in the laboratory on copper based oxides [24, 25] and molybdenum based sulfides [26, 27] which have been found to be hydrogen reservoirs. As a matter of fact different studies published on these catalysts deal with the relations existing between the active site unsaturation degree and the dienes hydrogenation activity and selectivity, as well as the existence of hydrogen H* reservoirs. It has been shown that the first hydrogen species introduced in the diene during the hydrogenation reaction is a hydride species coming from the solid . Indeed, it has been proposed that under helium+alkadiene, the titrated H* species are for one half H" species located in anionic vacancies and the second halfH + species (coming from OH groups) [24-27]. These species are inserted in the solid during the activation under H2 : O 2" M n+ D + H2 -9 OH-M n§ H" (with D 9anionic vacancy). 3.3. Oxidative dehydrogenation of propane on hydrogen reservoirs. The oxidative dehydrogenation of C3H8 has been performed on previously, at different temperatures under H2, in-situ reduced catalysts. On Figures 8 and 9, propene yield and selectivity are reported as a function of propane conversion, respectively on pretreated under
H2 at 473 K CeNi0.2 and on pretreated under H2 at 433 K and 473 K CeNi0.5. The two treatment temperatures (433 and 473 K) lead respectively to the creation of a slight and a large hydrogen reservoir as shown previously in Figure 5. The ODH of propane obtained on the H2 pretreated CeNi0.5 at 433 K is equivalent to that obtained on the untreated solid. Besides, at 648 K, an optimum yield of about 6.9 % can be obtained on the reduced CeNi0.5 at 473 K, while at the same temperature, a maximum yield of 5.4 % is obtained on the untreated solid (Figure 8). As shown on Figure 9, a similar effect is observed on CeNi0.2. Clearly, the presence of the large hydrogen reservoir (i.e. treatment at 473 K under H2) lead to a beneficial effect on the propene yield.
......~-.~ ......2 ~ i i 2 ~ : : 0
Figure 8 : propene yield (I, o) and selectivity (D, o) as a function of propane conversion on CeNi0.2 not treated (I, [3) and previously treated under H2 at 473 K (., o).
20 40 CONVERSION
Figure 9 : propene yield (m, o, 0 ) and selectivity ([2, o, 0) as a function of propane conversion on CeNi0.5 not treated (I, [2) and previously treated under H2 at 433 K ( 0 , 0) and at 473 K (., o).
The existence of the large hydrogen reservoir can be correlated to a partially reduced catalyst, as it has been published previously . So, the results obtained here confirm that for the ODH of propane, the catalyst works in a partially reduced state and that a redox mechanism is involved.
4. DISCUSSION. The catalysts CeNix possess this character of being hydrogenation and oxidation catalysts in agreement with their redox properties. Knowing that in previous studies it has been shown by work function measurements that propane reacts with O 2 species located at the surface of various oxide catalysts during the activation of the alkane , and taking into
390 account the results obtained in the present study, a hydrocarbon activation model can be proposed. By analogy to the heterolytic dissociation of hydrogen, a heterolytic dissociation of the alkane involving the abstraction of a H" species from the hydrocarbon can be envisaged on a low coordination site involving an anionic vacancy. This active site, < OH- M "+ H" + propene (with D : anionic vacancy). It is well known that the C-H bond activation can result from different mechanisms, and even if the abstraction of a hydride species from the alkane has already been proposed in the literature on solid super acid catalysts , the transfert of I-I+ species has been much more otten proposed. One must remark that, in the case of a heterolytic rupture, the two species (H', I-I+) can exist, but due to its high reactivity the hydride species is much more difficult to detect. In presence of 02 the hydride species will react violently forming finally water, this permits to consume and transform 02 into selective oxygen species and regenerate the active site. 2H- + I~O 2 ~ H20 +2e" + 2 D 8 9 + 2 e " + D - ~ O2 2 H" + 02 - ~ 0 2- q- H20 + D As it has already been proposed for the hydrogen treated solid , the hydroxyl groups can recombine together, forming also water : 2 OH"
--~ H20 + 02- + r-]
Of course, there is no direct relationship between the H* storage and the catalytic performance. And this is true whatever the reaction (oxidation or hydrogenation), mainly because (i) the H* species storage involves <>and <>phenomena whereas the activity is related to catalytic sites located at the <>of the solid, and (ii) activity and specially selectivity for a given reaction depend on specific kind of sites. It is highly probable that the sites interacting with the alkane are much more specific (structurally for example) than
391 those interacting with H2. Moreover, the aim of the present study is not to discuss the nature and charge of the cations involved in the catalytic site. One can only surmise from the results obtained that the catalyst works in a partially reduced state. Additional characterizations are in progress, devoted to highlighting the nature and structure of the active and selective site for the ODH reaction.
5. C O N C L U S I O N A beneficial effect of pretreating CeNixOv solids under H2 is shown for the oxidative dehydrogenation of propane. Dehydrogenation requires the abstraction of hydrogen from the hydrocarbon. We have shown that, in their partially reduced state, these solids are able to accept high quantities of hydrogen H* (H', I-I+). Moreover, the hydrogen H* content that the solid is able to store depends on the reduction degree of the solid (i.e. the treatment temperature under H2), and in particular, on the creation of anionic vacancies in the solid. Therefore, by analogy to the dissociation of H2, a mechanism of the alkane dehydrogenation is proposed, involving a heterolytic abstraction of a H" species by an anionic vacancy and of a IT species by an 0 2- species of the solid forming an OH" group.
2 3 4. 5 6. 7. 8.
9. 10. 11 12 13 14. 15 16. 17 18. 19. 20.
G. Busca, G. Centi, F. Trifiro and V. Lorenzelli, J. Phys. Chem., 90 (1986) 1337. A. Bielanski and J. Haber, in <
392 21. M. Breysse, M. Guenin, B. Claudel, H. Latreille and J. Veron, J. Catal., 28 (1972) 54. 22. G. Wrobel, C. Lamonier, A. Bennani, A. D'Huysser and A. Abouka'is, J. Chem. Soc. Farad. Trans., 92 (1996) 2001. 23. N. Boisdron, A. Monnier, L. Jalowiecki-Duhamel and Y. Barbaux, J. Chem. Soc. Farad. Trans., 91 (1995) 2899. 24. A. Sene, L. Jalowiecki-Duhamel, G. Wrobel and J. P. Bonnelle, J. Catal., 144 (1993) 544. And references therein 25. L. Jalowiecki, G. Wrobel, M. Daage and J. P. Bonnelle, J. Catal., 107 (1987) 375. 26. L. Jalowiecki, A. Aboulaz, S. Kasztelan, J. Grimblot and J. P. Bonnelle, J. Catal., 120 (1989) 108. 27. L. Jalowiecki, J. Grimblot and J. P. Bonnelle, J. Catal., 126 (1990) 101. 28. M. Daage and J. P. Bonnelle, Appl. Catal., 16 (1985) 335. 29. A. Pantazadis, A. Auroux, J.-M. Hermann and C. Mirodatos, Catal. Today, 32 (1996) 81. 30. G. Busca, G. Centi and F. Trifiro, J. Amer. Chem. Soc., 107 (1985) 7757. 31. G. Busca, G. Centi, F. Trifiro and V. Lorenzelli, J. Phys. Chem., 90 (1986) 1337. 32. G. Centi and F. Trifiro, Catal. Today, 3 (1988) 151. 33. J.C. Vedrine, J. M. Millet and J. -C. Volta, Catal. Today, 32 (1996) 115. 34. V.B. Kazansky, V. Yu. Borovkov, and L. M. Kustov, Proc. 5th International Congress on Catalysis, Berlin, 3 (1984) 3. 35. V.B. Kazansky, V. Yu. Borovkov, and A. Zaitsev, Proc. 9th International Congress on Catalysis, (Eds.M.J. Phillips and M. Ternan), Calgary, 3 (1988) 1426. 36. L. Jalowiecki-Duhamel, A. Monnier, Y. Barbaux and G. Hecquet, Catal. Today 32 (1996) 237. 37. a) H. Hatori, O.Takahashi, M. Takagi and K. Tanabe, J. Catal., 68 (1981) 132. b) O.Takahashi and H. Hatori, J. Catal., 68 (1981) 144.