Oxidative dehydrogenation of propane over vanadium and niobium oxides supported catalysts

Oxidative dehydrogenation of propane over vanadium and niobium oxides supported catalysts

Applied Catalysis A: General 184 (1999) 291±301 Oxidative dehydrogenation of propane over vanadium and niobium oxides supported catalysts Paolo Vipar...

249KB Sizes 0 Downloads 36 Views

Applied Catalysis A: General 184 (1999) 291±301

Oxidative dehydrogenation of propane over vanadium and niobium oxides supported catalysts Paolo Viparellia,*, Paolo Ciambellib, Luciana Lisic, Giovanna Ruoppoloa, Gennaro Russoa, Jean Claude Voltad a Dipartimento di Ingegneria Chimica, UniversitaÁ Federico II, Napoli, Italy Dipartimento di Ingegneria Chimica e Alimentare, UniversitaÁ di Salerno, Fisciano, SA, Italy c Istituto di Ricerche sulla Combustione, CNR, Napoli, Italy d Institut de Recherches sur la Catalyse, CNRS, Villeurbanne, France

b

Received 13 November 1998; received in revised form 6 April 1999; accepted 7 April 1999

Abstract The oxidative dehydrogenation of propane over niobium and vanadium oxides supported on high surface area TiO2 has been studied. The two different oxides have been investigated both as single and mixed supported phase. The vanadium containing catalysts are very active but poorly selective to propylene, while the presence of niobium increases the catalyst performances at low V/Nb ratio. Redox and acid properties of the catalysts have been investigated by isopropyl alcohol decomposition test. Results showed that strong acidity is associated with vanadia and niobia if present as single supported oxide. The interaction between vanadium and niobium leads to the formation of weaker acid centers. The presence of redox and acid sites at the same time is necessary for propane activation. At the same propane conversion, the selectivity to propylene increases when niobium, at a high coverage, is associated to vanadium with low coverage, while it decreases when associated to higher coverage of vanadium. This has been explained by differences in the local environment of vanadium sites with niobium sites at the surface of TiO2. The comparison of the same catalysts for oxidative dehydrogenation of ethane and propane shows that the formation of carbon oxides is more sensitive to the structure of the catalyst for ethane oxidative dehydrogenation. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Vanadia±niobia; Titania; Propane oxydehydrogenation; Redox processes; Acid properties

1. Introduction A process based on the oxidative dehydrogenation (ODH) of low cost saturated hydrocarbons could contribute to satisfy the growing demand of high *Corresponding author. Present address: Dipartimento di Chimica, Ingegneria Chimica e Materiali, I-67040 Monteluco di Roio (AQ), Italy; tel.: +39-862-434232; e-mail: [email protected]

purity light ole®ns [1]. Paraf®ns dehydrogenation in the presence of oxygen is thermodynamically favored due to water formation, but the selectivity to ole®ns is generally poor because of the lower reactivity of the paraf®n compared to that of the formed ole®n [2]. It has been reported that the features required to catalysts for ethane ODH are not the same for the ODH of higher alkanes [1]. This is associated to the reactivity of secondary C±H bonds present in propane

0926-860X/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 1 0 4 - 0

292

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

and butanes, but not in ethane. Moreover, as the presence of allylic carbon atoms in the formed ole®n should lead to easier production of COx , higher selectivity to ethylene is expected in the case of ethane ODH [3,4]. Most of the reported catalysts for the oxidative dehydrogenation of light paraf®ns are based on bulk or supported vanadium oxides [1]. V2O5 is active in many reactions of partial oxidation though the selectivity to the desired product is often low [5]. Selectivity can be usually improved by addition of other elements: V±Mg±O catalysts show high selectivity to ole®n in propane and n-butane ODH, V±Nb±O catalysts exhibit good productivity to propylene as well [1]. Regarding vanadium oxides supported catalysts, we have previously found that the addition of niobium to V2O5/TiO2 systems improves their catalytic performances in ethane ODH at low V/Nb ratio [6]. Depending on the reacting paraf®n, catalyst features such as vanadium coordination, redox and acidbase properties, in¯uence selectivity to ole®n differently. The presence of acid sites promotes the selectivity to ethylene in ethane ODH in contrast with propane or butane ODH, as the increase of alkane basicity with the chain length leads to easier desorption of short chain ole®nic intermediates from the catalyst surface [7]. In this paper, the catalytic performances of niobium and vanadium oxides supported on high surface area TiO2 were evaluated for propane ODH and compared with those previously obtained in the case of ethane ODH [6]. Furthermore, redox and acid character of the catalysts were investigated in order to understand their role in the selective oxidation of the two alkanes. 2. Experimental 2.1. Catalyst preparation Binary and ternary catalysts, respectively containing only supported vanadium or niobium oxide or both the oxides, were prepared by wet impregnation of high surface area (125 m2/g) pure anatase TiO2 (Tioxide specialties) with vanadium metavanadate (BDH Laboratory Supplies) and niobium ammonia complex (Companhia Brasileira de Metalurgia e Minerac,ao) as described in [6]. In the preparation of ternary catalysts,

the precursor salts of vanadium and niobium oxides were introduced at the same time in the water solution. The initial pH ranged from 3, when the niobium precursor was present in the solution, to 5 when only the vanadium precursor was present in the solution. After impregnation the materials were dried at 1108C and calcined at 5508C in ¯owing air for 4 h. The supported catalysts will be referred to with the code xVyNb/Ti where x and y are the nominal content (weight percentage) of V2O5 and Nb2O5, respectively. 2.2. Physico-chemical characterization The niobium oxide composition in the catalyst was determined by X-ray ¯uorescence with a Philips PW 1480 sequential spectrometer and the vanadium content was evaluated by manganometric titration [8]. The absence of any interference of Nb5‡ ions with the chemical analysis of vanadium was veri®ed. XRD analysis was performed with a PW 1710 Philips diffractometer. BET surface areas were obtained by N2 adsorption at 77 K with a Carlo Erba 1900 Sorptomatic. Isopropyl alcohol (IPA) decomposition tests were carried out in the temperature range 150±2208C with a quartz ®xed bed reactor fed by a 3% IPA/N2 mixture (total ¯ow ˆ 30 cm3/min). A gas-chromatograph equipped with two detectors (¯ame ionization and thermal conductivity detectors) was used to analyze the products. The catalyst sample was preheated at 2508C in ¯owing air before each experiment, then cooled down to 1508C. After N2 purging, the temperature was raised up to the reaction temperature and the IPA/N2 mixture was fed. 2.3. Catalytic activity tests Catalytic activity tests of propane ODH were carried out with a ®xed bed quartz microreactor operating under atmospheric pressure. Analysis of reactants and products was performed by on-line gas chromatography. A Delsi IGC 120 MB gas chromatograph equipped with a thermal conductivity detector was used to analyze permanent gases and H2O. Hydrogen was the carrier gas. Two columns, a 5A molecular sieve to separate O2 and CO and a Porapak-Q column to separate CO2 and H2O, were operated in parallel. The organic products were analyzed with a Delsi IGC

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

120 FB gas chromatograph, equipped with a ¯ame ionization detector. N2 was the carrier gas. Two columns were operated in parallel: a Durapak column to separate light hydrocarbons (methane, ethane, ethylene, propane and propylene) and a Carbowax column to separate oxygenates. The reactor was connected online to the gas-cromatograph with a hot box to prevent products from condensation. The feed composition was 2% propane, 4% O2, N2 balance. The contact time, determined using ¯ow rates at room temperature, ranged from 0.005± 1 g s cmÿ3 and the reaction temperature from 300 to 5008C. Further catalytic tests were carried out in the temperature range 450±5508C with a feed composition of 4% ethane, 2% O2, He balance on the same catalysts with a contact time ranging from 0.018 to 0.6 g s cmÿ3 in order to compare the effect of the reaction temperature on the ODH of ethane and propane. 3. Results In Table 1, the catalysts composition, surface area and theoretical surface coverage, estimated as in [6], are reported. The data show that the theoretical monolayer capacity was never exceeded even for 6V6Nb/Ti catalyst where traces of T or TT Nb2O5 phase [9] were detected by XRD analysis, suggesting that aggregation may occur when the coverage approaches the monolayer. No peaks due to TiO2 rutile phase were observed in XRD spectra of the catalysts indicating that the calcination temperature was low enough to inhibit the phase transition of TiO2 anatase. A small loss (17%) of the initial surface area of TiO2 was observed in all the catalysts.

293

3.1. IPA decomposition test IPA decomposition was used as model reaction to characterize the acid and redox properties of the catalysts. Isopropyl alcohol decomposition occurs following two parallel reactions: dehydration to propylene on acid centers or dehydrogenation to acetone on basic redox sites. The formation of di-isopropyl ether can also occur on acid centers at low temperature [10]. Results of the IPA decomposition test are reported in Table 2. In all the tests, propylene, acetone and diisopropyl ether were produced, while traces of propane were observed when vanadium was present in the catalyst sample. Table 2 shows that on V/Ti and V±Nb/Ti catalyst acetone is the main product at low temperature due to the lower activation energy for dehydrogenation with respect to dehydration, as reported by Pepe et al. [11]. At high temperature, mainly propylene is formed on 6Nb/Ti while on the vanadia containing catalysts the formation of acetone still occurs at 2208C. It has been reported [11] that pure V2O5 produces mainly propylene while acetone formation increases with the exposition of V = O bonds of vanadium oxide dispersed onto TiO2 anatase. No data are available on supported niobia though it was shown that niobic acid produces only propylene [12]. However, the high selectivity to propylene of niobia supported catalyst indicates a strong surface acidity, in agreement with NH3 TPD characterization [6], while the high dehydrogenation activity of vanadia based catalysts con®rms the strong redox properties found by TPR experiments [6]. The most active samples for IPA decomposition are the binary catalysts 6V/Ti and 6Nb/Ti. They give about the same IPA conversion to propylene at

Table 1 Catalysts composition and surface coverage of xVyNb/Ti catalysts Catalyst

V2O5 (wt.%)

Nb2O5 (wt.%)

V 2 O5 coverage

Nb2O5 coverage

Total coverage (a)

Surface area (m2 gCATÿ1)

6Nb/Ti 1V/Ti 6V/Ti 1V6Nb/Ti 6V6Nb/Ti

± 1.0 5.8 1.1 5.8

6.4 ± ± 6.7 6.9

± 0.065 0.38 0.065 0.38

0.37 ± ± 0.38 0.39

0.37 0.065 0.38 0.45 0.77

110 111 106 109 105

a

The value 1 corresponds to a theoretical coverage of the whole surface.

294

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

Table 2 Results of IPA decomposition test Catalyst

T (8C)

Conv. (%)

SACE (%)

SPropene (%)

SDIPE (%)

SPropane (%)

TiO2

150 180 200 220

0.13 2.6 9.6 24.8

54.3 8.5 3.5 2.6

45.7 78.4 87.3 94.6

0 13.1 9.2 2.9

± ± ± ±

6Nb/Ti

150 180 200 220

0.35 2.1 10.8 32.7

15.6 2.4 1.01 0.67

74.2 94.2 96.9 97.9

10.2 3.3 2.1 1.4

± ± ± ±

1V/Ti

150 180 200 220

1.1 1.5 2.5 5.9

89.4 69.7 44.4 24.4

4.7 23.4 46.3 64.7

6.0 3.9 2.5 1.8

± 3.0 6.8 9.0

6V/Ti

150 180 200 220

1.8 4.4 12.1 33.4

52.0 20.2 9.5 6.0

33.7 67.4 81.6 88.2

14.2 10.0 5.4 1.8

± 2.3 3.5 4.0

1V6Nb/Ti

150 180 200 220 150 180 200 220

1.3 2.7 11.9 22.6 1.4 3.2 6.6 18.2

82.9 31.3 8.4 8.8 81.0 28.9 15.9 10.7

11.9 52.6 79.7 82.3 13.9 59.7 73.1 79.7

5.2 13.0 8.9 2.9 5.1 7.9 6.5 3.1

± 3.1 3.0 6.0 ± 3.5 4.6 6.4

2208C (33.4 and 32.7%, respectively), the former also producing some acetone. As both Brùnsted and Lewis acid centers have been found on vanadia/titania catalysts [13,14] whilst only Lewis acidity is attributed to niobium oxide supported on TiO2 [15], it can be concluded that IPA dehydration can be activated by acid sites having different nature. With 1V/Ti the acetone yield was similar to that obtained with 6V/Ti while both IPA conversion and propylene yield increased with vanadium loading (from 3.8 to 29.5% propylene yield at 2208C), suggesting that the dehydration activity strongly depends on the vanadium content. On the other hand, although the vanadia binary catalysts and the vanadia/niobia ternary catalysts give quite similar acetone yields, propylene yield decreases with increasing vanadium content of ternary catalysts, i.e., increasing the V/Nb ratio (yield from 18.6 on 1V6Nb/Ti to 14.5 on 6V6Nb/ Ti at 2208C), in contrast with the results obtained

with the binary catalysts. Considering that the propylene yield was 32% at 2208C on 6Nb/Ti, it can be assumed that the high reactivity of niobium containing acid centers to form propylene is strongly reduced when both the metals are present on the catalyst surface, likely due to niobium interaction with vanadium. Such interaction was also evidenced by Raman characterization in [6]. According to this hypothesis IPA decomposition test carried out on the ternary catalysts provides an indication of the extent of the interaction between niobium and vanadium which, as expected, increases with increasing the V/Nb ratio. In the case of the 6V6Nb/Ti catalyst, in which a very high surface coverage is reached, one may suppose that the formation of Nb2O5 crystallites may affect the reactivity of IPA decomposition by decreasing the number of active sites on the catalyst surface. On the basis of the XRD analysis, we believe that for our

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

295

catalysts this effect was negligible, because only traces of Nb2O5 crystalline phases were detected. 3.2. Catalytic activity In order to verify the absence of homogeneous reactions under the conditions investigated, experiments with empty reactor and the same feed as for the catalytic tests were carried out. The results showed that homogeneous activity was negligible up to 5508C. In all the catalytic tests the oxygen conversion was always far from 100% and the temperature increase, due to exothermal reactions, was negligible. All catalyst samples produced C3H6, CO and CO2 with traces of C1, C2 hydrocarbons. No oxygenated products were detected in any case. Carbon balance was closed to within 2%. In Table 3 the performances of all the catalyst samples in propane ODH at 4008C are reported. The results refer to propane conversions close to 20% obtained at different contact times, in order to avoid any effect of a different propane conversion on the selectivity to propylene. Under our experimental conditions, 6Nb/Ti shows the highest selectivity to propylene and the lowest activity. The catalytic activity obtained with TiO2 was similar to that of 6Nb/Ti but the selectivity to propylene was very poor. Moreover the activity increases with vanadium content both in the absence and in the presence of niobium, as shown by the lower contact time required to obtain the same propane conversion. The selectivity to propylene decreases with increasing vanadium content, due to an higher selectivity to CO, the formation of CO2 being almost unaffected by the vanadium loading. This effect is much more noticeable for the ternary catalysts.

Fig. 1. Effect of the presence of niobium oxide in xV6Nb/Ti catalysts on the selectivity to propylene in the ODH of propane at 4008C.

In Table 3, it is also possible to observe the effect of the catalyst composition on the yield to propylene. The highest yield was obtained with 6Nb/Ti and 1V6Nb/Ti, showing a trend similar to that exhibited by the selectivity. Fig. 1 shows the values of propylene selectivity against the propane conversion achieved at 4008C on the binary (xV/Ti) and the ternary (xV6Nb/Ti). In the range from 5 to 20% propane conversion, the presence of Nb markedly enhances the selectivity to propylene at low vanadium content (more than 15% higher selectivity on 1V6Nb/Ti than on 1V/Ti), but not at higher loading. The in¯uence of the vanadium content in the series xV6Nb/Ti is shown in Fig. 2. The selectivity to propylene at 4008C decreases with respect to 6Nb/Ti in the whole range of propane conversion (5±20%) as the V2O5 weight percentage increases. The xV/Ti catalysts give selectivities to propylene similar to those reported for other vanadium oxide

Table 3 Results of propane oxidative dehydrogenation at 4008C Catalyst

6Nb/TiO2 1V/TiO2 6V/TiO2 1V6Nb/TiO2 2V6Nb/TiO2 6V6Nb/TiO2

 (gCAT s cmÿ3)

Conversion (%) O2

Selectivity (%) C3H8

C3H6

CO

CO2

C1 ‡ C2

C3H6

0.74 0.50 0.019 0.48 0.21 0.034

29.0 30.5 34.8 31.3 33.5 33.2

20.9 18.7 19.9 20.4 21.0 19.4

36.6 26.5 23.1 30.6 23.6 19.9

41.8 55.2 57.4 47.4 52.9 58.6

21.1 18.3 19.4 21.7 23.5 21.6

0.5 0.012 0.0 0.3 0.0 0.0

7.7 4.9 4.6 6.2 5.0 3.9

Yield (%)

296

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

Fig. 2. Effect of vanadium content of xV6Nb/Ti catalysts on the selectivity to propylene in the ODH of propane at 4008C.

based catalysts, even if lower than V/Mg/O systems [4]. However, it is noteworthy that with 6V/Ti, propylene productivity higher than with the other vanadia based catalysts [1] was achieved at 4008C, and that on 6Nb/Ti both selectivity and yield of propylene are lower than those reported for bulk Nb/V mixed oxides [1]. The data collected in Table 3 indicate that the formation of carbon oxides is responsible for the lower selectivity to propylene and that the formation of CO, much more than of CO2, seems to be in¯uenced by the catalyst composition. As an example, for very similar values of temperature, propane conversion and selectivity to CO2, the selectivity to CO varies from 41.8 on

6Nb/Ti to 58.6 on 6V6Nb/Ti (Table 3). Therefore, the distribution of CO and CO2 in the reaction by-products must be carefully examined at different experimental conditions. In Fig. 3(a) and Fig. 4(a) the propane conversion and the selectivity to propylene, CO and CO2 are reported as functions of the contact time for the two catalysts that exhibit the most different catalytic behavior, i.e., 6Nb/Ti and 6V/Ti. As it can be expected, for both catalysts the propane conversion increases with the contact time, while the selectivity to propylene decreases as result of the marked increase of the selectivity to CO, the formation of CO2 being less affected by the contact time. The formation of CO is more strongly favoured on 6V/Ti than on 6Nb/Ti in the whole range of contact time values. A similar but less marked effect was found for the ODH of ethane on the same catalysts, even if at 5508C and with an different alkane/O2 feed (Fig. 3(b) and Fig. 4(b)). This dependence of the contact time, that is a common feature of the most catalysts tested, suggests that CO2 can be formed directly from the alkane via a parallel reaction, while CO is mainly formed from the further oxidation of the alkene (or from the same reaction intermediate of the alkene) according to the following reaction path:

Fig. 3. Alkane conversion (*) and selectivity to alkene (*), CO (&) and CO2 () as a function of the contact time on 6Nb/Ti for (a) propane OXD at 4008C and (b) ethane OXD at 5508C [6].

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

297

Fig. 4. Alkane conversion (*) and selectivity to alkene (*), CO (&) and CO2 () as a function of the contact time on 6V/Ti for (a) propane OXD at 4008C and (b) ethane OXD at 5508C [6].

Obviously, the oxidation of propylene or of CO to CO2, even if occurring in lower extent, cannot be excluded. Similar reaction schemes, involving parallel and consecutive reactions, were proposed in literature [4,5,7], the selectivity to ole®n being determined by the ratio between the different kinetic constants. The effect of reaction temperature on the oxidative dehydrogenation of ethane and propane was also investigated. As examples, we report the performances exhibited by 6Nb/Ti (Fig. 5(a)) and 6V/Ti (Fig. 6(a)), the behavior of the other catalysts being intermediate

between these two cases. In Fig. 5(b) and Fig. 6(b) the results obtained on the same catalysts for the ODH of ethane are reported for comparison. In all the experiments we observed that the propane conversion increases and the selectivity to propylene decreases with the reaction temperature, all the catalysts favouring the reactions to carbon oxides at higher temperature. With respect to propane, ethane is activated at higher temperature (Fig. 5(b) and Fig. 6(b)) and the effect of the reaction temperature on the products distribution is less marked for all catalysts.

Fig. 5. Alkane conversion (*) and selectivity to alkene (*), CO (&) and CO2 () on 6Nb/Ti catalyst as a function of the temperature for (a) propane (contact time: 0.74 g s N cmÿ3) and (b) ethane (contact time: 0.6 g s N cmÿ3).

298

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

Fig. 6. Alkane conversion (*) and selectivity to alkene (*), CO (&) and CO2 () on 6V/Ti catalyst as a function of the temperature for (a) propane (contact time: 0.019 g s N cmÿ3) and (b) ethane ethane (contact time: 0.018 g s N cmÿ3).

On the other hand, the decrease of propylene selectivity with the various catalysts was different, depending on the different effect of the temperature on the formation of carbon oxides. For example, the selectivity to CO2 on 6Nb/Ti is about 30% at 3608C and decreases to 15% at 5208C, while it is mainly constant (about 20%) on 6V/Ti in the same range of temperature (see Figs. 5 and 6). In the case of the ternary catalysts (Fig. 7(a)), we observed that the presence of Nb in the catalyst composition results in a greater selectivity to CO2 at low temperature. This effect tends to disappear at increasing temperature and V/Nb ratio.

This result can be explained by supposing that the reactions reported in the scheme proposed above have different activation energies depending on the catalyst composition. Thus, at high V/Nb ratio the increase of propane conversion with the reaction temperature results from similar increase of the rate of Reaction (1) and Reaction (3) and from a higher increase of the rate of Reaction (2). On the contrary, at low V/Nb ratio the increase of the rate of Reaction (1) should be higher than the rate of Reaction (3). Therefore, the effect of temperature and of contact time con®rms that the sites

Fig. 7. Alkane conversion (*) and selectivity to alkene (*), CO (&) and CO2 () on 1V6Nb/Ti catalyst as a function of the temperature for (a) propane (contact time: 0.48 g s N cmÿ3) and (b) ethane (contact time: 0.3 g s N cmÿ3).

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

promoting the formation of C3H6 and CO are different from those promoting the formation of CO2. Moreover they indicate that the nature of these sites can be different when niobium is present in the catalyst formulation. Otherwise, these results can indicate that, after the surface activation of propane, the reaction can proceed via gas phase mechanism in the presence of niobium, according to the scheme reported by Albonetti et al. [3] for ethane ODH. Steps favoured at high temperature lead to the production of ethylene and CO whilst steps occurring on the catalysts surface lead to the total oxidation of the hydrocarbon. 4. Discussion The vanadia based catalysts studied in this work have high activity, although with a low selectivity, in propane ODH as shown by the comparison with the data of ole®n productivity reported by Cavani and Tri®roÁ [1] for a large number of catalysts. Propylene productivity of about 0.5 kg kgCATÿ1 hÿ1 at 4008C was obtained with the catalyst 6V/Ti even with a quite low selectivity. Ethylene productivity of about 1 kg kgCATÿ1 hÿ1 at 5508C was previously obtained with the same catalyst [6]. The V/Ti binary catalysts are able to activate propane due to their redox properties, the number of redox sites being higher at higher vanadium content. IPA tests con®rm our previous results [6]. However, increasing the vanadium loading results in slightly lowering the propylene selectivity, mainly as effect of consecutive CO formation (Table 3 and Fig. 1). Nevertheless, it has been pointed out that the high activity of 6V/Ti allows a higher yield of propylene

299

with respect to 1V/Ti. On the other hand, the not negligible activity of 6Nb/Ti in the ODH of propane indicates that Nb containing species can act as active and very selective sites. The low activity given by 6Nb/Ti is likely related to the low concentration of redox sites necessary for paraf®n activation, as indicated by previous characterization [6] and con®rmed by IPA decomposition test. Moreover, 6Nb/Ti is the most selective catalyst for propylene. (Table 3 and Figs. 1 and 2). Therefore, the coexistence of vanadium and niobium containing species on TiO2 surface is expected to affect the catalytic properties in propane ODH as previously found for ethane ODH [6]. The results obtained in this work have con®rmed this expectation. In fact the presence of Nb in 1V6Nb/ Ti enhances the performance of 1V/Ti, in agreement with the ®ndings for ethane ODH [6]. Further similarities of catalytic behavior in propane and ethane ODH can be recognized. In Table 4, rates of propane consumption and of propylene formation at 4008C and rates of ethane consumption and of ethylene formation at 5508C are reported for all catalysts investigated. The general trend of the catalytic behavior of vanadia and niobia based systems towards the oxidative dehydrogenation of both paraf®ns is quite similar: 1. 6Nb/Ti is the most selective catalyst for both ole®ns; 2. the catalytic activity increases with vanadium content much more than linearly either in the presence or in the absence of niobium; 3. the ratios between the rates of paraffin consumption on the binary catalysts 1V/Ti and 6V/Ti and on the ternary catalysts 1V6Nb/Ti and 6V6Nb/Ti are

Table 4 Rate of consumption of propane …rC3 H8 † and formation of propylene …rC3 H6 † evaluated at 4008C (this work), and rate of consumption of ethane (rC2 H6 and formation of ethylene …rC2 H4 † evaluated at 5508C [6] Catalyst

rC3 H8  107 (mol gÿ1 sÿ1)

rC3 H6  107 (mol gÿ1 sÿ1)

rC2 H6  107 (mol gÿ1 sÿ1)

rC2 H4  107 (mol gÿ1 sÿ1)

6Nb/Ti 1V/Ti 6V/Ti 1V6Nb/Ti 6V6Nb/Ti

1.67 2.57 85.8 3.27 39.1

0.88 0.97 29.6 1.49 17.1

3.09 9.11 262 13.8 172

2.35 4.17 104 8.53 62.4

300

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

about the same for ethane and propane, suggesting that they are activated by the same sites; 4. both propane and ethane reaction rates are higher on 1V6Nb/Ti with respect to 1V/Ti and the increase of both propane and ethane consumption rates on 6V6Nb/Ti with respect to 1V6Nb/Ti are lower than those observed on 6V/Ti with respect to 1V/Ti; 5. the addition of Nb to V/Ti catalysts has a promoting effect on the olefin selectivity at low V/Nb ratio whilst it depresses the activity at higher V/Nb ratio, only in part balanced by the increase of propylene selectivity in the case of propane ODH. These observations lead to the assumption that, when niobium and vanadium oxides are both present on the TiO2 surface, the active sites are different with respect to the V/Ti binary catalysts. However, it is noteworthy that the selectivity obtained for propane ODH are remarkably lower than those resulting from ethane ODH [6]. The small differences in propylene selectivity exhibited by the catalyst samples suggest that the ethane ODH is more sensitive to the catalyst composition than propane ODH. Moreover, all catalysts show a stronger dependence of the ole®n selectivity on the conversion in propane ODH than in ethane ODH as it can be argued by comparing the curves reported in Figs. 1 and 2 with the results showed in [6]. However, it must be pointed out that these results are likely affected also by the different oxygen/paraf®n feeding ratio in the two cases. In conclusion, all catalyst samples are able to activate ethane or propane with no remarkable differences between the two paraf®ns at suf®ciently high temperature, depending on the chain length. The selectivity to the ole®n seems to be more sensitive to the chain length of the hydrocarbon suggesting that ethylene formation depends much more on the catalyst structure than propylene formation. Moreover, this should indicate that at 5508C the reaction pathway of ethane ODH is still completely heterogeneous and that ethylene or ethyl intermediate formed by the breaking of the C±H bond of ethane evolves to oxygenated species according to the different surface acidity or reducibility of the catalyst. On the other hand, it was demonstrated that propylene reacts faster than propane over most of the catalysts reported in literature probably due to the easier oxygen insertion in the propylene molecule containing weaker C±H bonds compared to propane [4]. Moreover, it has been

reported that the enhancement of the surface acidity has a promoting effect on the selectivity to ethylene. This effect becomes less and less important with increasing the chain length of the alkane, turning into inhibiting effect towards butene formation in butane ODH [3,4]. Different hypotheses about the structure of the sites involved in the ODH reactions on vanadium based catalysts have been reported [2]. Either vanadyls V = O or V±O±V or V±O±M bridges have been proposed as active sites promoting the breaking of the ®rst C±H bond to form the adsorbed alkyl species. It has been also reported [16] that isolated tetrahedra VO4 promote the dehydrogenation while units containing bridging oxygen, more easily removable, are involved in the formation of oxygenated products. This model explains the effect of the metal±oxygen bond strength on the selectivity and the modi®cations induced by the presence of a cation on vanadium ions. Smits et al. [9] assumed that the active site in V/Nb/O catalysts is a surface vanadium atom and that optimal activity and selectivity are obtained when the vanadium site V* is in the situation V±O±V*±O±Nb. Results of the IPA decomposition test con®rm our previous ®ndings [6] that niobium interacts with vanadium affecting the redox and acidic properties, especially at low vanadium content. The formation of V±O±Nb±O±V bridges can be suggested, or the grafting of vanadium onto niobium oxide phase can be hypothesized. 5. Conclusions Vanadium and niobium oxides supported on TiO2 are active in the oxidative dehydrogenation (ODH) of propane. Although niobia/titania catalyst is also able to catalyze the ODH of propane, its low activity indicates that in vanadia containing catalysts a vanadium oxide species must be involved in the paraf®n activation as the activity increases with vanadium content, both in the absence and in the presence of niobium. The interaction between vanadium and niobium in the ternary catalysts modi®es the surface acidity of the sample leading to the formation of active centers different from those present on vanadia binary catalysts. Redox properties are less affected by this interaction and are probably involved in the alkane

P. Viparelli et al. / Applied Catalysis A: General 184 (1999) 291±301

activation and in the selective step of the reaction as well. Propane is more easily activated than ethane but lower propylene selectivity were obtained compared to ethylene. All catalysts, however, promote the same reaction path for both paraf®ns. The formation of carbon oxides in the ODH of ethane is more sensitive to the structure of the catalyst than in the ODH of propane due to the greater oxidability of propylene that can probably occur in the same extent on sites of different nature. References [1] F. Cavani, F. TrifiroÁ, Catal. Today 24 (1995) 307. [2] E.A. Mamedov, V. CorteÂs CorberaÁn, Appl. Catal. A 127 (1995) 1. [3] S. Albonetti, F. Cavani, F. TrifiroÁ, Catal. Rev. 38 (1996) 413.

301

[4] H.H. Kung, Adv. Catal. 40 (1994) 1. [5] B. Grzybowska-SÏwierkosz, Appl. Catal. A 157 (1997) 409. [6] P. Ciambelli, L. Lisi, G. Ruoppolo, G. Russo, J.C. Volta, in: G. Oyama, G. Lyons (Eds.), Studies Surf. Sci. Catal. 3rd World Congress on Oxidation Catalysis, 110, 1997, p. 285. [7] T. Blasco, LoÂpez-Nieto, Appl. Catal. A 157 (1997) 117. [8] P. Ciambelli, L. Lisi, G. Russo, J.C. Volta, Appl. Catal. B. 7 (1995) 1. [9] R.H.H. Smits, K. Seshan, J.R.H. Ross, Studies Surf. Sci. Catal. 72 (1992) 221. [10] B. Grzybowska-SÏwierkosz, Catal. Today 20 (1994) 165. [11] F. Pepe, C. Angeletti, S. De Rossi, J. Catal. 118 (1989) 1. [12] J. Cunningham, B.K. Hodnett, M. Ilyas, J. Tobin, E.L. Leahy, Chem. Soc. London, Faraday Disc. 281 (1981). [13] N. Y Topsùe, H. Topsùe, J.A. Dumesic, J. Catal. 151 (1995) 226. [14] R.M. Pittman, A.T. Bell, Catal. Lett. 24 (1994) 1. [15] J.M. Jehng, I.E. Wachs, J. Phys. Chem. 95 (1991) 7373. [16] P.M. Michalakos, M.C. Kung, I. Jahan, H.H. Kung, J. Catal. 140 (1993) 226.