TiO2 supported vanadyl phosphate as catalyst for oxidative dehydrogenation of ethane to ethylene

Applied Catalysis A: General 203 (2000) 133–142

TiO2 supported vanadyl phosphate as catalyst for oxidative dehydrogenation of ethane to ethylene P. Ciambelli a , P. Galli b , L. Lisi c,∗ , M.A. Massucci b , P. Patrono d , R. Pirone c , G. Ruoppolo e , G. Russo e a

Dipartimento di Ingegneria Chimica e Alimentare, Università di Salerno, via Ponte Don Melillo, 84084 Fisciano (SA), Italy b Dipartimento di Chimica, Università di Roma ‘La Sapienza’, Rome, Italy c Istituto di Ricerche sulla Combustione, CNR, Napoli, Italy d IMAI-CNR, Area della Ricerca di Roma, Monterotondo Scalo, Rome, Italy e Dipartimento di Ingegneria Chimica, Università di Napoli ‘Federico II’, Napoli, Italy Received 8 October 1999; received in revised form 27 January 2000; accepted 6 February 2000

Abstract Bulk and TiO2 supported VOPO4 has been investigated for the oxidative dehydrogenation of ethane. XRD, SEM, TG analyses and BET surface area measurements indicated that vanadyl phosphate is highly dispersed on the support up to mono-layer coverage. A fraction of vanadium is present as V(IV) in the calcined samples as evaluated by EPR and TPR techniques. Both reducibility and acidity of vanadium phosphate is strongly enhanced by deposition on TiO2 with respect to the bulk phase, as shown by TPR and NH3 TPD technique, respectively. The supported catalysts are active and selective in the oxidative dehydrogenation of ethane to ethylene in the temperature range 450–550◦ C, the mono-layer catalyst giving the best performances. Ethylene selectivity decreases with the contact time but increases with the temperature. The former effect indicates that ethylene is further oxidized to COx at high contact times. The effect of the temperature was attributed to the formation of V(IV), favoured at increasing temperature. This hypothesis was supported by TPR experiments carried out after catalytic tests at 550◦ C that indicated a significant increase of the fraction of V(IV) after the reaction. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Ethane oxidative dehydrogenation; Vanadium phosphate; Redox properties; Acid sites

1. Introduction In recent years a large number of papers emphasised the capability of various vanadium-based catalysts to favour the oxidative dehydrogenation (ODH) of C2 –C4 alkanes [1,2]. For these catalysts the co-presence of V(IV) and V(V) is believed to favour the selective oxidation to desired alkenes via a redox ∗ Corresponding author. E-mail address: [email protected] (L. Lisi).

process according to the Mars–Van Krevelen mechanism [3]: the hydrocarbon reacts with lattice oxygen reducing the catalyst which, in turn, is re-oxidized by oxygen present in the gas phase. However, on vanadium-based catalysts high selectivity is achieved only at low alkane conversion, resulting in alkene productivity far from that required for industrial application. This is also the case of bulk and TiO2 supported (VO)2 P2 O7 pyrophosphate, investigated as catalyst for the oxidative dehydrogenation of ethane [4] and propane [5]. The effect of supporting vanadium

0926-860X/00/$ – see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 0 ) 0 0 4 8 1 - 6

134

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

pyrophosphate on different oxides such as alumina, silica or titania has been widely investigated, TiO2 promoting the best dispersion [9–11]. It was found that the stronger interaction of titania with (VO)2 P2 O7 affects the vanadium reducibility leading to catalysts more active but less selective in the oxidation of n-butane to maleic anhydride [10,11]. Less attention has been addressed to other vanadium-phosphate phases, such as the vanadyl phosphate dihydrate, VOPO4 ·2H2 O [6]. It has a layered, tetragonal structure consisting of sheets of (VOPO4 )∞ connected to each other through molecules of water located in the interlayer space. The (VOPO4 )∞ sheets are composed of highly distorted VO6 octahedra sharing four oxygen atoms with four different PO4 tetrahedra groups. One of the two water molecules is coordinated to the vanadium atom in axial position, thus completing the VO6 vanadium octahedron; the other is anchored more loosely through H-bonds to the oxygens of the tetrahedral phosphate groups [7,8]. In this work we have investigated VOPO4 ·2H2 O either as bulk either as supported phase for ethane ODH. VOPO4 ·2H2 O has been supported in different amounts on TiO2 and the effect of the vanadium phosphate dispersion on the physico-chemical and catalytic properties has been studied. 2. Experimental 2.1. Preparation of materials VOPO4 ·2H2 O, thereinafter indicated as VOP, was prepared by refluxing V2 O5 in 85 wt.% H3 PO4 aqueous solution for 16 h according to Ladwig’s method [6]. VOP was also supported at various loadings by impregnating high surface area pure anatase TiO2 with different amounts of water solubilised VOP and drying at 80◦ C. The catalyst precursors were afterwards calcined at 550◦ C for 3 h in air flow. 2.2. Chemical analysis and physical measurements Vanadium and phosphorus content of the catalyst precursors was determined by dissolving a 100 mg sample with 5 ml of concentrated sulphuric acid in a water boiling bath and diluting the resulting solution to 100 ml. Vanadium was determined by

Inductively Coupled Plasma Emission Spectroscopy (ICPES) technique with a Varian Liberty 150 Model at λ=309–311 nm, while phosphorus was determined colorimetrically with a Perkin–Elmer 555 Model Spectrophotometer at λ=430 nm, according to the method described in [8]. A Philips PW 1100 diffractometer was employed for obtaining X-ray diffraction patterns (XRD) of the materials at room temperature. Ni-filtered Cu K␣ radiation was used and the 2ϑ measurements were accurate to 0.05◦ . BET surface areas were measured by N2 adsorption at 77 K with a Quantachrom CHEMBET 300 instrument. A Stanton Redcroft STA-801 simultaneous TG/ DTA thermoanalyser was used to study the thermal behaviour of the precursors (heating rate 10◦ C/min to 1000◦ C in air flow, Pt crucibles, Pt-Pt/Rh thermocouples). The electron paramagnetic resonance (EPR) spectra were obtained at X-band frequency with a Varian E-9 spectrometer equipped with a standard Oxford instrument low-temperature attachment. Scanning electron microscopy (SEM) was performed with a Philips XL30 apparatus equipped with an EDAX instrument. Temperature Programmed Reduction (TPR) with hydrogen and Temperature Programmed Desorption (TPD) of NH3 were carried out using a Micromeritics TPD/TPR 2900 analyser equipped with a TC detector and coupled with a Hiden HPR 20 mass spectrometer. In the TPR experiments the sample was reduced with a 2% H2 /Ar mixture (25 cm3 min−1 ) by heating at 10◦ C min−1 to 650◦ C. In the TPD experiments the sample was saturated with pure ammonia at room temperature for 1 h and, after purging with pure He for 2 h, it was heated (10◦ C mim−1 ) to 650◦ C in flowing He (25cm3 min−1 ). The samples were treated in flowing air at 550◦ C for 2 h. Catalytic activity tests were carried out with the experimental apparatus described in [12], equipped with a fixed bed quartz micro-reactor operating under atmospheric pressure. The reaction products were analysed with a Hewlett–Packard series II 5890 gas-chromatograph equipped with a thermal conductivity detector for the analysis of O2 , CO and CO2 and a flame ionisation detector for the analysis of hydrocarbons. The concentrations of O2 , CO and CO2 were also measured on line with a Hartmann & Braun

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

135

URAS 10 E continuous analyser. Water produced during the reactions was kept by a silica gel trap in order to avoid condensation in the cold part of the experimental apparatus. The feed composition was 4% C2 H6 and 2% O2 in a balance of He. The reaction temperatures were 450 or 550◦ C. The contact time ranged from 0.006 to 0.12 g s N cm−3 . Carbon balance was closed within 3% error in all experiments.

3. Results and discussion The list of catalysts investigated is reported in Table 1 (column 1), where VOP stands for VOPO4 ·2H2 O and Ti for TiO2 . The chemical composition of catalysts is reported in column 2 as VOP weight percentage of the catalyst precursor. In column 3 the VOP surface density, i.e. the number of vanadium phosphate molecules per square nanometer of the catalyst, calculated from the VOP content and the BET surface area, is also reported. The cell parameters of VOPO4 ·2H2 O [13] were used to evaluate the theoretical mono-layer coverage of TiO2 . According to this calculation it corresponds to a VOP loading of 9.5 wt.%. Therefore, Table 1 shows that the VOP content of the three supported catalysts ranges from lower to higher than mono-layer composition. 3.1. XRD and SEM analyses X-ray powder diffraction patterns of both precursors and catalysts revealed the presence of only TiO2 anatase phase indicating high dispersion of vanadium phosphate on the support. EDAX analyses confirmed the homogeneity of VOP dispersion in the sub mono-layer and mono-layer samples. A less homogeneous distribution of VOP was found for VOP/Ti-10.

Fig. 1. TG curves of TiO2 and VOP/Ti catalysts.

3.2. Surface area measurements The BET surface areas of the calcined catalysts are reported in Table 1. The values obtained for VOP/Ti-7 and VOP/Ti-9 are equal to that of pure TiO2 (125 m2 g−1 ). The slight reduction of surface area observed for VOP/Ti-10 is likely due to the above mentioned non-homogeneous distribution of VOP on the support. 3.3. Thermal behaviour The TG curves of TiO2 and VOP/Ti precursors are reported in Fig. 1, showing very similar trends. The weight change of TiO2 , due to water loss, occurs in two steps [14,15]. In the temperature range 20–175◦ C

Table 1 List of VOP catalysts with some characteristics Sample

VOP content (wt.%)

VOP surface density (units nm−2 )

BET surface area (m2 g−1 )

Hydration water (<175◦ C) (molecules nm−2 )

OH (T>175◦ C) (groups nm−2 )

TiO2 VOP VOP/Ti-7 VOP/Ti-9 VOP/Ti-10

– 100 7.0 9.6a 10.7

– – 1.7 2.3 3.0

125 0.8 125 125 110

9.4 – 8.4 8.3 8.3

11.6 – 9.8 9.2 8.4

a

This VOP content corresponds to theoretical mono-layer coverage of TiO2 surface.

136

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

titania loses hydration water, while above 175◦ C water is lost very slowly up to 700◦ C due to condensation of surface OH groups. The amount of water loss evaluated from TG analysis is reported in Table 1 as hydration water and OH groups (11.6 OH nm−2 ). A value of about 14 OH nm−2 has been reported by Haber et al. [16] for anatase TiO2 . The OH surface density of the supported materials is lower than that of TiO2 and decreases with increasing the VOP content indicating that an interaction takes place between VOP and OH surface groups of TiO2 . Taking into account that the presence of hydroxyl groups on VOP surface is not expected, as confirmed by the negligible weight loss in the temperature range 150–700◦ C [17], a ratio between OH surface groups of TiO2 and VOP can be evaluated. The density values of VOP molecules reported in column 3 very well agree with those of OH groups involved in the interaction with VOP, calculated as the difference between the OH groups of TiO2 and those of TiO2 supported VOP, thus indicating a stoichiometric ratio between VOP and Brønsted sites of the support in the formation of V–O–Ti bonds. At temperature higher than 700◦ C, while TiO2 does not give anymore mass loss, the VOP/Ti precursors begin to slowly decompose according to a redox process leading to release of O2 [16].

3.4. EPR measurements The EPR spectra of bulk VOP and the different VOP/Ti precursors show identical features indicating the presence of vanadium(IV) in a fraction not less than 5%. Spectrum of VOP/Ti-9 sample, chosen as representative, is reported in Fig. 2 (spectrum a). The signals are axially symmetric and exhibit a fairly good resolution, suggesting chemical equivalence and magnetic dilution of V4+ . The average magnetic parameters of the materials are reported in Table 2, together with those of VO(H2 O)5 2+ [18] and 1–2% V4+ /VOPO4 ·2H2 O [19] for comparison. The magnetic parameters of the precursors are almost superimposable to those reported for VO(H2 O)5 2+ , thus suggesting that tetravalent vanadium should be likely present as a VO2+ hydrated species hexa-coordinated to both lattice and water molecules oxygens. Furthermore, since the magnetic parameters of VOP/Ti-7, VOP/Ti-9 and VOP/Ti-10 precursors are practically

Fig. 2. EPR spectra of VOP/Ti-9 (a) at room temperature as precursor; (b) after air treatment at 550◦ C followed by immediate sealing of the cell; (c) at room temperature after ODH reaction at 550◦ C.

identical to those of 1–2% V4+ /VOPO4 ·2H2 O, the formation of a VO(H2 O)2+ –O–P species in the materials can be supposed. The EPR spectra of the precursors calcined at 550◦ C and those of the same materials after the ODH reaction at the same temperature were also recorded. As an example, in Fig. 2 the EPR spectra of the VOP/Ti-9 compounds are reported. When the VOP/Ti-9 precursor (spectrum a) is heated at 550◦ C and the water molecules are completely removed, the dehydrated solid (the catalyst) gives an EPR spectrum (spectrum b) indicating a new magnetic VO2+ site. The new VO2+ species exhibits an axial hyperfine line and the EPR parameters, reported in the last row of Table 2 as gk , g⊥ and A⊥ values, are quite similar to those of the two reference compounds (Table 2, rows 1 and 2) confirming the octahedral geometry of the new vanadyl group. However, the lower value of Ak , that could be produced from the saturation of the sixth co-ordination of vanadyl species with the oxygen of TiO2 support, should indicate a tetragonally distortion of the primary octahedral co-ordination.

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

137

Table 2 EPR parameters for some vanadium(IV)-containing species and for the VOP/Ti precursors and catalysts Species and catalysts

gk

g⊥

Ak (cm−1 )

A⊥ (cm−1 )

VO(H2 O)5 2+ [18] 1−2% V4+ /VOPO4 ·2H2 O [19] VOP/Ti-7 (precursor) VOP/Ti-9 (precursor) VOP/Ti-10 (precursor) VOP/Ti-9 (calc. 550◦ C)

1.936 1.935 1.934 1.932 1.932 1.936

1.982 1.977 1.975 1.975 1.975 1.990

−0.0178 −0.0175 −0.0175 −0.0175 −0.0176 −0.0161

−0.0070 −0.0063 −0.0076 −0.0076 −0.0076 −0.0066

After the catalytic experiment at 550◦ C, the material has been left in air and then submitted to EPR measurements. Signals evidencing the presence of two V4+ sites corresponding to hydrated and anhydrous phases of VOP/Ti-9 are clearly observed (spectrum c). This result indicates not only that a reversible hydration process occurs, but also that a pyrophosphate phase is not produced during the catalytic reaction. 3.5. TPR measurements In Fig. 3 the TPR profiles of VOP and VOP/Ti calcined catalysts are reported. All samples show a single peak except for a small shoulder appearing in the curve of VOP/Ti-7. It must be noted that the support undergoes a negligible reduction [12]. Unsupported VOP is reduced at high temperature (peak

Fig. 3. TPR curves of VOP (· · · ), VOP/Ti-10 (· · · —), VOP/Ti-9 (—) and VOP/Ti-7 (- - -).

value > 650◦ C) whereas the dispersion on TiO2 results in easier reduction, as shown by the lower values of peak temperature (Table 3). It can be noted that also the extent of reduction increases when VOP is supported. This could be attributed to limited diffusion of H2 across the VOP particle, however, TPR measurements carried out on bulk VOP with different particle dimension gave the same result of the first experiment suggesting that the reducibility of VOP has been modified by interaction with the support. All supported catalysts show values of V/H2 ratio ranging from 1 to 2 indicating the presence of both V4+ and V5+ in the fresh samples if the +3 oxidation state of vanadium after the TPR experiment is assumed. The presence of V4+ is in agreement with the results of EPR analysis. The amount of H2 consumed per vanadium is about constant up to the mono-layer coverage in contrast with the results obtained by Overbeek et al. [10] who found reducibility increasing with the vanadium phosphate coverage up to a pyrophosphate content corresponding to the mono-layer. The reducibility decreases when the mono-layer coverage is exceeded (VOP/Ti-10 catalyst) likely due to the formation of hardly reducible VOP aggregates. Also in this case our results do not agree with those of Overbeek et al. [10] indicating a different redox behaviour of vanadium ortophosphate and vanadium pyrophosphate. Each sample of calcined catalyst was treated at 550◦ C for 2 h in flowing air after TPR experiment and then a second TPR was carried out. The same result of the first experiment was obtained in every case indicating that the reduction/oxidation process is completely reversible. TPR of the catalysts samples before calcination were also carried out. The same TPR curves of calcined materials were obtained suggesting that the thermal treatment does not change the original oxidation state of vanadium in the hydrated compound.

138

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

Fig. 4. NH3 TPD curves of TiO2 (· – ·), VOP/Ti-10 (· · · · ), VOP/Ti-9 (—) and VOP/Ti-7 (- - -).

3.6. TPD measurements The ammonia TPD curves of VOP/Ti catalysts and TiO2 [12] and the relevant amounts of desorbed NH3 are reported in Fig. 4 and in Table 3. No desorption of NH3 was observed for unsupported VOP in our experiments, while Bagnasco et al. [17] reported NH3 desorption from 7.4×10−4 to 10.5×10−4 mol g−1 of hydrated VOP, due to ammonia intercalation between the VOP structure layers. In our case the thermal treatment at 550◦ C results in the complete loss of co-ordination water and, as a consequence, in the collapse of the layered structure. Therefore, NH3 can only be adsorbed on the external surface of VOP that is lower than 1 m2 g−1 resulting in undetectable amounts. Ammonia TPD spectrum of pure TiO2 shows an intense and broad signal [12] which is progressively

reduced as far as the titania surface is covered by VOP up to mono-layer coverage. The presence of VOP also modifies the TPD profile of the support showing the contribution of low strength acid sites not present on the TiO2 surface. If TiO2 surface is supposed to be completely covered in the mono-layer sample (VOP/Ti-9) the adsorption of ammonia can be attributed only to dispersed VOP which, as shown in Table 3, should have a lower density of acid sites with respect to TiO2 . As a consequence, in the VOP/Ti-7 sample either bare TiO2 surface and dispersed VOP should contribute to NH3 adsorption. By supposing a homogeneous dispersion of VOP in VOP/Ti-7 catalyst 27% of TiO2 surface can be considered available for ammonia adsorption and a value of 2.6×10−6 mol NH3 m−2 is expected, in good agreement with the experimental result. This further confirms the high dispersion obtained for the mono-layer and sub mono-layer samples. Above the mono-layer coverage the non-homogeneous distribution of VOP gives rise to a higher adsorption of NH3 likely due to partially uncovered TiO2 surface. 3.7. Catalytic activity tests In all catalytic tests only C2 H4 , CO and CO2 were produced, no oxygenated compounds being detected in the reaction products. Moreover, O2 conversion was always far from 100%. Unsupported VOP gave rise to near zero ethane conversion at 550◦ C, while at 580◦ C a 3% C2 H6 conversion with 95% selectivity to C2 H4 , decreasing to 65% at 10% conversion, was found. In Figs. 5–7 the conversion of C2 H6 and O2 , and the selectivity to C2 H4 , CO and CO2 at 450◦ C and 550◦ C are reported for VOP/Ti catalysts as a function of the contact time. It can be observed that the catalytic activity is dramatically enhanced when VOP is supported on titania,

Table 3 H2 uptake, V/H2 ratio, peak temperature resulting from TPR experiments and NH3 desorption resulting from TPD experiments Catalyst

H2 uptake×104 (mol g−1 )

V/H2 (mol mol−1 )

Tmax (◦ C)

NH3 desorption ×106 (mol m−2 )

VOP VOP/Ti-7 VOP/Ti-9 VOP/Ti-10

15.5 3.3 4.3 3.4

4.0 1.3 1.4 1.9

650 533 509 518

– 2.7 2.3 3.0

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

139

Fig. 5. O2 (䊉) and C2 H6 (䊏) conversion, and selectivity to C2 H4 (䊐), CO (1), CO2 (∇) as a function of contact time for VOP/Ti-9. (T=450◦ C (a), T=550◦ C (b)).

as the same ethane conversion is obtained on VOP/Ti catalysts at 550◦ C at about two orders of magnitude lower contact time when compared to bulk VOPO4 . Moreover, selectivity to ethylene at a given ethane conversion appears to be maintained on supported VOP. Specific activity of vanadium can be evaluated from catalytic activity data of bulk and supported VOP. At 550◦ C a rate of about 3×10−3 molecules of reacted ethane per atom of vanadium per second was calculated for bulk VOP on the base of vanadium surface density of VOPO4 . This value is one order of magnitude lower than that calculated for the supported

catalysts (Table 4) suggesting that the increase of catalytic activity is due not only to the enhanced surface area but also to the interaction between VOP and TiO2 . For all VOP/Ti catalysts the selectivity to C2 H4 decreases and that to CO increases by increasing the contact time, while that to CO2 , present in lower amount, shows a weak dependence on it. This suggests that C2 H4 produced by C2 H6 ODH is further oxidised to CO at high contact time while CO2 is mainly formed by ethane direct oxidation, in agreement with the results previously obtained with VOx /TiO2 catalysts [12].

Fig. 6. O2 (䊉) and C2 H6 (䊏) conversion, and selectivity to C2 H4 (䊐), CO (4), CO2 (5) as a function of contact time for VOP/Ti-7. (T=450◦ C (a), T=550◦ C (b)).

140

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

Fig. 7. O2 (䊉) and C2 H6 (䊏) conversion, and selectivity to C2 H4 (䊐), CO (4), CO2 (∇) as a function of contact time for VOP/Ti-10. (T=450◦ C (a), T=550◦ C (b)).

With respect to the effect of VOP loading on the catalytic performance, it should be first recalled that TiO2 investigated by us is weakly active in ethane ODH at 550◦ C [12], but its catalytic activity is negligible with respect to VOP/Ti catalysts. At 450◦ C VOP/Ti-7 and VOP/Ti-9 exhibit about the same activity as shown by the values of the rate of ethane consumption reported in Table 4. Otherwise, at 550◦ C the catalyst with the mono-layer coverage is the most active. The sample with VOP loading exceeding the mono-layer coverage shows lower catalytic activity. Michalakos et al. [4] tested bulk vanadyl pyrophosphate for ethane ODH in the temperature range 305–425◦ C obtaining lower ethylene selectivities with respect to our VOP based catalysts in the same range of ethane conversions reported in this work. Ethy−1 can be lene productivity lower than 10−3 kg kg−1 cat h ◦ evaluated from their data at 425 C while a value of

−1 (Table 4) is obtained with the about 10−2 kg kg−1 cat h VOP/Ti samples at a reaction temperature only 25◦ C higher. Moreover, the activity of VOP/Ti catalysts is similar to that of the most active VOx /TiO2 catalyst previously investigated by us [12,20] while the selectivity and, therefore, the ethylene yield is greater. The effectiveness of either vanadium–phosphorous interaction and active phase dispersion is confirmed by the good performances of V-␣Ti phosphate obtained by Santamar`ıa-González et al. [21]. The selectivity to ethylene increases with the reaction temperature for all samples (Fig. 8) Although this effect was not observed by Michalakos et al. [4] it must be noticed that they changed temperature and contact time at the same time in their experiments likely nullifying the two opposite effects. On the other hand, a similar behaviour in ethane ODH was observed by Blasco et al. [22], Le Bars et al. [23] and by Dejoz et al. [24] in the oxidation of n-butane for

Table 4 Rate of ethane consumption and ethylene formation evaluated at 450 and 550◦ C Catalyst

VOP/Ti-7 VOP/Ti-9 VOP/Ti-10

Ethylene productivity (kg kg h−1 )

rC2 H6 ×102 (mol mol V s−1 )

rC2 H4 ×102 (mol mol V s−1 )

450◦ C

550◦ C

450◦ C

550◦ C

450◦ C

550◦ C

0.10 0.09 0.05

0.7 1.13 0.88

0.4 0.3 0.2

3.2 3.3 2.4

0.3 0.2 0.1

2.0 2.3 1.6

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

141

Table 5 H2 uptake in TPR experiments H2 uptake×104 (mol g−1 )

Catalysts VOP/Ti-7 VOP/Ti-7 VOP/Ti-9 VOP/Ti-9 VOP/Ti-9

Fig. 8. C2 H4 selectivity as a function of C2 H6 conversion for the VOP/Ti-7(4), VOP/Ti-9 (䊊), and the VOP/Ti-10 (䊐) catalysts. (T=450◦ C (filled symbols), T=550◦ C (open symbols).

vanadium oxide supported catalysts. On the contrary, VOx /TiO2 catalysts promote the oxidation to COx at high temperature. The effect of the temperature could be explained by a possible modification of the active phase under the reaction conditions as reported by Santamar`ıa-González et al. [21] for V-␣Ti phosphate catalysts who observed the formation of ␣-VOPO4 under reaction conditions leading to an increased ethylene selectivity. In our case, however, any modification of active phase should not be involved since catalytic activity at 450◦ C was restored cooling down the catalyst after the test at 550◦ C. Furthermore, the transient period observed by Santamar`ıa-González et al. [21] during the ethane ODH was not detected for our catalysts suggesting that a phenomenon different from a structure rearrangement occurs to determine the different catalytic behaviour at high temperature. Since both Le Bars et al. [23] and Dejoz et al. [24] explained the increase of selectivity to ODH products at high temperature with a greater reducibility of vanadium, the effect of temperature on the catalytic properties of VOP/Ti samples was further investigated carrying out TPR experiments immediately after a catalytic test (12 h at either 450 or 550◦ C with 15% constant ethane conversion) cooling down the catalyst under He flow and without performing any thermal pre-treatment. The results, reported in Table 5 for VOP/Ti-7 and VOP/Ti-9, show that the H2 consump-

after after after after after

reaction at 450◦ C reaction at 550◦ C reaction at 450◦ C reaction at 550◦ C C2 H6 treatment at 550◦ C

3.2 2.5 4.3 3.2 2.1

tion of the fresh catalyst is unchanged upon ODH reaction at 450◦ C, while it is strongly decreased after reaction at 550◦ C. The relevant H2 uptakes reported in Table 5 indicate that the reduction of vanadium is promoted by high temperatures and that sites containing reduced vanadium are more active and selective, since the formation of ethylene is enhanced at higher temperature. Catalysts reduced by H2 in the TPR experiments were reoxidised at 550◦ C and then catalytic tests at 450 and 550◦ C were carried out. The results obtained in these tests were the same as obtained with fresh catalysts, suggesting that the reduction of VOP occurring during the reaction is reversible at all. In order to verify that vanadium phosphate can be reduced by ethane, a TPR of VOP/Ti-9 was performed after treating the sample at 550◦ C in flowing ethane and in the absence of oxygen. Ethylene, CO and CO2 were detected in the outlet reactor during the treatment until lattice oxygen was available. The H2 uptake calculated from the TPR curve after this treatment is lower than both that consumed by the fresh catalyst and that consumed by the catalyst after the ODH reaction at 550◦ C, indicating that ethane is able to reduce vanadium and that, as expected, the reduction is limited by the presence of O2 . The different behaviour observed by us [20] for VOx /TiO2 catalysts does not exclude that a redox mechanism occurs also in the absence of phosphorous as these catalysts could have reached an oxidation state that does not further affect the selectivity in the temperature range 450–550◦ C due to deeper vanadium reduction starting at lower temperature. According to this hypothesis, a less marked change of selectivity should be observed at T>550◦ C for VOP/Ti catalysts, nevertheless, the low structure stability of anatase TiO2 used as support did not allow us to verify it.

142

P. Ciambelli et al. / Applied Catalysis A: General 203 (2000) 133–142

4. Conclusions TiO2 supported VOPO4 catalysts are active and selective in the oxidative dehydrogenation of ethane giving one order of magnitude higher ethylene productivity than that reported for (VO)2 P2 O7 . The best ethylene yield was obtained with vanadium phosphate loading not exceeding the mono-layer coverage. The enhanced performances of supported VOP are only partially due to the increased surface area of the catalyst, since also the redox properties are modified. The reduction of vanadium phosphate by ethane during the ODH reaction results in improved catalytic performances. Acknowledgements P. Galli and MA. Massucci thank Ministero della Ricerca Scientifica e Tecnologica (MURST) for financial support. P. Patrono wish to thank the National Research Council for financial support. References [1] F. Cavani, F. Trifirò, Catal. Today 24 (1995) 307. [2] T. Blasco, J.M. Lopez Nieto, Appl. Catal. 157 (1997) 117. [3] D. Mars, D.W. Van Krevelen, Chem. Eng. Sci. Special Suppl. 3 (1954) 41. [4] P.M. Michalakos, M.C. Kung, I. Jahan, H.H. Kung, J. Catal. 140 (1993) 226.

[5] L. Savary, J. Saussey, G. Costentin, M.M. Betthar, M. Gubelmann-Bonneau, J.C. Lavalley, Catal. Today 32 (1996) 57. [6] G. Ladwig, Z. Anorg. Chem. 338 (1965) 266. [7] M. Tachez, F. Theobald, J. Bernard, A.W. Hewat, Rev. Chim. Min. 19 (1982) 291. [8] D.N. Bernhart, A.R. Wreath, Anal. Chem. 27 (1955) 440. [9] A. Ramstetter, M. Baerns, J. Catal. 109 (1988) 303. [10] R.A. Overbeek, P.A. Warringa, M.J.D. Crombag, L.M. Visser, A.J. van Dillen, J.W. Geus, Appl. Catal. A 135 (1996) 209. [11] R.A. Overbeek, A.R.C.J. Pekelharing, A.J. van Dillen, J.W. Geus, Appl. Catal. A 135 (1996) 231. [12] P. Ciambelli, L. Lisi, G. Ruoppolo, G. Russo, J.C. Volta, Studies Surf. Sci. Catal. 110 (1997) 287. [13] H.R. Tietze, Austral. J. Chem. 34 (1981) 2035. [14] S.R. Morrison, in: J.R. Anderson, M. Boudart (Eds.), Catalysis Science and Technology, Vol. 3, 1982, p. 209. [15] B.A. Morrow, Studies Surf. Sci. Catal. 57A (1990) 196. [16] J. Haber, A. Kozlowska, R. Kozlowski, J. Catal. 102 (1986) 52. [17] G. Bagnasco, L. Benes, P. Galli, P. Patrono, M. Turco, V. Zima, J. Thermal Anal. 52 (1998) 615. [18] R.N. Rogers, G.E. Pake, J. Chem. Phys. 33 (1960) 1107. [19] D. Ballutaud, E. Bordes, P. Courtine, Mater. Res. Bull. 17 (1982) 519. [20] P. Viparelli, P. Ciambelli, L. Lisi, G. Ruoppolo, G. Russo, J.C. Volta, Appl. Catal. A. 184 (1999) 291. [21] J. Santamar`ıa-González, M. Mart`ınez-Lara, M.A. Bañares, M.V. Mart`ınez-Huerta, E. Rodr`ıguez-Castellón, J.L.J. Jiménez-López, A. Fierro, J. Catal. 181 (1999) 280. [22] T. Blasco, A. Galli, J.M. López-Nieto, F. Trifirò, J. Catal. 169 (1997) 203. [23] J. Le Bars, A. Auroux, M. Forrisier, J.C. Vedrine, J. Catal. 162 (1996) 250. [24] A. Dejoz, J.M. López-Nieto, F. Melo, I. Vásquez, Ind. Eng. Chem. Res. 36 (1997) 2588.