Oxidative dehydrogenation of propane on vanadium supported on magnesium silicates

Oxidative dehydrogenation of propane on vanadium supported on magnesium silicates

Applied Catalysis A: General, 97 (1993) 159-175 159 Elsevier Science Publishers B.V., Amsterdam APCAT A2468 Oxidative dehydrogenation of propane o...

984KB Sizes 0 Downloads 32 Views

Applied Catalysis A: General, 97 (1993) 159-175

159

Elsevier Science Publishers B.V., Amsterdam APCAT

A2468

Oxidative dehydrogenation of propane on vanadium supported on magnesium silicates A. Corma, J.M. Lbpez Nieto and N. Paredes Instituto de Tecnologia Quimica, U.P. V.-C.S.I.C., Universidad Polite’cnica de Valencia, c/ Camino de Vera sfn, 46071 Valencia (Spain)

and M. PQrez

c/

CEPSA, S.A., Picas de Europa No. 7, Poligono Industrial S. Fernando de Henares II, Madrid (Spain) (Received 4 November 1992, revised manuscript received 11January 1993)

Abstract Sepiolite supported vanadium catalysts have been studied in the oxidative dehydrogenation of propane. Different vanadium species are present on the catalysts and the proportion changes with the preparation methods, support modifications, and vanadium content. At low vanadium content, isolated tetrahedral species are formed and both activity and selectivity to propene increase when increasing the vanadium content. At medium or high vanadium content the appearance of associated vanadium species can be observed and, while the activity still increases when increasing total vanadium, the selectivity to propene remains constant. MgVzOs and V206 crystallites are formed at vanadium content higher than 30 wt.% of VzOs, and, depending on the propane conversion level, the selectivity to propene remains constant or decreases when increasing the vanadium content. On a comparative basis, the oxidative dehydrogenation of propane bas also been carried out on a V105/Mg0 catalyst. From the comparison between catalytic properties in the oxidative dehydrogenation of propane on V/sepiolite and V / MgO catalysts a reaction network has been established and both type of catalysts show the same conversionselectivity behaviour. The nature of the sites responsible for selective and non-selective oxidation on V/sepiolite catalysts has been proposed.

Keywords: catalyst characterization

(XRD,

Raman);

oxidative

dehydrogenation;

propane;

se-

piolite as support; vanadium

INTRODUCTION

One of the driving forces of the refining and petrochemistry industry is the upgrading of cheaper feedstocks. In this line, it is of interest to produce highly to: Dr. J.M. Lopez Nieto, Instituto de Tecnologfa Qufmica, U.P.V.-C.S.I.C., Polit&nica de Valencia, c/ Camino de Vera s/n, 46071, Valencia, Spain. Fax. ( + 34-

Correspondence Universidad 6)3877996.

0926-860X/93/$06.00

0 1993 Elsevier Science Publishers B.V.

All rights reserved.

160

A. Coma et al. / Appl. Catal. A 97 (1993) 159-l 75

demanded C, and C!, alkenes from the corresponding alkanes. This can be achieved through a dehydrogenation process using Pt or Cr,O, as catalysts, and working at high temperatures in order to shift the equilibrium. The major drawback of this process, the frequent catalyst regeneration needed, due to the poisoning occurring when working at high temperatures, could be avoided if the alkenes are obtained through an oxidative dehydrogenation process. V-Mg-0 systems have been proposed as effective catalysts in the oxidative dehydrogenation of n-butane [l-3], propane [ 3-51, ethylbenzene [ 6,7], l-butene [ 7,8], and cyclohexane [ 91. Other vanadium based catalysts such as alkaline, alkaline earth, bismuth and rare earth vanadates have also shown good catalytic properties for the oxidative dehydrogenation of n-butane or propane [ 10-141. In the case of the V-Mg-0 catalyst, the exact nature of the active sites is still under debate. Kung et al. [2,3,12,14] have proposed that in order to obtain highly selective catalysts, the formation of V = 0 should be avoided, as it occurs when Mg,Vz08 species are formed. On the other hand, Volta et al. [4,5] have prepared pure and mixed Mg-V-O phases and proposed that the active and selective phase they found, i.e., a+MgZV207, is characterized by the presence of both, single and double vanadium-oxygen bonds in V-O-V pairs. However, when V/MgO catalysts are prepared, different Mg-V-O species are present [ 15,161, depending on the surface concentration of vanadium and the preparation conditions. In addition, isolated tetrahedral and/or octahedral vanadium species, their associated forms and magnesium orthovanadate have been proposed to exist at low and medium vanadium contents [ 6-8,13,17-191. Moreover, by carrying out X-ray photoelectron spectroscopy (XPS ) and secondary ion mass spectroscopy (SIMS) experiments on V,O,/MgO catalysts it has been observed that a small percentage of the MgO surface is occupied by vanadium species, and mono and bilayer vanadium oxide species are present

PO].

From this it appears that the main goal to be achieved is to find among the different vanadium and V-Mg-0 phases formed, the ones that are responsible for the oxidative dehydrogenation of alkenes. If this could be solved, then the following step, i.e. the preparation of the active phases on an optimized catalyst, could be achieved. Following this line, we have recently studied the influence of the acid-base character of the support on the catalytic properties of supported vanadium catalysts for the oxidative dehydrogenation of propane [ 131. It was proposed that the acid-base properties of the support or the metallic cations in the vanadates controls the proportion of the different vanadium species formed. In this way, on a support with acid character VZ05 is formed and catalysts with relatively lower selectivity are obtained. Meanwhile, on supports with basic characteristics such as MgO, Biz03, La,O,, or Sm,O,, orthovanadates are formed as the main phases, and relatively high selectivities to propene were achieved.

161

A. Corma et al. / Appl. Catal. A 97 (1993) 159-175

In this paper we present the results obtained during the oxidative dehydrogenation of propane, when vanadium has been supported on a carrier with basic properties such as sepiolite. This is a fibrous magnesium silicate which has been found to interact strongly with vanadium [ 21-281, up to such a point that it can be used for vanadium passivation in FCC catalysts. It will be shown here that magnesium vanadates are formed, and that the particular phases found depend on the sepiolite characteristics, as well as on the catalyst preparation procedure. Finally, the differences in catalytic activity and selectivity of the different vanadium species formed will be discussed. EXPERIMENTAL

Materials supports Natural sepiolite from Vallecas (Spain) with a general formula of Mg,S$015 ( 0H)z.6Hz0, has been used as starting material, and the chemical analysis is given in Table 1. Alkaline exchanged sepiolite was prepared following the procedure described previously [24-261 and 1.7 meq g-l of Na+ ions was introduced in the octahedral sheet of the silicate, replacing the magnesium ions located at the edges of the fibers in natural sepiolite. Alkaline exchanged as well as natural sepiolites, before and after calcination at 923 K (3 h), have been used as carriers, and their physicochemical characteristics are given in Table 2. For comparative purposes MgO was prepared by precipitation of MgS04*7Hz0 with an aqueous solution of ammonia and it was dried at 353 K (16 h) and calcined at 873 K, in air, during 6 h. Catalyst The catalysts were prepared by a ‘wet’ impregnation method of the different supports, with an aqueous vanadyl oxalate solution with a V,O,/oxalic acid TABLE 1 Chemical composition of natural vallecas-type sepiolite (wt.-% ) SiOz AI& Fe& CaO MgO Na,O KzO TiOz Weight loss on calcination at 1000 ’ C

60.31 1.88 0.48 0.17 25.48 0.12 0.26 0.22 10.88

162

A. Corma et al. /Appl.

Catal. A 97 (1993) 159-175

TABLE 2 Chemical compositions and surface area of catalysts Sample

s BJ!TO

V,06

(wt.-%)

Atomic ratio* V

Mg

Sepiolited

139

Calc-Sep’ Sep-N6 MgC

116 70.2 95.2

OV/Sepiolite SV/Sepiolite 7V/Sepiolite 12V/Sepiolite 25V/Sepiolite 37V/Sepiohte

127 92.1 82.4 66.7 53.2 32.3

2.50 6.67 12.2 24.5 37.4

1 1 1 1 1

25 9.0 4.6 2.0 1.1

20V/Cak-Sep 25V/Sep-Na

34.9 25.7

19.7 24.5

1 1

2.6 1.8

18V/MgO

88.8

18.1

1

Na”

_ 1.0

0.42

t t t t t t 0.75

13

“BET surface area ( m2 g-l). *Chemical analysis of the elements was done by atomic absorption spectrometry. Y: Traces, in accordance with Table 1. dNatural sepiolite. ‘Calcined sepiolite. ‘Na+-exchanged sepiolite.

molar ratio of 1: 3. The impregnated samples were dried at 363 K and 27 KPa. Finally they were calcined in air at 823 K, for 3 h. The natural, sodium exchanged sepiolites, calcined natural sepiolite and MgO containing vanadium will he referred as V/sepiolite, V/Sep-Na, V/Calc-Sep and V/MgO, respectively, preceded by a number that indicates the vanadium content expressed in wt.-% of V205. In this way, BOV/Calc-Sep indicates a catalyst supported on calcined sepiolite containing about 20 wt.-% of V205 (Table 2). Catalysts characterization The specific surface areas of the catalysts and supports were obtained by the BET method from the nitrogen isotherms at 77 K, using a value of 0.164 nm2 for the cross-section of nitrogen. X-ray diffraction (XRD) patterns were obtained using a Philips 1060 diffractometer operated at 36 kV and 20 mA employing nickel-filtered Cu Ka radiation (LO.1542 nm).

A. Corma et al. / Appl. Catal. A 97 (1993) 159-175

163

The laser Raman spectra (LSR) were recorded with a FT-Raman spectrometer incorporating a 4 W Spectrum 301 Nd: YAG laser and a liquid nitrogencooled germanium detector (Bio-Rad/Varian). Various laser power settings were tried for each sample so that the optimum power was selected in each case. 64 Scans were co-added to produce each spectrum for a total collection time of about two minutes each. The optimum power was selected as a compromise between signal and heating effects of the sample. The sensitivity was adjusted according to the intensity of the Raman scattering.

Catalytic studies The catalytic experiments were carried out in a fixed bed, continuous stainless steel tubular reactor (I.D. 20 mm; length 520 mm), equipped with a coaxial thermocouple for temperature profiling, and working at atmospheric pressure. Analysis of reactants and products were carried out by gas chromatography, using three columns: (i) Plot capillary, 50 m; (ii) molecular sieve 5A (1.5 x1/8 in.); (iii) Porapak Q (3.0 m~l/S in.). Catalyst samples of 1.0 to 2.0 g and particle sizes between 0.25 and 0.42 mm were mixed with variable amounts of Sic (0.59 mm particle size) to keep a constant volume in the catalyst bed (3 cm3). The flow of the reactants was varied to achieve different contact times ( W/F= 25-90 g,, h/molt, ). The reaction has been studied in the temperature interval 673-823 K and propane/ oxygen/helium molar ratios of 4/S/88 and 4/20/76. Blank runs for the different reactant mixtures have been carried out substituting the catalyst by Sic, at the maximum contact time and at the different reaction temperatures (673-873 K) studied in this work. The higher thermal conversion was obtained with a C3Hs/02/He molar ratio of 4/20/76 at 823 K and a total flow of 275 ml/min. In this case, 2.79% of propane was converted to carbon dioxide and propene. Differences in temperature along the catalytic bed never exceeded 8 K. RESULTSAND DISCUSSION

Catalysts characterization Vlsepiolite series The X-ray diffraction pattern of the natural sepiolite is characterized by the presence of the principal line at 28= 7.3 (JCPDS, 13-595). When natural sepiolite is calcined at 973 K (Fig. la) the appearance of magnesium silicate hydroxide, MgsSi12030(OH)4, as a result of the crystal water removal from sepiolite, is observed. The impregnation with vanadyl oxalate solution and subsequent calcination in air modifies the X-ray diffraction patterns of the resulting samples, showing low crystallinity (Fig. lb-f). At low vanadium con-

164

A. Cormu et al. /Appl.

Catal. A 97(1993) 159-175

Fig. 1. X-ray diffraction patterns of V/sepiolite catalysts: (a) OV/sepiolite, (b) 3V/sepiolite; (c) 7V/sepiolite; (d) lW/sepiolite; (e) 25V/sepiolite; (f) 37V/sepiolit.e. Symbols: Mg,Si12030(OH), [JCPDS, 26-12271 (A); monoclinic MgSiO, [JCPDS, 35-6101 (0); orthorombic MgSiO, [JCPDS, 19-7681 (=);/I-MgV,O, [JCPDS, 30-8021 (0).

tent, only the magnesium silicate hydroxide is observed by XRD. At intermediate vanadium content, both monoclinic and orthorombic MgSiOs are detected as well (Fig. Id-e). At the highest vanadium content studied (Fig. If), MgV206 together with MgSiOs are observed, while the Mg8Si12030(OH), phase is not detected anymore. Raman spectra of the V/sepiolite series are given in Fig. 2, while the frequencies corresponding to the observed bands are given in Table 3. At low vanadium content (7Vlsepiolite sample ) a broad band in the 700900 cm-’ range, centered at 807 cm-’ with a small band at 763 cm-’ is observed (Table 3 and Fig. 2a). While this small band cannot be assigned to magnesium vanadate phases [6,27-321, due to the broadness of the 700-900 cm-’ band the presence of magnesium vanadates cannot be completely ruled out. It has been proposed that an empirical correlation between the phenomenological V-O bond order and the frequency of the associated stretching motion exists [ 321. In this way, it has been proposed that a weak signal above 800 cm-‘, observed in vanadia silica mixed gel catalysts, corresponds to a V-O bond order less than the value found in VOi- [ 331. More recently, it has been

A. Corma et al. / Appl. Catal. A 97 (1993) 159-175

1000

000 RAMAN

600 SHIFT,

165

400 cm-’

Fig. 2. Raman spectra of V/sepiolite catalysts: (A) 7V/sepiolite; (B) 25V/sepiolite; (C) 37V/ sepiolite. Obtained at a laser power of 375 mW except (b) which was obtained at 94 mW. Symbols: y-V%(V); (VO,), (r);MgVzOs (m);V,O, (0).

reported in the V-Bi-0 system, that the symmetric stretch of a perfect y-VO, tetrahedron appears at 796 ( + 20) cm-’ [ 341. In addition to this, and according with the diatomic approximation, the following equation has been used to calculate the V-O length (R) as a function of the measured Raman stretching frequency Y(cm-‘) [ 341: R=0.52148 ln(21349/v(cm-‘))

(1)

Taking this into account, we propose that on 7V/sepiolite catalyst, symmetric VO, tetrahedron species have been formed and, from eqn. (l), V-O lengths of 1.71 and 1.74 A, corresponding to the 807 and 763 cm-’ hands, respectively, are obtained. These V-O lengths are close to 1.695-1.809 A, corresponding to V-O in a Mg,V,O, phase [ 351. In addition, the broadness of the band may indicate the presence of small differences in the symmetry of the VO, tetrahedron. At high vanadium content (37V/sepiolit.e sample) the bands observed (Fig. 2c and Table 3) clearly show the presence of MgV206 [ 6,27,30] as well as V,O,.

A. Cormo et al. / Appl.Catal. A 97 (1993)159-175

166 TABLE 3 Summary of the V/sepiolite Raman results Catalyst

Raman bands (cm-‘)

Phases”

3VJsepiolit-e 7V/sepiolite 12V/sepiolite

807,763 807,763 807,763 950 807,763 950 916 916,835,729,513,428,406 348,321,306,262 997,850,729,600 916,835,729,523,428,406 348,321,306,262 1010,898,872,839,775, 740,523,406,332,340

Y-V% Y-V% Y-V% ([email protected] )” Y-W ([email protected] ), MgVA MgVA

948,919,901 946,906,892,882,863,674 560,346,258,234,310 848,488,450,400,380 848,488,450,400,380

a-Mg,V,C, /?-NaVO,

25V/sepiolite

37V/sepiolite

SOV/Calc-Sep

25V/Sep-Na

lBV/MgC

VzCs MgVzCe 8-MgxVxC,

MgsVA M&V&

“Designation of the Raman bands according the literature data: y-VO1 (discussion in text); (VO; ), [32]; MgV206 [6,27,30]; Vz05 [6,27]; a-Mg,Vz07 [27];p-Mg2V,0, [27];BNaVO, [37]; Mg,[email protected] [6,27 1.

In samples with intermediate contents of vanadium, several vanadium species coexist. Indeed, the spectrum in Fig. 2b, obtained at lower laser power, and the values in Table 3 indicate the presence of isolated VO1 species, as well as associated vanadium species, such as polyvanadates, since [V100,,]6- and (VO, ), types show their most important bands at 970 [ 311 and 940 cm- ’ [ 321, respectively. In addition, the presence of a small band at 921 cm-l on the 25V/sepiolite sample may be assigned to the presence of a MgVz06 phase. Occelli and Stencel [ 271 and Occelli et al. [ 281 have observed in sepiolitesupported vanadium catalysts (with a 5 wt.-% of vanadium) calcined at 1153 K, the presence of mixed of MgV206 + /?-Mg2V207, while in our case crystalline phases of magnesium vanadates are obtained only at high vanadium loading. However, the lower calcination temperature used in this study could be responsible for the lower crystallinity observed. On the other hand, the presence of four-coordinate vanadium(V) on samples with lower loading vanadium (less than 3 wt.-% of vanadium) has been proposed [ 281. These results are in agreement with our data. Influence of sepiolite modifications When sepiolite is calcined at high temperature, a partial destruction of the

A. Corma et al. / Appl. Catal. A 97 (1993) 159-l 75

16’7

structure occurs, making the octahedral sheet of magnesium more accessible to vanadium. Then, when vanadium was incorporated on a previously calcined sepiolite (2OV/Calc-Sep sample), different magnesium vanadates were formed, in addition to MgsSi12030(OH) 4 and MgSiOs (Fig. 3b). The Raman spectrum (Fig. 4b) and the values from Table 3 show the presence of MgV206 together with a- and fi-Mg,V207 [ 27,301 and even also Mg,V208. Therefore, it is possible to conclude that if the support was previously calcined, a higher interaction between magnesium and vanadium occurs. By substituting part of the magnesium located at the edges of the fibers by Na+ ions, the basic character of natural sepiolite can be increased. This can influence the interaction of the vanadium with the support and consequently may influence the nature of the vanadium species formed. Indeed, when vanadium is supported on a Na+ exchanged sepiolite, differences with respect to V/sepiolite samples are observed by X-ray diffraction (Fig. 3~). While the presence of Mg,Si,,O,, (OH), and sepiolite structures are not observed any more, the presence of MgSiOB, and another diffraction lines resulting from vanadium phases are present on the 25V/Sep-Na catalyst. The poor crystallinity of the sodium containing sample, makes the assignation of new phases by XRD difficult. More information can be obtained from its Raman spectrum (Fig. 4a and Table 3). Thus, an intense band at 946 cm-‘, together with bands at 906,892,882,863,848,674,560,488,450,400,370,346,258 and 234 cm-’ are visible. Bands with frequencies around 950 cm-’ could be assigned to VO, species, as is the case for alkaline metal vanadates [ 36,371. In our case, the observed bands correspond top-NaVOs [ 371, which crystallizes in the symmetry space group Pnma-D2,, forming infinite double chains of VO, trigonal bipyramids extending along the b axis. In addition with this phase, the presence of Mg,V208 may also be proposed [ 6,271, but the presence of a-Mg,V,O,, with bands at 948,919 and 902 cm-l cannot be ruled out.

Fig. 3. X-ray diffraction patterns of the catalysts supported on modified sepiolite: (a) sepiolite calcined at 973 K, (b) 2OV/Calc-Sep; (c) 25V/Sep-Na. Symbols as Fig. 1.

A. Corma et al. / Appl. Catal. A 97 (1993) 159-l 75

1000

800 RAMAN

600 SHIFT,

400

cm”

Fig. 4. F&man spectra of the catalysts supported on modified Sepiolite: (A) 25V/Sep-Na; (B) ZOV/Calc-Sep; (C) 18V/MgO. Obtained at a laser power of 375 mW. Symbols: P-Na2V03 (0 );

MgV,Oci (ml; a-Mg,V,O, (A );8-Mg,[email protected], (*);

Mg,V,Os (U).

For comparative purposes the Raman spectrum of the 18V/MgO sample is showed in Fig. 4c. In this case the main phase detected is Mg3V20a, nevertheless smaller amounts of a-Mg,V,O, and /&Mg,Vz07 may also be present, in addition with MgO observed by X-ray diffraction. Catalytic results The catalytic results obtained during the oxidative dehydrogenation of propane on sepiolite supported vanadium catalysts are given in Table 4. Propene, carbon dioxide and carbon monoxide are the majoritary products, although, carbon monoxide is not detected at conversion levels below 3%. On the other hand, except on free vanadium catalysts (OV/sepiolite and OV/Sep-Na), methane and ethene are obtained when the reactant oxygen was consumed. In addition, partially oxygenated products were not observed on our catalysts. Since sensible differences in the specific surface areas exist in the V/sepiolite series (Table 2)) the influence of the vanadium content on specific activities, obtained at a contact time of 30 gC,,h/molc, and 773 and 823 K reaction temperatures, are showed in Fig. 5. It can be seen there that the higher the vanadium content on the catalysts, the higher the specific activity is.

A. Corma et al. /Appl.

169

Catal. A 97 (1993) 159-175

TABLE 4 Oxidative dehydrogenation of propane on sepiolite supported vanadium catalysts Reaction conditions: WIFz44.5

g,,h/molc,,;

T= 773 K, molar ratio C,/Ox/He=4/8/88

Sample

Conversion (% )

Selectivity to ‘&H, (W)”

CO/CO2 ratio

OV/sepiolite 3V/sepiolite 7V/sepiolite 25V/sepiolite 37V/sepiolite

12.4” 7.51 6.75 7.68 9.75

19.6 56.7 65.5 65.1 57.2

1.08 0.93 1.06 1.08 1.54

20V/Calc-Sep

6.15

77.6

0.89

10.7” 0.69

6.12 59.0

0.36 0

OV/Sep-Na 25V/Sep-Na

aW/F= 89 g,,h/molc, . bOther products are carbon monoxide and carbon dioxide, except on OV/sepiolite and OV/SepNa, in which the selectivities to methane+ethene are 5.3 and 3.1, respectively.

0

10

20

30

40

V2Os(wt%)

Fig. 5. Influence of vanadium content in V/sepiolite catalysts on the specific activity for propane conversion, obtained at 773 K (0 ) and 823 K (0 ). C3Hs/02/He molar ratio in the feed of 4/8/ 88.

With respect to the influence of the nature of the support, it can be said that similar catalytic properties between vanadium based catalysts supported on pure and calcined sepiolite are obtained, while on sodium-exchanged sepiolite samples the lower activities are obtained.

170

A. Corma et al. / Appl. Catal. A 97 (1993) 159-175

An inverse correlation between conversion of propane and selectivity to propene has been found, and the results obtained on different V/sepiolite catalysts fall in the same conversion-selectivity curve (Fig. 6). From the results given in Fig. 7 it appears that propene is a primary unstable product, carbon dioxide is a primary plus secondary, while carbon monoxide is a secondary product, formed from degradation of propene. Moreover, from the results showed in Fig. 7, it can also be concluded that the higher the vanadium content on the catalysts, the higher the degradation of propene and the greater the formation of carbon monoxide is. In Fig. 8 the influence of propane conversion on the CO/CO, ratio obtained on different V/sepiolite catalysts is showed. The higher the vanadium content of the catalysts, the greater the CO/CO, ratio is. For comparative purposes, the results obtained on sepiolite-supported catalysts have been compared with those on a MgO-supported vanadium catalysts (Fig. 8), in which Mg3V208 is the main magnesium vanadate phase. The greater differences in the CO/CO, ratio among all catalysts are observed between MgO- and sepiolite-supported catalysts. On the other hand, in all supported vanadium catalysts, no differences are observed when oxygen partial pressure in the reaction mixture is increased, as can be observed from the comparison between catalytic results obtained at C,H,/O, molar ratios in the feed of l/2 and l/5 (Table 5). However, on OV/ sepiolite, the higher the oxygen partial pressure, the higher the propane conversion is. Active sites for conversion of propane on V-sepiolite catalysts Calcined sepiolite, as well as MgO, are catalytically active for the transfor-

OI 0

20

conversion,

40

%

Fig. 6. Influence of propane conversion on the selectivity to propene obtained at 823 K on: 3V/ sepiolite ( A ) ,25V/sepiolite ( Cl ) , and 37V/sepiolite ( 0 ) catalysts. For comparative purposes, the results obtained on the lSV/MgO catalyst ( v ) are also included.

A. Corma et al. / Appl. Catal. A 97 (1993) 159-175

171

0B 0

IO-

3’

Jk

5

P

/

ap

/

-p, P

/

0,’

Do

ii

I

F II-

20

@

,/< 0, 0

5 lolp , 0

‘1

0

P

P

/o

/

/

I

/

/

/’

10

I

I

CONVERSION

0

40

20 %

Fig. 7. Influence of propane conversion on the yields of propene ( 0 ) , CO2 ( A ) and CO ( 0 ) : (A ) 7V/sepiolite; (B) 25V/sepiolite; (C) 37V/sepiolite. Reaction conditions: temperature, 823 K; C,/ O,/He molar ratio in the feed of 418188.

mation of propane, but carbon monoxide and carbon dioxide are the main products. When sepiolite is calcined before vanadium impregnation, the resulting catalyst shows a mixture of magnesium vanadates phases. The differences between calcined and non-calcined sepiolite-supported catalysts may be explained on the bases of magnesium accessibility and therefore more extensive Mg-V interactions. Indeed, when sepiolite is calcined, and the sepiolite structure is partially destroyed, the MgO segregates and a greater magnesium-vanadium interaction occurs. On the other hand, low activity is obtained on 25V/ Sep-Na which can be explained by the presence of sodium vanadate, P-NaVO,, which, as has been already proposed for other alkaline vanadates [lo], has a low activity and selectivity for the oxidative dehydrogenation of propane.

172

A. Coma et al. / Appl. Catal. A 97 (1993) 159-l 75

40

20

conversion,% Fig. 8. Influence of the propane conversion on the CO/CO2 ratio obtained on V/sepiolite catalysts: 3V/sepiolite (+); 7V/sepiolite ( A ); 25V/sepiolite ( 0 ); and 37V/sepiolite (0 ). For comparative purposes, the results obtained on the lSV/MgC catalyst (v ) are also included. Reaction conditions: temperature, 823 K; CJOJHe molar ratio in the feed of 4/8/88. TABLE 5 Influence of oxygen content in the feed on the catalytic properties of V/sepiolite series at 823 K Sample

C3/C2a

W/Fb

Conversion (%)

Selectivity to CaH, (% )

co/co, ratio

OV/sepiolite

418 4120 418 4120 418 4120 418 4120

89.0 89.0 29.5 36.1 59.0 59.0 59.0 59.0

25.0 65.5 13.6 13.6 25.0 25.3 32.2 33.7

22.6 3.85 57.0 54.3 43.2 42.4 31.2 29.8

1.22 1.05 1.22 1.17 1.38 1.33 1.92 1.76

7V/sepiolite 25V/sepiolite 37Vfsepiolite

“Molar ratio CsHs/O.JHe = 4/x/100 - (x + 4). bContact time, W/F, in g,,h/molc,.

The catalytic properties and the presence of different vanadium phases are related with the vanadium loading of the catalysts, as is presented in Fig. 9. For vanadium contents below 10 wt.-%, isolated VO, species were detected and their concentration increases linearly with total vanadium content. The evolution of activity and selectivity with vanadium loading indicate that isolated VO, species are highly active and selective for the oxidative dehydrogenation of propane.

A. Corma et al. / Appl. Catal. A 97 (1993) 159-175

0.8

a

0

10

I

I

20

30

173

V20,(WT%)

Fig. 9. Influence of the vanadium content of the catalysts on: (a) the appearance of vanadium phases; (b) the specific activity for propane conversion at 823 K, (c) the selectivity to propene obtained at 823 K and a propane conversion level of 4% (0 ) and 20% (m ); (d) the CO/CO, ratio obtained at 823 K and a propane conversion level of 20%.

For vanadium contents above 10 wt.-% of V205, the amount of isolated VO, species decreases, while MgV,O, and (VO; ), species are formed. From the fact that activity still increases in this range and selectivity remains constant (Fig. 9) it is clear that MgV,O, and (VO, ), species are also active and selective for the desired reaction. However, when the vanadium loading is 25 wt.-% or higher, V,O, is also formed. Since total conversion is still increasing but selectivity to propene decreases and the CO/CO2 ratio increases for these loadings, it appears that V,O, does oxidize propane and propene but mainly to carbon dioxide and carbon monoxide. In conclusion, on V-0-Mg pairs, in which V = 0 bonds are not present, as occurs in V-Mg based and on V/sepiolite catalysts at low vanadium loading,

174

A. Corma et al. / Appl. Catal. A 97 (1993) 159-l 75

the oxidative dehydrogenation occurs with high selectivity to propene. In addition to this, in the presence of V-O-V or VV pairs, with V= 0 double bonds, both the rate of H-abstraction from the alkane and the rate of degradation of the formed alkene increases, producing a corresponding decrease in selectivity. For this reason, although the vanadium in a tetrahedral position is less active than the vanadium in octahedral position, in the first case a greater yield to propene is obtained. In the V-sepiolite series, high CO/CO, ratios are obtained, indicating that both propene formation and propene degradation occurs, this effect being more important when vanadium content on the catalysts increases. On the other hand, the low CO/CO, ratio obtained on supported magnesium orthovanadate samples indicates that propane reacts by oxidative dehydrogenation to form propene and also by total combustion to give carbon dioxide. However, a lower propene degradation than on V/sepiolite exists. Nevertheless, and since the same selectivities are obtained on both catalytic systems, V/sepiolite appear as effective catalysts in the oxidative dehydrogenation of alkanes. ACKNOWLEDGEMENT

Financial support by the CICYT (Project MAT 607/91)is acknowledged. N.P. thanks Ministerio de Education y Ciencia for a fellowship. The authors thank Bio-Rad/Varian for assisting in the Raman measurements.

REFERENCES

6 7 8 9 10

11

12

H.H. Kung and M.A. Chaar, US Patent 4772319 (1988). M.A. Chaar, D. Patel, M.C. Kung and H.H. Kung, J. Catal., 105 (1987) 483. M.A. Chaar, D. Pate1 and H.H. Kung, J. Catal., 109 (1988) 463. D. Siew Hew Sam, V. Soenen and J.C. Volta, J. Catal., 123 (1990) 417. A. Guerrero-Ruiz, I. Rodriguez Ramos, J.L.G. Fierro, V. Soenen, J.M. Herrmann and J.C. Volta, in P. Ruiz and B. Delmon (Editors), New Developments in Selective Oxidation by Heterogeneous Catalysis (Studies in Surface Science and Catalysis, Vol. 72 ) , Elsevier, Amsterdam, 1992, p. 203. J. Hanuza, B. Jezowska-Trzebiatowska and W. Oganowski, J. Mol. Catal., 29 (1985) 109. A.V. Simakov, N.N. Sazonova, S.A. Ven’yaminov, J.P. Belonestnykh, N.N. Rozhdestvenskaya and G.V. Isagulyants, Kinet. Catal., 30 (1987) 598. A.V. Sirnakov and S.A. Veniaminov, React. Kinet. Catal. Lett., 28 (1985) 67. M.C. Kung and H.H. Kung, J. Catal., 128 (1991) 287. K. Seshan, H.M. Swaan, R.H.H. Smits, J.C. van Ommen and J.R.H. Ross, in G. Centi and F. Tritiro (Editors), New Developments in Selective Oxidation (Studies in Surface Science and Catalysis, Vol. 55), Elsevier, Amsterdam, 1990, p. 505. Y.-C. Kim, W. Ueda and Y. Moro-Oka in G. Centi and F. Trifiro (Editors), New Developments in Selective Oxidation (Studies in Surface Science and Catalysis, Vol. 55), Elsevier, Amsterdam, 1990, p. 491. D. Patel, P.J. Andersen and H.H. Kung, J. Catal., 125 (1990) 132.

A. Corma et al. / Appl. Catal. A 97 (1993) 159-175 13

14 15 16 17 18 19 20 21 22 23

24 25 26 27

28 29 30 31 32 33 34 35 36 37

175

A. Corma, J.M. tipez Nietq N. Paredes, M. Pkrez, Y. Shen, H. Cao and S.L. Suib, in P. Ruiz and B. Delmon (Editors), New Developments in Selective Oxidation by Heterogeneous Catalysis (Studies in Surface Science and Catalysis, Vol. 72), Elsevier, Amsterdam, 1992, p. 213. O.S. Owen, M.C. Kung and H.H. Kung, Catal. Lett., 12 (1992) 45. M.C. Kung and H.H. Kung, J. Catal., 134 (1992) 668. G.M. Clark and R. Morley, J. Solid State Chem., 16 (1976) 429. V.A. Fenin, V.A. Shveta and V.B. Kazanskii, Kinet. Catal., 16 (1975) 902. V.M. Vil’al’ba, K.N. Spiridonov and O.V. Krylov, Kinet. Catal., 20 (1979) 1077. O.V. Lapina, A.V. Simakov, V.M. Mastikhin, S.A. Veniaminov and A.A. Shubin, J. Mol. Catal., 50 (1989) 55. A.A. Rar, A.V. Simakov and S.A. Veniamihoov, Reac. Kinet. Catal. L&t., 39 (1989) 299. M.L. Occelli, in M.L. Occelli (Editor), Fluid Catalytic Cracking Role in Modern Refining (ACS Symposium Series, Vol. 375), American Chemical Society, Washington, 1988, p. 162. J.I. De Jong, Europ. Pat., 832 018 352 (1981). A. Corma, V. For&, A. Mifsuf and J. P6rez Pariente, in M.L. Occelli (Editor), Fluid Catalytic Cracking II (ACS Symposium Series, Vol. 452), American Chemical Society, Washington, 1991, p. 293. A. Corma, V. Fom&, A. Mifsud and J. P&z Pariente, Sp. Pat. 510 013 (1982). A. Corma, A. Mifsud and J. PBrez Pariente, Sp. Pat. 527 756 (1983). A. Corma, A. Mifsud and J. P6rez Pariente, US Pat. 4 542 002 (1985). M.L. Occelli and J.M. Stencel, in M.L. Occelli (Editor), Fluid Catalytic Cracking Role in Modem Refining (ACS Symposium Series, Vol. 375), American Chemical Society, Washington, 1988, p. 195. M.L. Occelli, R.S. Maxwell and H. Eckert, J. Catal., 137 (1992) 36. L.V. Kristallov, M.P. Tsvetkova and A.A. Fotiev, Russ. J. Inorg. Chem., 29 (1984) 990. L.V. Kristallov, A.A. Fotiev and M.P. Tsvetkova, Russ. J. Inorg. Chem., 27 (1982) 1714. W.P. Griffith and P.L.B. Lesniak, J. Chem. Sot. A, (1969) 1066. W.P. Griffith and T.D. Wickins, J. Chem. Sot. A., (1966) 1087. A. Wokaun, M. Schramil and A. Baiker, J. Catal., 116 (1989) 595. F.D. Hardcastle, I.E. Wachs, H. Eckert and D.A. Jefferson, J. Solid State Chem., 90 (1991) 194. N. Krishnamachari and C. Calvo, Can. J. Chem., 49 (1971) 1629. G.A. Kourouklis, A. Jayaraman, G.P. Espinosa and A.S. Cooper, J. Raman Spectrosc., 22 (1991) 57. L.V. Kristallov and L.A. Perdyaeva, Russ. J. Inorg. Chem., 36 (1991) 879.