Reactivity of vanadia with silica, alumina, and titania surfaces

Spectroscopic and Activity Studies on Vanadia Supported on Titania and Phosphorus-Modified Titania. Elmer C. Alyea, L. Jhansi Lakshmi, and Zhang Ju...
1 downloads 0 Views 751KB Size
Langmuir 1990,6, 801-806 high-resolution mass spectrometer was supported in part by National Science Foundation Grant BBS-8704089. R.D.B.thanks the Oklahoma State University Center for Water Research for fellowship support. We thank George L. Dorsey of Phillips Petroleum Co. for elemental anal-

801

yses and Gena Martin for help with octadecylsilica preparations. Registry No. SiO,, 7631-86-9; PEO-PPO block polymer, 106392-12-5;(EtO),Si, 78-10-4.

Reactivity of Vanadia with Silica, Alumir-a, and Titania Surfaces Margarita del Arco, Marl'a J. Holgado, Cristina Marth, and Vicente Rives* Departamento de Quimica Inorgbica, Universidad de Salamanca, Facultad de Farmacia, Avda. del Campo Charro, sln, 37007-Salamanca, Spain Received May 25, 1989. I n Final Form: November 17, 1989 Vanadia has been,supported (two monolayers) on silica, alumina, and titania (rutile) by melting V,O, at 973 K or by impregnation of these oxides with oxalic aqueous solutions of NH,VO, and further calcination at 773 or 973 K. Evolution of the phases thus formed and of the surface texture has been monitored by X-ray diffraction (XRD), electronic spectroscopy (ultraviolet-visible/diffuse reflectance, UVvis/DR), and nitrogen adsorption at 77 K. Formation of AlVO, is observed on alumina after heating the impregnated samples at 973 K and, to a smaller extent, when starting from V,O,. Vanadia XRD peaks are detected on silica (well-defined peaks for sample calcined at 773 K and poorly defined peaks for both samples calcined at 973 K)and on titania (but only after calcining at 773 K);in this case these peaks vanish after calcination at 973 K, probably because of an improved dispersion after melting of the vanadia. While sintering of unloaded alumina and silica does not take place at 973 K, the presence of vanadia deeply favors such a process, especially in the case of silica; sintering is already observed with bare rutile after calcination at 973 K, and in the presence of vanadia the behavior differs depending on the way (impregnation or melting) it had been incorporated. The results have been interpreted on the basis of previous data with similar systems using anatase and MgO as supports and with an acidity scale for binary oxides, the acidity decreasing in the order V,O, < SiO, C TiO, C Al,O, < MgO. Thus reaction between extreme oxides (alumina and magnesia on one side and vanadia on the other) is favored. The different results observed between the rutile- and the anatase-supported samples are due to the crystallographic fit between vanadia and rutile structures.

Introduction Vanadium oxides form a group of industrially important catalysts for partial oxidation of hydrocarbons, and much research has been done to understand the nature of active sites as well as the role played by the carrier (chosen to favor the formation of certain active sites) of the supported catalysts.' So, vanadia is supported on different carriers depending on the nature of the reaction to be catalyzed: on silica to oxidize naphthalene,2 on titania, mainly anatase, to oxidize o-xylene,' or on alumina to oxidize b e n ~ e n ein ; ~addition, other compounds containing vanadates are used to dehydrogenate ethylbenzene to styrene.' These catalysts are generally obtained by impregna-

* Author t o whom all correspondence should be addressed.

(1) Gellinp, P. J. In Catalysis; Bond, G. C., Webb, G. Eds.; Specialist Periodical Reports; The Royal Society of Chemistry: London, 1985: Vol. 7. D 105. (2) Weskkman, D. W. B.; Foster, N. R.; Wainwright, M. S. Appl. Catal. 1982,3,151. ( 3 ) Wachs, I. E.; Saleh, R. Y.; Chan, S. S.; Chersich, C. C. Appl. Catal. 1985,15,339. (4) Lucas, J.; Vanderwell, D.; Waugh, K. C. J. Chem. Soc., Faraday Trans. I 1981, 77, 15; 1981, 77, 31. (5) Hanuza, J.; Jezowaka-Trzebiatowska, B.; Oganowski, W. J. Molec. Catal. 1985, 29, 109. ~

0743-7463/90/2406-0801$02.50/0

tion of the support with aqueous or organic solutions containing a precursor salt of vanadium such as NH4V03, VO(acac),, or VOCl,.l.s Gas-solid reactions (using VOCl, as precursor) have been also reported to be effective;' it seems to be definitely established that the incorporation of vanadium species takes place through reaction with the hydroxyl groups of the s u p p ~ r t , and '~~ the solid thus obtained is sometimes washed to remove unreacted vanadium salt and then calcined to form vanadia. We have previously reported the preparation of V,O,/ TiO, (refs 9 and 10) and V20,/Mg0 (ref 11)systems using conventional impregnation and gas-solid reaction methods, as well as melting of vanadia on the support. As many other research groups, we have concluded that reac(6) Van Hengstum, A. J.; Van Ommen, J. G.; Bosch, H.; Gellinp, P. J. Appl. Catal. 1983,5, 207. (7) Bond, G. C.; Bruckman, K. Faraday Discuss. Chem. SOC.1981,

-.(8)Bond, G. C.; Kdnig, P. J. Catal. 1982, 77, 309.

72. 235. ---

(9) Del Arco, M.; Holgado, M. J.; Martin, C.; Rives, V. J. Catal. 1986,

998. 19. ---I

(10) Martin, C.; Rives, V. Adsorption Sci. Technol. 1985, 2, 241; J. Colloid Interface Sci. 1987,120, 469. (11)Del Arco, M.; Holgado, M. J.; Martin, C.; Rives, V. J. Mater. Sci. Lett. 1987, 6,616.

0 1990 American Chemical Society

802 Langmuir, Vol. 6,No. 4,1990

tivity of vanadium species with the support depends on the nature of the support, from the chemical point of view (basic or acid surfaces) and also on the crystallographic phase of the support (rutile or anatase). In the present paper, a systematic work is presented on the evolution of systems where vanadia has been incorporated onto the most frequently used supports (silica, alumina, rutile), and the results are compared with our previous results obtained with anatase and magnesia.'." Rutile was chosen instead of anatase and despite this seems to be more effective as a support for partial oxidation of hydrocarbons as we have also observed that rutilization of the vanadia/anatase systems takes place very easily when calcination is performed above 773 K, and we tried to avoid changes in the support nature when increasing the calcination temperature, to care only about changes in the supported phase. In addition, and for comparison purposes, samples were also obtained by melting vanadia on the supports, and, in the case of rutile, the effect of increasing amounts of vanadia was also studied. Experimental Section Materials and Methods. Three commercially available oxides were used as supports in the present study: silica (Aerosil-200, batch 394) and alumina (AluminumOxid C, batch RV0005) were both from Degussa (Frankfurt,F.R.G.),and titania (rutile,batch NP 85/10) was from Tioxide International (U.K.). All three oxides were kindly supplied by the manufacturers. Ammonium metavanadate was from Panreac (p.a., Spain),and vanadium pentoxide was obtained from NH,VO, after calcination in air at 773 K for 5 h. The X-ray diffraction diagrams (3O I 20 5 65O) were recorded in a Philips PW 1070 instrument, using nickel-filtered Cu K q radiation (A = 154.05 pm) and standard conditions. The electronic spectra (UV-vis/DR) of the samples in the range 880200 nm were obtained in a Shimadzu UV-240 spectrophotometer provided with a diffuse reflectance accessory and a Shimadzu PR-1 graphic printer, using MgO or parent supports as reference, with a slit of 5 nm. The nitrogen (S.C.O., 99.95%) adsorption isotherms (77 K) were measured in a conventional high-vacuum system (residual pressure less than lo-, N m-2) equipped with a mercury vapor diffusion pump, McLeod gauge, and grease-free stopcocks;pressure changeswere monitored with a MKS pressure transducer, and the system was calibrated with helium (S.C.O., 99.998%);the samples were outgassed in situ at 420 K for 2 h before the adsorption experiments were performed. Analysis of the isotherms for specific surface area and porosity was carried out with the assistance of a computer program developed by us1*and run in an Apple Macintosh computer. Sample Preparation. Preparationof the samples has been summarized in Scheme I. Supports were calcined overnight in air at 773 K. For every support, three sets of samples were obtained: two by impregnation with aqueous solutions of NH,VO, (containing a small amount of oxalic acid to aid dissolution of the salt), drying in air overnight at 380 K, and further calcination at 773 or 973 K (sets 1-773 and 1-973,respectively); another by melting the support and V205 at 973 K (set M). The relative amounts of NH,VO, or V,O, and the support were chosen in every case to yield two monolayers of the supported phase, from the specific surface area of the support after calcination at 773 K and the surface area occupied by a "molecule" of V205.13To check the effect of the vanadia loading on the properties of the final solids, in the case of series M samples prepared on rutile, four samples with 1, 2, 5, and 10 monolayers were prepared. For comparison,the unloaded supports were submitted to the calcination treatment at 973 K as well. For the samples, a code consisting of four fields is used: support (S, A, and T for silica, alumina, and titania, respective(12) Martin, C.; Rives, V.; Malet, P. Powder Technol. 1986, 46, 1. (13) Roozeboom, F.; Fransen, T.; Mars, P.; Gellings, P. J. 2.Anorg. A&. Chem. 1979, 449, 25.

del Arco et al.

Scheme I

(parent s w p o r t )

C~Iclnall0n/aIr/ouern1ghl/7?3K

NllqU03 lag. ol(allt actdl

I

ly), series (I or M for impregnation or melting),calcination temperature (in K), and number of monolayers of vanadia (2 for silica and alumina; 1, 2, 5, or 10 for rutile).

Results X-ray Diffraction Studies. The crystallographic phases detected by XRD in all samples studied here are given in Table I. After calcination a t 773 or 973 K, X-ray diffraction diagrams of the supports showed the presence of amorphous silica, y-alumina, and rutile as the only components. In samples corresponding to series 1-773, the peaks recorded correspond to the same phases of the support and to V,O,, and no peak due to any ternary M-V-0 (M = Si, AI, Ti) compound was recorded (see Figure 1 for aluminasupported samples). In the case of rutile, however, the peaks due to V,05 were fairly weak. After calcination a t 973 K (series 1-973), no appreciable change was observed in the X-ray diffraction pattern of the sample supported on silica, the intensities of the peaks due to V,O, were enhanced for the sample supported on titania, and new peaks, due to AlVO, in addition to those of V,O,, develop in the diagram of sample A-1-973-2 (Figure 1). The formation of aluminium vanadate upon reaction of vanadia and alumina has been previously reported by Roozeboom et al.13 The peaks due to the supports are stronger and narrower than in the diagrams of the bare supports calcined a t 973 K, indicating an improved crystallization. The behavior shown by samples corresponding to series M is fairly similar: AIVO, peaks (weaker than for sample A-1-973-2) and V,O, peaks were recorded for sample A-M-973-2, and for all other samples (except for T-M973-1) peaks due to V,O, were recorded in addition to peaks due to the supports. The peaks due to V,O, are stronger and narrower for sample S-1-773-2 than for the other two samples obtained on silica, indicating that the crystallinity (and/or the average crystallite size) of V,O, is larger after calcination a t 773 K than after calcination at 973 K. Some sort of equilibrium seems to exist between V,O, and AlVO, in samples obtained a t 973 K, as the peaks due to the former are more intense for sample A-M973-2 than for sample A-1-973-2,while the opposite behavior is observed for the peaks due to AIVO,. N, Adsorption Isotherms. Specific surface area values (as determined by the BET method) for all samples are indicated in Table I. For reasons given below, values are indicated as square meters per gram of catalyst

Langmuir, Vol. 6, No. 4, 1990 803

Reactivity of Vanadia with Different Carriers

Table I. Summary of Properties of Samples Studied in This Work sample s-773 s-973 S-1-773-2 S-1-973-2 S-M-973-2 A-773 A-973 A-1-773-2 A-1-973-2 A-M-973-2 T-773 T-973 T-1-773-2 T-1-973-2 T-M-973-1 T-M-973-2 T-M-973-5 T-M-973-10 0

color white white yellow ochreous ochreous white white greenish yellow greenish yellow orangish yellow white white yellow brown light brown brown brownish black black

crystallographic phases amorphous amorphous amorphous + V205 amorphous + V,O, amorphous + V20, r-A1203 y-A1203

y-Al,03 + V,O, y-A1203 V205+ AlVO, -pA120, V,05 + AIVO, rutile rutile rutile + V205 rutile + V205 rutile + V,On (weak) - ---, rutile v;O, rutile + V,05 rutile + V205

+ +

+

%

va

34.0 34.0 34.0 17.2 17.2 17.2 3.4 3.4 1.7 3.4 8.4 16.5

SBETb

sBmc

207.5 206.0 102.5 9.7 13.3 105.1 105.1 71.6 22.5 20.7 20.5 9.5 17.5 8.6 8.1 4.1 2.6 1.2

207.5 206.0 164.7 15.6 21.4 105.1 105.1 93.6 29.4 27.1 20.5 9.5 18.6 9.1 8.3 4.3 3.0 1.5

g of V/IOO g of support. m2 (g of catalyst)-'. m2 (g of support)-'

15

2

2.5

3

L

5

6 dlpm.10'

Figure 1. X-ray diffraction diagrams of samples prepared on alumina.

and of support. Nitrogen adsorption isotherms were recorded in the 0 5 P/PoI0.95 range for all samples. For silica, reversible type I1 isotherms and t-plots passing through the origin were obtained, indicating the lack of microporosity in this support calcined at 773 or 973 K.14 The same behavior is observed for samples S-1-773-2 and S-M973-2, but for sample S-1-973-2, the plateau in the isotherm extends in a wide range of relative pressure (0.20.6) and the t-plot, upon extrapolation, gives a positive zero intercept corresponding to adsorption on micropores amounting to an equivalent surface area of 4.3 m2 g-'. The cumulative surface area for this sample, as calculated following the method of Cranston and Inkley," was 5.4 m2 g-', coinciding with the value calculated from the slope of the t-plot for this sample. Summarizing, calcination at 773 and 973 K of unloaded silica does not modify the texture of the solid, but the simultaneous presence of vanadium gives rise to a marked sintering: SBET decreases to 50% upon calcination at 773 K (S-1-773-2) and to 5 % upon calcination at 973 K (S-1-973-2 and S-M973-2). In the case of samples obtained by impregnation, micropores, with equivalent surface area amounts of ca. 50% of the SBET, are developed after calcination at 973 K (Table I). The behavior observed for the samples obtained on alu(14)Lippens, B. C.; de Boer, J. H. J. Catal. 1965,4, 319. (15) Cranston, R. W.; Inkley, F. A. In Advances in Catalysis; Eley, D. D., Pines, H., Weisz, P. B., Eds.; Academic Press: New York, 1957; Vol. 9, p 143.

mina is quite similar to that of the silica-supported samples: type 11, hysteresis-free isotherms are obtained in all cases, but no micropores are developed. Again, calcination in the absence of vanadia does not decrease the specific surface area, but a decrease in SBET to 70% is observed for sample A-1-773-2 and of 20% for samples calcined at 973 K (Table 1). Contrary to the behavior of silica and alumina, unloaded rutile decreases its specific surface area to 50% when calcined at 973 K (Table I). The simultaneous presence of vanadia does not severely affect the SBmvalues for the samples obtained by impregnation (20.5 m2 g-' for the support calcined at 773 K and 17.5 m2 g-' for sample T1-773-2; 9.5 m2 g-l for the support calcined at 973 K and 8.6 m2 g-' for sample T-1-973-2, Table I), but SBET for sample T-M-973-2 is ca. 20% that of the parent support. For samples obtained by melting V,O, on TiO, containing increasing amounts of vanadium, a steady decrease of the specific surface area is observed as the vanadium content is increased. Diffuse Reflectance Spectra. The colors of the samples are indicated in Table I. Diffuse reflectance spectra of the samples were recorded vs MgO and vs the untreated supports. The spectra were coincident for samples supported on silica and alumina, but those supported on rutile showed an intense absorption at 320200 nm that is cancelled when rutile is used as the reference; this absorption is due to an 0,- Ti4+ chargetransfer process.16 The spectra of the samples supported on silica and alumina have been submitted to a deconvolution analysis, assumingGaussian-typecurves and keeping the number of components to the mininum available to achieve a good fit (Figure 2). For sample S-1-773-2, a broad band between 240 and 500 nm is recorded: components at 233, 307, 363, and 460 nm indicate the presence of V205.However, although for samples S-1-973-2 and S-M-973-2 the patterns of the spectra are similar, a shift toward larger wavelengths is observed, and the spectra can now by adjusted to four components at 235 f 1, 317,393 f 4, and 494 f 4 nm, indicating the presence of V4+ species coexisting with V205(ref 5) (Figure 2). For samples supported on alumina, three-component bands are found for sample A-1-773-2, centered at 237, 348, and 437 nm, due to V,05. However, for samples A1-973-2 and A-M-973-2, four components are needed to achieve a good fit with the experimental spectra, with

-

(16) Borgarello, E.; Kiwi, J.; Gratzel, M.; Pelizzetti, E.; Visca, M. J.

Am. Chem. SOC.1982,104,2996.

804 Langmuir, Vol. 6, No. 4, 1990 200

del Arco et al.

2w

i

.

-

s._-

E

Y E

i .-L Y

I A-1-973-2

A-1-773-2

*+++

++++ + + m++ +++ ++

1

E

b

x ++++

+++

++

++++++ +++++

+*,

0

3110

iz

200

100

E .-

~

3

100

S-1.973-2

+:

200

300 400 500 wavelengthhm

600

++

O

+++++

++++e$+ +*+ +++ ++ + +++

0

e+

+-

0 200

+++

+

400 500 wavelengthhm

300

600

Figure 2. Visible-ultraviolet (diffuse reflectance) spectra of selected samples: (+) component bands; (m) sum of component bands; (-)

experimental spectra.

maxima at 230, 315 f 2, 396 f 4, and 473 f 5 nm, Figure 2. This behavior is similar to that observed with the silica-supported samples. Finally, for samples supported on rutile, the spectra are dominated by a strong absorption centered at 410420 nm and a long absorption tail extending up to 800 nm that increases in intensity as the vanadium content increases, when using rutile as reference. The first band, at 410-420 nm, has been previously ascribed by us17 to the formation of peroxide species on the surface of rutile, stabilized by the presence of guest cations (in this case, vanadium). With low vanadium contents, a weak shoulder at ca. 470 nm is also recorded.

Discussion The main features of the samples have been summarized in Table I. From results above, it is clearly concluded that the three supports tested in this work react in very different ways with vanadium. This conclusion can be drawn from the changes observed in the specific surface areas of the samples and in their UV-vis/DR spectra. As already reported by Haber et a1.,18 no reaction seems to exist between silica and vanadia, and this support seems to behave quite closely to an "inert" support. No new V-Si-0 phase is observed, but a drastic decrease in S,, is measured upon calcination at 973 K, whichever (impregnation or melting) method is used to incorporate vanadium on the silica surface (9.7 and 13.3 m2 g-' respectively, vs 206 m2 g-' for the support calcined at 973 K). On the other hand, the XRD peaks due to V20, are very sharp for sample S-1-773-2 but not so well defined (and less peaks are recorded) in the diagrams of samples S-1-973-2 and S-M973-2. This behavior may be due to the melting of vana(17) Del Arco, M.; Holgado, M. J.; Martin, C.; Rives, V. Spectrosc. Lett. 1987, 20,201. (18)Haber, J.; Kozlowska, A,; Kozlowski, R. J. Catal. 1986, 102,52.

dia crystallites on heating at 973 K and their redispersion on the silica surface, thus leading to thinner vanadia crystallites that would probably block pores in the silica support and eventually ease agglomeration of the silica primary particles. Electronic spectra are fairly similar for samples S-1-773-2 and S-1-973-2, Figure 2, but a shift toward longer wavelengths is observed for samples calcined at 973 K. With regards to alumina, no reaction between vanadia and alumina seems to have taken place upon calcination at 773 K (sample A-I-773-2), but the presence of AlVO, is readily detected from the XRD patterns of samples A-1-973-2 and A-M-973-2; the amount of AlVO,, if compared to that of V20,, is larger in the former sample. Although this series of samples also shows a sharp decrease in their SBET (Table I), it is not as drastic as for samples obtained on silica. Differences are also observed in their UV-vis/DR spectra, Figure 2, probably due to the presence of AlVO,. In other words, the reaction observed between the support (alumina) and the supported phase (vanadia) leading to the formation of AlVO, seems to stabilize the surface of the former. Finally, with regard to rutile, the unloaded support and the samples obtained by impregnation reduce their specific surface area upon calcination at 973 K (Table I), but this surface area reduction is even larger in the presence of vanadia with samples belonging to series M, the surface area reduction steadily increasing as the vanadium content does. We have previously shown" that melting of vanadia on the surface of magnesia readily leads to formation of surface crystalline phases of MgBV20,and Mg2V20,, where [VvO,] species exist. On the contrary, by use of XRD no ternary phase (V-M-0) was detected on titania (neither rutile or anatase) or silica, and AlVO, is found on alumina in the present study. This different reactivity should be related to the surface acidity of the support

o

Langmuir, Vol. 6, No. 4, 1990 805

Reactivity of Vanadia with Different Carriers and of vanadia. Smith has recently proposed" an acidity scale for binary oxides from thermochemical data for oxo acid salts. According to this scale, where an arbitrary value of 0.0 is given to arameter Q (a measure of the oxide ability to accept 0 ions) for water, the values for MgO, Al,O,, TiO,, SiO,, and V20s are -4.5, -2.0, 0.7, 0.9, and 3.0, respectively; the lower the value of a, the larger the basicity of the oxide. Bearing in mind these data, it should be concluded that reaction of vanadia (the most acidic oxide among those here studied) with magnesia and alumina (the most basic oxides) should proceed very easily, while reaction with amphoteric or more or less acidic oxides (titania and silica) is hardly to be expected. The species formed on the surface of the support should be stable in the experimental conditions here used, although the temperature used to obtain samples A-1-973-2 and A-M-973-2 is fairly close to the temperature reported to destroy AlVO, to alumina and vanadia, and so vanadia is also detected by XRD in these samples. The changes observed in the specific surface area values of the samples and collected in Table I also deserve further comment. Catalysts formed by supported vanadium are generally described through the vanadium content in weight; however, this value is useless if the specific surface area of the catalysts is unknown. On the other hand, it is well established that these catalysts exhibit their best performances when the supported phase, preferably in the form of a monolayer, completely covers the surface of the support. In some cases, active sites in the support surface may lead to total oxidation, instead of partial oxidation, thus decreasing the selectivity toward the usually desired products. However, as our results clearly show, the presence of vanadia leads to a deep decrease in the specific surface area (specially with large vanadium contents), and then the final number of monolayers is usually larger than the number previously assumed from the starting SBmof the support. From data in Table I corresponding to systems with two monolayers of vanadia &e., the amount of vanadia correspondingto two monolayers assuming the specific surface area of the starting, unloaded support), this sintering effect is very marked on silica, with decreases to 50% and 5% on calcining at 773 and 973 K, respectively; for alumina, such values are 70% and 2090, respectively. The SBET decrease is probably lower in the case of the alumina-supported systems due to the formation of surface compounds (aluminum vanadates, as detected by XRD). With rutile, however, such a sintering is already observed upon calcination of the support at 973 K in the absence of vanadium (sample T-973), and no further sintering exists in the samples obtained by impregnation (20.5 m2 g-' for sample T-773 and 17.5 m2 g-' for sample T-1-773-2; with samples calcined at 973 K, 9.5 m2 g-' for sample R-973 and 8.6 cm2 g-' for sample T-1-973-2, Table I). This lack of interaction between rutile and vanadia, in systems obtained by impregnation similar to that studied here, has been already reported by Vejux and Courtine,' and Haber et ala2' However, with samples obtained by melting, a steady decrease of SBET is observed as the vanadium content increases (Table I, samples T-M-973-x, x = 1, 2, 5, 10). With vanadia-anatase systems? however, calcination at 773 K is enough to reduce SBpT to 50% of its original value without rutilization. This sintering does not seem to be related to the basicity of the support, according to

5600

I

I

A-1-773-2 o

F-

(19) Smith, D. W. J. Chem. Educ. 1987,64,480. (20) Vejux, A.; Courtine, P. J. Solid State Chem. 1978,23, 93. (21) Haber, J.; Machej, T.; Czeppe, T. Surf. Sci. 1985, 151, 301,

01

o

.

0.4

.

0.8

.

12

.

1.6

I o

.

a4

0.8

1.2

1.6

IO

I / nm

Figure 3. de Boer's t-plot for samples prepared on alumina

and silica.

the definition by Smith" that makes no difference between rutile and anatase, but is probably related to the crystallographic fit between vanadium octahedra and/or tetrahedra and the surface structure of the support. Jonson et al. have recently r e p ~ r t e da~study ~l~~ of silica- and alumina-supported vanadia catalysts obtained by impregnation with aqueous solutions of NH,VO,, calcination, and oxidation a t 773 K (i.e., a treatment similar to that given to our series 1-773 catalysts). According to these authors, V,O, is formed on silica with agglomerates containing Vw when the vanadium content is larger than 10%;these species develop from surface [VO,] units that become bridged through oxide ions as the vanadium content increases. In the case of the alumina-supported catalysts, polymeric chains of [VO,] units grow on the surface of the alumina support and then become linked, sharing edges as the vanadium content increases. The formation of thin surface layers of vanadia on alumina due to wetting of the support by the former has been described by Haber24and Knozinger et al.26 This dissimilar behavior (formation of chains growing from the surface in the case of alumina or formation of vanadia clusters from isolated vanadium-containing species in the case of silica) may account for the different reactivity observed with our samples. On the other hand, formation of "patches" of V-containing species on silica would ease the linkage between primary silica particles, thus favoring even more their agglomeration and decreasing the specific surface area (see Table I). For alumina, however, if chains grow from the surface of the particles, such an agglomeration may still leave mesopores unblocked, and then the decrease in the specific surface area would not be so large as for the silica-containing samples. Our results from nitrogen adsorption confirm these assumptions: the pore size distribution analysis indicates that mesopores with a diameter close to 9 nm, absent in bare alumina, develop in sample A-1-773-2, while the V-t de Boer plot yields a straight line for sample A-773. For sample A-1-773-2, the V-t de Boer plot deviates upward, indicating an increased uptake of nitrogen due to condensation in larger pores. On the contrary, in the case of the silica-supported samples, the V-t plots are very similar for samples S-773 and S-1-773-2, Figure 3, indicating that no new pores develop after vanadium incorporation, and the pores size distribution curves are qualitatively identical. If the specific surface area of the samples is referred to 1g of support instead of 1g of catalyst (see Table I), the decrease observed for sample S-I-7732 with respect to the SBElfvalue for unloaded silica is 20% (207.5 to 164.7 m2 g 1 while for sample A-1-773-2 (22) Jonson, B.; Rebenstorf, B.; Larsson, R.; Andemson, S.L. T.; Lundin, S. T. J. Chem. SOC.,Faraday Trans 1 1986,82, 767. (23) Jonson, B.; Rebenstorf, B.; Larsson, R.; Anderseon, S. L. T. J. Chem. SOC..Faraday Trans 1 1988.84. 1897. - - ,~ - (24) Haber, J. P h e Appl. Chem. 1984,56, 1663. (25) Leyrer, J.; Margraf, R.; Tagaluer, E.; Kn&inger, H. Surf. Sci. 1988,201,603.

806

Langmuir 1990,6, 806-816

it is only 11%; this difference may be due to the different way the primary particles are linked through the vanadium-containing surface species. The results obtained with the samples calcined a t 973 K (through impregnation or through melting) are practically coincident in these two sets of samples (silica- and alumina-supported), as melting of the supported phase will probably lead to a very effective blocking of the mesopores and a "decoration" of the primary particles of the support with some sort of vanadia film that will behave also as an adhesive agent, thus soldering the primary particles of the support. The results obtained with the rutile-supported samples are quite different from those with the silica- and the alumina-supported samples. First of all, the support itself reduces its specific surface area by calcination at 973 K (see Table I, samples T-773 and T-9731, and the S, values for the rutile-supported samples obtained by impregnation are fairly close to those for the unloaded support submitted to calcination at the same temperatures. However, larger SBET decreases are found for samples corresponding to series M, than with those corresponding to series I, despite the vanadium content being the same, Figure 4 (8.6 and 4.1 m2 g-l, respectively, for samples T-1-973-2 and T-M-973-2). It seems that the method of vanadium incorporation (impregnation or melting) still has an effect on the sintering/agglomeration process taking place during melting of vanadia; if vanadium has been incorporated via impregnation, the specific sur-

Q

*

0

'l-\1.973 'l-I-Y73

IO

20

51

Figure 4. Variation in the specific surface area of samples prepared on rutile with the vanadium content (labels correspond to the number of monolayers of V,05). face area is ca. twice, even after heating at 973 K, the value for the sample where the vanadium has been incorporated by melting, having the same vanadium content (3.4%) and having been calcined at the same temperature (973 K). Acknowledgment. Thanks are given to CICYT (Grant MAT88-0556) and Junta de Castilla y Le6n for finantial support. Gift of the oxide samples is also acknowledged. Registry No. V,05, 1314-62-1; TiO,, 13463-67-7.

Ethylidyne on the Rh( 100) Surface: A Theoretical Investigation Birgit Schiplttt and Roald Hoffmann' Department of Chemistry and Materials Science Center, Cornel1 University, Ithaca, New York 14853-1301

Mohamed K. Awad and Alfred B. Anderson Chemistry Department, Case Western Reserve University Cleveland, Ohio 441 06 Received July 26, 1989. In Final Form: November 17, 1989 The bonding of ethylidyne (CCH,) on the Rh(100) surface is analyzed by using the extended Huckel tight-binding approach and also the ASED-MOtheory with cluster models. The relative stabilities for ethylidyne bound at the on-top, bridging, and the 4-fold hollow sites of Rh(100) are discussed and compared with the well-characterized geometry of ethylidyne in the %fold hollow site of the Rh(ll1) surface. The theoretical indicators of bonding support the experimental assignment of a 4-fold hollow site for CCH,. In this geometry, the e orbitals of ethylidyne interact most with the surface, resulting in a better rhodium-carbon bond and a stronger carbon-carbon bond. A comparison is made between the bonding of ethylidyne to discrete transition-metal fragments and to the Rh(100) and Rh(ll1) surfaces. A 4-fold site certainly has the appropriate orbitals to bind strongly a CR fragment. Carbyne ligands are well-known ligands in organometallic chemistry.' The first compounds of this class to be synthesized and characterized were XM(CO),(CR) (X Permanent address: Department of Chemistry, University of Arhus DK-8000 Arhus C, Denmark.

= C1, I, Br; M = Cr, Mo, W; R = CH,, C,H,), 1, made by Fischer and co-workers in 1973.2 (1) (a) Fiecher, E. 0.; Schubert, U. J. Organomet. Chem. 1976,100, 59. (b) Schrock, R. R. Acc. Chem. Res. 1988, 19,342. (c) Nugent, W. A.; Mayer, J. M.Metal-ligand Multiple Bonds; Wiley: New York, 1988.

0743-7463/90/2406-0806$02.50/00 1990 American Chemical Society