J. Phys. Chem. 1987, 91, 4519-4524
4519
may be the primary monomeric units involved in the formation of the hydrocarbon products. The mechanism proposed in Figure 6 is more flexible in this regard, as OCH, can be formed without the intermediate formation of OC. It has been customary to discuss the reactions of OC only in terms of its role as an intermediate in the formation of hydrocarbons. Recent studies” of the kinetics of the synthesis reaction over a cobalt catalyst suggest that the presence of carbide (or other CO-derived species) on the catalyst surface reduces the methane selectivity and increases the occurrence of chain growth. It is possible, therefore, that the presence of carbide may also modify the surface properties of the catalyst, particularly, as suggested for cobalt by the above results, with respect to its hydrogenation ability.22
(iii) Formation and Reaction of OCH,. In common with other proposed mechanismslOyllFigure 6 shows the species OCH, to be the only intermediate leading to the formation of higher hydrocarbons, at least under conditions where no oxygenates are formed. The species can be formed by three different routes, their relative contribution depending on the hydrogen and even the H 2 0 partial pressures. The average H/C ratio of the intermediate on the catalyst surface is also dependent on the H2 partial pressure, as the various possibilities are proposed to be in equilibrium with each other, and the species MH,.
(22) This comment applies particularly for iron catalysts for which the gradual formation of specific irqn carbide phases during synthesis have been well established. See, for example, ref 9c, p 153.
(23) The yield of H2 calculated for this reaction is a lower limit because the presumably constant amount of H2 adsorbed on the catalyst is not included.
Acknowledgment. We thank Steven Smith and Ann Tibbett for their assistance with the gas chromatographic analysis. Registry No. Co, 7440-48-4; CO,630-08-0.
Structure of Vanadium Oxide on Supports As Measured by the Benzaldehyde-Ammonia Titration Method Miki Niwa,* Yoshihito Matsuoka, and Yuichi Murakami Department of Synthetic Chemistry, Faculty of Engineering, Nagoya University, Furo-cho, Chikusa- ku, Nagoya 464, Japan (Received: September 3, 1986)
In order to clarify the structure of vanadium oxide on supports, the benzaldehyde-ammonia titration (BAT) method was applied to various supported vanadium oxide catalysts. Prior to the measurements, an infrared study of the adsorbed benzoate ion on Ti02, Zr02, and Ce02 was carried out to justify the BAT method to measure surface cus sites. The exposed surface area of the supports Al2O3, TiO,, and Zr02 with different crystal phases was then measured, and the surface area of vanadium oxide supported was calculated by the difference between BET and exposed surface areas. On the other hand, the surface area of vanadium oxide on S O 2was measured after reduction at 773 K, because benzaldehyde was adsorbed on the reduced V2O3 but not on S O 2 . Based on these measurements, the relationship between percent coverage on support and surface V 2 0 5 concentration was obtained. The structure of supported V 2 0 5thus determined depended on the kind of support, but not significantly on the crystal phase. Except in small concentrations on A1203and Si02,vanadium oxide formed a multilayer. The support surface of A1203(y)was covered most effectively, and the average thickness of V 2 0 5 in 100% of the coverage was 3 layers. To the contrary, the Si@ surface was not covered effectively, and the average thickness attained up to 50 layers. Furthermore, Z r 0 2 and Ti02 showed intermediate behavior between these supports. It is shown that the coverage efficiency and thickness of the formed metal oxides are correlated with the electronegativity of the cations of supports.
Introduction Metal oxides are often supported on various supports to be used as industrial catalysts. Supports may play an important roIe in making the oxides well dispersed or spread on the catalyst surface. In order to study the dispersion or the spreading of metal oxides, specific surface areas of metal oxides have to be measured. However, this problem has not been extensively studied so far. Adsorption of oxygen at low temperature is a method derived by Weller’ to measure the surface of Cr203and Moo3. Nitric oxide is also used to measure the exposed surface of molybdenum oxide s ~ p p o r t e d .These ~ methods were proposed to study the supported molybdenum oxide catalysts because of their industrial importance. On the other hand, the present authors3 used the NO-NH3 rectangular pulse (NARP) method to measure the number of active sites on V205. When these situations are compared with (1) Weller, S . W. Acc. Chem. Res. 1983, 16,101. (2) Millman, W. S.; Hall, W. K. J . Phys. Chem. 1979, 83, 427. (3) (a) Miyamoto, A.; Yamazaki, Y.; Inomata, M.; Murakami, Y. J . Phys. Chem. 1981,85,2366. (b) Inomata, M.; Mori, K.; Miyamoto, A,; Murakami, Y. J. Phys. Chem. 1983, 87, 761, 754. I
,
,
those in the field of supported metal catalysts, one can notice that investigations on the supported metal oxides have not progressed significantly. We have previously proposed4 a benzaldehyde-ammonia titration (BAT) method that is applicable to the measurement of the exposed surface area (called BAT surface area hereafter) of supports A1203and Ti02. The surface area of the supported oxide itself is then calculated as the difference between BET and BAT areas, and the extent of coverage of support surface by metal oxide is thus measured. This method which is similar to back-titration is effective on supported metal oxides, since the amount of oxide supported is large enough to cover the support surface completely or significantly. A similar method by the use of the selective adsorption of carbon dioxide on A1203support has been r e p ~ r t e d . ~ Because the surface density of the adsorbed benzoate species on A203 is ca.2 nm-2, it covers the Al2O3surface almost completely.6 Because of this condition, the adsorbed benzoate species is an (4) Niwa, M.; Inagaki, S.;Murakami, Y. J . Phys. Chem. 1985,89, 3869. (5) Segawa, K.; Hall, W. K. J. Coral. 1982, 77, 221. (6) Niwa, M.; Inagaki, S.; Murakami, Y. J . Phys. Chem. 1985,89,2550.
0 1 9 8 7 American Chemical Society
4520
The Journal of Physical Chemistry, Vol. 91, No. 17, 1987
TABLE I: BET Surface Area and Site Density of Supports and Vanadium Oxides
Niwa et al.
r
BET surface support
area/m2 g-’
site d e n s i t y / n d
CeO, AIzOj( a+6)
66 45 78 54 6.2 29 1 81 8.9
2.82 2.87 3.05 1.90 3.10 0.14 0.02 2.47 0.06
ZrO, TiO,(a-r)
Ti02(r) Si02(HS) SiO,(LS) v2°3p
VZOS
5.4
excellent probe for the measurement of surface area of exposed support, because we could neglect the problem of experimental errors due to the supporting of metal oxides on specific sites. As mentioned in the present study, the BAT method can be applied also to C e 0 2 and ZrOz supports. Although the method is not usable on S O 2 , discrimination between S O 2 and reduced V 2 0 3 surfaces is possible. Supported vanadium oxides are industrially important catalysts in various processes. It has been known that the vanadium oxides supported on A1203,TiOz, SO2,’ and Mg08 are active in various reactions. Not only the kind of support but also its crystal structure influences activity and selectivity in catalytic reactions. For example, the V205/A1203(6 or x ) calcined at 1273 K is selective in the ammoxidation of aromatic compound^.^ The V205 supported on T i 0 2 (usually anatase phase) is a famous catalyst for the NO, removal from the exhaust gaslo and the oxidation of o-xylene into phthalic What makes different supports available for these reactions is of extreme interest. The purpose of the present study is therefore to investigate how the vanadium oxide is supported on different supports. The structure of vanadium oxide on the supports will be estimated based on the measurement of the coverage by oxide. The chemical properties, i.e., reaction activity, exposure of active sites, etc., will be disregarded.
Experimental Method Catalysts. Seven kinds of oxides were used as supports for vanadium oxide. TiOz(a-r) was obtained commercially from Nippon Aerosil Co. Ltd. P-25, which was proven to be an anatase with rutile phase structure (the content of anatase was about 90%). This was prepared by decomposition of TiCl, at a high temperature and was free from major impurities. Ti02(r) (purity, 99.9%) was supplied by Ishihara Sangyo Co. Ltd. and consisted of the rutile phase. Z r 0 2 was prepared from a ZrC120 solution. To the solution was added ammonium hydroxide to deposit the hydroxide, which was then dried and calcined in air at 773 K. CeOZwas obtained by decomposition of Ce(N03)3 at 773 K in air. Two kinds of SiOzwith high and low surface areas were supplied from Fuji-Davison Chemical, and these were denoted by Si02(HS) and -(LS), respectively. AlZO3(a+O) was obtained by calcining A1203(7+y) (a reference catalyst JRC-ALO- 1 supplied from Catalysis Society of Japan) at 1273 K, which was proven to be a mixture of the a and 6 crystal phases. BET surface areas of these oxides are summarized in Table I. Supported catalysts were prepared by an impregnation method from an oxalic acid solution of ammonium metavanadate, followed (7) (a) Lischke, G.; Hanke, W.; Jerschkewitz, H.-G.; Ohlmann, G.J. Catal. 1985, 91, 54. (b) Narayana, M.; Narasimhan, C. S.; Kevan, L. J. Chem. Soc., Faraday Trans. 1 1985, 81, 137. (c) Vorlow, S.;Wainwright, M. S.; Trimm, D. L. Appl. Carol. 1985, 17, 87. (8) Shakhnovich, G. V.; Belomestnykh, I. P.; Nekrasov, N. V.; Kostyukovsky, M. M.; Kiperman, S. L. Appl. Caral. 1984, 12, 23. (9) Murakami, Y.; Ando, H.; Niwa, M. J. Card. 1981, 67, 472. (10) Inomata, M.; Miyamoto, A.; Ui, T.; Kobayashi, K.; Murakami, Y. Ind. Eng. Chem. Product Res. Deu. 1982, 21, 424. (1 1) (a) Bond, G. C.; Bruckman, K. Faraday Discuss. Chem. SOC.1981, 72, 235. (b) Bond, G.C.; Konig, P. J . Catal. 1982, 77, 309. (12) Wachs, I. E.; Saleh, Y.; Chan, S. S.; Chersich, C. Appl. Caral. 1985, 15, 339.
1
4000
3500
3000
do0
1600
1400
1200
1000 lc7-1)
Figure 1. IR spectra on Ti02(a-r): (a) background; (b) upon adsorption of benzaldehyde a t 523 K; (c) by reaction with ammonia at 673 K; (d) after reaction with oxygen at 673 K.
by calcination at 773 K in a stream of oxygen for 3 h. In a few experiments, ammonium hydroxide was used in place of oxalic acid to check the method of dissolving ammonium metavanadate. The content of the V z 0 5is shown in mole percent; for example, V2O5/TiO2(10 mol %) consisted of 10 mol % V2Os and 90 mol
5% Ti02. Measurements. A pulse experiment was performed using a conventional apparatus. A liquid nitrogen trap was installed in front of the reactor in order to remove impurity oxygen. Silicon DC 550 column held a t 413 K or operating at temperatures ranging from 403 to 453 K was used for the separation of liquid products. Benzaldehyde and ammonia were used without further purification. In the BAT method by the pulse technique,6 20-40 mg of catalyst was pretreated at 673 K in the oxygen stream for 1 h. One microliter of benzaldehyde was injected per pulse until benzaldehyde was not adsorbed, and 10 cm3 of ammonia was then injected to form benzonitrile, until benzonitrile was not formed. Usually, 6 to 7 pulses of benzaldehyde were needed to saturate the surface with the benzoate species, and 3 pulses of ammonia were required for the complete conversion into benzonitrile. Surface density of the benzoate adsorbed was calculated from the sum of the obtained benzonitrile and BET surface area. Infrared spectra obtained upon the adsorption of benzaldehyde and the subsequent reaction with ammonia or oxygen were recorded in situ on a JASCO IR-G spectrophotometer. X-ray photoelectron spectra (XPS) were measured on VzOs/SiOz by Vacuum Generator ESCA LAB-5 in order to determine the oxidation state of vanadium. Temperature-programmed reoxidation experiment on the reduced V205/Si02was coupled with XPS measurements to confirm its oxidation state.
Infrared Spectrum of Adsorbed Benzaldehyde. In Figures 1-3 are shown spectra obtained on TiO,(a-r), ZrO,, and Ce02, respectively, upon adsorption of benzaldehyde and subsequent reaction with ammonia or oxygen. Adsorption of benzaldehyde at 523 K on TiOz clearly revealed the absorption bands ascribable to the benzoate ion which was bidentately bound with the coordination unsaturated (cus) site on the surface,6 i.e., Ph-Cs
M
The Journal of Physical Chemistry, Vol. 91, No. 17, 1987 4521
Structure of Vanadium Oxide on Supports
a E W
a W U
a U E
VI 3
V205 CONTENT I n o l % l
Figure 4. Dependence of BET (0)and BAT ( 0 )surface areas on the
V 2 0 5 loading on Ti02(a-r).
t.
4000
. . .
I
I
I
I
e
j
I
,
,
,
In
1
fkO0
3000
3500
)
1
%
I
1400
1600
'
1
1200
1
1
1000
("1)
Figure 2. IR spectra on Zr02: (a) background after a run of experiment, see text; (b) upon adsorption of benzaldehyde at 523 K; (c) after reaction with ammonia at 673 K.
L 4000
I
,
,
l
3500
~
>
,
II
3000
l
I
I
I
In
f¶OO
,
I
1600
I
1
1400
I
I
1200
I
I
1
1000
(cm-1)
Figure 3. IR spectra on Ce02: (a) background; (b) upon adsorption of benzaldehyde at 523 K; (c) after reaction with ammonia at 673 K.
It included absorptions ascribable to C=C (1590,1485, 1440 cm-I), C-H in the benzene ring (3050cm-I), and carboxylate (the asymmetric and symmetric stretching vibrations at 1500 and 1405 cm-I, respectively). Additionally, small shoulder peaks were observed at 1640 and 2900 cm-' which could be assigned to moncdentate benzoate ion and benzylalkoxide, respectively. When compared with on Al2O3(v T ) , ~absorptions of the carboxylate shifted to lower wavenumbers by 40-50 cm-I, while those of the benzene ring remained unchanged. The absorption bands of the benzoate ion were removed gradually upon reaction with ammonia at 673 K. The sDectrum thus obtained however still included small absorptions at 1600-1400cm-l. These residual absorptions seemed to be assigned to carbonate species since these diminished upon reaction with oxygen at 673 K for 30 min. Similar results were obtained also on ZrO, and C e 0 2 (Figures 2 and 3). However, spectra were contaminated by the presence of the carbonate included in the background. Of interest was that
+
the carbonate ion on Z r 0 2 at 1530 and 1440 cm-I was removed completely by the reaction with ammonia. The background spectrum free from the carbonate ion in Figure 2a was obtained by the experimental run of adsorption of benzaldehyde and subsequent reaction with ammonia. The exposure of benzaldehyde on the surface revealed the benzoate ion which could be removed by ammonia, as shown in Figure 2. On the other hand, the carbonate ion on the surface of C e 0 2 seemed to be strongly adsorbed, since successive admissions of ammonia (3 times) could not obtain an uncontaminated surface which revealed carboxylate bands broader than in the initial state. The strongly held carbonate ion on the C e 0 2 may be caused by its strong basic property.I3 It was concluded from the above experimental results that the adsorption of benzaldehyde yielded the benzoate ion, and the reaction with ammonia converted the adsorbed benzoate, some of carbonate species remaining unconverted on T i 0 2 and Ce02. Benzaldehyde-Ammonia Titration on Support. Saturated surface density of the benzoate ion was measured on supports, since the value was used to calculate the exposed surface area of support as shown below. In order to discriminate between surfaces of vanadium oxide and of supports, vanadium oxide is required to be kept oxidized, since the benzoate is not adsorbed on the V205 but on the reduced state V203.4 The surface of the oxide must be dehydroxylated, furthermore, because the cus site available for the adsorption of benzoate ion is believed to be formed upon dehydr~xylation.'~Based on these considerations, oxides were pretreated in dried oxygen for 1 h at 673 K. When treated in a flow of helium in place of oxygen, the surface density was decreased by 8 to 13% on T i 0 2 and ZrO,. Influences on the site density caused by various procedures were of interest to identify the cus site, but this was not sought in the present study. Site densities thus obtained are shown in Table I. These were fairly large on Ce02, A1203(a+B), ZrO,, and Ti02. One benzoate ion owned about 0.3 nm2 of the surface area, and this was close to its cross surface area.15 The high density therefore suggests that the surface is nearly satisfied with the benzoate ion. On the other hand, the density on S O 2 was very small. As discussed in a previous study on A1203,6 benzaldehyde is adsorbed due to a scheme similar to the Cannizzaro reaction. Therefore, benzaldehyde should be adsorbed accompanied by the formation of byproducts. Benzyl alcohol was observed upon adsorption on CeO, and Z r 0 2 at 523 K. However, no hydrocarbon byproduct was found on T i 0 2 and Al2O3(a+O). Molecular hydrogen might be formed on these oxides, as shown in the less acidic alumina.6 Measurements of the Exposed Surface Area of the Supports. Based on experimental results shown above, the BAT surface area, Le., exposed surface area of supports, was measured by the BAT method. This was calculated from the equation BAT(exposed (13) Niwa, M.; Furukawa, Y.; Murakami, Y. J . Colloid Interface Sci. 1982, 86, 260. (14) The formation of cus sites is explicit on alumina; see: Knozinger, H.; Ratnasamy; Rev,-Sci, Eng, 19,8, 31, (1 5) The molecular size of benzoate ion is regarded as being the same as that of benzene.
4522
Niwa et al.
The Journal of Physical Chemistry, Vol. 91, No. 17, 1987
-4 I
a
I '
a
0
u B
V2D5 CONTENT
10
Figure 5. Dependence of BET (0)and BAT ( 0 )surface areas on the V 2 0 5 loading on TiOz(r).
20
38
40
50
V2O5 C O N T E N T l m o l % l
[mol%)
Figure 8. Dependence of BET (0)and BAT ( 0 )surface areas on the V , 0 5 loading on CeOZ; on the sample prepared by using ammo-
am
nium hydroxide.
V2D5 C O N T E N T
lmol%l
Figure 6. Dependence of BET (0)and BAT ( 0 )surface areas on the VzOs loading on AI2O3(a+B).
I 01
N
E
a w LL
a
V205 C O N T E N T ( m o l % l
Figure 7. Dependence of BET (0)and BAT ( 0 )surface areas on the V , 0 5 loading on Zr02.
surface) area of support = amount of benzoate obtained &')/site density on support (m-'). The BAT and BET surface areas on vanadium oxides supported on TiO,(a-r), Ti02(r), Al,03(~+O), ZrO,, and CeOz are shown in Figures 4-8, respectively. The BAT surface area of the TiO,(a-r) support decreased by supporting vanadium oxide more readily than its total surface area obtained by the BET method (Figure 4). The BAT surface area became less than 10% of the total surface area at 13 mol % loading. The support surface was unexposed in about 15 mol % loading. In other words, the surface of TiO,(a-r) was covered with vanadium oxide under these conditions. In the case of V,OS/TiO2(r). the BAT surface area decreased significantly by increasing the content of V,Os, although the total surface area was kept almost constant. Similarly, the decrease in the BAT surface area was observed on the V2Os/Al2O3(a+O)(Figure 6). The abrupt increase of BET surface area in the low loading was not understood exactly. On the other hand, the BAT surface area of ZrO, did not become negligibly small even at 25-50 mol % loading, although it decreased gradually up to 10 mol % (Figure 7). On CeO,, shown in Figure 8, the support area derived from the above equation was in good agreement with the BET surface area. However, the
V2O5/Ce0, (more than 5 mol %) catalysts revealed clearly the crystal phase of CeV04. Such a spinel-type compound consisting of vanadium and support was not observed in any other catalysts. It was therefore considered that the benzoate species was adsorbed on the surface of CeV04 in almost the same surface density as on CeO,, and it was impossible to measure the surface area of exposed CeO, individually. The sample prepared by impregnation with a basic solution of ammonium hydroxide in place of oxalic acid also showed the consistency between the BAT and BET surface areas (Figure 8). Therefore, the formation of the spinel-type compound was not caused by the method of impregnation. Discrimination between SiO, and V,03 Surfaces. The BAT method was not applied to the SiO, support, since the benzoate was hardly adsorbed thereon. However, selective adsorption could be possible on the reduced V203/Si02,because the benzoate was adsorbed on V,03 but not on S O , . Two kinds of silica which possessed different BET surface areas were used to identify the influence of the pore volume on the supported condition. Experimental conditions for the reduction of the V,0s/Si02 were checked by the temperature-programmed reoxidation (TPRO). It was found that the V,Os supported on Si02 was reduced completely to V,O, at 873 K, but only to V,03 6-V,03 at 773 K, although unsupported V20s was reduced V , 0 3 at 773 K . The oxidation state as measured by the TPRO was the averaged one of vanadium cations in the surface as well as in the bulk, and the higher oxidation state on the V,0s/Si02 was considered to be caused by the low reducibility of vanadium oxide immersed in the bulk due to some effects by SiO, support. Different oxidizability caused by interaction with the support was observed also in the TPRO experiment, since the reduced vanadium oxide supported on SiO, was reoxidized at a lou7er temperature than unsupported V,03. XPS of the catalyst reduced at 773 K was then measured to identify the oxidation state of vanadium at the subsurface. It was found that the binding energies of V 2p3/, on the V,O,/SiO, (25 and 10 mol %) were 515.6 and 515.2 eV,I6 respectively, and these were almost in agreement with the reported value about the V3+ cation (515.5 eV).11317 In the present study, therefore, these catalysts were reduced by hydrogen at 773 and 873 K and subjected to the BAT measurements. Prior to the measurement, the reduced catalyst was reduced at 773 K in situ in the pulse experiment. The surface area of vanadium oxide thus measured was shown in Figure 9. Because the adsorption of benzoate on the SiO,(HS) was not neglected completely, this was corrected by assuming a linear combination of numbers of the benzoate on SiO, and on "203.
(16) Binding energy was determined as compared to CIS(184.6 eV) or Sizs (154.0 eV) for V205/Si02 (25 and 10 mol %), respectively. In the latter case, the CISpeak was not used, since this was abnormally shifted to higher binding energy probably because of electric charge of the sample. (17) Colton, R. J.; Guzman, A. M.; Rabalais, J. W. J . Appl. Phys. 1978, 49, 409.
The Journal of Physical Chemistry, Vol. 91, No. 17, 1987 4523
Structure of Vanadium Oxide on Supports
300
100
7 lBO
,
I i i'
JO'
,
a K W
a
a
8a U
K 3 Lo
W U
100
a
A-A
U K 3
0 k
VRNRDIUM CONTENT l n o l % l
Figure 9. Dependence of BET (0, A) and V203 (0,A) surface areas on the vanadium loading on SiO,(HS) ( 0 , O ) and SiO,(LS) (A, A); @,* reduced at 873 K.
The surface area of vanadium oxide supported on Si02increased at small loading and kept almost constant at more than 1 mol % loading. The difference of the reduction temperatures, 773 and 873 K, did not have influence on the behavior, which justified the conditions for the reduction of vanadium oxide. The difference between areas of vanadium oxide and of total catalyst should correspond to the exposed surface area of Si02support. It seemed that a difference was observed in the behaviors on two kinds of silica supports; however, this was caused only by the difference in surface areas of silicas. as shown below.
Discussion The substantial difference in structure between supported metal and metal oxide is based on the nature of these loaded materials. Metal and support may not be regarded as being strongly interacting because of different properties. Current attention to strong metal-support interaction (SMSI)IS has made one to reconsider this problem in detail. On the other hand, in supported metal oxides, oxides are loaded on supports (Le., oxides) with a similar property. More or less, oxides interact with supports through metal-oxygen bonds, and it has not been usually accepted that a single tetahedron or octahedron of oxide is completely isolated on supports. In addition to this situation, vanadium oxide used in the present study possesses a layer structure. Therefore, the dispersion of vanadium oxide is not appropriate in showing the structure of supported catalysts. Rather, the extent of monolayer covering or the thickness of oxide is preferable. Bond et al." and Gellings et al.I9 reported the unique activity of monolayer V 2 0 5on T i 0 2 which was prepared by the vapor phase supporting method or by the ion-exchange method. Dependence of the activity and selectivity in oxidation reactions on the thickness of vanadium oxide layer has been discussed by us.2o However, the present study has mainly dealt with the exposed surface area of supports. Therefore, the percent coverage on support surface by metal oxides will be used to indicate the structure of supported metal oxides. The coverage should be defined as coverage (%) = (surface area of vanadium oxide supported)/(BET surface area of the supported catalyst) X 100. As has been pointed out by Weller et al.,21metal oxides block the pores to decrease the surface area of supports. Therefore, the surface area of the support itself cannot be used to show the extent of coverage of the support surface by metal oxides. In addition, the coverage has to be compared in the same (18) (a) Tauster, S. J.; Fung, S. C.; Garten, R. L. J. Am. Chem. SOC.1978, 100, 170. (b) Vannice, M. A.; Garten, R. L. J . Catal. 1979, 56, 236. (19) (a) Roozeboom, F.; Mittelmeijer-Hazelger, M. C.; Moulljn, J. A,; de Beer, V. H. J.; Gellings, P. J. J . Phys. Chem. 1984, 84, 2783. (b) van Hengstum, A. J.; van Ommen, J. G.; Bosch, H.; Gellings, P. J. Appl. Catal. 1983, 5, 207. (20) Miyamoto, A.; Mori, K.; Inomata, M.; Murakami, Y. Proceedings of8th International Congress on Catalysis; Verlag Chemie: Weinheim, 1984; Vol. IV, p 285. (21) Garcia Fierro, J. L.; Mendiroz, S.; Pajares, J. A,; Weller, S. W. J. Catal. 1980, 65, 263.
1 10
I
I
20
30
I
V R N R D I U H SURF. CONC. lnm-21 Figure 10. Correlation between percent coverage by vanadium oxide and surface vanadium concentration on Al@3(7) (A), AI203(cu+B) (A), ZrO, (V),Ti02(r) (e),Ti02(a-r) (O), SiO,(HS) (a),and SiO,(LS) (m). The surface concentration of vanadium oxide on SiO, was calculated as the composition V203,6(SiO,(HS)) or V203.4(SiO,(LS)). The dotted line indicates the theoretical relationship in the monolayer condition.
TABLE 11: Formation of the Multilayer Vanadium Oxide coverage
av thickness in the coverage of
support
concn range/nm-2
efficiencyo
50%
100%
A1203(7) A1203(a+B)
>2 >1 almost all almost all almost all
1-'/4
1.5 2.5 2.5 2.5 4.5 25 25
3 13
Zr02 Ti02(r) Ti02(a-r) S O 2 (HS) SiO,(LS)
>OS
1-'/6 '/2-1/,0
'/3-]/9 1/4-1/8
]-'I50
1-'/50
b 15 6 50 50
"Calculated on the basis of the owned surace area of a vanadium species (10.6 X lo4 pm2). The unit value on A1203and SiO, shows the formation of the doping vanadium oxide, see text. *The surface of ZrO, was not covered completely.
surface concentration of vanadium oxide, because various supports with different surface areas were used. Percent coverages by vanadium oxide on supports including A1203(7) previously reported6 are plotted against the surface concentration of vanadium oxide in Figure 10. The coverage increased with increasing concentration of vanadium supported. The dotted line in Figure 10 shows the theoretical relation between them, vanadium oxide being supported in the monolayer.22 In small concentrations of vanadium oxide, the monolayer condition seems to be satisfied on A1203 and S O 2 . However, one cannot with certainty regard the structure as monolayer only from this finding, because isolated species of the vanadium oxide are known on A1203and Si02in these small concentrations. Inactive dissolved vanadium species on A1203was reported by The isolated tetrahedral vanadium oxide on S O 2 was reported to be a photocatalytically active species.23 Therefore, it is correct to conclude that the monolayer or the isolated vanadium species may be formed on A1203and Si02in the small concentration. On the other hand, in concentrations of more than 2 nm-2, all plots deviate from the theoretical relation of the monolayer structure. This deviation indicates clearly that the vanadium oxide is supported in the multilayer structure. We could estimate parameters about the formation of multilayer: (1) In what concentration is the multilayer formed (concentration range of multilayer supporting)? (2) In the multilayer condition, what extent of vanadium oxide covers effectively the exposed surface o f support (coverage efficiency)? (3) What thickness of metal oxide is formed in the half or complete covering of the support (22) In this paper, monolayer is defined as one sheet of V 2 0 slayers supported on oxides and discriminated from the situation where the isolated vanadium species corn letely cover the oxide support surface. One species of V 0 2 owns 10.6 X 10 pm2 of the surface area, which is calculated from the structure of V,05. (23) (a) Anpo, M.; Takahashi, I.; Kubokawa, Y. J . Phys. Chem. 1980,84, 3440. (b) Yoshida, S.; Tanaka, T.; Okada, M.; Funabiki, T. J . Chem. Soc., Faraday Trans. 1 1986, 81, 1 5 1 3 .
B
J. Phys. Chem. 1987, 91, 4524-4527
4524 TABLE 111: Parameters of Oxides oxide electronegativity A1203 10.5 ZrO, 12.6 TiO, 13.5 SiO, 16.2 V2OS 17.6
ZPC 5-9 4-1 1 4.7-6.2 1.8-2.2 (1.0-2.5)
surface (average thickness in the coverage of 50 or loo%)? These estimated parameters are shown in Table 11. On both A1203(7)and Al,O,(cu+O), V 2 0 j forms the multilayer in more than 2 to 1 nm-, of the surface concentration, and most effectively it covers the support surface. One-fourth to one-sixth of supported V 2 0 5covers the exposed support surface, and the remaining is piled on the supported V205. To the contrary, the surface of S O z is not covered effectively in more than 0.5 nm-, of the surface concentration. Only one-fiftieth of V 2 0 j contacts directly with the silica surface. V205 on ZrO,, Ti02(r), and TiO,(a-r) have intermediate behavior between them, and the multilayer is formed in almost all concentrations. Average thicknesses in the coverage of 50 and 100% are also indicative of the structures of supported vanadium oxides. It is remarkable that these parameters depend on the kind of support, but not significantly on the crystal structure. Two kinds of silica of different surface areas show the completely same parameters. The formation of the multilayer vanadium oxide should be correlated with the properties of oxides. Among various parameters applicable to the explanation of these behaviors, the electronegativity of cations in oxidesz4and zero point charge (ZPC),j may be considered. As shown in Table 111, these parameters of oxides which have been utilized to indicate the acidic and basic properties of oxides26are partially correlated each other but are somewhat different. Coverage efficiency and average thickness are correlated with the electronegativity of cations. The larger the electronegativity of support, the thicker the multilayer of supported oxide. In other words, the thickness of vanadium oxide (24) Tanaka, K.; Ozaki, A. J . Catal. 1967, 8, 1. (25) Parks, G. A. Chem. Reu. 1965, 177. ( 2 6 ) Tanabe, K. In Catalysis; Anderson, J. R . , Boudart, M., Eds.; Spring-Verlag: Berlin, 1981; Vol. 2, p 231.
becomes higher on the more acidic support. As properties of support and supported oxide become similar, the thicker oxide is supported, because vanadium oxide is the most acidic among them. Previous studies on supported molybdenum oxides,' have shown that molybdenum oxide is dispersed on A1203more than on SO,, and the monolayer is formed on Alz03in the small concentration of less than 15 wt %. Because this trend is similar to the present finding that Vz05is supported on AI203 in thinner layer than on S O 2 ,molybdenum oxide may be supported similarly to vanadium oxide. Because the method of preparation is reported to have an influence on the catalyst activity and selectivity, different structures may be anticipated on these catalyst systems. Furthermore, the dependence of the activity on the structure should be investigated. These studies are now in progress. Characteristics of the BAT method for the determination of exposed support surface are clarified in this investigation. This can be indicated as compared with COz-probe method: (1) The BAT method can be applied to supports AlZO3,TiO,, and ZrO,, whereas COz probe is applied to Al,03 only. (2) The surface density of the adsorbed benzoate ion is high, 2 to 3 nm-,; the support surface is filled with this species. This condition is different from the adsorption of C 0 2 whose surface density is less than 0.75 nm-2.4 In other words, specific adsorption sites are required for CO,, while not for benzoate ion. Therefore, experimental errors for the C02-probe method will be estimated when metal oxides are supported on the specific surface. ( 3 ) Carbonate ion is sometimes found on metal oxides as a contaminant, as shown above. The degree of evacuation therefore influences sensitively the adsorption of CO,. On the other hand, it is easy to obtain reproducible results for the benzoate adsorption. Acknowledgment. This work was supported by a Grant-in-Aid from the Ministry of Education, Science and Culture, Japan (No. 59470097). Registry No. V 2 0 5 , 1314-62-1; TiO,, 13463-67-7; Zr02, 1314-23-4; Ce02, 1306-38-3; benzaldehyde, 100-52-7; ammonia, 7664-41-7. (27) (a) Muralidhar, G.; Concha, B. E.; Bartholomew, G. L.; Bartholomew, C. H. J . Catal. 1984, 274. (b) Okamoto, Y . ;Imanaka, T.; Teranishi, S. J . Phys. Chem. 1981, 85, 3798.
Reaction of N-Alkyl-2-bromopyridinlum Ion in Micelles of Cetyltrimethylammonlum Hydroxide Hamad A. Al-Lohedan Department of Chemistry, King Saud University, Riyadh-1 1451, Saudi Arabia (Received: September 16, 1986)
Cationic micelles of cetyltrimethylammonium ion catalyze the reaction of OH- with N-alkyl-2-bromopyridiniumbromide (alkyl = n-C12H25,n-C14H29,n-C1,5H33). The variation of the first-order rate constant k, with [CTACl] can be fitted to the pseudophase ion-exchange model, but this model fails when the counterion of the surfactant is OH-, because the rate constants do not become constant when substrates are fully micellar bound and increase on addition of NaOH. However, with the more hydrophobic substrate (alkyl = n-CI6H3,)it appears to reach a limiting rate constant at high [CTAOH] or high [NaOH]. These observations can be explained on the assumption that the [OH-] in the micellar pseudophase increases with increasing [OH-] in the aqueous pseudophase; Le., it can be fitted to a mass-action model. Despite the apparent differences between these two models, they predict similar values for the second-order rate constants of reaction of a given anion in the two types of micelles.
Hydroxide ion reacts with N-alkyl-2-bromopyridiniumions (1) Cationic micelles of acetyltrito produce pyridone (2)'.
methylammonium chloride catalyze these reactions, and the effects increase markedly with increasing length of the n-alkyl group of
0022-3654/87/2091-4524$01.50/00 1987 American Chemical Society