J . Phys. Chem. 1985,89, 3869-3872 characterize the system at the steady state. With monochromatic light (A = 380 nm and a0= 1015cmW2 s-l) and for high &, both OH. and H202accumulate at the semiconductor surface, reaching concentrations of the order of l O I 3 and 1014 cm-2, respectively. At darkness, the lifetime of s), while H202 photogenerated OH,. radicals is very short remains stably bound to the T i 0 2 surface (probably as H02- at basic pH). Although this kinetic model is successful in interpreting the experimental results, the existence of parallel mechanisms affecting the photocurrent transient behavior cannot be excluded. Very recently, Anderman and Kennedy25assumed that local pH changes (25) Anderman, M.; Kennedy, J. H. J. Electrochem. SOC.1984,131,21.
3869
at the semiconductor-electrolyte interface, due to the depletion of physisorbed OH- ions, may be responsible for the photocurrent transients observed with a-Fe203 electrodes. New experiments are in progress in our laboratory in order to evaluate the real influence of this mechanism on the transient phenomena observed with Ti02. Acknowledgment. This work was partially supported by the CAICYT (Spain). Registry No. H20, 7732-18-5; TiOz, 13463-67-7; 02,7782-44-7; HzOz, 7722-84-1; OH., 3352-57-6; Na2S04,7757-82-6. (26) 'Handbook of Chemistry and Physics", 64th ed.; CRC Press: Boca Raton, FL, 1984.
Measurement of Exposed Surface Area of Supports on Supported Metal Oxide Catalysts Miki Niwa,* Shinji Inagaki, and Yuichi Murakami Department of Synthetic Chemistry, Faculty of Engineering, Nagoya University, Furo-cho, Chikusa- ku, Nagoya 464, Japan (Received: March 5, 1985)
A method for the determination of the exposed surface area of the support in supported metal oxide catalysts has been first proposed for the case of the supported vanadium oxide catalyst. The proposed method is based on the selective adsorption of benzoate species on the support and its high surface density. Experimental conditions for discriminating between surfaces of the support and the metal oxide were determined. The benzoate species was adsorbed at 523 K only on alumina and titania supports, but not on V205at all, which enabled us to calculate the exposed surface area of the support from the amount of the benzoate adsorbed and the site density on the support. On the other hand, the benzoate was adsorbed on reduced VzOl nonselectively as well as on A1203. However, it was oxidized into carbon dioxide at 623 K only on V 2 0 3 ,which was also available for determining the exposed alumina surface. The surface condition of supported vanadium oxides which was expected from the exposed surface area was in good agreement with that proposed previously. The agreement justified the present method for measuring the exposed surface area on supported metal oxide catalysts.
Introduction Metal oxides are supported on suitable solids as industrial catalysts in many cases. The surface of support is therefore either covered with metal oxide supported or exposed on the surface. However, it is difficult to discriminate between them, and no method for the purpose is available so far. A great advantage is anticipated if one could measure surface areas individually. It would provide us information about the dispersion of supported metal oxide and then enable us to calculate the turnover frequency of a catalytic reaction. This serious drawback in the investigation of supported metal oxide catalysts is easily understood if one compares the situation with the study of supported metal catalysts, because the latter study utilizes a selective adsorption of carbon monoxide or hydrogen on metals to measure their surface areas. A significant assumption upon which the method is based is the selective adsorption on metals irrespective of the structure or other physicochemical properties. It is, however, a problem for the quantitative measurement that carbon monoxide is adsorbed on the metal surface differently as linear, bridged, or twin species. Current studies' about the SMSI have also indicated a problem in connection with the adsorption experiment. In spite of these problems, the adsorption experiment is the most reliable and practical chemical method. On the other hand, there have been relatively few studies of oxide catalysts to date. A recent paper by Weller2 reviewed his own study of chromia and molybdenum oxide by utilizing oxygen chemisorption. In general, versatile probes have not been known for the oxide catalysts. (1) (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. (2) Weller, S . W. Arc. Chem. Res. 1983, 16, 101.
Knozinger3 proposed some criteria about the probes used in the characterization of metal oxide catalysts. These could be summarized as thermal stability, quantitative detectability, and site specificity. The benzoate species which the authors have studied on alumina4 satisfies these requirements. It is stabilized completely even at 673 K. Quantitative measurements can be performed precisely, since all the species are converted into benzonitrile by the reaction with ammonia. The site density on A1203is high, and all the exposed basic A 1 4 sites of alumina are available for the adsorption. These properties of the benzoate species are therefore adequate for the probe to measure the surface area of the alumina support. Characteristics of metal oxides which are catalytically important and which differentiate them from metals are acid-base propertiesS and oxidation activity. Generally speaking, supported oxides are eminently active for oxidation reactions, while supports are relatively inactive. It has been known that the oxidation activity is basically related to the metal-oxygen bond cleavage, and explained by the parameter heat of formation of metal oxide: On the other hand, metal oxides are classified experimentally into acid, base, and amphoteric catalysts. Therefore, acid-base properties or oxidation activity could be utilized for the differentiation. As previously reported,' the benzoate species is adsorbed stably on TiOzand Z r 0 2 as well as on A1203. It was found in a recent (3) Knozinger, H. In "Advances in Catalysis"; Eley, D. D., Pines, H., Weisz, P. B., Eds.; Academic Press: New York 1976; Vol. 25, p 184. (4) Niwa, M.; Inagaki, S.; Murakami, Y. J . Phys. Chem. 1985,89,2550. ( 5 ) (a) Tanabe, K. "Solid Acids and Bases"; Kodansha-Academic Press: Tokyo, 1970. (b) Ai, M. "Proceedings of the 7th International Congress Catalysis"; Tanabe, K., Seiyama, T., Eds.; Kodansha: Tokyo, 1981; p 1061. (6) Moro-oka, Y.; Ozaki, A. J . Catal. 1966, 5, 116. (7) Niwa, M.; Sago, M.; Ando, H.; Murakami, Y.J. Cafal. 1981,69,69.
0022-3654/85/2089-3869$01.50/00 1985 American Chemical Society
3870 The Journal of Physical Chemistry, Vol. 89, No. 18, 1985
Niwa et al.
.-
experiment* that this species was stabilized also on such basic oxides as CeOz and MgO, and the reaction with ammonia yielded benzonitrile rapidly. These amphoteric or basic metal oxides are used as typical supports. On the basis of these considerations, one could use the benzoate adsorbed species to measure the exposed surface area of support, if experimental conditions are selected to discriminate between the surfaces depending on the acid-base property or on the oxidation activity. Surface areas of supported metal oxide could be calculated as the difference between BET and exposed surface areas. As an example of this, vanadium oxide is selected as an active oxide, and supported on alumina and titania, because the active site of vanadium oxide can be measured by NO-NH3 reaction in the rectangular pulse method.9 The purpose of this study is then to first apply the benzaldehyde ammonia titration (BAT) method to the measurement of exposed surface area of alumina and titania, and to reveal the usefulness for the characterization. Experimental Method
V,05 was prepared by a thermal decomposition of ammonium metavanadate in a stream of oxygen at 773 K for 3 h. This was reduced by hydrogen at 773 K to obtain reduced oxides. VzO4 was obtained in a static system, and the oxidation state was determined gravimetrically in situ with a quartz microbalance. On the other hand, the oxidation state V203,16was determined by the amount of consumed oxygen in a temperature-programmed reoxidation (TPRO) experiments.I0 Supported catalysts were prepared by an impregnation method from an oxalic acid solution of ammonium metavanadate, followed by calcination at 773 K in a stream of oxygen for 2 h. A1203and T i 0 2 were obtained commercially from Sumitomo Chemical and Nippon Aerosil Co. Ltd., and these were proven to be y-phase and anatase-rutile phase structures, respectively. 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 a t 413 K and Porapak Q column at room temperature were used for the separation of liquid and gaseous products, respectively. Benzaldehyde and ammonia were used without further purification. In the BAT method by the pulse technique: 20 mg of catalyst h. Two was pretreated at 773 K in the oxygen stream for microliters of benzaldehyde were injected per pulse until benzaldehyde was not adsorbed, and 10 mL of ammonia was then injected to form benzonitrile, until benzonitrile was not formed. Usually, 6-7 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 obtained benzonitrile and BET surface area. Measurements of the number of V=O sites on the (010) face of V205 have been performed by the rectangular pulse method as described e l ~ e w h e r e . ~ Infrared spectra obtained upon the adsorption of benzaldehyde and the subsequent combustion were recorded in situ on a JASCO IR-G spectrophotometer. Results
Benzaldehyde-Ammonia Reaction on Constituent Oxides. In order to know the activity for the benzoate adsorption on constituent oxides, vanadium oxide and supports were used individually in this reaction. The temperature for the benzoate adsorption was varied with the subsequent ammonia reaction temperature kept constant at 673 K, since this temperature was required for complete conversion of the adsorbed species into ben~onitrile.~ As shown in Figure 1, the amount of benzonitrile obtained on V205 (8) Present authors, unpublished result. (9) (a) Miyamoto, A.; Yamazaki, Y.;Inomata, M.; Murahmi, Y. J. Phys. Ch” 1981,85,2366. (b) Inomata, M.; Mori, K.; Miyamoto, A.; Murakami, Y. Ibid. 1983, 87, 761, 154. (IO) Niwa, M.; Murakami, Y. J . Catal. 1982, 76,9.
I
EXPERIMENTRL RUN (TIME1 Figure 1. Activity of V205for benzaldehyde and ammonia reaction on using repeatedly at 673 K (0)and 523 K (A).
EA
I
I
I
558
800
650
ADSORPTION TEMP
7
/ K
0
2 N I S
-
C
\
BA
ADSORPTION TEMP
/
K
Figure 2. Dependence of site densities of benzoate (0)and toluene (A) on benzaldehyde adsorption temperature on A1203(a, top) and T i 0 2 (b, bottom).
increased gradually by repeated use at 673 K, while little activity was observed at 523 K for several runs. Because an excess amount of benzaldehyde was injected in order to saturate the surface with the species, the vanadium oxide should be reduced at 673 K. The gradual increase of the benzonitrile yield, therefore, suggested the dependence of the adsorption ability on the oxidation state of vanadium oxide. The amounts of benzonitrile obtained on vanadium oxides with different oxidation states were then measured; in these experiments the V z 0 5 was reduced by hydrogen in order to obtain the adjusted oxidation state. It was found that reduced vanadium oxide such as V203,16possessed a high surface density for adsorption of benzoate in compared with V204and V2OS(Table I).
Exposed Surface Area of Supports
The Journal of Physical Chemistry, Vol. 89, No. 18, 1985 3871
TABLE I: Surface Density of Benzoate Adsorbed on Vanadium Oxides with Different Oxidation States vanadium benzoate BET surf. benzoate
oxide
amount/ IOl9 g-'
area/m2 g-'
density/nm-2
2.89 0.14 0.03
13.9 8.4 5.4
2.08 0.17 0.06
v2°3.16 v204 v205
On the other hand, the activity of A1203was stable and unchanged by repeated use. Dependence of the site density on adsorption temperature was measured (Figure 2a), in which the reaction temperature with ammonia was kept a constant at 673 K. Density of sites was independent of the benzaldehyde adsorption temperature in the temperature range from 523 to 673 K. It was therefore found that the benzoate was adsorbed sufficiently at 523 K. Toluene was formed simultaneously with the adsorption of benzoate. A mechanism of the toluene formation on A1203was discussed in a previous paper.4 Almost the same conclusion could be drawn on TiOZ,because a similar temperature dependence was found as in Figure 2b. Measurement of Exposed Surface Area of A1203and T i 4 by Selective Adsorption of Benzoate. Because the benzoate is adsorbed at 523 K only on the support surface, the decreased amount of the benzoate adsorbed should be caused by the decrease of the exposed surface area of the support. Therefore, the exposed surface area can be calculated from the site density on the support and the decreased value on the supported catalyst, Le., surface area of exposed support = amount of benzoate obtained (g-')/site density on support (m-z) This method was first applied to the V205/A1203catalyst with different loadings. Benzaldehyde was adsorbed at 523 K, followed by elevating the temperature to 673 K to form benzonitrile upon subsequent ammonia injection. The exposed surface areas of support thus obtained are shown in Table 11, in which BET surface areas also are included as a comparison. In the case of VZO5/ Al2O3,the exposed surface area of alumina at 1 mol % loading was the same as the BET surface area, and decreased gradually with the loading up to 25 mol %. These were very small at more than 25 mol % loadings. BET surface area decreased with the content of V2O5 due to the loss of porosity of support. The surface area contributed by supported VzO5 was calculated as the difference between those of BET and the exposed alumina. Furthermore, the dispersion of V z 0 5 was given on the basis of the area of a crystallized VO2,,. The dispersion thus calculated was very high a t lower loadings. This indicates that the vanadium oxide can be dispersed on alumina in an idealized condition of a monolayer at about 5 mol % loading. During the elevation of catalyst bed temperature after the adsorption of benzoate, a small amount of C 0 2 was eluted. Amounts eluted were measured by collecting them at liquid nitrogen temperature, and shown in Table 111. The path of the carbon dioxide formation will be discussed below. Surface area of the (010) face on V2O5 was then measured by a rectangular pulse method in order to be compared with the total surface area of Vz05.9 As shown in Table 11, the surface of the (010) face of V z 0 5 had the maximum value against the VzO5
content. The degree of exposure of the (010) face on V205 was 55% at more than 25 mol % loadings, which was in approximate agreement with 50% of unsupported VzO5. The exposed surface area of Ti02 on which a monolayer of Vz05 was supported was then measured. Bleaching of the supported catalyst by ammoniacal solution removed easily the accumulated layer of vanadium oxide, and formed a monolayer of Vz05on TiOz.9 In this condition, the exposure of the (010) face is the greatest, and other faces are disregarded. The structure is, therefore, suitable for the comparison of surface areas obtained, because only the (010) surface area has to be considered. The exposed surface area of T i 0 2 was 30.7 mz g-I, and the area contributed by V2O5 was therefore 13.3 mz g-l. This was approximately in agreement with the (010) face area of 18.5 mz g-' which was measured by the rectangular pulse method. Measurement of Exposed Surface Area of A1203by Selective Combustion. As found above, the benzoate species was adsorbed on VzO3 as well as on Alz03. However, one can easily indicate a higher activity of V2O3 in the catalytic oxidation than that of AlZ03,since the vanadium oxide is known to be available for various oxidation reactions. Therefore, the discrimination of oxide surfaces was then tried by utilizing the different oxidation activities. Injection of excess amounts of benzaldehyde on V2O5 not only reduced the oxide surface but also saturated the active sites with the benzoate species. The combustion by oxygen was selected as the simplest oxidation reaction to remove only the adsorbed residue on V 2 0 3with the benzoate adsorbed on alumina unconverted. If this condition is established, the amount of the benzoate which remains on alumina is measured by the ammonia reaction. However, no product was obtained on V2O5 10 mol % loading on A1203catalyst at ammonia injection after the combustion by oxygen pulse at 673 K. The benzoate species adsorbed not only on V203 but also on A1203were converted completely by the injection of oxygen at 673 K. On the other hand, the injection of oxygen did not influence the benzoate adsorbed on pure A1203a t all, since the benzonitrile obtained on Alz03 was not decreased by the injected oxygen. It is therefore considered that the vanadium oxide catalyzes the combustion of organic molecules adsorbed on the support as well as on itself. Therefore, the combustion by oxygen was then studied by changing the oxidation temperature. After the benzoate was oxidized by oxygen above 573 K, benzonitrile was no longer produced on V2O5 10 mol % loading on A1203,while it was obtained on the catalyst which was oxidized at temperatures below 573 K (Figure 3). The temperature dependence was critical, and a subtle change of temperature around 573 K influenced greatly the amount of benzonitrile obtained. This was influenced also by the manner of pulse injection. Because the rapid injection of oxygen decreased the amount of benzonitrile, a high partial pressure of oxygen in the carrier gas could be necessary to remove adsorbed species. When air was used in place of the pure oxygen, the temperature dependence was not critical but only broad, as shown in Figure 3. The critical change of the benzonitrile amount by the combustion indicated that the combustion of the benzoate species adsorbed on Alz03became rapid abruptly above 573 K. It was therefore considered that the benzoate adsorbed on V z 0 3 was
TABLE 11: Exposed Surface Area of A1203 and Ti02 by Selective Adsorption of Benzoate surf. area/m2g-'
catalyst
benzoate amo~nt/lO'~ g-'
BET
exposed support
37.8 20.3 14.2 0.26 0.18
21 1 185 150 95.4 43.1
212 114 79.6 1.5 1.o
vanadium oxide
v202
(010)
dispersion4/%
71 70 94 42
0 16.3 49.7 51.5 23.4
0 118 60 36 9.3
13.3
18.5
V205/A1203b
1 5 10 25 50
0
V205/Ti02
monolayer 5.52 44.0 "The surface area owned by a crystallized V02,, is 10.6 X lo4 pm2.
30.7 mol %.
3812 The Journal of Physical Chemistry, Vol. 89, No. 18, 1985
Niwa et al.
TABLE III: Amount of Carbon Dioxide Formed During the Elevation of Temperature up to 673 K
catalvst
(C0,/7Y/10'9
0 0.76 1.51 0.22
A1203
V2OS/Al2O3, 5 mol% V205/Alz03, 10 mol % v2°5
4 A m ~ u noft ~C02 X '/,.
Q000 E
CGMBUST;ON TEMP i K Figure 3. Amount of formed benzonitrile as a function of combustion temperature with oxygen ( 0 )or air (0) and (#) on slow injection of
oxygen. converted almost completely, and the benzoate species remaining at this temperature corresponded to the fraction of the species adsorbed on the exposed alumina surface. The amount of benzonitrile thus obtained, 14.6 X loi9 g-l, was in good agreement with that obtained by the selective adsorption (Table 11). The discrimination between the active oxide and the support was possible also by the selective combustion, if the oxidation condition was adequately regulated. Behaviors of adsorbed species were followed by IR spectroscopy. As shown in Figure 4, adsorption of benzaldehyde on VzOs (10 mol %) on A1203a t 673 K revealed the spectrum of adsorbed benzoate which was bound to A1 or V: Ph-C;.
/Y\
,M
0 M = AI or V
The benzoate species was then subjected to oxidation successively at 573,623, and 673 K. The benzoate species remained inert at 573 K, but a large portion was consumed at 623 K. At 673 K, the benzoate was removed completely by the combustion. Since the behavior agreed with those in the pulse reaction, it was confirmed that the benzoate species was removed as carbon dioxide by the combustion above 623 K.
Discussion In a previous study about the structure of supported vanadium oxide catalyst, the authors9 used various techniques in addition to the rectangular pulse method using NO-NH, reaction. According to the conclusion, vanadium oxide is supported on alumina depending on the loading amount. At 1 mol % loading in VZOS, it is immersed into the support bulk so that the active V=O site does not appear. Active sites ascribable to the V=O species are observed at more than 2 mol 7'% loadings. The support surface is exposed up to 35 mol 5% of vanadium oxide, but not exposed beyond this loading. Additionally, it was found that the vanadium oxide could be dispersed as mono to double layers at lower loadings below 5 mol %. The exposed surface area obtained in the present study could provide a similar model of the structure, since the alumina surface was completely exposed upon loading of 1 mol
3500
3000
1800
1600
:400
1270
Figure 4. IR spectra obtained upon adsorption of benzaldehyde (a) and upon oxidation by oxygen (90 torr) successively at 573 K (b). 623 K (c), and 673 K (d) on V 2 0 5 IO mol % on Al,O,.
7% of V 2 0 s , and filled with the oxide at more than 25 mol %. It could be concluded that both studies indicated the structure of supported vanadium oxide catalyst consistently. A small amount of carbon dioxide was formed during the elevation of temperature up to 673 K, as shown in Table 111. If the carbon dioxide is caused by the combustion of the benzoate on alumina surface by neighboring vanadium oxide, this will be an experimental error for the determination of exposed surface area, and was evaluated to be 11% in the case of 10 mol % V 2 0 5 on A1203. However, carbon dioxide was formed also on the vanadium oxide, and its formation density was 0.22 nm-* on V 2 0 s 10 mol % on A1203, and less than or in the same order of 0.41 nm-2 on pure V 2 0 5 . Therefore, it seems that carbon dioxide is formed on the vanadium oxide surface, and the adjacent vanadium oxide does not influence the benzoate species on the support. In view of this, the experimental error caused by C 0 2 formation would be negligible. Findings about the complete removal of adsorbed species on supported catalyst by the combustion are indicative of phenomena which are interesting but difficult to explain. Combustion of the benzoate adsorbed on the support would be possible by postulating an oxygen spillover from the vanadium oxide to alumina. Although the spillover of oxygen is not clearly confirmed in any case, one might consider this. It may be possible that relatively inactive vanadium oxide will be activated at a high partial pressure of oxygen at higher temperature and catalyze the combustion of a neighboring benzoate species. As mentioned above, there are two possible discrimination methods for metal oxide catalysts. One method utilizes the selective and specific adsorption of benzoate species, while another depends on the superior activity in the combustion. What kind of surface property is utilized depends on the specific oxide and support. Advantage of the experimental methods is not determined only by this experiment, however. Other systems of supported oxide catalysts will be studied to establish the best method for determination of the exposed surface area of support by using benzaldehyde-ammonia titration (BAT) method.
Acknowledgment. This work was partially supported by a Grant-in-Aid from the Ministry of Education, Science and Culture, Japan (No. 59470097). Registry No. Vz05, 1314-62-1; AlzO3, 1344-28-1; Ti02, 13463-67-7; V z 0 3 , 1314-34-7; benzoate, 766-16-7.