6390
J . Phys. Chem. 1990, 94, 6390-6396 high in energy to be assigned to a Ru(III)-dicarbonyl species and also do not decrease in intensity with evacuation. For RuLi-X ESR does not show any significant Rull'-carbonyl species. However, an IR band at 2144 cm-' might indicate a Ru-carbonyl species of some type with the other expected frequency unresolved. Two-pulse ESEM data for the samples after D 2 0 adsorption show both deuterium and aluminum modulation and confirm that Ru(II1) ion is located in the vicinity of a framework aluminum and is coordinated to water. This indicates that the Ru(II1) dimers located in the /3-cages dissociate and coordinate to three water molecules. The rhombic symmetry of the ESR spectra after D20 adsorption could indicate a distorted octahedral symmetry with Ru(II1) coordinated to three zeolite framework oxygens from a six-ring and three oxygens from water molecules. From this data alone it is difficult to distinguish whether the Ru(II1) coordinated to water is in the a- or /3-cage or both.
assignment of the bands in the 1900-2100-~m-~range, a range as high as 2130-2140 cm-I has been assigned by at least two groups to ruthenium-dicarbonyl species with the ruthenium in the trivalent In this study, we observed bands as high as 2125-2140 cm-', which indicates a dicarbonyl species. Together with the ESR results which show the existence of Ru"'-carbonyl species, we assign these bands to RU"'(CO)~species. After desorption of C O at higher temperature there are significant changes in the IR spectra. Several different overlapping peaks are observed in the l900-21OO-cm-' region. The literature assignments in this region differ widely (Table Ill) so we can make no reliable assignment. In RuNa-X two pairs of bands are observed at 21 39,2078 cm-I and 2025, 2065 cm-I; both decrease in intensity upon evacuation. This is in agreement with the ESR result which shows two different species (Figure 2b) which decrease in intensity upon evacuation. Thus, these two pairs of bands are assigned to two Ru"'(CO)~ species, having possibly different locations near site 11. The carbonyl species with higher wavenumber bands (2139 and 2078 cm-l) could be situated closer to the center of the hexagonal window (site [I) where it experiences greater electronegativity from the oxygens in the six-ring. This can cause less electron backdonation from the metal to the x* antibonding orbital of C O and result in a higher vibrational energy of C0.34 The other carbonyl species with lower energies (2125 and 2065 cm-l) could be situated further from the center of the six-ring into the a-cage. The infrared spectrum of RuK-X is similar to that of RuNa-X, and two species are assigned to Ru"'(CO)~ with bands at 2142, 2077 cmi and 2132, 2062 cm-I in agreement with ESR data (Figure 2c) which show two carbonyl species after CO adsorption. The ESR spectrum of RuCa-X (Figure 2d) shows one dominant species and another less intense species; however, IR spectra of RuCa-X show only one Ru"'(CO)~species with bands at 2142 and 2076 cm-I. Possibly the less intense species seen by ESR is not resolved by IR. The bands at 21 59 and 2100 cm-' are too
There is no major difference in the ESR spectra after evacuation at 300 "C for Ru-X zeolite exchanged with [Ru(NH+I3' or [ R U ( N H ~ ) ~ C ~The ] ~ +near . absence of paramagnetic species after activation at 300 "C may be explained by the presence of Ru(II1) dimers. The Ru(lI1) dimers are suggested to be located in the @-cagesand to dissociate upon adsorption of CO, 02,or H 2 0 to form additional paramagnetic species. Several Ru( III)-carbonyl complexes were observed dependent on the cocation. The existence of Ru"'(CO)~ was characterized by IR and supported by ESR. Superoxide ions are observed to form after adsorption of O2on the evacuated Ru(II1)-carbonyl complex. A cocation effect was found to affect the formation and geometry of the Ru(III)-carbonyl complexes, the superoxide ions, and the Ru(II1)-water complexes. The ESEM results show that Ru(1II) is situated close to one framework aluminum nucleus and is coordinated to three waters after water adsorption.
(34) Douglas, 8. E.; McDaniel, D. H.; Alexander, J. J. Concepts and Models oJlnorganic Chemistry, 2nd ed.; Wiley: New York, 1983; p 427.
Acknowledgment. This work was supported by the National Science Foundation, the Robert A. Welch Foundation, and the Texas Advanced Research Program.
Conclusions
Transient Bleaching of Small PBS Colloids. Influence of Surface Properties Milica T. Nenadovit, Mirjana I. Comor, Vesna Vasiir, and Olga I. Mitic* "Boris KidriE" Institute, Vifica. P.O.Box 522, 11001 Belgrade, Yugoslavia (Received: July 26, 1989; In Final Form: February 14, 1990)
Small PbS colloids with a particle diameter of 40 A were prepared in aqueous solution, and their absorption spectra exhibit several maxima. Injection of electrons into these particles was achieved by using the pulse radiolysis technique. Excess electrons trapped on the surface lead to a blue shift in the absorption edge of colloids. The appearance of this shift depends critically on the method of colloid preparation. PbS and CdS colloids prepared at pH < 6 have long-lived bleaching, which disappears after several seconds. On the other hand, absorption bleaching does not appear after the addition of hydroxide ions to colloidal solutions (pH > 8). The existence of a hydroxide ion on the particle surface most likely removes surface defects on which electrons are trapped. PbS colloids prepared in the presence of 3-mercapto-1,2-propanediolhave an unstructured absorption spectrum, which is due to a wide particle size distribution (10-50 A). The redox potential for 40-APbS colloids was determined by an established equilibrium between zwitterionic viologen radicals and colloids. It slightly shifts to the cathodic direction when the number of electrons injected into the particles is increased.
Introduction
The photoelectrochemistry of small semiconductor particles has been an area of active research. Recently, there has been growing interest in ultrasmall particles that fall into the transition range between molecular and bulk pr0perties.l These small particles ( I ) For comprehensive reviews see: (a) Brus, L. J . Phys. Chem. 1986. 90, 2555. (b) Henglein, A. Top. Curr. Chem. 1988, 143, 113. (c) Paterson, M. W.: Micic. 0. I.: Nozik, A. J. J . Phys. Chem. 1988, 92, 4160.
have physical and chemical properties that are modified from their bulk value^.^-^ Of particular interest is their property to have photoinduced blue shifts in the absorption These (2) Weller, H.; Koch, U.; Gutierrez, M.; Henglein, A. Eer. Eunsenges. Phys. Chem. 1984, 88, 649. (3) Brus, L. E. J . Chem. Phys. 1984, 80, 4403. (4) Nedeljkovic, J. M.; Nenadovic, M. T.; Micic, 0. I.; Nozik, A. J . J . Phys. Chem. 1986, 90, 12.
0 1990 American Chemical Society 0022-3654/90/2094-6390$02.50/0
Transient Bleaching of Small PbS Colloids photoinduced blue shifts are important for potential applications involving nonlinear optical effects. The origin of the blue shift has been attributed to excess charged carriers on particles. Colloidal semiconductors are usually synthetized with very high density defects and disorders on surface. The process of diffusion of the conduction band electron from the bulk to the surface and subsequent trapping at defect sites is very rapid. This process has been estimated to occur in less than about 0.1 ps for CdS and PbS ( 9), the colloids show no detectable bleaching. The absorption spectrum of PbS-PVA remains the same under these pH changes. By decreasing the pH again to the starting value, the absorption bleaching appears. The appearance of long-lived bleaching is very sensitive to pH. On the other hand, it should be noted that the lifetime and intensities of bleaching are of poor reproducibility due to difficulty in strictly controlling the surface chemistry. We have also found that hydroxide ions cause a change in the fluorescence of the PbS particle. The colloids prepared at pH 5 have a weak broad emission from 600 to 800 nm (Figure lb) with a low quantum yield of about OS%, indicating that radiationless recombination of the charge carriers is the dominating process. However, addition of hydroxide ions to adjust the pH to pH 8.5 and IO increases the band edge fluorescence intensity by about a factor of 2 and 4, respectively. On the other hand, the long-wavelength fluorescence only very slightly increases compared to that for the slightly acid solution, and the maxima at lower energies disappear. The higher emission yields in the alkaline solution could occur through the removal of surface sites at which the radiationless recombination of charge carriers takes place. It seems that electrons that lead to bleaching are located on such surface defects. A similar observation has been made recently for CdS colloids for which a drastic increase in fluorescence intensity was found when particles were covered with cadmium h y d r ~ x i d e . ~ ' To confirm our observation that the bleaching of exciton absorption is pH dependent, we chose the CdS colloids that were studied earlier in detail.7b These colloids were prepared at pH 4-5 with two different particle diameters, 30 and 50 A. The transient absorption spectra after injection of electrons into the (24) Liu, C.; Bard, A. J. J . Phys. Chem. 1989, 93, 3232. (25) Fojtik, A.: Henglein, A,; Katsikas, L.; Weller, H. Chem. Phys. Lett. 1987, 138, 535. (26) (a) Koch, A.; Fojtic, A,; Weller, H.; Henglein, A. Chem. Phys. Lett. 1985, 122,507. (b) Bahnemann, D. W.; Kormann, C.; Hoffmann, M.R.J . Phys. Chem. 1987, 91, 3789. (c) Haase, M.; Weller, H.; Henglein, A. J. Phys. Chem. 1988, 92, 482. (27) Dimitrijevic, N . M.: Kamat, D. V. J . Phys. Chem. 1987, 91, 2096. (28) Kamat, P. V.; Dimitrijevic, N. M.;Fessenden, R. W. J. Phys. Chem. 1988, 92, 2324. (29) Kormann, C.: Bahnemann, D. W.; Hoffmann, M. R. J . Phys. Chem. 1988, 92, 5196. (30) Vucemilovic. M. 1.; Micic, 0. 1. Radiat. Phys. Chem. 1988, 32, 79. (31) Spanhel, L.; Haase, M.; Weller, H.; Henglein, A. J . Am. Chem. Soc. 1987, 109. 5649.
7
tl
-0.02 - O ' O ' r
I Figure 4. Time profile at 500-nm bleaching recovery of 50-A C d S after electron pulse (experimental conditions as in Figure 3).
CdS colloids are shown in Figure 3b. In our experiments, electrons are injected into particles from several strongly reducing species including e-(aq),Cd+, and C02- radicals, which are denoted simply as e-. CdS,,,,,)
+ ne-
-
(e-(l), Cd+, CdO) CdS,,,,)
(5)
For 30-A CdS, the intensities of bleaching were observed as a function of absorbed dose (2-50 Gy). It was found that they slightly increase with decreasing absorbed dose. This is due primarily to the competition of the electron transfer to the colloid with radical-radical reactions. At the lowest dose, the extinction coefficient for the bleaching at 450 nm is 2 X lo4 M-' cm-I, somewhat smaller than the value previously observed by Henglein and c o - w o r k e r ~ .The ~ ~ excess electron on CdS may be localized on at least two sites: (i) a defect site on the particle surface that can trap electrons, e-(l); and (ii) a cadmium atom in CdS, Cd+, or Cdo (reaction 5 ) . Together with the transient bleaching, Figure 3b also shows absorption for longer wavelengths (A > 490 nm for 30-A CdS and h > 530 nm for 50-ACdS), which only slightly changes in seconds and which can be attributed to the cadmium atom. In any case, bleaching is not the result of the presence of Cdo atoms in colloids, since after the bleaching recovery the absorption spectrum of Cdo remains. Trapped electrons lead to the bleaching. Electron localization on the CdS particle was observed earlier by Henglein and co-workers by using the pulse radiolysis techniq~e.'~ The bleaching observed in the present case was obtained for colloids prepared in slightly acid media. Close examination of kinetic traces (Figure 4) indicates that the recovery process for 50-ACdS is actually composed of several processes ( T ] , = ~ 2.8 ms and T i / 2 = 1 s). This observation may be understood if
6394
The Journal of Physical Chemistry, Vol. 94, No. 16, I990
NenadoviE et al.
A 110
a7
0.5
0.3
0.I
300
350 WAVEUNtlH Inm)
Figure 5. (a) Absorption spectra of a solution containing 2 X 10-4 M Pb(CH3C02)2and 1 X IO-) M RSH at different pHs; pH shown for each spectrum. (b) Absorbances as a function of pH at 270 nm. (c) Absorbances at 330 nm. (d) Absorbances at 350 nm. Solutions contained 2 X IO4 M Pb(CH3C02), and four different concentrations of RSH: (A) 5 X IO-" M, ( 0 ) 1 X IO-) M, ( X ) 5 X IO-) M, and (0) 1 X M.
we take into account that different defect sites can be created during the preparation of colloids. When bleaching disappears, some absorption remains, which is attributed to the Cd atom. It is a stable absorption in the absence of oxygen. CdS colloids prepared by adding Na2S to the solution containing C d S 0 4 and H M P showed no detectable bleaching. This was also observed earlier by Albery and co-workers.6 However, the absorption bleaching appears in the same colloidal solution after addition of H 2 S 0 4to decrease the pH below 7. This again confirms our observation that long-lived absorption bleaching is connected with the presence of defects on the surface that can be removed by metal hydroxide precipitated on the surface. Surface Capped Particles with Thiol. We have recently examined the radiolytic generation and control of extremely small RSH, CdS particles in the presence of 3-mercapto-l,2-propanediol, which forms a complex with Cd2+ ions." The particles thus obtained were found to strongly photobleach close to the absorption edge32and have a high yield of fluorescence,II which occurs by recombination of trapped charged carriers. I n the present work, RSH was added to PbS colloids to bind strongly with the particle surface in order to decrease the ag~~
(32) Hayes, D.; Meisel, D.; Micic, 0. 1. J . Colloid Polym. Sci., in press
glomeration of small colloids. It is known" that RSH forms a strong complex with lead ions, such as polynuclear complexes [Pb3(RS)*+, Pb2RS3+, Pb3(RS)42-] and mononuclear species [PbRS+, Pb(RS)2, and Pb(RS),-1. As is clear from the results in the l i t e r a t ~ r e quantitative ,~~ information on the extent of the formation of complexes is difficult to obtain. In order to collect information on the absorption of RSH-Pb*+ complexes, we examined the spectra of the solution at various initial RSH and Pb2+ concentrations and various pHs (no buffer or stabilizers). Figure 5 shows the absorption spectra and absorbances at maximum wavelengths of the solutions containing 2 X IO4 M Pb2+ total and several different RSH concentrations as a function of pH. At the highest RSH concentration shown in Figure 5 , most of the Pb2+form polynuclear complexes at pH > 7.0. On the other hand, at pH < 5.0, mononuclear complexes are formed. To avoid interference of complex absorption in the absorption spectrum of the colloids, we chose the condition where the absorption of complexes is very small. For preparation of colloid, we used a 1 X 10" M RSH solution at pH 4-5. (33) (a) De Brabander, H. F.; Tombeux, J. J.; Van Poucke, L.C. J . Coord. Chem. 1974.4, 87. (b) De Brabander, H. F.; Herman, G. G.; Van Poucke, L . C . Thermochim. Acta 19'14, IO, 385.
The Journal of Physical Chemistry, Vol. 94, No. 16, 1990 6395
Transient Bleaching of Small PbS Colloids
Figure 6. (a) Absorption spectra of 2 X IO-' M PbS colloids and 0.1% PVA after addition of 1 X RSH (pH 4) at various times: 0, 1, 2, 3, and 7 days. (b) Absorption and emission spectra of PbS colloids (he = 450 nm) prepared by mixing Pb2+and H2S (2 X IO-' M PbS) in the presence of RSH (1 X M) and followed by filtration through 20-A pore ultrafilter. Final concentration 7 X M PbS. foJ
4
1.51
I
\
U
-0.2
500
300
700
WAVELENGTH /nm
Figure 7. (a) Absorption spectrum (-) and derived transient spectrum (---) 100 after the electron pulse of dialyzed colloidal 2 X IO-' M PbS. Initial colloids were prepared in the presence of 1 X lo-' M RSH. (b) Transient difference spectrum obtained after the electron pulse of a N20-saturatedsolution of 2 X IO-' M PbS (dialyzed),0.1 M 2-propano1,
pH
4.
Absorbed dose 53 Gy.
The solution of PbS colloids stabilized with PVA (2 X lo4 M PbS, 0.1% PVA) was treated with RSH (1 X IO-' M). The absorption spectrum changes and excitonic peaks completely disappear (Figure 6a), while the long tail near the absorption edge appears. The PbS colloids, which were initially produced by mixing Pb2+ and H2S in the presence of RSH, have the same spectrum as these equilibrated PbS with RSH. These colloids have diameters of about 10-50 A and are very resistant to anodic photocorrosion. The filtration of these colloids through a 20-A ultrafilter shifts the absorption spectrum to the UV region (Figure 6b). The emission spectrum (Figure 6b) shows the presence of defect states on PbS-RSH colloids. It is also found that PbS-RSH is more strongly fluorescent by about a factor of 2 compared to PbS-PVA. The changes in optical absorption that the colloidal PbS-RSH capped with RSH undergo in the presence of excess electrons are shown in Figure 7. The half-life of the formation of bleaching at a concentration of 2 X IO4 M PbS is found to be 27 pus while the recovery of bleaching takes about 400 ms. The bleaching of particle absorption in Figure 7 does not occur in the vicinity of the absorption edge, as is the case with PbS-PVA. The reason for this most likely lies in the presence of different fractions of particles in solution when thiol is present. In the pulse radiolysis
19
20
iogn tr-/cm3)
2'
Figure 8. Redox potential of PbS colloids as a function of electron number injected into particles. Solutions contained 0.1 M 2-propanol, 5X M ZV,and PbS colloids with different concentrations: (0) 4.5 X IO4 M, (A) 3.5 X IO-' M, (A) 2.5 X IO4 M, ( X ) 2 X IO-' M, and (0) 1 X IO-' M. Absorbed doses in the range 4-130 Cy.
experiment, the fraction of dominating species due to its higher concentration reacts faster with radicals and creates bleaching. It should be noted that this fraction is not low molecular weight aggregates of PbS since their absorption spectra are very similar to that of monomeric PbS (absorption peak at 530 nm) or slightly shifted to longer wavelengths.2' We can only speculate what is the chemical identity of the species responsible for the bleaching. Monomeric species can also be excluded since the solution was dialyzed just before experiment. Absorptions of polymeric complexes of [Pbk(RS),](2k-")+with S2- with an unknown chemical composition are probably bleached. It seems that very small particles have a rather large solubility product that favors complexing and a change of particle nature. Redox Potential of Colloids. The position of the lowest empty electronic state (reducing power) of the PbS colloids can be determined by observing which redox couples could inject electrons into the semiconductor particles. We found that radicals that have a redox potential above the energy levels of conduction band edges for both 40-Aand 20-A particles can inject electrons into these colloids. However, the situation is different with methyl viologen, MV', radicals (E" = -0.44 V), which could not inject electrons into 20-A particles but do so for large particles. This reflects the effects of size quantization on the position of the conduction band edge. We chose zwitterionic viologen radicals ZV- (E" = -0.37 V)34to determine the position of the lowest empty electronic states of 40-A PbS. In our experiments, ZV- radicals were produced by fast reaction with (CH3)*COH radicals in a N20-saturated solution in which 2-propanol (0.1 M) was present. ZV
-
+ (CH3)#0H
ZV-
+ (CH3)$0
mZV
+ PbS",,,,,~
(6) The electron-transfer reaction from ZV- to 40-A PbS reaches equilibrium. mZV-
+ PbS,,,,,)
i-
(7)
The initial concentration of ZV- was always measured spectrophotometrically at 650 nm before the electron-transfer reaction took place. For the particle concentration used in this work, reaction 7 takes place in the time range 0.05-0.1 5 s. Under this condition, the redox potential of the particles, E(e-,,,,)) and that of the solution are equilibrated. E(e-(co,l))= -0.37
+ 0.059 log {[ZV]/[ZV-]J
(vs NHE) (8)
The variation of E(e-(mll))with the concentration of injected electrons is shown in Figure 8. E(e-,m,l)) depends slightly upon the initial ZV and PbS concentrations since ZV is adsorbed on the PbS-PVA particle surface. Different initial potentials were obtained by changing the absorbed radiation dose from 5 to 130 Gy; this, in turn, changed the concentration of ZV- initially formed. The magnitude of the injected electrons in our experi(34) Kalyanasundaram, K.; Gratzel, M.Coord. Chem. Rev. 1986,69, 57.
6396
J . Phys. Chem. 1990, 94, 6396-6399
ments ( 1019-1021 e-/cm3) is sufficiently large to fill all bulk and surface traps that lie below the Fermi level, merging the bulk Fermi level with the conduction band edge. At the same time, trapped electrons slightly increase the band gap, which we observed as conduction band movement in the cathodic direction. Conclusion Intensive red transparent colloidal 40-A PbS stabilized with PVA was prepared in an aqueous solution. The absorption spectrum of these particles exhibits bands that shift to higher energies by about 1.4 eV as compared to bulk material, due to quantization effects. Addition of thiol, which forms a strong complex with Pb2+, decreases the particle size. In that case, a broad size distribution of colloidal particles from I O to 50 A was found. Pulse radiolysis was used to characterize the excess electrons in ultrasmall semiconductor particles. Injected electrons
lead to a blue shift in the absorption edge. The appearance of this shift depends critically on the method of the colloid preparation. The presence of trapped electrons at surface defects results in the observed blue shift. Species that strongly interact with the semiconductor surface can control the appearance of the blue shift. Metal hydroxide completely suppresses the absorption shift since it covers the surface of PbS or CdS colloids and blocks the defects sites at which electrons are trapped. The position of the lowest empty electronic state of 40-A PbS was determined by an established equilibrium between the ZV- radicals and colloids and was found to be about -0.1 V. This value becomes slightly more negative with increasing number of injected electrons. Acknowledgment. We express our gratitude to Dr. T. Rajh for the synthetic efforts and for the TEM characterization and V. Ljubisavljevic for technical assistance.
Effect of Particle Size on CO Hydrogenation Activity of Silica Supported Cobalt Catalysts Sui-Wen Ho, Marwan Houalla, and David M. Hercules* Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260 (Received: August 28, 1989; In Final Form: March 5, 1990)
Two series of silica supported cobalt catalysts were prepared by incipient wetness impregnation, one by varying the calcination temperature (200-400 "C, 3 wt % Co) and the other by changing the cobalt loading (1-10 wt % Co). Examination by ESCA, XRD, and H2 chemisorption showed that Co30, is the dominant phase. The cobalt phase is reduced to cobalt metal at 400 "C. The cobalt particle sizes obtained from ESCA correlated well with those derived from H2 chemisorption and XRD line broadening. The turnover frequency of Co/Si02 for CO hydrogenation was invariant with cobalt dispersion in the range of 6-20s dispersion.
Introduction The effect of particle size on C O hydrogenation activity has been studied for most supported group VI11 metals. A trend of decreasing activity with increasing dispersion is generally observed for Nil, R u , ~Fe,3 and C O . ~ In . ~ the case of cobalt, CO hydrogenation activity was reported to be strongly dependent on particle size over a wide range of cobalt dispersion (6-80%).4 However, the catalysts measured used a number of supports, including those that are known to exhibit strong metal-support interaction, such as Ti02. In addition, the reducibility of cobalt was below 50% for most catalysts studied and was particularly low for the highly dispersed catalysts. As a consequence, the observed activity change cannot be unambiguously attributed to the structure sensitivity of C O hydrogenation on cobalt catalysts. The purpose of the present study is to "isolate" the effect of particle size in the case of the cobalt system from effects of other variables. To achieve this objective, the following steps were taken: ( 1 ) Silica was used as a support to minimize metal-support interaction. (2) The calcination conditions were selected to allow complete reduction of cobalt phase. The Co/Si02 catalysts were characterized by ESCA, XRD (X-ray diffraction), and hydrogen chemisorption. The cobalt ( I ) Bartholomew, C. H.: Pannell, R. B.; Butler, J. L. J . Caral. 1980, 65, 335. (2) Kellner. S. C.; Bell, A . T. J . Cafal. 1982, 75, 251. ( 3 ) Jung, H.-J.; Walker, P. L.; Vannice, M. A. J . Caral. 1982, 75, 416. (4) Reuel. R. C.; Bartholomew, C. H. J . Caral. 1984, 85, 78. ( 5 ) Lisitsyn, A. S.;Golovin, A . V.; Kuznetsov, V. L.;Yermakov, Y. U. 1. C, M o l . Chem. 1984, I . 1 1 5.
0022-3654/90/2094-6396$02.50/0
dispersions obtained were correlated with the activities for CO hydrogenation. Experimental Section Catalyst Preparation. The SiO, support (Cab-0-Sil, BET surface area 192 m2/g, pore volume 1.O cm3/g) was wetted with distilled water, dried at 100 "C for 16 h, and calcined at 400 "C for 8 h before use. Two series of catalysts were prepared by incipient wetness impregnation of the Si02support with cobalt(I1) nitrate (Fisher) solutions. The first series was prepared with various cobalt loadings. The nominal cobalt content (as cobalt metal) was varied from 1.0 to 10 wt % of the Si02support. The impregnated powders were dried at 100 "C for 16 h and then subjected to further treatment at 100 "C for 300 h. The mild "calcination" treatment was adopted because it yields better dispersed cobalt phase. The second series was prepared with a single cobalt loading (3 wt %); the dried powders (100 "C, 16 h in air) were heated in vacuo (few Torr) at various temperatures (200-400 "C). The first series will be designated as CoxSiA and the second Co3SiVy, where x is the cobalt loading (wt % Co), y stands for the calcination temperature, and A or V stands for the thermal treatment in air (A) or in vacuo (V). Catalyst Co3SiV400-E was prepared with ethanol as impregnating solvent instead of distilled water. The BET surface area remained relatively constant (190 f 10 m2/g) independent of treatment. Reduction of Catalysts. Catalysts were reduced under hydrogen flow (50 cm3/min; 99.999%) at 400 "C for 8 h. X-ray Diffruction ( X R D ) . X-ray powder diffraction patterns were taken by using a Diano XRD-6 diffractometer employing Ni-filtered Cu Ka radiation (1.5405 A). The X-ray tube was (C 1990 American Chemical Society