6494
J . Phys. Chem. 1989, 93, 6494-6501
Comparative Cluster Reaction Studies of the V, Nb, and l a Series Yoon Mi Hamrick and Michael D. Morse* Department of Chemistry, University of Utah, Salt Lake City, Utah 84112 (Received: January 12, 1989; In Final Form: April 5, 1989) Reactions of the group 5 transition-metal clusters (V,, Nb,, and Tan) have been investigated in the gas phase by using a fast-flow chemical reactor. Dissociative chemisorption of D2 and N2 is found to be size selective for these bare metal clusters, as well as for the cluster monoxides V,O and Ta,O and monocarbides V,C. The effect of a single atomic impurity (C or 0)wanes for clusters larger than 10 atoms and becomes negligible for clusters larger than 15-20 atoms. This loss of sensitivity to impurities is consistent with the loss of cluster size specificity, which occurs in the same size range. Evidence of structural isomerism is found for Nb9, Nbll, NbI2, and Talz, which react with N2 with biexponential reaction kinetics. The rate of chemisorption of ethane (C&) displays a steady increase with cluster size for V, and Nb,, but Tan clusters are only slightly reactive. Finally, the rate of addition of a second ligand to a metal clusters is considered, and examples are analyzed for Nb,N2 + N2
I. Introduction Metal clusters constitute a unique class of molecules which interpolate between the two extremes of the isolated metal atom and the bulk metallic surface. In recent years, the chemical and physical properties of the small metal clusters ( n = 2-40) and the approach of these properties to bulk values have become a focus of considerable attention.l-’ Spectroscopic studies of the smallest clusters (dimers and trimers) have provided electronic and geometrical information for a few s y s t e m ~ , 4although ~~ much is yet to be learned. For larger clusters ( n = 4-40) spectroscopic data are completely lacking except for a few isolated examples. Chemical reaction studies provide a useful (but indirect) probe into the structural makeup of these novel species and directly probe the approach of chemical behavior to the bulk metallic surface. The earliest reaction studies of iron, cobalt, and niobium clusters reacting with N2 and D2have shown sharp variations in reactivity as a function of cluster s i ~ e . ~ - Following ’~ these early studies considerable t h e o r e t i ~ a l ~and ~ ’ ~experimental2241 -~~ efforts have ( I ) Kaldor, A.; Cox, D. M.; Zakin, M. R. Adu. Chem. Phys. 1988,70,211. (2) Cox, D. M.; Reichmann, K. C.; Trevor, D. J.; Kaldor, A. J. Chem. Phys. 1988,88, 111. (3) Trevor, D. J.; Kaldor, A. In High Energy Processes in Organometallic Chemistry; Suslick, K. S., Ed.; American Chemical Society: Washington, DC, 1987; ACS Symp. Ser. No. 333, p 43. (4) Weltner, W.; Van Zee, R. J. Annu. Reu. Phys. Chem. 1984,35, 291. (5) Morse, M. D. Chem. Reu. 1986, 86, 1049. (6) Salahub, D. R. Adv. Chem. Phys. 1987, 69, 447. (7) Moskovits. M. Metal Clusters: Wilev-Interscience: New York. 1986. (8) Richtsmeier, S. C.; Parks, E. K.; Li;, K.; Pobo, L. G.; Riley, S. J. J. Chem. Phys. 1985, 82, 3659. (9) Parks, E. K.; Liu, K.; Richtsmeier, S. C.; Pobo, L. G.; Riley, S. J. J. Chem. Phys. 1985, 82, 5470. (IO) Liu, K.; Parks, E. K.; Richtsmeier, S. c.;Pobo, L . G.; Riley, S. J. J. Chem. Phys. 1985, 83, 2882, 5353. (1 1) Tievor, D. J.; Whetten, R. L.; Cox, D. M.; Kaldor, A. J. Am. Chem. SOC.1985, 107, 518. (12) Whetten, R. L.; Cox, D.M.; Trevor, D. J.; Kaldor, A. J. Phys. Chem. 1985, 89, 566. (13) Geusic, M. E.; Morse, M. D.; OBrien, S. C.; Smalley, R. E. Rev. Sci. Instrum. 1985, 56, 2123. (14) Morse, M. D.; Geusic, M. E.; Heath, J. R.; Smalley, R. E. J. Chem. Phys. 1985, 83, 2293. (15) De Heer, W. A.; Knight, W. D.; Chou, M. Y.; Cohen, M. L. Solid State Phys. 1987, 40, 93. (16) Cohen, M. L.; Chou, M. Y.; Knight, W. D.; de Heer, W. A. J. Chem. Phys. 1987, 91, 3141. (17) McAdon, M. H.; Goddard, W. A,, 111 J. Phys. Chem. 1987,91,2607. (18) Koutecky, J.; Fantucci, P. Chem. Reu. 1986, 86, 539. (19) Phillips, J. C. Chem. Reu. 1986, 86, 619. (20) Wahnstrom, T.; Rosen, A. Surf. Sci. 1987, 189, 788. (21) Rosen, A.; Rantala, T. T. 2.Phys. D 1986, 3, 205. (22) Rohlfing, E. A.; Cox, D. M.; Trevor, D. J.; Kaldor, A. Phys. Rev. Lett. 1985, 54, 1494. (23) Whetten, R. L.; Zakin, M. R.; Cox, D. M.; Trevor, D. J.; Kaldor, A. J. Chem. Phys. 1986, 85, 1697. (24) Cox, D. M.; Whetten, R. L.; Zakin, M. R.; Trevor, D. J.; Reichman, K. C.; Kaldor, A. in Aduances in Laser Science; Stwalley, W. C., Lapp, M., Eds.; American Institute of Physics: New York, 1986; AIP Conf. Proc. No. 146, p 43.
0022-365418912093-6494$01.50/0
concentrated on the effect on cluster size on various properties. Ionization potential^,^^-^^ electron affinities,2628 and magnetic propertiesBVM have been observed to depend on cluster size, in some cases quite strongly. More recently, reaction studies have been conducted on cluster ions to attempt to discern the effect of charge state on the reactivity of a given ~ l ~ ~ t e r .In~the ~ tradition - ~ ~ , ~ ~ , ~ ~ of surface science, saturation and desorption studies of small iron clusters have provided information concerning binding sites and cluster geometry, much as adsorption isotherm measurements have provided for metal ~ u r f a c e . ~ ’At - ~ ~the other end of the spectrum, just as metal atom reactions depend on the orientation of the reactant, the reactivity of small niobium clusters also shows a stereochemical selectivity which decreases with increasing cluster size.@ Evidence of cluster structural isomerism, which must occur for sufficiently large metal clusters, has recently been observed for small niobium cluster^^^,^^ and cluster cation^.^'.^^ Amid this recent expanse of intriguing and creative experiments, there has not yet been much effort focusing on the possibility of periodic trends in cluster reactivity. Since many chemical and physical properties are similar within a group or family, a comparative study of transition-metal clusters within a given family is warranted. In this paper we present a comparative study of the group 5 transition-metal cluster reactions. These metals, vanadium, niobium, and tantalum, are all highly refractory, share the same bulk metallic crystal structure (bcc), and react readily (25) Zakin, M. R.; Cox, D. M.; Whetten, R. L.; Trevor, D. J.; Kaldor, A. Chem. Phys. Lett. 1987, 135, 223. (26) Zheng, L.-S.; Karner, C. M.; Brucat, P. J.; Yang, S. H.; Pettiette, C. L.; Craycraft, M. J.; Smalley, R. E. J. Chem. Phys. 1986, 85, 1681. (27) Pettiette, C. L.; Yang, S. H.; Craycraft, M. J.; Conceicao, J.; Laaksonen, R. T.; Cheshnovsky, 0.;Smalley, R. E. J. Chem. Phys. 1988,88,5377. (28) Leopold, D. G.; Ho, J.; Lineberger, W. C. J. Chem. Phys. 1987,86, 1715. (29) Cox, D. M.; Trevor, D. J.; Whetten, R. J.; Rohlfing, E. A,; Kaldor, A. Phys. Reu. B 1985, 32, 7290. (30) Cox, D. M.; Trevor, D. J.; Whetten, R. J.; Rohlfing, E. A,; Kaldor, A. J. Chem. Phys. 1986, 84, 4651. (31) Elkind, J. L.; Weiss, F. D.; Alford, J. M.; Laaksonen, R. T.; Smalley, R. E. J. Chem. Phys. 1988.88, 5215. (32) Alford, J. M.; Weiss, F. D.; Laaksonen, R. T.; Smalley, R. E. J. Phys. Chem. 1986, 90, 4480. (33) Brucat, P. J.; Pettiette, C. L.; Yang, S.; Zheng, L.-S.; Craycraft, M. J.: Smallev. R. E. J. Chem. Phvs. 1986. 85. 4747. (34) Cox, D. M.; Reichmani, K. C.;’Trevor, D. J.; Kaldor, A. J. Chem. Phys. 1988, 88, 111. (35) Zakin, M. R.; Brickman, R. 0.; Cox, D. M.; Kaldor, A. J. Chem. Phys. 1988, 88, 3555. (36) Zakin, M. R.; Brickman, R. 0.;Cox, D. M.; Kaldor, A. J. Chem. Phys. 1988, 88, 6605. (37) Parks, E. K.; Nieman, G. C.; Pobo, L. G.; Riley, S. J. J. Chem. Phys. 1987, 86, 1066. (38) Parks, E. K.; Nieman, G. C.; Pobo, L. G.; Riley, S. J. J. Chem. Phys. 1988, 88, 6260. (39) Parks, E. K.; Weiller, B. H.; Bechtold, P. S.; Hoffman, W. F.; Nieman, G. C.; Pobo, L. G.; Riley, S. J. J. Chem. Phys. 1988, 88, 1622. (40) Song, L.; Eychmuller, A,; El-Sayed, M. A. J. Phys. Chem. 1988, 92, 1005. (41) Hamrick, Y.; Taylor, S.; Lemire, G. W.; Fu, Z.-W.; Shui, J.-C.; Morse, M. D.J . Chem. Phys. 1988, 88, 4095.
0 1989 American Chemical Society
Reactions of V, Nb, and Ta Clusters with H2 and N 2 to the point of lattice penetration?2 In this paper investigations of the reactions of V,, Nb,, and Tan with D2, N2, and C2H6are reported. Although the reactions of these metal clusters with D2 and N 2 have been previously studied, it is useful to compare data obtained on the same instrument since results are sensitive to source conditions. In particular, the cluster temperature may well differ from source to source. In addition, we have extended the size range of the clusters investigated to larger species in some cases and have studied the effects of impurity atoms by comparing the reactivity of V,C, V,O, and Ta,O with that of the parent bare metal clusters. 11. Experimental Section The metal cluster source and fast-flow chemical reactor used in the present study are quite similar to those used in previous investigation^.'^*^^^^^ A pulsed supersonic beam of metal clusters is produced by laser vaporization of a metal target within the throat of a pulsed supersonic nozzle. For this investigation niobium (Materials Research Corp., '/s-in. diameter), vanadium (Alfa, '/4-in. diameter), and tantalum (Aesar, 0.002-in.-thick foil) targets were used. Metal vaporization is achieved by focusing (100-cm focal length lens) the second-harmonic output (532 nm, 40-80 mJ/pulse) of a Q-switched Nd:YAG laser (Quantel Model 580-10) to a 0.5-mm-diameter spot on the metal target rod or disk. Helium, purified by a molecular sieve trap maintained at 77 K and regulated to a pressure of 125 psig, is pulsed over the metal target by a magnetically operated double-solenoid valve. The high density of helium in the 2-mm-diameter nozzle throat quenches the metal plasma, thereby forming clusters ranging up to 45 atoms in size. The helium carrier gas along with the newly formed clusters then expands into a fast-flow reaction tube (10 cm long, 1-cm i.d.) positioned approximately 5 cm downstream from the point of vaporization. The sudden decrease in density which accompanies expansion into the reaction tube then halts further clustering. Four equally spaced hypodermic needles protruding into the reaction tube are used to inject either reactant gas diluted in helium or pure helium as a control experiment. These needles are supplied from a common annular reservoir, which is itself supplied with gas from two independently pulsed solenoid valves (General Valve, Series 9) monitored by mass flowmeters (Teledyne Hastings Raydist, Model STSH-50). The average pressure in the reactor tube is estimated to be approximately 100 Torr of a 0.0-2.0% reactant/helium mixture. Alternate pulses of reactant/helium mixtures and pure helium control gas are injected into the flow tube to allow for a comparison between the bare cluster distribution and the product distribution. Following their passage through the reaction tube, the metal clusters expand into vacuum and pass through a skimmer into the ionization region of a reflectron time-of-flight mass spectrometer (TOFMS). Detection of the clusters and their reaction products is accomplished by photoionization time-of-flight mass spectrometry using an excimer laser (Questek, Model 2420) operating on the ArF (6.4 eV) or F2 (7.9 eV) transitions. The laser fluence is kept low, in the range of 5-50 MJ/cm2, to minimize multiphoton absorption and consequent fragmentation of the metal clusters. The resulting ions are then extracted perpendicular to the molecular beam axis, accelerated up a reflecting TOFMS, and collected on a microchannel plate detector (Galileo, Model 3025-B) coupled to a 50-ohm anode assembly. The signal is amplified (Pacific Instruments, Model 2A50), digitized (Transiac, now DSP Technology, Model 2001), and stored as a mass spectrum on a DEC 11/73 microcomputer. Typically 1000 experimental cycles are averaged for each reagent flow rate, resulting in a reactant mass spectrum and a control mass spectrum. Figure 1 shows a selected portion of the time-of-flight mass spectra for niobium clusters reacting with nitrogen. The figure shows both a control mass spectrum, where (42) Bond, G. C. Catalysis by Metals; Academic Press: London, 1962. (43) Geusic, M.E.; Morse, M. D.; Smalley, R. E. J . Chem. Phys. 1985, 82, 590.
The Journal of Physical Chemistry, Vol. 93, No. 17, 1989 6495 I
I
I
I
I
I
1
I
500
600
700
800
Mass (amu)
Figure 1. Reaction study of small niobium clusters reacting with N2. The control experiment, indicated by the dashed lines, provides the mass spectrum when only pure H e is injected into the flow tube. The solid trace shows the product distribution observed upon injection of 0.19 Seem of N2 diluted in pure helium carrier gas. Varying degrees of depletion of both the bare metal clusters and their monoxides, along with the growth of multiple nitride products, are observed.
only helium is supplied to the reactant needles, and the product mass spectrum, which results when a nitrogen mixture is injected into the flow tube. Nitride products (M,N2) are observed as well as depletion of the bare cluster signal. The relative reaction rate of these clusters is determined by monitoring the depletion of the bare metal clusters as a function of reactant flow rate. A kinetic analysis of the chemisorption and stabilization steps, assuming a pseudo-first-order process, gives3 where 0,is the natural logarithm of the depletion ratio for cluster size n, [R]is the reactant concentration in the reactor, t is the residence time in the reactor, and k, is the apparent reaction rate, which approaches the true forward rate constant in the limit of high helium pressures. Because of the pulsed nature of the experiment, an accurate determination of [R] and t is not possible. However, since t is constant and independent of cluster size, and [R] is proportional to the reactant flow rate (which we measure precisely), a relative measurement of the rate constant can be made. This is done by measuring D, as a function of the flow rate of reactant gas, FR,and fitting the data to the functional form The empirical rate constant k,,' differs from the apparent forward rate constant, k,, by a constant factor which involves the time spent in the fast-flow reaction tube and the proportionality constant which relates FR to [R]. The proportionality constant relating k, and k,,' is apparatus-dependent but should not depend on the metal cluster or reactant gas under investigation. As a result, it should only be necessary to scale our reported values of k,,' to obtain correct apparent forward rate constants k,. Note that we measure our reactant flow rates in standard cubic centimeters of pure reactant per minute (sccm), so the units of k,,' are sccm-I. Figure 2 shows the logarithm of depletion for Nb4 and Nb5 reacting with N2. The linear plots of D, as a function of flow rate demonstrate that indeed pseudo-first-order kinetics are followed for the addition of N 2 to Nb4 and Nb5. 111. Results and Discussion The group 5 metal clusters, V,, Nb,, and Tan, are particularly convenient for mass spectrometric studies, because the natural abundance of the major isotopes are very high (99.76% 51V, 100% 93Nb,and 99.988% IslTa). Even metal clusters as large as V,, or Ta30 are 93% 51V,0and 99.6% '81Ta30,respectively. The resulting simplification of the cluster mass spectra has permitted us to monitor the reactivity of the bare metal clusters, along with the reaction rates of the common impurity clusters V,O, V,C, and Ta,O. The group 5 metals are quite refractory and form clusters efficiently. Clusters ranging from n = 2 to 45 were routinely produced in our source. At first, tantalum seemed the exception
6496
The Journal of Physical Chemistry, Vol. 93, No. 17, 1989
I
-d g . -d -
I
I
I
TABLE I:
Cases of Biexponential Depletionsa
system
0
h
3
-1
-2
r
-3 -4
Hamrick and Morse
1
I
I
I
0.20
0.15
0.10
0.05
Nz flow rate (sccm) Figure 2. Logarithmic depletion of the bare metal cluster ion signals for
Nb4 and Nb5, plotted as functions of N2 flow rate (sccm). The linear relationship obtained here demonstrates the validity of pseudo-first-order kinetics in the addition of N2 to Nb,. The slope of each plot provides the effective rate constant k,,’, as discussed in the text.
.-
+ D2
-Nbn 102‘
I
I
I
I
I
5
10
15
20
25
Nb9 + N,b Nbg + N2‘ Nbg + D,b Nb,, + N2b Nb,, + N2C Nb,, + D2b Nb12+ N2* Nblz + N2C Nbt2+ Dzb Ta12+ N2b
CI
k,Id
c2
k2’
0.540 0.587 0.807 0.592 0.552 1.007 0.606 0.560 0.773 0.692
2.906 2.746 2.042 4.585 4.092 6.793 16.33 23.58 4.43 20.21
0.376
0.243 0.309 0.021 0.442 0.478 0.889
0.401
0.233 0.327 0.367 0.126 0.361 0.393 0.319 0.344
0.017
0.072 0.059 2.123
“Data have been least squares fitted to the formula [M,]/[M,],, = c1 exp[-kl’FR] + c2 exp[-k,’F,], which assumes two forms of M,, with fractional concentrations cI and c2 and effective rate constants k i and k i . All four parameters were determined by the least-squaresfit, so c1 and c2 are not constrained to sum to 1. Estimated errors for c, and c2 are f10-20%; errors in k,’ and k i are probably similar except for the slowest rates, which are more difficult to measure. ”Data obtained using low-intensity ArF (6.42 eV, 193 nm) radiation for photoionization. CDataobtained using low-intensity F2 (7.90 eV, 157 nm) radiation for photoionization. d k l ’ and k i are effective rate constants, give in units of sccm-I. a function of D2 flow rate shows a rapid.linear depletion followed by a slower linear depletion at high D2flow rates. This is consistent with two forms of Nb9 and NbI2,one of which is highly reactive while the second is relatively inert. In previous publications we4, and the Exxon group35have proposed that the two forms of Nb9 and Nb12correspond to different structural isomers which do not interconvert on a 100-ps time scale at 300 K. On this basis one expects a biexponential depletion of the form
Cluster size (n)
Figure 3. The effective rate constant k,,’, in units of sccm-l, displayed as a function of cluster size for D2chemisorption on V, and Nb, clusters. Note the similarity in reaction patterns for V, and Nb, and the disparate
reactivities of the two forms of Nb9 and Nb12. Estimated errors in the values of k,,‘, based on the least-squares fit to the data, are typically 110-20%, both here and in Figures 4, 5, and 7-9. to the rule since we initially only detected Tan up to n = 12. However, by redirecting the excimer ionization laser beam along the molecular beam axis, we were able to ionize the Tan clusters farther upstream. For heavy ions such as TaN+this was necessary, since it enabled a better compensation for the molecular beam velocity and allowed the heavy ions to reach the detector. Compensation for the molecular beam velocity by the use of electric deflection plates, although successful for the lighter ions, was not effective for masses above 2500 amu, presumably because the deflection plates were mounted 7 cm above the molecular beam axis. Rather than rebuilding our ion source, we found that down-axis photoionization was quite effective in improving the heavy ion transmission through our reflectron TOFMS. By this arrangement tantalum clusters as large as ‘81Ta45(mass 8142.66 amu) were detected. For the other metals the photoionization laser beam crossed the molecular beam at right angles; this limited the size of Nb, clusters detected to n 5 26 but was inconsequential for V,. We begin by discussing the chemisorption of D2on V,, V,C, V,O, and Nb,. This is followed by a more revealing study of the reactions of V,, V,C, V,O, Nb,, Tan, and Ta,O with N,, where a full comparison of vanadium, niobium, and tantalum is made. Next we compare the reactions of all three metals with ethane. Finally, we present a discussion on the feasibility of monitoring Nb5N4as an example. consecutive reactions using Nb5N2 N, A . D2 Chemisorption by V,, V,C, V,O, and Nb, Clusters. Eigure 3 displays the effective rate constant k,,‘ for the reactions of V, and Nb, with D,, as determined by fitting plots of D, (the logarithmic depletion, defined in eq 2.1) as a function of D2 flow rate according to eq 2.2. In all but two cases, Nbg and Nb12,an excellent linear correlation of D, with D, flow rate was obtained, indicating that the assumption of a pseudo-first-order kinetic scheme is correct. For Nbg and Nb12,however, a plot of D, as
+
-
where cI and c2 are the fractional concentrations of the two isomeric forms of M , (weighted by their photoionization cross sections), k,’ and k,’ are the effective rate constants of the two isomeric forms, and FR is the measured flow rate of the reactant gas (in this case D2). A nonlinear least-squares fit to this functional form provides values of cl, c,, kl’, and k,’. For Nbg and Nb,, reacting with D2 the resulting values of kl’ and k2/ are plotted in Figure 3 as distinct points and are given in Table I along with the corresponding coefficients cl and c2. In a previous publication4’ we reported the existence of two structural forms of Nbl,, yet we have plotted only one point for N b t l on Figure 3. As shown in Table I, a biexponential fit of Nbll reacting with D2 gives constants indicating the reactive form has nearly a 100% abundance, making it appear that only one structural form is present. When the same cluster is investigated with N2, however, the biexponential reaction kinetics are more obvious and are reproducible by using either an ArF or F2 photoionization laser. Although it is likely that the proportions of the two structural forms of Nb,, are sensitive to source conditions, we have attempted to hold these constant as we investigated all of the reactions reported in this paper. In the particular case of Nbll, we believe that two isomers are present but have nearly the same reaction rate with D,. On the other hand, these two isomers have very different reaction rates with N2, as indicated in Table I. Recent observation^^^ on the reactivity of Nb,+ with H2 reveal the same effect for Nbll+: two structural isomers are probably present, based on the observation of two limiting, fully saturated clusters (NbllH6+and NbllHI6+),but both react with nearly the same rate. It is interesting and unexpected that two structural isomers should exist for both Nbll and Nbll+ and that both should react with hydrogen with comparable rates. In contrast to niobium, we find no evidence of structural isomerism among vanadium clusters, V,, ranging from n = 3 to 32, in their reactions either with D, or with Nz. Three possibilities exist which could account for this observation. First, isomers could be present, but the isomers of a given cluster size could all react with the same rate. In view of the relatively wide variation of reactivity of small V, clusters (n < 15) with D, and N, as functions of cluster size (see Figure 3 and Figure 9, this explanation seems
Reactions of V, Nb, and Ta Clusters unlikely. Second, isomers may be present, but interconversion between isomeric forms is proceeding rapidly at the temperature of the flow tube. This would result in the observation of an averaged reaction rate, which would depend on the fractional time spent in each isomeric configuration. Finally, isomers may not be present, because the metal clusters successfully find the lowest potential energy minimum before they are cooled enough to be trapped in a particular configuration. These latter two possibilities seem likely for the vanadium clusters, particularly since vanadium is a much softer and lower melting metal than is niobium. For the larger vanadium clusters ( n > 20), however, it is plausible that multiple structural isomers may exist, and these may all react with nearly the same rate because all are likely to have the active site needed to chemisorb efficiently. Other investigators' find some evidence of cluster isomerism in V,, however, based on its approach to two different limiting levels of deuteration (V9D4and V9Dl,) at high deuterium concentrations. We have not repeated this experiment at high D2 flow rates and therefore cannot comment on the presence or absence of two limiting deuterides in our experiments. The presence of two limiting levels of deuteration could result from effects other than cluster isomerism, however, including such possibilities as kinetic bottlenecks or cluster isomerization following addition of one or more D2 molecules. In any case, we find that V, depletes with a single-exponential decay regardless of whether D2 or N, is the reactant. Figure 3 shows that there are some similarities in reactivity pattern for V, + D2 and Nb, + D,. The vanadium system shows pronounced even-odd alternations in reactivity, with vg, v8, VI,, and VI, significantly less reactive than their odd-numbered neighbors. The niobium system shows a less extensive pattern of even-odd alternation, with Nb8, Nb,o, the less reactive form of NbIz, and Nb16all relatively inert to D2. N o investigations of the reactivity of tantalum clusters, Tan, with D2 were made in the present study due to difficulties in cleanly separating the Tan and Ta,D, mass spectral features, particularly for clusters larger than Tala. An interesting question in metal cluster reaction studies concerns the effect of a single atomic impurity such as carbon or oxygen on the reactivity of a given metal cluster. For any potential commercial applications clusters will have to be supported, perhaps on oxide substrates such as AI,O, or S O z or on graphite. In any case, metal cluster oxides and carbides are likely to be formed in the course of commercial catalytic processes. For niobium clusters a qualitative observation has been that for certain clusters such as Nb, the oxide in a sense "poisons" the cluster, or makes it less reactive, while an opposite effect is seen with Nb8 where the oxide enhances the rea~tivity.'~Figure 4 displays the reactivities of v,, V,C, and V,O with D,, which were obtained with the naturally occurring V,C and V,O impurities that are present in our source. These impurities were found to be most intense immediately after the chamber had been opened to atmosphere and were found to diminish slowly as the metal surface was vaporized. Carbides probably form from laser vaporization/ fragmentation of residual diffusion pump oil in the vaporization mount, while oxides (which typically take much longer to clean up) are present in the metal surface itself. For reasons unknown, only vanadium was found to form significant quantities of cluster carbides. For the data displayed in Figure 4 care was taken to exclude high flow rate points where the buildup of V,D6 products could obscure the depletion of V,C and V,D8 and V,CDZ products could obscure the depletion of V,O. Since the carbide and oxide impurities were of approximately equal intensity in the mass spectrum, depletion of V,O may be artificially low due to buildup of V,CD2. Both depletion of V,O and buildup of V,CD2 are first-order in D,, so even constraining ourselves to the low flow rate points will not eliminate this possible artifact. For this reason reaction with N, (see section B) provides a more reliable comparison of the relative reactivities of V,C and V,O. With high flow rate points omitted by necessity, only the initial depletions could be measured for these imeuritv-containing clusters, so the reaction rates derived from these &dies are based only on the initial slopes of plots similar to those given in Figure 2. Thus,
The Journal of Physical Chemistry, Vol. 93, No. 17, 1989
-!I w
IO'
.P
loo
6497
'i
d
!
!
1
20 Cluster 10 size (n) 15
Figure 4. Reactivity patterns for the reactions of V,, V,C, and V,O with D2.For small clusters (n = 3-10) the reactivities for each of the three species differ dramatically, while for large n a convergence in reactivity is observed.
although one might expect isomeric forms of V,O and V,C to be more likely than for V,, we are sensitive only to the most reactive form of these impurity-containing clusters and therefore cannot observe structural isomerism in these systems. The presence of an impurity atom such as C or 0 significantly affects the reactivity of the vanadium cluster, as shown in Figure 4. Moreover, the identity of the impurity atom is quite important in determining the nature of the change in reactivity as compared to the pure metal cluster. For example, the addition of carbon to V, dramatically decreases its reactivity toward DZ,while just the opposite effect is observed for V,C as compared to V9. On the other hand, the corresponding oxides V 5 0 and V 9 0 show reactivities with D, that are quite comparable to those of the parent base metal clusters. Just as has been observed for niobium14and tantalum44clusters, the addition of an oxygen atom to V3 dramatically inhibits its reactivity, while the reactivity of V 8 0 is enhanced relative to v8. Beyond about 11 metal atoms the effect of the impurity atom begins to subside, as its effects are diluted with increasing cluster size. By VI5 there is little difference in D2 chemisorption rate between VI,, V&, and V 1 5 0 . B. N z Chemisorption by V,, V,C, V,O, Nb,, Tan,and Ta,O Clusters. In previous investigations of the chemisorption of N, on Nb,,14 Co,,l4 and TanNa significant cluster size specificity was observed, contrary to what has been observed for chemisorption of the isoelectronic C O m~lecule.'*~,'~ It has been sugge~ted'~ that molecularly chemisorbed CO is bound strongly to the metal clusters, so that the adduct can survive at the estimated 320 K temperature of the reaction tube, while molecularly bound N, desorbs at temperatures much lower than this. Thus, the metal cluster experiments are really probing the rate of dissociative chemisorption of N2, which has a significant cluster size specificity, presumably due to the severe geometrical or electronic requirements that it places on the metal cluster. Figure 5 displays the effective reaction rate, k i , for V,, Nb,, and Tan reacting with Nz, (44) a k i n , M. R.;Cox, D. M.; Kaldor, A. J . Chem. Phys. 1987,87,5046.
6498
The Journal of Physical Chemistry, Vol. 93, No. 17, 1989
,
I
Hamrick and Morse 1
I
I
I
I
i
-4
=
I
I
I
I
1
2
3
4
5
-3L -4
I IOIt
I
1
I
0.05
I
I
0.10
0.15
I
0.20
I
0.25
I
u.ju
N2 flow rate I
I
5
10
I
1
I
15 20 25 Cluster size (n)
l i 30
Figure 5. Reactivity patterns for the reactions of V,, Nb,, and Tanwith
N2. While niobium clusters offer the most dramatic display of cluster size specificity, several similaritiesin the overall reaction patterns of this group of metals are discernible. obtained by fitting depletion plots similar to Figure 2. The pattern of cluster reactivity for V, reacting with N2 is quite similar to that of V, reacting with D2 For small clusters (V4-Vl,) a pronounced even-odd alternation in rate is observed, followed by a gradual increase in rate up to the largest clusters investigated. Minor dips in reactivity at VISand VI9 are observed in reactions with either N2 or Dz.The magnitude of the oscillations in reaction rate is less for N2 chemisorption than for D, chemisorption. N o evidence of V, cluster isomerism was obtained in these studies. For the niobium clusters reacting with N2, marked biexponential depletions were observed for Nb9, NbIl, and Nb12,regardless of whether photoionization was done with ArF (6.42 eV) or F2 (7.90 eV) r a d i a t i ~ n . ~Both ’ sets of data have been independently fit to eq 3.1 for Nb9, N b l l , and Nb,,. The resulting values of cl, c2, kl’, and k2/ are given in Table I. For a given cluster size, the values of all four constants are in reasonable agreement for both sets of photoionization conditions. For these three species the reaction rates of both structural isomers are plotted in Figure 5 . The case of Nb12provides a dramatic example of the effect of structure on reactivity, with the reactivity of the active form at least 100 times larger than that of the inert form. In the context of structural isomerism in Nb,, it should be noted that ionization threshold measurements by the Exxon group2’ show all Nb, clusters to have thresholds defined to within +0.0/-0.10 eV, with the exception of Nb,,, for which the reported uncertainty was +0.0/-0.25eV. Now that structural isomers are known to exist for Nb,,, it is perhaps not surprising that the ionization threshold is less well defined in this system, since each isomer may well have a different vertical ionization threshold. The rate of N, chemisorption on tantalum clusters, Tan, for n I 1 1 shown in Figure 5 is very similar to that previously reported.44 We have, however, extended the range of investigation out to n = 30 by propagating the ionization laser down the molecular beam axis, thereby increasing our detection efficiency for the larger clusters, as described in the beginning of this section. Only Ta12shows evidence of a biexponential depletion with N,. A comparison of the depletion plots of Ta,, + N2 and Nb12+ N2
Figure 6. Logarithmic depletion of the bare metal cluster signals for Nb12 and Ta12reacting with N2, as functions of the N2 flow rate. In these examples, filled circles indicate data points obtained by using ArF radiation (6.42 eV) for photoionization, while open diamonds designate data points obtained by using Fz (7.90 eV) photoionization. Both Nblz
and Ta12show a strong departure from linear behavior. The fitted curves assume a biexponential reaction process, with two different forms of each cluster. Fitted constants are provided in Table I. is provided in Figure 6, and the fitted values of cl, c2, kl’, and k i are given for Ta,2 + N 2 in Table I. Comparing the reactions of all three metals with N2 in Figure 5, it is clear that the reactivity pattern for tantalum clusters reacting with N, is intermediate between the relatively weak size dependence seen with vanadium clusters and the much more dramatic pattern observed with niobium. With the exception of Tas,45the reactivity patterns for Tan and Nb, are similar for n ranging from 3 to 15, with reactivity minima at n = 8, 10, 12, 14, and 19 in both species and with both species showing structural isomerism for n = 12. Considering that the range of reaction rates k,,‘ is compressed for tantalum relative to niobium, it is possible that structural isomers exist for Tag and Tal, as well but are not sufficiently different in reactivity with N2 for unambiguous detection. Finally, the most glaring discrepancy among the three metals is that the inert nature of Nb16finds no correspondence in either VI6 or Tal,. Figure 7 displays the reaction rates, k,,’, for V,, V,C, and V,O reacting with N,. As was the case for these species reacting with D2, the impurity atom is found to significantly alter the chemical reactivity. The reactivity pattern for V,C + N2 shows a striking similarity to that for V,C + D,, as shown in Figure 4. In both cases reactivity minima occur for VsC, V8C, and VloC, with maxima at V7C and V9C. The reactivity patterns of V,O + D2 and V,O N, also show some similarities with minima at V60, V,O-V,oO, and Vl,O-Vl,O. As was found for the V,, V,C, and V,O clusters reacting with D2, above n = 15 the effect of the impurity atom is much reduced, as its concentration is increasingly
+
(45) It should be noted that both Ta5and Ta6show asymmetric broadening in the TOFMS, indicating that a significant contribution for these clusters arises from photofragmentation of larger species. (46) Cotton, F. A,; Wilkinson, G . Aduanced Inorganic Chemistry: A Comprehensiue Texr, 3rd ed.; Interscience: New York, 1972; p 923. (47) McElvany, S. W.; Creasy, W. R.; OKeefe. A. J . Chem. Phys. 1986, 85, 632.
The Journal of Physical Chemistry, Vol. 93, No. 17, 1989 6499
Reactions of V, Nb, and Ta Clusters
IO', -
---
1
vn + C2H6 *
-
I 1
18'
1
;
I
Ih
I I
1;
I
,
20
I
1
I
I
215
Cluster size (n)
11
-Tan *--.Tan0
*.
. . 5
-
IO
15
20
Cluster size (n)
j
Figure 9. The chemisorption of ethane on V, and Nb, shows no dramatic size dependence but only a gradual increase in reactivity with cluster size. At the highest flow rate of a 5.0% ethane/95.0%helium mixture tantalum clusters showed only a minimal depletion,so the trend in reactivity with ethane is V, > Nb, > Tan.
3b
Figure 7. The effect of adatom modification of cluster reactivity is shown here for V,, V,C, and V,O reacting with N1 The largest effects are seen for small clusters (n = 3-10). Note, for example, the large drop in reactivity when a carbon atom is added to V5.
1
10'
1
. . . ... --
r
-
---
+ Nz t N2
Cluster size (n)
Figure 8. An extended view of the effect of adatom modification of cluster reactivity is developed here for Tan and Ta,O reacting with N2. The effect of the oxygen atom gradually wanes, and the reactivity of Ta,O closely parallels that of Tan for n = 20-30.
diluted among the more abundant metal atoms in the cluster. Figure 8 presents a comparison of the reactivity of Tanand Ta,O toward NZ,in the spirit of previous work performed by the Exxon group.M In this study we have extended the range of Tan and Ta,O clusters studied from 3 In I11 to 3 In I30, in an effort to observe the convergence to the infinite dilution limit. The relative reactivities of Tan and Ta,O, measured for the small clusters, n 5 11, are in good agreement with previous work,44while the steady convergence to the infinite dilution limit continues with no major deviations up to n = 30. Beyond n = 13 the difference between Tan and Ta,O reactivities is no more than a factor of 1.5, while for n I 25 essentially no difference in reactivity is observed. C. C2H6Chemisorption by V,, Nb,, Tan,Con,and Ni, Clusters. In contrast to the dramatic dependence of reactivity on cluster size observed for the above-listed metal clusters reacting with D2 and N2, the V, and Nb, clusters show only a monotonic increase in reactivity with ethane as a function of cluster size. Figure 9 displays the fitted values of k,,','as determined for V, and Nb,
reacting with ethane. The nonselective pattern of reactivity is reminiscent of that obtained for most transition-metal clusters reacting with COS2Tantalum clusters appear to just begin to react with ethane at the highest flow rates of a 5% C2H6/He mixture, where minimal depletions are observed. The obvious trend in reactivity with ethane is V, > Nb, > Tan. Ethane was also found to react slowly with Con and Ni, clusters. In all cases where reaction occurred, weak product features were observed in the mass spectrum, which was obtained by photoionization with low-fluence (5-50 pJ/cm2) ArF (6.4 eV) radiation. The degree of dehydrogenation of C2H6for V, and Nb, was not a strong function of cluster size, with all cluster sizes showing M,C, M,CH, M,C2, M,C2H, and M,C2H, products. Of these products, the carbide and dicarbide were the most intense in the mass spectrum. Of course, some posibility of thermally induced dehydrogenation resulting from absorption of one or more ionization laser photons exists:* and this may partially explain the formation of unexpected products such as M,CH and M,C2H. Since M,CH4 has the same mass as M,O, we cannot determine whether a product of this composition is formed, nor can we determine whether the cluster oxides are reactive. D. Consecutive Addition Reactions of Metal Clusters. In the mass spectra which provide the raw data for the kinetic measurements discussed here, products are frequently observed which correspond to the addition of a second reactant molecule to the metal cluster. This indicates that consecutive addition reactions, such as
occur under the experimental conditions present in the fast-flow reactor. When such consecutive reactions occur, one would expect an initial increase in the observed concentration of M,N, followed by an exponential decay. Assuming that the concentration of the reactant gas, [N,], is constant during the time spent in the reaction tube, one may integrate the rate equations for reactions 3.2 to obtain
Converting the rate constants k , and k2 to effective rate constants k,' and k2/ by the relationships k i [ N 2 ] t= ki'FN2 and k 2 [ N 2 ] t= k i F N 2 and including the possibility of differing ionization cross sections (48) St. Pierre, R. J.; Chronister, E. L.; El-Sayed, M. A. J . Phys. Chem. 1987, 91, 5228.
6500
The Journal of Physical Chemistry, Vol. 93, No. 17, 1989
1
1.0
TABLE 11: Seauential Additions of N, to Nb.“
b
system Nb, + N, NbS + N2 Nb, + N2 Nbg + N,b
.-6 0.4 U
!L
0.2 0.0 1)
Hamrick and Morse
I
7
3
4
2 ? llou KJIC (sccm)
Figure 10. In addition to monitoring the depletion of bare metal clusters, the growth and decay of adducts, such as NbSN2,can be monitored as well. This enables subsequent reaction rates to be obtained, along with estimates of the ratio of photoionization cross sections for M, and M,N2. For this example, the measured data points for [NbS]/[Nbs],are given by open circles, and measured data points for [NbsN2]/[Nbsl0 are given by filled diamonds. Solid curves are least-squares fits to the data, described by the parameters of Table 11.
for M,N, and M,, we obtain the observed ratio of [M,N,],/[M,], as
where u(MnN2)and u(M,) are the photoionization cross sections of M,N2 and M,, respectively. Figure 10 displays the measured ratios [Nb,]/[Nb5l0 and [Nb5N2]/[Nb5],as functions of the N2 flow rate, FNz.The former ratio, [NbS]/[NbSlo,exhibits a pure exponential decay, as is also shown in the semilog plot of Figure 2. The ratio [Nb5N2]/[Nb5l0, on the other hand, shows a rapid rise followed by a slower decay, which is precisely the functional form predicted by eq 3.4. Since kl’ is independently determined by the decay curve of [Nb5]/ [Nb,],, it is constrained to that independently determined value, and the values of k i and the preexponential factor (u(M,,N2)/ u(Mn))(kl’ - k i ) ) are obtained by a least-squares fit of [NbSN2]/[Nb5],to the functional form of eq 3.4. The resulting fit, given by the solid curve in Figure 10, provides the values of k2/ and the photoionization cross-section ratio (a(M,N,)/a(M,)). We have performed similar analyses for the consecutive addition of N, molecules to Nb3, Nb5, Nb6, and Nb,. For the first three cases, Nb3, Nb5, and Nb6, addition of one N, molecule proceeds quite rapidly, and the second N, is added at a much slower rate, leading to strong Nb3N2,NbSN2,and Nb6N2peaks in the mass spectrum. In the case of Nbs, however, the first addition of an N 2 molecule is rate-limiting, and the addition of a second N, proceeds much more rapidly. As a result, k l ’ / ( k l f- k i ) is small and negative, and the NbsN2 feature is not very intense relative to Nb, under any conditions. For this case (kl’ < k2’),the exponential rise time for M,N2 corresponds to the constant k2/, but this is not easily measured since the overall intensity of M,N,is small. Fitted rate constants and photoionization cross-section ratios for the Nb3, Nb,, Nb6, and Nb, clusters reacting with N2 are given in Table 11.
IV. Conclusions A major goal of metal cluster reaction studies has been to characterize the reactivity of various metal cluster-ligand combinations, in the hope that ultimately a model of chemisorption may be constructed that can predict and explain the diverse array of chemical interactions on clusters and metal surfaces. Classifying the unique chemistry of metal clusters in the context of periodic properties is one fragment of information necessary for the construction of this universal model of chemisorption. The focus of this paper has been to delineate the differences and similarities in the reactivity of the clusters of the group 5 transition-metal family.
C
k,”
0.631 0.474 0.565 -0.134
11.40 16.54 11.09 0.1015
k2“ 1.35 0.697 0.274 1.811
k,’/k,‘ 0.119 0.042 0.025 17.84
~(NbnNd/ u(Nb,) 0.556 0.455 0.551 2.25b
’As described in the text, k2’ is obtained by fitting the depletion of the bare Nb, cluster, and this value is used in the fit of the growth and decay of the Nb,N2 adduct, which is fit to the formula [Nb,N2]/ [Nb,], = C[exp(-kiFNz)- exp(-kl’FN2)]. According to eq 3.4, C is given by C = (u(Nb,N2)/u(Nb,))(kl’/(kl’ - k i ) ) , from which we derive the ratio of photoionization cross sections a(Nb,N,)/u(Nb,). For these studies ionization was accomplished by low-intensity ArF (6.42 eV, 193 nm) radiation. the case of Nbg, k; is poorly determined because the addition of N2 to Nb8N2is much faster than the primary addition of N2 to Nbg. As a result, the concentration of NbgN2is never very large, and the uncertainty in k; and u(NbgN2)/u(Nb,) is quite large. This may account for the anomalous value of u(NbgN2)/u(Nbg)as compared to the other photoionization cross section ratios. ‘The effective rate constants k,’ and k; are given in units of sccm-I.
Although one-to-one correspondences in the reactivity patterns of these metals are not realized, several similarities are noted. The similarities are most obvious for Nb, and Tan reacting with N2 (see Figure 5). Cluster sizes 4, 8, 10, and to some extent 12 are relatively unreactive for both metals. Even minor dips in reactivity for cluster sizes 14 and 19 are present for both metals. Differences between the metals show up in Ta5,45for which the high reaction rate seems out of place, and in NbI6, which is uniquely unreactive in the size range 13-20 and which finds no correspondence in VI6 or Ta16. Although both V, and VI, are relatively inert, the reactivity pattern for vanadium clusters differs substantially from that of its heavier congeners. Unlike Nb, and Tan, the least reactive vanadium clusters are V4 and v6, and size-selective behavior is much suppressed, particularly above VI4. This should perhaps not come as a surprise, since it is generally true that the chemistries of second- and third-row transition metals are often quite similar but differ markedly from that of their first-row
counterpart^.^^ The effect of an impurity carbon or oxygen atom is now well-established for vanadium clusters. It is clear, at least for reactions with N2and D2, that the presence of an impurity carbon or oxygen atom modifies the reactivity of the parent V, cluster in a different way. In view of the electronegativity difference between carbon and oxygen it is not surprising that these impurities should affect the cluster reactivity in different ways. This tends to conflict with an earlier view4 that the impurity atom preserved the packing arrangement and that consequently the reactivity of M,X should track the reactivity of Mn+I. It is also notable that the presence of an impurity atom is indeed a local effect that abates with increasing cluster size, so that the reactivity approaches that of the bare metal cluster as cluster size is increased. This is most obvious in the comparison of the reactivities of Tan and Ta,O with N2, as shown in Figure 8. The observation of biexponential depletion kinetics for Nb9, Nb,,, NbI2,and Talz supports the hypothesis that these species possess structural isomers that differ in their reactivity with D2 or N2, or both. Structural isomerism has been previously reported in Nb9 and Nb12,35in Nbll+and Nb19+,31 and in C7+.47We have carefully searched for evidence of structural isomerism in V, and Con, without success. At this time all of the known clusters exhibiting structural isomerism correlate with materials that are extremely refractory in the bulk solid phase. On the basis of this correlation, we believe that lower melting point metals are less likely to be trapped in metastable configurations, or local minima on the multidimensional potential energy hypersurface, than are the more refractory metals. Presumably the cooling rate in the cluster formation region is insufficient to trap the less refractory metals in metastable configurations, or possibly these species may remain fluxional even down to the temperature of our fast-flow reactor, which is estimated at 320 K.I3
J. Phys. Chem. 1989, 93, 6501-6506 Finally, we note that it is possible to kinetically monitor the consecutive addition of ligands to metal clusters. We have investigated the addition of a second N2 molecule to Nb3N2,Nb5N2, Nb6N2, and Nb8N2 and have found that this secondary reaction proceeds much more slOwly than the first addition Of an N2 for Nb3, Nb5, and Nb6 but proceeds more rapidly for Nb8.
Acknowledgment. We are grateful for support of this research from the Science under Grant 8521050. Registry NO. H2, 1333-74-0; N2, 7727-37-9; C2H6, 74-84-0; V3C, 12070-1 1-0; V4C, 121674-52-0; VSC, 121674-53-1; V&, 12316-57-3; V7C, 55892-24-5; V&, 121674-54-2; VgC, 121674-55-3; VI&, 12167456-4; VjiC, 121674-57-5; VIZC, 121674-58-6; V,,C, 121674-59-7; VIdC,
6501
121674-60-0; V&, 121674-61-1; VI&, 121674-62-2; Vi7C, 12167463-3; vi&, 121674-64-4; V& 121674-65-5; v30, 64764-10-9; v40, 12306-38-6; VsO, 121674-40-6; V.50, 39350-97-5; V70, 121674-41-7; v@, 54651-68-2; v@, 12509-78-3; VloO, 121674-42-8; V l l O , 12167443-9; V120, 121674-44-0; Vl,O, 121674-45-1; VI40, 121674-46-2; VI50, 121674-47-3; V,,O, 121674-48-4; V,,O, 121674-49-5; V,,O, 12167450-8; VI@, 121674-51-9; T a 3 0 , 112510-43-7; T a 4 0 , 12059-92-6; T a 5 0 , 112510-44-8; Tab03 1 2 2 0 2 - 0 5 4 Ta@, 112510-45-9; TasO, 112510-46-0; T a 9 0 , 112510-47-1; TaioO, 112510-48-2; T a l l O , 112510-49-3; T a 1 2 0 , 71767-52-7; Tal@, 121674-22-4; T a i 4 0 , 121674-23-5; Tal@, 12167424-6; T a 1 6 0 , 121674-25-7; T a i 7 0 , 121674-26-8; TaI8O, 121674-27-9; T a i 9 0 , 121674-28-0; Ta200, 121674-29-1; T a 2 1 0 , 121674-30-4; Ta2,0, 121674-31-5; Ta230, 121674-32-6; Ta,,O, 121674-33-7; T a 2 5 0 , 121674-34-8; Ta2,0, 121674-35-9; T a 2 7 0 , 121674-36-0; T a 2 8 0 , 121674-37-1; Ta290, 121674-38-2; Ta300, 121674-39-3.
Genesis and Characterization by Laser Raman Spectroscopy and High-Resolution Electron Microscopy of Supported MoS, Crystallites Edmond Payen,+,$Slavik Kasztelan,t*s Sabine Houssenbay,+ Raymond Szymanski,g and Jean Grimblot*,+ Laboratoire de Catalyse HPtCrogPne et HomogZne, U.A. C.N.R.S.402, UniversitP des Sciences et Techniques de Lille Flandres- Artois, 59655 Villeneuve D'Ascq Cedex, France, L.A.S.I.R.,L.P. 2641, UniversitP des Sciences et Techniques de Lille Flandres-Artois, 59655 Villeneuve D'Ascq Cedex, France, and Institut Francais du PPtrole, 1 et 4, Avenue de Bois-PrPau, B.P. 311, 92506 Rueil-Malmaison Cedex, France (Received: January 18, 1989)
Sulfidation of Mo/y-Alz03based hydrotreating catalyst is a key step to activate the supported phase. In situ laser Raman spectroscopy (LRS) appears as a suitable technique for identifying intermediate species and characterization of the final state, Le., MoS, crystallites supported on A1203. A stepwise sulfiding procedure was used in varying different parameters, Le., hydration state of the oxide precursor, composition of the sulfiding mixture Hz/H2Sor N2/H2S,temperature, and time duration. The important role of these experimental conditions on the detection of transient species during sulfidation has been pointed out. Different intermediates such as oxysulfides, sulfido compounds, and MoS3 appear to be the dominant ones, whose existence depends on the previously mentioned parameters. The crystallite size of the MoSz, a function of the Mo loading and determined by HREM, influences the position and width of the -385-cm-' line.
Introduction The sulfidation of the Mo03/y-A1203or WO3/y-Al2O3hydrotreating catalysts, which are generally associated with a promoter, is the necessary step, usually performed directly in the refinery plant, to get the active working form of the catalysts. This sulfidation step produces supported MoS2 or WS2 particles which have been identified by several techniques.' However, a good knowledge of the nature and reactivity of the intermediate species formed during the complex transformation of the oxomolybdate entities into MoSz on the support surface is of considerable interest in order to optimize the catalytic properties of this active MoS2 phase. In situ LRS appears a suitable technique for such an investigation as it permits to detect both oxo and sulfido species. To date a large amount of results has been obtained on the oxide precursor2 whereas sulfidation has been less studied. Schrader and Cheng3 observed different surface species during the sulfidation of MoO3/y-AI2O3catalyst, Le., oxysulfides, a reduced phase and MoS2. Cobalt present as a promoter favors reduction of the oxomolybdate phase but leads to a more difficult s ~ l f i d a t i o n . ~ Since the work of Arnoldy et aL5 it appears that water, present in the hydrated precursor, influences the sulfidation rate. In a previous study6 we reported LRS observations on W 0 3 supported catalysts evolution when changing the nature of the sulfiding mixture, temperature, and time of reaction as well as
* To whom correspondence should be addressed.
' Laboratoire de Catalyse HBtBroggne et HomogLne. 'L.A.s.I.R. 5 Institut FranGais du Petrole.
0022-3654/89/2093-6501$0l.50/0
-
- -
the effect of oxide hydration. The following scheme was proposed: supported oxotungstate oxysulfide WS3 WS2. In this work, to complement this previous investigation, the nature of the intermediates formed during the sulfidation steps of molybdate/A1203based catalysts has been investigated by using in situ laser Raman spectroscopy (LRS). Several samples of different Mo loadings have been studied to monitor the effect of this parameter on the size of the MoS2 slabs after sulfidation. Indeed, both LRS and HREM (high-resolution electron microscopy) allow us to observe features of the MoSz final product.
Experimental Section Catalysts and Pretreatment. The catalysts studied in this work have been prepared by the pore filling method using y-A1203 extrudate (surface area, 238 m2.g-'; pore volume 0.6 g ~ m - and ~) heptamolybdate solutions. The solid was then dried at 380 K (1) Massoth, F. E. Advunces in Cutalysis; Academic Press: New York, 1978; Vol. 25, p 265. Gates, B. C.; Katzer, J. R.; Schuit, G. C. A. Chemisfry of Catalytic Processes; McGraw-Hill: New York, 1979; Chapter 5. (2) (a) Payen, E.; Kasztelan, S.; Grimblot, J.; Bonnelle, J. P. J . Rumun Spectrosc. 1986, 17, 233. (b) Stencel, J. M.; Makowsky, L. E.; Diehl, J. R.; Sarkus, T. A. J . Roman Spectrosc. 1984, 15, 282. (c) Jeziorowski, H.; Knozinger, H. J . Phys. Chem. 1979,83, 1166 and references therein for earlier literature. (d) Zing, D. S.; Makowski, L. E.; Tisher, R. E.; Brown, F. R.; Hercules, D. M. J . Phys. Chem. 1980, 84, 2898. (3) Schrader, G. L.; Cheng, C. P. J. Catul. 1983, 80, 369. (4) Schrader, G. L.; Cheng, C. P. J. Cutal. 1984, 85, 488. (5) Arnoldy, P.; Van Den Heijkant, J. A. M.; De Bok, G. D.; Moulijn, J. A. J. Cutal. 1985, 92, 3 5 . (6) Payen, E.; Kasztelan, S.; Grimblot, J.; Bonnelle, J. P. Catul. Today 1988,4, 57. Payen, E.; Kasztelan, S.; Grimblot, J. J . Mol. Strucf. 1988, 71, 174.
0 1989 American Chemical Society
