Photodissociation of Niobium-Carbon Clusters and Nanocrystals - The

Click to increase image size Free first page .... Hugh Harris and Ian Dance ..... Now You Can Watch Cellular Respiration with a Novel Nanoelectrode Pr...
0 downloads 0 Views 839KB Size
544

J. Phys. Chem. 1995, 99, 544-550

Photodissociation of Niobium- Carbon Clusters and Nanocrystals J. S. Pilgrim, L. R. Brock, and M. A. Duncan* Department of Chemistry, University of Georgia, Athens, Georgia 30602 Received: July 27, 1994; In Final Form: November 4, 1994@

Metal compound clusters containing niobium and carbon are produced in a laser vaporization pulsed-nozzle cluster source. The mass distributions of cations produced from this source are measured under different conditions. These cations are mass-selected in a specially designed reflectron time-of-flight mass spectrometer and photodissociated at various laser wavelengths. Mass distributions provide evidence for the preferential formation of the M8C1Z stoichiometry observed previously and associated with the “met-cars” cage structure. Additional abundant masses indicate the formation of face-centered-cubic crystallite stoichiometries. Photodissociation in both the met-cars and crystallite systems results in the formation of a broad distribution of fragment ions, with some enhancement in abundance for the smaller symmetric crystallites. This photochemistry is markedly different from that observed previously for metal-carbon clusters containing titanium, vanadium, molybdenum, or zirconium.

Introduction

Transition metal-carbon clusters exhibit a diversity of chemical bonding properties and a concomitant variety of structural motifs. At least three structural themes are now recognized for these systems: metallofullerenes,l-s metallocarbohedrenes ( “ m e t - c a r ~ ” ) , ~and - ~ ~cubic crystallite^.*^^^^-^^ These various structural patterns are linked in an intimate way with the electronic, optical, and magnetic properties of these clusters. There is exciting potential for these systems in the synthesis of “nanoscale” materials, but it is essential to understand the fundamental relationships between their geometries, bonding, and electronic structures. However, it is not yet clear which factors govern the bonding format found for a particular metal-carbon system. Indeed, some systems exhibit different apparent structures under different growth conditions. It is therefore important to develop a body of data for a variety of metal-carbon systems with which to define trends, if any, in their structures and bonding. One of the most significant observations arising from previous beam experiments is that there may be a competition between the formation of crystallites and met-cars clusters for certain metals. In the present report, we provide further investigation of this phenomenon in the niobium-carbon system. Castleman and co-workers were the first to report the growth of the unusual met-cars cluster stoichiometry M8C12?-21 In their early work, this stoichiometry was observed to be formed preferentially from laser-generated metal-carbon plasmas containing titanium, vanadium, zironium, or hafnium. More recently, the Castleman group has also observed the 8/12 stoichiometry for niobium-carbon clustersz1 and for mixed metal-carbon systems.18 An extremely stable cage structure with dodecahedral symmetry was proposed to explain this preferred stoichiometry, although theory has now suggested several other possible ~ t r u c t u r e s . ~In~ the - ~ ~case of zirconium, larger stoichiometries were suggested to adopt multiple cage structures.” Certain other metals (tantalum) were shown to prefer the growth of cubic crystallite carbide structures.’l Similar cubic crystallites were noted for metal nitrides and oxides.I3 Further work by Pilgrim and Duncan established that the same 8/12 M/C met-cars stoichiometry arises for iron,

* To whom correspondence should be addressed. ‘Abstract published in Advance ACS Absrracrs, January 1, 1995.

chromium, and molybdenum systems.23 However, perhaps most surprising, we also observed the formation of both the metcars stoichiometry and the cubic crystallite stoichiometry for titanium, vanadium, and zirconium cluster^.^^-^^ In particular, the 3 x 3 x 3 cubic crystal fragment, which has the stoichiometry M14C13, was found to be produced with the same efficiency as the M&l2 species. Moreover, photodissociation experiments indicated that this and other cubic crystallites, also called “nanocrystals”, decompose preferentially to produce the 8/12 met-cars s t o i c h i ~ m e t r y . ~These ~ - ~ ~experiments confirmed the strong stability of the met-cars clusters and also suggested that there is a structural relationship between the met-cars system and cubic crystals. In the present work, we focus on niobium-carbon clusters. As shown below, an extensive distribution of metal-carbon clusters can be produced in this system. However, the tendency for “magic number” stoichiometries is not as pronounced as it is for other metal-carbon mixtures. Castleman and co-workers have also investigated the mass distributions in the niobium system.21 They find that variation in the source conditions can dramatically influence the mass distributions observed, so that either the 8/12 cluster of the crystallite stoichiometries can be enhanced with a different choice of conditions. This behavior in cluster growth is unlike that observed in other metal-carbon systems, in which the mass distributions are relatively insensitive to the source conditions. The present photodissociationexperiments also indicate that niobium behaves differently than previously studied metals. The extent of photodissociation and the specific channels observed are significantly different from those observed for other systems. These observations suggest that the tendency for met-cars cage formation is reduced in this system and provide additional data delineating the region of the periodic table where this cluster pattern is prominent. Experimental Section

The clusters for these experiments are produced by laser vaporization in a pulsed-nozzle source. A rotating rod of niobium is vaporized with the focused output of a XeCl excimer laser (Lumonics, 308 nm) or the second harmonic of a Nd: YAG laser (Spectra Physics DCR-11,532 nm). The expansion gas for the experiment is helium seeded with 1% methane or acetylene. Metal-carbon cation clusters formed in the con-

0022-3654/95/2099-0544$09.00/0 0 1995 American Chemical Society

Photodissociation of Niobium-Carbon Clusters

J. Phys. Chem., Vol. 99, No. 2, 1995 545

Nb,CL

9.9

4

611

500

1000

I

a111l

15’00

Mass ( a m u ) Figure 1. Mass distribution of Nb/C cations formed from laser vaporization of a niobium rod with a helium expansion gas seeded with 1% methane and a Nd:YAG vaporization laser (532 nm). A relative maximum in intensity is observed for the 8/12 cluster and the 14/13 cluster, but many other masses are also observed with only slightly lower intensities. There are “clumps” of masses around each metal atom increment, and a significant drop in intensity beyond the N b l d C ~ + mass region. The relative intensities in this mass distribution are not strongly dependent on the vaporization laser power.

densation of the laser generated plasma are cooled by a supersonic expansion and sampled through a skimmer into a specially designed reflectron time-of-flight mass spectrometer. In the mass spectrometer, the cation clusters are pulse extracted from the molecular beam and undergo the full flight through the reflectron for the measurement of the distribution of cations formed by the source. For photodissociation experiments, these cations are mass-selected one at a time by their time-of-flight through an initial drift tube section. Photoexcitation takes place at the tuming point in the reflectron field, where a pulsed Nd: YAG laser interacts the cluster ion trajectory. The photodissociation wavelength is either 532 or 355 nm. Parent ions and resulting photofragment ions are mass-analyzed by their timeof-flight through a second drift tube section. A computer difference technique (photodissociation laser “on” minus “off”) is used to record photodissociation mass spectra. In these spectra, the mass-selected parent ion is shown as a negativegoing peak, indicating its depletion by the photodissociation. The photofragments are indicated as positive-going peaks. The complete details of this mass-selected photodissociation experiment are described e l ~ e w h e r e . ~ ~ Results and Discussion Cluster Growth. Figure 1 shows the mass distribution of niobium-carbon cluster cations produced by our source when using a 532 nm Nd:YAG laser for vaporization. Clusters are formed out to a total mass of about 1600 amu. While a number of different cluster stoichiometriesare observed, the masses seen do not correspond to all the statistically possible stoichiometries. Instead, the cluster masses appear as a series of “clumps” of mass peaks, with stoichiometries clustered about certain preferred ratios of the elements. There are no dramatic “magic numbers” in this distribution, but rather small local maxima occur near stoichiometries associated with each additional Nb atom mass. Because the mass of niobium is much greater than that of carbon the mass peaks are grouped about the niobium atom increments, but the number of carbon atoms is usually equal to, or greater than, the number of metal atoms. In this sense, this spectrum is dramatically different from those observed for other metal-carbon systems, where striking magic numbers appear at only a few precise stoichiometries. The most

doe,

14,13

,

Mass ( a m u )

zoo0

Figure 2. Mass distribution of Nb/C cations formed with the same conditions as in Figure 1, except that the vaporization laser is an excimer laser at 308 nm. The power level is approximately the same as that used in Figure 1. The prominent masses are now Nb/C = 4/4, 5/3, 6/7,9/9, and 14/13. As discussed in the text, all of these stoichiometries except 6/7 can be rationalized in terms of fcc crystallite fragment structures.

notable of the magic numbers observed previously are the M8Cl2 cluster, denoted as a metallocarbohedrene, or “met-cars” cluster by Castleman and co-workers, and the M14C13 “nanocrystal” cluster discussed previously by our laboratory. Both of these previously seen 8/12 and 14/13 stoichiometries are noticeable in the present distribution, but neither is as dramatically abundant as the peaks in other metal-carbon systems. The 8/12 peak is the third member of a triplet of peaks, after which there is a steep drop in intensity. The 14/13 peak is also one of several peaks in a multiplet, after which there is a sharp drop in intensity. Other noticeably intense peaks are found here for the M/C stoichiometries 4/4, 6/7, 7/9, and 10/12. Surprisingly, the distribution of cluster cation masses is changed when vaporization is accomplished with the 308 nm excimer laser. Figure 2 shows the mass distribution observed for this different wavelength when the total power of the vaporization laser is about the same as it was above. As in Figure 1, there are no dramatic magic number cation clusters. However, while the overall signal is about the same, some of the relative intensities in the spectrum have changed. Most noticeably, the 8/12 mass is significantly reduced in intensity, as is the 7/9 mass. Both of these mass peaks are now only one of many small features (nearly every possible M/C ratio is represented) in this region of the spectrum. On the other hand, the 5/3,9/9, and 14/13 stoichiometries are significantly greater in intensity than they were in Figure 1. In particular, the 9/9 stoichiometry has replaced the 8/12, 8/11, and 8/10 peaks, becoming prominent in the central region of the mass spectrum. The 14/13 peak is still one of a group of peaks with similar intensity, but the group has increased by about a factor of 2 relative to peaks following it. Castleman and co-workers have recently also studied this niobium-carbon system and found that the distribution of clusters can be shifted dramatically to emphasize different magic number peaks by adjustment of the relative concentration of the feed gas.21 The data in Figures 1 and 2 show that a similar effect can be caused by different vaporization laser wavelengths. Both experiments are understandable if there is no strong thermodynamic preference for the formation of any particular clusters based on an especially stable bonding configuration. If there are no strong thermodynamic effects, cluster growth will be affected more by kinetic factors such as the concentration

Pilgrim et al.

546 J. Phys. Chem., Vol. 99, No. 2, 1995

TABLE 1: Major Photofragments of NbNC#+ Parent Ions at 532 ma Fragments Formed (Columns) vs Parent Ion (Rows) 110 212 313 414 516 617 719 819 919 loll0 11/13 12/12 13/13

X X 617 X 719 X 8/12 X 919 X 1 0 / 1 o 11/13X 1 2 1 1 2 13/13 X 14/13 X

414 516

4/4

9/9

14/13

Figure 3. Idealized 4/4,9/9, and 14/13 crystallite fragment structures suggested by the mass distributions in Figure 2. The solid balls represent carbon, while the open balls represent niobium.

of the feed gas or the temperature in the growth region. The latter factor could explain in a qualitative way the different mass spectra observed for different vaporization laser wavelengths. The temperature in the growth region is determined by the cumulative effect of the vaporization efficiency and the absorption efficiency of the laser-generatedplasma, both of which may vary with the wavelength. Significant vaporization and initial plasma formation occur within less than a nanosecond on the leading edge of the laser pulse time profile, and the latter portion of the laser pulse may lead to heating of the plasma. This heating is expected to vary with the vaporization wavelength because of the absorption spectra of the different components in the plasma and because of the different energy content of photons at different wavelengths. The heating may also be expected to vary with the laser power, but the spectra shown here are not very sensitive to this variable, presumably because the absorption probability at both wavelengths is saturated. In the case of more stable clusters, such as the TigC12 species described previously, the cluster growth is more or less insensitive to the reactant concentrations and laser vaporization wavelength. The kinetic effects of the growth conditions appear to be outweighed by the thermodynamics of especially strong bond formation for certain special stoichiometries in these systems. On the other hand, the sensitivity to growth conditions observed here and in the Castleman work suggests that the niobium-carbon clusters do not have such large relative differences in their bonding for different stoichiometries. While the prominent clusters observed for the niobium system are not as dramatic as they are for other metal-carbide systems, they are interesting. The 8/12 stoichiometry is recognizable as the met-car cluster seen in other M/C systems. Likewise, the 14/13 stoichiometry is that associated previously with the 3 x 3 x 3 face-centered-cubic (fcc) crystallite structure. However, stoichiometries such as 414, 513,617, 719, and 919 have not been reported previously as prominent clusters. Of these, the 414 and 919 stoichiometries most likely represent the 2 x 2 x 2 and 2 x 3 x 3 fcc crystal fragments. These structures are shown in Figure 3 together with the 14/13 crystallite. The 513 species could be rationalized as an off-stoichiometry 2 x 2 x 2 crystallite, with one niobium atom substituted for a carbon atom. However, the 6/7 and 719 species, which also appear as relatively abundant masses, do not correspond to any simple geometric structure that we know. Photodissociation Experiments. Photodissociation experiments are another qualitative method for the identification of

X 314 X X X X X X X X X X x x x X X x x x X X X X

X X X X X X X X

X X X

x x x x x x X X X X 0 x x x x x 0

0 0 X X X X X X 10112 X X X 7/n 8/n X 1O/n

X

0

The parent ion is given in the row label. Major photofragments are indicated by an X in the stoichiometry given by the column label. An 0 indicates that the stoichiometry given by the column label is correct for the number of Nb atoms in the photofragment but that the number of carbons cannot be determined exactly. A number gives the stoichiometry of the photofragment that is closest to that given by the column label. Unresolved peaks are indicated with n.

250 Mass ( a m u )

500

Figure 4. Photodissociation mass spectrum of Nb&+ at 355 nm. There are fragments correspondingto the loss of neutral niobium atoms with various numbers of carbon atoms. There is a small, but reproducible, channel for the loss of a single carbon atom without accompanying metal (the 4/3 fragment). There is a slight local preference for the 2/2 fragment. The Nb+ fragment ion signal is offscale in the figure.

relatively stable clusters and/or patterns in cluster growth. We have therefore investigated the photodissociation behavior of a variety of the cluster cations shown here using the photodissociation wavelengths of 532 and 355 nm (Nd:YAG second and third harmonics). The fragment ions measured in these experiments are shown in Table 1. Selected fragmentationmass spectra are shown in Figures 4-7 for the parent ions 414, 919, 8/12, and 14/13. Figure 4 shows the photodissociation mass spectrum for the cluster Nb4C4+ at 355 nm. Essentially the same spectrum is observed at a photodissociation wavelength of 532 nm. The laser power for these experiments is about 20 mJ/cm2, and the spectra are believed to be the result of multiphoton absorption. At lower power, the overall signal is weaker, but the relative ratios in the fragment channels do not change significantly (see discussion below). The Nb+ fragment is by far the largest observed, and it is shown offscale in the figure. Additionally, essentially every possible combination of metal-carbon masses is observed as molecular fragments. The exceptions to this are that there are no fragments observed at Nb4C+, Nb4c2+, NbC3+, and NbC4+. Likewise, the other possible fragments with an excess of carbon relative to niobium (e.g., 213 and 214) appear with only weak intensity. The photodissociation method used here collects all the charged fragments resulting from multiphoton dissociation.

J. Phys. Chem., Vol. 99,No. 2, 1995 547

Photodissociation of Niobium-Carbon Clusters

I

500

1000

Mass ( a m u ) Figure 5. Photodissociationmass spectrum of NbSCg+ at 355 nm. As shown in the multiplet of depletionpeaks in the parent ion region, there is incomplete mass selection of the parent ion due to the limited resolution of the mass gate. "here are essentially no charged fragments observed larger than those in the 4/4 region. As in Figure 4, the Nb+ mass peak is offscale.

Thus, the photodissociation mechanism producing the fragment ions observed may be a series of consecutive events, a group of concerted parallel reactions, or a convolution of both kinds of processes. The latter mixed mechanism is perhaps the most likely. Therefore, the measurement of fragment ions does not allow us to determine the exact mechanism of the photochemistry. However, it is generally accepted that the thermodynamics of the bonding in the cluster cations is a significant factor in these kinds of photodissociation processes. Therefore, we look at the fragments formed from a variety of parent ions. If certain fragment ions are formed with enhanced intensity from many different parent ions, there is good circumstantial evidence that these fragment species are relatively more stable than others. In the present spectrum of Nb4C4+, the only prominent ion is Nb+. This is easy to understand if the metal-carbide clusters have higher ionization potentials (IPS) than the niobium atom. If this is true, and if the clusters fragment by loss of niobium, the charge will be on this atom, and the remaining carboncontaining cofragment, which has a higher IP, will be neutral. Whether fragmentation occurs in a sequence of events or as several concerted parallel channels, the average end result will be a large Nb+ fragment, which we observe. However, there is also another mechanism which would also produce Nb+. If neutral Nb atoms are produced as photofragments, these atoms could reabsorb light in the strong laser field and become ionized. Studies of the laser power of the fragmentationchannels indicate that this latter process is not likely to be very efficient, but we cannot rule it out completely as a contributor to the signal in the Nb+ channel. In the molecular fragment ions, the only species showing any enhancement at all is perhaps Nb&+. If we assume that the parent Nb&+ is indeed a 2 x 2 x 2 fcc species, then it is tempting to conclude that the 2/2 species represents half of this symmetric cube. The other interesting fragment observed here is the Nb4C3+ species, which must be formed by the elimination of a neutral carbon atom. This channel is weak, but reproducible, at both 355 and 532 nm dissociation wavelengths. This indicates that the IP of Nb4C3 is less than that of the carbon atom (11.26 eV). It is also noteworthy that carbon atom elimination has not been observed in the fragmentation of any other metal-carbon clusters. In every met-cars system, for example, the neutral fragments are either metal atoms or MCx species,22-26while metallofullerenespecies fragment by the loss of C 2 . l ~This ~ is one observation setting the niobium systems apart from the other metal carbides studied to date. Figure 5 shows the photodissociation mass spectrum of the

250

500

750

1000

Mass ( a m u ) Figure 6. Photodissociation mass spectrum of Nb&+ at 532 nm. As explained above, the limited resolution of the mass gate leads to incomplete isolation of the parent ion. Peaks corresponding to essentially every combination of metaYcarbon loss are observed. The 4/4 and 6/7 ions, which are also formed prominently in the nascent cluster distribution (see Figure 2), are slightly more prominent than other fragment masses.

ion Nb9C9+, which is one of the most abundant species in the nascent cluster distribution in Figure 2. This spectrum is taken at 355 nm, but again the fragment channels are unchanged at 532 nm or at lower laser power. The depletion region of this spectrum contains a triplet of peaks, rather than the desired single parent ion. This is because of the limited mass resolution available on the pulsed deflection plates which provide the mass selection. At the total mass of the parent Nb9C9+ peak (-950 m u ) , the two adjacent carbon peaks (Le., the ions Nb9Cs+ and NbsClo+) leak through the mass gate. The photodissociation laser is adjusted in time to overlap the 9/9 parent most efficiently, but some excitation of the other parent ions is unavoidable. Therefore, the fragment spectrum must be interpreted as arising from this distribution of three parent ions. As described above, the Nb+ channel is again the most prominent for this dissociation process. It is plotted offscale so that the molecular fragments can be presented more clearly. In the molecular fragments, there is virtually no intensity in mass channels larger than the group of peaks centered about Nb4C4+. The pattem of masses below that group is essentially the same as the fragments observed for Nb4C4+. There is no large magicnumbered fragment. The drop in intensity for fragments larger than 414 could be indicative that these species are relatively less stable, but it could also be argued that there was sequential fragmentation resulting in only those residual smaller fragments because of the available energy. Again, we must look at the general trends in the decomposition of other clusters to establish a trend in behavior. Figure 6 shows the photodissociation mass spectrum of the cluster Nb&12+, which has the stoichiometry as the met-car clusters studied previously for other metals. As shown in Figure 5 , we again have imperfect mass selection for this parent ion, and the depletion mass region contains three prominent peaks. In addition to the desired parent ion, we also have excitation of parent ions containing one additional and one fewer carbon atom. This fragmentation spectrum, like the ones above, also contains a relatively large intensity in the Nb+ channel, which is plotted offscale. As discussed above, this channel may arise because the various metal-carbon cofragments have higher ionization potentials on average than the niobium atom or from reionization of neutral Nb fragments. There are clumps of peaks at each niobium atom increment, with the number of carbon atoms varying in the vicinity of the 1:l stoichiometry. There are again no dramatic magic numbers. However, there is a small, but noticeable, intensity enhancement at the Nb4C4+ and

Pilgrim et al.

548 J. Phys. Chem., Vol. 99, No. 2, 1995

500

1000

1500

Mass ( a m u ) Figure 7. Photodissociation mass spectrum of Nb&l,+ at 532 nm. As above, the parent ion selection is incomplete. A distribution of fragments are observed, with a slight preference for masses near 4/4 and 617. The 9/9 mass, which is prominent in the nascent ion distribution, has very weak intensity here.

Nb&7+ channels. This broad distribution of fragment ions is dramatically different from the fragmentation observed for the other M8C12+ met-car clusters. In the titanium, niobium, chromium, and iron analogues, the dominant dissociation process is the loss of metal atoms. In what appears to be a sequential elimination process, the TigC12+parent, for example, produces the fragments Ti7C12+, Ti6C12+, and TisC12+.~~ The number of metal atoms lost has been discussed in terms of the relative stability of the various met-car clusters.22-26 In contrast to these first row transition metal systems, Mo&12+ and Zr8C12 dissociate by the loss of M, MC, and MC2 ~ n i t s . The ~ ~ most ,~~ abundant fragment from MogC12+ is Mo?C10+, while both 7/12 and 6/10 fragments are observed from ZrgC12+. Neither of these other second-row metals exhibits the extensive distribution of fragments seen for niobium, and neither has any evidence for abundant fragments at the 4/4 and 6/7 stoichiometries. In this sense, then, niobium is quite different from its second-row neighbors. Further evidence for the unusual behavior of niobium-carbon clusters is provided in the photodissociation of Nb14C13+, which is shown in Figure 7. As described above, the limited resolution of the mass gate allows several parent ions to leak through the mass gate, and these are shown as a multiplet in the depletion region of the spectrum. The general features of this spectrum are again similar to those already described. The atomic ion peak is large and is plotted offscale. There is a distribution of fragments, weighted toward the lower mass region. There is no strong magic number fragment, but again there are noticeable enhancements in intensity at the 4/4 and 6/7 fragment stoichiometries. The most dramatic fragment missing from this spectrum, or present only in low abundance, is NbgC12+. In every other metal-carbon system which we have studied, the 8/12 met-car stoichiometry is a prominent fragment from essentially every larger cluster containing enough metal and carbon atoms to make this fragment possible. An especially notable example is the dissociation of other M14C13+ clusters, in which the production of the 8/12 fragment has been explained by a surface reconstruction m e ~ h a n i s m . ~ ~ The - ~abundance ~,~~ of the 8/12 fragment from a variety of larger parent ions is found for titanium, vanadium, and zirconium clusters, in which the 8/12 stoichiometry is also a magic number in the nascent distribution formed in cluster g r o ~ t h . ~However, ~ - ~ ~ 8/12 is also produced as the main photofragment in the iron-carbon system, in which it is not a magic number in the growth di~tribution.~~ In the present spectrum, this fragment is completely lost in the broad distribution of other species.

Niobium-carbon clusters are therefore dramatically different from all the other metal-carbon systems in this region of the periodic table. The prominence of the 4/4 fragment ion again here reinforces the idea that decomposition proceeds by the production of smaller crystalline fragments. However, one would also expect to see the 9/9 species as a large fragment, but its abundance is quite low in this fragment spectrum. The safest overall statement for this system is that there is no strong preference for any dissociation pattern. There is certainly no preference for the 8/12 fragment seen for other met-cars systems. Photodissociation of other niobium-carbon species large enough to produce the 8/12 fragment (e.g., 12/12 and 13/13) also gives no evidence for strong formation of this cluster. Table 1 shows a summary of the other fragmentation processes observed for the various Nb/C cation clusters. In this table, the parent ion is indicated in the row labels, and photofragments observed are indicated by an X in the stoichiometry in the column label. An 0 indicates that the stoichiometry given by the column label is correct for the number of Nb atoms in the photofragment but that the number of carbon atoms cannot be determined exactly for that cluster because of the limited mass resolution. The parent ions studied here are the same as those which appear as relatively abundant peaks in the source distribution because these have enough density to study with photodissociation. The column labels in the table show that the most commonly found fragment ions are the same as the species which are abundant parent ions in the source distribution. Fragment stoichiometries not exactly the same as parent ions are indicated with appropriate M/C numbers. Thus, while there are no dramatic magic number clusters, the species formed in both cation cluster growth and fragmentation are, on the average, the same. These, then, are expected to be the most stable group of niobium-carbon cluster stoichiometries. As mentioned above, the M/C ratios in these systems are all nearly 1:1, with a tendency toward slight enrichment in carbon. Since there are no dramatic magic numbers in these fragmentation processes, it is interesting to examine the extent of dissociation in the various parent ions. If we compare the 9/9, 8/12, and 14/13 parent ions, it is apparent that the 9/9 species produces relatively more small fragment ions. However, this observation alone does not prove that there is more extensive fragmentation for this ion. The various fragments observed may be formed from either sequential processes or a number of concerted fission processes. If fission is occuning, the observation of small charged fragments may be favored because larger cofragments have higher ionization potentials. This idea can explain the observation of charged niobium atom, as described earlier, and it could conceivably explain the observation of small charged metal-carbide fragments. On the other hand, the 8/12 and 14/13 parent ions do produce a full range of larger charged fragments. This suggests that there is no severe IP problem in detecting these species. The observation of only small fragments from the 9/9 parent, therefore, may be caused because it fragments more extensively than the 8/12 and 14/13 ions. It could also be explained if the dissociation mechanism is different for 9/9 than it is for 14/13 and 8/12, Le., if 14/13 and 8/12 dissociate sequentially and 9/9 dissociates by cleavage of larger fragments. These observations suggest that the 8/12 and 14/13 species may be more stable than the 9/9 ion, even though the 9/9 ion is sometimes more prominent in the source mass distributions. This is an interesting question which could be tested by ab initio calculations on these systems. The absence of prominent magic numbered clusters here suggests that the relative bonding in the clusters here are similar. However, it is also interesting to compare the bonding in these

Photodissociation of Niobium-Carbon Clusters systems to that in other metal-carbide clusters. For example, the average bond energies calculated for the M8Cl2 and M14C13 clusters of titanium is in the vicinity of 5-6 eV per atom.34It would at first seem that strongly bound clusters such as these should not dissociate at 532 and 355 nm (2.33 and 3.49 eV, respectively). However, as we have discussed previou~ly?~ strongly bound clusters are heated significantly as they grow in the cluster source through the energy released in bond formation. This internal energy is so great that it cannot be cooled completely, even by supersonic nozzle sources. Therefore, these clusters which are internally hot can be caused to dissociate at low photon energies because absorption provides only the final increment of energy for the system, on top of the existing internal energy. This mechanism also is consistent with the insensitivity to the wavelength and the laser power used for dissociation. Once the final energy increment required for dissociation is obtained, the photodissociation channels are affected more by the unimolecular lifetime of the parent ion and the characteristic instrument time scale than they are by the slight energy difference in the two excitation wavelengths. The strongly bound titanium, vanadium, zironium, etc., metcars clusters and nanocrystal clusters which have been dissociated previously were all found to have fragment channels which were independent of laser power and ~ a v e l e n g t h .In ~ ~their dissociation dynamics, then, these niobium systems are similar to the other metal-carbide clusters studied. It therefore seems that their absolute bond energies are likely to be somewhat similar to the other systems studied (Le., perhaps several electronvolts per atom). With respect to the MgC12 met-cars stoichiometry, an additional comparison can be made to previously studied systems. The titanium and vanadium met-cars analogues fragment by the loss of only one or two metal atoms, while the zirconium and molybdenum systems lose M, MC, and MC2 moieties.22-26 However, the extent of fragmentation seen here for Nbgc12+ is significantly greater than that in the other second-row transition metal systems. It is more reminiscent of the multiatom elimination observed for FesClz+ and CrgC12+.~~ On the basis of the extent of fragmentation, it could be concluded that Nb&lZ+ has a stability more like the iron and chromium metcars analogues, which are believed to be relatively less stable than the titanium, vanadium, zirconium, and molybdenum systems. Again, this trend in bonding strengths could be tested with ab initio calculations on these various clusters. Conclusion Niobium-carbon clusters are unusual in several respects when compared to other metal-carbon clusters studied recently. There are no dramatic magic number patterns in these systems, but there is some tendency to produce the M8C12 and M14C13 cations observed for other metals. Additional stoichiometries produced with somewhat enhanced abundance are 414,513,617, and 919. All of these except 6/7 can be rationalized as fcc crystal fragments. However, the production of these various species is sensitive to the conditions in the cluster source, including the feed gas concentrations and the vaporization laser wavelength. The photodissociation channels of niobium-carbon clusters are likewise unlike those in other metal-carbide clusters. Again, there are no dramatic magic numbered fragment clusters produced, but there is a tendency to produce the fragments Nb&+ and Nb&7+. However, the ions Nb5c3+ and Nb9C9+, which are enhanced in the source cluster distributions, are not formed as abundant photofragments. There is also no evidence for the formation of NbgC12' as an important photofragment from larger clusters. Both the growth processes and

J. Phys. Chem., Vol. 99, No. 2, 1995 549 the photodissociationprocesses suggest that the NbsC12+ moiety is less stable than the corresponding met-cars clusters formed for other nearby fist- and second-row transition metals. While there are no dramatic magic numbers indicated here, both cluster growth and fragmentation experiments suggest that the most stable niobium-carbon clusters are Nb4C4+ and Nb6C7'. The 414 cluster is almost certainly the 2 x 2 x 2 fcc crystallite, but we are not able to suggest a simple structure for the 617 cluster. Ab initio calculations would be especially valuable to test the stabilities and structures for these particular clusters and to explain why niobium-carbon clusters in general are so different from other transition metal-carbon systems. Acknowledgment. We gratefully acknowledge support for this work from the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences (Contracts DEFG09-9OER14156 and DE-FG05-93ER14402). References and Notes (1) Curl, R. F.; Smalley, R. E. Sci. Am. 1991, 265, 54. (2) Chai, Y.; Guo, T.; Jin, C.; Haufler, R. E.; Chibante, L. P. F.; Fure, J.; Wang, L.; Alford, J. M.; Smalley, R.E. J. Phys. Chem. 1991, 91, 7564. (3) (a) Shinohara, H.; Sato, H.; Saito, Y.; Ohkochi, M.; Ando, Y. J. Phys. Chem. 1992,96,3571. (b) Shinohara, H.; Yamaguchi, H.; Hayashi, N.; Sato, H.; Ohkohchi, M.; Ando, Y.; Saito, Y. J. Phys. Chem. 1993, 97, 4259. (c) Bandow, S.; Shinohara, H.; Saito, Y.; Ohkohchi, M.; Ando, Y. J. Phys. Chem. 1993, 97, 6101. (4) McElvany, S. W. J. Phys. Chem. 1992, 96,4935. (5) Gillan, E. G.; Yeretzian, C.; Min, K. S.; ALvarez, M. M.; Whetten, R. L.; Kaner, R. B. J. Phys. Chem. 1992, 96, 6869. (6) Tanigaki, K.; Ebbesen, T. W.; Saito, S.; Mizuki, J.; Tsai, J. S.; Kubo, Y.; Kuroshima, S. Nature 1991,352, 222. (7) Fischer, J. E.; Heiney, P. A,; Smith,A. B. Acc. Chem. Res. 1992, 25, 112. (8) (a) Ruoff, R. S.; Lorents, D. C.; Chan, B.; Malhotra, R.; Subramoney, S. Science 1993,259, 346. (b) Moro, L.; Ruoff, R. S.; Becher, C. H.; Lorents, D. C.; Malhotra, R. J. Phys. Chem. 1993, 97, 6801. (9) Guo, B. C.; Kearns, K. P.; Castleman, A. W., Jr. Science 1992, 255, 1411. (10) Guo, B. C.; Wei, S.; Pumell, J.; Buzza, S.; Castleman, A. W., Jr. Science 1992, 256, 515. (1 1) Wei, S.; Guo, B. C.; Purnell, J.; Buzza, S.; Castleman, A. W., Jr. Science 1992, 256, 818. (12) Wei, S.; Guo, B. C.; Pumell, J.; Buzza, S.; Castleman, A. W.; Jr. J. Phys. Chem. 1992, 96, 4166. (13) (a) Chen, 2. Y.; Castleman, A. W.; Jr. J. Chem. Phys. 1993, 98, 231. (b) Zieman, P. J.; Castleman, A. W. J. Chem. Phys. 1991, 94, 718. 114) Guo. B. C.: Kerns. K. P.: Castleman, A. W. J. Am. Chem. SOC. 1993, 115,7415. (15) Guo, B. C.; Wei, S.; Chen, 2.; Kerns, K. P.; Pumell, J.; Buzza, S.; Cakeman, A. W., Jr. J. Chem. Phys. 1992, 97, 5243. (16) Chen, 2. Y.; Guo, B. C.; May, B. D.; Cartier, S. F.; Castleman, A. W., Jr. Chem. Phys. Lett. 1992, 198, 118. (17) Wei, S.; Guo, B. C.; Pumell, J.; Buzza, S. A.; Castleman, A. W. J. Phys. Chem. 1993, 97, 9559. (18) (a) Cartier, S. F.; May, B. D.; Castleman, A. W. J. Chem. Phys. 1994. 100. 5384. 1b) Cartier. S. F.: Mav. B. D.: Castleman, A. W. J. Am. Chem. SOC. 1994, 116, 5295. 119) Cartier. S. F.: Chen, 2. Y.; Walder, G. J.; Sleppy, _ _C. . R.; Castleman, A. W.'Science 1993, 260, 195. (20) Cartier, S. F.; May, B. D.; Toleno, B. J.; Pumell, J.; Wei, S.; Castleman, A. W. Chem. Phys. Letr. 1994, 220, 23. (21) Wei, S.; Guo, B. C.; Deng, H. T.; Kerns, K.; Pumell, J.; Buzza, S. A.; Castleman, A. W. J. Am. Chem. SOC.1994, 116, 4475. (22) Pilgrim, J. S.; Duncan, M. A. J. Am. Chem. SOC. 1993,115,4395. (23) Pilgrim, J. S.; Duncan, M. A.; J . Am. Chem. SOC. 1993,115,6958. (24) Pilgrim, J. S.; Duncan, M. A. J. Am. Chem. SOC. 1993,115,9724. (25) Pilgrim, J. S.; Duncan, M. A. In?. J. Mass Spectrom. Ion Processes, in press. (26) Pilgrim, J. S.; Duncan, M. A.; Advances in Metal and Semiconductor Clusters, Vol. III; Duncan, M. A., Ed.; JAI Press: Greenwich, CT, in press. (27) Yeh, C. S.; Afzaal, S.; Lee, S. A.; Byun, Y.; Freiser, B. S. J . Am. Chem. Soc., submitted. (28) Jin, C.; Haufler, R. E.; Hettich, R. L.; Barshick, C. M.; Compton, R. N.; Puretzky, A. A.; Dem' yanenko, A. V.; Tuinman, A. A. Science 1994, 263, 68.

550 J. Phys. Chem., Vol. 99, No. 2, 1995 (29) Pauling, L. Proc. Natl. Acad. Sci. U S A . 1992, 89, 8175. (30) (a) Lin, Z.; Hall, M. B. J. Am. Chem. SOC. 1992, 114, 10054. (b) Lin, 2.;Hall, M. B. J. Am. Chem. SOC. 1992, 115, 11165. (31) (a) Rohmer, M. M.; de Vaal, P.; Benard, M. J. Am. Chem. SOC. 1992, 114, 9696. (b) Rohmer, M. M.; Henriet, C.; Bo, C.; Poblet, J.-P. J . Chem. Soc., Chem. Commun. 1993, 1182. (32) Reddy, B. V.; Khanna, S. N.; Jena, P. Science 1992, 258, 1640. (33) Hay, P. J. J . Phys. Chem. 1993, 97, 3081. (34) Reddy, B. V.; Khanna, S. N. Chem. Phys. Len. 1993, 209, 104. (35) Chen, H.; Feyereisen, M.; Long, X. P.; Fitzgerald, G. Phys. Rev. Lett 1993, 71, 1732. (36) Grimes, R. W.; Gale, J. D. J . Phys. Chem. 1993, 97, 4616.

Pilgrim et al. (37) Lou, L.; Guo, T.; Nordlander, P.; Smalley, R. E. J . Chem. Phys. 1993, 99, 5301. (38) Khan, A. J . Phys. Chem. 1993, 97, 10937. (39) (a) Dance, I. G. J . Chem. SOC., Chem. Commun. 1992, 1779. (b) Dance, I. G. J . Chem. SOC.,Chem. Commun. 1993, 1306. (c) Dance, I. G. J . Am. Chem. SOC.1993, 115, 11052. (d) Dance, I. G. J . Am. Chem. SOC., in press. (40) Lou, L.; Nordlander, P. Chem. Phys. Lett., in press. (41) Comett, D. S.; Peschke, M.; LaiHing, K.; Cheng, P. Y.; Willey, K. F.; Duncan, M. A. Rev. Sci. Insfrum. 1992, 63, 2177. JP94 19306