4948
J . Phys. Chem. 1991, 95,4948-4950
Doping Bucky: Formation and Properties of Boron-Doped Buckminsterfullerene Ting Cuo, Changming Jin, and R. E. Smalley* Rice Quantum Institute and Departments of Chemistry and Physics, Rice University, Houston, Texas 77251 (Received: April I I , 1991)
Laser vaporization of a graphite pellet containing boron nitride powder was found to produce fullerenes in which one or more atoms of the hollow carbon cage was replaced by a boron atom. The clusters were probed as isolated positive and negative ions by ion cyclotron resonance techniques. The boron-doped carbon clusters were demonstrated to be fullerenes by the absence of odd-numbered clusters, and relative prominence of the clusters with 50,60, and 70 atoms, and the fact that they fragment by the successive loss of Cz, resulting in a shrinking of the cage. The boron-doped clusters were found to act as Lewis acids, readily chemisorbing one ammonia molecule per surface boron atom.
Introduction Laser vaporization supersonic cluster beam studies have shown that carbon appears to be unique in its ability to form fullerene cages.' Silicon, for example, does not form strong enough double bonds to stabilize the 3-connected network of the cage. Silicon clusters in the 2-100 atom size range therefore show no tendency to favor even-numbered clusters, whereas in carbon this even/odd alternation is one of the hallmarks of fullerene production? Boron nitride has a stable crystalline phase made up of two-dimensional hexagonal sheets much like graphite, leading to the suggestion that fullerenstype cages may be formed from BN units. However, since all fullerene cages must contain 12 pentagonal rings, it will be impossible to avoid B-B and N-N linkages in such a structure, and the resultant molecules would be expected to be unstable. Indeed, experiments performed earlier in this group with laservaporized boron nitride targets to form cluster beams revealed no indication of fullerene cage formation. Carbon does appear to be the only element capable of forming stable geodesic cage molecules. However, this does not necessarily mean that only pure carbon cages can be formed. It should be possible to substitute a heteroatom every once in a while into the carbon network without destabilizing the entire cage. This is particularly true since the fullerene cage is closed. Even though the heteroatom may not produce as energetically stable a cage as carbon, the barrier to removal of this heteroatom may be sufficiently high that the substituted cage will have high kinetic stability. This is the case with the various common ptype and n-type donor atoms deep in the bulk of semiconductor lattices like silicon, for example. Even so, there is the problem of how to put the heteroatom in the fullerene cage to begin with. The model originally put forward by this group for the formation of these cages3 involved the rearrangement of growing graphitic sheets so as to incorporate pentagons and thereby produce a curvature that would minimize the number of dangling bonds. As we have discussed elsewhere," this model nicely explains the role of the helium in the Kratschmer-Huffman recipeS for the production of C, in good yield. In this model the critical factor governing the yield of fullerenes of size near 60 is whether clusters growing into this size range have an adequate opportunity to anneal to their most energetically favored form. If, as we suspect, this form is the one that follows what we call the 'pentagon rule" (Le., the one that has the largest ( I ) (a) Curl, R. F.; Smalley, R. E. Science 1988,242, 1017. (b) Smalley, R. C. Supersonic Carbon Cluster Beams. In Atomic and Molecular Clusrers; Bemstein, E. R., Ed.; Studies in Physical and Theoretical Chemistry, Vol. 68; Elsevier Science: Amsterdam, 1990; pp 1-68. (2) Alford, J. M.; Laaksonen, R.T.; Smalley, R. E. J. Chem. Phys. 1991, 94, 2618 and references therein. (3) Zhang, Q.L.; OBrien, S.C.; Heath, J. R.; Liu, Y.; Curl, R. F.; Kroto, H. W.; Smalley, R. E. J. Phys. Chem. 1986, 90, 525-528. (4) Haufler, R. E.;,Chai, Y.;Chibante, L. P. F.; Conceicao, J.; Jin, Changming; Wang, Lai-Sheng; Maruyama, Shigeo; Smalley, R. E. Muter. Res. SOC.Symp. Proc., in press. ( 5 ) Kratschmer, W.; Lamb, L. D.; Fostiropoulos, K.; Huffman, D. R. Nuture 1990, 347, 354.
0022-365419 1/2095-4948%02.50/0
possible number of pentagonal rings while avoiding adjacent pentagons), then the fantastically high yields of Cboreported recently6 are readily understood. The C, soccerball structure is the inevitable result of slavish adherence to this pentagon rule during the growth process. The optimum pressure of helium is then the one that permits the most effective annealing during the critical early stages of growth. It does this by controlling the density of small carbon radicals close to the vaporizing graphite rod where the temperature is still high. Our estimate of the critical temperature range for effective annealing of these growing graphitic sheets4 is somewhere between 1500 and 3000 OC. From the standpoint of incorporating heteroatoms into the growing network, the critical problem will be having this substitution of the carbon net survive at these temperatures long enough to be trapped. The current lack of success in obtaining macroscopic amounts of metallofullerene such as C& may be due to just such a problem. The K atom may not stay attached to the growing carbon net long enough a t these high temperatures to be trapped. In the case of metals like K, Ca, and U, we currently suspect the atom must be able to stay attached to the side of the growing net, most likely by a charge/chargeinduced dipole interaction if the metal atom retains a net charge or a charge-transfer interaction if it is neutral. These are rather weak interactions for the job of holding a metal atom onto a writhing graphitic net at over 1500 'C. However, the likelihood of an atom like boron or nitrogen remaining attached to the carbon network should be much greater, since here it will be attached by strong covalent bonds as an integral part of the net. With the following experiments we provide the first evidence of successful incorporation of boron atoms into the carbon network of fullerenes. Laser vaporization was used on a target made from a mixture of graphite and boron nitride, and the analysis was performed o n the cluster ions levitated in a magnetic field. Even though no macroscopic amounts of these boron-doped fullerenes have yet been produced, they are shown here to be readily formed and highly stable.
Experimental Section The apparatus used for these studies was the Fourier transform ion cyclotron resonance (FT-ICR) device described extensively in earlier publications from this laboratory.**' Clusters were produced by laser vaporization of a graphite/boron nitride composite disk (1.2-cm 0.d.) prepared by mixing 15 wt % boron nitride powder (Fischer) with graphite powder (Poco Graphite Inc.) together with a graphite cement binder (Dylon Industries, Inc.). This composite disk was mounted in the minaturized supersonic pulsed cluster beam source described elsewhere.' Vaporization was accomplished with 10-20 mJ of the Nd:YAG second har(6) Whetten, R. L.; Alvarez, M. M.; Anz, S.J.; Schriver, K. E.;Beck, R. D.; Diederich, F. N.; Rubin, Y.;Ettl, R.; Foote, C. S.; Darmanyan, A. P.; Arbogast, J. W. Mal. Res. SOC.Symp. Proc., in press. (7) Maruyama, S.; Anderson, L. R.; Smalley, R. E. Rev. Sci. Inrrrum. 1990, 61, 3686.
0 199 1 American Chemical Society
The Journal of Physical Chemistry, Vol. 95, No. 13, 1991 4949
Letters
(a) As Selected.
60
580
620
660
700
740
780
820
860
Cluster Mass ( amu ) Figure 1. FT-ICR mass spectrum of positively (+I) charged clusters prepared by laser vaporization of a boron nitride/graphite composite target disk in a supersonic nozzle. The fine structure in the mass spectrum associated with each labeled clump reveals that boron has been substituted into the fullerene cage.
monic (532 nm) focused to a 0.1-cm-diameter spot on the target disk. Helium carrier gas was used, and the resultant supersonic cluster beam was directly injected into the ion trap of the FT-ICR apparatus, where the cluster ions were decelerated, stopped, and levitated in the 6-T solenodial magnetic field. The timing of the laser vaporization pulse was adjusted in synchrony with the onset of the supersonic helium pulse so that extensive residual ionization from the laser plasma remained. The cluster ions studied here, both positive and negative, were the residual ions thus obtained. After injection into the FT-ICR analysis cell, the clusters were thermalized by exposure to roughly 100 collisions with argon. Selective ejection of clusters from the cell, where appropriate, was accomplished by computer-crafted “SWIFT” rf waveforms as described in detail in earlier discussions of this apparatw2
Results and Discussion Figure 1 shows an FT-ICR mass spectrum of a section of the positive cluster ions produced by laser vaporization of the BNloaded graphite target disk. Although the detailed fine structure of the cluster ion peaks in this figure may be difficult for the reader to discern, it is clear that only even-numbered clusters are present, as labeled in the figure. More careful examination reveals that there is more fine structure to each of the major clumps than can be explained by carbon alone. Figure 2a shows the fine structure near the C,, clump more clearly. In addition to peaks at 720, 721, and 722 amu due to l2Cs0, 12Cs913C, and 12Cs813C2, respectively, there are clearly cluster ions present in significant abundance a t 719, 718, 717, 716, and 715 amu. These are due to carbon clusters that have been contaminated with one or more boron atoms, e.g., CS9Band CS8B2. Exact measurement of the relative importance of these various species in the ICR trap is complicated by the high natural abundance of the two stable isotopes of boron (20% log, 80% IIB). However, as demonstrated in earlier work from this group in studies of gallium arsenide clusters,8 it is possible to deconvolute the observed mass spectrum to obtain approximate relative compositions. The resulting compositions were found to vary somewhat shot to shot. The mass spectrum shown in Figure 2a was quite typical. In this case the approximate compositions were found to be 22% Cm, 21 % C ~ S B24% , C S ~ B18% ~ , C57B3,9% C56B4,4% CSSBS,and 2% C54B6. The most important aspect of Figure 1 is that only even-numbered clusters are observed. As mentioned in the Introduction, this is a signal feature of carbon cluster distributions for which the only compelling explanation yet offered is that they are, in fact, fullerenes, Le., closed geodesic spheres consisting of 12 (8) Wang,
L. H.;Chibante, L. P.F.;Tittel, F. K.;Curl, R. F.;Smalley,
R. E. Chem. Phys. Left. 1990, 172, 335.
(b) Reaction with Ammonia for 2 sec. c60
I
C5gB(NH3) C58B2 (NH3)2
700 710 720 730 740 750 760 770-700 790 800
Cluster Mass ( amu )
Figure 2. Detail view (a) of the FT-ICR mass spectrum of the 60-atom clusters produced as in Figure 1. All other clusters were selectively ejected from the ICR trap. The result of exposure of the selected clusters to ammonia vapor is shown in the bottom panel (b).
pentagons and an appropriate number of hexagons. The fact that they each appear here with roughly the same amount of boron substitution strongly supports the notion that the boron atoms have been “doped” into the fullerene cages. Notice that the borondoped cages still appear to be particularly stable with 50 and 60 atoms, just as with the corresponding pure carbon fullerenes. Similar cluster mass spectra taken in higher mass ranges reveal, as expected, that the boron-doped 70-atom clusters are also particularly abundant. The positive cluster distribution from this boron nitride/graphite composite target was checked out to over 200 atoms. Throughout the 44-200-atom size range only the even-numbered clusters were evident. The extent of boron doping appeared to increase roughly linearly with the cluster size, supporting the notion that boron incorporation into the growing carbon network under these source conditions is basically a statistical process. Another indication that the boron atoms have been doped into the carbon network of the fullerene cage is their surface chemistry. Any such surface boron atoms should be connected to three carbon atoms in the cage through single covalent bonds. Particularly in this case of the positive ions, there is no question that these boron atoms should act as strong Lewis acid sites. Figure 2b shows the result of a test of this prediction by exposure of the mass-selected 60-atom clusters to N H 3 at a pressure of 1 X 10” Torr for 2 s. Now it is quite clear that the boron-doped 60-atom (acidic) clusters have been rather effectively titrated by this ammonia base. Deconvolution of the isotopic fine structure on the observed ammoniated product peaks shows them to be dominated by the single compositions labeled in Figure 2. As is evident in this figure, the two borons in the doubly doped buckminsterfullerene cage appear to be acting independently. Judging from the size of the N H 3 group, these two boron atoms must be well separated from one another on the surface of the cage.
Photophysics One of the most vivid demonstrations of the stability of fullerene cages has been their photofragmentation behavior. The lowest
4950 The Journal of Physical Chemistry, Vol. 95, No. 13, 1991
Letters pentagons on the surface of the fullerene cage until two pentagons share an edge. The C2 fragment can then leave in a concerted process which simultaneously forms the next smaller fullerene. This is the “shrink-wrapping” mechanism that was used several years ago to prove that metal atoms had been captured in the fullerene cage.” In the case of Cm, a recent estimate’& of the amount of internal energy required for this process to proceed at a rate that competes well with cooling by infrared emission is over 30 eV. The photofragmentation behavior of borondoped fullerene is therefore quite a stringent test of the ability of the boron to stay doped in the cage. Figure 3 shows the result of laser photolysis of the positively charged mass-selected 60-atom clusters prepared, as above, by laser vaporization of a boron nitride/graphite composite. The top panel shows the FT-ICR mass spectrum of the 60-atom cluster prior to irradiation; the bottom shows the resultant spectrum after 2-s exposure to XeCl excimer laser pulses at a repetition rate of 50 s-I and an average fluence of 20 mJ cm-2 per pulse. Due to technical restraints, the laser beam profile only overlapped with roughly 30% of the cluster ions in the cell for this experiment. Even with this unfortunate problem, it was clear that those clusters which did ultimately fragment did so by successive loss of C2 units and that the boron-doped clusters were just as photoresistant as
(a) Cluster of 60 atoms a s selected.
J (b) After exposure to XeCl laser for 2 sec. 60
Cm.
.
580
600
620
640
660
680
.
,
700
.
,
720
.
740
Cluster Mass ( amu ) Figure 3. IT-ICR mass spectra showing result of photolysis of the
levitated 60-atom clusters. Top panel (a) shows the originally selected clusters prior to irradiation. Bottom panel (b) reveals the formation of daughter fragments by successive loss of C2. fragmentation channel for all fullerenes larger than C32has been found to be C2loss.9Jo Since C2is quite weakly bound compared to C3and larger carbon radicals, the dominance of this fragmentation channel must be due to the stability of the other part of the molecule left behind. With this thought in mind a mechanism has been proposed9 that involves the migration of the (9) OBrien, S.C.; Heath, J. R.; Curl, R. F.; Smalley, R. E. J. Chem. fhys. 1988,88,220. (10) (a) Radi, P. P.; Hsu, M.-T.; Rincon, M. E.; Kemper, P. R.; Bowers, M. T. Chem. Phys. L r r . 1990, 174, 223. (b) Radi, P. P.; Bum, T. L.; Kemper, P. R.;Molchan, M. E.; Bowers, M. T. J . Chem. fhys. 1988, 88, 2809.
Further such experiments are planned to follow this fragmentation process down to the range of Csz where the pure carbon fullerenes are found to burst abruptly. It will be interesting to find whether there is any intermediate point where loss of a boron-containing fragment becomes important. Conclusion These initial laser vaporization experiments provide strong support for the notion that fullerene cage molecules can be doped with heteroatoms. While these new cluster structures were produced here in microscopic amounts by laser vaporization, it may very well be possible to produce large quantities of similarly doped fullerenes by contact arc vaporization of appropriately compounded graphite rods. Such experiments are currently in progress.
Acknowledgment. This research was supported by the Office of Naval Research, The National Science Foundation, and the Robert A. Welch Foundation and utilized a cluster FT-ICR apparatus developed with substantial funding from the U.S. Department of Energy, Division of Chemical Science. ( 1 1) Weiss, F. D.;O’Brien, S. C.; Elkind, J. L.; Curl, R. F.; Smalley, R. E.J . Am. Chem. SOC.1988, 110, 4464.
