5763 J . Phys. Chem. 1991, 95, 5163-5168 + found, indicating that the 1(NH3),-(L),)H+, n m = 5 , are particularly stable. It is proposed that a central ammonium ion NH4+ is bound to four constituents in any combination of ammonia and L molecules. Studies of metastable unimolecular dissociation processes reveal that the ((NH3),.(L),)H+. n = 1, m = 2-4, cluster ions lose one L molecule whereas the {(NH~),,. (L),)H+, n = 2-1 1, m = 1-4, cluster ions lose one NH3 molecule in a few tens of microseconds. The experimental results suggest that the stable cluster ions ((NH3),,.(CH$N),IH+ and I(NH3),.(CH3CHO),)H+, n m = 5 , have completed solvation shell structures as expressed in structures I. + Acknowledgment. Support by the US.Department of Energy, Grant NO. DE-FG02-88-ER60668, is gratefully acknowledged. Registry No. NH3, 7664-41-7; NH4, 14798-03-9; CH,CN, 75-05-8; CH,CHO, 75-07-0. B. I. Dunlap,* D. W. Brenner, J. W. Mintmire, R. C. Mowrey, and C. T. White Theoretical Chemistry Section, Code 61 19, Naval Research Laboratory, Washington, D. C. 20375-5000 (Received: February 8, 1991; In Final Form: March 13, 1991) Local density functional methods and empirical hydrocarbon potentials are used to study the electronic and geometric structures of icosahedral Cm C,Hm and C,FW Compared to isolated C,, the C-H and C-F bond strengths of these saturated species are reduced by over 40% and over 15% from the C-H and C-F bond strengths in methane and in tetrafluoromethane,respectively. Similar C-H bond strengths are found for C,Hw and CsoHso Other C&, species are optimized by using empirical potentials. Outward bonding for all hydrogen atoms is not always energetically preferred. 1. Introduction C, has been isolated in macroscopic quantities in several laboratories.’” Carbon- 13 NMR of pure C, dissolved in various solvents shows that this molecule has one symmetry inequivalent atom*5 and therefore has the proposed buckminsterfullerene (BF) structure with icosahedral ( I h ) symmetry.’ Haufler et ale6have partially hydrogenated C, using Birch reduction to yield C60H36. C60H36is only the first in a series of new molecules that probably will be synthesized starting from c60. Perhaps methods will be devised to fully hydrogenate C, to CWH, and other routes found to yield CmFbO,which should be inert similar to poly(tetrafluoroethylene). These hypothetical fully saturated Ih molecules have already been studied theoretically within the Hartree-Fock (HF) approximation.* Stimulated by these results and possibilities, we have chosen to study the geometry and electronic structure of these and related molecules using empirical potentialsg and local density functional (LDF) calculations.’0 Of the many possible hydrogenated and fluorinated (260 structures, the fully saturated species, C,H, depicted in Figure 1, and the corresponding C,F, molecule, have the highest symmetry and hence are easiest to study by the all-electron, firstprinciples LDF methods herein. However, if intermediate steps in the transformation of c 6 0 to C(,OHw are considered, the high symmetry of the reactants and products is lost and a full geometry optimization of these intermediate species rapidly becomes impractical. In addition, if the proposed CaH36structure$ depicted in Figure 2, is made from a molecular modeling kit using flexible tubes of roughly equal length to form nearest-neighbor bonds, then the bonds connecting saturated carbon atoms bend slightly inward while the bonds connecting unsaturated carbon atom pairs of their four neighboring saturated atoms are quite planar. The inward curvature of this model for these saturated C-C bonds is as great as the outward curvature of a similarly constructed CWmolecule. The inward curvature for a fully saturated Ih C6OH60 is even greater. These results suggest that, rather than bonding all hydrogens to the exterior surface of the cage, lower energy isomers might be obtained by bonding a subset of these hydrogens to the interior surface of the molecule to better compensate local strains. However, putting one or a few hydrogen atoms inside the BF cage lowers both the tetrahedral (Th)symmetry” of the proposed Corresponding author. CsOH36structure6 depicted in Figure 2 and the I , symmetry of the C&, molecule depicted in Figure 1, making a first-principles study of these isomers difficult. We have optimized the geometries of some of these lower symmetry structures using recently introduced empirical hydrocarbon potentials9 as a practical way to address whether moving a subset of hydrogen atoms inside the Cm cage lowers the total energy of these clusters. We also have used these empirical potentials in a preliminary survey of the energetics of hydrogen addition to the exterior surfaces of some of these clusters. In section 2 we describe our LDF approach and compare our LDF results with experiment for CH4 and CF4. We also briefly review in section 2 our recently developed empirical hydrocarbon potentials. In section 3 we consider the icosahedral species, C,, CWHmand C,F, with all H and F atoms bonded to the outside of the cage. Optimized LDF geometries allow us to compare the C-H and C-F bond energies of C,H, and CWFa and to compare empirical potential and LDF geometries for CaHm In section 4 we use our LDF approach to study the recently synthesized? partially hydrogenated species CWH36at the optimized empirical potential geometry. In section 5 we consider a number of other CWHnspecies using empirical potentials in a preliminary (1) Kritschmer, W.; Fostiropoulos, K.; Huffman, D. R. Chem. Phys. Lett. 1990, 170, 167. (2) Kratschmer, W.; Lamb, L. D.; Fostiropolous. K.; Huffman, D. R. Nature 1990, 347, 354. (3) Taylor, R.; Hare, J. P.; Abdul-Dada, A. K.; Kroto, H. W. J . Chem. SOC.,Chem. Commun. 1990, 1423. (4) Johnson, R. D.; Meijer, G.; Bethune, D. S. J . Am. Chem. Soc. 1990, 112, 8983. ( 5 ) Ajie, H.; Alvarez, M. M.; Anz, S. J.; Beck, R. D.; Dicderich, F.; Fostiropoulos, K.; Huffman, D. R.; Kratschmer, W.; Rubin, Y.;Schrivcr, K. E.; Sensharma, D.; Whetten, R. L. J . Phys. Chem. 1990, 94, 8630. (6) Haufler, R. E.;Conceicao, J.; Chibente, L. P.F.; Chai, Y.; Byrne, N. E.;Flanagan, S.;Haley, M. M.; OBrien, S. C.; Pan, C.; Xiao, 2.;Billups, W. E.;Ciufolini, M. A.; Hauge, R. H.; Margrave, J. L.; Wilson, L. J.; Curl, R. F.; Smalley, R. E. J. Phys. Chem. 1990. 94, 8634. (7) Kroto, H. W.; Heath, J. R.; OBrien, S.C.; Curl, R. F.; Smalley, R. E. Nature 1985, 318, 163. (8) Scuseria, G. E. Chem. Phys. Lett. 1991, 176,423. (9) Brenner, D. W. Phys. Rev. B 1990, 42, 9458. (IO) Dunlap, B. I.; Connolly, J. W. D.; Sabin, J. R. J. Chem. Phys. 1979, 71, 3396, 4993. (11) Dunlap, B. 1.; Brenner, D. W.; Mowrey, R. C.; Mintmire, J. W.; Robertson, D. H.; White, C. T. Mater. Res. Soc. Symp. Proc., in press. This article not subject to US.Copyright. Published 1991 by the American Chemical Society Dunlap et al. 5764 The Journal of Physical Chemistry, Vol. 95, No. 15, 1991 TABLE I: C m B-is Set SMk# C-C bonds, A basis tvDe 1 tvw 2 91511 1.43 1.43 1.45 1.39 1.39 1.39 ll/6/l ll/7/l Eb, kcal/mol 218 El, hartrecs 201 197 -2.30 -1.72 -0.41 uSuccessivecolumns from the left are the primitive Gaussian basis sets in s/p/d notation, the two optimized bond distances, the atomization energy per carbon atom, and El as computed using eq 1. Bonds of type 1 are shared by a pentagon and hexagon while bonds of type 2 are shared by two hexagons. Figure 1. Icosahedral CmHm. U U Figure 2. Tetrahedral CmH36 with the 36 hydrogen atoms pointing outward. survey of the energetics of hydrogenation and isomerization on the Cso cage of these molecules. 2. Computational Methods Computationally, LDF‘s differ from H F in that they have a single one-electron (Fock) potential.I2 Thus, Abelian and nonAbelian groups are treated in the same fashion if one restricts the Fock potential to be symmetric under the point group of the nuclei. The trace of the Hamiltonian within each multidimensional (for non-Abelian groups) irreducible representation of the group is i n ~ a r i a n t ,reducing ’~ considerably the number of integrals involving this trace that must be actually computed.14 For example, consider the 60-equivalent-atom case of BF with one basis function on each atom. The 1830 matrix elements resulting from all unique orbital pairs in the 60 X 60 orbital products reduce by symmetry to 32 independent terms. The one-electron effective Hamiltonian is real; thus, the real and complex components of one-dimensional complex irreducible representations, which occur for the Thgroup used herein, can be treated as if they comprise a two-dimensional irreducible representation of a group. Maximal use of symmetry has been incorporated into a computer codeI5 that extends the practical a p roach of Sambe and FeltonI6 to variationally fitting the density’ and that uses products of solid-spherical harmonics and Gaussian functions as basis sets for both the one-electron orbitals and the LDF potential that define the orbitals.” The LDF used herein is the Perdew-Zunger’* (PZ) fit to the free- !? (12) Slater, J. C. Quanrum Theory of Molecules and Solids; McGrawHill: New York, 1974;Vol. 4. (13) Pitzer, R. M. J . Chem. Phys. 1972,58, 31 11. (14)Dunlap, B. 1. Adv. Chem. Phys. 1987, 69, 287. (I 5) Dunlap, B. 1.; R k h , N. Advances in Quantum Chemistry; Trickey, S.B., Ed.; Academic: New York, 1990,p 317. (16)Sambe. H.; Felton, R. H. J . Chem. Phys. 1975.62, 1122. (17)Dunlap, B. I. Phys. Rev. A 1990. 41, 1127. (18)Perdew, J. P.;Zunger, A. Phys. Rev. B 1981, 23, 5948. electron gas results of Ceperley and Alder.l9 Our previous LDF calculationsm on Cso used two orbital basis sets for the carbon atom. The first two are the 9s/5p and 1ls/6p orbital basis sets of van Duijneveldt2Iaugmented with a d function of exponent 0.57 bohr2 and contracted to 2,3/1,3/0,1 and 2,4/1,4/0,1. In this orbital contraction notation each angular momentum is separated by a slash, the first of the two numbers being the number of lowest energy atomic orbitals used and the second being the number of most diffuse Gaussian basis functions used. Two additional totally symmetric Gaussian basis sets are needed to fit the electron-electron LDF potential. Spherically symmetric atom-centered functions were obtained by scaling the orbital exponents appropriate for treating the direct Coulomb potential and for treating the weaker exchange-correlation potential,1° discarding exponents smaller than 0.3 and 0.1, respectively, for the two basis sets.22 The p fitting functions were 0.3, 0.7, 2.0 on carbon and 1.0 on hydrogen, with an additional exponent 5.0 added for the larger basis set. Bond-centered fitting functions of exponent 0.5 were added at the midpoint of each nearest-neighbor bond. The results of the previous study are included in Table I. The two bond lengths are determined by fitting the computed total energy on a grid of points and then computing the energy at the optimal geometry predicted from the fit. This process is iterated until the optimized bond distances are determined to within 0.01 A. Approximate A4 vibrational frequences can be determined by fitting the potential energy surface to simple polynomials in the various bond distances;1° thus, the greater the number of degrees of freedom, the more sensitive the fitted frequencies are to the exact nature of the fitting polynomial. Our previous calculations suggested that the bond distances are insensitive to further improvements in the orbital basis sets. For this work larger fitting basis sets were used. The C orbital basis set, an 1ls/7p basis21 augmented with a d exponent of 0.6 b ~ h r - * is, ~not ~ a significant change. The orbital basis set was contracted 2,3/ 1,2/0,1. The fitting basis sets were improved by attempting to minimize the absolute value of El El = Y2Jdr1 dr, i j ( r M r 2 ) - p(r2))/r12 (1) which is the difference in the self-Coulomb repulsion of all the electrons computed by using the fit to the density p interacting with the true density p less the self-Coulomb repulsion computed by using only the fitted charge density. This quantity is zero if the charge density is variationally fit without using a constraint.I0 The constraint that the integral of the fitted charge density equals the total number of electrons was used. Both constrained and unconstrained fits, although different, yield total energies that are identical within 10” hartrees. Using a constrained fit and minimizing E , leads to the best potential energy surfaces for a given orbital basis set.22 E l was reduced by a factor of 4 by including all scaled s-orbital exponents, without cutoff. Use of diffuse exchange-correlation fitting exponents required the use (19)Ceperley, D. M.; Alder, B. J. Phys. Rev. Lerr. 1980, 45, 566. (20)Dunlap, B. 1. Inr. J . Quanrum Chem., Quonrum Chem. Symp. 1988, 22, 257. (21)van Duijneveldt, F. B. IBM Rcsearch Report RJ945. 1971. (22)Dunlap, 8. 1.; Mei, W. N. J. Chem. Phys. 1983, 78,4997. (23)Huzinaga, S.,Ed. Gaussian Basis Sers for Molecular Calcularionr; Elsevier: Amsterdam, 1984. Structures of CMHM,CMFM,and CMH36 The Journal of Physical Chemistry, Vol. 95, No. IS, 1991 5765 TABLE II: Comparison of LDF md ExperimentalSSC-H a d C-F Avmge Bond Enet@e~(E,), Bond D ~ s ~ (R& c ~ s 8 d Ai Vibntionrl Frequencies (Y,) for CHI a d CF, CHI quantity Rc-x, A Eb, kcal/mol v , , cm-I exp 1.091 99 2916 CF, LDF 1.10 116 2990 exp 1.320 102 909 LDF 1.32 146 920 of variational weights to ensure numerical stability.24 Both Sp and 5d fitting exponents were used with exponents 0.25,0.37,0.7, 2.0, and 5.0 bohr-*. The bond-centered exponents were 1.0 and 0.33 bohr-* for the charge density and exchange-correlation basis sets, respectively. The calculated binding energy per atom Ebof CMdecreased slightly to 197 kcal/mol. The experimental enthalpy of formation (0 K) of carbon from graphite is 170.0 kcal/m01.~~ This overbinding is consistent with the fact that recent L D F s in general overbind most molecules.10 In going from the 11/6/1 basis to the 1 1/7/ 1 basis, the two A, vibrational frequencies are reduced from approximately 1820 and 710 to 1570 and 600 cm-I. Raman spectroscopy suggests that they experimentally occur at 1469 and 497 cm-'.26 The LDF-optimized CMbond distances are within 0.02 A of bond distances obtained in Hartree-Fock8u27-29and ~ e m i e m p i r i c aoptimizations. l~~~ The Hartree-Fock (HF) bond distances obtained by using the most complete basis sets are 1.37 and 1.45 A for bonds shared by two hexagons and a hexagon and pentagon, respectively.* A number of fixed-geometry LDF electronic structure calculations for BF have also appeared.3638 The primitive basis sets for H and F were the 6s and 12s/7p bases of ref 21. The polarization function exponents for H and F were chosen as I .O and 1.3 bohr-*, respectively. The contraction schemes for H and F were 1,4/0,1 and 1 ,S/ 1,4/0,1, respectively. Three 9 fitting functions scaled from the second, fourth, and sixth most diffuse p-orbital functions were used in the fitting bases. These bases are tested in calculations on CH, and CF4 in Table I1 and compared with experimental results for the JANAF tables.25 The average bond energies are computed by subtracting the separated atomic energies from the total energies of the molecules and dividing the result by 4. They are compared with numbers derived from the experimental enthalpies of formation at 0 KaZ5 Thus, the experimental energies exclude one-quarter of the zero-point vibrational energies of the molecules. Despite the fact that fluorine is the most overbound atom in LDF methods, six LDF reaction energies involving fluoromethanes agree with experiment to within 2 kcal/m01,~~ suggesting that relative energies are reliable. The bond distances and vibrational frequencies are also in excellent agreement with experiment. The empirical potential energy function9 used herein is based on a formalism orginally derived by Abel14 to describe universal (24)Dunlap, B. 1.; Cook, M. In?. J . Quantum Chem. 1986. 29, 767. (25)Chase, M. W.,Jr.; Davies, C. A.; Downey, J. R., Jr.; Frurip, D. J.; McDonald, R. A.; Syvemd, A. N. J . Phys. Chem. Re5 Data 1985,14 (Suppl. I). - I . (26)Bcthune, D. S.;Meijer, G.; Tang, W.C.; Rosen, H. J. Chem. Phys. Lett. 1990, 174. 219. (27)LUthi. H. P.;Almlaf. J. Chem. Phvs. Lett. 1987. 135. 357. (28) Disch, R. L.; Schulman, J. M. Ckem. Phys. Lett. 1986, 125,465. Schulman, J. M.; Disch, R. L.; Miller, M. A.; Peck, R. C. Ibid. 1987, 141, 45. (29)Fowler, P.W.; Lawretti, P.; Zansi, R. Chem. Phys. Lett. 1990, 165, 79. (30)Marynick, D.S.;Estreicher, S. Chem. Phys. Lcrr. 1986, 132, 383. (31)Newton, M. D.;Stanton, R. E. J. Am. Chem. Soc. 1986,108, 2469. (32)Stanton, R. E.;Newton, M. D. J . Phys. Chem. 1988, 92, 2141. (33)Haddon, R. C.; Brus, L. E.; Raghavachari, K. Chem. Phys. ha. 1986, 125,459. (34)Kataoka, M.; Nakajima, T. Tetrahedron 1986, 42,6437. (35) Lbzlb, I.; Udvardi, L. Chrm. Phys. Lett. 1987, 136,418. (36)Hale, P.D. J . Am. Chem. Soc. 1986, 108.6087. (37) Satpnthy, S. Chem. Phys. fer?. 1986, 130,545. (38) Rostn, A.; Wistberg, B. J . Chem. Phys.' 1989. 90,2525. (39)Dixon, D.A.; Andzelm, J.; Fitzgerald, G.; Wimmer, E.;Jasien, P.In Dcyiry FunCrioMI Methods In Chemistry; Labanowski, J., Andzelm, J., Eds.; Springer: New York, 1991;p 33. TABLE III: HF (Ref 8) a d LDF C&, a d C& POtellthl (I) c& ShYChvrl P 8 " e t c ~ " molecule CaH60 (1) CaHa (HF) CaHa (LDF) C a F a (HF) CaFa (LDF) tYDe 1 1.55 1.57 1.55 1.60 1.61 tYDe 2 1.54 1.55 1.54 1.59 1.60 md Empirical center%-X anale. dea 1.09 1.08 1.10 1.34 1.35 178 180 180 "Bonds of type 1 are shared by a pentagon and hexagon while bonds of type 2 are shared by two hexagons. TABLE I V Occupied a d Selected Unoccupied LDF One-Electron V a h e Levels of CJ&, in eVa 8 ti, -1.29 9 h, -8.81 3 a, -12.95 5 a, -1.73 6 g, -8.86 6 h, -14.32 5 g, -9.65 4 t2, -14.68 -5.45 8 h, -9.78 2 ti, -14.97 7 h, 7 g, -5.48 2 t2, -9.91 4 ti, -16.13 3 t2, -5.69 6 t2, -9.91 3 h, -16.79 7 gu -5.94 5 g, -10.03 3 g, -17.70 1 1 h, -6.31 4 g, -10.93 5 h, -18.01 3 ti, -6.74 6 ti, -11.04 3 gu -18.94 7 t2, -6.88 5 t2, -11.13 3 tiu -19.74 6 h, -7.15 7 h, -11.78 4 h, -20.46 7 ti, -7.36 4 a, -12.25 3 tiu -21.20 IO h, -7.43 5 ti, -12.54 2 a, -21.59 4 g, -12.58 6 g, -8.20 4 h, -12.79 5 h, -8.30 " A horizontal gap in the first column separates the occupied and virtual levels. features of chemical bonding and subsequently developed by TersofP' to describe bonding in covalently bonded solids. The potential energy is modeled by a short-range Morse-like pair potential where the attractive terms of the pair potential are coupled to bonding environments by a many-body empirical function. This many-body function is closely related to a normalized Pauling bond order and depends on bond angles and local atomic coordination and whether ?r bonds are part of a conjugated system. The function is fit to the lattice constants and binding energies for the diamond lattice, graphite sheets, hypothetical simple cubic and face-centered cubic lattices of carbon, and bond energies and lengths for molecular single, double, and triple carbon-carbon bonds and carbon-hydrogen bonds. The function provides a reasonable prediction of atomization energies for hydrocarbon molecules ranging from small ring systems to large aromatic molecules as demonstrated by Table IV of ref 9. Further properties and details of the potential form are found in ref 9. 3. Fully Saturated C a w and CwFw Table I11 compares optimized CMHMand c6oF@geometries in HF8 to those predicted by using empirical potential I of ref 9 and to our LDF geometries. The agreement is excellent between our calculations. In fact, the LDF energies at the LDF and empirical potential minima differ by only 0.02 hartree. The largest bond distance between H F and the LDF geometries is 0.02 A. The slight bending away from the radial direction of the originC-H angle in the empirical potential geometry is in the plane of the origin and the C-C bond shared by two hexagons and shortens the hexagonal H-H third-neighbor distance. The LDF electronic structure of CmHM is closed-shell, 4a,~tl,7t2,3tl,l 1hg7g,3t2,7g,7h,, where these highest occupied orbitals of each Ih symmetry are ordered according to increasing one-electron eigenvalues. The electronic configuration agrees with the H F result,8 and the one-electron energy levels are given in Table IV. C60F60 has L D F electronic structure 7a,14t1,14t2,1 5g,8tl,la,l 5g,8t2,22h,l 6h,, and the one-electron levels are given in Table V. Cm has LDF electronic structure (40)Abell, G.C. Phys. Reu. E 1985, 31,6184. (41) Tersoff, J. Phys. Reu. Let?. 1986, 56,632;1988, 61,2879. Dunlap et al. 5766 The Journal of Physical Chemistry, Vol. 95, No. 15, 1991 TABLE V: Occupied a d scketed Valence Levels of COF, in eVo 16 tiu -1.25 13 ti. 16 g8-1.69 12gu16 ti, -1.79 19 h, 24 h, -2.41 12g, 6 ti, 16 g, -3.08 18 h, 15 ti, -3.46 23 h, -4.28 7 a, 12 ti, 15 ti, -5.31 12 t2, 8 a, -6.01 5 ti, 1 1 h, 16 h, -9.50 17 h, 22 h, -9.64 1 1 ti, 8 tz, -9.93 I I g, 15 g, -9.97 IO h, 1 a, -10.05 1 1 tz, 14 g, -10.08 5 tz, 8 ti, -10.15 1 1 g, 15 g, -10.24 IO gg 7 t2, -10.26 16 h, 15 h, -10.30 14g, -10.31 9 h, 6 ti, -10.55 log, 15 ha 21 h, -10.61 14 h, -10.64 IO t2, 9 g, 13 g, -10.73 14 h, 7 ti, -10.93 14 t2, -10.97 9 g, 13 h, -11.14 4 ti, 13 ha 13t2, -11.18 20 h, -11.19 IO ti, 8 g, 14 ti, -11.21 13 g, -11.27 6 a, 12 h, -11.37 9 tz,, Unoccupied LDF One-Electron -11.38 -11.49 -11.52 -11.58 -11.72 -11.98 -12.14 -12.15 -12.17 -13.98 -14.12 -14.33 -14.33 -14.34 -14.40 -14.70 -14.70 -14.70 -14.83 -14.93 -15.34 -15.40 -15.44 -15.65 -15.88 -16.09 -16.20 -16.21 -16.41 -16.43 -16.59 -16.62 -16.71 9 tiu 8 h,. 8 g, 5a 12\, 4 ti, 8 tz, 8 ti, 7 h, 7g 1, 11 7 g, 7 t2, IO h, 7 ti, 4 a, 3 tip 6 g, 6 g, 6 h, 6 ti, 9 h, 3 ti, 6 ti, 5 h, 5 g, 8 h, 5 g, 5 t2, 7 h, 5 ti, 3 a, -17.27 -17.42 -17.57 -18.18 -18.38 -18.72 -18.80 -19.75 -19.89 -20.65 -21.03 -21.71 -22.56 -23.13 -23.77 -24.12 -31.26 -31.29 -31.59 -31.79 -31.89 -31.96 -32.40 -32.53 -32.87 -33.05 -33.07 -33.32 -33.50 -33.72 -33.93 -34.05 " A horizontal gap in the first column separates the occupied and virtual levels. ? I v x F 2 w " A horizontal gap in the first column separates the occupied and virtual levels. 2t,,4aC6t,,2t2,6t2,6g,10h 6g 6h,, and the one-electron levels are given in Table VI. The LDf: electronic configurations of C,F, and C, also agree with the HF res~lts.**~'-~~ Figure 3 gives the density-of-states Lorentz broadened by 0.2eV for these three molecules. The levels of the fluorinated compound are shifted down in energy relative to the other two because of the large electronegativity of fluorine. The one-electron LDF eigenvalue gaps between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) for Cbo,CaHm and C,F, are 1.68, 3.72,and 3.48 eV, respectively. The LUMO's for Cbo,CboH,, and CWFa are 7tl,, Sa,, and 8a,, respectively. Thus, HOMOLUMO excitations are electric-dipole forbidden in all three of these molecules. The LDF energetics, bond energies, and A vibrational frequencies are given in Table VII. The LDF C-I-! and C-F bond strengths have been reduced by 44% in going from CHI to CmHWand 16% in going from CF, to CboFbo. If perfect tetrahedral bonds surrounded each carbon atom in CboHbo,then our empirical potential would yield a C-H bond i P -30 L ~ I F w e 3. PZ one-electron occupied density of states for C,, C,H,, and C,F, Lorentz broadened by 0.2 eV. Scales are chosen to yield uniform maximum peak heights; all curves integrate to the total number of valence electrons. These molecules are all closed-shell and have a large electric dipole forbidden HOMO-LUMO excitation. TABLE VII: LDF C d F , and C a w and Average C-X Bond Energies E , (Obtained by Subtracting the Total Energy of C, from C& and Dividing by 60)and A. Vibrational Frequencies molecule kcal / mol vibrational frequencies, cm-I C60H60 65 123 3020, 1390, 1210, 830 970, 750, 510, 360 Eb* C , F TABLE VI: Occupied and Selected Unoccupied LDF One-Electron Valence Levels of Cu in eVo 1 1 h, -2.07 2 ti, -10.58 6 h, -15.70 7 ti, 8 ha -10.70 4 tz, -16.79 -2.25 3 ti, 5 g, -10.85 4 ti, -17.78 -3.15 7t1, 7 h, -11.26 3 h, -17.89 -4.26 4 g, -11.75 3 g, -19.42 6 h, -5.94 6 ti, -1 1.84 5 h, -19.90 6g -7.12 4 a, -12.44 3 gu -21.18 3 tz, -22.02 IO\, 5 tz, -12.47 -7.24 4 ha -23.04 5 ti, -13.05 5 h, -8.78 3 ti, -24.02 4 h, -13.25 6 g, -8.83 2 a, -24.53 4 g, -14.07 9 h, -9.09 6 12, -9.38 3 a, -15.34 5 gu -10.10 2 ti, -15.47 P'== -10 h TABLE VIII: Coordinates of the Symmetry Inequivalent C&% Atoms atom x, A Y,A z, A C C C 2.4879 3.2697 3.8962 4.0659 4.95 17 0.6636 1S280 0.7904 2.0222 1.0154 2.2277 1.2623 H H 0.0000 1.7755 O.oo00 strength decrease of 34% relative to CH4instead of the more than 40% computed with either LDF or empirical potentials. Thus, CmHm is destabilized because geometric constraints preclude perfect tetrahedral bonding patterns around each carbon atom. The strain introduced by closing the spherical carbon shell on itself might be severe enough that lower energy configurations could exist with a few of the hydrogen (and possibly even the larger and more ionic fluorine) atoms bonded inward. 4. ThSymmetric An isomer of C60H36with the 24 unsaturated carbon atoms paired up along bonds shared by a hexagon and a pentagon to form 12 double bonds that are as far apart as possible on the surface yields one of five equivalent Th structures for the molecule, depicted in Figure 2. In all five structures the double bonds, four at a time, are bisected by three mutually orthogonal planes passing through the center of the molecule, and each double bond is now associated with a different pentagon of the original buckminsterfullerene (BF). The remaining 36 saturated carbon sites are divided by the group into a set of 12 and a set of 24 C H groups. The smaller, symmetry-equivalent set of 12 C H groups lie in the planes that bisect the 12 double bonds and are second-nearest neighbors to double-bond carbon atoms. Each carbon atom of the larger, symmetry-equivalent set of 24 C H groups is a nearest neighbor tocarbon atoms involved in two different double bonds. Table VI11 gives the optimized coordinates of the symmetry-in- The Journal of Physical Chemistry, Vol. 95, No. 15, 1991 5767 Structures of CmHm, CmFm,and CmH36 TABLE I X Total Energies in eV Relative to C, for Various C&, Species Using Both Empirical Potentials hydrogen atoms total inward potential I, kcal / mol potential 11, kcat/ mol -4.03 -4.00 -80.51 -87.38 -87.10 -78.96 -52.12 -92.76 -143.96 -143.99 -148.73 -149.69 -1 48.24 -1 16.33 -4.19 -4.17 -83.20 -90. IO -89.84 -80.47 -49.62 -95.43 -146.04 -146.08 -150.71 -151.76 -1 50.95 -1 08.8 1 ~ 20 26 34 36 36 36 36 38 58" 58' 60 60 60 60 0 0 0 0 I 12 24 0 0 0 0 1 12 24 Nearest neighbors connected by a bond shared by two hexagons. 'Nearest neighbors connected by a bond shared by a hexagon and a pentagon. h ys h XP Figure 4. Tetrahedral CmH3,with one symmetry-equivalentset of 24 hydrogen atoms pointing outward and one set of 12 symmetry-equivalent hydrogen atoms, which are darkened, pointing inward. equivalent atoms of C60H36with all the hydrogen atoms pointing outward obtained by using potential I of ref 9. These coordinates refer to a coordinate system in which the 3-fold axes of the tetrahedral group point in the (l,l,l), (1,-1,-l), etc., directions. In contrast to the case for CmH60, comparison of our empirical potential energies for CWH3,with the energies that would result from assuming perfect planar bonding for the sp2 carbon atoms and perfect tetrahedral bonding for the sp' carbon atoms suggest little or no strain energy for this less saturated species. At this empirical potential geometry, the LDF electronic structure is closed shell, 6a,6eUl la,l le,22t,27t,.11 The CmH36LUMO is 7a,, and like all BF derivatives studied herein the HOMO-LUMO excitation of 3.75 eV is electric dipole forbidden. Relative to BF and separated hydrogen atoms the LDF average C-H bond energy at this geometry is 72 kcal/mol. This average bond energy is larger than that of CmHmand would only increase upon LDF geometry optimization. Nevertheless, it is significantly closer to that of CmHm than to that of CH4, and that result would be unlikely to change upon optimization. Using our empirical potentials, we have shown that putting any of the 36 hydrogen atoms inside the BF cage raises the energy." These results are included in Table IX. It costs a quite small amount of energy, less than 0.3 eV, to point one of the hydrogen atoms of C60H36inward. Pointing 12 hydrogen atoms attached to carbon atoms that are second neighbors to the double-bonded carbon atom pairs as shown in Figure 4. costs about 0.8 eV per atom. This interesting molecule is shaped more like a cube than the rather spherical isomer formed if all hydrogen atoms point outward, which is shown in Figure 2. This cubic isomer should be rather stable if synthesized, because 0.8 eV is much smaller than the average C-H bond strength. TABLE X Energy Gained by Adding the Two Hydrogen Atom To Get C a m Where Available, and the Average C-H Bond Strengtb (Calculated as in Table VII) Using Empirical Potential I no. of H atoms 2 34 36 38 58 60 E(CmHA - Eb, E ( C m H A eV kcal / mol -4.03 46.5 54.6 56.0 56.3 57.3 57.2 -6.86 -5.38 -4.73 5. Other BF Hydrides Table IX contains further results concerning the energetics of hydrogen addition and isomerization on the BF cage. These calculations use both empirical potentials from ref 9. Potential I generates bond distances more accurately and was used above, while potential I1 generates stretching force constants more ac~ u r a t e l y .Potential ~ I gives -422.62 eV for the total energy of Cm, which corresponds to a cohesive energy of 162.4 kcal/mol per carbon atom. Potential I1 gives -419.21 eV for the total energy of Cm, which corresponds to a cohesive energy of 161.1 kcal/mol per carbon atom. Table IX concerns relatively small changes in the hydrogen atom number and bonding for stoichiometries near c m , CmH36, and CmHbO. If the addition of two hydrogen atoms to Cm or subtraction of two hydrogen atoms from CaHa is considered, then the two processes can take place at any two of the 60 carbon atoms, leading to 32 distinct molecules in both cases. We have only considered altering the bonding of neighboring carbon atoms, which leads to two cases. The first is indicated by a superscript a in the table and corresponds to nearest neighbors connected by a bond shared by two hexagons. The second is indicated by a superscript b in the table and corresponds to nearest neighbors connected by a bond shared between a pentagon and a hexagon. The energy differences between these two cases is quite small, being less than 0.04 eV using either potential. We have also considered bonding hydrogen atoms to the inside of the carbon cage. This widens the scope of possible molecules greatly. To limit the scope of such studies, we have only considered pointing 1, 12, and 24 atoms inward. For one atom pointing inward we have included in Table IX only the results for the mast stable isomer. It is energetically favorable to point one hydrogen atom inward for CWHmbut not for CmH36. For more than one atom pointing inward, we have chosen to preserve Th symmetry. These molecules are significantly less stable than the corresponding molecules with all hydrogen atoms pointing outward except for the case of 12 atoms pointing inward in CmHm. Table X considers only molecules with all hydrogen atoms pointing outward. For the most stable isomer it lists the average C-H bond energy and, for the calculations that have been done, the energy gained by adding the last two hydrogen atoms. By a substantial margin, adding hydrogen atoms to complete the C60H36structure is energetically more favorable than going to higher levels of saturation. 6. Conclusions We have optimized the LDF Ih geometries of CpolCmHm, and C60Fm. These are all closed-shell molecules, with substantial HOMO-LUMO one-electron energy gaps, although the C-H and C-F average bond energies are reduced by 44% and 16% from those of methane and tetrafluoromethane. LDF bond energy considerations and empirical potential strain energy considerations suggest that a fully saturated CwH, molecule is somewhat less stable than C60H36. Empirical potentials suggest that CmHm would prefer to have at least one hydrogen atom point inward. For the case of C60H36bonding one hydrogen atom inward is almost neutral energetically. These results sugest that if appropriate chemical methods are developed, perhaps based on osmylation,42 adding substituents inside and outside BF could lead to (42) Hawkins, J. M.; Lewis, T. A.; Loren, S.D.; Meyer, A,; Heath, J. R.; Saykally, R. J. J . Org. Chem. 1990, 55, 6250. 5768 J . Phys. Chem. 1991, 95, 5768-5773 an incredibly rich set of new molecules of varying shape. Putting fluorine atoms inside the cage could also dramatically alter the response properties of C,Fm. Acknowledgment. These calculations were made using a grant of CRAY X-MP computer time from the US.Naval Research Laboratory. This work was supported in part by the Office of Naval Research (ONR) through the Naval Research Laboratory and through the ONR Chemistry and Materials Science Divisions Contracts NOOO14-91-WX-24150 and "14-91-WX-24154. Structures and Vibrational Frequencies of Cbo,C,o, and C8, Krishnan Raghavachari* AT& T Bell Laboratories, Murray Hill, New Jersey 07974 and Celeste McMichael Roblfing* Sandia National Laboratories. Livermore, California 94551 -0969 (Received: March 28, 1991) The structures and vibrational frequencies of C, C7&and CMhave been determined with the semiempirical MNDO method. Two highly stable minimum-energy structures have been identified for CE4. Of the two, a D6h form, which contains five distinct carbon atoms, is found to be slightly more stable than a Td form with four distinct carbons. Both isomers are found to be significantly more stable than either Csoor C7,,, consistent with the presence of a greater proportion of six-membered rings in CE4.An additional structure containing seven-membered rings has also been considered for CM,but is significantly less stable than structures containly only five- and six-membered rings. Finally, the geometries of C, C70, and the two stable forms of CE4have been completely optimized by using ab initio Hartree-Fock theory with the double-f 3-21G basis set. Introduction The recent macroscopic preparation' of spheroidal carbon clusters* has stimulated a wide variety of experimental and theoretical studies on these novel systems. C, and CT0have now been isolated by several different experimental groups and their properties have been m e a ~ u r e d . ~Many ' ~ theoretical studies have been reported on C, and a few on C70.'8-36Among the other E. Nature 1985, 318, 162. (3) KrHtschmer, W.; Fostiropoulos, K.; Huffman, D. R. Chem. Phys. Leu. 1990, 170, 167. (4) Taylor, R.; Hare, J. P.; AMul-Sada, A. K.; Kroto, H. W. J . Chem. Soc., Chem. Commun. 1990, 1423. ( 5 ) Aije, H.; Alvarez, M. M.; Anz, S. J.; Beck, S. D.; Diederich, F.; Fostiropoulos, K.; Huffman, D. R.; Kritschmer, W.; Rubin, Y.; Schriver, K.; Sensharma, D.; Whetten, R. L. J . Phys. Chem. 1990,94, 8630. (6) Bethune, D. S.;Meijer, G.; Tang, W. C.; Rosen, H. J. Chem. Phys. Len. 1990, 174, 219. (7) Bethune, D. S.;Meijer, G.; Tang, W. C.; Rosen. H. J.; Golden, W. G.; Seki, H.; Brown, C. A.; deVrics, M.S . Chem. Phys. Lett. 1991, 179, 181. (8) Cox, D. M.; Bchal. S.;Disko, M.; Gorun, S.; Greaney, M.; Hsu, C. S.; Kollin, E.;Millar, J.; Robbins, J.; Robbins, W.; Sherwood, R.; Tindall, P. J . Am. Chem. Soc. 1991,113.2940. (9) Haufler, R. E.; Conceicao, J.; Chibante, L. P. F.; Chai, Y.; Byrne, N. E.; Flanagan, S.;Haley, M. M.; OBrien, S.C.; Pan, C.; Xiao, 2.;Billups, W. E.; Ciufolini, M. A.; Hauge, R. H.; Margrave, J. L.; Wilson, L. J.; Curl, R. F.; Smalley, R. E. J . Phys. Chem. 1990, 94, 8634. (IO) Haufler, R. E.; Chai, Y.; Chibante, L. P. F.; Conceicao, J.; Jin, C.; Wang, L.-S.; Maruyama, S.;Smalley, R. E. Mater. Res. Soc. Symp. Proc., in press. (11) Arbogast, J. W.; Darmanyan, A. P.; Foote, C. S.;Rubin, Y.; Diederich, F. N.; Alvarez. M.M.; Anz, S.J.; Whetten, R. L. J. Phys. Chem. 1991, 95, 1 1 . (12) 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. Mater. Res. Soc. Symp. Proc., in press. (13) Haddon, R. C.; Schneemeyer, L. F.; Waszczak, J. V.;Glarum,S. H.; Tycko, R.; Dabbagh, G.; Kortan, A. R.; Muller, A. J.; Mujsce, A. M.; Rosseinsky, M.J.; Zahurak, S.M.; Makhija, A. V.; Thiel, F. A.; Raghavachari, K.; Cockayne. E.; Elser, V. Nature 1991, 350,46. (14) Tycko. R.; Haddon. R. C.; Dabbagh, G.; Glarum, S.H.; Douglas, D. C.; Mujsce, A. M. J . Phys. Chem. 1991, 95, 518. ( I S ) Allemand, P. M.; Koch, A.; Wudl. F.; Rubin, Y.; Diederich, F.; Alvarez, M. M.; Anz, S.J.; Whetten, R. L. J . Am. Chem. Soc. 1991, 113, 1050. (16) Yannoni, C. S.; Johnson, R. D.; Meijer, G.; Bethune, D. S.;Salem, J . R. J . Phys. Chem. 1991, 95,9. (17) Yannoni, C. S.;et al. Materials Research Society Meeting, Boston, 1990. 0022-3654/91/2095-5768$02.50/0 TABLE I: Bond Leugtbs (A) rad Bond Aagles (deg) for Cm param' MNDO HF/3-21G CI-CI 1.473 1.452 c1-Cl 1.402 I.370 cl-c3 1.469 1.447 c3-c3 1.389 1.356 c3-c4 1.478 1.458 C4C4 1.442 1.414 c4-cs 1.430 1.403 cs-cs 1.484 1.475 cI-cI-cI 108.0 108.0 c,-cI-cl 119.7 119.7 c&-C3 120.3 120.2 c3c2-c3 106.9 106.9 c1c3-c3 120.1 120.1 CZ-C3<4 108.5 108.3 c3-c3-c4 119.8 119.9 c3-c4-c4 108.1 108.2 c3-c4-cs 121.4 121.5 c4-c4-c5 121.6 121.3 c4-cs-c5 118.7 118.4 c4-c5-c4 115.6 116.3 For the numbering system, see Figure 3. possible clusters that have been seen in different experiments, Cs4 appears to be the next most significant fraction. However, only (18) Haddon, R. C.; Brus, L. E.; Raghavachari, K. Chem. Phys. Lcrr. 1W6,125, 459; Ibid. 1986, 131, 165. (19) Klein, D. J.; Schmalz, T. G.; Hite, G. E.;Seitz, W.A. J. Am. Chem. Soc. 108. . -. 1986. - .- -, . - 1301. .- - .. (20) Marynick, D. S.;Estreicher, S.Chem. Phys. Lerr. 1986, 132,2469. (21) Fowler, P. W.; Woolrich, J. Chem. Phys. Lea. 1986, 127, 78. (22) Disch, R. L.; Schulman, J. M. Chem. Phys. Lerr. 1986, 125, 465. (23) Newton, M. D.; Stanton, R. E. J . Am. Chem. Soc. 1986,108,2469. (24) Luthi, H. P.; AlmlBf, J. Chem. Phys. Lcrr. 1987, 135, 357. (25) Elser. V.; Haddon, R. C. Nature 1987, 325, 792. (26) Schulman, J. M.; Disch, R. L.; Miller, M. A.; Peck, R. C. Chem. Phys. Lerr. 1987, 141. 45. (27) Stanton, R. E.; Newton, M. D. J . Phys. Chem. 1988, 92, 2141. (28) Rosen, A.; Wastberg, B. J . Am. Chcm. Soc. 1988, 110, 8701. (29) Slanina, 2.;Rudzinski, J. M.; Togasi, M.; Osawa, E. J . Mol. Srrucr. 1989, 202, 169. (30) Weeks, D. E.; Harter, W. G. J. Chem. Phys. 1989,90,4144. (31) Fowler, P. W. J . Chem. SOC.,Faraday Trans. 1990,86, 2073. (32) Fowler, P. W.; Laueretti, P.;Zanasi, R.Chem. Phys. Leu. 1990,165, 79. -. 0 1991 American Chemical Society