Ab initio prediction of the structures and stabilities of the

Novel Beltlike and Tubular Structures of Boron and Carbon Clusters ..... ACS Publications Library Summit Series recently touched down in India for fou...
0 downloads 0 Views 1MB Size
J. Am. Chem. SOC.1991, 113, 9045-9054 tween the Cp ligands and the two guanosine residues as well as between the Cp ligands and the phosphate backbone (Figure 18). The modeling studies show that Cp2MoZ+-nucleobase bidentate coordination to mutually tipped guanosine residues such as those in c i ~ - ( P t ( N H , ) ~ ~ d ( ~ p isGalso ) ~ ]sterically ~*~ impaired. These observations are in agreement with results to be reported elsewhere67 which indicate that the binding of Cp,VCl,(aq) or Cp,MoCl,(aq) to DNA plasmids and the subsequent effects on DNA processing enzymes are greatly different from those of ~ i s p l a t h . ~While * ~ N M R dataprovide no evidence for strong Cp2MoCl,(aq) binding to various ribose/nucleobase-terminated model oligonucleotides, binding is observed to S'-phosphate terminated 0ligonucleotides.6~Thus, it is likely that the mechanism(s) of Cp2MX2antineoplastic activity is (are) quite different from (67) (a) Liu, A. H.; Kuo, L. Y.; Marks, T. J., manuscript in preparation. (b) Liu, A. H.; Marks, T. J., manuscript in preparation.

9045

that (those) of cisplatin. It if furthermore conceivable that mechanisms are different for different metallocene drugs, and these possibilities are presently under investigation. Acknowledgment. This research was supported by NSF (Grant CHE8800813). L.Y.K. was a Dee and Moody Predoctoral Fellow. We thank Dr. W. C. Finch for assistance with the solid-state 31P NMR spectroscopy and Dr. A. H. Liu for assistance with the molecular graphics. Supplementary Material Available: Tables of atomic coordinates (Tables 111, IV, and IX) and anisotropic thermal parameters for 3a, 4, and 5, hydrogen atom positions for 3a and 4, and IH-IH coupling constants and derived conformational popu~ations for 5, 6, and 7a (10 pages); listings of observed and calculated structure factors from the final cycles of least-squares refinement for 3a9 4, and 5 (78 pages). Ordering information is given on any current masthead page.

Ab Initio Prediction of the Structures and Stabilities of the Hyperaluminum Molecules: A130 and Square-Planar A140 Alexander I. Boldyred and Paul von R. Schleyer* Contribution from the Institut fur Organische Chemie, Friedrich-Alexander Uniuersitat Erlangen- Nurnberg, Henkestrasse 42, 0-8520 Erlangen, Germany. Received February 4, I991

Abstract: The concept of hypermetalation, characterized by molecules with unprecedented stoichiometries, is extended to the aluminum-oxygen combinations, A130 and A140. Their equilibrium geometries and fundamental frequencies, as well as those of the isolated reference species AIO, AIz, AI2+,AI20, A130+,AI30, AI3+, and A140, were calculated at HF/6-31G* and at various correlated levels, e.g., MP2(fu11)/6-31 G*.Extensive searches of possible structures and electronic states were carried out. The global minima are: linear A120 (Dah, IZ +), planar AI,O+ (D3,,,IA1'), planar A130 (the C, Y(,A1) and T(2B2)forms have nearly the same energy), tetrahedral AI? (Td,]Al),and planar AI40 (D4*,IAlg). All these species are very stable (with the exception of AI:+) with regard to all possible decomposition pathways. Representative dissociation energies (in kcal/mol at PMP4SDTQ/6-31 I+G*//MP2(full)/6-31G*+ZPE) are: A130+, into AI20 + AI+ (34.1); AI30, into A120 + AI (19.9); Al40, into AI30 + AI (45.5) or into AI20 + AI2 (37.1). Although the aluminum-oxygen attraction is largely ionic, aluminum-aluminum bonding contributes significantly to the stability of the hyperaluminum A130 and A140 molecules.

Introduction We have discovered a remarkable molecule, AI40, computationally. The oxygen is surrounded by four aluminum atoms in a square-planar ( D4& arrangement. The electronic structure, combining ionic and substantial metal-metal bonding, anticipates a large, new class of similar molecules. Such "hypermetalation" involving alkali metal stoichiometries exceeding normal valence expectations is now well-documented. Many of hyperlithium molecules (OLi4, OLi,, OLi6, NLi5, CLi5, CLi6, BLi5, BeLi,, BeLi,, Cs40, etc.) were discovered computationally.'-" Li30, Li40, and Li50 have been observed mass spectrometrically and the atomization energies determined.I2*l3 There is similar evidence for Na2C1,I4Na30, Na40, K30, Kg0,15J6 and CssO.l7 The 'suboxides" of rubidium and cesium, e.g., Rb902, Cs70, and Cs, have been c h a r a c t e r i ~ e d . ' ~ ~The '~ substantial stability of these molecules is due to the high degree of ionic character as well as bonding interactions between the ligand a t o m ~ . l - In ~ a sense, hypermetalated species can be regarded as metal clusters bound ionically to a centrally located "impurity" heteroatom. This contrasts with the usual situation in which the only bonding interactions are between the central atom and its attached atoms or ligands, e.g., in methane, where the ligand-ligand interactions are repulsive. Hence, the usual

'

Permanent address: Institute of Chemical Physics, USSR Academy of Sciences, Moscow V-334, Kosygin str. 4, USSR.

0002-7863/91/15 13-9045$02.50/0

valence theory, which does not include all the possible interatomic interactions as bonding possibilities, must be modified. ( I ) Schleyer, P. v. R.; Wurthwein, E.-U.; Pople, J. A. J . Am. Chem. Soc. 1982, 104, 5839.

(2) Schleyer, P. v. R. New Horizons of Quantum Chemistry; Lowdin, P.-O., Pulman, A., Eds.; Reidel: Dordrecht, 1983; pp 95-105. (3) Schleyer, P. v. R.; Wurthwein, E.-U.; Kaufmann, E.; Clark, T. J . Am. Chem. SOC.1983, 105, 5930. (4) Schleyer, P. v. R.; Tidor, B.; Jemmis, E. D.; Chandrasekhar, J.; Wurthwein, E.-U.; Kos, A. J.; Luke, B. T.; Pople, J. A. J . Am. Chem. SOC. 1983, 105,484. (5) Schleyer, P. v. R. Pure Appl. Chem. 1984, 56, 151. (6) Schlever. P. v. R.: Reed. A. E. J . Am. Chem. SOC.1988. 110.4453. (7) Reed,-A. E.; Schleyer, P.'v. R.; Janaschek, R. J . Am. C h e k Si.1991, 113, 1885.

( 8 ) Pewestorf, W.; Bonacic-Koutecky, V.; Koutecky, J. J . Chem. Phys.

1988.89. 5794.

(9) Fantucci, P.; Bonacic-Koutecky, V.; Pewestorf, W.; Koutecky, J. J . Chem. Phys. 1989, 91, 4229. (IO) Klimenko, N. M.; Musaev, D. G.; Gorbik, A. A.; Zyubin, A. S.; Charkin, 0.P.; Wurthwein. E.-U.; Schleyer, P. v. R. Kmrd. Khim. 1986, 12, 60 1

Savin, A,; Preuss, H.; Stoll, H. Reo. Roum. Chim. 1987, 32, 1069. Wu,C. H.J . Chem. Phys. 1976, 65, 3181. Wu, C.H.Chem. Phys. Lett. 1987, 139, 357. Peterson, K. I.; Dao, P. D.; Castleman, A. W., Jr. J . Chem. Phys. 1983, 79, 777. (15) Dao, P. D.; Peterson, K. 1.; Castleman, A. W., Jr. J . Chem. Phys. 1984, 80, 563.

(I I) (12) (13) (14)

0 1991 American Chemical Society

9046 J . A m . Chem. SOC.,Vol. 113, No. 24, 1991 a

PJ -Al

Al -4

2.539 (2.469) 12.4661

(3.138)

3.214

AI -0

Boldyrev and Schleyer n

1.697 (1.648) [ 1.6181

A10 (%I

A-0-4

I 1703

A1,O

(1733)

2.952

N

3.257

(Td;3A1) 131.4

1.887

2.877

(1.909)

(2.836)

A120 (C2v;3B2)

d 3.047 (3.005)

t 4-4

--AI (2.781) Al40 ( C Z v , I1:'Al)

b 92.1 (91.5)

1.757 (1.783)

1.912 (1.936)

Al 1.692 1.700

A

2.510

*' 121.0Figure 1. Optimized geometries of the aluminum oxides (bond length in A and valence angles in deg). Geometries at MP2(fu11)/6-3IG1 and experimental data are given in parentheses and square brackets, respectively; other data at HF/6-31G*.

Although having the usual "valence", SiLi4 also provides an instructive illustration. Metal-metal bonding contributes to the surprising preference for a nontetrahedral, Cb point group ge-

me try.^,' Hypermetalation should be a general phenomenon, exhibited by many if not all the metals. In fact, elements, M, which form stronger M-M bonds, may well be even better candidates. We have now chosen aluminum for further exploration and describe an ab initio examination of the structure and stabilities of the first examples of hyperaluminum molecules, A130 and A140. The former has already been detected among the reaction products of aluminum with oxygen in the gas phase.20 Matrix isolation (16) Wiirthwein. E . 4 . ; Schleyer, P. v. R.; Pople, J. A. J . Am. Chem. Soc. 1984, 106. 6913.

Fallgren, H.; Martin, T. P. Cfiem. Pfiys. Leu. 1990, 168, 233. Simson, A. Strucr. Bonding (Berlin) 1979, 36, 81. Borgstedt, H . U . Top. Curr. Chem. 1986, 134, 125. Cox, D. M.; Trevor, D. J.; Whetten, R. L.; Rohlfing, E. A.; Kaldor, A. J . Cfiem. Pfiys. 1986, 84, 465 1. (17) (18) (19) (20)

and gas-phase investigations of various aluminum-oxygen species also are e x t e n s i ~ e . ~ *A10, ~ ~ AIOz, AI20, Alz02,and A103 have (21) Linevski, M. J.; White, D.; Mann, D. E. J. Cfiem. Phys. 1964, 41, 542. (22) Snelson, A. J . Pfiys. Cfiem. 1970, 74, 2574. (23) Makowiecki, D. M.; Lynch, D. A., Jr.; Carlson, K. D. J . Pfiys. Cfiem. 1971. 75. 1963. (24) Marino, C. P.; White, D. J . Pfiys. Cfiem. 1973, 77, 2929. (25) Lynch, D. A., Jr.; Zehe, M. J.; Carlson, K. D. J . Pfiys. Cfiem. 1974, 78, 236. (26) Finn, P. A.; Gruen, D. M.; Page, D. L. Ado. Cfiem. Ser. 1976, No. ISR. . - -, 30. - -. (27) Zehe, M. J.; Lynch, D. A., Jr.; Kelsall, B. J.; Carlson, K. D. J. Phys. Cfiem. 1979, 83, 656. (28) Kelsall, B. J.; Carlson, K. D. J . Pfiys. Cfiem. 1980, 84, 951. (29) Sonchik, S. M.; Andrews, L.; Carlson, K. D. J . Pfiys. Cfiem. 1983, 87, 2004. (30) Bares, S. J.; Haak, M.; Nibler, J . W. J . Cfiem. Phys. 1985, 82,670. (31) Maltsev, A. A,; Shevelkov, V. F. Tepl.fiz. Vys. Temp., Acad. Nauk SSSR 1964, 2, 650. (32) Jarrold. M. F.; Bower, J. E. J . Cfiem. Pfiys. 1986, 85, 5373. (33) Jarrold, M. F.; Bower, J. E. J . Cfiem. Pfiys. 1987, 87, 1610.

Hyperaluminum Molecules: A130 and AI,O A120

9041

J . Am. Chem. SOC.,Vol. 113, No. 24, 1991

p

OW/6-31Q*

>

AE

n

d

b

Figure 2. Relative energies of the different forms of AI20 (a), A130 (b), A140 (c) at HF/6-31G*, and AI40 (d) at MP4SDTQ/6-31G*, Table I. Experimental and Calculated Dissociation Energies at Different Approximations Reaction Energies (kcal/mol) PUHG/ UMPZ(full)/ PMP2/ PMP3/ PMP4/ PMP2/ PMP3/ PMP4/ PMP4+ZPE/ reaction 6-31G* 6-31G* 6-31G* 6-31G8 6-31G* 6-311+G* 6-311+G* 6-31 I+G* 6-311+G* 1 2 3 4 5 6 7 8 9 IO +52.3 AI0 AI + 0 +97.2 +98.1 +89.6 +96.5 +101.7 +91.3 +100.8 +100.3 +14.2 A12 AI AI +26.9 +25.7 +26.8 +28.0 +25.9 +27.0 +28.3 +27.2 +30.8 AI2+ AI AI+ +26.6 +28.6 +30.2 +29.8 +30.1 +31.1 +30.7 +31.0 AI20 A I 0 AI +114.1 +150.0 +143.2 +137.1 +137.4 +142.5 +137.1 +138.0 +136.1 AI20 AI, 0 +152.3 +217.9 +215.6 +199.9 +205.9 +218.2 +201.5 +210.4 +209.1 +3.1 AI3+ AI2 AI+ +36.7 +34.5 +29.4 +31.9 +35.0 +29.9 +32.6 +32.2 AI,+ -AI2+ AI -1 3.7 +35.0 +26.1 +29.9 +28.7 +29.8 +30.1 +26.4 +29.8 +27.8 +37.6 AI,O+ A120 AI+ +36.3 +37.1 +34.0 +34.7 +34.1 +37.7 +33.5 AI30 A120 AI + I 3.3 +20.9 +21.8 +18.6 +20.3 +19.9 +20.5 +22.1 +19.9 A130 AI2 A10 +115.1 +141.6 +139.3 +130.7 +131.5 +128.8 +130.0 +128.6 +136.4 Al3+ AI+ -54.7 A140 AI30 AI +6.8 +52.2 +47.4 +44.0 +47.6 +45.6 +42.6 +46.0 +45.5 AI& A120 AI2 +5.9 +46.2 +43.5 +37.8 +41.7 +39.5 +34.3 +38.0 +37.1 ,I Reference 50. Reference 70. Reference 6 I .

--- ++ - ++ - ++ -- +++ -- +++ +

expt 11

+121.5 +35.7 +20.8 +127.6 +210.3 +30.0 +25.8

f 0.9O

* 1.00

f 6.9b f 2.00 &

4.00

f 8.1' f 8.1'

+

+

been identified.22v29In the gas phase, AI20, A130,20and A1,0, ( n = 2-26; m = 1, 2 ) have been demonstrated by mass spect r ~ m e t r y . ~ * - Ab ' ~ initio calculations are available for A10,38 A120,3e4iAI2O2,A102, and A103.40 (34) Ruatta. S. A.; Hanley, L.; Anderson, S. L. Chem. Phys. Lerr. 1987, 137, 5 .

(35) Jarrold, M. F.; Bower, J. E. J . Chem. Phys. 1987, 87, 5728. (36) Ruatta, S.A.; Anderson, S. L. J . Chem. Phys. 1988, 89. 273. (37) Ovchinnikov, 1. V.; Serebrennikov, L. A,; Maltsev, A. A. Zh. Fir. Khim. 1985, 59, 1558. (38) Lengsfield, 9. H.; Liu, 9. J . Chem. Phys. 1982, 77, 6083. (39) Wagner, E. L. Theor. Chim.Acta 1974. 32. 295. (40) Masip, J.; Clotet, A.; Ricart, J. M.; Illas; F.;'Rubio, J. Chem. Phys. Lea. 1988, 144, 373.

Despite the experimental detection of A130,20neither structural nor energetic data are available. We have not been able to find any mention of A140 in the literature. Computational Methods Geometries of AI2, AI2+, AIO, A120, AI,O+, AI,?, AI3+, AI:+, and using a poA140 were optimized by employing analytical larized split-valence basis set (6-31G*)43*Uat H F and at correlated MPZ(ful1) levels (UHF and UMPZ(ful1) for open-shell systems). Ge(41) Solomonik, V. G.; Sazonova, l. G. Zh. Neorg. Khim.1985,30,1939. (42) Schlegel, H. B. J . Compur. Chem. 1982, 3, 214. (43) Hariharan, P. C.; Pople, J. A. Theor. Chim.Acra 1973, 28, 213. (44) Frisch, M. J.; Pople, J. A.; Binkley, J. S.J . Chem. Phys. 1984, 80, 3265.

9048 J . Am. Chem. SOC.,Vol. 113, No. 24, 1991

Boldyrev and Schleyer

Table 11. Calculated Frequencies (cm-I) and Zero-Point Energies (ZPE, kcal/mol) of the Aluminum Oxide Species species symmetry 1

A I 0 (22+)

2

UHF/ 6-31G* 3 810 I .2 316 0.4 163 0.2 550 1 I2 1043 2.6 1 I95 264 97 2.4 764 426 424 2.3 257 252

4 (45.7) (0.0) (0.0) (0.0) (0.0) (741.4)

UMPZ(full)/ 6-31G* 5 789 1.1 366 0.5 176 0.3 513 90 988 2.4

(53.6) (66.3) (45.5) (60.1) (16.0) 105.8) (0.0) (1 4.4)

1.1

156 (0.0) 259 190.3) 78 (0.1) 0.8 403 (0.0) 509 (291.3) 168 (7.8) 235 (1.1) 2.9 759 (540.9) 421 (7.1) 233 (1.6) 170 (4.3) 326 (1.8) 86 (1 0.5) 2.9 450 383 146 169 66 1 149i 2.6 75 46 33 0.4 396 7351 533 IR intensities (KM/mol) are given in parentheses.

382 317 1.5 161 272 62 0.8 385 514 146 205 2.7 740 403 334 213 139 76i 2.6 429 397 153 164 643 102 2.7

ometries of A120 (0.3 and AllO (C,, T) were optimized also at MP2(full)/6-31+G*, but the resulting bond lengths (R(AI-O) = 1.742 A in AI20; R(AI,-O) = 1.997 A, R(A12-0) = R(AIl-O) = 1.839 A in A130) and valence angle (LAI,OAI~= 99.3' in AIIO) changed only by 0.02 A and 0 . 5 O , respectively, from those at MP2(fu11)/6-31G*. The results are summarized in Figure 1, which includes the experimental distances for A I 0 and AI2. Fundamental frequencies, normal coordinates, and zero-point energies (ZPE) were calculated by using standard FG matrix methods. The MP2(fu11)/6-31G* equilibrium geometries were used to evaluate electron correlation corrections in the frozen-core approximation by Moller-Plesset perturbation theory to full fourth order4s using 6-31G* and 6-31 I+G* basis sets (MP4SDTQ/6-31G* and MP4SDTQ/6-31 I+G*). The U H F wave functions for open-shell systems were projected to pure spectroscopic states (PUHF, PMP2, PMP3, and PMP4).' The (U)QCISD(T) level" was employed for some of the species as well. Optimizations and analytical frequency calculations at (45) Krishnan, R.; Pople, J. A. Inr. J . Quantum Chem. 1978, 14, 91. (46) Schlegel, H. B. J . Chem. Phys. 1986, 84,4530. (47) Pople, J. A.; Head-Gordon, M.; Raghavachari, K. J. Chem. Phys. 19a7,87, ~ 6 8 .

species symmetry 1

2

UHF/ 6-31G* 3 265 20 1 2.4 390 171 287 45 1 173 207 88 3.4 773 470 274 214 379 150 22 190 54 3.6 1037 541 216 59 329 135 47 153 74 3.7 370 228 172i 262 36i 330 161i 2.1 352 128 527 202 459 210 196 44 1 157 52

4

(0.0) (0.0) (0.0) (234.4) (29.1) (12.5) (0.0)

(781.6) (9.3) (9.4) (25.6) (14.3) (0.0) (1 2.0) (0.6) (1.6) . ,

UMPZ(full)/ 6-31G' 5

384 204 268 538 246 157 90 3.8 762 423 284 233 40 1 159 921 137 24 3.5

(840.7) (6.5) (3.4j (6.8) (3.7) (11.1) (1.3) (22.5) (4.3)

HF/6-31G* and at MP2(fu11)/6-31G* were carried out with the CADPAC Program." GAUSSIAN 88 (CONVEX version) was used to obtain the PMP2, PMP3, PMP4SDTQ, and QCISD data and for the natural population analy~is.4~The total energies at HF/6-31G* and at different correlated levels are given in Tables IS and 2s. Relative energies of the different forms of AI20, AI30, and A140 at UHF/6-31G* and A140 at MP4SDTQ/6-3 1G* are presented on Figure 2. Dissociation energies and harmonic frequencies are given in Tables I and 11, respectively.

Results Evaluation of the stabilities of A130 and A140 requires data for Al, 0, A10, AI,, A12+,and AI20. The A12+,All+, A130+,and Al,,+ cations also were examined to help understand the nature (48) CADPAC. Amos, R. D.; Rice, J. E. CADPAC: The Cambridge Analytic Deriuariues Package, Issue 4.0; Cambridge, 1987. (49) GAUSSIAN 88 (CONVEX version). Frisch, M.J.; Head-Gordon, M.; Schlegel, H. B.; Raghavachari, K.; Binkley, J. S.;Gonzales, C.; DeFrees, D. J.; Fox, D. J.; Whiteside, R. A,; Seeger, R.; Melius, C. F.; Baker, J.; Martin. R.;,Kahn, L. R.; Stewart, J. J. P.; Fluder, E. M.; Topiol, S.;Pople, J. A. Gaussian Inc.: Pittsburgh, PA, 1988.

Hyperaluminum Molecules: AIJO and AI,O

J . Am. Chem. Soc., Vol. 113, No. 24, 1991 9049

of the bonding in A130 and A140. We describe our results on these reference systems first. Neutral AIO. The ground state of A10 is known to be X22+ with an equilibrium distance of 1.6179 A, an equilibrium frequency we = 979.23 cm-I, and energy of dissociation Doo = 121.5 f 0.9 kcal/mol. We optimized the geometry of A I 0 ( W )( ( 1 ~ ) ~ (20)~(1 ~ ) ~ ( 3 uvalence )l configuration) at UHF/6-31G* and at UMP2(fu11)/6-3 IC*. Appreciable spin contamination was found at both levels ( ( S 2 )= 0.79). Spin-projecting the wave function to the pure spectroscopic state, 2Z+ ( (S2) = 0.750), reduced the total energy of A10 modestly, e.g., by 0.0046 au at PMP2 6-31G*. At UMP2(fu11)/6-31G*, the A I 4 bond length (1.648 Figure 1) is longer, and the harmonic frequency (789 cm-I, Table 11) is smaller, than experiment. The electron correlation correction to the dissociation energy is quite large, 40-50 kcal/mol; this value oscillated along the PMP2-PMP3-PMP4 series. At PMP4SDTQ/6-311 +G*//MP2(full)/6-3lG* + ZPE, the calculated Do is 20.6 kcal/mol and at QCISD(T)/6-311+G* 14.2 kcal/mol lower than the experimental value. The use of the experimental A I 0 distance has a negligible effect. Adding a second set of d-functions and one set of f-functions to the basis only improves the calculated Do to 113.5 kcal/mol ((U)QCISD(T)/6-31 l+G(2df)//MP2(full)/6-3lG* + ZPE at the experimental geometry). Neutral AI2. Dialuminum, AI2, has already been studied very carefully.s1-s8 According to the most extensive CASSCF-SOCI calculations with very large basis sets and relativistic corrections, the 3Z; ((lu8)2(luu)2(2u~)'(2uu)~) state lies 174 cm-I (0.6 kcal/mol) above the 'II, (( 1uJ2( 1u , ) ~ ( ~ u , )1~ ~( ~ground ) ~ )statess2 Aluminum clusters offer challenges for quantum chemistry. S C F convergence is often difficult to achieve, or leads to excited states. Open-shell states at U H F exhibit considerable spin contamination. Moreover, the potential energy surface for open-shell systems may not be smooth at both U H F and PUHF levels (see discussion in ref 59). We have chosen to employ the 3Z; state for AI2 as our reference; convergence is achieved easily and spin contamination is insignificant (e.g., at UHF/6-31G* ( S 2 ) = 2.025). The 0.6-kcal/mol difference in energy between the 3Z; and the ground 311ustates is negligible. Hence, correction of the dissociation energy estimates for A130 and A140 is hardly necessary (the inaccuracies in our De estimates are larger). The bond length of AI2 (3Z.J at UMP2(fu11)/6-31G* (2.469 A) is very close to the experimental value (2.466 A, Figure 1). Inclusion of correlation energy is essential to obtain reliable data for the calculated bond length, frequency, and dissociation energy of AI2 (3Zg-)* The AI2+ Cation. The AI2+ ion has been well studied experimentallym4 and theoretically.s6,6s,66All ab initio calculations predict X2Zg+ ((4uJ2(5uJ1) ground electronic state. Sunil and

A,

(50) Huber, K. P.; Herzberg, G. Molecular Spectra and Molecular Structure. Constants of Diatomic Molecules; Van Nostrand Reinhold: New York. 1979. (51) Bauschlicher, C. W., Jr.; Pettersson, L. G. M. J . Chem. Phys. 1987, 87,2198. (52) Bauschlicher, C. W., Jr.; Partridge, H.; Langhoff, S. R.; Taylor, P. R.;Walch, S. P. J . Chem. Phys. 1987, 86,7007. (53) Tse, J. S. J . Mol. Struct. (THEOCHEM) 1988, 165,21. (54) Basch, H.; Stevens, W. J.; Krauss, M. Chem. Phys. Lett. 1984, 109, 212. ( 5 5 ) Ginter, D. S.; Ginter, M. L.; Innes, K. K. Astrophys. J . 1964, 139, 365. (56) Upton, T. H. J. Chem. Phys. 1987, 86, 7054. (57) Douglas, M. A.; Hauge, R. H.; Margrave, J. L. J . Phys. Chem. 1983, 87. 2945. (58) Abe, H.; Kolb, D. M. Ber. Bunsenges. Phys. Chem. 1983, 87, 523. (59) Tse, J. S. J. Chem. Phys. 1990, 92, 2488. (60) Jarrold, M. F.; Bower, J. E.;Kraus, J. S. J . Chem. Phys. 1987, 86, 3876. (61) Hanley, L.; Ruatta, S. A.; Anderson, S. L. J. Chem. Phys. 1987,87, 260. (62) Cox, D. M.; Trevor, D. J.; Whetten, R. L.; Kaldor, A. J. Phys. Chem. 1988, 92, 421. (63) Fuke, K.; Nonose, S.; Kikuchi, N.; Kaya, K. 2.Phys. 1989,012,571, (64) Harrington, J. E.; Weisshaar, J. C. J . Chem. Phys. 1990, 93, 854.

Jordan6s obtained at CCD+ST(CCD)/ [ 12~9p3dlf/6~5p3dlfl 3.208-A bond length, 169-cm-I equilibrium frequency, and 31.6-kcal/mol dissociation energy for Al; (22g+), as well as 6.19and 5.92-eV vertical and adiabatic first ionization potentials (IP) for AIz (X311,). Bauschlicher et at CASSCF/SOCI/ [20~13p6d4f/6~5p3d2fl obtained similar results: re = 3.212 A, we = 169 cm-I, and De = 32.7 kcal/mol for A12+(2Zg+)and IPvm = 6.14 eV and IP,d = 5.90 eV for AI2(X3II,). These theoretical values may be compared with the available experimental data for AI2+: we = 178.8 cm-1,62and Do = 20.8 f 6.9 kcaI/moP and IPd(AI2) = 5.989 f 0.002 eV.@ There is good agreement between theoretical and experimental data (Do(A12+)is an exception but the experimental value needs to be refined). Our calculated re = 3.138 8, and we = 176 cm-I at UMP2(fu11)/6-31G* and Do(AI2+,2zg+) = 30.8 kcal/mol as well as IPad(AI2,'ng+) = 5.6 eV (both at PMP4SDTQ/6-31 l+G*//MP2(ful1)/6-31G* ZPE) agree reasonably well with the best ab initio and experimental data. Neutral AI20. Molecular A120 has also been investigated e ~ p e r i m e n t a l l y ~ ~ *and ~ ~ ~t h~ e~ ~ r~e't*i c~ a~ l l y . ~Gas-phase ~' IR measurement^^^,^^ indicate AI20 to be linear Dmhrbut matrix isolation experimentss7 were interpreted in favor of a bent C, geometry. Both of these structures have been considered theor e t i ~ a l l y ; ~geometry ~ - ~ ~ optimization at the HF/DZ and the HF/DZ+P levels led to a linear Dmh('Eg+)structure, whatever bond angle was assumed initially. We have examined three basic forms: linear A1-O-d (D-h), linear A I - A I d (C-,), and angular AI-0-AI (Cb), all in singlet and triplet states. Both linear structures should have closed-shell singlet ground electronic states. Linear AlOAl has a (1uJ2( 1~,)~(1~,)~(2~~)~(2u,)~( 1 ~valence ~ ) ~ configuration. In the corresponding triplet state, the 1T,-MO would be occupied by only one electron, and Jahn-Teller distortion to bent forms is expected. For similar reasons the C,, structure (( 1 ~ ) ~ ( 2 ~ ) ~ 1( ~ 3 )u~) (~4( u ) ~ (valence 2 a ) ~ configuration) also should be a singlet. We find the Dwh('E +) structure to be the AI20 global minimum. While the C,, 42') form is a local minimum, it is 94.1 kcal/mol higher in energy at HF/6-3 1G*. Nevertheless, this structure also is thermodynamically stable with respect to dissociation into A10 + AI and may be detectable in matrix isolation experiments. Starting with a 120' AlOAl angle, HF/6-31G* geometry o p timization in c2,symmetry led to the h e a r , Dmhstructure. But a 60' starting angle resulted in a local minimum with an equilibrium angle of 78.2'. This ('Al) singlet state has the (2a1)2(1b2)2( 1b1)z(3a1)2(2b1)2(2b2)o valence configuration. Similarly, a 91.5' angle was found for the (3B2)triplet state with the ( laJ2(2aJ2( 1b#( lb1)z(3a1)2(4a1)1(2b2)1 configuration. The C, (3B2) structure lies 66.8 kcal/mol above the ground state; the relative energy of C , (]Al) is even higher, 21 1.0 kcal/mol a t HF/6-31G*. At this uncorrelated level, the stability of the four isomers is: Dm,+ ('Bg+) < C,,('Z+) < CZJ3Bz) < C,(IA,) (see Figure 2a). Our calculated frequencies at HF/6-31G* for the h e a r Dmh form (see Table IV) correspond well with the prior theoretical values (531, 129, and 1012 cm-l@ and 527, 102, and 1057 cm-'39); our calculated A1-0 bond length also agrees.3w1 The frequencies calculated analytically at MP2(fu11)/6-31G* 513, 90, and 988 cm-I are even closer to the available experimental data 472, 100, and 992 cm-'.28*37A correction increases the AI-0 bond length 0.03 A on reoptimization (Figure I ) . Our calculated dissociation energy (MP4SDTQ/6-31 l+G*// MPZ(full)/6-31G* + ZPE) of AlzO into Alz 0,209.1 kcal/mol, agrees well with the experimental value, 210.3 f 4.0, but this is not the case for dissociation into A10 AI. Our best estimate is 8.5 kcal/mol larger than experiment. This reflects our difficulty in reproducing the experimental Doo (AIO) value (see above).

+

+

+

(65) Sunil, K. K.; Jordan, K. D. J. Phys. Chem. 1988, 92, 2774. (66) Bauschlicher, C. W., Jr.; Barnes, L. A,; Taylor, P. R. J . Phys. Chem. 1989. 93. 2932. (67) Bucher, A.; Stauffer, J. L.; Clemperer, W.; Wharton, L. J . Chem. Phys. 1963, 39, 2463. 1~

~

Boldyreu and Schleyer

9050 J . Am. Chem. SOC.,Vol. 113, No. 24, 1991 The AI3+Cation. The geometry and electronic structure of the related cation clusters may help to understand the stability of the A130+ and A130 species. The AI3+ion has been detected in mass and has been studied the~retically.~~ Hanley et. aL61 reported dissociation energies of AI3+ (into AI2 AI+) of 30.0 f 8.1 kcal/mol and (into AI2+ + AI) of 25.8 f 8.1 kcal/mol. Basch calculated the equilibrium geometries for several states.69 Three of these were linear and were described as "singlet 'Al (( la1)2(2a1)2(1 b1)2(2b1)2)",'triplets 3B1 ((lal)2(2al)2(lbl)2(2b,)1(3al)1)",and 'triplet 3A2(( la1)2(2a1)2( 1b1)'(2b1)I( 1b2)I)". In addition, a 'triplet 3B2((la1)2(2a1)2(lb1)2(lb2)1(3al)1)n possessed a strongly bent structure. However, the "3B1n and U3A2" triplets in ref 69 only are representations of the same 3rIgwave function; indeed, the calculated energies of these "states" are reported to be identical. The linear Dmh('E,') ((lu6)2(luu)2( 2 ~ ~ ) ~ ( 2 u , ) ~ valence ( configuration) form was found to be the most stable with a calculated dissociation energy, AI,+ ('E6+) AI2+ ('E,+) AI('P), of 39.2 kcal/mol. We examined AI3+in D3hand Dmhsymmetries (both as singlets and as triplets). Misleadingly, HF/6-31G* optimization gives the linear Dmh(lZ,+)structure (as was found in ref 69 at the CAS MC SCF level), but with the larger 6-31 1+G* basis set and with electron correlation we found the triangular D3h('Al') ((lal')'(le')4(1aF)2(2e')0 valence configuration) to be favored. The linear Dmh ('E +) structure is 31.3 kcal/mol higher in energy at MP4SDfQ/6-31 l+G*//MP2(fuI1)/6-31G*. The other singlet linear structures with (1 uJ2( 1a,)'( l ~ , ) ~ ( 2 a , and ) ~ ( I u )'(1 U , ) ~ ( ~ U 1, x,)2(2u,)0 )~( electronic configurations are much higker in energy. Similarly, the linear triplets with (1 uJ2( 1 ~ ~ ) ' (2u )2(1~,)2(2u,)0and (lu~)2(luu)2(2u~)2(l~u)1(2uu)1 electronic configurations as well as bent C, triplet (( la1)'(2a1)'( 1bl)'(1 b2)l(3al)I) lie more than 25 kcal/mol above the ground state at HF/6-31G*. The AI,+ cation has a D3h (]A1') global minimum according to our best calculations. The calculated dissociation energies for dissociation of AI3+ ( D j h )into AIz AI+, 29.9 kcal/mol, and into AI2+ AI, 31.0 kcal/mol, agree well with the experimental data, 30.0 f 8.1 and 25.8 f 8.1 kcal/mol, respectively6I Correlation corrections (which amount to 30 kcal/mol at MP4SDTQ/6-3 1G*!) are important for the accurate estimation of the AI3+cation dissociation energy (see Table 1). The A130+ Cation. The A130+ cation was detected in the mass spectrum of the reaction product of aluminum cluster ions, AIk+, with o ~ y g e n . ~Nonetheless, ~.~~ neither the structure nor the electronic state of A130+ is known. The vibrational spectra and the dissociation energy have also not been determined to our best knowledge. We investigated two AI$+ structural possibilities, with D3h (the oxygen atom is in the center of the AI3 triangle) and with C3, symmetry (the oxygen atom is at the vertex of a pyramid). Only the D3* singlet (( la1')'( le')4( la~)2(2al')2(2e')4(3e')o valence configuration) was found to be a minimum. Optimization in C,, symmetry (starting with a 109O AlOAl angle) led to the D3hplanar structure. The D j h structure for AI,O+ is expected because of the highly ionic bonding between 0'- anion and the three AI+ cations. Interestingly, the AI+ cation affinity of AI20 is only 5 kcal/mol more than that of AI2 (MP4SDTQ/6-311+G* ZPE). We have not examined the A130+ triplet, but this should not have a planar, trigonal structure. In Djh symmetry, both HOMO and LUMO have e-symmetry. An (2e3)(3e1)electronic configuration should lead to Jahn-Teller distortion. The Al30+triplet state should be higher than singlet because the HOMCbLUMO gap is quite large (9.3 eV). Neutral A130. While there is mass spectroscopic evidence for AI,O,ZO no detailed information, eg., on the structure, frequencies,

+

-

+

+

+

+

(68) Ray, U.; Jarrold, M. F.;Bower, J. E.;Krauss, J. S . Chem. Phys. Lett. 1989, 159, 221. (69) Basch, H. Chem. Phys. Lett. 1987, 136, 289. (70) Chase, M. W.,Jr.; Davies, C. A.; Downey, J. R.,Jr.; Frurip, D. J.; McDonald, R. A.; Syverund, A. N. JANAF Thermochemical Tables, 3rd ed.; Parts I and 11; J . Phys. Chem. Ref Data 1986.

U

Xi< Figure 3. Graph of the AloO flexible intramolecular rearrangement.

or energy, is available. The highest possible symmetry, D3,,, is not expected for A1,O since in the ( la1')'( le')4( lap)2(2a1')2(2e')4(3e')1 valence configuration only the electron occupies the degenerate 3e HOMO. Jahn-Teller distortion leads to Y-type or T-type geometries, both in C , symmetry. We designate these (C2,,Y)('AI: (~a1)2(2a1)2(~b~)z(~bl~*(3a~)2(4al)2(2b~~2~~~~~1 and (C,,T) ('B2: (la1)'(2a1)'( 1b2)2(lbl)Z(3a1)2(4al)2(2b1)2(2b2)1). Both were optimized, as were pyramidal (C,")('A1: (la1)'(le)4(2a1)2(la2)2(2e)4(3al)1) and planar (D3h) ('E: (lal')'( le")4( la/)2(2a1)2(2e')4(3e')1) forms (see Figure I). Appreciable spin contamination was found for all four structures ((S') = 0.79-0.85) in the initial examination at HF/6-31G*. After spin-projecting the wave function to pure spectroscopic states, the total energies of these structures of AI,O decreased by 0.0045-0.0065 au (at PUHF/6-31G*). At this level, the A130 C,,Y ('Al) form is a minimum and the C,,T (2B2)form is a saddle point on the intramolecular rearrangement (see Figure 3); the C,, ('AI) isomers is a local minimum, and the D3h('E) structure is a saddle point second order (two imaginary frequencies). At HF/6-31G* the barrier on the A130 intramolecular rearrangement is only 1.4 kcal/mol (1.1 kcal/mol with ZPE correction). As the relative energy of the C3, ('Al) isomer was quite high, 25.4 kcal/mol at UHF/6-31G*, this portion of the PES was not examined further. The D3h ('E) structure lies 10.9 kcal/mol above the most stable C,,Y ('A1) form. At the HF/6-31G* level, the stability of the four forms is C,,Y ('AI) > C2,,T ('B2) > D3h('E) > C,, ('Al) (see Figure 2b). structures were The lowest energy C,,Y, C,,T, and D3h reoptimized at MP2(fu11)/6-31G*, and single points were calculated at MP4SDTQ/6-31G* and at MP4SDTQ/6-311+G* (see Table I1 and Figure 1). At MP2(fu11)/6-31G*, the Cz,,T ('B2) structure is the global minimum, but the C,,Y ('A,) saddle point structure is only 1.1 kcal/mol higher in energy. This difference is even smaller (0.5 kcal/mol) at PMP4SDTQ/6-311+G* + ZPE, our highest level. Therefore, the A130 molecule is highly flexible with respect to distortion in the plane. A nonrigid model with large amplitude motions of the AI atoms should be used for describing the structure and spectra of this species. The D3h ('E) structure lies 26.4 kcal/mol (MP2(fu11)/6-3 IC*) above the global minimum; however, this 'E state ( D j hsymmetry) may require multiconfiguration treatment for a more reliable estimation of the relative energy. We have considered only two channels for dissociation of A130 into AI20 A1 and AI, AIO. All other possibilities would

+

+

Hyperaluminum Molecules: AIJO and A1,O

J . Am. Chem. Soc., Vol. 113, No. 24, 1991 9051

require much more energy. (Note that the AI-0 bond dissociation Figure 2c). The C,,I (3B2)triplet is actually the most stable A140 energy, 121.5 kcal/mol, is much larger than that of AI-AI, 35.7 species at uncorrelated levels. Of course, correlation favors singlet kcal/mol.) The data in Table I also show the degree to which over triplet states. Thus, the corresponding singlet, C,,I ('A'), also a minimum, is 10.0 kcal/mol higher in energy than the C,,I A130 is stable toward all possible dissociation pathways; this accords with its experimental observation.20 (3B2)triplet at HF/6-31G* but is 6.3 kcal/mol more stable at MP4SDTQ/6-31G*. At MP4SDTQ/6-31G* the C,,I singlet The calculated adiabatic IPS of A130 (2B2and 2AI) are both is 21.3 kcal/mol higher in energy than the most singlet D4hform. 5.1 eV a t PMP4SDTQ/6-311+G* + ZPE. None of the singlet, triplet, and quintet C,,I structures are The Ab2+ Cation. We consider the A142+cation because of the competitive energetically. The same is true of planar CJI forms ionic character of A140: the 02anion is surrounded by on Alq2+ (which have the A10- fragment coordinated to the vertex of the cage. We optimized singlet Alq2+in tetrahedral (Td) ((lal)2AI3+ triangle): the singlet ('Al) (( la1)2(2al)2(1b2)2(lbl)2(1t2)6(2al)2(2t2)0valence configuration) and in square-planar (D4h) (3a1)2(2b2)2(4a1)2(5a,)2(3b2)2(2bl)o), the triplet (3B2) (( la1,)2(leJ4( la2u)2(2alg)2(1 bIg)O)geometries. The Td form is (2a1I2(1b2)2(1b1)2(3al)2(2b2)2(4al)2(5al)2(3b2)1(6al)1), and the a minimum but the D4hform is a saddle point at HF/6-31G* (see quintet ('A2) (( la1)2(2a1)2(3a1)2(lb2)2(lbl)2(2b~)2(4a~)2~5a~~'Table 11). Because of Coulomb repulsion, A142+is not stable thermodynamically toward dissociation. (3b2)'(6al)1(2bl)'(la2)o) are 17.7, 18.5, and 19.0 kcal/mol less stable (at HF/6-31G*) than the C,,I (3B2)minimum. Neutral A140. The structure of molecular A140 cannot be HF/6-3 1G* optimization of square-pyramidal A140 in C , predicted on the basis of classical valence theory. Many possible symmetry (a 'Al state starting with the (( la1)2(le)4(2al)2geometries need to be considered. Our search strategy employed (3a1)2(2e)4(lb2)2(lbl)2(4al)0valence configuration) led to the a fragment approach. We divided A140 into two sets of closedplanar D4,, ('Alg) structure. Optimization of the alternative 3Bl shell moieties, A142++ 02and A10- + AI3+. Different ways of state also in C4, symmetry (starting with the ( le)4(2al)2combining these fragments were explored. Thus, the oxygen (3a1)2(2e)4(1b2)2(lbl)'(4al)'(3e)0 valence configuration) did result dianion, 02may be placed inside a A142+tetrahedron (the most in square-pyramidal structure, but the relative energy was 106.8 stable A142+structure) or may be coordinated to its face, edge, kcal/mol. or vertex. The lower D2d symmetry for A140 with oxygen in the To summarize, the initial search at HF(or UHF)/6-31G* led center also was considered. Likewise, the oxygen end of the A10to the following energy ordering for the various A140 species: C,,I fragment may be coordinated to the face, edge, and vertex of the (3B2)0.0 < D4,,('Alg) 8.5 < C,,I ('A,) 10.0 = C3,,I (3Al) 11.8 triangular AI3+cation (the most stable AI3+ structure). We also C2,,1 ('A2) 13.5 Cb,II (IAl) 17.7 = CbJ1 (3B2) 18.5 = CJI optimized geometries for singlet and triplet C , structures. These ('A2) 19.0 < D2,j ('AI) 43.2 < Td (3Al) 50.0 = c4, (3A2) 52.1 pyramidal forms may be considered to be an ionic complexes C3JI ('Al) 125.2 (see Figure 2c). comprised of a 02dianion coordinated to a square-planar A142+ The lowest energy forms were selected for refinement and were (D4,,)dication. Overall D4h symmetry is possible when the 02reoptimized at MP2(fu11)/6-3 lG* (Figure 2d). Electron correlies in the center of Alq2+(D4h). All the A140 structures which lation correction is known to favor singlets over triplets, and also survived after optimization are summarized in Figure 1. more compact structures. Indeed, at MP2(fu11)/6-3 lG* the First, consider the possibilities with the 02- inside a A+I: "cage". square-planar (D4,,) ('Alp) form is the global minimum. Both In Td symmetry, the eight oxygen dianion valence electrons of C , singlets, I and 11, are higher in energy (18.7 and 42.3 kcal/mol, 02must interact with the virtual ,4142' (Td) MO's of appropriate respectively), and neither one is a minimum at MP2(fu11)/6-31G*. symmetry. However, only two electrons are available to occupy The C2,,I triplet, which was the lowest energy A140 form at the doubly degenerate l e HOMO in AI40 (Td) ((lal)2(lt2)6HF/6-31G*, lies 21.3 kcal/mol above the D4h singlet a t (2al)2(2t2)6( le)2(3t2)ovalence configuration). Hence, Jahn-Teller PMP4SDTQ/6-31G* (calculated using the UHF/6-31G* geomdistortion is expected in the singlet state. (This leads to the (D4,,) etry). The relative energy of the triplet C3",I structure is 24.4 planar structure on optimization.) The (D4& singlet ('Alg) 1G*//HF/6-3 1G* . ( ( ~ a l g ) 2 ( ~ e , ) 4 ( ~ a ~ u ) 2 ( 2 a ~ ~ ) 2 ( ~ e , ) 4valence ( ~ b 2 , ) 2 ~kcal/mol ~ ~ l ~ ~ 2a ~t PMP4SDTQ/6-3 ~~~~o Our best estimates of the final relative energies (at configuration) structure is a minimum at the H F and IS the global PMP4SDTQ/6-31G*) are (in kcal/mol): D4h (]Alg)0.0 < C,,I minimum at our highest, correlated level (see Table 11). The (DU) ('A,) 15.0 < Cb,I (3B2)21.3 CJ (3A1)24.4. Hence, we predict ( 'Al) (( 1 1bJ2( 1e)4(2a1)2(2b2)2(2e)4(3a1)2( 3b2)0configuration) that A140 favors the unusual square-planar (D4*,'Alg) structure is lower in energy that the Td but has three imaginary frequencies. rather strongly. Our calculated frequencies at MP2(fu11)/6-31G1 The triplet (Td) (3Al) state is a minimum since each of the le (Table 11) may help identification in the gas phase and in the MOs is singly occupied (( laJ2( 1t2)6(2a,)2(2t2)6( le)2(3t2)0valence matrix isolation. configuration). Nevertheless, this state is 41.4 kcal/mol less stable than the (D4h)singlet at HF/6-31G* and is not a viable form. Bonding and Structure of the Aluminum Oxides This energy difference is even larger at correlated levels. Natural population and natural bond orbital a n a l y s e ~of~ ~ . ~ ~ All three structures with 02coordinated to the face (C3,,I), the various aluminum oxide species are summarized in Figure 4. the edge (CJII), and the vertex (C3,11) of tetrahedral A+I: were The systems are categorized into two group. The AI-0 bonds optimized in singlet and triplet states. The (C3,,I) structure is in the first group (AIO, A120 (D,,,,), and A130+) are essentially expected to be a triplet, because two electrons are available to fully ionic. Note the large values of the natural charges. Furoccupy the doubly degenerate e HOMO. Indeed, two imaginary thermore, the nonorthogonal Wiberg overlap populations (a frequencies (Table 11) are calculated for the singlet C3, form. The measure of the covalent contributions) between the central atom C J , , ~triplet (3A1 (( 1all2(1el4(2a1)2(3a1)2(2e)4(4al I2(3el2(4e)O and the ligands and especially between the ligands are very small. valence configuration) is a minimum only 3.2 kcal/mol higher Therefore, the chemical bonding in this group is not unusual; the in energy (HF/6-31G*) than the D4,,(IAlg) form. Upon optichief attractive interactions, largely electrostatic, are between the mization of the states, the singlet led to the Dlh minimum central oxygen atom and the aluminum ligands. and the triplet to the square-pyramidal (C,) (3B1)geometry. The According to classical theories of valence, neither A130 nor A140 C3,,11 singlet ('A,) (( lal)2(2a1)2(3a1)2( le)4(2e)4(4al)2(5al)2(3e)0 should be stable because the maximum "combining capacity" of valence configuration) lies 129.6 kcal/mol above the 0411 form oxygen is two. Even so, the nature of the A130+cation is not hard at HF/6-31G*. to understand. Aluminum is a very electropositive element (the The third set of AI& structure involves A10-, AI3+ interaction. electronegativity is 1.47 on the Allred-Rochow scale). Like H30+ At HF, the singlet 'Al ((la1)2(2al)2(lb2)2(lbl)2(3a,)2(4a1)2and especially Li30+, A130+ is a highly ionic complex in which (2b2)2(5a1)2(6al)2(2bl)o), triplet 'B2 ( ( l a 1 ) ~ ( 2 a , ) ~ ( 3lb2)2a~)~( three monovalent AI+ cations interact with an 02-dianion. The (1b1)2(2b2)2(4a1)2(2b1)2(5al)'(3b2)1). and quintet 5A2 ((lal)2(2aJ2( 1 b2I2(1 b1)2(3a1)2(4al)2(2b2)2(~al~'(6al~'(2bl~~~3b2)~(4b2)o) states of the &,I form (in which the AIO- fragment is coordinated (71) Reed, A. E.; Weinhold, F. QCPE Bull. 1985, 5, 141. to the edge of the triangular AI3+ cation) are low in energy (see (72) Reed, A. E.; Weinhold, F. Chem. Reu. 1988, 88, 899.

9052 J . Am. Chem. SOC.,Vol. 113, No. 24, 1991

Boldyrev and Schleyer

n s0.87

2a1‘=-0.2047

LUMO

1a2”=-0.3943 HOMO

9-

1e1--0.53O8

1a1’=-0.7483

Figure 5. MO scheme of AI3+ (D3h).

+0.19

N

0.0‘

‘1

\o . a s

N

/ I

-1.47 + 0 . 8 7 Al-0-Al 0 . : 1

0.44

tO.11

Figure 4. Natural charges and nonorthogonal natural atomic overlap populations of aluminum oxides (HF/6-31G*).

bonding in the neutral AI20 molecule is similar: AI+O*-AI+. The preference of AlzO for linear, and Al,O+ for a planar, trigonal geometry is due to the predominant electrostatic repulsion of the AI+ ligands. Even though each AI+ cation has two valence electrons at its disposal, the central oxygen atom octet already is ‘saturated” and AI-AI bonding cannot compete effectively. Some AI-AI bonding is present in the bent forms of A120, but these are much higher in energy. The bonding situation in AI,O and A140 is significantly different. As the oxygen retains its 2- character (a higher negative charge is unrealistic), the aluminum charges are reduced from the ca. 1 + values in A1,O and A130+. Hence, the electrostatic ligand-ligand repulsion is less important, and AI-AI bonding can contribute more effectively. The stability of both A130 and A140 is due to the attractive interactions among the ligands. This also is the case, but to a lesser extent, in the hypermetalated molecules

involving lithium, sodium, etc.’“ As shown in Figure 4, all stable AI,O and A140 structures benefit from relatively large overlap (nonorthogonal) populations between aluminum atoms. The reference data provided for the polyaluminum species, AI2, AI2+, AI;+, AI3+ (D,,,), and ( T d ) , provide calibration. Note that the C,,,Y structure of A130can be interpreted as a triple ion: a monovalent AI+ cation and a AI2+radical cation interacting with 02-. The possible alternative C, arrangement, in which 02- would interact with a triangular A132+cation, would not reduce the electrostatic repulsion as effectively. The same interpretation rationalizes the stability of the D4,, A140 versus the other possible forms. Examination of the molecular orbitals also helps clarify why some of these molecules have bonding ligand-ligand interactions while others do not. Figures 5 and 6 provide qualitative M O schemes for the AI3+and Al30+cations. The MOs of A130+ can MO’s and 02-AO’s: la,’-MO be derived from the AI3+ (D,,,) ( la,’-MO(A13+) + 2s(02-)), le’-MO (le’(Alf) + 2p,,2p (02-)), la,”-MO (la,”(A13+) 2p,(02-)), 2a,’-MO (lal’(A134) - 2s(02-)), 2e’-HOMO (le’(A13+) - 2p,,2p,,(02-)), 3e’-LUMO (2e’(AI3+) + 2p,(02-) and 2e’(A13+) - 2p,,(02-)), 2aF-MO (laT(A13+)- 2p,(02-)). Both le” and la; are pure ligand MOs. The four lowest AI,+ valence M O s (1a,’, 1e’ and 1a;) are O-AI bonding. The next two MO’s (2al’ and 2e’) are antibonding in this respect, but have AI-AI bonding character. On this basis, structure, because the 2e‘Al30+ can be said to prefer the D3,, HOMO is fully occupied by four electrons and the HOMOLUMO energy gap is very large. If the unpaired electron in the neutral A130 molecule occupies the 3e’-MO, C, distortion to Yor T-type structures occurs (see Figure 1). Relative to the D3* geometry, these distortions result in increased AI-AI bonding. If

+

J . Am. Chem. SOC.,Vol. 113, No. 24, 1991 9053

Hyperaluminum Molecules: A130 and AI+O

A142+ (D4h)

n le”=-O. 1078

36‘-

-0.1427 L U ~ O

20‘--0.4842

HOMO

2a1=-0.5433

1eU=-0.7734

9

dv

d

1.1‘--1.4943

Figure 6. MO scheme of A130+ (D3,,).

the unpaired electron occupies the 2a,”-MO, a C3, structure is found, but this minimum is higher in energy than the C2, forms because of the less effective overlap. The qualitative MO scheme for A130+ also provides insights in comparison with Li30+. The first eight valence electrons of both fill the lal’-, le’-, and la,”-MO’s which are central atomligand bonding. The next MO’s (2al’, 2e’, 3e’, 2a,”, and le”) have antibonding or nonbonding central atom-ligand character, but are ligand-ligand bonding. Hence, molecules in which such MOs are occupied may be stabilized by ligand-ligand interactions. Neutral LiJO and Al,O are examples. In Li30, the ninth valence electron occupies the 2a,’-MO, but in A1,O five “extra” valence electrons are available for the 2al’-, 2e’-, and 3e’-MO’s. A further increase in the number of valence electrons would lead to greater antibonding central atom-ligand character. After a certain limit in the number of electrons is reached, structures with the oxygen atom outside the M, or M4 moieties should be more stable than structures (like Li30, A130, Li40, and A140) with oxygen inside. (For example, P 4 0 might be expected to have oxygen outside the P4 cluster.) An investigation along these lines is in progress. Similar qualitative considerations apply to other cases with heteroatoms other than oxygen, and these are being studied as well.

lalg=-o.9762

Figure 7. MO scheme of AI:+

(D4*).

The qualitative M O schemes for D4,,symmetry, involving Aid2+ and A140, are given on Figures 7 and 8. Again, the MO’s of A140 may be derived from the MO’s of A142+and the AO’s of 02-:la, -MO (lalg(A142+)+ 2s(02-)), le,-MO (le,(A142+) + 2p,,2p,(b2-)), la2,-Mo ( la2u(A142+)+ 2p,(02-)), 2al -MO - le,(AI>+) + (lalg(A142+)- 2s(02-)), 2e,-MO (2pX,2p(02-) 2e,(A142+)), Ib -MO (pure Ib,g(A142+)i, Ib2 -HOMO (pure 1b2,(A142+)),a d 2a2,-LUMO ( la2,(A142+)-2p,(02-)). The first valence lalg-,le,,-, and la2,-MOs are central atom-ligand bonding and either bonding or nonbonding with respect to ligand-ligand interactions. In Li40, only these five valence MO’s are occupied; hence, this molecule is stable toward dissociation. In Li40 and in A140 these MO’s are comprised primarily of the 2s- and 2pAO’s of 02-.This dianion is stabilized by the field of Li42+or dication cage. In A140 (D4h)the next higher valence 2aI,-, 2eu-, 1b,,-, and 1 b2,-MO’s are fully occupied. All these MO’s are mainly ligand AO’s; only the 1bl,-MO is ligand-ligand antibonding. Hence, bonding AI-AI interactions contribute to the stability of A140. Because the LUMO of A140 as well as higher unoccupied M O s (up to 3a1,) also have predominate ligand-ligand bonding character, similar to M 4 0 molecules with electropositive ligands M and a greater number of valence electrons may also be stable toward dissociation. But, as discussed above, increasing the number of valence electrons will lead eventually to an increase in the antibonding central atom-ligand interactions. Alternative structures with the oxygen dianion outside of the M4 moiety will be favored increasingly. An example already is provided by the C3J triplet form of A140, in which the 02-dianion lies outside

9054 J . Am. Chem. SOC.,Vol. 113, No. 24, 1991 A140 (D4h)

2a2U=+0.0115 LUMO

Boldyrev and Schleyer HF/6-3 IC*. But correlation energy favors the singlet planar square D4hform preferentially. This also shows the importance of covalent, multicenter bonding in A140. Wholly ionic species have localized character; their relative energies typically are little influenced by correlation corrections. Finally, it is interesting to consider the stability relationship of the aluminum oxides along the A10-Al20-A130-Al~0 series (Table I). The first two members have very high dissociation energies (e&, well over 100 kcal/mol for loss of an A1 atom) because of the highly ionic bonding A P 0 - and A1+O2-AI+. As the maximum oxygen charge is 2- (note Figure 3; oxygen has nearly the same effective charge in A120, A130, and A1,0), the total electrostatic A 1 4 bonding does not improve when more A1 ligands are present. However, the ligand-ligand bonding does increase. Indeed, the energies of dissociation of A130 into A I 0 AI (19.9 kcal/mol) and A140 into Al,O AI (45.5 kcal/mol) or into A120 + AI2 (37.1 kcal/mol; all at PMP4SDTQ/6-31 I+G* ZPE) are similar to Do(A12)= 35.7 f 1.0 kcal/mol" (see Table I). (For AI,O, the dissociation into A10 and AI2 is less favorable for the combination of reasons discussed above.)

+ +

2eu=-0.3355

2alg=-0.4296

"x

1eu=-o.6765

lalg=-1.3379

Figure 8. MO scheme of A1,O (D4*).

the AId2+cage. This structure is only a few kcal/mol higher in energy than the singlet D4h global minimum at HF/6-3 1G*, but correlation increases this difference substantially. Other species which employ ligand-ligand interactions effectively may be also stable. The AI-A1 distances in C,,I and CJI structures of A140 ('A,) are quite short. These singlets lie above the D4h form by 1.4 and 9.1 kcal/mol at HF/6-31G*, respectively. Likewise, the triplet &,I is the most stable structure of A140 at

+

Conclusions The main conclusions of this study are the following. (1) The hyperaluminum A130 and A140 molecules are stable with respect to all possible modes of dissociation. A130 prefers planar triangular Y- and T-type geometries (these have nearly the same energy and this molecule is highly flexible with respect to in-planar intramolecular rearrangement). A140 adopts a square-planar singlet structure. The calculated frequencies for these species may help their identification in matrix isolation experiments. Several additional Al,O and A140 minima in both singlet and triplet states were identified, but these are significantly higher in energy at electron correlated levels. (2) Although we predict AI,O to have a substantially lower dissociation energy than A140 (e.g., 19.9 versus 45.5 kcal/mol, respectively, for loss of an AI atom), there only is mass spectrometric evidence for the former as yet.20 A deliberate search for A140 experimentally should be successful. (3) The high degree of ionic character as well as bonding interactions between the aluminum atoms is responsible for the stability of these hypermetalated species. AI-AI interactions are largely responsible for the greater stability of A140 over A130. (4) These findings should be general. Other hypervalent M,X molecules, where n is higher than the normal valence of the central atom X and M is an electropositive ligand (other than alkali or aluminum), are likely to be stable because of the combination of electrostatic and bonding ligand-ligand interactions. Acknowledgment. This work was facilitated by an Alexander-von-Humboldt Fellowship to A.I.B. and was supported by the Fonds Der Chemischen Industrie, the Stiftung Volkwagenwerk, the Deutsche Forschungsgemeinschaft, and the Convex Computer Corporation. Registry No. A130+, 136060-69-0;AI,O, 136060-68-9;AI4*+, 136060-67-8; A140, 136060-66-7.

Supplementary Material Available: Tables of total energies of different forms of aluminum oxide species at UHF/6-31G* (Table IS) and at different correlated levels (Table 2s) (4 pages). Ordering information is given on any current masthead page.