Theoretical Studies of Organometallic Compounds ... - ACS Publications

Feb 22, 1995 - groups.17bc·18-23 There are no ab initio studies of the ..... (2.092 Á),41 while a previously reported MNDO value is somewhat too short...
0 downloads 0 Views 802KB Size
Organometallics 1995, 14, 4263-4268

4263

Structure and Bonding of the Transition Metal Methyl and Phenyl Compounds MCHs and M C a s (M = Cu, Ag, Au) and M(CH& and M(C&)2 (M = Zn, Cd, Hg)' Iris Antes" and Gernot Frenking" Fachbereich Chemie, Philipps- Universitat Marburg, Hans-Meerwein-Strasse, 0-35032 Marburg, Germany Received February 22, 1995@ Quantum mechanical calculations of the geometries and metal-carbon bond dissociation energies using relativistic pseudopotentials with large valence basis sets for the metals are reported for MCH3 and MCsH5 (M = Cu, Ag, Au) and for M(CH3)2 and M(C&)z (M = Zn, Cd, Hg). The Cu-CH3 bond length calculated at the MP2 level is significantly shorter (1.866 than predicted in most previous studies. This is due to relativistic effects, which are important for accurate calculations of the geometries of copper compounds. The calculated is much longer than the experimental value of the Ag-C(pheny1) bond length (2.091 alleged silver aryl complex (1.902 reported by Lingnau and Strahle (Angew. Chem., Int. Ed. Engl. 1988,27, 436). The other theoretical bond lengths are in excellent agreement with experimental values. It seems unlikely that the measured compound is a silver aryl complex. The calculated metal-carbon dissociation energies at CCSD(T) are slightly lower than the experimental values. The calculations predict that the M-C bond strengths of the group 11 methyl and phenyl compounds have the order Au > Cu > Ag, while the group 12 elements have the order Zn > Cd > Hg. The NBO method and the topological analysis of the electron density distribution show that the metal-carbon bonds are strongly polarized toward the carbon ends. The M-C polarization decreases from the first to the second and third transition metal rows. The NBO analysis gives only one M-C bond for the M(CH3)z and M(CsH5)2compounds. The M-C bonds of the latter compounds are clearly more ionic than the group 11 methyl and phenyl compounds.

A)

A) A)

1. Introduction

The ubiquitous utility of organometallic reagants for synthetic purposes stands in striking contrast to the poor knowledge about the structures and particularly the bond energies of transition metal compounds.2 One example is the ongoing controversy about the "higherorder" and "lower-order" cuprates. Although cuprates are among the most versatile organometallic molecules in organic synthesis, the structure of the active species is still a topic of controversial discussion^.^ Another example concerns the structure of aryl-copper and aryl-silver compounds. In 1988 Lingnau and Strahle (LS) reported the first synthesis of monocoordinated Cu and Ag aryl compounds M(Ar) (M = Cu, Ag; Ar = 2,4,6P~~CGH A~ surprising ).~ feature of the X-ray structure analysis of the complexes was that the Cu-C and Ag-C + Present address: Institut fur Organische Chemie, Universitat Zurich, Zurich, Switzerland. ( 1 ) Theoretical Studies of Organometallic Compounds. XIV. Part XIII: Frenkine. G.: DaDDrich. S.: Ehlers. A. W.: Otto . M.: Vvboishchikov, S. F. ProceediGgs of 'the NATO 'Advanced Study 'Inktitute, Metal-Ligand Interactions: Structure and Reactivitv. Cetraro. Italv. SeptembG 5-16, 1994; Russo, N., Ed.; Kluwer Academic Pub1isher"s; Amsterdam, in print. @Abstractpublished in Advance ACS Abstracts, July 15, 1995. ( 2 )Bonding Energetics in Organometallic Compounds; Marks, T. J., Ed.; ACS Symposium Series 428; American Chemical Society: Washington, DC, 1990. ( 3 )(a) Lipshutz, B. H.; James, B. J . Org. Chem. 1994,59, 7585. (b) Bertz, S. H. J. Am. Chem. Soc. 1990,112,4031; 1991,113, 5470. (c) Stemmler, T.;Penner-Hahn,J. E.;Knochel, P. J.Am. Chem. Soc. 1993, 115, 348. (d) Stemmler, T. L.; Barnhart, T.M.; Penner-Hahn, J. E.; Tucker. C. E.: Knochel,. P.: . BBhme.. M.:. Frenkina. - G. Submitted for publication.

A;

bond lengths were nearly the same (Cu-C = 1.890 Ag-C = 1.902 The identification of the alleged metal complexes has recently been challenged by Haaland et al.5 These authors reexamined the results of LS,4 and they suggested that the observed molecules may be the bromine derivatives BrAr rather than M(Ar). They estimated that the Ag-C(ary1) bond length should be approximately 2.08 A rather than 1.902 A.5 The estimate was based upon experimentally known metalhydride and metal-chloride distances, which are 0.160.23 A longer for silver than for copper.6 Ab initio calculation of CuCH3 and AgCH3 predicted also that the Ag-CH3 bond is 0.27 A longer than the Cu-CH3 bond.5 This makes it highly unlikely that the Ag-C(ary1) distance is nearly as long as the Cu-C(ary1) bond. The use of ab initio methods to obtain accurate data for the structures and properties of molecules has become a routine in the chemistry of light-atom molecules. However, there is still the belief that heavyatom molecules, particularly transition metal complexes, cannot bs calculated with the same accuracy by quantum chemical methods. This assumption is not true. Transition metal complexes can be calculated very reliably using nonlocal gradient corrected density func-

A).

( 4 )Lingnau, R.; Strahle, J. Angew. Chem. 1988, 100, 409; Angew. Chem., Int. Ed. Engl. 1988,27, 436. ( 5 )Haaland, A.; Rypdal, K.; Verne, H. P.; Scherer, W.; Thiel, W. R. Angew. Chem. 1994, 106, 2515;Angew. Chem., Int. Ed. Engl. 1994, 33,-2443. (6) Huber, K. P.; Herzberg, G. Molecular Spectra and Molecular Structure. Vol. IV: Constants of Diatomic Molecules; Van Nostrand: New York, 1979.

0 1995 American Chemical Society

4264 Organometallics, Vol. 14, No. 9, 1995

tional theory (DFT)' or classical ab initio methods in combination with pseudopotentials.s It has been demonstrated in several studies of different transition metal complexes that standard ab initio methods using pseudopotentials give very accurate geometries and bond energies of heavy-atom molecule^.^^^ We reported recently about theoretically predicted structures of cuprates.1° In this paper we present calculated equilibrium geometries and metal-carbon bond dissociation energies of the methyl and phenyl compounds of the elements of group 11 MCH3 and MC&, (M = Cu, Ag, Au). We also calculated the compounds M(CH3)z and M(C6Hz)z of group 12 elements (M = Zn, Cd, Hg). The theoretically predicted bond lengths are given a t the MP2 level of theory. Bond dissociation energies are calculated at the MF'2 level and using coupled-cluster theoryll at the CCSD(T)12level. The bonding situation of the molecules has been examined using the natural bond orbital (NBO) partitioning scheme13 and the topological analysis of the electron density di~tributi0n.l~ The calculations reported here have been carried out with different pseudopotentials than in our previous studies.BJO The pseudopotentials developed by the group of Stoll and Preuss15have a much higher number of valence basis functions than the pseudopotential reported by Hay and Wadt.lG Thus, the valence basis set for the metals in this study has TZfP quality. Also, the pseudopotentials of group 12 elements Zn, Cd, and Hg suggested by Hay and WadtlGbhave a large core, i.e., the ( n - l)s2(n - Up6 outermost core electrons are included in the core potentials. The StoWPreuss pseudopotentials for Zn, Cd, and Hg have a 20-electron valence space.15 Another advantage of the latter pseudopotentials is that relativistic effects are included for Cu and Zn, while the Hay/Wadt pseudopotentials of the fourth row are nonre1ativistic.l6 It has been shown that relativistic effects cannot be neglected for accurate calculations of the bond lengths and bond energies of Cu (7) Li, J.; Schreckenbach, G.; Ziegler, T. J . A m . Chem. SOC. 1995, 117, 486. (8) Review: Frenking, G.; Antes, I.; Bohme, M.; Dapprich, S.; Ehlers, A. W.; Jonas, V.; Neuhaus, A,; Otto, M.; Stegmann, R.; Veldkamp, A.; Vyboishchikov, S. F. In Reviews in Computational Chemistry; Vol. 7, Lipkowitz, K. B., Boyd, D. B. Eds.; VCH: New York, in press. (9)(a) Ehlers, A. W.; Frenking, G. J. Am. Chem. SOC.1994, 116, 1514. (b) Veldkamp, A.; Frenking, G. J . Am. Chem. SOC. 116, 4937 (1994).(c) Neuhaus, A,; Veldkamp, A,; Frenking, G. Inorg. Chem. 1994, 33,5278. (d) Dapprich, S.; Frenking, G. Angew. Chem. 1995,107,383; Angew. Chem., Int. Ed. Engl. 1995,34, 354. (10) Bohme, M.; Frenking, G.; Reetz, M. T. organometallics 1994, 13, 4237. (11)Cizek, J. J. Chem. Phys. 1966, 45, 4256. (12)(a) Pople, J. A.; Krishnan, R.; Schlegel, H. B.; Binkley, J. S. Int. J . Quantum Chem. 1978,14, 545. (b) Bartlett, R. J.; Purvis, G. D. Ibid. 1978, 14, 561. ( c ) Purvis , G. D.; Bartlett, R. J. J . Chem. Phys. 1982,76,1910.(d) Raghavachari, K.; Trucks, G. W.; Pople, J. A,; HeadGordon, M. Chem. Phys. Lett. 1989,157,479.(e) Bartlett, R. J . Watts, J . D.; Kucharski, S. A.; Noga, J. Ibid. 1990, 165, 513. (13) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899. (14) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Oxford University Press: Oxford, 1990. (15) (a) Dolg, M.; Wedig, U.; Stoll, H.; Preuss, H. J . Chem. Phys. 1987, 86, 866. (b) Andrae, D.; Haussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1990, 77, 123. (16)(a)Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299. (b) Hay, P. J.;Wadt, W. R. J. Chem. Phys. 1985,82, 270. (17)(a) Barnes, L. A.; Rosi, M.; Bauschlicher, C. W. J. Chem. Phys. 1990, 93, 609. (b) Bauschlicher, C. W.; Langhoff, S. R.; Partridge, H.; Barnes, L. A. J. Chem. Phys. 1989, 91, 2399. ( c ) Sodupe, M.; Bauschlicher, C. W.; Lee, T. J. Chem. Phys. Lett. 1992, 189, 266.

Antes and Frenking

The methyl compounds M(CH3), of group 11 and 1 2 elements have been investigated previously by several g r ~ ~ p s . There ~ ~ ~are ~ no J ~ab- initio ~ ~ studies of the phenyl compounds M(CsH5), known to us. The geometry of Hg(Cs&)z has been optimized using MND0.24 2. Computational Methods The geometries of the molecules have been optimized at the Hartree-Fock (HF) and MP2 (M~ller-Plesset perturbation theory terminated at second orderIz5levels of theory. For the geometry optimizations an all-electron 6-31G(d) basis set was used for carbon and hydrogen.26 Relativistic pseudopotentials were employed for the metals.15 The ( n - l)s2(n- l)p6(n)s2(n - l)dx electrons of the metals are calculated explicitly. This means that the group 11 elements Cu, Ag, and Ag have 19 valence electrons and that the group 12 elements Zn, Cd, and Hg have 20 valence electrons. A large basis set [311111/22111/ 4111 has been used for the valence e1e~trons.l~ This basis set combination is denoted as basis set I. Improved total energies were calculated for the methyl compounds using coupled-cluster theory1' with single and double excitations and a noniterative estimate of the triple excitations CCSD(T).lZFor the latter calculations a 6-31+G(d) basis set is used for carbon and hydrogen.27 The valence basis set I of the metals is augmented by a set of f-type polarization functions.28 This basis set combination is denoted as basis set 11. The total energies of the phenyl compounds of Cu, Ag, and Au were also calculated at the CCSD(T) level using basis sets augmented by a set of f-type functions at the metals, but only 6-31G(d)at the other atoms. The total energies of the diphenyl compounds of Zn, Cd, and Hg are given at MPBA. The vibrational frequencies and zero-point energies (ZPE) were calculated at HFA. The ZPE corrections are scaled by a factor of 0.89.29 All structures reported here are minima on the potential energy surface, Le., the eigenvalues of the Hessian matrix are all positive. The calculations have been carried out using the program packages Gaussian 92,30T ~ r b o m o l eand , ~ ~ACES For the (18) (a) Schwerdtfeger, P.; Boyd, P. D. W.; Burrell, A. K.; Robinson, W. T.; Taylor, M. T. Inorg. Chem. 1990,29, 3593. (b) Schwerdtfeger, P. J . Am. Chem. SOC.1990,112, 2818. ( c ) Schwerdtfeger, P.; Boyd, P. D. W.; Brienne, S.; McFeathers, J . S.; Dolg, M.; Liao, M.-S.; Schwarz, W. H. E. Inore. Chim. Acta 1993.213. - , - - , 233. ~~(19) Kippel;, C.; Thiel, W.; McKean, D. C.; Coats, A. M. Spectmhim. Acta 1992,48A, 1067. (20) Reinhold, J.; Steinfeld, N.; Schiiler, M.: Steinborn. D. J . Organomet. Chem. 1992,425, 1. (21) Ziegler, T.; Tschinke, V.; Becke, A. J . Am. Chem. SOC.1987, 109, 1351. (22)(a) Chen, H.; Krasowski, M.; Fitzgerald, G. J. Chem. SOC.1993, 98, 8710. (b) Sosa, C.; Andzelm, J.; Elkin, B. C.; Wimmer, E.; Dobbs, K. D.; Dixon, D. A. J. Phys. Chem. 1992,96,6630. (c) Kaupp, M.; Stoll, H.; Preuss, H. J . Comput. Chem. 1990, 11, 1029. (d) Jayatilaka, D.; Amos, R. D.; Koga, N. Chem. Phys. Lett. 1989,163, 151. (23)Almenningen, A.; Helgaker, T. U.; Haaland, A.; Samdal, S.Acta Chem. Scand. 1982, A36, 159. (24) Rhodes, C. J.;Glidewell, C.; Agirbas, H. J . Chem. Soc., Faraday Trans. 1991, 87, 3171. (25)(a) Maller, C.; Plesset, M. S. Phys. Reu. 1934, 46, 618. (b) Binkley, J. S.; Pople, J. A. Int. J . Quantum Chem. 1975, 9, 229. (26) (a) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1971, 56, 2257. (b) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213. (27) Clark, T.; Chandrasekhar, J.; Spitznagel, G . W.; Schleyer, P. v. R. J . Comput. Chem. 1983, 4, 294. (28) The exponents for thefhnctions are: fcU= 1.3375;f4 = 1.3375; f A u = 1.1447;fZn = 1.500; fCd = 1.500; f H g = 1.500. (29) Hout, R. F.; Levi, B. A,; Hehre, W. J . J. Comput. Chem. 1982, 3, 234. (30) Gaussian 92, Revision C ; Frisch, M. J.; Trucks, G. W.; HeadGordon, M.; Gill, P. M. W.; Wong, M. W.; Foresman, J. B.; Johnson, B. G.; Schlegel, H. B.; Robb, M. A.; Replogle, E. S.;Gomperts, R.; Andres, J . L.; Raghavachari, K.; Binkley, J. S.; Gonzalez, C.; Martin, R. L.; Fox, D. J.; Defrees, D. J.; Baker, J.; Stewart, J. J . P.; Pople, J . A. Gaussian, Inc.: Pittsburgh, PA, 1992.

Organometallics, Vol. 14, No. 9, 1995 4265

Transition Metal Methyl and Phenyl Compounds

Table 1. Calculated and Experimental Metal-C Bond Lengths (A) of Group 11 and 12 Methyl and Phenyl Compounds our

sym HF/I

MP2A

other

exptl

CUCH~

C3"

1.970

1.866

Cu(CH3)zAgCH3 AuCH3 CUCsH5 AgC& AuCsH5 Zn(CH3)z Cd(CH3)z Hg(CH3)z Zn(C,&)z Cd(CsH& Hg(CsH5)z

D3h C3" CsU Czu Czu Czu Dah D3h D3h DM DM Dw

2.019 2.180 2.077 1.960 2.165 2.049 1.979 2.162 2.135 1.963 2.143 2.112

1.936;' 1.921;g 1.923;J 1.86h 1.917;i 1.927;' 1.963' 2.146;C2.06h 2.190d

1.922 2.111 2.017 1.850 2.091 1.981 1.925 1.912;' 1.910;e 1.94ok 2.123 2.119;' 2.123e 2.104 2.08Or 1.914 2.110 2.085

1.935'

molecule

1.928' 2.110b 2.093b 2.092'

'

Reference 34. Reference 35. ' MCPF level, ref 17b. H F level, ref 18a. e DFT level, ref 22a. f M P 2 level, ref 18c. 8 CCSD(T) level, ref 17c. DFT level, ref 21. * DFT level, ref 22b. J MP2 level, ref 10. CISD level, ref 22c. Reference 41. a

Table 2. Calculated and Experimental Metal-C Bond Dissociation Energies De and Doi (kcdmol) of Group 11 and 12 Methyl and Phenyl Compounds. Geometries are Optimized at MP2/I (see Table 1) molecule sym CuCH3 AgCH3 AuCH~ CuC&" AgCsHs" AuC&,~ Zn(CH3)zd Cd(CH3)zd Hg(CH3)zd Zn(C&)zd Cd(C&dzd Hg(C&)zd

HF/IIO MP2flIa De De D,

14.6 58.7 55.5 7.2 43.1 39.9 Csu 18.0 62.5 58.6 cpu 24.9 74.0 72.2 Czu 17.0 58.3 56.7 Czu 26.3 80.2 78.2 D3h 47.4 90.4 83.1 D3h 33.6 73.8 66.7 D3h 23.9 66.6 58.7 Du 70.6 123.2 118.9 D u 55.3 109.2 105.2 Dzd 45.1 103.3 99.0 C3" C3b

CCSD(T)/IIa De Do 52.1 40.8 58.1 65.1 53.8 72.4 81.1 67.8 60.5

48.9 37.6 54.0 63.3 52.2 70.4 73.8 60.7 52.6

other D,

exptl Do

48.4;e56.9h 53.3' 36.2;e42.3h 11.9

58.78

84.8' 67.2' 57.9 75.3'

The 6-31G(d) basis set is used for the phenyl compounds of Cu, Ag, and Au. For the diphenyl compounds of Zn, Cd, and Hg basis set I is employed. Reference 39. Taking the differences of the heats of formation, reference 38. Dissociation energy with respect to both methyl and phenyl groups, respectively. e MCPF level, ref 17b. f H F level, ref 18a. 8 MP2 level, ref 18c. DFT level, ref 21. * The Do values include the ZPE corrections.

value (1.923 A) given in our earlier study.1° Because the basis sets in these s t u d i e ~ 'were ~ ~ ? rather ~ large, it can be ruled out that the differences are due t o basis set effects. However, the latter three values have been obtained without the consideration of relativistic effects, while the present value was calculated with a relativistic pseud~potential.~~ Previous calculations gave Cu+CO bond lengths of CuCO+ r = 1.985 A (nonrelativistic) and 1.941 A (relativistic).17" It has also been shown that relativistic effects increase the Cu-CH3 bond energy by 3 k~al/mol.'~ The ~ present study suggests that the CuCH3 interatomic distance is significantly influenced by relativistic contributions. It should be pointed out that a previous DFT calculation21 gave a Cu-CH3 bond length of 1.86 which agrees with our value (1.866 A). The short DFT value for the Cu-CH3 distance is due to a fortuitous error cancellation. The DFT calculation was carried out using a local nonrelativistic functional.21 Nonlocal corrections would make the bond longer, and relativity would contract it. The conclusion about the importance of relativistic contributions is supported by the calculated Cu-C bond distance of dimethyl cuprate, Cu(CH3)2-. The experimental value for the Cu-C bond length is 1.935 A.34 Our previously reported value given by a nonrelativistic MP2 calculation is 1.963 A.lo The relativistic MP2A value given here is 1.922 A, which is in good agreement with experiment. The DFT results for Cu(CH& are very similar to our data (Table 1). The calculated MP2A values for the M-C bond lengths of AgCH3 (2.111 8) and AuCH3 (2.017 A) are shorter than calculated before at the MCPF level for AgCH3 (2.146A)17band at the HF level for AuCH3 (2.190 A).18aThe DFT value for the Au-CH3 bond distance is shorter (2.06 than given here. A local functional including relativistic effects was employed for the latter calculation.21 Nonlocal corrections would yield a longer Au-CH3 bond, which would be in agreement with the MP2 value reported here. Table 1shows that the M-C bond lengths of the M(CH& molecules (M = Zn, Cd, Hg) predicted at MP2/I are in excellent agreement with the experimental data. The experimental values have Other been taken from an IR spectroscopic experimental values using IR or electron diffraction are Because nearly the same as those given in Table of the good agreement between the MP2A values and the experimental data for the M-C distances of the M(CH3)2 compounds, we believe that the theoretical values for the Cu-CH3, Ag-CH3, and Au-CH3 bond lengths calculated a t MP2A should be accurate within f0.02 A. The calculated Cu-CsHs and Ag-C& bond lengths at MP2A are very interesting because of the controversy5 about the synthesis of the copper and silver aryl complexes reported by Lingnau and Strahle.4 The calculations predict that the Cu-C(pheny1) bond length is 1.850 A. The calculated Ag-C(pheny1) bond length is 2.091 A. The former value is in reasonable agreement with the experimental Cu-C(pheny1) bond length re-

A,

l.23136

topological analysis of t h e electron density distribution the programs GRID, CONTOUR, SADDLE, and GRDVEV were employed.33

3. Results and Discussion Table 1 shows the theoretically predicted and experimentally observed metal-carbon bond lengths. The M-C bond dissociation energies are shown in Table 2. The calculated Cu-CH3 bond length at MP2A (1.866 A) is significantly shorter than the previously reported distances at the MCPF level (1.936 A)17band at CCSD(T) (1.921 It is also shorter than the MP2 (31) (a) Haser, M.; Ahlrichs, R. J. Comput. Chem. 1989,10,104. (b) Ahlrichs, R.; Bar, M.; Haser, M.; Horn, H.; KBlmel, C. Chem. Phys. Lett. 1989, 162, 165. (c) Horn, H.; Weiss, H.; Haser, M.; Ehrig, M.; Ahlrichs, R. J . Comput. Chem. 1991, 12, 1058. (d) Haser, M.; Almlof, J.; Feyereisen, M. W. Theor. Chim. Acta 1991, 79, 115. (32)ACES II, an ab initio program system; Stanton, J. F.; Gauss, J.; Watts, J. D.; Lauderdale, W. J.; Bartlett, R. J. University of Florida: Gainesville, FL, 1991. (33) Biegler-Konig,F. W.; Bader, R. F. W.; Ting-Hua, T. J . Comput. Chem. 1982,3,317.

(34) Hope, H.; Olmstaed, M. M.; Power, P. p.; Sandell, J.; Xu, x. J. A m . Chem. Soc. 1985, 107,4337. (35) McKean, D. C.; McQuillan, G. P.; Thompson, D. W. Spectrochim. Acta 1980,36A, 1009. (36) (a) Kasiwabara, K.; Konakan, S.; Iijima, T.; Kimura, M. Bull. Chem. SOC.Jap. 1973,46,407. (b) McKean, D. C.; Duncan, J. L.; Batt, L. Spectrochim. Acta 1973, 29A, 1037. (c) Rao, S.; Stroicheff, B. P.; Turner, R. Can. J . Phys. 1960,38, 1516.

4266 Organometallics, Vol. 14, No. 9, 2995

Antes and Frenking

(a>

Figure 1. Calculated lowest energy conformations of (a) M(CH& (M = Zn, Cd, Hg) and (b) M(C&JZ (M= Zn, Cd, Hg).

ported by LS (1.890 A). The calculated Ag-C(pheny1) distance at MP2A is much longer, however, than the value reported by LS (1.902 A). Because of the good agreement of the other calculated metal-carbon bond lengths at MP2A with experimental values shown in Table 1, we think that the bond distance reported by LS refers to a different compound. The calculated results support the suggestion of Haaland et aL5 that the complex observed by LS is not a silver-aryl complex. It should be noted that the calculated Ag-C& bond length at MP2A is in excellent agreement with the estimate made by Haaland et al. for a silver-aryl bond The calculations predict that the length (2.08 metal-C(pheny1) bond lengths of the group 11 and 12 metals are 0.01-0.03 8, shorter than the respective metal-C(methy1) bond lengths (Table 1). The MP2A result for the Hg-C distance of Hg(CsH& (2.085 A) is in excellent agreement with the experimental value (2.092 A),41while a previously reported MNDO value is somewhat too short (2.002 Ahz4 The dimethyl compounds have been calculated with staggered (D3d) and eclipsed (D3h) conformations. The D3h form was in all cases more stable than the D3d form. However, the energy differences were very low a t all levels of theory ( ~ 0 . 0 5kcaYmol). This indicates nearly free rotation around the M-methyl bonds. The instantaneous symmetry of M(CH3)z (M = Zn, Cd, Hg) and Cu(CH3)z- is D3. This is in agreement with several experimental studies of these molecules which made it clear that the barrier of internal rotation is negligible.37 The calculations predict that in the diphenyl compounds the phenyl rings are orthogonal to each other (Dw symmetry, Figure 1). The M-C bond dissociation energies are shown in Table 2. The energies refer to the dissociation into the for Cu, metal atoms in the electronic ground state Ag, and Au; 'S for Zn, Cd, and Hg) and the methyl and (37)(a) Boyd, D. R. J.; Thompson, H. W.; Williams, R. L. Discuss. Faraday SOC.1950, 9, 154. (b) Butler, I. S.; Newbury, M. L. Spectrochim. Acta 1977, A33, 699 and references therein.

phenyl radicals, respectively. The experimental values for the dissociation energies of M(CH312 (M = Zn, Cd, Hg) and Hg(CsH& have been derived from the heats of formation of the fragments.38 The theoretically predicted bond energies at CCSD(T)iW/MPBAare higher than in previous theoretical studies except the DFT calculations,21but they are still lower than experimentally observed. The DFT results have been obtained using local DFT theory, which tends to overestimate the bond strength. On the other hand, relativistic effects are not included in the DFT calculations. Because relativistic effects increase the Cu-CH3 bond strength,ln the DFT value for the dissociation energy is subject to a fortuitous error cancellation. For M(C6H& (M = Zn, Cd, Hg) the dissociation energies could only be calculated at MP2A because of the size of the molecules. The theoretical dissocation energies predicted at the MPmI level are higher than at CCSD(T)AI (Table 2). The Do values for the methyl compounds at MP2AI are in very good agreement with experiment. We think that this is fortuitous. The dissociation energy for Hg(C&)2 predicted at MP2A is clearly too high (Table 2). We want to point out that the trend of the bond energies among the different series is the same for the MP2 and CCSD(T)results. Because the calculated differences of the bond energies appear to be quite reliable among a series, the MP2A results for the M(CsH5)z compounds may be used to estimate the dissociation energies for M = Zn and Cd. Using the experimental Do value of Hg(C&)z as reference data, the MP2A calculations predict that Doof Cd(CsH5)zshould be -81 kcdmol and Do of Zn(C6H5)z should be -95 kcaYmo1. It is interesting to note that the calculations predict for the bond strenbh of the methvl and Dhenvl com(38)The heat of formation a t 298 K was used. In order t o calculate Do values at 0 K, the differences of DHf were taken. In addition, the differences in the number of degrees of freedom of translational (3 pV = 1.8kcallmol) and rotational motion (1.5 pV = 0.9 kcal/mol) plus the work term (2 pV = 1.2 kcal/mol) were taken. The corrections refer to the loss of two groups. Lias, S. G.; Bartmess, J. E.; Liebman, J. F.; Holmes, J. L.; Levin, R. D.; Mallard, W. G. J.Phys. Chem. Ref: Data 1988, 17 (Suppl. 1).

Organometallics, Vol. 14, No. 9, 1995 4267

Transition Metal Methyl and Phenyl Compounds

Table 3. Calculated and Experimental First and Second Metal-CH3 Bond Dissociation Energies De and Do (kcal/mol) of M(CH& (M = Zn, Cd, Hg). Geometries are Optimized at MP2/I (see Table 1) ~

HFDI

MP2iII

molecule

diss.

D.

D,

Do

D,

D,

Zn(CH& Zn(CH& Cd(CH3)z Cd(CH& Hg(CH& Hg(CH3)z

first second

43.9 3.5 34.6 -1.0 35.2 -11.3

75.1 15.3 64.6 9.2 66.1 0.5

70.0 11.2 60.3 7.5 60.9 -0.4

65.9 8.2 56.2 4.5 56.0 -3.4

68.0a 16.6b

first second

first second

CCSD(T)fll

exptl

51.6c

7.W

* Reference 40. Reference 39. Taking the differences of the heats of formation, ref 38.

Table 4. Results of the NBO Analysis for the Metal-C Bonds and Partial Charges q at MP2/Ia

more covalent than the Ag-C and Cu-C bond. This becomes evident by the calculated polarization given by %M, the Wiberg bond order, BO, and the partial charges at the metal atoms (Table 4). The metal-C bonds have nearly pure s-character at the metals with very little d-contribution. This holds also for the phenyl compounds of Cu, Ag, and Au. The polarization of the M-C bonds toward the carbon end is higher, however, for the phenyl compounds than for the methyls. This is reasonable, because the latter molecules have sp2 carbon atoms, while in the methyl compounds the carbon atoms are sp3 hybridized. The NBO analysis gives only one metal-carbon bond orbital for the dimethyl and diphenyl compounds of Zn, Cd, and Hg. This means that the best Lewis structure of these molecules should be written in mesomeric forms:

~~

CUCH~ 2.000 AgCH3 2.000 AuCH~ 2.000 CuC6& 1.963 AgC&, 1.954 A u C ~ H ~ 1.974 Zn(CH3)zb 1.997 Cd(CH3)Zb 1.996 HgfCH3)zb 1.995 Zn(C&)zb 1.967 Cd(CsH5)zb 1.965 Hg(CsHS)zb 1.970

21.5 25.9 38.1 19.3 19.6 34.7 20.0 22.7 29.2 17.8 19.9 26.8

94.7 94.1 84.6 95.1 95.5 83.5 95.9 93.9 86.9 96.8 94.8 87.6

0.3 0.9 0.3 0.3 2.1 0.2 1.5 1.5 1.1 0.7 0.7 0.4

5.0 5.0 16.1 4.6 2.4 16.3 2.6 4.6 12.0 2.5 4.5 12.0

0.68 0.78 0.96 0.62 0.61 0.91 0.46 0.49 0.52 0.43 0.45 0.50

+0.58

+0.49 +0.26 +0.62 +0.60 +0.33 f1.33 +1.26 fl.10 f1.38 +1.33 f1.16

-1.22 -1.13 -0.93 -0.50 -0.45 -0.29 -1.33 -1.29 -1.22 -0.65 -0.60 -0.54

a Occ gives the occupancy of the M-C natural bond orbital; %M gives the metal part of the M-C bond; %s(M), %p(M), and %d(M) give the hybridization of the M-C bond a t the metal end; BO gives the Wiberg bond order; q(M) and q(C) give the atomic partial charges of the metal and the carbon atom of the M-C bond. The NBO analysis gives only one M-C bond.

pounds of group 11 elements the order Au > Cu > Ag. A different order is calculated for the first, second, and third row transition metal (TM) elements of group 12. Here, the bond strength of the dimethyl and diphenyl compounds decreases with the order Zn > Cd > Hg. Thus, the third row TM element of group 11 has the strongest bond, while the third row TM element of group 12 has the weakest bonds (Table 2). This is in agreement with the experimental values for the dissociation energies of the dimethyl compounds of Zn, Cd, and Hg (Table 2). We calculated also the first and second M-CH3 dissociation energies of the dimethyl compounds M(CHd2 (M = Zn, Cd, Hg). The results are shswn in Table 3. The calculations predict that the first dissociation energy is much higher than the second. This is in agreement with experimental studies of Zn(CH3)z and H ~ ( C H ~ ) Z . The ~ ~ -calculations ~O suggest that HgCH3 is hardly bound. We analyzed the bonding situation of the metal-alkyl and metal-aryl compounds using the NBO method.13 The results are shown in Table 4. The metal-carbon bonds of CuCH3, AgCH3, and AuCH3 are polarized toward the carbon end. The polarization decreases with the order Cu > Ag >> Au. The NBO analysis suggests that in the methyl compounds the Au-C bond is clearly ~

(39) (a) Armentrout, P. B.; Kickel, B. L. Organometallic Ion Chemistry; Freiser, S. B., Ed.; Kluwer: Dordrecht, in press. (b) P. Armentrout, personal communication, 1994. (40) McMillan, D. F.; Golden, D. M. Ann. Rev. Phys. Chem. 1982, 33. 493. (41) Vilkov, L. V.; Anashkin, M. G.; Mamaeva, G. I. J . Struct. Chem. USSR 1968,372. - - I

~~~

H3C-M+ CH3-

-

H3C- M+-CH3

The calculated atomic partial charges at the metal atoms of the dimethyl and diphenyl compounds of group 12 elements are significantly higher (>1.0)than those of the group 11elements. The polarization of the M-C bond toward the carbon end of the group 12 elements (only one bond) is even higher than for the group 11 molecules. This indicates that the group 12 dimethyl and diphenyl compounds are more ionic and less covalent than the group 11methyl and phenyl compounds. The polarization of the M-C bonds toward carbon has the order Zn > Cd >> Hg. The hybridization at the metal atoms has, in all cases, dominantly s-character of the M-C bonds and very little d-contribution, which is a little higher for the Au and Hg compounds than for the other molecules. The polar character of the metal-carbon bond is also revealed by the Laplacian of the electron density distribution, @@(r).Figure 2 shows the contour line for AuCH3, AuCsHs, Hg(CH&, and diagrams of 62@(r) Hg(CsH&. The difference of the shape of the Laplacian distribution of the M-C and C-H bonds reflects the different nature of the bonds. The C-H bonds exhibit a continuous area of charge concentration (@e(r) < 0, solid lines), while the charge concentration of the M-C bonds is localized a t the carbon ends. The Laplacian distributions of the other molecules are very similar and, therefore, are not shown here. 4. Summary

The theoretically predicted geometries a t the MP2II level of theory for the methyl and phenyl compounds MCH3 and MC& (M = Cu, Ag, Au), Cu(CH&-, and M(CH3)2 and M(CsH5)z (M = Zn, Cd, Hg) are in very good agreement with available experimental values. The relativistic effect upon the Cu-C bond length is remarkable. The calculated value using relativistic pseudopotentials at MP2II is clearly shorter (1.866 & than in most previous studies. The calculated Ag-C& bond length is significantly longer than the Ag-C distance of the silver-aryl complex reported by Lingnau and Strahle.4 It is highly unlikely that the measured complex is really a silver compound. The theoretical metal-carbon bond dissociation energies calculated at the CCSD(T)/IIlevel are slightly lower than experimental data. The calculated M-C bond strengths of the

4268 Organometallics, Vol. 14,No. 9, 1995

Antes and Frenking

/

Figure 2. Contour line diagrams of the Laplacian distribution %(r) at MP2/I for (a) AuCH3, (b) AuC&, (c) Hg(CH&, and (d) Hg(C&)z. Dashed lines indicate charge depletion (Vg(r)> 01, and solid lines indicate charge concentration(V+(r) 0).The solid lines connecting the atomic nuclei are the bond paths, the solid lines separating the atomic nuclei indicate the zero-flux surfaces in the plane. The crossing points of the bond paths and zero-flux surfaces are the bond critical points, n,. methyl and phenyl compounds have the order Au > Cu > Ag for the group 11elements. The order for the group 12 elements is Zn > Cd > Hg. The analysis of the bonding situation indicates that the metal-carbon bonds are polarized toward the carbon end. The polarization shows the TM order first row > second row >> third row. The dimethyl and diphenyl compounds of Zn, Cd, and Hg have metal-C bonds which are more ionic than the methyl and phenyl compounds of Cu, Ag,and Au. The best Lewis structure of the former compounds has mesomeric forms with only one M-C bond. The atomic partial charges a t Zn, Cd,

and Hg are clearly larger than at Cu, Ag, and Au. The metal part of the M-C bonds has mainly s-character.

Acknowledgment. We thank Prof. Tom Ziegler for helpful information concerning the DFT calculations reported in ref 21. This work has been supported by the Deutsche Forschungsgemeinschaft (SFB 260 and Graduiertenkolleg) and the Fonds der Chemischen Industrie. We acknowledge generous support and excellent service by the computer centers HRZ Marburg, HHLRZ Darmstadt, and HLRZ Jiilich. OM950141N