Cundari Th.R. -- Computational Organometallic Chemistry-0824704789

338	Czerw et al.

When M Rh, we are unable to locate isomer 3a as a minimum. With the BLYP and MP2 methods, 3a possesses one imaginary frequency and is hence a transition state; at the B3LYP level, the structure is a second-order saddle point. All computational methods predict a minimum corresponding to the nonclassical cis isomer 4a (Fig. 5). In addition, the DFT methods predict a minimum corresponding to trans isomer 5a, whereas MP2 fails to locate a minimum for the tetrahydride. However, the computed 4a–5a difference is more than 20 kcal/mol (Table 4) in favor of 4a. This result may be yet another manifestation of the strong trans influence exerted by H, which renders 5a with two hydrides as a trans pair disfavored (48,61). Structure 6 appears as a transition state with this B3LYP method; at the BLYP and MP2 levels, any attempt at locating a di-dihy- drogen stationary point failed. Although the calculations do not present a fully

FIGURE 5 Optimized geometries of H4M(PH3)2Cl isomers, M Rh and Ir. Bond

˚

lengths in A, angles in degrees. Phosphine groups omitted for clarity. BLYP: regular font; B3LYP: italics font; MP2: bold font.

A Computational Study Using DFT and MO Methods					339
TABLE 4 Relative Enthalpies (∆H, kcal/mol) of H4M(PH3)2Cl Species
Species	BLYP	B3LYP	MP2	MP4(SDTQ)	CCSD(T)
M Rh
4a	0.0	0.0	0.0
5a	21.0	22.3	a
M Ir
4b	0.0	0.0	0.0	0.0	0.0
3b	5.6	7.3	1.3	1.3	2.0

a Not a stationary point on the MP2 surface.

uniform picture, they clearly favor nonclassical over classical structures for H4Rh(PH3)2Cl. Lin and Hall have pointed out that the presence of contracted metal d orbitals will tend to favor the nonclassical isomers where metal–hydrogen electron transfer is minimized (61,62). Cis isomer 4a is hardly bound relative to trans-2a and H2, with the computed enthalpy for the formation reaction ranging from slightly negative (∆H 1.3, 0.5, and 1.2 kcal/mol with MP2, MP4(SDTQ), and CCSD(T), respectively) to positive (∆H 2.0 kcal/mol and 2.9 kcal/mol with B3LYP and BLYP, respectively). Since stronger electron-do- nating phosphines favor H2 addition, it is likely that the formation enthalpies for 4a will become more negative by a few kilocalories per mole, when alkylated phosphines are employed. However, ∆G for this bimolecular reaction will remain substantially positive, and the equilibrium for the formation of 4a will thus lie far toward the reactants (trans-2a, H2) under normal experimental conditions.

When M Ir, we locate the classical, four-hydride isomer, 3b, and the nonclassical cis isomer, 4b, as minima with all computational methods. With the singular exception of MP2, the methods agree that 4b is slightly more stable than 3b. The 3b–4b enthalpy difference (Table 4) is more than 5 kcal/mol with the DFT methods, but decreases to 2 kcal/mol or less at the highly correlated levels (MP4(SDTQ): 1.3 kcal/mol; CCSD(T): 2.0 kcal/mol). Lin and Hall found that the use of PH3 rather than PMe3 in calculations tended to favor the nonclassical isomers (62), but there are no indications of the nonclassical trans isomer 5b (or of 6) when the computational method used for geometry optimization includes electron correlation (63). Relativistic effects (destabilization of the 5d orbitals) should preferentially favor classical isomers (64), and, indeed, we could not locate the classical tetrahydride when M Rh (see earlier). There is NMR evidence pointing to a nonclassical structure for H4Ir(PiPr3)2Cl (65), in accord with the computational results (Table 4). According to the MO-based correlation methods, the seven-coordinate species 4b is moderately bound with respect to trans-2b

340	Czerw et al.

FIGURE 6 Optimized geometries for transition states 7 and 8. Bond lengths

˚

in A, angles in degrees. Phosphine groups omitted for clarity. BLYP: regular font; B3LYP: italics font; MP2: bold font.

A Computational Study Using DFT and MO Methods

341

and H2 (∆H 8.6 kcal/mol (MP2), 8.1 kcal/mol (MP4), 6.8 kcal/mol CCSD(T). However, the formation reaction is predicted to be essentially thermoneutral at the DFT levels [∆H 0.1 kcal/mol (BLYP), 1.9 kcal/mol (B3LYP)].

The transition state leading to the nonclassical cis isomer (7, Fig. 6) finds

˚

H2 at a large distance ( 2.6 A) from the metal center and only slightly activated

˚

(H–H 0.75 A). The transition state leading to 4 is only 1–3 kcal/mol above the reactants for both M Rh and M Ir. We have been unable to find a transition state, which leads directly to the classical isomer 3b or to the trans nonclassical isomer 5a from the separated reactants. However, 3b should be readily formed by intramolecular rearrangement. Transition state 8b (Fig. 6), which connects 4b and 3b, is located only 2.8 kcal/mol [CCSD(T)] above 4b. The classical tetrahydride 3b forms only a shallow minimum, since 3b and 8b are computed to be very close energetically [0.3 kcal/mol at CCSD(T)] and structurally (cf. Fig. 5 and 6) by all methods. On the MP2 surface for M Rh, the four-hydride species 8a represents the transition state for the degenerate interconversion of the two equivalent nonclassical cis isomers 4a; 8a is 13.0 kcal/mol higher in enthalpy than 4a.

4. CONCLUSIONS

All three computational methods used here for geometry optimizations (BLYP, B3LYP, and MP2) produce comparable structures for all the isomers. Bond lengths from MP2 are shorter than those obtained from DFT (Figs. 1–3, 5, 6); bond lengths from B3LYP tend to be slightly shorter than those from BLYP, probably reflecting the small admixture of Hartree–Fock exchange present in the B3 functional. There is also general agreement among the methods regarding the relative energies of isomers (Tables 1 and 3). In particular, for M(PH3)2Cl (MRh and M Ir) the singlet TPH3 structure is clearly the preferred isomer. It is noteworthy that the enthalpy differences among the M(I) and M(III) isomers predicted by the B3LYP method are very similar to those predicted by the far more elaborate CCSD(T) method (66). Large differences appear in computed reaction enthalpies for dihydrogen addition, with the MO-based methods [MPn, CCSD(T)] predicting considerably higher exothermicities, which translate into larger M–H bond energies. The MO-based results appear to be closer to the available (limited) experimental data, and the DFT methods thus underestimate the M–H bonding energies, although they do produce the better results for the intrinsic H–H bond enthalpy. The apparent ability of the MPn/CCSD methods to form stronger M–H bonds is on display in the Ir(V) complexes, where a very small enthalpy difference is predicted between classical and nonclassical isomers.

The structural and energetic influences exerted by bulky phosphines continue to be of interest. Unfortunately, the dramatic scaling of MPn/CCSD(T)

342	Czerw et al.

calculations with molecular size makes it impossible to perform these highly accurate calculations on large systems (67). DFT calculations scale less unfavorably with molecular size and would seem to be the method of choice for further investigations of such ‘‘substituent’’ effects.

ACKNOWLEDGMENTS

We gratefully acknowledge the National Science Foundation for financial support (CHE-9704304) and for a computer equipment grant (DBI-9601851-ARI). We thank Professor A. S. Goldman for stimulating discussions.

NOTES AND REFERENCES

1.PWNM van Leeuwen, JH van Lenthe, K Morokuma, eds. Theoretical Aspects of Homogeneous Catalysis, Applications of Ab Initio Molecular Orbital Theory. Dordrecht, The Netherlands: Kluwer, 1994.

2.DG Truhlar, K. Morokuma, eds. Transition State Modeling for Catalysis. ACS Symposium Series No. 721. Washington, DC: ACS, 1998.

3.L Deng, T Ziegler, TK Woo, P Margl, L Fan. Organometallics 17:3240–3253, 1998.

4.MT Benson, TR Cundari, ML Lutz, SO Sommerer. In: D Boyd, K Lipkowitz, eds. Reviews in Computational Chemistry. Vol. 8. New York: VCH, 1996, pp 145–202.

5.M Springborg, ed. Density-Functional Methods in Chemistry and Materials Science. London: Wiley, 1997.

6.J Burdeniuc, B Jedlicka, RH Crabtree. Chem Ber/Recueil 130:145–154, 1997.

7.K Nomura, Y Saito. Chem Commun 161, 1988.

8.T Sakakura, T Sodeyama, M Tokunaga, M Tanaka. Chem Lett 263–264, 1988.

9.W Xu, GP Rosini, M Gupta, CM Jensen, WC Kaska, K Krogh-Jespersen, AS Goldman. Chem Commun 2273–2274, 1997.

10.S Niu, MB Hall. J Am Chem Soc 121:3992–3999, 1999.

11.K Krogh-Jespersen, M Czerw, M Kanzelberger, AS Goldman. J Chem Info Comput Sci. In press.

12.A Veillard. Chem Rev 91:743–766, 1991.

13.MS Gordon. In: DR Yarkony, ed. Modern Electronic Structure Theory. Singapore: World Scientific, 1994, pp 311–344.

14.CW Bauschlicher Jr, SR Langhoff, H Partridge. In: DR Yarkony, ed. Modern Electronic Structure Theory. Singapore: World Scientific, 1994, pp 1280–1374.

15.RG Parr, W Yang. Density-Functional Theory of Atoms and Molecules. Oxford, UK: Oxford University Press, 1989.

16.WJ Hehre, L Radom, JA Pople, PvR Schleyer. Ab Initio Molecular Orbital Theory. New York: Wiley-Interscience, 1986.

17.JB Foresman, A Frisch. Exploring Chemistry with Electronic Structure Methods. Pittsburgh, PA: Gaussian, 1996.

18.M Head-Gordon, JA Pople, M Frisch. J Chem Phys Lett 153:503–506, 1988.

19.R Krishnan, MJ Frisch, JA Pople. J Chem Phys 72:4244–4245, 1980.

20.JA Pople, M Head-Gordon, K Raghavachari. J Chem Phys 87:5968–5975, 1987.

A Computational Study Using DFT and MO Methods

343

21.GD Purvis, RJ Bartlett. J Chem Phys 76:1910–1918, 1982.

22.AD Becke. Phys Rev A 38:3098–3100, 1988.

23.C Lee, W Yang, RG Parr. Phys Rev B 37:785–789, 1988.

24.AD Becke. J Chem Phys 98:5648–5652, 1993.

25.PJ Hay, WR Wadt. J Chem Phys 82:270–283, 1985.

26.TH Dunning, PJ Hay. In: HF Schaefer III, ed. Modern Theoretical Chemistry. New York: Plenum, 1976, pp 1–28.

27.R Krishnan, JS Binkley, R Seeger, JA Pople. J Chem Phys 72:650–654, 1980.

28.JS Binkley, JA Pople, WJ Hehre. J Am Chem Soc 102:939–947, 1980.

29.WJ Hehre, RF Stewart, JA Pople. J Chem Phys 51:2657–2664, 1969.

30.HB Schlegel. In: DR Yarkony, ed. Modern Electronic Structure Theory. Singapore: World Scientific, 1994, pp 459–500.

31.DA McQuarrie. Statistical Thermodynamics. New York: Harper and Row, 1973.

32.Gaussian 98 (Revision A.5). MJ Frisch, GW Trucks, HB Schlegel, GE Scuseria, MA Robb, JR Cheeseman, VG Zakrzewski, JA Montgomery, RE Stratmann, JC Burant, S Dapprich, JM Millam, AD Daniels, KN Kudin, MC Strain, O Farkas, J Tomasi, V Barone, M Cossi, R Cammi, B Mennucci, C Pomelli, C Adamo, S Clifford, J Ochterski, GA Petersson, PY Ayala, Q Cui, K Morokuma, DK Mailick, AD Rabuck, K Raghavachari, JB Foresman, J Cioslowski, JV Ortiz, BB Stefanov, G Liu, A Liashenko, P Piskorz, I Komaromi, R Gomperts, RL Martin, DJ Fox, T Keith, MA Al-Laham, CY Peng, A Nanayakkara, C Gonzalez, M Challacombe, PMW Gill, BG Johnson, W Chen, MW Wong, JL Andres, M Head-Gordon, ES Replogle, JA Pople. Pittsburgh, PA: Gaussian, 1998.

33.TA Albright, JK Burdett, M-H Whangbo. Orbital Interactions in Chemistry. New York: Wiley-Interscience, 1985.

34.N Koga, K Morokuma. J Phys Chem 94:5454–5462, 1990.

35.C Daniel, N Koga, J Han, XY Fu, K Morokuma. J Am Chem Soc 110:3773, 1988.

36.N Koga, K Morokuma. J Am Chem Soc 115:6883–6892, 1993.

37.DG Musaev, K Morokuma. J Organomet Chem 504:93–105, 1995.

38.MRA Blomberg, PEM Siegbahn, M Svensson. J Am Chem Soc 114:6095–6102, 1992.

39.J-F Riehl, Y Jean, O Eisenstein, M Pelissier. Organometallics 11:729–737, 1992.

40.P Margl, T Ziegler, PE Bloechl. J Am Chem Soc 117:12625–12634, 1995.

41.M-D Su, S-Y Chu. J Am Chem Soc 119:10178–10185, 1997.

42.YW Yared, SL Miles, R Bau, CA Reed. J Am Chem Soc 99:7076–7078, 1977.

43.A Dedieu, I Hyla-Kryspin. J Organomet Chem 220:115–123, 1981.

44.TR Cundari. J Am Chem Soc 116:340–347, 1994.

45.GP Rosini, F Liu, K Krogh-Jespersen, AS Goldman, C Li, SP Nolan. J Am Chem Soc 120:9256–9266, 1998.

46.K Krogh-Jespersen, AS Goldman. In: DG Truhlar, K. Morokuma, eds. Transition State Modeling for Catalysis. ACS Symposium Series No. 721. Washington, DC: ACS, 1998, pp 151–162.

47.PEM Siegbahn. J Am Chem Soc 118:1487–1496, 1996.

48.JD Atwood. Inorganic and Organometallic Reaction Mechanisms. Monterey, CA: Brooks/Cole, 1985.

49.CJ van Wu¨llen. J Comput Chem 18:1985–1992, 1997.

344	Czerw et al.

50.K Wang, GP Rosini, SP Nolan, AS Goldman. J Am Chem Soc 117:5082–5088, 1995.

51.We have not performed a thorough search of possible dimer configurations. The structures outlined in Figure 2 represent minima at BLYP, B3LYP, and MP2 levels of theory.

52.M Gupta, C Hagen, RJ Flesher, WC Kaska, CM Jensen. Chem Commun 2083– 2084, 1996.

53.RG Pearson. Acc Chem Res 4:152–160, 1971.

54.GS Hammond. J Am Chem Soc 77:334–338, 1955.

55.DR Lide, ed. CRC Handbook of Chemistry and Physics. 71st ed. Boca Raton, FL: CRC Press, 1990.

56.WHE Schwarz, EM van Wezenbeek, EJ Baerends, JG Snijders. J Phys B At Mol Opt Phys 22:1515–1523, 1989.

57.FA Cotton, G Wilkinson. Advanced Inorganic Chemistry. 6th ed. New York: Wiley, 1999.

58.G Kubas. Acc Chem Res 21:120–128, 1988.

59.PG Jessop, RJ Morris. Coord Chem Rev 121:155–284, 1992.

60.Z Lin, MB Hall. Inorg Chem 31:4262–4265, 1992.

61.Z Lin, MB Hall. J Am Chem Soc 114:6102–6108, 1992.

62.Z Lin, MB Hall. J Am Chem Soc 114:2928–2932, 1992.

63.Structure 5b is a minimum at the Hartree–Fock level; see Ref. 61.

64.J Li, RM Dickson, T Ziegler. J Am Chem Soc. 117:11482–11487, 1995.

65.M Mediati, GN Tachibana, CM Jensen. Inorg Chem 29:3–5, 1990.

66.RK Szilagyi, G Frenking. Organometallics 16:4807–4815, 1997.

67.BG Johnson, PMW Gill, JA Pople. J Chem Phys 97:7846–7848, 1992.

14

The Electronic Structure of Organoactinide Complexes via Relativistic Density Functional Theory: Applications to the Actinocene Complexes An(η8-C8H8)2

(An Th–Am)

Jun Li and Bruce E. Bursten

The Ohio State University, Columbus, Ohio

1. INTRODUCTION

The synthesis and characterization of the sandwich complex ferrocene, Fe(η5-C5H5)2, in 1951 stands as the starting point for modern organometallic chemistry (1). Since then, organometallic chemistry has been one of the most rapidly developing new fields in modern chemistry (2). In the development of modern organometallic chemistry, cyclopolyene and cyclopolyenyl systems of ηn-CnHn carbocyclics, such as C5H5 (Cp), C6H6 (Bz), C7H73 (Cht), and C8H82 (Cot) are among the most frequently used ligands for both transition metal and f-block element organometallic complexes (3,4). Transition metal and f-element sandwich complexes formed by these ligands have played a central role in understanding the fundamental bonding, electronic structures, and chemical properties of organometallic complexes.

Since the synthesis of ferrocene, a multitude of interesting ηn-CnHn (n 4, 5, 6, 7, 8) sandwich complexes of various transition metals and f-elements have been reported. Among these developments, perhaps the most interesting are the syntheses of bis(tetraphenylcyclobutadiene)-nickel and -palladium, M(η4-

345

346	Li and Bursten

C4Ph4)2 (M Ni, Pd) (5,6), bis(benzene)chromium(0), Cr(η6-C6H6)2 (7), and bis(cyclooctatetraene)uranium(IV), uranocene or U(η8-C8H8)2 (8), the structures of which are shown in Figure 1. The synthesis of uranocene, and its structure and chemistry via-a`-vis those of ferrocene, provided a remarkable example of the similarities and differences between organotransition metal chemistry and organo-f-element chemistry.

While sandwich complexes of the Cp, Cp* (C5Me5 ), and Bz rings are prevalent for transition metals, the development of the chemistry of actinide sandwich complexes has focused mainly on complexes of the Cot ligand; because of the larger size of the actinide metal atoms, complexes with two Cp or Cp* ligands always involve the coordination of additional ligands. Although uranocene and other actinocenes were synthesized in the late 1960s, actinide sandwich complexes of the cycloheptatrienyl (Cht) and benzene (Bz) ligands were unknown until very recently. In 1995, the first bis(cycloheptatrienyl) actinide complex, namely, the U(Cht)2 anion, was synthesized and characterized crystallographically by Ephritikhine and coworkers (9). Meanwhile, the bis(arene) actinide cationic complexes [Bz2An] and [(TBB)2An] (An Th, U) were also observed in the past several years by Pires de Matos, Marshall, and coworkers via mass spectrometry (10), which suggests that it might be possible to synthesize and isolate neutral bis(arene) actinide sandwich complexes. These discoveries have opened a new chapter in the chemistry of organoactinide sandwich complexes, and a comprehensive theoretical investigation is thus needed to provide systematic comparisons of the electronic structures of these unique new complexes.

To date, a great number of experimental and theoretical investigations have been carried out to elucidate the structures and chemical reactivities of transition metal sandwich complexes (11). The parallel interest in investigations of actinide sandwich complexes has revived again recently (4,12). We have found relativistic density functional theory (DFT) to be a superb tool for the elucidation of the

FIGURE 1 Structures of the M(CnRn)2 (n 4–8) sandwich complexes. The year in which the first member of each class of sandwich complex was synthesized is given below the structure.

Relativistic DFT and Organoactinide Complexes

347

electronic structural aspects of organoactinide complexes. We have recently used this methodology to explore aspects of the structure and bonding of a number of organoactinide sandwich complexes An(ηn-CnHn)2 (n 6, 7, 8) (13,14,15). In this chapter, we will discuss some practical aspects of the application of relativistic DFT to actinide complexes. We will then discuss some specifics of the bonding in organoactinide sandwich complexes, with particular emphasis on some recent detailed studies of the geometries, electronic structures, and vibrational properties of the actinocenes. Our goal is to demonstrate the utility of relativistic DFT in charting the future theoretical and experimental studies of organoactinide complexes.

2.THE CHALLENGES OF THEORETICAL ACTINIDE CHEMISTRY

The application of theoretical electronic structure methods to organoactinide complexes has long been hampered by some well-known challenges. Before we detail the computational methodology used in the present studies, we will briefly summarize some of the particular challenges inherent in the theoretical study of actinide complexes:

•Because of the importance of the An 5f orbitals in chemical bonding, the theoretical method chosen must be able to accommodate f orbitals, and, in order to investigate large actinide-containing systems, must do so with good computational efficiency.

•The radial distributions of the An 6s and 6p‘‘semicore’’ orbitals for the early actinides lie in the valence region. Thus, the calculations necessarily involve a great number of ‘‘valence’’ orbitals, including at least the An 6s, 6p, 6d, 5f, 7s, and 7p atomic orbitals in the valence space and a large number of valence electrons in the variational calculations (16,17).

•Dynamical electron correlation effects, i.e., the instantaneous correlation in the motions of electrons at short interelectronic distances, are so important for the heavy-element systems that exclusion of these effects in theoretical calculations of actinide complexes might lead to incorrect conclusions (18).

•The An 5f and 6d orbitals are very close in energy, so many low-lying, near-degenerate states exist for each given electron configuration. This complication poses a great challenge for conventional correlated ab initio calculations because a multiconfiguration and/or multireference scheme with large active configuration space is generally necessary to account for the nondynamical electron correlation effects arising from the degeneracy or near-degeneracy of different electron configurations.