Organometallics 1996, 14, 224-230
224
Trends in Structure and Bonding of Fischer Type Chromium Carbenes and Silylenes. A Density Functional Study Heiko Jacobsen and Tom Ziegler" Department of Chemistry, University of Calgary, 2500 University Drive N . W., Calgary, Alberta, Canada T2N IN4 Received July 11, 1994@ Density functional calculations were performed on a variety of carbene and silylene complexes involving the chromium pentacarbonyl fragment (CO)&r=ERz, with ER2 = CHZ, CFz, CC12, CMe2, CMe(OMe), SiHz, SiFz, SiClZ, SiMez, and SiMe(0Me). Further, a calculation on the base-stabilized silylene complex (CO)5Cr=Si(OPH3)Clz has been included. A bond analysis reveals that, compared with carbene compounds, silylene complexes form only weak n bonds with the metal center. Consequences for the coordination chemistry of silylene complexes as well as their reaction chemistry are discussed.
Introduction The concept of a double bond between transition metals and carbon constitutes one of the important elements in the field of organometallic chemistry. The notion of a metal-carbon double bond was first brought forward by Fischer and Maasboll in 1964. They suggested that the product of the reaction between LiMe and W(CO)6, followed by protonation and subsequent treatment with CH2N2, should be formulated as (C0)5W=CMe(OMe). The chemistry of transition metal carbenes2 developed rapidly, and these compounds have been established as valuable species in organic syntheses3 as well as in catalytic p r o ~ e s s e s . ~ The field of the higher homologues of Fischer carbenes was pioneered by Marks5 in the early seventies. Up to now, a larger number of germylene, stannylene, and even plumbylene systems have been structurally chara ~ t e r i z e d . ~Several ,~ modes of complexation are observed, two of which are displayed in Chart 1. We follow the notation of PetzGaand denote the structures as type I and type I1 complexes, respectively. In these arrangements, ML, represents the transition metal fragment, E is a group XIV element, X and Y represent o-bonded substituents, and B stands for a neutral Lewis base molecule. Structures of type I complexes with E = Ge, Sn reveal a trigonal planar coordination around the divalent E center in the solid state. The M-E bond lengths are noticeably shortened compared t o known M-E single bonds. An example of analogous type I and type I1 stannylene complexes6a Abstract published in Advance ACS Abstracts, November 15,1994. (1)Fischer, E.0.;Maasbol, A. Angew. Chem., Int. Ed. Engl. 1964, 3,580. (2) Dotz, K.H.; Fischer, H.; Hofmann, P.; Kreissl, F. R.; Schubert, U.; Weiss, K. Transition Metal Carbene Complexes; Verlag Chemie: Weinheim, Germany, 1983. (3) Dotz, K.H. Pure Appl. Chem. 1983,55, 1689. (b) Dotz, K.H. Angew. Chem., Int. Ed. Engl. 1984,23, 587. (c) Brookhart, M.; Studabaker, W. B. Chem. Rev. 1987,87,411.(d) Hegedus, L.S. Pure Appl. Chem. 1990,62,691.(e) Schmalz, H. G.Angew. Chem., Int. Ed. Engl. 1994,33,303. (4) Grubbs, R. H. In Comprehensive Organometallic Chemistry; Wilkinson, G., Stone, F. G. A. Abel, E. W., Eds.; Pergammon Press: Oxford, U.K., 1982; Vol. 8. ( 5 ) Marks. T. J. J . A m . Chem. SOC.1971.93.7090. (6)(a) Petz, W. Chem. Rev. 1986,86,1019:(bj Lappert, M. S.; Rowe, R. S. Coord. Chem. Rev. 1990,100,267. (7) Tokitoh, N.; Manmaru, K.; Okazaki, R. Organometallics 1994, 13,167. @
0276-7333/95/2314-0224$09.00/0
Chart 1
/x
ML,-E
MLm-E\Y I
/B
'YX I1
indicates an increase of the M-E bond length by about 10 pm under complexation of the Lewis base. For the ) ~ * Cand ~H~N, type I1 compound ( ~ - C ~ H ~ ) Z S ~ C ~ ( C O Brice Cotton8 describe the coordination around the tin center as an intermediate between a planar SnCrCz group with pyridine approaching perpendicularly and a tetrahedral SnCrCzN group. In contrast to germylene and its higher analogues, the area of transition metal silylene chemistry has been developed rather recently and represents a field of ongoing research a ~ t i v i t y .In ~ 1987, the cationic'O and neutral'l type I1 coordination compounds of silylenes were discovered. In the following years, Zybill and coworkers prepared a broad variety of base-stabilized silylene type I1 complexes12with the low-valent metal fragments Fe(C014 and Cr(C0)5. All these complexes are structurally well characterized, and for several of the compounds the solid state structures have been determined. Also a few type I silylene complexes have been i d e n t i f ~ e d . l ~Very ~ J ~ recently, the first structure of a cationic type I silylene complex has been reported.13b The reaction chemistry of the heavier carbene ana~,~ logues is still in an early process of d e ~ e l o p m e n tand not too many transformations involving silylene, germylene, and stannylene complexes are known in the literature. Most of the reactions deal with the modification of the E(B)XYunit of type I1 complexes. However, Zybill and co-workers reported the interesting Sila(8)Brice, M. D.; Cotton, F. A. J . A m . Chem. SOC.1973,95,4529. (9)Zybill, C. Top. Curr. Chem. 1991,160, 1. (10)Straus, D. A.; Tilley, T. D.; Rheingold, A. L.; Geib, S. J. J . Am. Chem. Soc. 1987,109,5872. (11)Zybill, C.; Muller, G. Angew. Chem., Int. Ed. Engl. 1987,26, 669. (12) (a) Zybill, C.; Muller, G. Organometallics 1988,7, 1368. (b) Leis, C.;Wilkinson, D. L.; Handwerker, H.; Zybill, C.; Muller, G. Organometallics 1992,11, 514. (c) Handwerker, H.; Paul, M.; Riede, J.;Zybill, C. J . Organomet. Chem. 1993,459,151. (13) (a) Leis, C.; Lachmann, H.; Muller, G.; Zybill, C. Polyhedron 1991, 10, 1163. (b) Grumbine, S. K.; Tilley, T. D.; Arnold, F. P.; Rheingold, A. L. J . A m . Chem. SOC.1994,116, 5495.
0 1995 American Chemical Society
Fischer Type Chromium Carbenes and Silylenes
Scheme 1
Wittig reaction12b of (CO)sCr=SiMez with dimethyl carbonate, yielding (CO)&r=C(OMe)2 and hexamethyltrisiloxane, [MezSiO-Is. Calculations on Fischer type carbenes possessing a fully saturated coordination sphere provide a formidable task for computational chemistry. The first computations reported for the complexes (CO)5Cr=CXY (X = OMe, NH2, Y = Me, OEt, Ph; X = NMez, Y = Me, Ph; X = SMe, Y = Me) have been performed using the Fenske-Hal1.meth0d.l~ On the basis of the result of a population analysis, qualitative trends in o-donor as well as n-acceptor contributions from the various ligands were evaluated. During the eighties, Fischer type complexes with transition metal carbonyl fragments became accessible for ab initio calculations. In 1981, Spangler15and co-workers reported the electronic structure and the optimized Ni=CH2 bond length of (c0)~NiCHz. Two years later, Nakatsuji and co-workers16a presented studies on the Fischer type compounds (ColaCr=CMe(OMe) and (CO)&=CMe(OMe), providing M=C bond energies and optimized M-C bond distances. These authors also investigated the silylene compound16b Cr(CO)&iH(OH), demonstrating the possible existence of a transition metal silylene complex. In their classical work on transition metal carbenes, Taylor and Hall17 made an essential contribution to the understanding of the nature of the bond in Fischer carbenes. The metalcarbon double bond is best described as o h dative interactions between two singlet fragments, as shown in Scheme 1. Recently, M6rquez and Fernandez S a d 8 presented fully optimized SCF geometries of the molybdenum complexes (C0)5Mo=EH2 (E = C, Si, Ge, Sn). They also employed the CASSCF method to get accurate values for the Mo=E bond distances and bond energies. Furthermore, MNDO calculations on chromium-based Fischer-type complexes of the type (C0)5Cr=EX2 (E = C, Si, Ge, Sn, Pb; X = H, C1) have been performed by Abroninlg and co-workers. Approximate density functional theory (DFT) has been proven to be a powerful computational tool in determining the structures and energetics of transition metal complexes.20 Thus, DFT seems to be the method of choice for careful investigation of the structures and bonding in Fischer type complexes. In connection with the generalized transition method, accurate total bond(14) (a) Block, T. F.; Fenske, R. F.; Casey, C. P. J . Am. Chem. SOC. 1976, 98, 441. (b) Block, T. F.; Fenske, R. F. J . Organomet. Chem. 1977,139, 235. (15) Spangler, D.; Wendoloski, J. J.; Dupuis, M.; Chen, M. M. L.; Schaefer, H. F., 111. J . Am. Chem. Soc. 1981, 103, 3985. (16) (a) Nakatsuji, H.;Ushio, J.;Han, S.; Yonezawa, T. J . Am. Chem. Soc. 1983, 105, 426. (b) Nakatsuji, H.; Ushio, J.; Yonezawa, T. J. Organomet. Chem. 1983,258, C1. (17) Taylor, T. E.; Hall, M. B. J . Am. Chem. SOC.1984, 106, 1576. (18) Mbrquez, A.; Fembndez Sanz, J. J . Am. Chem. SOC.1992,114, 2903. (19) Abronin, I. A.; Avdyuhina, N. A.; Morozova, L. V.; Magomedov, G. K.-I. J . Mol. Struct. 1991,228, 19. (20) (a) Ziegler, T. Pure Appl. Chem. 1991,28, 1271. (b) Ziegler, T. Chem. Rev. 1991,91, 651. (21) Ziegler, T.; Rauk, A. Baerends, E. J. Theor. Chim. Acta 1977, 43, 261.
Organometallics, Vol. 14,No. 1, 1995 225
ing energies2lyz2are available, and the bond energy can be analyzed in terms of steric as well as orbital interact i o n ~ .Our ~ ~ studies were prompted not only by the recent developments in the field of silylene chemistryg-l3 but also by the continued interest in an understanding of the physical properties of Fischer c a r b e n e ~ .In ~~ our first effort,z5awe investigated the role of the transition metal M as well as of the main group element E on the bonding in complexes of the type (C0)5M=EH2. We further discussed the influence of nonlocal corrections and relativistic effects on geometries and bond energies of carbene complexes,25bin order to judge the quality of our theoretical model. The present work deals with the influence of the modification of the R substituent in ER2 carbene and silylene systems. In contrast to the early Fenske-Hall studies, we will base our selection of ligands on experimentally known silylene ligands rather than on typical Fischer type carbenes. We further present calculations on a “real life” Fischer carbene. It is still an interesting question how well DFT is able to handle the molecular structures of these challenging and rather complex molecules. Finally, we investigate one silylene type I1 compound. To our knowledge, so far no theoretical studies have been reported on this type of complexes. We hope t o provide new insight into the different coordination as well as reaction chemistry of silylene complexes compared to their carbon analogues.
Computational Details The calculations in this work are based on approximate density functional theory within the local density approximationZ6(LDA) in the parametrization of Vosko, Wilk, and N ~ s a i r The . ~ ~exchange factor, G ~was , given the usual value of V 3 . Bond energies were evaluated by adding Becke’s nonlocal exchange correction2s as well as Perdew’s inhomogeneous gradient correction for correlationz9as a perturbation (LDA/NL). In a more sophisticated approach, the before mentioned corrections were added self-consistently3”(NL-
SCF). We utilized the AMOL program package for density functional calculations, which was developed by Baerends31and co-workers. The numerical integration scheme employed was that of te Velde3zand co-workers. A triple 5-STO basis33was used to describe the ns,np,nd,( n l)s, and ( n l)p shells of
+
+
(22) Ziegler, T.; Rauk, A. Theor. Chim. Acta 1977, 46, 1. (23) (a) Baerends, E. J.; Rozendaal, A. NATO ASZ 1986, C176,159. (b) Ziegler, T. NATO A S I 1991, C378, 367. (24) (a) Gandler, J. P.;Bernasconi, C. F. Organometallics 1989, 8, 2282. (b) Bemasconi, C. F.; Stronach, M. W. J.Am. Chem. SOC.1993, 115, 1341. (c) Bernasconi, C. F.; Sun, W. J . Am. Chem. SOC.1993, 115, 12526. (d) Bemasconi, C. F.; Flores, F. X.; Gandler, J. R.; Leyes, A. E. Organometallics 1994, 13, 2186. (25)(a) Jacobsen, H.; Ziegler, T. Znorg. Chem., submitted for publication. (b) Jacobsen, H.; Schreckenbach, G.; Ziegler, T. J . Phys. Chem. 1994,98, 11406. (26) (a) Gunnarsson, 0.;Lundquist, I. Phys. Rev. 1974, B10, 1319. (b) Gunnarsson, 0.; Lundquist, I. Phys. Rev. 1976, B13, 4274. (27) Vosko, S. J.; Wilk, M.; Nusair, M. Can. J . Phys. 1980,58,1200. (28) (a) Becke, A. J. Chem. Phys. 1986,84, 4524. (b) Becke, A. J. Chem. Phys. 1988,88,1053. (c) Becke, A. Phys. Rev. 1988, A38,3098. (29) (a) Perdew, J. P. Phys. Rev. 1986, B33,8822. (b) Perdew, J. P. Phys. Rev. 1986, B34, 7406. (30) (a) Fan, L.; Ziegler, T. J . Chem. Phys. 1991,94,6057. (b) Fan, L. Ph.D. Thesis, University of Calgary, 1992. (31) (a) Baerends, E. J.; Ellis, D. E.; Ros, P.E. Chem. Phys. 1973, 2,41. (b) Baerends, E. J. Ph.D. Thesis, Vrije Universiteit Amsterdam, 1975. (32) teVelde, G.; Baerends, E. J. J . Comp. Phys. 1992, 99, 84. (33) (a) Snijders, G. J.; Baerends, E. J. Vemoijs, P. At. Nucl. Data Tabl. 1982,26, 483. (b) Vemooijs, P.; Snijders, G. J.; Baerends, E. J. Slater Type Basis Functions for the Whole Periodic Table; Internal Report; Vrije Universiteit: Amsterdam, 1981.
Jacobsen and Ziegler
226 Organometallics, Vol. 14,No. I, 1995 Cr. For H, a double 5-STO basis33was extended by one 2pSTO polarization function. The ns and np orbitals of the remaining main group elements were described by a double C-STO augmented by one (n 1)d-STOpolarization function for second row elements, and one nd-STO polarization function for higher row elements. Electrons in lower shells were treated within the frozen core approximation as outlined by Baerends and co-workers. An auxiliary set34of s, p, d, f, and g type STO functions, centered on all nuclei, was used to fit the molecular density as well as present Coulomb and exchange potentials accurately in each SCF cycle. The geometry optimization procedure was based on a method developed by Versluis and Ziegler.34
+
Results and Discussion We will begin our discussion with a short introduction to our method of bond energy decomposition. Since we already presented a detailed account of the bond analysis for Fischer complexes,25awe restrict ourselves to a concise description. We analyze the Cr=E bond energy in (C0)5Cr=ERz complexes by considering the bond-forming reaction between the chromium pentacarbonyl fragment and the ER2 moiety: (CO),Cr
+ ER2 - (CO),Cr=ER2
(1)
The energy associated with reaction 1 is called the bondsnapping energy BEsnap.It can be decomposed into two main components, namely the steric repulsion term AEo and the electronic interaction term m i n t :
+ mint]
BEsnap= -[m0
AEo is in most cases dominated by the contribution of the so-called Pauli repulsion, which is directly related to the two-orbital three or four electron interactions between occupied orbitals on both fragments. Whereas AEo is mostly destabilizing in nature, the term AEint introduces the attractive orbital interactions between occupied and virtual orbitals on the two fragments. We will picture the bond as an interaction of two singlet fragments, as shown in Scheme 1. It has been argued25a that for carbene complexes with low-valent, late transition metals this bond description is most appropriate, even if the ER2 ligand has a triplet ground state. The bond energy BE is obtained when the bondsnapping energy is corrected by the preparation energy AEprep:
(3) Since the equilibrium geometry of the fragments usually differs from their arrangement in the framework of the final molecule, a small geometric preparation energy is required to get the fragments ready for bonding interaction. Further, the fragments have to have the electronic valence configuration, necessary for formation of the bond. The (CO)&r fragment and most of the ER2 ligands that we will encounter, namely CF2, CC12, CMe(OMe),SiH2, SiFz, SiC12, SiMe2,and SiMe(OMe),possess a singlet ground state and thus do not require an (34) Krijn, J.; Baerends, E. J. Fitfunctions in the HFS Method, Internal Report; Vrije Universiteit: Amsterdam, 1984. (35) Versluis, L.;Ziegler, T. J. Chem. Phys. 1988, 88, 322. (36) Mbrquez, A.; Fernandez Sanz, J. J.Am. Chem. SOC.1992,114, 10019.
electronic preparation. One of the exceptions is the methylene ligand, which needs to be electronically promoted:
-
CHz(3Bl)
CH2(lA1)
(4)
The triplet-singlet splitting of CH2 which is associated with reaction 4 amounts to 38 kJ/m01.~~ Considering the (Me)& ligand, we found that on the LDA/NL level of theory the triplet state is about 10 kJ/mol more stable than the singlet state. However, for this molecule one might further consider the following isomerization reaction: H3C-C-CH,
-,HzC=CH(CH3)
(5)
A 1,2-H-shift transforms dimethylmethylene into methylethylene, a process which we have calculated to be favorable by 289 kJ/mol. This value is comparable with the intrinsic ?G bond strength in ethylene, Dn(C2H4) = 312 kJ/m01.~*It will depend on the definition of the reference state whether the Cr-C bond in (CO)&r= CMez is to be considered as stable or not. Martinho Sim6es and B e a ~ c h a m pdifferentiate ~~ between metal-ligand bond dissociation enthalpies and metal-ligand bond enthalpy terms. For the latter, the molecule dissociates into the starred fragments which have the same configuration as in the initial complex and which are not allowed to relax. However, only dissociation enthalpies are experimentally accessible, whereas the determination of the enthalpy terms requires the aid of calculations. They further argue that bond strengths are better described by bond enthalpy terms rather than by bond dissociation enthalpies. The concept of bond enthalpy terms is closely related to our bond snapping energy BEsnap. The remaining differences due to zero point energy as well as to thermal corrections can be expected to be rather small. In the following, we will mainly use bond-snapping energies BEsnap as a measure for M=E bond strengths, and we only occasionally refer to the corrected bond energies
BE. Complexes (CO)sCr=CMe(OMe) and (C0)5Cr= SiMe(0Me). We have chosen (CO)&r=CMe(OMe) as the simplest model of a "real life" Fischer carbene complex. This compound was reported by Fischer and MaasbOl4O in 1967. We performed a full geometry optimization on the LDA as well as NL-SCF level of theory. Most crystal structures of Fischer carbenes show the ligand t o adapt a staggered conformation S with respect to the transition metal.41 We thus provided a starting geometry with a rotation angle CP = 45", as defined as follows:
+Lo
Me-0
Regarding the local orientation of the Me-C-OMe ligand, we have chosen a trans rather than a cis arrangement. This is in accord with the known structures of the closely related molecules (CO)&r=CMe(OEt)42aand (CO)&r=CPh(OMe).42b The optimized structures are displayed in la, and all bond distances
Organometallics, Vol.14,No.1, 1995 227
Fischer Type Chromium Carbenes and Silylenes
Table 1. Bond Analysis" for the Eclipsed E and the Staggered S Conformation of (CO)d3=CMe(OMe) LDtVNL mint
E-(CO)sCR=CMe(OMe) -391 S-(CO)Kr=CMe(OMe) -393
91 95
NL- SCF
BEsnap
mint
300 298
-366 -361
54 52
BE,, 312 309
Energies in kJ/mol.
W @ = 48'149"
LDA (NL-SCF)
la are reported in pm. We note as an important difference between the LDA and NL-SCF geometries that the selfconsistent inclusion of nonlocal corrections results in an elongation of the M-C bonds by 5-7 pm. The Cr=C bond lengths amounts to 194 pm or 199 pm, respectively. This value is significantly larger than for the = ~188 Cr=C bond in (CO)&r=CH2, ~ L D A ( C = C H ) pm25 = 193 pm.25b The experimentally and ~NL-scF(C=CHP) known bond lengths for Fischer carbenes range from 200 to 216 pm.41 The CFC distances in (CO)&r=CMe(OEt) and (CO)sCr=CPh(OMe)amount to 205.3and 204 pm, re~pectively.~~ Thus, our ~ N L - S C F value for the CFC bond is in good agreement with the experimental values. We can expect that further increase of the steric bulk of the carbene substituents will lead to an additional increase in the Cr=C separation. The C-0 distances as well as the rotational angle CP are less sensitive to the level of theory; both geometry optimization resulted in structures with CP slightly larger than 45". The question of the rotational barrier around the Cr=C bond has generated some interest in the literature.15J6J8 In fact, the first ab initio study on (C0)sNi=CH2 by Spangler and co-workers15focused on the value for the six-fold rotational barrier rather than on the Ni=C bond strength. We therefore also optimized the geometry for (CO)sCr=CMe(OMe) in an eclipsed conformation E,CP = 0". The results are displayed in lb. We note only minor structural changes of the Cr-C bond lengths. The carbene ligand now forms a slightly shorter bond with the transition metal fragment.
We next turn to the bond analysis of both the staggered S as well as the eclipsed E conformation of (CO)&r=CMe(OMe). The results are presented in Table 1. We find that both the LDA as well as the NLSCF geometries to be more stable in the eclipsed conformation by 2-3 kJ/mol. To explain these trends, we first consider the bond analysis for the NL-SCF geometries. As to be expected, the steric repulsion is reduced when the carbene ligand is rotated into the staggered conformation. We also observe a small decrease in orbital interaction, caused by the slight elongation of the Cr=C bond. The balance of these two effects, A(AEo) and A(hEint), determines whether the E or the S conformation will be more stable. If the steric bulk of the carbene ligand is increased, we can expect the value for A(AEo) to increase as well, which means that the steric repulsion is more effectively decreased under rotation. On the other side, A(AEint) will be less sensitive t o ligand variation on the carbenoid ligand, since it is mainly determined by the overlap of the empty 2p orbital a t carbon with the metal 3d orbitals. We therefore can expect that with increasing steric bulk of the carbene ligand a staggered arrangement will become more stable. In the case of LDA geometries, the situation is different. We now observe an increase in both steric repulsion as well as orbital interaction. Under rotation the methyl group of the methoxy ligand comes closer to one of the equatorial CO groups. This increases the steric repulsion, but also allows for some intramolecular H-0 stabilizing interaction. In case of the shorter Cr=C separation for the LDA geometries, this effect dominates the trends in A(AEo) and A(AEint). However, the net result again is a more stable eclipsed arrangement E. At this point, we like t o draw two main conclusions from our analysis presented so far. First, we observe that the value of the rotational barrier is very small, and when there is comparison with experimental solid state structures, intermolecular interactions, e.g. packing effects, might well compete with intramolecular interactions, as discussed above, in determining the most stable geometric arrangement of the molecule. The value of the rotational barrier amounts to less than 1% of the bond-snapping energy and can safely be neglected, when analyzing the Cr=C bond strength. Second, we point out that the LDA/NL value for BE,,, is already a reasonable approximation to the NL-SCF value. For (37) McKellar, A. R. W.; Bunker, P. R.; Sears, T. J.; Evenson, K. M.; Saykally, R. J.; Langhoff, S. R. J. Chem. Phys. 1983,79,5251. (38)Jacobsen, H.; Ziegler, T. J. Am. Chem. SOC.1994,116,3667. (39)Martinho SimBes, J. A.; Beauchamp, J. L. Chem. Rev. 1990, 90 629 - - I
LDA (NL-SCF)
@ = 0"
lb
(40)Fischer, E. 0.; Maasbol, A. Chem. Ber. 1967,100,2445. (41)Schubert, U.Coord. Chem. Rev. 1984,55,261. (42)(a) Kriiger, C.; Goddard, R.; Claus, K. H. 2.Nuturforsch. 1983, 38b, 1431. (b) Mills, 0.S.; Redhouse, A. D. J. Chem. SOC.A 1968, 642.
Jacobsen and Ziegler
228 Organometallics, Vol. 14, No. 1, 1995
the (CO)sCr=CMe(OMe) system, BE,,,,(LDA) value falls short by 12-13 kJ/mol, compared to BE,nap(NLSCF). The optimized LDA structure for (CO)&r=SiMe(OMe)is displayed in 2. We only observe minor changes
8
Table 2. Selected Optimized Bond Distances and LDA/NL Bond Energies for Various (CO)&r=ER* Systems ER2 for (C0)5CrER2
188 190 193 193 220 218 220 22 1
9
115
a
LDA
Q, =O"
2 in the chromium pentacarbonyl fragment, compared to (CO)sCr=CMe(OMe). It is of interest that for (CO)sCr=SiMe(OMe) the Cr-Si bond length is 2 pm shorter as compared to (CO)&r--SiH2, & r S i = 200 pm. We will return to this point in the next section. The bond-snapping energy amounts to BEsnap = 260 kJ/mol, with contributions from steric respulsion and orbital interaction as AEo = 165 kJ/mol and M i n t = -425 kJ/ mol. Thus, the Cr=Si bond strength is only 40 kJ/mol weaker than the Cr=C analogue. Nakatsuji and coworkers16bhave reported the Cr-Si bond to be 63 kJ/ mol weaker than the corresponding Cr=C link. It is, however, not obvious why the reaction and coordination chemistry of silylene complexes is so significantly different from that of their carbene analogues. To get a better understanding of this problem, we will now study a variety of carbene and silylene ligands, and analyze the bonding according to u and n bond contributions. Substituent Variation in (co)~Cr=CRgand ( C O h Cr=SiRg Complexes. We optimized the geometries of the Fischer type complexes (C0)5Cr=ER2, with ER2 = CH2, CF2, CC12, CMe2, SiH2, SiF2, and Sic12 and SiMe2. The substituents methyl and chloride were chosen since corresponding type I1 silylene complexes with these ligands are experimentally known and structural characterized.12b We included the fluoride-substituted ligands since a variety of halocarbene complexes including L,M=CF2 and L,M=CCl2 have been studied.43 The calculations were performed at the LDA level of theory with the ER2 ligand in an eclipsed conformation. Selected structural parameters and values for hEo, mint, and BEsnapare presented in Table 2. Considering the Cr=E distances, we observe that the distances of the calculated carbene ligands vary over a range of 7 pm, whereas the largest difference of two silylene-metal bonds only amounts to 3 pm. We also note that all silylene ligands provide similar contributions in AEo, whereas for the carbene ligands a larger variation in AEo is found. We recall that the steric repulsion term is mainly determined by the influence of the Pauli repulsion term. Compared to the 1s core of carbon, silicon possesses an extended electronic core ~~
~~
~
(43) Brothers, P. J.; Roper, W. R. Chem. Rev. 1988,88, 1293.
Bond distances" d(Cr=E) d(Cr-CO), 189 187 188 188 186 184 185 185
AP 113 135 141 88 157 166 172 162
bond energiesb mint BE,, -469 -348 -372 -390 -414 -353 -360 -439
356 213 231 302 257 187 188 277
Distances in pm. Energies in kJ/mol.
including p orbitals. Thus, for the SiRz ligands the silicon center causes the major contribution to the term hEo. Small changes in the electronic effects then lead to small variations in the calculated Cr-Si bond length. For the carbene ligands, on the other hand, the substituents provide the major contribution to hEo. This is one of the reason for the variation in the calculated Cr=C bond lengths. The especially short Cr=CH2 bond is also a result of electronic interaction, due to the fact that for the excited singlet methylene the acceptor orbital at carbon provides a better energetic match with the 3d donor orbitals of the metal fragment.25a At first, it might be puzzling that the larger CMe2 ligand is associated with the smaller value for hEo (Table 2). However, one has to keep in mind that hEo also depends on the Cr-E separation. This structural parameter in turn is determined by the balance between hEoand AEht. The CH2 ligand undergoes a very strong electronic interaction with the Cr(C0)5fragment. Thus, shortening of the Cr=C bond effectively strengthens the Cr=C bond, with the enhanced electronic interaction overcoming the increase in hEo. The C1-substituted ER2 ligands are associated with the largest value of hEo, since only C1, compared to the other systems, has an extended eight electron core. Another interesting structural parameter is the distance between the axial CO group and the chromium center. The LDA value for the Cr-C bond length in chromium hexacarbonyl amounts to 187 pm. Thus, for the carbene complexes one finds that the bond distance ~ ( c ~ - ciso slightly )~ elongated, whereas for the silylene complexes this bond is slightly shortened compared t o Cr(CO16. This trans effect might be the first indicator for differences in bonding between CR2 and SiR2 ligands, suggesting that silylenes form weaker n bonds. All ligands under investigation seem to form stable bonds with the chromium pentacarbonyl fragment. The bond strength of CMe2 is comparable to that of the CMe(OMe) ligand. Compared to the alkyl-substituted systems, the halo ligands form significantly weaker bonds. To our knowledge, the experimentally available metal carbene bond dissociation enthalpies are limited to three Mn(C0)5CXY molecules: D[Mn(C0)5+=CH21 = 401 31 kJ/mol, D[Mn(C0)5+=CHF] = 356 & 25 kJ/mol, and D[Mn(C0)5+=CF2] = 332 & 12 kJ/mol. All of these results were obtained by photoionization mass spect r ~ m e t r y . The ~ ~ positive charge of the manganese complexes will have a significant influence on the Mn=CR2 bond dissociation enthalpies, and we cannot compare our calculated bond energies for the neutral
*
(44) Stevens, A. E. Ph.D. Thesis, California Institute of Technology,
1981.
Fischer Type Chromium Carbenes and Silylenes 250
Organometallics, Vol. 14, No. 1, 1995 229
1
200 150
E
8-
100
50
n
Figure 1. Reduced u bond strengths and n bond strengths for various (CO)&r=ER2 complexes. chromium carbenes with the experimental results. We might, however, compare the effect of fluorination, that is A[D(Mn=CH2),D(Mn=CF2)1,and A[BE(Cr=CH2), BE (Cr=CF2)1. Since we compare with bond dissociation enthalpies rather than with bond enthalpy terms, we have to correct our BEsnapvalues to bond energies BE, according to eq 3. The geometric preparation energies for (CO)&r=CH2 and (C0)5Cr=CF2 amount to 10 and 4 kJ/mol, respectively. We further correct BE(Cr-CH2) with the experimental value for the triplet-singlet splitting of methylene. Thus, we obtain A[BE(Cr=CHz), BE(Cr=CF2)1= 99 kJ/mol. This result is in good agreement with the experimental value of A[D(Mn=CH2), D(Mn=CF2)1 = 69 f 43 kJ/mol. Assignment of u and rc Bond Strengths. In order t o obtain values for the u as well as the n component of the Cr-E double bond, we make use of the fact that we can decompose our electronic interaction energy due to different symmetry contributions:
Here, r represents the different irreducible representations of the point group of the molecule. We can associate the different terms from the right side of eq 6 with the intrinsic u and n bond strengths, Do,int and Dn,int,respectively. Orbitals that have the plane of the ER2 ligands as a nodal plane contribute to Dn,i,,t. Similarly, orbitals that lie in the plane of the ER2 ligand donate t o Do,int. Thus, we can break down AEint into only two terms as
We can further combine Dg,int and hEoto the so-called reduced intrinsic u bond strength25aDo,int:
We now are able to analyze our bond-snapping energy BEsnapor the Cr=E bond strength according to u and n contributions as ’‘snap
= D‘o,int + Dn,int
(9)
The results of our analysis are presented in Figure 1. We will first discuss the trends observed for the carbene ligands. CH2 posses by far the strongest n bond with Dn,int= 198 kJ/mol. The other carbene ligands possess n bond strengths around 120 kJ/mol. This can be
understood by noting that the F, C1, and Me substituents can donate electrons into the empty p orbital at the carbon center. This substituent donation competes with the back-donation for the metal fragment, with the consequence that substituted carbene ligands form weaker n bonds than methylene itself. Further, we see that, with the exception of the CMe2 ligand, all investigated carbene ligands show a higher n bond than u bond strength. This is in accord with the common notion that electron-withdrawing substituents enhance the ability of n acceptance, whereas alkyl substituents increase the capacity for u donation. If we now turn to the silylene systems, we find that all systems posses only a weak n bond around 70 kJ/ mol. The u component contributes the major part to the bonding interaction. In general, we can state that the different bond strengths of the ER2 ligands are determined by the variation in D’o,int rather than in Dn,int. For both the series of ligands we obtain a similar ranking of Dlo,int,EMe2 > EHz > ECl2 % EF2. The fact that silylene systems posses only a weak n bond holds an explanation why these species follow a trend to form Lewis base adducts and t o undergo type I1 complexation. In the next section, we will analyze a Lewis base adduct in more detail. Type I1 Complex (CO)&r=SiCl&PHs. We have chosen (CO)&r=SiC12*OPH3as a model compound of a type I1 silylene complex. The solid state structure of the related compound (CO)&r=SiC12*HMPA(HMPA = hexamethylphosphoric triamide) has been reported by Zybill and co-workers.12b The optimized structural parameters together with experimental values are presented in 3. The calculated geometric arrangements
LDA (Exp.)
3 of the (CO)&r fragments and the Sic12 ligand are in good agreement with the crystal structure. However, two structural parameters do not provide a good agreement between experiment and theory. First, the calculated Cr-Si bond length is 9 pm shorter than in the crystal structure. If we employ a nonlocal correction of 5 pm, as established for the (CO)&r=CMe(OMe) molecule, our Cr-Si distance comes close to the experimental result. However, comparing the Cr-Si distance in the corresponding type I and type I1 complexes, we find a bond elongation under Lewis base addition of 5 pm. This is only half the value found for a pair of analogous type Vtype I1 stannylene complexes.Ga The other structural parameter that is in disagreement with the experiment is the Si-0 separation. Our result is 13 pm too long compared to experiment.
Jacobsen and Ziegler
230 Organometallics, Vol. 14, No. 1, 1995 E
I
(CO),Cr=SiCl
O4H,
1
Figure 2. Bond analysis for (CO)&r-SiClz and its Lewis base adduct (CO)&r=SiC12*OPH3. An explanation for the poor agreement in the Si-0 bond lengths might be found in the difference of Lewis base strength of our model compound OPH3 and the experimentally employed OP(N(Me)z)3. Since the Lewis adduct bond is relatively weak, differences in Lewis base strength might well have major consequences on the Si-0 distance. Thus, if OP(N(Me)z)3is a stronger donor than OPH3, we can expect to find a shorter Si-0 for the triamide. Consequently, this should lead to longer Cr-Si bond, since the n bond character of the metalsilicon bond is decreased. This would also provide a reason why our theoretically obtained bond elongation is smaller than the experimental value. To support out argument, we analyze the energies of the 0-based donor orbitals for OPH3 and OP(NH2h as a measure for the Lewis base strength. We find that the donating orbital for the triamide is 0.16 eV higher in energy than for OPH3. The OP(NHz)3 donor orbital provides a better energetic match for the acceptor orbital of the (CO)&r=SiC12 fragment. Thus, a phosphoric triamide can indeed be considered as a somewhat stronger Lewis base than phosphine oxide. In Figure 2, we present a bond analysis for the (CO)&r=SiClyOPH3 molecule. We begin with the interaction of (C0)sCr with SiClz t o form the type I complex (CO)&r=SiC12. This process is favored by 188 kJ/mol, with a n contribution of Dn,int = 68 kJ/mol. In the next step we provide the system for Lewis base addition. The bond elongation as well as the pyramidalization of the silicon center requires on energy of 30 kJ/mol. The addition of the Lewis base finally stabilizes the system by 94 kJ/mol. We see that the final step not only overcomes the 30 kJ/mol of preparation energy but could also compensate for the complete loss of the Cr=Si n bond. Concluding Remarks The main difference between carbene and silylene ligands has been established in the very weak n bond strength of the latter. Lewis donation can favorably
compete with n bonding and as a consequence silylene complexes show a high tendency for type I1 complexation. Bulky substituents a t the silylene ligands are required to protect the reactive Cr=Si n bond. However, a compromise has to be found between steric protection of the silicon center and steric repulsion with the metal fragment. The Sila-Wittig reaction of (CO)&r=SiClz with OC(0Me)z has been proposed to proceed as a two step reaction. Experimental evidence clearly excludes a reaction mechanism in the sense of a concerted [2 21 cycloaddition.12b Our calculation support this notion and suggest that the polar addition of OC(0Me)Zto the silylene complex initializes the formation of the product (CO)&r=C(OMe)2. Although the n bond in silylene type I complexes is weak, our calculations indicate that those compounds form reasonably strong M-Si bonds and might be isolated without a stabilizing base. This has recently been demonstrated with the synthesis of [Cp*(PMe&Ru=SiMezl+, for which the first crystal structure of a base-free silylene complex without n-donor stabilization has been obtained.13b The main features of this compound are a planar dimethylsilene ligand as well as the shortest Si-Ru bond reported so far. The results of Fenske-Hall calculations13bon the model compound [Cp(PMe3)2Ru=SiH21f indicate a loss of electron density at the silylene ligand under coordination to the metal. This again is consistent with the fact that SiRz ligands form weak n bonds and have mainly to be considered as o donors.
+
Acknowledgment. This investigation was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and by the donors of the Petroleum Research Fund, administered by the American Chemical Society (ACS-PRFNo. 27023-AC3). Access to the IBM 6000 RISC computing facilities at the University of Calgary is acknowleged. OM940546S