interaction with single and double substitutions) levels are similar to those found in 1. ['] In [Ni(SiH,),] (4), a minimum with D,, structure is found. The binding energies, again relative to the ' S state of Ni and to the geometry-optimized Si,H ligands (C, symmetry, envelope conformation), are only slightly smaller than those in 2. Interestingly the Ni-Si and Si-Si distances are calculated to 220 and 259 pm, respectively. The Si-H bond lengths (147 pm) in 2 and 4 are identical. Several attempts were made to locate a D,, structure as a minimum on the potential energy surface for 2. In each instance the optimization collapsed back to the D,, minimum. Note that the binding energies for 2 and 4 are greatly underestimated at the H F Ievel; a similar situation exists for l.I1] Dynamical electron correlation apparently plays a critical role in setting the magnitudes.The situation for the congener [Ni(CH,),] is quite different. The geometry-optimized D,, structure, 5 (Scheme 3), possesses five imaginary frequencies. Furthermore, this structure lies 544kJmol-' higher in energy than the trisethylene complex (6)[15] with D,, symmetry at the HF level (BS I). The problem here is geometrical. The Ni-Cand C-C distances in 5 are 185 pm. This is somewhat short for a Ni -C bond, and is much too long for an effective C-C bond; note in Figure 1 that six electrons remain in three intra-ring bonding MOs. Although geometric factors dominate, there is also an electronic component at work. That there is a striking difference in the balance between CJ and x bonding for 5 and 6 on the one hand and the Si analogues on the other is not surprising-recall the similar difference between cyclo-0, and 3 O,, and cyclo-S, and 3 S, .
