Several levels of density functional theory, i.e., various combinations of exchange-correlation functionals and basis sets, have been employed to compute equilibrium geometries for a diverse set of 32 metal complexes from the first transition row, for which precise gas-phase geometries are known from electron diffraction or microwave spectroscopy. Most DFT levels beyond the local density approximation can reproduce the 50 metal-ligand bond distances selected in this set with reasonable accuracy, as assessed by mean and standard deviations of optimized vs observed values. The ranking of some popular functionals, ordered according to decreasing standard deviation, is BLYP ≈ HCTH > B3LYP > BP86 > TPSS ≈ TPSSh. Together with its hybrid variant, the recently introduced meta-GGA functional TPSS performs best of all tested functionals, with mean and standard deviations of -0.5 and 1.4 pm, respectively. Even smaller errors are found for a more compact but less diverse set of transition-metal mono- and dihalides, for which experimentally derived equilibrium geometries are available.

We compare apparent global minimum structures of three widely used empirical point-charge models for water. These are SPC/E (Simple Point Charge/Extended), TIP3P (Transferable Intermolecular Potential 3 Point), and TIP4P. We obtain results for (H 2 O) N with N e 25. We compute various quantities characterizing the clusters such as the average energy per molecule, the average oxygen-to-oxygen distance, the average hydrogen bond length and number, and the quantity µ 2 /V, where V is an effective cluster volume and µ is the cluster's dipole moment. The quantity µ 2 /V is interesting because it is approximately proportional to the free energy of solvation of a cluster in water. We also compare our genetic algorithm to the basin-hopping Monte Carlo method.

Results from Car-Parrinello molecular dynamics simulations are reported for [UO2(OH2)5]2+, UO2(NO3)2(OH2)2, and UO2(NO3)2(eta2-tmma) (tmma = tetramethylmalonamide) in the gas phase and in aqueous solution. The distances between uranyl and neutral ligands such as water and tmma are decreased by up to 0.2 angstroms upon hydration, whereas those between uranyl and the nitrate ion are increased by up to 0.08 angstroms. According to pointwise thermodynamic integration involving constrained molecular dynamics simulations, solvation facilitates the transition of the chelating nitrate ligand to a eta1-bonding mode: the free energy of UO2(eta2-NO3)(eta1-NO3)(OH2)2 relative to the bis-chelating minimum drops from 3.9 kcal/mol in vacuo to 1.4 kcal/mol in water. Optimizations in a polarizable continuum (specifically, the conductor-like screening model in conjunction with the zero-order regular approximation and triple-zeta Slater basis sets) can qualitatively reproduce the geometrical changes from explicit hydration.

59Co chemical shifts were computed at the GIAO-B3LYP level for [Co(CN)6]3-, [Co(H2O)6]3+, [Co(NH3)6]3+, and [Co(CO)4]- in water. The aqueous solutions were modeled by Car-Parrinello molecular dynamics (CPMD) simulations, or by propagation on a hybrid quantum-mechanical/molecular-mechanical Born-Oppenheimer surface (QM/MM-BOMD). Mean absolute deviations from experiment obtained with these methods are on the order of 400 and 600 ppm, respectively, over a total delta(59Co) range of about 18,000 ppm. The effect of the solvent on delta(59Co) is mostly indirect, resulting primarily from substantial metal-ligand bond contractions on going from the gas phase to the bulk. The simulated solvent effects on geometries and delta(59Co) values are well reproduced by using a polarizable continuum model (PCM), based on optimization and perturbational evaluation of quantum-mechanical zero-point corrections.