Direct experimental evaluation of charge scheme performance by a molecular charge-meter.

The remarkable advances accomplished in the past two decades in theoretical and computational capabilities have made the in silico study of complex chemical systems feasible. However, this progress is in strong contrast to the lag in experimental capabilities relating to the measurement of fundamental chemical quantities within convoluted environments such as solvents or protein milieu. As a result, many works rely extensively on predictions provided by ab initio methodologies without having independent experimental support. Such a proliferation of theory and computational approaches without being substantiated by appropriate experimental data is undesirable. The feasibility of using nickel-bacteriochlorophyll as a molecular potentiometer was recently demonstrated for the systematic evaluation of fragmental charge density transfer for metal complexes in solution, thus providing an experimental assay with high accuracy and sensitivity (better than +/-0.005 e(-); Yerushalmi, R.; Baldridge, K. K.; Scherz, A. J. Am. Chem. Soc. 2003, 125, 12706-12707). Here the experimentally determined fragmental charge density transfer values measured by the molecular potentiometer for metal complexes in solvent are used to provide, for the first time, an independent and critical experimental evaluation of theoretical approaches commonly used in determining atomic charges and fragmental charge density transfer among interacting molecular systems. Importantly, these findings indicate that the natural population analysis (NPA) charge analysis is highly robust and well-suited for determining charge transfer processes involving donor-acceptor coordination interactions. The majority of computational charge schemes fail to provide an accurate chemical picture for the whole range of systems considered here. In cases where the role of electronic correlation varies significantly among chemically related structures, as with mono- and biligated complexes, the widely used electrostatic potential fit-based methods for evaluating atomic charges may prove to be problematic for predictive studies. In such cases, alternative methods that do not rely on the net dipole moment or other higher multipoles of the system for determining charges should be employed.