While use of curvilinear coordinates such as bond lengths and bond angles is common in accurate spectroscopic and/or scattering calculations for triatomic and other small molecules, their use for large molecules is uncommon and restricted. For large molecules, normal-mode analysis is feasible but gives sensible results only if the dynamical or spectroscopic process being considered involves changes in angular coordinates, including ring deformations, which are so small that the motion can be approximated by its tangential component. We describe an approximate method by which curvilinear normal-mode-projected displacements and hence Franck–Condon factors, reorganization energies, and vibronic coupling constants, as well as Duschinsky (Dushinsky, Duschinskii) rotation matrices, can be evaluated for large systems. Three illustrative examples are provided: (i) to understand the nature of the first excited state of water, illustrating properties of large-amplitude bending motions; (ii) to understand the nature of the “boat” relaxation of the first excited state of pyridine, illustrating properties of large-amplitude torsional motions; and (iii) to understand the coupling of vibrational modes to the oxidation of bacteriochlorophyll-a, a paradigm with many applications to both chemical and biological electron transfer, illustrating properties of macrocyclic deformations. The method is interfaced to a wide variety of computational chemistry computer programs.
Density-functional theory (DFT) is widely used for studying large systems such as metals, semiconductors, and large molecules, with time-dependent density-functional theory becoming a very powerful tool for investigating molecular excited states. As part of a systematic study of both the intrinsic weaknesses of DFT and the weaknesses of present implementations, we consider its application to the one and two-dimensional conjugated π systems: polyacetylene fragments and oligoporphyrins, respectively. Very poor results are obtained for the calculated spectra, and polyacetylene is predicted by all functionals considered, including gradient-corrected functionals, to have a triplet ground state. The cause of this is linked to known problems of existing density functionals concerning nonlocality and asymptotic behavior which result in the highest-occupied molecular-orbital being too high in energy so that semiconductors and low-band-gap insulators are predicted to have metal-like properties. The failure of modern density functionals to predict qualitatively realistic molecular hyperpolarizabilities for extended systems is closely related.
1 April 1999 in Journal of the American Chemical Society
Various Green’s-function-based formalisms which express the current I as a function of applied voltage V for an electrode–molecule–electrode assembly are compared and contrasted. The analytical solution for conduction through a Hückel (tight binding) chain molecule is examined and only one of these formalisms is shown to predict the known conductivity of a one-dimensional metallic wire. Also, from this solution we extract the counter-intuitive result that the imaginary component of the self-energy produces a shift in the voltage at which molecular resonances occur, and complete analytical descriptions are provided of the conductivity through one-atom and two-atom bridges. A method is presented by which a priori calculations could be performed, and this is examined using extended-Hückel calculations for two gold electrodes spanned by the dithioquinone dianion. A key feature of this is the use of known bulk-electrode properties to model the electrode surface rather than the variety of more approximate schemes which are in current use. These other schemes are shown to be qualitatively realistic but not sufficiently reliable for use in quantitative calculations. We show that in such calculations it is very important to obtain accurate estimates of both the molecule–electrode coupling strength and the location of the electrode’s Fermi energies with respect to the molecular state energies.