Kohn-Sham theory (KST) is the "workhorse" of numerical quantum chemistry. This is particularly true for first-principles calculations of ground- and excited-state properties for larger systems, including electronic spectra, electronic dynamic and static linear and higher order response properties (including nonlinear optical (NLO) properties), conformational or dynamic averaging of spectra and response properties, or properties that are affected by the coupling of electron and nuclear motion. This Account explores the sometimes dramatic impact of the delocalization error (DE) and possible benefits from the use of long-range corrections (LC) and "tuning" of functionals in KST calculations of molecular ground-state and response properties. Tuning refers to a nonempirical molecule-specific determination of adjustable parameters in functionals to satisfy known exact conditions, for instance, that the energy of the highest occupied molecular orbital (HOMO) should be equal to the negative vertical ionization potential (IP) or that the energy as a function of fractional electron numbers should afford straight-line segments. The presentation is given from the viewpoint of a chemist interested in computations of a variety of molecular optical and spectroscopic properties and of a theoretician developing methods for computing such properties with KST. In recent years, the use of LC functionals, functional tuning, and quantifying the DE explicitly have provided valuable insight regarding the performance of KST for molecular properties. We discuss a number of different molecular properties, with examples from recent studies from our laboratory and related literature. The selected properties probe different aspects of molecular electronic structure. Electric field gradients and hyperfine coupling constants can be exquisitely sensitive to the DE because it affects the ground-state electron density and spin density distributions. For π-conjugated molecules, it is shown how the DE manifests itself either in too strong or too weak delocalization of localized molecular orbitals (LMOs). Optical rotation is an electric-magnetic linear response property that is calculated in a similar fashion as the electric polarizability, but it is more sensitive to approximations and can benefit greatly from tuning and small DE. Hyperpolarizabilities of π-conjugated "push-pull" systems are examples of NLO properties that can be greatly improved by tuning of range-separated exchange (RSE) functionals, in part due to improved charge-transfer excitation energies. On-going work on band gap predictions is also mentioned. The findings may provide clues for future improvements of KST because different molecular properties exhibit varying sensitivity to approximations in the electronic structure model. The utility of analyzing molecular properties and the impact of the DE in terms of LMOs, representing "chemist's orbitals" such as individual lone pairs and bonds, is highlighted.

In the framework of determining system-specific long-range corrected density functionals, the question is addressed whether such functionals, tuned to satisfy the condition -ε(HOMO) = IP or other energetic criteria, provide accurate electron densities. A nonempirical physically motivated two-dimensional tuning of range-separated hybrid functionals is proposed and applied to the particularly challenging case of a molecular property that depends directly on the ground-state density: the copper electric field gradient (EFG) in CuCl. From a continuous range of functional parametrizations that closely satisfy -ε(HOMO) = IP and the correct asymptotic behavior of the potential, the one that best fulfills the straight-line behavior of E(N), the energy as a function of a fractional electron number N, was found to provide the most accurate electron density as evidenced by calculated EFGs. The functional also performs well for related Cu systems.

Time-dependent density functional theory (TDDFT) computations are performed for 42 organic molecules and three transition metal complexes, with experimental molar optical rotations ranging from 2 to 2 × 10(4) deg cm(2) dmol(-1). The performances of the global hybrid functionals B3LYP, PBE0, and BHLYP, and of the range-separated functionals CAM-B3LYP and LC-PBE0 (the latter being fully long-range corrected), are investigated. The performance of different basis sets is studied. When compared to liquid-phase experimental data, the range-separated functionals do, on average, not perform better than B3LYP and PBE0. Median relative deviations between calculations and experiment range from 25 to 29%. A basis set recently proposed for optical rotation calculations (LPol-ds) on average does not give improved results compared to aug-cc-pVDZ in TDDFT calculations with B3LYP. Individual cases are discussed in some detail, among them norbornenone for which the LC-PBE0 functional produced an optical rotation that is close to available data from coupled-cluster calculations, but significantly smaller in magnitude than the liquid-phase experimental value. Range-separated functionals and BHLYP perform well for helicenes and helicene derivatives. Metal complexes pose a challenge to first-principles calculations of optical rotation.