Continuum solvation models are powerful tools for the modeling of condensed phases, and hence for approaching a realistic representation of complex chemical systems in a manner that is tractable for accurate quantum-chemical simulations. The SM12 solvation model, originally proposed by Marenich et al.\(^{1}\), is a Generalized-Born-Approximation (GBA) based solvation model. SM12 (and other GBA approaches) incorporate electrostatic effects as polarization energy due to partial charges while addressing non-electrostatic effects of the first-solvation shell in the form of atomic surface tension terms.\(^{2}\) Unique to SM12 among GBA approaches is the use of partial atomic charges that are basis-set independent and are derived from the accurate charge density. SM12 works in combination with the Hirshfeld-derived Charge Model 5 (CM5), and the ESP-derived Merk-Kollmann (MK) or ChElPG charges affording solvation energies for any element in the periodic table. The charge model 5 (CM5)\(^{3}\) variant was previously implemented for SM12 into the ADF code.\(^{4}\) However, the CM5 charges are not well-behaved for gradient evaluations, while the analytical gradients of the ESP-derived charges on a Cartesian grid are too expensive.
In this contribution, we present a revised SM12 implementation that uses Lebedeev grids to compute ESP-derived charges.\(^{5}\) These are atom-centered CHELPG (a-CHELPG) charges, that are computationally inexpensive, and well-behaved for energies and gradients. As a result, geometry optimizations are now feasible, and agree with finite-difference numerical gradients for molecules encompassing main-group, transition-metal, and f-block elements. In gas phase, the a-CHELPG charges yield dipole moments that are similar in scale to those from Cartesian grids. An assessment of the different charge options to compute solution-phase vertical excitation energies with SM12 will be presented.