How well do implicit solvation models represent intermolecular binding energies in organic-inorganic solutions?

BA Sorenson and SS Hong and HC Herbol and P Clancy, COMPUTATIONAL MATERIALS SCIENCE, 170, UNSP 109138 (2019).

DOI: 10.1016/j.commatsci.2019.109138

Computational methods, including density functional theory, are proving to be powerful approaches to tame the otherwise overwhelming selection of optimal species and processing conditions to fabricate hybrid organic-inorganic perovskite (HOIP) thin films via solution processing. In that processing, the choice of solvents is known to play a critical role in the quality of the resulting thin film, but their inclusion in a simulation dominates the overall size of the system, and hence the computational effort. This creates an incentive to understand the minimal representation of solvent medium necessary to adequately model the interactions between HOIP building blocks. These building blocks are chiefly the lead salt and cationic species dissolved in the processing solvent, whose interactions, monitored by calculations such as binding energies, govern nucleation and growth into thin films. We show that a simple implicit solvent model is surprisingly effective in terms of representing the intermolecular binding energies between lead salts and cation species in solution. Use of an implicit solvent produces binding energies that are typically within 2-3 kcal/mol of more accurate"all atom" models of solvent molecules. This is an important result since implicit solvents are roughly 100 times more efficient than explicit solvent models in terms of computational resource effort. We find that it is generally sufficient to use a Generalized Gradient Approximation for the DFT calculations, rather than more accurate, but more expensive, PW6B95 models.

Return to Publications page