Salt Versus Cocrystal: A Pragmatic First-Principles Computational Approach

Abstract

Distinguishing the salt versus cocrystal character in acid-base multi-component crystals is pharmaceutically important and can be challenging for both theory and experiment. On the theory side, commonly-used generalized gradient approximation (GGA) functionals in periodic density functional theory (DFT) artificially stabilize salt forms, sometimes even erroneously transferring the acidic proton from the acid to the base. The inclusion of exact exchange in hybrid density functionals reduces the problem. However, proton-transfer energies can be very sensitive to the specific fraction of exact exchange included and the optimal fraction is not always readily apparent. Here, a pragmatic approach which combines periodic GGA or hybrid DFT with local corrections computed on the acid-base pair at a higher level of theory is shown to predict the correct salt, cocrystal, or intermediate behavior for ten different acid-base multi-component crystals. This approach greatly reduces the sensitivity of the proton transfer energies to the underlying periodic DFT functional, and the correction increases the overall computational costs only marginally. On the other hand, it has limitations when applied to crystals where the acid-base pair is not uniquely defined.