Accuracy limit of non-polarizable four-point water models: TIP4P/2005 vs OPC. Should water models reproduce the experimental dielectric constant?

This article focuses on evaluating the performance limits of rigid, non-polarizable four-center (4C) water models by comparing two representative force fields: TIP4P/2005 (an Electronic Continuum Correction, ECC, model) and OPC (a non-ECC model that targets the dielectric constant). Using the comprehensive VA-Test, which assesses about 40 properties across various phases and thermodynamic states, TIP4P/2005 achieved a higher overall score (7.2) than OPC (6.3), despite OPC accurately reproducing the experimental dielectric constant of water. The study concludes that attempting to fit the dielectric constant in non-polarizable models compromises their global accuracy because the dielectric constant depends on both the potential energy surface (PES) and the dipole moment surface (DMS), whereas most other properties depend only on the PES. Additionally, the choice between ECC and non-ECC models influences ion charge scaling in electrolyte simulations to correctly reproduce Debye–Hückel limiting behavior. The authors suggest that TIP4P/2005 is near the performance ceiling for non-polarizable 4C models, and further improvements likely require polarizable or more complex models.