Testing recent charge-on-spring type polarizable water models. I. Melting temperature and ice properties

Area of Science:

• Computational chemistry
• Materials science
• Physical chemistry

Background:

• Accurate molecular models of water are crucial for understanding its properties.
• Polarization effects are significant in water models.
• Existing models often struggle to accurately predict phase transition properties like freezing points.

Purpose of the Study:

• To evaluate the freezing point accuracy of eight molecular water models employing the charge-on-spring (COS) method.
• To compare the performance of different polarization and repulsion schemes in these models.
• To identify which model features best correlate with accurate freezing point prediction.

Main Methods:

• Utilized thermodynamic integration (TI) and the Gibbs-Helmholtz equation to calculate free energy differences between liquid water and hexagonal ice (Ih).
• Employed TIP4P and SPC/E models as reference systems for TI calculations.
• Calculated freezing points for COS/G2, COS/G3, COS/B2, SWM4-DP, SWM4-NDP, BKd1, BKd2, and BKd3 models.

Main Results:

• All tested molecular models significantly underestimated the experimental freezing point of hexagonal ice (Ih).
• Calculated freezing points ranged from below 100 K (COS/B2) to 233 K (BKd3).
• The BKd3 model, which includes a variable size to approximate the temperature-density curve, yielded the best freezing point prediction among the studied models.

Conclusions:

• Current molecular water models, even those using the COS method, require improvement for accurate freezing point prediction.
• Accurately representing the temperature-density relationship of water is a key factor in improving freezing point predictions.
• Further development of water models is needed to capture the complex behavior of water and ice phases.