Is universal, simple melting point prediction possible?

Area of Science:

• Physical Chemistry
• Materials Science
• Computational Chemistry

Background:

• Melting point prediction for organic salts is crucial for synthesis.
• Experimental data reproducibility for organic salts can vary significantly (±5 to ±15 °C).

Purpose of the Study:

• To develop a universal, simple, and physically grounded prediction scheme for organic salt melting points.
• To enable accurate prediction of melting temperatures for computer-aided synthesis.

Main Methods:

• Analysis of experimental melting point data for 520 organic 1:1 salts.
• Development of two semi-empirical prediction schemes (five- and nine-parameter).
• Validation using DFT, COSMO, and COSMO-RS calculations, integrated into the IL-Prop module of COSMOtherm.

Main Results:

• Developed two prediction schemes with average errors of 33.5 °C and relative errors of 9.3% for melting points (-25 to +300 °C).
• Prediction schemes are suitable for high-throughput computational screening.
• Calculated quantities assessed via DFT and COSMO-RS.

Conclusions:

• The developed prediction schemes are valuable tools for identifying compounds with suitable melting points prior to synthesis.
• These schemes facilitate efficient computer-aided synthesis by reducing experimental efforts.