Reliable crystal structure predictions from first principles.

Area of Science:

• Solid-state chemistry
• Computational chemistry
• Materials science

Background:

• Crystal structure prediction (CSP) is crucial for understanding material properties.
• Current methods often rely on experimental data or computationally expensive empirical force fields (FFs).
• Developing accurate and cost-effective CSP protocols is an ongoing challenge.

Purpose of the Study:

• To develop an inexpensive and reliable protocol for molecular crystal structure prediction.
• To create an ab initio-based force field (aiFF) for crystal structure optimization without experimental input.
• To demonstrate the protocol's robustness across diverse molecular systems.

Main Methods:

• Generating a two-body, rigid-monomer ab initio-based force field (aiFF) from quantum mechanical calculations of molecular dimers.
• Utilizing crystal packing procedures to generate tens of thousands of plausible polymorphs.
• Optimizing polymorphs using the developed aiFF, comparable in cost to empirical FFs.
• Refining rankings with periodic density-functional theory (pDFT+D) calculations.

Main Results:

• The protocol successfully identified the experimental crystal structure within the top 20 predicted polymorphs for 15 diverse molecules.
• Post-ranking refinement using pDFT+D placed the experimental crystal structure at rank one for all tested systems.
• Alternative refinement using improved aiFFs also resulted in rank one predictions.
• The method proved robust and reliable for molecular crystal structure prediction.

Conclusions:

• The developed CSP protocol offers an inexpensive and reliable approach to predicting molecular crystal structures.
• The ab initio-based force field (aiFF) approach is a viable and accurate alternative to empirical force fields (FFs).
• This protocol has the potential to significantly advance crystal structure prediction research and applications.