Diffusion Monte Carlo Study of Para-Diiodobenzene Polymorphism Revisited

Area of Science:

• Computational Chemistry
• Materials Science
• Condensed Matter Physics

Background:

• Predicting molecular crystal polymorphism is crucial for materials design.
• Previous studies using diffusion Monte Carlo (DMC) for para-diiodobenzene (p-DIB) showed promise but lacked rigorous analysis of simulation parameters.
• Fixed-node DMC is a powerful method for electronic structure calculations but is computationally intensive.

Purpose of the Study:

• To rigorously investigate the impact of finite-size effects and nodal surface choice on predicting polymorph stability in molecular crystals using fixed-node DMC.
• To perform the largest feasible DMC calculations for p-DIB polymorphism on the K-computer.
• To understand and mitigate finite-size errors in DMC simulations for molecular crystals.

Main Methods:

• Employed fixed-node diffusion Monte Carlo (DMC) simulations.
• Utilized a density functional theory (DFT) nodal surface generated with the PBE functional.
• Performed calculations using both small (1x1x1) and large (1x3x3) simulation cells to assess finite-size effects.
• Analyzed finite-size errors using model periodic Coulomb (MPC) interactions and kinetic energy corrections (CCMH scheme).
• Investigated k-mesh convergence and employed the T-move scheme to manage computational stability.

Main Results:

• Finite-size effects significantly influence the nodal surface and polymorph stability predictions, especially in small simulation cells.
• The largest-scale DMC simulations to date (1x3x3 cell) were performed, yielding polymorph stability predictions consistent with experimental data.
• Finite-size errors in total energies were found to be larger than the energy differences between polymorphs, highlighting the importance of error cancellation.
• The T-move scheme proved essential for circumventing population explosions and time-step biases in massive DMC simulations.

Conclusions:

• Rigorous analysis of finite-size effects is essential for accurate first-principles prediction of molecular crystal polymorphism using fixed-node DMC.
• Previous predictions for p-DIB polymorphism may have been fortuitous due to unaddressed finite-size effects.
• Despite large computational costs and significant finite-size errors, accurate polymorph prediction is achievable through careful methodology and error cancellation.