Structural analysis and potential extraction from diffraction data of disordered systems by least-biased feature...

Area of Science:

• Computational physics and chemistry
• Materials science
• Statistical mechanics

Background:

• Determining atomic-scale structure and potentials from experimental data is crucial for understanding disordered systems like liquids and glasses.
• Existing methods may lack flexibility or struggle to integrate experimental constraints effectively.

Purpose of the Study:

• To propose a novel iterative approach for refining interaction potentials using experimental data.
• To develop a method that minimizes bias while ensuring consistency with experimental observations.

Main Methods:

• An iterative potential refinement method is introduced: u = ψ + θ·f, where u is the refined potential, ψ is a prior potential, θ are adjustable parameters, and f are features of atomic coordinates.
• Parameter updates rely on the difference between ensemble means of f from simulations and experimental data.
• The method minimizes Kullback-Leibler divergence and ensures consistency with experimental ensemble means.

Main Results:

• The approach successfully refines potentials and structures for Lennard-Jones liquid and SiO2 liquid/glass systems.
• Demonstrated effectiveness using both simulated and real experimental data.
• The method provides a least-biased potential consistent with experimental data and approximately minimizes the squared difference to the true potential.

Conclusions:

• The proposed method offers a flexible and robust way to extract or refine interaction potentials from experimental data for disordered materials.
• It enables the incorporation of prior knowledge and ensures potentials are consistent with empirical observations.
• This technique advances the study of liquids and glasses by improving potential accuracy and structure determination.