InfPolyn, a Nonparametric Bayesian Characterization for Composition-Dependent Interdiffusion Coefficients.

Area of Science:

• Materials Science
• Physical Chemistry
• Computational Modeling

Background:

• Interdiffusion coefficients are crucial for understanding material behavior but are difficult to determine.
• Challenges include underdetermined equations, limited data, and unknown coefficient forms.

Purpose of the Study:

• To develop a novel statistical framework, InfPolyn (Infinite Polynomial), for characterizing component-dependent interdiffusion coefficients.
• To improve the accuracy and flexibility of interdiffusion coefficient estimation in multi-component systems.

Main Methods:

• Developed InfPolyn, a generalized polynomial fitting method with enhanced model capacity.
• Integrated InfPolyn with the numerical inversion-based Boltzmann-Matano method for coefficient estimation.
• Validated the framework on ternary and quaternary systems with various predefined coefficient forms.

Main Results:

• InfPolyn demonstrated superior performance compared to State-Of-The-Art (SOTA) Boltzmann-Matano methods, achieving up to 10x greater accuracy.
• The model exhibited consistent and stable performance across different system complexities.
• InfPolyn required a small number of samples for reliable estimations.

Conclusions:

• InfPolyn offers a robust and accurate solution for determining interdiffusion coefficients.
• The framework enhances the understanding of diffusion processes in multi-component materials.
• This approach provides a flexible and efficient tool for materials science research.