Authors:

Liam Bielat

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Conductivity Data This project explores whether machine learning can be used to extract physically meaningful phonon scattering information from thermal conductivity data. In crystalline materials, heat is carried by lattice vibrations, and the way these vibrations scatter strongly influences thermal transport. Traditional analytical approaches can describe this behavior but often rely on difficult parameter fitting. Building on these ideas, this work uses synthetic thermal conductivity data generated from established transport models to train neural networks to predict underlying scattering parameters directly from thermal conductivity data. A key focus of the project is understanding how the temperature range of the input data affects the accuracy of the extracted coefficients, particularly the important role of low-temperature data in providing useful scattering information. By comparing predictions on synthetic and real material data, this project evaluates both the promise and the limitations of machine learning for interpreting thermal transport. More broadly, it shows how data-driven methods can complement physics-based modeling in the study of solid-state material thermal conductivity.

Source:

University of Oregon / 2026

Topics:

No topics listed

Co-authors:

Liam Bielat