Machine Learning and First-Principle Predictions of Materials with Low Lattice Thermal Conductivity

Area of Science:

• Materials Science
• Computational Materials Science
• Solid State Physics

Background:

• Developing materials with low lattice thermal conductivity is crucial for thermal management and insulation applications.
• Efficient thermoelectric materials are needed to convert waste heat into electricity.

Purpose of the Study:

• To computationally screen for new materials with exceptionally low lattice thermal conductivity (κL).
• To identify promising candidates for both thermal insulation and thermoelectric applications.

Main Methods:

• Utilized machine learning (ML) simulations to predict κL values for various compounds.
• Employed density functional theory (DFT) calculations to investigate electronic and transport properties.

Main Results:

• ML models predicted several A2CdX compounds (A = Li, Na, K; X = Pb, Sn, Ge) with κL < 1.0 W/mK.
• DFT calculations revealed that K2CdPb, K2CdSn, and K2CdGe exhibit a figure of merit (ZT) greater than 1.0.

Conclusions:

• The studied cadmium compounds show potential for advanced thermal management and insulation.
• K2CdPb, K2CdSn, and K2CdGe are identified as promising candidates for thermoelectric energy conversion.