Modeling thermoelectric performance of p-type Cu3SbSe4-based chalcogenide materials using decision trees and...

Area of Science:

• Materials Science
• Solid State Physics
• Computational Chemistry

Background:

• Cu3SbSe4-based materials are ternary chalcogenides with potential for efficient thermoelectric energy conversion due to their unique crystal structure.
• Their sphalerite super-lattice structure offers tunable characteristics but is limited by a restricted carrier concentration, hindering overall thermoelectric performance.
• Experimental enhancement of thermoelectric figure of merit is resource-intensive, necessitating predictive computational approaches.

Purpose of the Study:

• To develop and compare computational models for predicting the thermoelectric figure of merit (ZT) of Cu3SbSe4-based materials.
• To identify the most effective modeling approach for accurate ZT prediction, considering factors like temperature and dopant properties.
• To explore the influence of specific dopants (Sn, Fe) on the energy conversion efficiency of these thermoelectric compounds.

Main Methods:

• Modeling the figure of merit using Random Forest Regression (RFR) and a genetically optimized Support Vector Regression (GESVR) model.
• Utilizing temperature, dopant ionic radii, and concentrations as input features for the predictive models.
• Investigating the impact of Sn and Fe doping on Cu3Sb1-xSnxSe4 and Cu3Sb1-xFexSe2.8S1.2 compounds using the developed GESVR model.

Main Results:

• The GESVR model demonstrated superior performance compared to the RFR model, with significant improvements in correlation coefficient (188.04%), mean absolute error (30.18%), and root mean square error (42.36%) on testing samples.
• The study successfully predicted the influence of dopants on the thermoelectric properties of Cu3SbSe4-based materials.
• The developed models provide a precise and efficient method for exploring novel thermoelectric materials.

Conclusions:

• Genetically optimized support vector regression offers a highly accurate and efficient computational tool for predicting thermoelectric performance in Cu3SbSe4-based materials.
• This approach facilitates the accelerated discovery and optimization of materials for thermoelectric applications.
• The findings contribute to advancing green energy technologies and addressing the global energy crisis through improved thermoelectric materials.