Unfolding the connection between electronic and thermal transport

Various forms of electron and ion interactions in materials are broadly responsible for the observed electronic and thermal transport properties. Despite the overlapping origin, there is lack of clear connection between these two transport properties. In this work, we have used machine learning (ML) to establish a connection between otherwise independent electronic and thermal transport properties, whereinbonding features (BF)are found to be the bridging link.

By creating a comprehensive database, electronic and thermal transport properties of 135 materials are calculated. Analysis of independent prediction models for these transport properties reveals the potential of elemental and bonding descriptors in predicting them. Hence, to find a possible connection, we employed electronic transport properties along with chemical bonding driven elemental and structural descriptors to predict the thermal transport properties. These descriptors are electronegativity, volume, coordinationnumber, bond distances, and bond strength. Employing these descriptors, the prediction model for thermal transport property, gives root mean square error (RMSE) of 0.19/0.19 and R2 of 0.99/0.99 for the train/test data. The scatter plot for the variationof DFT versus ML predicted log-scaled is shown in Fig. 1 (a).

Fig. 1: (a) Scatter plot for DFT calculated versus ML-predicted log-scaled , (b) Contour plots showing the variation of  as a function of electronic transport properties and chemical bonding driven property.

This relationship with bonding is further substantiated by the contourplots, corresponding to the variation of  as a function ofelectronic transport and bonding properties. Among these, thecombination of mean bond distances (BD) and electronic transport properties partitions the dataset into the regions of low andhigh , as shown by the blue and red regions in Fig. 1 (b). Hence, the developed model can be used to screen materials with desired lattice thermal conductivity by looking at electronic transport and bonding attributes.

Reference : 

Unraveling the role of bonding chemistry in connecting electronic and thermal transport by machine learningJ. Mater. Chem. A, 2020, 8, 8716-8721

Website URL : http://mrc.iisc.ac.in/abhishek/