Machine learning accelerated design of 2D covalent organic frame materials for thermoelectrics

Abstract

Both thermodynamic stability and high thermoelectric figure of merit are required for organic thermoelectric materials for practical application. However, it is extremely time-consuming to design advanced organic thermoelectric materials via conventional theoretical calculation. In this work, combining first-principles calculations, machine learning fitted potential and solving phonon Boltzmann transport equation, we propose a new group of two-dimensional covalent organic frame semiconductors 2AL-PR-X (X = 2H, Ni, Pt, Zn) as promising organic thermoelectric materials. We found that the embedding metal atoms in the center can enhance remarkably the stability of porphyrin ring, demonstrated by evaluating the system total energy and phonon dispersion relationship. Moreover, the interaction between the metal atoms and the porphyrin ring will affect the phonon modes in the porphyrin ring, result in reduction in thermal conductivity. We found that 2AL-PR-Pt is the best thermoelectric candidate material among the studied 2AL-PR-X, which has a peak ZT of 0.32 at room temperature and can reach around 0.9 at 800 K. Our work provides strategy for the thermal transport regulation of two-dimensional covalent organic frame materials and demonstrates the large potential of 2AL-PR-X materials for thermoelectric applications.