The thermoelectric performance of new structure SnSe studied by quotient graph and deep learning potential

Abstract The structure predicting and structure screening of new structure SnSe are studied by quotient graph and deep learning potential. Using the quotient graph, we predict the new structure based on the feature of the experimental structure. Facing many structures generated by the quotient graph and reducing computation cost, the deep learning potential is used to optimize the structure. Through the electronic fitness function, it is found that the new structure with space group no. 64 shows a good electrical property of p type and n type. Based on the new structure, the n-type and p-type thermoelectric properties are studied by First-principles calculations and Boltzmann transport theory. The mechanical stability and thermodynamic stability of the new structure are checked, indicating that the new structure is stable. Due to the cyclic structure, the new structure shows large values of bulk modulus (B ∼ 62.82 GPa), shear modulus (G ∼ 43.78 GPa), Young’s modulus (E ∼ 106.59 GPa), Debye temperature (θ=291K), sound velocity (Va ∼ 2890 m/s) and low value of Gruneisen parameter γ ∼2.2, implying that strong Sn-Se atomic interactions and large lattice thermal conductivity. At 800K, the figure of merit of p type (1.83) is superior to that of n type (1.68). The main scattering process of both charge transport and thermal transport is dominated by optical phonons braches. This work not only demonstrates the excellent property of new structure, but also provides insights to understand the transport performance of the cyclic structure.