Theoretical investigation of the lattice thermal conductivities of II-IV-V2 pnictide semiconductors

Victor Posligua,Jose J. Plata,Antonio M. Márquez,Javier Fdez Sanz,Ricardo Grau-Crespo

2023-09-08

Abstract:Ternary pnictides semiconductors with II-IV-V2 stoichiometry hold potential as cost effective thermoelectric materials with suitable electronic transport properties, but their lattice thermal conductivities ($\kappa$) are typically too high. Gaining insight into their vibrational properties is therefore crucial to finding strategies to reduce $\kappa$ and achieve improved thermoelectric performance. We present a theoretical exploration of the lattice thermal conductivities for a set of pnictide semiconductors with ABX2 composition (A = Zn, Cd; B = Si, Ge, Sn; and X = P, As), using machine-learning based regression algorithms to extract force constants from a reduced number of density functional theory simulations, and then solving the Boltzmann transport equation for phonons. Our results align well available experimental data, decreasing the mean absolute error by ~3 Wm-1K-1 with respect to the best previous set of theoretical predictions. Zn-based ternary pnictides have, on average, more than double the thermal conductivity of the Cd-based compounds. Anisotropic behaviour increases with the mass difference between A and B cations, but while the nature of the anion does not affect the structural anisotropy, the thermal conductivity anisotropy is typically higher for arsenides than for phosphides. We identify compounds, like CdGeAs2, for which nanostructuring to an affordable range of particle sizes could lead to values low enough for thermoelectric applications.

Materials Science

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to understand and reduce the lattice thermal conductivity (\(\kappa\)) of II - IV - V2 - group phosphide semiconductor materials through theoretical research in order to improve the performance of these materials in thermoelectric applications. Specifically, although these materials have the potential for suitable electron - transport properties, their lattice thermal conductivity is usually too high, which limits their application as efficient thermoelectric materials. Therefore, researchers hope to find effective strategies to reduce the thermal conductivity by in - depth exploration of the vibration characteristics of these materials, thereby improving their thermoelectric performance.

To achieve this goal, the researchers adopted a machine - learning regression algorithm to extract force constants from density - functional - theory simulations and solve the Boltzmann transport equation of phonons, and carried out a theoretical exploration of the lattice thermal conductivity of a series of phosphide semiconductor materials with ABX_2 composition (A = Zn, Cd; B = Si, Ge, Sn; X = P, As). The research results not only match the existing experimental data well, but also have an improvement in prediction accuracy compared with previous theoretical work, with the mean absolute error reduced by about 3 Wm^-1K^-1. In addition, the study also found that the average thermal conductivity of Zn - based ternary phosphides is more than twice that of Cd - based compounds, indicating that the selection of different cations can significantly affect the thermal conductivity of materials. The paper also discussed the influence of nanostructures on thermal conductivity, pointing out that for some compounds such as CdGeAs_2, through nanostructuring within an appropriate range, a low enough thermal conductivity can be achieved, which is suitable for thermoelectric applications.