Halide perovskites represent a remarkably promising category of materials for the development of novel solar cells and opto-electronic devices. In this study, we used the neuroevolution machine learning potential to examine the thermal properties of CsPbX3 perovskites (X = Cl / Br / I). In alignment with previous research, our results disclose a pronounced correlation between lattice dynamics and thermal conductivity. Regardless of the specific perovskite variant, phonon band structure analysis revealed dynamical instability, characterized by overlapping phonon branches and densely packed optical branches beginning at imaginary low frequencies around 20i cm−1. We observed a significantly low lattice thermal conductivity of approximately 0.20W/(mK). This is due to the displacement of Cs atoms from their central positions and the deformation of the PbX6 octahedra, which results in lower phonon velocities and shorter phonon lifetimes.
Read full abstract