Machine learning-accelerated molecular dynamics calculations for investigating the thermal modulation by ferroelectric domain wall in KTN single crystals

Abstract

Ferroelectric perovskite materials, containing ferroelectric domain configurations, are promising thermal switching candidates in thermal management due to their fast response and efficient heat flow control. However, most of ferroelectric materials possess fixed or narrow Curie temperature range. In present study, the thermal-switching characteristics of ferroelectric potassium tantalate niobate (KTN) crystals, in which the Curie temperature can be adjusted by Ta/Nb ratio, are investigated by Machine Learning-Accelerated Molecular Dynamics calculations. Results show that temperature is an effective way to modulate thermal transport behavior. For 180°- and 90° DW, maximum thermal switching ratio are obtained at 300 K, with 1.80 and 1.89, respectively. Further modulation of thermal by the density of DWs strengthen the thermal switching effect. After introducing nine DWs in our calculation model, thermal switch ratio can be modulated to 2.19 and 2.29 for 180°- and 90° DWs configurations, respectively. Phonon anharmonicity investigation demonstrates that the decrease of phonon relaxation time of low frequency phonons (0–15 THz) are responsible for the difference of thermal conductivity between mono- and multidomain walls configuration. Large thermal switching ratio and flexible regulation of Curie temperature provide ferroelectric KTN crystal a broad application prospect in the field of intelligent thermal management.