Assessing phonon anharmonicity in KTaO₃ through stochastic methods and Machine Learning Force Field

The first ideas linking the ferroelectric phase transition to a long wavelength mode softening were laid down by Cochran in the 60′ and confirmed by Cowley a couple years later by studying the Strontium Titanate (SrTiO3) perovoskite. What was furthermore observed, is that around 0K a peculiar type of ferroelectric incipiency may furthermore occur, called “Quantum Paraelectricity”. The soft mode is here stabilized by zero-point quantum vibrations of strong anharmonic character. A concurring phenomenon is the plateau of the dielectric constant persisting at a high and fixed value down to 0K.

The Potassium Tantalate (KTaO3) perovskite is a good candidate for studying the quantum paraelectric state with full ab-initio techniques, due to its structural simplicity (cubic at all temperatures), a fair amount of experimental data and many microscopic models developed for more than 60 years.

The behavior of the Gamma point TO1 soft mode frqeuency in KTaO3 is thus here investigate in the temperature range between 0 and 300K by adopting an ab initio methodology that combines the Stochastic Self Consistent Harmonic Approximation (SSCHA) for assessing full phonon anharmonicity and the Machine Learning Force Field (MLFF) as implemented in VASP. This approach offers a fast and flexibile tool for the characterization of phonon-mediated phenomena beyond the harmonic approximation and for the study of phase transitions.