Thermal Transport through Defects in Rutile with Atomic Resolution STEM-EELS

Scanning transmission electron microscopy (STEM) with electron energy-loss spectroscopy (EELS) has been used routinely to study the chemical composition of samples with atomic precision for long time. In 2014 monochromation with full-width half-maximum below 10 meV in a modern aberration corrected setup has been reported [1]. Since then, the vibrations in materials are directly being studied with subatomic resolution [2-4]. Among others, this offers direct insights into the heat transfer mechanisms in materials on the atomic scale. Here we present our study of system of rutile (TiO2) with a planar defect. We report a direct local measurement of vibrational states across the defect, offering comprehensive methodology for studying of thermal transport across defects. The experimental results are complemented by extensive simulations of vibrational STEM-EELS using the Time Auto-Correlation of Auxiliary Wavefunction (TACAW) method [5], featuring our new computation engine Torched-TACAW. Through high energy resolution EELS, we comprehensively capture the energy-spectrum of vibrations. By utilizing an off-axis detector geometry, we selectively capture vibration modes with specific polarization vectors [6-8], allowing direct visualization of vibrational modes at the nanoscale. These measurements provide critical insights into heat transport mechanisms across crystallographic defects. Further understanding is developed by presented simulations based on the TACAW method. Presented theoretical results were obtained using machine-learned interatomic potential Orb2 [9] and our new simulation engine Torched-TACAW, which offers user-friendly and fast implementation of the TACAW method with (optional) GPU acceleration. Our integrated experimental and theoretical approach reveals the intricate thermal transport mechanisms in rutile, establishing a robust framework for investigating vibrational dynamics in defected and heterogeneous material systems. By combining high-resolution STEM-EELS with advanced computational modeling, we demonstrate a powerful methodology for understanding thermal properties at the nanoscale.