Understanding local physical properties of barium titanate ceramics depending on the synthesis conditions assisted by a machine learning algorithm

Barium titanate (BT) is widely used in electronic components, such as multilayer ceramic capacitors, owing to its high dielectric constant and piezoelectricity. The continuously increasing demand for more efficient devices has led to the development of smaller and more efficient BT ceramics by controlling the synthesis conditions. However, despite the decrease in size of BT grains to the nanoscale, the physical properties of BT ceramics have mostly been investigated on a macroscopic scale. In this study, we investigated the effects of synthesis conditions on local physical properties such as surface roughness, leakage current, and domain structure/switching using atomic force microscopy. We further applied machine learning analysis to effectively understand the effects of synthesis conditions on these physical properties; the analysis not only provides local information on BT ceramics but also a comprehensive understanding of their physical properties.