Deok-Hwang Kwon1
Korea Institute of Science and Technology1
Deok-Hwang Kwon1
Korea Institute of Science and Technology1
Combining the power of atomic structure imaging in transmission electron microscope (TEM) and current-voltage (I-V) measurement, in-situ I-V/TEM becomes a powerful tool to characterize the nanoscale of materials while applying voltage simultaneously. Particularly in electronic materials, its significant effectiveness was successfully demonstrated over decades. For example, one of the candidates suggested as the next electronic device for neuromorphic computing, the resistive switching phenomenon (also referred to as Memristor) has been highly investigated. With significant interest in a Neuromorphic computing device, massive research has been conducted on its mechanism and applications.<br/><br/>While the resistive switching phenomenon, which is a straightforward thing, occurs when an external potential is applied to a thin film oxide sandwiched between the top and bottom electrodes, its mechanism has remained controversial due to the lacking of direct evidence. A mobile defect such as oxygen vacancy has been frequently inferred to be the reason provoking the phenomenon; however, their physical form and distribution remained elusive. To resolve this issue, I employed in-situ I-V/TEM and directly investigated the switching mechanism in oxide materials. Various oxide thin films were studied as model systems, such as TiO<sub>2</sub>, SrTiO<sub>3</sub>, and SrFeO<sub>3</sub>. Since in-situ I-V/TEM enables the operation of electronic devices in TEM while measuring atomic structure changes simultaneously, the switching area was directly probed and visualized. Chemical analysis employing electron energy-loss spectroscopy was carried out, confirming oxidation state changes of transition metals. High-resolution TEM imaging and chemical information reveal the atomistic mechanism of resistive switching. Also, thermodynamic modeling is established. The deep insight found from these results sheds light on pathways to Memristor and Neuromorphic computing.