Researchers Uncover Electrothermal Multi Filamentary Dynamics Driving Stochasticity In Next Generation Volatile Memristor Semiconductors

SKKU researchers use thermal microscopy to reveal how multi-filamentary dynamics drive randomness in memristors, advancing TRNG and probabilistic computing.

By: AXL Media

Published: Mar 11, 2026, 5:06 AM EDT

Source: The information in this article was sourced from Sungkyunkwan University External Affairs Division

Direct Observation of Memristor Switching Dynamics

The development of next-generation semiconductors relies heavily on understanding the internal movement of ions, a process that has historically been difficult to observe in real time. A joint research team led by Professor Jung Ho Yoon has bridged this gap by utilizing scanning thermal microscopy to detect the heat signals generated during resistive switching. According to the study published in Advanced Functional Materials, this nanoscale characterization technique allowed researchers to measure Joule heating directly from the device surface, providing a clear window into the electrical properties of volatile memristors.

The Origin of Inherent Stochasticity

For years, the unpredictable nature of memristor switching was poorly understood, often attributed to the simple formation and rupture of a single filament. However, the thermal measurements revealed a much more complex reality involving the repeated appearance and disappearance of multiple localized hot spots. According to the research team, these hot spots are decisive evidence that multiple conductive filaments are simultaneously competing for current conduction. This multi-filamentary behavior, coupled with electrothermal effects, is the fundamental source of the stochasticity required for advanced computing.

Advancing True Random Number Generation

The inherent randomness of these devices makes them ideal candidates for information security applications that require completely unpredictable encryption keys. Leveraging their findings, the researchers implemented a bimodal true random number generator capable of producing both digital and analog random outputs. According to Professor Yoon, the team successfully demonstrated a full sequence of data encryption and decryption using these generated random numbers. This marks a significant step toward hardware-based security systems that are virtually impossible for traditional algorithms to crack.