Monolithic 3D memristor-TFT stack enables electrically programmable reservoir dynamics

October 6, 2026

Monolithic 3D memristor-TFT stack enables electrically programmable reservoir dynamics

Image: ChatGPT

Researchers from Seoul National University and Yonsei University have published a technical paper on a monolithically three-dimensionally integrated memristor-thin-film transistor stack for electrically programmable multimode reservoir computing. The work targets a core limitation of many hardware reservoir systems: their temporal response is often fixed or only narrowly adjustable by the intrinsic relaxation time of the material.

The demonstrated reservoir kernel combines Ru/HfO2/TiN memristors with In2O3 channel thin-film transistors in a vertical, monolithic 3D integration scheme. In this architecture, the TFTthin film transistorsTransistors whose active semiconductor is a deposited thin film, as in displays.s modulate the memristive conductance dynamics, enabling programmable temporal behavior rather than relying on a single built-in time constant.

That electrical controllability is the central result of the device stack. Instead of using a reservoir with one static dynamic regime, the integrated structure supports multimode operation across a wide timescale. The paper positions this as a hardware path for temporal information processing, where the internal dynamics of the system can be tuned to the task.

The reported structure is published in Nature Communications and is described as a monolithic 3D integration of memristors and TFTthin film transistorsTransistors whose active semiconductor is a deposited thin film, as in displays.s for reservoir computing. Further technical details beyond the device composition and its electrically programmable dynamics were not provided in the material.

Source: Semiconductor Engineering

More background: How Does It Work?

← All posts