3D Integration of functionally diverse 2D materials for optoelectronic reservoir computing.

Area of Science:

• Materials Science
• Nanotechnology
• Electronics Engineering

Background:

• Three-dimensional (3D) integration of non-silicon materials is advancing edge intelligence.
• This integration enables diverse functionalities like sensing, storage, and computing beyond traditional transistor scaling.
• Efficient information processing at the source reduces latency and power consumption.

Purpose of the Study:

• To demonstrate reservoir computing using 3D integration of In2Se3-based photodetectors and MoS2-based memtransistors.
• To leverage the unique properties of these materials for enhanced edge intelligence applications.
• To showcase the benefits of near-sensor processing for improved performance.

Main Methods:

• Constructed an optical reservoir using In2Se3 flakes of varying thicknesses for photoresponse variation.
• Employed programmable MoS2 memtransistors to convert photocurrent to photovoltages.
• Utilized a trained readout circuit based on MoS2 memtransistors for data processing.

Main Results:

• Achieved physical proximity between sensors and computing elements <50 nm, surpassing current packaging.
• Demonstrated near-sensor information processing for improved photoresponse calibration and speed.
• Successfully implemented reservoir computing in a vertically stacked heterogeneous material system.

Conclusions:

• The developed 3D stack represents a significant milestone in vertically stacked functional layers beyond silicon.
• The near-sensor and in-memory compute capabilities are crucial for next-generation edge applications.
• This work paves the way for advanced heterogeneous material integration in neuromorphic computing and AI.