Hybrid 2D-CMOS microchips for memristive applications.

Area of Science:

• Materials Science
• Electrical Engineering
• Nanotechnology

Background:

• Advanced electronic circuits rely on two-dimensional (2D) materials, but integration challenges limit device density and functionality.
• Previous research focused on large, isolated 2D devices, hindering practical applications due to low yield and variability.
• Monolayer 2D materials on silicon face challenges like pinholes and cracks, limiting integration density and computational demonstrations.

Purpose of the Study:

• To develop high-integration-density 2D-CMOS hybrid microchips for memristive applications.
• To overcome limitations of previous 2D material integration methods.
• To demonstrate the potential of these hybrid chips for in-memory computation and neuromorphic applications.

Main Methods:

• Fabrication of 2D-CMOS hybrid microchips by transferring multilayer hexagonal boron nitride onto silicon microchips with 180nm CMOS transistors.
• Integration of CMOS transistors for precise current control in hexagonal boron nitride memristors.
• Patterning of top electrodes and interconnections to finalize the hybrid circuits.

Main Results:

• Achieved high integration density of 2D memristors down to 0.053 µm² with an endurance of approximately 5 million cycles.
• Demonstrated in-memory computation through the construction of logic gates.
• Measured spike-timing dependent plasticity signals, suitable for spiking neural networks.

Conclusions:

• The developed 2D-CMOS hybrid microchips offer high performance and a high technology readiness level.
• This work represents a significant advancement towards integrating 2D materials into microelectronic products and memristive devices.
• The technology enables practical in-memory computing and paves the way for future neuromorphic hardware.