Self-selective van der Waals heterostructures for large scale memory array.

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Science

Background:

• Large-scale crossbar arrays are key for 3D memory and neuromorphic computing.
• Sneak currents and integration challenges hinder current crossbar memory cell designs.

Purpose of the Study:

• To introduce a novel self-selective memory cell for large-scale crossbar arrays.
• To overcome limitations of existing memory cells, such as process integration and destructive read operations.

Main Methods:

• Fabrication of a vertical heterostructure memory cell using hexagonal boron nitride (hBN) and graphene.
• Integration of non-volatile and volatile memory operations within hBN layers.
• Utilizing a graphene layer to control volatile filament diffusion.

Main Results:

• Demonstrated self-selectivity of 10^10.
• Achieved an on/off resistance ratio exceeding 10^3.
• Minimized sneak currents for practical large-scale memory operation.

Conclusions:

• The developed self-selective memory cell offers a viable solution for terabit-scale, energy-efficient 3D memory.
• This approach addresses fundamental challenges in crossbar array architectures.
• Enables practical readout margins crucial for advanced computing systems.