Compact Modeling of Pd-MoS2 Self-rectifying RRAM based on modulated Schottky barrier equation

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Engineering

Background:

• Artificial intelligence (AI) computing demands energy-efficient hardware accelerators for matrix-vector multiplication.
• Resistive random-access memory (RRAM) offers analog weight storage and crosspoint array compatibility but faces challenges with sneak-path currents in high-density arrays.

Purpose of the Study:

• To fabricate and characterize a self-rectifying RRAM device using Palladium (Pd) and Molybdenum Disulfide (MoS2).
• To develop a physics-informed compact model for evaluating sneak-path current suppression in RRAM arrays.
• To investigate the impact of material properties on device rectification and array-level performance.

Main Methods:

• Fabrication and experimental characterization of Pd-MoS2 based RRAM devices.
• Development of a physics-informed compact model incorporating modulated thermionic emission for Schottky barrier physics.
• Circuit-level simulations to evaluate sneak-path current suppression at the array level.

Main Results:

• The Pd-MoS2 device demonstrated asymmetric bipolar resistive switching due to Schottky barrier-controlled carrier injection.
• The compact model accurately reproduced experimental current-voltage characteristics, including a rectification ratio of ~60 and a memory window of 10^3.
• The model enabled predictive estimation of Schottky barrier height and rectification for various metal/MoS2 combinations.

Conclusions:

• The developed Pd-MoS2 self-rectifying RRAM device effectively suppresses sneak-path currents, improving array scalability and reliability.
• The physics-informed compact model provides accurate simulation capabilities for RRAM circuit design.
• This work offers a pathway towards more efficient and reliable RRAM-based hardware accelerators for AI applications.