A Study of the Variability in Contact Resistive Random Access Memory by Stochastic Vacancy Model

Area of Science:

• Materials Science
• Electrical Engineering
• Solid State Physics

Background:

• Resistive random access memory (RRAM) cell variability is a major obstacle for high-density RRAM array development.
• Stochastic oxygen vacancy generation and recombination are widely considered the primary cause of resistive switching variability in transition metal oxide films.

Purpose of the Study:

• To establish a stochastic model linking RRAM cell initial states to subsequent switching characteristics.
• To investigate the impact of oxygen vacancy concentration and distribution on RRAM performance using simulations.
• To propose a method for reducing variability in RRAM devices.

Main Methods:

• Analysis of experimental data to develop a stochastic model.
• Monte Carlo simulation combining a conduction network model and trap-assisted tunneling mechanism.
• Experimental verification using contact RRAM arrays.

Main Results:

• A stochastic model was established, accurately predicting RRAM characteristics based on initial states.
• The model demonstrated the influence of intrinsic oxygen vacancy concentration and distribution.
• A strong correlation between forming characteristics and initial states, linked to pre-existing oxygen vacancies, was verified.
• Measurement data from RRAM arrays showed good agreement with model projections.

Conclusions:

• Intrinsic oxygen vacancies are a key source of variability in contact RRAM devices.
• The developed model provides a comprehensive understanding of RRAM variability mechanisms.
• A reset training scheme can effectively reduce variability in subsequent RRAM states.