State dependence and temporal evolution of resistance in projected phase change memory

Area of Science:

• Materials Science
• Computer Engineering
• Solid State Physics

Background:

• Phase change memory (PCM) is crucial for in-memory computing and neuromorphic systems, leveraging its tunable resistance for computations and neural network synapses.
• However, PCM devices face challenges from resistance variations due to structural relaxation, 1/f noise, and temperature fluctuations.
• The projected PCM concept aims to reduce these variations by separating resistance storage from data retrieval.

Purpose of the Study:

• To develop a comprehensive device model for projected PCM.
• To capture and understand key device behaviors: resistance drift and state-dependent resistance.
• To provide insights into the role of interfacial resistance and guide material selection and device engineering.

Main Methods:

• Development of a physics-based device model for projected PCM.
• Incorporation of resistance drift (temporal resistance evolution) and state-dependent resistance (phase configuration influence).
• Experimental validation of the model using projected PCM devices with antimony and metal nitride in a lateral geometry.

Main Results:

• The developed model successfully captures resistance drift and state-dependent resistance in projected PCM devices.
• The study elucidates the significant impact of interfacial resistance on device performance.
• Experimental validation confirms the model's accuracy and utility.

Conclusions:

• The developed device model provides crucial insights into projected PCM behavior, addressing key challenges.
• The findings offer practical guidelines for selecting materials and engineering devices for improved PCM performance.
• This work advances the understanding and application of projected PCM in neuromorphic computing and memory systems.