ULTRARAM: An Emerging Memory Platform for NVM and Low-Power Neuromorphic Systems

Abhishek Kumar1, Musaibh Farooq Dar2, Peter D. Hodgson3, Dominic Lane3, Peter J. Carrington3, Evangelia Delli3, Richard Beanland4, Shruti Mehrotra5, James Ashforth-Pook6, Manus Hayne3, Avirup Dasgupta2
1University of California Berkeley, 2Indian Institute of Technology Roorkee, 3Lancaster University, 4Warwick University, 5GlobalFoundries, 6QuInAs Technology Limited


Abstract

ULTRARAM is an emerging memory technology exhibiting high endurance (> 10^7 P/E cycles), ultra-high retention (>1000 years), and ultra-low switching energy per unit area. This compound semiconductor-based non-volatile memory (NVM) works on the principle of triple-barrier resonant tunneling (TBRT) using InAs/AlSb heterostructures. Single memory cells are fabricated on GaAs and Si substrates. In addition, a physics-based compact model has been proposed to accurately capture the real-time trapping/de-trapping of charges in the floating gate (FG) and utilized for the synapse simulations. Array-level simulations and benchmarking highlight the promise of this technology as a next-generation non-volatile memory cells. Additionally, a circuit-level macro-model is employed to evaluate and benchmark the on-chip learning performance in terms of area, latency, energy, and accuracy of an ULTRARAM synaptic core.