Symposium DS06: Integrating Machine Learning with Simulations for Accelerated Materials Modeling
2023 MRS Fall Meeting Best Poster Awards

Symposium QT03: Higher-Order Topological Structures in Real Space—From Charge to Spin

Laura Bégon-Lours, IBM Research & ETH Zürich

Ferroelectric Hafnia Superlattices for Bio-Inspired Computing OnDemand

Written by Matthew Nakamura

Laura Bégon-Lours of IBM Research & ETH Zürich delivered an insightful talk emphasizing the intersection of artificial intelligence (AI) and environmental sustainability. She acknowledged AI’s aptitude in rapid learning and prediction and highlighted the imperative for energy-efficient solutions to align with greenhouse gas emission goals. Bégon-Lours detailed the innovative use of ferroelectric materials, specifically HZO-SL (ferroelectric HfO2/ZrO2 superlattices), in crafting synaptic weights for in-memory deep neural networks. Overcoming challenges such as large voltages and footprint issues, the research team achieved sub-volt programming and significantly reduced the footprint. The integration of HZO-SL devices into the Back-End-Of-Line of CMOS was successfully demonstrated, showcasing remarkable properties, including a large On/Off ratio, ultra-fast switching, linear readout, and high endurance. This work presents a promising outlook in applying these advancements in hardware for artificial neural networks, thereby supporting bio-inspired computing. The talk signifies a crucial step toward deploying AI responsibly utilizing cutting-edge technologies.


The comments to this entry are closed.