Samsung's computational approach zeroes in on 18 chalcogenide materials for next-gen SOM memory

With a clever approach, Samsung is pushing forward on developing Selector-Only Memory (SOM) tech: they’re using advanced computational modeling to zero in on the best chalcogenide materials for next-gen memory. Their research team went through over 4,000 different material combinations and narrowed it down to 18 that seem promising for real-life tests.

SOM tech could be a giant leap in memory design, combining the best of both worlds: it’s got the non-volatile perks of flash storage while keeping the speed of DRAM. It uses chalcogenide materials to act as both memory cells and selectors, doing away with the need for separate transistors that you’d find in more common phase-change or resistive RAM setups.

The team used complex simulations to examine bonding properties, how stable the materials are under heat, and how they behave electrically. They focused on things like how much the threshold voltage drifts and how stable the memory window is, which basically means how well it keeps on and off states distinct from each other.

Samsung is set to present these findings at the International Electron Devices Meeting in San Francisco this December. They believe this computational approach might have led them to high-performance materials that traditional lab testing might’ve missed.

This all builds on Samsung’s earlier work at IEDM 2023, where they showed off a 64Gbit OTS-based SOM with tiny 16nm memory cells.