Chinese researchers solve memory hurdle for future AI systems
What's the story
Chinese researchers have made a major breakthrough in the field of semiconductor technology. They have developed a way to make wurtzite ferroelectrics, a class of materials, much more durable. The advancement could help overcome a major reliability hurdle for high-performance computing and future artificial intelligence (AI) systems. The research was led by scientists from Xidian University in Xi'an, City University of Hong Kong, and Fudan University.
Milestone
Breakthrough in writing cycles
The team demonstrated over 10 billion writing cycles in wurtzite ferroelectrics, a class of materials that can switch between two electric states to store data.
The result was about 100 times the endurance previously achieved with the same material.
This major leap forward could help advance a potential next-generation memory technology and bring ferroelectric memory closer to practical use in future computing hardware.
Material potential
What are wurtzite ferroelectrics?
In recent years, wurtzite ferroelectrics such as aluminum scandium nitride (AlScN) have emerged as promising next-generation memory materials.
They offer fast switching speeds and potentially low energy consumption.
Importantly, AlScN is compatible with existing semiconductor manufacturing processes, making it easier to integrate into future memory devices.
However, the material has faced a major hurdle: deterioration after repeated electrical switching.
Durability improvements
New research pushes limits significantly
Existing AlScN devices have typically failed after about 100 million writing cycles, far from the billions needed for commercial use.
However, the new research by Chinese scientists has pushed this limit significantly.
The team's work could pave the way for more durable and reliable memory chips, making them suitable for high-performance computing and advanced AI systems.