Revolutionizing Memory Technologies: Advances in Phase-Change and Quantum Memory”
A collaborative effort involving Technologies researchers from Stanford University, TSMC, the National Institute of Standards and Technology (NIST), and the University of Maryland has resulted in the development of an innovative phase-change memory tailored for future artificial intelligence (AI) and data-centric systems.
Technologies
This memory technology relies on GST467, an alloy comprising four parts germanium, six parts antimony, and seven parts tellurium. The unique composition of GST467 affords it a rapid switching speed, a quality highlighted by Asir Intisar Khan, a postdoctoral scholar at the University of California Berkeley and visiting postdoctoral scholar at Stanford.
The GST467 alloy is strategically embedded within a superlattice structure, sandwiched between several nanometer-thin materials. This configuration enhances the memory’s performance, providing a low switching energy, excellent endurance, stability, and nonvolatility – the ability to retain its state for a decade or more. The researchers achieved these qualities by leveraging the nanoscale devices within the superlattice.
During testing, the memory demonstrated the avoidance of drift and operated at voltages below 1 volt. Eric Pop, a professor of electrical engineering at Stanford, emphasized the significance of this achievement, noting that while certain types of nonvolatile memory may be faster, they often operate at higher voltage or power. The phase-change memory developed by the team is remarkable for its ability to switch in mere tens of nanoseconds while operating at a voltage below one volt, striking a crucial balance between speed and energy efficiency.
The fabrication process of the superlattice is compatible with commercial manufacturing and offers the potential for increased density by stacking multiple layers vertically. This breakthrough opens up new possibilities for the development of memory technologies tailored to the evolving demands of AI and data-centric systems.
In a separate initiative, researchers at the University of Basel have created a miniature quantum memory element based on rubidium atoms enclosed in a tiny glass cell. This quantum memory, which can be produced en masse on a wafer, holds promise for supporting quantum networks requiring memory elements for temporary information storage and routing.
Initially, the rubidium atoms were contained in a handmade glass cell, several centimeters in size. To miniaturize it to just a few millimeters, the researchers heated the cell to 100 degrees Celsius to increase vapor pressure, ensuring a sufficient number of rubidium atoms for quantum storage. Exposure to a magnetic field of 1 tesla, significantly stronger than Earth’s magnetic field, facilitated the quantum storage of photons using an additional laser beam. This method enabled the storage of photons for approximately 100 nanoseconds.
The researchers envision further improvements, including storing single photons in the miniature cells and optimizing the glass cell design, paving the way for advancements in quantum memory technology.
