Developing STT-RAM as a Scalable High Density Universal Memory with High Speed and Low Power
Develop spin-transfer torque-based random access memory (STT-RAM) with enhanced performance and lower cost for use as a scalable, non-volatile universal memory solution with low power consumption for both the embedded and stand-alone memory markets.
Sponsor: Grandis, Inc.1123 Cadillac Court
Milpitas, CA 95035
As semiconductor technology scales toward ever-shrinking sizes, existing mainstream semiconductor memories based on electronic charge storage are encountering significant performance and reliability limitations. Furthermore, in today's advanced digital systems, various types of volatile and non-volatile memories-each with unique features-must be integrated together, thus complicating the system design. A scalable, universal memory technology that is non-volatile and consumes little power would therefore be a great enabler of simple system design and true mobility. In this project, spin-transfer torque (STT) switching will be used to provide a second-generation, scalable MRAM (magnetic random access memory) solution called STT-RAM. Derived from recent advances in the field of spintronics, STT switching utilizes a spin-polarized electron current flowing through a magnetic tunnel junction (MTJ) to directly switch the magnetization of a nanomagnet. STT switching is fast, consumes little power, and is intrinsically scalable to future semiconductor technology nodes. This ATP project will enable Grandis to explore a multi-faceted approach, including innovations in materials, devices and circuitry, to enhance the performance of STT-RAM with reduced cost. If successful, the program will strengthen U.S. leadership in this critical technology and will accelerate by at least two years the commercialization and adoption of STT-RAM as a scalable, non-volatile universal memory solution for both the embedded and stand-alone memory markets (with a total market size exceeding $100 billion annually).