A spintronics processor using magnetic tunnel junction devices
The operation of conventional charge-based electronic transistor systems is based on the storage and transfer of electrons in the form of an electric charge or a current. It has been projected that by about 2020, charge-based devices will reach fundamental limits of speed, size, energy, and noise immunity. Spintronics devices, in contrast, are based on the up or down "spin" of electrons rather than on charges. The state of a spintronics device can be changed by controlling the magnetic fields generated by currents on the device's input lines. These spin states can be mapped to the binary logic states of a digital computing system. Because of the unique characteristics of spintronics devices, a processor constructed from these new devices potentially could be faster, provide superior heat dissipation characteristics, and have greater device density than a processor constructed from conventional charge-based devices. Furthermore, the information stored in the processor would be non-volatile allowing it to immediately continue executing from its previous state when the power is cycled. It also is expected that these devices will be more resistant to all types of internally and externally generated noise than conventional transistor devices making them more resistant to radiation-induced upsets, power supply variations, and signal cross-talk.
The primary goal of this project is to demonstrate the feasibility of using spintronics-based magnetic tunnel junction (MTJ) devices to implement complex digital systems. The first step in this exploratory project will be to design and fabricate a single-bit full adder at the NSF-supported University of Minnesota Nanofabrication Center using the prototype MTJ devices previously demonstrated by the PIs. Actually fabricating and characterizing the adder will lead to an understanding of how to construct more complex computational elements from the magnetic gates. In parallel with this fabrication process, a multi-function bit-serial ALU will be designed and simulated that will take advantage of the programmability of the magnetic gates. The ultimate long-term goal beyond this exploratory feasibility study is to design a fully functional processor entirely from these novel non-charge-based spintronics devices.
Return
to: U of M Home
