Low energy magnetic domain wall logic

点击翻译按钮获取中文摘要

Heat dissipation is the most serious problem confronting modern electronics, and limits the clock rates and chip performance of microprocessors. The proposed work develops a magnetic logic device with zero static power losses and a projected dynamic power-delay product below that of contemporary digital logic. Low power operation is enabled by operating at a very low supply voltage, which reduces both the dynamic power when the devices are switched and circuit-wide dissipation in the interconnects. These benefits are obtained without increasing the ?off state? leakage current. The device relies on the integration of two proven technologies, current-induced domain wall propagation in a narrow soft ferromagnetic wire for the storage element, and a magnetic tunnel junction for the output element. The intellectual merit of this proposal is to understand the scaling limits in the technology, and to solve system-level issues such as fanout, clocking and circuit architecture. The scaling limits will be examined by studying domain wall propagation as a function of wire dimensions, both experimentally and computationally, to understand the reversal behavior as a function of current pulse amplitude and width and the geometrical and materials parameters of the wire. Circuit architecture and clocking schemes will be developed to enable integration of multiple devices, so that one device can drive one or more other devices, enabling complex logic operations to be performed, making use of the programmability and non-volatility of the device. The results from the project will not only demonstrate a new strategy for realizing logic, but will develop tools and architecture that can be applied to other nontraditional devices.The broader impacts of the proposal include the training of a graduate student and approximately two to four undergraduate students in a highly interdisciplinary materials science and electrical engineering field. In addition to training of personnel, the concepts and findings of this project will be incorporated into classes taught by the PIs, which will be available to the public through the MIT OpenCourseWare initiative, to high school teachers through summer experiences, and through a textbook and an associated teaching module for graduate students on magnetic tunnel junctions and spin transport. The goal of class development is to summarize the key device physics of post-silicon devices in a manner that is accessible to both graduate and undergraduate students.