SHF: Small: Pipelined and wireless ultra-low power straintronics: An acoustically clocked combinational and sequential nanomagnetic architecture

Virginia Commonwealth University, Richmond VA

Investigators

Abstract

Elliptical single-domain nanomagnets with two stable magnetization orientations are far more energy-efficient as logic switches than traditional transistors. However, the method employed to switch them must be energy-efficient as well in order to build ultra-low-power nanomagnetic logic and memory paradigms. It has been theoretically shown that using multiferroic (magnetostrictive-piezoelectric) nanomagnets, whose magnetization can be flipped with strain generated by a tiny electrostatic potential applied across the piezoelectric layer, results in a remarkably energy-efficient switching scheme. It reduces the dissipation in the switching/clocking circuit by four orders of magnitude at a clock rate of ~ 1 GHz compared to other nanomagnet switching schemes. While this is attractive, an unattractive trait of nanomagnetic logic chains is that in order to build a pipelined architecture and hence retain an acceptable bit transfer rate, each magnet must be clocked individually. This necessitates contacting each magnetic with a contact line, which imposes a Herculean lithographic burden. The PIs propose to overcome this problem completely by designing and fabricating a novel acoustic scheme for clocking that allows pipelining and at the same time does not require contacts to every magnet, thereby completely lifting the lithography burden. A surface acoustic wave (SAW) launched in the substrate, and slowed down with periodically placed masses, generates strain in an array of magnets in the correct sequence for bit transfer, as long as the spacing between the magnets is one quarter of the SAW?s wavelength. With this scheme, the energy dissipation in a gate operation at room temperature can be very low. This project will: (i) design combinational and sequential logic based on acoustically clocked magnetostrictive nanomagnets acting as logic switches, as well as perform extensive simulations using the stochastic Landau-Lifshitz-Gilbert (LLG) equation to understand and optimize reliability and fault tolerance in the presence of thermal noise; (ii) experimentally demonstrate pipelined unidirectional logic flow, and (iii) develop comprehensive coupled models for the switching dynamics of nanomagnets stressed by surface acoustic wave (SAW). This research will result in a novel computational paradigm whose astonishing energy efficiency combined with very little lithographic burden could enable the production of cheap, high yield and extremely low power processors. Such processors would consume so little energy that they can be run off the energy harvested from the environment. This could open up hitherto unimaginable applications such as medically implanted processors powered only by the motion of the patient's body, or processors that monitor the structural health of bridges and buildings while being powered by vibrations caused by wind or traffic. Integration of this research with education and mentoring will result in traditional training activities such as guiding two doctoral students who will gain multidisciplinary skills in advanced nanofabrication, nanocharacterization and modeling, as well as undergraduate projects on SAW devices and nanofabrication of magnetostrictive nanomagnets that will be mentored by the PI and co-PI?s doctoral students. Other innovative outreach programs will include holding workshops on nanomagnets and computing for high school students through the Math Science Innovation Center (MSIC) and incorporating diversity into outreach programs by hosting under-represented K-12 students in summer with the help of VCUs Richmond Area Program for Minorities in Engineering (RAPME) program. These students will perform nanolithography under supervision and study the magnetic structures they create with MFM.