FuSe2 Topic 2: Magnonic Combinatorial Memory and Logic Devices on a Silicon Platform

Information and communication technologies generate vast amounts of data that result in an urgent need for increasing the data storage density along with the functional throughput of semiconductor devices. With the support of the Future of Semiconductors (FuSe2) Program, Dr. Alexander Khitun and Professor Jacob Greenstein at the University of California - Riverside together with Professor Caroline Ross at the Massachusetts Institute of Technology will develop a new type of combinatorial memory and logic devices to provide a fundamental advantage over the existing devices in data storage density and data processing throughput.

The advantage can be achieved by utilizing phase in addition to amplitude as state variables in devices based on magnetic spin waves. The team will demonstrate prototypes of magnonic combinatorial memory that can store more bits of information than conventional magnetic memory with the same number of magnets. The team will also demonstrate prototypes for special task data processing such as Travelling Salesman Problem.

Broader impacts will focus on expanding the participation of a broad range of students in science and technology through activities and programs at both UC Riverside and MIT and by cooperation with community colleges. UCR is an accredited Hispanic Serving Institution and one of the most diverse universities in the USA. This project will also contribute to undergraduate and graduate STEM education.

The proposed combinatorial devices are expected to lead to revolutionary advances in a variety of practical applications including magnetic data storage and special task data processing.

Magnonic combinatorial devices comprise a magnonic n×n magnonic mesh with an electric part connected in an active ring circuit. The operation of magnonic combinatorial devices is based on the appealing property of the active ring circuit to self-adjust to the resonant path. The number of possible signal propagation paths in the mesh scales factorial (n×n)! with the size of the mesh.

The utilization of spin wave superpositions makes it possible to check all the paths at a time. Combinatorial devices can be used for data storage as well as accelerators for non-deterministic polynomial-time (NP) hardness problem solutions. There are working memory and logic prototypes based on ferrite films grown on garnet substrates.

To make these devices compatible with conventional technology, the team will demonstrate combinatorial memory and logic devices on a silicon platform. The results of this transformative research will add to the core knowledge in the areas of material science, signal processing, and computer engineering. The proposed magnonic combinatorial devices are expected to lead to revolutionary advances in a variety of practical applications.

The development of compact memory devices capable of storing all information generated by humankind in 1-inch × 1-inch matrix together with a functional throughout enhancement above 10^30 Ops/(m^2∙s∙μJ) (i.e., the throughput of all existing supercomputers combined) will have an enormous impact on the semiconductor industry as well as on the various fields of our life. Broader impacts will focus on expanding the participation of a broad range of students in science and technology through activities and programs at both UC Riverside and MIT and by cooperation with community colleges.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.