A New Spin on Computer Technology
By Mike Lotti
Spintronic computers, featuring zero boot-up time, ultra-low energy use and high processing speeds, aren’t available to consumers yet. But the University of Minnesota’s Center for Spintronic Materials, Interfaces, and Novel Architectures (C-SPIN) has been guiding a national “dream team” of researchers since 2013 to accelerate progress toward spintronic computing.
Here’s an overview of how this breakthrough technology works. Your computer, tablet, smart phone, and even calculator are basically machines that encode and process ones and zeros in the form of electric current. But all those electrons moving around cause heat, and it’s getting harder and harder to make electric devices small enough to meet the ongoing demand for more computing power in less space. The spintronic solution is to rethink ones and zeros as the “up” or “down” orientation of electrons in ultra-small magnets. Want a one? Make two magnets point “up.” Want a zero? Make them point in opposite directions. No moving electrons, very little heat, lots of room to cram magnets together. What’s not to like?
While the theory behind spintronic computing is solid, the technology to carry out the theory is still being developed in the lab. For example, C-SPIN researchers are exploring topics such as “What materials are best for recording a magnetic one and zero?” “What’s the most energy-efficient way to switch a one to a zero and vice versa?” “How can spin-based information be easily transferred from one part of a computer to another?”
Progress, the scientific (unpredictable) way
Until C-SPIN came into existence in 2013, very little comprehensive, integrated research had been devoted to these questions. Some potential spintronic materials were so new and under-studied that scientists didn’t even agree upon methods for testing them.
After two years of experimentation and theoretical development, the Center has made considerable progress. Dozens of new materials have been developed and tested, leading many to be discarded and a few to be further studied. Theorists have figured out why some materials have better spin properties than others, which has helped other scientists winnow the field of potential new materials. Many devices and proto-devices have been made to see how materials act when put next to each other, and in a few instances, C-SPIN researchers think that they know what some of the basic building blocks of a spintronic computer will be.
In 2013, for example, a C-SPIN researcher noticed that gadolinium oxide allowed for very low-energy “switching” when combined with cobalt — and he wasn’t even trying to find out anything about either material. Another C-SPIN researcher figured out why it happened, and now gadolinium oxide/cobalt is the focus of intense study and development within the Center.
How close are we to a spintronic computer?
In some ways, the industry is a third of the way there already. Computers have three basic parts: processing, short-term memory (also known as RAM) and long-term storage. Long-term storage – the type that exists on a hard drive – has been spin-based for the past 70 years. Why? It is “non-volatile,” which means that once you change a spin with a magnetic field, it will stay that way until another magnetic field comes along. In other words, you don’t need electric current running through mini-wires to keep a one a one and a zero a zero.
Processing and RAM, however, require billions (or trillions, for large computers) of switches every second, and materials for fast, efficient magnetic switching simply don’t exist yet.
According to Jian-Ping Wang, Director of C-SPIN and professor in the Department of Electrical and Computer Engineering, spintronic RAM devices are close to a commercial reality. In fact, spin-based RAM is already on the market for customers who want non-volatile short-term memory, but it’s expensive and slow. In the past two years, C-SPIN has put a lot of resources into developing “magnetic tunnel junctions” (or “MTJs”) which function as replacements for the current-driven transistors in today’s RAM. The most efficient RAM today uses four or five transistors to encode a one or a zero but one MTJ can perform the same function at a fraction of the energy and with more speed. And because these MTJs don’t need electrons running through them all the time, they last much longer than normal transistors. C-SPIN researchers have been focused on improving the quality of MTJs, figuring out ways to quickly and efficiently switch them and developing ways to get millions of them to work together.
Spin-based processing is further behind. Very few scientists even thought about spin-based processing before C-SPIN opened its doors, says Wang. No one even knew what the fundamental components should look like or how they would compare with standard, current-based devices. So C-SPIN has primarily invested its collective brainpower in designing and refining models for spin-based processing devices. As it heads into its third year, the Center will begin testing materials for processing performance and perhaps even constructing some prototypes.
2035 here we come
For Wang and his C-SPIN colleagues, this means that the world will probably see efficient and affordable spintronic RAM within five years and a nearly 100 percent spintronic computer 10 years after that. In the meantime, he thinks, mainstream (i.e. current-based) computers will adopt some spin-based components and become “hybrid” machines.
So, don’t look for the “Spintronic Special” at Best Buy just yet. But if you and Best Buy are still around in 2035, those new spintronic computers, tablets and smart phones will probably be worth a shopping trip.
Mike Lotti is communications associate for C-SPIN.