Fresh Spin On Logic

MIT team paves way for a spintronics microprocessor

3 min read

In the last few years, a new type of memory has begun to penetrate the market for nonvolatile data storage. The devices exploit a fundamental yet abstract and elusive property of ­electrons called spin. Because it ­underlies permanent magnetism, spin can be thought of as analogous to ­rotation, with a kind of nanoworld angular momentum. An electron’s spin is proportional to its magnetic momentum, and so when spin-polarized electrical currents flow through different types of magnetized metal, resistance changes can be exploited to store information.

Even more interesting would be a microprocessor using spin. In principle, a device that encoded information using the orientation of electrons could handle data thousands of times as fast as the present-day processors that rely only on charge. ”Instead of an electron being there or not there in the gate of a transistor—basically two pieces of information—think about an electron being able to hold a million pieces of information,” says David Awschalom, a physicist at the University of California at Santa Barbara who specializes in the development of magnetic semiconductors. In addition to being much faster, spintronics processors could be much smaller than present-day processors.

To incorporate spin technology into processors, however, researchers need to surmount the problem of making spin-polarized currents flow through semiconductors at room ­temperature. This has proved to be a perennial ­bugaboo because most semiconductor materials that have been tried function ideally at temperatures below -120 °C. But as surrounding temperatures rise, they lose their special magnetic properties, making them impractical for use in electronics and other consumer products.

A team led by Jagadeesh Moodera at MIT’s Francis Bitter Magnet Laboratory has inched closer to this goal by developing a magnetic ­material that can transmit spin currents ­without being chilled [see photograph, "Ambience"]. Consisting of indium oxide with trace amounts of magnetic chromium, the team's device rests atop a conventional silicon semiconductor and polarizes the spin of incoming electrons, which then flow directly into the chip. After traveling through the doped indium-oxide semiconductor, the spin-polarized electrons are read by a spin detector at the other end of the circuit, which determines the electrons' spin by accelerating them to high energies and scattering them (electrons of opposing spin states always scatter in different directions).

The indium-chromium mixture fulfills its function perfectly because when combined, these substances contain periodic ”gaps” in their molecular arrangement where oxygen atoms are missing. By modifying the character and extent of these gaps at the atomic level, Moodera can fine-tune the material's magnetic behavior to an unprecedented degree.

Stuart Wolf, a physicist at the University of Virginia in Charlottesville who has been involved in spintronics research for over a decade, describes Moodera's results as ”extremely encouraging....There have been similar reports of magnetism in other [semiconductor] substances, but the evidence reported in these earlier papers was not totally convincing,” he says. ”This work presents the most convincing evidence to date of high temperature magnetism in the oxides.” Despite Moodera's recent advances, however, Wolf cautions that considerable obstacles remain before spin-based processors can become a commercial reality.

The most pressing problem is a phenomenon called ”spin scattering,” which Moodera concedes is ”one of our biggest challenges.” Due to physical properties of the metal he used to build his electron-injecting device, the electrons' spin often changes slightly from the time they are injected to the time they are read by the spin detector. Such changes compromise the accuracy of any information that might be transmitted using this technique.

Moodera's quest for a material that will not have this effect is ongoing. Awschalom and other researchers are experimenting with laser-based techniques designed to compensate for this erroneous rotation.

Another problem: the materials used to manufacture spin-based chips and circuits are prohibitively expensive. A 4-megabit magnetoresistive RAM (MRAM) chip, for instance, costs US $25, while nonmagnetic RAM chips with the same capacity typically cost only around $5. ”The only way the cost is going to come down is if the volume of production goes up, and that's a relatively slow process,” Wolf says.

Awschalom likens spintronics to laser research several decades ago—its key players, he says, are on the verge of a breakthrough with future repercussions they can scarcely predict themselves. ”Today's technology is so inarguably successful that it's hard to imagine spintronics could do better,” he says. ”But if you could make computer processors a million times faster, with a hundred million times more memory, then you could make enormous impacts on every field—from pharmaceutical design to weather prediction.”

This article is for IEEE members only. Join IEEE to access our full archive.

Join the world’s largest professional organization devoted to engineering and applied sciences and get access to all of Spectrum’s articles, podcasts, and special reports. Learn more →

If you're already an IEEE member, please sign in to continue reading.

Membership includes:

  • Get unlimited access to IEEE Spectrum content
  • Follow your favorite topics to create a personalized feed of IEEE Spectrum content
  • Save Spectrum articles to read later
  • Network with other technology professionals
  • Establish a professional profile
  • Create a group to share and collaborate on projects
  • Discover IEEE events and activities
  • Join and participate in discussions