Future Tech: Computing With a Twist
Engineers begin to tap into the power of electron spin
By Neil Savage

Every appliance, from your toaster to your laptop computer, relies on a single aspect of subatomic physics: the negative charge of the electron. Charge is what makes electrical current flow through a maze of wires to do useful things, such as activating a heating element or encoding data. But another property of the electron, called spin, could greatly expand the particle's usefulness. Moving far beyond today's electronics, the emerging technology of spintronics may soon make it possible to store movies on a PalmPilot or build a radical new kind of computer.

IBM's spin-based M-RAM chip stores data without drawing power. Good-bye to dead batteries and long computer start-ups? Photograph courtesy of IBM

The principle behind this trickery is deceptively simple. Ignoring for a moment the weirdness of the quantum world, the electron can be thought of as a tiny rotating bar magnet with two possible orientations: spin-up or spin-down. Engineers can distinguish between spin-up and spin-down electrons by the corresponding orientation of their magnetic fields, north-up or north-down. Conversely, a properly applied magnetic field can flip electrons from one state to the other. In this way, spin can be measured and manipulated to represent the 0's and 1's of digital programming, analogous to the "current on" and "current off" states in a conventional silicon chip.

Broadly defined, the first spin-related technology was the compass, a piece of metal in which electron spins are mostly pointing in the same direction to generate a magnetic field. This field, in turn, attempts to align itself with Earth's magnetic pole. "We have used spin forever. Magnetism arises from the fact that electrons carry spin," says Sankar Das Sarma, a physics professor who heads the spintronics group at the University of Maryland in College Park. But exploiting the magnetic properties of the electron doesn't really qualify as spintronics, he says, until you start deliberately flipping the particle's spin back and forth and moving it from one material to another.

The first major breakthroughs in full-fledged spintronics came at IBM's Almaden Research Center about a decade ago, when materials scientists set out to find ways to cram more data onto computer hard drives. A hard drive uses an electrical charge to place tiny patches of magnetic field in the recording material; it then reads back the encoded data by measuring which way the field points at different locations.

The IBM project latched on to the work of two European scientific teams who had discovered a spin-related effect known as giant magnetoresistance in 1988. Starting with a magnetic material whose spins were all locked in one direction, the researchers had added a thin layer of metal and topped it off with another material in which the spins can flip. Current flowed easily from the top to the bottom of this composite if the spins were the same in both layers, but the current faced higher resistance if the spins were opposed. In theory, such a setup allowed a much more sensitive way to read back the data on a magnetic disk, but giant magnetoresistance seemed to occur only in expensive, pure crystals exposed to intense magnetic fields.

By 1991, the Almaden team found it could achieve the same effect in cheaper materials that responded to much weaker fields. The researchers eventually built a magnetic read head composed of one of these spintronic sandwiches. Magnetized patches on the spinning hard disk flip the spin state in the read head back and forth, transmitting digital data. A spintronic read head can detect much weaker magnetic fields than older devices can, so each bit of data can be much smaller. "It's the world's most sensitive detector of magnetic fields at room temperature," says Stuart Parkin of Almaden. Spintronics is why today's hard drives hold up to 100 gigabytes or more, compared to less than 1 gigabyte five years ago.

Fast laser pulses control an electron's spin. Peak heights denote how strongly the particle is tipped by each pulse. Photograph courtesy of David D. Awschalom/University of California at Santa Barbara

Now Parkin, along with researchers at Honeywell, Motorola, and the Naval Research Laboratory, is trying to create spin-based computer memory, called magnetic random access memory, or M-RAM, based on the same principles. A prototype design contains a series of tiny magnetic sandwiches placed on a silicon chip between crisscrossing arrays of wires. Electric current through the wires flips the spin, which stays put until it is changed again. Measuring the electrical resistance of a particular sandwich tells whether it represents a 1 or a 0.

In conventional desktop computers, random access memory-information that is available only while the device is turned on-is refreshed 60 times a second by a surge of electricity. M-RAM, in contrast, has almost no electrical demands. NASA is intrigued, because M-RAM could make it possible to build longer-lived spacecraft that perform more elaborate functions without requiring additional power. In more down-to-earth applications, M-RAM might lead to instant-on computers and cell phones with so much built-in memory that they could store entire conversations. "You could do all sorts of things that you can't do today, like have video on your PDA," says Parkin, who expects that IBM will be selling M-RAM by 2004.

Further ahead, spintronics could realize a long-sought, radical kind of data crunching known as quantum computing. According to the laws of quantum mechanics, an electron can be in both spin-up and spin-down states at the same time. That mixed state could form the base of a computer built around not binary bits but the quantum bit, or qubit. "It's not just a 1 or a 0 but any combination of a 1 and a 0. It's one of the first truly revolutionary concepts for computing that's come along in a long time," says David Awschalom, director of the Center for Spintronics and Quantum Computation at the University of California at Santa Barbara. Feed a problem into a quantum computer and instead of trying all possible results one at a time, it could calculate them all simultaneously. Barring any unforeseen breakthroughs, however, Das Sarma thinks it will be at least 50 years before anybody builds a quantum computer.

Long before then, the benefits of spintronics may spill over to other areas of electronics. Earlier this year, Awschalom and his colleagues at the University of California at Santa Barbara and Pennsylvania State University demonstrated that they could drag a cloud of electrons from one semiconductor material to another without disrupting the spin state of the cloud. This achievement points the way toward spin-mediated versions of transistors, the on-off switches that form the building blocks of just about every device powered by a battery or plugged into a wall outlet. "We were as surprised as anyone that it worked so well," Awschalom says.

Spintronics transistors might lead to faster, smaller, less-power-hungry versions of existing devices, but Awschalom also has a grander vision: "New science enables new technologies. And I think the most exciting ones will be things we haven't even imagined yet."

Sankar Das Sarma's group at the University of Maryland has written a helpful overview of spintronics, with links to the group's current work. See www.physics.umd.edu/ rgroups/spin/intro.html. David Awschalom of the University of California at Santa Barbara also has an extensive Web site: www.qi.ucsb.edu/awsch.

IBM developed the first practical spintronics devices, which use a physical principle called giant magnetoresistance. There's an online tutorial complete with animations at www.research.ibm.com/research/gmr.html.

If you really want to understand quantum computing, a good place to start is the University of Oxford's Center for Quantum Computation: www.qubit.org.