Spinning Out New Circuits
Magnetic semiconductors could lead to a future generation of computer logic
Photo: PDL Design/iStockphoto
15 April 2010—Tiny semiconductor dots could lead to a new type of circuit based on magnetism rather than current flow. At least that’s the hope of researchers who’ve made the dots and are hoping to build them into a workable device.
”We want to make it into a so-called nonvolatile transistor,” says Kang Wang, head of the Device Research Laboratory at the University of California, Los Angeles. Such a ”spintronic” transistor would retain its logic state in the absence of current and require less power to switch a bit, reducing the electrical power required by a computer chip by as much as 99 percent. Wang’s research, supported in part by Intel, was published in March in the online version of Nature Materials.
Where electronic transistors rely on the presence or absence of current to register the ones and zeros of digital logic, spintronic transistors depend on ”spin,” a quantum characteristic of the electron. Picture the electron as a rotating globe. When the north pole is pointing upward, that’s spin up; when pointing the other way, it’s spin down. When the spins of most electrons are aligned, the material is magnetic. When their spins are random, the material isn’t. An applied current can align or randomize the spins, allowing for spin-based switches.
To build spintronic circuits, engineers need a switchable magnetic material. So Wang and his colleagues made dome-shaped structures of 95 percent germanium with 5 percent manganese, a material known as a dilute magnetic semiconductor. Because the domes are just 30 nanometers across and 8 nm high, they confine electrons to quantum-scale dimensions, creating what are known as quantum dots. The researchers deposited the dots on silicon to make a complementary-metal-oxide semiconductor gate structure, then modulated the magnetism of the device by applying electric fields. ”We can now switch from nonmagnetic to magnetic,” Wang says.
When a negative bias is applied to the top gate, the density of holes—the positive counterpart of electrons—in the quantum dots increases. Even though the holes’ spin states are random, more holes will have one type of spin than the other, and because the holes are so tightly packed, the majority will impose their spin state on the minority, causing them all to align—that is, they impose ferromagnetic order. When a positive bias is applied, the density of holes declines until that order breaks down.
A practical device must work at room temperature or above. Many materials are ferromagnetic—retaining their magnetic state in the absence of an applied field—below certain temperatures but lose that characteristic when they warm up. The cutoff point is called the Curie temperature, measured in kelvins. Wang’s paper claims a Curie temperature in the material of over 400 kelvins, although his device worked only up to 100 K. ”We think we can go to 300 K [room temperature],” says Faxian Xiu, a postdoctoral researcher and the paper’s lead author.
Berend Jonker, head of the magnetoelectronic materials and devices section at the Naval Research Laboratory, made a similar device in 2002 using a thin film of manganese germanium and found that the Curie temperature of the material reached about 120 K. His group used a surface gate to control the hole density and turn the ferromagnetic order of the film on and off. He notes that using quantum dots rather than a thin film is a significant step forward. But he believes any higher-temperature ferromagnetism may be caused by nanosize clumps of material precipitating out of the compound. Such precipitates aren’t switchable and are therefore of no use for this application. ”The literature on ferromagnetic semiconductors over the past 10 years is full of claims of perfect samples with Curie temperatures well above 300 K, only to discover a year or so later that indeed there were precipitates that they did not previously detect,” Jonker says.
Xiu says the team has not ruled out the possibility of such precipitates but has further data, as yet unpublished, that make them believe they can reach room-temperature operation. Wang says it will also be a challenge to design a source and a drain, important components of a field-effect transistor, or FET. He hopes, however, to demonstrate a FET based on the quantum dots in the next few years. The International Technology Roadmap for Semiconductors calls for replacing traditional circuits with a new technology, perhaps based on spin, by 2020. Asked if they’ll meet the deadline, Wang and Xiu laugh. ”I don’t want to promise,” Wang says. ”I think we are working toward it. Put it that way.”
About the Author
Neil Savage writes about optoelectronics and other technologies, from Lowell, Mass. In the March 2010 issue of IEEE Spectrum, he reported on how engineers are working to lower the dose of CT scans while maintaining their diagnostic capability.