January 30, 2013
Purdue researchers designing devices that use 'electron spin' in computing platforms
Purdue researchers are part of a new center focusing on the development of future computing platforms based on new device technologies that use "electron spin" to represent information.
The Center for Spintronic Materials, Interfaces and Novel Architecture (C-SPIN) is one of six new centers funded by the Semiconductor Research Corporation and the Defense Advanced Research Projects Agency. C-SPIN, supported with a $29 million grant over five years, is led by the University of Minnesota.
The Purdue portion of the research is led by Kaushik Roy, the Roscoe H. George Professor of Electrical and Computer Engineering; Supriyo Datta, the Thomas Duncan Distinguished Professor of Electrical and Computer Engineering; and Anand Raghunathan, professor of electrical and computer engineering.
Purdue's role is to design and model the devices and architectures, or how devices are assembled into a functioning processor. The Purdue team will receive around $2.3 million for the research, which will involve about 10 graduate students.
Conventional computers use the presence and absence of an electric charge to represent ones and zeroes in a binary code needed to carry out computations. Spintronics, a new technology, uses the "spin state" of electrons to represent ones and zeros. Electrons surround an atom's nucleus in orbital "shells." In magnetic materials, unpaired electrons in the outermost orbitals are oriented in a certain way so that their "spin" is said to be either up or down.
Designing computers using spintronics will require a rethinking of devices, circuits, architecture and computing models, Roy says.
The new technology is not intended to replace conventional computer hardware, based on complementary metal-oxide-semiconductor transistors, or CMOS. Instead, the new technology is expected to work in conjunction with CMOS.
"There are certain things CMOS does very well, such as complex mathematical computations," Roy says. "However, there are other tasks CMOS has great difficulty with, such as facial recognition, which the human brain can easily accomplish."
One promising approach that the researchers are pursuing is to design electronic circuits that operate more like a human brain than conventional computers do. The Purdue researchers have proposed new concepts for transistor-like spintronic devices and shown that spintronic devices can mimic biological structures such as neurons and synapses.
"We're talking about a whole different class of computing, a brain-like processor," Roy says. Computers based on a "neuromorphic" computing model might be 10 to 100 times better at certain tasks such as handwriting and facial recognition compared with state-of-the-art CMOS technology.