Portland, Ore. -- A research group has announced a promising technique in spintronics that might be used in standard silicon chips in the near future. Spintronics combines today's charge-based data storage and communication with magnetic-based information using spin as the common coin.
The experimental group has managed to inject electrons with spin into silicon chips by virtue of a ferromagnetic semiconductor junction with silicon. "This breakthrough points the way to devices using spin injection into silicon as easily as it can be done in gallium arsenide today," said Igor Zutic, a professor at the State University of New York at Buffalo.
Spintronics marries electronics to magnetism by encoding electronic data with magnetic spin. "Tunable spin-tunnel contacts to silicon using a ferromagnetic material have been demonstrated by Byoung- Chul Min [a researcher at the University of Twente in the Netherlands working with his colleague Kazunari Motohashi at Sony Corp.]," said Zutic.
Theorists like Zutic recommend using techniques that have already been proved with gallium arsenide semiconductors to inject electrons with spin into silicon circuits. Zutic has created a cookbook of proposed techniques for spin injection and detection in silicon with colleagues Jaroslav Fabian at Germany's University of Regensburg and Steven Erwin at the U.S. Naval Research Laboratory.
Zutic predicts that his most promising technique, which he calls the spin-voltaic effect, will bring spintronics to standard silicon chips within a year. Zutic is cooperating with experimental scientists in Japan, who have already cast his spin-voltaic effect into experimental gallium arsenide chips. They plan to have silicon devices working by 2007.
"My theoretical work has been demonstrated experimentally in gallium arsenide by my collaborators at the Tokyo Institute of Technology," said Zutic. "Now scientists at the University of Tohoku (Sendai, Japan) are aiming to demonstrate similar structures in silicon devices by next year."
Using both polarities
Zutic calls his approach bipolar spintronics--because it uses spin carriers of both polarities (electrons and holes)--in contrast to unipolar devices like MRAMs.
Bipolar spintronics, Zutic maintains, exhibits larger current swings than unipolar, even when the very small bias voltages popular today are used. Together with the ease of manipulating minority carriers, bipolar spintronics enables active devices with controls not available in current charge-based-only electronics.
"The photovoltaic effect in solar cells enables light shining on a semiconductor to cause open-circuit voltage or short-circuit current, but if your carriers are polarized--meaning you add spin-degrees of freedom--then by changing the polarization of incoming light you can, for instance, change the direction of current flow in the semiconductor," said Zutic.
The spin-voltaic effect has already been demonstrated by Zutic's Japanese colleagues for the direct-bandgap semiconductor gallium arsenide (GaAs). The pro- cess is simplified for III-V optical materials like GaAs, because it is not necessary to incorporate additional ferromagnetic materials, since spin can be added with optical transport and resonance methods, such as using polarized light to generate spin. When polarized light illuminates a GaAs photovoltaic cell, the angular momentum is transferred from the photons to the electrons. Likewise, researchers can detect spin in GaAs by measuring the polarization of their light from recombination.
Silicon's indirect bandgap prevents researchers from using light to inject and detect spin-polarized electrons. Zutic proposes two solutions, which various groups are already attempting to realize in experimental chips. The most difficult method would be forming heterojunctions between silicon and direct-gap semiconductors such as gallium arsenide. Such junctions would enable optical methods in the gallium arsenide to inject and detect electron spin. Unfortunately, the lattice mismatch be- tween silicon and gallium arsenide is so great that it presents a substantial obstacle to realizing such junctions.
The easier method would be to add ferromagnets to silicon by doping the silicon semiconductors themselves with ferromagnetic materials. These ferromagnetic semiconductors would transfer magnetization to the spin polarization of the electrons passing through them--called elec- trical spin injection.
"We believe that these electrical means of injecting spin-polarized electrons into future silicon circuits could be even more efficient than the optical methods used with gallium arsenide today," said Zutic.
MRAMs already show how to electrically store magnetic spin in silicon semiconductors, and the junction recently demonstrated by the University of Twente and Sony shows how to electrically inject spin-polarized electrons into silicon. Zutic's cookbook should help improve those methods by allowing researchers to harness the spin-voltaic effect, as well as providing them with a blueprint for electrically detecting spin-polarized electrons. With storage, injection and detection of spin-polarized electrons, all the building blocks would be in place for a new era of spintronics devices, Zutic said.
"Until now, carrier-mediated ferromagnetism has only been done with III-V semiconductors such as GaAs," said Zutic. "But we think our methods can be successful with silicon too. We hope our experiments will lead to nonvolatile multifunctional silicon devices with tunable optical, electrical and magnetic properties." Zutic's proposals for spin injection and detection using the spin-voltaic effect could enable current amplification in spin transistors.
See related image