The future of computing might just come down to the honeycomb.
Researchers have pushed graphene’s ability to carry information using the spin of electrons to record distances. The results mean that the material—composed of honeycomb-like sheets of carbon atoms—may be ideal for future devices that use spin instead of electrical current to perform computations and carry signals.
Spin is a quantum mechanical property of many particles that responds to magnetic fields and corresponds to an intrinsic angular momentum. It can be a useful binary signal, because in the presence of a magnetic field it can be oriented in either of two ways—either parallel or antiparallel to the magnetic field.
Harnessing spin is nothing new—it is already used over short, vertical distances to store and read data in hard disks. But many researchers hope to find a way to use spin over longer distances. That could pave the way for new kinds of devices that perform computations using much less power than existing CMOS devices.
Such “spintronic” devices would bear some structural similarity to the traditional transistor, with an input and an output—sort of like the transistor’s source and drain—connected by an electron-carrying channel. However, the signal involved is carried by the magnetization of the channel rather than the flow of current. In a spintronic device, the input and output areas are magnetic. When a voltage is applied at the input, it skews the distribution of electron-spin orientations in the channel so that they more or less reflect the magnetization of the input. In one recently proposed spintronics device, this spin signal can then be used to alter the magnetization of the device’s output.
But the success of any spintronic device hinges in large part on what happens to spins once they make it into the channel. Interactions with the atoms that form the channel material can randomize electron spins, causing the original signal to fade away before it ever reaches the output side of the device. Finding a channel material that can carry spin signals over transistor-scale distances with few losses—a material that exhibits a long “spin relaxation length”—is thus a key goal in spintronics research.
By that measure, graphene is shaping up to be a winner. The material showed promise when the first spin relaxation length studies were published in 2007. But in the last few months, graphene’s lead over alternatives seems to be widening.
In December at the IEEE International Electron Devices Meeting, in San Francisco, Yunfei Gao, a graduate student at Purdue University, in West Lafayette, Ind., presented results showing that graphene could carry spin signals with a decay length of 5 micrometers, the longest distance yet observed for a material at room temperature. Room temperature is key, because it’s the starting point for proving practical devices.
The configuration that did the trick was actually a stack of seven layers of graphene. That configuration seemed to cut down on some of the unwanted electrical effects that can crop up due to interactions with the material the graphene must rest on. But devices containing more than seven layers seemed to be at a disadvantage, because less spin signal made it to the layers of the device with the least amount of interaction, Gao says.
The 5-µm result is rivaled by recent findings from a group led by Bart van Wees, a physics professor at the University of Groningen, in the Netherlands. In two recent papers, van Wees’s team has shown they can get spin relaxation lengths of 4.5 µm by placing graphene on a relatively uncommon substrate—boron nitride—or by physically suspending the graphene between two terminals over an air gap.
Graphene, says van Wees, is especially promising because electrons zip through it and interact little with the electric fields generated by the materials’ relatively low-mass carbon atoms. “I think at room temperature, there are no other materials that have come close,” he says, adding that the spin relaxation lengths of both metals and silicon top out at about a micrometer at room temperature.
There is some indication that graphene could do even better than what the Purdue researchers found. In research published in Nature Physics in July, a team of U.S. and French scientists measured spin relaxation lengths of 100 µm when graphene was cooled to just a few degrees above absolute zero. Cooling slows atoms, cutting down on some of the processes that can scramble electron spins as the particles move through the material.
Theory predicts that graphene could exhibit a similar relaxation length at room temperature, but so far no one understands exactly why the experimental results fall short. “I think that is the fundamental question at the moment,” van Wees says.
It is possible, says Gao’s advisor, electrical and computer engineering professor Joerg Appenzeller, that despite graphene’s exceptionalism in the transport department, it may not be the best material for future spin devices. He says his team at Purdue is also exploring copper, which is easier to fabricate and to integrate with other components. But he notes that copper doesn’t have another key advantage of graphene: the ability to boost the spin signal by increasing the concentration of electrons moving through the material.
“These room-temperature results are promising for future devices,” says Kanji Yoh of Hokkaido University, in Japan, an expert in both graphene and spintronics. But he stresses that research into potential materials, as well as for the devices they might eventually be used to build, is at an early stage. “People are still struggling to make a working device, [to figure out] what is the best working structure.”