Researchers one step closer to 'bootless' computer
- 18 October, 2010 21:14
Physicists at the University of California at Riverside have made a breakthrough in developing a "spin computer," which would combine logic with nonvolatile memory, bypassing the need for computers to boot up.
The new transistor technology, which one lead scientist believes could become a reality in about five years, would reduce power consumption to the point where eventually computers, mobile phones and other electronic devices could remain on all the time.
The breakthrough came when scientists at UC Riverside successfully injected a spinning electron into a resistor material called graphene, which is essentially a very thin layer of graphite, just like you might find in a pencil. The graphene in this case is one-atom thick.
The process is known as "tunneling spin injection." It involves laying down an electron in the graphene, which then represents a bit of data. By injecting multiple bits into the graphene, they can not only be stored in a nonvolatile state (without the need of electricity), but the data can be used for computations in the graphene itself.
Top image shows flow of electrons (dotted line) when no insulator is used. Flow of electron spin polarization is greatly improved (bottom image) when a magnesium oxide insulator is used as shown. (Image credit: Kawakami lab, U.C. Riverside)
If successful, the researchers will have created a chip that removes the input/output (I/O) bottleneck created by the system bus between a computer's CPU and a mass storage device such as a hard drive or solid state drive (SSD), also known as the Von Neumann Bottleneck .
One of the project's lead scientists, Roland Kawakami, an associate professor of physics and astronomy at UC Riverside, said the clock speeds of chips made using tunneling spin injection would be "thousands of times" faster than today's processors.
One of the major hurdles that remains involves finding a lower-power method to coax electrons into being flipped by a magnetic field, turning them into bits representing zeros or ones. Currently, the graphene spin technology requires more power than DRAM or SRAM to work, Kawakami said.
"If you can lower the energy needed, then you could lower the size of the supporting circuitry," Kawakami said. "What we're working on is a whole new concept. This will essentially give memory some brains."
The researchers also need to build out the circuitry. That will be the job of electrical engineers.
Kawakami's team has used a semi-conductor laser to essentially free-up electrons so they can be polarized and given a directional orientation, called "spin."
The electrons can either "spin up" or "spin down" and allow for more data storage than is possible with current electronics, according to the university. Once the electrons are polarized, they remain in place for the life of the chip, which in the case of graphene is almost practically an eternity.
"So it's the type of memory that can be very fast and it can be very durable. You're moving atoms. There's not a large magnetic field," Kawakami said. "I'm one of those researchers that really cringes at the thought of saying this [new technology] can be useful. I think for us, maybe within five years we can get one device working."
Kawakami's team is working on the electrical spin injection from a ferromagnetic electrode into graphene, which to date has been inefficient, he said. The spin lifetime of the electrons are thousands of times shorter than they should be from a theoretical perspective. "We would like longer spin lifetimes because the longer the lifetime, the more computational operations you can do," he said.
Kawakami's team has been able to lengthen the spin lifetime through the use of a nanometer-thick insulating layer, known as a "tunnel barrier," in between the ferromagnetic electrode and the graphene layer. They found that the spin injection efficiency increased dramatically, he said.
"We found a 30-fold increase in the efficiency of how spins were being injected by quantum tunneling across the insulator and into graphene," Kawakami said.
Kawakami said the research on spin computing is at a stage similar to that of the movement from vacuum tubes to transistors in the 1950s. Once one transistor was created, the floodgates to the modern computers were opened. Once a spin computing transistor has been created, in say about five years, he expects industrial support to ramp up and consumer products to follow in under 10 years.
Kawakami's team of three graduate-student researchers has for the first time joined with electrical engineers at the university, who are designing the circuitry that will carry the electrons through the graphene.
Graphene received broad notoriety earlier this month when the scientists who discovered its properties as the thinnest and strongest material known to mankind received the Nobel Prize in physics . Graphene is made up of carbon atoms and looks like chicken wire or lattice through an electron microscope.
To date, development of spin electronics has been geared entirely toward memory. Two years ago, another group of researchers at Rice University demonstrated a data storage medium made out of a layer of graphite only 10 atoms thick.
That technology has the potential to provide many times the capacity of current NAND flash memory, and can withstand temperatures of 200 degrees Celsius and radiation that would make solid-state disk memory disintegrate. That technology, for example, would be useful in satellites, which are constantly bombarded by the sun's radiation.
But the researchers, focused on combining the memory aspects with the computational capabilities of tunneling spin injection, are hopeful now that the right material is on hand.
"Things that have been missing are that the right material [graphene] hasn't been there on one hand, and on the other hand, the circuit computing design concept hasn't been there. It's like the chicken and the egg. One has to develop to give motivation to the other," Kawakami said.