Why you should care
A quiet revolution is brewing in supercomputing these days, one that could solve our energy woes and our need for greater computational power, all with the spin of an electron.
“It costs a nickel every time you flip that switch,” or so Grandma used to say when the children kept flickering the lights.
Turns out she was on to something, and today the potential for wasting energy with our fingertips is exponential. With Internet usage and data storage needs soaring the world over, those five-cent flicks are downright laughable when you consider the millions of energy-consuming clicks.
Headlines point fingers at tech giants like Facebook and Microsoft for energy consumption at overheating data centers, but it’s really our personal demand for greater computational power that’s to blame. And as we start storing more on the cloud and less on home computers, those demands will rise.
It’s hard to pinpoint exactly how much energy is being poured into most data centers, says Steve Hone of the Data Centre Alliance, because for every commercial data center there are at least 100 private ones. “The government would say that their CO2 emissions are the same as the aviation industry,” Hone explains, with the “average commercial data center using the same amount of power as a small town.”
Data centers, which use semiconductor transistor technology, are believed to draw nearly 2 percent of the world’s energy, but what’s more concerning is the amount of energy many of them waste. Some surveys have shown that only 6 percent to 12 percent of the energy drawn is used for actual computing, meaning 90 percent is wasted. Sixty percent alone can be lost on heat. Some tech firms are moving toward greener energy alternatives, and some of the energy used is being repurposed, but the industry has to find a way to eliminate the waste.
Since few of us are willing to give up our Google searches or eBay purchases, it’s a good thing that scientists are on the case. Research has focused largely on two branches of science: superconducting and spintronics.
An electron is one of the elementary particles of an atom. Electrons carry a negative charge and orbit the atom’s nucleus with an angular momentum known as spin.
Superconducting supercomputing, where cold temperatures are leveraged to get metal to a state where it offers no resistance to electrical current, could be the holy grail for reducing power demand. When resistance is high, it takes more energy to get things done; the more resistance drops in a superconductor, the less energy you need to feed your supercomputer.
Scientists have been striving to unlock the full potential of superconductivity for decades. It involves paired electrons, called Cooper Pairs, that carry charge but whose antiparallel spins cancel one another out (in other words, the electrons’ spin isn’t used for energizing the device). In order for the energy-efficient supercurrents to run through them, they must be paired.
Until now, science didn’t think combining superconducting and spintronics was possible.
The relatively new field of spintronics, meanwhile, has scientists working to harness the electron’s inherent spin to control electronic devices and enable faster processing. Spintronics does not currently involve superconductors, but without these materials, spintronics suffers from dissipation, meaning energy is lost before it can do any work.
Neither superconducting nor spintronics on its own is likely to meet our rising computational needs in an energy-efficient way, and until now science didn’t think combining the fields was possible.
All that has changed, thanks to the work of Dr. Jason Robinson and his collaborator Professor Mark Blamire, both at the University of Cambridge, as well as colleagues around the world such as Dr. Norman Birge at Michigan State University and Professor Jan Aarts at Leiden University. They have found a potential way to marry the fields into the brave new world of energy-efficient superconducting spintronics.
The point is that they have found a way to make a superconducting current spin in order to potentially revolutionize computing.
“We’ve discovered something that’s fundamentally very exciting and transformative,” says Robinson. “That’s a huge adventure and a privilege from a scientist’s point of view.”
In 2010, Robinson and his colleagues validated a theory by Drs. F.S. Bergeret, A.F. Volkov and K.B. Efetov, and proved they could transform superconducting Cooper pairs into ones able to carry both charge and spin. And just recently, and more important, they’ve proved that they can distinguish between these Cooper pairs with up-up spins and those with down-down spins — which is essential to spintronics.
If the science has your head spinning like Robinson’s Cooper pairs, fear not: The point is that they have found a way to make a superconducting current spin in order to potentially revolutionize computing.
So where’s the drumroll? The red carpet? Robinson laughs in response, pointing out that few want to try to wrap their heads around the science.
“Scientists are extremely interested in our work,” Robinson offers; he’s been granted 10 years of funding by the Royal Society to drive the technology forward.
But industry has yet to show much interest. “Industry is always a bit lethargic to pick up on new ideas, partly because they have something that already works,” he explains.
That interest will come once scientists have a working prototype, which is the next stage in Robinson’s research. “If we can create a prototype device that does everything a spintronics device does, but without dissipation,” he says, ”then there would be a lot more interest from industry.”
The world has come a long way since Commodore 64s and Atari Pong championships. Our vast electronic world of iPads, smartphones, laptops, virtual first-person-shooter games and the cloud is continuing to expand, taxing the capabilities and cost of today’s data centers.
If technology is just a prototype away from potentially unlocking a future of guilt-free Web surfing, data collection and billions of mouse clicks processed at greener data centers, then it must be time for us to roll out the red carpet for Robinson’s trip back to his lab.
And even if he has to flick a few light switches along the way, Grandma would surely approve.