After Silicon Valley: Magnetic Spintronics Could Bring Next-Gen Computing
Spin electronics joins a variety of new techniques for moving around binary data that could move beyond traditional silicon computer circuits entirely.
For many decades now, silicon has been the undisputed heavyweight champ of the semiconductor industry. They named an entire valley after it, after all.
But new research out of the University of Denver suggests that things might play out quite differently over the next few years. If all goes well, the center of computer industry could potentially relocate over the Rockies to Yttrium Iron Garnet Valley. It doesn’t have quite the same ring to it, but that’s progress for you.
The work of physicist Barry Zink and his colleagues at University of Denver’s Zink Research Lab has the potential to serve as the foundation for the next generation of computing technology. The idea, simply put, is to replace silicon with “spintronics” (spin electronics), an evolving information processing technique that harnesses the magnetic properties of electrons, and which is quickly making inroads into traditional computer processing areas.
“Most computer technology is based on silicon, and most people know that,” Zink told Seeker, speaking from his lab in Colorado. “But they might not quite appreciate what the details are.”
At the core of virtually every modern computer are tiny circuits made out of crystal silicon. Electrons moving through those circuits is how computers transmit and store information in the binary language of ones and zeros.
“This has done fantastic things over the last four decades, but as you’ve probably heard it’s approaching the end of road,” Zink said, referring to the miniaturization limits of silicon microchips.
In recent decades, the most prevalent approach to improving computers has been to simply make stuff smaller. Moore’s Law, formulated in 1965 by Intel founder Gordon Moore, holds that computing power doubles every 18 months, largely because of the steady progression in the number of components that can be squeezed onto a silicon microchip. The smaller the circuit, the more you can cram into a device. But despite various stopgap measures, silicon’s capacity for computing power will eventually be exhausted.
“People are now having trouble really improving that technology,” Zink noted.
WATCH: Moore's Law Is Ending... So, What's Next?
Part of the problem is that moving around charged particles, like electrons, produces heat, which ultimately limits processing speed. Spintronics attempts to resolve this dilemma by using the inherent magnetism of electrons to move information, instead of firing off blistering electronic charges all the time.
The basic technology itself has actually been around for a while — most traditional hard drives use principles of spintronics to move information. There’s an inherent limitation, however. The materials needed to support this spin transport have to be pretty much structurally perfect to transmit the data accurately, but such materials are difficult and expensive to create.
“The big, crazy thing about this paper we’ve just published is that we’ve shown the material can be totally disordered and still support spin transfer,” Zink said. “It’s really exciting. At the atomic level, it means we don’t have to come up with materials in which the atoms are perfectly stacked in little rows.”
The discovery is significant, Zink said, because the synthetic material his lab is currently using — yttrium iron garnet — is easier to develop than growing the silicon crystals that are currently used in computer processors. The research team’s report — which was produced with funding from the National Science Foundation, the Center for Integrated Nanotechnologies, and the US Department of Energy — is published in the latest edition of the journal Nature Physics.
Zink is quick to note that this work is still in its very early stages, but if all goes well, the technology should be able to power all kinds of next-generation computing devices.
“We’ll get better and faster computers, for sure,” he remarked. “Phones too, since they’re really just smaller computers. Make the processors better and faster, you make the phones better and faster.”
Christopher Leighton, editor of the journal Physical Review Materials, said that the new research concerning yttrium iron garnet is of high interest to other researchers in the field.
“The material is well known in the magnetism community,” said Leighton, a professor of chemical engineering and materials science at University of Minnesota. “It’s relatively simple and cheap to make such a form of this material, which is part of the reason for the excitement about this work.”
Leighton said that spintronics is an extremely busy area of research just now, in that the technique can be used to link together two chips without flowing any electrical current between them.
“Interconnecting chips could be done in principle by flowing spin currents — magnetic currents — instead of typical electrical currents, so finding materials where spins can flow over long distances becomes important,” Leighton said.
Zink and Leighton both noted that spintronics is just one of several technologies competing to unseat silicon from its digital throne.
So-called “2D materials” like graphene represent the ultimate level of scaling down, in that they’re literally sheets of material that are a single atom in thickness.
“There are myriad 2D semiconductors now, some of which look very promising,” Leighton said. “Learning how to make them over large areas, with low densities of defects, will be important.”
Another contender is gallium nitride, a semiconducting material that is commonly used in LEDs, and which some industrialists championed as a potential overall replacement for silicon. The basic pitch is that gallium nitride does what silicon does, but better.
“Gallium nitride could replace silicon in a more conventional nanoelectronics technology, with more favorable parameters,” Zink said.
Then there are the more far-out technologies. Researchers around the globe are looking into relatively exotic applications involving carbon nanotubes, organic biochips, and the perpetual mind-bender that is quantum computing.
So what technology or material is most likely to replace silicon in the coming years?
“I would love to be able to give you that answer, but I’d be too busy running around investing in those companies,” Zink said with a laugh. “Tell you what — once I invest, then I’ll give you the answer.”
Leighton said that while there is not yet a clear winner, it’s an exciting time for basic research in material sciences.
“It’s a very interesting situation,” he said. “The industry really needs some next-generation technologies to come through. I don’t know what the answers are, but I know they will emerge in some way from basic research on materials.”
Zink also cautioned against being too quick to brand silicon as obsolote.
“It’s kind of like that scene from Monty Python where they’re bringing out the corpses and the guy says, ‘I’m not dead yet!’” he said. “We know so much about silicon and how make it. Or it may be that there won’t be just one thing anymore. All of these technologies may find their own individual roles. That would actually be great for engineers like me. It gives us more things to try to figure out.”