A flexible, transparent memory chip created by researchers at Rice University. Image: Tour Lab/Rice University |
Researchers at Rice University are designing transparent,
two-terminal, 3D computer memories on flexible sheets that show promise for
electronics and sophisticated heads-up displays.
The technique based on the switching properties of silicon
oxide, a breakthrough discovery by Rice in 2008, was reported in Nature Communications.
The Rice team led by chemist James Tour and physicist
Douglas Natelson is making highly transparent, nonvolatile resistive memory
devices based on the revelation that silicon oxide itself can be a switch. A
voltage run across a thin sheet of silicon oxide strips oxygen atoms away from
a channel 5 nm wide, turning it into conductive metallic silicon. With lower
voltages, the channel can then be broken and repaired repeatedly, over
thousands of cycles.
That channel can be read as a “1? or a “0,” which is a
switch, the basic unit of computer memories. At 5 nm, it shows promise to
extend Moore’s Law, which predicted computer circuitry will double in power
every two years. Current state-of-the-art electronics are made with 22-nm
circuits.
The research by Tour, Rice’s T.T. and W.F. Chao Chair in
Chemistry as well as a professor of mechanical engineering and materials
science and of computer science; lead author Jun Yao, a former graduate student
at Rice and now a post-doctoral researcher at Harvard; Jian Lin, a Rice
postdoctoral researcher, and their colleagues details memories that are 95%
transparent, made of silicon oxide and crossbar graphene terminals on flexible
plastic.
The Rice laboratory is making its devices with a working
yield of about 80%, “which is pretty good for a non-industrial lab,” Tour says.
“When you get these ideas into industries’ hands, they really sharpen it up
from there.”
Manufacturers who have been able to fit millions of bits on
small devices like flash memories now find themselves bumping against the
physical limits of their current architectures, which require three terminals
for each bit.
But the Rice unit, requiring only two terminals, makes it
far less complicated. It means arrays of two-terminal memories can be stacked
in 3D configurations, vastly increasing the amount of information a memory chip
might hold. Tour said his laboratory has also seen promise for making
multi-state memories that would further increase their capacity.
Yao’s discovery followed work at Rice on graphitic-based
memories in which researchers saw strips of graphite on a silicon oxide
substrate break and heal when voltage was applied. Yao suspected the underlying
silicon oxide was actually responsible, and he struggled to convince his laboratory
colleagues. “Jun quietly continued his work and stacked up evidence, eventually
building a working device with no graphite,” Tour says. “And still, others
said, ‘Oh, it was exogenous carbon in the system that did it!’ Then he built it
with no exposure to carbon on the chip.”
Yao’s paper detailing
the silicon oxide mechanism appeared in Scientific
Reports in January.
His revelation became the basis for the next-generation
memories being designed in Tour’s laboratory, where the team is building
memories out of silicon oxides sandwiched between graphene and attached to
plastic sheets. There’s not a speck of metal in the entire unit (with the exception
of leads attached to the graphene electrodes).
The marriage of silicon and graphene would extend the
long-recognized utility of the first and prove once and for all the value of
the second, long touted as a wonder material looking for a reason to be, Tour
said. He noted the devices not only show potential for radiation-hardened
devices – several built at Rice are now being evaluated at the International
Space Station—but also withstand heat up to about 700 C. That means they can be
mounted directly atop integrated processors with no ill effects.
The laboratory is also building crossbar memories with
embedded diodes to better manipulate control voltages, Tour says. “We’ve been
developing this slowly to understand the fundamental switching mechanisms,” he
says. “Industries have flown in and looked at it, but we’re doing basic science
here; we don’t package things nice and pretty, so what they see looks
rudimentary.
“But this is now transitioning into an applied system that
could well be taken up as a future memory system,” he says.
Source: Rice University