H-P Thinks New 'Memristor' Could
Have Big Impact on Data Storage
By DON CLARK
Researchers at Hewlett-Packard Co. say they've built a new element of
electronic circuitry that had previously existed only in theory.
The company says the new component, known as the memory resistor, or
"memristor," could have a big impact on the way data are stored in
computers. Among other things, the development could help future memory chips
store information for long periods without electrical current,
eliminate the slow process of booting up computers and also sharply reduce
power consumption, the company says.
The notion of a memristor was postulated in 1971 by Leon Chua, an
electrical engineering professor at the University of California at
Berkeley. A working prototype was developed by a team of scientists the tech
giant's H-P Labs unit, who are disclosing their work in a paper being
published Wednesday in the journal Nature.
H-P's development comes at a time when many companies are racing to
find technologies to succeed a kind of chip known as flash memory, which
is widely used to store data in products such as digital cameras and
music players. Existing flash-memory technology is expected to lose its
usefulness within the decade, as companies keep shrinking the dimensions
of chip circuitry to store more data.
Assuming memristors can be manufactured efficiently and combined with
other kinds of circuitry, the technology could store data in a smaller
and more energy-efficient form than flash memory or other technologies,
said R. Stanley Williams, H-P's lead researcher on the project and an
expert in the field known as nanotechnology. In addition, while flash
chips lose their data after a year or so, a memristor has an unusual
ability to permanently recall the amount of electrical charge that flows
through it, he said.
The researchers constructed a prototype by placing a microscopic film
of titanium dioxide between two electrodes and applying a charge. "The
memristor remembers because what's happening is you actually change the
atomic structure of the memristor as charge flows through it," Mr.
Williams said. "That's different from any other device."
But memristors will join a number of promising new memory technologies,
some of which companies have worked on for years. For example, a joint
venture called Numonyx that was recently formed by Intel Corp. and
STMicroelectronics NV is betting on a technology called phase-change
memory.
Edward Doller, chief technology officer of Numonyx, described the H-P
paper as an interesting explanation of the behavior of some materials,
but questioned whether it could have as broad an impact as technology
his company is pursuing. "It doesn't really sound like a blazing new
discovery in my mind," he said.
Mr. Chua, though, is enthusiastic about H-P's work. He had originally
broached the idea of memristors by using a series of equations showing
that in addition to the three known circuitry elements -- the capacitor,
resistor and inductor -- there had to be another one with the property
of memory. He once developed a prototype, but it was battery-powered
and too large to be practical.
"I'm really excited, not so much because of the commercial applications
now, but because to me, it's really a paradigm shift," Mr. Chua said.
"It will wake up a lot of people."
Mr. Williams said H-P plans to work with chipmakers to get memristors
into computers and hand-held devices. Wolfgang Porod, an engineering
professor at the University of Notre Dame who's been briefed on the
memristor paper, predicts it may take a decade before electronics companies
figure out how to incorporate the technology.
"At this point, we see only a glimmer that a new kind of device like
this may open new applications," Mr. Porod said.
Write to Don Clark at don.clark@wsj.com |
