Excellent Ramtron (RMTR) Article:
Talks about their technology. Dated from sometime last year.
They will post earnings this week.
forbes.com
A computer in every shirt collar?
By Neil Weinberg
WALK THE HALLWAY at Celis Semiconductor Corp., a ten-person chip designer in
Colorado Springs, Colo., and you get a tour of the computer memory business. On one
side chip engineers from Matsushita and Mitsubishi crouch over workstations in dimly lit
offices. Across the hall: rivals from Korea's Hyundai and France's STMicroelectronics.
A padlocked rack of routers and electronic firewalls keep everyone honest.
Samsung, the world's largest memory maker, Siemens, NEC, Fujitsu and Hitachi have
descended upon Colorado Springs lately. What's the draw? Tiny Celis, along with
nearby parent Symetrix and crosstown rival Ramtron, have turned the town into the
fountainhead for a technology that could be, for the next generation of portable
computers and telephones, what DRAMs are for today's PCs.
The new technology is known as ferroelectric random access memory, or FRAM?as
distinguished from DRAM (for dynamic random access memory). The technology is
already doing duty inside millions of cell phones. The allure: the potential to combine the
best features of other memory chips without the drawbacks?to work at blazing speeds
yet consume far less energy and last far longer.
FRAMs could let digital cameras shoot one photo after another continuously, instead of
having to wait several seconds to save each image. In electronic organizers, FRAMs
could replace today's memory chips, make do with much smaller batteries and eliminate
hard-disk backup.
FRAMs will catch on in a big way in the next few years, particularly in low-end
stuff?inside videogames, air bags and VCRs. But some sophisticated new devices are
coming, too. Millions of FRAM-powered smart cards are bound for automatic fare
collection on subways and buses in Washington, D.C., Shanghai and elsewhere. The
chips for the credit-card-size devices, supplied by Nasdaq-listed Ramtron, sip so little
power that they run entirely on a radio signal from a turnstile-mounted reader. The
FRAMs wake up, deduct the fare and send it back to the reader in a tenth of a second.
The new chips are so cheap and durable that they may soon win work in no-stop toll
booths and package-tracking tags?even as button-sized chips identifying a shirt's
owner and telling the dry cleaner how much starch to use.
FRAMs cost as little as 20 cents apiece, yet Fujitsu expects the FRAM market to grow
from near zero last year to $10 billion in 2002. The technology one day may even
challenge DRAM chips in computers, a $16-billion-a-year business?although not
anytime soon.
The crystals inside FRAMs have 100 times the capacity to hold an electric charge as the
silicon dioxide in a DRAM, so in some theoretical sense FRAMs ought to be able to
store information 100 times as densely. But production techniques have a long way to
go. The densest FRAM chips, due this summer, will hold only 4 million bits of data (4
megabits), laughably primitive compared with the 64-megabit DRAMs now in use.
This misses the point. FRAM's real promise isn't as an understudy for DRAMs but as
the centerpiece of new wireless gadgets. "Our goal isn't to replace DRAMs but to use
FRAMs to create new portable markets," says Carlos Mazure, who runs ferroelectric
development at Siemens.
To be sure, high-tech is littered with memory might-have-beens that fell victim to their
own glitches and to blinding improvements in DRAMs and disk storage (see sidebar).
But FRAM's inherent advantages are clear.
DRAMs run at blazing speeds, yet, as countless cursing PC users know, a power
shutdown results in amnesia. Memory loss doesn't happen for two other kinds of
chips?flash memory and Eeproms; they run everything from microwave ovens to
wristwatches. But that feature means they are especially poky at storing new data.
FRAMs, by contrast, are as speedy as DRAMs, yet, like flash memory, retain their data
when the power is off. And FRAMs do it better. They can store data for years without
consuming any electricity. Compared with flash and Eeproms, they can download data
thousands of times faster, yet use one-tenth the power and run 100,000 times longer
before wearing out. A FRAM could handle word-processing functions for years; flash
memory would wear out almost immediately.
Ferroelectrics have been around for a century. The name aside, they contain no iron;
they are ceramic crystals that, when zapped with an electrical charge, take on a positive
or negative electric polarity. Work on computer uses began in earnest in the 1950s, but
foiled research powerhouses IBM, Bell Labs and Westinghouse. In 1973 IBM scientists
declared FRAMs a dead-end.
Early prototypes quickly "fatigued"?they stopped holding a charge or data. They also
interacted badly with the foundation of all other memory chips: silicon.
The resurrection came in the early 1980s when Carlos Araujo, a professor at the
University of Colorado, began working with industry veterans Larry McMillan and
George Rohrer. They formed Ramtron in 1982 to develop a new design. It eliminated
the bad silicon interaction by making a FRAM sandwich?putting the FRAM between
two microscopic electrodes to separate it from the silicon base.
Araujo and McMillan left Ramtron and set up Symetrix in 1988 to work on the fatigue
problem. They set a deadline: Find a solution by year-end 1991 or shut down. (Rohrer
left in 1987; he died last year at 62.)
Four months before their self-destruct date, they found a solution: a silicon substitute that
could crunch data for ten years without fatiguing?SBT, or strontium bismuth tantalate.
Araujo showed the material to Matsushita engineers, who were so struck by its
rock-solid structure that they sent a second crew to double-check the results. In 1995
Matsushita paid $5 million for 10% of Symetrix.
Matsushita's early FRAM success came not in memories but in high-density filters in cell
phones to block unwanted microwave signals. FRAMs shrank the six-chip modules that
unscramble phone signals to one-fiftieth their previous size. Matsushita now controls
more than half the world market for the modules and will make 100 million this year.
While Symetrix and its offspring, Celis, work with SBT, Ramtron has invested $150
million over the years in an older ferroelectric memory technology known as lead
zirconium titanate, or PZT. Ramtron already has sold several million PZT chips for
devices such as wireless meter readers.
Now Ramtron and Symetrix both hope to push into bigger markets. "If we can
overcome the barriers of time and money," says Symetrix's Araujo, "you'll soon see
products like PDAs (personal digital assistants) with long battery lives that can always
be hooked to the Net. You'll be able to download a 35-page report at your dinner table
in seconds."
Your Palm Pilot may soon look like an antique pocket watch. |
