Beyond E-Ink: How about E-microprocessor?
Hi Frank,
Here's an intriguing article that has all sorts of interesting implications:
techreview.com
[[Note: The holes in the text below are image frames available at the Website.]]
Print Your Next PC
Forget billion-dollar fabs. If Joe Jacobson has his way, you may be printing
cheap semiconductor chips on your desktop.
By Stephen Mihm
As Joseph Jacobson is fond of pointing
out, for all the gains in semiconductor
chip performance over the past few
decades, a typical integrated
circuit—the brains behind your
computer—is still far too expensive for
most people on the planet. “Look at the
way [a chip] is made,” he says,
punching the air with one hand while
directing a PowerPoint presentation
with the other. Fabricating a
high-quality logic chip like Intel's
Pentium processor, he points out, takes
“two weeks, seven days a week, 24
hours a day. Chip fabrication facilities
like the ones that Intel has are a $1.6
billion tool. And there are very few people on the globe who can touch that tool.”
Jacobson's solution: a “desktop fab” able to print circuits directly on a substrate,
such as plastic, without the expense and hassle of a multibillion-dollar
manufacturing facility. Jacobson, head of the Printed PC Group at MIT's Media
Lab, has already managed to print rudimentary but working transistors using an
“ink” consisting of nanometer-sized semiconductor particles. “Our goal is to follow
the trajectory silicon took, and start printing processors with perhaps several
hundred transistors, moving to thousands and then more,” says Jacobson. “We
should be able to demonstrate a very simple processor in the next 12 to 18
months.” And he predicts that printed logic chips with the speed and power of a
Pentium could eventually be possible, making microchips available for a fraction of
the time and expense associated with conventional manufacturing.
If Jacobson's vision becomes reality, it could change everything in computer
hardware. Printed electronics could be cheap enough to find their way into
everything from “wallpaper” able to display changeable images to
custom-designed logic circuitry. A chip fab on every desktop could bring about
the day when individuals download the architecture of integrated circuits the way
they download software today. It could, in short, transform hardware
manufacturing much the way the “open-source” movement has changed how
software is written. Indeed, at his most visionary, Jacobson contends printed logic
could give rise to an open-source hardware movement where chips are
custom-designed via the Internet and printed by the consumer in about the same
time it takes to print out a Web page. You could, says Jacobson, “download the
chip design from the Web, tie in some modifications from some guy in India, and
boom—out comes the device.”
It's lunchtime in Jacobson's lab, a windowless room with tangles of colored cable
dangling from the walls and ceiling and a row of chemical hoods set along one
wall. Jacobson's enthusiasm is contagious, and the cramped lab is obviously where
he and his handful of students spend most of their time, even when they're eating.
“What we're interested in is ‘give me a piece of plastic and in a few seconds I'll
give you back a Pentium,’ or something of that complexity,” he says between
mouthfuls. “I'm serious about that. Not slower than a Pentium; indistinguishable
from a Pentium.”
Coming from almost anyone else, such a claim would be hard to swallow. But the
35-year-old associate professor has the credentials to deliver the goods. After all,
when Jacobson joined the Media Lab in 1996, his immediate ambition sounded
nearly as outlandish. “I wanted to have a display [screen] that could be printed,”
Jacobson recalls. “I wanted something that was incredibly inexpensive, something
that would look like ink on paper.” Something, in other words, like “electronic
paper.”
His solution was a riff on research conducted at Xerox Palo Alto Research Center
(PARC) in the 1970s, where researchers had created microscopic balls that were
black on top, white on the bottom. An electric charge determined which side of
the balls rotated upward. With some clever wiring, the balls could be made to
form letters and words. Jacobson and a handful of MIT undergrads pushed the
idea in new directions. Rather than making balls of two colors, they fabricated
millions of tiny microcapsules, each containing a liquid mixture of oil, dark dye and
tiny shards of white pigment. They then layered the material onto flexible plastic
and sandwiched it between transparent electrodes on top and bottom. Depending
on the charge applied, the white shards migrate toward the top or bottom of the
sphere, and when activated in concert, the electrodes can force the ink into
recognizable patterns.
The rest is the stuff of venture-startup legends. E Ink was formed in 1997 with
several of Jacobson's students at the helm, and has since raised nearly $55 million
in private financing, forming deals with the likes of Motorola and Hearst
Publishing. Media and pundits alike have proclaimed the technology as the end of
paper as we know it. But what got lost in all the buzz over electronic paper is that
you still need electronics to drive the pixels (the ink) of the displays. The
prototypes built thus far by E Ink continue to rely on traditional (read: not cheap)
silicon chips to control the display. To reap the full benefits of the technology, you
need cheap, flexible electronic circuitry. E Ink has recently partnered with Lucent
Technologies, whose researchers have been working on ways to print organic
transistors onto flexible plastic substrates. (The two companies hope to unveil a
working prototype of the technology this fall.)
Jacobson, however, has even larger ambitions. Not only does he want to print the
relatively simple electronic circuitry required to control a display screen, he wants
to go the next step and find a way to fabricate high-quality logic on the order of a
Pentium using similar printing methods. Not only would you be able to “print” your
screen; you could, in a sense, print the PC itself—or at least its essential circuitry.
Inorganic Solution
Making a chip as powerful as a
Pentium by traditional means is not an
easy feat. While semiconductor makers
like Intel have learned to make
transistors smaller and smaller over the
past few decades, squeezing vastly
more performance into the
microprocessors, the basic mechanics
of chip making haven't changed much.
The base material remains silicon, sliced
into thin wafers. An insulating layer of
silicon dioxide goes on top of the
wafer; a thin layer of “photoresist” (a
light-sensitive material) is deposited on
the silicon dioxide. Light beams project the pattern of the circuit onto the
photoresist through a stencil; the pattern is then etched out by acids or reactive
gases. Additional layers of silicon are added, “dopants” such as boron or arsenic
are put into the mix, and finally the transistors are linked by means of tiny aluminum
wires.
The resulting microchips are a marvel of engineering and are largely responsible for
fueling the Information Revolution. Using multibillion-dollar manufacturing plants,
Intel and others can now make transistors as small as a few hundred nanometers
across (a nanometer is a billionth of a meter), packing tens of millions of them on a
single chip. The downside is that the several hundred manufacturing steps take
upward of two weeks and require clean rooms hundreds or thousands of times
more pristine than your average laboratory.
Last fall, Jacobson and his student Brent Ridley described in the journal Science
the first printed inorganic transistors. Several other research groups, most notably
at Lucent's Bell Labs and Cambridge University in Britain, have also printed
transistors. These groups, however, are using organic polymers; such materials
could have great promise in the electronics required to make cheap, flexible
displays. But organic transistors appear to be inherently limited in computing
speed. Jacobson's big breakthrough is that he and his colleagues at the Media Lab
have created liquid suspensions of inorganic semiconductors—the same class of
materials used in your Pentium chip—so that they can be used in a printing
process. In other words, rather than carving logic into a solid piece of silicon,
Jacobson is simply printing it onto a substrate.
Jacobson's optimism is justified by his group's rapid advances in synthesizing
“semiconductor ink.” Under normal conditions, semiconducting materials such as
silicon, cadmium selenide and gallium arsenide form bulk crystals with melting
points well over 1000 C. Jacobson and his team, however, have found a way to
synthesize a solution of tiny “nanocrystals” of 100 atoms or less. This
semiconductor ink can be patterned or printed onto a variety of substrates,
including thin sheets of plastic, at temperatures under 300 C. The particles,
Jacobson notes, are small enough to form 200-nanometer structures—about the
scale of complex integrated circuits like Intel's Pentium chip.
The suspension of nanoparticles is so similar to conventional inks that Jacobson
and his co-workers are able to use an inkjet printer manufactured by Hitachi to
fabricate tiny machines called MEMS, or microelectromechanical systems.
MEMS, which are one of the fastest-growing new areas in materials technology
(see “May the Micro Force Be With You,” TR September/October 1999), are
typically made using many of the same arduous techniques used to fabricate
conventional silicon microchips. Using the inkjet printer, Jacobson and his students
have managed to fashion both a working thermal actuator and a linear-drive motor
with features on the order of 100 micrometers by simply depositing hundreds of
layers of ink. And they are able to form the tiny machines without a clean room
and at temperatures well under 300 C.
The group has also used the inkjet printer to produce much more intelligent
radio-frequency identification tags. Others are also working on such tags but are
relying on logic using organic transistors. Jacobson thinks that the faster logic
possible with inorganics can make his version of the tags far more intelligent,
allowing companies to track everything from expensive goods to the packages in a
supermarket. A radio signal detector could read the devices, update them and
integrate them into inventory systems. A person could walk into a supermarket,
pick up some items and walk out, and the money would be automatically tallied up
and deducted from his or her bank account—and from the supermarket's
inventory system.
Using printed circuitry like that is just the beginning. Because the computer logic is
printed, it can be put on the surface of almost anything: soup can labels, textiles,
soda cans. “You could add intelligence to almost anything you want,” claims Colin
Bulthaup, one of Jacobson's students. “One thing we want to do is build a digital
camera in a business card: everything embedded into the card itself. There's no
reason to have all these clunky silicon chips. You can pattern your semiconductor,
your photodetector—all the materials together—and integrate them into a single
device, one that is incredibly small, incredibly cheap and incredibly quick to
produce.”
Making such devices using an inkjet printer, however, is still a far cry from printing
high-quality logic circuits. That requires fabricating transistors and other electronic
components at the scale of a few hundred nanometers—the level of precision in a
Pentium chip. For that, Jacobson has made use of polymer stamps that don't look
all that different from the stamps used to certify documents. In one version, the
stamp has the architecture of the circuit in positive relief and is dipped in the
nanoparticle ink; the circuitry is then transferred by hand onto a substrate. Also
promising is a negative stamp that “embosses” a thin layer of ink previously
deposited onto a plastic surface. The stamp's features push aside the ink at certain
points, forming whatever feature is engraved on the stamp at resolutions of 200
nanometers.
Pentium Challenge
This is all a mighty attractive vision. But
can printed electronics actually
compete with multibillion-dollar fabs in
making the exacting circuitry needed for
high-quality logic? Sigurd Wagner, for
one, doesn't think so. A professor of
electrical engineering at Princeton
University, Wagner is also pursuing
research into printed inorganic logic,
but he sees its promise in cheap
electronics that can be used over large
surfaces, not in taking on high-quality
microprocessors.
His goal, says Wagner, “isn't competing
with integrated-circuit technology; it's to go into an area that traditional integrated
circuits can't handle." Attractive applications include wallpaper that acts like a giant
display screen, electronics woven into textiles—even “electronic skin” covering an
aircraft that is able to respond mechanically to changing conditions.
Jacobson agrees that the short-term payoff will come in producing the cheap,
flexible electronics that could make such applications possible. “There are a huge
number of applications for incredibly inexpensive, low-power disposable logic on
plastic substrates,” he says. And for now, Jacobson's printed circuits are better
suited for these uses. For one thing, they are still far too slow for advanced logic
applications; while Jacobson's inorganic transistors are an order of magnitude
faster than the printed organic transistors made by Lucent and other research
groups, they're still 100 times slower than the best inorganic transistors made from
conventional techniques.
But making tomorrow's Pentium-like chips on a desktop fab remains the twinkle in
Jacobson's eye. That will take increasing the speed of the printed inorganic logic.
It's “likely a several-year research project,” he says, “but we believe it's doable.”
It's just the type of challenge and hugely ambitious project that Jacobson relishes.
It is the type of project that makes you rethink the possibilities of a very familiar
object. With E Ink, he is giving a new twist to a very old invention—the printed
page. Rather than throw out the newspaper, Jacobson wants to preserve its
virtues while updating it for the information age. And now he's rethinking the
fabrication of integrated circuits. If Jacobson can make his visions of printed
circuitry practical, he could change the meaning of “hardware” and replace the
multibillion-dollar semiconductor fab with something not so different from the
stamps that have been around for thousands of years.
While the rest of the computing industry attempts to drive down hardware prices
through mass production of a few standardized chips, Jacobson is going in the
opposite direction, trying to make every person the master—and
manufacturer—of his or her own logic.
Stephen Mihm produces the Web version of The New York Times Magazine.
|
