- Excerpt
NANOCOSM:
Nanotechnology and the Big Changes
coming from the Inconceivably Small
by William Illsey Atkinson
INTRODUCTION
The excerpt, below, is called "The Death of Digital Technology," and
features remarks by IBM's director of physical sciences, Dr. Thomas N.
Theis, about the limits of digital information storage and the coming
return to analog systems. This shocking admission highlights a battle
in nanoscience between the machinists, who foresee microscopic factories
and robots, and the naturalists, who use biomimicry to get viruses and
proteins to do their bidding.
NANOCOSM documents this philosophical divide, but mostly finds
unity in the world of the super small. Atkinson shows how scientific disciplines
are uniting at the nano level: biology, chemistry, physics, and engineering.
Also joining forces are the nations of the world; NANOCOSM features
contributions from the U.S.A., Japan, the U.K., Brussels, France, Canada,
Australia, among others.
Published by Amacom Books (a division of the American Management Association), NANOCOSM also
focuses on the evaluation and exploitation of this technology by entrepreneurs,
venture capitalists, and multinational firms in pharmaceuticals, information
technology, manufacturing, transportation, and other major industries.
The Death of Digital Technology
by William Illsey Atkinson
We now take you to Hall B of the DoubleTree Gateway in San Jose, California,
where Dr. Thomas N. Theis is announcing IBM's official position on nanotechnology.
Big Blue is constantly sensitive to an emerging commercial consensus,
whether among its customers or its competitors. Over the years it has displayed
a genius for running around to the front of an existing parade and taking
it over. IBM, Theis announces, has a goal: to remake its microtechnology
division into a nanotechnology division. Theis accepts the emerging standard:
For commercial as well as scientific purposes, the nanocosm ranges between
one and one hundred nanometers. That, he says, is the length at which size
really matters.
"Below 100 nanometers, the electron senses its quantum confinement and
regular electronics ceases to work," Theis says. He now plays that remarkable
Big Blue trump card: a depth of original research, applicable to nanotech,
that stretches back even before Eric Drexler coined the word, "nanotechnology." The
IBM milestones come rumbling out. The first scanning probe microscope,
the scanning tunneling microscope (or STM), invented at an IBM Europe basic
research lab in 1982. The AFM, or atomic force microscope, ditto: 1986.
Seizing, translating, and redepositing individual atoms: 1990. First intramolecular
logic circuit: 2001.
"Not how old some of this stuff is," Theis says. (The room is still and
silent; every eye in the audience is fixed on him.) "In fifteen or twenty
years, the science we are doing today will be just as commercially important
as those older discoveries are now." Nanotechnology, as Theis sees it,
is "a long-term game.... It can take decades for concepts to move from
lab science to products." But despite this time lag, it's an era that's
fabulously exciting: "There has never been, or will be, a time like this
in the whole history of science." Part of the reason for this is the ability
of physical science to slow or reverse its recent relative decline in federal
funding.
"I'm enormously enthusiastic," Theis says. "But
as people who are in some way involved in nanotechnology, everyone
here today must take steps to manage the hype. I'm worried that perception
and expectation are getting far ahead of reality. When bubbles burst,
there are tears.
"Not," he adds, "that I'm against bubbles." The room, located at the exact
financial, geographical, and spiritual center of Silicon Valley, erupts
into laughter. "It says much about our society," explains Theis when things
quiet down, "that it permits and encourages experimentation."
~ Micro and Nano Transistors ~
Theis proceeds to review IBM research.
The silicon transistor, he says, "is
already becoming a nanodevice." But silicon's unit hardware can't and
won't get much smaller. "We're at the dimensions where the devices won't
function if they shrink any more. Physics won't permit it."
As a possible alternative to silicon
transistors at the nanoscale, Theis shows us a slide of the FinFET,
a field effect transistor with a heat-radiating appendage sticking
out of it. This fin is only 20 nm (nanometers) thick. This qualifies
it as nanotech by Theis's accepted definition. In fact, nanotechnology
has already begun to creep into computer hardware almost unnoticed.
State-of-the-art heads for hard disk drives have layers that are laid
down on their surfaces with atomic-level control and tolerances of
only ten nanometers. Or consider "pixie dust" -- the engineers' nickname
for a one-atom-thick rhenium layer that, applied to a drive head, leads
to higher data retention at room temperature.
Theis's own lab has achieved data-storage densities of one terabit per
square inch. The device that makes this possible is the IBM Millipede,
which uses over a thousand individual AFM tips to simultaneously inscribe
ROM data on a hard substrate. Theis touts Millipede, as well as other technologies
such as standing-wave electronics that have the same function as Millipede,
as ushering in a new age.
"These things open up entirely new markets," he says. "They
aren't really about data-storage densities. They're about incredible
new things."
Tom Theis thinks that 1 terabit
per square inch is about the limit for data density that existing technology
can attain. Yet "existing technology" is
changing as we watch. "Things are accelerating," he says. "The newest
GameBoy has a faster central processing unit than a personal computer
with a Pentium IV chip. We're approaching the end of Moore's Law."
That "law," first propounded by
IT engineer Gordon Moore about forty years ago, states that computing
power per unit area doubles every eighteen months. Another way of stating
Moore's Law is that the cost of a given amount of computing power is
sliced in half every 18 months.
Moore's Law, Theis thinks, may have
another "ten, fifteen, even twenty
years yet to go. But silicon-based technology can't go on forever." In
other words, if Moore's Law is to hold, then at some point in the next
few years, something must replace silicon semiconductors. Theis thinks
that something is carbon-nanotube technology.
"These things are amazing," Theis says, referring to the hollow cylinders
of pure carbon, fifteen angstroms across, known as buckytubes. "Keep the
covalent bonds straight, and they conduct electrons like metals. Twist
the bonds, and they become semiconductors. But don't believe any claims
you hear about buckytubes revolutionizing information technology in a few
short years." For one thing, a buckytube transistor requires far more power
to modulate than a silicon-chip microtransistor. "Besides, you'd need ten-to-the-twelfth
carbon-based transistors on a single chip. That's a trillion -- one followed
by 12 zeros. To date, the record number for adjacent carbon-based nanotransistors
is all of two."
~ Self-Assembly ~
If miniaturization of components to the angstrom level is one goal of
nanotechnology, Theis tells us, another central goal is self-assembly.
It may be possible to mill, plane, and mold matter at the atomic level,
but it would be much more elegant to persuade nanoscale objects to put
themselves together.
"You need two kinds of information to build a snowflake," Theis suggests. "The
first is a tiny dust particle. This tells the impinging water molecules
how to minimize systemic energy. They use the dust particle as a matrix
on which to self-assemble. Yet even given this a priori condition, you
won't get any new self-assembled object unless ambient conditions are also
right. In other words, you also need environmental information.
"Right now at IBM Labs we're providing both informational sets and getting
honest-to-God self-assembly. We can create complex patterns and amazingly
regular arrays." "Self-assembly can occur with unusual metal alloys such
as silicon-germanium or iron-plutonium," Theis says. "But it need not be
limited to such exotic materials. Under the right circumstances, with necessary
information input both a priori and from the environment, many physical
systems will exhibit self-assembly. In fact, maybe most systems will."
Self-assembly is certainly not one
of IBM's core businesses, Theis admits. "Nonetheless,
this type of self-assembly process has been patented and looks very promising
for future manufacturing."
Theis's next remark makes my hands shake as I take note. This is it:
The Revolution. After sixty years of digital dominance, IT's central
paradigm is on the brink of reverting to a primitive, long-abandoned,
long-despised state. It is as if Big Blue's very hue is about to change
to Big Red. The bombshell: Digital may soon be going analog.
~ The Death of Digital ~
Theis sets off his H-bomb with a
dry theoretical query: How much digital information is necessary to
specify any given structure? "Our current
technology needs tens of gigabytes of data to specify a video file," Theis
intones. "Yet nature needs only three gigabytes to specify a human being." Here
he's referring to the 3,000,000,000 nucleotides that encode the human
genome. "Something is out of whack here. Obviously, our set of IT algorithms,
which is to say our whole conceptual understanding, is lacking. Under
current modes of storing data, to write a file specifying even a simple
living organism such as a paramecium would create a file that was unimaginably
huge. Yet nature does it effortlessly, and in less space than a pinpoint."
Yes, we can store and manipulate
data using digital electronics. But -- and here Tom Theis, staid R&D director in a company that defines staid,
holds clenched fists to the ceiling and shouts -- "We're just lousy at
it!"
Living things store information in 3-D at the atomic scale, something
of which humankind has only recently begun to dream. The simplest organism,
a virus that hardly meets the criteria for being alive, is billions of
times more complex than the most advanced IBM server.
"So!" he announces. "What's the conclusion?" The
future of computing, Theis hints strongly, is to depart from where
it's been these last five decades. It must escape from its digital
prison and compute as nature does: by analog means. Sooner or later,
probably sooner, ones and zeros will give way to computational values
that vary smoothly, with the steps between defining limits so tiny
they may as well not exist. We humans, who have carried the power of
natural brains to its greatest known limit, must go back to our own
source: nature. And nature, when she computes, does so using naturally
evolved analog techniques. Only we humans know from digital; it may
well turn out to be a passing fad.
Does anybody but me see what's going on here? Apparently not. The butt-kickers
and butt-kissers still lean forward, waiting for an opening to mock or
adore. But what I've just heard seems like Isaiah endorsing Baal, or George
W. Bush confessing he's an al Qaeda agent. So, retrospectively, here's
my take on Tom Theis's announcement.
A guy whose firm has for the last two generation been committed heart
and soul to digital data processing, has just publicly revealed his despair
with Ma Binary. He doesn't believe that digital computation will ever be
as good as living systems' analog/parallel computation, at least not at
most of the processes that really matter -- recognizing faces, creating
artwork, and the like. Simon Haykin came close to this concept at Purdue;
now I'm hearing it from one of the highest-ranking technical execs at IBM.
There's no future in digital. It won't wash. Ladies and gentlemen, place
your bets...elsewhere.
Shazam!
Copyright ©2003 by William Illsey Atkinson. All rights reserved.
Printed here with permission of the publisher, Amacom Books, http://www.amacombooks.org.
Please feel free to duplicate or distribute this file, as long as the
contents are not changed and this copyright notice is intact. Thank you.
