SWIMMING serenely through the freezing waters of the Antarctic, the winter
flounder is unaware of the interest it is provoking among scientists in cosy
labs on the other side of the world. It goes quietly about its business in seas
that ought to turn the fluid in its blood to ice. Yet the flounder doesn鈥檛
freeze.
The secret of the fish鈥檚 survival is a tiny antifreeze protein: as the
disordered water molecules in its blood begin to turn into the ordered structure
of ice, the protein moves in to break them apart.
Just as the flounder can hold molecular order at bay in its bloodstream,
researchers are learning to map and control the boundaries between ordered and
disordered materials. Eventually, they hope this will help them devise better
ways to preserve frozen foods or design materials with remarkable properties to
order.
Advertisement
But first the researchers aim to marry the processing power of silicon with
the flexibility of plastics to create cheap, bendy and disposable electronics.
This sounds simple, but at the molecular level these materials are complete
opposites. Plastics are a chaotic tangle of polymer chains. Silicon, meanwhile,
consists of neatly ordered crystals. To understand their properties when joined
together requires an appreciation of what happens when order meets
chaos鈥攚hich is exactly what occurs in a flounder鈥檚 blood.
If researchers can make the marriage work, it could lead to drug packaging
that sports unbreakable polymer display screens, for example. Touch the screen
and health information or medical advice could be at your fingertips. And there
would be countless uses for computer screens that could be worn on a sleeve or
rolled up and stuffed in a pocket鈥攁n electronic map for tourists, for
example, or a complete set of wiring diagrams for an engineer working at a
remote location.
But it needn鈥檛 stop with displays: this plastic-silicon combination could be
used to make speakers, microphones or a new generation of sensors to spot
pollutants or diagnose disease. Since we already know how to mass-produce
plastics and silicon chips, it should be a small step to making electronics as
cheap and disposable as burger wrappers.
Conventional electronic devices are about as rigid as they come. Most laptop
computer screens, for instance, are built from a layer of silicon sandwiched
between glass and liquid crystal polymers. The glass, which is about 100
micrometres thick, acts as an insulator and as a support layer. The silicon
layer above it is etched with thousands of tiny transistors that turn the
individual elements鈥攑ixels鈥攊n the polymer layer on and off. The
thick glass layer rules out any possibility of folding or bending these
devices.
The obvious way to build flexible electronics would be to switch to an
all-polymer structure. Researchers have already manufactured transistors from
thin layers of semiconducting polymers (快猫短视频, 24 September
1994, p 5). The problem with this approach is that electrons travel more slowly
in these polymers than in silicon. So electronic devices built from polymers run
hundreds of times more slowly than their silicon equivalents, limiting their
processing power.
But there is a way to get the best of both worlds. Researchers at Cornell
University in Ithaca, New York, and the Lawrence Livermore National Laboratory
near San Francisco have collaborated to create a prototype electronic device
built from silicon and polymers. Their trick is simple: replace the glass
support with a layer of plastic and make the silicon on top so thin that the
whole electronic device is flexible.
To do this, they deposit a layer of silicon circuitry just 50 nanometres
thick onto a flexible polyester support. 鈥淲e put such a thin layer of silicon on
it that it remained flexible,鈥� says Michael Thompson, a physicist at Cornell.
The result is a high-speed electronic device that can be bent like thin plastic.
Best of all, it offers manufacturers the ability to make large displays cheaply,
since they can use a manufacturing process much like that used in the printing
industry to run off thousands of metres of displays at a time. These sheets
could then be cut to size.
In fact, the Livermore researchers have so much faith in the future of
polymer-silicon devices that they left their jobs last September to
establish FlexICs, a company that, they hope, will mass-produce flexible
circuits. At the same time, the US National Science Foundation has awarded
$1.7 million to the Cornell researchers鈥攍ed by chemical engineer
Paulette Clancy鈥攆or research to discover what happens when polymers meet
silicon.
Thompson is still not quite sure how they managed to create a prototype
device. 鈥淲e played around with the variables and eventually found something that
works. We don鈥檛 know why it works.鈥� But he hopes their research will provide
some answers.
One of the biggest problems they had to overcome concerns the manufacturing
stage: how do you process two materials that are so different? 鈥淚f you鈥檙e making
a traditional semiconductor device there are something like 200 processing
steps,鈥� says Clancy. 鈥淨uite a few of them involve sticking the chip in an oven.
Do that to a polymer and it melts.鈥�
These high-temperature processing steps are vital. To make an array of
silicon transistors, a layer of amorphous silicon is first applied to a support.
But amorphous silicon is no good for building high-speed electronic devices
since electrons can鈥檛 move quickly through its random arrangement of atoms. So
it must be converted into its crystalline form before etching. The simplest way
to do this is to heat the silicon to about 800 掳C. As it cools, its atoms
align to form tiny crystals.
The crystalline silicon must also be 鈥渄oped鈥� with impurities if it is to be
used in transistors. Doping silicon with atoms such as boron provides extra
current-carrying electrons or holes which are needed to transport signals.
Again, the best way to get these dopants into the silicon is to heat the
material to high temperature.
But plastics such as polyester melt at about 120 掳C, so the researchers
had to find a way to heat the silicon while keeping the polymer layer cool. The
answer, they discovered last year, is short pulses of light from a high-power
laser.
When they beamed pulses of ultraviolet light lasting just 35 billionths of a
second onto a slab of polyester coated with a layer of silicon 40 nanometres
thick, they found that most of the light was absorbed in the top 10 nanometres
of the silicon. Although the silicon momentarily reached temperatures of about
1000 掳C, a thin layer of silicon oxide between the silicon and the polyester
acted as an efficient thermal barrier so that the polyester barely felt the heat
from the laser. The researchers discovered that the pulsed laser beam can also
be used to drive the dopant atoms into the silicon. The result of their
experiments was a flexible array of hundreds of silicon switches about 10
centimetres across.
To build devices such as interactive screens, the researchers plan to deposit
layers of polymer pixels on top of the silicon. Each pixel will then be
controlled by the transistors beneath it. Alternatively, the polymer layer could
be touch sensitive, or contain sensors designed to pick up traces of chemical
pollutants. When triggered, these sensors would generate electrical signals and
pass them on to the transistors for processing (see Diagram).
Although the researchers have solved some problems, there are more
fundamental difficulties ahead. If these devices are to work efficiently,
current should flow unimpeded from the silicon to the polymer and back again.
While current flows smoothly and quickly through crystalline silicon, to an
electron the polymer layer looks like an assault course. Polymer chains
generally form a tangled, spaghetti-like mess with large gaps between the
chains, and electrons find this difficult to negotiate鈥攖heir mobility in
the polymers is roughly 200 times lower than in silicon. Even crossing the
boundary from the polymer to the silicon beyond is a challenge. Put simply,
silicon-polymer electronics is a marriage made in hell.
So which materials make the best connections? Clancy believes that improving
relations between silicon and polymers involves understanding the boundary where
chaos meets order鈥攁 problem that the flounder encounters every day.
No one is entirely sure how the flounder鈥檚 antifreeze proteins work. Somehow,
they seem to spot where an ice crystal is forming and surround it鈥攁cting
as a barrier between the ordered ice and disorderly water molecules. It鈥檚 not
clear how the proteins position themselves correctly, because the ice-water
interface is blurred and difficult to define. So Clancy has spent years studying
the process in detail.
Molecule by molecule, she has watched ice crystals grow in the fish blood,
then drop away, then grow again. She has used the data to develop a computer
simulation which models how the protein suppresses crystal growth. Now she is
planning to apply her modelling expertise to mapping the order and disorder
present when unruly polymers are integrated with nicely ordered crystals of
silicon.
鈥淭he ultimate goal is to put a mathematical formalism around what constitutes
order,鈥� Clancy says. With computer modelling, she hopes to be able to simulate a
section of polymer film, for example, and get the program to characterise it
with an 鈥渙rder parameter鈥�. That should give some substance to predictions of its
properties鈥攁nd tell researchers how well it will perform as an electronic
device.
But what does 鈥渙rder鈥� mean? How do you simulate something that is neither
fully ordered or disordered? 鈥淲hen you鈥檙e in a crystal, or deep in a liquid,
those order parameters are very straightforward,鈥� Clancy says. Here, school-book
definitions apply: solids have a regular structure, with atoms or molecules in
well-defined, predictable positions. Liquids, on the other hand, contain
molecules that are randomly arranged and oriented.
However, these definitions don鈥檛 always apply. Polyethylene, for example,
contains both regions of order and disorder. 鈥淚n reality you have amorphous
domains and crystalline domains that are interconnected by some phases that are
neither crystalline nor amorphous,鈥� says Fernando Escobedo, the group鈥檚 polymer
modelling specialist. 鈥淚t鈥檚 likely that those regions might be important in the
properties of the polymer.鈥� And polycrystalline silicon is made from randomly
arranged chunks of well-ordered crystals鈥攐n one scale there鈥檚 order, on
another there鈥檚 disorder.
So the perfect simulation should work like a microscope: it should be able to
look at the electrons in the individual atoms. But it also has to look at the
material as a whole鈥攊ts mechanical properties and its response to heat,
for example. And if you want the materials to stick together, exchange electrons
and live happily ever after, the simulation must also model the interface where
they meet.
Mighty challenge
That鈥檚 an enormous amount to ask from any simulation. For instance, when
researchers have simulated polymer structures in the past, they have used simple
classical models to describe the way chains form. These models look at how
polymers bend, or map the bond angles between the molecules in the chain. Until
now, there has been no way to discover how quantum mechanics controls variables
such as conductivity. 鈥淭he computational effort of using anything other than a
classical function to model the forces between particles would be too much,鈥�
Clancy explains.
But with semiconductors like silicon, she says, it鈥檚 a different story.
鈥淭hey鈥檙e invariably modelled using a quantum mechanical approach.鈥� That鈥檚
because they are used for electronics, so people are interested in the energies
needed to move electrons through the material. In other words, says Clancy,
鈥渋t鈥檚 hard to describe the structure of polymers and silicon in the same way.鈥�
For each of the two types of materials, researchers have traditionally gone down
separate modelling paths: there is no model that can do both.
So Clancy has turned to Cornell colleague Michael Teter, who has developed
his own computer simulation of the behaviour of atoms and electrons that
includes the quantum mechanical subtleties, all without tying the computer in
knots.
Teter鈥檚 idea makes use of careful approximations and averaging tricks (
快猫短视频, 30 January 1999, p 34). 鈥淭here are some details that you
don鈥檛 need to look really hard at,鈥� he says, 鈥渁s long as you get their averages
accurately.鈥� The main property that Teter approximates is an atom鈥檚 electron
density.
With this approach, he has created a simulation that can handle hundreds of
atoms at a time, is a thousand times faster than traditional versions, yet no
less accurate. 鈥淧roblems that would normally run for a couple of years are down
to a day or two,鈥� he says. Teter can take any of the first 18 elements of the
Periodic Table, bring them together and watch them react to form new materials.
As luck would have it, these elements include silicon, carbon, hydrogen, oxygen
and nitrogen: all components of the new, polymer-silicon electronics.
With the development of his simulation almost complete, Teter and Clancy are
about to unleash it on the polymer-silicon combination. If all goes according to
plan, a whole new world of flexible electronics could unfold before them.
Others are unconvinced. Richard Friend of Cambridge University鈥檚 Cavendish
Laboratory believes that the polymer-silicon marriage is a diversion from
the real future of flexible electronics: all-plastic (快猫短视频,
10 July 1999, p 38). But Clancy suspects it is too soon to predict which
technology will win. 鈥淚t鈥檚 really about who is going to have the throughput, the
reliability, the flexibility and the manufacturability,鈥� she says.
But, she adds, this is about more than a dominant new technology. At the very
least, the project will teach researchers an enormous amount about how to model
order and disorder in the chemical universe. Manage this and they could discover
better methods to preserve frozen foods and medicines鈥攖o stop strawberries
turning soft when defrosted, for example, and to protect delicate vaccines
during storage鈥攐r find ways to design completely new materials to order.
Winning the technological race is small fry compared with the advances the
project could bring.