Although the future for polymer electronics looks bright, theres inevitably a cloud on the horizon. Its to do with size. While there would be some definite advantages in making computer chips out of p
Although the future for polymer electronics looks bright, theres inevitably a cloud on the horizon. Its to do with size. While there would be some definite advantages in making computer chips out of plastic, that change doesnt address the biggest problem currently facing silicon, which is that were rapidly approaching the limit for how small we can make circuit features without getting hampered by all sorts of quantum effects and other problems. Estimates of exactly when well hit this limit vary, but its likely to be within the next ten years. Whatever problems will arise then would apply also to polymer chips.
One possible solution, which was suggested back in 1974, would be to use individual molecules for the connectors, transistors and other elements of an integrated circuit. Moving electrons around within a single molecule would avoid some of the problems of trying to form clearly defined regions within a small area of crystal. It would also reduce the size of circuit elements by a factor of one million.
For many years, two main avenues of research dominated this field. One was based on using protein molecules to create computing circuits, the other relied on unusual organic crystals known as molecular metals. Although both looked promising, neither approach managed to make the big time. Once again though, polymers provided a solution. In 1996, James Tour of Rice University in Houston, Texas resurrected the idea of molecular computing by demonstrating that a polymer called polyphenylene could be used to make molecular wires. The following year, he managed to produce a molecular diode - an electronic device that allows current to flow through it in only one direction. Since then, Tour and other researchers have successfully managed to build molecular switching circuits, or gates, as well as molecular memories.
Polyphenylene is a fairly simple conjugated polymer with a backbone made up of linked carbon rings. The particular polymer compound that Tour uses is a specially fabricated variation called 2-amino-4-ethynylphenyl-4-ethynylphenyl-5-nitro-1-benzenethiol.
Far from being a second-best substitute for silicon, this molecular switch outperforms silicon in several ways. First, it shows a clearer difference between the current that flows when the switch is on and when its off. This difference, known as the peak-to-valley ratio, is about 1,000:1 in the molecular switch, but is normally less than 50:1 in silicon circuits. Not only is it much smaller than any switch that you could build in the solid state, says Tour, it has complementary properties, which in this case, if you want a large on/off ratio, it blows silicon away. More important though is that he believes the cost of molecular switches will be at least several thousand times lower than for traditional semiconductors.
One potential disadvantage is that the first molecular switches required very low temperatures of around -213C to operate. In the last few months though, molecular switches have been made to operate at room temperature. Admittedly, this was with a much lower peak-to-valley ratio, but this performance is bound to improve.
A much bigger problem is how to build complex circuits out of these molecular devices. One of the reasons why so much of the early effort went into researching protein-based computing was that proteins are good at self-assembly. All you had to do was to break a protein up into simpler subunits, put these into a test tube, shake the mixture and watch fully functional protein molecules self-assemble out of the soup. As it happened, this promising line of research didnt lead to practical computing devices, so the problem of assembly remains unsolved. Progress is being made - one of James Tours principle achievements was to find a way of bonding and positioning individual polyphenylene molecules on a rigid substrate. Even so, this is still a long way from being able to build anything as complex as a microprocessor, let alone do it on an industrial scale. Integrating these switches into a full-blown system where we need to address perhaps ten million of these devices remains a big challenge, says Tour. A big piece of the puzzle has been solved. Were looking at properties of single and very small packets of molecules, and now we need to learn how to string them together.
Other remaining problems include how to reduce overheating in these tiny and fragile molecular devices, and how to make them less vulnerable to faults. A conventional silicon chip may be vast by molecular standards, but at least that scale provides a built-in level of security - if the occasional silicon atom goes adrift, there are enough left to carry on working regardless. When you only have one molecule per function, it doesnt take much to produce a major failure.
In one form or another, this problem can only be solved by compromising on the miniaturisation and allowing a degree of redundancy. Even so, it should be possible to achieve a massive reduction in circuit size by switching to molecular designs, and Tour is hopeful that these molecular devices will find their way into computers quickly: It really looks like were going to have hybrid molecular- and silicon-based computers within five to ten years.
Even without the exotic technology of molecular computing, we may soon be able to fabricate plastic chips linked by plastic wires on a plastic circuit board.
The whole thing would drive a plastic display, be powered by a plastic battery and be packaged in a plastic case. Of course, it probably wont happen like that, but its a tempting prospect. Imagine a computer thats paper-thin, lightweight, waterproof and rolls up when youve finished using it. It would be a far cry from those polished wood panels that I bought just 20 years ago.