Why silicon computers rule

Media reports often predict the silicon chip will soon be outdated - but the claims are far ahead of reality. Alan Finkel explains.

A 45-nanometre process 300 mm silicon wafer from Intel Japan. The processors have improved power efficiency and performance compared with the preceding 65-nanometre processors. – YOSHIKAZU TSUNO / AFP / Getty Images

Judging from reports in the media, you might think the silicon chip is passé. Computers are about to be revolutionised by quantum chips, optical chips or chips made from graphene or even DNA!

Sorry to pour cold water on these dazzling ideas, but I don’t think silicon will be replaced any time soon. A recent article in Scientific American claimed “in optical computing, electrons do not carry information, photons do, and they do so far faster, at the speed of light”. That certainly makes it sound like it’s time to ditch the electrons in favour of photons.

But not so fast. Actually, photons and electrons end up transmitting information at the same speed. Please stay tuned for a short tutorial.

The speed of photons in optical fibres is the speed of light in glass, which is about two thirds of the speed of light in a vacuum. That’s impressively fast.

In metals such as copper, things are different. Instead of travelling at the speed of light the electrons vibrate rapidly about a stationary spot. When a voltage is applied from one end of the wire to the other the electrons start moving, but with extraordinary slowness. The aptly named “drift velocity” is less than one tenth of one millimetre per second. That’s slower than the rate honey oozes from a teaspoon into a cup of tea.

Within silicon chips, metals such as copper or gold are used to connect various groups of silicon transistors to each other. So how can metal wires inside computers carry information rapidly from point to point?

They are able to do so because the speed of the electrons is not relevant – the relevant speed is the voltage wave that carries the signal along. Just as boats bob up and down as an ocean wave passes by, so too with the voltage wave and the electrons.

To conceptualise the voltage wave at work, think of two people having a telephone conversation in different cities. Their phone lines are connected all the way by copper wires. When the first person speaks, the microphone converts the sound into an electrical voltage. This voltage creates an electric field that pushes the closest electron, which pushes its neighbour, and so on all the way to the listener in the other city. Although the individual electrons are drifting along at the speed of molasses, the rate their electric fields pass the voltage baton is stunningly fast – in fact it is about two thirds the speed of light. So the propagating wave of voltage is about as fast as the photons in an optical fibre.

Silicon-based computers keep on getting better and better, so the performance gap between silicon computers and the other contenders grows with time.

The result is that conventional copper wire and glass optical fibres convey information at the same speed, despite the speed of the individual photons being trillions of times faster than the speed of the individual electrons. It is true that for a variety of reasons an optical fibre carries more data per second than a copper wire, but a given bit of information does not travel from one end to the other any quicker.

This misconception about the relative speed of optical fibres is echoed in the predictions of an imminent revolution in which optical chips and other technologies will take over from silicon.

I’ve been hearing these pronouncements for the last 40 years – optical computers, DNA computers, protein computers, graphene computers or quantum computers. From time to time one or the other surfaces as the latest technology predicted to have greater speed and capacity than silicon computers.

In the meantime, silicon-based computers keep on getting better and better, so the performance gap between silicon computers and the other contenders grows with time. My observation prevails – silicon is king.

But it is important that scientists and engineers continue to work on alternatives to silicon computers – you never know where the next breakthrough will come from or what the spin-offs will be. My beef is that they should not be making claims that are far ahead of reality, and journalists should be comparing like with like, not photons with electrons.

Alan finkel 2014b.png?ixlib=rails 2.1
Alan Finkel is an electrical engineer, neuroscientist and Chief Scientist of Australia.
Latest Stories
MoreMore Articles