How small can you make a computer? (Engineering, Cambridge)

This is a question that’s been much in the minds of computer engineers recently, and the short answer is: very, very small indeed. Already by 2013, we had a fully working computer no bigger than a grain of sand. It’s designed to be inserted right into the eye to monitor glaucoma, which is why, with a nice touch of wit, it’s called the Micro ‘Mote’ after the biblical speck in the eye.¹ It’s got processing, data storage and even wireless communication, and is solar powered by light coming into the eye.

And the EU’s Pico-Inside project has made a simple logic gate that’s quite dramatically smaller than this. It has the logic power of fourteen transistors yet is made from just 30 atoms. That means it’s not only too small to be seen under any optical microscope; it’s too small to be seen by anything but the most powerful scanning tunnelling or atomic force microscopes. So you could fit about a quintillion of these little processors inside the Mote!

Back in the late 1960s, computer pioneer and Intel founder Gordon Moore noted something remarkable. In 1958 two transistors had been linked together in an integrated circuit inside a silicon crystal to create the first ‘silicon chip’. Ever since, Moore observed, the number of transistors that could be fitted inside a chip had doubled every year. Since then, electronic devices have gone on shrinking by the year in accordance with ‘Moore’s Law’.

The pace of miniaturisation has slowed down recently and the number of transistors doubles only every two years. Still, it has given us all the amazing ‘smart’ devices we have at our disposal today – tablet computers, phones with the computing power of a supercomputer of not so long ago, and so on. Every time someone suggests that miniaturisation has reached its limits, computer technologists seem to succeed in squeezing electronics into even smaller spaces.

So the question is, how much further can the shrink go? One might also ask, why would you want anything smaller?

It seems we are indeed coming close to just as far as we can go with conventional transistors. They are already down to the nanoscale – billionths of a metre (the size of viruses). But there may be problems with going much further. Transistors work as ‘gates’ turning on and off the flow of electrons. They are made of ‘semiconducting’ material that can be switched on and off to conduct electrons or block them off. But once the barrier gets down to about a nanometre (nm), quantum effects come into play. In particular, quantum tunnelling occurs. Quantum tunnelling is when an electron tunnels right through the barrier as if wasn’t there. (Actually it doesn’t ‘tunnel’ through at all, but simply disappears one side and reappears the other side.) If the gate can’t be shut to electrons because of quantum tunnelling, the transistor cannot work. The smallest transistors now are just over 30nm across, so this limit may soon be reached.

Transistors provide the logic gates on which computing depends: the yes/nos, and/ors, 0/1s. If transistors reach their limit, could logic gates be created in another way that circumvents this quantum limit? That’s what the Pico-Inside team and others are working on. Instead of trying to squeeze more and more computing power into an ever smaller space, they are starting from the bottom up, seeing if they can build a computer bit by bit from atoms, so that they can take advantage of quantum effects rather than being defeated by them. To move the atoms around to build such computers, they work with atomic force microscopes to nudge the atoms into place. So far, besides the 30-atom logic gate, they have assembled atoms to make vehicles, gears, wheels and even motors, each consisting of a single molecule. It’s long way before they create anything that remotely resembles a working computer, but the possibilities are clearly there.

One problem with these nanocomputers is not the processing power but all the peripherals. How can they be powered? How can they be kept cool? How can they communicate with other devices? It’s no good building a computer the size of a molecule if it then needs a wireless add-on a trillion times as big to send its data, or a solar cell or battery even bigger. And of course solar cells won’t work in dark places. So these are problems that need to be solved if nanocomputing is ever to become a reality.

Something even more dramatic than the nano-computer could be achieved by abandoning the simple logic gate of the transistorised computer altogether with ‘quantum computing’. The aim here isn’t to make a small computer, but to harness the power of quantum effects to achieve hugely faster speeds, which might end up meaning the same thing. And to make a quantum computer you have to scale things down anyway to the level where quantum effects come into play – that is, the level of atoms, electrons and even photons. Quantum computers, if they are ever built, will harness atomic or subatomic particles to be processing units.

The idea is that instead of the bits in a conventional computer, which are just 0 or 1, you use quantum bits or qubits, in which bits are superimposed by quantum effects to be not just 0 or 1 but can also be both at the same time. With conventional computers, the bits have to run through all possibilities sequentially when making a calculation. With qubits, they can all be tried simultaneously. That means a computer could solve a problem many millions of times faster than a conventional computer by working on problems in parallel.

In 2014, Canadian company D-Wave hit the front page of Time magazine with what they claimed was the first commercial quantum computer. The D-Wave is the size of a large wardrobe and works – but no one is sure yet whether it is actually a quantum computer. Neither are they yet sure what the benefits of such a computer might be. So far, it’s been suggested that it might help banks steal a march in financial dealings by making super-fast calculations, which would give it real commercial value, but its overall value to mankind is not yet so clear.

This is one of the issues with tiny computers: what’s the purpose? Why would you want a computer the size of a grain of sand, when it’s easy enough losing your super-slim mobile around the house? There are at least two answers to this.

The first is that devices the same size as your mobile could have dramatically increased computing power so they can do all kinds of fantastic things they can’t do at the moment. Critics say, though, that this is the wrong way to go about things. There is no need to increase the power of individual computers. Instead, you increase connectivity, so that the computing power of all computers linked into a network is used simultaneously, as in cloud computing. That way, the power of your individual terminal can be quite small, because it taps into the power of the cloud.

The second is that nanocomputers could be deployed to enable us to manipulate things on a nanoscale. The most exciting possibilities are inside the body. I’ve already talked about the Mote operating inside the eye. Nanocomputers could be inserted into the bloodstream to monitor blood flow, or help make other diagnoses with on-the-spot reports. One tiny computer might not do much, but swarms of nanocomputers swallowed in a simple pill might help break down cholesterol in the blood, or perform a swift inside job to remove kidney stones.

Other scientists talk about harnessing the power of organic molecules, to make biodegradable computers that can work inside living cells, maybe deactivating cancer cells, or delivering drugs to particular cells.

The vision of our bodies being continually repaired from the inside by swarms of unimaginably tiny computerised devices is certainly a beguiling one, if somewhat unnerving. It could be, if realised, the most astonishing breakthrough in medical care ever. And nanocomputers could be used in many other fields, from cleaning pipes from the inside to building drugs molecule by molecule.

All this is some way off, and the problems with building, powering and connecting with such devices are considerable. But 60 years ago who would have imagined computers could ever be so small and powerful that they could do things we now take for granted, such as connect to the internet from a tiny phone, pretty much anywhere in the world?

Personally, I couldn’t make a computer any smaller or more complex than an abacus right now, but there are people who can. But I’ve thought of an even better way: I could participate in the making the finest computer known – a human brain. Even when fully grown, the human brain is amazingly small for its power – the most powerful ever known. How do you fancy making a supercomputer tonight, darling?

Footnote

1 ‘And why beholdest thou the mote that is in thy brother’s eye, but considerest not the beam that is in thine own eye?’ (Matthew 7:3)