COMPUTING: Super-fast, tiny computers with the power and speed to make the current crop look Stone Age are not just the stuff of sci-fi - they're here, writes Karlin Lillington

Annoyed with the amount of space your computer hogs on your desktop? Imagine a PC so infinitesimally small that parts of it would only be visible under a microscope. Yet it would possess computing power and speed that will make the most muscular supercomputer of today look as limited as the vacuum-tube machines of the 1950s, computing's equivalent of a Stone-Age cart set next to a Ferrari.

This is the bizarre world of quantum computing, computing at the atomic level. It offers great promise - the ability to solve complex problems and analyse massive chunks of data that remain far beyond our ability today. But it also poses risks - a quantum computer could easily crack all the encryption, or mathematical codes, currently used by security agencies, governments, banks and other companies to protect their digital information.

Despite the Tomorrow's World sound of it all, it is here already. In the past decade, quantum computing as a discipline has moved from the textbook and research paper to reality. The first functioning quantum computer was built last year and researchers predict commercial quantum machines will perform specialised tasks in the next decade.

What makes a quantum computer potentially so extraordinarily powerful is something called parallelism. "A quantum computer can perform millions of simultaneous calculations in one run," says Prof Gerard Milburn, deputy director of Australia's Centre for Quantum Computing Technology (www.qcaustralia. org).

Parallel computing itself is not entirely new. It's been a growing research area within traditional computing for several years, where researchers have been experimenting with ways of putting lots of computer processors to work on a single problem.

They've done this by either building large multiple-processor computers, or by harnessing together the processors of many different computers connected on a network, an approach known as distributed computing. However, quantum computing promises to transport the whole concept to an entirely new dimension - literally. It exploits the extremely odd laws of physics that apply at the level of the very, very tiny, laws that were postulated by the physicist Albert Einstein in his theory of relativity.

For example, one of the features of quantum physics is that matter can be in any number of states simultaneously.

In the larger world of Newtonian classical physics that we see and feel, this cannot be so: a light, for example, is either turned on or off. But at the quantum level, a light could be both on and off and at all intermediary states, at the same time.

Take this to the world of computing, where all information is recognised by a computer as either the digit 1 or the digit 0. In the quantum world, there are far more possibilities.

"A quantum computer can store data as both 1 and 0 and all possible combinations in between," says Prof Milburn. It considers them all simultaneously and, when analysing data, calculates the results as all the possible variations that could happen.

"So a quantum computer can run many computing paths at once," he explains. "A quantum computer doesn't have to make up its mind until it's finished."

He explains this seemingly impossible situation another way. "The fundamental nature of quantum mechanics is that nature is fundamentally random. But it allows randomness and certainty to coincide, because if the world is just random, it would be completely unintelligible to us." Weird? Certainly.

"Quantum mechanics is weird, but that's the way the world works," he says with a laugh.

Because researchers are now able to manipulate matter at the atomic level, they are pursuing several possible channels for creating a working quantum computer. Prof Milburn, a leading researcher in the area who heads the work in one such channel, is spending a sabbatical break with the quantum computing laboratory at Cambridge University. He was visiting universities and in Ireland last week to give a lecture on his teams' efforts.

Their goal is to create quantum semiconductors - the processing brains of a computer - that are assembled from individual atoms. To do this, they are using phosphorus atoms set on a bed of silicon, the material traditionally used for manufacturing a semiconductor.

Described by Prof Milburn, the process sounds deceptively simple. Take a very thin bed of silicon and place a layer of hydrogen atoms on top. Holes are then punched through regularly back down to the silicon level, an atom at a time, to form regular rows of holes. Phosphine gas, comprising three hydrogen atoms and one phosphorus atom, is added to the holes. The hydrogen gas is evaporated, leaving a single phosphorus atom in each pocket, which has merged with the silicon beneath it. The result is a grid of hydrogen atoms that surround individual phosphorus atoms.

Many such layers make up the microprocessor. Prof Milburn says the team also has a method to create the minute network of wires that link all the phosphorus atoms into a true microchip, allowing data to be sent to it and received from it.

Each individual atom can either be in a 1 or 0 state or anything in between, due to the actions of the phosphorus nucleus, which is magnetically charged. It can spin between a positive (or 1) or negative (0) charge.

This tiny piece of information at the nucleus level, the smallest unit of quantum computing information, is called a quantum bit, or qubit.

Creating the tiny microporocessor is only the beginning, though, of creating a complex quantum machine. Scientists are also trying to figure out not just how data would be stored in one, but even more basically, how the data would be read.

Quantum data tends to move around and also can be changed quickly, so researchers need some way of netting all the data before it scatters and becomes unreadable. As a result, notes Prof Milburn, quantum computing "raises all sorts of questions about what you mean by information. We need new ways at looking at information, and at what an algorithm is."

Which all might sound impossibly futuristic, except that the semiconductor industry, after decades in which microprocessors grew ever smaller and more powerful, is about to hit a wall. An atomic wall, to be precise - companies such as Intel now build microprocessors at the molecular level, but within a decade, they must begin the move to the atomic level if chips are to continue to improve. Thus, Intel and Hewlett-Packard, for example, are following the work of Prof Milburn's team.

Prof Milburn expects to see the first commercial quantum machines, dedicated to a single task, such as drug analysis, in a decade.

But a general purpose quantum machine is, perhaps, two or three decades away. Which gives us plenty of time to find the room on our desks for it when it arrives.