IBM, the world's largest computer company, on Tuesday announced a breakthrough with the construction of a tiny "quantum" computer that could eventually be used to build extremely fast commercial computer systems.

A quantum computer is based on the internal interactions of atoms rather than electronic signals as in today's computers. Researchers at IBM's Almaden Research Center in California said they had built a simple quantum computer and used it to solve a very complex mathematical problem in just one step. Isaac Chuang, who led the IBM research team, said that it would have taken many steps to solve the same problem using a conventional computer.

"If we can scale this device into a larger one, it would lead to a very powerful system that could solve complex mathematical problems in days that would take a conventional computer system millions of years to solve," said Mr Chuang.

Although quantum computers are not suited for common business applications, for specialized problems such as weather forecasting, they would be very effective.

The IBM research work was presented at the Hot Chips 2000 conference held at Stanford University in heart of Silicon Valley. The experimental device uses five fluorine atoms embedded within a larger molecule. Using nuclear magnetic resonance instruments, similar to those found in hospitals for patient health diagnostics, the scientists can program the quantum computer for different tasks. As the fluorine atoms spin, their magnetic interactions process data in a similar way as the microprocessor found in PCs.

IBM said that the device proves commercial quantum computers are possible. These would operate millions of times faster than a supercomputer but are still ten to twenty years away.

However, Mr Chuang said that a hybrid computer that combines a quantum computer with a conventional computer system may be possible within five years.

"There are still further breakthroughs to be made but this is a field of enormous surprises," he added.

The IBM researchers used about $1m of equipment to build and operate the quantum computer. The next challenge is to lower costs by shrinking the electronic equipment into a handful of chips.

Computer performance doubles about every 18 to 24 months as described by Gordon Moore, a co-founder of Intel, in what is known as Moore's Law. But computer chips based on silicon are approaching fundamental physical limits that will constrain this rate of performance growth. Quantum computers are one of several novel computer technologies that researchers at IBM and at other organisations are investigating as potential replacements for silicon based chips.