By Caroline Humer

NEW YORK, Dec 19 (Reuters) - Researchers at International Business Machines Corp. IBM.N said on Wednesday they have demonstrated a calculation that could be used to break complicated codes, marking a small step in the advance of quantum computing, a technology based on quantum mechanics.

IBM scientists will publish details in the scientific journal Nature on Thursday of the demonstration of "Shor's Algorithm," a method of factoring numbers that was developed in 1994 by AT&T T.N scientist Peter Shor.

It was that algorithm, and the promise it holds for its ability to break large encryption codes, that spurred interest in quantum computing in the 1990s.

Quantum computing is one of several paths that researchers are taking as they strive to make smaller and smaller microchips. Under Moore's Law, which was set forth by Intel Corp. INTC.O co-founder Gordon Moore, the number of transistors on a chip doubles, or alternatively, data density doubles, every 18 months.

IBM said it has built a quantum computer based on seven atoms which, because of the physical properties of those atoms, are able to work together as both the computer's processor and memory. Previously the largest computer IBM had built was based on five atoms.

IBM scientists said that they were able to use the computer to show that Shor's algorithm works by correctly identifying 3 and 5 as the factors of 15.

"Although the answer may appear to be trivial, the unprecedented control required ... during the calculation made this the most complex quantum computation performed to date," Nabil Amer, manager of IBM Research's physics of information group said. A quantum computer is based on the spin of an electron or atomic nucleus.

In addition to encryption, other applications for quantum computing include data mining, or searching large databases for particular pieces of information. Amer said it is still unclear when quantum computers could become commercially available.

John Preskill, professor of theoretical physics and director of the Institute of Quantum Information at CalTech in Pasadena, California, said that the experiment brought quantum computing a baby step forward by revealing errors in the process.

"Part of the challenge of building large scale quantum computers is that they are very susceptible to error and we need to understand the types of errors that occur, as well, in order to know what's the most promising way of building quantum computers," Preskill said.

To put the size of the computer into perspective, Preskill explained that currently the fastest computers in existence, or supercomputers, could factor -- or find the smallest indivisible factors of -- a number that is 130 digits long in about a month. But they wouldn't be able to factor a 200-digit number.

A quantum computer could tackle that task, he said, but it would need to include thousands of quantum bits, or atoms. IBM scientists used the computer based on seven atoms to factor a two-digit number.