October 03, 2005

UA Physicist’s Predictions Proven True, Leads Industry in New Direction in Quest for Superior Computer Memory

By Chris Bryant

There are no crystal balls visible upon entering Dr. William “Bill” Butler’s office. Yet, theoretical predictions this physicist made in a scientific paper published in 2001 have been verified experimentally and may be key in development of the next generation of computer memory and hard drives.

Recent discoveries, made by Stuart Parkin at IBM and Shinji Yuasa at Japan’s National Institute of Advanced Industrial Science and Technology, have renewed excitement in the race to refine a new type of magnetic memory, known as MRAM. This new memory could, in principle, one day replace both the computer industry’s standard memory, known as RAM, and the computer’s hard drive, because it can operate as fast as RAM yet is non-volatile like the hard drive.

These 2004 discoveries, made by the pair of researchers working separately, were not accidental. They were the result of careful experiments aimed at verifying theoretical predictions made earlier by Butler in collaboration with colleagues Xiaoguang Zhang and Thomas Schulthess at Oak Ridge National Laboratory and James MacLaren at Tulane University.

As lead author of the paper, published four years ago in Physical Review, Butler performed computer calculations for a previously unknown electronic structure that incorporated multi atom-sized layers of iron and magnesium oxide, with an insulating barrier in between. The calculations predicted the materials’ behaviors under certain achievable conditions.

Butler, who directs UA’s Center for Materials for Information Technology, or MINT, and his colleagues were studying a phenomenon called electron tunneling – which allows an electron to occasionally defy the law of conservation and energy.

Dr. Arunva Gupta, professor of chemistry and chemical and biological engineering

Their paper predicted the change in the rate at which electron tunneling occurs when a magnetic field is applied. They found that this change could be made very large for a particular combination of materials arranged in a particular way.

This large change in tunneling rate leads to a large change in electrical resistance which can be very useful. One of the first applications of this discovery may be to disk drives which need a very tiny, but sensitive, magnetic field sensor, researchers say. Another application is to a new kind of memory called magnetic random access memory or MRAM. MRAM would work like the DRAM memory in today’s computers, but unlike DRAM, MRAM would be non-volatile so work would not be lost if the power failed. A third application might be to a new type of computer which could instantaneously reconfigure itself to perform a complex operation in the most efficient way.

“There’s hope that magnetic random access memory might be a replacement for the DRAM you now use in your computer,” Butler said. As every computer user experiences daily with DRAM, or dynamic random access memory, an extensive boot-up process is required and power outages can result in lost data. Not so with MRAM, Butler said.

“You could just turn your laptop or hard drive on, and it would be on immediately,” he said. “It wouldn’t have the boot-up process, and it would be better for power management as it would need less.” The power savings could greatly increase battery life for laptop and other portable computing devices.

“DRAM has to be refreshed,” Butler said. “That’s why you have to keep it plugged in, but MRAM doesn’t have to be refreshed. It’s what’s called non-volatile memory. When the electricity goes off, it ‘remembers’ the state it was in. In principle, it could be as cheap as DRAM, equal in density, equal in speed, and it should be nonvolatile.”

Despite the high hopes and extreme attention MRAM is getting from developers, it’s not yet on the market. “MRAM is a technology that has been promised to be coming out ‘this year’ for the last several years,” Butler said. “Lots of companies are working on it, lots of companies are making devices, but there is not a product you can buy.”

If a computer company Web site one day touts its latest machine as “equipped with MRAM,” a key hurdle in the development will likely be traced back to Butler and his colleagues’ 2001 calculations.

UA Home | Dialog Extra | UA News | Faculty/Staff Links | Legislative Updates