IBM scientist uses nanotech to create super memory

Von Arne Søiland

Without the research effort made by Stuart Parkin, it would not be possible to make iPods and other music-playing devices with hard disks. It was Parkin who, in 1991, discovered that the thickness of the non-magnetic layer between the magnetic layers in a hard disk creates large fluctuations between so-called parallel and antiparallel magnetization.

A few years later, in 1994, Parkin and his colleagues at IBM Research started using this giant magnetoresistive effect (GMR) to create read-write heads that multiplied the storage capacity in hard disks several fold.

The GMR technology, which is currently used by all hard-disk manufacturers, makes it possible to put your entire CD collection and then some in your pocket.

Simplifies 3D storage

Although IBM sold its hard-disk division to Hitachi three years ago, Parkin has continued his research into magnetic effects and "spintronics" (exploiting the spin inherent in electrons).

Parkin has been involved in the development of Magnetic RAM (MRAM), which combines the low cost of DRAM (dynamic RAM), the speed of SRAM (static RAM) and the stability of Flash memory. This technology and the varieties PCRAM (programmable conductor RAM) and FRAM (ferro-electric RAM) is now in production and consequently of less interest to the physicists at IBM Research.

Instead, Parkin, who was awarded the prestigious title "IBM Fellow" in 1999, has directed his attention to what he calls "Magnetic Race-Track Memory." The idea behind Race-Track is to store bits in 3D without having to read or write the information in 3D.

"3D-storage, for instance holographic storage, is hard to attain in practice as the technology is difficult to miniaturize," stated Parkin when Computerworld met him at the IBM research laboratory in Almaden, California, south of San Jose in Silicon Valley, earlier this summer.

He has not heard of Norwegian Opticom"s and Thin Film Electronics" failed attempts at developing 3D storage in polymers.

Hundred in a stack

In Race-Track, each bit is stored in a row of Us set side by side. Each bit is moved back and forth between the left and right side of the U, past a read-write head. This makes the storage density higher than in conventional storage media such as the hard disk, according to the highly renowned scientist.

The advantage of RAM, internal memory, is that it is very fast. The disadvantage is that the data disappear when the electricity is turned off (valid for SRAM and DRAM). The different Flash technologies (NAND, NOR etc) are also fast, and in addition non-volatile so that information is safe even without electricity.

The disadvantage of Flash is that the memory can only be overwritten a limited number of times. A limit of one million overwrites would for instance make it useless as internal memory in a PC.

Uncertain results

All these solid state-memories without moving parts have limited storage capacity. As for capacity, the hard disk is unbeatable, at least when magnetic tapes, which will handle the largest storage jobs in the future also, are disregarded.

Unfortunately, hard disks have a long access time. In addition, the moving parts limit the lifespan.

Parkin has immodestly set out to combine the best of all the different types of storage, while avoiding the disadvantages. He has proven the principles behind Race-Track, but he repeatedly stresses that nobody knows if it can be done. "If we knew the result, it wouldn"t be research," he paraphrases Einstein.

Moving walls

In magnetic storage media, information, or bits, is represented by points in the storage media having different magnetic orientation. The magnetic orientation is decided by whether the electrons are spin-up or spin-down.

One of the challenges Parkin is faced with is how to move spin without moving electrons. In IBM"s laboratory in Almaden, scientists show how this spin movement is done through so-called domain walls moving when subjected to electric currents.

Domain walls, which move, mean that bits of information can be moved back and forth over a stationary read-write head. And if bits are stored in a standing horseshoe, 3D storage is achieved.

Much electricity

The reader heads currently available can be used, but the challenge lies in creating a new type of write heads, explains Parkin.

Another problem facing Parkin is limiting the amount of electricity needed to move domain walls. At present, the amount of electricity used almost moves the atoms, an undesirable effect as it wears out the material.

In addition, the scientists argue over whether the domain walls have mass or not. Parkin belongs to the group who do not believe moving domain walls represent movement without mass, while other scientists maintain that this is the case.

Outside IBM

The scientists at IBM Research at Almaden merely shrug their shoulders at the term nanotechnology. Most of the scientists" efforts at the lab are on an atomic level, which makes nanometer a commonly used measuring unit.

In order to implement Race-Track, scientists want to use nanotraps, which are layers of metallic nanoparticles with a width of one atom or two atoms combined with a layer of polymer. Loaded particles in sizes one nanometers to three nanometers are thus spread in semi-conducting polymer materials.

This nano implementation is difficult as the electric resistance in the conductors becomes considerable with these thin layers. This comes in addition to the challenge of creating materials two atoms thick.

IBM has patented Race-Track, but the company will not be in charge of production if Parkin reaches his goal. Instead, the plan is to establish one or more companies that can develop and commercialize the idea. Talks with investment companies are already under way.

Moving atoms

Moving single atoms in tunnel microscopes paves the way for new possibilities in the electronics industry.

One of the advantages of Race-Track is that the storage density can be increased considerably without the integrated circuits having to be made more compact than today.

Circuits with conductors only 90 nanometers wide have been employed, and 35 nanometer is within reach. After that, the present lithographic methods get less applicable as the wavelength of light limits how narrow a conductor can be drawn.

Sure enough, by using liquid instead of air you can get a shorter effective wavelength on the light used to draw the circuits. Though, much narrower than 35 nanometer is hard to reach without reverting to a method called nanoimprint, which provides conductors with a width of 10 nanometers to 25 nanometers.

But if circuit designers are really to make the leap into the world of nano, where conductors are only a few atoms wide, engineers must move atoms.

Able to think outside the box

IBM"s scientists were the first to move single atoms. This was done using Scanning Tunneling Microscope, which IBM scientists shared the Nobel Prize for with Ernst Ruska from the Max Planck Institute in 1986.

"The alternative to STM is biology, that is grown electronic circuits," says Andreas Heinrich, scientist at IBM Research in Almaden.

In 1989, scientists at Almaden managed to isolate and move 35 xeon atoms by dragging them along the surface. Now, Henrich and his colleagues can lift atoms. This can be used for moving conductor molecules to a non-conducive surface.

The ability to move single atoms opens up completely new vistas, such as Magnetic Race-Track Memory, when developing the electronics of the future.