Sweating the small stuff Home

Terabit Storage Device

Plastic process produces puny pores

By Eric Smalley,
Technology Research News

Cramming lots of information into very small spaces means making and measuring infinitesimal containers for each bit of data.

Researchers from Tohoku University, the Japanese National Institute for Materials Science, and Pioneer Corporation in Japan have found a way to store huge amounts of data after figuring out how to make many tiny, inverted dots in a thin film of metal and determining how to sense the state of each dot.

The dots are as small as 10 nanometers in diameter and store one bit of information each. A nanometer is one millionth of a millimeter, or the equivalent of a line of 10 hydrogen atoms.

The researchers' prototype storage device packs 1.5 trillion dots per square inch, and so could store 1.5 terabits in one square inch of material, said Yasuo Cho, an associate professor of electrical engineering at Tohoku University in Japan. That's the equivalent of 48 million 250-page books, or 47 DVDs.

The storage material, a thin film of single-crystal lithium tantalate, is ferroelectric, meaning its atoms are aligned electrically, or polarized. Atoms in small sections, or domains, of the material can be polarized opposite to neighboring domains, and these two polarization states can represent the 1s and 0s of computing.

In contrast, today's disk drives are made from ferromagnetic materials, whose polarization is magnetic. The domains in ferromagnetic materials are sensitive to temperature, making very small domains unstable even at room temperature.

Domain size is not affected by temperature in the researchers' ferroelectric material, and the domain wall of a typical ferroelectric material can be as thin as one or a few lattice segments of the crystal, which is much smaller than is possible using ferromagnetic domains, said Cho.

In order to use these infinitesimal domains to store information, however, there must be a way to change the polarization states to write data to the media, and sense the state without affecting it to read the data.

Other research teams have used ferroelectric materials' piezoelectric behavior to read domains. Piezoelectric materials generate electricity when they vibrate and vibrate when they are subjected to an electric current; piezoimaging measures domains using the vibrations of a microscopic probe tip. But this measurement technique limits the size of the bits that can be measured and the speed at which they can be sensed, said Cho.

The researchers' measuring device, dubbed scanning nonlinear dialectic microscope (SNDM), overcomes these limitations, said Cho. The device uses an alternating electric field to measure the change in capacitance, or ability to store an electric charge, between domains, which reveals the different polarizations. "SNDM has sub-nanometer resolution, and is a purely electrical method," he said.

Using the prototype, the researchers were able to read 25 kilobytes, or thousand bytes, of data per second, said Cho. This is relatively slow -- it would take 10 seconds to retrieve a 250-page book at that speed, assuming 1,000 characters per page. It is possible to increase the read speed to 3.75 megabytes per second, said Cho. This would make it possible to retrieve the information contained in about 150 books in 10 seconds. Current disk drives have read speeds of about 20 to 50 megabytes, or million bytes, per second.

The researchers' prototype stores information 100 times faster than it can read it; the prototype has a write speed of 2.5 megabytes per second, said Cho. This could be increased to 125 megabytes per second, he added. Today's disk drives write data at about 10 to 35 megabytes per second.

The researchers are ultimately aiming to increase the amount of information they can store in the material to 4 thousand trillion bits, or 4 petabits, per square inch -- the equivalent of 125,000 DVDs worth of information. This assumes a domain size of 0.4 nanometers, which is an individual atom within the crystal lattice.

The researchers' scheme and the use of lithium tantalate are good ideas, said Rainer Waser, a professor of materials science and engineering at Aachen University in Germany. There are many questions, however, including how the researchers will increase the write speed, he said.

It is also much more difficult to come up with a new technology than to improve an existing one. "[Magnetic] hard drives are highly developed systems," said Waser. "Nevertheless, it is interesting to think along this road," he said.

The researchers current prototype is not accurate enough for practical applications, but further refinements should solve the problem, according to Cho. The system could be used in practical applications in five years, he said.

Cho's research colleagues were Kenjiro Fujimoto,Yoshiomi Hiranaga and Yasuo Wagatsuma from Tohoku University in Japan, Atsushi Onoe from Pioneer Corporation in Japan, and Kazuya Terabe and Kenji Kitamura from the National Institute for Materials Science in Japan. They published the research in the December 2, 2002 issue of Applied Physics Letters. The research was funded by the Japan Society for the Promotion of Science (JSPS).