Motorola plans to announce on Monday that it has manufactured prototypes of a flash-memory chip that relies on a thin layer of silicon crystals to retain data, a breakthrough that could help the flash industry overcome looming technical hurdles.
If the chip moves to the mainstream, flash-memory chips--used to store data and applications inside cell phones, industrial equipment and portable memory cards--would likely become cheaper or, conversely, more powerful compared to existing chips because manufacturers could squeeze more memory cells into a smaller space.
"We think we can achieve the same density with half the area," said Ko-Min Chang, manager of memory devices in Motorola's semiconductor product sector.
Flash memory can retain data up to 10 years because the transistor gates--the microscopic on/off switches inside chips--are wrapped in a layer of glass, or more technically, silicon dioxide. The layer traps electrons inside the gate: The size of the charge inside is then read as a 1 or 0 by the computer looking at it.
The silicon dioxide layer, however, is thick and will become increasingly difficult to shrink further over the next five years, a situation that researchers and analysts say could hamper the performance gains and profitability of flash.
Shrinking chips (and the subcomponents inside them) essentially lets manufacturers produce more chips out of a single wafer. Because the vast bulk of the cost involved in making chips is used up before the first chips are even produced, size is a major factor in profitability and volume production. Flash is a growing, multibillion-dollar industry, but profits can often be difficult to maintain because of harrowing price declines.
The nanocrystal layer is far thinner than the silicon dioxide, Chang said. In fact, it's not even solid, and more like frost on a window. Although silicon is traditionally an electrical conductor, at these levels the quantum nature of the material takes over, and it becomes an insulator, trapping electrons and thereby retaining data.
The yield, or number of useable chips per wafer, is also relatively high for an experimental chip, he added. Around 30 out of 33 chips per experimental wafer work, he said. If all goes well, Motorola could begin shipping samples of nanocrystal chips to manufacturers in 2005 with mass production starting six months or so later.
Still, the concept remains in the experimental stage. "One of the main challenges with nanocrystals is maintaining the uniformity," Chang said.
The company is also tinkering with an alternative, called Sonos, which substitutes a layer of nitride for the silicon dioxide. Right now, nanocrystals may have a slight edge. Motorola has yet to produce a Sonos prototype, and tentative data shows that these chips may not be as amenable to shrinking.
source | cNet News
