Small, Cheaper Flash Memory

Continued from page 1

Jim Handy, an analyst at Semico Research in Phoenix, AZ, says there is already a large demand for such inexpensive embedded memory. "What Freescale promises to do is open up the possibility to put larger memories onto an economically sized chip," he says.

The tricky thing about the Freescale process, Handy says, is distributing the nanocrystals. It's like "steam condensation on a window...if they put too much on, the droplets will start touching each other and linking together, which is the last thing that they want -- they want them to be all stand-alone. But if they put on too few, then they don't get a useful number of them. That's the secret sauce that they bring to the party."

Albert Fazio, director of memory technology development at Intel in Santa Clara, CA, says his company is also exploring technologies -- metal particles as well as silicon nanocrystals -- that could replace the polysilicon gates in flash memory. But such innovations probably won't be necessary, he says, until the end of the decade, by which time it will be difficult to shrink components further using current techniques. And flash memory leader Samsung shares that view, according to a company spokesperson.

By then, flash could have some stiff competition from other new nonvolatile memory technologies. Freescale, for one, later this year plans to commercialize an MRAM (magnetoresistive random access memory) chip, based on a type of memory that could eventually replace both the high-speed memory in a computer and flash memory in cell phones and digital cameras. MRAM stores information not by storing electrons but by changing the magnetic state of a material.

Meanwhile, Fazio says the most promising replacement for flash is something called phase-change memory, which uses lasers to switch a material between crystalline and amorphous states. This technology, he says, could scale down to the "5-to-10-nanometer range," a fraction the size of today's memory elements.