Dylan McGrath

EE Times

(03/08/2006 2:22 PM EST)

SAN FRANCISCO — Intel Corp. sees current flash memory technology sustaining until the end of the decade, pushing out the need for ``universal memory`` until 2010, according to Greg Komoto, manager of strategic planning for Intel`s flash memory group.

Speaking at a session at the Intel Developer Forum here Wednesday (March 8), Komoto said Intel continues to believe that ovonic unified memory (OUM), also known as phase-change memory, is the most promising nonvolatile memory alternative, more so than magnetic RAM (MRAM) or ferro-electric RAM (FeRAM), which are also being studied as potential replacements.

Komoto said OUM shows the most promise based on its scaling path, declining cost basis and the fact that it is bit-alterable.

Intel has been doing R&D on OUM with partner Ovonyx Inc. since 2000, when Intel took a stake in that company. STMicroelectronics NV has also been licensed Ovonyx technology for some time. Last year, Japan`s Elpida Memory Inc. licensed the technology, as did Korea`s Samsung Electronics Co. Ltd., as a potential replacement for flash and DRAM.

Intel executives have said in the past that the R&D effort has produced working OUM memories, but that cost has been an issue. Komoto said Wednesday that the R&D effort has OUM scaling visibility to 15 nanometers.

Meanwhile, NAND and NOR flash memory technologies continue to scale, pushing out the need for nonvolatile alternatives.

Intel worked with Azalea Microelectronics Corp. to develop an OUM test chip at the 0.18 micron node in 2002. Stefan Lai, vice president and co-director of Intel`s California Technology and Manufacturing Center told EE Times at that time that Intel would develop a 0.13 micron OUM prototype on its own.

Intel`s interest in OUM dates back more than 35 years to September 1970, when Intel co-founder Gordon Moore and OUM inventors from Energy Conversion Devices Inc. co-authored a technical article on the subject in an issue of Electronics.