Japan?s Elpida Memory Inc. and Sharp Corp. announced the formation of a Resistive RAM [ReRAM] development program which promises commercial devices as early as 2013, reported Nikkei Business Daily. The National Institute of Advanced Industrial Science and Technology [AIST], University of Tokyo and a number of semiconductor equipment companies are also joining the collaboration indicating the level of importance the project is being given within Japan?s semiconductor technology community.

Cell overview of Resistive RAMThe New Memory Project has been assigned under the newly created MCP & Flash Division reporting directly to Yukio Sakamoto, President of Elpida Memory Inc. as reported in a recent company reorganization announcement.

Resistive random access memory chips, or ReRAM as they?re called, consume far less power than conventional NAND flash memory and are capable of writing data 10,000 times faster. The companies fully expect ReRAM to displace NAND Flash is wide variety of applications.

No reference was made as to what resistive memory process technology is under consideration though AIST and University of Tokyo have been studying several varieties of experimental memory technologies for nearly a decade.  That the companies plan a commercial introduction in 2013 indicates that the technology chosen has already passed several internal technology audits and is robust enough to allow management the confidence to make the announcement. 

The project ties Elpida?s memory design and production expertise with Sharp?s thin film technology capabilities in photovoltaic solar cell production. A Resistive RAM cell is essentially a diode whose transfer characteristic can be changed [programmed] by applying a super voltage that?s slightly higher than that normally required for read operations.

ReRAM?s attractive attributes include:

  • Non volatile endurance greater than, or equal to 100K cycles
  • Zero power standby mode
  • Low Operational power
  • Symmetrical Read-Write Cycle times equivalent to – or better than DRAM
  • Generationally Scalable over Voltage and Geometry
  • Cost Effective Generational production
  • Are a Storage Class Memory [SCM] candidate

 
Finance & The Taiwan Connection
Elpida recently filed in early September for a 60 Billion Yen [$731 Million] convertible debenture to retool the Hiroshima memory fab for "40nm and below process for the R&D investments, etc., concerning the manufacturing of DRAM products using 30nm generation". Elpida is also planning to list shares on the Taiwan Stock Exchange in early 2011 as the "next step" in a larger strategy Sakamoto said in early October. The listing would raise money in Taiwan when the time comes to do deals, he said. Powerchip Technology, ProMOS Technologies, and Windbond Electronics are among possible targets that are thought to be under consideration. Sakamoto sees Samsung as Elpida?s chief competitor and now appears to be executing a well thought out strategic initiative to counter Samsung?s dominance.

What about Spansion?
All this places Spansion in a somewhat awkward position with their partner Elpida. Former joint venture of AMD and Fujitsu, Spansion just recently completed developing the industry?s first charge-trapping 1.8 V, SLC, 4-gigabit NAND flash memory. Samples should be shipping this quarter [4Q 2010]. Spansion is continuing the development of 1 and 2-gigabit products to satisfy mobile customer demands. Elpida plans to combine the Spansion NAND flash with Mobile RAMTM to sell into mobile consumer products with improved margins. It may well be speculated that hard data from the Elpida/Spansion collaboration may have concluded the decision to develop the resistive ram alternative…