What happened to next generation non-volatile memory? - Part 1
Paul Boldt posted on May 08, 2012 | | 3000 views
The road to next generation non-volatile memory (NVM) is windy to say the least. Bringing new technologies to market is never easy, but the task of replacing NANDMicron 25nm NAND, courtesy Micron Technology Inc. and IEEEMicron 25nm NAND Flash is proving to be particularly daunting.  To add insult to injury Flash hasshown considerable resiliency as it scales through more than one brick wall toMicron 25nm NAND smaller and smaller geometries.  Yes there may be higher bit errors, but that has only made Flash controllers and error correction technology more important. Certainly Apple put a stake in the ground, or at least hedged its bets for the next couple of years, with its December 2011 purchase of Anobit.

Among the first-generation contenders Phase Change Memory (PCM) had the early momentum to becoming a Flash replacement, while FRAM and MRAM went off to do their own thing.  PCM has, in the past year or so, been spotted in the wild in at least one socket normally occupied by NOR Flash.  While there was considerable fanfare around the spotting, the early press has certainly faded.  On the technical side there has been some recent discussion of serious scaling problems at future process nodes.  On the corporate side PCM is not having much fun either. First, it was part of Intel’s spin-off that formed Numonyx in 2008.  Then Numonyx was sold to Micron a scant few years later.  From here things have been pretty quiet.

So, is there anything on the horizon that will replace Flash?

What about some of the second generation NVM candidates? There are a bunch of them out there, including STT-MRAM and CMOx.  It is the latter of these two that will be considered in more detail here.  LIke others in its cohort, CMOx development has wandered a winding path.  Like all the candidates mentioned so far CMOx does not use the trapping of charge within a silicon structure.  In this case a conductive metal oxide is central to unit cells, which are in turn, organized in a passive rewritable cross-point array structure.  Without the need for a select transistor and the ability to stack memory cells, Unity looked at high density devices from the get-go.  There was a 2009 promise of a 64 Gb device appearing in the second half of 2010, with full production a year later.
So what happened to Unity in 2011? 

We will explore that in Part 2.  It certainly seemed like 2011 was starting on the right foot. Of all the possibilities for Unity as of February 5th, 2012  I would have placed a sale to Rambus pretty low on the list.  Stay tuned.