So far it's a snail's-pace race but there is hope that Samsung Electronics Co. Ltd. is going to pick up the pace now — possibly leaving Numonyx in its dust. However, it is unlikely that Samsung will dash Usain Bolt-like down the track, and there are pro and contra indicators for the take up of phase-change memory (PCM) otherwise known as phase-change random access memory (PRAM).

For both Numonyx, jointly owned by Intel and STMicroelectronics, and Samsung the technology derives from licenses and cooperation with Ovonyx Inc. formed as a spin-off from Energy Conversion Devices Inc., which performed some of the original research into phase-change in chalcogenide materials.

In both camps research intensified in the early part of this decade but has yet to produce a substantial revenue stream. This is slow progress by any measure.

Now Samsung has announced that it has begun production of a 60-nm 512-Mbit PRAM and is aiming it at mobile phone handsets and other battery-operated applications. The announcement comes almost three-years to the day after Samsung announced the existence of a prototype 512-Mbit phase-change RAM in September 2006. So it is clear that characterizing the data retention and reliability of such a memory has proved non-trivial.

Samsung has, however, jumped out of the starting blocks and got ahead of rival Numomyx.

Pro and contra indicators

Numonyx has been shipping samples of a 90-nm 128-Mbit phase-change memory since 2008 but there have been no signs of large-scale take up. And more recently there were indications that the Numonyx' timetable for the introduction of a 45-nm 1-Gbit phase-change memory on 45-nm silicon was running late (see Is 1-Gbit phase-change memory schedule slipping?).

In Samsung's case there is one ominous contra indicator to success. The company placed in its press release the phrase "Finer technology nodes will be applied in future-generations of PRAM to expedite further commercial adoption."

That emphasizes that the current 60-nm 512-Mbit device is not at the leading-edge or competitive with in terms of capacity. The supposedly hard-to-scale NAND flash is at 32-Gbit in 32-nm process technology thanks to Toshiba. This means that the 512-Mbit PRAM will more often than not be a proof-of-concept device intended to whet the appetite of developers.

What is also absent is any sense of when those finer technology nodes will arrive, any sense of roadmap, or that Moore's Law and scaling is driving phase-change down to finer geometries.

On the pro side Samsung, as well being the world's number two chip manufacturer, is also the world's number two mobile phone handset maker. That means that the company can start inserting the memory into its own phones and building volume production of phase-change memory.

Related links and articles

But both Samsung and Numonyx will have to step on the gas in terms of multilevel cell storage and getting on to leading-edge manufacturing process technology if this technology is going to make any sort of impact. Otherwise it looks set to become an interesting, but ultimately irrelevant, cul-de-sac similar to magnetic bubble memory — or a minority player such as ferroelectric random access memory.

If phase-change memory is really the good thing that Samsung is indicating, Numonyx will want to reply quickly with the announcement of 1-Gbit devices on 45-nm silicon and these memories will start to turn up as advantage-providers in consumer products. If Numonyx is slow to respond it will indicate that engineering and characterizing phase-change materials in each process node is just too much heavy-lifting to allow phase-change to catch up with flash.

And one of them main things that would demonstrate that phase-change memory is a contender would be the free flow of datasheet information. Numonyx has restricted the supply of the datasheet on the 128-Mbit Alverstone chip to its PCM "early access" club (see Numonyx rolls early access program for PCM). This has created the impression that phase-change memory has some sort of skeleton in its closet or something to hide.

Usain Bolt is good enough to showboat and if phase-change memory is good for battery-operated equipment its advocates should be prepared to show us. I fully expect Samsung to be conventional and open in the distribution of engineering information about its PRAM.