The Next Generation of Non-Volatile Memory

Wednesday, October 12, 2011

Emmett Jorgensen



Article by Ken Lee

Flash to the Future – The Next Generation of Non-Volatile Memory

Quickly, name a gadget that didn't even exist in the year 2000 but has since transformed our culture and the way we live today. 99% of you probably answered with MP3 player, tablet computer, GPS, e-Reader or smart phone.

What do all of these devices have in common? For one thing, they are all super portable, and that feature is due to the unbridled success of Flash memory.

The development and maturation of NAND Flash as an affordable, non-volatile, solid-state data storage solution has helped usher in an era of mobile technology. It has allowed us to invent gadgets that simply would not have worked if incorporated with spinning platter drives.

However, after a decade of revolutionizing the technology industry, Flash is reaching its limits for further development. In order to increase Flash's maximum capacity manufacturers have been shrinking the distance between transistors on flash chips over the years.

In 2000 Flash technology was manufactured using a 180-nanometer process. Today, modern NAND Flash cells are for the most part manufactured using a 32nm process, with some bleeding-edge manufacturers moving to a 24nm process. The only issue with this is that as Flash continues to shrink it becomes less and less reliable. As of April 2011, the theoretical minimum Flash cell size is 19nm. Beyond that point, the stability of Flash becomes highly suspect.

As we approach the limit of Flash technology, it is prudent to look towards the future and consider a couple of promising non-volatile, solid-state storage technologies that may succeed Flash memory in our mobile devices.

Phase-Change Memory

Like Flash, Phase-change memory (aka PRAM or PCM) is non-volatile, solid-state computer memory; meaning that it retains information when powered off and has no moving parts.

Unlike Flash which works by changing the electronic charge stored within gates to set a bit as a 1 or 0, PCM uses an electric current to produce heat which switches a chalcogenide glass between crystalline and amorphous states to set a bit as a 1 or 0.

PCM has several significant advantages over Flash:

  • PCM can effectively write data 30x faster. The memory element can change the state of a single bit from a 1 to 0.  In Flash if a bit is set to 0, the only way it can be changed to a 1 is by erasing an entire block of bits.
  • PCM can be scaled to 0.0467 nanometers without any loss of reliability.
  • PCM is more durable than Flash. Flash cells degrade quickly because the burst of voltage across the cell causes degradation. Once cells begin degrading, they leak electric charge causing corruption and loss of data. Flash memory is rated for about 5000 writes per sector, and most devices employ wear leveling to make them stable up to 1 million write cycles. PCM also degrades with use due to thermal expansion and metal migration but at a much slower rate. Theoretically PCM should endure up to 100 million write cycles.
  • PCM is suitable for use in more environments than Flash. Because Flash relies on trapped electrons to store information, it is susceptible to data corruption due to exposure to radiation. PCM exhibits a higher resistance to radiation and therefore can be used in space and military applications.

Magnetoresistive RAM

Magentoresistive RAM (aka MRAM) is another non-volatile, solid state technology has been in development since the mid 1990s. MRAM stores data using magnetic charge as opposed to electrical charge. MRAM is composed of pairs of miniscule ferromagnetic plates which make up the memory cells.

Each cell consists of two magnetic layers separated by an insulating layer. Each cell can be manipulated by an induced magnetic field which sets the polarity of the magnetic layers in parallel orientation or in an anti-parallel orientation. The different orientations determine whether the bit is set to a 1 or a 0.

MRAM has many significant advantages over Flash and PCM:

  • MRAM can be read and written to faster, and can be done on a much smaller scale. Like PCM, single bits can be changed from 1 to 0 without having to erase an entire block.
  • MRAM degrades substantially slower than either Flash or PCM.
  • MRAM could replace all memory in the future, making it a universal storage technology. It should offer speeds close to that of SRAM, with densities approaching that of DRAM, while being able to store information when power is removed like Flash or EEPROM.
  • Like PCM, MRAM also exhibits a higher resistance to radiation and therefore can be used in space and military applications that Flash is not suited for.

Final Thoughts

It should be noted that we can only speculate on when manufacturers will have to stop using Flash as the primary storage media in their products. Consider that in 2002, many experts assumed that Flash cells would not be stable when scaled past 45nm and predicted that Flash technology would need to be replaced by 2010. We know now that those predictions proved to be false.

Many experts today believe that technological breakthroughs, like implementing graphene, will allow the technology to be scaled down to 10nm without loss of stability.

If this is true, Flash may still be the dominant memory in mobile devices for many years to come. Even though emerging technologies like PCM or MRAM are vastly superior to Flash in many ways, PCM and MRAM are much more expensive to manufacture than Flash.

As long as Flash can remain a viable storage media there are too few incentives, and too much production costs for manufacturers to rush devices that use next generation memory into the market.

Ken Lee is a product manager at Kanguru Solutions specializing in data storage and duplication equipment.

Cross-posted from Kanguru Blog – Technology on the Move!

Possibly Related Articles:
Flash Storage Memory Hardware MRAM Phase Change Memory Non-Volatile Memory PCM
Post Rating I Like this!
jan baeyens Is there any particular reason why the memristor is not mentioned?
Best regards
Emmett Jorgensen You are correct, Memristor (as well as a couple of others) should be included in the discussion as well. Was more a matter of time and space constraints than anything else.
jan baeyens Txs for the feedback.
I'm really curious about the impact of the memristor.
Best regards
Victoria James Decades ago, Ken Olsen, the chairman and co-founder of DEC said, "When I was a teenager in the late 30s and early 40s, electronics wasn't a word. You were interested in radio if you were interested in electronics". I’m reminded of his quote every time I open a search engine to look for a new component or device.

The trouble with applying other people's new and innovative ideas is first finding out that those ideas exist. In many cases, we don't yet have words describing the concepts, we don't know where to look for them and we may not even know we should look for them. Search engines can’t tell us what to look for yet, they can only answer the questions we type in the box.

This problem is as applicable in design engineering and electronics in 2012 as it was when Olsen made his comment about radio. In my own sector – industrial portable memory, this is particularly true. As engineers, our response is often to simply re-invent the wheel - but unfortunately not every wheel we invent is as efficient as the first.

In memory, as in so many sub sectors of design engineering, developing your own device rather than opting for a specialist solution might be more costly than it seems at first glance. Similarly, choosing to adapt a commercially available memory solution, like a USB or SDHC card, can prove expensive in the long term. The answer isn’t as complicated as it seems – it’s a case of finding out what other people call the product you are looking for. If in the 30s Electronics wasn’t a word – I wonder what industrial memory will be called in a few decades’ time?
The views expressed in this post are the opinions of the Infosec Island member that posted this content. Infosec Island is not responsible for the content or messaging of this post.

Unauthorized reproduction of this article (in part or in whole) is prohibited without the express written permission of Infosec Island and the Infosec Island member that posted this content--this includes using our RSS feed for any purpose other than personal use.