Published in Memory

Intel and Numonyx achieve memory research milestone

by on09 November 2009


Phase-change memory uses far less voltage than NAND

Recently, Intel and its joint venture, multi-billion dollar NAND flash collaborate Numonyx announced a key breakthrough in the research of phase-change memory (PCM), a form of non-volatile memory attempting to compete in the global flash memory market.

For the first time in history, researchers were able to demonstrate a 64Mb chip that successfully enables the ability to stack multiple layers of phase change memory within a single die. The discovery of this vertical integration technique of memory cells will ultimately pave the way for building memory devices with much more density than conventional NAND flash and will benefit the global consumer market with reduced economical prices.

Over the last thirty years, many researchers have been excited about phase change memory for a number of reasons. For one, the nature of its highly switchable element structure between amorphous and crystalline states allows engineers to manipulate its thermoelectric properties for storing data in the forms of binary 0’s and 1’s. In layman’s terms, phase change memory is designed on a material called chalcogenide glass which is famous for its ability to be switched between two states with the application of heat. The crystalline state of chalcogenide glass has low electrical resistance and thus represents a 1. On the other hand, the amorphous state has high resistance against electrical current and thus represents a 0.



Intel and Numonyx researchers are now able to demonstrate a vertically integrated memory cell, which they call PCMS (phase change memory and switch). PCMS is comprised of one phase change memory element layered with a newly used Ovonic Threshold Switch (OTS) in a true cross point array. The ability to layer or stack arrays of PCMS provides the scalability to higher memory densities while maintaining the performance characteristics of PCM, a challenge that is becoming increasingly more difficult to maintain with traditional memory technologies.


Although PRAM hasn’t yet reached its commercialization stage for consumer electronic devices, nearly all of the prototype devices under development in research labs make use of a chalcogenide alloy of various elements called GST with the aforementioned heat-dependent properties. On another note, the time required to complete a phase transition from amorphous to crystalline and vice-versa is completely temperature-dependent. A typical crystallization time scale used is approximately 100ns.


Comparison table courtesy of HotHardware

In perspective, Intel and Numonyx see PCM as a much better alternative to NAND flash because it requires far less voltage. A typical NAND flash cell uses an electrical charge to store and read memory, whereas PCM uses heat on chalcogenide glass for the same purpose. While the advantages of PCM may seem positively feasible, there is still one unfortunate drawback to implementing the technology. The cost of production for manufacturers would require major equipment overhauls, and with the IT industry still in a recession, it may prove difficult to migrate to a new multi-billion dollar exploration investment when the current focus in mind is simply trying to stay alive. Intel and Numonyx have made it clear that their achievement is strictly a production milestone, not a commercial one.

“The results are extremely promising,” said Greg Atwood, senior technology fellow at Numonyx. “The results show the potential for higher density, scalable arrays and NAND-like usage models for PCM products in the future. This is important as traditional flash memory technologies face certain physical limits and reliability issues, yet demand for memory continues to rise in everything from mobile phones to data centers.”

Intel and Numonyx plan to present a paper on their achievement at the International Electron Devices Meeting (IEDM 2009) on Wednesday, December 9th at 9am EST.
Last modified on 09 November 2009
Rate this item
(0 votes)