Numonyx Memory Solutions

Discover the Next Transformation in Memory

By Clifford Smith
Emerging Technology Manager

In 1970, Gordon Moore, R. G. Neale and D. L. Nelson published a paper on amorphous memory titled, “Non-volatile, Re-programmable, Read-Mostly Memory is Here.” Thirty-eight years later, amorphous memory devices are not yet shipping in volume.

Were these authors wrong about amorphous memory? The answer: “No, just ahead of their time.”

Back in the 1970s, phase change memory (PCM) devices required a significant amount of power to drive the physical state change between an ordered, crystalline state and a disordered, amorphous state. In contrast, floating gate and capacitor-based memory technologies, which trap and store electrons, required less power and were cheaper to manufacture on the 10um lithographies available at that time. The result: SRAM, EPROM, EEPROM, NOR flash and NAND flash technologies moved beyond the R&D phase and into high volume manufacturing. Industry analysts report these memory technologies surpassed the $60B revenue mark in 2007.

PCM technology has the ability to take the best features of existing memory technologies and combine them into one device.

So why the resurgence in PCM now? As existing memory technologies shrink below 65nm lithography, they face scaling challenges which can be represented by a simple “glass and marble” analogy. Today’s memory cells act like a glass jar by holding marbles–fill the glass with marbles to store a “1” and empty the marbles to store a “0.” In the past, millions of electrons (aka marbles) were stored in flash floating gates or RAM capacitors (aka the glass jars). Due to shrinking lithographies, today’s memory technologies now store hundreds of electrons, which makes it more difficult to meet the reliability needs of existing applications. To summarize, the “glass jars” are getting too small.

In contrast, PCM devices do not store electrons, therefore the glass and marble analogy does not apply. A better analogy for phase change memory is to envision a body of water: freeze it to generate a crystalline (ice) structure or “1” and melt it to generate an amorphous (liquid) structure or “0.” In the 1970s, driving the physical state change of a chalcogenide material (the amorphous memory) was akin to freezing and melting a large body of water–think Pacific Ocean size. With today’s 65nm technologies, the volume of water is much smaller–think of your favorite nearby lake (Lake Tahoe if you live in California), and therefore the energy required to change the state is less. In short, today's PCM devices are more efficient in meeting power requirements compared to EEPROM, RAM and flash memories.

The next phase for PCM is to put it into the hands of customers. In addition to offering lower costs, PCM technology combines the best features of today’s memory technologies. Specifically, PCM is non-volatile like flash memory, plus it is bit/byte alterable like EEPROM devices. And, it offers read and write performance approaching that of today’s DRAM products. Because of this feature set, memory industry pundits dub PCM as the “unified memory.”

Numonyx 128Mb PCM Prototype on 90nm Technology In February 2008, Numonyx began shipping samples of its 90nm 128Mbit PCM to customers. With this milestone, Numonyx is enabling customers to take the next step in unifying their memory requirements and begin developing new applications and uses which benefit from phase change memory.

At Numonyx, memory is all we do. And we take pride in leading development of new memory technologies and solutions. We would agree with Mr. Moore, Mr. Neale, and Mr. Nelson–phase change memory is here!