SANTA CLARA, Calif. – Belgian microelectronics research center IMEC has announced a joint development program with Nantero Inc. to make carbon nanotube non-volatile memories with critical dimensions of less than 20-nm.
Although IMEC has an active program researching metal-oxide based resistive RAM, the organization expressed the hope that following successful research carbon nanotube (CNT) memory could be deployed as a replacement for DRAM.
Nantero (Woburn, Mass.) has been developing carbon nanotube (CNT) memory, which it calls NRAM, since 2000. The company seemed to make significant progress in the middle part of the decade. On its website the company claims that NRAM will be considerably faster and denser than DRAM, have substantially lower power consumption than DRAM or flash, and be as portable as flash memory.
Individual CNTs are about a nanometer in diameter and can be up to a millimeter long. However, Nantero forms a non-woven matrix of CNTs that are deposited onto a substrate that contains an underlying cell select device and array lines.
In 2006 the company announced it had fabricated and successfully tested a 22-nm memory switch based on mat-like composition of CNTs laid across and etched trench. In this configuration the membrane-like matrix of CNTs displays a bi-modal stability with different resistance states.
IMEC (Leuven, Belgium) has agreed to support Nantero in the manufacture, test and characterization of NRAM arrays with a focus on high-density next-generation memories.
"After review of the progress to date by Nantero and its manufacturing partners, we decided that this CNT-based non-volatile memory has multiple very attractive characteristics for next-generation highly scaled memory," said Luc Van den hove, CEO of IMEC, in a statement. "By taking a leadership position in this area of development, in partnership with Nantero, we will be able to bring substantial benefit to our member companies."
Nantero has already fabricated high-yielding 4-Mbit arrays of NRAM in CMOS production environments, with several important performance advantages: write speed has been shown to be as fast as 3 nanoseconds; endurance is expected to be unlimited and has been tested so far to over 10^12 cycles, with low operating power and superior high temperature retention, according to IMEC.
"Together, Nantero and IMEC can develop and demonstrate this form of memory for future applications below 20-nm such as terabit-scale memory arrays and ultra-fast gigabit-scale nonvolatile cache memories," said Jo de Boeck, CTO of IMEC. "NRAM holds clear promise in aggressively scaled non-volatile memory applications and, if we can demonstrate the suitable endurance and speed specifications, NRAM could even provide an alternative for DRAM that is facing scaling limitations beyond 18-nm."
Greg Schmergel, Nantero's co-founder and CEO, said: "Nantero is already working with world-leading manufacturers, and working with IMEC will enable these efforts to bring carbon nanotube memory to market to move even faster."
Nantero was included in the first three iterations of the Silicon 60 list of emerging technology startups published in 2004 and 2005. The company has raised more than $31 million in three rounds of funding and is backed by Globespan Capital Partners, Draper Fisher Jurvetson, Charles River Ventures and Stata Venture Partners.
@Kris, what is being bragged about for this memory is usually what is bragged about for MRAM (like speed and endurance) rather than for RRAM or ReRAM (low power, speed, endurance) so I wonder if its electrical specifications would be similar.
thank you Peter...make sense that it takes that long, material science is hard, typically takes 10-20 years for a new material to be implemented commercially...tough business, writing an app for iPhone is quicker ;-)...but if they make it work the payoff would be enormous...anyone from Nantero interested in giving a talk on this technology at emerging technologies event in Whistler in 2013? details at www.cmoset.com, Kris
Please note the company has been ploughing this furrow since 2000.
So while they may have been first , they have also been trying for a long time.
They have also worked with a number of semiconductor partners including LSI and ON semi.
When something doesn't emerge it suggest this technology is not coming to market easily. On the other hand IMEC executives give a somewhat glowing reference form the technology. Looks like one or more of the major semiconductor companies is interested (perhaps after losing interest in PCM) and is asking IMEC to do some expensive rendering of NRAM in silicon.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.