Note: Coatue was bought by AMD recently
A Big Name in Chips Helps Coatue in Molecular Memory Race
August 27, 2002
With a little development help from chip giant AMD and funding from Draper Fisher and Jurvetson and ITU Ventures, the founders of Coatue believe they will be the first to market with a molecular memory intended to replace today's Flash and DRAM.
Coatue's long-term goal is to eventually combine all the memory needed to run a PC — DRAM, SRAM and hard disk — into a 3-D memory stack that sits atop the microprocessor. But for now, the Woburn, Mass.-based company will have to be content with putting a 1 megabyte prototype into the hands of portable device manufacturers by year's end and having a multi-gigabyte molecular memory on the market by 2004.
But why are founders Avi Goldberg, Andrew Perlman and Aaron Mandell so confident of success when there are at least six other companies (including Opticom's Thin Film Electronics, which is 13-percent owned by Intel) working on similar technologies with similar timetables and yet another group working on totally different technologies with similar timetables all focused on the same $70 billion dollar market?
Their answer is simple: Coatue designed its chips around industry standard 180-nanometer (nm) interconnects, which means Coatue's memory can be made in today's semiconductor fabs today.
"That is one of our key advantages — the manufacturing compatibility," Perlman, Coatue's CEO, told NanotechPlanet." That's why we think we're well positioned to be first in the market."
Coatue's memory works by sandwiching a sheet of conductive organic polymer just a few molecules thick between a grid of perpendicular electrodes (wires) that resemble a street map. Where these wires intersect, a memory cell is created.
"Just by passing a positive current to write and a negative current to erase we can get a very, very large change in the resistance from a very 'off' state, which is in the order of mega-ohms to a very, very conductive state that can be as high as hundreds of ohms," Mandell, Coatue's CTO, told NanotechPlanet.
The engineers at Coatue have also found a way to dramatically increase storage density buy doping their polymer. This allows them to vary the conductive state of memory cells over a very wide range. In this way, Coatue is able to get up to six bits of memory per cell. By comparison today's DRAM holds one bit and Flash memories hold just two. On top of this, Coatue's memory is non-volatile for up to four years; stackable, meaning its chips will be comprised of multiple layers of memory cells instead of just one like today's silicon chips; use less power than today's chips and be potentially much faster. Eventually, the company believes it can achieve terahertz switching speeds.
Switching speed is the amount of time it takes to write or erase one bit of information. Today's fastest memory chip is the 1066 RAMBUS, which has a switching speed of 1066 megahertz. This is faster than one billionth of a second per switch.
On top of everything else, Coatue's memory is about 15-to-20 times cheaper per bit to produce because it so dramatically increases bit density. This allows manufacturers to glean more memory chips per wafer thus reducing overall production costs.
It is these qualities as well as the company's management team that caught the attention of ITU, an early stage venture capital firm focused on university developed technology. In fact, ITU liked Coatue so much they climbed aboard even before the AMD deal came through last year.
"What the team at Coatue had aggregated we viewed as one the most promising technologies in the memory space," Andrew Murray, a principal at ITU, told NanotechPlanet. "That, in conjunction with management's ability to execute, made it an attractive investment."
Although not intimately familiar with the company, Steve Cullen, In-Stat MDR's director of semiconductor research, is also bullish on its technology and prospects. "Conceptually, it's very simple and the nice part about simple is it's potentially very cheap."
And with competitors that include Hewlett-Packard, Nanosys, Rolltronics, ZettaCore and Molecular Electronics Corp., being cheap is the name of the game, Cullen said. Cost is the No. 1 issue facing next-generation memory producers. Technological improvements aside, if it can't be done cheaper manufacturer's won't buy it.
"In order to compete with DRAM you are competing with people who are willing to lose money," he said. "There have been a number of things that have come around over the last 30 years or so that were all going to replace DRAM but one by one they fell by the wayside and the problem was they just weren't cheap enough."
This is why Cullen views AMD's involvement with the company as a huge boost to its potential success. The assumption is AMD wouldn't be backing a technology it believed was too expensive to produce competitively. But since Morris Denton, AMD's director of corporate public relations would not comment on the partnership, except to say they are always on the lookout for promising technologies, it is hard to tell for sure.
But it's the relationship with AMD that could ultimately lead to Coatue's success. According to Perlman, part of the deal allows AMD to get first crack at Coatue's technology for the DRAM and Flash products. After that, at some undisclosed point in the relationship, Coatue will be free to license its technology to others.
"With the right partner you can get a lot more (than just money)," said Perlman, "and for us, it was getting strong technical development cooperation."
After it has product to market in 2004, Coatue will begin pursuing its ultimate goal of becoming a fabless chip maker similar to SanDisk through the development and independent production of a private label memory chip it has dubbed Madaket after a town on Nantucket Island. "Our goal, as a company, is to build a chip company," Perlman said.
nanoelectronicsplanet.com |
