Tuesday, January 25, 2005

Nantero sings a happy tune


PhysOrg.com is reporting a new deal between Brewer Science and Nantero to commercialize carbon nanotubes for nonvolatile memory in one more partnership in the traditional CMOS supply chain. It's part of Nantero's effort to remain all respectable and integrate with existing processes and products. You wouldn't want to frighten the semiconductor industry with news of its impending obsolescence. Dogs would lie down with cats and there would be panic in the streets.

As I wrote in a report for NanoMarkets last year, Nantero "has gone out of its way to make it clear that (it's) not some “crazy” nano company out to change everything by creating a manufacturing process that nobody else could reproduce. (CEO Greg) Schmergel’s marching orders to his people were that they were not to buy any 'weird equipment.'" The strategy works, of course, since Nantero has attracted some big venture capital bucks.

What's all the fuss about with Nantero? What the Boston-area company has laid claim to is its own first letter to add to RAM, and for some reason the letter N hadn't already been taken. So, NRAM it was, and here's how the nanotubes turn into memories: Suspend billions of nanotubes over a substrate that also contains tubes, then sound the call to prayer by flipping an electrical switch. The tubes on top bend in supplication toward the tubes on the bottom. And when they touch, they stay touched, even when the juice is turned off. Thank Mr. Van der Waals and his forces for that.

Why is this method of memory considered so great? It's because the tubes are, well, "nano," you can pack billions of them together, a kind of single-layer carpet of nanotubes upon nanotubes, and can be switched on and off in less than a nanosecond. And with Nantero's method, you don't have to worry about making all those billions of nanosoldiers line up correctly or even be the same size. The tubes are just scattered randomly, and regular semiconductor lithographic equipment will cut away the tubes that are not in the right place. Plus, we all know that nanotubes are strong and conduct better than Arthur Fiedler on the Fourth of July.

And according to Nantero, its process could theoretically conduct a memorable 10-gigabite symphony, a prototype of which has already been played. One potential problem, though, is that not all nanotubes were created equal. Some have different kinds of electrical properties that a random process just can't weave out – so there can be a few sour notes.

And still unanswered is how this data is read without eroding or destroying the information. The only explanation that's been given is that they're nanotubes, therefore incredibly strong and can't crack. That's never been truly put to the test, but I suppose it's safer to try it in computer memory than in a nanotube space elevator.

NanoBot Backgrounder
Revolutions happen to the unprepared
Memories, like the CMOS of my mind
Spintronics pioneer teaches online class


Anonymous said...

Hmmmm......sounds interesting. It would seem important though to have ensure that the distance spanned by the nanotube across the gap be need to be controlled in order for the electromechanical properties to be consistent.

This would suggest that alignment is truly needed, if standard mechanical models apply in the nano-realm. It would be interesting to see the data for this.....

Anonymous said...

I agree with the notion of the previous comment. If the length of the span that the nanotube is not controlled, then the electrical force required to bend the tube must be varying, which means that the switch voltage must be varying. A random array of nanotubes suggests the beam length is random, which suggests that the switch voltage must be random as well. One could design circuits to get a few bits in a large memory array to work, but it suggests a very low probability of most bits working.
On the other hand, is standard mechanical models don't work in this region, the above argument is not valid. Of course, if it is valid, then the fix would be to align the nanotubes.
Perhaps what is most likely is the company is doing nanotube alignment, and not telling anyone. That keeps the competition, like IBM, confused.

Anonymous said...

regarding the varying voltage , i feel we can maitain a standatrd size of tubes if we can remove tubes from unwanted area surely wea moderate their size using the same technique.

Anonymous said...

Notionally, the concept of removing the tubes, using conventional lithography and etch techniques, and maintaining the same diameter tubes, which is done at the laser ablation step prior to any semiconductor processing, are, in fact, controlled by two different processes. Hence, the statement posted above makes no intuitive sense.

Anonymous said...

Nantero seems to be reiterating the same thing - it has carbon nanotubes in a CMOS fab.....

When will it announce that it has a 1 megabit NVM working? Any insight?

Anonymous said...

My indirect sources at On Semi indicate that Nantero truly needs to align the nanotubes, despite what the CEO and CSO keep stating to the press.

Will be interesting to see if Nantero is really doing this (alignment) and throwing off the competition by suggesting a random assortment on nanotubes is sufficient, or if the management team at Nantero really has no clus as to the need for aligned nanotubes. Wish I was a fly on the BOD meeting room!!!

Anonymous said...

My sources at OnSemi suggest that Nantero is a flop....they in fact need to align nanotubes.

So much for a CEO whose claim to fame is a dot com start-up, and so much for the claim of a CSO whose claim to fame is to have his thesis advisor back another company (Nanosys) rather than his own start-up (Nantero)... any takers out there that Natero does not deliver a manufacturable NRAM in 2007?

Anonymous said...

Are you people serious? "Nantero is throwing off the competition by saying one thing and doing another" Trust me I'm sure the IBMs and Intels have all the resources at hand to do their on independent research

Anonymous said...


Seems to me that the January 2008 IEEE Spectrum has it correct - Nantero keeps promising a technology that can not be done in high volume manufacturing.

Of course, Brent Segal, Tom Rueckes, and Greg Schmergel may be telling the world otherwise, but it is clear that one cannot create order out of chaos, and in solid state memories, creating order and minimizing entropy are critical - and a random mat of carbon nanotubes is not minimization of entropy. Apparently, a couple of chemists in this company were not paying attention in thermodynamics class.

I concur with TJ Rodgers - both the government and the VC's have thrown money down a rathole. Interesting that the $30+ million invested in Nantero equates to $0.10 per American - ironically, the technology will prove itself to be not worth a dime. Mark my words.