The promises, perils for nanotech

Nantero is the kind of start-up that demonstrates the potential--and the lurking pitfalls--of nanotechnology.

This 10-employee company, based in Woburn, Mass., is aiming to revolutionize the memory business by replacing the traditional silicon and metal circuits inside conventional chips with a lattice of carbon nanotubes.

Ultimately, the use of these thin strands of carbon atoms, which are endowed with extraordinary properties, could lead to memory chips that would be cheap, extremely fast and likely to keep going far longer than anyone who buys them.

Nantero says it has invented a way to implant the nanotubes onto conventional silicon wafers using existing chipmaking equipment--a breakthrough that ideally could keep manufacturing costs low. In May, the company announced it had created a working 10-gigabit array of carbon nanotubes on the surface of a conventional 4-inch wafer.

"A local (chip fabrication facility) went under, and we got some of their equipment cheap," Nantero CEO Greg Schmergel said.

On the positive side, Nantero again underscores the notion that, despite ongoing skepticism, nanotechnology is for real.

Briefly put, nanotechnology is the science of making products out of components that measure 100 nanometers or less. (A nanometer is a billionth of a meter). This is more than "small for small's sake." At these dimensions, matter begins to behave differently. Electrical conductors, for example, can become insulators.

In this world, the carbon nanotube is the supreme celebrity. Invented by scientists at Japanese tech giant NEC, nanotubes are 1,000 times better at conducting electricity than is copper, more than 50 times stronger than steel and can insulate better (or dissipate less heat) than can diamonds, the world's best insulators.

To semiconductor companies, new and revolutionary technologies often represent new and revolutionary methods for losing billions of dollars.

They also bend, a property crucial to Nantero's approach. In the company's chips, parallel rows of carbon nanotubes are embedded in silicon. Suspended above them are perpendicular rows of nanotubes. When the upper and lower tubes aren't touching, electrical resistance in the upper tube is fairly high. (The tubes are all ultimately plugged into wires.) The computer reads the relatively high electrical resistance of the nanotube and interprets this as a 0.

Applying a small electrical charge, however, causes the top carbon nanotube to bow--like a guitar string on a fret--and, through van der Waals forces, bond with the lower tube. The bonding lowers the electrical resistance, which is interpreted as a 1. Millions of microscopic junctures--the meeting points between the carbon tubes--serving up 1s or 0s then lead to a digital photograph, or to any other complex data file.

In its experimental wafer, 98 percent of the junctures worked, Schmergel said. The wafer was heated to 200 degrees Celsius, and it still worked, he added.

While a lot more work needs to be done, the results are promising, he said. By late 2004 or early 2005, the company hopes to have 64-kilobit prototype memory chips out to potential licensees. (Like many semiconductor start-ups, Nantero will likely license its technology, rather than make chips itself.)

Now, the bad
However, major pitfalls also exist. For starters, Nantero is promoting carbon nanotubes. To semiconductor companies, new and revolutionary technologies often represent new and revolutionary methods for losing billions of dollars. In 2002, Advanced Micro Devices attempted to include "silicon on insulator," a design technique for reducing power consumption, in its chips. The result? One product was redesigned, and others were delayed.

Invite Michael Kanellos into your in-box Senior department editor Michael Kanellos scrutinizes the hardware industry in a weekly column that ranges from chips to servers and other critical business systems. Enterprise Hardware every Wednesday.

Of all the technological changes you could foist on a semiconductor maker, a new material such as carbon is the scariest, according to various researchers. There are just too many unknowns and variables, which is why many of the established companies say that they won't be looking at replacing traditional silicon technology for a decade or so. And this makes establishing a quick revenue pipeline difficult for a company like Nantero.

Going from the lab to the plant won't be easy, either. How do you plant billions of tubes on a 12-inch wafer in a mass-manufacturing environment? Where are we going to buy nanotubes? Right now, one of the main suppliers is Carbon Nanotechnologies, which charges $500 a gram on its e-commerce site. And what do we do with all the people trained in silicon development and manufacturing? These are just some of the questions chipmakers are going to ask.

There is yet another potential problem. Schmergel and two chemistry Ph.D.s from Harvard--Thomas Rueckes and Brent Segal--founded the company. Right now, nanotechnology experts--especially those at larger companies--tend to complain about the same thing: That too many academics are trying to commercialize their lab results too early. Although Rueckes and Segal are undoubtedly brilliant, they will face mounting skepticism.

Schmergel is also new to chipmaking. Before Nantero, he founded some Internet companies and spent time at consulting firms. This is the sort of resume that makes chip people cringe. Semiconductor execs generally hail from three foreign countries: China, Japan and Pennsylvania. They are usually engineers by training and had to claw their way up through the ranks of plant management, sales or chip design.

In other words, it's a slightly insular society. Regardless, Nantero could still succeed in the end.