Early Electronics Applications of Carbon Nanotubes

Better control of the processes for making and handling carbon nanotubes (CNTs) means they are now actually starting to show up in some early electronics applications. Suppliers of CNT gas sensors and atomic force microscope (AFM) probe tips now on the market are growing the tubes in situ with chemical vapor deposition (CVD) processes. Progress toward consistent and well-characterized CNT solutions for wet deposition of random networks is propelling the development of transparent conductive films and other biomedical sensors. And those developing CVD processes say they are making strides toward better control of properties for eventual use in semiconductor interconnects as well.

Nanomix Inc. (Emeryville, Calif.) has a commercial hydrogen sensor that warns of the danger of explosion on the market, and another industrial gas sensor and a CO₂ sensor for respiratory monitoring in the final stages of product development. However, Bill Perry, marketing vice president, said that the company is now concentrating its efforts on the much larger market for low-cost nitric oxide sensors for asthma monitoring, targeted for 2009. The gas is a biomarker for the airway inflammation that marks the progression of asthma, so people could easily monitor their condition to adjust their medication to avoid attacks.

1. A network of carbon nanotubes (CNTs) on a silicon surface can be used to detect gases such as carbon dioxide. (Source: Nanomix)

Nanotubes (Fig. 1 ) make potentially good sensors because they are already on the same nanoscale as many of the molecules and proteins of interest, and they create an exposed conducting channel, unlike CMOS, that can be in direct contact with the environment and all of the current flows on the surface of the tubes. Put particles that react with a target substance on this sensitive conductive framework and the electrical characteristics change in target-specific ways.

Nanomix's initial products are for very small niches, but the company has churned out well over a million of its basic chips with networks of CNTs sensitized for different materials (Fig. 2 ). It grows the random networks directly onto 150 mm wafers with a commercial CVD tool, and then sends the wafers out to a commercial foundry to add the electrode pattern on top. The chips are mounted on small circuit boards with the necessary supporting circuitry to complete the sensor. Perry said that the company could control the density, thickness and length of the tubes and, to some degree, the concentration of semiconductive vs. metallic CNTs as well, to adjust the sensitivity of the network. But what matters most is having a consistent process, so that the average electrical characteristics are known.

2. ‘Nano­maze’ of CNTs. (Source: Nanomix)

Going forward, however, Nanomix is working on a move that uses flexible substrates with printed electrodes for a range of biomolecular detections, including DNA, bacteria, viruses and glucose. The aim is to bring down costs, but to also speed up development. "You can certainly iterate faster if you're not using silicon," Perry noted.

Also now casting its lot for wet coating of random CNT networks for unique new medical sensors is the start-up Alpha Szenszor (Woburn, Mass.). Although cofounder Charles Lieber has been known to argue that nanowires allow for much better control of properties than CNTs, and his lab at Harvard has done much work on nanowire sensors, the company will now license Nantero Inc.'s (Woburn, Mass.) solution coating technology for random network CNTs. The company expects to have its first product for convenient low-cost detection of disease on the market in a very short time frame, and aims for eventually detecting biomarkers for everything from HIV and cancer to heart disease.

In August, Brewer Science Inc. (Rolla, Mo.) began selling a consistent semiconductor-grade CNT solution developed with Nantero. With this solution, researchers can take the material into a fab and begin trying out production of some of those intriguing one-off CNT devices they have been developing in the lab. Nantero has shown that the solution can be applied with a standard track and processed with technologies similar to conventional CMOS, but it is now also working with Hewlett Packard Co. (Palo Alto, Calif.) on thermal ink-jet printing of non-volatile CNT memory layers on flexible substrates for applications such as low-cost RFID tags.

Brewer Science will buy commercial CNTs from a number of qualified suppliers. It will purify them to >6 months. Characterization showed consistent films across the substrate and time at different densities. "We're showing data that it's fab compatible, and that the purification is a repeatable process," said Greg Schmergel, Nantero president. "Before this announcement, typically if you bought some of the material out there on January 1, and more on January 2, it would have different characteristics." The solution, however, like the available CNT supply, remains a mix of metallic and semiconducting nanotubes. Schmergel said that the availability of a material that can be used in the fab is generating interest in new CNT development projects. "Before, they knew that even if they spent the effort to develop a new device, they could never manufacture it," he noted.

Nantero has also developed a metrology method to make sure that the random CNT network does, in fact, make a continuous connection. It checks for conductive paths that light up where the metal CNT contact is charging under the SEM beam. Areas where the CNT film does not make connection to the metal do not show up in the SEM image.

Eikos Inc. (Franklin, Mass.), meanwhile, is demonstrating scalable production technologies for its transparent conductive films of CNT networks infused with binder that it said meet industry standards for electrical and optical performance and environmental stability. The company said that the printed CNT films may be better suited to flexible applications rather than the more brittle indium tin oxide (ITO), and look likely to be a viable lower-cost alternative within the next year or two, particularly for applications such as touchscreens and electronic paper that need patterning.

Eikos purifies CNTs, disperses them in a variety of solvents, prints them with a variety of conventional printing technologies on a range of substrates, and then infuses them with a polymer binder. The material has a sheet resistance of 100 Ω/sq with ~90% light transmission for films 550 nm thick, roughly similar to ITO for less demanding applications, said Chris Weeks, former business development manager for the product.

"Carbon nanotube's films are beginning to be equal or lower cost for some transparent conductive applications, especially where value-added patterning is needed," Weeks said. Although the single-walled carbon nanotubes (SWNTs) themselves are still expensive, only very small amounts are needed — the film can be as thin as 30 nm for 500 Ω/sq — and the films can be applied and patterned with low-cost printing processes. If the fragmented market for SWNTs consolidates as expected, and some big suppliers start producing in larger volumes, prices are likely to come down sharply over the next several years. "When single-wall tubes get down to the current cost of multi-wall tubes, CNT conductive films will be the lowest-cost alternative, even for antistatic coatings," he said.

"CNTs seem to be able to solve some real world problems," concured Lawrence Gasman, principal analyst at Nano Markets (Glen Allen, Va.), noting the better performance on flexible substrates than ITO and better durability outdoors than OLEDs for displays, signage and lighting.

Tight control of CNT size and alignment by in situ CVD growth is enabling production of commercial AFM probe tips by Xidex (Austin, Texas). The company said it promises tip angles within ±1° in each direction to ensure the probe can reach down into deep features. It is also selling tips with slightly lower tolerances at lower prices for researchers interested in simply higher resolution. CEO Paul McClure said that the company has now seen repeat sales in both markets — for measuring roughness, where the CNT tips offer better wear resistance, and inspecting high-aspect-ratio trenches, where only the long, thin tips can reach. "At 45 nm and 32 nm, there won't be an alternative in the scanning probe world to get inside these deep features," he noted.

Xidex worked with Sematech (Austin, Texas) to develop the tips, which it grows in a batch process directly on commercially available silicon cantilevers, an approach that aims to be scalable to eventually bring down costs. Xidex uses its own proprietary CVD process, presumably with technology related to its patent for growing, or at least starting, the CNT inside a precisely etched or micromachined hole to control its alignment and size, and then etching away that controlling template, if necessary. The company recently received an NSF grant to further develop precise site-selective CNT etching with a gas precursor-assisted e-beam-based process to improve the quality and yield of its process. The technology for cutting, cleaning and, where not wanted, removing CNTs would also be available for CNT fabrication and repair.

Xidex is also working on applying its controlled CNT growth technology to develop nanotube emitters for SEMs, looking to the electron field emission technology (FET) to improve brightness and resolution.

CVD tool supplier Atomate Corp. (Simi Valley, Calif.) argues that growing the CNTs directly on the substrate allows for better control of their properties. It said that the CVD process has recently been making rapid progress toward more consistency and higher percentages of single-walled and semiconducting tubes in the mix for semiconductor applications. "We shoot for deliveries within a year of some very simple high-power-density semiconductor devices using CNTs, with a number of different partners," said CEO Brian Lim.

Lim said that the company has started building 4 in. systems for research groups working on scaling their technologies, and customers are starting to talk about 6 and 8 in. systems. It is using a remote plasma system that favors the growth of semiconducting CNTs, and Lim said that he believes the lab results of 100% semiconducting CNTs can be scaled. Experience learning how best to adjust all of the usual variables, from gas flow rates and switching to temperature and vacuum, as well as better control of the catalyst and more uniform gas distribution across the wafer, is providing better consistency and control of defects.

However, the high process temperatures and metallic catalysts still make the in situ deposition incompatible with the CMOS processing required for volume manufacture. Lim said SWNT growth is now down to at least 500°C, and work is progressing both on new catalysts and containment systems, but the option of changing the CMOS process should not be ruled out. "I think CNT growth will eventually be part of a cluster tool," Lim said. "But to take advantage of some of the phenomenal properties of CNTs for some applications, there might have to be some compromises, like making the CMOS processes a little higher temperature or more tolerant of contaminates."