Storage capacity is critical for electronic musicians; the more, the better. This is particularly true of random access memory (RAM), which is used to

Storage capacity is critical for electronic musicians; the more, the better. This is particularly true of random access memory (RAM), which is used to hold data being actively processed. As the resolution of digital audio continues to increase, so must the amount of available RAM. But as I've written many times, conventional semiconductor RAM is fast approaching the limits of its capacity, so other approaches must be found if our ever-growing appetite for storage is to be sated.

One solution could come from the field of nanotechnology, which is concerned with constructing the smallest imaginable objects. Among the first such objects to be built were carbon nanotubes — which are cylindrical structures made from hexagonal rings of individual carbon atoms. With a diameter of about 1 nanometer (a billionth of a meter), nanotubes look like rolls of chicken wire, and they exhibit extra-ordinary strength and electrical properties, making them ideal for a variety of applications.

For example, an unexpected application is a novel form of electromechanical RAM. Originally conceived by Thomas Rueckes, then a graduate student at Harvard University, the new type of RAM is now being developed by a company called Nantero (, which was founded by Rueckes, Greg Schmergel, and Brent Segal in 2001. The result of their efforts is called nanotube RAM, or NRAM.

The basic idea is that strands of nanotubes are suspended over tiny gaps etched into a semiconductor chip (see Fig. 1). The strands are about 13 nm above the floor of the gaps, which are about 130 nm wide. The gaps contain electrodes on top of transistors, which produce electric fields. When a transistor's field is oriented in one way, the nanotubes above it are drawn downward until they touch the electrode. When the field is oriented the other way, the nanotubes remain suspended above the electrode. Those two states represent 1 and 0, respectively, which is highly reminiscent of the mechanical relays in the antique computers that originally inspired Rueckes's idea. Amazingly, the nanotubes stay in position even when the power is removed, allowing them to retain their data like flash memory.

Despite the simplicity of the concept, there were several practical hurdles to surmount. One was the high iron content of commercially available nanotubes. Nantero devoted much of its early work to developing a special filtration process that brought the iron content down to the parts-per-billion range. Then came the challenge of depositing the nanotubes on CMOS (Complementary Metal Oxide Semiconductor) wafers. Gas-vapor deposition requires temperatures that destroy the ancillary circuitry, and a suitable conventional solvent to use in spin coating is too toxic, so Nantero came up with a proprietary, nontoxic spin-coating solvent.

Once those obstacles were overcome, it was time to look for a manufacturing partner. In 2003, LSI Logic agreed to try making NRAM in its Gresham, Oregon, plant, and nine months later, the company had a working prototype. Now the trick is to increase the yield to the point of economic feasibility. LSI wants to use the technology as a replacement for the static RAM embedded in ASICs (Application Specific Integrated Circuits), because NRAM has the potential to be much smaller while consuming less power.

Ultimately, Nantero's goal is to match the speed of static RAM, the data density of dynamic RAM, and the nonvolatility of flash memory. In the short term, the company expects to meet two of those criteria, with data density lagging behind. Current prototypes have capacities in the megabit range, though chips with up to a terabit of storage should be possible eventually. In addition, NRAM consumes less power than other forms of RAM, while remaining more resistant to temperature extremes and electromagnetic fields. What more could electronic musicians ask of their memory?