Data storage: Playground for disruptive technologies
09/01/2006
The urgent need to push storage densities and access speeds for memory is even more intense than that faced by semiconductor technologists who have miraculously followed Moore’s Law for nearly 50 years. While semiconductor fabs have managed to double the number of devices per area every 18 months or so, storage technology developers have moved even faster, nearly doubling storage densities on disk drives every year for over a decade after developing giant magnetoresistance (GMR) technology.
The shrinks in both sectors are critical to keeping the economic engine churning, allowing prices to drop for existing products so that volumes swell and at the same time steadily opening up an ever-wider array of new applications.
Semiconductor developers have faced hundreds of “red brick walls,” such as the impossibility of patterning features smaller than the wavelength of the light in their steppers. Yet they have crashed through, with clever techniques like phase shift masks, optical proximity correction, and off-axis illumination, so that 90nm devices can now be mass-produced on 300mm wafers using 193nm lasers. By putting a little DI water between the lens and the wafer, they plan to push this to 45nm at least.
But over on the data storage side of the labs, the scientists and engineers face even tougher challenges. Disk-drive manufacturers are notorious for killing each other off with rapid price declines, and DRAM manufacturing is not much different. Tension runs high and impossible demands are commonplace. In other words, it’s a dreamland for the most innovative scientists and engineers! For example, they pioneered using the spin of electrons rather than charge for storage, moving from GMR to colossal magnetoresistance (CMR) and beyond to keep pushing the envelope. Clever statistical tricks enabled them to squeeze even more bits onto each inch of a track, and they figured out how to ride heads right on disk surfaces without triggering thermal aspersions. Scientists and engineers have made microdrives so rugged they can be used in portable consumer devices and can get knocked around without failing.
Solid-state memories, particularly flash, have surged ahead with dozens of innovations allowing multi-bit cells, deep trench capacitors to hold charge longer, and stacked capacitors. As current CMOS technology hits limits, memory developers have an assortment of potential disruptive solutions lined up, with phase-change memory the most likely to be used first. This works like the stored bits on a DVD, which switch from crystalline to amorphous and back again by varying the energy of laser pulses, except current is used to trigger the phase changes rather than light pulses.
As bit density on disks approaches the superparamagnetic limit, where thermal agitation of a few grains can cause the magnetic charge to flip, why not just pattern single domain-sized grains on a disk surface? While the idea seems outlandish-beyond the resolution of today’s optical stepper/scanners-it’s not stopping storage developers heading toward densities of terabits/square inch. Zvonimir Bandic et al. (see opposite) explain this process, which may result in the first volume market for nanoimprint lithography.
While fast-growing electronics markets are certainly due in large part to the efforts of semiconductor device developers, we can’t forget to give a lot of credit to the data storage technologists as well. Thanks for all the great memories!
 |
Robert Haavind
Editorial Director