Read head exploits optical pressure
Researchers at the University of Bucharest and the National Institute for Research and Development in Microtechnology (both in Bucharest, Romania) have developed a new way of reading out multiple gray-level optical information using...
Researchers at the University of Bucharest and the National Institute for Research and Development in Microtechnology (both in Bucharest, Romania) have developed a new way of reading out multiple gray-level optical information using a dense microelectromechanical systems (MEMS) cantilever array. The technique, which was designed for application to a single-chip computer system combined with an optical-data-storage device, has the advantage of being parallel rather than serial and is capable of fast read times. This development should result in extremely high data rates—on the order of gigabits per second, its inventors say. Though they have yet to demonstrate their system in the lab, the elements it relies on have already been proven experimentally. The next step would be to put them all together.
The researchers were partly inspired by work being done at Carnegie Mellon University (Pittsburgh, PA) to build a computer system on a single chip using a MEMS device as the hard drive.1 In the hard drive, each of an array of probes addressed a set of memory locations on a magnetic medium. The disadvantage of the system was that the entire probe chip had to be moved with respect to the magnetic medium for all of the memory locations to be addressed.
In the Romanian design, the actual data storage would be performed optically and then read out using an array of tiny (on the order of hundreds of microns long) free-standing cantilevers that can be pulled into electrical contact with the rest of the chip using electrostatic force.2 The researchers realized that, if the cantilevers were made sufficiently small and pliable, they would become sensitive not just to electrostatic pressure but to optical pressure. Though optical pressure is very slight—on the order of nanonewtons for moderate optical powers of less than 1 W—it only takes a few attonewtons to bend a cantilever.
In the device, the cantilevers bend towards a stop layer on the chip; when they make contact with it, a current flows (see figure). The contact and noncontact states can therefore be thought of as logic 1 and 0 respectively. However, what allows the system to be sufficiently informationally dense to be of interest is the fact that each cantilever can read out an optical signal with multiple bits (a gray-level signal). Though these bits are read out in series, the entire access time is still extremely short.
The bias voltage—which electrostatically assists the optical beam by pulling the cantilever close to the contact point—is scanned from low to high. A low voltage causes the cantilever tip to be far from contact: thus a large optical power is required to push it down all the way. At a higher voltage, the opposite is true: only a small amount of light enables contact. Thus, the gray level of the optical beam can be read by simply noting at which bias level the current first flows.
According to their calculations, the entire process to read out, say, a six-bit (64-level) value—quite achievable electronically, the researchers claim—would be less than a microsecond, compared to a millisecond or two for flash memories. With a modest array of 128 × 128 cantilevers, this would result in a memory of 12 kbytes that could be read out at 7 Gbytes/s. To date, the largest cantilever array, built for atomic-force microscopy, contains millions of devices at a density of 1 million/cm2. Such a chip could provide a storage density of more than 700 kbytes/cm2. Though impressive, none of these figures represents upper limits.
But, though the readout device is well thought out, the optical system is still in question. Though the researchers have proven that none of the demands of the readout device are impractical, they have yet to work them through for a specific optical storage medium or system architecture.
- L. Richard Carley et al., J. Appl. Phys. 87(9), 6680 (May1, 2000).
- Daniela Dragoman and Mircea Dragoman, Applied Optics 42(8), 1515 (March 10, 2003).