Terabit-per-square-inch hard drive areal densities, the goal for present
magnetic recording research, will probably not be achieved with giant
magnetoresistance (GMR) or with tunneling magnetoresistance (TMR) technology.
Much of applied physics is transduction--turning one thing into another
thing--and in the case of GMR this means sensing the orientation (0
or 1) of tiny magnetic domains with heads consisting of a sandwich of
alternating thin layers of magnetic and nonmagnetic atoms. The domain's
magnetism is transduced into a change in electrical resistance.
The multi-billion-dollar GMR industry is successful [with commercial
(areal) densities of about 15 Gb/in2] but, according to Stuart
Solin of NEC Research Institute, may not exceed an areal density of
about 150 Gb/in2 because of limitations imposed by magnetic
noise.
Solin's answer is to use the magnetoresistive effect in a nonmagnetic
semiconductor-metal composite read head, which is immune from magnetic
noise, has a faster response time than with GMR, and would entail a
fabrication process of perhaps 20 steps, as opposed to 200-300 steps
for GMR (or TMR, in which spins can "tunnel" from one orientation
to another). He believes that GMR or TMR will continue to furnish efficient
read heads until such time as higher areal densities are desirable,
at which time alternative approaches, such as the nonmagnetic sensor,
might be viewed favorably.
Solin (609-951-2610, solin@research.nj.nec.com) and his colleagues
are reporting the first mesoscopic device to employ the "extraordinary"
magnetoresistive (EMR) effect in a nonmagnetic sample-a Si doped InSb
quantum-well structure (see figure).
The areal density for this sensor exceeds 100 Gb/in2, while
the power signal-to-noise ratio is compatible with practical requirements
in working read heads. (Solin et al., Applied
Physics Letters, 27 May 2002.)