MULTILEVEL AND THREEDIMENSIONAL NANOMAGNETIC RECORDING - Bio-Inspired and Nanoscale Integrated Computing - page 180

and the anisotropy energy density, becomes comparable to the energy of thermal
activations, i.e., kT (which is approximately 25meV at room temperature).
Professor Stanley Charap and his graduate students at Carnegie Mellon
University were the first to use Monte Carlo-based numerical simulations to
estimate that the superparamagnetic limit becomes noticeable as the ratio of the
anisotropy energy and the thermal energy is reduced (approximately) below the
value of 40 [22]. For example, it was found that for the ratio value to increase from
40 to 60, the relaxation time changes from 72 seconds to over 3.6 10 9 years,
respectively. In other words, to avoid the superparamagnetic limit, the following
condition should be satisfied:

K u V
k B T >

40 60 ð 6 : 1 Þ

where K u and V are the anisotropy energy density and the volume of an average
grain, respectively. For this condition to be satisfied for typical CoCr-based
longitudinal media, the characteristic diameter of an average grain should be
greater than approximately 3 nm. Of course, one way to increase the ratio without
reducing the average grain size could be to increase the anisotropy energy density.
In fact, there are ultra-high anisotropy magnetic materials, such as L 10 and others,
that could be used to achieve the goal [23]. However, there is another fundamental
limit that makes it difficult to use these ultra-high anisotropy materials [24, 25].
A recording head is made of a soft magnetic material, and the maximum recording
field generated by the head is limited by the saturation magnetization of the
material. Unfortunately, the saturation magnetization in practical soft materials,
namely the 3D transition metals (Fe, Co, and Ni), is fundamentally limited by
approximately 26 kemu/cc, as described by the so called generalized Slater-
Pauling curve [26]. The curve shows that the magnetic moment per atom in
ferromagnetic alloys of the 3D transition metals could generally be predicated as
m=9.6 N d , where N d is the average number of d electrons per atom. This curve
works because it is energetically favorable to keep the majority-spin d-band full,
independent of alloy composition. Therefore, to further extend the areal density, it
is necessary to further reduce the grain size and thus further reduce the ratio in
Expression 1; this implies stepping in the superparamagnetic limit. The modern
laboratory demonstrations indeed indicate that the information becomes highly
unstable as the areal density is increased above approximately 200Gbit/in 2 [27].
That is why the multibillion data storage industry is currently searching for an
alternative technology that could extend scaling and thus areal data density much
beyond the current limits.

6.2.4.3. Temporary Solution/Patch in Layman's Terms. So called perpen-
dicular recording promises to defer the fundamental superparamagnetic limit for
several more years [28-32]. Ideally, this could bring the areal densities in
demonstration data storage systems to one Tbit-per-square-inch. Figure 6.4
illustrates the difference between the conventional

technology (longitudinal