Feasibility of lorentz readout of a high-density fast magnetic memory

A random-access magnetic-film memory with a desired bit density ρ
L
of 10
⁸
/cm
²
is postulated with writing and reading each to be performed with a deflectable electron beam within 1 μs. Curie point writing is suggested, while detection of the deflection of electrons due to the Lorentz force is proposed for readout. The transmission, reflection, mirror, and impact scanning methods of Lorentz microscopy are each explored as readout methods. Because of the absence of a heat-absorbing substrate, the transmission method is completely impractical, while the desired value of
\rhoL
is unattainable for the mirror and impact scanning methods because of the fields from other bits (ρ
L
must be reduced to
4 \times 10^{6}
/cm
²
for these methods). Moreover, the small attainable Lorentz deflection angles and the low brightness of conventional electron emitters necessitate further reduction of the bit density to values much less than ρ
L
for the reflection and mirror methods, unless a field-emission source is used. Excessive power dissipation forces a reduction of density below ρ
L
for the impact scanning method. The reflection method may be feasible if a sufficiently high fraction of elastically scattered electrons can be demonstrated, if a reliable field emission source can be employed, and if the sophisticated electron optics involved can be built and installed in ultrahigh vacuum.