Project 1: Spin Clusters and Nanomagnetic Information Processing
1.2. Nanomagnetic Spin-Logic Systems
The goal of this project is to study and develop new spin-logic
structures for information processing. These devices involve associative architectures
where logic processes are realized by neuromorphic networks, sparsely-connected
neural networks, and feedforward layered structures (perceptrons). The basic
principle is similar to neuromorphic networks used, e.g., in Hopfield-type parallel
pattern recognition. It is very similar to the functioning of the human brain;
the two basic states of a neuron (quiescent and firing) are analogous to the
two states of binary logic (0 and 1) or equivalently to the magnetization of
a nanodot (spin-up and spin-down). We plan to realize simple logic operations
by exploiting the switching behavior of nanodots.
Skomski, Sellmyer and Liou are varying the dot composition so as to tune the
anisotropy of the dots (perpendicular/in-plane, two-fold/four-fold, hard/soft)
and the switching behavior; the magnetostatic and exchange interactions between
the dots will be modified by choosing appropriate dot positions and matrix material
or substrates, respectively. As these dots and their interactions are understood,
digital logic gates will be constructed using the focused-ion-beam (FIB) system
to prepare the dots, with thermal annealing in the presence of a local magnetic
field provided by an MFM tip. A particularly simple example of a NOT-AND (NAND)
logic gate is shown in the figure.
 |
| Logic-gate structures: (a) calculated spin structure of a dot having a size slightly larger than the critical single-domain size, (b) well-defined binary states in an array of somwhat smaller dots, and (c) preliminary expermental realization using FIB. |