Reconfigurable Cellular Array Architectures for Molecular Electronics

摘要

This report is a compilation of largely unpublished work pertaining to reconfigurable cellular arrays for digital computation. They bear resemblance to both cellular automata and cellular neural networks, with the attributes of field programmable gate arrays. They are of potential interest to nano-scale / molecular-scale electronics approaches due to their simple, periodic arrangement. As such they address three critical issues at the smallest physical scales: (1) low-interconnect demand; (2) defect tolerance; (3) simplified construction through non-lithographic approaches (such as chemical self- assembly). The report exposes many facets of these arrays, including the ability to directly model their structures with artificial neural networks, which can be trained to implement digital functions directly. The report is intended to represent a snapshot of work against a very difficult problem, rich- in future research exploration opportunities.