Neuron-like transient phenomena in silicon p-i-n structures

Abstract Simulation of neuron transient phenomena by electronic processes in semiconductors is discussed from the point of view of hardware for new approaches to electronic processing of information which parallel the means by which information is processed in intelligent organisms. Neural networks in intelligent organisms are huge compared to supercomputers but the electrical processes are much slower because they involve ionic conduction. Recent work has uncovered a class of transient semiconductor phenomena which exhibit similarities to the way in which information is coded by neurons. The typical speed of the observed transients has been slow by the standards of electronic switching in computers. New results for such transients in silicon p-i-n structures show much faster switching times. Experimentally demonstrated features include logical operations, bifurcation, chaos and threshold behavior of firing. Further development along these lines could lead to fast, dense, neuronlike microelectronic networks that are simpler than artificial neural networks based on transistors.