Chaotic dynamics in nanoscale NbO2 Mott memristors for analogue computing

Title	Chaotic dynamics in nanoscale NbO2 Mott memristors for analogue computing
Authors	Suhas Kumar, John Paul Strachan, R. Stanley Williams
Magazine	Nature
Date	08/09/2017
DOI	https://doi.org/10.1038/nature23307
Introduction	Current machine learning systems often employ simplified neuron models, lacking the intricate nonlinear phenomena present in biological systems which exhibit spatio-temporal cooperative dynamics. Evidence suggests neurons may function at the 'edge of chaos,' a state crucial for complexity, learning efficiency, adaptability, and non-Boolean computation. Neural networks demonstrate enhanced computational complexity at this edge, and chaotic elements have been proposed for solving optimisation problems. The development of a controllable chaotic behaviour source, integrable into a neural-inspired circuit, could be vital for future computational systems. Previously, chaotic elements were simulated using complex transistor circuits, but a scalable electronic device demonstrating chaotic dynamics remained unrealised. This study introduces nanoscale niobium dioxide (NbO2) Mott memristors, smaller than 100 nanometres, exhibiting nonlinear transport-driven current-controlled negative differential resistance and Mott-transition-driven temperature-controlled negative differential resistance. Mott materials, with metal-insulator transitions dependent on temperature, act as electronic switches, providing history-dependent resistance. These memristors were integrated into a relaxation oscillator, revealing a tunable range of periodic and chaotic oscillations. Such memristors could enhance neural-inspired computation by generating pseudo-random signals, preventing global synchronisation, and aiding in global minimum searches during constrained optimisation. Specifically, incorporating these memristors into Hopfield networks can significantly enhance convergence efficiency and accuracy in solving computationally challenging problems.
Quote	Suhas Kumar, John Paul Strachan and R. Stanley Williams. Chaotic dynamics in nanoscale NbO2 Mott memristors for analogue computing. Nature. 2017. DOI: 10.1038/nature23307
Element	Niobium (Nb)
Materials	Oxides , Chemical Compounds
Topics	Nanotechnology and Nanomaterials , Machine Learning in Materials Design