Chaotic computing cell based on nanostructured phase-change materials

This paper presents and investigates a new architecture of a computational cell based on nanoparticles of the phase-change material Ge₂Sb₂Te₅. Such a cell is a chaotic array of nanoparticles deposited between closely spaced electrical contacts. The state of such a structure is determined by the resistance of the nanoparticle array, which depends on the phase state of each particle of the material. Simulation results show that the proposed structure has a number of electrical states switching features that cannot be achieved using a thin film architecture. The proposed architecture allows for smoother and more controlled switching of the resistance by electrical pulses. Simulation of the evolution of the cell state using complex control actions showed that the proposed structure can behave as an artificial convolutional neuron with horizontal connections and also as a multi-level memory cell. In addition, the proposed design is technologically simple to achieve and inexpensive to manufacture.