Prototyping a memristive-based device to analyze neuronal excitability

Abstract

Many efforts have been spent in the last decade for the development of nanoscale synaptic devices integrated into neuromorphic circuits, trying to emulate the behavior of natural synapses. The study of brain properties with the standard approaches based on biocompatible electrodes coupled to conventional electronics, however, presents strong limitations, which in turn could be overcame by the in-situ growth of neuronal networks coupled to memristive devices. To meet this challenging task, here two different chips were designed and fabricated for culturing neuronal cells and sensing their electrophysiological activity. The first chip was designed to be connected to an external memristor, while the second chip was coated with TiO2 films owning memristive properties. The biocompatibility of chips was preliminary analyzed by culturing the hybrid motor-neuron cell line NSC-34 and by measuring the electrical activity of cells interfacing the chip with a standard patch-clamp setup. Next, neurons were seeded on chips and their activity measured with the same setup. For both cell types total current and voltage responses were evoked and recorded with optimal results with no breakdowns. In addition, an external stimulation was applied to cells through chip electrodes, being effective and causing no damage or pitfalls to the cells. Finally, the whole bio-hybrid system, i.e. the chip interconnected with a commercial memristor, was tested with promising results. Spontaneous electrical activity of neurons grown on the chip was indeed present and this signal was collected and sent to the memristor, changing its state. Taken together, we demonstrated the ability of memristor to work with a synaptic/plastic response together with natural systems, opening the way for the further implementation of basic computing elements able to perform both storage and processing of data, as in natural neurons.