Implementation of an artificial spiking neuron with photoreceptor functionality using gas discharge tubes

Implementations of artificial neural networks increased power consumption by computing facilities. This can be solved by developing energy efficient neuromorphic hardware, where use of memristors is a promising route. Besides memristive synapse implementations, artificial neurons need to be created. Here, focus has been on developing threshold switches, among which perfect physical realization is yet to be found. In this work, we use gas discharge tubes (GDTs) as prototype threshold switching devices. We construct a leaky integrate-and-fire (LIF) neuristor, based on one GDT, and further extend its functionality by creating a Hodgkin–Huxley-like neuron (HH), based on two GDTs, capable of reproducing 24 neuronal modes. We exploit the light sensitivity of a GDT, caused by the photoelectric effect, and demonstrate light-induced inhibition of spiking in the LIF neuron, while in the HH-like neuron, we show the light-induced spiking threshold tuning and autonomous (without electrical input) spiking, combining photoreceptor and sensory neuron functionality. We propose the scheme to miniaturize gas discharge devices and reduce the operation voltage down to technologically relevant values. Employed approach offers an alternative path for developing modern neuromorphic electronic hardware and presents an opportunity to engage more researchers in the field by using off-the-shelf devices for neuronal circuit research.