Nanocomposite parylene-C memristors with embedded Ag nanoparticles for biomedical data processing

Neuromorphic networks adopt human brain mechanisms and create a promising way to solve various artificial cognitive problems. Memristors, novel circuit design elements, could play the role of hardware synapses in such systems. Memristors based on parylene, safe material for usage within the human body, have already shown their eligibility for neuromorphic applications. The embedding of metal nanoparticles into the parylene layer leads to even better memristive characteristics, but the usage of such memristors in neuromorphic systems is still questionable. In this paper, the characteristics of nanocomposite poly(chloro-p-xylylene) memristors are studied, and the possibility of their training following spike-timing-dependent plasticity rules is demonstrated. Furthermore, a simple spiking neuromorphic network based on such memristors is successfully simulated to solve the task of breast cancer data classification. The results pave the way towards bio-compatible and energy-efficient wearable neuromorphic systems based on parylene memristors.