Characterisation & modelling of perovskite-based synaptic memristor device

Neuromorphic computing architectures are required to execute several operations such as forgetting and learning behaviours with high-speed data processing. Due to the rapid advancement in technology, various transistor-based devices like field-effect transistor (FET), complementary metal-oxide-semiconductor (CMOS), etc. have the limitation to perform efficiently with a higher density of integration in combination with lower energy consumption. Consequently, there is a strong necessity for creating new devices with fast information storage, high-speed data processing, high density of integration, and low operating energy. Memristors are emerging as promising candidates as the next-generation technology which contains all the above-mentioned properties. According to previous literature, a nanoscale memristive device based on methylammonium lead iodide perovskite (CH3NH3PbI3) can be fabricated and characterised as a low power synaptic device. This study proposes the behavioural modelling of a perovskite-based synaptic memristor device with Glass/indium fin oxide (ITO)/SnO2/CH3NH3PbI3/Au structure for SPICE simulation in neuromorphic applications. We report an in-depth analysis of the physical model behind the creation of the p-i-n structure, induced by the ion drift in the perovskite layer. Furthermore, a SPICE Model is proposed to reproduce the observed behaviour of fabricated Glass/ITO/SnO2/CH3NH3PbI3/Au device and is able to mimic the neuromorphic learning and remembering process, similar to biological synapses. The proposed SPICE model will foster the potential of perovskite based synaptic devices by enabling large-scale circuit-level simulations thus allowing designers to explore the potential of this new device, for example in power-on-chip approaches and in an artificial neural network.