Lead Halide Perovskites for Low-Energy Consumption Neuromorphic Applications

Lead halide perovskites are highly promising materials for a wide range of optoelectronic applications, such as photovoltaics and LEDs. One major challenge for these applications is the efficient conduction of mobile ions in these materials, which causes unwanted hysteresis when a bias voltage is applied. However, this property of halide perovskite materials makes them highly interesting for applications in neuromorphic computing, where large resistance changes of a device upon applying a bias voltage are desirable. Specifically, the efficient conduction of ions can be utilized for bringing about non-volatile resistance changes to change synaptic weights or volatile resistance changes to mimic the firing of a neuron, all while expending only little energy. Here, we present our recent work on applying halide perovskites in devices for neuromorphic computing. We focus on addressing the challenge of downscaling of halide perovskite devices due to the high solubility of halide perovskites in polar solvents that are commonly used in lithography. Using our procedure, we demonstrate microscale halide perovskite artificial synapses and the first fully on-chip integrated halide perovskite artificial neuron. Owing to the small device size, the energy consumption of the synapse was in the sub-picojoule range, while the energy consumption of the neuron was on the order of tens of picojoules per spike. Both of these energy consumptions are similar to or even lower than those of analogous biological processes. Our design lends itself to further downscaling and we discuss how this would reduce the energy consumption even more. Moreover, the similarity of the artificial neuron and synapse device design allows easy integration in ultralow-energy consumption neuromorphic chips. These chips could potentially emulate the analog and parallel way that information is processed in the brain to achieve orders of magnitude lower energy consumption computation compared to digital computers.