Recently, self-powered optoelectronic synaptic devices
have attracted
great attention due to their bias-free and self-rectifying properties
for future computing systems. However, high energy consumption may
still be required to generate optical signals for the stimulation
of the systems. In this work, self-powered optoelectronic synaptic
devices are fabricated based on triple mixed cation perovskites with
excellent synaptic stability. Various synaptic functions are mimicked
in these devices, which are stimulated by a fully visible light spectral
range. These devices demonstrate the lowest energy consumption density
and the best consistency properties of all reported self-powered optoelectronic
synaptic devices to date. Color classification and speech recognition
are successfully implemented with high accuracy in these systems.
The results significantly promote the development of self-powered
systems in neuromorphic computing.