Publication date: 15th December 2025
Metal halide perovskite materials have been widely explored in the field of photovoltaics over the past two decades. Recently, they have also been gained traction in optoelectronic memories for neuromorphic applications, such as dynamic machine vision systems [1]. Their tuneable optical properties, solution-processability, compatibility with flexible substrates and their mixed electronic-ionic conductivity, especially their slow ion migration that gives rise to a significant current hysteresis, has brought them at the forefront of beyond-CMOS materials for emerging memories with neuromorphic functionalities.
Herein I will present results from our work with Bismuth-based perovskite films in both conventional sandwich device structures [2] and in a novel coplanar nanogap device configuration [3]. We tune the cation and the solvent to move the technology towards greener and more sustainable manufacturing avenues. Our 2-terminal devices show memristive characteristics that can be controlled both electrically and optically, showing both long- and short-term plasticity. We demonstrate learning and forgetting, potentiation and depression, paired-pulse facilitation by varying the intensity and wavelength of incident optical stimulus. Finally, we showcase application of our devices in reservoir computing and obtain >95% accuracy in simulated perceptron networks.
This work paves the way to employing greener materials, such as lead-free perovskites, that can be fabricated with low-cost scalable methods and are demonstrating multiple functionalities. The perovskite volatile optoelectronic memristive devices can be used as optical reservoirs in in-sensor reservoir computing systems for application in edge artificial intelligence (AI) that can be embedded in future wearable devices and the Internet of Things (IoT).
We acknowledge support from the UK Multidisciplinary Centre for Neuromorphic Computing (UKRI982), the UKRI Future Leaders Fellowship Grant “PHOTOMEM 2” (UKRI2061) and HORIZON Europe Project TEAM-NANO under grant agreement GA 101136388.
