Tunable Synaptic Plasticity in Inkjet‐Printed Oxide‐Based Heterojunctions for Hardware Authentication and Brain‐Inspired Optoelectronics

Optoelectronic synapses serve as the key element of neuromorphic vision systems, which replicate brain's efficient visual processing by integrating detection, memory, and computing into compact hardware. Optoelectronic synapses, particularly those based on amorphous metal oxides, exhibit synaptic plasticity, resembling that of the brain, upon optical stimulation, owing to their persistent photoconductivity. This work demonstrates that altering the elemental composition and varying the interfaces of solution‐processed amorphous indium gallium zinc oxide(a‐IGZO) based two‐terminal optoelectronic synapses enable control of PPC. These findings are supported by a systematic analysis of different synaptic behaviors, including spike‐timing‐dependent plasticity (STDP), and spike‐number‐dependent plasticity (SNDP), in printed devices with varying compositions. A robust STP to LTP transformation has been demonstrated, underscoring the effects of interface type (hole‐rich or hole‐deficient) on PPC dynamics. By utilizing the distinct fast and slow‐decay dynamics of the printed synapses, a proof of concept physically unclonable function (PUF) for low‐cost hardware authentication is demonstrated with a high pattern similarity score of 0.92. Finally, a software‐based artificial neural network has been developed using inkjet‐printed optoelectronic synapses, showing up to 512 discrete conductance states at 1 V, and recognizing handwritten digits in the MNIST datasets with an accuracy of 95%.