Reduction of current path of solution-processed organic photosynaptic transistors for neuromorphic computing

Solution-processed organic photosynaptic transistors (S-OPTs), inspired by the way biological nervous systems process visual information, offer several advantages such as large bandwidth, low latency, low energy consumption, tunable optoelectronic properties via molecular design, and applicability for simple and low-cost solution process at low temperatures. However, S-OPTs suffer high leakage current with undesirable current pathways, which is unavoidably a result of film formation over the entire substrate during solution processes. Herein, we propose a strategy of improving the photosensitivity of S-OPTs by patterning the organic semiconductor (OSC) film for application in smart and accurate optoelectronic systems. The OSC film of the device is simply patterned through selective evaporation contact printing. The OSC patterns with micrometer scale effectively contribute to reduction in the undesirable current paths and the resultant leakage current. Compared with conventional devices with nonpatterned OSC, our patterned S-OPT exhibits highly improved photosensitivity. Furthermore, our device demonstrates various types of synaptic characteristics, ranging from short- to long-term plasticity. By reducing the off-current level of the OSTs, hardware neural networks built using our patterned cells can successfully achieve recognition accuracy exceeding 90% for recognition of handwritten numerical images, which is comparable to those of ideal software systems. Thus, we believe that this study will introduce new avenues for fabrication of high-photoresponse S-OPTs and their utilization as essential building blocks for construction of neuromorphic systems.