Chemical Tuning of Charge Transport Polarity in Light-Emitting Transistors Enabling Neuromorphic Display, Computing, and Optocoupler

Résumé fourni par la source

Long-afterglow light-emitting transistors (LALETs) represent an efficient platform for neuromorphic computing by integrating in-memory processing, optical output, and noise-filtering functionality within a single device. Here, we have synthesized a series of donor-acceptor (D-A) conjugated copolymers comprising a diketopyrrolopyrrole (DPP) acceptor and oligo-thiophene (T) donors of tunable length to investigate the impact of charge-transport polarity on LALET performance. By terpolymerization, the D:A ratio was adjusted from 4:1 to 6:1, 8:1, and 12:1, yielding DPP4T, DPP6T, DPP8T, and DPP12T copolymers, respectively. Among them, DPP8T exhibited a favorable combination of high hole mobility (0.42 cm2 V-1 s-1) and strongly suppressed electron mobility, enabling high electroluminescence of 1189 cd m-2 while maintaining programming currents below 10 μA. The resulting LALET displayed dual operation mode, long-afterglow, and click-on upon employing the opposite gate stimuli. In long-afterglow mode, both electrical and optical postsynaptic signals enabled reservoir computing system with Fashion-MNIST recognition accuracies exceeding 90%. Furthermore, the intrinsic noise-filtering function of LALET emulated neuronal all-or-none firing, allowing a neuromorphic optocoupler to implement complex "stimulus-computing-decision-action" logic and to be used as the output layer for a reservoir computing framework toward a noise-filtering display. These results highlight the importance of the chemical modulation of the polarity in charge transport for high-performance neuromorphic devices.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.