Conventional machine vision architectures suffer from intrinsic energy inefficiency and latency bottlenecks due to the physical segregation of sensing, memory, and processing units. Emerging organic neuromorphic devices offer a promising solution, yet concurrently achieving strong photoresponse, ultralow-power operation, and reliable non-volatile memory remains a critical challenge. Herein, we integrate a heterostructured dielectric layer, poly (amic acid) (PAA)/hafnium oxide, into wafer-scale organic neuromorphic devices, significantly enhancing charge carrier mobility, photosensitivity, and memory performance. The optimized devices exhibit a suite of exceptional characteristics: high carrier mobility, ultralow light detection sensitivity (102 nW cm-2) with a fast response time (50 µs), minimal energy consumption (53 aJ per spike), long-term memory retention (50 000 s), and robust endurance. We further demonstrate programmable organic thin-film transistors (OTFTs) driving LED arrays, enabling repeatable light/mask-induced pattern writing/erasing and integration of sensing-memory-display functions. Linear Dynamic Range Adjustment (LDRA) simulations reveal ultrahigh dynamic range and superior imaging capabilities of PAA-optimized OTFTs. This hetero-dielectric strategy establishes a universal platform for organic neuromorphic electronics, addressing key gaps in adaptive edge vision systems and human-retina-like interactive meta-displays.
Jiang et al. (2026) studied this question.