Engineering Phase‐Separated Nanoclusters for Multilevel Negative Photoconductivity and Optically Driven Metaplasticity in Organic Heterojunction Transistors

ABSTRACT

Organic heterojunction transistors provide a promising platform for neuromorphic hardware, with engineered interfaces that enable versatile optoelectronic functionalities. However, uncontrolled shallow defects at the heterojunction interface induce stochastic carrier trapping and detrapping, destabilizing weight evolution, and preventing the realization of complex bio‐functions. Here, we report a phase‐separation engineering strategy that embeds discrete nanoclusters within organic heterojunction transistors. The nanoclusters serve as spatially distributed deep trapping centers, effectively passivating interface disorder defects, thereby enabling multilevel conductivity and biomimetic metaplasticity. Consequently, the device exhibits a wide memory window of 35 V and a high ON/OFF current ratio (>10 ⁵ ). It further demonstrates tunable negative photoconductivity (NPC) featuring 131 distinct levels with 1000 s retention. Neural network simulation using experimentally extracted weight updates yields a recognition accuracy of 96.86% on the MNIST dataset. Furthermore, deep‐trap saturation induces fully optically driven metaplasticity at zero bias, enabling contrast‐sensitive encoding via light intensity‐dependent gain control reminiscent of Matthew‐effect‐inspired selective reinforcement. This work provides a versatile and general platform for tailoring interfacial carrier dynamics, paving the way for next‐generation neuromorphic and biomimetic vision hardware.