【Member Papers】Interface-Defect Synergy Engineering of Amorphous Ga₂O₃ for Voltage-Tunable Dual-Mode Optoelectronic Devices: Self-Powered Imaging and Neuromorphic Vision

      Researchers from Chongqing Normal University, Northeast Normal University, University of Electronic Science and Technology of China and the University of Wollongong have published a paper titled “Interface-Defect Synergy Engineering of Amorphous Ga₂O₃ for Voltage-Tunable Dual-Mode Optoelectronic Devices: Self-Powered Imaging and Neuromorphic Vision” in the academic journal Advanced Functional Materials.

Background 

      Traditional von Neumann architecture physically separates photodetection, memory and computing units, leading to high data transmission latency and excessive power consumption, which limits lightweight intelligent vision hardware. Integrating photoelectric sensing and synaptic cognitive functions into a single device is a core route to break computing bottlenecks, yet the two functions have contradictory demands on carrier regulation: high-sensitivity detection requires efficient carrier separation, while synaptic behavior relies on charge trapping to realize gradual conductance modulation, which is hard to balance simultaneously. Most existing dual-mode devices adopt multi-heterojunction composite structures with difficult band offset control, complex fabrication and uncontrollable interface defects. Amorphous Ga₂O₃ owns an ultra-wide bandgap and can be fabricated in large area at low cost, making it a candidate for solar-blind detection, while controllable interface defect regulation schemes are still absent. This work proposes an interface-defect synergy engineering strategy. In vertical ITO/GaOₓ/Al architecture, oxygen vacancies and asymmetric potential barriers are precisely introduced via in-situ interfacial redox reactions. Only a single active layer is adopted to realize two voltage-switchable working modes: zero-bias self-powered solar-blind imaging and low-power synaptic response, balancing high-sensitivity weak light detection and low-power learning-memory behavior, and offering a simple feasible solution for monolithic integrated intelligent vision hardware.

Abstract

      Integrating disparate optoelectronic functionalities—ranging from self-powered photodetection to neuromorphic computing— within a unified, single-material architecture is pivotal for the next generation of bias-programmable intelligent vision systems. Here, we report a voltage-tunable, monolithic platform based on amorphous GaOₓ (ITO/GaOₓ/Al) that achieves bias-programmable dual-mode operation, uniting self-powered solar-blind photodetection with low-power optoelectronic synaptic functionalities. The dual-modality originates from the synergistic coupling between asymmetric interfacial potential barriers and oxygen-vacancy engineering, which is in situ induced by interfacial redox reactions at the GaOₓ/Al junction. Benefiting from this precise interface-defect co-engineering, the device exhibits exceptional self-powered solar-blind detection with high responsivity, a rapid response of ∼100 µs, and a detection limit down to 13 nW. Under a low bias, it emulates bio-realistic synaptic dynamics—including paired-pulse facilitation and learning–forgetting–relearning processes—with an ultralow energy footprint of 18 pJ. Reversible voltage modulation enables both single-pixel and 10 × 10 array devices to bridge the gap between high-contrast transient imaging and long-term visual memory. Moreover, a neuromorphic visual system constructed from this platform performs robust “detection–storage–recognition” tasks even under external noise. This study establishes a transformative route for bias-programmable optoelectronic integration, effectively uniting sensing and cognitive functions within a simplified, high-performance single-material system.

Highlights

      ① An original interface-defect synergy engineering strategy for amorphous GaOₓ is proposed. Oxygen vacancies are generated via in-situ redox at Al interface, and dual voltage-switchable modes are realized with only one active layer without multi-heterostructure;

      ② Under zero-bias self-powered mode, the device achieves an ultra-weak detection limit of 13 nW, response time around 100 µs, and solar-blind/visible rejection ratio up to 2.6 × 10³;

      ③ Under low bias, the device accurately emulates biological synaptic behaviors including paired-pulse facilitation, short/long-term plasticity and learning-forgetting process, with ultra-low energy consumption of 18 pJ per synaptic event;

      ④ A 10 × 10 crossbar array is fabricated, which can separately realize high-contrast self-powered solar-blind imaging and long-term image memory retention;

      ⑤ A neuromorphic visual system is constructed, which stably completes image detection, storage and recognition even under noise disturbance with stable recognition accuracy of 93%.

Working Mechanism

      To elucidate the mechanism underlying the dual-mode operation, the researchers further constructed a series of control devices for systematic verification. The results demonstrate that variations in interfacial potential barriers induced by different electrode configurations determine the direction of photogenerated carrier separation under zero bias, enabling self-powered photoelectric conversion. Meanwhile, oxygen vacancies generated at the Al/GaOₓ interface act as carrier trapping centers, inducing persistent photoconductivity under low bias and thereby enabling synaptic memory functionality. These findings clearly distinguish the distinct roles of interfacial barriers and defect states in the two operating modes, providing a physical basis for voltage-programmable multifunctional optoelectronic integration.

Conclusion

      In summary, we have developed a voltage-tunable, vertical optoelectronic device based on a single amorphous Ga₂O₃ layer to realize bias-programmable dual-mode operation within a monolithic system. By leveraging an interface-defect co-engineering strategy, the synergistic coupling between asymmetric interfacial potential barriers and oxygen-vacancy modulation enables the seamless reconfiguration of the device between self-powered solar-blind photodetection and low-power optoelectronic synaptic functionalities. Benefiting from this heterojunction-free architecture, the platform exhibits exceptional sensing performance, including high responsivity, a rapid response time of ∼100 µs, and ultraweak signal detection limits down to 13 nW. Simultaneously, the device emulates energy-efficient synaptic plasticity with a lower power consumption of 18 pJ per event and superior operational stability. This implementation of voltage modulation facilitates the transition from high-contrast transient imaging to persistent visual memory in both single-pixel and array configurations, effectively bridging the functional gap between signal acquisition and neuromorphic perception. This work not only establishes a streamlined, single-material route toward multifunctional optoelectronic integration but also provides a robust hardware foundation for the development of energy-efficient, intelligent optoelectronic systems for future cognitive photonics, neuromorphic visual networks, and large-scale integrated hardware vision systems.

Project Support

      The authors gratefully acknowledge support from the National Natural Science Foundation of China (Grant Nos. 62574029 and 12304102), the Natural Science Foundation of Chongqing (Grant Number CSTB2023NSCQ-MSX0479), the Science and Technology Research Project of Chongqing Municipal Education Commission (Grant Nos. KJZD-M202600502, KJQN202300536 and KJQN202400558) and the Chongqing Normal University Foundation (Grant Nos. 23XLB029, 23XLB002, and S202510637007).