Abstract
Perovskite-based synapses are regarded as promising candidates for mitigating the von Neumann bottleneck and advancing the field of visual bionics. However, they suffer performance degradation under long-term environmental exposure. Thus, developing a performance-oriented restoration strategy is essential to realize their practical potential. Herein, we fabricate an ion-gel-gated CsPbBr3/graphene synaptic device in which the perovskite is pre-treated with TFSI- anions to delay performance degradation. After one year tracing in natural conditions, the recognition accuracy declined from 96.13% to 67.8%. To address this, we propose a photoelectronic synergistic modulation strategy to restore performance. By fine-tuning the gate voltage from 5 mV to 10 mV, the photocurrent increases by ~34.8%, reaching ~3.43 μΑ, and the recognition accuracy is restored to approximately 93.01%, close to its initial level. This approach enhances the photoelectric response and plasticity linearity of the device by promoting rapid transfer of photogenerated holes and suppressing defect-trapping through a transient electric field. This study presents a low-cost, non-reconstructive, and environment friendly functional-recovery strategy, achieves the reactivation from the degradation condition of the synaptic devices, and offers a promising pathway toward stable and controllable neuromorphic hardware systems.
| Original language | English |
|---|---|
| Pages (from-to) | 3497-3504 |
| Number of pages | 8 |
| Journal | Photonics Research |
| Volume | 14 |
| Issue number | 8 |
| DOIs | |
| Publication status | Published - 2026 |
| Externally published | Yes |
Fingerprint
Dive into the research topics of 'Synergistically modulated CsPbBr3/graphene photoelectric synapses for enhanced image recognition performance'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver