Fabry–Pérot Resonant Memristor for Neuromorphic Photonic Computing

A Fabry–Pérot enhanced resistive random-access memory (FP-RRAM) comprising an Ag/BFO/ITO stack integrated with a distributed Bragg reflector (DBR) is demonstrated, enabling dual electrical and optical memory readout. Well-separated cavity resonances allow reliable optical discrimination between high- and low-resistance states. Cross-sectional transmission electron microscopy (TEM) coupled with energy-dispersive X-ray spectroscopy (EDS) mapping reveals spatially localized Ag migration and oxygen-vacancy filament formation, collectively modulating the effective permittivity of the BFO layer. Momentum-resolved reflectance measurements and finite-difference time-domain (FDTD) simulations further validate that resistive switching tunes the cavity eigenmode through changes in the effective refractive index. Combined electrical and optical characterization exhibits pronounced hysteresis and bistability, with robust retention and synchronized modulation observed up to a 50-ms timescale. By employing the Maxwell–Garnett effective medium approximation, ionic reconfiguration with optical response, revealing sharp increases in both Ag and oxygen-vacancy volume fractions at the set threshold, with Ag migration dominating the switching kinetics is quantitatively correlated. This integrated structural, optical, and theoretical analysis directly links nanoscale filament dynamics with macroscopic photonic modulations. Additionally, the FP-RRAM demonstrates linear optical weight updates with enhanced state resolution for precise synaptic modulation, leading to higher recognition accuracy than purely electrical programming. These results establish FP-RRAM as a reliable, electrically programmable, and optically addressable memory platform, offering a promising avenue for neuromorphic and photonic computing applications.