The fast evolution of synthetic intelligence (AI) computing calls for progressive reminiscence applied sciences that combine high-speed processing with energy-efficient information storage. Right here, we report a mixed-dimensional photomemory machine primarily based on a CsPbBr3/Al2O3/MoS2 structure, leveraging perovskite quantum dots (PQDs) as a photoactive floating-gate layer, a tunable Al2O3 dielectric, and a 2D MoS2 channel. Optical and electrical characterizations, together with steady-state and time-resolved photoluminescence (PL), Kelvin probe drive microscopy (KPFM), and current-voltage measurements, reveal the interaction of dielectric thickness and interfacial results in governing cost switch effectivity. By optimizing the Al2O3 thickness to five.5 nm, we obtain exact management over cost switch dynamics, enabling an optimum cost switch charge with minimal optical vitality (~sub-pJ) to retailer a single optimistic cost within the PQDs. The machine reveals distinctive optoelectronic efficiency, together with an almost linear correlation between incident photon quantity and common photocurrent (Iph(avg)) over two orders of magnitude, multilevel storage functionality, and a reminiscence window with a excessive On/Off ratio. These findings set up a strong platform for next-generation perovskite-based photomemories, providing insights into energy-efficient, high-performance optoelectronic programs for superior AI chip purposes.
