Séminaire de Tao Xu (Université de Shanghai)

Semitransparent organic photovoltaics (ST-OPV) offer transformative potential for building-integrated applications by simultaneously harvesting energy and transmitting daylight. However, their development is fundamentally constrained by the efficiency-transparency trade-off governed by intricate photon management. To overcome this limitation, this work introduces a multiscale optical engineering strategy integrating physics-enhanced deep learning guided optical design for global optimization with localized near-infrared (NIR) plasmonic nanostructures for selective photon harvesting. This synergistic approach achieves a breakthrough in balancing high visible transparency and efficient NIR photoelectric conversion, delivering a state-of-the-art light utilization efficiency of 6.09% in ST-OPV, with a power conversion efficiency of 14.97% and an average visible transmittance of 40.70%. The devices further demonstrate excellent color rendering and effective infrared heat-insulation, aligning with practical architectural requirements. Overall, this work establishes an optical engineering paradigm bridging computational design and nanophotonic control, accelerating high-performance ST-OPV deployment in sustainable buildings.