Achieving high diffraction efficiency in LCoS-SLMs via neural-network-based precise fringe-field compensation

Liquid crystal on silicon (LCoS) devices have been widely applied in optical beam steering. Increasing demands for advanced applications are driving LCoS technology toward larger deflection angles and higher diffraction efficiency, while these two performance metrics are difficult to achieve simultaneously due to the fringe-field effect. To address this issue, we have trained a neural network to achieve precise compensation for the fringe-field effect and have experimentally validated the approach using a commercial LCoS device. The results demonstrate that this method achieves an efficiency over 37.8% with a continuous diffraction angle up to 10° @ 1550 nm, representing the largest deflection angle and highest efficiency yet reported for LCoS devices. This work paves the way for the practical application of LCoS in fields such as LiDAR, holography, and optical switches.