光子晶体激光器掩埋纳米孔结构自动量取OA

The geometric morphology of buried nano-hole structures plays a critical role in the optical feedback characteristics and device stability of photonic crystal surface-emitting lasers(PCSELs),in which size and uni-formity are key parameters for performance evaluation.In practical characterization,cross-sectional scanning electron microscopy(SEM)is commonly employed by cleaving devices along specific crystallographic direc-tions.However,due to the buried nature of the nano-holes and the large field of view required for array-level inspection,individual holes occupy only a small fraction of the image,leading to low contrast,blurred boundar-ies,and complex morphologies that hinder automatic identification and accurate measurement.To address these challenges,an automatic hole identification and geometric measurement method for PCSEL cross-sec-tional SEM images is presented.A semantic segmentation model is employed to automatically extract hole re-gions under different hole geometries and imaging conditions,based on which hole dimensions are measured in a unified and automated manner.The selected model achieves an Intersection-over-Union(IoU)of approxi-mately 93%on the validation dataset,with an F1-score of about 95%.The automatically measured results show good agreement with manual measurements,with a relative uncertainty of approximately 1%-5%across different size ranges.In addition,the proposed approach enables quantitative evaluation of dimensional unifor-mity and local non-uniformity of buried nano-hole arrays along the lateral direction,providing useful structural information at the array scale.These results demonstrate that the proposed method improves measurement effi-ciency and reliability,offering a practical tool for dimensional evaluation and process monitoring of buried nano-hole structures in PCSEL devices.