Three-dimensional topography measurement of curved diffractive microstructures based on laser scanning confocal microscopy stitching

As core optical components in spectrometer systems, curved diffraction gratings require precise control of both macroscale substrate curvature accuracy and microscale periodic structure parameters, as these topographic features govern optical performance. However, existing methods struggle to simultaneously provide large-range and high-resolution measurements, making high-precision measurement of macro-to-micro scale profile topography challenging. In this paper, we propose a laser scanning confocal microscopy (LSCM) combined with subaperture stitching method to measure curved grating microstructures. Feature points in overlapping subaperture regions are extracted using the Scale Invariant Feature Transform (SIFT) algorithm. Mismatches are filtered via an improved Random Sample Consensus (RANSAC) algorithm, establishing reliable feature matching point pairs between adjacent subapertures. A weighted 3D Single Value Decomposition-Iterative Closest Point (SVD-ICP) point cloud registration method is used to solve the spatial transformation matrix and complete the subaperture stitching of the curved microstructure. Finally, comprehensive characterization is performed on both the macroscopic substrate and microscopic diffraction parameters of the stitched grating. Experimental results demonstrate that the proposed method achieves sub-pixel accuracy in reconstructing millimeter-scale convex spherical gratings, without stitching misalignment phenomenon. The substrate surface error measures 196.55nm PV and 33.85nm RMS, and the line profile characterization of microscopic diffractive structure is in good agreement with the white light interferometry measurement: mean grating period deviation is below 0.8%, and mean blazed angle deviation is below 1.3%. This method realizes high-precision cross-scale 3D topography measurement of curved diffractive microstructures, providing a reliable measurement means for accurate characterization and process optimization of diffractive microstructures.