A model-based parameter optimization strategy for dual-rotating-compensator mueller matrix ellipsometers

The dual-rotating-compensator Mueller matrix ellipsometer is a vital tool for characterizing thin films and nanostructures. However, its measurement precision is highly sensitive to the system configuration parameters. Random noise generated during the modulation-demodulation process propagates to the Mueller matrix elements through Fourier transformation, becoming a primary factor that limits the system precision and repeatability. Using an analytical variance model, this study reveals the random-error propagation mechanism and evaluates the individual and coupled effects of configuration parameters. A multidimensional synergistic hierarchical optimization algorithm is then developed to efficiently identify the optimal configuration through distribution-guided sparse sampling. Numerical simulation results demonstrate that the optimized configuration reduces the total theoretical variance of the Mueller matrix by 50.6% and 30.0% compared to the conventional univariate and multivariate condition-number-based configurations, respectively. Experiments further confirm that a realizable configuration close to the theoretical optimum reduces measurement variance and improves repeatability. This result alleviates the problem of certain elements of the Mueller matrix being overly sensitive to random noise, providing new optimization strategies and theoretical references for achieving high-precision, highly repeatable full Mueller matrix characterization.