Femtosecond laser micro-machining of three-dimensional surface profiles on flat single crystal sapphire

In this study, a regression model is proposed using the machine learning (ML) method for femtosecond laser micro-machining of three-dimensional (3D) surface profiles on flat single-crystal sapphire. Based on the ML regression model, a quantitative relationship is established between the femtosecond laser processing parameters and the processed 3D functional curved surface topography. The combination of laser processing parameters (i.e. laser fluence F0, hatch dh, and scanning speed v) is also optimized according to the predicted parameters (i.e. surface quality Ra, ablation depth Z, and material removal rate MRR). Finally, the optimized combination of laser processing parameters was employed to process 3D surface profiles on flat single-crystal sapphire sheets. The maximum predicted profile error of the 3D structures was approximately 12.6% without any prior optimization experiments. Both the numerical simulation and experimental results demonstrate the feasibility of the proposed ML regression model in optimizing laser processing parameter combinations to achieve the desired 3D surface geometries on single-crystal sapphire by femtosecond laser micro-machining without trial-and-error.