Source And Mask Co-optimization Based On Immersed Full Physical Model

With the development of semiconductor integrated circuits and the prediction and advancement of the key dimensions of the chip by Moore's Law,lithographers have to propose and use more accurate models and introduce better optimization algorithms in order to explore the potential of optical resolution enhancement.However,various factors in the actual production process restrict the manufacturability of the chip,reducing the yield and productivity efficiency of the integrated circuit.The key to solving the above problems is to establish a more accurate model and further improve the optimization efficiency of the algorithm.By referring to a large number of domestic and foreign literatures,this paper adds the imaging process of optical projection in photoresist in the partially coherent light vector model,including the refraction,reflection and standing wave formation of electromagnetic waves,which further improves the model and The degree of agreement between the actual situation.In addition,we apply a semi-implicit update rule based on the level set algorithm,which can synthesize the source and mask faster than the traditional explicit algorithm.This experiment not only improves the imaging accuracy,but also has higher computational efficiency under the same conditions of optimization object,and the synthetic projection image has better effect.The research content of this paper mainly includes:The scalar model and vector model based on partially coherent light illumination system are constructed and compared to analyze the imaging effect and inverse optimization effect.Then build a layered media model based on the vector model,compare the similarities and differences of imaging,and focus on the analysis of the ability of inverse optimization to the complex target image.Define and use micro-process parameters to improve process manufacturability.Secondly,the model-based inverse optimization framework and the rule-based inverse optimization framework are built.By comparing the steepest gradient descent method and the stochastic gradient descent method,the optimal performance algorithm is selected.Finally,a semi-implicit optimization framework based on level set is built,and the vector model based on layered media is used for inverse optimization,which is compared with the experiment using the steepest gradient descent method under the same conditions.In this paper,the model and inverse optimization of immersion projection lithography system are studied experimentally,and a vector imaging model based on layered media is built.The imaging process is further described and the image precision is improved.And combined with the semi-implicit iterative scheme and the additive operator splitting algorithm,the efficiency of inverse optimization is improved,which has certain theoretical research value.