Extreme ultraviolet lithography mask multilayer film defect morphology reconstruction method

The invention relates to an extreme ultraviolet lithography blank mask multilayer film defect morphology reconstruction method. According to the method, blank mask multilayer film defect space images in multiple illumination directions are adopted to represent phase information of image missing, and the relation between the multilayer film defect space images and morphology parameters of multilayer film defects is constructed through a trained convolutional neural network model. The method comprises the steps of firstly simulating a space image set of multilayer film defects in different illumination directions to construct training set data, and then inputting the training set data into a convolutional neural network model for training. And inputting an actually measured mask multilayer film defect space image set into the trained model to obtain defect morphology parameters. According to the method, the relationship between the defect space images of the multilayer film in multiple illumination directions a