Method for predicting defect in semiconductor device

A method for predicting a defect in a semiconductor device, the method comprising:

• calculating a first probability that particles will be generated in a semiconductor element of the semiconductor device by radiation;
• calculating a second probability that damage will occur in the semiconductor element due to the particles;
• generating a training data set using input data and simulation data, the input data comprising damage data generated using the first probability and the second probability and comprising at least one of a position in which the damage will occur and an amount of the damage, an impurity concentration of impurities doped in at least a portion of the semiconductor element, and structural data of the semiconductor element, and the simulation data comprising electrical characteristics of the semiconductor element obtained as a result of a simulation based on the input data;
• training a machine learning model based on the training data set to generate a defect prediction model;
• applying the defect prediction model to a design of the semiconductor device to generate a defect prediction output;
• generating an updated design of the semiconductor device based on the defect prediction output of the defect prediction model;
• and
• manufacturing the semiconductor device based on the updated design.