Open-ended design creation tasks involve complex cognitive activities, such as problem understanding and idea generation. Previous studies using microstates indicate that design cognitive processes are associated with different temporal correlations; however, the temporal dynamics of EEG microstate sequences remain unknown. This study aims to classify EEG signals into six main design creation cognitive states (rest, problem understanding, idea generation, rating idea generation, idea evaluation, and rating idea evaluation) through microstate analysis and a deep learning model. Encoders were used to extract features from estimated temporal dynamics obtained from EEG microstate analysis, which were fed to a fully connected convolutional neural network (CNN). The classification performance was: 85.2% sensitivity, 98.0% specificity, 91.3% precision, and 86.9% F-score. Results showed that the proposed method effectively segmented complex design cognitive processes and classified design cognitive states. This study provides neurophysiological insights into brain dynamics associated with design tasks.