A Bidirectional Deep Neural Network for Accurate Silicon Color Design

• Published in: Advanced materials (Weinheim) 2019-12, Vol.31 (51), p.e1905467-n/a
• Main Authors: Gao, Li, Li, Xiaozhong, Liu, Dianjing, Wang, Lianhui, Yu, Zongfu
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Color; Design optimization; Electric dipoles; Inverse design; Iterative methods; Machine learning; Magnetic resonance; Materials science; nanophotonics; Nanostructure; Neural networks; Silicon; structural color
• ISSN: 0935-9648; 1521-4095
• DOI: 10.1002/adma.201905467

Silicon nanostructure color has achieved unprecedented high printing resolution and larger color gamut than sRGB. The exact color is determined by localized magnetic and electric dipole resonance of nanostructures, which are sensitive to their geometric changes. Usually, the design of specific colors and iterative optimization of geometric parameters are computationally costly, and obtaining millions of different structural colors is challenging. Here, a deep neural network is trained, which can accurately predict the color generated by random silicon nanostructures in the forward modeling process and solve the nonuniqueness problem in the inverse design process that can accurately output the device geometries for at least one million different colors. The key results suggest deep learning is a powerful tool to minimize the computation cost and maximize the design efficiency for nanophotonics, which can guide silicon color manufacturing with high accuracy. Nanophotonic device design relies heavily on time‐consuming electromagnetic simulation and iterative optimization. This study reports the training of deep neural networks that can perform both forward modeling and inverse design of one million different structural colors. The adoption of a tandem network can solve the nonuniqueness problem in the inverse design process with high efficiency and accuracy.