Multicomponent materials are microwave-absorbing (MA) materials composed of a variety of absorbents that are capable of reaching the property inaccessible for a single component. Discovering mostly valuable properties, however, often relies on semi-experience, as conventional multicomponent MA materials' design rules alone often fail in high-dimensional design spaces. Therefore, we propose performance optimization engineering to accelerate the design of multicomponent MA materials with desired performance in a practically infinite design space based on very sparse data. Our approach works as a closed-loop, integrating machine learning with the expanded Maxwell-Garnett model, electromagnetic calculations, and experimental feedback; aiming at different desired performances, Ni surface@carbon fiber (NiF) materials and NiF-based multicomponent (NMC) materials with target MA performance were screened and identified out of nearly infinite possible designs. The designed NiF and NMC fulfilled the requirements for the X- and Ku-bands at thicknesses of only 2.0 and 1.78 mm, respectively. In addition, the targets regarding S, C, and all bands (2.0-18.0 GHz) were also achieved as expected. This performance optimization engineering opens up a unique and effective way to design microwave-absorbing materials for practical application.
Keywords: Maxwell−Garnett model; machine learning; microwave absorption; multicomponent materials; performance optimization.