Research on high-entropy spinel microwave absorption materials: Exploration of machine learning and experimental integration

Abstract

With the advancement of electronic communication technology, the intensity of electromagnetic radiation is increasing, and traditional wave-absorbing materials no longer meet the demands in various current environments. Spinel ferrite is one of the earliest materials used for absorbing electromagnetic waves, and the introduction of "high entropy" and controlling the degree of entropy disorder can achieve performance regulation. Nonetheless, the lengthy testing period and high costs associated with trial and error have posed challenges, limiting the development of high entropy microwave absorbing materials. This study utilized machine learning techniques to construct a high-entropy spinel microwave absorption property database. The design of Co_xNi_0.4-xCu_0.2Zn_0.2Mn_0.2Fe₂O₄ (X = 0, 0.1, 0.2, 0.3, 0.4) was informed by the SHAP plots of the GBR model. Machine learning demonstrates that factors such as cobalt and nickel content, particle size, and others significantly influence microwave absorption performance. The experimental result manifests it and Co content regulate the microwave absorbing performance through adjusting the defect density and particle size. The Co_xNi_0.4-xCu_0.2Zn_0.2Mn_0.2Fe₂O₄ with optimized Co content achieved a reflection loss (RL) of −45.32 dB, and an effective absorption bandwidth (EAB) of 6.48 GHz. This approach significantly reduced the research period, presented a novel research methodology for other scholars, and expedited the research progress in the field of microwave absorbing materials.