Pushing the limits of microwave absorption capability of carbon fiber in fabric design based on genetic algorithm

Abstract

The field of electromagnetic wave absorption (EWA) requires the adaptability, tenability, and multifunction of high-performance materials in the future. The design and preparation of EWA materials aiming at performance requirements is the latest research hotspot. Here, a performancedriven strategy for simultaneously coordinating different target performances was proposed to optimize the structure of the periodical long continuous carbon/glass fiber fabric (PCGF) materials through algorithm and simulation. The optimized structure of the PCGF not only improves the impedance matching, but also introduces the induced orientation effect for a high cooperative loss of conductivity, resonance, and periodic structure. The flexible PCGF shows a broad effective absorption bandwidth (EAB) of 32.7 GHz covering a part of the C-band and the whole X-, Ku-, K-, and Ka-bands with a thickness (d) of only 0.92 mm and a density of 5.6×10 −4 kg•cm −3 . This highly designable fabric is promising for the EWA practical application owing to integrating the characteristics of good flexibility, acid and alkali resistance, bending resistance, excellent mechanical properties, and easy large-scale preparation.

Performance Optimization Engineering of Multicomponent Absorbing Materials Based on Machine Learning

“…By inputting the depolarization factor, , content of the absorbent, and the measured EM parameters (absorbent/matrix and matrix), the intrinsic EM parameters of the absorbent can be obtained using the MG model. The EM parameter of NiF04/paraffin was selected as the matrix (ε m ), and the EM parameters of NiF12 as the effective EM parameters (ε eff ).…”

“…Given that the MG model is a nonlinear equation, different designs are generated by the sampling-bounded space of designable factors, and the variables can be scaled by one of the dimensions, so the shape size and content of NiFs are the second and third designable factors, respectively. Different from the Ni surface, the intrinsic EM parameter of NiF08 (Figure S6) is obtained by inputting the depolarization factor N j ( x , y , z ) of NiF08, paraffin as ε m and NiF08/matrix as ε eff .…”

“…Our data-driven framework integrates the process of acquiring intrinsic and equivalent EM parameters, giving the strategy a growth feature, as multicomponent MA materials can be continuously improved by introducing new components. The multiple types and shapes of absorbents have been introduced in order to increase the designable factors and further expand the MA performance space, which includes Fe@CF (Figures S10–S14), WS 2 (two-dimensional (2D) materials; Figures S15 and S16), Co nanospheres (three-dimensional (3D) materials, Figures S17 and S18), glass fiber, and CF . Next, the designable factors of multi-absorbents were taken as codes, which were optimized until the fitness could no longer be improved, in order to achieve the different objectives (Figure a).…”

“…The multiple types and shapes of absorbents have been introduced in order to increase the designable factors and further expand the MA performance space, which includes Fe@CF (Figures S10−S14), WS 2 (two-dimensional (2D) materials; Figures S15 and S16), Co nanospheres (threedimensional (3D) materials, Figures S17 and S18), glass fiber, and CF. 24 Next, the designable factors of multi-absorbents were taken as codes, which were optimized until the fitness could no longer be improved, in order to achieve the different objectives (Figure 5a). The designed NMC-GMA has already demonstrated MA performance that meets the objectives (8.0−18.0 GHz), and this approach will also address many bottlenecks regarding Sband, C-band, and all bands (2.0−18.0 GHz).…”

“…Yet, the process for designing these materials for self-adaptive EM parameters is cumbersome and time-consuming, − due to the absence of general design principles between designable factors of absorbents and EM parameters. In this case, the expanded Maxwell–Garnett (MG) model , and machine learning (ML) − can provide significant advantages for accelerating the design process. The MG model describes the relationship between the designable factors and the EM parameters, and the expanded MG model increases the application field .…”