Introduction

Massive antenna systems in 5G base stations and automotive radars for autonomous driving are characterized by high component density and complex packed arrangements. This makes them more susceptible to mutual coupling interference and environmental scattering1,2,3. Furthermore, the advancement of various detection technologies has greatly increased the challenges of radar stealth for sensitive facilities and equipment. Microwave absorbing materials provide an effective solution to these issues4,5,6. However, conventional coatings or rigid microwave absorber structures are difficult to deploy efficiently and flexibly on the complex surfaces and deformable components present in these applications. While some commercial solutions such as microwave-absorbing silicone sheets and open-cell foams offer specific advantages like low-profile or broadband performance and limited conformability to simple curves, they inevitably involve inherent performance trade-offs and remain inadequate for applications requiring higher levels of adaptability (Supplementary Table S1). Consequently, conformal microwave absorbers that combine robust broadband high-efficiency absorption with adaptive compatibility for complex curvatures have emerged as a promising research frontier for addressing electromagnetic compatibility (EMC) in highly integrated electronics and enabling low observability for target objects7,8.

To address these emerging challenges, researchers have explored several potential solutions. The most prevalent approach leverages the bendable properties of conductive films combined with flexible dielectric layers to create a series of conformal metasurface absorbers9,10,11,12. Some 3D-printed absorbers designed with lightweight and bendable substrates can also conform to curved surfaces13,14. However, due to intrinsic limitations in structural flexibility, these absorbers are typically restricted to developable surfaces such as cylindrical planes15. Additionally, if conformality is achieved by bending a flat absorber rather than manufacturing it at a predefined angle, internal stress buildup within the structure can occur. Stretchable absorbing materials offer a promising alternative, yet they inevitably suffer from structural deformation and reduced electromagnetic parameters upon conformal application, leading to reduced absorption efficiency16. To date, few solutions have succeeded in optimally balancing structural adaptability and performance retention17,18, highlighting an urgent need for new design paradigms. Meanwhile, research on metamaterial absorbers has expanded from static designs to dynamic response systems, with circuit-tuning strategies being particularly prominent19,20,21. Such approaches enable effective control over absorption frequency and amplitude, and in some cases even exhibit unique dependencies on input power and waveform22,23,24. However, integrating these advanced functionalities with robust conformal deployment on complex curved surfaces poses even greater difficulties.

Chainmail, a type of armor composed of interlinked metal rings, has been widely used for centuries due to its unique combination of robustness and wearability (Fig. 1a)25. In recent years, advanced smart chainmail fabrics have attracted considerable attention in the field of mechanical metamaterials26,27. However, their potential for electromagnetic wave manipulation remains largely unexplored. Here, we boldly transfer insights across disciplines to propose a chainmail-inspired microwave metamaterial absorber. The absorber requires only slight movement or rotation of the interconnected rigid units to adapt to various complex surfaces, thereby avoiding the negative effects of stretching or bending. This characteristic allows it to maintain minimal degradation in absorption performance (Fig. 1c), offering a promising solution to the aforementioned challenges. Notably, the protective capabilities of chainmail largely depend on its weaving pattern. A sparse weave provides lightweight protection against slashing or cutting, whereas a denser weave offers enhanced defense against piercing attacks (Fig. 1b). Motivated by this principle, we integrate the microwave absorber with elastic bands to fully exploit its fabric-like flexibility, enabling dynamic adjustment of the unit arrangement density (Fig. 1d). The resulting reconfigurable system offers an approach to modulating microwave absorption and bandwidth expansion, with the added advantage of sustaining specific configurations without sustained driving power28,29,30.

Fig. 1: Design concept and function of the absorber.
Fig. 1: Design concept and function of the absorber.
Full size image

a Characteristics of chainmail and its components. b Defense effects achieved by different weaving methods. c Structure of the absorber unit and the conformal application. d Absorber states corresponding to different operating frequencies.

In this work, we designed a chainmail unit composed of a lossy double-layered cross at the center enclosed by a lossless rectangular frame. The absorber array was fabricated as a single piece using a dual-nozzle fused deposition modeling (FDM) printer. In its fully expanded planar state, the absorber demonstrates a polarization-insensitive broadband effective absorption bandwidth (EAB) from 6.2 to 17.6 GHz. When conformed to cylindrical, saddle, and composite spherical surfaces, the absorber maintains good performance with only 0.049 average absorptivity degradation due to the deformation-free unit design, which shows a significant robustness advantage over conventional absorbers. The dynamic transition between expansion and contraction enables the absorber to cover a broad frequency range of 4.6–18 GHz, further extending its ultrabroadband absorption capability. Moreover, we systematically analyze the broadband absorption mechanism and dynamic switching process of the metamaterial using effective medium theory31, Lorentz dispersion model32, and characteristic mode analysis (CMA)33. The prototype successfully addresses long-standing limitations of conventional microwave absorbers, including: (i) difficulties in adaptive deployment on complex curved surfaces, especially non-developable ones, (ii) angular sensitivity of absorption performance in curved conformal applications, and (iii) thickness-bandwidth tradeoff governed by the Rozanov limit for static response material. This provides insights into the development of ultra-wideband, conformal, and switchable microwave absorbers.

Results

Design of chainmail units

The circular ring represents the most classic unit configuration in chainmail, offering simplicity and lightweight properties. However, its high degree of rotational freedom makes it difficult to maintain fixed orientations, increasing performance instability and design complexity of absorber34. To address this issue, we replaced the circular rings with cubic rings (Type 1). This modified structure ensures stable vertical placement of units on flat surfaces while facilitating efficient performance design and optimization (Fig. 2a). More importantly, it enables direct fabrication using basic FDM technology. The optimized prototype of this design achieves an EAB approaching 13 GHz, but exhibits notable polarization sensitivity (Fig. 2b). Furthermore, the required narrow width and thickness of the square rings for optimal performance (Supplementary Fig. S1) result in insufficient mechanical strength (Fig. 2e) and low 3D printing quality. While adding crossbeams (Type 2) partially mitigates these issues (Fig. 2c, f, and Supplementary Fig. S2), the improvement remains unsatisfactory.

Fig. 2: Design of chainmail units.
Fig. 2: Design of chainmail units.
Full size image

a Unit structure improvement process. Reflection loss for different polarization angles: b Type1, c Type2, d Type3. Force-displacement curve during compression of a single unit: e Type1, f Type2, g Type3.

Polarization insensitivity typically requires a structure with fourfold rotational symmetry. However, to accommodate interconnection, the mutually perpendicular edges of the transformed cubic rings are not coplanar. This inherent conflict prevents the entire framework from being constructed solely with lossy materials, leading to the development of the improved Type 3 design. This configuration consists of a lossy double-layered cross at the center and a lossless rectangular frame. By reducing the permittivity of the anisotropic components, the polarization insensitivity is enhanced, with a bandwidth discrepancy of only 0.3 GHz between the TE and TM polarizations and slight fluctuations in absorptivity (Fig. 2d). Although its areal density increases from 0.53 kg/m² and 0.74 kg/m² to 1.7 kg/m², this value remains comparable to that of honeycomb absorbers, indicating that its lightweight performance is superior to that of most absorbing materials (Supplementary Table S1)35,36,37,38,39. Moreover, the thickness of the structure is reduced from 7 mm to 5.5 mm (Supplementary Fig. S3), while its mechanical strength improves nearly tenfold compared to the prototype (Fig. 2g). Ultimately, the absorption performance of the unit was successfully improved, while the increased number and width of beams and pillars in the Type 3 structure also provided additional enhancements in mechanical strength and fabrication feasibility.

Microwave absorption mechanism

Figure 3a presents the absorber sample produced in a single-step, support-free FDM printing process. The cubic unit cells form a 2D square lattice through quadrant nesting (Supplementary Fig. S4). The external frame employs pure PLA (ε = 2.7), while the internal cross structure consists of lossy PLA with 10 wt% CNT filler. The SEM imaging confirms uniform CNT distribution within the matrix, ensuring stable electromagnetic parameters. Due to the high conductivity of CNTs, the permittivity of CNT-PLA exhibits pronounced dispersion and a large loss tangent (Supplementary Fig. S5), which serves as the dominant source of energy dissipation in the absorber. After comprehensive optimization considering electromagnetic, mechanical, and fabrication requirements, the final structural parameters were determined as l1 = 14 mm, d1 = 1.5 mm, h1 = 2 mm, and h2 = 1.5 mm (Supplementary Fig. S3).

Fig. 3: Absorption performance and mechanism.
Fig. 3: Absorption performance and mechanism.
Full size image

a Composition of absorber. b Effective permittivity of absorber, with insets showing the loss distributions at representative frequencies. c Effective permeability of chainmail, with insets showing the current distributions at representative frequencies. d Reflection loss under TE and TM polarization from 2 to 18 GHz, and angle-dependent absorptivity for oblique incidence at 8 GHz. e Modal significance of PEC chainmail absorber (gray region highlights the significant modes). f Radiation patterns of different modes.

To elucidate the absorption mechanism, we further analyzed the chainmail unit characteristics through effective medium theory. The two-port S-parameters of the unit were obtained through simulation and converted into equivalent permittivity and permeability using the Nicholson-Ross-Weir (NRW) method (Supplementary Note 1)40,41. It should be noted that, unlike conventional composite materials which typically exhibit Debye relaxation behavior (Eq. (1) where εs is the static permittivity, ε∞ is the optical-frequency permittivity, f is the frequency, and τ is the relaxation time), the structure of the unit significantly affects its dielectric response (Fig. 3b). The chainmail unit exhibits a pronounced loss peak around 10 GHz, corresponding to a typical Lorentz resonance (Eq. (2) where f0 is the resonance frequency, and δ is the damping factor). This phenomenon is half wavelength resonance caused by the crossed-dipole antenna shaped lossy parts42,43. It is further evidenced by the centered high-loss distribution at resonant frequency. Additionally, the dual-layer structure generates relatively strong counter-parallel currents near 8 GHz and 14 GHz, leading to artificial magnetic resonance (Fig. 3c)44,45. These equivalent magnetic loss effects facilitate the transition of permittivity to the ideal range, effectively broadening the absorption bandwidth (Supplementary Fig. S6)46,47.

$$\varepsilon (f)={\varepsilon }_{\infty }+\frac{\left({\varepsilon }_{s}-{\varepsilon }_{\infty }\right)}{1+i2\pi f\tau }$$
(1)
$$\varepsilon (f)={\varepsilon }_{\infty }+\frac{\left({\varepsilon }_{s}-{\varepsilon }_{\infty }\right){4{\pi }^{2}f}_{0}^{2}}{{4{\pi }^{2}f}_{0}^{2}+i2\pi f\delta -4{\pi }^{2}{f}^{2}}$$
(2)

The proposed absorber demonstrates excellent impedance matching (Supplementary Fig. S7), achieving a −10 dB EAB of 11.4 GHz (6.2–17.6 GHz). Remarkably, it maintains over 90% absorption efficiency for waves with TE or TM polarizations near 50° oblique incidence at 8 GHz (Fig. 3d). Across the entire operating band, the incident angles at which the absorption remains above 80% reach up to 50° for TE polarization and 60° for TM polarization (Supplementary Fig. S8). This good oblique incidence stability is primarily attributed to the dipole-based structure, which can be systematically analyzed using CMA (see Supplementary Note 2 for details)48,49. The ground plane introduced when the chainmail used as an absorber slightly modifies the current distribution on the structure. As shown in Fig. 3e, this alteration shifts the modal significance (MS) peak from ~10 GHz, which was originally associated with a strong dielectric resonance, down to around 8.5 GHz. The new peak position corresponds closely to the frequency demonstrating strong absorption and stable performance under oblique incidence. Multiple significant modes exhibit similar effective areas concentrated between 8 and 10 GHz. The corresponding radiation patterns (Fig. 3f) demonstrate distinct directional characteristics of the simultaneously excited modes. Clearly, their complementary spatial coverage can effectively cover the upper half-space, suggesting the absorber’s excellent wide-angle performance.

Deformability analysis

To ensure optimal conformal adaptability of the absorber on complex curved surfaces, the chainmail units must maintain sufficient rotational freedom to prevent being stuck. We analyzed the tilting capability of the unit along the four directions where the structure is most prone to deformation (Fig. 4a), corresponding to 0° (1#), 90° (2#), 45° (3#), and 135° (4#) relative to the X-axis in Fig. 3a. Due to the structural anisotropy, the tilting modes in directions 1 and 2 exhibit noticeable differences, whereas those in directions 3 and 4 are identical (Fig. 4b). In direction 1, the tilt angle increased with larger l1, smaller d1, or reduced overall thickness (Fig. 4c). Meanwhile, tilting in direction 2 demonstrated a two-stage mechanism. Initially, l1, d1, and h1 collectively determine whether the bottom edge of the blue unit will be obstructed by the central axis of the gray unit during tilting. The regions containing spheres in Fig. 4d indicate configurations where this obstruction occurs. Once the unit crosses the central axis, the tilt angle became solely dependent on h1 and d1 (Fig. 4e), typically achieving greater inclination.

Fig. 4: Deformability of the chainmail in principal directions.
Fig. 4: Deformability of the chainmail in principal directions.
Full size image

a Four most susceptible tilting directions (labeled 1–4). b Schematic diagram of tilt mode (blue: active units; gray: constrained units). c Tilt angle analysis for direction 1. d Tilt angle in direction 2 when stuck by the central axis. e Tilt angle in direction 2 when it is possible to cross the central axis. f Achievable size ranges for full overhang in direction 3 and 4. g Conformal demonstration of the chainmail absorber on a human hand.

The motion of the units inclined in directions 3 or 4 is more complex due to the involvement of a certain amount of torsion. We therefore estimated the feasible size range for complete overhanging, as shown in Fig. 4f. Clearly, the available deformation space in these directions is larger compared to directions 1 and 2. Detailed angle functions for each direction are provided in the Supplementary Note 3. The proposed design achieves a 41.2° tilt in direction 1, a 60° tilt in direction 2, and complete overhanging in directions 3 and 4. When freely placed, the sample conformed well to fingers, palms, and wrists, demonstrating good conformal adaptability (Fig. 4g).

Absorption performance under conformal conditions

To validate the applicability of the proposed chainmail absorber, we conformally mounted it onto various metallic curved surfaces and measured its radar cross-section (RCS) reduction to evaluate absorption performance. First, we demonstrated its application on a typical semi-cylindrical surface and compared it with three conventional absorbers: Salisbury screen, metasurface, and honeycomb (Fig. 5a). Full-scale simulations were performed and cross-verified with measured monostatic RCS data from fabricated prototypes to confirm reliability. As shown in Fig. 5b and Supplementary Fig. S9, the high agreement between simulated and measured results confirms that the effectively designed chainmail units are uniformly distributed in a tangential arrangement on the cylindrical surface, consistent with the modeling. The trends and peak positions of the monostatic RCS curves closely match those of the planar absorption curves, demonstrating a high level of performance stability. The structure also exhibits high absorption efficiency in bistatic RCS simulations, achieving a maximum reduction exceeding 30 dB, which is comparable to its monostatic RCS performance (Fig. 5c).

Fig. 5: Conformal absorption performance on complex curved surfaces.
Fig. 5: Conformal absorption performance on complex curved surfaces.
Full size image

a Structural comparison between the chainmail absorber and three conventional absorbers on semi-cylindrical surfaces. b Comparison of the monostatic RCS of a metal semi-cylindrical surface (height: 295 mm, radius: 93 mm) before and after being covered by the absorber. c Comparison of the 3D bistatic RCS simulation results of a metal semi-cylindrical surface before and after being covered by the absorber. d Comparison of the measured absorption bandwidth of the chainmail absorber with full-size, equivalent medium, and conventional absorbers’ simulation results. e Comparison of the measured average absorptivity of the chainmail absorber at different frequency bands with full-size, equivalent medium, and conventional absorbers’ simulation results. Photographs of the absorber conformal to a saddle surface (f) and a spherical (two convex and two concave)-planar combined surface (g), with insets showing the Gaussian curvature distribution. Comparison of the monostatic RCS of a metal saddle surface (h) and a metal spherical-planar combined surface (i) before and after being covered by the absorber.

Comparative studies with conventional absorbers of the same thickness further highlight its advantages. In the planar state, the absorption performance of the competing structures has already been optimized for the best bandwidth. Except for the Salisbury screen, which has reached its upper limit of 10 GHz, the bandwidths of the metasurface and honeycomb are slightly superior to that of the chainmail absorber (Supplementary Fig. S10). However, their performance degraded substantially when conformed to curved surfaces. Benefiting from their initial performance advantage, the metasurface and honeycomb exhibit slightly higher average absorptivity than the chainmail absorber in the C band. However, none of the conventional structures could match the chainmail’s performance in the X and Ku bands (Fig. 5d). Compared to their original performance, the chainmail absorber’s average absorptivity decreased by only 0.049, whereas the Salisbury screen, metasurface, and honeycomb suffered reductions of 0.144, 0.138, and 0.103, respectively (Supplementary Fig. S11). In terms of bandwidth retention, the chainmail absorber maintained ~80% absorption (93.78%) across its designed frequency range with only minor discontinuities, while also exhibiting the most extensive 90% absorption zones (63.25%). In contrast, all conventional structures demonstrated numerous frequency bands with performance degradation (Fig. 5e). In addition, we provide a performance comparison (Supplementary Table S2) with typical flexible absorbers10,14,50,51,52. The results indicate that after being conformed to a cylindrical surface, the proposed absorber demonstrates better performance in both average absorptivity and bandwidth retention compared to other structures, underscoring its potential for conformal absorption applications.

For non-developable surfaces with non-zero Gaussian curvature, non-stretchable planar structures cannot conform perfectly without distortion. Unsurprisingly, Salisbury screen and metasurface (with resin film substrate) and honeycomb (aramid paper-based) face insurmountable challenges under such conditions. Even flexible rubber materials can hardly achieve firm conformal adhesion to non-developable surfaces through stretching (Supplementary Fig. S12). In contrast, the ample spacing between chainmail units allows it to conform seamlessly to complex 3D curvatures by simply bonding its edges and key inflection points. When applied to highly curved non-developable surfaces, such as saddle (Fig. 5f) and spherical-planar combined surface (Fig. 5g), the absorber achieved high-efficiency absorptivity of 86.6% and 90.3% within the 4–18 GHz band (Fig. 5h, i and Supplementary Fig. S13). These experimental results comprehensively demonstrate that the proposed chainmail absorber can adapt to various complex surfaces while offering better absorption performance retention compared to conventional structures. Moreover, as the chainmail absorber is free from internal stresses and requires no pre-designed rigid dielectric layers of specific curvature, it can be flexibly and repeatedly deployed on arbitrary surfaces with minimal preparation.

Switchable design and dynamic absorption performance

Since the structure of the chainmail unit remains unchanged and only minor localized deformations occur in the array during conformal adaptation, the absorption performance remains stable. Moreover, this design allows for a considerable range of motion for the units (with side lengths ranging from 24 cm to 30 cm). By precisely controlling the movement spacing between units, we can accordingly shift the operating frequency band of the absorber, much like chainmail with varying densities is suited for defense against different types of attacks.

Conventional mechanically reconfigurable metamaterials typically require a dedicated control module for each unit, often resulting in a complex and bulky overall structure. Leveraging the fabric-like properties of chainmail metamaterials, we designed a switchable absorber (Fig. 6a, Supplementary Fig. S14) that uses only a single stepping motor to control the entire array’s dimensions. First, four elastic bands pass through the middle layer of the outermost chainmail units. Then, four locating pins are inserted into the outermost empty slots at the four corners of the chainmail absorber. The elastic bands are pre-stretched (to suppress lateral deformation) and secured to the pins. This setup allows equal distance movement of the four sides with the combination of motor-driven pins and elastic bands. Furthermore, once the desired performance state is achieved, no continuous power supply is required, resulting in a low-power design.

Fig. 6: Switchable design and dynamic absorption performance.
Fig. 6: Switchable design and dynamic absorption performance.
Full size image

a Schematic of the frequency-switchable absorption system (Gray arrows indicate the direction of movement for the locating pins). b Expansion and contraction process of the absorber (with a step size of 1 cm for the side length). Green represents the areas that have not yet moved during expansion, while red represents the areas that have not yet moved during contraction. c Simulated reflection loss heatmap of the absorber with uniformly varied side lengths. d Loss distribution for contraction and expansion states at 5.5 GHz and 8 GHz. e Effective permittivity of the absorber with different side lengths. f Effective permeability of the absorber with different side lengths. Measured reflection loss of the absorber with different side lengths during expansion (g) and contraction (h).

Reflection loss heatmap of the absorber with different side lengths is shown in Fig. 6c. As the absorber size decreases, the primary absorption peak near 8 GHz gradually shifts toward 5.5 GHz, thereby extending the low-frequency absorption performance. The overall bandwidth initially broadens to 12.5 GHz before narrowing to 2.5 GHz. During the switching process, the effective absorption frequency range covers 4.6 GHz to 18 GHz. Notably, the optimal switched superposition performance (24 cm + 27 cm configurations) has already surpassed what can be achieved by a fixed non-magnetic absorber with the same thickness (actual need 5.733 mm). This confirms that the proposed dynamic switching strategy successfully circumvents the Rozanov limit (Eq. (3) where λ is the wavelength, R is the reflectivity, and dmin is the minimum thickness achievable at the corresponding reflectivity)53. Full-wave simulation results show that the loss distribution, electric field distribution, and magnetic field distribution patterns at the absorption peak frequencies remain nearly identical before and after switching, with only intensity variations observed (Fig. 6d, Supplementary Fig. S15). The variation in effective electromagnetic parameters further highlights this characteristic. As the size decreases, the permittivity gradually increases in magnitude but ultimately follows a Lorentz resonance trend (Fig. 6e, Supplementary Fig. S16). And the resonant frequency points show only minor shifts (Supplementary Table S3). Meanwhile, since the gap between the upper and lower layers remains unchanged, the magnetic resonance behavior induced by parallel currents exhibits even less pronounced variation (Fig. 6f).

$${d}_{\min }=\frac{1}{4{\pi }^{2}}\left|{\int }_{\lambda min }^{\lambda max }{{{\mathrm{ln}}}}\left(R(\lambda )\right)\right|{{{\rm{d}}}}\lambda$$
(3)

According to the quarter-wavelength theory54, the above properties and constant thickness enable the structure to achieve a smooth transition during dynamic switching while facilitating targeted design optimization. Benefiting from this advantage, despite the simple mechanism dominated by four elastic bands causing non-uniform changes in the chainmail array during expansion and contraction (Fig. 6b, Supplementary Video S1), the performance variations remain relatively uniform. In practical dynamic switching experiments, the absorption peak and bandwidth shift trends align well with simulations, and the absorption performance in fully expanded and contracted states can be stably and consistently reproduced after switching (Fig. 6g, h). To thoroughly validate its stability, five consecutive switching tests were conducted, with the results illustrated in Supplementary Fig. S17. The absorption curves under different side-length states exhibit a narrow 95% confidence interval and a standard deviation below 0.015 across most of the operating band, indicating excellent reproducibility of reconfigurable performance. Furthermore, a demonstration of 100 switching cycles was carried out as shown in Supplementary Video S2. Highly consistent deformation patterns were observed throughout the tests. After the experiment, the absorber successfully passed load-bearing validation, sustaining 75 kg in the z-direction and 5 kg in the x-direction without failure and degradation of absorption performance (Supplementary Figs. S18, S19). In summary, the proposed structure demonstrates satisfactory robustness for practical applications.

As shown in Supplementary Table S4, compared with typical voltage-controlled55, thermally tuned56, and circuit-tuned switchable absorbers57,58,59,60, the chainmail absorber offers a much wider tunable frequency range. In comparison with other types of mechanically reconfigurable structures61,62, particularly those relying on thickness variation for band switching63,64,65, the proposed structure achieves a markedly reduced profile, with merely 20% of their thickness. Except for its slower switching speed relative to circuit-tuned designs, the chainmail absorber generally exhibits excellent overall performance, including lightweight, broadband operation, low power consumption, and low cost. These attributes endow it with unique advantages in certain scenarios. For example, future advanced driver-assistance systems rely on both high-frequency automotive radar and low-frequency radio access technologies to realize higher-level autonomous driving, where the two systems face distinct EMC/EMI requirements66,67. When driving in suburban or unfamiliar areas, radar serves as the primary means of environmental sensing, and chainmail absorbers deployed at appropriate positions can operate in the high-frequency band to suppress echoes caused by multipath effects. Once entering highways or dense urban traffic, V2X communication becomes dominant, and the absorbers can be switched to the low-frequency band to mitigate communication interference. They can also be applied in microwave anechoic chambers, where they provide targeted coverage for fixtures and auxiliary equipment and switch to the corresponding test band to reduce interference from non-test objects. As an extension, we envision an application scenario where the chainmail absorber is embedded within a dual-layer aircraft skin. Owing to its lightweight nature, the absorber would have minimal impact on the payload capacity of unmanned aerial vehicles (UAVs), ensuring their ability to perform long-endurance missions. In typical radar systems, low-frequency radars are employed for long-range detection, while high-frequency radars are used for short-range detection and target locking68,69. Accordingly, during defensive surveillance missions at long distances, the absorber can be switched to operate in the low-frequency band, whereas for close-range reconnaissance tasks, it can be switched to the high-frequency band. Simulations show that by switching the expansion/contraction states, a full-scale UAV can achieve significant RCS reduction at specific frequencies (Supplementary Fig. S20), with the head-on RCS decreasing by 26.1 dB and 15.6 dB at 5.5 GHz and 8 GHz, respectively. It also impairs the reliability of inverse synthetic aperture radar (ISAR) imaging (Supplementary Fig. S21), enhancing the UAV’s stealth capabilities against detection.

Discussion

This study presents a conformable and switchable microwave metamaterial absorber inspired by chainmail structure, which demonstrates satisfactory comprehensive advantages in both multifunctionality and practicality (Supplementary Table S5). On one hand, the dipole-like resonance characteristics of the chainmail units enable the equivalent electromagnetic parameters to fall within the optimal range over a broader spectrum, achieving broadband absorption from 6.2 to 17.6 GHz with a thickness of mere 5.5 mm. On the other hand, the unique rigid-yet-flexible nature of interlocking chainmail structure ensures deformation-free lossy components when conformed to curved surfaces. As a result, it exhibits excellent adaptability and stability on both developable and non-developable surfaces, with the absorptivity decreasing by only 0.049. Furthermore, leveraging the fabric-like properties of chainmail, we developed a lightweight, mechanically reconfigurable system comprising chainmail, elastic bands, locating pins, and a single motor-driven module. Dynamic modulation of unit density allows the absorber’s operational frequency to be tuned across 4.6–18 GHz. Overall, this work extends the application of 3D-printed smart fabrics into the field of microwave absorption. The design strategy and systematic mechanism analysis provide new insights for developing advanced conformable and switchable microwave absorbers.

Methods

Materials and fabrication

PLA pellets 4032D is produced by NatureWorks LLC., diffusion oil TSF-96-1000 is produced by GE Toshiba Silicone Co., Ltd., and CNT is provided by Shandong Dazhan Nanomaterials Co., Ltd. The lossy 3D printing filament was custom-fabricated. CNT-PLA pellets were prepared by blending 10 wt% CNT with 89 wt% PLA in a twin-screw extruder (Ruiming SJZS-10B), with the addition of 1 wt% diffusion oil to enhance flowability. These pellets were then extruded into filaments with a diameter of ~1.75 mm using a single-screw extruder (3DPANY HT400). The lossless pure PLA filament was purchased from Lanbo printing Consumables Co., Ltd. Finally, the chainmail absorber sample comprising 14 units per column was fabricated using a dual-nozzle FDM printer (Wiiboox D400). The key printing parameters were set as follows: printing speed of 30 mm/s, layer height of 0.2 mm, nozzle diameter of 0.4 mm, nozzle temperature of 200 °C, and heated platform temperature of 55 °C. Supplementary Video S3 demonstrates the 3D printing process.

Characterization

The cross-sectional morphology of the PLA and CNT-PLA filament was characterized using a field emission scanning electron microscopy (Hitachi S4800). The compressive performance of various units in different directions was evaluated via electronic universal testing machine (NSS CMT5105). The complex permittivity was measured using a vector network analyzer (Agilent E8363A) via the transmission-reflection method. The waveguide fixture was selected over coaxial fixture to mitigate air gap-induced measurement inaccuracies, as the latter requires more stringent dimensional tolerances for the toroidal samples’ inner and outer diameters. The test samples were 3D-printed rectangular blocks with precisely controlled dimensions (length: 22.86 mm, width: 10.16 mm, thickness: 2 mm). The permittivity of CNT-PLA measured in X-band was extrapolated to 2–18 GHz via Cole-Cole model for design optimization (Supplementary Fig. S5)70. The reflection loss of absorber was measured by an improved arch system with a pair of broadband horn antennas (1–18 GHz). The RCS of metal surfaces before and after being covered by the absorber was measured in a microwave anechoic chamber, using four pairs of standard gain horn antennas (4–6, 6–8, 8–12, 12–18 GHz).

Simulation

All electromagnetic simulations were performed in CST Microwave Studio. Frequency-domain solver simulations were performed to analyze both reflection loss and effective electromagnetic parameters. The chainmail unit edges were aligned parallel to the X and Y axes and perpendicular to the Z axis. The X and Y boundaries were set as unit cells, the bottom Z boundary was set as an electric boundary, and the top was defined as an open space. Multilayer solver was used to compute the characteristic modes of the PEC chainmail. Except for the bottom, which was set as an electric boundary, all other boundaries were defined as open space. Integral equation solver was used to simulate the RCS of the metal surface before and after absorber coverage. The Salisbury screen has a sheet resistance of 350 Ω/sq. The metasurface features a sheet resistance of 44.9 Ω/sq with a unit period of 8 mm, square loop width of 1.1 mm, and inter-element spacing of 0.33 mm. The honeycomb structure exhibits a sheet resistance of 500 Ω/sq with a cell edge length of 1.8 mm and wall thickness of 0.25 mm. The permittivity of dielectric layer is 1.1 and the permittivity of the honeycomb wall is 3.4. All absorbers have a thickness of 5.5 mm. Asymptotic solver was used to simulate the RCS and perform ISAR imaging calculations for a 1:1 UAV model equipped with the chainmail absorber (replaced by equivalent medium layer). The motion analysis module in SolidWorks was employed to simulate the tilting behavior of chainmail units under gravitational loading (Fig. 4b). The hardware configuration of the workstation used for the simulations is as follows: CPU: AMD EPYC 9474F, GPU: RTX 3090, RAM: 1.5 TB DDR5.