Spherical CIP (purity > 99.9%) with an average particle size of 2-3 μm was purchased from BASF SE (Ludwigshafen, Germany). LM (EGaIn, 78.5 wt.% Ga/21.5 wt.% In) was supplied by Beijing Zhongke Yannuo New Materials Technology Co., Ltd. Reagent-grade glucose (99.9%) and polyethylene glycol (PEG-400) were obtained from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Deionized (DI) water was provided by Beijing Jindong Tian Zheng Precision Chemical Co., Ltd. All chemicals were of analytical grade and used as received without further purification.

Figure 1A schematically illustrates the synthesis of CIP@C@LM core-shell particles. CIP (1.0 g) was dispersed in DI water (200 mL) containing glucose (9.0 g) under mechanical stirring (350 rpm) for 12 h. The precursor was magnetically collected, dried at 60 °C for 8 h, and subsequently calcined in air at 300 °C for 2 h (heating rate: 3 °C min^-1, furnace cooling) to yield CIP@C particles. The obtained powder was resuspended in DI water (200 mL) containing PEG-400 (1 mL) and stirred for 2 h. Separately, LM (0.2 g) was sonicated in DI water (25 mL) at 600 W for 15 min to generate microdroplets. This LM suspension was added to the CIP@C dispersion and mechanically stirred for 8 h. During this process, intermittent ultrasonication (200 W, 30 °C) was applied. The CIP@C@LM sample was prepared using four ultrasonication cycles (15 min per cycle, with 30 min intervals), giving a total ultrasonication duration of 60 min. For comparison, CIP@C@LM1 and CIP@C@LM2 were prepared with total ultrasonication durations of 30 min and 120 min, respectively, under otherwise identical conditions. The final product was magnetically separated and dried at 60 °C for 8 h.

To prepare the coatings, the synthesized powders were blended with a silicate-based high-temperature adhesive (3:2 mass ratio) and homogenized. The slurry was applied via brush-coating, followed by conditioning in a humidity chamber [25 °C, 60% Relative Humidity (RH)] for 24 h. A stepwise thermal curing profile was implemented: 60 °C for 8 h, 80 °C for 6 h, and 100 °C for 2 h.

Morphological characteristics were analyzed using field-emission scanning electron microscopy (SEM, Hitachi S-4800) and transmission electron microscopy (TEM, JEOL JEM-F200). The phase composition was investigated by X-ray diffraction (XRD, Cu Kα₁ radiation, λ = 1.54056 Å) with a 2θ scanning range from 5° to 90° at a scan rate of 10 °/min. Fourier transform infrared spectroscopy (FTIR, 4,000-400 cm^-1) was conducted using a Bruker Vector33 spectrometer. Raman spectroscopy was performed on a Horiba LabRAM HR Evolution spectrometer with a 532 nm laser. The chemical states of the elements were analyzed by X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-Alpha, USA). Electron paramagnetic resonance (EPR) spectra were recorded using a Bruker EMX-10/12 EPR spectrometer. Magnetic properties were measured at room temperature with a vibrating sample magnetometer (VSM, Model 3105, East Changing Technologies, China). Electromagnetic properties in the frequency range of 2-18 GHz were characterized using a vector network analyzer (VNA, Agilent E5071C) via the coaxial line method. For this, CIP-based powders were mixed with paraffin wax (mass ratio of 3:2) and pressed into toroidal rings (inner diameter: 3.04 mm, outer diameter: 6.95 mm, thickness: 2.00 mm). Electrochemical corrosion and impedance measurements were carried out using an electrochemical workstation (CHI760E, Shanghai Chenhua, China), employing a three-electrode system in a 3.5 wt.% NaCl solution. The working electrode was a glassy carbon electrode, with an Ag/AgCl reference electrode and a platinum mesh counter electrode. Tafel polarization curves were recorded within a voltage range of -0.5 V to 0.5 V vs. Open circuit potential (OCP), at a scan rate of 5 mV/s. Thermal infrared radiation intensity of coatings was assessed using a thermal infrared imaging device (KREVOR FLIR ONE3, FLIR Systems, Inc., USA).

Figure 1B-Q track the morphological evolution and interface assembly of the CIP@C@LM heterostructure. The pristine CIP [Figure 1B] exhibits a characteristic spherical geometry with relatively smooth surfaces. Upon C encapsulation, a textured, conformal coating emerges [Figure 1C], confirming the successful deposition of the C interlayer, which serves as a buffer against core oxidation. Subsequently, the introduction of LM via ultrasonication triggers a dramatic morphological transformation, yielding a well-defined core-shell architecture enveloped by a dense array of nanorod (needle-like) nanostructures [Figure 1D]. This hierarchical growth leads to a discernible increase in particle size, with the size distribution shifting notably toward the 2-3 μm range [Supplementary Figure 1]. High-magnification SEM images [Figure 1E] reveal that the surface nanorods are self-assembled into rhomboidal patterns, uniformly distributed across the particle surface with discernible interparticle gaps. TEM analysis [Figure 1F-H] provides deeper insight into this microstructure, clearly resolving the radially oriented, rod-like crystallites that constitute the outer shell. Statistical analysis indicates that these crystallites possess an average length of ~62 nm [Supplementary Figure 2]. The crystalline nature of the shell is corroborated by high-resolution transmission electron microscopy (HRTEM) [Figure 1I], where clear lattice fringes with d-spacings of 2.407 Å and 2.645 Å are observed. These correspond well to the (111) and (301) planes of orthorhombic α-GaOOH^[36]. Additionally, the precise dimensions of these nanostructures are detailed in the magnified view [Figure 1J]. Acquired from this specific region, the corresponding selected area electron diffraction (SAED) [Figure 1K] shows distinct polycrystalline diffraction rings. The calculated d-spacings from these concentric rings are 2.645, 2.407, 2.259, 1.575, and 1.489 Å, which are indexed to the (301), (111), (002), (312), and (103) planes of the standard α-GaOOH phase (JCPDS PDF No. 54-0910)^[37], respectively. To elucidate the elemental distribution and interfacial chemistry, high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy-dispersive X-ray spectroscopy (EDS) mappings were performed [Figure 1L-Q]. The images distinctly delineate a sandwich-like structure: the Fe-rich core, the C interlayer, and the Ga/O-dominated outer shell. Notably, an O-enriched region is observed overlapping with the Fe/C interface, implying that partial surface oxidation of CIP occurred during the aqueous processing. Crucially, while Ga and O signals are strongly colocalized in the outer shell, the In signal is remarkably weak. This phenomenon can be attributed to the thermodynamically driven selective oxidation of the EGaIn alloy. Since Ga has a greater tendency to undergo oxidation than In^[38], it preferentially migrates to the surface to undergo hydrolysis and oxidation, forming the voluminous α-GaOOH framework. The unoxidized In segregates into isolated metallic nanodroplets, which are subsequently removed by intense ultrasonic shear and washing processes, resulting in a diminished In signal in the final powder. Collectively, these microstructural and compositional analyses confirm the successful construction of a hierarchical magnetic-dielectric heterostructure, where the abundant heterogeneous interfaces and unique acicular shell are anticipated to play a critical role in optimizing electromagnetic response.

Figure 2 summarizes the phase composition and defect chemistry of CIP, CIP@C, and CIP@C@LM. The XRD patterns [Figure 2A] show that CIP exhibits the characteristic peaks of metallic Fe (body-centered cubic, bcc) at 2θ = 44.7° and 82.3°, indexed to the (110) and (211) planes (JCPDS PDF No. 06-0696)^[39]. After C coating, the diffraction features become noticeably weakened, and the Fe (211) reflection at 82.33° is strongly suppressed, indicating a reduced crystallinity. Upon subsequent LM-assisted treatment, distinct new peaks emerge at 2θ = 21.5° and 37.2°, assignable to orthorhombic α-GaOOH (110) and (111) (JCPDS No.54-0910)^[40]. In addition, minor peaks at 2θ = 33.5° and 34.1° are detected, which match α-FeOOH (130) and (021) (JCPDS PDF No. 29-0713)^[41]. Combined with the nanorod shell morphology as shown in Figure 1M-P, these results indicate that aqueous ultrasonication drives α-GaOOH growth while partially converting the Fe surface into α-FeOOH. FTIR spectroscopy [Figure 2B] further supports the formation of the oxyhydroxides. The bands at 930 cm^-1 and 638 cm^-1 are attributed to the bending vibrations of -OH···O=^[42] and the stretching vibrations of Ga-O^[40], respectively, corroborating the presence of α-GaOOH. Raman spectra [Figure 2C] further probe the surface oxyhydroxide species and the C shell. For pristine CIP, the distinct Raman scattering signals detected at ~286 cm^-1 verify the existence of α-FeOOH^[43], and the signals at ~528 cm^-1 and ~650 cm^-1 corresponded to γ-FeOOH phase^[44]. The absence of γ-FeOOH peaks in XRD suggests a poorly crystalline and/or ultrathin surface layer. After C coating, the FeOOH-related peaks are markedly suppressed, while the carbon D and G bands emerge at ~1,330 cm^-1 and ~1,590 cm^-1[45], confirming C encapsulation. Following LM treatment, the α-FeOOH band strengthens and slightly upshifts, whereas the γ-FeOOH features weaken, indicating α-FeOOH enrichment near the interface. A new peak at ~394 cm^-1 appears, consistent with the A₁g mode of α-GaOOH^[41]. Furthermore, the redshifted D and G bands indicate increased defect density and interfacial charge transfer within the heterostructure.

As shown in Figure 2D-H, the XPS results further confirm the surface composition and bonding environment of the synthesized composites. Figure 2D clearly identifies the presence of C, Ga, In, and O signals in CIP@C@LM. The Fe 2p region [Figure 2E] shows remarkably weakened and less-resolved features. This attenuation is anticipated due to the limited escape depth of photoelectrons. In the C 1s spectrum [Figure 2F], peaks deconvoluted at 284.9, 286.9, and 288.4 eV correspond to C-C, C=O, and O-C=O species^[46], respectively. The Ga 2p spectrum [Figure 2G] exhibits characteristic spin-orbit doublets components at 1,117.7 eV (Ga 2p_3/2) and 1,144.6 eV (Ga 2p_1/2)^[47], with a binding energy separation of ~26.9 eV, indicative of the Ga³⁺ oxidation state in the α-GaOOH lattice. In the O 1s spectrum [Figure 2H], the dominant peak at 531.2 eV is characteristic of Ga-O lattice oxygen^[48], indicating that the outer surface is dominated by Ga-based oxide/oxyhydroxide species formed during the sonochemical hydrolysis. To assess defect states and unpaired electrons, EPR measurements were performed [Figure 2I]. A distinct resonance signal at g-factor (g) = 2.003, characteristic of O_v trapped with unpaired electrons^[49], is detected. Pristine CIP shows a weak O_v response, consistent with the presence of a thin native oxidized surface layer (e.g., FeOOH). C coating produces negligible change. However, LM treatment leads to a pronounced increase in the O_v signal intensity. This surge is consistent with the formation of oxyhydroxides (α-FeOOH and α-GaOOH) enriched with defect sites, likely originating from the rapid, non-equilibrium crystal growth driven by the high-energy sonochemical/aqueous conditions. Collectively, the XRD, FTIR, Raman, XPS, and EPR results confirm the coexistence of α-GaOOH and interfacial α-FeOOH. Furthermore, they demonstrate the defect-rich nature of the shell/interface^[50].

As shown in Figure 3, the concurrent growth of nanorods α-GaOOH/α-FeOOH shells can be attributed to interfacial oxidation, hydrolysis, and heterogeneous nucleation. First, regarding the Ga evolution [Figure 3A]: Upon ultrasonic dispersion, the liquid Ga droplets undergo interfacial oxidation to form a dynamic Ga₂O₃ skin (Equation 1). The high-intensity acoustic cavitation continuously renews the gas-liquid and liquid-solid interfaces, accelerating mass transport and disrupting the oxide skin. This promotes the hydration of Ga₂O₃ into low-solubility Ga(OH)₃ (Equation 2)^[51], which subsequently converts to α-GaOOH via a dissolution-recrystallization and dehydration process^[52] (Equation 3). The cavitation environment also introduces abundant defect sites (e.g., O_v) and active hydroxyl groups on the Ga-based surface. Simultaneously, the Fe pathway is modulated by the C shell. The micropores and defects in the C coating act as diffusion valves, allowing limited access of H₂O and dissolved O₂ to the CIP surface [Figure 3B]. This restricts the anodic dissolution of Fe⁰ (Fe⁰ → Fe²⁺/Fe³⁺) (Equation 4) to a rate-controlled regime, preventing the structural collapse of the magnetic core [Supplementary Figure 3]. Coupled with the cathodic reduction of dissolved oxygen, hydroxide ions (OH^-) are generated (Equation 5), establishing a localized alkaline microenvironment that facilitates the hydrolysis of Fe ions into FeOOH intermediates (Equations 6-7). Crucially, the synergistic growth is driven by structural and energetic compatibility. Since both α-GaOOH and α-FeOOH crystallize in the isostructural diaspore/goethite framework^[53] (built from edge-sharing octahedral chains), the in-situ formed Ga-based species serve as ideal templates. The Ga₂O₃/Ga(OH)₃ interphase not only lowers the interfacial energy but also provides a hydrophilic, hydroxylated surface that facilitates the heterogeneous nucleation of Fe(OH)_x. Furthermore, the defect sites (O_v) serve as active anchoring centers, lowering the nucleation barrier and promoting the epitaxial-like assembly of α-FeOOH along the α-GaOOH lattice. However, this self-assembly process is strictly governed by reaction kinetics. As schematically illustrated in Figure 3C, the morphological evolution is highly dependent on the ultrasonication duration: sparse initial growth is observed at 30 min, which develops into dense, ordered nanorod arrays at 60 min [Figure 1D and E], while prolonged sonication to 120 min leads to disordered and detached nanostructures. Correspondingly, increasing the ultrasonication duration induces a morphological transition from ordered nanorod arrays to disordered textures, eventually leading to discrete, detached nanostructures [Figure 3D].

(1) $$ 4 \mathrm{Ga}+3 \mathrm{O}_{2}=2 \mathrm{Ga}_{2} \mathrm{O}_{3} $$

(2) $$ \mathrm{Ga}_{2} \mathrm{O}_{3}+3 \mathrm{H}_{2} \mathrm{O}=2 \mathrm{Ga}(\mathrm{OH})_{3} $$

(3) $$ \mathrm{Ga}(\mathrm{OH})_{3}=\mathrm{GaOOH}+\mathrm{H}_{2} \mathrm{O} $$

(4) $$ \mathrm{Fe}^{0} \rightarrow \mathrm{Fe}^{2+} / \mathrm{Fe}^{3+} $$

(5) $$ \mathrm{O}_{2}+\mathrm{H}_{2} \mathrm{O}+\mathrm{e}^{-}=\mathrm{OH}^{-} $$

(6) $$ \mathrm{Fe}^{3+}+3 \mathrm{OH}^{-}= \mathrm{FeOOH}+\mathrm{H}_{2} \mathrm{O} $$

(7) $$ 4 \mathrm{Fe}^{2+}+\mathrm{O}_{2}+8 \mathrm{OH}^{-}=4 \mathrm{FeOOH}+2 \mathrm{H}_{2} \mathrm{O} $$

To characterize the EMW absorption performance, RL and impedance matching are generally calculated based on experimentally measured f dependent electromagnetic parameters. According to the transmission line theory, the relevant calculation formulas are expressed as follows^[54]:

(8) $$ R L=20 \lg \left|\frac{Z-1}{Z+1}\right| $$

(9) $$ Z=\frac{Z_{i n}}{Z_{0}}=\sqrt{\frac{\mu_{r}}{\varepsilon_{r}}} \tanh \left(j \frac{2 \pi f d}{c} \sqrt{\mu_{r} \varepsilon_{r}}\right) $$

Here, Z represents the normalized input impedance relative to free space, Z_in is the input impedance of the material, Z₀ is the characteristic impedance of free space, d is the absorber thickness, and ε_r (ε_r = ε’ - jε”) and μ_r (μ_r = μ’ - jμ”) are the complex permittivity and permeability. Figure 4 compares the EMW absorption performance of CIP, CIP@C and CIP@C@LM. For pristine CIP, the EMW absorption remains modest, with an RL_min of -16.29 dB at 18 GHz (d = 5.5 mm) and an EAB_max of 4.17 GHz (13.83-18 GHz) at a d of 2.0 mm. This limited performance is attributable to the low dielectric response of the CIP/paraffin composites at a filler ratio of 60 wt.%. The C coating significantly improves the attenuation capability, with CIP@C delivering an RL_min of -52.68 dB at 14.88 GHz (d = 1.68 mm). More importantly, after the LM-assisted conversion, the final CIP@C@LM heterostructure achieves an RL_min of -52.60 dB at 14.24 GHz (d = 2.31 mm), together with an EAB_max of 7.56 GHz (10.44-18 GHz) at a d of 2.08 mm. The EMW absorption maps of CIP@C@LM show a double-peak feature, indicating the concurrent contributions of multiple loss processes. Furthermore, EMW absorption performance deteriorates when the ultrasonication time is either insufficient or excessive, which can be attributed to inappropriate microstructures as shown in Figure 3C. When the treatment time is too short, it is difficult to form effective heterointerfaces for interfacial polarization. Meanwhile, too long treatment generates redundant non-conductive phases and lowers electrical conductivity [Supplementary Figure 4]. The corresponding electromagnetic-parameter comparison further confirms that the optimized oxidation state at 60 min results in dielectric-loss-dominated attenuation [Supplementary Figure 5].

Overall, as shown in Figure 5A-C, the EMW absorption maxima of all samples follow the quarter-wavelength (l/4) cancellation principle. When the absorber thickness satisfies^[55]:

(10) $$ d_{m}=\frac{n \lambda}{4}=\frac{n c}{4 f_{m} \sqrt{\left|\varepsilon_{r} \mu_{r}\right|}}(n=1,3,5 \ldots) $$

where d_m is the matching thickness of RL_min, c is the velocity of light in a vacuum, n is the odd-order matching number and f_m is the matching frequency corresponding to the absorption peak. The reflected wave at the absorber/air interface is approximately out of phase with the incident wave, leading to destructive interference and a pronounced absorption peak. Accordingly, to achieve effective EMW absorption at lower f_m, a larger d_m is generally required to satisfy the phase cancellation condition. This thickness-f correlation also indicates that the absorption behavior of the system is governed not only by intrinsic dielectric/magnetic losses, but also by the combined effects of phase delay during wave propagation and impedance matching. As the d_m increases, the effective propagation path lengthens and multiple reflections/interference become more prominent, driving the peak to lower f and, in some cases, giving rise to secondary peaks. Notably, CIP@C@LM exhibits dual absorption peaks at a fixed d_m, suggesting that the material simultaneously satisfies the phase cancellation condition at different orders (e.g., n = 1 and n = 3), resulting in multi-order matching. In addition, concurrent intrinsic loss processes, such as polarization relaxation and magnetic resonance, further enable different f windows to reach more favorable impedance matching and energy dissipation^[56]. Meanwhile, impedance matching is a prerequisite for efficient EMW absorption because it determines the fraction of incident electromagnetic energy that enters the absorber rather than being reflected at the surface. When the normalized input impedance Z approaches 1, more EMW can penetrate the material; in practice, the region 0.8 < Z < 1.2 is commonly regarded as indicative of good matching^[57]. As shown in Figure 5D-F, the normalized input impedance curves reveal a frequency-dependent matching behavior after heterostructure construction. Compared with CIP@C, CIP@C@LM exhibits somewhat weakened impedance matching in the low-frequency region, whereas more favorable matching behavior is observed in the middle and high frequency regions, particularly within its main EAB of 10.44-18 GHz [Supplementary Table 1]. This improvement is mainly attributed to the α-GaOOH-rich shell, which reduces excessive dielectric storage and alleviates the impedance mismatch caused by overly high ε’ in the middle and high frequency regions. Meanwhile, the introduced heterogeneous interfaces, hydroxyl-related dipoles, and defect-associated polarization centers maintain sufficient dielectric loss, thereby enabling a better balance between electromagnetic-wave entry and internal dissipation in the relevant absorption region. Beyond wave entry, EMW absorption fundamentally relies on the dissipation of the transmitted energy through dielectric and magnetic losses, ultimately converting EMW energy into heat. A key descriptor of dissipation strength is the attenuation constant α, which quantifies the exponential decay rate of the wave amplitude during propagation in the medium. Physically, α is closely associated with intrinsic loss terms and is further amplified by multiple reflections and an extended effective propagation path within the absorber, as expressed by^[58]:

(11) $$ \alpha=\frac{\sqrt{2} \pi f}{c} \times \sqrt{\left(\mu^{\prime \prime} \varepsilon^{\prime \prime}-\mu^{\prime} \varepsilon^{\prime}\right)+\sqrt{\left(\mu^{\prime \prime} \varepsilon^{\prime \prime}-\mu^{\prime} \varepsilon^{\prime}\right)^{2}+\left(\mu^{\prime} \varepsilon^{\prime \prime}-\mu^{\prime \prime} \varepsilon^{\prime}\right)^{2}}} $$

As shown in Figure 5G, α differs substantially among the three materials. CIP exhibits the lowest α across the measured range, indicating the weakest overall dissipation capability. In the 11-15 GHz region, α of CIP@C@LM exceeds that of CIP@C, demonstrating a stronger attenuation ability in this band, consistent with its improved absorption response. To benchmark the performance of CIP@C@LM, we further compare it with representative CIP-based composites reported recently^{[3,6,7,8,9,11,14,25]} [Figure 5H]. In comparison, despite the simpler structural design and the relatively low filler loading (60 wt.%), our CIP@C@LM still delivers a competitive EAB_max of 7.56 GHz, underscoring the effectiveness of the interfacial architecture in balancing impedance matching and attenuation. For a normalized evaluation of the bandwidth performance, the fractional bandwidth is calculated as FBW = (f_H - f_L) / [(f_H + f_L)/2] × 100%, where f_H and f_L represent the upper and lower frequencies of the effective absorption bandwidth, respectively. Based on the EABmax ranges shown in Figure 4G-I, the FBW values of CIP, CIP@C, and CIP@C@LM are 26.2%, 37.9%, and 53.2%, respectively. Notably, CIP@C@LM achieves the highest FBW of 53.2% over 10.44-18.00 GHz, confirming its substantially broadened effective absorption coverage after heterostructure construction. Loading-dependent measurements confirm that, in the present system, strong and broadband absorption is achieved at 60 wt.% [Supplementary Figure 6].

The EMW absorption behavior is intrinsically governed by the dielectric and magnetic loss capabilities, as detailed in Figures 6 and 7. As shown in Figure 6A, VSM measurements reveal high M_s for all samples. As expected, introducing non-magnetic shells progressively dilutes the magnetic phase, leading to a stepwise decrease in M_s from 203.3 emu/g (CIP) to 187.1 emu/g (CIP@C) and 179.1 emu/g (CIP@C@LM). The corresponding coercivities (H_c) are 0.29, 49.96, and 60.97 Oe, respectively. The increase in H_c indicates enhanced resistance to magnetization reversal after interfacial reconstruction. Although α-FeOOH is present only as a minor interfacial phase, its formation at the CIP surface introduces an additional magnetic component into the heterogeneous interface, thereby enriching the local magnetic structure of CIP@C@LM^[59-61]. To assess the magnetic loss contribution, the eddy-current loss coefficient (C₀) is commonly used and defined as C₀ = μ”(μ’)^-2f^-1[62], where μ’ and μ” are the real and imaginary parts of the relative complex permeability, respectively. When C₀ remains nearly constant with f, C₀ is considered dominant. As shown in Figure 6B, magnetic loss in the 2-18 GHz range is mainly associated with eddy-current loss and natural resonance in pristine CIP. For CIP@C@LM, C₀ shows only minor variation as f increases, suggesting that eddy-current loss governs the magnetic dissipation, while a weak resonance feature appears in the 10-13 GHz range, indicative of a relatively modest natural resonance contribution. The f-dependent μ_r further supports these interpretations. In Figure 6C, μ’ follows the order CIP > CIP@C@LM > CIP@C. The decrease in μ’ after C encapsulation originates from the dilution of the magnetic CIP phase, whereas the higher μ’ value of CIP@C@LM relative to CIP@C indicates that the α-FeOOH-containing heterogeneous interface contributes to the enhanced effective permeability response after LM-assisted interfacial reconstruction. The μ” spectra [Figure 6D] show that CIP@C exhibits the highest μ” in the 2-6 GHz region, whereas CIP dominates in the 6-18 GHz range, confirming that pristine CIP retains the strongest intrinsic magnetic loss at higher f. However, despite its favorable magnetic loss, CIP alone still delivers limited EMW absorption because the composite suffers from insufficient dielectric loss and poor impedance matching under the given loading, highlighting the need to enhance the dielectric contribution. Consistently, the magnetic loss tangent (tan δ_μ = 0.24~0.35, Figure 6E) of CIP@C remains relatively high below ~13 GHz, while CIP shows the largest tan δ_μ at 13-18 GHz, reflecting the high-f nature of its dominant resonance-related losses. For CIP@C@LM, the frequency-dependent μ” and tan δ_μ responses reflect the modified permeability behavior and retained magnetic-loss contribution after interfacial reconstruction. Together with its higher μ’ than CIP@C, these results indicate that the minor interfacial α-FeOOH enriches the local magnetic structure and contributes to the magnetic response of CIP@C@LM through the heterogeneous interface.

As for the dielectric loss contribution, the pristine CIP exhibits the lowest ε’ at a filler loading of 60 wt.%, reflecting its weak dielectric contribution and the difficulty in establishing an effective polarization/conduction response in the paraffin matrix as shown in Figure 7A. After introducing the C interlayer, the real part of complex permittivity (ε’) increases markedly, indicating enhanced charge-storage capability and stronger interfacial/electronic polarization. Upon further formation of α-GaOOH, ε’ decreases to some extent, suggesting that the outer oxidized/hydroxylated shell partially suppresses the overall dielectric response by reducing the effective conductivity. The evolution of imaginary part of complex permittivity (ε”) [Figure 7B] more directly reflects dielectric dissipation. CIP shows the smallest ε”, indicating negligible dielectric loss. In contrast, CIP@C exhibits a pronounced enhancement in ε” and presents a clear relaxation feature in the 9-13 GHz range, implying activation of polarization relaxation processes after C encapsulation. Notably, CIP@C@LM displays a further, substantial increase in ε”, with the most prominent relaxation peak located at 11-15 GHz and reaching a maximum value of 3.5. This enhanced dielectric dissipation is mainly associated with the α-GaOOH-rich nanorod shell, which increases the heterogeneous interfacial area and introduces polar Ga-O/OH environments, thereby promoting Maxwell-Wagner interfacial polarization^[63]. Under an alternating electromagnetic field, charge carriers accumulate at heterogeneous boundaries due to contrasts in conductivity and permittivity, leading to dynamic charge redistribution and polarization relaxation; the associated lag of polarization behind the field converts electromagnetic energy into Joule heating, thereby enhancing attenuation and suppressing reflection and secondary interference. To further deconvolute the dielectric-loss origins, ε” was separated into the conduction-loss term (ε_c”) and polarization-loss term (ε_p”) based on the Debye relaxation framework (method described in Supplementary Section 8). As shown in Figure 7C, all three samples exhibit relatively weak ε_c”, implying that long-range charge transport is not the dominant dissipation pathway. This observation aligns with the largely amorphous and discontinuous nature of the C layer, as well as the insulating effect of the α-GaOOH-rich shell, which further suppresses electrical conductivity. In contrast, the trend of ε_p” [Figure 7D] closely follows that of ε” for CIP@C and CIP@C@LM, indicating that polarization-dominated loss is the primary contributor. Here, interfacial polarization mainly arises from the effective dipole layers formed at boundaries with large contrasts in permittivity/conductivity. Meanwhile, oxygen-containing functional groups [Figure 2B] and O_v-related defect dipoles [Figure 2I] within the α-GaOOH/α-FeOOH-containing shell/interface introduce localized dipole moments and provide additional pathways for defect-induced polarization relaxation. Although the reduced ε’ leads to slightly weakened matching at low frequencies, it improves the impedance balance in the middle and high frequency absorption region. Together with the enhanced ε” and ε_p”, this result indicates that the α-GaOOH-rich shell enables favorable wave entry while retaining strong polarization-dominated dissipation.

To visualize relaxation behavior, Cole-Cole plots (ε” vs. ε’) were analyzed according to Debye theory, using the fitting relation^[64]:

(12) $$ \left(\varepsilon^{\prime}-\frac{\varepsilon_{s}+\varepsilon_{\infty}}{2}\right)^{2}+\left(\varepsilon^{\prime \prime}\right)^{2}=\left(\frac{\varepsilon_{s}-\varepsilon_{\infty}}{2}\right)^{2} $$

where ε_s and ε_∞ are the static and high-f limit permittivity, respectively. An ideal single Debye relaxation corresponds to one semicircle, deviations from a single semicircle generally indicate multiple relaxation processes. As shown in Figure 7E, CIP, CIP@C, and CIP@C@LM display approximately 2, 3, and 4 semicircles, respectively, suggesting progressively more complex relaxation behavior with increasing interfacial/defect complexity. The larger number and broader distribution of semicircles for CIP@C@LM indicate coexisting polarization processes and more developed interfacial polarization network enabled by the enriched surface dipoles and abundant heterogeneous interfaces. Finally, the dielectric loss tangent (tan δ_ε, see Figure 7F) provides a direct comparison with tan δ_μ [Figure 6E]. In the principal absorption frequency range of 10-16 GHz, tan δ_ε is substantially higher than tan δ_μ, indicating that dielectric loss dominates the electromagnetic attenuation of CIP@C@LM. This result is consistent with the enhanced polarization-loss component ε_p” and the increased number of Cole-Cole semicircles, confirming that abundant polarization-relaxation processes are primarily responsible for its improved EMW absorption performance. Magnetic loss, as reflected by the μ” and tan δ_μ responses, remains a secondary contribution, while optimized impedance matching further facilitates the broadband absorption of CIP@C@LM.

To substantiate the broadband and wide-angle EMW attenuation capability of CIP@C@LM composite, we further evaluated its stealth-related performance through radar cross-section (RCS) simulations^[65] [Figure 8]. Figure 8A presents the reference model constructed with a perfect electric conductor (PEC) plate, while Figure 8B shows the unmanned aerial vehicle (UAV) model coated with the as-prepared absorber, highlighting its potential for practical platform-level EMW suppression. In general, the RCS quantitatively represents the intensity of reflected EMW detected from a target under coating conditions; a smaller RCS corresponds to weaker backscattering and thus improved stealth performance. Figure 8C and D present the RCS simulations at a specific resonance f. The coated model achieves minimum RCS values (with respect to the PEC reference) of -18.18 dB m² at 11.21 GHz and -30.81 dB m² at 14.24 GHz, representing a significant reduction in radar backscattering. The 3D far-field scattering patterns at 11.21 GHz and 14.24 GHz [Figure 8E and F] further visualize this effect, where the scattering “lobe” area is markedly reduced for the coated model, confirming effective suppression of reflected waves at both f. Such a combined attenuation strategy suggests that this material can effectively regulate radio-wave absorption under complex operating conditions, supporting its promise for practical stealth-oriented EMW mitigation applications^[66].

Based on the above analysis, the EMW absorption mechanism in CIP@C@LM can be summarized as shown in Figure 9. First, the hierarchical nanorod/porous architecture extends the transmission path of incident waves via multiple scattering and reflection, thereby increasing the dissipation probability. Second, the intrinsic magnetic performance of the CIP core contributes to the magnetic loss by natural resonance. The reconstructed heterostructure helps induce the eddy-current dissipation by regulating the local electromagnetic environment. Furthermore, the minor interfacial AFM α-FeOOH introduces an additional magnetic component, thereby enriching the local magnetic structure and contributing to the effective magnetic response of CIP@C@LM. Third, dielectric loss is significantly enhanced by the synergistic effect of interfacial and dipolar polarizations. The multiple heterogeneous boundaries (CIP/C, C/α-GaOOH, α-FeOOH/α-GaOOH) induce strong Maxwell-Wagner relaxation, while the abundant defects (O_v) and functional groups within the oxyhydroxide shell serve as active dipolar centers, converting electromagnetic energy into heat. The integration of these loss channels enables the CIP@C@LM composite to achieve superior broadband absorption.

With the expanding deployment of EMW absorbing materials, their corrosion resistance and heat-dissipation capability under harsh environments have become increasingly critical. The formation of C and α-GaOOH on the CIP surface can act as an effective protective barrier, thereby improving the oxidation resistance and corrosion tolerance of CIP. Figure 10A and B schematically illustrate the electrochemical measurement configuration and the corresponding Tafel polarization curves obtained in 3.5 wt.% NaCl solution using glassy C working electrodes coated with CIP, CIP@C, and CIP@C@LM, respectively. Compared with pristine CIP, the polarization curves of CIP@C and CIP@C@LM shift toward more positive corrosion potentials (E_corr: -0.415 V → -0.319 V → -0.301 V), accompanied by a pronounced decrease in corrosion current density (I_corr: 3.16 × 10^-6 A → 1.99 × 10^-7 A → 1.28 ×10^-7 A). This trend indicates that both the C shell and the α-GaOOH shell promote the formation of a compact protective film, which suppresses charge transfer at the metal/electrolyte interface and thus retards the corrosion process. The fitted parameters derived from Tafel^[67] analyses are summarized in Supplementary Table 2. To complement the Tafel polarization analysis, electrochemical impedance spectroscopy (EIS) measurements were further performed. As shown in the Nyquist plot in Figure 10C, compared with CIP and CIP@C, CIP@C@LM exhibits a more pronounced impedance response. More importantly, the Bode impedance modulus plots in Figure 10D show that CIP@C@LM possesses a markedly higher |Z| value (impedance modulus) in the low-frequency region, indicating enhanced resistance to electrolyte penetration and interfacial charge transfer. Together with the positively shifted E_corr and reduced I_corr values obtained from the Tafel analysis, these EIS results further support the improved corrosion resistance of CIP@C@LM under the present testing conditions. The heat-dissipation capability of the coatings is compared in Figure 10E. All coatings were heated to an initial temperature of approximately 97 °C and the surface temperature evolution was tracked within 40 s. The CIP coating develops visible bulging during cooling, indicative of thermal-stress accumulation caused by interfacial mismatch and a relatively high effective thermal expansion. In contrast, CIP@C and CIP@C@LM coatings remain intact during the cooling process, indicating better coating integrity under the present test condition and more efficient heat dissipation. Quantitatively, after 10 s, the surface temperatures of CIP, CIP@C, and CIP@C@LM decrease to 90.9, 82, and 68.7 °C, respectively. After 40 s, the temperatures further drop to 64.2, 46.7, and 37.7 °C, demonstrating a substantially accelerated cooling rate for the coated architectures. These results suggest that the introduction of the C interlayer and the α-GaOOH shell reduces the interfacial thermal resistance and facilitates heat transport across the coating, thereby enhancing the heat-dissipation capability of the composite coating.