Direct-Write Printed Epoxy Composites with Layered Gradient Structure: Shape Memory and Electromagnetic Shielding Performance

Abstract

To address the growing problem of electromagnetic pollution, the development of intelligent, multifunctional electromagnetic shielding materials is essential. The objective of this work is to fabricate an intelligent, low-reflection and high-absorption electromagnetic shielding composite via direct ink writing. In this study, epoxy resin (EP) was employed as the matrix, with nickel powder (Ni), multi-walled carbon nanotubes (MWCNTs), and silver powder (Ag) serving as functional fillers. Direct-ink printing enabled the fabrication of uniformly structured composites and layered gradient-structured composites. By precisely varying the filler content through layer-by-layer printing, the gradient-structured composite exhibited an increasing electrical conductivity gradient and a decreasing magnetic permeability gradient along the direction of electromagnetic wave incidence. Comprehensive characterization of microstructure, electrical, magnetic, and dielectric properties, and electromagnetic shielding effectiveness revealed that the uniformly structured composites exhibited higher total shielding effectiveness (SE_T) and reflection coefficient (R) with increased electrical conductivity. The layered gradient-structured composite achieved an electrical conductivity of 5.44 S/m and an SE_T of 17.74 dB, with the R value reduced to 0.53. Compared to the highly conductive homogeneous composite used in the bottom layer (R = 0.87), this represents a reduction in reflectivity of approximately 39.1%, thereby mitigating secondary pollution from excessive reflection. Under a DC voltage of 200 V, all composites recovered their original shape within 63 s, with shape fixity (R_f) and recovery (R_r) ratios exceeding 92%. This strong shape memory capability supports conformal coating on complex devices and facilitates material recycling, offering a practical foundation for next-generation multifunctional electromagnetic shielding materials.

1. Introduction

The rapid evolution of communication technologies and the widespread use of electronic devices have led to a dense accumulation of information-carrying electromagnetic waves in the environment. This phenomenon not only causes electromagnetic interference (EMI) in communication systems but also poses numerous health risks, such as dizziness, insomnia, memory impairment, and palpitations [1,2]. Following water, air, and noise pollution, electromagnetic waves have emerged as a major pollution source. Consequently, there is an urgent need to develop shielding materials capable of effectively absorbing or reflecting electromagnetic waves when applied to electronic casings and communication cables.

Electromagnetic shielding materials are typically classified into metallic materials and conductive polymer composites. Metallic materials suffer from drawbacks such as high density and poor corrosion resistance, limiting their use in lightweight, durable applications. By contrast, polymers offer low density, high free volume, and long-term stability, making them attractive alternatives [3]. To meet the minimum shielding requirement, materials must achieve an electrical conductivity of at least 1 S/m [4]. Conventionally, conductivity enhancement in polymer composites relies on the incorporation of conductive fillers. However, simply increasing the filler content often raises costs and disrupts structural continuity. Therefore, research focus has shifted toward structural design strategies that maximize conductivity at lower filler loadings. Reported architectures primarily include homogeneous [5,6], segregated [7,8,9], layered [10,11,12,13], and porous structures [14,15,16].

Another critical challenge is the impedance mismatch between highly conductive shielding materials and air, which causes strong reflection of electromagnetic waves and secondary pollution [17]. Thus, merely increasing electrical conductivity is insufficient for high-performance shielding. Accordingly, recent research efforts have concentrated on designing conductive networks and tailored structures to balance absorption and reflection. For instance, Ma et al. [18] used the directional freezing method to fabricate three-dimensional (3D) magnetic MXene/Fe₃O₄@aMWCNT aerogels. Decorating Fe₃O₄ onto acidified multi-walled carbon nanotubes (aMWCNTs) enhanced interfacial polarization and magnetic loss at low filler loadings, yielding a reflection coefficient (R) of 0.31 and tunable shielding effectiveness (SE) between 24.5 and 51.6 dB. Xiong et al. [19] prepared a conductive TEMPO-cellulose nanofiber/carbon black/Vinasse activated carbon (TCA) composite via electrostatic adsorption, freeze-drying, and hot-pressing techniques. By constructing an inner conductive layer combined with an orderly stacked porous layer, a hierarchical conductive network was formed, which achieved an SE of 36.4 dB at small thickness and retained 35.1 dB after 1000 bending cycles. Voronin et al. [20] fabricated AgNW/PEDOT:PSS composite films (AgNW: silver nanowires) with a double-layer sandwich structure, enhancing SE to ~49.34 dB while maintaining a light transmittance of 66.33%. Moreover, the composite film demonstrated excellent long-term stability under corrosive conditions.

As an additive manufacturing technique, direct ink writing (DIW) offers advantages such as high material utilization and the ability to realize complex structural designs. It has been widely adopted in the fabrication of polymer-based electromagnetic interference (EMI) shielding composites, allowing more flexible and convenient structural design of materials [21,22]. For instance, Wang et al. [23] fabricated graphene nanoplate/unsaturated polyester resin composites via direct-ink-printing, where shear forces aligned graphene nanoplates, enhancing conductivity along the print direction to 69.9 S m⁻¹. The material also exhibited excellent thermal dissipation and electromagnetic shielding performance. Lv et al. [24] developed a PEDOT:PSS–EP composite coating using DIW by optimizing the ink’s rheology and printability. The coating achieved a high electrical conductivity of 41.50 S·cm⁻¹ and an adhesion strength of 99.3 kPa, demonstrating great potential for EMI shielding applications. However, most current studies employing DIW still focus primarily on uniform or simple layered structures, overlooking the unique capability of the technique to tailor complex architectures through layer-by-layer deposition. In this context, An et al. [25] utilized DIW to fabricate polydimethylsiloxane (PDMS)-based functional composites by alternately printing different layers, constructing sandwich structures such as ASA (Al₂O₃/SCF/Al₂O₃) and SAS (SCF/Al₂O₃/SCF). This approach enabled structural optimization of the shielding performance, yielding an EMI shielding effectiveness of up to 35.15 dB. Therefore, adopting DIW in this study is both necessary and important. It allows us to go beyond conventional uniform or simple sandwich designs. Instead, we can realize a multi-component, multifunctional composite with a gradient-layered structure.

Over the recent years, shape-memory polymers have been explored as matrices for intelligent, multifunctional electromagnetic shielding materials, incorporating carbon-based and metallic conductive fillers. These composites combine high SE with shape memory, self-healing, and tunable EMI shielding properties [26,27]. Li et al. [28] produced GNP/PLA Janus membranes by hot pressing, achieving a maximum EMI SE of 68 dB and enabling shape memory pre-deformation through a non-contact magnetic field. Jiang et al. [29] fabricated anisotropic GFF/EVA composites with X-band SE exceeding 40.8 dB axially but below 8.1 dB vertically. Simple rotation allowed tunable shielding, while the shape memory effect supported dynamic SE regulation. Zhao et al. [30] prepared a self-healing, magnetically responsive PVA/EGaInSn-Ni composite hydrogel via ultrasonic crosslinking, achieving a total SE (SE_T) of 65.8 dB, which improved to 75.2 dB after long-term storage, with rapid self-healing within 10 s. Despite these advances, reports on shape-memory polymers in electromagnetic shielding applications remain limited, and the fabrication processes are often complex.

In this study, the objectives are to develop a layered gradient-structured composite material with high shielding efficiency, low reflection, and excellent shape memory performance via direct ink writing technology. epoxy resin (EP), known for its shape memory properties, was selected as the matrix. Nickel (Ni), multi-walled carbon nanotubes (MWCNTs), and silver (Ag) were incorporated as functional fillers to respectively enhance magnetic loss, construct a conductive network, and improve electrical conductivity, while acetylene carbon black (ACB) served as a conductive thickening agent. Mechanical stirring, filtration, and vacuum filtration were employed to prepare an ink with excellent printability. uniformly structured composites were fabricated via 3D direct-write printing, and gradient filler distributions yielded gradient-structured composites with layered structures.

The effects of filler type and content on the layered gradient structure, microstructure, electrical conductivity, hysteresis loops, complex permittivity, complex permeability, and electromagnetic SE of the composites were systematically investigated using four-point probe measurements, scanning electron microscopy (SEM), energy dispersive spectrometry (EDS), vibrating sample magnetometry (VSM), and vector network analysis (VNA). Furthermore, the electrically responsive shape memory properties, including shape fixity ratio, recovery ratio, and recovery time, were evaluated under a DC voltage of 200 V. These analyses revealed the gradient-structured design established a positive electrical conductivity gradient and a reverse magnetic permeability gradient along the wave incidence direction, thereby reducing internal wave reflection. Under external heating, magnetic, or electric fields, the composites adhered to and reverted to their original shape upon re-stimulation, demonstrating recyclability and excellent shape memory capability.

By innovatively integrating direct ink writing with multi-layered gradient structural design, this study achieves a multiple dissipation mechanism of “absorption-reflection-reabsorption” for electromagnetic waves within the material. This approach maintains high shielding performance while significantly reducing the reflection coefficient. This strategy provides a novel pathway for intelligent, low-reflection shielding materials.

2. Experimental Material and Methods

2.1. Experimental Materials

Epoxy resin (EP51, molecular weight (Mw) = 185, T_g ≈ 120 °C), methyl hexahydrophthalic anhydride (MHHPA, Mw = 170), and N, N-dimethylbenzylamine (BDMA) were purchased from Shanghai McLean Biochemical Technology Co., Ltd. (Shanghai, China). Polypropylene glycol (PPG) was obtained from Guangzhou TaiRui New Materials Online Store (Guangzhou, China). Acetylene carbon black (ACB, average particle size ≈ 40 nm) was supplied by Suzhou Sheng’ernuo Technology Co., Ltd. (Suzhou, China). Nickel powder (Ni, average particle size ≈ 0.25 µm) and flake silver powder (Ag, average diameter ≈ 3-5 µm) were procured from Qinghe County Huiguang Metal (Xingtai, China). MWCNTs (length 10–30 µm) were purchased from Shanghai McLean Biochemical Technology Co., Ltd. (Shanghai, China).

2.2. Synthesis and Characterization

2.2.1. Preparation of Printing Ink

Table 1. Composition of EB/NM-A Composite Materials.

Ink	EP (g)	ACB (g)	Ni (g)	MWCNTs (g)	Ag (g)
EB-6N	20	2.4	1.2	0	0
EB-10N	20	2.4	2.0	0	0
EB-14N	20	2.4	2.8	0	0
EB-10N0.5M	20	2.4	2.0	0.1	0
EB-6N1M	20	2.4	1.2	0.2	0
EB-1.5M	20	2.4	0	0.3	0
EB-10A	20	2.4	0	0	2.0
EB-20A	20	2.4	0	0	4.0
EB-30A	20	2.4	0	0	6.0

presents the filler ratios for the EB/NM-A composite materials, where E, B, N, M and A denote EP, ACB, Ni, MWCNTs, and Ag, respectively; EB denotes the EP/ACB composite; and NM denotes the Ni/MWCNT hybrid filler. To construct an impedance-graded structure, Ni, MWCNTs, and Ag were incorporated into the EP/ACB (EB) matrix at controlled ratios. Ni was employed as a magnetic filler to provide magnetic loss. To investigate the effect of Ni content on magnetic loss capability, three samples were prepared: EB-6N, EB-10N, and EB-14N. MWCNTs, acting as one-dimensional conductive carbon fillers, were introduced to enhance electrical conductivity. To explore the synergistic effect between Ni and MWCNTs, two hybrid-filler samples, EB-10N0.5M and EB-6N1M, were fabricated. Additionally, EB-1.5M, containing only MWCNTs without metallic fillers, was prepared to evaluate the conductive contribution of MWCNTs alone. Ag, a two-dimensional flaky conductive metal filler, was added to substantially increase the electrical conductivity. To examine the influence of Ag content on conductivity and total shielding effectiveness (SE_T), samples denoted as EB-10A, EB-20A, and EB-30A were produced. The numbers in all sample labels indicate the mass fraction (wt%) of the respective filler relative to EP.

Figure 1 is the flowchart for the preparation of EB/NM-A printing ink. The preparation procedure was as follows: 20 g of EP base was weighed and placed into a stirring cup. Then, 0.6 g of PPG was added at a mass ratio of 100:3. The mixture was preheated in a drying oven to 78 °C until complete dissolution. MWCNTs (if required) were added and ultrasonicated for 10 min. Subsequently, 9.2 g of MHHPA and 0.2 g of BDMA were added sequentially at mass ratios of 185:85 and 100:1 relative to EP, respectively. Then, 2.4 g of ACB was incorporated and stirred for 60 min. Next, nickel powder (if required) was added and stirred for an additional 30 min. Finally, flake Ag powder (if required) was added and stirred for 30 min. The ink was filtered through an 80 μm filter cloth and vacuum-degassed for 10 min. The prepared ink was transferred to a syringe and stored below 0 °C.

2.2.2. Fabrication of Uniform and Gradient-Structured Composites

The printing inks prepared in Section 2.2.1 were deposited onto silicone substrates using a 3D printer (Architect^®Sparrow, Hangzhou GeNova Biotechnology Co., Ltd., Hangzhou, China). Printing was performed using a 0.4 mm nozzle with four deposited layers. These uniformly structured composites were designated as EB-6N, EB-10N, EB-14N, EB-10N0.5M, EB-6N1M, EB-1.5M, EB-10A, EB-20A, and EB-30A.

To construct a layered gradient composite, direct-write printing was employed to sequentially vary filler content layer by layer. This yielded the NMA composite material, containing Ni, MWCNTs, and Ag, with a four-layer gradient structure, which included top layer: high-Ni (EB-14N), second layer: Ni/MWCNT composite I (EB-10N0.5M), third layer: Ni/MWCNT composite II (EB-6N1M), and bottom layer: high-Ag (EB-20A).

Figure 2a illustrates a schematic of the NMA layered gradient structure, Figure 2b depicts the printing process, and Figure 2c displays the printed EB-6N1M and EB-20A layers (third and fourth layers, respectively). Elongated specimens of this gradient composite were fabricated for property testing, while U-shaped specimens were printed for assessing electrically responsive shape memory effect.

The printed samples were placed in a vacuum drying oven and subjected to the following curing procedure: heating to 90 °C and holding for 30 min, then holding at 120 °C for 1 h, followed by holding at 140 °C for 30 min, and finally holding at 160 °C for 2 h.

2.3. Experimental Characterization

The cross-sectional microstructure of the prepared samples was examined using SEM (SU70 microscope, Hitachi, Naka, Japan). Elemental distribution was analyzed using EDS (EDS-EBSD, AMETEK, Berwyn, USA). Hardness and load–displacement curves were obtained using a nanoindenter (TI980, Hysitron, Karlsruhe, Germany) on polished specimens (20 × 20 × 1.6 mm³) with a 2 × 2 indentation matrix.

Hysteresis loops were measured at room temperature using VSM (EZ-7, Microsense, Lowell, USA) on samples (4 × 4 × 1.6 mm³) under an applied field of −20,000 to 20,000 Oe.

Electrical resistance was measured using a four-point probe tester (DMM6500, Keithley, Cleveland, OH, USA) on bar specimens (20 × 1.6 × 5 mm³). Five replicate samples were prepared, and each underwent five replicate tests. The mean and standard deviation (SDs) are calculated.

Volume conductivity (σ) was calculated using Equation (1) [31]:

$$\mathsf{\sigma } ={\displaystyle \frac{L}{SR}}$$

(1)

where σ is the electrical conductivity (S/m), L is the sample length (m), S is the cross-sectional area (m²), and R is the measured resistance (Ω).

Complex magnetic permeability, electromagnetic SE, and complex permittivity were measured in the X-band (8.0–12.0 GHz) using a vector network analyzer (VNA N5234A, Keysight Technologies, Santa Rosa, CA, USA) via the coaxial transmission method on samples (22.8 × 10.1 × 1.6 mm³). For NMA composites, the EB-14N layer was set as the incident side. Three replicate samples were prepared and each was tested once. The measurement result closest to the mean of the three was selected.

The complex magnetic permeability was derived from the following expression [32]:

$$\mathsf{\mu }={\mathsf{\mu }}^{\prime }-j{\mathsf{\mu }}^{″}={\mathsf{\mu }}_0{\mathsf{\mu }}_{\mathrm{r}}\left(1-j\tan {\mathsf{\delta }}_{\mathsf{\mu }}\right),$$

(2)

where ${\mathsf{\mu }}_0$ is the permeability of free space; ${\mathsf{\mu }}^{\prime }$ is the real part of the complex permeability, denotes the ability of the magnetic medium to store magnetic field energy; ${\mathsf{\mu }}^{″}$ is the imaginary part, denotes the magnetic hysteresis loss and eddy-current loss of the material; $\tan {\mathsf{\delta }}_{\mathsf{\mu }}$ is the magnetic loss tangent, defined as the ratio of lost magnetic energy to stored magnetic energy; and ${\mathsf{\mu }}_{\mathrm{r}}$ is the relative permeability.

The SE parameters were calculated from S-parameters as follows [33]:

$${SE}_{\mathrm{T}}={SE}_{\mathrm{A}}+{SE}_{\mathrm{R}}+{SE}_{\mathrm{M}},$$

(3)

$$T={\left|S_{21}\right|}^2$$

(4)

$$R={\left|S_{11}\right|}^2$$

(5)

$$1=R+T+A$$

(6)

When the total shielding effectiveness (SE_T) exceeds 10 dB, ${SE}_{\mathrm{M}}$ is generally ignored for multiple reflection losses. In this case:

$${SE}_{\mathrm{T}}={SE}_{\mathrm{A}}+{SE}_{\mathrm{R}}$$

(7)

$${SE}_{\mathrm{R}}=10\lg \left(1-R\right)$$

(8)

$${SE}_{\mathrm{A}}=10\lg \left({\displaystyle \frac{T}{1-R}}\right)$$

(9)

$${SE}_{\mathrm{T}}=10\lg T,$$

(10)

where SE_T is the total shielding effectiveness, SE_R is the reflection shielding effectiveness, SE_A is the absorption shielding effectiveness, and R is the reflection coefficient, T is the transmission coefficient, and A is the absorption coefficient.

The complex permittivity was calculated from the following formula [34]:

$$\mathsf{\epsilon }={\mathsf{\epsilon }}^{\prime }-j\mathsf{\epsilon }”={\mathsf{\epsilon }}_0{\mathsf{\epsilon }}_{\mathrm{r}}\left(1-j\tan {\mathsf{\delta }}_{\mathsf{\epsilon }}\right),$$

(11)

where ${\mathsf{\epsilon }}_0$ is the absolute permittivity of free space, ${\mathsf{\epsilon }}^{\prime }$ is the real part of the complex permittivity, denotes the charge-storage capability of the material; $\mathsf{\epsilon }”$ is the imaginary part, denotes the dielectric loss factor;$\tan {\mathsf{\delta }}_{\mathsf{\epsilon }}$ is the loss tangent, defined as the ratio of energy lost to energy stored per cycle; and ${\mathsf{\epsilon }}_{\mathrm{r} }$ is the relative complex permittivity.

For shape memory testing, a U-shaped specimen (Figure 3a) was heated, bent to 90°, and cooled to fix the temporary shape. Three replicate samples were prepared and each was tested once. The measurement result closest to the mean of the three was selected.

The shape fixity ratio ($R_{\mathrm{f}}$) was calculated using Equation (8). Electrical stimulation was applied using a 200 V DC power supply, and the shape recovery ratio ($R_{\mathrm{r}}$) was determined using Equation (9) [31]:

$$R_{\mathrm{f}} = {\displaystyle \frac{{180-\mathsf{\theta }}_1}{90}}\times 100\%,$$

(12)

$$R_{\mathrm{r}}={\displaystyle \frac{{\mathsf{\theta }}_2-{\mathsf{\theta }}_1}{90}}\times 100\%,$$

(13)

where ${\mathsf{\theta }}_1$ is the angle formed between the two sides of the U-shaped specimen after cooling and removal of external force, and ${\mathsf{\theta }}_2$ is the angle after shape recovery (Figure 3b).

3. Results and Discussion

3.1. Microstructure

Figure 4 illustrates the fracture surfaces of uniformly structured EB/NM-A composites with different filler configurations. In EB-14N (Figure 4a), which contains only zero-dimensional fillers, the distribution is relatively uniform, but the large inter-particle spacing prevents the formation of continuous conductive pathways. Incorporation of MWCNTs (one-dimensional fillers) in EB-10N0.5M (Figure 4b) and EB-6N1M (Figure 4c) bridges the zero-dimensional fillers (Ni), synergistically constructing a 3D conductive network. Among these, EB-6N1M exhibits enhanced connectivity within its internal conductive network. In contrast, when a high content of MWCNTs alone was introduced, the filler spacing in EB-1.5M (Figure 4e) became smaller than that in EB-14N, but noticeable agglomeration occurred, making uniform dispersion more challenging. In EB-20A (Figure 4d), flake-shaped Ag particles are uniformly dispersed across the matrix, contributing to enhanced conductivity.

Figure 5 shows the microstructural characterization of the NMA composite with layered gradient structure. Figure 5a presents a schematic of the NMA structure. The SEM images of the interfaces between different filler layers (Figure 5b–d) and the overall cross-section (Figure 5e) show no distinct boundaries between layers. This is because the entire four-layer structure was cured together, allowing the epoxy matrices to cross-link simultaneously. Polymer chains interdiffused and entangled at the interfaces, creating a seamless network without visible defects. Elemental mapping (EDS) confirms the layered design. In Figure 5f,g Ag is concentrated in the bottom layer, while Ni is distributed across the upper three layers, with concentrations decreasing from top to bottom, exhibiting a clear layered structure. In summary, SEM confirms the NMA composite’s structural integrity, and EDS verifies its gradient composition. This demonstrates that the gradient distribution of fillers within the EP was successfully achieved through layer-by-layer direct printing.

3.2. Mechanical Properties

Figure 6 shows the load–displacement curves of EB/NM-A composites with different uniform filler structures. The curves exhibit three distinct phases: (i) the loading phase (ascending portion), (ii) the load-holding phase (horizontal plateau at the peak), and (iii) the unloading phase (descending portion). It can be seen that the hardness of the composites is primarily governed by Ni content rather than MWCNTs or Ag. As the Ni content decreases, the curves shift rightward, indicating reduced hardness.

The straight-line segment length during the load-holding phase reflects indentation hardness. EB-14N shows the shortest length and the highest hardness, attributed to Ni particles acting as hard metallic reinforcements that effectively resist indenter’s penetration. By contrast, EB-20A (with Ag as metallic filler) exhibits an indentation hardness of 0.19 GPa, lower than that of EB-14N and EB-10N0.5M, likely due to differences in Ag particle size and volume fraction.

Overall, EB-14N demonstrates superior hardness and is recommended as the surface layer for NMA gradient composites to minimize surface damage when encapsulating electronic devices.

3.3. Electrical Properties

To establish an increasing gradient conductivity, MWCNTs were introduced at concentrations of 0.5 wt.% and 1 wt.% into EB-10N and EB-6N, yielding EB-10N0.5M and EB-6N1M as absorber layers, respectively. To enhance the conductivity of the reflective layer, MWCNTs were added at 1.5 wt.% to an EB base containing 12% ACB, yielding the EB-1.5M sample. Subsequently, Ag powder was incorporated at 10, 20, and 30 wt.% into the same EB base, resulting in EB-10A, EB-20A, and EB-30A reflective layers.

Figure 7a shows the electrical conductivity of the absorber layers. Without MWCNTs, the electrical conductivity does not increase significantly with increasing Ni content. The conductivity of EB-6N is 0.33 S/m, while EB-14N shows a slight improvement of 0.32 S/m, which is attributed to the discontinuous conductive pathways formed by zero-dimensional Ni and ACB fillers. The incorporation of MWCNTs markedly enhances conductivity, with that of EB-10N0.5M and EB-6N1M reaching 2.28 S/m and 6.18 S/m, respectively. The one-dimensional MWCNTs synergistically bridge Ni and ACB, forming an effective 3D conductive network, significantly enhancing the electrical conductivity of the composite materials [35]. Furthermore, the length of MWCNTs (20-30 μm) is significantly smaller than the printing nozzle diameter (410 μm), ensuring uniform distribution without anisotropy along the printing path.

Figure 7b shows the electrical conductivity of the reflector layers and NMA. EB-1.5M exhibits a conductivity of 6.80 S/m, only marginally higher than that of EB-6N1M, likely due to the limited interconnection efficiency between one-dimensional and zero-dimensional fillers. By contrast, two-dimensional flake-shaped Ag powder as conductive filler significantly improves conductivity: EB 10A, EB-20A, and EB-30A exhibit electrical conductivities of 7.37, 15.80, and 17.27 S/m, respectively. The modest increase from EB 20A to EB-30A (1.47 S/m) indicates supersaturation of Ag content, with conductive pathways approaching completion. Considering the balance between performance and cost, further increasing the Ag content to 30 wt% (EB-30A) yields only a marginal improvement in conductivity while significantly raising both material cost and weight, thereby limiting its potential for weight-sensitive applications. Therefore, EB-20A was selected as the optimal reflective layer to balance conductivity and minimize electromagnetic wave penetration.

The forward-guided NMA layered composite exhibits an electrical conductivity of 5.44 S/m. Due to its multilayer architecture, the resistance in all layers except the EB-20A reflective layer is higher than that of the single EB-20A layer, resulting in reduced overall conductivity compared to the standalone reflective layer.

3.4. Magnetic Properties

Figure 8a shows the hysteresis loops of the EB/NM-A composites. As the Ni content decreases, the magnetic saturation strength drops from 0.335 emu/g in EB-14N to 0.168 emu/g in EB-6N1M, confirming that NMA composites exhibit an inverse magnetic permeability gradient. The hysteresis loops of EB-14N, EB-10N0.5M, and EB-6N1M display negligible residual magnetism or coercivity, indicating that Ni powder behaves as a paramagnetic filler. This characteristic facilitates the conversion and dissipation of electromagnetic wave energy into thermal energy within the composite.

The magnetic loss power density (P), defined as the energy dissipated per unit time and volume in an alternating magnetic field, is expressed as follows:

$$P=\pi f{\mathsf{\mu }}_0{\mathsf{\mu }}^{″}{H}_{\mathrm{m}},$$

(14)

where f is the frequency of the applied alternating magnetic field, μ₀ is the vacuum permeability, and H_m is the amplitude of the applied field.

Figure 8b–d present the frequency-dependent curves of the real (μ′) and imaginary parts (μ″) of magnetic permeability and the magnetic loss tangent (tanδ_μ), respectively. As shown in Figure 8b-d, μ′ of EB-14N decreases noticeably over the X-band, while μ″ remains at a relatively high level between 0.12 and 0.16. This indicates that EB-14N shifts from magnetic energy storage to magnetic energy dissipation at higher frequencies, a characteristic that thereby favors impedance matching between air and the material. Therefore, placing EB-14N as the surface layer promotes electromagnetic wave entry and enhances absorption loss.

For the uniformly structured EB/NM-A composites, both μ′ and μ″ generally increase with higher Ni content. Accordingly, EB-10N0.5M and EB-6N1M were respectively selected as the second and third layers, to construct an decreasing magnetic permeability gradient structure. Thus, the magnetic loss is greatest at the surface layer of NMA and decreases with depth. The resulting NMA exhibits μ″ and tanδ_μ values in the ranges of 0.08–0.14 and 0.06–0.12 across the X-band, which are comparable to those of EB-10N0.5M and remain close to the high-magnetic-loss EB-14N (μ″: 0.12–0.16; tanδ_μ: 0.08–0.12). Notably, around 10 GHz, the μ″ of NMA slightly surpasses that of EB-14N. These results confirm that the decreasing magnetic permeability gradient design enables the composite to achieve effective magnetic loss performance within the X-band.

Additionally, at higher frequencies, fluctuations appear due to magnetic dispersion during magnetization relaxation under an external magnetic field. The experimental frequency range (8.0–12.0 GHz, X-band) lies near the ultra-high frequency domain (108–1010 Hz), where magnetic dispersion is primarily governed by natural resonance [36,37].

3.5. Electromagnetic Shielding Effectiveness

NMA composites exhibit positive magnetic permeability gradients in one direction and negative gradients in the opposite direction. To validate the electromagnetic shielding mechanism of this gradient design, their SE was further investigated. For homogeneous composites, the electrical conductivity and thermal conductivity are positively correlated, as described by Equations (15) and (16) [38]:

$$S{E}_{\mathrm{A}}=131.43t\sqrt{f\mathsf{\mu }\mathsf{\sigma }},$$

(15)

$$S{E}_{\mathrm{R}}=168.2+10\lg {\displaystyle \frac{\mathsf{\sigma }}{f\mathsf{\mu }}},$$

(16)

where t is the sample thickness, f is the frequency, μ is the magnetic permeability, and σ is the electrical conductivity.

Figure 9 illustrates the electromagnetic shielding performance of the EB/NM-A composites. The curves of SE_T, SE_R, SE_A and R over 8–12 GHz for both NMA and EB/NM-A composites are presented in Figure 9a–d. For EB/NM-A materials, SE_T, SE_R, SE_A and R all increase with rising electrical conductivity. As shown in Figure 9e,f at 8 GHz, EB-14N exhibits an SE_A of only 5.78 dB, an SE_R of only 2.86 dB, with an R value of 0.48, while EB-20A achieves an SE_T of 24.92 dB, and an R value of 0.87. With SE_A > SE_R and a relatively low R value, EB-14N shows an absorption-dominated shielding mechanism. However, due to its low conductivity (0.65 S/m), its total shielding effectiveness (SE_T = 8.64 dB) is insufficient for high-performance shielding requirements. After introducing MWCNTs to build a more complete three-dimensional conductive network, the conductivities of EB-10N0.5M and EB-6N1M increase to 2.28 S/m and 6.18 S/m, respectively. Their SE_T values rise accordingly, while R remains in the range of 0.6–0.7, indicating a gradual transition to reflection-dominated shielding. The EB-20A sample, containing a high Ag content, achieves a markedly higher conductivity (15.80 S/m) and total shielding effectiveness (SE_T = 24.92 dB). However, SE_R and R increase significantly (SE_R = 9.19 dB, R = 0.87). This is because the extremely high conductivity induces severe impedance mismatch, which completely shifts the shielding mechanism to reflection dominance and leads to secondary electromagnetic reflection pollution.

For the layered gradient NMA composite, although its bottom layer contains the highly conductive EB-20A, the total SET (17.74 dB) is still lower than that of a single EB-20A layer (24.92 dB). This is not a performance deficiency, but rather a deliberate outcome of the design aimed at high shielding efficiency with low reflection. By constructing an increasing electrical conductivity gradient and a decreasing magnetic permeability gradient, the transition zones formed by the surface and intermediate layers (EB-14N, EB-10N0.5M, EB-6N1M) effectively improve the impedance matching at the air–material interface, creating an impedance-graded structure.

Moreover, the reflection coefficient (R) of NMA exhibits clear frequency dependence (Figure 9d), showing an overall descending trend across the X-band. When the frequency exceeds approximately 9 GHz, the R value of NMA falls below 0.5, further confirming the effectiveness of the layered gradient structure in achieving low-reflection electromagnetic shielding in this frequency range. In summary, this design achieves a balance between an absorption-dominated shielding mechanism and high shielding performance.

Figure 10 illustrates the electromagnetic shielding mechanism of NMA composites. When electromagnetic waves interact with the material, the surface layer—characterized by low electrical conductivity and high magnetic loss—allows wave penetration while promoting extensive absorption. As the waves propagate inward, the conductivity increases progressively across layers. Upon reaching the highly conductive bottom layer, strong reflection occurs, followed by reabsorption as the waves pass through the high magnetic loss layer.

This repeated “absorption–reflection–reabsorption” process of electromagnetic waves between layers enhances overall absorption loss, thereby increasing SE_A and SE_T, while simultaneously reducing wave reflection (SE_R) and minimizing secondary electromagnetic pollution. In summary, direct-write printing technology enables precise control of filler gradients, allowing the conductivity and magnetic loss to be tailored layer by layer. This design approach mitigates the high reflectivity issues associated with high conductivity and overcomes the limitations of low conductivity, such as elevated transmittance and poor shielding performance.

3.6. Dielectric Properties

The dielectric performance of the composite, which governs its polarization and loss response to electromagnetic waves, is thus closely related to its shielding effectiveness. Figure 11a–c present the frequency-dependent curves of the real part of complex permittivity (ε′), imaginary part of complex permittivity (ε″), and dielectric loss factor (tanδ_ε) for EB/NM-A composites, respectively. According to Equation (17), the imaginary part ε″ consists mainly of conductive loss and polarization–relaxation loss, where the former is proportional to electrical conductivity and the latter arises from dipole reorientation and interfacial charge relaxation [39].

$${\mathsf{\epsilon }}^{″}={\displaystyle \frac{\mathsf{\sigma }}{2\pi f{\mathsf{\epsilon }}_0}}+{\mathsf{\epsilon }}_{relax}^{″},$$

(17)

where σ represents the electrical conductivity, ${\mathsf{\epsilon }}_0$ is the permittivity of free space, and ${\mathsf{\epsilon }}_{relax}^{″}$ is the polarization relaxation loss.

As shown in Figure 11a,b both ε′ and ε″ increase with electrical conductivity. This is because higher conductivity enhanced migration of free charge carriers, which facilitates electronic polarization and thereby increases the polarization strength. For the uniform EB/NM-A composites, the increase in conductivity is the primary reason for the growth in ε″, contributing dominantly to conductive loss. In contrast, for the layered gradient NMA composite, the increase in ε″ benefits not only from its conductivity but also from its multi-layer design, which introduces abundant heterogeneous interfaces and enhances polarization-relaxation loss.

Furthermore, as seen in Figure 11c, although the conductivity of NMA is much lower than that of EB-20A, their dielectric loss tangent (tanδ_ε) values are quite close, differing by only about 0.05. This confirms that the layered gradient structure of NMA strongly excites interfacial polarization. The multiple interfaces induce strong interfacial polarization, which enhances the dissipation of electromagnetic wave energy. The dissipation intensity is determined by the number of interfaces, highlighting the role of structural design in improving dielectric loss and overall shielding performance.

3.7. Electro-Responsive Shape Memory Properties

Figure 12a,b illustrate the electro-responsive recovery behavior of EB-14N and the shape memory performance of EB/NM-A composites. Under a 200 V DC voltage, the pre-deformed U-shaped specimen gradually returns to its original shape during energization. This electro-responsive shape memory effect is governed by a thermal mechanism: electrical energy is converted into Joule heat, and once the composite reaches its glass transition temperature (T_g), shape recovery is initiated [40,41].

As shown in Figure 12b, EB-14N requires 63 s for complete recovery, whereas EB-20A reduces the recovery time to 32 s. This indicates that higher electrical conductivity lowers resistance, accelerates Joule heating, and significantly improves shape recovery ratio. For the layered gradient NMA composite, the recovery time is 44 s, lying between those of EB-14N and EB-20A, reflecting the integrated electro-responsive behavior across its layers.

In terms of shape fixity ratio, NMA achieves a value of 94.6%, which is comparable to that of EB-20A (92.8%), indicating satisfactory temporary-shape retention. The shape recovery ratio of NMA is 95.8%, slightly lower than those of EB-14N (97.3%) and EB-20A (98.2%). This minor decrease may stem from internal constraint stresses caused by asynchronous recovery between layers, which partially counteracts the driving stress for shape recovery. Nevertheless, recovery ratios remain above 95%, confirming reliable shape-recovery capability.

Overall, both the shape fixity and shape recovery ratios of all composites exceed 92%, demonstrating excellent shape-memory performance. These results highlight the suitability of EB/NM-A composites for conformal encapsulation of complex devices, while their recyclability supports sustainable applications [42].

4. Conclusions

Overall, this study establishes a gradient-structured, multifunctional shielding composite that combines strong EMI absorption, reduced reflection, and robust shape memory, offering a practical route for next-generation smart and sustainable electromagnetic shielding materials. The main conclusions can be summarized as follows:

• Material preparation and structure: Using epoxy as the matrix, ACB as a conductive thickener, and Ni, MWCNTs, and Ag as functional fillers, a layered gradient-structured NMA electromagnetic shielding composite was successfully prepared by direct ink writing. SEM and EDS analyses confirmed uniform filler dispersion within the matrix, gradient distribution across the NMA layers, forming an layered gradient-structured with strong interlayer bonding.
• Electrical, Magnetic, and Dielectric Properties: The layered gradient architecture of NMA effectively integrates the electrical, magnetic, and dielectric properties across its layers. It exhibits an increasing electrical conductivity gradient (from 0.65 S/m at the surface to 5.80 S/m at the bottom, with an overall value of 5.44 S/m) and a decreasing magnetic loss gradient (tan δ_μ ranges from 0.08–0.12 to 0.01–0.07 through the thickness, with an overall range of 0.06–0.12). Furthermore, the multiple heterogeneous interfaces introduced by the layered structure enhance interfacial polarization, leading to significantly improved dielectric loss.
• Electromagnetic shielding performance: The layered NMA composite achieved superior shielding performance, with SE_T = 17.74 dB and R = 0.53. Its unique dual-gradient structure enables an “absorption–reflection–reabsorption” shielding mechanism, effectively reducing secondary electromagnetic pollution compared to conventional single-layer shielding materials.
• Shape memory performance: The NMA composite demonstrates favorable electro-responsive shape-memory properties, achieving shape-fixity and shape-recovery ratios of 94.6% and 95.8%, respectively. Under a DC voltage of 200 V, it recovers its original shape within 44 s, highlighting its strong potential for conformal encapsulation of complex devices and recyclable applications.