Optimization of Magnetic Filler Loading and Interstitial Dielectric Percolation for Tunable Triboelectric–Electromagnetic Hybrid Generators

Abstract

In this study, a material-driven strategy is presented to realize tunable triboelectric–electromagnetic hybrid generators while overcoming the form-factor limitations of conventional magnet-assisted systems. A magneto-dielectric hybrid generator (MDHG) was constructed using a soft magnetized dielectric composite, where NdFeB microparticles were embedded in an Ecoflex matrix and activated by pulse magnetization, allowing a single compliant layer to operate simultaneously as a triboelectric contact medium and a magnetic flux source coupled to a coil. The magnetic filler loading was systematically optimized to elucidate the trade-off between enhanced electromagnetic induction and a non-monotonic triboelectric response governed by dielectric polarization, surface potential, and interfacial energetics. To selectively strengthen the triboelectric branch without sacrificing electromagnetic output, nanoscale BaTiO₃ was introduced as an interstitial dielectric phase to promote polarization-active pathways and suppress screening-driven charge-utilization loss. Under contact–separation operation, the optimized MDHG produced triboelectric outputs up to a V_OC of 400.40 V and I_SC of 56.95 μA, while the electromagnetic branch delivered up to a V_OC of 260.04 mV and I_SC of 0.89 mA, corresponding to 2.87- and 2.62-fold increases in triboelectric V_OC and I_SC over pristine Ecoflex. Finally, the hybrid signatures enabled a wearable smart-skin interface capable of decoupling touch occurrence, intensity, and counter-material identity.

1. Introduction

The rapid evolution of next-generation wearable electronics and Internet of Things (IoT) systems is driving a fundamental paradigm shift in device architecture, where the power supply is no longer a peripheral component but a primary design constraint determining system viability [1,2]. As on-body technologies move toward continuous operation across diverse environments, power solutions must remain reliable under variable mechanical conditions and frequent user-driven motion [3,4]. In these scenarios, minimized thickness, lightweight form factors, and mechanical compliance are as essential as the available energy, because seamless integration with curvilinear and dynamically moving human surfaces requires both electrical stability and mechanical adaptability [5,6]. Conventional electrochemical batteries impose rigid form factors and recurring maintenance burdens, and they raise environmental concerns related to disposal and operational safety in wearable and distributed deployments [7,8]. These limitations have accelerated interest in energy-harvesting technologies that can replace or augment batteries by scavenging ubiquitous ambient energy spanning biomechanical kinetics and environmental vibrations [9,10,11]. Mechanical energy harvesting is particularly aligned with wearables because the source is naturally co-located with the user and accessible across daily activities without dedicated infrastructure [12,13]. Furthermore, such systems can extend beyond power generation by enabling self-powered sensing, where the transduced electrical signals function as intrinsic information carriers for real-time perception and human–machine interaction without auxiliary power sources [14,15]. However, the widespread deployment of energy-autonomous wearable and IoT ecosystems remains constrained by practical challenges, particularly robust power management under intermittent energy inputs and the long-term operational reliability of harvesters in dynamic real-world environments [16,17,18]. These deployment-driven constraints motivate harvesting platforms that can deliver usable electrical outputs under irregular mechanical stimuli and provide a wider operating margin for energy management and sensing.

Among various harvesting mechanisms, triboelectric nanogenerators (TENGs) operate on the coupling of contact electrification and electrostatic induction and have become a representative platform for self-powered systems due to their material versatility and structural adaptability [19,20,21]. TENGs can be realized using a wide selection of polymers, metals, and composites, which supports device customization for skin-conformal electronics and soft form factors [22,23]. Governed by Maxwell’s displacement current, TENGs can generate high-voltage outputs under low-frequency and irregular mechanical excitations, which is advantageous for triggering high-threshold readout circuits and forming distinct signal patterns in sensing scenarios [24,25]. These characteristics have enabled innovations in wearable power modules, tactile interfaces, and interactive smart skins, where electrical signatures can be correlated with contact events and mechanical interactions [26,27]. Moreover, triboelectric charge generation can be sensitive to the electronic affinity and surface states of the counter material, opening avenues for interaction-aware sensing beyond simple energy harvesting [28,29]. Despite these advantages, TENGs behave as high-impedance capacitive sources, and practical operation often suffers from mismatch with low-impedance electronics and energy storage elements, limiting charge transfer efficiency and utilizable power for current-intensive loads [30,31]. This limitation becomes particularly important in realistic energy management, where rectification, storage charging, and load interfacing typically favor lower source impedance and higher deliverable current [32,33].

To mitigate these electrical constraints, hybrid energy-harvesting architectures have been explored by integrating multiple transduction mechanisms within a single platform to improve practical power usability and broaden functional bandwidth [34,35]. In the context of triboelectric systems, hybridization has been pursued with piezoelectric, thermoelectric, pyroelectric, and photovoltaic mechanisms, where complementary source characteristics and multi-domain responsiveness can improve energy delivery and extend functionality under shared stimuli [36,37,38]. Among these strategies, coupling with electromagnetic generators (EMGs) is particularly compelling because EMGs, based on Faraday’s law of induction, typically provide current-dominant outputs with relatively low internal impedance [39,40]. Accordingly, triboelectric–electromagnetic hybridization is not only an energy strategy but also a system-level route to multifunctional self-powered transduction.

Despite these synergistic advantages, the structural implementation of TENG–EMG hybrids often faces a critical compliance mismatch due to the conventional reliance on rigid bulk permanent magnets to generate sufficient magnetic flux [41,42]. Integrating such heavy and stiff inclusions into soft wearable systems increases device volume and introduces geometric constraints, and it can induce mechanical stress concentrations under repeated deformation that compromise long-term durability and skin-conformal integration. To circumvent these issues, flexible magnetic composites formed by dispersing magnetic powders in soft polymer matrices offer a route to internalize magnetic functionality within compliant layers and reduce reliance on bulky external magnets [43]. However, optimizing a single composite layer for simultaneous triboelectric and magnetic operation introduces competing physical constraints. Limited magnetic filler loading can preserve dielectric strength and surface charge retention for triboelectric operation, but it restricts remanent magnetization and yields insufficient electromagnetic output [44]. Conversely, higher magnetic loading can intensify electrostatic screening and interfacial charge redistribution, reducing effective triboelectric surface potential and degrading charge utilization at the contact interface [45]. Therefore, developing a material-driven strategy that maintains robust triboelectric performance even within high-loading magnetic composites remains a critical engineering challenge for fully flexible hybrid generators.

In this study, a material-centric solution is presented through a bulk-magnet-free, tunable hybrid platform established on an Ecoflex composite layer that embeds high-loading Nd₂Fe₁₄B (NdFeB) microparticles as the functional magnetic component. Through an in situ pulse magnetization process, the composite film is activated to function as an intrinsic remanent flux source coupled to a coil, enabling a compact single-layer architecture that simultaneously serves as the triboelectric contact medium and the electromagnetic induction source. While increasing the magnetic filler content is essential for enhancing electromagnetic output, the results reveal a trade-off where excessive conductive loading promotes electrostatic screening and charge leakage, thereby constraining the triboelectric potential. To resolve this dichotomy, nanoscale BaTiO₃ (BTO) is introduced as an interstitial dielectric phase to construct a hierarchical dual-filler microstructure. This interstitial dielectric percolation strategy fills the void spaces between magnetic microparticles, reinforcing polarization-active pathways while suppressing leakage formation. This microstructural engineering enables selective amplification of triboelectric output without sacrificing the remanent magnetic flux required for EMG, ensuring balanced high-performance harvesting. Finally, the optimized hybrid outputs are leveraged to demonstrate a wearable smart-skin interface capable of multimodal tactile perception. By analyzing the complementary electrical signatures from the two mechanisms, the system decouples complex mechanical stimuli, enabling simultaneous recognition of touch occurrence, impact intensity, and counter-material identity for advanced human–machine interaction. Overall, this work provides a scalable material design route for composition-tunable triboelectric–electromagnetic hybrids that improve practical energy delivery while enabling multimodal self-powered tactile sensing.

2. Materials and Methods

2.1. Materials

A two-part silicone elastomer (Ecoflex 00-30, Smooth-On, Inc., Macungie, PA, USA) was used as the polymer matrix. NdFeB micromagnets (MQFP-B-20076-089, NEO Performance Materials Korea Inc., Seoul, Republic of Korea) and barium titanate powder (BaTiO₃, BTO, Sigma-Aldrich, St. Louis, MO, USA) were employed as functional fillers. Polylactic acid (PLA) filament (Bambu Lab, Shenzhen, China) was used to fabricate casting molds by 3D printing. Copper (Cu) wire with a diameter of 0.1 mm was used to wind the coil for electromagnetic energy harvesting. Aluminum (Al) electrodes were prepared from commercially available Al tape. All materials were used as received without further purification.

2.2. Fabrication of the Magneto-Dielectric Hybrid Generator (MDHG)

To prepare the composite layer, BTO powder was first added to Ecoflex Part B and mechanically stirred for 10 min. NdFeB powder was then introduced and mixed for an additional 10 min. Subsequently, Ecoflex Part A was added, and the mixture was further stirred for 10 min to obtain a homogeneous mixture. The filler contents were adjusted according to the designed compositions of each sample. A casting mold (target thickness: 1 mm) was fabricated using a 3D printer (Bambu Lab P1S, Bambu Lab, China) with PLA filament. The prepared mixture was poured into the mold and cured in a convection oven at 60 °C for 3 h. After curing, the film was demolded to obtain the Ecoflex/NdFeB/BTO composite layer. For magnetic activation, the composite film was magnetized using a pulse magnetizer (ST-2550, MTS, Eden Prairie, MN, USA), yielding a soft magneto-dielectric composite layer (SMDC). The SMDC was then laminated onto a bottom Al electrode. Finally, a top module consisting of an Al electrode integrated with a Cu coil layer was aligned and assembled with the SMDC to form the magneto-dielectric hybrid generator (MDHG), enabling simultaneous triboelectric and electromagnetic energy harvesting under continuous mechanical operation.

2.3. Material Characterization

The morphology of NdFeB and BTO, and the cross-sectional morphology of SMDC were examined using high-resolution field-emission scanning electron microscopy (HR FE-SEM, Carl Zeiss, Oberkochen, Germany). Elemental distribution and composition were analyzed by energy-dispersive X-ray spectroscopy (EDS, Oxford Instruments, Abingdon, UK). Crystalline phases of the functional fillers and SMDC were identified using X-ray diffraction (XRD, Bruker, Oberkochen, Germany). Magnetic characteristics of the SMDC films, including M–H hysteresis loops and magnetic field, were measured using a vibrating sample magnetometer (VSM, Lake Shore Cryotronics, Westerville, OH, USA) and a gauss meter (TM-197, Tenmars Electronics Co., Ltd., Taipei, Taiwan). Surface electrical characteristics of the composite films were evaluated by Kelvin probe force microscopy (KPFM, Suwon, Republic of Korea) using an EFM-type conductive cantilever and a highly oriented pyrolytic graphite (HOPG) reference.

2.4. Electrical Output Measurement

Electrical outputs of the MDHG were evaluated under contact–separation excitation. Open-circuit voltage (V_OC) and short-circuit current (I_SC) of TENG and EMG were recorded using an electrometer (6514, Keithley Instruments Inc., Cleveland, OH, USA) connected to a multi-channel data acquisition (DAQ) system (PCI-6220, NI, Austin, TX, USA). The outputs were also measured across external load resistances (up to 1 GΩ) to determine the load-dependent behavior. The instantaneous output power was calculated using Ohm’s law, and the optimal load resistance was determined from the resistance-dependent measurements. To apply a controlled mechanical input, an electrodynamic shaker (Labworks Inc., LW139.138-40, Costa Mesa, CA, USA), driven by a function generator (33120A, Agilent Technologies Inc., Santa Clara, CA, USA), was used.

3. Results and Discussion

3.1. Architecture and Coupled Working Mechanisms of the MDHG

The MDHG (magneto-dielectric hybrid generator) was designed to integrate triboelectric and electromagnetic energy harvesting within a single elastomer-based platform by employing a soft magnetized dielectric composite film (SMDC) as a dual-functional layer. The overall strategic framework of the MDHG, encompassing the SMDC synthesis, integrated architecture, and the synergistically coupled transduction mechanisms, is visually summarized in Figure 1.

The overall fabrication process for the SMDC is summarized in Figure 1a. To promote uniform dispersion of fillers and stable film formation, a stepwise mixing process was adopted. Specifically, BTO was initially dispersed in Ecoflex Part B because its lower viscosity effectively facilitates particle wetting and suppresses the inherent agglomeration tendency of high surface energy of BTO. After BTO pre-dispersion, NdFeB was introduced and further mixed, followed by the addition of Ecoflex Part A to complete the crosslinkable precursor. For NdFeB-only samples (soft magnetized composite, SMC), NdFeB was directly mixed with Ecoflex Part B prior to adding Part A. The NdFeB loading was controlled by the Ecoflex:NdFeB weight ratio (1:1 to 1:5), and the BTO amount was defined as parts by weight (pbw) with respect to the total mass of the Ecoflex + NdFeB composite (1–15 pbw), corresponding to 1–15 parts of BTO per 100 parts of the Ecoflex + NdFeB composite. The mixed precursor was cast into a 3D-printed mold (thickness: 1 mm) and thermally cured at 60 °C for 4 h to yield the composite film. The curing temperature was selected primarily to preserve the dimensional fidelity of the PLA mold and ensure batch-to-batch reproducibility of the film geometry. In preliminary trials, accelerated curing at elevated temperatures (≥70 °C) caused noticeable thermal deformation of the PLA mold, which distorted the cavity geometry and prevented reliable reproduction of the intended film shape and thickness. Therefore, 60 °C was adopted as a robust processing condition that maintains mold integrity while providing consistent curing, enabling a fair comparison of composition-dependent electrical outputs. To enable EMG hybridization, the cured film was magnetized using a pulse magnetizer that applied a transient magnetic field of 5 T, aligning magnetic domains and leaving remanent magnetization in the NdFeB phase to form the SMDC [46].

The magnetically activated SMDC exhibited a stable surface magnetic flux density, which was quantified using a gaussmeter, as presented in Figure 1b. For an SMDC film with dimensions of 40 × 40 × 1 mm³, the average magnetic flux density measured above the top surface was 5.44 mT, confirming that the layer can act as an effective magnetic source for inductive transduction during operation. This quantified magnetic field supports the device concept illustrated in Figure 1c, in which the SMDC simultaneously serves as the triboelectric contact layer and the permanent magnetic component coupled to the coil. In the assembled device, the SMDC functions as the tribo-negative contact layer laminated on the bottom Al electrode. The top Al layer functions as both the tribo-positive counter surface and the current-collecting electrode during repeated contact–separation. In parallel, the Cu coil integrated with the top electrode experiences a time-varying magnetic flux produced by the magnetized SMDC during the same mechanical operation, thereby enabling electromagnetic induction. This coil-coupled architecture eliminates the need for an external bulky magnet while preserving device flexibility.

The coupled working mechanism under contact–separation operation is described in Figure 1d as a synchronized TENG-EMG process. Upon periodic contact and separation between the top Al electrode and the SMDC surface, the MDHG generates two concurrent electrical outputs: a triboelectric signal driven by contact electrification and electrostatic induction and an electromagnetic signal driven by time-varying magnetic flux through the coil.

In the TENG mode, contact electrification establishes opposite surface charges (±σ) at the interface, and subsequent separation produces an electric potential difference that drives electron flow through the external load to balance the electrostatic field. For a contact–separation TENG, the open-circuit voltage (V_OC) is expressed using Equations (1)–(4):

$$V=-{\displaystyle \frac{Q}{A{\epsilon }_0}}(x(t)+{\displaystyle \frac{d}{{\epsilon }_r}})+{\displaystyle \frac{\sigma x(t)}{{\epsilon }_0}}$$

(1)

$$V_{OC}={\displaystyle \frac{\sigma x(t)}{{\epsilon }_0}}$$

(2)

$$V_C={\displaystyle \frac{{\varphi }_M-{\varphi }_D}{e}}$$

(3)

$$\sigma \propto {\epsilon }_0{\epsilon }_r[{\displaystyle \frac{{\varphi }_M-{\varphi }_D}{d}}]$$

(4)

where x(t) is the instantaneous air-gap distance, d and ε_r are the dielectric thickness and relative permittivity of the composite layer, A is the effective contact area, ε₀ is the permittivity of vacuum, Q is the transferred charge through the external circuit, σ is charge density on the surface, and Φ_M and Φ_D are the effective work functions of the metal and the composite surface, respectively [47,48]. The short-circuit current (I_SC) of TENG (I_TENG) arises from charge redistribution through the external load in response to the time-varying electrostatic potential during contact–separation cycling, and its polarity reverses between the separating and approaching processes. During contact, the SMDC/Al interface establishes the triboelectric charge distribution (Figure 1d(i)). As the contacted layers begin to separate, the increasing gap distance builds the potential difference and produces a transient current (Figure 1d(ii)). At the maximum separation distance, the net current diminishes as the system approaches electrostatic equilibrium (Figure 1d(iii)). As the two layers approach again, the potential difference reverses, generating an opposite-polarity current and completing one triboelectric cycle (Figure 1d(iv)).

In the EMG mode, the motion of the magnetized SMDC relative to the coil changes the magnetic flux (Φ_B) through the coil, generating an induced voltage according to Faraday’s law (Equation (5)):

$$V_{EMG}=-N{\displaystyle \frac{\Delta {\varphi }_B}{\Delta t}}$$

(5)

where N is the number of turns. In this work, the time-varying flux was captured using a copper coil with a 4 cm diameter and 1000 turns. The measured DC resistance was 132.76 Ω. The inductance measured using an LCR meter was 14.48 mH at 100 Hz, 14.43 mH at 1 kHz, and 14.34 mH at 10 kHz, indicating minimal frequency dependence in the tested range. Based on these values, the coil’s electrical time constant was approximately 110 μs, supporting an adequate electrical response relative to the contact–separation motion. The electromagnetic current, denoted as I_EMG, is generated when the induced voltage from the changing magnetic flux drives charge flow through the external load. The current magnitude is governed by the coil’s internal resistance, together with the external load impedance, and its polarity reverses as the direction of flux change switches between separation and approach. When the device is in contact, the coil–SMDC geometry sets a baseline flux linkage (Figure 1d(i)). As the contacted layers begin to separate, the flux linkage changes, and an EMG signal is generated (Figure 1d(ii)). At the maximum separation distance, the induced signal diminishes as the flux approaches an extremum (Figure 1d(iii)). As the layers approach again, the flux change reverses, producing an opposite-polarity EMG signal (Figure 1d(iv)).

3.2. Structural Characterization of NdFeB, BTO, and the SMDC

The structural and compositional characteristics of NdFeB, BTO, and the resulting SMDC were comprehensively examined to describe the material basis of the MDHG, as detailed in Figure 2.

Figure 2a shows that the XRD pattern of the magnetic particles agrees well with the Nd₂Fe₁₄B reference (PDF#39-0473), indicating the crystalline phase of the NdFeB filler. In addition to the dominant NdFeB reflections, weak peaks assignable to Nd₂O₃ (PDF#43-1023) and Fe₂O₃ (PDF#33-0664) are also observed, which is consistent with the formation of a native oxide layer on NdFeB under ambient exposure. Figure 2b shows that the NdFeB particles have irregular morphologies with a characteristic size in the micrometer range, and the magnified image indicates a representative size of ~5 μm. The corresponding EDS maps show that Nd and Fe signals are distributed over the particle body, whereas the oxygen signal appears preferentially near the particle boundaries. Taken together with the oxide-related XRD features, this observation supports the presence of an oxide-enriched surface region (native passivation shell) rather than solely uniform oxygen adsorption.

For the dielectric filler, Figure 2c shows an XRD pattern that matches the perovskite BaTiO₃ reference (PDF#31-0174), indicating the intended high-dielectric ceramic phase of the BTO particles. Figure 2d further shows nanoscale BTO particles with elemental composition comprising Ba, Ti, and O, with the primary particle size being in the order of ~100 nm, while some degree of agglomeration is observed due to their high surface energy. Such nanoscale high-dielectric fillers are expected to occupy interstitial regions more effectively in a multi-filler polymer composite, which can be beneficial for enhancing the effective permittivity and stabilizing charge-related behavior in the triboelectric layer [49].

The presence of both NdFeB and BTO within the Ecoflex matrix was further supported by the structural and elemental signatures of the SMDC layer. Figure 2e shows that pristine Ecoflex does not exhibit distinct diffraction peaks (Figure S1), which is consistent with its predominantly amorphous/polymeric nature, whereas the SMDC exhibits pronounced diffraction features originating from the embedded inorganic constituents. Given the high NdFeB loading, NdFeB-related reflections dominate the SMDC pattern. Nevertheless, BTO-associated reflections remain discernible, indicating that the high-dielectric ceramic phase was retained after composite processing and thermal curing. In parallel, Figure 2f provides spatially resolved compositional information through SEM-EDS mapping. The Si signal represents the silicone matrix, while the Fe and Ti signals correspond to NdFeB and BTO, respectively. The Fe and Ti maps are broadly distributed across the inspected region rather than confined to isolated domains, suggesting that both fillers were incorporated throughout the SMDC without pronounced macroscopic segregation. The co-localized detection of Fe- and Ti-associated signals within the same cross-sectional field supports a dual-filler microstructure essential for simultaneous magnetic functionality and dielectric modulation.

3.3. Effect of NdFeB Content on TENG and EMG Outputs

The effect of NdFeB loading in the silicone-based magnetic composite (SMC, Ecoflex + NdFeB) layer was systematically investigated. Figure 3 summarizes how the NdFeB loading in the SMC layer governs both the triboelectric and electromagnetic behaviors of the MDHG by simultaneously modulating surface electrical properties, dielectric polarization, and magnetic flux density.

As illustrated in Figure 3a, NdFeB particles are inferred to possess an insulating native oxide surface layer that provides self-passivation and preserves interparticle electrical isolation. This interpretation is consistent with the oxide-related reflections in XRD and the oxygen-enriched signal localized near particle boundaries in EDS (Figure 2a,b). This structural feature is critical in high-loading composites because it can suppress direct particle-to-particle electrical contact. Consequently, it mitigates the formation of long-range conductive pathways, thereby maintaining the insulating nature of the triboelectric layer even at elevated magnetic filler contents.

The NdFeB loading produced a non-monotonic trend in the triboelectric output. Figure 3b shows that both the open-circuit voltage and short-circuit current increased from the pristine Ecoflex film and reached a maximum at an Ecoflex:NdFeB weight ratio of 1:2, yielding V_OC = 303.71 V and I_SC = 44.69 μA. Compared with the pristine sample at 1:0 (V_OC = 139.69 V and I_SC = 21.70 μA), the 1:2 composition exhibited a pronounced enhancement, indicating an optimal balance between dielectric polarization benefits and electrostatic charge utilization. Relative to 1:0, the 1:2 sample exhibited a 2.17-fold increase in V_OC and a 2.06-fold increase in I_SC. At higher loadings, the triboelectric output decreased and approached the pristine-level magnitude. The 1:4 and 1:5 samples yielded V_OC of 149.94 V and 147.36 V, and I_SC of 31.42 μA and 28.49 μA, respectively. This behavior implies that simply increasing the magnetic filler fraction does not continuously improve the triboelectric performance, suggesting the existence of an optimal composition.

In contrast to the peaked triboelectric output, the effective dielectric constant of the composite layer increased monotonically with NdFeB loading. Figure 3c shows a continuous rise in relative permittivity as the NdFeB content increased, which is consistent with a microcapacitor-network interpretation of conductor-like inclusions dispersed in a dielectric elastomer (Figure S2). In this framework, each isolated NdFeB particle acts as a floating microelectrode, and neighboring particles separated by a thin Ecoflex gap form numerous microcapacitors. As the particle number density increases, the population of such interparticle capacitors increases, thereby raising the effective capacitance of the composite. This trend can be rationalized by the capacitance relation C = ε₀ε_rA/d, where the interparticle gap (d) and effective facing area (A) between adjacent particles define each microcapacitor contribution. Accordingly, the resulting increase in the equivalent capacitance is reflected as an increased effective permittivity. In addition to this geometric microcapacitor effect, interfacial polarization, often described within the Maxwell-Wagner-Sillars (MWS) framework, can further enhance the apparent permittivity in heterogeneous dielectrics due to charge accumulation at the conductor-insulator interfaces under an applied AC field [50].

Notably, the monotonic rise in permittivity did not coincide with a noticeable increase in dielectric loss, even at the highest loading (1:5). The loss tangent remained sufficiently low (Figure S3) to support stable TENG operation, implying that the composite remained below the electrical percolation threshold. This behavior is consistent with the presence of native-oxide-enabled self-passivation on NdFeB particles, which preserves interparticle electrical isolation and suppresses DC leakage currents under high filler loading (Figure S2). As a result, the abrupt increase in leakage current commonly observed in conductive-filler composites is avoided. Consequently, the composite functions as a polarization-efficient dielectric medium in which capacitive coupling and interfacial polarization dominate, rather than forming a leaky conductive network. Although higher permittivity often correlates with higher TENG output by increasing effective capacitance and charge storage during contact-separation [51,52], the present results deviate from this simple trend: the TENG output peaks at 1:2 and then decreases at higher NdFeB contents despite a continued increase in permittivity. This divergence suggests that surface-governed factors beyond bulk permittivity increasingly constrain the effective interfacial potential at high NdFeB loading, thereby necessitating surface-sensitive analysis.

To rationalize this behavior, the surface potential and work function of SMC were evaluated using Kelvin probe force microscopy (KPFM). Figure 3d reveals that the surface potential follows a trend similar to the TENG output, with the strongest tribo-negative potential observed in the 1:2 sample. Specifically, the surface potential reached −1864 mV in the 1:2 sample, whereas the high-loading samples returned close to the pristine (1:0) value of −355 mV, showing −365 mV at 1:4 and −364 mV at 1:5 ratios. Figure 3e further shows the work function increase (ΔΦ_D) in SMC relative to the pristine state. The maximum increase occurred at 1:2 with ΔΦ_D = +1.50 eV, while only minor increases were observed at 1:4 (ΔΦ_D = +0.31 eV) and 1:5 (ΔΦ_D = +0.01 eV). In KPFM, the sample work function is related to the contact potential difference (V_CPD) by Φ_sample = Φ_tip − eV_CPD. Accordingly, the work function increase relative to the pristine state can be expressed as ΔΦ_D = −eΔV_CPD. Physically, the large ΔΦ_D at the 1:2 ratio indicates an increased energetic offset between Al and the SMC surface, which strengthens the driving force for electron transfer from Al to SMC during contact electrification. However, at higher loadings (1:4, 1:5), this driving force diminishes despite the high permittivity, consistent with enhanced electrostatic screening and lateral charge redistribution enabled by dense near-surface conductive inclusions. Taken together with the divergence between the monotonic permittivity increase and the peaked KPFM-derived metrics, these results indicate that electrostatic screening becomes increasingly influential at high NdFeB loading, reducing the effective interfacial potential available for contact electrification and electrostatic induction [53].

The magnetic response and EMG output exhibited a distinct monotonic enhancement with increasing NdFeB loading. Figure 3f shows that the B-H hysteresis loops broadened as the NdFeB fraction increased, indicating higher magnetic content and increased remanent magnetization after pulse magnetization. This stronger remanence is expected to generate a larger magnetic flux density in operation, thereby increasing the time-varying flux through the coil during contact-separation motion and enhancing inductive output (Equation (5)). Consistently, Figure 3g shows that the EMG output increased with NdFeB loading. For example, the 1:5 sample (V_OC = 461.09 mV and I_SC = 1.55 mA) significantly outperformed the 1:1 sample (V_OC = 86.05 mV and I_SC = 0.26 mA), confirming that higher magnetic filler loading provides a practical route to strengthening electromagnetic induction. Notably, the triboelectrically optimal 1:2 composition still delivered a substantial EMG signal (V_OC = 252.52 mV and I_SC = 0.82 mA), demonstrating a balanced operating point for hybridization.

Collectively, the NdFeB loading functions as a quantitatively tunable material parameter that governs the balance between triboelectric charge generation and electromagnetic induction in the hybrid generator. As the magnetic filler fraction increases, the device response progressively transitions from a triboelectric-dominant regime to an EMG-favored regime, reflecting a composition-controlled shift in the dominant transduction pathway. Within this tunable trade space, the TENG output exhibits a distinct maximum at the intermediate 1:2 ratio. Accordingly, the 1:2 composition was selected as the baseline for subsequent SMDC optimization, as it maximizes triboelectric output while maintaining a meaningful electromagnetic contribution for hybrid operation.

3.4. BTO-Enabled Dielectric Filling for Enhanced Triboelectric Output

Following the NdFeB-loading optimization in the SMC baseline, BTO was incorporated to form the silicone-based magnetized dielectric composite (SMDC, Ecoflex + NdFeB + BTO). This interstitial dielectric filling strategy was implemented to selectively amplify triboelectric energetics, specifically surface potential, effective work function, and charge retention while maintaining the magnetic flux coupling required for EMG performance.

The structural concept presented in Figure 4a depicts nanoscale BTO occupying the interparticle voids formed among micrometer-scale NdFeB fillers. This hierarchical dual-filler arrangement is expected to increase polarization-active interfaces and promote connective dielectric pathways in the interstitial volume, a behavior often discussed in terms of interstitial percolation of high-k ceramics in polymer matrices.

Figure 4b shows that the triboelectric output increased with BTO content, defined as parts by weight (pbw) with respect to the total mass of the SMC baseline. TENG outputs reached a maximum at 5 pbw, followed by a decline at higher loadings. This optimum-type dependence is commonly observed in dielectric-enhanced triboelectric layers. Moderate loading increases charge storage capability and electrostatic induction, whereas excessive loading induces filler aggregation and increases dissipative loss, thereby diminishing the effective dielectric contribution to triboelectrification. Accordingly, 5 pbw represents a practically relevant regime in which interstitial dielectric filling is maximized without triggering pronounced degradation mechanisms.

The dielectric measurements support this interpretation. As shown in Figure 4c, the effective permittivity of SMDC increased up to 5 pbw and then decreased at 10 and 15 pbw. In parallel, the loss tangent increased sharply at high BTO contents, as shown in Figure S4. This behavior is consistent with the strong tendency of BTO nanoparticles to agglomerate because of their high surface energy, which becomes more severe at high loading. Agglomeration reduces the effective interfacial area that contributes to polarization and introduces locally intensified fields and dissipative pathways, resulting in reduced permittivity and increased dielectric loss. Therefore, the 5 pbw condition can be viewed as an interstitial filling window that enhances dielectric polarization efficiently while avoiding aggregation-driven losses.

Beyond bulk dielectric properties, Figure 4d provides an electronic-energy perspective on the enhanced triboelectrification. The effective work functions of the triboelectric composite layer progressively increased from pristine Ecoflex to SMC and further to SMDC, with the values of 4.46, 5.96, and 6.55 eV, respectively. In contact–separation TENGs, the electron-transfer tendency at contact can be strengthened when the energetic offset between the metal electrode and the dielectric surface increases, which can raise the attainable surface charge density and the resulting V_OC, as described in Equations (2) and (4). Considering the work function of the counter material (Al, 3.80 eV), the work function difference (Φ_M − Φ_D) widened from 0.66 eV for pristine to 2.16 eV for SMC and 2.75 eV for SMDC. This trend supports a material-level rationale that BTO increases tribo-negative character and improves charge retention under cyclic contact–separation, thereby reinforcing the electrostatic induction process. Although NdFeB and BTO were embedded functional fillers, the outermost contact interface was still dominated by the Ecoflex matrix. Accordingly, the enhanced triboelectric output is not interpreted as a direct surface-chemistry replacement, but rather as an effective increase in tribo-negativity arising from a KPFM-consistent interfacial-energetic shift driven by near-surface dielectric polarization and improved charge retention in the composite.

The practical consequence of this combined dielectric–electronic reinforcement is summarized in Figure 4e. Relative to pristine Ecoflex, the BTO-filled SMDC delivered a 2.87-fold increase in V_OC and a 2.62-fold increase in I_SC. Notably, this enhancement was achieved by introducing a small high-k additive into the interstitial volume, indicating that triboelectric performance can be improved without relying on additional structural complexity or sacrificing magnetic functionality. In this regard, the interstitial filling approach is particularly attractive because it strengthens the triboelectric response through a material-level modification of the composite layer rather than device-level reconfiguration. This outcome directly supports the central thesis of this work, namely that interstitial dielectric percolation serves as an effective tuning strategy for maximizing triboelectric output while preserving magnetic functionality for hybrid operation.

Crucially, the electromagnetic output remained nearly invariant upon BTO incorporation, as shown in Figure 4f. At the optimized BTO content, the EMG outputs remained within 2.93% in V_OC and 8.53% in I_SC of the SMC baseline, indicating no meaningful penalty in electromagnetic induction. This near-invariance is consistent with the fact that BTO constitutes only a minor fraction of the composite and therefore does not substantially reduce the NdFeB component responsible for remanent magnetization and flux generation. Accordingly, the time-varying magnetic flux coupled to the coil was effectively preserved across the tested BTO contents, even though the triboelectric response was noticeably enhanced. This pronounced TENG enhancement with negligible EMG penalty highlights the practical advantage of the interstitial dielectric filling strategy for tunable triboelectric–electromagnetic hybrid generators. Notably, BTO incorporation effectively strengthens the triboelectric transduction of the composite layer while leaving the electromagnetic induction pathway at a comparable level. This hybrid-compatible improvement elevates triboelectric output without requiring additional device-level complexity or sacrificing the magnetic functionality required for coil-coupled EMG operation. As a result, the triboelectric pathway can be enhanced on top of an EMG-capable baseline, enabling the hybrid output characteristics to be tailored in a material-driven manner without compromising either transduction pathway.

3.5. Comprehensive Performance Characterization of the Optimized MDHG

The fully optimized hybrid generator was evaluated to benchmark the final TENG and EMG performances and to clarify their load, force, displacement, and frequency dependencies.

Under a periodic contact–separation operation at 1 Hz, the device delivered stable cyclic outputs over 8 s, corresponding to eight consecutive cycles (Figure 5a). The optimized TENG output reached a V_OC of 400.40 V and an I_SC of 56.95 μA, while the EMG output reached a V_OC of 260.04 mV and an I_SC of 0.89 mA. The simultaneous generation of a high-voltage triboelectric signal and a current-capable electromagnetic signal under identical actuation confirms the intended hybrid complementarity of the MDHG.

The load-dependent measurements further emphasize that the two branches behave as fundamentally different source types. For the TENG, the voltage increased with load resistance, and the power peaked at a very high resistance (Figure 5b), reaching 7.01 mW at 5 MΩ. This behavior is consistent with the intrinsically high internal impedance of contact–separation TENGs, where the measured voltage approaches its open-circuit limit under large external resistance and the power output typically peaks at a resistance comparable to the effective internal impedance. In contrast, the EMG exhibited a maximum power at a much lower load resistance (Figure 5c), reaching 144.56 μW at 200 Ω. This low-resistance optimum strongly suggests that the EMG internal impedance was dominated by the coil resistance in the present low-frequency regime, where the inductive reactance is comparatively small. Accordingly, the observed power maximum near a few hundred ohms is physically consistent with resistance-dominated matching in low-frequency electromagnetic induction.

To assess practical energy storage, capacitor charging tests were conducted using a 22 μF capacitor (Figure 5d), where the charging voltage V_Cap(t) can be directly compared among TENG-only, EMG-only, and hybrid operation. The instantaneous stored energy in the capacitor was calculated using Equation (6):

$$E={\displaystyle \frac{1}{2}}C{V}_{Cap}^2$$

(6)

Because V_Cap(t) sets the attainable storage voltage relevant to downstream electronics, while E_Cap(t) quantifies the net usable energy accumulated in the storage element, both metrics were analyzed to rigorously evaluate charging performance. Under an identical charging duration of 300 s, the hybrid configuration charged the 22 μF capacitor to 8.75 V, exceeding the TENG-only case at 7.61 V and the EMG-only case at 1.57 V. This voltage increase leads to a disproportionately larger energy gain due to the quadratic dependence on voltage. The hybrid mode stored 842.18 μJ, whereas the TENG-only and EMG-only modes stored 637.03 μJ and 27.11 μJ, respectively. Notably, the hybrid energy also exceeded the arithmetic sum of the separately measured single-source energies over the same 300 s window (637.03 μJ + 27.11 μJ = 664.14 μJ) by 26.81%. This behavior can arise in rectified capacitor charging because the effective operating point is determined by the instantaneous capacitor voltage together with the nonlinear conduction of the rectifier and the source–load impedance interactions. In the hybrid circuit, the TENG and EMG outputs are independently full-bridge rectified and then combined at a common DC node to charge the same capacitor (Figure S5). Therefore, the charging trajectory in the hybrid case is not a simple linear superposition of two isolated charging tests; rather, the simultaneous contribution can increase the net charge throughput delivered to the capacitor within the same actuation window. Consistent with this interpretation, the delivered charge at 300 s (Q_Cap = C · V_Cap) increases from 167.42 μC in the TENG-only mode to 192.50 μC in the hybrid mode, corresponding to an additional 25.08 μC. Overall, the results indicate complementary roles during rectified charging: the TENG establishes a higher storage-voltage baseline, while the EMG primarily contributes to additional charge throughput. In practice, rectification losses and impedance matching limit ideal additivity. Nevertheless, the observed increase in V_Cap(t), E_Cap(t), and Q_Cap supports enhanced net charging in the hybrid configuration under identical actuation conditions.

The mechanical response tests further differentiate the governing factors of each output. The TENG voltage increased with applied force and gradually approached saturation (Figure 5e). This behavior is consistent with a contact-area-mediated mechanism. Increasing force enlarges the real contact area and improves interfacial intimacy, thereby increasing effective charge generation and electrostatic induction until the interface approaches a practical contact limit. In this regime, additional force yields diminishing gains because the contact area and charge density approach a maximum determined by surface morphology and material properties [54]. The EMG current increased with displacement and showed high linearity (Figure 5f), with R² = 0.9906. This near-linear scaling is physically reasonable because a larger displacement increases the magnetic flux excursion experienced by the coil during each cycle, thereby strengthening the induced electromagnetic response. This linear scaling supports the use of the EMG current as a quantitative indicator of displacement in hybrid operation, complementing the force-sensitive triboelectric response. Finally, the frequency response clearly differentiated the two transduction mechanisms (Figure 5g). The TENG V_OC remained largely insensitive to frequency over 1–5 Hz with variations within 10%, while a maximum reduction of 11.78% was observed at 15 Hz. This reduction is plausibly associated with an experimental artifact in which the effectively transferred force slightly decreases as the actuator frequency increases in the current setup, rather than a fundamental limitation of the contact–separation TENG mechanism. In contrast, the EMG I_SC increased with frequency, consistent with faster motion increasing the rate of magnetic flux change and thus strengthening electromagnetic induction under otherwise similar displacement conditions.

Overall, Figure 5 demonstrates that the optimized MDHG simultaneously delivers a voltage-dominant TENG branch and a current-capable EMG branch, while each output exhibits distinct and interpretable dependencies on load and mechanical inputs. These results suggest the practical utility of the MDHG both as a hybrid power source for energy storage and as a self-powered sensing platform in which force-, displacement-, and frequency-related information can be encoded into complementary electrical signatures.

3.6. Self-Powered Wearable Smart-Skin Demonstrations for Multimodal Sensing

The optimized MDHG, based on the optimized SMDC layer, was leveraged to demonstrate multimodal tactile perception capabilities. Figure 6 presents an application-level implementation of this optimized MDHG as a self-powered wearable sensor for smart-skin systems, capable of decoupling complex tactile stimuli into distinct electrical readouts. Here, the optimized SMDC was selected to balance KPFM-derived triboelectric interfacial energetics with remanence-driven, coil-coupled flux generation, thereby preserving both transduction pathways for hybrid operation.

Figure 6a illustrates the architecture of the wearable MDHG sensor designed to discriminate distinct tactile parameters, which include touch occurrence, touch intensity derived from displacement, and counter-material identity. This discrimination is achieved by utilizing the complementary characteristics of the hybrid outputs of MDHG. The triboelectric branch functions primarily as a contact indicator and material-sensitive channel, whereas the electromagnetic branch serves as a displacement-sensitive channel quantifying pressure-induced deformation. To ensure compatibility with on-body applications, the device adopts a flexible arch geometry. This arch structure confers dual mechanical advantages. First, its intrinsic restoring force facilitates stable and repetitive contact-separation cycles without external mechanical guides. Second, its compliance effectively transduces external pressure into structural deformation of the SMDC layer relative to the coil, thereby modulating the magnetic flux for EMG output generation. Here, the optimized SMDC layer serves as both the triboelectric contact medium and the magnetized flux source for coil-coupled EMG. Consequently, the platform naturally decouples tactile information, where the TENG output signals interfacial contact and material properties, while the EMG output reflects the magnitude of deformation. This division of roles underpins the multimodal perception capabilities of the sensor and enables the simultaneous inference of multiple touch attributes from a single self-powered device.

The separability of these hybrid signals is validated by comparing weak and strong contact regimes. As illustrated in Figure 6b, the device establishes surface contact under weak contact conditions and generates a distinct triboelectric response. However, due to limited structural deformation, the relative motion is insufficient to induce a significant magnetic flux change. This results in a negligible EMG current near the noise floor. Conversely, Figure 6c demonstrates that the substantial deformation of the magnetized composite relative to the coil under strong contact induces a pronounced EMG current. Notably, the contact-related TENG voltage remains comparable between the weak-state value of 4.44 V and the strong-state value of 4.74 V. This modest change indicates that both conditions realize a broadly comparable contact state, with only a limited increase in effective contact area or interfacial intimacy under stronger pressing. In contrast, the EMG current exhibits a significant dependency on pressure as it increases from 16.25 μA under weak contact to 59.13 μA under strong contact. These observations confirm a practical mechanism for signal decoupling, where the triboelectric signal serves as a binary indicator of touch occurrence, and the electromagnetic signal provides a quantitative measure of touch intensity. Thus, the device avoids signal redundancy and instead offers complementary signatures that enable reliable discrimination between contact status and mechanical intensity.

Beyond contact intensity, the MDHG extends its sensing capabilities to material recognition by leveraging the sensitivity of contact electrification to the counter-surface identity, as depicted in Figure 6d. When interacting with five different counter materials, namely Al, Cu, PI, PET, and PTFE, the TENG waveforms exhibit distinct variations in polarity and amplitude corresponding to the relative positions of the materials in the triboelectric series. Practically, material discrimination is achieved through two waveform descriptors. First, the polarity and asymmetry of the positive and negative peaks serve as robust qualitative indicators that reflect the directionality of charge transfer and electrostatic induction. Second, quantitative features allow for the differentiation of materials with similar peak magnitudes. Specifically, the peak-to-peak voltage (V_PP) captures the overall waveform spread. Furthermore, the ratio of positive-to-negative peak magnitudes (R_PN) reflects peak imbalance and thus the relative dominance of one polarity over the other. Importantly, an R_PN value greater than 1 indicates a larger positive peak magnitude than the negative peak magnitude. Using these descriptors, PET is readily identifiable because its response exhibits a markedly larger amplitude, placing it in an isolated signal region of the feature space. The remaining materials form a denser amplitude cluster but still exhibit distinguishable combinations of V_PP and R_PN, enabling additional separation even when peak magnitudes partially overlap. This converts raw voltage waveforms into compact material fingerprints suitable for lightweight decision rules in wearable sensing.

Collectively, Figure 6 establishes that the MDHG-based hybrid sensor supports three layers of tactile inference within a single self-powered architecture. These layers include touch detection through TENG response, pressure quantification through EMG magnitude, and material discrimination through triboelectric waveform analysis. This integrated capability underscores the practical value of the tunable triboelectric–electromagnetic hybrid concept. By employing a material strategy that enhances triboelectric performance without compromising EMG functionality, the device realizes a multimodal tactile sensor capable of simultaneously capturing contact and material cues with reduced signal interference and improved interpretability.

4. Conclusions

In this study, a material-level framework for tunable triboelectric–electromagnetic hybrid generators was established by integrating magnetic filler loading optimization with interstitial dielectric percolation within a single soft composite layer. The resulting SMDC functions as a dual-purpose medium that simultaneously enables triboelectric contact electrification and provides an embedded remanent-flux source for coil-coupled electromagnetic induction. This approach successfully realizes a compact and compliant hybrid architecture without the need for external bulky magnets. The investigation into NdFeB loading defined a composition-controlled trade space in which electromagnetic induction enhances with increasing magnetic content, whereas triboelectric performance is constrained by electrostatic screening and lateral charge redistribution at high loadings. To overcome this limitation, nanoscale BaTiO₃ was introduced as an interstitial dielectric phase. This hierarchical dual-filler engineering effectively increases polarization-active interfaces and widens the energetic offset between the metal and the composite, thereby selectively amplifying the triboelectric pathway while preserving the magnetic flux required for EMG hybridization. Under identical actuation, the fully optimized MDHG delivered complementary hybrid outputs with a V_OC of 400.40 V and an I_SC of 56.95 μA in the triboelectric branch, and a V_OC of 260.04 mV and an I_SC of 0.89 mA in the electromagnetic branch. Notably, compared to the pristine Ecoflex matrix, the optimized triboelectric output corresponds to a 2.87-fold increase in V_OC and a 2.62-fold increase in I_SC, confirming the efficacy of the interstitial dielectric percolation strategy. Consistent with their fundamentally different source characteristics, the TENG branch reached a maximum power of 7.01 mW at 5 MΩ, while the EMG branch peaked at 144.56 μW at 200 Ω. In practical energy storage tests using a 22 μF capacitor under an identical 300 s charging window, the hybrid operation increased the stored energy to 842.18 μJ, exceeding both the TENG-only and EMG-only cases and surpassing the arithmetic sum of the two single-source energies by 26.81%, which directly evidences an additive hybrid advantage in storable energy. Collectively, these results position the SMDC strategy as a scalable design principle for compact hybrid energy harvesters. Furthermore, the successful demonstration of a flexible smart-skin interface, capable of concurrently resolving touch and material identity via the triboelectric branch while quantifying structural displacement via the electromagnetic branch, validates the practical versatility of this framework for advanced self-powered wearable perception systems.