A Multifunctional Eutectogel for Wide‐Temperature and Intelligent Speech Recognition

ABSTRACT

The practical application of flexible sensors is often constrained by limited mechanical properties and a narrow operating temperature range, particularly under subzero conditions. Herein, a multifunctional eutectogel is fabricated via one‐step thermal polymerization of methacrylic acid and functionalized adenine in a binary solvent system comprising water and a deep eutectic solvent (DES). This gel exhibits high strength, remarkable fatigue resistance (withstanding 10 000 electrical cycles), and stable performance across a broad temperature range. By leveraging hydrophobic domains and a multi‐physical crosslinking strategy, a synergistic balance between mechanical flexibility and structural stability is achieved. The hydrophobic structures are mainly contributed by hydrophobic interactions, π–π stacking, hydrogen bonding, and electrostatic interactions, while the flexible hydrophilic network incorporates multiple physical interactions, endowing the material with outstanding mechanical properties. The eutectic network formed in the binary solvent system enables reliable operation across a wide temperature range (−20°C to 80°C). As a multifunctional sensor, it accurately monitors human motion and functions as a bendable supercapacitor, retaining a high areal capacitance of 495 mF·cm⁻² and 94.8% capacitance after 1400 charge–discharge cycles. When integrated with a deep learning algorithm, the system enables real‐time data processing and multilingual speech recognition, demonstrating considerable potential for next‐generation wearable electronics.

1. Introduction

Gel‐based materials have emerged as promising candidates for flexible electronics, owing to their three‐dimensional network structures, inherent flexibility, and efficient ion transport capability, which closely resemble biological tissues [1]. However, unlike biological hydrogels, synthetic hydrogels often suffer from poor mechanical properties, severely limiting their application in fields that demand high strength and toughness, such as wearable sensors and energy storage devices [2]. To overcome these limitations, various strategies have been developed to significantly enhance the mechanical performance of synthetic hydrogels, including double‐network (DN) gels [3], nanocomposite (NC) gels [4], topological [5] or slide‐ring gels [6], and dual‐crosslinked gels [7]. Unfortunately, although some individual properties of these gels can rival or even surpass those of living tissues, they often exhibit inadequate long‐term stability and a limited operational temperature range. At subzero temperatures, they are prone to freezing, while at elevated temperatures, water loss leads to significant performance degradation. While organogels and ionogels can partially mitigate these issues, their practical application in wearable systems is often hindered by drawbacks such as solvent toxicity, volatility, and high manufacturing costs. Therefore, the development of novel flexible materials that combine high strength, toughness, fatigue resistance, and a broad operating temperature range remains a critical yet challenging goal.

The introduction of supramolecular interactions, including non‐covalent and dynamic covalent bonds, as energy dissipation units into polymer gels can effectively enhance material toughness and fracture elongation [8]. This is achieved through the dissociation and reconstruction of chain‐bound motifs, which facilitate efficient energy dispersion and stress transfer. For instance, Lei et al. [9] developed an all‐polymer hydrogel utilizing abundant cluster hydrogen bonding, electrostatic interactions between PIL chains and anionic PSS shells, along with sparse chain entanglements. This system served as a multi‐responsive and self‐powered wearable sensor, demonstrating high strength, toughness, low hysteresis, and good electrical conductivity. Although physically cross‐linked gels can improve toughness, sliding friction between polymer chains may negatively affect stress transfer [10], leading to compromised mechanical performance, such as reduced responsiveness and fatigue resistance under large deformations. Additionally, at low temperatures, freezing can prevent rapid response to external strains, cause baseline drift, and reduce detection sensitivity. Numerous methods have been developed to expand the operating temperature range of materials. Some researchers have disrupted the ordered arrangement of water by incorporating hydrophobic proteins or constructing specialized coatings, thereby lowering the freezing point of the material [11, 12]. For example, Zhang et al. designed a durable interpenetrating coating with excellent anti‐icing and de‐icing performance by combining a fluorinated amphiphilic ionomer with a biomimetic binder [13]. Incorporating organic solvents to disrupt the ordered arrangement of water is also an effective strategy for lowering the freezing point. For example, the researcher combined a binary solvent with inorganic salts, achieving excellent properties. This enables wireless transmission of distress signals through human‐machine interaction even at −78°C [14]. Incorporating inorganic salts such as LiCl or NaCl can effectively inhibit ice crystallization by disrupting hydrogen bonding between water molecules via strong ion hydration [15]. Alternatively, ionic liquids (ILs), organic salts with low melting points, can be applied to prepare ionogels, thereby circumventing the freezing issues common in hydrogels [16]. However, the addition of inorganic salts inevitably induces a salting‐out effect, where excessive ions may disrupt the three‐dimensional network of the hydrogel and sacrifice its mechanical properties. Meanwhile, the use of organic solvents and ILs raises concerns regarding toxicity, ionic conductivity, and cost, limiting their practical applicability [17, 18, 19].

Deep eutectic solvents (DESs), a class of low‐eutectic‐temperature liquids formed through hydrogen bonding between hydrogen bond donors (HBDs) and hydrogen bond acceptors (HBAs), have shown great potential in the design of functional gels due to their good biocompatibility, tunable composition, and cost‐effectiveness [20]. Several DES‐based anti‐freezing hydrogels, referred to as eutectogels, have been developed for use in wearable strain sensors [21, 22, 23]. However, the high viscosity of pure DESs tends to hinder ion conduction within eutectogels and increases interpolymer chain friction, leading to reduced fatigue resistance, resilience, and ultimately lower sensitivity in strain sensing [24]. It has been reported that retaining a certain amount of water in eutectogels can reduce the viscosity of DESs and enhance ionic conductivity, thereby facilitating the fabrication of highly sensitive strain sensors [10]. Moreover, the presence of water does not compromise the anti‐freezing properties of eutectogels. In our previous work, a highly resilient and fast‐responsive ionic hydrogel was fabricated by incorporating functionalized adenine, which enabled the connection between a “soft” hydrogen‐bonded cross‐linked polymer network and “rigid” domains reinforced by dual hydrogen bonds, hydrophobic interactions, and π–π stacking [25]. Nevertheless, the salting‐out effect of inorganic ions and limited fatigue stability continue to restrict its broader application. To date, to the best of our knowledge, studies on the role of hydrophilic‐hydrophobic composition and multiple physical interactions in regulating the mechanical and electrical properties of deep eutectic gels remain limited. Therefore, a systematic investigation into the effects of DES content, amphiphilic component design, and network structural evolution on the overall performance is essential for advancing the development of novel eutectogel materials.

Therefore, we fabricated a versatile eutectogel with high strength, toughness, fatigue resistance, and a broad operating temperature range via a facile thermally initiated polymerization strategy. The gel was synthesized from hydrophilic methacrylic acid (MAA) and a hydrophobic di‐hydrogen‐bonding monomer, adenine (BA‐A), uniformly dispersed in a binary solvent system comprising water and a deep eutectic solvent (DES). The surfactant cetyltrimethylammonium bromide (CTAB) was introduced to facilitate the dissolution of the BA‐A monomer and promote the formation of stable hydrophobic associations and electrostatic attractions within the eutectogel network. These interactions serve as dynamic physical cross‐links that rapidly dissociate and reorganize under stress, endowing the material with excellent elasticity and outstanding anti‐fatigue performance. Meanwhile, multiple physical interactions, such as hydrogen bonding, electrostatic forces, and chain entanglements, between the flexible hydrophilic polymer chains and the eutectic solvent contribute to remarkable toughness. The binary solvent system enables stable performance across a wide temperature range (−20°C to 80°C) and ensures reliable ionic conductivity. Moreover, the gel exhibits a fast response speed (400 mm/min) and exceptional fatigue stability, showing no baseline drift over 10 000 electrical loading‐unloading cycles. As a multifunctional sensor, the eutectogel can accurately detect human motion signals and even subtle facial expressions. It also functions reliably as a supercapacitor under various bending conditions. Notably, we integrated this flexible sensor with artificial intelligence technology to achieve real‐time recognition, processing, and analysis of multiple languages, reaching a classification accuracy of 94.21%. In summary, this work provides a promising strategy for developing more intelligent multifunctional eutectogels, showing great potential to advance the field of flexible and wearable electronics.

2. Results and Discussion

2.1. Schematic Illustration of Hydrogel Fabrication Process and Molecular Interaction Networks

Figure 1 schematically illustrates the fabrication process, structural design, and potential applications of the eutectogel. First, a deep eutectic solvent (DES) was formed by mixing choline chloride (HBA) and urea (HBD) at a molar ratio of 1:2, with the reaction maintained at 70°C for approximately 2 h (Figure S1). This DES was then combined with water to create a binary solvent system, into which cetyltrimethylammonium bromide (CTAB) was added to form a micellar solution. The composite P(MAA‐BA‐A) hydrogel was fabricated via micellar copolymerization of assembled BA‐A in CTAB micelles and methacrylic acid (MAA). With the addition of CTAB, uniformly dispersed BA‐A/CTAB micelles as nano‐sized reinforcing domains were bound together by dynamic hydrogen bonds, π–π stacking, and hydrophobic association interactions. Subsequently, thermally initiated free‐radical polymerization was carried out with the hydrophilic monomer methacrylic acid (MAA), yielding an amphiphilic copolymer with both hydrophilic and hydrophobic segments. During polymerization, the hydrophobic groups of BA‐A tend to aggregate, forming hydrophobic microdomains, while MAA interacts with water molecules to constitute hydrophilic regions. These hydrophobic microdomains further assemble via hydrophobic association and incorporate multiple physical interactions, such as hydrogen bonding and π–π stacking, which can rapidly dissociate and reconfigure under external force, enabling efficient energy dissipation. This dynamic behavior imparts the hydrogel with unique mechanical properties and stimulus‐responsive characteristics. Meanwhile, the abundant active groups (─COOH) on the hydrophilic segments facilitate the formation of dense dynamic hydrogen bonds within the polymer network and enhance chain entanglement, thereby reinforcing the gel structure [26]. The presence of water molecules helps mitigate interchain friction and alleviates stress concentration. Furthermore, the uniform distribution of the deep eutectic solvent throughout the eutectogel not only provides high ionic conductivity but also confers anti‐freezing capability, ensuring broad temperature adaptability [27]. This synergistic structural design endows the eutectogel with excellent mechanical stability, electrical conductivity, and environmental tolerance, making it a promising candidate for flexible and intelligent electronic devices.

FIGURE 1.

Composition, fabrication, and applications of the P(MAA‐BA‐A) eutectogel. Schematic diagram of the gel's structure, featuring hydrophilic‐hydrophobic domains stabilized by multiple physical interactions (e.g., hydrogen bonding, electrostatic forces, and π–π stacking). Illustration of its multifunctional applications in motion sensing, supercapacitors, and deep learning‐based speech recognition.

2.2. Morphological and Structure Characterization of Eutectogels

To investigate the chemical interactions and structural evolution within the hydrogel components, a series of characterization techniques was employed, including Fourier transform infrared (FTIR) spectroscopy, X‐ray diffraction (XRD), X‐ray photoelectron spectroscopy (XPS), and scanning electron microscopy (SEM), as summarized in Figure 2. The FTIR spectrum of eutectogel is presented in Figure 2a. Compared with the spectra of the hydrogel prepared without using deep eutectic solvent (DES), it can be seen from the infrared image that the ─OH peak of ─COOH redshifts from 3456 to 3442 cm⁻¹, and the ─C═O peak shifts from 1714 to 1695 cm⁻¹, confirming that DES forms hydrogen bonds with the main chain [28]. Meanwhile, characteristic peaks related to urea can be observed in the eutectic gel, specifically the C═O stretching vibration at 1670 cm⁻¹ and the N─H bending vibration at 1636 cm⁻¹ [29, 30]. The presence of these features, along with a noticeable shift, suggests the protonation of the NH₂ groups of urea within the eutectogel, which enhances the strength and rigidity of the N─H bonds. Since no free urea was added to the P(MAA‐BA‐A) eutectogel, these spectral changes confirm the successful incorporation and interaction of the DES components within the polymer network. The XRD pattern of the P(MAA‐BA‐A) eutectogel (Figure 2b) exhibits a broad, amorphous halo with no sharp crystalline peaks, consistent with its polymeric and non‐crystalline nature. Notably, the shift of peaks in the XRD graph and the decrease in the intensity of the amorphous peaks confirm that the addition of DES has formed new hydrogen bonds and ionic bonds with the system, promoting the uniformity of the network structure [31].

FIGURE 2.

Structural and compositional analysis of the eutectogels. (a–c) Spectroscopic characterization: (a) FTIR, (b) XRD, and (c) XPS survey scans. (d–f) High‐resolution XPS spectra of (d) C 1s, (e) N 1s, and (f) O 1s. (g–i) Morphological characterization: SEM images of (g,h) the 20% DES gel at different magnifications and (i) the pure hydrogel without DES. Scale bars: 50 µm (g) and 10 µm (i).

The molecular interactions were further corroborated by Raman spectroscopy (Figure S2). In the P(MAA‐BA‐A) eutectogel, the peaks observed between 2800 and 3000 cm⁻¹ are attributed to symmetric and asymmetric C─H stretching vibrations of hydrogel constituents, while the broad multi‐mode band in the 3100–3700 cm⁻¹ region corresponds to O─H stretching of liquid water. Characteristic Raman shifts at 714 and 3044 cm⁻¹ are assigned to C─C and C─N stretching vibrations from choline chloride, respectively [32]. The chemical composition of hydrogels with different formulations was further analyzed by X‐ray photoelectron spectroscopy (XPS). The full survey spectra (Figure 2c) confirm the presence of C, N, and O in both gels, with an additional Cl signal detected only in the 20% DES‐containing eutectogel. High‐resolution C 1s spectra (Figure 2d) of P(MAA‐BA‐A) gel show peaks at 284.80, 286, and 288.8 eV, corresponding to C─C, C═O, and O─C═O bonds originating from MAA and choline chloride. The N 1s spectrum (Figure 2e) reveals that N⁺ and N─H species in the eutectogel derive from CTAB, BA‐A, and urea. After the addition of DES, the binding energies of N⁺ and N─H exhibit a noticeable shift, indicating the presence of electrostatic and hydrogen bonding interactions between the N⁺ of choline chloride, the N─H of urea, and the polymer backbone. The broadening and increased intensity of these peaks further suggest a higher relative content of these nitrogen species. The O 1s spectrum (Figure 2f) of pure P(MAA‐BA‐A) hydrogel displays peaks at 531.8 and 533 eV, assigned to C─O─C and C─O bonds, respectively. With the addition of DES, these peaks shift to 532.6 and 534.55 eV, confirming strong electronic interactions between DES constituents and the polymer matrix [33].

Collectively, these results verify the successful synthesis of the P(MAA‐BA‐A) eutectogel, in which abundant intra‐ and intermolecular hydrogen bonds form between the DES and polymer chains, contributing to the enhanced mechanical properties of the dual‐network gel. SEM images of the gels with and without 20% DES are shown in Figure 2g–i. A well‐defined honeycomb‐like porous structure with uniform pore size can be clearly observed in the eutectogel, which results from the solvation effect that promotes homogeneous polymer chain distribution. This uniform porous architecture is highly beneficial for rapid ion transport, laying the foundation for the subsequent application of the material as a highly sensitive sensor [34]. As shown in Figure S3, it can be observed that the addition of DES makes the network structure of the gel more uniform and compact, which corresponds to its enhanced mechanical properties and fatigue resistance [35].

2.3. Characterization of Mechanical Properties

The mechanical properties of a material are fundamental to its suitability for various application scenarios. Since these properties are primarily governed by compositional design, a series of samples with systematically varied components was prepared to accurately understand the mechanical behavior of the hydrogels [36]. First, the influence of BA‐A content on the mechanical performance was investigated. As shown in Figure 3a–c, as the BA‐A content increased from 0 to 0.2 mmol, the toughness (0.46–2.5 MJ/m³), strength (235–446 KPa), and fracture elongation (368%–916%) of the hydrogel all improved significantly. This enhancement can be attributed to the incorporation of BA‐A, which introduced additional physical cross‐linking points and increased the cross‐linking density within the hydrogel network. Upon deformation, abundant hydrogen bonds between PMAA chains, along with ionic and hydrophobic associations, were disrupted and subsequently reformed, effectively dissipating energy and hindering crack propagation. This energy‐dissipation mechanism contributed to the improved mechanical performance [31, 37]. However, when the BA‐A content reached 0.4 mmol, all the mechanical metrics declined, likely due to overcrowding of the PMAA chains, which restricted segmental mobility and resulted in an overly rigid network. Therefore, the optimal BA‐A content was determined to be 0.2 mmol.

FIGURE 3.

Mechanical performance of the P(MAA‐BA‐A) eutectogel as a function of composition. (a–c) Effect of BA‐A content: (a) stress–strain curves, (b) fracture stress and strain, (c) elastic modulus and toughness. (d–f) Effect of DES content: (d) stress–strain curves, (e) fracture stress and strain, (f) elastic modulus and toughness. (g) Visual evidence of mechanical robustness, including high stretchability, multi‐axial deformation, and puncture resistance.

The influence of the deep eutectic solvent (DES) content on the mechanical properties was further investigated by systematically varying its proportion in the binary solvent system (0%, 10%, 20%, 30%, 40%). As shown in Figure 3d–f, the fracture elongation initially increased and then decreased with rising DES content. When the DES content increased from 0% to 20%, the tensile strength, fracture elongation, and toughness improved from 233 to 446 KPa, 450% to 916%, and 0.6 to 2.5 MJ/m³. This enhancement can be attributed to the introduction of a moderate amount of DES, which reinforces the network through additional hydrogen bonding and electrostatic interactions, thereby increasing the interchain distance and improving the material's ability to withstand external stress [38, 39]. However, when the DES content was further increased from 20% to 40%, the overall mechanical performance declined. This reduction is ascribed to the disruption of hydrophobic associations acting as physical cross‐linking points. In the eutectogel system, hydrophobic association is driven by the hydrophobic BA‐A units on the P(MAA‐BA‐A) chains and the hydrophobic tails of the surfactant CTAB. These BA‐A units repel water molecules, leading to the formation of hydrophobic clusters. An increase in DES content reduces the water content, which impairs hydrophobic association and results in a lower cross‐linking density. Although hydrogen bonding and electrostatic interactions are strengthened, the overall network integrity is compromised due to the loss of physical junctions, ultimately causing structural collapse and degraded performance. Therefore, based on a comprehensive evaluation of the mechanical parameters, the gel with optimal composition was selected for subsequent studies. As illustrated in Figure 3g, this hydrogel can fully restore its original shape even when stretched to 9 times its initial length, subjected to lateral stretching, or punctured by sharp forceps. The retention rates of stress and strain are 76.1% and 86.2%, respectively (Figure S4). This further confirms its excellent mechanical strength.

2.4. Structural Evolution and Energy Dissipation Mechanism of Hydrogels

To investigate the internal network disruption and energy dissipation mechanism of the gel, cyclic loading‐unloading tests were performed at room temperature with a fixed stretching rate of 50 mm/min under varying strains (Figure 4). To evaluate the mechanical behavior under different deformation levels, the hydrogel was subjected to cyclic stretching at strains of 50%, 100%, 200%, 300%, 400%, and 500%. As shown in Figure 4b–d, the toughness and dissipated energy of the hydrogel remained highly stable and gradually increased across all strain levels [40]. From 50% to 500% strain, the toughness increased by more than 40‐fold, while the dissipated energy rose by approximately 77‐fold. The hydrogel also exhibited a resilience efficiency of up to 84%, even after 100% strain. This excellent recovery is attributed to the rapid dissociation of hydrophobic interactions, π–π stacking, and hydrogen bonds within the hydrophobic domains under stress, along with effective energy dissipation via hydrogen bonds and electrostatic interactions between flexible polymer chains. Upon release of the external force, the micellar structures rapidly reassemble, allowing the gel to recover its original state and thereby endowing it with high toughness and elastic recovery. To further examine the evolution of the hydrogel network structure, cyclic tests were conducted at 25% strain intervals. As illustrated in Figure 4e, as the strain increased progressively from 25% to 450%, the hysteresis loops remained relatively small but showed a gradually increasing trend. This indicates that the disruption of the internal network precedes macroscopic failure and becomes more pronounced with increasing elongation. Moreover, the noticeable overlap between consecutive hysteresis loops demonstrates the hydrogel's excellent self‐recovery capability. The relationship between dissipated energy (U _hys) and strain is plotted in Figure 4f. The continuous increase in U _hys with rising strain suggests the absence of significant defects in the internal network structure, which is consistent with the observations from SEM results (Figure 2g–i).

FIGURE 4.

Mechanical and rheological performance of the P(MAA‐BA‐A) eutectogel. (a) Tensile recovery schematic. (b,c) Cyclic loading and energy dissipation at varying strains. (d) Resilience. (e,f) Cyclic performance at incremental strains. (g,h) Strain‐ and frequency‐dependent viscoelasticity. (i) Thermal mass retention. (j) Fatigue resistance over 1000 cycles.

To elucidate the flow characteristics and structural behavior of the material, dynamic rheological tests were conducted. Initially, by fixing the frequency and varying the strain, the hydrogel was ensured to remain within the linear viscoelastic region for measurement. As shown in Figure 4g, with increasing strain, both the storage modulus (G′) and loss modulus (G″) of the hydrogel containing the deep eutectic solvent (DES) were higher than those of the DES‐free hydrogel, indicating that electrostatic interactions significantly enhance the elastic behavior of the material. The crossover point of G′ and G″ for eutectogel occurred at 44.3% strain and 311.6% strain, respectively, suggesting enhanced hardness and a reduced maximum bearable strain, which aligns with the trend observed in the mechanical performance. As illustrated in Figure 4h, within the linear viscoelastic region, G′ consistently exceeded G″, confirming the gel's solid‐like elastic behavior [41]. Overall, the dynamic mechanical and rheological results demonstrate that the multiple physical interactions within the system endow the gel with excellent fatigue resistance and dynamic responsiveness, supporting its potential as a wearable sensor for human motion monitoring. Furthermore, the non‐volatile nature of the DES, combined with the abundant hydrogen bonding and electrostatic interactions formed with the polymer chains, effectively locks the DES within the gel and enhances its moisture retention capability. As displayed in Figure 4i, the mass retention of the DES‐containing gel was superior to that of the pure hydrogel across different temperatures. This improved stability also ensures the structural integrity of the network during long‐term cyclic loading. As clearly shown in Figure 4j, even after 1000 loading‐unloading cycles, the stress retention remained close to 100%, underscoring the outstanding fatigue resistance of the eutectogel. Subsequently, the gel was photographed before and after 1000 cycles. Under the same fixture gap (13.87 mm), no significant deformation of the gel was discernible after 1000 cycles, indicating excellent fatigue resistance (Figure S5). This is mainly attributed to the hydrophobic association crosslinking points and abundant physical interactions such as hydrogen bonds and ionic bonds within the system. These interactions effectively disperse stress and stabilize deformation, thereby exhibiting excellent fatigue resistance [42].

2.5. Sensing Performance of Eutectogel

The deep eutectic gel exhibits excellent mechanical properties, remarkable flexibility, and outstanding fatigue resistance. Unlike water, the DES itself is ionically conductive, eliminating the need for additional salts. The gel demonstrates a conductivity of 8.6 mS·cm⁻¹ at room temperature (Figure S6), making it suitable for use as a resistive strain sensor [43]. When integrated into an LED circuit, the gel causes the bulb's brightness to vary with deformation (Figure S7), which can be attributed to the narrowing of ion transport pathways and the prolonged migration distance under strain, leading to increased resistance [44]. After adding DES, the gel has a wider range of application temperatures. This is mainly due to the hydrogen bond interaction between DES and water, which breaks the ordered arrangement of water molecules, making the gel less prone to freezing at low temperatures and less likely to lose water at high temperatures, thus ensuring that the gel maintains good performance within a wide temperature range [45]. As shown in Figure S8, the eutectic gel has good conductivity in the temperature range of −20°C to 80°C. The conductive circuit maintains stable LED brightness across a broad temperature range, and the gel retains excellent mechanical properties both at room temperature and −20°C (Figure S9). The strain sensitivity of the eutectogel, quantified by the gauge factor (GF), is calculated using the formula GF = (ΔR/R₀)/ε, where R₀ and R represent the initial and real‐time resistance, respectively. The GF values are determined to be 0.38 in the 1%–9% strain range, 0.47 for 10%–90% strain, and 0.69 for 100%–400% strain. More importantly, the sensor delivers highly stable and reliable resistance signals. Repeated stretching to identical strain levels yields nearly identical resistance responses (Figure 5a–f). The sensor's performance under different strain rates was also investigated. As shown in Figure 5g, it generates distinguishable and consistent signals under both slow (50 mm·min⁻¹) and high strain rates (100–400 mm·min⁻¹), highlighting its rapid response and high sensitivity.

FIGURE 5.

Sensing performance of the P(MAA‐BA‐A) deep eutectic gel under various tensile strains. (a–c) Relative resistance change (ΔR/R₀) and (d–f) the corresponding gauge factor (GF) in different strain ranges: 1%–9%, 10%–90%, and 100%–400%. (g) Relative resistance changes at a fixed 50% strain under different stretching rates. (h) Cyclic stability test over 10 000 loading‐unloading cycles at 50% strain.

To evaluate its potential as a wearable motion detector, a gel patch was attached to a finger to monitor bending movements (Figure S10). Bending the finger stretches the gel, increasing its resistance, and the ΔR/R₀ value rises with the bending angle. Consistent signals were obtained when the finger was bent to the same angle at different speeds, confirming the sensor's high reliability [46]. Similarly, the gel accurately tracked motions of other body parts, including the elbow, wrist, knee, and neck. Its high sensitivity also enables the detection of subtle muscle movements such as smiling, frowning, and swallowing. When attached to a volunteer's corresponding areas (Figure S11), the sensor captured distinct signal patterns for different micro‐expressions and swallowing actions with excellent repeatability. Given that practical applications involve repeated loading–unloading cycles, the cyclic stability of the strain sensor was evaluated. As shown in Figure 5h, the sensor exhibits highly reproducible signals over 10 000 consecutive cycles at 50% strain, attesting to its exceptional cyclic durability. In addition, no significant signal drift was observed during constant strain loading, indicating that the eutectogel‐based sensor responds rapidly and generates stable signals, making it highly suitable for real‐world applications. Compared with other hydrogel sensors, in this study, the gel sensors were subjected to strain tests of 1%–9%, 10%–90%, and 100%–400%. They demonstrated excellent detection ranges. Moreover, the deep eutectic solvent (DES) endowed the gel with good electrical conductivity and freeze resistance. Therefore, this sensor exhibited outstanding comprehensive performance in detecting strain, freeze resistance, and electrical conductivity (Figure S12 and Table S3).

2.6. Electrochemical Properties of the Eutectogel‐Based Supercapacitors

Based on its excellent mechanical properties, high ionic conductivity, and superior environmental stability, the binary‐solvent eutectogel shows great potential as an electrolyte for flexible supercapacitors [47]. An integrated supercapacitor was assembled by sandwiching the eutectogel electrolyte between two activated carbon/carbon cloth electrodes. Hydrogen bonds formed between the ─COOH groups on the polymer chains and the ─OH/─COOH groups on the activated carbon electrodes result in strong interfacial adhesion, effectively reducing the ion transport resistance at the electrolyte/electrode interface [48]. Electrochemical impedance spectroscopy (EIS) was employed to evaluate the interfacial contact resistance. As shown in Figure 6a,b, the eutectogel‐based supercapacitor exhibits a low interfacial resistance of 2.15 Ω, significantly lower than that of the DES‐free hydrogel supercapacitor (17.01 Ω), indicating that the binary solvent system facilitates faster ion transport. The electrochemical performance of the supercapacitor was further investigated by cyclic voltammetry (CV) and galvanostatic charge–discharge (GCD) tests. The CV curves measured at different scan rates (Figure 6c) maintain a nearly rectangular shape even at high rates, suggesting rapid electrochemical response and excellent capacitive behavior. The GCD profiles collected at current densities ranging from 1.1 to 5.5mA·cm⁻² exhibit nearly symmetric triangular shapes (Figure 6d), which confirms a highly reversible charge–discharge process at the electrode‐electrolyte interface. Furthermore, the device delivers a high areal capacitance of 495 mF·cm⁻², the energy density and power density are 68.75 µWh·cm⁻² and 550 µW·cm⁻², respectively [49]. The electrochemical stability of energy storage devices under mechanical deformation is a critical parameter for practical applications.

FIGURE 6.

Supercapacitor performance of the P(MAA‐BA‐A) eutectogel. (a) Device schematic. (b) EIS curves of eutectogel at different DES content. (c) CV curves measured under different scan rates. (d) GCD curves at different current densities. (e) Bending stability. (f,g) Temperature‐dependent GCD responses and capacitance retention. (h) Series/parallel operation.

Owing to the outstanding flexibility of eutectogel and its strong adhesion to the carbon electrodes, the supercapacitor maintains excellent structural integrity without delamination or interlayer slippage even under bending, overcoming the limitations of conventional hydrogel‐based devices. When bent to different angles (45°, 90°, and 180°), the supercapacitor shows no significant degradation in electrochemical performance (Figure 6e). Thanks to its wide operating temperature range, the device also exhibited stable GCD profiles between −20°C and 80°C (Figure 6f). Even at −20°C, it maintained a capacitance retention rate of 67.7% (Figure 6g), along with bending resistance and the capability to power electronic devices, such as illuminating an LED (Figure S13). (Figure 6h). Remarkably, the deep eutectic gel‐based supercapacitor demonstrates outstanding cycling stability, retaining 94.8% of its initial capacitance after 1400 charge–discharge cycles (Figure S14). To meet the voltage requirements of practical applications, supercapacitor devices can be connected in series or parallel. When two supercapacitors are connected in series, the amount of charge stored in each supercapacitor is equal, and the total voltage is the sum of the voltages of the individual cells, thereby increasing the voltage to 2.0 V. When connected in parallel, the terminal voltage of each cell is the same, and the total charge is the sum of the charges of the individual cells, thus the equivalent capacitance is expanded to 160%. (Figure 6f), demonstrating excellent scalability for high‐voltage devices. Compared with other reported materials, our eutectic gel exhibited distinct advantages in electrochemical cycling stability, mechanical cycling stability, and areal capacitance (Figure S15, Table S2). These outstanding properties underscore its great potential as an electrolyte for flexible supercapacitors.

2.7. Deep Learning‐Assisted Language Recognition System

Deep learning‐assisted speech recognition systems have garnered significant research interest for enabling personalized and intelligent human–computer interaction [50]. As illustrated in Figure 7, a eutectogel sensor integrated with a deep learning model was implemented to construct an interactive system capable of recognizing complex multilingual speech. Specifically, when a volunteer vocalized phrases in different languages, the electrical signals captured by the sensor attached to the throat were transmitted to an algorithmic model for training and testing. Furthermore, we tested volunteers of different genders, ages, and accents to verify the influence of these variables on the electrical signals. Different volunteers were asked to say “Ni Hao” (Hello). It was observed that factors such as different volunteers and genders introduced certain variations in the signals; however, these variations did not affect the overall signal trend. This confirms that the gel material exhibits good accuracy for speech recognition and can be generalized for use among diverse populations (Figure S16). With increasing training iterations, the training loss gradually decreased while the test accuracy improved, achieving the goal of robust signal recognition. To evaluate the performance of the designed preprocessing and recognition system along with the hybrid lightweight algorithm, tests were conducted on electrical signals corresponding to the word “Hello” spoken in five languages: Chinese, English, French, Japanese, and Korean (Figure 7a,b). The acquired signal data were used to train a deep learning model comprising a convolutional neural network (CNN) and a gated recurrent unit (LSTM) [51, 52]. For each language, 50 signal data samples were acquired and preprocessed, and the dataset was divided into training set, validation set, and test set at a ratio of 8:1:1. In this framework, the CNN extracts discriminative features from the input signals, while the CNN‐LSTM captures sequential dependencies in the data, as depicted in Figure 7c,d. Although the electrical signals corresponding to different languages exhibit high similarity, they can still be distinguished based on characteristic peak intensities, which serve as the foundation for building the classification system.

FIGURE 7.

Deep learning‐assisted multilingual speech recognition using the P(MAA‐BA‐A) eutectogel sensor. (a) Schematic of the system for recognizing the word “Hello” in five different languages: Chinese, English, French, Korean, and Japanese. (b) Logic diagram of the CNN‐LSTM model used for data classification. (c,d) The test loss and accuracy of the CNN‐LSTM model after each training epoch. (e) Confusion matrix and overall prediction accuracy of the model.

Throughout the testing process, prediction parameters were determined from electrical signals transmitted within a 5‐s window. The model provides relevant feature values for each recognition prediction. As shown in Figure 7c,d, after several training epochs, the recognition accuracy of the proposed model gradually attains saturation. The system convergence approaches 1, indicating that the model is reaching a stable state. Figure 7e presents the confusion matrix comparing the ground truth and predicted values during speech recognition. The average prediction accuracy reaches 94.21%, demonstrating the system's capability to accurately distinguish between different languages. This outcome provides a valuable reference for future real‐world applications in intelligent sensing.

3. Conclusion

In summary, this study presents a multifunctional eutectogel fabricated via a facile one‐step thermal polymerization strategy, which integrates high strength, toughness, fatigue resistance, and a broad operating temperature range. The incorporation of hydrophobic domains and a multi‐physical crosslinking strategy enables rapid dissociation and reconstruction of the network under external force, endowing the material with remarkable elasticity and anti‐fatigue performance. The eutectic network formed in the binary solvent system ensures reliable performance across a wide temperature range (−20°C to 80°C), a rapid response speed of 400 mm/min, and excellent stability against fatigue. Owing to the intrinsic ionic conductivity of the DES, the hydrogel exhibits an excellent conductivity of 8.6 mS·cm⁻¹. As a motion sensor, the eutectogel demonstrates high sensitivity, capable of monitoring subtle facial expressions as well as movements from various body parts. Moreover, it retains reliable operation over more than 10 000 electrical test cycles, underscoring its suitability for long‐term health monitoring applications. In addition, the material functions effectively as a supercapacitor for energy storage, maintaining excellent performance over a broad temperature range, with a high areal capacitance of 495 mF·cm⁻² and a capacitance retention of 94.8% after 1400 charge–discharge cycles. By analyzing resistance signals, a deep learning‐assisted platform was further established using the hydrogel system to accurately recognize different languages, validating its potential as an advanced intelligent speech recognition system. Therefore, this work provides a versatile platform that advances epidermal electronics and paves the way for next‐generation wearable devices, including e‐skin, supercapacitors, and language recognition systems for human–computer interaction.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll73889‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author on reasonable request.