LCST-phase-separated porous liquid metal-filled hydrogel actuators with fast electro-response, enhanced strength, and low electric field

Abstract

Electro-responsive hydrogel actuators (ERHAs) are promising candidates for soft robotics due to their capability for exhibiting large, reversible deformations. However, their application potential is constrained by the requirement for high driving electric field strength (E), insufficient mechanical robustness, and slow actuation response. Here, to simultaneously address these limitations, we design an ionic hydrogel with integrated liquid metal (LM) and thermoresponsive LCST behavior. The porous architecture is readily constructed by the LCST-induced phase separation process. LM inclusion not only enables sensitivity to low E but also reinforces mechanical properties of the otherwise weakened porous hydrogel. The resulting actuator achieves a large bending angle of 88.1° within 32 seconds under a low electric field of 0.25 V mm⁻¹. This represents the fastest electro-response reported to date among ERHAs operating below 1 V mm⁻¹, a threshold widely recognized as safe for human exposure. Furthermore, we demonstrate its versatility in executing diverse underwater tasks, including object manipulation, encapsulation, and directional locomotion. This facile yet effective strategy for constructing mechanically robust, fast-response hydrogel composites offers new avenues for the development of next-generation soft robotic systems.

---

New concepts

Electro-responsive hydrogel actuators (ERHAs) are promising for soft robotics and biomedical devices. However, achieving fast response, mechanical robustness, and low electric field strength (E) actuation in one system remains a key challenge. Currently, no design concept exists that can simultaneously improve all three properties. Here, we propose a conceptually new material design strategy that integrates liquid metal (LM) microparticles into a porous ionic hydrogel matrix. The porous structure, formed via LCST-induced phase separation, promotes rapid ion transport, while LM microparticles strengthen the hydrogel network through Ga³⁺–carboxylate coordination and enhance low-E responsiveness via Ga³⁺ and dynamic electrical double layers. This strategy enables concurrent improvements in actuation speed, mechanical strength, and responsiveness to low E, as demonstrated in representative LM-hydrogel systems. The concept provides a generalizable framework for developing next-generation ERHAs with enhanced performance, making them well-suited for diverse soft robotic applications.

Introduction

Hydrogel actuators, owing to their inherent softness, high water content, and biocompatibility, have emerged as promising candidates for applications in soft robotics, artificial muscles, and stimuli-responsive systems.^1–3 Among various actuation mechanisms, such as thermal,^4,5 pH,^6,7 and light responses,^8,9 electro-responsive hydrogel actuators (ERHAs) have drawn increasing attention due to their ability to respond rapidly and remotely under mild conditions with simple electrical inputs. These hydrogels leverage ionic migration and electric field-induced osmotic pressure imbalances to generate deformation, offering excellent controllability and repeatability.^10–13 Harnessing these advantages, ERHAs have been widely explored in diverse fields such as soft robotics,^10,14,15 artificial muscles,¹² biomedical actuators,^16,17 drug delivery systems,^18,19 and underwater manipulation.^20,21

For ERHAs to be promising in practical applications such as biomedical and wearable devices, advancements are needed in three critical areas: achieving faster actuation kinetics, reducing the required driving voltage, and improving mechanical robustness. However, the existing design strategies for ERHAs cannot simultaneously improve these three key aspects. For example, the most widely adopted strategy to accelerate actuation involves the introduction of porous structures,^{10,12,22–24} which enhance ion transport, swelling kinetics, and deformation dynamics. However, the formation of porous networks often compromises the mechanical properties of the hydrogel. An alternative approach to improve responsiveness and mechanical properties involves incorporating functional nanofillers such as carbon nanotubes (CNTs),²⁵ MXene,¹⁴ and graphene.²² These fillers not only facilitate faster ion migration but also enhance mechanical toughness through filler fracture and pull-out during deformation. Despite these advantages, these composite ERHAs still suffer from relatively slow actuation responses, with response times on the order of several minutes. Efforts to reduce actuation E,²⁶ including tuning crosslinking density,¹¹ adjusting monomer composition,¹³ modifying electrolyte ionic strength,²⁵ and employing wrinkled nanomembrane electrodes,¹⁵ have not been reported to enhance either response speed or mechanical properties. Consequently, achieving a synergistic improvement in response speed and mechanical properties, while simultaneously enabling lower E operation, remains a significant and unresolved challenge in the development of next-generation ERHAs.

In this work, we propose a simple yet effective materials design strategy to overcome the aforementioned limitations by developing a novel platform of LM-infused porous ionic hydrogels (LMHs). These hydrogels exhibit thermoresponsive LCST behavior, enabling thermally induced gelation without the need for external porogens or complex fabrication processes.^27–29 Upon heating above the LCST, the LM–polymer aqueous mixtures spontaneously form a highly porous network, achieving a porosity of up to 52% even with the incorporation of 5 wt% LM microparticles (LMMPs). The LMMPs, composed of an eutectic gallium–indium alloy, endow the system with three synergistic advantages: (i) enhanced actuation kinetics through improved ionic conductivity, (ii) in situ release of Ga³⁺ ions that reinforce the hydrogel matrix via dynamic coordination interactions, and (iii) increased electromechanical responsiveness, allowing for effective actuation under reduced E. The tensile strength of the hydrogel is increased by more than twice (from 31.4 kPa to 81.7 kPa) after introducing 2 wt% LMMPs. Under a low actuation E of 0.25 V mm⁻¹, the actuator achieves a large bending angle of 88.1° within 32 seconds, representing the fastest electro-actuation response among reported ERHAs operating below 1 V mm⁻¹. Moreover, the actuator remains functional at even lower E, reaching a bending angle of 30° under a low E of 0.125 V mm⁻¹.

As summarized in Table S1, unlike traditional design strategies for ERHAs that typically address only one or two of the aforementioned limitations, our proposed design concept enables simultaneous enhancement of hydrogel mechanical properties, response time, and responsiveness to low electric fields. Capitalizing on these multifunctional properties, we demonstrate the versatility of our composite system through three proof-of-concept underwater soft robotic applications: (1) an artificial limb capable of displacing a load twentyfold greater than its own weight, (2) a tentacle-like actuator for object encapsulation, and (3) a fully soft robot capable of directional walking over a distance of 14.9 mm in 31 seconds using asymmetric LMH legs. This integrated design approach, combining thermo-induced gelation and LM-induced ion coordination, offers a new pathway for developing high-performance hydrogel actuators tailored for next-generation soft robotic systems.

Results and discussion

Designing hydrogel networks

To fabricate highly porous, electrically driven hydrogel actuator composites, it is essential to rationally design the molecular structure of the ERHAs. In this study, we developed an ionic polyelectrolyte with LCST phase transition behavior. The ionic groups impart electro-responsiveness to the actuator by enabling mobile ion diffusion under an applied electric field. When the ionic polymer solution is heated above its LCST, the polymer chains undergo phase separation and precipitate from the solution. Subsequent solvent evaporation leaves behind pore spaces between the aggregated polymer chains. Furthermore, coordination bonds formed between the ionic groups and metal ions stabilize the porous structure upon cooling to room temperature.

Previously reported strategies for producing porous responsive hydrogel actuators typically rely on the incorporation of PEG macromolecules as pore-forming agents or the use of freeze–thaw cycles to induce porosity.^30,31 However, these approaches are relatively complex, requiring either the removal of pore-forming agents or multiple freeze–thaw cycles over several days (typically at least three). In contrast, the method of utilizing LCST-induced phase separation to generate a highly porous structure is significantly more straightforward, involving only heating the polymer solution above the LCST followed by drying (Fig. 1A). Notably, the drying process also increases the density of noncovalent crosslinkers, which may enhance the mechanical properties of the hydrogel.

Fig. 1 Designing an ionic hydrogel with integrated liquid metal (LM) and thermoresponsive LCST behavior. Schematic representation of (A) the mechanism of phase separation and (B) the preparation of the composite electro-responsive hydrogel actuators.

To fulfill the aforementioned requirements, methacrylic acid (MAA) was selected as the monomeric unit due to its ionizable carboxylic acid groups, which are essential for enabling electrical actuation in hydrogels (Fig. S2 and Fig. 1B). With a pK_a of 4.65, MAA exists predominantly in its anionic form under neutral conditions (pH ≈ 7), thereby providing ionic conductivity. To impart thermoresponsive properties, oligo(ethylene glycol) methacrylate (OEGMA) was copolymerized with MAA, enabling the resulting copolymer to exhibit LCST behavior. Although the homopolymer of OEGMA typically displays an LCST around 90 °C, the incorporation of MAA can lower the LCST due to acid–ether hydrogen bonding interactions which could avoid the formation of bubbles during the drying process. According to previous reports, the LCST of the poly(MAA-co-OEGMA) copolymer is ∼42 °C.²⁸ Upon heating above the LCST, the initially transparent solution turned opaque (Fig. S3). Additionally, the MAA unit is capable of forming strong coordination bonds with trivalent metal ions such as Fe³⁺, which serve as dynamic crosslinkers to stabilize the porous network by generating densely coordinated ionic domains.

Another key aspect of our design strategy is the incorporation of LMs, which not only act as crosslinkers to strengthen the matrix but also act as nanofillers to facilitate ion transport. Unlike commonly employed nanofillers such as graphene and CNTs, LMs can be readily dispersed in aqueous media to form nanoscale droplets without adding surfactants.³² This inherent dispersibility makes them highly compatible with aqueous polymer systems, thereby enabling homogeneous integration into the hydrogel matrix.³³ Additionally, LMMPs can release multivalent ions (Ga³⁺) and form an electrical double layer (EDL), which contributes to ionic transport and improved deformation amplitude.^20,34 Moreover, the incorporation of LMMPs markedly enhances the electro-responsiveness of the hydrogel composite, particularly under low electric field strengths (details are discussed later).

Unlike previous studies that primarily characterized the electrical stimulus using applied voltage, this work employs electric field strength (E), calculated by dividing the applied voltage by the distance between the two electrodes. This approach provides a more accurate and physically meaningful measure of the actual electrical stimulus experienced by the actuator. Voltage alone does not account for variations in electrode spacing, which can significantly affect the local field intensity and, consequently, actuation performance. By using E, we enable a more reliable comparison across different studies and geometries, ensuring that the reported actuation behavior reflects intrinsic material performance rather than device-dependent parameters.

It should be noted that in our design, all of the soft actuator components are biocompatible. Both poly(methacrylic acid) (PMAA) and poly(oligo(ethylene glycol) methacrylate) (POEGMA) homopolymers have been widely investigated and reported to exhibit excellent biocompatibility in biomedical applications.^35–37 Gallium particles themselves have demonstrated intrinsic biocompatibility, largely attributed to the formation of a stable gallium oxide skin that passivates the metal–solution interface and mitigates ion release.^38,39 Together with the low electrical driven field, our designed hydrogels are highly attractive for future biomedical applications.

Improving mechanical properties of porous ionic hydrogels

Our design concept, which integrates ionic hydrogels exhibiting LCST behavior with LMMPs, not only facilitates the formation of a porous structure but also significantly enhances the mechanical properties. This strategy directly addresses the common trade-off between actuation responsiveness and mechanical robustness in hydrogel actuator systems. To validate this, we first investigated the microstructure of LM-ionic hydrogel composites with varying LM contents. The SEM image shows that the phase-separated hydrogels manifest as a percolating porous structure with pore dimensions in the micrometer scale (Fig. 2A–D and Fig. S4). Herein, the authors named the LM-filled hydrogel with 2 wt% LMMPs LM2. LM0 is denoted as pure poly(MAA-co-OEGMA). At lower LM concentrations (LM0, LM2, and LM5), the hydrogels exhibit a relatively open and interconnected porous network with small, consistent pore sizes. Notably, the average pore diameter (∼3.6 μm) remains nearly unchanged among these three samples. However, at higher LM loadings (LM10 and LM20), the pore structure becomes more sparsely distributed, with larger pore diameters of 5.4 μm and 7.1 μm, respectively. This trend is quantitatively reflected in Fig. 2E, where the porosity gradually decreases with increasing LM content (from 79.1% to 19.1%). When it comes to LM40, there are no pores observed (Fig. S5). These observations suggest that moderate LM inclusion maintains the structural integrity of the porous network, whereas excessive LM content disrupts the homogeneity of the gel structure, possibly due to aggregation or phase separation, leading to reduced porosity and enlarged pores (Fig. 2E). The highest porosity of 79.1% achieved in our hydrogel samples is comparable to that of previously reported highly porous hydrogel systems.⁴⁰ While several ERHAs with porous structures have been reported, their porosity values were not quantified.^13,22,26,41 To the best of our knowledge, this is the first study to report the porosity of porous ERHAs.

Fig. 2 Physicochemical properties of LM-ionic hydrogel composite actuators. The SEM images of the cross-section of (A) LM0, (B) LM2, (C) LM5, and (D) LM20. (E) The bar chart of porosity and the curve of pore diameter with different LM content. (F) Stress–strain curves and (G) toughness and Young's modulus column charts of hydrogels with different LM content. Sample size n = 3.

We further investigated the effect of metal coordination and LM crosslinking on the mechanical properties of the networks. Poly(MAA-co-OEGMA) without the Fe³⁺ and LM crosslinkers is soluble in water and couldn’t form hydrogels, which could be attributed to the low stability of hydrogen bonding in water. The H-bond network results in the fragile property of the polymer, which makes it hard to hold to conduct a tensile test. When Fe³⁺ is introduced into the polymer network, the tensile strength, toughness, and Young's modulus of the hydrogel were measured to be 32.2 kPa, 27.1 kJ m⁻³, and 50.3 kPa, respectively (Fig. 2F and G). This can be attributed to the high coordination strength of the carboxyl–Fe³⁺ coordination complexes.⁴² The mechanical properties of this hydrogel, only crosslinked by metal coordination bonds, are comparable to those of most hydrogel actuators.^43,44

To further enhance the mechanical properties of the formed porous hydrogels, LMMPs were introduced into the network system to enable the formation of hydrogen bonds between the polymer chain and the surface of LMMPs. As shown in Fig. S6, Fourier transform infrared spectroscopy (FTIR) spectra of LM0, LM2, and LM5 were analyzed to investigate the interaction between LMMPs and the hydrogel network. As the LMMP content increased from 0 to 2 wt%, the O–H stretching band exhibited a slight red shift from 3375.5 cm⁻¹ to 3373.5 cm⁻¹, indicating an increase in hydrogen bonding interactions. Meanwhile, the COO⁻ band slightly red-shifted from 1645.8 to 1642.4 cm⁻¹, implying potential coordination between carboxyl groups and Ga³⁺ ions from the LMMPs.^45,46 Increasing the mass fraction of LM from 0 to 2 wt% resulted in a significant enhancement in the mechanical performance of the hydrogel. Specifically, the tensile strength increased from 32.2 kPa to 81.7 kPa, while the Young's modulus was elevated from 50.3 kPa to 86.6 kPa, indicating improved stiffness of the polymer network due to LMMP-induced physical cross-linking (Fig. 2F and G). Durability is also an important indicator for evaluating the practicality of an actuator.^47,48 LM2 hydrogels were subjected to 50 consecutive stretching–releasing cycles at 40% strain (Fig. S7), during which the stress–strain curves showed good overlap without obvious deterioration, confirming the cycle stability of the hydrogel. The toughness of the hydrogel also improved substantially, increasing from 27.1 kJ m⁻³ to 51.5 kJ m⁻³. Although the introduction of LMMPs slightly compromised the elongation at break, the LM5 and LM10 samples still maintained sufficient stretchability with strains of 80.4% and 46.6%, respectively. Notably, the Young's modulus increased progressively with LM content, reaching 149.4 kPa for LM5 and 218.7 kPa for LM10. In contrast, the toughness exhibited a decreasing trend beyond LM2, dropping from 40.8 kJ m⁻³ for LM5 to 21.6 kJ m⁻³ for LM10, indicating a trade-off between stiffness and energy absorption capacity. These results collectively demonstrate that moderate incorporation of LMMPs enhances both the strength and energy dissipation capacity of the hydrogel.

To elucidate the influence of the porous structure on the mechanical properties of the hydrogels, non-porous samples were prepared by removing water under vacuum at room temperature, which is well below the LCST. The non-porous LM0 hydrogel exhibited a fracture strain of 356% and a tensile strength of 79 kPa, whereas its porous counterpart showed significantly reduced values of 135% and 32 kPa, respectively (Fig. S8). A similar trend was observed for LM2. The non-porous LM2 hydrogel displayed a fracture strain of 182% and a tensile strength of 118 kPa, in contrast to the porous LM2 sample, which exhibited only 124% tensile strain and 82 kPa tensile strength. This mechanical degradation can be attributed to the presence of pores within the hydrogel network, which act as structural defects that facilitate stress concentration and promote crack propagation, ultimately compromising both tensile strength and extensibility.^49,50 Together with the increase from ∼32 kPa to ∼82 kPa when adding LM to the porous network, these data support that LM primarily compensates for porosity-induced weakening by reinforcing the network, while the porous architecture and drying-induced densification account for the elevated strength but lower elongation.

The LMH exhibited rapid swelling behavior, as evidenced by the steep initial increase in the swelling ratio (SR) within the first 60 minutes (Fig. S9). The hydrogels reached equilibrium swelling within approximately 1–2 hours, with the final SR strongly dependent on LM content. Specifically, LM0 displayed the highest equilibrium SR of ∼850%, while the SR progressively decreased with increasing LM loading, reaching ∼450% for LM2 and ∼350% for LM40. This trend can be attributed to the reduced porosity at higher LM contents, which lowers the water uptake capacity. Nevertheless, all hydrogels showed a fast swelling response, indicating that the porous internal structure still facilitates efficient water diffusion and penetration into the polymer matrix.

Fast response and lower driving electric field strength

To meet the requirements of diverse device applications, particularly in biomedical systems, ERHAs must exhibit rapid response under E of ≤1 V mm⁻¹, a threshold generally considered safe for human exposure.⁵¹ However, the previously reported ERHAs sensitive to low E (≤1 V mm⁻¹) exhibit slow response (≤1° s⁻¹).^{11,13,20,22,23,25} Our design concept aims to address this issue by introducing LMs into porous ionic hydrogels. The response of hydrogel actuators under milder stimuli will be significantly enhanced by the synergistic effect of a highly porous structure and introduced LM both of which will accelerate ion diffusion kinetics. Furthermore, the deformation degree under the low E will also be enhanced owing to the presence of LM which would increase the number of ions and the cation drift. As shown in Fig. 3, we achieved a large bending angle of 88.2° and a fast bending speed of 2.76° s⁻¹ under a low electric field strength of 0.25 V mm⁻¹. Even at a lower field strength of 0.125 V mm⁻¹, the actuator still exhibited a bending angle of 23.2° with a speed of 0.29° s⁻¹, demonstrating excellent sensitivity to mild electric stimuli (Fig. S13).

Fig. 3 Electro-responsive bending deformation of LMH. (A) Definition of the bending angle (θ) of the ERHAs. (B) The optical images of the continuous process of LM2 bending towards the cathode (bidirectional). The scale bar is 10 mm. (C) Bending angle of the LMH with different LM content under an applied E of 0.25 V mm⁻¹ in a 0.2 M NaCl solution. (D) Bending angle of LM2 as a function of the applied E in a 0.2 M NaCl solution. (E) Bending angle of LM2 as a function of the concentration of NaCl solution under an E of 0.25 V mm⁻¹. (F) and (G) Comparison of driven electric field strength (E), bending angle (θ), and bending speed performance of our LMH with other ERHAs.

In this ERHA, the actuation mechanism is primarily governed by asymmetric osmotic swelling induced by ionic migration under an applied electric field (Fig. S10A).^11,17,22 The hydrogel network, composed of poly(MAA-co-OEGMA), contains abundant carboxylic acid groups, which ionize in aqueous environments to form negatively charged polyanionic chains. These fixed charges are counterbalanced by mobile cations (including Fe³⁺ and H⁺) within the hydrogel matrix. Upon the application of an electric field in NaCl solution, the mobile cations (Na⁺, as well as H⁺ which dissociated from the carboxyl group) migrate toward the cathode, while the mobile anion (Cl⁻) migrates toward the anode, leading to an ionic imbalance across the hydrogel. The simulation results shown in Fig. S10B reveal distinct ion migration behavior under an applied electric field. Specifically, Na⁺ and H⁺ ions migrate toward and accumulate near the cathode, reaching concentrations of approximately 0.10405 mol L⁻¹, while Cl⁻ ions migrate toward the anode, with a final concentration of approximately 0.10138 mol L⁻¹. This results in a higher ionic concentration, and thus higher osmotic pressure on the anode-facing side compared to the cathode side. Consequently, the anode-facing region undergoes greater swelling, and the hydrogel bends toward the cathode.

The effects of electro-response actuation of LMH were fully investigated, which include variations in LM content, voltage, and NaCl concentration.⁵² The hydrogel was fixed at one end and deformed under an applied electric field. The bending angle (θ) is measured between the vertical reference line and the line connecting the free end and the fixed end of the hydrogel (Fig. 3A). Fig. 3B and Movie S1 illustrate the electrically actuated bending deformation of LM2. As shown in Fig. 3C, the θ increased more rapidly and reached higher values as the LM content in the LMH increased from 0 to 2 wt%. This enhancement is attributed to the increased concentration of mobile Ga³⁺ ions released from LM droplets and the formation of a more responsive microstructure. Electrochemical impedance spectroscopy (EIS) results reveal that their incorporation enhances the dielectric response of the hydrogel (Fig. S11). A higher dielectric constant reduces ion–ion coulomb interactions and facilitates ion dissociation and migration, while interfacial polarization around LM microdomains promotes local field concentration.²⁴ On further increasing the LM content, the θ and response speed decrease because enhanced stiffness provided by excessive LMMPs hinders the bending movement of the LMH (Fig. 3C and Fig. S12A). Therefore, a small amount of LMMPs is beneficial to the electro-response actuation of LMH (more details in SI S1 and S2).

Fig. 3D shows the effect of E on the bending actuation of LM2 in 0.2 M NaCl solution. Increasing the applied E from 0.125 V mm⁻¹ to 0.375 V mm⁻¹ resulted in a significant enhancement in bending amplitude and speed. The LM2 hydrogels exhibited rapid bending deformation in which complete bending (88.1°) was performed within 32 s under applied E of 0.25 V mm⁻¹. In contrast, the bending angle was about 23.2° under an applied E of 0.125 V mm⁻¹ at the period of 85 s. The higher electric field intensified ion migration, thereby increasing the osmotic pressure differential across the hydrogel film and accelerating the bending motion.

Notably, the incorporation of LMMPs markedly improves the electro-actuation performance of LMH even under low electric fields, with a 37% increase in bending angle and a 111% increase in bending speed relative to the pure LMH (Fig. S13). This enhancement could be attributed to the formation of dynamic EDLs around the LM microdroplets, which could locally increase the concentration of mobile ions and facilitate more efficient ion migration.³⁴ Such effects promote faster cation transport, thereby lowering the actuation threshold and enable responsive behavior under sub-1 V mm⁻¹ conditions (SI S1 and S2).

The influence of the ionic strength of electrolyte solutions on the bending behaviors of LM2 was also determined (Fig. 3E). The surrounding electrolyte concentration markedly influenced the actuator's performance. Hydrogels immersed in higher NaCl concentrations (up to 0.4 M) exhibited faster and greater bending deformation, which is reflected in a bending angle of 82.9° and a bending speed of 3.19° s⁻¹ (Fig. 3E and Fig. S12B). This trend is likely due to improved ionic conductivity and enhanced ion exchange rates between the hydrogel and external solution, which amplifies the internal osmotic pressure imbalance. However, excessive applied voltage and electrolyte concentration lead to the generation of a large number of bubbles at the cathode and anode, which seriously affect the observation of LMH bending (Fig. S14). Although the applied voltage and electrolyte concentration were reduced to mitigate the effect of bubbles (applied E of 0.25 V mm⁻¹ and NaCl solution of 0.2 M), the LMH still exhibited a large bending angle and rapid speed (a bending angle of 88.2° and a bending speed of 2.76° s⁻¹, Fig. S12C and S15).

Cyclic actuation performance of the LM2 hydrogel was evaluated under a periodically reversed electric field of 0.25 V mm⁻¹ in NaCl solutions of varying concentrations (0.05 M to 0.4 M), as shown in Fig. S16A–D. The LM2 actuator exhibited periodic and reversible bending in response to the polarity of the applied electric field. Specifically, the actuator bent toward the cathode during the positive phase and reversed direction under negative phase, demonstrating excellent polarity-sensitive deformation. As the NaCl concentration increased from 0.05 M to 0.4 M, the time required to complete one full bending cycle decreased from 215 s to 76 s, indicating enhanced actuation responsiveness. This improvement is attributed to increased ionic mobility and the more rapid establishment of osmotic pressure gradients across the hydrogel network.

Interestingly, under 0.05 M NaCl, the actuator displayed a gradual increase in bending angle over successive cycles. This phenomenon may be attributed to the progressive swelling of the hydrogel matrix under continuous electric field stimulation. One possible explanation is the leaching of Fe³⁺ ions from the network, particularly in low ionic strength environments where the scarcity of compensating cations may promote the dissociation of ionic crosslinking points. The resulting decrease in crosslinking density leads to increased water uptake and a softer network, thereby amplifying the deformation over time. In contrast, hydrogels actuated in higher NaCl concentrations (0.1–0.4 M) showed consistent bending angles across cycles without noticeable drift, suggesting improved ionic screening and more stable ionic crosslinking under high ionic strength conditions. To further assess durability, extended cyclic actuation tests of the LM2 hydrogel were performed using 0.2 M NaCl (Fig. S16E), where reversible bending was maintained over more than 20 cycles. In addition, inductively coupled plasma mass spectrometry (ICP-MS) analysis revealed that electroactuation led to detectable release of Ga species, with the bath concentration reaching 16.6 mg L⁻¹ compared to 1.01 mg L⁻¹ in the control (Table S2). Although this confirms partial leakage of LM into the surrounding medium, the absolute concentration remains far below reported acute toxicity thresholds of gallium salts and is consistent with previous studies highlighting the overall low toxicity and biocompatibility of Ga-based liquid metals.⁵³ These results highlight the hydrogel's capability for stable, reversible, and long-term bidirectional actuation, making it a strong candidate for applications in soft robotics, artificial muscles, and programmable electroactive systems.

As a result, LMHs with higher porosity exhibit shorter response times, larger bending angles, and faster actuation rates compared to other ERHAs (Fig. 3F and G and Fig. S17 and Table S3).^{11,13,16,20,22–25} The introduction of LMMPs not only enhances the crosslinking density and mechanical robustness of the hydrogel but may also influence the ionic migration dynamics, further contributing to the actuator's fast and reversible deformation behavior under low-voltage stimulation.

Soft robotic applications

The LMH exhibits diverse electro-actuation behaviors in underwater environments, enabled by its engineered porous architecture and the functional incorporation of LMMPs. Leveraging its rapid, polarity-sensitive bending and excellent underwater compatibility, the hydrogel can be integrated into a wide range of motion tasks, including force transmission, object manipulation, and autonomous locomotion. Fig. 4A and Movie S2 demonstrate the LMH strip functioning as an artificial soft arm to push a floating ball. Despite weighing only 60 mg, the hydrogel successfully displaces a 1200 mg ball, which is 20 times heavier than itself within 8 seconds under an applied electric field. The highly porous network facilitates rapid water uptake and swelling, while LMMPs enhance the mechanical robustness and electrical responsiveness required for effective actuation. To mimic biological functionality, a tentacle-like actuator is constructed using two LMH legs and a buoyant PU foam body. When voltage is applied, the legs undergo large-angle bending, enabling them to wrap around and grasp an object underwater (Fig. 4B and Movie S3). In Fig. 4C, an underwater soft robot is demonstrated using two LMH legs with different LM content (LM2 and LM5). As described previously, the LM2 leg with higher ionic mobility and swelling kinetics exhibits greater bending amplitude. Upon alternating the polarity of the electric field, the asymmetric deformation of the legs results in directional propulsion, driven by frictional interaction with the ground. This allows the robot to achieve a forward displacement of 14.9 mm within 31 seconds (Fig. 4E and Movie S4). These demonstrations collectively highlight the LMH's potential as a multifunctional soft actuator for underwater robotics, capable of rapid, reversible, and programmable motion.

Fig. 4 Applications of the LMH. (A) The LMH actuator functions as a soft arm to push a floating ball in water. (B) A bionic soft robot mimicking tentacle-like motion to wrap an object. (C) The optical photos of the locomotion of an underwater soft walking robot. The scale bars are 10 mm. (D) Schematic diagram of the soft robot. (E) Displacement curves of three parts of the soft robot.

Beyond the proof-of-concept demonstrations, the proposed fabrication route also offers potential for large-scale manufacturing and system integration. Heating aqueous formulations containing polymers, LM particles, and ions, followed by simple drying, provides a straightforward fabrication route that eliminates the need for complex templating or multistep treatments, offering clear advantages for large-scale manufacturing and scalability. In addition, the low driving electric field (≤0.25 V mm⁻¹) is compatible with portable power supplies and conventional electronic circuits, thereby facilitating seamless integration of the actuators into complex soft robotic architectures or biomedical devices. These attributes indicate that the present design strategy not only advances material performance but also holds promise for translation towards practical applications.

Conclusions

In summary, this work presents, for the first time, a strategy to overcome the longstanding trade-off between mechanical properties, response speed, and actuation stimulus in ERHAs by introducing a novel LM-based porous ionic hydrogel platform. Without relying on additional pore-forming agents or post-processing steps, the hydrogel spontaneously forms an interconnected porous network, which facilitates rapid ion transport and enhances actuation efficiency. The introduction of LMMPs significantly improves the mechanical strength and electrical responsiveness of the hydrogel, while maintaining a high porosity of 71.9%. Under a low E of 0.25 V mm⁻¹, the hydrogel actuator exhibits a record-high actuation capability coefficient among all existing ERHAs operating below 1 V mm⁻¹, a threshold widely recognized as safe for human exposure. Furthermore, the LMH enables diverse underwater robotic applications, including object pushing, adaptive grasping, and directional locomotion. These results highlight the synergistic role of structural porosity and functional fillers in boosting the actuation performance of hydrogel systems. Future studies will focus on systematically evaluating the biocompatibility of this composite system and exploring its potential in advanced biomedical and wearable applications.

Experimental

Materials

Gallium and indium (99.99%) were purchased from Changsha Santech Materials Co., Ltd, China. S-FOAM 50 PU foam was purchased from Barnes. Except gallium, indium, and S-FOAM, all chemicals were obtained from Sigma-Aldrich. To remove inhibitors, OEGMA (500 g mol⁻¹) and MAA were passed through a basic alumina column immediately before use. Azobisisobutyronitrile (AIBN, 12 wt%) was recrystallized from ethanol and dried overnight prior to use.

Synthesis of poly(MAA-co-OEGMA)

The procedure for preparation of well-defined poly(MAA-co-OEGMA) copolymers via free-radical polymerization is illustrated in Fig. S2. As a representative example, OEGMA 500 (1.9 mL), MAA (1.198 mL), and AIBN (1.68 mg) were dissolved in anhydrous N,N-dimethylformamide (DMF, 10.0 mL). The flask was sealed with a rubber septum and the solution was deoxygenated by sparging with nitrogen for 15 min. After polymerizing for 24 h at 70 °C, the reaction was cooled to room temperature and exposed to air. The viscous polymer solution was precipitated into diethyl ether. The precipitate was heated to 70 °C in a vacuum environment to remove residual DMF and diethyl ether.

Preparation of the LM-filled cross-linked polymer

EGaIn was fabricated by mixing gallium (75.5 wt%) and indium (24.5 wt%). To prepare LMMPs, 100 μL of EGaIn and 2 mL of deionized water were added into a small vial, followed by probe ultrasonication for 10 min (900 W, 25 kHz, 30% amplitude, pulse on 3 s, pulse off 3 s, BXT-900W). An appropriate amount of FeCl₃ was dissolved into pure water to form an aqueous solution with concentration fixed at 0.5 mol L⁻¹. Aqueous solution of 1 mL of poly(MAA-co-OEGMA) (0.15 g of mL⁻¹), certain volume of LMMPs solution (4.8–192 μL) and 10 μL of FeCl₃ aqueous solution were mixed in a small vial and stirred for 5 min. Then the solutions were poured into a PTFE mold and placed at 70 °C for 3 h to dry and form a cross-linked polymer. Finally, the samples were immersed in deionized water for 2 h to reach the equilibrium. Non-porous LM hydrogels were prepared by drying out the water under vacuum at room temperature for 6 h, followed by placing the samples in a fume hood under ambient conditions for 12 h to ensure complete drying.

Material characterization

The microstructure of the hydrogels was examined by using a JEOL JSM-6490LA SEM. The hydrogels to be observed were frozen in liquid nitrogen for 45 s and fractured carefully. The porosity (P) of LMH can be estimated from SEM images using the formula P (%) = A_pore/A_total × 100, where A_pore is the total area of the pores, and A_total is the total area of the SEM image. The porosity analysis and pore diameter measurements were conducted via ImageJ.

To measure the water content values, LMHs were taken in a beaker and immersed in DI water until attaining the maximum swelling ratio. Then, the swollen hydrogels were taken out of the water and weighed after removing the excess water on the surface. The swollen ratio (SR) was calculated from the following equation SR (%) = (W_s − W_d)/W_d × 100, where W_s and W_d are the weights of the swollen hydrogel after a certain time of soaking and the corresponding dried hydrogel, respectively.

Uniaxial tensile measurements of the hydrogels (20 mm × 4 mm × 0.4 mm) were performed using an Instron universal test instrument (Model 5566, Instron Instruments, USA) with a 10 N load cell at a velocity of 20 mm min⁻¹ at room temperature. The hydrogels in the tensile test were all soaked in DI water for 2 h.

EIS results of hydrogels were measured using a PalmSens4 electrochemical workstation (PalmSens BV, The Netherlands). The frequency was set to 0.2–10⁵ Hz. Graphite foil sheets were used as electrodes attached to both sides of the hydrogel, and the electrodes were connected to the electrochemical workstation for testing. The dielectric constants of the samples were calculated by using the parallel plate capacitor equation ε = ε_rε₀ = C/tS, where C is the capacitance obtained from the EIS test, t is the thickness of the sample, and S is the cross-sectional area of the sample between the electrodes.

Electro-responsive bending actuation tests

The size of the LMH used to investigate bending behavior was 20 mm × 4 mm × 0.4 mm (length × width × thickness). Two graphite electrodes in NaCl solution (0.05–0.4 M) were placed parallel at a distance of 40 mm and were linked to a voltage direction shift circuit. This circuit can provide programmable power-on time, adjustable voltage, and variable current direction and is shown in Fig. S1. Then, the hydrogel strip was vertically placed in the center between the two electrodes in the solution. The authors used a camera to take photos of the bending behavior of the samples every 5 s. The applied electric voltage was 5, 10, or 15 V, thus the applied E is 0.125, 0.25, and 0.375 V mm⁻¹, since E can be calculated using E = V/d, where V is the applied voltage and d is the distance between the electrodes. For the fabrication of the soft robot, LM2 was used to perform the item pushing and item wrapping. The walking robot was made with a buoyant PU foam body and two LMH legs with different LM content (LM2 and LM5). The legs and the body were assembled with PU foam. All the applications were carried out in an acrylic transparent water tank with parallel graphite electrodes spaced 50 mm apart under the applied E of 0.3 V mm⁻¹ (voltage of 15 V).

ICP-MS leaching test

LM2 hydrogel samples were immersed in 40 mL of 0.2 M NaCl solution at 25 °C between two parallel graphite electrodes (electrode gap: 40 mm). An alternating electric field of ±0.25 V mm⁻¹ with polarity switching every 40 s was applied for up to 60 min. Following actuation, 20 μL of the bath solution was collected and diluted 100× in 0.5% HCl for ICP-MS analysis. A control group was prepared by diluting 0.2 M NaCl solution 100× with deionized water. ICP-MS measurements were performed at the UNSW ISSEAU-ICP Laboratory using an ICP-MS NEXION 5000-ParkinElmer instrument.

Author contributions

Q. Z. contributed to the investigation, methodology, the development of software and validation of this work. Z. J. contributed to the conceptualization. Q. Z. took the lead in formal analysis, data curation and visualization with the support of H. L. and X. B. Q. Z. and Z. J. performed the experiments and measurements with the support of Y. G., J. W., and L. G. Q. Z. wrote the original draft, while S-Y. T., W. L., H. D., and Z. J. contributed to reviewing and editing. W. L. and Z. J. directed the team and led the funding acquisition, project administration, resource management, and supervision. All authors have read and approved the final manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its supplementary information (SI). The SI includes device schematics, synthesis procedures, characterization details, swelling kinetics, EIS data, ICP-MS analysis, simulation results, additional mechanical and electro-responsive data, comparison with reported ERHAs, and theoretical derivations. Supplementary information is available. See DOI: https://doi.org/10.1039/d5mh01365h.

Acknowledgements

The authors acknowledge the use of the facilities at the UOW Electron Microscopy Center. The authors are grateful for the financial support provided by the University of Wollongong. This work was funded by the Australian Research Council (ARC) Discovery Project Grant DP230100823 (to W. L.) and the Discovery Early Career Researcher Award DE220101102 (to Z. J.).

References

1. F. M. Cheng, H. X. Chen and H. D. Li, J. Mater. Chem. B, 2021, 9, 1762–1780.
2. W. Li, Q. Guan, M. Li, E. Saiz and X. Hou, Prog. Polym. Sci., 2023, 140, 101665.
3. M. Yang, Y. Liu, G. Duan, Z. Liang, Y. Huang, C. Zhang, X. Han, C. Ma, S. He and S. Jiang, J. Chem. Eng., 2024, 498, 155157.
4. H. Liu, H. Chu, H. Yuan, D. Li, W. Deng, Z. Fu, R. Liu, Y. Liu, Y. Han, Y. Wang, Y. Zhao, X. Cui and Y. Tian, Nanomicro Lett., 2024, 16, 69.
5. L. Xu, Y. Yang, J. Li, W. Pan, T. Li, S. Chen and J. Ding, Mater. Lett., 2023, 350, 134889.
6. X. Q. Wang, K. H. Chan, W. Lu, T. Ding, S. W. L. Ng, Y. Cheng, T. Li, M. Hong, B. C. K. Tee and G. W. Ho, Nat. Commun., 2022, 13, 3369.
7. Y. Sun, L. Chen, Z. Zeng, T. Wang, K. Zhang, Y. Sun, L. Yang, X. Feng, Q. Sun, C. Ma, S. Xiao, Z. Liu, J. Liu and X. Yang, Adv. Funct. Mater., 2024, 35, 2410348.
8. C. Qian, Y. Li, C. Chen, L. Han, Q. Han, L. Liu and Z. Lu, J. Chem. Eng., 2023, 454, 140263.
9. L. Fan, X. Sun, X. Wang, H. Wang and J. Liu, RSC Adv., 2019, 9, 13026–13032.
10. C. Li, S. Zheng, X. Hao, W. Hong, Q. Zheng and Z. Wu, Sci. Adv., 2022, 8, eabm9608.
11. H. Jiang, L. Fan, S. Yan, F. Li, H. Li and J. Tang, Nanoscale, 2019, 11, 2231–2237.
12. Y.-W. Kang, J. Woo, H.-R. Lee and J.-Y. Sun, Smart Mater. Struct., 2019, 28, 095016.
13. Y. Shin, M. Y. Choi, J. Choi, J. H. Na and S. Y. Kim, ACS Appl. Mater. Interfaces, 2021, 13, 15633–15646.
14. J. Zeng, X. Jing, L. Lin, G. Wang, Y. Zhang and P. Feng, J. Colloid Interface Sci., 2024, 677, 816–826.
15. J. Ko, C. Kim, D. Kim, Y. Song, S. Lee, B. Yeom, J. Huh, S. Han, D. Kang, J.-S. Koh and J. Cho, Sci. Robot., 2022, 7, eabo6463.
16. G. H. Kwon, Y. Y. Choi, J. Y. Park, D. H. Woo, K. B. Lee, J. H. Kim and S. H. Lee, Lab Chip, 2010, 10, 1604–1610.
17. C. Dong, A. Carnicer-Lombarte, F. Bonafe, B. Huang, S. Middya, A. Jin, X. Tao, S. Han, M. Bance, D. G. Barone, B. Fraboni and G. G. Malliaras, Nat. Mater., 2024, 23, 969–976.
18. Y. Zhao, A. C. Tavares and M. A. Gauthier, J. Mater. Chem. B, 2016, 4, 3019–3030.
19. J. H. Ha, H. H. Shin, H. W. Choi, J. H. Lim, S. J. Mo, C. D. Ahrberg, J. M. Lee and B. G. Chung, Lab Chip, 2020, 20, 3354–3364.
20. J. Zhang, J. Liao, Z. Liu, R. Zhang and M. Sitti, Adv. Funct. Mater., 2023, 34, 2308238.
21. D. Morales, E. Palleau, M. D. Dickey and O. D. Velev, Soft Matter, 2014, 10, 1337–1348.
22. S. Lee, M. Kim, J. Choi and S. Y. Kim, Mater. Today Chem., 2023, 29, 101434.
23. Y. Wang, N. Alizadeh, M. Barde, M. L. Auad and B. S. Beckingham, ACS Appl. Polym. Mater., 2022, 4, 971–979.
24. J. Yang, W. Huang, K. Peng, Z. Cheng, L. Lin, J. Yuan, Y. Sun, N. J. Cho and Y. Chen, Adv. Funct. Mater., 2024, 34, 2313725.
25. Z. Ying, Q. Wang, J. Xie, B. Li, X. Lin and S. Hui, J. Mater. Chem. C, 2020, 8, 4192–4205.
26. K. Rotjanasuworapong, N. Thummarungsan, W. Lerdwijitjarud and A. Sirivat, Carbohydr. Polym., 2020, 247, 116709.
27. M. A. Jolfaei, Y. Zhao, G. M. Spinks and Z. Jiang, Chem. Commun., 2024, 60, 14407–14410.
28. Z. Jiang, S. M. Seraji, X. Tan, X. Zhang, T. Dinh, M. Mollazade, A. E. Rowan, A. K. Whittaker, P. Song and H. Wang, Chem. Mater., 2021, 33, 7818–7828.
29. J. Liu, W. Xu, Z. Kuang, P. Dong, Y. Yao, H. Wu, A. Liu and F. Ye, J. Mater. Chem. C, 2020, 8, 12092–12099.
30. Y. Jian, B. Wu, X. Yang, Y. Peng, D. Zhang, Y. Yang, H. Qiu, H. Lu, J. Zhang and T. Chen, Supramol. Mater., 2022, 1, 100002.
31. Y. Yang, T. Wang, F. Tian, X. Wang, Y. Hu, X. Xia and S. Xu, Macromol. Rapid Commun., 2021, 42, e2000749.
32. H. Lu, S.-Y. Tang, Z. Dong, D. Liu, Y. Zhang, C. Zhang, G. Yun, Q. Zhao, K. Kalantar-Zadeh, R. Qiao and W. Li, ACS Appl. Nano Mater., 2020, 3, 6905–6914.
33. S.-Y. Tang and R. Qiao, Acc. Mater. Res., 2021, 2, 966–978.
34. F. M. Allioux, M. B. Ghasemian, W. Xie, A. P. O'Mullane, T. Daeneke, M. D. Dickey and K. Kalantar-Zadeh, Nanoscale Horiz., 2022, 7, 141–167.
35. N. M. Smeets, E. Bakaic, M. Patenaude and T. Hoare, Acta Biomater., 2014, 10, 4143–4155.
36. M. Reindl, V. Zach and S. P. Schwaminger, ACS Appl. Mater. Interfaces, 2025, 17, 16644–16657.
37. N. Tiwari, D. Kumar, A. Priyadarshani, G. K. Jain, G. Mittal, P. Kesharwani and G. Aggarwal, J. Drug Delivery Sci. Technol., 2023, 82, 104319.
38. C. Zhang, B. Yang, J. M. Biazik, R. F. Webster, W. Xie, J. Tang, F. M. Allioux, R. Abbasi, M. Mousavi, E. M. Goldys, K. A. Kilian, R. Chandrawati, D. Esrafilzadeh and K. Kalantar-Zadeh, ACS Nano, 2022, 16, 8891–8903.
39. W. Zhang, H. Guo, X. Zhang and H. Li, Mater. Today Commun., 2025, 47, 112976.
40. E. Nicol, Biomacromolecules, 2021, 22, 1325–1345.
41. C. Yang, Z. Liu, C. Chen, K. Shi, L. Zhang, X. J. Ju, W. Wang, R. Xie and L. Y. Chu, ACS Appl. Mater. Interfaces, 2017, 9, 15758–15767.
42. S. Y. Zheng, H. Ding, J. Qian, J. Yin, Z. L. Wu, Y. Song and Q. Zheng, Macromolecules, 2016, 49, 9637–9646.
43. H. Hu, D. Li, T. Salim, Y. Li, G. Cheng, Y. M. Lam and J. Ding, J. Mater. Chem. C, 2024, 12, 1565–1582.
44. B. Bin Asghar Abbasi, M. Gigliotti, S. Aloko, M. A. Jolfaei, G. M. Spinks and Z. Jiang, Chem. Commun., 2023, 59, 7141–7150.
45. Z. Zhang, L. Tang, C. Chen, H. Yu, H. Bai, L. Wang, M. Qin, Y. Feng and W. Feng, J. Mater. Chem. A, 2021, 9, 875–883.
46. M. Wang, O. J. Rojas, L. Ning, Y. Li, X. Niu, X. Shi and H. Qi, Carbohydr. Polym., 2023, 301, 120330.
47. S. Guo, S. Zhang, H. Li, S. Liu, J. J. Koh, M. Zhou, Z. Sun, Y. Liu, H. Qu, Z. Yu, Y. Zhang, L. Yang, W. Chen, C. He, C. Lee, D. Mao, S. K. Ravi, Y. Lai and S. C. Tan, Matter, 2025, 8, 101785.
48. G. Yin, J. Wu, L. Ye, L. Liu, Y. Yu, P. Min, Z. Z. Yu and H. B. Zhang, Adv. Funct. Mater., 2024, 35, 2314425.
49. H. Li, C.-w Wu, S. Wang and W. Zhang, Mater. Lett., 2020, 266, 127504.
50. Y. Alsaid, S. Wu, D. Wu, Y. Du, L. Shi, R. Khodambashi, R. Rico, M. Hua, Y. Yan, Y. Zhao, D. Aukes and X. He, Adv. Mater., 2021, 33, e2008235.
51. I. I. C. o. E. Safety, IEEE C95.1-2019, 2019, pp. 1–312 DOI:10.1109/IEEESTD.2019.8859679.
52. T. Shiga and T. Kurauchi, J. Appl. Polym. Sci., 1990, 39, 2305–2320.
53. S. Chen, R. Zhao, X. Sun, H. Wang, L. Li and J. Liu, Adv. Healthcare Mater., 2023, 12, e2201924.

---