Reversible Thermoactuation Unlocks Minimally Invasive Implantation and Retrieval of Soft Bioelectronics

ABSTRACT

Minimally invasive delivery of bioelectronics is currently limited by the irreversibility of deployment, rendering device retrieval traumatic and hindering clinical translation. Here, for the first time, we introduce a novel thermoresponsive, reversible‐actuating polymer (Trap) that enables both minimally invasive implantation and retrieval. Trap exhibits a mechanistically unique dual‐crystalline competition between (110)‐oriented low‐entropy crystals and (100)‐oriented high‐entropy crystals. The competitive crystallization governs bidirectional, stress‐free shape memory within a human‐compatible window (10°C–37°C), enabling rapid (<3 s), fatigue‐resistant, and large reversible strain (∼30.17%). The solid–solid switching between two nanocrystalline states provides a robust and tunable actuation mode, allowing Trap to transition reversibly between compact 1D and functional 2D/3D geometries without mechanical loading. This materials’ innovation directly enables microinvasive deployment and retraction of Trap‐based neural electrodes through the same small incision (∼5 mm), as well as autonomous helical self‐assembly and thermal detachment on peripheral nerves, achieving stable electrophysiological interfacing over weeks to months. This work establishes a material‐centered framework for reversible biointerfaces, resolving the conflict between surgical invasiveness and device retrievability.

Addressing the “easy‐in, hard‐out” dilemma of implantable electronics, a reversible actuating polymer (Trap) is developed. By harnessing a cooperative dual‐crystalline switching strategy, Trap achieves robust, stress‐free reversibility (∼30.17%) within the body's thermal window. This enables the fabrication of neural electrodes that can be delivered via microcatheters and, crucially, refolded for safe retrieval.

1. Introduction

Advanced bioelectronic devices and engineered constructs have been successfully developed for nearly all major human tissues, including the heart [1, 2], vasculature [3], bone [4], spinal column [5], brain [6], and ocular tissues [7]. However, placing these devices or engineered constructs inside the body still relies on invasive surgical procedures. For instance, implanting cortical surface electrodes typically requires large craniotomy [8], and subcutaneous electromyography recording often involves extensive skin incisions [9]. Such procedures carry substantial risks, including tissue damage, inflammation and infection [10, 11, 12, 13, 14].

Emerging concepts from soft robotics [5, 15], 4D printing [16] and injectable materials [17, 18] offer promising routes toward minimally invasive delivery, enabling devices to be introduced through small incisions or needles in a compact, temporary configuration and subsequently expanded in situ into their functional geometries. Some seminal works have highlighted the efficacy of these approaches. For instance, Zhao et al. [19] elegantly exploited developed 4D‐printed hydrogel scaffolds exhibiting swelling‐stiffening behavior, which enables multidimensional transformation (1D–3D) for minimally invasive delivery; Proctor et al. [15] successfully integrated flexible thin‐film electrode arrays with soft robotic, creating large‐area electrocorticography (ECoG) devices capable of shape reconfiguration via integrated fluidic actuators. However, despite the success of these delivery mechanisms, a fundamental limitation remains: the deployment of these structures is generally irreversible. Current research prioritizes the stage of implantation while often neglecting the retrieval process. As a result, the removal of the post‐monitoring device still requires invasive open surgery, which severely limits clinical transformation and long‐term application.

The combination of minimally invasive delivery with biodegradable materials has been proposed to avoid the need to take it out [20, 21, 22]. However, the biodegradation process inevitably destabilizes electrical performance, and the long‐term biosafety of degradation byproducts remains a concern. Consequently, there is a critical need for strategies that enable both minimally invasive implantation and retrieval, ensuring surgical trauma minimally and high‐quality signal acquisition throughout the device's operational lifetime.

Shape‐memory polymers (SMPs) have emerged as ideal candidates for such adaptable biointerfaces, owing to their intrinsic shape tunability [23]. However, the conventional shape memory effect is typically one‐way (irreversible), involving a transition from a kinetically trapped temporary state to a thermodynamically stable equilibrium state upon stimulation [24]. This mechanism restricts devices to a single‐use deployment action, facilitating implantation but fundamentally lacking the reversibility required for retrieval. Although reversible (two‐way) actuation has been achieved in certain liquid crystalline networks [25] and semi‐crystalline polymers [26], these systems typically suffer from critical limitations in clinical scenarios. Most operate at actuation temperatures significantly exceeding physiological safety limits (≥43°C) [27], posing severe risks of thermal necrosis to surrounding tissues. Furthermore, the actuation mechanism in conventional reversible SMPs typically relies on the transition of driving phase, from mutually conversion of the ordered crystal domain and the completely disordered amorphous coil [28, 29]. This complete microstructural dissociation often results in structural discontinuity, leading to fatigue and sluggish response rates. Consequently, it remains a formidable challenge to develop a material system capable of stress‐free, rapid, and fully reversible actuation within a mild, biologically compliant thermal window remains.

Here, we address this problem by introducing a neural interface platform based on thermoresponsive, reversible‐actuating polymers (Trap), which operate via a novel dual‐crystalline competition mechanism. Distinct from conventional polymers that rely on a disordering melt transition, Trap harnesses a reversible solid‐solid phase transition between a stable low‐entropy crystalline phase and a metastable high‐entropy phase. This cooperative crystallographic switching enabled rapid, fatigue‐resistant actuation within a biologically compliant thermal window (10°C–37°C). Exploiting these unique thermodynamics, the device can be programmed into a transient one‐dimensional shape for microcatheter delivery and autonomously expands into a robust functional structure upon exposure to endogenous body heat (37°C). Crucially, this transition is fully reversible. Localized cooling can trigger enthalpy‐driven recrystallization, allowing the device to be refolded and safely retrieved with minimal secondary trauma. This work enables applications in multiple scenarios, such as minimally invasive transcutaneous catheters, minimally invasive electrophysiological signal monitoring interface and so on. This work provides a robust material strategy for resolving the retrieval challenges of soft bioelectronics, enabling a closed‐loop surgical workflow from minimally invasive implantation to atraumatic extraction.

2. Results and Discussion

To address the limitations between minimally invasive implantation and traumatic retrieval, we engineered a thermoresponsive, reversible‐actuating polymer (Trap) scaffold capable of on‐demand morphological reconfiguration. Trap is designed to undergo a fully reversible transformation between a compact delivery state and a functional expanded state, thereby enabling both atraumatic deployment and minimally invasive extraction to minimize postoperative complications (Figure 1a). Trap is synthesized through a catalyst‐free polycondensation reaction of citric acid (CA), sebacic acid (SA), and 1,12‐dodecanediol (DC) (Figure 1b). Specifically, DC was selected as the primary crystallizable segment. Its long alkyl chains facilitate the formation of ordered crystalline domains, which is essential for the shape memory switching. CA serves as the multifunctional crosslinking center to establish a robust elastomeric network. And SA acts as a flexible spacer to modulate the crystallization kinetics and mechanical compliance. The chemical structure of Trap prepolymer is confirmed using fourier transform infrared spectroscopy (FTIR) and ¹H‐nuclear magnetic resonance (¹H‐NMR) (Figures S1 and S2).

FIGURE 1.

The synthesis and programming process of thermoresponsive reversible‐actuating polymers. a) Schematic illustration for deformation of Trap from 1D to 2D based on temperature response and the operation of minimally invasive implantation and retrieval. b) Synthetic scheme of Trap prepolymer. c) DSC thermogram showed a broad melting transition temperature (T_high = 35°C–45°C) during heating scan and crystallization transition temperature (T_low = 0°C–22°C) during cooling scan. d) Programming and shape changing process of Trap. e) Molecular mechanisms for the thermoresponsive reversible actuating of trap.

By adjusting the ratio of CA:SA:DC, Trap can achieve reversible volume contraction and recovery between a high‐temperature actuation (T_high = 35°C–45°C) and a low‐temperature actuation (T_low = 0°C–22°C) (Figure 1c and Figures S3 and S4). The broad, tunable actuation temperature range means that Trap can undergo reversible shape changes under both body temperature and comfortably low thermal stimuli, thereby minimizing potential tissue damage.

To probe the reversible shape‐memory behavior, Trap is programmed into a “rod‐like” shape. Figure 1d and Figure S5 illustrate the reversible shape‐memory behavior of Trap. Heating Trap in 60°C while maintaining deformation force allowed changing of this temporary shape. And at −30°C, it locked into a temporary rod‐like shape. After the force was removed, the rod‐liked Trap was transferred to water at a high temperature of 37°C, where it immediately reverted to a flat shape. The flat Trap was then transferred back to water at a low temperature of 10°C, where it returned to its rod‐like shape.

Essentially, as a semicrystalline polymer, Trap exhibits its reversible shape memory behavior due to crystal‐induced elongation (CIE) and melt‐induced contraction (MIC). We envision that the programming of Trap can proceed via the scheme in Figure 1e. Here, four temperatures (T_prog, T_high, T_low, and T_fix) are defined as T_prog > T_high > T_low > T_fix. These correspond to the programming temperature (T_prog), the high (T_high) and low (T_low) temperature for reversible actuation, and the shape fixing temperature (T_fix). In this process, Trap undergoes complete crystal melting at T_prog (T_prog < T_m), resulting in an amorphous state which it can be programmed permanent shape A into temporary shape B. Subsequently, maintaining deformation force, Trap is cooled to the shape fixing temperature (T_fix < T_c), where directed crystallization occurs along the strain axis. This process establishes a crystalline network that stabilizes the deformed configuration, thereby defining the temporary shape B. During the reversible shape memory cycle, heating Trap to T_high triggers partial melting of the crystalline domains, resulting in a structural reconfiguration of the temporary shape (shape B → shape A’). The residual oriented crystallites act as a structural framework and internal stress reservoir, directing the recrystallization process and facilitating the recovery of the original configuration upon cooling to T_low (shape A’ → shape B’). As such, reversible actuation between Shape A’ and Shape B’ occurs as the temperature switches across T_high and T_low. Based on this mechanism, simple reheating Trap above T_m induces complete recovery to the permanent Shape A. During this process, the disappearance of the oriented crystalline domains effectively erases the stored shape memory. The material can then be reprogrammed into a new actuation mode by applying a different deformation force in the next programming cycle.

To validate the proposed mechanism, in situ WAXS and SAXS analyses were employed to monitor the evolution of the microstructural features of Trap. In Figure 2a,b, compared with the unstretched (or unprogrammed) samples, the stretched (or programmed) samples exhibited an enhanced scattering intensity along the equatorial direction at T_low, indicating the presence of crystalline preferential orientation. This demonstrates that the material transitions from an isotropic to a highly anisotropic state [30]. The azimuthal intensity distribution revealed that the broad and diffuse peak observed in the unstretched sample indicated a disordered arrangement of molecular chains, whereas the stretched sample exhibited sharp double peaks at 0° and 180°, confirming the uniaxial orientation of crystals along the stretching direction [31] (Figure 2c). To quantify this alignment, the Herman's orientation factor (f) was calculated based on the azimuthal intensity I(φ) by Equation (1) [32]:

$$\begin{array}{c}\hfill f=\frac{3〈co{s}^2\phi 〉-1}{2},where〈{\cos }^2\phi 〉=\frac{{\int }_0^{\frac{\pi }{2}}I\left(\phi \right){\cos }^2\phi \sin \phi d\phi }{{\int }_0^{\frac{\pi }{2}}I\left(\phi \right)\sin \phi d\phi }\end{array}$$ (1)

FIGURE 2.

Evolution mechanisms of microstructures in the reversible shape memory behavior of Trap. a,b) 2D WAXS (i) and 2D SAXS (ii) patterns of unstretched (left) and stretched (right) Trap. c) Azimuthal scattering curves of 2D WAXS patterns obtained by unstretched and stretched Trap. d) In situ 2D‐WAXS patterns at T_high = 37°C and T_low = 5°C, and their corresponding 1D‐WAXS scattering curves in the insert map. e) Structural changes occurring during the thermoresponsive reversible actuating cycle obtained by in situ 2D‐SAXS. f) Reversible shape memory cycle of Trap.

This substantial structural ordering was quantitatively supported by the Herman's orientation factor, which rised from f = 0.05 in the unprogrammed state to a highly oriented f = 0.66 after programming. The Herman's orientation factor of the programmed sample (f = 0.66) is significantly closer to 1 compared to the unprogrammed state (f = 0.05), which proves that the programming process successfully triggers a transition from a random, disordered molecular arrangement to a highly oriented, anisotropic crystalline structure aligned along the stress axis. Such a high degree of orientation is crucial for stabilizing the temporary shape. These highly oriented crystallites serve as robust physical crosslinking points [33], which restrict molecular chain mobility and form a rigid, low‐entropy crystalline framework that effectively stabilizes the temporary shape at T_low.

Temperature‐dependent 1D‐WAXS analyses revealed that the solid‐solid crystalline phase transition served as the structural basis for the bidirectional shape memory effect observed in this work. Fundamentally, the two‐way shape memory behavior within the 5°C–37°C range originates from a thermally induced competitive crystallization process. This process is governed by the Gibbs free energy Equation (2) [34]:

$$\mathrm{G}=\mathrm{H}-\mathrm{TS}$$ (2)

As shown in Figure 2d, according to the crystallographic database [35], the two diffraction peaks located at 2θ = 20° and 2θ = 22° correspond to the (100) and (110) planes, respectively. At 5°C, the free energy was dominated by the enthalpy term (H). The system minimizes H by forming the low‐entropy crystalline phase (Phase II) [36], as evidenced by the sharp diffraction peak at 2θ = 22°. This phase II was attributed to a lamellar stacking oriented perpendicular to the polymer backbone, which stabilized the temporary shape by robust molecular alignment [37]. Upon heating to 37°C, the entropy term (‐TS) outweighed the enthalpic contribution [38]. Phase II partially melted, accompanied by the emergence of a high‐entropy metastable phase (Phase I). This phase I was characterized by a broad diffraction peak at 2θ = 20°, which corresponded to the (100) reflection of a loosely packed chain arrangement. The expanded intermolecular spacing in Phase I enabled maximizing conformational entropy (S) to power the recovery of shape closing to the permanent shape. The reversible exchange between these phases thus serves as a direct evidence of this enthalpy‐entropy trade‐off, confirming that the actuation is structurally regulated by the solid‐solid phase transition.

In situ 2D‐SAXS (Figure 2e) further elucidated the hierarchical mechanism governing the reversible shape transformation. At 5°C, the 2D‐SAXS pattern exhibited meridional arc‐like scattering, indicating that Phase II crystals adopted a periodic lamellar stacking arrangement perpendicular to the strain direction [39]. This long‐period ordering was fully consistent with the uniaxial molecular orientation revealed by WAXS. Upon heating to 37°C, the scattering pattern evolved into an isotropic ring, suggesting that the long‐range lamellar order was disrupted while the local crystallinity was retained. During repeated cooling/heating cycles (5°C ↔ 37°C), the 2D‐SAXS patterns displayed a fully reversible transition in anisotropy. Collectively, these observations demonstrate that the reversibility originates from a thermally regulated nanocrystalline ordering: at T_low, cooperative chain alignment locks the temporary shape through enthalpy‐minimized packing, whereas at T_high, disruption of the lamellar order triggers an entropy‐driven recovery to the permanent shape.

To quantify the reversible shape‐memory behavior, dynamic mechanical analysis (DMA) was employed to monitor the strain evolution of Trap during temperature switching. Figure 2f presented five reversible shape memory cycles of Trap. Under stress‐free conditions, a reversible strain of up to 30.17% was achieved, surpassing that of most previously reported reversible shape memory polymers (Figure S6a). The folding angle experiment demonstrates that Trap maintains a stable, reversible shape‐memory capability during 100 continuous cooling‐heating cycles, exhibiting excellent fatigue‐resistant properties (Figure S7). Furthermore, Trap is currently the only shape memory material capable of reversible transformation within the human body's comfortable temperature range. This property is crucial for protecting a patient's surrounding tissues during surgical procedures (Figure S6b). The excellent reversible actuation performance is attributed to the stable solid–solid phase transition between two crystalline orientations. Unlike conventional shape‐memory polymers, which require complete melting and recrystallization of actuation domains to achieve reversible cycling [30]. This process prone to fatigue due to structural discontinuity. However, the bidirectional switching demonstrated here arises from a solid–solid transition between two distinct crystalline phases. During heating‐cooling, the partial melting of Phase II cooperates with the nucleation of Phase I to achieve entropy‐driven recovery without passing through a fully disordered state. This cooperative crystallographic switching reduces the free‐energy barrier and suppresses hysteresis, thereby providing a structural basis for rapid, robust, and fatigue‐resistant actuation.

As shown in Figure 3a, Trap exhibited an exceptional bidirectional shape memory capability under stress‐free conditions, enabling effortless transformation between 2D and 3D configurations, which holded promise for minimally invasive implantation and removal. The two‐step curing process of Trap enabled the change of its permanent shape. After the first curing, a soft 2D sheet was obtained, which still contains some unreacted active groups. Consequently, during the second curing, new ester linkages were formed, allowing to design new permanent 2D or 3D architectures. For instance, a flat 2D sheet can be programmed into a 3D helix, followed by a second curing stage that sets its new permanent shape (Figure 3b).

FIGURE 3.

Diverse reversible actuation behaviors of the Trap and potential biomedical applications. (a) Programmable deformation of designed 2D flat patterns. The left column and right column represented the initial configuration and the deformed 3D structures, respectively (including flower, butterfly, box and chair). (b) Schematic illustrations and photographs for the permanent shape programming of Trap. (c–f) Schematic illustrations and photographs of multi‐dimensional structures undergoing the permanent shape programming and potential applications as intravascular stent (c), soft tissue defect support (d), Spinal cord stimulation electrode (e) and multi‐channel electroencephalogram electrodes (f). All of them can be minimally invasive implanted and removed by reversible actuation behaviors. Scale bars:10 mm.

Inspired by natural morphologies and biomedical needs, various Trap‐based devices were fabricated to demonstrate its programmable and thermally responsive capabilities. To align with simulated clinical application scenarios, 10°C was used as the T_low to ensure the animals' comfort during cooling. In fact, effective retrieval can be triggered at any temperature within its crystallization window (T_low = 0°C–22°C). 10°C was specifically selected to as the optimal operational temperature for its mild physiological impact and high recovery efficiency. It should be noted that although irrigation with 10°C normal saline solution may cause localized surface temperature fluctuations, temporary vasoconstriction and nerve block, these effects are fully reversible within minutes (Figures S8–S10), thereby confirming the physiological safety of this actuation method. The physiological safety of employing a 10°C cooling medium for transient localized thermal intervention is well‐documented in the literature [40, 41, 42], endowing Trap with broader application potential. Some proof‐of‐concepts in vitro was conducted to evaluate the feasibility of using Trap as a scaffold for minimally invasive implantation and removal. Figure 3c depicted a 3D hollow tube as a vascular stent. The 3D hollow tube was first transformed into a 1D solid cylinder and delivered to the target vascular site. Under the stimulation of body temperature, the 1D solid cylinder expanded to support the vessel. Subsequent localized cooling induced contraction back to the 1D solid form, enabling its minimally invasive retrieval. Figure 3d illustrated a spherical scaffold formed by bonding the ends of a triple‐cross pattern. Upon heating, the six edges bend outwarded from the center, giving rise to a 3D hollow spherical scaffold, representing its permanent shape. This structure facilitated minimally invasive filling of soft‐tissue defects. Subsequently, localized cooling was achieved, that enabled the hollow sphere to be reversibly retrieved in a minimally invasive manner. Figure 3e displayed a 3D rolled structure designed for spinal cord stimulation electrodes. The 3D rolled construct can be delivered to the lesion site within the spinal column and subsequently unfolded into a planar configuration upon exposure to body temperature, enabling electrical stimulation of the spinal cord. Upon localized cooling, Trap re‐rolled into its compact form, allowing for minimally invasive retrieval. Figure 3f illustrated a mimosa‐inspired 2D structure designed for large‐area, multichannel electrocorticography (ECoG) recording. The folded, mimosa‐like Trap electrode can be inserted through a small slit in the skull and positioned onto the cortical surface, where the leaf‐like structure unfolded spontaneously at body temperature. For retrieval, localized cooling is applied, inducing refolding of the structure and enabling its removal through the same minimal incision. Notably, all the scaffolds and microelectrodes were capable of completing body temperature–triggered recovery to their permanent shapes and cold water–induced programmable deformation within a few seconds.

The combination of reversible actuation capability and minimally invasive implantation was demonstrated through both in vitro and in vivo experiments. In vitro, a 2D sheet was rolled along its longitudinal axis into a 1D rod‐like form and fixed at −30°C. When immersed in water at 37°C, the 1D rod‐like Trap rapidly unfolded into its 2D configuration; after transferring it to 10°C water, the Trap quickly returned to its 1D rod‐like shape (Movie S1 and Figure 4b). Next, the feasibility of the 1D–2D–1D shape transformation of Trap was verified in vivo through endogenous thermal stimulation (Figure 4a,c and Movie S2). The Trap were sterilized using ethylene oxide (EtO), which did not affect its transition temperature or performance (Figure S11). The temporary 1D rod‐like Trap was implanted beneath the skin through a small incision using a minimally invasive clip (Figure 4c‐i). Warm water (37°C) was injected subcutaneously into Sprague Dawley (SD) rats to create a deformation space and to ensure uniform heating of Trap (Figure 4c‐ii). After thermal stimulation for 3 s, Trap unfolded into a 2D planar configuration (Figure 4c‐iii). When the warm water was withdrawn and replaced with 10°C cold water, Trap rapidly recovered its 1D rod (Figure 4c‐iv, 4c‐v). This process demonstrates the potential of Trap as a subcutaneous electrode that can be implanted and retrieved in a minimally invasive manner.

FIGURE 4.

The minimally invasive implantation and retrieval process of Trap as microelectrode, along with the recorded electromyographic signals (EMG). (a) Schematic illustration for the minimally invasive implantation and retrieval of Trap. (b) Digital images revealing the shape changing delivery procedure of Trap. (c) Photographs showing the feasibility of Trap electrode to complete 1D–2D deformation subcutaneously in SD rats. (d,e) Comparative digital images of traditional electrode implantation (d) and minimally invasive electrode implantation (e). The process represent includes incision (i), electrode placement (ii), suturing (iii), wound healing after 1 week (iv), and electrode removal (v). (f) EMG signals detected by the Trap electrode on first day and seventh day. (g) Signal‐to‐noise ratio (SNR) of the EMG signals detected by the Trap electrode with different time. Scale bar: 5 mm.

As a proof of concept, a four‐channel Trap electrode was implanted into the left hindlimb muscle of SD rats via a minimally invasive approach for electromyographic (EMG) signal measurement (Figure S12). Trap electrode was encapsulated with SU‐8, selectively exposing the electrode sites. The corresponding leakage current data confirms its high insulation capacity (Figure S13). As shown in Figure 4e and Figure S14, despite the electrode's area being similar to the muscle area, it can be inserted through a small 5‐mm incision and then expanded on the tissue. The incision size was significantly smaller compared to traditional subcutaneous electrode implantation (Figure 4d‐i, e‐i), which not only reduced the risk of surgical infection but also promoted faster recovery. As shown in Figure 4e‐iv, within 1 week post‐surgery, the incision in the rat had completely healed, with no impact on walking, whereas the wounds in rats subjected to traditional surgeries were still healing (Figure 4d‐iv). After 1 week, Trap electrode was successfully removed by gently separating the surrounding fascia and applying local cooling, allowing for its minimally invasive retrieval (Figure 4e‐v). The Trap electrode successfully recorded 7 days of electromyography signals from the left hindlimb of freely moving rats (Figure 4f and Figure S15). With all channels demonstrating reliable performance, spatiotemporal surface EMG potential maps were built (Figure S16). It revealed the highly similar signal patterns observed across different channels during free locomotion. These indicated excellent inter‐channel consistency and confirm the stable recording capability of the Trap‐based bioelectrode interface. In addition, the signal‐to‐noise ratio remained comparable 1 week after implantation, further demonstrating the consistent and reliable signal‐recording capability of the Trap electrode (Figure 4g). Otherwise, the good biocompatibility and biosafety of the Trap electrode can be evaluated by the high viability of the cells on the surface of Trap electrode (Figures S17 and S18). Histological evaluation staining revealed mild inflammation around the Trap electrode three days after implantation, which resolved by 1 week, further confirming the excellent biocompatibility of the Trap electrode (Figure S19). Therefore, the developed Trap provides a new strategy for both minimally invasive electrode delivery and retrieval, showing great potential for biomedical applications.

Apart from the size of the surgical incision, there is another important point that also needs to be taken into consideration. Electrode implantation procedures are often accompanied by iatrogenic neural injury, arising from mechanical stretching or compression during manual manipulation [43, 44, 45]. Conventional shape‐memory electrodes can self‐wrap around nerves after implantation [46], which helps reduce damage caused by surgical tools. However, once their functional lifespan ends, these electrodes still need to be manually removed from the nerve. An overlooked issue is that the unloading process of electrode poses a substantial risk of causing additional injury to the nerve. Encourage by the permanently programmable shape and stress‐free bidirectional shape‐memory behavior of Trap, we further explored its use as a neural interface electrode to enable damage‐free assembly and gentle removal on the sciatic nerve. Under stress‐free conditions, Trap can autonomously wrap around or detach from delicate neural tissues such as the sciatic nerve, enabling gentle fixation and release.

As shown in Figure 5b and Movie S3, the permanent shape of Trap was programmed into a helical configuration to achieve a conformal electrode–nerve interface. Next, Trap was temporarily deformed into a 2D sheet, allowing insertion between the sciatic nerve and surrounding muscle tissue. Upon mild irrigation with 37°C water, Trap rapidly recovered its helical shape and spontaneously wrapped around the sciatic nerve. When exposed to 10°C cooling, the helix promptly unwounded and detached from the nerve. This design minimized potential neural damage induced by rigid surgical tools and enabled a stable, conformal electrode‐nerve interface without additional fixation. Subsequently, the helical Trap electrode was implanted onto the surface of the rat sciatic nerve for in vivo recording of action potentials. The recorded signals can serve as feedback inputs for closed‐loop prosthetic control, enabling the investigation of the relationship between neural stimulation and motor response. The experimental setup for stimulation and evoked potential recording was illustrated in Figure 5a. Two hook‐shaped platinum (Pt) electrodes were used for stimulation, and the helical Trap electrode served as the recording interface. A series of monophasic rectangular pulses with amplitudes ranging from 0.1 to 0.3 mA, a constant frequency of 15 Hz, and a pulse width of 100 µs were delivered to the sciatic nerve through the Pt hooks. During stimulation, action potentials were evoked and propagated along the nerve to the motor endplates, triggering muscle contractions and subsequent limb movement. The corresponding compound nerve action potentials (CNAPs) recorded by the wrapping Trap electrode were shown in Figure 5c. All recorded signals exhibited a stimulation artifact followed by an evoked CNAP. For each stimulus, the CNAPs displayed nearly identical amplitudes and waveforms. Figure 5d compared three representative evoked CNAPs at different stimulation currents. The waveforms remained consistent, while the peak amplitudes (∼490, ∼607, and ∼825 µV) increased proportionally with the applied current (0.1, 0.2, and 0.3 mA), consistent with previous reports [47, 48]. After the stimulation were completed and localized cooling was applied, allowing the electrode to autonomously detach from the sciatic nerve and thereby preventing additional nerve damage during the removal process (Figure 5b). Trap's thermally induced ability to autonomously wrap and unwrap from the sciatic nerve makes it highly suitable for acute intraoperative nerve monitoring, providing critical nerve protection during high‐risk orthopedic procedures such as acetabular and pelvic fracture surgeries [49], hip arthroscopy [50] and so on.

FIGURE 5.

Application of the Trap electrode for monitoring of sciatic nerve action potentials. (a) Schematic diagram of the in vivo experimental setup to detect sciatic nerve compound nerve action potentials (CNAPs). (b) Photographs of the spiral Trap electrode autonomously wrapping and unwrapping around the sciatic nerve. (c) Recorded CNAPs evoked by varying current (0.1, 0.2, and 0.3 mA). (d) Enlarged view of the comparison between the three evoked CNAPs. (e) Sciatic nerve signals recorded over 4 months by the spiral Trap electrode. (f) Signal‐to‐noise ratio (SNR) of the CNAPs detected by the spiral Trap electrode with different time.

Furthermore, the long‐term stability of Trap was also assessed. Following prolonged incubation in both PBS and lipase solutions, Trap exhibited no significant mass loss and change of pH (Figure S20), demonstrating the long‐term stability of its crosslinked network and confirming its suitability for chronic in vivo monitoring. Stable electrophysiological signals from the sciatic nerve were successfully recorded in freely moving rats over a period of 4 months, showing a high and stable signal‐to‐noise ratio (Figure 5e). The long‐term EMG recording owes to the stable interfacial adhesion between the Trap and the Au layer (Figures S21–S23), coupled with the highly conformal and stable contact established at the Trap‐nerve interface. Meanwhile, the pressure applied by the Trap electrode to the sciatic nerve is significantly lower than the damage threshold (Figure S24). Because the spiral structure relaxes as the Trap swells, the actual pressure exerted becomes even smaller over time. Therefore, the sciatic nerve is free from the risk of mechanical compression (Figure S25).

Histological analysis confirmed the excellent biocompatibility of the Trap electrode, revealing no signs of severe inflammation or tissue damage or significant increase in fibrous capsule thickness. (Figures S26–S28). The permanent shape‐programmable property of Trap enables the fabrication of customized helical electrodes tailored to different nerve sizes. This design ensures conformal and stable contact with the nerve while avoiding mechanical compression. In addition, the stress‐free and reversible shape memory behavior of Trap eliminates the need for forced mechanical manipulation, offering a low‐trauma strategy for neural interfacing. Under mild and controllable thermal stimulation, the material can form a safe and conformal electrode–nerve interface, and can also autonomously and controllably detach the interface when needed. This capability provides a promising pathway toward safer neural interfaces and reduced postoperative complications.

3. Conclusions

In summary, we have developed a material‐centered solution to the longstanding challenge of minimally invasive implantation and on‐demand retrievability in bioelectronics by engineering a thermoresponsive, reversible‐actuating polymer (Trap). Distinct from conventional shape‐memory systems that rely on a disordering melt transition, Trap operates via a novel dual‐crystalline competition mechanism. This process involves a reversible solid‐solid phase transition between an enthalpically stable phase and an entropically favored metastable phase. Trap enables rapid (<3 s), stress‐free and fatigue‐resistant actuation (∼30.17% reversible strain) strictly within a biologically compliant thermal window (10°C–37°C). Base on these unique thermodynamic attributes, Trap demonstrated a closed‐loop minimally invasive workflow for neural interfaces. The device can be compacted into a high‐stiffness rod for precise microcatheter delivery, autonomously deployed upon exposure to body heat. And crucially, It also can on‐demand refolded via localized cooling for atraumatic retrieval. This capability effectively bridges the gap between advanced soft electronics and surgical safety, minimizing postoperative complications. Ultimately, the dual‐crystalline strategy presented here establishes a versatile framework for designing next‐generation recylable bio‐implants, offering a paradigm shift from permanent fixation to reversible interaction in clinical applications.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: adma73685‐sup‐0001‐SuppMat.docx.

Supporting File 2: adma73685‐sup‐0002‐MovieS1.mp4.

Supporting File 3: adma73685‐sup‐0003‐MovieS2.mp4.

Supporting File 4: adma73685‐sup‐0004‐MovieS3.mp4.

Data Availability Statement

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