Conformal and adhesive gel for stable electrophysiology on hairy animals without shaving

Abstract

Non-invasive collection of weak electroneurographic signals is crucial for cognitive neuroscience. However, animal scalps are often covered by dense hair, which presents a substantial barrier in the scalp-electrode interface for accurate electrophysiological acquisition. Although scalp hair removal is common practice in clinical settings, it has detrimental effects on the test subjects and is infeasible in many non-clinical scenarios. Here we report a hair-adaptable and adhesion-tunable (HAAT) scalp-electrode interface that can accurately acquire electroneurographic signals from hairy skins of animal and human. By synthetically integrating dynamic covalent bonds and ionic conduction channels in the copolymer poly(sodium thioctate-co-sulfobetaine methacrylate), we implement the key functionalities of the HAAT interface including penetration of dense hair layers, strong conformal skin-electrode adhesion, on-demand painless detachment, and sensitive neuroelectric coupling. We achieve stable and precise electroneurographic signals acquisition from human, monkey and mouse’s scalps with various hair densities. The accurate event-related potential monitored during monkey vision-attention tasks facilitates the study of monkey brain activity. These results demonstrate the HAAT interface as an outstanding platform for conducting non-invasive research on animal cognition.

Hair hampers scalp electrophysiology, yet shaving is often impractical. Here, the authors demonstrate an adhesive gel for high-quality EEG via dynamic covalent bonding and ion conductive polymers.

Introduction

The non-invasive and precise collection of electroneurographic signals from the brain, such as electroencephalogram (EEG) and event-related potential (ERP), is a crucial technology widely applied in cognitive neuroscience^1–4. A primary challenge in developing non-invasive brain neural electrodes lies in establishing a robust and conformal interface with the scalp, a prerequisite for minimizing signal artifacts and ensuring reliable neuroelectric coupling^5,6. This challenge becomes particularly pronounced in test subjects, e.g., rodents and primates, whose scalps are covered with amounts of hair. The hair layer creates gaps at the scalp-electrode interface and significantly hinders stable and intimate contact. While hair removal is an established pre-operative protocol in clinical settings such as craniotomy, where patient compliance is implicitly secured by therapeutic necessity, this approach fails to translate to many scenarios. For example, pediatric subjects have very delicate scalps, so hair removal can trigger psychological and physiological harm. In cognitive neuroscience research, hair removal from healthy volunteers encounters substantial cultural, comfort, and esthetic barriers. In animal behavior studies, fur and feathers serve vital biological functions, rendering their removal both impractical in field studies and against the ethical treatment of laboratory animals. Thus, there is a pressing need for non-invasive brain neural interfaces capable of overcoming the challenges posed by hair while ensuring stable and conformal contact with the scalp.

Current strategies involve two types of non-invasive neural interfaces to achieve hair adaptability for brain study (Supplementary Fig. 1a, b). Solid electrodes^7–10 have been designed with specific structures, such as claw configurations, to address this issue. However, these solid electrodes were fixed by external pressure on the scalp, where the continuously strong force often leads to extreme discomfort. Fluid interfacial materials^11,12 such as saline bypass hair and conform to skin wrinkles. Nevertheless, they require EEG caps to fix electrodes, and introduce risks of positional shifts as well as signal crosstalk in high-density arrays. The development of a non-fluid and imperceivable interface for non-invasive neural electrodes to comfortably and efficiently acquire electrophysiology at hairy regions of animal scalps remains challenging.

We envision a hair-adaptable and adhesion-tunable scalp-electrode interface, termed HAAT, with the following features: (1) the capability to penetrate dense hair layers and conform to skin wrinkles, for example, through liquid-solid phase transitions; (2) robust mechanical and electrical coupling to the scalp for reliable recording; and (3) tunable adhesion for on-demand painless detachment. Biogels including gelatin and poly(N-isopropylacrylamide)^13–15 are capable of penetrating dense hairs through fluid-gel transition. However, they exhibit either weak adhesion, causing unexpected detachment, or excessively strong adhesion, inducing hair loss or skin irritation (Supplementary Fig. 1c, d, Supplementary Table 1). Polymers with dynamic covalent bonds in the backbone have emerged as promising bioadhesives with potential hair-adaptability. Dynamic dissociation and formation of the covalent bonds facilitate fluid-gel transitions, strong interfacial bonding, and tunable adhesion modulated by external stimuli. Notably, disulfide-based poly(lipoic acid) (PLA) has recently been reported as a recyclable surgical superglue due to its high-strength adhesion and excellent biocompatibility^16–18. Leveraging these properties, a dynamic covalent polymer incorporated with a disulfide bond emerges as a suitable interface for non-invasive neural electrodes for hairy animals, one that not only adapts to hair but also offers robust and customizable adhesion. Despite their outstanding mechanical properties and biocompatibility, poly(lipoic acid) based gels have found limited use in electrophysiological recording, highlighting the critical need for material innovation to fully realize the HAAT design.

Result

Mechanical design principle of HAAT

Ensuring a mechanically robust scalp-electrode contact is crucial for the non-invasive acquisition of high-quality electroneurographic signals from the brain (Fig. 1a). The robustness of the electrode-skin adhesion can be characterized by the critical peeling force (F_c) that is needed to break the contact^19,20. According to the Irwin-Kies relation:

$$F_c\propto {\left(\frac{G_C\cdot A}{C}\right)}^{\frac{n}{n+1}}$$ 1

where G_c, C, A and n are the critical fracture energy, compliance of the electrode material, contact area, and a peeling mode-dependent constant, respectively. G_c is further broken down into the intrinsic fracture energy (G₀) the dissipative term (G_D) that depends on specific peeling conditions including peeling rate ($\stackrel{°}{a}$), temperature (T), etc.

$$G_c=G_0+G_D=G_0\left(1+f\left(\stackrel{°}{a},T,etc.\right)\right)$$ 2

Fig. 1. Design of hair-adaptable and adhesion-tunable scalp-electrode interface (HAAT).

a Schematic of the hair-adaptable and adhesion-tunable scalp-electrode interface (HAAT) for animal cognition study. b Working mechanism of hair adaptability and tunable adhesion of HAAT. 1) Dynamic bond dissociation allows hair penetration; 2) Dynamic bond reassociation and interfacial covalent bonding allow strong adhesion; 3) Interfacial dynamic interaction failure allows hair-friendly removal. c Photographs of HAAT penetrating different densities of hair, varying from dense hair from monkey and mouse to sparse hand hair. Scale bar is 1 cm.

Therefore, large modulus (low C) and strong interfacial bonding (large G₀) favor stable adhesion^21–23.

An ideal HAAT goes through three steps in its life cycle: 1) low-modulus fluid state for hair penetration; 2) high-modulus and adhesive gel state for stable skin-electrode interface; 3) non-adhesive state for hair-friendly removal (Fig. 1b). Dynamic polymers undergo fluid-gel transitions triggered by temperature, ultraviolet irradiation or solvent^24–26, facilitating hair penetration and conformal attachment to skin. The stimuli-triggered formation and dissociation of covalent bonds in the polymer backbone and on polymer-skin interfaces alter the modulus and interfacial fracture energy, allowing both strong adhesion and easy on-demand detachment of the electrode. Dynamic polymers augmented with ionic conductivity, therefore, represent an ideal and generalizable strategy to implement HAAT for non-invasive electroneurographic sensing on the skins of varying hair densities (Fig. 1c).

Fabrication of HAAT

We demonstrate the HAAT principle using the copolymer of sodium thioctate (ST) and sulfobetaine methacrylate (SBMA), hereafter denoted as poly(ST-co-SBMA) (Fig. 2a). The disulfide bonds in the backbone can be formed and cleaved by heating and reducing reagents, giving the copolymer its dynamic behavior. We use ST, instead of its conjugate α-lipoic acid, to enhance the water dispersibility of the monomer and hydrophilicity of the polymer¹⁶. The PSBMA subunit plays several critical roles. First, the zwitterions in the sidechain create ionic conduction channels, promoting sensitivity to electrophysiological signals. Skin interfacial impedances of the HAAT electrodes are comparable or even lower over a broad range of frequencies than commercial electrodes, which is attributed to the improved bio-interface as well as sufficiently high ionic conductivity (Supplementary Figs. 2, 3). Second, it suppresses the spontaneous ring-closing depolymerization of PST initiated by terminal sulfur radicals^16,27,28. Electron paramagnetic resonance (EPR) spectroscopy, Fourier transform infrared (FTIR), Raman, and X-ray diffraction (XRD) indicate that the disulfide bonds of ST undergo homolytic cleavage upon heating, generating thiyl radicals to initiate the polymerization of SBMA. Simultaneously, ST diradical coupling occurs, resulting in the formation of poly(ST-co-SBMA) (Supplementary Figs. 4–10). To further increase the modulus, ferric ions are added as crosslinkers through the formation of coordination bonds (Supplementary Fig. 11). This copolymer thus enlists multiple interactions, including coordination, hydrogen bonds, electrostatic interactions, and dynamic disulfide bonds, to synergistically improve its modulus, stretchability, adhesion, and ionic conductivity (Fig. 2a, Supplementary Figs. 12–14). By adjusting the temperature and concentration of disulfide and coordination bonds, the mechanical properties of poly(ST-co-SBMA) can be effectively controlled.

Fig. 2. Hair adaptability of HAAT induced by dynamic bonding.

a Chemical structure of poly(ST-co-SBMA). b Logarithm of storage and loss modulus (lg G’ and lg G”) and the loss factor (tan δ = G”/G’) from 25 to 95 °C. c A top-view scanning electron microscopy (SEM) image of the HAAT applied on hairy rat skin. d Cross-sectional SEM image of the HAAT applied on porcine skin. e Variable-temperature-variable in-situ Raman spectra of HAAT at 25, 35, 50, 65, 90 and 95 °C. υ(S–S) and υ(S-H) denote the S–S and S-H stretching modes, respectively. f, g 2DCOS synchronous and asynchronous spectra generated from the Raman spectra during heating. The warm color (red) represents positive intensities, while the cold color (blue) represents negative intensities.

Hair adaptability

Poly(ST-co-SBMA) exhibits a reversible fluid-gel transition within the range of body-friendly temperature (Supplementary Fig. 15). Rheological analysis reveals that as the temperature increases, both the storage modulus (G’) and the loss modulus (G”) decrease, crossing at 62 °C (Fig. 2b). Differential scanning calorimetry (DSC) indicates a phase transition from gel to fluid at the same temperature¹³ (Supplementary Figs. 16, 17). The transition temperature is promoted and suppressed by the addition of disulfide and Fe³⁺, respectively (Supplementary Figs. 18–21). By optimizing the disulfide and Fe³⁺ concentrations, we achieve a fluidic poly(ST-co-SBMA) at slightly above body temperature, which easily penetrates thick hairs (Fig. 2c, Supplementary Figs. 22, 23 and Supplementary Movie 1) and fills the microstructures of the skin, thereby achieving conformal contact (Fig. 2d).

Raman spectroscopy further reveals the temperature-driven dynamic behavior of the disulfide bonds. Temperature-dependent Raman spectroscopy is conducted over the range of 25 to 95 °C. Of note, the dissociation of disulfide bonds does not lead to substantial mass loss, as revealed by the thermogravimetric analysis (TGA, Supplementary Fig. 24). Intensities of the S–S stretching modes at 509 and 527 cm⁻¹ decrease with rising temperature, while the intensity of the S-H stretching at approximately 2536 cm⁻¹ slightly increases²⁹ (Fig. 2e). Two-dimensional correlation spectroscopy (2DCOS) provides additional insight^23,30, with synchronous and asynchronous spectra revealing the sequence of changes: 509, 527 → 2536 (→ denotes preceding) (Fig. 2f, g). According to Noda’s rule, this progression indicates the cleavage of disulfide bonds into thiol groups upon heating. The resulting reduction in polymer chain length drives the transition to a fluid state with lower modulus. Conversely, during cooling, the reformation of disulfide bonds lengthens the polymer chains, restoring the HAAT to a gel state with a higher modulus (Supplementary Fig. 25).

Tunable adhesion

Tunable adhesion and on-demand detachment are crucial features of HAAT (Fig. 3a). To this end, we first investigate the effect of Fe³⁺ and ST contents on the modulus and interfacial toughness of poly(ST-co-SBMA). When Fe³⁺ concentration is low, higher Fe³⁺ content increases modulus by forming more Fe³⁺-carboxylate coordination bonds; accordingly, interfacial toughness increases by the contribution of the dissipative term (G_D)¹⁹. At high Fe³⁺ concentration, excessive crosslinking reduces polymer chain mobility^31,32, limiting the movement of adhesive groups to the interface and reducing interfacial toughness through the reduction of G₀ (Fig. 3b). Initial increase of ST content raises the proportion of dynamic disulfide bonds, leading to a less rigid polymer backbone and lower modulus. Concomitantly, the more abundant adhesive groups increase G₀, dominating over decreasing G_D and leading to enhanced interfacial toughness. With increasing ST content, increased rigidity limits the availability of adhesive groups on the interface, leading to reduced interfacial toughness (Fig. 3c). By tuning Fe³⁺ and ST content, we simultaneously optimize G₀ and G_D, achieving the largest interfacial toughness of 523 N/m.

Fig. 3. Tunable adhesion of HAAT induced by dynamic disulfide bond.

a Schematic illustration of strong adhesion between HAAT and skin enabled by disulfide bonds and non-covalent interactions. b, c Modulus and interfacial toughness of HAAT tuned by the content of Fe³⁺ (b) and ST (c). d Schematic illustration of adhesion reduction triggered by detachment solution, where dynamic bonding breaks down. e Comparison of interfacial toughness between HAAT and porcine skin before and after detachment solution treatment. Data are presented as mean values ± standard deviation (SD) (n = 5 independent samples). f Comparison of Raman spectra before and after detachment solution treatment, showing the cleavage of disulfide bonds. g Photographs of the hairy skin before and after HAAT (top) and commercial electrode (bottom) removal. No redness occurs with HAAT. h Photographs of the scalp before and after HAAT removal by detachment solution. No hair loses by HAAT. Scale bar is 0.5 cm. i Comparison of interfacial toughness tuning range (Г_max/Г_min) and hair adaptability of HAAT with literatures^31,45–51.

Next, we demonstrate that chemically triggered cleavage of the dynamic bonds drastically reduces interfacial toughness, enabling painless detachment of the electrode (Fig. 3d). This is achieved by delivering a detachment solution consisting of glutathione (GSH) and NaCl to the skin-gel interface. GSH cleaves the disulfide bonds both at the interface and within the polymer backbone, resulting in reduced intrinsic adhesion and cohesion by breaking the polymer main chains³³. NaCl disrupts the interfacial non-covalent dynamic interactions, such as electrostatic interactions and hydrogen bonding³⁴. PBS solution (simulated sweat) alone cannot detach HAAT from the scalp, and large adhesion loss only occurs in the presence of reducing agents capable of cleaving disulfide bonds (Supplementary Movie 2). These observations suggest that interfacial disulfide bonds are the determining factor of interfacial toughness. Upon the delivery of the detachment solution, interfacial toughness between HAAT and porcine skin drastically decreases from 523 N/m to almost zero under a 90-degree peeling test (Fig. 3e, Supplementary Figs. 26, 27). The Raman intensities of υ(S–S) at 509 and 527 cm⁻¹ decrease substantially upon addition of the detachment solution, confirming the cleavage of disulfide bonds (Fig. 3f). These results highlight the on-demand detachment of HAAT without skin irritation or hair loss (Figs. 3g top, 3h, Supplementary Movie 3), which is advantageous over commercial electrode (Fig. 3g bottom, Supplementary Fig. 28). Notably, the detachment solution modulates the interfacial toughness (Γ) by over 50 times, ranking the highest among both hair-in-adaptable and adaptable EEG gels in literature (Fig. 3i, Supplementary Fig. 29 and Supplementary Table 2). In the cytotoxicity assessments of HAAT (Supplementary Fig. 30), the fluorescent images exhibit no observable signs of apoptosis or necrosis. The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay further demonstrates a higher proliferation rate of cells on HAAT compared to the control group after 3 days of culture, indicating excellent biocompatibility of HAAT.

Hairy scalp EEG recorded by HAAT electrode

Human and animals of different evolution levels are important research subjects in brain cognition^2,35–37. Animals, particularly primates and rodents, possess structural and functional similarities with humans, making them excellent models for uncovering the physiological mechanisms and cognitive processes of the human brain. Electroneurographic signals, such as scalp electroencephalogram (EEG), offer prominent evidence for various brain activities^38,39. To evaluate the acquisition ability of scalp EEG, HAAT electrode is first applied on human occipital region that is covered by thick hair (Fig. 4a). Compared to commercial EEG pastes, HAAT is able to penetrate hair without the need of additional pressure, tightly and stably adhere to scalp during prolonged usage in the presence of simulated sweat, and is readily removed by a detachment solution (inset of Fig. 4c, Supplementary Figs. 31–33). Characteristic brain waves (α, β, γ, θ) are distinguished after filtering the raw signals⁴⁰ (Fig. 4b, Supplementary Fig. 34). The power spectral density (PSD) of α and β waves obtained by HAAT is higher than that by commercial paste, highlighting its superior performance in scalp EEG acquisition on human^41,42 (Fig. 4c, Supplementary Fig. 35). HAAT is also able to penetrate the denser and finer hair of monkey, and construct a stable bio-interface enabling accurate EEG acquisition without the need of pressing (Supplementary Movies 4, 5). Harnessing the positional stability of HAAT, we subsequently utilize it in a 16-channel EEG system (O1, O2, P3, and P4) to concurrently record monkey EEG from different brain regions without signal crosstalk (Fig. 4e, f, Supplementary Figs. 36, 38), demonstrating its potential for high-density EEG mapping. Accurate acquisition of scalp EEG from animals with smaller heads and denser hair, such as mice, remains an outstanding challenge. Commercial EEG pastes fail to penetrate mouse fur and acquire an EEG signal even in the presence of applied pressure (Supplementary Fig. 39, Supplementary Movie 6). In contrast, the HAAT electrode successfully detects auditory evoked potentials (AEP) from the mouse scalp, demonstrating their superior hair adaptability (Fig. 4h, Supplementary Figs. 40, 41). A side-by-side impedance comparison between commercial gel and HAAT on skin and hairy skin is summarized in Fig. 4i, Supplementary Figs. 42.

Fig. 4. Hairy scalp EEG recorded by HAAT electrode.

a Schematic illustration and a photograph of an EEG recording from a human’s hairy scalp. b The EEG recorded from the occipital region by the HAAT electrode shows clear θ, α, β, and γ waves. c PSD of α and β waves obtained by the HAAT electrode. The inset photograph shows the HAAT electrode’s capability to penetrate hair and adhere to the scalp without the need for pressure. d Schematic illustration and a photograph of an EEG recording from a monkey’s hairy scalp. e Location map of 16 channels on monkey scalp. White and orange channels were recorded by HAAT and commercial EEG paste, respectively. f 16 channels of raw monkey EEG by HAAT array electrode (top) and comparison of raw monkey EEG recorded by HAAT (Channel O2, orange) and commercial paste (Channel PO8, gray) (bottom). g Schematic illustration and a photograph of an EEG recording from a mouse’s hairy scalp. h Raw mouse EEG recorded by HAAT electrode. i Impedance comparison between commercial gel and HAAT on skin and hairy skin.

ERP monitoring in a monkey’s vision-attention task

To further demonstrate the capability of HAAT in hairy scalp EEG electrodes, we incorporate it into the monitoring of weak electroneurographic signals to study a monkey’s cognitive process. Vision is a primary mode of interaction with the external environment, and has been extensively studied in cognitive neuroscience³⁶. A vision-attention task is designed to investigate attention’s influence on the visual processing of monkey (Fig. 5a). In this task, monkey is tasked with detecting a bright dot that emerged after a bright annulus over 4 h (Fig. 5b). Throughout this process, the event-related potential (ERP), which mirrors brain’s responses to distinct sensory, cognitive, or motor events, is meticulously monitored by HAAT electrode. ERP components are extracted and averaged over repeated trials to minimize noise and highlight signals associated with the events. Stimuli-evoked C1 (50–100 ms), P1 (100–130 ms), and N1 (120–200 ms) components are observed under eight different conditions⁴³ (Supplementary Figs. 43–50, Supplementary Table 3), indicating the high quality of signals and reflecting stimuli-evoked working processes. Heatmaps of neighboring channels (P3, P4 by HAAT electrodes and PO7, PO8 by commercial electrodes) demonstrate comparable ERP patterns under these conditions (Fig. 5c–e). The signal-to-noise ratio (SNR) in the early stage of the task is slightly higher than that in the end period, indicating that HAAT achieves equivalent performance throughout the entire task (Supplementary Figs. 51, 52). These further confirm that HAAT can serve as an advanced alternative to commercial pastes for ERP collection.

Fig. 5. ERP monitoring in a monkey’s vision-attention study.

a Experimental design of a vision-attention task to investigate the hemispheric laterization of attention processes in a monkey’s brain. The inset shows a representative ERP recorded by the HAAT electrode. b The experiment process of the vision-attention task. c Comparison of ERP heatmaps for two groups of neighboring channels (P3, P4 with HAAT electrode and PO7, PO8 with commercial electrode) under eight different conditions. d, e Statistical histogram (d) and boxplot (e) of the SNR by commercial paste (Channel PO8, gray) and HAAT (Channel O2 and P4, orange). In the boxplot, the central line indicates the median, the box bottom and top edges indicate the 25 and 75 percentiles, respectively. The whiskers extend to the maximum and minimum values. Following the Bootstrap Method, 5000 SNR values are obtained for comparison of signal quality by different electrodes. f Contralateral (f₁) and ipsilateral (f₂) ERP under the conditions of Annulus Left and Annulus Right (n = 1460). ERPs are presented as mean values ± standard deviation (SD). The arrow points to P1. g Contralateral processing of visual information in the brain.

To investigate the monkeys’ visual processing, ERP signals are compared across different stimulus (annulus and dot) and locations (left and right) in the task requiring the monkeys to detect the position of the target. Under the condition of Annulus Right (the annulus was presented on the right visual field), a pronounced P1 component is observed in the contralateral channel (on the left hemisphere of the brain) (Fig. 5f₁, orange), while the P1 component barely rises in the ipsilateral channel (on the right hemisphere brain) (Fig. 5f₂, orange). This aligns with the principle of contralateral processing of visual information⁴⁴ (Fig. 5g). However, under the condition of Annulus Left, not only the contralateral channel (Fig. 5f₁, blue) but also the ipsilateral channel (Fig. 5f₂, blue) exhibits a significant P1 component, suggesting a stronger response to left-side stimuli. This indicates a lateralized attention bias toward the left visual field. The ERP amplitudes (P1/N1) for Annulus Left (Figs. 5f₁ and 5f₂, blue) are larger than those for Dot Left (Supplementary Fig. 53, blue) conditions, because annulus stimuli are larger and brighter than dot stimuli. This task demonstrates the feasibility of using HAAT electrodes for recording ERP signals in hairy regions, underscoring their potential applications in cognitive neuroscience research. The HAAT electrode could be further used to explore exogenous attention effects on ERP by exploring more complex task conditions, such as setting the annulus as a cue and the dot as a target appearing in the same experimental trial.

Discussion

In this work, we propose a hair-adaptable and adhesion-tunable scalp-electrode interface achieved by incorporating dynamic covalent bonding into ionically conductive polymers. Poly(ST-co-SBMA) serves as a proof-of-concept of HAAT, with which we demonstrate non-invasive electroencephalographic recording on the hairy scalp for an animal cognition study. The fluid-gel phase transition facilitates facile hair penetration and adaptability to the scalp surface. Dynamic covalent bonds provide robust adhesion during bio-signal acquisition and allow for on-demand removal via a detachment solution. HAAT electrode successfully acquired high-quality electroneurographic signals from hairy-skin regions of human, monkey, and mouse, highlighting its potential for studying brain cognition. We believe that the present design can be extended to other dynamic covalent bonding polymers for the development of hair-adaptable and adhesion-tunable scalp-electrode interfaces.

Methods

Materials

DL-α-Thioctic acid (LA), Sulfobetaine methacrylate (SBMA), Iron chloride hexahydrate (FeCl₃·6H₂O), Sodium chloride (NaCl) and Glycerol are obtained from Aladdin. Glutathione (GSH) and Sodium hydroxide (NaOH) were obtained from Innochem. All purchased chemicals are of analytical purity. The commercial electroneurographic electrode was purchased from (X-1, Hangzhou Xunda). The commercial EEG paste (for human, monkey and mouse) and conductive gel were purchased from (Ten20, Weaver and Company) and (GT10, GREENTEK), respectively.

Synthesis of HAAT

HAAT was prepared via the following steps. Firstly, LA (4 g) and NaOH (0.7755 g) were dispersed uniformly in deionized water (10 mL) to prepare the ST solution. The ST solution was stored at 4 °C in the dark. Secondly, FeCl₃·6H₂O (2 g) was dissolved in deionized water (5 mL) to form an FeCl₃ aqueous solution. Thirdly, a certain amount of FeCl₃ aqueous solution, ST (2 mL), SBMA (2 g), glycerol (1 mL) was added into deionized water (2 mL) and then stirring in the dark to form the precursor solution of HAAT. The specific contents of each component are listed in Supplementary Tables 4, 5. Finally, the precursor solution was polymerized at 90 °C for 210 min to form the HAAT. Additionally, GSH (0.5 g) and NaCl (0.1 g) were dispersed uniformly in deionized water (9.4 g) as the detachment solution (5 wt% GSH and 1 wt% NaCl).

Characterizations of HAAT

The cross-sectional morphology of the HAAT was observed by SEM (SU-8010, HITACHI). EPR is measured at 9.853 GHz on a Bruker A300 instrument. Raman spectroscopy is performed by LabRAM HR Evolution (Horiba) equipped with an objective (50 ×, Olympus). The wavelength of the excitation laser is 532 nm. FTIR spectra of the monomers and the freeze-dried HAAT are collected on a Fourier transform infrared spectrophotometer (IRAffinity-1, SHIMADZU) over the spectral range from 4000 cm⁻¹ to 500 cm⁻¹. XRD of LA monomer and the freeze-dried bulk HAAT is collected on a diffractometer (X’ Pert Pro MPD, Panalytical). The thermal Imaging Camera (FLIR E54) is used for real-time observation of the temperature of HAAT. DSC is obtained on a METTLER TOLEDO DSC1 instrument at a ramp rate of 3 K/min. The rheological testing is conducted on a rheometer (MCR 302, Anton Paar). After frequency sweep (Supplementary Fig. 18a) and amplitude sweep (Supplementary Fig. 18b), the temperature ramp is conducted at the condition of 40% strain and 1 rad/s with the ramp rate of 5 K/min. Thermogravimetric analysis (TGA) is performed with a heating rate of 10 K/min from room temperature to 700 °C in N₂ atmosphere (STA200, HITACHI).

Mechanical and adhesive performance measurements

The samples for mechanical performance were prepared in a Telfon mold (50 mm × 25 mm × 1.5 mm). The precursor solution was poured into the mold before heating. After heating and cooling, HAAT with a thickness of about 1 mm was transferred from the mold and cut into strips (approximately 20 mm × 10 mm × 1 mm) for the tensile test. The tensile tests were conducted on a tensile tester (F105-IM, MARK-10). The tensile test was performed at a speed of 60 mm/min.

Heated HAAT was brushed on dried porcine skin or other substrates. The width of the specimen was about 10 mm. The sample is used for the 90-degree peeling test after cooling to room temperature. A tape was attached to the top surface of HAAT with the help of clamps, gauze and superglue (Supplementary Fig. 26). The peeling test was performed at a speed of 20 mm/min.

Cytotoxicity test

The cytocompatibility of HAAT was evaluated using an extract-based cell culture method. Human skin fibroblasts (HSFs) were seeded in a 96-well plate at an initial density of 5000 cells per well and cultured overnight to allow cells to attach and adapt. Subsequently, sterilized HAAT extract (prepared by dissolving 2 g of HAAT in 1 mL of culture medium) was added to the wells. A control group consisting of HSFs cultured in standard medium without HAAT extract was also established. Cell viability and cytotoxicity were assessed after 24, 48, and 72 h. At each time point, the pristine medium was removed, and the cells were stained using a Calcein-AM/PI Cytotoxicity Assay Kit. Fluorescent images were captured using a Confocal Laser Scanning Microscope (Stellaris 5, Leica) to visualize cell morphology. Cell viability was further conducted using an MTT kit (Thermo Fisher). All parallel holes were set in five groups for each test.

Subjects

One male macaque monkey (Macaca rhesus DN, aged 8 to 9 years at the beginning of the experiments and weighing 10 to 11 kg) participated in the experimental tasks. To immobilize the animal’s head during behavioral training, a titanium post was surgically attached to its skull using bone screws. The procedure involved inducing general anesthesia with ketamine (10 mg/kg) and maintaining it under isoflurane (1.5°–2.0%). All procedures followed the guidelines outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Approval for the experiments was obtained from the Institutional Animal Care and Use Committee of Beijing Normal University.

C57BL/6 J male mouse, aged 3–6 months and weighing 25–30 grams (Charles River Laboratories), was used in this experiment. The mouse was allowed to acclimate to the environment for at least one week prior to the experiment. It was housed under standard conditions of 22 ± 2 °C, 50 ± 10% humidity, and a 12-h light/dark cycle. For procedures requiring anesthesia, the mouse was anesthetized using 3% isoflurane delivered via an inhalation mask. Anesthesia was maintained throughout the procedure, and the mouse was monitored for signs of distress. Afterward, the mouse was placed in a recovery room at 22 ± 2 °C, with regular monitoring of their body temperature and overall health. All experimental procedures were approved by the Animal Ethics Committee of Beijing Normal University and were conducted in accordance with national animal welfare regulations.

Vision-attention task

In the vision-attention task, a monkey was required to detect a bright dot (as a target) that appeared following the presentation of a bright annulus (as a cue). To initiate a trial, the monkey fixated on a 0.3° fixation point (FP) for 300 ms (within a 3° window). The main part of the trial lasted 2.5 s, during which the monkey had to maintain fixation on the FP (within a 3° window). After a 350 ms delay following the main part of the trial, a bright annulus appeared on a gray background, either 5° to the left or right of the fixation point, and lasted for 20 ms. 580 ms after the disappearance of the annulus, a bright dot appeared, also 5° to the left or right of the fixation point, and lasted for 20 ms.

The animal was required to maintain fixation until the end of the trial, at which point a response prompt display appeared, presenting two bright dots 5° to the left and right of the fixation point, indicating that the animal should make a saccade to one of the locations³⁶. During this response phase, the animal was required to make a saccade to the location it believed the target had appeared, and maintain fixation for 150 ms. Within 0–2000 ms after the appearance of the response prompt display, trials in which the animal made a saccade to the correct location were counted as correct and rewarded with water; trials in which the animal broke fixation or responded incorrectly were not rewarded.

In most trials, both the cue (bright annulus) and the target (bright dot) stimulus appeared. In a smaller number of trials, only the target stimulus appeared, meaning that at the time the cue appeared, only the fixation point was presented. Whether a reward was given in these trials depended on the animal correctly detecting the target’s location. In some trials, only the cue was presented, meaning that at the time the target appeared, only the fixation point was present. In these cases, rewards were given randomly.

Stimulus details

Visual stimuli were generated using a stimulus generator (ViSaGe; Cambridge Research Systems) controlled by a custom C++ program developed in our lab. These stimuli were presented on a 22-inch CRT monitor (Dell, P1230, 1200 × 900 pixels, mean luminance 29.4 cd/m², 100 Hz refresh rate). The viewing distance was 114 cm. In the vision-attention task, all stimuli were presented on a gray background. A bright annulus appeared on either the left or right side of the screen, positioned 5° from the fixation point. The annulus had a contrast of 0.9, with an inner diameter of 2.2° and an outer diameter of 2.5°. Additionally, a bright dot was also displayed on either the left or right side of the screen, 5° from the fixation point. This dot had a contrast of 0.56 and a diameter of 0.35°. The positions of the annulus and the dot were independent of each other.

Data recording and analysis

To evaluate the electrical performance, we measured the skin contact impedance of the HAAT electrode and commercial electrode by an electrochemical workstation (CHI660E) in the range from 1 Hz to 100 kHz with the applied voltage of 20 mV.

The human and mouse EEGs in the present work were recorded by a neuroelectric signal recording box (Human SpikeBox, Backyard Brains). For dual-channel EEG recording, two electrodes (Au cup electrodes with commercial EEG paste or HAAT) were placed on occipital region (Supplementary Fig. 32), and a commercial ECG electrode was placed behind the ear as a ground electrode. For mouse EEG recording, two HAAT electrodes were placed mouse scalp, one of which is a reference (Supplementary Fig. 40). The hairy scalp EEG was recorded by the box with a bandpass filter (1 Hz to 50 Hz).

The monkey EEG signal processing was performed on the collected data using MATLAB (v.2023a) for fundamental signal analysis (independent component analysis/superposition average/fast-Fourier transform/filter). A 16-channel EEG cap was used to record monkey EEG, with one reference on the forehead (placed as Fig. 4e). Raw data were acquired with a 128-channel system (Blackrock Microsystems). The raw data were downsampled to 500 Hz and subsequently band-pass filtered between 0.3 and 50 Hz. Trials exhibiting excessive signal variability were excluded from the analysis. For each trial, the main segment of 2500 ms and the preceding 100 ms will be included in the analysis. Additionally, the results of all valid trials for each condition were averaged to obtain the ERP.

The SNR value is calculated according to the equation below.

$${SNR}_{ERP}=20{log}_{10}\left(\frac{{RMS}_{POST}}{{RMS}_{PRE}}\right)$$

SNR_ERP represents the signal noise ratio of the ERP, RMS_POST, and RMS_PRE represents the root mean square of the ERP signal and baseline, respectively. Following the Bootstrap Method, 5000 SNR values are obtained for comparison of signal quality by different electrodes.

Statistics and reproducibility

All experiments were repeated independently with similar results at least three times.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

L.Y., M.C., D.X., and N.L. designed the project and wrote the manuscript. L.Y., M.C., and S.W. carried out experiments and analyzed the experimental data. J.Q. suggested in mechanical experiments and analysis. W.H. suggested in the electroneurographic signal collection and analysis. M.C. and L.Y. performed the animal vision-attention task and analyzed the data. Y.W. suggested in the animal vision-attention task and data analysis. D.S. suggested in the manuscript writing. D.X. and N.L. supervised the project. All authors reviewed and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks Huiliang Wang, who co-reviewed with Mengmeng Yao; Chi-Ching Kuo, Huiliang Wang, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data supporting the results in this study are available within the paper and its Supplementary Information. Source data are provided with this paper. More detailed information is available from the corresponding authors upon request. Source data are provided with this paper.

Code availability

Data analysis made use of inbuilt functions in MATLAB. All parameters used for analysis are available in Methods and Supplementary Information. All codes for the electroneurographic signal analysis used in this study are available from the corresponding authors on request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70093-z.