A flexible and biocompatible trilayer coaxial heterogeneous structure microfiber electrode for long-term electrophysiological recordings in freely moving mice

Abstract

Neuroscience and neural engineering face the critical challenge of accurately capturing and interpreting electrophysiological signals for understanding brain function and developing neural prosthetics. Here, we develop a trilayer coaxial heterogeneous structure flexible neuronal electrode, rSF–Au–PC, which addresses these challenges through its innovative design and superior performance. Developed via a multi-step, large-scale fabrication process, the rSF–Au–PC electrode features an adjustable diameter and low specific impedance (7.67 MΩ at 1 kHz), ensuring precise signal capture. It also boasts high charge storage capacity (51.16 mC cm⁻²), high charge injection capacity (19.11 mC cm⁻²), and high signal-to-noise ratio (14.48 dB after three weeks), which are essential for reliable electrophysiological signal recording. The electrode's remarkable biocompatibility and robust electrochemical and mechanical stability make it suitable for long-term use, outperforming conventional tungsten wire electrodes in chronic in vivo applications. This advancement holds significant implications for neuroscience applications, particularly those requiring extended electrophysiological surveillance, and may pave the way for future innovations in neural prosthetics and diagnostic technologies.

---

Haolun Wang

Haolun Wang is currently an associate professor at Sichuan University. He received his PhD degree from the University of Electronic Science and Technology of China in 2020. He was a joint PhD student at Tsinghua University from 2015 to 2017 and the University of California, Los Angeles (UCLA) from 2018 to 2020. His research interests focus on nanofibers, flexible materials, and electroactive polymers.

Introduction

The domain of bioelectronics has seen exponential growth, especially in the development of neural interfaces technologies aimed at achieving efficient communication with the nervous system. Constructing a stable and efficient neural interface hinges on the use of high-quality neural electrodes that must possess low tissue-electrode impedance, excellent flexibility, superior biocompatibility, and outstanding stability to bridge neurons with backend electronic systems.^1,2 To this day, a variety of novel materials combined with advanced manufacturing technologies have yielded numerous types of neural electrodes, including metal electrodes,^3–5 metal oxide electrodes,^6,7 two-dimensional (2D) material electrodes,^8,9 carbon-based material electrodes,^10,11 and conductive polymer electrodes.^12–14 However, existing neural electrodes still face several challenges that need further optimization, especially in long-term recordings of neuronal electrophysiological signals in freely moving mice.

One of the primary limitations encountered in the field of neural interface technology is the elevated Young's modulus of current neural electrodes, particularly those composed of metallic or alloy materials, which results in a significant mismatch with the mechanical properties of brain tissue. This mismatch can lead to electrode tissue misalignment during the animal's daily activities, subsequently triggering biocompatibility issues, and even inflammation or immune rejections.^15–17 Secondly, the dimensions of existing neural electrodes are frequently excessive, impeding the minimization of tissue trauma associated with the implantation process.¹⁸ Thirdly, some composite neural electrodes, such as those made from graphene or carbon nanotubes, have higher electrode-tissue impedance compared to commercial tungsten wire electrodes or platinum–iridium alloy electrodes, resulting in lower signal-to-noise ratios of the recorded signals.^19,20 Finally, the production of some neural electrodes is predicated on photolithographic processes, which are rather intricate and not amenable to large-scale production and widespread application.²¹ The development of novel neural electrodes that possess high electrical conductivity, extreme flexibility, superior biocompatibility, precisely tunable dimensions, and scalability in manufacturing holds significant importance for the pressing needs in neuroscience and brain science for long-term in vivo electrophysiological signal recording.²²

As one of the alternative materials, silk fibroin, leveraging its superior mechanical strength and flexibility, solubility in solutions, tunable physical properties, high biocompatibility, ease of processing, and degradability, has been extensively utilized in the development of flexible electronic and bioelectronic devices.²³ However, there are limited reports on large-scale production of silk fibroin-based flexible neural electrodes for long-term in vivo electrophysiological signal recording. This scarcity is primarily due to the complexity of integrating directed one-dimensionalization with conductive modification for silk fibroin, which often hinders large-scale production. Moreover, the long-term in vivo electrophysiological signals recording demands electrodes with stringent collective dimensions, as well as exceptional electrochemical and mechanical stability, and biocompatibility. Current processes often struggle to achieve a comprehensive control over the geometric dimensions and the mechanical and electrical properties of silk fibroin electrodes.²⁴ Therefore, developing a fabrication process to achieve the large-scale production of silk fibroin-based neuronal electrodes with high conductivity, high flexibility, and high biocompatibility holds significant promise for long-term in vivo electrophysiological signals recording.

In this work, we integrated and proposed a fabrication process that includes silk fibroin extraction, phase-separated wet chemical spinning, magnetron sputtering, and chemical vapor deposition (CVD) to achieve the large-scale production of flexible neuronal electrodes based on geometrically adjustable silk fibroin fibers with a trilayer coaxial heterogeneous structure. This electrode features adjustable diameter, low specific impedance (∼7.67 MΩ at 1 kHz), high charge storage capacity (∼51.16 mC cm⁻²), and high charge injection capacity (19.11 mC cm⁻²), along with excellent electrochemical and mechanical stability. Moreover, this electrode exhibits excellent biocompatibility, with no inflammation or rejection reactions in local tissues after implantation for up to three weeks. This electrode was successfully used to record electrophysiological signals in vivo, exhibiting a higher signal-to-noise ratio (SNR) than commercial tungsten wire electrodes. After ten weeks, it still sensitively and stably recorded neuronal spike discharges. This electrode is of significant importance and practical value for in vivo neuroscience experiments, especially for paradigms requiring long-term electrophysiological signal recordings in freely moving mice.

Results and discussion

First, silkworm cocoons were subjected to degumming, drying, and dissolution processes to prepare silk fibroin spinning precursors.^25–27 The wet spinning apparatus used for the fabrication of regenerated silk fibroin (rSF) fibers is schematically depicted in the Fig. 1a. The spinning precursor solution was loaded into a syringe, from which it was extruded into a room temperature deionized water coagulation bath under the propulsion of an injection pump. Guided by a rotating shaft at a constant speed, the resultant fibers were collected onto a roll collector, forming a roll of rSF fiber substrate (Fig. 1b). The rSF fibers substrate was transferred to a vacuum drying chamber set at 40 °C and dried overnight. Subsequently, the substrate underwent parameter-controlled magnetron sputtering to deposit a conductive gold layer of varying thicknesses, resulting in the formation of rSF–Au doublelayer coaxial heterostructure fibers. Finally, a Parylene-C insulating layer was grown via CVD, yielding the rSF–Au with post-CVD Parylene-C insulation (rSF–Au–PC) trilayer coaxial heterostructure fibers (Fig. 1c).

Fig. 1 Fabrication process and presentation of the rSF–Au–PC electrode. (a) Regenerated silk fibroin (rSF) fiber substrate was prepared via wet chemical spinning, where the silk fibroin precursor solution in a 27G syringe was extruded into a deionized water coagulation bath and collected on a winding roll at 120 rpm; (b) with corresponding physical photographs and optical microscope images; (c) subsequent layer-by-layer fabrication process, including the preparation of the conductive layer Au via magnetron sputtering and the insulating/biocompatible coating Parylene-C via CVD, culminating in the formation of a trilayer coaxial heterogeneous structure rSF–Au–PC electrode; (d) the impact of different syringe parameters during the wet chemical spinning process on the diameter of the rSF substrate fibers; (e) the impedance of rSF substrate fibers with the same thickness of the Au conductive layer decreases with increasing diameter; (f) SEM images of rSF–Au–PC electrodes with different diameters.

We conducted a detailed study on the impact of various spinning process parameters on the diameter of fibers. The results indicated that the fiber diameter was significantly larger for the 23G needle compared to the 27G needle under the same conditions. The influence of coagulation bath length on the diameter of silk fibers was found to be relatively minimal, particularly with the use of a 27G needle (Fig. 1d). The variation in fiber diameter upon entry into the coagulation bath at a consistent winding speed was not significant, suggesting that smaller diameter fibers set more rapidly upon immersion in the coagulation bath. For the 23G needle, winding speeds below 40 rpm resulted in rapid deposition of the fibroin solution, while speeds above 120 rpm were too high for adjusting fiber position. To prevent connections between multiple fibers, the electrode diameter used in this study was 8–10 micrometers, achieved with a 27G needle at a winding speed of 30 rpm. We further investigated the effect of diameter on the impedance of rSF–Au–PC electrodes with the same Au layer of 120 nm. The results demonstrated that the impedance of rSF–Au–PC electrodes decreased with increasing of their diameters, this is attributed to the large conductive area resulting from the large diameter. Particularly, rSF–Au–PC electrode with a 20 µm diameter exhibiting an impedance as low as approximately 10 kΩ at the frequency of 1 kHz (Fig. 1e). Fig. 1f presents the micromorphology of rSF–Au–PC electrodes formed under various spinning parameters. The results indicate that the surfaces are smooth and the diameters are uniform, which is conducive to the vertical implantation and precise positioning of the electrodes during subsequent in vivo electrophysiological experiments.

Fig. 2a presents XRD patterns of the three types of fiber structures. The rSF fibers are primarily composed of two crystalline structures, Silk I and Silk II. Silk I structure is predominantly α-helical, with its corresponding diffraction angle 2θ around 9.5°, while Silk II structure is mainly β-sheet conformation, with the main diffraction peaks near 20.2° and 25.2°. This indicates that during the wet spinning process, rSF successfully transitioned from an α-helical structure to a β-sheet structure.^28,29 The XRD pattern of the silk fibers post-Au sputtering shows some changes relative to the original, with the broad peaks corresponding to the main crystalline structures of the silk fibers becoming less pronounced, and the appearance of an Au diffraction peak near 38°, confirming the presence of the gold layer. The energy-dispersive X-ray spectroscopy (EDS) analysis of rSF–Au–PC revealed the presence of characteristic Au peaks and a significant amount of carbon content. This is attributed to the chemical CVD process, which results in a uniform and dense deposition of Parylene-C on the surface of the gold layer (Fig. 2b). The SEM images of the rSF fibers show uniformly smooth micromorphology without any cracks or grooves (Fig. 2c). The corresponding EDS elemental mapping revealed the distribution of C, Au, and Cl elements. Wherein, the carbon content was significantly lower than that in the conductive adhesive of the substrate due to the Parylene-C coating, resulting in a darker appearance for the C element mapping in the fiber region (Fig. 2d). The mappings for Au and Cl elements exhibited uniform distribution, which can be attributed to the uniform formation of the magnetron sputtered Au layer and the CVD Parylene-C layer (Fig. 2e and f). The EDS elemental mapping of rSF and rSF–Au fibers reveals characteristics of the intermediate manufacturing process. Specifically, rSF, being a pure silk fibroin structure, exhibits a minimal presence of C, Au, and Cl elements (Fig. S1 and S2). In contrast, the rSF–Au fibers demonstrate a uniform distribution of Au elements throughout the structure (Fig. S3 and S4). Fig. 2g–i respectively display the knotting results of rSF, rSF–Au, and rSF–Au–PC fibers at the microscale. The bending radius of the three types of fibers can reach 10 µm, 50 µm, and 50 µm, respectively, demonstrating their excellent flexibility.

Fig. 2 Microscopic morphology and elemental analysis of the rSF–Au–PC electrode and its process products. (a) XRD spectra of rSF, rSF–Au, and rSF–Au–PC; (b) EDS point analysis elemental spectrum of rSF–Au–PC; (c) SEM images of rSF–Au–PC and corresponding (d) C, (e) Au, and (f) Cl elemental mapping images; (g–i) SEM images of rSF, rSF–Au, and rSF–Au–PC fibers after knotting at the microscopic level.

Electrical performance is the most important technical indicator for flexible neural electrodes, with low impedance and high charge injection capacity (CIC) being beneficial for efficient recording of high-quality neuronal electrical signals.^30–33 To closely simulate the electrochemical impedance of electrodes implanted in mouse brain tissue, phosphate-buffered saline (PBS) was employed as the electrolyte in this work. This approach allowed for a more accurate representation of the in vivo conditions, facilitating the assessment of electrode performance under conditions that mimic the physiological environment of the brain. We investigated the effect of different Au layer thicknesses (227 nm, 284 nm, 378 nm, and 454 nm) on the specific impedance (SI) of rSF–Au–PC electrodes, as well as the SI of commercial tungsten wire electrode from the same frequency range for comparison (Fig. 3a and S5). It was observed that as the Au layer thickness increased, the corresponding impedance values decreased. Particularly, at a frequency of 1000 Hz, the impedance of the rSF–Au–PC electrode with 378 nm-thick Au layer was found to be close to that of commercial tungsten wire electrode, and the impedance of the rSF–Au–PC electrode with 454 nm-thick Au layer was lower than that of commercial tungsten wire electrode. We have statistically analyzed the SI data at 1 kHz frequency for five types of electrodes, the results showed that the rSF–Au–PC electrode with a 454 nm-thick Au layer had the lowest SI, with a value less than 1 × 10⁷ Ω cm², proving its superior impedance performance (Fig. 3b). Stability is also crucial for long-term in vivo electrophysiological experiments. We further investigated the electrical and mechanical stability of the rSF–Au–PC electrode. In the highest 1000 cycles of cyclic voltammetry testing, the integral area of the CV curves of the rSF–Au–PC electrode remained almost unchanged, demonstrating its excellent electrochemical stability within the water reduction potential (Fig. 3c). Furthermore, we conducted cyclic bending tests on the rSF–Au–PC electrode. The results showed that after 1000 bending cycles, the impedance of the rSF–Au–PC electrode at 1 kHz frequency was still below 1 × 10⁶ Ω (Fig. 3d). We applied periodic square wave current excitation to the rSF–Au–PC electrode under a three-electrode system and recorded its voltage response. The results showed that the rSF–Au–PC electrode exhibited a fast capacitive voltage (V_a) of 0.28 V and a maximum polarization voltage (V_p) of 1.20 V (Fig. 3e). Based on the data and the geometric area of the rSF–Au–PC electrode, we calculated its CIC to be 19.11 mC cm⁻² as following equation:³⁴

where I_C is the current pulse applied, t_C is the pulse width, and GSA is the geometric surface area. We compared this value with the state-of-the-art neural interface electrodes reported in the literature (Table 1), including PEDOT:PSS–Au–Parylene electrode,³⁵ graphene fiber electrode,³⁶ CNT fiber electrode,³⁷ and PDMS-Pt–Ir electrode etc.³⁸ The results show the highest CIC for rSF–Au–PC electrode (Fig. 3f), indicate that the rSF–Au–PC electrode can inject charge more effectively, which is crucial for achieving efficient electrical signal transmission and neural stimulation.

Fig. 3 Electrical performance assessment of the rSF–Au–PC electrode. (a) The impact of varying Au layer thicknesses on the alternating current impedance within the frequency range of 10 Hz to 10 kHz for the rSF–Au–PC electrode; (b) comparison of specific impedance at 1000 Hz between rSF–Au–PC electrodes and commercial tungsten wire electrodes; (c) cyclic voltammetry testing over 1000 cycles within the aqueous stability window to characterize the electrochemical stability of the rSF–Au–PC electrode; (d) impedance variation of the rSF–Au–PC electrode after 1000 bending cycles; (e) voltage transient test for calculation of the charge injection capacity of the rSF–Au–PC electrode; (f) comparative analysis of the charge injection capacity specific impedance at 1 kHz, and geometrical area of the rSF–Au–PC electrode with those of state-of-the-art neural interfacing electrodes as reported in the literature.

Table 1 Key parameters comparison of neural electrodes include the rSF–Au–PC electrode and state-of-the-art neural interfacing electrodes as reported in the literature

Material	Geometrical surface area [μm^2]	Specific impedance at 1 kHz [MΩ μm^2]	Charge storage capacity [mC cm^−2]	Charge injection capacity [mC cm^−2]	Reference
RSF–Au–parylene C	78.5	7.67 ± 0.51	51.16 ± 6.46	19.11	This work
PI–IrOx	23 223.44	17.70	13.9 ± 3	2.2 ± 0.7	Tao Sun^30
Multifunctional hydrogel (MH)–Au	7 100 000	1136	7.64	0.195	Ming Yang^40
CNT fibers	1450	20.44 ± 8.2	372 ± 56	6.52	Flavia Vitale^37
PtIr	17 000	451 ± 13.9	1.2 ± 0.08	0.15
PDMS-Pt–Ir	14 500 ± 500	116	10 ± 2	0.14 ± 0.04	David A. Roszko^38
Pt–Ir	72 700	356.23	2.6 ± 0.4	0.06 ± 0.01
Cryogel–PPy	390 000		109.47 ± 19.71	0.168	Tianhao Chen^31
PEDOT:PSS–Au–Parylene	1400	7	25 ± 2.5	15 ± 2	Venkata Suresh Vajrala^35
IrOx	282 600	197.82	23.77	3.95	Jiahui Wang^41
CF	706	1490.72	0.91	0.12	Xuefeng Fu^36
GF	706	35.64	832.03	9.96
aCNTF	706	91.5	223.86	9.43
fCNTF	706	114.8	156.21	3.74
PtSi–PEDOT:PSS	23.8	7.14	15	1.4	Ines Muguet^42

---

For neural electrodes, particularly those used for long-term in vivo recordings, biocompatibility is of utmost importance as it directly affects the success of the recordings and the accuracy and reliability of the signals. It is essential to verify the biocompatibility of the rSF–Au–PC electrode before conducting animal experiments. It is well-known that the modulus matching between neural electrodes and brain tissue is one of the keys to improving biocompatibility. In this work, we first used finite element analysis (FEA) to simulate the stress distribution of the rSF–Au–PC electrode and a tungsten wire electrode after implantation into brain tissue when subjected to minor relative motion with the brain tissue (Fig. S6). The FEA simulation can effectively demonstrated the inevitable effects of the animal's daily activities on the electrodes themselves and the brain tissue during chronic experiments. The results indicated that during a minor displacement of 6.67-micrometer, the rSF–Au–PC electrode, benefiting from a Young's modulus more closely matched to brain tissue and its excellent flexibility, would naturally bend along the direction of displacement outside the tissue. This effectively reduced the movement amplitude of the intra-tissue portion of the electrode, thereby decreasing the compression of the tissue by the electrode. This would translate as a whole along the direction of displacement when subjected to relative movement, which could compress the implanted brain tissue and thus affect its biocompatibility (Fig. 4a). The stress distribution cloud diagrams revealed that under a 6.67-micrometer displacement, the stress at the implantation end of the rSF–Au–PC electrode was only 16% of that at the implantation end of the tungsten wire electrode, proving that the former is more capable of adapting to the mechanical fluctuations introduced during the animal's daily activities, thereby reducing the likelihood of immune responses (Fig. 4b). The gold layer is considered the primary contributor to the elastic modulus of the rSF–Au–PC electrodes. The nanoindentation experiment results showed a modulus of the rSF–Au–PC electrode approximately 1.5 GPa (Fig. 4c). In contrast, the modulus of the tungsten wire electrodes was about 400 GPa,³⁹ which is approximately 260 times higher than that of the rSF–Au–PC electrode.

Fig. 4 Mechanical property analysis and biocompatibility assessment of the rSF–Au–PC electrode. (a) Comparative finite element analysis between the rSF–Au–PC electrode and commercial tungsten wire electrodes, (b) where a minor displacement is applied to both to obtain stress magnitude and distribution within the electrode and simulated tissue. (c) Modulus measurement data of the rSF–Au–PC electrodes, where the gold layer is considered the main contributor to the elastic modulus. (d) Localized enlarged images of histological sections and pathological analysis of the rSF–Au–PC electrode implanted in mouse brains from one day to three weeks to investigate its biocompatibility.

We further implanted rSF–Au–PC and tungsten electrodes into the hippocampal CA1 region of ten mice and collected brain tissue samples from the implantation area for immunohistochemical analysis at 1 day, 3 days, 1 week, 2 weeks, and 3 weeks post-implantation (Fig. S7 and S8). The results showed no significant tissue necrosis or inflammatory cell infiltration, and there were no positive intervals for inflammatory responses, demonstrating the excellent biocompatibility of the rSF–Au–PC electrode (Fig. 4d).

A multi-channel electrode system, which allows for precise control over the depth of electrode insertion into the brain region, was employed for long-term in vivo neuronal signal recording. The system integrates rSF–Au–PC and tungsten electrodes within polyimide tubes, facilitating comparative signal detection (Fig. 5a and S9). Similar to biocompatibility testing, we selected the hippocampal CA1 region for neural signal recording (Fig. 5b), and the mice with implanted electrodes were housed separately to ensure the integrity of the experimental setup and the well-being of the animals (Fig. 5c). Concurrently, we conducted a comparative analysis of the in vivo impedance between rSF–Au–PC electrodes and tungsten wire electrodes across the four channels tasked with detecting neuronal signals, both prior to and following an three weeks period of signal detection. Particularly, channels 1 and 2 depict the impedance statistics for the tungsten wire electrodes, whereas channels 3 and 4 present the corresponding data for the rSF–Au–PC electrodes. We measured in vivo impedance for the four channels using the Intan system, results indicate that the average in vivo impedance of the rSF–Au–PC electrodes are significantly lower than that observed for the tungsten wire electrodes, suggesting a superior electrical performance of the silk protein-based electrodes over the course of the study (Fig. S10). After a two-week period post-electrode implantation, the spike firing data revealed that the sensitivity for spike detection on the two channels equipped with rSF–Au–PC electrodes markedly exceeded that of the two channels with tungsten wire electrodes (Fig. 5d). From the two-week in vivo electrophysiological experiments, it can be observed that the signal-to-noise ratio (SNR) for the tungsten wire electrodes was 8.43 dB and 9.21 dB, while for the rSF–Au–PC electrodes it was 8.98 dB and 13.31 dB (Table S1). After three weeks of electrode implantation, the SNR for the tungsten wire electrodes was 9.96 dB and 9.76 dB, and for the rSF–Au–PC electrodes it was 12.72 dB and 14.48 dB (Table S2). The primary reason for the increased SNR of the rSF–Au–PC electrodes after three weeks may be due to changes in the cellular state during recording. Spike sorting analyses confirmed that the SNR of the neural signals captured by the rSF–Au–PC electrodes was significantly higher compared to those recorded by the tungsten wire electrodes (Fig. 5e). Following a three-week period post-electrode implantation, it was consistently observed that the rSF–Au–PC electrodes exhibited heightened sensitivity for capturing high-frequency neuronal spike firing (Fig. 5f and g). To further validate the long-term in vivo neural signal recording capability of the rSF–Au–PC electrodes, the tungsten wire electrodes were removed, and all channels were connected to the rSF–Au–PC electrodes. The rSF–Au–PC electrodes continued to record neuronal signals effectively after ten weeks, with the neuronal firing amplitude remaining stable between the seventh and tenth weeks (Fig. 5h). These data indicating that the rSF–Au–PC electrodes possess superior long-term signal recording capabilities compared to tungsten wire electrodes, particularly for electrophysiological signal recordings in freely moving mice.

Fig. 5 In vivo electrophysiological experiments. (a) Configuration of the 4-channel electrode and photographs of the device, (b) schematic diagram and (c) experimental photographs of electrode implantation in the hippocampal CA1 region; in vivo spike firing recordings from the four channels, including channels 1 and 2 – tungsten wire electrodes, and channels 3 and 4 – rSF–Au–PC electrodes (d) signal status two weeks after implantation, (e) and spike sorting results; (f) signal status three weeks after implantation, and (g) spike sorting results. (h) The spike sorting results from the hippocampal CA1 region, which are stably recorded during 7–10 weeks.

Conclusion

In conclusion, a trilayer coaxial heterogeneous structure flexible neuronal electrode, known as rSF–Au–PC, has been developed through a multi-step and large-scale fabrication process. These electrodes not only demonstrate adjustable diameter and low specific impedance (∼7.67 MΩ at 1 kHz), but also possess high charge storage capacity (∼51.16 mC cm⁻²), high charge injection capacity (19.11 mC cm⁻²), and remarkable biocompatibility, which are crucial for capturing precise electrophysiological signals. The robust electrochemical and mechanical stability of our electrodes ensures reliable performance over extended periods. Moreover, the rSF–Au–PC electrode has exhibited enhanced performance relative to conventional tungsten wire electrodes in long-term in vivo electrophysiological signal recording, presenting significant implications for neuroscience applications, especially those necessitating prolonged electrophysiological surveillance. This advancement may indeed lay the groundwork for future innovations in neural prosthetics and diagnostic technologies.

Materials and methods

Materials and equipment

Silver paint (Silver Print 60%, M.E. Taylor, Engineering, Inc., USA), isoflurane (RWD Life Science Co., Ltd), Super Bond C&B super glue (Kunshan Ri Jin), Vetbond (3M, USA), paraformaldehyde saline (domestically produced, analytical pure), saline (0.9% sodium chloride in water, Qingdao Heno Bioengineering Co., Ltd), hydrogen peroxide solution (3% solution, Guangdong Hengjian Pharmaceutical Co., Ltd) are commercial available. Hot air gun (Shenzhen Chenzhou Technology Co., Ltd), Autolab (Metrohm Swizz Ltd), universal stereomicroscope (Beijing Padiwei Instrument Co., Ltd), animal respiration anesthesia equipment, stereotactic apparatus, and mini handheld cranial drill were procured from RWD Life Science Co., Ltd.

Characterization

In this work, a high-resolution field-emission scanning electron microscope (SEM, Zeiss, Germany) was utilized to observe the morphological characteristics of the silk fibroin fiber surface after treatment. Additionally, to investigate the structural changes of silk fibroin during dissolution in CaCl₂-FA and subsequent wet spinning, as well as to present the results of surface treatments on RSF, X-ray diffraction analysis (XRD) was performed on three types of fiber structures: regenerated silk fibroin (rSF), silk fibroin post-magnetron sputtering with gold (rSF–Au), and rSF–Au with post-CVD Parylene-C insulation (rSF–Au–PC).

Experimental section

Detailed experimental processes are provided in the SI.

Author contributions

J. Y. H. and S. L. conceived the idea. X. W. C., H. Q. and S. L. conceived the project and designed the experiments. J. Y. H. and X. L. L. fabricated the electrode. J. Y. H., X. L. L. and S. L. contributed to the production parameter adjustment. J. Y. H., X. L. L., J. J. J., J. B. W., S. D. Z., Y. C. L. and Y. C. L. contributed to micromorphological characterization. J. Y. H. and X. L. L. contributed to electrochemical experiments. J. Y. H., H. L. W. and S. L. contributed to mechanical experiments and FEA analysis. X. L. L. and S. L. contributed to theoretical analysis. J. Y. H. and X. L. L. contributed to biocompatibility test. J. Y. H., X. L. L., H. L. C., H. Q. and X. W. C. contributed to in vivo experiments and electrophysiological signal analysis. X. W. C. and S. L. supervised all the activities. All the authors discussed and commented on the manuscript.

Conflicts of interest

The authors declare no competing interests.

Ethical statement

All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of Guangxi University and experiments were approved by the Animal Ethics Committee of “Guangxi University Laboratory Animal Ethics Committee”.

Data availability

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

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d5nr02957k.

Acknowledgements

We acknowledge the following funding sources: National Natural Science Foundations of China (No. 32127801 and 62104051), Natural Science Foundation of Guangxi Province (No. AE31200115), and the Sichuan University Interdisciplinary Innovation Fund. We also acknowledge the Center for Instrumental Analysis at Guangxi University for providing research facilities and resources, as well as the Wuhan ServiceBio Technology Co., Ltd for their assistance with immunohistochemistry experiments.

References

1. G. Hong and C. M. Lieber, Nat. Rev. Neurosci., 2019, 20(6), 330–345.
2. G. Schiavone, X. Kang, F. Fallegger, J. Gandar, G. Courtine and S. P. Lacour, Neuron, 2020, 108, 238–258.
3. C. Won, U.-J. Jeong, S. Lee, M. Lee, C. Kwon, S. Cho, K. Yoon, S. Lee, D. Chun, I.-J. Cho and T. Lee, Adv. Funct. Mater., 2022, 32, 2205145.
4. D. Byun, S. J. Cho and S. Kim, J. Micromech. Microeng., 2013, 23, 125010–125010.
5. S. Yoo, R. Kim, J.-H. Park and Y. Nam, ACS Nano, 2016, 10, 4274–4281.
6. F. Zhang, L. Zhang, J. Xia, W. Zhao, S. Dong, Z. Ye, G. Pan, J. Luo and S. Zhang, Adv. Sci., 2022, 10, 2204467–2204467.
7. O. Karatum, E. Yildiz, H. N. Kaleli, A. Sahin, B. Ulgut and S. Nizamoglu, Adv. Funct. Mater., 2022, 32, 2109365.
8. W. Wu, H. F. Ge, L. Zhang, X. J. Lei, Y. Yang, Y. Fu and H. Feng, Chem. Res. Toxicol., 2020, 33, 2953–2962.
9. M. Xiao, X. Y. Li, S. Pifferi, B. Pastore, Y. Liu, M. Lazzarino, V. Torre, X. W. Yang, A. Menini and M. L. Tang, Nanoscale, 2022, 14, 10992–11002.
10. H. Yang, Z. Y. Qian, J. J. Wang, J. Y. Feng, C. Q. Tang, L. Y. Wang, Y. Guo, Z. W. Liu, Y. Q. Yang, K. L. Zhang, P. N. Chen, X. M. Sun and H. S. Peng, Adv. Funct. Mater., 2022, 32, 38.
11. K. Z. Wang, C. L. Frewin, D. Esrafilzadeh, C. C. Yu, C. Y. Wang, J. J. Pancrazio, M. Romero-Ortega, R. Jalili and G. Wallace, Adv. Mater., 2019, 31(15), e1805867.
12. R. T. Hassarati, W. F. Dueck, C. Tasche, P. M. Carter, L. A. Poole-Warren and R. A. Green, IEEE Trans. Neural Syst. Rehab. Eng., 2014, 22, 411–418.
13. D. N. Heo, S. J. Song, H. J. Kim, Y. J. Lee, W. K. Ko, S. J. Lee, D. Lee, S. J. Park, L. G. Zhang, J. Y. Kang, S. H. Do, S. H. Lee and I. K. Kwon, Acta Biomater., 2016, 39, 25–33.
14. C. A. R. Chapman, E. A. Cuttaz, J. A. Goding and R. A. Green, Appl. Phys. Lett., 2020, 116(1), 010501–010501.
15. J. Yu, W. Ling, Y. Li, N. Ma, Z. Wu, R. Liang, H. Pan, W. Liu, B. Fu, K. Wang, C. Li, H. Wang, H. Peng, B. Ning, J. Yang and X. Huang, Small, 2021, 17(4), 2005925.
16. Y. Zhou, H. Yang, X. Wang, H. Yang, K. Sun, Z. Zhou, L. Sun, J. Zhao, T. H. Tao and X Wei, Microsyst. Nanoeng., 2023, 9, 88–88.
17. X. Shang, W. Ling, Y. Chen, C. Li and X. Huang, Small, 2023, 19, 2302241–2302241.
18. D. Yi, Y. Yao, Y. Wang and L. Chen, J. Manuf. Process., 2024, 126, 185–207.
19. W. Yang, Y. Gong and W. Li, Front. bioeng. biotechnol., 2021, 8, 622923–622923.
20. A. Campbell and C. Wu, Micromachines, 2018, 9, 430–430.
21. A. Tanwar, H. A. Gandhi, D. Kushwaha and J. Bhattacharya, Mater. Today Chem., 2022, 26, 101153.
22. Y. Li, N. Li, N. D. Oliveira and S. H. Wang, Matter, 2021, 4, 1125–1141.
23. B. W. Zhu, H. Wang, W. R. Leow, Y. R. Cai, X. J. Loh, M. Y. Han and X. D. Chen, Adv. Mater., 2016, 28, 4250–4265.
24. D. L. Wen, D. H. Sun, P. Huang, W. Huang, M. Su, Y. Wang, M. D. Han, B. Kim, J. Brugger, H. X. Zhang and X. S. Zhang, Microsyst. Nanoeng., 2021, 7(1), 35.
25. Y. Ma, T. Huang, Z. Hu, J. Yu, L. Liu, L. Lin, M. Chen, R. Jia, X. Li, Z. Wang and Y. Fan, Carbohydr. Polym., 2025, 348, 122867–122867.
26. J. Xiao, Y. Li, J. Wang, Y. Xu, G. Zhang and C. Leng, Metals, 2023, 13(4), 745.
27. F. Zhang, Q. Lu, X. Yue, B. Zuo, M. Qin, F. Li, D. L. Kaplan and X. Zhang, Acta Biomater., 2015, 12, 139–145.
28. Y. Wang, W. Wu, R. Liu, B. Niu, X. Fang, H. Chen, M. A. Farag, L. S. Wang, G. Wang, H. Yang, H. Chen and H. Gao, Food Chem., 2025, 463, 141179–141179.
29. Y. Y. Choi, M. J. Jang, B.-D. Park and I. C. Um, Int. J. Mol. Sci., 2023, 24, 11485.
30. T. Sun, T. Tsaava, J. Peragine, C. Crosfield, M. F. Lopez, R. Modi, R. Sharma, C. Li, H. Sohal, E.H. Chang and L. Rieth, Acta Biomater., 2023, 159, 394–409.
31. T. Chen, K. S. K. Lau, S. H. Hong, H. T. H. Shi, S. N. Iwasa, J. X. M. Chen, T. Li, T. Morrison, S. K. Kalia, M. R. Popovic, C. M. Morshead and H. E. Naguib, Acta Biomater., 2023, 171, 392–405.
32. J. Ding, Z. Chen, X. Liu, Y. Tian, J. Jiang, Z. Qiao, Y. Zhang, Z. Xiao, D. Wei, J. Sun, F. Luo, L. Zhou and H. Fan, Mater. Horiz., 2022, 9, 2215–2225.
33. Y. Lu, H. Lyu, A. G. Richardson, T. H. Lucas and D. Kuzum, Sci. Rep., 2016, 6, 33526.
34. S. F. Cogan, Annu. Rev. Biomed. Eng., 2008, 10, 275–309.
35. V. S. Vajrala, K. Elkhoury, S. Pautot, C. Bergaud and A. Maziz, Biosens. Bioelectron., 2023, 227, 115182–115182.
36. X. Fu, G. Li, Y. Niu, J. Xu, P. Wang, Z. Zhou, Z. Ye, X. Liu, Z. Xu, Z. Yang, Y. Zhang, T. Lei, B. Zhang, Q. Li, A. Cao, T. Jiang and X. Duan, Front. Neurosci., 2021, 15, 771980–771980.
37. F. Vitale, S. R. Summerson, B. Aazhang, C. Kemere and M. Pasquali, ACS Nano, 2015, 9, 4465–4474.
38. D. A. Roszko, S. Mirkiani, N. Tyreman, D. Wilson, A. Toossi and V. K. Mushahwar, IEEE Trans. Bio-Med. Eng., 2022, 354–365.
39. C. L. Kolarcik, S. D. Luebben, S. A. Sapp, J. Hanner, N. Snyder, T. D. Y. Kozai, E. Chang, J. A. Nabity, S. T. Nabity, C. F. Lagenaur and X.T. Cui, Soft Matter, 2015, 11, 4847–4861.
40. M. Yang, P. Chen, X. Qu, F. Zhang, S. Ning, L. Ma, K. Yang, Y. Su, J. Zhang, W. Jiang, T. Yu, X. Dong and Z. Luo, ACS Nano, 2023, 885–895.
41. J. Wang, Z. Xiang, G. G. L. Gammad, N. V. Thakor, S.-C. Yen and C. Lee, Sens. Actuator A-Phys, 2016, 249, 269–275.
42. I. Muguet, A. Maziz, F. Mathieu, L. Mazenq and G. Larrieu, Adv. Mater., 2023, 35(39), 2302472.

---