I. INTRODUCTION

Electroceuticals that interface with nerves can be applied to both the central and peripheral nervous systems. Neural electrodes targeting the central nervous system typically interface by being implanted into the brain, whereas those designed for the peripheral nervous system establish contact by wrapping around the nerve. Neural electrodes that encircle peripheral nerves are referred to as cuff-type electrodes, and securing these electrodes firmly to the nerve is critically important. If a cuff-type electrode is attached too loosely, it may fail to acquire neural signals; conversely, if it is attached too tightly, it can damage the nerve or even lead to nerve death ^{[1, 2]}.

Fig. 1 illustrates various nerve electrode fixation methods reported in previous studies. Fig. 1(a) illustrates the suture-type fixation method, which is the most fundamental approach for securing a cuff electrode onto a nerve. This method demands significant time and effort from the surgeon and often necessitates assistance from additional personnel. One major drawback is the difficulty in controlling the appropriate suture tension. If the suture is too loose, the electrode may fail to maintain sufficient contact with the nerve, leading to inadequate signal acquisition. Conversely, if the suture is overtightened, it may compress or even damage the nerve tissue ^[3]. Additionally, due to the thin and flexible nature of the nerve electrode, the suture holes can be torn during the fixation process. Fig. 1(b) illustrates a ratchet-based fixation method ^[4]. In this approach, a wide slit is formed across most of the electrode's width, and an insertion tab with ratchet teeth on both sides is designed to pass through the slit in only one direction, effectively locking into place. This mechanism enables the surgeon to secure the electrode to the nerve with relative ease. However, the ratchet teeth are spaced at fixed intervals, resulting in discrete steps for adjusting the electrode-to-nerve fit. This limits the resolution of contact gap adjustment. In addition, since the slit occupies the majority of the electrode's width, routing conductive traces must be confined to the narrow regions between the slit edge and the electrode edge. This constraint introduces design complexity and limits layout flexibility. Fig. 1(c) illustrates the slit-insertion fixation method ^[5]. In this approach, a U-shaped slit is formed in the substrate, and the nerve becomes engaged with the protruding structures created by the slit geometry, enabling passive and automatic fixation. Despite its simplicity, this method presents several drawbacks: it is mechanically unstable and prone to dislocation, offers a minimal contact area between the electrode and the nerve, and may cause nerve folding due to interaction with the electrode edges. Fig. 1(d) illustrates a clip-based fixation method for securing the electrode to the nerve ^[6]. This approach enables the surgeon to attach the electrode with relative ease by utilizing a mechanical clip. However, the clip occupies a substantial three-dimensional volume and must be suspended from the nerve, which imposes a significant mechanical load on the nerve tissue.

Fig. 1. Various previous fixation methods for cuff-type neural electrodes: (a) suture-method, (b) ratchet-method, (c) slit-insertion method, (d) clip-method.

As demonstrated in previous studies, a simple and convenient fixation method is required to secure electrodes to nerves effectively. Specifically, the electrode should be fabricated on a thin and flexible substrate, without any bulky, clip-like three-dimensional structures. It should provide maximal contact with the nerve's outer surface, and the electrode-to-nerve gap should be continuously adjustable, rather than limited to discrete steps. Moreover, the fixation process should be simple enough to be performed by a single physician. To enable multimodal biosignal acquisition-such as sensing core body temperature in addition to neural activity-a temperature transducer must also be integrated into the electrode.

The local temperature around a neural electrode serves as a sensitive biomarker of inflammatory and metabolic activity. Peripheral nerves, which have diameters of approximately 2 mm or less, are structurally delicate and highly susceptible to inflammation. Even mild inflammatory responses can lead to nerve conduction loss or tissue degeneration, ultimately resulting in permanent nerve injury. The integrated temperature sensor enables continuous postoperative monitoring of localized temperature elevations that may indicate inflammatory responses around the implanted device. Such inflammatory reactions can persist or recur for days to months after implantation, emphasizing the need for long-term thermal surveillance. Recent chronic wireless implant studies have demonstrated that long-term temperature sensing can effectively track in vivo inflammatory trajectories, supporting the translational value of this approach for peripheral implants ^[7]. Furthermore, the integration of co-localized thermometry can extend beyond neural interface monitoring to broader health applications. The system can provide real-time detection of fever associated with infection or systemic illness, and it can also track physiological temperature variations related to ovulation for reproductive health management. These capabilities illustrate the versatility of temperature-integrated electrodes, enabling both localized tissue monitoring and general health assessment in freely moving subjects.

To address these requirements, this study proposes a thin, flexible neural electrode that incorporates a novel nerve-fixation mechanism along with an integrated temperature sensor.

II. NOVEL NEURAL ELECTRODE STRUCTURE AND FIXATION MECHANISM

Fig. 2 shows the architecture of the proposed neural electrode and its fixation onto the nerve. Fig. 2(a) presents the top-view layout of the thin, flexible electrode. The electrode consists of the following six components: circuit connection region (F1), nerve contact region (F2), ratchet teeth (F3), plastic deformation element region(F4), temperature transducer area (F5), and etched hole (F6). The nerve contact region (F2) includes four metal contact patterns (e1-e4) that are exposed to make surface contact with the outer surface of the nerve. Among these four metal patterns, two can be selectively used for neural signal acquisition, or alternatively, two can be configured for current stimulation and the other two for signal recording. Each metal pattern (e1-e4) is mapped one-to-one to the corresponding patterns (i1-i4) in the circuit connection region (F1). The width of each neural contact metal pattern (e1-e4) is 0.5 mm, and the spacing between adjacent patterns is 1.5 mm. The length of the patterns is 10 mm, which is sufficient to fully surround nerves with diameters of up to 3 mm. In previous animal experiments, the diameter of the rat sciatic nerve was measured to be approximately 2 mm, and the electrode design incorporated a margin to accommodate nerves with slightly larger diameters. The temperature transducer area (F5) is designed for mounting a surface-mount thermistor (0.6 $\times$ 0.3 $\times$ 0.3 mm, NTCG064EF104FTBX, TDK Corporation, Tokyo, Japan), electrically connected through pins i5 and i6 of the circuit connection region (F1). Thus, the circuit connection region (F1) exposes six signal lines (i1-i6) for electrical connection with external circuits. It is designed with a line width of 0.3 mm, spacing of 0.2 mm, and length of 4 mm, making it compatible with commercial connectors (505110-0692, Molex LLC, Lisle, IL, USA). The triangular-shaped sawtooth structure (F3) has a base length of 0.5 mm and a height of 1 mm. A Ti wire is inserted into the plastic deformation element region(F4), and the cross-sectional width between points CC3 and CC4 is 0.3 mm. Etch holes were introduced during the fabrication process to reduce the time required to release the neural electrode from the substrate. Each etch hole is square-shaped, with dimensions of 50 $\mu$m $\times$ 50 $\mu$m.

Fig. 2. Proposed neural electrode (schematic, not to scale): (a) top-view layout of the electrode structure, (b) illustration of the electrode fixed onto the nerve.

Fig. 2(b) shows a schematic of the proposed electrode's nerve fixation result. The electrode is secured to the nerve through the following steps: Step 1. Pre-fold the plastic deformation element region(F4) using a tool, aligning the folds precisely at the designated positions F10 and F20. The sections F11 and F21 should be pre-folded upward along the +z direction, forming an angle of approximately 45$^\circ$ with respect to the xy-plane. Step 2. Wrap the nerve contact portion (F2), which includes exposed metal surfaces, around the outer surface of the nerve to ensure surface contact. Step 3. Loosely fold the plastic deformation element region(F4) by approximately 180$^\circ$ so that it engages with the ratchet teeth (F3). This mechanism allows the distal end F9 to move in the D1 direction while preventing backward motion (D2). If the target nerve has a relatively large diameter, the F3 section may also bend when F11 and F21 are folded. Step 4. Using two forceps, gently grasp the folded region of the plastic deformation element (F9, F11, F21) with Forceps 1, and the opposite end F8 with Forceps 2. Pull F8 toward the D2 direction while maintaining a gentle hold with Forceps 1, ensuring tight contact between the electrode and the nerve. Step 5. Finally, firmly press Forceps 1 to fully fold the plastic deformation element region(F4) at the predefined fold lines F10 and F20. This action brings the surfaces F9 and F30 into contact, locking the structure in place and preventing relative motion between the electrode and the nerve.

III. NEURAL ELECTRODE FABRICATION

Fig. 3 illustrates the fabrication process flow of the proposed neural electrode. The left column presents cross-sectional views along the CC1-CC2 line, corresponding to the flat regions of the device as marked in Fig. 2(a), while the right column shows cross-sections along CC3-CC4, which include the embedded titanium (Ti) wire for fixation. The detailed process flow is as follows.

Fig. 3. Cross-sectional views of the fabrication process for the proposed neural electrode.

(p10) Trench formation: V-shaped trenches are etched into the (100) silicon substrate using KOH anisotropic wet etching to define channels that accommodate the Ti wires. The trench for the titanium (Ti) wire extends to the wafer edge. If not properly aligned with the $\langle$110$\rangle$ direction of the (100) silicon wafer, the trench becomes wider than designed due to anisotropic etching errors. To precisely determine the $\langle$110$\rangle$ direction on the silicon wafer, a fan-shaped test pattern was etched using KOH anisotropic etching ^[8]. The trenches were then aligned to the experimentally determined $\langle$110$\rangle$ direction, and the alignment error was kept within 0.01$^\circ$.

(p20) Sacrificial layer deposition: In the final step, a sacrificial layer is deposited to enable the release of the fabricated electrode structure. This layer consists of 500 nm of aluminum over a 100 nm gold base layer, deposited using thermal evaporator (UTS-150, Ultech Co., Ltd., Republic of Korea). During electrochemical etching, the aluminum is selectively dissolved by applying a constant electrical potential. However, since etching proceeds non-uniformly across the device, some regions may become fully released while others still retain portions of the aluminum layer. In such cases, areas with remaining aluminum may become electrically isolated, interrupting the flow of current required for complete etching. To prevent this, the underlying gold layer remains intact throughout the process and serves as a continuous conductive path, ensuring that all regions maintain electrical connectivity until the release is complete. To verify proper trench formation, a cross-sectional SEM analysis was performed. Fig. 4 presents an SEM image of the V-groove trench formed on the silicon wafer through anisotropic etching.

Fig. 4. Images of the silicon wafer after trench formation: (a) photograph of the wafer, (b) cross-sectional SEM image of the formed trench.

(p30) Ti wire placement and filling: The Ti wire inserted into the Si trench was straightened using a rolling process on a stainless-steel jig. Trenches patterned on the Si wafer extended from edge to edge across the wafer. Instead of using a single long Ti wire, multiple Ti wire segments were inserted into the long trenches from edge to edge. Subsequently, liquid polyimide (PI) was dispensed at the wafer edge of each trench, and capillary action drew the PI along the trench. When excess PI was applied, it was wicked away with an absorbent placed at the wafer edge. Through these steps, the trenches containing the Ti wires became completely filled with PI, yielding a planarized surface across the entire wafer.

(p40) Bottom PI patterning: A base layer of PI is spin-coated and patterned to define the lower structure of the flexible electrode. All polyimide (PI) layers used in the electrode fabrication were formed using a photosensitive polyimide (PSPI, HD-4100, HD MicroSystems, Parlin, NJ, USA) ^{[9, 10]}. Because PSPI is a photosensitive polyimide, it can be directly patterned through photolithography without the need for an additional photoresist coating.

PSPI (HD-4100) was spin-coated on the wafer using a spin coater (SSP-200) at 1,000 rpm for 10 s followed by 2,000 rpm for 30 s. Following this, a two-step soft bake was performed on a hotplate (SHAMAL HHP-411, AS ONE Corp., Osaka, Japan) at 85 $^\circ$C for 90 s and 95 $^\circ$C for 90 s. A mask aligner (MDA-600S, MIDAS SYSTEM Co., Ltd., Daejeon, Republic of Korea) exposed the wafer to UV light at 28 mW for approximately 7 s. Development used cyclopentanone as follows: i) spray at 100 rpm for 5 s, ii) puddle at 0 rpm for 15 s (steps i and ii were repeated twice), iii) co-application of cyclopentanone and AZ 1500 Thinner rinse for 5 s at 1,000 rpm, iv) AZ 1500 Thinner rinse at 1,000 rpm for 10 s, and v) spin-drying at 3,000 rpm for 15 s. Finally, thermal curing was performed on a hotplate at 200 $^\circ$C for 30 min and 375 $^\circ$C for one hour.

Polyimide has been extensively evaluated for biomedical use. According to Constantin et al. (Materials, 2019), various PI materials have shown non-toxic and stable performance in vitro and in vivo, with successful medium-term implantations such as retinal and intraneural electrodes ^[11]. Overall, the accumulated evidence supports the biocompatibility and bio-stability of PI as a substrate for neural interfaces, although comprehensive long-term clinical data remain limited.

(p50) Metal patterning: Gold (Au) patterns were formed to define the neural electrodes, interconnect lines, and contact pads. A 500-nm Au film was deposited directly onto the PI base layer without a Ti or Cr adhesion layer. No delamination of the Au film was observed throughout fabrication and subsequent processing. Conventional photolithography was not suitable for this process, as the bottom polyimide (PI) layer had a thickness of approximately 10 $\mu$m-substantially greater than the typical photoresist (PR) thickness of 1.5 $\mu$m-resulting in a large step height that interferes with uniform PR coating and resolution. To overcome this limitation, the Au patterns were formed by thermal evaporation through a shadow mask (Youngjin Astech Co., Ltd., Gumi, Republic of Korea) fabricated from a 40 $\mu$m-thick nickel (Ni) sheet.

(p60) Top PI patterning: A second PI layer is applied and patterned to encapsulate the underlying metal traces while leaving the circuit interface region (F1) and neural interface region (F2) exposed for electrical access.

(p70) Structure release: The electrode structure was released by removing the aluminum sacrificial layer through an electrochemical process known as anodic metal dissolution ^[12]. The wafer was immersed in a 1 M NaCl aqueous solution. The aluminum layer on the wafer was connected to the anode through a 100 $\Omega$ series resistor, while a platinum electrode, serving as the cathode, was immersed in the same solution. A constant voltage of 2 V was applied between the two electrodes. Initially, a closed electrical loop was established through the power supply, the series resistor, the aluminum layer, the NaCl solution, and the platinum electrode. As the aluminum layer gradually dissolved electrochemically in the NaCl solution, current flowed through the circuit. Part of the applied 2 V potential dropped across the resistor (V_r), and the remainder appeared as the potential difference (V_w) between the aluminum and platinum electrodes. At the start of the process, V_w was approximately 1.2 V. After about 4 hours, when the aluminum had been completely etched, V_w increased to 2 V. Once the aluminum forming the current path was fully dissolved, the closed loop was disconnected, and the current dropped to zero. Consequently, V_r across the resistor fell to 0 V, while V_w rose to 2 V. Therefore, when a sudden increase of V_w to 2 V is observed during voltage monitoring, it indicates that the aluminum etching process has been completed.

Finally, a temperature transducer (thermistor, 0.6 mm $\times$ 0.3 mm $\times$ 0.3 mm, NTCG064EF104FTBX, TDK) for body temperature measurement was mounted. To ensure electrical insulation from the surrounding tissue, the entire thermistor was passivated with PI. The thermistor was bonded to the metal pad using an anisotropic conductive adhesive (ACA, DELO MONOPOX AC268, DELO Industrie Klebstoffe GmbH & Co. KGaA, Windach, Germany) by thermo-compression at 190 $^\circ$C for 6 s.

To electrically insulate the thermistor from surrounding biological tissues, the thermistor was passivated with a polyimide (PI) layer. A PI plate larger than the outer dimensions of the thermistor was placed on top of it. Liquid PI was then injected along the sides of the thermistor, allowing capillary forces to fill the gap between the substrate and the top PI plate. The assembled film was soft-baked at 200 $^\circ$C for 30 min and subsequently cured at 375 $^\circ$C for 60 min to complete imidization.

Fig. 5 shows the fully fabricated neural electrode, with key functional regions and their outer dimensions indicated. The outer dimensions of the designated regions are approximately 29 mm $\times$ 15 mm. The thickness of the PI/Au/PI laminate layer is approximately 15 $\mu$m.

Fig. 5. Photograph of the fully fabricated neural electrode.

IV. ELECTRODE CHARACTERIZATION

Fig. 6 shows the test results of the fabricated electrode's fastening function on a nerve-mimicking wire. The wire used in the test had a diameter of 1.5 mm. The electrode was secured to the wire by folding the Ti wire, which serves as the plastic deformation element of the electrode. Owing to its very thin structure, the fabricated electrode appears slightly transparent.

Fig. 6. Fabricated neural electrode secured to a nerve-mimicking wire by folding a plastic deformation element.

To evaluate the electrochemical characteristics of the neural contact of the fabricated electrode, impedance, charge storage capacity (CSC), and charge injection capacity (CIC) were measured. For the temperature transducer characterization, sensitivity, linearity, and measurement error over the operating temperature range were assessed, along with long-term leakage current measurements to verify the passivation performance of the thermistor.

Fig. 7(a) shows the impedance of the fabricated electrode as a function of frequency, measured using electrochemical impedance spectroscopy (EIS) with a Gamry Reference 600 device (Gamry Instruments, Warminster, PA, USA) ^[13]. Since the primary spectral power of neural signals lies in the kilohertz range, the electrode impedance is often represented by its value at 1 kHz. For an electrode measuring 0.5 mm in width and 10 mm in length, the measured impedance at 1 kHz was 530 $\Omega$, corresponding to 2.64 k$\Omega$$\cdot$mm². This value falls within the range reported in the literature for implantable neural interfaces. For example, González-González et al. developed thiol-ene/acrylate SMP-based cuffs with Au/TiN electrodes showing impedances of 1-3 k$\Omega$ (at 1 kHz), while Dong et al. reported PEDOT:PSS recording electrodes on a parylene-C/Au platform with approximately 1 k$\Omega$ (at 1 kHz) ^{[14, 15]}.

Fig. 7. Electrochemical characterization of the fabricated electrode: (a) impedance spectrum curve measured from electrochemical impedance spectroscopy (EIS), (b) charge storage capacity (CSC) curve measured from cyclic voltammetry (CV), and (c) charge injection capacity (CIC) curve measured from current pulse measurements.

Fig. 7(b) shows the cyclic voltammetry (CV) results for the fabricated electrode ^[16]. The voltage was swept at a rate of 100 mV/s using a Gamry Reference 600 device. The CSC in the positive region of the curve was 79 $\mu$C/cm², while that in the negative region was 151 $\mu$C/cm², yielding a total CSC of 230 $\mu$C/cm².

Fig. 7(c) shows the charge injection capacity (CIC) characteristics, which are commonly used to evaluate the stimulation limit of an electrode ^[17].

In addition to the results obtained under the unwrapped condition, when the electrode was wound around an insulated wire mimicking a nerve, the impedance slightly increased to 577 $\Omega$ (2.87 k$\Omega$$\cdot$mm²), and the charge-storage capacities were CSC_A = 61.7 $\mu$C/cm² and CSC_C = 164.5 $\mu$C/cm², resulting in a total CSC of 226.2 $\mu$C/cm². The observed variability in the measurements reflects both reproducibility limits and changes in the interface conditions between the electrode and PBS. As a representative characterization of the electrode itself, measurements performed with the electrode laid flat in PBS are preferable, because they exclude structural interface variations introduced by the mounting geometry and thus better capture intrinsic electrode properties. When the electrode is wrapped around a nerve phantom, part of the electrode surface is masked by the phantom rather than being fully exposed to PBS, which alters the effective electrochemically active area.

A stimulation current pulse was applied, and the resulting electrode voltage was measured using an oscilloscope (MSOX4104A, Keysight Technologies, Santa Rosa, CA, USA). The water window, defined as the stable potential range within which water electrolysis does not occur, was conservatively set to $\pm$0.6 V. The negative boundary (-0.6 V) was reached when a current of 700 $\mu$A was applied, corresponding to a CIC of 14 $\mu$C/cm².

The charge injection capacity (CIC) was measured by fixing the current pulse width at 1 ms while gradually increasing the current amplitude and recording the resulting electrode voltage. When the voltage reached the water electrolysis potential of 0.6 V, the applied current was 700 $\mu$A, which represents the upper limit of the charge injection level applicable for neural stimulation at a 1 ms pulse width. Heo et al. applied a biphasic current pulse to stimulate the rat sciatic nerve, using a pulse amplitude of 300 $\mu$A and a pulse duration of 100 $\mu$s ^[18]. Thompson et al. also reviewed 33 studies related to neuroplasticity and rehabilitation and reported that most studies used biphasic pulses with an amplitude of 800 $\mu$A and a duration of 100 $\mu$s ^[19]. When normalized to a 1 ms pulse duration, this corresponds to an equivalent current of approximately 80 $\mu$A. In our experiments on rat sciatic nerves, the rheobase current and chronaxie time were measured to be 40.8 $\mu$A and 100 $\mu$s, respectively. Taken together, these findings indicate that a charge injection capability corresponding to approximately 800 $\mu$A at a 1 ms pulse duration lies within a practically applicable range for sciatic or vagus nerve stimulation.

Fig. 8 shows the output response and measurement error of the thermistor integrated into the fabricated electrode as a function of temperature. The electrode was placed in a temperature chamber (PKK-50, Kambic d.o.o., Semič, Slovenia), and measurements were performed at 1 $^\circ$C intervals over a range of 34 $^\circ$C to 43 $^\circ$C. A platinum thermometer (CTP5000-250, WIKA, Klingenberg, Germany) and a precision temperature reading device (UT-ONE B03B, Batemika d.o.o., Ljubljana, Slovenia) served as reference sensors to determine the true applied temperature. The thermistor was connected to a custom-designed ROIC, and its output was digitized using a time-to-digital converter (TDC) ^[20].

Fig. 8. Thermistor characterization results: (a) TDC output as a function of reference temperature with first-order interpolation, and (b) temperature measurement error of the thermistor relative to the reference temperature.

Fig. 8(a) shows the TDC output as a function of the applied temperature. First-order linear interpolation yields the regression equation y = 810.35x - 31403, with a correlation coefficient (R²) of 0.998. The sensitivity, defined as the output change per 1 $^\circ$C variation, is 810.35 digits/$^\circ$C, and the nonlinearity is 1.8 % FS. Fig. 8(b) presents the errors relative to the first-order interpolation, all of which are within $\pm$0.1 $^\circ$C. These errors mainly arise from the nonlinearity of the thermistor, the nonlinearity of the readout circuit, and electrical noise. The observed errors conform to the performance standards for medical thermometers specified in ASTM E1112-00 ^[21]. According to this standard, the permissible error is $\pm$0.1 $^\circ$C for 37.0-39.0 $^\circ$C, $\pm$0.2 $^\circ$C for 35.8-<37.0 $^\circ$C and > 39.0-41.0 $^\circ$C, and $\pm$0.3 $^\circ$C for temperatures below 35.8 $^\circ$C or above 41.0 $^\circ$C.

To evaluate the long-term stability of the thermistor passivation, leakage current was monitored over a 150-day period. The fabricated electrode was immersed in phosphate-buffered saline (PBS), with a voltage of 1.8 V applied to the thermistor, and a platinum electrode placed in the PBS to measure leakage current. The PBS was left at room temperature, which was around 25 $^\circ$C on average during the experiment. Throughout the 150 days, the leakage current remained below an average of 1 nA, indicating stable insulation performance.

Table 1 compares the function and performance of the electrode developed in this study with those reported in previous works. While ratchet fixation is more convenient than conventional surgical suturing, it inevitably leaves a gap between the nerve and the electrode ^[4]. This gap is determined by the pitch of the ratchet teeth and can only be adjusted in discrete increments. The slit-insertion method, in which the nerve is placed between protrusions formed by a U-shaped slit in the substrate, is simple but prone to loosening, and the small contact area between the electrode and the nerve can result in poor signal quality ^[5]. The three-dimensional clip method provides complete engagement with the nerve but places a bulky structure on it, imposing an excessive mechanical load ^[6]. The electrode developed in this study does not require a separate three-dimensional clip, enables broad contact with most nerve structures, and accommodates nerve diameters of varying sizes in a continuous manner. While previous studies on electrodes employing novel fixation methods featured a single sensing mode, the electrode developed in this study supports a multi-sensing mode that also enables body temperature measurement.

Electrochemical parameters such as electrode impedance, CSC, and CIC are key metrics for evaluating suitability in neural interface applications. All impedance values were normalized to an equivalent electrode area of 1 mm² using the inverse proportionality relationship between impedance and electrode area ($R \times A$ = constant).

Although the electrode presented in this study demonstrates several advantages over previously reported designs, it still shares a fundamental limitation common to other wrapping-type electrodes. During implantation, the electrode must be manually wrapped around the nerve, and the contact pressure must be carefully controlled-insufficient tension may lead to unstable electrical coupling, whereas excessive pressure can cause nerve compression or injury. To overcome this limitation, future studies should focus on developing quantitative or self-adjusting fixation mechanisms that can automatically achieve optimal wrapping force while minimizing dependence on the operator's skill.

The mechanical robustness of the Ti wire-based wrapping and locking mechanism was evaluated through repeated deformation and preliminary retention testing. Although the device is designed for single-use fixation, a complete fold-unfold cyclic test showed fracture after about 30 cycles, confirming sufficient strength for implantation while indicating a finite fatigue limit of the plastically deformable wire. In addition, a PBS-wetted electrode wrapped around a nerve phantom exhibited no backward slippage or disengagement during handling, demonstrating stable fixation. Nevertheless, future quantitative evaluations-such as measuring pull-out force and slip onset load-are required to further validate the fixation performance.

V. CONCLUSIONS

This study presents a thin, flexible neural electrode that integrates a plastically deformable Ti wire fixation mechanism with an embedded temperature sensing element. The fixation method enables continuous adjustment of the electrode-to-nerve gap, ensuring secure contact without the discrete gap limitations of bulky three-dimensional clips or conventional ratchet-based designs. A novel fabrication approach was developed for mounting sub-millimeter Ti wires onto a 15 $\mu$m-thick PI-based flexible substrate. The fabricated electrode was successfully applied to a nerve-mimicking wire, demonstrating reliable fixation performance. Electrochemical testing yielded an impedance of 530 $\Omega$ (normalized: 2.64 k$\Omega$$\cdot$mm²), a CSC of 230 $\mu$C/cm², and a CIC of 14 $\mu$C/cm², which are within the range applicable to neural interfaces. The integrated thermistor exhibited high linearity (R² = 0.998) and sensitivity (810.35 digits/$^\circ$C) across 34-43 $^\circ$C, with errors within $\pm$0.1 $^\circ$C, satisfying ASTM E1112-00 standards. Long-term PBS immersion confirmed stable passivation, with leakage currents remaining below 1 nA over 150 days. By combining multimodal sensing with a simple, surgeon-friendly fixation mechanism in a compact, flexible form factor, the proposed electrode addresses key limitations of existing designs. This platform holds strong potential for future implantable neural interfaces requiring both stimulation/recording and continuous physiological monitoring.