Flexible, Conductive Fabric‐Backed, Microneedle Electrodes for Electrophysiological Monitoring

Microneedle‐based electrodes have attracted significant attention due to their potential applications in high‐quality, long‐term electrophysiological monitoring. In this study, micromoulding technology is used to develop a flexible, conductive fabric microneedle (CF–MN) electrode by bonding a conductive textile to polymeric microneedle arrays. The resulting electrode conforms to the curved surfaces of the body, and the use of a conductive textile backing reduces metallization time and costs by 50% over double‐sided dry electrodes. The electrode‐skin impedance of the CF–MN electrodes is significantly lower than that of gel electrodes over a wide frequency range, and tests on healthy volunteers, both at rest and under a range of ambulatory conditions, showed that the electrodes are capable of acquiring electrocardiography (ECG) and electromyography (EMG) signals that have a comparable kurtosis and signal‐to‐noise ratio to those recorded using conventional electrodes, but without the need for skin preparation or the use of a wet gel. The biomechanical properties of the electrode are also characterized. This is the first demonstration of CF–MN electrodes that can ultimately be integrated into clothing, which in turn can facilitate personalized cardiac health management by enabling continuous, long‐term ECG/EMG monitoring during daily activities.


Introduction
3] Electrocardiography (ECG) and electromyography (EMG), which measure electrical signals from the heart and skeletal muscles, respectively, are amongst the simplest and most widely used procedures for screening cardiovascular diseases (including DOI: 10.1002/admt.202301606arrhythmias, fibrillation, or heart failure), [4] and muscular activity (e.g., for motor disability and neuromuscular disorder detection, or for prosthesis control). [5]The use of lightweight, wearable devices to acquire these surface biopotential measurements is an attractive solution for applications in health and lifestyle; these monitors usually incorporate electrodes, an electronic acquisition system, signal processing software, and machine learning algorithms. [2]lectrodes are crucial for obtaining physiological signals by converting ionic current to electronic current. [6]The electrode-skin impedance (ESI) is a key parameter of such electrodes, [1] with undesirably high ESI leading to lower signal quality and signal-to-noise ratio (SNR).Wet gel electrodes are commonly used in clinical and research settings, and use an electrolytic gel to establish a low impedance connection with skin.This gel tends to become dry over time, and the use of wet electrodes also often requires skin preparation such as shaving or abrasion to further reduce the skin-electrode impedance. [7]However, skin preparation can be uncomfortable, time-consuming, and may cause skin irritation or infection.This can be especially problematic for premature babies, people with sensitive skin, and the elderly. [8,9]ry electrodes offer improved comfort, stability, and signal quality compared to wet gel electrodes, making them a promising solution for long-term monitoring, sports and lifestyle applications, and telemedicine. [10,11]Researchers have proposed several types of dry electrodes, including those made from stiff materials, soft/flexible materials, and textiles.[14] Conductive fabrics woven, embroidered, or knitted into clothes may also be used as electrodes.Nonetheless, the issue of stretching and the potential for loose skin-electrode contact during movement represent significant challenges that can result in motion artifacts. [15,16]These challenges pose a major obstacle to obtaining electrophysiological signals that are free from artifacts.The CF-MN electrode described here can offer a potential solution to some of these issues.E hc is the half-cell potential between electrode and skin; R hc and C hc are the charge transfer resistance and equivalent capacitance; R sc is the series resistance of gel, moisture, and sweat; R sc and C sc are the equivalent resistance and capacitance of epidermis layer; Rtissue is the resistance of dermis and underlying tissues.
25][26][27][28][29] Of particular interest is the development of flexible microneedle electrodes, which are more tolerable for long-term applications as they can conform to the curved body surface and lessen the possibility of causing pain or irritation.These electrodes could be integrated into textiles or "smart garments" and their flexible nature could also reduce the risk of needle failure.Figure 1 illustrates the use of equivalent circuit models for these three electrode types, comprising resistors (R) and capacitors (C) arranged in parallel and series combinations to approximate the electrical impedance of the skin.These models aim to capture the effective behaviors of the skin's complex structures and the properties of its different layers in contact with the electrodes.
9][30][31][32][33][34][35][36][37][38][39][40] Micromoulding technology is a simpler and cost-effective alternative that enables the production of sharp and precise microneedles.We have previously demonstrated a double-sided micromoulding procedure for the production of sharp-tipped microneedle elec-trodes that also incorporated through-wafer vias to facilitate metallization at the wafer scale. [41]n this study, we adapt that micromoulding process to introduce flexible biopotential dry electrodes based on conductive fabric-backed microneedles with a sharp tip, called CF-MN.These electrodes can record high-quality signals from the heart and muscle and are mechanically robust by maintaining good contact.This simple fabrication process is the first-ever report on the fabrication of CF-MN through micromoulding technology for surface electrophysiological monitoring and further has the advantage of only requiring metallization on the front face of the electrode, reducing metallization costs and time by ≈50% compared with standard, double-sided electrodes.We combine the CF-MN electrode with specialized ECG and EMG recording systems to enable biopotential recording with high SNR and low motion artifacts.Tests on healthy volunteers, both at rest and under a range of ambulatory conditions, show that the electrodes are capable of acquiring ECG and EMG signals with quality comparable to those recorded using commercial wet electrodes.Our proposed technology is expected to facilitate personalized cardiac health management by enabling continuous, longterm ECG/EMG monitoring during daily activities.

Mechanical Characterization
The flexibility and compressibility of CF-MN arrays were assessed utilizing an Instron tensile testing machine (Model 5565, Instron, Buckinghamshire, UK).Three samples with dimensions of 2 × 6 cm 2 were prepared for machine testing.A setup was developed for stretching and bending tests as described in reference, [27] and data was recorded at 1 mm min −1 with five samples per second.
An axial compression test for strength conformance by applying pressure to the array with a probe that moved downward at a speed of 0.01 mm −1 s until a maximum force of 35N was reached The mode of failure of the microneedles was characterized using a scanning electron microscope and force/displacement curve analysis.

Skin Insertion Test
Skin insertion tests were performed using a microneedle array and an impact inserter device. [34]Ex vivo human abdominal skin, discarded after cosmetic surgical procedures, was obtained following ethical approval by the Clinical Research Ethics Committee of the Cork Teaching Hospitals (CREC) and informed consent from all patients (approval reference ECM 4 (aa) 06/08/13).This was incubated in PBS before being clamped in a customized 3Dprinted jig for controlled stretching.A total of five plastic discs of 1 mm thickness were placed underneath the skin sample to stretch and dome the surface into a shape approximating that of in vivo skin.The CF-MN electrode was applied to the skin sample with an impact energy of ≈0.38 J cm −2 and 5 N residual force. [34,36]

In Vivo Electrode-Skin Impedance Characteristics
In accordance with policies and procedures of the Clinical Research Ethics Committee (CREC) of the Cork Teaching Hospitals, the biomechanical and electrical characteristics of the electrodes were measured in a small cohort of healthy volunteers.The study protocol was approved by CREC (ECM 3 (ZZ) 10/11/20), and all participants provided written consent.
The electrode skin impedance was measured using an Agilent E48980A LCR meter and customized LabVIEW software.The standard wet gel and CF-MN electrodes were placed on the anterior and posterior antebrachial region of the forearm to avoid hair shaft interference and maintain consistent conditions.Both types of electrodes were applied with light pressure using thumb and index fingers.The impedance was recorded from six healthy volunteers over a wide frequency range of 20 Hz-1 kHz, with a 100 Hz interval.Although ECG bandwidth typically ranges from 0.05 Hz to 150 Hz, in this case, ESI was measured from 20 Hz due to equipment limitations.The skin surface was not shaved or abraded prior to recording, and measurements were performed at room temperature.

Skin Compatibility Test
To check for skin irritation, CF-MN, and wet-gel electrodes were placed adjacent to each other on the forearm of a healthy subject with a 5 cm distance between them and left for 48 h, while a medical grade tape was placed over the electrodes for additional support.The skin was then examined for signs of erythema. [48]

ECG Recording: Hardware and Signal Analysis
The performance of the CF-MN electrodes was assessed by recording ECG signals from the subjects in static and dynamic conditions using the customized dual-channel ECG device previously reported. [42]Schematic, PCB, CAD files, and PCB readout module details can be found in Figure S2A-D (Supporting Information).To test the anti-artifact capability of the CF-MN electrodes and wearable recording system, ECG signals were recorded simultaneously with CF-MN and gel electrodes on three volunteers (two males and one female: age -35 ± 12 years; height -168 ± 7 cm; weight -68 ± 24 kg).The volunteers wore dual-channel ECG wireless recording systems with CF-MN and gel electrodes in the lead-I configuration.Medical grade tape was used to stabilize the recording system during ECG signal recording.
The recordings were performed at 500 samples per seconds (sps) for 1 min before and after exercise and for 2 min and 30 s during exercise (walking at 2, 4, and 6 km h −1 ) as recommended by the American Heart Association and Nyquist criterion. [43]The two sets of electrodes were placed adjacently on the left and right chest, with a distance of ≈6.5 ± 0.4 and 6.4 ± 0.6 cm between the centers of the electrodes for left and right sides, respectively.Kurtosis (kSQI), signal-to-noise ratio (SNR), and the standard deviation (SD) of the QRS complex were used to assess the performance of the CF-MN electrodes.A kurtosis larger than five indicates clean, sinus rhythm ECG, [44] while a kurtosis ≈5 indicates muscle artifact, and a kurtosis lower than five indicates baseline wander and powerline interference. [45]A low kSQI usually indicates low-frequency noise such as Gaussian (thermal observation) noise, baseline wonder, and power-line interference.
where kSQI represents the kurtosis as a signal quality index, μ x and  are empirical estimates of the mean and standard deviation of x respectively, and x is the ECG signal.
The SNR was calculated as follows: where A signal pp and A noise pp represent the peak-to-peak ADC counts of the QRS wave and noise in the isoelectric region between the T and P waves, respectively.
SD of the QRS complex could quantify the overall variability and dispersion of the amplitude values within the complex, which could be an indicator of the presence of noise.A higher SD suggests a greater variation and was often associated with a noisier signal.Conversely, a lower SD indicates less variability and a cleaner signal. [29]Labview 2020 SPI (32-bit) was used to calculate the values of kurtosis, SNR, and SD of the QRS complex.

EMG Recording: Hardware and Signal Analysis
A dual-channel EMG sensor (SEN0240 from DFRobot/ OYMotion) was used for the simultaneous recording of EMG   signals from both the wet gel and CF-MN electrodes.The EMG sensor was integrated with an ESP32-WROOM-32D microcontroller using a general-purpose PCB board for wireless transmission and processing, enclosed in a 3-D printed housing and powered via USB.A custom-built Labview GUI software was used for real-time display and recording of the data.The device was lightweight (54.54 grams) and compact for ease of EMG signal recording (see Figure S3A,B, Supporting Information).This recording system was used to evaluate the performance of the CF-MN electrode against that of the gel electrode.CF-MN and wet gel electrodes were applied ≈6 cm apart in the flexor digitorum profundus (FDP) region of the forearm. [46]The EMG data was recorded from three subjects (three males with age, height, and weight of 32 ± 6 years, 167 ± 5 cm, and 79 ± 14 kg, respectively) for ≈70 s while subjects were performing various tasks like opening and closing of the fist, and flexing of the pinky, ring, middle, index, and thumb fingers.A 20 Hz high-pass FIR digital filter and full signal rectification were performed to remove the low-frequency motion noises and aggregate the negative part of the recorded raw EMG data.The high pass filter from 20 Hz was selected because normally 80% of the muscular energy is concentrated between 20 Hz and 450 Hz. [47] Meanwhile, a 50 Hz digital notch filter was used to remove the power line interference.The root mean square (RMS) values of the signal and noise data of all extracted segments from each EMG signal were calculated to reflect the average amplitude.The SNR was estimated as follows: where RMS signal and RMS Noise depict the root mean square values of signal and noise regions when the subject contracted and relaxed muscles.The data analysis involved the use of Origin Pro (Origin 2021b) to perform calculations such as mean, standard error, and paired t-tests.Additionally, Labview 202 SPI (32-bit) was used to calculate root mean square (RMS) and SNR.

Materials
Sylgard 184 polydimethylsiloxane (PDMS) was purchased from Dow Corning Inc., USA.Medical grade EPO-TEK 353ND epoxy resin and medical grade MED-H20E conductive epoxy was acquired from Epoxy Technology Europe Ltd.Conductive fabric (NCS95R-CR, a nylon ripstop fabric covered by nickel, copper, silver, and a waterproof layer) with a thickness of 127 μm ± 15%, DC surface resistivity of <0.01 Ω sq −1 ., and temperature ranges of −40 to 100 °C was sourced from Marktek Inc. USA.Medical grade, double-sided, a conductive adhesive (ARcare 90 366, with a thickness of 1.3 μm and low volume resistance of ≈10Ω) was received as a sample from Adhesives Research, Inc., USA.We also bought Micropore Medical Tape, thickness 25 000 μm, from 3 m Ireland Limited, Ireland that was used for attaching the dry electrodes onto the skin.3 m Red Dot (2239) Monitoring Electrodes were procured from Promed, Tulligmore, Ireland.Male ECG snap buttons were purchased from PMSOEHT Professional Accessories, China.

Design, Fabrication, and Assembly of Electrodes
The flexible microneedle array structure for CF-MN electrodes was fabricated using a micromoulding process as shown in Figure 2. To create a mold, a 100 mm diameter silicon wafer was used as a master template, on which was etched an array of 500 μm tall microneedles consisting of eight (263) silicon planes, a base of (212) planes, and a height: base diameter aspect ratio of 3:2. [6,41]The needle-to-needle pitch was 1750 μm and the tip radii were typically between 50 and 100 nm.PDMS was mixed according to the manufacturer's instructions, poured over the master template, and left to degas at room temperature for 24 h.The PDMS was then cured at 80 °C for 1 h, cooled, and peeled from the master template.Thereafter, a biocompatible epoxy resin (EPO-TEK 353ND) was mixed in a 10:1 ratio and poured onto this front mold.This material has previously been used for microneedle manufacture. [17,35,41]It was left in a vacuum chamber to degas the epoxy at 200 mTorr for ≈15-20 min.The mold was then placed on a vacuum table for ≈30-45 min, during which time a spatula was used to spread the epoxy resin and fill the microneedle pits.
While excess epoxy was removed after the microneedle pits were filled, in order to improve the adhesion of conductive fabric to the microneedles a thin layer of epoxy resin (≈40% of the original deposited volume) was left to remain on the front mold surface.To form a rear backing layer and define the geometrical dimensions of the electrodes, a circular ring shape with internal and external diameters of 18 and 19.14 mm was created using Microsoft PowerPoint and saved as a JPG file (Figure S1, Supporting Information).The file was then imported into Cricut Design Space (v7.20.88) and cut into the conductive fabric using a Cricut Maker cutting machine (CXPL301) equipped with a fine point blade.Next, the cut CF sheet was placed over the filled front mold and left on a vacuum table for ≈30-45 min.A petri dish cover and weight were placed over the rear mold.After curing in an oven at 80 °C for an hour, the CF and microneedles were detached from the front mold.This process resulted in a substrate thickness of ≈165 ± 5 μm that had good flexibility.
Before metallization, the electrodes were cleaned with isopropyl alcohol and dried with a nitrogen gun.A Temescal UEFC-4900 e-beam system was used to deposit an adhesion layer of 20 nm thick titanium, followed by a 150 nm thick gold layer.The evaporation process was performed exclusively on the front side of the flexible CF-MN, while the electrical connection was established from the front to the rear side of the conductive fabric via the gap created by cutting out a circular ring from the fabric as presented in Figure S1 (Supporting Information).
The electrodes were cut from the rear backing layer using a scalpel or fine blade.For testing purposes, they were then affixed to 3 m Red Dot Monitoring Electrodes, from which the wet gel pad was removed, using conductive adhesive.Sterilization was performed with UV-ozone for 5 min at a power of 1380 mW cm −2 before the electrodes were packed and stored in plastic cases.

Materials, Structural, and Fabrication Approach
Figure 2A shows the basic structure of the CF-MN electrode, which includes arranging ≈100 microneedles on an 18 mm di-ameter conductive fabric.The microneedles have a tip radius of ≈5 μm, height of 500 μm, and pitch of 1750 μm.The NCS95R-CR fabric worked well once the amount of epoxy resin necessary to ensure bonding between the rigid microneedle and the fabric was optimized.Bonding was confirmed by casting epoxy resin onto a fabric sample (5 × 7 mm 2 ), curing it, and analyzing it using SEM. Figure S4A,B (Supporting Information) showed that the epoxy has diffused into the fabric, indicating a good bond between the fabric and the epoxy resin.Ti/Au metal layers were used for signal transmission on the front face of the electrodes, including the microneedles and sidewalls.This single-sided metallization approach reduces processing time and cost by 50% compared to earlier studies. [23,41]s depicted in Figure 2B, the CF-MN electrode, when placed on the skin, demonstrates the capability of the microneedles to effectively traverse the stratum corneum (SC) and establish direct contact with the epidermis.This contact is achieved without coming into contact with the underlying dermis and deeper tissues, which are home to nerves and blood vessels.The manufacturing process of the CF-MN electrode is illustrated in Figure 2C.
Figure 3A-D illustrates the fabricated CF-MN electrode, showcasing the metal coating on the front face (Figure 3A) as well as the nonmetallized rear side (Figure 3B).The figures also demonstrate the successful reproduction of the master template structures.Figure 3D and Figure S5, Table S1 (Supporting Information) show that the mean tip diameter of 4 ± 2 μm is consistent with previous studies [36,49] and sharp enough to penetrate the skin.The robustness of the fabrication process was confirmed by Figure 8.The 3 s segment of recorded ECG signal using CF-MN and gel electrodes of subject 1. A) Electrode placement scheme for ECG signal recording using CF-MN and gel electrodes.B) ECG signal in a sitting position before exercise, C) ECG signal when subject was standing on a treadmill before the exercise, D) ECG signal when subject was walking at 2 km h −1 , E) ECG signal when subject was walking at 4 km h −1 , F) ECG signal while subject walking at 6 km h −1 , G) ECG signal when standing on the treadmill after exercise, H) ECG signal when subject was sitting on a chair after exercise.r: linear correlation coefficient estimated for 3 s signal for distinct postures.
repeatedly micromolding the same silicon master template over 100 times, without any deformation or loss of morphology.

Flexibility Test
Figures 4A,B illustrate the experimental setup and the stressstrain curve concerning stretchability.In Figure 4B, it can be observed that three samples displayed a stretchability of (3.6 ± 0.6)%.Notably, Figure 4B shows a distinct reduction in stress at 11 N mm −2 , suggesting a failure in the fabric and epoxy.However, it is worth mentioning that the microneedles themselves remained intact, as depicted in the inset image of Figure 4A.Bending tests as shown in Figure 4C,D demonstrate that the electrodes may be deformed into shapes with small radii of curvature without damage or delamination of the materials.It should also be noted that forces or deformations such as those investigated here should never arise during routine application and use of the electrodes, thereby allowing for a significant margin of safety during normal wear.

Compression Test
Compression force measurements were conducted on a 16mm diameter array of 52 microneedles.The test revealed that Figure 9. Quantitative comparison of 15 s of regular segments for each subject in sitting and standing positions before exercise, walking at speeds of 2, 4, and 6 km h −1 , and standing and sitting after exercise, from left to right, subject 1, 2, and 3; A) kurtosis estimated for each subject; B) SNR calculated for each subject; C) standard deviation (SD) of QRS complex for each subject.
microneedles remained intact even under forces significantly higher than those needed for skin penetration, as shown in Figure 5 (Left).The combination of a polymeric microneedle tip and a strong flexible base prevented the microneedles from breaking, instead causing them to bend as shown in Figure 5 (Right).

Skin Insertion Test
To verify skin penetration, 1% w/v methylene blue dye was applied to the skin after the insertion test and left for 2 h before being cleaned off.Methylene blue is a surgical dye with a high protein affinity, which binds to the epidermal tissue but not the hydrophobic stratum corneum layer.It is commonly used to assess microneedle penetration, as these blue marks indicate a suc-cessful breach of the stratum corneum layer. [40]In a sample of 16 microneedles, all 16 were able to penetrate the skin with a residual force of 5 N, indicating a 100% penetration efficiency, as shown in Figure 6.This suggests that a maximum force of ≈0.31 N is required for a microneedle to penetrate the skin, which is in agreement with an earlier study. [36]

Electrode Skin Impedance Measurement
The CF-MN electrode demonstrates a low impedance due to its microneedles effectively penetrating the highly resistant outer layer of the skin, known as the stratum corneum (SC).This penetration results in the formation of a coupling capacitance, which contributes to the overall impedance at the interface between the electrode and the skin.The impedance measurements were conducted within a frequency range of 20 Hz-1000 Hz, with data points collected at 20 Hz intervals.In Figure 7A, the impedance values (mean ± SE) obtained from six subjects using CF-MN and gel electrodes are displayed.Figure 7B illustrates the impedance values at specific frequencies, providing a clear representation of how the impedance changes as the frequency increases.To provide further insight, Figure 7C,D visually depict the appearance of microneedles after removal from one subject.Additionally, Figure S6A (Supporting Information) provides the impedance data for each individual subject.
The ESI analysis showed that the CF-MN electrodes with Au surface had superior performance compared to the gel electrodes, especially at low frequencies (≤700 Hz), with a lower estimated standard error.The impedance of both CF-MN and gel electrodes decreased with the increase in frequency.The microneedle structure and Au metallization contributed to an ESI of only (36.4 ± 2 kΩ) at 20 Hz for CF-MN electrodes, which was substantially lower than that of the gel electrode (152.4 ± 30 kΩ).Moreover, throughout the lower tested frequency range (<700 Hz), CF-MN electrodes consistently displayed lower total impedance than gel-based electrodes, and also had lower impedance values compared to previous studies [25][26][27]30,40,41,50,54], and their dry nature makes them more suitable for long-term use. These advantagesmake CF-MN electrodes suitable for recording high-quality electrophysiological signals with short frequency ranges and low amplitude.[51]

Surface Resistance, and DC Offset Voltage,
The surface resistance measured using a four-probe system (Figure S6B, Supporting Information) was ≈0.5 Ω, which is consistent with an earlier study. [52]This implies that the high conductivity of the Au metalized electrode allowed for high-quality signal recording.The schematic representation of the four-probe system for measuring surface resistance and DC offset voltage is illustrated in Figure S6B,C (Supporting Information), respectively.The wet electrode had a DC offset voltage of 12 mV, while the CF-MN electrodes exhibited DC offset voltages of ≈9 mV, which was lower than in an earlier study. [52]

Skin Compatibility Test
The materials used for CF-MN electrodes are biocompatible.A 48-h continuous test on a human volunteer showed no inflammation or allergies, and marks caused by the microneedles

Electrophysiological Signal Measurement
Figure S8 (Supporting Information) illustrates the setup for ECG recording.As described in Section 2.3, seven activity states were considered and the data is shown in Figure 8.It is evident that the signals obtained from both types of electrodes exhibit a remarkable resemblance.The key features of the P, QRS, and T waves are clearly captured and visible under both static and dynamic conditions, with the exception of the ECG recorded at 6 km h −1 , where the P and T waves appear distorted and necessitate further processing.However, the R peak remains distinctly identifiable in this scenario.The Pearson correlation coefficient (r) [53] confirms that there is substantial similarity of more than 80% between the ECG signals acquired by the gel electrodes and the CF-MN electrodes across all activities, excluding walking at 6 km h −1 .
The slight variation in this particular case can be attributed to factors like motion artifacts, electrode movement, and muscle artifacts.Notably, the similarity observed during the sitting activity exceeds 87%, surpassing the findings of a recently published study. [54]These outcomes serve to affirm the effectiveness and efficiency of the newly developed electrodes.Overall, the kurtosis values were generally higher for dry electrodes compared to wet electrodes.The kurtosis tended to decrease with increasing activity intensity, with the lowest values observed at the highest speed of 6 km h −1 .However, the kurtosis values increased after exercise, particularly in the standing and sitting positions.These findings suggest that factors such as electrode type and activity level can influence the shape and distribution characteristics of the recorded ECG signals.
Based on the analysis presented in Figure 9A, the paired t-test results indicated no statistically significant difference (p > 0.05) in the average kurtosis values, except for certain cases.For subject one walking at 4 km h −1 and subject two in sitting and standing positions, higher kurtosis values were observed for the CF-MN electrodes.This could potentially be attributed to initial good contact with the skin and subsequent settling of the electrode against the skin over time.
The kurtosis values of the ECG signals recorded during various activities, except for walking at 6 km h −1 and subject three standing after exercise, were consistently higher than five.This suggests that the ECG signals remained relatively unaffected by artifacts like muscle interference, baseline wander, and powerline interference.However, for all subjects walking at 6 km h −1 and subject three standing after exercise, the kurtosis values were lower than five.This indicates that respiration and muscle activity had an impact on the ECG signals.Overall, the analysis demonstrates that the kurtosis values provide insights into the quality and characteristics of the ECG signals, with deviations observed in specific situations where factors like electrode settling and physiological activities influence the signal properties.The overall mean SNR, estimated for all three subjects, for both wet and dry electrodes varied across different activities.
The study found that motion artifacts had a significant impact on the ECG signal quality during physical activity (Figure 9B).The SNR tended to decrease with increasing activity intensity.As the speed increased, the ECG cycle became shorter, and signal quality tended to decrease due to myoelectric signals and noise from electrode friction, motion artifacts, and breathing, with the lowest values observed at the highest speed of 6 km h −1 .The CF-MN electrodes showed no statistically significant difference (p ≥ 0.05) in SNR compared to gel electrodes, except for subject one when walking at 6 km h −1 and standing after exercise, and subject two in a sitting position after exercise and standing before exercise, potentially due to the flexibility of the CF-MN electrode.However, the mean SNR values estimated for CF-MN were lower for all subjects when walking at 4, or 6 km h −1 , and standing on the treadmill, likely due to electrode displacement.
The mean SD values of the QRS complex for both wet and dry electrodes were also determined for various activities and exhibited slight variations across different activities and types of electrodes.The differences between wet and dry electrodes were minimal, with generally higher SD values observed at higher speeds.However, overall, the SD values remained relatively low, indicating that the shape and distribution characteristics of the recorded ECG signals were close to a normal distribution.These findings suggest that the influence of artifacts and nonstationary noise on the ECG signals was minimal, ensuring the quality and reliability of the recorded data.Overall, the study suggests that CF-MNbased dry electrodes can acquire clinically acceptable ECG signals during a range of activity levels.
Images were also taken to assess the impression left on the skin after removing the electrodes.Minor erythema marks (Figure 10A) were observed after the removal of both commercial and microneedle-based electrodes, consistent with previous findings. [41]These marks were highly transient and typically disappeared within 30 min as shown in Figure 10B.Digital microscopy images of CF-MN electrodes (Figure 10C,D) confirmed the microneedles' integrity without any signs of damage, detachment, or breakages during or after use.
For EMG recording as described in Section 2.4, the electrodes were placed on the forearm of the subjects (Figure 11A, see Figure S9, Supporting Information).Because the muscle activities were relatively intense, the raw data still showed obvious artifacts, and digital filtering methods were carried out to improve signal quality as described earlier.Figure 11B shows a ≈140 s segment recording, after filtering, that depicts EMG signals arising from the various hand and finger movements.Figure 11C illustrates the full rectified signal.The calculated SNR values in Figure 11E were derived from the signal and noise regions.
As presented in Figure 11B-I inset, the CF-MN electrode exhibited an SNR comparable to that of commercial wet electrodes across all activities except for thumb movement.The reason for this is the inability to place both sets of electrodes close enough to simultaneously pick up the activity of the flexor pollicis longus (FPL) muscle of the forearm, which is responsible for flexing the thumb.Nevertheless, the data suggests that the flexible CF-MN electrode can effectively record EMG signals associated with finger movements, followed by computationally inexpensive signal processing.The flexibility of the CF-MN allows it to conform closely to the muscles being monitored, enabling selective EMG signal recording and elimination of crosstalk.This is particularly important when multiple muscles are located in small spaces, making the flexible CF-MN a superior choice for EMG recording.

Conclusion
The CF-MN electrodes proposed in this work are suitable for wearable biopotential recording, and the acquisition of ECG and EMG signals has been demonstrated using customized electronic systems and a small sample of healthy volunteers.Using metrics such as SNR, SD of QRS complex, and kurtosis, it has been shown that these signals were comparable in quality to those obtained using conventional wet electrodes, but without the need for skin preparation or the use of wet gel.The low ESI, high mechanical strength, flexibility, and comfort of the CF-MN also represent significant performance advantages over standard wet electrodes and rigid microneedle electrodes.In addition, the use of a conductive textile as a rear backing layer means that metal deposition is required on just the front face of the electrode, thereby reducing metallization time and cost by 50% over comparable devices.
While no skin irritation or microneedle damage was observed during 48 h of wearing the electrodes during this assessment, longer-term studies will be necessary to fully evaluate their comfort and performance over extended periods.Future studies will include also involve recording electrophysiological signals from a greater number of subjects using CF-MN electrodes in order to gain further statistical proof of performance.It is also anticipated that the performance of CF-MN electrodes can be enhanced by optimizing the assembly process and investigating the effects of electrode coating on the recording capabilities.Due to its high performance and low cost, this electrode shows potential for a wide range of applications, including sports and lifestyle monitoring, medical diagnostics, human-computer interaction, and long-term health assessments.

Figure 1 .
Figure 1.Schematic representation and equivalent electrical model for A) conventional wet electrodes, B) dry electrodes, and C) CF-MN electrodes.E hc is the half-cell potential between electrode and skin; R hc and C hc are the charge transfer resistance and equivalent capacitance; R sc is the series resistance of gel, moisture, and sweat; R sc and C sc are the equivalent resistance and capacitance of epidermis layer; Rtissue is the resistance of dermis and underlying tissues.

Figure 2 .
Figure 2. A) Schematic view of the structure and material of the CF-MN electrode.B) Principle of minimally invasive electrophysiological signal acquisition by a flexible CF-MN electrode.C) (I-VI) Manufacturing process flow of the CF-MN electrode.

Figure 3 .
Figure 3. A) Front side of the metallized wafer; B) rear side of the conductive fabric (nonmetallized); C) image of the detached electrode; D) SEM image of 500 μm tall CF-MN microneedle array.

Figure 4 .
Figure 4. Illustration of the stretchability and bending tests.A) Experimental set-up for stretchability; inset: SEM Image after stretching of the sample; B) stress-strain relationship; C) experimental set-up for bending; D) force versus displacement curve of bending for one of the samples.

Figure 5 .
Figure 5. Representative examples of microneedle compression data.Left: force-displacement data obtained following compression of a CF-MN array using a cylindrical steel probe.Inset: experimental set-up.Right: post-compression SEM micrograph of the microneedles after application of a 35 N load to the array.

Figure 6 .
Figure 6.Skin penetration was assessed using methylene blue dye staining and an ex vivo human skin model: stained skin sample confirms successful microneedle penetration.

Figure 7 .
Figure 7. A) ESI spectra of clinical grade gel and CF-MN electrodes in the frequency range from 20 Hz to 1000 Hz with a step size of 20 Hz (n = 6 subjects, mean ± standard error); B) ESI values at specific spectra ( = 6 subjects, mean±SE); C,D) SEM micrographs of the CF-MN electrode after ESI measurement.

Figure 10 .
Figure 10.Effects of electrode wear on the skin: A) impressions on the chest of a subject immediately after removal of MN and gel electrodes; B) 30 min after removal of electrodes; C) 18 mm CF-MN flexible electrode after use; D) close-up view of four of microneedles, no deformation or breakage of the tip is observed.

Figure 11 .
Figure 11.A) Electrode placed scheme on the forearm, B) Acquired EMG signal from eight gesture classes: hand open, pink finger extension, ring finger extension, middle finger extension, index finger extension, thumb extension, hand closed, C) Full rectified EMG signal, D) smooth EMG signal, each peak illustrates the distinct activities, E) SNR estimated for each subject while performing different activities.
Figure 9A-C shows three quantitative methods for analyzing the artifact level in recorded ECG: kurtosis, SNR, and standard deviation (SD) of QRS amplitude over a 15-s period.Data are presented in Figure 9A-C.All three methods indicate that CF-MN electrodes can acquire high-quality ECG signals.The overall mean kurtosis values, computed for all three subjects, for both wet and dry electrodes, were calculated across different activities.