Noble-Metal-Free Recyclable Electronic Nanoinks for Wireless Wearable Sensors

Abstract

Electronic waste (e-waste) severely pollutes ecosystems in addition to depleting scarce material resources, highlighting the urgency to develop sustainable and recyclable electronics. While various metal nanoparticles have been tested in literature regarding built-in recyclability for electronics, critical knowledge remains unclear on how recycling processes affect their properties, since oxidation and contamination of recycled nanomaterials may compromise the functional and reliable performance of remanufactured devices. This study aims to fill this knowledge gap by systematically investigating the behaviors of metal particles at different remanufacturing stages and by developing an effective, printing-enabled, remanufacturing route using fully recyclable, noble-metal-free, conductive inks. Recyclability of the printed conductors is investigated in terms of electrical properties across multiple reuse cycles, achieving ~90% recovery of electrical conductivity after 3 reuse cycles and ~70% recovery after 6 reuse cycles. As proof of concept, a wireless strain-sensing platform is designed for real-time monitoring of small strains generated by the human body, highlighting potential for wearable human-machine interface applications.

Introduction

Driven by the rapid expansion of the Internet of Things (IoT), it is projected that by the year 2035, there will be at least 1 trillion electronic devices worldwide.¹ Moreover, continued advances in smart and wearable sensing technologies could potentially accelerate device proliferation beyond current projections.^2–5 Once these devices reach their end-of-life, disposal of them will create a substantial volume of e-waste. Moreover, current e-waste recycling techniques often involve resource-intensive operations, requiring high capital investments, harsh chemicals, and processing temperatures above 1,000 °C.⁶ As a result, only a small fraction of the global e-waste is recycled and remanufactured. For example, global e-waste generation reached a record high in 2022 at 62 billion kg, while only 22.3% was formally collected and recycled.⁷ These challenges underscore the need to design and develop sustainable electronic devices that are engineered for recycling and remanufacturing.

To address the limitations of existing recycling techniques, additive manufacturing (AM) has gained attention as a room-temperature processing route that rapidly converts bulk waste materials into printable nanoparticle/microparticle inks that can be processed into upcycled products with customizable form factors, such as wearable sensors,^{8, 9} energy harvesters,^{10, 11} energy storage devices^12–14 and magnetic devices.^{15, 16} In addition to manufacturing innovations, recent advances in substrate engineering use transient, biodegradable materials to enable facile recovery of metals by selectively degrading the polymers.^17–20 However, a key challenge is that the recycled materials often exhibit degraded functional properties (e.g., electrical, optical, and magnetic) after multiple reuse cycles.^{21, 22} Thus, a knowledge gap remains regarding the factors governing this property degradation and how these factors can be deliberately manipulated in recycled materials to enable effective remanufacturing. While some studies have investigated the recyclability and electrical properties of silver-based conductive inks and pastes,^23–25 silver nanoparticles are expensive,^{26, 27} have low abundance in the Earth’s crust (< 0.1 ppm),²⁸ and are toxic to biological organisms.²⁹ Hence, to achieve sustainability and circularity in the manufacturing of printed electronics, it is imperative to explore alternatives such as non-noble metals and, more importantly, to understand the recyclability and functional properties of these alternative materials across multiple recycling phases.

In this study, we have developed an AM-enabled, fully recyclable Ni-based conductive nanoink for wearable printed electronics (Figure 1). The recyclability of the printed conductors was evaluated based on their printability and electrical properties across three reuse cycles. Furthermore, we introduced a cost-effective solvent-dissolution strategy to restore the electrical properties of recycled conductors, enabling stable and consistent device performance during reuse. As proof of concept, we fabricated a recyclable, flexible strain sensor and integrated it with a low-energy Bluetooth module for real-time wireless motion monitoring, highlighting the potential for wearable applications in human-machine interface devices.

Figure 1.

The recycling and remanufacturing process for 3D-printed strain sensors.

Results and discussions

The fabrication of the strain sensor and development of the wireless sensing platform are schematically illustrated in Figure 2A and comprise two steps: ink formulation and 3D printing, followed by integration with a Bluetooth low-energy module for wireless data transmission. To formulate the ink, polyvinylpyrrolidone (PVP) is used as the polymeric binder, deionized water serves as the solvent, and Ni particles and multi-walled carbon nanotubes (MWCNTs) act as the conductors. For enhancing the printability of the inks, a planetary centrifugal mixer is used for homogeneous mixing. The ink is then printed onto various flexible substrates (e.g., paper or polyimide) in user-defined sensor designs. After the printed sensors reach their end-of-life, the printed conductors can be mechanically recovered with a high yield of ~95% (Figure S1, Supporting Information) and reformulated into recycled inks for future processing (Figure 1).

Figure 2.

Ink formulation, printability, and surface properties of the printed sensors across three reuse cycles; A. Schematic of the ink formulation, device fabrication, and wireless strain sensing; B. Optical image showing the pristine and recycled inks; C. Extrusion-based printing of the conductive ink on a paper substrate (scale bar: 1 mm); D. Print accuracy of the inks in terms of width deviation; E. Surface profiles of the 3D printed lines; F. Surface topology of a 500 μm × 500 μm surface of the printed sensor from cycle 1; G. Average surface roughness of the printed lines across three reuse cycles.

We first study the ink behavior and printability of the Ni-based nanoinks. The optical image of the formulated inks from the pristine Ni particles and recycled Ni particles is shown in Figure 2B. All the inks exhibit homogeneous dispersion of the functional particles, with no visible particle sedimentation or phase separation. Furthermore, both the pristine and recycled inks have high dispersion stability during storage. By optimizing the particle volume fraction and adding PVP as a polymeric binder, we engineered the ink formulation to achieve improved colloidal stability (Figure S2). The inks show no visible sedimentation/phase separation of the particles with a shelf life of 3 weeks (56 wt% Ni, 4 wt% MWCNT), indicating the inks remain printable. Figure 2C shows the smooth and continuous extrusion of the conductive ink on a flexible paper substrate. The print accuracy of the pristine and recycled inks is determined by measuring the deviation in the width of printed traces from the set nozzle size (Figure 2D). The low deviation in the width of ~15% across three different reuse cycles indicates that the formulated inks are highly printable with sufficient printing accuracy. Furthermore, the inks exhibit consistent and reproducible surface profiles both before and after recycling (Figure 2E). The average thickness of the printed traces across three reuse cycles is shown in Figure S3 of the Supporting Information. We further investigated the effects of recycling on the surface topology of the printed sensors. Figure 2F shows the surface topology of a 500 μm × 500 μm region of the sensor fabricated from the pristine ink. The surface topology of the recycled sensors is shown in Figure S4 in the Supporting Information. We estimated the arithmetic mean height (S_a) to quantify the surface roughness of the printed sensors (Figure 2G). S_a represents the average of the absolute distances of the surface points from the mean plane and is estimated using Equation 1.³⁰

$$S_a=\frac{1}{MN}{\sum }_{j=1}^N{\displaystyle {\sum }_{i=1}^M\left|Z\right|}\left(x_i,y_j\right)$$ (1)

Here, M is the column number, and N is the number of rows on the surface, and Z (x_i, y_j) is the deviation of the surface irregularities from the base plane. The roughness of the printed lines increased slightly by approximately 9% after one recycling cycle and 14% after two recycling cycles. This increase in roughness can be attributed to an increase in particle-size polydispersity resulting from neck formation and grain growth during sintering of the conductive traces.³¹

We then investigate how different recycling phases affect the electrical and microstructures of the printed inks. Figure 3A shows the scanning electron microscopy (SEM) image of the pristine Ni particles. The Ni particles exhibit a multi-granular, flower-like morphology composed of individual Ni particles. A high electrical conductivity of the sensors was demonstrated from the pristine Ni particles in Figure 3B. After sintering at 300 °C for 1 hour, most of the polymeric binders are decomposed, resulting in an electrical conductivity of 2.7 × 10⁴ S m⁻¹ for the pristine particles. However, after two cycles of recycling and reuse of the Ni particles, a decreasing trend in electrical conductivity is observed. There was a sharp decline in electrical conductivity after two cycles of recycling. A similar declining trend was also observed for pure Ni nanoink without MWCNT (from 1.92 × 10⁴ S m⁻¹ to 54.9 S m⁻¹). The deterioration in electrical performance may be attributed to either oxide formation at elevated sintering temperatures,³² or the presence of residual organics at the particle interface due to the decomposition of polymeric binders at the sintering temperature.³³ To validate our hypothesis, we performed X-ray diffraction (XRD) on the pristine and recycled Ni particles (Figure 3C). From the diffraction pattern, it is evident that the crystalline face-centered cubic (fcc) structure of Ni has remained intact, with characteristic peaks at 44.58°, 51.9°, and 76.54° for both the pristine and recycled Ni particles.³⁴ This confirms the absence of oxide formation and rules out any bulk phase degradation of the particles.

Figure 3.

Morphology and electrical properties of the pristine and recycled sensors without any intermediate washing steps; A. SEM image of the pristine Ni particles (scale bar: 1 μm); B. Electrical conductivity of the printed sensors across three recycling cycles without intermediate washing steps; C. X-ray diffraction (XRD) patterns for Ni particles from cycles 1–3; D. SEM image of the recycled Ni particles showing possible non-conductive layers on the particle surface (scale bar: 1 μm); E. Optical image of the pristine and recycled Ni particles.

Hence, the reduced electrical performance may be attributed to non-conducting layers on the surfaces of Ni particles formed during the decomposition of polymeric binders during sintering, rather than to Ni oxidation under the experimental conditions in this study. The highly heterogeneous structure of these residues enables them to bind strongly to metal particles, encapsulating or blocking the particle surface and creating non-conductive layers.³⁵ These non-conductive layers would increase contact resistance and disrupt electron transport across particle–particle junctions, decreasing the overall electrical conductivity. The SEM image in Figure 3D of the recycled Ni particles shows these non-conductive layers, marked by black spots on the particle surface. This mechanism of carbon residue formation has been well reported for various metal catalysts in the literature.³⁵ The presence of these organic residues can be further confirmed by the visual color of the Ni particles after recycling (Figure 3E). While the pristine Ni particles have a distinct gray color, after multiple recycling cycles, they progressively turn black due to the accumulation of organic residues.

To remove organic residues from the surface of the Ni particles, we used a simple solvent-dissolution method with a low-cost organic solvent, N,N-dimethylformamide (DMF).³⁶ Owing to its strong solvating ability, DMF can effectively dissolve and strip residual organic compounds that are attached to the particle surface.^37–39 After recovering the sensor particles, they were washed 3-5 times with DMF. Before drying at 70 °C for 24 hours, the particles were washed 2 times with deionized water. Figure 4A shows the optical image of the Ni particles before and after the washing steps. After washing, the distinct gray colour of the Ni particles is regenerated, confirming the removal of the organic residues. The spent DMF solvent turned completely black after the washing (Figure 4B). Furthermore, the surface morphology of the Ni particles after solvent treatment shows a substantial reduction in the non-conducting layers (Figure S5). The regeneration of the Ni particles is further confirmed by the restoration of the electrical properties of the printed sensors, resulting in an 90% recovery of their initial electrical conductivity after 3 cycles of reuse and 70% recovery after 6 cycles of reuse (Figure 4C). The observed decline in electrical conductivity after 6 cycles of reuse can be attributed to progressive degradation of the metal nanoparticles during prolonged reuse and repeated sintering cycles. Although Ni exhibits relatively high resistance to oxidation compared to other base metals, the literature indicates that repeated thermal processing can still promote oxidation and microstructural changes that adversely affect electrical transport.^{40, 41} Furthermore, the stability of the electrical properties was investigated by monitoring the resistance of the printed sensors from different reuse cycles over 7 days in ambient conditions (Figure 4D). The negligible increase in the normalized resistance change (ΔR/R₀) confirms that the printed sensors exhibit stable electrical properties with no performance degradation.

Figure 4.

Washing-enabled regeneration of the electrical properties of the recycled sensors; A. Optical image showing the Ni particles as recovered (left) and after regeneration (right); B. Optical image showing the black spent DMF solvent; C. Electrical conductivity of the printed sensors across six reuse cycles with intermediate washing steps; D. Electrical stability of the printed sensors across six reuse cycles over 7 days; E. Optical image of the flexible strain sensor printed on a polyimide substrate and encapsulated with PDMS (scale bar: 5 mm); F. SEM image showing the Ni particles embedded within a percolating MWCNT network (scale bar: 1 μm).

To enhance the flexibility of the strain sensors, we added a small amount of MWCNTs to the ink formulation. Due to the high aspect ratio (~10⁶), elastic moduli (~1TPa), and tensile strength (~100GPa) of the MWCNT,⁴² MWCNT improves the mechanical robustness of printed sensors, minimizing microcrack formation under applied strain. Figure 4E shows the optical image of the strain sensor printed on a flexible polyimide substrate and encapsulated with polydimethylsiloxane (PDMS) to provide an additional protective layer. The SEM image in Figure 4F shows the dense percolation network of Ni particles and MWCNTs in the printed sensor.

The overall design of the wireless transmission system is shown in Figure 5. The resistance change signal from the Ni-MWCNT sensor in response to applied strain undergoes two stages of signal conditioning. In stage 1, the strain-induced resistance change signals are fed into a Wheatstone bridge to generate a differential voltage signal. In stage 2, this differential voltage signal is sent to an HX711 instrumentation amplifier/ADC module (Hemobllo, 24-bit ADC), which both amplifies the signal and converts it to a digital output. The digitized signal is then sent to a microcontroller unit (MCU) (HiLetgo, 2.4 GHz Dual Mode Bluetooth ESP-WROOM-32), which can locally store the processed data on an SD card and simultaneously transmit it wirelessly via a Bluetooth Low Energy (BLE) Generic Attribute Profile (GATT) service. The BLE GATT service can continuously stream the strain measurements. On the user interface, a Python client running on a computer and/or a smartphone connects as the BLE central and records the received data in real time. This wireless architecture enables continuous, low-latency strain monitoring with a single 5 V power source driving the entire system.

Figure 5.

Schematic diagram of the design of the overall wireless transmission system.

We integrated the design principles of the wireless transmission system shown in Figure 5 with the fabricated strain sensor and the required electronic components to develop the wearable monitoring system, as illustrated in Figure 6A. The sensor’s flexibility enables conformal attachment to the human body (e.g., wrist, finger, knee). Before investigating wireless strain-sensing applications, we verified the sensitivity and mechanical stability of the strain sensor. The sensitivity of the strain sensor was estimated by measuring the gauge factor (GF), a direct measure of its sensitivity to applied strain. The GF can be estimated from Equation 2, where ΔR is the change in resistance, R₀ is the initial resistance, and ɛ is the applied strain.

Figure 6.

Proof-of-concept demonstration of the wireless sensing system based on a fully recyclable Ni-MWCNT strain sensor; A. Optical image of the setup for the wireless strain sensing system; B. Change in relative resistance of the strain sensors under applied strain; C. Change in relative resistance of the strain sensor during 500 cycles of stretching and releasing at 0.8% strain; Response signals of the Ni-MWCNT strain sensor during D. wrist bending at different angles; E. making a fist.

$$GF=\frac{\raisebox{1ex}{$\text{Δ}R$}\!\left/ \!\raisebox{-1ex}{$R_0$}\right.}{\epsilon }$$ (2)

From Figure 6B, the GF of the Ni-MWCNT sensor was characterized by monitoring the percent change in resistance under varying strain loads (0-1%). The relatively high GF of ~12 was determined by linear curve fitting, compared with conventional metal foil strain gauges (BF350-3AA, DAOKAI) with a GF of ~2. This higher sensitivity can be attributed to the higher stiffness of the Ni-MWCNT matrix compared to those of conventional sensor materials such as constantan.⁴³ More importantly, the Ni-MWCNT strain sensor exhibited negligible changes in sensitivity after recycling, with a GF of ~11, indicating the highly recyclable nature of these printed conductors without any downcycling/degradation of their functional properties. Although pristine Ni and recycled Ni sensors demonstrated a higher GF of ~20 and ~18, respectively (due to the higher stiffness of Ni⁴⁴), they were not mechanically robust, resulting in microcrack formation during device testing. Figure S6(A–B) shows the surface morphology of the Ni sensor before and after cyclic stretching and releasing at 0.8% strain. Due to the Ni sensor’s lower mechanical strength, the device failed after only 50 cycles. The surface morphology of the Ni sensor shows visible microcrack formation, disrupting the percolation network (Figure S6B). In contrast, the Ni-MWCNT sensor exhibited remarkable repeatability and mechanical stability even after 500 cycles of stretching and releasing at 0.8% strain, as shown in Figure 6C. The resistance of the Ni-MWCNT sensor exhibited <5% change after 500 cycles at the unstretched state, whereas we did observe that after 150 cycles, the stretching-induced resistance change increased to >12% at the stretched state and then reached a stable plateau at 200-500 cycles. In addition, the sensor’s surface morphology shows no visible micro-crack formation during the cyclic strain loading (Figure S7), further confirming the robustness and flexibility provided by MWCNT. Furthermore, the 3D-printed sensor generated consistent, reproducible electrical signals over 500 consecutive cycles (inset of Figure 6C). The stable electrical response of the Ni–MWCNT strain sensor indicates that the conductive matrix is mechanically robust, which helps suppress microcrack formation and prevents irreversible damage to the sensor.

Furthermore, to evaluate the performance of the Ni-MWCNT sensor under real-world conditions, we investigated its chemical and thermal stability (Figure S8). Both the pristine and recycled MWCNT sensors exhibited moderate moisture stability at 70% relative humidity (RH). Although the relative resistance increased during the first 9 hours, it stabilized thereafter, and the overall change in resistance was <15% (Figure S8A). This increase can be attributed to the particle surface interacting with moisture and gradually reducing film conductivity.^{45, 46} In addition, the Ni-MWCNT sensors exhibited excellent thermal stability over 20 cycles of heating and cooling between 50 °C and 25 °C (covering the human body temperature range) (Figure S8B).

To investigate the sensing application of the Ni-MWCNT strain sensor, the sensor was integrated into a wireless system design, as illustrated in Figure 5, for real-time monitoring of human motion. To characterize the sensing performance of the wireless platform, we measured the sensor’s response time under 0.8% strain (Figure S9). The sensor’s response time was estimated at approximately 960 ms. The sensor was attached to a volunteer’s wrist using 3M double-sided tape. Once the volunteer bent his wrist at varying angles, the effective conductive path would be stretched, generating an electrical signal in the form of a resistance change (ΔR/R₀). The MCU module can then collect, process, and convert the electrical signal generated by the sensor in real-time and transmit the data wirelessly via the low-power BLE transmitter module, which is connected to a smartphone/PC with a Bluetooth receiver. As shown in Figure 6D, the response signal amplitude increased as the volunteer increased the angle of wrist bending. A larger bending angle produces greater strain and stretching in the sensor, resulting in a higher signal intensity. Moreover, the sensor’s resistance returned to its initial values as the wrist returned to the unbent state. Similarly, the sensor could detect when different gestures (e.g., making a fist) are performed, as shown in Figure 6E. The sensor could generate highly stable, replicable electrical signals (with negligible change in signal amplitude) when the volunteer repeated the same actions. Furthermore, to evaluate the effect of multiple reuse cycles on the sensing performance, we integrated a recycled Ni-MWCNT sensor with the wireless system. The recycled sensor maintained its sensing performance and produced repeatable signals when the volunteer made the same gesture (e.g., making a fist), with no performance degradation (Figure S10A). The recycled sensor also maintained its mechanical robustness, as the recycled Ni-MWCNT sensor exhibited <5% change in resistance at the unstretched state after 500 cycles of stretching and releasing at 0.8% strain with reproducible signal outputs (Figure S10B).

While this study demonstrates the feasibility of recycling and reusing printed conductors in a circular, sustainable manner for wearable and wireless electronic sensors, the potential health concerns of these materials warrant further assessment. Future research is needed to develop recycling strategies for non-toxic, biocompatible, and environmentally friendly metals, such as Ti or Mg, to evaluate the broader use of recyclable materials for human-machine interface-compatible devices.

Conclusion

This study presents fully recyclable nanoinks for the direct printing of wearable sensors. The formulated ink shows high colloidal stability, and the nanoparticles can be easily recovered and reused across multiple recycling cycles without requiring incineration processes of e-waste. In particular, we observed that the direct recycling of metal nanoparticles can result in a sharp decline in electrical conductivity due to the buildup of organic residues, which can be addressed by solvent washing. A simple wash-and-reprint workflow mitigates performance degradation and downcycling of the conductive particles, allowing the sensors to remain recyclable while maintaining high electrical conductivity (~1.6 × 10⁴ S m⁻¹ after six cycles of reusing) and stable sensing performance (a gauge factor of ~11). Furthermore, by integrating a BLE microcontroller module, a wireless sensing prototype is developed for in-situ, real-time wearable applications that enable human motion sensing. This prototype is capable of detecting minor strains across various bending angles and gestures, highlighting its potential in human-machine interface applications.

Experimental section

Materials

Polyvinylpyrrolidone (PVP, average molecular weight, 360 kDa) and N, N-dimethylformamide (DMF) were purchased from Sigma-Aldrich Corp. Nickel powder (99.8%) was obtained from Thermo Scientific and used as received. Multi-walled carbon nanotube (MWCNT) powder was purchased from Alfa Chemistry and used as received. Deionized water was used as the solvent for all the ink formulations.

Ink formulation

PVP was used as the polymeric binder, and deionized water was used as the solvent for the ink formulation. 2.63 g of PVP was dissolved in 50 g of deionized water to make a 5 wt% PVP solution. To formulate the ink, 4.5 g of Ni particles and 0.3 g of MWCNT were combined with 3.35 g of the 5 wt% binder solution. A planetary centrifugal mixer (Thinky AR-100) was used for homogeneous mixing. The mixer operated at 1200 rpm for 15 minutes to homogenize the dispersion. The formulated ink was then transferred to a 2 mL ink cartridge for printing. For pure Ni ink, no MWCNTs were added, with all other steps remaining the same.

Printing

All printing processes in this study were conducted using a direct ink writing (DIW) printer. The print settings were 0.27 mm pass spacing, 0.2 mm z height, 500 mm min⁻¹ printing speed, 50 mm trim length, 1 mm trace penetration, 0.1 mm anti-stringing distance, 0.13 mm piston kick, 0.01 soft start ratio, 1 soft stop ratio, and 0.16 rheological setpoint. A flexible polyimide substrate (Kapton) was used for printing. Before printing, the substrate was cleaned with isopropyl alcohol and de-ionized water. To improve the ink’s adhesion to the substrate, the substrate was plasma-cleaned for 15 minutes using a plasma cleaner (Harrick Plasma, PDC-001-HP) to introduce hydrophilic surfaces.

Sintering

After the printing, the printed patterns were thermally sintered in a vacuum tube furnace (MTI Corporation, OTF-1200X) equipped with dual-zone heating. Sintering was performed at 300 °C with a temperature ramp of 4 °C min⁻¹. Sintering was carried out for 1 hour under continuous purging with forming gas (95% N₂, 5% H₂) at a flow rate of 30 standard cubic centimeters per minute to maintain a reducing environment and prevent oxidation.

Recycling

For recycling of metal sensors, the conductive trace was removed by mechanical abrasion, followed by dispersing in deionized water for probe sonication. The dispersion was probe-sonicated (Qsonica Q500) at 50% amplitude with 2-second pulses for 5 minutes to break up the large aggregates that formed during sintering-induced grain growth. After breaking the aggregates, they were rinsed with de-ionized water and then washed 3-5 times with DMF to remove the organic residues from the particle surface. The particle-to-solvent mass ratio was maintained at 1:5 during the washing steps. To accelerate the removal, the dispersion was probe-sonicated at 20% amplitude with 2-s pulses for 1 minute. After that, the particles were rinsed twice with deionized water before being dried at 70 °C in a vacuum oven for 24 hours. Once the particles are dried, they can be mixed with the required amount of binder solution, as reported in the ink formulation, and reformulated into recycled inks for subsequent reuse. For the control samples, the particles were washed three times with water only, without any DMF treatment, and then dried in a vacuum oven at 70 °C for 24 hours. The recycling yield is calculated through Equation 3. Here, m_D is the mass of the deposited particles on the substrate during printing, and m_R is the mass of recovered particles from each sensor.

$$Recyclingyield(\%)=\frac{\left({\mathit{m}}_{\mathit{D}}\mathbf{-}{\mathit{m}}_{\mathit{R}}\right)\times 100}{{\mathit{m}}_{\mathit{D}}}$$ (3)

Measurement and Other Characterization

Microscopic images were acquired using an Olympus BX53M System Microscope (bright-field mode) and a Hitachi S-4700 field emission scanning electron microscope (accelerating voltage 2 kV, beam current 10 mA, working distance 3 mm). A Rigaku MiniFlex 6G X-ray Diffractometer was used to acquire XRD results (Cu Kα radiation, 40 kV, 20 mA). The thickness of the printed patterns on the polyimide substrates was measured using a Filmetrics Profilm3D optical profilometer. The dimensions of the patterns are given in Table S1 in the Supporting Information. A four-probe system (Ossila, four-point probe) was used to measure the conductivity of the printed pattern, minimizing contact resistance. Temperature, oxidation conditions, and humidity were maintained at the same levels throughout all the experiments, unless otherwise specified. To characterize the sensors’ humidity stability, a controlled-humidity chamber was used, with the relative humidity maintained at 70%. For the thermal stability analysis, the sensors were subjected to heating (50 °C) and cooling (25 °C) cycles. The resistance of the sensors was measured after thermal equilibrium was reached during both heating and cooling.

Statistical Analysis

Unless stated otherwise, all data are presented as the mean ± standard deviation (SD) from at least three replicate samples. OriginLab and ImageJ were used for data analysis.

Testing the device on a human arm

The experiments involving the strain sensor on a human arm, as described in Figure 6, do not require Institutional Review Board (IRB) approval, as the experiments did not involve any medical procedures, personal data collection, or interventions that posed risk or discomfort. The experiments were performed with the volunteer’s consent.

Data Availability

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