Fabrication of High-Resolution, Flexible, Laser-Induced Graphene Sensors via Stencil Masking

Abstract

The advent of wearable sensing platforms capable of continuously monitoring physiological parameters indicative of health status have resulted in a paradigm shift for clinical medicine. The accessibility and adaptability of such portable, unobtrusive devices enables proactive, personalized care based on real-time physiological insights. While wearable sensing platforms exhibit powerful capabilities for continuously monitoring physiological parameters, device fabrication often requires specialized facilities and technical expertise, restricting deployment opportunities and innovation potential. The recent emergence of rapid prototyping approaches to sensor fabrication, such as laser-induced graphene (LIG), provides a pathway for circumventing these barriers through low-cost, scalable fabrication. However, inherent limitations in laser processing restrict the spatial resolution of LIG-based flexible electronic devices to the minimum laser spot size. For a CO₂ laser–a commonly reported laser for device production–this corresponds to a feature size of ~120 μm. Here, we demonstrate a facile, low-cost stencil-masking technique to reduce the minimum resolvable feature size of a LIG-based device from 120 ± 20 μm to 45 ± 3 μm when fabricated by CO₂ laser. Characterization of device performance reveals this stencil-masked LIG (s-LIG) method yields a concomitant improvement in electrical properties, which we hypothesize is the result of changes in macrostructure of the patterned LIG. We showcase the performance of this fabrication method via production of common sensors including temperature and multi-electrode electrochemical sensors. We fabricate fine-line microarray electrodes not typically achievable via native CO₂ laser processing, demonstrating the potential of the expanded design capabilities. Comparing microarray sensors made with and without the stencil to traditional macro LIG electrodes reveals the s-LIG sensors have significantly reduced capacitance for similar electroactive surface areas. Beyond improving sensor performance, the increased resolution enabled by this metal stencil technique expands capabilities for scalable fabrication of high-performance wearable sensors in low-resource settings without reliance on traditional fabrication pathways.

INTRODUCTION

Graphene, a monolayer of hexagonally-packed carbon atoms, demonstrates unique electrical, mechanical, and thermal properties that have resulted in significant academic and industrial interest for a diverse range of applications.(Tiwari et al., 2020) Traditional synthesis methods (e.g., mechanical (Novoselov et al., 2004) / liquid-phase (Li et al., 2019) exfoliation, chemical vapor deposition (F. Liu et al., 2022)) often involve complex and/or resource-intensive procedures that can restrict broader utilization of graphene. By comparison, laser-induced graphene (LIG) represents a paradigm shift for graphene synthesis via a simplified, one-step laser-based process to synthesize and pattern a porous, three-dimensional (3D) graphene structure that preserves much of the unique material properties of graphene. This low-cost, rapid, and scalable production method has resulted in utilization of LIG in areas traditionally associated with graphene (e.g., energy storage (Li et al., 2018; Ma et al., 2020; Peng et al., 2015), catalysis (Cheng et al., 2022; Zhang et al., 2018), water purification(Gupta et al., 2021; Singh et al., 2017; Tittle et al., 2018)) and rapid adoption in emerging areas such as wearable sensing (Carvalho et al., 2018; Chen et al., 2021; Clark and Ray, 2023; Lee et al., 2022; Ling et al., 2020; J. Liu et al., 2022; Seo et al., 2020; Vivaldi et al., 2021; Wang et al., 2021; Xuan et al., 2018; Yang et al., 2020).

The landscape of laser-induced graphene (LIG) has evolved dramatically since the initial report in 2014 (Lin et al., 2014) marked by sophisticated advancements in preparation methods and a deepened comprehension of the complex structure-property-processing relationships that dictate material performance (Vivaldi et al., 2021). This evolution is exemplified by the expansion from initial demonstrations on polyimide and poly(ether imide) to a broad spectrum of carbonaceous materials including lignin, cellulose, and various polymers that can be converted into LIG via laser ablation (Chyan et al., 2018). This expansive library of substrate materials is largely attributed to an enhanced understanding of how various laser parameters impact LIG formation and influence the resulting chemical and mechanical properties (Duy et al., 2018). Concurrently, additional methods for tuning the properties of LIG have been discovered through strategic modifications to the substrate, either pre- or post-laser processing (Clerici et al., 2016). Methods such as substrate surface pre-treatment (Yoon et al., 2020) or functionalization with metal nanoparticles (Hui et al., 2019) or carbon nanotubes (Settu et al., 2021) can enhance the sensing performance of LIG. Additionally, modulating the environmental conditions surrounding the substrate, such as employing an inert atmosphere, enables tuning the wettability of LIG (e.g., from hydrophilic to hydrophobic) (Li et al., 2017). Laser-induced graphene, a field characterized by its dynamic and rapid pace of innovation, continues to evolve with novel techniques. For example, recent studies report the use of an air assist cone to create a localized atmosphere during laser scribing to achieve LIG with wettability properties similar to that obtained in a controlled atmospheric chamber (Dallinger et al., 2023). A similar report describes altering the pulsing density and output power during laser rastering under ambient conditions to pattern LIG with distinct superhydrophobic, parahydrophobic, and superhydrophilic regions (Nasser et al., 2020).

Though LIG processing methodologies have advanced substantially, challenges remain, particularly in attaining high-resolution features and optimizing material performance. For example, the resolution limit of commercial CO₂ lasers (de la Roche et al., 2023; Lin et al., 2014; Mikheev et al., 2022), which represents a common laser class used for LIG patterning, constrains the design space to features with resolutions typically greater than ~100 μm. Similarly, efforts to tune laser parameters (e.g., laser power) to enhance material properties (e.g., conductivity) often inversely affect other parameters (e.g., thickness(Duy et al., 2018)). Addressing such limitations is essential for continuing to realize the promise of LIG as a sensing material, particularly in wearable sensing applications where high spatial resolution is critical for device miniaturization and optimization of on-body performance. Current efforts center on utilizing laser-based processes that utilize shorter wavelength lasers (e.g., green (Hayashi et al., 2021), blue (Stanford et al., 2020), or UV (Carvalho et al., 2018)) or faster pulsing (e.g., femtosecond (Hong et al., 2022)) to produce a narrower line width with reported obtainable feature sizes ranging from 6.5 μm to 50 μm. Alternative approaches utilize hybrid processes that rely upon conventional cleanroom-based photolithography to generate high-resolution patterns in a photoresist (e.g., SU-8), which are then converted to LIG via laser processing (Beckham et al., 2021). Although these innovations result in improvements in achievable resolutions of patterned LIG features (Joanni et al., 2022; Raza et al., 2022; Yong et al., 2016), the reliance upon controlled environments, specialized equipment, and relatively expensive materials inhibits broad deployment. Other methods employ mask-based patterning of CVD synthesized graphene to achieve higher resolution, albeit necessitating advanced synthesis processing capabilities and costly equipment for plasma etching (Yong et al., 2016) or UV ozonation (Wu et al., 2017). This underscores the necessity for innovative, low-cost approaches that offer enhanced feature resolution without introducing additional burdens that would further limit the broad accessibility enabled by laser-based processing.

This paper introduces a rapid, low-cost approach to LIG patterning that enables the fabrication of feature sizes below the native resolution of a CO₂ laser via the use of metal stencils. This approach bypasses the need for customized laser configurations and specialized equipment by utilizing a standardized, commercially available component (metal stencil) found in the production of conventional flexible and rigid printed circuit boards (PCBs). This process can repeatably pattern LIG with a minimum achievable feature size of 45 ± 3 μm (vs. 120 ± 20 μm without stencil-masking). This stencil-masked LIG (s-LIG) patterning approach produces higher-quality patterned features by reducing or eliminating the laser raster induced edge effects. As a result of the enhanced spatial resolution, this reported method expands the design space for LIG-based components (e.g., sensors). We demonstrate the practicality and effectiveness of this technique via fabrication of common sensors utilized in LIG-based wearable platforms including temperature and dual electrode chemical sensors. We further highlight the potential of the expanded design space by demonstrating fabrication of a low-capacitance microarray sensor not readily produced via current LIG-based patterning methods. Collectively, these results demonstrate the promise this approach holds for enhancing the rapid and facile fabrication of high-quality, high-fidelity LIG-based devices via conventional laser-based processes.

METHODS

LIG Synthesis and Patterning.

A laser engraving process using an unmodified, commercial CO₂ laser cutter (Epilog Mini 24 30W, Golden, CO) with a 10.6 μm wavelength laser under ambient conditions (air atmosphere, ~55 % relative humidity, 72 °C) formed each sensor. Rinsing both surfaces of polyimide (PI) sheets (0.003” thickness Dupont Kapton, American Durafilm, Holliston, MA) with isopropyl alcohol (IPA, Certified ACS Plus, Fisher Chemical) followed by deionized (DI) water (18.2 MΩ-cm ultrapure water via Milli-Q Direct-Q 3 system, MilliporeSigma, USA) removed surface contaminants in preparation for laser engraving. Tape secured each PI sheet to the reusable metal stencil (stainless steel, 0.006” (~152 μm) thickness, Metal Etch Services Inc., San Marcos, CA) prior to laser engraving. Neodymium magnets (various sizes, K&J Magnetics, Pipersville, PA) placed near smaller sensor features maintained firm contact between the PI/metal stencil assembly and the laser bed during processing. Rastering the CO₂ laser in a rectangular pattern over the open areas of the metal stencil transformed the exposed PI surface into graphene via the following pattern specific settings: (A) microfeatures: two passes using autofocus; 28% power, 20% speed, 1200 DPI; (B) connection pads: one pass using autofocus; 23% power, 25% speed, 1200 DPI. Removal of the PI sheet after laser processing completed the sensor fabrication process. Cleaning the metal stencil via sonication in DI water for 30 min prepared the stencil for subsequent sensor production runs. The stencils show no loss of efficacy after over 6 months of use unless the features are physically bent or damaged from misuse.

Material Characterization of LIG

A digital microscope (Keyence VHX-7000) enabled optical characterization of LIG sensors and dimensional analysis via linewidth measurements of distinct LIG microarray bands and stencil band openings (n=60). Secondary electron imaging on a FEI Helios 660 dual-beam focused ion beam SEM (FIB-SEM) instrument (5.0 kV accelerating voltage) enabled microstructural characterization of the LIG sensors via scanning electron microscopy (SEM) image analysis. A micro-Raman RXN system from Kaiser Optical Systems (785 nm NIR laser, 50 μm slit) supported Raman analysis of the LIG graphene (50x magnification, 6.7 mW power, 60 s).

Electrical Characterization of LIG

A digital multimeter (Keithley DMM 7510 ½) and a four-point probe (Signatone S-302, SP4) enabled collection of sheet resistance measurement data of the thin bands of the microarray electrodes fabricated with conventional LIG and s-LIG patterning. Values reported comprise average resistance measurements resulting from maintaining contact between the probe and sample for 30 s. Averaging addressed variations resulting from the laser raster process that resulted in distinct LIG microarrays (n=3) and distinct bands (n=3) on the same LIG microarray fabricated with one stencil pattern.

Temperature Measurements

Recording current using a digital multimeter (Keithley DMM 7510 ½) under an applied constant voltage (1 V) from a source meter (Keithley 2450 SourceMeter) enabled collection of temperature data from LIG temperature sensors. Measurement of sensor resistance at specific temperatures (30, 35, 40, 45, and 50 °C) in a sequential manner (5 min dwell time at each temperature) enabled creation of a calibration curve specific to each sensor. The combination of known supply voltage and measured current allows for precise calculation of resistance and, therefore, temperature via this pre-established calibration curve. A FLIR One Pro (Teledyne FLIR, Wilsonville, OR) and infrared thermometer (Ryobi, Fuchu, Japan) provided external validation of measured temperatures used to establish sensor calibrations.

Electrochemical Measurements

A portable potentiostat (PalmSens4, PalmSens BV, Houten, Netherlands) enabled recording of all electrochemical measurements via a three-electrode measurement system with LIG serving as the working electrode, counter electrode, and pseudo-reference electrode. All solutions were prepared or diluted in 0.1 M KCl. Constant potentials of +0.6 V (vs. carbon pseudo-reference electrode at first working electrode) and −0.6 V (vs. carbon pseudo-reference electrode at second working electrode) were utilized for all dual electrode amperometric measurements of varying concentrations of Fe(CN)₆^3-/4- (0, 0.1, 0.25, 0.5, 0.75, 1, 2, and 5 mM). A syringe pump (Harvard Apparatus Pump 11 Elite; 0.5 mL/min flow rate) connected to a straight channel, adhesive-based microfluidic device (225 μm channel height x 4 mm width, fabricated according to previous reports (Clark et al., 2022)) facilitated these sequential flow-based measurements (recorded in triplicate per dual electrode; n=3 electrodes). Chronoamperometry measurement potential was held at +0.6 V for 60 s vs. carbon pseudo-reference electrode. Custom adhesive reservoirs (3M 467; area ~24.5 mm²) confined solutions to the working, counter, and reference electrodes for cyclic voltammetry (CV) and chronoamperometry measurements (in 0.1 M KCl) of both the microarray and macro electrodes fabricated with and without stencils. Calculation of capacitance values utilized established methods (Klunder et al., 2019). For the stability study, conventional LIG and s-LIG microarray electrodes were fabricated on different designated days and stored in ambient conditions (~21 °C, 62% relative humidity, in a covered petri dish) and measured using CV on the same day. Additional CV measurements in 1 mM Fe(CN)₆^3-/4- enabled characterization of sensor performance and a scan rate study. Demonstration of microarray electrode performance (with stencil) utilized CV measurements of varying concentrations of Fe(CN)₆^3-/4- (25, 50, 100, 250, 500, and 1000 μM). Each measurement was replicated (n=3) using independent electrodes (i.e. unique sensor). The evaluation of scan rate data enabled electroactive surface area calculations via the Randles-Sevcik equation (Muzyka and Xu, 2022):

$$i_p=0.4463nFAC{(\frac{nFvD}{RT})}^{\frac{1}{2}}$$

where $i_p$ is the current maximum (A), $n$ is the number of electrons transferred in the redox event, $A$ is electroactive surface area (cm²), $F$ is Faraday constant (C mol⁻¹), $D$ is diffusion coefficient (cm²/s), $C$ is concentration (mol/ cm³), $v$ is scan rate (V/s), $R$ is ideal gas constant (J K⁻¹ mol^𢈒1), and $T$ is temperature (K).

RESULTS AND DISCUSSION

High Resolution Patterning via Stencil-Masked Laser Induced Graphene

The simple, low-cost stencil-masked LIG (s-LIG) technique presented here (Figure 1) enables high-resolution patterning (< 100 μm) of LIG beyond the native resolution of a commercial CO₂ laser cutter while simultaneously enhancing both the quality and uniformity of the patterned LIG. This patterning approach leverages a reusable metal stencil to precisely mask the laser beam during the laser raster process. Directly fixing the substrate (typically a polyimide (PI) sheet) to the stencil surface via adhesive eliminates registration errors during processing. Magnetically securing the substrate/stencil assembly to the laser bed eliminates substrate delamination during processing, such as during the patterning of fine features. As with subtractive machining processes, this fixturing is essential for ensuring the highest resolution of patterned features both for a given design and in the scalable manufacture of sensors via processing of the full laser bed (i.e. 24 in width, 12 in depth). In contrast to conventional cleanroom processes, the conclusion of laser processing represents the complete manufacturing process for LIG patterning with devices ready for either direct deployment (e.g., strain, temperature sensors) or subsequent modification steps (e.g., biosensing).

Figure 1.

LIG sensor fabrication process via (A) conventional LIG processing and (B) s-LIG patterning. Sub-panels highlight optical images (80x magnification) of a LIG-based (i) temperature sensor, (ii) dual electrode, and (iii) microarray electrode, respectively. Scale bars are 200 μm. (C) Photograph of s-LIG patterned sheet of multiple microarray electrodes. Scale bar is 1 cm.

Figure 1 presents representative sensor examples to highlight the observable improvements in feature resolution enabled by s-LIG patterning. Visual comparison of a temperature, multi-electrode, and a microarray electrode sensor patterned via conventional laser engraving (Figure 1A) and our s-LIG process (Figure 1B) reveal differences in resolution and pattern quality. In the absence of a stencil, the patterned features across all sensor designs exhibit dimensional variations (e.g., variable width), defects, or poorly-defined edges which can limit the minimal resolvable feature for a given design (as with the temperature and multi-electrode sensors). In contrast, s-LIG patterned sensors demonstrate a higher resolution for resolvable features as well as an increased dimensional fidelity and enhancement in overall LIG quality. For the laser system in this work, the data indicate the minimal achievable resolution improved from 120 ± 20 μm (n = 60) to 45 ± 3 μm (n = 60) indicating the stencil resolution rather than laser beam diameter governs the resolution of a patterned feature. Concomitantly, utilization of a stencil reduces variations within a patterned feature (i.e. uniform widths of patterned lines) to those inherent in the stencil (e.g., average width of stencil openings defining tines of microarray is 52 ± 4 μm, n = 60). Drawing from common practice in similar fields (e.g., additive manufacturing (Schmutzler et al., 2016)), utilization of a unique scaling factor for a given stencil design can mitigate the deviation between the nominal and theoretical feature dimensions. Additionally, the s-LIG method translates well to larger scale manufacturing through the ability to pattern multiple sensors concurrently (Figure 1C).

The observed improvements resulting from the s-LIG process also extend to the electrical properties of the LIG. Characterization of sheet resistance for conventional LIG and s-LIG indicates the stencil-masked method yields a lower sheet resistance than the conventional processing approach (60 ± 10 Ω sq⁻¹ vs. 1700 ± 400 Ω sq⁻¹ respectively, n = 3). The s-LIG features a high degree of uniformity within a given device (50 ± 10 Ω sq⁻¹, n = 3 devices). Table 1 contextualizes the reported improvements of the s-LIG process reported here with conventional and alternative patterning processes reported in existing literature. Although some methods achieved higher resolution (Beckham et al., 2021; Hong et al., 2022; Stanford et al., 2020), the reported s-LIG method presents a low-cost alternative. This method, utilizing a conventional CO₂ laser, distinguishes itself with exceptional resolution while eliminating the need for expensive equipment and infrastructure, departing from traditional approaches that mandate clean rooms (Beckham et al., 2021) and high-resolution lasers (Carvalho et al., 2018; Hayashi et al., 2021; Hong et al., 2022; Li et al., 2022; Stanford et al., 2020).

Table 1.

LIG Fabrication Techniques and Resulting Properties

Method	Resolution (μm)	Sheet Resistance (Ω sq−1)	Ref.
Conventional CO2 LIG Processing
10.6 μm laser on PI	~100	35	(Lin et al., 2014)
10.6 μm laser on PI	120 ± 20	1700 ± 400	This work
Mask-Based Processes
10.6 μm laser on cleanroom-produced photoresist	10 – 20	~120	(Beckham et al., 2021)
Patterned synthesized CVD graphene from vertical magnetic-field-assisted UV ozonation through sapphire, nickel, and steel stencils	29	-	(Wu et al., 2017)
Metal stencil-masked fabrication on PI (s-LIG)	45 ± 3	60 ± 10	This work
Patterned synthesized CVD graphene using oxygen plasma etching with laser-cut mask, lamination transfer, and chemical etching	55	-	(Yong et al., 2016)
Patterning the graphene ink via laser-cut stencils	~200		(Hu et al., 2023)
UV Laser
Frequency tripled 355 nm laser on PI sheet sandwiched between two microscope slides	~25	Resistivity 68 μΩ m−1	(Joanni et al., 2022)
1 μs pulsed irradiation from 355 nm diodepumped laser	50	Resistivity 24 Ω cm−1	(Carvalho et al., 2018)
Visible Laser
SEM-mounted 405 nm fiber-coupled laser at varying pulse widths	~13 (~20-25)	~1000 (108)	(Stanford et al., 2020)
522 nm laser pulses on PDMS at two distinct focuses using fs laser system and translation stage	~ 100	Conductivity 51.6 S/m	(Hayashi et al., 2021)
fiber-coupled 405 nm laser irradiation on PI with Au metal precursor	150	-	(Li et al., 2022)
800 nm fs laser with temporal shaping, plasma ejection confinement, and silver nitrate doping using 6-axis stage	6.5	0.0004	(Hong et al., 2022)

Material Characterization of s-LIG

The cross-sectional image of conventional LIG results in a surface comprising loose graphitic flakes transitioning to a partially-converted region of the PI substrate (Figure 2A). Scanning electron microscopy (SEM) images of conventional LIG (Figure 2B, Figure SI1A) maintain the characteristic porous and fibrous microstructure associated with LIG (Lin et al., 2014; Vivaldi et al., 2021; Ye et al., 2018). Given the known conductivity of fibrous LIG (Duy et al., 2018), the pronounced charging phenomena observed at the tips of fibrous features in both samples (i.e. lighter regions) indicates reduced conductivity, which may result from incomplete electrical pathways between fibers and the grounded sample holder. In contrast, the s-LIG patterning yields a compact structure embedded in the PI substrate with minimal surface protrusion observed in the cross-sectional image (Figure 2C). SEM images of s-LIG (Figure 2D, Figure SI1B) exhibit a distinctive morphology with a discernable valley observed within the porous matrix suggestive of a unique topological characteristic attributable to fabrication process. The valley ridges transition from the conventional morphology to a bottom surface of planar graphite with reduced charging. The increased conductivity of the planar graphite is attributed to its hexagonal lattice structure, in contrast to the less ordered structure in the fibrous regions (McCreery, 2008; Singh et al., 2023; Wilcox et al., 2007). The valley morphology forms across various changes in laser parameters (i.e., power, speed, DPI; Figure SI2), including non-optimal settings with only partial LIG formation.

Figure 2.

(A) Cross sectional optical microscopy images and (B) SEM images of conventional LIG microarray. (C) Cross sectional optical microscopy images and (D) SEM images of s-LIG microarray. Optical microscopy images were collected 1000x magnification and SEM at 3500x magnification.

These findings align with decreased sheet resistance of the LIG fabricated with a s-LIG patterning. Fibrous topology has been shown to be a result of increased degassing rate from increased irradiation time (Singh et al., 2023). In conventional LIG formation, the temperature increases beyond the beam diameter of laser due to the thermal conductivity of PI which makes graphene conversion wider than irradiation zone (Singh et al., 2023). The metal stencil in the s-LIG process mitigates this but still results in the fibrous regions on the edges, hypothesized to result from heating and cooling effects from the edge of the metal stencil. However, Raman spectroscopy confirms the composition and chemical structure of the LIG produced by both processes remains consistent as these data show the characteristic peaks at ~1310 cm⁻¹ and ~1600 cm¹ (Figure SI3) (Lin et al., 2014; Ye et al., 2018; Yong et al., 2016). As a consequence, the observed improvements in resolution and material performance from the s-LIG process result primarily from these topological differences.

Performance Benchmarking of s-LIG Device Fabrication

A flexible resistance-based temperature sensor (or resistance temperature detector, RTD)(Kulyk et al., 2022; Kun et al., 2021; Nag et al., 2022; Zhang et al., 2023) serves as a representative example of a LIG-based sensor commonly deployed for wearable, on-body measurements. Typically fabricated from metallic thin-films via cleanroom-based processes (Krishnan et al., 2018), RTDs transduce temperature variations by measuring changes in electrical resistance. In the context of wearable sensing, RTDs with high-density, serpentine geometries (to improve sensitivity while reducing self-heating effects and improving flexibility/stretchability(Lacy, 2011)) support localized temperature measurements for sports applications (Soomro et al., 2024), diagnostics (Mirjalali et al., 2022), and environmental responses (Wu et al., 2023) as well as calibrating other on-body sensors (e.g., electrochemical sensors) (Gao et al., 2016). With the emergence of LIG, recent reports highlight the growing utilization of rapidly produced, low-cost LIG-based RTDs (Gao et al., 2016; Le et al., 2022; Nag et al., 2022; Yang et al., 2023; Zhu et al., 2021); however, the planar resolution of such sensor demonstrations is constrained by the beam diameter of a conventional CO₂ laser. By contrast, s-LIG enables a reduction in the width of the serpentine traces of a LIG-based RTD sensor. The resulting increase in resistive path length can improve sensor sensitivity within the same form-factor. For the demonstration sensor reported here, current-voltage (I-V) measurements enable calculation of device resistance via Ohm’s law (14 ± 2 kΩ; n = 10 devices; Figure SI4). Sequential temperature measurements (continuous recording, 30–50 °C temperature range, 5 °C increment; Figure SI5) enable establishment of calibration curves (Figure 3A). Minimal variation in the slopes of the calibration curves from different s-LIG sensors (Figure SI6) indicate consistent sensitivity across devices with the temperature coefficient of resistance (TCR) (Nag et al., 2022) calculated to be −0.027 ± 0.002% °C⁻¹. The TCR value is comparable to other LIG and graphene-based temperature sensors reported in literature (Kun et al., 2021; Yang et al., 2023). Sensor performance demonstrated close agreement (Figure 3B) with measurements from a commercial handheld IR thermometer and FLIR thermal camera (Figure SI7).

Figure 3.

Calibration curve for s-LIG temperature sensors. Symbols and error bars represent average and standard deviation of measurements from distinct temperature sensors (n=3). Resistance calculations from average of current (I) values continuously recorded over a 60 s interval (+1 V source voltage). (B) Plot of measured temperature vs. reported hot plate temperature highlights representative s-LIG sensor performance as compared to a FLIR thermal camera and a handheld IR thermometer.

A dual electrode sensor serves as a representative example of a common electrochemical sensor deployed in wearable sensing applications. Here, the sensor design features two equivalent working electrodes centrally positioned to a shared carbon pseudo-reference electrode and counter electrode. Such sensors, when equipped with two independently controlled working electrodes, support multi-analyte detection (Jadon et al., 2016; Pakchin et al., 2017), differential measurements and blank subtraction (Nguyen et al., 2023), and signal amplification (Lai et al., 2009; Mathew et al., 2020). This sensor geometry can represent a demanding design for fabrication via the conventional laser engraving process. Although miniaturization of the sensor can increase sampling accuracy (i.e. smaller probe area), the inherent resolution limit of the LIG processing restricts the minimum intra-electrode spacing. Reduction beyond this limit will merge individual electrodes resulting in either a short circuit or introduce measurement errors. In contrast, s-LIG enables rigid separation of electrical components beyond the resolution minimum, which in turn preserves sensor function.

Validation of the s-LIG sensor performance utilizes a well-established reversible redox probe (Fe(CN)₆^3-/4-) as the test analyte. Simultaneous application of constant positive and negative potentials to the sensor working electrodes enables concurrent measurement of the Fe(CN)₆^3-/4- probe oxidation and reduction currents. A straight channel, adhesive-based microfluidic device (Figure SI8) facilitates precise, sequential delivery of increasing Fe(CN)₆^3-/4- concentrations to the sensor. Triplicate measurements of eight concentrations over a 33 ± 3 min period allow for comprehensive assessment of s-LIG sensor performance. As shown in Figure 4A, the sensor exhibits stable, repeatable performance with the current returning to initial baseline levels across the extended measurement period, indicating no discernable signal degradation. The observation of a modest increase in cathodic current responses at higher concentrations (1-5 mM) suggests potential electrode surface activation or the gradual formation of reduced graphene oxide (rGO) on the working electrode held at a negative potential, consistent with reported rGO production methods (Compton and Nguyen, 2010). Both working electrodes demonstrate rapid, synchronous responses to changes in analyte concentrations (Figure 4B, singular replicate, 0–1 mM Fe(CN)₆^3-/4-); however, the anodic and cathodic current responses differ in magnitude despite the initial solution containing equal concentrations of both Fe(CN)₆^3-/4- redox states. Minimal discrepancies in CV responses (Figure SI9) suggest that factors beyond inherent variations in individual working electrodes contribute to this observed phenomenon, such as the presence of surface defects resulting in different electrocatalytic behavior for the redox states or the detection of molecules by the second electrode initially oxidized by the first working electrode, leading to a concentration increase of that specific redox state. Both working electrodes yielded linear responses (Figure 4C) across replicate devices (n=3, Figure SI10). The minor variability (20 ± 6% relative error) in the cathodic calibration curve likely stems from the aforementioned factors.

Figure 4.

(A) Representative amperograms simultaneously measured with two working electrode of the s-LIG dual electrode across triplicate measurements of 0, 0.1, 0.25, 0.5, 0.75, 1, 2, and 5 mM Fe(CN)₆^3-/4- in 0.1 M KCl. (B) Zoomed in view of amperometric plot for concentrations from 0-1 mM Fe(CN)₆^3-/4-. (C) Corresponding calibration curve of Fe(CN)₆^3-/4- from amperometric measurements. Working electrode 1 (WE1) was held at a constant potential of +0.6 V and WE2 was held at a constant potential of −0.6 V vs. carbon pseudo-reference electrode. Symbols and error bars represent average and standard deviation of triplicate measurements averaged over a 10 s interval from a single device (n=3).

Expanded Design Space for LIG Sensors: Microarray Electrodes

Microarray electrodes have attracted significant interest in electrochemical analysis due to well-established advantages including reduced capacitive-charging currents, mitigated IR drop, and improved mass transport efficiency resulting in steady-state diffusion currents (Matsue, 1993). This superior performance to macro-scale electrodes stems from the strategic arrangement of microelectrodes within the array, leveraging the benefits of both high current densities typically associated with microelectrodes and the sufficiently high currents observed in larger electrodes (Huang et al., 2009). Here, s-LIG enables the fabrication of electrochemical sensors with an exemplar design that incorporates a pseudo-reference and counter electrode alongside a microarray working electrode comprised of ten thin bands (45 μm x 2.5 mm). Measurement of CVs for a range of Fe(CN)₆^3-/4- concentrations (25 μM – 1 mM, Figure 5A) establishes the performance metrics of these microelectrode arrays for analytical applications. The derived calibration curve (Figure 5B) showcases a linear response complemented by a minimal replicate error.

Figure 5.

(A) Representative CVs at varying concentrations of Fe(CN)₆^3-/4- in 0.1M KCl using s-LIG microarray electrodes. (B) Corresponding calibration curve of Fe(CN)₆^3-/4- from CV current values. Symbols and error bars represent average and standard deviation of measurements from distinct microarray electrodes (n=3). All potentials were scanned at 100 mV/s vs. carbon pseudo-reference electrode.

Fabrication of microarray electrodes and macro-scale electrode counterparts, designed with equivalent geometric areas (Figure SI11), via s-LIG and conventional LIG patterning processes enable quantitative assessment of s-LIG microarray analytical performance. CVs collected in an electrolyte solution (Figure 6A) exhibit a significant reduction in capacitive current for microarrays compared to the macro electrodes independent of fabrication method (conventional LIG vs s-LIG). These data (Figure SI12) show an order of magnitude reduction in capacitance for s-LIG microarrays (7 ± 3 μF cm⁻²) relative to conventional LIG counterparts (60 ± 20 μF cm⁻², Figure 6B). The enhanced spatial resolution associated with s-LIG patterning enables a geometric design that effectively disperses electroactive regions (i.e. large, current-dense areas) resulting in a more efficient charge distribution. The resulting reduction in capacitance is advantageous for sensing applications requiring lower limits of detection and enhanced sensitivity as this yields a reduced background current.

Figure 6.

(A) Representative CVs of 0.1M KCl at a scan rate of 100 mV/s using macro and microarray electrodes fabricated with conventional LIG and s-LIG patterning. (B) Scaled CV plots of 0.1M KCl using conventional LIG and s-LIG microarray electrodes. Representative CVs of 1 mM Fe(CN)₆^3-/4- in 0.1M KCl using conventional LIG and s-LIG (C) macro and (D) microarray electrodes at a scan rate of 25 mV/s. Representative CVs at varying scan rates (25, 50, 100, 200, 400, 500, 700, 1000 mV/s) of 1 mM Fe(CN)₆^3-/4- in 0.1M KCl using (E) conventional LIG and (F) s-LIG microarray electrodes.

The microarray electrode capacitance values compare favorably with other reports of pure graphene materials (range 2-60 μF cm⁻²) (McCreery, 2008). The reported values of electrodes formed from other carbon-based materials such as carbon composite electrodes (Antiochia et al., 2004; Klunder et al., 2019; McCord et al., 2021) often approach the values orders of magnitude higher which aligns with the values of the LIG macro electrodes. We note that as a large decrease in capacitance appears in the macro electrode–s-LIG yields 280 ± 50 μF cm⁻² compared to conventional LIG 600 ± 100 μF cm⁻²– the improvement in capacitance cannot solely be attributed to the enhanced spatial resolution. Other factors including surface roughness, presence of incomplete electrical pathways within the electrode, and pseudo-capacitance effects from other redox reactions are known to contribute to capacitance (McCreery, 2008). The morphology characterized in the SEM and cross-sectional images further validates the cause of the increased capacitance in the conventional LIG. The flaky and fibrous topography, spanning the entire surface of the conventional LIG electrode, reveals greater surface roughness compared to the smoother, planar regions observed with the s-LIG. Moreover, the conventional LIG processing approach yields a thick LIG layer above the substrate surface, which likely contributes to a greater quantity of incomplete electrical pathways within the electrode. This combination of factors often results in LIG-based devices optimizing performance to target high capacitance values suitable for enabling use in microcapacitors (1-46 mF cm⁻²) rather than sensing (Lin et al., 2014; “Preparation of high-performance flexible microsupercapacitors based on papermaking and laser-induced graphene techniques,” 2022). As a consequence, the enhanced performance of the s-LIG fabrication process suggests the potential for enabling the tuning of capacitance to suit a broad spectrum of potential applications.

CVs collected with the macro electrodes (Figure 6C) and microarrays (Figure 6D) exhibit comparable peak currents in the presence of the redox molecule Fe(CN)₆^3-/4-, despite substantial differences in capacitive backgrounds. Baseline-subtracted peak currents for the conventional LIG and s-LIG macro electrodes were 1.3 ± 0.2 μA and 1.0 ± 0.1 μA, respectively, and 1.0 ± 0.3 μA and 1.6 ± 0.2 μA for the corresponding microarray electrodes. The shapes of the CVs, however, differ appreciably between the two cases. The macro electrodes display typical profiles indicative of Nernstian behavior (Elgrishi et al., 2018), while the microarray electrodes exhibit sigmoidal shapes associated with steady-state currents, suggesting non-diffusion-limited processes (Matsue, 1993). Such steady-state behavior, which is commonly observed in microarrays, provides stable measurements and predictable electrochemical responses, with advantages that include enhanced reproducibility, simplified data analysis, and improved signal-to-noise ratios. CV studies at different scan rates provide additional insights into this observed electrochemical behavior (Figure SI13, Figure 6E, F). The s-LIG microarrays yield notably higher currents (130% ± 50%) than the conventional LIG microarrays across all scan rates and demonstrate a return to Nernstian behavior at higher scan rates (400-1000 mV/s) as characterized by substantially sharper peaks compared to the counterpart conventional LIG microarray. Similar trends are observed in the macro electrode dyad. Here, the s-LIG electrodes exhibit sharper and larger magnitude current peaks (Figure SI13). These findings suggest that the s-LIG patterning enhances the electrochemical performance of both macro and microarray electrodes, particularly at higher scan rates, potentially due to improved electron transfer kinetics and more efficient mass transport. Additionally, s-LIG and conventional LIG demonstrate stable signal response without degradation when stored in ambient conditions (Figure SI14) for 14 days. Conventional LIG exhibited statistically significant signal degradation (p = 0.029 by two tail t-test, Table SI1) after 17 days.

Further analysis of this scan rate data using the Randles-Sevcik equation (Muzyka and Xu, 2022) reveals that the electroactive surface areas (ECSAs) of the electrodes are generally higher than their geometric areas (Figure SI15, Figure SI16). The s-LIG microarrays yield geometric areas (0.0090 ± 0.0006 cm²) comparable to macro electrodes from conventional LIG (0.0105 ± 0.0001 cm²) and s-LIG (0.0096 ± 0.0001 cm²). The conventional LIG microarray exhibits a larger geometric area (0.025 ± 0.004 cm²) due to unconfined LIG formation and laser resolution limitations. The ECSAs of the s-LIG microarray (0.021 ± 0.001 cm²) and the conventionally fabricated macro electrode (0.0265 ± 0.0004 cm²) are more than twice the geometric areas as a consequence of the three-dimensional, porous nature of LIG, which increases overall electrode surface area to enable electrochemical reactions to occur in pores and crevasses. Interestingly, the microarray fabricated with conventional LIG has the highest geometric area (0.025 ± 0.004 cm²) but the lowest ECSA (0.00786 ± 0.00009 cm²), which may be attributed to the presence of a greater number of incomplete electrical pathways. By comparison, the s-LIG macro electrode demonstrates the highest ECSA (0.035 ± 0.001 cm²) among all electrodes, suggesting the metal stencil impacts ECSA beyond effects attributed solely to size (Wirojsaengthong et al., 2024). This may result from a greater number of complete electrical connections as a consequence of both the compact morphology and presence of additional lateral pathways (compared to the microarray electrodes). The electrochemical performance using chronoamperometry (Figure SI17) of conventional LIG macro electrodes demonstrated the highest sensitivity (59 ± 3 μA cm⁻² mM⁻¹; n = 3 electrodes), exceeding that of the s-LIG macro electrodes (37 ± 5 μA cm⁻² mM⁻¹) despite having higher ECSA, which could result from increased time for diffusion to the electrode surface in chronoamperometry (Baronas, 2017). In contrast, the performance of the s-LIG microarray electrodes revealed significant improvement in sensitivity compared to conventional LIG microarrays from 31 ± 2 μA cm⁻² mM⁻¹ to 43 ± 1 μA cm⁻² mM⁻¹ (p = 0.005 by two tail t-test, Table SI2), consistent with other improvements observed using the s-LIG method.

CONCLUSIONS

Here we report a facile and cost-effective process for enhancing the resolution of laser-induced graphene patterning by employing commercial metal stencils to overcome the intrinsic resolution limits of conventional CO₂ laser-based processing. This stencil-masking technique not only increases the resolution of the LIG patterning process but also improves the electrical performance and uniformity of the resultant devices. Performance benchmarking highlights the utility of the s-LIG patterning for representative classes of LIG-based sensors, including temperature and dual-electrode electrochemical sensors. A notable advancement afforded by this technique is the ability to reliably fabricate intricate patterns with increased resolution, thereby unlocking new possibilities in device design–exemplified by the creation of microarray electrodes that demonstrate enhanced electrochemical capacitance. The findings of this study establish the stencil-masked fabrication strategy as a scalable and accessible pathway to the production of high-resolution, high-fidelity LIG-based devices in low-resource settings, with potential applications spanning from advanced electronics to wearable sensors.

Highlights:

• Introduced stencil-masked patterning process for Laser-Induced Graphene (LIG) sensors.

• Reduced LIG sensor feature size from 120 μm to 45 μm using CO2 laser.

• Process yields LIG sensors with enhanced electrochemical performance.

• Expanded design possibilities for wearable LIG-based sensors.

• Process offers cost-effective, scalable technique for democratizing sensor innovation.