Rapid, One-Step Synthesis of Laser-Scribed Graphene Sensor for Self-Activated Chemiresistive NO₂ Detection

Abstract

Recently, laser-scribed graphene (LSG) has gained significant attention in sensor applications due to its cost-effectiveness, simplicity, environmental friendliness, excellent conductivity, and high thermal stability. Compared with many other graphene fabrication methods, LSG stands out as a promising approach for developing advanced sensors. In this study, we present a rapid, one-step synthesis for fabricating a highly sensitive and selective LSG sensor for self-activated chemiresistive detection of nitrogen dioxide (NO₂). Sensors were characterized using field emission scanning electron microscopy (FE-SEM), high-resolution transmission electron microscopy (HR-TEM), Raman spectroscopy, X-ray diffraction (XRD), X-ray photoemission spectroscopy, Hall measurement, and a thermographic infrared (IR) camera. The LSG-based chemiresisitve sensor exhibited high electrical conductivity with a sheet resistance of around 23.4 ± 0.8 Ω/sq at room temperature and highly sensitive, selective, and fully reversible responses to NO₂ without external heating. Three-dimensional porous graphene plays an essential role in long-term reliability (42 days) in its self-activated state. Also, the sensor demonstrated a detection limit of parts per billion (ppb) levels of 2.68 ppb with high reliability. This rapid, one-step synthesis method provides a promising approach for the scalable production of high-performance graphene-based sensors for gas detection.

1. Introduction

With rapid economic development, concerns about environmental pollution have grown significantly, particularly regarding the impact of air quality on public health. − Among various pollutants, colorless and odorless gases pose a challenge as they often remain undetectable without specialized equipment. Long-term exposure to these gases has been linked to severe health risks, necessitating stringent monitoring and regulation. For instance, excessive inhalation of carbon monoxide (CO), a colorless and odorless gas, can cause irreversible brain damage and even death without immediate awareness. Similarly, toxic gases such as ammonia (NH₃), CO, and nitrogen dioxide (NO₂) are lethal at parts per million (ppm) concentrations. Beyond acute toxicity, prolonged exposure to low concentrations of NO₂ can lead to reduced lung function and damage to the spleen and liver. These findings underscore the critical need for effective detection systems capable of identifying harmful gases that are imperceptible to the human senses. As urbanization continues to accelerate, the development of advanced detection technologies becomes essential for safeguarding public health. Effective air quality management systems are critical for identifying and addressing these hidden threats in our environment.

Chemiresistive gas sensors operate by detecting changes in electrical resistance caused by the interaction between gas molecules and the semiconductor sensing layer, facilitated by electron exchange with metal electrodes. These sensors rely on various sensitive materials, including metal oxides, transition metal dichalcogenides (TMDCs), and carbon-based materials such as graphene and graphene oxide (GO). , The sensing performance is highly dependent on factors such as the type of sensing material, target gas, operating temperature, pressure, and relative humidity (RH). In particular, metal oxides such as SnO₂, WO₃, and ZnO have been widely used and commercialized as sensing materials due to their ability to facilitate charge transfer during gas adsorption and desorption. However, the charge transfer within these metal oxides is highly sensitive to temperature variations, making it challenging to maintain stable performance in diverse thermal environments. In addition, most commercial chemiresistive sensors have faced challenges such as low sensitivity and slow response times in low and high humidity ranges. These limitations not only hinder their performance but also increase the complexity and cost of developing supporting circuits. To avoid these bottlenecks, it is essential to develop sensing materials with improved sensor characteristics and simplified interfaces. Graphene has been considered a promising candidate for sensing applications for decades. , One of the most appealing features of graphene-based chemiresistive sensors is their high sensitivity at room temperature (RT) combined with ultralow power consumption (<μW). , However, sensors fabricated using chemical vapor deposition (CVD) and micro-electro-mechanical systems (MEMS) manufacturing processes face significant challenges, including nonuniformity and low stability in their responses. This variability in sensor-to-sensor performance severely limits the practical application of these gas sensors in real environments. Furthermore, such fabrication methods generally increase the complexity and cost of scaling up the sensor production process, posing additional barriers to commercialization. , In contrast, laser-scribed graphene (LSG) fabrication methods involve direct laser writing on graphene oxide (GO) or polymer substrates, offering the significant advantage of enabling rapid and cost-effective production of graphene-based sensors. − Recent advances demonstrate that LSG sensors benefit from the high surface area and porous three-dimensional structure generated by laser scribing, which improve gas diffusion and adsorption, leading to enhanced sensitivity and selectivity for analytes such as NO₂. The use of diverse carbon sources and composite materials permits precise control over morphology and electronic characteristics, enabling the development of flexible, stretchable sensors suitable for real-time environmental monitoring. , Additionally, integration with other nanomaterials or application of surface modifications can further improve the sensitivity and selectivity. Notably, the recently studied porous three-dimensional (3D) LSG is fast, economical, and provides numerous active surface sites for gas–solid interactions, making it highly suitable for gas sensing applications. Flexible and stretchable gas sensors based on LSG and their nanocomposites include in situ laser-assisted synthesis and precise patterning that create highly conductive, porous 3D structures suited for wearable applications. MXene/LSG composites fabricated via single-step laser irradiation show enhanced conductivity, flexibility, strain sensitivity, and cyclic stability, making them promising for use in next-generation environmental and health sensors. Standalone stretchable LSG platforms leverage LSG’s mechanical strength and electrical properties for multifunctional sensing, with scalable fabrication offering significant future opportunities in flexible electronics and wearable technology. In this study, we present a rapid, one-step synthesis for fabricating a highly sensitive and selective LSG sensor for the self-activated chemiresistive detection of NO₂. The LSG-based chemiresistive sensor is synthesized via a direct CO₂ laser scribing process on a polyimide (PI) film, resulting in the formation of interconnected porous graphene structures. The sensors were characterized using field emission scanning electron microscopy (FE-SEM), high-resolution transmission electron microscopy (HR-TEM), Raman spectroscopy, X-ray diffraction (XRD), X-ray photoemission spectroscopy (XPS), Hall effect measurement, and thermographic infrared (IR) camera. The repeatability, response time, and long-term stability of the LSG-based chemiresistive sensor for NO₂ were investigated. Importantly, the self-activated chemiresistive behavior of the sensor eliminates the need for additional functionalization or external stimuli, thereby simplifying the detection process and improving the robustness of the sensor.

2. Experimental Section

2.1. Fabrication of Laser-Scribed Graphene on Polyimide Film

The laser-scribed graphene (LSG) was prepared using 127 μm-thick polyimide Kapton films on a slide glass supplied by DuPont (Wilmington, DE), which was first cleaned with ethanol and deionized water. The films were scribed under ambient conditions using a Q-switched pulsed CO₂ laser beam operating at a wavelength of 10.6 μm with a maximum output power of 100 W. The laser beam is continuously and directly delivered along the predefined paths, which were previously designed as lines in a CAD file. We investigated the influence of focusing parameters of the quality of LSG. The results revealed that positioning the polyimide film at a defocused distance of 10 mm below the laser focal plane produced optimal performance characteristics. During the experiment, the scanning speed of the equipment was set to 100 mm/s, with a line step width of 0.01 mm and a repetition frequency of 90 kHz. The laser power was systematically varied between 12 and 36 W to investigate its influence on the self-heating properties of LSG. This automated method enables cost-effective fabrication of glass slide-sized chemiresistive sensors (1.5 cm × 2.5 cm) at approximately 50 cents per sensor, eliminating the need for additional steps such as chemical treatments for cleaning or surface modification. Herein, we only included material costs (substrates), manufacturing costs (laser machining), labor, and overhead costs related to fabrication, and did not consider component packaging. The entire process, including design, fabrication, and cleaning, is completed in under 10 min per sensor.

2.2. Characterization

The surface morphology of LSG on polyimide substrates was characterized using a Merlin FE-SEM (Carl Zeiss Co. Ltd.) and HR-TEM operated at 200 kV. The crystallinity and phase properties of LSG samples were evaluated using glancing angle XRD (GAXRD; Rigaku D/Max-2500) with a fixed incident angle of 2°. Raman spectroscopy was performed using a Horiba Jobin Yvon Labram HR UV–visible–NIR spectrometer equipped with a 514 nm Ar ion laser (Horiba Scientific). Hall effect measurements are employed to determine the sheet resistance (R _S) of graphene by analyzing the Hall coefficient (R _H), enabling the calculation of the carrier density and mobility (μ). The Hall coefficient is derived from the measured Hall voltage (V _H), applied current (I), and magnetic field (B) using the relationship: μ = R _H/R _S. The R _S, measured through van der Pauw techniques, is then utilized to calculate the μ. The chemical states of constituent elements in graphene were characterized by using XPS with a Thermo Fisher ESCALAB 250i system. All binding energy values were calibrated by referencing the C 1s peak of adventitious carbon at 284.8 eV. The temperature profiles and thermal images of LSG sensing lengths (1–9 mm) under applied direct current (DC) voltages were obtained by using a FLIR T62101 infrared thermal imaging camera (FLIR T420, 320 × 240 IR resolution), which provides high-resolution thermal mapping and precise temperature measurement capabilities. The thermal distribution of the LSG sensor was performed using COMSOL Multiphysics Software (version 5.4) to calculate the heat transfer at the junction between the electrodes and sensing layer.

2.3. Sensing Characteristic Evaluation

The samples were characterized under controlled relative humidity (14% RH) using a Keithley 2420 SourceMeter and a cylindrical chamber with a volume of 63.37 cm³. Sensing measurement system has a fast gas flow rate of 1000 sccm passing in 4 s (16.67 cm³/s). In this study, the sensor response (R) is defined as R (%) = [(R _air/R _gas) – 1] × 100 (oxidizing gases) and R (%) = [(R _gas/R _air) – 1] × 100 (reducing gases). R _air and R _gas denote resistance in the air and target gas environments, respectively. A fixed total gas flow rate of 1000 cm³/min was maintained throughout the system. For self-heating-based gas sensing evaluations, a Keithley 2420 SourceMeter was used to apply DC voltages (1–7 V) across graphene electrodes of LSG. For gas sensing measurement of LSG-based sensor, the quartz cylinder for gas flow was maintained a constant RT of 24 °C in a tubular furnace without external heating. To evaluate the long-term stability of the sensor, R ₀ (initial resistance) and response for 10 ppm of NO₂ were conducted over a 35-day period with DC voltage (7 V) applied every 3 days. In this study, for sample-to-sample variations (Table S2 and Figure S3), we included error bars representing standard deviations or confidence intervals based on resistance data (3 and 10 measurements) for sensors manufactured so far, when exposed to NO₂ gas at different concentrations and in air. However, the batch-to-batch variation was not considered in this study because it is not in the form of a sensor array.

3. Experimental Results and Discussion

3.1. Fabrication of LSG-Based Chemiresistive Sensor

Compared to conventional graphene synthesis techniques, LSG offers a simplified, cost-effective, and environmentally benign fabrication approach while exhibiting excellent electrical conductivity and high thermal stability. These advantages establish the LSG as a versatile material for advanced sensor applications, combining fabrication efficiency with robust functional performance. The LSG-based chemiresistive sensors were fabricated by direct CO₂ laser writing of polyimide, as illustrated in Figure a. Graphene was successfully patterned on polyimide/slide glass with high uniformity and density using a femtosecond laser system integrated with a computer-controlled program. Figure b shows the photograph image of sensor configuration in which the electrodes and sensing layers are monolithically integrated, demonstrating a unified structural design that eliminates interfacial resistance between components for efficient self-heating behavior. Systematic power variation trials during laser scribing of polyimide (PI) substrates revealed two critical thresholds: (i) Threshold power for graphene formation: 10 W (minimum energy required for polyimide to graphene conversion) and (ii) maximum permissible power: 45 W (beyond which laser-scribed graphene (LSG) exhibits delamination due to excessive thermal stress). This power-dependent behavior highlights a nonlinear relationship between laser energy input and LSG structural integrity, which is critical for optimizing fabrication parameters in sensor applications.

1.

(a) Schematic of the fabrication of laser-scribed graphene (LSG) with flexibility. (b) Photograph image of LSG-based chemiresistive sensor on polyimide film with the electrodes and sensing layers. (c) Top-view FE-SEM image of LSGs fabricated with various laser powers. Inset images by FE-SEM show the thickness of LSG according to powers (12–36 W).

3.2. Morphology and Structure of LSG

The surface morphology of LSG fabricated under varying laser powers (12–36 W) was characterized using FE-SEM. As seen in Figure c, with increasing laser power, the laser-scribed lines demonstrate improved edge definition and expansion in line width. Moreover, higher laser power drives the formation of graphene with hierarchical porosity, where the graphene flakes are converted into porous graphene with micro and nanoscaled holes, reflecting controlled photothermal processing dynamics. These porous micro/nanoscaled holes underscore the correlation between laser parameters and graphene morphology, critical for optimizing functional properties in applications for gas sensing. The formation of porous graphene, which is power-dependent, occurs only within approximately 60 μm of the 127 μm-thick PI Kapton film, demonstrating a depth-limited conversion process during laser-scribed graphitization. The nanostructure and crystallinity of the LSG flakes were characterized by HR-TEM and XRD experiments. LSG flakes exhibiting few-layer structural features; the edge analysis in Figure a provides confirmation of these layered features. Moreover, the surface of LSG exhibits ripple-like wrinkled structures, which have been demonstrated to enhance the electrochemical performance of devices through increased active surface area and improved charge transport kinetics. In addition, the nanoscale ripples correspond to exposed edges of graphene layers. These structural features likely originate from localized thermal expansion during pulsed laser irradiation, a process driven by rapid photothermal energy conversion. The highlighted red box in Figure a is an enlarged region intended to illustrate the lattice spacing. The average lattice spacing of 0.34 nm, as seen in Figure a (inset), corresponds to the interplanar distance of the (002) crystallographic planes in graphene. This value aligns precisely with XRD data. In addition, consistent with previously reported studies, , the HR-TEM image reveals LSG with the unusual polycrystalline structure, characterized by disordered grain boundaries that reflect nonequilibrium growth kinetics during rapid laser processing. This nonequilibrium growth kinetics can account for the curvature of the graphene layers, leading to the porous structure. Figure b shows out-of-plane θ–2θ (theta) XRD patterns of LSG on polyimide films according to various laser powers. The image inserted in Figure b shows a representative 36 W sample (1 cm × 1 cm) prepared for XRD analysis. There is one characteristic peak in the XRD pattern of the graphene (around 2θ = 27°), which was assigned to the (002) plane. The other two characteristic peaks in the XRD pattern are polyimide related with (200) and (006) planes. The graphene exhibited diffraction peaks nearly identical to those of polyimide, indicating retention of the structure of carbon atoms and molecules after laser-driven graphitization. Furthermore, increasing laser power enhanced both the sharpness and intensity of diffraction peaks, reflecting improved crystallographic alignment. sensor applications. Raman spectra of the LSG according to laser power are shown in Figure c. The peaks of LSG at 1342 and 1583 cm^–1 corresponded to sp³ (D band) and sp² (G band), respectively. The G band is universally observed in carbon-based materials, corresponding to the in-plane vibrational modes of ordered sp²-hybridized carbon networks. In contrast, the D band serves as a structural property, with its intensity directly correlating to the density of defects and disordered domains within the carbon lattice. The intensity ratio (I _D/I _G) provides a quantitative metric for assessing crystallographic disorder, where higher ratios indicate increased structural imperfections, such as vacancies, grain boundaries, or defects. The I _D/I _G of LSG was found to be 1.42 at 12 W, 1.14 at 20 W, 1.29 at 28 W, and 1.56 at 36 W, respectively. As laser power increases (except for 12 W), the G peak becomes more distinct and sharper, and the intensity ratio (I _D/I _G) increases, reflecting enhanced defect density within the graphene lattice. This structural imperfection optimizes surface reactivity and charge transfer kinetics, which are critical for enhancing sensor sensitivity and detection efficiency. In addition, the Raman spectrum of LSG shows the presence of a 2D peak observed at 2688 cm^–1. The G and 2D bands correspond to the in-plane vibrational mode of sp²-hybridized carbon networks and the stacking order of graphene along its c-axis, respectively. Because the I _2D/I _G obtained from the Raman spectrum range from 0.69 to 1.82 depending on the laser power, the synthesized LSG is implied to have a bilayer and multilayer structure with relatively low defect density. In previous studies, the peak shift in the Raman spectra of graphene is related to the change in the electronic and structural properties, and the 2D peak for carbon-based materials indicates the presence of graphene. The ratio I _2D/I _G is used to compare the number of graphene layers and is inversely proportional to the number of graphene layers. As the laser power increases, the I _2D/I _G is increased, which shows a few graphene layers. The chemical composition of the LSG was investigated using XPS. As shown in Figure d, the LSG exhibited no detectable environmental impurities by photothermal and photochemical reactions during laser processing. High-resolution C 1s XPS spectra of LSG with 36 W of laser power revealed four dominant peaks: sp² C–C bond at 284.6 eV, sp³ C–C bond at 285.4 eV, C–OH bond at 286.3 eV, and C–O bond at 287.5 eV. The sp² C–C peak became dominant in LSG, accounting for 80.6% as a major functional group. This result suggests that PI film-based LSG has a higher degree of carbon enrichment and a lower degree of oxygen suppression upon laser irradiation, which would be attributed to the highly functionalized structure of graphene. Therefore, LSG was dominated by sp² carbon, which agreed well with the Raman and XRD results.

2.

(a) HR-TEM image of LSG treated at 36 W, (b) XRD pattern of LSG according to the laser powers, (c) Raman spectrum of LSG according to the laser powers, (d) high-resolution C 1s XPS spectra of LSG treated at 36 W, (e) sheet resistance, and (f) electrical conductivity of LSG on polyimide (PI) substrates fabricated with various laser powers.

In order to investigate the electrical conductivity (σ) of LSG, samples were consistently fabricated for each laser power. As shown in Figure e, the sheet resistances of LSG corresponding to increasing laser powers were measured as 24.5, 23.9, 23.4, and 22.4 Ω/sq, respectively. A gradual decrease in the sheet resistance was observed as the laser power increased. The phenomenon described can indeed be attributed to laser power-driven graphitization, which enhances charge carrier mobility by increasing the degree of graphitization and the size of sp²-hybridized carbon domains. During laser irradiation, high localized temperatures cause the transformation of carbon materials into more ordered graphene-like structures with larger graphene flakes and fewer defects. Electrical conductivity values were determined using the formula: σ = 1/(R _S × t). The thickness (t) of the LSG was measured from the cross section (60 μm) of FE-SEM. As the laser power increased, the overall electrical conductivity of the LSG is observed to increase significantly. This increased graphitization improves the structural properties and decreases electron scattering at grain boundaries, effectively reducing resistance and increasing electrical conductivity. As graphitization progresses, charge carriers (electrons or holes) move more freely with less impediment, leading to an enhanced conductivity. In addition, it was confirmed that the carrier concentration also gradually increased as the laser power increased. Higher laser power tends to increase the defect density and promotes graphitization, which can modify the carrier concentration by introducing more doping sites or defects that provide additional free carriers. The previous studies have shown that laser treatment can significantly influence carrier density in graphene, with reported carrier concentrations reaching on the order of 10¹² cm^–2. , This increase in carrier concentration improves electrical conductivity, as more charge carriers contribute to current flow. The correlation of laser power’s impact on LSG conductivity is closely linked to self-heating and current crowding effects. Increasing laser power generally enhances LSG conductivity by promoting higher degrees of graphitization, producing larger graphene flakes, and reducing defects. This enhances the charge carrier mobility of LSG, which increases electrical conductivity and, consequently, leads to improved self-heating. In Figure f, it can be seen that increasing laser power leads to an increase in electrical conductivity. However, the overall electrical conductivity of LSG is observed to increase significantly. This phenomenon may be attributed to enhanced charge carrier mobility resulting from laser power-driven graphitization, where increased sp²-hybridized carbon domains improve structural properties and reduce electron scattering at grain boundaries.

3.3. Self-Activated LSG

To investigate the electrical and thermal properties of LSG, surface temperature distribution analysis was performed on a self-activated LSG (10 × 25 mm² rectangular geometry) under applied voltages using a high-resolution infrared thermal imaging camera. The IR imaging system enabled spatially resolved mapping of localized Joule heating phenomena with resistivity variations. Figure a presents the temperature–voltage characteristics of the LSG with sensing layer lengths from 1 to 9 mm, fabricated using different laser powers. Depending on the applied voltage, the LSG exhibits a thermal distribution ranging from 25 to 440 °C. Notably, the 1 mm-length LSG shows a maximum surface temperature of approximately 430 °C at an applied voltage of 7 V. The 1 mm-length LSG specimen demonstrated a maximum surface temperature of ∼430 °C under 7 V applied bias (Figure b), with thermal gradients reflecting localized resistivity variations in its porous network. The LSG exhibited stable thermal distribution over 30 min at applied voltages. However, under exceeding voltages of 9 V, localized Joule heating generated excessive thermal stress, resulting in rapid electrical pathway degradation and eventual disconnection within 15 s. This extreme Joule heating behavior arises from current crowding effects at microstructural discontinuities between electrodes and the sensing layer. To understand the enhanced heat distribution of the LSG, these current crowding effects were confirmed through COMSOL simulations, which showed that the thermal effect was enhanced in the porous microstructure, as shown in Figure c. A direct comparison between COMSOL simulation results (169 °C) and experimental temperature data (166 °C) for LSG sensors often shows that when assumptions and parameters are properly set, the simulation can yield temperature values closely matching experimental results. Table S1 summarizes the typical parameters used for simulating graphene self-heating in COMSOL Multiphysics. These parameters are adapted to the graphene sample dimensions and electrical conditions and benchmarked with experimental conditions for accurate modeling of Joule heating and temperature rise effects. Enhanced NO _X vapor detection at relatively low operating temperatures (∼200 °C) is attributed to the superior oxidizing capability of NO_X species compared to ionized oxygen adsorbates [O₂ , O^–]. The maximum sensor response for 10 ppm of NO₂ of the LSG sensor fabricated at 36 W laser power was achieved at an applied voltage of 7 V. Moreover, the 3-mm-length LSG-based chemiresistive sensor demonstrated excellent long-term stability at RT over 42 days under a bias voltage of 7 V. From these results, the 3-mm-length LSG sensor fabricated with 36 W laser power and operated at 7 V was determined to be an optimized sample with stable electrical conductivity for application as a self-activated chemiresistive sensor (Figure d).

3.

(a) Change of surface temperatures for LSG-based chemiresistive sensor with various lengths (sensing layer) according to applied voltages. (b) Thermographic camera images of LSG sensor fabricated at 36 W laser power under an applied voltage of 7 V. (c) COMSOL simulation of the temperature distributions in the porous microstructure. (d) Plots of the temperature of LSG sensor under the applied voltage of 7 V for 42 days. The inset shows thermographic camera images of the LSG-based chemiresistive sensor during the stability test.

3.4. Typical Responses of Self-Activated LSG

Recently, graphene-based chemiresistive sensors have emerged as leading candidates for next-generation sensing platforms due to their exceptional surface-to-volume ratio, which enhances molecular adsorption kinetics and charge transfer efficiency. However, these carbon-based materials face persistent challenges in achieving simultaneous selectivity and reversibility (full recovery) for ideal sensing performance at RT. These limitations originate from nonspecific molecular adsorption at structural defects and inherently slow desorption kinetics. To address these challenges, thermal energy modulation plays a major role in enhancing reaction rates. To evaluate the practical viability and operational reliability of the self-activated LSG-based chemiresistive sensor, we characterized its response to 10 ppm of NO₂ vapor under varying bias voltages (1–7 V) at RT. Prior to sensing measurements, the fabricated sensors underwent an 80-h electrical aging process at a 7 V bias to stabilize thermal and electrical baselines, ensuring reproducible gas sensing performance. As shown in Figure a, the maximum sensor response was achieved at a bias voltage of 7 V. This result suggests that the optimal operating temperature for NO₂ vapor detection is approximately 160 °C, which is consistent with previous reports on chemiresistive sensor performance. Therefore, the operating temperature influences the adsorption kinetics at the sensor’s active surface, thereby modulating the overall sensor response. At temperatures below 160 °C, the NO₂ adsorption and surface reaction of LSG are less promoted due to insufficient thermal energy, making it difficult to overcome the activation energy barrier. Consequently, the response amplitude of the LSG-based chemiresistive sensor increases with temperature, reaching its maximum at 160 °C. However, when the operating temperature exceeds 200 °C, desorption processes at the active surface are expected to dominate, leading to a decrease in the response to NO₂. Figure b shows the sensor responses to various gases, including 10 ppm of NO₂, CH₄, CO, H₂S, NH₃, C₆H₆, SO₂, H₂, C₂H₅OH, and CH₂O. Notably, the LSG-based chemiresistive sensor exhibits a high response of 12.5% to NO₂. The outstanding sensitivity and selectivity of the LSG-based chemiresistive sensor for NO₂ are attributed to the efficient gas diffusion enabled by its large specific surface area and the optimized sensing temperature of 160 °C. The enhanced response to NO₂ vapor at relatively low temperatures (<200 °C) arises from the fact that NO₂ acts as a stronger oxidizing agent than ionized oxygen species (O₂ and O^–). In contrast, the responses to other gases below 200 °C are negligible because of insufficient thermal energy. This demonstrates the improved selectivity achieved through chemical sensitization, which generates active reaction sites (NO₂ ) at relatively low temperatures. Moreover, to achieve further improved gas selectivity, we are currently developing high-performance sensor arrays (3 × 3 metrics) based on LSG and noble metal nanoparticle/LSG composites for future applications. As shown in Figure c, the responses of the LSD-based chemiresistive sensor at 20, 50, and 90% relative humidity (RH) were approximately 12.3, 12.8, and 12.7%, respectively.

4.

(a) Response curves of LSG-based chemiresistive sensor for 10 ppm of NO₂ according to various applied voltages at room temperature (RT). (b) Response patterns of LSG-based chemiresistive sensor to 10 gases (10 ppm of NO₂, CH₄, CO, H₂S, NH₃, C₆H₆, SO₄, H₂, C₂H₅OH, and CH₂O) at RT at 7 V according to ΔR (%) = [(R _air/R _gas) – 1] × 100 for oxidizing gases or ΔR (%) = [(R _gas/R _air) – 1] × 100 for reducing gases, respectively. (c) Response curves and (d) response and base resistance (R ₀) of the LSG-based chemiresistive sensor to 10 ppm of NO₂ at RT with 7 V under a relative humidity (RH) atmosphere ranging from 20 to 90%. (e) Resistance changes of LSG-based chemiresistive sensor with self-heating (160 °C) for NO₂ concentration ranging from 1 to 20 ppm. (f) Plot of the detection limit of LSG-based chemiresistive sensor for NO₂.

Under high RH conditions, water molecules adsorbed on the LSG surface donate electrons to the valence band, thereby decreasing the number of holes. As a result, the R ₀ of the LSG sensor increases due to the greater separation between the Fermi level and the valence band. The temperatures of the LSG sensor at various voltages are 27 °C at 1 V, 43 °C at 3 V, 82 °C at 5 V, and 160 °C at 7 V, respectively. As illustrated in Figure , as the temperature increases in air, oxygen molecules capture more electrons, shifting the Fermi level of graphene further toward the valence band and thereby decreasing the sensor’s initial resistance. Consequently, the resistance changes as the applied voltage increases. Overall, the LSG sensor exhibits a weak dependence on environmental humidity, which can be attributed to the relatively weak binding force of OH^– at elevated temperatures (Figure d). Figure e presents the real-time response curves of the LSG-based chemiresistive sensor as a function of the NO₂ concentration, ranging from 1 to 20 ppm. As a result, the sensor demonstrated excellent selectivity and sensitivity toward NO₂ vapor, as well as complete recovery, in comparison with other chemical vapors. In general, the times to reach 90% variation in resistance upon exposure to detecting gas and air are defined as the 90% response time and recovery time. We indicated the response of the sensor at a point of 90% response time using the following formula: ΔR = [(R _gas/R _air) – 1] × 100 (%) for reducing and oxidizing gases. As shown in Figure S1, the 90% response time and recovery time are 400 and 900 s. The LSG-based chemiresistive sensor exhibited fast response time and full recovery within a short period of time upon exposure to 10 ppm of NO₂ vapor. For comparison with previous graphene sensors, the outstanding sensing characteristics of the LSG-based chemiresistive sensor are summarized in Table . Notably, the sensor consistently returned to its baseline resistance (R ₀) within 1000 s for NO₂, which can be attributed to the sufficient thermal energy provided at 160 °C. To quantitatively evaluate the theoretical limit of detection (LOD) of the LSG-based chemiresistive sensor for NO₂ gas, the sensor responses were plotted as a function of gas concentration on a linear scale, as shown in Figure f. The LOD for NO₂ was determined by using a linear fit of the response data. The LOD represents the minimum concentration of an analyte that can be reliably detected by the sensing element with a specified probability. Each response was measured three times to obtain the average value, thereby ensuring reliability and accuracy. The LOD was calculated using eqs and .

$${\mathrm{RMS}}_{\mathrm{noise}}=\sqrt{\frac{S^2}{N}}=0.026$$ 1

$$\mathrm{LOD}(\mathrm{ppb})=3\times \frac{{\mathrm{RMS}}_{\mathrm{noise}}}{\mathrm{slope}}=2.68$$ 2

5.

Schematic illustration of the expected NO₂ sensing mechanism of a self-activated LSG.

1. Summary of Room-Temperature Graphene-Based Gas Sensor Properties and Their Fabrication Methods.

sensing material	fabrication method	response (%)/concentration (ppm)	limit of detection (ppb)	response/recovery time (s)
SnO2–rGO composite	solvothermal	3/80	209	>5.6/14.1
rGO/SnO2	CVD + RIE	17/10	15.70	>500/>4000
TiO2 NPs/PrGO	hydrothermal	8.2/10	114
3D SnO2/RGO	electrospinning	3/4.5	20	900/>1600
GO/rGO	laser microfabrication	18.23/100	230	>500/>1000
graphene nanomesh	ethanol-chemical vapor deposition	11/10	15	900/>6000
alkali-lignin-assisted edge-oxidized graphene (EGO-AL)	inkjet print (direct writing)	6.2/10	12.7	3000/>5000
laser-induced graphene (LIG)	CO2 laser	2.5/10		500/>1000
metal (Ag)-doped LIG	CO2 laser	2/1000		>500/>1000
LIG	CO2 laser	6/1	4	134/388
LIG-PI aerogel	CO2 laser	36/1	2.3	36/286
laser-scribed graphene (LSG) (this work)	CO2 laser	7/1	2.68	400/750

RMS_noise is the root-mean-square standard deviation of the noise. Based on 350 data points (N) and a standard deviation (S) of 0.4846 from the baseline curve of the LSG-based chemiresistive sensor, the calculated RMS_noise value is 0.026, which fits well with eq . The LOD for NO₂, derived from the calculated RMS_noise, is 2.68 ppb at a signal-to-noise ratio of 3, as defined by eq . Experimental validation of measured gas detection near the LOD for LSG sensors demonstrates that the experimentally determined LOD closely matches or slightly exceeds the calculated theoretical LOD, confirming reliable sensitivity near the theoretical limit (Figure S2). In this study, the response curve shows significant background noise with a nonsmooth profile, indicating challenges in distinguishing gas signals near the LOD for NO₂. To address this, advanced filtering methods such as moving average and Savitzky–Golay filters and environmental controls for temperature, humidity, and electromagnetic shielding are being implemented. These improvements are expected to yield smoother curves, reduced noise, and more accurate LOD measurements for LSG sensors. Nevertheless, this detection limit is significantly lower than the national environmental standard for NO₂, which is 200 ppb as a 1 h average. The previous publication shares certain similarities with our work, particularly in the use of porous laser-induced graphene (LIG), its self-heating properties, and ultrasensitive NO₂ detection. However, there are also significant differences from our results. For example, in the referenced study, highly sensitive nanomaterials such as MoS₂ and reduced GO (rGO)/MoS₂ were dispersed onto the LIG sensing region to fabricate an ultrasensitive chemiresistive NO₂ gas sensor. Benefiting from the large surface area and high porosity of LIG, the abundant and specific active sites of MoS₂, and the potential formation of rGO/MoS₂ p–n heterojunctions, the sensor exhibited relatively fast response/recovery speeds and excellent selectivity, even at slightly elevated temperatures. In contrast, our study employed a simple direct-writing approach using a high-power CO₂ laser (36 W) to fabricate an improved porous LSG, eliminating complex synthesis processes. In addition, the electrode and sensing layer design in our work was optimized to maximize the current crowding effect. As a result, compared to the reported self-heating temperature of 39.8 °C at 7 V for LIG in the above study, our LSG sensor achieved a substantially higher temperature of 160 °C at the same voltage (7 V). This elevated temperature of 160 °C is particularly favorable for NO₂ detection, leading to an enhanced sensing performance. Therefore, laser processing parameters such as power, scan speed, and focus point critically influence the microstructure and defect density of LSG, which directly affect its gas sensing performance. In detail, LSG sensors excel in high sensitivity, fast response times, and wide detection ranges due to their porous 3D structure and tunable properties via laser parameters like power and speed. , For example, the increased scan speeds enhance hydrophilicity, high conductivity, and multifunctional sensing performance through hybrid structures, such as LSG/MoS₂, enabling selectivity for gases. These properties make them suitable for wearable and real-time environmental monitoring applications. , Adjusting these parameters can shift LSG morphology from sheet-like to needle-like or porous structures, impacting the active surface area and sensor sensitivity. For example, needle-like LSG fabricated at 0.6 W demonstrated the highest NO₂ response and ultralow detection limits, highlighting the importance of optimized laser conditions for enhanced sensor function. ,

3.5. NO₂ Sensing Mechanism of Self-Activated LSG

As illustrated in Figure , the capture of electrons by oxygen molecules induces a shift of the Fermi level of the graphene toward the valence band. Similarly, NO₂ molecules, as strong oxidizing agents, withdraw electrons from the graphene surface. Consequently, the Fermi level of the graphene shifts further toward the valence band, increasing the number of holes available in the valence band and thereby causing a significant decrease in sensor resistance upon exposure to NO₂. Once NO₂ is removed from the environment, the equilibrium of surface species is restored, the Fermi level shifts back toward the conduction band, and the electrical resistance of the sensor returns to its baseline value in air. Moreover, self-activated temperatures enhance the energy available for carrier excitation and gas–surface interactions. In NO₂ gas sensing, defect sites and functional groups on graphene serve as active sites that facilitate enhanced gas adsorption and electron transfer at elevated temperatures. These interactions induce changes in the Fermi level position and electrical resistance, which constitute the fundamental basis of the sensor signal. When the sensor is exposed to NO₂ gas, a complex series of electron transfer reactions occurs due to interactions between the NO₂ molecules and oxygen species preabsorbed on the p-type surface layer of graphene. These reactions involve NO₂ capturing electrons from both the preabsorbed oxygen species (O₂ , O^–, and O^2–) and the sensing material itself (eqs and ). This electron transfer process increases the concentration of holes, which are the majority carriers in p-type graphene, thereby enhancing the electrical conductivity of the sensor.

$${\mathrm{NO}}_2(\mathrm{gas})+{\mathrm{e}}^{-}\to {\mathrm{NO}}_2^{-}(\mathrm{ads})$$ 3

$${\mathrm{NO}}_2(\mathrm{gas})+{\mathrm{O}}_2^{-}+2{\mathrm{e}}^{-}\to {\mathrm{NO}}_2^{-}(\mathrm{ads})+2{\mathrm{O}}^{-}$$ 4

3.6. Long-Term Stability of Self-Activated LSG

Long-term stability is a critical factor for the practical application of chemiresistive sensors. As shown in Figure a, the base resistance (R ₀, black) and the resistance change after NO₂ reaction (R _NO2 , red) of LSG sensors were measured with 10 ppm of NO₂ at RT under a 7 V bias for 35 days in dry air. It is noted that our long-term stability sensing system is designed to automatically record resistance values in air and NO₂ at fixed intervals every 3 days over a 35-day period. The total response variation was approximately 12.5 ± 0.4% air for 35 days. To assess the repeatability of the LSG sensor, the LSG-based chemiresistive sensor was exposed to 10 ppm of NO₂ vapor over five successive cycles for a total of 29,000 s (Figure b). A consistent response amplitude of 13.5 ± 0.6% for NO₂ vapor was observed, indicating excellent repeatability and stability. Therefore, LSG plays a crucial role in enhancing long-term reliability as well as improving the thermal and sensing properties of the sensor. It is noteworthy that the response and recovery times of the LSG-based chemiresistive sensor are significantly improved compared to previously reported NO₂ sensors based on carbon-based materials. A detailed comparison between the LSG-based chemiresistive sensor and previously reported RT graphene-based gas sensors is provided in Table . − It should be noted that the sensitivity, response, and recovery times, and detection limit for NO₂ of the LSG-based chemiresistive sensor are significantly improved compared to those of the reported RT graphene-based gas sensor and LIG-based gas sensors for NO₂ at RT. ,

6.

(a) Baseline resistance (R ₀, black) and the resistance change after NO₂ reaction (R _NO2 , red) of LSG sensors were measured with 10 ppm of NO₂ at RT at 7 V bias for 35 days in dry air. (b) Response curves of LSG-based chemiresistive sensor upon exposure to 10 ppm of NO₂ vapor in five successive cycles, demonstrating the long-term stability of LSG-based chemiresistive sensor.

However, several previous LIG sensor studies have demonstrated a superior detection performance for NO₂ gas compared to the results of this study. Yang et al. demonstrated the stretchable LIG-based NO _X sensors that achieved a NO₂ LOD of approximately 4 ppb at RT. The devices featured needle-like LIG structures and moisture-resistant encapsulation, exhibiting reasonably fast response (134 s) and recovery times (388 s) for 1 ppm of NO₂. Also, Ji et al. reported highly graphitized microporous LIG sensors showing a response of about 36% to 1 ppm of NO₂ at room temperature and an ultralow LOD of around 2.3 ppb, representing the state of the art for carbon-based NO₂ sensors. Therefore, these results demonstrate the outstanding performance of LSG (or LIG)-based sensors, which feature 3D porous morphology, high conductivity (>10³ S/m), and RT operation without heaters, enabling power consumption below 1 mW. These characteristics promise to position LSG as a cutting-edge sensor in low-power environmental monitoring applications. It turns out that our LSG gas sensor, fabricated using high-power lasers to enhance the porosity of micro-/nanostructures, exhibits fast response times and high sensitivity, outperforming existing LIG gas sensors, which feature fluffy, fiber-like, or needle-like microstructures produced by low-power lasers. ,

When compared with previously published Brunauer–Emmett–Teller (BET) measurement results (Figure S4), the LSG sensor shows the highest specific surface area. , Furthermore, unlike conventional synthesis processes such as chemical vapor deposition (CVD), laser microfabrication, the hydrothermal method, and the electrospinning technique, the LSG method does not require prolonged ultrasonication or stirring steps. This makes it a simple, cost-effective, stable, and reliable approach for fabricating chemiresistive sensors. Therefore, LSG sensors have promising use in flexible wearable devices for stress, temperature, and humidity monitoring, thanks to their mechanical strength, conductivity, and fast response times. Their low-cost, facile fabrication, and ability to support multifunctional sensing (e.g., combining biochemical, temperature, and strain signals) further distinguish them from conventional sensors, enabling applications in environmental monitoring, healthcare diagnostics, and smart wearable technology.

4. Conclusion

In this study, we demonstrated the fabrication and application of LSG in a fully reversible chemiresistive sensor for NO₂ gas detection. The LSG was produced by direct CO₂ laser writing on a polyimide film, resulting in a porous structure with excellent electrical conductivity and high surface area, as supported by both Joule heating measurements and COMSOL simulations. These analyses revealed that the porous LSG architecture plays a crucial role in maintaining a stable operating temperature of 160 °C, which is essential for long-term sensor reliability. The long-term thermal stability of the LSG-based chemiresistive sensor significantly enhances its NO₂ sensing performance. Remarkably, the sensor maintained a consistent temperature of 160 °C during a 42-day stability test, without any observable changes in thermal distribution, thereby demonstrating exceptional durability and operational stability. As a result, LSG was successfully integrated into a fully reversible and self-activated gas sensor platform for NO₂ detection. The LSG-based chemiresistive sensors exhibited highly sensitive and selective detection of NO₂, achieving low detection limits in the several parts per billion range, which is well below typical environmental standards. This high sensitivity, combined with excellent selectivity and full recovery, underscores the sensor’s suitability for trace-level NO₂ monitoring. We propose that the self-activated LSG-based chemiresistive sensor developed in this study holds significant promise for a wide range of electronic applications, including smart clothing and wearable electronics, that require robust thermal endurance and long-term operational stability. The simple, scalable, and cost-effective nature of the LSG fabrication process further supports its potential for integration into next-generation flexible and wearable sensing platforms.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c10232.

• Additional COMSOL simulations, resistance, response, and BET analysis results: COMSOL simulation of self-heating in LSG sensors (Table S1); raw resistance data of LSG sensors under varying RH (Table S2); measurement of response and recovery times of the LSG sensor (Figure S1); response curves of the LSG sensor near the LOD (Figure S2); resistance changes of self-activated LSG sensors for indicating response variations (error bars) (Figure S3); and comparison of BET measurement results of previously published LIGs (Figure S4) (PDF)

All authors contributed to the conception and design of the study and have approved the final version of the manuscript. S.B., C.-H.L., C.-B.P., and H.G.M. performed most of the experiments, including material preparation, data collection, device fabrication, gas sensing measurements, and data analysis. S.B. and H.G.M. wrote the first draft of the manuscript. M.-J.P., J.-S.S., and S.K. contributed to the stability measurement. C.-H.L. and H.G.M. helped in revising the manuscript. All authors discussed the results and commented on the manuscript.

The authors declare no competing financial interest.