Upcycled Graphene Oxide Nanosheets for Reversible Room Temperature NO₂ Gas Sensor

Abstract

Graphene oxide (GO) nanosheets, as one of the most studied graphene derivatives, have demonstrated an intrinsically strong physisorption-based gas–matter behavior, owing to its enhanced volume–surface ratio and abundant surface functional groups. The exploration of efficient and cost-effective synthesis methods for GO is an ongoing task. In this work, we explored a novel approach to upcycle inexpensive polyethylene terephthalate (PET) plastic waste into high-quality GO using a combination of chemical and thermal treatments based on a montmorillonite template. The obtained material had a nanosheet morphology with a lateral dimension of around ~2 µm and a thickness of ~3 nm. In addition, the GO nanosheets were found to be a p-type semiconductor with a bandgap of 2.41 eV and was subsequently realized as a gas sensor. As a result, the GO sensor exhibited a fully reversible sensing response towards ultra-low-concentration NO₂ gas with a limit of detection of ~1.43 ppb, without the implementation of an external excitation stimulus including elevating the operating temperature or bias voltages. When given a thorough test, the sensor maintained an impressive long-term stability and repeatability with little performance degradation after 5 days of experiments. The response factor was estimated to be ~11% when exposed to 1026 ppb NO₂, which is at least one order of magnitude higher than that of other commonly seen gas species including CH₄, H₂, and CO₂.

1. Introduction

Two-dimensional (2D) materials, composed of single or a few atomic layers, have emerged as promising options for gas sensing applications due to their outstanding sensitivity, superior selectivity, and minimal power consumption [1,2,3,4,5,6]. The interaction between gas molecules and the surface of 2D materials is likely distinct from the traditional chemisorption mechanism, as it appears to adhere more closely to the physisorption model, with no participation from adsorbed oxygen [1,6,7]. Specifically, the direct interaction between gas molecules and the surface of a material instigates the formation of electrical dipoles at the interface between the gas and the material, facilitated by mechanisms of charge transfer [1,3,7,8,9]. The quantity of gas molecules adhering to the surface hinges upon the adsorption energy of the two-dimensional (2D) material, whereas the extent and direction of charge transfer are contingent upon the positions of electronic bands within the 2D material relative to those of the gas molecules [7,8,9]. As a consequence, the emergence of interfacial electrical dipoles prompts a significant reorganization of charge carriers within the ultra-thin 2D material matrix, ultimately resulting in discernible fluctuations in electrical resistance, which serve as the transduction output signal [1,6,7,8,9]. The early exploration of graphene and its derivatives demonstrated the potential of developing such physisorption-based gas sensors that operate at room temperature [2,10,11,12]. However, the high surface adsorption energies of these materials often result in extended periods for desorption or, in certain instances, failure during the recovery phase of targeted gas molecules. This renders the sensors permanently nonfunctional or exhibiting impractically long recovery kinetics [10,13,14,15].

The gas reactivity of low-dimensional carbon materials, such as carbon nanotubes and graphene, is heavily reliant on the density of defects induced by surface functionalization with oxygen groups. These groups play a pivotal role in the electrical response to gas molecules [16,17,18]. Consequently, precise control over the introduction of oxygen functional groups is essential for enhancing the performance of graphene-based sensors, as it amplifies the adsorbate binding energy and promotes charge transfer at reactive sites, since pristine graphene gas sensors often encounter challenges during the recovery process in room temperature experiments without thermal treatment or light excitation [14,19,20,21]. A method for optimizing the oxidation degree as well as the homogeneous distribution of oxygen groups on graphene surfaces remains to be fully established. Nonetheless, the most popular synthesis method for graphene oxide (GO) until now is Hummer’s method, which was developed back in 1958 [22]. In this method, the oxidization of graphite is achieved through the harsh treatment of graphite powder with a concentrated H₂SO₄ solution containing KMnO₄ and NaNO₃ [2,23,24,25]. The disadvantages of this method include a complex procedure and the use of relatively expensive carbon sources. The flash graphene method is an emerging synthesis approach for upcycling inexpensive carbon sources, such as coal, carbon black, rubber tires, and mixed plastic wastes, into high-quality graphene and its derivatives with a flash joule heating process in less than one second [26,27,28]. Despite its high efficiency, the flash graphene method involves high-voltage electric discharge with large capacitor banks, which introduces potential safety risks into conversion operations [29,30,31]. Different from the Flash Joule Heating method, solvolysis is carried out through nucleophilic substitutions involving an excess of solvent molecules. Plastic solvolysis has been conducted using various solvent systems, such as methanol, polyols, and ammonia, typically at temperatures around 300 degrees Celsius, with the addition of a catalyst to facilitate depolymerization [32,33,34]. This process yields a complex mixture of products including alcohols, aldehydes, and others, leading to challenges in separation due to their diverse chemical properties and similar boiling points [35,36,37].

In this work, we successfully synthesized graphene oxide nanosheets through a combination of chemical and thermal treatments using a montmorillonite template. Different from the Hummers method, inexpensive PET plastic wastes such as plastic bags and beverage bottles instead of graphite powder are used as the carbon sources in this novel synthesis method. Subsequently, the morphology, crystalline system, chemical composition, semiconductor characteristics, and material band structures of the obtained GO nanosheets were all investigated and extensively discussed in this work. Finally, the material was realized as a gas sensor to further explore its physisorption-based gas–matter interaction. As a result, the GO sensor exhibited an excellent room temperature sensing performance towards ultra-low-concentration NO₂ gas in a fully recoverable manner, reaching a limit of detection (LOD) of 1.43 ppb. When given a thorough test, the sensor demonstrated extraordinary long-term stability and selectivity [38].

2. Materials and Methods

2.1. Material Synthesis and Preparation

The synthesis of graphene oxide (GO) followed a multi-step process. Initially, PET plastic was extracted from plastic beverage bottles, and the central, thin portions were obtained by removing the head and tail rigid segments (Figure S1). The subsequent cleaning involved stirring the waste fragments in a 500 mg/L NaOH solution, followed by triple washing with deionized water and vacuum drying at 80 °C for 5 h. The cleaned waste fragments were then combined with 16 g of organically modified montmorillonite (OMMT-N30), and the mixture was subjected to heating at a rate of 15 °C/min to 700 °C (Figure S2). Refluxing with fuming nitric acid at 110 °C for 3 h and then going through another thermal treatment at 400 °C resulted in a product consisting of montmorillonite and graphene oxide, which was confirmed by XRD and Raman analysis (Figure S3).

2.2. Material Characterization

The lateral dimensions and thicknesses of the GO flakes were measured using Bruker Atomic Force Microscopy Dimension Icon (Bruker Corporation, Billerica, MA, USA) equipped with a Multi75E-G tip. The GO crystal structure was characterized using a JEOL 2100F equipped with an EDS detector at an accelerating voltage of 200 kV. XRD was characterized using a Bruker D4 ENDEAVOR (Bruker, Billerica, MA, USA) equipped with a monochromatic Cu Kα radiation source (λ = 0.154 nm). HAADF-STEM images were obtained using a JEM-ARM300F GRAND ARM microscope operating at an acceleration voltage of 300 kV. Raman spectra were acquired using a HORIBA LabRAM HR Evolution spectrometer with an excitation wavelength of 532 nm, while photoluminescence (PL) spectra were recorded using a diode laser with an excitation wavelength of 405 nm. Additionally, X-ray photoelectron spectroscopy (XPS) and its valence data were collected using a Thermo Scientific K-Alpha XPS spectrometer equipped with a monochromatic Al Kα X-ray source (hυ = 1486.7 eV). Additionally, ultraviolet–visible (UV–Vis–NIR) spectroscopy was performed using a Cary 500 spectrometer, and Tauc plots of the samples from the UV–Vis data were calculated to identify the optical bandgap.

2.3. Sensor Fabrication and Measurements

The transducing substrates utilized in this study comprised 200 pairs of gold interdigital electrodes (IDEs) purchased from HORX Sensortech, Melbourne, Australia. Before drop-casting onto the transducing substrate, a solution of graphene oxide (GO) was prepared in 10 mL of ethanol and sonicated for 10 min to ensure the uniform dispersion of the nanoflakes in the ethanol solution. Subsequently, a suspension containing 12 mg/mL of graphene oxide black powder was drop-casted onto the transducing substrate to create the NO₂ sensor. This deposition was performed within a 4.2 mm diameter exposed area while maintaining a temperature of 25 °C to keep the electrical conductivities of the samples within the measurable range using a multimeter. Gas sensing measurements were conducted at room temperature in a dark environment. The sensor’s resistance was monitored during the gas sensing experiments using an Agilent 34401A digital multimeter, facilitated by a minimized probe stage built inside a customized gas chamber. Additionally, a computerized multi-channel gas calibration system was used to regulate the incoming gas stream at a constant flow rate of approximately 100 standard cubic centimeters per minute (sccm) into the gas chamber through the inlet. To maintain a gas pressure of 1 standard atmosphere within the chamber, a back pressure regulator (Hy-Lok Oceania, Melbourne, Australia) was connected to the outlet.

2.4. Statistical Analysis

A smoothing technique involving a five-data-point average window was employed on the raw XPS data to create the fitted curve. The limit of detection (LOD) was determined using the formula (3.3 × σ)/S, where σ represents the standard deviation of the signal-to-noise response, and S denotes the slope of the calibration curve for the dynamic response.

3. Results

3.1. Material Characterization

The graphene oxide (GO) nanosheets were synthesized from plastic wastes through chemical reactions (details are presented in the Material Synthesis and Preparation section). From the atomic force microscopy (AFM) image shown in Figure 1a, a typical GO nanosheet showed an ultra-thin sheet morphology with a lateral dimension of ~2 µm and a thickness of ~3.11 nm, which falls into the main range of statistical distribution between 1.8 and 2.3 µm for the lateral dimension and between 2 and 4 nm for thickness (Figure S4). Such a sheet-like morphology was further confirmed by the transmission electron microscope (TEM) image shown in Figure 1b. An X-ray diffraction analysis (XRD) was utilized to study the crystal structure of the obtained GO; as shown in Figure 1c, a substantial sharp peak was found at ~12° of 2-theta, which can be associated with the (001) plane [39,40,41,42]. From the high-resolution transmission electron microscopy (HRTEM) image with the corresponding selected area electron diffraction (SAED) pattern (Figure 1d and the inset), the GO nanosheet exhibited a polycrystalline nature with a d-spacing of 0.74 nm, measured as the (001) plane, which is in good agreement with previous studies as well as our XRD result [40,41,42,43,44]. The elemental composition of the material was investigated using electron spectroscopy (EDS); as shown in Figure 1e, the EDS mappings of C and O elements were identified and matched well with the electron image (Figure 1e), revealing that the formed graphene nanosheet was simultaneously oxidized during the synthesis process, eventually converting it to graphene oxide.

X-ray photoelectron spectroscopy (XPS) was conducted to study the chemical compounds in the GO nanosheet (Figure 2). In the C 1s spectrum shown in Figure 2a, three peaks were observed at the binding energies of ~285, ~286.6, and ~288.9 eV, which can be assigned to the non-oxygenated ring C–C, the carbonyl C=O, and the carboxylate carbon O=C–O, respectively [39,45,46,47]. On the other hand, according to the O 1s spectrum shown in Figure 2b, the XPS peaks of the C=O and C-O bonds were found at ~531.1 and ~532.4 eV, respectively. Before deconvolution, Shirley’s background subtraction method was employed. This step was necessary due to the occurrence of inelastic scattering events that are experienced by the photoelectrons during their transit from the excitation point to the sample surface. These events contribute to the inelastic background intensity, which must be subtracted to isolate the areas corresponding to the individual peaks. The XPS spectra of C 1s and O 1s were fitted using Gaussian–Lorentzian (GL30) functions to delineate the respective components, utilizing the characteristic binding energy (BE) values for carbon and oxygen 1s photoelectrons obtained from the relevant literature [45,47,48,49,50]. It is worth noting that the energy peak of C-O was significantly larger than that of the C=O peak, suggesting that a single bond between carbon and oxygen atoms considerably outweighs a double bond in graphene oxide, which is consistent with previous reports [51,52,53,54].

Throughout history, Raman spectroscopy has served as a tool for investigating the structural and electronic properties of graphite materials. It offers valuable insights into defects, represented by the D band, and the in-plane vibration of sp2 carbon atoms, denoted by the G band. According to the Raman spectrum shown in Figure 3a, the graphene oxide nanosheet exhibited a D band and G band at 1370 and 1593 cm⁻¹, respectively, which can be ascribed to the vibrational mode of A_1g and E_2g for the graphene oxide [55,56,57,58]. Furthermore, the Raman spectrum of the GO revealed that the intensity of the D band was slightly less than that of the G band, indicating a moderate structural disorders due to the harsh oxidation condition in the novel synthesis process [59]. Given the presence of D band peaks in the Raman spectra, defects and impurities were anticipated, and the calculated I_D/I_G ratio in this study was estimated to be approximately 0.75, leading to a lack of the pristine hexagonal structure of graphene, which is consistent with the findings of prior studies [59,60].

To further study its semiconductor characteristics, a photo luminescence (PL) measurement was conducted on the GO nanosheet, revealing a sharp PL intensity peak at ~557 nm (Figure 3b). According to the UV–Vis spectra shown in the inset of Figure 3c, the GO nanosheet showed a broad optical absorption peak at ~350 nm with an extended tail covering the range of the visible light spectrum. From the corresponding Tauc plot, derived from the optical absorption coefficient (αhv)² against the optical energy hv (Figure 3c), the energy band gap of the GO sheet was calculated to be ~2.41 eV, which is in a good agreement with the estimation of the PL result shown in Figure 3b. In addition, based on the valance band spectra shown in Figure 3d, the energy gap between the Fermi level and the valence band maximum (VBM) was determined to be 0.44 eV, suggesting that the obtained GO sample is a p-type semiconductor [61,62,63]. Given the Mott–Schottky plot shown in Figure 3e, the energy level of the conduction band minimum (CBM) for the GO nanosheet was further estimated to be ~0.2 eV vs. a reversible hydrogen electrode (RHE), resulting in the overall energy band structures for the GO sample that are presented in Figure 3f [3,61,62,63].

3.2. Room Temperature Reversible Gas Sensor

To confirm the gas sensing performance of the sample, a graphene oxide sensor was fabricated by depositing ~40 μL of material solution on a silicon-based interdigital transducer (IDT) substrate, in which 200 pairs of interdigital electrodes (IDEs) were designed with 10 µm gaps to further enhance the gas–matter interaction behavior of the nanosheet material (more details are given in the Materials and Methods section). An air-tight chamber was custom-designed to house the prepared sensor substrate. Within this chamber, a set of programmable mass flow controllers (MFCs) was connected to the inlet, allowing for adjustable gas flow from the target gas cylinders into the chamber (refer to Figure S5). The chamber’s exhaust was directed to a fume hood through a ¼ inch Teflon tube. Simultaneously, the electrical resistance of the sensor was continuously measured using a high-resolution multimeter via the integrated probe stages within the chamber. The assessment of the gas sensing performance was conducted using a response factor, which was calculated using the formula (R_analyte − R_balance)/R_balance × 100%, in which R_analyte signifies the electrical resistance of the sensor when exposed to the analyte gas species, while R_balance represents the baseline resistance in ambient air [64,65,66].

The sensor was tested at room temperature without any additional stimulus including excitonic light sources or bias voltages. In this work, NO₂ was chosen as the analyte gas owing to its intrinsic paramagnetic nature originating from its unique open-shell electron configuration [1,21,67,68]. From the dynamic response shown in Figure 4a, the GO sensor demonstrated an excellent recoverable sensing response towards ultra-low-concentration NO₂ gas ranging from 315 ppb to 1026 ppb at room temperature. The response exhibited in an almost linear trend, reaching a response factor of ~11% at 1026 ppb NO₂. Given its significantly enlarged surface–volume ratio and abundant oxygen functional groups such as hydroxyl (-OH) and carboxyl (-COOH) groups, the GO nanosheets act as electron acceptors with a strong affinity for electrons from the polar NO₂ molecules in the ambient environment, forming interfacial dipole moments between the material surface and gas molecules. Such a physisorption-based gas–matter interaction behavior causes a significant re-distribution of the charge density on the material surface, resulting in a reduction in electrical resistance for the GO sensor corresponding to the concentration of the adsorbed NO₂ gas molecules [1,6,7]. As the NO₂ gas desorbed from the sensor, its resistance gradually returned to the base level during the recovery phase (Figure 4a). Despite the absence of an external stimulus, the sensor still demonstrated complete response and recovery patterns when exposed to the low-concentration NO₂ gas at a reasonably fast rate. From Figure 4b, the sensor’s response/recovery time were measured as the elapsed time until a 90% change in the full magnitude of the response factor, reaching approximately 27 and 115 min for the response and recovery, respectively, when the sensor was exposed to 315 ppb NO₂. The sensor was further tested with other commonly seen gas species at industrially meaningful concentrations, including CH₄ (1%), H₂ (1%), and CO₂ (1%), under the same test conditions as NO₂. According to Figure 4c, the GO sensor was highly selective to NO₂ gas compared to the other gas species, which is probably due to its pure physisorption-based sensing mechanism as well as the distinct electronegativities of different kinds of gas molecules [1,6,7]. From the long-term stability and repeatability results shown in Figure 4d and 4e, respectively, the sensor demonstrated little performance degradation after the 5-day long-term stability test and an almost identical response towards 1026 ppb of NO₂ gas during three cycles of continuous experiments over 12 h. In addition, the sensor’s limit of detection (LOD), derived from three times the ratio of the noise level to the response slope of NO₂ gas, was estimated to be 1.43 ppb (Figure 4f). The response slope was calculated from a linear regression model established based on the response datasets of the sensors under various NO₂ concentrations [1,6,7].

4. Conclusions

We successfully obtained 2D graphene oxide nanoflakes through the chemical and thermal treatment of a carbon source in a controlled environment. Through morphological analysis, ultra-thin graphene oxide, with the predominant size being around 2 µm, exhibited a thickness of about 3.11 nm. According to the structural and chemical composition analysis, most of the C atoms in the pristine graphite framework were broken down and invaded by O atoms after experiencing high electric discharge from the capacitor bank and thermal annealing, leading to a crystal phase transition from an initially hexagonal to polycrystal coordination. The bandgap energy of the flash graphene oxide was expanded to ~1.43 eV, which was wider compared to the zero bandgap of pure graphene. To evaluate the gas sensing performances, the change in electrical resistance of annealed graphene oxide was investigated upon the exposure to NO₂ gas at room temperature without the application of external stimuli including light excitation and voltage biasing. A response factor of ~11.84% was found for 0.000126% NO₂ gas with an almost linear trend for concentrations between 0.0000315% and 0.000126%. In addition, a high degree of repeatability and full reversibility were demonstrated. Furthermore, the response magnitude of 0.000126% NO₂ was at least one order of magnitude larger than that of several commonly seen gases including H₂ (1%), CH₄ (1%), and CO₂ (1%), revealing the excellent selectivity for NO₂, which is rarely seen in pure semiconducting gas sensors. We believe that the physisorption mechanism governing the gas adsorption behavior over the surface of the ultra-thin annealed graphene oxide produced from the flash method enabled the fully reversible, highly selective, and room-temperature NO₂ gas sensing performance, similar to the observations of the gas interactions with many types of 2D graphene oxide. This work demonstrated the great potential of annealed graphene oxide in high-performance physisorption-based gas sensing, indicating that it could be a suitable for use in developing low-cost and power-saving devices for Internet of Things (IoT) applications.