Enhanced Sensitivity of NO₂ Gas Sensor Utilizing Fe₂O₃-Embedded ZnO Nanostructures

Abstract

This paper introduces a streamlined three-step synthesis method for crafting porous Fe₂O₃/ZnO nanofibers (NFs). Initially, Fe₂O₃ nanoparticles (NPs) were synthesized using the hydrothermal method. Subsequently, PVP NFs laden with Fe₂O₃ NPs and zinc salt were synthesized via an electrospinning method. Finally, porous Fe₂O₃/ZnO NFs were fabricated through calcination, resulting in an average diameter of approximately 100 nm. Gas-sensing experiments illuminate that the porous Fe₂O₃/ZnO NFs exhibit outstanding sensitivity, selectivity, and robust long-term stability. Although the response magnitude decreased under high relative humidity (RH) due to competitive adsorption, the sensor maintained distinct detectable responses towards NO₂ vapor at an optimum temperature of 225 °C. Particularly noteworthy is the substantial enhancement in NO₂ sensing properties observed in the Fe₂O₃/ZnO composite compared to pure ZnO NFs. This enhancement can be ascribed to the distinctive microstructure and heterojunction formed between Fe₂O₃ and ZnO.

1. Introduction

The escalating issue of air pollution, driven by expanding industrialization, particularly highlights NO₂ as a significant contributor emitted from industrial processes and vehicular combustion of fossil fuels. Recognized as an irritating red-brown oxidizing gas, NO₂ poses adverse environmental consequences, including the initiation of acid rain and contributions to climate change [1,2,3]. Its presence in ambient air necessitates urgent monitoring due to its potential threat to the ecological balance. As public awareness of environmental pollution grows alongside rapid industrial development, NO₂ emerges as a major threat. Therefore, there is imperative for us to develop NO₂ gas sensors with characteristics such as high sensitivity, selectivity, and a low detection limit to mitigate its impact on the environment effectively [4,5]. The profound impact of NO₂ on the environment and human health, stemming from burning procedures in vehicles and industrial activities, amplifies the urgency to develop cost-effective sensors for effective monitoring. Continuous exposure to NO₂ raises serious health concerns, including respiratory infections and potentially permanent lung lesions [6,7]. Hence, it is critical to maintain NO₂ concentrations below the defined permissible exposure level of 50 ppm set by U.S.EPA [8]. In response to these pressing challenges, researchers are compelled to prioritize the development of affordable and efficient NO₂ sensors. These sensors must reflect the immediate need to safeguard both the environment and human well-being by enabling rapid NO₂ detection to address associated health risks and ecological pollution.

Metal oxides, valued for their cost-effectiveness and ease of manufacturing, are widely used in diverse applications, including sensors for monitoring toxic substances and in photocatalytic wastewater degradation [9,10]. Among these, semiconducting metal oxides (SMOX) like ZnO stand out due to their n-type semiconductor characteristics, high electron mobility, and chemical stability, making them effective materials for NO₂ gas sensors [11]. The gas-sensing properties of ZnO can be finely tuned through various methods such as diverse nanostructures, doping, and loading techniques [12,13,14,15,16]. In NO₂ detection, electron transfer from the metal oxide leads to the formation of nitrite, consequently altering the resistance of n-type materials. Researchers explore a range of preparation methods including evaporation, deposition, solvent thermal, and chemical vapor deposition to optimize ZnO’s properties [17,18,19]. Acknowledging the profound impact of microstructure and surface state on gas sensor performance, studies focus on designing nanostructures, regulating exposed facets, and modulating surface defects. Nanostructured metal oxides, especially those with high specific surface areas and multi-channel structures, exhibit superior gas sensitivity.

The practical application of ZnO sensors currently faces challenges due to inherent limitations in sensitivity and response speed [20,21,22]. There is an urgent need to optimize their nanostructure and composition to significantly enhance their sensing capabilities. Notably, ZnO NFs and related one-dimensional (1D) materials have shown superior performance in gas sensing compared to bulk counterparts. This improvement is attributed to their larger specific surface area and the presence of more reactive sites, which enhance their gas detection efficiency. Additionally, doping emerges as a highly effective strategy for enhancing the gas sensing performance of ZnO sensors. By carefully modulating local structures and coordination environments through doping, researchers have achieved notable improvements in sensitivity and response speed. This multifaceted approach underscores ongoing efforts to overcome the limitations of ZnO sensors and further advance their practical utility in gas-sensing applications [23]. Hsu et al. reported that CuO-doped ZnO NFs were synthesized using sol-gel and electrospinning techniques. At a CuO-to-ZnO ratio of 0.15:1, the H₂S response was enhanced by 25% compared to pure ZnO, reaching 83.98% at 1 ppm H₂S and 200 °C, with reliable recovery and reproducibility. Fan et al. prepared a ZnO@CNF gas sensor, synthesized via electrospinning, preoxidation, and carbonization, which exhibits enhanced performance in detecting ammonia at room temperature. The integration of ZnO NPs onto carbon NFs significantly boosts sensitivity and selectivity while lowering the operational temperature [24]. Lee et al. augmented H₂ sensing performance by synthesizing Co₃O₄-loaded ZnO NFs with varying Co₃O₄ concentrations. Among the compositions tested, the 0.95ZnO-0.05Co₃O₄ sensor exhibited the highest response of ~133 to 10 ppm H₂ gas, showcasing superior selectivity. The heightened H₂-sensing capability of the optimized sensor stemmed from the synergistic effects between ZnO surface/grain boundaries and Co₃O₄ [25]. Naderi et al. used ZnO/CdO NFs, synthesized via electrospinning and calcination, to form n-n type I heterojunctions with a straddling energy band gap, which exhibits high sensitivity and selectivity to NO gas, detecting concentrations as low as 1.2–33 ppm at an optimal temperature of 215 °C, with negligible influence from humidity. With excellent repeatability, long-term stability, and rapid response times, this sensor holds promise for monitoring toxic gases in industrial, urban, and potentially exhaled breath environments [26,27].

Certainly, increasing surface area, doping, and defect engineering have emerged as promising strategies to enhance the gas-sensing performance of ZnO sensors. However, the journey to construct optimized NO₂ gas sensors and unravel their intrinsic sensing mechanisms remains an ongoing challenge. Deeper research in this area is essential, considering the current complexities, as it necessitates a thorough understanding of the intricate interplay between nanostructure, composition, and the dynamic processes involved in gas sensing. Despite the challenges, persistent exploration holds the key to unlocking innovative solutions and advancing the development of highly efficient and reliable NO₂ gas sensors.

In this study, distinct from atomic-level doping, we fabricated a heterostructured composite by utilizing a cost-effective electrospinning method to successfully synthesize ZnO NFs with embedded Fe₂O₃ NPs. The results reveal significant changes in the morphology of the ZnO NFs and the creation of numerous surface O_V defects upon the introduction of embedded Fe₂O₃ NPs. Our investigation extended to their sensing behavior towards NO₂ gas, demonstrating that 4Fe-ZnO exhibits the most robust gas-sensing performance. This is characterized by a high response (78 for 5 ppm NO₂), long-term stability, and superior selectivity compared to ZnO sensors without Fe₂O₃ NPs embedding. Importantly, our work experimentally establishes the correlation between the amount of embedded Fe₂O₃ and the concentration of oxygen vacancies. Additionally, our gas sensitivity experiments underscore the excellent gas-sensitive properties of the material, attributed to the synergistic effects of embedded Fe₂O₃ and oxygen vacancies. It is noteworthy that the adsorption energy for NO₂ on ZnO is greatly enhanced after embedding Fe₂O₃ and introducing O_V compared to pristine ZnO, indicating their crucial roles in optimizing the sensing performance of ZnO-based sensors.

2. Experiment

2.1. Preparation of Fe₂O₃ NPs

The Fe₂O₃ NPs were synthesized using the hydrothermal method, as illustrated in Figure 1a. Anhydrous ferric chloride powder (0.162 g, FeCl₃, DUKSAN PURE CHEMICALS, Seoul, Republic of Korea) was added to 50 mL of deionized (DI) water and dissolved through magnetic stirring for 15 min. Sodium hydroxide beads (0.08 g, NaOH, DAEJUNG Chemicals, Gyeonggi-do, Republic of Korea) were introduced into this solution, followed by an additional 15 min of dissolution. The solution was then transferred to a Teflon-lined autoclave and placed in a convection oven at 150 °C for 5 h. After the reactor cooled to room temperature, the sample was transferred into a 50 mL conical tube and centrifuged at 3000 rpm for 5 min. To obtain pure nanoparticles, the supernatant was decanted and the precipitate was redispersed in 30 mL of a 1:1 (v/v) mixture of ethanol and deionized (DI) water. The mixture underwent ultrasonic dispersion for 10 min, followed by centrifugation under the same conditions. This washing procedure was repeated five times to effectively eliminate residual ions. The resulting Fe₂O₃ NP powder was collected in a 50 mL vial and dried in a convection oven at 50 °C for 24 h. The final yield of the dried powder was approximately 80 mg. Finally, the powder was redispersed in 20 mL of ethanol via ultrasonic agitation for 30 min and stored for further use.

2.2. Preparation of Fe₂O₃-Embedded Porous ZnO NFs

Fe₂O₃-NP-embedded porous ZnO NFs were fabricated via electrospinning, as depicted in Figure 1b. To prepare the electrospinning precursor solution B, 7 mL of ethanol and 3 mL of N,N-Dimethyl formamide (DMF, DUKSAN PURE CHEMICALS) were mixed to form a solvent. A mixture of 0.6 g zinc nitrate hexahydrate (Zn(NO₃)₂, 99%, Alfa Aesar, Ward Hill, MA, USA) and 0.3 g citric acid (C₆H₈O₇, 99+%, Alfa Aesar) was then added with magnetic stirring for 1 h. To achieve the desired solution viscosity, 0.75 g polyvinylpyrrolidone (PVP, Mw = 1,300,000, Sigma-Aldrich, St. Louis, MI, USA) powder was added. A specific amount of solution A was added to solution B, with 0 mL, 0.2 mL, 0.4 mL, and 0.6 mL corresponding to sensor 0Fe-ZnO, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO, respectively.

For electrospinning, the syringe needle was connected to a high-voltage power supply, and the collector was grounded. The distance between the needle and collector was set at 25 cm. Then, the solution in the syringe was injected at a velocity of 0.005 mL/min by applying a 15 kV positive bias to the needle. After 30 min, a white PVP fiber containing (hereafter referred to as PVP NFs) Fe₂O₃ NPs and Zn salt was deposited on the substrate. Finally, the collected white fibers were calcined at 500 °C at a moderate rate of 10 °C/min for 2 h in air to obtain pure ZnO and Fe₂O₃-NP-embedded ZnO NFs, as shown in Figure 1c.

2.3. Characterization

The morphologies of the following samples were characterized using field emission scanning electron microscopy (FE-SEM, Hitachi S8010, Tokyo, Japan). Their crystal structures were determined from the X-ray diffractograms obtained using a glancing angle (5°) high-resolution X-ray diffractometer (XRD, D/Max-2500, Rigaku, Tokyo, Japan) with Cu Kα radiation (λ = 1.5418 Å) within scattering angles of 20–80°. The detailed morphology of the NFs and their crystal structures was examined by field emission transmission electron microscopy (FE-TEM, Jeol 2100F) equipped with energy-dispersive X-ray spectroscopy (EDS) for qualitative chemical analysis of the corresponding nanotubes. X-ray photoelectron spectroscopy (XPS, Thermo Scientific, Waltham, MA, USA, Kα) was employed to measure the binding energies of the samples. The specific surface area and pore size distributions of the samples were calculated through N₂ gas adsorption using the Brunauer–Emmett–Teller (BET, BELSORP-max, MicrotracBEL Corp., Osaka, Japan) and Barrett–Joyner–Halenda (BJH) surface analytical instruments (Osaka, Japan), respectively.

2.4. Fabrication and Measurement of Gas Sensor

The process began with patterning interdigital electrodes (IDE) onto a silicon wafer coated with a SiO₂ layer, as depicted in Figure S1. Initially, a 10 nm titanium (Ti) layer was deposited onto the substrate, followed by a 100 nm gold (Au) layer to serve as an adhesion and conductive electrode. The gap between neighboring electrodes was maintained at 10 μm. In sample preparation, three milligrams of the synthesized material were dispersed in 3 mL of ethanol through ultrasonication for one minute. Subsequently, 0.5 mL of the resulting solution was applied onto the IDE chips using a spray gun. The IDE chip was then heated to 60 °C to facilitate solvent evaporation from the mist application.

A gas-sensing setup was established to evaluate the response of the NO₂ sensor, as depicted in Figure S1. The IDE chip, coated with NFs, was connected to a source meter (B2901A, Keysight Technologies, Santa Rosa, CA, USA) via a gold wire within the sensing chamber. The sensor temperature was precisely regulated using a quartz tube furnace equipped with a PID controller, ensuring stability within ±1 °C via a thermocouple positioned near the sensor. The chamber was heated to the operational temperature, and the sensing chamber’s temperature was systematically adjusted from 150 °C to 250 °C in 50 °C increments to identify the optimal operating temperature for sensing performance. To assess sensitivity to NO₂ gas, sensors were exposed to increasing concentrations of NO₂ gas (ranging from 1 ppm to 5 ppm) for 200 s. Between exposures, the chamber was purged with synthetic air for 500 s to eliminate residual gases. For selective sensing analysis, gases including ethanol, acetone, toluene, methanol, TEA, NH₃, and CO were introduced into the chamber. Various gases (99.99%) used for assessing sensing performance were acquired from Samjung Special Gas (Seoul, Republic of Korea). Specifically, standard calibration gas cylinders were utilized for all analytes to ensure precise concentration control. A 5 ppm NO₂ cylinder (balanced in synthetic air) was used for the target gas sensing. For selectivity tests, 100 ppm standard gas cylinders (balanced in synthetic air) were employed for all interfering gases, including ethanol, acetone, toluene, methanol, TEA, NH3, and CO. These gases were diluted with additional synthetic air, and their flow rates were regulated using Mass Flow Controllers (MFCs). To clearly distinguish between gas types, the response of the sensors (R) was defined as R_g/R_a for the oxidizing gas (NO₂) where resistance increases (R_g > R_a), and as R_a/R_g for reducing gases (ethanol, acetone, toluene, methanol, TEA, NH3, and CO) where resistance decreases. Response and recovery times were recorded until the resistance change in the sensing chip reached 90% of its saturation value. Before experimentation, the gas underwent a drying process to reduce its humidity level to 0.5%. This was achieved using a DRIERITE laboratory gas-drying unit from W.A. Hammond Drierite Co. LTD (Xenia, OH, USA). Subsequently, the gas flow was divided into two streams, each regulated by a Mass Flow Controller (MFC). One stream contained the humid air, while the other was previously dried. By blending these streams, humidity levels were adjusted from nominal 0% (experimentally maintained at <0.5% RH) to 80% (equivalent to 79.2% relative humidity). The RH levels were monitored using a commercial digital humidity sensor (SK-140TRH, SATO Japan, Tokyo, Japan) before conducting any sensing measurements. For a visual representation of this process, please refer to Figure S1.

3. Results

3.1. Characterization

The low-magnification and magnified surface morphologies of the synthesized samples were analyzed using SEM, as depicted in Figure 2 and Figure S2. Figure S2 illustrates the uniform morphology of the PVP NFs before calcination and the samples after calcination. Figure 2a displays high-magnification SEM images of the PVP NFs before calcination, revealing a smooth surface with an average diameter of 100 nm. Subsequently, Figure 2b–e present corresponding SEM images of the calcined 0Fe-ZnO, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO, respectively. Upon calcination, the diameter of 0Fe-ZnO decreased to approximately 100 nm. The calcined NFs exhibited depressed and rough surfaces with numerous nanograins, promoting increased gas sensitivity by enhancing gas adsorption and desorption through enlarged contact surfaces. Conversely, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO displayed a rougher surface morphology with aggregated sub-grains compared to 0Fe-ZnO after calcination, featuring uniform diameters transformed into porous and bumpy structures comprised of numerous small sub-grains. Although these sub-grains consisted of ZnO and Fe₂O₃, discerning their compositions via SEM was challenging due to their similar morphologies.

XRD was employed to determine the composition and crystal phase of the acquired samples. As depicted in Figure 3, the prominent diffraction peaks of XRD patterns for all samples are consistent with hexagonal Wortzite-phased ZnO (JCPDS 89-1397), indicating the clear presence of ZnO. However, in addition to the ZnO peaks, the peaks of Fe₂O₃ (JCPDS 89–2810) were not observed. As the content of Fe₂O₃ is relatively low, some peaks partially overlap with those of ZnO, resulting in weaker diffraction peaks. The diffraction peaks at 2-theta (2θ) of 31.73, 34.38, 36.21, 47.48, 56.54, 64.78, 66.30, 67.87, and 69.01° correspond to the (100), (002), (101), (102), (110), (103), (200), (112), and (201) crystal planes of hexagonal ZnO.

Figure 4 illustrates the TEM images of a single 4Fe-ZnO NF. Similar to the SEM images, the NF depicted in Figure 4a exhibits a porous and agglomerated structure consisting of numerous small nanograins. A high-resolution TEM image is presented in Figure 4b, where distinct fringe patterns representing the crystal structures of ZnO and Fe₂O₃ are observable within each grain. The inter-planar spacings of the neighboring fringes measure 0.281 nm, corresponding to the (100) lattice planes of ZnO, and 0.270 nm, corresponding to the (104) lattice planes of Fe₂O₃. Furthermore, the Selected Area Electron Diffraction (SAED) pattern (Figure S3) confirmed the polycrystalline nature of the NFs. The diffraction rings were indexed to the (100) plane of ZnO and the (104) plane of Fe₂O₃, which aligns perfectly with the inter-planar spacings measured in the HRTEM analysis. A scanning TEM (STEM) image of the 4Fe-ZnO NF is depicted in Figure 4c. Additionally, Figure 4d–f display the EDS elemental mapping profiles of O, Fe, and Zn, respectively. From these images, it can be concluded that the 4Fe-ZnO NFs are primarily composed of O, Zn, and Fe. Although the precise quantification of Zn:Fe ratios via EDS was limited by the geometric curvature of the nanofibers, the intensity of the Fe signal exhibited a clear increasing trend proportional to the nominal precursor loading, confirming the successful modulation of the Zn:Fe composition. The uniform distribution of Fe signals indicates that the Fe₂O₃ nanoparticles are well dispersed on the ZnO surface without significant agglomeration. No other elements were observed in the profile images.

The XPS full survey spectra depicted in Figure 5a illustrate the elemental composition of the four samples. While 0Fe-ZnO exclusively exhibits Zn and O components, the presence of Fe in 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO indicates the successful embedding of Fe₂O₃ in NFs. Figure 5b reveals characteristic peaks at around 1044 and 1021 eV, corresponding to Zn 2p_1/2 and Zn 2p_3/2 of Zn²⁺, respectively [28]. A noteworthy observation is the slight positive shift in the binding energies of Zn 2p compared to those in 0Fe-ZnO. Furthermore, Figure 5c displays the Fe 2p XPS spectrum of the samples, showcasing distinct peaks at approximately 710 eV and 724 eV, attributed to Fe 2p_3/2 and Fe 2p_1/2, respectively, with the exception of 0Fe-ZnO [29]. This observation suggests an electrical interaction between Fe₂O₃ and ZnO, leading to electron transfer from ZnO to Fe₂O₃ and subsequent changes in binding energy [30,31].

To assess the porous characteristics and specific surface area of the samples, nitrogen adsorption–desorption isotherms were conducted on both the PVP NFs and the four samples. Figure 6a–e illustrate the porous properties and corresponding pore size distribution of each sample. The specific surface area of the PVP NFs and the four samples varied with the Fe₂O₃ content and microstructure. Specifically, their specific surface areas were measured as 2.7 m²/g, 129.03 m²/g, 135.37 m²/g, 136.55 m²/g, and 137.49 m²/g, respectively. Correspondingly, the total cumulative pore volumes were determined to be approximately 0.37 cm³g⁻¹ for the calcined samples, further confirming their porous nature. Notably, the specific surface area of the calcined NFs significantly exceeded that of the PVP NFs. Moreover, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO demonstrated type-IV isotherms with H2 hysteresis loops, indicative of a mesoporous structure present in all obtained samples. This mesoporous structure, facilitated by the calcination process, enhances gas diffusion and the utilization rate of sensitive materials, which are crucial factors for gas-sensing applications. The inset shows corresponding BJH plots, confirming the presence of pore sizes in all samples except the PVP NFs. These BET results underscore the significance of pore size and volume in influencing sensing performance.

3.2. Gas-Sensing Properties

To determine the optimal operating temperature, we investigated the sensing performance of our fabricated sensors under dry air conditions (approx. 0% RH). Figure 7a illustrates the responses of the 0Fe-ZnO, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO sensors to 5 ppm NO₂ across temperatures ranging from 150 to 250 °C. Our findings reveal that sensor responses to 5 ppm NO₂ are contingent upon operating temperature, exhibiting a characteristic pattern of initial increase followed by a decline against temperature. This pattern reflects the delicate balance between molecular activity and adsorption–desorption capacity. Specifically, both the 0Fe-ZnO and 2Fe-ZnO sensors demonstrate an initial increase in response towards 5 ppm NO₂, peaking at 200 °C before declining at higher temperatures. The 4Fe-ZnO and 6Fe-ZnO sensors exhibit an optimal operating temperature around 225 °C, where they achieve maximum response values before experiencing a decline at higher temperatures. This shift to a higher optimal temperature compared to pure ZnO (200 °C) is attributed to the enhanced adsorption energy of NO₂ on the Fe₂O₃/ZnO heterojunction. The stronger binding at the interface requires higher thermal energy to facilitate effective desorption and maintain rapid reaction kinetics. Beyond their optimal temperatures, sensors encounter challenges related to oxygen adsorption and gas diffusion as the operating temperature increased, thereby diminishing the rate of chemical reactions and subsequently reducing sensor response. To ensure a fair evaluation of each sensor’s intrinsic capability, we compared the sensing performances at their respective optimal operating temperatures (200 °C for 0Fe-ZnO and 2Fe-ZnO, 225 °C for 4Fe-ZnO and 6Fe-ZnO) rather than at a single fixed temperature. This approach allows for the assessment of the maximum potential of each sensing material. Notably, the 4Fe-ZnO sensor outperforms the 0Fe-ZnO sensor at higher temperatures, displaying an 18-fold increase in response at its optimal operating temperature. This improvement can be attributed to the formation of a heterojunction between Fe₂O₃ and ZnO. At this interface, charge carriers are efficiently separated, leading to enhanced gas-sensing properties such as increased sensitivity and selectivity. Moreover, the heterostructure facilitates improved adsorption and desorption kinetics of gas molecules, thereby augmenting the overall gas-sensing performance compared to pure ZnO.

Figure 7b illustrates the relationship between the baseline resistance (R_a) of the 0Fe-ZnO, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO sensors and the operating temperature. It is important to highlight that the R_a sequence of the four sensors at identical temperatures follows the following order: 6Fe-ZnO > 4Fe-ZnO > 2Fe-ZnO > 0Fe-ZnO. Notably, the R_a value of the 0Fe-ZnO sensor is significantly lower than that of other three sensors. The R_a of all sensors exhibits typical semiconductor behavior with respect to operating temperature, where sensor resistance decrease is inversely proportional to temperature. Initially, at room temperature, the conduction of the sensing layers involves only a minimal number of free electrons. However, as the operating temperature increases, more free electrons participate in conduction, leading to a decrease in resistance. Across all sensors, R_a decreases as operating temperature rises, attributed to the increased transition of electrons into the conduction band at higher temperatures. This characteristic behavior of resistance–temperature curves is typical for surface-controlled gas sensors.

The 0Fe-ZnO, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO sensors were also exposed to 100 ppm concentrations of various interfering gases, including ethanol, acetone, toluene, methanol, TEA, NH₃, and CO, at their respective optimal operating temperatures. The results are illustrated in Figure 7c. As observed in the radar chart, the sensors exhibited negligible responses to both organic (VOCs) and inorganic gases (NH₃, CO), despite the high concentration (100 ppm) of these interferents. Specifically, the response of the 4Fe-ZnO sensor to NO₂ was distinctly higher compared to its response to other gases, demonstrating excellent selectivity against not only VOCs, but also common environmental pollutants like CO and NH₃. Among these gases, the response of the 4Fe-ZnO sensor to NO₂ was notably higher compared to its response to other interfering gases. This observation suggests the sensor’s potential for excellent selectivity towards NO₂. Consequently, transient response measurements were conducted across different NO₂ concentrations.

Figure 7d–g present the transient response curves of all sensors to NO₂ concentrations ranging from 1 to 5 ppm, and, subsequently, from 5 ppm back to 1 ppm, conducted at their optimal operating temperatures. It is evident that all sensors exhibited a strong response to NO₂, with response values increasing alongside NO₂ concentration. Notably, the 4Fe-ZnO sensor demonstrated the highest response value, reaching 78, surpassing the other sensors. As the concentration decreased again, all sensors displayed a reduction in response to NO₂. In Figure 7h, plots depict the relationships between NO₂ vapor concentrations and the responses of the three sensors. These responses showed a gradual increase up to 5 ppm, attributed to NO₂ surface saturation.

Figure 7i illustrates the sensing transient curve of the 4Fe-ZnO sensor with low NO₂ gas concentrations ranging from 0.1 to 0.8 ppm. The corresponding responses curve as a function of low NO₂ concentration, which is depicted alongside. Even under low concentration conditions, the 4Fe-ZnO sensor exhibited observable responses to NO₂, particularly becoming apparent at 0.4 ppm. However, due to the presence of noise, responses below 0.4 ppm were relatively weaker. This suggests that the 4Fe-ZnO sensor holds significant potential as a NO₂ gas sensor even at low concentrations. These findings highlight the promising applicability of the 4Fe-ZnO sensor in detecting low concentrations of NO₂, indicating its suitability for various practical applications where precise detection of trace amounts of NO₂ is crucial.

The response/recovery times of the 0Fe-ZnO, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO sensors were found to be 22 s/198 s, 38 s/271 s, 56 s/310 s, and 96 s/399 s, respectively, as shown in Figure 7j. The 0Fe-ZnO sensor showed fast response/recovery time compared to its counterparts. This phenomenon can be attributed to several factors, including the presence of a heterojunction in the composite material. The heterojunction between Fe₂O₃ and ZnO introduces additional interfaces within the composite material. These interfaces can slow down the diffusion of gas molecules, leading to prolonged response and recovery times compared to pure ZnO, which lacks such interfaces. Secondly, the heterostructure may alter the surface chemistry and electronic properties of the sensing material. The presence of Fe₂O₃ in the heterostructured composite could introduce defects and trap sites on the surface of ZnO, which could affect the adsorption and desorption kinetics of target gas molecules. This alteration in surface properties could contribute to the longer response and recovery times observed in the heterostructured composite.

Figure 7k illustrates the repeatability transient curves of the 4Fe-ZnO sensor, obtained by subjecting it to repeated exposures to 5 ppm NO₂. Notably, no significant changes in the response–recovery kinetics were observed even after nine continuous reversible cycles, indicating excellent repeatability and reproducibility. To further evaluate the long-term stability of the 4Fe-ZnO sensor, it was exposed to 5 ppm NO₂ for 30 days, with sampling conducted every 6 days. As depicted in Figure 7l, the sensor exhibited negligible changes in both response and resistance after the 30-day sensing period, indicating robust long-term stability. This consistent electrical performance confirms that the sensing material retains its structural and chemical integrity without undergoing degradation or corrosion upon repeated exposure to NO₂ gas.

RH plays a crucial role in evaluating the sensing performance of semiconductor gas sensors in practical applications. To investigate the influence of RH on the 4Fe-ZnO sensor, the sensor response to 5 ppm NO₂ at various RH levels was tested at 225 °C, as depicted in Figure 8. As RH increased from 0% to 80%, both resistance and response noticeably decreased. This decrease can be attributed to the increased adsorption of water molecules on the surface of the sensing materials. The presence of these water molecules could obstruct the absorption of oxygen molecules, thereby impacting the reaction between NO₂ and oxygen ion species [32,33]. Despite the significant drop in response with increasing RH, even at 80% RH, the 4Fe-ZnO sensor still exhibited observable and distinct responses. However, it is worth noting that the influence of water molecules introduced more pronounced noise signals into the system. This observation indicates that, while water molecules interfere with the sensing reaction via competitive adsorption, the 4Fe-ZnO sensor retains its functionality up to 80% RH. Furthermore, the robust recovery and long-term stability shown in Figure 7k,l suggest that the humidity effect is reversible and does not induce permanent structural degradation of the sensing material.

The performance of the 4Fe-ZnO sensor was compared with other reported ZnO-based NO₂ sensors, and the findings from the literature surveys are summarized in

Table 1. Comparison of NO₂-sensing properties between our sensors and previous reports.

Sensing Materials	Con. (ppm)	Tem. (°C)	Res. (Rg/Ra)	Ref.
ZnO/ZnSe	8	200	10.42	[34]
Au-ZnO	1	150	31.4	[35]
ZnO	100	180	23.3	[36]
ZnO	2	200	1.29	[37]
ZnO	50	220	36.64	[38]
ZnO	50	270	29	[39]
ZnO	5	195	1.04	[40]
ZnO	5	250	3.3	[41]
ZnO	10	200	74.68	[42]
ZnO	100	200	37.2	[43]
Fe2O3/ZnO	5	225	78	This work

. Our study demonstrates that the 4Fe-ZnO sensor presented herein exhibits enhanced sensing performance, thereby positioning it competitively alongside other NO₂ gas sensors reported in the literature.

3.3. Gas-Sensing Mechanism

When ZnO functions as a gas sensor, the redox reaction of the gas on its surface induces a shift in the carrier concentration, leading to a variation in the material’s space charge layer thickness and, consequently, a change in resistance [44]. As a prevalent n-type metal oxide semiconductor, ZnO interacts with oxygen molecules in the ambient air. These molecules become adsorbed on the surface, capturing electrons from the conduction band near the surface. This process occurs at different operating temperatures, giving rise to the formation of distinct ionic states, namely $O_2^{-}$, $O^{-}$, and $O^{2-}$. In specific temperature ranges, $O_2^{-}$ dominates at temperatures below 100 °C, $O^{-}$ between 100 and 300 °C, and $O^{2-}$ at temperatures exceeding 300 °C. In the context of this study, with an operating temperature set at 225 °C, the prevalent surface-bound oxygen ion species are predominantly $O^{-}$ [45].

$$O_2\left(\mathrm{gas}\right)\to O_2\left(\mathrm{ads}\right)$$

(1)

$$O_2\left(\mathrm{ads}\right)+e^{-}\to O_2^{-}\left(\mathrm{ads}\right)(\mathrm{T}<100°C)$$

(2)

$$O_2^{-}\left(\mathrm{ads}\right)+e^{-}\to {2O}^{-}\left(\mathrm{ads}\right)(100°C<\mathrm{T}<300°C)$$

(3)

$$O^{-}\left(\mathrm{ads}\right)+e^{-}\to O^{2-}\left(\mathrm{ads}\right)(\mathrm{T}>300°C)$$

(4)

As depicted in Figure 9, O₂ molecules adsorb onto the material surface at the optimal working temperature of 225 °C, trapping electrons to generate $O^{-}$ (ads). This process results in the formation of an electron depletion layer, creating a potential barrier and subsequently increasing the resistance of ZnO materials. In the presence of NO₂ gas, the NO₂ molecule undergoes a reaction with the electrons in ZnO, leading to the formation of ${NO}_2^{-}$ (ads) [46], as expressed below.

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

(5)

The observed response mechanism of ZnO/Fe₂O₃ to NO₂ gas can be explained based on the work functions and semiconductor properties of ZnO and Fe₂O₃. A proposed energy band structure diagram of the ZnO/Fe₂O₃ heterojunction is elucidated schematically in Figure 9. The work function of Fe₂O₃ is greater than that of ZnO, both belonging to n-type semiconductors [47,48]. ZnO and Fe₂O₃ form an n-n heterojunction, where the Fermi level (E_F) of ZnO is higher than that of Fe₂O₃. This energy level difference prompts electrons to transfer from ZnO to Fe₂O₃ upon contact, resulting in the establishment of a depletion layer on the ZnO side and an accumulation layer on the Fe₂O₃ side [49]. Furthermore, band bending and potential barriers are formed at the intergranular boundaries of Fe₂O₃/ZnO due to the electron transfer and lead to a higher resistance state of the Fe₂O₃/ZnO material. In the presence of air, O₂ is absorbed on the surfaces of both Fe₂O₃ and ZnO, creating depletion layers in both materials. The depletion layer on the ZnO side consists of contributions from the n-n heterojunction and the adsorption of oxygen molecules [50]. Notably, the lower electron concentration in the ZnO side weakens its ability to absorb oxygen molecules from the air. Upon exposure to NO₂, the potential barrier in ZnO intensifies, leading to an increase in the height of the barrier. This is attributed to the consumption of electrons from the ZnO surface, causing the depletion layer region to broaden. Consequently, the resistance of the ZnO material further increases. In summary, the response of ZnO/Fe₂O₃ to NO₂ gas is intricately linked to the electron transfer, the formation of depletion layers, and the modulation of potential barriers, collectively influencing the electrical conductivity of the material in the presence of the target gas.

Additionally, the remarkable gas sensitivity exhibited by 4Fe-ZnO can be attributed to the synergistic effects of a significantly high specific surface area and a mesoporous structure. The sensing mechanism is visually represented in the schematic diagram in Figure 9. The porous Fe₂O₃/ZnO composite, with multiple exposed facets, facilitates a sensitive variation in the depletion region and provides abundant active sites. Consequently, NO₂ molecules in the air can easily diffuse and rapidly adsorb on the surface due to the interconnected pore channels within the porous structure. This enhances electron capture, further widening the electron depletion layer. Therefore, the structure and connectivity of the pores play a crucial role in determining the gas sensitivity performance.

In this study, the 4Fe-ZnO composite, comprising numerous sub-grains synthesized via electrospinning and thermal calcination processes, serves as the host matrix for the proposed NO₂ gas sensor. Fe₂O₃ NPs are incorporated onto the NFs to augment its sensing capabilities. Analysis of XPS spectra concerning the oxygen content reveals a pivotal mechanism for enhancing sensing performance attributed to the appended Fe₂O₃, as depicted in Figure 10. In the XPS spectra, the peaks corresponding to O 1s of the sensors are situated around 530 eV, further discernible into three distinct peaks representing lattice oxygen (O_L), oxygen vacancy (O_V), and adsorbed oxygen (O_C) [51]. O_L denotes oxygen ions within the material lattice with marginal influence on gas-sensing responses. O_V signifies oxygen ions within oxygen-deficient regions, crucially affecting the material’s gas adsorption capacity. O_C encompasses chemisorbed oxygen species, including dissociated oxygen species ($O_2^{-}$, $O^{-}$, and $O^{2-}$) and ${OH}^{-}$. Quantitative analysis of O 1s XPS spectroscopy provides valuable insights into the relative proportions of oxygen species across all samples [52], as summarized in Figure 10. Recent studies regarding the rigor of XPS O 1s peak assignment [53,54] suggest that the component at ~531 eV, often labeled as oxygen vacancies (O_V), may largely stem from surface hydroxyls or chemisorbed oxygen species associated with surface defects. While we acknowledge this debate, in the context of this study, this component is utilized as a comparative indicator to observe the relative trend of surface active states, which correlates consistently with the enhanced sensing performance. Remarkably, due to disparities in Fe₂O₃ cation content within NFs, the proportions of O_V and O_C exhibit variations among the samples. A higher total content of O_V and O_C implies increased involvement of surface-chemisorbed oxygen species in redox reactions, resulting in amplified resistance changes and enhanced gas sensor responses. Consequently, the O_V rate in O 1s peaks for 0Fe-ZnO, 2Fe-ZnO, 4Fe-ZnO, and 6Fe-ZnO are assessed at 21.04%, 21.77%, 23.12%, and 22.43%, respectively. This variation in oxygen vacancy content is attributed to the trade-off between interface density and agglomeration. The introduction of Fe₂O₃ creates heterojunction interfaces with ZnO, where the lattice mismatch induces strain that promotes the formation of oxygen vacancies to lower the system’s energy. However, in the 6Fe-ZnO sample, excessive Fe loading leads to the agglomeration of nanoparticles, reducing the effective interfacial area available for defect formation. As a result, 4Fe-ZnO exhibits the optimized defect density, which significantly contributes to its improved sensing performance of 4Fe-ZnO.

Furthermore, it is acknowledged that both insufficient and excessive incorporation of Fe₂O₃ NPs within the NFs are speculated to adversely affect gas responsiveness. The porous NFs synthesized in this study comprise a plethora of sub-grains, as illustrated in Figure 9. The gas-sensitive properties of these NFs are contingent upon the polycrystalline nature of the sub-grains. The boundaries between these sub-grains act as potential barriers to the flow of charge carriers. Oxygen species adsorbed on the surface of sub-grains give rise to oxygen species according to reactions (3). A barrier to the flow of electrons emerges at the boundaries of adjacent sub-grains [55]. In this scenario, upon introduction of the target gas, the width of the barrier expands. As the height of the barrier varies, so does the resistance of the sample.

4. Conclusions

In summary, we successfully synthesized Fe₂O₃/ZnO composites using a combination of hydrothermal and electrospinning methods. These composites were then evaluated for their gas-sensing capabilities towards NO₂. Impressively, the Fe₂O₃/ZnO composites exhibited exceptional sensitivity to NO₂ vapors, particularly at 225 °C. Notably, the response to 5 ppm NO₂ gas reached approximately 78. This heightened sensitivity can be attributed to the synergistic effects of Fe₂O₃ and ZnO, as well as the formation of an n-n heterojunction. These findings underscore the potential of the n-n Fe₂O₃/ZnO composite sensor as a promising candidate for high-performance gas-sensing applications.