Highly Selective Gas Sensor Based on Litchi-like g-C3N4/In2O3 for Rapid Detection of H2

Hydrogen (H2) has gradually become a substitute for traditional energy, but its potential danger cannot be ignored. In this study, litchi-like g-C3N4/In2O3 composites were synthesized by a hydrothermal method and used to develop H2 sensors. The morphology characteristics and chemical composition of the samples were characterized to analyze the gas-sensing properties. Meanwhile, a series of sensors were tested to evaluate the gas-sensing performance. Among these sensors, the sensor based on the 3 wt% g-C3N4/In2O3 (the mass ratio of g-C3N4 to In2O3 is 3:100) showeds good response properties to H2, exhibiting fast response/recovery time and excellent selectivity to H2. The improvement in the gas-sensing performance may be related to the special morphology, the oxygen state and the g-C3N4/In2O3 heterojunction. To sum up, a sensor based on 3 wt% g-C3N4/In2O3 exhibits preeminent performance for H2 with high sensitivity, fast response, and excellent selectivity.


Introduction
Hydrogen (H 2 ) is a potential clean energy that has rich application prospects in automobile, aerospace, and other fields [1]. However, due to its flammable and explosive characteristics, H 2 can very easily cause disaster during utilization and storage [2]. Therefore, an effective survey of H 2 has become an important problem to be solved. At present, detection methods of H 2 include gas chromatography, mass spectrometry, gas sensors, and so on [3]. In the above methods, gas sensors have become the best choice, with the advantages of low cost, simple preparation, and small size [4]. Furthermore, the type of gas sensor include semiconductor oxide [5], catalytic combustion [6], solid electrolyte [7,8], etc. Among them, a semiconductor oxide sensor is widely used in people's lives because of the outstanding performance [9]. So far, many semiconductor oxides have been researched as sensing-materials in H 2 sensors, such as TiO 2 , ZnO, SnO 2 , NiO, and In 2 O 3 . Some reported H 2 sensors based on different semiconductor oxides are summarized in Table 1.
In 2 O 3 is a common semiconductor with a band gap of 3.6 eV, which is broadly used in gas sensors [10]. Up to now, preparing In 2 O 3 with distinct morphologies has proved to be an available measure to heighten the performance of gas sensors. However, a single In 2 O 3 -based sensor has the defects of low response and poor selectivity in practical applications [11]. Many researchers have made great efforts to enhance the gas-sensing performance, such as doping noble metals [12], fabricating heterojunctions [13], and constructing 2D materials or polymer composites [14,15]. Graphitic carbon nitride (g-C 3 N 4 ) is a typical polymer semiconductor with a band gap of 2.76 eV, which has been used in gas sensors owing to high chemical stability, large surface area, and good catalytic function [16,17]. It is reported that In 2 O 3 -based sensors can elevate gas-sensing performance through the g-C 3 N 4 composite. Liu et al. prepared In 2 O 3 , and g-C 3 N 4 was compounded by way of MOF, and the response value of the sensor achieved 294 to 100 ppm NO x at RT [18]. Sun et al. synthesized g-C 3 N 4 /In 2 O 3 through a calcination annealing process, and the response value of the sensor reached 1405 at 119 • C to 100 ppm formaldehyde [19]. The enhanced performance of the In 2 O 3 sensor can be attributed to the formation of g-C 3 N 4 /In 2 O 3 heterojunction different oxygen species content [20]. In this paper, a litchi-like g-C 3 N 4 /In 2 O 3 composite was successfully prepared by the hydrothermal method. Additionally, the morphology and composition of C 3 N 4 /In 2 O 3 were characterized by XRD, SEM, TEM, and XPS. Sensors based on different amount of g-C 3 N 4 composite were fabricated to investigate their gas-sensing specifics. Among them, the performance of the sensor based on 3 wt% g-C 3 N 4 /In 2 O 3 was significantly improved for H 2 , giving it the merits of a fast response and excellent selectivity. All chemicals are purchased from Aladdin Reagent and are analytical grade without being further purified for use. In a typical process, 147.5 mg of InCl 3 ·4H 2 O was dissolved in 15 mL of deionized water, then 15 mL of glycerol was added and stirred until it was homogeneous. Afterwards, 520 mg of Na 3 C 6 H 5 O 7 ·2H 2 O was added into the above solution and stirred vigorously for 20 min. Finally, 250 µL of NaOH (0.1 M) aqueous solution was added slowly to form a uniform solution. The obtained solution was transferred to an autoclave lined with 50 mL PTFE for a hydrothermal reaction at 190 • C for 16 h. After being cooled to room temperature, the precipitate was centrifuged with deionized water and absolute ethanol for several times. The powder was collected and dried at 80 • C overnight. The dried sample was placed in an Al 2 O 3 boat and calcined in a muffle furnace at 400 • C for 2 h (2 • C/min) to obtain peachcore-like pure In 2 O 3 .

Synthesis of Litchi-like g-C 3 N 4 /In 2 O 3
Firstly, 20 g of urea was added into a lidded crucible and heated in a muffle furnace at 550 • C for 2 h (2 • C/min) to prepare g-C 3 N 4 . Then, the prepared g-C 3 N 4 was dispersed in 15 mL deionized water and sonicated for 2 h to ensure dispersion. Subsequently, 147.5 mg of InCl 3 ·4H 2 O and 15 mL of glycerol was added the above solution in turn and stirred until homogeneous. Thereafter, 520 mg of Na 3 C 6 H 5 O 7 ·2H 2 O was added and stirred vigorously for 20 min. Finally, 250 µL of NaOH (0.1 M) aqueous solution was added to form a uniform solution. The subsequent process was the same as the preparation of pure In 2 O 3 . According to the different contents of g-C 3 N 4 , 1 wt% g-C 3 N 4 /In 2 O 3 , 3 wt% g-C 3 N 4 /In 2 O 3 , and 5 wt% g-C 3 N 4 /In 2 O 3 composites were prepared, respectively.

Material Characterization
The crystal structures of as-samples were determined by an X-ray diffractometer (XRD, Rigaku Miniflex 600 X, Cu Kα1 radiation, λ = 1.5406 Å) operated at 40 kV and 15 mA. The morphology characteristics of samples were characterized by a scanning electron microscope (SEM, PHENOM SCIENTIFIC ProX G5, The Netherlands) and a transmission electron microscope (TEM, FEI Tecnai G2 F30, USA). X-ray photoelectron spectroscopy (XPS, Thermo escalab 250Xi, USA) was used for the chemical composition analysis of samples.

Fabrication and Measurement of Gas Sensors
The device structure of the gas sensor is shown in Figure 1. As-prepared samples were mixed with deionized water to form a uniform slurry and coated on an alumina ceramic tube as the sensing layer. The coated sensing layer was baked under an infrared lamp for 15 min. Then, the ceramic tube was calcined at 300 • C for 1 h (2 • C/min). Finally, the heating wire was passed through the ceramic tube and welded to a six-legged base to make a gas sensor. In addition, the components of the test gas are the target gas and the component gas, wherein the standard value of the target gas is 1% mol/mol and that of the component gas is nitrogen.

Material Characterization
The crystal structures of as-samples were determined by an X-ray diffrac (XRD, Rigaku Miniflex 600 X, Cu Kα1 radiation, λ = 1.5406 Å) operated at 40 kV mA. The morphology characteristics of samples were characterized by a scannin tron microscope (SEM, PHENOM SCIENTIFIC ProX G5, The Netherlands) and mission electron microscope (TEM, FEI Tecnai G2 F30, USA). X-ray photoelectro troscopy (XPS, Thermo escalab 250Xi, USA) was used for the chemical comp analysis of samples.

Fabrication and Measurement of Gas Sensors
The device structure of the gas sensor is shown in Figure 1. As-prepared s were mixed with deionized water to form a uniform slurry and coated on an a ceramic tube as the sensing layer. The coated sensing layer was baked under an i lamp for 15 min. Then, the ceramic tube was calcined at 300 °C for 1 h (2 °C/min). the heating wire was passed through the ceramic tube and welded to a six-legged make a gas sensor. In addition, the components of the test gas are the target gas component gas, wherein the standard value of the target gas is 1% mol/mol and the component gas is nitrogen. The evaluation of gas-sensing performances was carried out in a static test (50% RH, 25 °C), as shown in Figure 2. The heating current of the sensor was prov the DC-regulated power supply, and the resistance value was recorded by the m ter. The response value is defined as R = (Ra − Rg)/Rg × 100%, wherein Ra and Rg resistance value of the sensor respectively exposed to clean air and target gas. Ad ally, the response/recovery time is defined as the time for 90% of the resistance ch  The evaluation of gas-sensing performances was carried out in a static test system (50% RH, 25 • C), as shown in Figure 2. The heating current of the sensor was provided by the DC-regulated power supply, and the resistance value was recorded by the multimeter. The response value is defined as R = (R a − R g )/R g × 100%, wherein R a and R g are the resistance value of the sensor respectively exposed to clean air and target gas. Additionally, the response/recovery time is defined as the time for 90% of the resistance change.

Material Characterization
The crystal structures of as-samples were determined by an X-ray diffrac (XRD, Rigaku Miniflex 600 X, Cu Kα1 radiation, λ = 1.5406 Å) operated at 40 kV mA. The morphology characteristics of samples were characterized by a scann tron microscope (SEM, PHENOM SCIENTIFIC ProX G5, The Netherlands) and mission electron microscope (TEM, FEI Tecnai G2 F30, USA). X-ray photoelectr troscopy (XPS, Thermo escalab 250Xi, USA) was used for the chemical com analysis of samples.

Fabrication and Measurement of Gas Sensors
The device structure of the gas sensor is shown in Figure 1. As-prepared were mixed with deionized water to form a uniform slurry and coated on an ceramic tube as the sensing layer. The coated sensing layer was baked under an lamp for 15 min. Then, the ceramic tube was calcined at 300 °C for 1 h (2 °C/min) the heating wire was passed through the ceramic tube and welded to a six-legged make a gas sensor. In addition, the components of the test gas are the target gas component gas, wherein the standard value of the target gas is 1% mol/mol and the component gas is nitrogen. The evaluation of gas-sensing performances was carried out in a static tes (50% RH, 25 °C), as shown in Figure 2. The heating current of the sensor was prov the DC-regulated power supply, and the resistance value was recorded by the m ter. The response value is defined as R = (Ra − Rg)/Rg × 100%, wherein Ra and R resistance value of the sensor respectively exposed to clean air and target gas. A ally, the response/recovery time is defined as the time for 90% of the resistance ch

Characterization of Material Structure
The crystal phase of the as-prepared samples was obtained by XRD, as shown in Figure 3. The g-C 3 N 4 has two peaks at 2θ angles of 13.36 • and 27.38 • , which are related to the tris-triazine units and aromatic systems, respectively [26]. The diffraction peaks of pure However, the diffraction peaks of g-C 3 N 4 are not obviously observed in the XRD pictures of g-C 3 N 4 /In 2 O 3 composites, probably due to the low content of g-C 3 N 4 [27].

Characterization of Material Structure
The crystal phase of the as-prepared samples was obtained by XRD, Figure 3. The g-C3N4 has two peaks at 2θ angles of 13.36° and 27.38°, which a the tris-triazine units and aromatic systems, respectively [26]. The diffract pure In2O3 at 2θ angles of 22.37°, 30.99°, 32.61°, 45.61°, 50.25°, 57.20° and 58.1 to the crystal planes (012), (104), (110), (024), (116), (214) and (300) of In 22-0366). However, the diffraction peaks of g-C3N4 are not obviously obs XRD pictures of g-C3N4/In2O3 composites, probably due to the low content of The SEM images of pure In2O3 and 3 wt% g-C3N4/In2O3 are shown in Fi In2O3 exhibits a peachcore-like structure with a diameter of about 250 nm in The morphology of 3 wt% g-C3N4/In2O3 possesses a distinctive litchi-like st the diameter of 3 wt% g-C3N4/In2O3 is only 130 nm, as shown in Figure 4c,d. the morphology of In2O3 was changed from peachcore-like to litchi-like, and of 3 wt% g-C3N4/In2O3 is smaller about 100 nm than that of pure In2O3. Moreo of 3 wt% g-C3N4/In2O3 is more uniform than that of pure In2O3. The crysta as-prepared samples were calculated by the Scherrer formula: is the X-ray wavelength of 0.15406 n half-width of the diffraction peak, and is the Bragg diffraction angle. The of pure In2O3, 1 wt% g-C3N4/In2O3, 3 wt% g-C3N4/In2O3, and 3 wt% g-C3N4/In 15 nm, 14.4 nm, 13.9 nm and 12.6 nm, respectively. Based on the above res ferred that the g-C3N4 may inhibit the growth of In2O3 crystal, thus leading logical changes [28]. The SEM images of pure In 2 O 3 and 3 wt% g-C 3 N 4 /In 2 O 3 are shown in Figure 4. Pure In 2 O 3 exhibits a peachcore-like structure with a diameter of about 250 nm in Figure 4a,b. The morphology of 3 wt% g-C 3 N 4 /In 2 O 3 possesses a distinctive litchi-like structure, and the diameter of 3 wt% g-C 3 N 4 /In 2 O 3 is only 130 nm, as shown in Figure 4c,d. Apparently, the morphology of In 2 O 3 was changed from peachcore-like to litchi-like, and the diameter of 3 wt% g-C 3 N 4 /In 2 O 3 is smaller about 100 nm than that of pure In 2 O 3 . Moreover, the size of 3 wt% g-C 3 N 4 /In 2 O 3 is more uniform than that of pure In 2 O 3 . The crystallite sizes of as-prepared samples were calculated by the Scherrer formula: where K is Scherrer constant of 0.9, λ is the X-ray wavelength of 0.15406 nm, β is the half-width of the diffraction peak, and θ is the Bragg diffraction angle. The average size of pure In 2 O 3 , 1 wt% g-C 3 N 4 /In 2 O 3 , 3 wt% g-C 3 N 4 /In 2 O 3 , and 3 wt% g-C 3 N 4 /In 2 O 3 is about 15 nm, 14.4 nm, 13.9 nm and 12.6 nm, respectively. Based on the above results, it is inferred that the g-C 3 N 4 may inhibit the growth of In 2 O 3 crystal, thus leading to morphological changes [28].  The microstructure of 3 wt% g-C3N4/In2O3 was characterized by TEM and HRT as shown in Figure 5. Through TEM in Figure 5a, we can clearly see that the 3 g-C3N4/In2O3 presents a litchi-like structure with uniform size. Furthermore, HRTE Figure 5b is manifested by the different lattice spacings of In2O3, which plainly indic In2O3 with a highly crystalline form. The three lattice spacings are 0.12 nm, 0.18 nm 0.17 nm, which are assigned to the (012), (110) and (104) planes of In2O3, respectivel addition, the element mapping is shown in Figure 5c-h. From element mapping, it ca explicitly noticed that the In, O, C, and N elements are evenly distributed; this ca evidence that the 3 wt% g-C3N4/In2O3 composite was successfully prepared. To analyze the chemical composition of pure In2O3 and 3 wt% g-C3N4/In2O3, the is shown in Figure 6. As shown in Figure 6a, the In 3d of pure In2O3 and 3 The microstructure of 3 wt% g-C 3 N 4 /In 2 O 3 was characterized by TEM and HRTEM, as shown in Figure 5. Through TEM in Figure 5a, we can clearly see that the 3 wt% g-C 3 N 4 /In 2 O 3 presents a litchi-like structure with uniform size. Furthermore, HRTEM in The microstructure of 3 wt% g-C3N4/In2O3 was characterized by TEM and HRT as shown in Figure 5. Through TEM in Figure 5a, we can clearly see that the 3 g-C3N4/In2O3 presents a litchi-like structure with uniform size. Furthermore, HRTEM Figure 5b is manifested by the different lattice spacings of In2O3, which plainly indic In2O3 with a highly crystalline form. The three lattice spacings are 0.12 nm, 0.18 nm 0.17 nm, which are assigned to the (012), (110) and (104) planes of In2O3, respectivel addition, the element mapping is shown in Figure 5c-h. From element mapping, it ca explicitly noticed that the In, O, C, and N elements are evenly distributed; this ca evidence that the 3 wt% g-C3N4/In2O3 composite was successfully prepared. To analyze the chemical composition of pure In2O3 and 3 wt% g-C3N4/In2O3, the is shown in Figure 6. As shown in Figure 6a, the In 3d of pure In2O3 and 3 g-C3N4/In2O3 have two strong peaks at 451.6 eV, 444.1 eV and 451.7 eV, 444.2 eV, w To analyze the chemical composition of pure In 2 O 3 and 3 wt% g-C 3 N 4 /In 2 O 3 , the XPS is shown in Figure 6. As shown in Figure 6a, the In 3d of pure In 2 O 3 and 3 wt% g-C 3 N 4 /In 2 O 3 have two strong peaks at 451.6 eV, 444.1 eV and 451.7 eV, 444.2 eV, which correspond to In 3d 5/2 and In 3d 3/2 , respectively [29]. The O 1s peak spectrums in Figure 6b are decomposed into three fitting peaks around 532.1 eV, 531.2 eV, 529.5 eV and 532.2 eV, 531.2 eV, 529.6 eV, wherein the fitting peaks are assigned to the hydroxyl (OH) or chemisorbed oxygen (O C ), oxygen vacancy (O V ), and lattice oxygen (O L ) [30,31]. Furthermore, the O C , O V , and O L content of pure In 2 O 3 and 3 wt% g-C 3 N 4 /In 2 O 3 are about 8.56%, 23.59%, 67.85% and 9.14%, 33.42%, 57.44%, respectively. It is worth noting that the content of O C and O V increases with the introduction of g-C 3 N 4 . It may be one of the reasons why the gas-sensing performance of 3wt% g-C 3 N 4 /In 2 O 3 -based sensors has improved [32,33]. The C 1s spectrum displays three peaks at 288.3 eV, 286.3 eV, and 284.8 eV, as shown in Figure 6c. These three peaks of C 1s respectively belong to the sp 2bonded carbon (N-C=N) and the sp 3fitted carbon bond from surface defects of g-C 3 N 4 and carbon atoms (C-C) [34]. However, the N 1s peak appears to be a weak peak at 398.8eV, as shown in Figure 6d, which may be caused by the low content of g-C 3 N 4 [35]. s 2023, 22, x FOR PEER REVIEW 531.2 eV, 529.6 eV, wherein the fitting peaks are assigned to the hydroxyl (OH isorbed oxygen (OC), oxygen vacancy (OV), and lattice oxygen (OL) [30,31]. Fur the OC, OV, and OL content of pure In2O3 and 3 wt% g-C3N4/In2O3 are about 8.56% 67.85% and 9.14%, 33.42%, 57.44%, respectively. It is worth noting that the con and OV increases with the introduction of g-C3N4. It may be one of the reason gas-sensing performance of 3wt% g-C3N4/In2O3-based sensors has improved [3 C 1s spectrum displays three peaks at 288.3 eV, 286.3 eV, and 284.8 eV, as sho ure 6c. These three peaks of C 1s respectively belong to the sp 2-bonded carbon and the sp 3-fitted carbon bond from surface defects of g-C3N4 and carbon at [34]. However, the N 1s peak appears to be a weak peak at 398.8eV, as shown 6d, which may be caused by the low content of g-C3N4 [35].

Gas-Sensing Properties
To determine the optimal operating temperature of the sensors, the differe based on g-C3N4/In2O3 composites with different proportions were evaluated fr to 300 °C, as shown in Figure 7. When the operating temperature reaches 275 sponse value is 180% to 100 ppm H2 of 3 wt% g-C3N4/In2O3 sensor in Figure 7a 3.5 times that of pure In2O3 sensor. The operating temperature can be explaine ing to the desorption equilibrium of the gas molecules and the chemical reactio When the working temperature is too low, the gas molecules do not have en energy and kinetic energy to react on the In2O3 surface, so the adsorption capa gas is reduced. However, when the operating temperature is too high, the gas

Gas-Sensing Properties
To determine the optimal operating temperature of the sensors, the different sensors based on g-C 3 N 4 /In 2 O 3 composites with different proportions were evaluated from 225 • C to 300 • C, as shown in Figure 7. When the operating temperature reaches 275 • C, the response value is 180% to 100 ppm H 2 of 3 wt% g-C 3 N 4 /In 2 O 3 sensor in Figure 7a, which is 3.5 times that of pure In 2 O 3 sensor. The operating temperature can be explained according to the desorption equilibrium of the gas molecules and the chemical reaction kinetics. When the working temperature is too low, the gas molecules do not have enough heat energy and kinetic energy to react on the In 2 O 3 surface, so the adsorption capacity of the gas is reduced. However, when the operating temperature is too high, the gas molecules adsorbed on the In 2 O 3 surface will have a high activity and escape before the electron carrier transfer, Sensors 2023, 23, 148 7 of 12 resulting in a reduced response [3,36]. In addition, the response values of different sensors were tested with different concentrations of H 2 (10~1000 ppm) at 275 • C. As exhibited in Figure 7b, the response value of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor is greatly improved compared with other sensors within the whole range of H 2 concentrations. With the increase in the H 2 concentration, the rising trend of the response value curve is gradually stable, which may indicate that the sensor is gradually saturated. The response/recovery time of the different sensors at 275 °C is shown in Figure 8ad. It is not hard to notice that the increase of g-C3N4 content, which may have a positive action in the sensor. Although g-C3N4 is beneficial to the response and recovery characteristics of the sensor, excessive g-C3N4 may play a reverse role. Furthermore, the 3 wt% g-C3N4/In2O3 sensor exhibits fast response/recovery time (2 s/2.4 s) to 100 ppm H2. From Figure 8e, even if it is in a high concentration H2 atmosphere, the 3 wt% g-C3N4/In2O3 sensor also demonstrated extremely fast response/recovery time. The above results reveal that the 3 wt% g-C3N4/In2O3 sensor has great potential in practical application. The Ra of different sensors at different temperatures are shown in Figure 8f; the Ra decrease with the increase in temperature. In addition, the Ra of sensors based on g-C3N4/In2O3 is lower than that of the pure In2O3 and decreased with the increasing amount of g-C3N4 in the composites. The reason for the increase in conductivity is firstly due to the increase in oxygen vacancy in g-C3N4/In2O3. The second reason is attributed to the C-N bond breaking, which releases a lot of electrons [37]. Based on the above reasons, it can be proven that the sensor based on g-C3N4/In2O3 has high conductivity. The response/recovery time of the different sensors at 275 • C is shown in Figure 8a-d. It is not hard to notice that the increase of g-C 3 N 4 content, which may have a positive action in the sensor. Although g-C 3 N 4 is beneficial to the response and recovery characteristics of the sensor, excessive g-C 3 N 4 may play a reverse role. Furthermore, the 3 wt% g-C 3 N 4 /In 2 O 3 sensor exhibits fast response/recovery time (2 s/2.4 s) to 100 ppm H 2 . From Figure 8e, even if it is in a high concentration H 2 atmosphere, the 3 wt% g-C 3 N 4 /In 2 O 3 sensor also demonstrated extremely fast response/recovery time. The above results reveal that the 3 wt% g-C 3 N 4 /In 2 O 3 sensor has great potential in practical application. The R a of different sensors at different temperatures are shown in Figure 8f; the R a decrease with the increase in temperature. In addition, the R a of sensors based on g-C 3 N 4 /In 2 O 3 is lower than that of the pure In 2 O 3 and decreased with the increasing amount of g-C 3 N 4 in the composites. The reason for the increase in conductivity is firstly due to the increase in oxygen vacancy in g-C 3 N 4 /In 2 O 3 . The second reason is attributed to the C-N bond breaking, which releases a lot of electrons [37]. Based on the above reasons, it can be proven that the sensor based on g-C 3 N 4 /In 2 O 3 has high conductivity.
Stability and reproducibility are important indicators of a sensor's performance, and the evaluation result of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor at 275 • C is shown in Figure 9. As shown in Figure 9a, the dynamic curves of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor is exhibited. Within the variation range of H 2 concentrations, the response and recovery characteristics of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor are stable. Meanwhile, the detection limit was tested to 1 ppm H 2 , and the response value is 10%. Reproducibility was evaluated by continuous exposure to 50 ppm H 2 for five cycles, as shown in Figure 9b. In the five-cycle experiment, the response/recovery time and response value were basically stable without significant change. By assessing the gas-sensing properties of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor, it was sufficienly proven that the sensor has fine stability and reproducibility. Stability and reproducibility are important indicators of a sensor's performance, the evaluation result of the 3 wt% g-C3N4/In2O3 sensor at 275 °C is shown in Figure 9 shown in Figure 9a, the dynamic curves of the 3 wt% g-C3N4/In2O3 sensor is exhib Within the variation range of H2 concentrations, the response and recovery characteri of the 3 wt% g-C3N4/In2O3 sensor are stable. Meanwhile, the detection limit was teste 1 ppm H2, and the response value is 10%. Reproducibility was evaluated by continu exposure to 50 ppm H2 for five cycles, as shown in Figure 9b. In the five-cycle experim the response/recovery time and response value were basically stable without signifi change. By assessing the gas-sensing properties of the 3 wt% g-C3N4/In2O3 sensor, it sufficienly proven that the sensor has fine stability and reproducibility.  Stability and reproducibility are important indicators of a sensor's performance, and the evaluation result of the 3 wt% g-C3N4/In2O3 sensor at 275 °C is shown in Figure 9. As shown in Figure 9a, the dynamic curves of the 3 wt% g-C3N4/In2O3 sensor is exhibited. Within the variation range of H2 concentrations, the response and recovery characteristics of the 3 wt% g-C3N4/In2O3 sensor are stable. Meanwhile, the detection limit was tested to 1 ppm H2, and the response value is 10%. Reproducibility was evaluated by continuous exposure to 50 ppm H2 for five cycles, as shown in Figure 9b. In the five-cycle experiment, the response/recovery time and response value were basically stable without significant change. By assessing the gas-sensing properties of the 3 wt% g-C3N4/In2O3 sensor, it was sufficienly proven that the sensor has fine stability and reproducibility. Selectivity is a momentous parameter for the practical application of the sensor. Various gases were tested at 275 • C for the appraisement of different sensors, as presented in Figure 10. Among them, 1 wt% g-C 3 N 4 /In 2 O 3 and 3 wt% g-C 3 N 4 /In 2 O 3 sensors exhibit similar selectivity characteristics, which possess a higher response value to 50 ppm H 2 than other gases (50 ppm SO 2 , 50 ppm NH 3 , 50 ppm CO 2 , 50 ppm CO, 50 ppm CH 4 , 100 ppm NO 2 ). Besides, the response value of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor is three times that of the 1 wt% g-C 3 N 4 /In 2 O 3 sensor to 50 ppm H 2 . Meanwhile, the response value of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor to 50 ppm H 2 is 7.5 times that of 50 ppm SO 2 (the secondary response gas). As we all know, CO, CH 4 , and H 2 are common fuel gases, and the response values of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor to H 2 are 20 times and 40 times that of CO and CH 4 of 50 ppm, respectively. The improvement of H 2 selectivity is due to the better dispersion of litchi-like 3wt% g-C 3 N 4 /In 2 O 3 compared with pure In 2 O 3 , which makes H 2 with the smallest molecular size easy to diffuse [38]. In addition, at 275 • C for the sensor based on the 3 wt% g-C 3 N 4 /In 2 O 3 surface, the adsorption energy of H 2 gas is much higher than that of other gases [22]. From the above results, the 3 wt% g-C 3 N 4 /In 2 O 3 sensor exhibits excellent selectivity in H 2 detection.
Selectivity is a momentous parameter for the practical application of the sensor Various gases were tested at 275 °C for the appraisement of different sensors, as pre sented in Figure 10. Among them, 1 wt% g-C3N4/In2O3 and 3 wt% g-C3N4/In2O3 sensor exhibit similar selectivity characteristics, which possess a higher response value to 5 ppm H2 than other gases (50 ppm SO2, 50 ppm NH3, 50 ppm CO2, 50 ppm CO, 50 ppm CH4, 100 ppm NO2). Besides, the response value of the 3 wt% g-C3N4/In2O3 sensor i three times that of the 1 wt% g-C3N4/In2O3 sensor to 50 ppm H2. Meanwhile, the respons value of the 3 wt% g-C3N4/In2O3 sensor to 50 ppm H2 is 7.5 times that of 50 ppm SO2 (th secondary response gas). As we all know, CO, CH4, and H2 are common fuel gases, and the response values of the 3 wt% g-C3N4/In2O3 sensor to H2 are 20 times and 40 times tha of CO and CH4 of 50 ppm, respectively. The improvement of H2 selectivity is due to th better dispersion of litchi-like 3wt% g-C3N4/In2O3 compared with pure In2O3, which makes H2 with the smallest molecular size easy to diffuse [38]. In addition, at 275 °C fo the sensor based on the 3 wt% g-C3N4/In2O3 surface, the adsorption energy of H2 gas i much higher than that of other gases [22]. From the above results, the 3 wt% g-C3N4/In2O sensor exhibits excellent selectivity in H2 detection. The long-term stability is shown in Figure 11, which was evaluated by continuou exposure to 50 ppm H2 at 275 °C within 2 weeks. It is distinctly perceived from Figure 11a that the response value to 50 ppm H2 is not significantly changed. Moreover, the transien curves for different days are shown in Figure 11b-d, and it is noteworthy that the air re sistance (Ra) and response features of the sensor are basically stable. In view of the abov assessment, it can be assured that the 3 wt% g-C3N4/In2O3 sensor has good long-term stability for H2 detection. The long-term stability is shown in Figure 11, which was evaluated by continuous exposure to 50 ppm H 2 at 275 • C within 2 weeks. It is distinctly perceived from Figure 11a that the response value to 50 ppm H 2 is not significantly changed. Moreover, the transient curves for different days are shown in Figure 11b-d, and it is noteworthy that the air resistance (R a ) and response features of the sensor are basically stable. In view of the above assessment, it can be assured that the 3 wt% g-C 3 N 4 /In 2 O 3 sensor has good long-term stability for H 2 detection.

Gas Sensing Mechanism
In order to understand the enhanced properties of the H2 sensor based on g-C3N4/In2O3, the gas-sensing mechanism of the In2O3 sensor was analyzed. In air at different temperatures, the oxygen molecules on the surface of In2O3 materials will become O2 -(<147 °C), O − (147 °C-397 °C), O 2-(>397 °C), as follows in Equations (2)-(5) [29].

Gas Sensing Mechanism
In order to understand the enhanced properties of the H 2 sensor based on g-C 3 N 4 /In 2 O 3 , the gas-sensing mechanism of the In 2 O 3 sensor was analyzed. In air at different temperatures, the oxygen molecules on the surface of In 2 O 3 materials will become O 2 In this paper, the optimum operating temperature was determined to be 275 • C, so most of the oxygen molecules are converted into O − . When the In 2 O 3 sensor exposure to H 2 , the H molecules are oxidized by O − to form H 2 O and free electrons as in Equation (6) [22]. The reaction process is shown in Figure 12a. The elevated gas-sensing performance of sensor based on 3 wt% g-C 3 N 4 /In 2 O 3 may be attributed to the morphology characteristic, the oxygen state and the g-C 3 N 4 /In 2 O 3 heterojunctions. The first reason is the distinctive litchi-like morphology of 3 wt% g-C 3 N 4 /In 2 O 3 . Through observing SEM and TEM photos, the 3 wt% g-C 3 N 4 /In 2 O 3 has smaller size and better dispersion than pure In 2 O 3 . The performance of the sensor based on 3 wt% g-C 3 N 4 /In 2 O 3 may be improved due to the unique morphology configuration. The second reason can be attributed to the adjustment of the oxygen state. T results show that the OV and OC contents in 3 wt% g-C3N4/In2O3 are higher than In2O3. For the sensor based on 3 wt% g-C3N4/In2O3, a large amount of OV provid active sites for the adsorption of active oxygen, and the increase in OC content i that more chemically adsorbed oxygen participates in the redox reaction [19]. Th the performance of the 3 wt% g-C3N4/In2O3 sensor may be improved. The last r the formation of the g-C3N4/In2O3 heterojunction [19]. The work functions and ba of g-C3N4 and In2O3 are W = 4.3 [39], Eg = 2.76 eV [16] and W = 5.0 [13], Eg = 3.6 respectively, as exhibited in Figure 12b. The electrons flow from g-C3N4 to In2O new equilibrium of the Fermi level. Therefore, the sensor based on 3 wt% g-C3N shows excellent response characteristics [40]. The second reason can be attributed to the adjustment of the oxygen state. The XPS results show that the O V and O C contents in 3 wt% g-C 3 N 4 /In 2 O 3 are higher than pure in In 2 O 3 . For the sensor based on 3 wt% g-C 3 N 4 /In 2 O 3 , a large amount of O V provides more active sites for the adsorption of active oxygen, and the increase in O C content indicates that more chemically adsorbed oxygen participates in the redox reaction [19]. Therefore the performance of the 3 wt% g-C 3 N 4 /In 2 O 3 sensor may be improved. The last reason is the formation of the g-C 3 N 4 /In 2 O 3 heterojunction [19]. The work functions and band gaps of g-C 3 N 4 and In 2 O 3 are W = 4.3 [39], E g = 2.76 eV [16] and W = 5.0 [13], E g = 3.6 eV [10], respectively, as exhibited in Figure 12b. The electrons flow from g-C 3 N 4 to In 2 O 3 to the new equilibrium of the Fermi level. Therefore, the sensor based on 3 wt% g-C 3 N 4 /In 2 O 3 shows excellent response characteristics [40].