Low-Operating-Temperature NO2 Sensor Based on a CeO2/ZnO Heterojunction

CeO2/ZnO-heterojunction-nanorod-array-based chemiresistive sensors were studied for their low-operating-temperature and gas-detecting characteristics. Arrays of CeO2/ZnO heterojunction nanorods were synthesized using anodic electrodeposition coating followed by hydrothermal treatment. The sensor based on this CeO2/ZnO heterojunction demonstrated a much higher sensitivity to NO2 at a low operating temperature (120 °C) than the pure-ZnO-based sensor. Moreover, even at room temperature (RT, 25 °C) the CeO2/ZnO-heterojunction-based sensor responds linearly and rapidly to NO2. This sensor’s reaction to interfering gases was substantially less than that of NO2, suggesting exceptional selectivity. Experimental results revealed that the enhanced gas-sensing performance at the low operating temperature of the CeO2/ZnO heterojunction due to the built-in field formed after the construction of heterojunctions provides additional carriers for ZnO. Thanks to more carriers in the ZnO conduction band, more oxygen and target gases can be adsorbed. This explains the enhanced gas sensitivity of the CeO2/ZnO heterojunction at low operating temperatures.


Introduction
Nitrogen dioxide (NO 2 ) is a harmful gas that threatens human survival [1]. Vehicle exhaust fumes and boiler exhaust emissions are among the principal sources of man-made NO 2 . NO 2 is a brownish-red, highly reactive gaseous substance that is very harmful to the human body. Even after only a short exposure to nitrogen dioxide, lung function can be impaired [2]. If exposed for a long time, the chance of respiratory infections increases and can lead to permanent organic lesions in the lungs [1,3]. Furthermore, NO 2 is harmful to the environment and can pollute water, soil, and the atmosphere. Therefore, the rapid and accurate detection of NO 2 is critical for human health and environmental protection; research on nitrogen dioxide sensors is very important.
In recent research on gas sensors, there is no doubt that sensors based on metal oxide semiconductors (MOSs), graphene [4], polymer nanofibers [5], metal organic frameworks (MOFs) [6], and molecularly imprinted polymers [7] have received the most attention. Compared to other solution strategies, MOS-based sensors provide a cost-effective solution for the rapid deployment of gas detection due to their low power consumption, simplicity of preparation, and ease of integration into electronic devices. Numerous MOS-based chemiresistive gas sensors (ZnO [8][9][10], WO 3 [11,12], SnO 2 [13,14], TiO 2 [15], etc.) have been proven to be used for efficient gas sensing. As a wide bandgap semiconductorsensitive material, ZnO is a critical component of contemporary gas sensor research due to its cheap cost, high sensitivity, simplicity of manufacture, and miniaturization. However, the disadvantages of MOS-based gas-sensing materials include excessively high operating temperatures and low selectivity, which limit their use in practical engineering applications.

Preparation of ZnO and CeO2/ZnO
Preparation of the ZnO seed layer on substrates: Typically, an appropriate amount of Zn(CH3COO)2·2H2O was added to ethanol and vigorously stirred for 20 min as a suspension; in order to prepare the ZnO seed layer on the substrate, a 5.4 mg/mL suspension was spin-coated on Al2O3 substrates patterned with Au interdigital electrodes (IDEs). Then, drying treatment was carried out at 200 °C for 20 min in a thermostat drier to stabilize the seed layer.
The synthesis process of the pure ZnO nanorods array: The pure ZnO nanorods array was fabricated by a hydrothermal reaction. In 100 mL of deionized water, dispersions of Zn(NO3)2·6H2O (3.56 g) and hexamethylenetetramine (HMTA) (1.67 g) were prepared to provide homogenous solutions A and B, respectively. Then, solution A was gently swirled into solution B. The solution was then transferred to a 50 mL Teflon-lined autoclave vessel with a prepared substrate and maintained at 95 °C for 12 h. Finally, samples were annealed in the air atmosphere for 2 h at a heating rate of 5 °C min −1 and an annealing temperature of 400 °C.
CeO2 is made by grinding cerium dioxide powder. The preparation method of CeO2/ZnO is similar to that of the ZnO nanorods array, except with additional anodic electrodeposition coating steps on the ZnO nanorods array in a two-electrode setup at room temperature. IDEs with the ZnO nanorods array were used as the anode, with Pt foil as the cathode. Before anodization, IDEs with the ZnO nanorods array were ultrasonically cleaned in acetone, ethanol, and deionized water, respectively. CeO2 was electrochemically grown on a substrate in a 2 M Ce(NO3)3·6H2O solution at a current density of 20 mA cm −1 for 1 min, 2 min, 3 min, and 5 min, respectively. Additionally, the corresponding samples were referred to as CeO2/ZnO-1, CeO2/ZnO-2, CeO2/ZnO-3, and CeO2/ZnO-4, respectively. Finally, the samples were annealed in the air atmosphere for 2 h at a heating rate of 5 °C min −1 and an annealing temperature of 400 °C.

Characterization
X-ray diffraction (XRD; Rigaku SmartLab, Osaka, Japan) patterns were used to analyze the phase and crystal structure using a Rigaku SmartLab system with Cu K incident radiation (λ = 1.54056 Å, 20°-80°). Field emission scanning electron microscopy (FESEM; Thermo Scientific Various G4 UC, Brno, Czech Republic), transmission electron microscopy (TEM; Thermo Scientific Talos F200X G2 operated at 200 kV, Brno, Czech Republic), The synthesis process of the pure ZnO nanorods array: The pure ZnO nanorods array was fabricated by a hydrothermal reaction. In 100 mL of deionized water, dispersions of Zn(NO 3 ) 2 ·6H 2 O (3.56 g) and hexamethylenetetramine (HMTA) (1.67 g) were prepared to provide homogenous solutions A and B, respectively. Then, solution A was gently swirled into solution B. The solution was then transferred to a 50 mL Teflon-lined autoclave vessel with a prepared substrate and maintained at 95 • C for 12 h. Finally, samples were annealed in the air atmosphere for 2 h at a heating rate of 5 • C min −1 and an annealing temperature of 400 • C. CeO 2 is made by grinding cerium dioxide powder. The preparation method of CeO 2 /ZnO is similar to that of the ZnO nanorods array, except with additional anodic electrodeposition coating steps on the ZnO nanorods array in a two-electrode setup at room temperature. IDEs with the ZnO nanorods array were used as the anode, with Pt foil as the cathode. Before anodization, IDEs with the ZnO nanorods array were ultrasonically cleaned in acetone, ethanol, and deionized water, respectively. CeO 2 was electrochemically grown on a substrate in a 2 M Ce(NO 3 ) 3 ·6H 2 O solution at a current density of 20 mA cm −1 for 1 min, 2 min, 3 min, and 5 min, respectively. Additionally, the corresponding samples were referred to as CeO 2 /ZnO-1, CeO 2 /ZnO-2, CeO 2 /ZnO-3, and CeO 2 /ZnO-4, respectively. Finally, the samples were annealed in the air atmosphere for 2 h at a heating rate of 5 • C min −1 and an annealing temperature of 400 • C.

Characterization
X-ray diffraction (XRD; Rigaku SmartLab, Osaka, Japan) patterns were used to analyze the phase and crystal structure using a Rigaku SmartLab system with Cu K incident radiation (λ = 1.54056 Å, 20 • -80 • ). Field emission scanning electron microscopy (FESEM; Thermo Scientific Various G4 UC, Brno, Czech Republic), transmission electron microscopy (TEM; Thermo Scientific Talos F200X G2 operated at 200 kV, Brno, Czech Republic), and highresolution transmission electron microscopy (HRTEM; 200 kV) were used to examine the sample morphologies. TEM attachments were also used to measure the energy-dispersive X-ray spectroscopy (EDS) analysis. An X-ray photoelectron spectrometer was used to analyze the surface chemical elements (XPS; KRATOS Axis Supra, Kyoto, Japan). A UV-Vis spectrophotometer was used to obtain UV-visible diffuse reflectance spectra (UV-Vis; TU-1901, Beijing, China). The electrical signals of the sensors were tested by using a digital source meter (Keithley 2450, Beaverton, OR, USA). Detailed gas-sensitive test methods are available in the Supporting Information.

Morphological and Structural Characteristics
The X-ray diffraction (XRD) patterns of synthesized pure ZnO, pure CeO 2 , and CeO 2 /ZnO-2 nanomaterials are presented in Figure 2a. The typical wurtzite hexagonal peak type of ZnO can be seen from it, and no phase transition from anatase to rutile is observed. Which have corresponded with standard PDF card (JCPDS #79-2205) [23], and could observe the pattern exhibits typical diffraction peaks at 2θ = 31.79 • , 34.44 • , 36.28 • , 47.57 • , 56.64 • , 62.90 • , and 68.00 • in all CeO 2 /ZnO composites' XRD patterns, respectively ( Figure 2b) [24,25]. In Figure 2b, all specimens showed the ZnO and CeO 2 phases, and no other phases were found except for the effect of the substrate. Even at higher Ce loading concentrations, no significant peak shifts were noticed, suggesting that perhaps the Ce ions were not incorporated into the ZnO lattice after a two-hour heat treatment at 400 • C. Zn 2+ has an ionic radius of 0.74, which is comparable to Ce 4+ (0.87) but much less than Ce 3+ (1.01). As a result, Ce 4+ substitution for Zn 2+ is possible but not observed in our studies, most likely due to the heat treatment temperature (400 • C) being too low for a solid-state reaction to occur [26,27]. A more pronounced peak of CeO 2 is observed at 28.5 • as the loading concentration of Ce gradually increases ( Figure 2c). The absence of CeO 2 peaks in the samples with less Ce content indicates that the less crystalline CeO 2 nanoparticles were uniform in size and did not form clusters or granulate [28].
used to analyze the surface chemical elements (XPS; KRATOS Axis Supra, Kyoto, Japan). A UV-Vis spectrophotometer was used to obtain UV-visible diffuse reflectance spectra (UV-Vis; TU-1901, Beijing, China). The electrical signals of the sensors were tested by using a digital source meter (Keithley 2450, Beaverton, OR, USA). Detailed gas-sensitive test methods are available in the supporting information.

Morphological and Structural Characteristics
The X-ray diffraction (XRD) patterns of synthesized pure ZnO, pure CeO2, and CeO2/ZnO-2 nanomaterials are presented in Figure 2a. The typical wurtzite hexagonal peak type of ZnO can be seen from it, and no phase transition from anatase to rutile is observed. Which have corresponded with standard PDF card (JCPDS #79-2205) [23], and could observe the pattern exhibits typical diffraction peaks at 2θ = 31.79°, 34.44°, 36.28°, 47.57°, 56.64°, 62.90°, and 68.00° in all CeO2/ZnO composites' XRD patterns, respectively ( Figure 2b) [24,25]. In Figure 2b, all specimens showed the ZnO and CeO2 phases, and no other phases were found except for the effect of the substrate. Even at higher Ce loading concentrations, no significant peak shifts were noticed, suggesting that perhaps the Ce ions were not incorporated into the ZnO lattice after a two-hour heat treatment at 400 °C. Zn 2+ has an ionic radius of 0.74, which is comparable to Ce 4+ (0.87) but much less than Ce 3+ (1.01). As a result, Ce 4+ substitution for Zn 2+ is possible but not observed in our studies, most likely due to the heat treatment temperature (400 °C) being too low for a solid-state reaction to occur [26,27]. A more pronounced peak of CeO2 is observed at 28.5° as the loading concentration of Ce gradually increases ( Figure 2c). The absence of CeO2 peaks in the samples with less Ce content indicates that the less crystalline CeO2 nanoparticles were uniform in size and did not form clusters or granulate [28].  In Figure 2d,e, the UV-Vis spectrum pattern of pure ZnO, CeO 2 , and CeO 2 /ZnO-2 is shown. The results show that CeO 2 /ZnO-2 has good absorption of light in the UV wavelength range. According to previous studies, both CeO 2 and ZnO are direct bandgap semiconductors, which means that electrons from the valence band in both materials can jump directly to the conduction band [29,30]; the Kubelka-Munk equation was used to compute the bandgap (E g ) of these materials [31]. The (F(R)hv) 1/2 -hv curve and accompanying tangent line were determined using the UV-Vis analysis data, with the intersection of the tangent line and the x-axis being the sought bandgap. Based on the results of the calculations it is clear that the E g for ZnO and CeO 2 is 3.12 and 2.73 eV, respectively. The CeO 2 bandgap energy value was lower than that reported for bulk cerium oxide (3.15-3.2 eV) [32]. As a result, the CeO 2 /ZnO-2 bandgap was measured to be 3.02 eV, which was somewhat lower than that of as-prepared ZnO (3.12 eV). This finding is significant because it suggests that the CeO 2 /ZnO heterojunction can give more e − , which could lead to improved gas-sensing capability.
The shape of the ZnO nanorods arrays is shown to be affected by the addition of Ce by the FESEM. Figure 3a shows that the pure ZnO nanorod arrays grow uniformly and are randomly oriented in all regions. The nanorods contact each other where they cross, as can be seen. This provides a path for the electrical signal to travel between the nanorods. The ZnO nanorod arrays have a flat surface and the nanorod diameters range from 50 to 100 nm ( Figure 3a). Overall, the morphologies of all the CeO 2 /ZnO nanorods arrays were similar (Figure 3b). When the ZnO nanorods are magnified 100,000 times, little nanosheet structures can be clearly observed on their surface (Figure 3b). An EDS spectrum shows that the ZnO nanorods have a Ce-rich composition on their outer layer based on the distribution of these three elements (O, Zn, and Ce) (Figure 3c-f).
directly to the conduction band [29,30]; the Kubelka-Munk equation was used to compute the bandgap (Eg) of these materials [31]. The (F(R)hv) 1/2 -hv curve and accompanying tangent line were determined using the UV-Vis analysis data, with the intersection of the tangent line and the x-axis being the sought bandgap. Based on the results of the calculations it is clear that the Eg for ZnO and CeO2 is 3.12 and 2.73 eV, respectively. The CeO2 bandgap energy value was lower than that reported for bulk cerium oxide (3.15-3.2 eV) [32]. As a result, the CeO2/ZnO-2 bandgap was measured to be 3.02 eV, which was somewhat lower than that of as-prepared ZnO (3.12 eV). This finding is significant because it suggests that the CeO2/ZnO heterojunction can give more e − , which could lead to improved gas-sensing capability.
The shape of the ZnO nanorods arrays is shown to be affected by the addition of Ce by the FESEM. Figure 3a shows that the pure ZnO nanorod arrays grow uniformly and are randomly oriented in all regions. The nanorods contact each other where they cross, as can be seen. This provides a path for the electrical signal to travel between the nanorods. The ZnO nanorod arrays have a flat surface and the nanorod diameters range from 50 to 100 nm (Figure 3a). Overall, the morphologies of all the CeO2/ZnO nanorods arrays were similar (Figure 3b). When the ZnO nanorods are magnified 100,000 times, little nanosheet structures can be clearly observed on their surface (Figure 3b). An EDS spectrum shows that the ZnO nanorods have a Ce-rich composition on their outer layer based on the distribution of these three elements (O, Zn, and Ce) (Figure 3c-f).  TEM and HRTEM images of CeO 2 /ZnO-2 are shown in Figure 3g,h. Pure ZnO has been discovered to have a nanorod-like nanostructure with distinct boundaries. After ZnO and CeO 2 were compounded no significant changes in nanorod size were observed, and CeO 2 clusters could be observed on the surface of the nanorods. The lattice fringes of ZnO and CeO 2 cross and overlap in the composite material, suggesting that a heterojunction structure is formed. The HRTEM image (Figure 3h) shows interplanar spacings of about 0.243 nm and 0.308 nm, which are close to the (101) plane of ZnO and the (111) plane of CeO 2 .
The prepared materials' elemental composition and chemical state were further investigated by utilizing XPS techniques. The investigated XPS spectrum of the prepared sensing material is shown in Figure 4a, which contains mainly C, O, Ce, and Zn peaks.
Under the same experimental conditions, the appearance of C1s (284.8 eV) validates the analysis. It is shown in Figure 4b that Zn2p 3/2 and Zn2p 1/2 have binding energies of 1021.9 eV and 1045.1 eV, respectively, which are consistent with the values of prepared pure ZnO [33]. Then, we compared the Ce3d energy spectra of CeO 2 /ZnO-2 and pure CeO 2 in Figure 4d. The peak labeled (*) is for the Ce 4+ state and the other peaks labeled (#) are characteristic of the Ce 3+ state, suggesting that the majority of the Ce ions are in the Ce 4+ state [34]. This suggests that in cerium oxides the Ce ion exists in both Ce 3+ and Ce 4+ states and that the corresponding binding energies of these two valence states in the XPS spectra are close. The properties of the Ce 3d final state may be traced to the six peaks at 882.3, 888.7, 898.2, 900.7, 907.6, and 916.6 eV labels generated by three pairs of spinorbit doublets [35]. Furthermore, two distinct peaks with binding energies of 884.7 and 903.2 eV in the CeO 2 /ZnO composites' spectra showed the presence of Ce 3+ surface states in the produced CeO 2 /ZnO composites even though Ce 3+ -containing compounds were not found in the XRD pattern, most likely due to the material's extremely low concentration of Ce 3+ -containing compounds [36]. O1s spectra are shown in Figure 4c, and the peaks at 529.4, 530.2, and 532.2 eV are attributed to lattice oxygen (O L ) and oxygen vacancy (O V ) in ZnO, CeO 2 , and the adsorbed oxygen (O Ads ) on the sensing materials' surface [37,38]. ZnO and CeO2 cross and overlap in the composite material, suggesting that a heteroju tion structure is formed. The HRTEM image (Figure 3h) shows interplanar spacing about 0.243 nm and 0.308 nm, which are close to the (101) plane of ZnO and the (111) pl of CeO2.
The prepared materials' elemental composition and chemical state were further vestigated by utilizing XPS techniques. The investigated XPS spectrum of the prepa sensing material is shown in Figure 4a, which contains mainly C, O, Ce, and Zn pea Under the same experimental conditions, the appearance of C1s (284.8 eV) validates analysis. It is shown in Figure 4b that Zn2p3/2 and Zn2p1/2 have binding energies of 102 eV and 1045.1 eV, respectively, which are consistent with the values of prepared pure Z [33]. Then, we compared the Ce3d energy spectra of CeO2/ZnO-2 and pure CeO2 in Fig  4d. The peak labeled (*) is for the Ce 4+ state and the other peaks labeled (#) are charac istic of the Ce 3+ state, suggesting that the majority of the Ce ions are in the Ce 4+ state [ This suggests that in cerium oxides the Ce ion exists in both Ce 3+ and Ce 4+ states and t the corresponding binding energies of these two valence states in the XPS spectra close. The properties of the Ce 3d final state may be traced to the six peaks at 882.3, 88 898.2, 900.7, 907.6, and 916.6 eV labels generated by three pairs of spin-orbit doublets [ Furthermore, two distinct peaks with binding energies of 884.7 and 903.2 eV in CeO2/ZnO composites' spectra showed the presence of Ce 3+ surface states in the produ CeO2/ZnO composites even though Ce 3+ -containing compounds were not found in XRD pattern, most likely due to the material's extremely low concentration of Ce 3+ -c taining compounds [36]. O1s spectra are shown in Figure 4c, and the peaks at 529.4, 53 and 532.2 eV are attributed to lattice oxygen (OL) and oxygen vacancy (OV) in ZnO, Ce and the adsorbed oxygen (OAds) on the sensing materials' surface [37,38].

Gas-Sensing Properties
For the purpose of comparing the gas-sensing ability of the sensitive materials prepared in this study, we tested pure ZnO, pure CeO 2 , and CeO 2 /ZnO composites prepared by controlled deposition times for 1, 2, 3, and 4 min. The resistance of the pure-CeO 2-based gas sensor was far above the range of our existing equipment and its response performance to the gas could not be measured. In subsequent performance tests, pure-ZnO-nanorodarray-based sensors and a series of CeO 2 /ZnO-composite-based sensors were mainly tested. It is important to note that the relative humidity (RH) of the test environment can greatly affect the performance of the sensors, as water in high-humidity air can significantly affect the adsorption of sensitive materials to the target gas or O 2 , and cause changes in the baseline resistance of the sensor, thus affecting the results [39]. Consequently, testing was done in an environment with generally consistent temperatures (25 ± 2 • C) and relative humidity (30 ± 5% RH).
First, the influence of operating temperature on the sensor's performance was investigated briefly. Due to the limited conditions of the test apparatus, the resistance of the sensitive material under testing needed to be controlled in the range of 0-200 MΩ. However, we found that the resistance of pure CeO 2 was still too high to exceed the test range. Even when the temperature was increased beyond the heater power range (450 • C), the resistance was still too high. Therefore, all performance tests were carried out on pure ZnO and CeO 2 /ZnO only. In Figure 5a, gas sensitivity tests were carried out over the test temperature range of RT-300 • C for 1 ppm NO 2 , and it was found that all CeO 2 /ZnO samples showed an appreciable response at room temperature relative to the pure ZnO optimum operating temperature at 300 • C and reached optimum operation at 120 • C, followed by a gradual decrease in response intensity with increasing temperature. The best response performance at low temperatures of all samples was obtained for CeO 2 /ZnO-2.
gas sensor was far above the range of our existing equipment and its response perfor-mance to the gas could not be measured. In subsequent performance tests, pure-ZnO-nanorod-array-based sensors and a series of CeO2/ZnO-composite-based sensors were mainly tested. It is important to note that the relative humidity (RH) of the test environment can greatly affect the performance of the sensors, as water in high-humidity air can significantly affect the adsorption of sensitive materials to the target gas or O2, and cause changes in the baseline resistance of the sensor, thus affecting the results [39]. Consequently, testing was done in an environment with generally consistent temperatures (25 ± 2 °C) and relative humidity (30 ± 5% RH).
First, the influence of operating temperature on the sensor's performance was investigated briefly. Due to the limited conditions of the test apparatus, the resistance of the sensitive material under testing needed to be controlled in the range of 0-200 MΩ. However, we found that the resistance of pure CeO2 was still too high to exceed the test range. Even when the temperature was increased beyond the heater power range (450 °C), the resistance was still too high. Therefore, all performance tests were carried out on pure ZnO and CeO2/ZnO only. In Figure 5a, gas sensitivity tests were carried out over the test temperature range of RT-300 °C for 1 ppm NO2, and it was found that all CeO2/ZnO samples showed an appreciable response at room temperature relative to the pure ZnO optimum operating temperature at 300 °C and reached optimum operation at 120 °C, followed by a gradual decrease in response intensity with increasing temperature. The best response performance at low temperatures of all samples was obtained for CeO2/ZnO-2.  Only when the sensor has good selectivity to the target gas can it be used to distinguish the gas to be measured. Therefore, the gas selectivity of the device is a key indicator for measuring the gas sensitivity of the sensor. Figure 5b shows the results of the selectivity tests of the prepared sensor for different gases. Sensors based on CeO 2 /ZnO-2 composite material and pure ZnO were used to test six different gases under the same test conditions. The gases tested were NO 2 , ethanol, acetone, methanol, formaldehyde, and CO 2 . The concentration of NO 2 gas was 1ppm, and the concentration of other gases was 10 ppm. It can be seen that the sensor is very selective and that its response is far better to NO 2 than any other test gas.
The CeO 2 /ZnO-2-based sensor's response to NO 2 concentrations ranging from 1 ppm to 5 ppm at 120 • C and 25 • C is shown in Figure 5c,d, respectively. As the concentration rises so does the sensor's response; even at gas concentrations as low as 1 ppm the CeO 2 /ZnO-2-composite-based sensor has a substantial response. In these two lowtemperature experiments, an excellent linear connection between sensor response and target gas concentration was observed (120 • C: R 2 = 0.986; RT: R 2 = 0.978).
Considering that humans have an odor threshold of 0.5 ppm or less for NO 2 [3], the gas-sensing performance of the CeO 2 /ZnO-2-based sensor to lower concentrations of NO 2 at 120 • C was further investigated. Figure 6a shows the change in sensor resistance with gas concentration from 100 ppb to 900 ppb. Figure 6b shows a linear fit of the response to NO 2 at operating temperatures of 120 • C and 25 • C, respectively. The sensitivity responses of the sensor at 120 • C and RT are shown in Figure 6b.
ditions. The gases tested were NO2, ethanol, acetone, methanol, formaldehyde, and CO2. The concentration of NO2 gas was 1ppm, and the concentration of other gases was 10 ppm. It can be seen that the sensor is very selective and that its response is far better to NO2 than any other test gas.
The CeO2/ZnO-2-based sensor's response to NO2 concentrations ranging from 1 ppm to 5 ppm at 120 °C and 25 °C is shown in Figure 5c,d, respectively. As the concentration rises so does the sensor's response; even at gas concentrations as low as 1 ppm the CeO2/ZnO-2-composite-based sensor has a substantial response. In these two low-temperature experiments, an excellent linear connection between sensor response and target gas concentration was observed (120 °C: R 2 = 0.986; RT: R 2 = 0.978).
Considering that humans have an odor threshold of 0.5 ppm or less for NO2 [3], the gas-sensing performance of the CeO2/ZnO-2-based sensor to lower concentrations of NO2 at 120 °C was further investigated. Figure 6a shows the change in sensor resistance with gas concentration from 100 ppb to 900 ppb. Figure 6b shows a linear fit of the response to NO2 at operating temperatures of 120 °C and 25 °C, respectively. The sensitivity responses of the sensor at 120 °C and RT are shown in Figure 6b. For sensors that need to be used for practical gas detection, response and recovery time are critical factors in determining if such a sensor is abnormal. The response/recovery curves of the CeO2/ZnO-2-based sensor to 1 ppm NO2 at RT (25 °C) and 120 °C are shown in Figure 7a,b. Response and recovery time are generally defined as the time required to achieve 90% of the ultimate steady resistance change [40]. The response and recovery time of the CeO2/ZnO-2-based sensor at RT can be calculated as 24.8 and 79.2 s, respectively, as shown in Figure 7b. The rapid response/recovery exhibited by the prepared sensor at room temperature test conditions suggests that reversible surface reactions can occur throughout the gas-sensing reaction at room temperature. As shown in Figure 7a, the reaction and recovery times for the CeO2/ZnO-2-based sensor at 120 °C can be calculated to be 104 and 417.6 s, respectively. The response/recovery speed is slower than that at room temperature, but the response value has improved substantially. It can be observed that the gas sensitivity of the CeO2/ZnO-2-based sensor is greatly improved at low temperatures compared with pure ZnO and that it can respond quickly and sensitively to NO2 gas For sensors that need to be used for practical gas detection, response and recovery time are critical factors in determining if such a sensor is abnormal. The response/recovery curves of the CeO 2 /ZnO-2-based sensor to 1 ppm NO 2 at RT (25 • C) and 120 • C are shown in Figure 7a,b. Response and recovery time are generally defined as the time required to achieve 90% of the ultimate steady resistance change [40]. The response and recovery time of the CeO 2 /ZnO-2-based sensor at RT can be calculated as 24.8 and 79.2 s, respectively, as shown in Figure 7b. The rapid response/recovery exhibited by the prepared sensor at room temperature test conditions suggests that reversible surface reactions can occur throughout the gas-sensing reaction at room temperature. As shown in Figure 7a, the reaction and recovery times for the CeO 2 /ZnO-2-based sensor at 120 • C can be calculated to be 104 and 417.6 s, respectively. The response/recovery speed is slower than that at room temperature, but the response value has improved substantially. It can be observed that the gas sensitivity of the CeO 2 /ZnO-2-based sensor is greatly improved at low temperatures compared with pure ZnO and that it can respond quickly and sensitively to NO 2 gas at a lower working temperature. In addition, the long-term stability of the gas-sensitive material was tested at 120 • C, which determines the lifetime of the sensor (Figure 7c). The sensor's response changed slightly after long-term stability testing (10 days). These experimental results show that sensors based on CeO 2 /ZnO-2 composites provide the possibility of NO 2 gas detection at low temperatures. The gas-sensitive performance of the sensors was also tested at different relative humidity (RH) levels, with both pure-ZnO-and CeO 2 /ZnO-2-based sensors showing a significant deterioration in gas-sensitive performance as the RH rises at 120 • C (Figure 7d). Overcoming the effect of humidity on gas-sensitive properties is also a problem that needs to be further addressed in subsequent studies.
ity of NO2 gas detection at low temperatures. The gas-sensitive performance of the sensors was also tested at different relative humidity (RH) levels, with both pure-ZnO-and CeO2/ZnO-2-based sensors showing a significant deterioration in gas-sensitive performance as the RH rises at 120 °C (Figure 7d). Overcoming the effect of humidity on gassensitive properties is also a problem that needs to be further addressed in subsequent studies.

Sensing Mechanism
Based on the above experimental results and published studies, we propose the following hypothesis for the gas-sensing mechanism and the gas-sensitive enhancement mechanism in this work.
ZnO is a typical n-type MOS that is widely accepted. When used as a gas-sensitive material, it will absorb oxygen in the air. Due to the difference in electronegativity, the adsorbed oxygen will take away electrons from the ZnO surface and form ionic states (O2 − , O − , and O 2− ) depending on the operating temperature; <150 °C, 150 to 400 °C, and >400 °C

Sensing Mechanism
Based on the above experimental results and published studies, we propose the following hypothesis for the gas-sensing mechanism and the gas-sensitive enhancement mechanism in this work.
ZnO is a typical n-type MOS that is widely accepted. When used as a gas-sensitive material, it will absorb oxygen in the air. Due to the difference in electronegativity, the adsorbed oxygen will take away electrons from the ZnO surface and form ionic states  [17]. The formulas were expressed as follows: A potential barrier (∆ϕ) is formed on the surface of the material when oxygen begins to adsorb to its surface. As a result, the ZnO resistance increases. The ZnO's resistance value will stay constant after the adsorbed oxygen on the material's surface achieves saturation. This stable resistance value is called the baseline resistance. When ZnO is in contact with an oxidizing gas such as NO 2 , the gas molecules will react with oxygen ions and the ZnO surface to obtain electrons, which causes the thickness of the space charge layer on the ZnO surface to increase, ∆ϕ becomes larger, and the surface resistance increases ( Figure  8a). The formulas were expressed as follows: A potential barrier (Δφ) is formed on the surface of the material when oxygen begins to adsorb to its surface. As a result, the ZnO resistance increases. The ZnO's resistance value will stay constant after the adsorbed oxygen on the material's surface achieves saturation. This stable resistance value is called the baseline resistance. When ZnO is in contact with an oxidizing gas such as NO2, the gas molecules will react with oxygen ions and the ZnO surface to obtain electrons, which causes the thickness of the space charge layer on the ZnO surface to increase, Δφ becomes larger, and the surface resistance increases (Figure 8a). The formulas were expressed as follows: (a) (b) Figure 8. (a) The gas-detecting mechanism of the ZnO-based sensors for NO2 gas is shown schematically. (b) Schematic representations of the proposed increased gas-sensing process for CeO2/ZnObased NO2 sensors. EF is the Fermi level; EC and EV are the conduction and valence band edges, respectively (Table S1).
Following the manufacturing of the heterojunction, an internal built-in field from the CeO2 zone to the ZnO zone will develop at the interface due to charge accumulation, as shown in Figure 8b [41,42], while decreasing the likelihood of electrons in the ZnO conduction band transferring to CeO2 owing to the existence of a potential barrier at the interface [43,44]. With the effect of the internal built-in field and the potential barrier, the ZnO side of the heterojunction accumulates more electrons, increasing the carrier concentration significantly. This enables the material to absorb more oxygen ions and target gases without the need for additional excitation conditions. Figure 8. (a) The gas-detecting mechanism of the ZnO-based sensors for NO 2 gas is shown schematically. (b) Schematic representations of the proposed increased gas-sensing process for CeO 2 /ZnObased NO 2 sensors. E F is the Fermi level; E C and E V are the conduction and valence band edges, respectively (Table S1).
Following the manufacturing of the heterojunction, an internal built-in field from the CeO 2 zone to the ZnO zone will develop at the interface due to charge accumulation, as shown in Figure 8b [41,42], while decreasing the likelihood of electrons in the ZnO conduction band transferring to CeO 2 owing to the existence of a potential barrier at the interface [43,44]. With the effect of the internal built-in field and the potential barrier, the ZnO side of the heterojunction accumulates more electrons, increasing the carrier concentration significantly. This enables the material to absorb more oxygen ions and target gases without the need for additional excitation conditions.

Conclusions
In this work, we successfully synthesized a NO 2 gas sensor based on a CeO 2 /ZnO heterojunction by using simple electrodeposition followed by the hydrothermal method. The heterostructure consists of an array of ZnO nanorods and tiny CeO 2 nanocrystals. At low operating temperatures, CeO 2 coupling significantly improves the sensing ability of ZnO for NO 2 . This is mainly due to the large increase in carrier concentration of the sensing material caused by the built-in field formed after the construction of the heterojunction. In addition, we found that compared to other samples, CeO 2 /ZnO-2 heterojunctions are the best for optimizing gas-sensitive properties when used as gas-sensitive materials. In summary, a series of CeO 2 /ZnO n-n type heterostructured gas-sensing materials were synthesized by electrodeposition with low operating temperatures and good stability at low temperatures (RT, 120 • C), leading us to explore them as promising NO 2 -sensitive materials. Author Contributions: K.S.: investigation, formal analysis, data curation, and writing-original draft; G.Z.: investigation, conceptualization, supervision, writing-review and editing, and funding acquisition; H.C.: investigation, formal analysis; S.L.: conceptualization, supervision, writingreview and editing, and funding acquisition. All authors have read and agreed to the published version of the manuscript.