Enhanced High-Temperature (600 °C) NO2 Response of ZnFe2O4 Nanoparticle-Based Exhaust Gas Sensors

Fabrication of gas sensors to monitor toxic exhaust gases at high working temperatures is a challenging task due to the low sensitivity and narrow long-term stability of the devices under harsh conditions. Herein, the fabrication of a chemiresistor-type gas sensor is reported for the detection of NO2 gas at 600 °C. The sensing element consists of ZnFe2O4 nanoparticles prepared via a high-energy ball milling and annealed at different temperatures (600–1000 °C). The effects of annealing temperature on the crystal structure, morphology, and gas sensing properties of ZnFe2O4 nanoparticles are studied. A mixed spinel structure of ZnFe2O4 nanoparticles with a lattice parameter of 8.445 Å is revealed by X-ray diffraction analysis. The crystallite size and X-ray density of ZnFe2O4 nanoparticles increase with the annealing temperature, whereas the lattice parameter and volume are considerably reduced indicating lattice distortion and defects such as oxygen vacancies. ZnFe2O4 nanoparticles annealed at 1000 °C exhibit the highest sensitivity (0.13% ppm–1), sharp response (τres = 195 s), recovery (τrec = 17 s), and linear response to 100–400 ppm NO2 gas. The annealing temperature and oxygen vacancies play a major role in determining the sensitivity of devices. The plausible sensing mechanism is discussed. ZnFe2O4 nanoparticles show great potential for high-temperature exhaust gas sensing applications.


Introduction
Hazardous exhaust gases such as nitrogen dioxide (NO 2 ) and sulfur dioxide (SO 2 ) are the major atmospheric pollutants [1]. The European Union's (E.U.) ambient air quality directives have set the hourly NO 2 concentration threshold as 200 µg/m 3 [2]. According to the European Environment Agency (EEA) report published in 2016, NO 2 pollution was responsible for 71,000 premature deaths in the E.U. [3]. Thus, it is important to detect the emission and subsistence of NO 2 in indoor and outdoor air. The main source of NO 2 pollution is the exhaust emissions as a result of the combustion processes in motor vehicles and manufacturing industries [4]. The direct inspection of the exhaust emissions requires devices that can detect NO 2 at high temperatures, i.e., usually ≥500 • C [5]. In this regard,

Materials and Methods
Iron(III) oxide (Fe 2 O 3 nanopowder) and zinc oxide (ZnO nanopowder) obtained from MilliporeSigma (Merck KGaA, Darmstadt, Germany) were used to prepare ZnFe 2 O 4 nanoparticles. ZnFe 2 O 4 nanoparticles were synthesized by high-energy ball milling (HEBM) process using a SPEX™ 8000M Mixer/Mill™ (SPEX ® SamplePrep, New Jersey, NJ, USA). The ball mill was equipped with a 500-cc stainless steel vessel containing stainless steel balls for mechanical milling of Fe 2 O 3 and ZnO. The mass ratio of steel balls and chemical powders was fixed at 50:1. HEBM was performed under ambient conditions for 2 h at 600 rpm. The product was subsequently vacuum annealed at 600, 800, and 1000 • C for 2 h, and characterized. Corresponding to the annealing temperature (600-1000 • C), the samples were abbreviated as ZnFe 2 O 4 -600, ZnFe 2 O 4 -800, and ZnFe 2 O 4 -1000, respectively.
The crystal structure of the annealed ZnFe 2 O 4 nanoparticles was studied with a STOE STADI P X-ray diffractometer (XRD) (STOE & Cie GmbH, Darmstadt, Germany) using a Cu Kα irradiation source (λ = 1.5406 Å). The samples were scanned in the 2θ range of 10 • -90 • with a scan rate of 2 • /min. The crystallite size (D) is determined by the Scherrer's formula [26] (D = Kλ/Bcosθ), where K is a numerical factor referred to as the crystallite-shape factor with an approximate value of 0.89, λ is the wavelength of the X-rays, B is full-width at half-maximum of the most intense (311) diffraction peak in radians, and θ is the Bragg angle. The experimental lattice parameter (a), X-ray density (ρ xrd ), and the specific surface area (S A ) are also calculated from the XRD data of annealed ZnFe 2 O 4 nanoparticles, as described elsewhere [27,28].
The microstructure and surface morphology of ZnFe 2 O 4 nanoparticles were studied with a JEOL JSM-6510 scanning electron microscope (SEM) (JEOL Ltd., Tokyo, Japan). The elemental composition of ZnFe 2 O 4 nanoparticles was determined with the energy-dispersive X-ray spectroscopy (EDS) (JEOL Ltd., Tokyo, Japan).
Thick-film chemiresistor-type gas sensors were fabricated by mixing an appropriate amount of ZnFe 2 O 4 -600, ZnFe 2 O 4 -800, and ZnFe 2 O 4 -1000 nanoparticles with absolute ethanol to make a thick slurry, which was subsequently drop-coated onto alumina micro-hotplates with vapor-deposited platinum (Pt) contacts. The devices were placed in a vacuum oven at 80 • C for 2 h to dry and stabilize the sensing element. The chemiresistor-type devices were installed in a gas sensing chamber fitted with the electrical connections and the gas inlet and outlet. The measurements were performed with a Keithley 6517A electrometer. The sensor responses were measured simultaneously at 600 • C for 100-400 ppm of NO 2 gas. The sensor response (S) is defined as S(%) = (R g -R a ) × 100/R a , where R a and R g are the resistances in air and (100-400 ppm) NO 2 gas. Figure 1 shows the XRD pattern of as-synthesized ball-milled ZnFe 2 O 4 nanoparticles, referred to as BM-ZnFe 2 O 4 . The HEBM process yields crystalline BM-ZnFe 2 O 4 nanoparticles with a cubic spinel lattice structure as indicated by the presence of characteristic (311) diffraction at 35.22 • (2θ) position. The crystallite size (D) of as-synthesized BM-ZnFe 2 O 4 nanoparticles is 9.30 nm. The lattice parameter (a) is calculated to be 8.445 Å, which is in agreement with the values reported for spinel ZnFe 2 O 4 nanostructures in the literature and the standard value of bulk ZnFe 2 O 4 (a = 8.441 Å) [29][30][31]. The lattice parameter of as-synthesized BM-ZnFe 2 O 4 nanoparticles is slightly higher (~0.05%) than the standard value that may be inherent to the ball-milling process because an increase in the lattice parameter of ball-milled ZnFe 2 O 4 samples has been reported earlier [32][33][34].  Figure 1 shows the XRD pattern of as-synthesized ball-milled ZnFe2O4 nanoparticles, referred to as BM-ZnFe2O4. The HEBM process yields crystalline BM-ZnFe2O4 nanoparticles with a cubic spinel lattice structure as indicated by the presence of characteristic (311) diffraction at 35.22° (2θ) position. The crystallite size (D) of as-synthesized BM-ZnFe2O4 nanoparticles is 9.30 nm. The lattice parameter (a) is calculated to be 8.445 Å, which is in agreement with the values reported for spinel ZnFe2O4 nanostructures in the literature and the standard value of bulk ZnFe2O4 (a = 8.441 Å) [29][30][31]. The lattice parameter of as-synthesized BM-ZnFe2O4 nanoparticles is slightly higher (~0.05%) than the standard value that may be inherent to the ball-milling process because an increase in the lattice parameter of ball-milled ZnFe2O4 samples has been reported earlier [32][33][34]. Theoretically, the cation distribution in a perfect normal spinel ZnFe2O4 unit cell is (Zn 2+ )tet[Fe 3+ ]octO4, i.e., the tetrahedral (A) and octahedral (B) sites are solely occupied by Zn 2+ and Fe 3+ cations, respectively [35]. However, in nanocrystalline ZnFe2O4 the contrary distributions of Zn 2+ and Fe 3+ cations on both A and B sites are observed [36,37], which form mixed (or random) spinel structure. According to Chinnasamy et al. [32], the slight increase in the lattice parameter is attributed to the lattice expansion caused by the occupation of B sites by Zn 2+ ions. Thus, the XRD pattern of assynthesized BM-ZnFe2O4 nanoparticles indicates the formation of a mixed cubic spinel lattice. Nonetheless, BM-ZnFe2O4 nanoparticles are annealed at different temperatures to examine the effect of annealing on the crystal structure evolution, lattice parameter, crystallite size, and morphology of ZnFe2O4 nanoparticles.

Results and Discussion
High-temperature annealing is an important step in the fabrication of ZnFe2O4 nanoparticles, as it renders stability and improves the physical properties of ZnFe2O4 [38,39]. Figure 2 shows the XRD patterns of ZnFe2O4 nanoparticles annealed at 600, 800, and 1000 °C for 2 h. XRD patterns are refined Theoretically, the cation distribution in a perfect normal spinel ZnFe 2 O 4 unit cell is (Zn 2+ ) tet [Fe 3+ ] oct O 4 , i.e., the tetrahedral (A) and octahedral (B) sites are solely occupied by Zn 2+ and Fe 3+ cations, respectively [35]. However, in nanocrystalline ZnFe 2 O 4 the contrary distributions of Zn 2+ and Fe 3+ cations on both A and B sites are observed [36,37], which form mixed (or random) spinel structure. According to Chinnasamy et al. [32], the slight increase in the lattice parameter is attributed to the lattice expansion caused by the occupation of B sites by Zn 2+ ions. Thus, the XRD pattern of as-synthesized BM-ZnFe 2 O 4 nanoparticles indicates the formation of a mixed cubic spinel lattice. Nonetheless, BM-ZnFe 2 O 4 nanoparticles are annealed at different temperatures to examine the effect of annealing on the crystal structure evolution, lattice parameter, crystallite size, and morphology of ZnFe 2 O 4 nanoparticles.
High-temperature annealing is an important step in the fabrication of ZnFe 2 O 4 nanoparticles, as it renders stability and improves the physical properties of ZnFe 2 O 4 [38,39]. Figure 2 shows the XRD patterns of ZnFe 2 O 4 nanoparticles annealed at 600, 800, and 1000 • C for 2 h. XRD patterns are refined using Match! (version 3.11.1.183) and FullProf programs for phase identification from X-ray powder diffraction. All samples exhibit a crystalline structure with the characteristic diffractions corresponding to the following miller indices: (111), (220), (311), (222), (400), (422), (511), (440), (620), and (533), which conform to the crystallography open database card number 230-0615 [40]. XRD results substantiate the formation of the cubic spinel ferrite structure with the Fd-3m space group. Also, the XRD patterns align well with the reported literature for ZnFe 2 O 4 nanoparticles [30,41,42]. The absence of additional diffraction peaks corresponding to the impurities or unreacted oxides also reveals the formation of a single-phase cubic spinel lattice [40].  [40]. XRD results substantiate the formation of the cubic spinel ferrite structure with the Fd-3m space group. Also, the XRD patterns align well with the reported literature for ZnFe2O4 nanoparticles [30,41,42]. The absence of additional diffraction peaks corresponding to the impurities or unreacted oxides also reveals the formation of a single-phase cubic spinel lattice [40]. Figure 3a-c shows the most intense diffractions of the characteristic (311) plane in ZnFe2O4 nanoparticles annealed at different temperatures. The XRD data were used to calculate the crystallite size (D), lattice parameter (a), interplanar distance (d311), volume (V), X-ray density (ρxrd), and specific surface area (S) of the annealed ZnFe2O4 nanoparticles. Table 1 presents these structural parameters for different samples. The effect of annealing is obvious because of the changes in position and breadth of (311) diffraction peak as a function of the annealing temperature, which reveal variations in the crystallite size and lattice parameter. The position of (311) shifts to a higher 2θ value as the annealing temperature increases.   Table 1 presents these structural parameters for different samples. The effect of annealing is obvious because of the changes in position and breadth of (311) diffraction peak as a function of the annealing temperature, which reveal variations in the crystallite size and lattice parameter. The position of (311) shifts to a higher 2θ value as the annealing temperature increases. As shown in Figure 3d, the crystallite size of ZnFe2O4 nanoparticles increases with the annealing time, which is certainly comprehensible because annealing results in grain growth, and the higher the temperature, the greater is the crystallite size [43][44][45]. The annealing at 600 °C doubles the crystallite size of ZnFe2O4@600 nanoparticles compared to as-synthesized BM-ZnFe2O4 nanoparticles. While annealing at 800 °C results in a further increase in the crystallite size, the crystallite sizes of ZnFe2O4@800 and ZnFe2O4@1000 nanoparticles are comparable. Thus, the little difference in the crystallites sizes of samples treated at 800 and 1000 °C means the rate or degree of annealing decreases with the increasing crystallite size [46].  On the other hand, the lattice parameter decreases with the increase in annealing temperature, as shown in Figure 3e. Although this is contrary to the findings reported earlier that demonstrate an increase in the lattice parameter upon high-temperature annealing [47][48][49], the lattice parameter and crystal structure essentially depend on the processing method and conditions. As discussed above, in the starting sample, as-synthesized BM-ZnFe2O4 nanoparticles exhibit a random spinel structure with a certain degree of inversion that is inherently observed for the ball-milled ZnFe2O4 nanoparticles [32,33]. The reduction in the lattice parameter of annealed ZnFe2O4 nanoparticles can be explained by the redistribution of cations and crystal defects (oxygen vacancies).

Sample T (°C) D (nm) a (Å) V (Å 3 ) d311 (Å) ρxrd (g/cm 3 ) S (m 2 /g)
During the high-temperature annealing process, both Zn 2+ and Fe 3+ cations may alter positions that influence the crystal structure. For instance, Lemine et al. [50] demonstrated that a decrease in the lattice parameter (from 8.448 to 8.427 Å) was caused by the redistribution of cations within the interstitial sites. However, this can also be attributed to the crystal defects [39,51]. It is a well-known fact that high-temperature annealing induces lattice defects and distortions. Furthermore, in the case of nanocrystalline ZnFe2O4, it is believed that Zn 2+ ions due to their volatile nature escape from the As shown in Figure 3d, the crystallite size of ZnFe 2 O 4 nanoparticles increases with the annealing time, which is certainly comprehensible because annealing results in grain growth, and the higher the temperature, the greater is the crystallite size [43][44][45]. The annealing at 600 • C doubles the crystallite size of ZnFe 2 O 4 @600 nanoparticles compared to as-synthesized BM-ZnFe 2 O 4 nanoparticles. While annealing at 800 • C results in a further increase in the crystallite size, the crystallite sizes of ZnFe 2 O 4 @800 and ZnFe 2 O 4 @1000 nanoparticles are comparable. Thus, the little difference in the crystallites sizes of samples treated at 800 and 1000 • C means the rate or degree of annealing decreases with the increasing crystallite size [46]. On the other hand, the lattice parameter decreases with the increase in annealing temperature, as shown in Figure 3e. Although this is contrary to the findings reported earlier that demonstrate an increase in the lattice parameter upon high-temperature annealing [47][48][49], the lattice parameter and crystal structure essentially depend on the processing method and conditions. As discussed above, in the starting sample, as-synthesized BM-ZnFe 2 O 4 nanoparticles exhibit a random spinel structure with a certain degree of inversion that is inherently observed for the ball-milled ZnFe 2 O 4 nanoparticles [32,33]. The reduction in the lattice parameter of annealed ZnFe 2 O 4 nanoparticles can be explained by the redistribution of cations and crystal defects (oxygen vacancies).
During the high-temperature annealing process, both Zn 2+ and Fe 3+ cations may alter positions that influence the crystal structure. For instance, Lemine et al. [50] demonstrated that a decrease in the lattice parameter (from 8.448 to 8.427 Å) was caused by the redistribution of cations within the interstitial sites. However, this can also be attributed to the crystal defects [39,51]. It is a well-known fact that high-temperature annealing induces lattice defects and distortions. Furthermore, in the case of nanocrystalline ZnFe 2 O 4 , it is believed that Zn 2+ ions due to their volatile nature escape from the lattice during thermal treatment that successively results in oxygen vacancies [39,52]. Thus, a decrease in the lattice parameter (from 8.445 Å for BM-ZnFe 2 O 4 to 8.420 Å for ZnFe 2 O 4 @1000) is attributed to the cationic redistribution (distortion) and lattice compression caused by escaping Zn 2+ ions and oxygen vacancies.
Consequently, the interplanar distance and the volume of the annealed ZnFe 2 O 4 nanoparticles are reduced as a function of the annealing temperature. On the other hand, the X-ray density increases (from 5.321 g cm -3 for BM-ZnFe 2 O 4 to 5.368 g cm -3 for ZnFe 2 O 4 @1000) with the increase in annealing temperature. However, as shown in Table 1, the specific surface area is reduced to 48 m 2 g -1 due to an increase in the crystallite size of the annealed ZnFe 2 O 4 nanoparticles. These results demonstrate that ZnFe 2 O 4 @1000 and ZnFe 2 O 4 @800 nanoparticles have bigger crystallite size and smaller specific surface area, but the greatest number of defect sites (as oxygen vacancies) and a geometrically frustrated [53] or distorted cubic spinel crystal structure compared to as-synthesized BM-ZnFe 2 O 4 nanoparticles. Figure 4a-c shows the SEM images of ZnFe 2 O 4 nanoparticles annealed at different temperatures. An increase in the annealing temperature (to 1000 • C) results in a more compact surface, as shown in Figure 4c: the micrograph of ZnFe 2 O 4 @1000 nanoparticles. On the other hand, ZnFe 2 O 4 @600 nanoparticles annealed at 600 • C (Figure 4a) show less compact surface morphology with smaller particle size and relatively less aggregation of nanoparticles into clusters. The ZnFe 2 O 4 @800 nanoparticles demonstrate a similar surface morphology with slightly larger aggregates of nanoparticles, as shown in Figure 4b. lattice during thermal treatment that successively results in oxygen vacancies [39,52]. Thus, a decrease in the lattice parameter (from 8.445 Å for BM-ZnFe2O4 to 8.420 Å for ZnFe2O4@1000) is attributed to the cationic redistribution (distortion) and lattice compression caused by escaping Zn 2+ ions and oxygen vacancies. Consequently, the interplanar distance and the volume of the annealed ZnFe2O4 nanoparticles are reduced as a function of the annealing temperature. On the other hand, the X-ray density increases (from 5.321 g cm -3 for BM-ZnFe2O4 to 5.368 g cm -3 for ZnFe2O4@1000) with the increase in annealing temperature. However, as shown in Table 1, the specific surface area is reduced to 48 m 2 g -1 due to an increase in the crystallite size of the annealed ZnFe2O4 nanoparticles. These results demonstrate that ZnFe2O4@1000 and ZnFe2O4@800 nanoparticles have bigger crystallite size and smaller specific surface area, but the greatest number of defect sites (as oxygen vacancies) and a geometrically frustrated [53] or distorted cubic spinel crystal structure compared to as-synthesized BM-ZnFe2O4 nanoparticles. Figure 4a-c shows the SEM images of ZnFe2O4 nanoparticles annealed at different temperatures. An increase in the annealing temperature (to 1000 °C) results in a more compact surface, as shown in Figure 4c: the micrograph of ZnFe2O4@1000 nanoparticles. On the other hand, ZnFe2O4@600 nanoparticles annealed at 600 °C ( Figure 4a) show less compact surface morphology with smaller particle size and relatively less aggregation of nanoparticles into clusters. The ZnFe2O4@800 nanoparticles demonstrate a similar surface morphology with slightly larger aggregates of nanoparticles, as shown in Figure 4b.  The image analysis of the scanning electron micrographs (via WSxM freeware [54]) shows the size distribution of ZnFe 2 O 4 nanoparticles and the respective histograms are presented as insets in Figure 4a-c. ZnFe 2 O 4 @600 nanoparticles exhibit narrow size distribution with an average aggregate size of 100.2 nm, while ZnFe 2 O 4 @800 and ZnFe 2 O 4 @1000 nanoparticles reveal a relatively broad size distribution and an average aggregate size of 143.8 and 146.5 nm, respectively. Figure 5 shows three-dimensional surface micrographs and topographic profiles of the sensing layers composed of ZnFe 2 O 4 nanoparticles annealed at different temperatures. ZnFe 2 O 4 @600 surface exhibits a relatively smooth profile and roughness (Figure 5a). On the other hand, ZnFe 2 O 4 @800 (Figure 5b) and ZnFe 2 O 4 @1000 (Figure 5c) nanoparticles demonstrate higher roughness, greater particle size, and cluster formation. Thus, both X-ray diffraction and microscopic results indicate that ZnFe 2 O 4 nanoparticles annealed at 800 and 1000 • C exhibit bigger crystallite size and a compact surface microstructure compared to those annealed at 600 • C. The image analysis of the scanning electron micrographs (via WSxM freeware [54]) shows the size distribution of ZnFe2O4 nanoparticles and the respective histograms are presented as insets in Figure 4a-c. ZnFe2O4@600 nanoparticles exhibit narrow size distribution with an average aggregate size of 100.2 nm, while ZnFe2O4@800 and ZnFe2O4@1000 nanoparticles reveal a relatively broad size distribution and an average aggregate size of 143.8 and 146.5 nm, respectively. Figure 5 shows three-dimensional surface micrographs and topographic profiles of the sensing layers composed of ZnFe2O4 nanoparticles annealed at different temperatures. ZnFe2O4@600 surface exhibits a relatively smooth profile and roughness (Figure 5a). On the other hand, ZnFe2O4@800 ( Figure 5b) and ZnFe2O4@1000 (Figure 5c) nanoparticles demonstrate higher roughness, greater particle size, and cluster formation. Thus, both X-ray diffraction and microscopic results indicate that ZnFe2O4 nanoparticles annealed at 800 and 1000 °C exhibit bigger crystallite size and a compact surface microstructure compared to those annealed at 600 °C.  Table 2 presents the elemental composition of the annealed ZnFe2O4 nanoparticles. Compared to theoretically calculated values (wt.% or at.%), annealed ZnFe2O4 nanoparticles exhibit variations. As the annealing temperature increases, the relative percentage of Fe increases while the proportions of Zn and O decrease. A decrease in the oxygen content with increasing temperature is attributed to the oxygen vacancies and lattice defects. In addition, the atomic ratio of Fe/Zn is found to be 2.07, 2.20, and 2.21 for ZnFe2O4@600, ZnFe2O4@800, and ZnFe2O4@1000 nanoparticles, respectively. The increase in the Fe/Zn ratio as a function of annealing temperature may be attributed to the volatile nature of Zn 2+ ions [39], as discussed earlier. Thus, the results are consistent and exhibit the microstructure evolution in the annealed ZnFe2O4 nanoparticles as a function of annealing temperature.    As the annealing temperature increases, the relative percentage of Fe increases while the proportions of Zn and O decrease. A decrease in the oxygen content with increasing temperature is attributed to the oxygen vacancies and lattice defects. In addition, the atomic ratio of Fe/Zn is found to be 2.07, 2.20, and 2.21 for ZnFe 2 O 4 @600, ZnFe 2 O 4 @800, and ZnFe 2 O 4 @1000 nanoparticles, respectively. The increase in the Fe/Zn ratio as a function of annealing temperature may be attributed to the volatile nature of Zn 2+ ions [39], as discussed earlier. Thus, the results are consistent and exhibit the microstructure evolution in the annealed ZnFe 2 O 4 nanoparticles as a function of annealing temperature.  Figure 6 demonstrates the NO 2 gas response of ZnFe 2 O 4 nanoparticles annealed at different temperatures. The sensor measurements are performed at 600 • C. The as-synthesized BM-ZnFe 2 O 4 nanoparticles based chemiresistive devices are not stable at 600 • C and do not show a measurable response to NO 2 gas. On the other hand, all the annealed ZnFe 2 O 4 samples show a significant measurable response to 100-400 ppm NO 2 , as shown in Figure 6. The sensor responses are generally saturated after~4 min of exposure to the different concentrations of NO 2 gas. ZnFe 2 O 4 @1000 nanoparticles exhibit the highest NO 2 gas response, which is attributed to their greater stability at elevated temperatures and the presence of a large number of lattice defects. ZnFe 2 O 4 @600 and ZnFe 2 O 4 @800 nanoparticles also exhibit significant gas response at 600 • C. Peng et al. [55] recently demonstrated that the gas sensing properties of ZnFe 2 O 4 nanoparticles could be enhanced by controlling the oxygen vacancies and that ZnFe 2 O 4 nanoparticles with more oxygen vacancies revealed superior gas (acetone vapors) sensing performance at 280 • C. Thus, the higher NO 2 response of ZnFe 2 O 4 @1000 nanoparticles may be attributed to the oxygen vacancies resulting from high-temperature annealing of nanoparticles.  Figure 6. The sensor responses are generally saturated after ~4 min of exposure to the different concentrations of NO2 gas. ZnFe2O4@1000 nanoparticles exhibit the highest NO2 gas response, which is attributed to their greater stability at elevated temperatures and the presence of a large number of lattice defects. ZnFe2O4@600 and ZnFe2O4@800 nanoparticles also exhibit significant gas response at 600 °C. Peng et al. [55] recently demonstrated that the gas sensing properties of ZnFe2O4 nanoparticles could be enhanced by controlling the oxygen vacancies and that ZnFe2O4 nanoparticles with more oxygen vacancies revealed superior gas (acetone vapors) sensing performance at 280 °C. Thus, the higher NO2 response of ZnFe2O4@1000 nanoparticles may be attributed to the oxygen vacancies resulting from hightemperature annealing of nanoparticles.  Figure 7 shows the calibration curves obtained by plotting the maximum gas response of the annealed ZnFe2O4 nanoparticles as a function of gas concentration. All ZnFe2O4 samples exhibit a linear response in the concentration range of 100-400 ppm as demonstrated by the straight lines in Figure 7. The sensitivity of ZnFe2O4-based chemiresistive devices can be calculated from the slope of   Figure 7. The sensitivity of ZnFe 2 O 4 -based chemiresistive devices can be calculated from the slope of a straight line. The sensitivity of NO 2 sensors follows the order: ZnFe 2 O 4 @1000 > ZnFe 2 O 4 @800 > ZnFe 2 O 4 @600, which describes the effect of annealing temperature on sensor performance. An increase in annealing temperature improves the NO 2 sensing properties of ZnFe 2 O 4 nanoparticles. Therefore, ZnFe 2 O 4 @1000 nanoparticles exhibit 2.0-fold and 3.2-fold high sensitivity compared to ZnFe 2 O 4 @800 and ZnFe 2 O 4 @600 nanoparticles, respectively. Nanomaterials 2020, 10, x FOR PEER REVIEW 9 of 14 a straight line. The sensitivity of NO2 sensors follows the order: ZnFe2O4@1000 > ZnFe2O4@800 > ZnFe2O4@600, which describes the effect of annealing temperature on sensor performance. An increase in annealing temperature improves the NO2 sensing properties of ZnFe2O4 nanoparticles. Therefore, ZnFe2O4@1000 nanoparticles exhibit 2.0-fold and 3.2-fold high sensitivity compared to ZnFe2O4@800 and ZnFe2O4@600 nanoparticles, respectively.  Figure 8 shows the kinetics of ZnFe2O4-based chemiresistive gas sensors. The response (τres) and recovery (τrec) times of the annealed ZnFe2O4 nanoparticles are estimated from their response to 300 ppm NO2 gas. All samples show fast response and recovery times. The response times are in the range of 145-195 s and follow the order: ZnFe2O4@800 > ZnFe2O4@600 > ZnFe2O4@1000. Thus, ZnFe2O4@1000 nanoparticles exhibit slightly longer response (τres) times compared to ZnFe2O4@600 and ZnFe2O4@800 nanoparticles. The recovery times are sharp (i.e., ≤20 s) for all samples and all sensors exhibit 100% recovery to their original state. At elevated temperatures, the recovery times are generally faster [10]. Overall, ZnFe2O4@1000 nanoparticles exhibit excellent NO2 gas sensing properties such as high sensitivity, good response kinetics, and linear response in the tested concentration range. The chemiresistive gas sensors function on the principles of changes in resistance of the sensing element when test gas molecules interact with the semiconductor surface [9]. Figure 9 demonstrates the gas sensing mechanism of ZnFe2O4 nanoparticles. ZnFe2O4 is an n-type semiconductor [56]. In principle, when the ZnFe2O4-based chemiresistive device is exposed to air at elevated temperatures, active oxygen species are adsorbed on the surface of ZnFe2O4 nanoparticles. As shown in Figure 9a, O2 molecules are physisorbed (O2 − ) at low temperatures (<200 °C) and subsequently chemisorbed (O -  Figure 8 shows the kinetics of ZnFe 2 O 4 -based chemiresistive gas sensors. The response (τ res ) and recovery (τ rec ) times of the annealed ZnFe 2 O 4 nanoparticles are estimated from their response to 300 ppm NO 2 gas. All samples show fast response and recovery times. The response times are in the range of 145-195 s and follow the order: ZnFe 2 O 4 @800 > ZnFe 2 O 4 @600 > ZnFe 2 O 4 @1000. Thus, ZnFe 2 O 4 @1000 nanoparticles exhibit slightly longer response (τ res ) times compared to ZnFe 2 O 4 @600 and ZnFe 2 O 4 @800 nanoparticles. The recovery times are sharp (i.e., ≤20 s) for all samples and all sensors exhibit 100% recovery to their original state. At elevated temperatures, the recovery times are generally faster [10]. Overall, ZnFe 2 O 4 @1000 nanoparticles exhibit excellent NO 2 gas sensing properties such as high sensitivity, good response kinetics, and linear response in the tested concentration range. Nanomaterials 2020, 10, x FOR PEER REVIEW 9 of 14 a straight line. The sensitivity of NO2 sensors follows the order: ZnFe2O4@1000 > ZnFe2O4@800 > ZnFe2O4@600, which describes the effect of annealing temperature on sensor performance. An increase in annealing temperature improves the NO2 sensing properties of ZnFe2O4 nanoparticles. Therefore, ZnFe2O4@1000 nanoparticles exhibit 2.0-fold and 3.2-fold high sensitivity compared to ZnFe2O4@800 and ZnFe2O4@600 nanoparticles, respectively.  Figure 8 shows the kinetics of ZnFe2O4-based chemiresistive gas sensors. The response (τres) and recovery (τrec) times of the annealed ZnFe2O4 nanoparticles are estimated from their response to 300 ppm NO2 gas. All samples show fast response and recovery times. The response times are in the range of 145-195 s and follow the order: ZnFe2O4@800 > ZnFe2O4@600 > ZnFe2O4@1000. Thus, ZnFe2O4@1000 nanoparticles exhibit slightly longer response (τres) times compared to ZnFe2O4@600 and ZnFe2O4@800 nanoparticles. The recovery times are sharp (i.e., ≤20 s) for all samples and all sensors exhibit 100% recovery to their original state. At elevated temperatures, the recovery times are generally faster [10]. Overall, ZnFe2O4@1000 nanoparticles exhibit excellent NO2 gas sensing properties such as high sensitivity, good response kinetics, and linear response in the tested concentration range. The chemiresistive gas sensors function on the principles of changes in resistance of the sensing element when test gas molecules interact with the semiconductor surface [9]. Figure 9 demonstrates the gas sensing mechanism of ZnFe2O4 nanoparticles. ZnFe2O4 is an n-type semiconductor [56]. In principle, when the ZnFe2O4-based chemiresistive device is exposed to air at elevated temperatures, active oxygen species are adsorbed on the surface of ZnFe2O4 nanoparticles. As shown in Figure 9a, O2 molecules are physisorbed (O2 − ) at low temperatures (<200 °C) and subsequently chemisorbed (Oand O 2-) at elevated temperatures (>200 °C) by capturing mobile electrons (e -) from the surface. This The chemiresistive gas sensors function on the principles of changes in resistance of the sensing element when test gas molecules interact with the semiconductor surface [9]. Figure 9 demonstrates the gas sensing mechanism of ZnFe 2 O 4 nanoparticles. ZnFe 2 O 4 is an n-type semiconductor [56]. In principle, when the ZnFe 2 O 4 -based chemiresistive device is exposed to air at elevated temperatures, active oxygen species are adsorbed on the surface of ZnFe 2 O 4 nanoparticles. As shown in Figure 9a, O 2 molecules are physisorbed (O 2 − ) at low temperatures (<200 • C) and subsequently chemisorbed (Oand O 2-) at elevated temperatures (>200 • C) by capturing mobile electrons (e -) from the surface. This leads to the formation of a charge depletion layer on the surface of ZnFe 2 O 4 nanoparticles. Afterward, the surface is exposed to different concentrations of NO 2 gas and NO 2 being an electron-withdrawing molecule [57] further extracts mobile e − from the surface or interacts with the chemisorbed oxygen species, as shown in Figure 9b. Consequently, the density of major charge carriers (e − ) decreases, and the thickness of the depletion region increases, which increases the resistance of the device. The redox reactions taking place on the surface of ZnFe 2 O 4 nanoparticles are depicted in Figure 9.
Nanomaterials 2020, 10, x FOR PEER REVIEW 10 of 14 leads to the formation of a charge depletion layer on the surface of ZnFe2O4 nanoparticles. Afterward, the surface is exposed to different concentrations of NO2 gas and NO2 being an electron-withdrawing molecule [57] further extracts mobile e − from the surface or interacts with the chemisorbed oxygen species, as shown in Figure 9b. Consequently, the density of major charge carriers (e − ) decreases, and the thickness of the depletion region increases, which increases the resistance of the device. The redox reactions taking place on the surface of ZnFe2O4 nanoparticles are depicted in Figure 9. Considering the mechanism described above, it is important to understand the behavior of ZnFe2O4 nanoparticles annealed at different temperatures. It is believed that semiconducting metal oxides with more oxygen vacancies adsorb a large number of active oxygen species, which in turn facilitates the surface redox reactions with the target gas molecules and improves the gas response [55]. Thus, oxygen vacancies and lattice defects play a major role in determining the gas response of ZnFe2O4 nanoparticles. Therefore, ZnFe2O4@1000 nanoparticles exhibit the best NO2 gas sensing properties despite their slightly bigger crystallite size and smaller specific surface area. Table 3 shows a comparison of ferrite-based chemiresistive NO2 sensors. The results demonstrate the potential of stable ZnFe2O4@1000 nanoparticles for high-temperature gas sensing applications.  Figure 9. The gas sensing mechanism of ZnFe 2 O 4 nanoparticles-based chemiresistive gas sensors.
Considering the mechanism described above, it is important to understand the behavior of ZnFe 2 O 4 nanoparticles annealed at different temperatures. It is believed that semiconducting metal oxides with more oxygen vacancies adsorb a large number of active oxygen species, which in turn facilitates the surface redox reactions with the target gas molecules and improves the gas response [55]. Thus, oxygen vacancies and lattice defects play a major role in determining the gas response of ZnFe 2 O 4 nanoparticles. Therefore, ZnFe 2 O 4 @1000 nanoparticles exhibit the best NO 2 gas sensing properties despite their slightly bigger crystallite size and smaller specific surface area. Table 3 shows a comparison of ferrite-based chemiresistive NO 2 sensors. The results demonstrate the potential of stable ZnFe 2 O 4 @1000 nanoparticles for high-temperature gas sensing applications. Sputtering 600 50-300~4 ‡ -- [63] † The response (S) is reported for the highest tested concentration of NO 2 gas: S(%) = (R g -R a ) × 100/R a . ‡ If not reported as S(%), the response is measured as: S = R g /R a .

Conclusions
In summary, this study presents the effects of annealing temperature on the microstructure evolution and gas sensing properties of ZnFe 2 O 4 nanoparticles. A high-energy ball-milling procedure is used to prepare pure ZnFe 2 O 4 nanoparticles that are annealed at different temperatures (600-1000 • C). ZnFe 2 O 4 nanoparticles exhibit a random spinel lattice structure that is distorted during high-temperature annealing. The XRD results show an increase in the crystallite size, but a reduction in the lattice parameter and volume that is attributed to the presence of lattice defects as oxygen vacancies. The oxygen vacancies play a major role in controlling the sensitivity of ZnFe 2 O 4 nanoparticles. Thus, ZnFe 2 O 4 @1000 nanoparticles (annealed at 1000 • C) reveal the superior gas sensing properties with the highest sensitivity, good response kinetics, and linear response toward 100-400 ppm NO 2 gas. This is the first example of a ZnFe 2 O 4 @1000-based chemiresistive device showing significant gas response and stable sensor performance at 600 • C.