Synthesis of ZnAl2O4 and Evaluation of the Response in Propane Atmospheres of Pellets and Thick Films Manufactured with Powders of the Oxide

ZnAl2O4 nanoparticles were synthesized employing a colloidal method. The oxide powders were obtained at 300 °C, and their crystalline phase was corroborated by X-ray diffraction. The composition and chemical structure of the ZnAl2O4 was carried out by X-ray and photoelectron spectroscopy (XPS). The optical properties were studied by UV-vis spectroscopy, confirming that the ZnAl2O4 nanoparticles had a direct transition with bandgap energy of 3.2 eV. The oxide’s microstructures were microbars of ~18.2 nm in size (on average), as analyzed by scanning (SEM) and transmission (TEM) electron microscopies. Dynamic and stationary gas detection tests were performed in controlled propane atmospheres, obtaining variations concerning the concentration of the test gas and the operating temperature. The optimum temperatures for detecting propane concentrations were 200 and 300 °C. In the static test results, the ZnAl2O4 showed increases in propane response since changes in the material’s electrical conductance were recorded (conductance = 1/electrical resistance, Ω). The increases were ~2.8 at 200 °C and ~7.8 at 300 °C. The yield shown by the ZnAl2O4 nanoparticles for detecting propane concentrations was optimal compared to other similar oxides categorized as potential gas sensors.


Introduction
For decades, the preparation processes of inorganic materials have had great relevance for obtaining optimal microstructures for potential technological applications. Traditionally, the solid-state reaction technique (or ceramic method) has been the most commonly used to synthesize different types of materials [1]. However, some research groups have reported alternative routes of synthesis, such as wet chemistry, which is simple, economical, easy to apply, and can be very efficient for obtaining high purity inorganic materials at low the solution was kept at around 70 • C, measuring the temperature with an Extech 403,255 infrared thermometer. The purpose of using microwave radiation in the synthesis process is to elucidate its effects on ZnAl 2 O 4 's microstructure. After evaporation, the obtained paste was dried in air at 200 • C for 8 h, and the precursor material was afterward calcined at 300 • C. The calcination was done in a Novatech programmable oven with temperature control (Tlaquepaque, Mexico), starting at room temperature and reaching 300 • C at a heating rate of 100 • C/h for 5 h.

Physical Characterization of ZnAl 2 O 4 Powders
Purity and crystallinity of the ZnAl 2 O 4 powders were characterized by X-ray diffraction at room temperature using a Panalytical Empyrean device with CuKα radiation (λ = 1.546 Å). The diffraction was done using a 2θ continuous scan from 10 • to 80 • with 0.026 • -steps at a rate of 30 s per step. The powders' microstructure calcined at 300 • C was analyzed by means of a field-emission scanning electron microscope (FE-SEM, Tescan MIRA 3 LMU, Mexico City, Mexico) with an acceleration voltage of 10 kV in a high vacuum. The shape and size of the nanoparticles were studied with a transmission electron microscope (TEM, Joel JEM-ARM200F, Mexico City, Mexico) in image mode. For this, a representative amount (0.010 g) of the powders was placed in a vial containing methanol. Subsequently, the powders and the methanol were dispersed by an ultrasonic generator for 10 min and deposited on a 300-mesh copper grid containing a Formvar/carbon membrane.

Chemical and Optical Characterization of the ZnAl 2 O 4
Chemical composition and surface analysis of the ZnAl 2 O 4 calcined at 300 • C were studied using an X-ray photoelectron spectroscopy system (XPS SPECS, Berlin, Germany) equipped with a monochromatic Al Kα 1 (hv = 1486.7 eV) X-ray source and a hemispherical energy analyzer (PHOIBOS 150). To determine the value of the forbidden bandwidth and the material's percentage of reflection, a Cary 5000 UV-vis NIR spectrophotometer (Agilent Technologies, Santa Clara, United States) was used, equipped with a polytetrafluoroethylene (PTFE) integration sphere.

Pellets and Thick Films Preparation for Gas Sensing Tests
For the detection tests in propane atmospheres, pellets and thick films were made from the powders calcined at 300 • C. For the pellets, 0.350 g of ZnAl 2 O 4 powders were compressed using a manual hydraulic press (Simplex Ital Equip-25 tons, Mexico City, Mexico), applying a 10-ton load for 1 min. The dimensions of the pellets were 12 mm in diameter and 0.5 mm in thickness. For the thick films, 0.1 g of ZnAl 2 O 4 powders was used, which was dispersed with an ultrasonic generator (Branson 2510 Ultrasonic) in a plastic container containing ethyl alcohol. The dispersed powders were then dropwise placed in a ceramic ring so that there was a contact between the radially separated electrodes at two tips ("two-tips method"). The dimensions of the thick films were 0.3 mm in diameter and 0.5 mm in thickness.
During the sensing tests in static and dynamic propane atmospheres, changes in the material's electrical resistance (in DC) were expected depending on the propane concentration, the operating temperature, and the testing time (dynamic response). For that, two ohmic contacts were placed on the pellets' surface using colloidal silver paint (Alfa Aesar, 99%, Mexico City, Mexico). Subsequently, the pellets were placed in a small box (V = 19 cm 3 ) placed inside the measurement chamber (see Figure 1a). The purpose of this small box was to ensure fast response times. The box had two holes: one through, which the electrodes were inserted, and which also functioned as a gas outlet; the other orifice worked as the gas inlet (see Figure 1b). The gas was then removed through the evacuation system installed in the vacuum chamber (which had a pressure of 10 −3 torr). In the case of the static tests (as a function of the concentration and temperature), the total volume (10 L) of the vacuum chamber was used. The partial pressure of the test gases was monitored by an electronic detector (Leybold TM20, Oerlikon Leybold Vacuum, Cologne, Germany). The change in electrical resistance was measured with a digital multimeter (Keithley 2001, Cleveland, OH, USA) coupled to a control and acquisition system using the software LabView (National Instruments). For controlling the propane flow during the dynamic tests, a pair of massflow regulators (2600 and 10 cm 3 /min) were used (Brooks Instruments, GF100CXXC-SH452.6 L and GF100CXXC-SH40010C), at a total flow rate of 500 cm 3 /min. To estimate the static response (S) as a function of concentration and operating temperature, the relative difference of the electrical conductances (1/electrical resistance, Ω) was considered according to the formula S = (G G − G O )/G O [5], where G G and G O are the pellets' electrical conductance in propane and air, respectively.
For the dynamic tests with the thick films, these were placed in a quartz tube inside a tubular-type oven with programmable temperature control (Lindberg Blue M). The measurements were made at 200 and 300 • C by the two-tips method employing high purity platinum wires with a diameter of 0.006 mm. The control of the gas concentrations was made in the same manner as in the pellet detection tests.  Figure 2 shows an X-ray diffractogram of powders calcined at 300 • C. The diffraction peaks were located at 2θ angular positions: 18

XRD Analysis
where a is the lattice constant, λ is the wavelength of the radiation (1.5406 Å), (h,k,l) are the Miller indices, and θ is the Bragg angle. Considering all the reflections, the estimated lattice parameter for the ZnAl 2 O 4 was 8.060 Å with a standard deviation of 0.007 Å [24]. This result is consistent with those reported in the literature for the same compound [18,20,21,24]. The material's crystalline phase was identified by means of the PDF file 05-0669 corresponding to the cubic phase of the ZnAl 2 O 4 , with a spinel-like structure and an Fd-3m (227) spatial group [22].  The ZnAl2O4′s cell parameter was calculated using the equation [23,24]: where a is the lattice constant, λ is the wavelength of the radiation (1.5406 Å), (hkl) are the Miller indices, and θ is the Bragg angle. Considering all the reflections, the estimated lattice parameter for the ZnAl2O4 was 8.060 Å with a standard deviation of 0.007 Å [24]. This result is consistent with those reported in the literature for the same compound [18,20,21,24]. The material's crystalline phase was identified by means of the PDF file 05-0669 corresponding to the cubic phase of the ZnAl2O4, with a spinel-like structure and an Fd-3m (227) spatial group [22]. The synthesis method employed in this work was successful for obtaining the spinellike crystalline structure of the ZnAl2O4 at a relatively low-temperature (300 °C) compared to other works where the same oxide was synthesized but using other preparation methods. For example, zinc aluminate has been synthesized via a coprecipitation approach followed by a heat treatment at 900 °C [17,18]. ZnAl2O4 powders were obtained by heating a precursor in the air at 900 °C (pyrolysis) [20]. MAl2O4 (M = Ni, Cu, Zn) spinels have been prepared by the sol-gel autocombustion method in the temperature range of 1000-1200 °C [22]. In the present study, the use of ethylenediamine contributed to the formation of the crystalline phase at relatively low temperatures. This amine is a ligand that can bind to metal ions forming coordination compounds. When the ethylenediamine reacts with a transition-metal nitrate, this produces a coordinated complex whose conformation can act as a directing agent of structure [5].

Analysis of Chemical and Optical Properties
To determine the chemical composition of the calcined oxide at 300 °C, the XPS technique was applied. Figure 3a shows the elements on the material's surface. All spectra taken from the deep levels corresponding to Al 2p, Zn 2p, and O 1s were aligned with respect to the adventitious carbon C 1s centered at 284.8 eV and deconvolved with the software AAnalyzer (RDATAA, Mexico City, Mexico) [25]. The XPS spectrum associated with the Zn 2p ( Figure 3b) was deconvolved considering two doublets: The first one at 1020 and 1043 eV associated with Zn 2+ ions located at the tetrahedral sites of the ZnAl2O4 spinel network; the second one at 1021.3 and 1044.3 eV associated with Zn 2+ ions located at the octahedral sites in the spinel network [26,27]. Figure 3c shows the XPS spectrum The synthesis method employed in this work was successful for obtaining the spinellike crystalline structure of the ZnAl 2 O 4 at a relatively low-temperature (300 • C) compared to other works where the same oxide was synthesized but using other preparation methods. For example, zinc aluminate has been synthesized via a coprecipitation approach followed by a heat treatment at 900 • C [17,18]. ZnAl 2 O 4 powders were obtained by heating a precursor in the air at 900 • C (pyrolysis) [20]. MAl 2 O 4 (M = Ni, Cu, Zn) spinels have been prepared by the sol-gel autocombustion method in the temperature range of 1000-1200 • C [22]. In the present study, the use of ethylenediamine contributed to the formation of the crystalline phase at relatively low temperatures. This amine is a ligand that can bind to metal ions forming coordination compounds. When the ethylenediamine reacts with a transition-metal nitrate, this produces a coordinated complex whose conformation can act as a directing agent of structure [5].

Analysis of Chemical and Optical Properties
To determine the chemical composition of the calcined oxide at 300 • C, the XPS technique was applied. Figure 3a shows the elements on the material's surface. All spectra taken from the deep levels corresponding to Al 2p, Zn 2p, and O 1s were aligned with respect to the adventitious carbon C 1s centered at 284.8 eV and deconvolved with the software AAnalyzer (RDATAA, Mexico City, Mexico) [25]. The XPS spectrum associated with the Zn 2p ( Figure 3b) was deconvolved considering two doublets: The first one at 1020 and 1043 eV associated with Zn 2+ ions located at the tetrahedral sites of the ZnAl 2 O 4 spinel network; the second one at 1021.3 and 1044.3 eV associated with Zn 2+ ions located at the octahedral sites in the spinel network [26,27]. Figure 3c shows the XPS spectrum associated with Al 2p, which was deconvolved using two doublets located around 72 and 73.8 eV at the tetrahedral and octahedral sites, respectively, and associated with Al 3+ ions. The XPS spectrum associated with O 1 s (Figure 3d) was deconvolved using three singlets located at 529.3, 531.2, and 532.6 eV. The 529.3 and 531.2 peaks were associated with the O 2− ions on the O-Zn and O-Al bonds, so it can be assumed that these ions are located at the ZnO x and AlO x sites (tetrahedral or octahedral) in the spinel [28]. The peak located at 532.6 eV was associated with the chemisorbed oxygen on the material's surface [29,30]. associated with Al 2p, which was deconvolved using two doublets located around 72 and 73.8 eV at the tetrahedral and octahedral sites, respectively, and associated with Al 3+ ions. The XPS spectrum associated with O 1 s (Figure 3d) was deconvolved using three singlets located at 529.3, 531.2, and 532.6 eV. The 529.3 and 531.2 peaks were associated with the O 2-ions on the O-Zn and O-Al bonds, so it can be assumed that these ions are located at the ZnOx and AlOx sites (tetrahedral or octahedral) in the spinel [28]. The peak located at 532.6 eV was associated with the chemisorbed oxygen on the material's surface [29,30]. The percentage of reflection and the forbidden bandwidth were determined from UV-vis spectra. Figure 4 shows the UV-vis reflectance spectrum of ZnAl2O4, where an increase in the percentage of reflection from 400 to 700 nm towards infrared and a strong absorption from 400 nm towards ultraviolet can be observed. To determine the value of the forbidden bandwidth, the Kubelka-Munk (K-M) theory was considered in which the absorption coefficient is proportional to the reflectance [31]. Hence, from Tauc's formula, (ℎ ) ≈ (ℎ − ) , and if ≈ ( ) = ( ) , where R is the reflectance and n = 1/2 for a direct transition, the forbidden bandwidth's value Eg for the spinel ZnAl2O4 was obtained from the graph of ( ( )ℎ ) versus ℎ . As can be seen in Figure 4 (see inset), the value was 3.2 eV, which is less than the theoretically predicted value but similar to that experimentally reported [32,33]. The percentage of reflection and the forbidden bandwidth were determined from UV-vis spectra. Figure 4 shows the UV-vis reflectance spectrum of ZnAl 2 O 4 , where an increase in the percentage of reflection from 400 to 700 nm towards infrared and a strong absorption from 400 nm towards ultraviolet can be observed. To determine the value of the forbidden bandwidth, the Kubelka-Munk (K-M) theory was considered in which the absorption coefficient is proportional to the reflectance [31]. Hence, from Tauc's formula, where R is the reflectance and n = 1/2 for a direct transition, the forbidden bandwidth's value E g for the spinel ZnAl 2 O 4 was obtained from the graph of (F(R)hv) 2 versus hv. As can be seen in Figure 4 (see inset), the value was 3.2 eV, which is less than the theoretically predicted value but similar to that experimentally reported [32,33].    These photomicrographs show the multidirectional growth of a large number of microrods, which are distributed throughout the material's surface (similar to a tapestry, see Figure 5a,b). These microrods are positioned next to each other, almost in an orderly manner, causing a coated surface of the microstructure, as shown in the inset of Figure 5a. It     These photomicrographs show the multidirectional growth of a large number of microrods, which are distributed throughout the material's surface (similar to a tapestry, see Figure 5a,b). These microrods are positioned next to each other, almost in an orderly manner, causing a coated surface of the microstructure, as shown in the inset of Figure 5a. It These photomicrographs show the multidirectional growth of a large number of microrods, which are distributed throughout the material's surface (similar to a tapestry, see Figure 5a,b). These microrods are positioned next to each other, almost in an orderly It can also be observed that the microrods grow from areas located on microplates that act as a substrate. Such microrods agglomerate in a way that they tend to form a sphere-like geometry (see Figure 5b,c). Analyzing several SEM images, the length of the microrods was estimated in the range of 0.048 to 1.05 µm, with an average size of~0.678 µm and a standard deviation of~0.20 µm (see Figure 6). Observing another area of material's surface, it is corroborated that the compound's microstructure is formed by microrods that grow due to the effect of the temperature and the ethylenediamine concentration used during the synthesis process. The diameter of the microrods was estimated at~0.398 µm on average. In general, it is observed that the surface of the ZnAl 2 O 4 shows the same morphology, which can be seen in Figure 5d. In particular, the use of ethylenediamine as a way to produce 1D structures, such as rods, tubes, and wires, has been previously reported in the literature [34,35]. In our case, ethylenediamine acts as a template that is first incorporated into the inorganic network to later escape from it forming morphologies, such as the ones shown in Figure 5. Therefore, the presence of ethylenediamine determines the geometric features of the formed nuclei, which grow during the heat treatments [5]. We also have reported in previous works that by using ethylenediamine during the synthesis process, it is possible to obtain different types of morphologies [5,[36][37][38], like microrods, nanorods, microspheres, microoctahedra, and nanoparticles, which have been successfully proved as potential gas sensors for the detection of CO, CO 2 , and C 3 H 8 atmospheres [39]. Additionally, other authors, such as Michel et al. [40], used ethylenediamine for the synthesis of the semiconductor oxide CoSb 2 O 6 in order to obtain nanoparticles to be applied in the detection of CO 2 and O 2 . In general, the microstructure shown in Figure 5 follows the crystallization principles described by LaMer and Dinegar [41]. can also be observed that the microrods grow from areas located on microplates that act as a substrate. Such microrods agglomerate in a way that they tend to form a sphere-like geometry (see Figure 5b,c). Analyzing several SEM images, the length of the microrods was estimated in the range of 0.048 to 1.05 µm, with an average size of ~0.678 µm and a standard deviation of ~0.20 µm (see Figure 6). Observing another area of material's surface, it is corroborated that the compound's microstructure is formed by microrods that grow due to the effect of the temperature and the ethylenediamine concentration used during the synthesis process. The diameter of the microrods was estimated at ~0.398 µm on average. In general, it is observed that the surface of the ZnAl2O4 shows the same morphology, which can be seen in Figure 5d. In particular, the use of ethylenediamine as a way to produce 1D structures, such as rods, tubes, and wires, has been previously reported in the literature [34,35]. In our case, ethylenediamine acts as a template that is first incorporated into the inorganic network to later escape from it forming morphologies, such as the ones shown in Figure 5. Therefore, the presence of ethylenediamine determines the geometric features of the formed nuclei, which grow during the heat treatments [5]. We also have reported in previous works that by using ethylenediamine during the synthesis process, it is possible to obtain different types of morphologies [5,[36][37][38], like microrods, nanorods, microspheres, microoctahedra, and nanoparticles, which have been successfully proved as potential gas sensors for the detection of CO, CO2, and C3H8 atmospheres [39]. Additionally, other authors, such as Michel et al. [40], used ethylenediamine for the synthesis of the semiconductor oxide CoSb2O6 in order to obtain nanoparticles to be applied in the detection of CO2 and O2. In general, the microstructure shown in Figure  5 follows the crystallization principles described by LaMer and Dinegar [41].

TEM Analysis
For a more detailed analysis of the individual particles that made up the ZnAl2O4 oxide, transmission electron microscopy (TEM) was employed. Figure 7 show typical bright-field TEM images of the oxide calcined at 300 °C.

TEM Analysis
For a more detailed analysis of the individual particles that made up the ZnAl 2 O 4 oxide, transmission electron microscopy (TEM) was employed. Figure 7 show typical bright-field TEM images of the oxide calcined at 300 • C. In the TEM images, dark areas are observed that are attributed to the poor transmission of electrons caused by the agglomeration of nanoparticles on the material's surface. It is mentioned in the literature that, when using ethylenediamine during the synthesis process, particle agglomeration occurs, where the thermal treatment (in our case at 300 °C) plays a key role in obtaining agglomerated areas of nanoparticles, as shown in Figure  7a-d [37,39]. In a more detailed study of the TEM images, we corroborated our SEM analysis where the rod growth and nucleation are observed, as well as other nanoparticles with no apparent shape (Figure 7a). Furthermore, it can be seen that the nanoparticles agglomerated one with the other until the nanorod morphology shown in Figure 7b was obtained. Regarding Figure 7c,d, it is observed that the surface of the ZnAl2O4 is made up of irregularly shaped nanometric-sized particles, which are joined by coalescence. The visible dispersion of the nanoparticles is attributed to the preparation technique for the TEM analysis, which required that the ZnAl2O4 was dispersed by ultrasound. The particle size, in a range from 5 to 35 nm (with an average of ~18.2 nm and a standard deviation of ~6.5 nm), was estimated by examining several TEM images (Figure 8). In the TEM images, dark areas are observed that are attributed to the poor transmission of electrons caused by the agglomeration of nanoparticles on the material's surface. It is mentioned in the literature that, when using ethylenediamine during the synthesis process, particle agglomeration occurs, where the thermal treatment (in our case at 300 • C) plays a key role in obtaining agglomerated areas of nanoparticles, as shown in Figure 7a-d [37,39]. In a more detailed study of the TEM images, we corroborated our SEM analysis where the rod growth and nucleation are observed, as well as other nanoparticles with no apparent shape ( Figure 7a). Furthermore, it can be seen that the nanoparticles agglomerated one with the other until the nanorod morphology shown in Figure 7b was obtained. Regarding Figure 7c,d, it is observed that the surface of the ZnAl 2 O 4 is made up of irregularly shaped nanometric-sized particles, which are joined by coalescence. The visible dispersion of the nanoparticles is attributed to the preparation technique for the TEM analysis, which required that the ZnAl 2 O 4 was dispersed by ultrasound. The particle size, in a range from 5 to 35 nm (with an average of~18.2 nm and a standard deviation of~6.5 nm), was estimated by examining several TEM images (Figure 8). Figure 9 shows high-resolution TEM images (HRTEM) of the ZnAl 2 O 4 nanoparticles obtained at 300 • C, a typical electron diffraction pattern of the oxide, and a histogram that corroborates the interplanar spacing of the ZnAl 2 O 4 . In Figure 9a,b, it can be seen that the material's surface consisted of very fine and irregular nanoparticles (~20 nm in size). The borders between the individual nanoparticles were drawn with yellow-dotted lines. We could corroborate in these individual nanoparticles the presence of the oxide's crystalline planes. Selecting some areas of the particle surfaces, we could estimate the distance between the planes at~0.47 nm, which corresponded to the crystalline plane (111) of the spinel's crystal structure, as shown in Figure 9a-d. Figure 9e shows a typical electrondiffraction pattern of the ZnAl 2 O 4 nanoparticles, where the characteristic reflection rings of a nanometric polycrystalline material are observed. Accordingly, the first six reflections corresponding to the crystalline planes (111), (220), (311), (400), (331), and (422) were indexed, confirming the sample's local crystallinity, consistently with the XRD study (see Figure 2).  Figure 9 shows high-resolution TEM images (HRTEM) of the ZnAl2O4 nanoparticles obtained at 300 °C, a typical electron diffraction pattern of the oxide, and a histogram that corroborates the interplanar spacing of the ZnAl2O4. In Figure 9a,b, it can be seen that the material's surface consisted of very fine and irregular nanoparticles (~20 nm in size). The borders between the individual nanoparticles were drawn with yellow-dotted lines. We could corroborate in these individual nanoparticles the presence of the oxide's crystalline planes. Selecting some areas of the particle surfaces, we could estimate the distance between the planes at ~0.47 nm, which corresponded to the crystalline plane (111) of the spinel's crystal structure, as shown in Figure 9a-d. Figure 9e shows a typical electrondiffraction pattern of the ZnAl2O4 nanoparticles, where the characteristic reflection rings of a nanometric polycrystalline material are observed. Accordingly, the first six reflections corresponding to the crystalline planes (111), (220), (311), (400), (331), and (422) were indexed, confirming the sample's local crystallinity, consistently with the XRD study (see Figure 2).    Figure 9 shows high-resolution TEM images (HRTEM) of the ZnAl2O4 nanoparticles obtained at 300 °C, a typical electron diffraction pattern of the oxide, and a histogram that corroborates the interplanar spacing of the ZnAl2O4. In Figure 9a,b, it can be seen that the material's surface consisted of very fine and irregular nanoparticles (~20 nm in size). The borders between the individual nanoparticles were drawn with yellow-dotted lines. We could corroborate in these individual nanoparticles the presence of the oxide's crystalline planes. Selecting some areas of the particle surfaces, we could estimate the distance between the planes at ~0.47 nm, which corresponded to the crystalline plane (111) of the spinel's crystal structure, as shown in Figure 9a-d. Figure 9e shows a typical electrondiffraction pattern of the ZnAl2O4 nanoparticles, where the characteristic reflection rings of a nanometric polycrystalline material are observed. Accordingly, the first six reflections corresponding to the crystalline planes (111), (220), (311), (400), (331), and (422) were indexed, confirming the sample's local crystallinity, consistently with the XRD study (see Figure 2).

Gas Sensing Tests
The results of the static detection tests on the ZnAl 2 O 4 pellets at different operating temperatures and propane concentrations are shown in Figure 10. The propane concentrations were 1, 5, 50, 100, 200, 300, 400, and 500 ppm, while the temperatures were 100, 200, and 300 • C. As can be seen in Figure 10, the ZnAl 2 O 4 shows an increase in its response as the propane concentration and the working temperature increase during the test. However, at 100 • C, the pellets did not show changes in electrical resistance with increasing concentration and operating temperature (see Figure 10a,b). This behavior is attributed to the fact that the relatively low-temperature is not enough for the reaction of the pellets' surface and the test gas. This caused that at 100 • C, oxygen desorption did not occur [38,39] and, therefore, a low response of the material. Some authors also observed the poor response that shows some semiconductors at low temperatures (in our case at 100 • C), such as ours. They attributed such phenomenon to the fact that the oxygen species present (usually O − 2 [2]) do not react with the test gas nor with the pellets' surface (that is, adsorption and desorption processes do not occur) [5,10,12,17], provoking the absence of changes in the material's electrical resistance [5,36] (see Figure 10). On the contrary, when increasing the operating temperature to 200 and 300 • C, a very significant increase in the response of the pellets was recorded (see Figure 10a,b). The increases in response are largely due to the fact that, with increasing thermal energy, O − oxygen species are easily adsorbed on the surface of the pellets [2,42]. These species are more reactive and capable of causing a greater speed in the mobility of charge carriers, having as a consequence an improvement in sensitivity, which in turn promotes an increase in the response of the ZnAl 2 O 4 . Therefore, by increasing the temperature to 300 • C, the response increased 2.7 times more than that measured at 200 • C (2.8 at 200 • C and 7.8 at 300 • C, both at 500 ppm of propane). The response increase is associated with the increase in the concentration of the propane gas that reacts with chemisorbed oxygen (ionic oxygen) on the material's surface due to the rise of temperature. That caused variations in the semiconductor's electrical resistance (or conductivity) and the increase of the sensor's response [5,14,15,36]. The trend shown in Figure 10 is consistent with the results of similar oxides reported in the literature [2,5,36,43]. as ours. They attributed such phenomenon to the fact that the oxygen species present (usually O [2]) do not react with the test gas nor with the pellets' surface (that is, adsorption and desorption processes do not occur) [5,10,12,17], provoking the absence of changes in the material's electrical resistance [5,36] (see Figure 10). On the contrary, when increasing the operating temperature to 200 and 300 °C, a very significant increase in the response of the pellets was recorded (see Figure 10a,b). The increases in response are largely due to the fact that, with increasing thermal energy, O oxygen species are easily adsorbed on the surface of the pellets [2,42]. These species are more reactive and capable of causing a greater speed in the mobility of charge carriers, having as a consequence an improvement in sensitivity, which in turn promotes an increase in the response of the ZnAl2O4. Therefore, by increasing the temperature to 300 °C, the response increased 2.7 times more than that measured at 200 °C (2.8 at 200 °C and 7.8 at 300 °C, both at 500 ppm of propane). The response increase is associated with the increase in the concentration of the propane gas that reacts with chemisorbed oxygen (ionic oxygen) on the material's surface due to the rise of temperature. That caused variations in the semiconductor's electrical resistance (or conductivity) and the increase of the sensor's response [5,14,15,36]. The trend shown in Figure 10 is consistent with the results of similar oxides reported in the literature [2,5,36,43].
As seen in Figure 10a, at a propane concentration of 50 ppm, and at 200 and 300 °C, the response increased significantly. However, when the concentration was increased to 200, 300, 400, and 500 ppm, no significant changes were observed. Even at 300 °C, there was no change in the response from 400 to 500 ppm. This could be explained by the saturation of the sensor's surface. Desorption and adsorption of gas molecules were in equilibrium at these concentrations, which meant that additional molecules could not be adsorbed directly at the sensor's surface, causing a delayed contribution to resistance changes. Figure 10. Response of ZnAl2O4 pellets as a function of (a) propane concentration, (b) operating temperature. Figure 10. Response of ZnAl 2 O 4 pellets as a function of (a) propane concentration, (b) operating temperature.
As seen in Figure 10a, at a propane concentration of 50 ppm, and at 200 and 300 • C, the response increased significantly. However, when the concentration was increased to 200, 300, 400, and 500 ppm, no significant changes were observed. Even at 300 • C, there was no change in the response from 400 to 500 ppm. This could be explained by the saturation of the sensor's surface. Desorption and adsorption of gas molecules were in equilibrium at these concentrations, which meant that additional molecules could not be adsorbed directly at the sensor's surface, causing a delayed contribution to resistance changes.
Dynamic tests on pellets and thick films made from ZnAl 2 O 4 powders were carried out alternatively in air-propane atmospheres by the two-tips method using a direct current signal (DC). The variation of the electrical resistance was monitored while injecting synthetic air (20% O 2 , 80% N 2 ) as a stabilizing gas for the pellets (500 mL/min of synth. air) and the films (1500 mL/min of synth. air). Propane's concentration was 1000 ppm in balance with N 2 (air-propane) at 200 and 300 • C in both cases. Material's dynamic response was measured as follows: first, the pellets and the thick films were stabilized in synthetic air for 5 min at each temperature; subsequently, propane was injected for 5 min, provoking changes in the electrical resistance of the pellets and the thick films; then, the propane was withdrawn leaving the pellets and the thick films stabilizing in synthetic air for 5 min up to reaching the initial values of the electrical resistance. This process was repeated cyclically for each temperature (200 and 300 • C). Figures 11 and 12 show the change in electrical resistance as a function of time at the given temperatures and propane concentration. As can be observed, the cycle repetitions are indicative of good reproducibility. In addition, as expected, in both cases, the oxide powders showed a decrease in electrical resistance (or an increase in conductivity) when the propane was introduced into the measurement chamber. The change was more noticeable at 300 • C. According to Figure 11a,b, the variations in electrical resistance were estimated at 0.126 MΩ (on average) at 200 • C, while at 300 • C it was calculated at 6.677 MΩ. The response and recovery times were calculated according to reference [44], which considered the changes in electrical resistance as a function of time. As in reference [45], we considered 90% of the total response and 10% of the saturation values in air. Therefore, when applying these criteria to the results shown in Figure 11, we estimated the response and recovery times at 200 • C were 206 and 160 s. At 300 • C, response and recovery times were 176 s and 205 s. was measured as follows: first, the pellets and the thick films were stabilized in synthetic air for 5 min at each temperature; subsequently, propane was injected for 5 min, provoking changes in the electrical resistance of the pellets and the thick films; then, the propane was withdrawn leaving the pellets and the thick films stabilizing in synthetic air for 5 min up to reaching the initial values of the electrical resistance. This process was repeated cyclically for each temperature (200 and 300 °C). Figures 11 and 12 show the change in electrical resistance as a function of time at the given temperatures and propane concentration. As can be observed, the cycle repetitions are indicative of good reproducibility. In addition, as expected, in both cases, the oxide powders showed a decrease in electrical resistance (or an increase in conductivity) when the propane was introduced into the measurement chamber. The change was more noticeable at 300 °C. According to Figure 11a,b, the variations in electrical resistance were estimated at 0.126 MΩ (on average) at 200 °C, while at 300 °C it was calculated at 6.677 MΩ. The response and recovery times were calculated according to reference [44], which considered the changes in electrical resistance as a function of time. As in reference [45], we considered 90% of the total response and 10% of the saturation values in air. Therefore, when applying these criteria to the results shown in Figure 11, we estimated the response and recovery times at 200 °C were 206 and 160 s. At 300 °C, response and recovery times were 176 s and 205 s. Similar trends were obtained on the thick films. The results are depicted in Figure  12a,b, where changes in electrical resistance are observed as a function of time at different operating temperatures. It can be verified that the effect of temperature and propane concentration played an important role during the tests. As in the pellets' case, the changes in the electrical resistance fluctuated due to the reaction that occurred between propane and oxygen on the thick film's surface at the given operating temperature, causing a decrease in electric resistance as a function of time. The average variation of the electrical resistance at 200 °C was 174.03 kΩ, with a response time of 192.40 s and a recovery time of 11.07 s, while at 300 °C it was 96.05 kΩ, with a response time of 74.06 s and a recovery time of 8.7 s. Again, as in the case of the dynamic response on the pellets, we considered references [44,45] to calculate the response and recovery times. The response depicted in Figures 11 and 12 is commonly observed in type-n semiconductor oxides [33] when they are exposed to atmospheres similar to ours [14,43]. Again, we consider that the increase in operating temperature (in our case from 200 to 300 °C) was a determining factor in the pellet's response due to the presence of oxygen during the tests. References [2,46] report that several oxygen species appear as a consequence of the temperature during gas detection experiments: when testing at temperatures below 150 °C, the predominant oxygen species is O , while at temperatures greater than 150 °C, the predominant oxygen species are O and O , which are more reactive than O [5,36,46]; oxygen reactions produced at temperatures above 150 °C (in our case 200 and 300 °C) cause an increase in the reaction kinetics of the gas molecules on the pellet surfaces and thereby a sensitivity increase. It has been reported that other factors that affect the test gas chemisorption are both the presence of oxygen and the material's microstructure (size and morphology) [47] since if the particle size is less than 2 times the thickness of the outer layer (LS), the adsorbed oxygen species are those that lead to a greater material's response [43,48], as occurred in our case. Similar trends were obtained on the thick films. The results are depicted in Figure 12a,b, where changes in electrical resistance are observed as a function of time at different operating temperatures. It can be verified that the effect of temperature and propane concentration played an important role during the tests. As in the pellets' case, the changes in the electrical resistance fluctuated due to the reaction that occurred between propane and oxygen on the thick film's surface at the given operating temperature, causing a decrease in electric resistance as a function of time. The average variation of the electrical resistance at 200 • C was 174.03 kΩ, with a response time of 192.40 s and a recovery time of 11.07 s, while at 300 • C it was 96.05 kΩ, with a response time of 74.06 s and a recovery time of 8.7 s. Again, as in the case of the dynamic response on the pellets, we considered references [44,45] to calculate the response and recovery times.
The response depicted in Figures 11 and 12 is commonly observed in type-n semiconductor oxides [33] when they are exposed to atmospheres similar to ours [14,43]. Again, we consider that the increase in operating temperature (in our case from 200 to 300 • C) was a determining factor in the pellet's response due to the presence of oxygen during the tests. References [2,46] report that several oxygen species appear as a consequence of the temperature during gas detection experiments: when testing at temperatures below 150 • C, the predominant oxygen species is O − 2 , while at temperatures greater than 150 • C, the predominant oxygen species are O − and O 2− , which are more reactive than O − 2 [5,36,46]; oxygen reactions produced at temperatures above 150 • C (in our case 200 and 300 • C) cause an increase in the reaction kinetics of the gas molecules on the pellet surfaces and thereby a sensitivity increase. It has been reported that other factors that affect the test gas chemisorption are both the presence of oxygen and the material's microstructure (size and morphology) [47] since if the particle size is less than 2 times the thickness of the outer layer (L S ), the adsorbed oxygen species are those that lead to a greater material's response [43,48], as occurred in our case.
In the case of the pellets, the sensor's resistance varied from~8 MΩ without gas tõ 1 MΩ in propane at 300 • C ( Figure 11b); the ratio R 0 /R G was 8. Instead, for the thick films, at the same temperature and propane concentration, the change in the resistance varied from~480 MΩ to~380 MΩ ( Figure 12b); the ratio R 0 /R G was 1.26. The increased pellet response could be attributed to the preparation of the sensor: pressure was applied to the powder, which should decrease the average distance between the particles. The current between the grains must be strongly controlled by the Schottky barriers between the grains of the material. On the other hand, the thick films were drip prepared. As a consequence, their particles had, on average, a greater distance (i.e., there was less contact) between them. Therefore, higher electrical resistance was generated with the associated tunneling effect. Another effect of the grain compaction was observed in the response times: there was greater gas mobility through the film (associated with more pores) compared to the pellets.
In general, the response of a semiconductor material (such as the one studied in this work) is based on the mechanism that involves changes in the conductance (or electrical resistance) produced by the electron transfer that occurs during chemical adsorption [8,14,43]. Depending on the semiconductor type (p or n), the surface charge carriers may increase or decrease, as appropriate [14]. The mobility and variations of the charge carriers depend on the temperature [49], the morphology, and the particle size of the semiconductor (which in our case was~18.2 nm) [47,49,50]. This means that the interaction between the oxygen and the semiconductor's surface will be more active due to the effect of temperature. That will provoke more significant changes in the semiconductor's electrical resistance, therefore, meaning a greater response, as happened in this work.
The sensitivity of the ZnAl 2 O 4 prepared in this work was compared with previous studies on propane gas sensing. For example, thin-films of SnO 2 showed a sensitivity of 0.7 at a propane concentration of 300 ppm at a temperature of 300 • C [51]. At the same experimental conditions, thin-films of ZnO exhibited a response of 0.4 [52]. The trirutile-type ZnSb 2 O 6 [53] and CoSb 2 O 6 [54] showed responses of 1.3 and 4.8, respectively, to 300 ppm of propane at temperatures between 250 and 300 • C. The cited studies do not report information on response and recovery times. The results obtained in the present study highlight the synthesized spinel as a promising material in gas sensing applications.

Conclusions
In this work, ZnAl 2 O 4 nanoparticles were synthesized by the microwave-assisted colloidal method, which is a simple and economical process that does not require sophisticated equipment for obtaining both nanometric particle sizes and the crystalline phase at low temperatures. The powders were studied chemically by XPS, which provided information on the oxidation states of the Zn and the Al. The bandgap of the oxide was determined by UV-vis, obtaining a value of 3.2 eV. The oxide nanoparticles showed an increase in sensitivity as the propane concentration and the operating temperature were also raised.
The maximum sensitivity value in the static test was~7.8 for a propane concentration of 500 ppm at 300 • C. The dynamic tests on pellets and thick films of ZnAl 2 O 4 revealed that with the increase in operating temperature and test gas concentration, the response of the nanoparticles improved substantially, which was reflected in the changes in their electrical resistance. The best response was at a temperature of 300 • C with short response and recovery times. According to our results, ZnAl 2 O 4 nanoparticles can be a viable option to be used as a propane sensor.