Gas Sensors Based on Porous Ceramic Bodies of MSnO3 Perovskites (M = Ba, Ca, Zn): Formation and Sensing Properties towards Ethanol, Acetone, and Toluene Vapours

In this work, the gas-sensing functionality of porous ceramic bodies formed by the slip casting technique was studied using perovskite nanoparticles of an MSnO3 system (M = Ba, Ca, Zn) synthesized by a chemical route. The performance and reliability of the sensitive materials in the presence of different volatile organic compounds (acetone, ethanol, and toluene), and other gases (CO, H2 and NO2) were analysed. The ZnSnO3, BaSnO3, and CaSnO3 sensors showed sensitivities of 40, 16, and 8% ppm−1 towards acetone, ethanol, and toluene vapours, respectively. Good repeatability and selectivity were also observed for these gaseous analytes, as well as excellent stability for a period of 120 days. The shortest response times were recorded for the ZnSnO3 sensors (e.g., 4 s for 80 ppm acetone) with marked responses to low concentrations of acetone (1000 ppb). These results are attributed to the porosity of the sensitive materials, which favours the diffusion of gases, induces surface defects, and provides greater surface area and good sensitivity to acetone, as is seen in the case of ZnSnO3.


Introduction
Metal oxide (MOx) gas sensors are of high interest due to their small size, low cost [1], and rapid response compared with traditional instruments, such as mass spectrometry and gas chromatography [2], hence the interest in investigating, in a more systematic and rigorous way, the sensitivity, selectivity, and stability of these materials in the presence of a particular gas. This type of sensor can be used in environmental protection [3], safety instruments in laboratories and industry [4], and in the field of medicine for the early diagnosis of diseases [5], among other applications. MOx micro/nanostructures, e.g., based on TiO 2 , ZnO and SnO 2 , have been used as gas-sensitive materials for several decades due to their unique physicochemical properties. However, the lack of selectivity and high operating temperatures (200-500 • C) of these oxides [6] have led to the development of special MOx to further promote the physicochemical properties and thus better expand the practical applications of these materials. The MOx improvements include their doping/decoration with noble metals (e.g., Pd, Ag, Pt, and Au) [7,8], surface functionalization [9,10], and formation of composites (e.g., MOx-MOx, polymer-MOx, and MOx-carbon nanotubes (CNTs)) [11]. These MOx modifications have shown that the detection performance of a gas is mainly mediated by the properties of the surface of the sensitive material and its combination with multiple components that act synergistically to increase the sensitivity, selectivity, and response rates during gas detection. Other factors to consider are morphology and size casting, allowed the formation of small disc-shaped bodies from concentrated suspensions of nanoparticles of the MSnO 3 system (M = Ba, Ca, Zn). The shaped discs were sintered to improve their densification and mechanical properties while preserving their stoichiometry. In the sintered bodies, the structure, morphology, chemical compositions, and gas-sensing properties were studied. The experimental results show that the bodies of MSnO 3 (M = Ba, Ca, Zn) exhibit excellent gas-sensing performance towards ethanol, acetone, and toluene vapours. The lowest measurements for these gaseous analytes in humid environments (10-30% relative humidity) were performed at 1000 ppb.

Raw Material
The preparation of the compounds and their corresponding structural analysis can be found in [56]. Specifically, particles with sizes between 20 and 100 nm of the MSnO 3 system (M = Ba, Ca, Zn) synthesized by the complex polymeric method were used. The surface areas of these particles were determined using a surface area and pore analyser Quantachrome NOVA 1000e; the results were 38,913 m 2 ·g −1 , 4056 m 2 ·g −1 , and 2560 m 2 ·g −1 for BaSnO 3 , CaSnO 3 , and ZnSnO 3 , respectively.

Sensor Fabrication
To form the gas-sensitive material (a disc-shaped ceramic body), the slip casting technique was used. This method includes the preparation of aqueous suspensions (slurries) with the ceramic powders of interest. The proportions of distilled water and solid content (i.e., MSnO 3 perovskites (M = Ba, Ca, Zn)) were experimentally determined by viscosity (   3 of 23 as developed without complex instruacuum chambers) and with repeatable sensitive materials. This method, slip ped bodies from concentrated suspen-Ba, Ca, Zn). The shaped discs were sinl properties while preserving their stoiorphology, chemical compositions, and ental results show that the bodies of ing performance towards ethanol, acents for these gaseous analytes in humid rformed at 1000 ppb.
corresponding structural analysis can between 20 and 100 nm of the MSnO3 lex polymeric method were used. The using a surface area and pore analyser 913 m 2 ·g −1 , 4056 m 2 ·g −1 , and 2560 m 2 ·g −1 ped ceramic body), the slip casting techation of aqueous suspensions (slurries) ions of distilled water and solid content erimentally determined by viscosity (ɳ gy for the preparation of colloidal susrties in [57]. The rheological tests were m Thermo Scientific with a concentric of the suspensions were determined to O3, and CaSnO3 systems, respectively. er-type disc-shaped moulds (70 mm in s sucked into the areas in contact with ct packed particles, which grew in the act material formed was removed from ram of the methodology used to obtain ly described, is presented in Figure 1a. bolite RHF 1600 electric furnace at tempycnometer (Ultrapyc gas pycnometer mine the density in the sintered bodies. g Equation (1), described in the ASTM

× 100%
(1) eight (g); and S is the suspended weight ge the relationship of the volume of the hology, and pore size distribution were D8 Advance operated at 40 kV and 40 aman spectroscopy (Raman-Horiba oscopy (XPS-Kratos Axis Supra specce, Manchester, UK), scanning electron ) vs. solid content (ϕ) curves, using the methodology for the preparation of colloidal suspensions and the testing of their rheological properties in [57]. The rheological tests were performed using a Haake Mars III rheometer from Thermo Scientific with a concentric cylinder configuration. The optimal solid contents of the suspensions were determined to be 13.7, 19.2, and 21.3 vol% for the BaSnO 3 , ZnSnO 3 , and CaSnO 3 systems, respectively. Subsequently, the slurries were poured into plaster-type disc-shaped moulds (70 mm in diameter and 30 mm in thickness). The liquid was sucked into the areas in contact with the walls of the mould, forming a layer of compact packed particles, which grew in the suspension from the walls of the mould. The compact material formed was removed from the mould and left in a desiccator for 24 h. A diagram of the methodology used to obtain the green body of the sensitive material, previously described, is presented in Figure 1a.
The ceramic specimens were sintered in a Carbolite RHF 1600 electric furnace at temperatures between 1000-1500 • C for 1 h. A helium pycnometer (Ultrapyc gas pycnometer 3000, Anton-Paar, Graz, Austria) was used to determine the density in the sintered bodies. The apparent porosity (AP%) was calculated using Equation (1), described in the ASTM Standard C20 and in [58].
where D is the weight in air (g); W is the soaked weight (g); and S is the suspended weight (g). The apparent porosity expresses as a percentage the relationship of the volume of the open pores in the specimen to its exterior volume. Their structure, chemical composition, morphology, and pore size distribution were studied by X-ray diffraction (XRD-Bruker, AXS D8 Advance operated at 40 kV and 40 mA, Cu Kα radiation, Karlsruhe, Germany), Raman spectroscopy (Raman-Horiba XploRa, Kyoto, Japan), X-ray photoelectron spectroscopy (XPS-Kratos Axis Supra spectrometer, with Al/Ag monochromatic X-ray source, Manchester, UK), scanning electron microscopy (SEM-Carl Zeiss, Auriga Series, Oberkochen, Germany), and mercury porosimetry (Micromeritics, AutoPore IV 9500 model, Norcross, GA, USA), respectively. The XPS spectra were deconvoluted with CasaXPS software version 2.3.24 (Computer aided surface analysis for X-Ray photoelectron spectroscopy, Casa Software Ltd., Devon, UK) using a Gaussian/Lorentzian (85/15) product function after subtraction of a Shirley nonlinear sigmoid-type baseline. The sensitive materials (sintered ceramic bodies) were manually integrated using silver paint to improve adhesion on the platinum electrodes that make up the system. The integrated sensor is shown in Figure 1b. spectrometer, with Al/Ag monochromatic X-ray source, Manchester, UK), scanning electron microscopy (SEM-Carl Zeiss, Auriga Series, Oberkochen, Germany), and mercury porosimetry (Micromeritics, AutoPore IV 9500 model, Norcross, US), respectively. The XPS spectra were deconvoluted with CasaXPS software version 2.3.24 (Computer aided surface analysis for X-Ray photoelectron spectroscopy, Casa Software Ltd., United Kingdom) using a Gaussian/Lorentzian (85/15) product function after subtraction of a Shirley nonlinear sigmoid-type baseline. The sensitive materials (sintered ceramic bodies) were manually integrated using silver paint to improve adhesion on the platinum electrodes that make up the system. The integrated sensor is shown in Figure 1b.

Sensor Testing
The sensors based on sintered bodies were tested using a system equipped with 6 chambers connected in parallel (each with a volume of approximately 3.2 mL) with continuous flow, equipped with mass flow controllers that allowed a mixture of dry/humid air. Calibrated analyte gases (acetone, ethanol, toluene, CO, H2, and NO2 purchased from Praxair) were used to obtain the desired concentration. Moisture was generated by bubbling, and the relative humidity (RH) control inside the gas chamber was monitored with an evaluation kit (EK-H4, Sensirion AG, Stäfa, Switzerland) equipped with a humidity sensor. An adjustable heating plate (Combiplac, JP SELECTA) was used to set the temperature of the sensitive material, which was monitored during the test.
The measurement of the DC resistance of the sensors was achieved with an electrometer (Keithley 6517B, Germering, Germany) with a multiplexer relay to simultaneously monitor several sensors (see the diagram of the test system in Figure 2, adapted from [59]). The sensor response was defined as Ra/Rg, where Ra and Rg are the resistance in dry/humid air and the resistance after exposure to the gas, respectively. The sensors were exposed to the target gas for a period of 5 min in the phase of the identification of the optimal temperatures and responses to gases. Subsequent tests (i.e., at the optimal temperature and with the selected gases) were performed by exposing the sensors to the target gases for 10 min. The sensors were tested over a period of 2 months, during which each sensor accumulated 400 h of operation under the different conditions (gases, temperatures, and humidities) used. The tests were performed in duplicate for each sensitive material to evaluate the repeatability of the sensors.

Sensor Testing
The sensors based on sintered bodies were tested using a system equipped with 6 chambers connected in parallel (each with a volume of approximately 3.2 mL) with continuous flow, equipped with mass flow controllers that allowed a mixture of dry/humid air. Calibrated analyte gases (acetone, ethanol, toluene, CO, H 2 , and NO 2 purchased from Praxair, Swindon, UK) were used to obtain the desired concentration. Moisture was generated by bubbling, and the relative humidity (RH) control inside the gas chamber was monitored with an evaluation kit (EK-H4, Sensirion AG, Stäfa, Switzerland) equipped with a humidity sensor. An adjustable heating plate (Combiplac, JP SELECTA, Barcelon, Spain) was used to set the temperature of the sensitive material, which was monitored during the test.
The measurement of the DC resistance of the sensors was achieved with an electrometer (Keithley 6517B, Germering, Germany) with a multiplexer relay to simultaneously monitor several sensors (see the diagram of the test system in Figure 2, adapted from [59]). The sensor response was defined as R a /R g , where R a and R g are the resistance in dry/humid air and the resistance after exposure to the gas, respectively. The sensors were exposed to the target gas for a period of 5 min in the phase of the identification of the optimal temperatures and responses to gases. Subsequent tests (i.e., at the optimal temperature and with the selected gases) were performed by exposing the sensors to the target gases for 10 min. The sensors were tested over a period of 2 months, during which each sensor accumulated 400 h of operation under the different conditions (gases, temperatures, and humidities) used. The tests were performed in duplicate for each sensitive material to evaluate the repeatability of the sensors.

Optimum Loading of Solids in the Suspensions
The amount of MSnO3 powders (M = Ba, Ca, Zn) to be used to form the suspensions was determined by graphs of viscosity (η) vs. volumetric fraction of solids (φ) (Figure 3). For the measurement of viscosity in the suspensions, a shear rate of 100 s −1 was used, a value commonly used in casting techniques. In this way, a value close to the maximum particle packing fraction was sought, in which a small increase in φ leads to a significant increase in viscosity. To find this value, in Figure 3, a tangent line to the curve of the relative viscosity increase was drawn in the region where → ∞ such that this line, intercepting the baseline (i.e., the minimum viscosity), allowed us to determine the values of the maximum optimal load (φm) of solids used in this work. The volumes of solids were 13.7%, 20.8%, and 19.2% for the suspensions containing BaSnO3, CaSnO3, and ZnSnO3 powders, respectively. The suspensions with optimal filler content were poured into plaster moulds, as described in Section 2.2.

Optimum Loading of Solids in the Suspensions
The amount of MSnO 3 powders (M = Ba, Ca, Zn) to be used to form the suspensions was determined by graphs of viscosity (η) vs. volumetric fraction of solids (ϕ) (Figure 3). For the measurement of viscosity in the suspensions, a shear rate of 100 s −1 was used, a value commonly used in casting techniques. In this way, a value close to the maximum particle packing fraction was sought, in which a small increase in ϕ leads to a significant increase in viscosity. To find this value, in Figure 3, a tangent line to the curve of the relative viscosity increase was drawn in the region where η r → ∞ such that this line, intercepting the baseline (i.e., the minimum viscosity), allowed us to determine the values of the maximum optimal load (ϕ m ) of solids used in this work. The volumes of solids were 13.7%, 20.8%, and 19.2% for the suspensions containing BaSnO 3 , CaSnO 3 , and ZnSnO 3 powders, respectively. The suspensions with optimal filler content were poured into plaster moulds, as described in Section 2.2.   Figure 4a shows the relative densities and the apparent porosity of the stannate ceramic bodies subjected to heat treatments at temperatures between 1000 and 1500 °C. The densification rates for the bodies were different and depended on the chemical nature of the stannate. The three systems achieved densifications greater than 75% from 1000 °C;  Figure 4a shows the relative densities and the apparent porosity of the stannate ceramic bodies subjected to heat treatments at temperatures between 1000 and 1500 • C. The densification rates for the bodies were different and depended on the chemical nature of the stannate. The three systems achieved densifications greater than 75% from 1000 • C; the ZnSnO 3 system obtained an appreciable increase in its relative density of~90% from 1300 • C, and the densifications for the BaSnO 3 and CaSnO 3 systems were greater than 90% after 1400 • C. The apparent porosity, an indirect measure of the estimation of the degree of porosity of the sintered bodies, decreased with increasing temperature; at 1300 • C, the AP% presented values around 11% for the three systems. These results, as well as the degree of densification of the sintered bodies, allowed the selection of the appropriate temperature to thermally treat the shaped bodies that were used as sensitive materials. The selection criteria for this temperature were, on the one hand, to have densification values that would allow manual adaptation of the ceramic body to the electrode system shown in Figure 1b, and, on the other hand, to obtain an apparent porosity with similar values between samples and that favours the interaction between the sensor surface and the test gas. Therefore, the selected sintering temperature was 1300 • C for the three systems; at this temperature, the desired degrees of densification and values between 60-70% for the volume of permeable pore space (voids) were obtained. The volume of permeable pore space (voids) is a porosity value that only takes into account permeable pores, those that dry and then become wet when the sample is placed in water and boiling; other pores it does not recognize. Therefore, the effect of the porosities present in the sensitive materials was studied by mercury porosimetry and nitrogen absorption-desorption (Supplementary Materials). The pore size distribution curves obtained by mercury intrusion porosimetry show that all of the sensitive materials (MSnO 3 (M = Ba, Ca, Zn)) sintered at 1300 • C presented a monomodal distribution, with average pore diameters ranging from 1600 to 5000 nm, as shown in Figure 4b.      Figure 5 shows the XRD patterns of the ceramic bodies sintered at 1300 °C manufactured from the suspensions of MSnO3 perovskite powders (M = Ba, Ca, Zn). The indexed diffraction peaks for each pattern correspond to BaSnO3, CaSnO3, and ZnSnO3, with JCPDS card Nos. 15-780, 31-312, and 28-1486, respectively. The Raman spectra of the ceramic bodies of the MSnO3 system (M = Ba, Ca, Zn) measured at room temperature are shown in Figure 6. For the BaSnO3 sample, bands at 258, 413, 455, 549, 650, and 831 cm −1 are observed, similar to those identified in the study by James et al. [60]. According to the XRD pattern presented above, only the crystalline phase of BaSnO3 is observed, with an ideal cubic structure belonging to the space group Pm3m (Oh). However, this should not show the first-order Raman active mode (this was The Raman spectra of the ceramic bodies of the MSnO 3 system (M = Ba, Ca, Zn) measured at room temperature are shown in Figure 6. For the BaSnO 3 sample, bands at 258, 413, 455, 549, 650, and 831 cm −1 are observed, similar to those identified in the study by James et al. [60]. According to the XRD pattern presented above, only the crystalline phase of BaSnO 3 is observed, with an ideal cubic structure belonging to the space group Pm3m (O h ). However, this should not show the first-order Raman active mode (this was predicted by group theory according to the irreducible representation 3F 1u (IR) + F 2u (silent) [60,61]). Therefore, the observed Raman bands can be assigned to the vibration modes of the SnO 6 octahedron, which has O h symmetry, in the distorted cubic structure of BaSnO 3 . The six fundamental vibrations of the SnO 6 octahedron are the symmetrical stretching mode v 1 A 1g , asymmetric stretching modes v 2 E g and v 3 F 1u , asymmetric bending mode v 4 F 1u , symmetric bending mode v 5 F 2g , and inactive mode v 6 F 2u [62]. In our study, the observed Raman activity can be attributed to defects that affect the translational periodicity of the crystal lattice in the sample, as well as oxygen vacancies.

Structural, Compositional, and Morphological Studies
In the Raman spectrum of the CaSnO 3 sample, the bands located at 165, 184, 229, 251, 280, 359, 445, and 703 cm −1 are in agreement with the previous studies of Maul et al. [63] and Redfern et al. [64], in which a harmonic analysis was performed. According to the authors, the bands between 100 and 300 cm −1 have a higher contribution of Ca 2+ and Sn 4+ atoms and correspond to vibrations of Ca-SnO 3 and O-Sn-O and a small contribution of apical O 2− . For frequencies greater than 300 cm −1 , the greatest contribution comes from oxygen ions, which are related to the torsion and stretching modes of SnO 3 and Sn-O [65], respectively.
Finally, the Raman spectra corresponding to the ZnSnO 3 sample show first-order vibrational active modes that agree well with the values reported at 477 cm −1 (E g ), 636 cm −1 (A 1g ), and 778 cm −1 (B 2g ) for the structure SnO 2 [66]. The most prominent Raman peak, located at 636 cm −1 , was also observed by Mayedwa et al. [67]; this peak corresponds to the distinctive Raman shift of ZnSnO 3 and is attributed to stretching vibrations of short M-O bonds, also coinciding with the XRD results reported in the same study with JCPDS card 28-1486. The weak bands at 501 and 700 cm −1 could correspond to infrared (IR) modes that can become weakly active when structural changes induced by disorder and size effects or even the presence of oxygen vacancies are introduced [68][69][70]. lent) [60,61]). Therefore, the observed Raman bands can be assigned to the vibration modes of the SnO6 octahedron, which has Oh symmetry, in the distorted cubic structure of BaSnO3. The six fundamental vibrations of the SnO6 octahedron are the symmetrical stretching mode v1A1g, asymmetric stretching modes v2Eg and v3F1u, asymmetric bending mode v4F1u, symmetric bending mode v5F2g, and inactive mode v6F2u [62]. In our study, the observed Raman activity can be attributed to defects that affect the translational periodicity of the crystal lattice in the sample, as well as oxygen vacancies. In the Raman spectrum of the CaSnO3 sample, the bands located at 165, 184, 229, 251, 280, 359, 445, and 703 cm −1 are in agreement with the previous studies of Maul et al. [63] and Redfern et al. [64], in which a harmonic analysis was performed. According to the authors, the bands between 100 and 300 cm −1 have a higher contribution of Ca 2+ and Sn 4+ atoms and correspond to vibrations of Ca-SnO3 and O-Sn-O and a small contribution of apical O 2− . For frequencies greater than 300 cm −1 , the greatest contribution comes from The chemical composition (the valence states of metal ions) of the surface of the ceramic bodies of the MSnO 3 system (M = Ba, Ca, Zn) was additionally investigated by XPS, as shown in Figure 7.
The binding energies (BEs) obtained from the XPS spectra were calibrated by referring to the C 1s signal at 284.8 eV, corresponding to adventitious physisorbed carbon oxide. High-resolution XPS spectra of the Ba-3d, Ca-2p, Zn-2p, Sn-3d, and O-1s regions are shown in Figure 7. The Ba-3d 5/2 and Ba-3d 3/2 BEs in the BaSnO 3 sample are located at 779.3 and 794.6 eV (Figure 7a), respectively, and separated by 15.3 eV, confirming the presence of Ba 2+ species [71]. The states of Ca-2p 3/2 and Ca-2p 1/2 at 346.1 eV and 349.6 eV (Figure 7b), respectively, whose separation from each other is 3.5 eV, are attributed to Ca 2+ [72,73] in the CaSnO 3 sample. The bands at 1021.7 and 1044.8 eV, separated by 23.1 eV, are assigned to the Zn-2p 3/2 and Zn-2p 1/2 of the Zn 2+ in ZnSnO 3 , respectively (Figure 7c) [74]. The spectra in Figure 7d reveal two peaks, at 486.3 and 494.7 eV with a separation of 8.4 eV, for the Sn-3d state in the stannates, which are attributed to Sn-3d 5/2 and Sn-3d 3/2 , respectively. The maximum separation of Sn-3d in this study (8.4 eV) coincides with that observed for the Sn-3d reported for SnO 2 [75][76][77]. The presence of Sn(II) can be ruled out due to the absence of subpeaks between the symmetrical peaks Sn-3d 3/2 and Sn-3d 5/2 as a result of deconvolution. The O-1s spectra (Figure 7d) show wide and asymmetric peaks and can be resolved into three fitting peaks in the 528-534 eV region, which has been observed in previous investigations of compounds with perovskite structures [30,31]. The first peak, at a lower BE of 529.3 eV in BaSnO 3 , 529.6 eV in CaSnO 3 , and 530.2 eV in ZnSnO 3 , is assigned to lattice oxygen (O L ), and the other two components can be attributed to oxygen vacancies (O V ) in the region between 531.2-531.5 eV and chemosorbed oxygen (O Chem ) in the band at 532.7 eV. The results suggest a significant proportion of oxygen vacancies with respect to the lattice oxygen that is similar for all samples (1.17 O V /O L for BaSnO 3 , 1.13 for CaSnO 3 , and 1.10 for ZnSnO 3 ). These oxygen vacancy defects trap electrons from the conduction band of the perovskites to form abundant adsorbed oxygen ions and to favour gas-sensing performance [78].
oxygen ions, which are related to the torsion and stretching modes of SnO3 and Sn-O [65], respectively.
Finally, the Raman spectra corresponding to the ZnSnO3 sample show first-order vibrational active modes that agree well with the values reported at 477 cm −1 (Eg), 636 cm −1 (A1g), and 778 cm −1 (B2g) for the structure SnO2 [66]. The most prominent Raman peak, located at 636 cm −1 , was also observed by Mayedwa et al. [67]; this peak corresponds to the distinctive Raman shift of ZnSnO3 and is attributed to stretching vibrations of short M-O bonds, also coinciding with the XRD results reported in the same study with JCPDS card 28-1486. The weak bands at 501 and 700 cm −1 could correspond to infrared (IR) modes that can become weakly active when structural changes induced by disorder and size effects or even the presence of oxygen vacancies are introduced [68][69][70].
The chemical composition (the valence states of metal ions) of the surface of the ceramic bodies of the MSnO3 system (M = Ba, Ca, Zn) was additionally investigated by XPS, as shown in Figure 7.   Figure 8 shows the SEM micrographs of the surface of the ceramic bodies formed with BaSnO 3 , CaSnO 3 , and ZnSnO 3 powders sintered at 1300 • C. In the three samples, a high porosity is observed. The BaSnO 3 sample has the smallest grain size (<500 nm) with a rhombohedral morphology (see Figure 8a). The largest grain size (~1 µm) was presented by the solids formed with CaSnO 3 and ZnSnO 3 powders, presenting spheroidal morphologies

Gas Sensing Properties
To investigate the sensing properties of the MSnO 3 porous bodies (M = Ba, Ca, Zn), the optimal operating temperatures (T op ) of the sensors were initially determined (i.e., the temperature at which the sensor response was highest). A temperature range of 180 to 300 • C was chosen to test the response of the sensitive materials to 80 ppm acetone, ethanol, and toluene. As a result of the tests, the average R a /R g response of each porous body exposed to acetone is shown in Figure 9a. The response of the ZnSnO 3 sensor increased appreciably, reaching a maximum at 270 • C, while the BaSnO 3 and CaSnO 3 sensors showed low response to temperature variation. When the samples were exposed to ethanol (Figure 9b), the BaSnO 3 and ZnSnO 3 sensors had a greater response with increasing temperature, reaching a maximum value at 270 • C. The CaSnO 3 samples produced a weaker response than the BaSnO 3 samples. In Figure 9c, with toluene gas, the sensors did not reach an optimal response in the range of working temperatures; however, the CaSnO 3 sensor presented a stronger response than the other sensors, although temperatures higher than the range studied here were required. Figure 9a,b show that the response curves for the ZnSnO 3 and BaSnO 3 sensors exposed to acetone and ethanol, respectively, increased with increasing working temperature and then gradually decreased. The relative optimal working temperature (T op ) can be explained by the cooperation of two opposite effects: (1) an increasing probability of activated detection reactions at low temperatures (e.g., for ZnSnO 3 to acetone, between 180 and 270 • C) and (2) an increasing probability of adsorbed gas molecules to desorb before the detection reactions occur when the temperature is higher (e.g., for ZnSnO 3 to acetone, over 220 • C) [79].

Gas Sensing Properties
To investigate the sensing properties of the MSnO3 porous bodies (M = Ba, Ca the optimal operating temperatures (Top) of the sensors were initially determined (i. temperature at which the sensor response was highest). A temperature range of 180 °C was chosen to test the response of the sensitive materials to 80 ppm acetone, eth and toluene. As a result of the tests, the average Ra/Rg response of each porous bod posed to acetone is shown in Figure 9a. The response of the ZnSnO3 sensor increase preciably, reaching a maximum at 270 °C, while the BaSnO3 and CaSnO3 sensors sh low response to temperature variation. When the samples were exposed to ethano ure 9b), the BaSnO3 and ZnSnO3 sensors had a greater response with increasing tem ture, reaching a maximum value at 270 °C. The CaSnO3 samples produced a weak sponse than the BaSnO3 samples. In Figure 9c, with toluene gas, the sensors did not an optimal response in the range of working temperatures; however, the CaSnO3 s presented a stronger response than the other sensors, although temperatures highe the range studied here were required. Figures 9a, b show that the response curves f ZnSnO3 and BaSnO3 sensors exposed to acetone and ethanol, respectively, increased increasing working temperature and then gradually decreased. The relative op working temperature (Top) can be explained by the cooperation of two opposite effec an increasing probability of activated detection reactions at low temperatures (e.g ZnSnO3 to acetone, between 180 and 270 °C) and (2) an increasing probability of ads gas molecules to desorb before the detection reactions occur when the temperat higher (e.g., for ZnSnO3 to acetone, over 220 °C) [79].  To test the selectivity of the sensors, they were exposed to fixed concentrations (80 ppm) of various volatile organic compounds, known as VOCs (acetone, ethanol, and toluene), among other gases (CO, H 2 , and NO 2 ). The responses of the MSnO 3 sensors (M = Ba, Ca, Zn) measured at T op = 270 • C were compared in Figure 9d. It can be clearly observed that the ZnSnO 3 sensor presented the strongest response in the presence of acetone gas. The BaSnO 3 sensor responded very well to ethanol gas, although less than ZnSnO 3 ; however, it suggested being more selective than ZnSnO 3 due to the low responses obtained for the other gases. The CaSnO 3 sensor, although it was not tested at its optimal temperature since that would be above 300 • C, showed a good response to toluene gas compared to the other VOCs.
The responses of the sensors based on porous bodies of MSnO 3 (M = Ba, Ca, Zn) showed superior gas detection of ethanol, toluene, and acetone. The typical curves of the response of the sensors to these gases as a function of the concentration are shown in Figure 10. There was a good linear relationship between the response and the concentration of tested gases. This linear fitting relationship provides an experimental basis for practical applications. Correspondingly, the sensitivity (i.e., the variation in the sensor response as a function of the change in the tested gas concentration from 5 to 80 ppm) showed higher values for ZnSnO 3 to acetone (~40% ppm −1 ), BaSnO 3 to ethanol (~16% ppm −1 ), and CaSnO 3 to toluene (~8% ppm −1 ). To test the selectivity of the sensors, they were exposed to fixed concentrations (80 ppm) of various volatile organic compounds, known as VOCs (acetone, ethanol, and toluene), among other gases (CO, H2, and NO2). The responses of the MSnO3 sensors (M = Ba, Ca, Zn) measured at Top = 270 °C were compared in Figure 9d. It can be clearly observed that the ZnSnO3 sensor presented the strongest response in the presence of acetone gas. The BaSnO3 sensor responded very well to ethanol gas, although less than ZnSnO3; however, it suggested being more selective than ZnSnO3 due to the low responses obtained for the other gases. The CaSnO3 sensor, although it was not tested at its optimal temperature since that would be above 300 °C, showed a good response to toluene gas compared to the other VOCs.
The responses of the sensors based on porous bodies of MSnO3 (M = Ba, Ca, Zn) showed superior gas detection of ethanol, toluene, and acetone. The typical curves of the response of the sensors to these gases as a function of the concentration are shown in Figure 10. There was a good linear relationship between the response and the concentration of tested gases. This linear fitting relationship provides an experimental basis for practical applications. Correspondingly, the sensitivity (i.e., the variation in the sensor response as a function of the change in the tested gas concentration from 5 to 80 ppm) showed higher values for ZnSnO3 to acetone (~40% ppm −1 ), BaSnO3 to ethanol (~16% ppm −1 ), and CaSnO3 to toluene (~8% ppm −1 ).    Initially, the responses were recorded by repeatedly exposing and purging to 80 ppm of the test gases (ethanol, toluene, or acetone) for three consecutive cycles. Then, the concentrations were varied from 60 to 5 ppm and again to 80 ppm. Reversible cycles were observed, in which the sensitive materials maintained their base electrical resistance (in air). Figure 11 also indicates the magnitude of the electrical resistance (plotted logarithmically), which was on the order of MΩ for the ZnSnO 3 sensor and GΩ for the BaSnO 3 and CaSnO 3 sensors. These magnitudes can be influenced by the bulk and grain boundaries of the porous bodies [80,81]. It should be noted that the three perovskites showed a reduction in electrical resistance when in contact with reducing gases such as acetone, ethanol, and toluene, which indicates n-type semiconductor behaviour [27,28]. indicates the magnitude of the electrical resistance (plotted logarithmically), which was on the order of MΩ for the ZnSnO3 sensor and GΩ for the BaSnO3 and CaSnO3 sensors. These magnitudes can be influenced by the bulk and grain boundaries of the porous bodies [80,81]. It should be noted that the three perovskites showed a reduction in electrical resistance when in contact with reducing gases such as acetone, ethanol, and toluene, which indicates n-type semiconductor behaviour. [27,28]. Considering the practical application of the sensors, the accuracy of long-term detection must be guaranteed. This stability was studied by measuring the average response of each sensor system exposed to the test gas at 80 ppm during the first 3 days and then at 7, 10, 14 and 120 days. As shown in Figure 11d, the long-term stability of the sensors had an almost constant response, which confirmed the high stability of sensors based on porous bodies of the MSnO3 system (M = Ba, Ca, Zn). The small variations in the response can be explained by the degree of passivation of the oxygen species adsorbed on the surface of each sensor, which over time can cause a difference in the resistance measured in air and the test gases (ethanol, toluene, and acetone).
Response time and recovery are important parameters of gas sensors and are considered when these devices are implemented in scenarios where real-time detection requires a "rapid response" according to the area of application. The response time (tres) was Considering the practical application of the sensors, the accuracy of long-term detection must be guaranteed. This stability was studied by measuring the average response of each sensor system exposed to the test gas at 80 ppm during the first 3 days and then at 7, 10, 14 and 120 days. As shown in Figure 11d, the long-term stability of the sensors had an almost constant response, which confirmed the high stability of sensors based on porous bodies of the MSnO 3 system (M = Ba, Ca, Zn). The small variations in the response can be explained by the degree of passivation of the oxygen species adsorbed on the surface of each sensor, which over time can cause a difference in the resistance measured in air and the test gases (ethanol, toluene, and acetone).
Response time and recovery are important parameters of gas sensors and are considered when these devices are implemented in scenarios where real-time detection requires a "rapid response" according to the area of application. The response time (t res ) was defined as the time required to reach 90% of the resistance difference after the injection of a test gas, and the recovery time (t rec ) was defined as the time required for the sensor to recover 90% of its resistance in air. The characteristics of the dynamic response of the sensors based on BaSnO 3 and ZnSnO 3 porous bodies were investigated and showed a rapid response to ethanol and acetone gases, respectively, operating at 270 • C, as seen in Figure 12 (normalized data).
Molecules 2022, 27, x FOR PEER REVIEW 15 of 23 defined as the time required to reach 90% of the resistance difference after the injection of a test gas, and the recovery time (trec) was defined as the time required for the sensor to recover 90% of its resistance in air. The characteristics of the dynamic response of the sensors based on BaSnO3 and ZnSnO3 porous bodies were investigated and showed a rapid response to ethanol and acetone gases, respectively, operating at 270 °C, as seen in Figure  12 (normalized data). The values tres and trec for the sensitive materials ZnSnO3 and BaSnO3 were calculated at 80, 20, and 5 ppm of the test gases at Top = 270 °C, as indicated in Table 1. The results in Table 1 show that the ZnSnO3 sensor exposed to acetone gas exhibited faster responses (between 4 and 117 s) than the BaSnO3 sensor exposed to ethanol gas (between 72 and 219 s). A clear difference was observed in the response and recovery values, which suggested the importance of the porosity factor since the response and recovery were faster for the ZnSnO3 samples with larger pores; the open pores were effectively filled with chemisorbed species. Table 1. Response (tres) and recovery (trec) times of the BaSnO3 and ZnSnO3 sensors exposed to different gas concentrations of ethanol and acetone, respectively, at operating temperature of 270 °C. Results based on Figure 12. 80  72  1385  4  1285  20  98  1125  23  1159  5  219  1068  117  696   Table 2 summarizes the results of the main detection performance parameters of the porous bodies based on MSnO3 perovskites (M = Ba, Ca, Zn) compared with those corresponding to some previously published studies based on MSnO3 structures (M = Ba, Ca, Zn). The sensors obtained in this study showed results comparable to those of previous studies considering the gas concentrations used in each study. It is important to mention that the reports in Table 2 generally used more complex instrumentation in the elaboration of the sensor (e.g., physical vapor deposition (PVD) sputtering process) with respect to The values t res and t rec for the sensitive materials ZnSnO 3 and BaSnO 3 were calculated at 80, 20, and 5 ppm of the test gases at T op = 270 • C, as indicated in Table 1. The results in Table 1 show that the ZnSnO 3 sensor exposed to acetone gas exhibited faster responses (between 4 and 117 s) than the BaSnO 3 sensor exposed to ethanol gas (between 72 and 219 s). A clear difference was observed in the response and recovery values, which suggested the importance of the porosity factor since the response and recovery were faster for the ZnSnO 3 samples with larger pores; the open pores were effectively filled with chemisorbed species. Table 1. Response (t res ) and recovery (t rec ) times of the BaSnO 3 and ZnSnO 3 sensors exposed to different gas concentrations of ethanol and acetone, respectively, at operating temperature of 270 • C. Results based on Figure 12.  Table 2 summarizes the results of the main detection performance parameters of the porous bodies based on MSnO 3 perovskites (M = Ba, Ca, Zn) compared with those corresponding to some previously published studies based on MSnO 3 structures (M = Ba, Ca, Zn). The sensors obtained in this study showed results comparable to those of previous studies considering the gas concentrations used in each study. It is important to mention that the reports in Table 2 generally used more complex instrumentation in the elaboration of the sensor (e.g., physical vapor deposition (PVD) sputtering process) with respect to the method reported here or, in some cases, for the materials based on powders, did not carry out a control for the forming of the sensor film. The sensitivity of the ZnSnO 3 samples prepared in this study was higher than those of the majority of previously published studies, even without incorporating second-phase materials such as Ag, Au, or SnO 2 , which are commonly used to enhance the sensitivity of intrinsic materials [11]. The ZnSnO 3 samples in this study also showed a short response time, although the recovery time was longer than those in other studies. In the case of the BaSnO 3 samples, the highest responses for similar ethanol concentrations could be connected to the higher operating temperatures (see Line 8, Table 2). It is also noted that the use of rare earth elements, such as La and Gd, in the works of Bhattacharya et al. [82,83] yielded significant improvements in the response and operating temperature of the BaSnO 3 sensors compared to those of the samples prepared in this work. No reports were found on the detection of toluene vapours using sensors based on CaSnO 3 .

Gas Sensing Mechanism
The gas detection mechanism of stannates and many MOx semiconductors is controlled by the surface [89,90]. The most accepted model for explaining the sensitivity of semiconductors to gases postulates that the resistance changes are due to the species and the amount of chemisorbed oxygen on the surface. When sensors based on n-type semiconductors such as BaSnO 3 and ZnSnO 3 are exposed to air, the electrical resistance of the material is controlled by the concentration of adsorbed oxygen species (O 2− , O − or O 2− ). Previous studies suggest that for working temperatures below 100 • C, the majority of oxygen ions exist in the form of O 2− ; in the range of 100-300 • C, O − ions are the stable oxygen species; beyond 300 • C, the dominant oxygen species is O 2− ions [86]. Generally, these oxygen species trap electrons and act as dispersion centres, effectively reducing the conductivity of the semiconductor. When the sensor was exposed to ethanol and acetone gases at temperature T op (270 • C), these gases reacted with the adsorbed oxygen species, reducing their concentration, and thus increasing the conductivity of the semiconductor, i.e., reducing the electrical resistance, as observed in the measurements shown in Figure 11a-c. The occurring reactions can be explained as follows: This indicates that the ethanol and acetone molecules are adsorbed on the surface of the porous body and react with the oxygen ions to produce CO 2 and H 2 O. In this process, the electrons are released back to the conduction band, which results in a substantial increase in the density of charge carriers on the surface. This reduces the width of the semiconductor depletion layer and the potential barrier height [86], and as a consequence, the resistance of the sensitive material. This process is developed in the same way at the reaction sites of the remaining surface of the sensitive material, including the pores that are occupied by the diffusion of the gas until stabilizing. When the sensors are re-exposed to an air environment, the target gas is desorbed from the surface of the material, and the oxygen captures electrons from the conduction band to form oxygen ions, which increases the width of the electron depletion layer. The resistance of the sensing material returns to the initial value, as shown in Figure 11a-c.
In this work, the shaped bodies of BaSnO 3 and ZnSnO 3 presented good performance due to their porous structure, showing average pore sizes at both the macro-and nanolevels. In particular, the ZnSnO 3 bodies were favourable because they had slightly larger macropores (5000 nm) than the BaSnO 3 (1600 nm) and CaSnO 3 (1740 nm) bodies, which would contribute to the diffusion of the gas in the bulk (volume) of the body. This difference in pore size was also observed, for example, in electron microscopy images of sections of the ZnSnO 3 and BaSnO 3 bodies, as seen in the Supplementary Materials ( Figure S1). In addition, the ZnSnO 3 samples also had pores with lower magnitudes (<10 nm in diameter), as did the BaSnO 3 and CaSnO 3 samples (Supplementary Materials Figure S2). Thus, the three bodies had a relatively similar surface area at the nanometric level. These characteristics, i.e., larger macropores and nanopores, also seemed to favour the response speed of the ZnSnO 3 bodies (see Table 1).
The ZnSnO 3 sensor also exhibited better responses than BaSnO 3 and CaSnO 3 in this study in terms of the operating temperature. Temperature plays an important role in the gas-solid interactions at the surface of sensitive materials. In this case, it was observed that the selected temperature (270 • C) for the tests with different gases and concentrations was closer to the optimal value of the ZnSnO 3 sensor (for acetone and ethanol) than that of the BaSnO 3 sensor (only for ethanol) and CaSnO 3 (which seems to need higher thermal stimulation for acetone, ethanol, and toluene); see Figure 9. Hence, the ZnSnO 3 sensors performed better for the tested VOCs.

The Influence of Humidity
Currently, the development of MOx sensors for the detection of gases is of great interest due to increasing demands for environmental protection and their potential use in the diagnosis of diseases. Therefore, these sensors are used in humid environments due to the presence of water vapour in the environment. Considering that humidity is a negative factor for gas-sensing properties, we sought to confirm the influence of humidity on the sensitivity of the BaSnO 3 and ZnSnO 3 sensors towards ethanol and acetone vapours, respectively. Figure 13 shows the resistance changes (plotted logarithmically) for the BaSnO 3 and ZnSnO 3 sensors at different concentrations of ethanol and acetone, respectively, in an atmosphere with humidity (10 and 30% RH) at an operating temperature of 270 • C. Clearly, an increase in RH caused the resistance values of the sensors to decrease, especially in air, affecting the response and recovery rates. This occurred because the water molecules reacted with the chemisorbed oxygen species (O − ) on the surface of the sensing material, which provided more electrons and reduced the reference resistance, resulting in a decrease in sensitivity [91]. This interaction in the ZnSnO 3 and BaSnO 3 sensors can be described by the following equations: Molecules 2022, 27, x FOR PEER REVIEW 18 of 23 resulting in a decrease in sensitivity [91]. This interaction in the ZnSnO3 and BaSnO3 sensors can be described by the following equations: H2O + O − (ads) + 2Zn → 2(Zn-OH) + e − (5) Figure 13. Dynamic response-recovery cycles in different relative humidity conditions. (a) BaSnO3 and (b) ZnSnO3 sensors, exposed for 10 min to various concentrations of ethanol and acetone, respectively, at operating temperature of 270 °C.
With respect to the previous observations and the consideration of the influence of OH species on the adsorption kinetics, we investigated the effect of humidity on the response and recovery rates of our sensors. Based on the responses of the sensors, the response and recovery times were calculated against different levels of RH with concentrations of 20 and 5 ppm of the target gases at 270 °C. Table 3 summarizes the results, indicating that under conditions of high humidity, the response takes longer because the water molecules adsorbed on the surface of the sensitive material can act as a barrier that prevents the adsorption of oxygen and gas molecules [92]. This results in a decrease in the active surface, reducing the sensitivity. In contrast, the recovery time was faster in humid than in dry conditions, specifically with BaSnO3 sensors. The faster recovery time may be due to an obstruction in the smallest pores by the OH molecules that interacted with the O − species, causing the target gas molecules to be located mainly on the surface and not internally (in the bulk) and making their desorption faster and less difficult than when no humidity was introduced (Table 1). Table 3. Response (tres) and recovery (trec) time of the BaSnO3 and ZnSnO3 sensors exposed to different gas concentrations of ethanol and acetone, respectively, in humid atmosphere, at operating temperature of 270 °C. Results based on Figure 13.  With respect to the previous observations and the consideration of the influence of OH species on the adsorption kinetics, we investigated the effect of humidity on the response and recovery rates of our sensors. Based on the responses of the sensors, the response and recovery times were calculated against different levels of RH with concentrations of 20 and 5 ppm of the target gases at 270 • C. Table 3 summarizes the results, indicating that under conditions of high humidity, the response takes longer because the water molecules adsorbed on the surface of the sensitive material can act as a barrier that prevents the adsorption of oxygen and gas molecules [92]. This results in a decrease in the active surface, reducing the sensitivity. In contrast, the recovery time was faster in humid than in dry conditions, specifically with BaSnO 3 sensors. The faster recovery time may be due to an obstruction in the smallest pores by the OH molecules that interacted with the O − species, causing the target gas molecules to be located mainly on the surface and not internally (in the bulk) and making their desorption faster and less difficult than when no humidity was introduced (Table 1). Table 3. Response (t res ) and recovery (t rec ) time of the BaSnO 3 and ZnSnO 3 sensors exposed to different gas concentrations of ethanol and acetone, respectively, in humid atmosphere, at operating temperature of 270 • C. Results based on Figure 13.

Conclusions
In the present study, highly uniform porous bodies based on MSnO 3 perovskites (M = Ba, Ca, Zn) were formed using the slip casting technique. Considering the structural advantages of the shaped bodies, the porous surfaces of the sensing materials had a large number of adsorption sites for gas molecules and were favourable for the diffusion of these molecules. The sensitivities of the BaSnO 3 , CaSnO 3 , and ZnSnO 3 sensors to ethanol, toluene, and acetone were 16, 8, and 40% ppm −1 , respectively, at an operating temperature of 270 • C, where the ZnSnO 3 sensor exhibited the shortest response times (e.g., response time and recovery of 4 s and 1285 s, respectively, for 80 ppm acetone) as well as excellent medium-term stability (120 days). The improved gas detection properties of porous bodies, particularly ZnSnO 3 , with respect to other previously reported studies can be attributed to the porous structure, pores at both the macro (~5000 nm) and nano (<10 nm) levels, favouring the surface area and surface defects of the sensitive material.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27092889/s1. Figure S1: Scanning electron microscopy images and illustration of the gas sensing mechanism of the (a) BaSnO 3 and (b) ZnSnO 3 sensors; Figure S2