II. EXPERIMENTAL

To obtain BaSnO₃ with the Pechini method, a mixture of ethylene glycol and citric acid was first heated to 70 °C, as indicated in a previous paper.^{Reference Ochoa-Muñoz, Ponce and Rodríguez-Páez21} The precursors C₄H₆BaO₄ and SnCl₂·2H₂O, in the amounts 8.4 and 7.4 g, respectively, were added to the ethylene glycol and citric acid solution. At room temperature, the resulting solution was brought to a basic pH and heated to 140 °C, stirring constantly until a resin was formed. The resin was then precalcined at 350 °C and the precalcined material was heat treated within a range of temperatures (T ≤ 1000 °C) to obtain the BaSnO₃. The powders obtained at the end of the synthesis processes were characterized by x-ray diffractograms (XRD) and SEM.

To obtain BaSnO₃ by the controlled precipitation method, SnCl₂·2H₂O and C₄H₆BaO₄ were used as precursors. These were dissolved separately in 0.1 M aqueous HNO₃ solutions and the pH value was increased up to a basic pH by adding NH₄OH. The obtained suspensions were then mixed and left to age for 24 h. Finally, the suspension was washed with an aqueous solution of diethylamine. The wet solid obtained was dried at 100 °C and subsequently treated at 1250 °C for 2 h. After the synthesis process, the powders were characterized using IR Spectroscopy, XRD and SEM (JEOL 5910LV; JEOL Ltd., Tokyo, Japan). To obtain the IR Spectrum, a Thermo Nicolet IR200 (Thermo Fisher Scientific, Pittsburgh, Pennsylvania) was used. The x-ray diffractograms were obtained using the X TEL-X-OMETER TEL-580 with a Cu source (λ = 1.54 Å) and a 30 kV voltage (Tel-Atomic, Jackson, Michigan).

A paste was then prepared using an organic binder (glycerol) and the powders were synthesized. Porous thick film samples were made by painting onto a 96% dense insulating alumina substrate and were placed in a vacuum chamber at its base pressure of 10⁻⁶ mm Hg. Electrodes of platinum or gold, with an interdigitated shape, were then deposited by sputtering. The electrode was a platinum 200 nm film and was deposited over a previously deposited 25 nm Ti layer.^{Reference Aldao, Schipani, Ponce, Joanni and Williams22} For defining the electrodes, the substrates with the metal films were placed in a homebuilt micromachining system with a 20 W, frequency tripled Nd:YVO₄ laser (λ = 355 nm). The interelectrode distance was 20 μm. After painting, samples F _p (Pechini sample) and F _cp (controlled precipitation) were heat treated for 1 h in air at 100 °C. This treatment was carried out to evaporate the organic binder and to improve the film adhesion to the alumina substrate. Later, the samples were preheat treated up to 380 °C using a heating rate of 1 °C/min. The average thickness of the films was 0.1 mm. According to the geometry of our sensors, and taking into account that the electrical path in the interelectrode distance is shorter than that present everywhere else, an increase in the sensor thickness will only increase the response time, since it will take longer for gases to reach the interdigitated path.

The BaSnO₃ film was placed in the sample holder cell of a static flow system (Fig. 1). The electrical conductivity behavior of the films, as function of temperature, was determined with a multimeter (Agilent 34401A; Keysight Technologies, Palo Alto, California). Conductivity versus temperature measurements were carried out (for the F _p and F _cp films) raising and then decreasing the temperature of the samples in the range 25–200 °C at a rate of ∼1 °C/min with the sample first kept in vacuum (10⁻⁴ mm Hg). To gain confidence in sensor properties, resistance (at 20 Hz) versus time curves were measured while changing the vacuum (10⁻⁴ mm Hg) to an O₂ or CO atmosphere (at different O₂ and CO concentrations of 2, 4, 6, 8, 10, 20, and 40 mm Hg) and, after having reached quasisaturation, impedance was measured in a frequency interval of 20 Hz–1 MHz using an impedance analyzer HP4284A. Later, the samples were exposed again to vacuum. The sensitivity of the BaSnO₃ films was determined from 230 to 540 °C. It was calculated as, the resistance of the sensors in a gaseous atmosphere divided by their resistance in vacuum.

FIG. 1.

Diagram of the equipment used to determine the gas sensing capacity of sensing films.

III. CHARACTERIZATION OF SYNTHESIZED POWDERS

A. XRD measurements

The diffractograms of Fig. 2 correspond to the solids obtained with the Pechini method, treated at 1000 °C, and to the samples obtained by controlled precipitation method, treated at 1250 °C for 2 h, since for treatments below 1000 °C these samples have primarily the crystalline phase of SnO₂ but, when treated at higher temperatures, specifically at 1250 °C, crystallization yields pure BaSnO₃ as the main phase.^{Reference Ochoa-Muñoz, Ponce and Rodríguez-Páez21} Both samples show BaSnO₃ (PDF 15-780) as the main crystalline phase.

FIG. 2.

XRD diffractograms of Ba–Sn system solid samples synthesized by the controlled precipitation and the Pechini methods, heat treated at 1250 and 1000 °C, respectively, for 2 h.

B. SEM images

Figure 3 shows SEM micrographs corresponding to BaSnO₃ thick films formed with ceramic powders obtained using controlled precipitation (a), and the Pechini method (b). Both samples were heat treated at 380 °C. In the case of the F _cp film there is a very wide distribution in the grain size, and the images demonstrate the presence of grains larger than 5 μm [Fig. 3(a)]. With the F _p film, the grains are more uniform in grain size [Fig. 3(b)]. It can also be seen that the F _cp film has a grain size of ∼3.5 μm, greater than the ∼1.5 μm for the F _p film.

FIG. 3.

SEM photographs of BaSnO₃ thick porous films at different magnifications. (a) F _cp film and (b) F _p film. Both films were made by the screen printing technique.

C. Fourier transform infrared spectroscopy (FTIR)

To obtain information on the evolution of the functional groups during crystallization of the ceramic powder, the dry powders from each synthesis method were characterized using IR spectroscopy. Figure 4 shows IR spectra corresponding to samples obtained by the controlled precipitation method, heat treated at 100, 500, 800, 1000, and 1250 °C. The results indicate the presence of hydroxyls in the solid, bands around 3445 cm⁻¹ showing the presence of absorbed water in the samples, and ${\rm{N}}{{\rm{O}}_3}^ -$, a band located at 1361–1395 cm⁻¹.^{Reference Ochoa-Muñoz, Ponce and Rodríguez-Páez21,Reference Socrates23} The most evident band is that located at ∼630 cm⁻¹, corresponding to a vibrational mode of the Sn–O group of the ${\rm{Sn}}{{\rm{O}}_{\rm{3}}}^{2 - }$,^{Reference Ochoa-Muñoz, Ponce and Rodríguez-Páez21,Reference Amalric-Popescu and Bozon-Verduraz24} and that might be used to demonstrate the existence of BaSnO₃. The band at 1422 cm⁻¹ is associated with BaCO₃,^{Reference Ochoa-Muñoz, Ponce and Rodríguez-Páez21,Reference Socrates23,Reference Nakamato25} so that its intensity decreased at high temperatures. This band appeared in most of the spectra indicating that the BaCO₃ was found in the samples as an unwanted additional phase, which was required to be removed.

FIG. 4.

IR spectra corresponding to solid samples of the SnCl₂–Ba(CH₃COO)₂–H₂O–NH₄OH system, controlled precipitation method, heat treated at different temperatures.^{Reference Ochoa-Muñoz, Ponce and Rodríguez-Páez21}

Figure 5 shows the IR spectra corresponding to solid samples obtained by the Pechini method, calcined at 350 °C and then heat treated at 500–1000 °C. The spectra show a band at 3440 cm⁻¹ that can be attributed to the vibrational mode of the tension in the H–OH, while the band at 1630 cm⁻¹ corresponds to the flexing mode of the water. The band at 1430 cm⁻¹ may be associated with the vibration of residual organic C–H groups.^{Reference Socrates23,Reference Nakamato25} The band at ∼630 cm⁻¹ seen in the spectra corresponds to Sn–O.^{Reference Amalric-Popescu and Bozon-Verduraz24} This band, considered to be a characteristic of BaSnO₃, is most evident in the treatment at 1000 °C.

FIG. 5.

IR spectra corresponding to solid samples of the SnCl₂–Ba(CH₃COO)₂–H₂O–NH₄OH system, Pechini method, heat treated at different temperatures.

IV. MEASUREMENTS AND RESULTS

The sensitivity of BaSnO₃ thick films was determined from 230 to 540 °C. Sensitivity was calculated as, the resistance of the sensors in a gaseous atmosphere divided by their resistance in a vacuum. The concentrations of O₂ and CO were varied between 2 and 40 mm Hg for both types of samples, F _p and F _cp. No further changes in the electrical behavior were found during exposure to gas concentrations greater than 40 mm Hg, thus 40 mm Hg was chosen to be the highest concentration studied. Figures 6 and 7 show the sensitivity of the sensors to O₂ and CO, respectively. Although, both samples show a good sensitivity to O₂ at 300 °C, F _p presents a better response (an order of magnitude greater than that of F _cp). For the F _p samples the sensitivity increases with the gas concentration in the range of 230–540 °C.

FIG. 6.

Sensitivity curves of different oxygen concentrations, (a) F _p film and (b) F _cp film.

FIG. 7.

Sensitivity curves for different CO concentrations, (a) F _p film and (b) F _cp film.

From Fig. 6, we can conclude that the films made with powders obtained using the Pechini method present a better reproducibility, while the best operating temperature, for oxygen detection, is around 300 °C. This represents a notable difference found with respect to other authors who have reported higher temperatures for maximum sensitivity to O₂.^{Reference Cerda, Arbiol, Dezanneau, Díaz and Morante2,Reference Aldao, Schipani, Ponce, Joanni and Williams22}

To analyze the electrical behavior of the F _p samples, it will be assumed, as is generally accepted, that barrier formation between grains is responsible for sensor conductivity and that these barriers have a Schottky-type nature. As a first approach, for barrier height calculations, researchers have regularly considered a thermionic mechanism that leads to a conductivity of the type

(2)$$G = {G_0}\,{\rm{exp}}\left( {{{e\varphi } \mathord{\left/ {\vphantom {{e\varphi } {kT}}} \right. \kern-\nulldelimiterspace} {kT}}} \right)\quad ,$$

where eφ is the barrier height, T is the temperature, and k is the Boltzmann constant. This equation reflects an activated process due to intergranular barriers. As is commonly done, barrier height is estimated by plotting conductance (ln G) versus the reverse function of temperature (1/T). Thus, an activation energy (E _a) can be obtained from the slope of a fitting straight line.

Figure 8 shows the conductance in vacuum conditions of F _p samples versus the inverse of temperature, in cycling experiments. Specifically, conductance was measured by raising and then decreasing the temperature in vacuum, reaching about 200 °C. In this temperature range, the room temperature conductance value is recovered after the cycle implying that no oxygen adsorption/desorption takes place during the heating and cooling processes.

FIG. 8.

Arrhenius plot for conductivity of the film in vacuum. The sample was heated to 200 °C and then cooled in vacuum to ensure no diffusion into the grains. An effective activation energy of 0.52 eV was calculated.

At high enough working temperatures, gas in–out diffusion can take place.^{Reference Batzill and Diebold26} Indeed, it has been detected by the authors in previous works that an oxygen rich atmosphere alters the density of vacancies in tin oxide for temperatures greater than 300 °C.^{Reference Ponce, Aldao and Castro27–Reference Malagu, Carotta, Giberti, Guidi, Martinelli, Ponce, Castro and Aldao29} As a result of this diffusion process, the density of vacancies changes and from this point the slope of the conductivity curve can no longer be associated with the barrier height.^{Reference Schipani, Aldao and Ponce30,Reference Ponce, Castro and Aldao31} Accordingly, the effective activation energy obtained was ${E_{{F_{\rm{p}}}}} = 0.52\;{\rm{eV}}$ for the F _p sample. The activation energy value was found with the sample in vacuum and considering that (i) no gas adsorption/desorption takes place during the heating and cooling processes, and that (ii) no oxygen diffusion into or out of the grains occurs during the heating or cooling within the temperature range of the study.

For nonoverlapped potential barriers, the intergrain capacitance is that of two Schottky barriers in series. In a one-dimensional approximation, the resulting capacitance when barriers are nonoverlapped, C _no, related to that corresponding to a sample with overlapped barriers, C _o, can be expressed as

(3)$${{{C_{{\rm{no}}}}} \over {{C_{\rm{o}}}}} = {z \over {2\Lambda }}\quad ,$$

where z is the grain size and Λ is the depletion layer width. This capacitance enlargement can be interpreted, using Eq. (3), as the consequence of depletion widths that are narrower than the grain radius due to a higher doping.

To gain confidence in the present studies, capacitance measurements were carried out at 550 °C in an oxygen atmosphere and in vacuum. An atmosphere rich in oxygen will decrease the amount of oxygen vacancies, widening the potential barriers. This can lead to overlapped potential barriers. In this case, capacitance is lower and has a mild dependence with bias, as observed. Thus, from Eq. (3) and for an average grain size of 1 µm, the depletion layer can be estimated as

$$\Lambda = {z \over 2}\left( {{{{C_{\rm{o}}}} \over {{C_{{\rm{no}}}}}}} \right) \cong 1.63 \times {10^{ - 7}}\;{\rm{m}}\quad {\rm{.}}$$

On the other hand, in a Schottky barrier, the height (ϕ) and width (Λ) are directly related^{Reference Sze32,Reference Kao33} :

(4)$$\phi = {{e{N_{\rm{d}}}} \over {2\varepsilon }}{\Lambda ^2}\quad ,$$

where N _d is the donor density. Using the activation energy previously calculated and the depletion layer width, donor density can be estimated, N _d = 3.9 × 10²² 1/m³.

The conductance under vacuum was measured for F _cp samples. The conductance of the sample was much lower compared to that of F _p; this is not expected as grains are three times bigger on average than in the F _p sample, so the conductance should have been higher. We present below a possible explanation for this finding. Even so, samples F _p and F _cp were both exposed to a CO atmosphere. Only films formed using powders obtained with the Pechini method (F _p) responded appreciably to CO. The maximum sensitivity was reached at different times, depending on the working temperature. The response time at the highest sensitivity point was reached in 3–4 min.

To analyze the sensor behavior when samples were exposed to a CO atmosphere, measurements with different concentrations were obtained for samples F _p and F _cp from 230 to 540 °C. In the following step, CO (from 2 to 40 mm Hg) was incorporated into the gas chamber. Only the samples prepared using the Pechini method presented an appreciable sensitivity and reproducible responses. The maximum sensitivity value was detected at 450 °C with no change for CO concentrations greater than 40 mm Hg.

It is known that carbon monoxide reacts with the oxygen adsorbed on the tin oxide surface according to these equations,^{Reference Ponce, Malagu, Carotta, Martinelli and Aldao34}

$${{\rm{O}}_2}\left( {\rm{g}} \right) + 2( \equiv {\rm{S}}) + 2{\rm{e'}} \leftrightarrow 2{{\rm{O'}}_{{\rm{ads}}}}\quad ,$$

$${{\rm{O'}}_{{\rm{ads}}}} + {\rm{CO}}\left( {\rm{g}} \right) \leftrightarrow {\rm{C}}{{\rm{O}}_{{\rm{2ads}}}} + {\rm{e'}}\quad {\rm{,}}$$

$${\rm{C}}{{\rm{O}}_{{\rm{2ads}}}} \leftrightarrow {\rm{C}}{{\rm{O}}_2}\left( {\rm{g}} \right) + ( \equiv {\rm{S}})\quad ,$$

where (≡S) corresponds to a site on the surface where oxygen can be adsorbed. This reduction in the amount of oxygen makes the sample conductivity higher. This means that the barrier height and the depletion width (Λ) decrease if oxygen is previously adsorbed on the surface.

In Table I, a clear distinction can be seen between sensitivities corresponding to F _p and F _cp samples. For the Pechini-made films (F _p), the high sensor response to oxygen at 300 °C contrasts with the low response to CO at that temperature. Conversely, at 450 °C the high sensitivity is observed to CO and the low one to oxygen. The F _cp sensor meanwhile shows almost no difference in sensitivity between gases at these temperatures. This lack of selectivity and low sensitivity could be explained taking account of the fact that the controlled precipitation, having a 3.5 μm average grain size, could not allow grain boundaries to reach complete equilibrium with the atmosphere, even at the highest temperatures. Oxygen could not in/out-diffuse to reach the grain boundary completely, in a way that the inner part of the interface would remain unaltered in spite of a changing atmosphere. This “trapped oxygen” hypothesis can also explain the lower conductance of the F _cp sample.

TABLE I.

Sensitivity comparison between F _p and F _cp samples at 40 mm Hg of oxygen and CO.

Interestingly, an anomalous response was found at temperatures greater than 250 °C for F _p samples exposed to CO. A great increase in the electrical resistance value for exposure times greater than 10 min was observed, as shown in Fig. 9. This striking resistance response could be due to a catalytic reaction of CO produced by the platinum in contact with the film. To test this hypothesis, substrates with Au contacts were prepared by following the same procedure as that used for Pt contacts. The anomalous effect was not found in samples prepared with Au contacts (Fig. 10).

FIG. 9.

Sensitivity curves at three different temperatures for F _p samples with platinum electrodes. This anomalous effect was observed above 250 °C at all CO concentrations tested. At 300 °C, the film resistance remains almost constant for 10 min, and then it increases very rapidly.

FIG. 10.

Sensitivity curves of F _p samples with gold and platinum electrodes. The anomalous response only appears when platinum electrodes are used.

During the CO exposure for Au electrodes, no changes in the resistance were found for films prepared with either the controlled precipitation method or the Pechini method. Figure 10 shows the electrical resistance behavior when F _p samples (interdigitated electrodes of Au and Pt) were exposed to 10 mm Hg at 300 °C.