As evaluated by SEM pictures, the morphology of Cu-SnO₂, WO₃ and In₂O₃ films presented in Figure 1 shows a porous morphology comprising aggregates with uniform size distribution (2–5 μm for Cu-SnO₂ and about 10 μm for In₂O₃). The WO₃ films show variations in the size of the aggregates from 2 μm to 10 μm and this can be due to the lower deposition temperature (350 °C) compared to the deposition temperature of the other films (400 °C). The WO₃ films deposited at 400 °C have a more developed porosity than the films prepared at 350 °C, but the adhesion to the substrate was poor, hence, a compromise had to be accepted. The morphology of the films deposited at a certain temperature depends mainly on the rate of evaporation, spreading, precipitation and decomposition reaction. For this reason different deposition parameters has been used for the preparation of the films (Table 1) in order to obtain porous morphology which plays an important role in the adsorption of the gas molecules [19]. The thickness of the films varies from 7 to 10 μm as determined by film cross-section.

Supplementary details about the deposition optimization process of the films are described elsewhere [20-22].

The topographic 3D views of Cu-SnO₂, WO₃ and In₂O₃ films are shown in Figure 2. The pictures, realised on a surface of 1 μm per 1 μm, provide information about the shape, the size of the grains and their distribution in the aggregates. Furthermore, by means of appropriate software the mean roughness (Ra) can be calculated. It can be seen that all the films have porous morphology comprising grains with sizes ranging from 100 to 250 nm. The roughness of the Cu-SnO₂ seems to be the highest (21 nm), followed by In₂O₃ (17 nm) and WO₃ (7 nm). Hence, no major differences between the films can be noted at this scale.

For a more precise evaluation of the roughness of the films, which could eventually provide a better comparison between the samples, it would be preferable to scan a larger surface. Unfortunately, this could not be successfully accomplished due to the specific morphology of the films, which present cauliflowers-like aggregates that are not connected between them at the surface, so, differences in the height of the aggregates and the “space” between them brought signal instabilities and consequently no good quality pictures could be obtained.

The crystallinity of the films was analysed by the XRD technique, as presented in Figure 3. The films exhibit crystalline phases with peaks corresponding to tetragonal rutile (Cu-SnO₂), monoclinic (WO₃) and cubic (In₂O₃) phases [20-22]. No supplementary peaks (except those of the substrate) are detected, proving the purity of the films. The average crystallite sizes are calculated according to Debye-Scherrer formula [23] and are about 7–10 nm for Cu-SnO₂ and 20–30 nm for WO₃ and In₂O₃. The small crystallite size of Cu-SnO₂ is also suggested by their broad XRD peaks. For gas sensor applications the size of the crystallite is very important since improved sensitivity has been reported for materials which have crystallite size similar to twice of the space charge layer (2L) [12]. In our case the Cu-SnO₂ has the closest value to 2L (6 nm).

Supplementary information concerning structural homogeneity of the films was provided by Raman spectroscopy studies. In Figure 4 the Raman spectra and cartographies of Cu-SnO₂, WO₃ and In₂O₃ are shown. All the present peaks are indexed to rutile SnO₂ [24], monoclinic WO₃ [25] and cubic In₂O₃ [26,27] in good accordance with the literature. Furthermore, these results validate the XRD ones with respect of the phase crystallization. To obtain the Raman cartographies approximately 100 spectra were collected for each film on a surface of 120 μm × 120 μm. The image was further obtained by the integration of the highest peak (632 cm⁻¹ for Cu-SnO₂, 810 cm⁻¹ for WO₃ and 307 cm⁻¹ for In₂O₃) in each studied point.

It can be seen that the films are quite homogeneous in structure but they also present variations in the peak intensities which can be perhaps related to their morphology (especially to the grain size) and to the crystallinity. This type of Raman mapping gives an overall view of the surface structure of a material which is advantageous compared to the literature [28] where generally only one Raman spectrum is presented and sometimes the evaluation of a small structural variation is delicate to be exploited.

Figure 5a presents the response (R_air/R_H2s) of the films to 10 ppm H₂S as a function of the operating temperature. The maximum response (2500) is showed by Cu-SnO₂ films at 100 °C. At this temperature the other films exhibit low response (about 6). They show better response (1200 for WO₃ and 75 for In₂O₃) than Cu-SnO₂ (13) at higher temperature (200 °C).

Comparing to the literature, the Cu-SnO₂ prepared in this work presents higher response at lower operating temperature than some other Cu-SnO₂ materials fabricated with different techniques [8]. In the same way, the performances of WO₃ are superior to those reported in [29]. On the contrary, the response obtained in [10] is higher than that of our films, but with the disadvantage of a recovery process possible only by applying heating pulses at 250 °C. Concerning the In₂O₃ films in the detection of H₂S, there are just few dedicated papers in the literature [30-32], and we can say that our film response is not remarkable.

The differences between the film responses studied in this work can be due to the nature of the studied oxide and their morphological and structural properties. The accepted mechanism for the detection of reducing gases consists in the reaction of chemisorbed oxygen species ( O 2 −, O⁻, O²⁻) on the film surface and the gas molecules (H₂S). As a consequence of this reaction the electrons are released, the film resistance decreases (R_H2s) and the response is improved. In addition, for Cu-SnO₂ the influence of the copper dopant has to be taken into account since without dopant the sensitivity of SnO₂ is very low [33]. In this direction several authors have proven by different techniques [34,35] the formation of CuS due to the reaction between Cu or CuO with H₂S. The CuS metallic nature (low resistance) allows great sensor response to be achieved. The choice of a proper dopant for WO₃ and In₂O₃ could beneficially improve their performance.

The reproducibility of the film response at their optimum operating temperature (Figure 5b) shows very good and satisfactory values for WO₃ and Cu-SnO₂, respectively. The In₂O₃ films present significant response differences between successive gas exposure cycles and this can be due to the incomplete recovery of the film resistance within the selected period of 60 minutes air purging. A good recovery of the resistance and implicitly of the response was be achieved by a long period (one night) of air exposure. Long recovery times for In₂O₃ have also been reported by other authors [36,37].

The ability of detecting a desired gas from a mixture of gases is an important performance property of a sensor. For Cu-SnO₂ the 100 °C optimum temperature was selected to study the cross-sensitivity to other reducing (SO₂) and oxidising (NO₂) gases, while for WO₃ and In₂O₃ the 200 °C temperature was used. The selection of the temperature corresponds to the temperature offering the maximum response to H₂S. As highlighted in Figure 6, all the films present significantly higher response for H₂S than for NO₂ or SO₂. Hence, we can affirm that the films can provide selectivity in an eventually H₂S detection from a mixture comprising the three studied gases (H₂S, NO₂ and SO₂).

The films (Cu-SnO₂, WO₃ and In₂O₃) were deposited on Pt partially coated alumina using the ESD technique as presented in Reference [22]. The parameters used for the deposition of the films are described in Table 1. The precursors were dissolved in ethanol in order to obtain a 0.05 M solution which was atomisated by applying a high voltage (7-8 kV) between a metallic nozzle and a heated and grounded substrate. The formed aerosol (spray) comprises highly charged droplets which are directed to the substrate under an electrostatic force. At the impact, the droplets loose their charge, and then spreading, drying, and decomposition of the precursor solution occur. In this way a thin layer was formed on the substrate surface and to improve their crystallinity annealing treatment was performed. Several films of each composition have been prepared to validate the reproducibility of film deposition in terms of morphology, structure, homogeneity, and gas responses.

The film morphology was studied using a JOEL JSM 580LV scanning electron microscope and the topography was evaluated with an atomic force microscope (Nanoscope IV-Multimode Veeco Instruments, USA) operating in a tapping mode regime. Raman measurements were performed at room temperature using a Jobin-Yvon microspectometer having a He-Ne excitation source (wavelength 632.8 nm). The sample surface was visualized by an optical microscope which allowed the selection of specific zone for structure evaluation. Raman spectra were recorded by scanning a part of the sample surface (120 μm × 120 μm) in 10 μm steps in both X and Y-axes. By integration of the desired Raman peak, the structural cartography of the surface was obtained.

The gas sensing measurements were carried out in a closed quartz tube furnace. The temperature was measured using a type K thermocouple and was controlled by a PID temperature regulator (JUMO dTRON 16.1). The resistance measurements were performed using a two-point probe method with an electrometer (KEITHLEY 6514). The gas response of a film (S) is defined as the ratio of R_air/R_gas for H₂S and SO₂ (reducing gases) and R_gas/R_air for NO₂ (oxidizing gas), where R_air represents the electrical resistance of the film in synthetic air and R_gas represents the resistance during the gas exposure.

To determine the temperature at which the H₂S response was maximum, the films were exposed to 10 ppm H₂S in a temperature range of 100 to 300 °C. Firstly the temperature was set to the desired value and the films were purged with synthetic air for 60 min. Next, the films were exposed to H₂S for 30 min, followed by regeneration with synthetic air for 60 min. This cycle was repeated about four times during a day (in order to ensure the reproducibility). The responses of the films to 20 ppm SO₂ and 1 ppm NO₂ were evaluated at the selected operating temperature (temperature at which the film shows the maximum response).

The concentration of gases was fixed by adjusting the flow rates of a target gas and a carrier gas (synthetic air) in the way to maintain a constant total flow rate of 100 mL/min. The gas bottles were provided by Air Products (France) and the concentration of the gases was controlled using mass flow controllers (MFC, BROOKS Instruments, 5850 TR).