Figure 2 shows the XRD patterns of mixed xWO₃(1-x)Y₂O₃ powders for 2θ = 20°∼65°. The main peaks for all samples are the same as the x is changed (matched with 1-0486, 2-0308, 5-0363, 44-0357, and 44-0399 JCPDS file numbers). A higher level of doping (x ≤ 0.85) shows a significant change in chemical composition as Y₆WO₁₂ is produced. Furthermore, the distance between similar atomic planes (d-spacing) in WO₃ and Y₂O₃ is changed with Δd variation less than 10% as x is decreased. Maximum d-spacing of each sample occurs at a mean peak of XRD pattern and also can be determined from the Bragg Equation (nλ = 2dsinθ for n = 2, Bragg angel = θ, and λ = 1.542 Å). In the case of x = 1 and at θ = 23.184°, d-spacing from the XRD result is 3.8293 Å, whilst calculated d-spacing from the Bragg Equation is 3.9168Å.

The average of crystallite size was determined from XRD results based-on the Scherrer Equation (Scherrer Constant, K = 0.94 for spherical crystals with cubic symmetry [28] and λ = 1.542 Å from XRD results): (1.a) FWHM = K λ L   cos θthen: (1.b) L = K λ ( FWHM )   cos θ

Since K and λ are constant, the crystallite size is only related to the position (2θ) and the Full-Width Half-Maximum (FWHM) value of the peaks. Table 1 shows the maximum, minimum, and average crystallite sizes of the samples determined from Equation (1.b). Note that the calculation was carried out for the first 12 peaks of the XRD results. Variations into the crystallite sizes in respect to Y₂O₃ concentration can be interpreted as the changes into FWHM values and a shift into position angel of the peaks. Figure 3 shows that the slope of the XRD patterns as well as peak positions are changed due to the level of dopant. Increasing the slope leads to a higher value of integral breadth, this in return results in an increased FWHM value. Decrease of x, resulted in slightly displacement of the peak position, 2θ, to higher values, which in return causes a reduction of cosθ. However, this reduction is not significantly changed the value of crystallite size, L, in the equation (1.b). Furthermore, the value of shift in the position of peaks is not unique for the whole pattern and varies from one peak to another. The maximum displacement of the peak position observed, with respect to the basic powder, happened at the fourth highest peak of WO₃ from 2θ = 34.074° (x = 1.0) to 2θ = 34.371° (x = 0.80). The plot of crystallite dimension versus molar content of WO₃ is depicted in Figure 4. The line shows that the sizes are almost fitted with the exponential decay curve.

Figure 5 shows TEM micrographs of the powders obtained from pure WO₃, and 5 wt.%, 10 wt.%, and 20 wt.% Y₂O₃ doped into WO₃. TEM images show groups of nanosized particles ranging below 85 nm (mean particle size, 63 nm). The sizes of particles were larger than the crystallite sizes measured by XRD because the scattering measures the average lengths between the defects. Spherical crystallites in undoped samples (Figure 5a) and 5 wt.% doped (Figure 5b) have sizes ranging from 22 nm to 85 nm and almost show an angular hexagon. At higher levels of doping (Figure 5c and 5d), the aggregates of particles become elongated and rod-shape with an average lengthened and broadness of 65 nm and 15 nm, respectively. However, reduction of particle sizes due to the dopant concentration has been reported by the other researchers [10,20,29].

The surface microstructure changes of films that were about 15 μm thick over a 5 μm × 5 μm area due to the level of doping were studied. AFM micrographs of the undoped, and 10 wt.% and 20 wt.% Y₂O₃ doped into WO₃ films are shown in Figure 6. Analysis was run under two channel simulated mode, channel A for topology (Z sensing) and channel B for deflection. The films show an extremely smooth surface with a roughness of about 0.6 μm. From grain size analysis of AFM and for the given area of the films, the average size of 3.324E + 4 nm² ± 5% and average diameter of 200 ± 35 nm were observed. The AFM micrographs confirmed that the Y₂O₃ concentration was not an important factor for the porosity and roughness of the sensing area.

Figure 7a shows a SEM micrograph of the cross section film layer deposited by screen printing. A 10 μm thick Pt electrode and a 15 μm thick sensing layer can be observed. The surface of sensing area in Figure 7b shows porosity as the typical microstructure of screen printed sensing films. The surface porosity observed was not a ruling factor of the doping concentration as it was also demonstrated in AFM results. The porosity of the sensing layer definitely resulted from the annealing temperature, such that the films annealed at lower temperature exhibit a compact flat surface with no crystalline structure, and for film annealed at higher temperature, the surface becomes rough and granules appear, and the average grain size increases, while the film becomes more porous [11,20,30]. Therefore, the porosity doesn’t change since the annealing temperature in this experiment was kept at 650 °C. The average grain size measured in the SEM images is very close to those obtained from AFM results.

Electric resistance of the samples was measured at various temperatures ranging from 100 °C to 525 °C before applying gas. Figure 8 shows the variation of film resistance versus temperature for all samples, where the resistances converted to the logarithm was plotted against temperature for the given range. The results show the resistance of all the samples decreased with increasing temperature within the given temperature range. A magnificent increase in the resistance of films can be observed due to the concentration level of Y₂O₃. The increase of resistance due to the molar concentration of yttria can be interpreted as an increasing in the depletion layer thickness. Since some site of W is replaced with Y, as it was illustrated in XRD results, the majority of carrier concentration in W is decreased, resulting in an increase in Schottky barrier height at the grain boundaries. The minimum resistance observed at T = 502 °C was 0.723 kΩ for undoped WO₃, while it was 6,456 kΩ for 20 wt.% added Y₂O₃ at T = 508 °C.

Figure 9 shows the setup experiment for measuring response of the samples. A dc-voltage is applied across the heater (V_h) and is changed corresponding to the sensor temperature. A fixed dc voltage is also applied to the V_in and voltage across the R_L (V_RL) is measured. The maximum value of V_RL is located at minimum value of sensor resistance, R_S. The load resistance, R_L, is chosen to prevent the sensor from saturation. Then R_S as well as sensitivity of sensor (S) is determined from the Equations (2) and (3), respectively: (2) R S = V in − V RL V RL × R L (3) S = R S , a R S , gwhere R_S,a is the resistance of sensor in air and R_S,g is the resistance of sensor in presence of gas.

The sensitivity prosperities of samples in the presence of methane were tested inside a test chamber and nitrogen was used for purging. It is well known that the sensitivity of a gas sensor is extremely dependent on the operating temperature of the sensor as well as the concentration of doping [14–20]. Such dependencies for detection of 1,000 ppm CH₄ are depicted in Figure 10. From this figure, the maximum sensitivity of the samples is located at about 405 °C for pure film. It can be seen that the concentration of yttria up to 10 wt.% shifts the maximum sensitivity to a lower operating temperature, but maximum sensing in samples with more than 10 wt.% yttria content take place at operating temperature above 405 °C. Those shifts can be explained as follow:

The gas absorbance mechanism of WO₃, as an n-type semiconductor material, is obviously dependant on the amount of chemisorbed oxygen ions on the surface [11]. The chemisorption of a reducing gas such as CH₄, needs the electrons from the surface of the film which is resulted in a change in the surface charge (e.g., OH⁻→O²⁻, O⁻→O²⁻, or O²⁻→2O⁻). In the case of complete reaction, methane molecules react with oxygen ions trapped between grains boundaries caused to increase charge carriers and resulted to decrease film resistance: (4) CH 4 + 4 O − → CO 2 + 2 H 2 O + 4 e −

Furthermore, increasing concentration of Y₂O₃ caused a change in the activation energy (E_a) and slop of Schottky barrier (Δφ) over the operating temperature range. Since the conductivity of the film is related to the activation energy, height of Schottky barrier, and operating temperature (σ∼exp(–eΔφ/kT), σ∼exp(E_a/kT)); any changes either in operating temperature or carrier concentration would result in changes in the film conductivity.

Figure 11 shows the responses of the samples versus different concentrations of CH₄ (50 ppm ∼ 20,000 ppm), measured at the temperature corresponding to the maximum response to 1,000 ppm CH₄. It can be seen that the responses rapidly increase due to increasing CH₄ concentration up to about 1% (10,000 ppm) and then a slightly change in the response at higher concentration of methane can be observed. Undoped (x =1) sensor shows greater response than the doped samples at about 3,000 ppm and below, but at higher concentration, 5 wt.% (x = 0.95) and 10 wt.% (x = 0.90), doped sensors show higher response with respect to the other samples. From the TEM and XRD results, it can be seen that at a concentration of yttria higher than 10 wt.%, crystallography of WO₃ shows the particles are agglomerated and tendto a rod-like structure, due to the contents of yttria and sintering temperature, as the yttria is a well known sintering aid factor [31,32]. Below 10 wt.% content, the grain sizes are increased, so that the 10 wt.% yttria content can modify the operating temperature as well as the sensor response. At concentrations above 10,000 ppm of methane, the rate of recombination and generation of surface charges would be constant; therefore no more reaction between CH₄ molecules and surface charges happens and the sensor would be saturated. Concentration of yttria higher than 10 wt.% causes a high gradient of barrier potential, which wouldlead to high surface resistivity and lower sensitivity.

While the sensors are exposed to target gas, the sensor resistance decreases rapidly with time to be stable at a value in which the complete reaction happens. On the other hand, the complete reaction takes a place at a certain temperature so that it would be incomplete below and above that temperature, as it was illustrated at Figure 10. The time taken from the exposure until the response is stable is called the response time (t_r), while the taken time throughout removing gas until the response is stable at its minimum value is called the recovery time (t_rr). Figure 12 shows the response of samples during exposure to different concentrations of CH₄ measured across the load resistor. From this figure and with respect to Figure 9, it can be seen that during exposure, higher concentration of methane causes V_RL increase and voltage across R_S is decreased rapidly with almost the same response time for each sample, while at higher concentration of CH₄ recovery time is extended.

The chemisorption of C₄H₁₀ during a complete reaction with trapped negative ions of oxygen produces water vapor and carbon dioxide. If the operating temperature is not sufficient or the oxygen ions are limited, soot or carbon monoxide may also be formed: (5) 2 C 4 H 10 + 13 O 2 − → 8 CO 2 + 10 H 2 O + 13 e −where the negatively charge carriers, produced during the reaction, cause a fast degradation in the resistance of the sensor.

The temperature dependency of the reaction and the optimized operating temperature, where an almost complete reaction take place, are investigated in the presence of 1,000 ppm iso-butane in a temperature range between 100 to 500 °C. Figure 13 shows such a relation between the operating temperature and the normalized resistance of the samples at 0.1% of C₄H₁₀ air content, in which the pure film has the higher sensitivity than the others, meanwhile a lower operating temperature was observed in 10 wt.% doped film. Figure 14 shows the sensitivity of the samples exposed to a range of C₄H₁₀ at their optimized operating temperature, where the 10 wt.% film shows a magnificent sensitivity to above 0.2% of iso-butane content. Furthermore, x = 0.9 sample shows an almost linear of sensitivity for whole range of the gas concentration.

To investigate the response and recovery time of the samples, C₄H₁₀ was applied in different concentrations from 100 to 5,000 ppm. Figure 15 shows those responses of the samples during exposure to the gas. From this figure and Figure 14, it seems that the responses of pure and 5 wt.% doped sample are almost the same for the whole range of the applied gas. When the concentration of iso-butane is increased above 1,000 ppm, an overshooting in response of the pure film is observed, while the response of 10 wt.% added yttria shows a higher sensitivity than the other films in the steady state. From Figure 15 it can be seen that the peak of overshooting is increased due to increasing of the gas concentration. Such overshooting in gas concentration is considered to show up as a result of competition between diffusion and reaction for the gas molecules having entered from the surface to the bulk [33]. In consequence of a higher Schottky barrier due to increasing the ratio of dopant, the overshooting is eliminated in a higher concentration of yttria. Also, it can be seen that the films with dopant concentration above 15 wt.% show a poor response and recovery times.