Ethanol sensor based on microrod-like La-doped barium stannate

Due to its importance in human health and environmental monitoring, ethanol sensing has generated large research interest across the globe. In this work, we report a simple co-precipitation method to synthesize La-doped BaSnO₃ and evaluate its ethanol sensing properties. Electron microscopy and X-ray diffraction analyses revealed formation of crystalline BaSnO₃ having rod-like shape, few microns in length and 1–2 µm in width. The optimal sensing performance was achieved when operated at 220 °C for 4 wt% La-doped BaSnO₃ microrods for which a response as high as 48 was obtained against 100 ppm of ethanol exposure, whereas the undoped BaSnO₃ exhibited its best performance (although much lower response) at 260 °C. Multiple characterization techniques revealed that the enhancement of the sensor performance by incorporation of La was due to changes in the physico-chemical properties like specific surface area, oxygen content and electronic bandgap of the BaSnO₃. The high repeatability, high selectivity to ethanol, fast response and recovery times, and low detection limit of 0.01 ppm suggest good potential for the 4% La-doped BaSnO₃ film as a low cost ethanol sensor.

The sensing of volatile organic compounds (VOCs) is a major concern of modern-day research due to their significant role in our day-to-day life [[1]]. Ethanol is a VOC whose detection, especially at low concentration, is useful in fermentation, food packaging, pharmaceuticals and traffic safety [[1]–[8]]. To identify drunk drivers, ethanol sensors are used by traffic police across the globe since the exhaled breath of humans who have consumed ethanol is expected to contain more ethanol than others [[2]]. Naturally, many researchers have attempted to develop ethanol sensors using a variety of techniques and materials of which resistive sensors using metal oxide semiconductors (MOS)-based nano/microstructures and composites are perhaps the most widely explored [[9]–[14]]. In recent years, ABO3 type perovskite metal oxides (A and B represent two different metals), such as BaSnO3 (BSO) have generated good research attention for sensor applications [[15]–[28]]. However, the performance of these sensors has been somewhat limited due to their relatively poor sensitivity, selectivity, etc. Naturally, there is an increasing demand for research to improve the performance of BSO-based sensors.

Some researchers demonstrated that the method of synthesis of BSO could influence the sensing performance of BSO and that some method produces BSO with morphology that exhibits better sensor performance than that by other methods [[14], [29]–[32]]. Others found that the sensing performance of BSO could be improved by doping with suitably chosen cations since doping can alter the semiconductor band structure [[29]–[36]]. For example, Lu et al. [[33]] reported a sharp change in the electrical conductivity of BSO by Sb doping whereas Liu et al. [[34]] observed the same with Gd doping. Recently, our group [[36]] also reported the effect of Gd doping on ethanol sensing properties of BSO. Some researchers showed La doping to be effective in controlling the properties of BSO and other semiconductors [[31], [35], [37]]. However, we could not find any studies on the influence of La doping on the ethanol sensing properties of BSO synthesized by co-precipitation method.

In view of the above, herein, we report the fabrication of an ethanol sensor based on BSO synthesized by co-precipitation method whose performance has been enhanced by doping it with La ions. The ethanol sensing properties of BSO and La-doped BSO (LBSO) were tested as a function of La dopant weight fraction at different temperatures to find out the optimum dopant fraction and optimum operating temperature of the sensor.

Tin (IV) chloride pentahydrate (SnCl4·5H2O), barium nitrate (Ba(NO3)2, lanthanum nitrate hexahydrate (La(NO3)3·6H2O), and oxalic acid (C2H2O4·2H2O) were all procured from Aladdin Chemicals, USA with 99.999% purity and used as received.

The detailed method for the synthesis of BSO can be found in a previous study by our group [[36]]. Briefly, 1 M oxalic acid solution was added to a 50 ml aqueous solution of 25 mmol Ba(NO3)2 and 25 mmol SnCl4·5H2O prepared through stirring for 30 min. For the homogenous precipitation of complex oxalates, the obtained solution was stirred for an additional 2 h following the addition of oxalic acid. The precipitate was extracted by centrifuging the obtained solution and dried at 90 °C for 12 h. The complex precursor oxalates were then annealed in air at 700 °C for 4 h and cooled to room temperature. Both heating and cooling were performed very slowly at the rate of 1 °C/min.

The procedure for synthesis of La-doped BSO differed from above only in the first step in which along with Ba(NO3)2 and SnCl4·5H2O, appropriate amounts of La(NO3)3·6H2O were also added in water. Six samples were prepared by varying the amount of La(NO3)3·6H2O and were named as BSO (undoped), 1LBSO (1 wt% La), 2LBSO (2 wt% La), 3LBSO (3 wt% La), 4LBSO (4 wt% La), and 5LBSO (5 wt% La), where the numbers preceding the LBSO samples indicate the weight fraction of La dopant.

To confirm the phase composition of samples, X-ray diffraction (XRD, Bruker D8 Advance diffractometer, USA) was performed using Cu-Kα radiation (λ = 1.54 Å) at room temperature. A 10 kV accelerating voltage was used to record the electron micrographs using a field emission scanning electron microscope (FESEM, FEI Nova NanoSEM 430, USA). The samples were further characterized by a High-Resolution Transmission Electron Microscopy (HRTEM, Tecnai G2 F20 S-TWIN microscope) along with Selective Area Electron Diffraction (SAED) under an accelerating voltage of 120 kV. UV–Vis diffused reflectance spectroscopy (UV-DRS, Shimadzu UV-3600 spectrophotometer, Japan) was used to measure and compare the electronic bandgaps of the obtained samples. N2 adsorption-desorption isotherm plots (Micrometrics ASAP 2020 surface area analysis apparatus, USA) were used to evaluate the Brunauer-Emmett-Teller (BET) specific surface area of the as-prepared samples. The chemical compositions of various elements present on the surface of BSO and 4LBSO were confirmed using X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB250Xi, USA) performed with an Al-Kα source radiating X-rays of energy 1486.7 eV. Further, the binding energies were calibrated using the position of the C1s peak at 284.9 eV originating from the adventitious carbon signal [[38]].

The ethanol sensors were fabricated by a method reported earlier [[35]]. A thin film of the BSO/LBSO, with an average thickness of 20 µm, coated on the outer surface of an alumina tube (Fig. 1) acted as the sensor which was prepared by drop-casting a terpineol suspension of the samples followed by drying for 12 h at 80 °C. A multimeter connected to two gold contacts deposited at both ends of the alumina tubes, measured the electrical resistance of the film while the temperature of the sensor was regulated by a Ni-Cr coil passing through the alumina tube. The sensing data were recorded at 60% relative humidity unless otherwise specified. The sensor response (S) was measured [[14], [36]] as follows:

Graph

where Ra and Rg stand for the electrical resistance of the sensor in air and in presence of ethanol, respectively.

Graph: Fig. 1 Schematic diagram of the sensor

Mastech MS8226 DMM multimeter, China, interfaced with a desktop PC, was used to measure and record the voltages across the reference load while DC power supply regulator, Maisheng MT153DH, China, was used to regulate the voltage across the Ni–Cr heating element.

The XRD patterns in Fig. 2 show presence of peaks at 2θ values of 30.69°, 43.96°, 54.56°, and 63.91°, which correspond to (110), (200), (211), and (220) planes of cubic BaSnO3 phase, respectively, in accordance with PDF#15-0780. A closer look at Fig. 2 also reveals that the intensity of the strongest peak of BSO corresponding to (110) planes is slightly reduced and its position is slightly shifted to lower 2θ angle upon La doping indicating a slight increase in the inter-planar spacing. For peaks due to other planes this change is not so prominent due to their lower intensity. Interestingly, the best sensing performance (discussed later) was obtained from the 4LBSO sample and hence all data except sensing data shown hereafter in the manuscript refer to this sample only. The crystallite size of BSO, as measured by Williamson-Hall method was found be ~ 21.3 nm in good agreement with previous literature [[32], [36]], whereas it decreased monotonously as the doping concentration increased to 4 wt% (4LBSO) for which the sizes were estimated to be 19.1 nm, 18.7 nm, 17.1 nm, and 16.2 nm for 1LBSO, 2LBSO, 3LBSO and 4LBSO, respectively. A slight increase of the crystallite size (16.8 nm) was observed for 5LBSO compared to that of 4LBSO. The reduction of the crystallite size is consistent with the reduced intensity of the strongest XRD peak of BSO that results in a broadening of the peak width.

Graph: Fig. 2 XRD patterns of BSO and La-doped BSO samples

The FESEM micrographs of all the synthesized samples are shown in Fig. 3 where (a) and (b) represent micrographs of BSO and 4LBSO where one can clearly see that most of the BSO are formed as rod-like structures having widths and lengths on the order of a few microns. One can see that both the length and width of these rods are smaller for LBSO than those of BSO. The width distribution of the particles in these two samples, as obtained by analysis of several micrographs, are presented as insets within their micrographs which reveals that the average width of BSO microrods is ~ 1.9 µm which reduces to ~ 1.1 µm for 4LBSO. For completeness of the data, we have also shown the FESEM micrographs of the other doped samples such as 1LBSO, 2LBSO, 3LBSO and 5LBSO in Fig. 3c–f, respectively, although these images are recorded with a slightly higher magnification. One can see rodlike structures in all these samples but the rods seem thicker and less well-defined. Since the optimum sensing performance was obtained with 4LBSO (discussed later), we have not performed any further analysis of the images of other doped samples.

Graph: Fig. 3 FESEM micrographs of a BSO, b 4LBSO, c 1LBSO, d 2LBSO, e 3LBSO, and f 5LBSO. Insets in (a) and (b) represent the width distribution of microrods

TEM images in Fig. 4a, b show that the surface of the microrods of BSO and 4LBSO is quite rough and porous in nature. The lattice resolved TEM images in Fig. 4c, d show presence of (110) plane having d-spacing of 0.2937 nm and 0.2986 nm, for BSO and 4LBSO, respectively. This is consistent with the observation of the most intense peak in XRD plot for (110) planes and also in accordance with the small shift in its position from BSO to 4LBSO. The selected area electron diffraction (SAED) patterns in Fig. 4e, f show the presence of rings due to different crystalline planes such as (110), (200) and (220) all of which were also observed by XRD.

Graph: Fig. 4 a, b TEM micrograph; c, d high resolution TEM micrographs, and e, f SAED patterns of BSO and 4LBSO, respectively

Since the interaction of a semiconductor with UV–visible light can provide information on its electronic properties (such as bandgap), which are believed to contribute to the semiconductor's sensing performance, we performed UV-DRS measurements on the films of all the samples which are shown in Fig. 5a and the corresponding Tauc plots are presented in Fig. 5b. All the doped samples showed strong and intense absorption around 280 nm along with minor shifts in the position (Fig. 5a). The intercepts of the Tauc plot on the x-axis was used to estimate the bandgap [[40]] but due to overlapping nature of the Tauc plots, the intercepting region of the optimized sample (4LBSO), is separately shown as inset of Fig. 5b where one can clearly see its bandgap is of 3.9 eV. From the intercepts, the bandgap energy of the undoped BSO was estimated as 3.3 eV, in good agreement with reported literature [[20]], whereas that for other doped samples were found to be 3.70 eV, 3.75 eV, 3.78 eV, 3.90 eV, and 3.82 eV for 1LBSO, 2LBSO, 3LBSO, 4LBSO, and 5LBSO, respectively. Thus, the bandgap of the doped samples monotonously increased as the concentration of La dopant increased except for 5 wt% doped sample, for which it reduced slightly from that of 4LBSO. The increase in the bandgap upon La doping could be due to reduction in the crystallite size as well as band filling caused by partial reduction of the Sn4+ (4d105s05p0) cations within BSO, consistent with previous literature [[42]].

Graph: Fig. 5 a UV-DRS spectra, and b Tauc plots for BSO and LBSO. The inset in b is an expanded Tauc plot for 4LBSO showing the intercept on x-axis

N2 adsorption-desorption isotherms of BSO and 4LBSO are shown in Fig. 6a. It can be observed that both BSO and 4LBSO exhibit type IV isotherms although it was found that BSO exhibits an H4 hysteresis loop while 4LBSO exhibits an H1 hysteresis loop. Hence, it is evident that the 4LBSO has higher pore size uniformity along with facile pore connectivity compared to pure BSO. Figure 6b shows the variation of specific surface areas (estimated from BET N2 adsorption-desorption isotherms) with variation in La dopant weight fraction. It can be seen that the specific surface area of the 4LBSO sample (70.58 m2/gm) is remarkably higher than that of the BSO sample (32.6 m2/gm) which is not surprising if we recall the FESEM image of 4LBSO in Fig. 3b where the 4LBSO rods were observed to have much smaller width and length compared to those of undoped BSO. All other doped samples exhibited surface area less than that of 4LBSO but more than that of undoped BSO arguably because of a doping induced formation of defects and rough surface. However, doping above 4 wt% may lead to aggregates which might have led to a slight decrease in the specific surface area of 5LBSO sample.

Graph: Fig. 6 a N2 sorption isotherms of BSO and 4LBSO, and b Plot of specific surface areas of different BSO samples (as estimated by BET isotherms) against La dopant fraction

The survey XPS spectrum of 4LBSO sample in Fig. 7a shows peaks corresponding to photoelectrons emitted from Ba, Sn, La and O and C indicating a clean sample. The Ba 3d, Sn 3d, La 3d, and O 1 s spectra obtained from 4LBSO are shown in Fig. 7b–e whereas the O1s spectrum obtained from undoped BSO is shown in Fig. 6f for comparison. The peaks at 779.1 eV and 794.5 eV in Fig. 7b originates from Ba 3d5/2 and Ba 3d3/2 core levels, respectively [[31]]. Peaks at 486.5 eV and 494.5 eV (Fig. 7c) confirmed the presence of Sn representing Sn 3d5/2 and Sn 3d3/2 states, respectively [[31]]. As the La content in the 4LBSO sample is expected to be low, the La 3d spectrum presented in Fig. 7d shows two pairs of weak peaks emerging from a noisy background signal. The relatively stronger pair of peaks at 835.2 eV and 852.5 eV correspond to photoelectrons emitted from La 3d5/2 and La 3d3/2 core levels whereas the weaker pair of peaks, 3.5 eV apart from the main peaks, at 838.9 eV and 855.8 eV, are satellite peaks originating from incomplete d-orbitals causing a continuous transition of electrons [[35], [39]]. The O1s spectra in Fig. 7e, f were both fitted with three components positioned at 531.8 eV, 530.9 eV, and 529.2 eV, corresponding to O2−, O−, and O2− species, respectively [[43]]. A closer look further reveals that the oxygen content is much higher in 4LBSO than that in undoped BSO. This can be further visualized from Table 1 which summarizes the elemental composition of BSO and 4LBSO as calculated from the XPS survey.

Graph: Fig. 7 XPS spectra of 4LBSO: a survey, b Ba 3d, c Sn 3d, d La 3d, e O 1s, and f O 1s spectrum of undoped BSO

Table 1 Variation of elemental composition of BSO and 4LBSO

The response of undoped and La-doped BSO to 100 ppm ethanol at different temperatures are shown in Fig. 8a. All sensors were found to sense ethanol in the temperature range of 150–350 °C and peaked in the range 220-260 °C. Incorporation of La also lowered the optimum temperature for the sensor from 260 °C to 220 °C. Further, the highest response of 48 was obtained for the 4LBSO sample against 100 ppm ethanol exposure at 220 °C. The variation of the sensor response as a function of temperature is induced due to the change in the resistance of the sensor with temperature [[45]].

Graph: Fig. 8 a Response of the undoped and La-doped BSO sensors to 100 ppm ethanol at different operating temperatures, b Variation of the response of 4LBSO samples to 100 ppm of various gases with operating temperature, c Selectivity of BSO and 4LBSO sensors against 100 ppm of various gases at 220 °C, and d Dependence of 4LBSO sensor to 100 ppm ethanol on relative humidity

The selectivity of 4LBSO was also evaluated by comparing the response of 4LBSO sensor to 100 ppm of a number of VOCs, namely formaldehyde (HCHO), acetaldehyde (CH3CHO), acetone (CH3COCH3), methanol (CH3OH), ethanol (C2H5OH), acetic acid (CH3COOH), ammonia (NH3), trimethylamine ((CH3)3NH3), benzene (C6H6), and toluene (C6H5CH3) at different operating temperatures. It is further evident that 4LBSO has its highest response of 48.6 to ethanol when compared with those of other VOCs (Fig. 8b), while the 2nd highest response of 8.4 was observed against acetaldehyde vapors indicating its good selectivity towards ethanol. Interestingly, it can be observed that the optimal operating temperature of 4LBSO-based sensor was similar for all tested VOCs.

Figure 8c further compares the responses of all the undoped and doped BSO sensors against 100 ppm of each of the VOCs at 220 °C in column graph where the best selectivity of the 4LBSO sensor to ethanol is clearly established as its response is much higher to ethanol than to any other VOCs. Since the response of a semiconductor sensor often drops under the influence of humidity, the effect of relative humidity on the response of 4LBSO-based sensor to 100 ppm ethanol is illustrated in Fig. 8d. It can be observed that the response of 4LBSO-based sensor decreased with an increase in the relative humidity and that the response increased significantly in a dry atmosphere. This phenomenon can be attributed to a decrease of oxygen adsorption on the sensor at higher relative humidity due to presence of moisture (H2O) molecules, on the surface of gas sensor resulting in a reduced resistance [[46]–[50]].

The dynamic response-recovery plot for the 4LBSO sensor against ethanol at different concentration is shown in Fig. 9a where one can see that the sensor has a very low limit of detection of 0.01 ppm of ethanol for which it exhibits a response of 1.6. The variation of response of sensors based on 4LBSO as a function of ethanol concentrations is shown in Fig. 9b. The following equation (Eq. 2) gave a good fit to the data:

2 $${\text{log}}_{10}S=0.37405{\text{*log}}_{10}C+0.89731$$

Graph

where C and S stand for concentration of ethanol (in ppm) and response, respectively.

As observed in Fig. 9c, the sensor was able to reproduce its response to 100 ppm ethanol for five consecutive cycles of exposure suggesting good repeatability of the sensor. The response and recovery times of the 4LBSO sensor was calculated as shown in Fig. 9d in which the response time was defined as the time required by the sensor to achieve 90% of the final resistance in ethanol whereas recovery time was defined as the time required to achieve 90% of the final resistance of the sensor in air (Ra). The recorded response and recovery times of 4LBSO to different concentrations of ethanol were found to be similar to that observed for 100 ppm ethanol. It can be seen that the sensor was able to respond to the presence of ethanol in 5 s and recover its original resistance (in air) in 12 s, both of which are quite fast.

Graph: Fig. 9 a Dynamic response curves to different ethanol exposure, b Plot of the response versus ethanol concentration, c Five consecutive cycles of response-recovery plots, and d Expanded plot of the response-recovery times. All data shown here is for 4LBSO sensor operated at 220 °C against 100 ppm ethanol

Figure 10 shows a schematic of the sensing mechanism occurring on the surface of pure BSO and 4LBSO-based sensors. As mentioned earlier, BSO being an n-type semiconductor its resistance is affected by electron transport properties [[51]]. On exposure to ethanol, an electron depletion layer is formed resulting from electron withdrawal caused by adsorption of oxygen on the surface of the sensors [[52]]. Oxygen species adsorbed on the surface of the sensor is significantly affected by the operating temperature of the sensor. O2− dominates below 147 °C, O− species dominate within the range of 147–397 °C, and above 397 °C, O2−dominates [[34], [45], [53]]. Since 4LBSO sensor performed best at 220 °C, we can say that in the present work, the target VOCs predominantly react with O−species. The amount of adsorbed oxygen is higher in 4LBSO sensor compared to BSO due to a spillover phenomenon caused by the presence of La3+ ions [[54]–[56]]. As a result of this phenomenon, the decomposition reaction of the analyte VOC takes place at sensor surface by transfer of electrons from adsorbed oxygen on the La3+ ions.

Graph: Fig. 10 Schematic of the gas sensing mechanism of the sensors

The reactions taking place on the surface of sensing material pertaining to the detection of ethanol have been described as below [[52], [57]]:

3 $${\text{O}}_2\left(\text{gas}\right)\to {\text{O}}_2\left(\text{ads}\right)$$

Graph

4 $${\text{O}}_2\left(\text{ads}\right)+e^{-}\to {{\text{O}}_2}^{-}\left(\text{ads}\right)$$

Graph

5 $${{\text{O}}_2}^{-}\left(\text{ads}\right)+e^{-}\to 2{\text{O}}^{-}\left(\text{ads}\right)$$

Graph

6 $${\text{C}}_2{\text{H}}_5\text{OH}+6{\text{O}}^{-}\left(\text{ads}\right)\to 2\text{C}{\text{O}}_2+3{\text{H}}_2\text{O}+6e^{-}$$

Graph

A response of the sensor is obtained by measuring the change in the resistance caused by the release and re-entry of the electrons, as shown in Eq. (6), to the conduction band of the sensor material. The lowering of resistance caused by the release of electrons due to reactions shown in Eqs. (3)–(6) indicates a significant electrical response from both BSO and 4LBSO.

The enhanced sensitivity of 4LBSO compared to that of BSO originates from the enhanced adsorption of oxygen ions adsorbed on the surface of the former which we have confirmed from XPS in Fig. 7e and f. The larger adsorption of oxygen ions on 4LBSO compared to that of BSO is the result of its larger specific surface area (as expected due to smaller crystallites and thinner microrods), and larger number of oxygen vacancies created by La3+ ions. Although the bandgap is slightly increased for 4LBSO due to La3+ doping, but its electrical resistance is expected to greatly reduce (by several orders of magnitude) from that of BSO due to creation of numerous oxygen vacancies as reported by Huang et al. [[42]]. When 4LBSO comes in contact with normal atmosphere, large number oxygen ions adsorb on its surface by readily extracting electrons from its conduction band in large numbers which in turn greatly reduces its electrical conductivity. Upon exposure to ethanol vapor, the adsorbed oxygen species (much larger in number than that of BSO) interact with ethanol molecules thereby releasing the "pre-extracted" electrons back to the 4LBSO which results in a large increase in its electrical conductivity (large response of the sensor).

Well crystalline microrod-like structures of BSO and La-doped BSO were prepared by a co-precipitation method. Average width of the 4LBSO microrods was about 1 micron whereas their length was around 3 µm. FESEM micrographs showed that the size of the microrods decreased upon doping with La which was further confirmed by an increase in the specific surface area as measured by N2 adsorption-desorption isotherms. When exposed to ethanol vapor, best response (48.6) was obtained at 220 °C for the sample that was doped with 4 wt% La (4LBSO). The 4LBSO sensor showed both high response and high selectivity to ethanol. Together with its high repeatability, fast response and recovery times, and low limit of detection (0.01 ppm; response 1.6), the 4LBSO film shows very good promise as a low cost ethanol sensor. XRD and XPS confirmed the success of doping as well as an excess amount of oxygen in 4LBSO. Further, XPS along with FESEM, TEM, and BET specific surface area results confirm that smaller particles of 4LBSO helped in adsorption of large quantities of oxygen species on its surface resulting in the improved performance. As confirmed through UV-DRS, the increased content of adsorbed oxygen also resulted in an enhanced bandgap of 4LBSO lowering its electrical conductivity. Overall, this results in an increased response of the sensor when exposed to external analyte such as ethanol. Hence, we have shown that La-doped BSO is a potential candidate for ethanol sensor although further studies on its long time stability against aging, humidity, etc. are to be conducted for commercial use. Future studies can also focus on lowering the operating temperature of the BSO sensor through control of synthesis techniques, doping, and heterostructures.

This project was funded by National Natural Science Foundation of China (Nos. 61671019 and 61971003).

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.