Introduction Chemo-resistive gas sensors are promising for distributed air quality monitoring and medical diagnostics due to their compact and low-cost design as well as high sensitivity to detect analytes down to ppb concentrations. Yet, their widespread commercial use has been hindered due to their limited selectivity over interferants in the above applications. Most challenging is ethanol since established metal-oxide sensors (e.g., SnO2 [1], WO3 [2] or TiO2 [3]) respond to it and due to its omnipresence indoors (especially in hospitals and distilleries) from hand disinfection, distillation and cleaning agents, reaching >20 ppm concentrations [4]. This often exceeds trace level concentrations of target analytes such as metabolic acetone [5] during breath analysis (500 ppb in healthy humans [6]) or carcinogenic benzene in indoor air (8 h exposure limit of 50 ppb in the EU) by orders of magnitude. Therefore, removing the ethanol interference is crucial for establishing commercial use of such sensors.A simple approach to mitigate ethanol interference is to combine sensors with a modular filter [7]. Particularly suitable are catalytic filters [8] that convert active interferants continuously to sensor-inactive species, while target analytes remain unscathed. Here, we systematically optimize the surface properties (i.e., basicity) of catalysts at the nanoscale for the highly selective removal of ethanol in gas mixtures at high relative humidity (RH). Method Catalysts (e.g., WO3, SnO2, Fe2O3 and ZnO) were prepared by scalable flame spray pyrolysis and air annealed at 500 °C for 1 h to ensure thermal stability [2]. Catalytic conversion was assessed between 100 – 350 °C for ethanol and acetone (and for ZnO additionally with formaldehyde, 1-butanol, isopropanol, methanol, isoprene, toluene, benzene and H2 and CH4) at 90% RH with a gas mixing setup connected to calibrated gas standards and evaluated by proton transfer reaction time-of-flight mass spectrometry (PTR-ToF-MS) and a Quintron Breath Tracker for H2 and CH4. The best performing filter (i.e., ZnO) was then installed as packed bed upstream of a flame-made Si-doped WO3 [9] sensor and exposed to 1 ppm acetone as well as 5 - 20 ppm ethanol. The sensor response is defined as S = Rair/Ranalyte -1, where Rair and Ranalyte are the sensing film resistances in air and during analyte exposure, respectively. Results and Conclusions Figure 1 shows the catalytic performance of (a) WO3, (b) SnO2, (c) Fe2O3 and (d) ZnO towards 1 ppm acetone (circles) and ethanol (triangles) at 90% RH [2]. The catalysts were chosen due to their distinctly different surface acidity and basicity, where the basicity increases in the order WO3 < SnO2 < Fe2O3 < ZnO [10]. The highest selectivity was achieved for basic ZnO [10], where complete ethanol conversion is achieved at 200 °C while acetone remains unaffected until 260 °C. In fact, acetone is known to coordinate primarily with Lewis acid sites [11] that are abundantly present on acidic oxides (e.g., WO3 [10]), while ethanol is converted preferentially on surface-adsorbed hydroxyl-related species of basic oxides [12]. Note that at 260 °C, the ZnO removes also other interferants (e.g., formaldehyde, 1-butanol, isopropanol, methanol and isoprene) while it leaves toluene, benzene, H2 and CH4 intact. So it may be used to detect the latter analytes selectively over ethanol as well (Figure 2).To validate the filter performance, a packed bed of ZnO particles at 260 °C was coupled to a Si/WO3 sensor and tested with 1 ppm acetone as well as 5, 10 and 20 ppm ethanol [2]. Without the activated filter (Figure 3a), the sensor clearly detects 1 ppm acetone with high response (i.e., 8.6), but is sensitive to ethanol as well (i.e., response of 7.37 at 20 ppm). In contrast, with the activated filter (Figure 3b), the interference of ethanol is reduced significantly, for example by 88% for 20 ppm ethanol. The residual response can be attributed to formation of H2 during ethanol oxidation [12].
Read full abstract
Aug 27, 2021
Joachim Ezeji + 1
Open Access