In-Situ DRIFT Spectroscopy on a Resistive NOx Dosimeter – How Can the Non-Linear Electrical Behavior be Explained?

Abstract

Introduction The detection of low NOx concentrations is still on focus for air-quality monitoring. The direct detection of the total dose of NOx is beneficial since the air pollution limits are given as hourly and annual mean values. A gas dosimeter for direct detection of the dose of NOx was presented in [1,2]. In the so-called sorption phase, NOx molecules are adsorbed on the functional film and the electrical properties, e.g. the resistance, change. During this phase, the characteristic sensor curve corresponds to the correlation between NOx dose and resistance change. The sensor curve is linear until a certain loading state is reached. As soon as the characteristic curve gets non-linear, described often as saturation of the NOx adsorption [1], the regeneration of the sensor is necessary. During the thermally induced regeneration, the sorbed NOx species is desorbed from the surface. Afterwards, a new sorption cycle can begin.For further understanding of the sensing mechanism and the effects of non-linearity, NOx dosimeters are investigated by in-situ DRIFT (Diffuse Reflectance Infrared-Fourier-Transformed)-Spectroscopy [3]. The electrical sensor signal and the formation of adsorbed species are investigated in-situ directly on the dosimeter. The non-linear behavior of the electrical properties occurring at a certain sensor signal will be investigated at various film thicknesses [4] and will be combined with the sorption properties of the material. Materials and Methods As functional material potassium-permanganate impregnated on lanthanum-stabilized alumina powder (K/Mn-La-Al2O3) [1] is investigated. It is applied by screen-printing on an alumina substrate equipped with interdigitated Au electrodes and fired at 650 °C. Different layer thicknesses of the functional film are achieved by several screen printing steps. The sensors are mounted in a gas purgeable in-situ DRIFTS cell (see Fig. 1) [3] and heated by an integrated platinum heating structure to 350 °C. They are exposed to synthetic air (20 % O2 and 2 % humidity in N2 balance) and varying ppb NOx concentrations are added. The NOx concentration is analyzed by a chemiluminescence detector (CLD). The electrical behavior of the sensor is measured by impedance spectroscopy (f = 200 Hz, U eff = 200 mV) and the resistance R can be calculated since the R||C equivalent circuit is valid. The sensor signal is defined as the relative resistance change (R-R 0/R 0), with R 0 being the resistance in base gas and R the resistance during NOx exposure. At the same time, Drift-spectra of the sensor are measured by a FTIR Spectrometer and the absorbance -log(I/I 0), with the intensity I 0 in base gas and the intensity I during NOx adsorption, is calculated. The sensors are regenerated by heating up to 650 °C. Results and Discussion An in-situ DRIFTS result during exposure to three peaks of 50 ppb NOx with NOx pauses in-between is shown in Fig. 2. The NOx dose in ppm∙s is calculated by integrating the CLD NOx signal. The dose increases linearly during NOx exposure and stays constant during dosing pauses. The sensor signal shows the same behavior: ΔR/R 0 increases linearly when 50 ppb NOx are present and stays constant during NOx absence. The calculated absorbance shows a clear peak at a wave number of 1370 cm-1 occurring during NOx exposure. It can be attributed to the formation of nitrates on potassium sites of K/Mn-La-Al2O3 and is characteristic for the sorption phase [3]. The calculated peak area A E of this nitrate peak is presented in Fig. 2, too. The peak area A E and therefore the amount of formed nitrates increases linearly during NOx addition and remains constant during NOx pauses. The electrical properties of the dosimeter can be related directly to the NOx dose and to the formed nitrates. One important point is, that all curves increase linearly during NOx exposure and no saturation effects can be observed for the sensor signal and the nitrate formation. In Fig. 3, the sensor is exposed to three peaks of 390 ppb NOx. ΔR/R 0 gets non-linear after reaching 40 %, the sensor seems to saturate. The slope during NOx exposure is non-linear and the signal in NOx pauses decreases. In contrast, A E behaves almost linear. It is assumed that the electrically observed non-linearity cannot be explained by a saturation of the adsorption capability of K/Mn-La-Al2O3. It seems that the NOx adsorption capability is not exhausted, but the electrical properties show a saturation effect. Conclusions The results show that the sensor behaves like a dosimeter with linear characteristics between NOx dose, sensor signal and formed nitrate species until a certain sensor signal of around 40 % is reached. At higher relative resistance changes, the electrical signal becomes non-linear, whereas the adsorption capability itself, i.e. the nitrate formation, remains linear and the NOx adsorption on K/Mn-La-Al2O3 is not saturated. This behavior will be investigated by in-situ DRIFTS in dependence on the film thickness.