Uncertainty models and influence of the calibration span on ambient air measurements of NO2 by chemiluminescence

Abstract

An assessment of uncertainty in the hourly and annual limit values of NO₂ measurements by molybdenum NO₂-to-NO conversion followed by chemiluminescence detection was carried out at 3 different span concentrations (100, 300 and 700 nmol mol⁻¹). The uncertainty of the linearity term was calculated for each span concentration by considering (i) a zero-and-span calibration and (ii) a multipoint calibration. Two uncertainty models were applied for the overall uncertainty estimation: (i) the Standard EN 14211 and (ii) a mechanistic model that considers the NO₂ reduction in the converter. The main difference between these models stems from considering or not the possible covariances derived from interactions between NO_x and NO concentrations and the converter's efficiency. For both models, the span determined whether or not it was possible to meet the quality objective requested by the EU Air Quality Directives in the annual limit value when no linearity corrections were performed in environments with NO_z/NO₂ ratios ≤ 0.04. In environments with significant amounts of NO_z species (NO_z/NO₂ ≥ 0.12), the expanded uncertainty can easily be higher than the data quality objective if bias' corrections are not or cannot be applied.