Uncertainty estimation in chemical measurements: carbon dioxide determination at atmospheric concentration.

The present work deals with the evaluation of measurement uncertainty in the determination of carbon dioxide (CO2) concentration in atmosphere, given the high relevance of this greenhouse gas that influences earth climate. In order to carry out CO2 measurements, non dispersive infrared (NDIR) analysers are usually employed as they are stable and scarcely affected by interferences from other air components or pollutants. Typical uncertainty sources are the resolution of the analyser, its time drift and the contributions due to instrument calibration, which is required in order to produce traceable measurement results. The calibration uncertainty takes into account the uncertainty of the composition of the calibration gas mixtures, the instrument repeatability and the possible or residual lack of fit of the adopted mathematical model.

Stability of a NDIR analyser for CO2 at atmospheric concentration.

Carbon dioxide monitoring is significant in the environmental field since this gas plays an important role in the greenhouse effect. In order to determine CO2 concentration and to develop simulation models, it is necessary to carry out measurements which are accurate and comparable in time and space, i.e. SI-traceable. Non-dispersive infrared (NDIR) analysers are employed for CO2 measurements, as they are precise and stable. In order to achieve traceability, such instruments have to be characterized and calibrated. At the Istituto di Metrologia "G. Colonnetti"--CNR, a procedure for calibrating NDIR analysers for CO2 at atmospheric level was developed, which enables to calculate a correction for the analyser output. In addition, a complete uncertainty analysis was carried out and a correct traceability chain was established. The goal of the present work is the study of the stability of a NDIR analyser by repeating calibrations during three years and comparing the correction curves obtained to identify a proper re-calibration interval for such analysers. The investigated instrument has good repeatability and reproducibility, hence satisfactory stability during time, as shown by the short-term and long-term compatibility of calibration curves.