Model-based measurement uncertainty estimation in amperometric dissolved oxygen concentration measurement

In this paper, uncertainty sources in amperometric dissolved oxygen ( DO) concentration measurement are explored and the ISO GUM uncertainty estimation procedure based on a detailed measurement model is presented. The procedure is applied to two different commercial amperometric DO measurement instruments of galvanic type differing in cathode and membrane area and membrane thickness. The complete uncertainty budgets of several typical measurement processes of the two instruments are discussed. From this comparison evidence is provided that the deciding influential factors may be different for the investigated instruments under otherwise comparable measurement conditions, even though the instruments follow the same working principles. Furthermore, the uncertainty as well as the uncertainty budget of the same instrument under different measurement conditions may differ significantly. In this study, variations in the relative expanded uncertainty between U = 0.8% and U = 9% ( k = 2) were observed for the same instrument under different conditions. At DO concentrations lower than below 4 mg l(-1) ( depending on other conditions), the background current of the sensor becomes the dominating uncertainty source. At DO concentrations above that range, a variety of factors become relevant depending on the specific conditions, for instance stirring speed and membrane properties. The high importance of the cathode and membrane area, membrane material and membrane thickness on the uncertainty is demonstrated. Based on these results, a set of recommendations for DO sensor design is formulated.