Designing Calibrators Preparation Methodology for Instrumental Quantifications

Ricardo Bettencourt da Silva, Chemistry and Biochemistry Department, Faculty of Sciences of the University of Lisbon, Lisboa, Portugal
Roksana Wójcicka, Chemistry And Biochemistry Department, Faculty Of Sciences Of The University Of Lisbon, Lisboa, Portugal
M. Filomena Camões, Chemistry And Biochemistry Department, Faculty Of Sciences Of The University Of Lisbon, Lisboa, Portugal
Cristina Oliveira, Chemistry And Biochemistry Department, Faculty Of Sciences Of The University Of Lisbon, Lisboa, Portugal

Calibrators¹ quality plays a major role in instrumental quantifications uncertainty regardless of used approach for uncertainty estimation. The ‘bottom-up’ approach for uncertainty evaluation involves estimating uncertainty from the regression model after checking calibrators quality requirements. The alternative ‘top-down’ evaluation of the measurement uncertainty based on in-house validation data involves estimating uncertainty from measurements intermediate precision and trueness; i.e. regression model is not used to estimate measurement uncertainty. Nevertheless, if calibrators quality is too weak, global performance is affected. In extreme cases, highly uncertain calibrators values can make difficult the modelling of performance throughout the calibration range from measurements precision and mean errors estimated in some quantity values.

This work presents a tutorial to define calibrators preparation methodology aiming at ensuring this component will become a minor uncertainty component in ‘bottom-up’ uncertainty evaluations or will not affect ‘top-down’ uncertainty modelling throughout the calibration range from performance estimated at a small number of levels.

This tutorial is implemented in a user-friendly and validated MS-Excel spreadsheet.

The proposed calibrators preparation methodology was successfully applied to the determination of nitrites in drinking water by molecular spectroscopy, between 0.1  mg L^-1 and 0.4 mg L^-1., allowing estimating results with an expanded relative uncertainty ranging from 2.1 % to 3.1 %. Measurement uncertainty estimation quality was checked through the analysis of four control standards distributed along the calibration range, in ten independent calibrations performed in different days summing up 40 controls. The metrological compatibility of estimated and reference values of control standards proved the adequacy of the measurement model.

¹ - Calibrators are frequently designated ‘calibration standards’.

R. J. N. Bettencourt da Silva, Spreadsheet for designing valid least-squares calibrations: A tutorial, Talanta 148 (2016) 177-190.