|
VAM Project MU3: Controlling Uncertainties in Extraction and Recovery
Complete curve fitting of extraction profiles using iterative model algorithms and non-linear least squares fitting reduces uncertainties in recovery estimates.
The complete paper forming the body of this report forms deliverable MU3 d in the VAM 2000-2003 programme.
The paper incorporates material originally described in report LGC/VAM/2002/017 part 2, describing the implementation of basic models for extraction profile modelling using diffusion models. Part of this material was first presented in poster form in June 2002 at the Eurachem/CITAC workshop 'Measurement Traceability and Uncertainty in Analytical Chemistry: Meeting the requirements of ISO/IEC 17025'.
The paper reports the use of numerical approaches to modelling extraction profiles, and shows that the approach substantially reduces statistical prediction uncertainties compared to those obtained on the basis of a three-point extrapolation from the later part of the extraction curve. In particular, the statistical uncertainty in predicted final extract value is as low as 0.3% of the value (0.11 in 31 % (w/w)),based on residual standard deviation. This is at least a factor of four better than the best uncertainty attainable from the three-point extrapolation, assuming good fit and the same precision. This improvement is likely to be reduced when full account is taken of residual correlation, but still represents a considerable gain.
|