Uncertainty associated with the measurement of mineral micropollutants in natural waters and in waste waters: differences observed between analytical methods
during proficiency testing schemes D. De Kuyssche (1), R. Charpentier (2), M. Penhoat (1), E. Courtade (1), P. Guarini (2)
(1) USTL – IUT A de Lille 1, Licence Professionnelle « Métrologie en mesures Environnementales et Biologiques », Villeneuve d’Ascq, [email protected]
(2) AGLAE Association, accredited proficiency testing provider, [email protected]
Goals: - highlight and quantify the differences between analytical methods (trueness and precision)
- assess the impact of these differences on the decision making for regulations
Data Results from proficiency tests organised by AGLAE between 2005 and 2010
Analysis of about 20 metals (Al, As, B, Ba, Be, Cd, Cr, Cu, Fe, Hg, Mn, Mo, Ni, Pb, Sb, Se, Sn, Sr and Zn)
Repeated tests at different concentration levels (around 12)
High number of laboratories who participated in the tests (about 120)
Results
Impact on the regulation values
No significant differences between clean waters and waste waters Major tendency: ICP-AES < ICP-MS < AAS in oven 2 exceptions: Sn and Se, for which AAS in oven << ICP
Way of data processing
For the differences between results (trueness): ANOVA with normally distributed random variables
For the deviations between precision values: for each analytical method, a model of reproducibility variations (CVR%) is calculated according to the concentration level
Deviations between results (trueness)
Clean waters
0,069 0,149 0,261
0
5
10
15
20
25
30
35
40
45
0 0,1 0,2 0,3 0,4 0,5 0,6
Deviations between precision values
Major tendency: AAS (in oven and in flame) reproducibility is less satisfactory than ICP (AES and MS) reproducibility ICP-MS is more reproducible than ICP-AES for low concentration levels No significant difference between clean waters and waste waters
CVR%
Concentration in ppm
Example of the analysis of Al
in clean waters
Waste waters
Example of the analysis of Al (AAS in oven)
in clean waters
Al
As
Cd
Cr
Ni
Pb
Sb
Se
Sn
0,90
0,92
0,94
0,96
0,98
1,00
1,02
1,04
1,06
ICP-AES ICP-MS SAA-four
As
Cd
Pb
Se
Sn
0,90
0,92
0,94
0,96
0,98
1,00
1,02
1,04
1,06
ICP-AES ICP-MS SAA-four
Re
lati
ve t
rue
ne
ss m
/ M
Methods
Doubt zone: zone in which the risk to misclassify a sample as ‘compliant’ or ‘non-compliant’ is higher than 5%
The more reproducible the results for an analytical method are, the less expanded the doubt zone around the regulation value will be The analytical methods which give higher results reduce the risk to classify as ‘compliant’ a ‘non-compliant’ sample The analytical methods which give lower results reduce the risk to classify as ‘non-compliant’ a ‘compliant’ sample
AAS in oven
AAS in oven
AAS in oven ICP-AES ICP-MS
5% of risk to classify a sample with a concentration level equal to the upper bound as ‘compliant’ whereas it is not
5% of risk to classify a sample with a concentration level equal to the lower bound as ‘non-compliant’ whereas it is compliant
Regulation value
Lower bound
Upper bound
0
5
10
15
20
25
30
35
40
45
0 0,1 0,2 0,3 0,4 0,5 0,6
95% confidence level
chosen model