In general, every measured value is subject to uncertainty. A measurement result as an approximate value for the actual value of a measured variable must therefore also take into account a measurement uncertainty. This limits a value range within which the actual value of the measured variable lies with a certain probability. In our case, this probability is 95 % (k=2) by using the “expanded, combined measurement uncertainty”, represented by “U”. DIN EN ISO/IEC 17025:2018-03 requires the specification of a measurement uncertainty, e.g. if we are to confirm conformity with a regulation/standard.

Measurement uncertainties determined in the IGR

You can assume the following maximum measurement uncertainties for the analyses received from the IGR, unless the test report contains detailed information on the measurement uncertainty. If individual measurement uncertainties are specified, these naturally apply.

a) Analyses using ICP-OES according to DIN 51086-2 (specification of U in rel. %)

• Concentration range 100 % – 10 %: 4 %
• Concentration range <10 % – 1 %: 8 %
• Concentration range <1 % – 100 ppm: 20 %
• Concentration range <100 ppm – 10 ppm: 30 %
• Concentration range <10 ppm: 120 %

b) Analyses using SEM-EDX according to DIN ISO 22309 (specification of U in rel.-%)

• Concentration range 100 % – 1 %: 30 %
• Concentration range <1 % – 0.1 %: 100 %

Explanation: Concentrations < 0.1 % are not stated in the test report for SEM-EDX examinations

c) Density determination using the H. Scholze method … (specification of U in rel.-%)

• Concentration range 2.4700 g/cm³ – 2.5600 g/cm³: 0.05 %

Explanation: The concentration range includes standard soda-lime glass and crystal glass and applies to pure glass without glass defects such as air bubbles or impurities. The standard deviation is of course well below the above-mentioned measurement uncertainty.

Background information

When determining the measurement uncertainty, all relevant sources of uncertainty must be taken into account. There is no prescribed procedure for determining the measurement uncertainty. We use software for this purpose which, in addition to reproducibility within the laboratory, also takes into account the multiple analyses of a certified reference material. In practice at the IGR, the expanded, combined measurement uncertainty is strongly influenced by the quality of the certified reference material, as reliable standards are only available on the market for a few of the materials we analyse.

And when is the standard deviation used?

The specification of the standard deviation ‘σ’ required by many bodies is a statement on precision (e.g. for multiple determinations). If a body requires the specification of the standard deviation ‘σ’, this is to be understood as an additional requirement (in accordance with DIN EN ISO/IEC 17025:2018 paragraph 5.4).

Example of measurement uncertainty

Imagine you have to comply with a limit value of 100 ppm for lead in the product for four samples. We analyse 80, 95, 105 and 120 ppm. So two samples are OK and two are not, right? No, not necessarily. If the measurement uncertainty is correspondingly high, it may be different.

For our example below, let’s assume that we have a measurement uncertainty of 10% (relative to the value analyzed). This results in the following possibilities: