LAWR

13 Annual Guide 2015 | SOURCE TESTING ASSOCIATION e limit of detection is expressed as the mean analytical blank value (bave) plus three times the standard deviation of the analytical blank (sb). LOD = bave + 3 sb where LOD is the detection limit; bave is the mean analytical blank value; sb is standard deviation of the analytical blank. NOTE 3 In this document the limit of detection should preferably be calculated from the analytical blank bave. If this is not possible, the limit of detection can be calculated from the signal to noise ratio according to 8.1 of EN 1948-3:2006 (resp. 10.5 of this document)." e reference to ISO 9169 is quite inappropriate as this is a continuous monitoring standard and doesn't apply to a chromatographic system! Conclusions In reality, with so little specificity, laboratories perform their own process for estimating the LOD and LOQ. However as an anonymous survey of UK laboratories by the Environ- ment Agency (EA) found, reported LODs could vary by up to two orders of mag nitude for the same parameter. A critical aspect of many emission measurements is that the determi- nands of interest, by their nature are present ubiquitously in the environment. is leads to the real LODs being more often than not determined by blank contributions and vari- ability in these background levels. Laboratories take extensive precautions to minimise the risk of contamination but often instru mental detection limits are dominated by these blank values. is reinforces the need to ensure that the sampling process and method performance combine to establish that the critical level of interest is at least 10x the LOD. It is equally important to realise that the uncertainty of measurement close to the LOD will be much higher than would be obtained at higher levels. As an example, if a value is reported as 0.01, but the LOD is 0.01, then the confidence in the result being 'real' is very poor. I have seen customers over-interpret results in this region, saying perhaps that 0.02 is twice 0.01 (when in reality these could be the same result). e use of the LOQ helps here in that the precision of the method is significantly better at this higher level. e critical level of interest should be at least at the LOQ level. e STA laboratory group worked with the EA and UKAS to propose a common approach to the determina- tion of LOD. A draft document was prepared that required paired blanks (in the media of interest) be analysed inde- pendently in ten batches. A statistical analysis of the data then yields a robust LOD. If no analytes are detected in the blanks, low spikes are used (circa 3-5 times the expected LOD). is approach was first developed in the UK drink- ing water industry, and was made a mandatory aspect of the EA's MCERTs scheme for soil and water testing. Interest- ingly, when this approach was mandated, laboratories that could 'magically' achieve LODs an order of magnitude better than others, suddenly changed to reporting similar values! Unfortunately neither the EA nor UKAS are able to prescribe this approach for stack work, so at present the draft is yet to be taken further. David Wood STA Laboratory Group Chairman SAL Laboratories BSEN1948-4:2010 Stationary source emissions — Determination of the mass concentration of PCDDs/PCDFs and dioxin-like PCBs Part 4: Sampling and analysis of dioxin-like PCBs "3.10 Limit of detection LOD – minimum value of the measurand for which the measuring system is not in the basic state, with a stated probability NOTE 1 e detection limit, also referred to as capa bility of detection, is defined by reference to the applicable basic state. But it may be different from "zero", for instance for oxygen measurement as well as when gas chromatographs are used. [Adapted from EN ISO 9169:2006, 2.2.10 [12]] NOTE 2 e measurement value can be distinguished from the analytical blank value with a confidence of 99%. 6.3.2 Detection limit 6.3.2.1 General e detection limit shall be less than 0.04 ng/m 3 6.3.2.2 Determination based on laboratory filter blanks Determine the detection limit from the standard deviation of at least 10 laboratory filter blanks using equation below. where S lfb is the standard deviation of laboratory filter blanks in ng; m – is the mean of laboratory filter blanks in ng; m i is the individual filter blank in ng; n is the number of analysed filters e minimal detectable mass of BaP is calculated using the equation below: D M = t n–1;0,95 x S lfb where D M is the minimal detectable mass of BaP in ng; t n–1;0,95 is the Student factor for n measurements and a 95% confidence interval; S lfb is the standard deviation of laboratory filter blanks in ng 6.3.2.3 Determination based on the signal-to-noise ratio Perform a chromatographic analysis without injecting any solution. Keep the chromatographic parameters as used for the calibration and detection of BaP. Calculate the detec - tion limit as three times the average of the height of the noise at the retention time of BaP +/- ten times the peak width at half peak height at the lowest calibration level. Σ S lfb = (m – – m i ) 2 n – 1 n i=1

• Cover