In HPLC why we do not use other ratios? For example, in case of LOQ, why not the ratio of 5, 6, 7, 8, 9 considered?
3 sigma for LOD depends on the theory of the normal distribution: from wikipedia (https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule): In the empirical sciences the so-called three-sigma rule of thumb expresses a conventional heuristic that "nearly all" values are taken to lie within three standard deviations of the mean, i.e. that it is empirically useful to treat 99.7% probability as "near certainty". The "three sigma rule of thumb" is related to a result also known as the three-sigma rule, which states that even for non-normally distributed variables, at least 98% of cases should fall within properly-calculated three-sigma intervals.
For limit of quantification there are other definitions. It is an operative quantity. 10 (or 9) are conventional figures. see attached files, especially EURACHEM guideline
3 sigma for LOD depends on the theory of the normal distribution: from wikipedia (https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule): In the empirical sciences the so-called three-sigma rule of thumb expresses a conventional heuristic that "nearly all" values are taken to lie within three standard deviations of the mean, i.e. that it is empirically useful to treat 99.7% probability as "near certainty". The "three sigma rule of thumb" is related to a result also known as the three-sigma rule, which states that even for non-normally distributed variables, at least 98% of cases should fall within properly-calculated three-sigma intervals.
For limit of quantification there are other definitions. It is an operative quantity. 10 (or 9) are conventional figures. see attached files, especially EURACHEM guideline
As Dr. Polesello points out, 3 times signal to noise (S/N) has become the de facto standard for LOD since it provides as close to a statistical certainty as you can reasonably expect. Back many years ago it was common to see 2 sigma calculations (2 times signal to noise) for LOD from instrument manufacturers, but those have gone by the wayside. If your noise is well defined (such as in chromatography) then it certainly is justifiable to consider 2 x S/N as your LOD.
10 times S/N is only one definition of LOQ. Note that LOQ is not as readily defined as is LOD For example, in regulatory work it is common to see LOQ defined as your lowest calibration standard, which is not the same as 3.3 times LOD. If your signal is clean then you can certainly run a standard lower than 10 x S/N, which would give you a lower LOQ.
Three and ten times baseline noise for LOD and LOQ always give optimistic values for the performance of an analytical method because they depend only on the separation and detection step, and ignore variation in sample preparation and injection.
Much more robust and realistic estimates for LOD and LOQ come from the uncertainty of the calibration, and the best estimates of all are from repeated analyses of real samples with demonstrated relative standard deviations in results of better than 33% and 10 %.
I generally agree with Peter. For LOQ we can fix a minimum acceptable uncertainty. You have to plot uncertainty respect to concentration and find at which concentration you get the acceptable uncertainty. Which uncertainty? depends on the aim of the method. Validation is demonstrating that the method is fit for purpose. For Water Monitoring in EU we stated an extended method uncertainty of 50% at EQS level (with a coverage factor of 2). This is also the approach of EURACHEm guideline. For LOD I prefer the method linked to the standard deviations of the background, linked to the statistic of detection. The measurement of blank concentrations in some analytical techniques can be difficult as the instrumental software or hardware may impose peak detection threshold values or peak smoothing algorithms etc., which suppress small signals. This occurs most often for chromatographic methods. When this situation is encountered, it is normal to artificially increase the signal using one of the following methods:
• Use a real sample containing a very low, but measurable concentration of the analyte.
• Fortify a sample that contains no analyte to a very low, but measurable concentration.
• Dilute a sample extract containing a higher concentration of the analyte to achieve the required very low but measurable concentration.
It should be noted that when uncorrected blank signals are used to calculate the limit of detection, increasing the absolute concentration of the blank as above will inevitably produce a higher value for the estimate of the limit of detection.
- Badges
- Science method