Limit of detecion and Limit of quantification (Case study)

[along]

Dear all,

Recently, I tried to estimate the limit of detection of three organophosphorus pesticide namely methamidophos, acephate and monocrotophos by carried out a series of injection (5ppb, 10 ppb, 15 ppb and 20 ppb). I found out that at the concentration of 5 ppb (standard in blank matrix), I am not able to detect any peaks and at 10 ppb, the peaks were in good shape with S/N ratio of 23.6, 14.2 and 3.7 respectively.

Can I concluded that my LOD was 10 ppb?

I did carried out recovery test at 10ppb level and the results were pretty good >90%, similarly no peaks were detected for the recovery test at 5 ppb level.

My question is:

How should I express my LOD and LOQ?

Thanks and regards.

[along]

Dear all,

Someone did mentioned to me before the method to estimate LOD from the standard calibration curve, can anyone please let me know where to locate the literature or paper that discussed on this issue, I need more information on how to derive LOD from calibration curve.
Thanks

[Peter Apps]

Hi Kenny

The signal: noise ratio is a parameter from a single run, and it gives you a hopelessly optimistic estimate of the LOD or LOQ of a method.

The LOQ depends on the repeatability (or reproducibility) of the whole method. Even if you are just injecting standards as you did, this includes the injection repeatability as well as the signal:noise. For a complete pesticide method it includes sample homogeneity, extraction, clean up, and concentration of the sample, on top of the calibration and the injection and signal:noise.

You can test this for yourself with the solutions that you have: do a series of seven injections and calculate the relative standard deviations of the peak areas. You will find that the rsds are higher than the S:N figures that you give. For the peak with an S:N of 3.7 the expectation is that up to three of the chromatograms will not have peaks above the noise level.

Please do the forum a favour and post the results if you do this test. It is a pet subject of mine since I pulled two back to back all-nighters to redevelop a method for which someone had estimated repeatability by measuring S:N on a chromatogram !!

Peter

[jbennett]

Hi Kenny,

The discussion about LOD and LOQ is one that has been going on for a long time, and is one that is sure to be controversial. Many USEPA environmental methods specify the procedure listed in 40CFR136 to determine the Method Detection Limit (MDL), and the LOQ is often a multiple of the MDL. From a practical standpoint, the LOQ is the lowest calibration standard.

The 40CFR136 MDL is defined as the level where you can be 99% certain that the analyte will be detected. Over the years there has been much discussion that this procedure does not work well in practice, and that it is actually a procedure that approximates Currie's Critical Level and not a Limit of Detection. EPA has set up a workgroup consisting of experts in the field to recommend a 'better" way to determine MDL's and they have been meeting for about a year.

Hope this helps.

Jack