Chromatography Forum

Hello. I am trying to setup glycol analysis method. I need to know how to determine MDL. What are the steps involved. I have already created the calibration curve. How to determine MDL. Thank you

Keep injecting lower levels until you cannot get required S/N.

I assume that you are developing/validating a new test method, otherwise there's be no need to establish MDL.

On the other hand, my supervisor demanded we establish MDL for an ethanol assay we were validating where the active level was 62%. So MDL was a zillion times lower, so very insignificant. But he was fearful of writing a justification, so we just did it (was a few ppm).

USP monograph for glycerin has test requirements for ethylene glycol/DEG contamination at under 0.1% (which in my opinion is silly, as anyone adulterating glycerin to make money would adulterate at much higher levels).

Keep injecting lower levels until you cannot get required S/N.

I assume that you are developing/validating a new test method, otherwise there's be no need to establish MDL.

On the other hand, my supervisor demanded we establish MDL for an ethanol assay we were validating where the active level was 62%. So MDL was a zillion times lower, so very insignificant. But he was fearful of writing a justification, so we just did it (was a few ppm).

USP monograph for glycerin has test requirements for ethylene glycol/DEG contamination at under 0.1% (which in my opinion is silly, as anyone adulterating glycerin to make money would adulterate at much higher levels).

Lower levels of what. The highest concentration in our samples is 100ppm. So what do i inject to get MDL. How will i know i have the required signal/noise ratio.

Lower and lower concentrations of your analyte in your matrix. If it's glycol in water, you need to make standards that are lower and lower in glycol concentration until you get to where your signal is reliably just above the noise. The "best" rule of thumb that most of us use is S/N ratio of 3 for limit-of-detection.

With a mass spec detector, I've had instances where a S/N = 1.75 was good enough for me to be able to tell that what I was seeing was really the analyte. BUT, that was with a very powerful detector. When you're using a flame, the S/N of 3 is probably a good way to go.

Lower and lower concentrations of your analyte in your matrix. If it's glycol in water, you need to make standards that are lower and lower in glycol concentration until you get to where your signal is reliably just above the noise. The "best" rule of thumb that most of us use is S/N ratio of 3 for limit-of-detection.

With a mass spec detector, I've had instances where a S/N = 1.75 was good enough for me to be able to tell that what I was seeing was really the analyte. BUT, that was with a very powerful detector. When you're using a flame, the S/N of 3 is probably a good way to go.

Like how low we are talking for example 0.1 PPM, 0.2 PPM etc

40 CFR Appendix B to Part 136 - Definition and Procedure for the Determination of the Method Detection Limit-Revision 1.11]40 CFR Appendix B to Part 136 - Definition and Procedure for the Determination of the Method Detection Limit-Revision 1.11 is the official EPA method. Its been updated a few times. I prefered the earlier versions based on the standard deviation of first method blanks and then mdl spiked method blanks. Here is a pdf link Definition and procedure for the Determination of the Method Detection Limit, Revision 2

40 CFR Appendix B to Part 136 - Definition and Procedure for the Determination of the Method Detection Limit-Revision 1.11]40 CFR Appendix B to Part 136 - Definition and Procedure for the Determination of the Method Detection Limit-Revision 1.11 is the official EPA method. Its been updated a few times. I prefered the earlier versions based on the standard deviation of first method blanks and then mdl spiked method blanks. Here is a pdf link Definition and procedure for the Determination of the Method Detection Limit, Revision 2

OK i will check this out. Thank you

Like how low we are talking for example 0.1 PPM, 0.2 PPM etc

That's the question only you can answer and that's exactly what we're telling you to find out.

Make a solution at 10ppm, can you easily differentiate the signal from the noise? Great! Now make 5ppm and do the same, and test smaller concentrations until YOU find the concentration that you would feel comfortable stating is the MDL based on what you are able to observe and quantify.

Zoraku has summed up the repeated suggestions of the other posters perfectly!

You can always post chromatograms if you need further help on the S/N as you approach the MDL

Like how low we are talking for example 0.1 PPM, 0.2 PPM etc

That's the question only you can answer and that's exactly what we're telling you to find out.

Make a solution at 10ppm, can you easily differentiate the signal from the noise? Great! Now make 5ppm and do the same, and test smaller concentrations until YOU find the concentration that you would feel comfortable stating is the MDL based on what you are able to observe and quantify.

OH OK Thank you

Zoraku has summed up the repeated suggestions of the other posters perfectly!

You can always post chromatograms if you need further help on the S/N as you approach the MDL

Thank you

Hi,

Since you have a calibration you could always use the alternative method based on the ICH's Standard Deviation of the Response and the Slope. The methodology is explained on page 4.24 of the Fusion QbD Method Validation Users Guide http://http://www.ciinformatics.co.uk/? ... ent_id=647

The Limit of Quantitation can be calculated similarly (page 4.25). Typically these values are determined during a combined experiment that determines the methods Accuracy, Linearity, and Range. If you are validating the method you will need to determine these as well.
For more information on Fusion QbD Method Validation software start here http://www.ciinformatics.co.uk/products ... -software/

Kind regards,

Bob