Dear Mahesha Manjunatha Poojary

Are you fallowing standard method or a method developed by yourself?

Anyway, fist you have to optimize the GC Chromatogram. To perform this,

1. You have to create good GC program (temperature program) to separate the mixture of compounds which you are looking for analysis.

2. You have to estimate the instrument detection limit. To perform this,

a. you have to measure the noise level of the instrument.

b. Estimate minimum amount of analyte (compound) correspondent to five time of the noise level. (This may the estimated method detection level (EMDL))

3 calibration curve

a. Prepare series of calibration solution using standard compound (from EMDL).

b. Inject them into GC and plot response against the concentration.

c. Identify linier dynamic range.

4 Determination of method detection limit (MDL)

At present, MDL is the single value applied to report minimum level of analyte that can be reported with 95% confident limit (so, no need to define both LOD, LOQ). Please refer the attachment to determine MDL (this example is for water, but you can apply for all sample metric)

5 recovery study

You have to introduce known amount of compound to your sample metrix and analyse the sample using experimental procedure and calculate the recovery level (this should be within 80% to 120%)

If you complete these steps you can validate your method.

Piyal

Hi Mahesha,

I'd ask first, why are you validating the method and what's important in the analysis? I  used to work in a pharmaceutical analytical lab and we operated a sliding scale of validation from a simple check for LOD to full ICH validation, depending on the type of analysis.

Thanks

Mark

To develop a method for quantification of compounds in the sample you need to know the limitation and quality of the method.

1) you need to know how low you can see the compound in the final extract (LOD, that you can distict the presence of the analyte from the background), how low you can quantify with good confident (LOQ, you are confident on quantity you declare the quantity in the sample 99% of the time) .. You need to know how true of the amount of anlyte in the sample (accuracy, you know how much you added and see the result how close to the expected value). You need to know how precise of the method e.g., get the value close to the expected value no matter how many times you do. Since you you calibration standard to estimate the amount in the unknown by comparing the response vs. the expected amount, you need to know how good the relationship of response vs concentration (linearity). The calibration curve that you construct is base on the linear relationship and it may have limitation where if the response of the unknow is beyond this point, the result may not be accurate. Say you make a calibration curve of the detector response of standard from 1 to 100 ppm, if the response is falling in this range, you will be confident to determine the amount of anlyte in the sample in this range. If the resonse is 2 times more than the response of standard 100 ppm, the amount of unknow may not be 200 ppm because it may not be linear and you did not prove it.

2) for the calculation, I would use the lowest standard in the curve that you can see the good signal above the noise and inject it several times (minimum 7) and determine the standard deviation. The LOD is 3 x std dev and LOQ is 10 x std dev. You know the concentration of the analyte in that standard solution (wt/vol) then you can convert to the concentration of analyte in the sample (wt/wt) based on how you prepare the sample. To calculate the accuracy, you spike the samples with the know amount at least 3 reps and determine the mean of the amount found vs the expected x 100. Good accuracy is 80-120%.  You determine the standard deviation of the result and calculate the relative standard deviation (RSD) by dividing standar deviation with the mean x 100. RSD is the the precision (good method is less than 20%)

Hi, I am using the method stated by others (but not optimized). Now I am optimizing the same method with respect to few variables such as temperature and volume.  

Again, whatever you do with the final method, you still need to prove all that parameters, e.g., accuracy, precision, LOD, LOQ, linearity and so on.

Thanks Narong for your answers. They are really very useful. Now my question is while spiking which kind of matrix I have to use? Can I use a simple solvent such as water?

Regarding the Matrix, you will have to use the same matrix as the one in which your sample is resuspended or diluted in. In other words, it is the sample solvent without the actual active ingredient that you want to measure. Hope this will help. 

Many thanks Dr. Ruan. Actually, I am analysing the compounds which are dissolved/diluted in isoctane. However, when you consider the application of the developed method, it can be used for an oil sample containing the same analytes which I am studying (dissolved in isooctane). Now tell me the matrix for spiking should be isooctane or oil?

If I understand it correctly your samples will be in isooctane, if this is correct isooctane should be used as the matrix.

Yes, Thank you very much Dr.Ruan.. 

. 

Your sample is in Isooctane, but what has happened before that. Say your "real" sample is oil and you extract and do the cleanup process and final solvent is isooctane, you may extract other stuff from the sample and follow thru the final extract. In GC, these matrix will affect the vaporization and stability of your analyte in the "hot" injector port and will "protect" your analyte from degradation. If you inject only analyte in solvent, your analyte will degrad faster in the injetor port because it does not have matrix to protect. The result is you will get higher recovery than expected. You can easily check this out by making the blank matrix (sample with no analyte goes to the sample prep to the end. Then spike analyte in the matrix and the clean solvent. Inject both and compare the response, in theory, you should have the same response, but in reality it may not.

Thanks Narong. I think for recovery studies it will be better to spike the analyte over a matrix.

If you need a good method, you need to do it. You cannot say you can run a marathon by using the treadmill in door. You need to run on the real track.

He he he perfect!  :) :)

All test methods even standard methods shall be validated, the validation experimental design depends on nature of analyte and concentration range, accuracy, precision and linearity shall be studied but LOD and LOQ only measured in trace and ultra trace analysis. you can find calculation methods for all parameter in IUPAC guide lines