Calibration GCMS?

[pollypocket78]

Hello everybody,

I would like an advice about calibration of my GCMS.

My activities are : Analysis of complexe mixtures of volatiles (determination of the molecule + % area) and simple QC control of simple or mono-molecule components, using again my MS library to check if it is the good molecule, and check the %Area of the main peak.

I absolutely do not use quantitative method for my mixtures.

Before any analysis, every day, I perform an autotuning, to tune the machine.

Do I need to create a calibration method for my GCMS? If yes, how?

My company is going through the process of ISO certification, would this calibration be detrimental to it?

Thanks very much and looking forward to hear about answers...

[Don_Hilton]

What is the objective of your analyis and can you demonstrate that that the objective is met with a desired degree of certanty?

[pollypocket78]

My objective is double :

- QC chemicals and oils if needed

- Make an accurate analysis of a complex mixture, determining the different molecules present in it + their %Area (no precise quantification).

For the ISO part, they want that all the machines present in the QC process will be calibrated... Is autotune enough considering that I don't pay attention to precise quantification, (not using quantitative methode of the GCMS) ?

[Don_Hilton]

You will need to be more specific about what is meant by QC.

You are testing for some specific property or properties of the chemicals and oil to determine that some particular quality is met for all and/or each. For ISO you will need to state that - and then demonstrate how you determine that the instrument is capable of meeting of providing correct answers.

If you use area percent, you use quantification. If a decision is based on the number, you need a way to determine that the numbers are good enough for use in deecision making. How big an error in that number would be a problem? Can you demonstrate that your numbers are that good? Can you get the same results today that you got yesterday and will get next month - thus providing consistant data for consistant decisions and consistant product quaity?

While autotune will give an indication of the operating condition of the mass spectrometer, it says nothing about the condition of the GC. An injection of some standard mixture to show that a compund can make it from the GC vial to the detector and give expected signal to noise and a library searchable spectrum may be adequate to demonstrate instrument sensitivity. If you are looking for specfic kinds of compunds, one of that kind of compund might be a good QC check. This would be similar to the intstallation qualification check sample.

If I were auditing a method, my concern would not be the mass spec, but rather the GC. It is easy to foul an inlet or column so that it loses sensitivity or resolving power. And in the worst case, the losses are selective for particular compunds. If your analysis is for just any kind of contaminant, the Grob mixture might be a good check.

You do not have to run the checks every day, but if a check fails data since the last passed qc are brought into question or are invalidated. You need to run the checks often enough that 1) you do not have too much data invalidated by a bad QC and 2) the knoledge of an instrument problem is timely so that corrective action may be taken with regard to materials that would have been incorrectly QC'ed based on poor instrument performance.

[pollypocket78]

Dear Mr Hilton,

Thank you so much for your time and concern.

I am a young engineer in the Fragrances industry, working in Asia for a local company, handling alone the analytical service and it is sometimes quite questioning!!

So, basically, When we receive Raw materials for the fragrances mixing, we do organoleptic tests (olfaction), if it is fine, then the raw material is validated, if there is a doubt, I use the GCMS.

I usually check if the peaks are ok, that it is the good molecule, and the area %, to see the possible impurities.

I use the machine to analyse complex fragrances as well, but not in a QC objective, that's why I am less concerned about quantitation (on a MS point of view).

I totally agree and understand that I would need to have a calibration method to test the QC part, as I have no idea if the area% are relevant or not, and our wish to go through ISO process will lead me to create this method.

But I have no idea about what mixture to inject, and how to perform it. Again on a GC quantitative point of view.
My idea was maybe to weigh 10 molecules that we use regularly, choosen to cover the spectrogram of my analytical method.
But do I have to have them being extremely pure (ordered from an organic synthesis lab instead of taking our own materials which quality is sometimes questionable)?
And how do we perform this calibration? Different concentrations, do we use the MS, I have no idea :S

Thanks a lot again it is nice to be able to exchange with experimented professional specialists...

[Don_Hilton]

Keep it as simple as you can. If you wiegh out some compunds and mix them, just purchase them from a good vendor and and specified purity. 98% pure or better should be good enough. (You specify minimum purity in your method.)

Take a look at the Grob mix. While these are not flavor related compunds, it has a wide variety of functionality and you can purchase this premade with a certificate of analyis from most or all chromatographic suppliers. For your method,you inject at a concentration where you can easily see peak shape and measure peak areas. The area for the nitrophenol and nitro aniline should be constant. This tests for acid/base sensitivity in the column. Peaks should not tail. Specify limits that allow your column to age a bit.

If you are not familiar with the Grob mixture, look around for it. I did a quick search and this link has some descritpion of the kinds of things the mixture tests: http://www.griffinanalytical.com/app_note_6.html. The article is part of the promotion for a particluar instrument - but it does describe what the grob mix tests.

Use functions in the instrument software to measure peak tailing, signal to noise, and to obtain numbers for peak resolution. Peak resolution will probably not be computed by your instrument software, but you can make a spreadheet to compute resolution based of peak widths and retention times.

[pollypocket78]

Thanks a lot for your infos and sorry not to have kept you update, but I was busy.

I think I will indeed weigh myself a mixture of synthetic molecules, with a minimum purity of 98% for each (based on certificates of analysis from suppliers). Then inject the mixture at different concentration, to be able to define the ideal one.

Then I will see what numbers I can get from my software concerning peak tailing and resolution, and define a good range within the calibration is acceptable. Do I need anything else as parameter?

About quantification, what do I do exactly? Knowing the exact weigh of each components, and their purity, how do I infer from my mixture spectra that the calibration is ok? By measuring the area%(and infering from that an actual %), that I will compare to the actual % that I really weighed? And after that maybe define a range that allows me to say it's good?

Thank you

[Don_Hilton]

If you weigh a mixture of molecules, I would suggest that you inject at a particular concentration and measure the area percents in the TIC over a specified mass range ( you pick it it might be 45 to 350 ) and look to have the ratios of areas remain constant within some reasonable percentage. If you pick something like 10%, it will be a specification that will be difficult to miss. If you make a calibration curve, use one compund as an internal standard - same concentration in all solutions and make up solutions with a few other compunds and show that when you build the calibraiton curve it is linear. The specification might be an r value of .95 or better.

Try a few runs over several days. And as you collect data you can see if the specifications are too tight. If they are much wider than what you obtain from your instrument, do not tighten the specifications unless it becomes clear that the specifications are wide enough to allow unacceptable instrument performance be accepted. As the instrument ages, it may be difficult to meet specifications set with a newer instrument.

[pollypocket78]

Thank you very much for your answer and concern. Couldn't reply before because I was not notified of your reply in my mailbox, don't know why

Indeed I will try to make trials with internal standard, will let you know.

I think making calibration curves is important to at least determine the efficiency of the figures provided by the instrument.

Thank s a lot again and have a good day!