calibration curve in GC-MS

[sauder]

Hi,


I noticed a phenomenon that is quite difficult for me to understand:
Sometimes, when I make a calibration curve, in the same range of concentration (1-200 ppm) and with the same system (GC with mass detection), my calibration curve is linear when I use a polar colum, and quadratic when I use an apolar one.
It seems to be compound dependant because I don't see that for all the compound I have tested.
Have you ever heard of such phenomenon, and if you did, is there an explanation of this thing ?

thank you for your answers.


Gabriel CAZORLA
Ingénieur de recherche
SCP

[GOM]

Hi Gabriel,

My first thought is that, given it is the same injection port and detector, then you may have adsorption sites on your non-polar column. Try confirming by running an alkane standard set on both columns and see if the problem goes away?

Regards,

Ralph

[sauder]

Thank for your answer.

Just to illustrate what I'm talking about, here is the columns I have used:

Apolar column : 100% PDMS, 60*0.25mm*0.25µm
Polar column: PEG, 60*0.25mm*0.25µm

With the apolar column and for this compound : farnesol (CAS 4602-84-0) I obtained the most beautiful quadratic curve, and for this one : limonene (5989-27-5), the calibration is linear. When I used the polar one, this 2 products give a linear calibration curve.

I would like to test another apolar column, just to confirm the result I have.
Is it common with apolar column to have adsorption sites? I mean for the compound I have tested, I don't see any difference in the shape of the peaks. I thought if we had some particular retention behavior I would observe some peak tailing, or shouldering peaks... But as I don't have a great knowledge about the retention process in GC, I'm certainly don't take this problem under the right angle, so every tips are welcome



Best regards


Gabriel

[Peter Apps]

Hi Gabriel

Yes. it is strange that you see symmetrical peaks if adsorption is the cause of the quadratic fit.

What injection conditions are you using; split or splitless, and if split what is the split ratio ?.

Could you post the calibration graphs so that we can see where the deviation from linearity is.

Are you calibrating by peak area or peak height ?. If peak height, the problem may be that 200 ng (1 microliter splitless injection) of farnesol is ocerloading the non-polar column. Could you post a chromatogram - instructions of how to do it are in a sticky at the top of the LC forum.

Peter

[GOM]

Hi,

Unlike limonene, farnesol contains an OH group, which is consistent with the problem being due to adsorption.

Does the curve look different if you dilute your calibration standards by a factor of 5? Alternatively, assuming that you have no carryover, what does your curve look like if you inject your 200ppm standard in between each of your other calibration ones?


Regards,

Ralph

[Peter Apps]

The -OH group on the farnesol will make it more susceptible to both adsorption, and overloading on the non-polar column. Adsorption will give deviations from linearity at the bottom of the calibration, overloading will give deviations at the top, especially if peak heights are measured rather than areas.

What kind of MS are you using ?, in ion traps space charge effects (or the automatic changes in conditions to reduce them) can reduce the response at higher concentrations.

Peter

[sauder]

Thank again for all your answers

I use a Simple Quad MS system (agilent GC6890 with a 5973 MS detector)
My split ratio is 1/50 and I calibrate with peak areas.

For the chromatograms I will post it as soon as I understand how to do
I guess that it is not very difficult but I never done this ....

Could you give me some publications which deals about retention process in GC/MS with apolar column and -OH effects in such conditions?
Regards


Gabriel

[GOM]

Hi Gabriel,

Out of interest could you do a peak symmetry test for us on your lowest level farnesol calibration peak from the non-polar column?

Zoom into the peak on screen and print it out (window 2 on the 5973 ChemStation I think).
Draw in the baseline under the peak. Now drop a vertical from the peak apex to the baseline. Draw a horizontal line across the peak at 5% of the peak height. Now measure the distance along this line from the front edge of the peak to the vertical line (A). Next measure the distance along this line from the vertical line to the back edge of the peak (B).

From this the tailing factor is (A+B)/2A. If it is 1.0 then you have a symmetrical peak.

This link may help http://www.sonoma.edu/users/e/eckd/chem ... a_peak.pdf

Thanks,

Ralph

[sauder]

Hi,

I calculated the peak symetry for the farnesol with the method you gave me.

For the polar column the peak symetry for the farnesol is 1.105 and for the apolar one the peak symetry 1.35

Is it difference significant?


Thank you very much



Gabriel

[GOM]

Hi Gabriel,

Thank you for doing that. I am not ruling anything else out - I am just trying to confirm or eliminate adsorption. As Peter requested it would still be useful to see your chromatogram and calibration graphs (or even the data for them).

The tailing factor that you have calculated does tell us that there is a slight difference between the columns and that your non-polar column is giving less symmetrical slightly tailing peaks. Could you also do a similar thing at 10% of the peak height only this time just give us the ratio of A/B? I'll have you cutting out and weighing peaks yet!

Can you just re-assure me that nothing else was changed apart from the column?

If we still assume adsorption and that the cause is your column and not your injection port then things to try are (in no particular order)

1. New column
2. Take 2 metres off your column - in case it has contamination at the front end.
3. Inject 1000ppm (at 50:1 split) between each calibration to saturate any sites. It can work as long as you just check first that there isn't carryover.
4. Another trick that I have used( but it depends on the analysis) is to add another component to your sample at quite a high level. This component is selected to contain a functional group and acts in a sacrificial way to bind adsorption sites.

Now you are going to tell us that you found out that you diluted your standards incorrectly

Regards,

Ralph

[chromatographer1]

Another possibility is that the surface of your apolar column has sites which may be catalytically dehydrating the farnesol to a degree (injection temperature can contribute to this reaction). As you put more sample on the column this would give a higher response at the detector per sample concentration, a curved response plot.

The shape of your peaks is expected from the description you give. One would expect to see see some peaks from the dehydration reaction unless dimerization was the preferred end product reaction and then the peaks, if any, might be very late to elute and not seen in your normal oven programming.

Just a shot in the dark.

best wishes,

Rod

[sauder]

Thank again for your help.


@GOM

If we still assume adsorption and that the cause is your column and not your injection port then things to try are (in no particular order)

1. New column
2. Take 2 metres off your column - in case it has contamination at the front end.
3. Inject 1000ppm (at 50:1 split) between each calibration to saturate any sites. It can work as long as you just check first that there isn't carryover.
4. Another trick that I have used( but it depends on the analysis) is to add another component to your sample at quite a high level. This component is selected to contain a functional group and acts in a sacrificial way to bind adsorption sites.

1. I will test another column as soon as possible. I think I will test another manufacturer juste to make a comparaison
2. I'll try this too
3. For my calibration, I use a sample which consist of 26 products (24 allergen and 2 IS). Maybe 10 of them have some hydroxy groups. Which products can I use to saturate the actives sites? For the carry over : between each standards we make a blank, and we never see some strange peaks in this blank run.
4. I think it's the more easy solution to test, as I said in point 3 I already use a quite complex sample, so what kind of coumpound may I use to make this saturation?


For the determination of symmetry at 10% of the peak :

With the apolar column:1.20
With the polar one :1.11

And I repeat that nothing change apart the column, and the day of the experiment

@ chromatographer1 : What can I do against this kind of thing ? a lower injection temperature? desactivation of this active sites?


regards


Gabriel

[chromatographer1]

Yes on both counts, but I would first lower the injection temperature and see if any change in the calibration curve occurs.

Methanol or ethanol or IPA might be an additional solvent to add to your samples.

I would not inject a silylation reagent onto a MS.

You might try another brand or type of apolar column, especially if you are using a low bleed version presently.

The surface chemistry might be different and be the cause of the problem.

best wishes,

Rod

[Peter Apps]

Hi Gabriel

Saturating active sites and similar measures is based on the deviations from linearity being due to adsorbtion somewhere in the system, which we have not established. The slight asymmetry of the peaks on the non-polar column suggests some modest activity on that column, but an asymmetrical peak does not necessarily have a smaller area than a symmetrical one. For adsorption to cause smaller peaks it has to remove analyte, not just retard its progress down the column.

You still have not told us whether the calibration on the non-polar column deviates from linearity for the high or the low concentrations. This is essential information. If you cannot post the calibration graphs, at least tell us the equations.

Peter

[GOM]

Hi Gabriel,

Thank you for doing that. Overall you do appear to be getting more tailing on your non-polar column. This interaction isn't large but could be significant for this 20pg injected peak.

Unfortunately we haven't disproved column adsorption or other interaction yet either. To answer your questions

Item 3. From your description so far am I correct in assuming that you are looking at the controlled fragrance component list? Why not just use farnesol as the injection between runs?

Item 4. I think that your run will be too complex to fit a sacrificial one in so I don't think that it is a suitable suggestion for this application.

Hope that helps,

Ralph