GC/MS - response factors

[ekster]

Hai,
I am performing Pyrolysis-GC/MS on natural organic matter. As these complex samples contain hundreds of different compounds, it is impossible to measure response factors for them all. Looking for information on RFs, I encounter loads of useful information and even RF prediction models for the GC technique with FID detectors, but virtually nothing on GC/MS (R)RFs. To what extent can the information on RFs of substances using a FID be applied to those using GC/MS?
For now, I'm just looking for the answer on a hopefully very simple question: will the peak area of two different compounds at similar concentrations be on a per molecule or on a per mass/carbon atom basis? In other words, after determining peak areas, would I have to use molecular weight while quantifying?
I would also greatly appreciate any reference to publications on this issue.

Thank you, Joeri Kaal
(I recognise that a large amount of properties of molecules influence their detection sensitivity, but please keep things simple, I'm a geograoher)

[Peter Apps]

Hi Joeri

The simplest answer to the question is "neither" - different molecules fragment differently and therefore give different signal intensities on both a signal per mass and a signal per carbon atom basis. When an MS is used as a detector it is standard practise to calibrate using the analyte itself, or a very close analogue (say an isotope labelled standard).

If your main aim is to quantitate a large number of compounds rather than identify them (which will be a challenge in itself) you will do better with an FID than with a mass spec. It's cheaper and a lot less trouble to run as well.

Peter

[ekster]

Thank you for the reply Mr. Apps,

I understand there isn't a correct answer to my question, response factors is a very delicate issue. However, I would like to know what basic experience is: would a molecule of C30-alkane yield a similar contribution to peak area as a molecule of benzene for example, or would the much higher mass of the alkane result into a larger peak area?
Using a FID detector, the alkane would more or less yield a peak 5 times higher than benzene, as the RF is mainly C related. Using a MS as the detector, "different molecules fragment differently and therefore give different signal intensities on both a signal per mass (1) and a signal per carbon atom basis (2)". Would it be possible to say if the first or the second is the more likely, if we stick to the C30-alkane/benzene example?
I hope I'm not getting on your nerves already...

Regards, Joeri

[tom jupille]

Would it be possible to say if the first or the second is the more likely, if we stick to the C30-alkane/benzene example?

Neither.

To a first approximation, the signal depends on the number of ions getting to the detector per unit time. That will be affected by the ionization efficiency of the analytes, which in turn will depend on the details of the structure and the ionization conditions (voltage, temperature, etc.) as well as on the MS tuning.

[Peter Apps]

Hi Joeri

The comparison between aliphatic hydrocarbons and aromatics is a useful one.

In the ion source of the mass spec, a tiny fraction of the molecules is ionised, each ionised molecule picks up one unit of charge, then it fragments and the charge ends up on one of the fragments, which is the only fragment from that molecule that is detected. The uncharged molecules and the other fragments from the charged molecule disappear down the vacuum pump.

Because a single molecule can give only one unit of signal no matter what its molecular structure or weight, the signal per ionised molecule is independent of the nature of the molecule. To your question "would a molecule of C30-alkane yield a similar contribution to peak area as a molecule of benzene" the direct answer is Yes, but you are not analysing single molecules, you are analysing huge numbers of them.

The strength of the signal for a sample of a substance (in other words a very very large number of molecules) depends on how many of the molecules get ionised in the source. This depends on the ionisation cross section for the substance - which depends on molecular structure and molecular weight. It is higher for aromatics than for alkanes of a given molecular weight but without dredging up the data I am not sure whether it would be higher for C30 than for benzene. In a closer comparison a C30 alkane would give a weaker signal than a C30 aromatic (a PAH) because ithe alkane has a smaller ionisation cross section.

In this respect the difference betwen an FID and an MS is that in an FID the molecules are first fragmented and then the fragments are ionised, so a large molecule that gives many fragments also gives many ions, while in a mass spec ion source the molecules are first ionised and then fragmented, so each molecule only gives one ion.

Far from getting on my nerves, your question has made me think more clearly about something that I had taken for granted, so thank you.

Peter

[ekster]

Before I start reading some general organic chemistry textbooks, I wish to reply to the intelligible information given by Peter as it invokes some doubts.

I've been looking for datasets containing relative EI ionization cross sections iin the literature and it seems that experimental values are limited to LMW components. Then I encountered a "model" of Fitch and Sauter (Calculation of relative electron-impact total ionization cross-sections for organic molecules, Analytical Chemistry 55: 832-835) from 1983 which allows for the calculation of molecular cross sections based on the cross sections of individual atoms. Applying these calculations (the variety taking hybridization states into account, Eq. 3), the results match quite well with the calculations that can be made on a site of the NIST for benzene, propane, ethane etc. However, applying these equations to naphthalene and C10-alkane (equal C numbers), the C10-alkane's ionization cross section (30 squared Angstroms) is 50% higher than that of naphthalene (20 squared Angstroms). I believe this contradicts your statement Mr. Apps, and therefore I would like to know where my mistake is. Or if I'm just overinterpreting general concepts to complex chemistry issues?

Thank you both, Joeri

(this forum is a great initiative!)

[Peter Apps]

Hi Joeri

For some reason I cannot download the full text of the paper that you cite, but I did manage to look at the first page.

Does Fitch and Sauter's regression line include a range of chemical types ? - aromatics as well as aliphatics ?.

The reason to expect that aromatics will ionise more easily than aliphatics is that aromatics have clouds of delocalized pi electrons that are relatively easily detached from the molecule by a 70 eV electron flying past.

Why is it that you are interested in response factors ? You need to worry about response factors only if you are doing quantitative analyses.

This thread might catch the eye of an MS expert if you post it under hyphenated techniques.

Peter

[ekster]

Hello Mr. Apps and hopefully any MS expert willing to take the time to read this thread (concerned with the estimation of GC/MS-response factors of molecules with widely differing masses and structures),

Fitch and Sauter's model is build on data encountered in the literature: a total of 179 compounds, among which alkanes (up to pentane) and benzenes. Hybridization can be accounted of, but the carbon atoms in benzene (sp2 hybridized) have a similar "coefficient" (read: contribution to ionisation cross section, as the model is additive) as the C atoms at a double bond in an alkene, and hence the conjugation of the electrons in the pi-bond is (which would be relatively easily displaced as you say) is not taken into account. In fact, a sp3-hybridized C (alkanes) contributes more to the cross-section of a molecule than a sp2-hybridized one (and I lack the theoretical background to understand why).
The reason why I'm interested in ion. cr. sections is that I'm currently comparing solid-state 13C NMR with Py-GC/MS of a peat core from Spain. I apply a model to allocate the different "C types" (alkyl C, O-alkyl C, aromatic C, etc.) to structural components (carbohydrates, proteins, lignin, etc.). The distribution of C among these structural components is regarded "quantititave" on a mass basis. I am comparing this distribution with the one obtained by Py-GC/MS, and despite all the cavities of this approach I do get very nice results. But this depends on the way a quantify the Py-GC/MS trace, which depends on a.o. ionization cross-section.
Actually, I appear to overestimate aromatic substances in the Py-GC/MS trace, which could have to do with the pi-bonding of them (giving aromatics a lower ionisation cross-section than the model of Fitch and Sauter assumes).

Really appreciate your efforts,
Joeri

[ekster]

(giving aromatics a lower ionisation cross-section than the model of Fitch and Sauter assumes).

Sorry, on the contrary: the ionization cross section of C atoms in benzene may be higher than the value given by the model of Fitch and Sauter.


And, for clarity, with:

I apply a model to allocate the different "C types" (alkyl C, O-alkyl C, aromatic C, etc.) to structural components (carbohydrates, proteins, lignin, etc.)

I refer to the deconvolution of the NMR spectra.

Regards, Joeri

[Peter Apps]

Hi Joeri

How do you distinguish between aromatic carbons, O-Cs, and aliphatic carbons etc ? I presume that you are looking at the mass spectra. If so you will be able to identify the compounds with a reasonable degree of certainty. You could then measure their response factors by injecting a known quantity of representative members of each class. This might work better than trying to model the response factors.

Is the relationship that you are finding robust to the pyrolysis conditions ? - I would expect that changing the heating rate and maximum temperature would change the chromatographic profile quite dramatically for a complex non-volatile material like peat.

Peter

[ekster]

Peter,

Well, in NMR they are of course distinguished by chemical shifts, and by Py-GC/MS they are indeed distinguished on their retention times and mass spectra.
Your advice to produce a mixture of known "representative" compounds of their corresponding compound classes is gratefully taken. However, at the moment I do not have a GC/MS at my disposal, which will change in a few months.

Pyrolysis is performed at the Curie-Point of the metal wire on which the sample is pressed. In this case were talking about flash pyrolysis.

I knew I shouldn't have mentioned what my project is about!

Kind regards, Joeri