"Quantitating" without a Standard in GCMS

[Karen01]

I have situation where I need to ID peaks (formula good enough) and and hopefully reasonably quantitate (decent ballpark number) a class of hydrocarbons by GC MS (EI or Charge transfer APGC) that differ by # of carbons (and branching... though IDing the branching does not matter)

The class of compounds I am looking at does show a small parent ion in EI and a bigger one in charge transfer APGC so, as I know the formulas i am looking for, peak ID for formula is RELATIVELY straight forward.

I don't have standards for the compounds/peaks I have to ID which makes the quantitation not so simple.

I do however have a known well characterized compound of the same class from which I can make solutions of known concentration.

When using FID, I know that for these compounds, FID response will be reasonably proportional to the number of carbons... So the obvious approach would be to ID by GCMS and "quantitate" by FID using the response of the known and the difference in the # of carbons between the two... but that assumes complete separation of the peaks of interest from everything else on FID ... and I'm seeing indications that may not be the case.

In MS, even with then same carbon number, i worry that degree and position of branching could affect parent ion intensity significantly ... but maybe i'm wrong

So is there some MS equivalent to teh FID assumption about response factors based on # of carbons that can be used? Or is there some other way to get make a correlation to get a ballpark quantitation?

Thanks,
- karen

[Don_Hilton]

It is common to use the total ion chromatogram from a GC-MS (at least in EI mode) like the FID trace for estimating composition of a mixture. And, there are discussions of the validity of such measurements on this forum.

The assumption that the total ion chromatogram would give responses proportional to concentrations of analytes would imply that we are assuming the same ionization efficiency, perhaps increasing with the size of the molecule, for all compounds (at least within a class) and that we count all the ions. If you want to look at the molecular ion only, you would have to assume that it is formed in the same proportion for all compounds. Examination of the spectra of the normal hydrocarbons, for example, shows a fairly healthy molecular ion for the smaller hydrocarbons and a very small intensity molecular ion of the larger hydrocarbons. And, you see more and a greater number of fragment ions being formed in the larger hydrocarbons. This suggests that the response factors for molecular ions will vary widely – and will not be useful without some form of calibration. Even with APGC, which reduces fragmentation, you will need to capture the areas from fragments.

This leaves the question of ionization efficiencies across the compound class? Soft ionization techniques give widely varying responses for compounds, so this depends on the class of compounds you are looking at and the diversity of compounds in that class. So with APGC, proceed with caution.

[Karen01]

It is common to use the total ion chromatogram from a GC-MS (at least in EI mode) like the FID trace for estimating composition of a mixture. And, there are discussions of the validity of such measurements on this forum.

I did a search on "total ion chromatogram" in this forum and could not find the discussions... Could you provide links to either inside the forum or out ?

Update: I did find one:
viewtopic.php?t=9944&p=47482

After reading that I came away that doing it by FID carbon counting my error would likely be less than 10%. I'm still not sure how much potential error there is likely be trying to use an EI TIC like an FID in terms of # of carbons.

The assumption that the total ion chromatogram would give responses proportional to concentrations of analytes would imply that we are assuming the same ionization efficiency, perhaps increasing with the size of the molecule, for all compounds (at least within a class) and that we count all the ions.

The carbon number number can vary by 7 from my 'standard' ... We are talking about chains in the 20s.

Given how easily hydrocarbons fall apart, for TICs there would obviously be more ions for larger molecules, but I would expect branching could impact that number by affecting fragment sizes because of ion stabilization, though i don't know how significantly.

Has there been any work done to correlate # of carbons (thus MolWt) to TIC response for hydrocarbons? For unbranched I would think there could be a decent relationship. Perhaps that was discussed in the the threads you are referring to? It would be nice if the TIC peak area was proportional to Mol Wt (or carbon number) in the class.

If the ion production differences are not big in the range I'm looking at, or is directly proportional to carbon #, then perhaps I could just quantitate against the single standard I have on a mass basis. If not I would need a relationship between Carbon # and TIC area.

Of course to use TICs as with FID, I also have to separate the compounds from everything else... in that case, if I can do that, as I said the most straightforward way is to ID the peaks by MS using the parent ion and quantitating by FID.

If you want to look at the molecular ion only, you would have to assume that it is formed in the same proportion for all compounds. Examination of the spectra of the normal hydrocarbons, for example, shows a fairly healthy molecular ion for the smaller hydrocarbons and a very small intensity molecular ion of the larger hydrocarbons. And, you see more and a greater number of fragment ions being formed in the larger hydrocarbons. This suggests that the response factors for molecular ions will vary widely – and will not be useful without some form of calibration.

Does that not imply the same thing for TIC peak areas unless they are directly proportional to carbon #?

Thanks,
- Karen

[Don_Hilton]

Whether you go with FID or MS for quantitation depends on how good you need results to be. For some, variation from C20 to C27 would "only" be a variation of about 50%. For others it would be a giant variation of nearly 50%!!!

A point for clarification: In the MS we may get many fragments for a type of molecue in the spectrometer, but for an individual molecule, we get only one ion formed and, thus only one ion reaching the detector for an individual molecule present. (And out of all molecues present in the spectrometer, few are actually ionized.) Any relationship between TIC area and carbon number, if it really exists, would have to be a difference in ionization efficiency related to the carbon number. And, the mass range acquired has an impact on the TIC. If the spectra area acquired from m/z 45 and up strong signals for many smaller ions are not measured. And there are many GC/MS runs made with acquistion starting at m/z 45 to avoid background signal from CO2. So if you are lucky, large molecues which would have more fragments, have more fragments above the lower acquistion limit - and there is, then, a bias in the total ion chromatogram to have more response for larger molecues.

I suspect that most people who use TIC area percent to estimate percent composition are closer to the opinion that the above variation is "only" 50%. And, if the mean molcuar weight is in the middle of that range, the error in estimating percent composition may be less than 25% for any particular compund.

[Karen01]

A point for clarification: In the MS we may get many fragments for a type of molecue in the spectrometer, but for an individual molecule, we get only one ion formed and, thus only one ion reaching the detector for an individual molecule present.

Thanks... That makes sense... i was thinking of it the wrong way...

So using the parent would likely much more uncertain... which means using TIC or FID for quantitation are the only real options, and implies I have to separate the targets from everything else.

Thanks,
- Karen

[Amirav]

Karen
TIC peak area in EI is proportional to the hydrocarbon amount (weight) the same as in FID. However, the molecular ion in EI is reduced (significantly) with the hydrocarbon molecular weight and with the degree of its branching. Branched hydrocarbons that can form tertiary ion provides particularly small molecular ion, much less than the linear chain isomer of it. In addition, the effect of response uniformity is erroded by ion source peak tailing above C20 and this effect also depends on the ion source temperature. The hotter the ion source temperature the less is the tailing but similarly the smaller is the molecular ion. GC-MS with Supersonic Molecular Beams provides much more uniform response than standard GC-MS but even with it the molecular is lower by x2-3 for the branched hydrocarbons. If you want more information on this topic please write your Emaill address
Amirav

[Don_Hilton]

Amirav: I have used the TIC as having similar response to the FID, but have never seen work showing that this is valid (other that it has generally worked across the range of compunds of interest to me at the time). Do you have any references to published work that supports this practice? If so, I'd love to do some reading. Thanks.

[lmh]

I'd also be grateful for more evidence. We've come unstuck with this assumption in the past, and found that area percents of fatty acid methyl ester standard mixes by TIC from EI do not match the numbers on the certificate of analysis. They're not terribly out (i.e. within that 50%), but they were far enough to render the data unfit for our particular purpose.

The other problem is that if you can't use FID because the peaks are inadequately separated (see original post), they may not be adequately separated in TICs either.

[Amirav]

Don-Hilton and Imh
The EI ionization cross section is uniform and compound independent (depends about on the number of electrons in the compounds hence on the sample weight). Thus, for small and volatile compounds the TIC behaves the same as FID response for organic compounds such as small hydrocarbons. As the sample size is incrased two additional factors erode the TIC response uniformity: a) quadrupole ion transamisson and ion detector response dependence on mass (some reduction with mass); b) Ion source peak tailing. Thus, "TIC as having similar response to the FID" is valid only for small and volatile molecules. For a FAME like methyl stearate we found with a 5975 that for the same comcentration the TIC response per ng is lower by a factor of about 3 than for hexadecane and this number depends on the ion source temprature (and even on the FAME concentration).
We have a unique GC-MS with supersonic molecular beams (5975-SMB) that has uniform response to all compounds the same as FID since it uses a fly-through EI ion source. In TIC when one encounters coelution one can perform RSIM on the molecular ion and multply its area by the % of its abundance in the TIC as measured with a standard thereby obtain the amount.
We recently submitted a paper titled "Measurement and Optimization of Organic Chemical Reaction Yields by GC-MS with Supersonic Molecular Beams" This paper discusses the topic of TIC reponse uniformity with examples and I will be happy to Email it to whoever will ask me by my Email amirav@tau.ac.il
Amirav

[mckrause]

Karen is talking about hydrocarbons, as per her original post. FAME is definitely NOT a normal hydrocarbon. Any time you have heteroatoms in your molecule the TIC will not track an FID since the fragmentation is not conventional. For straight chain or branched aliphatic hydrocarbons the TIC will approximate an FID. Yes, Karen, this means that you have to separate the compounds if you want good correlation - consider it a fancy FID with similar limitations.

Please bear in mind that most GC injectors have mass discrimination - it is exaggerated using TIC. What we do to keep things "real" is run a hydrocarbon locator standard - for example, C8-C40 by 2 carbon units (C8, C10, etc). This allows you to approximate response factors within boiling point ranges, and helps correct for injector discrimination. While this is not strictly "standardless", it is a minimal standard (one injection) that allows you to derive a bit better semiquantitative result.

[lmh]

so can I get this clear: A TIC from conventional EIC isn't universally compound independent because hetero-groups make a huge difference and a FAME can differ by a factor of 3 from a similar-sized alkane (Amirav). The situation can be improved by sophisticated ion sources (Amirav). Within a single class of compounds, e.g. alkanes, a TIC may approximate fairly well to the result seen in FID (Amirav and mckrause) but even there, there is some discrimination which can be countered by running a single external standard mix, once, and using a single-point calibration curve (mckrause)?

Have I got this right?

[Amirav]

Imh
Ion source peak tailing affects all compound groups. While polar compounds are affected more than non polars even for linear chain alkanes the larger (>C30) compounds will be reduced by ion source peak tailing in addition to other factors such as syringe discrimination (which equally affects FID).
Note also that ion source peak tailing is like an iceberg thus a small chromatographic tailing can hide a big signal loss due to very long time tail.
For this reason the vendors increased the upper ion source temperature and advertized it for PAHs (Agilent app note). However, while this works well for PAHs that has dominant molecular ions in alkanes hydrocarbons increased ion source temperature practically kills the molecular ions.
In summary, one can use a chromatogram with standards in the needed volatility range and it can serve as a first order correction.
Amirav