Chromatography Forum

Hi All,

We have MS library from NIST. For qualitative identification, how much the minimum percentage should be matching with the library, so we can sure that is those compound?

thank you
ito

It is not possible to identify a compound only from a library search match, no matter how good the fit is. You would need supplementary information about what is likely to be in the sample, does the retention match a literature value ?, if you inject the pure substance does it elute at the same time ?, if you spike a sample with the pure substance does it co-elute, if there are hetero-atoms can you see them with a selective detector such as a flame photometric or electron capture detector ?, do you have any spectroscopic data.

Peter

I wouldn't report anything below 75% unless I looked at both spectra and was pretty sure they were the same. Sometimes with smaller compounds a lot of the library spectra is below 40 m/z where i don't scan so that gives me a false low %.

Sometimes I can look at the spectra and tell that I have two coeluting compounds because one match and the other are complimentary of the average spectra.

I've also seen related compounds both give a 95% match or 1 give a 95 or 1 give a 94.

In my industry, flavors, A lot of time if I have a few close matches and see which one is a known flavor ingredient and has a FEMA #.

I can also pull out the old n-alkanes and do a good old Kovats index charting.

Back when I was working as an applications chemist, I had a samle come in to the lab with the note that the potential customer would be intersted in the instrument if we showed match scores better than 90%. The sample was a mint oil. I knew the outcome of this experiment even before I ran the sample. There were many peaks - most or all of them terpene related - and many had multiple library hits, each hit with a score above 90%. Probably above 95%. And, given the similarity of terpenes one to another, this is no surprise.

There are a few important things to remember on library searches.

1) The match score is a mathematical calculation in which each spectrum is handled as a vector and the match score relates to the angle between the spectra. (The vectors each have a dimension for each mass in the spectrum and the magnitude for the vector in that dimension is the intensity for that ion.) Thus, in a spectrum wiht many masses showing, it is possible to actualy have two spectra with an ion as the base peak in one spectrum and to present in the other to show a relatively high match score.

2) The spectrum of a compund depends on the formation of an ion with sufficient stability to pass from the ion sorce to the detector. In some cases two different compunds will be ionized and wiht rapid rearangement or fragmentation, the ions seen in the spectrum are identical or are so nearly identical that it is difficult to discriminate the compunds based on examination of the spectrum. There are some gamma lactones that I have liked to use as teaching examples.

3) There are far fewer spectra in the libraries than there are compunds in the world. Many of the compunds that do not have spectra recorded in libraries will have spectra that have similarity to spectra already in the libraries.

I have seen many cases of sepctra with high match scores and the compounds were no the same and many cases where the match scores were poor, but based on other information, I knew that the compund actually was that in the library spectrum showing the poor match score. You always need additional information. Sample history may be sufficient or you may need to isolate compunds and obtain additional chemical information.

ito_miura
For proper identification you must be sure that the compound eluted and that the sample compound has abundant molecular ion or otherwise the library identification may not be trustworthy.
The best way to be sure about NIST library identification is to further identify the elemental formula of your sample via its isotope abundance. We have the Tal-Aviv Molecule Identifier software (TAMI) that inverts the experimental isotope abundances and automatically correlates them with NIST results to provide much higher confidence level in the identification or to suggest that the library is incorrect and in that case provide a list of elemental formula with declining order of matching factors. You can find further information on the TAMI software in its Aviv Analytical website
http://www.avivanalytical.com/Isotope-Abundance.aspx

If your GC-MS is of Agilent you are invited to send me your Chemstation or MassHunter file and I will analyze it for you with the TAMI software
Amirav
amirav@tau.ac.il

Yep terpenes give me a hard time too they were what I had in mind when I said related compounds give me multiple hits with high % matches as well as some D/L optical isomers, some cis/trans isomers, and some structural isomers. I kept getting hits of ortho cymene when I knew it was definately para cymene.

... and of course isomers have exactly the same accurate mass, and exactly the same isotope pattern.

Peter, it is right to compare with other refference/literature. In our industry, our sample mostly like scale deposit (extractives). Until now, we only guest the component of the sample with look the matching percentage, compare with other sample in the current place that may current that scale/deposit.

MSCHemist, why the library spectra below 40 m/z can gives a false low %? Is it % matching with library?

Don_Hilton, thank you for the knowledge you share...

ito

Peter, it is right to compare with other refference/literature. In our industry, our sample mostly like scale deposit (extractives). Until now, we only guest the component of the sample with look the matching percentage, compare with other sample in the current place that may current that scale/deposit.


ito

How important is it for you to identify the components ? If the identity of the compound is significant you should at least get a standard of the compound that is top of the library search results, and run that to confirm retention time.

Peter

Boy I had a tough one today. I had a big glomp of terpenes and sequiterpenes all cooeluting on a wax column (If Id known it was all hydrocarbon I'd of used a db-5). Dozens of high matches of terpenes I'd never even heard of before.