Screening for Unknowns

Martin Perkins

1st May 2013

For years, targeted analysis has been very much in vogue. Someone gives you a list of compounds to look for in a given matrix and you report how much of each that you can measure. Of course, there may be other compounds present that you (or someone else) ought to be interested in, but in targeted analysis, you just don’t see them (out of sight, out of mind).

Things are changing though; and, in the future, more analysts will have an obligation to look for and report everything of significance that we can see in our samples – not just the compounds on the target list.

Target analysis is a piece of cake in comparison to screening for unknowns, so things are about to get very interesting.

Accurate mass, mass spectrometry (GC and LC) is the “Gold Standard” when it comes to screening for unknown compounds, but as yet, relatively few labs have the instrumentation needed; and equally important, staff have yet to acquire the requisite skills in mass spectral interpretation.

As a consequence, for now, most labs will have to make the most of the instruments and skills they have to hand.

For GC amenable samples, this means that the best available option for screening “known unknowns” is to run GC/MS EI/CI in full scan, acquire spectra, then perform a NIST library search for the EI data. To get clean spectra, you would be well advised to use mass spectral deconvolution to overcome inevitable co-elution problems.

To the best of my knowledge, CI libraries don’t yet exist, so with these spectra, you still have to manually compare the spectra of unknowns to those available in books and literature for tentative identification.

Once you have a tentative identification, you then need to find a source of the pure compound (if available) and compare your unknown with full-scan spectra and the retention indices data for your standard. This takes time and is a lot of work.

Of course, this assumes that the set of unknown compounds you are working to identify, have all previously been characterised by someone else. When you come across an “Unknown, unknown” (something not previously reported in the literature or libaries) then you really are stuck – unless you have access to higher resolution mass spectrometers; tools that enables you to work out molecular formulae from first principles.

In the past, I have used this simplified approach for; screening “known unknowns” in environmental samples, for TIE and EDA based studies; identifying biologically active causative compounds following on from bioanalytical screening to target “hot” fractions.

I also worked closely with many researchers in Europe, on several DEFRA and EU Framework 6/7 funded projects, to standardise analytical techniques for doing this; making use of GC-MSD and a GC-Ion Trap, to generate both EI and CI, full scan, spectra.

This approach worked OK up to a point, but you are obviously limited by the content of libraries for spectral matching and by the commercial availability of standard compounds needed to confirm tentative identifications. I have copies of plenty of papers on this subject (to which I contributed to the analytical work).

In later studies, I also used a GCxGC TOF at the Free University of Amsterdam, Institute for Environmental Studies, working with Jacob de Boer and Pim Leonards.

Much of this work was published. The use of GCxGC-TOF-MS gave much more separation power and accurate mass capability.

The identification of unknowns by non-accurate mass LC-MS instruments, (usually ion traps) can be done by structural elucidation, however, this takes a long time and you need to have a very good understanding of collision-induced dissociation (CID) – which is tricky.

Libraries in LC-MS, (excluding TOF and Orbitrap), are poorly advanced, sparsely populated and quite unreliable due to the variety of conditions that can be applied to the both samples and standards (i.e. ion source conditions and chromatographic parameters).

For non-accurate mass LC-MS libraries to be useful, you must standardise both your ion source conditions and general analytical conditions, to ensure that you analyse both standards (to generate your libraries) and samples, under the same conditions.

As you may guess from what I have written here, I believe that accurate mass LC/TOF MS and/or GC TOF-MS are pretty much essential and need to be used in addition to existing nominal mass instruments, if you have to analyse for unknowns on a routine basis.

I know of plenty of good publications on this topic, mainly focused on the analysis of environmental samples (water, sediments, biota and the like). The determination is always pretty much the same but the approach to the initial sample prep will be different to that used in targeted analysis as you have to also have non-selective, “catch all”, sample prep methods.

If you are interested in discussing this further, please do not hesitate to contact me on +44 (0) 1223 279210, or email