Getting Unknown Mixture Spectra to Yield Their Secrets

I was recently tasked with an interesting challenge by a client and it had me exercising spectral interpretation muscles I have not used in a long time. They sent me a dozen unknown mixture spectra to analyze prior to my teaching an on-site course on the topic at their facility (for the advantages of on-site FTIR training, including free customization, click here http://www.spectros1.com/course_custom.html ). These spectra are perhaps the most difficult to interpret because they are unknowns and because in mixture spectra it can be difficult to figure out what functional groups give rise to what peaks. As I was wading through these spectra I became conscious of the process I was following, and since I had some success I thought I would share that process with my readers.

In my Infrared Spectral Interpretation I course (outline here: http://www.spectros1.com/c-spectral-i.html ) I teach attendees a 12-step program for successfully interpreting spectra. I followed the 12 steps for each spectrum, but in several cases I got to the end of the process without having made much progress. This is when Step 12, "Get Help", comes into play. I found the first thing I did after completing my analysis of a difficult unknown mixture spectrum was to do a library search. In one case the search was of high quality and allowed me to identify the main component in an unknown.

In a few other cases the library search was inconclusive. This is when I hit the literature. I have published a book on Infrared Spectral Interpretation (more info here: http://www.spectros1.com/books.html) and I also have on my bookshelf a number of IR spectral interpretation books by other authors, some of which are more far ranging than my introductory text. Between these books I was able to narrow down some of the unknowns to categories of molecules. For instance, that several of the samples contained carboxylates.

After this I looked up reference spectra of possibilities in a specific chemical class in an infrared spectral atlas. Such an atlas is a collection of infrared spectra organized by functional group. My favorite infrared spectral atlas is the comprehensive 3-volume collection published by Aldrich Chemical (more info here: http://www.sigmaaldrich.com/catalog/ProductDetail.do?N4=Z286001ALDRICH&N5=SEARCH_CONCAT_PNOBRAND_KEY&F=SPEC). This compendium contains over 18,500 spectra organized into 53 functional groups. The beauty of this atlas is that you can look at many spectra of the same type of molecule together and quickly learn the pattern of peaks that is diagnostic for that functional group. I did this with a few of the unknowns to become more familiar with the spectra of functional groups that the library search suggested were present in a sample. If I wanted to look up a specific reference spectrum the Aldrich Spectral Atlas could be used for that. However, Aldrich also sells the Aldrich Spectral Viewer (details here: http://www.sigmaaldrich.com/labware/learning-center/spectral-viewer.html). This is an electronic collection of 11,000 infrared spectra that can be searched by compound name or functional group. I find that if I have to look up the spectrum of a specific compound the Spectral Viewer is faster than the Spectral Atlas. The spectral viewer is nice because the spectra are in color, the display limits can be altered, and peaks can be picked and marked. By looking up spectra of functional groups I made progress, and in a few cases by looking up specific spectra I was able to identify specific compounds in a mixture. In the end, I was able to identify specific molecules in a number of the unknowns. However, in a few cases I was only able to suggest what functional groups might be present in a sample.

So, the key then to analyzing unknown mixture spectra is to execute the first 11 steps of the 12-Step interpretation strategy I have discovered. Then exercise the "Get Help" step by using library searches, spectral atlases, and the spectroscopy literature. Using these techniques unknown mixture spectra can be convinced to yield some of their secrets.