Let me begin by stating my experiences with elucidations on organic molecules. I have led over 60 training sessions attended by chemists and Spectroscopists from Academia to industry around the world. Some more successful than others—this came down to expertise of the attendee. I have also conversed with dozens of chemists/Spectroscopists, experts and non-experts, at conferences, onsite visits and through email on this amazing subject.
By the way, I am not going into how I got into elucidation while in graduate school, especially helping all those poor synthetic chemists solve their unknowns. 🙂
How about this for pressure? Imagine standing before an audience and asked to elucidate a structure from NMR data especially since you have never seen the data before. Well I have done that and on at least 10 occasions that I can remember. I am not trying to toot my horn but I want to express that my experience in this matter is applicable.
Taking a step back, the one thing all my training encounters have in common is success is judged by how much practice and time one denotes to this task. Just like anything new, practice through repetition makes things better and easier to handle. It is not that easy and not every chemist/Spectroscopist realizes the complexity behind teaching this process.
I am going to try to list all the possible nuances an NMR elucidator must take note of for any unknown whether they are aware of it or not. I want to talk about MS but I also want to keep the blog short. Maybe in the future another blog will follow on this subject.
1. The toughest elucidations and the ones I do not like doing are samples that have mixtures, salts, polymers or rotamers. Trying to pick out what belongs to what compound is a big headache especially when the compounds exist at an approximately equal ratio to each other. A big ouch.
2. Size of the structure can complicate matters. I have heard peptides and protein elucidation are tough. I managed once to do an elucidation of a peptide (~1200 Da), however, I knew beforehand what the end product so I would not classify this as a true unknown elucidation.
3. Knowing when to use what combination of NMR experiments is crucial. There are hundreds of choices for NMR experiments and some are more useful than others are. It is so easy to waste your time and instrument time. Experience and versatility count here.
4. Learning how to interpret the NMR data. I find the bulk of any elucidation falls under this classification. There are so many points to cover here so I will try to list as many as I can think of.
a. The tough stuff comes from differentiating between true signals and artifacts. Experience really counts here.
b. Dealing with overlapping 2D NMR correlations in a crowded region.
c. Deciding whether a correlation in an HMBC experiment is showing a 2J or 3J separation.
d. Missing signals due to symmetry, broad peaks, low S/N or paramagnetic metals present.
e. Few 2D correlations from a structure with a high RDBE (Ring and Double Bond Equivalence or element of unsaturation) or containing few hydrogens. Lots of possible candidate structures here.
5. Narrowing down to a single Molecular Formula. I find combining NMR with MS data very applicable here.
6. Deciding the carbon multiplicity (C, CH, CH2, CH3) should be easy to do. Knowing the MF could be helpful here and vice versa. Having this information lets you know how many exchangeable atoms are present.
7. Searching across a library/database of chemical shifts and coupling constants like C-F. Having information on the starting material or derivatives can speed things up. Absolutely an easy thing to do has the potential to save lots of time.
8. Deciding carbon bond order (hybridization) – should I consider sp carbons? I rarely run into sp carbons but it is always a thought in the back of my mind.
9. I like to ask myself these questions when dealing with heteroatoms. Do I have bonds between heteroatoms or bonds between heteroatoms of the same type? Do I have a pentavalent nitrogen? I find IR data to be helpful here. If I have a salt, which atom(s) is charged?
10. Exclude (or bias) possibilities based on NMR shifts. If I have a carbon at 200 ppm, then I immediately jump to a C=O group whereas a carbon at 120 ppm I exclude C=O as a possibility. Biasing the data can be a bad thing sometimes.
11. Apply an internal filter. I do this without realizing I’m doing this. For example, if I see a list of 6 aromatic carbons, I jump to a benzene ring. On an initial pass, I commonly exclude 3 and 4 membered rings and even the higher up rings such as 8-10 membered rings as I rarely run into these. I also try not to draw fragments and structures that do not have crossing (concatenation) bonds. Knowing whether the unknown is natural or not can help sometimes.
12. I like to draw out possible fragments and candidate structures especially since it helps to get a visual on things. The next step is to rank the list and see what can be eliminated. Logics and reasoning do the trick here.
13. With a candidate structure or structures, I like to go back and see if the couplings make sense and thus verify the elucidation. Nothing worse to the ego than passing on a wrong structure.
14. If there is any stereochemistry in the structure, additional experiments like NOESY data are needed. Remember for every chiral carbon, there 2^n possibilities where n is the number of carbons.
Just a final note. In my experience, there are two classifications for unknowns for small molecules: synthetic/impurity and natural unknown.
A synthetic/impurity unknown generally has additional information, that is, information on the starting material and probable product(s). It can be a bad thing to know this and I have spoken with some elucidators who choose not to know anything about the starting material and expected product(s) as this could bias the elucidator and send them on the wrong track. That being said, due to time constraints, it is so tempting to use this information and speed up an elucidation.
Natural unknowns are tough especially if there is very little material. The sample size can restrict the types and acquisition time of NMR experiments. This is very similar to impurity identification. Dereplication (searching a database) can help assuming it is not a novel compound.