I really enjoyed this paper from Merck’s Process R&D group, but some readers will be saying “Yeah, but that’s just because you really enjoy photochemistry reactions”. The latter part is true, but it’s the sort of paper that we need to help drain some of the voodoo out of all the exciting photochemistry work that’s going on these days. I also like good process-chemistry papers, partly out of a sense of relief that I didn’t have to do all the work involved.
In this case, they’ve got an ingenious experimental setup to investigate what’s going on in the photochemical reaction: they’ve rigged up an LED light source inside their NMR probe, so the reaction can be turned off and on while monitoring the species by whatever form of NMR you like. The reaction studied is shown at left, a catalytic ring formation of an exo-enol ether (2a) from an alkynol. As that scheme shows, there are several steps and several possibilities for things to move away from the desired product, so getting a handle on the factors involved is crucial.
The catalyst is tetraphenylcyclopentadienone iron tricarbonyl, and a CO has to come off the metal center to give you an active catalyst. As the paper details, there are several ways to do that – heat, a co-oxidant, and light being the three main categories. Combinations of the first two tended to give mixtures of all the possible products along with starting alkyne, and if you pushed the conversion the percentage of the desired product got even lower, indicating that it probably wasn’t a stable species under these reaction conditions. So the group turned to photochemical activation.
One of their first moves when trying that route was to carefully saturate the solution with nitrogen case, to rid it of all traces of carbon monoxide (which you’d expect to be an inhibitor of active catalyst generation). But that reaction, while it worked up to a point, was far less efficient than just running the thing in a closed vessel without bothering with the purge. That’s puzzling, and that’s the result that led to the decision to go Full Mechanistic on the reaction to figure out what was going on. The problem is, the enol ethers are chromatographically sensitive (making LC/MS difficult), and analyzing the free CO levels was no easy task, either. So they tried LED-NMR, and it proved to be just the thing.
Several useful insights: the quantum yield for the reaction is just about 3 exactly – that is, three molecules of product were formed for each photon absorbed by the catalyst. When they let the reaction go under LED light for a bit and then turned the light off, the reaction did not actually stop: it just slowed down along a nice exponential curve, and could be restarted by turning the LED back on. This seems to be the active iron catalyst being formed by the illumination and going on to turn over three times (on average) before recombining with its CO ligand and getting deactivated again. The turnover-limiting step is deprotonation of the oxygen and its attack on the alkyne, and it warmed my heart to see this illustrated by a series of aryl-substituted reactions showing a beautiful Hammet plot, from methoxy to trifluoromethyl, just like it was the 1960s or something. (There’s a reason some of these physical organic techniques have become classics – they work).
A particularly nice experiment featured 13C NMR measurements using a catalyst that had been loaded with 13C-enriched CO ligands instead of the regular sort. This not only proved the turnover hypothesis in the light/dark reaction conditions, but also showed why you want a closed system that traps the CO in solution: it turns out that the active catalyst can deactivate itself completely by losing another CO, and that having CO around in solution, although it can eventually recombine with the active species, also keeps that active species from being irreversibly destroyed through loss of the second CO ligand. In fact, the team went on to show that the reaction yield depended on the fill volume of the container: the less headspace, the higher the yield! That’s not a variable that you might have stumbled on without some insight into the mechanism, but it has a very strong effect on the reaction.
This is just the sort of thing that needs to be done to turn reactions into reliable tools – the paper finishes up with an impressive range of products produced under their optimized conditions. I look forward to seeing others pick up on the LED-NMR technique to advance the photochemistry field, which (as this paper shows) has been getting serious respect and effort in the process R&D community. Good stuff!