Some colleagues of mine and I were talking the other day about the number of catalytic photochemistry reactions there are out there now, and how there must be many more of them that haven’t quite been worked out. Now I see that Frank Glorius and his group at Münster are proposing a way to start doing that in a more high-throughput fashion.
Angewandte Chemie calls it “catalyst speed dating”, but then, they would, wouldn’t they? The approach is a mechanistic one – instead of looking for products, the group is looking to see whether particular sorts of substrates can interact with a given catalytic system. They use luminescence spectroscopy to probe a key quenching step in the catalytic cycle(excited catalyst forming a new substrate radical), and see what sorts of starting materials make the cut. (The emission spectra of the catalysts will change noticeably if they can interact with the substrate). This not only gives you some mechanistic insight into what’s going on with the photocatalyst, but it also immediately suggests which systems to use for a more traditional product-based catalytic screen. Note that you don’t have to know the various redox potentials, triplet-state energies, etc. – you just set loose an array of potential substrates and let them tell you.
They ran over a hundred substrates past on of the standard iridium photocatalysts, and identified seven that showed a good interaction. Two were known and had been deliberately included as positive controls, and two more turned out to be known in the literature as well, but three structures were not yet described in catalytic photochemical reactions: benzotriazole, 1-trifluoromethylbenzotriazole, and 4-methoxyphenol. The second of these turned out to be interfering with the luminescence assay, but they took the two remaining compounds, and a set of their derivatives, into another round of luminescence screening with several other known photocatalysts.
These led to combinations that were then tried out with test photochemical reactions (acylation, bromination), and these did, in fact, lead to new transformations. It’s worth noting, though, that in each case there were some outliers – reactions that screened well but did not turn out to be the most efficient combinations, or reactions that could produce product via a different mechanism. The excited-catalyst test that was used in the initial screen is (as the authors note) only one part of the catalytic cycle, albeit a crucial one (and this makes one wonder if some other screening protocol could be developed for other parts of the cycle as well).
Similarly, the paper makes the point that this sort of “mechanistic screening” could well be useful in other reactions systems entirely. The current reaction discovery schemes tend to be mechanism-agnostic, and are just looking for new products. This could be a good complement – given a particular mechanism, what can be done with it? Redox photochemistry would seem to be a particularly good fit, given the diversity of the transformations that have been accomplished with it, but there are others out there as well. And we need all the reactions we can get. . .