Medicinal chemists and molecular biologists are used to screening huge numbers of compounds for possible biologic activity. Think phage display — http://en.wikipedia.org/wiki/Phage_display, etc. etc.
Now a chemist is using the technique to discover new reactions (Science vol. 333 pp. 1387 – 1388, 1423 – 1427 ’11 — the 9 September issue). 17 substrates (all of a similar mass — which turns out to be crucial) are put into each well of an array. Then 1 of twelve ligands was put into each well of a column, and one of 8 metal catalysts was put into each well of a row in the array. Then the array is then sealed and heated at 100 C for 18 hours. Obviously all sorts of reactions could happen, but because ligands and substrates were all of similar mass all they had to do was mass spectrometry to look for covalent linkage of ligand to substrate. Two new reactions were discovered this way — copper catalyzed alkyne hydroamination and two nickel catalyzed hydroarylation.
Very clever, but to me this shows how very incomplete our understanding of chemistry really is. The senior author (John Hartwig) is an organometallic maven, having written a 1160 page monster tome on the subject. Yet, his knowledge was not sufficient to predict these reactions. I’m sure all sorts of post hoc propter hoc explanations will be forthcoming for why the particular combinations of catalyst, ligand and substrate worked. It isn’t likely that such explanations will be forthcoming for the reactions that didn’t work.
All is not lost, the explanations of the reactions which did work will be added to the theoretical toolkit and used to understand (and hopefully predict) future new reactions.