Excluding O2 can also be critical for reactions using stoichiometric amounts of metals. Modi et al. found that O2 reacted with Grignard reagents faster than H2O.  Researchers at Johnson and Johnson studied the conversation of an Mtr-protected arginine derivative to the corresponding ketones using Grignard reagents.  (The 3-methoxy-2,3,6-trimethylbenzylsulfonyl (Mtr) protecting group is removed by TFA and other strong acids. Other protecting groups have superseded the use of the Mtr group for arginine residues. .) The J&J researchers found that when flasks were opened for sampling O2 was inadvertently introduced and led to the formation of the benzo[d]thiazole trimer shown [above]; in this case, one molecule of O2 consumed 3 molecules of benzothiazolyl magnesium chloride. ...By adding the benzothiazole to a solution of excess t-BuMgCl the generation of the trimer was minimized.Anderson goes on in the safety notes to comment that Grignard reactions in the pilot plant (just like all pilot plant reactions, IMO) should be run under positive nitrogen pressure to exclude O2 and prevent fires.
[I didn't remember the bit about molecular oxygen and Grignards, but Michael Smith's "Organic Synthesis" has a nice summary of the reaction:
"Another electrophilic reagent that reacts with Grignard reaagents is molecular oxygen, which gives a hydroperoxide anion as an initial product. This anion reacts with additional Grignard reagent to give an alkoxide, and hydrolysis liberates the alcohol has the final product."Huh. I'll have to go back and reread that chapter.]
Also interesting: the paper that Neal Anderson references first (Modi et al., in the above passage) talks about how they had to rigorously exclude oxygen from the quench of the Grignard reaction:
Our observations of the effect of traces of oxygen on the conjugate addition of Grignard reagents to 1 led us to investigate the possibility that the low, erratic yields that we had encountered on conjugate addition of Grignard reagents to the steroidal ∆4,6-3-ketone system also might be the consequence of the intermediate enolates reacting with oxygen faster than they protonate during quenching of the reaction....
[explanation of reaction....] The reactions were quenched by injection of water or aqueous hydrochloric acid with rigorous exclusion of air.... The isolation of products of 1,6-addition of the Grignard reagents to the steroidal 3-keto-∆4,6-diene system in yields that were at least three times greater than had been observed previously again illustrates that the carbanions formed by conjugate addition of Grignard reagents to enones or to dienones faster with oxygen than they react with water.I confess that, on lab scale, there's been little desire on my part to keep my quenches rigorously air-free (even though I should, for safety reasons.) I guess I will now. (I wonder how often alcohols are seen as Grignard side products?)