A recent blog post from Derek Lowe talked about the fun of quenching a reaction, thinking you are finished, only to find the chemical has its own agenda and that apparently involves climbing out of the flask and redecorating your hood. It is quite startling how using the reagent in the normal part of the reaction (when you are encouraging it to react) ends up going off without a hitch, but when you quench it to destroy the excess, well, that is when the fun really begins.
The poster child for this type of behavior is phosphorus oxychloride (POCl3). I have had the opportunity to see it in action several times and it is always surprising how suddenly it takes off.
The problem with this particular reagent is that, though it reacts violently with water, at low temperatures it is sluggish, at least partly due to not being soluble. The unsuspecting chemist sees no vigorous reaction, so adds more. What happens is you get a build-up of reagent, skulking in the bottom of the flask as far from the horrible water as it can get, but eventually, enough mixing occurs to initiate reaction, temperatures rise and suddenly you have all the reaction happening at once and giving the best impression of that fountain display in Las Vegas that it can.
It seems like the safest thing to do: default behavior when you are playing it safe is to cool the vessel and add it in slowly and for most quenches this works admirably. Certainly, you don’t want the heat in the system getting out of control. The issue is to make sure that each drop you add is reacted before you add the next one and, though it always feels strange, that is why you should add your phosphorus oxychloride quench into warm (i.e. room temperature) water with plenty of stirring and careful temperature monitoring.
It does not help, of course, that a lot of procedures call for use of POCl3 as solvent and thus in large excess, which in turn leads to there being a lot of it left at the end to dispose of. It would be preferable from a safety standpoint, not to mention environmentally, to use less in the first place. Using a cosolvent such as toluene reduces it some, though half of a large excess is still pretty much in the large excess category. Similar to the way that dimethylformamide (DMF) activates thionyl or oxalyl chloride in acid chloride formation, DMF can activate POCl3 as well. Another procedure that I have had a lot of success with is the procedure of Morris Robins [see Can. J. Chem. 1981, 59, 2601], which accelerates the reaction by addition of chloride ion, in the form of something like benzyltriethylammonium chloride. This reaction can be done in acetonitrile and takes you from refluxing in neat phosphorus oxychloride for several hours to 50 to 80 C with the reaction complete in a few hours at most (I had one that was done in 30 minutes and would start to decompose if it ran too long). A lot of the problems with this type of chlorination or dehydration procedure is that the initial reaction is fast but loss of the phosphorus intermediate (somewhere between phosphate and PO2Cl2) is slow. This is also why sometimes you can start this reaction and it seems to have done something, but when you stop it (prematurely), the intermediate just hydrolyzes back to the starting material and you have wasted your time. The chloride increases the rate of the slow step. And you can get away with uses several equivalents of POCl3 instead of the less than scientific ‘lots’.
What is interesting about the frequency of incidents with this is a thread with common accidents in chemistry. They happen when you think the danger is passed. With very dangerous and toxic chemicals like hydrogen cyanide, everyone is very careful and accidents are rare as a result. It is when you are underestimating what is in your flask that you can get into trouble.