UK Technical Process Safety Forum (formally UK Chemical Reaction Hazards Forum)
 |  Previous |  Home |  Next |  SiteMap |
ICRA Label
Safe Surf
Miltronic DHTML menu
Creative Commons License

The Incident

A routine preparation of a Grignard reagent.  Approximately 20-30% of the bromobenzene derrivative had been added when the reaction suddendly exothermed and the majority of the contents were ejected from the flask.

The Investigation

Several concerns were found wrong with this preparation.  The Grignard was not 100% attended; the magnesium was very old and known to be problematic; only iodine was used to initiate the reaction; the reaction had been left on a steam bath (with the tap turned off).


Lessons Learnt

It is imperitive that a Grignard should not be left unattended.  The potential for stalling is well known.  Initiating with iodine is often problematic (even more so with THF, as was used here) - the solvents have to be vary dry and the magnesium be of good bright quality.  Here, we know that the magnesium was very old and rather dull so a difficult initiation was expected.  Initiation is far better started with either methyl iodide (typical initiation time is 9 seconds!), or if the presence of a small amount of methyl magnesium iodide gives rise to problems downstream, then dibromoethane is also a good initiator - the mixture needs to be low on solvent volume, and unstirred when the dibromoethane is added, one then just waits until the mixture is fizzing nicely (elimination of ethylene) before continuing.  Alternativly initiation can be started by using some Grignarde reagent from the previous batch (only good for campaigns!)  By leaving the reaction on a steam bath, it is possible that a trace of steam was leaking by the valve, causing the mixture to reflux, and give the impression that the reaction was under way.

Since this incident, the laboratory now only uses methyl iodide or dibromoethane to initiate; also if possible dry stirs the magnesium overnight; monitors the reaction throughout, taking a sample for analysis by gc (looking for starter dissapearance) at least every 10% of halide addition.

Back to Top