A simple distillation of dichloromethane from the solution of product was in progress on a steam can. The total material to be stripped down was around 70 litres of solution - reducing down to around 4 litres of concentrate, when petroleum ether was to be added to precipitate the product. The 20 litre flask had been topped up a couple of times as the distillation progressed, and some three quarters of the total volume had been charged. The distillation was progressing normally and was briefly left to run. There was then a large bang, with lots of white fumes. The fire alarm caused all employees to evacuate.
Once the fumes had cleared and all was safe, the employees returned to find that there had been one large explosion in the fume cupboard.
- The glass sash was destroyed.
- There had been an ejection of debris at such a speed to...
- Break another glass sash opposite
- Break several power sockets on the opposite cupboards
- Completely smash a 10 litre buckner flask that was also in the cupboard
- The fume cupboard walls (made of strong Trespar) were broken
- The adjustable feet of the cupboard frame were pushed right up (downward force of the explosion)
- Naturally the main flask was destroyed
The material involved was Teterahydro-4H-pyran-4-one oxime tosylate (in dichloromethane).
It was noted that 4 previous (and uneventful) campaigns of this material were (possibly) made slightly different...
- Early campaigns used a water batch at 60°C to remove the DCM
- The later campaigns did not show what heat was used.
- All campaigns had been smaller (half the size or less)
As all the material had been destroyed, it was necessary to make the material for testing. This was done with extra care, avoiding any heating of the material. In addition, the need to make this material afforded the chance to split the batch several times and vary the work up slightly - so as to be able to test for process variations.
As it turns out, the variations were not needed. The isolated material was shown to be highly unstable, either as a strong concentrate in dichloromethane or as the dry solid. DSC of the material showed onset temperatures in the region of 87°C. No melting/dissolving enodtherm was observed - any such event was lost in the large exotherm. A further test in a "Reactive Screening Device" with the sample in a titanium bomb showed the presence of an exotherm from 60°C onwards, in the next 20 seconds the temperature had risen to 90°C, and then (in the space of one data sample) the pressure jumped from just above ambient to over 45bar. Clearly this mimicked the event in the laboratory and explained how the glass had achieved such high velocities. Further tests indicated that the decomposition was probably autocatalytic.
The cause of the explosion was the very rapid decomposition of the compound, this was enhanced by the facts that...
- The batch size had increased - meaning longer exposure to heat
- The change from steam bath to water bath meant that there was no chance of losing heat back to the heating medium, once the exotherm set in. Plus the increased utility temperature further accelerated the decomposition, once the majority of solvent had been removed.
- The material shows autocatalytic behaviour, thus the scale up meant that longer (and undesirable) exposure to heat was involved.
- The material had not been tested, before as the temperature involved were low, and the product was not going to be distilled (which would have triggered a DSC test).
It is envisaged that the solvent was almost completely removed, allowing the contents temperature to rise and then the subsequent autocatalytic large gas producing exotherm blew up the flask and everything in it's path.
It was probably very fortunate that no-one was very near at the time of the blast. The RSD times suggest that any one working nearby, would have had little warning of impending doom.
- A minor change in the process (from water bath to steam bath) was a major factor leading to the explosion
- The original process was an accident waiting to happen, the smaller scale and use of water bath allowed the experiment to be completed without an incident.
- The standard "toughened glass" fume cupboard sash is not suitable for retaining any flying fragments, once the sash breaks then the majority of flying debris is free to penetrate some distance. An immediate program of adding an approved safety film to all working fume cupboards is now in place.