Chlorine dioxide (ClO2) has been used for decades as a bleaching agent, as a biocide and in water treatment.  As one would expect from the formula, it is an unstable compound and decomposes in the gas phase at concentrations above 10% to chlorine and oxygen, liberating approx 100 kJ/mole.  The liquid and solid forms are prone to unpredictable and explosive decomposition.

Researchers at Los Alamos National Laboratory in USA were experimenting with ClO2 as a biocide, and generated the gas by the standard methods of passing a mixture of chlorine and nitrogen through a packed bed of sodium chlorite flakes at ambient temperature.  This reaction has a heat of reaction of approx 200 kJ/mole ClO2.

During a series of experiments, which were carefully designed to minimise hazards, the researchers noticed a dramatic rise in temperature and wisely evacuated the room and activated the automatic shut-off valve for the feed.  Shortly afterwards, an explosion destroyed the experiment and did considerable damage to the fume-hood.

It transpired that in the experiment, the ClO2 had condensed as a liquid.  Only a few ml of ClO2 had condensed, but the liquid has roughly one-third the power of TNT, hence the damage concerned.  Fortunately the temperature rise had given sufficient warning for the experimenters to escape before the explosion occurred.  For more details see Chem and Eng News, Dec 9, 2002, pp 4 and 5.

This is perhaps a warning to all who deal with compounds with halogen-oxygen bonds of the potential hazards of these compounds.  I still see many academic reports of preparations of perchlorate salts of organic compounds, followed by drying.  Without knowing the properties of the product, this can be extremely dangerous.  I believe that all papers which describe such salts should have a safety warning and refer to the extensive literature on the subject, for example Kirk-Ohmer’s Encyclopaedia of Chemical Technology and Bretherick’s Handbook of Reactive Chemical Hazards, the latest edition of which was 1999 in the hard copy (Butterworth-Heinemann).