Comment by kragen

Yes, that's why I said, "dry magnesium, no". A pile of dry magnesium powder only burns at the surface as air diffuses into it. If you have it mixed with the oxidizer so that the flame can propagate through the whole mixture, it will, and the propagation speed is determined by factors like the reaction speed, gas production, and thermal conductivity. The reaction speed in turn is governed by the particle size, since the reaction only takes place at particle surfaces; it goes to completion faster when particle size gets smaller.

Small hydrogen balloons do not in fact go bang; they just create small conflagrations. What goes bang are small balloons filled with a near-stoichiometric mixture of hydrogen and oxygen, such as you get from the simplest forms of water electrolysis.

The stoichiometric mixture of magnesium with water is 1.36 grams of water per gram of magnesium (which is 1.74g/cc, so this 58-wt%-water mixture is 70% water by volume), the enthalpy of formation of H₂O is -285.83kJ/mol, and the enthalpy of formation of MgO is -601.6kJ/mol. So this reaction:

    Mg + H₂O → MgO + H₂

yields 315.8kJ/mol, which is to say, 315.8kJ per 24.3 grams of magnesium, or per 58 grams of mixture, about 5.4MJ/kg, about an 18% higher energy density than TNT. And the hot hydrogen gas produced will carry the heat produced by the reaction into nearby areas, igniting them and resulting in a flame propagation velocity that's higher than thermal conduction alone.

For a large enough particle size, you won't get an explosion, and you may even lose most of your water as steam; but for a small enough particle size and an oxidizer concentration close enough to stoichiometric, you will. Some nanothermites consisting of magnesium nanoparticles with an oxidizer such as iron oxide even reliably detonate.

So, it's a potential safety hazard, but it seems like one that should be easy enough to guard against.