In case anyone is tempted to follow in my footsteps.
I've been experimenting with 2-part polyurethane foam.
Previously I've been using the 1 part aerosol can variety. The results have been marginally OK, but it's horrible to work with. It's messy, sticky (i.e. doesn't flow), hard to get voidless pours etc. Also it expands too much, leaving a foam that's a little too sparse. So therefore needs a constrained (i.e. nearly airtight) mould to get a decent density.
In contrast the 2 part polyurethane pours and sets like a dream. Great density even unconstrained. Great internal structure.
But I've found a major problem: Shrinkage.
After a couple of weeks, my 2-part-polyurethance test halos have shrunk to about 2/3 of the cast thickness (and hence are now 50% denser - too dense). It seems that the blowing agent (the gas that does the foaming, and ends up in the closed cells in the foam) is a flourocarbon with a 23degC (74F) boiling point. i.e. the heat from the reaction boils this, hence causing the expansion. The flaw is that sometime later, once the ambient temp drops below this, it falls below the boiling point, condenses, and therefore shrinks, pulling the foam in with it. Not helpful!
In contrast, the 1-part-in-an-aerosol can stuff doesn't seem to shrink at all. I think because the blowing agent in that is butane, which has a 0degC (32F) boiling temp. It might shrink if left in an ice rink for a long term, but when actually worn, I doubt it would get close to that.
I'm thinking about trying to dissolve butane in the 2-part stuff. Might need to heat it (before mixing) to boil the existing fluorocarbon blowing agent out, and then freeze it, so I can mix in some liquid butane. PITA! Then the question remains as to how well it's foam and set near 0C.