> They can’t use foam for flotation at that depth. There is a concrete like “synthetic foam” if you want unpowered rise to surface but it’s expensive. I read early on they wanted carbon to be light enough to get back up by releasing weights on powerloss.

This would make sense. The first bathyscapes (FNRS-2, Trieste) used the same iron ballast method via electromagnets, but bouyancy was provided by an external tank filled with a liquid of lower density than water -- gasoline.

I hadn't heard of the foam before, but I looked to see how modern deep submersibles provided bouyancy. The first one I checked was the one I'm most familiar with from my childhood fascination with the Titanic -- DSV Alvin, commissioned in 1964 and taken by Robert Ballard on the very first expedition to the wreck -- and it uses such a foam developed around the same time. A minor correction, though: the foam is called "syntactic foam"; the term shares its etymology with "syntax": from Greek σύν (sún, “together”) and τάξις (táxis, “arrangement”). These foams are composite materials containing what they call "cenospheres" or "microballoons": tiny, rigid, hollow spheres. Why use a large steel ballast filled with flammable hydrocarbons when you can use thousands of much smaller ones made out of an assortment of materials from ceramic to titanium to polymers?