Re: Thermodynamics fail
A cleverer idea would be 92% filling the inside of a hollow metal (e.g. silver) sphere with water and leaving the rest as vacuum to deal with the expansion. Then you'd have high conductivity, phase change and no dilution. Maybe use an internal matrix structure to help conductivity and add strength, allowing the walls to be thinner.
It's so obvious that I assume someone's done it already and probably patented it.
The dangers of over thinking. I mean, I know that yours are better engineered and more effective