The efficiency of most existing heat-driven refrigerator and heat-pump designs is kind of disappointing. This isn't a fundamental thermodynamic issue; the Carnot efficiency for a gas burner at 2000 C running a refrigerator offering a 40 C temperature drop is close to the electric efficiency. In principle, it could be a more efficient way to use stored hydrogen than a fuel cell. In practice, it isn't. Maybe actually taking advantage of such high temperature differences requires very good refractory alloys. I would be curious to know the absolute efficiency proposed here rather than just a rough estimate of relative efficiency.