408, the sealed system idea sounds interesting. Were it not for the outstanding toxicity of nickel carbonyl, that would be the obvious process to go with. According to a little googling, nickel carbonyl has an LC50 of 3ppm for 1/2 hour exposure (about 1/2 mg)... that makes cyanide look benign. Anyway, that's a good idea, I'm sure something sealed could be designed that uses convection/density differences to pump CO/Ni(CO)4 through the system from the cool to the hot ends.
ardbeg, I hadn't run into that idea yet, thanks. Apparently those phenolic oximes are widely used in mining to selectively separate copper, nickel, cobalt, etc. Lots of good patent data to look at.