Harold M. Agnew (Editorial, 13 Sept., p. 1475) has proposed that the United States buy annually from Russia the supply of tritium expected to be needed for U.S. nuclear weapons in about 10 to 12 years and argues that this approach would be substantially cheaper than either the accelerator approach to producing tritium or its production in commercial-type light-water power reactors in the United States.
I agree and, as a result of discussions with responsible Russian officials, believe that one can satisfy U.S. need for tritium in this way at about one-tenth the cost of domestic production.
Furthermore, by carrying one production design through to detailed planning and obtaining requisite permits, one answers possible concerns about a sudden cutoff of the supply of Russian tritium, in view of the buffer provided by the 12.3-year half-life of existing tritium stocks. I have studied tritium production options for the Department of Energy, and these are my personal conclusions.
Gustave K. Kohn (Letters, 25 Oct., p. 481) gives no indication of the cost of tritium production by low-voltage glow discharge, if indeed tritium is produced by such an approach. The world price of tritium is some $2 per curie, and Kohn tells us that 5 kilowatts for "hundreds of hours" produces "10's of nanocuries." If we assume the duration is 400 hours and that the "tens" are 30, then we see that 2000 kilowatt-hours ($100 at $0.05 per kilowatt-hour) produces 30 nanocuries, for a cost of $3 billion per curie.
Although the reactor approach and the accelerator may be 10 times more costly than buying tritium from Russia, a factor of a billion is out of sight.
Richard L. Garwin
IBM Fellow Emeritus,
Post Office Box 218,
Yorktown Heights, NY 10598, USA