Skip to content

Dead End: The Cryotron

Abstract

In the mid-1950s the cryotron, a superconducting switch invented by the MIT engineer Dudley Allen Buck (1927–1959), looked like the component that would replace the vacuum tube and shrink a computer to a cubic foot. It worked only in liquid helium, switched slowly, and was overtaken by the silicon transistor and the integrated circuit. Buck died at 32 in May 1959, and the cause of his death is still disputed. Superconducting logic came back twice, as IBM’s Josephson-junction programme of 1964 to 1983 and in today’s quantum computers.

Dudley Buck

Buck was born in San Francisco on 25 April 1927, the son of a postal clerk, and spent his teens with a grandmother in Santa Barbara, building amplifiers for dances and concerts and working as a radio amateur. From 1944 he studied electrical engineering in a Navy programme in Seattle, graduating in 1947, and then served until 1950 with naval intelligence in Washington. There he met people from Engineering Research Associates (ERA), the St. Paul firm that built the Atlas codebreaking computer for the Navy, and with colleagues he built a relay computer, ABEL, on which Atlas programmers could practise.

From 3 July 1950 he was a research assistant at MIT, first in Jay Forrester’s Whirlwind project, which was then developing magnetic-core memory; his 1952 master’s thesis dealt with the subject (see Whirlwind, SAGE, and Core Memory). He also advised the intelligence services: he attended MIT’s exclusive 1952 conference on machine translation, a CIA paper of 1953 records contacts with the Soviet trade agency Amtorg, and another document lists him as an informal NSA collaborator. He invented a content-addressable memory, which retrieves data by its contents instead of its address, and took his doctorate at MIT in 1958.

The Switch

The cryotron exploits the fact that a magnetic field destroys superconductivity. A straight tantalum wire is wound with a coil of niobium; immersed in liquid helium, a few degrees above absolute zero, both become superconducting. A current through the niobium coil produces a magnetic field that returns the tantalum to its normal, resistive state, so one current switches another, as in a relay or a tube. Buck’s notebook records the idea on 15 December 1953 (the HNF dates the invention to February 1954); he had working devices by 1955 and filed a patent in July of that year.

The attraction was size. In 1956 Life gave him a full page, showing him with a vacuum tube in one hand and a cryotron in the other, and told readers the device might lead to a computer of one cubic foot. In 1954 Bell Labs’ transistor computer TRADIC was still the size of a wardrobe (see From Vacuum Tubes to Transistors); military planners imagined a cryotron computer small enough to fit in the nose of a missile. Computer firms and the consultancy Arthur D. Little, MIT’s neighbour, took up the idea; RCA built cryotron circuit boards in the 1950s and an integrated cryotron circuit in 1965, and in 1957 John Bremer at General Electric developed thin-film cryotrons of lead and tin. In December 1958 Buck and Kenneth Shoulders presented “An Approach to Microminiature Printed Systems,” on building such circuits from deposited thin films, at the Eastern Joint Computer Conference.

May 1959

In April 1959, during a brief thaw in the Cold War, a delegation of Soviet computer experts toured the United States and visited Buck’s laboratory at MIT; he declined to demonstrate his invention, and they left disappointed. That spring he was working on thin films. His notebook for 18 May 1959 records an experiment with boron trichloride, a highly toxic gas. That evening he felt unwell, in the night he developed a fever and could no longer move, and he was taken to hospital in Winchester near Boston, where antibiotics did not help. He died on the morning of 21 May 1959.

The cause was never established. The historian David Brock has suggested careless handling of chemicals; the biography The Cryotron Files (2018) by Iain Dey and Buck’s son Douglas Buck leans towards poisoning by the KGB. The computer pioneer Louis Ridenour died the same day, which fed speculation, although Douglas Buck has said there is no evidence connecting the two deaths.

Why It Failed

The cryotron had two problems that no amount of engineering removed. It needed liquid helium, which made every machine a refrigeration plant. And it was slow: a cryotron switched by moving its gate between the superconducting and the normal state, which took far longer than switching a transistor. Meanwhile the silicon transistor grew faster and cheaper every year, and from 1959 the integrated circuit offered the miniaturisation the cryotron had promised, at room temperature (see The Integrated Circuit Revolution). Industry worked on the cryotron for a few more years, but no cryotron computer ever reached the market.

Afterlife

Superconducting logic returned in a faster form. In the Josephson junction a current tunnels between two superconductors through a thin insulating barrier; Juri Matisoo at IBM demonstrated a Josephson switch under 800 picoseconds, and IBM pursued Josephson-junction computers from 1964 until it abandoned the programme in 1983, again defeated by silicon. Today superconducting circuits are the basis of the leading quantum computers, which run at even lower temperatures than Buck’s cryotrons (see Quantum Computing).

📚 Sources