A measuring bench of 1957 in a semiconductor laboratory in Tokyo, on a grey bench a clear hexagonal plastic box with a round lid, inside it a tiny diode with two thin wire legs, two small closed cream card boxes lying face down beside it, behind them a curve tracer with a round glowing screen on which a single green curve rises, falls and rises again, a hand in a white coat sleeve turning a black knob, a vacuum flask at the edge with a thin white vapour rising, daylight from the left.
What it was used for
From 1956 Leo Esaki at Sony investigated interband tunneling in semiconductor diodes. He names the use himself in his 1973 Nobel lecture: negative resistances matter because of possible signal amplification. The Semiconductor History Museum of Japan names high-frequency oscillation, amplification and high-speed switching as fields of use. A tunnel-diode memory was demonstrably built: a 1962 publication reports the design and construction of a memory and associated circuits which ran continuously for several months at a 10 Mc/s clock frequency, with a complete read-write cycle of 100 nanoseconds. Later, per the Japanese museum, a tunnel-diode amplifier for 14 to 14.5 GHz was mounted on the INTELSAT V communication satellite.
What was new
Esaki places the device himself, with a caveat, in his Nobel lecture: the tunnel diode is "probably the first structure in which the negative resistance effect was observed". He says this within a talk on tunnel structures; right afterwards he shows a similar characteristic in a metal-oxide-semiconductor tunnel junction. It was achieved with an extremely thin, heavily doped junction; he gives a calculated width of approximately 200 Å, confirmed by capacitance measurements. He names the forerunners himself: Zener proposed interband tunneling in 1934, Chynoweth and McKay examined silicon junctions of low-voltage breakdown in 1957, and his own group first obtained a backward diode that conducted better in reverse than forward. The finding came out of a failure analysis: per the Japanese museum it arose while investigating large quantities of defective products in transistor production for radios. His talk in October 1957 drew little attention, according to the same source.
What the sources leave open: Esaki describes the first observation in the Nobel lecture without a date; July 1957 is given by the Semiconductor History Museum of Japan. The result was published in Physical Review 109 (1958) 603 and presented at the Brussels conference of 1958, proceedings 1960. Unit numbers and prices for tunnel diodes are not documented; no source opened names by company the tunnel-diode computers often attributed to RCA, IBM or General Electric.