1951 · Grown-Junction Transistor · Work 818 of 1,871
A germanium crystal of 1951 on a workbench beside a crystal-pulling furnace in a telephone research laboratory in New Jersey, on white filter paper a silvery grey rod two centimetres thick with two fine pale lines running across it near the middle, beside it a thin sawn cross-section slice, a black plastic block one centimetre long with three thin wires held in watchmaker's tweezers, hands in white coat sleeves, a magnifier lamp, behind them soft in focus the glass bell of the furnace with a crucible glowing orange, daylight through tall windows.
What it was used for
The grown-junction transistor came from Bell Telephone Laboratories: they pulled single crystals of germanium and, from April 1950, dropped two successive doping pellets into the melt while pulling, producing an n-p-n sequence with a thin base inside one piece of crystal. The device sat in a hard plastic bead about 3/16 inch across and amplified very small signals on very little supply power: the Bell Laboratories Record of August 1951 cites an audio oscillator needing 6 microamperes at 0.1 volt. There was no series production at first. In May 1952, according to Radio-Electronics, only Western Electric reported making junction transistors, and expressly on an experimental basis only, with an output of less than 100 per month, while total US output that month was nearly 8,500 transistors, almost all point-contact types.
What was new
Against the point-contact transistor it replaced, the gain was in noise: the Bell Laboratories Record of August 1951 states that the noise figure is 1000 times less than that of its predecessor, and gives 10 to 20 db at 1000 cps. The same house journal notes that collector capacitance limits the frequency response at full gain to a few kilocycles, but that with a suitable impedance mismatch it stays flat to at least one megacycle with a useful amount of gain. Crystal pulling itself was not new; the Computer History Museum lists Czochralski's 1917 method as a forerunner. What was new was dropping dopants into the melt during the pull, making the junction inside the grown crystal. Shockley had filed the junction-transistor theory on 26 June 1948; US 2,569,347 issued on 25 September 1951.
What the sources leave open: Shockley's 1949 theory paper was not accessible; the year appears only indirectly in a Bell paper of 1 July 1951. The frequently quoted 6 per cent yield appears in a museum footnote to the reprint, not in the Bell article itself, and is therefore omitted. No source gives unit numbers or a price for the M1752 development type.