Mesa Transistor
A mesa transistor is a diffused-junction transistor in which the individual device is isolated by etching away the surrounding semiconductor, leaving a raised, flat-topped plateau like a mesa in a desert landscape. The wafer is diffused across its whole surface to form the base and emitter layers, then the unwanted material is etched down through the junctions so each transistor stands alone on its own pedestal, with the collector junction exposed around the sides. Developed at Bell Labs and taken to market by Texas Instruments and Fairchild in the late 1950s, the mesa structure made high-frequency diffused transistors manufacturable in quantity, and Fairchild's 2N696 and 2N697 family, which first shipped in 1958, became one of the most recognizable silicon mesa products.
Etching a plateau solved a real manufacturing problem: how to turn a wafer that is diffused everywhere into many separate, small junctions. A wafer diffused across its whole surface has a base layer under every square millimetre, so cutting it into chips without cutting the layer back would leave a large collector junction area, with heavy capacitance and leakage and no way of making a small, fast device. By masking the active area and etching around it, the junction area shrinks to the size of the plateau, the collector capacitance drops, and the device becomes faster. Early etch masks were crude, but with photolithography, applied to transistors by Lathrop and Nall at the US Army's Diamond Ordnance Fuze Laboratories in 1957 and 1958, the mesas could be defined reliably across a whole wafer at once.
The price of the mesa was an exposed junction, and that weakness defined the next step in the field. The etch cuts through the collector-base junction, so the junction ends on the bare sidewall of the plateau, where moisture, ions and dust can change surface charge and open leakage paths. Devices that tested well on the bench could drift or fail in service, and manufacturers kept them in hermetically sealed metal cans with careful cleaning and sometimes protective coatings. Jean Hoerni at Fairchild identified this contamination weakness as a core reliability limit and sought a way of leaving the junction buried under a protective layer rather than uncovered, an idea that grew into his 1959 planar process.
Fairchild's 2N696 and 2N697 show the mesa at its commercial moment, and they should be described carefully. Fairchild Semiconductor, founded in September 1957 by the Traitorous Eight, shipped diffused-base silicon mesa transistors in 1958, and the 2N697 became a long-running n-p-n type with a metal can package that was used in military and industrial circuits. The 2N696 and 2N697 are mesa devices, not planar ones; the planar 2N1613 followed in 1960 as the same kind of electrical product with a better surface. A frequent confusion is to date the planar process to the same devices because Fairchild's best-known products were around at the same time, when in fact the mesa products came first and the planar types answered their weaknesses.
The mesa idea is far from extinct, even though the mesa transistor was largely replaced. Etched mesa structures remain common in high-voltage power diodes and thyristors, in photodiodes and avalanche photodiodes, in many laser diodes and optical modulators, and in some radio-frequency transistors, where isolating a device by etching is simpler than doping around it. Modern mesa devices differ in having passivation: a glass or dielectric layer covers the sidewall, solving the exposure problem that early transistor engineers could only avoid by sealing the whole device in a can. The history of the mesa is thus a story about a good idea paired with an unsolved surface problem, and about how long a surface can take to solve.
Read mesa transistor as the practical, high-frequency diffused transistor made by etching a plateau, brilliant at making small junctions and honest about the contamination risk that the planar process was invented to fix.