To the Digital Age: Research Labs, Start-up Companies, and the Rise of MOS Technology

Introduction

1 Introduction The MOS (metal-oxide-semiconductor) transistor, the fundamental element in digital electronics, is the base technology of late-twentieth-century and earlytwenty -first-century America. Through it digital electronics have entered almost every area of American life, first through the calculator, then through the digital watch, and finally through the microprocessor. The rise of the MOS transistor has made what was once ludicrous commonplace. In Woody Allen’s 1969 movie, Take the Money and Run, his character responds to the question ‘‘Have you ever had any experience in running a high-speed digital electronic computer?’’ by replying that his aunt has one—an absurd assertion given the state of computing in 1969. Where would his aunt get the $5 million or so necessary to buy one? Would she have it in a cavernous basement with raised floors and special cooling equipment? Would she oversee the teams of programmers and technicians required to use it? Would she use it for calculating missile trajectories, computing payrolls for thousands of employees, or storing her recipes? MOS technology has ruined a good joke. The adjectives that in the 1960s served as reminders of computers’ foreignness to people’s everyday experience are gone, tautological today. Of course a computer is high-speed, digital, and electronic. What else would it be? Today’s image of a computer is of a 2 Introduction personal computer, vastly more powerful than the largely forgotten 1969 behemoth , easily affordable, and relatively simple to use. Even if the aunt of today were too timid to buy one, she would likely own dozens of computers simply by having a car, a camera, a sewing machine, or almost any other household appliance of recent vintage. MOS technology is central to understanding this transformation of the computer and digital electronics. The density of MOS circuitry led to low-cost semiconductor memory and the microprocessor, making a personal computer possible. From the first popular personal computer, the Altair 8800 in 1975, MOS technology development has continued apace, making possible not only more and more powerful personal computers, but also the ever cheaper microcontrollers that have made digital electronics ubiquitous. In 1999, the semiconductor industry sold more than 5 billion microcontrollers and 328 million microprocessors.1 Although the term revolution is used promiscuously in regard to semiconductor electronics, with the precise nature of the revolution left to the reader’s imagination, the MOS transistor has been revolutionary in a way that can be well defined: it overthrew the previously dominant bipolar transistor. The bipolar transistor, invented by William Shockley in 1948, was the first technologically important transistor and for roughly two decades was the default meaning of the word ‘‘transistor.’’ The first transistor radio, the first transistorized television, the first integrated circuit, and the computer Woody Allen had in mind in 1969 all used bipolar transistors. In the 1960s, after some in the industry had begun work on the MOS transistor, a debate raged over whether it would find a place, and if it did, what the division of labor between the two transistors would be. While partisans on each side predicted the demise of the other technology, a sensible moderate view was that there would be room for both. In 1997, Gordon Moore, the head of research and development (R&D) at Fairchild Semiconductor in the 1960s and the cofounder of Intel, noted that contrary to his expectations in the 1960s, the final split was over 99 percent MOS.2 As Moore suggests, it has been a most improbable triumph. If one were to travel back through time to the late 1950s and try to explain to semiconductor engineers the course their field would take, one would be hard-pressed to know which they would find more incredible, the growth of semiconductor elec- 3 Introduction tronics or the fact that the MOS transistor was the vehicle of that growth. A 1950s engineer with a sense of history would have recognized the MOS transistor as a close relative of several of Bell Labs’ missteps on its way to inventing the transistor in 1947. Everything that happened in the intervening years reinforced the view that the MOS transistor was a failure. One could certainly build a transistor that way, but it would be much slower and much less reliable. An MOS transistor did not represent progress. It was not the wave of the future. The future would belong to new semiconductor materials, such as gallium...