When you hold a contemporary chip up to the light and realize you’re looking at billions of copies of something that was initially created by two people in a lab in New Jersey who weren’t even sure their employers wanted it, you feel a certain kind of humility. Beneath every phone, laptop, and server rack in use today is that bizarre origin story. The MOSFET is the device, and dissecting one layer by layer reveals nearly all of the reasons why silicon prevailed.
Naturally, the breakdown begins with silicon. A crystal created with almost perfect atomic order that serves as a stable foundation for everything placed on top of it—not the untidy, impure variety found in sand. In order to encourage electrons to behave reliably, engineers then dope some areas with minuscule amounts of foreign atoms, such as phosphorus or boron. It sounds a lot like flavoring food. In actuality, it is what transforms inert silicon into a device that can precisely direct electrical current.
The gate, which operates more like a faucet than a door, is located above that. Here, a tiny voltage is supplied to open or close the current flow between two additional regions, the source and the drain, without the need for any moving parts, the sluggish warm-up time, or the glowing filaments that afflicted vacuum tubes for decades. An incredibly thin oxide layer, an insulating screen that prevents the switching mechanism from short-circuiting itself, separates the gate from the channel below. It turns out that this device only functioned because of that layer.
Mohamed Atalla and Dawon Kahng, who were employed at Bell Laboratories in 1959, are primarily responsible for this discovery. Atalla had already devoted years to researching the electrical misbehavior of untreated silicon surfaces, a phenomenon known as surface states that had thwarted previous attempts to construct this type of transistor since the 1920s. Kahng, a physicist who had recently completed his doctorate, was given the actual construction task. They had a functional device by November of that year. It wasn’t quick.
According to modern accounts, it operated about 100 times more slowly than the bipolar transistors Bell was already producing. One noteworthy aspect of this history is that Bell Labs itself reportedly remained indifferent to the concept for years afterward, preferring the transistor technology it had already made investments in.
In retrospect, the hesitancy seems almost endearing. The true advantage of the MOSFET was not its speed but rather its ease of mass production and exceptional density packing, which bipolar transistors were unable to match at scale. Sensing what Bell had been reluctant to do, Fairchild, RCA, and General Microelectronics all launched their own MOS research programs in a matter of years. The density curve that was subsequently codified as Moore’s Law was made possible by the fact that MOS chips were already more affordable and denser than their bipolar competitors by the middle of the 1960s.
