MOSFET Scaling

MOSFET scaling is the practice of making the transistors on an integrated circuit progressively smaller from one technology generation to the next, which is the engine behind the decades-long improvement in chip density, speed, and cost per function. The central idea is deceptively simple: shrink the dimensions of the metal-oxide-semiconductor field-effect transistor — its length, width, and the thickness of its gate insulator — and more transistors fit in the same area, each switches faster, and the cost of each falls. This relentless shrinking is the physical substance behind the famous observation that the number of transistors on a chip grows exponentially over time.

Scaling works because a smaller transistor is not merely a miniature copy but a device with genuinely better characteristics, at least while the shrinking is done carefully. A shorter channel lets charge travel between the transistor's terminals more quickly, so the device can switch faster. A smaller device occupies less area, so more of them fit on a die, raising density and lowering cost per transistor. A thinner gate insulator gives the gate stronger control over the channel. When all the dimensions and associated parameters are reduced together in a coordinated way, these benefits compound, which is why generations of smaller transistors have been simultaneously denser, faster, and cheaper.

The disciplined version of scaling follows rules that specify how all the transistor's parameters should change together, most famously the constant-field scaling described by Robert Dennard and colleagues. Simply shrinking one dimension while leaving others fixed creates problems; the insight of systematic scaling was that dimensions, voltages, and doping should be reduced in concert so that the internal electric fields stay roughly constant as the device shrinks. Following such rules kept the transistor well-behaved across many generations and tied the shrinking of size to proportional improvements in speed and power, which is what made scaling a predictable, plannable path rather than a gamble each generation.

Scaling eventually runs into fundamental physical limits, and much of the modern history of chip technology is about fighting those limits. As transistors became extremely small, the gate lost some of its ability to fully control the channel, insulators became thin enough that charge leaked through them, and the power dissipated per unit area rose to levels that threatened to overheat the chip. These problems meant that the simple, clean scaling of earlier decades could no longer continue unmodified. The industry responded with new materials, new transistor geometries, and design changes, which are best understood as ways to keep the benefits of scaling going after the straightforward approach stopped working.

The significance of MOSFET scaling is that it turned transistor miniaturization into a sustained economic and performance engine for the entire electronics industry. Because smaller transistors meant cheaper, faster, denser chips on a regular cadence, whole industries could plan around steady improvement, designing products that assumed next year's chips would be better and cheaper. Scaling is the mechanism that made that expectation reliable for decades, and understanding it is understanding why electronics got relentlessly more capable and less expensive year after year.

Read MOSFET scaling as the shrink that pays three ways at once: smaller transistors that are denser, faster, and cheaper when their dimensions and voltages are reduced in coordinated fashion, an engine of steady improvement that drove the industry for decades until the physics of very small devices forced new materials and shapes to keep it alive.