FinFET

Every computer chip consists of billions of tiny electrical switches. These switches are called transistors and know only two states: current flows or current doesn’t flow. This is how the zeros and ones that a computer computes with come about. For a long time, these switches were built flat: a thin conducting path on the surface of the silicon, with a control electrode above it. In a FinFET, the conducting path instead stands upright like a narrow fin. This is exactly where the name comes from: “fin” refers to something like a fish fin or ridge.

Why flat transistors hit their limit

Chips have been getting smaller for decades. Smaller switches mean more computing power in the same area and less power consumption per computing step. But starting around 2010, a problem arose with flat transistors. They became so short that the control electrode could no longer cleanly switch off the current flow.

The result is called leakage current: even in the switched-off state, current would seep through the channel. With billions of transistors, this adds up to real wasted heat. Phones got hot, batteries drained faster, data centers needed more cooling. Simply continuing to shrink the chips would therefore have made them worse instead of better.

The FinFET solved this problem and thereby made the continuation of miniaturization possible. Starting around 2012, the industry gradually switched over to it. Almost every smartphone and every graphics card from the 2010s is full of FinFETs. Without this design, the performance leap of the last ten years would not have happened this way.

The fin and the grip around the channel

A transistor has three important parts. Current comes in from the source, arrives at the drain, and the channel lies in between. A control electrode, called the gate, decides whether the channel conducts or blocks. In a flat transistor, this gate touches the channel only from above. Control is thus one-sided and becomes weak with very short channels.

In a FinFET, the channel is etched out of the silicon and stands up as a thin, tall wall. The gate is laid over this wall and wraps around it from three sides: left, right, and top. You can picture it like a hand that doesn’t just press a garden hose from above but grips it all the way around. This lets the hose be pinched shut much more cleanly.

This three-sided control strongly suppresses leakage current. In addition, several fins can be built side by side and switched together. This gives you more current capacity without using up more base area. That’s why some manufacturers also call this design Tri-Gate.

From the smartphone to the successor technology GAAFET

You never see a FinFET with the naked eye, but you use them constantly. Processors in phones, laptops, and game consoles from recent years are based on this design. The graphics chips on which AI models are trained were also long manufactured using FinFET technology. In company news, the term usually comes up with chipmakers like TSMC, Intel, or Samsung.

FinFETs are often confused with figures like “5 nanometers” or “3 nanometers.” These numbers today are purely marketing names for a manufacturing generation and no longer a genuine measurement. A FinFET, by contrast, describes the shape of the component, not its size. The two are related but mean different things.

By now, the FinFET is also hitting its limits, because the fins can hardly be made any thinner. The successor technology is called GAAFET, or nanosheet. There, the gate wraps around the channel from all four sides instead of just three. Since around 2022, the major manufacturers have been switching to it for their most advanced chips. Even so, FinFETs won’t disappear anytime soon, since they will remain in use for cheaper chips for a long time to come.