From FinFET to GAA — Why the Transistor Had to Go Vertical

Why did a flat transistor grow a fin, and then become a stack of thin slabs? With three tools — short-channel effects, DIBL, and the natural length — this piece shows that FinFET and GAA nanosheets were not a matter of taste but a corner physics painted the industry into. No prior knowledge assumed.

The Day the Flat Switch Stopped Closing

A transistor is a sluice gate opened by voltage (the basics are in MOSFETs from the Ground Up). Put voltage on the gate and a channel — a path for current — forms underneath it; take the voltage away and the path disappears. From the 1960s until roughly 2010, that gate was flat: a thin insulating film laid on the silicon surface, with a gate electrode on top. One plank pressed down on the waterway from above.

Pressing from one side only is fine as long as the waterway is long. The trouble starts when you shrink it. The shorter the gate gets, the more something other than the gate starts having a say over the channel. That something is easy to name: the high voltage sitting on the drain, at the exit end.

This is the short-channel effect, and it is fair to say that both FinFET and GAA (gate-all-around) exist for the sole purpose of defeating it. Going vertical was not an aesthetic choice. It was the last exit left.

The Analogy: From Pressing a Plank to Closing Your Hand

Picture shutting off a garden hose. The planar transistor is a plank pressed down from above. Directly beneath it the hose flattens, but the sides keep their shape. If the hose is long, the flattened stretch is long enough to stop the water — but keep cutting the hose shorter and the un-flattened sides alone will carry water straight to the outlet.

The fix is intuitive. Stop pressing with a plank and close your hand around it instead. Three fingers squeeze far better than one plank, and a full grip closes it almost completely.

• Plank from above → planar MOSFET
• Squeezed from three sides → FinFET (the channel stands up like a fin)
• Wrapped on all sides → GAA / nanosheet (the channel becomes a thin slab)

Those three lines are the entire structural history of leading-edge silicon from 2011 to today. The rest of this article translates "why does adding gated sides help?" into numbers you can design with.

What Short-Channel Effects Actually Look Like

Measured on real silicon, short-channel effects show two faces.

Threshold voltage roll-off: even devices from the same process show a lower $V_{th}$ the shorter their gate is. They are supposed to share one "closing voltage," yet the short ones open early. A device whose gate came out a couple of nanometers short due to process variation is, electrically, a different transistor.

DIBL (Drain Induced Barrier Lowering) is nastier: on one and the same transistor, raising the drain voltage lowers the threshold. The gate has done nothing, yet the voltage at the exit reaches sideways and pulls down the energy barrier at the channel entrance. The gatekeeper is holding the gate shut while someone on the far side tugs it open a crack.

DIBL becomes a number as soon as you measure the threshold at two drain voltages and take the difference.

In plain words — how many millivolts the threshold drops when you raise the drain voltage by one volt. The unit is mV/V, and smaller means the gate is still in charge. $V_{th,\mathrm{lin}}$ is the threshold measured at low drain voltage, $V_{th,\mathrm{sat}}$ the one measured at high drain voltage.

Why is this fatal? Because off-state leakage depends exponentially on $V_{th}$ — roughly 60 to 90 millivolts per decade at room temperature. Let DIBL shave a few tens of millivolts off the threshold and standby power doubles, or grows by an order of magnitude. Short-channel effects are not a story about losing a little performance. They are a story about a chip that heats up while it is doing nothing.