Electricity/MOSFETs/Inside a MOSFET

Inside a MOSFET

Metal, oxide, semiconductor — the name is a description of the sandwich. Run it once and you can see where the channel comes from, why nothing crosses the gate, and why the gate stays on after you let go of it.

The name is the recipe. Metal on top, an oxide under it — silicon dioxide, which is glass — and semiconductor under that. FET is field-effect transistor, which is the mechanism: a field does the work, not a current.

Press the button and watch the middle of the drawing.

Inside, while the gate goes positive
gate at 0 V
As built, it cannot conduct. Source and drain are two islands of n-type silicon in a p-type body. Between them is p-type silicon, which means two junctions back to back — and one of them is always the wrong way round, whichever direction you push.

Source and drain start as two islands of n-type silicon in a p-type body, which is two junctions back to back — one of them always the wrong way round, so nothing gets through in either direction.

Then the gate goes positive. The field reaches through the glass, pushes the body's holes down and away, and pulls electrons up to the underside of the oxide. Enough of them arrive and that thin strip stops behaving like p-type silicon and starts behaving like n-type: a wire, joining the two islands, made out of a piece of the body.

That is the channel, and the word enhancement on a datasheet means exactly this — there is nothing there until the gate makes it.

Nothing crossed the glass

Worth saying twice, because it is the fact everything else rests on. No charge moved from the gate into the silicon. It cannot — that is what an insulator is. The gate only held a voltage, and the field did the rest.

Which is where the good news stops.

An insulated gate remembers

A gate that current cannot get into is a gate that current cannot get out of either.

Let go of the gate and see what happens
gate straight to the pin
Pin low, lamp off. Watch what happens after the pin stops driving — that is the frame that matters, not the one where the lamp comes on.

Two things fall out of that, and both are on every MOSFET schematic you will ever see:

  • A pull-down resistor, gate to source, usually 10 kΩ. It is the only thing holding the part off while the board is resetting and every pin is an input.
  • The gate is never left floating. Not while you test something, not for five minutes. A floating gate drifts to whatever the nearest static charge suggests, and a half-open MOSFET on a real load gets hot with nobody having asked it to.

The diode you did not fit

One more thing is built in and it is not optional. The body of the silicon is connected to the source inside the package, and the body and the drain are a junction — so there is a diode across every MOSFET, from source to drain, permanently.

Fit an N-channel part the right way round and it points backwards and you never see it. Fit one the wrong way round, or ask it to block current flowing the other way, and the diode conducts regardless of the gate. A MOSFET is a one-directional switch, and that is why.

It also earns its keep: it is the diode that catches the flyback spike from a coil, which is why a MOSFET switching a relay survives things a bare transistor does not.

Edit this page — content/fundamentals/electricity/inside-a-mosfet.mdx

Discuss this article

Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.

Browse Fundamentals on the forum