[Coda]

Yep, that's a pretty good explanation of the phenomenon.

The middle layer is suuuuper interesting, actually. It's a pure (undoped) wafer of silicon crystal. Silicon has some WEIRD properties. Silicon is not exactly a metal, and it's not exactly a non-metal. It's a metalloid. It's not exactly an electrical conductor, but it's not exactly an electrical insulator. It's a semiconductor. You've probably heard that term before when talking about computers; well, you're about to learn what that actually means.

Pure silicon crystal, with no impurities whatsoever, has a very neat arrangement of electrons. Every atom has exactly the number of electrons it wants to have, and it's energetically stable, so the electrons don't have much motivation to go anywhere. It acts as a weak insulator.

But that's assuming that all of the electrons ARE where they belong.

It takes more energy than a normal electric current to dislodge those electrons from their happy place of stability, but it doesn't take THAT much -- it takes less than an insulator would take. And conveniently, photons near the visible light spectrum have enough energy to do the job. And when the electron is forced out of place, if it has somewhere it can go -- such as into a circuit -- then it leaves a positively-charged hole behind. That hole is unstable. There's a pressure to pull another electron from nearby into the hole. But that means THAT electron leaves a hole behind that needs to be filled. So eventually that hole has to get filled by an electron from outside.

But that's the behavior of pure silicon. If you mix in just a few atoms of phosphorus, then you have extra electrons. The region gains a small net positive charge, because those extra electrons will readily leave. If you mix in a few atoms of boron, then you have extra holes. The region gains a small net negative charge, because those holes will accept extra electrons readily. (Yes, it sounds a little bit counterintuitive that adding atoms with too many electrons will cause it to have a positive charge, and vice versa. This is because the phosphorus atom is neutrally-charged normally but when you mix it in, it loses the extra electron, thereby becoming positively-charged.)

So by sandwiching undoped silicon between a negatively-doped ("n-type") region on top and a positively-doped ("p-type") region on the bottom, all of the mobile electrons or holes in the undoped silicon that MIGHT allow conducting electricity are pulled away into the doped regions -- the silicon is "depleted" (hence "depletion zone" in the picture) of charge carriers, leaving a potential difference that further resists electrons passing through. And if a loose electron DOES get in there (perhaps because it was hit by a photon, eh?), it follows the bias of the electric field. So the current can only flow one way through it.