Where guide 3 left off: pure silicon barely conducts
From guide 2 you carry the picture that decides everything: a semiconductor like silicon has a full valence band, an empty conduction band above it, and a band gap — a forbidden step, about 1.1 eV wide for silicon — that an electron must jump before it can carry current. That gap is the whole difference between silicon and a metal, which has no gap at all, and a diamond insulator, whose gap of 5.5 eV is simply too tall to leap. Silicon sits in the sweet middle: the step is real, but not impossibly high.
In perfectly pure silicon the only way to conduct is to borrow thermal energy and kick an electron across that step by luck. Every electron that makes the jump leaves an empty seat behind in the valence band — an electron-hole pair. This is why a pure semiconductor is called an intrinsic semiconductor: its handful of carriers come only from its own thermally-excited electrons, in exactly matched pairs of one electron and one hole. And there are only a handful. At room temperature silicon manages about 10^10 carriers per cubic centimetre — which sounds like a lot until you remember there are about 5 x 10^22 silicon atoms in that same cubic centimetre. Roughly one atom in a trillion has lent out a carrier.
N-type: slip in an atom with one electron to spare
Silicon is group IV: each atom shares its four valence electrons in four covalent bonds, and every electron is tied up in a bond. Now the trick of doping: replace one silicon atom in the lattice with a group-V atom such as phosphorus, which brings five valence electrons to a neighbourhood built for four. Four of them settle into the four bonds as usual. The fifth has no bond to join — it is left dangling, held to its phosphorus core only weakly. This kind of dopant, generous with an electron, is a donor.
How weakly is that fifth electron held? In band language the donor creates a filled level sitting just below the conduction band — for phosphorus in silicon, only about 0.045 eV below it, versus the full 1.1 eV band gap. Remember that room-temperature thermal energy is about 0.026 eV. So where the full gap is a step an electron clears only by rare luck, the tiny donor step is one it clears almost as fast as it is offered: at room temperature essentially every donor has already given up its electron to the conduction band. Doping does not lower the whole gap — it plants a convenient stepping-stone right below the top of it.
Now count. Add just 10^16 phosphorus atoms per cubic centimetre — about one phosphorus for every five million silicon atoms, a few parts per million — and you flood the conduction band with roughly 10^16 free electrons. Compare that to the 10^10 you had intrinsically and the free-electron count has jumped a millionfold, so the conductivity jumps by roughly the same factor. Because the current is now carried overwhelmingly by negative electrons, this is an n-type semiconductor. The electrons are the majority carriers; the few thermal holes still lurking are the minority carriers.
P-type: take an atom with one electron too few
Run the same idea the other way. Replace a silicon atom with a group-III atom such as boron, which brings only three valence electrons to a site that wants four. Three of its bonds fill normally; the fourth bond is left one electron short — a vacancy in the bonding network. This dopant, hungry for an electron, is an acceptor, and it creates an empty level sitting just above the valence band (again only about 0.045 eV up for boron in silicon). A valence electron next door needs only a tiny thermal nudge to hop up and fill it — and when it does, it leaves behind a hole in the valence band.
Here is the picture that trips up every beginner, so slow down. A hole is not a positive particle you can hold — it is the absence of an electron, and it moves because its neighbours move. Picture a full theatre row with one empty seat. A person shuffles over to fill it; now the seat they vacated is empty. Someone fills that; and so the empty seat travels down the row in the direction opposite to the people. The hole behaves exactly like a real particle carrying positive charge, drifting the opposite way to electrons in an electric field. Because current here is carried by these positive holes, this is a p-type semiconductor: holes are the majority carriers.
Counting carriers, mobility, and the honest limits
Conductivity is bookkeeping: how many carriers you have, each times how much charge it carries, times how nimbly it moves. In symbols, conductivity = (electron count)(charge)(electron mobility) + (hole count)(charge)(hole mobility). In intrinsic silicon both counts are equal and both terms tiny. Doping smashes that symmetry — in n-type the first term dwarfs the second; in p-type, the reverse — and lets you dial the carrier count over many orders of magnitude just by choosing how much dopant to add. That deliberate control, not raw conductivity, is the superpower.
Mobility deserves an honest footnote. In silicon, electrons move about three times more nimbly than holes (roughly 1400 versus 450 cm^2 per volt-second), because a hole's motion is really the laborious shuffle of a whole row of valence electrons. That asymmetry is not a curiosity — it is why n-channel transistors switch faster than p-channel ones, and why chip designers must size the two types differently. And just as in guide 3, mobility falls as temperature rises and as you add more dopant, because hotter atoms and more impurity sites scatter the carriers more.
Silicon band picture (energy increases upward)
conduction band ---------------------------- empty at 0 K
^ . donor level (P): ~0.045 eV below CB -> n-type
BAND | gap = 1.1 eV
GAP | o acceptor level (B): ~0.045 eV above VB -> p-type
v
valence band ========================== full at 0 K
intrinsic Si : ~10^10 carriers / cm^3 (electrons = holes)
silicon atoms: ~5 x 10^22 atoms / cm^3
add ~10^16 P / cm^3 (about 1 P per 5,000,000 Si atoms)
-> ~10^16 free electrons / cm^3
-> ~1,000,000x more conductive, and now n-typeTwo honest limits. First, a doped (extrinsic) semiconductor is only well-behaved across a middle temperature range — the extrinsic region where the carrier count sits pinned at the dopant level. Go cold enough and carriers "freeze" back onto their donors; go hot enough and thermal generation across the full gap swamps the doping, and the material lapses back toward intrinsic behaviour with all its temperature sensitivity. Second, this whole clean story rests on staggering purity: the background impurities must be far below the intended dopant level, or you cannot control which carrier wins. Semiconductor silicon is among the purest matter humans make.
Put n and p together: the p-n junction
Neither n-type nor p-type alone does anything magical — each is just a resistor you can tune. The magic starts when you join them. Fashion a single silicon crystal that is n-type on one side and p-type on the other, and you have made a p-n junction, the atom from which nearly all electronics is built. At the boundary, the two populations cannot ignore each other. Electrons, crowded on the n-side, diffuse across into the p-side; holes, crowded on the p-side, diffuse into the n-side. Where they meet they recombine — an electron drops into a hole — and both vanish.
That mutual clearing-out leaves behind a thin depletion region straddling the junction, swept empty of mobile carriers. But the fixed ionised cores stay: exposed positive donor ions on the n-side, exposed negative acceptor ions on the p-side. Those uncovered charges set up a built-in electric field pointing from n to p — a voltage hill, about 0.7 V for silicon — that pushes back on any further diffusion. The system settles into equilibrium when the field's push exactly balances the diffusion's shove. You now have a lopsided device: it is far easier to send current one way across that hill than the other.