The puzzle: 25 orders of magnitude
Guide 1 handed us a fact and a mystery. The fact: resistivity (and its inverse, conductivity) is an intrinsic property, and across materials it spans about twenty-five orders of magnitude — copper sits near 1.7 x 10^-8 ohm-m, while window glass and diamond hover up around 10^14 to 10^16 ohm-m. That is the widest range of any property in all of materials science. It let us sort everything into three camps: conductor, semiconductor, and insulator. The mystery: guide 1 measured this split but never explained it. Why do the very same electrons that pour through copper sit frozen and useless in glass? The answer is not about atoms taken one at a time. It is about what happens to their electron energy levels when 10^23 atoms crowd together into a solid. That story is band theory.
Recall from the atomic-structure rung that a lone atom holds its electrons on sharp, discrete energy rungs, and the Pauli exclusion principle forbids any two electrons from sharing the exact same state. Now bring two atoms close: each shared rung must split into two slightly different rungs so the electrons don't collide. Bring 10^23 atoms together in a crystal and each original rung splits into 10^23 rungs, packed so tightly — all within an eV or two — that they smear into a near-continuous ribbon of allowed energies. That ribbon is an energy band. Picture one atom as a single seat at each musical pitch; a mole of atoms is a whole grandstand of seats so densely packed that the tiers become a smooth ramp. And between one band and the next there can lie a stretch of energies that no electron is allowed to have at all — a forbidden gap.
Two of these bands do all the work. The electrons that decide conduction are the outermost — the valence electrons you met in the bonding rung. The band they fill when the material is at rest is the valence band; the next empty band above it is the conduction band; the forbidden stretch between them is the band gap. That is the whole cast. Whether a material conducts comes down to just two questions: is the valence band completely full, and if it is, how tall is the jump up to the empty conduction band across the band gap? This is also the bonding rung finally cashing out — a metallic bond's shared electron sea versus a covalent or ionic bond's locked-down electrons turn out to be two different band pictures.
Full or not, and how tall the jump
Start with a metal. In a metal the highest occupied band is only partly filled — sodium's single valence electron only half-fills its band — or a full band overlaps an empty one. Either way there are empty allowed states sitting immediately beside filled ones, at essentially zero energy cost. Apply the faintest voltage and electrons simply slide sideways into those neighbouring empty states and drift along: that drift is current. This is the metallic bond's delocalized electron sea seen from the energy side — no gap to climb, empty seats always within reach. That is exactly what makes a metal a conductor.
Now an insulator. Here the valence band is completely full and the conduction band completely empty, separated by a large gap — several eV, about 5.5 eV in diamond. A completely full band carries no current at all, and this is the point people miss: it is like a fully booked train carriage where every seat is taken, so no passenger can shuffle anywhere, and there is no net motion no matter how hard you push. To conduct, an electron must be lifted clear across the gap into the empty conduction band above — and at ordinary temperatures almost none can make it. A semiconductor is the same picture with a small gap: silicon's is about 1.1 eV, germanium's about 0.67 eV. The step is low enough that a thin trickle of electrons, borrowing energy from heat, do jump — so a semiconductor conducts weakly, parked between the two extremes. This is the plain meaning of the slogan: the band gap is a step the electron must jump to conduct. A metal has no step; an insulator's step is a wall; a semiconductor's is just a stair you can occasionally clear. Notice how the covalent bond in silicon, which locks each electron into a shared pair, shows up here as a full valence band with a modest gap above it.
BAND PICTURE OF THE THREE CLASSES (energy increases upward)
[##] = states filled with electrons [ ] = empty states
METAL SEMICONDUCTOR INSULATOR
[ ] [ conduction ] [ conduction ]
[ empty ] [ empty ] [ empty ]
[############] <-- ....~1 eV gap... ====~5 eV gap====
[## filled ##] [## valence ##] [## valence ##]
[# (partly) #] [## full #] [## full #]
[############] [#############] [#############]
no gap: empty small step: a few big step: almost
states touch electrons hop at no electron ever
filled ones room temperature hops across
-> CONDUCTOR -> weak conductor -> INSULATOR
copper silicon (pure) diamond
~1.7e-8 ohm-m ~1e3 ohm-m ~1e14 ohm-mTwo carriers, and the water line
Look closer at what a jump leaves behind. When an electron is lifted across the gap, it abandons an empty seat down in the valence band. That once-full band is now not quite full — the electrons around the empty seat can shuffle into it, and as they do, the empty seat itself appears to drift the opposite way, behaving for all the world like a mobile positive charge. We call it a hole. It is not a real particle; it is the collective motion of many electrons around a missing one, dressed up as a single positive carrier because that is far easier to track. So exciting one electron creates two carriers at once: the electron up in the conduction band and the hole down in the valence band. In a pure crystal they are born strictly in pairs — the defining mark of an intrinsic semiconductor.
How much current a carrier actually delivers depends not only on how many there are but on how freely each one moves — its electron mobility, the drift speed it picks up per unit of electric field. Conductivity is the clean product of the two: the number of carriers, times the charge on each, times the mobility (with a matching term for the holes). That product splits the labour perfectly. A metal has a colossal number of carriers — every atom donates one or more — with a moderate mobility. A semiconductor has extremely few carriers but often a perfectly decent mobility. It is the tiny carrier count, not sluggish electrons, that makes pure silicon's resistivity (~10^3 ohm-m) sit eleven orders of magnitude above copper's (~1.7 x 10^-8 ohm-m). The conductivity equation lets you see at a glance which lever a given material is pulling.
One last piece of vocabulary you will meet everywhere: the Fermi level. Loosely, it is the 'fill line' of the electrons — at absolute zero, the energy below which every allowed state is occupied and above which every state is empty, exactly like the water line in a tank filled to a certain height. In a metal that fill line lands inside a band, right among empty states, so conduction is free. In an intrinsic semiconductor or an insulator the Fermi level sits roughly in the middle of the gap, so the nearest empty states are a whole gap away. That orientation is all you need for now — but hold on to it, because in the doping guide the Fermi level becomes the working tool: nudging that water line up toward the conduction band or down toward the valence band is precisely what turns a semiconductor into n-type or p-type material.
One number, many consequences
Because conduction in a semiconductor depends on electrons borrowing heat to clear the gap, temperature matters enormously — and here is where a real number makes the whole idea click. The fraction of electrons with enough thermal energy to make the jump scales roughly like exp(-Eg / 2kT), where kT is the available thermal energy: about 0.026 eV at room temperature, roughly one-fortieth of an eV. Feed in silicon's 1.1 eV gap: Eg / 2kT is about 1.1 / 0.052, which is about 21, and exp(-21) is around 10^-9 — about one electron in a billion clears the step. Now feed in diamond's 5.5 eV: the exponent is about 106, and exp(-106) is around 10^-46 — effectively zero, forever. That single exponential — the gap divided by the thermal energy — is the engine behind guide 1's fan of resistivities across 25 orders of magnitude.
Now watch the direction of the temperature effect, because it is the whole surprise. Heat a semiconductor and you liberate exponentially more carriers, so its resistivity falls as it warms — a hot semiconductor is a better conductor. A metal does the exact opposite: it already has all its carriers, so heating cannot add any; it merely makes the lattice atoms vibrate harder, and those vibrations scatter the drifting electrons, so a metal's resistivity rises with temperature. Same heat, opposite response, because one material is starved of carriers and the other is starved of a clear path. Impurities play the same two-faced game, and the reason a perfect, cold, pure crystal would in principle conduct with no loss at all is the whole subject of the next guide. For now, just hold the contrast: warmth helps a semiconductor and hurts a metal.
Two closing honesties. First, the reason semiconductors rule electronics is not their pure behaviour — one electron in a billion is a feeble conductor. It is that adding just a few of the right impurity atoms per million, called doping, injects extra electrons or extra holes and swings conductivity by orders of magnitude on command, letting us build n-type and p-type material and press them together into the p-n junction at the heart of every diode, solar cell, and transistor — the subject of guide 4. Second, the band gap is not a metaphor but a measurable energy: shine light on the material and photons carrying more energy than the gap get absorbed as they kick electrons across, while lower-energy photons pass straight through. That is why silicon (gap 1.1 eV, down in the infrared) looks opaque and grey while diamond (gap 5.5 eV, up in the ultraviolet) is transparent to every colour of visible light — and it is exactly the effect a solar cell harvests to turn sunlight into current.