Band Theory: Metals, Semiconductors & Insulators

From a handful of orbitals to a continuous band

You have already done the hard part of this guide — twice. In the molecular-orbital rung you saw two atomic orbitals combine into one bonding level and one antibonding level; in the delocalization guide you watched three p orbitals on ozone spread into a delocalized pi system of three levels, and then read the promise out loud: push the same idea to thousands of atoms and the levels merge into continuous bands. This guide cashes that promise. Band theory is not a new theory at all — it is MO theory with the atom count turned up to the absurd numbers found in a real crystal.

Keep the one counting rule that has survived every step: N orbitals in, N orbitals out. Line up two sodium atoms and their 3s orbitals make two levels. Line up a hundred and you get a hundred levels, fanned out between the most-bonding combination at the bottom (no nodes, electron density flowing smoothly atom to atom) and the most-antibonding at the top (a node between every pair). Now line up the roughly 10^23 atoms in a visible grain of sodium. You still get exactly that many levels — but spread across the same finite energy range, so they are crammed unimaginably close together, separated by gaps far smaller than any thermal jiggle. We stop drawing them as rungs and start drawing them as a single smeared block of allowed energies: a band. The band is wide when the orbitals overlap strongly and narrow when they barely touch.

Gaps, the valence band, and the conduction band

A solid usually has more than one band, because each set of atomic orbitals spreads into its own band — the 3s orbitals make one band, the 3p orbitals another, and so on. Here is the decisive new feature that two-atom diagrams could not show you: bands need not touch. Between the top of one band and the bottom of the next there can be a band gap, a stretch of energy in which no orbital exists and therefore no electron is allowed to sit — a forbidden zone, like a missing floor in a building. Whether a gap opens, and how wide, depends on the atoms and how they bond; this is the single number that will sort every solid in the next section.

Fill the bands the way you have filled every level since the Aufbau rung: from the bottom up, two electrons per state. Two bands then earn names. The highest band that still holds electrons at low temperature is the valence band; the next band above it — the lowest one that is empty or only partly filled — is the conduction band. The whole question of electrical conductivity reduces to one thing: can an electron find an empty state right next to it to move into? An electron jammed into a completely full band is like a car in a bumper-to-bumper traffic jam with no gaps — it cannot go anywhere, because every neighbouring slot is taken. Only an electron with empty states nearby can accelerate under an applied voltage and carry current.

Three solids, one number: metal, semiconductor, insulator

Now the payoff. Whether a solid is a metal, a semiconductor, or an insulator comes down almost entirely to band filling and gap size. A metal has a valence band that is only partly filled — or one that overlaps the conduction band so there is effectively no gap at all. Either way there are empty states right next to the filled ones, so electrons flow at the gentlest push. That is exactly the metallic bond you met earlier, now sharpened: the 'sea of delocalized electrons' is band language for a partly filled band. Sodium's half-filled 3s band is the textbook case, and it explains a clue that puzzles beginners — a metal conducts *worse* as you heat it, because hotter atoms jostle more and trip up the already-free electrons.

An insulator sits at the opposite extreme: its valence band is completely full, its conduction band completely empty, and a *wide* gap — several electron-volts — yawns between them. The full band is the gridlocked traffic jam; to make any electron move you must lift it clear across the forbidden zone into the empty conduction band, and thermal energy at room temperature simply cannot pay that toll. Diamond, with a gap near 5.5 eV, is the classic example: full valence band, vast gap, no current, and (because that gap is larger than the energy of visible light) perfectly transparent. A semiconductor is the honest surprise of the whole subject: it is *not* a third kind of matter, just an insulator with a small gap, around 1 eV. Silicon (about 1.1 eV) has the very same diamond structure as diamond itself — the only difference is that one number.

                  METAL            SEMICONDUCTOR        INSULATOR
                 (e.g. Na)        (e.g. Si, ~1.1 eV)   (e.g. diamond, ~5.5 eV)

  conduction   [...empty....]    [....empty....]      [.....empty.....]
    band       [..PARTLY....]          ^                      ^
               |  FILLED:   |      ~1 eV gap            BIG gap (several eV)
               | no real gap|          v                      v
  valence      [..filled....]    [.....full....]      [......full.....]
    band

  conducts:    yes, freely      a little; MORE         no (at room T)
               (worse hot)      when heated

Watch what that small gap does. At absolute zero a semiconductor's valence band is full and it behaves as an insulator. Warm it to room temperature, though, and a few electrons borrow enough thermal energy to leap the narrow gap into the conduction band — and just as importantly, each departing electron leaves behind an empty slot, a hole, in the valence band. Now *both* bands are partly occupied: the lonely electrons up top can move, and electrons in the valence band can shuffle into the holes below, so a little current flows. This gives the fingerprint that separates a semiconductor from a metal: a semiconductor conducts *better* when heated, because heat manufactures more carriers, while a metal conducts worse. The band theory reading of conductivity is the whole story in one diagram.

Doping: tuning a semiconductor on purpose

Pure silicon, relying only on thermally-leaping electrons, is a feeble conductor — a trickle of carriers. The idea that transformed it into the foundation of every chip is doping: deliberately stirring in a trace of a different element, often just one foreign atom per million silicon atoms, to control the carriers on purpose. That whisper of impurity changes the conductivity by orders of magnitude. Recall that silicon — like its zinc-blende-structured cousins, the diamond-like semiconductors GaAs and GaN — has four valence electrons and bonds to four neighbours, using every electron exactly. Doping works by quietly breaking that perfect electron count.

1. n-type: replace an occasional silicon with phosphorus, which has five valence electrons. Four of them slot into the normal bonds; the fifth has nowhere to go, is only loosely held, and slips easily up into the conduction band. The dopant donates extra mobile electrons — negative carriers — so we call it n-type. In band terms, phosphorus adds filled donor levels just below the empty conduction band, so almost no energy is needed to free that fifth electron.
2. p-type: instead replace a silicon with boron, which has only three valence electrons. Now one bond is short an electron — a hole, an empty slot in the valence band. Electrons from neighbouring bonds can hop into the hole, which makes the hole itself appear to drift along like a positive carrier, so we call it p-type. In band terms, boron adds empty acceptor levels just above the full valence band, easy for valence electrons to step up into, leaving mobile holes behind.
3. Join them: press a p-type region against an n-type region and you get a p-n junction — a one-way valve for current, the diode. Stack three regions as n-p-n or p-n-p and you have a transistor, the switch-and-amplifier that, copied billions of times on one wafer, is the literal hardware of every computer and phone. Add light and a p-n junction becomes a solar cell or an LED.

The bridge to materials chemistry

Band theory is where the solids rung quietly hands you over to the materials chemistry that ends this whole ladder. Everything you built here — close packing, the unit cell, the structure types, the Born-Haber energy that holds a lattice together — was setting the stage; band filling and the gap are what turn an arrangement of atoms into something that does a job. The very same diagram that classifies copper, silicon and diamond is the starting point for engineered semiconductors, for solar cells and LEDs, and for the carefully tuned band gaps of designer materials.

The picture also bends in surprising ways once you shrink things or push them hard, which is the doorway to the final rung. Crush a crystal so small that it holds only a few thousand atoms — a quantum dot — and the band stops behaving like a continuous smear: with too few atoms, the levels no longer crowd together, the effective gap widens, and the colour the dot emits depends on its size, a vivid throwback to the discrete molecular orbitals you started from. Cool certain materials and the resistance vanishes entirely (superconductivity), something simple band theory does not predict. So end on the honest note that has run through every rung: band theory is a magnificent, load-bearing model — the backbone of modern electronics — but it is a model, and the frontier of materials chemistry lives partly in the places where it bends or breaks.