4. Solid-State Physics: From Single Atoms to Useful Materials

Now we have all the pieces — fields, currents, quantum mechanics — to understand how silicon, the most important industrial material in the world, actually works.

4.1 The lattice: atoms arranged regularly

Most of the materials we care about in electronics are crystalline: the atoms sit in a regular, repeating, three-dimensional pattern called a lattice. Common arrangements:

• Simple cubic. Atoms at the corners of a cube. Rare in nature.
• Body-centered cubic (BCC). Corners plus one in the middle. Iron at room temperature.
• Face-centered cubic (FCC). Corners plus one on the center of each face. Copper, aluminum, gold, silver — most familiar metals.
• Diamond cubic. Two interpenetrating FCC lattices. Diamond, silicon, germanium.

The spacing between adjacent atoms in the lattice is called the lattice constant. Silicon's lattice constant is 5.43 Å (5.43 × 10⁻¹⁰ m, or 0.543 nm) — almost the same as the wavelength of an electron in silicon, which is no coincidence and is why quantum mechanics matters here.

Apartment-building analogy. Imagine a vast apartment block where every unit is identical and arranged in a perfect 3D grid. The grid is the lattice; each unit is an atom; the spacing between units is the lattice constant. Now, the electrons are not the apartments themselves — they are the residents who can wander. Some residents are tightly bound to one apartment (core electrons of each atom). Others can drift between neighboring apartments (valence electrons). And in a metal, some can wander the entire building (conduction electrons).

Defects in the lattice are like missing apartments, weirdly oriented apartments, or extra apartments wedged in (vacancies, dislocations, interstitials). They affect how electrons move through the material — and, as we will see, deliberately introduced defects in the form of dopants are how we make semiconductors useful.

4.2 Bands: what happens when many atoms get together

We met this idea briefly in section 3.3: in a single atom, electrons occupy discrete energy levels. In two atoms close together, each level splits into two. In a crystalline solid containing $10^{23}$ atoms, the levels smear into continuous bands of allowed energies.

The two most important bands for understanding electronics:

• Valence band. The highest band that is fully occupied (or nearly so) at zero temperature. Electrons here are "stuck" — there are no nearby empty states for them to jump into, so they cannot easily move under an applied voltage.
• Conduction band. The next band up, normally mostly empty. Electrons in the conduction band have plenty of empty neighboring states, so they roam freely under a voltage.

Between the two there is usually a band gap — a forbidden range of energies where no electron states exist at all. Electrons cannot live in the gap.

Parking-garage analogy — the canonical one. Picture a giant multi-story parking garage. The ground floor is full of parked cars, jammed bumper-to-bumper, and they cannot move because every spot is taken — that is the valence band. The roof deck is wide open, with plenty of room to drive around — that is the conduction band. Between them is a sealed wall of variable height — the band gap.

• Conductors (metals). No wall: ground floor and rooftop overlap. Cars on the ground can drift up and start driving anywhere. Many free, mobile electrons. High conductivity.
• Insulators (glass, ceramics). Tall thick wall (band gap of 5+ eV). Practically no cars ever make it up to the rooftop, even at room temperature. Almost no free electrons. Almost no conductivity.
• Semiconductors (silicon, germanium). A short wall (about 1 eV in silicon). At absolute zero, no cars on the rooftop. At room temperature, thermal energy gives a few cars enough kick to leap over the wall. A small but nonzero population of conduction-band electrons. Modest conductivity, between metals and insulators. And — this is the important part — that conductivity is enormously sensitive to small perturbations. That sensitivity is what makes them useful.

The exact size of the band gap, and how electrons populate the bands, is what determines whether a material is a metal, insulator, or semiconductor. Silicon's gap is 1.12 eV at room temperature. Germanium's is 0.67 eV — narrower, so germanium has more thermal carriers at room temperature, which is one reason germanium transistors had so much "leakage" and were unstable in heat. Diamond's gap is about 5.5 eV — a true insulator. GaN, the semiconductor inside modern fast USB-C chargers, has a 3.4 eV gap — wide enough to handle hundreds of volts before breaking down.

4.3 Electrons and holes

At room temperature in pure silicon, a small fraction of valence-band electrons get thermally kicked up to the conduction band, where they can carry current. But notice: when an electron leaves the valence band, it leaves behind an empty seat. That empty seat — a missing electron in an otherwise full band — behaves remarkably like a positive mobile charge. We give it a name: a hole.

This is not just bookkeeping. Holes really do behave like positive carriers. Apply a voltage to a piece of pure silicon and you get both a current of electrons (drifting in the conduction band toward the positive terminal) and a current of holes (drifting in the valence band, in the opposite direction, also adding to the current). For pure silicon at room temperature, the populations are equal: $n = p = n_i \approx 1.5 \times 10^{10}$ electrons (or holes) per cubic centimeter. That density gives silicon a tiny but measurable conductivity even when undoped — about $4 \times 10^{-6}$ S/m, ten orders of magnitude less than copper's. Pure silicon is almost an insulator.

Bus-station analogy for holes. Imagine a long row of bus seats, all occupied. A passenger gets up and walks to the back. Now there is one empty seat — but the passengers behind take turns shuffling forward into it. From a god's-eye view, you see a kind of bubble of emptiness traveling backward through the row, even though no individual passenger took a long trip. The "empty seat" effectively moves — and if the seats are charged, that bubble of emptiness is itself effectively charged. That is what a hole is.

4.4 Doping: the magic that makes electronics possible

Here is the move that built the entire electronics industry. Take pure silicon and replace one atom in a million with an atom of phosphorus. Phosphorus has five valence electrons, one more than silicon. Four of those go into bonds with neighboring silicon atoms; the fifth is loose, easily kicked into the conduction band. Now you have a semiconductor with a population of free electrons proportional to the dopant concentration — typically a million times higher than intrinsic silicon. We call this n-type silicon ("n" for "negative" carriers, i.e., electrons).

Replace a silicon atom with a boron atom — boron has only three valence electrons. It is missing one to complete the bonding. The missing electron is essentially a hole, ready to move. Now you have p-type silicon, with mobile holes as the majority carriers.

Doping levels are absurdly low. A "heavy" doping is around $10^{19}$ atoms per cubic centimeter — roughly one part in a thousand. A "light" doping is $10^{15}$ — one part per million. This is why fabs need cleanrooms with parts-per-trillion contamination control. A few unwanted iron atoms in the wrong spot of a wafer kills the chip.

Choir analogy for doping. Imagine a thousand-voice choir all singing in unison — no single voice stands out. That is intrinsic silicon: pure, but quiet, with all electrons in the valence band singing along. Now sneak in two tenors who sing slightly above the chorus pitch (donor atoms in n-type doping). You can now hear the tenors floating above the unison; they are mobile in a way the chorus is not. Sneak in a couple of bass-stoppers who skip notes (acceptor atoms in p-type), and you create silences in the choir that propagate around as the singers shift in and out — those silences are the holes.

Doping is selective. Modern fabs use ion implantation to fire dopant ions at exactly chosen energies into precisely defined regions of the wafer, building up the n-type and p-type regions of every transistor. We will see how this builds diodes and transistors in Chapter 1.

4.5 The Fermi level: the water-line of energy

Here is a useful concept that ties it all together. In a piece of material in thermal equilibrium, electrons occupy the available energy states according to a probability distribution called the Fermi-Dirac distribution:

$f(E) = \frac{1}{1 + e^{(E - E_F)/kT}}.$

The number $E_F$ is the Fermi level. At absolute zero, all states below $E_F$ are filled (probability 1) and all above are empty (probability 0). At room temperature, the transition is smeared over a range of about $kT \approx 0.026$ eV — small compared to the band gap.

Water-tank analogy. Picture the available electron states as a series of horizontal shelves in a tank. The Fermi level is the water level. Below the line, shelves are full; above it, empty (with a fuzzy boundary at room temperature).

In intrinsic silicon, $E_F$ sits in the middle of the band gap. In n-type silicon, the extra electrons push $E_F$ up — close to the conduction band. In p-type silicon, the missing electrons (holes) drag $E_F$ down — close to the valence band.

When you bring two doped regions into contact (n-type and p-type, say, to form a diode), the Fermi levels equalize. The conduction and valence bands have to bend across the junction to make this happen. That bending creates an internal electric field across the junction — the built-in potential — which is the heart of how a diode works. More on this in Chapter 1.

4.6 Mobility, drift, diffusion: how carriers actually move

Apply an electric field $E$ to a doped semiconductor. The mobile carriers (electrons in n-type, holes in p-type) feel the force $qE$ and accelerate. But they do not run free for long: after about a picosecond, on average, they collide with a vibrating atom (a "phonon") or with an impurity, lose their momentum, and start over. The average drift velocity is

$v_d = \mu E,$

where $\mu$ is the mobility — a measure of how slippery the material is for that carrier.

Note two key facts. First, electrons are about three times more mobile than holes in silicon. This asymmetry is why n-channel transistors are intrinsically faster than p-channel transistors, and why CMOS digital circuits typically size their PMOS transistors 2–3× larger than their NMOS partners — to compensate. Second, GaAs has 6× the electron mobility of silicon, which is why GaAs has historically been used for the very-fast circuits in radio receivers and amplifiers.

Carriers can also move by diffusion — flowing from regions of high concentration to low, like ink dispersing in water. Diffusion does not need an electric field; it is driven by random thermal motion and statistical drift toward uniformity. The diffusion current density is

$J_n^{diff} = qD_n \frac{dn}{dx},$

where $D_n$ is the diffusion coefficient.

Drift vs diffusion. Think of drift as wind blowing a leaf along — you need a force. Think of diffusion as molecules of perfume spreading from a bottle: even with no wind, statistical motion alone causes the perfume to fill the room. Both effects are happening in every semiconductor device. A working transistor relies on both: in the channel of a MOSFET, drift dominates; across the depletion region of a diode, drift; through the base of a BJT, diffusion. The Einstein relation $D/\mu = kT/q$ links the two coefficients — they come from the same thermal motion underneath.