Skip to content

Chapter 3b — Solid State Physics Prerequisites

Open in Jupyter (browser)

"A solid is the densest, most boring, and most useful state of matter." — anonymous condensed matter lecturer

Chapter 3 took you from isolated atoms to bonded crystals: you learnt what a unit cell is, drew a few Bravais lattices, met reciprocal space, and computed a structure factor. Chapter 4 then taught you quantum mechanics in finite systems — a particle in a box, a harmonic oscillator, the many-electron problem and the exponential wall. What is glaringly absent is the bridge between the two: how does quantum mechanics behave in an infinite periodic potential? That bridge is solid state physics, and without it nothing in the rest of the book makes sense. You cannot read a DFT output, configure a k-point grid, interpret a band gap, fit a machine-learning potential to a phonon spectrum, or design a graph neural network that respects translational symmetry, unless you know the four or five ideas that this chapter installs.

A note on reading order. This chapter is labelled 3b because conceptually it expands Chapter 3 (descriptive crystallography) into its quantum counterpart, but it does use a small amount of Chapter 4 material — chiefly the time-independent Schrödinger equation, eigenvalues of Hermitian operators, and bra–ket notation. If you are following the linear path (Path A of the learning path), the recommended order is Ch 3 → Ch 4 → Ch 3.5 → Ch 5: this is the order that uses the least forward-referenced material. If you prefer the original "geometry first" ordering Ch 3 → Ch 3.5 → Ch 4, the bits of Ch 4 we lean on are brief enough that you can skim the relevant single-particle Schrödinger sections (4.2 and 4.3) and come back; nothing in this chapter requires the many-electron material from Chapter 4.5 onwards.

Why a working materials simulator needs this chapter

Every method in Tier 1 of this book — DFT (Ch 5, Ch 6), molecular dynamics (Ch 7), MLIPs (Ch 9), graph neural networks (Ch 10) — is shaped by one mathematical fact: the potential a particle sees inside a crystal is periodic. From that single observation flow:

  • Bloch's theorem, which says you only ever need to solve the Schrödinger equation inside one unit cell, with a k-dependent boundary condition. This is exactly the structure every plane-wave DFT code (VASP, Quantum ESPRESSO, ABINIT, CP2K) exploits — and the reason k-point convergence is the single most common parameter you will tune in Chapter 6.

  • Band structure, which tells you whether a material is a metal, a semiconductor, or an insulator. We will derive band structures in two complementary limits — nearly free electrons (weak potential) and tight binding (strong, localised potential). Both limits appear in real codes: plane-wave DFT is the nearly-free-electron extreme; localised-orbital DFT (SIESTA, FHI-aims, NWChem-LCAO) is the tight-binding extreme. MLIPs that learn from local environments — SchNet, NequIP, MACE — are effectively learning an implicit tight-binding model, as we shall see in Chapter 9.

  • Phonons, the quantised lattice vibrations. In Chapter 7 you will run molecular dynamics: atoms wobble around their equilibrium positions in a thermostatted simulation. The Fourier transform of that wobble is a phonon spectrum. In Chapter 8 the vibrational free energy — built from phonon densities of states — is what makes a phase stable at finite temperature. And in Chapter 9 you will benchmark MLIPs against DFT phonon dispersions, because matching phonons at the \(\Gamma\) point is the strictest test of an interatomic potential.

  • The free-electron and Sommerfeld models, which are conceptually the simplest band structure of all: \(E(\mathbf k) = \hbar^2 k^2/2m\), no lattice at all. This is the "jellium" that the local density approximation in DFT (Ch 5) takes as its reference. Every time you write xc=LDA in an input file you are betting that the local electron density looks, to leading order, like a uniform gas. Knowing why that bet works — and when it fails — is impossible without the Sommerfeld machinery.

  • Defects and doping, which are how real materials acquire their useful properties. In Chapter 6 you will compute defect formation energies; in the capstone project you will screen dopants for a target band gap. Both rest on the hydrogenic donor/acceptor picture developed here.

Roadmap

This chapter has seven content sections and an exercise set.

  1. Bloch's theorem (§3b.1). The cornerstone. We prove it from scratch starting from translational symmetry, derive the crystal momentum, define the first Brillouin zone, and introduce the band index. About 4000 words.

  2. Nearly free electrons (§3b.2). Weak periodic potential, empty lattice, degenerate perturbation theory at zone boundaries, gap opening, physical picture in terms of standing waves.

  3. Tight binding (§3b.3). The opposite limit. LCAO ansatz, 1D monatomic chain (\(E = -2t\cos ka\)), then graphene with its Dirac cones. Working Python code that diagonalises the graphene Hamiltonian along K–\(\Gamma\)–M–K.

  4. The free electron gas and Sommerfeld (§3b.4). 3D density of states, Fermi energy, finite-temperature corrections, electronic specific heat. Numerical evaluation for copper.

  5. Phonons (§3b.5). Classical lattice dynamics: 1D monatomic chain (\(\omega = 2\sqrt{K/m}|\sin(ka/2)|\)), then 1D diatomic chain with its acoustic and optical branches, then the 3D dynamical matrix.

  6. Specific heat: Einstein and Debye (§3b.6). Why Dulong–Petit fails, why Einstein's tweak is not enough, why Debye's \(T^3\) law works at low temperature. The Debye temperature is the single most useful number you will compute for a new material.

  7. Defects and band engineering (§3b.7). Point defects, shallow donors, the effective mass approximation, alloying and strain. The shortest section, but a forward reference to almost every applied chapter.

  8. Exercises (§3b.8). Eight problems, with solutions in admonitions, half analytical and half numerical.

What you need

From Chapter 0: linear algebra (eigenvalues of \(2\times2\) Hermitian matrices), Fourier series and Fourier transforms, the gradient and Laplacian, partial derivatives, complex exponentials. From Chapter 3: Bravais lattices, the reciprocal lattice, the first Brillouin zone, the structure factor. From Chapter 4: the time-independent Schrödinger equation, eigenvalues of Hermitian operators, bra-ket notation, the harmonic oscillator, and the fact that commuting Hermitian operators share an eigenbasis.

What you do not need

We will not quantise the phonon field in second-quantised language. We will not develop the full theory of the magnetic Brillouin zone, time-reversal symmetry, or topological invariants. We will not do anything with phonon–phonon scattering, polarons, superconductivity, or strong correlations. Those belong in a serious solid-state course; we want only the minimum that makes the rest of the handbook readable. Five pages of Ashcroft and Mermin, faithfully.

When you finish this chapter, you will be able to read a band-structure plot, recognise an acoustic branch, write a tight-binding Hamiltonian, and explain — at three in the morning, to a sceptical reviewer — why your DFT calculation needed an \(8\times 8\times 8\) Monkhorst–Pack mesh. Onwards.