Bloch’s theorem states that electronic wavefunctions in a periodic crystal can be written as a plane wave times a function with the same periodicity as the lattice. That statement turns an infinite crystal with \(\sim 10^{23}\) electrons into a problem on a single unit cell. Without it, plane-wave density functional theory (DFT) would not exist.
The Problem: An Infinite Number of Atoms
A real crystal has an astronomically large number of atoms. Solving the Schrödinger equation for that many particles is hopeless. Crystals save the day through periodicity: atoms repeat under lattice vectors \(\mathbf{R}\), so the potential satisfies
$$ V(\mathbf{r} + \mathbf{R}) = V(\mathbf{r}) $$for every \(\mathbf{R}\). The single-particle (or Kohn–Sham) Hamiltonian then commutes with every lattice translation operator \(T_{\mathbf{R}}\). Eigenstates can be chosen as simultaneous eigenstates of \(H\) and of the translations. Bloch’s theorem is that algebraic fact made explicit — and the reason electronic structure of solids is computable.
Statement of Bloch’s Theorem
Solutions of the Schrödinger equation for a periodic potential can be chosen as
$$ \psi_{\mathbf{k}}(\mathbf{r}) = e^{i \mathbf{k}\cdot\mathbf{r}}\, u_{\mathbf{k}}(\mathbf{r}), $$where \(u_{\mathbf{k}}(\mathbf{r}+\mathbf{R}) = u_{\mathbf{k}}(\mathbf{r})\) and \(\mathbf{k}\) is the crystal momentum. A wavefunction of this form is a Bloch wave. Equivalently, under a lattice translation the wavefunction only picks up a phase:
$$ \psi_{\mathbf{k}}(\mathbf{r} + \mathbf{R}) = e^{i \mathbf{k}\cdot\mathbf{R}}\, \psi_{\mathbf{k}}(\mathbf{r}). $$The density and all observables are unchanged — only a phase changes — so every unit cell looks the same.
flowchart LR A[Infinite crystal<br/>periodic V(r)] --> B[Bloch theorem] B --> C[Solve u_k(r)<br/>in one unit cell] B --> D[Sample k over<br/>Brillouin zone] C --> E[Plane-wave DFT] D --> E E --> F[Bands, DOS, total energy]
Derivation Intuition: Why the Bloch Form Must Hold
You do not need group theory to see why the Bloch form is forced. Define the lattice translation operator by \(T_{\mathbf{R}} f(\mathbf{r}) = f(\mathbf{r} + \mathbf{R})\). Because \(V(\mathbf{r} + \mathbf{R}) = V(\mathbf{r})\), this operator commutes with the Hamiltonian: \([H, T_{\mathbf{R}}] = 0\). Commuting operators share a common set of eigenfunctions, so energy eigenstates may be labeled by the eigenvalues of \(T_{\mathbf{R}}\).
Those eigenvalues cannot be arbitrary. Applying two successive translations gives \(T_{\mathbf{R}} T_{\mathbf{R}'} = T_{\mathbf{R}+\mathbf{R}'}\), so the eigenvalues \(\lambda(\mathbf{R})\) must form a representation of the additive group of lattice vectors: \(\lambda(\mathbf{R})\,\lambda(\mathbf{R}') = \lambda(\mathbf{R}+\mathbf{R}')\). The continuous solutions of that functional equation are pure phases, \(\lambda(\mathbf{R}) = e^{i \mathbf{k}\cdot\mathbf{R}}\), for some real wavevector \(\mathbf{k}\). Hence any simultaneous eigenfunction of \(H\) and the translations satisfies
$$ \psi(\mathbf{r} + \mathbf{R}) = e^{i \mathbf{k}\cdot\mathbf{R}}\, \psi(\mathbf{r}). $$Define \(u(\mathbf{r}) \equiv e^{-i \mathbf{k}\cdot\mathbf{r}}\,\psi(\mathbf{r})\). Substituting the phase relation immediately shows \(u(\mathbf{r}+\mathbf{R}) = u(\mathbf{r})\), so \(\psi\) factors as a plane wave times a lattice-periodic function — the Bloch form. The label \(\mathbf{k}\) is not put in by hand; it is the quantum number that classifies how the wavefunction transforms under crystal translations.
Why periodic boundary conditions work. Born–von Karman conditions on a large crystal of \(N_1 \times N_2 \times N_3\) cells require \(\psi(\mathbf{r} + N_i \mathbf{a}_i) = \psi(\mathbf{r})\), which discretizes \(\mathbf{k}\) into a fine mesh inside the Brillouin zone. As \(N_i \to \infty\) that mesh becomes continuous and recovers every Bloch state of the infinite crystal. Periodic BCs are the finite-size expression of the same translational symmetry — not an ad hoc numerical trick. On one unit cell one imposes periodicity on \(u_{\mathbf{k}}\) while the plane-wave factor supplies the crystal-momentum phase: exactly the boundary conditions used in plane-wave DFT.
Decoding the Bloch Wave
The Bloch form has two parts, and understanding each is the key to the whole theorem.
- The plane-wave envelope \(e^{i \mathbf{k}\cdot\mathbf{r}}\) is a smooth, extended wave that carries the electron across the crystal. It encodes the long-range, delocalized character of an electron in a solid.
- The periodic part \(u_{\mathbf{k}}(\mathbf{r})\) carries the atomic-scale detail — the way the wavefunction wiggles and concentrates near each nucleus within a single unit cell.
An electron in a crystal is neither free nor bound to one atom: it is a modulated wave that propagates through the lattice while retaining local atomic structure. That is why electrons travel through a perfect crystal without scattering. Scattering requires a perturbation that breaks translational symmetry (phonons, impurities, surfaces). In a perfect lattice, Bloch states are exact eigenstates and therefore stationary.
Crystal Momentum vs True Momentum
The label \(\mathbf{k}\) is called the crystal momentum, but it is not ordinary momentum \(\mathbf{p} = -i\hbar\nabla\). Ordinary momentum is conserved only when the potential is completely translation-invariant, i.e. free space. In a crystal the potential is only discrete-translation invariant, so the conserved quantity is the crystal momentum \(\hbar\mathbf{k}\), defined only up to a reciprocal lattice vector \(\mathbf{G}\).
Concretely, the expectation value of true momentum in a Bloch state is
$$ \langle \psi_{\mathbf{k}} | \mathbf{p} | \psi_{\mathbf{k}} \rangle = \hbar\mathbf{k} + \langle u_{\mathbf{k}} | \mathbf{p} | u_{\mathbf{k}} \rangle, $$which need not equal \(\hbar\mathbf{k}\). Under Bragg scattering from the lattice, an electron can absorb or emit a reciprocal lattice vector of momentum \(\hbar\mathbf{G}\) while remaining in an equivalent Bloch state. Crystal momentum is therefore conserved only modulo \(\hbar\mathbf{G}\). That is the microscopic origin of umklapp processes in transport and of the zone-boundary gaps that appear when a weak periodic potential is turned on.
Adding any reciprocal lattice vector to \(\mathbf{k}\) gives a physically identical state:
$$ \psi_{\mathbf{k}+\mathbf{G}}(\mathbf{r}) \;\text{is equivalent to}\; \psi_{\mathbf{k}}(\mathbf{r}). $$The reciprocal lattice is the Fourier-space counterpart of the real-space crystal lattice, defined so that \(e^{i \mathbf{G}\cdot\mathbf{R}} = 1\) for all lattice vectors \(\mathbf{R}\). Because of this redundancy, we only ever need to consider \(\mathbf{k}\) values inside a single primitive cell of the reciprocal lattice — the first Brillouin zone. Every distinct electronic state can be labeled by a \(\mathbf{k}\) inside this zone and a band index \(n\) that counts the discrete solutions at each \(\mathbf{k}\).
| Concept | Real space | Reciprocal space |
|---|---|---|
| Fundamental vectors | Lattice vectors \(\mathbf{R}\) | Reciprocal vectors \(\mathbf{G}\) |
| Repeating cell | Unit cell | Brillouin zone |
| Periodic quantity | Potential \(V(\mathbf{r})\) | Bloch state label \(\mathbf{k}\) |
| Conserved quantity | — | Crystal momentum \(\hbar\mathbf{k}\) (mod \(\hbar\mathbf{G}\)) |
| Role | Where atoms sit | Where \(\mathbf{k}\)-sampling happens |
The Brillouin Zone
The first Brillouin zone (BZ) is the Wigner–Seitz cell of the reciprocal lattice: the set of points closer to \(\mathbf{G}=\mathbf{0}\) than to any other reciprocal lattice vector. It is the natural domain for \(\mathbf{k}\) because every Bloch state outside the zone is identical to one inside, after a suitable redefinition of the periodic part \(u_{\mathbf{k}}\).
High-symmetry points (\(\Gamma\), \(X\), \(L\), \(K\), …) depend on the Bravais lattice. Band plots trace \(E_n(\mathbf{k})\) along paths connecting them because gap and effective-mass extrema almost always sit on symmetry lines. Total energy, charge density, and density of states require integrating over the full zone, replaced in practice by a discrete k-point mesh (e.g. Monkhorst–Pack). Metals need denser meshes than large-gap insulators because the Fermi surface cuts through the zone.
BZ volume is \((2\pi)^3 / \Omega\), with \(\Omega\) the real-space cell volume. Larger cells produce smaller zones — the geometry behind band folding, and why a supercell calculation samples a shrunk zone for the same physical material.
From Infinite Crystal to a Single Unit Cell
Bloch’s theorem collapses the infinite crystal into two finite tasks:
- Solve within one unit cell. Because \(u_{\mathbf{k}}\) is periodic, the eigenvalue problem for each fixed \(\mathbf{k}\) lives on a single cell with periodic boundary conditions. Substituting the Bloch form into the Schrödinger (or Kohn–Sham) equation yields an equivalent problem for \(u_{\mathbf{k}}\) on that cell, with \(\mathbf{k}\) a parameter in the kinetic operator.
- Sample over \(\mathbf{k}\). Physical quantities — total energy, charge density, density of states — are Brillouin-zone integrals, approximated by a finite k-point grid.
Cost scales with atoms in the unit cell and with mesh density — not with sample size. Choosing the mesh well is a core convergence parameter in any DFT study. Supercell calculations of defects or surfaces use the same machinery: the cell is enlarged, the BZ shrinks, and the defect is formally repeated — an approximation controlled by supercell size.
Band Structure, Band Folding, and the Free-Electron Limit
At each \(\mathbf{k}\), eigenvalues form a discrete ladder labeled by band index \(n\). Plotting \(E_n(\mathbf{k})\) along high-symmetry paths yields the electronic band structure:
- The band gap (or its absence) classifies metals, semiconductors, and insulators.
- Same-\(\mathbf{k}\) vs different-\(\mathbf{k}\) extrema distinguish direct from indirect gaps.
- Band curvature sets effective masses and carrier mobilities.
All of these are direct consequences of energy as a function of crystal momentum \(\mathbf{k}\).
Band folding. Enlarge the real-space cell and the BZ shrinks. Doubling a lattice constant halves the zone and folds old zone-boundary states back to \(\Gamma\). A two-atom conventional cell of a monatomic crystal therefore shows twice as many bands in half the zone — identical physics, redrawn. That is why supercell dispersions look crowded and why primitive cells are preferred for clean band plots. It is also why zone-boundary phonons of a primitive cell can appear at \(\Gamma\) in a doubled cell.
Free-electron limit. If \(V=0\), eigenstates are pure plane waves with \(E = \hbar^2 |\mathbf{k}|^2 / 2m\). In the empty-lattice construction those parabolas are redrawn in the reduced-zone scheme of a chosen Bravais lattice: each time a free-electron sphere crosses a Bragg plane, the branch folds into the first BZ. A weak periodic potential opens gaps at those boundaries by hybridizing plane waves that differ by \(\mathbf{G}\). Nearly-free-electron metals sit near this limit (\(u_{\mathbf{k}}\) almost constant); tight-binding materials sit at the other extreme (\(u_{\mathbf{k}}\) peaked on atoms). The Bloch form covers both.
Plane-Wave Expansion, DFT, and Quantum ESPRESSO
The connection to modern computation is direct. Because the periodic part \(u_{\mathbf{k}}(\mathbf{r})\) is periodic, it can be expanded exactly as a Fourier series over reciprocal lattice vectors:
$$ u_{\mathbf{k}}(\mathbf{r}) = \sum_{\mathbf{G}} c_{\mathbf{k},\mathbf{G}}\, e^{i \mathbf{G}\cdot\mathbf{r}}, $$and therefore the full Bloch wave is a single sum of plane waves labeled by \(\mathbf{k}+\mathbf{G}\):
$$ \psi_{\mathbf{k}}(\mathbf{r}) = \sum_{\mathbf{G}} c_{\mathbf{k},\mathbf{G}}\, e^{i (\mathbf{k}+\mathbf{G})\cdot\mathbf{r}}. $$This is the mathematical heart of plane-wave DFT. The unknowns become the coefficients \(c_{\mathbf{k},\mathbf{G}}\). Kinetic energy is diagonal in this basis; a local potential couples \(\mathbf{G}\) to \(\mathbf{G}'\) through its Fourier components \(V(\mathbf{G}-\mathbf{G}')\). The size of the basis is controlled by a single, systematically improvable parameter: the plane-wave cutoff energy \(E_{\mathrm{cut}}\), which retains all \(\mathbf{G}\) with \(\hbar^2 |\mathbf{k}+\mathbf{G}|^2 / 2m \le E_{\mathrm{cut}}\). Increasing the cutoff monotonically improves accuracy, a property that makes plane-wave methods exceptionally robust and easy to converge.
In Kohn–Sham DFT one solves, at each mesh \(\mathbf{k}\), a self-consistent eigenvalue problem for the occupied Bloch orbitals, builds the density from \(|\psi_{n\mathbf{k}}|^2\), updates the effective potential, and iterates. Periodic cell, k-grid, plane-wave basis, SCF — Bloch’s theorem as an algorithm.
This is the approach of Quantum ESPRESSO: ecutwfc, K_POINTS, and the lattice vectors are the practical knobs on the objects above. Plane-wave codes represent an infinite crystal with a finite basis on a finite k-grid for exactly this reason. For how these methods run at scale, see our computational engines page and related explainers on the blog.
Run it on Simatra
Simatra runs plane-wave DFT — the direct computational embodiment of Bloch’s theorem — on GPU-accelerated clusters engineered for materials modeling. On our GPU-Opt-V2 instances you get up to 5x faster convergence across k-point grids and plane-wave cutoffs, and you can model supercells of up to ~2,000 atoms for defects, interfaces, and disordered systems where strict periodicity is broken locally. Choose the open-source Quantum ESPRESSO engine or Simatra’s native KRONOS engine, a C++20, GPL-3.0 DFT code with CUDA/HIP/Metal backends, detailed on our computational engines page. Begin a free trial with $100 in credits at app.simatra.io and compute your first band structure today.