Nearly Free Electron Model and Band Gaps

Introduction

The free-electron model produces a smooth parabola E = ħ²k²/(2m), but real crystals can have forbidden energy intervals. The nearly free electron model keeps mobile wave-like electrons and adds a weak periodic potential from the ions. The largest qualitative change occurs where two free-electron waves have equal energy and the lattice can scatter one into the other: a Brillouin-zone boundary. Their mixing forms two standing-wave patterns with different interaction energies, splitting the original crossing into a gap.

Core explanation

In a one-dimensional lattice of period a, the smallest nonzero reciprocal vector is G = 2π/a. At the first-zone edge k = π/a, the free-electron states at +π/a and −π/a have the same kinetic energy because E depends on k². The periodic potential can transfer reciprocal vector G, connecting them: k−G = −π/a. Away from the edge their energies differ, so a weak perturbation has a comparatively small effect; at exact degeneracy, mixing can be strong even when the periodic potential is weak.

Write the lattice potential as a Fourier series with coefficient V G at reciprocal vector G. In the two-state basis of the degenerate travelling waves, the matrix at the boundary has equal diagonal energies E₀ and off-diagonal entries V G and its complex conjugate. Its eigenenergies are E₀ ± V G , so the local gap is 2 V G in this simple convention. If a source defines the real-space potential as 2V cos(Gx), then the Fourier coefficients at ±G are V and the gap is 2 V . If the same source instead calls the full cosine amplitude V, its off-diagonal coefficient is V/2 and the gap is V . Stating the Fourier convention avoids a factor-of-two error. A MIT materials problem and solution shows the two-state matrix and its zone-boundary split.

The two combinations can be pictured as standing waves, proportional in a convenient phase choice to cos(πx/a) and sin(πx/a). One concentrates electron density near a set of positions in the periodic potential; the other concentrates it between them. Their potential energies differ, creating the gap. Which combination is lower depends on where the potential minima are placed. It is safer to say they differ in overlap with the ionic potential than to assert that a particular sine function is always the lower state without specifying the origin and potential sign.

In an extended-zone plot, the two free parabolas would cross at the relevant Bragg plane. The periodic potential creates an avoided crossing . In a reduced-zone plot, branches are folded inside the first zone and separated at the boundary. A gap between two branches at a particular k is a direct local gap. The material's fundamental band gap is the minimum energy difference between its highest occupied valence state and lowest empty conduction state, which may occur at different k values and need not equal this simple boundary gap. Filling matters too: a gap at a zone edge cannot make a material insulating if another partially occupied band crosses the Fermi level.

The model's weak-potential assumption is most plausible for simple metals with relatively delocalised electrons. Strongly directional covalent bonding, narrow d bands, multiple atoms per cell and electron correlation require more detailed orbital or many-body treatments. Even then, its core lesson persists: periodicity scatters waves coherently, and near degeneracies that scattering can rearrange electronic energies substantially. The MIT lecture on the nearly-free-electron model connects the diffraction picture with gap formation and extended versus reduced zone plots.

Step-by-step reasoning

1. Identify the real-space period a and reciprocal vector G = 2π/a. 2. Locate k values where E free(k) = E free(k−G). 3. Write a two-state coupling matrix with V G off diagonal. 4. Diagonalise it to find E₀ ± V G at exact degeneracy. 5. Compare the resulting local gap with electron filling and other bands before classifying the solid.

Visual explanation

Draw two free-electron parabolas after folding into one zone. They meet at k = π/a. Replace the meeting point by two curves bending away from one another, with a vertical bracket labelled 2 V G . Beneath them sketch two standing waves over a row of ions: one peaks on sites, the other between sites, making their different potential energies visible.

Real-world analogy

Two runners moving in opposite directions can have the same speed but different interactions with a repeating obstacle course. A periodic array of gates couples their routes at a special rhythm. The resulting standing patterns experience the gates differently and split into two characteristic energy costs. The analogy captures degeneracy and periodic coupling, though electrons are quantum waves rather than runners.

Real-world example

Band calculations for metals and semiconductors show avoided crossings where orbital states of compatible symmetry mix. The nearly-free-electron model gives a clean prototype: near a Bragg plane the dispersion bends and a local gap can appear. Interpreting a real semiconductor also requires locating the actual valence maximum and conduction minimum and checking whether they share k.

Why?

Why does a weak potential have its greatest effect at a zone boundary? At a degeneracy, two states have no free-electron energy penalty for mixing. Their coupling therefore splits them to first order in V G ; far from degeneracy, the energy mismatch suppresses mixing and the shift is usually smaller.

Common misconception

"Any zone-boundary gap proves the solid is an insulator." A solid is insulating in a simple band picture only if its occupied bands are full and no other band crosses the chemical potential. A local avoided crossing may coexist with a metallic Fermi surface elsewhere.

Worked example

Question: In a one-dimensional nearly-free-electron model, the Fourier coefficient coupling the two degenerate waves at a zone boundary has magnitude V G = 0.15 eV. What are their energies relative to the original crossing E₀, and what is the local gap?

Reasoning: At exact degeneracy the two-state matrix has diagonal entries E₀ and off-diagonal coupling V G. Its eigenvalues are E₀ − V G and E₀ + V G . Each is shifted by 0.15 eV from the uncoupled crossing, in opposite directions. Subtracting the lower from the upper gives 0.30 eV. The result assumes other bands and Fourier components can be neglected near this crossing.

Answer: E₀ − 0.15 eV and E₀ + 0.15 eV, separated by 0.30 eV.

Quick check

1. Why can k = π/a and k−G = −π/a mix especially strongly when G = 2π/a? Answer: They have equal free-electron energies, and the periodic potential has a Fourier component that connects wavevectors differing by G.

Exam focus

Draw an avoided crossing and derive the 2 V G result with the Fourier convention stated. Distinguish a local gap at one k from the material's fundamental gap. Include band filling before predicting metallic or insulating behaviour.

Advanced insight

More generally, a periodic potential couples an infinite ladder of waves k+mG. The two-state approximation works when just one pair is nearly degenerate and all other coupled states are energetically distant. Symmetry can prevent mixing between some crossing states, allowing protected crossings rather than gaps. The simple model therefore illustrates the mechanism, not an unconditional rule that every line crossing must split.

Summary

At a Brillouin-zone boundary, free-electron waves related by a reciprocal vector can be degenerate. A weak periodic potential mixes them into standing-wave combinations and splits their energies. In a two-state approximation the local gap is 2 V G , but the whole band structure and electron filling determine actual conductivity.

Practice questions

1. What reciprocal vector connects +π/a and −π/a? Answer: G = 2π/a. 2. What is the zone-boundary energy split when the off-diagonal Fourier coefficient is V G? Answer: 2 V G in the stated two-state convention. 3. Why is an avoided crossing stronger near degeneracy? Answer: The uncoupled states have little or no energy mismatch, so coupling mixes them efficiently. 4. Does a local zone-boundary gap guarantee an insulating solid? Answer: No; other partially filled bands can still cross the chemical potential.