The Fermi Energy

Definition

Electrons are fermions: they obey the Pauli exclusion principle, so no two electrons can occupy the same single-particle quantum state (a state is fixed by its wavevector k and its spin, ↑ or ↓). Take the conduction electrons of a metal and cool them to absolute zero. They cannot all collapse into the lowest-energy state — exclusion forbids it. Instead they fill states from the bottom up, one pair (spin up + spin down) per orbital, occupying every level until they run out of electrons.

The energy of the last state filled is the Fermi energy:

E_F = (ℏ²/2m)(3π²n)^(2/3)

where:

• ℏ = 1.055 × 10⁻³⁴ J·s — reduced Planck constant (h/2π)
• m = 9.109 × 10⁻³¹ kg — free-electron mass (or the effective mass m* in a band)
• n = N/V — number of conduction electrons per unit volume (m⁻³)

Notice what is not in this equation: temperature. The Fermi energy is a purely quantum-mechanical, zero-temperature quantity. It is set entirely by how densely the electrons are packed. This picture — a gas of non-interacting electrons obeying Fermi statistics — is the Sommerfeld free-electron model (Arnold Sommerfeld, 1927), which rescued Paul Drude's 1900 kinetic theory of metals from its catastrophic overestimate of the electronic heat capacity.

The Fermi sphere and Fermi surface

It is cleanest to think in momentum space (k-space). Each allowed state is a point on a fine grid, and the energy of a free electron is E = ℏ²k²/2m, which depends only on the magnitude of k. So filling states from lowest energy upward means filling a solid sphere in k-space. At T = 0 that sphere is completely full inside and completely empty outside. Its radius is the Fermi wavevector:

k_F = (3π²n)^(1/3)      E_F = ℏ²k_F²/2m      v_F = ℏk_F/m

The Fermi surface is the boundary of this filled region — the sphere of radius k_F. For a real crystal, band structure warps this sphere into intricate shapes (necks, pockets, sheets), and mapping those shapes is a central goal of condensed-matter physics because the Fermi surface controls nearly every electronic property. But the free-electron sphere is the essential starting point.

The Fermi velocity v_F is the speed of the fastest electrons at T = 0. For copper v_F ≈ 1.6 × 10⁶ m/s — about 0.5% of the speed of light — even though the metal is stone cold. This is a striking, purely quantum result: the electrons are moving fast not because they are hot, but because exclusion has forced them into high-momentum states.

How the electron gas fills up — step by step

1. Start with one electron. It drops to the lowest state, k = 0, energy E = 0.
2. Add electrons. Each new pair (↑ and ↓) must go into the next unoccupied orbital, i.e. a slightly larger |k|, because the lower ones are taken. Energy climbs as k².
3. Keep going to N electrons. The occupied states fill a sphere of radius k_F in k-space. Counting states — two spins per k-cell of volume (2π)³/V — gives N = V·k_F³/3π², which inverts to k_F = (3π²n)^(1/3).
4. Read off the top. The highest occupied energy is E_F = ℏ²k_F²/2m = (ℏ²/2m)(3π²n)^(2/3).
5. Warm it slightly. At finite T the sharp step in occupancy smears over an energy width ~k_B·T. Only electrons in that thin shell near E_F can be excited — the rest are locked in place by exclusion.

The Fermi–Dirac distribution

At temperature T the probability that a state of energy E is occupied is the Fermi–Dirac distribution:

f(E) = 1 / (exp[(E − μ)/k_B·T] + 1)

where μ is the chemical potential (the Fermi level). At T = 0 this is a perfect step: f = 1 below E_F and f = 0 above it. As T rises, the step softens over a window of width ~k_B·T centered on μ. Because room temperature gives k_B·T ≈ 0.026 eV while E_F ≈ 7 eV, the step is only very slightly rounded — the electron gas is strongly degenerate, behaving almost as if it were at absolute zero.

Why only electrons near E_F matter

This is the single most important consequence of the Fermi energy. Consider what a classical physicist expected: every one of the ~10²³ conduction electrons per mole should carry (3/2)k_B of thermal energy, giving a huge electronic heat capacity. Experiment showed the electronic contribution is roughly a hundred times smaller. The resolution is Pauli exclusion.

An electron deep inside the Fermi sea cannot absorb a small amount of thermal energy k_B·T, because every nearby state it might jump to is already occupied. Only electrons within about k_B·T of E_F have empty states available just above them. The fraction of electrons that can respond is therefore roughly:

fraction ~ k_B·T / E_F = T / T_F ≈ 300 / 81,000 ≈ 0.4%   (copper, room T)

That small active fraction explains a whole family of observations at once:

• Electronic heat capacity is linear in T: C_el = γT with γ = (π²/3)g(E_F)k_B², not the classical constant (3/2)Nk_B.
• Electrical and thermal conduction are carried only by electrons in that shell — the transport current is a slight imbalance of the Fermi sphere in k-space.
• Pauli paramagnetism: only the near-E_F electrons can flip spin in a magnetic field, so the metallic spin susceptibility is small and nearly temperature-independent, χ ∝ g(E_F).
• Screening and plasmons are set by the response of the Fermi surface, with a Thomas–Fermi screening length that shrinks as g(E_F) grows.

Density of states at the Fermi energy

The density of states g(E) counts how many electron states lie in an energy interval dE per unit volume. For a 3D free-electron gas it rises as the square root of energy:

g(E) = (1/2π²)(2m/ℏ²)^(3/2) · √E        (per unit volume, both spins)

Evaluated at the top of the filled sea this gives the compact relation g(E_F) = 3n/(2E_F). This one number governs the low-temperature behavior of the metal: the linear heat-capacity coefficient γ, the Pauli susceptibility, and the screening length all scale with g(E_F). A high g(E_F) is also the precondition for itinerant magnetism (the Stoner criterion U·g(E_F) > 1) and for BCS superconductivity, whose gap depends exponentially on g(E_F). The dimensionality matters: g(E) ∝ √E in 3D, is constant in 2D, and diverges as 1/√E at band edges in 1D.

Fermi energies of real metals

Metal	n (10²⁸ m⁻³)	E_F (eV)	T_F (10⁴ K)	v_F (10⁶ m/s)
Sodium (Na)	2.65	3.24	3.77	1.07
Potassium (K)	1.40	2.12	2.46	0.86
Copper (Cu)	8.47	7.00	8.16	1.57
Silver (Ag)	5.86	5.49	6.38	1.39
Gold (Au)	5.90	5.53	6.42	1.40
Aluminium (Al)	18.1	11.7	13.6	2.03
Lithium (Li)	4.70	4.74	5.51	1.29
Zinc (Zn)	13.2	9.47	11.0	1.83

Every Fermi temperature in that last column is of order 10⁴–10⁵ K — thousands of times room temperature and above the melting point of the metal. That is the quantitative meaning of "the electron gas is degenerate."

Worked example — copper

Copper has one conduction electron per atom, mass density 8960 kg/m³ and atomic mass 63.5 u, giving an electron density n = 8.47 × 10²⁸ m⁻³. Then:

k_F = (3π²n)^(1/3)
    = (3 · 9.870 · 8.47×10²⁸)^(1/3)
    = 1.36 × 10¹⁰ m⁻¹

E_F = ℏ²k_F² / 2m
    = (1.055×10⁻³⁴)² · (1.36×10¹⁰)² / (2 · 9.109×10⁻³¹)
    = 1.13 × 10⁻¹⁸ J
    = 7.0 eV

T_F = E_F / k_B = 1.13×10⁻¹⁸ / 1.381×10⁻²³ = 8.2 × 10⁴ K

v_F = ℏk_F / m = 1.055×10⁻³⁴ · 1.36×10¹⁰ / 9.109×10⁻³¹ = 1.57 × 10⁶ m/s

So copper's least-bound conduction electrons already carry 7 eV of kinetic energy and move at 1.6 million metres per second at absolute zero, without any heating whatsoever. The degeneracy pressure of this electron gas, P = (2/5)n·E_F ≈ 3.8 × 10¹⁰ Pa (about 4 × 10⁵ atmospheres), is a large part of what makes the metal hard to compress.

Degeneracy pressure and white dwarfs

Because exclusion forces electrons into high-momentum states, the gas resists compression even at T = 0: squeeze the volume and n rises, E_F rises as n^(2/3), and the total kinetic energy climbs. Differentiating gives a genuine pressure with no thermal origin at all:

P = (2/5) n E_F ∝ n^(5/3)   (non-relativistic degeneracy pressure)

Scale this idea to a dying star. In a white dwarf, gravity crushes matter until electron degeneracy pressure halts the collapse. As the star grows more massive the electrons become relativistic (v_F → c), the pressure softens to P ∝ n^(4/3), and the balance fails above the Chandrasekhar limit of about 1.4 solar masses (Subrahmanyan Chandrasekhar, 1931). The very same Fermi-energy physics that sets copper's bulk modulus also sets the maximum mass of a white dwarf star.

History

Enrico Fermi and Paul Dirac independently derived the quantum statistics of identical fermions in 1926, building on Wolfgang Pauli's 1925 exclusion principle. In 1927 Arnold Sommerfeld applied Fermi–Dirac statistics to the electron gas in metals, replacing the classical Maxwell–Boltzmann distribution that Drude had assumed in 1900. The immediate payoff was resolving the "missing heat capacity" puzzle — the electronic specific heat is small precisely because only the ~T/T_F fraction of electrons near E_F can be thermally excited. Llewellyn Thomas and Fermi developed the statistical (Thomas–Fermi) model of the atom over 1926–1927 using the same filled-Fermi-sea idea, and the concept became the backbone of the modern band theory of solids.

Common misconceptions

• "The Fermi energy depends on temperature." No — E_F is defined at T = 0. What drifts with temperature is the chemical potential μ(T) ≈ E_F[1 − (π²/12)(k_B·T/E_F)²], and in loose usage "Fermi level" refers to μ. In metals the drift is negligible.
• "Electrons at the Fermi energy are moving fast because the metal is hot." The Fermi velocity (~10⁶ m/s) is a zero-temperature quantity coming from exclusion, not from heating. Thermal motion adds only a tiny correction.
• "Fermi energy is the same as the work function." The work function is the energy to remove an electron at E_F to the vacuum (roughly 4–5 eV for metals); E_F is measured from the bottom of the conduction band. They are different quantities.
• "E_F is the average electron energy." The average energy of the free-electron gas at T = 0 is (3/5)E_F, not E_F. The Fermi energy is the maximum occupied energy.
• "In a semiconductor the Fermi level sits at an occupied state." In an intrinsic semiconductor the Fermi level lies in the middle of the band gap, where there are no states at all — it is the reference energy of half-occupation, not a physically occupied level.
• "All conduction electrons carry the current equally." Only the imbalance of the Fermi surface matters — a small shift of the whole sphere in k-space. Electrons deep in the sea contribute nothing net because their opposite-momentum partners cancel them.