The Library · PhysicsPlate № 278 · Folio II
ILL. № 278
PHYS
Plate — The Boltzmann Distribution

The Boltzmann Distribution

At thermal equilibrium, a state's probability is proportional to e^(−E/kT) — the exponential decay that organizes every microscopic ensemble.
Suggested next → Logarithms & Exponentials · MATH · T4
Facets
  • Pᵢ ∝ exp(−Eᵢ/k_BT), the exponential suppressionnot yet tested
  • The partition function and equilibrium thermodynamicsnot yet tested
  • Molecular speeds, reaction rates, and thermal physicsnot yet tested
  • Classical limit of Bose-Einstein and Fermi-Diracnot yet tested
The brief

In 1868, Ludwig Boltzmann — a young Austrian physicist — derived a remarkably simple result. In a gas at thermal equilibrium with temperature T, the probability of finding a molecule with energy E is proportional to e^(−E/k_BT). The formula is short, the derivation is short, and the consequences are the entire structure of equilibrium statistical mechanics. The Boltzmann distribution underpins every chemical reaction rate (Arrhenius's law), every spectroscopic line intensity, every magnetic susceptibility calculation, every semiconductor carrier concentration, every astrophysical model of stellar atmospheres. It is one of the most-applied formulas in physics, and it crystallizes the deep idea that higher-energy states are exponentially less likely.

Consider a system with discrete energy levels E₁, E₂, E₃, … in thermal equilibrium with a reservoir at temperature T. The Boltzmann distribution gives the probability of finding the system in state i: Pᵢ = (1/Z)·exp(−Eᵢ / k_B·T), where k_B is Boltzmann's constant (1.381 × 10⁻²³ J/K) and Z = Σ exp(−Eⱼ / k_B·T) is the partition function, the normalization that ensures the probabilities sum to one. The distribution can be derived from maximum-entropy reasoning (subject to the constraint of a fixed average energy, the Boltzmann distribution maximizes the entropy) or from microscopic counting (the macrostate with the most microstates dominates). Salient features: the exponential suppression of high-energy states means that most of the population lies near the lowest energies; the temperature sets the scale — at high T, more states are appreciably occupied; at low T, the system collapses into the ground state. The Maxwell-Boltzmann distribution of molecular speeds in a classical gas is a special case: P(v) ∝ v²·exp(−mv²/2k_B·T), peaking at v_p = √(2k_B·T/m). The partition function Z is more than a normalization — it encodes the entire equilibrium thermodynamics. Free energy F = −k_B·T·ln Z; internal energy U = −∂(ln Z)/∂β (where β = 1/k_B·T); entropy S = −k_B·Σ Pᵢ·ln Pᵢ; heat capacity follows by another derivative. Quantum statistics modify the Boltzmann result for indistinguishable particles: Bose-Einstein statistics for bosons (allowing many particles in the same state) and Fermi-Dirac statistics for fermions (with Pauli exclusion). The classical Boltzmann form is the high-temperature, low-density limit of both. Chemical equilibrium constants, vapor pressures, reaction rate constants (the Arrhenius equation k = A·exp(−E_a/k_B·T)), semiconductor carrier densities, stellar atmospheric line strengths — all derive from the Boltzmann distribution applied to the appropriate microscopic system.

Why nowChemistry texts commonly introduce Arrhenius factors in reaction kinetics. Semiconductor models use Boltzmann statistics in non-degenerate regimes and Fermi–Dirac statistics more generally. Plasma physics — fusion-reactor design, stellar atmospheres, ionospheric modeling — runs on Boltzmann-distributed populations of ionization states. Atmospheric physics uses the Boltzmann factor to compute the height-dependent density of atmospheric gases. Markov-chain Monte Carlo methods in machine learning sample from Boltzmann-like distributions; energy-based models (including Hopfield networks, Boltzmann machines, and the diffusion models behind today's image generators) are direct conceptual descendants of Boltzmann's exponential suppression. The little formula Boltzmann derived in 1868 has, in retrospect, quietly underwritten an enormous fraction of applied physical science.