Mean Free Path

Definition

Picture one molecule in a gas. It flies in a straight line until it strikes another molecule, then ricochets off in a new direction, flies again, strikes again — an endless zig-zag. The mean free path λ is the average length of those straight segments: the mean distance travelled between one collision and the next.

Kinetic theory (Clausius, 1858; refined by Maxwell) gives the compact result

λ = 1 / (√2 · n · σ)

where the symbols are:

• λ — mean free path, in metres (m).
• n — number density, molecules per cubic metre (m⁻³). For an ideal gas n = P/(k_B·T).
• σ — collision cross-section, in square metres (m²). For hard spheres of effective diameter d, σ = πd².
• √2 — a dimensionless factor from averaging the relative speed of two moving molecules; the average relative speed is √2 times the mean speed of a single molecule.

Substituting the ideal-gas density n = P/(k_B·T) gives the equivalent pressure form, which is often the most useful in the lab:

λ = k_B · T / (√2 · π · d² · P)

with k_B = 1.381 × 10⁻²³ J/K the Boltzmann constant, T the absolute temperature (K), and P the pressure (Pa). This form immediately shows the two levers: λ is inversely proportional to pressure and directly proportional to temperature.

Where the √2 comes from

A first, naive estimate imagines the target molecule as a moving disk of cross-section σ sweeping out a cylinder. In time t it sweeps a volume σ·v̄·t and hits every molecule whose centre lies inside — that is n·σ·v̄·t collisions — so the path between collisions would be λ = 1/(n·σ). This is the Clausius estimate, and it is wrong by a factor of √2.

The fix, due to Maxwell, is that the other molecules are moving too. What matters is the mean relative speed between the flying molecule and its targets, not the mean speed of a single molecule. For a Maxwell–Boltzmann distribution the mean relative speed is

v̄_rel = √2 · v̄

Because the effective collision rate scales with v̄_rel while the distance travelled scales with v̄, the ratio introduces exactly one factor of √2 in the denominator, giving λ = 1/(√2·n·σ). The molecule collides more often than the static picture predicts, so it travels a shorter distance between hits.

Worked example — air at room temperature

Let us compute λ for air at room temperature and atmospheric pressure (T = 298 K, P = 101 325 Pa) — the conditions the canonical "68 nm" figure refers to.

Step 1 — number density. Using n = P/(k_B·T):

n = 101325 / (1.381e-23 × 298)
  ≈ 2.46 × 10²⁵ molecules / m³

Step 2 — cross-section. Air is mostly N₂ and O₂, with effective collision diameter d ≈ 0.37 nm = 3.7 × 10⁻¹⁰ m:

σ = π d² = π × (3.7e-10)²
  ≈ 4.30 × 10⁻¹⁹ m²

Step 3 — mean free path.

λ = 1 / (√2 × 2.46e25 × 4.30e-19)
  ≈ 6.7 × 10⁻⁸ m  ≈ 68 nm

The answer lands at about 68 nm — roughly 200 molecular diameters. (At 0 °C the same diameter gives ≈ 61 nm; at fixed pressure λ grows with temperature, and published values span roughly 60–70 nm depending on the exact diameter and temperature assumed.) So even though each molecule collides billions of times per second, it spends nearly all of its life in free flight over gaps hundreds of times its own size.

Step 4 — collision frequency. The mean molecular speed of air (M ≈ 0.029 kg/mol) is v̄ = √(8RT/πM) ≈ 467 m/s, so the collision rate is z = v̄/λ ≈ 467 / (6.8 × 10⁻⁸) ≈ 6.9 × 10⁹ collisions per second.

Mean free path across regimes

Environment	Pressure	Approx. λ
Air at sea level (STP)	101 kPa	~68 nm
Air at 30 km (stratosphere)	~1 kPa	~7 µm
Air at 100 km (thermosphere edge)	~0.03 Pa	~0.1 m
Rough vacuum (mechanical pump)	~1 Pa	~7 mm
High vacuum (diffusion pump)	~10⁻⁴ Pa	~70 m
Ultra-high vacuum (UHV chamber)	~10⁻⁸ Pa	~700 km
Interstellar medium (~1 atom/cm³)	~10⁻¹⁵ Pa	~10¹² m (~10 AU)

The span is astonishing: from tens of nanometres at sea level to astronomical-unit scales in space. Because λ ∝ 1/P, every factor of ten drop in pressure lengthens the path tenfold.

Why it matters — transport coefficients

The mean free path is the hidden microscopic length behind every gas transport property. Simple kinetic-theory arguments (a molecule carries momentum, energy, or its own identity a distance ~λ before a collision randomizes it) give:

Property	Kinetic-theory estimate	What λ transports
Viscosity η	η ≈ (1/3)·ρ·v̄·λ	Momentum
Thermal conductivity κ	κ ≈ (1/3)·ρ·c_v·v̄·λ	Kinetic energy (heat)
Self-diffusion D	D ≈ (1/3)·v̄·λ	Molecular identity (mass)

Here ρ is the mass density (kg/m³), v̄ the mean speed (m/s), and c_v the specific heat per unit mass (J/(kg·K)). One of the most beautiful predictions of kinetic theory follows immediately: because η ≈ (1/3)·ρ·v̄·λ and λ ∝ 1/n while ρ ∝ n, the product ρ·λ is independent of density — so gas viscosity does not depend on pressure. Maxwell derived this in 1866 and, initially disbelieving it, verified it experimentally. It is why a squeeze of the tyre pump feels the same whether the reservoir is nearly full or nearly empty (at fixed temperature).

The Knudsen number — when fluids stop being fluids

Compare λ to the size L of whatever the gas is flowing through and you get the dimensionless Knudsen number:

Kn = λ / L

It tells you which physics applies:

Kn range	Regime	Description
Kn < 0.01	Continuum	Navier–Stokes valid; gas is a smooth fluid
0.01 – 0.1	Slip flow	Continuum with velocity slip at walls
0.1 – 10	Transitional	Neither continuum nor free-molecular; needs kinetic (Boltzmann/DSMC) treatment
Kn > 10	Free-molecular	Molecules hit walls far more than each other

This is not academic. A MEMS gas sensor with 100 nm channels, a satellite skimming the thermosphere, gas transport through a shale nanopore, or the residual gas in a vacuum tube can all sit at Kn > 1, where treating the gas as a continuous fluid gives badly wrong drag, heat transfer, and pressure. Reentry aerodynamics and semiconductor deposition chambers both live and die by the Knudsen number.

A little history

Rudolf Clausius introduced the concept of a mean free path in 1858 to answer a sharp objection to kinetic theory: if molecules fly at hundreds of metres per second, why does an opened perfume bottle take minutes, not milliseconds, to be smelled across a room? Clausius's answer was that molecules do not fly freely — they collide constantly, and diffusion is a slow random walk with step length λ. James Clerk Maxwell then put the theory on a statistical footing in the 1860s, supplying the √2 factor and the pressure-independence of viscosity. The mean free path became one of the first quantitative bridges between the invisible molecular world and measurable macroscopic quantities, and it remains a cornerstone of statistical mechanics and rarefied-gas dynamics today.

Common misconceptions

• "The mean free path depends on temperature." Only through density. At fixed number density n, λ = 1/(√2·n·σ) has no T in it at all. The temperature dependence you see in λ = k_B·T/(√2·π·d²·P) comes entirely from holding pressure fixed, which lets n fall as T rises. (Higher T also speeds molecules up, but that shortens the time between collisions, not the distance.)
• "Forgetting the √2." The naive λ = 1/(n·σ) ignores that targets are moving. It overestimates λ by 41%. Always include the √2 for a gas of identical molecules.
• "σ is the physical area of a molecule." σ = πd² uses the collision diameter d — the centre-to-centre distance at contact — so σ is the area of a circle of radius d, four times a molecule's own cross-sectional area πr². It is an effective interaction area, not a geometric footprint.
• "Mean free path and mean free time are the same thing." The mean free path λ is a distance; the mean free time τ = λ/v̄ is a time. Their inverse gives the collision frequency z = v̄/λ = 1/τ.
• "Viscosity rises with pressure." For a dilute gas it does not — η ≈ (1/3)ρv̄λ is pressure-independent because ρ and λ cancel. This famous counterintuitive result only fails at very high densities where the gas is no longer dilute.
• "λ is the same for every molecule." It is a statistical average. Individual free paths follow an exponential distribution P(ℓ) = (1/λ)·e^(−ℓ/λ), so many are much shorter and a few much longer than λ.