Graham's Law: Why Light Gases Escape Faster

The wall-collision flux: where the law is born

Graham's Law falls straight out of the kinetic theory of gases. Picture a container of gas at temperature T with number density n (molecules per m³). Every molecule is careening around with a distribution of speeds. The quantity that matters for effusion is the effusive flux — how many molecules per second strike a unit area of the wall from the inside. Kinetic theory gives a beautifully simple answer:

• Φ = ¼ n ⟨v⟩ — the number of molecules hitting unit wall area per second,
• where ⟨v⟩ = √(8k_BT/πm) is the mean molecular speed,
• k_B = 1.381 × 10⁻²³ J/K is Boltzmann's constant, m the molecular mass in kg.

The factor of ¼ comes from averaging the perpendicular velocity component over all molecules moving toward the wall — one factor of ½ for the hemisphere of inward-moving molecules, another from the angular average of cos θ. Now cut a hole of area A in that wall, small enough that a molecule passing through it never notices the hole is there. The effusion rate — molecules escaping per second — is simply the flux times the area:

• R = ¼ n ⟨v⟩ A = ¼ n A √(8k_BT/πm).

Everything on the right is fixed except the molecular mass. Hold n, T, and A the same and the rate depends only on ⟨v⟩ ∝ 1/√m. That is Graham's Law: R ∝ 1/√m ∝ 1/√M, where M is the molar mass.

From speeds to the √M ratio

Because m and molar mass M are proportional (M = N_A m, with Avogadro's number N_A = 6.022 × 10²³ /mol), the mass dependence is identical whether you use molecular or molar mass. Comparing two gases at the same T, n, and hole area, every prefactor cancels and you are left with the pure ratio Graham wrote down:

• R₁/R₂ = ⟨v₁⟩/⟨v₂⟩ = √(m₂/m₁) = √(M₂/M₁).

Plug in real numbers. Hydrogen (M = 2.016 g/mol) versus oxygen (M = 32.00 g/mol): the ratio is √(32.00/2.016) = 3.98. Hydrogen effuses nearly four times faster. Helium (4.003) versus methane (16.04): √(16.04/4.003) = 2.00 — helium leaks exactly twice as fast. The He-vs-SF₆ headline number is √(146.06/4.003) = 6.04.

It is worth stressing which speed matters. Kinetic theory carries three characteristic speeds, all ∝ 1/√m:

• most probable speed v_p = √(2k_BT/m),
• mean speed ⟨v⟩ = √(8k_BT/πm) ≈ 1.128 v_p,
• root-mean-square speed v_rms = √(3k_BT/m) ≈ 1.225 v_p.

Effusion is governed by ⟨v⟩, the true arithmetic mean, because it is the mean speed that sets the wall-collision flux. But since all three share the identical 1/√m scaling, the ratio for two gases is the same no matter which you pick — which is exactly why Graham's Law is so robust.

The Knudsen condition: what 'small hole' really means

Graham's Law is exact only in the free-molecular or Knudsen regime, where the hole diameter d is much smaller than the mean free path λ — the average distance a molecule travels between collisions. The governing dimensionless group is the Knudsen number:

• Kn = λ/d, with λ = k_BT / (√2 π D² P),
• where D is the molecular diameter (~0.3–0.4 nm) and P the pressure.

At atmospheric pressure and room temperature, air's mean free path is only about 68 nm. So a true effusion hole must be smaller than that — a genuine molecular pinhole. When Kn ≫ 1, a molecule approaching the hole flies straight through with no clue there is even an opening; the escape is set purely by how often molecules randomly happen to be heading at the aperture, i.e. by the flux ¼n⟨v⟩.

If instead the hole is large (Kn ≪ 1), you no longer have effusion — you have hydrodynamic (viscous) flow. The gas moves as a collective fluid, driven by the pressure difference, and the mass dependence weakens or vanishes. That is why a punctured car tire deflates at nearly the same rate regardless of gas: the hole is enormous compared to λ, so it is bulk flow, not effusion. Graham's clean √M law lives only in the tiny-hole limit.

Graham's plaster plugs and the birth of the law

Thomas Graham, a Scottish chemist at University College London, published his effusion studies in 1846 (building on diffusion work from the 1830s). He didn't have nanometer-scale drilled apertures — instead he used porous plugs: plates of plaster of Paris, graphite, or fine stucco riddled with microscopic channels that, at the reduced pressures he worked with, acted as bundles of molecular pinholes. By timing how fast different gases passed through, he found the escape rate was inversely proportional to the square root of the gas density.

Since at fixed T and P density ρ is proportional to molar mass (ρ = PM/RT from the ideal gas law), rate ∝ 1/√ρ ∝ 1/√M — the two statements are identical. Graham's genius was recognizing a quantitative regularity decades before the kinetic theory of Clausius (1857) and Maxwell (1860) explained why. His law became a prediction that the not-yet-mature molecular theory had to reproduce — and it did, exactly, through the ¼n⟨v⟩ flux. Graham is also honored in Graham's law of diffusion and in colloid chemistry (he coined 'colloid' and 'dialysis').

Enriching uranium: Graham's Law on an industrial scale

The most consequential application is gaseous diffusion enrichment. Natural uranium is 99.3% ²³⁸U and only 0.72% fissile ²³⁵U. Converted to uranium hexafluoride, UF₆ (the only convenient uranium gas), the two isotopes give molecules of mass 349.03 u (²³⁵UF₆) and 352.04 u (²³⁸UF₆). Graham's Law predicts a per-stage effusion-rate ratio of:

• √(352.04/349.03) = 1.00429.

An enrichment of just 0.43% per stage — agonizingly small. To go from 0.72% to weapons-grade (>90%) ²³⁵U requires roughly 1,000–1,400 stages cascaded in series, each barrier a porous nickel membrane with holes near the mean free path. The Manhattan Project's K-25 plant at Oak Ridge, built for exactly this in 1944–45, was the largest building in the world at the time, covered 44 acres, and drew hundreds of megawatts. The theoretical ideal separation factor √(M₂/M₁) is degraded in practice by back-diffusion and finite Kn, so real barriers achieved only ~0.3% per stage. Gaseous diffusion has since been superseded by gas centrifuges (which exploit mass difference via the far stronger centrifugal M-dependence), but the founding physics was pure Graham.

Everyday effusion and things it explains

You don't need Oak Ridge to see Graham's Law. A helium balloon deflates in a day or two while an air-filled latex balloon lasts a week — the latex rubber is a molecular sieve, and He (M = 4) effuses through its micropores far faster than N₂ (28) and O₂ (32). Mylar balloons last far longer because their aluminized film has essentially no molecular-scale holes for effusion.

• Isotope separation of neon and hydrogen in the lab uses cascaded effusion for the same √M reason, historically important for producing deuterium-depleted or heavy-isotope samples.
• Vacuum leak detection: helium is the tracer of choice partly because its tiny mass makes it effuse through the smallest cracks fastest — a helium mass spectrometer sniffs out leaks a heavier gas would never reach.
• The Knudsen cell (effusion cell) in materials science measures a substance's vapor pressure by weighing how fast atoms effuse from a heated crucible through a tiny orifice — the rate directly gives P via R = ¼n⟨v⟩A.
• Planetary atmospheric escape: light gases at the top of an atmosphere effuse into space fastest, which is a major reason Earth retained heavy N₂ and O₂ but lost most of its primordial hydrogen and helium — Jeans escape is effusion writ across a planet.

Subtleties, misconceptions, and the diffusion trap

The single biggest confusion is between effusion and diffusion. Graham's Law is written most cleanly for effusion — escape through a molecular pinhole with no collisions. Diffusion, the spreading of one gas through another, involves constant intermolecular collisions and its mass dependence is only approximately 1/√M, corrected by collision cross-sections and reduced masses. Many textbooks state 'Graham's Law of diffusion,' which is fine as a rough rule but not the exact √M relation.

• Temperature drops out of the ratio. Both gases share the same T, so it cancels — a hotter gas effuses faster in absolute terms (R ∝ √T), but the ratio between two gases at the same T is temperature-independent.
• It's mean speed, not kinetic energy, that differs. A common trap: at equal T both gases have the same average kinetic energy (½m⟨v²⟩ = 3/2 k_BT). It is precisely because energy is equal that the lighter molecule must move faster — v_rms ∝ 1/√m. Equal energy, unequal speed, is the whole engine of the law.
• The hole must stay tiny. As pressure rises or the hole grows, Kn falls below 1 and you cross into viscous flow, where Graham's √M scaling breaks down and pressure-driven Poiseuille flow takes over.
• Real membranes blur it. Porous barriers have a distribution of pore sizes and some surface interaction, which is why practical enrichment stages fall short of the ideal √(M₂/M₁) separation factor.

Effusion vs bulk diffusion: two mass-dependent transport processes often confused

Property	Effusion (Graham's Law)	Diffusion
Geometry	Escape through hole ≪ mean free path	Spreading through a background gas
Collisions in transit	None — molecules stream freely	Constant intermolecular collisions
Mass scaling	rate ∝ 1/√M (exact)	≈ 1/√M, modified by collision cross-section
Controlling regime	Knudsen (Kn = λ/d ≫ 1)	Continuum (Kn ≪ 1)
He vs SF₆ speed ratio	6.04 (matches √(146/4))	Slower, distorted by mutual collisions
Rate driven by	Wall-collision flux, ¼n⟨v⟩	Concentration gradient, Fick's law