Mean Free Path

Take a single nitrogen molecule at room temperature. It moves at roughly 470 meters per second. That is faster than a rifle bullet. Yet it takes that molecule about a hundred nanoseconds to cross a single centimeter. This is because it collides with neighbors roughly a billion times per second. Between collisions, it travels only a few tens of nanometers. That short zigzag distance is the mean free path. The mean free path matters because it sets the length scale for transport. Molecules carry momentum, energy, and mass through a gas over this scale. It explains why gases diffuse slowly despite molecular speeds being enormous. It explains why viscosity and thermal conductivity are pressure-independent over a wide range. It also explains why vacuum systems behave so differently from gases at atmospheric pressure. It is the geometric bridge between molecular size and macroscopic transport. This article builds the mean free path from the collision frequency and collision cross-section you already know. You will see the full derivation with the corrected relative-speed factor. Then you will see how it scales with pressure, temperature, and molecular size. Two worked examples ground the formula in real numbers. These range from air at STP to interstellar gas.

Key takeaways

The mean free path $\lambda$ is the average distance a molecule travels between successive collisions.
It is built from the mean speed divided by the collision frequency: $\lambda = \bar{c}/z$ .
Including the correct relative-speed factor gives $\lambda = k_BT/(\sqrt{2}\,\sigma p)$ .
$\lambda$ is inversely proportional to pressure and collision cross-section, but independent of molecular mass.
At fixed pressure, $\lambda$ increases with temperature because the gas expands and molecules spread out.
Key idea: the mean free path is simply "how far a molecule coasts, on average, before its next collision" — the ratio of speed to collision rate, corrected for the fact that both partners in a collision are moving.

The core insight

The mean free path emerges from a single geometric idea: a molecule sweeps out a thin cylindrical "collision tube" of cross-sectional area $\sigma$ as it moves, and it collides whenever another molecule's center lies inside that tube; dividing the distance traveled per second by the number of collisions per second — with a $\sqrt{2}$ correction for the fact that targets are moving too — gives the average coasting distance between collisions.

The physical picture

Imagine a single tagged molecule of diameter $d$ moving through a sea of otherwise stationary molecules of the same diameter. As it travels, it sweeps out an imaginary cylinder of cross-sectional area $\sigma = \pi d^2$ — the collision cross-section, defined so that any molecule whose center falls within this tube triggers a collision. In one second, the tagged molecule travels a distance equal to its mean speed $\bar{c}$ , sweeping a cylinder of volume $\sigma \bar{c}$ . The number of collisions per second, the collision frequency $z$ , is this swept volume times the number density of targets, $z = \sigma \bar{c} \mathcal{N}$ , where $\mathcal{N} = N/V$ is molecules per unit volume. The mean free path is then the total distance traveled per second divided by the number of collisions suffered per second:

\lambda = \frac{\bar{c}}{z}

But this simple picture has a subtlety. Real target molecules are not stationary — they move with the same distribution of speeds as the tagged molecule. What matters for collisions is not the tagged molecule's absolute speed but its speed relative to the other moving molecules. Averaging over the Maxwell–Boltzmann distribution, the mean relative speed between two molecules of the same mass is larger than the mean speed of either one alone, by a factor of $\sqrt{2}$ :

\bar{c}_{\text{rel}} = \sqrt{2}\,\bar{c}

This means the true collision frequency is $z = \sqrt{2}\,\sigma \bar{c}\, \mathcal{N}$ , and since $\lambda$ is the distance traveled (at speed $\bar{c}$ , not $\bar{c}_{\text{rel}}$ ) between collisions, the $\sqrt{2}$ lands in the denominator of $\lambda$ , not the numerator — a point worked through carefully below.

Figure

A molecule's collision tube: sweeping out cross-section σ over successive free flights

The core relations

\sigma = \pi d^2

where: $d$ is the effective collision diameter (sum of molecular radii for unlike species).

z = \sqrt{2}\,\sigma \bar{c}\,\mathcal{N}

where: $z$ is the collision frequency of one molecule (collisions per second), $\mathcal{N} = N/V$ is number density, $\bar{c} = \sqrt{8k_BT/\pi m}$ is the mean speed.

\lambda = \frac{k_BT}{\sqrt{2}\,\sigma p}

where: $k_B$ is Boltzmann's constant, $T$ is temperature, $p$ is pressure, $\sigma$ is the collision cross-section.

\lambda = \frac{1}{\sqrt{2}\,\sigma \mathcal{N}}

where: this is the number-density form, equivalent to the pressure form via the ideal gas law $p = \mathcal{N}k_BT$ .

Derivation[From collision frequency to the mean free path]

Goal: derive $\lambda = k_BT/(\sqrt{2}\sigma p)$ starting from the collision frequency and the definition of mean free path as distance-per-collision.

Step 1 — Set up the collision tube. Consider a tagged molecule of diameter $d$ moving with mean speed $\bar{c}$ through a gas of identical stationary molecules with number density $\mathcal{N}$ . A collision occurs whenever the center of a target molecule lies within distance $d$ of the tagged molecule's center, so the effective target is a circle of cross-sectional area

\sigma = \pi d^2

In one second, the tagged molecule sweeps a cylindrical volume of length $\bar{c}$ and cross-section $\sigma$ , namely $V_{\text{swept}} = \sigma \bar{c}$ .

Step 2 — Count collisions per second (naive, stationary-target version). The number of target centers inside this swept volume is $\sigma \bar{c}\,\mathcal{N}$ . If targets were truly stationary this would be the collision frequency $z_0 = \sigma \bar{c}\,\mathcal{N}$ .

Step 3 — Correct for the motion of the targets. Targets are not stationary; they share the same Maxwell–Boltzmann speed distribution as the tagged molecule. What determines whether a collision happens per unit time is the relative speed between the tagged molecule and each target, not the tagged molecule's speed alone. For two molecules of the same mass $m$ , each with speed distributed according to Maxwell–Boltzmann, the mean of the relative speed is

\bar{c}_{\text{rel}} = \sqrt{2}\,\bar{c}

This result follows because the relative velocity $\vec{v}_{\text{rel}} = \vec{v}_1 - \vec{v}_2$ of two independent, identically distributed Maxwellian velocities is itself Maxwellian, but with an effective mass $\mu = m/2$ instead of $m$ ; since $\bar{c} \propto \sqrt{1/m}$ , replacing $m$ with $m/2$ multiplies the mean speed by $\sqrt{2}$ . The correct collision frequency is therefore

z = \sigma \bar{c}_{\text{rel}}\,\mathcal{N} = \sqrt{2}\,\sigma \bar{c}\,\mathcal{N}

Step 4 — Form the mean free path. By definition, $\lambda$ is the average distance traveled by the tagged molecule (at its own mean speed $\bar{c}$ , since that is the actual pathlength it covers) divided by the number of collisions it suffers per second:

\lambda = \frac{\bar{c}}{z} = \frac{\bar{c}}{\sqrt{2}\,\sigma \bar{c}\,\mathcal{N}} = \frac{1}{\sqrt{2}\,\sigma \mathcal{N}}

Notice that $\bar{c}$ cancels exactly: the mean free path does not depend on how fast the molecule moves, only on how densely packed the targets are and how large the cross-section is. This is expected — $\lambda$ is a purely geometric quantity, a mean distance, not a time.

Step 5 — Convert to pressure and temperature. Using the ideal gas law in the molecular form $p = \mathcal{N} k_B T$ , so $\mathcal{N} = p/(k_BT)$ , substitute into Step 4:

\lambda = \frac{1}{\sqrt{2}\,\sigma \left(\dfrac{p}{k_BT}\right)} = \frac{k_BT}{\sqrt{2}\,\sigma p}

\boxed{\lambda = \frac{k_BT}{\sqrt{2}\,\sigma p}}

Worked examples

Example 1 — Nitrogen gas at STP (foundational). Find $\lambda$ for N₂ at $T = 298\ \text{K}$ , $p = 1.00 \times 10^5\ \text{Pa}$ , using collision diameter $d = 3.75 \times 10^{-10}\ \text{m}$ (a standard tabulated value for N₂). Cross-section:

\sigma = \pi d^2 = \pi (3.75\times10^{-10}\,\text{m})^2 = 4.42\times10^{-19}\ \text{m}^2

Mean free path:

\lambda = \frac{k_BT}{\sqrt{2}\,\sigma p} = \frac{(1.381\times10^{-23}\,\text{J K}^{-1})(298\,\text{K})}{\sqrt{2}\,(4.42\times10^{-19}\,\text{m}^2)(1.00\times10^{5}\,\text{Pa})}

Numerator: $1.381\times10^{-23} \times 298 = 4.116\times10^{-21}\ \text{J}$ . Denominator: $\sqrt{2}\times 4.42\times10^{-19}\times 1.00\times10^{5} = 6.251\times10^{-14}\ \text{J m}^{-1}$ .

\lambda = \frac{4.116\times10^{-21}}{6.251\times10^{-14}}\ \text{m} = 6.58\times10^{-8}\ \text{m} \approx 66\ \text{nm}

Sanity check: the mean speed of N₂ at 298 K is $\bar{c} = \sqrt{8k_BT/\pi m} \approx 475\ \text{m s}^{-1}$ . The collision frequency is then $z = \bar{c}/\lambda \approx 475/(6.58\times10^{-8}) \approx 7.2\times10^{9}\ \text{s}^{-1}$ , i.e., about 7 billion collisions per second — consistent with textbook values for air at room conditions. This confirms $\lambda \sim 60$ – $70$ nm is the right order of magnitude, roughly 175 times the molecular diameter.

Example 2 — Vacuum system design (applied, harder). A chemist is running an electron-diffraction experiment and needs the mean free path of residual N₂ gas to exceed 10 cm so that gas-phase molecules don't scatter the electron beam. To what pressure must the chamber be evacuated, at $T = 298\ \text{K}$ , using the same $\sigma = 4.42\times10^{-19}\ \text{m}^2$ ? Rearranging the boxed result for pressure:

p = \frac{k_BT}{\sqrt{2}\,\sigma \lambda}

With $\lambda = 0.10\ \text{m}$ :

p = \frac{(1.381\times10^{-23})(298)}{\sqrt{2}(4.42\times10^{-19})(0.10)}

Numerator (same as before): $4.116\times10^{-21}\ \text{J}$ . Denominator: $\sqrt{2}\times 4.42\times10^{-19}\times 0.10 = 6.251\times10^{-20}\ \text{J m}^{-1}\cdot\text{m}^{-1}$ ... carrying units carefully: $\sqrt{2}\sigma\lambda = 1.4142 \times 4.42\times10^{-19}\,\text{m}^2 \times 0.10\,\text{m} = 6.251\times10^{-20}\ \text{m}^3$ .

p = \frac{4.116\times10^{-21}\ \text{J}}{6.251\times10^{-20}\ \text{m}^3} = 6.58\times10^{-2}\ \text{Pa}

Converting to more familiar vacuum units: $6.58\times10^{-2}\ \text{Pa} \times \frac{1\ \text{torr}}{133.3\ \text{Pa}} \approx 4.9\times10^{-4}\ \text{torr}$ , i.e., a high-vacuum regime (roughly $10^{-4}$ torr). This is achievable with a good turbomolecular pump, confirming the experiment is technically feasible without requiring ultra-high vacuum. Sanity check: since $\lambda \propto 1/p$ , going from $10^5$ Pa (Example 1, $\lambda \approx 66$ nm) to $6.58\times10^{-2}$ Pa is a pressure drop by a factor of $1.52\times10^{6}$ , and indeed $66\ \text{nm} \times 1.52\times10^{6} \approx 0.10\ \text{m}$ — consistent.

Limiting cases & common traps

Situation	Behavior of $\lambda$	Why
Pressure $p \to 0$ (high vacuum)	$\lambda \to \infty$	Fewer targets per unit volume; molecules coast freely
Larger cross-section $\sigma$ (bigger molecules)	$\lambda$ decreases	Bigger "target area" means more frequent hits
Higher $T$ at fixed $p$	$\lambda$ increases	Gas expands ( $\mathcal{N} = p/k_BT$ falls), fewer targets per volume
Higher $T$ at fixed $\mathcal{N}$ (fixed volume, sealed container)	$\lambda$ unchanged	$\lambda = 1/(\sqrt{2}\sigma\mathcal{N})$ has no explicit $T$ dependence
Heavier molecules, same $\sigma$ , same $p$ , $T$	$\lambda$ unchanged	Mass cancels out of $\lambda$ entirely

The chemist behind it

The modern kinetic-theory treatment of the mean free path traces to Rudolf Clausius, who in 1858 first introduced the concept explicitly to explain why gas diffusion is so much slower than molecular speeds would suggest — resolving an objection that had been raised against the kinetic theory itself. Clausius's original estimate treated all target molecules as stationary; it was James Clerk Maxwell, shortly afterward, who supplied the full velocity-distribution treatment and derived the $\sqrt{2}$ relative-speed correction used here, linking the mean free path rigorously to the distribution that bears his name.

Origins · 1 minRudolf Clausius and how this was discovered

Rudolf Clausius

1822–1888 · Introduced the mean free path concept in 1858 to defend kinetic theory against the "gases should diffuse instantly" objection

Ask the tutor: How does the mean free path plug into the formula for a gas's viscosity or thermal conductivity?

Why does the mean free path depend on pressure but not on molecular mass?

The mean free path is a purely geometric quantity: it measures how far a molecule travels, on average, before another molecule's center enters its collision tube. Pressure controls the number density of targets, $\mathcal{N} = p/k_BT$ , which directly sets how crowded the "obstacle course" is. Molecular mass affects how fast a molecule moves and thus how often per second it collides, but that same mass-dependence cancels when you divide distance traveled by collisions per second — leaving $\lambda$ mass-independent.

Does the √2 factor matter numerically, or is it just a fine correction?

It is not negligible: omitting it overestimates $\lambda$ by 41%. The naive stationary-target picture gives $\lambda_0 = 1/(\sigma\mathcal{N})$ , while the correct treatment gives $\lambda = 1/(\sqrt{2}\sigma\mathcal{N})$ , smaller by a factor of $1/\sqrt{2} \approx 0.707$ . Since collision frequencies, viscosities, and diffusion coefficients all depend on $\lambda$ , the √2 correction is essential for getting transport coefficients right to within a few percent of experiment, not just as a theoretical nicety.

How is the mean free path related to the collision frequency from the prerequisite topic?

They are reciprocally linked through the mean speed: $\lambda = \bar{c}/z$ , where $z$ is the collision frequency (collisions per second) built earlier from the same cross-section $\sigma$ and number density $\mathcal{N}$ . If you already have $z = \sqrt{2}\sigma\bar{c}\mathcal{N}$ from the collision-frequency discussion, the mean free path follows immediately by dividing $\bar{c}$ by $z$ — the speed cancels, leaving a formula with no explicit speed dependence at all.

Why is the mean free path important for real gases and everyday phenomena?

It sets the length scale for diffusion, viscosity, and thermal conductivity — a gas can only transport momentum or heat efficiently over distances shorter than $\lambda$ per collision, so transport coefficients scale with $\lambda$ . It also defines the boundary between continuum flow and molecular (Knudsen) flow: when $\lambda$ becomes comparable to the size of a container or pore, as in aerogels, vacuum systems, or the upper atmosphere, gas behavior changes qualitatively from what everyday fluid intuition predicts.