Collision Frequency and Collision Cross-Section

Picture a single nitrogen molecule at room temperature, darting through air at roughly 470 metres per second. It does not travel far before slamming into a neighbor. In one second it suffers about seven billion such collisions. This staggering number is not a curiosity — it sets the timescale for every chemical reaction, every pressure equilibration, and every diffusive mixing process in the gas phase. Kinetic theory lets us calculate this collision rate from first principles, using nothing more than molecular size, mass, temperature, and pressure. The key geometric idea is the collision cross-section: an effective target area that converts a molecule's speed into a rate of encounters. This article builds the collision frequency step by step. You will see why relative speed, not individual speed, governs collisions. Then you will derive the collision density for both single-component and mixed gases. Finally you will connect these results to the mean free path and to gas-phase transport properties.

Key takeaways

Collision frequency $z$ is the number of collisions one molecule suffers per second; it scales with cross-section, density, and speed.
The relevant speed is the mean relative speed $\bar{v}_{rel} = \sqrt{2}\,\bar{v}$ , not the mean speed of a single molecule.
The collision cross-section $\sigma = \pi d^2$ treats molecules as hard spheres of diameter $d$ ; it sets the effective target area.
Collision density $Z_{AA}$ or $Z_{AB}$ counts total collisions per unit volume per second, essential for reaction rate theory.
Both $z$ and $Z$ depend on pressure through number density, so raising pressure increases collision rates linearly.
Key idea: a moving molecule sweeps out a cylindrical collision volume each second, and multiplying its cross-sectional area by the relative speed and target density gives the collision frequency directly.

The core insight

A molecule "collides" whenever the center of an approaching partner enters a disk of area $\sigma = \pi d^2$ swept out at the mean relative speed $\bar{v}_{rel}$ , so the collision rate is simply this swept volume per second times the number density of targets.

The physical picture

Imagine a single gas molecule, modeled as a hard sphere of diameter $d$ , moving through a sea of identical stationary-seeming partners. As it travels a distance $\bar{v}\,\Delta t$ in time $\Delta t$ , it does not need to touch another molecule's center to collide — contact occurs whenever the centers of the two spheres come within a distance $d$ of each other (assuming both have the same diameter). This means the effective target is not the molecule's own cross-sectional area $\pi(d/2)^2$ , but rather a circle of radius $d$ , giving cross-section

\sigma = \pi d^2.

As the molecule moves, it sweeps out an imaginary cylinder of cross-sectional area $\sigma$ and length equal to the distance traveled. Any molecule whose center lies inside this cylinder gets struck. The number of collisions in one second is therefore the volume of this cylinder swept per second, multiplied by the number density of target molecules, $\mathcal{N} = N/V$ . The subtlety is that all molecules are moving simultaneously, not just the one we are tracking. Two molecules approach each other with a relative velocity that is generally larger than either individual velocity, because both partners contribute to closing the gap. Kinetic theory shows that averaging over the Maxwell–Boltzmann distribution for both particles gives a mean relative speed

\bar{v}_{rel} = \sqrt{2}\,\bar{v}

for collisions between identical molecules, where $\bar{v}$ is the ordinary mean speed of a single molecule. This factor of $\sqrt{2}$ is the single most important correction that distinguishes the naive "swept cylinder" estimate from the correct kinetic-theory result.

Figure

The collision cylinder swept by a moving molecule in one second

Once we know the collision rate for a single molecule, we scale up to the whole gas. Every molecule is colliding at the same average rate $z$ , so the total number of collisions per unit volume per second, called the collision density $Z_{AA}$ , is obtained by multiplying by the number density and dividing by two — otherwise each collision would be counted twice, once from each partner's perspective.

Figure

Collision frequency versus pressure at fixed temperature for N2 at 298 K

The core relations

\sigma = \pi d^2

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

\bar{v}_{rel} = \sqrt{2}\,\bar{v}, \qquad \bar{v} = \sqrt{\dfrac{8RT}{\pi M}}

where $\bar{v}_{rel}$ applies to identical molecules; $M$ is molar mass.

\bar{v}_{rel,AB} = \sqrt{\dfrac{8RT}{\pi \mu}}, \qquad \mu = \dfrac{M_A M_B}{M_A+M_B}

where $\mu$ is the reduced molar mass for collisions between unlike species A and B.

z = \sigma \bar{v}_{rel}\,\mathcal{N}

where $z$ is the collision frequency of one molecule (collisions per second); $\mathcal{N}=N/V$ is number density.

Z_{AA} = \tfrac{1}{2}\, z\, \mathcal{N} = \tfrac{1}{\sqrt2}\,\sigma \bar{v}\, \mathcal{N}^2

where $Z_{AA}$ is the total collision density among identical A molecules (collisions per unit volume per second).

Z_{AB} = \sigma_{AB}\,\bar{v}_{rel,AB}\,\mathcal{N}_A \mathcal{N}_B

where $Z_{AB}$ is the collision density between unlike molecules A and B (no factor of 1/2, since each collision is distinct).

Derivation AFrom the Maxwell–Boltzmann distribution to the collision frequency

Goal: derive $z = \sigma \bar{v}_{rel}\mathcal{N}$ and hence $Z_{AA}$ , starting from the swept-cylinder picture and the relative-speed distribution.

Step 1. The swept cylinder. Consider one tagged molecule moving at speed $v$ for a short time $dt$ . It sweeps a cylindrical volume of cross-sectional area $\sigma = \pi d^2$ and length $v\,dt$ , so the swept volume is $\sigma v\, dt$ . Any target-molecule center inside this volume results in a collision. If target molecules were stationary and uniformly distributed with number density $\mathcal{N}$ , the number of collisions in time $dt$ would be $\sigma v \mathcal{N}\, dt$ , giving a naive collision frequency $z_{naive} = \sigma v \mathcal{N}$ .

Step 2. Why relative speed matters. In reality every "target" molecule is also moving, with its own velocity drawn from the Maxwell–Boltzmann distribution. What matters for a collision is the rate of approach, i.e. the magnitude of the relative velocity $\vec{v}_{rel} = \vec{v}_1 - \vec{v}_2$ . If both molecules have mass $m$ and independent Maxwell–Boltzmann velocity distributions, the relative velocity is itself Maxwell–Boltzmann distributed, but with an effective mass equal to the reduced mass $\mu = m/2$ (for identical particles). Because the speed distribution depends on $\sqrt{m/T}$ , replacing $m$ by $m/2$ is equivalent to scaling all speeds by $\sqrt{2}$ .

Step 3. Averaging over the relative speed distribution. The mean speed formula from the Maxwell–Boltzmann distribution is

\bar{v} = \sqrt{\dfrac{8k_BT}{\pi m}}.

Substituting the reduced mass $\mu = m/2$ for identical-particle collisions,

\bar{v}_{rel} = \sqrt{\dfrac{8k_BT}{\pi(m/2)}} = \sqrt{2}\sqrt{\dfrac{8k_BT}{\pi m}} = \sqrt{2}\,\bar{v}.

This confirms that the correct replacement in Step 1 is $v \to \bar{v}_{rel} = \sqrt{2}\,\bar{v}$ , not simply $\bar v$ itself.

Step 4. Collision frequency of a single molecule. Combining Steps 1 and 3, the properly averaged collision frequency for one molecule becomes

z = \sigma\,\bar{v}_{rel}\,\mathcal{N} = \sqrt{2}\,\sigma\,\bar{v}\,\mathcal{N}.

This is the number of collisions any single molecule undergoes per second, on average.

Step 5. Total collision density. To get the total number of collision events per unit volume per second among $\mathcal{N}$ identical molecules, multiply the per-molecule rate $z$ by the number density $\mathcal{N}$ — but this counts every collision twice (once from each partner's viewpoint), so divide by 2:

Z_{AA} = \frac{1}{2} z \mathcal{N} = \frac{1}{2}\sqrt{2}\,\sigma\,\bar{v}\,\mathcal{N}^2 = \frac{\sigma \bar v}{\sqrt2}\mathcal N^2.

Step 6. Unlike molecules. For collisions between distinct species A and B, there is no double counting, because a collision between an A and a B molecule is a single, unambiguous event type. The reduced mass is now the true reduced mass $\mu = m_A m_B/(m_A+m_B)$ , and

Z_{AB} = \sigma_{AB}\,\bar{v}_{rel,AB}\,\mathcal{N}_A\mathcal{N}_B, \qquad \bar v_{rel,AB}=\sqrt{\dfrac{8k_BT}{\pi\mu}}.

\boxed{z = \sigma\sqrt{2}\,\bar v\,\mathcal N, \qquad Z_{AA}=\dfrac{\sigma \bar v}{\sqrt2}\mathcal N^2, \qquad Z_{AB}=\sigma_{AB}\bar v_{rel,AB}\mathcal N_A\mathcal N_B}

Worked examples

Example 1 — Collision frequency of N₂ at room conditions (foundational). Take nitrogen gas, $M = 28.02\times10^{-3}\ \text{kg mol}^{-1}$ , at $T = 298\ \text{K}$ and $p = 1.00\ \text{atm} = 1.013\times10^{5}\ \text{Pa}$ . Use collision diameter $d = 3.75\times10^{-10}\ \text{m}$ (a standard tabulated value for N₂), so

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

Mean speed:

\bar v = \sqrt{\dfrac{8RT}{\pi M}} = \sqrt{\dfrac{8(8.314\,\text{J mol}^{-1}\text{K}^{-1})(298\,\text{K})}{\pi(28.02\times10^{-3}\,\text{kg mol}^{-1})}} = 475\ \text{m s}^{-1}.

Relative speed:

\bar v_{rel} = \sqrt2\,\bar v = 1.414 \times 475\ \text{m s}^{-1} = 672\ \text{m s}^{-1}.

Number density from the ideal gas law:

\mathcal N = \dfrac{p}{k_BT} = \dfrac{1.013\times10^{5}\,\text{Pa}}{(1.381\times10^{-23}\,\text{J K}^{-1})(298\,\text{K})} = 2.46\times10^{25}\ \text{m}^{-3}.

Collision frequency:

z = \sigma \bar v_{rel}\mathcal N = (4.42\times10^{-19}\,\text{m}^2)(672\,\text{m s}^{-1})(2.46\times10^{25}\,\text{m}^{-3})

z \approx 7.3\times10^{9}\ \text{s}^{-1}.

So each N₂ molecule collides about 7 billion times per second at 1 atm and room temperature — consistent with the plot above. Collision density:

Z_{AA} = \tfrac12 z\mathcal N = \tfrac12(7.3\times10^9\,\text{s}^{-1})(2.46\times10^{25}\,\text{m}^{-3}) \approx 9.0\times10^{34}\ \text{m}^{-3}\text{s}^{-1}.

Sanity check: $z^{-1} \approx 1.4\times10^{-10}\ \text{s} = 0.14\ \text{ns}$ , an extremely short but physically reasonable time between collisions for a dense gas of small, fast molecules.

Example 2 — Cross-collisions in a He/Ar mixture (applied, harder). A vessel at $T = 500\ \text{K}$ and total pressure $p_{tot} = 2.00\ \text{atm}$ contains an equimolar mixture of helium ( $M_{He}=4.00\times10^{-3}\ \text{kg mol}^{-1}$ , $d_{He}=2.60\times10^{-10}\ \text{m}$ ) and argon ( $M_{Ar}=39.95\times10^{-3}\ \text{kg mol}^{-1}$ , $d_{Ar}=3.40\times10^{-10}\ \text{m}$ ). Find the He–Ar collision density.

Collision diameter for unlike pairs (arithmetic mean, standard hard-sphere mixing rule):

d_{AB} = \tfrac12(d_{He}+d_{Ar}) = \tfrac12(2.60+3.40)\times10^{-10}\,\text{m} = 3.00\times10^{-10}\ \text{m}.

\sigma_{AB} = \pi d_{AB}^2 = \pi(3.00\times10^{-10}\,\text{m})^2 = 2.83\times10^{-19}\ \text{m}^2.

Reduced mass:

\mu = \dfrac{M_{He}M_{Ar}}{M_{He}+M_{Ar}} = \dfrac{(4.00\times10^{-3})(39.95\times10^{-3})}{(4.00\times10^{-3})+(39.95\times10^{-3})}\ \text{kg mol}^{-1} = 3.636\times10^{-3}\ \text{kg mol}^{-1}.

Mean relative speed:

\bar v_{rel,AB} = \sqrt{\dfrac{8RT}{\pi\mu}} = \sqrt{\dfrac{8(8.314)(500)}{\pi(3.636\times10^{-3})}}\ \text{m s}^{-1} = 2.15\times10^{3}\ \text{m s}^{-1}.

Partial pressures: since the mixture is equimolar, $p_{He}=p_{Ar}=1.00\ \text{atm} = 1.013\times10^5\ \text{Pa}$ .

Number densities:

\mathcal N_{He} = \mathcal N_{Ar} = \dfrac{1.013\times10^{5}\,\text{Pa}}{(1.381\times10^{-23}\,\text{J K}^{-1})(500\,\text{K})} = 1.467\times10^{25}\ \text{m}^{-3}.

Collision density:

Z_{AB} = \sigma_{AB}\bar v_{rel,AB}\mathcal N_{He}\mathcal N_{Ar}

= (2.83\times10^{-19})(2.15\times10^{3})(1.467\times10^{25})(1.467\times10^{25})\ \text{m}^{-3}\text{s}^{-1}

Z_{AB} \approx 1.31\times10^{35}\ \text{m}^{-3}\ \text{s}^{-1}.

Sanity check: this is comparable in magnitude to the pure N₂ result despite lower total density, because the light He atoms move very fast ( $\bar v_{rel}$ here exceeds the N₂ value by a factor of about 3), showing how strongly $Z$ depends on the reduced mass through $\bar v_{rel}\propto 1/\sqrt\mu$ .

Limiting cases & common traps

Situation	What happens	Why
$T\to$ high	$z, Z \propto \sqrt T$	speeds scale as $\sqrt T$ , cross-section unchanged
$p\to$ high (fixed $T$ )	$z\propto p$ , $Z\propto p^2$	$\mathcal N \propto p$ ; $Z$ has two density factors
Light species (e.g. He, H₂)	large $\bar v_{rel}$ , large $z$ per collision partner	speed $\propto 1/\sqrt{M}$ or $1/\sqrt\mu$
Large molecules	large $\sigma$ , large $z$	$\sigma = \pi d^2$ grows with size
Two different species	use $Z_{AB}$ , no factor of 1/2	each A–B pair is a distinct, non-duplicated event

The chemist behind it

The hard-sphere collision model traces back to the kinetic theory developed by James Clerk Maxwell and Ludwig Boltzmann in the 1860s–1870s, but it was Max Trautz and, independently, William Lewis who in 1916–1918 turned collision theory into a quantitative tool for chemical kinetics. They combined the collision frequency derived here with a Boltzmann factor for activation energy and a steric factor, producing the collision theory of reaction rates — a direct ancestor of transition state theory. Their insight was that a bimolecular rate constant should be proportional to $Z_{AB}$ per unit concentration, multiplied by the fraction of collisions energetic and correctly oriented enough to react.

Origins · 1 minWilliam Cudmore McCullagh Lewis and how this was discovered

William Cudmore McCullagh Lewis

1885–1956 · British physical chemist who, alongside Max Trautz, formulated collision theory linking gas-kinetic collision frequency to chemical reaction rates

Ask the tutor: How does this collision density connect to the mean free path formula, and how would you use both to estimate a gas's viscosity?

Why is the cross-section pi d squared rather than pi (d/2) squared?

Collision occurs when the centers of two spheres, each of diameter $d$ , approach within a distance $d$ of each other — that is when their surfaces touch. The effective "target circle," as seen by an incoming molecule, therefore has radius $d$ , not $d/2$ , giving area $\sigma = \pi d^2$ . If you instead used each molecule's own geometric cross-section $\pi(d/2)^2$ , you would undercount collisions by a factor of four, since you would be ignoring that the target's own finite size also contributes to the collision radius.

How does collision frequency relate to the mean free path?

The mean free path $\lambda$ is the average distance a molecule travels between collisions. Since a molecule moving at $\bar v$ suffers $z$ collisions per second, the time between collisions is $1/z$ , so $\lambda = \bar v/z = 1/(\sqrt2\,\sigma\mathcal N)$ . This connection, covered in the mean free path topic, shows that collision frequency and mean free path are two faces of the same kinetic picture — one measures rate, the other measures distance.

Does the hard-sphere model overestimate or underestimate real collision rates?

Real intermolecular potentials are soft, with attractive tails (van der Waals forces) that can enhance the effective cross-section at low temperature, and repulsive cores that are not perfectly hard. The hard-sphere diameter $d$ is typically fit to viscosity or diffusion data at a reference temperature, so it implicitly captures some of this softness, but a single fixed $d$ cannot be exact across all temperatures. More refined models use temperature-dependent effective cross-sections from realistic potentials like Lennard-Jones.

Why does Z_AB not include the factor of one half found in Z_AA?

The factor of one half in $Z_{AA}$ exists purely to prevent double-counting: multiplying $z$ (per-molecule rate) by $\mathcal N$ (number of molecules) counts each A–A collision from both partners' perspectives. In an A–B mixture, an A molecule colliding with a B molecule is never double-counted this way, because A and B are distinguishable — the product $\mathcal N_A \mathcal N_B$ already counts each distinct pair exactly once.

How is collision density used beyond kinetic theory itself?

Collision density is the geometric core of collision theory for bimolecular reaction rates, where the rate constant is written as $Z_{AB}$ times a Boltzmann activation factor times a steric factor. It also underlies transport property estimates, since viscosity, thermal conductivity, and diffusion coefficients all depend on how frequently molecules exchange momentum, energy, or identity through collisions — the general transport framework builds directly on $\sigma$ , $\bar v_{rel}$ , and $\mathcal N$ introduced here.