How would I prove that the number of collisions by gas molecules on a wall of unit area per unit time is $r = \frac 1 4 n \left< v \right>$ using the Maxwell-Boltzmann speed distribution function? $r$ is the rate of collisions, $n$ is the number of molecules per unit volume, and $\left< v \right>$ is the average speed of the molecules.

  • $\begingroup$ You don't need to actually know the distribution function for this, just that its isotropic. It might help if you add what $r$ is to your question (presumably rate of particle effusion through a hole in some container). $\endgroup$
    – jacob1729
    Commented Jun 8, 2018 at 18:28
  • $\begingroup$ What quantity is represented by $r$? (To my knowledge, $r$ is not a common sign for a thermodynamic quantity.) $\endgroup$ Commented Jun 8, 2018 at 18:28
  • $\begingroup$ I think $r$ is the rate of collisions of the gas molecules. $\endgroup$
    – Alex Wang
    Commented Jun 8, 2018 at 18:32

1 Answer 1


So imagine the entire space $z > 0$ is filled with this gas and you are looking at collisions on the plane $z=0$ in some time $t$. Considering the infinite space gives us translational invariance, so look at a small circle of area $dA$ centered at $(x, y) = (0, 0)$. Use spherical coordinates with $\phi$ being a polar angle down from $x=0, y=0, z>0$ and $\theta$ being an azimuthal angle counterclockwise from $y=0, x > 0,$ both measured in radians.

Somewhere out in space we have a small volume $dV = r^2\sin\phi~dr~d\phi~d\theta$ and it has a number of particles $dn = \rho~dV.$ What fraction of those pass through your square $dA$?

Well there's two concerns. One is the amount of solid angle occupied by the area $dA$. The particles at $(x, y, z) = (1, 0, \epsilon)$ for small $\epsilon$ see this area nearly edge-on and have a lot more trouble getting into it than the particles at $(x, y, z) = (0, 0, 1)$, which see it perpendicular. This boils down to a factor of $\cos \phi$, if you work it out.

In addition there's the more obvious concern that even if it were parallel, the chance of the particle having the right direction goes like $dA / 4\pi r^2.$

Finally it needs to have a velocity sufficient to traverse the distance in the given time. The easiest way to do this is with a cumulative distribution function: $F(s)$ is the probability that the speed $S$ is less than or equal to some value $s$, with $F(0) = 0$ and $F(\infty) \to 1.$ So you've got an expression for the rate $\eta$ of collisions, $$\frac{\eta}{dA} = \int_0^\infty dr~\int_0^{\pi/2}r~d\phi~\int_0^{2\pi}r~\sin\phi~d\theta~~\frac{\cos\phi}{4\pi r^2}~[1 -F(r/t)].$$You can immediately complete the angular integrals:$$\frac{\eta}{dA} = \frac{1}{2}~2\pi~\frac{1}{4\pi}~\int_0^\infty dr ~[1 - F(r/t)].$$So there's your mysterious $1/4$. As for the rest, you may first substitute $s = r/t, dr = t~ds $ to find $$\frac{\eta}{t~dA} = \frac14 \int_0^\infty ds~[1 - F(s)],$$ then integrate by parts, raising $ds \to s$ and lowering $[1 - F(s)] \to -f(s)~ds$ to find an expression $\int s~f(s)~ds = \langle S \rangle,$ and you just check that the boundary terms are both zero.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.