Given that the probability to find a particle in a state $|i\rangle$ is $$ p_i \propto \exp(-\beta E_i) $$ one can conclude that for classical particles (:=particles obeying the classical kinetic energy relation) with mass $m$ the velocity distribution is given by $$ P(|\mathbf{v}|)=\sqrt{\frac{2}{\pi}} {(\beta m)}^{3/2}\exp\left(-\beta m \frac{|\mathbf{v}|^2}{2}\right) $$ Now consider a cubical box with Volume $L^3$. Which pressure exert $N$ classical particles confined in this box on the six identical walls of the confinement?

One can arrive at the answer $$ p=\frac{N}{\beta V} $$ by assuming that

  1. There are no interactions between particles
  2. Collisions with the walls are elastic and that the impulse of the particle after a collision with a wall is $\mathbf{p'}=\mathbf{p}-2\mathbf{p}\mathbf{n}$. Where $\mathbf{n}$ is normal to the wall.
  3. Every particle is traveling for a distance $L$ before it collides with a wall. Particles only travel in a straight path from Side $S_i$ to the corresponding opposite side $S_i'$. They never travel from one side to an adjecent side.

Now the logic is easy. Force exerted on one side of the cube by one particle with velocity $v$ is just $$ F(v)=\frac{\Delta p}{\Delta t}=\frac{2mv}{2L/v} $$ Force exerted on the side by one particle on average $$ \langle F\rangle=\frac{2m}{2L}\langle v^2\rangle=\frac{m}{L}\int_{\mathbb{R}^+}v^2P(v)dv=\frac{3}{\beta L} $$ Which leads to the total pressure exerted on this side $$ p=\frac{N}{3}\frac{\langle F\rangle}{A}=\frac{N}{\beta V} $$ Which by symmetry is the same for all six sides.

Note that not $N$ but $N/3$ particles travel from one side to it's opposite side, since there are six sides in total.

Can someone arrive at the result without using the third assumption, that I made up? I would consider this as a purely geometric problem and can not think of a clever way to tackle it. I want a treatment that also allows particles to travel from one side to one of it's adjacent side.


We assumed that the particles are non-interacting. Let us pick a particle at random. This particle has with $p_i$ probability an energy of $E_i$. Assume this energy is also a given. Now, we have a substitute assumption for your third:

  1. Any particle has equal probability of being headed to any direction. In $\Bbb R^3$, this means that probability that a particle is headed to a direction is $dP(\theta+d\theta,\phi+d\phi)=\sin(\theta)\,d\theta\,d\phi\ /\ 4\pi$.

This last assumption is not really an assumption than a symmetry argument. Now assume we have a particle with velocity $\mathbf v$. This particle's contribution to the total pressure on the cube's upper face (that is, $(x,y,L)$) is,

$$p_z = \frac{m (\mathbf v\cdot\hat z)^2}{L} = \frac{m}{L}\,(v\cos(\theta))^2$$

This follows since this particle has speed $v_z=\mathbf v\cdot\hat z$ on $+\hat z$ direction, and it goes through a total path (along $\hat z$) of $L$ in the process.

We calculated the probability of this event above.

$$p=\langle p_z \rangle=\int p_\, dP(\theta+d\theta,\phi+d\phi) = \int_0^\pi \frac{mv^2}{L}\,\cos^2(\theta)\,\sin(\theta)\,d\theta\ /\ 2$$ $$p=\frac{mv^2}{6L}\,(\cos^3(0)-cos^3(\pi)) = \frac{mv^2}{3L}$$

The deal is, assuming only $N/3$ particles are moving in $\hat z$, and the ones that do have velocities only on $\hat z$ is identical to the calculations made by assuming all particles ($N$) have an arbitrary velocity vector.

  • $\begingroup$ Thank you, this is what I was looking for. But you say "This follows since this particle has speed $v_z=v⋅\hat{z}$ in $+\hat{z}$ direction, and it goes through a total path (along $\hat{z}$) of $L$ in the process." Why is the path along $\hat{z}$ equal to $L$? This assumes that the particle starts at the bottom of the box. On average this is not the case though. $\endgroup$ – TheoreticalMinimum Nov 17 '19 at 20:57
  • $\begingroup$ I think I answered this last point for myself: Pick a random particle with speed $v$ and look at it after a collision where the upper wall received an impulse of $\Delta p=2p_z=2mv_z=2mv\cos(\theta)$. The totally elastic collision did not change the velocity $v$. How long does the particle need to again hit the upper wall? The answer is $\Delta t=\frac{2L}{v_z}=\frac{2L}{v*cos(\theta)}$, since the particle has to travel down to the wall on the floor and than back up again, a total $\hat{z}$-distance of $2L$. So the force on the upper wall is indeed $F=\frac{\Delta p}{\Delta t}=mv_z^2/L$ $\endgroup$ – TheoreticalMinimum Nov 17 '19 at 21:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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