Let's say we have a box with a non-permeable wall separating the box in half. There is gas on the other side of the wall. Now we remove the wall so that the gas can diffuse to the other half of the box.

enter image description here

It is said that the entropy of the gas increases because the molecules now have more room to move, and therefore there are more states that the gas can be in. I can understand this well.

But the change in entropy is also defined as follows:

$\displaystyle \Delta S = \frac{Q}{T}$

Where $T$ is the temperature of the gas and $Q$ is the change in heat of the system. But if we look at this definition, why did the entropy change for the gas inside the box? By just removing the wall, the kinetic energy of the molecules does not change, therefore the temperature does not either. We also didn't add any heat to the system, so $Q$ is zero as well. So why did the entropy change?

  • $\begingroup$ You might enjoy how The Feynman Lectures (46-4 Irreversibility) analyzes a similar situation: removing the barrier in a box that has "white" molecules on one side and "black" molecules on the other side. $\endgroup$
    – Ron Burk
    Jan 19 at 10:10

This is a classic example of making sure you know what your equations actually mean. You are thinking along the lines of $$Q=T\Delta S$$ Where your change in entropy determines the heat exchange. This is not the right way to view the equation.

The actual meaning of the equation is "if you have reversible heat flow $Q$ at temperature $T$ for a reversible process, then you have a change in entropy $\Delta S$". Since there is no heat flow, this equation does not tell you anything useful. All you can say is there is no entropy change due to a heat flow.

  • 3
    $\begingroup$ better to say: " The actual meaning of the equation is "if you have reversible heat flow Q at temperature T in a reversible process then you have a change in entropy ΔS"" $\endgroup$
    – hyportnex
    Oct 17 '18 at 13:24
  • $\begingroup$ @Aaron Stevens In your answer, reversible means internally reversible which in turn means, there is no 'unaccounted heat transfer' or any other dissipation effect.Am I understanding it right(its difficult for me to understand reversible processes)? $\endgroup$
    – Mohan
    Oct 19 '18 at 16:48

The formula you used, $$\mathrm{d} S=\frac{\mathrm{d}Q_{\mathrm{reversible}}}{T}$$is a simplification for situations where all other factors are constant. Here, you're better off using $$\Delta S=nR\ln\frac{V_2}{V_1}$$ ($R$ is the gas constant, and $n$ is the number of moles) It's a more sensible formula since the only thing you're changing is volume: the temperature and energy are both maintained as constants.

Eventually, the entropy did change, but the formula you used wasn't the right one, and hence it suggested that entropy wouldn't change.

The Wikipedia page on entropy has a list of useful formulas, you'll need to choose the appropriate one for different situations, depending on what factors you're maintaining as constants.

  • 1
    $\begingroup$ Your first formula should have a $T$ not a $\Delta T$. The issue the OP is making is simply that in this formula $\Delta Q$ is the heat transfer to do the process reversibly (I like to write $\Delta Q_{rev}$ whereas this process is irreversible. If they found a reversible process with the same endpoints, their formula would work. $\endgroup$
    – jacob1729
    Oct 17 '18 at 14:54
  • $\begingroup$ @jacob1729 yeah, the reversible thing is a good way to describe the situations where that formula is relevant. I'll make that edit in a moment. Also, now that you mentioned it, when I compare this to the usual calculus version that I like using, the $\Delta T$ for temperature does seem inappropriate. $\endgroup$
    – user191954
    Oct 17 '18 at 14:56
  • $\begingroup$ Interesting! About the second formula you use, $\Delta S=nR\ln\frac{V_2}{V_1}$: I looked this formula up on Wikipedia article on Joule expansion. It involves integrating pressure $P$ with respect to the volume. But if we integrate pressure with respect to volume, doesn't this mean we did work? On my lecture notes there is even a question: "If a gas inside a box is allowed to expand into the whole box adiabatically without doing work, how does the entropy change?". There correct answer is that it increases. But how is this formula correct when the gas expands into vacuum without doing work? $\endgroup$
    – S. Rotos
    Oct 26 '18 at 17:06

A statistical mechanics perspective. The phase space of an atom in the gas is described by the microstates, made from position and momentum values, $(x, y, z, p_x, p_y, p_z)$.

The allowed position states are constrained by the dimensions (l, w and d) of the box, $0<x<l$, $0<y<w$, $0<z<d$.

When you increase the box size, you have increased $W$, the number of microstates and therefore increased the entropy of the gas, according to the Boltzmann entropy definition,

$S=k_B \ln W$


All previous answers get close. The definition of $S$ in thermodynamics (the statistical mechanics view is more straightforward in this case, but this was not the question) is: $$dS = dQ_{rev} / T$$ which means that the infinitesimal change in entropy going from state A to state B is tantamount to the heat $dQ$ exchanged along any reversible transformation going from state A to state B, divided by the temperature $T$ at which this happens. Since in this case we are going to a final state which has the same temperature as before, like you said, it is smart to directly use an isotherm curve from $V1$ to $V2$. Now, in an isotherm going from $V1$ to $V2$ the heat exchanged is: $$\Delta Q_{rev}=nRTln(V2/V1)$$ where $R$ is the gas constant (see wikipedia for derivation if this is not clear) so that in our case, integrating $dS$

$$ \Delta S=\Delta Q_{rev}/T = nRln(V2/V1)$$ which is the formula Chair used - but it is not a magical formula, it comes from your definition. Your definition is NOT a simplification (you wrote the integrated version, the right one is the "differential" one, but whatever), it is the right definition! The tricky point is that $Q$ is not the heat of your process, but the heat of $any$ process starting and ending in the same states you have. Then you just need to find a smart way to compute it.

More in general, since you can always reach two points of the PV diagram with two transformation, an isothermal one and a constant-volume one, you can always write the entropy of a perfect gas going from A to B as:

$$n C_v ln(TB/TA)+nRln(VB/VA)$$ where $C_v$ is the specific heat of the gas at constant volume. This formula is again a general formula (amongst many others for the perfect gas) but it can be easily derived from your definition!

The entropy change does not depend on the actual heat exchanged in the transformation, it comes from the simple fact that you are changing state. However, to compute it, we use the heat along a different path. That is all!


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.