I'd like to derive the $S(U,V,N)$ function. In the lecture, we were using the $S=k_b \log(\Omega)$, some combinatorics, approximation, $6N$ dimensional spheres, etc. But I'd like to avoid that if it's possible. So I think I should do it from the first and second law of thermodynamics: $$\mathrm{d}U=T\mathrm{d}S-p\mathrm{d}V+\mu\mathrm{d}N$$ $$\mathrm{d}S=\frac{1}{T}\mathrm{d}U+\frac{p}{T}\mathrm{d}V-\frac{\mu}{T}\mathrm{d}N$$ And with $pV=Nk_bT$ and $U=\frac{f}{2}Nk_b T$, I could get that $$\mathrm{d}S=\frac{f}{2}k_b \frac{N}{U}\mathrm{d}U+k_b\frac{N}{V}\mathrm{d}V-\frac{\mu}{T}\mathrm{d}N$$ But I can't get rid of the $\frac{\mu}{T}$. How could I do it?

  • $\begingroup$ Just compute the partition function. For an ideal gas it’s a trivial Gaussian integration. $\endgroup$
    – JamalS
    Commented Jun 12, 2018 at 9:53
  • $\begingroup$ @JamalS I never heard of that. I think I will learn about that in the statistical physics class. $\endgroup$
    – Botond
    Commented Jun 12, 2018 at 10:05
  • 1
    $\begingroup$ Try to use some Maxwell relations maybe? $\endgroup$ Commented Jun 12, 2018 at 11:24
  • $\begingroup$ That third equation is likely derived from, or maybe just a differnt form of the Gibbs-Durham equation - it can further be modified for heat per unit particle.. When you do this properly, you can also find the state of the entropy. $\endgroup$ Commented Apr 13, 2019 at 2:12

1 Answer 1


From the definition of chemical potential:

\begin{equation}\mu = \left(\frac{\partial U}{\partial N}\right) = cRT \end{equation} where c = 3/2 for a monatomic ideal gas, c = 5/2 for a diatomic ideal gas.

To verify the consistency of the derivation:

From the “fundamental equation” of an ideal gas:

\begin{equation} PV = N RT \end{equation}

\begin{equation} U=c N RT\end{equation}

if you rewrite the equations of state as

\begin{equation} \frac{1}{T}=\frac{cN R}{U}=(∂S/∂U)_{V,N}\end{equation}

\begin{equation} PT=\frac{NR}{V}=(∂S/∂V)_{U,N} \end{equation}

By integrating both, the first equation gives:

\begin{equation} S(U, V, N) =c N R \ln(U) +f(V, N)\end{equation} and the second

\begin{equation} S(U, V, N) =N R \ln(V) +g(U, N) \end{equation}

Which are consistent only if we can write S in the form:

\begin{equation}S(U, V, N) =cN R \ln(U) +N R \ln(V) +f(N) \end{equation}

where $f(N)$ is a function of N alone namely:

\begin{equation}f(n) = c RN \end{equation}

which accounts also for the fact that $S$ must be extensive in $U, V$, and $N$.

the arguments of the logarithms must be dimensionless (otherwise, the logarithm doesn’t make sense)

\begin{equation}S(U, V, N) =cN R \ln \frac{U}{Nk_1} +N R \ln \frac{V}{Nk_k2} +f(N) \end{equation}

Here $k_1$ and $k_2$, are constants with the dimensions of energy per mole, volume per mole


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.