1
$\begingroup$

Imagine $N$ oscillators with only two possible energies, $\epsilon_0$ and $ \epsilon_1$, with $\epsilon_1 > \epsilon_0$. Taking $\epsilon_0 = 0$ for now

I showed $\Omega(q\epsilon_1) = \frac{N!}{(N-q)!q!}$ and then

$$\frac{\partial S}{\partial q} = k \log(N/q - 1) $$

How can I use the above equation to show that

$$U = N\epsilon_1\frac{e^{-\epsilon_1/(kT)}}{1+e^{-\epsilon_1/(kT)}} $$

I tried moving the $\partial q$ over to the right, and then have $dS = dU/T$, but i wasn't getting anything meaningful.

$\endgroup$

1 Answer 1

2
$\begingroup$

You should use $U=q\epsilon_1$.

With the total number of particles $N$ being constant, we have: $$\frac{\partial S}{\partial q}=\epsilon_1 \frac{\partial S}{\partial E}=\frac{1}{T}\epsilon_1\tag{1}$$ As you said: $$\frac{\partial S}{\partial q}=k_B\ln(N/q - 1)=k_B\ln(N\epsilon_1/q\epsilon_1 - 1)=k_B\ln(N\epsilon_1/U - 1)\tag{2}$$ $$\to \,\,\,\,\,U(T)=N\epsilon_1\frac{e^{-\epsilon_1/(kT)}}{1+e^{-\epsilon_1/(kT)}}$$

$\endgroup$
2
  • $\begingroup$ and if $\epsilon_0$ does not equal 0, do I just repeat but with $U = q(\epsilon_0 + \epsilon_1)$? $\endgroup$
    – user34032
    Feb 14, 2014 at 18:15
  • $\begingroup$ No, $U=\sum_iq_i\epsilon_i$. In that case you'd better find the (canonical) partition function and calculate $U$ from that. $\endgroup$
    – Mostafa
    Feb 14, 2014 at 18:58

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.