6
$\begingroup$

In Wikipedia's article on Fermi Gases, they have the following equation for the chemical potential:

$$\mu = E_0 + E_F \left[ 1- \frac{\pi ^2}{12} \left(\frac{kT}{E_F}\right) ^2 - \frac{\pi^4}{80} \left(\frac{kT}{E_F}\right)^4 + \cdots \right]$$ where $E_0$ is the potential energy per particle, $k$ is the Boltzmann constant and $T$ is the temperature.

I don't understand how they get the third term, particularly the 1/80 factor. I've often seen this equation expressed just to the tau^2 term, and I understand how to get the 1/12 factor from a binomial expansion of $$\left( 1 + \frac{\pi^{2}}{8} \left(\frac{\tau}{E}\right)^{2}\right)^{-2/3} $$

However, I've tried continuing the binomial expansion and cannot figure out why the factor is 1/80.

$\endgroup$
2
  • $\begingroup$ Sorry about the erroneous edit - our bad. (+1 btw :) $\endgroup$
    – user10851
    Commented Jul 1, 2013 at 6:37
  • $\begingroup$ No worries; I should've written it in Latek to start. Unfortunately I had somewhere I needed to be. $\endgroup$
    – MattS
    Commented Jul 1, 2013 at 6:40

1 Answer 1

2
$\begingroup$

I not sure how you obtained the last expression. The standard Sommerfeld expansion (for details, see e.g. Ashcroft & Mermin) gives a slightly different result, which is $$ E_{F} \approx \mu\left[1+\frac{\pi^{2}}{8}\left(\frac{k_{B}T}{\mu}\right)^{2}\,\right]^{2/3} \approx \mu\left[1+\frac{\pi^{2}}{12}\left(\frac{k_{B}T}{\mu}\right)^{2}\right] $$ to leading non-trivial order in $k_{B}T/\mu$, i.e., $O\big((k_{B}T/\mu)^{2}\big)$. (I have set $E_{0}=0$ in the expression from Wikipedia.)

We can invert this relation by substituting $\mu = E_{F}\left[1 + c_{2} (k_{B}T/E_{F})^{2} + \cdots \right]$ into the above. That is,

$$ E_{F} = E_{F}\left[1 + c_{2} (k_{B}T/E_{F})^{2} + \cdots \right]\left\{1+\frac{\pi^{2}}{12}\left(\frac{k_{B}T}{E_{F}}\right)^{2}\left[1 + c_{2} (k_{B}T/E_{F})^{2} + \cdots\right]^{-2}\right\}. $$

Comparing the zeroth order terms in $k_{B}T/E_{F}$ on both sides of the above equation, we simply obtain $E_{F}=E_{F}$. Comparing the second order terms, we have $0 = \frac{\pi^{2}}{12} + c_{2}$. Hence

$$ \mu = E_{F}\left[1-\frac{\pi^{2}}{12}\left(\frac{k_{B}T}{E_{F}}\right)^{2}\right] $$ up to $O\Big((k_{B}T/E_{F})^{2}\Big)$.

To determine the next-order correction to $\mu$, you should include one higher order term in the Sommerfeld expansion, which gives

$$ E_{F} \approx \mu\left[1+\frac{\pi^{2}}{8}\left(\frac{k_{B}T}{\mu}\right)^{2}+\frac{7\pi^{4}}{640}\left(\frac{k_{B}T}{\mu}\right)^{4}\,\right]^{2/3}. $$

Expanding this up to $O\big((k_{B}T/\mu)^{4}\big)$, then substituting $\mu = E_{F}\left[1 + c_{2} (k_{B}T/E_{F})^{2} + c_{4} (k_{B}T/E_{F})^{4}+\cdots\right]$ (where we have already determined $c_{2}$ before) into it, and then matching the coefficients of both sides up to $O\Big((k_{B}T/E_{F})^{4}\Big)$ will lead to the desired result.

$\endgroup$
15
  • $\begingroup$ Where does $\mu = E_{F}\left[1 + c_{2} (k_{B}T/E_{F})^{2}\right]$ come from? $\endgroup$
    – MattS
    Commented Jul 1, 2013 at 9:02
  • $\begingroup$ I'm expanding $\mu/E_{F}$ in powers of a small quantity $k_{B}T/E_{F}$, and we know that it is equal to one at zeroth order because $E_{F}=\mu$ at $T=0$. $\endgroup$
    – higgsss
    Commented Jul 1, 2013 at 9:11
  • $\begingroup$ I'm sorry, I'm still not quite following. I'm trying to repeat your analysis and get to $\mu = E_{F} \left[1 - \frac{\pi^{2}}{12}\left(\frac{k_{B}T}{E_{F}}\right)^{2}\right]$, but I'm not getting it. $\endgroup$
    – MattS
    Commented Jul 1, 2013 at 9:13
  • $\begingroup$ Substitute $\mu = E_{F}\left[1 + c_{2} (k_{B}T/E_{F})^{2} + \cdots \right]$ into $E_{F} = \mu \left[1 + \frac{\pi^{2}}{12} \left(\frac{k_{B}T}{\mu}\right)^{2} \right]$ and match the coefficient of both sides up to $O\Big((k_{B}T/E_{F})^{2}\Big)$. $\endgroup$
    – higgsss
    Commented Jul 1, 2013 at 9:23
  • $\begingroup$ I tried that, and I didn't achieve anything. What is $O\big((k_{B}T/\mu)^{2}\big)$? $\endgroup$
    – MattS
    Commented Jul 1, 2013 at 9:27

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.