Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have got so far the 2D density of states as $g(\epsilon)=\frac{Am}{\pi\hbar^2}$ where $A$ is the area of the "square" and $m$ is the the electron mass. Then I have found an expression for the the chemical potential for the gas using: $N={\int_0^\infty}g(\epsilon)n_Fd\epsilon$, where $n_F$ is the fermi distribution function. This spits out $N=\frac{Am}{\pi\hbar^2}k_BT\ln(1+e^{\frac{\mu}{k_BT}})$, and taking the limit of $T$ goes to zero we get $\epsilon_F=\frac{N\pi\hbar^2}{Am}$, so $\mu=k_BT\ln(e^{\frac{\epsilon_F}{k_BT}}-1)$. This is where I start to get a little stuck. I am thinking I should next work out $<E>=\int_0^\infty{n_F}{\epsilon}g(\epsilon){d}\epsilon$ However I think this is going down the wrong route (or I can't get the algebra right) as I have a hint to use the result $\int_{-\infty}^{\infty}dx\frac{x^2e^x}{(e^x+1)^2}=\frac{\pi^2}{3}$. If anyone has a hint on where I should be going then it'd be much appreciated!

share|cite|improve this question
Do you want this "at" T = 0, or at T small, but nonzero? If it's the latter, you want to look up "Sommerfeld expansion". – Lagerbaer Apr 9 '13 at 17:30
Low T limit, couple of things confusing me if you could take a quick look below – Dmist Apr 10 '13 at 13:29
up vote 1 down vote accepted

The procedure to obtain the specific heat of a free electron gas in arbitrary dimension is related to the so-called Sommerfeld expansion that is applied to integrals of the form

$$I(\mu,T)=\int_{-\infty}^{+\infty} d\epsilon \, f(\epsilon) n_F(\epsilon) $$

where $n_F(\epsilon)$ is the Fermi-Dirac distribution and $f(\epsilon)$ is an arbitrary function which vanishes at $\epsilon \rightarrow -\infty$ and diverges as a power law when $\epsilon \rightarrow +\infty$.

This technique uses the fact that the derivative of the Fermi function is peaked around the chemical potential $\mu$ and broadened by $k_BT$ [and indeed collapses to $-\delta(\epsilon-\mu)$ at $T=0$]. Therefore, if the function behaves as above, which is true for $f(\epsilon)=\epsilon g(\epsilon$), we can integrate by parts to get

$$\int_{-\infty}^{+\infty} d\epsilon \, f(\epsilon) n_F(\epsilon)=-\int_{-\infty}^{+\infty} d\epsilon \, F(\epsilon) \dfrac{\partial n_F}{\partial \epsilon}$$ where $$ F(\epsilon)=\int_{-\infty}^{\epsilon}d\epsilon' f(\epsilon')$$ and expand $F(\epsilon)$ around $\epsilon=\mu$ in Taylor series $$F(\epsilon)=F(\mu)+\sum_{k=1}^{+\infty} \dfrac{F^{(k)}(\mu)}{k!}(\epsilon-\mu)^k$$ with $F^{(k)}(\mu)$ being the $k$-th derivarive of the function $F$. Observing that $(-\partial n_F/\partial \epsilon)$ is normalized to unity and that is an even function (so that only terms even in $k$ contribute) we can go back to the original function $f(\epsilon)$ and using $x=\beta(\epsilon-\mu)$ $$I(\mu,T)=\int_{-\infty}^{\mu}d\epsilon f(\epsilon)+\sum_{k=1}^{+\infty} \dfrac{f_k}{\beta^{2k}}\dfrac{d^{2k-1}}{d\epsilon^{2k-1}}f(\epsilon)\Bigg|_{\epsilon=\mu}$$ with coefficients $$f_k=\int_{-\infty}^{+\infty} dx \dfrac{x^{2k}}{(2k)!}\left(-\dfrac{d}{dx} \dfrac{1}{e^x+1}\right)dx$$ Using this you should not have any problem to find the mean kinetic energy for the free 2DEG and from there the specific heat.

share|cite|improve this answer
Okay, after looking at it for quite a while I'm pretty happy bar a couple of things, firstly the function $f(\epsilon)$ should vanish for epsilon goes to minus infinity, however in this case $f(\epsilon)=g(\epsilon)\epsilon$ and the density of sates is just a constant, so f diverges at minus infinity? Also, I'm confused why the integral's upper limit can be changed to $\mu$ (or even to $\epsilon_F$ as I've seen in some places). Is it that the fermi function goes quickly to zero within a couple of $k_BT$ of the Fermi energy (or chemical potential) and T is small so limit is pretty much $\mu$ – Dmist Apr 10 '13 at 12:23
The DOS is not constant everywhere. It's zero below the single particle ground state energy, which we set to 0. So $f(\epsilon) = 0$ for $\epsilon <0 $. It's only constant above that. Next, in the limit of very low $T$, if you do the expansion, you'll find that $\mu = \epsilon_F$. I think there's a typo in the post in the definition of $F(\epsilon)$. I think it should read $F(\epsilon) := \int_{-\infty}^\epsilon d\epsilon' f(\epsilon')$. – Lagerbaer Apr 10 '13 at 15:47
@Lagerbaer: Yep, there is a typo. I fixed it. Thanks! – DaniH Apr 10 '13 at 17:07
@Dmist: I forgot to mention that the integral you give as hint appears at first order in the Sommerfeld expansion. – DaniH Apr 10 '13 at 21:32
Ahh okay, so we set it to zero for $\epsilon$ below zero but that doesn't matter anyway as we can neglect any terms which don't depend on T as that's all we're interested in for the heat capacity. I get out $C/N=\frac{(\pi{k_B})^2N}{6\epsilon_F}T$ which gives the total heat capacity of a metal the required form of $AT^3+BT$ (I am kind of assuming this dependence on T is general for d dimensions) – Dmist Apr 11 '13 at 10:48

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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