3
$\begingroup$

I am trying to calculate the internal energy of a free electron bas in a box in $D$ dimensions. To calculate the density of states, I used the following formula:

$$g(E) = \int \frac{d^Dk}{(2\pi)^D} \delta(E(k)- E(k'))$$

From this, I found that the density of states is $g(E) = \gamma E^{\frac{D}{2}-1}$, where $\gamma$ is a constant in terms of $D$, $\pi$, $\hbar$, etc. Using this, I plan to calculate the energy for finite temperatures. Hence, I used to the following equation:

$$U = \int Eg(E) f(E) dE$$

In the three dimensional case, I know for sure that $f(E)$ is the Fermi-Dirac distribution, which enables us to use the Somerfield expansion formula. In $D$ dimensions, does the following hold? $$f(E) = \frac{1}{e^{\beta(E - \mu)} + 1}$$

In other words, is there a $D$-dimensional version of the Fermi-Dirac distribution, or is the Fermi-Dirac distribution always of the form $$f(E) = \frac{1}{e^{\beta(E - \mu)} + 1}$$ regardless of dimension?

$\endgroup$
2
  • 3
    $\begingroup$ It's always of that form. $\endgroup$ Oct 13 '21 at 0:32
  • $\begingroup$ In your formula for the density of states, the Dirac distribution should be $\delta(E(\vec k)-E)$. $\endgroup$
    – Christophe
    Oct 13 '21 at 11:11
1
+50
$\begingroup$

Fermi-Dirac distribution depends only on the energy - it is a consequence of the grand-canonical ensemble applied to particles, obeying Pauli exclusion principle; a state of energy $E$ can be either occupied, with weight $e^{-\beta(E-\mu)}$ or empty with weight $1$ so that the partition function for this state is $$Z=e^{-\beta(E-\mu)}+1,$$ whereas the probabilities of the occupied and empty states are $$ P_1=\frac{1}{Z}e^{-\beta(E-\mu)}=\frac{1}{e^{\beta(E-\mu)+1}}=f(E),\\ P_0=\frac{1}{Z}=\frac{e^{\beta(E-\mu)}}{e^{\beta(E-\mu)+1}}=1-f(E). $$

The statistical physics arguments leading to the grand canonical distribution are independent on the number of dimensions - the only place where the dimensions appear is when neglecting the energy contributions due to the surface encolising the system, scaling as $R^{D-1}$, in comparison to the volume that scales as $R^D$. This works for arbitrary integer $D$ (although some peculiarities may appear in systems with $D=1$).

$\endgroup$

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.