next up previous
Next: General Case: How to Up: Ensembles and Thermodynamic Potentials Previous: Entropy, Energy and Free

Grand Canonical Ensemble

We have both energy E and the number of particles N vary!
\begin{figure*}
 \InputIfFileExists{grand_canonical.pstex_t}{}{}\end{figure*}
The same considerations as for canonical ensemble (substitute N for E).

Probability of a state:

\begin{displaymath}
\Prob(A_{i,N}) = \frac{1}{\Xi}\exp(-E_i/kT + \mu N/kT)\end{displaymath}

Here

\begin{displaymath}
\Xi = \sum_{i,N} \exp(-E_i/kT+\mu N/kT)\end{displaymath}

is called grand partition function , and

\begin{displaymath}
\mu = -\frac{1}{T}\left(\frac{\partial S}{\partial N}\right)_{E,V}\end{displaymath}

is called chemical potential . It is an intensive variable (why?)

Another formula:

\begin{displaymath}
dS = \frac{1}{T}dE - \frac{\mu}{T}dN +\frac{p}{T}\,dV\end{displaymath}

so

\begin{displaymath}
dE = \mu\,dN + T\,dS -p\,dV,\quad \mu =\left(\frac{\partial
 E}{\partial N}\right)_{S,V}\end{displaymath}

Yet another formula: since A=E-TS,

\begin{displaymath}
dA = \mu\,dN - S\,dT -p\,dV, \quad \mu =\left(\frac{\partial
 A}{\partial N}\right)_{T,V}\end{displaymath}

Averages:

\begin{displaymath}
\left\langle f\right\rangle = \frac{1}{\Xi}\sum_{i,N} f_{i,N} \exp(-E_i/kT+\mu N/kT)\end{displaymath}

$\Omega$-potential (compare to free energy A!)

\begin{displaymath}
\Omega = -kT\ln\Xi\end{displaymath}

Average entropy:
\begin{multline*}
\left\langle S\right\rangle = -k\sum_{E,N} \Prob(E,N)\ln\bigl(...
 ...gle E\right\rangle}{T} - \frac{\left\langle N\right\rangle\mu}{T}\end{multline*}
Or:

\begin{displaymath}
\Omega = E - TS - N\mu\end{displaymath}

$\Omega$-potential is a Legendre transform of E!  
 \begin{displaymath}
 d\Omega = -S\,dT-P\,dV-N\,d\mu\end{displaymath} (1)
So $\Omega=\Omega(V,T,\mu)$
Consequence:
$\Omega$-potential is just the product of volume and pressure:

\begin{displaymath}
\Omega = -PV
 \end{displaymath}

Proof:
In thermodynamic limit

\begin{displaymath}
T\to\mathit{const},\quad \mu\to\mathit{const},\quad \Omega\sim V\to\infty
 \end{displaymath}

Therefore

\begin{displaymath}
\Omega = \omega(T,\mu)V
 \end{displaymath}

Then

\begin{displaymath}
\omega = \left(\frac{\partial\Omega}{\partial V}\right)_{T,\mu}
 \end{displaymath}

and from (1)

\begin{displaymath}
\omega = -P
 \end{displaymath}



© 1997 Boris Veytsman and Michael Kotelyanskii
Tue Sep 9 22:39:08 EDT 1997