1056 25 Equilibrium Statistical Mechanics. I. The Probability Distribution for Molecular States
The most probable distribution can now be written
Nj
N
z
gje−βεj (25.3-4a)
The probability of finding a randomly selected molecule in molecule energy leveljis
equal to the fraction of molecules that are in this level:
pj
Nj
N
1
z
gje−βεj (probability of levelj) (25.3-4b)
The name “partition function” was chosen becausezis related to the way in which
the molecules are partitioned among the possible molecular states. The German name
for a partition function isZustandsumme, literally translated as “state sum” or “sum over
states.” This name is more descriptive than “partition function” and is occasionally used
in the English language, with or without translation. The German name is the reason
for using the letterzto denote the molecular partition function, but the letterqis also
used by some authors.
Each term in Eq. (25.3-3) is proportional to the number of states in that level, so we
can write the molecular partition function as a sum over states instead of a sum over
levels:
z
∑
i
e−βεi (sum over molecule states) (25.3-5)
The molecular partition function can be interpreted as a sum of probabilities of states,
with the probability of a state of zero energy set equal to unity. Its value can be thought
of as an effective total number of states accessible to a molecule under the given
conditions.
The Parameterβ
In order to determine what the parameterβrepresents, we substitute the expression for
Njgiven in Eq. (25.3-4a) into the constraint of Eq. (25.2-14b):
1
z
∑
j
εjgje−βεj〈ε〉
E
N
U
N
(25.3-6)
where we have again equated the system energy eigenvalueEto the thermodynamic
energyU. Equation (25.3-6) can be written in another way, using a mathematical trick
that identifies a derivative and interchanges the order of summing and differentiating.
U
N
1
z
∑
j
εjgje−βεj−
1
z
∑
j
∂
∂β
(
gje−βεj
)
−
1
z
∂
∂β
⎡
⎣
∑
j
gje−βεj
⎤
⎦−^1
z
∂z
∂β
−
(
∂ln(z)
∂β
)
V
(25.3-7)
The energy eigenvalues depend on the volume, soVmust be held fixed in the differ-
entiation. The interchange of summing and differentiating an infinite series is valid if
the series is uniformly convergent. This means that the series converges at least at a
given rate for all values of the variables involved. We assume this to be the case.