Derivation of Bose
Distribution
This document will try to document the way we compute the
probability of occupation by n indistinguishable particles of the ith energy
level which has g sub-levels. I have
found that the best way to keep track of the occupations of the various g
sub-level populations is to assign a number to each sub level and then tabulate
the sums of these numbers for all possible indistinguishable sets of sums. This can be motivated by assuming that each
of the sub-levels has a very small magnetic energy splitting and then the sum
of these energy values will distinguish it from other sub-levels. In the end, we can simply remove the
splitting but the sub-level distribution will
remain.
The number of different sums we can get for n particles
assigned to levels with g degeneracy is the binomial factor
(1.1)
If the Bose character of the particle is to be interesting,
then we should have many more particles than there are degeneracies, i.e.
n>>g. The opposite situation is
when n<<g very low density gas
which characterizes Maxwell-Boltzmann statistics.
The total number of ways that all N indistinguishable
particles with ni in each level with
gi sub-levels or degeneracies is the product of all the
binomial factors for the ni+gi-1
terms
|
|
(1.2)
|
Use Stirling's approximation for the
factorials:
|
|
(1.3)
|
|
|
(1.4)
|
The number of particles, ni,
in each energy level εi
will be such that the value of W is maximized taking into account that the
total number of particles is constant at N and the total energy is constant at
E.
We wish to find the values of each ni for which W is a maximum taking into account that the
total number of particles, N, is constant and the total energy is
constant. To find ni while
keeping N and E constant, we use Lagrange multipliers for the particle sums and
the total energy. Then the equation from which to find ni becomes:
|
|
(1.5)
|
It is easier to find the maximum if we first take the natural
logarithm of the expression for w.
Using equation (1.4) in
equation (1.5)
we obtain:
|
|
(1.6)
|
where I've dropped the square root terms. If we also make the assumption that ni>>gi
then we can further simplify the above expression. Let’s rename the factorial expression on the
left to be w(ni,gi)
|
|
(1.7)
|
We will need the derivative of wi with respect to ni:
|
|
(1.8)
|
Including the particle sums and energy sum derivative terms
in equation (1.5)
we have:
|
|
(1.9)
|
taking the exponential of both sides we have:
|
|
(1.10)
|
Solving for ni we
obtain:
|
|
(1.11)
|
Equation (1.11) has the form of the Bose distribution and is
better recognized with the substitution
|
|
(1.12)
|
so that we have:
|
|
(1.13)
|