T1 = T2
We also saw that the average kinetic energy of a system was related
to the temperature via
![]()
where kB is called Boltzmann's constant and has a
value of kB = 1.381 x 10-23 J/K. Recall that
(3.8) is the condition of thermal equilibrium. Thus, we should be able
to relate it to temperature. Noticing that the energy is involved in the
denominator of the relationship, we are lead to the definition
|
|
(4.1)
|
We call t the fundamental temperature.
It is related to the normal temperature in Kelvin by
t = kB T
Notice that the fundamental temperature has units of energy. This allows
us to relate our definition of entropy to the classical one
where S is the classical entropy.
Example:
Assume that Ul > U2,
and that an amount of energy DU is extracted
from S1 and placed in S2.
Then the total entropy change Ds is
![]() |
(4.2)
|
Since Ul > U2, we have t1 > t2, and so the quantity on the right is positive, thus showing that the total change of entropy is positive when energy flows from a hotter system to a cooler one.
U = Ul + U2 = U1,0 + U2,0
The multiplicity function is then
![]()
and contains, as one of the accessible states, the original state g(U1,0)
g(U2,0). Since there are other states also accessible
now, we see that, in general, g(U) ³
g(U1,0)g(U2,0). Recall
that the entropy is defined by s = ln g(U),
so this conclusion leads to the law of increase of entropy
|
|
(4.3)
|
Suppose that dU is the uncertainty
in U. We can look at the density of states in a given system. Let
D(U) be the number of states per unit interval of energy.
Then
g(U) = D(U) dU
and
s(U) = ln D(U) + ln dU
In many cases, we find that the total number of states is proportional
to 2N. If the total number of states is of order N times
some average particle energy, D, then
.
For this case, we see that
s(U) = N ln(2) - ln(N) - ln(D) + ln(dU)
In most other cases, we find that the total number of states in a system
is proportional to UN dU.
So the entropy can be written as
|
|
(4.4)
|
Typically, the uncertainty in U will be less than 1. Thus, we see that in both cases, the first term, N ln(U) or N ln(2), will dominate the entropy.
