9.6 Entropy (Young §20.7, Blundell §13.

7, chapter 14)
• Processes are irreversible when they increase the amount of randomness or disorder in the
universe.
• This is related to the number of configurations of the system, on a microscopic scale,
i.e., the number of microstates of the system.
• For example free expansion of a gas increases the amount of space the molecules can
occupy, so there are many more microstates of the system than there were before. Since
each microstate is equally probable (there is no reason to favour one over another), we
can say that the system is more disordered or random than before the expansion.
• The quantitative measure of disorder used in thermodynamics and statistical mechanics
is called entropy. It has the symbol S and it is a state property of a system. State
property because it depends on the state (not on how it got to that state).
• Entropy is an extensive property because if there are two identical systems, each with
entropy S, then the total entropy is 2S.
• But what should be the definition of entropy in terms of thermodynamic quantities?

9.6.1 Thermodynamic definition of entropy

|QC |
• Recall that the efficiency of an engine is given by e = 1 . For a Carnot engine,
QH
TC
recall that this expression can be shown to be equal to eCarnot = 1 .
TH
QC TC
• Since QC < 0, it follows that = for a Carnot cycle.
QH TH
QH QC
• Rearranging this, we have + = 0 for a Carnot cycle.
TH TC
X Qrev
• This can be written as , where the subscript “rev” stands for reversible process
T
cycle
(since the Carnot cycle is reversible).
• We can replace the sum by the integral around the Carnot cycle, which gives
I
d¯Qrev
= 0. (214)
T

• But we proved mathematically at the beginning of this course that an equation of that
form is true if and only if the same integral between any two limits is independent of path.
• Hence Z B
d¯Qrev
A T
is independent of the path taken between states A and B.
d¯Qrev
• Then is an exact di↵erential, which we can use to define the entropy:
T
d¯Qrev
dS = . (215)
T

109
 • This also implies that
Z B
d¯Qrev
S(B) S(A) = . (216)
A T
with S the entropy of the system.

• Note that S is a state function as it depends on the state but not on the path taken by
the system to arrive at that state.

• But in fact equation (214) is valid for any reversible cycle. To see this consider that any
reversible cycle can be constructed by combining many Carnot cycles, as in Fig. 53.

Figure 53: Young fig. 20.19. Any reversible cycle can be approximated by a series of Carnot
cycles. So equations (214), (215) and (216) are valid for any reversible process.

• The path integral over each Carnot cycle in Fig. 53 vanishes according to equation (214).
The path integral over the arbitrary reversible cycle also vanishes because it is equal to
the sum of all the small path integrals about the Carnot loops, each of which vanishes.

• Why? Because we always go around in the same direction (clockwise, say) and so the
middle portions (shown by the thin lines) make a positive contribution in one Carnot
cycle and a negative contribution in the next Carnot cycle, and these contributions exactly
cancel out. So we are left with just the thick line.

• By unsing arbitrarily many very “thin” Carnot cycles (isotherms arbitrarily close to-
gether), we can fit the cycle shown in Fig. 53 to arbitrary precision.

• Thus, equations (215) and (216) are valid for any reversible process.

• What do these results say about the change in entropy of the Universe during a reversible
process?

• Well, for the process to be reversible, the heat cannot be transferred across a finite tem-
perature di↵erence, so the system is in thermal equilibrium with the surroundings, at
temperature T .

• The system absorbs Qrev at (system) temperature T and the surroundings reject Qrev at
(surroundings) temperature T .

110
 • Hence, for a reversible process
Z B Z B
Qrev Qrev
SUni = Ssys + Ssur = = 0. (reversible process), (217)
A T A T
where the subscripts “Uni”, “sys” and “sur” stand for Universe, system and surroundings,
respectively.

• So for a reversible process the entropy of the Universe does not change.

9.6.2 Entropy change during an irreversible process and the 2nd law of TD in
terms of entropy
• Let’s consider what happens in an irreversible process by going through similar arguments
|QC | TC
except that now e = 1 < eCarnot = 1 .
QH TH
QC TC
• Since QC < 0, it follows that < .
QH TH
QH QC
• Rearranging this, we have + < 0.
TH TC
• Generalizing to a thermodynamic cycle (i.e. a series of processes that returns to the same
state), we have I
d¯Qirr
< 0, (218)
T
which can be compared with equation (214).

• Thus in general we can write

I
d¯Q
0, (219)
T
where the “=” is for reversible processes and the “<” is for irreversible processes.

• But what we really want is to know how entropy changes for any process (not necessarily
cyclic).

• Consider Fig. 54, which shows a cycle made up of an irreversible and a reversible process.

Figure 54: Blundell fig. 14.1.

111
 • The cycle is irreversible
I so we can apply equation (218) to this case by replacing dQirr
d¯Q
with dQ, which gives < 0. Thus,
T
I Z B Z A Z B Z B
d¯Q d¯Qirr d¯Qrev d¯Qirr d¯Qrev
= + = < 0. (220)
T A T B T A T A T
Z B Z B
d¯Qirr d¯Qrev
• Hence, < . But the RHS is just the change in entropy S = S(B)
A T A T
S(A) (from equation 216). Therefore, we obtain
Z B
d¯Qirr
< S, (221)
A T

• In general, A ! B can be irreversible or reversible, so we can write

Z B
d¯Q
 S, (222)
A T

where, again, the “=” corresponds to a reversible process and the “<” corresponds to an
irreversible process.

• In the limit of small changes, equation (222) simply becomes

d¯Q
 dS . (223)
T

• Now let’s consider the system to be thermally isolated, so there is no heat flow between
the system and surroundings. A special case of this is taking the system to be the entire
Universe, so let’s consider that case (to fix ideas).

• In that case, d¯Q = 0 and thus the LHS of equation (222) vanishes, and we are left with

SUni 0 (entropy form of the 2nd law of TD). (224)

• Equation (224) is another form of the 2nd law of TD. Every process makes the total (net)
entropy of the Universe go up (irreversible process) or stay the same (reversible process).
No process ever makes the total entropy of the Universe go down.

9.6.3 Calculating the change in entropy during an irreversible process

Z B
d¯Qrev
• Equation (216), S = , gives the change in entropy of the system and is valid
A T
for a reversible process. But how can S be calculated for an irreversible process?

• There is a trick! Recall that S is a state function, so equation (216) can always be used
to calculate the change in entropy.

• However, if the process to go from state A to state B is irreversible, then in order

Z B to use
d¯Qrev
equation (216) to calculate S = S(B) S(A) we need to calculate the RHS
A T
for a reversible process.

112
 • Any reversible process will do because all have the same efficiency as a Carnot engine (as
we showed above using Fig. 53).

Example 1 :
(Blundell 14.7a) Consider n moles of a gas, initially confined within a volume V and held at
temperature T . The gas is expanded to a total volume ↵V , where ↵ is a constant, by (a) a
reversible isothermal expansion and (b) removing a partition and allowing a free expansion
into the vacuum. Both cases are illustrated in Fig. 55. Assuming the gas is ideal, derive an
expression for the change of entropy of the gas in each case.

Figure 55: Blundell fig. 14.9.

Solution:
Z B
d¯Qrev
a) We need to calculate . For an isothermal process E = 0 for an ideal gas. So
A T
the 1st law gives
dV
d¯Qrev = d¯Wrev = P dV = nRT (225)
V
Then we obtain
Z B Z ↵V
d¯Qrev dV 0
S= = nR = nR[ln(↵V ) ln(V )] = nR ln ↵ = N k ln ↵. (226)
A T V V0

b) The answer is exactly the same as in part (a) because the initial and final states are the
same. The change in entropy does not depend on the path (in P V -space or any other
parameter space).

113