Chapter 4

2nd Law of Thermodynamics

4.1 General statement of the law

The First Law is an empirical statement regarding the conservation of

energy.

The Second Law is concerned with the maximum fraction of heat that
can be converted into useful work.

The second law may be stated in several different ways, such as :

a) Thermal energy will not spontaneously flow from a colder to a

warmer object. (What is thermal energy?)

b) The entropy (defined below) of the universe is constantly

increasing.

Thus, the second law is not a conservation principle (as in the 1st Law),
but rather is a law defining the direction of flow of energy. In the
following we will see that entropy and energy are closely related

1
4.2 Entropy
Entropy is a state function defined by (per unit mass)

dq rev
ds ≡ (defined for a reversible process)
T Note: entropy is also represented by the symbol φ

The second law defines entropy as a state function (see Petty, Section 6.1) and
permits the following statements:

a) For a reversible process the entropy of the universe remains constant.

b) For an irreversible process the entropy of the universe will increase.

Thus, a more general definition of entropy is

dq
ds ≥
T
Note that the Second Law does not address anything specifically about the
entropy of the system, but only that of the universe (system + surroundings).

http://en.wikipedia.org/wiki/Second_law_of_thermodynamics

Chap. 6 in Petty is short and sweet. He has a good explanation of the

second law and its relation to thermodynamic equilibrium in Section 6.2.
Read this carefully.

Key Fact: Within any isolated system that is not at equilibrium, the net
effect of any active process is always to increase the total entropy of the
system. A state of equilibrium in therefore reached when the total entropy
of the system has achieved it maximum possible value. At this point, no
further evolution of the system state variables is possible.

2
 Definition of reversibility (revisited)
A system process is defined as reversible if a system, after
having experienced several transformations, can be returned to
its original state without alteration of the system itself or the
system's surroundings.
1. A reversible transformation will take place when a system moves by
infinitesimal amounts, and infinitesimally slowly, between equilibrium
states such that the direction of the process can be reversed at any
time.
2. Remember that in a reversible process the deviation from equilibrium
is infinitesimal. [Refer to the work of expansion problem considered
previously in Section 3.6.]
3. In a reversible process, the entropy of the universe (i.e., the system
plus surroundings) remains constant.

We can examine reversible processes theoretically, but do reversible

processes actually take place in the atmosphere?

Thermodynamic process can be classified into one of three categories:

1. natural
2. reversible
3. impossible.

Natural processes are more or less irreversible. Examples include:

• friction - associated heating warms the surroundings (frictional heating in a

hurricane is real)
• unrestrained expansion (expansion of a gas into a vacuum) - again the
surroundings are modified
• heat conduction in the presence of a temperature gradient (surface
heating/cooling)
• chemical reactions (e.g., the combination of two atoms of H and and one
atom of O in the production of H20)
• turbulent mixing and molecular diffusion of pollutants and aerosols
• freezing of supercooled water
• precipitation formation - removes water and heat from an air parcel
• mixing between a cloud and the subsaturated atmosphere
• deliquescence behavior of NaCL

3
A closer look at entropy:
In the p-V diagram below , isotherms are distinguished by differences in
temperature and the adiabats by differences in potential temperature θ. There is
another way of distinguishing differences between adiabats. In passing from one
of the adiabats (θ1 or θ2) to another along an isotherm (this is actually one leg of
the Carnot cycle, see also the appendix), heat is absorbed or rejected, where the
amount of heat ∆qrev depends on the temperature of the isotherm. It can be
shown that the ratio ∆qrev /T is the same no matter what isotherm is chosen in
passing from one adiabat to another.

Therefore, the ratio ∆qrev/T

is a measure of the
difference between the
adiabats – and this is also
the difference in entropy s.

Using the definition of entropy from Eq. (4.1), the first law can be expressed as

dq = Tds = du + pdα.

When a substance passes from state 1 to state 2, the change in entropy is

found by integrating (4.1):

2 dq rev
∆s = s 2 − s1 = ∫
1 T

4
 The relation between s and θ -- Another cool derivation
How is entropy related to the more commonly-used atmospheric variables?

Combine the equation of state, pα = RT, with the first law in the form
dq = cpdT - αdp.

We can then write

dq dT dp
ds = = cp −R
T T p
Taking the log differential of Poison's Eq. (potential temperature) we can write

dθ dT dp
cp = cp −R
θ T p

Since (4.2) and (4.3) have identical right-hand sides (RHS), they can be equated:

dq dθ
ds = = cp = c p d ln θ
T θ

Now we have a more intuitive definition of entropy!

Thus the entropy function can be expressed in terms of potential temperature as

dθ
ds = c p (differential entropy)
θ

or s = cp lnθ + const.

From this we can see that transformations in which entropy is constant

are also processes in which the potential temperature of an air parcel is
constant. Such processes are called isentropic (adiabatic) processes.

Analyses using the variable θ are similarly called isentropic analyses, and
lines of constant θ are termed isentropes. An example of an isentropic
analysis, and a corresponding temperature analysis, is shown in Fig. 4.2.

5
 a) Temperature analysis

Conversion between
T and θ, as shown in
a vertical cross tropopause

Pressure (mb)
section of each

Height (km)
Solid contour lines
Upper panel: isotherms
cold front
Lower panel: isentropes
The difference: T → θ transformation
b) Isentropic analysis

Note how the tropopause is better

defined in the insentropic analysis.

Pressure (mb)
Height (km)
Fig 4.2. Analysis of (a) temperature and (b)
potential temperature along a vertical
section between Omaha, NE and
Charleston, SC, through the core of a jet
stream. In each panel, wind speed in m s-1
is indicated by the dashed contours. Taken
from Wallace and Hobbs (1977). OM CB NA AT CH
Horizontal distance (sounding location)

4.3 A generalized statement of the second law (review)

Calculation of entropy requires an equivalent reversible process. [But all natural
processes are irreversible since they move a system from a nonequilibrium state
toward a condition of equilibrium.]

The second law can be stated more generally in terms of the following
postulates:

1) There exists a function of state for a system called entropy s.

2) s may change as the system: (a) comes into thermal equilibrium with its
environment or (b) undergoes internal changes within the body. The total entropy
change ds can be written as the sum of external (e) and internal (i) changes

ds = (ds)e + (ds)I

3) The external change (ds)e is given by (ds)e = dq/T.

4) For reversible changes, (ds)i = 0, and for irreversible changes, (ds)i > 0.

6
 Thus,
ds = dq/T for reversible changes
ds > dq/T for irreversible changes

Combining these two gives the generalized form of the first law as

Tds ≥ du + pdα, (4.4)*

where the equality refers to reversible (equilibrium) processes and the

inequality to irreversible (spontaneous) transformations.

Examples

4.4.1 Some idealized entropy change processes

[Note: in the examples below we are beginning with the first law dq = du + pdα or
dq = cpdT - αdp]

a) isothermal expansion of an ideal gas

For an isothermal process du = 0 and the work of expansion (determined

previously) is

∫pdα = nRTln(α2/α1).

∆s = ∆q/T = nRln(α2/α1).

[Proof: Since α = RT/p, dα = (R/p)dT - (RT/p2)dp. Then -pdα (=dw) = -RdT +

RTdlnp]

If the final specific volume α2 is greater than the initial α1 then the entropy change is
positive, while for a compression it is negative.

7
b) adiabatic expansion of an ideal gas

For a reversible adiabatic expansion dq=0 and the entropy change

is ds=0. This is the isentropic process defined previously.

c) heating of an ideal gas at constant volume

By defintion, dα=0. Then

ds = dqrev/T = cvdT/T = cvdlnT.

8
d) heating of an ideal gas at constant pressure

For a reversible process

ds = dqrev/T = cpdT/T = cpdlnT.

e) entropy changes during phase transitions

For a phase transition carried out reversibly,

∆s = ∆htransition/Ttransition.

[Recall that ∆h = L = cp∆T for a phase change . . . which occurs at

constant pressure.]

9
4.4.2 A more comprehensive example: The entropy change in an
irreversible process

Consider the isothermal expansion of an ideal gas: p = 1 atm, T = 273.1 K, and V

= 22.412 liters per mole.

Let this system expand isothermally against a constant external pressure of 0.5
atm. The final volume is 44.824 liters and the work done is

pext(V2-V1) = 0.5(22.412) = 11.206 L atm = 271.04 cal = 1135 J (1 cal = 4.187 J).

This is the heat that must be supplied from an external reservoir to maintain
isothermal conditions.

Since this process is irreversible, the entropy change of the system is not dq/T.
Rather, we must find a reversible process from the initial to final state.

In this case we refer to Example (a) above (isothermal expansion) in which

∫dqrev = RTln 2 = 1573 J.

The change in entropy of the reversible process is thus ∆q/T = 1573 J / 273.1 K =
5.76 J K-1.

4.4.3 The phase change entropy

At 273.15 K (0 °C) the entropy of melting of water is

Lil/Tf = 3.34 x 105 J kg-1 / 273.1 K = 1223 J K-1 kg-1,

while at 373.1 K the entropy of vaporization is

Llv/T = 2.25 x 106 J kg-1 / 373.1 K = 6031 J K-1 kg-1.

Note the large difference between these two entropies. Why?

This entropy change is due primarily to two effects: (1) the entropy associated with
the intermolecular energy and (2) configurational entropy.

Further explanation: For the conversion of ice into water there is little change in
the intermolecular entropy term and an increase in configuration entropy in
transforming to a slightly less ordered system. However, in evaporation there is a
large change in intermolecular entropy (the molecules in the gas are spaced far
apart and are subject to little interaction compared to molecules in the liquid
phase) as well as a large change in configurational entropy in going from a
somewhat ordered liquid to a nearly completely disordered gas.

10
Example:
Calculate the change in entropy when 5 g of water at 0 °C are raised to
100 °C and then converted to steam at that temperature. We will assume
the latent heat of vaporization is 2.253x106 J kg-1 at 100 °C. (Note that
we will use the extensive forms – capital letters – since mass is involved.)

Step 1: Compute the increase in entropy resulting from increasing the

water temperature from 0 to 100 °C:
373
∆S1 = S 373 − S 273 = ∫ dQrev T
273

Here, dQrev = m(dqrev) = mcwdT where m is mass and cw is the specific

heat of water. It we assume cw to be constant at 4.18x103 J kg-1 K-1
then

373
∆S1 = S373 − S 273 = (0.005 kg )(4.18 × 103 J kg −1 K −1 ) ∫ dT T
273

= 20.9 ln(373/273) = 6.58 J K-1

Step 2: Compute the change in entropy from conversion of 5 g of

water to steam, which involves a latent heat term.

This is

∆S2 = mLvl/T = (.005 kg)(2.253×106 J kg-1)/373 K

= 30.2 J K-1.

The sum of these components gives the total increase ∆S

∆S = ∆S1 + ∆S2 = 6.58 + 30.2 = 36.78 J K-1.

11
4.5 The free energy functions

The first law is a conservation statement . . .

The second law governs the directions of thermal energy transfer and also permits
the determination of the reversibility of a process.

It is desirable to have a function or set of functions which will describe for a

system the likelihood of a given process and the conditions necessary for
equilibrium.

Since there are really only two basic thermodynamic functions (u and s), we can on
the basis of convenience define additional functions that may be based on u or s

[Wait a minute -- It may not be clear why u and s are so

basic. Think about this.] Is this true???

These functions can then be used to define equilibrium conditions for processes to
be considered later.

4.5.1 Helmholtz free energy

The Helmholtz free energy is defined as

f ≡ u - Ts.

In differential form, we have

df = du - Tds - sdT (4.5)

Combining this with Eq (4.4) (Tds=du+pdα -- recall that the equality implies the
reversible condition here) gives

df = -sdT – pdα

If a system is in equilibrium and both T and α are constant, then df = 0. For a

system which undergoes a spontaneous (irreversible) process, we have

df < -sdT - pdα

and df < 0. Thus, a system at constant T and volume (α) is in a stable equilibrium
when f attains a minimum value. For this reason, the Helmholtz free energy is
sometimes called the thermodynamic potential at constant volume.

12
4.5.2 Gibbs free energy

In this case we will derive the Gibbs free energy from the First Law using the form
(after all, let’s see how fundamental Gibbs free energy is):

dq = du + pdα.

Integration between the limits associated with a phase change, we get

L ≡ ∫ dq = ∫ du + ∫ pdα

Assuming p=const and through some simple rearrangement we can obtain

q2
L = T ∫ dq = T ( s 2 − s1 ) = u 2 − u1 + p (α 2 − α1 )
T
q1

where the subscripts 1 and 2 denote the two phases. Rearranging to combine
like subscripts yields the following equality regarding the energy between the two
phases:

u1 + pα1 - Ts1 = u2 + pα2 - Ts2

Based on the above, the Gibbs free energy is defined as (per unit mass)

g = u - Ts + pα [= f + pα]

In differential form,

dg = du - Tds - sdT + pdα + αdp. (4.6)

Again, we can use (4.4) (Tds = du + pdα, in reversible form) to obtain

dg = -sdT + αdp.

In this case, if T and p are constant, for a body in equilibrium we have dg = 0.

For an irreversible process,

dg < -sdT + αdp.

Thus, dg < 0 in an irreversible, isobaric, isothermal transformation. Gibbs free

energy is also called the thermodynamic potential at constant pressure.

We will find that g is very useful for phase changes which occur at constant T
(isothermal) and p (isobaric).

13
4.5.3 The free energy functions and total work

At this point it is instructive to relate g and f to the external work that a

system can perform under various conditions. So far we have assumed that the
only work term is that of expansion, pdα. There are other forms of work that we
will consider, however. [Recall the strange ∑ei term in Eq. (3.3) at the top of
page 3, Chapter 3: Du = q + ∑ ei]

For example, the creation of a surface in the nucleation (formation) of water

droplets and ice crystals will be of interest to us.

In this more general form, the First Law can be written as

dq = du + dwtot,
and for a reversible transformation
Tds = du + dwtot (4.7)
where the total work is dwtot.
If we combine the above with (4.5) and assume an isobaric condition, we find
dwtot = -df - sdT.
Furthermore, for an isothermal process,
dwtot = -df.

Thus, the total external work done by a body in a reversible, isothermal, isobaric
process is equal to the decrease in Helmholtz free energy of the body.

If da (this variable is a and not α) is the external work done by a unit mass of a
body over and above any work of expansion (pdα), i.e.,

da ≡ dwtot - pdα ,

then we can use (4.6), (4.7) and the above to write

da = -dg -sdT + αdp.

For an isothermal, isobaric process,

da = -dg.

The thermodynamic functions f and g have important applications in problems

involving phase changes in the atmosphere. In particular, these functions will be
utilized later in this course when we consider the formation (nucleation) of water
droplets from the vapor phase.

[In other words, don’t forget about g!]

14
Appendix: The Carnot Cycle: Highlights
The Carnot cycle may be one the the most
popular examples used in the study of increasing p
(general) Thermodynamics. ( See Petty,
pp. 143-149).

The Carnot cycle illustrates several aspects of

the Second Law, and also defines
thermodynamic efficiency.
p2
The Carnot cycle is a sequence of 4
component processes, two isothermal and
two adiabatic.

These component cycles are interlaced as

follows:

reversible isothermal expansion at T = T1

reversible adiabatic compression at θ = θ1 p1
reversible isothermal compression T = T2
reversible adiabatic expansion θ = θ2
Fig. A.1 illustrates these paths as they
would appear on a skew-T, ln p diagram.
(note, Tsnonis uses a p-V diagram.)

The quantitative measures of work, internal energy change, and heat input
along each leg are detailed in Petty, pp. 143-149. Take some time to examine
these.

From the Carnot cycle, the thermodynamic efficiency can be defined as

T2
Ethermo = 1 −
T1

Efficiency is zero when T1 = T2, and is maximized when T2 << T1.

15
 Tsnonis mentions two postulates that originate from this, and these are alternate
statements of the Second Law:

Kelvin’s postulate: It is impossible for a thermal engine to accomplish work at

only one temperature (p. 52, Tsnois).

Clausius’s postulate: A transformation that permits heat transfer from a cold

body to a hot body is impossible (p. 53, Tsonis). (Is this not a
restatement of the 2nd Law?)

Recall that the First Law does not address the possibility of transformations; it
only quantifies them, even if they are impossible. (Think of the First Law as
the smart person who has no common sense, and the Second Law as the
wise person who has abundant common sense.)

Question for discussion: How does the Carnot Cycle illustrate the way
in which a heat pump (or refrigerator) works.

Another example problem:

Calculate the change in entropy when 1 mol of an ideal diatomic gas

initially at 13 °C and 1 atm changes to a temperatu re of 100 °C and a
pressure of 2 atm.

16
Find the change in air pressure if the specific entropy decreases by 50 J kg-1 K-1
and the air temperature decreases by 5%.

ds = cpdT/T - Rdp/p
Rearrange: dp/p = (cp/R)dT/T - ds/R
Integrate: ln(pf/pi) = (cp/R)ln(0.95Ti/Ti) - ∆s/R
Insert values: ln(pf/pi) = (1005.7/287.05)ln(0.95Ti/Ti) - (-50)/287.05
= 3.504 (-0.0513) - 0.174 = -0.00576
pf/pi = 0.994
pf = 0.994pi

HW problems
• Notes: Problems 1-4
– Hint for No. 1a: Internal energy has 3
components

17