Statistical Thermodynamics: Molecules to Machines

Venkat Viswanathan
May 20, 2015

Module 2: Classical Thermodynamics

Learning Objectives
Introduction of the basic concepts that underlie classical thermodynamics, which gives the governing laws for macroscopic matter.

First two laws of thermodynamics.

Thermodynamics driving forces that arise from the second law of

thermodynamics.

Classical thermodynamics applied to phase coexistence.

Key Concepts:
First and Second Laws of Thermodynamics, thermodynamic driving
forces, ensembles, extensive and intensive variables, Legendre transform,
Euler equation, Phase equilibrium.

statistical thermodynamics: molecules to machines

First Law of Thermodynamics

The first law of thermodynamics, also known as the conservation of
en- ergy principle, provides the basis for studying the relationships
among the various forms of energy and energy interactions. The law
states that although energy assumes many forms, the total quantity of
energy is con- stant, and when energy disappears in one form it
appears simultaneously in other forms.
The energy of a system is measured by its internal energy U , and is
defined as the net sum of the kinetic and potential energies of all the
molecules in a macroscopic material. The internal energy is an
extensive property, meaning that the internal energy depends linearly
on the size of the system.
According to the First Law of Thermodynamics any change in the
internal energy occurs due to the addition of heat Q and work W , according to the relation:
dU = Q + W

(1)

The difference in the notation of the change of internal energy (dU

) from the addition of heat (Q) or work (W ), lies in the fact that
internal energy is a state variable, while heat and work are path
dependent. However, once heat and work are applied, they are
indistinguishable on a macroscale.

Second Law of Thermodynamics

The origins of the second law lies in a statement that it is impossible
for any self-acting process or machine to produce net flow of heat from
a region of low temperature. The law states that an isolated system
always proceeds from an ordered to a more disordered state, or else it
undergoes no change at all.
The level of order from a system is measured by the entropy S,
and according to the second law the entropy of the universe tends to
a maximum (Rudolf Claussius, 1865). This is stated mathematically
as:
(dS)U ,V ,N 0

(2)

Where the subscripts U , V and N indicates constant internal

energy, volume and number of particles, respectively.

Thermodynamics Driving Forces

A thermodynamic system can be defined according to their extensive
variables, i.e., variables that vary linearly with the system size. For example, we define the internal energy U to be a function of extensive
vari-

Figure 1: The total internal energy of

a system containing two subsystems,
is the sum of the internal energies of
each subsystem

statistical thermodynamics: molecules to machines

ables S, V , and N = (N1, N2, ..., Nr ), where r is the number of

chemical species in the system. Therefore, we have U = U (S, V , N ).
Differential changes in the extensive variables leads to a differential
change in the internal energy, such that:
.
.
.
U
U
U
dU
dS
dV
dNi
(3)
.
.
.
=
S,V ,Nj=i
S
V
Ni
+
+
V ,N1,...,Nr

S,N1,...,Nr

The partial derivatives are intensive variables, i.e. variables that do

not depended on the size of the system. We define these as:
.

U
.
S

.
p
V S,N1,...,Nr

T
V ,
N1,...,Nr

U
.
Ni

i
S,V ,Nj=i

(4)

Where T is temperature, p is pressure, and i is chemical potential

of species i.
The change in internal energy is written as:
r

dU = TdS pdV +

(5)

idNi

i=1

Since U is a state variable, we can choose any path to find a

differen- tial change. Equation (5) states that any change in the
internal energy of the system, is the result of a quasi-static heat flux
(TdS), mechanical work (pdV ) and chemical work (idNi) to the
system.
Equivalently, we can define the system according to their entropy
S = S(U , V , N ), resulting in a differential change:
r
1
p
i
.
dS =
dU
dV
dNi
(6)
T
T
+
i=1
T
Which is useful to discuss the second law of thermodynamics.
Consider a closed, isolated system which cannot exchange heat,
work,
or particles with its surroundings (Fig. 2). We initially prepare the
system such that part of the system (subsystem 1) has parameter
values U (1), V
(subsystem
2).

( 1)

, and N

(1 )

, and analogous for the rest of the system

Making the statement that the entropy is additive, i.e. S(1) + S(2) =
S), we can write:

dS =

p(1)

1
dU
+

T (1)
1
dU
T (2)

dV

(1)

(2)

(1)

T (1)
+

p(2) dV
(2)

T (2)

(2)

r. (1)
i

(1)
i=1 T
r. (2)i

i=1

T (2)

dN

(1 )

dN
i

(7)

Figure 2: Closed, Isolated System containing two subsystem

statistical thermodynamics: molecules to machines

Since the total system is closed and
isolated,(1)we can write dU (2) =
2
(
)
dU (1), dV (2) = dV (1), and dN
= dN .
i

Therefore, we can write:

dS =

1
T
T
(1)
r
.

i=1

dU
+

(2)

p(2)
.
dV (1)

(1)

p(1)

T (2)

T
(1)

(1)

(2)

(1)

.
(1)

i
T

(8)

dNi

(2)

According to the second law of thermodynamics, any spontaneous

process requires dS 0, thus striving for the maximum value at equilibrium. The condition at equilibrium satisfy the condition dS = 0 for
arbitrary small changes dU (1), dV (1), and dN (1):

Thermal equilibrium occurs when T (1) = T

(2)

Mechanical equilibrium occurs when p(1) = p(2)

Chemical equilibrium (no reactions) occurs when

(1)

(2)

Notice that although thermal and mechanical equilibrium implies

spatial homogeneity of T and p, the condition of spatial homogeneity i
should not be interpreted as homogeneous concentration or density of
species i (vapor-liquid equilibrium in a box or a structured fluid like a
1
block
- Fig.
(1)3 ). (2)
(1) copolymer
Consider
the
case
p
=
p ,
(2)

i
i , and the partition
(1)
between subsystems is rigid and impermeable (dVi (1) = dN = 0).

If T (1) > T (2) the condition dS 0 is met only if dU (1) < 0

If T (1) < T (2) the condition dS 0 is met only if dU (1) > 0

Heat spontaneously flows from high temperature to low.

(1)

In the case T (1) = T (2),

Kell Mortensen, Damian A. Hajduk,

and Sol M. Gruner. Phase Behavior of
Polystyrene Poly(2-vinylpyridine) Diblock Copolymers. Macromolecules,
29(8):28572867, January 1996. ISSN
0024-9297

(2)

(1)

= 0).

If p(1) > p(2) the condition dS 0 is met only if dV (1) > 0

If p(1) < p(2) the condition dS 0 is met only if dV (1) < 0

Inhomogeneities in pressure leads to the spontaneous expansion of

high pressure regions into low pressure regions.
In the case T (1) = T (2), p(1) = p(2):
(2)

1
Mark F. Schulz, Ashish K. Khandpur, Frank S. Bates, Kristoffer Almdal,

= , and the partition between

subsystems is movable but impermeable (dNi

(1)

Figure 3: Phase behavior of

polystyrene-poly (2-vinylpyridine)
diblock co-polymer. Transmission
elec- tron micrographs show lamellar
phase (fraction polystyrene f = 0.4),
gyroid phase (f = 0.38), and
cylindrical phase (f = 0.35). The
experimental phase diagram is
shown in the lower right
panel.

(1)

If i 1 > i the condition dS 0 is met only if dNi < 0

( )
If
(2)
(1)
< i the condition dS 0 is met only if dNi > 0
i
Matter flows from high chemical potential to low (not always high

Figure 4: A schematic of the

underlying molecular mechanism for
heat conduc- tion in a crystal. The
atoms in a crys- tal fluctuate due to
thermal energy; the
higher the temperature, the larger the
fluctuations. At short time t1, a crystal has a region
high
temperature,
indicated
by theofred
atoms.
The fluc-

concentration to low).

tuations of the hot atoms (red) leads to the neighboring cold atoms (blue)
heating up due to interactions. As time progresses from t1 through t4, the hot
spot dissipates and cools down, leading to an increase in the neighboring temperature. The process of heat transport from high temperature to low temperature due to molecular fluctuations is conduction.

Thermodynamic Ensembles
The governing thermodynamic potential of the closed, isolated system
is the entropy S(U , V , N ), i.e. the control or canonical variables are
U , V , and N . However, the closed, isolated system is only one
possible thermodynamic system. In many instances, it is desirable to
have differ- ent control variables, where we replace one (or more)
extensive variable with their conjugate intensive variable (Fig. 6).
For convenience, we start from the equation of state U (S, V , N ),
which is equivalent to S(U , V , N ). The extensive variables and their
conjugate intensive variables are related through the relationships
.
ST =

U
. V
S

Figure 5: Four different ensembles and

their canonical variables
,N1,...,Nr

V p = .

U
. S,N1,...,Nr
V

Ni i = .

U
.
Ni

S,V ,
Nj=i

(9)

A thermodynamic ensemble is defined by the canonical variables of

the system. Changing from one ensemble to another amounts to
shifting from one thermodynamic potential to another that depends
on the new canonical variables. Replacing an extensive variable for its
conjugate intensive variable is effectively replacing the control
variable with the slope of the control variable.
The mathematical method of performing this change of variables is
called a Legendre Transform. The method is defined according to the
following steps:

A function y = f (x) is defined as a list of x, y, pairs, i.e. the plot of

y(x) graphs (x1, y1), (x2, y2), (x3, y3),...

Equivalently, it is possible to express the same function in terms of

the tangent slope c(x) and the corresponding y-intercepts, i.e. the
points (c1, b1), (c2, b2), (c3, b3),...

For a change dx in the x-variable, the change dy is given by:

.
.
dy
dx = c(x)dx
dy =
dx

(10)

From this point on the curve, we can extend a line back to x = 0 to

find the y-intercept b(x), such that
y(x) = c(x)x + b(x) b(x) = y(x) c(x)x

(11)

The resulting function for the series of intercepts versus the slopes
(i.e. b = b(c)) contains the same information as y = y(x).

A change in b(c) is given by db = dy cdx xdc = xdc.

Example. Consider the function y = (x 2)2. We use a Legendre

Transform to shift the variable from x to the slope c. The slope of the
function is:
c
c = 2(x 2) x = + 2
(12)
2
The Legendre transform gives us a new function
b(c), given by:
c2 c2
c2
b = y cx
2c = 2c
(13)
=
4
2
4
We note that this Legendre transform correctly relates back to the
original function by noting that
db

2 = x
(14)
dc
2
All of the thermodynamic potentials that define specific ensembles
are Legendre transforms of the original potential U = U (S, V , N ):
Transform from S to T for the closed, isothermal ensemble gives
the Helmholtz free energy:
.
F=U

U
.
S

S = U TS

(15)

V ,N

which has a differential change:

r

dF = dU TdS SdT = SdT pdV +

idNi
(16)

i=1

Transform from S to T and V to p for the closed, isothermal,

isobaric ensemble give the Gibbs free energy:
G=U

U
. V ,N S

Figure 6: Plot of the function y(x) =

(x 2)2 and its corresponding
Legen-

U
.

V
S
which has a differential change:

V = U TS + pV

(17)

S,N

dG = dU TdS SdT + pdV + V dp

r
.
= SdT + V dp +
idNi

(18)

i=1

Euler Equation
The internal energy is a homogeneous, first-order function, which requires that the internal energy must increase linearly with the size of
the system. Mathematically, this is written as:
U (S, V , N ) = U (S, V , N )

(19)

dre transform b(c) =

4
2c

where is an arbitrary scaling factor for the system size. If we take

the derivative of this equation with respect to , we arrive at:
U (S, V , N )

= U (S, V , N )

The left-hand side is written as:

.
.
U (S, V ,
U (S, V ,
(S)
.
.
N )

V
N )
+
,N
.
.
(V )
r (S)
U (S, V , N )
(Ni)

(20)

(V )
S,N

TS pV

.
i=1
.

(Ni)

S,V ,Nj=i

iNi
+

i=1

(21)

Therefore, the total Legendre Transform equals zero, which is the

Euler equation:
r
.
U = TS pV +
iNi
(22)
i=1

Phase Equilibrium
Consider a single component system that obeys the van der Waals
equa- tion of state:
p=
NRT
V Nb

aN 2
V

RT
vb

a
v2

(23)

where v = V /N is the molar volume.

Although a and b are considered empirical constant, there are
micro- scopic justifications for the van der Waals constants:
The constant b represents the hard-core excluded volume of the
molecules in the system.
The constant a determines the strength of the two-body attractive
interaction between molecules in the system
For simplicity, we write the dimensionless equation of state:
p
pb

1
v
1

(24)

v 2
, thus reducing the number

=
, = v , and = a
of
v
a
R
b
bR
T
parameters to just a to capture theT temperature dependence.
Consider the closed, isothermal system containing a van der Waals
fluid in vapor-liquid equilibrium (Fig. 7). Our goals are to find the
limits of stability of each phase, the conditions where the phases
coexist, and the properties of the two phases in coexistence. The
simple example of phase coexistence demonstrates the essential
issues at work in more complex, multi-component, multiphase systems.
where
p

Figure 7: Closed, isothermal system

containing a van der Waals fluid in
vapor-liquid equilibrium

Limits of Stability
Thermodynamic stability requires that T =
.

limit of stability of a single pure phase occurs when:

.
.
2
1
p
=
a =0
(v
v a

1)2 +

> 0, and the

p T

(25)

v 3
This gives the cubic equation:
. 2
.
3
v + 2a v 2v + 1 = 0

(26)

Equation (26) has three solutions:

The first solution is less than one (v < 1) and is unphysical since
the pressure p diverges as v 1+.
The two solutions greater than v = 1 identify the limits of
stability of the vapor and liquid phases.
The two solutions are approach each other with decreasing a , until
that meet at the critical point:
v c = 3

and

Phase Coexistence

a
=

27
8

Figure 8: Plot of p versus v for a =

3.679. The Maxwell construction for
coexistence states that vapor-liquid coexistence occurs when the areas A1 =
A2. The dashed segment of curve indicates conditions where a single phase
is unstable

(27)

The equilibrium conditions are (v) = (l) and p(v) = p(l) = pcoex,
and the molar Helmholtz free energy is written as:
= f .
dv = f0 p
0
.

+
dv
f
v
NRT
a .
= log (v a
a fdv
.
1)
1

= f0

0
v 1
+
v
v 2
f =

Figure 9: Plot of f versus v for a =

3.679. The common-tangent construction for coexistence states that vaporliquid coexistence occurs when a
single

(28)

where f0 is an unspecified constant of integration (reference state).

The Euler equation for a one-component system is:
N = U ST + pV = F + pV = G

line is simultaneously tangent to

points on the v f curve. The
dashed segment of curve indicates
conditions
where a single phase is unstable

(29)

or in dimensionless form, we have:

v
= = f + pv = f0 + log(v 1) 2
(30)

RT
v 1
v
Setting the chemical potentials in vapor and liquid phases equal, is
equivalent to:

(
v
)

two

27

Figure 10: Plot of versus v for a =

(v)
=

.
.
p(v)dv + pcoex v (v) v (l) =

(l)
v
(l)

(31)

3.375 = 8 (critical point, blue), a

=
3.527,
a
= a = 3.679, a = 3.831, and
3.982 (red). The dashed segments indicate conditions where a single phase is
unstable.

This is rewritten as:

.
. v
(v)
pcoex v (v) v (l)
p(v
=
)dv

(32)

v
(l)

As written, this leads us to the Maxwell construction for vaporliquid phase coexistence, which provides a graphical method of
identifying phase coexistence (Fig. 8).
A second way to view the coexistence condition is to note:
Figure 11: Plot of versus p for
a =
3.679. Coexistence conditions are de-

(v) =
f(v)
(v)

p = p
f
=

(l)

(l)

(v)
f

f(v

v (v) =
v f(l)
(l

)
)

f(
l)

termined by the point where the curve

crosses itself along the vapor and liq-

v (
l)

uid branches. The dashed segment of

(33)
v

This development gives the common-tangent construction (Fig. 9,

10). The most direct method of visualizing the coexistence
condition is
plot versus p. Numerically calculating the equilibrium
conditions is
easiest using the plot of versus p, since the conditions involve a
single point of crossing (Fig. 11, 12).
The conditions on the v p plane that correspond to the
coexistence points form the binodal curves. The limit of stability of the
pure vapor and pure liquid phases form the spinodal curves. These
curve form the phase diagram, which predicts the conditions of
coexistence of the vapor and liquid phases (Fig. 13).

References
Mark F. Schulz, Ashish K. Khandpur, Frank S. Bates, Kristoffer Almdal, Kell Mortensen, Damian A. Hajduk, and Sol M. Gruner. Phase
Behavior of Polystyrene Poly(2-vinylpyridine) Diblock Copolymers.
Macromolecules, 29(8):28572867, January 1996. ISSN 0024-9297.

curve indicates conditions where a single phase is unstable.

Figure 12: Plot of versus p for

a =
27
3.375 = 8 (critical point, blue), a =
3.527, a = 3.679, a = 3.831, and a
= 3.982 (red). The dashed segments
indi- cate conditions where a single
phase is unstable.

Figure 13: Plot of p versus v for

a =
27
3.375 = 8 (critical point, blue), a =
3.527, a = 3.679, a = 3.831, and a
= 3.982 (red). The solid black curves
are the binodal curves (coexistence),
and the dashed black curves are the
spinodal
curves
(limit
of
metastability).