Chap.

1: Introduction
Introduce yourself.

Thermodynamics is a funny subject. The first time you go through it, you
Go through first page of syllabus & homework instructions. dont understand it at all. The second time you go through it, you think you
understand it, except for one or two small points. The third time you go
[Pair] Why learn Thermodynamics?
through it, you know you dont understand it, but by that time you are so
used to it, it doesnt bother you any more. Arnold Sommerfeld
Explain active learning.
1. Whats classical thermodynamics (power developed from heat) ? A difficult course that requires abstract thinking, logical
reasoning, and mathematical derivation.
(1) A set of general laws that describes static states of macroscopic systems;
(2) Independent of microscopic or molecular-level behavior of matter;
(3) Based on a few number of postulates, from which the whole subject is developed through mathematical transformation and logic.
Overview of course structure on the third page of syllabus.
Important: Review your undergraduate calculus, especially for multivariable functions.
Review your undergraduate thermodynamics.
Go through the lecture notes BEFORE doing your homework.
Ask me questions.

dA
A
vs.
, no A !
dx
x
A
A
For A( x, y ), dA
dx
dy.
x
y

A vs. dA,

2. Three steps of using classical thermodynamics to solve engineering problems

(1) Problem definition and modeling
Specify internal constraints and boundary conditions; use idealization to make the problem tractable.
(2) Application of thermodynamics laws and mathematical reasoning
(3) Evaluation of property data
Classical thermodynamics doesnt specify everything about the system; extra data (heat capacity, equation-of-state, ) are
needed that are unique characters of the matter under study. These are obtained through experiments, computer simulations, or
theoretical calculations using statistical mechanics.
NIST Webbook: http://webbook.nist.gov/chemistry/fluid/

Chap. 2: Basic Concepts (I)

1. System subject of experiment;
Environment everything else;

Universe

Boundary separates the system from its environment.

(1) System interacts with environment through the boundary.
(2) Classification of boundaries and systems: [TPS] Read through and come up with questions.
Rigid vs. movable work interactions;
Adiabatic vs. diathermal heat interactions;
Impermeable, semi-permeable, and permeable mass transfer.
Closed system

Open system

Isolated system enclosed by rigid, adiabatic, and impermeable boundary.

Simple system no internal boundaries (no rigid, adiabatic, or impermeable walls), and not subject to external force fields (no
potential energy) or inertial forces (no kinetic energy).
Oil
Composite system two or more simple systems.
Air
Phase a uniform region within a simple system.

Water

Restraints barriers within a system that prevent some changes from occurring within the time span of interest.
Examples: high activation energy in a simple system, internal boundaries in a composite system, etc.
H2 + O2 H2O

Chap. 2: Basic Concepts (II)

2. Thermodynamic state determined by a set of thermodynamic properties (state functions) required to reproduce the system.
State functions depend only on the present conditions of the system, however reached.
Classification of thermodynamic properties:
Primitive vs. derived measurability;
Intensive do not depend on system size (N: number of moles); denoted by small letters (except T, P, Cv, Cp).
Extensive scale with system size; denoted by capital letters.
Molar properties = Extensive / N
Unit Conversion: http://www.onlineconversion.com/
How many properties needed?
Postulate I: For closed simple systems with given internal restraints, there exist stable equilibrium states which can be
characterized completely by two independently variable properties in addition to the masses of the particular chemical species
initially charged. -- This is NOT the Gibbs phase rule!
Which two properties? Determined by thermodynamics.
Example: T and P in a mixture of H2 and O2.
For VLE of H2O in a closed container, P=Pvp is a function of T. (The other independent variable is
the amount of liquid or vapor.)
Reaching the stable equilibrium state
Postulate II: In processes for which there is no net effect on the environment, all systems (simple and composite) with given
internal restraints will change in such a way as to approach one and only one stable equilibrium state for each simple subsystem. In
the limiting condition, the entire system is said to be at equilibrium.
3. Change of state change in the value of at least one thermodynamic property, caused by interactions with environment or altered
internal restraints. Usually refers to systems going from one stable equilibrium state (where no change of state
can occur) to another.
Path all states the system traverses during a change of state.
Quasi-static path all states on the path are equilibrium states.
Thermodynamic process two end states, interactions with environment, and the path (usually quasi-static).

Chap. 2: Basic Concepts (III)

Reversible process a process is reversible if a second process could be performed so that the system and all elements of its
environment can be restored to their respective initial states, except for differential changes of second order.
A reversible process
is frictionless;
has no gradient in temperature, pressure, velocity, concentration, etc.;
is never more than differentially removed from equilibrium;
is driven by forces whose imbalance is differential in magnitude;
traverses a succession of equilibrium states;
can be reversed at any point by an infinitesimal change in external conditions;
when reversed, retraces its forward path, and restores the initial state of system and surroundings.

Chap. 3: The First Law (I)

1. Work (W)
Expansion/Contraction work change in system volume V.

P,
V

A
F due to Pout

Taking the gas in the cylinder as the system. The infinitesimal amount of work done on the system (by the external force F )
is W Fdl. The displacement dl d V A >0 for system contraction (dV 0) and 0 for system expansion (dV 0).
How to
V1
calculate?
The above minus sign is taken such that the work done on the system (W) is positive when the system is compressed (dV<0).
According to Newtons 3rd law, the work done by the system on the environment () is equal and opposite to the work done by the
environment on the system (+). This sign of work is taken as a convention.
V2

With F P out A, we have W P out dV . When P out P, we also have W PdV W PdV .

Shaft work WS no change in system volume.

Adiabatic work interaction no heat transfer between the system and its environment.
Work usually depends on path, but adiabatic work does not.
Postulate III: For any states 1 and 2, in which a closed system is at equilibrium, the change of state represented by 12 and/or
the reverse change 21 can occur by at least one adiabatic process, and the adiabatic work interaction between this system and its
environment is determined uniquely by specifying the end states 1 and 2.
E2 E1 = Wa1 2

E is a derived, extensive property of the system, usually called total energy.

Conservation of energy

2. Internal energy (U) energy contained in the molecules comprising the system. It includes molecular motions (translation, rotation,
vibration), intramolecular and intermolecular interactions, but not the energy due to macroscopic position or movement of the
system. Statistical mechanics is needed to relate these molecular motions and interactions to macroscopic properties of the system.
E = Kinetic energy + Potential energy + U For a simple system, E = U.
(inertial effects) (body force fields)
U is a derived, extensive property of the system. Its absolute value cannot be measured.
3. Heat (Q)
Q (E2E1) W = E W for a closed system.

W = Wa when Q0.

Convention: Q>0 if heat is added to the system to increase its energy.

Work and heat are energy in transit

across the boundary between the
system and its environment; they are
never stored or contained in the system.

Chap. 3: The First Law (II)

4. Thermal equilibrium no heat interaction between subsystems separated by a non-adiabatic wall.
How do we know if there are heat interactions? [Pair] Use thermometer (from any change of the liquid level in it).
Postulate IV: If the system sets of A-B and A-C each are at thermal equilibrium, then B-C are at thermal equilibrium.
Systems that are at thermal equilibrium have the same thermometric temperature ().
Convention: when A > B, heat goes from A to B. Thus,

dE
0.
d

5. The 1st law

E Q W for a closed system. E 0 for an isolated system.
How about open systems? [TAPPS] Read through and explain to your partner: What are states 1 and 2, and the term underlined?

A small amount of materials Nin (B) enters the open system A. The latter is enclosed by a
-surface. Q and W are not related to Nin.

Applying the 1st law to the closed composite system ANin, we get
E2 (E1einNin) Q W PinvinNin E E2 E1 Q W (einPinvin )Nin
Q

A
B
Nin @ Pin, molar volume vin,
and molar energy ein

If both A and B are simple systems, then EU, einuin. Defining enthalpy HUPV
(or huPv), we obtain the internal energy change of the open system as
U Q W hinNin
For multiple streams, U Q W hin N in hout N out ( N in 0, N out 0).
in

out

6. Ideal gas (IG)

PV = NRT and U(N,T) H(N,T)
In general, Cv

1 U
1 H

and C p
For IG, du Cv dT , dh C p dT , and C p Cv R.
N T V
N T P

Chap. 3: The First Law (III)

Examples 3.3 & 3.4 (slightly modified): [Read through the problem statement and make a drawing] An impermeable piston divides an
adiabatic cylinder of cross-sectional area S into two compartments A and B. Initially, A and B have the same length of 10cm, and both
are filled with gaseous helium at 300K. PAi2bar and PBi1bar. The piston is then moved quasi-statically to a final position where
PAfPBfPf. Assume that the gases are ideal with Cv12.6 J/(molK) and ignore the energy change of the piston and cylinder. What are
the final temperatures TAf and TBf if: (1) the piston is diathermal and frictionless; (2) the piston is diathermal and frictional; (3) the
piston is adiabatic and frictionless; (4) the piston is adiabatic and frictional.
For both (1) and (2), the composite system AB can be viewed as an isolated simple system (not if the gases are different). UAUB is
conserved and (for IG) depends only on temperature TATB. Therefore, TAfTBf300K.
For (3), applying the 1st law to A (closed system of IG undergoing adiabatic expansion/compression):
dU A QA WA

[TAPPS] Explain the following

derivation to your partner.

dU A N A Cv dTA

WA PA dVA (This is true only for a reversible process, which is quasi-static.)

VA

T
N A RTA
dVA N A Rd A
PA
PA

PA dTA TA dPA
NA R
PA 2

P T
dT
dP
Cv R A R A f Af
TA
PA
PAi TAi
UA+UB is conserved: U Ai U Bi U Af U Bf

Cv

R 1

. (This also gives PV

C p Cv

Const.)

U Af U Ai N ACv TAf TAi , same for B

N A TAf TAi N B TBf TBi 0.

Total volume is conserved: N A TAi PAi N B TBi PBi N A TAf Pf N B TBf Pf , N A RTAi PAi LAi S , and N B RTBi PBi LBi S .
Solving the above five equations (only NAS and NBS are relevant), we get Pf1.5bar, TAf267K, and TBf366K.
What if we use the 1st law on B instead of A? We get TBf353K! Why? The piston cannot be frictionless in order to be
moved quasi-statically without interacting with the environment! By applying the 1st law on A(B) as above, we implicitly
assume that the friction is on B(A) side.
For (4), we can only estimate TBf (366353)/2 360 K, since we dont know how the friction is distributed.

Chap. 3: The First Law (IV)

Example 3.5: [Read through the problem statement] A 4m3 storage tank containing 2m3 of liquid is about to be pressurized with air
from a reservoir through a valve at the top of the tank. The air in the reservoir is maintained at 100bar and 300K. The gas space above
the liquid contains initially air at 1bar and 280K. When the pressure in the tank reaches 5bar, the liquid transfer valve is opened and the
liquid is ejected at the rate of 0.2 m3/min while the tank pressure is maintained at 5bar. Neglect heat interactions at the gas-liquid and
gas-tank boundaries. It may be assumed that the gas above the liquid is well mixed and that air is an ideal gas with Cv20.9 J/(molK).
What is the air temperature when the pressure reaches 5bar (step 1) and when the liquid has been drained completely (step 2)?
How to choose the system and why? [Pair] Take air region in the tank as the (open) system.
[TAPPS] Explain the following derivation to your partner.
For step 1: Const. V 2m3

Air @
100bar, 300K

2m3

dU Q W hin N in
hin is constant and N in dN
dU d( Nu ) Ndu udN

du
dN
h ui N
. Integrate from the initial state i in

.
hin u N
hin u N i

PV
T 382 K.
RT
[TAPPS] Explain the following derivation to your partner.
For IG, hin u C pTin CvT and N

For step 2: Const. P 5bar

dU Ndu udN Q W hin dN

W PdV P( Ndv vdN )

Note that dh du Pdv
dh
dN
h hi N

.
. Integrate from the initial state i (the end of step 1) in
hin h N
hin h N i
For IG, hin h C p Tin T and N

2m3, 1bar,
280K

PV
T 336 K.
RT

Chap. 4: The Second Law (I)

1. Heat engine a closed device that converts between heat and work. It always undergoes a cyclic process such that any net effects
appear only in the external systems.
A (A>B)
QA5
E
(1)

(2)

E
(3)

WE5

QA6

E
(4)

QB5

WE6
E

(5)

QB6

A process is allowed () if we see it in real

systems. A process is not possible () if it
violates one or more of the postulates.

(6)

B
(1)
(2) (violates Postulate II)
(3)

(4) (identical to (2) if combined with (3))

(5) , but all the 3 cases with the reverse of any direction of heat interactions are
(6) (examples: refrigerators, air conditioners, etc.)

The 2nd law (conclusion from (4)): It is not possible to convert heat into an equivalent amount of work; some amount of heat must
be transferred to a second body or the environment in the process of converting heat into work.
Take engine (E) as the system:
WE work on the system: WE5<0, WE6>0.

QA (QB) Heat on the system from A (B): QA5>0, QA6<0, QB5<0, QB60.

Use arrow to indicate direction; all the numbers besides arrows are positive.
Net effects of combining (5) and (6) with A unchanged (i.e., QA5QA6): total heat from B on the system QB5QB6, total work on
the system WE5WE6 QB5+QB6) (conservation of energy).
When WE5WE6 < 0, such a combination is equivalent to (4) and is therefore not possible. Thus, it is required that WE5 WE6.
Define efficiency (): 5

WE 5
W
0, 6 E6 0.
QA 5
QA 6

Then, 5<1, 61, and for the above case of combination 56 (when 56, we call them reversible engines or Carnot engines).

Chap. 4: The Second Law (II)

2. Thermodynamic temperature (T)
Reversible engines have the (same) maximum efficiency (maximum work obtained or minimum work needed), which depends on
Q
both A and B, i.e., B 1 f , . (Use combination of (5) and (6) to prove). [Do one case]
QA
[TAPPS] Explain the following derivation to your partner.

WE2

QA1 QA 2
WE1 WE 2 WE 3

QC 2 QC3
QB1 QB3
QB1
f A , B
QA1

WE3

QC 2

f A , C
QA 2
QB3
Q

f A , C f C , B B1

QB3
QA 2
QA1
f C , B

f A , B
QC 3
f i
f j , i
f j

A (A>C>B)
QA2
QA1

E2

WE1

QC2

E1

C
QC3

QB1

E3
QB3
B

=1+ f A , B

f A f B
f
A

Carnot cycle ideal gas undergoes a thermodynamic cycle consisting of the following 4 processes:
(1) reversible, isothermal expansion; (2) reversible, adiabatic expansion; (3) reversible, isothermal
compression; (4) reversible, adiabatic compression.

Carnot

TA TB
TA

Chap. 4: The Second Law (III)

3. Entropy (S)

f T QB f , f B TB QA QB 0.

QA
TA
TA
TB
f A

In differential form:

QA
TA

QB
TB

0 For any reversible cycle:

Q
0.
T rev

depends only on states 1 and 2, not on the path.

T rev
Q
Q
Define S
, or d S
.
T rev
T rev

The above can be used to calculate S . When a system undergoes an irreversible process between two equilibrium states,
S is evaluated by applying the above to an arbitrarily chosen reversible process that accomplishes the same change of
state as the actual process.
S is a derived, extensive property of the system.
Example 4.3: Prove that S0 for an adiabatic process of a closed system.

QI0, WI
State 1

State 2

Q
QII, WII,
Proof: If the process (I) is reversible, then S
0.
reversible
T rev
Otherwise, we can construct a reversible process II that brings the system from 21.
During the cycle of III, total work done on the system is WIWII, and total heat to the system is QII (WIWII).
If QII0, this is equivalent to the heat engine (4), which is not possible.
If QII0, this is a reversible cycle because there is no net effects on both the system and environment. Process I must therefore
be reversible, which is contradictory.
\ QII0 S210 S120.
S0 also for an isolated system. Since the universe is an isolated system, Suniverse Ssystem Senvironment 0.

Chap. 4: The Second Law (IV)

4. Fundamental equations (FE) combining the 1st and 2nd laws
For a closed, simple, n-component system, U = U(S,V,N1,,Nn) (see postulate I).
n
U
U
U
S
V
d
d
dN i .
dU

S V , N i
V S , N i
i 1 N i S ,V , N
j[i ]

Subscript Ni means that all the number of moles

are hold constant; subscript Nj[i] means that all
the number of moles, except Ni, are hold constant.

Equation of states (EOS)

U

Define chemical potential i

i S ,V , N j [ i ]

dU T dS PdV i dN i .
U
U
, P
T

S V , N i
V S , N i

Entropy representation of the FE: S = S(U,V,N1,,Nn). [TAPPS] Explain the following derivation to your partner.

1 S
P

x z y
U
S
S

xyz 1 rule: 1.

T U V , N i T
V S , N i U V , N i V U , N i
y z x y z x
EOS

S
i U
S

FE contains all thermodynamic information about a

T N i S , V , N U V , N i
N

i U , V , N j [ i ]

j[i ]

system, while two EOS are needed to recover FE.

i
1
P
For IG, with d U NC v d T and d V NR d T P ,
dS dU dV dN i .
T
T
i T
this can be used to calculate S .
Example 5.1 (modified): Given EOS PvRT and ucRT (where c is constant), derive FE.

1
P
cR
R
du dv
du dv
T
T
u
v
Integrating from a reference state (denoted by subscript "0"), we get

ds

s s0 cR ln

u
v
U cV
R ln
S S 0 NR ln c .
u0
v0
U 0 V0

For monatomic IG, Cvc3/2.

Chap. 4: The Second Law (V)

5. Reversible work of expansion/compression in flow systems
P1, v1

Take the fluid in the device as the system:

P2, v2
Device

d U Q W hin N in hout N out

Steady flow: d U 0 and N in N out N
Q W hout hin N h2 h1 N .

Take a differential change of the fluid N in the device (between 1 and 2): d H Q W .
[TAPPS] Explain the following derivation to your partner.
For reversible process: Q T d S

dH dU PdV VdP
dH TdS VdP
dU T dS PdV
P2

This is used directly with the understanding that

the two assumptions employed here are valid.

W V d P N vd P W N vd P Ws W
P1

P2

N vd P.
P1

Engine work or Shaft work

(Work done by the device on the fluid)

v2

Total work done on the fluid N is Wfluid N P d v.

v1

Wfluid W N P d v N vd P N d(Pv ) N Pv
1 1 P2 v 2 .
v2

P2

v1

P1

Mass flow work on N (done by the fluid

after N at 1 and the fluid ahead of N at 2)

Example 4.6: Assume IG, diathermal wall, and reversible process. What is the total work done by the compressor to evacuate the tank?
[TAPPS] Explain the following derivation to your partner. In particular, what do the underlined parts mean?

Ws N vd P
Pa

P
N: differential amount of gas
W s N RT ln a .

RT
processed by the compressor;
Pt

For IG: v
P
Pt: tank pressure during the process.
Vt
N dN
d Pt
RT
0

0
0
Pt
Pt
P
W s Vt ln d Pt Vt Pt ln
Pt d ln t Vt P0 10 5 J.
P0
P0

Pa
Pa P
Pa
0

Pt

Vt=1m3,
P0=1bar

Pa=1bar,
T=300K
C

Chap. 4: The Second Law (VI)

6. Entropy balance for open systems
d Ns CV
dt

Q , j S 0.
Ns
G
fs
j T , j

CV: control volume (the region of space identified for analysis of open systems). fs: flowing streams. out in.
N in the first term represents the number of moles in CV. N in the second term represents the flowrate of the corresponding flowing
stream; it has nothing to do with the N in the first term. Likewise, the two s are completely different.
The third term represents the rate of entropy change of surroundings (due to the heat transfer between CV and its surroundings), where
Q is the rate of heat transfer into the system through the j th part of the control surface, T the temperature of the surroundings
, j

, j

associated with the j part of the control surface, and SG the rate of entropy generation.
th

7. Ideal work and Lost work

In a steady-state flow process, the minimum amount of work required or the maximum amount of work produced by the change of state

of the fluid is obtained when SG 0, i.e., when the process is completely reversible (both internally reversible within CV and reversible
.
heat transfer between CV and its surroundings). In such cases, with a uniform surrounding temperature T , we get Q (rev) T Ns

Substituting it into the 1st law,

d Nu CV

The lost work is defined as Wlost

Q W , we get the rate of ideal work W W (rev) Nh

T Ns
.
Nh
ideal

fs

fs
fs
fs
dt
W Wideal . It is the work wasted as the result of irreversibilities in a process. For a steady-state flow

Q T S 0.
process with a uniform surrounding temperature T , we have Wlost T Ns

G
fs

Chap. 5: Calculus (I)

1. Euler relation
[TAPPS] Explain the following derivation to your partner.
For 0, U ( S , V , N i ) U ( S , V , N i )
U ( S )
U ( V )
U ( N i )

U
Take the derivative with respect to , we get
( S )
( V )

i ( N i )
U
U
U
U
U
U

T,

P,

i TS PV i N i U .
( S ) S
( V ) V
( N i ) N i
i
Gibbs-Duhem equation

Euler relation d U T d S S d T P d V V d P i d N i N i d i
i

dU T dS PdV i dN i
i

S d T V d P N i d i 0.
i

For a single-phase, n-component, simple system, only n1 of the n2 intensive variables (T,P,i) are independent.
2. Legendre transform (LT)
y(0)(x) [U(S)]

For fixed V and N i (i 1, , n ), U U ( S ).

How to represent the FE using T

U
as the independent variable instead of S ?
S

d y (0 )
, then y ( 0 ) ( x ) x y (1) y (1) ( ) y ( 0 ) x
dx
d y (1)
The function y (1) ( ) is equivalent to y (0 ) ( x ), since x
.
d
Define

U ( S ,V , N i ) F (T , V , N i ) U TS

H ( S , P , N i ) U PV

1st order LT

G (T , P , N i ) U TS PV i N i
i

2nd order LT

y(1)
x [S]

Chap. 5: Calculus (II)

Total differential: d U T d S P d V i d N i

A thermodynamic potential expressed in terms of its

independent variables is referred to as the fundamental
equation (or generating function).

d F d U d(TS ) S d T P d V i d N i
i

d H d U d(PV ) T d S V d P i d N i
i

d G d U d(TS ) d(PV ) S d T V d P i d N i
i

3. Transforming derivatives
For an n-component mixture, there are totally (n2)(n1)2 independent 2nd order partial derivatives of the FE.
n1 for pure substance 3 independent 2nd order partial derivatives.

1 H
T 2G
s
Cp

T

N T P
N T 2 P
T P
1 V
1 2G
p

V T P V T P

1
V

isobaric heat capacity

isobaric thermal expansion coefficient

1 2G
V

isothermal compressibility

V P 2 T
P T

Calculus involved:

x
1
y
y z

x z
f
f z

x y z y x y
x z y
1
y z x y z x

chain rule

f
f

x
y x

x
y

x
y

Maxwell relation

xy

f yx

Chap. 5: Calculus (III)

Thermodynamic square
V

U
S

For total differentials, use if an arrow is pointing to the independent variable.

e.g., d U T d S P d V

G
F
S

T P
T V

G
H

For Maxwell relations, use if two arrows are pointing in different directions from the lower 2 variables.
e.g., T P

V S
S V

Order of the 3 variables must form a right angle; switch the lower 2 variables;
fill in the 4th variable (at a corner); check the sign.

Jacobian transformation

For f ( x , y ) and g ( x , y ), define

x y

( f , g )

( x, y )
g

x y

Properties:

( f , y)
f

x y ( x , y )
( f , g )
( g , f )
( f , g )

( x, y )
( x, y )
( y, x)
1
( f , g )

( x, y )
( x, y )
( f , g )
( f , g ) ( f , g ) ( z , w)

( x, y )
( z , w) ( x, y )

y x

f g f g

.
x y y x y x x y
g

y x

Example: [TAPPS] Explain the following derivation to your partner.

( s, v)
s v
s v

( s, v)
(T , P )
s
T P P T P T T P
Cv T
T
T
T
(T , v )
(T , v )
v
T v

(T , P )
P T
Tv p
1
v v
.

C p v T T

Cp
T
v T
T P T P
2

Chap. 5: Calculus (IV)

x
Procedure to reduce
, where x , y , z U , F , G , H , , S ,V , T , P :
y z
Bring potentials to numerator, and eliminate them using total differentials;

U
P

U G P G

S
V
T
P

P G
P G

G
G

P S
P V
P
G
G

S P
V P

T
T
V S

V S P

S
P V
T
P
T
T

S
S

S P
V P

G
, d vdP sdT .
N

Bring S to numerator, and eliminate it using Maxwell relations, Cv, or Cp (put dT under dS);

S
V
S

T
P T T P , S T

N
S
P S
V P V
C
p

T
T P
T

Bring V to numerator, and eliminate it using p or T;

T
P

V
P V

T
Cv C p

T T
p

Tv p 2

Goal: Express all derivatives in terms of p, Cp, T, and S, T, P, V, N (not in any derivative).

Chap. 6&7: Equilibrium & Stability Criteria (I)

1. Classification of equilibrium states
Depends on perturbations and constraints

Stable

Unstable

Metastable

Neutral

2. Extrema principles
Consider a closed system with uniform T and P, which is in thermal and mechanical equilibrium with its surroundings (reversible heat
and work interactions with the surroundings).

dU PdV TdS 0.

This applies to any process that the system undergoes (for reversible processes, the equal sign is taken).

Q surr

T
T
Q TdS
nd
surr
The 2 law requires dS dS 0
The 1st law gives dU Q PdV

dS surr

surr

For an isolated system to be at stable equilibrium, the entropy must have an maximum value with respect to any allowed variations.
Example: An isolated system of (1) and (2) with an internal wall that is diathermal, movable, and permeable to all n components.
[TAPPS] Explain the following derivation to your partner.

S S S
(1)

(1)

(2)

(2)

1
P (1)
i (1)
1
P (2)
i (2)
(1)
(1)
(1)
(2)
(2)
0 (1) U (1) V (1) N i (2) U (2) V (2) N i (2)
T
T
T
T
i T
i T

Constraints: U U (1) + U (2) 0

V V (1) + V (2) 0
N i N i (1) + N i (2) 0, for i 1,, n.
T (1) T (2)
Phase
P

1
P

1
S (1) (2) U (1) (1) (2) V (1) i(1) i(2) N i (1) 0 P (1) P (2)
T
T
T
T
i T
T
(1) (2) equilibrium
i
i
Diathermal, rigid, impermeable wall Thermal equilibrium. Diathermal, rigid, semi-permeable wall Membrane equilibrium.
(1)

(2)

(1)

(2)

Chap. 6&7: Equilibrium & Stability Criteria (II)

Equivalently, for a stable equilibrium system at constant S, V, and Ni, the total internal energy must be a minimum.
(Analogue: surface area and volume of a sphere.)

Consider a closed, pure system at fixed S , we have U (V ). Criterion of stability for U then requires
U 0 U (V V ) U (V V ) 2U (V )

For small V dV , Taylor expansion the above to 2nd order

U (V ) UV dV

UVV
U
2
2
dV U (V ) UV dV VV dV 2U (V )
2
2

UVV dV 0 UVV 0
2

1
2
2

Similarly, for fixed N , U N du d 2u 0, where d 2u uss ds 2usv dsdv uvv dv . For a stable equilibrium, du 0, d 2u 0.
2

1
u
T
T
Consider T ( s, v) dT
dT sv dv
ds
dv uss ds usv dv ds
uss
uss
v s
s v

1
usv 2
2
2
Substitute ds into d u and simplify d u
dT uvv dv 0
uss
uss

uss 0, uss uvv usv 2 0.

2

3. Stability criteria for other thermodynamic potentials U [ ] U x , where

1
1
U 2 [ ]
x

2
2

x
x 2

U
U [ ]
and x
,
x

2 F
2 F
2 0, 2 0;
T V
V T
2 H
2 H
2 0, 2 0;
P S
S P
2G
2G
0,

2
2 0.
T

P
P T

For extensive variables

For intensive variables

Chap. 6&7: Equilibrium & Stability Criteria (III)

4. For stable states
T
0 Cv 0,
uss
Cv
usv 2 2 f
1
P
uvv
2
0 T 0,

uss v T
v T v T
C p Cv

Tv p 2

Cv .

P
2
3
5.
0 is the limit of intrinsic stability, also called spinodal curve. In such a case, F 0 and F 0 is required for stability.

T
P
Both
0 and
v T

2P
2 0 define the critical point.
v T

RT
a
Example 7.2: Determine the spinodal curve and critical point for van der Waals fluid, with EOS given by P
2 , where a and
vb v
b are constants.

RT
2a
P
3 0

2
(v b)
v
v T

Spinodal curve

2P
2 RT
6a
4 0
2
3
v
v T (v b)
8a
27bR can be used to

a
estimate a and b
Substitute into EOS P Pc
2

27b
8Tr
3
T
P
v
Define Tr , Pr , and vr , the EOS becomes Pr
2.
3vr 1 vr
Tc
Pc
vc
Solve T and v from the above v vc 3b, T Tc

Chap. 6&7: Equilibrium & Stability Criteria (IV)

At constant T , d vdP 0 vdP 0 Pv 0 Pdv. For VLE, L V Equal-area construction.
0

P
0

3
d

a 1
b

e
2

f
g

3
h
i

0 0

1
j

4
2

0
2

5
v

0
1,4

5
5

In the above plots, 0~5 represent points on the van der Waals loop, and a~j represent the areas with efh.
[TAPPS] Express 0 at points 0~5 in terms of the areas a~j, and explain the above -v and -P curves.

Chap. 8: Properties of Pure Materials (I)

1. Theorem of corresponding states
Compressibility factor z

For most fluids, zc

Pv
(for IG z 1, for most liquids z 0.01 ~ 0.2).
RT

Pc vc
0.27 0.04 (for vdw fluid zc 0.375).
RTc

Theorem of corresponding states All fluids, when compared at the same Tr and Pr, have approximately the same compressibility
factor, i.e., z(Pr,Tr). This is supported by experimental observations, and can be derived using statistical mechanics.
Pitzer et al.: z z (0) Tr , Pr z (1) Tr , Pr ,
where the acentric factor log10 Pvp, r

Tr 0.7

1 measures the difference in structure between the materials of interest and

a spherically symmetric gas (e.g. argon), and is tabulated for lots of materials.
2. EOS
Cubic EOS: RK, RKS, PR,

These are based on vdw EOS, applicable to certain materials within a wide range (both vapor and liquid), and mainly empirical.
Virial EOS
z 1 B v C v2

pressure explicit P(T,v) or z(T,v)

1 BP C P 2

volume explicit v(T,P) or z(T,P)

where B

B
C B2
, C
,
RT
( RT ) 2

virial coefficients depend only on T .

This applies only to gases within low to intermediate pressure range, and can be derived from statistical mechanics.
3. Evaluating changes in properties between two states

Need: Cp (or Cv) data and PvT EOS (which provides p and T).
Path: (T1, P1, v1) (T1, P0, v) (T2, P0, v) (T2, P2, v2).
Why? For most substances, Cp (Cv) is measured at P1bar. It can therefore be viewed as for IG (denoted by superscript 0), which
depends on T only and Cp0 = Cv0 + R.

Chap. 8: Properties of Pure Materials (II)

Example 8.3: Calculate s using a volume-explicit EOS and Cp0(T) data.
Choose T and P as independent variables, then
0
T2 s
P2 s
s
s

d
P
d
T

T1 T
0 P T T dP
P1 P

T T1
P
2
0

C p0
s 0
s
v

,
T
P T T1
T P T P
s

P1

T2 C p
P2 v
v
dT
dP
dP

T1 T
0
T P T T1
T P T T2

v
Problem:
diverges at P 0 !
T P
dB
R
1

v
e.g. z 1 BP v RT B
.
BR RT
dT
P

T P P
R
Solution: Add and substract dP, that is,
0P
0
P1 v
T2 C p
P2 v
R
R
P
P
s

d
dT
dP R ln P 01 R ln P 02
P

0
0
T1 T
T P P T T1
T P P T T2
P
R ln 1
P2

4. Departure function (Residual property) difference between the property of interest in its real state at specified (T,P) and in an IG
state at the same (T,P).
[TAPPS] Explain the following derivation to your partner. In particular, how to get the underlined term?
P v
R
s R (T , P ) s (T , P ) s 0 (T , P )
dP.

0
T P P

h R (T , P )
s

v
At const. T , d h T d s v d P T
v dP v T
dP
T P
P T

h R (T , P ) h (T , P ) h 0 (T , P ) h (T , P ) h (T , P 0)

P0

dh

v
v T
d P.
T P

Chap. 8: Properties of Pure Materials (III)

RT

d P.
0
P

The above can be directly used for volume-explicit EOS. For pressure-explicit EOS, the integration needs to be over v.
g R (T , P ) g (T , P ) g 0 (T , P ) h R (T , P ) Ts R (T , P )

[TAPPS] Explain the following derivation to your partner. In particular, how to get the underlined terms?

Pv
zRT
vd z zd v
dP dz dv
P
At const. T , d P RT

2
RT
v
v
P
z
v
R
P v
P
z
v
v
1
dP
dz
dv
dv
g (T , P )

d P z 1
z 1 z 1
z 1 ln z z 1 .
0
0
1

RT
P
z
v
v
RT P
RT

v R v v 0 ( z 1)
R

P
gR
hR
z 1
hR
z 1 P g RT

dT
dP d

g R vR
hR
P
RT
RT 2
P T v
RT 2
T
v

d
d
d

P
T

2
RT
RT
RT

zRT
R z
P

T
z
v
T v v T v

g R RT
v
v z dv
gR
dv
1 z

1
z 1 ln z z 1

T
RT
v
z T v
T

v v

v z dv
hR
z 1

2
T
RT
T

v v
v z d v
v
sR hR g R
dv

z 1 T

1
ln
z 1

R
RT
v
T v v

( z 1)T d v
ln z .

v v

Dont want to do the calculations by hand? Use http://people.ds.cam.ac.uk/pjb10/thermo/pure.html.

Chap. 9: Mixtures (I)

1. Partial molar property
Molar fraction: xi

Ni
N
i,
Nj N

i , j , k : index of component;

xi 1.

means sum over all n components.

Z
Partial molar property: zi
, where Z is an extensive property (V , S , F , G , H , U ).

N i T , P , N j [ i ]

zi z i for pure i
gi i

subscript Nj[i] means holding all number of moles, except for i, constant.
(1) F U TS f i u i Tsi ;
0

H U PV hi u i Pvi .

d G

N k
dG SdT VdP k dN k d i
si d T v i d P k
dN k k d
.

N i T , P , N
k
k N i T , P , N
k
N i T , P , N j [ i ]

j[i ]
j[i]
(2)zi is an intensive property, thus depends on T, P, and x1, , xn1 (the composition of the mixture).
n 1
z
z
z
dzi i d T i dP i

T P , x
P T , x
j 1 x j

dx j .

T , P , x [ j , n ]

(3) Z is a function of T, P, and N1, , Nn, thus

Z
Z
dZ
dT
dP z j dN j .

T P , N j
P T , N j
j

subscript x means holding all molar fractions constant;

subscript x[j,n] means holding all molar fractions
constant except for components j and n.
[TAPPS] Explain the following derivation to your partner.

At constant T , P , x[i , n ], and N , divide both sides by d xi , we get:

Z
N j
x
zj
N zj j
N zi z n .

j
j
xi T , P , x [ i , n ], N
xi T , P , x [ i , n ], N
xi T , P , x [ i , n ], N
Therefore,
Z N
z
1 Z

zi z n .

x
x
N
x

i
i T , P , x [ i , n ]
i T , P , x[ i , n ], N
T , P , x [ i , n ]

Chap. 9: Mixtures (II)

Since z is a function of T , P , and x[ n ],
n 1
n 1
n 1
z
z
z
z
z

dz
d
d
d
d
d
d
T
P
x
T
P
z
x
z

i
i
i
n d xi

T P , x
P T , x
T P , x
P T , x
i 1 xi T , P , x [ i , n ]
i 1
i 1
dxn
z
z

dT
d P z i d xi .
T P , x
P T , x
i

(4) Z is an extensive property, and is a function of T, P, and N1, , Nn Euler relation: Z z i N i z z i xi .

i

(5) Gibbs-Duhem relation:

N dz
i

x dz
i

Z
Z

dT
dP ,

T P , N j
P T , N j
z
z

dT
d P , or
T P , x
P T , x

zi
0.

j T , P , x [ j , n ]

x x
i

Can be used to: calculate one unknown partial molar property from the others that are known;
check the thermodynamic consistency for a given data set;
smooth the only available set of data that are inconsistent.
2. Mixing rules
For cubic EOS: (empirical)

For virial EOS: (derived from theory)

Bm yi y j Bij .

bm xi bi ;

a m xi ai1 2 , or
i

a m xi x j aij , where aij 1 ij ai a j and ii 0.

i

Explain the double-sum for 1-, 2-, 3-, and n-component systems.

yi: for vapor composition,

exchangeable with xi.

For IG heat capacities:

C v0, m yi C v0,i and C p0 , m yi C p0 ,i .

i

Chap. 9: Mixtures (III)

Example 9.2: Derive an expression forvk using R-K EOS.

V
vk

N k T , P , N j [ k ]
R-K EOS: P

N k T ,V , N j [ k ]

.
P

V T , N j

RT
am
NRT

v bm
T v v bm V Nbm

Mixing rules: a m xi ai
i

N 2 am

TV V Nbm

N a m N i ai
i

N 2 a m 2V Nbm
NRT
P

2
2
V T , N j
T V 2 V Nbm
V Nbm

N 2 am

2 N i ai a k 2 N a m a k
N k T ,V , N
i

j[k ]

Nbm
bm xi bi Nbm N i bi
bk

i
i
k

T ,V , N j [ k ]

P
Nbk
1
1

RT

2
TV
V Nbm V Nbm
N k T ,V , N j [ k ]

2 N am ak
N 2 a m bk
.

2
V Nbm
V N bm

3. Mixing function

Z mix Z (T , P , N 1 ,..., N n ) N j z j (T , P , x1 ,..., xn1 ) z mix z (T , P , x1 ,..., xn 1 ) x j z j (T , P , x1 ,..., xn1 )

j

where superscript + denotes the reference state. If we choose it to be the pure component at the same temperature, pressure, and
state of aggregation of the mixture, then for component j: T+T, P+P, xj+1, xi+0 (ij), and thuszj+zj(T,P).
Partial molar mixing function

Z mix
z i

N i T , P , N j [ i ]

N j z j z j
j

z i z i .

N i

T , P, N j[i]

Chap. 9: Mixtures (IV)

Euler relation: Z mix N i z i z mix xi z i ;
i

n 1
z mix
z mix
z mix

Total differential: d z mix

d
T
d
P
d xi .

xi
i 1
T
P, x
P T , x
T , P , x [ i , n ]

z mix is analogous to z . For example,

Gibbs-Duhem re lation:

z
z
i xi d z i Tmix dT Pmix dP , or

P,x

T , x

z i
x

i i x
j

0.

T , P , x [ j , n ]

4. Ideal-gas (IG) mixtures and Ideal (ID) solutions

Pure IG i: i IG (T , P ) RT ln P g i IG (T , P 1)

IG mixture, component i: i (T , P , y ) RT ln pi g i (T , P 1) definitions of pure IG, IG mixture, and ID solution

ID solution, component i: i ID (T , P , x ) RT ln xi g i (T , P )

where P is the total pressure, and pi yi P is the partial pressure of i.

IG

(1) Proof for pure IG

IG

[TAPPS] Explain the following derivation to your partner.

i IG
RT
i RT ln P g i (T , P 1)

P
P T
IG
Pv RT .

We also know
v

P T

T G
G T
G TS
H
T P

Gibbs-Helmholtz relation:

2

2
2
T
T
T
T P
IG

IG

T
h

2
T
T P

i IG

h IG is a function of T only.

IG
i T
d g i IG (T , P 1)

RT ln P g i IG (T , P 1)

T
T
i IG RT ln P g i IG (T , P 1) gives an IG.

P d T

Chap. 9: Mixtures (V)

(2) For IG mixtures

[TAPPS] Explain the following derivation to your partner.

i IG
RT
RT
yi

RT

pi
P v IG yi vi IG
.
P T , y
P
i

IG
vi

i IG T
hi IG

2
T
T

P , y

IG
IG
IG
hi is a function of T only hi hi .
d g i IG (T , P 1)

T
dT

u i hi Pvi u i IG is a function of T only u i IG u i IG .

(3) For ID solutions

[TAPPS] Explain the following derivation to your partner.

g i T , P
i ID
ID

vi
P
P T , x
T , x

ID
ID
do not depent on composition. vi vi , hi hi .
ID
i ID T
g i T , P T
h

i2
T
T
T

P , x
P , x

ID
ID
g i g i RT ln xi si si R ln xi . Here vi , hi , g i , and si are for pure i at T and P.

Taking the reference state to be pure components at the same T , P , and state of aggregation as those of the ID solution,
ID
ID
ID
ID
ID
g ID xi g i RT xi ln xi , smix
R xi ln xi , hmix
0, vmix
0.
we get z mix
z ID xi zi , that is: g mix
i

Also: g i

ID

RT ln xi , si

ID

R ln xi , hi

ID

0, vi

ID

0.

Chap. 9: Mixtures (VI)

5. Fugacity and fugacity coefficient
IG

(1) i RT ln fi g i (T , P 1) for component i in a mixture

definition of fugacity
IG
i RT ln f i g i (T , P 1) for pure i

-- fi ( f i ) is equivalent to i , but is more convenient to manipulate than i ;

-- It has the unit of pressure;
f
f
-- lim i 1, lim i 1;
P0 P
P0 y P
i
ln fi
--
P

1 i
vi
vi
ln f i
,
.

RT
P
RT
P
RT

,
T
y
T
T , y

For pure i: the departure function i i IG RT ln

[TAPPS] Explain the following derivation to your partner.

fi
, where i IG RT ln P g i IG (T , P 1).
P

ln f i P
hi hi IG

T
RT

P
f
For i in a mixture: i i IG RT ln i , where i IG is now for IG mixture at T , P , and y .
yi P

ln fi yi P

P, y

hi hi IG
hi hi IG
.

RT 2
RT 2

f
f
(2) Fugacity coefficient: i i , i i .
yi P
P
ln i

1
vi

RT ln i

RT
P
T , y

RT

vi
d P. Similarly, RT ln i
P

RT

vi
dP.
P

Chap. 9: Mixtures (VII)

[TAPPS] Explain the following derivation to your partner.
If we have a pressure-explicit EOS, then

V
V V
P 1
V P
P
RT
P

P
V
RT
P
V
RT

d
d
d
d

0 N i
N i
0 P
N i
0 d ln P
P
V T , N j

,
,
,
,
,
,
T
P
N
T
P
N
T
V
N
j[i ]
j[i]
j[i]

V P
V
P
V P
RT
RT

d V RT ln z.

V
RT
V
RT
P
d
d
ln
d
ln

0
N i

V
V
N i T ,V , N j [ i ]

T ,V , N j [ i ]

RT ln i

(3) Lewis or Randall rule

For ID solution, iID RT ln xi g i (T , P ) RT ln xi RT ln f i g iIG (T , P 1) fi ID xi f i .

(4) Choose pure i in IG state at T and P 1 as the reference state, then

G mix N i g i N i g iIG (T , P 1) N i RT ln fi
i

RT ln fi is the partial molar mixing Gibbs free energy ( g i ) with the chosen reference state.
Gibbs-Duhem relation:

g mix
g mix
dT
dP
T P , y
P T , y

y d RT ln f
i

With g mix yi g i yi g i g iIG (T , P 1) , we get

i

h h IG
v
yi d ln fi yi i 2i d T yi i d P.
RT
RT
i
i

Chap. 9: Mixtures (VIII)

6. Activity, Excess function, and Activity coefficient
(1) RT ln ai g i (T , P , x1 ,..., xn 1 ) g i (T , P , x1 ,..., xn1 ).
-- The activity is used for condensed (liquid or solid) phase;
-- RT ln ai g i with a chosen reference state that is at the same temperature as the real solution;
f
-- ai i ;
fi

-- Taking the reference state to be pure i at the same T, P, and the state of aggregation as those of the real solution, fi f i (T , P ).
ID

For ID solution, RT ln ai ID g i RT ln xi ai ID xi .
(2) Excess function difference between the property of interest in its real state and an ID solution state that is at the same T, P,
and composition as the real state.

Z EX Z Z ID N j z j z jID zi EX zi zi ID .
j

For example: f

EX

u EX Ts EX , h EX u EX Pv EX , g EX h EX Ts EX .

EX
EX
ID
EX
Excess mixing function: Z mix
Z mix Z mix
Z N j z j Z ID N j z j Z EX z i zi EX , z mix
z EX .
j
j

EX
hmix .
Regular solution: si EX 0 for all components g EX h EX hmix

Athermal solution: hi EX 0 for all components.

(3) [TAPPS] Explain the following derivation to your partner.

a
f
f
i i i IDi
xi xi f i
fi
ID

EX

RT ln ai g i g i g i
Gibbs-Duhem relation:

x d ln
i

EX

RT ln xi g i

EX
g i RT ln i

hmix
v
d T mix d P.
2
RT
RT

EX
ln
h i
h i
i

2
RT
RT 2
T P , x
.

EX
ln i
v i
v i

RT
RT
P T , x

Chap. 9: Mixtures (IX)

(4) Two-suffix Margules equation
Assume a regular (sEX=0) binary solution with hmix being symmetric in composition, we then have
EX
EX
g mix
hmix
hmix

Cx1 x2 , where C is proportional to 1 T .

RT
RT
RT
ln 1 Cx22 and ln 2 Cx12 .

[TPS] Derive the following and gmix.

(5) Henrys law applies to adsorption of gases in liquids or solids.

In general, A

ai
f
i for a chosen reference state. As the real state approaches the reference state, A 1.
xi xi fi

Consider a binary solution of A and B. If we choose the reference state for solute A to be its infinite dilution (denoted
by **) in solvent B at the same T and P of the real solution, then as xA 0, A** 1. In this limit, we also have
xB 1 and thus B 1 (the reference state for B is still pure B at the same T , P , and state of aggregation as those of
the real solution).
fA ** depends only on T and P , and we denoted it by HA (Henry's constant).
In the region of small xA , we then have (approximately) fA HA xA xA A f A , where A lim A , and fB xB f B .
xA 0

Chap. 15: Phase Equilibrium and Stability (I)

1. Gibbs phase rule: For a simple system with n species and phases, in which reactions do not occur, we have
degree of freedom
f
2
(
n1
) (
1
)n 2n.
(# of independent intensive variables)
T and P # of independent mole
# of chemical potential
fractions in each phase
constraints for each species
e.g.: VLE for pure materials Pvp(T);
triple point of pure materials a unique point;
VLE or LLE for binary systems T and x1 are independent.
2. Solving phase equilibrium problems

fi fi

for all species i.

(1) Some cubic EOS can be used for both vapor and liquid phases to calculate fugacity coefficients;
(2) Activity coefficients can be used for liquid phases, i.e., fi L f i i xi or fi L Hi xi .
(3) For pure i in a condensed phase (liquid or solid) at T and P,

P v
Pvp ,i v
P v
1
1
1
f i Pi P exp i d P P exp i d P exp i d P
P
0
0
RT P
RT P

vp ,i RT P
P v
i (T , Pvp ,i ) exp i
P
vp ,i RT

P vi

P
P
T
P
d
(
,
)
exp
d P Pvp ,i .

vp ,i
Pvp ,i
i
vp ,i
RT

Poynting correction factor

Some approximations: Pvp , i (T ) is small (T , Pvp , i ) 1;

V
i

vi P Pvp , i
P vi

The condensed phase is nearly incompressible exp

d P exp
;
RT

Pvp ,i RT

Both vi and P Pvp , i

P vi

are small exp

d P 1.
Pvp ,i RT

Chap. 15: Phase Equilibrium and Stability (II)

(4) Raoults law An IG mixture in equilibrium with an ID solution (under the above approximations) Pyi Pvp ,i xi .
Modified Raoults law An IG mixture in equilibrium with a real solution (under the above approximations) Pyi Pvp ,i xi i .
(5) For VLE of pure materials,

d ln f

d ln f

ln f V
ln f V
ln f L
ln f L

dT
dP
dT
dP
hV hL

dP
T P
P T
T P
P T

V
L
d T VLE T v v
h h IG ln f
v
ln f

RT 2
T P
P T RT

d ln Pvp

d 1 T

hvp
R z vp

Antoine equation of the form ln Pvp A

This is called Clausius-Clapeyron equation when z vp

Pvp v V v L
RT

1.

B
is generally used to calculate Pvp (T ), with A, B and C being speciesT C

specific co nstants.
3. Phase transitions
For pure materials, the 1st-order phase transition refers to a point of change where the first and higher order derivatives of with
respect to T and P (e.g., h and v) have a discontinuity; the 2nd-order phase transition is where the second and higher order
derivatives (e.g., Cp and T) have a discontinuity, but not the first order ones.
4. Stability of a binary liquid mixture

2G

x
A
A A
0

2
N A T , P , N B N A T , P , N B x A T , P N A T , P , N B

A
NA

0.

N
xA
N N A xB
xA T , P

N2
N
N A T , P , N B N A

T , P , N B

g mix
g A
g B
g mix xA g A xB g B xA g A g A xB g B g B
g A g A xA
g B g B xB

xA T , P
xA T , P
xA T , P

Chap. 15: Phase Equilibrium and Stability (III)

g
g
g
x g
Gibbs-Duhem relation: xA A xB B 0 B A A
xB xA T , P
xA T , P
xA T , P
xA T , P
2 g mix
g A
g B
g A
xA




1
0.
2
xB
x A T , P x A T , P x A T , P x A T , P
2 g mix
Spinodal cu rve:
0.
2
x

T , P

2 g mix

2
xA
Critical point:
3
g mix
x 3
A

0
T , P

0
T , P

Cotangent line in the

A
Binodal curve: A
A B A B

plot of g mix vs. xA .

B B

Chap. 10: Statistical Thermodynamics (I)

1. Why statistical thermodynamics? estimate the
macroscopic properties from the behavior of molecules.
(1) Statistical thermodynamics deals with large
number (~1023) of molecules at thermodynamic
equilibrium states.
(2) It is based on a few number of postulates.
(3) It is the basis of molecular simulation (MC
and MD), which is independent and
complementary to experiment and theory.
2. Classical atomic system N identical spherical
particles each with position ri and momentum pi.
Phase space all possible (micro)states of a
N-particle system. Each state is defined by a 6Ndimensional vector that specifies ri and pi for each
particle i.
Configuration space {rN};
Momentum space {pN}.
3. Quantum mechanics and classical approximation
Heisenberg uncertainty principle the positions
and momenta of molecules cannot be specified
exactly at the same time.
The energy of molecules is not distributed
continuously, but divided up into discrete (quantum)
states m with energy level Em.

Chap. 10: Statistical Thermodynamics (II)

Postulate of quantum partitioning for an isolated system, the probability of occupying a state m, Pm, depends only on Em relative
exp Em
to the total energy of the system, i.e., Pm ( Em )
, where 1kBT is a constant and
exp

n
the summation is over all the states.

Classical approximation replace the discrete sets of states and energies by continuous distributions. This can be done safely when
Em+1Em << kBT.
translational modes: T > 50 K; rotational modes: T > 100 K; vibrational modes: need quantum treatment even at 300 K.
Under the classical approximation, the Hamiltonian of the system, which is equal to the total energy of the system, can be written as
N

the sum of the kinetic energy and the total potential energy of the system, i.e., r N , p N
the system in time follows Newtons second law of motion.

p
i 1

2m

2
i

r N , and the evolution of

Chap. 10: Statistical Thermodynamics (III)

4. Molecular dynamics simulations
dp i

F
r

i
r
i
r N (t t )
r N (t )
dt

N
p
d
r

p N (t )
p
(
)

t
t
i
i

mi dt

(1) Force fields model for particle interactions

Fij (r )

du ( r ) rij
, where r = rij .
dr r

Pairwise additivity assume only two-body interactions (not good at

liquid densities): r N u (rij ).
i

j i

12 6
Lennard-Jones 12-6 effective pair potential: u r 4 .
r
r
LJ

Reduced units: * 3 , *

, T*

k BT

, P*

P 3

. avoid redundant calculations; easy to spot errors.

(2) Periodic boundary conditions and Minimum image convention

Interaction potential is often truncated and shifted to: reduce computation; be used with PBC.
(3) Verlet algorithms
(4) What to compute? A macroscopic observable (e.g, T, P,) is the time average of its microscopic (instantaneous) counterpart.

Chap. 10: Statistical Thermodynamics (IV)

5. Ensemble A set of allowable states each with a probability corresponding to that appeared in the time evolution of the system at
equilibrium, subject to the macroscopic thermodynamic constraints imposed on the system.
Pm(rN,pN)drNdpN is the probability of finding the N-particle system in the vicinity of state m specified by (rN,pN) in phase space.
Postulate of ergodicity time average of any thermodynamic quantity A is equal to its ensemble average A Pm (r N , p N ) Am (r N , p N ).
m

(1) Microcanonical ensemble constant N, V, and E for each state.

The probability of being at any allowable state m is given by Pm = 1/Q(N,V,E), where the partition function
1
dr N dp N (r N , p N ) E is the total number of allowable states, and h is the Plancks
Q( N , V , E ) En ( N ,V ), E
3N
N !h
n

Classical approximation
constant. Note that the N particles in the system are indistinguishable, and piri h (here ix,y,z).
Connection to classical thermodynamics: S k B ln Q( N , V , E ) .
(2) Canonical ensemble constant N, V, and T; different states can have different energies.
View the ensemble as an isolated set of Ns systems, each having V and N. Use Nm to denote the number of systems at state m.

For a given distribution of {Nm}, the total number of different ways to arrange these Ns systems is t N n ! N n ! .
n
n

Now, lets look for a distribution {Nm*} such that t is maximized, i.e.,
constraints: N n N s and
n

N E
n

ln t
N m

0. This must be done under the following two

N m*

Et , where Et is the total energy. We use Lagrange multipliers and to transform this

0.
ln t N n N s N n En Et
n

n
Nm*

With the Stirling approximation: ln N ! N ln N N (for large N ), this reduces to:

constrained optimization problem into the following unconstrained one:

N m

ln N s ln N m* Em 0 N m* N s exp exp Em .
N n ln N n N n ln N n N n N n En
N m n
n
n
n
n
Nm*
N m* exp Em

,
The probability for a system to be at state m under this most probable distribution is therefore Pm ( Em )
Ns
Q( N ,V , )

Chap. 10: Statistical Thermodynamics (V)

1
where Q( N , V , ) exp exp En
dr N dp N exp (r N , p N ) is the canonical partition function.
3N
N !h

P (E ) 1
n

[TAPPS] Explain the following derivation.

Classical approximation

The ensemble average of energy is E En Pn ( En )

n

exp En

Q( N , V , )

dr

dp N (r N , p N ) exp (r N , p N )

dr

dp N exp (r N , p N )

ln Q

.
V , N

E
n exp En
1 ln Q
E
n V N

.
Define the instantaneous pressure n n , then n Pn ( En )
V , N
Q( N ,V , )
V N
n
E
2 ln Q

V N
V , N

E

1 ln Q
1 2 ln Q
1 E

,N

V , N

,N

,N

V , N
1
P
U
, where k B is the Boltzmann's constant.
Compare with the thermodynamic relation
T
P
k BT
V T
T V
[TAPPS] Explain the following derivation.

T
n Pn ( En ) 1 dn Pn ( En ) 0 n ln Pn ( En )dPn ( En ) d n Pn ( En ) ln Pn ( En )
exp En
1
Pn ( En )
En ln Pn ( En ) ln Q( N , V , T )
Q( N , V , T )

1
E En Pn ( En ) d E En dPn ( En ) Pn ( En )dEn ln Pn ( En ) ln Q dPn ( En ) Pn ( En )n dV (at const. N )
n

k BTd Pn ( En ) ln Pn ( En ) dV
S k B ln Q ( N ,V , E ) in microcanonical ensemble.
n

Universal
E
Compare with the thermodynamic relation dU TdS PdV S k B Pn ( En ) ln Pn ( En ) k B ln Q( N , V , T )
.
T
n
Use the thermodynamic relation F U TS F k BT ln Q( N , V , T ).

Fundamental Equation

Chap. 10: Statistical Thermodynamics (VI)

Fluctuations: E 2

E2 E

k BT 2 NCv .

E N1 2 E N

(3) Grand canonical ensemble constant , V, and T; different states can have different energies and number of particles.
View the ensemble as an isolated set of Ns systems, each having T and V. Use Nm(N) to denote the number of systems at state
m having N particles. For a given distribution of {Nm(N)}, the total number of different ways to arrange these Ns systems is

t N n ( N ) ! N n ( N )! .
n,N
n,N

Now, lets look for a distribution {Nm*(N)} such that t is maximized, i.e.,
constraints: N n ( N ) N s ,
n,N

( N ) En ( N ) Et , and

n,N

ln t
N m ( N )

0, under the following three

N m* ( N )

( N ) N Nt where Nt is the total number of particles. Similarly,

n,N

Lagrange multipliers , , and are used to transform this constrained optimization into the following unconstrained one:

0.
ln t N n ( N ) N s N n ( N ) En ( N ) Et N n ( N ) N N t
N m ( N )
n,N

n,N

n,N
Nm* ( N )
The results show that the probability for a system having N particles to be at state m under the most probable distribution is
Pm ( Em , N )

N m* ( N ) exp Em exp N

, where Q( , V , ) exp En exp N Q( N , V , ) exp N

Ns
Q( , V , )
n,N
N 0

is the grand canonical partition function.

d E d En ( N ) Pn ( En , N ) d Pn ( En , N ) ln Pn ( En , N ) dV d N
n,N

n,N

Compare with the thermodynamic relation dU TdS PdV dN

k BT

Use the thermodynamic relation PV U TS N PV k BT ln Q( , V , T ).

Chap. 10: Statistical Thermodynamics (VII)

The way to calculate fluctuations in simulations.

Fluctuations: [TAPPS] Explain the following derivation.

N2
N2

N 2 2N N N

N2 N

1
1
N 2 exp En ( N ) exp N

Q( , V , T ) n , N
Q( , V , T )

Q( , V , T )

N exp E ( N ) exp N
n

n,N

N
N

ln Q( , V , T )
N Q( ,V , T )
N
N
k BT

,V

,V

T ,V
,V

kT N
V 2 P

T .
Thermodynamic relation

N2 B

2
N V T , N
v
N T ,V

6. Factoring the partition function

Q( N ,V , T )

px 2 p y 2 pz 2
1

dpx dp y dpz exp

2
N !h3 N
mk
T
B

px 2
1

dpx exp

N !h3 N
2mk BT

1 2 mk BT
Q( N ,V , T )

N ! h2

3N 2

N N1 2

[TAPPS] Explain the following derivation to your partner.

1
1
dr N dp N exp (r N , p N )
3N
N !h
N !h3 N

Using the Gaussian integral:

,V

N 2
pi
N
i 1
dp exp 2mkBT

dr N exp (r N )

dr

exp (r N )

3N

dr

exp (r N )

dx exp ax bx
2

N
N
dr exp (r )

px, py, and pz are independent.

b2
exp (for a 0), we get
a
4a

Z ( N ,V , T )
h
,
where
the
thermal
deBroglie
wavelength

N !3 N
2 mk BT

ln Z ( N , V , T )
and the configuration integral Z ( N , V , T ) dr N exp (r N ) . P k BT
.
V

T ,N

EOS

Chap. 10: Statistical Thermodynamics (VIII)

7. Monatomic ideal gas
N

1 V
No interactions between molecules r 0 Q ( N , V , T )
.
N ! 3
N

IG

1 V exp( )
V
Q ( ,V , T ) Q ( N , V , T ) exp N

exp 3
3

N 0
N 0 N !

IG

IG

PV k BT ln Q IG ( , V , T )
ln Q IG ( , V , T )
N

( )

,V

PV N k BT .
IG
ln Q ( , V , T )

, where exp .

Note that N here is the number of particles, not number of moles!

R = NAkB where NA is the Avogadros number.

8. Maxwell distribution of velocity

Consider a particle in the system with a potential (generated by other particles in the system) acting on it. The probability that this
[TAPPS] Explain the following derivation to your partner.
particle has a momentum between p and pdp is given by

p2
1
p2
p2
exp
exp
exp

dr exp (r )

2mk BT
2m
2m
h3

dp
dp
dp, where p p .
P (p)dp
32

1
p2
p2
2 mk BT

dp exp
dr exp (r )
dp exp 2m
2m
h3

Gaussian distribution.
Integrate over the angles in spherical coordinates, where dp p 2 sin dpd d , we get

p2
4 exp

2
2mk BT 2

2
P ( p )dp d d P(p) p sin dp
p dp.
32
0
0
2 mkBT
Under this Maxwell distribution of velocity, the ensemble-average of kinetic energy per particle is

2
p
p2
3
P( p )dp k BT .

0 2m
2m
2

Chap. 10: Statistical Thermodynamics (IX)

9. Monte Carlo simulations
Used to estimate high-dimensional integrals, such as the ensemble average of A(r N ) which is independent of p N . In such cases,

dr dp A(r ) exp (r , p ) dr

d
r
d
p
exp

(
r
,
p
)

NVT

A(r N ) exp (r N )

Z ( N ,V , T )

(1) Sample mean method (Random sampling)

x2
x2 f ( x )
To estimate an integral I f ( x)dx
P( x)dx, where P ( x) is an arbitrary probability density function, we can generate
x1
x1
P
(
x
)

random numbers [ x1 , x2 ] that follows the distribution of P( x), then I

For example, if P ( x)

f ( )
.
P( )

1
, then I x2 x1 f ( ) .
x2 x1

Problem of estimating A using the sample mean method for liquids (which are at high densities): exp (r N ) is negligibly
small for most configurations generated due to the overlaping of particles.
How about generating configurations with P(r )
N

exp (r N )
Z ( N ,V , T )

? We don't know Z ( N ,V , T ) !

(2) Metropolis method (Importance sampling)

This method uses a Markov chain to generate a series of configurations with a probability proportional to P(r N ), i.e., we don't need
to know Z .
Let's denote the probability for the system being at state m by Pm , and the probability for the system to go from state m to n by mn .
For an ergodic system (where each state can be accessed through finite number of transitions from any other state), these two
probabilities are related by Pm mn Pn (balance condition) and
m

mn

1, where the sum is over all the states of the system. Any

transition matrix mn that satisfies these two requirements will give us the correct density of states Pm , i.e., repeated applications
of mn to an arbitrary probability distribution (initial configuration) make the resulting probability distribution converge to Pm .
This is a property of Markov chain. Note that the balance condition leaves the correct distribution Pm invariant.

Chap. 10: Statistical Thermodynamics (X)

Let's use a stronger requirement of detailed balance : Pm mn Pn nm

P
m

One solution of mn

mn

Pn nm Pn .
m

mm 1 mn

n m
where the underlying matrix of the Markov chain mn is
,
is:
P
P

m
n

for
and

n
m
symmetric, i.e., mn nm .
mn mn
P
P
P
P

for

n
m
mn n m

(3) Detailed balance

Consider the transition from state o to state n.
We can write on on con , where on is the
probability of proposing this transition and
con is the probability of accepting this
transition. The detailed balance then requires
c
P
Po on con Pn no cno on n no , which is
cno Po on
satisfied by using the acceptance probability
P
Pacc con min 1, n no . The detailed
Po on
balance is a sufficient but unnecessary
condition.

In canonical ensemble, Pn Po exp ,

where n o . With on no , we thus
have Pacc min 1, exp .

(4) Procedure of Monte Carlo simulations

Chap. 11: Lattice Theories for Binary Mixtures of Liquids (I)

1. Small-molecule liquids
(1) Approximations
1 lattice site 1 molecule (No empty sites). Total number of lattice sites NNANB, where NA (NB) is the number of molecules of
species A (B).
The lattice is rigid (vmix0 and GF).
Only nearest neighbor interactions. We use z to denote the number of nearest neighbors of a lattice site (lattice coordination
number). NAA, NBB, and NAB denote the total number of nearest A-A, B-B, and A-B pairs,
respectively. Note that zNA NAB 2NAA, zNB NAB 2NBB.

(2) General formalism

zN A
zN
BB

BB B AB AA
N AB U AA U BB N AB U ( N AB ),
2
2
2

zN
BB
.
where AA is the interaction energy of a nearest A-A pair, U AA AA A is the total energy of pure A, and AB AA
2
2
Define g ( N A ,N B ,N AB ) as the total number of ways to arrange the N A and N B molecules on the lattice such that there are N AB

U AA N AA BB N BB AB N AB AA

nearest A-B pairs;

g(N

,N B ,N AB )

N AB

N!
. Then,
NA ! NB !

Q( N , V , T ) g ( N A ,N B ,N AB ) exp U ( N AB ) Q pure g ( N A ,N B ,N AB ) exp N AB , where Q pure exp U AA U BB ;

N AB

N AB

NAB

F ln Q( N ,V , T ) Fmix ln Q( N ,V , T ) ln Q pure ln g ( N A ,N B ,N AB ) exp N AB .

(3) Ideal solution
ID
For ideal solution, 0. Therefore, Fmix
ln

N
N!
N A ln N A N B ln N B N ln N N xA ln xA xB ln xB , where xA A .
NA ! NB !
N

(4) Two-suffix Margules equation

Assume random mixing of A and B, then N AB N AB*

zN A N B
N!
EX
Fmix ln
N AB* Fmix
Nz xA xB G EX .
N
NA !NB !

Chap. 11: Lattice Theories for Binary Mixtures of Liquids (II)

2. Flory-Huggins theory for polymer solutions
We now have nA chains of length NA and nB small-molecules (solvent), and the total number of lattice sites N nANA+nB.
Lets put all the chains into the lattice first; the rest of the lattice is then filled with solvent molecules.
For the insertion of the i th chain, the total number of available sites for its 1st segment is N (i 1) N A . The 2nd segment must be
placed at an empty site that is a nearest-neighbor of the 1st segment. Assuming random mixing of the occupied and empty sites, the
probability of finding an empty site in the lattice at this moment is

N (i 1) N A 1
. So the total number of ways for placing the
N

N (i 1) N A 1
th
z. Similarly, the total number of ways for placing the j 1 segment (j 1, ,N A 1) is
N
N (i 1) N A j
z; this ignores the long-range (along the chain backbone) interactions, i.e., the chain can fold back onto itself .
N
N 1
N A 1
N (i 1) N A j z A N (i 1) N A !
N A 1
th

The total number of ways for placing the i chain is therefore i N (i 1) N A z
.

N
N iN A !
N
j 1
2nd segment is

1 nA
1 z
i
The total number of ways for placing all the chains is given by

nA ! i 1
nA ! N

nA N A 1

N!
.
N nA N A !

n
n
z

The entropy of the system is S k B ln k B nA ln A nB ln B nA N A 1 ln .

N
N
e

n N
n

S mix S S (nB 0) k B nA ln A A nB ln B .
This does not depend on z.
N
N

With the random mixing of A and B, U mix U U AA U BB

f mix

Fmix
k BT
ln 1 ln 1 1 .
N
NA

znA N A nB
n N
z
Nk BT 1 , where
and A A .
N
k BT
N