2012 Spring

Thermodynamics and Statistical Physics
http://rays.ustc.edu.cn/Business/Teaching/TSPhysics/2012Spring/
Huinan Zheng
hue@ustc.edu.cn
University of Science and Technology of China
2012 Spring
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
1 / 324
1IO I
I
I
9On14p
, 2008.
Greiner, Walter, L. Neise and H. Stocker, Thermodynamics and statistical
mechanics, Springer, New York, 1994.
Landau, L. D. and E. M. Lifshitz, statistical physics, part 1, 3rd edition,
Butterworth-Heinemann, Oxford, 1980.
K v6R[On < 1964
(1979).
Bases for grade

I
The homework is definitely and unquestionably the most important part of this
course
Your final grade will be based on your work with the following weighting,
Homework (%)
20
..
.
30
Huinan Zheng (USTC)
9On
Final exam (%)

80
..
.
70
00 :51 :25 2012 Spring
2 / 324
Vg I
nn
6
n
nnNX
*NX
O
9/
O/n
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
3 / 324
Thermodynamics I
1
Equilibrium and State Quantities

Introduction
Systems, phases and state quantities
Equilibrium and temperature the zeroth law of thermodynamics
Kinetic theory of the ideal gas
Pressure, work and chemical potential
Heat and heat capacity
The equation of state for a real gas
Specific heat
Changes of state reversible and irreversible processes
Exact and inexact differentials, line integrals
The Laws of Thermodynamics

The first law
Carnots process and entropy
Entropy and the second law
Insertion: Microscopic interpretation of entropy and of the second law
Global and local equilibrium
Thermodynamic engines
Eulers equation and the Gibbs-Duhem relation
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
4 / 324
Thermodynamics II
3
Phase Transitions and Chemical Reactions

Gibbs Phase Rule
Phase equilibrium and the Maxwell construction
The law of mass action
Application of the laws of thermodynamics
Thermodynamic Potentials
The principle of maximum entropy
Entropy and energy as thermodynamic potentials
The Legendre transformation
The free energy
The enthalpy
The free enthalpy
The grand potential
The transformation of all variables
The Maxwell relations
Jacobi transformations
Thermodynamic stability
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
5 / 324
Introduction
Introduction I
systems consisting of very many particles theoretical description
atoms and molecules is gases, fluids, solids or plasmas
the quantum gas of electrons in semiconductors or metals
burned-out suns
universe
common and very general laws
thermodynamic equilibrium
the microscopic point of view of statistical mechanics
classical thermodynamics the concepts of thermodynamics are very general
and to a great extent independent of special physical models
the task of thermodynamics
appropriate physical quantities
the state quantities characterize macroscopic properties of matter, so-called
macrostate
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
6 / 324
Introduction
Introduction II
relate these quantities by means of universally valid equations the equations
of state and the laws of thermodynamics
set up relations which seem to have general validity independent of the special
physical system under consideration the axiomatic laws of thermodynamics
certain state quantities formulate and substantiate the laws of
thermodynamics the energy law and the entropy law
supplemented by a variety of empirical established relations between the state
quantities the equation of state valid only for special physical system
sufficient to specify a few state quantities state variables all other state
quantities have certain uniquely defined values
not give reasons
based on only a few empirical theorems restriction
the central concept of heat by means of the statistical, thermal motion of
particles is not a subject of thermodynamics
microscope regime
equilibrium state
infinitesimal changes of state
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
7 / 324
Systems, phases and state quantities I

thermodynamic system an arbitrary amount of matter uniquely and
completely described by specifying certain macroscopic parameters
container physical walls surroundings
isolated systems the total energy E , the particle number N, and the volume
V
closed systems N, V , and, when in equilibrium with its surroundings the
energy will assume an average value which is related to the temperature of
the system or of the surroundings, T
open systems when in equilibrium with its surroundings mean values of
the energy and the particle number are assumed which are related to the
temperature and the chemical potential (defined below)
homogeneous
heterogeneous
phase
phase boundary
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
8 / 324
Systems, phases and state quantities II

a typical example a closed pot containing water, steam and air the phase
boundary in this case is the surface of the water one speaks of a gaseous
phase (steam and air) and of a liquid water
state quantities the macroscopic quantities which describe a system
the energy E , the volume V , the particle number N, the entropy S, the
temperature T , the pressure p, and the chemical potential
the charge, the dipole momentum, the refractive index, the viscosity, the
chemical composition, and the size of phase boundaries
the number of state quantities which are necessary for a unique
determination of a thermodynamic state is closely related to the number of
phases of a system (cf., the Gibbs phase rule)
it is sufficient to choose a few state quantities (state variables), such that all
other state quantities assume values which depend on the chosen state
variables
the equations which in this way relate state quantities are called equations of
state specified by empirical means
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
9 / 324
Systems, phases and state quantities III

In particular, we refer to the concept of the ideal gas, as a model for real
gases, but which allows for reliable assertions only in the limit of low density
two classes of state quantities
I
I
extensive (additive) state quantities

intensive state quantities
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
10 / 324
Equilibrium and temperature the zeroth law of

thermodynamics I
temperature is specially introduced for thermodynamics closely connected
with the concept of (thermal) equilibrium
equality of temperature of two bodies is the condition for thermal equilibrium
between these bodies
thermodynamic state quantities are defined (and measurable) only in
equilibrium
equilibrium state the one macroscopic state of a closed system which is
automatically attained after a sufficiently long period of time such that the
macroscopic state quantities no longer change with time.
Suppose that, in an isolated system, one brings two partial systems, each
formerly in equilibrium, into thermal contact (no exchange of matter) with
each other. Then one observes, in general, various processes which are
connected with a change of the state quantities, until after a sufficiently long
time a new equilibrium state is attained. One calls this state thermal
equilibrium.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
11 / 324

thermodynamics II
As experience has shown, all systems which are in thermal equilibrium with a
given system are also in thermal equilibrium with each other. Since this
empirical fact, which we will use as the foundation of our definition of
temperature, is very important, once calls it also the zeroth law of
thermodynamics
temperature a common intensive property systems in thermal equilibrium
with each other
a system whose thermal equilibrium state is uniquely connected with an easily
observable state quantity (i.e., a thermometer), is brought into thermal
equilibrium with the system whose temperature is to be measured
global and local equilibrium
the procedure of measuring the temperature is connected with an equation of
state, i.e., the relationship between the observed state quantity (volume,
resistance) and the temperature
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
12 / 324

thermodynamics III
exploit the fact that many different kinds of gases behave similarly if they are
dilute
define the thermodynamic temperature T with the
help of the volume of such a dilute gas as
T = T0
V
V0
(1.1)
at constant pressure and constant particle number

Today one usually takes the melting point of ice as T = 273.15K, where the
unit is name in honor of Lord Kelvin.
Historically, the unit of temperature was fixed by
defining the temperature of the melting point of
ice as 0 C and that of boiling water as 100 C (at
atmosphere pressure), which is Celsius scale.
The conversion to Fahrenheits scale is y [ C] = (5/9)(x[ F] 32)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
13 / 324

thermodynamics IV
If one plots the volume of a dilute gas versus the temperature in C, one
finds a crossing point with the abscissa at the temperature 273.15 C.
an idealized system (an ideal gas) the volume of which is just V = 0m3 at
the absolute temperature (simply called as the temperature) T = 0K
the kinetic theory of gases
the mean kinetic energy of particles
microscopic meaning
stationary state connected with an energy flux the surroundings have a
different temperature
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
14 / 324
Example the ideal gas I

noninteracting particles in 1664, R. Boyle, and
shortly later (1667) but independently from him, E.
Mariotte
pV = p0 V0 ,
T = const.
not until 1802 did Gay-Lussac

V =
T
V0 ,
T0
p = const.
relationship holds between pressure, volume and temperature

pV
p0 V0
=
= const.
T
T0
since the expression PV /T is an extensive quantity it must, under the same
condition, increase proportional to the particle number, i.e.,
pV
p0 V0
=
= Nk,
T
T0
Huinan Zheng (USTC)
pV = NkT or p = kT
9On
(1.2)
00 :51 :25 2012 Spring
15 / 324
Example the ideal gas II

where k = 1.380658 1023 J K1 is the Boltzmanns constant of proportionality.
This is the ideal gas law.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
16 / 324
Kinetic theory of the ideal gas I

temperature mean kinetic energy of the particles
velocity v
For a equilibrium state on average there will always be the same number of
particles in a certain interval d 3 v of velocity, although individual particles
change their velocities
it makes sense to ask for a probability that a particle is in the interval d 3 v ,
i.e., to speak of a velocity distribution in the gas which does not change with
time in thermodynamic equilibrium
the number of particles dN(v ) in the velocity interval around v
dN = Nf (v )d 3 v ,
f (v ) =
1 dN
N d 3v
(1.3)
where f (v ) is the velocity distribution, and of course it must hold that

R +
f (v )d 3 v = 1
the pressure of the gas originates from the momentum transfer of the
particles when they are reflected at a surface A
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
17 / 324
Kinetic theory of the ideal gas II

assume the z-axis of our coordinate system to be perpendicular to the area
A, a particle of velocity v which hits that area transfers the momentum
p = 2mvz
particles with velocity v hit the surface element A during a time dt all the particles, inside a parallelepiped
with basis area A and height vz dt
on the other hand,
dN = N
dV
f (v )d 3 v
V
(1.4)
if dV /V is the fraction of the total volume occupied by the parallelepiped

it holds that dV = Avz dt
the impulse per area A is
dFA dt = 2mvz dN = 2Nmvz2 f (v )d 3 v
Huinan Zheng (USTC)
9On
Adt
V
(1.5)
00 :51 :25 2012 Spring
18 / 324
Kinetic theory of the ideal gas III

the total pressure
Z
Z
Z +
Z +
1
N +
p=
dFA =
dvx
dvy
dvz f (v )2mvz2
A
V
(1.6)
since the gas is at rest, the distribution f (v ) cannot depend on the direction
of v , but only on |v |
R
R +
we can write the integral 0 dvz as 21 dvz ,
Z
pV = mN
d 3 v f (v )vz2
(1.7)
this integral represents nothing than the mean square of the velocity in the
direction perpendicular to the surface
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
19 / 324
Kinetic theory of the ideal gas IV

this mean value, however, has to be the same in all spatial directions because
of the isotropy of the gas,
Z

d 3 v f (v )vz2 = vz2 = vx2 = vy2
(1.8)
or, since v 2 = vx2 + vy2 + vz2 ,
2 1
2 1
2
2
2
v =
vx + vy + vz
vz =
3
3
(1.9)
we finally have
1
2 2
v = N hkin i
(1.10)
3
3

Here hkin i = 12 m v 2 is the mean kinetic energy of a particle
compare with the ideal gas law, Equation (1.2), it obviously holds that
hkin i = 32 kT
the quantity kT exactly measures this mean kinetic energy of a particle in an
ideal gas
pV = mN
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
20 / 324
Example Maxwells velocity distribution I

The functional form of the velocity distribution the isotropy of the gas f (v
can be a function only of |v |, or equivalently of v 2 the velocity distributions of
the single components (vx , vy , and vz ) are independent of each other
f (vx2 + vy2 + vz2 ) = f (vx2 )f (vy2 )f (vz2 )
(1.11)
The only mathematical function which fulfills the relationship (1.11) is the
exponential function, so that we can write f (v ) = C exp av 2 , where the
constants C and a must not depend on v , but may be arbitrary otherwise
assume f (v to be normalizable a < 0 Gaussian distribution of the velocity
components C can be determined from the normalization of the function f (vi
for each component,
Z +
Z +

dvi f (vi ) = C
dvi exp avi2
1=
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
21 / 324
Example Maxwells velocity distribution II

where ap
is written with a > 0 the value of this integral is well known,
i.e., C = a/ from Equation (1.10)
Z

kT =m vz2 = m d 3 v f (v )vz2
Z +
Z +
Z +
=m
dvx f (vx2 )
dvy f (vy2 )
dvz vz2 f (vz2 )
Z
=m
p
/a,
dvz f (vz2 )vz2
Z

a +
dvz exp avz2 vz2
=m

r Z +

a
=2m
dvz exp avz2 vz2
0
r
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
22 / 324
Example Maxwells velocity distribution III

substitute x = avz2 , we find, with dvz =
r
kT =2m
a 1
2a a
1
dx
,
2 a x
dxe x x
in terms of the -function, which is defined by

Z +
(z) =
dxe x x z1
0
With (1/2) =
, (1) = 1, and the recursion formula (z + 1) = z(z).

1
3
1m
m
kT =m
=
or a =
2
2 a
2kT
a
The velocity distribution in an ideal gas for a component vi therefore reads

r
r

a
m
mvi2
2
f (vi ) =
exp avi =
exp
(1.12)
2kT
2kT
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
23 / 324
Example Maxwells velocity distribution IV

and the total distribution is

m 3/2
mv 2
f (v ) =
exp
2kT
2kT
(1.13)
These expressions are normalized and fulfill Equation (1.11). We will see
Equations (1.12) and (1.13) follow from first principles in the framework of
statistical mechanics.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
24 / 324
Pressure, work and chemical potential I

other state quantities
amounts of matter in terms of particle number N multiples of Avogadros
number NA = 6.0221367 1023 the atomic mass unit u
1u =
1 12
m C
12
(1.14)
i.e., via the mass of one atom of the carbon isotopic 12 C. Avogadros number
is just the number of particles with mass 1u which altogether have the mass
1g,
NA =
1g
= 6.0221367 1023
1u
(1.15)
The quantity NA particles is also called 1 mole of particles
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
25 / 324
Pressure, work and chemical potential II

If a system consists of several kinds of particles, for instance N1 , N2 , . . ., Nn
particles of N species, the so-called molar fraction X is a convenient quantity
for measuring the chemical constitution,
Ni
N1 + N2 + + Nn
P
It always holds that i Xi = 1
Xi =
(1.16)
The pressure, in terms of a force which acts perpendicular to a known area A

p=
F
A
(1.17)
We therefore have [p] = N m2 = Pa as the unit. It has the same dimension

as energy density, since
1N m2 = 1kg m s2 m2 = 1J m3
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
26 / 324
Pressure, work and chemical potential III

For the ideal gas, the pressure is the product of the particle density and the
kinetic energy of the particles, i.e., the temperature. Therefore p = 23 e,
where e = hEkin i is the (kinetic) energy density of the ideal gas.
The energy the total energy of a system E the mean energy per particle
E /N the concept of work from mechanics
W = F i d s
(1.18)
if F i is the force exerted by the system and d s is a small line element. The
minus sign in Equation (1.18) is purely convention in thermodynamics we
count energy which is added to a system as positive, and energy which is
subtracted from a system as negative.
If one pushes the piston a distance d s further into the
volume against the force exerted by the system, the
amount of work needed is just
W = pAds > 0
(1.19)
Now Ads = dV is just the decrease of the gas volume

dV < 0 in the container,
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
27 / 324
Pressure, work and chemical potential IV

We may only consider an infinitesimal amount of work, since the pressure
changes during the compression. To calculate the total compressional work
one needs an equation of state p(V ).
It is a general property of the energy added to or subtracted from a system
that it is the product of an intensive state quantity (pressure) and the change
of an extensive state quantity (volume)
If the system contains an electric charge q, which gives rise to an electric
potential . If one wants to add another charge dq with the same sign to the
system, one has to perform an amount of work,
(1.21)
W = dq
The locally defined electric potential is the intensive quantity which describes
the resistance of the system against adding another charge, just as the
pressure is the resistance against a compression.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
28 / 324
Pressure, work and chemical potential V

Electric or magnetic dipole moment
Wel =E d D el
(1.22)
Wmag =B d D mag
(1.23)
Consider the work necessary to add another particle to a thermodynamic

system Our system should maintain equilibrium after adding the particle
The particle being added has to have a certain energy that is comparable to
the mean energy of all the other particles Define
(1.24)
W =dN
as the work necessary to change the particle number by dN particles.
The intensive field quantity is called the chemical potential
If the system consists of several particle species, each species has its own
chemical potential i , and dNi is the change in the particle number of species
i This is valid as long as the particle species do not interact with each other.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
29 / 324
Pressure, work and chemical potential VI

All different kinds of work have the generic property that they can be
converted into each other without restrictions. There is no principle objection
that these conversions do not proceed completely, i.e., with a rate of 100%,
although real energy converters always have loses.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
30 / 324
Heat and heat capacity I

The situation is completely different with another kind of energy which is of
principle importance for thermodynamics: heat
Heat is a special form of energy [Earl Rumfort, 1798; Davy, 1799; R. J.
Mayer, 1842]
It is a daily experience that work performed on a system (of mechanical or
electrical origin) often increase the temperature,
Q = CdT
(1.25)
C total heat capacity of the system

Formerly, one had the thermochemical calorie as the unit for heat the
amount of heat warms 1g water from 14.5 C to 15.5 C
C1g H2 O, 15 C = 1cal/ C
Joule (1843-49) showed that 1cal of heat is
equivalent to 4.184 Joule of mechanical work.
The SI unit for the heat capacity is also J/K.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
31 / 324
Heat and heat capacity II

Microscopic picture heat is energy which is statistically distributed over all
particles If the particles move in a completely disordered and statistical
manner, it is obviously not possible to extract all the kinetic energy by a
simple device
to change work into heat simpler than to gain utilizable work from heat
(thermodynamic engine)
One can define an intensive quantity, the specific heat c, via
C = mc
(1.26)
with m being the mass of the substance. It is also possible to define the
specific heat on a molar basis, C = ncmol , with n = N/NA . The quantity cmol
is the molar specific heat
cV and cp the specific heats at constant volume and constant pressure
cH2 O = 4.184J K1 g1 holds at constant atmosphere pressure
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
32 / 324
The equation of state for a real gas I

It is in general sufficient to fix a few state variables for a system. Then all
other quantities assume values that depend on the state variables.
pV =p0 V0 ,
T = const.
(1.27)
p = const.
(1.28)
or
V =
T
V0 ,
T0
As a standard example for a general equation of state connecting all relevant

variables the ideal gas law
pV =NkT
Huinan Zheng (USTC)
(1.29)
9On
00 :51 :25 2012 Spring
33 / 324
The equation of state for a real gas II

For a real gas, we can determine the value of Boltzmanns constant with the
help of Equation (1.29). In most cases one takes N = NA particles, i.e., just
one mole, and obtains
NA k =R = 8.31451J K1 mol1
(1.30)
R is named the gas constant.

One often assumes equations of state to be polynomials of a variable. If
Equation (1.29) is correct for low pressures (p 0), the ansatz
pV = NkT + B(T )p + C (T )p 2 +
(1.31)
should be more sophisticated equation of state for larger pressures.

As a first approximation terminates the virial expansion Equation (1.31) after
the linear term the coefficient B(T ) can be determined experimentally.
B(T ) the first virial coefficient
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
34 / 324
The equation of state for a real gas III

In terms of low density
pV = NkT + B 0 (T )
N
+ C 0 (T )
V
N
V
2
+
(1.32)
also called the virial expression

characteristic quantities the range of interaction r0 , and the depth of the potential U0
The potential vanishes for large interparticle
distances all these gases resemble an ideal
gas for low densities (large mean particle distance)
The potential has an attractive region for medium distances (' r0 ) while it is
strongly repulsive for small distances, caused by a large overlap of the atomic
electron clouds, as long as a chemical bond between these atoms is not
possible (no mutual molecular orbits)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
35 / 324
The equation of state for a real gas IV

Beyond this region, the atoms feel an attractive force (van der Waals
interaction)
simple gases quite similar behaviour
The equation of van der Waals (1873) the proper volume of the particles
V V Nb where b is a measure for the proper volume of a particle
Pid preal +p0 where p0 is the so-called inner pressure
Both dependences are in crude approximation proportional to the particle density N/V , so that p0 =
a(N/V )2 , where a is a constant.
van der Waals equation therefore reads
"
2 #
N
p+
a (V Nb) = NkT
V
(1.33)
a and b are material constants, which are mostly cited per mole and not per
particle
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
36 / 324
The equation of state for a real gas V

According to Equation (1.33), the pressure is
shown as a function of the volume for T =
const.. The best parameters a and b for water
are taken.
Typical values for the constants a and b are
given in Table
Material
H2
H2 O
N2
O2
CO2
a (Pam6 mol2 )
0.01945
0.56539
0.13882
0.13983
0.37186
b (103 m3 mol1 )
0.022
0.031
0.039
0.032
0.043
An often used approximation for van der Waals equation is obtained by

setting N/V in the inner pressure equal to that of an ideal gas, N/V p/kT

p2
p+
a
(V Nb) = NkT
(1.34)
(kT )2
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
37 / 324
The equation of state for a real gas VI

or
pV =
NkT
+ pNb
pa
1 + (kT
)2
(1.35)
For low pressure and high temperature we have pa/(kT )2 1,

a
p +
pV = NkT + N b
kT
(1.36)
expressed by the virial expansion, with B(T ) = N(b a/(kT )].
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
38 / 324
Specific heat
Specific heat I
The specific heats cV as well as
cp can be considered functions of
the state variable T and p, which
are the most easy to control experimentally. For dilute gases
(p 0), the specific heats are
mainly independent of pressure
and even approximately independent of temperature (at least for
rare gases)
Specific heat the ability of a substance to absorb energy in a statistically
distributed way this ability increases with the number of degrees of freedom
of a particle
The behavior of cp and cV for ammonia. The saturation line corresponds to
the gas liquid phase transition
The specific hear at constant pressure cp , is always larger than that at
constant volume cV
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
39 / 324
Specific heat
Specific heat II
cp and cV increase strongly, if one
approaches the phase transition from
gaseous to liquid ammonia (saturation
line) at constant pressure and decreasing
temperature.
cp and cV increase also (continuously)
with increasing pressure.
For liquids and solids, one almost always
quotes the value of cp which is easier to
measure.
Liquids show quite different dependences
on pressure and temperature (except, e.g.,
Hg and H2 O, with cp const.)
For metals, the law of Dulong and Petit
(1819) holds. All metals have the constant specific heat cp = 25.94JK1 mol1
over a wide range of temperature.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
40 / 324
Specific heat
Specific heat III

measurement of specific heat at very low temperatures quantum mechanics
and quantum statistics
Investigate an equation of state corresponding to an ideal gas equation for
solid. In this case,
V (T , p) =V0 [1 + (T T0 ) (p p0 )]
(1.37)
i.e., by a linear approximation. The constants and ,

1 V
=
V0 T p=p0

1 V
=
V0 p
(1.38)
(1.39)
V =V0
are called the coefficient of expansion (at constant pressure) and the
compressibility (at constant temperature), respectively.
The order of 105 K1 , while the order of 1011 Pa1 even small
changes in the temperature at given constant volume may effect a very high
pressure, i.e., large forces.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
41 / 324
Changes of state reversible and irreversible processes I

It is a daily experience that a process in an isolated system proceeds by itself
until an equilibrium state is reached irreversible
It is characteristic for irreversible processes that they proceed over
nonequilibrium states.
Processes which proceed only over equilibrium states are called reversible
an idealization nonexistent, strictly speaking
Reversible changes of state, however, can be simulated by small
(infinitesimal) changes of the variables of state sufficiently slowly
comparable to the relaxation time of the system quasi reversible
For every small step of the process the system is in an equilibrium state with
definite values of the state quantities
It is in general not possible to attribute values to the state quantities during
an irreversible process.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
42 / 324
Example isothermal expansion I

A constant temperature is practically realized by a heat bath
Irreversible process removing the external force
Reversible process decreasing the force at each step only by an infinitesimal
amount and wait for the establishment of equilibrium in the new situation
Consider ideal gas, p = NkT /V , the total amount of
work performed in the expansion of the system
Z
V2
W =
1
V2
pdV = NkT
V1
V1
dV
V
V2
= NkT ln
(1.40)
V1
This reversible work of the system is the maximum
work that can be extracted from the system there
is no way to gain more work from a system than reversible work
Real expansions, of course, lie between the extreme
cases of the completely irreversible expansion (W =
0) and the completely reversible expansion (W =
NkT ln(V2 /V1 ))
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
43 / 324
Example isothermal expansion II

This is also the case if we consider isothermal compression. For a reversible
process we need in this case the work
Z
V1
W =
2
pdV = NkT ln
V2
V2
V1
= NkT ln
>0
V2
V1
The work performed in the isothermal expansion depends on the way in which
the process is carried out, although the initial and final states are the same
A special case of the daily experience the work performed in a process, and
also the transferred heat, not only depends on the initial and final state of
the system, but also on the way of performing the process This, however,
means that work and heat are not suited to describing a macroscopic state in
a unique way They are not state quantities Mathematically, they are not
exact (i.e. total) differentials
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
44 / 324
State function I
A state quantity as a function of certain variables of state (e.g., T , p, etc.) a state function
As experience shows, the number of variables of state which are necessary to
uniquely determine a state, depends on the possible kinds of energy which
the system can absorb or emit
For many systems there are, for instance, heat Q and mechanical work
Wmech , as well as chemical energy Wchem To any of these kinds of energy
belongs a variable of state (e.g., T , V or N), and it is sufficient to determine
these three quantities to fix all other state quantities
Various particle species discussed in detail when studying the
thermodynamic laws and Gibbs phase rule
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
45 / 324
State function II
Restrict ourselves to functions of two state variables, for example
z = f (x, y )
(1.41)
One has often to be content with an implicit equation

f (x, y , z) = 0
(1.42)
It is characteristic of state quantities, and hence of state function They

depend only on the values of the state variables, but not on the way (i.e., on
the procedure) in which these values are assumed
If one changes the state variables by dx and dy with respect to the initial
values x and y , as is done in reversible change of state, for the change of z
we have

f (x, y )
f (x, y )
dx +
dy
(1.43)
dz =
x y
y x
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
46 / 324
Exact and inexact differentials, line integrals I

Start from a state function
z = f (x, y )
(1.44)
with the differential

f (x, y )
f (x, y )
dz =
dx +
dy
x y
y x
(1.45)
More general and mathematically convenient notation

df (x) = f (x) d x
(1.46)
where the vector d x = (dx, dy ).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
47 / 324
Exact and inexact differentials, line integrals II

A property of such total differentials which has enormous importance for
thermodynamics, is that the corresponding original function (state function)
can be obtained, up to an additive constant, via line integration along an
arbitrary curve
Z
f (x) f (x 0 ) =
f (x) d x
(1.47)
C
The curve C leads from x 0 = (x0 , y0 ) to x = (x, y ). If x(t) with t [0, 1] is

a parametric representation of this curve, the explicit calculation is done via
Z
f (x) f (x 0 ) =
dtf (x(t))
0
d x(t)
dt
(1.48)
When a given differential is total, or equivalently, under what conditions the

integration in Equations (1.47) or (1.48) does not depend on the integration
contour
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
48 / 324
Exact and inexact differentials, line integrals III

The existence of a potential a simple condition for deciding whether a force
F possesses a potential f (x)
F =0
(1.49)
or
Fy
Fz
= 0,
y
z
Fx
Fz
= 0,
z
x
Fy
Fx
=0
x
y
(1.50)
If F = f is valid, Equation (1.50) reduce to

2f
2f
= 0,
y z
zy
2f
2f
= 0,
zx
xz
2f
2f
= 0 (1.51)
xy
y x
This means nothing else than the right to interchange the sequence of
differentiation, which holds certainly for a function f (x, y , z) which is totally
differentiable.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
49 / 324
Example A simple differential I

Consider the differential
F d x = yxdx + x 2 dy
It is not exact, since then
Fy
Fx
= x 2x = x
y
x
(1.52)
should vanish, which is not the case. However

F d x = ydx + xdy
(1.53)
is exact, since
Fx
Fy
=11=0
y
x
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
50 / 324
Example A simple differential II

To this end we integrate along the contour
C1 = (x(t), y (t)) = (x0 + t(x x0 ), y0 + t(y y0 )),
t [0, 1]
and we have, according to Equation (1.48),

Z
f (x, y ) f (x0 , y0 ) =
n
dt [y0 + t(y y0 )](x x0 )
0
o
+ [x0 + t(x x0 )](y y0 )
1
=y0 (x x0 ) + (y y0 )(x x0 )
2
1
+ x0 (y y0 ) + (x x0 )(y y0 )
2
=xy x0 y0
Huinan Zheng (USTC)
9On
(1.54)
00 :51 :25 2012 Spring
51 / 324
Example A simple differential III

By differentiating, we readily confirm that

f
f
= y and
=x
x y
y x
Now we show that the same result is obtained via another contour C2

(t, y0 ),
t [x0 , x]
C2 = (x(t), y (t)) =
(x, t),
t [y0 , y ]
where we have to add the integrals over both parts of
the curve
Z x
dt(y0 1 + t 0)
f (x, y ) f (x0 , y0 ) =
x0
Z y
+
dt(t 0 + x 1)
y0
=y0 (x x0 ) + x(y y0 )
=xy x0 y0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
52 / 324
Example The integrating factor I

It is possible to construct an exact differential from a nonexact differential
F (x) d x by multiplication with an appropriate function g (x). Let g (x)F (x) d x
be the corresponding total differential, then we have for n variables (cf.
Equation 1.50))
[g (x)Fk (x)] =
[g (x)Fi (x)] ,
xi
xk
i, k = 1, 2, . . . , n
(1.55)
One calls g (x) the integrating factor. Again, let the differential be given,
F d x = yxdx + x 2 dy
We now try to determine g (x, y ) in such a way that
g F d x = g (x, y )yxdx + g (x, y )x 2 dy
is an exact differential. Then

g (x, y )x 2 =
[g (x, y )xy ]
x
y
Huinan Zheng (USTC)
9On
(1.56)
00 :51 :25 2012 Spring
53 / 324
Example The integrating factor II

Try to solve this via the product ansatz g (x, y ) = g1 (x)g2 (y ). Insert this in
Equation (1.56), it follows that
2xg1 (x)g2 (y ) + x 2 g2 (y )
dg2 (y )
dg1 (x)
= xg1 (x)g2 (y ) + xyg1 (x)
dx
dy
Divide this equation by xg1 (x)g2 (y ) 6= 0, it follows after rearranging terms that
1+
x dg1 (x)
y dg2 (y )
=
g1 (x) dx
g2 (y ) dy
(1.57)
It should be valid for all combinations x and y , which is fulfills only if both sides
of the equation have constant values, i.e., if
1+
x dg1 (x)
y dg2 (y )
=C =
g1 (x) dx
g2 (y ) dy
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
54 / 324
Example The integrating factor III

From
C 1
d ln g1 (x)
=
dx
x
and
d ln g2 (y )
C
=
dy
y
it follows that
ln g1 (x) = (C 1) ln x + K1
and
ln g2 (y ) = C ln y + K2
or
g (x, y ) = g1 (x)g2 (y ) = x C 1 y C K ,
K = e K1 +K2
Only want to determine a special function g (x, y ), we may choose C = 0,

K1 = K2 , i.e., g (x, y ) = x 1 . Thus our differential is now
gF dx =
1
(xydx + x 2 dy ) = ydx + xdy
x
This is just the differential of Equation (1.53), which we already know to be an

exact differential.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
55 / 324
Exercise
Exact and inexact differentials
Consider the differential
F d x = (x 2 y )dx + xdy
(1.58)
R
Is it exact? Calculate Ci F d x where Ci are the
contours from (1, 1) to (2, 2) in Figure. If it is not
an exact differential, what is the integrating factor?
Determine the original function.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
56 / 324
The first law
The first law I

Heat a special form of energy the discovery of R. J. Mayer (1842) The
perception of heat as energy which is statistically distributed among the
particles of a system [Clausius, 1857] statistical concept of the mean square
of the velocity and the ideal gas law from kinetic theory
The principle of conservation of energy experience asserts the assumption
that this principle is correct in macroscopic as well as in microscopic
dimensions
Thus we can assign an internal energy U to each macroscopic system
For an isolated system, U is identical to the total energy E of the system
known from mechanics or electrodynamics
If the system is able to exchange work or heat with its surroundings, an
energy law holds which is extended with respect to mechanics or
electrodynamics.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
57 / 324
The first law
The first law II

The change of the internal energy for an arbitrary (reversible or irreversible)
change of state is given by the sum of the work W and heat Q exchanged
with the surroundings.
First law: dU = W + Q
(2.1)
The work and heat exchanged with the surroundings in a small change of
state may depend on the way in which the procedure takes place they may
not be exact differentials Therefore we write for the changes to
distinguish them from exact differentials
Once again we explicitly remark that, e.g., the work has the form
Wrev = pdV only for reversible processes for irreversible processes it may
be that Wirr = 0.
The same holds for the exchanged heat Qrev = cV dT is only valid for
reversible processes
There exist many formulations for the first law of thermodynamics, which are
all equivalent
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
58 / 324
The first law
The first law III

1
The internal energy U of a system is a state function The total energy

content of a system is always the same for a given macroscopic state
There is no perpetuum mobile of the first kind A perpetuum mobile of the
first kind is an engine which permanently generates energy, but does not
change its surroundings
The change of the internal energy for an arbitrary infinitesimal change of state
is a total differential
The energy law holds independently from the procedure for reversible as well
as irreversible changes of state
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
59 / 324
The first law
Example Internal energy and total differential I

We have already derived for an ideal gas
pV = NkT =
2
N hkin i
3
where hkin i was the mean kinetic energy per particle.

In the case of the ideal gas the particles possess only kinetic but no potential
energy. U = hEkin i = N hkin i
U=
3
NkT
2
Huinan Zheng (USTC)
(2.2)
9On
00 :51 :25 2012 Spring
60 / 324
The first law
Example Internal energy and total differential II

Consider a container with an ideal gas at constant volume in a heat bath of
temperature T . If the temperature is changed by dT ,
dU = W + Q
On the other hand, the work exchanged with the surroundings is, because of
W = pdV = 0 V = const.
equal to zero. (Changes of state at constant volume are called isochoric).
Hence
dU = CV (T )dT
(2.3)
Here we have used the heat capacity CV at constant volume. Note that Q
can be integrated for the present procedure. For dilute gas the specific heat
is constant,
U(T ) U(T0 ) = CV (T T0 )
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
61 / 324
The first law
Example Internal energy and total differential III

The total heat capacity is proportional to the particle number, CV = NcV , we
find
U(T ) U(T0 ) = NcV (T T0 )
(2.4)
where cV is the constant specific heat per particle of the ideal gas. By
comparison with Equation (2.2) one obtains
cV =
3
k
2
orCV =
3
Nk
2
respectively.
With the help of Equation (2.3) we can determine the internal energy of real
gases from their measured specific heat. Quite generally,

U
CV =
T V
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
62 / 324
The first law
Example Adiabatic equations for the ideal gas I

We now ask for the relationship between the temperature and the volume of
an ideal gas if there is no heat exchanged with the surroundings an
adiabatic process
According to the first law, with Q = 0 and Wrev = pdV it holds that
dU = Wrev = pdV
for a reversible adiabatic process
From Equation (2.2) it holds that dU = CV dT . Therefore,
CV dT = pdV
Insert the ideal gas law for p(V , T ) we obtain
CV dT =
NkT
dV
V
(2.5)
This is a differential equation which describes the relationship between V and

T for adiabatic changes of state.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
63 / 324
The first law
Example Adiabatic equations for the ideal gas II

Since CV = const. we can integrate Equation (2.5) by separation of variables
from an initial state (T0 , V0 ) to a final state (T , V ),
Z
T0
CV dT
=
Nk T
V0
CV
T
V
dV
ln
= ln
V
Nk T0
V0
If we insert CV = 32 Nk and rearrange terms,

T
T0
3/2
=
V0
V
(2.6)
With the help of the ideal gas law, equivalent equations for relations between
p and V or p and T can be derived, respectively, for reversible adiabatic
process, e.g.,

T
T0
5/2
Huinan Zheng (USTC)
p
=
p0
and
p
=
p0
V
V0
5/3
9On
(2.7)
00 :51 :25 2012 Spring
64 / 324
The first law
Example Adiabatic equations for the ideal gas III

A specific process an adiabatic process exactly as for processes with
constant temperature (isotherms), constant pressure (isobars), or constant
volume (isochores) a variable of the ideal gas equation can be eliminated.
The total entropy of the system is constant (isentropes) for adiabatic,
reversible processes
The adiabates (isentropes) (pV 5/3 = const.) in a PV diagram are steeper
than the isotherms (pV = const.)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
65 / 324
The first law
Work and heat I

The first law holds independently of whether a change of state is reversible or
irreversible,
dU = Wrev + Qrev = Wirr + Qirr
(2.8)
The isothermal expansion of an ideal gas the absolute value of performed

work is larger for reversible processes than for irreversible processes
Compression the required work for a irreversible process is always larger
than for the reversible process
Taking account into the sign, that
Wirr > Wrev = pdV
(2.9)
In an irreversible process the absolute value of the (negative) waste heat

radiated off is always larger than in a reversible process and less heat is
needed
(2.10)
Qirr < Qrev

Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
66 / 324
The first law
Work and heat II

In other words, for reversible processes one requires the least work or
produces the most work especially, while for irreversible processes a part of
the work is always converted into heat which is radiated out of the system
Simultaneously the entropy of the system increases This increase of
entropy, however, cannot be reversed
Cyclic thermodynamic processes For a cycle, where a working material
regains its initial state after a series of changes of state, the equation
I
dU = 0
(2.11)
has to be fulfilled
If such a cycle nevertheless performs utilizable work, obviously a
corresponding amount of heat (extracted from the surroundings) has been
converted into this work.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
67 / 324
Carnots process and entropy I

This cycle, with an ideal gas as the working material,
was presented by Carnot in 1824.
The Carnot process is performed in four successive reversible steps, illustrated in a pV diagram
Step 1. Isothermal expansion
p1
V2
=
V1
p2
(2.12)
Consequently, it holds that

U1 = W1 + Q1 = 0
(2.13)
With the help of Equation (1.40),

Q1 = W1 = NkTh ln
V2
V1
(2.14)
This is amount of heat exchanged with the heat bath in the first step.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
68 / 324
Carnots process and entropy II

Step 2. Adiabatic expansion
V3
=
V2
Th
Tc
3/2
(2.15)
Since Q2 = 0 (for adiabatic processes), the work performed in the

expansion is taken from the internal energy,
W2 = U2 = CV (Tc Th )
(2.16)
For an ideal gas, CV = 3Nk/2 is a constant independent of temperature and

volume.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
69 / 324
Carnots process and entropy III

Step 3. Isothermal compression
p3
V4
=
V3
p4
(2.17)
The work performed during the compression is, because U3 = 0, submitted

to the heat bath in form of heat,
U3 =W3 + Q3 = 0
(2.18)
V4
Q3 = W3 = NkTc ln
V3
(2.19)
This is the amount of heat absorbed by the heat bath in this step.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
70 / 324
Carnots process and entropy IV

Step 4. Adiabatic compression
V1
=
V4
Tc
Th
3/2
(2.20)
Since Q4 = 0 it follows
W4 = U4 = CV (Th Tc )
(2.21)
The total energy balance of the process

Utotal = Q1 + W1 + W1 + Q3 + W3 + W4
{z
} | {z } |
{z
} | {z }
|
1
(2.22)
If we insert Equations (2.14), (2.16), (2.19), and (2.21), we immediately

recognize that indeed Utotal = 0, as it should be a cycle.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
71 / 324
Carnots process and entropy V

We have Q1 + W1 = 0 and similarly Q3 + W3 = 0, and furthermore
W2 = W4 . In addition
Q1 = NkTh ln
V2
V1
Q3 = NkTc ln
V4
V3
(2.23)
On the other hand, according to Equations (2.15) and (2.20), we have

V4
V3
=
V2
V1
or
V2
=
V1
V4
V3
1
(2.24)
Then, however, we have, according to Equation (2.23), that

Q1
Q3
+
=0
Th
Tc
(2.25)
This equation is valid not only for our special Carnot process, but according
to all experience, for any reversible cyclic process.
The quantity Q/T is also known as reduced heat.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
72 / 324
Carnots process and entropy VI

By infinitesimal parts of process, we may obviously
write instead of Equation (2.25)
I
Qrev
=0
(2.26)
T
The reduced heat Qrev /T is contour-independent and thus an exact
differential. In other words, 1/T is the integrating factor of the nonexact
differential Q.
Also if one integrates from state 1 to state 2 and back
again, it follow that
I
Z
Z
Q
Q
Q
=
+
=0
(2.27)
T
CA T
CB T
If one reverses the direction of integration on the curve CB (i.e., changes
R2
sign), one realizes that the integral 1 Q/T is contour-independent.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
73 / 324
Carnots process and entropy VII

Divide the arbitrary cycle into a sequence of infinitesimal Carnot-like parts
(N ), as shown in figure. All dashed parts are passed twice by
neighboring processes, but for each process in opposite direction
As can be experimentally confirmed, Qrev /T is an exact differential not only
for ideal gases, but for any other reversible thermodynamic process.
There has to exist a state function, the total differential of which is Q/T
the entropy S,
Qrev
,
dS =
T
Z
S1 S0 =
0
Qrev
T
(2.28)
In the TS plane the Carnot process is just a rectangle T = const. in steps 1

and 3 and S = const. in steps 2 and 4.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
74 / 324
Carnots process and entropy VIII

The Carnot process effectively performs a work W ,
W =W1 + W2 + W3 + W4
V4
V2
NkTc ln
= NkTh ln
V1
V3
(2.29)
and using Equation (2.24)

W = Nk(Th Tc ) ln
V2
= (Q1 + Q3 )
V1
(2.30)
This is a negative quantity. Hence, W is the work performed by the gas.

A Carnot engine is an engine which transforms heat into work.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
75 / 324
Carnots process and entropy IX

The efficiency of this engine the ratio between the heat transformed into
work and the total heat absorbed,
=
|W |
Q1 + Q3
Q3
=
=1+
Q1
Q1
Q1
(2.31)
If we insert Equation (2.25), we have

=1
Tc
Th Tc
=
Th
Th
(2.32)
There is in principle no heat engine with a better efficiency than

Equation (2.32). This impossibility of constructing such an engine leads us to
the formulation of the second law of thermodynamics.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
76 / 324
Entropy and the second law I

Entropy the state quantity was introduced by R. Clausius in 1850 defined
via Equation (2.28) as the amount of hear reversibly exchanged at a
temperature T
It holds (sign!) that
Qirr < Qrev = TdS
(2.33)
Especially for isolated systems, we have Qrev = 0.

In an isolated system the entropy is constant in thermodynamic equilibrium
(reversibility!), and it has a extremum because dS = 0. Every experience
confirms that this extremum is a maximum
All irreversible processes in isolated systems will lead into equilibrium are
connected with an increasing entropy, until the entropy assumes its
maximum, when equilibrium is reached.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
77 / 324
Entropy and the second law II

Second law For isolated systems in equilibrium it holds that
dS = 0,
S = Smax
(2.34)
and for irreversible processes it holds that

dS > 0
(2.35)
In irreversible processes the system strives for a new equilibrium state.

The entropy of a system can also decrease, if the system exchanges heat with
its surroundings For isolated systems, however, Q = 0 and in this case
Equation (2.34) is correct.
The entropy is obviously an extensive quantity
dU = Qrev + Wrev = TdS pdV + dN + dq +
(2.36)
The entropy just fits in the set of extensive state quantities (S, V , N, q, . . .)
which describe the change of the internal energy under the influence of
intensive, locally definable field quantities (T , p, , , . . .).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
78 / 324
Entropy and the second law III

If the function U(S, V , N, q, . . .) is given, we can determine T , p, , , . . . via

U
U
U
T =
, p =
, =
,
(2.37)
S V ,N,q,...
V S,N,q,...
N S,V ,q,...
The function U(S, V , N, q, . . .) gives one complete knowledge of the system
the fundamental relation Equations (2.37) are the corresponding
equations of state.
The intensive state quantities are therefore nothing but the derivatives of the
fundamental relation with respect to the corresponding extensive state
quantities.
One may also denote the entropy as a function of the other extensive state
quantities S = S(U, V , N, . . .).
The entropy is actually a new notion in thermodynamics
The state of equilibrium is defined as the state of maximum entropy - dS = 0.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
79 / 324
Example Entropy of the ideal gas I

We want to determine the entropy of an ideal gas at constant particle number
as a function of T and V For a reversible change of state the first law reads
dU = TdS pdV
(2.38)
for dN = 0.
With the equations of state
U=
3
NkT ,
2
pV = NkT
for an ideal gas we can solve Equation (2.38) for dS,

dS =
3
dT
dV
Nk
+ Nk
2
T
V
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
80 / 324
Example Entropy of the ideal gas II

Starting from a state T0 , V0 with entropy S0 = S(T0 , V0 ), we integrate this
equation,
3
T
V
S(T , V ) S(T0 , V0 ) = Nk ln
+ Nk ln
2
T0
V0
" #
3/2
V
T
=Nk ln
T0
V0
(2.39)
Substitute V T /p,
"
S(T , p) S(T0 , p0 ) = Nk ln
T
T0
5/2
p0
p
#
The equation does not contain the full dependence on N for systems with a
variable particle number
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
81 / 324
Example Entropy of the ideal gas III

To this end, we should have added the term dN in Equation (2.38), and we
would have to know the function (N, V , T ). Nevertheless, we may
conclude that the entropy has to be proportional to the particle number N,
i.e.,
(
" #)
5/2
p0
T
(2.40)
S(N, T , p) =Nk s0 (T0 , p0 ) + ln
T0
p
where s0 (T0 , p0 ) is an arbitrary dimensionless function of the reference state
(T0 , p0 ) will be directly calculated in statistic treatment.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
82 / 324
Microscopic interpretation
Microscopic interpretation I
The second law establishes a very close connection between statistical and
phenomenological point of view
All isolated physical systems converge toward an equilibrium state, where the
state quantities do not change anymore after a certain relaxation time This
process never reverses itself
The entropy is the state quantity that uniquely characterizes this tendency.
H-theorem (pronounced Eta-theorem, H = Greek Eta) by Ludwig Boltzmann
in 1872
In mathematical statistics, one can uniquely assign to an arbitrary random
event a measure for the predictability of this event. This function is
commonly denoted by H and is called uncertainty function
The uncertainty associated with an arbitrary nonequilibrium velocity
distribution can only grow or at least remain equal as a function of time.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
83 / 324
Microscopic interpretation II
The Maxwell-Boltzmann velocity distribution (equilibrium distribution) is
characterized by a maximum of the uncertainty function. The prediction of
the momentum of a particle in as gas at given temperature is associated with
the largest uncertainty.
Analogously, a homogeneous distribution of the particles in coordinate space
is associated with the largest uncertainty with respect to the prediction of the
coordinates.
An important consequence of the H-theorem an arbitrary (nonequilibrium)
distribution of particles changes, after a sufficiently long period of time, into
the Maxwell-Boltzmann velocity distribution and the latter is the only
possible equilibrium distribution
This seemed to contradict invariance with respect to time reversal, which is a
known principle of classical mechanics resolved by statistical investigation
Classical mechanics a motion of N particles is uniquely determined by the
3N coordinates and 3N momenta (q , p ) of the N particles at a certain time
the set (q , p ) is also called the microstate of the system one point in a
6N-dimensional space the phase space
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
84 / 324
Microscopic interpretation III

Consider the diffusion of a gas in an initial state (q (t0 ), p (t0 )) from a
smaller into a larger volume. If one is really able to reverse all momenta in
the final state (q (tf ), p (tf )) and to prepare a state (q (tf ), p (tf )), the
process would in fact be reversed.
Statistical point of view this is an event with an incredibly small probability.
There is only one point (microstate) in phase space which leads to an exact
reversal of the process. The great majority of microstates belonging to a
certain microstate, however, lead under time reversal to states which cannot
be distinguished macroscopically from the final state
The fundamental assumption of statistical mechanics all microstates which
have the same total energy can be found with equal probability the
microstate (q , p ) is only one among very many other microstates which
all appear with the same probability
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
85 / 324
Microscopic interpretation IV
We denote by the number of these microstates compatible with a given
macrostate The number of available microstates for a gas which
homogeneously occupies a volume V is overwhelmingly larger than the
number of available microstates compatible with a smaller volume
If we characterize the macrostate by the volume V available to our
N particles, the number of microstates (V ) available to one
particle is proportional to V . For N particles which are independent
from each other, the number of microstates available to each
particle have to be multiplied, so that (V ) V N .
The gathering of gas particles in one half of the container is impossible in
statistical terms it may be possible, but unimaginably improbable.
The number of microstates compatible with a given macrostate is a
quantity very similar to the entropy of this macrostate The larger , the
more probable is the corresponding macrostate, and macrostate with the
largest number max of possible microscopic realizations corresponds to
thermodynamic equilibrium.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
86 / 324
Microscopic interpretation V
Deviations from the equilibrium state are, for a finite number of particles, not
impossible (as thermodynamics claims), but only extremely improbable.
A connection between the entropy and the number of microstates compatible
with a macrostate tot = 1 2 and Stot = S1 + S2 only on mathematical
function the logarithm S ln
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
87 / 324
Example Microstates in a simple system I

Consider a container with volume V , which is homogeneously filled with N
particles of a gas in equilibrium
Imagine the container to be divided into two compartments with volume V1 and V2 , where V1 + V2 = V ,
and with N1 and N2 particles (N1 + N2 = N), respectively.
We set V1 = pV and V2 = qV . For the fractions p and q of the total volume
we of course have p + q = 1.
The total number of microstates compatible with N particles and the total
volume V , according to the binomial theorem,
tot (V , N) V

N
X
N
= (V1 + V2 ) =
V N1 V NN1
N1 1 2
N
(2.41)
N1 =0
On the other hand,

1 (V1 , N1 ) V1N1
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
88 / 324
Example Microstates in a simple system II

and
2 (V2 , N2 ) V2N2
are just the number of microstates compatible with the partial volumes V1
and V2 and the particle numbers N1 and N2 , respectively.
We may interpret the expression

N
(V1 , V2 , N, K ) =
V1K V2NK
K
as the number of microstates of the situation where K particles are in V1 and
N K particles in V2 . Furthermore we interpret

(V1 , V2 , N, K )
1 N
= N
(pV )K (qV )NK
pK =
K
tot (V , N)
V

N K NK
=
p q
(2.42)
K
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
89 / 324
Example Microstates in a simple system III

as the probability that there are just K particles in the fictitious volume V1
and N K particles in V2 .
With Equation (2.42) we can now immediately find the average particle
number in volume V1 .
=
K
N
X
pK K =
K =0
N
X
N
Kp K q NK
K
(2.43)
K =0
K
We write Kp K as p p
p , so that
N
X N K NK
p q
= p (p + q)N = Np(p + q)N1
K =p
K
p
p
K =0
= Np, or K
/N = V1 /V .
Since p + q = 1 we have K
This is of course obvious, since equilibrium corresponds to a homogeneous
distribution of particles.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
90 / 324
Example Microstates in a simple system IV

We can even calculate the fluctuations around this value.
For our example, with a total of
N = 20 particles and a volume
V1 = 0.6V . The measure square
definition, is defined as
)2 =
(K )2 (K K
N
X
)2 =
pK (K K
K =0
N
X
2
pK K 2 K
(2.44)
K =0
since
N
X
= 2K
2
2pK K K
K =0
and
N
X
pK = 1
K =0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
91 / 324
Example Microstates in a simple system V

The calculation of Equation (2.44) proceeds as in Equation (2.43), but with
K 2pK =

p
2
pK
and we have
2

(p + q)N (pN)2
(K )2 = p
p

=p
pN(p + q)N1 (pN)2
p

=p N(p + q)N1 + pN(N 1)(p + q)N2 (pN)2
We again insert p + q =, it follows that
(K )2 =pN + p 2 N(N 1) (pN)2 = pN p 2 N = Np(1 p) = Npq
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
92 / 324
Example Microstates in a simple system VI

This means that the width of the distribution, measure via
q
K = (K )2
increases with N. The relative width, i.e., the width K with respect to
the average particle number in the volume pV , is then
r
K
Npq
q 1
= Np = p N
K
This represents the fluctuation relative to the (mean) particle number.
Small deviations in small spatial regions are quite natural.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
93 / 324
Concerning the equal probability of all microstates

We started from the example of a gas gathered in a part of the volume of a
container, with values (q (t0 ), p (t0 )) for the coordinates and momenta.
We then assumed that in the course of time the coordinates and momenta
statistically attain some values and that we therefore find, by counting all
possible values (), the macrostate with max as the equilibrium state.
Exactly at this point, however, we assumed that all microstates (q , p )
compatible with our macrostate have equal rights, i.e., equal probability.
In an isolated system the macrostate is characterized by the total energy, the
particle number and the volume of the system. We determine by counting
all microstates compatible with these values of E , N and V .
Example It may of course be possible that these
microstates do not have equal probabilities.
Anyhow, we see that the equal probability of all compatible microstates is an assumption which can only be
justified by an experimental examination of the consequences
Up to now, no one has invented an experiment which disapproves this
assumption.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
94 / 324
Microstates belong to the macroscopic equilibrium state

A system which is initialized at time t0 in a nonequilibrium state assumes,
with high probability after a certain relaxation time, only microstates
(q (t), p (t)) which belong to the macroscopic equilibrium state since their
number is by far larger than the number of all other microstates.
For if certain microstates are never reached (not even approximately), we
actually must not count them in the calculation of !
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
95 / 324
Alternative statements and some consequences I

The second law of thermodynamics in different ways
1
There is no perpetuum mobile of the second kind. A perpetuum mobile of the

second kind is an engine which does nothing but perform work while cooling a
heat bath. Thus it is an engine which transforms heat into work with 100%
efficiency.
Each isolated macroscopic system wants to assume the most probable state,
i.e., the state which is characterized by the largest number of microscopic
realization possibilities.
Some consequences for the state variables T , p, , ,

. . . in an isolated system in equilibrium by both laws
Imagine the completely isolated system to be divided
into two parts
U1 + U2 =U = const.,
S1 + S2 = S = const.,
V1 + V2 =V = const.. N1 + N2 = N = const.,
Huinan Zheng (USTC)
9On
...
00 :51 :25 2012 Spring
(2.45)
96 / 324
Alternative statements and some consequences II

Remember the first law for a reversible change of state for both partial
systems,
dU1 = T1 dS1 p1 dV1 + 1 dN1 +
dU2 = T2 dS2 p2 dV2 + 2 dN2 +
(2.46)
Because of Equation (2.45), dU1 + dU2 = 0, dS1 = dS2 , dV1 = dV2 , . . .,

that
0 = (T1 T2 )dS1 (p1 p2 )dV1 + (1 2 )dN1 +
(2.47)
Since the change of the variables S1 , V1 , N1 , . . . in system 1 underlies no

restrictions, Equation (2.47) is true only if it separately holds that
T1 = T2 ,
p 1 = p2 ,
1 = 2 ,
(2.48)
These are necessary conditions for thermodynamic equilibrium.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
97 / 324
Alternative statements and some consequences III

If an isolated system is in equilibrium, it has everywhere the same constant
temperature, the same pressure and the same chemical potential, etc.
If there is, however, a real wall instead of the fictitious wall separating the
partial systems which, for instance, prohibits a change of volume or of the
particle number dN1 = 0, dV1 = 0 - then only the condition
T1 = T2
(2.49)
remains.
Correspondingly, the conditions (2.48) hold also separately or in combination,
if the partition is permeable only for certain changes of the state variables.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
98 / 324

Global equilibrium If a system is in thermodynamic equilibrium, i.e., if it has
everywhere the same temperature, the same pressure, and the same chemical
potential.
Local equilibrium If one can divide the whole system into small partial
systems which still contain very many particles and which are individually
approximately in thermodynamic equilibrium, these partial systems also can
be described by thermodynamic state quantities. These quantities will vary
from partial system to partial system.
The differences in temperature, pressure, and chemical potential affect heat
flow, volume changes, and particle fluxes. These fluxes are driven by the
corresponding potential differences and cause a compensation of these
potentials which leads, for an isolated system, to global equilibrium during
the course of time.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
99 / 324
Thermodynamic engines I
Cyclic heat engines play an extraordinarily large role in technics
The experience summarized in the second law asserts that the work
performed in reversible processes is smallest and the heat largest,
Wirr > Wrev = pdV ,
Qirr < Qrev = TdS
(2.50)
For the reversible or irreversible expansion (compression) of an ideal gas we

were able to verify the first inequality explicitly.
Quite analogously one understands the second inequality.
If we now have a cyclic engine which leads the working material back to the
initial state after one cycle, it holds
I
dU = 0
(2.51)
and thus
0 = Wrev + Qrev = Wirr + Qirr
Huinan Zheng (USTC)
9On
(2.52)
00 :51 :25 2012 Spring
100 / 324
Thermodynamic engines II
Of all possible processes, reversible processes produce
the largest amount of utilizable work, and requires the
smallest amount of work for given heat exchange Q.
The best efficiency of transforming heat into work is
reached by an engine working in reversible way
Each engine needs a heat reservoir (T = Th ) from
which to extract heat energy and a second reservoir
(T = Tc ) to absorb the waste heat of the process,
i.e., to cool the engine.
An engine which works with only one reservoir cannot
perform utilizable work in a cyclic process.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
101 / 324
Thermodynamic engines III

According to the first law
0 = W + Qh + Qc
The efficiency =
(2.53)
|W |
Qh
irr < rev =
W
Qh + Qc
=
Qh
Qh
(2.54)
Since the engine shall work reversibly, it holds that

Qh = Th dS,
Qc = Tc dS
(2.55)
dS 6== 0, although only equilibrium states occur. The engine is not an

isolated system!
We have
=
|W |
Th Tc
=
Qh
Th
Huinan Zheng (USTC)
(2.56)
9On
00 :51 :25 2012 Spring
102 / 324
Thermodynamic engines IV
Perpetuum mobile of the second kind Engine
A works here in the reverse direction. Engine
B works with higher efficiency, so that there
remains an amount of work WB WA .
WA =A QhA ,
WB =B QhB ,
QcA =QhA WA ,
QcB =QhB WB
(2.57)
Now adjust the engine in such a way that QhA = QhB = Qh , then there will
be no change in the hot reservoir on a long time scale. Then
QcA =(1 A )Qh > QcB = (1 B )Qh
(2.58)
since B > A . Thus the heat

Qc = QcA QcB = (B A )Qh
Huinan Zheng (USTC)
9On
(2.59)
00 :51 :25 2012 Spring
103 / 324
Thermodynamic engines V
is effectively drawn off from the cold reservoir. The engine performs the work
WB WA = (B A )Qh
(2.60)
while cooling the cold reservoir.

The vain efforts lasting for centuries to construct such an engine, which does
not contradict the energy law but rather the entropy law, resulted in the
recognition that QC = WB WA = 0, or
Th Tc
(2.61)
A = B =
Th
for all reversible processes at given Th and Tc .
Consider the work diagrams of some processes. The work performed per cycle corresponds to the hatched area
I
I
W = pdV = TdS
(2.62)
It is exactly as large as the difference of the
heats Qh = Th S and Qc = Tc S.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
104 / 324
Thermodynamic engines VI
Real processes, as for instance in an Otto-cycle engine, deviate more or less
from this diagram.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
105 / 324
Exercises I
Mixture temperatures
Calculate the range of possible final temperatures Tf in equilibrium for a system
consisting of two partial systems A and B, if A and B have initial temperatures
TA , TB and heat capacities CA , CB which are independent of temperature.
To this end, consider the limiting cases of a totally irreversible process (W = 0
and a totally reversible process (Wmax ). Calculate the mechanical work that one
can maximally extract from this system and the change of entropy of the partial
systems in the irreversible case. The final temperature Tf lies in between
q
C +C
Tf = A B TACA TBCB for reversible process
and
Tf =
C A TA + C B TB
CA + CB
Huinan Zheng (USTC)
for totally irreversible process
9On
00 :51 :25 2012 Spring
106 / 324
Exercises II
Room radiator
A room shall have a temperature of 21 C when the outdoor
temperature is 0 C. Calculate the relation between the
costs of heating if the room is heated with
1
electricity (100% efficiency)
a heat pump between T1 and T2 , if a fraction of

the energy is lost in the heat pump.
According to the scheme,

Q1 :
W :
Q2 :
Q3 :
the
the
the
the
heat flux which is extracted from the outside

power supplied to a heat pump
heat flux to supply the room
emission of heat from the room to the surroundings
Obviously,
Q3 = (T2 T1 )
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
107 / 324
Exercises III
By the first law, the energy fluxes (with respect to the heat pump) obey
W + Q1 + Q2 = 0
(2.63)
For reversibly working pump

Q1
Q2
+
=0
T1
T2
It thus follows that

T1
W + Q2 1
=0
T2
(2.64)
In the stationary case the heat flux Q2 has to compensate exactly the loss of
heat Q3 . Thus, with Q2 = Q3 we have

T1
T1
T1
W = Q2 1
= Q3 1
= (T2 T1 ) 1
T2
T2
T2
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
108 / 324
Exercises IV
If we consider that losses occur in the heat pump, the power Weff supplied to the
pump has to be larger by exactly these losses, so that

T1
W = Weff (1 ) = (T2 T1 ) 1
T2
For heating with a heat pump we therefore need the work

T2 T1
T1
hp
=
1
Weff
1
T2
When heating with a electricity directly, the power has to compensate the
dissipated heat Q3 ,
W el = (T2 T1 )
The relation between the powers is that

hp
Weff
1
T1
=
1
W el
1
T2
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
109 / 324
Eulers equation and the Gibbs-Duhem relation I

We assume that the system contains K particle species (chemical
components), each of which has, of course, a separate particle number and
chemical potential. Then
dU = TdS pdV +
K
X
i dNi
(2.65)
i=1
In general, an extensive state variable is proportional to the absolute size of

the system. This means that
U(S, V , N1 , . . . , NK ) = U(S, V , N1 , . . . , NK )
(2.66)
if is the enlargement factor. One calls functions which have this property
homogeneous functions of first order.
The intensive variables are homogeneous functions of zeroth order of the
extensive variables,
T (S, V , N1 , . . . , NK ) = T (S, V , N1 , . . . , NK )
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(2.67)
110 / 324
Eulers equation and the Gibbs-Duhem relation II

Consider an infinitesimal increase of the system ( = 1 + with 1),
U
U
U
S +
V +
N1
S
V
N1
U
+ +
NK
NK
U((1 + )S, . . .) =U +
(2.68)
If we insert this into Equation (2.66) and consider that according to

Equation (2.65)
U
= T,
S
U
= p,
V
U
= 1 ,
N1
...,
U
= K
NK
(2.69)
it follows that
!
U((1 + )S, . . .) = U + U = U + TS pV +
i Ni
(2.70)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
111 / 324
Eulers equation and the Gibbs-Duhem relation III

i.e., Eulers equation is valid,
X
U = TS pV +
i Ni
(2.71)
In other words, from Equation (2.66) it follows that Equation (2.65) may be
trivially integrated.
This is by no means obvious since according to Equation (2.69) T , p and i
are functions of S, V and Ni .
If we calculate the total differential of Eulers equation,
X
X
dU = TdS pdV +
i dNi + SdT Vdp +
Ni d i
(2.72)
i
and compare this with Equation (2.65), obviously the condition

X
0 = SdT Vdp +
Ni d i
(2.73)
must always be fulfilled. One calls Equation (2.73) the Gibbs-Duhem relation.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
112 / 324
Example Chemical potential of the ideal gas I

For only one particle species the Gibbs-Duhem relation has the form
0 = SdT Vdp + Nd
or
d (T , p) =
S(T , p)
V (T , p)
dT +
dp
N
N
If we here replace S(T , p) by Equation (2.40), with the help of V (T , p) = NkT /p,
(
" #)
5/2
T
p0
dp
d (T , p) = s0 k + k ln
dT + kT
(2.74)
T0
p
p
Since is a state quantity and has thus a complete
differential, we may integrate Equation (2.74) along
an arbitrary contour from (T0 , p0 ) to (T , p).

Z T
5
T
s0 k + k ln
dT
(T , p) (T0 , p0 ) =
2
T0
T0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
113 / 324
Example Chemical potential of the ideal gas II

Z
+ kT
p0
dp
p
(2.75)
and find the result

5
T
(T , p) =(T0 , p0 ) s0 k(T T0 ) kT ln
2
T0
5
p
+ k(T T0 ) + kT ln
2
p0
" #
5/2
T
p
=(T0 , p0 ) kT ln
T0
p0

5
s0 k(T T0 )
+
2
(2.76)
To add one particle to an ideal gas of temperature T and pressure p in

equilibrium, we have to summon the energy (T , p) according to Equation (2.76),
no matter how many particles are present before.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
114 / 324
Exercise
Eulers equation for an ideal gas
Show that for an ideal gas Eulers equation
U = TS pV + N
(2.77)
holds in general, as long as (T0 , p0 ) and s(T0 , p0 ), the additive, undetermined

constants of the chemical potential and the entropy, fulfill a certain relation.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
115 / 324
Gibbs Phase Rule
Gibbs Phase Rule I

We start from an isolated system which contains K different particle species
(chemical components) and P different phases (solid, liquid, gaseous, . . .).
Each phase can be understood as a partial system of the total system and
one can formulate the first law of each phase, where we denote quantities of
the i th phase by superscript i = 1, . . . , P.
For reversible changes of state we have
dU (i) = T (i) dS (i) p (i) dV (i) +
K
X
(i)
(i)
l dNl ,
i = 1, 2, . . . , P
(3.1)
l=1
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
116 / 324
Gibbs Phase Rule
Gibbs Phase Rule II

Altogether we therefore have P(K + 2) extensive state variables. If the total
system is in equilibrium, we have in addition the following conditions for the
intensive state quantities, cf. Equations (2.45)-(2.48),
T (1) = T (2) = = T (P)
p (1) = p (2) = = p (P)
(1)
(2)
(P)
l = l = = l , l = 1, . . . , K
Thermal equilibrium
Mechanical equilibrium
Chemical equilibrium
(3.2)
Equation (3.2) is a system of (P 1)(K + 2) equations. Thus we only require

(K + 2)P (K + 2)(P 1) = K + 2
(3.3)
extensive variables to determine the equilibrium state of the total system.

This number is independent of the number of phases.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
117 / 324
Gibbs Phase Rule
Gibbs Phase Rule III

If we now consider that exactly P extensive variables (e.g., V (i) , i = 1, . . . , P)
determine the size of the phases, one needs
F = (K + 2) P
(3.4)
intensive variables. Equation (3.4) is named after J. W. Gibbs and is called

Gibbs phase rule.
For instance, let us think of a closed pot containing a vapor. With K = 1 we
need 3 (= K + 2) extensive variables for a complete description of the
system, e.g., S, V , and N. One of these (e.g., V ), however, determines only
the size of the system. The intensive properties are completely described by
F = 1 + 2 1 = 2 intensive variables, for instance the pressure and the
temperature. Then also U/V , S/V , N/V , etc. are fixed and by additionally
specifying V one can also obtain all extensive quantities.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
118 / 324
Gibbs Phase Rule
Gibbs Phase Rule IV

If both vapor and liquid are in the pot and if they are in equilibrium, we can
only specify one intensive variable, F = 1 + 2 2 = 1, e.g., the temperature.
The vapor pressure assumes automatically its equilibrium value. If one wants
in addition to describe the extensive properties, one has to specify for
instance V li and V v , i.e., one extensive variable for each phase, which
determines the size of the phase.
If there are vapor, liquid, and ice in equilibrium in the pot, we have
F = 1 + 2 3 = 0. This means that all intensive properties are fixed
pressure and temperature have definite values. Only the size of the phases
can be varied by specifying V li , V s , and V v . This point is also called triple
point of the system.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
119 / 324
Gibbs Phase Rule
Gibbs Phase Rule V

If a system consists of various particle species, reactions between particles,
are often possible, which transform one species into another. For example,
2H2 + O2
2H2 O
(3.5)
In general we can write such reaction equations as

a1 A1 + a2 A2 + +
b1 B1 + b2 B2 +
(3.6)
The numbers ai and bi are the stoichiometric coefficients from chemistry.

Equation (3.6) is a condition for the particle numbers NA1 , NA2 , . . . and NB1 ,
NB2 , . . ..
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
120 / 324
Gibbs Phase Rule
Gibbs Phase Rule VI

Let dR be an arbitrary number of the type of equation (3.6). Then it must
hold that
dNA1 = a1 dR
dNA2 = a2 dR
..
.
dNB1 =b1 dR
dNB2 =b2 dR
..
.
Huinan Zheng (USTC)
(3.7)
9On
00 :51 :25 2012 Spring
121 / 324
Gibbs Phase Rule
Gibbs Phase Rule VII

As we already know, the equilibrium condition for an isolated system reads
1
p
1 X
dS = dU + dV
i dNi = 0
(3.8)
T
T
T
i
However, if U and V are constant in such a system, we have from

Equation (3.8) the condition
X
i dNi = 0
(3.9)
If we insert the dNi from Equation (3.7) into Equation (3.9), we have after
dividing by the common factor dR
X
X
ai i =
bj j
(3.10)
i
If we have, for instance, R reaction equations, we can formulate an extended

Gibbs phase rule,
F =K +2P R
Huinan Zheng (USTC)
(3.11)
9On
00 :51 :25 2012 Spring
122 / 324
Gibbs Phase Rule
Example Clausius-Clapeyron equation I

We want to derive a general equation to determine the vapor pressure of a
liquid in equilibrium with its vapor.
We have the following equilibrium conditions for two partial systems which
can exchange energy, volume, and particles,
Tli = Tv ,
pli = pv ,
li = v
If the equation of state is known and if we assume T and p to be given, we

can calculate li and v . The equation
li (T , p) = v (T , p)
(3.12)
yields a dependence between T and p
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
123 / 324
Gibbs Phase Rule
Example Clausius-Clapeyron equation II

If we change the temperature by dT in Equation (3.12), the vapor pressure
also has to change by a certain amount dp to account for equilibrium. For
the corresponding changes d li and d v , it must hold that
d li (T , p) = d v (T , p)
This can be expressed with the Gibbs-Duhem relation SdT Vdp + Nd = 0
in the following way
Vli
Sli
dT +
dp
Nli
Nli
Sv
Vv
d v (T , p) =
dT +
dp
Nv
Nv
d li (T , p) =
or with sli = Sli /Nli , vli = Vli /Nli and analogously for the vapor,
dp(vli vv ) =dT (sli sv )
dp
sli sv
=
dT vli vv
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
124 / 324
Gibbs Phase Rule
Example Clausius-Clapeyron equation III

This is the Clausius-Clapeyron equation.
At a give evaporation temperature, this entropy difference corresponds to an
amount of heat Qli v , which has to be added to evaporate all particles
from the liquid into the vapor phase.
Changing to the corresponding intensive variables and referring to a certain
amount of material, e.g., per particle or per mole,
sv sli =
Qli0 v
Sli
Sv
=
Nv
Nli
T
Now Qli0 v = Qv /Nv Qli /Nli is the amount of heat required to evaporate
one particle.
In many cases and for not too large temperature differences, this evaporation
heat may be considered to be constant.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
125 / 324
Gibbs Phase Rule
Example Clausius-Clapeyron equation IV

With the evaporation heat per particle Qli0 v we thus obtain
Qli0 v
dp
=
dT
T (vv vli )
(3.13)
where vv and vli are the volume per particle in the vapor and liquid phases,
respectively.
For an ideal gas the volume attained by
NA particles at room temperature and a
pressure of one atmosphere is 22, 400cm3 ,
while a liquid like H 2 O attains only a
volume of 18cm3 under these conditions.
Thus, in many cases vv vli and one has
Qli0 v
dp
'
(3.14)
dT
Tvv
However, the approximation vv vli becomes vary bad if one approaches the
critical point.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
126 / 324
Gibbs Phase Rule
Exercise
Vapor pressure of a liquid
Determine the vapor pressure of a liquid in equilibrium with its vapor under the
assumption that the evaporation heat per particle does not depend on pressure or
temperature and that the vapor behaves as an ideal gas.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
127 / 324
Phase equilibrium and the Maxwell construction I

The isotherms of van der Waals equation

N 2a
p + 2 (V Nb) = NkT
V
(3.15)
show regions of negative pressure as well as mechanically unstable regions having p/V > 0, where the
gas wants to compress itself. Both cases are certainly
unphysical.
These contradictions can be resolved by considering the phase transition from
gas to liquid.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
128 / 324
Phase equilibrium and the Maxwell construction II

In equilibrium between vapor and liquid, however, a certain vapor pressure pv
is established, which we have already calculated for an ideal gas from the
equilibrium conditions,
pli = pv ,
Tli = Tv ,
li (T , p) = v (T , p)
(3.16)
The vapor pressure pv is solely a function of temperature and does not

depend on the vapor volume, so that one obtains a horizontal isotherm in the
pV diagram.
It is remarkable that neither the density of the liquid (given by N/V2 ) nor the
density of the vapor (given by N/V1 ) changes during this phase transition.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
129 / 324
Phase equilibrium and the Maxwell construction III

The internal energy U(T , V ) at fixed particle number is a state function,
which depends only on the volume for a given temperature. For constant
temperature we therefore have (integrating Equation (2.36) for dN = 0)
Z
V2
U = T (S2 S1 )
p(V )dV
(3.17)
V1
Since U has an exact differential, it cannot matter whether U is calculated

along the direct path of constant pressure or along the van der Waals
isotherm, for which the following holds
p(V ) =
NkT
aN 2
2
V = Nb
V
(3.18)
In the first case we simply have

U1 = Q pv (V2 V1 )
Huinan Zheng (USTC)
9On
(3.19)
00 :51 :25 2012 Spring
130 / 324
Phase equilibrium and the Maxwell construction IV

where Q = T (S2 S1 ) is the latent heat of the phase transition. And in
the case of the van der Waals isotherm we have

V2 Nb
1
1
U2 = Q NkT ln
N 2a
(3.20)
V1 Nb
V2
V1
From the condition U1 = U2 ,
V2 Nb
N 2a
pv (V2 V1 ) = NkT ln
V1 Nb
1
1
V2
V1

(3.21)
One can, in principle, determine the unknown pressure pv as well as the also
known volumes V1 and V2 .
Equation (3.21) can be understood far more
easily. It tells us that the area pv (V2 V1 )
of the rectangle betwen V1 and V2 below
the unknown vapor pressure equals the
area below the van der Waals isotherm.
Maxwell construction
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
131 / 324
Phase equilibrium and the Maxwell construction V

Boundary of the phase coexistence region and critical point K Above the
critical temperature (the critical isotherm) a distinction between the gaseous
and liquid states is no longer reasonable!
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
132 / 324
Phase equilibrium and the Maxwell construction VI

The critical point is characterized by the fact that both derivatives vanish
(saddle point)

p
2 p
= 0,
=0
(3.22)
V Tcr ,Vcr
V 2 Tcr ,Vcr
or
2aN 2
NkTcr
= 0,
+
2
(Vcr Nb)
Vcr3
NkTcr
2aN 2
6 4 =0
3
(Vcr Nb)
Vcr
(3.23)
Thus
Vcr = 3Nb
(3.24)
and
Tcr =
8a
27kb
Huinan Zheng (USTC)
(3.25)
9On
00 :51 :25 2012 Spring
133 / 324
Phase equilibrium and the Maxwell construction VII

It finally follows with van der Waals equation from Vcr and Tcr that
pcr =
a
27b 2
(3.26)
The critical state quantities are therefore uniquely determined by the

parameters a and b. Hence, for all gases one should have
pcr Vcr
3
= = 0.375
NkTcr
8
(3.27)
Experimentally one finds for Equation (3.27) numbers between 0.25 and 0.35,
which once again confirms the qualitative usefulness of van der Waals
equation.
On the other hand, a measurement of the critical data of a gas yields a
comfortable method for determining the parameters a and b.
Metastable delayed condensation or delay boiling for isotherm
superheated liquid or supercooled vapor for isochor
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
134 / 324
The law of mass action I

Let us consider a vessel containing a mixture of ideal gases which mutually
react, for instance, according to Equation (3.6).
To take a concrete example,
H2 + Cl2
2HCl,
U = 92.3
kJ
mol(HCl)
(3.28)
The purely thermal energy content for an ideal gas of N particles at

temperature T was U = 3NkT /2. This energy does not take into account
additional internal energies of different particle species due to their internal
structures, different masses, etc.
For instance, two molecules, H2 and Cl2 , differ from two HCl molecules by
the chemical binding energy which is released in the reaction. Thus
3
Ui (T , pi , Ni ) = Ni i + Ni kT ,
2
pi V = Ni kT
(3.29)
Where the energies i define the different zero points of the energy scales of
the respective particles.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
135 / 324
The law of mass action II

The difference 2HCl H2 Cl2 , for instance, is the binding energy difference
between two molecules of HCl and one molecule of H2 and Cl2 , respectively.
The constants i appear also in the chemical potentials of the ideal gases
(
" #)
5/2
i (T0 , p0 )
T
p0
i (T , pi ) = i + kT
ln
(3.30)
kT0
T0
pi
Rewrite
P the equation for i (T , pi ) using p0 /pi = (p0 /p) (p/pi ) = p0 /(pXi )
(p = i pi is the total pressure of the system and X = pi /p = Ni /N is the
molar fraction of the component i)
(
"
#)
5/2
i (T0 , p0 )
T
p0 1
i (T , pi ) =i + kT
ln
kT0
T0
p Xi
(
" #)
5/2
i (T0 , p0 )
T
p0
=i + kT
ln
+ kT ln Xi
kT0
T0
p
=i (T , p) + kT ln Xi
Huinan Zheng (USTC)
9On
(3.31)
00 :51 :25 2012 Spring
136 / 324
The law of mass action III

Insert Equation (3.31) into the equilibrium condition, Equation (3.10), and
we obtain in general
X
X
ai i (T , pi ) =
bj j (T , pj )
i
ai i (T , p)
bj j (T , p) = kT
exp
1
kT
bj Xj
ai i (T , p)
bj j (T , p) =
ai Xi
XBb11 XBb22
XAa11 XAa22
(3.32)
Equation (3.32) is the law of mass action, which determines the equilibrium
concentration of productions XB1 , XB2 , . . . and reactants XA1 , XA2 , . . . in a
chemical reaction according to Equation (3.6).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
137 / 324
The law of mass action IV

One often writes for the lefthand side in Equation (3.32)
X
X
1
K (T , p) = exp
bj j (T , p)
ai i (T , p)
kT
j
(3.33)
This is the equilibrium constant of the reaction at the total pressure p and
the temperature T . We know the chemical potentials i (T , p) for all
pressures and temperatures if we determine them once for a standard pressure
p0 and a standard temperature T0 see Equation (3.30). To this end, we
form the ratio of K (T , p) to K (T0 , p0 ) and find, with Equation (3.30) that

1
1
K (T , p) =K (T0 , p0 ) exp
kT
kT0
P
P
" # j bj i ai
5/2
T
p0
(3.34)
T0
p
with =
Huinan Zheng (USTC)
b j j
ai i , the energy gained or required per reaction.

9On
00 :51 :25 2012 Spring
138 / 324
The law of mass action V

Pressure dependence and temperature dependence Ammonia synthesis
Concentration dependence of the chemical potential of component i
i (T , p, Xi ) =i (T , p, 1) + kT ln Xi
(3.35)
also for ideal solutions.

One can retain the form of Equation (3.35) also for nonideal systems, if one
transforms the term kT ln Xi to include the activities kT ln fi Xi , i.e., if one
introduces effective concentrations.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
139 / 324
Exercises I
Raoults law, increase of boiling point
Calculate the dependence of the vapor pressure of a solvent on the concentration
of a (not very volatile) dissolved substance and the resulting increase of the
boiling point. Consider the vapor and the solution to be ideal.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
140 / 324
Exercises II
Vapor pressure
Calculate the change of the vapor pressure in a liquid, if a nonsoluble gas is mixed
with the vapor.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
141 / 324
Exercises III
The law of Henry and Dalton
Calculate the relation between the pressure of a gas above a nonvolatile solvent
and the concentration of the dissolved gas in the solvent.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
142 / 324
Exercises IV
Vapor pressure of a mixture
Calculate the vapor pressure of a mixture of two solvents as a function of the
molar fraction of solvent 1. Assume that Raoults law is valid for the partial
pressure.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
143 / 324
Exercises V
Osmotic pressure
A solvent with a dissolved material is separated from
the pure solvent by a diaphragm which is only permeable for the solvent. Calculate the pressure difference
between the systems as a function of the concentration
Xm of the dissolved material. Assume the solution to
behave ideally.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
144 / 324
Application of the laws of thermodynamics I

We want to calculate the internal energy U(T , V ) of a real gas. The exact
differential of U reads

U
U
dU =
dT +
dV
(3.36)
T V
V T
We have already identified the expression U/T |V = CV (T , V ) as the heat
capacity.
The exact differential of the entropy S

S
S
dS =
dT
+
dV
T V
V T
(3.37)
for which, on the other hand

dS =
Qrev
dU + pdV
1
=
= CV dT +
T
T
T
Huinan Zheng (USTC)
9On

1 U
p
+
dV
T T V
T
00 :51 :25 2012 Spring
(3.38)
145 / 324
Application of the laws of thermodynamics II

also is valid. By comparing coefficients one finds

S
1 U
1
S
1 U
p
= CV =
and
=
+
T V
T
T T V
V T
T V T T
Since S has an exact differential, it must hold that

1 U
2S
=
V T
V T T V T

1 U
p
2S
+
=
=
T V
T T V T T V
(3.39)
(3.40)
Performing the differentiations yields, with

2U
2U
=
V T
T V
Huinan Zheng (USTC)
(3.41)
9On
00 :51 :25 2012 Spring
146 / 324
Application of the laws of thermodynamics III

the result that

1 2U
1 U
p
1 2U
1 p
= 2
+
+
T V T
T V T T T V
T2
T T V

U
1 p
=
p

V T T T V
If we insert Equation (3.43) into Equation (3.36) we have

p
dU = CV (T , V )dT + T
p dV
T V
Since dU is an exact differential one has

p
CV
=
T
p

V T
T
T V
V
Huinan Zheng (USTC)
9On
(3.42)
(3.43)
(3.44)
(3.45)
00 :51 :25 2012 Spring
147 / 324
Application of the laws of thermodynamics IV

For an ideal gas one has, for instance
p(T , V , N) =
NkT
V
(3.46)
and thus
T

p
p =0
T V
Huinan Zheng (USTC)
so that

CV
=0
V T
9On
(3.47)
00 :51 :25 2012 Spring
148 / 324
Exercises
Internal energy of the van der Waals gas
Calculate the internal energy of a van der Waals gas as a function of temperature
and volume at constant particle number.
Entropy of the van der Waals gas

Calculate the entropy of a van der Waals gas as a function of temperature and
volume at constant particle number.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
149 / 324
The principle of maximum entropy I

All spontaneous (irreversible) processes in an isolated system increase the
entropy, until the maximum is reached for the equilibrium state,
dS = 0,
S = Smax
(4.1)
On the other hand, we know from mechanics, electrodynamics, and quantum

mechanics that systems which are not isolated want to minimize their energy.
For instance, mechanical systems want to assume a state with a minimum
of potential energy.
However, if one adds the heat energy created, the total energy is not changed.
This leads us to the assumption that the striving for minimum energy can be
traced to the striving for maximum entropy.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
150 / 324
The principle of maximum entropy II

We consider an isolated system containing two
partial systems. We remove a certain work W1 <
0 from system 1, e.g., a difference in potential energy. The partial system shall not exchange any
heat with the surroundings during this process.
For this reversible process we have
Q1 = TdS1 = 0
(4.2)
hence the entropy S1 stays constant.

If we now hand over a fraction of the work W1 as heat and a fraction
(1 ) as work to partial system 2, we have
dU2 =Q2 + W2 = dU1 = W1 > 0
(4.3)
Q2 = W1 => 0,
(4.4)
Huinan Zheng (USTC)
W2 = (1 )W1
9On
00 :51 :25 2012 Spring
151 / 324
The principle of maximum entropy III

If the heat is passed over to system 2 while the temperature stays constant,
the following holds,
Q2 =TdS2 > 0
(4.5)
The total entropy of the isolated system has obviously increased through the
transformation of work from partial system 1 into heat in partial system 2,
and the internal energy of partial system 1 has decreased.
A nonisolated system at constant entropy (Q = 0) heads for a state of
minimum energy. At least a part of the work W1 is transformed into heat.
The principle of minimum energy can be derived from the principle of
maximum entropy.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
152 / 324
Entropy and energy as thermodynamic potentials I

The entropy or the internal energy, respectively, are the central state
quantities. If they are known as functions of the natural variables
(U, S, V , N, . . .) of an isolated system, it is guaranteed that also all other
thermodynamic quantities are completely known.
If we know U(S, V , N, . . .), it holds that
dU =TdS pdV + dN +

U
U
T =
,
p
=
T
V
V ,N,...
(4.6)

U
, =
, ...
N S,V ,...
S,N,...
(4.7)
so that also the temperature, pressure, and chemical potential are known as
functions of the natural variables.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
153 / 324
Entropy and energy as thermodynamic potentials II

A similar assertion holds also for the entropy
1
p
dU + dV dN
T T
T

1
S
S
S
p
=
=
,
, =
, ...
T
U V ,N,... T
V U,N,...
T
N U,V ,...
dS =
(4.8)
(4.9)
Equations (4.7) and (4.9), respectively, are the equations of state of the
system.
On the other hand, knowing all equations of state we may calculate the
entropy and the internal energy, respectively, as functions of the natural
variables by integration.
Just like the potential energy of mechanics, the entropy gives information
about the most stable (equilibrium) position of the system.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
154 / 324
Example The entropy of the ideal gas I

Consider the entropy of the ideal gas, as given in Equation (2.40),
(
" #)
5/3
T
p0
S(T , p, N) = Nk s(T0 , p0 ) + ln
T0
p
if we rewrite this in terms of the independent variables U, N, and V using
U = 23 NkT and pV = NkT (U0 = 32 N0 kT0 and p0 V0 = N0 kT0 ,
respectively), we obtain
(
S(U, V , N) =Nk s(U0 , V0 , N0 )
"
+ ln
Huinan Zheng (USTC)
N0
N
5/2
U
U0
3/2
9On
V
V0
# )
(4.10)
00 :51 :25 2012 Spring
155 / 324
Example The entropy of the ideal gas II

All equations of state of the ideal gas can be obtained by partial
differentiation according to Equation (4.9),

3
1
S
1
= = Nk
U V ,N T
2
U

S
1
p
= = Nk

V U,N T
V

S
=
N U,V
T
(
" #)
5/2
3/2
N0
U
V
5
=Nk s0 + ln
k
N
U0
V0
2
(4.11)
(4.12)
(4.13)
If one inserts Equations (4.11) and (4.12) into (4.13), one gets for the
chemical potential
" #

5/2
5
T
p0
(T , p) = kT
s0 kT ln
(4.14)
2
T0
p
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
156 / 324
Example The entropy of the ideal gas III

which coincides up to an additive constant with Equation (2.76).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
157 / 324
The Legendre transformation I

The extensive state variables U, S, V , N, . . . are very useful for isolated
system, where they assume constant values in equilibrium, but in practice, for
instance in a heat bath, these state variables are not appropriate.
It is, for example, experimentally far easier to control, instead of the entropy,
the corresponding intensive variable, the temperature.
Out aim is therefore, for example, in the case of the internal energy
U(S, V , N, . . .), to perform a transformation from the entropy S to the
intensive variable T = (U/S)|U,V ,... .
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
158 / 324
The Legendre transformation II

The transformation we need is the Legendre transformation, which is well
known from classical mechanics. There one uses this transformation to
replace the generalized velocities q in the Lagrange function L(q , q ) by
the new variables p = L/ q , the generalized momenta.
X
H(q , p ) =
q p L(q , q )
(4.15)
The proof is simply given by differentiation

X
L
L
dq
d q
dH =
p d q + q v dp
q
q

X
L
=
q dp
dq
q
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(4.16)
159 / 324
The Legendre transformation III

Let us first restrict ourselves to functions of one variable. Assume f (x) to be
a function of the variable x, with the total differential
df =
f
dx = p(x)dx
x
(4.17)
The task of the Legendre transformation is to find

a function g (p) of the new variable p = f 0 (x),
which is equivalent to the function f (x), i.e.,
which contains the same information. One must
be able to calculate g (p) unambiguously from the
function f (x) and vice versa.
We consider the intersection of the tangent to f at
the point (x0 , f (x0 )) with the y -axis. The tangent
has the following equation,
T (x) = f (x0 ) + f 0 (x0 )(x x0 )
Huinan Zheng (USTC)
9On
(4.18)
00 :51 :25 2012 Spring
160 / 324
The Legendre transformation IV

The intersection with the y -axis g = T (0) therefore is
g (x0 ) = f (x0 ) x0 f 0 (x0 )
(4.19)
One calls the function g (x) for an arbitrary point x the Legendre
transformation of f (x),
g = f xp
with p =
f
x
(4.20)
We differentiate Equation (4.20),

dg = df pdx xdp
(4.21)
If one inserts Equation (4.17) for df , one has

dg = xdp
(4.22)
Thus, g can depend only on the variable p.

Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
161 / 324
The Legendre transformation V

To calculate g (p) explicitly, we have to eliminate x in Equation (4.20),
g (x) = f (x) xf 0 (x)
(4.23)
with the help of the equation

p = f 0 (x)
(4.24)
This, however, is only possible, if Equation (4.24) can be uniquely solved for
x, i.e., if there exists the inverse f 01 to f 0 . Then one can insert
x = f 01 (p)
(4.25)
into Equation (4.23), and one obtains explicitly the function

g (p) = f f 01 (p) f 01 (p)p
(4.26)
It is therefore evident that a unique Legendre transform exists only if

Equation (4.24) represents a bijective mapping. The function f 0 (x) has to
be strictly monotonic for Equation (4.24) to be invertible.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
162 / 324
The Legendre transformation VI

One can reconstruct the original function f (x) from the Legendre
transformation in a unique way. According to Equation (4.20), it holds that
f (p) = g (p) + xp
(4.27)
In this equation can uniquely replace p by x. According to Equation (4.22),

we have
x = g 0 (p)
(4.28)
Since f 0 (x) is strictly monotonous, the inverse function (4.25) is also strictly
monotonous. Therefore Equation (4.28) can be uniquely solved for p(x).
This can be inserted into Equation (4.27), and we uniquely reobtain the
function f (x).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
163 / 324
The Legendre transformation VII

The generalization of the Legendre transform to functions of several variables
is obvious. For instance, f (x, y ) is given.
df = p(x, y )dx + q(x, y )dy
where we have put

f
p(x, y ) =
,
x y
(4.29)

f
and q(x, y ) =
y x
(4.30)
If the variable x is to be replaced by p, one forms

g (x, y ) = f (x, y ) xp
(4.31)
with the total differential

dg = df pdx xdp = xdp + qdy
Huinan Zheng (USTC)
9On
(4.32)
00 :51 :25 2012 Spring
164 / 324
The Legendre transformation VIII

where g in only a function of p and y . To calculate g (p, y ) explicitly, the
first of Equations (4.30) has to be invertible for all values of y . Analogously,
one can replace both variables x and y by p and q.
h(x, y ) = f (x, y ) xp yq
(4.33)
To evaluate h(p, q) explicitly, one must be able to solve the system of

equations (4.30) for x(p, q) and y (p, q).
Examples
f (x) = x 2
f (x) = x
Reverse transformation
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
165 / 324
The free energy
The free energy I

We start from the internal energy.
F = U TS = pV + N
(4.34)
which is called the free energy or Helmholtz potential. The total differential
of U reads, by employing Eulers equation (2.71)
dU = TdS pdV + dN +
(4.35)
Correspondingly,
dF = dU SdT TdS = SdT pdV + dN +
(4.36)
Hence, the free energy is a function of T , V , N, . . ., which contains exactly

the same information as the internal energy U.
One obtains from Equation (4.36) the equations of state

F
F
F
S =
, p =
, =
, ...
T V ,N,...
V T ,N,...
N T ,V ,...
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(4.37)
166 / 324
The free energy
The free energy II

Consider a nonisolated system in a heat bath of constant temperature T . The total system (including
the heat bath) must be isolated. The second law applies,
dStot = dSsys + dSbath 0
According to the first law
dUsys = Qsys + Wsys ,
(4.38)
dUbath = Qbath + Wbath
(4.39)
Since the total system is isolated, for reversible process it must hold that
Qsys = Qbath
Huinan Zheng (USTC)
and Wsys = Wbath
9On
(4.40)
00 :51 :25 2012 Spring
167 / 324
The free energy
The free energy III

When discussing the second law we have discussed the following inequalities,
which are also valid for partial systems
TdS = Qrev Qirr
and Wrev Wirr
(4.41)
We therefore have
rev
irr
dUsys TdSsys = Wsys
Wsys
(4.42)
For a given temperature

rev
irr
dFsys = d (Usys TSsys ) = Wsys
Wsys
(4.43)
The change of the free energy dFsys of the system at constant temperature
(isothermal process) represents the work done by on the system in a
reversible process.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
168 / 324
The free energy
The free energy IV

For reversible processes,
dSbath = dSsys =

Qsys
1
rev
=
dUsys Wsys
T
T
(4.44)
Insert this into Equation (4.38), it follows for isothermal reversible processes
that
rev
rev
TdStot = TdSsys dUsys + Wsys
= dFsys + Wsys
=0
(4.45)
or for irreversible processes, respectively,

irr
TdStot = TdFsys + Wsys
0
(4.46)
For isothermal systems the free energy has an importance quite analogous to
the entropy for isolated systems.
Let the work performed by Wsys = 0, then the entropy of the isolated total
system has a maximum if and only if the free energy of the isothermal
particle system has a minimum.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
169 / 324
The free energy
The free energy V

In particular, processes which diminish the free energy happen spontaneously
and irreversibly in an isothermal system.
dF = d (U TS) = dU TdS 0
for W = 0 and T =const. (4.47)
The free energy yields a combination of the principle of maximum entropy

and minimum energy.
In general, an isothermal system which does not exchange work with its
surroundings strives for a minimum of the free energy. Irreversible processes
happen spontaneously, until the minimum
dF = 0,
F = Fmin
(4.48)
is reached.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
170 / 324
The free energy
Examples I
Precipitation from a solution The precipitation of magnesium carbonate
from a hydrous solution when mixing solutions which contain separately
magnesium ions and carbonate ions,
2
Mg2+
aq + CO3aq MgCO3solid
happens spontaneously, since the cost in energy U 25.1kJ/mol is much

smaller than the gain in entropy, which is approximately T S 71.1kJ/mol
at room temperature.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
171 / 324
The free energy
Examples II
Free energy of the ideal gas We solve Equation (4.10) for U(S, V , N),

U(S, V , N) =U0
N
N0
5/3
V0
V
2/3

2 S
exp
s0
3 Nk
(4.49)
Now we form
F = U TS
(4.50)
We have to express S in Equation (4.50) by T . This happens via

T =

5/3 2/3

N
V0
2 S
2
U
=
U
s
exp
0
0

S N,V ,...
N0
V
3 Nk
3Nk
This equation has to be solved for S(T , V , N),

(
"
3/2 5/2 #)
3 NkT
N0
V
S(T , V , N) =Nk s0 + ln
2 U0
N
V0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
172 / 324
The free energy
Examples III
If we insert this into Equation (4.50) for S, it follows that
(
"
3/2 5/2 #)
3
3 NkT
N0
V
F (T , V , N) = NkT NkT s0 + ln
2
2 U0
N
V0
or, with U0 = 23 N0 kT0 , that
" #)
(
3/2
T
N0
V
3
s0 ln
F (T , V , N) =NkT
2
T0
N
V0
(4.51)
F (T , V , N), as well as U(S, V , N) or S(U, V , N), contains all the equations

of state.

F
S(T , V , N) =
T V ,N

T
N0
V
=Nk s0 + ln
(4.52)
T0
N
V0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
173 / 324
The free energy
Examples IV

F
NkT
p(T , V , N) =
=

V T ,N
V

F
(T , V , N) =
N T ,V
" #)
(
3/2
T
N0
V
5
s0 ln
=kT
2
T0
N
V0
(4.53)
(4.54)
Together with the reverse transformation,

U(T , V , N) =F (T , V , N) + TS =
Huinan Zheng (USTC)
3
NkT
2
9On
(4.55)
00 :51 :25 2012 Spring
174 / 324
The free energy
Examples V
Qualitative difference between internal energy and free energy. U(T , V , N) is
well known to us,
3
U(T , V , N) = NkT
2
(4.56)
The difference is that F (T , V , N) contains the full thermodynamic

information about the system, as well as U(S, V , N). By contrast, in
U(T , V , N) information is lost. One cannot determine the entropy from
U(T , V , N) without the help of other equations of state.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
175 / 324
The enthalpy
The enthalpy I
In chemistry, processes at constant (atmosphere) pressure are of special interest. The corresponding Legendre transformation,
H = U + pV = TS + N
(4.57)
Equation (4.57) defines the enthalpy, which is also a

thermodynamic potential, in the variable S, p and N.
The total differential of the enthalpy reads
dH = dU + pdV + Vdp = TdS + Vdp + dN +
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(4.58)
176 / 324
The enthalpy
The enthalpy II
If the enthalpy H(S, p, N, . . .) is known, all other state quantities may be
obtained by partial differentiation, as for U and F ,

H
H
H
T =
, V =
, =
, ...
(4.59)
S p,N,...
p S,N,...
N S,p,...
As with all other thermodynamic potentials the enthalpy can in principle be
calculated for any system.
It is special useful for isobaric (p = const., dp = 0) and adiabatic (Q = 0)
systems.
The total work reversibly exchanged with the surroundings is
rev
rev
rev
Wtot
= Wvol
+ Wother
. With the help of the first law,
dU = Q + W = Q + Wother pdV , for reversible changes of state at
constant pressure, that
dH|p = d (U + pV )|p = (dU + pdV + Vdp)p = dU|p + pdV |p
rev
|p
dH|p = Q|p + Wother
Huinan Zheng (USTC)
(4.60)
9On
00 :51 :25 2012 Spring
177 / 324
The enthalpy
The enthalpy III

For isobaric changes of state, the change of the enthalpy is just the amount
of heat exchange with the surroundings plus the exchanged utilizable work,
which is not simply volume work against the constant external pressure.
Consider especially an isobaric and adiabatic system with p = const. and
Q = 0. Then
rev
dH|p,ad = Wother
|p,ad
(4.61)
This corresponds to Equation (4.43) for the change of the free energy in an
isothermal system.
irr
irr
Again, for irreversible processes Wother
Wother
is valid and thus

rev
irr
dH|p,ad = Wother
|p,ad Wother
p,ad
(4.62)
irr
If, especially Wother
= 0, i.e., if no utilizable work is performed, we have
dH 0
Huinan Zheng (USTC)
(4.63)
9On
00 :51 :25 2012 Spring
178 / 324
The enthalpy
The enthalpy IV
In an adiabatic, isobaric system, which is left to its own, irreversible processes
happen, which decrease the enthalpy, until in equilibrium a minimum of the
enthalpy is reached,
dH = 0,
H = Hmin
(4.64)
Examples
1
Enthalpy of the ideal gas
Calculation of the equations of state from the enthalpy
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
179 / 324
The enthalpy
The enthalpy V
If we add an amount of heat Q to a system at constant volume, we have,
with W = 0,
dU = Q|V
(4.65)
For the specific heat at constant volume it holds that

Q
U
CV =
=
dT V
T V
(4.66)
However, if the heat Q is added under constant pressure, generally the

volume of the system will change, and a certain work will be performed,
dU = Q|p pdV
(4.67)
The internal energy is not very appropriate to describing this process. At

constant pressure, however, Equation (4.67) can be simply put in the form
dH = d (U + pV ) = Q|p
(4.68)
which is quite analogous to Equation (4.65).

Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
180 / 324
The enthalpy
The enthalpy VI
The specific heat at constant pressure therefore is

Q
H
Cp =
=
dT p
T p
(4.69)
We have for an ideal gas

H(T , p, N) =
5
NkT
2
(4.70)
Thus
Cp =
5
Nk
2
(4.71)
3
Nk
2
(4.72)
while
CV =
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
181 / 324
The enthalpy
The enthalpy VII

In chemistry the enthalpy plays an important role. On the other hand, many
reactions happen so fast that an exchange of heat with the surroundings is
nearly impossible (Q = 0). With the help of the enthalpy one can easily
decide in this case whether a certain chemical reaction is possible and
whether it happens spontaneously under given conditions. One simply
compares the sum of the enthalpies of the reaction products with that of the
reactants. If H = Hproducts Hreactants is negative, i.e., H 0, the
reaction happens spontaneously and irreversibly.
Exothermal and endothermal.
Example
Reaction enthalpy Theorem of Hess
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
182 / 324
The free enthalpy
The free enthalpy I

For systems with given temperature and pressure,
G =U TS + pV
(4.73)
The corresponding thermodynamic potential is the free enthalpy introduced

by J. W. Gibbs (1875), for which reason it is also called the Gibbs potential.
The total differential of the free enthalpy reads
dG =dU TdS SdT + pdV + Vdp
= SdT + Vdp + dN +
(4.74)
Consequently, G indeed depends only on T , p, and N.

If the function G (T , p, N) is known, we can obtain all further quantities by
partial differentiation,

G
G
G
S =
,
V
=
,
=
, ...
(4.75)
T p,N
p T ,N
N T ,p
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
183 / 324
The free enthalpy
The free enthalpy II

Using Eulers equation, which must be fulfilled in any case, we can identify
the Gibbs free enthalpy somewhat more explicitly. Eulers equation (2.71) for
a system of one species reads
U = TS pV + N
(4.76)
By comparison with Equation (4.8),

G = N
(4.77)
Hence, G is directly proportional to the particle number, and the free

enthalpy per particle is identical with the chemical potential.
To understand the meaning of the free enthalpy, we
form an isolated system consisting of the isothermal,
isobaric system and its surroundings (the heat bath).
Then it holds that
Huinan Zheng (USTC)
(4.78)
9On
00 :51 :25 2012 Spring
184 / 324
The free enthalpy
The free enthalpy III

The heat bath, which is of no further interest, can be excluded. It holds (for
reversible processes) that
dSbath = dSsys =
1
rev
(dUsys + pdVsys Wother
)
T
(4.79)
or (for irreversible processes),

rev
TdStot = TdSsys dUsys pdVsys + Wother
0
(4.80)
which can also be written as follows (the subscript sys is omitted)

rev
irr
dG = d (U TS + pV ) = Wother
Wother
(4.81)
The change in the free enthalpy is just the work performed by the system in
an isothermal, isobaric reversible process, without the volume work against
the constant pressure.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
185 / 324
The free enthalpy
The free enthalpy IV

In an isothermal, isobaric system which is left to its own, irreversible
processes happen until a minimum of the free enthalpy is achieved,
dG = 0,
G = Gmin
(4.82)
Exergonic and endergonic
Example
Free enthalpy of the ideal gas
Exercise Gibbs-Helmholtz equation

Show that the free enthalpy and its derivative with respect to temperature can be
related to the enthalpy of the system via the Gibbs-Helmholtz equation

G
(G /T )
2
H =G T
= T
T p
T
p
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
186 / 324
The free enthalpy
The free enthalpy V

If the isothermal, isobaric system consists of several chemical components
(particle species), then it holds according to the extended Euler
equation (2.71), that
X
G=
i Ni
(4.83)
i
If reactions between the particles are possible,

a1 A1 + a2 A2 +
b1 B1 + b2 B2 +
(4.84)
then for the changes in particle numbers dNAi of species Ai and dNBj of
species Bj , the relations dNAi = ai dN and dNBj = bj dN have to be valid,
where dN is a common factor. The equilibrium condition for an isothermal,
isobaric system can be written as,
X
i dNi = 0
(4.85)
dG =
i
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
187 / 324
The free enthalpy
The free enthalpy VI

also, it follows that
X
X
dG =
Ai ai +
Bj dN = 0
i
(4.86)
Since dN is arbitrary, we must have

X
X
Ai ai =
Bj
i
(4.87)
a relation we had derived earlier from other considerations (cf.

Equation (3.10)).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
188 / 324
The free enthalpy
The free enthalpy VII

We already know the dependence of chemical potential on concentration in
dilute solutions or gases (cf. Equation (3.31)),
i (T , p, Xi ) = i (T , p, 1) + kT ln Xi
(4.88)
If we insert this into Equation (4.86) we have
X
X
dG =
0Bj bj
0Ai ai dN
j
#
b
b
(XB1 ) 1 (XB2 ) 2
dN
+ kT ln
a
a
(XA1 ) 1 (XA2 ) 2
"
(4.89)
where 0Bj = Bj (T , p, 1) and 0Ai = Ai (T , p, 1). Since G /N = G /N, it is

independent of the change in the particle number dN, we can put dN = 1
and obtain
(T , p, XA1 , . . . , XB1 , . . .) =G 0 (T , p)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
189 / 324
The free enthalpy
The free enthalpy VIII

"
(XB1 ) 1 (XB2 ) 2
+ kT ln
a
a
(XA1 ) 1 (XA2 ) 2
#
(4.90)
In equilibrium we must have G = 0, or

b
b
(XB1 ) 1 (XB2 ) 2
G 0 (T , p)
= exp
a
a
kT
(XA1 ) 1 (XA2 ) 2
(4.91)
The equilibrium condition G = 0 for an isothermal, isobaric system leads

directly to the law of mass action. In particular, we now have identified the
equilibrium constant more explicitly. It is determined by the free enthalpy
difference G 0 between the products and the reactants at standard
concentration.
Equation (4.91) contains also an exact information of the principle of Le
Chatelier, according to which an equilibrium state changes under exertion of
a force in such a manner that the system gives way to the force.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
190 / 324
The free enthalpy
The free enthalpy IX

We define the equilibrium constant appearing in Equation (4.91)

G 0 (T , p)
K (T , p) = exp
kT
Using the free reaction enthalpy per particle
X
X
G 0 (T , p) =
0j bj
0i aj
j
(4.92)
(4.93)
we obtain

v
ln K (T , p)
= kT
p
T
(4.94)
if we denote the change of the volume per particle at a given p and T by
X
X
X
X 0
v =
j bj
0i aj =
vj0 bj
vi0 ai
(4.95)
p
j
Huinan Zheng (USTC)
T
9On
00 :51 :25 2012 Spring
191 / 324
The free enthalpy
The free enthalpy X

In the same manner we find for the temperature dependence of the
equilibrium constant

ln K (T , p)
h
= kT
T
p
if we denote the reaction enthalpy per particle by h. We have
0
0
X
X
j
h = T 2
bj
aj i
T
T
T
j
(4.96)
(4.97)
because it holds in general, according to the Gibbs-Duhem relation, that

s
= 2
(4.98)

T T p
T
T
due to /t = s = S/N, and furthermore H = U + pV = N + TS.
Hence h = H/N = + Ts is the enthalpy per particle.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
192 / 324
The grand potential
The grand potential I

Just as a fixed temperature is established by a heat bath, the chemical
potential can be fixed via a particle bath. Just as the exchange of heat with a
heat bath leads to a constant temperature in equilibrium, the exchange of
particles with a particle reservoir leads to a constant chemical potential.
Since such a particle exchange is in most case also connected with a heat
exchange and the particle reservoir thus also acts like a heat bath, we want
to transform the internal energy in the variables S and N to the new variables
T and ,
= U TS N
(4.99)
The corresponding potential is called the grand potential.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
193 / 324
The grand potential
The grand potential II

The total differential reads
d = dU TdS SdT dN Nd = SdT pdV Nd (4.100)
The remaining thermodynamic quantities can be calculated by differentiating
the grand potential,

S =
, p =
, N =
(4.101)
T V ,
V T ,
T ,V
Because of Eulers equation
U = TS pV + N
(4.102)
the grand potential is identical with pV ,

= pV
(4.103)
This potential is especially suited for isothermal systems with a fixed chemical potential.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
194 / 324
The grand potential
The grand potential III

If we combine the heat bath and the system under consideration to a total,
isolated system, it holds that
(4.104)
In the reversible case we can express dSbath by changes of system quantities,

rev
TdSbath = TdSsys = (dUsys dNsys Wother
)
(4.105)
If we insert this into Equation (4.104), we obtain

rev
irr
dUsys = TdSsys dNsys = Wother
Wother
(4.106)
At constant temperature and constant chemical potential this is equivalent to

rev
irr
d = d (U TS N) = Wother
Wother
Huinan Zheng (USTC)
9On
(4.107)
00 :51 :25 2012 Spring
195 / 324
The grand potential
The grand potential IV

If we leave the system to its own without performing work, W = 0, it strives
for a minimum of the grand potential,
d 0
(4.108)
which is achieved in equilibrium,

d = 0,
Huinan Zheng (USTC)
(4.109)
= min
9On
00 :51 :25 2012 Spring
196 / 324

The Legendre transformation for all variables in U would read
= U TS + pV N
(4.110)
However, according to Eulers equation it must always hold that

U = TS pV + N
(4.111)
so that this potential vanishes identically, 0.

The simultaneous transformation of all variables is thus of no relevance.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
197 / 324
The Maxwell relations I

The total differential of the internal energy reads
dU =TdS pdV + dN

U
U
=
dS
+
S
V
V ,N
dV +
T ,N

U
dN
N T ,V
(4.112)
Since
!

U
S V ,N
S,N
!

U
V T ,N
=
S
(4.113)
V ,N
it immediately follows, e.g., that

p
T
=
V S,N
S V ,N
Huinan Zheng (USTC)
9On
(4.114)
00 :51 :25 2012 Spring
198 / 324
The Maxwell relations II

In a systematic way, it follows from Equation (4.112) that

T
p
T

=
,
=
,
V S,N
S V ,N N S,V
S V ,N

p

=
N S,V
V S,N
(4.115)
Corresponding relations exist for the free energy F (T , V , N),

dF = SdT pdV + dN

p
S

S
=
,
=
,
V T ,N
T V ,N
N T ,V
T V ,N

p

=
N T ,V
V T ,N
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(4.116)
(4.117)
199 / 324
The Maxwell relations III

Analogously it holds for the enthalpy H(S, p, N) that
dH =TdS + Vdp + dN

V
T

V

T
=
,
=
,
=
p S,N
S p,N N S,p
S p,N N S,p
p S,N
(4.118)
(4.119)
For the free enthalpy G (T , p, N) one has

dG = SdT + Vdp + dN

S
V
S

=
,
=
,
p T ,N
T p,N
N T ,p
T p,N

V

=
N T ,p
p T ,N
Huinan Zheng (USTC)
9On
(4.120)
(4.121)
00 :51 :25 2012 Spring
200 / 324
The Maxwell relations IV

Finally for the grand potential (T , V , ) we have

S
V
d = SdT pdV Nd

p
S
N
p
N
=
,
=
,
=
T V , T ,V
T p, T ,V
V T ,
T ,
(4.122)
(4.123)
Relations (4.115), (4.117), (4.119), (4.121), and (4.123) are call Maxwell
relations.
There exists a simple device which allows for a
quick overlook of the potentials and their variables and which yields the Maxwell relations.
Thermodynamic rectangle.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
201 / 324
The Maxwell relations V

Examples
1
Heat capacities
Joule-Thomson experiment
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
202 / 324
Jacobi transformations I
Pure transformations of variables in the same physical quantity.
Example
Calculation of Cp from the entropy
In general, the Jacobi determinant for the transformation from the variables
(x1 , x2 , . . . , xn ) to the new variables (u1 , u2 , . . . , un ) is defined by
u1
x
u12

x1
J(x1 , x2 , . . . , xn ) = .
..

un
x1
u1
x2
u2
x2
..
.
un
x2
..
.
u1
xn
u2
xn
(4.124)
..
.
un
xn
it is also denoted as
J(x1 , x2 , . . . , xn ) =
Huinan Zheng (USTC)
(u1 , u2 , . . . , un )
(x1 , x2 , . . . , xn )
9On
(4.125)
00 :51 :25 2012 Spring
203 / 324
Jacobi transformations II
According to the rules for multiplication of determinants
(u1 , u2 , . . . , un ) (w1 , w2 , . . . , wn )
(u1 , u2 , . . . , un )
=
(w1 , w2 , . . . , wn ) (x1 , x2 , . . . , xn )
(x1 , x2 , . . . , xn )
(4.126)
This is nothing but a generalized chain rule. For n = 1 Equation (4.126)

simply reads
du dw
du
=
dw dx
dx
(4.127)
The Jacobi determinant of the reverse transformation

1
(x1 , x2 , . . . , xn )
(u1 , u2 , . . . , un )
=
(u1 , u2 , . . . , un )
(x1 , x2 , . . . , xn )
(4.128)
is just the inverse of the original transformation.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
204 / 324
Jacobi transformations III

Especially

u
x1 x2 ,x3 ,...,xn
u

x1

(u, x2 , . . . , xn )

=
=
(x1 , x2 , . . . , xn )

0
1
..

0

1
(4.129)
Example
Joule-Thomson coefficient
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
205 / 324
Thermodynamic stability I
The thermodynamic equilibrium states are characterized by a maximum of
the entropy, or by a minimum in the various thermodynamic potentials,
respectively.
One can immediately derive relations, the so-called conditions for
thermodynamic stability, which can be directly determined from the second
derivative of the potentials.
For instance, for the isothermal compressibility it must hold that

1 v
0
=
V p T
(4.130)
This means that in equilibrium a spontaneous decrease of the volume

(V < 0) effects an increase of the pressure (p > 0), so that the system
moves by itself back towards the equilibrium state.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
206 / 324
Thermodynamic stability II
Similarly the heat capacities must be Cp , CV 0, so that a spontaneous
temperature increase which would correspond to an energy increase
cannot happen. These two conditions are special requirements which follow
from the Braun-Le Chatelier principle
If a system is in stable equilibrium, then all spontaneous changes of
the parameters must invoke processes which bring the system back
to equilibrium, i.e., which work against these spontaneous changes.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
207 / 324
Statistical Mechanics I
5
Number of Microstate and Entropy S

Foundations
Phase space
Statistical definition of entropy
Gibbs paradox
Quantum mechanical counting of
Ensemble Theory and Microcanonical Ensemble

Phase-space density, ergodic hypothesis
Liouvilles theorem
The microcanonical ensemble
Entropy as an ensemble average
The uncertainty function
The Canonical Ensemble

General foundation of the Gibbs correction factor
Systems of noninteracting particles
Calculation of observables as ensemble averages
Connection between microcanonical and canonical ensembles
Fluctuations
Virial theorem and equipartition theorem
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
208 / 324
Statistical Mechanics II
For better understanding: canonical ensembles as the mean value of all possible
distributions
Applications of Boltzmann Statistics

Quantum Systems in Boltzmann Statistics
Paramagnetism
Negative temperatures in two-level systems
Gases with internal degrees of freedom
Relativistic ideal gas
The Macrocanonical Ensemble

Fluctuations in the macrocanonical ensemble
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
209 / 324
Foundations
Foundations
Macroscopic properties equations of state empirically
From the point of view of physical insight a little unsatisfactory
microscopic considerations
Macroscopic quantities of state mean values of microscopic properties
Statistics define in an exact way the process of taking mean values a
connection between the atomistic, microscopic theory of matter and
macroscopic thermodynamics.
Entropy connected to the number of possible microstates in a given
macrostate (U, V , N), S ln .
Make the notion number of microstates more precise ensemble theory
The macroscopic quantities are defined as mean values of microscopic
quantities, weighted with probability densities.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
210 / 324
Phase space
Phase space I
Microstate classical system generalized coordinates q (t) and momenta
p (t) the set (q , p ), = 1, 2, . . . , 3N 6N-dimensional space phase
space
A definition point in this phase space exactly corresponds to one microscopic
state of motion of the whole system.
Phase-space trajectory
q =
H
,
p
p =
H
q
(5.1)
In a closed global system, in which the Hamiltonian does not depend

explicitly on time, the total energy
E = H(q (t), p (t))
(5.2)
is a conserved quantity, which along the phase-space trajectory (q (t), p (t))

always assume the same time-independent value, E .
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
211 / 324
Phase space
Phase space II
In general, the time dependence of an observable quantity A(q (t), p (t), t)
is given by
3N
dA A X
=
+
dt
t
=1
A H
A
+
p
q p
p

(5.3)
which, using Equation (5.1), can be rewritten as,

3N
dA A X
=
+
dt
t
=1
=
A
A H
q
q
p q
A
+ [A, H]
t
(5.4)
Here we have used the Poisson bracket [A, H] as an abbreviation for the sum.
Equation (5.4) describes a (6N 1)-dimensional hypersurface in phase space.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
212 / 324
Phase space
Phase space III

Example (Harmonic oscillator)
The Hamiltonian of a harmonic oscillator in one dimension reads
H(q, p) =
1
p2
+ + Kq 2
2m
2
(5.5)
Since H does not explicitly depend on time the total

energy is a conserved quantity.
H(q, p) =
p2
1
+ + Kq 2 = const.
2m
2
is just an ellipse in one-particle phase space which is determined by the value of E .

The hypersurface also reflects the phase-space distribution of many equal systems
at one moment. A collection of such phase-space points (systems) which are
consistent with certain macroscopic properties is called an ensemble.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
213 / 324
Phase space
Phase space IV
The phase-space elements d 3N q d 3N p, is call a
phase-space cell. And in general, phase-space volumes shall be abbreviated with the letter (which
must not be confused with frequency). Consequently, the short notation of a phase-space cell
d 3N q d 3N p is d .
For example, for phase-space volume between the ellipses corresponding to E
and E + E , respectively, we have
Z
Z
=
dq dp =
d
(5.6)
E H(q,p)E +E
E H(q,p)E +E
In the same way, according to Equation (5.2), we can relate an area

Z
(E ) =
d
(5.7)
E =H(q,p)
to the energy hypersurface.

Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
214 / 324
Phase space
Phase space V
Consider a closed system, characterized by the natural variables E , V , and N.
As a measure of the number of microstates, we can use the area of energy
surface, and assume (E , V , N) to be proportional to this surface,
Z
(E , N, V )
with (E , V , N) =
d
(5.8)
(E , V , N) =
0
E =H(q ,p )
Let (E , V , N) be the total phase-space volume, the boundary of which is
given by the energy hypersurface E = H(q , p ) and the walls of the
container in coordinate space,
Z
(E , V , N) =
d 3N q d 3N p
(5.9)
H(q ,p )E
For small E , the volume between two energy surfaces with energy E and
E + E is given by

E
(5.10)
= (E + E ) (E ) =
E V ,N
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
215 / 324
Phase space
Phase space VI
On the other hand, according to Cavalieris theorem,
(5.11)
= (E )E
which by comparison with Equation (5.10) gives

(E ) =
(5.12)
Thus
(E , V , N) =
(E , V , N)
1
=
0
0 (E )
(5.13)
where is given by Equation (5.9).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
216 / 324
Statistical definition of entropy I

In thermodynamic equilibrium the most probable macroscopic state is the one
which corresponds to the largest number of consistent microstates.
The basic principle all microstates with the same total energy appear with
the same probability.
Consider a closed system which consists of two subsystems
E = E1 + E2 = const.
V = V1 + V2 = const.
N = N1 + N2 = const.
dE1 = dE2
dV1 = dV2 (5.14)
dN1 = dN2
The total number of all microstates of the total system (E , V , N) results as

the product of these numbers for the subsystems, if the latter can be
considered to be statistically independent,
(E , V , N) = 1 (E1 , V1 , N1 )2 (E2 , V2 , N2 )
Huinan Zheng (USTC)
9On
(5.15)
00 :51 :25 2012 Spring
217 / 324
Statistical definition of entropy II

The most probable state, i.e., the equilibrium state, is the one with the
largest number of microstates, i.e., = max and d = 0.
d = 2 d 1 + 1 d 2
(5.16)
d ln = d ln 1 + d ln 2
(5.17)
or
The equilibrium condition reads

d ln = 0
Huinan Zheng (USTC)
ln = ln max
9On
(5.18)
00 :51 :25 2012 Spring
218 / 324
Statistical definition of entropy III

Consider the system from a purely thermodynamic point of view,
S(E , V , N) = S1 (E1 , V1 , N1 ) + S2 (E2 , V2 , N2 )
(5.19)
The total difference is

dS = dS1 + dS2
(5.20)
We are in equilibrium exactly if the entropy is maximal,

dS = 0 S = Smax
(5.21)
Comparison of Equation (5.17) with Equation (5.20), and Equation (5.18)

with Equation (5.21) reveals the complete analogy of ln and entropy.
S(E , V , N) = k ln (E , V , N)
Huinan Zheng (USTC)
9On
(5.22)
00 :51 :25 2012 Spring
219 / 324
Statistical definition of entropy IV

Remark: In general, we would have
S(E , V , N) = k ln (E , V , N) + const.
(5.23)
However, since S is only defined up to a constant, the constant can be

absorbed in S.
Equation (5.22) is of fundamental significance for statistical mechanics It
allows one, at least in principle, to calculate all thermodynamic properties of
a given many-body system using the Hamiltonian H(q , p ).
The thermodynamic
variable at the same

1
S
=
,
T
E V ,N
potential S(E , V , N) as a function of the natural

time gives us the equation of state via

p
S
S
=
,
=
(5.24)
T
V E ,N
T
N E ,V
The practical calculation of is by no means trivial and only the general

ensemble theory will give us a practicable method for the calculation of more
complicated systems.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
220 / 324
Statistical definition of entropy V

Constant 0 energy surface
Quantum mechanically it is possible to identify with the discrete numbers
of microstates, which are fixed by the quantum numbers. In this case,
Equation (5.22) gives the absolute entropy without an additive constant.
The entropy S = 0 corresponds to a system which can assume only one
exactly defined microstate ( = 1). In practice, such systems are, for
instance, ideal crystal at the temperature T = 0 The statement that such
systems at T = 0 have the entropy S = 0 is also called the third law of
thermodynamics.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
221 / 324
Example Statistical calculation of the entropy of the ideal

gas I
The Hamiltonian of the ideal gas reads
H(q , p ) =
N
3N
X
X
p 2
p2
=
2m =1 2m
=1
According to Equation (5.9),

Z
(E , V , N) =
(5.25)
d 3N q d 3N p
H(q ,p )E )
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
222 / 324

gas II
Since the Hamiltonian does not depend on the coordinates of the particles,
Z
(E , V , N) = V N
d 3N p
(5.26)
H(p )E )
The remaining
integral is just the volume of a 3N-dimensional sphere of
radius 2mE . The condition H(p ) E explicitly reads

3N
X
p2
2mE
2
=1
The volume VN (R) of an N-dimensional sphere with the radius R reads

Z
Z
N
VN (R) = P
dx1 dxN = R P
dy1 dyN
N
=1
Huinan Zheng (USTC)
N
=1
x2 R 2
9On
y2 1
00 :51 :25 2012 Spring
223 / 324

gas III
The last integral no longer depends on R but only on the dimension N of
space,
VN (R) =R N CN
Z
CN = P
N
=1
y2 1
dy1 dyN
where CN is just the volume of the N-dimensional unit sphere.

It is well known that
Z
dx exp(x 2 ) =
and thus
Z
dx1
Huinan Zheng (USTC)

dxN exp (x12 + + xN2 ) = N/2
(5.27)
9On
00 :51 :25 2012 Spring
224 / 324

gas IV
Now the integrand of Equation (5.26) depends only on
R = (x12 + + xN2 )1/2 . By transforming to polar coordinates in
N-dimensional space, we express the volume dx1 dxN by spherical shell,
dx1 dxN |shell = dVN (R)|shell = NR N1 CN dR
The Equation (5.26) is equivalent to
Z
NCN
R N1 dR exp(R 2 ) = N/2
0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
225 / 324

gas V
With the substitution R 2 = x the integral just gives the -function,
Z
1
NCN
dx x N/21 e x = N/2
2 0
As per definition,
Z
(z) =
dx x z1 e x
(5.28)
(5.29)
So we have calculated CN ,
CN =
N/2
(5.30)
N
2 (N/2)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
226 / 324

gas VI
With the recursion formula for the -function
(z + 1) = z(z)
and (1/2) = , e.g.,
(5.31)
3/2
4 3
V3 (R) = R 3 3 1 =
R
3
22
Correspondingly, the volume of an N-dimensional sphere is given by
VN (R) = R N
Huinan Zheng (USTC)
N/2
N
2 (N/2)
9On
00 :51 :25 2012 Spring
227 / 324

gas VII
For (E , V , N) this gives,
(E , V , N) =
3N/2
(2mE )3N/2 V N
3N
2 (3N/2)
(5.32)
Using Equation (5.13) we get

(E , V , N) =
1
1 N 3N/2
=
V
(2m)3N/2 E 3N/21
0 E
0
(3N/2)
So the entropy of the ideal gas is

1 N 3N/2
S(E , V , N) = k ln
V
(2m)3N/2 E 3N/21
0
(3N/2)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(5.33)
(5.34)
228 / 324

gas VIII
Equation (5.34) can be consider simplified if N 1, so that
E 3N/21 E 3N/2 , and using Stirlings formula,
ln (n) (n 1) ln(n 1) (n 1) n ln n n
1/N
With the new constant = 0 ,

(
"
3/2 #)
3
V 4mE
S(E , V , N) = Nk
+ ln
2
3N
Huinan Zheng (USTC)
9On
(5.35)
00 :51 :25 2012 Spring
229 / 324

gas IX
Now we can confirm first that the entropy, Equation (5.35), gives the right
equations of state of the ideal gas, and second that the constant k is
Boltzmanns constant,

S
1
3
3
1
=
or E = NkT
= Nk

T
E V ,N
2
E
2

p
S
Nk
=
=
or pV = NkT
(5.36)
T
V
V
E ,N
Equation (5.35) leads to a contradiction? the entropy an extensive

quantity?
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
230 / 324
Gibbs paradox
Gibbs paradox I
With the aid of Equations (5.35) and (5.36),

V
3
+ ln
(2mkT )3/2
S(T , V , N) = Nk
2
(5.37)
Consider a closed system consisting of two containers

which are separated by a wall and which contain two
different ideal gas A and B under the same pressure
and at the same temperature. If the wall is removed,
both gases will spread out over the whole of the container until a new equilibrium situation is reached.
The internal energy remains constant. The temperature and pressure does
not change either. The entropy, however, increases by a certain value.
mixing entropy,
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
231 / 324
Gibbs paradox
Gibbs paradox II
Before the removal of the separating wall
(0)
(0)
(0)
Stotal = SA (T , VA , NA ) + SB (T , VB , NB )
(5.38)
and afterwards we have

(1)
(1)
(1)
Stotal = SA (T , VA + VB , NA ) + SB (T , VA + VB , NB )
(5.39)
By using Equation (5.37), the entropy difference becomes

(1)
(0)
S =Stotal Stotal

VA + VB
VA + VB
=NA k ln
+ NB k ln
VA
VB
Huinan Zheng (USTC)
9On
(5.40)
00 :51 :25 2012 Spring
232 / 324
Gibbs paradox
Gibbs paradox III

We can undergo the same considerations for two identical gases. The initial
entropy Equation (5.38) is still right, while now, in the final situation, we have
(1)
Stotal = S(T , VA + VB , NA + NB )
(5.41)
In this case we receive exactly the same entropy difference S > 0. This
cannot be correct!
Without any change we can bring in the separating wall again and recover
the initial situation. Thus, in the case of identical gases the removal of the
separating wall is a reversible process and we must have S = 0.
In classical mechanics the particles are distinguishable one can enumerate
them Quantum mechanically, the atoms are completely indistinguishable.
Obviously it is the fact that classical particles are in principle distinguishable,
which leads to Gibbs paradox
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
233 / 324
Gibbs paradox
Gibbs paradox IV
If we count the microstates (E , V , N) of a given macrostate (E , V , N) we
must take into account that in reality the particles are not enumerable
(distinguishable). Two microstates may be considered different, if they
differ in more than the enumeration of the particles.
Instead of
(E , V , N) =
(E , V , N)
0
(5.42)
we now try a new definition of ,

(E , V , N) =
1 (E , V , N)
N!
0
(5.43)
where, of course, the calculation of (E ) by (E ) = /E has not been

changed.
The factor 1/N! in Equation (5.43) is also called the Gibbs correction factor.
Its presence proves that classical statistics with distinguishable particles
very quickly leads to contradictions with experiment.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
234 / 324
Gibbs paradox
Gibbs paradox V
Instead of Equation (5.35), we have
(
"
3/2 #)
3
V 4mE
S(E , V , N) = Nk
+ ln
k ln N!
2
3N
Here for N 1 we can also use Stirlings formula (ln N! N ln N N), and
we get
(
"

3/2 #)
4mE
5
V
(5.44)
S(E , V , N) = Nk
+ ln
2
N
3N
Entropy is now indeed an extensive quantity, since in the argument of the
logarithm there remains only the intensive variables V /N and E /N.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
235 / 324
Gibbs paradox
Gibbs paradox VI
If we now calculate the entropy difference in our mixing experiment using
Equation (5.44), we obtain

5
V
3/2
S(T , V , N) = Nk
+ ln
(2mkT )
(5.45)
2
N
because of E = 32 NkT .
According to Equations (5.38) and (5.39) for different gases, we obtain, using
Equation (5.45),

VA + VB
VA + VB
S =NA k ln
+ NB k ln
VA
VB
which means that Equation (5.40) remains unchanged.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
236 / 324
Gibbs paradox
Gibbs paradox VII

For identical gases we get from Equations (5.41) and (5.38), using
Equation (5.45),

5
VA + VB
3/2
S =(NA + NB )k
+ ln
(2mkT )
2
(NA + NB )

VA
5
+ ln
(2mkT )3/2
NA k
2
NA

5
VB
3/2
NB k
+ ln
(2mkT )
2
NB
(5.46)
Since the pressure and temperature do not change during the mixing process,
and since in the initial state we have thermal and mechanical equilibrium,
VB
VA + VB
VA
=
=
NA
NB
NA + NB
Huinan Zheng (USTC)
(5.47)
9On
00 :51 :25 2012 Spring
237 / 324
Gibbs paradox
Gibbs paradox VIII

is valid, and thus
S = 0
(5.48)
as it should be.
However, in the case of distinguishable objects (e.g., atoms which are
localized at certain grid points), the Gibbs factor must not be added. In
classical theory the particles remain distinguishable. we will meet this
inconsistency more frequently in classical statistical mechanics.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
238 / 324
Quantum mechanical counting of I

We now consider the ideal gas once again, by counting
the quantum mechanical states of distinguishable particles. In this way, it becomes possible to get a value
for the unit surface (unit volume) 0 in phase-space,
which up to now has been unknown.
The one-particle states have the wavefunction
nx ,ny ,nz =A sin(kx x) sin(ky y ) sin(kz z)
ny y
nz z
nx x
sin
sin
,
=A sin
L
L
L
nx , ny , nz = 1, 2, . . . , (5.49)
with the corresponding single-particle energy

nx ,ny ,nz =
(~k)2
~2 2
h2
=
(kx + ky2 + kz2 ) =
(n2 + ny2 + nz2 )
2m
2m
8mL2 x
(5.50)
The state of motion (microstate of a particle) is fixed by the quantum

numbers nx , ny , nz . Each single-particle state corresponds exactly to a point
in the 3-dimensional (nx , ny , nz )-space.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
239 / 324
Quantum mechanical counting of II

A given energy E for the particle
corresponds to a
spherical shell with radius Lh 8m. However, only

integer grid points which lie on this sphere are possible
single-particle states for this energy .
The corresponding N-particle problem is now directly
solved by occupation of the single-particle states with
N particles. The total energy is then determined by
the 3N quantum numbers of the occupied states,
E=
3N
h2 X 2
ni
8mL2
(5.51)
i=1
We now deal with a 3N-dimensional space and a (3N 1)-dimensional energy

sphere. It may be possible that more, fewer, or even no grid points at all lie
on the energy sphere according to how many ways there are to subdivide the
number = 8mL2 /h2 into a sum of 3 (altogether, 3N) quadratic numbers.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
240 / 324
Quantum mechanical counting of III

We use the notation which is already known from atomic physics and
chemistry, (state)occupation for instance (1)9 , i.e., the quantum number 1
occurs 9 times.
The example of N = 3 particles, i.e., 9 quantum numbers n1 , . . . , n9 .
I
The lowest state all quantum numbers are ni = 1. The corresponding

dimensionless energy E = 8mL2 E /h2 is
E = 9 = 9 12 ,
= 1,
(1)9
(5.52)
The next higher state is reached if one quantum number assumes the value 2
while all the others remain at the value 1,
E = 12 = 8 12 + 1 22 ,
= 9 = 9choose1,
(1)8 (2)1
(5.53)
In the correct quantum statistics the indistinguishability of the particles must

be taken into account ab initio!
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
241 / 324
Quantum mechanical counting of IV

I
The next higher states of the 3-particle system are given as,
State
E = 15 = 7 12 + 2 22
E = 17 = 8 12 + 1 32
E = 18 = 6 12 + 3 22
E = 20 = 7 12 + 1 22 + 1 32
E = 21 = 5 12 + 4 22
9
= 36
2

9
=9
1

9
= 84
3

9 8
= 72
1 1

9
= 126
4
Configuration
(1)7 (2)2
(1)8 (2)0 (3)1
(1)6 (2)3
(1)7 (2)1 (3)1
(1)5 (2)4
is indeed a very irregular function of the parameter E . On the average, however, strongly
increases with energy.
The irregularities of simply express the fact that
the system cannot assume arbitrary energy states,
but can absorb or emit energy only in discrete
quanta.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
242 / 324
Quantum mechanical counting of V

This property of the system becomes especially important if the typical
energy ( kT ) of a particle at temperature T is lower than or equal to the
energy differences which can be estimated by E = 1 = 8mL2 E /h2 or
E h2 /8mL2 . case only for very small system or very low temperatures
In reality, there is always an exchange of energy with the environment, so
that, strictly speaking, the only interesting number for us is the number of
states averaged over an energy interval E .
We first calculate the number of all microstates
(grid points) in the interior of the energy sphere,
X
(E , V , N) =
(E 0 , V , N)
(5.54)
E 0 E
It is easy to denote a mean function corresponding to . Differentiation of

with respect to the energy yields the mean density of states, which gives
the mean number of states per energy interval,
g (E , V , N) =
Huinan Zheng (USTC)
(E , V , N)
E
9On
(5.55)
00 :51 :25 2012 Spring
243 / 324
Quantum mechanical counting of VI

This is in completely analogy to the classical calculation of the energy surface
(E ) by means of (E ) = /E .
We can immediately
write down the volume of a 3N-dimensional octant of a
sphere with radius E = (8mEL2 /h2 )1/2 . Now we have, of course, 23N
octant (instead of 8 in the case of N = 1), and with L2 = V 2/3 instead of
Equation (5.29) we now have
#
3N "
1
3N/2
3N/2
E
(E , V , N) =
3N
3N
2
2 2
N
V
(2mE )3N/2

(5.56)
=
3N
3N
h3
2 2
Therefore we get
g (E , V , N) =
Huinan Zheng (USTC)
=
E
V
h3
N
(2m)3N/2 3N/21
E
3N
2
9On
(5.57)
00 :51 :25 2012 Spring
244 / 324
Quantum mechanical counting of VII

Observe that now g is no longer a dimensionless number, but rather is the
density of states, i.e., the mean number of states per energy interval E .
The mean number of states of energy E , (E ), is then given by,
(E ) = g (E )E =
=
E
V
h3
N
(2m)3N/2 3N/21
E
E
3N
2
(5.58)
To eliminate the dependence on E we rewrite E and E using

E =
8mEL2
h2
(5.59)
as E and E . We obtain
N

3N/2
(2m)3N/2
V
h2

(E ) =
E 3N/21 E
2/3
h3
8mV
3N
2
=
3N/2 1 3N/21
3N E
E
2
3N
2
Huinan Zheng (USTC)
9On
(5.60)
00 :51 :25 2012 Spring
245 / 324
Quantum mechanical counting of VIII

We can subdivide ln into (3N/2 1) ln E + ln E . Because
N ln E ln E , we can set ln E 0 and thus E 1. (E )
describes the states in an energy shell at E of width E 1.
Such an energy shell must be taken into account in quantum mechanics,
because the number of states is step function-like due to the shell effects.
Thus we get
(E )
3N/2 1 3N/2
3N E
2
3N
2
(5.61)
Rewrite for E , (E ) reads

(E )
Huinan Zheng (USTC)
V
h3
N
(2mE )3N/2

3N
2
9On
(5.62)
00 :51 :25 2012 Spring
246 / 324
Quantum mechanical counting of IX

The absolute entropy of an ideal gas, by considering the Gibbs correction
factor 1/N!, reads
(
"
3/2 #)

5
V
4mE
S(E , V , N) = Nk
+ ln
(5.63)
2
Nh3
3N
The constant of Equation (5.45) has been identified with h3 .
Equation (5.63) is called Sackur-Tetrode equation.
Our general prescription for the calculation of the absolute entropy of a
classical system including the Gibbs correction now reads
S(E , V , N) = k ln (E , V , N)
(5.64)
with
(E , V , N) =g (E , V , N)E
g (E ) =
Huinan Zheng (USTC)
(E )
E
9On
00 :51 :25 2012 Spring
247 / 324
Quantum mechanical counting of X

1
(E ) =
N!h3N
Huinan Zheng (USTC)
d 3N q d 3N p
(5.65)
H(q ,p )E
9On
00 :51 :25 2012 Spring
248 / 324
Exercise Equations of state of the ideal gas

Use Equation (5.63) to calculate some properties of the ideal gas.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
249 / 324
Phase-space density, ergodic hypothesis I

All microstates of the energy surface of a closed system can be assumed with
equal probability. the basic postulate of statistical mechanics.
For nonclosed systems, however, it could very well be true that microstates
with a certain energy are more probable that microstates with another energy.
The microstates may no longer be counted equally, but must be multiplied by
a weighting function (q , p ) which depends on the energy of the state.
The probability density is called the phase-space density. It can be
normalized to 1,
Z
d 3N q d 3N p (q , p ) = 1
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(6.1)
250 / 324
Phase-space density, ergodic hypothesis II

If now f (q , p ) is any observable of the system, e.g., the total energy
H(q , p ) or the angular momentum L(q , p ), then in general one will
observe a mean value hf i of this quantity in a given macrostate, in which
each microstate (q , p ) contributes corresponding to its weight (q , p ),
Z
hf i = d 3N q d 3N p f (q , p ) (q , p )
(6.2)
The quantity hf i is thus called the ensemble average of the quantity f and
the phase-space density is the weighting function of the ensemble.
In the case of closed system, is given by
1
(E H(q , p ))
(6.3)
(E )
The phase-space density of a closed system corresponds to a certain ensemble of possible microstates
and is called a microcanonical ensemble, which we denote by the index mc.
mc =
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
251 / 324
Phase-space density, ergodic hypothesis III

For practical calculations, one sets

const. E H(q , p ) E + E
mc =
0
otherwise
The constant is determined by normalization,
Z
Z
d 3N q d 3N p mc = const.
(6.4)
d 3N q d 3N p = 1
(6.5)
E H(q ,p )E +E
We have (without the Gibbs factor 1/N!)

1
const. = (E , V , N)h3N
Huinan Zheng (USTC)
9On
(6.6)
00 :51 :25 2012 Spring
252 / 324
Phase-space density, ergodic hypothesis IV

Since the factor h3N will appear very frequently, from now on instead of
Equation (6.1) we will write
Z
1
d 3N q d 3N p (q , p ) = 1
(6.7)
h3N
and, respectively, for Equation (6.2),
Z
1
hf i = 3N
d 3N q d 3N p f (q , p ) (q , p )
h
(6.8)
The phase-space density now is a dimensionless number.

The phase-space density of the microcanonical ensemble reads, normalized
(again without Gibbs factor),
1
E H(q , p ) E + E
mc =
(6.9)
0 otherwise
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
253 / 324
Phase-space density, ergodic hypothesis V

Basic to ensemble theory is the further assumption that all thermodynamic
quantities of state can be written as an ensemble average of a suitable
microscopic observable f (q , p ).
The time dependence of the actual phase-space trajectory is of no importance
for the ensemble average. In equilibrium, all thermodynamic (macroscopic)
observables are independent of time.
In principle,
1
f = lim
T T
dtf (q (t), p (t))
(6.10)
If one could mathematically prove that the time average essentially leads to the same result as the ensemble average, then our preceding assumptions could be
founded purely microscopically.
ergodic hypothesis.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
254 / 324
Liouvilles theorem
Liouvilles theorem I
Examine some general properties of the phase-space density ( , p ).
The ensemble average for a system in thermodynamic equilibrium must be
time independent. - the phase-space density must not explicitly depend on
time. In this case (/t = 0), one deals with stationary ensemble.
To ensure complete generality we also want to allow for an explicitly time
dependence in (q , p , t).
Along the phase-space trajectory, the phase-space density changes with time.
The temporal change can, in general, be written according to Equation (5.4),
d
(q (t), p (t), t) =
(q (t), p (t), t) + [, H]
dt
t
(6.11)
If we consider a phase-space volume , each phase-space point of this volume

element can be assumed to be the starting point of a phase-space trajectory.
In the course of time, all systems will move to different
phase-space points, mapping the volume element at
time t to another volume element 0 at time t 0 .
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
255 / 324
Liouvilles theorem
Liouvilles theorem II
In this process no points are lost and also no points are gained. The mapping
can therefore be interpreted as the flux of an incompressible fluid without
sources or sinks.
Z
Z
d = v nd
(6.12)
t
where v is the flux velocity, which of course given by the vector (q , p ).

Equation (6.12) can be rewritten as,

Z
d
+ (v ) = 0
t
(6.13)
The divergence here reads explicitly,

(v ) =
Huinan Zheng (USTC)

3N
X
(q ) +
(p )
q
p
=1
9On
(6.14)
00 :51 :25 2012 Spring
256 / 324
Liouvilles theorem
Liouvilles theorem III

Thus, along a phase-space trajectory the continuity equation
+ (v ) = 0
t
(6.15)
holds, since the considered volume is arbitrary.

On the other hand, from Equation (6.14) we obtain, using Hamiltons
equations of motion,

3N
X
q
p
q +
p +
+
q
p
q
p
=1

3N
X
H
H
=
q p
p q
=1

3N
X
2H
2H
+
q p
p q
=1
(v ) =
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(6.16)
257 / 324
Liouvilles theorem
Liouvilles theorem IV
or
(v ) =[, H]
(6.17)
We now have for Equations (6.15) and (6.11),
d
=
+ [, H] = 0
dt
t
(6.18)
The total time derivative of the phase-space density vanishes along a

phase-space trajectory. Liouvilles theorem (1838).
For stationary ensembles (/t = 0), it follows that

3N
X
H
H
[, H] =
=0
q p
p q
=1
(6.19)
is a constant of the motion.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
258 / 324
The microcanonical ensemble I

Consider N identical copies of a closed system (an ensemble), each with the
macroscopic natural quantities of state (E , V , N).
Subdivide the energy surface into equally large surface elements i . Each of these surface elements
contains a number ni of systems (bundles of microstates). All together, we must of course have
X
N =
ni
(6.20)
i
The number of systems ni in a certain surface element i corresponds exactly to the weight of the
corresponding microstate in the ensemble. The
number ni /N can be interpreted as the probability for microstate i to lie in i .
The probability pi = ni /N therefore corresponds to the expression
(q , p )d 3N q d 3N p in the continuous formulation.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
259 / 324
The microcanonical ensemble II

The total number of ways W {ni } to generate a certain distribution {ni } is
given by
N!
W {ni } = Q
i ni !
(6.21)
if i runs over all surface elements.

Let i be the probability of finding one system within the surface element
i . Then the probability of having exactly ni systems in i is (i )ni . The
probability Wtot {ni } of finding a distribution {ni } on the surface element i
is given by
Wtot {ni } = N !
Y (i )ni
i
Huinan Zheng (USTC)
(6.22)
ni !
9On
00 :51 :25 2012 Spring
260 / 324
The microcanonical ensemble III

Now we ask for the most probable distribution {ni } of the N systems over
the phase-space cells (surface elements). Approximately by ln n! n ln n n
for ni , we calculate
X
ln Wtot = ln N ! +
(ni ln i ln ni !)
i
=N ln N N +
[ni ln i (ni ln ni ni )]
(6.23)
If ln Wtot is maximal, the total differential must vanish. Since the number N
is constant, it must hold that
X
d ln Wtot =
(ln ni ln i )dni = 0
(6.24)
i
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
261 / 324
The microcanonical ensemble IV

The differential of Equation (6.20) multiplied by an unknown Lagrange
multiplier reads
X
d N =
dni = 0
(6.25)
i
This equation is added to Equation (6.24), resulting in

X
(6.26)
as a condition for an extreme value of ln Wtot . Now all dni are independent
from each other, if subsequently Equation (6.20) is fulfilled by a convenient
choice of the Lagrange multiplier . Therefore
ln ni = + ln i
Huinan Zheng (USTC)
or ni = i e
9On
(6.27)
00 :51 :25 2012 Spring
262 / 324
The microcanonical ensemble V

Now one of the basic assumptions of statistical physics is that all microstates
(all phase-space points) are, in principle, equally likely, and thus have the
same probability i . The i s are simply proportional to the corresponding
surface element i . If all surface elements are chosen to be of equal size
and very small, the number ni of systems must be equal in all surface
elements.
A constant phase-space density on the energy surface is the most probable
one for a system.
If instead of the energy surface we consider a very thin energy shell between
E and E + E , and we have

ni
const.
H=E
pi = =
0
otherwise
N

const. E H(q , p ) E + E
mc =
(6.28)
0
otherwise
pi is understood to be the probability of finding a system in the ensemble in
the microstate (surface element) with number i.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
263 / 324
Entropy as an ensemble average I

An ensemble average
Z
1
hf i = 3N
d 3N q d 3N p f (q , p )(q , p )
h
(6.29)
The microcanonical phase-space density is given by

1
E H(q , p ) E + E
mc =
0
otherwise
(6.30)
On the other hand,

S(E , V , N) = k ln (E , V , N)
(6.31)
The entropy is therefore

Z
1
S(E , V , N) = 3N
d 3N q d 3N p mc (q , p ) [k ln mc (q , p )]
h
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(6.32)
264 / 324
Entropy as an ensemble average II

To prove this, insert Equation (6.30) into Equation (6.32), and bear in mind
that ln = 0 for = 0,

Z
1
1
1
3N
3N
S(E , V , N) = 3N
k ln
(6.33)
d qd p
h
E H(q ,p )E +E
The integrand is a constant on the energy shell and can therefore be pulled
out of the integral
Z
1
1
S(E , V , N) = k ln 3N
d 3N q d 3N p
(6.34)
h
E H(q ,p )E +E
Inserting here Equation (5.65) (without the Gibbs factor 1/N!, which has
already been omitted in Equation (6.29)), we get
S(E , V , N) =
1
k ln = k ln
(6.35)
as it should be.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
265 / 324
Entropy as an ensemble average III

Equation (6.32) is thus a slightly more complicated formulation of
Equation (6.31). It has the great advantage that it can be easily
transformed to other phase-space densities.
In general, we write
S = hk ln i
(6.36)
Thus, entropy is the ensemble average of the logarithm of the phase-space

density.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
266 / 324
The uncertainty function I

Consider an experiment that involves random events, e.g., throwing dice with
different possible outcomes. Let There be a probability pi attached to each
of the i possible outcomes of the experiment. In the case of an ideal one
has pi = 1/6 for i = 1, . . . , 6. In a series of N casts (N ), all
numbers from 1 to 6 will appear an equal number of times, namely, on the
average exactly ni = pi N times.
If instead of the ideal die we take a manipulated one, which for instance may
have the probabilities p1 = p2 = p3 = p4 = p5 = 1/10 and p6 = 5/10, then a
series of casts will yield the number 6 five times as often as any one of the
other numbers.
The result of a cast with the modified die can thus be predicted with a larger
certainty than that of an ideal die.
In the extreme case of a special die with p1 = p2 = p3 = p4 = p5 = 0 and
p6 = 1 one can even predict with absolute certainty what the next throw will
yield.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
267 / 324
The uncertainty function II

Whether is there a unique measure for the predictability (uncertainty) of a
random event which also can be used to compare different kinds of random
events.
The uncertainty function H should only be a function of the probabilities pi
of the events.
H = H(pi ),
i = 1, . . . (possible outcome of the experiment)
(6.37)
The uncertainty of a sure experiment should be H = 0. Thus,

H(p1 , p2 , . . .) = 0
for p1 = 0, . . . , pi1 = 0, pi = 1, pi+1 = 0, . . .
(6.38)
In addition, the measure for the uncertainty must be independent of the

enumeration of the pi .
H(. . . , pi , . . . , pk , . . .) = H(. . . , pk , . . . , pi , . . .)
Huinan Zheng (USTC)
9On
(6.39)
00 :51 :25 2012 Spring
268 / 324
The uncertainty function III

In the case of the dice we have already seen that the equal distribution
pi = const. is obviously the one with the largest uncertainty.
H = Hmax
for p1 = p2 = = const.
(6.40)
The uncertainty H(I AND II) of an experiment that consists of the logical
conjunction of two experiments I and II, with uncertainties H(I) and H(II) is
given by
H(I AND II) = H(I) + H(II)
(6.41)
if experiments I and II are independent.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
269 / 324
The uncertainty function IV

Now one can prove mathematically that the conditions (6.37) to (6.41)
uniquely fix the uncertainty function. It reads
X
H(pi ) =
pi ln pi
(6.42)
i
Condition (6.37) is trivial, while Condition (6.38) is at once clear because

ln pi = 0 for pi = 1, and because pj ln pj = 0 for pj = 0. Condition (6.39) is
trivial as well, since in Condition (6.42) the enumeration of the summation
index may be changed. Condition (6.40) is now easily shown by forming the
complete differential,
X
dH =
(ln pi + 1)dpi
(6.43)
i
This must vanish for H = Hmax , but because of

X
pi = 1
(6.44)
i
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
270 / 324
The uncertainty function V

the pi are not all independent from each other. This extreme-value problem
leads to the statement pi = const. with the aid of the Lagrange multipliers.
Finally there results condition (6.41) with the probabilities pi for experiment I
and qk for experiment II, since the probabilities of statistically independent
events multiply in the case of AND-combinations,
XX
H(I AND II) =
(pi qk ) ln(pi qk )
i
XX
XX
(pi qk ) ln pi
(pi qk ) ln qk
=
i
pi ln pi
qk ln qk
=H(I) + H(II)
Huinan Zheng (USTC)
9On
(6.45)
00 :51 :25 2012 Spring
271 / 324
Examples and exercises I

Example (Motion in one dimension)
We consider a particle which can move only in the x-direction. Let the particle be
restricted to the interval 0 x a and let it move statistically forward and
backward.
The probability density (x) of finding the particle at
the point x is given by

1/a 0 x a
(x) =
0
otherwise
The corresponding uncertainty is
Z a
H=
dx ln = ln a
0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
272 / 324
Examples and exercises II

If the interval is enlarged by a factor > 1, we obtain

1/a 0 x a
0
(x) =
0
otherwise
and hence
H0 =
dx 0 ln 0 = ln a + ln
Thus, the uncertainty increases in an exactly defined manner, if > 1 (or

decreases, in the case < 1).
Example (The ultrarelativistic gas)

Now we want to calculate the thermodynamic properties of an ultrarelativistic
classical gas with the aid of the microcanonical ensemble.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
273 / 324
Examples and exercises III

Such a gas consists of massless particles that move with the velocity of light (e.g.,
photons).
= (p 2 c 2 + m2 c 2 )1/2 = |p|c
for m = 0
This ultrarelativistic gas is also used frequently as an easily calculable model for
particles with mass m 6= 0, if the available energy per particle mc 2 , or
equivalently, if the temperature is very high, so that the rest energy mc 2 can be
neglected compared to the kinetic energy.
Harmonic oscillators
Calculate the thermodynamic properties of a system of N classical distinguishable
harmonic oscillators with frequency in the microcanonical ensemble.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
274 / 324
The Canonical Ensemble I

The microcanonical ensemble is especially suited for closed systems with the
natural variables E , V , and N. Only in the most simple cases is this really
possible.
Therefore we now want to reflect on the probability distribution (phase-space density) of a system
at a given temperate (a system S in a heat bath
R). The total energy of the whole system
E = ER + ES
(7.1)
By definition, the heat bath is very large compared to the system itself,

ES
ER
= 1
1
(7.2)
E
E
Since now it is no longer the energy ES which is fixed, but the temperature,
the system S will be able to assume all possible microstates i with different
energies Ei with a certain probability distribution.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
275 / 324
The Canonical Ensemble II

However, we expect that microstates with very large Ei will appear only very
scarcely.
The probability pi , of finding the system S in a certain microstate i with the
energy Ei , will be proportional to the number of microstates S (Ei ), if S is a
closed system.
Analogously, pi is proportional to the number of microstates in the total
closed system for which S lies in the microstate i with the energy E .
Obviously, this is just equal to the number of microstates of the heat bath for
the energy E Ei , since S only assumes one microstate i,
pi R (ER ) = R (E Ei )
(7.3)
If the heat bath is very large, we can expand R with respect to Ei .

k ln R (E Ei ) k ln R (E )
Huinan Zheng (USTC)
[k ln R (E )] Ei +
E
9On
00 :51 :25 2012 Spring
(7.4)
276 / 324
The Canonical Ensemble III

For a very large heat bath the first two terms in Equation (7.4) are sufficient,
since then we have E ER Ei . However,
SR
1
[k ln R (E )] =
=
E
E
T
(7.5)
Insertion of Equation (7.5) into Equation (7.4) and exponentiation yields

Ei
(7.6)
R (E Ei ) R (E ) exp
kT
The number of microstates of the heat bath thus decreases exponentially
with the energy of the system.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
277 / 324
The Canonical Ensemble IV

Since E = const., R (E ) is a constant, and the probability pi reads

Ei
pi exp
kT
(7.7)
Here again, all microstates with the same energy Ei have the same
probability, only now the energy is no longer fixed, but for a fixed
temperature the system S can be on any of the possible energy surfaces.
P
pi can be normalized to 1, so that i pi = 1,

Ei
exp kT

pi = P
(7.8)
Ej
exp
j
kT
Here the sum
Huinan Zheng (USTC)
extends over all microstates (phase-space points).
9On
00 :51 :25 2012 Spring
278 / 324
The Canonical Ensemble V

In continuous
Equation (7.8) reads (i (q , p ) and
R notation
P
1
3N
3N
d
q
d
p),
3N
j
h
c (q , p ) =
h3N
exp [H(q , p )]
d 3N q d 3N p exp [H(q , p )]
(7.9)
Here we have abbreviate the frequently appearing factor 1/kT by .

Equations (7.8) and (7.9) yield the canonical phase-space density, which we
denote by the index c.
In the canonical ensemble the energy Ei of the system is not fixed. We
consider an ensemble of N identical systems. Each of these systems is in
some microstate at a fixed time. Let there be just ni systems in each
phase-space cell i . Then
X
ni
(7.10)
N =
i
Just as in the microcanonical case, pi = ni /N is the probability that

microstate i appears in the ensemble of N systems.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
279 / 324
The Canonical Ensemble VI

In the case of a system at a constant temperature, indeed all possible
microstates i and thus all possible energies Ei can be assumed with the
probability pi , but of course in equilibrium a certain mean value of energy will
be established, which we want to denote by U.
X
U = hEi i =
pi Ei
(7.11)
i
or with pi = ni /mathcalN,
X
NU =
ni Ei
(7.12)
In addition to Equation (7.10), Equation (Eq:7/12) is thus another condition

for the distribution {ni }.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
280 / 324
The Canonical Ensemble VII

Now a distribution of the systems {ni } over the phase-space cells i can be
achieved in many different ways, as we have already seen in the
microcanonical case. However, now we do not deal with surface elements on
the energy surface, but with the phase-space elements i in the total phase
space.
The probability of the distribution {ni } in the microcanonical case is
W {ni } = N !
Y (i )ni
i
(7.13)
ni !
Again, i is the probability of finding one microstate within the cell i .

Just as in the microcanonical case, we again look for the most probable
distribution {ni } of the systems over the phase-space cells. But in this
extreme value problem for W {ni } we now have two boundary conditions for
the {ni }, namely Equations (7.10) and (7.12).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
281 / 324
The Canonical Ensemble VIII

At first we form the logarithm of Equation (7.13)
X
ln W {ni } N ln N N
[(ni ln ni ni ) ni ln i ]
(7.14)
For an extremum of ln W the total differential must vanish,

X
d ln W {ni } =
(7.15)
We form the differentials of Equations (7.10) and (7.12) and multiply these
with the unknown factors and
X
dni = 0
(7.16)
i
Ei dni = 0
(7.17)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
282 / 324
The Canonical Ensemble IX

Equations (7.16) and (7.17) are now added to Equation (7.15),
XX
(ln ni ln i + Ei )dni = 0
i
(7.18)
Then
ln ni = + ln i Ei
or ni = i e e Ei
(7.19)
Equation (7.10) can be used to determine the factor e . We make use of the
fact that the elementary probabilities i for equally large phase-space cells
must be equal.
pi =
exp(Ei )
ni
=P
N
j exp(Ej )
Huinan Zheng (USTC)
(7.20)
9On
00 :51 :25 2012 Spring
283 / 324
The Canonical Ensemble X

We see that in this way we end up exactly with the form (7.8), only the
factor must still be determined from Equations (7.11) or (7.12),
P
i Ei exp(Ei )
U = hEi i = P
(7.21)
j exp(Ej )
This means, that if one fixes a certain mean energy U for the system, the
factor is a function of U.
We define the abbreviation
X
Z=
exp(Ei )
(7.22)
The quantity Z is so-called canonical partition function.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
284 / 324
The Canonical Ensemble XI

Now the entropy should result as the ensemble average of the quantity
k ln c , thus written in continuous notation,
Z
1
S = hk ln c i = 3N
d 3N q d 3N p c (q , p ) [k ln c (q , p )] (7.23)
h
The continuously written partition function reads
Z
1
Z = 3N
h
(7.24)
and the continuously written pi , c (q , p ) become

c (q , p ) =
Huinan Zheng (USTC)
exp [H(q , p )]
Z
9On
(7.25)
00 :51 :25 2012 Spring
285 / 324
The Canonical Ensemble XII

It follows for the entropy, according to Equation (7.23),
Z
1
S = 3N
d 3N q d 3N p c (q , p ) [kH(q , p ) + k ln Z ]
h
(7.26)
Now the first term in the square bracket (up to the coefficient k) yields
exactly the definition of the ensemble average of H, namely hHi, while the
second term (ln Z ) does not depend at all on the phase-space point and may
therefore be brought in front of the integral.
S = khHi + k ln Z
(7.27)
We have, according to Equation (7.21),

S = kU + k ln Z
(7.28)
Now we form S/U = 1/T , where we must, however, take care of the fact
that (U), as well as k ln Z ((U)), are functions of U.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
286 / 324
The Canonical Ensemble XIII

The mean energy U can, of course, be identified with the internal energy U,
thus
S
1
=
= kU
+ k +
(k ln Z )
T
U
U
U
(7.29)
Now one has
(k ln Z ) =
(k ln Z )
U
(7.30)
#
"
X
k
Ei exp(Ei ) = kU
(k ln Z ) =
(7.31)
and
so that Equation (7.29), reduces to

S
1
1
=
= k =
U
T
kT
Huinan Zheng (USTC)
9On
(7.32)
00 :51 :25 2012 Spring
287 / 324
The Canonical Ensemble XIV

The Lagrange multiplier from Equation (7.17) is really 1/kT .
However, Equation (7.28) is of great importance beyond the determination of
. If one rewrites it, using = 1/kT ,
U TS = kT ln Z
(7.33)
We know from thermodynamics that

F (T , V , N) = U TS
(7.34)
is the free energy of the system. So we have derived the following important
statement,
F (T , V , N) = kT ln Z (T , V , N)
(7.35)
This relation in the canonical ensemble at a given temperature is completely

analogous to the relation
S(E , V , N) = k ln (E , V , N)
(7.36)
from the microcanonical ensemble.

Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
288 / 324
The Gibbs correction factor I

We introduced a correction of the number of microstates (E , V , N) by the
factor 1/N! in the microcanonical ensemble,
Z
d (E , V , N) =
E HE +E
d 3N q d 3N p
h3N
Z
nd (E , V , N) =
E HE +E
d 3N q d 3N p
N!h3N
(7.37)
Here the index d stands for distinguishable and nd for nondistinguishable

particles.
This correction can be directly taken over to arbitrary ensembles, if
everywhere the infinitesimal phase-space volumes are replaced accordingly,
d d (E , V , N) =
Huinan Zheng (USTC)
d 3N q d 3N p
d 3N q d 3N p
(E
,
V
,
N)
=
nd
h3N
N!h3N
9On
00 :51 :25 2012 Spring
(7.38)
289 / 324
The Gibbs correction factor II

In the case of the canonical ensemble the phase-space density is given by
(r 1 , . . . , r N , p 1 , . . . , p N ) =
1
exp [H(r 1 , . . . , r N , p 1 , . . . , p N )]
Z (T , V , N)
(7.39)
The partition function Z (T , V , N) is here given by

Z
Zd (T , V , N) =
d 3N q d 3N p
exp(H)
h3N
(7.40)
and for nondistinguishable particles, by

Z 3N
d q d 3N p
Znd (T , V , N) =
exp(H)
N!h3N
(7.41)
The phase-space density (r 1 , . . . , r N , p 1 , p N ) for the distinguishable particles

denotes the probability density for particle 1 to be at r 1 and to have the
momentum p 1 , etc..
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
290 / 324
The Gibbs correction factor III

This can be used to calculate the probability density for any particle to be at
r 1 with momentum p 1 , etc.. One has only to sum up the probability densities
for arbitrary renumberings of the particles,
X
nd (r 1 , . . . , r N , p 1 , p N ) =
d (r P1 , . . . , r PN , p P1 , p PN )
(7.42)
P
The sum extends over all permutations (P1 , . . . , PN ) of (1, . . . , N). Now we
require that
H(r P1 , . . . , r PN , p P1 , p PN ) = H(r 1 , . . . , r N , p 1 , p N )
(7.43)
is valid for all permutations. This immediately yields

d (r P1 , . . . , r PN , p P1 , p PN ) = d (r 1 , . . . , r N , p 1 , p N )
(7.44)
Equation (7.42) becomes

nd (r 1 , . . . , r N , p 1 , p N ) = N!d (r 1 , . . . , r N , p 1 , p N )
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(7.45)
291 / 324
The Gibbs correction factor IV

We have not only found the foundation for the generalization of the Gibbs
correction factor to arbitrary ensembles, but with Equation (7.43) we also
have a criterion on systems to which it may be applied. These are systems
whose Hamiltonians are invariant under a different enumeration of the
coordinates and momenta.
The Hamiltonian of the ideal gas,
H=
N
N
X
X
p 2Pi
p 2i
=
2m
2m
i=1
(7.46)
i=1
fulfills this condition. Here in the last term the index Pi stands for an
arbitrary permutation of the numbers i.
An example where condition (7.43) is not fulfilled
H=
N
N
X
X
p 2i
1
2
+
m 2 (r i b i )
2m
2
i=1
Huinan Zheng (USTC)
(7.47)
i=1
9On
00 :51 :25 2012 Spring
292 / 324
The Gibbs correction factor V

Example (The ideal gas in the canonical ensemble)
The Hamiltonian of the ideal gas is
H=
3N
X
p2
2m
=1
The definition of the partition function with the Gibbs factor reads
Z
1
Z (T , V , N) =
d 3N q d 3N p exp[H(q , p )]
N!h3N
R
The integral d 3N q simply yields thep
factor V N , if V is the volume of the
container. With the substitution x = /2mp , all these
Example (The ultrarelativistic gas)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
293 / 324
The Gibbs correction factor VI

Example (Harmonic oscillators in the canonical ensemble)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
294 / 324
Systems of noninteracting particles I

The calculation of systems for which the Hamiltonian is a sum of one-particle
Hamiltonians,
H(q1 , . . . , q3N , p1 , . . . , p3N ) =
N
X
h(q , p )
(7.48)
=1
is especially easy in the canonical ensemble.

The partition function reads (with Gibbs factor)
Z
1
Z (T , V , N) =
d 3N q d 3N p exp[H(q , p )]
N!h3N
N Z
1 Y
=
d 3 q d 3 p exp[h(q , p )]
N!h3N =1
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(7.49)
295 / 324
Systems of noninteracting particles II

The integral can be interpreted as a one-particle partition function (N = 1),
Z
1
Z (T , V , 1) = 3
d 3 q d 3 p exp[h(q, p)]
(7.50)
h
so that
Z (T , V , N) =
1
N
[Z (T , V , 1)]
N!
(7.51)
for nondistinguishable particles, and

N
Z (T , V , N) = [Z (T , V , 1)]
(7.52)
for distinguishable particles, respectively.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
296 / 324
Systems of noninteracting particles III

Let us now consider the phase-space density of the total system
exp[H(q , p )]
Z (T , V , N)
1 exp[h(q1 , p1 )] exp[h(q2 , p2 )]
=
N!
Z (T , V , 1)
Z (T , V , 1)
N =
(7.53)
Up to the Gibbs factor, the probability N (q , p ) of finding the N particles

exactly at the phase-space point (q, p) equals the product of all probabilities
of finding a certain particle in a certain one-particle microstate.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
297 / 324
Systems of noninteracting particles IV

For a system of noninteracting particles, the probability of finding a particle
at the coordinate q with the momentum p, is given by the distribution
1 =
exp[h(q, p)]
Z (T , V , 1)
(7.54)
We can interpret such a system as an ideal ensemble. Each individual of the

N particles forms a system by itself and at a given time occupies a certain
one-particle microstate. All the other particles of the system form the heat
bath at a given temperature.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
298 / 324
Systems of noninteracting particles V

Example (The ideal gas)
Example (Mean velocity and most probable velocity)
Calculate the most probable, the mean,
and the root mean square absolute velocity in the ideal gas, using the velocity distribution

m 3/2
1
f (v ) =
exp mv 2
2kT
2
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
299 / 324
Systems of noninteracting particles VI

Example (Velocity distribution of an evaporating gas)
Calculate the velocity distribution f (v ) of particles evaporating from a hole in a
container with an ideal gas at temperature T .
Assume that the equilibrium in the interior of
the container is not disturbed by the evaporating
particles. In addition, calculate the mean velocity in z-direction and the mean squared velocity
of the evaporating particles, as well as the rate
R = d 2 N/dtdA of the particles leaving the container per unit time and hole area.
Show that in general R = 14 (N/V )hv i is the mean absolute value of the velocity
in the interior. What force acts on the container due to momentum conservation?
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
300 / 324
Calculation of observables as ensemble averages I

We assumed that all observables can be written as the mean value over the
ensembles with respect to an appropriate function f (q , q ),
Z
1
hf (q , p )i = 3N
d 3N q d 3N p (r , p )f (r , p )
(7.55)
h
The entropy is given as the ensemble average of fS (q , p ) = k ln (q , p ),
S = hk ln i
(7.56)
On the other hand, from Equation (7.56) we can determine the

thermodynamic potentials S(E , V , N) (microcanonical) and F (T , V , N)
(canonical). Therefore Equation (7.56) already contains all thermodynamic
properties of the system.
For the first step it is sufficient to calculate S(E , V , N) or F (T , V , N) from
Equation (7.56). All other thermodynamic quantities follow just as in
Chapter 4.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
301 / 324
Calculation of observables as ensemble averages II

With the aid of Equation (7.56) one can also obtain observables, which
thermodynamics does not tell us anything about. For instance, the
phase-space density is such an observable,
*
+
N
Y
(r 1 , . . . , r N , p 1 , . . . , p N ) = h3N
(r i r 0i )(p i p 0i )
(7.57)
i=1
The delta functions in Equation (7.57) just cancel the integral in

Equation (7.55) and yield the integrand at the points r 01 , . . . , r 0N , p 01 , . . . , p 0N .
Strictly speaking, Equation (7.55) stands for a general mapping of the
phase-space density on the real numbers.
The phase-space distribution of the particle i results from
(r 1 , . . . , r N , p 1 , . . . , p N ) (distinguishable particles),

i (r , p) = h3 (r i r )(p i p)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(7.58)
302 / 324
Calculation of observables as ensemble averages III

For noninteracting systems i (r , p) is identical to the corresponding
single-particle distribution (r 1 , p 1 ).
In the same way, one obtains the density of particles i in coordinate space,
i (r ) = h(r i r )i
(7.59)
or the momentum distribution of particle i,

i (p) = h(p i p)i
(7.60)
The total particle density in coordinate space is

* N
+
X
(r ) =
(r i r )
(7.61)
i=1
and the total momentum distribution is

* N
+
X
(p) =
(p i p)
(7.62)
i=1
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
303 / 324
Calculation of observables as ensemble averages IV

Observe the different normalizations of the quantities (7.59)-(7.62),
Z
Z
3
d r i (r ) = d 3 p i (p) = 1
Z
Z
d 3 r (r ) = d 3 p (p) = N
(7.63)
(7.64)
Another very interesting quantity is the distribution of the relative distances

of two particles, or the relative momenta. These follow from
fik (r ) = h(r |r i r k |)i
(7.65)
The distribution fik (r ) is the probability density for finding the particle i and
k at a separation r . The distribution of the absolute relative momenta reads
fik (p) = h(p |p i p k |)i
Huinan Zheng (USTC)
9On
(7.66)
00 :51 :25 2012 Spring
304 / 324
Calculation of observables as ensemble averages V

The mean distance of the particle i and k is
Z
hrik i = h|r i r k |i =
rfik (r )dr
(7.67)
Analogously, the mean relative momentum of particles i and k is given by

Z
hpik i = h|p i p k |i =
pfik (p)dp
(7.68)
0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
305 / 324
Calculation of observables as ensemble averages VI

Example (i (r ) for the ideal gas)
Example (The law of atmosphere)
Consider an air column above the surface of earth with
a basis area A. Calculate the density distribution of the
particles in the column under the influence of gravitation, at a given temperature T . Assume that the air
behaves like an ideal gas and assume gravity to be
constant.
Example (Relative momenta in the ideal gas)

Calculate the distribution of the absolute values of the relative momenta of two
particles in an ideal gas.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
306 / 324
Calculation of observables as ensemble averages VII

Example (Mean distance of two particles)
Calculate the probability density of finding two
particles of an ideal gas
at a distance r , if they
are contained in a spherical container of radius K .
What is the mean distance of two particles in
this sphere?
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
307 / 324
Connection between microcanonical and canonical

ensembles I
The canonical ensemble gives essentially the same result as the
microcanonical ensemble, although the possible microstates are very different
in both cases.
The probability of finding a system of the canonical ensemble in the
microstate (q , p ), is (without the Gibbs factor),
dp =
1
1
(q , p )d 3N q d 3N p = 3N exp [H(q , p )] d 3N q d 3N p
h3N
h Z
(7.69)
This probability is constant on the energy surface H(q , p ) = E (basic

postulate of statistical mechanics).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
308 / 324

ensembles II
Therefore we can easily calculate the probability of finding a system in any
microstate with an energy between E and E + E (the integrand is constant
in this case),
Z
1
1
d 3N q d 3N p
(7.70)
dp(E ) = exp(E ) 3N
Z
h
E H(q ,p )E +E
The last integral is just the number of all microstates in the energy shell of
thickness dE . With the number of states carrying an energy
H(q , p ) E ,
Z
1
d 3N q d 3N p
(7.71)
(E , V , N) = 3N
h
H(q ,p )E
the integral in Equation (7.70) assumes the value
Z
1
d 3N q d 3N p =
dE = g (E )dE
3N
h
E
E H(q ,p )E +E
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
(7.72)
309 / 324

ensembles III
if g (E ) is the density of state. The probability of finding a system in a thin
energy shell between E and E + dE , is therefore given by
dp(E ) = p(E )dE =
1
g (E ) exp(E )dE
Z
(7.73)
Of course, the partition function Z can also be expressed with the aid of g (E )
Z
1
Z (T , V , N) = 3N
h
Z
= dE g (E ) exp(E )
(7.74)
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
310 / 324

ensembles IV
In the case of discrete quantum mechanical states, the density of states g (E )
is replaced by the degeneration factor gE , which denotes the number of
quantum mechanical states which have exactly the same energy.
Equation (7.73) and (7.74) can be directly transformed to quantum
mechanics. Then
p=
gE
exp(E )
Z
(7.75)
is the probability for the quantum mechanical system to assume one of the
gE energy states E .
However, this does not yet have anything to do with quantum mechanics,
since in this case we are also dealing with distinguishable particles.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
311 / 324

ensembles V
The theory which has been presented up to now is called Maxwell-Boltzmann
statistics. This theory starts from enumerable particles and is afterwards
corrected with the Gibbs factor, if in real life the particles are
indistinguishable. In true quantum mechanics this will be no longer
necessary.
Equation (7.74) provides a very interesting relation between the canonical and
microcanonical ensembles. On the one hand, we have calculated the density
of states g (E ), which is tightly connected to , and on the other hand, the
integration in (7.74) yields the canonical partition function (free energy).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
312 / 324

ensembles VI
If Z (, V , N) is known, one can use this to calculate g (E ). We consider the
analytic continuation of Z () (V and N = const.) to the complex -plane.
Then
Z
Z () =
dE g (E )e E
(7.76)
0
is an analytic function of , if Re > 0. The relation Re > 0 guarantees

that the integrand in Equation (7.76) is bounded for all energies and that the
integral exists. Equation (7.76) is just the Laplace transformation of the
function g (E ).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
313 / 324

ensembles VII
For the Laplace transformation one can also write
down a reverse transformation. With = 0 + i 00
( 0 , 00 real) and 0 > 0 (arbitrary),
Z
0 +i
d Z ()e
E 0
0 i
d e (E
E )
E )
g (E )dE
(7.77)
d 00 e (
+i 00 )(E 0 E )
=e
Huinan Zheng (USTC)
e (E
=i
0 i
Z
d
0 i
we obtain
Z 0 +i
0 +i
0 (E 0 E )
2i(E 0 E )
9On
(7.78)
00 :51 :25 2012 Spring
314 / 324

ensembles VIII
where we have used the well-known formula
this into Equation (7.77) yields
Z
0 +i
d Z ()e E =
0 i
dE e
R
infty
(E 0 E )
dx e ikx = 2(k). Insert
2i(E 0 E )g (E )
=2ig (E 0 )
(7.79)
Hence we have found the inverse of the Laplace transformation

1
g (E ) =
2i
0 +i
d Z ()e E
(7.80)
0 i
The real part Re = 0 in this case is arbitrary but it must hold that 0 > 0
(so that Equation (7.76) remains analytic).
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
315 / 324

ensembles IX
Example (The ideal gas)
Example (Density of states for N harmonic oscillators)

Calculate the density of states g (E ) for a system of N harmonic oscillators from
the partition function.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
316 / 324
Fluctuations
Fluctuations I
Start with Equation (7.73)
pc (E ) =
1
g (E ) exp(E )
Z
(7.81)
where pc (E ) is the probability density of finding a system at a given

temperature ( = 1/kT ) at the energy E .
The density of states g (E ), is in general a function which strongly increases
with E (g (E ) E N , N ). By contrast, the Boltzmann factor e E
decreases exponentially with energy.
Therefore pc (E ) must be a function with a maximum. The maximum, at E ,
corresponds to the most probable energy. It is determined by

pc (E )
1 g
=
g exp(E ) = 0
(7.82)
E
Z E
or

1 g
1
=
g E E =E
kT
Huinan Zheng (USTC)
(7.83)
9On
00 :51 :25 2012 Spring
317 / 324
Fluctuations
Fluctuations II
The number of states in a constant small energy interval E obeys the
relation = g E , so that for Equation (7.83) we can also write

ln
S
1
1
or
(7.84)
=
=

E E =E
kT
E E =E
T
The most probable energy E of the canonical ensemble is thus identical to
the fixed energy E0 = const. of the microcanonical ensemble.
Equation (7.84) is just the prescription according to which the temperature in
the microcanonical ensemble has to be calculated at a given energy, cf.
Equations (4.8) and (4.9). Now the maximum of the function pk (E ) at E is
simultaneously the mean value hE i of all possible energies.
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
318 / 324
Fluctuations
Fluctuations III
This can be seen in the following way,
Z
1
hE i =U =
dE g (E ) exp(E )
Z 0
1
Z =
ln Z ()
=
Z
(7.85)
which because of F = kT ln Z and = 1/kT can also be rewritten as

follows,

F
F
2
hE i =U = +
= kT
kT
T kT
F
=F T
(7.86)
T
Using F /T |V ,N = S, this implies
hE i =U = F + TS
Huinan Zheng (USTC)
(7.87)
9On
00 :51 :25 2012 Spring
319 / 324
Fluctuations
Fluctuations IV
i.e., the mean value hE i is also identical to the fixed energy E0 of the
microcanonical ensemble
In the canonical ensemble the most probable
energy E is identical to the mean value of
all energies hE i and corresponds to the fixed
given energy E0 of the microcanonical ensemble.
The standard deviation from the mean value hE i is defined by
2 = hE 2 i hE i2
(7.88)
We differentiate Equation (7.85) with respect to ,

Z
1
U
=
dE g (E )E 2 exp(E )
Z 0
Z
2
1
+ 2
dE g (E )E exp(E )
Z
0
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
320 / 324
Fluctuations
Fluctuations V
= hE 2 i hE i2
(7.89)
With = 1/kT we obtain from Equations (7.88) and (7.89) for the standard
deviation

U
2
2 U
=
= kT
= kT 2 CV
(7.90)
T V ,N
The relative width is the ratio of to the mean energy U = hE i,
2
1p 2
=
=
kT CV
hE i
hE i
U
(7.91)
The total heat capacity, however, is proportional to the particle number N, as

is the internal energy U,

1
=O
(7.92)
hE i
N
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
321 / 324
Quantum Statistics I
10
Density Operators
Fundamentals
Pure and mixed states
Properties of the density matrix
The density operators of quantum statistics
11
The Symmetry Character of Many-Particle Wavefunctions
12
Grand Canonical Description of Ideal Quantum Systems
13
The Ideal Bose Gas

Ultrarelativistic Bose gas
14
Ideal Fermi Gas

The degenerate Fermi gas
Supplement: Natural units
15
Applications of Relativistic Bose and Fermi Gases

Quark-gluon plasma in the Big Bang and in heavy-ion collisions
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
322 / 324
Real Gases and Phase Transitions I

16
Real Gases
For absorption: Mayers cluster expansion
Virial expansion
17
Classification of Phase Transitions

Theorem of corresponding states
Critical indices
Examples for phase transitions
18
The Models of Ising and Heisenberg
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
323 / 324
The Models of Ising and Heisenberg
Huinan Zheng (USTC)
9On
00 :51 :25 2012 Spring
324 / 324

2012 Spring

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

2012 Spring

Hochgeladen von

Copyright:

Verfügbare Formate

Thermodynamics and Statistical Physics

Huinan Zheng (USTC)

00 :51 :25 2012 Spring

Bases for grade

Huinan Zheng (USTC)

Final exam (%)

00 :51 :25 2012 Spring

Huinan Zheng (USTC)

00 :51 :25 2012 Spring

Equilibrium and State Quantities

The Laws of Thermodynamics

00 :51 :25 2012 Spring

Phase Transitions and Chemical Reactions

Huinan Zheng (USTC)

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Huinan Zheng (USTC)

00 :51 :25 2012 Spring

Equilibrium and State Quantities

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Systems, phases and state quantities

Systems, phases and state quantities I

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Systems, phases and state quantities

Systems, phases and state quantities II

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Systems, phases and state quantities

Systems, phases and state quantities III

extensive (additive) state quantities

Huinan Zheng (USTC)

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Equilibrium and temperature the zeroth law of thermodynamics

Equilibrium and temperature the zeroth law of

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Equilibrium and temperature the zeroth law of thermodynamics

Equilibrium and temperature the zeroth law of

Huinan Zheng (USTC)

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Equilibrium and temperature the zeroth law of thermodynamics

Equilibrium and temperature the zeroth law of

at constant pressure and constant particle number

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Equilibrium and temperature the zeroth law of thermodynamics

Equilibrium and temperature the zeroth law of

Huinan Zheng (USTC)

00 :51 :25 2012 Spring

Equilibrium and State Quantities

Equilibrium and temperature the zeroth law of thermodynamics

Example the ideal gas I

not until 1802 did Gay-Lussac

relationship holds between pressure, volume and temperature

Equilibrium and State Quantities

Equilibrium and temperature the zeroth law of thermodynamics

Example the ideal gas II

Huinan Zheng (USTC)

00 :51 :25 2012 Spring