You are on page 1of 448

Preface

I.

xiv

The Nature ofThermodynamics


1.1

What is ther1nodyn

1.2

Definitions

ics?

1.3

T he kilomole

1.4

Limits of the continuum

1.5

More definitions

1.6

Units

I.7

Temperature and the zeroth law of thermodynamics

1.8

Temperature scales

6
6

Problems

I0

12

16

Contents

Yl

2.

19

Equations of State
2.1

Introduction

2.1

Equation of state of an ideal gas

21
22

2.3

23

s' equation for a real gas

2.4

P-v-T surfaces for real substances

2.5

Expansivity and compressibility

2.6

An application
Problems

25
27

29

31

3.

The First Law ofThermodynamics


Configuration work

3.2

Dissipative work

3.3

Adiabatic work and internal energy

3.4

Heat

3.5

Units of heat

3.6

The mechanical equivalent of heat

3.7

Summary of the first law

3.8

Some calculations of work

40
41

43

Problems

4.

37

3.1

45
46

46
47

48

Applications of the First Law


4.1

Heat capacity

4.2

Mayer's equation

4.3

Enthalpy and heats of transformation

4.4

Relationships involving enthalpy

4.5

Comparison of

4.6

Work done in an adiabatic process


Problems

64

SI

53

54

and h

59

61
61

57

35

Contents

5.

VII

5.1

The Gay-Lussac-Joule experiment

5.2

The joule-"Thomson experiment

5.3

Heat engines and the Camot cycle


Problems

6.

69
72
74

80

The Second Law ofThermodynamics


6.1

Introduction

6.2

The mathematical concept of entropy

6.3

lrreversible processes

6.4

Carnot's theorem

6.5

The Clausius inequality and the second law

6.6

Entropy and available energy

6.7

Absolute temperature

6.8

Combined first and second laws


Problems

7.

67

Consequences of the First Law

85

87
88

89

91
94

97

98
I03

I 04

I 07

Applications of the Second Law


I 09

7.I

Entropy changes in reversible processes

7.2

Temperature-entropy diagrams

7.3

Entropy change of the surroundings fc>r a eversible process

7.4

Entropy change for an ideal gas

7.S

The Tds equations

7.6

Entropy change in irreversible processes

7. 7

Free expansion of an ideal gas

7.8

Entropy change for a liquid or solid


Problems

Ill

I I0

Ill

113

I 21

1 22

118

Ill

Contents

VIII

8.

Thermodynamic Potentials
81

Introduction

8.1

The Legendre transformation

8.3

Definition of the thersrnodynamic potentials

8.4

The Maxwell relations

8.5

The Helmholtz function

8.6

The Gibbs function

8.7

Application of the Gibbs function to phase

8.8

An application of the Maxwell relations

8.9

Conditions of

Problems

9.

.
.

1 29
1 30
1 31

1 34
1 34

1 16

e equilibrium

1 37

142

1 43

1 46

..

.
.

.
.

.
.

'

...

::
..

. . :.

.
.

..

.
:
. .

. .. .

The Chemical Potential and Open

149

Systems
9. I

The chemical potential

9.2

Phase equilibrium

9.3

The Gibbs phase rule

9.4

Chemical reactions

9.5

Mixing processes
Problems

I 0.

127

15I

ISS
1 57

160

1 62

166

. .:
.

.. .

. .

. .

. . .:

. ...

.< <

. .

.
. .

The Third Law of Thermodynamics


I 0.1

Statements of the third law

I 0.2

Methods of cooling

I 0.3

Equivalence of the statements

175

I 0.4

Consequences of the third law

178

Problems

1 79

I7I

1 74

... ..

..

. ..

: .

:.

1 69

..:.:.

-: : :-:-;:
: : .- : :.:.. .

:.

..

..
. ;

Contents

II.

ix

11 1

Basic assumpti ons

11.2

Molecular flux

ll.l

Gas pressure an d the ideal ps law

I1.-4

Equ

186
188

190

.
'

11.5

Specific heat capacity of an ideal gas

I 1.6

Distribution of molecular spee ds

II.7

Mean free path and collision frequency

I 1.8

Effusion

191

193
198

10 I

processes
Problems

203

208

213

Statistical Thermodynamics
12.1

Introduction

12.2

Coin-tossing

ll.l

Assembly of distinguishable

11.4

Therpanoclynamic probability and e ntropy

12.5

Quantum states and energy le\'els

12.6

Density of quantum states

..

..

.
.

..

215
116
es

111
113

215

229

23 I

Problems

13.

183

of enersr

II.9

12.

181

inetic Theory of Gases

The

..

.
.

..

Classical and

.
.

'

uantum Statistics

I ].1

Boltzmann statisti cs

13.1

The method of Laaan multipliers

13.3

The Boltzmann

1].4

The felstfliDirac

ll.S

The

235

bution

138

ZaiZ
244

136

233

Contents
13.6

Dilute gases and the Maxwell-Boltzmann distribution

13.7

The connection between classical and statistical


thermodynamics

14.

Comparison of the distributions

13.9

Alternative statistical models

253

254

2S7

The Classical Statistical Treatment


of an Ideal Gas

261

14.1

Thermodynamic properties from the par tition function

14.2

Partition function for a gas

14.3

Properties of a monatomic ideal gas

14.4

Applicability of the Maxwell-Boltzmann distribution

14.5

Distribution of molecular speeds

14.6

Equipartition of energy

14.7

Entropy ch ange of mixing revisited

14.8

Max well's demon


Problems

15.

248

13.8

Problems

246

265
266
268

269

270
271

273

275

The Heat Capacity of a Diatomic Gas


279

15.1

Introduction

15.2

The quantized linear oscillator

I 5.3

Vibrational modes of diatomic molecules

I 5.4

Rotational modes of diatomic molecules

15.5

Electronic excitation

15.6

The total heat capacity


Problems

263

289

287
288

279
282
284

277

Contents

16.

XI

The Heat C
Introduction

16.1

Einstein's theory of the heat

16.3

Debye's theory of the heat capacity of a solid

of a solid

296

305

308
agnet

17.3

Properties of a spinl/2

17.4

Adiabatic demagnetization

17.5

Negative temperature

17.6

Ferromagnetism
Problems

agnetism

307

etism

318

321

325

328

33 I

Bose-Einstein Gases
ation

333

18.1

Blackbody

18.2

Properties of a photon gas

18.3

Bose-Einstein condensation

18.4

Properties of a boson gas

18.5

Application to liquid helium


Problems

293

30 I

Introduction

17.1

19.

193

The Thermodynamics of
17. I

18.

291

16.1

Problems

17.

of a Solid

338
340

345
347

349

Fermi-Dirac Gases
19.1

The Ferwni energy

19.2

The calculation of p.(T)

353

355
357

liS

XII

Contents

Free electrons in a metal

1 9.4

Properties of a fermion gas

1 9.5

Application to white dwarf stars

364
367

370

Problems

20.

36 1

19.3

lnformationTheory

373

20. 1

Introduction

20.2

Uncertainty and information

20.3

Unit of information

20.4

Maximum entropy

20.5

The connection to statistical thermodynamics

20.6

Information theory and the laws of thermodynamics

3 75

375

3 79
381
384
386

20.7

Maxwell's demon exorcised

387

388

Problems

. ....:.:.,.">'
;

. .
.:-...
. :
..:

':::

.
.<
..=:.-;.
;;:;..;..
'I';
.: :::
':
":
..:.:

:I'
'
.'':
<-'
<

...,
.
.-:
.-:
.- .:;
..-:
::. .=:
:
..:::
'.,;
::=-:::
.-=
:;;.::
. .
..
.:
-:
,,
;.
.;
..
..
.;:
;.
:
.
xo
:..
:-.
....
.
. =..
.:::
..:
..:
.:
.,
;.
::

..: :
.,::
.:
:
::::--::

;::
:.-:
- ::.::: :=
:
:...
: v.-:

:; ;:;:::
;:;
::::;:;
::
;:;
::
-: ::; ;:.:- :::
:=:
;:
: ;:::?: ;:::: ::::
::
::: :::;::
:;:
:::
::
=: . :::;:; :::::::
::
:: ::-.; :
: :::
:==
!:::: ::: ::
:: .:;:. :;: ::::::.
:
"='
:=:::;;
:"
::.
: : ;:::
$=:-=

:
:-:..
...

;:=:
::.::
,-.v
<.==- . .:::
z

:::.:.:
:.
:::
;:
: ::...
z{!
...:::
:.;
::
-.-:.::::. "';:
:
::::.:...::
:::::: .=:.:,;:.:
:::::::
==
=
:
.
-:-:
:::
: :f.
:::::
:

=:
:
:: ..:x
:-:v
:-J'..=
:- : <
-.:;
>=.:y..
.
:
:
=:
--:
;::::
:::.
=
=:::: :
:
: :=
:
: ::
,.:
::.
:- -:. .
.:;
:-=
:
:-=::=

:::
:

'"

:
:

Appendices

A.

Review of Partial Differentiation


39 1

A. I

Partial derivatives

A.l

Exact and inexact differentials


Problems

39 I

393

3 99

40 I

B.

Stirling's Approximation

C.

Alternative Approach to Finding the


Boltzmann Distribution

D.

Various Integrals

407

403

:
::

xiii

Contents

Bibliography

41 3

Classical thertnodynamics

Kinetic theory of gases


Statistical mechanics
Special topics

41 3

4 14
41 4

41 5

Answers to selected problems

Index

424

417

This book is intended as a t ext for a one -sem est e r u nder gr aduat e cour se in
t herm al p hysics Its obj ective is to provid e third- o r fo urt h year p hysics stu
dent s wit h a solid intr oduct ion to t he cl assical and st at ist ical t heor ies of t her
m ody namics. No prep arat ion is ass umed beyond colle ge-level general phy si cs
and advanced calcul us. A n acquaintance with prob ab il ity and st at i stics is h elp ...
f ul b ut is by no means necessary.
The curr ent pr actice in many coll e g e is to <>ffer a cour se in cl assical
therm odyn am ics with littl e or no ment ion of the st atist ical t heory or vic e
versa. Th e argument is t hat it is in1p ossible to do j ustice t o b ot h in a one
semester cour se. On t he b asis of m y own te ach i ng exp er ience I str ongly dis
agree. The standar d treatment of temperature work, heat entropy, etc. oft en
seems to t he student l ike an endle ss coll ect i<l n of p art ial der ivatives t hat shed
.

..

..

only limited light on the underlying physics and can be abbreviated. The fun

of classical therm(>dynamics can easil y be grasped in little


semester. leaving a n1 pl c tin1e t o gain a reasonably thorough

damental concepts
more than half -a

of the statistical method.


statistical t h erm odynam i cs

understanding
Since

n ot

structure the entire

good r e as ons not


providing a

The p h ysics

XIV

course

subsumes the classical results

why

around the statistical approach? There are

classical t h e < l ry is ge ne ral

simple._ and d irec t


kind of visceral, int uit iv e com pr e hen si on of t herm al processes.
st udent not confronted wit h thi s rem ark abl e phenom enol ogical
t o do so. The

'I

Preface

XV

conception is defi ni teiy

deprived. Tt) be sure, the inadequacies of c l assic al

ther

modynamics becorne apparertt upon close scrutiny and invite inquiry about a

more fundamen t al descripti on This, of course, exactly reflects the historical


.

development of the subject. If only the statistical picture is presented, however,

it

is my observation that the student

fails to appreciate fully its more abstract

concepts, given no exposure to the related classical ideas first. Not only do clas

si c al and st a tistical the rtn odyn amics in this se11se corr1 plemcnt each oth er, they

also bea u ti fully illustrate the physicist's perpetu al stri vi ng for descriptions of
gr eater power , el egance, uni ver s ali ty, and freedon1 from ambiguity.

Chapters 1 th rough

10

r epres e nt a fairly traditional introduction to the

classical theory. Early on emphasis is placed on the advantages of expressing

the fu!tdamental la\vs in terms of state variables, qu anti tie s whose differentials

are exact. Accordingly, the sea rch for i nt egr ati ng factors for the differentials of
work and heat is discussed. The elaboration of the first law is follow ed by
chapters on app lic ati ons and consequences. Entrop y is presented both as a

useful mathematical variable and as a phenomenological construct necessary


to explain why there are processes permitted by the first law tha.t do not occur

in nature. Calculations are then gi ven of the change in entropy for variclJS

reversible and irreversible. processes The therm od ynamic p otentials are


.

br oach ed vi a the Legen d re transformation fo llov. in g elucidation of the ratio


'

nale for having precisely four such quantities. The conditions for stable

equ i

librium are examined in a section that rarely appears in undergraduate texts.

Modifications of fundarnental relations to deal with open systems are treated

in Chap ter

9 and the third l aw is given its due in Chapter 10.

The kinetic theory of gases, treated in Chapter 1 1 , is

con

cerned with the

molecuJar basis of such thermody n amic prop erties of gaes as the tempera

ture, pressure, and thermal energy. It represe11ts, both logically and historically,
the tran sition between classic a l thermodyn am i cs and the s tat istical t heo ry

The unde rlying principles of equilibrium statistical thern1odynamics are

introduced in Chap ter 12 through consideration of a simple c oin tossing


-

experiment. The basic concepts are then defined. The statistical interpretation
of a system con taining many molecttles is observed to require a knowledge of
the properties of the in div idua l molecules making up the system. This infor

mation is furnished by the <.tuantum rnechanical notions of energy levels, quan


tum states, and intermolecular ft1rces. In C h apt er 13, the explication of

classical and quantum statistics and the derivation of the particle distribution
functions is based on the method of Lagrange multipliers. A di scussio of tl1e

connection betwe en cla ssic a l an d statistical the rm o dynamics c om pl etes the


development of the mathematica1 formulation of the statistical theory.

Chapter 14 is devoted to the statistics of an ideal gas. Chapters 15 througl1 19

present important examples of the application of the statistical method "The


.

last chapter introduces the student to the basic ideas of inforn1ation theory
and

offers the intriguing thought

that statistical t hermo dynam ics is but a spe-.

cial case of some deeper, more far-reaching set of physical principles.

Preface

xvi

Throughout the book a serious attempt has been made to keep the level
of the chapters as uniform as possible. On the other hand. the problems are
intended to vary somewhat more widely in difficulty.
In preparing the text, my greatest debt is to my students, whose response
has provided a practical filter for the refinement of the material presented
herein.
A.H.C.
Drew University

ACKNOWLEDGMENTS
In addition to my students at Drew University, I owe .thanks to two colleagues
and friends, Professors Robert Fenstermacher and John OJlom, who have
encouraged me at every turn during the writing of this book. I am indebted to
.

Professor Mark Raizen of the University of Texas at Austin. who reviewed the
manuscript and used it as the text in his thermal physics course: his comments
were invaluable.

I am especially grateful to Professor Roy S. Rubins of the University of


Texas at Arlington for his thoughtful and thorough critique. I also received
useful feedback from other reviewers, whose suggestions contributed substan
tially to an improved text. Th ey are AnjtJm Ansari_ University of Illinois at
Chicago; John Jaszczak, Michigan Technol(>gical University: David Monts,

Mississippi State University; Hugh Scott, Oklahoma State University: Harold


Spector, Illinois Institute of Technology at <:hicago; Zlatko Tesanovic, Johns
Hopkins University.
I thank my editor Alison Reeves and her assistants, Gillian Buonanno
and Christian Botting, for their support, guidance, and patience. Production
editors Richard Saunders and Patrick Burt of WestWords Inc. were particu
larly helpful. FinaiJy, I am extremely graleful to Heatl1er Ferguson. who turned
my lecture notes into a first draft, and to l.,ori Carucci and her daughters
Amanda and Brigette. who prepared the final manuscript.
Without all of these people, the book would never have seen the light
of day.

1. 1

What Is Thermodynamics?

1.2

Definitions

1.3

The Kilomole

1 .4

Limits of the Continuum

1 .5

More Definitions

1.6

Units

I. 7

Temperature and the Zeroth Law

1 .8

of Thermodynamics

10

Temperature Scales

12

1.1

WHAT IS THERMODYNAMICS?

Thermodynamics is the study of heat in the field of physics. The central con
cept of thermodynamics is temperature. Since temperature is not expressible
in terms of basic mechanical quantities such as mass, length, and time, it is evi
dently a fundamental notion that sets thermodynamics apart from other
branches of physics.
The development of ther1nodynamics provides some of the most fasci
nating chapters in the history of science. It began at the start of the Industrial
Revolution when it became important to understand the conversion of heat
into mechanical work. The experin1ents of Joule, Hirn, and others, and the the
oretical studies of Helmholtz resulted in the principle of the conservation of
energy when applied to therrnal phenotnena. The principle became the first
law of thertnodynamics. Mayer postulated the equivalence of heat and work
and made an estimate of the mechanical equivalence of heat.
The subsequent progress of thermodynamics owes much to the

Frenchman, Sadi Carnot, whose treatise of 1824 led io one of the most far

reaching principles of physical science: Carnot's theorem, which is, in effect,


the second law of thcrrnodynamics. Actually, the principle emerged before the
first law. It was the outgrowth of Carnot's interest in the practical question of
the efficiency of steam engines. His \vork, put in simple mathematical form by
Clapeyron, attacked the more fundamental problem of the efficiency of heat .

engines in general.
The concept of entropy began to appear quite early in papers by
Clausius and William Thomson (Lord Kelvin), but it was not until

1865

that

Clausius saw fit to give it a name and a full definition. Later, Nernst and
Planck added a third law, which is a statement about the behavior of thermo
dynamic quantities, including entropy, at the temperature of absolute zero.
In this period classical thermodynamics was worked out in essentially its
present form. It is a phenomenological theory, describing the macroscopic prop
erties of matter, most of which are amenable to direct measurement. No assump
tions are made about the fine structure of material substances. No attempt is
made to explain

underlying causes or to provide a mechanistic description. As a


3

The Nature ofThermodynamics

Chap. I

...

'.

y;.:..:,.
.
,
:: :

'

.. .

Figure 1.1 Ludwig


Boltzmann, 1844-1906 whose
work led to an understanding
of the macroscopic world on
the basis of molecular
dynamics. (Courtesy of
American Institute of
Physics/Emilio Segre Visual
Archives.)

..

::t. '

' .
..

}"'

'J
.

consequence" the theory has the advantages c.>f great s implicit y.. broad gene rality

and a close connection between experimental results and familiar concepts. The

noted experimentalist

W. Bridgman said .. ''The J aws of thermodynamics have a

different feel from most of the other laws of physics. There is something palpably
verbal about them

they smeJI of their hum<tn origin:

Toward. the end of the nineteenth century, \vhen the aton1ic nature of
.,

matter began to be understood .. ways were found to e x pr ess the pressure. tem

perature, and other macroscopic p roper t ies of a gas in terms <)f a ver a ge values
su ch

of the properties of molecules,

as the 1 r kinetic energy. The r es ul ts. with

which Maxwelrs name is closelv asS()Ciated. came to he c<tlled the kinetic the.,

ory of gases.
The kinetic theory eventually expanded int o the f ar more comprehen
sive statistical mechanics (or statistical thermodynamics) of BoJtztnann and

Gibbs, and ultimately encompassed the ideas of quantum mechanics. The statistical approa ch takes account of the molecular C()nstitution of matter and

reveals a deeper foundation on which the laws of thermodynamics exist . The

connection between the classical and statistical theories is associated \Vith the
fact that macroscopic measurem e nt s arc averages of the hehavior of astro

nomical numbers of particles.


Thermodynamics is by no means

closed sub.iect. In recent years it has

been possible to ext en d the statistical theory

to

include nonequilibrium

processes and even nonlinear effects. At the present time. the re1ationship

Sec. 1.2

Defrnit!ons

between thermodynamics and information theory, and the study of chaotic


behavior in thermodynamic systems, attract a great deal of attentiott. Modern

ideas, furthermore, have been shown to be applicable to a wider variety of


phenomena than hitherto suspected.
Thermodynamics has captured the imagination of many of the greatest
minds of science. Einstein, who was captivated by the subject, wrote: "A theory
is the

more impressive the greater the simplicity of its premises, and the ntore

extended its area of applicability. Therefore, the deep impression which classi
cal thermodynamics made upon n1e. It is the only physical theory of universal
content concerning which I am convinced that, within the applicability of its

basic concepts, it will never be overthrown.''That is quite a statement, indeed!

1.2 DEFINITIONS
In developing the basic ideas of thermodynamics the originators of the theory
were careful not to be so concise as to render definitions sterile. Tne following
brief definitions will be illustrated by examples in future sections.

Thermodynamics can be described as the study of equilibrium

in which

temperature is an

properties

important variable. All of the words in italics need

to be defined.

In thermodynamics we are concerned with a S}'Stem, some portion of the


physical world. The system could be a container of gas, a piece of metal, a mag

net. The system must not interact chemically with the vessel that contains it.

(The behavior of a liquid must not be influenced by the test tube that holds it.)
A system may exchange energy with other systems, which constitute the

roundings

comprtse a

open

sur

of the given system. The system, together with its surroundings,

untverse.

We classify S)'Stems as to whether they are open, closed, or isolated. An


system can exchange mass and energy with its surrottndings. A

closed

system cannot exchange mass with its surroundings, but can exchange energy
in other forms. An isolated system cannot exchange mass or energy in any
form with its surroundings; it is completely cut off from other systems.
The quantities we use to describe the macroscopic behavior of a system

are called

properties,

observable characteristics of a system. Other names are

thermodynamic variables or thermodynamic coordinates. An extremely impor


tant concept is that of a state variable, a property whose differential is exact.
Properties are extensive or intensive. An

extensive

property is propor

tional to tl1e mass. An example is the volume V: if the mass is doubled, the vol
ume is doubled (assuming that the density remains constant). An

intensive

property is independent of the mass. Temperature Tis an intensive property;


its value is not affected by a change of mass. Pressure
ther examples of intensive properties.

P and density pare fur

Chap. I

The Nature ofThermodynamics

Inherently extensive properties are given by capital letters. Inherently


intensive properties are denoted by lowercase letters. There are two important
exceptions: The temperature Tis always capitalized to avoid confusion with the
timet: and the pressure P is capitalized to distinguish it from the probability p.
An extensive property can be converted to an intensive property by
dividing by the mass. This is called a specific value:
Specific value -

value of the extensive property


-

mass of the system

In this text, we will go back and forth between extensive properties and their
corresponding specific values.

1.3 THE KILOMOLE


'

The kilomole is a unit of mass defined as follows:


I

kilomole - mass m in kilograms equal to the molecular weight.

Thus one kilomole of oxygen gas (02) is equal to 32 kg. (The mole is a
unit of mass more familiar to chemists: one mole is equal to the mass in grams.
Thus a mole of oxygen gas is 32 g.)

1.4 LIMITS OF THE CONTINUUM

We tacitly assume that classical thermodynam ics is a continuum theory, that


properties vary smoothly from point to point in the system. But if all systems
are made up of atoms and molecules (as our definition of the kilomole
implies), it is reasonable to ask: how small a volume can we be concerned with
and still have confidence that our continuum theory is valid?
We can obtain an approximate answer to this question by invoking
Avogadro's law, which states that at standard temperature and pressure (0C
3
and atmospheric pressure), one kilomole of gas occupies 22.4 m and contains
6.02 x 1026 molecules. (The latter is called Avogadros number, NA . ) Then
.3
22.4 m kilomole -l

m3

This molecular density is sometimes called Loschmidt's number. Using it we can


easily show that a cube one millimeter on each side contains roughly 1016 mole
cules, whereas a cube one nanometer (1 0-9 tn) on a side has a very small prob-

Sec. 1.5

More Definitions

ability of containing eve11 one molecule. We can therefore be rasonably cer


tain that classical thermodynamics is applicable down to very small macro
scopic (and even microscopic) volumes, but ultimately a limit is rea ched where

our theory will break down .

1.5 MORE DEFINITIONS

The state of a system is defined as a condition uniquely specif i e d by a set of


properties. Examples of such properties are pressure, volume, and tempera
ture. The question arises: how tnany properties are required to specify the
state of a thermodynamic systerrt? By "required number" we mean that every
time a system with the gi v en properties is subjected to a particular environ

ment, every feature of its sllbsequent behavior is identical. In classical


mechanics, if the displacement and velocity of a system are known at some
instant of time, as well as the forces acting on it, the behavior of the system is
predicted for all times. Most frequently, the t hermodynamic state of a single
component system is also specified by two independent variables.

An equilibrium state is one in wh ic h the properties of the system are un i


form throughout and do not change with time unless the system is acted ttpon
by ex t ern al influences. A non-equilibriltm state characterizes a system in which
gradients exist and whose p r ope rties vary with time (the atmosphere and the

oceans are examples). State variables are properties that describe equilibrium
states. An equation of state is a functional relationship among the state vari

ables for a system in equilibrium. A path is a series of states through which a

system passes.

In introductory physics, a 'change of state" is frequently used to den ote a


transition from a liquid to a gas, or from a solid to a liq1Jid, etc. In therlnodynam

ics, such a change is referred to as a charzge of phase or phase tratzsformation.

If the pressure .P, the volume V, and the t en1 pe ra t u re Tare the state vari

ables of the system, the equation of state takes the form


f(P, V,

T)

0.

( 1.1 )

This rel a tions hip reduces the number of independent variables of the system
from three to two. The function f is assun1ed t o be given as part of the specifi
cation of the system. It is customary to represent the state of such a system by
a point in three-dimensional p V- T space. The equation of state then defines a
surface in this space (Figure 1.2). Any point lying on this surface represen t s a
state in eq uilib rium. In thermodynamics a state automatically means a state in
eq uilibr iu m unless otherwise specified.
The boundary bet\\'een a system and its surroundings th roug h which
changes may be imposed is called a system wall. An adiabatic wall is a boundary
tha t permits no heat interaction with the sttrrou n dings The word comes from
..

The Nature ofThermodynamics

Chap. l

T
te
ta
s
m
iu
r
ib
il
u
q
e
--,
_
/-

---f(P. V. T)
v

Surface in P- V- T
space. Points on the surface
represent equilibrium states of
the system whose equation of
0.
state is f ( P. V. T )

Figure 1.2

I'

ia
ad
is
m
ste
sy
ed
lat
iso
An
...
h
ug
o
r
h
g
t
in
go
t
no
..
ng
ni
ea
m
s.
the Greek adiabato
y
l
ri
e
ssa
c
ne
t
no
is
ll
wa
c
ati
iab
ad
an
th
wi
t
m
s
e
y
s
a
r.
ve
we
Ho
d.
batically containe

isolated. Mechanical interactions can take place th rough adiabatic walls. For
e xample material can be removed or added the volume can change, a mag
ne ti c field can be applied, etc. A diatltert11al rail is a boundary that freely allows
,

heat to be exchanged The Greek word diatl1ern-zos me a ns Hheat through.'


If a system s t rong l y interacts \\'tth its walls. the pr<lhlcm becomes compli

cated. Think of panc ake batter stick ing to the griddlt!. or sulfuric acid 1n an

iron ve ssel !

A process is a c h an ge of s t ate expr c ,sed in terms of a path along the

e qu ation of state surface in Figure 1.2. In a c_vclicaiJJrt)cess. the initi al and final

states are the same. A qltasi-static prl)cess is a process in \\'hich. at each instant,

the system departs only infinitesimally fr<)m an equilihrium state. That is.
changes of state are described in te rms of differentials. An example is the

gradual compression of a gas.


A reversible prtJcess

is

a proces s \\'hose directi<)n can l)e reve rsed hy an

infinitesimal change in some property. It is a quasi-static process in which no


di ssipati ve forces such as friction are p rese nt All re versible processes are
.

quasi-static but a quasi-static process is not necessarily reversible. For exam


..

ple, a slow lea k in a tire is quasi-static but not reversih1c. A reve rs i ble process

is an idealization: friction is al w a y s present. An irre\'ersihle pr<>cess in v ol ve s a


finite change in a prope rty in a given step and includes dissipation (energy

J oss ) All natural processes are irreversible.


.

In many proce sses some IJT<)perty of the


,

s yst e n1

remains constant. An

isobaric process is a process in \\'hich the pr\!ssure is constant . If the

con stant the process


,

is

v ol ume

is

iS(JCIInric. And if th temperature d <)e s n 't cl1 an ge the


.

process is called isotlrertnal.

Suppose a piston enclosing a gas is immersed in a heat bath so that the

gas is kept at constant temperature. The gas is slowly co mp resse d (Figure

1.3).

Sec. 1.6

Units

,.
p

diathertnal wall
.. .
. .
.. . .
.
.
.

. . .
.
.. .
.
.
. . .. . . .
. . ..
.
.
;t
.
.
, ,

. .

.
0

piston

. . " .
.
..
.
.
.

'

.' .;

..

.: .> :.>.
.

v . .
. .

:0 :
. .
.
.. . .
.. ". .
.
.
.
. . .
.. .... .
. ..
. .

.
.

..

(P;, V,)
v

heat bath

gas

(a)

(b)

Figure 1.3

Example of a quasi-static, reversible and isothermal process.


(a) A piston slowly compresses a gas held at constant temperature. (b) A P-V
diagram representing the process; (P;, V;) and (P1, V1) denote the initial and
final states, respectively.
The process is quasi-static, reversible, and isothermal and can be described by
a path in a

P- V diagram connecting the initial and final states through a series

of infinitesimal changes.

If, instead of

gradual compression, the piston is given a violent push,

sound "'aves or shock waves and turbulence are generated, accompanied by


stro11g temperature and pressure gradients. The process is an irreversible, non
equilibrium process that cannot be represented by a path in the

P- V plane.

Only the end points can be plotted, representing the initial state and the final
state reached after equilibrium has been eventually reestablished.

1.6 UNITS
The lntertlational System of Units (SI) will be used, almost exclusively. The
system is based Oil the fundamental units of length, mass, and time

the meter

(m), the kilogram (kg) and the second

(s). respectively. In this system, the unit

of pressure is the pascal (Pa), equal to

Nm -2

Other units of pressure in common use ar e


1 ba r -- 105 Pa,
1 atm<>sphere (atm)
1 torr

1.01 X

lOS Pa.

133.3 Pa.

The atmosphere and the torr (named after Torricelli) are derived units. The
atmosphere is based on the use of a manometer to measure pressure (Figure 1.4).

Chap. I

10

The Nature ofThermodynamics

vacuum

p
h

.
: .
.
\. .
. .
.
.
.
.
> .
.

.
.
..-- m crcu rv

.
.
.
.
.

:
..
:
.
.:
.

:
.
.
. . . . . . . .. . ,..,,, :. .. . . . . . . .. . .
.
.
.. . .. ... .:.. . ...... . . . . . . ... . . . . .
. ...
. . . .. .. . . . ... ... . .
.
. . ..
:

. . .


.
:
.

.
.

..

'<

'

'

.,

:..

Figure 1.4 A simple mercury


manometer used to measure
the atmospheric pressure.

'

The he ight of a column of mercury is balanced by the force exerted by

the gas on the liquid surface. A p re ss u re of 1 atmosphere cau es the mercury


to rise to a height of

76 em. That

is. P

hpg. where P is the pressure of the

atm osphe re his the height of the mercury column. pis the density of m ercury

and g is the acceleration of gravity. Thus

1 atm

(0.76m) 1.36

The torr is then defined

as

,.

,.

..,

'

__

..,,

, , .
'"'

. ,

7fi0

9 .8

s-

1.01

105

Pa.

of 1 mm of mercury:

atm

133.3 Pa.

..s,.,

...
\ \4 ...., ,,
..:-';.,
, t.:.
W" '. #t::--T "C

. .\IT-"'
<('C 'U.-.2
., ,..
I -..-',.TY"
"'E:I
' C\
.
_,,-
,.. ._,
,.- .._ .....
....
o""...,....__.,.--;

"""
:WV
- "'
.
' ....,
_..-
_..
.._,.
,
,

. 7..
_
....
..
,
, '
.
. "
.
..
X
...
..._

-'_
-

;,
'f
-..

..
'
'
'
'
o
'
'

r:m
tw.
,.
__.

(J,
::
.,
,
,..
,
_,..
_,
,..
'4"'

'

">o+.
':oo o ., .... , \oo'.., o

O:
'ot-;-''- , .. t oil' _ , ,_ .,. ";,
: ' - " :o:,;: , 'YT ;. , ; \,- .' ,:.. ' ".J '

o o". "': ', .,.


o
N' ,..,
, - -

..,_.c. r.-,.
,.

'

l O"

the pressure

1 t(1ff

kg

.,.,,,.\- ' ,

..

...

..

..._

;."J> -

'

1.7 TEMPERAT UREANDTHE ZEROTH LAW


OF THERMODYNAMICS
Temperature is

mor e suhtlc property than pre s s u re Its ori gin is the so-called
.

zeroth Jaw of thermodynamics.* The zeroth law is based on

e xpe ri men ts

(as

are all physical laws) and is concerned with properties of systems in thermal

eq uili briu m .. that is, systems in eq u il i b rium connected by a diathermal wall.


The law states:
If tvo s_vstems are separate(r in ther1nal e(Jllilihriul1l vitlz
they

are i11 eqttilibrilll11

a tl1ird syste111.

-vith each fJTher.

Let systems A B. and


..

tainer. We shall use C

(_each consist of a 1nass of flutd in an insulated conas a reference. We choose to describe the state of each

.
*There are some \\'ho hc1icve that this should n)t be dignified by ca11ing it a . Jaw."
.

Temperature and t:he Zeroth Law ofThermodynamics

Sec. 1.7

(a)

II

(b)

Figure l.S Systems in therntal contact: (a) systems A and C in


equil ibritun ; (b) systems B and C in thermal equilibrium.

thermal

system in terms of the state variables P and V. We place A in contact with C


through a diathermal wall as in Figure 1.5(a).
The system C can be thought of as a thermometer. Its state is given by
the pair of variables ( , Yc). From observations, if we choose a particular
value for PA, then VA will be uniquely deter111i ned (only one variable is inde
pendent). The condition under which A and C are in equilibrium may be
expressed by the equation

(1.2)
where F1 is some function of the four variables. We assume that this equation
can be solved for Pc:

(1 .3)
Next we place system B and C in thetttlal contact (Figure 1.5(b)). For equilibrium,
(1. 4)
or

(1.5)
Equating Equations (1.3) and (1.5), \Ve obtain the condition under which A
and B are separately in equilibrium \vith C:

(1.6)
But, according to the zeroth l aw A and B are then in equilibrium with each
other, so that
,

(1.7)

The Nature ofThermodynamics

Chap. I

11

Solving for P_4 we obtain


..

( 1 .8)
Now, Equation ( 1 .6) can also be solved for PA in p ri ncipl e :

PA = g ( V4 , PB .. VH .. Vc ) .

( 1 .9)

as
ere
wh
les,
iab
var
r
fou
by
ed
1in
ern
P"
det
s
i
t
tha
t
es
t
a
s
.9)
1
(
on
i
at
u
q
E
Equ atio n ( 1 .8) says that it is a fun ctio n of t)nl y thre e. TI1i s can only mea n tha t
the functions f1 and f2 in Equ atio n ( 1 .6) cont ain Vc in such a form that it can
cels out on the two sides of the e qua tion . f<)f e x am p l e
,

!t

f2

<> 1 < PA ..

vA ) ' ( vc )

4J J ( p8 , VH ) ( ( .\1( )

11 ( vc ) ..

+ 7J ( VC )

When the cancellation is performed w e ha ve


..

( 1 . 1 0)

Extending the argument to additional systems, we get

For any system in thermal equili brium with a given system .. we can choose to
w ri t e

( 1. 12)
where we define T as the empirical

e qu a t i on

tetnperature and Equation ( 1 . 1 2 ) is the

of state of t he system. Equat ion ( 1 . 1 1 } th e n says that systetns in ther


m a l equilibrium with one a n o th e r h ave the s a me temperature. Thus tenzpera
ture is a property of a sys t em that determ ines if thermal equili brium e xi sts \\'ith
some other syst em .
The next section contains a scheme f,)r me asuri n g the temperature. Let
C be a thermometer, ..' an arbi trari l y sclecttd standard body. Then the relative
temperature of the systems A a n d B c a n bt compared without bringing them
into contact \Vith each other.

1 .8 TEMPERATURE SCALES

To assign a numerical value to the tcn1pcrature of a system. we choose as a ther


mometer a substance that has a so-called tJzernzometric propert.,v tha t changes
with temperature and is easily measured. A n example is the volume of a fluid
that expands on heating (think of the familiar liq ui d in glass thermometer).
-

Sec. 1 .8

Temperature Scales

13

We choose a thermometric property X that is linearly relate d to the tem

perature over as wide a range as possible:


X =

Here

and

Prior to

(1.13)

aT + b.

are constants. We choose reference points to llefine the scale.

1954, two

fixed points were used, the ice and steam points of water.

The scale was determined by assigning the numerical values ooc


Celsius) to the ice point and l00C to the steam point.

(C

for

An excellent choice for the thermometric property X is the pressure of a

gas. The constant-volume gas them1ometer, shown in Figure

1.6, is a practical

method for measuring the temperature of an object.


Several observations were made when this method was used (Fig
ure

1 .7):

a.

The P- Tcurve is very nearly linear over a wide range of temperature.

b. The curve is increasingly linear as the pressure decreases.


c.

A linear extrapolation of the plot gives

0 at

- 273.l5C. This

turns out to be true for all gases although the slope of the curve is differ
ent for different gases.

I
-

capillary

---.....--

....
:. ,.._,
..
.
. :.
.

"

.
.
.

. .

.. .. .
.
'
.
.. . . .
... .

.
.

rigid
_....,.__ diathermal
wall

. ..:...
.
: : : .
': :
:
.. ..
.

-. -. :: .
.
. ..

.
..

.
. ,'
0

. .. .. ..... ..
......
.. . .. .
. :: ..: .
: .
.. ... . . ! .
.:.
. .
.
0

. .

'

. . . . ,
. ..;:: . : .;
. . .. .

-: ....
.
. . ..
. .
'
; .,.:..:
.
. ...

.
. .
: .: :
.

..

. ..
'

.
. . .

_
_.....

_
_

,,

.
.

o ,

. . .
-:
..
. : ..

.
.

'
.
.
.
: :

'o

,o.

::
. ..: . :0.. ...

.. . :. .
..
.
..;_: ...
..: .. ..
,

0
0

. .

.
....
.

.
.

.
.
.
.
.... :
.
.
. . .>'
: .. : ..
.
..;...: .. ..
. .. '
.
..

.. . . . ... .
. ..: .
. .
::::

. .

. . ...
. :
. . -.- .
.
.
. .... . .
..0

.
.
" '
o

.
.
.
.

.
.
.

p
gas

. < : :
::
.. .. .
.. :. . , .
.. :
.
... ..

::
:_

.
.
:
:
:

.
;:
:. .
-.: . : .
. ,.: . ...
:
... .
. ... . . .
. .. .
.
' ... .
. ... . .
- .; ; - ::
..
:... ...>.'
...
. .: .
. .
: :-.-:... : .
.... .....
.
.. .:. ...'...'
.. ,
' .: ..
. .. .
:-.
-: :-.. . .
. . ......
.. .. ... . ..
:-..:
-=

.
.......... ... .. ,.
. . ... "' .
.
.
-: :
\
'::::

:
.
.
:
:
:
.
.
.
.
.
. :-:- ....
,

.
.

..-:. .. ........
. :: .. , : .
.. . .

... ...: .: . :

.. .. .

.
.
.
.
'
. : :.. '-:
. . . .. . .. .
. . . .. .......
..
..
.. . . . .
.:- :: :.

..:-.::::::
.
::::::... -:: : :.. .::::-::: --=:
:
. .., ,:.. . ... '' ... . ..... . . 0
. :- :
:: :?: . :_:-:: :
:
..

.
.

.
.
. ..
.
.
. .
.

. ..

.
.

'

mercury
manometer

..

.
.

.
.
.

.
.

<
.

. .
.

..
.

.
. . .
.
.

..

.
.

. .

.
.

.
.

,.o.

. .
.
.
.

. . .

:.:::
; : :

. .

Figure 1.6 Simplified constant-volume gas thermometer. The pressure is


given by P
hpg, as in Figure 1 .4.
=

Chap. I

14

The N ature ofThermodynamics

I
_
_
_
_
_
_
_
...._
__.
_
_
_...
.
__.
_
__
;...

Figure 1.7

Pressure versus temperature for a gas

ther1nometer. The ice and steam points of water are


fixed points. The temperature at zero pressure is
obtained by linear extrapolation.

For a gas thermometer. then.


P

==

aT

b.

The zero pressure point gives


()

273 . 1 5a

b.

Thus
P

af7 ( c'C )

We can now choose an a bsol ut e


ice and steam points. We want

sca l e such

==

273. 1 51.

( 1 . 1 4)

that there are 1 00C between the

aT ( K. )

( 1. 1 5 )

where K is t h e abbreviation for kelvin. the unit of temperature on the absolute


scale (no degree sign ). It follows that
( 1 . 1 6)
Now we need only one reference poi nt to dc.finc the slope a in Equ a t i on ( 1 . 1 5 ) .
In 1 954 the reference point was taken to be the triple point of water. the pres
sure and temperature at which ice and liq uid \Vater coexist in eq uilibrium
with saturated vapor. The tri ple point temJ>erature of water is 0.01 :)C, that is
0.01 degrees above the i cc point at atrilospheri<; pressure. Equations ( 1 . 15 )
and ( 1 . 1 6) give

Sec. 1 .8

15

Temperature Scales

T(K)
--

273.16

( 1 . 17)

Prp '

where Prp is the pressure at the triple point. A better definition, taking into
account the convergence of th.e P-T curves for gases at low temperature
(Figure 1.8), is

( 1 . 18)
The curves of the figure are constructed in the following way. An amount of
gas in the bulb of a constant-volume gas thermometer is exposed to a cell con
t ai nin g water at the trip le point. The pressure is recorded. With the volume
held constant, the bulb is surrounded with boiling water at atmospheric pres
sure, the gas pressure is measured, and the temperature at the steam point, T5 ,
is calculated usin g Equation ( 1 . 17). Then some of the gas is removed from the
bulb, the lower triple point pressure is measured, and the new values of the
pressure and temperature at the steam poi n t are found. The process is continued and the resulting curve is extrapolated to the ordinate. Although the
curves for different gases have different slopes, they are discovered to inter
sect in the limit as PT P approaches zero. The triple point reference is precisely
reproducible and absolute zero is precisely defined.

373.8

t-

373.6

;;;;
;;; ================= He2
H
37 3.0

...---..J..-----"--'--'"'L-.

20

Figure 1.8

40

60

80

100

120

Prp(torr)

Readings of a constant-volume gas therrnometer for the


temperature of condensing steam as the density of the gas (and hence PrP)
is lowered. The extrapolated value of T5 is the same for all gases.

Chap. I

16

u se.

Two other temperature scales are in

The Nature ofThermodynamics

The familiar Fahrenheit scale is

related to the Celsius scale by the equation

( 1 .19)
On this scale the ice point is 32F and the steam point is 2 1 2F. The Rankine
scale is derived from the Fahrenheit scale:
( 1 .20)
The Rankine scale is widely used in enginc.!ering: it does not use the ""degree''
symbol.

We shall see later that it is possible t o define the absolute scale of tem
perature independently of the thermometric properties of gases.

. ..

..

..

.. .

.
.

.
'

'

.
.

.
.

P RO B L E M S

1-1 Classify the follo\\'ing systems as ope n_ closed. or isolated:

(a) A mass of gas in a con tainer \Vith rigid. impermeable. di athe rm a l walls.
(b) A mass of gas in a container \Vith rigid, impermeable. adia b a t i c wa11s.

(c) A sugar solution enclosed hy a men1branc permeahJc only to water that is


immersed i n a large container of wa1cr.

12 U si ng the terms de fi n ed in the chapter. characterize the foJiowing processes as

completely as possible:
(a) The temp e ra t u re of a gas. enclosed i n a cy l i nd e r provided with a frictionless
piston . is slowly increased. The pressur remains constant .
(b) A gas, enclosed in a cyl inder provided with a piston, is slowly ex pande d . The
temperature remains constan t . ll1ert is a force of friction bet,veen the cylin
der wall and the piston .
(c) A gas enclosed in a cylinder provided \Vith a frictionless piston is quickly
compressed.
(d) A piece of hot metal is thro\vn into cold \\'ater. (Assume that the system is
the metal, which n e it h er contracts nor expands. )
(e) A pendulum with a frictionless su pport swings back and forth .
(f) A bullet is s topp ed by a targe t .
1-3 On a plot of volume versus temperat ure. dra\\' and label l ines indicating the fol
lowing processes.. each pr t)C c c d i n g from t he san1e initial state . ).

(a) An isothermal expansion.


(b) An isothermal compress i on.

(c) A n isochoric increase in temperature.

Problems

17

1-4 Estimate the pressure you exert when standing on the floor. Express your answer

in atmospheres, in pascals, and in torr. Repeat the calculation for spiked heels.

1-S Let the resistance R of a piece ot wire be a thermometric property for measur

ing the temperature T. Assume that

R = aT + b,

where a and b are constants. The resistance of the wire is found to be 5 ohms
when it is at the temperature of melting ice and 6 ohms when it is at the temper
ature at which water boils at atmospheric pressure . If the ice point is taken as
100 and the boiling point as 500 on a particular scale, what is the temperature
5.4 ohms?
on this scale when R
=

1-6 The following table gives the observed values of the pressure P of a gas in a con
stant-volume gas thermometer at an unknown temperature T and at the triple

point of water as the mass of the gas used is reduced.


P-n (torr)
P (torr)

1 00
1 27.9

200
256.5

300
385.R

400
5 1 6.0

Determine T in kelvins to two decimal places by considering the limit


PrP-+0 ( pI Pr p ) . What is this temperature in C?
1-7 The resistance R of a doped germanium crystal is related to the temperature T

through the equion

log R

4.70

3 .92 log T

when R is in ohms, T is in kelvins, and the logarithm is taken to the base 1 0. In a


liquid helium cryostat, the resistance is measured to be 218 ohms. What is the
temperature?
1-8 A thern1ocouple consists of two wires made of dissimilar metals that are joined
together to form an electrical circuit. A thermal electromotive force (emf) e is

generate(J when the two j unctions are at different temperatures. When one junc
tion is held at the ice point and the other is at a Celsius temperature of T, the
therntometric function is given by
2
E = aT + bT ,

where e is in millivolts. Calculate the constants a and b for a thertnocouple read


ing 60 m V at 200C and 40 mV at 40occ. What temperature corresponds to a
reading of 30 m V?
1-9

(a) Consider the linear relationship between the thermometric prop e rty
the temperature T given by

X and

X = aT + b.

Suppose that the ice and steam points are used as fixed points with temper
atures of oo and 100 respectively. Show that
T

100

X - XI

X.( _ XI .

Chap. I

18

(b )

The Nature ofThermodynamics

If, instead, the thertnometric function is chosen as

aln .Y

h,

show that on the new scale

1-10 At what temperature do the Fahrenheit and Kelvin scales give the same reading?
1-11 The temperature of the normal boiling point of nitrogen is

77.35

K. Calculate

the corresponding value of the temperature on the (a) Celsius. (b) Fahrenheit,
and (c) Rankine scales.

2. 1

Introduction

21

2.2

Equation of State of an Ideal Gas

22

2.3

Van Der Waals' Equation for a Real Gas

23

2.4

P-v-T Surfaces for Real Substances

25

2.5

Expansivity and Compressibility

27

2.6

An Application

29

19

2. 1 INTRODUCTION
In thermodynamics, we add to the basic dimensions of mass, length and time
the "fourth dimensio n" of temperature. We saw in Chapter 1 that the concept
of temperature is intimately associated with the. notion of systems in thermal
equilibrium. Perhaps the simplest example of a thermodynamic system is a
homogeneous fluid say, a gas or vapor.* Its state, as we have observed , can
be described by an equation of the fottn
f ( P, V , T)

0.

(2. 1)

The equation of state connects the three fundamental state variables, only two
of which are independent. It is an expression of the results of experiments.
Every system has its own equation of state.
We note that the equation of state does not involve time. Classical ther
modynamics deals mainly with systems in thertnal, mechanical, and chemical
equilibrium that is, thermodynamic equilibrium and does not concern itself
with the rate at which a process takes place.
The mechanical variables occur in "canonically conjugate" pairs, one an
extensive variable, the other intensive. Thus the volume V (extensive) is "con
jugate" to the pressure P (intensive). The situation is analogous to that in
mechanics, where the generalized position coordinates qk and the generalized
momenta Pk form canonically conjugate pairs.
In thertnodynamics, the quantity PdV is the differential of work (with
units of energy). We are led to ask if there is an extensive state variable canon
ically conjugate with the temperature whose product has the unit of energy.
This very important question will be addressed in Chapter 6.

* The laws of thennodynamics apply to all forms of matter, but they are most easily formu
lated and understood when applied to gases.

21

Chap. 2

22

Equations of State

1.2 EQUATION OF STATE OF AN IDEAL GAS


The equation of state of a system composecl of m kg of a gas whose molecular
weight is M is given approximately b):

(2.2 )
.

where R is a universal constant having the same value for a11 gases:

Since

( 2.3)

mj M is the number of kilomoles of the gas, we can write

PV

nRT.

(2.4)

This equation is called the equation of sta1e of an ideal gas or perfect gas. It
includes the laws of BoyJe't Gay-Lussac Charles.. and Avogadro .. which were
discovered over a period of 200 years.
In 1 8 1 1 Avogadro postulated that at a given temperature and pressure
equal volumes of all gases con tain e.quaJ numbers of molecules. The reason for
this is that the molecules in a gi ven sample of gas have negligible volumes
compared with the volume of the sample itself.
In Equation (2.4) we note that the extensive variable V divided by n the
number of kilomoles of the gas.. is the specific volume v. Thus the equation of
state can be writ ten Pv
RT. The projections of the surface f ( P, v, T) 0
on the P-v plane, the P- T plane and the v- T plane are sh ow n i n Figure 2. 1 . We
use whatever diagram is most appropriate for the process we are in terested in.
=

==

..

Vz

(a)
Figure l.l

::;

;.__--- -

(h)

...._

___.

_
_
_
_

(c)

Diagrams for an ide al gas. I n ( a ) the isothern1s are eq uilateral hyperbolae:


in ( b ) the isochores are straight l i nes: and in (c) t h e isobars are also straight lines.

Sec. 2.3

23

Van Der Waals' Equation for a Real Gas

2.3 VAN DERWAALS' EQUATION FOR A REAL GAS


The characteristic equation of an ideal gas represents the behavior of real
gases fairly well for high temperatures and low pressures. However, when the
temperature and pressure are such that the gas is near condensation, important deviations from the ideal gas law are observed.
Among the numerous equations of state that have been introduced to
represent the behavior of real gases, that of van der Waals is especially inter
esting because of its sim.plicity and because it satisfactorily describes the
behavior of many substances over a wide range of temperature and pressure.
Van der Waals derived his equation from considerations based on kinetic
theory, taking into account to a first approximation the size of a molecule and
the cohesive forces between molecules. His equation of state is
a

P + 2 ( v - b) -= RT,
v

(2 .5)

where a and b are characteristic constants for a given substance. For


b 0, Equation (2.5) reduces to the equation of state for an ideal gas.
a
The tettn aj v2 arises from the intertttolecular forces due to the overlap of
electron clouds. The constant b takes into accottnt the finite volume occupied
by the molecules; its effect is to subtract from the volume term.
2
Multiplication of Equation (2.5) by 1J yields the equation
=

Pv3 - ( Pb + RT) v2 + av - ab

= 0.

(2.6)

The result is a cubic equation in v with three roots, only one of which needs to
be real. In Figure 2.2 some isotherms calculated from the van der Waals equa
tion have been drawn.
constant. The correction
As T increases, the curves approach Pv
terms in the van der Waals equation become less important. For T < Tc , there
is a local maximum and minimum value of P. The transition between the two
types of curves is a curve having an inflection point CP, the so-called critical
curve, T = Tc . Imagine that the volume is decreased by loading a piston con
fining the gas in a cylinder. Suppose one moves along an isotherttl for which
T < Tc . After we reach a maximum of the cJrve, the pressure begins to fall.
This is an unstable region since the pressure no longer increases as the volume
is diminished.
Actually, this portion of the isotherm is not really traversed at all
because the gas undergoes a change of phase. * Assume that the compression
=

* The transition from a gas to a liquid, from


change" or a .. phase transformation."

liquid to

solid, etc. is termed a "phase

Chap. 2

14

Equations of State

d
Figure 2.2

Isotherms for a
van der Waals gas.

b ...... a

--- -----+

begi ns at p oi nt a i n the diagram. Part of the gas begins to liquefy at b and the
pressure remains constant as the volume is further decreased as long as the
temperature is held constant. Between b and d. liquid and vapor are in equi
librium. Finally, at d liquefaction is complete. Thereafter the curve rises steeply
because it takes a large increase of pressure to compress a liquid.
The curves with no extrema ( T > Tc ) can have no regions of this kind.
Above Tc it is impossible to liquefy a gas, n<, matter how large the pressure is.
The interface between a liquid and a vapor is not discernible.
The critical values Vc , Tc . and Pc of a st1bstance can be expressed in terms
of the constants a and b that appear in van der Waals' equation. For T Tc .
=

(2.7)
The po i n t at which the tangent to the curve is horizontal is given by

iJP
av

0
=- =

RTc
--

( Vc -- b ) --

2a
-

(2.8)

Vc ,

and the point of inflection is the point at which the rate of change
ts zero:

of the

slope

=
= 0

2RTc

3
( vc - b )

6a

4.
vc

(2.9)

Sec. 2..f

P-v-T Surfaces for Real Substances

15

Solving these equations yields


Vc

Tc

3b,
8a

21Rb '
(2. 10)

Evidently, Pc vc

RTc for a van der Waals gas at the critical point. The rela

tions of Equation (2.10) suggest that the equation of state of a van der Waals
gas can be written in terms of the critical variables, eliminating the constants a
and b. This is indeed the case (see Problem 2-5).

1.4 P-v-T SURFACES FOR REAL SU BSTANCES


Real substances can exist in the gas phase only at sufficiently high tempera
tures and pressures. At low temperatures and high pressures transitions occur
to the liquid phase and the solid phase. The P-v- T surface for a pure substance
includes these phases as well as the gas phase.
Figure 2.3 is a schematic diagram of the P-v-T surface for a substance
that contracts on freezing.
Notice the regions (solid, liquid, gas or vapor) in which the sttbstance can
exist in a single phase only. Elsewhere two phases can exist simultaneously in
equilibrium, and along the so-called triple line, all three phases can coexist .

._._,

,_

CP

.--..

Jiqui

CP
gas
vapor

Figure 2.3

freezing.

---

.-

---

._

--..,

triple line
_. ----

_
_
_
_
_

P-v- T surface for a substance that contracts on

Chap. 2

16

Equations of State

S-L

S-L
CP

liquid

CP

liquid

solid
solid
vapor

TP

vapor

TP
T

------ T

(a)

(b)

p. T diagrams for (a) a su bs t ance that contracts on freezing: and

Figure 2.4

(b) a substance th at e x p a nds on free zi n g.

The physical distinction between t he various phases is straightforward. A solid


has a definite volume and shape. A liquid has a fixed volume but not a fixed
shape. A gas has neither

a fixed volume nor shape. A vapor is a gas at any tem

perature less than the critical temperature.


The projections of the surfaces on the

P- T plane are of special in terest

( Figure 2.4). The L-V curve is the vapor pressure curve, for which the liquid
and vapor phases coexist in equilibrium. It is also known as the hsaturated
,
vapor, or boiling-point curve. The S-L curve is the freezing point curve. and
the

S-V curve is the sublimation curve.


At T

Tc the specific volumes of the saturated liquid and vapor

become equal. For T

>

Tc there is no separa tion into two phases; the interface

between the liquid and the vapor is indiscern i ble. By raising the temperature
a bove Tc then increasing the pressure. and finally reducing the temperature,

one can m ake an "end run" around the critical point. In this way, it is possible
to m ove from a vapor to a liquid without crossing the vapor pressure curve.
No critical point can exist for soJid-Iiquid equil ibrium (the S-L curve).
That is because solids and liquids possess diffe rent symmetry properties. A
normal liquid is isotropic, whereas a solid has a cryst alline structure whose ori
entation defines a particular set of directions. The transition from one sym metry to another is strictly a discontinuous process.
For C02 the critical temperature is

3 l .0C

.
and the critical pressure is

72 atmospheres. The triple point occurs at - 56.6C a nd about 5 atmospheres.


When heat is supplied to solid co (dry ice) at atmospheric pressure.. it sub

limes and changes directly to the vapor phase.

17

Expansivity and Compressibility

Sec. 2.5
v

I
I
I

I
I
I
I
I
t
I

lOOOC

l.S

Sketch of the specific volume of water as a


function of temperature.

For H 2 0 the critical point is 374.2C and 207 atmospheres. The triple
3
point is 0.01 oc and 4.6 torr (6.0 X 10- atm ) . Thus the pressure must be

reduced to a low value to reach the triple point.

Water and a few other substances expand on solidif}ing. Water expands


about 10 percent on freezing. A block of ice floats with nine-tenths of its vol
ume below the sutface of the 'Nater and one-tenth above. (An iceberg floats
with one-seventh of its volume above the surface; the difference is because sea
water is denser than fresh water and common ice is full of air b1Jbbles.) If

water contracted on freezing, the results would be disastrous, but ice floats.
During a change of state, there is always a change in volume but the tempera
ture and pressure are constant. For example.. at the freezing point of water, the

10 percent, as noted. But at the steam point the expansion


is enormous: 1 cm3 of liquid water at l00C becomes 1600 c.m3 of steam at this
change in volume is

temperature. Changes of state are accompanied by changes in molecular


forces. In Figure 2.5 the specific volume is sketched as a function of tetnpera

ture for water. Since v is the reciprocal of the density, it follows that the ma:<i
mum density of water occurs at 4C.

2.5 EXPANSIVITY AN D COMPRESSIB!LITY


Suppose that the equation of state of a given substance is written in the form
v

v ( T, P ) .

(2. 1 1 )

Chap. 2

18

Equations of State

Taking the differential, we obtain

dv

ilt'

tiT +

:.=

iiT

clv

\ ri P

tl P.

(2. 12)

(The subscript denotes the q ua ntit y held (:onstant . ) Two important measur
able quantities are defined in terms {)f the pa rti al derivatives. The expansivity.
or coefficient of volume e xp an sion is given by
.

f3

rlV \
.l .
aT ; P

1
=

t'

--

(2. 1 3)

This is the fractional change of volume resulting from a change in temperature,


at constant pressure. Simi l a rl y the isothermal compressibility is defined as
,

iJv

(2.14)

i1 P .,.

the fractional change in volume as the pressure change with the temperature
held constant. The negative sign is used since the volume always decreases wi t h
increasing pressure (at consta n t tempera t ure ): that is. the partial derivative is

inherently negative, an d the isothermal compressibility is a positive q uantity.


For()an ideal gas, v = RTI P. and
1

f3 = p = r

(2. 1 5 )

The higher the temperature the I<)WCr the expansivity. AJso,


K

--

l'

. RT \
-

,
P_

(2. 1 6 )

The isothern1al compressi b i l i ty of an ideal gas is small at high pressures.


.
For a l iqu id or a solid .. in contrast to a gas {3 and K are nearly constant over
a fairly wi de range of temperature and pressure. This experimental fact allows
us to develop an approximate equation of st ate that is useful in many applica
tions. Substituting Equation (2. 1 3 ) and ( 2. 1 4 ) in Equation (2. 1 2 ), we obtain
tl v

=-

{3vd7'

( 2. 1 7 )

I<" Vd P .

We assume t h a t the volume change is sm a i J when the temperature and pres


sure - are changed so that to a first approximation, v v0 (a constant) and f3
and K are constants. Then
..

..

Sec. 2.6

An Application

29

Integrating, we have
v

dv

{3v0

T
To

dT

KV0

dP,

so that

v0 [1 + {3 ( T

T0 )

K ( P - P0 ) ].

(2.18)

This is an approximate equation of state for a liquid or a solid. The volume


increases linearly with an increase in temperature and decreases linearly with
an tncrease tn pressure.

Suppose that the temperature of a block of copper is increased from


127C to 1 37C. We wish to know what change in pressure would be necessary
to keep the volume constant. For copper in this temperature range,

Then, Equation

(2.18) gives
10-5
T =
(
10)
2
K
10-1
7.6 X

{3
=

5.2

6.8

107 Pa

680 atm !

2.6 AN APPLICATION
Suppose that we wish to calculate the decrease in pressure of a fluid when it is

cooled from T1 to T2 We know that the equilibrium states of the fluid are
fixed by specifying two of the state variables that are related by some equa
tion of state

f ( P, V, T)

0.

(2 .19)

We assume that the process in which the fluid is changed from an equilibrium
state ( P1 , T1 ) to another equilibrium state ( P2 , T2 ) is isochoric that is, the

volume is unchanged. We suppose for a moment that the fluid . is cooled


reversibly. This could be achieved by placing in contact with the fluid a series

of large bodies (reservoirs) ranging in temperature from T1 to T2 to effect a


quasistatic cooling through a sequence of equilibrium states. For any one of
these states we may write Equation (2.1 9), or, solving for P,

P (v T ) .

(2.20)

Chap. 2

30

Equations of State

Taking the differential , w e ob ta in

aP
a t'

c/P =

dv

iJP

aT

(2.2 1 )

dT.
t'

Since we are assuming that the cool i ng takes pl ace at constant volume, the first
term on t he right-hand side is zero. Thus
,.

r,

p,

pl

iJ P
,..

ar

T.

'
1.

(2.22)

dT.

Unfo rtuna tely.. the in te gran d is unknown. However.. using the cyclical relation

given in Appendix A, we have

aP
aT

--

aP 1 r

,.

dl'

Th us

riP

riT

ar

av

iJ V

a,v

aP

---

aT

riP

--

Substituting t his result in Equat ion (2.22) we obtain


P,-

PI

{3
=

( T*\-

Tl

( 2. 2 3 )

if {3 and K are i ndependent of T This is negative i f T., < T1 indicating that the
pressure is reduced on cooling. Note that this result could have been obtai ned
immediately from Equation (2. 1 8).
In practice, the cooling would not be reversible because one simply heats
the fluid and lets it cool. The result wou ld be large temperat ure gradients within
the fluid itself and between the fl uid and its surroundings. Thus the intermedi
ate states are no t equilibrium states and the equation of state cannot be applied.
However_ the initial and fi nal states ar equilibrium states, and i t doesn't
matter how we go from state 1 to state 2 to determi ne the change in the state
P2
P, i s independent of the path, bei n g
variable P. This is because a P
deterr11i ned only by the end points. We can choose any convenient path to cal
culate changes in state functions for processes between a pair of eq ui li briu m
states. The most con ve nie nt path is a reversible path.
..

Problems

ll

Ou r fundamental state variables P, v, and T are all exact differe ntia ls


whose integrals are independent of the path . Fotltlulating all thertnodynamic
relations in terttts of state variables will be a goal of the theoretical develop

ment; this is called the condition ofintegrability. When we encounter a thertno


dynanlic quantity whose differential is imperfect, we will seek an integrating
factor to give us a quantity that we prefer to work with. Appendix A includes a
comprehensive discussio1 of exact and inexact differen tials.

P RO B L E M S

2-1 How many kilograms of helium gas are contained in a vessel of I liter volume at

50C if the pressure is one atmosphere? (The atomic weight of helium is 4.)

2-2 A tank of volume 0.5

m3 contains oxygen at a pressure of 1.5

106 Pa and a

temperature of 20C. Assume that oxygen behaves like an ideal gas.


(a) How many kilomoles of oxygen are in the tank?
(b) How many kilograms?
(c) Find the pressure if the temperature is increased to 500C.
(d) At a temperature of 20C, how many kilomoles can be withdrawn from the
tank before the pressure falls to 10 percent of the original pressure?

23 A cylinder provided with a movable piston contains an ideal gas at a pressure


P1 , specific volume v1 , and temperature T1 The pressure and volume are simul

taneously increased so that at every instant P and v are related by the equation
P

A = constant

= Av,

(a) Express A in terms of the pressure P1 , the temperature T1 , and the gas con

stant R.
(b) Make a sketch representing the process in the P-v plane.
(c) Find the temperature when the specific volume has doubled, if T1

200 K.

2-4 For an ideal gas the slope of an isotherrn is given by

iJP
av

and that of an isochore is


aP
aT

=
t

'

Show that these relations give P v = RT, the equation of state.

2-S Noting that at the critical point, the three roots of the van der Waals equation .

are equal (i.e., ( v


Vc )3 = 0), show that the critical values of the specific vol
ume, temperature, and pressure are given by Equation (2. 10). Show that, in
-

Chap. 2

ll
v'

terms of the reduced quantities

vfv" , T'

T/Tc

and

Equations of State

P'

P/Pc .

the van

der Waals equation becomes

P'

-r .

3-

,
..

I
-

2-6 Usi ng the Berthelot equation of state


P

RT

1J

- -

.,

--

Tv-

show that
IJ(

)b.

Tc =

(Hint: As noted in the text iJ PI rl l '

Compare the numerical value of


in the fo llowi ng table:

Ba

'J 2 1bR

P
c

fZtiR
T2b 3b
t

..

0 and il PI riv2

RTc_ - / Pc vc. with

0 at the critical point.)

the experimental values given

Substance
He
H,
Q.,
co,
H ., O
-

..

3.06
J.27
:\.42
J.6)
4.29

2-7 Using the Dieterici equation of state.


p

RT

v - b

e a/ RTr

show that
a

Pr = - 2 2 "
4e b
and find the numerical value of RTc
u la t ed experimenta] values"?

2-8

(a)

How docs this com p a re with the tab

Making use of the cyc l ica l re lation ( Eq u a t io n ( A .7) in Appendix A ) . fi nd

the expansivity f3 of

(b)

/ Pc Vc .

substance obeying the Dieterici e q uation of state i n

Problem 2-7.

At h ighe r temperatures and l arge specific vol utnes ( low densities) all gases
a pproxi ma t e an ideal gas. Show that for large values of T and v. the e x pres s ion
for {3 obtained in ( a ) goes over to the corresponding equation for an ideal gas.

2-9 Show that in general

ap
-

aP

dK

-aT

0.

Problems

]3

1-10 A hypothetical substance has an expansivity /3


2bTfv and an isotbertnal com
pressibility K = afvt where a and b are constants. Show that the equation of
=

state is

bT2 +

aP

constant.

Z-11 Suppose that

f3

v-a
Tv

K =
'

3 ( v - a)
4Pv

Show that the equation of state is

4
3'
P (v - a)
where

AT

a and A are constants.

2 12 Show that and K are infinite at the critical point.


2-13 A glass bottle of nominal capacity 250 cm3 is filled brim full of water at 2QOC. H the
bottle and contents are heated to SOOC, how much water spills over? (For water,

f3 ::: 0.21

10- 3 K - 1 . Assume that the expansion of the glass is negligible.)

3. 1

Configuration Work

3.2

Dissipative Work

37

40

.
.

'

..

3 .3

Adiabatic Work and Internal Enetgy

3.4

Heat

43

3.5

Units of Heat

45

3.6

The Mechanical Equivalent of Heat

46

3.7

Summary of the First Law

46

3.8

Some Calculations ofWork

47

41

35

3.1 C ONFIGURATION WORK

Configuration work is the work in a reversible process given by the product of


some intensive variable (P, for example) and the change in some extensive
variable (V, for example). Let y represent such an intensive variable and X the
corresponding extensive variable. In the most general case, where more than
one pair of variables may be involved,
llW

y; d X; , i

1, 2, . . n.

(3.1)

'

Here ltW represents the work done by virtue of infinitesimal changes in the
extensive variables X;; 71W is not an exact differential, as indicated by the
notation (see Appendix A). The variables X1, X2, etc. are said to determine
the configuration of the system, and

'

is called configuration work.


Consider the configuration work done by the system consisting of a gas
enclosed in a cylindrical container by a piston (Figure 3.1). The gas expands
against the pressure associated with the external force F (possibly due to the
weight of the movable part of the piston).lne work done by the gas is
aW

F dx

PA d.t

P dV.

Here dx is the displacement of the piston, A is its cross-sectional area, P is the


pressure (the intensive variable), and V is the volume (the extensive variable).
Since P and dV are positive, ltW > 0, denoting work done by the sytem. This
sign convention will be used throughout the text. We assume that the process
is reversible and that the system is in equilibrium at all times. No friction (dis
sipation) is involved. The integration is nontrivial since P P(V) in general.
=

37

The First Law ofThermodynamics

Chap 3

38

F=

dx

PA

----

--

----

'

final

level

--

original level

--: -------

Fipre 3.1

Expansio.n
of an enclosed gas
against the pressure
produced by an
external force.

TABLE 3.1

Various examples of configuration work.

Svs tem

Intensive Variable

gas

..

Extensive Variab1e

V (volume)
A (area)
q ( c har ge )
M ( manctization)
P (polarization)

P (pressure)
r (surface te n sio n )
(electromotive force)
8 ( ma g n et i c fi e l d )
E (electric field).

liquid. or solid

film
electrolvtic cell

magnetic material
dielectric material

...

PdV
I'dA
Edq
Bt/M
Ed'P

In thermodynamics configuration work is often called PdV work for


obvious reasons. However. there are many examples of configuration work,
....

'

some of which are listed in Table 3.1. For a surface film of liquid._ the reversible
work required to increase isothern1ally the surface area of the film by an

amount

dA

is aW

fd A._ where

is the surface tension. In an electrolvtic


cell such as a lead storage battery._ the configuration work is the product of the
=

electromotive force

of the cell and the amount of charge transferred in the

chemical reaction . Magnetic work is done <)n a magnetizable material by a


magnetic field.

I( for

example the material is placed in a solenoid that gener

ates a field B, the incremental work done is

lldM, where dM is the change in

total magnetization of the material. Analogously. an electrically susceptible


material inserted between the plates of a parallel-plate capacitor experiences
a change of total electrical polarization

tiP: the

work is

EdP,

where E is the

electric field strength in the region between the plates.


The configuration work done in the change of volume of an ideal gas can

be easily calculated in some special cases. For an isochoric process.. dV


aW

0. so

0. For an isobaric process,


\ ..

PdV

liV
I
I

=:

P(Vh- Vn).

(3.2)

Sec. 3.1

Configuration Work

39

----

3.2 Work done


Fi
in an isobaric process.

Here Va is the initial volume and Vb the final volume. The work done is evidently the area under the curve in a P-V diagram. lbis is the shaded rectangle
of Figure 3.2 for the case of an isobaric pr oce ss
For an iscthert11al process, the temperature is constant, the P-V curve for
an ideal gas is a hyperbola, and
.

nKT

v, dV
Vo

V,

nRT In V.

(3.3)

(Figure 3.3).
In both the isobaric and the isothermal process, if Vb > Va, the process is
an expansion, and work is done by the system (W > 0). If Vb < V,, the process
is a compression, and work is done on the system (W < 0). This important
sign convention will be followed throughout the text.

Figure 3.3

Work done
in an isothertnal
process involving an
ideal gas.

Chap. 3

40

The First Law ofThermodynamics

p
A

Figure 3.4 Work is


path-dependent; the
area under curve A is
differe nt from the area
under curve B.

I
'

I
I

I
I

VII

It must be emphasized that work is not


a state variable. This is easily seen in Figure

property of the system: W is not

3.4. Since f P dV is the area under

the curve, different results are obtained for paths A and B. Thus
a.

aWis not exact.

b. W depends on the path of integration.


c.

The integral around a closed path is not zero: that is,

:: : :::::::*;:: :?.::::=:::-:-:::: :=:::s: :/:: : :: . :::

aW

* 0.

):::::::=->.::-=.:: :;:

. '......
... . ....:
. ....
. . ...
.. ....
. .
. ..
... . . ...
.. ......
, . ..
.
..... ...
. .

....
...
.. ...
. . . ......... . .:.. . .
.
. ..: =:.::.....
'I'' ' :
.: .::.::-:::
..
: :
:: : :: : :
:.::::; : :::: :,:: . : :.
:: :::::. ::::::=:=:;::

:(:
:
:::.
:: .::.::: ::.:.
:
: :: ::=
:
:.. .:.::
:,
::
=:{"- ::
<

:::x::: :: :::
:: ::": .
.

....

...

3.2 DISSIPATIVE WORK


Dissipative work is work done in an irreversible process. Such work is always
done on the system. In the general case both configuration 'W'ork and dissipa
tive work may be done in a process. The total work is the algebraic sum of the
two kinds of work. If a process is to be reversible. the dissipative work must be

zero. Since a reversible process is necessarily quasistatic then to specify that a


process is reversible implies that the process is quasistatic.
There are two well-known examples of dissipative work. The first is stir
ring work. Consider a stirrer immersed in

fluid. The stirrer and the fluid

together constitute the system. The stirrer is attached to a shaft projecting


through the wall of the container and an external torque is applied at the
outer end of the shaft (Figure

3.5).

Regardless of the sense of rotation of Lhe shaft, the external torque is


always in the same direction as the angular displacement of the shaft, and the
work done by the external torque is always negative (work is always done on

the system). The work is given

by
aw

=:
-

Td8.

(3.4)

Sec. 3.3

Adiabatic Work and Internal Energy

.
..

41

. .

,.. .'
-.; ;:: ....... ...
. .
.. . .... .
.... :::
.._.,'-"'-""
.

. . .. .
..

./'..-/'""'"""'
.

......
:. ......
.

..:.>

. . ... .
.. . . . '
. '. ..
.

..

..

.
..

Figure 3.5 Stirring


work. A stirrer is
inunersed in a fluid and
an external torque is

.
.

.
.

:.

.:. ...

..

.
...., ..
. ..
. :

.
.

::

...
.
..
.

.
.
.

...
,.

. ..
..
.
.<.
.

. .

applied.
I

I
R
Figure 3.6 Electrical
work. A current is
passed through a

resistor.

Clearly, reversing the sense of the torque

reverses the direction of the angu

lar displacement d8.


A second example of dissipative work is the work needed to maintain an

electric current I in a resistor of resistance R (Figure 3.6). Quantitatively,

(3.5)
Here ttW is the work done in time dt on the system comprising the current and
the resistor. Reversing the direction of current flow does not affect the sign of
the work.

3.3 ADIABATIC WORK AND IN TERNAL ENERGY

A fundamental result of the greatest importance is arrived at by considering


adiabatic processes between two equilibrium states of a system. Having
defined configuration work and dissipative work, we can think of different
adiabatic paths between two states a and b of the system. In the P- V diagram
of Figure 3.7, the path labeled 1, denoted by the solid line, could represent stir
ring work (at constant volume), followed by an adiabatic compression of the
fluid. (The system is assumed to be thermally insulated from its surroundings.)

Chap. 3

42

The First Law ofThermodynamics

Figure 3. 7

Three
different adiabatic
paths representing
processes in which an
isolated system is
carried from state a to
state b.

-------+

Correspondingly, path

portrayed by the dashed line, could depict an

adiabatic compression followed by electrical work at constant volume. T he


dotted line of path 3 could delineate somt combination of mechanical and
electrical work performed simultaneously.
Now, it is an experimental fact that the work done in the three processes is
the same. To be sure, experiments have not been performed for all possible adia
batic processes between al1 equilibrium states of all possible systems. However,
the whole edifice of thermodynamics is consistent with the conclusion that
The total work done i11 all adiabatic processes be/ween any tvo
equilibriLlnz states is i1z dependen t

of the path.

"

tJ

lf"''ad

is path-independent and

lfWad is an exact differential.

(We can remove the bar

from the differentiaL but only if we restrict ourselves to adiabatic work.) It fol
lows that we can define a property of the system, a state variable U, such that

ll

,,

In differential form,
(3.6)
We note that dU is the negative of the adiabatic work done bj' the system, so it
is the adiabatic work done on the system. We define U as the internal energy of
the system. Thus Equation (3.6) states that the increase in internal energy is

Sec. 3.4

43

Heat

equal to the work done on the system in any adiabatic process. Alternatively,
the work done on the system (with no heat flow) results in an increase in its
internal energy.
To summarize, experiments show that an energy function of state can be
defined by the amount of work done in an adiabatically enclosed system. This

is not to say, however, that any two states of a system can be coupled by an adi
abatic path.

3.4 HEAT

We wish now to extend our ideas to processes that are not adiabatic. If the sys
tem is not contai11ed within adiabatic walls, it is nevertheless possible to bring
about a given change of state in quite different ways. "-\ beaker of water may
be brought from 20C to l00C by electrical work perforrned on a resistor
immersed in it, or, alternatively, by lighting a Bunsen burner under it. The lat
ter process inyolves no work at all (Figure 3.8).
For any change between given states, d U can always be uniquely deter
mined by carrying out an experiment under adiabatic conditions, for which
dU = -llWad If the conditions are not so specialized, then dU =1: - t1Wad' in
general. Instead we may write the equation
dU

7/Q- aW,

or

t!Q

ltW - 7/Wad.

'

(a)
Figure 3.8

(b)

Two ways of raising the temperature of water:


(a) by electrical work; (b) by applying heat.

Chap. 3

44

The First Law ofThermodynamics

In other words, we can define heat flow as follows:


Heat flow into the system is equal to the total work done by the system
minus the adiabatic work done.

Then
aQ

==

dU + dW,

(3.7)

which is the first law of therr11odynamics in differential form. In words, the first
law states that
The heat supplied is equal to the increase in intertzal energy of the system
plus the work done by the system. Energy is conserved if heat is taken
into account.

Note that heat is not a property (state variable) of the system; only the inter
nal energy is.
It can be shown that the quantity liQ exhibits the properties that are
commonly associated with heat. These properties are summarized as follows:
1.

The addition of heat to a body change its state.

2.

Heat may be conveyed from one body to another by conduction, convec


tion_ or radiation.

3.

In a calorimetric experiment by the tnethod of mixtures, heat is con


served if the experimental bodies are adiabatically enclosed.
We note that since U is a state variable

dU

0.

That is, in a cyclic process, the change in internal energy is zero so that the
work performed is equal to the heat absorbed by the system.
One useful way of looking at the first law follows from transposing the
terms in Equation (3.7):

dU

71Q- aW.

The increase in internal energy of a system equals the heat flow into the
system minus the work done by the system. Think of a bank: aQis a deposit
itW a withdrawal. What's left in the bank is d U. The sign convention for aQ
is 7/Q > 0 for heat flow into a system, and 71Q < 0 for heat flow Ollt of the
system.

Sec. 3.5

Units of Heat

45

The temperature of a body alone is what determines whether heat will


be transferred from it to another body with which it is in contact, or vice versa.

A large btock of ice at ooc has far more internal energy than a cup of hot water;

yet when the water is poured on the ice some of the ice melts and the water
becomes cooler, which si

tes that energy has passed from the water to the ice.

When the temperature of a body increases, it is customary to say that


heat has been added to it; when the temperature decreases it is customary to
say that heat l1as been removed form it. When no work is done, U

Q,

which says that the internal energy change of the body is equal to the heat
transferred to it from the s11rroundings. One definition of heat is:
Heat is energy transferred across the boundary of a system
temperature difference only.

as

a result of a

This definition is not entirely satisfactory, however. Changes of state (e.g., from
ice to water or from water to steam) involve the transfer of heat to or from a
body witlzout any change in temperature. It should also be kept in 111ind that heat
transfer is not the only way to change the temperature of matter: a body that has
work done on it may become hotter as a result, and a body that does work on

something else may become cooler. A more precise definition of heat is:

Heat is the change in internal energy of a S)'Stem when no work is


done on or by the system.

3.5 UNITS OF HEAT

Since heat is a form of energy, the correct SI unit of heat is the joule (J).
However, the kilocalorie is widely used. It is defined as follows:
1 kilocalorie (kcal)

the heat required to r&ise the temperature

1 kg of \Vater from

of

14.5C to 15.5C.

The relationship between the joule and tlte kilocalorie is:


1J

4
2.39 X 10- kcal,

1 kcal

4184 J.

The calorie that dieticians use is the same as the kilocalorie and is some
times written "Calorie" to distinguish it from the ordinary, smaller calorie
associated with the cgs S)'Stem of units.

Chap. 3

46
. .
:.:.:_:.

.
.

:.

. ..

..

. .

.
. . ..

The First Law ofThermodynamics


.

'

.
.

'

. .

'

3.6 THE MECHANICAL EQUIVALEN T OF HEAT

We have noted that if the configuration \York and the dissipative work are
both zero.

or the heat flow in increases the internal energy. Suppose. on the other hand.
that the configuration work is zero and that work is done with a stirrer
immersed in a fluid kept at constant volume and thermally insulated. Then

where Wd is the dissipative work done on the system and is inherently negative.
The right-hand sides of the equations in the two examples give rise to the same
change in internal energy. Thus work and heat are .. equivalent'' in this sense.
However, heat and work can be differentiated from a microscopic view
point. When energy is added to a system in the form of heat, the random motion
of the constituent molecules is increased. Consider gas in a cylinder. If heat is
added through diathe1tt1al walls, this increases the random kinetic energy of the
molecules which means a rise irt temperature. Now let the gas be confined in a
cylindrical piston with adiabatic walls. When the piston is pushed in the mole
cules striking the piston are accelerated. in the direction of its travel. However.
any organized motion initially imparted to these m<>lecules is rapidly random
ized by collisions, either with the walls or with other gas molecules. Again the
increase in random kinetic energy appears as an increase in temperature.

3.7 SUMMARY OF THE FIRST LAW


1.

Energy is conserved. Heat is energy transferred to a system causing a


change in its internal energy minus any work done in the process.

2.

The quantity U is a generalized store of energy possessed by a thermody


namic system which can be changed by adding or subtracting energy in
any form.

3.

The internal energy U is a state variable: it is e x te nsiv e

4. The first law can be expressed in differtntial form as


aQ

dV + aw.

5. For a reversible process, aW is so lely


71Q

==

dV +

con figura tj on (lo PdV') work, so that

PdV.

The First Law ofThermodynamics

Chap. 3

48

If the expansion is isobaric d P


= nR

( T,,

0 and

T,,)

P ( V,,

V.,) .

Again, work is done by the system.

Equation (3.8) is especia lly useful for deter m in ing the work done

compressing a solid. Suppose that

10 g block of copper is isothermally (and

quasistaticaJJy) compressed at a t emperature of 0C. If the initial pressure is


1 atm and the final pressure is

1000 atm. how much work is done on the block

of copper? We assume that whereas the pressure change is large. the volume
cha n ge is small,

V,,

=::;

V.,

V. Then
aw

KPVtiP

and
,

'

PdP

KV
=

-.

. I' .

For copper. K

7.5

lO

( P,,- - Po-) .

(3.10)

...,

Pa- 1 at OOC and the density pis 8.93 g( em)-'. Thus


10 g

Since Pa2 << P,?, and

12

__

1.01 x 108 Pa we find that

a comparatively small amount of work.


At this stage of our study we cannot say much about what happens to
the heat and the internal energy in these processes. We need some more basic
concepts, which we shaH introduce in subsequent chapters.

PROBLEMS

3-1 Ten kilomoles of an ideal gas are cotnpressed isothermally and reversibly from a
pressure of 1 atmosphere to 10 atmospheres at 300 K. How much work is done?
3-2 S t e am at a constant pressure of 30 atn1ospheres is adn1itted to the cylinder of a
steam engine. The length of the stroke is 0.5 m and the diameter of the cvlinder
is 0.4 m. How much work is done by the steam per str oke?
.,

Problems

49

3-3 An ideal gas originally at a temperature

and pressure P1 is compressed

7;

reversibly against a piston to a volume equal to one-half its originai volume. The
temperature of the gas is varied during the compression so that at each instant
the relation P

(a)
(b)
(c)

A V is satisfied, A

constant.

Draw a diagram of the process in the P- V plane.


Find the final temperature T2 in ternts of T1
Find the \York done on the gas in terms of n, R. and T1

3-4 Ice at ooc and at a pressure of 1 atmosphere has a density of 916.23 kg m-:l,
while water under these conditions has a density of 999.84 kg m-:t. How much
work is done against the atmosphere when 10 kg of ice melts into water?

3-5

(a)

Derive the general expression for the work per kilomole of a van der Waals
gas in expanding reversibly and at a constant temperature T from a specific
volume v1 to a specific volume v2

(b)

Find the work done when 2 kilomoles of steam expand from a volume of
30m

volume

of

60 m-'

at

temperature

of

l00C. Take

3.19 x 10-2 m3(kilomole) -t.


lQ-" Jm3(kilornole) -2 and b
Find the work of an ideal gas in the same expansion.
a =

(c)

to

5.80

==

3-6 The temperature of an ideal gas at an in.itial pressure P1 and volume

Vj

is

increased at constant volume until the pressure is doubled. The gas is then
expanded isother1nally until the pressure drops to its original value, where it is
compressed at constant pressure until the volume returns to its initial value.

(a)
(b)

Sketch these processes in the P- V plane and in the P- T plane.


Compute the work in each process and the net work done in the cycle if
n

2 kilomoles, P1

2 atm, and V1

4 m.

37 Two kilomoles of a monatomic ideal gas are at a temperature of 300 K. The gas
expands reversibly and isothermally to twice its original volume. Calculate (a)
the work done by the gas; (b) the heat supplied. (Note that for an ideal gas UrxT,
so U

0 in an isothermal process.)

3-8 A gas is contained in a cylinder fitted with a frictionless piston and is taken from
the state a to the state b along the path acb shown in figure 3.9. 80J of heat flow
into the system and the system does 30J of work.

Figure 3.9 P- V
diagram for
Problem 38.

d
v

The First law ofThermodynamics

Chap. 3

50

(a) How much heat flows into the systetn along the path adb if the work done
by the gas is 10 J?

(b) When the system is returned from b to

along the curved path, the work

done on the svstem is 20 J. What is the heat transfer?


.,

(c) If Ua

0 and Ud

40 J, find the heat absorbed in the processes ad and db.

3-9 A volume of 10 m3 contains 8 kg of oxygen at

temperature of 300 K. Find the

work necessary to decrease the volume to 5 m)

(a)
(b)
(c)
(d)
(e)

At constant pressure.

At constant temperature.
What is the temperature at the end of the process in (a)?
What is the pressure at the end of the process in (b)?
Sketch both processes in the p V plane.
..

3-10 The pressure on 100 g of nickel is increased quasistaticalJy and isothermally


from zero pressure to 500 atm. Calculate the work done on the material, assum
ing that the density and isothermal compressibility remain constant at the val
ues of8.90

103kgm-3and6.75

10-11Pa-1,respectively.

3-ll During the ascent of a meteorological balloon its volume increases from 1 m3 to
1.8 m3, and the pressure of the ideal gas inside the balloon decreases from 1 bar
to

bar-(1 bar= 105 Pa). The internal energy of the gas is U =BOOT joules,

where Tis in kelvins.

(a)
(b)
(c)
(d)

Assume that in this process (the ascent) V

AP

B. Find A and B.

If the initial temperature is 300 K what is the final temperature?


..

How much work is done by the gas in the baJJoon?


How much heat does it absorb?

'

4.1

Heat Capacity

53

4.2

Mayer's Equation

54

4.3

Enthalpy and Heats ofTransformation

57

4.4

Relationships Involving Enthalpy

59

4.5

Comparison of u and h

61

4.6

Work Done in an Adiabatic Process

61

51

4.1 HEAT CAPACITY

Q is

Consider a process in wrhich heat

added to a system, causing it to change

from one equilibrium state to another with an accompanying rise in tempera


ture !l T. The heat capacity C of a system is defined .as the limiting ratio of the
heat absorbed divided by the temperature increase:

lim

r ....o

Q
T

71Q
.
dT

The heat capacity is not truly a derivative beca.use

aQ

(4.1)

is not an exact differential. We interpret

Q is not a state variable and

'llQ

as an infinitesimally small

flow of heat and dT as the resulting change in temperature. Furtherttlore,


"heat capacity" is a misnomer because it implies th at

system is capable of

holding heat, which it is not; a term connoting the storage of internal energy
would be more appropriate.
The heat required to produce a given temperature change is propor
tional to the mass of the system. We define the specific heat capacity, often
abbreviated to "specific heat," as the heat capacity per unit mass:

c=

1
n

tJQ
dT

(4. 2)

The SI units are J kilomole -l K -t.


The specific heat capacity depends on the conditions under which the
heating takes place

that is, on what parameters are held constant during the

heating process. There are two important cases:

(1)

the specific heat c11, where

the heat is supplied at constant volume; and (2) the specific heat

cp,

in which

the heat is added at constant pressure. Thus


llq

dT

and

cp

(4.3)

53

Chap.4

54

Applications of the First Law

c,

.l kilomole-1

K -J

400

200

800

600

1 000

1200

7"( K)
Plots of c,. and cp for copper versus temperature at a
pressure of one atmosphere. (Adapted from Thermodynamics.. Kinetic

Figure 4.1

Theorv, and Staristica/ Therntt>dvnanlics 3rd edition by F.W. Sears and


..

..

G.L. Salinger.)

The distinction between the two heat capacities is important chiefly for gases
because of the large coefficients of thermal expansion.
In general. the specific heats are functions of temperature and can be
regarded as constants only over limited ranges of temperature. As an example.,
consider the specific heats of a solid such as copper shown in Figure
low temperatures,

Cp

::=

cr

and both ten(l toward zero as the temperature

approaches zero. For high temperatures,

25

10 J kilomole- 1 K

4.1. At

c,.

approaches the constant value of

Many solids exhibit this behavior an en1pirically

established fact known as the law of Dulong and Petit. to be discussed in


Chapter 16.

4.2 MAYER'S EQUATION


We wish to find the relationship betweert

c,.

and

cp

for an idea) gas. For a

reversible process. the first law is


dU
or in terms of intensive variables

..

tiLt
Since

7/Q- Pd..

aq

(4.4)

Pdt'.

is a function of the state variables. \Ve can write. say.

(4.5)

Sec. 4.2

Mayer's Equation

55

Here we assume that the equation of state is

Pv

RT.

(4.6)

From Equation (4.5),

du

au
av

v
d

au
aT

dT.

(4.7)

Combining Equations (4.4) and (4.7), we have

tlq

au
aT

dT

(4.8)

To obtain cv, we divide this equation by dT and hold the volume constant so
that dv =0. The result, which holds for any reversible process, is
tlq

au
aT

dT

(4.9)

We shall see that for an ideal gas, u


u(T) only. This follows from the
Gay-Lussac-Joule experiment (see Chapter 5). Thus
=

au
av

=0

(4.10)

for an ideal gas. The substitution of Equations (4.9) and (4.10) in Equation
( 4.8) gives

tlq

cvtJT + Pdv.

(4.11)

Differentiating Equation (4.6), we have

Pdv

vdP

RdT.

(4.12)

Using this in Equation (4.11) leads to

tlq

(ct.

R)dT - vdP.

Now

Cp=

tlq

dT

p'

(4.13)

56

dP

Applications of the First Law

Chap.4

O. Then
llq
=

dT

c,. +

c,. -+

R.

c1'1.,

or
Cp

(4.14)

This relation is known as Mayer's equation.* It states that over the range of

variables for which the ideal gas law holds, the two specific heat capacities differ by the constant R.
We introduce the ratilJ of specific heat capacities
Cp

y=

c,.

(4.15)

For a monatomic gas at room tempera t u re

For a diatomic gas at room temper at ur e

values will be derived in Chapters


R

cp

cp

==

..

i R.

cr

cr

..

and y =

R. andy

14 and J S. The difference cp

c,.

1.67.

1.40. These
is equal to

8.31 x 103 J kilomole -I K- 1 for a wide range ()f temperatures. For solids

the difference is about 5 x 1 O' J kiJomole K- 1 at 1200 K by way of comparison.


..

Finally. from the foregoing analysis we can calculate the internal energy

of an ideal gas. From Equation

( 4.9) we can write

(' v =

since

11

ar

depends solely on Tin the case of an ideal gas. Then

or
T
ll

u0

-T

CvdT.

If cv is independent of the temperature.

(4.16)
*Julius Robert von Mayer. a German physician. Jiscovercd the equivalence of heat and

work and first enunciated the principle of conservation of energy

( 1842 ).

Sec. 4.3

Enthalpy and Heats ofTransformation

57

The specific heat capacity at constant volume is nearly constant over a region
of temperature extending from approximately 200 K to 1200 K for most gases.
Thus Equation (4.16) is applicable for many practical problems.

4.3 ENTHALPY AND HEATS OF TRANSFORMATION

The heat of transfotmation is the heat transfer accompanying a phase change.


A change of phase is an isothetultlal and isobaric process and entails a change

of volume, so work is always done on or by a system in a phase change. Thus

is the specific work involved in the process. Now

du

aq- Pdv

or, for a finite change,


.

'

so that

I
I

(4.17)

I
.
.

Here l is the latent heat of transformation per kilomole associated with a

given phase change. Let


h

+ Pv.

The quantity h is called the specific enth.1/py. The origin of the word is the
Greek verb thalpein meaning ''to heat." Since

u,

P, and v are all state variables,

h is also a state variable. (If two variables have exact differentials, their prod

uct and sum also have exact differentials.)


With this definition, Equati on

(4.17) becomes

which is to say that the latent heat of transformation is equal to the difference
in enthalpies of the two phases.
Consider now the three possible phase changes. Let 1 denote a solid, 2 a
liquid, and 3 a vapor. Then .

Applications of the First Law

Chap.4

58

I 12

/23

/11

h'' - /z' solid to liqttid (fusion).

h'"

h'" - h'

h"--+ liquid to vapor

'

(vaporization),

solid to vapor (sublimation) .


.

Here the first superscript is associated with the .'iecond subscript; that is, dou-

ble prime is associated with the subscript 2 denoting a liquid etc. This is
merely a notational convention.
Since enthalpy is a state function..
,

tlh = (.)
the change of enthalpy in a
diagram as in Figure 4.2.

cyclical process is zero. This can be depi cted in a p T


..

Consider a cyclical process around the triple point and close enough to it
so that only changes in the enthalpy occur during phase transitions. Then
solid

:.

vapor (heat flows in)

..

dhl

vapor liquid (heat flows out) ..

h 2

liquid sol id (heat flows out).

h3

=:

Here

CP

solid
liquid
Figure 4.2 Pressure
temperature plot for a
substance that contracts on
freezing. The circle denotes a
cycJicaJ process around the
triple point for which the total
change of enthalpy is zero.

ST.Rl'
vapor

/13
I:.2
/21

1-r:,

= -lr2

Sec. 4.4

Relationships Involving Enthalpy

TABLE 4.1
substances.

and fusion for various

Latent heats of

Latent Heat of Vaporization


at Steam Point,

Mercury
Alcohol
Gasoline

Latent Heat of Fusion

1 atm

538 kcal/kg
63
207
95

Water

59

at Melting Point,

1 atm

80kcal/kg

Water

3
5
77

Mercury
Lead
Aluminum

so

or

(4.18)
(All these latent heats are intrinsically positive.) This is an important result.
Some values are given in Table 4.1. The latent heat of sublimation for water is
538 + 80
618 kcal kg-1
=

The SI unit for latent heat is J kg -t. Thus, for example,


8
kcal
53
kg

4184

J
kcal

2.25

106

J
kg

4.4 RLATIONSHIPS INVOLVING ENTHALPY


In Section 4.2 we assumed that

u = u(v, T).

Whereas the selection of

from among the three fundamental stat variables

P, v,

v, T

and Twas arbitrary,

we will find that this choice is a natural one both from an experimental and
mathematical viewpoint. Similarly, the natural choice in the case of the thermodynamic variable

h is
h

h (T, P).

(4.19)

Then the analysis proceeds in a manner exactly parallel to what was done in
the case of the internal energy. Here
aq
h

du + Pdv,

= u + Pv,

(4.20 )
(4.21)

Sec. 4.6

Work Done in an Adiabatic Procass

61

since, from Equation (4.27), h depends on T only. Hence we can write


T

h - h0= cpdT = cP ( T

T0)

To

if

Cp is a constant. For an ideal gas, then, the specific enthalpy is given by


h=h0+cp(T-T0).

::: : ::;::: ::::::::::-:;


: :::::: ::::= :: .:.:: : :;::::.:::::
: .::;: :::. ::: :: : ::::=: ::: :: ::-:
=: ::.: :: :;::=
-::: ::::::: ::

(4.29)

::*:::.:=;:::::.:::=:::.:=\:: .::=: ;:::.:.:'::::: :.:::::::::::::.: :::.: :::: :::::::: ': :: :: .:.::::::::::. ::;::.;:.:: ::: :::: :;:::.: :.::: .:.: :::.: :: : ..:::: ;: : ::

4.5 COMPARISON OF

AND h

Table 4.2 summarizes the parallel expressions involving the internal energy
and the enthalpy.

TABLE .2
enthalpy.

Analogous relations involving the internal energy and the

Reversible process:

Internal energy u

Enthalpy h

du

dh

Cv

tlq- Pdv
dU
dT

Cp

ilq + vdP
ah

llq c"dT
au
0
av r

Ideal gas:

Pdv

ar

ilq cpdT- vdP


iJh
0
iJP T
=

h- h0

c,.dT

T.,

4.6 WORK DONE IN AN ADIABATIC PROCESS


We have seen that the specific work

P( v2 - v1), where

v1

done in an isobaric process is

is the initial specific volume and is the final volume.

Similarly, for an isotherlttal process in which the substance is an ideal gas,

(4.30)
We now wish to find the specific work done in an adiabatic process involving
n ideal gas. Setting llq
0 in Equation (4.28), we obtain
=

vdP = cpdT.

.-

(4.31)

62

Chap.4

Applications of the First Law

We also have
7/q
which for 7/q

c,.dT + Pdv

0 yields
Ptlv

c,.dT.

(4 32 )
.

Dividing Equation (4.31) by Equation (4 32 ) gi ves


.

vdP

cp
-

Pdv

-y ..

--

-----

c,.

:__

or
dp

(/1)
.
-y

- =

( 4.33)

'l'

jt>

This equation can be easily integrated to give


( 4.34)

where K is a constant of in t e g ra t ion . This is the relationship between the pres

sure and volume for an adiabatic process involving an ideal gas. Since y > 1

it follows that P falls off more rapidly with v for an adiabatic process than it

does for an isothermal process (for which J:lv

==

constant).

The work done in the adiabatic process is


V2

Ptlv

1C =

v-Ydv

==-

1
(Kv Y)

1)2

. .

1-'VI

--

VI

v,

Pv1 at bot h limits: if we use K


Now, K
P1v{Y at the lower limit 'e ohtain
K
=

P2vl at the upper limit and

lV -=

For an expansion v2

> v1

..

lV >

1 -y

t.

0. and the work is done bj the gas: for

pression the work is done by the surroundings

reve rs i ble adiabatic process.

1-' = u!

ful expression for an ideal gas.

( 4.35)

---- r P2L2
),
- PI v IJ1

lt2

==

cr(T1

<Jn
-

com

the gas. Note that for a

T2) .. which is another use

Sec. 4.6

Work Done in

an Adiabatic

Process

As an example of an adiabatic expansion, consider an ideal mona


tomic gas enclosed in an insulated chamber with a movab le pis ton The ini. tial values of the state variables are P1 = 8 atm, V1 = 4 m3 and T1 = 400 K.
.

The final value of the pressure after the expansion is P2

1 atm. We wish to

find \12, T2, W, and d U. For an ideal monatomic gas, the ratio of specific
heats 1 is

5/3.

For an adiabatic process,

Then T2 is found from the ideal gas law:


13.9
4

(400)=174K.

As the gas expands, the temperature and the pressure drop (no heat flows into
or out of the system).
The work done by the system in the ex pansion is

---

= 2.74
Finally, since

106 J.

0, dU = - W= -2.74

106 J. The internal energy

decreases. This result could also be found by

using the expression

AU= Cv(T2 - T1 ):
3R

(8

=-2.74

PtYt 3R
1.01

105

400
X

4) 3
- (174- 400)
2;

106 J.

Work is done by the system at the expense of internal energy.

Applications of the First Law

Chap.4

64

PROBLEMS

4-1 The specific heat capacity

c,.

of solids at Jow temperature is given by the Debye

T3 Jaw:
c I)

T
=A
0

The quantity A is a constant equal to 1 .4 x 1 o- J kilomole -I K- 1 and 6 is the


Debye temperature, equal to 320 K for NaCI.

(a) What is the molar specific heat capacity at constant voll:Jme of NaCI at 10 K
and at 50 K?

(b) How much heat is required to raise the temperature of 2 kilomoles of NaCl
from 10 K to 50 K at constant volume?

(c) What is the mean specific heat capacity at constant volume over this tem
perature range?

4-2 The equation of state of a certain gas is { P + b) v


aT + ht + u0
energy is given by u

RT and its specific internal

(a) Find Cr.


(b) Show that c,.

c,.

R.

4-3 Show that

au
(Refer to Equation ( 2.13 ). )

4-4 Show that

dU

K
= PK'V- (cp- Cv) -.
aP ,
f3
(Refer to Eq uat i ons (2.13) and ( 2.14 ). )

4-5 When a material is heated .. it cxperiencl!s a very slight increase in mass since
E
me:.. Estimate the fractional change in mass when the temperature of a
block of copper is rai sed from 300 K to 400 K. Take cp for co.pper to be
2.6 x. 104 1 kiJomolc- 1 K- 1 Its atomic weight is 29.
=

4-6 Consider oxygen as an ideal gas w i th c,.


(5/2)R. Suppose that the tempera
ture of 2 kilomoles of 0" is raised from 27C to 227C.
(a) VVhat i s the increase in inte rnal energy?
(b) What is the increase in enthalpy?
=

4-7 The specific heat capacity at constant pressure of a gas varies with the tempera
ture according to the expression
(

( P == tl

+ hT

'

'
2
T
-

Problems

65

where a, b, and c are constants. How much heat is transferred in an isobaric

process in which a kilomole of gas experiences a temperature increase from T

to2T?

4-8 Show that the following relations hold for a reversible adiabatic expansion of an
ideal gas :

(a) TV1-l

(b)

T
pt-lh

constant.

consta nt

4-9 An ideal gas undergoes an adiabatic reversible expansion from an initial state

(T1, v1) to a final state


(a) Show that

(T2,

v2).

In

T, \

_::

Tl

( 'Y

1 ) In

v1
v2

where y is the ratio of specific heats.

(b)

If TJT1

2/5 and

vJv1

2. show that the final state is not accessible from

the initial state via any adiabatic reversible process involving any known
ideal gas.

4-10 The equation of state for radiant energy in equilibriunt with the temperature of
the walls of a cavity of volume Vis P

equation is U

(a)

(b)

aT4V.

aT4/3, where a is a constant. The energy

Show that the heat supplied in an isotherrnal doubling of the volume of the

cavity is (4/3)aT4V.

Sho\\ that in an adiabatic process, VT3 is constant.

4-11 An ideal diatomic gas, for which

cv

5R/2

..

occupies a volume of 2m3 at a pres

sure of 4 atm and a temperature of 20 C. The gas is compressed to a final pres

sure of 8 atm. Compute the final volume, the final temperature, the work done.
the heat released.. and the change in i_nternal energy for:

(a) A reversible isothermal compression .


(b) A reversible adiabatic compression.
4-12 An ideal monatomic gas having an initial pressure P0 and an initial volume V0
undergOC!s an adiabatic expansion in which the volume is doubled.

(a) Show that the fmal pressure is 0.314 Pr}


(b) Show that the ch ange in enthalpy in the process is

-0.93 P0VQ.

4-13 One ki l ogram of ice at -20C is heated to ooc and melted. The resulting water
is heated to l00C, vaporizes, and the steam is furth er hated to 4QO>C. Assume
that all the processes are isobaric. Calculate the c hange in enthalpy in kilocalo
ries and in

joules.

(The specific heats at constant pressu;e of ice, liq uid water,

and steam are 0.55, 1.00, and 0.48 kcal kg-1, respectively.)

4-14 An auto mob ile tire is inflated to a pressure of 270 kPa at


After three hours of high speed
-

driving the

the beginning of a trip.

pressure is 300 kPa. What is the

internal energy change of the air in the tire between pressure measurements?
Assume that air is an ideal gas with a constant specific hea t capacity Cv

and that the internal volume of the tire remains constant at 0.057 m3.

5R/2

Chap. 4

66
4-15 Shortly after detonation the fireball of

Applications of the First law

uranium fission bomb consists of a

sphere of gas of radius 15 m and temperature 3 x lD-c; K. Assuming that the

e xpa n sion is adiabatic and that the fireball remains sphe ri c a L estimate the
radius of the ball when the temperature is 3000 K. (Take y =

1.4.)

5.1

The Gay-Lussac oule Experiment

69

5.2

The oule Thomson Experiment

72

5.3

Heat Engines and the Carnot Cycle

74

67

5.1 THE GAYLUSSAC-JOULE EXPERIMENT

We have postulated that in general u

u(T, v ) . The first efforts undertaken

to measure the dependence of the internal energy of a gas on its volume were
made by Gay-Lussac and later by Joule in the middle of the nineteenth cen
tury. The results of the experiments showed that the internal energy is a func
tion of T only and does not depend on the volume v. This, like many other
properties of gases, is on ly approximately true for real gases and is assumed to
hold exactly for ideal gases.
Before describing the experiments, we consider the following arguments.
Using the cyclical and reciprocal relations for partial derivatives given in
Appendix A, we can write

au
iJv

aT
-

av

dlt

II

aT

For a reversible process, we have (Equation

(5.1)

,,

(4.9))

au

Ct

aT

Thus Equation (5.1) gives


aT

...

dU

av

= -c
1.

av

(5.2)

ll

Equation (5.2) implies that if we can measure the change in temperature of a


gas as the volume increases while the internal energy is held constant, we can

69

Consequences of the First Law

Chap. S

70

diaphragm
I

as sa
...

mpl

I
.-----

vl - vo

vo. To
-

thermal insula t1on"

Figure 5.1

vacuum

Setup for the Joule experin1cnt prior to rupture of the diaphragm.

determine the dependence of the internal energy on the volume. The question
is: how can we keep

11

constant duri ng the expansion? Since

ev i dently we want an adiabatic process in \\'hich no work is done.


The pr inci ple used by Joule is shown chematically in Figure 5.1. A sam
ple of the gas unde r investigation is enclosed in one
insulted vesseL separated by

d i a phragm from an initially evacuated cham

ber . The t empe r atu re of the gas is measured w i th

diaphragm is rup t u red

the

gas under goes

tial specific volume of the gas.

c1

portion of a thermal1y

thern1ometer. When t he

.free eplpansiv11. He re

V0

i s the ini

the final specific volume: and T1 are the ini

tial and final temp e r a t u re s respectively. In the


.

e xpa n sion

no work is done and

the internal energy is unchanged since no nergy is lost or gain ed . * The fina]
temperature which can he measured. is
u, oT
Vo

av

dv.
u

from basic mathematics. The subscript on the d e ri va t i ve is appro priat e since

the internal

e ne rgy

coefficient.
Joule

\\'as

is held c onsta n t. The integrand is kno\\'n as TJ, the

unable

to

measure

any

temperature

change

Joule

for

air.

Subsequent experitnents using various gases ind ica t e that the temperature
d e cr e as es bv at most a verv sn1all amount. Summarizing these resul ts
.

a v.

0.001 K kilomole m-3.

is finite. However. the rupture of the diaphragm


results in a rapid expansion during which the gas is so highly nonuniform that we cannot assign a
distinct value to the pressure. See Sect ton 7.7 for a further discussion of free expansion.
)jl

Although

<

..

no external work is done. Pdv

Sec. 5.1

The Gay-Lussac Joule Experiment

71

It can be shown (Problem 5-3) that for a van der 'aals gas,

Since the constant

is zero for an ideal gas, \Ve conclude that 11 is exactly zero

in that case:

iJu

'TJ =

dV

= 0

and

(5.3)

(T)

for an ideal gas.


This very in1portant experimental result can be proved theoretically.
Consider Equation (4.8), which is valid for any reversible process. If \Ve divide
the equation by the temperature T, we obtain

au
T

T aT

dV

+ P dv.

(5.4)

In Appendix A we note that an inexact differential can be made exact if an inte


grating factor can be found. We shall see that the integrating factor

1/T makes

the left-hand side of Equation (5.4) an exact differential, a function of the state
variables

T and v. * If d z

M dx + N dy, then the condition that dz be exact is


iJM
-

iJN
=

ay

iJx

In this case, the condition is


1

au
av TaT

a 1 au
aT T av

'

where the subscripts on the partial derivatives have been omitted for conve
nience. Carrying out the differentiation, we have
+

iJ2u

TaTav

1 iJP
+-

--

TaT'

or

au
av

__

iJP

aT

- P.

(5.5)

This fact is intimately associated with the introduction of the concept of entropy. as will

be shown in Chapter 6.

Consequences of the First Law

Chap. 5

71

s
a
g
l
a
e
d
i
n
a
r
o
F

RT

..

so that

du
dV

-P

--

dT

RT
-

RT

(5.6)

0.

==

of
ity
lid
va
e
th
s
on
d
p
n
e
de
f
oo
pr
e
Th
v.
on
nd
pe
This proves that u does not de
the assumption that aq/T is an exact differential and the applicability of the

analysis to an irreversible process. These matters will be discussed in Chapter 6.

5.2

THE JOULE-THOMSON EXPERIMENT

The Joule-Thom son experimen t also known as the Joule-Kelvin experiment"*


gets the same result as the Jou l e e xp e ri m ent in a different way. It is important
._

because the temperature drop in Joule "s experiment is difficult to measure.


The experiment involves a throttling process, performed by forcing gas
th rough a porous plug to a region of lower pressure in an insulated cylinder
(Figure 5.2). A continuous throttl i ng process can be pe rformed by a pump that
mai n tai ns a constant high pressure on one side of a porous wall and

constant

low pressure on the other side. The objective is to measure the temperature
difference T2
llq

and

T1

S ince the p rocess takes place in an insulated cylinder,

u2 - u1

1v =

0' The specific work done in forcing the gas

t hrough the plug is


0
=
w1

P1

dv

P1,v1 ..

li l

while the work done by the gas in the

ex p a n sio n

P2 ti'v

UJ2 =

==

is

P2v2

The tota l work is therefore

(5.7)

WiJJiam Thomson became Lord Kelvin at the end of the nineteenth centun.

Sec. 5.2

initial
state

The Joule-Thomson Experiment

7l

,...____,

__..-

porous plug

.. .

final
state

(b)
Figure 5.2

Principle of the Joule-Thomson porous plug


experiment. Here (a) corresponds to the initial state and (b) to the
final state, with the piston in the fully advanced position.

It follows that

or
(5.8)
Thus a throttling process occurs at constant enthalpy.
In the Joule-Thomson experiment, the pressure and temperature of the
gas before passing through the plug are kept constant and the pressure on the
other side is varied while the corresponding temperature is measured. The results
give a locus of points on a T-P diagram representing the isenthalpic state of
the gas (Figure 5.3). The numerical value of the slope of the curve at any point
is called the Joule-Thomson coefficient , where

aT
-

aP

(5.9)

The gas is cooling when /L is positive and heating when J.L is negative. The point
0 is called the inversion point.
where JL
=

The experimental result is that for most gases over a reasonably wide
range of temperatures and pressures, the T.P curve is approximately flat and

Chap. 5

74

Consequences of the First Law

tnvers1on

po1nt

I
I

I
t
I
I
J

Figure 5.3 A curve of


constant enthalpy for the
throttling process.

J.L

==

0. It can be shown in a manner completely parallel to the analysis of the

Joule experiment that

ah

. riP

-cp

iJT
aP

11

Thus, for an ideal gas


ah
aP

and

h(T).

(5.10)

The results of the two experiments are equivalent: that is.


dll

iJh

iJv

aP .,.

0.

Again, the result can be proved theoretically.

5.3 HEAT ENGINES AND THE CARNOT CYCLE

The heat engine is a highly useful concept in thermodynamics and reminds


one of the subject's origins. A heat engine is a system that receives an input of
heat at a high temperature.. does mechanical work., and gives off heat at a
lower temperature.
Before discussing a practical heat engine.. we consider two extreme cases.
In Figure 5.4(a) .. work is done

on

the system and is converted to heat. The heat

in turn is used to raise the internal energy <>fa heat reservoir. In this case, the
heat out (of the system

M) is equal to the work applied because the state of

the system is unaltered. Examples are stirring heat or the heat generated by
the flow of current through a resistor.

Sec. 5.3

Heat Engines and the Carnot Cycle

heat reservoir

Q
w

--

(W

75

machine or sub5tance

Q)

( W < Q)

(a)

(b)

Figure 5.4

The concept of a heat engine. In (a), work


is done on the system and is converted to heat. In (b),
heat is extracted from a reservoir and is converted to
mechanical work. This configuration is not possible.

Case (b) is the reverse of c.ase (a). Heat is extracted from a reservoir and
is converted to mechanical work by the machine. The question is: can the work
done by the system be equal to the heat in? Can the conversion be 100 percent
efficient? The answer is no. The second law of thermodynamics states unequiv

ocally that it is impossible to construct a perfect heat engine. (This is sonle


times referred to as the Kelvin statement of the second law, which will be
discussed in the next chapter.)
Thus case (b) must be modified as shown in Figure 5.5. Heat is absorbed
from the higher temperature reservoir, work is done, and heat is given up by the
engine to a lower temperature reservoir. The engine works in a cycle, so that the
state of the system is the same at the end of the cycle as at the beginning.

Fig11re S.S Modification of


Figure 5.4b to include a cold
temperature reservoir.

Q2

heatabsorbed

Qt

heat given off

uthermal pollution,

Consequences of the First law

Chap. 5

76

The efficiency of the engine 11 is equal to the work done by the system

(the machine or working s ubst ance ) divided by the heat absorbed Q2:*
JW!

=
71

==

-,Q

output
input

(5.11)

,
In other words the efficiency is the ratio of -\vhat you. get" to what you pay. ..

Applyi ng the first law to

the syste m

we have

In writing this, we arc mindful of our sign convention in which the heat flow
into the system and the work done b.,v the system are p os i tive q uantit ies..
whereas heat flow out of the system and W()fk done on the system are inher
ently nega tive We assume that the engine operates in a cycle. Since the state
.

of the system is unaltered in a cyclical process, there is no change in its internal

energy (recall that U is a state variable and U =


becomes!

in a cycle). The first law

(5.12)
Subs t itut in g this

in Equation (5. t 1 )

we

obt a in

(5.13)

We now consider by far the nl<)St important case of an engine operating


in a cycle

namely. the Carn<)t engine. We assume a four-step reversible

process in which the ""working substance'' is an ideal gas. The four steps are an

(a to b in a P-l' diagram), an adiabatic expansion (b to


c). an isothermal compression (c tel tl). and an adiabatic compression (d to a ) ,
returning the gas to its original state (Figures 5.6 and 5.7).
Referring to Tahle 5.1. we sec that the t<)tal work done in the cvcle is
isothermal expansion

.,

w
Note that

p dV

=:

"''

* 0: the

W' + w,

work

must

W"

==

p t/V4

be greater than zero if the engine is to

be useful. We should remind ourselves that because P and


ables,

dP

0 an d

* The efficiency

#tiV

71 is not to

Some texts writ

lV

{}2

0, hut t hi s does

he mistaken for the


-

value signs often leads to confusion.

Q1 n1caning "

'

n o t imply that

.Joule coefficient TJ.

::: ;Q! - Qt/. The

V are
p dV

state vari

om i ssion of

the

0.

absolute

Sec. 5.3

Heat Engines and the Camot Cycle

77

::
r:: .: ::::.
: ... :. : -:: : _ ! -;: :- ..
: -: :. :o:: ..:-. :::.;-
.::.:.: . :;<
.
:<, .
-
. . .. . ..<
,.

...
..
.. .,, ....
....
- .- : > ;s-. , :
.

<-;- . ,:. - , ..,.? .v

if.::'
X.
!
,.
: :ro'Y.o "
=::;:::
: : :
...-,= : :;,;

; ,;
::
oo-;' =' :f.
' ':JJ'

'.-:;: :.:
'o

'
:::.
=
. <':..:
A

: .:.- . : '.:.-7
;-

>:
.
;
:
:.;;

'
:

.
.
=.

.
.
.

. . ........
.. ..r.)'4';;
_ ., ,.". '
..
- ,v.,
.,. . .,,
.. .v
..,
,.,. __,
..;..-.<Zo\
. " .,
:
.-'.>.
.,..
,x

..;A
A.:::x
. (.. .:J., . .;
..., . ;,.,,, v
:
.
,.
:
.
;

;
,
:'
;-:
:
:
:.
.

.
'
.
.
..

x
.
.

<
.

.
-..
..
.
.
.
,
... ..
...

.
?.
..
.
..;'. ..:-.. ..\:.::
,
. ...... ..

. . , .:on
-,,.....
.. -.,
..

.
., ..&
'
.
.
..
..
..

,,
.
.
.

.
.

.
..
.

)
,
.
.

.
o.....,
'\
, N,
, ' ' .o,...
, . .... ........,
oC- .
. ......-:..
....,.... ... . ..
, ,.
, -,
,... .. ...

..,-..
, ..., ...... ..
....-,.
...
.

. . . . . . ..,
.

,.
.

0
.
::=; :

:
;
=*
!
:
::::::
:

::
:
;
::
-[:
::

:
:
=
.;
:;;; .
:}
:{

.
.
.

. ;-,:.-. -.'.>:..-:.;;...:.
: -:;.o:
:..""'
.>.. ':$;..A;'(..'
. ..; :..
. ..
.
;;

"X...,.
; - ..,,. . .-:....:-;,.
v . v .; .
eo
'_.. ,,
...._
:- . ,;;.,.
o
.l:- A
.;,
!:
'
"
r
.
,

.
,.

<
;

6
X
J'
-'
..
O.
.
;
o.,

'
'
. ..,, .. . -Y
o
o
'(

<
;
:

'
il
.
. ..,
,r. _,
:
.o
:
'
1':":"1":

.
'
lb : ..o :.
:.,.
..
: ''...1.:
, : A
':' .,!. .

'

.
..

"

"

'

,
.
'!
'
,
/'., _,,
.
_.y_,,
_.

'A
A
.;o,.r

'

'

.
h'
A

.
:oro'""
._
,
r,.

A.t"

'
'
-
o
J
'
'7

),.. , (o .:<'
,
,T
.
)<;o
,,. , , ,o -'o'
u , L
;Jit;
o. ;.o o- ,
0 , ,,

.
:
7.;-:::.<
:;;
....
:

...
;.,>:.
0:.....:.):;;
:'
-,:.!,
:
' -:..
. .. .,. f',
-"
: ......
. o ,.. . . .....
.... .... >o:.
Y.
.
,

.
.
1. ,:

. <"
A; . ......
':'-' o .;;,;
N(.."' 0 ')0
. .:'.':'
.;,.. .;; ;.<

'

.
.

.
.

:.; ...
" . ..>.<,
. ;,
-(:.(
.> .
\;
=\
X
,-. ..-=
' .: :-0 A'
. . :j.o. .
....
- ":
... - "=-;
.

.. .:,. ._.x
. .A. :.:,
. '
.
. ,, . ..,.;.)
>
i'\: :0

;;
::
c
:
:c;"
:<
=. ...''
-!

>
..,_>
:'.
"
.

;
=-

<

;
..A
,
'

!
"

!
0

'

'
.!-':. -:>-,;,..:
- : .(<
' :
:
:
'.h'

''-'t. .:.- . o, '.v


J
.;.
.
Y
.'\
.
.
.-..' L
. ...
,,.,__
' '
..
';,. '' ""
'
'

'
'
<I
"
No.
,

.
A

)
'

, ' ',Jii

A ',,' .''
o
)i;
, , ,
, , o:
:y. o'<l' oTo'o'/'_', "-' "-J\' o'V"o'"
'(,..;.. 0

-;
.:..: .; .;

!.:<\.
: ;
-'.. .: , ..
. :::

:
,
.
o
\
1
:
:

.
:

;. . :O::
...:
.
,

v >"..> . ,.....
.-..:
. . <\!
' o, . .'0 '. , y,o, .?....-..... ..... .....; :!-.

.
.

..0
.

,.
.
<

.
.

, .. ,'

...,. . ... -... -

, : .
. , ) .; ,.,
...: .-.; :-:-x.
.... .-.
. :
. .:-: .:
:
::
:.
..

:.:

:,:;<
: ,..;;: :,

.
.

,
,
,
'
'
,
-::_
.
'\.
.
".t
...
,

.
.I< --
. . . '"
.
.r
.
oV'.'''"
...
9f.
. , . .. .'v,.
"

.
...
_,
,.A..
.
.

'""
..
,
.
".' t. .".-'.'.

'"
""f"
V"

oo
oo
"

..
..
o

X
..
.
..,v

.
'

.
..
...
,
...
.

t
.;
<

,
,
,
.
,.
.
.
.)

.
..
.-.
..

.
.
.

,
.
.

.
, .. . < . . .,,
.. .. . ..t[!tL,' '" ',., . .. .. '1 . o . .
;.
.
-:
:

?:..: ..-,.: ...;


..
:
;

:.-.
:
...
,
..-._,;. ' '
:-:
:-:t.::.
<
:l
:
l
.
;
.
.
:
.
:.'. . .. .
.
..
.
.

.
-..;

,,
v.X.
.:..t
X'
. .,
,-- ,)'..,...,..
.:
<.. ,
...,"
".....,.....
. ,. :.:<
:
:,
.
. ..'< )
,
-""'
- V., .., .v
..... x,....
... .A
'
.
,.....
'
'"'
vv-1+
. \....,..,. >'

.
.
.
,,.,
......
..:.
' ', ' '.1'.0. . .. ...,,...
,
.
&:
..,
..
,

..

.
.
.
.

.
.
v
:.o..,
.. 6 . (...' ,.x,,.
.-
..x_
. . :o . .
...
.........
,; .-,..,., ,,r
; . ....ol':.,
, .." ' ;r.
<.., ..,. .
. .
, "'. . . '-.-.

'

. v
.;
.
.
.:
,
:
.
A
.{!
<

:
>
7

'
.
:
.
,
..
,

""

'
-"'.,,

, ::
...
,0 * ,...
.

"'!'

.
.
A.
:-

,. . . '-

0
,,
. .,
....
"
.
....

,
.. ,...

. ?.i
. ' )0 . 5-'>'.t.
,,,..., i' .- -...- , .,. v, ,:.,'.>
,..:.Y<'f..
.

W
,
:l!'.:.:
_,.
.,

"
<1.'.
('\
'

,
'L
.

.,.
.
.

- ., ....:J,..
, ,. -':'
. :
r. ;... ./..
. . .,...,,..,o,...,
.r....,.....
. .
0
)

.
"-.;.
Y."
. : :: ::: :.;;
" .;:
;
;:
.,. ..
<:;_>
.
-.
":;
.
.
-;,-.. := :
.;::
::::

:
..... ....,
.":it
.<
.

.,0 'V
:...:..>:. ...'-:': :
.

..

,
.
.

x-:
0:
,.
.
"
.
.

,.
'

'
'
.

.
'
'

:
"
*"
#
"
A
v,
,
'li
.

J<
.
,

'
"
'",
o

,...

,
"

.,
' . . ....v
.y
,

\'
'. ....
...:.-.
..._
,
,_.+
=-- . )' $
.. ,
.r.
......'rol'.
:...

. .,
. ?S
. ....
. . "'......''....... ..,... .,.-
.
.
.
>
.
._:.
. ., ....-::t,
.
,

'-"-o-
..

...
:.0.

<''"-'
,""
..;, '
"
"'.e;.' NV'
.
,Jo,o:'..
'''<',.
W
"#:''
.
.

'
.,

.
...
-:;Y."!'"
"'=.:,(,:r-<'
'. -'":'":>::Ci;:..;
,
.
.
.
::;;;;:;.::;:"'?;:'. : ::
:- .:::: :::. :*::
-:
=,::: :=:;
' ..'...
. ...-;,o..,
,. . . . :1) :,C:
. : >. ....
,
..;..
o -,.o,,
.
".
'.
,

-...
r
,
.
(

..
A
.
.

.
.
.
.
.
.

.. ... . .,'i., . ,.
..,_
.
'1
'
...
0
..,-. 0

.
. 4
.,...,

.
.
.

'If.

: :

.
.

.
;.
:
:
:
v
._:-!o_t,;vo -:
""0
:o
:

:
:
:
:
..
.
:
;

.
>

-;,
:

'

:
;

.
.
.
.
.
J ' -":"
'\1 oo oo .' :.- ... ... .o
. ... ..o...o .. o.....
.... ..,....
"""'
...,,0. ,... : . > o ..
..,,Yo
.
..
. . 0 ...,.
'.
.
,

.
. '..
. ...

..
. ..... ... . . . ....
. . . ... .. -'
,,

: ?. i

..
.

..

.. ._

. ,_

_,,
,.

..

-::.:A -'.!
,_.
,
v o,

'

: :..:.

. ,

-..
..-
.- "--l
.

.._.

o roo

oo

oo

..

'
'

'
' -!\

_.

.o.
..

..

.,.

.-_.,

.. ..

....

YV .-'.A.
"'

..

..:.' .

.,

..

..

...

......

...

. .

..

...

'o'
o
...
.
0 .. ..
o

::;x
.
.

...

..

'o

....

..

, ..

...

0 .. .....
.

'

....

.
.

,_
..

..

...

....

..
- .

....

.1.

..

..

...

''
"'

..

'.

..

...

insulator

insulator

isotherrnal
expanston

adiabatic
expanston

isothermal
compresston

adiabatic
compresston

(a-+ b)

(b .... c)

(c-+ d)

(d-+ a)

Figure 5.6

The four steps of a Carnot cycle.

p
Q-.

(adiabat)

r2
(adiabat)

5.7 P-V diagram for a


Carnot cycle. The slopes of the
isotherrns and adiabats are
exaggerated.

TABLE 5.1

Heat transfer and work for the four steps of a Carnot cycle.
Heat

Process
a

Isothermal expansion

.. c

Adiabatic expansion

d
d-+a
c

lsothertnal compression
Adiabatic compression

Q2

>

O(in)

0
Q1 < 0 (out)
0

Work

0 (done by system)
W' > 0 (done by system)
W1 < 0 (done on system)
W" < 0 (done on system)
w2

>

Chap. 5

Consequences of the First Law

aQ -ttw.

(5.14)

78

We can write the first law


dV

On the isotherms. for an ideal gas d U


alone. Thus aQ

Now

b and

l!W. Here Q2

0 since U depends on the temperature

W2 and Q1

are on the same adiabat py-r

W1. Hence

(>

0 since vb >

(<

0 since V:1

=:

(5.15)

< -).

constant. Since P

(5

1 6)

nRT/V we

have

so that Tv;--

constant and
(5.17)

Similarly, a and d lie on the same adiabat. so


TV

If we divide Equation

(5.17) by

y-

Tv
I
./, ..

Equation (5.18) and take the

(5.18)
1/(y-

l)th

root. we obtain

V,,

v:.
-

(5.19)

v:,

Combining Equations (5.15), (5.16), and (5.19), we obtain

Ql
-

Q:
Using this expression in Equation

==

r1

(5.20)

T:..

(5.13) gives
(5.21)
-

for an ideal gas. Since T1

<

T2, it follows that 11

< 1.

Sec. 5.3

Heat Engines and the Carnot Cycle

79

The work perforttted during the adiabatic steps of the cycle can be calcu
lated using Equation (4.35). However, we note that the use of Equations
(5 .11), {5 .15), and (5.21) yields W, the total work done, which is the quantity of
greatest interest. It can be easily shown that this is just the area enveloped by
the curves in the P- V diagram delimiting the cyclical process.
It is important to emphasize that a Carnot engine operates between

only two reservoirs and that it is reversible. Also, if a working substance other
than a11 ideal gas is used, the shape of the cunyes in the P- V diagram will, of
course, be different. Carnot theorized that the efficiency given by Equation
(5.21) is the maximum efficiency for any engine that one might design. We
note that the efficiency would be 100 percent if we were able to obtain a low
temperature reservoir at absolute zero. However, this is forbidden by the
third law.
A Carnot refrigerator is a Carnot engine in reverse (Figure

5.8). Work is

done on the engine with the result that heat is removed from a low tempera
ture reservoir {the interior of the refrigerator) and delivered to a high temper
ature reservoir {the surrounding room). In this case we define a coefficient of
performance

as the ratio of the heat Q1 extracted from the low temperature

reservoir to the work done on the system

again, ..'what you get" divided by

"what you pay." Because Q1 is positive (heat tlo\v into the system) and W is
negative (work done on the system), we introduce a minus sign in order to
make the coefficient of perfor111ance a positive quantity:

(5.22)

The last step follows from Equation (5.20), which applies to the Carnot refrig
erator as well as to the Camot engine. Since T1 < T2, the coefficient of perfor
mance of a refrigerator, unlike the thermal efficiency of a heat engine, can be
made much larger than unity.

T,

...

--

Figure 5.8 The concept of a


Carnot refrigerator.

Chap.5

80

Consequences of the First Law

refrigerant
(heat out to surroundings)

(e.g., freon)

high
pressure

condenser

high
pressure
vapor

liquid

compressor
throttle
--

low
pressure

low

liquid/

pressure

vapor

vapor

evaporator
long capillary tube

Figure 5.9

Ot

(heat in

from refrigerator space)

Schemati c diagram of a typical refrigerator.

A schematic dia gram of a typical refrigerator is shown in Figu re 5.9. The


refrigerant is a substance chosen t<> be a saturated liquid at the pressure and
temperature of the condenser. The liquid undergoes a throttling process in

which it is cooled and is partially vaporized. The vaporization is completed in


the evaporator: the heat of vaporization is heat absorbed by the refrigerant
from the low temperature reservoir (the interior refrigerator space). The low
pressure vapor is then adiabatically compr(ssed and isobarically cooled until
it becomes a liquid once again.

Refrigerators are designed to extract as much heat as possible from a


cold reservoir with small expenditure of work. The coefficient of performance
of a household refrigerator is in the range 5 to 10.

PROBLEMS

S-t The Gay-Lussac-Joule experiments demonstrated that

if l'

=0

for an ideal gas. Show that this result in1plies that

ilh

,iJP

0.

(Hint: Write t h differential dh and assunte that h

h(P . T)

and u

( 1'. T).)

Problems

81

52 Assume that the specific internal energy of an ideal gas is given by

u
(a)
(b)

u0

cv(T - Tv), u0

constant.

Show that the Joule coefficient 71 is zero.


Show that the Jouie-Thomson coefficient JL is zero.

S-3 The specific internal energy of a van der Waals gas is given by

u0

'

a
cvT - -, u(), a constants.
v

(a) Find an expression for 71 Sho\v that 11 = 0 if a == 0.


(b) Find an expression for the specific enthalpy h as a function of v and T.
(c)
(d)
(e)

Show that JL

RTv

Cp

( V- b)

v
Cp

Calculate the isothertnal compressibility


Show that if

0,

K =

RT

, and p.

for the van der Waals gas.

0.

54 Show that

(a)
(b)
(c)
5-S

iJh
aT
ah

=
v

av

ar
av

Cp 1-

=
h

{3p,

JLCp
-

VK

J.L
--- --

v(

p,/3 - K)

Carefully sketch a Carnot cycle for an ideal gas on


(a) a u-v diagram;
(b) a u-T diagram;
(c) a u-h diagram;
(d) a P T diagram.
-

S-6 A Carnot engine is operated between two heat reservoirs at temperatures of


400 K to 300 K.

(a)

If the engine receives 1200 kilocalories from the reservoir at 400 Kin each
cycle, how much heat does it reject to the reservoir at 300 K?

(b)

If the engine is operated as a refrigerator (i.e., in reverse) and receives 1200


kilocalories from the reservoir at 300 K, how much heat does it deliver to
the reservoir at 400 K?

(c)
(d)

How much work is done by the engine in each case?


What is the efficiency of the engine in (a) and the coefficient of perfor
mance in (b)?

5-7 Fifty kg of water at ooc must be frozen into ice in a refrigerator. The room tem
perature is 20C. The latent heat of fusion of water is 3.33 x lOS J kg-1 What is
the minin1um power required if the freezing is to take place in one hour?

S-8 W hich gives the greater increase in the efficiency of a Carnot engine: increasing
the temperature of the hot reservoir or lowering the temperature of the cold
reservoir by the same amount?

Consequences of the First Law

Cha p. 5

81

S-9 An idea ) monatomic gas is the w or k ing substance of

Carno t engine. During

the isothermal expa n s ion the volume doubles. The ratio of the final volume to
the initial volume in the adiabatic expansion is 5. 7. The \\ork output of the
engine is 9 x 1 0" J in each cycle . Compute the temperature of the reservoirs
between which the engine ope ra t e s. ( A ssutne

1 kilomole).

5-10 A h e t engine ha ving two kilotnolcs of an ideal ntonatomic gas as the working
substance under goes a revers ihle cycle of t.>peratio ns n1ade up of the four steps
shown in Figure 5.1 0. Here ( T, V, )
'
( T. ,)
( 400 K. 10m ). ( T. V-l)
.

(a)

( 300 K. 2 m . ) . ( T". V )
'
( 300 K. I 0 m ). Find:

( 400 K. 2 m .)

the heat flow in each step:

(b) the work done by the gas in the cyc le :


(c) th e thermal efficiency of the engine.
5-11 Durin g some integral number of cy cles a rve rsible engine works between three
.

heat reservoirs. lt absorbs

Q joules

Qj joules

of heat from a reserv oir at T, .. also absorbs

of heat from a reservoir at T> de l ivers

Q. j o u) es

at T. and performs W j oules of n1echanical \\'ork. If T1

T'

300 K. Q1

1200 J. a n d "'

-=

of heat to
=

400 K. T:

rese rvoi r
=

200 K.

200 J. find Q a nd Q.

5-12 A stude nt clain1s to have co ns tr u cted

a ptototype cyc lical eng ine that operates

between two reservoirs at temperatures of 500 K and 350 K. respectively. In

each cycle the engine e xt racts 6000 J of heat fro m the high temperature reser
voir. rejects 900 J of heat to the lo\\: ten1prature reservoir and per forms 2100 J

of work on the surround i n gs Would you be prepared to give financial support to


.

the development of the engine._,

5-13 A heat pump is a device t h at can he used in winte r to \\'arm a house by refriger
a t i ng the ground .. the outdoor a i r or the \\'ater in the mains. A Carnot engine.
op er a t i ng in reverse as a heat pump, extracts heat Q: from a cold re s er voir at
t e m pe rat u re T and delivers heat Q to a reservoir at a higher temperature T: in
.

e ach cvcle .

(a) D e fi ning the coefficient of performance c as the ratio of what y ou get to


.. what you pay:' derive a n e x pression for c for a carnot heat pump.
(b) On a day when the outside temperature is 32 F. the interi or of a h ouse is
maintained at a ten1perature of oX:- F by means of a Carnot heat pump

'"

Calculate its coefficient of perforn1ance.

p
2
..... -isotherm

isochore -

Figure 5.10

P- V diag ra m for

a four-step cycle.

-/

isochore

isotherm
'-----
---.-.+-

\ .'

Problems

83
p
c

adiabat

Q,

Fig11ae S.ll

P-V diagram for

...-.

I
I
I
l

adiabat
a

..
..._
__.
..

___

the Otto cycle.

v,

5-14 The behavior of a four-stroke gasoline engine can be approximated by the so


called Otto cycle. shown in Figure 5.11. The processes are as follows:
e a

isobaric intake (at atmospheric pressure)

adiabatic compression (compression stroke)

)l

b--.c

isochoric increase of temperature during ignition (Gas combustion


is an irreversible process. here replaced by a reversible isochoric
process in which heat is assumed to flow in from a reservoir.)
adiabatic expansion (power stroke)
isochoric decrease of temperature (exhaust valve opened)

a--.e

(a)

isobaric exhaust (at atmospheric pressure)

Assume that the working substance is an ideal gas and show that the effi ..
ciency is given by

'vhere

(b)

vnft'h is the compression ratio of the engine.

Calculate ,1 for the realistic values

Note: Since )'

2:

r =

5 andy

1.5.

1 "Ne want r as large as possible. The maximum practical obtain

able value is approximately 7. For greater values, the rise of temperature upon
compression is large enough to cause an explosion

before

the advent of the

spark. This is called preignition. For diesel engines preignition is not a problem
and higher compression ratios are possible. This is partly the reason that diesel
engines are inherently more efficient than gasoline engines.

Consequences of the First Law

Chap. 5

84

5-15 Consider the Joule cycle, consisting of two iobars and two adiabats (Figure 5.12).

Assume that the working substance is an ideal gas with constant specific heat
capacities Cp and c, Show that the efficiency is
..

1J=l-

pl

()'-I

li)'

p,

2..,_

.
_ ..
..
..._

Figure 5.U P- V diagram for


the Joule cycle.

6.1

Introduction

87

6.2

The Mathematical Concept of Entropy

88

6.3

Irreversible Processes

89

6.4

Carnot's Theorem

91

6.5

The Clausius Inequality and the Second Law

94

6.6

Entropy and Available Energy

97

6. 7

Absolute Tern perature

6.8

Combined First and Second Laws

98
103

85

6.1 INTRODUCTION

The first law of thermodynamics is remarkable in its generality, simplicity, and


utility. Yet our statement of it is in some respects less than satisfactory.
Furthermore, in our discussion of heat engines in the previous chapter, we
hinted that the first law does not constitute a complete theory because certain
processes that it perntits do not occur in nature.
The problems are as follows. First, classical thermodynamics is concerned
with states of equilibrium and various processes connecting them. We have
seen that there is a very substantial advantage in dealing with state variables,
changes that are expressible as exact differentials. Then the exact process by
which a final state is reached from a given initial state is immaterial; the transi

tion is i n de pen dent of the particular path taken. However, two of the three
quantities appearing in our statement of the first law the work perfottned
and the heat exchanged are inexact differentials. The question arises: Is there
any way in which we can write the first law in terms of state variables only?
A second concern is that our theory emphasizes reversible processes
that is, quasi-static processes in which no dissipative forces such as friction are
present. But reversible processes are idealizations whereas real processes are
irrev
. ersible, and the first law draws no distinction between them. This leads us
to ask: Is there some state variable by which we can distinguish between a
reversible and an irreversible process?
Finally, in Chapter 5 we noted that certain processes are impossible that
are by no means prohibited by the first law. It is not possible to construct a
machine that convet1s all the heat it extracts from a reservoir into useful
mechanical work. It is not possible to transfer heat from a cold body to a hot
body without supplying energy in the forrn of work. The first law says nothing
about these prohibitions, these "principles of impotency," as they have been
called. Is there, then, something missing from our theory? Do we require a sec
ond fundamental law for a complete description of our world as it actually is?
The answer is yes to all of these questions. We will begin by introducing
the concept of entropy as a useful mathematical variable. Phenomenological
observations will give the concept its physical meaning.
87

The Second Law ofThermodynamics

Chap. 6

88

6.2 THE MATHEMATICAL CONCEPT OF ENTROPY


The first law in its most general forn1 (f<Jr a cfosed system) is
dU

7/Q-

( 6.1)

7/W.

Neither aQ nor ztW is an exact differential: only their difference is exact. We


saw that for a reversible process the work is configuration work alone which
can be expressed as
aW,

PdV.

-=

where the subscript denotes a reversible process. Thus


a"'
---

(6.2)

t/V.

Now.. the volume Vis a state variable and tf\l is an exact differential. By multi
plying llW, by the integrating factor l/ P. \Ve have replaced an inexact differen
tial with an exact one.
Is there an i n tegra t i n g factor for liQ or liQ, that \villlead to an t!Xact dif
ferential? We note that Pis an intensive state

ia ble while ZIW, is an exten

va r

sive variable. The quantity aW,jP is therefore extensive. Also, two of the three
fundamental state variables

P, V,

and T appear in Equation

( 6.2). This leaves

as the \'unused'' fundamental state variable and suggests the relation


..

aQ,
..

( 6.3)

(/ 5.

That is, the reciprocal of the ahsoJute temperature is an integrating factor that
permits the replacement of the inexact differential aQ, by the exact differen
tial dS. Equation

(6.3) is in fact Clausius's definition of the e11tr(JPY S. The term

is derived from the Greek word en-+ trepei11 for turning,, or Hchanging.
When we substitute Equations (6.2) and (6.3) in Equation (6.1) we obtain

( 6.4)
for a reversible process. Thus we have succeeded in \vriting the first Jaw
entirely in terms of state variables.
There

are

(I)
(6.4)

t wo remaining concerns:

exact differential: and (2) Equatio_n

\\'C

have not proved that dS is an

appears to hold for reversible

processes only. This would seem to indicate that we have replaced a com
pletely general statement of the first law with one that is restricted to a special
case. Both of these problems will be addressed later in this chapter.

Sec. 6.3

Irreversible Processes

89

6.3 IRREVERSIBLE PROCESSES


We have already encountered irreversible processes in discussing dissipative
work. Consider the following processes and the results observed in nature:

1. A battery will discha:-ge through a resistor, ieleasinog energy (Figure 6.1 ).


The reverse will not happen: adding energy to the resistor by heating will
not cause the battery to charge itself.
2. Two gases, initially in separate adjoining chambers, will mix uniformly
when the partition separating the chambers is removed (Figure 6.2). The
gases will not separate spontaneously once mixed.
3. A gas in one portion of a chamber is allowed to undergo a free expan
sion into an evacuated section of the chamber (Figure 6.3). The gas will
not compress itself back into its original volume.
4. Heat flows from a high temperature body to a low temperature reservoir
in the absence of other effects (Figure 6.4). The reverse process does not
take place. If T2

>

T1 initially, the body comes to temperature T1

.. . . . .
'.. . .. .. . ...
. . ..
. . . ...
.: : : : ;:

..______,/ I

Figure 6.1

Battery

discharging current through a


resistor in a medium to which
heat can be supplied.

. . . .'
: :. : : < .

:.:. . ::::.'
. .. :

...
:' : :- ;. .
. .. .. . ..
. . ...
. =: : .: .
'. : ; : .
. . . ..
0 . ..
.' : . .
0 .. .
. .
. . .. .
. :-:- ..
:.0.:.::
. '..=
-.
. 0ooo 0 oo
. .. .

) :

::: ':
.. . .. .
....
.. ... . .' .
. . ...
;..:-:
'
.
.:-;
::.:. ..:-:
. ..... :(...
. .... ... .
... :"%"
. .....
. ..
.:...
vv .
.
.

.
.

..
.

.
...
.......
. - . ..

:( -:.
.._ . . .
.. .
. ...
. . -.-.
:. . >:. .,.. ... '-'1' , . . . .. ..
....... .. .. .
. ,...
... .....
, .. , . . . . .... . . ..., ....
,. :. . .
.
. ...
..
.
.
.
...
.
.
,

..

.
... ........

'

.
.

.
.
.:

..
.
......
.
.
.
..
.

.
.'
.;-''
'
'
.

...

:
:
.
;

.:.
.
.
.
.

.
..

...
.

-...
.

,
...
.

-:
.

;
..

;
-:-:

.
-:
.

.
.
.

.
..
.
.

.
.
.
.
.
,
..
..
.
.
. ... .. . . ..
. .. . .......
.. .. ''""'
.. . . '. ' . ..
. .. . .. '\' . . . .. ... . . . . . . ..
. . . . .
.. ..
' .'. (...
.
.. . . .
. . ... ...... .. . ...
',.. ....... .... , .
.
.

.
.
..
....

.
.
.
.

,.
.
.
.
)'
....

,
.
..

.
.
. ..
.

.
.
,
'
.

.
.
...
.

'
.

)'
'
.
.
.
.
.


.
.
.
-

.
.
.

:
:
:
:
.: :: ;:j?,;= :;
: :::.:: ::
: : : ::=: : -::=::: ::;-=:::::::
: : . : :::: ::::;. :.:;:.: ? :::.:.:::-: :;:::::::: :( :_:_; . = : :
::

; {
,

::

:; ; ;.:::: : :

'

::: :!:-

. .. . . . . . ... .. . .. . .. . .. . . . - '. . .. ...


. ..
..
. .. ..
..._;;. : :-:. : -: . ;. :-. : :.. : . : .. : . . . ;. . . :-:- .: . . . . : : :-:-: :- : : : . : . .. '. >. : .-: -:-:
.
.
.

..

. : . . :: : / , ...:.-: ::-: . :: . . :: : :-:-:-:: . . ::=:- -: :: :.;:: : : :. ... :: -::.:, ; ; , :;. ;::. :. : .::: :: -:: :> ::
.
. .
0 . . : o , . .. . ..
..
. . . . .. . .. . . . . . .. . ...
. . ... . .. . .. :
, "

-: -:X : ' .: ..... :.' v


' ,: : . :- :- : -: :-:-: :: : : -: :: . : :' :>:.-:. .. ; . : : -- :-.- : -: .:.:
7
:<
:
: .' . -:-: .
:-:
.'
.
-:
:
Y
-;
; :: . :
:

:
:
.
.
.
.
.

.
.
.
.
. . .. : "" .. ., .-..: . . ... .... .... . .... . . :: ::.;.: ::
.
. >. . : :-: .: ;... : : -- :- .. : :. -: . .;. - : .....

::_;.
.::::::. .:
. o:o::::.
: : :0:0
.;.;
': :
:. ...
. .
..... -...
: :-:.: -'ooo
.
0 0
.. .
.. o

:-:.

'oO .'o0o .0
_
. . ....... .

:;

. .. . . .
. . ..

. .
:
: . 0 . 0:
0.: 0 . .0:.
.
o o

.o,o. o:o,O o. :o
O
o
ooooo:-oo ,

::>: ):

'
:-:- > .
: ' ...:'-!:-:
.. 0 . . 0. . 0
o 0
0 0o
o' ooO
o
o

0 .

<:.<}0

0
X

=-o.:oooo.o:

.o

"

::
)oo>:oooo o

....:-.: .:.;.
,..
:
:: :=
;:
:
..... . . :.
k-.<
7 : ..
.--: o :0:;,.,.
..
, o:.
.; . "<
.. .
::: . :
0 oo0 -o :
. . 0 >. :/

;:::
:. :0::::: 0
0
0 0

:0. .

. . ..

;:

:S ::;:

o:' :00 0-:0 .

.
:;::::
:
; .
..
. ..-.., .. .....
. .. .....
.....
.
.
.
..-
.
.
.. .. ...
.. .
, .. ..
. .. .
. . . -:

-:
:-::
. . ....
, ........: .. . ..
. . ... . . . . . . .. .
.
. .. . .. .-.. .-. . -. . .. . . ., . . . .
. . . ....,...:-- . ;,;.
.
.. . .
.

:
.
{

.
.
.
.
:
:: ::; =} ..-:: -::: ;: .:: :: : ' .::.: .:::;:: : .;. , :;.::, : : ;. : : :. '::: ::::-::.':' ;: ::.
.: :: :: :. :::.. .::;: - . : -: ::::: :::::::;::}
.; . :::-: :.: :: ::::< :
.
.
.
.
.
.
.
.

.
:

.
,
.
..
.
.
.
.
..
.
.
: .;. . . : ... .; ,. ....
,. : .
. . . ....-.. .. ...... .. .-:
:
..
. . ,.;..; .....,. .. . :.- . .
. .. ........
... .
. ..
. , ..
.....:...:....... . .;.
-.;
.. .....;.....
. ...
,.
...,.,..:.. : .... X..:;:. .". .
: . . . ,.7 . ......,..
.
.

. ..
'
.
.

::: ' .. : ;.t,;- . i : : .; : ;-: - ; ;- .: : . : -:- : ::. :: :: ;


- : ; . . <' ; :- : .: :- = :- : : :- : ..; . .. . -:-:-..: -: .:-. = ; : : : :.::: _,-:- ;- .: : : :.::: ::: : -::- . : ; :;;.

..

(a)
Fig111e 6.2

:.

(b)

1\vo different gases, (a) before mixing, and (b) after mixing.

Sec. 6.4

Camot's Theorem

91

T
Figure 6.6

Schematic
diagram of a device forbidden

by the Kelvin-Planck

statement of the second law. It


is impossible to have W

Q.

6.4 CARNOT'STHEOREM

We have said that Carnot argued that efficient engines must operate as closely
as possible to a Carnot cycle. Using the Clausius statement of the second law,
we can prove Carnot's theorem which states:

No engine operating between mo reservoirs can be more efficient than a


Carnot engine operating between those same two reservoirs.
We consider two engines, M (Carnot) and M' (hypothetical), and
assume 17'

>

11 That is,

(6.5)
The engines are depicted schematically in Figure

6.7.

Since a Carnot engine is reversible, it may be driven backward (as a

refrigerator) by the work from M'. We can arrange the position of the adiabats
T,

...

M'

r,
(a)
Fi811re 6.7

(b)

Two engines: (a) a Camot engine M and (b) a

hypothetical engine M' with efficiency assumed to exceed that


of the Carnot engine.

92

Chap. 6

The Second Law ofThermodynamics

Q,

M'

"'

M
'

Figure 6.8 The two engines


of Figure 6.7 harnessed
together, with the hypothetical
engine driving the Carnot
engine configured as a
refrigerator.

so that in one cycle of each engine M uses exactly as much work as M' pro
duces. Thus I WI
I W' 1. The Hcomposite" engine is shown in Figure 6.8. Paying
careful attention to signs we note that
=

W'l1'I,
I - ;Q'I-IQ
I
2

and

or

The equality

f WI

j W' I therefore gives

or

Taking account of the signs in Equation (6.5) and noting that

; W' i, we

see that jQ2 > IQ2'1 and therefore Q11 > Q1'j. We conclude that the compos'
I Q 1 I from the cold reservoir
ite engi1;1e does no work but extracts heat I Q1
and delivers an amount Q2 - Q2': to the hot reservoir. This conclusion vio, lates the Clausius statement, so our hypothetical engine cannot exist. It fol-

lows

77'.
We can show that if the Clausius statemnt is violated, the Kelvin-Planck
statement is also violated. In this sense the two statements are equivalent.

that TJ

>

Sec. 6.4

Carnot's Theorem

93

M'

(a)

(b)

Figure 6.9

(a) A composite engine in violation of the Clausius


statement; and (b) the equivalent engine, in violation of the
Kelvin-Planck statement.

Referring to Figure

6.9, we consider

a hypothetical machine M' that, in viola

tion of the Clausius statement, extracts heat from a cold reservoir and delivers
the same amount of heat to the hot reservoir (no work is done on the
machine). The Carnot engine M absorbs heat Q2 from
work

W, and rejects heat Q1 The work done is lV

ite engine,

no

Q2

the hot reservoir, does


+

Q1 For the compos

heat is exchanged at the cold reservoir, but heat Q2 + Q1 is

extracted from the hot reservoir and an equal amount of work is done. This
violates the Planck-Kelvin statement.

Continuing the proof of Carnot's theorem, we have shown that 11 > 11'
Suppose that the hypothetical engine were reversible. Then
TJc

>

(6.6)

1J,,

where the subscript C denotes the Carnot engine and

the reversible hypo

thetical engine. Since both engines are reversible, the Carnot engine could
have been used to drive the other engine backward (as a refrigerator), attd we
would have the result
1Jc
If both Equations

<

TJr

(6.7)

(6.6) and (6.7) are1o be satisfied,


TJc

'Ylr

(6.8)

Chap. 6

94

The Second Law ofThermodynamics

That is to say, all reversible engines operating between the same reservoirs
have the same efficiency 11 = 1

T1/T2 Irreversible engines will have a lesser

efficiency. This is Carnot's theorem.

6.5

THE CLAUSIUS INEQUALITY AND THE SECOND LAW

For a Camot cycle we have seen that (Equation

5.20):
(6.9)

The quantity Q/T is known as the Carnot ratio. From this we deduced the effi

ciency 'T1 = 1

T1/T2 and proved that the efficiency of all reversible cycles has

this value. Applying the Carnot ratio to an infinitely narrow Carnot diagram
(finite temperature difference but infinitesimally small quantities

of heat

extracted and rejected), we obtain

( 6.1 0)
We consider an arbitrary reversible cycle and represent it as

continu

ous contour in a P-V diagram. We replace the process by infinitely narrow


Carnot cycles (Figure

6.10).

The fact that the continuous contour is replaced by a sequence of infini


tesimal saw-toothed steps is not problematical: the adjacent adiabats cat1cel
each other, leaving only the contributions at the boundary. By an extension of
Equation

(6.10), we obtain
,...._

l!Q,

l!Q, = 0

_._....

( 6.11)

Figure 6.10

P- V diagram of a

reversible cycle. The process is


replaced by infinitesimal
Camot cycles.

L....-..--------.

Sec. 6.5

The Clausius Inequality and the Second Law

95

where the integration is carried out over the whole corttour. The subscript

emphasizes the reversible nature of the cycle. Since the integrand is the differ
ential

dS of the entropy, it instailtly follows that dS is an e.tact differential and

S is a state variable.
Consider next an irreversible cycle. An irreversible Camot cycle that
operates between the same temperatures and produces the same work has a
smaller efficiency 11' than the efficiency 71 of a reversible Carnot cycle. That is,
11' < TJ, which implies that

or

(6.12)
For an infinitely narrow Carnot diagram we have

instead of Equation

(6.10). Following the

same reasoning as before, it is clear

that for an arbitrary cycle that is partly or wholly irreversible, we must have

'
Q
l!
T

0
.

<

(6.13)

Combining Equations (6.11) and (6.13) we obtain the famous Clausius

71Q

<

'

inequality

(6.14)

0'

where the equal sign holds for reversible processes. Equation

(6.14) is some

times taken as a statement of the second law.

Finally, we consider the change in entropy in an irreversible process. Let


2 be an irreversible change and 2-+ 1 be any reversible path connecting

the two states in the P- V diagram of Figure

ttQ
-

2aQ
\.

lzt

6.11. Then

Equation

(6.14) gives

The Second Law ofThermodynamic

Chap. 6

96
p

_,

__

...

I
I
Figure 6.11

An irreversible
change (dashed line) and a

reversible path connecting


states 1 and 2.

---- ----

Chang ing the order of integration and the sign of the second

27/
Q

\.

<
-

term we h ave

F
y

( irrc:vcrsihk)

Thus

or_ in

differential

form.

( 6.15)

Again, the equality

i gn

holds for

reversible process, and the inequali ty

appropriate for an irreversible p rocess

Suppose now that the sys t e m is is()]ated. Then

5'
for

a finite process. We

..,

is

()

aQ

0 an d d5;

0 or

( 6.16)

(isolated system)

conclude that

The entrop) (Jf an isolated s_vstetn increases in any irreversible prlJCess

and is unaltere(i i11 any reversible pr()Cess. This is the principle o.f increas-

1ng entropy.

It is to be noted that this statcn1cnt refers to net entropy changes. It does

not say that the entropy of part of the system cannot decrease. If. for example.

lower

heat flows from body A to body B at a


contained in an adiabatic enclosure, 5

ofaQA), but

uSA+ 58 will st il l

..

temp era t ure with bo t h bodies


,

is then nega ti v e ( becau se of the sign

he posi ti v e ( figu re 6.12).

Sec. 6.6

Entropy and Available Energy

97

Figure 6.12 Two bodies in


thermal contact in an
adiabatic enclosure.

adiabatic container

The adiabatic enclosure contains everything that interacts during the


process and we can define this assembly as our universe. (This is not to be
confused with the real universe, which may or may not constitute an isolated
system.) If.. then, our universe consists of a system and its su rr oundings, it fol
lows that

Suniverse

Ssystem + LlSurroundings

>

0.

(6.17)

The fact that the entropy of an isolated system can never decrease in a
p rocess provides

direction for the sequence of natural events. The laws of

mechanics are second-order equations in time t and are unaltered by the


replacement oft with - t. As far as these equations are concerned, all physical

processes can run backward as well as forward. Clearly this is not so; the story

of Humpty Dumpty illustrates that processes go only in the direction of


increasing entropy. It is for this reason that the law of increasing_ entropy is

often described as p roviding "the arrow of time'' for the evolution of natural
processes.

6.6 ENTROPY AND AVAILABLE ENERGY


Another way of ex p ressi ng the results of the second law is in te1tns of what is
known as available energy. It is impossible to utiliZe all the i n terna l energy of a
body for the production of mechanical \Vork, since work c;an only be obtained
by extracting heat from the body and giving it to an

e n g i ne

whose efficiency is

less than unity. Suppose that the temp er a ture of a body is T2 and that a

reve rsible engine works between T2 a nd T1, which is tt1e tem perature of a large
reservoir to \Vhich the engine can give up hea t and is also the lowest ten1pera
ture available (Figure 6.13). Then, if the body gives an amount of heat

the engine, only part of it

namely aQ( 1

T1/T2)

aQ

to

can be converted into

mechanical work. If the engine is irreversible, still less work can be obtained.
The available energy is defined to be liQ( 1
energy is T1llQ/T2.

T1/T2), while the unavailable

Chap. 6

98

The Second Law ofThermodynamics

dQ
M

dW

Figure 6.13

Diagram
showing the amount of heat
available to do mechanical
work:ltW
ltQ(l
T1/T2).
=

T1

Corresponding to the principle that the entropy always increases in a


spontaneous proess there is the principle that the available energy always

decreases in an irreversible cvcle. We could state that

.,

There exists no process that catz i1zcrease the available energ}' in the universe.
The property of increasing entropy is eqttivalent to saying that energy is

always being degraded into forms that are more and more difficult to use for
the production of work.
The significance of entropy becomes clearer when we consider heat to
be the energy of molecular motion. The lack of complete availability of that
energy for the production of work is due to the randomness of the m olecular

motion. It is impossible to reduce the motion of each molecule simultaneously


to zero by the action _of forces acting on the hody as a

whol e

and so it is impos

sible to extract all the heat energy from a body Thus the property of increas
.

ing entropy means that the molecular motion of an isolated system always
tends to become more random, and the entropy can be thought of as a mea
sure of the Hrandomness" of the internal motion of a system. This connection
between entropy and randomness will be made quantitative by means of sta
tistical thermodynamics.

6.7 ABSOLUTETEMPERATURE
The Carnot c ycle together with Carnots theorem can serve as the basis for
d efin ing an absolute temperature scale. Carnot's theorem shows that the ratio

Q1jQ2 has the same value for

all reversible engines that operate between the

same temperatures. The special fact about a reversible Carnot cycle is that the
efficiency is independent of the nature of the working substance. It can there

fore be used to define an absolute scale of temperature as follows.

Sec. 6.7

Absolute Temperature

99

Let 8 denote an empirical temperature, based on the ex


pansion of an
ideal gas. Divide a Carnot C}'Cie into tw o subcycles, each using
the same mater

ial (Figures 6.14 and 6.15).


Let the path a -+ b in a 8- V diagram represent an isothern1al expansion
during which heat Q2 flows into the system. Similarly, c -+ d is an isothet1nal

compression during which heat Q1 flows out of the system. For the cycle abcda,

a
v
t

6.14 Representation in
the 8- V plane of a Camot cycle
divided into two sub-cycles.

Bt

""
-

,._,

----

------------

L----- v

e.l

Figure 6.15 The tw Ca ot


engines in series dep1cted tn
the 8- V diagram of Figure 6.14.

The Second Law of Thermodynamics

Chap. 6

100

f is an unknown function of the two temperatures 91 and 82.


lenge is to find f. To do th is consider cycle ahefa. For this subcycle,

where

Our

chal

S im ilarly for the subcy cle fecdf,


,

Thus

or

Because the right-hand is a function of fJ1 and 82

only, the

left-hand side must

also depend on these variables alone. This is only possible if

and

Then

independent of the properties of any given substance. Here

<I> is another

unknown function. Since the scale of temperature is arbitrary.. we can introduce


a thermodynamic temperature scale using <P itself instead of 8. Kelvin suggested

Ac/> ( 0)

constant,

( 6. 1 8)

so that

IQ
i

I :

TI

---

! Q,!
,

- I

T_,

( 6.19)

This ratio is the same as the ratio of the temJ)eratures obtained in our analysis
of the Carnot cycle.. in which the working substance is an ideal as.

Sec. 6.7

but

Absolute Temperature

101

The problem remains to determine the function cP ( (}) . The proof is long,

IS

worth the effort. Using the definition of the entropy, we can write the

first law in the following form (in terms of intensive variables):

ds

l,(du + Pdv).

(6.20)

This equation is true in general, as wiiJ be discussed in Section

and v to be the fundamental independent variables and let u

dtt

au
aT

dT +
v

Substituting this expression in Eq uation

ds
But also,s

dU

1
=

dT +

T iJT

au
av

dv.

6.8. We take T
u(T, v). Then
(6.21)

(6.20) we have
1

au

dV

+
T

P dv.

(6.22)

s(T, v), so that

ds
It follo,vs from Equations

as
aT

(6.23)

dT +
v

(6.22) and (6.23) that


au
T aT
1

as

iJT

(6.24)

'
v

and

as

au

dV

av

(6.25)

+P.

Now

a
av

as

--

iJT

we can therefore differentiate Equation


important relation

a as
aT av

'

(6.24) with respct to v nd Equation

Combined First and Second Laws

Sec. 6.8

101

where A' is a constant of proportionality. An appropriate choice of A' makes 8

equal to T; we can then use T to represent both the empirical temperature and
the equivalent thermodynamic temperature. The definition of the Kelvin scale
is completed by assigning to T1 in Equation (6.19) tlte value of 273.16 K .. the
temperature of the triple point of water. For a Carnot engine operating
between temperatures Tand 71, we have

IQf

(6.30)

The smaller the value of Q, the lower the corresponding value T. The smallest
value of

is zero and the corresponding value of T is zero, called absolute

zero. In a Carnot cycle, heat is transferred during the isothermal processes.

Hence, if a system undergoes a reversible isothermal process without heat


transfer, the ten1perature at which the process takes place is absolute zero.
This is a fundamental definition of absolute zero.

6.8

COMBINED FIRST AND SECOND LAWS

We saw early in the chapter that for a reversible process,


dU

where

llQ,

71Q,

T dS and 7/W,

ltW,

T dS

(6.31)

P dV,

P dV. In its most general fortn, the first law can

be written
dU

aQ - aW

The second law states that T dS

aQ,

>

(general).

aQ

(6.32)

for an irreversible process. We

can write

llQ,
where

ZIQ

(irreversible),

(6.33)

is a positive quantity. Substitutin.g this in Equation (6.31) gives


dU

aQ

aW,

(irreversible).

Comparing Equation (6.34) with Equation (6.32) we see that

(6.34)

aW

<

aW,

or

more specifically,

aW

aW,

(irreversible).

(6.35)

This is exactly what we expect: the total work done by the system is the
reversible configuration work (the useful mechanical work) plus the (negative)

The Second Law ofThermodynamics

Chap. 6

104

dissipative work associated with frictional forces. The latter appears as heat in

Equation (6.33). Therefore,


dU

==

aQ- aw

aQ,

(6.36)

(aW,- e).

or
dU

(general).

T dS - P dV

(6.37)

Evidently Equation (6.37) has the same universality as Equation (6.32). Its
just that

7/Q is

identifiable with TdS

and

t1W w i t h PdV

only for

reversible

process. In fact. Equation (6.37) is not restricted to a process at all: it simply


expresses a relationship among the state variables of a system and the differ
ence between the values of these variables for two neighboring equilibrium
states.
Consider two irreversible pr<)cesscs. For the free expansion of a gas.
al\1

==

0 but PdV is finite. S imil arly

ztQ

==

0 but TdS has a nonzero value (the

entropy increases in the process). For adiabatic stirring, aQ 0 but TdS =f. 0
and dS > 0. Also. dV
0 hut aW =f. 0, since stirring work is done.
Equation (6.37) is by far the most important relation in classical thermo
=

==

dynamics.

PROBLEMS

6-1 Jt\ (arnot engine operates


to be an ideal gas. Take y
rn u m

volun1e (c anJ

on

I kg of rne:thane ( CH-l), \vhich we sha1l consider

1.35. The rat10 of the maximum volume to the rnini


th e <arnot cycle Jiagran1 of Figure 5.7) is 4 and t h e

a on

cycle efficiency is 25 percent. Finu the entropy increase of the methane during
the isothertnal expansion.

62 Find the chn!gc in entropy of the system during the follo\ving processes:
(a) i kg of \Vater is heated reversibly by an electric heat1ng coil from 20C to
80 c ( c p
l cal g I c c : =: 4. 1 8 X ] 0 3 J kg - I K I ) .
(

==

..

(b) 1 kg of ice at ooc and 1 atn1 pressure melts at the same temperature and
5
pressure. (The l aten t heat of fusion is 3.34 x l 0 J kg- 1.)
(c) 1 kg of steam at lOOoc and 1 atm pressure condenses to \Vater at the same tem
perature and p r e ss u r e (The latent he a t of vaporization is 2.26 x 106 J kg- 1.)
.

6-3 The Jow temperature specific heat

of

a diamond varies with temperature

according to
3

kilomole -I

K -l

.
.

'

Problems

105

where the Debye temperature 8

2230 K. What is the entropy change of 1 g of

dia m ond when it is heated at constant volume from 4 K to 300 K?

(The

atomic

weight of carbon is 12.)

6-4 An electric current of 1 A flows far 10 s in a resistor of resistance 25 oh1ns. The


resistor is submerged in a large volume of water, the temperature of which is
280 K. What is the change in the entropy of the resistor? Of the water?

6-S A thermally insulated resistor of 20 ohms has a current of 2 A passed through it


.

for 1 second. It is initially at 20C. The mass of the resistor is 5 g; cP for the resistor is 850 J kg -l K -I.

(a)
(b)

What is the temperature rise?


\\'hat is the entropy change of the resistor and the universe?

(Hint: In the actual process, dissipative \Vork is done on the resistor. I magine a
reversible process taking it between the same equilibrium states.)

6-6 An inventor claims to have developeq an engine that takes in 108 J at a temper
ature of 400 K, rejects 4 x 107 J at a temperature of 200 K, and delivers 15 kilo
watt hours of mechanical work. Would you advise investing money to put this
engine on the market?

6-7 Derive an expression for the entropy of an ideal gas

(a)

As a function of

T and V.

(b) As a function ofT and P.

Assume that the specific heats of the gas are constants.

6-8 An ideal monatomic gas undergoes a reversible expansion from specific volume
v1

to specific vo l ume

(a)
(b)
(c)

v2.

Calculate the change in specific entropy

s if the expans i on is isobaric.

Calculate as if the process is isothermal.


Which is larger? By how much?

6-9 A kilomole of an ideal gas undergoes a reversible isothermal expansion from

volume of 5 liters to a volume of 10 liters at a temperature of 20C.

(a)

What is the change in entropy of the gas? Of the universe?

(b) What are the corresponding changes of entropy if the process is a free

expansion?

610

(a)

Show that for reversible changes in temperature at constant volume,

cv

(b)

T(iJsfiJT)v.

3
bT for

aT +
a metal at low temperatures. Calculate the
Assume that C0
variation of the specific entropy with temperature.
=

6.11 Consi er a van der Waals gas.


Show that cv is a function ofT on ly. (Hint: use Equation (6.26)).

(a)
(b)

Sho w that the specific internal energy is

(c)

Show that the specific entropy is


S ==

Cv
T

dT

+ R In

(V

b)

+ S0

Chap. 6

106

The Second Law ofThermodynamics

6-12 When there is a heat flow out of a system during a reversible isothermal process
the entropy of the system decreases. Why doesn't this violate the second law?

613 One kilomole of

monatomic ideal gas is

cycle shown in Figure 6.16. Here PJ

==

around the reversible closed


3
3
10 atm. V1
2 m , and V2
4m

ca rr ied

Calculate the change in entropy for each leg of the cycle and hence show that
the entropy change for the complete cy cl e is zero.

p
I

isochore ---r-

adiabat

---..
-..

Figure 6.16 P- V diagram for


Problem 6.13.

isobar

7 .I

Entropy Changes in Reversible Processes

I 09 .

7.2

Temperature-Entropy Diagrams

I I0

7.3

Entropy Change of the S_urroundings for a Reversible


Process

II I

7.4

Entropy Change for an Ideal Gas

I 12

7.5

T he Tds Equations

I 13

7.6

Entropy Change in Irreversible Processes

118

7. 7

Free Expansion of an Ideal Gas

121

7.8

Entropy Change for a Liquid or Solid

122

107

7.1 ENTROPY CHANGES IN REVERSIBLE PROCESSES


.

To illustrate the calculation of the entropy change, we first consider reversible


processes. In terms of specific quantities, the first law for a reversible process is

aq,

du

(7.1)

Pdv,

or

aq,
T

du
=

T + T

dv

ds.

(7.2)

We shall examine some special cases.


1. Adiabatic process: aq,

0, ds

0, s

constant. A reversible adiabatic

process is an isentropic (constant entropy) process. We note in passing


that an irreversible adiabatic process is not isentropic.
2. lsothertnal process:

2
ztq,
1

q,

(7.3)

3. Isothertnal (and isobaric) change of phase:

(7.4)

Here lis the latent heat of transformation .


.

4. Isochoric process: We assume that u


v

= constant in an isochoric process, u

du

u( v, T)

in general. Since

u(T), as for an ideal gas, and

cvdT. Thus

Sz - St

2
I

Cv

dT

.
109

Applications of the Second Law

Chap. 7

110

If ct, is constant over the temperature range

T2

T1 we have

(7.5)
S. Isobaric process: It is convenient to use the specific enthalpy in an iso
baric process. Since

d/1

du + Pdv + vdP.

(7.2) yields

Substitution in Equation

h h(P. T) in general. Since P =constant here, h


only and dh
cpdT. Then

Assume

==

h(T)

Cp

if

.
.

.
.

Cp

: _ .. , > _

Cp)n

=:

T2

(7.6)

is constant.

dT

.
.

. .

:
.

.
.

7.2 TEMPERATURE-ENTROPY DIAGRAMS


The total quantity of heat transferred in

reversible process from state 1 to

state 2 is given by
2

(/,

Ttl,..

This equals the area under a curve in a T-s dtagram. For a reversible iothermal
process Tis a constant and the curve is a horizontal line. For
batic (and isentropic) process. dJ

reversible adia

{J if T =I= 0 and the curve is a vertical line.

Consider a T-s diagram for a Carnot cycle (Figure 7.1 ). Recall that a--+ b
is an isothermal expansion b
..

compression, and d

an adiabatic expansion. c

an isothermal

an adiabatic compression. lt follows that the T-s dia

gram is a simple rectangle for a Carnot cycle. The area under the curve is

Sec. 7.3

Entropy Change of the Surroundings for a Reversible Process

III

Figure 7.1

r)

T-s diagram for a

t----

--

Carnot cycle.

Since f du

0, it immediately follows that


w =

T ds

q,2

q,,.

7.3 ENTROPY CHANGE OF TH E SURROUNDINGS FORA


REVERSIBLE PROCESS

In every process in which there is a reversible flow of heat between a system


and its surroundings, the temperatures of the systetn and the surroundi;tgs are
essentially equal, differing only by dT. The heat flow Ollt oftlte surroundings at
every point is equal in magnitude and opposite in sign to the heat flow into the
system (Figure

7.2). Thus
dssystem

dssurroundings

==

dsuniverse

But

7lq,
=

T + dT

surroundings

surroundings

( lts ) surroundings

surroundings
T+ dT

Figure 7.2

Heat flow into a


system from its surroundings
in a reversible process.

system
T

dq, > 0

Applications of the Second Law

Chap. 7

112
Thus

Ids I surroundings

I tis luniverst!

and

- ! tis ystcm

==

o.

In any reversible process, the entropy change of the universe is always


ero. TI1is means that any change in entropy of the system will be accompanied
by an entropy change in the surroundings equal in magnitude but opposite in

sign. Entropy is conserved in

a rev e rs ible

process. Of course, onl y

idealized -

processes are reversible; all natural processes are irreversible and entropy is
not conserved in general.

7.4 ENTROPY CHANGE FORAN IDEAL GAS


With liu

==

ccdT in Equation

(7.2), we have
(7.7)

for a reversible process. For an ide_al gas P/T

ds

==

c,.

tlT
T

dv
v

R/v. so
.

Integrating, \Ve have

+ R In

if

cr

v'"'

(7.8)

is constant. In any thermal process. (lnly the cl1ange in entropy is impor

tant. In chemistry, however it is important to be able to assign an absolute


value to the entropy. Statistical thermodynamics will make such an assignment
possible.

(7.8) holds only for an ideal gas, it has characteristics


typical of practically all solids. liquids, and gases: ( 1) if only the temperature is
varied. the higher the temperature rise the greater the increase in entropy: (2)
Although Equation

if only the volume is varied, the l a rg e r the volume expansion. the greater the
entropy tncrease.

The implications of these results are important. Consider the isentropic


expansion of a gas. An increase of volume provides a positive contribution to
an- entropy change. From a molecular point of view, the available energy levels
of the system become more closely spaced as the volume increases and the mol
ecules tend to become more randomly distributed among them. The entropy

The Tds Equations

Sec. 7.5

II 3

tends to increase. To keep the entropy constant, there must

be a compensating

negative contribution to the entropy change. which is provided by a decrease in


temperature. The drop in temperature that takes place when a gas expands adi
abatically (and reversibly) may therefore be regarded as an effect that is
needed to offset the volume increase, in order to keep the entropy constant.

7.5

THE Tds EQUATIONS

From the combined first and second laws, expressed as

Tds

du

Pdv,

(7.9)

we can obtain some powerful results known as the

''Tds

equations." They

involve writing the specific entropy as a function of two independent coordi


nates, that is, two of the three fundamental state variables

P, v, and T. The

equations are:

(7.10)
av
a

aT
av
The
ferred in

(7.11)

aT
aP

Tds equations
T

have a variety of uses:

and integrating;

(7.12)

reversible process (aq,

dividing by

cJ<

cp

(3)

(1)

they give the heat trans

Tds); (2) the entropy can be obtained by


the equations express the heat flow or

entropy in ter1ns of measurable properties such as cp, {j,

K,

T, etc; (4) they can

be used to deterntine the difference in the specific heat capacities cp and cv;
and

(5) the equations can provide relations bet\veen pairs of coordinates in a

reversible adiabatic process in which

ds

0.

Th! derivation of the Tds equations is straightforward. The key is noting


that the entropy is

state variable whose partial derivatives satisfy the condi

tion for an exact differential. The first

Tds equation was essentially derived in

Section 6.7 in the steps leading to Equation (6.26), from which the final result
can easil}' be found. Here we shall derive the second
the derivation of the third as a problem.
Let

T and P be the independent variables. Then


Tds

du + Pdv.

Tds equation and leave

Applications of the Second Law

Chap. 7

114

The enthalpy ish

+ Pv, so that

du

dh - Pdt\ - vdP.

Thus

Tds

dh - vdP

rill
=

CIT

dT +

or

ds

With s

ah

iJT

ah

aP

dT
P

dP.

(7.13)

s ( T, P). we have

ds

as
=

dT

aT

as

aP

dP.

(7.14)

Since T and Pare independent._ it follows that


as

1
-

aT

ah

(7.15)

AT

and
as
iJP

ah

riP

(7.16)

v .

The differential ds is exact. Therefore we can equate the mixed second-order


partial derivatives of s:
a
aP

a2 s

as
iJT

p T

aP aT

Substituting Equations (7 .15) and

--

aT aP

as

iJT

aP

T p

(7 .16) atld carrying out the differentiation"

we get
1

a2h

a2h

T iJP aT

av

T aT aP

ah

-v.

--

iJT

aP

Sec. 7.5

The Tds Equations

liS

Two of the terms cancel, and we have

ah
=

iJP

-T

av

iJT p

v,

(7.17)

analogous, incidentally, to Equation

(6.26).
cp. Using this and Equation (7.17)
For a reversible process, (ahjiJT)p
=

in Equation (7.13), we obtain the result

dP.
Finally, since

(avj iJT) P

v{3, we have
T ds

cptlT - Tv/3 dP.

As an example, suppose that one kilomole of an ideal gas undergoes a

P1 to P2 We wish to find the


quantity of heat transferred in the process. For an ideal gas, {3
1/T. Thus

reversible isotherrnal change in pressure from

Tds

cP.T. - vdP

RT
=

dP.

Then

T ds

q, =

P2dP

- RT

P,

J'

- RT In

P2

}'1

Heat is absorbed if P2 < P1 and evolved if P2 > P1


An important application of the Tds equations is the determination of
cp - Cv, the difference in the specific heat capacities of a given substance.
Equating the first and second Tds equations, we have

Cp dT - Tvf3 dP
Solving for

dT

dT, we get

TJ3
dv
K(Cp - Cv)

cvdT

dv.

(7.18)

Tv/3
dP
(Cp - Cv)

aT
av

dv
P

aT
dP.
aP v

Applications of the Second Law

Chap. 7

116
Thus

aT

T{3
--

and

Tv{3
= --( cP c )

ciT

--

aP
Solving the first equation for

l'

cp - ell

and using the reciprocal relation gives

or

(7.19)

The equation for


since

Cp

(aTIaP) l: gives

the same result. This equation is noteworthy

is measured in experiments.. whereas

cv

be calculated from theory the equation affords

is difficult to measure but can


a

comparison of the two. Also..

the quantities on the right-hand side of the equation are all positive, showing
that

cP

cP

ct.

is always greater than


=

It is left to the student to show that

c r

R for an ideal gas.

Equation (7.19) can be used to calculate the specific heat capacity


.

of a solid. As an example, let us calculate


ph eric pressure,
K =

9.5

Cp

29

c,.

for copper at 1000 K. At atmos

103 J kilomole -l K-

1 ..

{3 = 6.5

10

V
n

nz

np

63.6 kg kilomole- 1
--

--

X 103 kg

8.96

m-3

7.1

10- m-1 kilomole- 1

Hence

Cl,

and

2
10-1 Pa-1 The specific volume in m3 kilomole-1 can be found

from the atomic weight of copper and its density:

v =

C1

Cp

= 29

2
Tv{3
-

103

29

3.2

X 10 -

103

( 100()) (7.1

26

X 10-3) (6.5 X 10-5)2

- ------

9.5 x
x

I0-12

lW J kilomole-1 K-1

117

The Tds Equations

Sec. 7.5

This high-ternperatttre value is very nearly equal to 3R, a result that will be
explored in detail in Chapter 16.

.
Finally, we consider a reversible adiabatic process ds = 0. Equation

(7.12) gives

Cp_dv = - C,J< dP

f3

{3v

(7.20)

'

or
---

OV

iJP

=K

Cv

Cp

In analogy with the isothermal compressibility K we introduce the adiabatic


compressibility

Ks = -V

Since 'Y

>

1,

Ks

<

av
aP

--

(7.21)

K; the adiabatic compressibility is less than tt1e isothermal

compressibility because the increase in pressure causes the temperature to


rise. The temperature increase, in turn, results in an expansion that partially
offsets the compression associated with the pressure increase. It folJows that in

the adiabatic case the volttme change is less and Ks is smaller.

It is interesting to note that the speed of sound in a gas is given by

c=

1
-

pK

(7.22)

'

where p is the density of the gas. The adiabatic compressibility is used since the
process of compression and rarefaction takes place very fast; there is no time
for heat to be exchanged with the surroundings.
Using Equation (7.22), we can estimate the speed of sound in air. If air is

regarded as an ideal diatomic gas, then 'Y


atmospheric pressure, and p

c=

:::=:

;;;

1.4,

K ::=

1/P

=:::

10-5 Pa-1 for

1.2 kg m -3 Thus
1.4

l/2
-

340 ms-1

This compares favorably with the measured value of 343 ms -t at 20C.

Chap. 7

118
:>

.
.

::::.

.
.
.

.
.

Applications of the Second Law

.
.

. .

.
_...;
. ::

7.6 ENTROPY CHANGE IN IRREVERSIBLE PROCESSES


How do we calculate the entropy change for an irre\'ersible process when the
entropy is only defined in terms of reversil)le heat flow? We resort again to an
earlier argument. The entropy is a state variable. and the entropy difference is
the same between any two equilibrium states regardless of tl1e nature of the
process. Thus we can find as for an irreversible process by choosi11g any con
venient reversible path from the initial to the final state and be certain that it
is equal to the change produced by the actual. irreversible path.
Consider the example of Figure 7 3 A body at temperature T, is in thetmal
.

contact with a single reservoir at temperature T2 with T2 > T1 The system is

allowed to come to thermal equilibrium with the reservoir. The process is irre
versible since there is a finite temperature difference; the process cannot be
reversed by an infinitesimal change. We assttme that the process is isobaric. Then

"IF
u

Here we are using an equation reprcse11ting the first law for

reversible

process. That is" we are effecting a simple heat exchange by bringing up a


whole series of reservoirs between r1 and 72 (keeping the pressure constant)

in suh a way that the body passes through a series of equilibrium states. Then
llq,

ds =

== c 1,

dT

so that
T.,
-

bodv

.
.

"'

TJ

...--

.
.

. .

Q
T..,
-

Figure 7.3 Body initially at


temperature T1 in thermal
contact with a rese rvoir at
temperature T2

.
.

( 12

.
.

:
.

..

.
.

r, )

>

.
.

..

. .
.

..

.
.

..
.
.
. .
. .
'
.
.
.
. .
.
.

.
.

. ..

..

..

.
.

.
.

. .
.

\ rervoir

surroundings

.
.

==

system

Sec .. 7.6

Entropy Change in lrre\ersible Processes

119

while

As far as the reservoir is concerned, the process is isothermal (the tempera


ture of the reservoir remains constant) as well as isobaric. The hea flow out of
the reservoir is

-cp(T2- T1).

Now

( S ) universe

( S ) body

( S ) reservoif'

or

(7.23)

It is easily seen that the entropy of the universe will always increase, whether
T2

>

T1 or

T2

< T1 Let x

TJT1 Then the terrn in brackets can be written

Taking the derivative of f(x) and setting it equal to zero shows that the func
tion has a maximum or a minimum at
x =

1, where the function is zero. At

1 the second derivati.v e is positive, so the extreme value is a minimum.

Hence f(x) is always positive except when T2


heat exchange at all.

T1 in which case there is no

ltl the example, the increase in entropy of thf! body is larger than the
decrease in entropy of the reservoir and the entropy of the universe is greater
at the end of the process than at the beginning. In all real changes, the entropy
will always increase; in other words, entropy is created in the process. The addi
tional entropy is an onus that the actual universe must bear. For an indefi
nitely expanding universe, there is no known process that can cause the
entropy to decrease.*
To illustrate the calculation of entropy changes in irreversible processes,
we consider the following examples.

EXAMPLE

Suppose that 0.5 kg of water at 90C is cooled to 20C, the temperature of the
surrounding room. The specific heat capacity at constant pressure of water is
*

However9 a universe that expands without limit is different from a finite. closed system.

There is no final equilibrium state and energy conservation may not apply.

Applications of the Second Law

Chap. 7

110

1
4180 J kg-1 K- The change in entropy of tl1e system (the water) is the change

that would occur if the "ater were cooled reversibly from T1


T2

293K:

(S)system

r aQ,

r!

mer

dT
=

7,

mcp

In

363K to

T,

i
1

The entropy of the system decreases. The change in entropy of the surround

ings is equal to the heat transferred to the room divided by the temperature of

the room, which remains constant. The sign is positive since the heat flows into
the room:
T

nzcr

dT =

mcp

363- 293

Therefore
(4s)universe =

(S)system +

(.\')surroundings=

-448+499 = +51 J K-1

The entropy of the universe has increased.

EXAMPLE2

4 x 105 J flows through a diathermal wall separating a high temper


ature heat reservoir at TH
500 K from a low temperature heat reservoir at
Heat Q

==

TL

200 K. The p ai r of reservoirs constituting the system is thermally insu

lated from its surroundings. Then

( s) universe= ( dS) S\'Stem

Qr
-

TH

Q,
TL

( s) reservoir H
-

( s) reser\'oir

4 X l(t5

4 X 105

500

200

-- +

= +

/.

1200 J K -I .

Note that if the two reservoirs are at tl1e same temperature .. no heat is
exchanged and

(AS)

erse

un i v

0.

Sec. 7.7

Free

Expansion of an Jdeal Gas

Ill

EXAMPLEl
A 5 kg mass falls to the ground from a height of 50 m;The temperature is con

stant at 20C. 'The process is irreversible but we can imagine the mass being
slowly and reversibly lowered by a string and pulley arrangement. Since no
heat exchange is involved,

( 4S)system

0, but
W,

mgh

We note that for the surroundings, which are ttnchanged if the mass is small,
T(S)

Q,

U + Wr

0 + mgh.

7. 7 FREE EXPANSION OF AN IDEAL GAS

We have seen that in the free expansion of an ideal gas, du


t h at aw

0 and llq

0 so

0 also. Referring to Figure 7.4, we observe that the equations

describing the equilibrium end states are

P0

RT0
=

Vo

and

A reversible isothermal process would be described by the equation


Pv

constant or P0v0

P1v1 However, the latter equation does not describe

the free exp ans i on because P1 is initially zero and the process is irreversible.

vacuum

gas

(a)
Figure 7.4

(b)

Free expansion of an ideal gas: (a) initial state; and (b) final state.

Applications of the Second law

Chap. 7

Ill

Nonetheless, the entropy change can be calculated by assuming that a


reversible, isothermal expansion takes place between the initial and final
states of the system. Thus,

ds

Cv

dT
T

+ R

dv
V

' y I

=0

so

( ds)svstem
.

v1

R ln

(7.24)

'

Vo

and

( ds) universe

(7.25)

( ds) system > 0..

since the system is isolated (adiabatically insulated). In an irreversible free


xpansion, the available energy levels become more closely spaced .. leading to
greater randomness and increased entropy.
We note that in a reversible. isothertnal expansion .. the work done would be
I

u1,

Since du

0,

w,

RT0 In

v,

'Vn

q, so that
..

.is

q,
=

Tu

R In

vJ
Vo

In a free expansion no work is done but the change in entropy in the irre
versible process is as if work

were

done in a reversible, isothermal process

between the same end points.


If the situation involves the mixing of gases

( P1

=F

0)" the problem is

quite different and will be treated in Section 9.5.

7 .8

ENTROPY CHANGE FOR A LIQUID OR SOLID

The equation of state of

liquid or solid is, to a

v = v0[1

first approximation,

P(T- T0) - K(P- P0)]

(7.26)

Problems

Ill

(see Section 2.5). The second Tds equa ti on is:

(7.27)

Setting

av
iJT

'llQ
. {)IJ'

we obtain

Integrating, we have

- voP(P - P0).

s-

(7.28)

The entropy increases if the temperature increases and decreases if the press ure Inc reases.

PROBLEMS

7-1 The latent heat of fusion of ice at a pressure of 1 atm and ooc is
3.348 x 105 J kg-1 The density of ice under these conditions is 917 kg m-3 and
the density of water is 999.8 kg m -J. If 1 kilomole of ice is melted, what will be

(a) the work done?


(b) the change in internal energy?
(c) the change in entropy?

7-2 Ten kg of water at 20C is converted to ice at -l0C by being put in contact
with a reservoir at -l0C. The process takes place at constant pressure. The
heat capacities at

constant

pressure of water

and

ice

are 4180 and

2090 kg-1 K-1, respectively. The heat of fusion of water is 3.35

lOS J kg-1

Calculate the change in entropy of the universe.

7-3 Calculate the change in the entropy of the universe as a result of each of the fol
lowing processes:

(a) A copper block of mass 0.4 kg and heat capacity 150 JK -l at l00C is placed
in a lake at l0C.

Applications of the Second Law

Chap. 7

124

(b)

The same block at l0C is dropped from a height of 100m into the lake.

(c) Two similar blocks at 1 ()(r)C and 1 ooc are joined together. (Hint: See
Problem

(d)

7-8.)

One kilomole of a gas at 0(" is expanded reversibly and isothermally to


twice its initial volume.

(e) One kilomole of

gas at ooc is expanded reversibly and adiabatically to

twice its initial volume.

7-4 Suppose that the specific heat c a pa city cp of the body discussed in Section 7.6 of
the text is 10 J kg-1 K- 1 and T,

200 K. Assume that the mass of the body is 1

kg.

(a) Calculate the change in entropy of the hody and of the reservoir if
T.,

..

(b)

400 K .

Make the same calculations for T2

==

100 K.

(c) Find the entropy change of the universe in both cases.


7-5 One kilomole of an ideal gas undergoes a free

e x pa

nsion tripling its vol um e.


.

What is the entropy change of

(a) th e gas?
(b) the universe?
7-6 An ideal gas has a specific heat given by

c,.

==

BT. wh e r e A and Bare con

stants. Show that the change in entropy per kilomo1e in going from state

( o1 T1)
..

to state ( lh T.,) is
..

;ls

A In

t
I

8(1

Td + R In

v,

l' I

7-7 A 50 kg hag of sand at 2S'C' fa ll s 10 m onto the pavement and comes to an


abrupt stop. Neglect any transfer of heat het\\'ccn the sand and the surroundings
and assume that the thermal capacity of the sand is so Jarge that its temperature
is unchanged.

(a) What is the dissipative w o r k done on the sand?


(b) What is the changt! in the internal energy of the sand'?
(c) What is the entropy change associated \Vith this U at constant T?
(The sand docs no \\'Ork as it defor1ns when it hits the pavemnt: only its
shape changes.. not its volume.)

7-8 Two equal quantities of water, each of mass

and at temperatures T1 and T2, are


adiabatica11y mixed together, the pressur remaining constant.

(a) Show that the entropy change of the universe is


. I
5
nzcp
In
-

'[I

+
1

] \
__

-:

2\/T1 T

where cp is the specific heat capacity of the water at constant pressure.

(b)

Show

th a t

.s > 0 fo r any

finite

temperatures T1 and T.

(Hint:

(a- b)?> Oforallrcalaandb.)

7-9 Tw o identical blocks of copper are held at constant vo l um e with a constant hea t
capacit y CF
380 J K 1 One is at an initiat tempe ra ture of 320 K the other at a
temperature of 280 K. The two h1ocks arc thermally isolated and placed in con
=

tact with each other. What is the entropy change of the system?

..

Problems

125

7-10 (a) Assuming that P and v are independent, show that in a reversible process,
as

Cv
=

aP

iJT

and

iJP

as
av

cp

ar

ri'lJ

(b) Use these results to derive the third Tds equation (Equation (7.12)).
7-11 Use the first Tds equation to calculate the entropy for a van der Waals gas.
7-12 Calculate the difference in the spe.cific heat capacities cp - cv of copper
4.9 x 10-5 K-1, K = 7.7 x 10-12 Pa-1, and v = 7.1 x 10-3
at 300 K. Take {3
m3 kilomole-1 How does your answer differ from the result of Section 7.5?
=

7-13 (a) Show that the difference between the isothermal and adiabatic compressibilities is
K- K.\' =

--

Cp

(b) What is this difference for a monato'.:tic ideal gas?


7-14 Using the fact t h at dvjv is an exact differential, prove that
iJ/3

riP

dK
-

iJT

.
p

7-lS (a) Show that the Joule coefficient may be written

dV

Cv

(b) Show that the Joule-Thomson coefficient may be written


J.L

iJT
aP

v
cp

(T/3- 1).

(c) Using these results,find 11 and 1-L for a van der Waals gas and show that both
are zero for an ideal gas.

7-16 Calculate the velocity of sound in liquid He4 using Equation (7.22). T'-.ke
y = 1.48, p = 162kgm-3,andK = 9.43 x 10-8Pa-1.
7-17 (a) The temperature of a block of copper is increased from T0 to T without any
appreciable change in its volume. Show that the change in its specific entropy is
.:1s =

Cp

In T,

vo{32

(T

- T0).

(b) Calculate sin units of J kg -l K -I if the temperature increases from 300 K to


-12
x
K-1,
x
10-5
Cp
=
K-t,
f3
=
4.9
K
=
7.7
3901 kg-1
10
and
310 K. Take
.
Pa The density of copper is 9.85 x 1 cY kg m- 3.
-

;;.
:;::; ::
.
;
:'...::.::;. .:.;
:
:
;
::
:

::::

:;:.
:; ::::::::

fiX

:-;
: :
:

IC

8.1

Introduction

129

8.2

The Legendre Transformation

130

8.3

Definition of the Thermodynamic Potentials

132

8.4

The Maxwell Relations

134

8.5

The Helmholtz Function

134

8.6

The Gibbs Function

136

8. 7

Application of the Gibbs Function to Phase


Transitions

137

8.8

An Application of the Maxwell Relations

142

8. 9

Conditions of Stable Equilibrium

143

127

.
.

.
.

.
.

.
.

8.1 INTRODUCTION
In the context of the first law, we have defined two functions of the state vari
ables with the dimensions of energy: the internal energy U and the enthalpy H.
Since neither of these is well suited to the analysis of certain processes, it
will be convenient to introduce two additional functions the Helmholtz
function F and the Gibbs functio11 G. Because of their role in detertttining the
equilibrium states of systems under prescribed constraints, they are known as

thermodynamic potentials, in analogy with the potential energy in mechanics.


The enthalpy, Helmholtz function, and Gibbs function, are all related to
the internal energy and can. be derived from it using a procedure known as a
Legendre differential transformation . To see how this is done, we consider the
combined first and second laws, written as
dU

(8.1)

TdS - PdV.

The two independent variables S and V are intrinsically extensive quantities.


The two intensive variables, T and - P, are said to be canonically conjugate to
them. That is, the canonically conjugate pairs are
T, S

and

- P, V.

Note that T and S are thermal variables, whereas P and V are by nature
mechanical variables.
In Equation (8.1) \\'e assume U

dU

au
\ as

U(S, V) so that

dS +
v

au
av

dV.

(8.2)

Comparing Equations (8.1) and (8.2), we see that

au
as

=T '

au
av

-P.

(8.3)
129

Chap. 8

130

Thermodynamic Potentials

However" the selection of the two independent variables is a matter of choice.


There are four possible ways in which a thermal variable can be paired with a
mechanical variable:
(8.4)
We consider the feasibility of defining a thermodynamic potential fclr each of
the four pairs.

8.2 THE LEGENDRE TRANSFORMATION

Consider the function Z

Z ( .r v) and write the differential

..

d Z ( .r y )

X tl .t + Y d y.

(8.5)

where x, X and},' Yare, by definition, canonically conjugate pairs. We wish to


replace (x, y) by (X, Y) as independent variables. To do this we transform
the function Z to a fun ction lv/. The equatiorl of transformation is
Nf(X. Y)

Z- xX- y},P.

(8.6)

Then
tiM

tiZ- X d.t -- Y llv

.\

dX

- v liY.

Since. by Equation (8.5) the first three ter1ns on the right-hand side sum to
zero, we have
{/NI

= -.r

dX-

dY .

Thus Equation (8.6) is the transformatic)n that takes us from

(8.7)
a

function of one

pair of variables to the other. Equations (8.5) and (8.7) gi ve recipr(JCity rela

tions as follows:
az

ax

==

az

X.

ay

aM

aM

ax -x.

aY

If we wish to replace onl y

lJne

==

- Y

(8.8)

( 8.9)

()f th e variables. say }', by its canonically

conjugate variable Y, we must consider the function

N(.r_ Y)

)'Y.

(8.1 0)

Sec. 8.2

The Legendre Transformation

Ill

Then
dN

dZ

Y dy - y dY

and, using Equation (8.5), we obtain


dN

X dx

dY,
y

(8.11)

with reciprocity relations


iJN
ax

X'

aN
aY

=
-

(8.12)

y.*

The equations of transfortnation, Equations (8.6) and (8.1 0), may look
mysterious at first glance. Howe\'er, they are based on a very sirnple idea: a
curve in a plane can be equally well represented by pairs of coordinates (point
geometry), or as the envelope of a family of tangent lines (line geometry), as

depicted in Figure 8.1.


Z()'), where y is the
Suppose that the curve is given by the relation Z
independent variable. Consider a tangent line that goes through the point (y, Z)
Y. If the Z intercept is N, the equation of the line is
and has a slope. d ZI dy
=

Z-N
=

y-

0,

Figure 8.1 A curve


represented as the envelope of
a family of tangent lines.

*It is note\\'Orthy that in classical mechanics. the transfonnation from the Lagrangian to
the Hamiltonian is

Here the qJc, q"

cegendre transfortnation:

and P1c are generalized coordinates, velocities, &nd momenta, respectively.

Chap. 8

Ill

Thermodynamic Potentials

or

which is Equation (8.10). If now we differentiate this expression. we have


dN = dZ

dY - Y dv.
.

Y dj', so

But dZ =

dN

vd}"

which is the reciprocal relation

.v = -

dN
dY

Thus the Legendr e transformation is a mapping from a

(}', Z) space to a

( Y, N) space, from a point representation <>f a curve to a tangent line repre

sentation. The extension of the transformation to functions of more than one


independent variable is straightforward. as shown.

8.3

DEFINITION OF THE THERMODYNAMIC POTENTIALS


U(S V) to the enthalp.y H(S P), we
r epla ce V by its conjugate varia hie - P. Thus

1. To transform the internal energy


.

Since

dU

=-

TdS

P)dV.

we make the fo ll owing associations with the variables of Equations (85)


and

(8. 1 0):
Z=U,

Equations

X=T

t=S.

Y=-P.

v==V_

(8.1 0), (8.11) and ( 8.12) immediately give


,

N=H.

Sec. 8.3

Definition of the Thermodynamic Potentials

Ill

H = U + PV,
dfl=TdS+VdP,
aH

OH
aP

(8.13)

=V

These are the fundamental relations involving the enthalpy.

2. Next we replace the thennal variable ..in the function U(S, V) by its
conjugate variable T. This leads to a new potential, the Helmholtz func
tion F. The Legendre transformation is
U(S, V)-+ F(T, V).
With the identification
z =U,

-P ,

X =V,

y = T,

S,

N =F,

we obtain the expressions

F
dF
OF

U- ST,

= -P '

av

(8.14)

P dV- S dT,

OF
aT

= -S.
v

3. Finally, in U (S, V) we replaceS by its conjugate T and V by its conjugate


-

P, yielding the Gibbs function G, the last thettnodynamic potential

required by the pairing of thermal and mechanical variables. The trans


formation
U(S, V) G(T, P)
is effected by using Equations (8.5) and (8.6) and setting

Z = U,

T,

x =

S,

y =V,

Y = - P,

G.

The resulting expressions are


G =U- ST + PV,
dG = S dT+V dP,
-

OG
iJT

= -S '
P

OG
aP

=V.
r

(8.15)

Thermodynamic Potentials

Chap. 8

134

These potentials have interesting propcrtie which will be discussed later


in this chapter.

8.4 THE MAXWELL RELATIONS


Each of the four thermodynamic potentials is a state variable whose differen
tial is exact. We can use the condition for exactness discussed in Appendix A ..
which states that the value of a mixed second partiaJ derivative is independent

of the order in which the differentiation is applied. As an example_ we consider


dU

Tt/S

au

P)dV

riS

au

dS +

dV.

riV

The exactness of dU immediately gives

--

av as

==

--

av

--

.:=:

aP

iJS

\'

t.

The equality of the first derivatives,

av

a.s

. .

'

is known as a Maxwell refa1ion. I t s utility lies in the fact that each of the par
tials is a state variable that can be integrated along any convenient reversible
path to obtain differences in values of the fundamental state variables
between given equilibrium states.
A summary of relationships involving the thermodynamic potentials,
including the MaxweJI relations, is presented in Table 8.1.

8.5 THE HELMHOLTZ FUNCTION


We have seen that a key property of the internal energy is that the change in U
is the heat flow in an isochoric reversible p rocess. Similarly, the change in the
enthalpy H is the heat flo\\' in an isobaric

reve rs ible

process. We ask: With

what quantities can we associate changes in F and G as a result of reversible


processes?
Suppose that a system is in thermal contact "'ith a reservo ir environment
that is at a constant temperature T. Let tht system undergo a process from
some initial state to a final stat e According t\) the second law,
.

......

So> 0.

The Helmholtz Function

Sec. 8.5

TABLE 8. I

135

Relationships involving the thermodynamic potentials.

Thermodynamic

Independent

Reciprocity

Maxwell

Potential

Variables

Relations

Relations

T=

S, V

lntemal energy

as

av

aP
-

as

a2u
=

av

'

au

-P

dU = T dS - P d'l

ar

au\

--

av as

F= U- TS

-P

dG = -SdT + V dP

U- TS + PV

as

s
=

--

riP as
aP

as

aF

ar

av

ar

V=

--

av

ar

aG

aP

--

-T

av ar
av

as

aG

liP

a2F

aF
=

a2H

aP

S= -

T,P

Gibbs function
=

S dT - P dV

aP

aH

S=-

T,V

dF

as

V=

dH = T dS + V dP

U + PV

Helmholtz function

T=

S.P

Enthalpy

av

ar

riH

ar

a2G
aP ar

H - TS

where S is the entropy change of the system and So is the entropy change
of the reservoir. If the reservoir transfers heat to the system, then
So
Q/T and therefore,
=

<

T S,

where the equality sign applies if the process is reversible. If we substitute this
in the first law expressed in the form
.
W

AU +

Q,

we obtain
W

or, since F

<

dU f. T S,

TS,
W

<

(no change in n.

(8.16)

Thus the change in the Helmholtz function in an isotherttla l reversible process


is the work done on or by the system. More generally, the decrease in F equals

Chap.8

136

Thermodynamic Potentials

the maximum energy that can be freed in an isothermal process and made
available for work.* The function is therefo re often called the HelmJtoltz free
energy. (Elsewhere it is frequently given the symbol A. an abbreviation for the
German word for

work, Arbeit.)

If the system held in thermal cont act with the rese rvoir has a uniform
pressure throughout its volume and the latter is also held constant, then the
work performed will be zero and

(8.17)

(8.18)
where i and f denote the initial and final values, respectively. Hence, if the
Helmholtz function is a minimum, any change in the state of the system would
increase

F,

which would be cont radi c tory tl Equati on

the condition for

equilibriun1 in

(8.18). It follows that

system in thermal contact with a reservoir

and kept at cons t an t volume is

(8.19)

with Fa minimunz. S ince T and V are the most convenient fundamental state
variables for the purposes <)f statistical Jnechanics, as

we

shall see. the

Helmholtz function becomes a very significcrnt means of specifying a system's

properties.

8.6 THE GIBBS FUNCTION


The Gibbs function is useful in prohlems in which T and P are the fundamen
tal variables. Consider a system in a surrounding environment that C< )ns t itutes

a temperature and pressure reservoir. By this we mean that the reservoir is so


large that its temperature and pressure remain unch anged. Most chemic a)
reactions and some phase changes (icc melting in a heakcr e x p osed to the
atmosphere for example) take p lace in this 'vay. As before,
.

wi th

P :!1 v.

Actually. it is not necessary that the temperature remain constant during the process: it is

only required that the two end points be at tcmperatur('

T.

Sec. 8.7

Application of the Gibbs Function to Phase Transitions

137

reverstb
_ le.
Co1nbining these expressions, we hav
e

&U + P V

T AS < 0.

Since
G

U + PV- TS

'

and T and Pare constant here, we obtain

( AG)r.P

<

0,

(8.20)

or

(8.21)
Spontaneous processes occur in the direction of decreasing

G.

A system in

thettnal contact with a heat and pressure reservoir moves to a state of mini

mum G for which


dG

(8.22)

0.

It can be shown that if nonmechanical forces (electric forces, magnetic


forces, etc.) act on

system, doing work "''"m' then


Wnm <

aG

(no change in Tor

P).

(8.23)

The decrease in the Gibbs function is equal t<.l the maximum energy that can
be freed in an isothermal, isobaric process and made availabJe for nonmechan
ical work. Again, T and Pneed not be fixed tltroughout the process; they need
only have the same initial and final values.
Table

8.2

lists the applicable conditions for various system states or

processes.

8. 7 APPLICATION OF THE GIBBS FUNCTION TO PHASE


TRANSITIONS

Consider a system consisting of the liquid and vapor phases of some sub
stance in equilibrium at temperature T and pressure P. Let n'{ be the number
of kilomoles in the liquid phase and n'i' the number of kilomoles in the vapor

Thermodynamic Potentials

Chap. 8

138

TABLE 8.2*
processes.

Conditions on thermodynamic variables for different systems or

Valid

Valid

Equilibrium

Equation

Inequality

Condition

dU :S 0
dll 0
dF s 0
dG :50
dS 0

L'
Minimum H
finimum f'
Minimum (j

State of System
or Type of

Process

dS
dS
tiT
dT
71Q dU

V constant
Sand P cons ta n t
T and V constant
T and P constant
Sand

==

=-=

Adiabatic

dV 0
dP 0
dV
0
dP 0
+ P d\1'
=

==

Minimum

MaximumS

___;.
--.1

......
.
_

'

.
. . .

____

._...
_....
.: . . . . .. ..
.
.

. . . .
.
.

n.,

..

...

..
.
.
. .. . .
.
.
::._::.- . .. . .:.: ..: : . .
.
.
.
..
. . '. -
. : .:.'. .t: . .
.. '
.
' .
.. .
:
.
..
:. . . . :
' . .
.

Figure 8.2

,.

Liquid and vapor


phases of a substance in
equilibrium at temperature T

..
-

. :

;.
.

..

and pressure P (a) initial

'

'

..

..

. ..

. .

..
.

.
.

.
.

.
. .

'

. .

.
.

. .

. : . .

(a)

state: (b) final state.

(b)

phase. 'Tne state of the system is defined in terms of the variables


(T, P,

Consider a second state differing from the -first only in the

n]', n'(').

number of kilomoles of liquid and vapor and defined by

(T. P. n3, n2') (Figure

8.2). Mass is conserved so that


1z

}' +

j"

== n

2' .

(8.24)

''
"'
We define g and g as the specific Gibbs functions of the liquid and vapor.
respectively, associated with the particular substance under investigation.
Noting that the Gibbs function is an extensive variable. we have for the two
states:
G1
G2

== 11

' g
,,

=: n

"

+ n
+

tg

"'

,,,

,,,

112 g .

(8.25)
(8.26)

7.2 in Tht'rnlodynanJics tnd Statistical Mechanics by P. L. Landsberg.


Publ icat i ons New York, 1990.

*Adapted from Table


Dover

is that

section 4.3: one. t\VO. and three prim es denote the soJid.liquid.
and vapor phases. respectively. Here 1 refers to the initial state and 2 to the final state.
t

The notation

used in

Sec. 8.7

Application of the Gibbs Function to Phase Transitions

139

liquid
dP
Figure 8.3 Relationship
between temperature and
pressure for a iiquid and vapor
in equilibrium. The derivative
dPIdT is the slope of the
vaporization curve.

vapor

""

dT

Suppose that a reversible transition takes place from state 1 to state 2. Since

( G)r,P

0 for a reversible process, it follows that

G1

G2. Equating

Equations (8.25) and (8.26) and using Equation (8.24), we find that
g"

g'".

(8.27)

The specific Gibbs function is the same for the two phases. This is true for all
phases in equilibrium, that is, for all points on the curve of the phase transfor
mation (Figure 8.3).
Since at a temperature 1' + dT and a pressure P + dP we still have
equilibrium, it follows that g" + dg''
Equation (8.27), we have
dg"

g"'

dg"'. Combining this with

dg'" .

Using the expression for the differential previously derived, we can write

-s

"

dT +

"

dP

"'
s

dT +

'"
v

dP,

or
"
"
( s ' - s ) dT

(v

"'

- v

"

) d P.

Thus
dP
-

dT

s"'
=

--

"

---

"'

- v

"

(8.28)

Chap. 8

140

Thermodynamic Potentials

From the definition of entropy

(8.29)
where t23 is the latent heat of vaporization. Since heat is absorbed as a liquid
becomes a vapor, f23 is positive and
Equation

(8.28) gives

dT 23

"'

> s

' Substituting Equation (8.29) in

'

v"

(8.30)

(liquid-vapor).

----

T ( v"'

This is the famous Clausius-Clapeyron equation. It gives the slope of the curve
denoting the boundary between the liquid and vapor phases, that is the vapor
ization curve. Similar expressions hold for the sublimation and fusion curves:

dP
dT

dT

--

T(v'''

1.1

dP

13

12

v')

f12

--

"
T ( t'

v')

(solid-vapor),

(8.31)

(solid-liquid).

(8.32)

The latent heats in these expressions are positive, and the slopes are all posi
tive for substances that expand on melting. A notable exception is water,
which contracts when ice melts into liquid; for this case

( dPI dT) 12

< 0.

The Clausius-Clapeyron equation combined with the appropriate equa


tions of state, can in principle yield equati(>ns for the phase transformation

curves. A simple example is the vaporization curve describing, say the conver
sion of liquid water to stea m Here
.

v"' >> v''

dP

dT .,,

:=:

....

(see Chapter

f 2.1

"'
t'

If we treat the vapor as an ideal gas.

v "'

::::::=

RT
-

p '

so that

dP
dT

::...:::

t23 P
-

R T2.

2) and so

Application of the Gibbs Function to Phase Transitions

Sec. 8.7

141

If f2 is assumed to be temperature-independent, upon integrating we obtain


3

(8.33)

where

(T0, P0)

denotes some fixed poin.t on the curve.*

The Clausius-Clapeyron equation can also help us understand why the

(273.15 K) is 0.01 K below the triple point (273.16 K).


Solving for dT in Equation (8.32), we have

ice point of water

T( v" - v')

dT

t12

dP.

For the small temperature change involved, we can write


4T

T(v" - v')
=

f12

and assume that the specific volumes

v''

and

v'

are constant. This equation

gives the change in temperature as the pressure is increased from the low
value at the triple point

(Prp

4.58 Torr) to atmospheric pressure, the pres

sure at which the normal melting point of water is defined.


The density of pure liquid water is 1000 kg m -J and the density of ice is
916 kg m-3 Noting that the specific volume in m3 kg-1 is just the reciprocal of
1
1.00 X 10-3m3 kg-1 and v' 1.09 X 1 0 3 m3 kg- ,
the density, we have v"
:=

to three significant figures. The latent heat of melting for water is


f12

3.34

liP

1051 kg-1 Writing Prp in Pascals, we have


=

Patm - Prp

1.01

105 Pa

610 Pa

::::

1.01

105 Pa.

Then

AT

273(1.00

10-3

1 . 09

3.34

10-3) (1.01

lOs

lOS)
..

-0.0074 K.

The presence of dissolved air in a mixture of ice, and water further lowers the
temperature at which ice melts by 0.0023 K. The total reduction of temperature
below the triple point of water is therefore 0.0097 K, or approximately 0 01 K.
..

*The more generaJ solution is


P

P0 exp

Chap. 8

141

Thermodynamic Potentials

8.8 AN APPLICATION OF THE MAXWELL RELATIONS


As an example of the use of a Maxwell relation, consider an ideal gas that

undergoes an isothertttal reversible change from pressure P0 to pressure P. By


definition of the entropy,
aQ,

T dS.

We express S as a function of the fundamental state variables of the prob


lem

namely, T and P:

S(T, P).

Then

dS

as
ar

Since the process is isothermal, dT

as
aP

dT +
p

dP.
r

0, and

'

dP.

ap

'

From Table 7.1, we select the Maxwell relation


<

as

riP

iJ\'
-

- -

a1

..

Thus

l/Q,

-T

av

iJT

dP.
r

For an ideal gas,

riV

aT

nR
-

,)

P '

and
p

Pn

Sec. 8. 9

Conditions of Stable Equilibrium

143

If P > Po, the negative sign indicates tltat heat flows out of the system in the
process.

8.9 CONDITIONS OF STABLE EQUILIBRIUM


The second law states that in spontaneous processes the entropy increases.
Maximum entropy corresponds to thermodynamic equilibrium. Furthermore,

maximum entropy is a point of stable equilibrium: if the system is perturbed


slightly from its equilibrium state, it will return to that state spontaneously. We
wish to examine the implications of this fundamental law of nature.
Since U = U( S, V), it follows that S == S(U, V). The maximization of S
is a problem in multivariate calculus.* A theorem states that the function

f ( x,

1.

y)

has an absolute maximum if, for all points in the domain,


< 0,

fxx

2. D

>

0,

where

(In

fxy

this

for example,

iJ2fI ox a y.) Thus the conditions on S are

1.

section, the subscripts indicate differentiation;

Suu

2. D

< 0,

Suu Svv

(Suv )2

>

0.

From the combined first and second laws,

dS

Su dU

Sv dV

I
=

dU +

dV.
T

(8.34)

Thus

Su

1
=

'

Sv

(8.35)

'
T

Taking the differential of the first of these equations, we obtain

Calculus: One and Several Variables,


E. Hille, Wiley. New York, 1982, pp. 792 ff.

See, for example,

4th edition, by S.

L.

Salas and

Chap. 8

144

Thermodynamic Potentials

Then
riU

Suu

iiT

--

r.,

Since
au

Cr

riT ,.

"

it follows that
(8.36)

The temperature Tis always positive. Therefore the first condition for
ity, Suu

<

0 implies that

stabil

..

(8.37)
Now let''s take the
tion (8.35):

d iffe r e nti als of the two equations labeled Equa

Svv dU

Sl't: dV

where D- Suv

dT

dP

Cramers rule:

V
d

We can solve for dV (or tiU) using

tiT
f2

(8.38)

-------

Sv\-' - (Scv )2. Expanding the determinant in the numerator

and rearranging terms. we obtain

D tlV

Svu
-

T
-

dP

(S"'u

--

Suu

P)

dT
"'T

Sec. 8.9

Conditions of Stable Equilibrium

145

Then

iJP

T '

or

D=

Suu

aP

av

Now
av

1
I(

aP

'

and

from Equation (8.36), so that


1

(8.39)

Since Cv, T, and V are all positive, the second condition for stability, D > 0,
implies that

{8.40)

I(> 0.

We conclude that for thermal stability, the addition of heat to a body


causes the temperature to increase:

Cv

For mechanical stability, a decrease of pressure causes an increase in vol.ume:

iJV
aP

> 0.
r

That is, for a closed system, stable-thermodytzamic equilibrilJffi consists of ther


mal and tnechanical equilibrium.
There are a few exceptions to this otherwise general result. At critical
points the heat capacities and the compressibility can diverge and the stability
conditions are then violated. It is also interesting to note that long-range
forces can give rise to negative heat capacities. Such modifications have by no
means been fully investigated as yet.

Thermodynamic Potentials

Chap.8

146

PROBLEMS

8-1 A van der Waals gas and an ideal gas are originally at the same pressure, tem
perature .. and volume. If each gas undergoes a reversible isothermal compres
sion.. which gas will experience the greater change in entropy?

8-2 Show that for the an ideal gas

(a) f

C0(T

To)

c]

In

To
.

(b) g = c,.(T

T0)

c,.T In

- RT In

Vo .

p
+ RT ln p,
'

soT.
s0T.

8-3 A cylinder contains a piston on each side of which is one kilomole of an ideal
gas. The walls of the cylinder are diathermal and the system is in contact with a
heat reservoir at a temperature of 0C. The initial volumes of the gaseous sub
systems on either side of the piston are 12 liters and 2 liters. respectively. The
piston is now moved reversibly so that the final volumes are each 7 liters. What
is the change in the Hemhohz potential? (Note that this is the work delivered to
the system by the reservoir.)

8-4 Derive the following equations:


(a) F

(b) Cr

U + T

aF

ar

-T

(reversible process)-
.

(c) H

G
-

(d) C,

rJG
aT

a 2G
T
2
ar

p:

(reversible process):

(The third relation is the Gibbs-Helmholtz equation. alluded to in Chapter

10.)

8-S The Helmholtz function of a certain gas is


F

n2a
=

nRT ln(V- nb) + J(T).

where J is a function of T only. Derive an expression for the pressure of the gas.

8-6 The Gibbs function of a certain gas is


G
where A 8,

C.

nRT lnP

A + BP

CP2
2

DP3
+ --3 .

and D are constants. Find the equation of state of the gas.

Problems

8-7

147

The specific Gibbs function of a gas is given by


'
g

RT ln

- AP.

Po

where A is a function of T. Find expressions for:

8-8

the equation of state;

(a)
(b)
(c)

the specific Helmholtz function.

(a)

A van der Waals gas undergoes an isothetrnal expansion from specific vol

the sp ecific entropy ;

ume v1 to specific volume

v2

Calculate the cha nge in the specific Helmholtz

function.

(b)

Calculate the change in the specific internal energy in terrns of

8-9 Start with the first Maxwell relation lited in Table

t't

and v2

8.1 and derive the second by

using the cyclical and recipirocal relations (Appendix A) and the identity

az

ay

ax

ay I oz

iJx

1.

(The remaining Maxwell relations can be derived in a similar manner.)

8-10

(a)

Prove that
Cp

and use the result to show that

(b)

Prove that Cp for an ideal gas is a function ofT o nly

8-11 ln Figure 2.2 depicting the isotherms for a van der Waals gas, the shaded areas
were shown by Maxwell to be equal. The path

bed is the segment of an isotherrn

along which liquid and vapor are in equilibrium. Prove Maxwell's result by not
ing that Ag

0 for a phase change (P and T constant) and calculating f

8-12 The P-v diagram in Figure 8.4 shows two neighboring isotherrns in the region of
a liquid-gas phase transition. By considering a Carnot cycle .between tempera
tures T and T + d1. in the region shown, derive the Clausius-Clapeyron equa
tion

dP/dT

f23/T(v).

Here

vis the

specific volume change between gas

and liquid.

8-13 The equations of the sublimation and the vaporization curves of a particular
material are given by
In

In

0.04
0.03

6fT

(sublimation),

4/T

(vaporization),

148

Chap. 8

Thermodynamic Potentials

T
.

. .

.;- .
.
. .
. ....

T
Figure 8.4

P- v diagram for

dT

.
.

. .:...

. .
. ..
.

= const.

..
.

. . . ..

. : ...-: . . .. . ..
.
.
.
.

.#

..

..

.. .
. .

.
.

= const.

'

an infinitesimal Carnot cycle.

---- ------

t'

where P is in atmosp heres

(a) Find the temperature of the tri ple point.


(b) Show that the specific latent heats of vaporization and sublimation are 4R
and 6R, resp ective ly (You may assume that the specific volume in the vapor
.

phase is much larger than the specific volume in the liquid and solid

phases . )

(c) Find the latent heat of fusion.


8-14 (a) Calculate the slope of the fusion curve of ice in PaK _, at the normal melting
point . At this temperature, the heat of fusion is 3.34 X 1 a Jkg. i and the
change in specific volume on melting is - 9.05 x 1 o- m3kg- 1
(b) Ice at 2C and atmospheri c pressure is compressed i so th e rmally Find the
-

pressure at which the ice starts to melt (in atmospheres).

e
en

9 .I

The Chemical Potential

lSI

9.2

Phase Equilibrium

ISS

9.3

The Gibbs Phase Rule

157

9.4

Chemical Reactions

160

9.5

Mixing Processes

162

149

9.1 THE CHEMICAL POTENTIAL

Until now we have confmed our discussion to closed physical systems, which
cannot exchange matter with their sutToundings. We turn our attention in this
chapter to open systems, in which the quantity of matter is not fixed.
Suppose that

dn kilomoles of matter are introduced into a system. Each

kilomole of added matter has its own internal energy that is released to the
rest of the system, possibly in a chemical reactio11. The added energy is propor
tional to

dn and may be written as p,dn. The quantity #L is called the chemical

potential.
The chemical potential is associated with intertttolecular forces. An elec
trically polarized molecule experiences a Coulomb attraction when it is
brought into the vicinity of another such molecule.* 'Ibis force is expressed as
a negative potential energy, a sort of "potential well." As the new particle
approaches its neighbor, it gains kinetic energy while losing potential energy.
The kinetic energy is imparted to other particles through collisions, so the sys
tem gains internal energy in the process.
Consider a motionless molecule infinitely distant from other molecules.
Its kinetic energy and potential energy are both zero. The molecule is moved
into the force field of a second molecule. This can, in principle, be done slowly
so that the kinetic energy is negligibly small. Left by itself, however, the mole
cule picks up kinetic energy equal in magnitude to the depth of the potential
well (Figure 9.1). Quantitatively,
E=K+V (r),
where E is the total energy, K is the kinetic energy, and V(r) is the potential
energy; r is the distance between the molecule and its neighbor. At r = oo,
K=

0, and V

0, so E

0 everywhere. At r = r0,

See Section 11.1 for a discussion of the molecular interaction potential.


lSI

The Chemical Potential and Open Systems

Chap. 9

152

V(r)

Figure 9.1 Schematic


diag ram of a potential well
due to intermolecular forces.
A more descriptive graph is
given in Section 11.1.

I
I K
I
I

Energy is conserved but a conversion from potential energy to kinetic energy


takes place. The kinetic energy is added to the internal energy of the system.
It is reasonable to ask what the magnitude of J.L is. In a standard labora
tory experiment sulfuric acid is added to water, producing an increase in tem
perature. Imagine that 10-5 kilomoles of acid at room temperature are added
to a liter of water

(5.56

x Jo-:! kilomoles). also at room temperature. The tem-

perature is observed to rise

0.1 C. We wish to determine the chemical poten

tial of acid in water.


We shall assume that the interaction among the acid molecules is small
compared with their interaction with tl1e water molecules (both acid and
water molecules are eJectrically polarized). 'llle specific heat capacity of water
is

7.52

104 J kilomole- 1

K -1 Thus the heat gained by the water through the

addition of the acid is

lnc,T

(5.56

)(7.52

X 10

10-')(0.1)

4181.

Since the mass of the water is more than 103 times the mass of the acid .. we can
ignore the heat capacity of the latter. The chemical potentiaL then is

"
,-

41X

1()-)
_

7
4.1 8 x 10 J kilomole-

1
_

The sign i s negat i ve because heat is transferred jro111 the acid to the \Vater.
We are interested in the chemical energy per particle, which is essentially
the depth of the potential well. Since a kil()ffiole has 6.02 x 1026 n1olecules.
this value is -6.94 x 10 20 J or -0.43 e\t'. (One electron volt is equal to

1.6

10-19 joules.) Most chemical potentials are of this order of magnitude.

To account for the effect of adding ma ss to

system. we need to add a

term to our fundamental equation of thermodynamics:


dU

=-

Ttl'

P dV

J.Ldn.

(9.1)

Sec. 9.1
Here

153

The Chemical Potential

dn is

the increment of mass added (in kilomoles) and J.L is the chemical

potential in joules per kilomole. If, in an open system, U

iJS

dS

au

OV

V,n

dV

S.n

U (S, V,

au
dn

n)

and

(9.2)

dn,
S,V

then
iJU

J.L=

an.

(9.3)

s.v

That is, the chemical potential is defined as the internal energy per kilomole
added under conditions of constant entropy and volume.
If there is more than one type of particle added to the system (say
types), then Equation

(9.1) becomes

dU

TdS - PdV +

j=l

J.Li

(9.4)

d nj,

with

(9.5)
The subscript nk means that all other n's except ni are held constant.
Another way of seeing the relationship between U and J.Li is to integrate
Equation

(9.4). This

can be done by using Euler's theorem for homogenous

functions. Euler's theorem states that if

Af(x, y, z)

f(Ax, Ay, Az),

(9.6)

at
y-ay

(9.7)

then

at
f =X
ax

at
.
Z
az

The theorem can be easily proved by differentiating Equation (9.6) with


respect to

and then setting

equal to unity. Now U

U(S, V,

n1

nm)

Suppose the amounts of all the types of substance, called constituents, in the
system were doubled or halved or, more generally, changed by the factor

without changing any of the fundamental state variables. Then the extensive
variable U would be changed by

and all the independent, extensive state

The Chemical Potential and Open Systems

Chap. 9

154

variables would also be changed by the factor A. Thus U is a homogeneous


function and Euler's theorem can be applied to it:

au

au

+ v

as

\-"

111

rlv

,/l 1

,= 1

s"

iJU

lli

rin,

(9.8)

, v.fl J..
)'
.

From the differential of U we know that


au
as

V.nJ

iJU

T ..

av

=
I

au
..

s.v.n ..

S.n,

= J.lj .

(9.9)

Substituting the relations of Equation (9.9) in Equation (9.8) we have

ST- PV +

Jl;n;.

..

(9.10)

r= I

Recall that the Gibbs function is defined as

==

ST + PV It is

therefore immediatelv clear that

111

(9.11)
If only one constituent is present .. then G

==

p.11 or JL ==

Gjn: so 1-L in this case is

simply the Gibbs function per kilomole of the substance.


Finally.. if we take the differential of Equation (9.1 0). we obtain
.

Equating Equations

(9.4) an d (9.12) we have


..

Ill

SdT- VdP

+
(= I

11;dJ.L,

0..

(9.13)

a relation known as the Gibbs-Duhem equation. Taking the differential of


Equation

(9.11) gives

Note that

G(T P n1
..

..

(9.14)
.
.

11,,). Thus if we have

process that takes place

at constant temperature and pressure, the first two terms of Equation

(9.13)

Sec. 9.2

Phase Equilibrium

155

vanish so the sum also vanishes. Equation

(9.14)

then yields the important

result

(9.15)

We now have the mathematical tools to look at some applications.

.
.
.

.
.

.
.
.

..

.
.

'

9.2 PHASE EQUILIBRIUM


We Ylish to find the condition of equilibrium for two subsystems under particle
exchange where the subsystems are two phases of the saane substance. An
example would be ice melting in liq.uid water; an exchange of H20 molecules
occurs between the two subsystetns_ We assume that the total system is enclosed
in a rigid adiabatic wall (Figure 9.2). Since there can be heat and particle flow
across the boundary between the two phases, the boundary will move as the
process evolves. The various interactions are subject to the foUowing conditions:

nA

n8

VA

V8

UA

U8

=n = constant

= V =constant
=

constant

(conservation of mass),

(9.16)

(conservation of volume),

(9.17)

(conservation of energy).

(9.18)

The total internal energy is constant, according to the first law, since no heat
flows into or out of the combined system and no work is done. At equilibrium,
the entropy of the combined system will be a maximttm:

SA

S8

Let all the quntities n, V, U, and

dS

(maximun1).

(9.19)

S change by infinitesimal amounts.. Then

dS.4 + dSB

0.

phase boundary

Figure 9.2 1\vo phases of a substance in thermal


equilibrium.

(9 20)
.

156

Chap. 9

The Chemical Potential and Open Systems

But

(9.2 1 )
and

(9.22)
We can eliminate three infinitesimals by invoking Equations

(9.16), ( 9.17), and

(9.18):

Combining these ex pressions with Equations (9.21) and (9.22), and substitut-

ing in Equation (9.20) gives the result


1
-

(9.23)

--

Since Equation

(9.23) must be true for

ar b i t ra ry

increments

dUA, d\lA dnA all


..

of the coefficients must be zero:


TA

PA

J.L 11

PH

(thermal equilibrium) ..

T8

(9.24)

(mechanical equilibr ium)

J.L B

( diffusive tquilibrium)

(9.25)

(9.26)

We conclude that if we have two phases or s u bsys t ems in thermal and


mechanical equil i bri um then they will also be in
.

diffitsive equilibri u m ( equi

librium against particle exchange) if their chemical potentials are equal.


Suppose that the two interacting systems are not yet in equilibriu m

Then the entropy of the combined system \l:ill not have reached its max i mum
value bu t will still be increasing:

(9.27)
Consider a state very close to an equilibriurr1 state and differing from it only in
that there is a small excess of mass 'zA(>O) in s ubsystem A over the equ i l ib

rium value

nA,

and a correspondi n g defici (>f n1ass

d11A in subsystem B. As

Sec. 9.3

The Gibbs Phase Rule

157

the combined system moves toward equilibrium, subsys


tem A will give up its
excess mass with an accompanying entropy change

(9.28)
Simultaneously, the entropy change of subsystem B as a result of acquiring the
additional mass will be

(9.29)

The temperature is the same in both expressions. Substituting Equations


(9.28) and (9.29) in Equation (9.27) gives

or
(9.30)

J.LA > J.LB

Thus, in the approach to equilibrium, heat energy must flow from the
hotter to the cooler body, volume will be gained by the body under lower pres
sure at the expense of the other, and mass will tend to flow from the body of
higher chemical potential toward the one of lower chemical potential. All of
these tendencies follow from the second Jaw of thertnodynamics.

9.3 THE GIBBS PHASE RULE


Consider a system in equilibrium with k constituents in 7r phases

We note the

following:
1. Only one gaseous phase can exist because of diffusion .
.

2. Several liquids can coexist in equilibrium if they are immiscible.


3. Several solids can coexist.
4. Only rarely do more than three phases of a given constituent coexist.

*The symbol
here

with 3.14. .

1r,

traditionally used to denote the number of phases, is not to be confused

158

Chap. 9

The Chemical Potential and Open Systems

Let P.l be the chemical potential of the ith constituent in the yth phase.
According to Equation (9. 1 1) the G i bbs functi on is

(9.31)
where n? is the number of kilomoles of the ith constituent in the yth phase.
For equilibrium

(dG)r.P

0. so Equation (9.15) gives

1-L?
dn?
..,..

0 1 and P fixed.

i=] y=l

(9.32)

It also follows from the Gibbs-Duhem equation (Equation (9.13)) that for
constant temperature and pressure.

(9.33)

t=J y= I

If we consider the special case of a closed ystem in which mass

dn; of the ith

constituent is transferred from one phase t(> another with the total 1nass of the
constituent unchanged.. then
-

"

dll?
--

:=:

0, i

(9.34)

1, 2. . . . k.

'Y = 1

This equation is simply a statement of the conse rvati on of mass: No con


stituent is created or destr oy ed.
Suppose that there are two constituents ( i

(y

cr,

1. 2 ) and two phases

J3). Then Equation (9.32) becomes

and Equation ( 9 34) giv es


.

(9.36)
.

Combining these e q ua tion s gives

(9.37)
Since the differentials dn1u and in2o are arbitrary, the coefficients must vanish.
Hence
(9.38)

Sec. 9.3

The Gibbs Phase Rule

159

In other words, the equality of the chemical potentials for the two phases of
each of the constituents is the condition for equiljbriutn with T and P fixed.
From the conservation of mass it follows that the number of kilomoles of
each constituent is fixed provided no chemical reaction occurs. Thus

(9.39)
where

n1

and

n2

are the given number of kilomoles of constituents

and 2,

respectively. It. is convenient to introduce for each constituent the concept of


the kilomole fraction, the ratio of the number of kilomoles of a given phase to
the total number of kilomoles of all constituents in that phase:
.

. .

(9.40)

From Equations (9.39) and (9.40), it is evident that

(9.41)
Therefore, in our example of k

2, 1r

2, the state of the system is detertnined

by four independent variables: T, P, x1a, and x2a. tlowever, since there are two
equilibrium conditions ( Equation (9.38)) , the ttumber of variables is reduced to
two.
The generalization of this specia l case leads to the so-called Gibbs phase
rule. For a multiconstituent, multiphase system with T and P fixed, the condition for equilibrium is
II. Q

r-i

ll.f3 - ll. y
r-i

- ,.-,

J.L/r'

1' 2,

k.

( 9 42 )

7T.

(9.43)

The kilomole fractions are:

xY
l

n;y

i= 1

'

1' 2 ..k;
.

1'

1' 2

...

,,Ir

Thus, there would b k1r k ilomole fractions were it not for the identity
k

x?
i

(9.44)

1,

which subtracts 1r kilomole fractions from the total. Counti ng the two prescribed
auantities. T an d P. there are therefore 2 + k1r

1T

inde oenden t variables

The Chemical Potential and Open Systems

Chap. 9

160

and k ( 1T

l) equilibrium conditions (Equation (9.42)) . The number of

remaining .. degrees of freedom," called the variance f, is

==

[2

k7i- 1T]- [k(1T-

J)l

or

1r +

(9.45)

(no chtmical reaction).

This is the usual form of the Gibbs phase rule.


If f

==

0. the system has zero variance and is completely determined. If

1, the system is monovariant and is not determined until one additional

variable is specified.
Consider the following examples:
1. A homogeneous fluid with one constituent and one phase. Here k
71'

1,

2. Thus we can choose T and P arbitrarily. This is the case of.

an ideal gas for instance.


2. A homogeneous system consisting ot two chemically different gases in

the same phase. Here k

2,

r.

I, f

3. Here we can choose

T, P. and

one kilomole fraction.


3. Water in equilibrium with its saturated vapor. In this case k

1. Thus we can choose only the temperature

trarily. Equilibrium is defined by

1,

7T

==

2,

(or the pressure) arbi

p<")int on the vaporization curve for

water; specification of either Tor P gives the value of the other variable.
4. Ice, liquid water, and water vapor in equilibrium. Here k

1,

1T =

3,

0. The three phases can coexist in equilibrium onJy for the fixed set

of values T and P defining the triple point.

9.4 CHEMICAL REACTIONS

As we have noted, the chemical potential is a measure of the chemical energy per
kilomole that a substance can generate in a reaction. Chemical thermodynam
ics involves, in part, the measurement of th e chemical potentials of compounds.
A chemical reaction is usually described by an equation, such as

(9.46)
Such an equation can be generalized

as

(9.47)

Sec. 9.4

Chemical Reactions

161

which states that a certain number of molecules of the initial reactants will com
bine to yield a certain number of final product molecules. Here the Mi denote
chemical symbols and the vi are so-called stoichiometric coefficients. The coeffi

ciettts are positive or negative integers; negative values represent initial reactants
and positive values correspond to final products. Thus, in the previous example,
and

vn,c)
-

+ 2.

For a chemica! reaction to occur, more than one substa11ce must be present
either initially or fmally. The number of kilomoles ni of each of the constituents
will change in a manner consistent with the conservation of atomic species. In

the example, for every two kilomoles of hydrogen and one kilomole of oxygen
that disappear, two kilomoles of water will appear. The change in the number of
kilomoles will therefore be proportional to the stoichiometric coefficients:

Suppose that a chemical reaction takes place under conditions of con


stant temperature and pressure. This would occur when , say, the reactants are
immersed in a water bath at atmospheric pressure. However, it is not neces
sary that the temperature and pressure be constant throughout the process. It
is sufficient that they return to their initial values after the reaction happens,
which is virtually always the case. Under these Ctlnditions, the Gibbs-Duhem
equation (Equation

(9.13)) states that the chemical potentials of the reactants

are fixed. Then, at equilibrium

(dG)r.P

0, so that

where the j's are constants. Finally, since dnioc"i' we can write

(9.48)

for chemical equilibriztm in a reaction. Once initiated, most chemical reactions


proceed spontaneously either until some reactant runs out or until the reac
tion arrives at equilibrium. If the reaction is being used to determine the chem
ical potentials of the constituents, the experiment will be designed to reach
equilibrium. In the reaction of Equation

(9.46), we obtain the useful result


(9.49)

This relation can serve as a check on the measured values of the individual p,'s.

The Chemical Potential and Open Systems

Chap. 9

162

9.5 MIXING PROCESSES

In Section 9.2 we considered the problem of particle exchange between two


phases of a given substance. Here we shall address the phenomenon of the
mixing or

interdiffusion

of two different gases. We assume that the mixing

takes place at constant temperature and pressure.. so that the Gibbs function is
again a central concept in the description of the process.
Dalton's law of partial pressures states that the pressure Pj of the jth con
stituent gas of a mixture of gases is given by
(9.50)
where xi is the kiJomole fraction of the jth constituent gas defined in Section 9.3 ..
and Pis the pressure of the mixture.* Equation (9.50) is a definition of the par
tial pressure; the relation foiiO\\IS immediately from the ideal gas law for which

nRT
1
v

:=

nRT
v

since then

l!liaat
. tl
,

p.
J

p
11

'

The Gibbs function for a mixture is


(9.51)

where gj is the specific Gibbs function for the jth constituent. To calculate the

specific Gibbs function for an ideal gas we start with the first law expressed in
the forrn
Tlls

With

v =

cpdT

vdP.

RT/P, we obtain
ds

= c

dT
T

dP
Rp.

We omit the superscript on the kilomole fraction because there is only one phase in this

process. Here

Ln,
J

n.

the totaJ number of kilomoles of the system.

Sec. 9.5

Mixing Processes

16]

Integration gives
s

Cp In T - R In P + s0

Since
g

= u

+ Pv

Ts

Ts,

and

it follows that
g

cpT - cpT ln T

Kf In P

Ts0

h0,

or
g

RT(ln P

cp),

(9.52)

where c/J is a function that depends on T only.


We can now calculate the change in the Gibbs function as a result of the
mixing of two ideal gases (Figure 9.3). The initial Gibbs function (before mixing) is

(9.53)
where
gli

RT(ln P +

</>1),

Ku

RT(In P +

cPz),

so that

(9.54)
Figure 9.3

The mixing of two


ideal gases. (a) The initial
state; the gases are at the same
temperature and pressure and
are separated by a diaphragm.
(b) The final state; the
diaphragm has been removed,
allowing the gases to mix. The
final temperature is T and
the final pressure is the sum
of the partial pressures.

: initial state

T,P

T,P

diaphragm
(a)

final state f

T, P

P1

P2

(diaphragm removed)
(b)

The Chemical Potential and Open Systems

Chap. 9

164

Although c/>1 and c/J2 are functions of the same temperature T, the functions
themselves are different because

Cp

will in general not have the same value for

the two gases.


The final value of the Gibbs function is

with

Since T doesn t change in the mixing process, c/>1 and c/>2 will be the same for
the final state as for the initial state. Using P1
g11

RT( ln P + </>1 +Inx1)

g21

x1P and P2

x2P, we obtain

RT(ln P + c/J2 +Inx2),

so
G1

n1RT(In P

</>1 +In x1) + 1z2RT(ln

4>2 +In x2).

(9.55)

Suppose that we define


J.l

RT(ln P + 4> + lnx)


g

==

RTin x.

Then

Comparing this with Eq u ati on (9.11) we see that J.L is the chemical potential!
Since g1 and g2 are characteristics of the constituents, they are the same for the
initial and final states and we can drop the subscripts on gin Equation (9.53),
which becomes

(9.56)
Our final result is therefore
G

G,. - G,

nt (p.1

g,) + n2(J.l2

g2)

or

(9.57)

Sec. 9.5

Mixing Processes

165

==

a(aG)
aT

Hence
(9.58)
Clearly

as > 0, as must be the case.

Entropy increase is to be expected when two different gases are mixed.


But what if the two gases are the same? Does the removal of a diaphragm sep
.
aratmg the two halves of a volume V filled with one kind of gas change its
entropy or not? Then

with

so that

and
LlS

nR In 2 > 0.

(9.59)

Sttppose that the properties of gas 1 approach those of gas 2. The pre\'l
ous result gives a finite entropy increase. However, if the two gases are identi
cal there can be no change in entropy when the diaphragm is removed
( LlS

0). This is the Gibbs paradox.

TI1 resolution will be fully discussed when we treat statistical thermody


namics. Suffice it to say now that there is a positive change in entropy when
different, distinguishable gases are mixed. The transition from differettt to
identical gases is not a continuous change, so Equation (9.59) does not apply
to the case where the gases are the same.

The Chemical Potential and Open Systems

Chap. 9

166

.
.

'

.c

'

.
.

PROBLEMS

9-1 For a one-component open system


tlU

==

TdS

PcJV

dn.

+ 11

(a) Use Eule r s theorem for homogeneous functions to show that

(b) Prove that (dG) r.r

==

:-::

TS

PV

+ JLn.

J.Ldn or equivalently,
aG

J.l

fin

-:--

r.P

(c) Prove the alternative definition of the ch e nt ical potent i al


iiF

-:--

li=

dn

f.\

9-2 (a) E xpress the chemical pote n t ia l of an ideal gas in terms of the te 1n pe rature T
and the volume V.

J.L

==

crT

<,.1 In T

(Hint: find the e n t ro py S

RJ' 1n V

snT

( r. ): use G
.

+ constant.

TS

Pl' and write

Gjn.)
(b) S imilarly. find 1-L in terms of T and P. Sho\v that the chemical po1entia1 at the
fixed temperature r v aries with press ur e as
J.1. =

J.L

-=

JJ..o +

( p\
RT In P,

\\'here J.Lo is the value of J.l at the rcfc_!rcncc poi n t ( P.,. T). This

ex p res si on

is

of great use in chemistry.

9-3 A con t a i ne r of volurne \t' is divided by p artit i on s into three parts containing one
kiJomole of helium gas two kilomoles of neon gas. and t hree kilomoles of argon
.

gas, respectively. The te m pera t ure of each gas is initiaHy 300 K and the pressure
is 2 atm. The p artit ions arc removed and the gases diffuse into one a n other

(a) Calculate the kilomolc fraction and the partial pre ssure of ea ch gas in the
mtxturc.

(b) Calculate the cha n ge of the Gihhs function and the change of the entropy of
the system in the mixing process.
9-4 Show that for

closed system consisting of t\\o phases coex isti n g in equilibrium

at a temperature T and under

p ressure P.

;,p
iV

T dP

==

--

Here dP/dT is the s l ope of the ph as e equ il i brium curve.

Problems

167

9-5 A mixture of gold :1nd thallium can exist in equilibrium with four phases pre
sent: solution, vapor, solid gold, and solid thallium. What is the variance? (This
point is kno\vn as the

eutectic point.)

9-6 The Gibbs phase rule can be generalized to systems in which there occur

chemical reactions:
f

1T

2.

Determine the number of degrees of freedom at equilibrium of a chemically


reactive system containing solid sulfur

S and the three gases 02, S02, and S03

The elements S and 02 appear in the reactions

3
02

-+

S03.

9-7 (a) Show that


as

dP

v
-

.
T

--

H.n

(b) For an ideal gas, the entropy S(H, P, n) has the form

S(H, P, n)
where <I>(H,

nR In

<P(H,n)
P

n) is a function of the enthalpy and the number of kilomoles.

Use the result of part (a) to derive the equation of state.

9-8 A container is initially separated by a diathermal wall into two compartments of


equal volume. The left compartn1ent is filled with 1 kilomol of neon gas at a
pressure of 4 atmospheres and the right with argon gas at 1 atmosphere. The
gases may be considered ideal. The whole system is initially at temperature
T 300K, and is thermally insulated from the outside world. Suppose that the
=

diathermal wall is removed.

(a) Wl)at is the nev temperature of the system? l11e new pressure?
(b) What is the change in the Gibbs function of the system?
(c) What is the change in the entropy of the system?

9-9 (a) Show that for an open system with one component,

dG

-SdT

VdP

JLdn.

(b) Using this result. calculate G for a van der Waals gas assuming a fixed
amount of material at a given, fixed temperature. Show that

-nRT ln(V-

nb)

n2bR
2n2a
V
nb- V

C(T),

where the integration constant C(1) is, in general, different for different
temperatures.

9-10 Show that, during a first-order phase transition:


(a) The change of entropy of the system undergoing the transition is a linear
function of the volume change.

The Chemical Potential and Open Systems

Chap. 9

168

(b) The change of internal energy is given by


AU

d In T

L 1

d In p

where L is the latent heat of transfom1ation.

9-11 (a) Sh ow that the Gibbs-Duhem equation 1eads to the relation

L x;dJ..l.;

( 7. P constant)

where

X;

is the kilomole fraction of the ith component.

(b) Sh ow that for a two-component system


-

d )n

Xl

dIL2

--

T.P

(c) Show that for a two-component liquid phase whose vapor can be treated as
an ideal gas mixture,
.

d In P...
L

d In

x1

r.P

.. d In

x2

r.t

where P1 and P2 are the partial pressures of the components of the vapor,

and

x1

and

x2

are the kilomoJe fractions of the liquid. (This is known as the

Duhem-Margules

q ua tio n )
.

ICS

I 0.1 Statements of the Third Law

171

I 0.2 Methods of Cooling

174

I 0.3 Equivalence of the Statements

175

I 0.4 Consequences of the Third Law

178

'

169

.
.

.
.

. .

_ .;,

.
.

...

.
.

I 0.1 STATEMENTS OF THE THIRD

. .
.

.
.

.
.

LAW

The third law of thermodynamics is concerned with the behavior of systems in


equilibrium as their temperature approaches zero. The definition of entropy
given by

Tf1Q
'+So
S=
0 T

(10 1 )
.

is incomplete because of the undetermined additive constant

S0, the entropy at

absolute zero. In this short chapter we shall introduce a principle that will enable
us to determine

S0 The

principle, which was discovered by Nernst in

1906, is

often referred to as Nemst's postulate or the third law of ther1110dynamics.


As long as we deal only with differences of the entropy, a knowledge of

S0 is unnecessary. However, the absolute entropy is by no means unimportant.


Suppose, for example, that we wish to determine the change in the Gibbs func
tion. Since

dG = -SdT + PdV, it is not suffjcient to know dS; we must know

S itself. In chemistry, moreover, knowing the absolute entropy

makes it possi

ble to calculate the equilibrium constants of a chernical reaction from the ther
mal properties of the reactants.
We have seen that a spontaneous process can occur in a system at con
stant temperature and pressure if the Gibbs function decreases. The Gibbs
function is related to the enthalpy by the equation
reciprocity relation

S = - ( aGI aT) p,

G = H - TS.

Using the

we obtain an expression known as the

Gibbs-Helmholtz equation:

aGG=H+T
aT

(10.2)

If this relation is applied to the initial and final states of a system undergoing
an isothermal process, it takes the form

(10.3)
171

The Third Law of Thermodynamics

Chap. I 0

171

--

-.----

Figure 10.1 Variation of .lG


and tlH in the vicinitv of
absolute zero.

L.....----

which shows that the change in enthalpy and the ch ange in the Gibbs function
are equal at T

0 for an isobaric process. Based on the results of experi

ments, Nernst postulated that as the temperature tends to zero .. not only do

llG and llH become equaL but their temperature rates of change approach

zero as well:
A(G)
lim
r-t,
AT

()

ri(H)

lim

iiT

T-+0

I'

0.

( 10.4)

This behavior is illustrated in FiQure 1 0.1 .


We may write the first of the two equations as
.

.....

iiGl

lim

ac1

il1'

T -+0

==

aT

0.

( 1 0.5)

where the subscripts 1 and 2 refer to the initial and fi11al states, respectively.
From the reciprocity re]ation we have

J1jm, ( 5' 1

-+I,

5 ) = 0.

( 10.6)

This is the Nernst formulation of the third law :


All reactio1zs in a litJUi{/ (Jr solid in tl1er1nal equilibriutn take place lvith no
change o.f entrop_r in the J1e(hhJriJlJlJtl o_f'a/JsrJ/ute zero.

Although the Nernst theorem is quoted for solids

and liquids, it is believed to

hold for all sy ste ms in equilihrium states. (ll1ere arc certain quantum systems

that constitute gas-like

aggregations at absolute zero.)

In 1911 Planck extended Nernst's hypothesis by assuming that it holds


for G1 and G2 separately. Specifically, he prOJ.1osed that
lim G(T)
ro

==

lim

T o

lf(T),

(I 0. 7)

Sec. I 0.1

173

Statements of the Third Lav1

and

I.1m iJG

r-.o

Ilffi

aT p

ali

aT

r-.o

(10.8)

These statements lead to an important conclusion. For convenience, we tem


porarily introduce a variable <I>

T dG
iJT

H. Equation (10.2) then becomes

-<I>= 0.

(10.9)

By Equations (10.7) and (10.8), we have in the limit when

<I>
Adding the term

a<I>

and

iJT

0,

0.

(10.10)

T ( aHI aT) P to both sides of Equation ( 10.9), we get


T

d<l>

<I>

, aT p

dH

ar p'

or

a<t>
aT p

<I>
T

--

ctJfT

aH
aT

T = 0 is the same as the


(iJ<I>/ aT)p, which is zero, according to Equation (10.10). Therefore,

By L'Hopital's rule, the limit of

lim

r_.o

for

aH
aT p

==

0.

( 10.11)

limit of

(10. 12 )

Finally, by Equation (10.8), we obtain the result

G
lim .a
r-.o aT p
Since

(iJGfiJT)p

0
.

(10.13)

-S, it fqllows that


lim S

T-+0

0.

(10.14)

This is Planck's statement of the third law. In words:

The entropy of a true equilibrium state of a system at absolute zero is zero.

Chap. 10

174

The Third Law ofThermodynamics

This statement. stronger than the Nernst theorem. needs qualification. Certain

glasses have a nonvanishing entropy as the temperatre tend to absolute


zero. This makes sense.. since statistical thermodynamics assoctates entropy

with disorder. and glass is a disordered structure. The Planck statement can be
shown to be valid for every pure crystalline solid. Quantum statistics is necessary for a complete understanding of the absolute entropy of a system.
Still another statement of the third law is:
It is inlplJSSible

(0

retlllc.:e the le111perature oj'a swvstet11 to absolute zero

llSi11g a finite nttnzber lJfprocesses.

In Section 10.3 this so-called unattainability principle will be shO\\'n to be


equivalent to Nernst"s postulate.

..

'

'

. .

.
.

I 0.2 METHODS OF COOLING


A widely used method for cooling

substance is to isolate it from its sur

roundings (with no heat flow in or out) and reduce its temperature in an adia
batic reversible process. Then work is done J,y the system solely at the expense
of its internal energy. This can be accomplished by varying some parameter
such as the magnetic field. The process is called adiabatic demagnetization and
is discussed in Chapter 17.
In the 1970s laser coolitlg was proposed a process in which the decelera
tion of a beam of atoms is accomplished t'y directing

laser beam so as to

oppose the atomic heam. Under the action of the laser beam, momentum is
transferred to an atom by absorpti<>n of a photon. The laser light frequency is
chosen to excite the lowest order at(lmic transition. If the laser frequency is lower
than the atomic resonant frequency and the atom is moving against the laser
beam. the laser frequency in the rest frame of the atom is Doppler shifted
,
toward resonance. The key point is that the laser light pushes on , <>nly those
atoms that are moving into the laser beam.
-

Following absorption , the atom spontaneously emits the photon in a random direction. This absorption and isotropic reradiation result in an average
driving force in th e direction of the incident light. Since the force is pr(JportionaJ

to the velocitv, the atomic ffi(ltion is slov.'ed and the atoms are therebv cool ed .

In this process, the atomic vapor is cooled without becoming

liquid or a solid.

Various methods, including laser detuning, hav e been used to compen


sate the changing Doppler shift as the atorns decelerate so as to keep them
near resonance. Three-dimensional cooling results if ()rthogonal pairs of laser
beams are used in a so-called h<.>ptical m<.>lasses" configuration.

A thorough understanding <lf the cooling mechanisms involved has


made it possible to conduct novel experiments on cooled atoms and to achieve
temperatures of a fraction of a microkelvin. the lowest ever measured.
'

Sec. 10.3
I O.l

Equivalence of the Statements

175

EQUIVALENCE OF THE STATEMENTS

By considering the adiabatic reversible process, we can show in a few teps the
equivalence of two statements of the third law. We assume that absolute zero
is unattainable and prove that in a change from equilibrium state

1 to equilib

rium state 2,

(10.15)
where the naught subscript refers to T

0. We can examine the process in the

entropy-temperature plane. The shape of the S versus T curve can be found


from the relation

TztQ

So +

(10.16)

Now Cv

(71QfdT)v

and Cp

(71Q/dT)p. We assume

that either P or Vis

held constant and drop the subscript. Thus

S =So+

T dT
0

(10.17)

CT .

We note that the Debye law for the heat capacity of a solid gives Cva.T3 In
this case, S increases \Vith T according to some power law, as Figure 10.2

reflects. The curve S1 applies to an initial value of the variable parameter and
s2 to some final value.

Figure 10.2

Entropy
temperature diagram for a
hypothetical cooling process in
which S1 < 52 (The entropy
curve for the initial value of
the varied parameter lies
below that of the final value.)

____

L
1
I
I
I
I
l

_
____
____

I
I
I
I
t
I
I
T

Chap. I 0

l76

The Third Law ofThermodynamics

Cooling from T1 to T2 by an adiabatic (and isentropic) reversible process


occurs along the horizontal line connecting the sl and s2 curves. Then s2

sl

gtves

(10.18)

Transposing terms.. we have


Tz

dT

(10.19)

()

Now, Cis always positive: if it were negative, a heat gain would result in a low
ering of the temperature.. and the attainment of absolute zero would be triv
ially easy!

501) > 0, as shown in Figure 1 0.2 and because C > 0, there


If ( S02
must be some temperature Tj for which the right-hand side of Equation
(10.19) is zero. Then
-

( 1 0.20)

where

T is the final temperature corresponding to T,.

But

T must be zero if

Equation (10.20) is to hold. Thus absolute zero could be achieved through the

adiabatic change from state 1 to state 2 if 50 > S01 This violates the unattain
ability principle. We conclude that

(10.21)
Thus we need to redraw the 5- T di a gram with S02
abat s2

sl gives Equation

r2

dT C,- J1- -

0 there is some value of 1"2


hand side of Equation ( 10.22) is zero. Then
..

5.,1 (rigure 10.3). The adi

(10.18) as before .. \Vhich can be rewritten as

If (SoJ - So2) >

<

( 10.22)

say T2"

such that the left

Sec. I 0.3

Equivalence of the Statements

177

Figure 10.3

Entropy
temperature diagram for a
hypothetical cooling process in
which sl > s2.

- - ---------------

S'r.1 _____ L_________


I
f
I
I
I
t
I
I
I
I
I
r1"

I
I
I
I
I
I
I

------11---.

----
=

r1

T2" T2

0. This also vioiates the unattainability principle and we


conclude that for this case
requiring that T1"

So2 >Sot

(10. 23)

The only way that Equations (10.21) and (10.23) can both be true is if

(10.24)
which is the Nernst postulate. According to Equation (10.24) the curves of
2
and S1 versus T must come together at T- 0. The final sketch, Figure 10.4,
nicely illustrates the equivalence of the two statements of the third law. The
curves have been drawn to reflect the Planck statement, though the proof does
not place the entropy at absolute zero.
A series of isothermal and adiabatic processes (demagnetizations) are
represented by the zigzag path in the figure. Each successive process reduces
the temperature. It is clear, however, that because the curves intersect at

Figure 10.4 Entropy


tetnperature diagram for an
actual cooling process. The
unattainability of absolute
zero is illustrated by the
indefinitely increasing number
of steps required to achieve a
given temperature reduction
as absolute zero is approached.

The Third Law of Thermodynamics

Chap. I 0

178

0, an infinite num ber of steps wouJd bt required to reach absolute zero.

\\7e conclude tha t it is impossible to achieve

te m peratu re of absolute zero by

any practical method. This fundamental fact of nature should not be espec i al l y
disquieting, however. given that cry(lgenics has enabled us to get ex t re m ely
close to T

0.

I 0.4 CONSEQUENCES OF THE THIRD LAW

1. Expansivit.'
The expansivity (coefficient of volume expansion) is defined as

av

From Section 8.5 we have the Max\\'ell relation

'
-

--

iiT

, (J p

I'

The Nernst postulate implies that the change in e nt ro p y is zero in any


process at absolute zero: that is.

lim
/'_,.()

(J
--

jp

(1

0.

r
'

Since the volume V remai n s finite as T


Jim [3

T --+(l

==

0, it is therefore true that

0.

( 10.25)

We conclude that in the case of all kno\\'n solids, the expansivity


approaches zero as the temperature approaches absol ut e zero.

2. Slope of the phase transf'or11IlllirJn curtcs

In con n ect i on with any phase change that takes place at low ten1perature
we can invoke the Clausius-Clapeyron cquat !_<)n (Section

tl p

--

8.7):

==

-- ---

The Nernst theorem states that

lim 5

T--+0

0.
,

Since V is not zero for so-caJJed firstorder transitions.. it follows that

Problems

179
dP
.
I1m
T --+0 dT

0.

(10.26)

The slope of the boundary between two phases is zero at absolute zero.
This has been verified for all known sublimation curves, for the vaporiza
tion curve of Helium II, and for the fusion curve of solid helium.
1
Experimental results show that for solid helium, dP/dT'XT , which
approaches zero at a very fast rate indeed.
3. Heat Capacity
From the definitions of the heat capacities at constant volume and con
stant pressure, we have

Cv = T

as

Cp = T

ar v'

as
ar p

Integrating, we obtain
T

S- S0 =

dT
Cv
,
T
0

S- S0 =
0

dT

Cp
T.

The left-hand sides of the equations will remain finite down to absolute
zero, where, by the third law, they will vanish. The right-hand sides must

behave in the same way. The integrals show that the heat capacities must
decrease to zero at least as rapidly as T. Othenvise the integrals would
diverge because of the lower limit of T = 0. We conclude that
limCv
T --+0

0,

lim Cp
T -+0

0.

(10.27)

That is, the heat capacities approach zero as the temperature approaches
absolute zero. With respect to this behavior, extensive experiments pro
vide a full confirmation of the third law.

PROBLEMS

10-1 Show that the heat capacity at constant volume can be written
Cv
It follows that as T

0, InT

-X)

as
=

dInT v.

and S

0, so that Cv

Equation (10.18). By the same argument, Cp

0 as T

0.

0, in confirmation of

The Third Law of Thermodynamics

Chap. I 0

180

102 Show that. in view of the third law. neither the id e al gas equation nor the van
0.
der Waals e quation can hold at T
=

103 (a) Show that the third Jaw cannot prdict the value of the isothermal com
pressibility K at absolute zero. (It remains finite.)

(b) The isothermal bulk modulus B is the reciprocal of K.


8

aP
- v -av

Show that

riBlim
T o dT
.....

0.

te-4 According to De bye's theory of heat capacities. to be discussed in Chapter 16. th e


specific heat capacit y of an insulating solid at very low temperatures is given by

12

c,.

where

c,.

1r.t

-- -

0.01 K and T

tJD

is in the usual units of J kilomol..!

such a solid at T

K 1 Find the specific entropy of


I OK, given that 8D
300K.
1

10-5 (a) Show that if c\.


hTU at low temptratures. the third law requires that
a> 0.
..
(b., If ( \.'
aT + hT-"' at lo\\ temperatures, calculate the variation of the
=

entropy with temperature.

10-6 Consider a solid whose equation of state is


PV
where

.f(\1).

AU.

f ( V) is a funct io n of the volume nnlv and A is a constant. Show that

Cv OasT o.
10-7 A low-temperature physicist \\'ishes

to

publish his experimentaJ result that the

heat capacity of a nonmagnetic dielectric material between 0.05K and 0.5K


varies as AT112
publication?

BT3 As editor of the journaL should you accept the paper for

ases

I 1.1 Basic Assumptions

183

I 1.2 Molecular Flux

186

I 1.3 Gas Pressure and the Ideal Gas Law

188

I 1.4 Equipartition of Energy

190

11.5 Specific Heat

191

acity of an Ideal Gas

I 1.6 Distribution of Molecular Speeds

193

I I.7 Mean Free Path and Collision Frequ en cy

198

I 1.8 Effusion

201

I I.9 Transport Processes

203

Ill

11.1 BASIC ASSUMPTIONS


In the nineteenth century, scientists realized that in order to go beyond the
limitations of classical thermodynamics, we must take into account the
structure of matter_ The recognition that bulk matter is composed of parti
cles led to the kinetic theory, which assumes that atoms and molecules obey
the same laws of mechanics that apply to macroscopic systems. This assump
tion was ultimately challenged by quantum theory and, indeed, statistical
thermodynamics is best treated from the quantum mechanical point of
vtew.

Nonetheless, kinetic theory yields far deeper insight into such concepts

as pressure, internal energy, and specific heat than classical therttlodynamics


provides, and is therefore well worth investigating. In addition, it gives

an

explanation of transport processes such as viscosity, heat conduction, and dif


fusion that is quite satisfactory for most purposes.
Kinetic theory, as the name implies, is concerned witl1 the motion of mol
ecules rather than the forces between them. It is therefore especially suitable

for describing the behavior of gases at relatively lo\V pressures.

The. fundamental assumptions of the kinetic theory of gases are:


1. A macroscopic volume contains a large number of tnolecules.
What do we mean by this statement? We know that Avogadro's number is

NA

6.02 x

lc6 molecules kilomole-1

and that at standard temperature and pressure (STP), <lne kilomole of a


gas occupies a volume of 22.4 m1. This gives a molecular density of about
3 x 1025 molecules m-3 or 3 x 1019 n1olecules cm-3 Even at high vacu

ums down to 10-11 atmospheres of pressure, at ordinary temperatures


there are still many millions of molecules in each cubic centimeter.

2. The separation of molecules is large compared with molecular dimensions


and with the range of intermolecular forces.

183

The Kinetic Theory of Gases

Chap. II

184

Take a volume of 22.4 m3, divide it into cubical cells with one molecule in
each ce)J. The volume of each cell (at

3
22.4 m f6.02

is

STP)

<

2fl
0
1

!!:::;

30 x to--27m3 Thus the distance between molecules in this case is equal

to the cube root of this volume.. or roughly 3 X Io-"J m. This is a factor of


about 10 greater than the diameter of a typical gas molecule.

We must also consider the range of the forces of interaction between


molecules.. which

._

are

electrical in nature. The force between two mole-

cules is strong and repulsive at separations approximately equal to the

molecular diameter and is associated with the overlap of electron clouds.

At greater distances the force is attractive and is due to induced electri

cal dipoles. The behavior is often characterized by the total potential


energy
d

12
-

(11.1)

where fl. and d are empirically determined parameters. Equati<>n

is known as the Lennard-Jones. or

6--12""

potential. and is pJotted in

Figure 11.1. The curve has a minimum at


V ( r)

( 11.1)

fd

2 Lh

1.12..

where

-E. To the left of the minimum. the curve is very steep; to the

right it is comparatively flat. Typically. E lies in the range of 2 x 10 4


2
eV, and d v aries between 2.2 x 10 -Jn m and 4.5 x 10-10
to
5
X
10eV

m. For molecular separations of an order of magnitude greater. the inter


action force is evidently negligibly small.

In a liquid or solid, the molecules have nearest neighbors at a distance

only slightly greater than d. Thus there is substantial interaction between


them. In contrast. the molecules of a p.as under normal conditions are

most of the time traveling freely through space. An ideal gas is a good
10

6
\l(r)i.1E

,
-

Figure 11.1

Variation of the

intermolecular potential V ( r)

with the intermolecular


distance r for a pair <>f gas

molecules.

()
_.,

'

--

\.

-J.o

1.0

1.4

rltl

l.R

Sec. I 1.1

Basic Assumptions

185

approximation at these low densities; dry air at

STP deviates from the

. ideal gas law by iess than 0.1 percent. For high density gases, on the ott,er
hand, it is essential to include the intermolecuiar forces, as

is done in the

van der Waals model.


3. No forces exist between molecz,les except those associated with collisions.

That is, molecules move in straight-line paths between collisions with


constant velocities.

Each molecule moves forever in a zigzag path, for it

collides many million times per second with other moecules of the gas
or with the molecules that make up the walls of the container. 'lltese col
lisions are so brief that in our analysis we entirely ignore their duration.
F urtherrnore, the distance a molecule tnoves between collisions can lie
between zero and thousands of times the molecular diameter. However,
he very short and very long distances are unttsual. The average distance

between collisions is called the mean free path. In oxygen at

STP, the

mean free path is about 10-7m.


4. The collisions are elastic.

Both momentum a11d mechanical energy are conserved in the collision

process.

S. The molecules are uniformly distributed within a container.

6. The directions of the velocities of the molecules are uniformly distributed.


The last two assumptions characterize the randomness of the motion.
The basic idea of kinetic theory is that the n1olecules of the gas are moving
rapidly and randomly, colliding \Vith each otl1er and with the walls of the con

tainer, thereby creating

pressure due to their countless impacts. We are con

cerned with a gas that is in thertnal equilibrium at a fixed teu1perature and in


mechanical equilibrium with the walls. It follows from these assumptions that

the average molecular speed and the probability distribution of speeds are
unchanging.
v +

Let f( v )dv be the fractional number of molecules in the speed range v to

dv. The probability density function f ( v) has the dimensions of v- 1 since

f(-v )dv

1.

The mean speed is given by


X

v=

vf( v )dv,

(11.2)

and the mean square speed is


V"'
.,

(11.3)

Chap. 'I I

186

The Kinetic Theory of Gases

The square root of

v2

is ca11ed the root mean sqt1are or rms speed:


( 11.4)

The nth moment of the distribution is defined as

We shall also be interested in the most probable speed, obtained by differentiat


ing f ( t') and setting the derivative equa1 to zero. The speed distribution wiJI be
derived in Section 1 1 .6. We can develop the fundamental aspects of the kinetic
theory, however, without knowing the exact form of the distribution function.

11.2 MOLECULAR FLUX

In many applications of kinetic theory.. we need to know the molecular flux,


the number of molecules striking a unit area per unit time. Consider a slanting
cylinder of slant height vdt at angle fJ m ea su red with respect to the normal to
the elementary area dA (Figure

11.2). We want to find the number of mole

cules that strike the area dA in time dt.


z
/

'

/
I

udt

'

/
/
.

./

/
/

/
'

dA

,_.---.,.'---- . ---+-------...
'

.......

.....

/cb

--<....

Figure 11.2

Slant cylinder geometry used to calculate the

number of molecules that strike the area dA in time dt.

\'

Sec. 11.2

Molecular Flux

We define n

187

NJV, the number of molecules per unit volume. Now, the

number of molecules with speeds v to v + dv in the cylinder is ( v) dv


X (volume of the cylinder). Since the volume of the cylinder is (d A cos8)( vdt),
the number of molecules with speeds v to v + dv in the cylinder is
n v
::ra

f ( r) dv

cos8 dA dt. Only a few of these molecules are traveling toward

dA. Let the velocity of each molecule be represented by a vector, and imagine
that all the velocity vectors are moved parallel to themselves to a common ori
gin (Figure 11.3). Assumptio11 6 states that the head ends are distributed uni
formly in a spherical shell of radius v and thickness dv. The element of area of
the surface of this sphere is

du

v2 sin8

d8 d</J,

with

sphere

Therefore, the fraction of the total number of molecules with angular distribu
tions 8 to 8 + dO,</> to </J + dcp, is

sinO d8 d<P

du
---

sphere

----

41T

du

The number of molecules in the cylinder at time t traveling with speeds v to

dv toward dA (i.e., the 8, 4> molecules) is


dN

sin8 d8 dl/>
.
nv f(v)dv cos8dAdt

(11.5)

47T

;:s::

Figure 11.3

(a) Randomly
oriented molecular velocities;
(b) the same vectors referred
to a common ortgtn.

dv
(a)

(b)

188

Chap. II

The Kinetic Theory of Gases

The corresponding particle flux is by definition

dA dr

411"

Tr/2

sin8 cos8d8

vf ( v) dv

==

<t>

(11.6)

(11.5) in Equation (11.6) and integrating we obtain

Substituting Equation

<I>

dN

(11. 7)

()

or

(11.8)
where vis the mean speed, defined in Equation
.

.
.

. .

..

(11.2).

11.3 GAS PRESSURE AND THE IDEAL GAS LAW

Next we consider a single molecule colliding with a surface. The normal com
ponent of the change of momentum is (Figure

mt' cosO

n1v

cosfJ)

1 1 .4)

2mv cos8.

The change of momentum dp produced by all the molecules with speeds

'

dv traveling toward an element of the surface dA


dp

v to

in time dt is

2mv cosO dN

mv

cosfJ

ll1'f(t')dvcos6 dA

dt

sin6 dO dQ>

----

nmdA
=

---

27T
1

dt

.,
lf. -

v2f ( v )dv
()

tic/>

sin8cos28 d9
()

From Newton's second law and the defi11ition of the pressure P. we know that
dp

dAdt

P.

189

Gas Pressure and the Ideal Gas law

Sec. 1 J.3

dA

6
------

'
I
l
I
\
I

\
I
I

Figure 11.4 Change in


momentum of a molecule
striking a surface element dA.

m"

mv

'

so that

P
We note that

nm
-=a

1
-nmif.

(11.9)

3-

N m/V. Then

PV

1
=

1Vmv

2
=

(11.10)

-Nmv

The right-hand side is two-thirds of the average translational kinetic energy of


the molecules of the gas.
Equation (11.10) is highly suggestive of the ideal gas law

PV

nRT.

(11.11)

The number of kilomoles is simply related to N:

no. 0 f k.l
1 omo 1es

no. of molecules
=

N
_

no. of molecules kilomole- 1

-N-A'

where NA is Avogadro's number. Thus

PV

NkT,

(11.12)

where

Here k is Boltzmann's constant; it is the ratio of two universal constants and is


therefore also a universal constant. We see that Equation

(11.12)

will agree

190

Chap. I f

The Kinetic Theory of Gases

with Equation (11.10). the result we derived from kinetic theory if we make
the association

NkT

==

I
-Nmv2 ..
3

or

(11.13)

Thus, the theory has provided a statistical interpretation of the absolute tem
namely the te m pe ra tu re is proportional to the aver

perature of a dilute gas

..

age kinetic energy of a molecule. The r esu lt is one of the great insi ghts given to

us by the kinetic the or y

From Eq u ation (11.13) it is apparent that the mean kinetic energy is

independent of the pressure .. the volume.. or the molecular species. At a given

temperature, the mean kinetic energies of different gases are all the same.

It is instructive to note that at room ten1perature (293 K),


.

(11.14)

\Ve can compa re this energy with E. the depth of th e potential

I>

well"' dis

cussed at the beginning of the chapter There we observed tha t E has a mag
.

nitude ranging from about 0.0002

eV

to ().05

e V.

I 1.4 EQUIPARTITION OF ENERGY


Let the mol ec u lar velocity have cornponents ( vx
v2

=-=

v2,

vY, vz ) .

Then

v2Y + v2,.

since taking an average is a linear op e r at i on By assumption, there arc no pre


.

ferred directions, so

-2
V

-2

== V

::_:

-2
V

-=

I ,
- V".
3

Thus the mean kinetic energy per molecule aSS(lCiatcd with any one component
the x-component, for example
I

is

-- nzv.;.X

'\

1
- n1v

..

l
=

kT.

Sec. I 1.4

Equipartition of Energy

191

stnce

-mv2
2

3 ,
-kr
2 ,

(11.15)
.

It follows that the average translational kinetic energy associated with each

component of velocity is one-third th totai and has a magnitude


Each component represents a

of kT /2.

degree of freedom f of a mechanical system. An

atom or a monatomic molecule, for example, has three translational degrees of


freedom so that f

3.

A diatomic molecule is modeled as tv.'o masses connected by a s pri ng

As such, it has seven degrees of freedom: three translational, two rotational,


and two vibrational. There are two rotational degrees of freedom, because
t\\o angles are needed to specify the orientation of an axially symmetric mol

ecule. The vibrational motion is represented by a one-dimensional harmonic


oscillator. For simple harmonic motion, the energy is partly kinetic and partly
potential; the average values of the two are equal, and both are expressible as
quadratic forms. Thus these are two vibrational degrees of freedom.
The equipartition theorem (proved in Chapter 14) states that (1) there is
one degree of freedom associated with every form of energy
quadratic fortn, and

(2)

of the energy of

the mean value

sponding to each degree of freedom is

expressible in

a molecule corre

kT/2. Thus the total average energy of a

molecule 8 is f kT/2. It follows that the internal energy of a gas must be the
total average energy of its N molecules. That is,

(11.16)-.
..

or

1_RT.

(11.17)

11.5 SPECIFIC HEAT CAPACITY OF AN IDEAL GAS


For a reversible process the specific heat capacity at constant volume is given by

From Equation

au
=

(11.17), this gives


Cv

d f__RT
T
2
d

The Kinetic Theory of Gases

Chap. II

192

TABLE

11.1

Ratio of specific heat capacities of various

values
the velocity of so u nd in the gas.

gases at temperatures close to room temperature. The


a re obtained hy measuring
Gas

Gas

Gas

co,-

He

1.66

H"'

1.40

Ne

l.M

0"

1.40

NH

Ar
Kr

1.67
1.69

f\.,
co

1.40
1.42

CH_.
Air

Mayer's equation for an ideal gas is

Cp

and the ratio of specifi heats

f-R
2

R. Hence,

c,. +

j' + 2\

R=

is

y=
For

cp

1.29
1.33
1.30
1.40

Cp
C I'

f.

mo n ato mi c gas,

c,.

= iR.

Cp

:;R.
I-

'Y = 1 =

1.67 .

...

These results are in good agreement with mfasured values at tem per atures in
the vicinity of room temperature ( Ta ble 11.1 ).
For a diatomic eas..
'-

'Y = 7 =

1.28.

Here the agreement is poor: the number of degrees of freedom is too large.
We try
-

.f

5.

c,.

= 2_R.

c I'

==

7
-

..

y =

7
-

==

1.40.

Good agreement is obtained wi th this adjustment. Evidently near room tem


perature the rotational degrees of freedom

(Jr

dom are excited, but not blJth. Maxwell

C(lnsidered

the vibrational degrees of free

the resolution of this

q ue st io n the most pressing challenge confronting the kinetic theory. It took


statistical thermodyn ami cs to provide the final answer (see Ch ap ter 15).

Sec. 11.6

Distribution of Molecular Speeds

193

Two other observations are noteworthy. For large molecules f will be


large and y 1. In general,

<

}'

<

1.67,

in pretty fair agreement with experimental resu.lts.

The kinetic theory predicts that


We have seen that this is not true

c.,

and

cr

are constants for an ideal gas.

at low temperatures

when

both must

approach zero. Kinetic 1 heory, which is a classical theory, is incapable of


accounting for this behavior. A quantum mechanical description is required.

DISTRIBUTION OF MOLECULAR SPEEDS

11.6

We have yet to derive_ the probability distribution of molecular speeds from


which various averages can then be found. In Chapter 14, the so-called
Maxwell-Boltzman distribution will be determined from a knowledge of the
distribution of particles among energy levels. We incl11de here a more

physi

cally satisfying derivation that makes reference to the molecular collisions

through which the speed distribution is maintained.*


In a dilute gas, it is reasonable to assume that collisions are. binary
events, involving only two molecules. Simultane,Jus collisions of more than
two molecules are exceedingly rare. In a l'inary collision, the vel(lcity of parti
cle 1 is changed from

v1 to v and the

velocity of particle 2 is changed from

v2

to v. (Here we use boldface symbols to denote vectors.) For identical point


masses, conservation of momentum implies coplanar collisions, with
+

Vr

v2

v]

(11.18)

v2.

The assumptions of kinetic theory ensure that under equilibrium conditions,


for each binary collision there is an inverse collision in which the original
velocities are restored (Figure 1 1.5). The velocity vectors are collinear with the
original set.
If we let F ( v)
range

v to

1 and 2 per

dv

be the number of molecules with velocities in the

dv, then the number of collisions occurring between molecules


unit time will be aF ( v1) F ( v2), whe.re a is a constant of propor
+

tionality.
For tlte inverse collisions, the number wili be a' F ( v' 1) F ( v' 2). Because
the velocity distribution is unchanged by the total of all the elastic collisiorts,
the two rates must be the same. A.lso, the collisions will be completely equiva
lent when viewed in

center of mass frame of reference, so a

*See. for exampJe. Thermal Ph vsic'i,


York.l991.
..

by

M. Sprackling, A1nerican

a'. T hus

Institute of

Physics. New

The Kinetic Theory of Gases

Chap. II

194

th

..

(a)

v.,

v,

(a) A binary
molecular collision; (b) the

Figure 11.5

(b)

corresponding inverse collision.

or
(11.19)
Since kinetic energy is conserved and
.,

.,

==

l v 12

..,

.,

v,

we also have

(11.20)

vl- + v2- = vl- -1- v2-

The functional relation of Equation ( 11.19), with the constraint of Equation


{! 1.20), can be solved by inspection.* A solution is
F ( v)

A exp (

- crv

v),

(11.21)
.

where A and

are constants. This is the Maxwell-Boltzman velocity distribu-

tion. To obtain a

speed

distribution, we let N ( v )d1) be the number of mole-

cules whose speeds lie in the range


Figure

to

v +

dv, irrespective of direction. As in

11.3, we can visualize the velocity vectors as displaced parallel to them

selves with their ends at a common point. Since the directions are random, the
velocity distribution has spherical symmetry. 'Therefore N ( v )dv is the number

*The solution can also be found hy


Chapter 13.

the method of Lagrange multipliers discussed in

Sec. I 1.6

Distribution of Molecular Speeds

195

of velocity vectors whose tips lie \Vi thin the spherical shell between radii v and

dv. It follows that

(11.22)
The constants A and a can be found from two conditions. The total number of

particles N is given by

X.

N=

N(v)dv,

(11.23)

0
.

and the total energy of the molecules in the container is

(11.24)

To determine

a we

can divide Equation

Equation (11.22). The result

7.

r
t
a
v4e dv

v2N(v)dv
0

by Equation

(11.23) and use

is

3kT

(11.24)

v2e -avdv

2
1
)
8a
1T

(11.25)

2a

2
1
J
4a

--

..

t-J(v)dv

1T

(The values of the definite integrals are given in Appendix D.) Thus, a

mj2kT.

The constant A can be then found from Equation (11.23):


X.

N (v)dv

N=

(11.26)

47T A
0

Hence

A=

12

1T

In

--

2rrkT

!t/2

N.

(11.27)

Finally,

N(v)dv

41TN

21rk T

The function N( v) is shown schematically in Figure 11.6.

( 1 1 . 28}

The Kinetic Theory of Gases

Chap. II

196

N( l')
r,

Graph of the
Maxwell speed distribution for
three temperatures:
T3 > T2 > T1.

Fig11re 11.6

If we divide Equation (11.28) by N \Ye normalize N(v) and obtain a


N ( v) I .V that can then be used to obtain
probability density function f (v)
=

some useful average values. As noted in Section 11.1, the definition of the

mean value of vn is

v"

V11j' ( v) dv,

where f ( v) is the probability density f u nc ti t:>n

Three

speeds that characterize

the Maxwell-Boltzmann distribution arc of 5pcciaJ importance:


'

1.

Mean speed:

..

vN('-')tlv

--

2r.kT

8kT
1Tn1

( 1 I .29)

--

sqttare speed:
The mean square speed is given

41T --

by

1'11

21TkT

3kT_
-

nz

2. Root mean

47T -

1
-

.
.

v4enl'.tlv

Sec. I I .6

Distribution of Molecular Speeds

197

The root mean square ( rtns) speed is, then,


3kT
m

(11.30)

3. Most probable speed:


This is the value

Vm

for which N (v) j N is a maximum, or for which

d
-(v2e-av)
dv

a=

"

The derivative is zero for

vm

'

m
--

2kT.

12
, which yields
1
)
a
1/
(

112
2kT
=

112
kT
1.414 m

(11.31)

It is evident that (Figure 11.7)


Vm

: V:

V,ms
=

For a nitrogen molecule

1.414 : 1.596:

1 : 1.128

(m

4.65

1.732

1.225.
X

(11.32)

10-26 kg) at T = 273

K,

vrms

493 ms -t, which is slightly greater than the speed of a sound wave in nitrogen
gas. Numerical values of the mean speed and root rnean square speed for vari
ous gases are given in Table 11.2.

N(v)
N

0.6
0.4

- ...
--6---"'- ....
---+---

----'-_....__._ --+-- -6--+----.---+---4---+-_.... .....____.___

o
0

Fi

2.0

0.4

0.8

1.0

1.2

1.6

11.7

The Maxwell-Boltzmann speed distribution function


at a particular temperature showing the most probable speed vm,
the mean speed v, and the root mean square speed Vrms

The Kinetic Theory of Gases

Chap. 11

198
TABLE 11.2 Molecular
Speeds at 273 K. in ms 1
Gas

v,ms

1690
1210
570
530
450
420
400

1840
1310
620
580
490
460
430

'

There are other speed distributions of less interest to us one is the num
ber of particles whose t-component of velocity lies in th e range

vr f()

v.\

dv,

regardless of the values of the other components:

(11.33)
( Fi g ure 11.8). Clearly

vx

vY

vz since the velocity directions are assumed to

be un i formly distributed over all a ng 1 e s in s p ace

I I. 7

MEAN FREE PATH AND COLLISION FREQUENCY

We are interested in ho\\' far the molecules of a gas travel bet\\'cen collisions
and how often they undergo collisi<)ns. These concepts are called .. respectively,
the mean free path ( and the coJJisi<)n frequt-ncy f ..
Suppose that the radius of

m ole cul e of the gas is

R. Then a collision

occurs when one molecule approaches another \\'ithin a center-to-center dis


tance 2R. In time t a moving molecule sweeps out a cylindrical exclusion''
volume of length vt and cros-s-sectional area

a =

rr( 2R)2 (Figure 11.9). If the

nei gh bo ri n g molecules were at rest. the nun1ber <)f molec ules with centers in
this vol ume would be

na vr

-
.

and the mean free path would therefore he

VI
-

na vr

nu

This answer is only approximately correct hecause we have used the _mean
speed v for all the molecules instead of performing an inte gration over the
Maxwell-Boltzman speed distribution. If that is done, and the most probable

Sec. 11.7

Mean Free Path and Collfsion Frequency

199

N(v_r)

--------- ------- V

Maxwell-Boltzman distribution function for the x


component of velocity for three temperatures: T.1 > T2 > T1 (The
Figure 11.8

The

distributions for they- and z-components are identical with this


distribution.)

speed is used in place of v, then


e

( 11 34)

--;::=--- ,

nu
7T==
'

which is the correct expression for the mean free path. The corresponding c<?l
lision frequency is

f -

Vm

(11.35)

Note that the mean free path is independent of the molecular speed and
therefore of temperature. Since the nun1ber of molecules per unit volume

!:! is
--

directly proportional to the pressure, the mean free path increases as the pressure decreases. For an oxygen molecule, 2R
temperature and pressure,n

:;

3.6 x 10-10 m. Under standard

2.7 X 1025m-andvm
-=.-;

5 7 x lo-s
.

370ms-1.Thus

and

The average time between collisions, the reciprocal of the collision frequettcy,
is approximately 1.5 x 10-10 s. Note that f is the order of 300R, validating our
kinetic theory assumption that the separation of molecules is large compared
with molecular dimensions.

The Kinetic Theory of Gases

Chap. II

200
2R

I.

'

'

Figure 11.9

Exclusion volume swept out in time t by a


particle of radius R traveling with average speed v.

The actual distance that a molecule \\ill travel between collisions is sub
ject to a large statistical variation, independent of time. Let the probability
that a collision will occur in a distance tlx he
collision will
-

alLt.

not

adx.

Then the probability that a

occur \\'hen the molecule travels a distance

dx

will be 1

If p(J) is the probability that a molecule \\'i1J travel a finite distance'

before making a

co1Jision, then p(J.-) ( 1


adt) is the probability that a
cule wi11 travel a distance x + dt before eXJ)eriencing a collision . Thus

Expanding p(:r
have

a Tay lo r se ries a nd r et a i ning the first two terms, we

dx) in

p(J:)

mole

d p ( .'( )
1- --,J.r

lX

p(.r)- aJJ(.t)tlx.

or

clp(x)
- -

-- -

tlx

aJ> ( .r ).

Integrating. we obtain

\\'here A is a constant. Since p(O) must he

p(x)
To determine the

f(x)

m ean

I d p( .t) I dxI

"'

1 and

free path. we need the probability density function

p(x). Because p(.r)


f f(.r)dx. it follows
since f{t) 1nust he positive. Then

associated \Vith the probability

that f( .x )

==- e

u n it v

ae

ax,

Sec. 11.8

201

Effusion
1.0

p(x)

N
05
No .

--------

0.5

1.5

1.0

xlf.

x=

xf(x)dx=

so that
(11.36)
The probability p(x) can be interpreted as the ratio N I N0, where N0 is
the initial number of particles in a molecular beam, and N is the number of
particles remaining in the beam after a distance

x. Figure

11.10 is a graph of

this so-called "survival equation." We see that the fraction of molecules with
free paths longer than f is

e -t

or 37 percent, while the fraction with free paths

shorter than the mean is 63 percent. Equation (11.36) shows how

beam of

particles is attenuated due to collisions as it passes through a low density gas.


Use of this curve is independent of the starting point: the probability of a colli
sion in a length t is independent of where the last coJlision occurred.

11.8 EFFUSION

If a container of gas, surrounded by a vacuum, has a small hole in its wail, the
contained gas will leak through the hole into the surrounding space. If the hole
is sufficiently small, the equilibrium state of the gas in the container will not be
appreciably affected. In that case, the number of molecules escaping through
the hole will be the same as the number that would impinge upon the area of the
hole if the latter were not present. The emergence of molecules through the
hole is a process known as

ion.

How small must the hole be? Molecules move unimpeded through a dis
tance approximately equal to the mean free path f. Thus the density of mole
cules at a given point is maintained by the flow of molecules from the surface

202

Chap. II

The Kinetic Theory of Gases

2
of an imagined sphere of radius { and a rea 47Tf . If the flow from the area A of

a circular hole is to be cut off \Vithout making much of a difference to the mol'-

ecular density at the center of the sphere, then A << 47Tf2. Let the diameter of

2
the hole be D. Then rrD2/4 << 41Tf or /.)2 << 16f2.
2
somewhat arbitrarily as D2 < O.l6f or D ( 0.4f. At
atmospheric pressure, e

We may interpret this


room temperature and

10-7m. Thus the opening must be extre me ly small

if the sta t e of the gas in the container is to remain virtually undisturbed. Since
the mean free path is inversely proportional to the pressure, it follows that
there is an upper limit to the pressure for which effusion occurs. Effusion is
essentially a low pressure phenomenon.
In Section 11.2 we found that the number of molecules with speeds in
the range

to

v +

dr that strike a surface, per unit area and per unit time, is

( 11.37)
Here tz,.tl-v is the nun1ber of molecules per unit volume with speeds lJetween
and

v +

dv. Assu1ning that the molecules have a Maxwell--Boltzn1an speed

distribution, we know that

( 1 l.38)
Substituting this in Equation ( 11.37 )_we have
mv'-!2k
.

T d V.

(11.39)

This is the number of molecules that escape through the hole in the container
wall that have speeds from

t'

to

v +

(/I', per unit area and per unit time.

\Ve wish to determine the mean speed 1)c of escaping molecules.


Following the usual procedure, we multiply

ble values of

l\

1'

by <P,.{/v_ integrate over all possi

and divide by the i n t e g ral of 1.dV. The last step is necessary to

ensure that the weighting function used in calculating the mean value is a true
probability density function. Thus
.::t

vp

--

V4e -nzvj2kTdV

vet> v dv

)':

0
X

cl>vdv

3eV

mn-;2kT

dV

"'lJ

Values of the integrals arc given in Appendix D. The result is

1J

(J

==

3 21rkT

--

111

l,i2

( 11.40)

Sec. 11.9

Transport Process\!s

203

Comparing this with the mean speed v of molecules within the container
(Equation

(11.29)), we note that ve is slightly higher than v:


-

Ve

31i=

1.18v.

(11.41)

This is because the faster molecules have a greater probability of ''finding" the

hol e in a given tim e

Effusion has found an important application in experimental physics.


Molecules allowed to escape from a small hole in a container into surround
ings maintained at low pressure can be collimated to form a well-defined mol
ecular beam. (Standard methods involving electric or magnetic fields that are

used to guide charged particles are not ap pli cable to neutral molecules. ) The

effusion technique makes it possible to study individual molecules whose


interactions with other molecules are negligibly small. Molecular beams have
also been used to verify the Maxwell-Boltzman velocity distribution
.

Another use of effusion is in the separation of isotopes. The principle is

based on the fact that the rate of effusion of a molecule is dependent on its mass.

1j_V/4;
(8kTj1riii)112

The rate per unit area is the molecular flux cf>


PV

NkT,

so

n =

N/V

P/kT

and v

for an ide al gas,


Combi ning these

exp ressions we see that


,

<I>

(11.42)

--;::==-:::=::: .

21rmkT

Thus light molecules escape m ore ra pidly than heavier ones.

A gas mixture of two is otopes is permitted to effuse through tiny holes in

th e wall of a container. After a period of time the relati ve concetttration 9f the


heavier isoto pe will exceed that of the lighter iotope in the container. In the
gas that has effused , the reverse will be the case. It is interesting to note tltat
the effusion method was one of the methods used to separate the isotop es
218
in the making of the aton1ic bomb.
U235 and U
.

I I. 9 TRANSPORT PROCESSES
Transport processes involve the re action of

gas to small departures from

thermal equilibri um due to gradients in the medium. From a knowledge of the


average speed and the mean free path of the molecules. it is rossible to gi"e an
elementary derivation of three common p roperties of a dilute gas: its vi scosity,

thermal conductivity, and coeffi ci en t of dif fusion

In a transp<Jrt process, each molecul e conveys momentum , energy, and


mass (the molecttle itself) from one collision to the next. At each collision it
acquires the average properties of the molecules in the neighborhood of the

Chap. t I

204

The Kinetic Theory of Gases


-

--d

'-

. .

\'

z=O

--

--

z =-d

.
.

._.

.
.

0
.

..

.
.

Figure 11.11 Flow of a viscous fluid bet\\'een a


stationary lower plate.

movin g

upper plate and a

collision. Hence as it travels along the next free path, it carries a sample of the
properties of the gas in the vicinity of its previous collision. If there are gradi
ents present the properties in the region of its next collision may differ from
those carried by the molecule. During the next collision the molecule will give
up this difference to the gas in the new region. In their motion from one colli
sion to the next, the molecules will couple the properties of the gas at one
point with those at another. The ultimate resttlt is an equalization of the prop
erties throughout the gas.TI1is is the e s s en ce of transport processes.
Consider a fluid such as a gas between two large parallel plates. The
upper plate is moved horizontally while the lower plate remains at rest
(Figure 11.11 ). Since real fluids are viscous the fluid is pulled parallel with
the moving plate. The magnitude of the 11uid velocity

in the l1orizontal

ll

direction changes from the speed of the upper plate to zero at the stationary

plate. In other words, a velocity gradient d1tldz exists in the z direction. per
pendicular to the plates.
Owing to the gradient .. the molecules that are continually moving from
positive values of z are passing through the z == 0 plane.. transferring their
higher momenta in the x-direction t(> the layer below.. thereby increasing the
average momentum <)f the layer. Similarly the molecules moving from nega
tive values of

through the plane are carrying \\'ith them the lower momenta

corresponding to the region where they last underwent a collision. The net
momentum transfer per unit time through unit area gives the tangential stress
P-:.x (the viscous force per unit area) .. \vhich a layer below the t)t plane exerts on
..

the fluid above it due to the velocity gradient.


For small gradients we expect a relatic)n of the form

(11.43)

where the proportionality fact or TJ is the C<lefficient of viscosity of the fluid.


Viscosity has tradi tionally been measured in units of dyne

s em-, called a

Transport Processes

Sec.ll.9

205
z

cos

l
I
f
I
I
I
I

I
I
I
I

dy
X

dx
Figure 11.12 Molecule crossing the z
0 plane after its
last collision at a distance e cos8 above the plane.
=

"'poise," after Poiseuille*, who studied viscous flow. The Sl unit is Pa


is equal to

10 poise.

which

The coefficient of viscosity for a dilute gas can be calculated using


kinetic theory and Newton's second law. We must first determine the average
distance z above the xy plane at which a molecule makes its last cc,Jlision
before crossing the plane. From Figure 11.12, z= f cos8. Then z can be found
by multiplying this quantity by the incremental flux d4>8 in the 8 direction,
integrating over 8, and dividing by the total flux <1>. Using Equations
and

(11.7)

(11.8), we obtain

z=

--=
-

(11.44)

---

Thus, on the average, a molecule crossing, the xy plane makes its last collision
before crossing at a distance equal to two-thirds of a mean free path.
If

u(O)

is the forward velocit}' of the fluid at z =

0,

then at a distance

2tj3 above the plane, the fluid velocity is

2
-e
3

du

(11.45)

Jean Louis Marie Poiseuille. a French anatomist.. was interested in the physics of the circu

lation of the blood. His work on the flow of liquids through capillaries was published in

1842.

Chap. II

206

The Kinetic Theory of Gases

Here we assume that the gradient does not vary appreciably over the distance
of a mean free path. The momentum of a molecule with this vei<)City in the
.x-direction is mtt(2f/3) and the total momentum transferred across the plane
per unit time and per unit area by a11 of the molecules crossing from above is

1
r .
- n tn v u (0 )
4=

2 lill
.
-;; f
, d

Similarly, the total momentum transporte.d per unit time and per unit area
across the plane by the molecules crossing from below is
1

n1n v

4=

u ( 0)

2 du
.
-f
3 (Jz

The net rate of momentum transport per unit area is the difference between
these two quantities, or
l

till
-- 111n v f
d
3
::
-

-.

e 4 ua l to

From Newton ,s second law.. this is

the viscous force per unit area

given by Equation ( 11.43 ) It follo\vs that


.

TJ

J
-.
-- 111111'{.
1
=

( 11.46)

..

R ecal l i ng that ('

b e w ri t te n

==

H/rrna)
-I,
--

we

see

t h at the co ef ficie nt of viscositv can

.-

T}

1
:..::.

/1r

6'v

mi'

(11.47)

The mean kinetic energy of


Thus vx ( T/ n1) 1/2 and

ffi()(ecule is proportional to the te 1nper atur e.

)
1
(con st T
fT

(11.48)

This result may seem surprising. Intuitively, \VC might expect the viscosity to

decrea se with i ncreas ing temperature and to increase with i n creas i n g collision
cross section. But higher temperatures mean higher molecular speeds and a
small cross section means a large mean free path. B oth factors contribute to a
more effective transfer of momentum from one fluid laver to the next and hence
to a higher viscosity. This behavior differs markedly from that of the viscosity of
..

.,

Sec. 11.9

Transport Processes

207

liquids, which decreases rapidly with increasing temperature. That is because in


a liquid the molecules are closer together and momentum transport across a

surface involves interaction forces between molecules on the two sides of the
surface as well as molecular motion across the surface.
An especially useful expression for calculating the viscosity is obtained by
solving for n in Equation ( 11.9) and substituting the result in Equation (11.46).
-

Then, assuming that v

::.::: ( v)2,we

get

')")'=>
.,

For oxygen under standard conditions

11

(1.01 X lo-'Pa)(5.7 X

Pf

(P

( 11.49)
=

1.01

8
10- m)

(420ms-1)

105 Pa,

273K),

14 X

10- Pa s.

This compares with the measured value of close to 19 x 10-6 Pa s.TI1e discrep
ancy is due to the approxin1ate nature of kinetic theory, which treats the mole

cules of a gas as hard spheres. Also, the factor 1/3 in Equation

(11.46) is not to be

taken too seriously, because it depends on how various quantities are averaged.

The calculation of the them1al c9nductivity proceeds in a similar way. The


only difference is that it is the thermal energy that is transported rather than the
momentum. The upper and lower plates in Figure 11.11 are both assumed to be

stationary but are held at different temperatures so that there is

temperature

gradient rather than a velocity gradient in the gas. The themtal conductivity
results from the net flux of molecular kinetic energy across a surface. The aver
age kinetic energy of a molecule is cvT / NA, where cv is the specific heat capacity

and NA is Avogadro's number. The thert11al conductivity A is found to be


A=

Cv

-3ve
_

NA

1T

--

CvV

(11.50)

--

lVAa.

It is interesting to note that the ratio of the thertnal conductivity to the viscosity is
A

cv
-

so that
AM= l
1JCv

'

where M is the molecular weight of the gas. Measured values are closer to 2,

but kinetic theory gives the right order of magnitude.

The Kinetic Theory of Gases

Chap. II

208

Finally.. the transport of mass across a surface as a result of random mol


ecular motion gives rise to diffusion if a concentration gradient exists. The
treatment of diffusion is complicated when there is more than one type of molecule present, but its main features can be understood by considering the
motion of molecules in a single constituent gas.. a phenomenon known as self

diffusion. The coefficient of self-diffusion

l> can be easily calculated by mean

free path arguments like those used in the discussion of ''iscosity and heat con
ductivity. The result is
v

Since!!=

2 n(T

(11.51)

3Pjmv2 and both Vand (v)l:2

1
are proportional to T

..

2,

( const) T3 2

at a fixed pressure an d

( CfJnst)
=

at a fixed temperature. The ratio of the roefficients of viscosity and self


diffus ion is

--

.=:

11

t1n1

--

where p is the mass density of the gas. Thus


Dp
-

17

Experimentally. DpfTI l ie s

in

I.

the range 1.3 to 1.5. The extent of the agreement

between theory and experiment must be COilsidered satisfactorv in view of the

approximations made in our physical model.

IIIIII

PROBLEMS:

'
11.1 The distribution of particle speeds of a certain hypotheti cal gas is given by

tv' ( l' )d v
where A and

v.)

are constants.

A ne-,.

r..

d v..

Problems

209

(a)

Determine A so that f ( v)

(b)

Find v and

i.e.

; .f ( v )dv

V,m.

(c) Differentiate

N ( v) IN is a true probability density function;

1. S ketch f ( v) versus v.

in tem1s of v0

f ( v)

with respect to

the most probable speed

and set the result equal to zero to find

Vm.

(d) The standard deviation of the speeds from the mean is defined as

- _), 1/"2.
(

u=

v-v -

where the bar denotes the mean value. Show that


u

[ v 2 - ( v) 2] J :'2

in gener al . What is ufor this problem?

11-2 At standard temperature and pressure the mean speed of hydrogen molecules is
1. 70 X 1 OJ ms -l. What is the particle flux?
11-3 What is the number density of molecules at a temperature of 77 K in an ultra
high vacuum at 10-10 torr?
11-4 Compute the

rms speed of helium atoms at 2 K; of

and of mercury atoms at 100C.

nitrogen molecules at 27C;

11-S Compute the mean energy in electron volts and the rms speed in ms -t of an elec
tron at 1000 K. At 10,000 K, what fraction of the speed of light is the rrns speed?
11-6 At what values of the speed does the iax wel l speed distribution have half its
maximum value? Give your answers as a constant time

11-7

(a)

Show that the Maxwell speed distr ibuti on can be written as

N ( X) dX
where x

(b)

(kT,Im) 112

4N

1T

x-e-',. dX,
.,

vfvm.

Show that the number of molecules with speeds Jess than some specified

speed v0 is given by

)_.X.,
where x0

N er/( Xo )

v0fv,. Here erf(x0)


erf(x)

. xoe

1T

.. 1
a

'

is the error function, defined by

--;:=
1T

()

u du.

(c) Find the fractional number of molecules -with speeds less than
118 Compute

vm, v, and v,ms

vm.

for an oxygen molecule at 300 K. What are the corre

sponding values at lOJ)OO K?

11-9

(a)
(b)

Show that the mean speed of a nitrogen molecule at 300 K is 476 ms- 1

What is the ratio of the probability of finding a nitrogen molecule w hose

speed is 476 ms-

at 300 K to the probability of fi nding a nitrogen molecule

with the same speed at 600 K?

The Kinetic Theory of Gases

Chap. II

110
11-10 Find the

rms

free path in terms of the mean free path. What is the most probable

free path ? Use t

1e

.t'tas

the probability density function.

11-11 For carbon dioxide (molecular weight 44! molecular diameter 4.6 X 10 10m) at
1 atm and 300 K, find the mean free path f using Equation (11.34 ). What is the
ratio of f to the molecular diameter? What is the collision frequency?

11-12 The number density of the gas. mainly hydrogen that fiBs interstellar space is
one molecule per cubic centimeter ( W' m l). If the moleci.Ilar diameter is 10

10

m .. what is the mean free path in interstellar space? What is the collision fre
quency of a hydrogen mo1ecule in collisions per century if the tetnperature of
interstellar space is

10 K?

11-13 A beam of oxygen molecules start tog e th er The pressure is 1.8 x 1 OJ to rr and
.

the temperature is

3.6

300 K. The diameter of an oxygen molecule is approximately

10-10 m. How 1ong will half the molecules remain unscattercd? (Assume

that all the particles have a speed equal to the mean speed.)

11-14 A thin-walled vessel containing 1 liter of carbon dioxide is kept at a tempera


ture of 0C. The gas slowly leaks out through a circular hole of diameter 100 p,m.
The outside pressure is low enough that leakage back into the vessel is negligi
ble. (The diameter Of a CQ molecule l" abOUt 4.6 X } 0 HIm and ifs moleCUlar
weight is 44 kg kilomole- '.)

(a) Estimate the upper limit of the pressure in the container for effusion to
occur through the hole.

(b) Estimate the time, starting

at

this pressure for the pressure to drop to one..

half its initial value.

11-15 A vessel is divided into two parts of equal volume by means of a plane partition,
in the middle of which is a very small hole. Initially both parts of the vessel con
tain ideal gas at a ternperature of 300 K and a lo\\ pressure P. The temperature
of one-half of the vessel is then raised to 600 K while the temperature of the
other half remains at 300 K. Determin the pressure difference in terms of P
between the two parts of the vessel \vhen steadv conditions are achieved .

11-16 A vessel has porous walls containing many tiny holes. Gas molecules can pass
through these holes by effusion and then be pumped off to son1e collecting
chamber. The vessel is filled with a dilute gas consisting of two types of mole
cules that have different masses

nz1

and

nl

by virtue of the fact that they contain

two different isotopes of the same atom. The concentrations of these molecules
are c1 and c2 respectively. (The concentration

c,

is the ratio of the number of

molecules of type ito the total number of molecules.) The concentrations can be
kept constant in the vessel by providing a steady flow of fresh gas through it so
as to replenish any gas that has effused. If

'
c1

and

'
c:

denote the concentrations

of the l\\'O types of molecules in the collecting chamber. \\'hat is the ratio

c/ Ic1 '?

11-17 Diffusion can be regarded as a random walk prohlem in \Vhich the successive
displacements of a gas molecule are statistically independent. In this model. the
distance IJ traveled by a molecule after N displacements is related to the mean
free path ( by the expression L2
tance L is given by t

N f. and the time

required to move a dis

N fj t;. Using these relations estimate how long it \vould

Problems

211

take a molecule in a room of macroscopicnlly "still'' air with unifortn tempera


ture and pressure to move

distance of 5 meters.

11-18 The experimental value of the viscosity of argon gas is found to be


22.0 x 10-6 Pa s at l5C and atmospheric pressure. The atomic weight of

argon is 39.94. Estimate the diameter of an argon atom.

11-19 The radius of an air molecule is approximately 1.8 x 10-10m, its mass is about
4.8 x 10-26 kg, and it has 5 degrees of freedom. The molecular weight of air is 29
and its density is 1.29 kg m- 3 under standard conditions.
(a) Estimate the values of the coefficient of viscosity TJ, the therrnal conductiv
ity A, and the diffusion coefficient D under these conditions.

(b) Check that for air,


and

Dp
=

11

1.

Also, check that these quantities are indeed dimensionless.

12.1 Introduction

215

12.2 Coin-Tossing Experiment

216

12.3 Assembly of Distinguishable Particles

221

12.4 Thermodynamic Probability and Entrop)#

223

12.5 Quantum States and Energy Levels

225

12.6 Density of Quantum States

229

213

12.1 INTRODUCTION

The union of the energy and entropy concepts led to a science of therttlody
namics that combines great generality with reliability of prediction. Classical
thermodynamics, however, provides little insight into lvhy thermal phenom
ena occur the way they do.
The object of statistical thermodynamics is to present a particle theory
resulting in an interpretation of the equilibrium properties of macroscopic sys
tems. The foundation upon which the theory rests is necessarily quantum
mechanics. Fortunately, a satisfactory theory can be developed using only the
quantum mechanical concepts of quantum states, energy levels, and intermole
cular forces.
The central idea is the probability density function applied to a large col
lection of identical particles. A thermodynamic system is regarded as an
assembly of submicroscopic entities in an enormous nutnber of ever-changing
quantum states. The basic postulate of statistical thermodynamics is that all
possible microstates of an isolated assembly are equally probable.
A few new concepts are needed. We shall use the term assembly to

denote a number N of identical entities, such as molecules, atoms, electrons,


photons, oscillators, etc. The word system is used synonymously.
The macrostate of a system, or configuration, is specified by the number
of particles in each of the energy levels of the system. Thus

is the number of

particles that occupy the jth energy level. If there are n energy levels, then
n

....
=1

N.

The macrostate of statistical thermodynamics is another word for the therttlo


dynamic state of the classical theory, specified by a pair of state variables.
A microstate is specified by the number of particles in each energy state.

In general, there will be more than one energy state (i.e., quantum state) for
each energy level, a situation called degenerac}' A microstate is the most spe
cific description one can get. In general, there will be many, many different
215

Statistical Thermodynamics

Chap. 12

216

microstates corresponding to a given macrostate. We \\ill be interested in the


number of microstates but not in their detailed specification.
The number of microstates leading to a given macrostate is called the
thermodynamic probabili(v. It is the number of ways in which. a given configu

ration can be achieved. This is an Hunn<>rmalized"" probability. an integer


between zero and infinity" rather than a number between zero and one. For the
kth macrostate, the thermodynamic probability is taken to be

1L'k

A true prob

ability Pk could be obtained by dividing wk by the total number of microstates

.n available to the system. For much of what we shall do, the thermodynamic
probability will be adequate.

12.2 COIN-TOSSING EXPERIMENT

By way of introducing the centra) features of statistical thermodynamics" we


shall apply some elementary concepts of probability theory to a coin-tossing
experiment.
We assume that we have N coins that we toss on the floor and then exam
ine to detertnine the number of heads N1 and the number of tails N2
We consider the case of N

N - N1

4. Using the terrns we last introduced" we can pre

pare a table listing the possible outcomes of the experiment (Table 12.1 )
Each macrostate is defined hy the number of heads and the number of
.

tails. There are five. For each macrostate there are one or more microstates. A
microstate is specified by the state. heads

\)T

tails. of each coin: it is the most

detailed description possible. We are interested in the lllllnber of microstates


for each macrostate. the thermodynamic J'robability. The true probability is
the number divided by the total number of microstates. Thus

( 12.1)
where
'

k=l

1vk

==

16.

( 12.2)

We might also wish to calculate the average occupation numbers


this case, the average number of heads and tails. Let j

==

in

I or 2. where N1 is the

number of heads and N2 the number <>f tails. Let Njk be the occupati<>n number
for the kth macrostate. Then the average occupation number is
.

N.J

Njkwk
-

Ltvk
k

Nj*wk
-

NikPk

(12.3)

Coin-Tossing Experiment

Sec. 12.2

TABLE 12.1

217

Possible outcomes of a coin-tossing experiment using four coins. We introduce the

language of statistical therm

mics.

TherJnoMacrostate

Macrostate

Label

Specification

Microstate

dynamic

True

Probability

Probability

N.

Coin 1

Coin 2

Coin 3

Coin 4

1/16

4/16

6/16

T
H

T
T
H

H
T
H

.;.

T
4

4/16

1/16

The average number of heads is therefore

16
Similarly, N2

2, as we would expect. Then N1 + N2 4


Figure 12.1 is a plot of the thertnodynamic probability
=

ber of hellds N1 The curve is symmetric about N1

N.
w

versus the num

2, the most probable

configuration.
Suppose now that we want to perform the coin-tossing experiment with
a larger number of coins. We need a way of computing the thermodynamic
I'

probability without tabulating the actual state of each coin. We assume that we
haveN distinguishable coins and ask: h<>w many ways are there to select from

the N candidates N1 heads and N - N1 tails? The answer is given by the bino
mial coefficient
N

N!
=

-----

Nt!(N- N1)!

(12.4)

Chap. 12

218

Figure 12.1

Thermodynamic
probability versus the number
of heads for a coin-tossing

Statistical Thermodynamics

experiment with four coins.

This result can easily be seen by noting that there are N ways to pick the first
coin, leaving ( N

1) for the second, ( N

2) for the third, etc. Thus the total

number of ways of picking N1 heads is


N!

However, this counts the N1! permutations of the N1 coins separately. The
ordering of the coins is not important, s<) we divide by this factor, giving
Equation (12.4).
If we increase the number of coins from four to eight .. we obtain the plot

12.2. We see that the smallest value of w is again unity, bt its maxi
mum value has grown to 70 the peak has become considerably sharper.
of Figure

The maximum value of the thermodynamic probability is of paramount


interest to us. Its calculation is straightforward for N
peak occurs at N1

8. We know that the

N /2. Thus, Equation ( 1 2.4) gives

'IVmax

8!
=

4!4!

==

70.

Suppose now that we perform a more ambitious experiment, with


N

1000. In this case,


100()!
'LOmax

==

500! 50_0_! '

(12.5)

and the computation becomes much more challenging. Fortunately, for such
large numbers we can use Stirling's remarkable approximation (Appendix B)
and write
ln n! ::::: n In n - n.

Sec. I 2.2

Coin-Tossing Experiment

219

50 40

Figure U.2

Therm odynamic

t--

10 1--

probability versus the number


of heads for a coin-tossing
experiment with eight coins.

N1

Taking the natural logarithm of both sides of Equation {12.5), we then obtain
In

Wmax

ln( l OOO ! ) - 2ln(500!)

693.

But
log10

x =

(log10 e)ln

x =

0.4343 In

x,

so
log 1 o

Wmax

(0.4343)(693)

300.

Thus
Wmax

10300

(The error in using Stirling s approximation here is extremely small: even for n as
'

small

as

50, the error is less than 2 percent, and decreases rapidly

For N

1000 we find that

Wmax

as n

gets larger )
.

is an astronomically large 011mber. We

conclude, then, from our coin-tossing experiments that the peak in our curve
of thermodynamic probability versus the number of heads always occ u rs at

N/2 but grows rapidly and becomes much sharper as N increases. Also, t h e tails
of the curve always terminate at

w =

1. The result is what we would expect:

for ..'honest" coins, we get ever closer to an exact 50-50 split between heads
and tails as the number of coins gets larger. In other words, the most probable
con fig uration

is that of total randomness.

Conceptually, this behavior for N

1000 is shown in Figure 12.3. The

most nearly random configuration (macrostate) N1

always occurs.
.

N2 is the 011e that almost

Statistical Thermodynamics

Chap. 12

220

totally random region

10300

---

----

----

----

ordered region

ordered region
I

Figure 12.3 Thermodynamic


probability versus the number

1000

of heads for a coin-tossing


experiment wit 1000 coins.

The ''ordered regions' almost never occur:


pared with

Wmax

is extremely small com

We are led to a very important conclusion

namely, that the

total number of microstates is very nearly equal to the maximum number:

!l

The figure a1so suggests

( 12.6)

method for finding

the curve is zero. Thus we need to find

'lVm",.

At the peak the slope of

u( /\'1 ), take the derivative of the func

tion with respect to N1 and set the result equal to zero. This wi11 give 1 at

10

u'm ax

and hence N2 The outc<Jmes will then be known for the most prob

able state.
For the thermodynamic probJem .. the outcomes' arc the c,ccupation
numbers for each of n

energy levels. Our task is to find 10( N1 N2


..

..

Ni ... N,)

..

possibly subject to constraints.. and set its derivative equal to zero. In this way
we wil1 obtain the particular set { ',.} of occupation numbers corresponding to
the most probable ntacrostate. The most probable macrostate is the equilib

rittm state of the assembly. This is the fundamental problem of statistical thernlo
dynamics

to determine the equilibrium state of the syste m

Before proceeding we need to extend our counting formula'" of


Equation

(12.4) from two "levels"' (heads and tails) to

11

levels. We sk: in how

many ways can N distinguishable objects be arranged if they are divided into
n

groups with N1 objects in the first group N2 in the second. etc.? The answer is

10 =

N!

N!
'

NI'N'
2 Nn

II

N,!
i =I

(12.7)

Sec. 12.3

111

Assembly of Distinguishable Particles

represents the extended product, analogous to the sum...

Here the symbol


.

mation sign
Equation

(12.7)

can be easily understoocl. Suppose we want to place

distinguishable objects in

three boxes, with N1 in tlte first box, N2 in the sec


ond, and N3 in the third. The number of ways of selecting N1 (out of N) for
box 1 is

N!
=

N1
We are left with

-----

N1!(N- N1)!.

N - N1 objects. We can choose N2 of them for box 2 in


(N- N1)!
=

---

---

==

ways. The number of ways of putting the rerr1airting


Thus the number of ways

w of

._

N3 objects in box 3 is one.

selecting objects to occupy the three boxes is

the product of the three factors:

w=

N!
---

N, ! (N - N1 ) !

(N

N1 ) !

N2!'A)!

Xl=

N!
N1 ! N2! N3! .

---

Equation (12.7) follows by induction.


One additional important point: the coin-tossing model assumes that
the coins are

distinguishable (by a date, perhaps, or a mint mark). The ther

modynamic analogy is with particles in a lattice, whose location distin


guishes them. This would be the case for a crystalline solid, as an example.
But for other cases, such as molecules in a gas, the particles are identical
and indistinguishable and the number of microstates available to the
assembly will be correspondingly fewer. We need to develop statistics for
both situations.

12.3 ASSEMBLY OF DISTINGUISHABLE PARTICLES

Th.e constituents of the system under study (a gas, liquid, or solid, for exam
ple) are considered to be a fixed number N of distinguishable particles,
occupying a fixed volume V. Vle limit ourselves to isolated systems that do
not exchange energy itt any form with the surroundings. This implies that
the internal energy U is also fixed. The macrostate in which the system finds
itself is defined by

(N, V, U), ira correspondence with a classical thermody


namic state defined by (n, V, T), say, where n is the mass in kilomoles. Note
that the temperature Tis simply a measure of the internal energy for a sys
tem in thermal equilibrium.

222

Chap. 12

Statistical Therntodynamics

We assume that the particles are weakly ittteracting. By this we mean


that they interact sufficiently so that the system is in thermal equilibrium, but
do not experience strong coupling through \!lectromagnetic or nuclear forces.
We seek the distribution among energy levels

{Ni}

for an equilibrium

state of the system. That is.. we want to determine the number of particles

Ni

with energy ei for all n energy levels of the system, subject to the restrictive
conditions
ll

"""-

j =l

Nj

}l

(conservation of particles).

( 12.8)

(conservation of energy)

(12.9)

ll

1\/iei

j=}

where Nand U are consta n ts


As with our coin-tossing experiments, we can clarify these ideas with a
.

simple example. Consider three part i c l es labeled A. B. and C distributed


,

among four energy l e ve l s 0 e.. 2e and 3e, such that the total energy is 3e. Thus
,

..

..

.l

'

LN,

LNje;

and

i::. (I

3e.

f =U

The possible microstates and macrostates are listed in Table 12.2. The
occupation numbers are: N0 particles with energy 0, N1 with energy e_ N with

TABLE 12.2

Microstates and macrostates for N

3. U

3,.; with

t:::1

j. j

0, I, 2, 3.

Thermo
Macrostate

Macrostate

Microstate

dynamic

True

Label

Specification

Speci fie at ion

Probability

Probabilitv

0
0
3e

0
3c
0

()

()

_F.

F.

()

2f

2e

k
t

()

2
3

..,

.)

0.

0
0

o'

0.3

2
0

0.6

0.1

Thermodynamic Probability and Entropy

Sec. 12.4

123

TABLE 11.3 Distribution of


three particles among four energy
levels with total energy U 3e.
=

3e
a
e
0
k

1
1
1

3
0

1
3

2
6

3
1

energy 2e, and N3 with energy 3s. The total number of microstates is 10 and the
number of possible macrostates is 3. For the most probable macrostate,
the number of available microstates is 6. Another way of displaying the results
is given in Table 12.3, where the occupation numbers are listed for each of the
energy levels.
sor
"di
st
mo
the
t
tha
ent
evid
is
it
ll,
sma
are
e
her
s
ber
num
the
ile
Wh
dered" macrostate is the state of highest probability. For the very large nJm
ber of particles in a physically meaningful assembly, this state will be sltarply
defined and will be the observed equilibrium state of the system.
In classical thermodynamics we have seen that as a system proceeds
toward a state of equilibrium the entropy increases, and at equilibrium attains
its maximum value. Similarly, our statistical model suggests that systems tend
to change spontaneously from states with low thertnodynamic probability to
states with high thermodynamic probability (large number of microstates).
Thus the second law of thermodynamics is a consequence of the theory of
probability: the world changes the way it does because it seeks a state of
higher probability!

12.4 THERMODYNAMIC PROBABILITY AND ENTROPY

It was Boltzmann who made the connection between the classical concept of
entropy and the thermodynamic probability. His argument is as follows.
Assume that the entropy is some function of w, that is,
S

f(w).

(12.10)

HereS and ware properties of the state of the system (state variables). To be
physically useful, f ( w) must be a single-valued, monotonically. increasing
function.
Consider two subsystems, A and B (Figure 12.4). The entropy is an
extensive property; like the volume, it is doubled when the mass or number of

Chap. 12

224

system

Statistical Thermodynamics

system B

Figure 12.4 Two systems


with entropies SA and S8.

particles is doubled. Thus the combined entropy of the two subsystems is sim
ply the sum of the entropies of each subsystem:

or
( 12.11)
However, one subsystem configuration can be combined with the other to give
the configuration of the total system. That is.
(12.12)
This follows from the fact that independent probabilities are multiplicative. To
return to our coin-tossing experinlent._ suppose that the two subsystems each
consist of two distinguishable coins. The possible configurations of the subsys
tems are listed in Table 12.4. The. probability of two heads in subsystem A is
clearly 1/4, as it is in subsystem B. From Table 12.1, the probability of obtaining
all heads when fottr coins are tossed is 1/16, which is equal to 1/4

><

1/4. Thus

Equation (12.12) holds. and therefore


(12.13)
TABLE 12.4 Possible configura
tions of tossed coins for subsystems
A and B, each consisting of two coins
A

HH

HH
HT
TH
TT

HT
TH
TT

Sec. JlS

Quantum States and Energy Levels

225

..

:..

:..
.

.
.:..

- : ,

.:.

-: :-

.;

,. : o::': -:;: ...

'/-::::::. ;-:
.

..

. .
.
..

..

. t

..
.

-:: ::fJ: :ft


'

..

.::

.
.

;>

..

U.S

Ludwig
Boltzmann's tombstone,
showing his famous equation.
(lnstitut fur Theoretische
Physik der Universitat
Wien/Central Library for
Physics in Vienna.)

.
..

.
- .: ..r
.

. .

Combining Equations (12.11) and (12.13), we obtain

f( WA )

f( Wa )

f( WAWB )

(12.14)

The only function for which this statement is true is the logarithm. Therefore

S =kIn

w,

(12.15)

1-

where k is a constant with the units of entropy. It is, in fact, Boltzmann's


c o n stant :

1.38 X 10-23 J K-1

Boltzmann's tombstone in Vienna bears this fundamental relation as its

inscription (Figure 12.5).

12.5

QUANTUM STATES AND ENERGY LEVELS

We have tacitly assumed that the energy levels of our assembly of particles fotin
a discrete set rather than a continuum. This is a consequence of the particles'
confinement in a container of finite volume. Only in the case of completely

_#

226

Chap. 12

l
I

_j

Statistical Thermodynamics

!...._
... ---JI

L.____.l

83 = 5

Figure 12.6

Energy levels and quantum states. 'The degeneracy g1 is the


number of quantum states whose energy level is e1.

Hfree" particles can a continuous spectrum t.>f energy exist. To understand how
the discreteness arises, we must resort to quantum mechanics.
In quantum theory to each energy level there corresponds one or more
quantum states described by a wave function '}1. For so-calJed stationary

states, ')1 will be a function d e pendent on the position coordinates and the
time.* When there are several quantum states that have the same energy the
states are said to be degenerate. The quantum state associated with the lowest
energy level is called the groLlnd state of the system: those that correspond to

high e r energies are called excited states.

The energy leveJs can be thought of as a set of shelves at different

heights while the quantum states correspond to


(Figure

12.6).

set of boxes on each shelf

For each energy level ;the number of quantum states is given

by the degeneracy gi.

C<ln side r

As an instructive example..

mechanics, that of

particle of mass

n1

basic proble m in quantum

in a one-dimensional box with infi

nitely high walls. (The particle is confined within the region 0

c X :5

L.)

Within the box the particle is free s u bje ct e< i to no forces except those associ
ated with the walls of the container (Figure 12.7).
ll1e time-independent part of the wave functi<)n 1/J( .r) is

measure of

the probability of finding the particle at a positi on .r in th e box. The problem of


finding

.P ( .r)

is analogous to the classical problem of transverse waves on a

string with both ends fixed. The "'ave function sa ti s fyin g the boundary condi
tions

t/1 ( 0)

0 and

l/1 ( LJ)

0 is

t/l(.r)

A in k.r, 0

<

.r

LJ.

( 12.16)

with

= II

7i

l.

11 =

J 2 3. .

*A It hough 'V is timcdependent. the obscrvahle quantity '1'.

the complex conjugate of 'I'.

(12.17)

'I'

is constant in time.

Is

Sec. 12.5

Quantum States and Energy Levels

127

00

12.7 A one

x=O

dimensional potential well.

x=L

The boundary conditions simply state that the wave function must vanish out
side the box. The probability of finding the particle there is zero.
Now, the de Broglie relation of wave m!chanics is
p

==

1tk,

where p is the m o ment um of the particle, k is the wavenumber, and 1i is

Planck's constant divided by 21T. The particle's kinetic energy is therefore

(12.18)

Substituting Equation (12.17) in Equation (12.18), we obtain

(12.19)

Here the integer

is the quantum number of the one-dimensional box. The

important result is that the energy is proportional to the square of the quan
tum number.
We can easily extend this result to the case of a three-dimen s i onal box
with dimensions Lx, Ly, and Lz, for which the energy becomes

2
7T21i2 nx2
ny2
nl.
+
+
.
2
2
2
2m Lx
Ly
Lz
In this case, any particular quantum state is designated by three qua n tu m num
bers nx, ny, and nr.. If Lx = Ly = L, = L, then
e
J

(n 2+ n 2+ nZ 2)
X

.V

nJ'
.2

(12.20)

ni is the total quantum number for states whose energy level is et The
energy levels depend only on the values of n/ and not on the individual values
of the integers ( n "' n V' n z) .
where

First three states of a three-dimensional infinite potential well.

TABLE 11.5

Energy State

Level

j
j

Statistical Thermodynamics

Chap. I 2

228

Ground state

First excited state

Second excited state

The volume Vof a cubical box equals L3 so/_


, 1T*"1t- -,
V
n
1 2nl

e. =
1

3
3

V2,'-' and hence

..,

3
6
9

(Ll.l)
( 1. 1.2 ): ( 1.2J ): ( 2.1.1 )
( L2.2): (2,1.2); (2.2J)

..

(12.21)

This result applies to a container of any shape whose dimensions are large
compared with the de Broglie wavelength

2rrn/ p. We observe that as the vol

ume decreases, the value of the jth energy level increases.


The lowest energy leveLj
n?

3 and

1, is that for which

nx

n_,.

11

This is the ground state. There is only one set of quantum numbers

(n_,

. ..

that has this energy. Thus the lowest level is nondegenerate and g1
dgeneracy for the first three states is shown in Table

the degeneracy increases fairly rapidly ith

12.5. As
increasing j.

1. Then

n\

..

n::)

1. The

a general rule,

Now each energy level is occupied by a number of particles

!vj.

Returning to our analogy of shelves with boxes on them, the degen eracy gi of
level j is the number of boxes on the jth shelf. The number of particles in any
one box is the number in a particular quantum state. Those particles on any

one shelf are in different states but have the same energy. Our interest is in the
total number of particles in the boxes at the various JeveJs.. that is.. in the occu

pation numbers

numbers { Ni] then defines a macrostate of the

N1. The set of

system. A fictitious example is sho\\'n in Figure 12.8.


It is instructive to estimate the quantum

n umb e r II;

for a on e li te r vol
-

ume of helium gas at room temperature. The mass of a helium atom is


6.65 x I0-27kg,and1liter
2ti2

iT2

2m

1 054

2(6.65

X
x

==

1 0 m.Alson

1 0- 34
27
10- )

==

1.054

x 10-34Js.Thus

At room temperature the mean kinetic energy of a helium gas atom is


.

e1

kT

( 1.38

23
10- ) (293)

4.04

10-11 j

::=::=

2.5

10-2 eV.

Sec. 12.6
e3

Density of Quantum States

'-_....._j

81 l

229

83

==

NJ

Nt

8t

Distribution of e1even particles among three energy levels.

Then
nJ

2.5

;;:::-!
-

X 10

1
2 12
=

to-21

X 10

Thus, for the vast majority of the molecules of a gas at ordinary temperatures,
the quantum numbers nj are large indeed. In other words, most of the mole
cules occupy highly excited states. This means that the energy levels are very
closely spaced and that the discrete spectrum may be treated as an energy
continuum. That is the subject of the next section .

12.6

DENSITY OF QUANTUM STATES

Under the conditions that the quantum numbers are large and the energy levels are very close together, we can regard then's and thee's

as

fotrning a con

tinuous function rather than a discrete set of values. We are interested in

finding the density of states g(e).

Dropping the subscripts in Equation ( 1 2. 2 1) , using 1i

ing for n2, we obtain

n2

n X2 +

nY2

ny, nz
( nx, ny, nz )

The possible values of nx,

+ n l2

8mV213
=

h2

aU positive integers

R2

h/2Tr,

and solv

(12.22)

correspond to points

space. According to Equation (12.22), for a


in a cubic lattice in
given value of e, the valttes of nx, ny, nz that satisfy this equation lie on a
sphere of radius R. A cross-section through the positive octant is shown in
Figure 12.9.
We let g (e)de be the number of quantum states whose energy lies in. the
range

to

de.

This is the number of states whose quantum numbers

( nx, ny, nz ) lie within the infinitesimally thin shell of the octant of a sphere with
radius proportional to the square root of theenergy (Figure 12.10). Evidently,

g(e)de

dn(e)
n(e +de) - n(e) "'"'
de.
de

(12.23)

Statistical Thermodynamics

Chap. 12

130

Figure 12.9

A cross-section

through the positive octant of

radius R showing lattice points


corresponding to the quantum

num bers nx

and ny-

./ .

R/

"--

_,__
_
.,..

________

11.\

Here we have expanded the function n(e +

de) in a Taylor series and

retained only the first two terms. Now n (e) is the number of states cont ained
within the octant of the sphere of r a d ius R; that is,

n(e)

1 4

8 3 7T R
.

1T
=

8m
V h2

3/2
.

3/2

e .

( 1 2 24)
.

so that

g(e)de

Equation

dn(e)
=

de

de=

(12.25)

(12.25) is not quite the result w seek. It takes into account the

translational motion only of a particle of the assembly. But quantum parti

cles may have spin as well. The result for g ( e )dr, is valid for bosons of zero
spin (bosons are particles witl1 integer spin). For fermions of spin one-half
(fermions are particles with half-integer spin) there are llvo sp in states for

each t ranslat i o nal state. Thus we must multiply Equation


factor y5:

(12.25) by a spin

(12.26)
where Ys

1 for spin zero bosons and 'Ys

2 for s pin one-half fermi<lns.

For molecules .. states associated with rotation and vibration mav ex ist
-

in addition to translation and sp i n . We shall consider such cases in later


c h ap t e r s
.

231

Problems

..

'
'

..
.

e +de
.
..

..

Figure 12.10

Shell enclosing
quantum states with energies
between e and e + de.

..

.
..

PROBLEMS
12-1 Suppose that you flip 50 "honest" coins.

(a)

How many microstates are there? Give your answer as a factor of the order
of unity times an integral power of 10.

(b)

Ho\\' many microstates are there corresponding to the most probable


macrostate?

(c) What is the true probability of achieving the most probable macrostate?
Note: Use a calculator that gives you n! or a table of gattuna functions
(f(n + 1)
n!). Stirling's approximation will not give you sufficient accuracy.
=

1.2-2 Do Problem 12-1 for the case of 100 coins.


12-3 Using whatever computing means you have at your disposal, investigate the
error in Stirling's approximation

In n!

n In n

n.

(You might display the percentage error as a functionof n for n

>

10, say. Be

creative.) For what value of n is the error less than 2 percent (approximately)?

12-4 Suppose you flip 1000 coins. We found that the thermodynamic probability of
getting 500 heads and 500 tails is w = lQ-'00 Show that for the case of 600 heads
9
is
smaller
by
a
factor
of
400
tails
w
and
10
ll-5 For N distinguishable coins the therntodynamic probability is
w=

N!
--

N1 ! ( N

where Nl is the number of heads and N

(a)

N1 ) ! '

N1 the number of taiis.

Assume that N is large enough that Stirling's approximation is valid. Show


that lnw is a maximum for N1 = N /2.

(b)

Show that

Wmax

=::

eN1"2.

Chap. 12

232

Statistical Thermodynamics

12-6 Consider a model thermodynamic assem hly in which the allowed (nondegener
ate) states have energies 0. r.. 2r.. 3r.. 1l1e assembly has four distinguishable
(localized) particles and a total energy U

6e.

(a) Tabulate the nine possible distributions of the four particles among the
energy levels ne where
..

(b) Eval u ate wk


(c)

0 1 ......

n =

..

for each of the ma crostates and calculate fl

Calculate the average occupation numbers

,;v,

L N 1{'/0
J

of the four particles in the energy statS.

12 7 Six distinguishable particles are to be distributed among energy levels 0 e, 2e 3e


6e.
..... each with a degeneracy of three. The 1 otal energy is U
(a) Tabulate the 1 1 po ss ible macrostates nf the assembl y.
(b) Calcu late 1v (the number of microstates) for each of the 11 macrostates as
a factor times 3tl.
(c) What is the total number of available microstates fl for the assentbly?
(d) Find the at'erage occupa ti on numher for each energy level.
(e) What is the sun1 of the average occupation numbers (it should be six)?
..

..

12-8 Two distinguishable part i cle s are to be distributed an1ong nondegenerate energy
2.
levels 0 e and 2e such that the tota l energy U
(a) What is th e entropy of the assen1hly?
(b) If a di st i ng u i s h abl e part ic l e \\ith zerc energy is added to the system .. show
that the entropy of the assembly is i nc reased by a factor of 1 .63.
=

..

U-9 In reference to E quati on ( 12.1.+) prove th a t if


.

j(xy)
wher e

f(xl

j(y).

and y are indepe nd e n t variables. then .f' ( x)

C In

"'

==

c ons tan t .

(Hint: Take parti al derivatives of the equation.)


12-10 (a) Tabul ate the values of the quantum numbers n, n , ..
energy levels of a pa r ticl e in a contain er of volume V
(b) What is the d egen eracy g of each level?

,-

for the 12 lowest

L'.

(c) Find the energy of each level in units of 1r2it2j2n1L2


(d) Are the energy levels equally s p aced ?
12-11 Calculate the value of 11; in \vhich an o x ygen atom confined to a cubi cal box 1 em
on each side will h ave the same ene rgy as the lo\\'est energy available to a helium
10
n1 on each side.
atom confined to a cubical box 2 x 10
12-U Consider a gas consisting'of one ki lo m o le of helium atoms at standard tempera
ture and

the de g e neracy g( e)

for the energy level


1 ). What is the approx in1 ate ratio of g (E) to the num

pressure. Calculate

( 3/2) kT

(take y,

ber of atoms N?

..

dSSICd

an

13.1 Boltzmann Statistics

235

13.2 The Method of Lagrange Multipliers

236

13.3 The Boltzmann Distribution

238

13.4 The Fermi-Dirac Distribution

242

13.5 The Bose-Einstein Distribution

244

13.6 Dilute Gases and the Maxwell-Boltzmann


Distribution

246

13.7 The Connection Between Classical and


Statistical Thermodynamics

248

13.8 Comparison of the Distributions

253

13.9 Alternative Statistical Models

254

233

13.1 BOLTZMANN STATISTICS

We wish to detertnine the equilibrium configuration for an assembly of N distin

guishable noninteracting particles subject to the constraints (Equations


and

(12.9))

(12.8)

,_,Ni

j=l

N,

(13.1)

(13.2)
Here

is the number of particles with single-particle energy sj; N and

U are

assumed to be fixed quantities. Our goal is to find the occupation number of


each energy level when the thertnodynamic probability is a maximum.
We need to allow for degeneracy. Consider the first energy level, j

1.

The number of ways of selecting N1 particles fro1n a total of N to be placed in


the first level is

N
N1

N!
-----

N1! ( N

N1 ) ! .

We ask: in how many ways can these N1 particles be arranged in the first level?
There are g1 quantum states in the first level, so for each particle there are g1
choices. That is, there are (g1)N possibilities in all. (Suppose that g1 = 2 and
N1

3. For each particle there are two choices, so there are

23

choices total

(Figure 13.1).)
Thus the number of ways to put N1 particles into a level containing g1
distinct options is

N
N!g1 ,
--

Nl ! ( N

N1 ) ! .

235
'1

Chap.

236

Stitte b

State a
123

123
3

12
23
31

Figure 13.1 The eight


arrangements of three
distinguishable particles in two
quantum states, a and b. The
particles are labeled 1, 2 and 3 .

Classical and Quantum Statistics

13

.,
'-

23
31
12

.,

For the second energy level .. the situation is the same, except that there are
only ( N - N1) particles remaining to deal with:

Continuing the process" we obtain

- N'.

g /V 'g N,g N3

II

g '".

'
- '
N.
.
'
N
N)'N2'N3'
j=J
j
1

(13.3)

Having developed the statistics 1v8 for the so-called Boltzmann distribution, our

task is to maximize w, subject to the conditions of Equations ( 13.1) and ( 13.2).

13.2

THE METHOD OF LAGRANGE MULTIPLIERS

The so-called method of Lagrange multipliers is tailor-made for our problem.


Suppose we wish to maximize a function
the constraint 4>( x

...

y)

f ( .x y)
..

of two variables subject to

constant. We must satisfy the condition

df

rif
=

ax

dx

at
av

dv

0.

Sec. 13.2

If

The Method of Lagrange Multipliers

237

dx and dy were independent, we would conclude that iJfjax

iJffiJy

= 0.

However, they are not independent but are related by the constraint equation,
which can be differentiated to give

dt/>

iJc/J
dx +
dy
ax
oy

de/>
=

0.

It follows that

iJffiJx
iJcfJfiJx

iJffiJy
iJt/J/!J y

If we let the common ratio be the constant

iJf
ax

iJ</J
a
ax

= 0,

iJf
ay

a, we then have

iJtjJ
a ay

(13.4)

= 0.

These are exactly the expressions we would get if we attempted to maximize


the function

f +

ac/J without the constraint. The method adds a new unknown

a called the Lagrange multiplier; it is the price we pay for the constraint.

The extension to a function of

variables is straightforward. Let

(13.5)
be subject to the condition

(13.6)
For a maximum, df

0. Since cJ> is a constant, dc/J = 0. The method gives

iJf
OX

iJc/>
a
dX

'

1, 2, 3

= 0, l =

n.

(13.7)

'

(13.6), constitute a set of n + 1 equa


xn
tions that can in principle be solved for the independent variables x1, x2,

These equations, together with Equation

and the Lagrange multiplier


.

If there are

(13.2)),

two

a.

constraint relations, as in our case (Equations

cP

.p

cf>( X1, X2,


t/I(Xt, x2,

(13.1) and

(13.8)

(13.9)

Xn ,
Xn ,

Classical and Quantum Statistics

Chap. I 3

138

we must introduce two arbitrary multipliers a and {3. In this case.. we have a set of
n

+ 2 equations

the two constraint equatio11s and the

at

- + a

dX;

a
f3-:il.ri

..

dX1

0,

1 =

2.. 3 . .

equations of the form

(13.10)

. 11.

Thus the problem is fully specified. The multipliers are often called undeter
mined," but they can be found.

13.3 THE BOLTZMANN DISTRIBUTION

Because In

is a monotonically increasing function of

equivalent to maximizing
Equation

u1.

1v

..

maximizing In

is

The logarithm is much easier to work with. From

(13.3) we have
..

"

In

lV

ln.N!

N.
In
I
...,

II
cJ.
'"''

(13.11)

i= l

Since we are concerned with very large numbers, we


Stirling's approximation and write

safely use

can

In

w =

InN!

(13.12)

+
I

We needn't approximate InN! because it is a constant that vanishes ltpon dif


ferentiation.
In applying the method of Lagrange multipliers.. we take partial deriva
tives with respect to a particular valueNj of the set (N,,N'!. ..... N; ... Nn). Thus

(13.13)

where

(13.14)
.

So

aN.J

(13.15)

Sec. 13.3

The Boltzmann Distribution

239

In working out the derivatives, we note that the only terms of the sums whose
j.
derivatives with respect to are other than zero are those for which i
Therefore, Equation (13.15) reduces to
=

(13.16)
=0

Then

or

(13.17)
Equation (13.17) gives, for every energy level of the system, the number of
particles per quantum state for the equilibrium configuration of the assembly.
The result, obtained under the assumption of distinguishable particles, is
called the Boltzmann distribution function.
The constants a and f3 are related to the physical properties of the
assembly. We consider {3 first. If we multiply Equation (13.16) by and sum
over j, we obtain

or

The two terms on the left-hand side appear in the expression for In
Equation (13.12). Making the substitution, we get
ln

In N! + N -aN- {3U.

SimpJifying, we have
In

w =

C- {3U,

of

240

Chap. 13

Classical and Quantum Statistics

where C is a constant. Using the statistical definition of entropy,


S

k In

1v

S0

(13.18)

k{3U,

where 50 is the constant kC.


From classical theory,
riS

dU +

riV

as
av

dV,

g1vtng

as
au

From Equation

T.

(13.19)

(13.18),
( 13.20)

It follows that

{3

.
kT

(13.21)

The association of the classical expression with the statistical expression for
the entropy suggests that the temperature is a Lagrange multiplier!* Note that
the partial derivative in Equation

(13.19)

is taken with the volume held con

stant. The constancy of V is implied when \Ye write Equation (13 .2 ) , since
e/x v-213 as we saw in Chapter

12.
Substituting Equation (13.21) in Equati<>n ( 13.17), we obtain

of a can easily be found from the first constraint equation


(Equation ( l 3 1 ) ) :

The value
_

*Some texts give

{3

J Ik T. The

Lagrange multiplier. which is arbitrary.

positive sign comes from the choice of sign of the

Sec. 13.3

The Boltzmann Distribution

241

so that

a
e =

N
----

'

and hence

=
1

=
gj

T
k
e;/
Ne

(Boltzmann distribution),

--

(13.22)

where

fi is the probabilit)' of occupation of a single state belonging to the jth

energy level.
The sum in the denominator has a special significance. It is called the
partition function or sum-over-states (Zustandsumme in Ge1tt1an), and is rep..
resented by the symbol Z:
,,

z
j=!

T
k
i
,
e
e
.
i

(13.23)

The partition function depends on the temperature and on the parameters


that detertnine the energy levels and quantum states.
If the energy levels are crowded together very closely, as they are for a

gi may be replaced by g(e)de, the number


of states in the energy range from e to e + de. Correspondingly, can be
rep1aced by N ( e)de, the number of particles in the range e to e + de. We then
gaseous assembly, the degeneracy

obtain the continuous distribution function

f(e)

N(e)
g(e)

T
I<
/
;
Ne-

----

k
)
/
P.
g(e e Tde

(13.24)

The equation is completely analogous to the expression for the discrete


. energy spectrum.
It is worth noting that the occupation numbers are fully determined by
the temperature and the volume (whiclt established the energy levels and the
degeneracies). The set of occupation numbers, in turn, that maximize w specify
the equilibrium macrostate. Thus two state variables define a thermodynamic
state, exactly as in the classical theory.
The maximization of

does not absolutely require the use of Lagrange

multipliers; an alternative method is given in Appendix C.

Chap.

241

Classical and Quantum Statistics

13

.
:: .: : : :=-:
::
: .....::- -.: : :
-: -:.-:. - ::.:.:.-:. -.
. ..-:.:
::
::..
-::
: :..-:
- .:...
.: .:
:...
: ...
:- . :

.:
.:-:
--:;::
:. ::.:;.:: ...
:::::
: . :.: -:- :.:..-.:- ..
....
. .
:
:
..
:-:
-::-:
.:

:::. : .;

:
:.....
:
.
::

:
:..:. <
-:-
::;v

:.:-:.;

:.
:
::1':..: :
.:

:. :-. :-.. :
::-.:
: .-:.:
-.::
- .
:
-.:
:. ..
-: :-: -.
:- .:.
-..:
:
:.
:- :
:...
: ::.
::::::.:.:.:::.
. .:-: -: :-:.. :
::. ..
: :::
;:::..::: .
-.:. ::
::
:
:: :.::
:::.:- .-:...::.. .:.
:: . .:-. ::
:
..:.::
::.;;

: -:-.:
-:;
:.::.<.::-.
- .-:.- : -: ...:..-:. -.::
.

::
..
.

.
.

..

.
.
.

.
.

::

:
.:
:

.
.

.
.

..

..
.
.
.
.
-
..

.
.
.
,
,
.

.
.

.
.

.
.

.
....
:
:

:..
.
.

.:.. ..:
.:
- ::
.. :
.
.

13.4 THE FERMI-DIRAC DISTRIBUTION


As we have discussed .. Boltzmann statistics can be applied when the
particles of the assembly are distinguishable
a solid lattice. In this section we assume
and

fl)

say, by their positions i n

that the particles are identical

indistinguishable, and (2) that they obey the Pauli exclusion principle. This

means that no quantum state can accept ffi(>re than one particle, taking spin
into account. Such particles have half-integer spin and are given the generic
name

fermion.

Examples of fermions are electrons, positrons, protons, neu

trons. and muons. An important application of Fermi-Dirac statistics is to the


behavior of free electrons in metals and semi-conductors.
Invoking once again our picture of energy level shelves with quantum state
boxes resting on them., we emphasize that in the Fermi-Dirac case, one particle
or no particle occupies a given state so that

N;

gi for all j (Figure 13.2).

<

In determining the thermodynamic probability for a given macrostate we


note that the group of 8; states is divisible into two subgroups:
are to contain one particle and

(g1 - Ni)

N; of the states

of the states must be unoccupied. The

counting problem is precisely the same as that of the coin-tossing experiment ..


where we divided N particles into

N1

heads and

N - N1

tails and obtained

N!

Here

(13.25)
for the jth energy level. The total number of microstates corresponding to an
allowable configuration is simply the product of the individual factors of the
forrtl of Equ_ation

(13.25) for all the levels:

IJ

To maximize

WFo

gI.'.

(13.26)

we follow exactly the same procedure as in the case of

Boltzmann statistics. We use the method of Lagrange multipliers assuming


.

} th
energy
lever

Figure 13.2

1 l

___

__,) {

f l

) L.

._I
___.lj 1L...,_-..Jj

__

e1

NI

Fern1ions in quantum states with energy

- 7

gi

The Fern1i-Dirac Distribution

Sec. 13.4

143

conservation of particles and energy, to calculate the occupation numbers for


the equilibrium macrostate. Taking the logarithm of Wro, we have
In

wFo

(13.27)

'

'

Using Stirling's approxi111ation, this becomes

In wro

(g;
InN;+
g;g;
In
N;
N;
[g;

N;)ln(g;- i'V;) + (g;

N;)]

'

[g; In g; - N; In N;

(g; - N;) ln(g, - N;) ].

(13.28)

'

'

a(In Wfo )
----

aNJ.

iJf/1
iJc/J
+a
+
aN:.J
aN.J

0,

1, 2, .

n.

lead to
-

N.ln N +

iJNI.

'

,__

(g;

N;)ln(g; - N;)

'

+a aN

. N; + f3 aN
'
J

(13.29)
.

'

Here we have taken into account that

g; In g;

aN.I

0,

since the g;'s are not functions of the N;'s. Working out the derivatives, we obtain

Ni
l

(gi- )
\,

-I

or

NI

<'

Chap.

244

Classical and Quantum Statistics

13

or

(13.30)
As before, we set

{3

- 1/kT.

H owever the determination of the other


.

Lagrange multiplier by summing over the N/s doesn't work for this distribu
tion. Provisionally, we associate

with the chemical potential J.L divided by

kT,

and reserve for later the physical interpretation of this connection. Thus we set

(13.31)
so that

( 13.30) becomes
(13.32)

The corresponding result for the con tinuous energy spectrum is


1

(13.33)

13.5 THE BOSE-EINSTEIN DISTRIBUTION

Not all elementary particles o bey the Pauli exclusion principle; ph<>tons are
the most notable exception. Thus we need st at is tics for indistingtlishable parti

cles, any number of which can occupy a given quantum state. Such particles
have zero or integer spin and are called

bosons.

The sta tist i cs developed for

this case leads to the Bose-Einstein distribution, named after its inventors.
Counting microstates here is slightly more complicated than it is for
Ferrtli-Dirac statistics. For the jth energy level there will be g1 quantttm states
containing a total of N1 identical particles with no restriction on the number of
particles in each state. It is convenient to depict the arrangement of the N1 parti
cles among the g1 states by

(g1

1) _partitions or lines and N1 dots (Figure 13.3).

jth
energy
level

Fig11re 13.3

N-I = 14

Bosons in quantum states with energy

e,.

Sec. 13.5

The Bose-Einstein Distribution

145

Now, we can obtain new microstates b' shuffling the lines and dots while keep

gi and fixed. We ask: in how many ways can the


1) symbols (lines and dots) be arranged into (gi- 1) lines and

ing the numbers

(+gi

dots? Again we have the binomial problem of the coin-tossing experiment.


Clearly the answer is

W =

(+gj-1)!
!(gj- 1)!

for the jth energy

level. The

(13.34)

---

--

of

number

microstates for a

given

macrostate is,

then, the extended product

(+gj-1)!

}. g}

j=l

Taking the logarithm of Equation

In

WsE

(13.35)

(13.35), we have

=
'

'

Invoking Stirling's approximation, we obtain


In

w8E

[ ( lV;

g;

1) In ( N;

8; - 1 )

- ( N;+ 8;- 1) - N; In N;

1)+ (g;-1)]

+ N;- (g;- l)ln(g,

[(N;

+ g;-

l)ln(N;+g;-1) - N; InN;- (g;- l)ln(g;-1)] .


(13.37)

With the constraints taken into consideration, the method of Lagrange multi
pliers gives
a (In

WeE

a NJ.

+a

dc/J
aN.J

J3

iJfi
aN.1

0, j

1, 2, 3, .

n,

which leads to

iJN1.

l.

( N;+g;

1 )In ( N;+g;

1) I

(13.38)

Chap. 13

246

Classical and Quantum Statistics

or

g -

N1

N;

g;

This gives

In

NJ + gJ

Q
= -a- tJej.

N1

Neglecting unity compared with Nj + gi we obtain the result


..

With f3

-1/kT and

a =

----

(13.39)

J.lfkT. we have

(Bose-Einstein distribution).

For the continuous energy spectrum

(13.40)

..

(13.41)

13.6 DILUTE GASES AND THE MAXWELL-BOLTZMANN

DISTRIBUTION
Since elementary particles have half-integral or integral spins.. the particles of
a gas are either fermions or bosons: they are clearly one or the other.
However, it is useful to consider another kind of statistics pertinent to a so

called dilute gas. The word --dilute means that for all energy levels.. the occu
pation numbers are very small compared with the available number of
quantum states (most quantum states
N , << ,.

are

empty). We assume that

for all j.

This condition holds for real gases except at very low temperatures.

(13.42)

Dil ute Gases and the Maxweii-Bolamann Distribution

Sec. 13.6

247

If in this region very few states are occupied at all, it is extremely


unlikely that more than one particle will occupy a given state. Thus it is irrelevant whether or not the particles obey the Pauli exclusion principle, and we
might therefore expect Fermi-Dirac and Bose-Einstein statistics to be approx
imately identical in the dilute gas limit. This is indeed the case, as can easily be
seen. Recall Equation (13.26),

(13.43)

Wfo=

Now

gj!
----

(gj- )!

gj(gj - l)(gj- 2)... (gj-

l)(gj- Nj)!

._

J
N

so that

(13.44)

The approximate value is slightly greater than the exact value since factors
such as (g - 1}, (gi- 2), etc. are written as Ki

(gi

Ni- 1)!

(13.45)

!(gj- 1)! '

where

(gj

- 1)!

(gj

- l)(gj

- 2)... (gj

- )(gj- 1)!.

We see that in the numerator of Equation (13.45) there are

(gi- 1)! so that for << gi,

terms ahead of

which is slightly less than the exact expression. Thus

gfj(gj- 1)!
i !(gj- 1)!

(13.46)

Classical and Quantum Statistics

Chap. 13

248

It is apparent that for a dilute gas Ferrni-Dirac and Bose-Einstein statistics


give virtually the same therrnodynamic probability This ""classical limit_" called
.

Mttrwell-Boltzmann statistics, was investigated long before the development of


quantum mechanics. Note the difference between Maxwell-Boltzmann statistic&
Equation (13.46), and Boltzmann statistics_ Equation (13.3 ):

(13.47)

The much larger Boltzmann proba bility includes the pernltltation N! tlf the N
identifiable particles, giving rise to additional microstates.
The distribution of particles among e11ergy levels can be found by the
usu al method of Lagrange multipliers. However, the result can be written
down immediately by obser ving that

'lVMB

and

1.08

differ only by a c<>nstant

the factor N!. Since maximizing the thermo tlynamic probability involves tak
ing derivatives and t he derivative <>f

constant is zero_ we get pre cisely the

Boltzmann distribution
Nj
!'""""'

N e-

This should not come as

cJ/kT
-

(Max\vcii-B<)Itzmann distribution).

( 13.48)

surprise: two functions differin g only by a constant

will have maximum values at the same point.


Because the two distributions are identical, Boltzmann statistics and
MaxweJJ-BoJtzmann statistic s are frequently C<>nfused with each other.
Boltzmann statistics assumes distinguishable (localizable) particles a11d there
fore has limited application_ largely to solids and some liquids. For gases. either
Fertni-Dirac or Bose-Einstein statistics applies_ depending on the spin of the
particles. Maxwell-Boltzmann statistics is a very useful approximation for the
special case of a dilute gas_ which is

good model for a real gas under most

conditions. It so happens that the cor responding distribution is the same as the
Boltzmann distribution.

13.7 THE CONNECTION BETWEEN CLASSICAL AND


STATISTICAL THERMODYNAMICS
The statistical expression for internal energy is

Sec. 13.7

The Connection Between Ciassical and Statisticai Thermodynamics

149

Taking the differential, we have

(13.49)

where si is some function of an extensive property X such as the volume:

Since dei -

(dBi/dX)dX, it follows that

Nde
}
J

..,
=

dei

.J dX

dX.

Let

Then

dU

edN
1
I

Y dX.

For two states with X the same, dX

0 and

(13.50)

The classical analog to Equation (13.49) is

dU

TdS- YdX,

so that

(dU)x

TdS.

Comparing Equations (13.50) and (13.51), we see that

(13.51)

Classical and Quantum Statistics

Chap. 13

250
Accordingly,

Furthermore, since aQ,

e dN
J
J

T dS and aW,

aQ,

}' dX it follows that

and

(13.52)

The first equation states that heat transfer is energy resulting in a net redistri
bution of particles among the available energy levels. involving no work. In
Figure 13.4 heat added to a system shifts particles from lower to higher energy

levels.
The second part of Equation (13.52) can be interpreted in terms of an
adiabatic reversible process (no heat flow) in which work is done

ott

the sys

tem (note the negative sign). An increase in the system"s internal energy could
therefore be brought about by a decrease in volume with an associated
increase in the e1 's. The energy levels are shifted to higher values with no redis
tribution of the particles among the levels (Figure 13.5 ).
The relationships of the two parts of Equation ( 13.52) provide a deeper
understanding of the fundamental concepts <>f thermodynamics.

'

'

'

'
'

N,

--

'

'

-.... ...- .....

'

"

, ..... ....

'

'

--

'

'

'

'

Figure 13.4 Heat added to a


system moves particles from
lower to higher energy levels.

.... .

'

'

'

'

NJ

'

'
'

'
'

'

'

---

'

'

'
'

Figure 13.5

Work done on a

system moves energy levels to


hig her values.

'

'

'

Sec. 13.7

The Connection Between Classical and Statistical Thermodynamics

251

We turn our attention next to the chemical potential, the quantity appear
ing thus far in two of the distribution functions. In Chapter 9 we saw that when

matter is added to or taken from an open system, the change in internal energy is
dU

TdS- PdV +.p,dN,

(13.53)

where 1-L is the chemical potential defined on a per particle basis.* Now the

Helmholtz function is defined as F


dF

U- T S so that

TdS- PdV + p,dN- TdS- SdT,

or
dF

-SdT- PdV + JLdN.

Therefore,
JL

iJF
=

aN

(13.54)

T.V

Let us calculate firstS and then F fo r Maxwell-Boltztnann statistics:

(13.55)

with

Then

k ln

w =

N,.In
g.
J
1

--- k
,_..

(13.56)

Substituting the distribution function of Equation (13.55) in Equation (13.56),


we obtain

*Equation (9.1) is dU

T dS- P dV

is Avogadro's number. Thus, if we


p,dN;p, is the chemical potential energy per particle.

defined as energy per kilomole. But n


de fi ne J.' as JJ.*/N A then p,*dn

+ JJ,*dn, where J.L* is the chemical potential

NINA, where N,.

Classical and Quantum Statistics

Chap. 13

252
or

==

u
T

Nk ( ln

In

1)

(13.57)

Then

F
Using Equation

U- TS

-NkT(ln

In N +

1).

(13.58)

(13.54) we find that


J.L

- k T ( ln Z- In N + 1) +

NkT
N

or

11 =
r-

kT

In

N
--

( 13.59)

For gases, the partition function Zis proportional to the volume as we shall see
in Chapter 14. Thus /J- is an increasing function of both the number of particles
per unit volume and the temperature. This makes sense since particles flow from
higher to lower chemical potential and fr{>rn hi gher to lower concentrations. The
difference in chemical potential is a true potential energy: for two systems it is
the potential barrier that will bring the systetns into diffusive equilibrium.
An example is the coexistence of two r>hases of the same substance

a liquid and its vapor at the same temperature. In Chapter

say..

9 we observed that

the chemical potentials for two phases of a given constituent must be equal for
the system to be in diffusive equilibrium. This means that there is no net transfer
.

of particles from one phase to the other: the total number of particles/Vis a constant. This is precisely the condition that allowed

us

to determine the Lagrange

multiplier a. Thus the association of a with Jl/k Tis completely plausible.


Finally we note that Equ ation
,

(] 3.59) can be written


==

e-LikT

(13.60)

Substituting this in the Maxwe11-B())t7.mann distribution gives

--

---

'

e(t,- J.t}rkT.

(13.61)

We are now in a position to summarize a11d compare the distribution functions


we have derived in the previous sections of this chapter.

153

Comparison of the Distributions

Sec. 13.8

I I 18

I IIIII-I II 111M

IU

S
N
IO
T
U
IB
R
T
IS
D
E
H
T
F
O
N
O
IS
13.8 COMPAR

The distribution functions for identical indistinguishable particles can be rep


resented by the single equation

Ni

gi

(13.62)

-----

e<e,- p,)fkT +

'

where
+ 1 for FD statistics
a =

1 for BE statistics

0 for MB statistics

For

<<

gi, the denominator in Equation (13.62) is very large compared with

unity, and the MB distribution is an approximation to the FD and BE distribu


tions, as we have seen. Plots of y
N;/g1 versus x ( ei -t) /kT are shown in
-

Figure 13.6.

A word about the sign of JL is appropriate. The value of the chemical

potential in some sense is arbitrary. When particles are being transferred from
one system to another, it is important to establish a common zero-energy ref

erence point from which all potential energies are measured. The problem is
exactly analogous to defining a reference level for the gravitational potential
in classical mechanics or for a potential well in quantum mechanics. Thus
ei

p,, the thei"mal energy of a particle, is the energy measured relative to the

lowest point of the potential well. If a particle which itself has zero energy is
inserted into a system, the ensuing interactions with its neighbors are most
often attractive, in which case the chemical potential is negative. When

MB

BE

--

FD
Figure 13.6 Comparison of
the FD, BE and MB
distributions.

-3

-2

-1

X=

0
ei

1
-

kT

J.'

Chap. I 3

254

Classical and Quantum Statistics

systems are not exchanging particles. we can choose any convenien1 reference
level, which is usuallv taken as zero.
.

Examining the curves of Figure 13.6

1. BE curve: y

1/(ex- 1 ) :

.t

for

The distribution is undefined for


wher e

centrate) in regions

curve: y
For .x -

2. FD

\VC

..

note the fo)Jowing:

)0 y ) x and for lar ge


..

..

x,

i=s

e-x.

< 0. Particles tend to condense (con

e is small. that is, in the lo\ver energy states.


;
-x
+ l ) ; for x = 0, y = 1/2, and for x large, y :=e .

1/(e

y- ) 1. A t the lower levels with eJ


J.L negative, the quantum states are nearly un i form ly populated with one particle per state.
x,

3. MB curve: The curve y

e-x lies between the BE and FD curves and is

only valid for y << 1. This is the dilute gas r egi on : many.. many states are
unoccupied. Note that

g,.
=

g
,

--

N;
At high temper at ure s

mately unity so that

k1r

( cJ

e- AI

( E.

c. ,

C'

{.;;)

<

c. r

1) << I and the cxponentia1 is approxi


gJ. Since - tends to increase with increasing
-

energy.. so does the occupa t ion number. At /(J' temperatures.. on the


other hand, the population of the )<)Wcr states is fa\'ored: for
g, = g1 N; > - where i < j.
..

StatisticaJ equilibrium is a balance between the rando mizin g forces of


thermal agitation, tending to produc e a uniform populati on of the energy lev

els.. and the t endency of mechanical systetns to sink to the states of lowest
energy.. resulting in highl y ordered populatir>ns. For a negativ e chemical poten-

tial, Equation

(13.60) gives

When many states are available

to

P-

kr

JJ.

kT.

the system, Z is large ()nd 1-L is highly nega

t i ve. A particle seeks a position <>f lo\vest potential energy choosing to be


where the chemical p ot e nt ia l has its largest negative value.

Having developed the statistics and equilibrium state distribution func

tions for all situations of physi c l interest.

\\'e are

no\\' in a positi on to apply

the results to some specific prc)hlems in subsequent chapters.

13.9 ALTERNATIVE STATISTICAL MODELS

The statistical approach d i sc ussed in the previous t\\'O chapters describes an

assembly of Nparticles with total energy U. whe re both Nand U are fixed. The
assembly is

con si der e d

to he isolated in the sense that it does not exchange

Sec. 13.9

Alternatjve Statistical Models

255

energy in any form with an external system. Such an assembly is known as

microcanonical ensemble.
In many experiments the system under investigation is not isolated. In
chemical reactions, for example, the reaction vessel is (lften held at constant
temperature by being kept in thertnal contact with a heat reservoir such as a
water bath that absorbs energy generated in the reaction. In this case it is con
venient to consider an ensemble of

large number

say

NA

of identical

assemblies in contact with a reservoir. The ensemble can be regarded as a sin


gle assembly in a heat reservoir provided by the remaining NA

1 assemblies.

The reservoir fixes the temperature of the system rather than the energy. The
individual members of this so-called canonical ensemble can exchange heat
and work with the reservoir and hence the energy of the system can fluctuate.
This is in contrast with the microcanonical ensemble in which there is no inter
action with a reservoir at all.
In spite of this difference, the two et1sembles give essentially the same
results. The reason for this coincidence can be seen as follows. The assemblies
of the canonical ensemble are separate macroscopic objects, like distinguish
able particles, and can therefore be assumed to obey Boltzmann statistics, with
the distribution

N.I

N . -e;fkT
gIe
' Z
z

(1 3 63)

=:

The corresponding distribution for the ensemble involves the energy states of
a single sample assembly, whicl1 alone has a very large number of states with a
given energy. The allowed values of the energy E are extremely close
together, so that the degeneracy

g(E)dE,

gi

can be replaced by a density of states

the number of one-assembly states (microstates) with energy

E and E
becomes N(E)j1VA
energy E. Thus

between

+
=

dE. Sums can be replaced by integrals and NilN


P(E), the probability that the sample assembly has an

-E/kT

ZA

IX

(13.64)

The density of states g (E) is a rapidly increasing fttnction of

E,

whereas

the exponential factor is a rapidly decreasing function. For definiteness, sup


pose that g()

En; then
(13.65)

P(E) is found by setting its derivative equal to zero; the result


(3/2)NkT for an ideal
U. But U
ives n = Ejk T at the peak E
(3/2)N, a very large number. Thus one factor in P(E)
monatomic gas, so n

The maximum of

Chap.

256

13

Classical and Quantum Statistics

is large and increasing at an astronomically high rate and the other is extremely
small and decreasing. As a result

P( E) has an exceptionally sharp peak.


determining the probable fluctuation of E around
..

We are interested in
the mean value

U. To do this it is convenient to expand the logarithm of P(E)


in a Taylor series about U:
ln

P()

In P(U)

dlnP()

dE

E .:.u

. E=U

(- U)

2
U) + .

(13.66)

Here
In P(E)

nln

dlnP()
---

lnP()

------

d E2

The first derivative is zero at E,

1
kT.

---

dE
2
d

-- EfkT,

:=.

.....

.
2
-

U and the second derivative is

2j3N k2T2

Using these values and neglecting terms higJ1er than second orderwe find that

(13.67)
where
3
u 2 :::;: - Nk2T2.
2

(13.68)

This is the usual form of the Gaussian distribution in terms of the standard
deviation

u ,which

our purpose is the

is a measure of the peak width of

P(E).

More useful for

fractional width
1
-

--

Thus for a kilomole of an ideal gas. the deviation from the n1ean value

( 13.69)

U is of

the order of one part in 1013 It follows that predictions of the physical proper
ties of a system using the canonical ensemble are virtually identicalwith those
derived from the microcanonical ensemble in which the energy fluctuation is
zero. When fluctuation effects are important, the canonical ensemble yields

Problems

157

information not obtainable from the microcanonical ensemble. Otherwise the


two are closely related and either one can be derived from the other.

In the canonical ensemble we remove the restriction that the energy U is


fixed. ., but the assemblies are still assumed to have a fixed number

N of parti

cles. As a further generalization, \Ve can consider assembl ies with an indefinit e
number of particles. The resulting grand canonical ensemble is a system i n co n
tact with a reservoir with which it can exchange not only energy but particles

as well. In this model only the mean number of particles is fixed. The grand
canonical ensemble is useful when the constraint of a fixed number of parti
cles is too restrictive, as in the problem of chemical equil i b rium among a num

ber of different species.


Sumn1arizing the three approache the microcantlnical ensemble treats
a single material sample of volume V consisti ng of an assembly of

N particles

with fixed total energy U. The independent variables are V, N, and U The cen
tral concept is the distribution Nil N, the fractional number of p art icles occu
pying the jth energy level. The method is the least abstract of the thtee. -It is
-

mathematically the simplest but is completely adequate for solving the prob
,

lems usually discussed i11 undergraduate texts.


The canonical ensemble considers a collection of NA i dentical assem
blies, each of volume V. A single assembly is assumed to be in contact through

1 assemblies.
a diathermal wall with a heat reservoir of the remaining NA
The independent variables are V, N, and T, wi1ere Tis the temperature of the
-

reservoir. The av eragi ng process involves the probability distribution function

P(E)

N(E)/NA. The

approach is more mathematically abstract than the

microcanonical ensemble, but one can be derived from the otl1er.


'The most general and most abstract of the three models is the grand
canonical ensemble, which consists of open assemblies that can exchange both
energies and particles with a reservoir. The independent variables describing the
ensemble are V, T, and 1-L where 1-L is the chemical potential. The grand canonical
ensemble permits the treatment of more complicated problems than does either
of the other methods and is correspondingly more mathemati cally demnding.

PROBLEMS

13-1 Use the method of Lagrange multipliers to solve the following problems:
(a) A rectangle has a base x and a height y, where x + y 8. Find the values of
x andy that give the maximum area.
=

(b)

Find the area of the largest rectangle that can be inscribed in the ellipse
X

az

y2

fj

What percentage of the area of the ellipse does the rectangle occupy?

Chap. 13

258

Classical and Quantum Statistics

13-2 Five identical noninteracting particles occupy the kth energy level, which is ten
fold degenerate. How many possible microstates are there if

(a) the particles are bosons?


(b) the particles are fermions?
13-3 A model therrnodynamic system in which the allowed nondegenerate states
6e.
have energies 0. e., 2e .. 3e, . , consists of .four parti c les with total energy U
.

Identify the possible distributions of particles, evaluate n

U'k,

and work

out the average occupation numbers for the various energy Jevels

(a)
(b)

when the particles are gaseous bosons:


when the particles are gaseous ferrnions.

13-4 Show that for a system of lV particles obeying Maxwell-Boltzmann statistics, the
occupation number for the jth energy l eve l is give n hy

e.I

.1.

13-5 Show that it is poss i bl e to write the the rmodynamic probability in the general
form

where
-+--

a =

for FD statistics

1 for BE statistics

0 for MB statistics
13-6 Substitute the Maxwell-Boltzmann distribution function into the equation
Le_;dNi

T dS

to obtain

Hint: note that

LdN1

0 and dN1 In

N-

-'
,

13-7 For the Fermi-Dirac distribution .. sketch

d Ni In

lv;/g1 versus e,

for T

0 and for T

sJightly greater than zero.

13-8 Show that for a system of a large number N of bosons at very lo"' temperature
(such that they are all in the nondegenerate lowt!st energy state
chemical potential varies with temperature according to

kT
J..L

r4o

--

0), the

Problems
13-9

(a)

259
At room temperature (300 K), what is the equilibrium number of bosons
per state (N ilgi) for an energy ei
(1) 0.001 e\' above the chemical potential J.L?
(2) 0.1 eV above the chemica) potential p,?

(b)

At room temperature

(300 K), what is the equilibrium number of fermions

per state ( / gi) for an energy ei

(I) 0.1 eV below the chemical potential p,?


(2) 0.1 eV above the chemical potentia) JL?

13-10

(a)

Using results from Chapter 9, sho\\' that

iJN Jv.v
(b)

It follows from the statistical definition of the entropy that

IJ.
Alnw=-k AN.
T
Consider a system with a chemical potential p,

0.3 eV. By what factor is

the number of possible microstates of the system increased when a single


particle is added to it at room temperature?

13-11 A system has energy levels at e

0, e

300k, and e

600k, where k is

Boltzmann,s constant. The degeneracies of the levels are 1, 3, and 5 respectively.

(a)

Calculate the partition function, the relative populations of the energy lev

els, and the average energy, all at a temperature of 300 K.

(b)

k
At what temperature is the population of the energy level at 600 equal to
the population of the energy level at 300k?

13-12 A system with two nondegenerate energy levels e0 and e1 (et

>

eo

>

0) is popu

lated by N distinguishable particles at temperature T.

(a)

Show that the average energy per particle is given by


u =

(b)

U
N

Show that when T

and when T

1 +

e-!Jle
A
t
P f;
e-

de

o,

ljkT.

0,

)lo )0 ,

(c) Show that the specific heat at constant volume is


cv

(d)

Compute

cv

in the limits T

versus e/kT.

-:fkT
e

- ---

kT

(1

--

e-,;/kT)2

0 and T --+ ::,: and make a careful sketch of

cv

rea

14.1 Thermodynamic Properties from the Partition


Function
14.2 Partition Function for a Gas

263
265

14.3 Properties of a Monatomic Ideal Gas

266

14.4 Applicability of the Maxwell-Boltzmann Distribution

268

14.5 Distribution of Molecular Speeds

269

14.6 Equipartition of Energy

270

14.7 Entropy Change of Mixing Revisited

271

14.8 Maxwell's Demon

273

261

14.1 THERMODYNAMIC PROPERTIES FROM THE PARTITION


FUNCTION
The power of statistical thertnodynamics is beautifully demonstrated in the
application of the theory to an ideal gas. We shall begin by showing that all the
thermodynamic properties can be expressed in terms of the logarithm of the
partition function and its derivatives. Thereafter we need only to evaluate the
partition function for a gas to obtain its thermodynamic properties.
In Section

13.7

we obtained the following relations for the Max,vell

Boltzmann distribution, valid for a dilute gas:

u
=

Nk (ln Z -

NkT(ln

In

Z- InN+

p, = - kT( ln Z - In

N)

1),
1),

(14.1)
(14.2)
(14.3)

To these we can add the other thermodynamic potentials and the pressure.

1. Internal energy
Using the Maxwell-Boltzmann distribution function, we can write

(14.4)
with

Z=

(14.5)

Differentiating Equation (J 4.5) with respect to T, we have


iJZ
-

aT

kT

(14.6)
263

The Classical Statistical Treatment of an Ideal Gas

Chap. 14

164
Since

e(V),

V constant

keeping

means treating e1 as a constant in

tuting it in Equation ( 14.4)we obtain


N

az

.,

(kT)

aT :

or

2.

a In z

1VkT2

(14.7)

aT

Gibbs function
The calculation of G is trivially simple; in Chapter 9 we saw that

J.LN

for a single component system. Using Equation (14.3), we have


G
3.

-NkT(In Z- InN).

Enthalpy
The enthalpy is defined

TS

+ PV, so that G

Substituting Equations
H

as

U + PV and
H
TS_ or

the Gibbs function is

G + TS.

( 14.1 ) ( 14. 7),


,

U + NkT, or
H

4.

( 14.8)

and ( 14.8) in this equation yields

Nk T 1 + T

a In z

(14.9)

ar

Pressure
A reciprocity relation from Section

8.5 gives

aF
==

Taking the derivative of Equation

av

-r

( 14.2)

with respect to

V with

T held

constant gtves

iJ ln Z
NkT
a\'

(14.10)

Sec. 14.2

Partition Function for a Gas

265

14.2 PARTITION FUNCTION FOR A GAS

The definition of the partition function is

Suppose that gi

1 for all j. For convenience, we assume temporarily that the

sum runs from 0 to n

1 and we take e0

0. Then

Suppose that the energy levels are far apart. What does Z look like? For low
'
temperatures, with the ei s differing significantly from one another, the expo
nentials in the series get successively smaller quite rapidly, whereas at higher
temperatures the falloff is more gradual (Figure 14.1 ) The partition function is
a measure of the states available to the system and the distribution gives the
occupation numbers, which are proportional to the factors e-e/kT. We see that
.

for low T, essentially all the particles are in the ground state; for high T, the
populations of the excited states are sigitificant.
For a sample of gas in a container of macroscopic size, the energy levels are
very closely spaced and many states are available to the system (Figure 14.2).
It follows that the energy levels can be regarded as a continuum and we can
use the result for the density of states derived in Chapter 12. In Equation (12.26)
we use a spin factor l's

1 since the gas is composed of molecules rather than

-t=lkT
e J

\
\
\

'

low T

'

'

'

'

1\
\

high T
'

'

'

,,

,
,_
.... ....
.

. ... ...

,
' -..
., ,

._,_._,_

...... ......
.... .... __

.._.__ ___,_____...
=__

so

el

(a)
Figure 14.1

e2

ej

:
___._____...._,..

_
...._____
_

60

e1

(b)

Successive terms of the partition function sum for (a) low


temperature; and (b) hi gh temperature. The energy level spacings are
comparatively large for a small volume.

83

ei

166

Chap. 14

The Classical Statistical Treatment of an Ideal Gas


- &ikT

low T

high T

1
'

'

'
I
'

'

.
.

I
I

I
'
'

I
I

'

E
J

(a)

I
I

I
I
I
I

.
I
'

I
)
I
.

'
'

I
'

'

'

I
I

I
I
I
I

e
1

(b)

Figure 14.2

Successive terms of the partition function sum for (a) IO\\'


temperature; and (b) high temperature. Here the energy levels are very
closely spaced when the volume is large.

spin 1/2 particles; that is the quantization associated with the spin does not
apply. Thus

g(e)de

27TV
h3

---

1'2 l/2d
e.
ml e

(14.11)

Then

Z=

(I
.

The integral can be found in tables and is

This gives the partition function for a gas under the assumption that the
energy levels are so closely spaced that they form a continuum:

(14.12)

14.3 PROPERTIES OF A MONATOMIC IDEAL GAS


The partition function depends on both the volume V and the temperature T.
We need In Z and its partial derivatives:

Sec. 14.3

Properties of a Monoatomic Ideal Gas

In Z

= 2" In T

In V

a In

2 In

2mk
h2

'
V

3 1
2 r

--

Substituting these expressions in the equations of Section

PV

NkT

and

a In Z

iJT

av

267

14.1 gives

3
zNkT.

(14.13)

However familiar these equations are, their appearance here is an exhilarat


ing development. It represents the capability of statistical thermodynamics
to produce relations that occur in classical thermodynamics as empirical
generalizations.
The calculation of the entropy is even more dramatic. It shows that the
statistical theory can give results not at all obtainable in classical thermodynam
ics. Using Equation (14 1 2) and U
.

(3/2)NkT in

Equation

V (21TmkT)312

(14.1), we obtain
(14.14)

This formula is known as the Sackur-Tetrode equation for the entropy of a


monatomic gas. A rearrangement of terms yields

3
Nk - In T
2

+ In

S0,

(14.15)

where

S0
Recalling that s

Nk

= S/n and Nk/n


s

Cv

2
=

In

ln

R, we obtain

R In

so,

(3/2)R. This is the equation for the specific entropy we obtained


earlier. We note that s0 is an undeterrnined constant in the classical theory.
T he Sackur-Tetrode equation shows us how to calculate the number S0,
where

cv

which could only have been previously estimated on empirical grounds. We

The Classical Statistical Treatment of an Ideal Gas

Chap. 14

268

therefore now have two different ways of evaluating the entropy <>f an ideal
monatomic gas. From measurements of the heat capacity at various tempera
tures we can make an empirical detertnination and we can also make a theo
retical calculation by using Equation (14.15). Excellent agreement has been
found of the results obtained bv the two methods. We note.. however that
Equation (14.15) is obviously not valid down to absolute zero since S does not
-'

approach zero as T

0.

14.4 APPLICABILITY OFTHE MAXWELL-BOLTZMANN


DISTRIBUTION

Maxwell-Boltzmann statistics is valid u11der the dilute gas assumption

<< gi. How good is this approximation f<>r gases under average conditions?

We can write Equation (14.12) in the form

where

1z0

2 1rm kT
=

312

h2

has the dimension of the reciprocal of the volume (Z is a pure number) .. and is
called the quantzlm ctJncentratiolz of the gas. Thus we can express the Maxwell
Boltzmann distribution as

N-}

gj

N
--e

'<JJ /k.T

- r

Consider helium gas under standard

and T

273 K, so that

N
V

1
-- e
n0

-
J

;kT

( STP) conditions. Then 111

6.65

><

10--27 kg

The quantity N/V, the number of particles per unit volume, can be f<>und from
Avogadro "s Ia\\:

Sec. 14.5

Since

Distribution of Molecular Speeds

269

e/kT is of the order of unity and so therefore is

e-e/kT,

The result indicates that on the average only

few states in a million are occu-

pied. Helium gas is thus highly dilute under normal conditions. When
we say that the gas is in the

Ni

<< gj,

classical regime. An ideal gas is defined as a gas of

noninteracting molecules in th classical regime.

14.5 DISTRIBUTION OF MOLECULAR SPEEDS

The Maxwell-Boltzmann distribution of particles among energy levels leads


directly to the speed distributions of molecules of an ideal gas. For a contin
uum of energy levels,

N(e)de

f(e)g(e)de,

N(e) de is the number of particles having


de. For the Maxwell-Boltzmann distribution

where

(14.16)
energies between

and

(14.17)
where, for an ideal gas,

Z=V

2TTmkT
h2

---

312

(14.18)

The density of states is given by

(14.19)
for molecules. Combining these equations we obtain

(14.20)

Chap. 14

270

Here

The Classical Statistical Treatment of an Ideal Gas

e is the single particle kinetic energy.lbus


1

Ci

- mv2

de

=:

mvdv.

and

N ( e)de N ( v )dv, where


N ( v )dv is the number of particles whose speeds are in the range v to v + dv.

Converting from kinetic energy to speed. we have

The computation gives the Maxwell speed distribution

N(v)dv

4 N
=

r.

n1

(14.21)

21rkT

This is precisely the result derived in Section 11.6 by using a more heuristic
argument. The various averages discussed there need not be repeated here.

i!18881
8(.
1118

14.6 EQUIPARTITION OF ENERGY


In the kinetic theory t1f gases" an important result is that

(14.22)
where e is the average energy of a molecule and f is the number of degrees of
freedom of its motion. Thus. for a monatomic gas there are three degrees of
freedom one for each direction of the molecu]e ,s translational motion._ and

Nf:

( 3/2) NkT. It should be possi b l e to obtain Equation ( 14.22) from

statistical thermodynamics and, indeed. that is the case.


Let

be a parameter such that e = e( z). Each

a degree of freedom. For example, if


velocity.. then z

l'x

is to be associated with

is the _r-component of the molecular

v and e = e ( v,{). From Equation ( 11.33) it is evident that


x

we can \Vrtte

(14.23)
where N(z)dz is the number of molecules with parameter
z +

in the range

z to

dz and A is independent of z. From the definition of an average, we have

Sec. 1..4.7

Entrop)' Change of Mixing Revisited

271

e(z)N(z)dz
e(z)

----

(14.24)

N(z)dz
where the denominator insures that N (z) is a probability density function. We
note in passing that (z) = UIN.
We assttte that e(z) is a quadratic function of z, that is, e(z) = az2,
where a is a constant. This is certainly the case when eo: vx2 Therefore
Equation (14.24) gives
oc

7r

112

where a= ajk T. Thus


(14.25)
The principle of the equipartition of energy states that for every degree of
freedom for which the energy is a quadratic function, the mean energy per par
ticle of a system in equilibrium at temperature T is (1/2)kT.

14.7 ENTROPY CHANGE OF MIXING REVISITED

In Section 9.5 we used the results of classical thermodynamics to determine


the change in entropy due to the mixing of two different gases. We considered
a volume V divided into two equal parts by a partition. On one side of the par
tition there are n1 = x1n kilomoles of an ideal gas and on the other side there
are n2 = x2n kilomoles of a different gas, both gases being at the same tem
perature and pressure. The partition is removed, each gas diffuses into the
other, and a new equilibrium state is obtained in which both gases occupy the
total volume. The change in entropy as a result of the mixing process was
found to be
(14.26)
where x1 and x2 are the kilomole fractions of the constituents.

Chap. 14

271

The Classical Statistical Treatment of an Ideal Gas

Let's examine the problem from the point of view of statistical thermo

dynamics. First. we observe that n

That is,

x1

= Nk/R, so that

is the fractional number of molecules of gas 1 and x2 is the corre

sponding quantity of gas 2, with Xi +

x2

= 1.

The effect of the mixing is to itzcrease the number of configurations

available to the system by

facto r of

au'=

N!
N) .'Nt
.

N!

(14.27)

the number of ways of arranging the N molecules with N1 in one container and
N2 in the other. The correspt)nding entropy change due to mixing is
S

=k ln

=kIn N! - k ln(x1N)!

A1v

k l n(x N ) ! .

Using Stirling,s approximation we obtain

or

(14.28)
Noting that Nk

=nR, we see that this is precisely the classical result.

We ask: For what proportion of the two constituent molecules is the.

entropy of mixing a maximum? Let x1


u

.lS
k
N

In

=x and x2 = 1

x, so that

(I - x)ln(l - x).

To maximize u. we take
du

-d-x

=0 = In

Thus xj ( 1 - x)

+ I

1 and x

1
=

1
_

ln ( I

- x)
.

x
_

=ln

- .\"

1/2. The entropy change due to mixing is there

fore a maximum when the two gases are present in equal proportions. This is

Sec. 14.8

Maxwell's Demon

273

in no way a surprise since the mixing problem is exactly equivalent to the

tossing experiment.

coin

The statistical analysis of mixing sheds light on the Gibbs paradox


described in Section 9.5. Clearly, if the gases are identical, there is no increase
in the number of microstates when the partition is removed and the occupa
tion numbers remain the same. As we have seen, the sharp distinction between
distinguishable and nondistinguishable particles is fundamental in quantum
mechanics and in statistical thermodynamics.The Gibbs paradox was viewed
as a serious problem in the purely classical context in which it first arose.

ELL'S DEMON

14.8

Jatnes Clerk Maxwell viewed the secottd law of thermodynan1ics as statistical


even though it was left to Boltzmann to develop its theory. Maxwell, however,
illustrated the statistical nature of the law with his famous ''demon." In 1867
he suggested a conceivable way in which if two things are in contact, the hotter
"could take heat from the colder without external agency."*
Consider a gas in a container divided by an adiabatic diaphragm into two
sections, A and B (Figure 14.3). The gas in A is assumed to be hotter than the

0
0

Fi

14.3 Maxwell's demon allows a molecule to


pass from A to B if its kinetic energy is less than the
average molecular kinetic energy in B. Passage from B
to A is allowed only for molecules whose kinetic
energies exceed the average kinetic energy per molecule
in A. The demon is thumbing his nose at the second law
of thern1odynamics.

1867. Reprinted in C.G. Knott, Life and Scientific


Work of Peter Guthrie Tait, Cambridge University Press, 1911.
*

J.C. Maxwell to P.G. Tait, 11 December

Chap. 14

274

The Classical Statistical Treatment of an Ideal Gas

gas in B. From the molecular point of view, the higher temperature means a
higher average value of the kinetic energy <)f the molecules in A compared
with those in B. Maxwell wrote 4N<lW conceive a finite being who knows the
paths and velocities of all the molecules by simple inspection but who can do
no work except open and close a hole in the diaphragm by means <>f a slide

without mass.'' Maxwell s beitlg is assigned t<l open the hole in the diaphragm
to let through an approaching molecule in A when that molecule has a veloc
ity less than the

rm.s

velocity of the molecules in B. At the same time, a mole

cule from B is allowed to pass through the hole into A only when its velocity
exceeds the

rms

velocity of the molecules in A. These two procedures are car

ried out in such a way that the number of molecules in A and B do not change.
As a result, Hthe energy in A is increased and that in B is diminished; that is,
the hot system has got hotter and the cold colder and yet no work has been
done; only the intelligence of a very observant and neat-fingered being has
been employed." In this way the being vi<llates the second law. 'Only

o
o

0
X

0
X

X
o

.I

0
Maxv.etrs

X
X

Demon

X
X

X
Maxwtfi

Demon later

Figure 14.4 Maxw e ll s demon in action. In this


version the demon ope rates a valve. allowing one
'

species of a two-component gas (hot or cold)


through a partit ion separating the gas from an
initially evacuated chamber. Onlv fast m ol e cul es are
allowed through .. resulting in a cold gas in one
chamber and a hot gas in the other. (Reproduced
.

from Gasser. R.P.H. and Richards. W.G., l:.ntropy and


Energy Levels, Clarendon Press.. Oxford 1974 .. by
..

pertnission of Oxford University Press.)

we

Problems

275

can't," added .Maxwell. "not being clever enough." Kelvin immediately gave
the being the name ''demon" by \Vhich it has been knoNn ever since.

Maxwell's demon has attracted an enormous amount of attention since


it appeared well over a century ago. It has been said that the ''demon is catlike;

several times it has jumped up and taken on a ne\v character after being left
for dead."* Because of its connection with information theory and computer
science, the subject of the demon continues to generate irtterest. We shall
return briefly to it in the last chapter.

PROBLEMS

14-1 (a) Calculate the entropy S and the Helmholtz function F for an assembly of
distinguishable particles.
(b) Show that the total energy U and the pressure Pare the same for distin
guishable particles as for molecules of an ideal gas while S is different.
Explain why this makes sense.

14-2 Show that for an assembly of N particles that obeys Maxwell-Boltzmann statis
tics, the occupation numbers for the most probable distribution are given by:

14-3 (a) Show that for an ideal gas of 1V molecules,

where

(kT)Si2
-

(b) Forei

(3/2)kT, T

300K,P

103 Pa,andm

10-26kg,calculategi/hJ.

14-4 Calculate the entropy of a kilomole of neon gas at standard temperature and
pressure. The mass of a neon atom is 3.35 x 10-26 kg.
14-5 Calculate the chemical potential in electron volts for a kilomole of helium gas at
27
standard temperature and pressure. The mass of a helium atom is 6.65 x 10- kg.

See

Maxwell's Demon: Entropy, Information. Computing,

by H.S. Leff and A. F. Rex,

Princeton University Press, 1990. This bock contains a chronological bibliography and reprints of

25 papers on the subject.

Chap. 14

276

The Classical Statistical Treatment of an Ideal Gas

14-6 A tank contains one kilomole of argon gas at 1 atm and 300 K. The mass of an
argon atom is 6.63 x to-:?tt kg.
(a) What is the internal energy of the gas in joules? What is the average energy
of a molecule in e V?

(b) What is the partition function Z?


(c) What is the chemical potential J.! in e v:
(d) What is Nijg1?
14-7 The partition function of a system that obeys Maxwell-Boltzmann statistics is
aVT\ where a is a constant. (..alculate U. P, and S.
given by Z
=

14-8 An ideal monatomic gas consists of N atoms in a volume V. The gas is allowed to
expand isothern1ally to fill a volume 2V. Show that the entropy change is Nk In 2.

ea

I I

15.1 Introduction

279

15.2 The Quantized Linear Oscillator

279

15.3 Vibrational Modes of Diatomic Molecules

282

15.4 Rotational Modes of Diatomic Molecules

284

15.5 Electronic Excitation

287

15.6 The Total Heat Capacity

288

177

..

15.1

0'>
001!1

)')1>
..

----

INTRODUCTION

n
u
co
en
e
w
m
le
b
ro
p
y
rn
o
t
h
a
e
lv
so
re
n
ca
el
d
o
m
l
ca
ti
is
at
st
e
th
therefore that
s
di
s
s.
ga
c
o
at
di
a
to
s
jc
arr
yn
od
errn
th
l
ca
si
as
cl
of
n
io
at
lic
p
ap
e
t
in
d
re
te
e
th
t
ve
g
to
tls
fa
gy
er
en
of
on
iti
rt
a
i
p
qu
e
cussed in Chapter 11, the pr inciple of
p
re
sc
di
is
th
of
n
io
at
an
pl
ex
e
Th
y.
cit
pa
ca
at
he
c
ifi
ec
sp
observed value of the
e
th
ing
fac
ge
en
all
ch
nt
rta
po
im
st
e
mo
th
be
ll
to
we
ancy was considered by Max
statistical theory. In thi s chapter we shall see how the problem is solved.

15.2

THE QUANTIZED LINEAR OSCILLATOR

Until now we have confined our attention to systems of particles that have
translational degrees of freedom only. To deal with particles such as diatomic
molecules, we need to investigate so-called

internal degrees of freedom such as

vibrations, rotations, and electronic excitations. We shall address vibrations frrst.


We consider an assembly of N one-dimensional harmonic oscillators. We
assume that the oscillators are

loosely coupled in that the energy exchange

among them is small. This means that each particle can oscillate nearly inde
pendently of the others. We further assume that eaclt oscillator is free to

vibrate in one dimension on ly with some natural frequency

v.

From quantum mechanics, the single particle energy levels are

given by

(15.1)
Note that the energies are equally spaced and that the ground state has "zero
point" energy equal to hv/2. The states are nondegenerate in that

gi

For the internal degrees of freedom, Boltzmann statistics

1 for allj.

applies

. The

assumption may seem questionable since Boltzmann statistics characterizes

distinguishable particles (localized in a crystal lattice, for example) and would


279

280

Chap. IS

The Heat Capacity of a Diatomic Gas

therefore appear to be inappropriate for the treatment of an assembly of


diatomic molecules. However.. in the dilute gas approximation .. the number of

translational quantum states is so much larger than the number of particles

(gi

>>

N;)

that the great majority of states are unoccupied, a few are occupied

by a single particle, and virtually none have a population greater than one.
Thus, in our treatment of the internal degrees of freedom of diatomic molecules, \Ve can regard the particles
tum states that they occupy.

differentiated by the translational quan-

as

We begin by evaluating the partition function


:r.

gJe jkT

j'=O

The temperature at which kT

J-+ t/2lh /KT.

(15.2)

-0

ht' is caJ1ed the characteristic temperature 8:

JJv

k .

Using this in Equation (15.2), we have


2.

..

e-o 2r

(1

e-eflT(1

e;r

+ e

e/T +

+ e

2e1r

+ .

(e-etr)2

. .

+ ...

The sum in this expression is just an infinite geometric series of the form
I

which equals ( 1
times

(1

+ y

y i .....
2

y) -'.This can easily be seen by taking the product ( 1

2 +
y
-

). Here

.v

z =

atT

1-e

-B;i

( 15.3)

..

the fraction of the total number of particles \\ith energy


=

1 is

.Y)

exp( -0/T) so

We shall also be concerned with the occupation numbers


tribution for g,

e,.

or

with

Ni/ N

The Boltzmann dis-

(15.4)

Sec. I 5.2

The Quantized Linear Oscillator

281

The exponent of the term outside the parentheses can be written


-

()

kT

2T

+
(15.4)

Thus Equation ,

.
1 hv
hv
+
]+2 kT
2kT

hv

j kT

j T.

(15.5)

becomes
(15.6)

A sketch of Equation (15.6) for two temperatures shows that the lower the tem

. perature, the more rapidly the occupation numbers decrease withj (Figure 15.1).
At higher temperatures, more particles populate the higher energy levels.

Next we compute the internal energy of the assembly of oscillators. The

expression we obtained for U in terms of the derivative of In w (Equation


(14.7)) applies here because the Boltzmann distribution is the same as the
Maxwell-Boltzmann distribution. We have

(15.7)
v

Substituting Equation (15.3) in this equation and carrying out the differentia---tion, we obtain

1.0

1
Nk6 - +
2

0.8

8/T

- 1

(15.8)

1.0

9
2

0.8

28

0.6

0.6
NJ
N

NJ
N

0.4

0.4

0.2

0.2

1
(a)

(a) T

15.1
=

(b)

Fractional occupation numbers for quantized linear oscillators with


8/2, and (b) T = 28.

Sec. I 5.3

Vibrational Modes of Diatomic Molecules

283

or

(15.9)
At very high temperatures T/8 >> 1 or 9/T << 1 and
2

1
T

That is, the heat capacity approaches the constant Nk. In the low temperature
limit, forT/8 << 1 or 8/T >> 1, we have e81T >> 1 and

so that
2

-8/T

The rate at which the exponential factor approaches zero as T--+ 0 is greater
than the rate of growth of (8/T)2 so that
Cv

as

0,

consistent with the third law. The variation of Cv wit Tis shown schematically

in Figure 15.3.

k
....
..
N

-----------------

Cv

Figure 15.3 Variation with


temperature of the he.at
capacity of an assembly of
linear oscillators.

...,_ ----_._.

___.__
..

____

28

The Heat Capacity of a Diatomic Gas

Chap. 15

284

At an extremely low temperature

T c(<

0 or

kT

<<

hv;

the oscillators

(particles) are Hfrozen "" into the ground state., in that virtually none are ther
mally excited at any one time. We have U

Nhv/2

==

and C\.

0. This is the

extreme "quantum limit.'" The discrete quantum nature of the states totally
dominates the phy sical properties.
At the opposite extreme of temperature for which

we reach a "classical limit."" Here the expressions for U and

hv.

>> 8 or

Cv

kT

>>

hv,

do not involve

Planck"s consta11t .. the scale of the energy level separation, is irrelevant.

Oscillators of any frequency have the same average energy


dence of

U on kT is associated

k T.

The depen

with the classical law of energy equipartition:

each degree of freedom contributes

NkT/2

to the internal energy. But the

question left unanswered by the classical law is: when is a degree of freedom
excited and when is it frozen out?
The total internal energy of a diatomic molecule is made up of four- con.

tributions that can be treated separately: ( 1) lhe kinetic energy associated.with


the translational motion of the center of the mass of the molecule (this is
(3/2)kT .. the same as that for a monatomic molecule); (2) the rotational
energy due to the rotation of the two atoms about the center of mass of the
molecule; (3) the vibrational motion of the two atoms along the axis joining
them; and (4) the energy of excitation of the atomic electrons. The last three
are internal modes of possessing energy. Because the four contributions can
vary independently it follows that the partition function is a product of the
corresponding factors. This can be

se e n

by n<)ting that the internal energy is a

function of the logarithm of Z.


Having considered the vibrational motion in Section 15.3. we turn our
attention to the rotational contribution.

15.4 ROTATIONAL MODES OF DIAT OMIC M OLECULES


The rotation of a diatomic molecule is modeled as the motion of a quantum
mechanical rigid rotator. The rotation takes place about an axis thr<lugh the
center of mass of the molecule and perpendicular to the line joining the two
atoms. We take I as the moment of inertia of the molecule about this axis:
J.LTo2 where J..L = m 1m2/ ( n11 + 1112) is the reduced mass of the t\\'O atoms
I
=

..

and r0 is the equilibrium value of the distance between the nuclei.


Quantum mechanics states that the all<>wed values of the square of the
angular momentum are
27T and

0 1 2 3 . .
..

..

energy is ( 1/2) 1

..

2
w "

I (I + 1)
.

1f2 where tz is Plancks constant divided by


..

Recall frotn classical mechanics that the r<ltational

where w is the angular velocity. The angular momentum L

is I w so the energy is L 2/21. The quantized energy levels are therefore

e,

I (I + 1 )

2i .

(15.10)

Sec. I 5.4

Rotational Modes of Diatomic Molecules

285

We define a characteristic temperature for rotation:

(15.11)
so

that
(15.12)

Experimental values of 8rot are given in Table 15.1 for several gases; they are
found from infrared spectroscopy, in which the energies required to excite the
molecules to higher rotational states are measured.
The energy levels of Equation (15.12) are deg\!nerate; quantum mechantcs gtves

g,

(2/ + 1)

states for Ievell corresponding to different possible directions of the angular


momentum vector. Given these results, we can write down the partition func
tion for rotation:

(15.13)

The important quantity in this expression is the argument of the exponential.


For T << 8r01, virtually all the molecules are in the fe\v lowest rotational states,
and the series can be truncated with negligible error after the first two or three
tertns. It can be st1own that in this. case both the internal energy and the heat
capacity at constant vblume vanish as the temperature approaches zero (see
Problem 15-4)

For all diatomic gases except hydrogen, the rotational characteristic tem
perature is of the order of 10 K or less. Since these gases have liquefied

TABLE IS. I Characteristic temper


ature of rotation of diatomic molecules.
Substance

8rot (K)

H2
0.,
N2
HCI
co
NO
Cl2

85.4
2.1
2.9
15.2
2.8
2.4
0.36

I
Chap. 15

286

The Heat Capacity of a Diatomic Gas

(indeed, solidified) at such low ten1peratures it is always true that T >> Brot at
ordinary temperatures. Hence a great many closely spaced energy states are
excited and the sum in Equation
write
dx

(15.13) may be replaced by an integral. We


x as a 'ontinuous variable. Note that

l (l + 1) and treat
(21 + 1 )dl. In this approximation,

Z=

[exp( -9roJT)x]

dx

()rot

(15.14)

This result is too large by a factor of 2 for homonuclear molecules such as

H2,

02, and N2 For two identical nuclei, two opposite positions of the molecular
axis correspond to the same state of the molecule. This slight modification has
no effect on the thermodynamic properties

<lf the system such as the internal

energy and the heat capacity.


Using Equation (15.14) in
a ln Z
ar

(15.15)

we find that for T >> Hr01,

Urot

NkT.

(15.16)

and
Cv.rot

T >>(}rot

N k.

(15.17)

us, at significantly high temperatures, the rigid rotator exhibits the eq uipar

tltion of energy between two degrees of free{lom.


At very low temperatures such that Orot >> T, the partition function.
Equation (15.13), can be expanded as
Z

3e- 2Bro)T

5e

--

h#J,.,,/T + .

Retaining only the first two terms, we can write

Since the exponential term is small compared with unitv. we can use. the
approximate relation
--

In ( 1

c)

e,

<< 1

Sec. I S.S

Electronic Excitation

187

Nk ..,..._ __ _

..
____
_
...
,..-.-...

_
_
--- __ ._.... _ _ __

Cv. rot

Variation with
temperature of the heat
capacity of an assembly of
rigid rotators.
Fig11re 15.4

28rot

and obtain the result


In Z

JT.
29ro
3e-

Hence
a In z
iJT

28roJT
e-

61Vk8rot

'

(}rot>>

(15.18)

and
CV,rot

au
aT

3Nk

1'J
._

rot

28 JT
e- n .

....
....
... ... T
8 rot//

(15.19)

Again, we see that Cv approaches zero as T approaches zero, as it must. Fig


ure 15.4 shows schematically tl1e contribution of rotation to the heat capacity

as a function of temperature.

I 5.5 ELECTRONIC EXCITATION

The electronic partition function can e written

(15.20)
where g0 and g1 a re respectively. the degeneracies of the gro un d state and the
,

first excited state, and Be is the energy separation of the two lowest states

divided by Boltzmann's constant k. For most gases, the higher electronic states
are not excited (6e

--

120,000

K for hydrogen) and Z

g0 Thus the partition

Chap. IS

288

The Heat Capacity of a Diatomic Gas

function at practical temperatures makes no contribution to the external


energy or heat capacity.

'

I 5.6 THE TOTAL HEAT CAPACITY


We are finally in a position to account fully for the behavior of Cl/ as a func
tion of T for a diatomic gas at temperatures in the usual range of interest.
Assuming that electronic excitations are negligible.. we have

Cv

==

C'v.tr + Cv.rol + CV.vih

Thus, at ordinary temperatures,

or

C,

Nk -

(15.21)

Table 15.2 gives values for 8,ih for several diatomic gases.
At the lowest temperatures at which the system exists in the gaseous
the

phase,

(3/2)Nk

molecular

(3/2)1zR

motion

is

solely

translationaL

contributing

to the heat capacity. A.s the temperature increases and

approaches Or01.. rotational quantum states are excited. EventuaJiy, for


T >> 8r(lt\ the rotational contribution beomes equal to nR. corresponding to
two

rotational degrees of freedom. Therefore, the heat capacity steps up to

( 5/2) nR

in the region 8n,1 << T << 8,.; Since room temperature lies within

this range.. this is the value usually given for the heat capacity of a diatomic gas.
------- -

TABLE

15.1

Characteristic tempera

tures of vibration of diatomic molecules.


Substanc

H1
02
N2
HCl
co
NO
Cl2

8-Jtb ( K;
6140

2239
3352
4150

3080

2690
810

Problems

289

'

'

'

'

., -

I
'

I
I
'

.
'

Cv
nR

-4
1

(
I

10

50

100
500 1000
8rot Temperature (K)

5000

Figure 15.5

Values of CvfnR for hydrogen (H2) as a


function of temperature. The temperature scale is
logarithmic. (Adapted from Thermodynamics, Kinetic
Theory, and Statistical Thermodynamics, 3rd edition. by
F.W. Sears and G.L. Salinger. Addison-Wesley. 1975.)

At elevated temperatures, the higher vibrational states are excited and


the heat capacity exhibits two additional degrees of freedom, rising asymptot
ically to a value of (7/2) nR for T >> 8vib The characteristic temperature of
vibration depends on the bond strength between the two atoms of the mole
cule and on their masses. In some cases, the diatomic molecules disassociate as
the vibrational energy overcomes the bonding energy.
Experimental values of the heat capacity at constant volume are close to
predicted values over a wide range of temperature. Values of Cv/nR are
shown for hydrogen on a logarithmic scale in Figure 15.5. An understanding of
the temperature variation of the heat capacity of diatomic gases is surely an

outstanding triumph of quantum statistical theory.

PROBLEMS
15-1

(a)

quantum states

IS-2

/ N

Calculate the fractional number

Nil N

(j

0,

1, 2)

of oscillators

in

the three lowest

forT= 8/4 and forT= 8.

(b)

Sketch

(a)

For a system of localized distinguishable oscillators, Boltzmann statistics

versus j for the two temperatures.

applies. Show that the entropy S is given by

-k ln

N.J
N .

The Heat Capacity of a Diatomic Gas

Chap. 15

190

(b)

Substitute the Boltzmann distribution in the previous result to show that

(c)

ln Z.

Nk

Using the expressions derived in the text for U and T, prove that

S
where 8

hv/ k.

8/T

==

N k - 8IT
-

- 1

- In ( 1

H1'
e- 1

Examine the behavior of S as

T approaches zero.

153 Consider 1000 diatomic molecules at a temperature Ovib/2.

(a)
(b)

Find the number in each of the three lowest vibrational states.


Find the vibrational energy of the system.

15-4 (a) In the low temperature approximation of Section 1544 show that the
Helmholtz function for rotation is

Ff(lt

(b) Use the reciprocal relation 5

..

same approximation. Note that

e - 8"-L/T

3 N kT

(aF/aT)v to find the entropy Srot in the


S 0 as T--+ 0., in agreement with the third
-

law.

155 As an alternative evaluation of Zrot to that given in the text, assume that for
T >> 6rot the numbers I in the sum of Equation ( 15.13) are large compared with
unity and replace the summation by integration with respect to I. Show that
Zrot

T/ (Jro1

156 Use the data of Table 15.1 to determine

r0

..

the equilibrium distance between the

nuclei.. for

(a)
(b)

an H2 molecule
a CO molecule.

15-7 Consider a diatomic gas near room temperature. Sho\v that the entropy is
S

312
512
7
V -T
2 rrn1k
Nk -+In
N
2
h2
28rot

if the atoms of the diatomic molecule are identical.

lS-8 For a kilomole of nitrogen (N2) at standard temperature and pressure. compute
(a) the internal energy U; (b) the Helmholtz function F; and (c) the entropy S.
15-9 Using the relation
p

==

NkT

In

itV

show that the equation of state of a diatomic gas is the same as that of a
monatomtc gas.

15-10 Calculate the specific heat capacity at constant volume for hydrogn (H2) at
T
2GOO K.
=

fj;fY!i?t

= \),: :\?::::,':/"'

..

16.1 Introduction

293

16.2 Einstein's Theory of the Heat Capacity of a Solid

293

16.3 Debye's Theory of the Heat Capacity of a Solid

296

291

l!!lliill

'

16.1 INTRODUCTION

'

'
'

'

I
'

f
'

e
t
f
o
le
p
m
a
x

a
is
ic
p
to
is
th
,
15
r
te
p
a
h
C
f
o
t
c
je
b
su
e
densed matter. Like th
cs
m
a
ch
e
m
l
a
tc
s
tt
a
st
r
o
d
e
e
n
e
th
d
n
a
ry
o
e
th
c
ti
e
in
k
l
a
ic
ss
a
cl
f
o
shortcomings

nt
ta
ns
co
at
ty
Ci
pa
ca
at
he
c
ifi
ec
sp
e
th
at
th
ed
rv
se
In 1819, Dulong and Petit ob
-t K -I
e
ol
m
lo
ki
J
10"
X
9
2.4
y
el
at
im
ox
pr
ap
is
s
lid
so
volume of all elementary
the
d
by
ine
pla
ex
be
n
ca
ult
res
is
Th
e.
ur
rat
pe
tem
om
ro
ar
at temperatures ne

a
lin
as
id
sol
the
of
m
ato
ry
eve
g
atin
tre
by
rgy
ene
of
ion
rtit
ipa
equ
of
ple
princi
ear oscillator with six degrees of freedom and then assOCiating an energy of
3R, in agreement
(l/2)kT with each degree of freedom. Then cv = 6 (R/2 )
=

with the observation of Dulong and Petit.


However, additional measurements showed that the specific heat of
solids varies with temperature, decreasing to zero as the temperature
approaches zero. This behavior cannot be explained by the "freezing" of
degrees of freedom when the temperature is decreased since the specific heat
varies gradually with temperature and does not exhibit abrupt jumps by any
multiple of R/2 (in contrast to the specific heat of a diatomic gas). Even at
room temperature the specific heat capacities of certain substances such as
beryllium, boron, carbon, and silicon were found to be much stnaller than 3R.
.
Quantum statistics is needed to explain these discoveries.

I .2 EINSTEIN'S THEORY OFrHE HEAT CAPACITY


OFA SOLID
,

It was Einstein who developed the first reasonably satisfactory theory of the
specific heat capacity of a solid. He assumed that the crystal lattice structure of
a solid comprising N atoms can be treated as an assembly of 3N distinguish
able one-dimensional oscilators (three oscillators for each atom since the
atoms of a solid are free to move in three dimensions). From Equation (15.8),
the internal energy of a solid made up of N atoms is
.
293

The Heat Capacity of a Solid

Chap. 16

294
U

1
3N kfJE -2

1
/
8
e 1 T

where 8E is the Einstein temperature given by fJE


constant volume is

v-

ar v

3Nk

(16.1)

hv/k. The heat capacity at

8
e ' /T

(J

eBEIT

( 16.2)

1 )2

For temperatures very large compared with the Einstein temerature.


Cv 3Nk

(16.3)

3nR.

Thus the high temperature limit of Einstein's equation gives the value of
Dulong and Petit. The failure of their law becomes evident when we examine
the low temperature limit. For 8F./T >> 1,

( 16.4)
As T approaches zer(l, Cv

powers the growth of

a lso
,

( OE/T)-.

goes to zero since the exponent ial decay over.

Einstein s theory also explains the lov.' heat capacities of some elements

at mo d erately high temper at ures. Jf an elen1 ent has a large Einstein tempera
will be large even for temperatures well above absolute
l1vj k must be very large
zero and Cv wil l be small. For such an element (JE

ture, the ratio

8E/T

..

and accordingl y

must be larg e. For an oscillator with force constant

and

reduced mass p., the oscillator frequency is


..

v =

A large frequency value suggests

2rr

( 16.5)

s mall reduced mass or a large force con

stant. corresponding to lighter elements an(l elements that produce very hard
crystals. The theory correctly predicts the failure of the law of Dulong and Petit

for those elements. As an example, the h ea t capacity of diamond approache s

3Nk only at e xtremely high te mp e rat ures

(8:

1450 K for diamo nd )

The essential behavior of the specifi c heat capacity of solids is incorpo

rated in the ratio

OE/T. When

this ratio is large_ the p artit i on function reduces

to the zero-point term\ i m ply i n g that all the atoms are in the ground state and
th e vibrational degrees of freedom are not excited ( Equation

(15.6)). The spe

cific heat remains close to zero since sm all temperature increases are not suffi

cient to excite a si gn ificant number of at()ffiS to the first vibrational state.

When 8E/T is smalL the difference b e t we en e ne rgi es corr espo ndi ng to various

Sec. 16.2

295

Einstein's Theory of the Heat Capacity of a Sotid

vibrational states is small contpared with thettnal energies. Thus the vibra
tional states can be approximated by

an

energy continuutn and treated by

classical theory. For values of 8JT between the two extremes, there is a transi
tion region of partial excitations.
A consequence of tlte fact that c,,; N depends

only

on the ratio 8EfT is

that a single measurement of the heat capacity at one temperature detertnines


its value at all other temperatures. ln additit:n, different elements at different
temperatures will possess the same specific heat capacity if the ratio 8EfT is
the same in each case. The elements are said to be in ''"corresponding states."
The graph in Figure 16.1 clearly shows this behavior.
Careful measurements of heat capacities show that Einstein's model gives
results that fall slightly below experimental values in the transition range of
8EfT between the two limiting values. The discrepancy can be seen in Figure 16.2
in which the heat capacity of lead is shown for temperatures in the range 0-50 K.
2.5'

2.0

t--

1.5 .......
1r'
I kilomole-1 K--l
cv

Pb

Ag

Zn

6. Cu

1.0 .....

+AI
0.5

'

0.5

1.0

1.5

2.0

2.5

Figure 16.1

The specific heat capacity of various solids as a function of T/9E. Note


(NAJN)Cv.(Adapted from Elements of Statistical Th ermo dynamic s'
that Cv
zud edition, by L.K. Nash, Addison-Wesley, 1972.)
=

2.5

r-

2.0

t-

Figure 16.2

The specific heat


capacity of lead showing the
disagreement with Einstein's
theory at low temperatures.
(Adapted from Elements of

104
J kilomole-1 K-1
c, X

0.5

..,_

Statistical Thennodynamics,

Einstein

2nd edition, by L.K. Nash,


Addison-Wesley,

1972.)

25

50

T(K)

Tl8e

The Heat Capacity of a Solid

Chap. 16

196

16.3 DEBYE'STHEORY OFTHE HEAT CAPACITY OFA SOLID


The disagreement between Einstein's result and the experimental data is due
to the fact that Einstein's assumptions about the atoms in a crystal do not
strictly apply to real crystals. The main problem lies in the assumption that a
single frequency of vibration characterizes all 3N oscilJators. Debye improved

on Einstein s theory by considering the vibrations of a body as a whole..


regarding it as a continuous elastic solid. He associated the internal energy of
the solid with stationary elastic sound aves. Each independent mode of
vibration is treated as a degree of freedom.
In Debye's theory a solid is viewed as a phonon gas. Vibrati<>nal waves
are matter waves each with its own de Broglie wavel ength and associated par
ticle. The particle is called a phonon .. with characteristics similar to those of a

photon. We are interested in determining the number of possible wavelengths


or frequencies within a given range.
For quantum waves in a one-dimensional box. we saw in Chapter 12 that
the wave function is l/J
A sin kx, where
=

I" 2 .. 3.. .

n =

( 16.6)

Here A is the de Broglie wavelength and /_j is the dimension of the box. Using
the fundamental equation of \vave motion,

where cis the wave velocity and

the frequency, \ve obtain

tl

2L
=

v.

('

If we consider an elastic solid as a cuhe of volume V

n =

v.

L3, we get

( 16. 7)

nx2 + 1zy2 + n j/ . The quantum numbers tlx, n.r, and IZ:


where, in this case n2
are positive integers. Thus the possible values that they can assume occupy the
=

first octant of a sphere of radius

(Figure 16.3).

Sec. 16.3

Debye's Theory of the Heat Capacity of a Solid

297

dn

n.{

16.3 A shell of
thickness dn of an octant of a
sphere of radius n.

Let

g(v)dv

be the number of possible frequencies in the range

to

Since n is proportional to v, g(v)dv is the number of positive sets of


integers in the interval n to n + dn
that is, within a shell of thickness dn of

dv.

an octant of a sphere with radius n:

Substituting Equation

(16.7)

for n, we obtain

(16.8)
In a vibrating solid, there are three types of waves: one longitudinal with
velocity c1 and two transverse with velocity c1 All are propagated in the same
direction. When all three waves are taken into account, Equation (16.8)
becomes

g( v)dv

4n.Y

3
c,

3
c,

v2dv.

(16.9)

Since each oscillator of the assembly vibrates with its own frequency, and
we are considering an assembly of 3N linear oscillators, there must be an
upper limit to the frequency spectrum. The maximum frequency

vm

is deter

mined from the fact that there are only 3N phonons:

3N

(16.10)

The Heat Capacity of a Solid

Chap. 16

298

Combining this result with Equation (16.9). we get

(16.11)

Equation (16.10) provides us with some insight into the cutoff frequency and
wavelength. Since

vm (N/V)113 and A min r:x.lfvm, it follows that


.

\:

/\mtn
.

1/3

The minimum possible waveJength is determined by the average interatomic


spacing. Thus the structure of the crystal sets a lower limit to the wavelength;
shorter wavelengths do not lead to new modes of atomic vibration.
The principal difference between Einstein's description and Debye's
model is in the assumption about the frequency spectrum of the lattice vibra
tions. This is shown graphically in Figure 16.4. More rigorous calculations have
led to more complex spectra.
Now, there is no restriction on the number of phonons per energy level
jhv, where j is an integer. Thus phonons .are bosons. *This means that the occu
pation numbers must be given by the Bose-Einstein distribution. For the con
tinuum , Equation (13.41) applies:

N(e)

l
=

g(e)

-----

e(t-p.)/kT-

1.

I I
I I

..

g(v)

...
...
..

,,

3N

g(v)

.
.

. . . ..
. . . . . . :. '.
. . .. .
,
:
:

:
:
.
.
. .. . . .
. .

..,
:

.
.
.
...

.
.
.

.
. . . . 2:
:::. ::.
. L
.

.
. : .

;:

.
.
.
.
-
. : . . .; : . : : . .: :--:

.- . ...
... . - .
==:::l. ;:- . -

..

.
.

.
:'.

.
.

..

.. .

. .

,,

._

(a)

'

:>
:

.
:.

..

. .
:

..

..

: . . .
... .

:
,

>

'

.
.. :

; ; .-:

.'

.
.

::

(b)

Frequency s pectra of crystal vibrations: (a) Einstein model:

(b) Debye model.

Figure 16.4

*Some physicists like to call them "phony bosons."

Jl

Sec. 16.3

Debye's Theory of the Heat Capacity of a Solid

299

In this expression the chemical potential J.L must be set equal to zero. This is
because the total number N of phonons is not an independent variable but

rather is determined by the volume and temperature of the particular crystal


being considered. Specifically,

N is

the number of phonons that causes the

Helmholtz function to be a minimum at equilibrium. Since p.


follows that J.L

0. With e

( aF1aN)r,v, it

hv, we have

N(v)dv

g(v)dv
/kT - ,
h
e
1

(16.12)

"

N( v) dv is the number of phonons with frequencies in the range v to


dv. When we substitute Equation (16.11) in this result, we obtain

where

N(v)dv=

v<vm

- --- --

(16.13)

>Jim

The total energy of the phonons in the frequency range

to

hv N(v)dv. Hence the internal energy of the assembly is


U

dv

is

hvN(v)dv

9Nh
=-JIm 3

v3dv

"m

/kT
hv
e

(16.14)

1.

(We leave out the constant zero-point energy since this terrn has no effect on
the heat capacity. Its value is calculated in Problem

16-2.)

The Debye temperature 80 is defined as

(16.15)
That is, the Debye temperature is proportional to the cutoff frequency
Some values are given in Table 16.1
To obtain the heat capacity, we need to differentiate Equation
with respect to the temperature. Now

d
dT

h
fkT
v
e

hv
-

h,
fkT
e

kT2 ( eh"/kT
.

l )2

"m

(16.14)

300

Chap. 16

The Heat Capacity of a Solid

TABLE 16.1 Debye tem


peratures of some materials.
Substance

80 ( K)

Lead
Mercury
Sodium
Silver
Copper
Iron
Diamond

88
97
172
215
315
453
1860

Thus

Cv
We let

x =

dU
dT

hvjkT and Xm

9Nh2

( 16.16)

-v m- 3"--Tk -:!

hvmfkT

80/T. With the change of variable we

have

(16.17)
.

high temperatures. 00/T << 1 and .t << 1. So e""


1 =
1 in the numerator so the integral becomes
tor and ex
For

.r

in the denomina-

=e-

..,

8o J
T

Hence

Cv=9Nk

T 3 1 Oo J

80

3Nk.

This is the law of Dulong and Petit. For low temperatures 00/T is large and we
can let the upper limit of the integral be infinity. Then

( 16.18)

Problems

301

and
T

6o

(16.19)

T3 law. It is valid when the temperature is


lower than about 0.180, which means for most substances about 10-20 K. The

This equation is known as Debye's

relation gives a better fit to experimental data at very low temperatures than
the Einstein model, and is valid for

all monatomic solids. When the tempera

ture is above the De bye temperature, the heat capacity is very nearly equal to
the classical value 3Nk. For temperatures below the Debye tetnperature,
quantum effects become important and Cv decreases to zero. Note that dia

mond, with a Debye temperature of

1860 K, is a Hquantum solid" at room

temperature.
Recent work has centered on the behavior of solids at low temperatures.

3
Experiments suggest that amorphous materials do not follow the Debye T
law even at temperatures below 0.0180. There is more yet to be learned.

PROBLEMS

16-1 The partition function of an Einstein solid is


Z

e-erJ2T
=

rr'
E

-8

where 8E is the Einstein temperature Treat the crystalline lattice as an assembly


..

of 3N distinguishable oscillators.

(a)
(b)
(c)

Calculate the Helmholtz function F.

Calculate the entropy S..

Show that the entropy approaches zero as the tem per ature goes to absolute
zero. Show that at high temperatures, S 3Nk[1 + ln(T/9E}]. Sketclt
Sf3Nk as a function ofT/BE.

16-l Show that the inclusion o the zero-point energy in Equation (16.14) gives a
term (9/8)Nk60 that is added to the integral. Since the terrn is a constant, it does
not contribute to the heat capacity.

16-3

(a)

Referring to Equation (16.14) and Problem 16-2, show that the internal
energy of the assembly of pbonons can be written

U
where

D(80/T)

Nk00
g

3NkTD

Oo
T

is one of several Debye functions. Its definition is

The Heat Capacity of a Solid

Chap. 16

302

9n
D
T
(b)

---

---

For very low temperatures. the upper limit of the integral can be taken to be
infinity. Show that

1=

--

15

.
rm
te
by
rm
te
g
in
at
gr
te
in
d
an
-.r
e
of
s
er
w
by expanding the integrand in po
n.
ow
kn
l
el
w
is
m
se
su
ho
w
s
rie
se
ite
fin
in
an
The result can be expressed as
Note

(c)

that D( x)

0.

Show that at low temperatures

11"

r3

r 4

16-4 (a) The heat capacity can be expressed in terms of the Debye function

D(90/T) by noting that

x ex

fll, .'T

(ex

J )2

---

tlx

and integrating by parts. Sho\\' that

(d) Calculate and plot' CvI ( 3N k) for 1.1 H0

0.2. 0.4. 0.6 0.8, and J .0. The


..

cor

responding values of the De bye function are:

0.2

0.4

0.6

0.8

1.0

0.117

0.354

0.496

0.608

0.674

16-S The partition function of a Dehye solid is

ln Z

==

-9

,8o

Hp/T
a

x2

Jn( 1

e-x)dx.
-

Show 1hat the Helmholtz function can be written

9Nk7 In( I

8
o1r
e)

3NkTD(00/T).

16-6 The experimental value of c,. for diamond is 2.68 x 1{)3 J kilomole -l K- 1 at a
temperature of 207 K. For diamond the Einstein temperature is 1450 K and the
Debye temperature is 1860 K. Calculate c,. at 207 K using the Einstein and
Debye models and compare the results with the experimental value.

Problems
16-7

(a)

303
The Debye temperature

80 can

be determined from measurements of the

speed of sound in the solid. Use Equation (16.10) to show that


.a

un -

1/3

3N

he

'

where

(b)

Show that
-

00(K) -

2.51

10

-2

1/3

c,

where p is the density in kg m-3, M is the molecular weight, and c is the


sound speed in ms-1 For copper, p = 8.95 x 103m-3, M
63.6 and
=

= 2.24

l<Y ms-1

Calculate

90.

16-8 The equation of state for a monatomic solid is

Pv
where

f(v)

is the specific internal energy of the lattice

is the specific volume and

= fu,

vibrations of the crystal. Tne so-called Gruneisen constant r is important in


solid state theory.
(a) Show that

(b)

Bv
.
CJ<

Show that

y = 1

fBT,

where i' is the ratio of specific heats.

(c)

The Grtineisen constant is defined as

f
Assun1e that

80 in the

d ln

00
dIn v

Debye model is a function of

v only and

how that

this definition leads to the result of (a). Hint: Use Maxweil's relation

as
av

aP
aT

t'

and assume that the specific entropy sis a function of T/80.

.::> :
(\l :;;';;
..-..

::::_:

.;::::::: ::::::::::::::::

ICS

17.1 Introduction

307

17.2 Paramagnetism

308

17.3 Properties of a Spin-1/2 Paramagnet

315

17.4 Adiabatic Demagnetization

318

17.5 Negative Temperature

321

17.6 Ferromagnetism

325

. -.

lOS

.
:-::::::,..:::<.;>.>::::.::.. ::
.

:-

..: .: .:

.;.;.:-::-:::::-:-:-:-:::: ;.;. ;.;. :-:-::. : ..


...::: :
:
-

:-

::

:
- :- :

;
.

: .. :.

. /:: :::
::
: - :-: : . > .
llliiil
iiiiBBii
818i!B

17.1 INTRODUCTION
Magnetism comprises phy sical phenomena involving magnetic fields and their
effects on rrtaterials. All magnetic fields are due to electric charges in motion.
On an atomic scale, individual atoms can give rise to magnetic fields when
their electrons have a net magnetic moment as a result of their orbital and/or
spin angular momentum.
Each electron of an atom contributes a m.agnetic dipole that experiences
a torque when placed in an external magnetic field. The phenomenon of para
magnetism results from the tendency of unpaired dipole moments to align
themselves parallel to the applied field. Paramagnetism in solids occurs in com
pounds containing transition metal ions that have either incomplete d shells
(the iron, palladium and platinum groups) or incomplete f shells (the lan
thanide and actinide groups). Both the net spin and orbital magnetic moments
may contribute to paramagnetism, but for an ion with an incomplete d shell, the
effective orbital moment may be "quenched" by electrostatic interactions with
its neighboring ions, leading to a predominant spin moment. Paramagnetism is
therefore due mainly to the spin angula momenta of the electrons.
In an external magnetic field, all the orbiting electrons of an atom speed
up or slow down, depending on the direction of the field. In either case, a
change in each e]ectron's magnetic moment is produced antiparallel to tl1e
field. The effect is known as diamagnetism. It is present in all atoms, but is usu
ally much weaker than paramagnetism.
In ferromagnetic materials, each atom has a comparatively large dipole
moment caused primarily by uncompensated electron spins. Interatomic
forces produce parallel alignments of the spins over regions containing large
numbers of atoms. These regions, or domains, have a variety of shapes and
sizes (with dimensiotts ranging from a micron to several centimeters), depend
ing on the material sample and its magnetic history. The domain moments are
generally randomly oriented, so the material as a whole has no magnetic
moment. Upon application of an external field, however, those domains with
moments in the direction of the applied field increase their size at the expense
of their neighbors, and the internal field becomes much larger than the external

307

The Thermodynamics of Magnetism

Chap. 17

308

field alone. When the external field is removed .. a random domain alignment
does not usually occur, and a residual dipole field remains in the sample. This
effect is called hysteresis. The only elements that are ferromagnetic at room
temperature are iron.. nickel, and cobalt. At sufficiently high temperatures.. a
ferromagnet becomes paramagnetic.
The modern quantum mechanical theory of magnetism was to a large
extent developed by J. H. Van V leck. One C)f the trium phs of this theory is the
remarkable agreement between the observed and theoretically calculated
magnetic moments of paramagnetic salts. "These are the magnetic systems of
primary interest in statistical thermodynamics.

17.2 PARAMAGNETISM

Most experiments on magnetic materials are performed at constant pressure


and involve insignificant volume changes. Thus.. pressure and volume varia

tions can be ignored, and equilibrium states can be represented by a relation


involving three thermodynamic variables: the magnetic field B, the magneti
zation M, and the temperature T. The methods of statistical thermodynamics
are appropriate since the alignment of large numbers of atomic magnets

is a

statistical process.

Paramagnetic effects are produced in a crystal when some of its atoms


have a net magnetic dipole moment associated ith the electron orbital angu
lar momentum, the electron spin .. or both. To understand the relationship
between the magnetic moment and the angular momentum. it is instructive to
consider first the hydrogen atom and then give the result for more complex
atoms or ions. The

basic ideas are the same for other systems as for hydrogen.

In the quantum-mechanical description, the magnitude of the orbital

angular momentum vector of an isolated hydrogen atom is V /(/ + 1 )1i,


where the allowed values of the quantum number I are 0. 1 2 . . The z com..
ponent of the orbital angular momentum is n111i where -I < m1 < I. It fol
..

lows that there are 2/ + I values of

m1 for a given value of /. The magnetic

dipole moment p, is proportional to the angular momentum: the constant of


proportionality is ef2m, where e is the charge of the electron and m( is its

mass. Thus

ett

and

<

m1 <

l.

Sec. 17.2

Paramagnetism

309

e1if2me is the so-called Bohr magnton and g is a number of the


Here J.Ls
order of unity referred to as the Lande g factor; g is difficult to deterrttine
=

exactly but is known to be

1 for orbital ang ular momentum.

The analogous expressi ons for the electron's intrinsic angular momen

tum or spin are

IPI

s(s + 1),

=KILn

-s < ms <

s'

2 for spin. For an electron the spin quantum number s


is 1/2, so that ms can assume only two values, -1/2 and 1/2. (Both m1 and ms
where g is very close to

vary i n integral steps.)


The magnitude of the total angular momentum , the vector sum of the
orbital and spin parts, is

ll

Vj(j + 1)1i,

where the possible values of j are


s J. The z component of angul ar momentum has

f fi - sf + 1, ... ll +
the value mili, where the possible values of mi are
-

s ,

j, j

1, ... j

1, j.

These relations are restricted to one orbital electron. For a many-electron


atom, similar expressions apply, with capital letters instead of small letters to
indicate that they pertain to the tot al number of particles. The generalization is

jp.

(17.1)

KILB v'i( J + 1),


-J < m < J,

(17.2)

Sl, IL St + 1, . . IL + Sl and the


. . J- 1, J. Here LandS are the total

where the possible values of J are fL

allowed values of mare -1,-J + 1,


orbital and total spin quantun1 numbers.* For simplicity, the so-called mag
netic quantum number m is written without a subscript. Since the total angular
.

momentum includes spin,

J can

be integral or half-integral, depending on

whether the atom has an even or an odd number of electrons, respectively. To


a good approximation, the Lande g factor is given by

J(J

1)

S(S + 1)

L(L

*For paramagnetic ions in solids it is often convenient to

+ 1)

use

(17.3)

an effective spin

S' to

describe the magnetic behavior of the lowest group of energy levels. In this case we may write

fl'l

'
g

p.s

S'(S'

1),

w here g', the effective value of g, is close to 2 for many ions with incomplete d shells. However, we
shall continue to use S and g rather than S'

and g'.

The Thermodynamics of Magnetism

Chap. 17

310
l

p.

H
H

Figure 17.1 Magnetic moment


p, in a magnetic field B.

A dipole with magnetic moment p. placed in an external magnetic field


by
en
giv
N
e
qu
tor
a
ce
i
en
per
e
x
l
wil
1)
17.
re
gu
(Fi
ion
ect
dir
z
the
in
ing
int
po
B
the vector prod uct

p, X

( 17.4)

B.

The magnetic p otential energy when the dipole is in the angular posi t ion 8
(the angle b e tween B and p,) is the work that must be done to rotate the
dipole from its zero energy p os i t ion fJ

TT/2 t o the given posi tion 0:

NdO

sin Bd8

J.LB

rr/2

TT

B cos e

( 17.5)
By substituting Equation ( 17.2) in this equation, we get

( 17.6)
the magnetic energy of the atom in the state designated by the quantum num
be r

nz. *

As an example, for an electron with a single net electron spin there

would be only t wo poss i bl e energies.. wit h


n1

-1/2 (spin antipara1lel to

n1 =

1/2 (spin parallel to

B). Note that the

B) and

energy is lowest when the

dipole moment is parallel to the field.


In a paramagnetic material. because the atoms occupy definite sites in
the crystal lattice they can be regarded as distinguishable particles and there
fore obey Boltzmann statistics. Accordingly, the probability

in an energy state

e,,

P,11

that an atom is

is equal to the fractional occupation number correspond

ing to that energy level (Equation (13.22)). Since the degeneracy is removed
*

\Ve used the

classical argument

to derive

as the result obtained from quantum theory.

E quat ion ( 17.) )

hut Equation

( 17.6) is the

same

Sec. I 7.2

Paramagnetism

311

by the applied magnetic field, the degeneracy factors are unity


for all possible
states. Thus

'

(17.7)

where

Z-

(17.8)

m=-1

It follows that the mean z component of the magnetic moment of the atom is

(17.9)
with
I

T
Jl"'11Bm/k
e
Z=
.
m=-J

(17.10)

We can write the expression for J.Lz more compactly as follows. Differentiating
Equation (17.10) with respect to B, we obtain
,

az
aB T

kT m=--J

gJLsmeg8Bm/kT.

Then

kT dZ
P-

dB T

kT

d In Z
iJB

(17.11)

The next step is to evaluate the partition function Z. We introduce the


abbreviation
KJLsB

kT '
and write
J

Z=

eflm.

m=- J

(17.12)

312

The Thermodynamics of Magnetism

Chap. 17

Sett ing

e", we have
J

Xm

X J

--

X J+ 1

X1 -I

:c1

m=--1

The terms in parentheses constitute a finite geometric series whose sum is


1 -

x2J+'

1-

Hence

e-JTJ

1 - e(:!J + 1 >TJ
1

efl

----

e11

-----

If we multiply both the numerator and the denominator by e-.,;2 we obtain


..

the following result:

Z=

<1 + 1/2)71
e

''
1;2)'1
e
I"

12

e -TJ.

e"

sinh ( J +

! )11

sinh YJ/2

4.

(17.13)

Fmally,. using Equation (17.11).. we have


p,l.

kT

a In Z
aB

kT

(J + ! )cosh(J
sinh(J +

iJ

11
aB
+

a In Z
d1)
1

)11
-

!)17

a In Z

cosh(71/2)
----

sinh(71/2)

Therefore,

(17.14)
where

1
1
J +- coth J -#
2
2

11

- coth -1J

is called the Brillouin function. Its variation with 71 is sketched in

(17.15)
Figure 17.2.

Sec. 17.2

Paramagnetism

313
J

1.0 1--

712
J
j

Figure 17.2 The Brillouin


function for several values of J.

112

----'..._
__
_
_
..._
_
---.
....
.
..

Let us examine the behavior of 81 ( 11) for large and small values of its
argument. For 1J >> 1, the hyperbolic cotangent approaches unity and

J+

1
2

1
-

1,

, >> 1,

(17.16)

corresponding to magnetic saturation. For small values of y,

l11y + y 3; hence
BJ ( , )

Sd

1
J+2

1
J

1
(J+ !)11

-1
1 1
1
1
+- J+- TJTJ
2
3
I 1)

1
-

31

71

+
2 TJ 6

11

11

1
1+-,

12

]2 + J + - - -

J+l
3

,,

, << 1.

(17.17)

From Equations (17.12), (17.14), (17.16), and (17.17), we obtain


KJLBB >>

(17.18)

kT,

and

'

KILsB <<

k T.

(17.19)

The Thermodynamics of Magnetism

Chap. 17

314

Of particular interest is the second region, coresonding to high te

peratures and a weak magnetic field. The magnettzatlon M of a magnetic


material is defined as the total mean dipole moment per unit volume. Thus, for

a material sample of N atoms and volume


N g2J.LB2J (J +
--

1) B

( 17.20)

3VkT

This is the experimentally observed Curie law discovered by Pierre Curie.,


which states that the magnetization of a paramagnetic solid is proportional to
the ratio of the magnetic field to the temperature.
For B

1 T, * T

300 K and g

8JLsB
kT

5
(86
. 2 X 10- eV/K) (3

-----

----

----

102 K)

so the Curie law certainly applies under these conditions. Even at


g

2 and J

K. with

1/2, a field of 1 T (a comparatively strong laboratory magnetic

field) would produce Jess than 25 percent magnetic saturation.


The paramagnetic salts most widely used contain paramagnetic ions sur
r o un de d

by a large number of nonmagnetic particles. A typical example is

Cr2(S04)3

K2S04

24 H20 (chromium p()tassium alum). Its magnetic prop-

erties are due solely to the chromium atoms that exist in the crystal as ions. The
chromium ion cr+

t +

has three unpaired electron spins and therefore a mag

netic moment of 3 J.tB Besides the two chrotnium ions there are 4 sulfur atoms,
2 potassium atoms, 40 oxygen atoms, and 48 hydrogen atoms. Hence there are a
total of 94 particles that are nonmagnetic. TI1e magnetic ions are so widely sep
arated in the molecule that the interaction between them is negligibly small. At
the same time. the effect of the orbital motions of the valence electrons is
quenched by the fields of neighboring ions. What remains is a net electron spin.
The configuration of the chromium atom jCr is

5
( Ar) ( 4s) 1 ( 3d) Argon

has 18 electrons in closed shells, so chromium has an additional6 electrons

one in the 4s subshell and 5 in the


4s electron and two of the

3d

3d subshell. In the chromium ion Cr-t-+


3d electrons that

produce pragnetism.

Since the total orbital angular momentum is essentially zero J


(1/2 for each of the unpaired electrons). With g
z

the

electrons are misS'iag. Thus the structure of

Cr+++ is (Ar) (3d)3. It is the three

in the

==

3/2

2 the mean dipole moment

direction is

..
:

( 17.21)

1 T

*The SI unit

kg/ A

of

the magnetic

s2. Also. 1 T

field 8

104 gauss.

is the tesJa

(T):

in terrns

of

fundamental unit

Sec. 17.3

Properties of a Spin- I /2 Paramagnet

Figure 17.3

The mean

magnettc moment

tn

l15

Uillts of

the Bohr magneton for t he

chromium ion cr+++ in


chromium potassium alum.
The s olid curve is the

theoretical result. The


experimental values are those

reported by W. E. Henry, Phys.


Rev. 88,561 (1952).
In Figure 17.3

JLz/#Ls is plotted against BjT for chromium potassium alum.

Note the excellent agreement between the theoretical result and the experi
mental data.

17.3

PROPERTIES OFA SPIN-1/2 PARAMAGNET

The simplst paramagnetic system is an atom or ion with a spin-1/2 ground state
and zero orbital angular momentum. Then J = 1/2 g = 2, and there are just two
energy levels. The thermodynamic properties of a two-level system exhibit the
,

same features as those of more complex systems and are easier to calculate.
For this case, the partition function of Equation (17.13) takes the fortn
Z=

sinh 11

11 =

--

TJ
.h
SID

'

2p,BB
kT

(17.22)

This can be simplified by using the identity


.

sinh 11 = 2 sinh

1J

'T1

cosh

in the numerator and canceling one of the factors with the denominator, giving
Z = 2 cosh

where

11

==

2 cosh

kT ,

( 1 7 . 23 )

J.L8B. The tota l magnetic energy U8 can then be found from


a In Z
aT

(17.2 4)

The Thermodynamics of Magnetism

Chap. 17

316

This is Equation

calculated for the Maxwell-Boltzmann distribution

(14.7),

and also valid here. (Boltzmann statistics and Maxwell-Boltzmann statistics


differ by a constant factor, but the distributions are the same.) Then
iJ In Z

iJT

kT

sinh(ker)

---

cosh ()"
0

so that

U8

Since tanh(O)

Ne tanh

0 and tanh( x )

( 17.25)

k.r .

1. the energy is evidently

Ne at T

and approaches zero asymptotically at high temperatures. This behavior,


shown in Figure 17.4, is markedly different from that of the corresponding
energy of an assembly of harmonic oscillators treated in Section

15.2. The lat

ter increases without limit as the temperature increases. Here higher tempera

tures produce increased randomization <>f the dipole moments and the
magnetic potential energy goes to zero.
The magnetic contribution to the heat capacity also has distinctive char
acteristics, given by

au8
iJT

H.N

t:

Nk

--

kT

.,

sech-

(17.26)

kT .

This expression can be put in an alternative form from which the limiting
behavior is more readily seen (Problem

17 -X):

e2tikT +

kT

(17.27)

()r--

---

-0.5

t--

- 1.0
Fige 17.4
of a spin-1/2

Internal energy
paramagnet.

()

'

kT

Sec. 17.3

Properties of a Spin- I /2 Paramagnet

317

Then it is clear that


c

2e
=:;

kT/B

kT

<< 1,

and the exponential factor reduces C8 rapidly to zero as the temperature


decreases, in accordance v;ith the third law. At the other limit,

kT/e
For

efkT in

known as a

>> 1.

the v icinity of unity, the heat capacity has a fairly sharp peak

Schottky anomaly

(Figure 17.5). The anomaly is useful for deter

mining energy level splittings of ions in rare-earth and transition-group met


als. Like the energy, the heat capaci ty is qualitatively different from other
systems in its variation with temperature. The most important thertttodynamic
property of two-level systems is the entropy. For Boltzmann statistics, .

( 17.28)
Substituting Equations (17.23) and (17.25) in this expression, we get

Nk

In 2 cosh

Figure 17.6 is a plot of S/Nk versus

8
-

kT

e
kT

tanh

e
kT

( 17.29)

kTje.

At low temperatures,

Cs
Nk

H eat capacity of
_
a spin-1/2 paramagnet.

Figtre 17.5

kT
e

The Thermodynamics of Magnetism

Chap. 17

318

s
.'Vk

0.6

t--

0.4

t-

0.2

I--

0
Figure 17.6

Entropy of a
spin -1/2 paramagnet.

f
"'--__.___....___.._____._

kT
E

and
e
- tanh
kT
kT
so s

e
I

-----

kT

..

().

high temperatures. the se con d term in curly brackets approaches


zero, cosh ( ej k T) 1 and S Nk In 2. Th is is exactly what we would expect.
At the upper temperature limit 1v = 2/v the number of equally probable
.N k In 2. This corresponds to a pattern of ran
microstates, and S = k In 2i,.
dom dipole orientations\ involving equal numbers of parallel and antiparallel
magnets in any chosen direct ion In this disordered state the entropy is a max
imum. As T 0 all the di po les are in the lower energy state pointing in a
direction parallel to the applied magnetic field. There is only one possible
n1icrostate so the thermodynamic probability 10 is 1 and S k In 1
0.
The foregoing discussion is significant because many systems can be
treated as two-level svstems.
At

..

..

"'
...
....;.(
..
.
. ..
*
.
.
<!...
.,

17.4 ADIABATIC DEMAGNETIZATION

Paramagneic salts have been used to a tt a in very low temperatures with a


process known as adiabatic lielnagJzetizatilJil, alluded to in Chapter 10. The
method rests on the fact that the entropy is a monotonically increasing func
tion of kT/c that is. of Tl B.
When the magnetic field is increased the degree of alignment of the
magnetic moments increases and the disorder of the spin system decreases,
thereby lowering the entropy. The entropy is also lowered if the temperature
decreases again because the moments t.end to line up. But if the field is

Sec. 17.4

Fi

Adiabatic Demagnet!zation

319

17.7

Entropy versus
temperature for two values of
the magnetic field B1 and 82,
wjth > B1 The spin systeJn

goes from state a to state b at

I
I
I
I
I
I

a constant temperature 1j as
the magnetic field is increased.
When the magnetic field is
reduced to its original value
along the adiabatic path from

b to c, theemperature is

lb
I
I
t
I
I
T

reduced to its final value T1.

reduced adiabatically, without changing the entropy, the temperature must


drop if the same degree of disorder is maintained.

This method is illustrated in Figure 17.7. A paramagnetic salt is first


cooled to a temperature of 1 K or less by contact with liquid helium. A mag

netic field acting on the salt is then increased from B1 to B2 under isothermal
conditions at temperature T;. The spins are in a lower energy state so heat is
evolved and the entropy is lowered. At this point the salt is thermally isolated
so that no heat is exchanged with the surroundings and the field is gradually
reduced from B2 to B1. Since the entropy is constant,

At very low temperatures the heat capacity of the spin system is much larger
than that of the crystal lattice, so the final temperature of the paramagnetic

salt is only slightly greater than T1. Temperatures of less than 1 mK have been

achieved with this technique. The reason a final temperature of absolute zero
cannot be reached is discussed in Chapter 10.

Demagnetization cooling can be understood by combining results from


classical thermodynamics with the statistically derived Curie law. The work
done by the paramagnetic salt when an applied magnetic field changes by an
amount dB is 7/W

dV

MdB where M is the total magnetization (

0, the Helmholtz function takes the form

dF

-SdT- MdB.

If F is considered a function of T and B,

dF

iJF
=

ar

dT +

OF

aB

dB.
r

MV). With

310

Chap. I 7

Since F is a state

The Thermodynamics of Magnetism

variable dF is a perfect differentiaL so that

From this we immediately obtain the Maxwell relation

as

()8

riT

(17.30)

The magnetization Misgiven approximately by the Curie law

( 17.31)
where. a is a pos itiv e constant. Taking the derivative with respect to T while
keeping

B constant, we have
riS

---

aB
--

( 17.32)

--

'

Hence the entropy always decreases with an isothermal increase in the mag

netic fie l d

Figure 17.7, we

For the adiabatic Jeg in

can use the fundamental relation

dU
Here it is convenient to letS

T dS
or, for

TtiS- MdB.

S(T, B). Then

(15
iJ7'

dT

CIS

aB

dB.

reversible process,

T ds

cRdT +

where C8 is the hea t

Setting dS

riS

. iJB

dB,
T

capacity of the system in a constant magnetic


0 and using Equation ( 17.30). \Ve find that
dT
T

field.

(17.33)

Sec. 17.5

Negative Temperature

321

Because thermal motions tend to disalign the tnagnetic dipoles of the para
magnetic material, the magnetization decreases with increasi11g temperature.
Therefore

(oM/oT) 8 is negative and consequently the temperature drops

as

the magnetic field is lowered, according to Equation (17.33). This is known as


the

magnetocaloric effect.

17.5 NEGATIVETEMPERATURE
The two-level case can be used to discuss an extension of the notion of tem
perature. As noted, the energy of the lower lvel (p, paralll to B) is
B
p.
==:'
e1
ts
to
B)
le
ral
B
tpa
ant
el
(p,
lev
-p.8 whereas that of the upper
.
80
n
tto
tta
exc
the
,
2e
ts
B
els
lev
of
n
tio
ara
sep
the
t
tha
so
e,
for
be
as
p.8
e
We set
gap. The number of particles in the lower level e is N0 and the number in the

upper level e is N1, with


N0 + N1

(17.34)

N,

and

IA.YTO

N
-

T
/k
e
e '

N1

N
=

/kT
-:
.
e

(17.35)

The total magnetic energy of the system is


(17.36)
The ratio N1/ N0 is
N
,
No

-2e/kT
e
,

which leads to the following expression for the temperature:


T

l
=

2e

k In N0- In N1

(17.37)

In a state of stable equilibrium the occupation number N0 of the lower energy


level is greater than the occupation number N1 of the higher energy level, so
the absolute temperature N1 is positive in this equation.
Imagine now that the direction of the applied magnetic field is suddenly
reversed. The elementary magnets that were parallel to the original field and

Chap. 17

322

The Thermodynamics of Magnetism

in the lower energy state are now antiparallel to the field and in the higher
energy state. Conversely, the magnetic moments originally in the higher

are now in the lower energy state. A popttlation inversion has


taken place. In Equation (17.37) the sign changes and the temperature T
energy state

becomes negative!

To be sure, after some relaxation time has passed, the moments in the
higher energy state will flop over into the new low-energy state and a new
state of equilibrium will be established. But immediately after the field is
reversed, the temperature of the spin systen1 will be negative.

The situation can be made clearer by examin i ng how the entropy

ad

temperature vary with the total energy. The entropy of the two-level system is
S k ln w, where
=

'U.

Using Equat i ons

N!
=

,"'t,! Nl!

(17.38)

(17.34) and (17.36), we can write


)

N
2

:=;

NE

N,

Substituting t hes e relations in Equation


1)

1 -

1Ve

l + Ne .

(17.39)

(17.40)

( 17.3R), we have
N!

!\'

--

""

1+

1Vc

'

The use of St irli n g s approximation leads to an e x pr e ssion for the entropy in


terms of the dimensionless q ua n tity ,t
'

U/.'Vc, (Problem

17-10):

( 1 7 .4 1 )
E videntl y x varies from

-1 to

1: in this intl!rval

0 (Figure

S/Nk is an even function of x

17.8).
Referring to Equation ( 17.39). we see that when U / N E = - 1, l
N,
a nd N1
0. all of the particles occupy the lower energy level. Similarly. when
1, >
N, the upper level is fully occupied. In both
U IN e
0, and N1
with a maximum at x

==

==

cases T

0.

==

When the particles are equally distrihuted between the two levels, the

entropy reaches its maximum v a lue of Nk In 2. Here the temperature sud

denly j umps from +x to -as the direction of the field is reversed.

Sec. 17.5

Negative Temperature

In 2

s
Nk

----------

323

I
J
I
I
J
I
I
I
I
I
I
r
I
I
I
I
I
I

0.5

T>O

T<O

-----------

-1

-0.5

x =

Figure 17.8

Entropy

as a

0.5

u
iVe

function of energy for a two-level system.

To obtain a more detailed picture of the relationship between the tem


perature and

the energy, we can calculate


procity fortttula (see Chapter 8):
T=

T as a function of x by

usin

g a reci

au

as

Then
as

au

as

Ne

ax

2e

In

1-

1 +X

(17.42)

UjNe. Clearly shown is the abrupt


In Figure 17.9, kT/2s is plott ed versus x
shift from +oo to oo when slightly more particles occupy the upper energy level.
=

Three conditions must be satisfied if a system


perature:

(1)

is to have a negative tem

the system must be in thermal equilibrium, (2) the energy spec

trum of the system must have a finite upper bound, and (3) the system

must be

energetically isolated from states that are at a positive temperature. The sec
rule

s out systems such as


upper limit to its possible energy.

ond condition

a harmonic oscillator, which has no

The Thermodynamics of Magnetism

Chap. 17

324
3

---------- --------

I
I
'

'

2,__

I
I
l

I
'

kT

-----

----

-1

--

t
1

-1 ....

---- --

- -------

{)

I
J

-2

.
I
I
I
I

.,._

------
x =

Figure 17.9

u
Ne

Temperature as a function of energy for a two Jcvel


..

system. Note the jump in temperature from


particles are in the higher energy state.

+x

to

-x

when more

In paramagnetic materials the interaction between the magnetic ions


and the crystal lattice, though weak, is sufficiently large that the substance can
not exist in a state of population inversion for any appreciable length of time.
However, in a series of elegant experiments. Purcell and Pound in 1951 used
nuclear magnetic resonance techniques to show that a ngative temperature
could exist in a nuclear spin subsystem. In

lithium fluoride crystal, nuclear

spins were aligned by a strong magnetic field. At a positive temperature

lithium nuclei absorbed energy supplied by a radio-frequency oscillator with

angular frequency equal to the excitation frequency Zejn. When the magnetic
field was reversed, the investigators detected resonant emission of electro
magnetic energy instead of resonant absorption. indicating a population inver
sion and a resultant negative temperature. The relaxation time necessary to
reestablish thermal equilibrium between the spins and the crystal lattice was
2-5 minutes, long enogh to demonstrate the existence of the inverted popula

tion. In the experiments only the spin subsystem was affected by the field
.
reversal; the whole crystal remaind at laboratory temperature
In radio astronomy.. negative temperature systems have been used to
amplify very weak radio-frequency signals. Also. the action of a laser depends
on a population inversion, which is sustained by some external influence. In
the case of a ruby laser, the method is optical pumping using light radiation .

Sec. 17.6

Ferromagnetism

315

::: :::::.::
::
::::;:;:;:;:;:::::::::::::::::: :;:::: ::=: .::.::::::::: :::;:::: ::::::::.:::=:: :::: ::=:.::::-: =::::
:: :;:: ::::::::::::::: ;:;:::::.; ;:;::::::::: ::.:: : :;::::.::::;:::::;:::;:::::: =:;:: ::::::::::::::::::: :::::;:;:::::>:::;::::
.: :;.;
: :::::::.:::::.::: :::::::::::::;:: ;::.:::

17.6 FERROMAGNETISM

Pierre Weiss was the first to understand that underlying the domain structure
of ferromagnetism is its fundamental atomic nature. The atoms of a ferromag
net have a net magnetic moment and these couple together to forttl the
domains. Much of the previous discussion of this chapter is applicable to ferro
magnetism, with one important modification.

A ferromagnet has a magnetic moment even when it is not in an external

field, a property called spontaneous magnetization. This observation led Weiss


to postulate the existence of an internal field proportional to the magnetiza
tion, and to replace B with the sum of the external and internal fields:

B-+B

+AM,

(17.43)

where A is an undetermined constant. From Equation

(17.14)

we know that

the magnetization is

NJLz

For the case J

1/2 and g

(17.44)

2, this becomes

2J.L8B
kT '

TJ =

(17.45)

where the Brillouin function is

(17.46)
This can be simplified at once by using the identities
coth 71

cosh,.,

11

cosh 11 +

11

cosh 11- 1

sinh 11'

Then

and

M=

Np,B
V

tanh

tsB
kT

(17.47)

The Thermodynamics of Magnetism

Chap. 17

326

With the replacement indicated in Equation ( 17.43 ), we find that

(17.48)

From this equation we can Jearn a great deal about ferromagnetic behavior.
To investigate spontaneous magnetization we set B equal to zero in
Equation(17.48) and obtain a transcendental equation jor M:
..

N J.LB
=

tanh

AJL8M

(17.49)

---

kT

If we set
.

VkT
--

(17.50)

Equation (17.49) becomes

g = tanh

g"
(II

( 17.51)

Fi gu re 17.10 is a graphical' solution of this equation. Nonzero solutions evid en tly exist for

0 in the range 0

<:

0 < 1. The c orrespondi n g temperature

range is 0 < T < -where

( 1.7 52)
..

This critical temperature is cal1ed the Curie tt:mperature or Curie point.

1.0

t.

r--

'
/

/
'

Figure 17.10

Graphical
solution of Equation ( 17.51 )
The dashed curve is the Jeft

hand side, a straight curve


\\'ith unit slope. The solid

curves represent the right


hand side for the indicated
values of 0 Non-zero

intersections occur onlv for

go < I.

O.H

().6

t--

tanh ? 0
()
)
fo
(

0.
0.2 .....,._.
o

s;...__
.,_
....__
......-_...___
...
...___

0.5

J.O

1.5

0.5
1.0
2.0

Sec. 17.6

Ferromagnetism

At T

317

0 the magnetization takes on its saturation value

Np,8fV. As T

increases, Mdecreases, until at the Curie point an abrupt change in the mater
ial properties takes place. This can be seen b}? solving for the temperature in
Equation

(17.49):
(17.53)

Assuming that MV/NILB is small, we can approxim ate the inverse hyperbolic
function using
tanh-1x !!! x +

x3

X<<

3'

1,

and write

(17.54)
With a little algebra this becomes
-NP,a

M=

We note that as T

Tc,

(17.55)

the spontaneous magnetization vanishes abruptly:

M 0 with infinite slope. For T

-+

T,, the ferromagnetic material becomes

paramagnetic. This is an example of a second-order phase transition. A plot


of the reduced magnetization as

function of T/Tc is shown in Figure

17.11,

along with experimental values for nickel.


We have seen ho\v the spontaneous magnetization varies with tempera

ture in the absence of an applied field. In the paramagnetic region, T >

1;;, we

would like to know how M varies with T in an external field B. At tempera

tures above the Curie point, the magnetization is small compared with its sat
uration value. Therefore the hyperbolic tangent is small in Equation

(17.48)

and can be set equal to its argument. The resultant approximate expression for
the magnetization is
M

N'
,-s2B

Vk(T

(17.56)

Tt)'

*A second-order phase transitio!l occurs when

system passes through a critical region

corresponding to incipient instability. Here the Curie temperature is the ferromagnetic critical
point and the phase transition consists of an order-disorder transition of the electron pairs.

The Thermodynamics of Magnetism

Chap. 17

318

1.0

M\l
vJ.J.H

Figure 17.11

Magnetization
of nickel as a function of
temperature. The

-+-r----r-

0.8

1----

0.6

....,___ -+---+----t-\o--1

0.4

.,__- -'--+---.---t-

0.2

1-----

experimental values were


r eported by P. Weiss and R.
Forrer The solid curve is the
1/2.
th eoretical result for J

0 '-0

-+---+---t---t1
-'--

__._

0.2

_
_
_
.
_
__.

0.4

0.6

0.8

1.0

1/T(

This relation is the Weiss-Curie law\ which is a modification of Curie s law dis
cussed in Section
Equation

17 .2.
(17 56 ) suggests a method for
.

determining the Curie tempra-

ture. For a given material measurements are made of the magnetic susceptibil
ity defined as X

( B/JLoM

1)

For temperatures well above the Curie

point, a plot of experimental values of

+ .t -I versus T gives a straight line


.

whose intercept with the temperature axis is proportional to J: Except for


mi nor departures at very high temperatures linear behavior is observed and
.

extrapolation is straightforward. For many .paramagnetic substances 7;_. is less


than a kelvin so that Curie s law gives an adequate representation at all but the
lowest temperatures.
A more extensive treatment of ferromagnetism leads to refinements of
these results and attempts to incorporate a detailed description of domains
and hysteresis in the physical model.
I

PROBLEMS
17-1 Show that the Bohr magncton is given by JJ.H =
in a hydrogen atom moves

en/211lt'' A ssume that the electron


in a circular orbit of radius a about the proton and

that its angu l a r momentum is 1i. The magnetic moment J.L is the product of the
.

electron current and the area swept out by the orbiting electron.

172

(a)

For the hydrogen atom of Problem 17-1 show that the orbital magnetic

:
dipole moment is ea w/2, \\'here w is the electrons angular velocity.

(b) Show that the torque produced by a n1agne tic field para11el to the plane of

;,
the orbit is ea wB 12.
'

(c) By equating th e Coulomb force and the mass times the cent ripet al acceleration, show that

w=

Problems

(d)

329

Find values for the angular velocity, torque, and orbital magnetic moment
11
m; let B
10X
5.29
=
a
1 T.
for a hydrogen atom, where
=

172

Suppose that the hydrogen atom of Problems 17-1 and 17-2 is subjected to a
magnetic field B perpendicular to the plane of the orbit . It can be shown that

the forces caused by


by ap,

B result in a decrease in the orbital magnetic moment given

e2a2Bf4mt. For the relatively large magnetic field of 1 T, what is

llJ.L/JL

in parts per million? (Titis is the mechanism responsible for diamagnetism .)

173

The paramagnetic salt iron ammonium alum has the magnetic ion Fe+++. The
spin system has S

(L

O).Thus J

5/2

and the orbital angular momentum is quenched

5/2andg

2.Findthe mean dipole moment zof the para

magnetic salt in a magnetic field of 1 tesla at a temperature of 2 K. What is the


saturation value of liz?

17-S

In a paramagnetic solid containing N particles, the total magnetic moment

JLr.TOTAL is 'NP-z The Curie law is often written in the form


TOTAL_

ILl

CcB
T '

where C, is the so-called Curie constant, given by

Consider a kilomole of a paramagnetic material made up of molecules whose


electron cloud has zero orbital angular momentum, spin 1/2, and total angular
momentUIIl J = 3/2. Compute Cc.

17-6

A sa11tple of paramagnetic material containing 1()25 atoms is placed in an exter

nal magnetic field of 1 tesla at room temperature. Assume that the electrons of

each atom have zero orbital angular momentum and spin angular momentum of
1/2. Thus J = 1/2 and g = 2.

(a)
(b)
177

Find the total magnetic moment of the sample.

What would be the total magnetic moment at a temperature of 0.1 K?

Show that the magnetic moment

able by proving that dM is

17-8

an

M of

a paramagnetic material is a state vari

exact differential. (Use the Curie law

MB/T.)

Consider a two-level system with an energy 2e separating the upper and lower

states. Assutne that the energy splitting is the result of an external magnetic field

B. Given that the total magnetic energy is


Us= -Ne. tanh

kT

show that the associated heat capacity is

2s

thereby verifying Equation

(17.27).

eafkT

Chap. 17

330

The Thermodynamics of Magnetism

17-9 The energy levels of a localized particle are 0, e and 2e. The middle level is dou
..

bly degenerate and the other Jevels are nondegenerate.

(a) Write and simplify the partition functi<,n.


(b) Find the total ene rgy the heat capacity and the entropy of a system of these
,

particles. Sketch these properties as a function of the temperature and com


pare them with the corresponding properties of the spin-1/2 system.
17-10 Prove Equation (17.41) starting with Equation (17.40) and using Stirling's
approximation. Show that the entropy is

a 1n ax i mum

when the total energy U is

zero, corresponding to equal populations of the two energy levels of the system.
17-11 Verify Equation (17.42).. referring to Equation (17.41 ).
17-12

(a) For iron find the saturation value of the magnetization. (The atomic weight
of iron is 55.9 and its density is 7880 kg m- '.)
1042 K).
(b) Estimate the magnitude of the internal field AM for iron (T,
=

17-13 What is the magnetization of iron in an external magnetic field of 1 tesla at a


temperature twice the Curie temperature?

18.1 Blackbody Radiation

333

18.2 Properties of a Photon Gas

338

183 Bose-Einstein Condensation

340

I 8.4 Properties of a Boson Gas

345

18.5 Application to Liquid Helium

347

331

Statistical thermodynamics is applicable to radiant energy as well as material


particles. It is a familiar observation that a hot body loses heat by radiation.
The energy loss is attributable to the emission of electromagnetic waves from
the body. r"Ibe distribution of the energy flux over the wavelength spectrum does
not depend on the nature of the body but does depend on its temperature.
Here we are concerned with the therxttodynamic properties of electro
magnetic radiation in therntal equilibrium. The radiation can be regarded as a
photon gas. We consider an enclosure or cavity of volume Vat a constant tem
perature T. The walls of the cavity are thermally ittsulated and perfectly reflect
ing. Since the system is isolated, it has a fixed energy U. However, the photons
emitted by one energy level may be absorbed at another, so the number of
photons is not constant. This means that the restriction

Ni =

N does not

apply. Correspondingly, the Lagrange multiplier a that was determined by this


condition is zero and e-a = 1.
Photons are bosons of spin 1 and hence obey Bose-Einstein statistics.
The number of photons per quantum state is therefore given by Equation
(13.40) with p. set equal to zero (recall that
N.I

a =

p.fkT):

For a continuous spectrum of energies,

(18.1)

The energy of a photon is hv, so this equatio can be written


N(v)

(18.2)
333

Chap. 18

334

Bose-Einstein Gases

g(v)dv is the number of quantum states with frequencies in the range v


to v + dv. This number was obtained in Chapter 16 for a phonon gas; however,
Here

it must be doubled in this application because in a photon gas there are two
states of polarization corresponding to the two independent directions of
polarization of an electromagnetic wave. Each photon may be in either polar

( 16.8) becomes

ization state. With this modification Equation

g(v)dv

81rV
1

(18.3)

dv is the number of photons in

v--dv.

where c in this case is the speed of light.

( v)dv in the range v to v


this range times the energy hv of each:
The energy

u(v)dv

N(v)dl.'

hv.

(18.4)

But

N(v)

g(v)j(11).

Hence

( 18.5)
Thi is the Plack radiation formula. It gives the sp
ectral distribution of the
.
radtant energy mstde the encJosure. i.e., the energy pe
r unit frequencv.
The blackbody spectrum is often expressed in terms of
the waelength.
Then

li(v)dv ll(A)dA
x

and

dv

c
-

dA
,
,.\2

or

The wavelength spectrum is therefore

>._5

ehc/AkT

'

(18.6)

Blackbody Radiation

Sec. 18.1

335

u(A)

Fagure 18.1 The wavelength spectrum of blackbody


radiation energy for three temperatures: T1 > T2 > T3

(Figure 18.1). Here u(A) is the energy per unit \Vavelength.


Prior to Planck's analysis, various empirical formulas existed. All of them
can be found from Equations (18.5) or (18.6). The Stefan-BoltZinann law
states that the total radiation energy is proportional to T4 (the area under the
curves of Figure 18.1}.
The total energy density (energy per unit volume) is
(18.7)

Setting x

hcjltkT, we have

The integral has the value 1T4/15 (see Appendix D).Thus


_U
v

aT4'

(18.8)

where

Since radiation inside the cavity is continually absorbed and emitted by


the inner surface, we can relate the ener. gy per unit volume, which moves at
the speed of light

c, to the energy emitted per unit area of the surface per unit

time. The latter is the power per unit area or energy flux. In Chapter 11 we

Chap. 18

336

found the particle flux to be

(vj4)n
=

Bose-Einstein Gases

where vis the mean speed and

is the

number of particles per unit volume. In a similar way, the energy flux

is

( c/4) ( UjV). Therefore,

(18.9)
where

(18.10)

Equation

(18.9)

is known as the Stefan-Boltzmann Jaw, and

is called the

Stefan-Boltzmann constant. The theoretical and experimental values of

agree to within at least three significant figures.


The wavelength

Amax at which u(A) is a maximum satisfies a relation

known as Wien 's displacement law. It can be found by setting the derivative of

u(A) equal to zero or, equivalent1y, by minimizing the denominator in


Equation (18.6):

With .t

hc/AkT. this

reduces to

This is a transcendental equation whose numerical solution is 4.96. Hence

( 18.11)

(18.11) is Wien s displacement law. The experimental values of the


constant on the right-hand side and of the Stefan-Boltzmann constant can be
Equation

used to determine the values of h and k, assu1ning that

is known.

For long wavelengths.l1c/AkT << 1 in Equation (18.6), and the exponen


tial can be approximated by the first two tertns of its Taylor series expansion:

ehc/AkT

===

1 +

he
AkT.

Sec. i 8.1

Blackbody Radiation

337

Then

u(A)dA

;o.::

81rkT
V
dA..
4
A

(18.12)

This is the so-called Rayleigh-Jeans formula, which exhibits an ''ultraviolet


catastrophe" as the wavelength approaches zero (u(A) becomes infinite).
For short wavelengths,
1
>>
hc/AkT
e
and

u(A)dA

z::

87ThC
5

.
-hc/AhT
e
dA

(18.13)

This is Wien's law, valid in the short wavelength regiort (Figure 18.2).

To summarize, the total blackbody radiant energy per unit volume


increases with the fourth power of the temperature, and the wavelength of the
peak of the radiation curve u(A) is inversely proportional to T.

The temperature of the Sun's surface is approximately 6000 K, and Amax


is 483 nm, a wavelength in the visible range of the electromagnetic spectrum.
At the surface temperature of the Earth, roughly 300 K, Amax is about lOJLm,
which is in the infrared region.
The cosmic background microwave radiation that perrneates all of space
was discovered by accident in 1964 by Amo Penzias and Robert W. Wilson
while making measurements of radio signals at a wavelength of 7.35 centime

ters (4 GHz). They observed a nonzero value for the radiation in every direc
tion in which they pointed their radio telescope. In 1990 the Cosmic
Background Explorer (COBE) satellite measured the background radiation

u(A)
I

\4
\

Rayleigh-Jeans

VJien

Figure 18.2

..,..._Planck

Sketch of

Planck's law, Wien's law and

the Rayleigh-Jeans law.

Chap. 18

338

Bose-Einstein Gases

energy

density
500

;...
___.,
..._
__._
_
__.__
____
_
___
_
....
...
__,
_

400

200

600

frequency (GHz)
Figure 18.3 Cosmic background measurements by the
COBE satellite. The smooth curve is the theoretical
blackbody spectrum at 2. 735 K. The energy density is in
eV m-3 per gigahe rtz (Adapted from The Inflationary
Universe by Alan H. Guth, Addison-Wesley, Reading.
.

Massachusetts, 1997.)

at a large number of wavelengths. The data fit the theoretical Planck curve

18.3). The cosmic background radi


ation is evidently a blackbody with a temperature of 2.735 :: 0.06 K.

with almost unbelievable precision (Figure

18.2 PROPERTIES OFA PHOTON GAS

It is instructive to ask: what is the average value of the ratio hv Ik Tin a volume
Vat temperature T? The number of photons having frequencies between
and

dv is found by combining Equations (18.2) and (18.3):


N(v)dv

81rV
==

eh,./kT

(18.14)

-- ---

The total number of photons in the cavity is determined by integrating this


expression over the infinite range of frequencies. The result is

N
where the substitution x

R17.Y

kT
he

,.

x dx
ex

1-

(18.15)

hvjkT has been made. The integral has the numer

ical value 2.404. Hence

2.02

107 T3

V,

(18.16)

Sec. 18.2

Properties of a Photon Gas

339

where Tis in kelvins and Vis in m3 To find the mean energy ofthe photons in the
cavity, we divide the total energy U
7.55 x 10-16 T4 V J (Equation (18.8)) by
=

N. The result is

3.74 X

10-23 T

=:;

2.7 kT.

Thus, the average value of h vjk T is ofthe order of unity.

The heat capacity is easily calculated using Equation (18.8):

v-

au

ar

v.

(18.17)

Since this relation holds down to a temperature of absolute zero, it can be used
to determine the absolute entropy:

S=

kT
0

he

45

V.

(18.18)

Thus both the heat capacity and the entropy increase with the third power of
the temperature.
'

For an open system, we know that

dU

TdS - PdV

JLdN.

Substituting the differential of the Helmholtz function F

TS in this

equation gives

dF

-SdT- PdV + p,dN.

(18.19)

Therefore,

J.L=

aF

aN

T.V

(18.20)

For the photon gas,

(18.21)
Thus, as we discussed in the previous section, the chemical potential JL is zero,
because F does not depend explicitly on N.

Bose-Einstein Gases

Chap. 18

340

Finally Equation (18.19) also yields


=
P

aF
aV

(18.22)

r.N

The pressure of the photon gas on the walls of the cavity is therefore

(18.23)

Note that this differs from the relation P

=:=

2U/3 V that obtains

for an ideal

gas of weakly interacting particles (see Chal>ter 11). Equation (18.23) can also

be obtained from electromagnetic theory if it is assumed that the radiant


energy in the enclosure is isotropic.

18.3 BOSE-EINSTEIN CONDENSATION

In this section we shall be concerned with a gas of noninteracting particles


(atoms or molecules) of comparatively large mass such that quantum effects
only become important at very low temperatures. The particles are assumed to
comprise an ideal Bose-Einstein gas. The discussion is relevant to 4He_ which
undergoes a remarkable phase transition known as Bose-Einstein conde11sa
tion. This phenomenon is intimately related to the superfluidity of liquid

helium at low temperatures.


As we noted previously bosons are particles of integral spin that obey
Bose-Einstein statistics. There is no limit t(> the number of bosons that can
occupy any single particle state. We consider an ideal boson gas consisting of N
bosons in a container of volume V held at absolute temperature T. The Bose
Einstein continuum distribution is

N(e)

(18.24)

---

g e

..

Our initial concern is determining how the chemical potential J..L varies
with the temperature. We shall adopt the convention of choosing the ground
state energy to be zero. At T

0 all N bosons will be in the ground state since

0
there is no restriction on the number of bosons in a given state. Setting e
in Equation (18.24) .. we see that if .f(e) is to make sense J.L must be intrinsi
=

..

cally negative. Furthermore, J.L must be zero at a temperature of absolute zero


and only slightly less than zero at nonzero low temperatures.. assuming N to be
a large number.

Sec. 18.3

Bose-Einstein Condensation

341

At high temperatures, in the classical limit of

Boltzmann distribution applies:

a dilute

gas,

the Maxwell-

(18.25)
From Chapter

14,
1-L =

kT

z
ln -N ,

(18.26)

where

z =

2TrmkT 312
v.
2
h '

(18.27)

Thus

2TrmkT 312 V

_JL_ =-In

kT

(18.28)

As an example, for one kilomole of a boson gas comprising 4He atoms at stan
dard temperature
J.L

kT

=-In

and pressure, we have

27T(6.65

2
10- 7) ( 1 38
.

10-23) (273)

(6.63 X 10-34)2

312

22.4
6.02 X 1()26

-----

= -12.43.
The chemical potential per se is --0.29 eV. In comparison, the average energy
of

an

(3/2)kT = 0.035 eV at 273 K. Also,


12.4 = 13.9; substituting this value in Equation (18.25)

ideal monatomic gas atom is

JL) /kT
1.5 +
gives f ( e ) = 9.2 x 10-7, confirming

(e

!he validity of the dilute gas assumption.

In this classical limit,

27TinkT

h2

312 V

(18.29)

is a positive number that increases with temperature and decreases with the

particle density NjV.

The obvious way to determine the temperature dependence of JL is to

use the conservation of particles condition that led to the statistical definition

342

Chap. 18

Bose-Einstein Gases

of the chemical potential in the first place. For the continuum appr<>Ximation,
this is
X

N(e)tle

f(e)g(e)de

N.

(] 8.30)

For g(e) we have the result of Equation (12.26) with the spin factory_, equal
to unitv:
.,

g(e )de

4
=

2 7T V

h3

3/2

I /l

e de.

(1 8 3 1 )
.

Thus

e112de

e(e-p._t/kT

(18.32)

There is a significant flaw in this fortnulation. In using the integral approximation

of Equation (18.30) rather than the sum, the ground state e = 0 is left out. This
terrn, in the sum of particle states at low temperature, is the largest tetm of all. It

is omitted here because the density of states function g (e) depends one' :! which

is zero for the ground state. Under ordinary circumstances this doesnt matter..

since, fore small compared ith kT. the omission of the one term makes a negli
gible difference in the result. But at low temperatures, bosons condense into this

lowest state and its occupation becomes much greater than for any other state.
We can surmount the difficulty in the foJJowing way. The tota1 number of
bosons consists of Nc. in the ground state and Nex in the excited states. Hence
(18.33)
Since the ground state is excluded from Equation ( 1 8.32) the integration only
gives the number of bosons in excited states. Therefore

(18.34)
The integral can be solved numerically to obtain a relationship between J.L the
temperature,. and the particle density. It is n1orc instructive. however.. to note
..

that at temperatures very close to zero, N.,

-==

N. We can therefore pute

Equation (18.24) and write


N=::

1
e

JJ./kT

0 in

Sec. 18.3

Bose-Einstein Condensation

343

Then

for N large. (Even for a small macrosc opic system N will be a huge number.)
For low temperatures, then, we can safely put exp(- JL/ kT) equal to unity in
Equation (18.34). With the change of variable x

e/kT, we get

(18.35)

---

0 ex-

The value of the integral is 2.612

TT/2. Thus
3/2

(18.36)

The so-called Bose temperature T8 is the temperature above which all


the bosons should be in excited states. Thus we set Nex
N and T = T8 in
-

Equation (18.36) and obtaitt

N = 2.612V

(18.37)

Solving for T8, we get

To=

Jil

2/3

(18.38)

21rmk 2.612V

T8, all the bosons are in excited states. As T falls below T8, an increas
0, all the bosons
ing number of bosons occupy the groltnd state until at T
are in this state and N0 = N. The fractional number of bosons in the ground

For T >

state 1s

(18.39)

Dividing Equation (18.36) by Equation (18.37), we have

No
-=1N

T
T8

3/2

(18.40)

Chap. 18

344

Bose-Einstein Gases

What is a typical value of the Bose temperature? Consider a boson gas


made up of 6.02 x 1023 4He atoms confined to a volume of 22.4 x 10-3m3.
The mass of a 4He atom is 6.65

Io--27 kg. Using Equation

(18.38)

we find

that

X
l<f3
3
( 2. (}12) ( 22.4 X 10 )
6.02

In the case of helium, the gas liquifies at


before the temperature is reduced to

4.21

2/3
=

0.03 6

K.

Kat atmospheric pressure, long

0.036 K.

In fact, all real gases liquefy

before their Bose temperature is reached.

N0/ N and NeJ N versus T jT8 are shown in Figure 18.4. A corre
sponding graph of p,/ kT8, numerically calculated, is shown in Figure 18.5. *The
Plots of

sudden "collapse" into the ground state at very low temperature is called the
Bose-Einstein condensation. It is not a condensation in geometrical space but
rather in what might be called momentum spac e.
For many years physicists believed that a Bose-Einstein condensate did
not exist in nature. In 1995 researchers created a nearly pre condensate by

1.0
Nn
0.5
Figure 18.4

Variation with
temperature of N0/ N and

NcJ N

for a boson gas.

Nl'X
N

.N
t--

() -------r.. ------
)
1
-

T
Ts

JJ.

kT8

-0.5 ._

-1.0
Figure 18.5
boson gas.

Variation with temperature of I-ll kTn for

*See Problem 18-10 for the method of calculating p./ kT8 versus T/TH.

Sec. 18.4

Properties of a Boson Gas

345

cooling a vapor of rubidium atoms to a temperature of 1.3 x 10-7

K. The

achievement was hailed as "the most interesting development in atomic


phy sics in a decade,'' opening up an entirely new field of investigation.*

18.4

PROPERTIES OFA BOSON GAS

The bosons in the ground state do not contribute to the internal energy nor to
the heat capacit y. The first term in the sum I si is Noe0 For T < Ta, No may
be large but e0 = 0, and for T > T8, N0

0 in any case.
For temperatures above the Bose temperature, all the bosons are in
excited states and we may expect the internal energy to approach (3/2) NkT as
=

the temperature is increased. Below the Bose temperature the number of


bosons in the excited states is Nex = N(T/T8)3f2. As a first approximation, we
assume that each of these bosons will have a thermal energy of the order of
kT.Thus
3/2 .

(T < T8),

indicating that U varies as T512 below the Bose temperature. A more nearly
exact result is obtained by noting that
00

U =

eN(e)de.

(18.41)

Using Equations (18.24) and (18.31), we find that

21TV

2m J/2 oo

h2

e312de
(
)/kT

e e-p,

(18.42)

As we have seen, if we choose the energy of the ground state to be zero, the
chemical potential JL of the boson gas is very close to zero for temperatures
below the Bose temperature. Setting p, = 0 and making the substitution
x =

e/kT, we

obtain
'

2
1T

kT

2'1TmkT 3/2
h2

x312dx
0

eX -

(18.43)
'

*See the Richtmyer Memorial Lecture, "Bose-Einstein Condensation in an Ultracold

Gas,n by Carl E. Wieman, American Journal of Physics 64, pp. 847-855

(1996).

Chap. 18

346

Bose-Einstein Gases

the

The definite integral is equal to the product of

gamma function

f(S/2) 3TT1i2J4 and the Riemann zeta function {,(5/2)


(
expression for T11, Equation (18.38), we get
=

0.770 N kT

3/2

(T

T
'B

<

1.34. Using the

(18.44)

Tn),

not greatly different from our approximation.


It is a simple matter now to obtain the heat capacity:

(18.45)
Note that Cv is proportional to T

312

A graph of the heat capacity as a function of temperature is shown in


Figure

18.6. The

curve has a change in slope at T

T1h and the heat

capacity

1.92 Nk. At higher temperatures, Cv


approaches the classical value of (3/2) Nk.
has its maximum there, equal to

The absolute entropy at temperatures below the Bose tempt:rature can


be calculated from the heat capacity, since Equation

(18.45) holds at absolute

zero:

(T

The entropy goes to zero at T

(18.46).

T8).

() as it should, according to the third law of

thermodynamics. The Helmholtz function F

<

U - T S is

3/2

-0.51 NkT

T B.

(T

<

T8),

(18.47)

2.0
J

--

- .. - -

......

Cv 1.0
Nk
0.5
Figure 18.6 Variatio with
temperature of the heat
capacity of a boson gas.

t--

-------

Ts

Sec. 18.5

Application to Uquid Helium

347

from which we can deterrnine the pressure, using the reci procity relation
P

(aFjaV)r.N Expressing Fin terms of the volume V using Equation

(18.38) for T 8, we have

3(2
V,

(18.48)

from which

(T

<

T8).

(18.49)

The pressure of a boson gas at low temperatures is proportional to T512 and is

independent of the volume. This is because the bosons in the ground state
e =

0 have no momentum and therefore make no contribution to the pres

sure. It is easy to show that P

(2/3)UfV, the same relation obtained for a

classical ide al gas of noninteracting particles.

I 5 APPLICATION TO LIQUID HELIUM


The phase diagram of ordinary heliu1n (4He) is shown in Figure 18.7. The sub
stance has a critical point at a temperature of 5.25 K and exhibits unique

behavior in the vicinity of 2 K. Above the critical temperature 4He can't exist

as a liquid. When helium gas is compressed isother1J1ally at a temperature


below the critical temperature but above 2.18 K, it condenses to a liquid phase

30
p

(atm)

....._

20
10 .,__

solid
He
A-line
liquid
He II

liquid
He I

0 L...-.-..J..-.----..+-0
1
2
3
2.18 K

CP
4

He gas
6

T(K)

5.25 K

Figure 18.7 Phase diagram of 4He. The two liquid phases are
labeled I and II.

Chap. 18

348

Bose-Einstein Gases

called helium I. When the vapor is compressed at temperatures beJow 2.18 K,


a liquid phase called helium II re sul ts. Helittm II is a superfluid. He I and He II
can coexist in equilibrium over a range of pressures and temperatures defined
by the so-called lambda line. He II remains a liquid down to absolute zero.
Solid helium cant exist at pressures below 25 atm and it can't exist in
,

equil i b rium with its vapor at any pressure or temperature. Helium has two
triple points. At one of them liquid He I and liquid He II are in equilibrium
with solid helium. At the other, called th e A-point the two forms of liquid
helium are in equilibrium with the vapor.
The transition between ordinary liquid helium He I and superfluid He II
can take place at any point along the A-line. A graph of the heat capacity ver
sus temperature for the two phases has the general sha pe of the Greek letter ,\
(Figure 18.8). As the figure shows. the heat capacity does not change continu
ously. Its variation with temperature is vastl y different in the two phases.
The properties of l iquid He II are unique. Its viscosit y is virtuall y zero
hence it is called a superfluid. In the two-tluid theory of He II it is assumed
that, at temperatures below the A-pont.. liquid He 11 is a mixture of a no1n1al
fl uid having viscosity and a superfluid that has no viscosity. The proportion of
the superfluid is zero at the A-point a n d increases to

u nity

as the temperature

approaches zero.
Since the forces of interaction between atoms are weak in liq ltid helium
we might assume that liquid He ll i s an ideal boson gas, to a first appro xima

tion.* If we identify the A-point with the Bose temperature, the two-fluid model
s uggests

that the supc rfl uid component co n si sts of the ) atoms in 1 he ground

f
I
4

liquid
3

--

liquid
f-Ie I

I
.

He II

I.

..
.

c,
Nk

2
I
Jt--

----i

....
.

_j__
..

__ __

-.,

,.,_!

____

""
'

.. r( K)

--..

2.1H K
Figure 18.8

The heat capacity anomaly of He.

*It might seem strange to

apply a

theory for an ideal gas to a liquid. However. in some

ways, liquid helium behaves ore like a gas than a liquid.

Problems

349

state while the normal component is identified with the Nex atoms in the excited

states. As the temperature falls below T8, an increasing number of atoms popu
late the ground state, and the superfluid component becomes predominant.
Are T8 and the A-point the same? The volume of a kilomole of liquid 4He
is 27 X 10-3m3 so that the concentration is

Using the value of 6.65 x 10-27 kg for the mass of the 4He atom, we can com
pute T8 from Equation (18.38). The result is
T8

3.1 K,

as compared with 218 K for the A-point. We conclude that the superfluid prop
erties of liquid He II could be attributed, in part at least, to a Bose-Einstein
like condensation at the A-point. The ideal boson gas model, however, is only
an approximation to act1.1al liquid helium that neglects interatomic interac
tions. The complete picture is considerably more complcated.

\ '

PROBLEMS

18-l (a) Calculate the total electromagnetic energy inside an oven of volume 1m3
heated to a temperature of 400F.
(b) Show that the thermal energy of the air in the oven is a factor of approxi
mately 1010 larger than the electromagnetic energy.

18-l (a) Calculate Amax for the Earth, assnnting the Earth to be a blackbody.

(b)

Calculate the temperature at which the hntnan eye is most sensitive to a

wavelength of 500 nm.

18-3 Assume that the radiation from the Sun can be regarded as blackbody radia
tion. The radiant energy per wavelength interval has a maximum at 480 nm.

(a)
(b)

Estimate the temperature of the Sun .


Calculate the total radiant power emitted by the Sun. (The radius of the Sun

is approximately 7 X

lei m.)

18-4 (a) Find the frequency at which the radiant energy per unit frequency interval
of a blackbody is a maximum. How does this compare with the frequency at

(b)

which the radiant energy per unit wavelength interval is a maximuJn?


Find the frequency at which the cosmic background radiation is a ma?Cimum
and compare your result with that shown in Figure 18.3. Assume that

2.7 K.
I

Chap. 18

350

Bose-Einstein Gases

18-5 (a) Calculate the number of photons in equilibrium in a cavity of volume 1


273 K.
held at a temperature T

(b)

Compare this number with the number of molecules the same volume of an
ideal gas contains at STP.

18-6 Assume that the universe is a spherical cavity with radius 101n_ m and tempera
ture 2.7 K. How many thermally excited photons are there in the universe?
187 For photons. where there is no restricti<)n on the total number of )Jarticles, the
partition function is independent of N, as are the physical properties of a photon
gas. The partition function

for a single oscillator is given by (see Section 15.2)

In

-ln(1-

z =

e h ...

lkT),

ignoring the zero-point energy. The number of single particle (photon) states in
a volume V in the frequency range

v to

dv is

Therefore, the partition function Z of the photon gas is the sum over states
given

(a)

by

Show that integration by parts Jeads to the equation


In Z
where x

B1rV
he

hvfkT.

(b) Referring to Appendix D. show that


-

In z

87T> k T

45

he

18-8 Use the result of Problem 1H.7 to show that

(a)

kT2

consistent with Eq u ati o n

iJ In Z

aT

( 18.8):

v.v

(b)

Jl

kT

iJ

tn-z
=

riN

V.T

confirming the statement in Section 18.1:

(c)

u
=

+kIn Z

321T5k

45

he

T3V.

Problems

351

the same result as Equation (18.18); and

(d)

kT

which differs from the result P

In

<Jl/

v ,

T.N

obtained in Chapter 11 for an

( 2/3) ( U /V)

ideal gas of molecules.

18-9 In classical thermodynamics, it is found that

T iJP
iJT

For blackbody radiation, the energy density


does the pressure P
ential equation for

u
::s

( 1/3)

u.
1:11;;&

P.

u =
==

UfV depends on T only, as

Use the above classical result to obtain a differ-

whose solution is the Stefan-Boltzmann law.

18.5 is a plot of/ kT8 versusT /T8. In the region 0 < T /Tn < 1, p,/ kT8 is
essentially zero. ForT/T8 > 1, Equation (18.34) applies, with Nex set equal toN

18-10 Figure

(all N of the bosons are in excited states). Using the definitions x

p,fkT8, 11

= T /T8,

ejkT8,

show that Equation (18.34) reduces to


2.315

..

xl.'2dx

oex "-1
(

-&.) /

-,

which can be evaluated numerically to give as a function of 1J (Figure

18.5).

2T8 is approxi
18-11 The chemical potential of a boson gas at a temperature T
mately -0.8 kT8 Deterinine the mean number of bosons f (e) in single particle
=

states having energies of (a) 0, (b) 0.5 kT13,

18-12

(a)

(c)

2.0 kTu, (d) 2.0 kTn. (e) 3.0 kT8

Find the chemical potential of a kilomole of 4He gas at STP. Express your
answer in joules and in electron-volts.

(b)

Use Equation (18.24) to show that the mean occu11ancy of a single particle
state having energy

18-13

(a)

(3/2) leT is 8.8

10-7 at STP

An ideal boson gas consists of 4He atoms whose Bose temperature is 0.087 K.
Find the boson concentration, the number of bosons per cubic meter.

(b)

What percentage of the bosons are in the ground state at a temperature of


to-2 K?

18-14 In a Bose-Einstein condensation experiment, 107 rubidiun1-81 atoms were


cooled down to a temperature of 200 nK. The atoms were confined to
1!\
of approximately lo-- m3

(a)
(b)
(c)

volume

Calculate the Bose temperature 7!3


Determine how many atoms were in the ground state at 200 nK.
Calculate the ratio kTje0, where T
energy

e0

is given by

200 nK and \vhere the ground state

Bose-Einstein Gases

Chap. 18

352

18-15 Hydrogen freezes at 14 K and boils at 20 Kat atmospheric pressure. The density
of liquid hydrogen

is 70 kg m-).

Hydrogen molecules are bosons. No evidence

has been found for Bose-Einstein condensation in hydrogen. How do you


account for this?

18-16 Show that for temperatures below the Bose temperature of a boson gas,
(a) Cv
( d) P

(5/2) UjT, (b) S


(2,/3) UjV .

(5/3) UfT.

(c ) F

(2/3) U,

18-17 A system of N bosons of mass tn and zero spin is in a container of volume V at a


temperature T

>

0. The number of particles is

21TV

2m

/,
_, ..

1 '2

e' de
( - )lkT

e t P.

'

(Equation (18.32 ) ). In the dilute gas approximation, exp (

J.L! kT) >> 1, and the

Bose-Einstein distribution becomes the Maxwell-Boltzmann distribution.


Evaluate the integral in this approximation referring to Appendix D, and show
that
exp(- p.,/kT)

d
=

2
where A= h/(21TtnkT)1-' is the de Broglie wavelength of the particles' thermal
motion and d
(V I N) 11\ It follows that in the classical limit the average dis
==

tance

d between the particles is very large. compared to A.

19.1 The Fermi Energy

355

19.2 The Calculation of JL T

357

19.3 Free Electrons in a Metal

361

19.4 Properties of a Fermion Gas

364

19.5 Application to White Dwarf Stars

367

353

19.1 THE FERMI ENERGY

Fermi-Dirac statistics governs the behavior of indistinguishable particles o f


half-integer spin called fermions. Fermions obey the Pauli exclusion principle,
which prohibits the occupancy of an available quantum state by more than
one particle. We consider an ideal gas comprising
each of mass

m, in a container of volume

N noninteracting fermions,

V held at temperature T.

The Fer111i-Dirac distribution is

(19.1)

or

N(e)

(19.2)

in the continuum approximation. The right-hand side of Equation

(19.2)

is

referred to as the "Fern1i function"; it gives the probability that a single parti
cle s tate

e will

For

be occupied by a fermion. Clearly, 0

JL, f ( e )

<

f (e )

<

1.

has the value 1/2 at any temperature. Here J.L is the

chemical potential, which is a function of temperature. Its value at T

0, that

is JL ( 0), is called the Fermi energy. The Fermi energy is also written as eF.
Consider the Fermi function at a temperature of absolute zero.
,

Evidently, at T

0,

(e

JL(O) )/kT

if e <

oo

if e >

p,(O)
(0)

355

Fermi-Dirac Gases

Chap. 19

356

Correspondingly

0 tf E

>

JL(O)

This tells us that at T = 0 all states with energy e


states with

>

( 19.3)

<

JL(O) are occupied and all

(0) are unoccupied. At absolute zero fermions will occupy

the lowest energy states available. The exclusion principle says that only one
fettnion is allowed per state. So all

N particles will be crowded into theNlow

est energy levels.


It follows that only one configuration (microstate) is possible for the
who]e assembly at T
S

0. Thus the thermodynamic probability w is 1 and

k ln w = 0. The vanishing of the entropy at absolute zero is consistent

_ ction at T
with the third law of thermodynamics. The Fermi fun

19.1. We ask: how does the Fermi energy JL(O)

0 is shown in

m N, and V?
0. We need g(e),
It obviously doesn't depend on the temperature since T
the density of states of Chapter 12 once again. For particles of spin 1/2, such as
Figure

depend on
=

electrons, the spin factor is 2 and

(19.4)

For conservation of particles,

N;.
I

or

:X

N(e)de
0

Substituting Equations
p.,(O)

f(e)g(e)de

N.

(19.5)

()

( 19.3) and ( 19.4) in Equation ( 19.5), we get


J/2

2n1

N=

p.,(O)

e'ide

87T V
= --

(19.6)

f)

()

2m

/(e)
'

(T

0)

I ,._.____-.

Figure 19.1 The Fermi


function at T
0.
=

t(O)

Sec. I 9.2

357

The Calculation of p,(T)

lt
u
s
e
r
e
h
t
in
a
t
b
o
e
w
,
Solving for #J-(0)
h

#J.(O)

2/3

3N

2m

temperature

t
rm
c
F
a
e
c
u
d
o
tr
in
For convenience, we
s
a
n
e
tt
ri
w
e
b
n
ca
is
Th
F.
T
k
p
&
p,(O)
=

(19.7)
TF

such

that

2/3

(19.8)

--

27Tmk

'

1.504V

analogous to the Bose temperature given in Equation

19.2 THE CALCULATION OF

(18.38).

p,(T)

Figures 19.2 and

19.3 show how the Fermi function changes with temperature.

To obtain these curves, we must determine J.L(T). The calculation is consider


0. We have
ably more complicated than it was for T
=

2
3/
2m

:o

'

elf2de

+1

0et JL

(19.9)

Let the integral be I in this expression and equate Equation (19.6) with
Equation

(19.9). The result is

e(s-p.)jkT + 1.

/(e)
1.0 ......

.-- (T

--- - ---

.
---------5
0

Fig11re 19.2 The Fermi


function at T
0.2TF.
=

(19.10)

0.5

1
I
I
I

1.0

0)
T

p,

0.2 TF
0.96#L(O)

e/(0)

358

Chap. 19

f(e)
1.0

0"

Fermi-Dirac Gases

(T =O)

- - - -

T
1 J.L
I
I

1.2 TF
-0.45(0)

I
-----------

--1
.
.
.

Figure 19.3 The Fermi


function at T
1.2 TF.

0.5

The plus sign in the denominator of

1.0

e/(0)

integrand makes all the difference in


our ability to evaluate the integral. It can be evaluated numerically. But we can
the

find a very good infinite series approximation for p,(T) in the following way.
Let

I=

(19.11)

where

')

and

f (e )

I nt egra ting E q uati on

1
==

) k r

eF.-p._.-

+ 1

( 19.1]) by parts, we ot,tain


df(e)

de

Now F ( oc)

f(s)f (e)
..

F(O)

is i nfin ite

but f ( x)

11

==

0 and its de ca y is exponential,

0 at the upper litnit. At the

Jo\ver limit

0, so the product is zero. Hence


:/_

F (e)

d_((t-;)

de

ll.

so

/(0) is finite but

The Calculation of J.t(T)

Sec. 19.2

359

19.1 shows that at T 0, the functio11 f(e) has a zero


JL(O), where the slope is infinite. Thus the
slope everywhere except at e
derivative d.f (e)/de is a Dirac deita function. At temperatures less than TF but
A quick look at Figure

greater than zero, we might expect tl1e derivative to be


function peaked at

e=

p,. This suggests that we expand

kind of 'fuzzy" delta

F(e)

in a Taylor series

about p, since the only significant contributions to the integral! will be in the
vicinity of e

JL. * The expansion is

(19.12)
The contribution to the integral of the linear term is zero since

df (e) /de

is

symmetric about JL and the areas under the curve to the left and right of p, will

cancel. Thus

(19.13)
Evaluating the derivative

df(e)/de in the integrand, we obtain

=
I

We set

(e

1
kT

kT
F( e )e<e-,u.)J
-de
[e(e-,u.)/kT + J ]2

(19.14)

-----

J.L) I kT in the integrand, so that


I=

F(y)eYdy
.
2
(eY
+ 1)
-p.jkT
:G

0 where y

The factor

eY is already negligible

for

at least at low temperatures, which is the region of concern. Hence

e<

Bf,

at

we can safely replace the lower limit by

co

J.L/kT, and it's small

Then, using Equation

(19.12), we

get

*This method of evaluating the chemical potential and the internal energy for low temper
atures is known

as

the Sommerfeld expansion.

Fermi-Dirac Gases

Chap. 19

360

7T'2/3.

The first integral has the value 1 and the second

Thus, reminding our

selves of Equation ( 19.10), we have

A little algebra gives

JL

JL(O)

7T2 kT

kT

-2/3

1 + 8

-li_
:z:o ,.,., ( Q )

Finally, we replace J.L in the ''correction term'' by J.L (0)

1T2

1-L(O)

==

kTF, to obtain

(19.15)

A comparison of this approximate expression with the Hexact'' numerical eval

0 at T fTF
1.1 p, (0).
instead of

uation is shown in Figure 19.4. The approximation gives J.L

instead of 0.999, and JL

-0.85 J.L (0)

at T/TF

1.5

1.10

We especially note that J.L is positive for temperatures below the Fertni

temperature and negative for higher temperatures. As the temperature


increases above

TF, more

and more of the fermions are in the excited states

and the mean occupancy of the ground state falls below

1.0

0.5
J.L

(0)

-0.5
Figure 19.4

Exact and

approximate calculations of
JL/p,(O) versus T/TF.

-1.0

1/2. In this region,

....

approximate

t--

exact
0.5

'
'

1.5
\
\
\
\
\
\
\
\
\
\

T!TF

Sec. 19.3

Free Electrons in

Metal

361
1

f(O)

_
-e-p./kT+l

<

2'

which implies that

JL
kT
or J-L

<

<

0'

0. We also note that the situation is different for

boson gas, where J.L

is negative at all temperatures and is zero at absolute zero.


At high temperatures the fermion gas approximates the classica1 ideal

gas. In the classical limit,

JL

-kT In

(19.16)

with
Z
N

27TmkT
h2

312

.
N

(The spin degeneracy factor is 2 for fermions.) For


large negative value and exp(- J.LfkT) >>

(19.17)
T

>>

TF, JL/kT takes

on a

1. As an example, consider a kilo

mole of 3He gas atoms (which are fermions) at standard temperature and
pressure (Problem

T/TF

19-2). The Fermi temperature is 0.()69 K, so that


-12.7
3900. Using Equations (19.16) and (19.17), we find that J.L/kT

and exp(- JL/kT)

3.3 x lfr'\. The average occupancy of single partitle states

is very small, as in the case of an ideal dilute gas obeying the Maxwell
Boltzmann distribution.

19.3 FREE ELECTRONS IN A METAL


Statistical tl1ermodynamics provides profound insights into the behavior of
conduction electrons in metals at moderate tentperatures. Electrons are spin

1/2 fermions. Each atom

in the crystal lattice of the metal is assumed to part

with some number of its outer valence electrons, \Vhich can then move freely

about in the metal. There is an electric field due to the positive ions that varies

widely from point to point. Ho\vever, the effect of the field is canceled out
except at the surface of the metal where there is a strong poteittial barrier,

called the work function, tht draws back into the metal any electron that hap
pens to make a small excursion outside. The free electrons are therefore con

fined to the interior of the metal as gas molecules are confined to the interior

of a container. We speak of the electrons as an electron gas.

Chap. 19

362

Fermi-Dirac Gases

<h
.
,
-

1--------.-----

......

._--

-Fern1i level

J.L(T)

Figure 19.5 Potential well for


free electrons in a metal

lr

In this model, the free electrons move in a potential box or well whose
walls coincide with the boundaries of the specimen. They occupy energy states
up to the so-called Fermi level, which is the chemical potential p.(T). The work

cf> is the energy required to remtlve an electron at the Fermi level


from the metal surface. The depth of the potential well is equal to J.L(T) + 4J
function

(Figure 19.5).
The Fermi level of the free electrons in most metals at room temperature
J.L(O). It is often assumed
is only fractionally less than the Fermi energy eF
that the two are equaL and this leads to confusion. The Fermi level, strictly
speaking, is p,(T), which is an approximation to the Fermi energy valid for
T << TF.
A more realistic picture of the potential well is given in Figure 19.6,

which shows how the potential varies in the vicinity of the positive ions in the
crystal lattice. The periodicity leads to a band structure in the density of quan
tum states, which is the foundation of semiconductor physics.
To get an idea of the magnitude of the quantities in the electron gas
-

modeL we consider the free electrons in silver\ which is monovalent (one free

I
I
I

I
I
I
'

Figure 19.6

-'-----'-----1.--.L-

Sketch of th potential well showing

periodicities associated with the positive ions of the


crystal.

Sec. 19.3

Free Electrons in

Metal

363

electron per atom). The density of silver is 10.5 x Hf kg m-3 an


d its atomic
weight is 107. The concentration is therefore

Since silver is monovalent, this is also the electron concentration. The Ferttti
energy is given by Equation (19.7):

ep

J.L(O)

h2

2/ 3

3N

=- 2m
81rV

Herem is the electron mass, equal to 9.11 x 10-31 kg. So

e F

(6.63

3
2
2/
3 X 5.90 X 10 8

10-34)2
X

------

2 X 9.11

- 8.85 X 10
_

10-31

_ 19

J X

87T

1 eV
1.6 X 10-

19

5.6 eV.

The Fermi temperature is


T.

eF
k

5.6 eV

=
8.62

10-5 eV K-1

= 65 '000 K.

The ratio T/TF at room temperature is

T
TF

300

6.5 X 10

= 0.00462.

At room temperature, therefore, the electron gas is in the so-called degenerate


region T << TF. The chemical potential (the Fermi level) can be found from
Equation (19.15). The computation gives

J.t(T)

0 999eF
.

This shows why J.L(T) is often identified with

eF.

The work function tP depends on the metal and the condition of its sur
face and is typically of the order of 3-4 eV. At very high temperatures, some of

Fermi-Dirac Gases

Chap. 19

364

the free electrons may have sufficient energies to leave the metal. 1his results
in thermionic emission. The condition for their escape is
,

where Px is the component of the electron's momentum normal to the surface


of the

metal

19.4 PROPERTIES OF A FERMION GAS


The function

N ( e)d

is the number of fer1nions in the single particle ene.rgy

range e to e+de. We know that

N(e)de

f(e)g(e)ds,

where the Fermi function f ( s ) and the degeneracy function

g( s)

the curves of Figure 19.7 for a temperature in the range 0

r << TF.

<

behave like

The product of the two curves gives N (e) versus E as shown in Figure 19.8.
The electrons crowd around the Fermi energy because the degeneracy
increases with energy; there are threfore n1ore available quantum states to be
occupied with 0 or 1 electron( s) per state.

(It

is the electrons in the tail of this

distribution that have the best chance of escaping from the metal in the free
electron model.)

/(e)

(T
....,_.

0)

___ ----

I
I

'--

..__
_
__.;:....
_...
..
..

____
_

(a)
Figure 19.7

F.

L...... --- - --- ---..

(h)

The variation with single particle eergy of (a) the Fermi function._
and (b) the degeneracy function. The curves are sketched for 0 < T << TF.

t;

N(e)

,
,

,'

'

__

...

I
I
I

--

(T
I

0)

I
Figure 19.8 The energy
distribution of fem1ions for
0 < T << TF.

The internal energy of a gas of N fermions is


oo

eN(e)de

ef(e)g(e)de

47TV

3
2
/
2m
h

e312de

oo

e(e-J.L)/kT + 1

An approximate evaluation of the integral can be carried out in the same


manner in which we evaluated the similar integral in Section
of order T4 is included, the result is

3
At T

0, U

2
2
57r T

(3/5)Nc;F;

7T4

19.2. If the term

(19.18)

this energy is large because all the electrons must

occupy the lowest energy states up to the Fermi level. The average energy of a
free electron in silver at T

e(O)

0 is

u (0)
N

5eF

3 .

5 (5.6 eV)

3 4 eV.
.

Note that the mean kinetic energy of an electron, even at absolute zero, is two
orders of magnitude greater than the mean kinetic energy of an ordinary gas
molecule at room temperature.
The electronic heat capacity Ce can be found by taking the derivative of
Equation

(19.18):
e

dT

+ .

(19.19)

For temperatures that are small compared with the Fern1i temperature, we can
neglect the second term in the expansion compared with the first and obtain

Chap. 19

366
1r1
=

---

Fermi-Dirac Gases

kT

Nk

( 19.20)

BF

For silver at room temperature.

t:

5.6eV

2.2 X 10

Thus the electronic specific heat capacity is 2.2


explain s

Nk.

1 o- 2 R. This small value

a puzzle. Metals have a specific heat capacity of about 3R,

the same

as

for other solids. It was originally believed that their free electrons should
contribute an additional

( 3/2) R associated with their three translational degrees

of freedom. Our last calculation shows that the contribution is negligible.


Wh)' is it so small? While the kinetic energy of the electrons is much
greater than the thermal energy of electrons in a gas the energy of the elec
trons changes only slightly with temperature

(dU/dT

is smalJ). Only those

electrons near the Fermi level can increase their energies as the temperature is
raised, and there are precious few of them.
At very low temperatures the picture is different. From the Debye the
ory, Cvcx.T3 and so the heat capacity of a metal takes the form

where the first term is the electronic contribution and the second is associated
with the crystal lattice. At sufficiently low temperatures, the former can domi
nate, as the sketch of Figure 19.9indicates.
Whereas we have emphasized free electrons in metals most of our
results apply to any ideal gas of fermions. Using the fact that the reversible

3/v'k

1--

--

--

- -- - - - - - -- -- - -.,17 -2 -

heat

capa_cuy

Celect onic

Figure 19.9
..

Sketch of the heat capacity of a melal as a

function of temperature showing the electronic and lattice


contributions.

Sec. 19.5

Application to White Dvnrf Stars

367

heat flow into a gas at constant volume is given by TdS


late the entropy from Equation

c dT'

Therefore

0 at T

Vk

(19.21)

0, as it must be. The Helmholtz function F

U - TS is

T'

{19.19):

1T2

Cl'dT, we can calcu

+ .

--

7T4

+ .

(19.22)

The ferttlion gas pressure is found from

It is left as Problem

iJF
-

av

r.N

(19.23)

19-9 to prove that


p

57r2 T 2

2NkTF
5 v

+ . .

(19.24)

Comparison with Eq uation (19.18) sho\vs that P


(2/3)UfV.
For silver we found that N/V
5. 9 X 1028 m-3 and TF 65,000 K. Thus
=

5
Given this tremendous pressure, we can appreciate the role of the surface
potential barrier in keeping the electrons from evaporating from the metal.
-

19.5 APPLICATION TO WHITE DWARF STARS

Very high pressures, of the order of 1017 atm, exist in the degenerate electron gas
of a white dwarf star. It is this pressure that prevents its gravitational collapse.
A typical star has a hot core with a temperature of the order of

107

K.

The heat energy is supplied by thertnonuclear reactions. The atoms are com
pletely ionized (kT is 900 e V at tttis terr1perature), creating a huge electron
gas. In young stars the electron pressure is sufficient to withstand the weight of
the material pressing on the center, thereby preventing collapse. In old stars,
the hydrogen at the core has run out, fusion has stopped, and the core cools.
However, the loss of gravitational energy results in an increase in the kinetic

Fermi-Dirac Gases

Chap. 19

368

In
t.
fse
of
a
lly
ti
pa
r
is
s
es
oc
pr
g
in
ol
co
e
th
d
an
energy of th e electrons and ions
n
tro
ec
el
e
h
t
by
d
te
e
n
v
p
r
e
is
t
in
po
n
i
a
r
t
a
e
c
nd
e
yo
b
se
white dwarfs the colla p
of
us
di
ra
e
th
d
an
n
Su
e
th
of
s
as
m
e
th
ve
ha
ly
l
gas pressure. White dwarfs typ i ca
the Earth.

The pressure of the electron gas in Si rius B.


mated us ing the formula

white dwarf. can be esti-

=
P

(19.25)

We need the fol low ing characteristics of Si riu s B:

Mass M

Radius R

V o) umc V

==

2.09

::< IO'f' kg

5.57

::<

106m

7. 2 3

::<

1 0 t J
'l

We assume that nuclear fu sion has ceased

"
'.

that all the core hydroge n has

been converted to helium. We further assume that the helium is completely

io nized

that each He atom has Jost two electrons. In addition, we suppose

that the number of nucleons in th e star is equal to its mass di vi ded by the mass
of a nucleon. Thus

No. of nucl e ons

2.09

==

1 o-'u

I . 66 X l ()

'l-

..

'

1 . 26

1 a,.
-

--

Since there are four

n uc l eons

and t\VO electrons in a helium a torn, the number

of electrons N in the fermion gas of Sirius B is 0.63 x 1057 Then wi t h


1n e

9. 11 X 1 0

31

kg,

3N
81rV
Then TF

eF! k

3.9

5.33

10<) K. If 7

..

Io-141

==

0.33MeV.

107 K. then the condition T << TF is

satisfied.
To a fair approximation. the ele ct ron gas p ressure is given by Equ a t i on

( 19.25). Thus

5- 7.23

100

( 5. 3 3 x l 0

14

1 .8

1 02 Pa

1. 8

1 0 17 atm.

We "d lik e to show that this pressure is sufficient to prevent further collapse.

Sec. 19.S

Application to White Dwarf Stars

369

A white dwarf is stable when its total energy is a n1inimum. The energy is

(19.26)
where Ue is the energy of the electron gas and Ugrav is the gravitational energy.

We need to express each in ternts of the star's radius


tive of U with respect to
ForT<< TF,

R and set it equal to zero.


3

But

R, then take the deriva

3 It

3
81rV

(4/3)1rR3, so
(19.27)

where

(19.28)
The gravitational energy of a solid sphere of mass M and radius
known from classical mechanics to be

U1rav

b
=

'

is

(19.29)

where

(19.30)
Thus

(19.31)
Minimizing this function, we find that

Rmin

2a
=

.
b

(19.32)

Fermi-Dirac Gases

Chap. 19

370

Substituting the values of a and b from Equations (19.28) and (19. 3()), respec
tively, we find that (see Problem 19-5)
( 19.33)
Since the observed value of the radius of Sirius B is roughly equal to this, we
conclude that the star has contracted down to its stable minimum size and is
truly a white dwarf.

.
White dwarfs are at present viewed as comprising a core of degenerate

electron gas surrounded by a nondegenerate (T > TF) outer layer Nuclear


.

reactions have stopped. The star is cooling slowly. The time before it becomes
invisible is estimated to be 1010 years. Since the age of the universe is believed
to be 1.5 X toH> years it follows that very few white dwarfs have had time to
,

become invisible.
The st abi li ty of white dwarfs is but one example of the very extensive use
of quantum statistics in the st udy of st ellar evolution.*

PROBLEMS

19-1 Assume that for T

3TE- t he value of the chemical potential is -5.6 eF.

Calculate the value of the Fermi function

of e/ et of (a)

0, (b) 0.5, (c) I .0. and ( d ) 2.0.

f (e) at the temperature T for values

19-2 Consider a ki lom ol e of -He gas atoms under STP tonditions.


(a) What is the Fenni temperature of the gas?

(b)

Calculate J.t/kT and exp(- 11/ kT).

(c) Find the average occupancy


energy of (3/2) kT

19-3

.f ( f; ) of

a sin gle particle state that has an

or a system of n oni nteract i ng electrons. show that the probability f (e) of find

A above the chemical potentialJ.L is the


same as the probability of finding an electron absent from a state with energy .6.
belo' J.L at any given temperature 7
mg an electron in a state with

e n e rg y

19-4 (a) Verify that the average energy per rrmion is ( 3/5 )r.r at absolute zero bv
..
making a direct cal cu la t i on of U (0)/V.

(b)

Similarly, prove that the average speed of a fermion gas particle at T


(3/4)tF where the Fermi v d oc i ty v1 is defined by r.r
.

0 is

l/2)ml:l

this section was motivated by Chapter 12 of Statistical Ph\sics bv


W.C. V Rosser, Ellis Horwood. 1982. Rosser gives an exten si ve treatment of white dwrfs and
*The discussion of

neutron stars.

Problems

371

SectJo 9.2. Wnte F(e)


(2/5)e512 and expand F(e) in a Taylor series
a
b
o
u
t
JL, retamm the first thre terms only.
You will get an expression for u in
terms
=

approximately
Ce

where A can

be A

be determined by

AT,.

experiment. For gold, the constant A is found to

0.73 J kilomole-1 K-2 Compare this result to the value obtained from

Equation (19.20). (The atomic weight of gold is 197 and its density is

18.9 x lQl kg m-3.)


197

(a)
(b)
(c)

19-8

Calculate eF for aluminum assuming three electrons per aluminum atom.

1000 K, JL differs from eF by less than 0.01 %.


(The density of alumincm is 2.69 x lQl kg m-3 and its atomic weight is 27.)

Show that for alutninum at T

Calculate the electronic contribution to the specific heat capacity of alu


minum at room temperature and compare it to 3R.

In sodium there are approximately 2.6 x 1028 conduction electrons per cubic
meter, which behave as a free electron gas. Give an approximate value for the
electronic specific heat of sodium at room temperature. (The atomic weight of
sodium is

23.)

In the series expansion force, show that at this temperature the

cubic terrn is negligible compared with the linear terrn.

19-9 Derive the expression for the fermion gas pressure, Equation (19.24), using
Equations (19.22) and (19.23).

1910 Calculate the isothermal compressibility of the ferrnion gas consisting of the

free electrons in silver. Compare your answer with the experimental value for

silver of 0.99 x 10-11 Pa-l.

1911 Show that the gravitational energy of a uniformly solid sphere of mass M and
(3/5) GM2/R. (Consider the l?'avitational interaction
radius R is Ugrav
=

.
between a solid sphere (core) of radius rand a surroundmg shell of thtckness dr.)

19-U For the white dwarf star Sirius B, do the computation leading to Rmin

:==

7 X lW m.

Use Equations (19.28) and (19.30) with the appropriate parameter values.

19-13 Consider the collapse of the Sun into a white dwarf. For the Sun, M
R 7 x lOS m, V 1.4 x 1027 m3
=

(a)
(b)
(c)

2 x 10 30 kg,

Calculate the Fermi energy of the Sun's electrons.

What is the Fermi temperature?


What is the average speed of the electrons in the fermion gas (see Problem

19-4). Compare your answer with the speed of light.


(d) What is the density of the electron gas? Compare it with the density of water.
(Note: For a star with mass greater than a critical value of 1.44 times the mass of

the Sun, the collapse continues beyond the white dwarf stage as gravitational
attraction overcomes the electron pressure. The result is a very dense neutron
star that eventually ends in a supernova explosion.)

20.1 Introduction

375

20.2 Uncertainty and Information

375

20.3 Unit of Information

379

20.4 Maximum Entropy

381

20.5 The Connection to Statistical Thermodynamics

384

20.6 Information Theory and the Laws


ofThermodynamics

20.7 Maxwell's Demon Exorcised

386
387

373

20.1 INTRODUCTION
No reasonably up-to-date introduction to statistical therrnodynamics would
be complete without some mention of inforn1ation theory. We have seen that
the entropy is a measure of the degree of randomness or disorder of a system.

A system that is in its single lowest quantum state is one in perfect order. On
the other hand, a system that is likely to be in any of a ntltnber of different

states is a disorderly system. The greater the number of states it might occupy,
the greater the disorder. Disorder, in this sense, implies a lack of infoiiitation
regarding the exact state of the system. A disordered system is one about
which we lack complete information. The reason for briefly discussing infor
mation theory is that it casts further light on the meaning of entropy.

20.2

UNCERTAINTY AND INFORMATION

In his classic 1948 paper, A

Mathematical Theory of Communication, Oaude

Shannon laid down the foundations of information theory.* It was further devel
oped by Leon Brillouint and applied to statistical thermodynamics by E.T.
Jaynes.* Shannon's single greatest accomplishment was to provide a mathemati
cal measure of inforrnation. The cornerstone of the theory is the observation
that information is a combination of the certain and the uncertain, of the
expected and the unexpected. The first part carries no real inforrnation; the mea
surable infortttation is contained in the second part, the unexpected. The degree
of surprise generated by a certain event

say, one that has already occurred

is

zero. If a less probable event is reported, the information conveyed is greater.


The infortnation should therefore increase as the probability decreases.
Uncertainty is reduced by relevant inforn1ation. If I throw a die rnany
times, I can expect that the average score will be 3 1/2. If I throw it once there
*C.E. Shannon, Bell System

Tech. J. 27,379 and 623 (1948).


L. Brillouin, Science and Information Theory, Academic Press, New York, 1956.
tE.T. Jaynes, "Information Theory and statistical mechanics" Phys. Rev. 106,620 (1957) .
t

.,

375

Information Theory

Chap. 20

376

is an equal probability of getting any number between

1 and 6. But if someone

tells me before I throw it that the same number 3 appears on all six sides, my
uncertainty about the outcome of the experiment is totally removed. I can ask

the question, before I throw the die, how much uncertainty is there about the
outcome? Or. correspondingly. how much information is transmitted by the
actual result?
For a given experiment, consider a set of possible outcomes whose probabilities are p1, p2..

tity H (p1

p11)

. . Pn Shannon discovered that it is possible to find a quan

that measures in a unique way the amount of uncertainty

represented by the given set of probabilities. Only three conditions are needed

H (p1

to specify the function

p,) to within a constant factor. They are:

1. H is a continuous function of the p;.

2. If all the p;'s are equaL

increasing function of

p;

n.

= l jn; then

H(ljn, ... , 1/n)

is a mo-notonic

(With equally likely outcomes.. there is more

uncertainty when there are more possible events.)

3. If the possible outcomes of a particular experiment depend on the possi

ble outcomes of

subsidiary experiments, then

is the sum of the

uncertainties of the subsidiary experiments.


The last condition is known as the composition law and needs some elab

oration. Consider three possible outcomes of an experiment with probabilities

1/6. We note that the sum of the probabilities is


1/3, and p3
Pt
1/2, P2
1 as must be the case. Now consider a second experiment broken down into
=

two successive experiments. Let the first subsidiary experiment have two

equally probable outcomes, with p1

1/2 and P2 = 1/2. Let the second sub

sidiary experiment have two possible outc<>mes with probabilities


and

Pi

p{

2/3

1/3 (Figure 20.1 ). We note that the probability of two independent

outcomes occurring together is the product of the probabilities of the separate

outcomes. We also observe that in the two cases the final outcomes have the..
same probabilities. Thus.. condition 3 requires that
1

H 2'3'6
l/2
single

expenment

l/3

1 1

22
l/2

two

expenments
SUCCCSSI\'C

l/6

(20.1)

l/2

213

l/ 3

l/3

116

Figure ZO.l Decomposition of an experime nt For the first expe rime nt.. the
pro babi liti es of the outcomes are p1
1/2 p
1 /3 .. and P3
1/6. In the
second experiment .. the first subsidiary experiment has probabilities p1
1/2
1/3.
and P2 = 1/2 while the second has probabilities J't' = 2/3 and Pi
.

..

..

Uncertainty and Information

Sec. 20.2

377

The coefficient of the second term on the right-hand side


is 1/2 because the
second outcome occurs only half the time (see Problem 20-1

).

Shannon showed that the simplest choice for H consistent w


ith the three
conditions is

H (Pt

P,)

i=l

(20.2)

1 (p;),

where the function f is unknown. Since f (Pi) is a continuous function, it suf


fices to detetntine /for the special case of equal probabilities. For P;
1 / n for
=

all i, Equation (20.2) gives

(20.3)

We note that condition 2 requires that


d

f
n
dn

>

(20.4)

0.

Invoking the composition law, condition 3, we consider the case where


the total number of equally possible outcomes of the first subsidiary experi
ment is r and that of the second subsidiary experiment is s. Thus n
rs and
=

--

... , -

Using Equation

1
1
1
1
=H --, .
H - , ... , - - H , . ..,
11
n
s
s
rs
1

rs

(20.5)

(20.3) we obtain
1

rf -

Let R

1/r and S

sf

1
-

rsf

1
TS

(20.6)

1/s. Equation (20.6) then becomes

f(R)
R
Setting g(R)

5f(S)

(1/ R)f(R), g(S)


g(R)

1
f(RS)
.
RS

(20.7)

(1/ S)/(S), etc., we have


g(S)

g(RS).

If we differentiate this equation first with respect to R and then with respect to

S, we obtain

Chap. 20

378

g'(R)

Sg'(RS)

g'(S)

Rg'(RS)

Information Theory

and

From these relations, it follows that


Rg'(R)

(20.8)

Sg'(S).

Since R and S are independent variables, the only way in which Equation (20.8)
can be satisfied is for both sides to be equal t<l the same constant:

Rg'(R)

(.A

constant)..

(C

constant).

or

g( R)

A In R

Recalling the definitions of R and g(R) we see that this gives


A

-In

r +

C
r

If the probability is
C

1, the uncertainty must be zero. That is f ( 1)

0, so

0 .. and for equal probabilities..

nf

-A ln

ll

(20.9)

n.

The remaining issue is the sign of


(20.4), we find that

A. Applying the condition of Equation


(),so A must be negative. Setting A
K with K

A/n >
1/p we obtain
positive, and writing n
-

..

f ( p)

K p h1 p.

(20.10)

Using this result in Equation (20.2), we obtain the expression for the uncer
tainty we have been seeking:
"

p;
i =I

where K is a positive constant.

ltl p;.

(20.11)

Sec. 20.3

Unit of Information

379

Let's check our example of the composite experiment. We have


1

1
2 1
+ -ll
2
3'3

--

K(-0.346 - 0.346)

K
-

( -0.270- 0 366 )
.

l.OlK.

Also, for equal probabilities, p;

1/n for all i, andH

K In

n. It follows that

Equation (20.11) satisfies all the conditions imposed on the measure of


uncertainty.
It is instructive to consider the

binar)' case of two possible outcomes


an experiment with probabilities p1 and p2 such that p1 + P2 = 1. Then
H/K

=
=

P1

In

P1 - P2 In P2

Pt

In

Pt

(1 -

of

(20.12)

Pt ) ln( l - Pt)

What isH when p1 is either 0 or 1? We need L'Hopital's rule in the form


.

llDl

X0

( )
V (X )
u x

'

ltm

(x )
.
'
v (x )

X-+0

liere

lim(x In

X-+0

In x
.
ltm
1

X-+0

Thus the uncertainty is zero when either

1.1m

1
i

X-+0

0.

2
x

p1 or p2 is zero and the other is con

sequently unity. Also, if we differentiate Equation (20.12) with respect to p1


and set the derivative equal to zero, we find that H/K has a maximum at
p1

1/2 ( = p2). These results are illustrated in Figure 20.2.

These inferences can easily be extended to the case of n possible outcomes.


In particular, we can use the method of Lagrange multipliers to maximize H sub-

ject to the cortdition

p;

1 (Problem 20.2). The result isHmax

K In

n.

20.3 UNIT OF INFORf1ATION

We need to choose a value for K so that H has some convenient unit. Consider
the simplest possible experiment with equally likely outcomes
or the binary digits 0 or 1. Then

heads or tails,

Information Theory

Chap. 20

380

HIK

Figure 20.2

Uncertainty

1'----

measure for the case of two

____.

___ ___

0.5

possibilities with probabilities

0.5

()

Pl and P2

H
-

1
1
1
1
- In - - - In
2
2
2
2

In 2.

If we use 2 as the base of the logarithm and take K

==

1 we obtain

We call the unit of information a bit for Bin a ry dig .IT.*


Consider the follo\\'ing examples:

1. Decimal digit
Here

Thus a decimal digit contains about 3 1/3 bits of information.


2. Foregone conclztsion
The message ''the sun will rise between midnight and noon tomorrow'
has a probability of one so that H

0 hits.

3. Coin flipping
Flip a coin three times. The information will be three times that associ
ated with one flip of the coin. Thus H

3 1og2 2

==

3 bits. This can be

*The ubit'' was coined by the statistician John Tukey during a lunchtime discussion at
AT&T Bell Laboratories.

Sec. 20.4

Maximum Entropy

381

TABLE 20.1
Known probabilities
of occurrence for
symbols in the
English alphabet.
Symbol

P;

blank
E
T
0
A

0.200
0.105
0.072
0.065
0.063

J,Q,Z

0.001

written H

log2 23 By extension, flipping a coin N times yields N bits

of information.
4. English alphabet

Suppose that a message comprises words using the 26 letters of the


alphabet and
equal, then H

a
=

blank, totaling 27 characters. If the probabilities were


log2 27

4. 76 bits per character. If, on the other hand,

we use kttown probabilities of occurrence, H is reduced to 4.03 bits per


character (1able 20.1). Because of the redundancy built into the struc

ture of the language, the actual information content is in the range of 1

to 2 bits per character. The calculation involves the use of conditional


probabilities.

t0.4 MAXIMUM ENTROPY

Ne have seen that the amount of uncertainty represented by a discrete proba


>ility distribution is
n

H (Pt

Pn)

K
i= I

p; In p;,

(20.13)

vhere K > 0. In Section 20.5 it will be shown that, apart from the constant,
his is just the expression for the entropy in statistical thermodynamics.
.

llerefore, we can consider the tet111s "entropy" and "uncertainty'' to be syn


nymous. Indeed, entropy is a tetnl used in information theory.

The connection between entropy and uncertainty. is illustrated by the

Jllowing example. Imagine that an insulated container is divided into two

Chap. 20

382

Information Theory

chambers by a partition. Initially there is gas on one side of the partition and
none on the other. Suppose the partition is suddenly removed so that the gas
expands to fill the whole container. We know from our study of the free expan
sion of a gas that the internal energy is unchanged but the entropy increases.
When the molecules of the gas are all in one of the two chambers, the
entropy is less than when they are distributed throughout the container. Our
knowledge of the possible positions of the molecules is greater when they are
all on one side of the partition than when they are somewhere in the entire
container. The more detailed our knowledge is concerning a physical system,
the less uncertainty we have about it .. and the less the entropy is. Conversely,
more uncertainty means greater entropy.
To make inferences based on only partial information. it is necessary to
determine the probability distribution that has maximum entropy subject to
what is already known. Suppose that we know the mean value of son1e partic
ular variable x that can assume the discrete values .xi, i

1" 2, .

..

n. The

mean

value is

..

.t

(2().14)

PiX,.

where the unknown probabi1ities satisfy the condition

II

:-

(20.15)

In general, there will be a large number of probability distributions

{p,, P2 ... Pn} consistent with the infortnation given- Our problem is to deter
mine that distribution which maximizes the uncertainty (i.e. the e-ntropy). We
..

introduce the Lagrange multipliers a and {3 and follow the procedure we used
in Chapter 13. We write

(20.16)
where

(20.17)
I

and

(20.18)

Sec. 20.4

Maximum Entropy

383

Then

_,;

P; In P;

a
+a-
apj

'

or

- K In Pi - K

Pi

+a +

Pi

f3xi = 0.

Solving for In Pi and simplifying the constants, we obtain

or

We can determine the new Lagrange multipliers A and p, from the constraints.
Equation (20.17) gives
e-A

1
=

---

so that

Pi=

(20.19)

We define the partition function


z

(20.20)

Then
(20.21)

and
A- In Z.

(20.22)

Chap. 20

384

Information Theory

Invoking the remaining constraint Equation (20.18) we can derive an expres


sion from which we can determine J.L. Noting that
az

we have
1 az

z dJ.L

a
-- In Z.

i}J.L

(20.24)

Equation (20.21) gives the probabilities that maximize the uncertainty.


Thus

K(J.Lx +In Z).

(20.25)

This result can easily be generalized to include additional information associ


ated with the knowledge of the mean value of other variables.

20.5 THE CONNECTION TO STATISTICAL


THERMODYNAMICS
The randomness or

disorder of a system implies an uncertainty regarding

its state. Boltzmann described the entropy as a measure of the Hmissing'"


information.
The similarity between the development of the last section and the pre
vious chapters is striking. Consider a system of N distinguishable particles
obeying Boltzmann statistics. Assume that the quantum states are nondegen
erate. Then the thermodynamic probability is

U;=

N!
--

..

(20.26)

Sec. 20.5

The Connection to Statistical Thermodynamics

385

and the entropy is

=:;

N InN- N-

N1

J\j [lnN-

In

N;]

N1

Then

(20.27)
Imagine that the system is in perfect order, meaning that all the particles
are in the lowest quantum state. Then
N1

1 and

N3

N2
=

Nn

0,

and the sum in Equation (20.27) reduces to one term equal to zero. Hence the
entropy is zero for the perfectly ordered state.
A disordered system would be Jikely to be in any number of different

quantum states; the larger the number of states available to it, the greater the
disorder. If

N;

1 for N different states and Ni

0 for all other available

states, then
N

kN

kN In N

(20.28)

This function is positive and increases with increasing N.

If we associate

N;/ N with the probabilit; Pi that Ni particles are in the jth

quantum state, it follows from Equation (20.27) that


.

kN

Pi I n Pi

kN
=

H.

(20.29)

Chap. 20
...

386

Information Theory

e
th
d
an
s
ic
am
yn
od
m
er
th
al
ic
ist
at
st
of
Apart from a constant, the entropy
lly
fu
e
in
rm
te
de
to
pt
m
te
at
e
w
If
e.
m
sa
e
th
e
entropy of information theory ar
e
th
en
th
n,
io
at
rm
fo
in
al
rti
pa
ly
on
n
ve
gi
e
ar
t
bu
the macrostate of a system
k
lac
r
ou
of
re
su
ea
m
a
be
d
ul
wo
in
ga
d
ul
wo
expected amount of information we
is
it
,
ng
ssi
mi
is
on
ati
m
m
inf
the
e
us
ca
Be
.
tem
sys
e
th
of
te
e
sta
th
of knowledge of
the
,
ver
we
Ho
.
ure
rat
lite
the
in
y
rop
ent
neg
or
py
tro
en
ve
ati
neg
led
cal
s
sometime
constant of proportionality between Sand H in Equation (20.29) is positive.
The Boltzmann distribution for nondegenerate energy states is
e-e,/kT

N
j

--

'

where

If we substitute the distribution in Equation (20.27), we obtain


N 1
Smax =T Z

The internal energy is

=
.

Thus

Sm

,.
+ N k ln Z.

This expression has exactly the same form as Equation (20.25). Hence the
internal energy plays the role of a variable whose mean value is known.
"Partial information'' in this case is knowledge of the energy of the system.

20.6 INFORMATION THEORY AND THE LAWS OF


THERMODYNAMICS

We liave seen that the thermo-dynamic concept of entropy as defined by


Clausius is physically related to the information theory concept of entropy as
defined by Shannon. This characterization is clearly more than a mere analogy.

Sec. 20.7

Maxwell's Demon Exorcised

387

Accordingly, some in vesti g ators have attempted to show that the fundamental
equations of equilibri un1 thermodynamics can be de ri ved from infottnation
theory. In classical ther modynamics heat, temperature, and work are taken as
primary concepts, while energy and entropy are derivative ideas. The informa
tion theory treatment inverts the procedure. Heat is introduced without the
need to define it in terms of reservoirs, adiabatic walls, etc. Information theory

is free from all artificial constructs such as heat baths or Camot engines.The
zeroth, first, second, and third laws become consequences rather than

premises. The third law, for example, is true by definition since, for a perfectly
ordered state at absolute zero, there i s no missing information (S

0).

Infotittation theory is thus an alternative way of presenting the postu


lates of thermodynamics. Its elucidation as such is beyond the scope of this

text. Suffice it to say that its fa r rea ching consequences unify our understand
-

ing of thermal phenomena.

20.7 MAXWELL'S DEMON EXORCISED


How seriously Maxwell took his demon is hard to say. In a11y case, he neither

conducted nor pr omot ed experiments to test his hypothesis What is certain is


.

that the paradox posed by his demon botltered generations of physicists.


In the early 1900s it was noted that Bro\\ nian agitation of the trap door
'

(recall t hat it is massless and frictionless) would result in a random opening


and closing of the door, thereby rendering ineffective its long-term operation .
After the d evelop ment of quantum mechanics in the 1920s, it

\vas

suggested

that the uncertainty principle might play a role in the prob lem: later it \vas
shown that this would not be the case for heavy molecules at reasonably low
pressures.

In his seminal1929 paper on Maxwell's demon, Szilard observed that an


amount of energy at least as great as whatever amount of energy can be gained,
must necessarily have been ex1:.ended by the demon in his detection, inspec

tion, and routing activities. Demers and Brillouin (1944-51) called attention to

the fact that in an isolated enclosure it would be impossible for the demon to
see

the individual molecules.To make the molecule.s visible against the back

ground blackbody radiation, the demon 'Nould have to use a flashlight. The
entropy produced in the irreversible

ope ration

of the flashl ight would always

exceed the entropy destroyed by the demon's sorting procedure.Thus a real


demon could not produce a violation of the second law of thermodynamics.
Richard Feynman, looking at the demon as a mechanical system, wrote
"If \Ve assume that the specific heat of the den1on is not infinite, it must heat
up . .. Soon it is shaking from Brownian motion so much that it can't tell
.

whether it is coming or going, much less whether the molecules are coming or

going so it does not work."


,

Information Theory

Chap. 20

388

Information theory provides a new framework in which to view the

paradox. A knowledge of the momentum and position of the molecules co st


tutes the information needed to achieve a given entropy decrease. But thts

IS

counterbalanced by the entropy increase 11ecessary to acquire this informa


tion. Lehringer observes that about 1<Y4 bits of information are re quired to

reduce the entropy of 1 kilomole of gas by 1 calorie/K.


Recently.. Landauer* and Bennettt have interpreted Maxwelrs demon as
a computing automaton, subject to limitations imposed by the rules of data
processing and computational logic. Their view is that the gathering of infor
mation by the demon does not nece ss arily increase the entropy. But after the
demon has used the information to allow the passage of fast molecules.. he
must forget it if he is to continue doing his j(Jb. It can be shown that erasing the
inforr11ation raises the entropy more than any previous lowering that arises
from its use. The association of computation with thermodynamics suggests
that computational procedures might fruitfully be expressed in terms of the
behavior of some thermodynamic system.
In The Natltre ofThermodynan1ics B rid gma n states that If the Maxwell
..

demon had been invented vesterdav instead of in t he last centurv I b elieve he


.

"'

...

would not have caused as much consternation.- Perhaps not. but the language of statistical thermodynamics is beginning to permeate fields as diverse
science, complexity theory biology, and animal behavior.

as computer

Maxwell's demon may be exorcised, but the ideas that conjured him up live on .

;..:-:.. :--. :-:.:- - ;-.:


:
!.:: : . .: :. :
: :

:,..:.. -.. :?
: '..-..;,.
.
.::
...... . .
.. .: =" ::-: ..... ....., ..... . .. ..:v;, . .. .......... .. .. .
-....
'.
- "<" x: - .--.

_,

...

.. . ... .
..

..

..
. ..

. . . .. . .
. ... .. . . .. . .
.... .. ..
.

. .... ... .

.
.

. . ... ....
. . . . .. . .

. .. . .

. . -

PROBLEMS

20-1 Refe rri ng to thecompound experiment of Section 20.2 show explicitly that

=H

Il
--

:?

..

1
.,

4-

1
2 I
+-H
2
3' 3 .
-

20-2 Use the method of Lagrange multipliers to maximize H su bj ec t to the condition


n

p1

==

1. Show that Hm;r'( = K In

n.

i= 1

20-3 Make a table of the logarithn1s to thl! base 2 of the numbers from 1 to 10. In
adjoining columns list the logarithms to the base e and to the base 10 .

. .; * R. Land aue r computation: A Fundamental Physical \'ic\\.H A.fa.lue/15 De1non: Entropy.


Information. Computingt edited hy H.S. Leff and A.F. Rex. Princeton lJnivc:rsity Press. Princeton.
N
Jersey, 1990, pp. 260 ff
,

ew

C.H. Bennett The Thermodynamics

of Computation

A Revie\\

Ibid. pp. 213 ff


.

P. W. Bridgman. The Nature ofThermodynamic.ft, Peter Smith. Gloucester Mass.

1969.

Problems

389

Note that

20-4 Consider a loaded die. Let the probabilities of throwing the numbers 1, 2, 3, 4 , 5,'
and 6 be 0.1, 0.1, 0.1, 0.1, 0.1, and 0.5, respectively. Calculate the un certait1ty H in
bits for a throw of the die. What would H be if the die were "honest"?
205 A perfect shuffle of a deck of 52 playing cards is one in which all of the possible
orderings of the cards are equally probab le.
(a) How much information h1 bi ts is there in a perfect shuffle?

(b) HO\\' many times do you have to cut and interleave a deck in order to
achieve something approximating a perfect shuffle?

246.3 K. Assume the


''alphabef' of possible values as being multidimensional: e hundreds of
degrees being one dimensio n, the tens value a second d i nsion etc. Associated

206 Suppose that a measurement of ten1perature gives T

with each dimension is a probability. Reasonable values might be

PH(2)

l;

p10(4)

0.2;

p1 ( 6)

0.1; Po.t( 3)

0.1.

With these values, calculate the amount of information in the measurement

20-7 Consider a substitution cryptogram in which, for each letter of the alphabet,
some other letter is substituted. The number of possible keys is 26!. The cryp
togram can be viewed as a compound experiment consisting of two parts: x, the
communication of the clear text, andy, the choice of a key from one of the 26!
possibiliti es. Thus the total information associated with the compound e"peri
ment is

H ( x)

log 26! bits. A substitution cryptogram of 40 letters can usually

be solved. What does this imply about the maximum information in bits per let
ter of an English messag e ?

20-8 Imagine that system 1 has probability p}1> of being forced into a state j and sys
tem 2 has probability p2> of being forced h1to state k. Then

H0>

K pf0 In

pfl) H<2>

K pk<2> In P
k

2>
<
c .

Each state of the composite system consisting of systems 1 and 2 can then be
labeled by the pair of numbers j, k. Let the probability that the composite sys
tem is found in this state be P;k Then

K Pjlc In Ppc
j
k

If the two systems are only weakly interacting, so that they are statistically inde
pendent, then Pik

p}1> Pk(2). In this case show that


H

flO>

H < 2>

209 P.T. Landsberg (Thermodynamics and Statistical Mechanics, Dover, New York,
1990) draws a distinction between entropy and disorder. For a system of
tinguishable states, he defines the disorder D(n) as

n dis

Chap. 20

390

D(n)
Here

5'(n) is the

entropy and

S(n)

.:

ln

Information Theory

is a constant. The rate of ch ange of disorder can

be found by diff ere ntiating this equation with respect to the time. Show that

.
D(n)
.

S( n)

S (n)

n-- l
-JD(n).

ln

Here the dots denote time derivatives. Since the second term in the brackets can
conce iv ably be greater than the first tem1 the result suggests that it is possible

to have decreasing disorder even though the entropy increases. The expression
has a possible application to the growth of biological systems.

A. I PARTIAL DERIVATIVES

Consider a function of three variables

f (x, y, z)

0.

(A.l)

Since only two variables are independent, we can write

y(x, z). Then


dx

ax
dy
ay z

ax
az

dz,

x(y, z)

and

(A.2)

and

dy

ay
=

ax ..

dx

....

ay
dZ

dz.

(A.3)

The subscripts denote the variables that are held constant in the differentia
tion and are often omitted. Substituting Equation (A.3) in Equation (A.2), we
obt a n

391

Review of Partial Differentiation

Appendix A .

392
ax

dx

If we choose

..

ay

av

.. ax \

ay

dx +
ax :

ay

az

ax
az

dz.

(A.4)

and z as the independent variables, then Equation (A.4) holds

for all values of dx and dz. Thus, if dz

0 and dx * 0. we have
1

iJ.x

av
..

(A.5)

dV

..

"'

...

a.x

...

This expression is known as the reciprocal relation.


If dx

0 and dz :f:. 0, we get


ax

dX
-

--

av

..

..

...

az

az

"

(A.6)

The reciproca-l relation gives


1
a..
.

ri.r

\'

Substituting this in Equation (A.6) yields


ax
=

rlV ..
..

Equation

-1.

(A.7)

...

(A.7) is the cyclical rule, or c,vclical relation.

A function

lt

of three variables x y, and z can be written as a function of

only two variables when they are related through Equation (A.l), that is.. when
only two variables are independent. Thus
li = ll ( X,

y)

(A.8)

Alternatively, we can write


.r =

(A.9)

.X(ll, v) .

Then

dx

rlX
-au

du
\'

ax
+

av
..

dv.
.,

(A.lO)

Sec. A.l

Exact and Inexact Differentials

193

If we divide Equation (A.lO) by dz while holding u constant, we obtain

ax
az

ay

ax
ay

dZ

(A.ll)

This is the chain rule of differentiation.

A.l EXACT AND INEXACT DIFFERENTIALS


Consider the function z
b

z(x). The fundamental theorem of integral C<'llculus is


b

(A.12)

If the integration is taken around the closed path C (a to b and b to


b

dz

dz

dz

+
b

dz -

dz

M(x, y)dx

0.

(A.l3)

This can be generalized to two independent variables. Let

dz

a), we have

z(x, y). Then

N(x, y)dy.

(A.l4)

We wish to determine the condition under which the integral of dz around a


closed path is zero in this case.
Consider the integral of aNfiJx over a rectangle A in the x-y plane. From
Figure A.l, the integral is

aN
-dxdy
ax
A

=
c

iJ
N
-dxdy
ax

[N(b, y) - N(a, y)]dy.

=
c

y
d

---- ....

--

----

---

A.l Contour C
enclosing the rectangular
area A.

----

--- ----

I
I
I
I
a

l
I
I
\
b

(A.15)

Review of Partial Differentiation

Appendix A

394

N(x, y)d)'

We now evaluate theintegral

around the path C in a counter

clockwise direction:

N(x, y)dy

N(b, y)tiy

N(a, y)dy

+
d

[N(b,y)- N(a,y)]dy.

(A.l6}

The contribution along the horizontal segments of the .Path are zero since
dy = 0 there. Combining Equations (A.15} and (A.l6) gives

aN
-dxdy
A

dX

Ndy.

(A.17)

Similarly, it can be shown that

aM
-dxdy
ay

Mdx.

(A.l8)

Adding these two equations we obtain the result

(Mdx

N dy)

aN
aM
- dxdy.
ax
ay

d., -

.(.

(A.l9)

It follows that the condition under which the integral of dz around a closed
path is zero is

aM

aN
ax

iJv
"'

(A.20)

When this condition. is satisfied, the differential dz is said to be


Note also that

d -

a;.
. d.,'(

(A.21)

and
.

av a.r

--

.,

ayiJx

exact.

a-z
-

axay

az

ax ay

Sec. A2

Exact and Inexact Differentials

395

Identifying the partial derivatives in Equation


obtain Equation (A.20).

(A.21) with M and N, we again

dz is exact, we wish to be able to integrate it to obtain


z. If M = M(x) and N N(>'), the problem is trivial:
If the differential

M(;:)dx

z =

N(y)dy.

(A.22)

However, this will not be the case in general. We define

Mdx,

u
y

the partial integral of .M \Vith y fixed. That the function N

- iJufay depends on

y only can be seen as follows:


au
a
Nay
ax

aN
ax

aN

a au
ax ay

d.X

a a
Mdx
ay ax y

aN aM
= 0.
ay
ax

The last step is a consequence of our assumption that dz is exact.


As a trial solution, we take

z=u+

LV-

au
dy
ay
au
dv
Nay

.Y

Mdx + g(y).

(A.23)

This gives us a recipe for finding z: carry out the integration indicated in
Equation (A.23), take the differential dz including the tertn

g'(>)dy, and com


pare the result with the original differential to find g'. A final integration gives g.
As an example, consider the exact differential
dz

(x + y)dx + (x + 2y)dy.

(A.24)

Review of Partial Differentiation

Appendix A

396
Then

(x + y)dJ. + g(.Y)

z =

So

dz

:cdx + vt/:( + xtly

+.

..

g'(y )d)'

(x + .v)dx + (x + g'(}')jdv.

Comparing this with the original differential, we see that

g(y)

2
y

g' (y)

2y, so that

C. Therefore
"'

x-

==

+ .r v + .y- +

.;

(A.25)

C.

Alternatively.. we could have written

.
"

Ntl_v

(A.26)

h(x)

which leads to the satne result.


What if dz is inexact? Consider the example

dz
which doesn't satisfy Equation
that

=:

vdx

(A.20)

..

xdv_

the criterion for exactness.lo indicate

dz is not exact, we write

dz

vtlx
-

.rdv.
.

This cannot be integrated. Ho\\'ever, we 11otice that if we multiply the equation


by

2
1/y

..

we obtain an exact differential:

vdx

dz
-

.,
-

.,

yThe term

1/ v2
..

is known as an

XlfV
.,

..,

X
-

v
.

...

i11tegrating .factor. It can be proved that it is

always possible to find an integrating factor for an inexact differential that is a


function of two independent variables. and hence to obtain a differential that
is integrable. This fact is of great importance in thermodynamics.

Sec. A.2

Exact and Inexact Differentials

397

An integrating factor is not unique; for a given inexact differential, many


integrating factors can often be found. The following are all integrating factors
for the differential ydx - xdy:

A formal method exists for fi nding an integrating factor J.L. Let


dw

ILliZ

tMdx + p,Ndy.

(A.27)

If dw s exact, then

or
iJt
oM
-M + p,
ay
ay

iJL
=

ax

N + p,

iJN
ax

In general, this cannot be solved for J.L However, suppose that J.L is
of x only. Then <Jp,/ ay

function

0, and
aM

J.L ay

dp.
dx

J.L

iJN
ax'

or
d t
-

aM

1
=

aN
-

ay

a.:r

dx.

Integrating, and setting the constant of integration equal to zero, we obtain

p,

exp

iJkl

!.J

ay

iJN
ax

dx .

(A.28)

Note that if the differential is exact to begin with, the integral is zero and

1.
In pr actice the formal integration of Equation (A.28) is unnecessary, as
we can see from the following example. Let

IL

dz

sinydx

cosydy.

Review of Partial Differentiation

Appendix A

398
Then

For exactness
a
-(J,siny)
av

iJ
=

ax

(tcosy).

or
rl
ay
If JL

JL(X), iJL/iJy

stny + J.L cosy

= 0, and

diL/JL

==

ax

d:c, or p,

cosy.

=ex.

Thus

which is exact. If we choose to select J-L = Jl.(y), then J.L = 1/sin..v and

dw

d.t + C<)t Vd}'


..

Thus we have obtained two integrating factors; usuaiJy one is easier to calcu
late than the other.
From this analysis it is evident that the differential

dz

Md:c + Ndy is

exact if the equivalent statements hold:

1.

aM
-

aN
=

ox

dV
"'

2.

dz

(A.29)

dz is independent of the path.

3.
a

This means that if a problem can be formulated in terms of exact differentials,


or differentials made exact through the use of an integrating factor. then the
path of integration between the end points is arbitrary.. and we are free to
select one that leads to a simple evaluation.
In thern1odynamics all state variables are, by definition .. exact differen
tials whose integrals are path-independent. However, there exist certain ther-

Sec. A2

Exact and lnexaa Differentials

399

.modynamic quantities such as the differentials of work and heat wh


ose inte
grals depend on the path of integration. Thus the search for appropria
te inte
grating factors that allow us to express fundamental relationships in
tetrns of
variables of state is an importa1.1t goal of the mathematical theory.

PROBLEMS

Al Test the following differentials for exactness. For the cases in which the differen
tial is exact find the function z(x, y).

(a) dz
(b) dz
(c) dz
A-2

=
=

2x lnydx

(x2jy)dy.

(y - l)d x + (x - 3)dy.
(2y3- 3x)dx - 4xydy.

(a) Show that dz ydx + (x + 2y)dy is exact and int egrat e it to find z(x, y).
(b) Show that az xdx + (x + 2 y)dy is inexact.
(c) Integrate the two previous differentials counterclockwise around the trian=

gle whose vertices are the three points (0,0), (1,1), and (0,1) .

A-3 Consider the differential dz


(2xy- 3)dx
(a) Show that dz is an exact differential.
=

xdy.
.,

z (x, y).

(b)

Find z

(c)

Referring to Figure A.2, evaluate the line integral

dz. from (1,0) to (0,1) along

Path A: lines parallel to the axes;


Path B: circular arc;
Path C: straight line diagonal.

(0. l)

( 1 1)

8
c
Paths of
integration for Problem A-3.

Flgure A.l

(0,0)

(1. 0)

Review of Partial Differentiation

Appendix A

400

A4 In the following cases show that liz is not an exact differential. Find an integrat
ing factor such that dw = JL71Z is an exact differential and check that t his is so.
(a) i1z
2y dx + 3x dy.
=

(b) az
(c) llz

y dx + (2x
(x4 + y4)dx

y2)dy.

xy3dy.

A-S Find an integrating factor p.. and integrate du;


llz

(ycos3x

1 )dx

,.uJz, where

sinx cos2x dy.

i.
.

'

. . :.
.. ... -::

.
.
.
.
.
.
.
.
'
.
: . . : :. - .
, ,
. ... . :- .
. - ; . ..
...
..
;- ' .

' .

.. : :
'

:: ::::

'

'

-: . >,
: . ".
-.
.:

'.

'
. : :
. :::

.
.

: . .

-..

.
' -

- -

.
.

.
.
.
- - . . . ' '.

: ;
- :

..

'

'

_.

'

..
.

The definition ofn!, the factorial of a positive integern, is


n! = n(n- l)(n-

2) 1.

(B.l)

From this and the fact that


(n + 1)! = (n + l)n!,

(B.2)

it follows that 1 ! = 0! = 1.
We are interested in the natural logarithm of n! when n is a very large
number. From Equation

(B.l) .we see that


n

In tz! = In 1 + In 2 + In 3 +
'

Equation

(B.3)

+ Inn =

Ink.

(8.3)

k=t

is clearly the area under the step curve shown by the dashed

lines between n = 1 and n = n in Figure B.l. The rectangles all have unit
width; the height of the first is In 2, that of the second is In 3, etc. This area is
approximately equal to the area under the smooth curve y = Inn between the
limits 1 andn ifn is large. For small values ofn the area under the step curve is
appreciably different from and exceeds that of the smooth curve. As

401

402

Appendix 8

Stirling's Approximation

=Inn
"

v
"'

ln6 ......
In 5
In 4

r--I
1

ln3 .....
In 2 .....

I
I
I
I
I
I
I
I
I

r--

1
I
-

I
I
I
I
I
I
I

--

I
I

I
I
I

I
I

I
I

..___
___
...__
_
""""'_
_.__
_
........_

__._

Figure B.l

Graph of In n versus

___

n.

increases, the smooth curve flattens out and the difference becomes negligibly
small. Thus we can approximate Equation {8.3) by

Inn!

= n Inn - n + 1.

Inn dn = (n Inn - n)

Sincen is assumed large, we can neglect unity and get


In n!

=' n

Inn - n.

(B.4)

This is Stirling's approximation, valid for large n. For n = 10, Equation (B.4)
gives a value low by 1 3.8 percent and. for

= 30, a value low by 3.5 percent.

As an example, we can use Stirling's formula to estimate 52!, the number


of possible rearrangements of cards in a standard deck of playing cards. Here
In 52! :::; 52 ln 52 - 52 = 153.
Then
log1052! = (0.4343)ln 52!= (0.4343)(1 53 ) = 66,
so

:.

/:

'

, ...

:':
.
'

'
.
.
'
-.
. . .. '
- :.
. _ '. . .
. . .

:
:
:: :

<

'

. ..
-

..:
. >-::.

.
: .
_

'

..
.

...:
.

'

: .
.

: -
_ . ': -'
. ;)
: ' \:
. : :_ . _ .
'

. ..

I
ann

The expression defining Boltzmann statistics is

(C.l)

We wish to show that

N
J

Ki

ex

e!Je'

(C.2)

To simplify the argument we assume no degeneracy for any energy level, so


that gi

1 for allj. Then Equations


w=

(C.l) and (C.2) become


N!

. . A..I '
1Yn

--

-- --

Nl ' . . . N-'

(C.3)

and

(C.4)
403

404

Appendix C

Alternative Approach To Finding the Maxweii-Boltzman Distribution

We also assume, initially at least, that the energy levels are equally spaced. We
wish to make a minimal change in the given configuration while holding Nand

U constant.* Thus we transfer one particle from level j to level k (gaining


energy) and one particle from leveJ I to level k (losing the same amount of
energy) (Figure C.l). Then the new (primed) configuration is:

Note that we still have

The new value of

N1

and U

N_,e1.

is then

as given by Equation (C.l), is the predominant (equilibrium) configura


tion for which ln w reaches a maximum value, then the relatively infinitesimal
If

w,

change in configuration described previously produces essentially no change


tv'. Equating Equations (C.3) and (C.5) .. we obtain
in w. Thus we write w
=

N!
;! Nk! N,!

-------

Nl !

Nn!

-------

N, !

( N,

N!
I )_! ( Nc + 2) ! ( N,

------

I) !

Nn! '

so

or

J!

N,!
(N1-l)! (fVt-1)!

-------

-------

:=

e, ----

'

1
Ek --1
F.i t-----

Figure C.l

levels.

*Here N
system.

Adjacent energy

N1 is the total number of particles and U

:L ,vie; is the total energy of the

405

or

But N1c >> 2 or 1, so

to a very good approximation. This can be written

(C.6)

If we bad chosen i,j, k instead of j, k, l, we would have obtained


NI

so we evidently have a continuing geometric series

(C.7)
Now, if the levels are not equally spaced, but are given by

(s1

sk)

(ek - ei)

p
=

q'

then it can easily be shown that Equation (C.6) is replaced by

so that

or

406

Appendix C

Alternative Approach To Finding the Maxweii-Boltzman Distribution

Again, we would have found equalities like this if we had used i,j, k instead of
j, k, I, etc. Thus the function must be a constant

call it - {3. Finally, we have

N-1

or

or

(C.8)
which is the result we have been seeking. The Boltzmann distribution and the
Maxwell-Boltzmann distribution are identical.

..

- - . :.
. .:
:
..
.
.

.
.-

.
_, ,

+co
_2
r
e - a dx

11'

1/2

-oo

(D.l)

(0.2)
-oo

+'lO
x2n+le-ax2dx

0.

(D.3)

-oo

00

xne-axdx
0

n!
.
.
l
+
an

(0.4)

(For Equations (D.2), (0.3), and (D.4), n


oo

s-ld
X
X

e 0

s >

0.

0, 1, 2, 3 .... )

(0. 5)

407

408

Appendix D

t
J;xn- e-xdx.

Here f(n) is the gamma function: r(n)

f(n

1)- nf(n)

Various Integrals

11!,

wherenisapositiveinteger.

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

:X

l(a) is the zeta function: ' (a)

-a

n=l

l(3/2) = 2.612.

,(1) = oo.
(( 2 )

((3)

7T2/6

l(5/2). = 1.341.

1.645.

1.202.

((7/2) = 1.127.

((4) = 1T4/90 =-1.082.

((9/2) = 1.055.

3dX

--

---

2
)
1
(ex

(0.6)

(D.7)

15

=-

(0.8)

1.

(D.9)

---

eXdx
-x(ex+l) ..
X

.,

---

---

(0 .10)

409

Note that

smce

an integer,

---

------

and therefore the integrand is seen to be an even function of .t.


In evaluating integrals

ated with the Maxwell-Boltztnann speed dis-

tribution, consider the integral

r
e-a dx,

> 0.

The equation could equally well be written

so that

- (il+f>
e a
dxdy.

Changing the integration variables to polar coordinates, we have

dxdy

rdrd6.

The region of integration is the first quadrant. Thus


11/2
/2

r
e-a rdrd8

MID

0
=

7T
-

il
e-a rdr

1T

2 0

2a

(D.ll)

410

Appendix 0

Various Integrals

-- 'Tr e -ar
_

4a

1T

Taking the square root, we obtain

I=

(0.12)

Integrals of the form

X-ne- uxdx,
'

.,

t 2 3
'

'

'

can be found by differentiating Equation (D.12) with respect to a:

dl
-

da

x2e-'xdx

t)

v; a-.;.:.

J'"'

and

da--

()

Successive differentiation leads to the general formula Equation


y

For odd powers of x

(0.2).

in the algebraic term of the integrand, we can set

x2 For example,

xe

J=

ax

dx

e-aydv
..

411

1
=

2a

e-

00

1
-

2a.
Then

dJ
-

da
With successive differentiations, we can generate the general for1nula

(0.13)

CLASSICAL THERMODYNAMICS
C.J. Adkins, Equilibrium Thermodynamics, 3rd edition, Cambridge University Press,
Cambridge, 1983.
W.P. Allis and M.A. Herlin, Thermodynamics and Statistical Mechanics, McGraw-Hill,

New York, 1952.


M. Bailyn, A Survey ofThermodynamics, American Institute of Physics, New York, 1994.
R.P. Bauman, Modem Thermodynamics with Statistical Mechanics, Macntillan, New
York, 1992.
P.W. Bridgman, The Nature o[TI1ermodynamics, Harper, New York, 1961.
H.B. Callen, Thermodynamics, Wiley, New York, 1960.
G. Carrington, Basic Thermodynamics, Oxford University Press, Oxford.1994.

E. Fermi, Notes on Thermodynamics and Statistics, University of Chic.ago Press,


Chicago, 1966.

E. Ferrni, Thermodynatnics, Dover, Nw )ork, 1956.


C.B.P. Finn, Thermal Physics, 2nd edition, Chapman and Hall, London, 1993.
W. Greiner, L. Neise, H. Stocker, Thermodynamics and Statistical klechanics, Springer
Verlag, New York, 1995.

W. Kauzmann, Thermodynamics and Statistics: lVith Applications to Gases, Benjamin,


New York, 1967.
413

414

Bibliography

C. Kittel and H. Kroemer. Thermal Physics, 2nd edition, Freeman, San Francisco. 1980.
R. Kubo, Tl1ermodvnamics, John Wilev and Sons, New York, 1960.

P.T. Landsberg. T hermodynamics and Statistical Meclzanics, Dover, New York, 1990.
D.F. Lawden, Principles ofThermodynamics, Wiley, New York, 1987.
M.C. Martin, Elements of Thermodynamics, Prentice-Hall. Englewood Cliffs, New
Jersey. 1986.
P.M. Morse, Thermal Physics, 2nd edition, Benjamin. New York. 1969 ..
A.P. Pippard, The Elements ofClassical Thermodynamics, Cambridge University Press,
Cambridge, 1987.

M. Planck, Treatise on Thermodynamics, 3rd edition, Dover, New York, 1 945.


H. Reiss, Methods ofThermodynatnics, Dover .. New York, 1996.
.

B.N. Roy, Principles of Moden1 Thermodynamics, Institute of Physics, London, 1995.


F.W. Sears and G.L. Salinger, Thermodynanzics, Kinetic Theory, and Statistical
Thermodynamics, 3rd edition, Addison-Wesley, Reading, Massachusetts. 1975.

A. Sommerfeld.. Thermodynatnics and Statistical Mechanics, Academic Pres New


York 1964.
M. Sprackling, Heat and Thermodynamics, MacMillan, London, 1993.
M. Sprackling, Thermal Physics, American Institute of Phy sics, New York .. l991.
M.V. Sussman

Elementary General Thermodynamics, Addison-Wesley, Reading.

..

Massachusetts, 1972.

H. C. Van Ness, Understanding T h ermodynam ics, Dover, New York .. 19 69 .


M.W. Zemansky and R.H. Dittman. Heat and Thermod vnamics, 7th edition .. McGraw
__

Hill, New York 1997.

: .

::

.:

'

': .

. :

KINETIC THEORY OF GASES


L. Boltzmann, Lectures <>n Gas Theory, Dover, Ne\\' York. 1995.
C. H. Collie, Kinetic Theory and Entropy, Longman London, 1982.

W. Kauzmann, Kinetic Theory

of Gases, W.A. Ben.jamin, New York, 1966.

STATISTICAL MECHANICS
R. Bowley and M. Sanchez. Introductory Statistical Mechanics, Clarendon Press,
Oxford, 1996.

R.H. Fowler and E.A. G uggenhei m Statistical Tht,rmodynamics, Cambridge University


.

Press, Cambridge, 1952.


C. Garrod, Statistical Mechanics and Thermodynamics. Oxford University Pres
Oxford, 1995.
-

T. Grenault Statistical Physics 2nd edition, Chapman and Hall Londo 1995.
.

Bibr

415

T.L. Hill, A n Introduction to Statistical Thermodynamics, D


over, New York, 1986 .
K . Huang, Statistical Mechanics, 2nd edition, Wiley, New Yor
k, 1987.

E.L. Knuth, Introduction to Statistical Thennodynamics, M


cGraw-HilL New York, 1966.
R. Kubo Statistica/ Mechanics, North-Holland,Amster
dam, 1965.
,

L.D. Landau and E.M. Lifschitz, Statistical Physics, A


ddison- Wesley, Reading,
Massachusetts, 1958.

R .B. Lindsay, Introduction to Physical Statistics, Wile


y, New York, 1941.
L.K. Nash, Elements of Statistical Thermodynamics,
2nd edition, Addison-Wesley
Reading, Massachusetts, 1972.

'

R .K. Pathria, Statistical Mechanics, Pergamon Press, Oxford, 1972.


M. Plischke and B. Bergerson, Equilibrium Statistical Physics, 2nd edition, World
Scientific, Singapore, 1994.

F. Reif, Fundamentals of Statistical and Thermal Physics, McGraw-Hill, New York,


1965.
W .G.V. Rosser, An Introduction to Statistical Physics, 2nd edition, Horwood,
Chichester, UK,l985.
K. Stowe, Introduction to Statistical Mechanics and Thermodynamics, Wiley, New York,
1984.
R.C. Tolman, Principles of Statistical Mechanics, Oxford University Press, Oxford, l938.
G.H. Wannier, Statistical Physics, Dover, New York, 1966.

::;:;:;:;: ;.;.;:;:;:: :::;: ;:;:;:;:,:;.; ::;:;:,:; :;:;:.:: ::::;:; ::;:;:;:;:;: ; ::


;:
::;:
;.:;:;:.:;.;:;::::;:;:: ::;:;:;:;.;:;:;:;:;:::
:;:
: ::: : ;:;:;:;:.: :::; :;: ;..
;:::;::;::;: : ;:;:;:;:;:;:;:::;;: :.:;:;:;:;:;.;:;;

SPECIAL TOPICS
P. Coveney and R. Highfield, The Arrow of Time, W.H. Allen, London, 1990.
J.S. Dugdale, Entropy and Its Physical Meaning, Taylor and Francis, London, 1996.
J.D. Fast, Entropy, McGraw-Hill, New York, 1962.
M . Goldstein and I.F. Goldstein, The Refrigerator
Press, Cambridge, Massachusetts, 1993.

d the Universe, Harvard University

D.L. Goodstein, States of Matter, Dover, New York, 1985.


H.S. Leff and A.F. Rex, editors, Maxwell's Demon: Entropy, Information, Computing,
Princeton University Press, Prinr.eton, New Jersey, 1990.
M. Mott-Smith, The Concept of Heat and Its Workings Simply Explained, Dover, New
York, 1962 .
J.R. Pierce, An Introduction to Information Theory: Symbols, Signals and Noise, 2nd
edition, Dover, New York, 1980.
G. Raisbeck, Information Theory, M.I.T. Press, Cambridge, M3ssachusetts, 1964.
C. E. Shannon and W. Weaver, The Mathenzatical Theory of Communication, University
of Illinois Press, Urbana, Illinois, 1963.
M.W. Zemansky, Temperatures Very Low and Very High, Dover, New York, 19t

Chapter 1

11

(a)

closed;

(b)

1-2

(a)
(b)
(c)
(d)
(e)
(f)

quasistatic, reversible, isobaric.

isolated;

(c)

open.

quasistatic, irreversible, isotherntal.


irreversible, adiabatic (if fast enough so no heat is exchanged).
irreversible, isochoric.
reversible, isobaric, isothertttal, isochoric.
irreversible, adiabatic.

lS

260.

1-6

348.65 K; 75.50C.

17

4.00 K.

1-8

a=

1-10

574.

111

-195.80C;- 320.44F; 139.23 R.

0.5 mV/C;b

-lo-J mV/(,C)2; T =

430.3C.

Chapter 2
10-4kg.

1-1

1.5

Z.Z

(a)

0.308 kilomole;

(b)

9.86 kg;

(c)

3.96

l<t Pa;

(d)

0.277 kilomole.

417

Answers to Selected Problems

418
2-3

(c) 800K.
2.67.

2-7
Chapter

3-1

3.69.
3
-5.74

7
10 J.

3-Z

1.91

3-3

(b) T1/4;
-

lfr' J.

(c)

(3/8) nRT1

92 1.
.

3-5

(c) 4.3 0 x tonJ.

(b) 4.26 x 106J;


(b) 3.12

10 J.

3-7

(a) 3. 45

106 J;

3-8

(a ) 601;

(b) -70J;

3-9

(a) -3.12

3-to

9.71 x

3-11

(a) A= -1.6 x
(d) 3.6 X 1 0 J.

Chapter

(b) 3.45 X 1 06 J.
(c) 50J,lOJ.

(b) -4.32 X 1(} J:

105 J;

(c) 150K;

(d) 1.25

105 Pa.

to-2 J.
to-)m1Pa 1.8

2.6mJ;

(b) 270K;

(c) 6 x l04J:

4-1

(a) 59.2 J kilomole -I K


(c) 37301 kilomo le 1

4-S

l0-1 approximately.

4-6

(a) 8.3 1 x 1 06 J;

4-7

Q = n[aT

--

t.

7400 J kilomole

1 K -1;

(b) 1.85 x lfr.:; J;

..

4-11

(a) 1 m, flU
4.48 X 1 Q- J.

4-13

3.66

106 J.

4-14

4.28

10' J.

4-lS

696 m.

(b) 11.6 x 1otl J.

(3/2)bT2 - cj2T].
=

0,

5 62
.

x lOi J

(b) 1.22 m, 357K. -4.48 X 10=' J,

Chapter 5

5-3

(a) 11 = aj cv2; (b) h u0 + c,.T - 2ajv


(d) K = [RTv/(v- b)2- 2afv2]-1

S-6

(a)

5-7

340 watts.

-900kcal

(b) -1600kcal

RTvj (v

(c) 300kcaL-400kcal;

b);
(d) 25/o,3.

It is more efficient to lower the temperature of the cold reservoir.

Answers to Selected Problems

19

5-9

71 K, 227 K.

5-10

(a) 2.49 X HfJ, 10.70


(b) 2.68 x 10" J; (c)

5-11

200J, -1200J.

5-13

(a) T,j(T2

5-14

HfJ, -2.49

20.3%.

106J, -8.02

HfJ;

1)); (b) 14.6.

(b) 55.3%.

Chapter6
6-1

294 JK-.

6-2

(a) 1074JK-1; (b) 1220JK-1; (c) -6060JK-1

6-3

127JK-1
0, 0.893JK-1

6-5

(a) 19 K; (b)

6-9

(a) 5.76

6-10

(b) aT

0.27 JK-1, 0.27 JK-1

10"1JK-1,0;

(b) 5.76

1()-JK-1,5.76

lfr'JK-1

+ bT3/3.

Chapter?
71

(a) -165J; (b) 6.03 X 106J; (c) 2.21

7-l

660JK-1

7-3

(a) 6.3JK-l; (b) 1.38JK-1; (c)

7-4

(a) 6.93JK-1, -5.0JK-1; (b) -6.93 JK-1; 10.0 JK-1


(c) 1.93JK-1,3.07JK-1

75

(a) 9.13

7-7

(a) -4.9

7-9

1.7 JK-1

711

Cv

7-12

664Jkilomole-1K-1

7-13

(b) 0.4/P.

7-16

331 ms-1

7-17

(b) 12.5Jkg-1K-1

X
X

1Q-1JK-1;
lfr JK-1;

In T + R ln( v

b)

o;

lfrlJK-1

(b) 4.9

11 JK-1;

s0

8-1

The change is greaterfor an ideal gas.

8-3

-1.62
nKT +

10'J.

BP

CP2

+ DJ>-'.

let'JK-1

(d) 5.73

(b) 9.13

ChapterS

1()-1 JK-1;

(c) 16.4JK-1

(e) 0.

410

Answers to Selected Problems

8-8

(a) - RT

8-13

(a) 200 K;

v.,-

ln

+ a

v,-

l
-

v1

(c) 2R.

---

v2

(b)

v1

1
-

Chapter 9

93

(b) - 1 50

9-5

0.

9-6

2.

9-8

(a)

107 15

300 K, 2.5 atm ;

10

JK-1

(b) -1.56

(c)

lOfl J:

5.19

lOJ JK- 1

Chapter 10

10-4

2.38

10-8 J kilomole -I K- 1 23.8 J kilomole- 1 K _,_

105

(b) a(T- T0)

Chapter 11

111
11-2
113
11-4

(b/3)(T3-

T03).

2.45

(c)

(a) Nj

2
v0 ;

1.14

1Q28 m - 2s - 1

1.25

101J m- 3.

-1
ms ,

112

(b)

517

ms

v0,

-1

v0;

v0;

(d) 1.41'V.

215 ms -1

11-S

0.129 eV, 2.1 3

11-6

0.682

11-7

(c)

118

394 ms

119

(b) 1.50.

11-10

v2e, zero.

105 ms-1 about 0.2/o.

(kT/m) 'I 2.31 (kT/nt) '12

0.427.
-

1,

445 ms-

1,

483 ms- 1: 2278 ms

2571

1111 3.86

10-8m_ 83.9,

11-12 2.00

1013 m_ one collision everv 20 centuries.

11-13 4.0

6.16

ms- 1 2790 ms- 1


_

109 s -1
"'

to-ss.

11-14 (a) 14 Pa;

(b) 975 s.

1115 0.485 P.

112
2
1
)
m
m
11-16 ( c2/c1 ) ( /
11-17 About 10 days.
1118 2.80

10 -tom.

1119 (a) 1.10

10-5 Pa

7.39

10-3 Jm

1s-'K-1 8.53

10-6

2
m s -1

Answers

to

Selected Problems

421

Chapter 12

12-1

(a)

1.13

1015;

(b)

1.26

101 4 ;

(c)

0.112.

12-2

(a)

1.27

10:;

(b)

1.01

1029;

(c)

0.0796.

U-3

49.

1?-6

(b) 84; (c)

12-7

(c)

336,798;

1'-8

(a)

1211 8.66
12-U 5.6

2.73, 1.64, 0.908, 0.454, 0.195,0.065, 0.013.

(d)

In 3.
1 07

1.33, 1.00, 0.71,0.48,0.29, 0.14, 0.05.

letc;2, lcY6

Chapter 13

13-1

(a)

4,4;

13-l

(a)

2002;

13-3

(a)

1.33,1.00, 0.78, 0.44, 0.22, 0.11, 0.11;

13-9

(a)

25.3,0.0211;

13-10

(b) lQ-c;.

13-11

(a)

(b) 2ab,64o/o.
(b)

252.

(b)

(b)

1, 1, 1, 1, 0, 0, 0.

0.980, 0.0202.

2.78, 1:1.10:0.675,265

k; (b)

587 K.

Chape
t r 14

14-1

(a) U/T + Nk In Z, -NkT In Z.

14-3

(b)

14-4

1.44

14-5

- 0.29eV.

14-6

(a)
(d)

14-7

2.66

108

l(}c; JK-1

3.74

106J,0.0388e\l;

2.22

10-8

(b) 6.06

10"1;

(c)

-0.417eV;

4NkT,NkTfV,Nk[5 + ln(aVT4,f.V)].

Chapter 15

ts-1

(a)

0.982, 0.0180, 0.00033, 0.632, 0.232, 0.085.

15-3

(a)

865,117,16;

15-4

(b) 6Nk{8roJT)exp( -28roJT) ( 1 + T/28rot)

15-6

(a)

7.53

to-ll

15-8

(a)

5.67

106 J;

15-10 2.48

(b)

m;

(b)

656 k 6.

(b)

11.2

-4.34

104 J kilomole--t K -t.

x
X

10-11
107 J;

m.

(c)

1.79

105 JK-1

Answers to Selected Problems

422

Chapter 16
16-6

1.11 x 1()l J kilomole

16-7

(b)

K-t

(Einstein) 2.68

1oJ J

kilomole -t K -I (Debye).

293 K.

Chapter 17
-1
1016 s 9. 24

4.13 X

10-24 Nm, 9.24 X 10-24 Am2

172

(d)

173

2.12 x to-6

17-4

2.27 X 10 4 evT-1, 2.90 X 10-4 eVT-1

175

1.87

17-6

(a)

0.212 JT-1:

1712

(a)

7.87 ?<

17-13

507 Am-1

104 JKT-2

(b)

92.7 JT--1

l<P Am-1: (b)

1550T.

Chapter 18
X

10-5 J.

18-1

(8) 3.89

18- 2

(a)

18-3

(a) 6040 K; (b) 4.65 x tc0 megawatts.

18-4

(a)

10.6 mic rons ;

(b)

5800 K.

2.82 kT/h, smaller by a factor of 1.76;

(b)

1.58 X 1011Hz.

1
1
x
1
0 1
(b) photons/molecules= .7
-

1014:

(a) 4.1 x

18-S

'
'

Approximately 107

18-6
18-11

(a)

2.032:

18-12

(a)

-4.67 x 10-20 J. -0.292 eV.

(b)

1.092:

(c) 0.685;

(d)

0.327;

(e)

0.176.
.

18-13

(a)

1.0 1 X 1026 m-3;

1814

(a)

8.59 X 1 0-6 K;

(b) 96/o.
(c)

(b) 9.96 X 1 0(,;

242 .

Chapter 19
191

(a)

0.134

(b)

19-2

(a)

0.070

K; (b)

0.116;

(c)

0.0997;

(d)

-12.7, 3.28 X 105;

0.0735.

(c) 6.80 X 10-7

19-6

0.648 J kilomole-J K -1 Equation ( 19.20) gives a value 12 /o lower than the

- experimental value.
1.88 eV;

(c) 1.89 x 1 oJ 1

197

(a)

19-8

331 J kilomole -t K -I.

kiJomoJe -l K

ce/3R

0.046.

19-10 2.9 X 10-11 Pa 1


- -

19-12
19-13

7.2 x 1 06 m

(a)

0.21

MeV; (b) 2.4 x

109 K;

(c)

2 x

-1
lOS ms ;

(d)

0.39 kg m-3.

Answers to Selected Problems

413

Chapter20

2.161 bits, 2.583 bits.


20-5

(a) 226 bits; (b) 5.

206

1.33 bits.

20-7

2.5 bits/letter.

Appendix: A

Al

(a)

(b)

Z =

x2

lny +

y (X

C.

3)

X +

C.

(c) Inexact.
A-l

(a)

z =

xy + y2 +

C.

(c) 0, 1/2.
A-3

A -4

AS

(b)

2
z = x y - 3x + C.

(c)

3 fo r all three paths.

(a)

2
lf
y
3
1
1
o
r
x

(b)

y.

(c)

x-.

(a)

sec2x, w = ysin.t - tanx + C.

Absolute temperature, 14,98-103

Berthelot equation of state, 32

Absolute zero, 14

Binomial coe fficient 217


,

entropy at, 172-173

Blackbody radiation. 333-338

heat capacity at, 179

Blackbody spectrum, 335-338

unattainability of. 17 4

Bohr magneton. 308-309

Adiabat, 22

Boiling, 13,140, 147

Adiabatic boundary, 7

Boltzmann distribution, 238-241

Adiabatic compressibility, 117

Boltzmann factor. 403-406

Adiabatic de magnet iza tion 174, 177,


.
318-321

Boltzmann's constant. 189

Adiabatic expansion of an ideal gas,

Bose-Einstein condensation, 340-345

Boltzmann statistics. 235-236


Bose-Einstein distribution. 244-246.

61-62
Adiabatic process, 8, 62

253-254

Adiabatic work, 40-43, 61-63 104

Bose-Einstein statistics. 245.333

Alternative statistical models,

Bose temperature. 343

254-257

Boson. 230.244-246

Arrow of time, 97

Boson gas

Assembly, 215,221-223

chemical potentiaL 341. 3-44

Atmosphere (unit) 9

energy, 345-346

Atmospheric pressure 9

entropy, 346

Available microstates. 216

heat capacity, 346

Average speed,l85, 196,202-203

Helmholtz function. 346

Avo gadro s law, 6, 22

partition function. 341

Avogadro's number, 6

pressure.347

'

414

Index

425

Boundary, 7-8

of thermal conductivity,207

Boyle's law,22

of viscosity,204-208, 211

Brillouin functicn,312-315

Coin-tossing experiment, 216-221

Bulk modulus,180

Collision cross section,198-200


Collision frequency, 199

Calorie (unit), 45

Collisions with wall,188-189

Canonical ensemble,255

Combined first and second laws, 103-104

Canonically conjugate pairs, 21,130

Comparison of distributions, 253-254

Camot cycle,74-79

()opressibuuty

and Qausius-Clapeyron equation,


147-148

adiabatic,117
of ideal gas, 28

efficiency,76,91-94

isothertnal,28

with ideal gas,76-79

Compression ratio, 83

temperature-entropy diagram,110-111

Concentration,252

and thertnodynamic temperature, 98-103


Carnot engine,76

Condensation, 23,340-345
Condition of integrability,31

Carnot heat pump, 82

Configuration,215

Camot refrigerator,79-80

Configuration work,37-40

Carnot's theorem,91-94,98-100

Conservation of energy, 40

Celsius temperature scale,13,16

Constant volume gas thertnometer, 13-15

Chain
rule of differentiation,393
.
. Change of phase,7, 23,2 6, 137-141
Characteristic temperature

Constituents, 153,157-160
Constraints, 237,238
Cooling

De bye,180,299

laser, 174

Einstein,294

magnetic,174,318-321

for linear oscillator,280

Cosmic background radiation, 337-338

for rotation, 285

Corresponding states, 295

for vibration, 280

Critical constants, van der Waals gas, 23-25

Charles' law, 22

Critical point,23-25,27

Chemical equilibrium, 160-161

Critical pressure,24-25

Chemical potential,151-155,244,251-252

Critical temperature,23-25

of boson gas, 341,344

Critical volume, 24-25

of dilute gas, 251,263

Cross section,198-200

of ideal gas, 166

Curie constant,329

of feranion gas, 355- 361

Curie law,314

multicomponent system, 153

Curie point, 32.6-328

of photon gas, 333

Curie temperature,326

Chemical reactions, 160-161,167

Curie-Weiss Ia"', 327-328

Classical distribution, 248

Cutoff frequency,297

Classical limit,247

Cycle

Classical statistics, 248

Carnot, 74-79,98-100

Clausius-Clapeyron equation't 140-141,147

Joule,84

Clausius inequality. 94-96

Otto,83

Clausius statement of the second law,90

Cyclical process, 8, 44,58

Closed system, 5

Cyclical rule,392

Coefficient
binomial. 217

Dalton's law of partial pressures, 162

of diffusion, 208

de Broglie wavelength, 227, 296,352

of expansion

(see also Expansivity), 28

of performance,79

Debye functions, 301, 302


Debye T3 law, 64, 180

Index

416
Debye temperature, 180, 299

Einstein temperature, 294

Deby e theory of specific heat capacity,

Einstein theory of specific heat capacity


293-295

296-301

Electrolytic cell,38

Degeneracy, 215
Degeneracy function, 229-230

Electromagnetic radiation, 333

Degenerate region, 363

Electromotive force, 38

Degrees of freedom

Electron gas, 361-364,367-368

electronic excitation, 279,287-288

Electronic excitation, 279.. 287-288

Gibbs phase rule, 160

Electronic heat capacity, 365-366

internal, 279

Electrons in a metal, 361-364

rotational, 191,279,284-287

Empirical temperature, 12, 103

translational. 19 1 , 279

Energy

vibrational, 191 279,282-284

Demagnetization adiabatic, 174, 177,

conservation of.. 40
equipartition of. 190-191, 270-271
Ferrni, 364

318-321

Density, 5, 10

frc e. 1 36

Density of states, 229-231

gravitationaL 369

Diamagnetism, 307. 329

heat. 53

Dia thertnal boundary, 8

of ideal gas, 56

Diatomic gas, 56.192,279-289

internal, 42 263

Diatomic molecules

kinetic, 151-152, 190

electronic excitation, 287

levels, 215,225-229

rotational modes, 284-287

magnetic. 310

translational modes, 28H

pot entiaL 151-152.310

vibrational modes, 282-284

states,225-229

Dielectric polarization. 38
Dieterici equation of state. 32
Differentials

ze ro

poi nL 279

Enthalpy
definition, 57, 132-133

exact, 42,393-400

of dilute gas. 264

inexact 37, 40,393-400

heats of transformation, 57-59

Diffusion, 208

of ideal gas. 60

Diffusive equilibrium. 156

Joule-Thomson experiment. 60

Dilute gas, 246 248


-

Entropy

Disorder, 220, 384. 388 - 389

at abs olute zero, 172-173

Dis si pati ve work,40-41

and available energy, 97-98

Distinguishable particles, 221, 235-236

Boltzmann definition. 223-225

Distribution function

of dilute gas, 252 .. 263

Boltzmann, 238-241,403-406

disorder, 220 .. 384,388-389

Bose-Einstein 244-246,253-254

of ideal g as, 112 113

Fern1i-Dirac, 242-244,253-254

information theory, 381,384-386

Maxwell-Boltzmann, 246-248.253-254

irreversibility, 95-96,118-121. 121-122

speed. 1 9J- 1 98
velocity, 194

Ncrnst heat theorem, 172


pr i ncipl e of increase of. 96-97

velocity components. 190

reversibilitv, 88 95,109-110

..

.,

Duhem-Margules equation, 168

second law 97

DuLong-Petit law 54, 293, 300

third law, 172-173

of van dcr Waals gas, 125


Efficiency of heat engine, 76, 91-94
Effusion, 201-203

Equations of state, 21- 31


Berthelot. 32
.

Index
Dieterici, 32
ideal gas. 22
liquid or solid, 29

van der WaaJs gas, 23-25


Equilibritun
chemical, 160-161

mechanical,145,156
stable, 143-145
state,7
thermal, 10-12,145,156
thermodynamic, 145
Equipartition of energy, 190-191,270-271

Error function,209
Euler's theorem, 153
Eutectic point, 157
Exact differential, 42,393- 400
Exclusion principle, 242, 356

Expansion coefficient,28 (see also


E xpansivi ty)
Expansion, free, 70, 104, 121-122
Expansivity, 28

427
Free expansion, 70,104, 121-122
Freezing, 26
Fusion, 26,140,148
Gamma function, 231,408
Gas constant, 22

Gas thermometer, 13-15


Gaussian integral,409
Gay Lussac Joule experiment, 69-72
-

Gay-Lussac law, 22
Gibbs-Duhem equation, 154
Gibbs function, 136-137
chemical potential, 154-155
of dilute gas, 264
of ideal gas, 146
open systems, 151-165
phase transitions, 137-139
thi rd

law, 171-173

Gibbs-Helmholz equation, 146,171


Gibbs paradox, 165,273
Gibbs phase rule, 157-160,167

of ideal gas, 28

Gradient, 9,204

of liquid or solid, 28-29

Grand canonical ensemble, 257

Extensive variable, 5

G rav itatio na l energy, 369


GrUneisen constant, 303

Fahrenheit temperature scale, 16


Fernti-Dirac distribution, 242-244,253-254

Harrnonic oscillator, 279-282

Ferrni-Dirac statistics, 242, 355

Heat

Fern1i energy, 355-357

definition, 43-45

Fernti level, 362

inexact di fferential, 44

Fermjon, 230,242-244
Fernlion gas

latent, 57,140
mechanical equivalent of, 46

chemical potential, 355-361

reservoir, 74, 75, 79, 81,82,89, 118-120

electronic heat capacity, 365-366

sign convention for, 44

energy,364

statistical interpretation, 250

entropy, 367
free electrons in a metal, 361-364

units of,45
Heat capacity, 53,191-193

Helmholtz function, 367

at absolute zero, 179

pressure,367

anomaly of liquid heliwn, 348

Fernli temperature, 357

at constant pressure, 53-54,116,191-193

Fermi velocity, 370

at constant volume, 53-54, 116,191-193

first law of thermodynamics, 44,46

Debye theory, 296-301

combined vlith second law, 103-104

of a diatomic gas, 191-193,279-289

FIXed-point temperature, 14

Einstein theory, 293-295

Fluctuations, 255-256
ux,186-188,202-203,205,336

electronic, 365- 366


of an ideal gas, 54-56,191-193

Free energy

Mayer's equation, 54-56

Gibbs, 136-137

ratio, 56,65, 192-193

Helmholtz, 134-136

of a solid, 293- 301

Index

428
of a spin-1/2 paramagnet, 317

definition 42
..

Heat e n gin e.. 76

dilute gas. 263

Heat flow, direction of, 44

ideal gas, 56, 191

Heat of transfortnation

relation to partition function, 263-264

fusion, 26, 140

Inverse collision, 193

sublimat ion 26, 140

Inversion point, 74

vaporization, 26, 140-141

Irreversible process.. 8 89-90, 118-121

..

Heat pump. 82

di ssipat i ve w ork , 40 41

Helium, liquid and solid, 347-349

entropy change, 95-96, 118- 121 , 121-122

Hel mholtz function

Ciibbs function, 136-137

and chemical poten tial. 251

Helmholtz function, 135-136

definition, 133 .. 1 34 -1 36

Isen tropic compressib ilit y 117

of dilute gas, 252, 263

Isobar, 22

of ideal gas, 146

Isobaric process, 8

H y steresis.. 308

Isochore. 22
Isochoric process, 8

Ice phase diagram, 26

Isotope se par ation by effusion , 203

Ice point. 13

Isotherm, 22

Ideal gas

critical, 23-24

adiabatic process. 62

Isothern1al com pressib ility. 28

and Carnot cycle, 74 -79

Isot hermal process, 8, 47

com pr essibility, 28
diatomic, 56

Joule (unit), 45

energy. 56

Joule co effici ent .. 70-71

e n thalpy. 60

Joule cvcle. 84

entropy, 112-113

Joule-Thomson coefficient, 73

e quation of state. 22

Joule-Thomson e xperi me n t 60 72-74 (see

..,.

expansivity, 28

also Throttling process)

Gibbs func tion . 146


heat capacity, 56

Kelvin-Joule experiment, 72-74

Helmholtz fun ction, 146

Kelvin-Planck statement of second law, 75.

internal energy. 56

90

Joule coefficient, 71

Kelvin temperature scale, 14-15

Joule-Thomson coefficient, 73-74

Kilocalorie (unit). 45

Increase of entropy, 96 -9 7

Kilom ole (un it ) , 6

Indistinguishable particles, 242. 244

Kilomole fraction, 1 59

Inequality of Clausius, 94 -96

Kinetic energy, 151-152, 190

Inexact differential 37 40, 393-400

Kinetic theory of gases. 183-208

Inflecti o n point, 23, 24

assumptions, 183-185

Informati on

ideal gas, 188-189

bit, 380

entropy, 381

Lagrange multipliers 236-238,257

maximum entropy. 381-384

Lambda li ne 348

uncertainty. 375-379

Lambda point . 348-349

lnfortnation theory. 375-388

connection to thermodvnamics. 384-387


.,.

In tegrati n g factor, 88, 396- 398

Lase r cooli ng .. 17 4
Latent heat 57. 140-141 (see also Heat of

transformation)'

Intensive variable, 5

Law of D uLong and Petit, 54 293, 300

Intermolecular forces, 151-152.183-185

Law s of thermodynamics

Internal energy

combined first and second, 103-104.

Index
first, 44, 46
second,96-97
third,171-179
zeroth,10-12

Legendre transformation, 130-132


Lennard-Jones potential, 184 (see also 612 potential)
Linear oscillator,279 -282

Liq uefaction of gas, 24


Liquid, 25
entropy change, 122-123
equation of state, 29

419
Microstate, 215,217
Microstates, sum over, 216,241,263-264
uing,l62-165,271-273
Molecular diameter, 184

Molecular flux, 186-188


Molecular weight, 6
Moments of a probability distribution,186
Most probable speed,197

Negative temperature, 321-324


experimental demonstration, 324
Negentropy, 386

Loschmidt's number, 6

Nernst statement of third law,172

Macrostate, 215,217

Occupation number. 216

Magnetic cooling,174,177,318 -321

Open systems, 151-165

Magnetic dipole moment,308-315

Otto cycle,83

energy, 310

paramagnetic salt, 307,314


saturation,313, 314

Paramagnetism, 308 -314


Brillouin function,312- 315

Magnetic domains, 307

Curie constant,329

Magne tic field, 38, 174, 307-310,318-321,

Curie law, 314

325-328

energy, 316

Magnetic saturation, 313,314

entropy,317-318

Magnetic susceptibility, 328

heat capacity, 318

Magnetism, theory ot 308-328

mean magnetic dipole moment, 311-315

Magnetization, 314,325- 328

spin-1/2 system,315-318

spontaneous, 325

Partial differentiation, 391-393

total, 38, 319

chain rule,393

work,38

cyclical rule,392

Manometer, 9-10
Maxwell-Boltzmann distribution, 248,
253-254,403-406
Maxwell-Boltzmann speed distribution,

193-198,269-270

reciprocal relation,392
Partial pressure, 162
Partition function,241,263-264
De bye solid,302
dilute gas, 265-266

Maxwell-Boltzmann statistics,248

linear oscillator,280

Maxwell-Boltzmann velocity distribution,

rotational,285-286

194

vibrational,280

Maxwell relations, 134,135,142-143

Pascal (unit),9

Maxwell's demon,273-275,387-388

Path,7

Mayer's equation,54-56

dependence,30

Mean free path,198-200

independence, 30

Mean kinetic energy,190

Pauli exclusion principle, 242, 256

Mean magnetic dipole moment,311-315

Performance,coefficient of,79

ean speed,l85,196,202-203

Phase,change of, 7, 23,26,137-141

Mean square speed, 185, 196

Mechanical equilibrium, 145, 146

(see also

Phase transforntation)

Phase diagram

Mechanical equivalent of heat, 46

helium,347

Melting,26

substance that contracts on freezing, 26

Microcanonical ensemble,255

substance that expands on freezing, 26

Index

430
water, 26

intensive, 5

Phase equilibrium,25-27,140-141,155-157

P-1' diagram, 22 .. 26

Phase rule, 157-160,167

P-vdiagrarn,22.. 24

Phase transformation (or transition), 7. 23.


26 .. 137-141

P-v T surface, 25

P honon 298

Quantum concentration, 268

Photon. 333- 334

Quantum density of states,229-231

Photon gas, 333-340

Quantum number, 227-229

chemical potential. 333,339

Quantum of energy, 333

energy, 335

Quantum states, 225-229

entropy, 339

Quasi-static process. 8

heat capacity,339
Helmholtz function,339

Radiation, blackbody,333-338

pressure,340

Rankine temperature scale, 16

Planck curve,335,337,338

Ratio of specific heat capacities 56,65

Planck radiation fortnula .. 334-335

Rayleigh-Jeans formula 337

Planck statement of third law, 172-173

Reactions! chemical, 160-161, 167

Poise (unit), 205

Reciprocal relation of differentiation. 392

Polarization, dielectric. 38

Reciprocity relations, 130, 133, 135

Population inversion, 322,324

Reduced mass, 284

Porous plug experiment,72-74

Reduced variables, 32

Potential energy, 121, 151-152. -184

Refrigerator, Carnot, 79-80

gravitationaL 369. 371


magnetic, 310

Reservoir, heat. 74,75, 79. 81 82. 89.


118-120

Potential well, 151-152, 184,226-227

Resistance thermometer. 17

Pressure
critical,24-26

Reversiblc processes.. 8
entropy change, 109-11 0, 111-112

dilute gas, 264

Riemann zeta function, 408

ideal gas, 22

Rigid rotator,224-285

kinetic theory evaluation. 188 -189

Root mean square speed, 186 196-197

partial, 162
radiation, 340

Sackur-Tetrode equation 267

units ot 9

Saturated liquid. 26

van der Waals gas,23

Saturated vapor, 26. 160

Probability density function, 185

Saturation,magnetic,313,314

Probability thermodynamic, 216-217

Schottky anomaly, 317

Process, 8

Second law of thermodynamics, 96-97

adiabatic, 8

Clausius statement, 90

cyc li ca l ,

and first law, 1 03 -104

free expansion. 70.104. 121-122

increase of entropy, 97

irreversible, 8. 89-90 .. 118-121

Kelvin-Planck statement 90

isentropic,112-113

Self-diffusion. 208

isobaric 8

Sign convention for work and heat 37 .. 41

isochoric, 8

6-12 potential, 184 (see also Lennard-Janes

..

isothermaL 8
quasi-static, 8

potential)

Solid. 25

reversible, 8, 109-110

entropy change .. 122-123

throttling, 72-74

equation of state, 29

Property, 5
extensive, 5

specific heat capacity, 54.. 293- 301


Solid helium, 347

431
Index
7
2
5
,
e
2
S
a
h
p
d
li
o
S
359
sion,
expan
ld
mrner
fe
So
117
gas,
in
d
spee
Sound
.
also
(see
93
191-1
53,
ty,
capac
i
beat
c
Specifi
Heat capacity)
,
6
1
1
,
4
5
3
5
,
e
r
u
s
s
e
r
a t constant p
191-19 3
3
9
1
1
9
1
,
6
1
1
,
4
5
3
5
,
e
at constant volum
9
3
-1
1
19
,
6
5
4
5
s,
a
g
i
a
e
of an id
Mayer's equation, 54-56
ratio, 56, 65, 192-193
6
e,
b
l
a
v
ari
e
v
si
n
te
x
e
n
a
f
o
e
lu
a
Specific v
Spectral density,334
Speed
average,185,196
distribution, 193-198,269-270
mean,185, 196
mean square,185,196
most probable, 1 97
root mean square,186, 196-197
of sound in gas, 117

Superfluid, 348
Surface film, 38

97
,
s,
5
g
undin
Su1ro
Survival equation, 201
8
2
3
,
c
i
t
e
n
g
a
m
,
y
t
li
i
Susceptib
System
closed, 5

isolated, 5
open, 5, 151 165
and surroundings, 5, 9 7
-

thermodynamic, 5

Tds equations, 1 1 3-117


Temperature
absolute, 14 98 103
absolute zero,171-178
,

Bose,343
Celsius, 13, 16
characteristic, 280, 285
critical, 23
Debye, 180,299

Spin angular momentum, 307

Einstein, 294

Spin degeneracy factor, 230

empirical, 12, 103

Spin-1/2 paramagnet, 315-31.8

Fahrenheit, 16

Stability of white dwarf stars, 369-370


Stable equilibriuu1, 143-145
entropy, 96, 138
Gibbs function, 136-138
Helmholtz function, 135-136, 138
mechanical, 145

therntal, 145
thermodynamic, 145
Standard deviation, 209
State variable,7
Statistical entropy, 223-225
Stltistics
Boltzanann, 235-236
Bose-Einstein, 244-246
classical, 248
Fermi-Dirac, 242
Maxweli-Boltztnann, 248
Steam point, 13

Fernti, 357
fixed point, 14
ice point,13
Kelvin, 14
negative, 321-324
Rankine, 16
steam point, 13
thermodynamic, 10-12,103
triple point, 14
Temperature-entropy diagram, 110-111
Temperature scale, 12-16,103
absolute, 14, 103
linear, 13-14
logari tlunic, 18
Thern1al conductivity, 207
Theranal equilibrium, 10-12,145,156
Thertual stability, 145
Thermionic emission, 364

Stefan-Boltzmann constant, 336

Thermocouple,17

Stefan-Boltzntann law, 236

Thermodynamic coordinates, 5

Stirling's approximation, 218,401-402


Stirring work, 40- 41

Therntodynamic equilibrium, 21,145

Stoichiometric coefficients, 160-16 .l

Therrnodynamic laws, 10-12,44,46,96-97,


171-179

Stress, 204

Thermodynamic potentials, 132-138

Sublimation, 26, 140, 1 47

Therntodynamic probability, 216-217

Sum over states, 241

Boltzmann, 235-236

Index

432
Bose-Einstein. 244-246

specific, 6

and entropy, 223-225

state. 5

Fermi-Dirac. 242

Variance, 160

Maxwell-Boltzmann, 248

Velocitv

.,

Thertnodynamic system. 5

distribution" 194

Thertnodynamic temperature. 10-12. 103

gradient. 204

Thermometer

l='ernti, 370

constant volume gas, 13-15

fluid, 204

resistance, 17

Viscosity 204-208

thermocouple, 17

coefficient of. 204-208,211

Thermometric property, 12, 13, 17

Volume

Third law of thermodynamic 171-179


Nernst statement, 172

critical, 24
expansivity. 2R

Planck statement 172-173

specific 27

unattainability staten1ent, 174, 177

v-1, diagram, 22

Throttling process, 72-74


Time s arrow, 97

Water

Torr (unit), 9

critical point 27

Total magnetization, 38 319

heat of fusion. 59

Transforrnation._ latent heat of, 57. 1 40-141

heat of vaporization, 59

Transfortnation, Legendre, 130-132

ice point 13

Transport processes. 203-208


diffusion, 208

pressure-temperature diagram, 26
specific heat capacity, 65

thermal conductivity._ 207

steam point 13

viscosity.. 204-208

triple point, 14. 15


vapor pressure. 26

Triple line. 25
Triple point, 14.58

Wave function. 226

of carbon dioxide. 26

White dwarf stars, 367-370

of water, 14, 15

Wicn 's displacement law, 336

T-s diagram, 110-111

Wicn 's law, 337

Two-fluid theory. 348

Work

Two-level system. 315-318

adiabatic, 41-43. 61-63


c:arnot cycle 76-79

Ultraviolet catastrophe. 337

configuration, 37-40

Unattainabilitv statement of third law. 174 .

dissipative.. 40-41

.,

177

Uncertainty.. information theory 375-379

electrical.. 41

Units, International System of (Sl ) .. 9

free expansion, 70. 104


inexact differentiaL 37

Universal gas constant, 22

isobaric. 48

Universe (system and surroundings). 5. 97

isothermaL 47
path-dependent. 40

Van der Waals gas. 23


characteristic constants. 23

path-independent (adiabatic). 42
sign convention for. 37

critical constants. 24. 25

statistical interpretation. 250

equation of state. 23

stirring. 40-41

Vapor, 24-26, 160


Vapor pressure. 26

Zero-point energy. 279

Vaporization. 26, 140, 147 160


Variable

Zeroth law of thermodvnamics. 10-12


Zeta function. 408

extensive, 5
intensive. 5

"'

Zustandsunune, 241

CONVERSION FACTORS

QUANTITY
Length

Mass

CONVERSION
1m= 100

em

3.281 ft

39.37 in

1 kg= 10 g

= 2.205lbm
Volume
=

103 liter

35.31 ft-'
')

Force

1 N = 1 kg m s- 2

10 dyne
-

= 0.2248lbr

Pressure

1 Pa = 1 N m-2
=

10 dyne cm-2

9.872 x 10-6atm
10

..

'

bar

= 7.502 x to-; torr


-

Energy

14.50 x lo-- psia

lJ=lNm
=

=:

=:

107 erg

2.390 x 10-4 kcaJ


6.242 X 101R eV
9.478 X 10- Btu