Peter Atkins - Four Laws That Drive The Universe (2007)

     
This page intentionally left blank

FOUR LAWS
   
Peter Atkins
1
3
Great Clarendon Street, Oxford ox 
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide in
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Oxford is a registered trade mark of Oxford University Press
in the UK and in certain other countries
Published in the United States
by Oxford University Press Inc., New York
© Peter Atkins 2007
The moral rights of the author have been asserted
Database right Oxford University Press (maker)
First published 2007
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
without the prior permission in writing of Oxford University Press,
or as expressly permitted by law, or under terms agreed with the appropriate
reprographics rights organization. Enquiries concerning reproduction
outside the scope of the above should be sent to the Rights Department,
Oxford University Press, at the address above
You must not circulate this book in any other binding or cover
and you must impose the same condition on any acquirer
British Library Cataloguing in Publication Data
Data available
Library of Congress Cataloging in Publication Data
Data available
Typeset by SPI Publisher Services, Pondicherry, India
Printed in Great Britain
on acid-free paper by
Clays Ltd, St Ives plc
ISBN 978–0–19–923236–9
1 3 5 7 9 10 8 6 4 2
P R E FAC E
Among the hundreds of laws that describe the universe, there

lurks a mighty handful. These are the laws of thermodynamics,
which summarize the properties of energy and its transformation
from one form to another. I hesitated to include the word ‘ther-
modynamics’ in the title of this little introduction to this bound-
lessly important and fascinating aspect of nature, hoping that you
would read at least this far, for the word does not suggest a light
read. And, indeed, I cannot pretend that it will be a light read.
When in due course, however, you emerge from the other end
of this slim volume, with your brain more sinewy and exercised,
you will have a profound understanding of the role of energy in
the world. In short, you will know what drives the universe.
Do not think that thermodynamics is only about steam
engines: it is about almost everything. The concepts did indeed
emerge during the nineteenth century when steam was the hot
topic of the day, but as the laws of thermodynamics became for-
mulated and their ramifications explored it became clear that the
subject could touch an enormously wide range of phenomena,
from the efficiency of heat engines, heat pumps, and refrigera-
tors, taking in chemistry on the way, and reaching as far as the
processes of life. We shall travel across that range in the pages
that follow.
vi Preface
The mighty handful consists of four laws, with the numbering

starting inconveniently at zero and ending at three. The first
two laws (the ‘zeroth’ and the ‘first’) introduce two familiar
but nevertheless enigmatic properties, the temperature and the
energy. The third of the four (the ‘second law’) introduces what
many take to be an even more elusive property, the entropy,
but which I hope to show is easier to comprehend than the
seemingly more familiar properties of temperature and energy.
The second law is one of the all-time great laws of science, for
it illuminates why anything—anything from the cooling of hot
matter to the formulation of a thought—happens at all. The
fourth of the laws (the ‘third law’) has a more technical role,
but rounds out the structure of the subject and both enables
and foils its applications. Although the third law establishes
a barrier that prevents us from reaching the absolute zero of
temperature, of becoming absolutely cold, we shall see that
there is a bizarre and attainable mirror world that lies below
zero.
Thermodynamics grew from observations on bulk matter—as
bulky as steam engines, in some cases—and became established
before many scientists were confident that atoms were more than
mere accounting devices. The subject is immeasurably enriched,
however, if the observation-based formulation of thermodynam-
ics is interpreted in terms of atoms and molecules. In this account
we consider first the observational aspects of each law, then dive
below the surface of bulk matter and discover the illumination
that comes from the interpretation of the laws in terms of con-
cepts that inhabit the underworld of atoms.
Preface vii
In conclusion, and before you roll up the sleeves of your mind

and get on with the business of understanding the workings of
the universe, I must thank Sir John Rowlinson for commenting
in detail on two drafts of the manuscript: his scholarly advice
was enormously helpful. If errors remain, they will no doubt be
traced to where I disagreed with him.
C O N T E NTS
Preface v
1. THE ZEROTH LAW: The concept of temperature 1
2. THE FIRST LAW: The conservation of energy 21
3. THE SECOND LAW: The increase in entropy 47
4. FREE ENERGY: The availability of work 79
5. THE THIRD LAW: The unattainability of zero 101
Conclusion 123
Further reading 125

Index 127
Symbol and unit index 131
1. THE ZEROTH LAW
The concept of temperature
T he zeroth law is an afterthought. Although it had long been
known that such a law was essential to the logical structure
of thermodynamics, it was not dignified with a name and num-
ber until early in the twentieth century. By then, the first and sec-
ond laws had become so firmly established that there was no hope
of going back and renumbering them. As will become apparent,
each law provides an experimental foundation for the introduc-
tion of a thermodynamic property. The zeroth law establishes the
meaning of what is perhaps the most familiar but is in fact the
most enigmatic of these properties: temperature.
Thermodynamics, like much of the rest of science, takes terms
with an everyday meaning and sharpens them—some would say,
hijacks them—so that they take on an exact and unambiguous
meaning. We shall see that happening throughout this introduc-
tion to thermodynamics. It starts as soon as we enter its doors.
The part of the universe that is at the centre of attention in
thermodynamics is called the system. A system may be a block
of iron, a beaker of water, an engine, a human body. It may
even be a circumscribed part of each of those entities. The rest
of the universe is called the surroundings. The surroundings are
where we stand to make observations on the system and infer
its properties. Quite often, the actual surroundings consist of
a water bath maintained at constant temperature, but that is
a more controllable approximation to the true surroundings,
the rest of the world. The system and its surroundings jointly
make up the universe. Whereas for us the universe is every-
thing, for a less profligate thermodynamicist it might consist
of a beaker of water (the system) immersed in a water bath
(the surroundings).
4 The Zeroth Law: The concept of temperature
A system is defined by its boundary. If matter can be added to

or removed from the system, then it is said to be open. A bucket,
or more refinedly an open flask, is an example, because we can
just shovel in material. A system with a boundary that is imper-
vious to matter is called closed. A sealed bottle is a closed system.
A system with a boundary that is impervious to everything in the
sense that the system remains unchanged regardless of anything
that happens in the surroundings is called isolated. A stoppered
vacuum flask of hot coffee is a good approximation to an isolated
system.
The properties of a system depend on the prevailing condi-
tions. For instance, the pressure of a gas depends on the vol-
ume it occupies, and we can observe the effect of changing
that volume if the system has flexible walls. ‘Flexible walls’ is
best thought of as meaning that the boundary of the system is
rigid everywhere except for a patch—a piston—that can move in
and out. Think of a bicycle pump with your finger sealing the
orifice.
Properties are divided into two classes. An extensive property
depends on the quantity of matter in the system—its extent.
The mass of a system is an extensive property; so is its vol-
ume. Thus, 2 kg of iron occupies twice the volume of 1 kg
of iron. An intensive property is independent of the amount
of matter present. The temperature (whatever that is) and the
density are examples. The temperature of water drawn from a
thoroughly stirred hot tank is the same regardless of the size
of the sample. The density of iron is 8.9 g cm−3 regardless
of whether we have a 1 kg block or a 2 kg block. We shall
meet many examples of both kinds of property as we unfold
The Zeroth Law: The concept of temperature 5
thermodynamics and it is helpful to keep the distinction in

mind.
So much for these slightly dusty definitions. Now we shall use

a piston—a movable patch in the boundary of a system—to
introduce one important concept that will then be the basis for
introducing the enigma of temperature and the zeroth law itself.
Suppose we have two closed systems, each with a piston on one
side and pinned into place to make a rigid container (Figure 1).
The two pistons are connected with a rigid rod so that as one
moves out the other moves in. We release the pins on the piston.
If the piston on the left drives the piston on the right into that sys-
tem, we can infer that the pressure on the left was higher than that
on the right, even though we have not made a direct measure of
1. If the gases in these two containers are at different pressures, when the
pins holding the pistons are released, the pistons move one way or the
other until the two pressures are the same. The two systems are then in
mechanical equilibrium. If the pressures are the same to begin with, there
is no movement of the pistons when the pins are withdrawn, for the two
systems are already in mechanical equilibrium.
the two pressures. If the piston on the right won the battle, then
we would infer that the pressure on the right was higher than
that on the left. If nothing had happened when we released the
pins, we would infer that the pressures of the two systems were
the same, whatever they might be. The technical expression for
the condition arising from the equality of pressures is mechanical
equilibrium. Thermodynamicists get very excited, or at least get
very interested, when nothing happens, and this condition of
equilibrium will grow in importance as we go through the laws.
We need one more aspect of mechanical equilibrium: it will
seem trivial at this point, but establishes the analogy that will
enable us to introduce the concept of temperature. Suppose the
two systems, which we shall call A and B, are in mechanical equi-
librium when they are brought together and the pins are released.
That is, they have the same pressure. Now suppose we break the
link between them and establish a link between system A and a
third system, C, equipped with a piston. Suppose we observe no
change: we infer that the systems A and C are in mechanical equi-
librium and we can go on to say that they have the same pressure.
Now suppose we break that link and put system C in mechanical
contact with system B. Even without doing the experiment, we
know what will happen: nothing. Because systems A and B have
the same pressure, and A and C have the same pressure, we can
be confident that systems C and B have the same pressure, and
that pressure is a universal indicator of mechanical equilibrium.
Now we move from mechanics to thermodynamics and the
world of the zeroth law. Suppose that system A has rigid walls
made of metal and system B likewise. When we put the two
systems in contact, they might undergo some kind of physical
change. For instance, their pressures might change or we could

see a change in colour through a peephole. In everyday language
we would say that ‘heat has flowed from one system to the other’
and their properties have changed accordingly. Don’t imagine,
though, that we know what heat is yet: that mystery is an aspect
of the first law, and we aren’t even at the zeroth law yet.
It may be the case that no change occurs when the two sys-
tems are in contact even though they are made of metal. In
that case we say that the two systems are in thermal equilib-
rium. Now consider three systems (Figure 2), just as we did
when talking about mechanical equilibrium. It is found that if
A is put in contact with B and found to be in thermal equi-
librium, and B is put in contact with C and found to be in
thermal equilibrium, then when C is put in contact with A, it
is always found that the two are in thermal equilibrium. This
rather trite observation is the essential content of the zeroth law of
thermodynamics:
if A is in thermal equilibrium with B, and B is in thermal equilib-

rium with C, then C will be in thermal equilibrium with A.
The zeroth law implies that just as the pressure is a physical

property that enables us to anticipate when systems will be in
mechanical equilibrium when brought together regardless of
their composition and size, then there exists a property that
enables us to anticipate when two systems will be in thermal
equilibrium regardless of their composition and size: we call this
universal property the temperature. We can now summarize the
statement about the mutual thermal equilibrium of the three
systems simply by saying that they all have the same temperature.
A
A
B C B C
A A C
B C B
2. A representation of the zeroth law involving (top left) three systems that
can be brought into thermal contact. If A is found to be in thermal equilib-
rium with B (top right), and B is in thermal equilibrium with C (bottom
left), then we can be confident that C will be in thermal equilibrium with
A if they are brought into contact (bottom right).
We are not yet claiming that we know what temperature is, all we
are doing is recognizing that the zeroth law implies the existence
of a criterion of thermal equilibrium: if the temperatures of two
systems are the same, then they will be in thermal equilibrium
when put in contact through conducting walls and an observer
of the two systems will have the excitement of noting that
nothing changes.
We can now introduce two more contributions to the vocabu-
lary of thermodynamics. Rigid walls that permit changes of state
when closed systems are brought into contact—that is, in the
language of Chapter 2, permit the conduction of heat—are called
diathermic (from the Greek words for ‘through’ and ‘warm’).
Typically, diathermic walls are made of metal, but any conduct-
ing material would do. Saucepans are diathermic vessels. If no
change occurs, then either the temperatures are the same or—
if we know that they are different—then the walls are classified
as adiabatic (‘impassable’). We can anticipate that walls are adia-
batic if they are thermally insulated, such as in a vacuum flask or
if the system is embedded in foamed polystyrene.
The zeroth law is the basis of the existence of a thermometer,
a device for measuring temperature. A thermometer is just a
special case of the system B that we talked about earlier. It is
a system with a property that might change when put in con-
tact with a system with diathermic walls. A typical thermometer
makes use of the thermal expansion of mercury or the change
in the electrical properties of material. Thus, if we have a sys-
tem B (‘the thermometer’) and put it in thermal contact with
A, and find that the thermometer does not change, and then
we put the thermometer in contact with C and find that it still
doesn’t change, then we can report that A and C are at the same
temperature.
There are several scales of temperature, and how they are
established is fundamentally the domain of the second law (see
Chapter 3). However, it would be too cumbersome to avoid
referring to these scales until then, though formally that could
be done, and everyone is aware of the Celsius (centigrade)
and Fahrenheit scales. The Swedish astronomer Anders Celsius
(1701–1744) after whom the former is named devised a scale
on which water froze at 100◦ and boiled at 0◦ , the opposite of
the current version of his scale (0◦ C and 100◦ C, respectively).
The German instrument maker Daniel Fahrenheit (1686–1736)
was the first to use mercury in a thermometer: he set 0◦ at the
lowest temperature he could reach with a mixture of salt, ice,
–300 –200 –100 0 100
Celsius
–400 –200 0 200
Fahrenheit
100 200 300 400
0
Kelvin
3. Three common temperature scales showing the relations between them.

The vertical dotted line on the left shows the lowest achievable tempera-
ture; the two dotted lines on the right show the normal freezing and boiling
points of water.
and water, and for 100◦ he chose his body temperature, a readily
transportable but unreliable standard. On this scale water freezes
at 32◦ F and boils at 212◦ F (Figure 3).
The temporary advantage of Fahrenheit’s scale was that with
the primitive technology of the time, negative values were rarely
needed. As we shall see, however, there is an absolute zero of tem-
perature, a zero that cannot be passed and where negative tem-
peratures have no meaning except in a certain formal sense, not
one that depends on the technology of the time (see Chapter 5).
It is therefore natural to measure temperatures by setting 0 at
this lowest attainable zero and to refer to such absolute tem-
peratures as the thermodynamic temperature. Thermodynamic
temperatures are denoted T , and whenever that symbol is used
in this book, it means the absolute temperature with T = 0
corresponding to the lowest possible temperature. The most
common scale of thermodynamic temperatures is the Kelvin
scale, which uses degrees (‘kelvins’, K) of the same size as the

Celsius scale. On this scale, water freezes at 273 K (that is, at 273
Celsius-sized degrees above absolute zero; the degree sign is not
used on the Kelvin scale) and boils at 373 K. Put another way,
the absolute zero of temperature lies at −273◦ C. Very occasion-
ally you will come across the Rankine scale, in which absolute
temperatures are expressed using degrees of the same size as
Fahrenheit’s.
In each of the first three chapters I shall introduce a property

from the point of view of an external observer. Then I shall enrich
our understanding by showing how that property is illuminated
by thinking about what is going on inside the system. Speaking
about the ‘inside’ of a system, its structure in terms of atoms and
molecules, is alien to classical thermodynamics, but it adds deep
insight, and science is all about insight.
Classical thermodynamics is the part of thermodynamics that
emerged during the nineteenth century before everyone was fully
convinced about the reality of atoms, and concerns relationships
between bulk properties. You can do classical thermodynamics
even if you don’t believe in atoms. Towards the end of the nine-
teenth century, when most scientists accepted that atoms were
real and not just an accounting device, there emerged the ver-
sion of thermodynamics called statistical thermodynamics, which
sought to account for the bulk properties of matter in terms of
its constituent atoms. The ‘statistical’ part of the name comes
from the fact that in the discussion of bulk properties we don’t

need to think about the behaviour of individual atoms but we do
need to think about the average behaviour of myriad atoms. For
instance, the pressure exerted by a gas arises from the impact of
its molecules on the walls of the container; but to understand and
calculate that pressure, we don’t need to calculate the contribu-
tion of every single molecule: we can just look at the average of
the storm of molecules on the walls. In short, whereas dynamics
deals with the behaviour of individual bodies, thermodynamics
deals with the average behaviour of vast numbers of them.
The central concept of statistical thermodynamics as far as we
are concerned in this chapter is an expression derived by Lud-
wig Boltzmann (1844–1906) towards the end of the nineteenth
century. That was not long before he committed suicide, partly
because he found intolerable the opposition to his ideas from
colleagues who were not convinced about the reality of atoms.
Just as the zeroth law introduces the concept of temperature
from the viewpoint of bulk properties, so the expression that
Boltzmann derived introduces it from the viewpoint of atoms,
and illuminates its meaning.
To understand the nature of Boltzmann’s expression, we need
to know that an atom can exist with only certain energies. This
is the domain of quantum mechanics, but we do not need any
of that subject’s details, only that single conclusion. At a given
temperature—in the bulk sense—a collection of atoms consists
of some in their lowest energy state (their ‘ground state’), some
in the next higher energy state, and so on, with populations that
diminish in progressively higher energy states. When the pop-
ulations of the states have settled down into their ‘equilibrium’
populations, and although atoms continue to jump between

energy levels there is no net change in the populations, it turns
out that these populations can be calculated from a knowledge of
the energies of the states and a single parameter, ‚ (beta).
Another way of thinking about the problem is to think of a
series of shelves fixed at different heights on a wall, the shelves
representing the allowed energy states and their heights the
allowed energies. The nature of these energies is immaterial: they
may correspond, for instance, to the translational, rotational, or
vibrational motion of molecules. Then we think of tossing balls
(representing the molecules) at the shelves and noting where
they land. It turns out that the most probable distribution of
populations (the numbers of balls that land on each shelf) for a
large number of throws, subject to the requirement that the total
energy has a particular value, can be expressed in terms of that
single parameter ‚.
The precise form of the distribution of the molecules over
their allowed states, or the balls over the shelves, is called the
Boltzmann distribution. This distribution is so important that it
is important to see its form. To simplify matters, we shall express
it in terms of the ratio of the population of a state of energy E to
the population of the lowest state, of energy 0:
Population of state of energy E

= e−‚E
Population of state of energy 0
We see that for states of progressively higher energy, the popu-

lations decrease exponentially: there are fewer balls on the high
shelves than on the lower shelves. We also see that as the para-
meter ‚ increases, then the relative population of a state of given
energy decreases and the balls sink down on to the lower shelves.
They retain their exponential distribution, with progressively
fewer balls in the upper levels, but the populations die away more
quickly with increasing energy.
When the Boltzmann distribution is used to calculate the
properties of a collection of molecules, such as the pressure of
a gaseous sample, it turns out that it can be identified with the
reciprocal of the (absolute) temperature. Specifically, ‚ = 1/kT,
where k is a fundamental constant called Boltzmann’s constant.
To bring ‚ into line with the Kelvin temperature scale, k has the
value 1.38 × 10−23 joules per kelvin. 1 The point to remember is
that, because ‚ is proportional to 1/T , as the temperature goes
up, ‚ goes down, and vice versa.
There are several points worth making here. First, the huge
importance of the Boltzmann distribution is that it reveals the
molecular significance of temperature: temperature is the para-
meter that tells us the most probable distribution of populations
of molecules over the available states of a system at equilibrium.
When the temperature is high (‚ low), many states have signif-
icant populations; when the temperature is low (‚ high), only
the states close to the lowest state have significant populations
(Figure 4). Regardless of the actual values of the populations,
they invariably follow an exponential distribution of the kind
given by the Boltzmann expression. In terms of our balls-on-
shelves analogy, low temperatures (high ‚) corresponds to our
throwing the balls weakly at the shelves so that only the lowest
1. Energy is reported in joules (J): 1 J = 1 kg m2 s−2 . We could think of 1 J

as the energy of a 2 kg ball travelling at 1 m s−1 . Each pulse of the human heart
expends an energy of about 1 J.
4. The Boltzmann distribution is an exponentially decaying function of

the energy. As the temperature is increased, the populations migrate from
lower energy levels to higher energy levels. At absolute zero, only the
lowest state is occupied; at infinite temperature, all states are equally
populated.
are occupied. High temperatures (low ‚) corresponds to our

throwing the balls vigorously at the shelves, so that even high
shelves are populated significantly. Temperature, then, is just a
parameter that summarizes the relative populations of energy levels
in a system at equilibrium.
The second point is that ‚ is a more natural parameter for
expressing temperature than T itself. Thus, whereas later we shall
see that absolute zero of temperature (T = 0) is unattainable in
a finite number of steps, which may be puzzling, it is far less
surprising that an infinite value of ‚ (the value of ‚ when T = 0)
is unattainable in a finite number of steps. However, although ‚ is
the more natural way of expressing temperatures, it is ill-suited to
everyday use. Thus water freezes at 0◦ C (273 K), corresponding
to ‚ = 2.65 × 1020 J−1 , and boils at 100◦ C (373 K), corresponding
to ‚ = 1.94 × 1020 J−1 . These are not values that spring readily off
the tongue. Nor are the values of ‚ that typify a cool day (10◦ C,
corresponding to 2.56 × 1020 J−1 ) and a warmer one (20◦ C,
corresponding to 2.47 × 1020 J−1 ).
The third point is that the existence and value of the fun-
damental constant k is simply a consequence of our insist-
ing on using a conventional scale of temperature rather than
the truly fundamental scale based on ‚. The Fahrenheit, Cel-
sius, and Kelvin scales are misguided: the reciprocal of tem-
perature, essentially ‚, is more meaningful, more natural, as
a measure of temperature. There is no hope, though, that it
will ever be accepted, for history and the potency of simple
numbers, like 0 and 100, and even 32 and 212, are too deeply
embedded in our culture, and just too convenient for everyday
use.
Although Boltzmann’s constant k is commonly listed as a
fundamental constant, it is actually only a recovery from a
historical mistake. If Ludwig Boltzmann had done his work
before Fahrenheit and Celsius had done theirs, then it would
have been seen that ‚ was the natural measure of temperature,
and we might have become used to expressing temperatures in
the units of inverse joules with warmer systems at low values
of ‚ and cooler systems at high values. However, conventions
had become established, with warmer systems at higher tem-
peratures than cooler systems, and k was introduced, through
k‚ = 1/T , to align the natural scale of temperature based on
‚ to the conventional and deeply ingrained one based on T .
Thus, Boltzmann’s constant is nothing but a conversion factor
between a well-established conventional scale and the one that,
with hindsight, society might have adopted. Had it adopted ‚
as its measure of temperature, Boltzmann’s constant would not

have been necessary.
We shall end this section on a more positive note. We have
established that the temperature, and specifically ‚, is a param-
eter that expresses the equilibrium distribution of the molecules
of a system over their available energy states. One of the easiest
systems to imagine in this connection is a perfect (or ‘ideal’) gas,
in which we imagine the molecules as forming a chaotic swarm,
some moving fast, others slow, travelling in straight lines until
one molecule collides with another, rebounding in a different
direction and with a different speed, and striking the walls in a
storm of impacts and thereby giving rise to what we interpret
as pressure. A gas is a chaotic assembly of molecules (indeed,
the words ‘gas’ and ‘chaos’ stem from the same root), chaotic in
spatial distribution and chaotic in the distribution of molecular
speeds. Each speed corresponds to a certain kinetic energy, and so
the Boltzmann distribution can be used to express, through the
distribution of molecules over their possible translational energy
states, their distribution of speeds, and to relate that distribu-
tion of speeds to the temperature. The resulting expression is
called the Maxwell–Boltzmann distribution of speeds, for James
Clerk Maxwell (1831–1879) first derived it in a slightly different
way. When the calculation is carried through, it turns out that
the average speed of the molecules increases as the square root
of the absolute temperature. The average speed of molecules
in the air on a warm day (25◦ C, 298 K) is greater by 4 per cent
than their average speed on a cold day (0◦ C, 273 K). Thus, we
can think of temperature as an indication of the average speeds of
molecules in a gas, with high temperatures corresponding to high
average speeds and low temperatures to lower average speeds

(Figure 5).
5. The Maxwell–Boltzmann distribution of molecular speeds for mole-

cules of various mass and at different temperatures. Note that light mole-
cules have higher average speeds than heavy molecules. The distribution
has consequences for the composition of planetary atmospheres, as light
molecules (such as hydrogen and helium) may be able to escape into space.
A word or two of summary might be appropriate at this

point. From the outside, from the viewpoint of an observer
stationed, as always, in the surroundings, temperature is a
property that reveals whether, when closed systems are in
contact through diathermic boundaries, they will be in thermal
equilibrium—their temperatures are the same—or whether
there will be a consequent change of state—their temperatures

are different—that will continue until the temperatures have
equalized. From the inside, from the viewpoint of a microscop-
ically eagle-eyed observer within the system, one able to discern
the distribution of molecules over the available energy levels, the
temperature is the single parameter that expresses those popu-
lations. As the temperature is increased, that observer will see
the population extending up to higher energy states, and as it is
lowered, the populations relax back to the states of lower energy.
At any temperature, the relative population of a state varies expo-
nentially with the energy of the state. That states of higher energy
are progressively populated as the temperature is raised means
that more and more molecules are moving (including rotating
and vibrating) more vigorously, or the atoms trapped at their
locations in a solid are vibrating more vigorously about their
average positions. Turmoil and temperature go hand in hand.
2. THE FIRST LAW
The conservation of energy
T he first law of thermodynamics is generally thought to be
the least demanding to grasp, for it is an extension of the
law of conservation of energy, that energy can be neither created
nor destroyed. That is, however much energy there was at the
start of the universe, so there will be that amount at the end.
But thermodynamics is a subtle subject, and the first law is much
more interesting than this remark might suggest. Moreover, like
the zeroth law, which provided an impetus for the introduction
of the property ‘temperature’ and its clarification, the first law
motivates the introduction and helps to clarify the meaning of
the elusive concept of ‘energy’.
We shall assume at the outset that we have no inkling that there
is any such property, just as in the introduction to the zeroth law
we did not pre-assume that there was anything we should call
temperature, and then found that the concept was forced upon
us as an implication of the law. All we shall assume is that the
well-established concepts of mechanics and dynamics, like mass,
weight, force, and work, are known. In particular, we shall base
the whole of this presentation on an understanding of the notion
of ‘work’.
Work is motion against an opposing force. We do work when
we raise a weight against the opposing force of gravity. The mag-
nitude of the work we do depends on the mass of the object, the
strength of the gravitational pull on it, and the height through
which it is raised. You yourself might be the weight: you do work
when you climb a ladder; the work you do is proportional to
your weight and the height through which you climb. You also
do work when cycling into the wind: the stronger the wind and
the further you travel the greater the work you do. You do work
24 The First Law: The conservation of energy
when you stretch or compress a spring, and the amount of work

you do depends on the strength of the spring and the distance
through which it is stretched or compressed.
All work is equivalent to the raising of a weight. For instance,
although we might think of stretching a spring, we could connect
the stretched spring to a pulley and weight and see how far the
weight is raised when the spring returns to its natural length. The
magnitude of the work of raising a mass m (for instance, 50 kg)
through a height h (for instance, 2.0 m) on the surface of the
Earth is calculated from the formula mgh, where g is a constant
known as the acceleration of free fall, which at sea level on Earth is
close to 9.8 m s−2 . Raising a 50 kg weight through 2.0 m requires
work of magnitude 980 kg m2 s−2 . As we saw in the footnote
on p. 14, the awkward combination of units ‘kilograms metre
squared per second squared’ is called the joule (symbol J). So,
to raise our weight, we have to do 980 joules (980 J) of work.
Work is the primary foundation of thermodynamics and in
particular of the first law. Any system has the capacity to do work.
For instance, a compressed or extended spring can do work: as
we have remarked, it can be used to bring about the raising of
a weight. An electric battery has the capacity to do work, for
it can be connected to an electric motor which in turn can be
used to raise a weight. A lump of coal in an atmosphere of air
can be used to do work by burning it as a fuel in some kind of
engine. It is not an entirely obvious point, but when we drive
an electric current through a heater, we are doing work on the
heater, for the same current could be used to raise a weight by
passing it through an electric motor rather than the heater. Why
a heater is called a ‘heater’ and not a ‘worker’ will become clear
The First Law: The conservation of energy 25
once we have introduced the concept of heat. That concept hasn’t

appeared yet.
With work a primary concept in thermodynamics, we need a
term to denote the capacity of a system to do work: that capacity
we term energy. A fully stretched spring has a greater capacity to
do work than the same spring only slightly stretched: the fully
stretched spring has a greater energy than the slightly stretched
spring. A litre of hot water has the capacity to do more work
than the same litre of cold water: a litre of hot water has a greater
energy than a litre of cold water. In this context, there is nothing
mysterious about energy: it is just a measure of the capacity of a
system to do work, and we know exactly what we mean by work.
Now we extend these concepts from dynamics to thermodynam-

ics. Suppose we have a system enclosed in adiabatic (thermally
non-conducting) walls. We established the concept of ‘adiabatic’
in Chapter 1 by using the zeroth law, so we have not slipped
in an undefined term. In practice, by ‘adiabatic’ we mean a
thermally insulated container, like a well-insulated vacuum flask.
We can monitor the temperature of the contents of the flask by
using a thermometer, which is another concept introduced by
the zeroth law, so we are still on steady ground. Now we do some
experiments.
First, we churn the contents of the flask (that is, the system)
with paddles driven by a falling weight, and note the change
in temperature this churning brings about. Exactly this type of
experiment was performed by J. P. Joule (1818–1889), one of

the fathers of thermodynamics, in the years following 1843. We
know how much work has been done by noting the heaviness
of the weight and the distance through which it fell. Then we
remove the insulation and let the system return to its original
state. After replacing the insulation, we put a heater into the
system and pass an electric current for a time that results in the
same work being done on the heater as was done by the falling
weight. We would have done other measurements to relate the
current passing through a motor for various times and noting the
height to which weights are raised, so we can interpret the com-
bination of time and current as an amount of work performed.
The conclusion we arrive at in this pair of experiments and in
a multitude of others of a similar kind is that the same amount
of work, however it is performed, brings about the same change of
state of the system.
This conclusion is like climbing a mountain by a variety of
different paths, each path corresponding to a different method
of doing work. Provided we start at the same base camp and
arrive at the same destination, we shall have climbed through the
same height regardless of the path we took between them. That
is, we can attach a number (the ‘altitude’) to every point on the
mountain, and calculate the height we have climbed, regardless
of the path, by taking the difference of the initial and final alti-
tudes for our climb. Exactly the same applies to our system. The
fact that the change of state is path-independent means that we
can associate a number, which we shall call the internal energy
(symbol U ) with each state of the system. Then we can calculate
the work needed to travel between any two states by taking the
6. The observation that different ways of doing work on a system and

thereby changing its state between fixed endpoints required the same
amount of work is analogous to different paths on a mountain resulting
in the same change of altitude leads to the recognition of the existence of a
property called the internal energy.
difference of the initial and final values of the internal energy,

and write work required = U (final) − U (initial) (Figure 6).
The observation of the path-independence of the work
required to go between two specified states in an adiabatic system
(remember, at this stage the system is adiabatic) has motivated
the recognition that there is a property of the system that is a
measure of its capacity to do work. In thermodynamics, a prop-
erty that depends only on the current state of the system and
is independent of how that state was prepared (like altitude in
geography) is called a state function. Thus, our observations have
motivated the introduction of the state function called internal
energy. We might not understand the deep nature of internal
energy at this stage, but nor did we understand the deep nature of
the state function we called temperature when we first encoun-
tered it in the context of the zeroth law.
We have not yet arrived at the first law: this will take a little
more work, both literally and figuratively. To move forward,
let’s continue with the same system but strip away the thermal
insulation so that it is no longer adiabatic. Suppose we do our
churning business again, starting from the same initial state and
continuing until the system is in the same final state as before.
We find that a different amount of work is needed to reach the
final state.
Typically, we find that more work has to be done than in the
adiabatic case. We are driven to conclude that the internal energy
can change by an agency other than by doing work. One way of
regarding this additional change is to interpret it as arising from
the transfer of energy from the system into the surroundings due
to the difference in temperature caused by the work that we do
as we churn the contents. This transfer of energy as a result of a
temperature difference is called heat.
The amount of energy that is transferred as heat into or out
of the system can be measured very simply: we measure the
work required to bring about a given change in the adiabatic
system, and then the work required to bring about the same
change of state in the diathermic system (the one with thermal
insulation removed), and take the difference of the two values.
That difference is the energy transferred as heat. A point to note
is that the measurement of the rather elusive concept of ‘heat’ has
been put on a purely mechanical foundation as the difference in
the heights through which a weight falls to bring about a given
change of state under two different conditions (Figure 7).
We are within a whisper of arriving at the first law. Suppose
we have a closed system and use it to do some work or allow a
7. When a system is adiabatic (left), a given change of state is brought

about by doing a certain amount of work. When the same system under-
goes the same change of state in a non-adiabatic container (right), more
work has to be done. The difference is equal to the energy lost as heat.
release of energy as heat. Its internal energy falls. We then leave

the system isolated from its surroundings for as long as we like,
and later return to it. We invariably find that its capacity to do
work—its internal energy—has not been restored to its original
value. In other words,
the internal energy of an isolated system is constant.
That is the first law of thermodynamics, or at least one statement

of it, for the law comes in many equivalent forms.
Another universal law of nature, this time of human nature, is
that the prospect of wealth motivates deceit. Wealth—and untold
benefits to humanity—would accrue to an untold extent if the
first law were found to be false under certain conditions. It would
be found to be false if work could be generated by an adiabatic,
closed system without a diminution of its internal energy. In
other words, if we could achieve perpetual motion, work pro-
duced without consumption of fuel. Despite enormous efforts,
perpetual motion has never been achieved. There have been

claims galore, of course, but all of them have involved a degree
of deception. Patent offices are now closed to the consideration
of all such machines, for the first law is regarded as unbreakable
and reports of its transgression not worth the time or effort to
pursue. There are certain instances in science, and certainly in
technology, where a closed mind is probably justified.
We have a variety of cleaning-up exercises to do before we leave

the first law. First, there is the use of the term ‘heat’. In everyday
language, heat is both a noun and a verb. Heat flows; we heat. In
thermodynamics heat is not an entity or even a form of energy:
heat is a mode of transfer of energy. It is not a form of energy, or a
fluid of some kind, or anything of any kind. Heat is the transfer
of energy by virtue of a temperature difference. Heat is the name
of a process, not the name of an entity.
Everyday discourse would be stultified if we were to insist on
the precise use of the word heat, for it is enormously convenient
to speak of heat flowing from here to there, and to speak of
heating an object. The first of these everyday usages was moti-
vated by the view that heat is an actual fluid that flows between
objects at different temperatures, and this powerful imagery is
embedded indelibly in our language. Indeed, there are many
aspects of the migration of energy down temperature gradients
that are fruitfully treated mathematically by regarding heat as the
flow of a massless (‘imponderable’) fluid. But that is essentially a
coincidence, it is not an indicator that heat is actually a fluid any

more than the spread of consumer choice in a population, which
can also be treated by similar equations, is a tangible fluid.
What we should say, but it is usually too tedious actually to
say it repeatedly, is that energy is transferred as heat (that is, as
the result of a temperature difference). To heat, the verb, should
for precision be replaced by circumlocutions such as ‘we con-
trive a temperature difference such that energy flows through a
diathermic wall in a desired direction’. Life, though, is too short,
and it is expedient, except when we want to be really precise, to
adopt the casual easiness of everyday language, and we shall cross
our fingers and do so, but do bear in mind how that shorthand
should be interpreted.
There has probably been detected a slipperiness in the preceding

remarks, for although we have warned against regarding heat as
a fluid, there is still the whiff of fluidity about our use of the term
energy. It looks as though we have simply pushed back to a deeper
layer the notion of fluid. This apparent deceit, though, is resolved
by identifying the molecular natures of heat and work. As usual,
digging into the underworld of phenomena illuminates them.
In thermodynamics, we always distinguish between the modes
of transfer of energy by observations in the surroundings: the
system is blind to the processes by which it is provided with or
loses energy. We can think of a system as like a bank: money can
be paid in or withdrawn in either of two currencies, but once
inside there is no distinction between the type of funds in which

its reserves are stored.
First, the molecular nature of work. We have seen that at an
observational level, doing work is equivalent to the raising of
a weight. From a molecular viewpoint, the raising of a weight
corresponds to all its atoms moving in the same direction. Thus,
when a block of iron is raised, all the iron atoms move upwards
uniformly. When the block is lowered—and does work on the
system, like compressing a spring or a gas, and increases its inter-
nal energy—all its atoms move downwards uniformly. Work is the
transfer of energy that makes use of the uniform motion of atoms in
the surroundings (Figure 8).
Now, the molecular nature of heat. We saw in Chapter 1
that the temperature is a parameter that tells us the relative
numbers of atoms in the allowed energy states, with the higher
energy states progressively more populated as the temperature
8. The molecular distinction between the transfer of energy as work (left)

and heat (right). Doing work results in the uniform motion of atoms in the
surroundings; heating stimulates their disorderly motion.
is increased. In more pictorial terms, a block of iron at high

temperature consists of atoms that are oscillating vigorously
around their average positions. At low temperatures, the atoms
continue to oscillate, but with less vigour. If a hot block of iron
is put in contact with a cooler block, the vigorously oscillating
atoms at the edge of the hot block jostle the less vigorously
oscillating atoms at the edge of the cool block into more vig-
orous motion, and they pass on their energy by jostling their
neighbours. There is no net motion of either block, but energy
is transferred from the hotter to the cooler block by this random
jostling where the two blocks are in contact. That is, heat is the
transfer of energy that makes use of the random motion of atoms in
the surroundings (Figure 8).
Once the energy is inside the system, either by making use
of the uniform motion of atoms in the surroundings (a falling
weight) or of randomly oscillating atoms (a hotter object, such
as a flame), there is no memory of how it was transferred. Once
inside, the energy is stored as the kinetic energy (the energy due
to motion) and the potential energy (the energy due to position)
of the constituent atoms, and that energy can be withdrawn
either as heat or as work. The distinction between work and heat
is made in the surroundings: the system has no memory of the
mode of transfer nor is it concerned about how its store of energy
will be used.
This blindness to the mode of transfer needs a little further
explanation. Thus, if a gas in an adiabatic container is com-
pressed by a falling weight, the incoming piston acts like a bat
in a microscopic game of table-tennis. When a molecule strikes
the piston, it is accelerated. However, as it flies back into the gas it
undergoes collisions with the other molecules in the system, and

as a result its enhanced kinetic energy is quickly dispersed over
them and its direction of motion is randomized. When the same
sample of gas is heated, the random jostling of the atoms in the
surroundings stimulates the gas molecules into more vigorous
motion, and the acceleration of the molecules at the thermally
conducting walls is quickly distributed over the entire sample.
The result within the system is the same.
We can now return to the faintly enigmatic remark made earl-
ier that an electric heater is better regarded as an electric worker.
The electric current that is passed through the coil of wire within
the heater is a uniform flow of electrons. The electrons of that
current collide with the atoms of the wire and cause them to
wobble around their mean positions. That is, the energy—and
the temperature—of the coil of wire is raised by doing work
on it. However, the coil of wire is in thermal contact with the
contents of the system, and the vigorous motion of the atoms
of the wire jostle the atoms of the system; that is, the filament
heats the system. So, although we do work on the heater itself,
that work is translated into heating the system: the worker has
become heater.
A final point is that the molecular interpretation of heat
and work elucidates one aspect of the rise of civilization. Fire
preceded the harnessing of fuels to achieve work. The heat
of fire—the tumbling out of energy as the chaotic motion
of atoms—is easy to contrive for the tumbling is uncon-
strained. Work is energy tamed, and requires greater sophis-
tication to contrive. Thus, humanity stumbled easily on to
fire but needed millennia to arrive at the sophistication of
the steam engine, the internal combustion engine, and the jet
engine.
The originators of thermodynamics were subtle people, and

quickly realized that they had to be careful when specifying how a
process is carried out. Although the technicality we shall describe
now has little immediate relevance to the first law at the level of
our discussion, it will prove to be of vital significance when we
turn to the second law.
I alluded in Chapter 1 to science’s hijacking of familiar words
and adding a new precision to their meaning. In the current
context we need to consider the word ‘reversible’. In everyday lan-
guage, a reversible process is one that can be reversed. Thus, the
rolling of a wheel can be reversed, so in principle a journey can
be traversed in reverse. The compression of a gas can be reversed
by pulling out the piston that effected the compression. In ther-
modynamics ‘reversible’ means something rather more refined: a
reversible process in thermodynamics is one that is reversed by an
infinitesimal modification of the conditions in the surroundings.
The key word is infinitesimal. If we think of a gas in a system
at a certain pressure, with the piston moving out against a lower
external pressure, an infinitesimal change in the external pres-
sure will not reverse the motion of the piston. The expansion is
reversible in the colloquial sense but not in the thermodynamic
sense. If a block of iron (the system) at 20◦ C is immersed in a
water bath at 40◦ C, energy will flow as heat from the bath into
the block, and an infinitesimal change in the temperature of the

water will have no effect on the direction of flow. The transfer
of energy as heat is not reversible in the thermodynamic sense
in this instance. However, now consider the case in which the
external pressure matches the pressure of the gas in the system
exactly. As we saw in Chapter 1, we say that the system and its
surroundings are in mechanical equilibrium. Now increase the
external pressure infinitesimally: the piston moves in a little. Now
reduce the external pressure infinitesimally: the piston moves
out a little. We see that the direction of motion of the piston is
changed by an infinitesimal change in a property, in this case
the pressure, of the surroundings. The expansion is reversible
in the thermodynamic sense. Likewise, consider a system at the
same temperature as the surroundings. In this case the system
and its surroundings are in thermal equilibrium. If we reduce the
temperature of the surroundings infinitesimally, energy flows out
of the system as heat. If we increase the temperature of the sur-
roundings infinitesimally, energy flows into the system as heat.
The transfer of energy as heat is reversible in the thermodynamic
sense in this instance.
The greatest amount of work is done if the expansion of a
gas is reversible at every stage. Thus, we match the external
pressure to the pressure of the gas in the system and then
reduce the external pressure infinitesimally: the piston moves
out a little. The pressure of the gas falls a little because it now
occupies a greater volume. Then we reduce the external pressure
infinitesimally, the piston moves out a little more and the
pressure of the gas decreases a little. This process of effectively
matching the external pressure to the falling pressure of the gas
continues until the piston has moved out a desired amount and,
through its coupling to a weight, has done a certain amount
of work. No greater work can be done, because if at any stage
the external pressure is increased even infinitesimally, then the
piston will move in rather than out. That is, by ensuring that
at every stage the expansion is reversible in the thermodynamic
sense, the system does maximum work. This conclusion is
general: reversible changes achieve maximum work. We shall draw
on this generalization in the following chapters.
Thermodynamicists are also subtle in their discussion of the

quantity of heat that can be extracted from a system, such as
when a fuel burns. We can appreciate the problem as follows.
Suppose we burn a certain amount of hydrocarbon fuel in a
container fitted with a movable piston. As the fuel burns it
produces carbon dioxide and water vapour, which occupy more
space than the original fuel and oxygen, and as a result the piston
is driven out to accommodate the products. This expansion
requires work. That is, when a fuel burns in a container that is
free to expand, some of the energy released in the combustion
is used to do work. If the combustion takes place in a container
with rigid walls, the combustion releases the same amount of
energy, but none of it is used to do work because no expansion
can occur. In other words, more energy is available as heat in the
latter case than in the former. To calculate the heat that can be
produced in the former case, we have to account for the energy
that is used to make room for the carbon dioxide and water
vapour and subtract that from the total change in energy. This
is true even if there is no physical piston—if the fuel burns in a
dish—because, although we cannot see it so readily, the gaseous
products must still make room for themselves.
Thermodynamicists have developed a clever way of taking
into account the energy used to do work when any change, and
particularly the combustion of a fuel, occurs, without having to
calculate the work explicitly in each case. To do so, they switch
attention from the internal energy of a system, its total energy
content, to a closely related quantity, the enthalpy (symbol H).
The name comes from the Greek words for ‘heat inside’, and
although, as we have stressed, there is no such thing as ‘heat’ (it is
a process of transfer, not a thing), for the circumspect the name
is well chosen, as we shall see. The formal relation of enthalpy,
H, to internal energy, U , is easily written down as H = U + pV,
where p is the pressure of the system and V is its volume. From
this relation it follows that the enthalpy of a litre of water open to
the atmosphere is only 100 J greater than its internal energy, but
it is much more important to understand its significance than to
note small differences in numerical values.
It turns out that the energy released as heat by a system free
to expand or contract as a process occurs, as distinct from the
total energy released in the same process, is exactly equal to the
change in enthalpy of the system. That is, as if by magic—but
actually by mathematics—the leakage of energy from a system
as work is automatically taken into account by focusing on the
change in enthalpy. In other words, the enthalpy is the basis of a
kind of accounting trick, which keeps track invisibly of the work
that is done by the system, and reveals the amount of energy that
is released only as heat, provided the system is free to expand in
an atmosphere that exerts a constant pressure on the system.
It follows that if we are interested in the heat that can be
generated by the combustion of a fuel in an open container,
such as a furnace, then we use tables of enthalpies to calculate
the change in enthalpy that accompanies the combustion. This
change is written H, where the Greek uppercase delta is used
throughout thermodynamics to denote a change in a quantity.
Then we identify that change with the heat generated by the sys-
tem. As an actual example, the change of enthalpy that accompa-
nies the combustion of a litre of gasoline is about 33 megajoules
(1 megajoule, written 1 MJ, is 1 million joules). Therefore we
know without any further calculation that burning a litre of gaso-
line in an open container will provide 33 MJ of heat. A deeper
analysis of the process shows that in the same combustion, the
system has to do about 130 kJ (where 1 kilojoule, written 1 kJ, is
one thousand joules) of work to make room for the gases that are
generated, but that energy is not available to us as heat.
We could extract that extra 130 kJ, which is enough to heat
about half a litre of water from room temperature to its boiling
point, if we prevent the gases from expanding so that all the
energy released in the combustion is liberated as heat. One way
to achieve that, and to obtain all the energy as heat, would be to
arrange for the combustion to take place in a closed container
with rigid walls, in which case it would be unable to expand and
hence would be unable to lose any energy as work. In practice, it
is technologically much simpler to use furnaces that are open to
the atmosphere, and in practice the difference between the two
cases is too small to be worth the effort. However, in formal ther-

modynamics, which is a precise and logical subject, it is essential
to do all the energy accounting accurately and systematically.
In formal thermodynamics the differences between changes in
internal energy and enthalpy must always be borne in mind.
The vaporization of a liquid requires an input of energy
because its molecules must be separated from one another. This
energy is commonly supplied in the form of heat—that is, by
making use of a temperature difference between the liquid and
its surroundings. In former times, the extra energy of the vapour
was termed the ‘latent heat’, because it was released when the
vapour re-condensed to a liquid and was in some sense ‘latent’
in the vapour. The scalding effect of steam is an illustration. In
modern thermodynamic terms, the supply of energy as heat is
identified with the change in enthalpy of the liquid, and the term
‘latent heat’ has been replaced by enthalpy of vaporization. The
enthalpy of vaporization of 1 g of water is close to 2 kJ. The
condensation of 1 g of steam therefore releases 2 kJ of heat, which
may be enough to destroy the proteins of our skin where it comes
in contact. There is a corresponding term for the heat required
to melt a solid: the ‘enthalpy of fusion’. Gram-for-gram, the
enthalpy of fusion is much less than the enthalpy of vaporization,
and we do not get scalded by touching water that is freezing to ice.
We saw in Chapter 1 in the context of the zeroth law that

‘temperature’ is a parameter that tells us the occupation of the
available energy levels of the system. Our task now is to see

how this zeroth-law property relates to the first-law property
of internal energy and the derived heat-accounting property of
enthalpy.
As the temperature of a system is raised and the Boltzmann
distribution acquires a longer tail, populations migrate from
states of lower energy to states of higher energy. Consequently,
the average energy rises, for its value takes into account the ener-
gies of the available states and the numbers of molecules that
occupy each one. In other words, as the temperature is raised,
so the internal energy rises. The enthalpy rises too, but we don’t
need to focus on that separately as it more or less tracks the
changes in internal energy.
The slope of a graph of the value of the internal energy plot-
ted against temperature is called the heat capacity of the system
(symbol C ). 1 Substances with a high heat capacity (water is an
example) require a larger amount of heat to bring about a given
rise in temperature than those with a small heat capacity (air is
an example). In formal thermodynamics, the conditions under
which heating takes place must be specified. For instance, if the
heating takes place under conditions of constant pressure with
the sample free to expand, then some of the energy supplied
as heat goes into expanding the sample and therefore to doing
work. Less energy remains in the sample, so its temperature rises
less than when it is constrained to have a constant volume, and
therefore we report that its heat capacity is higher. The difference
1. To be almost precise, the heat capacity is defined as C = (heat sup-

plied)/(resulting temperature rise). The supply of 1 J of energy as heat to 1 g
of water results in an increase in temperature of about 0.2◦ C.
between heat capacities of a system at constant volume and

at constant pressure is of most practical significance for gases,
which undergo large changes in volume as they are heated in
vessels that are able to expand.
Heat capacities vary with temperature. An important experi-
mental observation that will play an important role in the fol-
lowing chapter is that the heat capacity of every substance falls
to zero when the temperature is reduced towards absolute zero
(T = 0). A very small heat capacity implies that even a tiny trans-
fer of heat to a system results in a significant rise in temperature,
which is one of the problems associated with achieving very low
temperatures when even a small leakage of heat into a sample can
have a serious effect on the temperature (see Chapter 5).
We can get insight into the molecular origin of heat capacity by
thinking—as always—about the distribution of molecules over
the available states. There is a deep theorem of physics called the
fluctuation–dissipation theorem, which implies that the ability of
a system to dissipate (essentially, absorb) energy is proportional
to the magnitudes of the fluctuations about its mean value in a
corresponding property. Heat capacity is a dissipation term: it
is a measure of the ability of a substance to absorb the energy
supplied to it as heat. The corresponding fluctuation term is
the spread of populations over the energy states of the system.
When all the molecules of a system are in a single state, there
is no spread of populations and the ‘fluctuation’ in population
is zero; correspondingly the heat capacity of the system is zero.
As we saw in Chapter 1, at T = 0 only the lowest state of the
system is occupied, so we can conclude from the fluctuation–
dissipation theorem that the heat capacity will be zero too, as
is observed. At higher temperatures, the populations are spread

over a range of states and hence the heat capacity is non-zero, as is
observed.
In most cases, the spread of populations increases with
increasing temperature, so the heat capacity typically increases
with rising temperature, as is observed. However, the relationship
is a little more complex than that because it turns out that the
role of the spread of populations decreases as the temperature
rises, so although that spread increases, the heat capacity does
not increase as fast. In some cases, the increasing spread is bal-
anced exactly by the decrease in the proportionality constant
that relates the spread to the heat capacity, and the heat capacity
settles into a constant value. This is the case for the contribution
of all the basic modes of motion: translation (motion through
space), rotation, and vibration of molecules, all of which settle
into a constant value.
To understand the actual values of the heat capacity of a sub-
stance and the rise in internal energy as the temperature is raised
we first need to understand how the energy levels of a substance
depend on its structure. Broadly speaking, the energy levels lie
close together when the atoms are heavy. Moreover, translational
energy levels are so close together as to form a near continuum,
the rotational levels of molecules in gases are further apart, and
vibrational energy levels—those associated with the oscillations
of atoms within molecules—are widely separated. Thus, as a
gaseous sample is heated, the molecules are readily excited into
higher translational states (in English: they move faster) and, in
all practical cases, they quickly spread over many rotational states
(in English: they rotate faster). In each case the average energy
of the molecules, and hence the internal energy of the system,

increases as the temperature is raised.
The molecules of solids are free neither to move through space
nor to rotate. However, they can oscillate around their average
positions, and take up energy that way. These collective wob-
blings of the entire solid have much lower frequencies than the
oscillations of atoms within molecles and so they can be excited
much more readily. As energy is supplied to a solid sample, these
oscillations are excited, the populations of the higher energy
states increase as the Boltzmann distribution reaches to higher
levels, and we report that the temperature of the solid has risen.
Similar remarks apply to liquids, in which molecular motion
is less constrained than in solids. Water has a very high heat
capacity, which means that to raise its temperature takes a lot
of energy. Conversely, hot water stores a lot of energy, which is
why it is such a good medium for central heating systems (as well
as being cheap), and why the oceans are slow to heat and slow to
cool, with important implications for our climate.
As we have remarked, the internal energy is simply the total
energy of the system, the sum of the energies of all the molecules
and their interactions. It is much harder to give a molecular
interpretation of enthalpy because it is a property contrived to do
the bookkeeping of expansion work and is not as fundamental a
property as internal energy. For the purposes of this account, it
is best to think of the enthalpy as a measure of the total energy,
but to bear in mind that that is not exactly true. In short, as the
temperature of a system is raised its molecules occupy higher and
higher energy levels and as a result their mean energy, the inter-
nal energy, and the enthalpy all increase. Precise fundamental
molecular interpretations can be given only of the fundamental

properties of a system, its temperature, its internal energy, and—
as we shall see in the next chapter—the entropy. They cannot
be given for ‘accounting’ properties, properties that have simply
been contrived to make calculations easier.
The first law is essentially based on the conservation of energy,

the fact that energy can be neither created nor destroyed. Con-
servation laws—laws that state that a certain property does not
change—have a very deep origin, which is one reason why sci-
entists, and thermodynamicists in particular, get so excited when
nothing happens. There is a celebrated theorem, Noether’s the-
orem, proposed by the German mathematician Emmy Noether
(1882–1935), which states that to every conservation law there
corresponds a symmetry. Thus, conservation laws are based on
various aspects of the shape of the universe we inhabit. In the
particular case of the conservation of energy, the symmetry is
that of the shape of time. Energy is conserved because time is
uniform: time flows steadily, it does not bunch up and run faster
then spread out and run slowly. Time is a uniformly structured
coordinate. If time were to bunch up and spread out, energy
would not be conserved. Thus, the first law of thermodynamics
is based on a very deep aspect of our universe and the early
thermodynamicists were unwittingly probing its shape.
3. THE SECOND LAW
The increase in entropy
W hen I gave lectures on thermodynamics to an undergrad-
uate chemistry audience I often began by saying that no
other scientific law has contributed more to the liberation of the
human spirit than the second law of thermodynamics. I hope
that you will see in the course of this chapter why I take that view,
and perhaps go so far as to agree with me.
The second law has a reputation for being recondite, notori-
ously difficult, and a litmus test of scientific literacy. Indeed, the
novelist and former chemist C. P. Snow is famous for having
asserted in his The Two Cultures that not knowing the second
law of thermodynamics is equivalent to never having read a work
by Shakespeare. I actually have serious doubts about whether
Snow understood the law himself, but I concur with his sen-
timents. The second law is of central importance in the whole
of science, and hence in our rational understanding of the uni-
verse, because it provides a foundation for understanding why
any change occurs. Thus, not only is it a basis for understanding
why engines run and chemical reactions occur, but it is also a
foundation for understanding those most exquisite consequences
of chemical reactions, the acts of literary, artistic, and musical
creativity that enhance our culture.
As we have seen for the zeroth and first laws, the formulation
and interpretation of a law of thermodynamics leads us to intro-
duce a thermodynamic property of the system: the temperature,
T , springs from the zeroth law and the internal energy, U , from
the first law. Likewise, the second law implies the existence of
another thermodynamic property, the entropy (symbol S). To
fix our ideas in the concrete at an early stage, it will be helpful
throughout this account to bear in mind that whereas U is a
50 The Second Law: The increase in entropy
measure of the quantity of energy that a system possesses, S

is a measure of the quality of that energy: low entropy means
high quality; high entropy means low quality. We shall elaborate
this interpretation and show its consequences in the rest of the
chapter. At the end of it, with the existence and properties of T ,
U , and S established, we shall have completed the foundations
of classical thermodynamics in the sense that the whole of the
subject is based on these three properties.
A final point in this connection, one that will pervade this
chapter, is that power in science springs from abstraction. Thus,
although a feature of nature may be established by close observa-
tion of a concrete system, the scope of its application is extended
enormously by expressing the observation in abstract terms.
Indeed, we shall see in this chapter that although the second
law was established by observations on the lumbering cast-iron
reality of a steam engine, when expressed in abstract terms it
applies to all change. To put it another way, a steam engine
encapsulates the nature of change whatever the concrete (or cast-
iron) realization of that change. All our actions, from digestion
to artistic creation, are at heart captured by the essence of the
operation of a steam engine.
A steam engine, in its actual but not abstract form, is an iron

fabrication, with boiler, valves, pipes, and pistons. The essence of
a steam engine, though, is somewhat simpler: it consists of a hot
(that is, high temperature) source of energy, a device—a piston
The Second Law: The increase in entropy 51
or turbine—for converting heat into work, and a cold sink, a

place for discarding any unused energy as heat. The last item, the
cold sink, is not always readily discernible, for it might just be the
immediate environment of the engine, not something specifically
designed.
In the early nineteenth century, the French were anxiously
observing from across the Channel England’s industrialization
and becoming envious of her increasing efficiency at using her
abundant supplies of coal to pump water from her mines and
drive her emerging factories. A young French engineer, Sadi
Carnot (1796–1832), sought to contribute to his country’s eco-
nomic and military might by analysing the constraints on the
efficiency of a steam engine. Popular wisdom at the time looked
for greater efficiency in choosing a different working substance—
air, perhaps, rather than steam—or striving to work at danger-
ously higher pressures. Carnot took the then accepted view that
heat was a kind of imponderable fluid that, as it flowed from
hot to cold, was able to do work, just as water flowing down a
gradient can turn a water mill. Although his model was wrong,
Carnot was able to arrive at a correct and astonishing result:
that the efficiency of a perfect steam engine is independent of
the working substance and depends only on the temperatures at
which heat is supplied from the hot source and discarded into the
cold sink.
The ‘efficiency’ of a steam engine—in general, a heat engine—
is defined as the ratio of the work it produces to the heat it
absorbs. Thus, if all the heat is converted into work, with none
discarded, the efficiency is 1. If only half the supplied energy is
converted into work, with the remaining half discarded into the
surroundings, then the efficiency is 0.5 (which would commonly

be reported as a percentage, 50 per cent). Carnot was able to
derive the following expression for the maximum efficiency of
an engine working between the absolute temperatures Tsource and
Tsink :
T sink
Efficiency = 1 −
T source
This remarkably simple formula applies to any thermodynam-
ically perfect heat engine regardless of its physical design. It gives
the maximum theoretical efficiency, and no tinkering with a
sophisticated design can increase the efficiency of an actual heat
engine beyond this limit.
For instance, suppose a power station provided superheated
steam to its turbines at 300◦ C (corresponding to 573 K) and
allows the waste heat to spread into the surroundings at 20◦ C
(293 K), the maximum efficiency is 0.46, so only 46 per cent
of the heat supplied by the burning fuel can be converted into
electricity, and no amount of sophisticated engineering design
can improve on that figure given the two temperatures. The only
way to improve the conversion efficiency would be to lower the
temperature of the surroundings, which in practice is not pos-
sible in a commercial installation, or to use steam at a higher
temperature. To achieve 100 per cent efficiency, the surroundings
would have to be at absolute zero (Tsink = 0) or the steam would
have to be infinitely hot (Tsource = ∞), neither of which is a
practical proposition.
Carnot’s analysis established a very deep property of heat
engines, but its conclusion was so alien to the engineering
prejudices of the time that it had little impact. Such is often
9. The Kelvin (left) and Clausius (right) observations are, respectively,

that a cold sink is essential to the operation of a heat engine and that heat
does not flow spontaneously from a cooler to a hotter body.
the fate of rational thought within society, sent as it may be to

purgatory for a spell. Later in the century, and largely oblivious of
Carnot’s work, interest in heat was rekindled and two intellectual
giants strode on to the stage and considered the problem of
change, and in particular the conversion of heat into work, from
a new perspective.
The first giant, William Thomson, later Lord Kelvin (1824–
1907), reflected on the essential structure of heat engines.
Whereas lesser minds might view the heat source as the crucial
component, or perhaps the vigorously reciprocating piston,
Kelvin—as we shall slightly anachronistically call him—saw
otherwise: he identified the invisible as indispensible, seeing
that the cold sink—often just the undesigned surroundings—
is essential. Kelvin realized that to take away the surroundings
would stop the heat engine in its tracks. To be more precise,
the Kelvin statement of the second law of thermodynamics is as
follows (Figure 9):
no cyclic process is possible in which heat is taken from a hot

source and converted completely into work.
In other words, Nature exerts a tax on the conversion of heat into

work, some of the energy supplied by the hot source must be paid
into the surroundings as heat. There must be a cold sink, even
though we might find it hard to identify and it is not always an
engineered part of the design. The cooling towers of a generating
station are, in this sense, far more important to its operation than
the complex turbines or the expensive nuclear reactor that seems
to drive them.
The second giant was Rudolph Clausius (1822–1888), work-
ing in Berlin. He reflected on a simpler process, the flow of
heat between bodies at different temperatures. He recognized the
everyday phenomenon that energy flows as heat spontaneously
from a body at a high temperature to one at a lower tempera-
ture. The word ‘spontaneous’ is another of those common words
that has been captured by science and dressed in a more precise
meaning. In thermodynamics spontaneous means not needing to
be driven by doing work of some kind. Broadly speaking, ‘spon-
taneous’ is a synonym of ‘natural’. Unlike in everyday language,
spontaneous in thermodynamics has no connotation of speed:
it does not mean fast. Spontaneous in thermodynamics refers to
the tendency for a change to occur. Although some spontaneous
processes are fast (the free expansion of a gas for instance) some
may be immeasurably slow (the conversion of diamond into
graphite, for instance). Spontaneity is a thermodynamic term
that refers to a tendency, not necessarily to its actualization. Ther-
modynamics is silent on rates. For Clausius, there is a tendency
for energy to flow as heat from high temperature to low, but the
spontaneity of that process might be thwarted if an insulator lies
in the way.
Clausius went on to realize that the reverse process, the trans-

fer of heat from a cold system to a hotter one—that is, from a sys-
tem at a low temperature to one a higher temperature—is not
spontaneous. He thereby recognized an asymmetry in Nature: al-
though energy has a tendency to migrate as heat from hot to cold,
the reverse migration is not spontaneous. This somewhat obvi-
ous statement he formulated into what is now known as the Cla-
usius statement of the second law of thermodynamics (Figure 9):
heat does not pass from a body at low temperature to one at high
temperature without an accompanying change elsewhere.
In other words, heat can be transferred in the ‘wrong’ (non-

spontaneous) direction, but to achieve that transfer work must
be done. That is an everyday observation: we can cool objects in
a refrigerator, which involves transferring heat from them and
depositing it in the warmer surroundings, but to do so, we have
to do work—the refrigerator must be driven by connecting it
to a power supply, and the ultimate change elsewhere in the
surroundings that drives the refrigeration is the combustion of
fuel in a power station that may be far away.
The Kelvin and the Clausius statements are both summaries
of observations. No one has ever built a working heat engine
without a cold sink, although they might not have realized one
was present. Nor have they observed a cool object spontaneously
becoming hotter than its surroundings. As such, their statements
are indeed laws of Nature in the sense that I am using the term as
a summary of exhaustive observations. But are there two second
laws? Why is Kelvin’s, for instance, not called the second law and
Clausius’s the third?
10. The equivalence of the Kelvin and Clausius statements. The diagram
on the left depicts the fact that the failure of the Kelvin statement implies
the failure of the Clausius statement. The diagram on the right depicts the
fact that the failure of the Clausius statement implies the failure of the
Kelvin statement.
The answer is that the two statements are logically equiva-

lent. That is, Kelvin’s statement implies Clausius’s and Clausius’s
statement implies Kelvin’s. I shall now demonstrate both sides of
this equivalence.
First, imagine coupling two engines together (Figure 10). The
two engines share the same hot source. Engine A has no cold
sink, but engine B does. We use engine A to drive engine B.
We run engine A, and for the moment presume, contrary to
Kelvin’s statement, that all the heat that A extracts from the hot
source is converted into work. That work is used to drive the
transfer of heat from the cold sink of engine B into the shared
hot sink. The net effect is the restoration of the energy to the
hot sink in addition to whatever engine B transferred out of its
cold sink. That is, heat has been transferred from cold to hot with
no change elsewhere, which is contrary to Clausius’s statement.
Therefore, if Kelvin’s statement were ever found to be false, then

Clausius’s statement would be falsified too.
Now consider the implication of a failure of Clausius’s state-
ment. We build an engine with a hot source and a cold sink, and
run the engine to produce work. In the process we discard some
heat into the cold sink. However, as a cunning part of the design
we have also arranged for exactly the same amount of heat that
we discarded into the cold sink to return spontaneously, contrary
to Clausius’s statement, to the hot source. Now the net effect of
this arrangement is the conversion of heat into work with no
other change elsewhere, for there is no net change in the cold
sink, which is contrary to Kelvin’s statement. Thus, if Clausius’s
statement were ever found to be false, then Kelvin’s statement
would be falsified too.
We have seen that the falsification of each statement of the
second law implies the other, so logically the two statements are
equivalent, and we can treat either as an equivalent phenom-
enological (observation-based) statement of the second law of
thermodynamics.
An interesting side issue is that the discussion so far enables us

to set up a temperature scale that is based purely on mechanical
observations, with the notion of a thermometer built solely from
weights, ropes, and pulleys. You will recall that the zeroth law
implied the existence of a property that we call the temperature,
but apart from the arbitrary scales of Celsius and Fahrenheit,
and a mention of the existence of a more fundamental thermo-

dynamic scale, the definition was left hanging. Kelvin realized
that he could define a temperature scale in terms of work by using
Carnot’s expression for the efficiency of a heat engine.
We shall denote the efficiency, the work done divided by heat
absorbed, of a perfect heat engine by ε (the Greek letter epsilon).
The work done by the engine can be measured by observing
the height through which a known weight is raised, as we have
already seen in the discussion of the first law. The heat absorbed
by the engine can also, in principle at least, be measured by
measuring the fall in a weight. Thus, as we saw in Chapter 2,
the transfer of energy as heat can be measured by observing how
much work must be done to achieve a given change of state in
an adiabatic container, then measuring the work that must be
done to achieve the same change in a diathermic container, and
identifying the difference of the two amounts of work as the
heat transaction in the second process. Thus, in principle, the
efficiency of a heat engine can be measured solely by observing
the rise or fall of a weight in a series of experiments.
Next, according to Carnot’s expression, which in terms of
ε is ε = 1 − Tsink /Tsource , we can write Tsink /Tsource = 1 − ε, or
Tsink = (1 − ε)Tsource . Therefore, to measure the temperature of
the cold sink we simply use our weights to measure the efficiency
of an engine that uses it. Thus, if we find ε = 0.240, then the
temperature of the cold sink must be 0.760Tsource .
This still leaves Tsource unspecified. We can choose a highly
reproducible system, one more reliable than Fahrenheit’s armpit,
and define its temperature as having a certain value, and use
that standard system as the hot source in the engine. In modern
work, a system in which pure liquid water is simultaneously in

equilibrium with both its vapour and ice, the so called triple point
of water, is defined as having a temperature of exactly 273.16
K. The triple point is a fixed property of water: it is unaffected
by any changes in the external conditions, such as the pressure,
so it is highly reproducible. Therefore, in our example, if we
measured by a series of observations on falling weights the effi-
ciency of a heat engine that had a hot source at the temperature
of the triple point of water, and found ε = 0.240, we would be
able to infer that the temperature of the cold sink was 0.760 ×
273.16 K = 208 K (corresponding to −65◦ C). The choice of
the triple point of water for defining the Kelvin scale is entirely
arbitrary, but it has the advantage that anyone in the galaxy can
replicate the scale without any ambiguity, because water has the
same properties everywhere without our having to adjust any
parameters.
The everyday Celsius scale is currently defined in terms of the
more fundamental thermodynamic scale by subtracting exactly
273.15 K from the Kelvin temperature. Thus, at atmospheric
pressure, water is found to freeze at 273 K (to be precise, at about
0.01 K below the triple point, at close to 273.15 K), which corre-
sponds to 0◦ C. Water is found to boil at 373 K, corresponding to
close to 100◦ C. However, these two temperatures are no longer
definitions, as they were when Anders Celsius proposed his scale
in 1742, and must be determined experimentally. Their precise
values are still open to discussion, but reliable values appear
to be 273.152 518 K (+0.002 518◦ C) for the normal freezing
point of water and 373.124 K (99.974◦ C) for its normal boiling
point.
A final point is that the thermodynamic temperature is also

occasionally called the ‘perfect gas temperature’. The latter name
comes from expressing temperature in terms of the properties of
a perfect gas, a hypothetical gas in which there are no interactions
between the molecules. That definition turns out to be identical
to the thermodynamic temperature.
It is inelegant, but of practical utility, to have alternative state-

ments of the second law. Our challenge is to find a single succinct
statement that encapsulates them both. To do so, we follow Clau-
sius and introduce a new thermodynamic function, the entropy,
S. The etymology of the name, from the Greek words for ‘in
turning’ is not particularly helpful; the choice of the letter S,
which does from its shape suggest an ‘in turning’ appears, how-
ever, to be arbitrary, being a letter not used at the time for other
thermodynamic properties, conveniently towards the end of the
alphabet, and an unused neighbour of P , Q, R, T , U , and W, all
of which had already been ascribed other duties.
For mathematically cogent reasons that need not detain us
here, Clausius defined a change in entropy of a system as the
result of dividing the energy transferred as heat by the (absolute,
thermodynamic) temperature at which the transfer took
place:
heat supplied reversibly

Change in entropy =
temperature
I have slipped in the qualification ‘reversibly’, because it is impor-

tant, as we shall see, that the transfer of heat be imagined as
carried out with only an infinitesimal difference in temperature
between the system and its surroundings. In short, it is important
not to stir up any turbulent regions of thermal motion.
We mentioned at the start of the chapter that entropy will
turn out to be a measure of the ‘quality’ of the stored energy.
As this chapter unfolds we shall see what ‘quality’ means. For our
initial encounter with the concept, we shall identify entropy with
disorder: if matter and energy are distributed in a disordered way,
as in a gas, then the entropy is high; if the energy and matter are
stored in an ordered manner, as in a crystal, then the entropy is
low. With disorder in mind, we shall explore the implications of
Clausius’s expression and verify that it is plausible in capturing
the entropy as a measure of the disorder in a system.
The analogy I have used elsewhere to help make plausible
Clausius’s definition of the change in entropy is that of sneez-
ing in a busy street or in a quiet library. A quiet library is the
metaphor for a system at low temperature, with little disorderly
thermal motion. A sneeze corresponds to the transfer of energy
as heat. In a quiet library a sudden sneeze is highly disruptive:
there is a big increase in disorder, a large increase in entropy. On
the other hand, a busy street is a metaphor for a system at high
temperature, with a lot of thermal motion. Now the same sneeze
will introduce relatively little additional disorder: there is only a
small increase in entropy. Thus, in each case it is plausible that
a change in entropy should be inversely proportional to some
power of the temperature (the first power, T itself, as it happens;
not T 2 or anything more complicated), with the greater change
in entropy occurring the lower the temperature. In each case,

the additional disorder is proportional to the magnitude of the
sneeze (the quantity of energy transferred as heat) or some power
of that quantity (the first power, as it happens). Thus, Clausius’s
expression conforms to this simple analogy, and we should bear
the analogy in mind for the rest of the chapter as we see how to
apply the concept of entropy and enrich our interpretation of it.
A change in entropy is the ratio of energy (in joules) trans-
ferred as heat to or from a system to the temperature (in kelvins)
at which it is transferred, so its units are joules per kelvin
(J K−1 ). For instance, suppose we immerse a 1 kW heater 1 in
a tank of water at 20◦ C (293 K), and run the heater for 10 s, we
increase the entropy of the water by 34 J K−1 . If 100 J of energy
leaves a flask of water at 20◦ C, its entropy falls by 0.34 J K−1 . The
entropy of a cup (200 mL) of boiling water—it can be calculated
by a slightly more involved procedure—is about 200 J K−1 higher
than at room temperature.
Now we are ready to express the second law in terms of the
entropy and to show that a single statement captures the Kelvin
and Clausius statements. We begin by proposing the following as
a statement of the second law:
the entropy of the universe increases in the course of any spontan-

eous change.
The key word here is universe: it means, as always in thermo-

dynamics, the system together with its surroundings. There is
1. Power is the rate of supply of energy; it is measured in watts (W), with

1 W = 1 J s−1 (1 joule per second). Thus, a 1 kW heater switched on for 1 h
(3600 s) supplies 3600 kJ of energy.
11. The interpretation of the Kelvin and Clausius statements in terms of

entropy. An engine like that denied by Kelvin’s statement (left) implies a
reduction in entropy. A process like that denied by Clausius’s statement
(right) implies a net reduction in entropy because the decrease in the cold
body is greater than the increase in the hot body (and there are no other
changes).
no prohibition of the system or the surroundings individually

undergoing a decrease in entropy provided that there is a com-
pensating change elsewhere.
To see that Kelvin’s statement is captured by the entropy state-
ment, we consider the entropy changes in the two parts of a heat
engine that has no cold sink (Figure 11). When heat leaves the
hot source, there is a decrease in the entropy of the system. When
that energy is transferred to the surroundings as work, there is
no change in the entropy because changes in entropy are defined
in terms of the heat transferred, not the work that is done. We
shall understand that point more fully later, when we turn to the
molecular nature of entropy. There is no other change. Therefore,
the overall change is a decrease in the entropy of the universe,
which is contrary to the second law. It follows that an engine with
no cold sink cannot produce work.
To see that an engine with a cold sink can produce work, we

think of an actual heat engine. As before, there is a decrease in
entropy when energy leaves the hot sink as heat and there is no
change in entropy when some of that heat is converted into work.
However, provided we do not convert all the energy into work,
we can discard some into the cold sink as heat. There will now
be an increase in the entropy of the cold sink, and provided its
temperature is low enough—that is, it is a quiet enough library—
even a small deposit of heat into the sink can result in an increase
in its entropy that cancels the decrease in entropy of the hot
source. Overall, therefore, there can be an increase in entropy of
the universe, but only provided there is a cold sink in which to
generate a positive contribution. That is why the cold sink is the
crucial part of a heat engine: entropy can be increased only if the
sink is present, and the engine can produce work from heat only
if overall the process is spontaneous. It is worse than useless to
have to drive an engine to make it work!
It turns out, as may be quite readily shown, that the fraction
of energy withdrawn from the hot source that must be dis-
carded into the cold sink, and which therefore is not available
for converting into work, depends only on the temperatures of
the source and sink. Moreover, the minimum energy that must
be discarded, and therefore the achievement of the maximum
efficiency of conversion of heat into work, is given precisely by
Carnot’s formula. 2
2. Suppose q leaves the hot source as heat: the entropy falls by q /Tsource .
Suppose q is discarded into the cold sink: the entropy increases by q /Tsink .
For the overall change in entropy to be positive, the minimum amount of heat
to discard is such that q /Tsink = q /Tsource , and therefore q = qT sink /Tsource .
That means that the maximum amount of work that can be done is q − q ,
Now consider the Clausius statement in terms of entropy. If

a certain quantity of energy leaves the cold object as heat, the
entropy decreases. This is a large decrease, because the object is
cold—it is a quiet library. The same quantity of heat enters the
hot object. The entropy increases, but because the temperature
is higher—the object is a busy street—the resulting increase
in entropy is small, and certainly smaller than the decrease in
entropy of the cold object. Overall, therefore, there is a decrease
in entropy, and the process is not spontaneous, exactly as
Clausius’s statement implies.
Thus, we see that the concept of entropy captures the two
equivalent phenomenological statements of the second law
and acts as the signpost of spontaneous change. The first law
and the internal energy identify the feasible change among all
conceivable changes: a process is feasible only if the total energy
of the universe remains the same. The second law and entropy
identify the spontaneous changes among these feasible changes:
a feasible process is spontaneous only if the total entropy of the
universe increases.
It is of some interest that the concept of entropy greatly trou-
bled the Victorians. They could understand the conservation of
energy, for they could presume that at the Creation God had
endowed the world with what He would have judged infallibly as
exactly the right amount, an amount that would be appropriate
for all time. What were they to make of entropy, though, which
somehow seemed to increase ineluctably. Where did this entropy
or q (1 − Tsink /Tsource ). The efficiency is this work divided by the heat supplied
(q ), which gives efficiency = 1 − Tsink /Tsource , which is Carnot’s formula.
spring from? Why was there not an exact, perfectly and eternally
judged amount of the God-given stuff?
To resolve these matters and to deepen our understanding of
the concept, we need to turn to the molecular interpretation
of entropy and its interpretation as a measure, in some sense,
of disorder.
With entropy as a measure of disorder in mind, the change in

entropy accompanying a number of processes can be predicted
quite simply, although the actual numerical change takes more
effort to calculate than we need to display in this introduction.
For example, the isothermal (constant temperature) expansion
of a gas distributes its molecules and their constant energy over a
greater volume, the system is correspondingly less ordered in the
sense that we have less chance of predicting successfully where a
particular molecule and its energy will be found, and the entropy
correspondingly increases.
A more sophisticated way of arriving at the same conclusion,
and one that gives a more accurate portrayal of what ‘disorder’
actually means, is to think of the molecules as distributed over
the energy levels characteristic of particles in a box-like region.
Quantum mechanics can be used to calculate these allowed
energy levels (it boils down to computing the wavelengths of the
standing waves that can fit between rigid walls, and then inter-
preting the wavelengths as energies). The central result is that as
the walls of the box are moved apart, the energy levels fall and
become less widely separated (Figure 12). At room temperature,
12. The increase in entropy of a collection of particles in an expanding

box-like region arises from the fact that as the box expands, the allowed
energies come closer together. Provided the temperature remains the same,
the Boltzmann distribution spans more energy levels, so the chance of
choosing a molecule from one level in a blind selection decreases. That
is, the disorder and the entropy increase as the gas occupies a greater
volume.
billions of these energy levels are occupied by the molecules,

the distribution of populations being given by the Boltzmann
distribution characteristic of that temperature. As the box
expands, the Boltzmann distribution spreads over more energy
levels and it becomes less probable that we can specify which
energy level a molecule would come from if we made a blind
selection of molecules. This increased uncertainty of the precise
energy level a molecule occupies is what we really mean by the
‘disorder’ of the system, and corresponds to an increased entropy.
A similar picture accounts for the change in entropy as the
temperature of a gaseous sample is raised. A simple calculation in
classical thermodynamics based on Clausius’s definition leads us

to expect an increase in entropy with temperature. That increase
in molecular terms can be understood, because as the temper-
ature increases at constant volume, the Boltzmann distribution
acquires a longer tail, corresponding to the occupation of a wider
range of energy levels. Once again, the probability that we can
predict which energy level a molecule comes from in a blind
selection corresponds to an increase in disorder and therefore to
a higher entropy.
This last point raises the question of the value of the entropy
at the absolute zero of temperature (at T = 0). According to
the Boltzmann distribution, at T = 0 only the lowest state (the
‘ground state’) of the system is occupied. That means that we
can be absolutely certain that in a blind selection we will select
a molecule from that single ground state: there is no uncertainty
in the distribution of energy, and the entropy is zero.
These considerations were put on a quantitative basis by Lud-
wig Boltzmann, who proposed that the so-called absolute entropy
of any system could be calculated from a very simple formula:
S = k log W
The constant k is Boltzmann’s constant, which we encountered
in Chapter 1 in the relation between ‚ and T , namely ‚ = 1/kT,
and appears here simply to ensure that changes in entropy
calculated from this equation have the same numerical value as
those calculated from Clausius’s expression. 3 Of much greater
3. If humanity with prescient wisdom had adopted ‚ as a natural measure of
temperature, Clausius’s expression would have been written change in entropy =
‚qrev and Boltzmann’s S = log W; entropy would have been a pure number. We
can but weep for lost simplicity.
significance is the quantity W, which is a measure of the number

of ways that the molecules of a system can be arranged to
achieve the same total energy (the ‘weight’ of an arrangement).
This expression is much harder to implement than the classical
thermodynamic expression, and really belongs to the domain
of statistical thermodynamics, which is not the subject of this
volume. Suffice it to say that Boltzmann’s formula can be
used to calculate both the absolute entropies of substances,
especially if they have simple structures, like a gas, and changes
in entropy that accompany various changes, such as expansion
and heating. In all cases, the expressions for the changes in
entropy correspond exactly to those deduced from Clausius’s
definition, and we can be confident that the classical entropy
and the statistical entropy are the same.
It is an incidental footnote of a personal history that the equa-
tion S = k log W is inscribed on Boltzmann’s tombstone as his
wonderful epitaph, even though he never wrote down the equa-
tion explicitly (it is due to Max Planck). He deserves his constant
even if we do not.
There are various little wrinkles in the foregoing about which

we now need to own up. Because the Clausius expression tells
us only the change in entropy, it allows us to measure the
entropy of a substance at room temperature relative to its value at
T = 0. In many cases the value calculated for room temperature
corresponds within experimental error to the value calculated
from Boltzmann’s formula using data about molecules obtained

from spectroscopy, such as bond lengths and bond angles. In
some cases, however, there is a discrepancy, and the thermo-
dynamic entropy differs from the statistical entropy.
We have assumed without comment that there is only one
state of lowest energy; one ground state, in which case W = 1 at
T = 0 and the entropy at that temperature is zero. That is, in
the technical parlance of quantum mechanics, we assumed that
the ground state was ‘non-degenerate’. 4 In some cases, that is not
true, and in such cases there may be many different states of the
system corresponding to the lowest energy. We could say that the
ground states of these systems are highly degenerate and denote
the number of states that correspond to that lowest energy as
D. (I give a visualizable example in a moment.) If there are D
such states, then even at absolute zero we have only 1 chance in
D of predicting which of these degenerate states a molecule will
come from in a blind selection. Consequently, there is disorder
in the system even at T = 0 and its entropy is not zero. This non-
zero entropy of a degenerate system at T = 0 is called the residual
entropy of the system.
Solid carbon monoxide provides one of the simplest exam-
ples of residual entropy. A carbon monoxide molecule, CO, has
a highly uniform distribution of electric charge (technically, it
has only a very tiny electric dipole moment), and there is little
difference in energy if in the solid the molecules lie . . . CO CO
CO . . . , or . . . CO OC CO . . . , or any other random arrangement
4. In quantum mechanics, the term ‘degeneracy’, another hijacked term,

refers to the possibility that several different states (for instance, planes of
rotation or direction of travel) correspond to the same energy.
of direction. In other words, the ground state of a solid sample of

carbon monoxide is highly degenerate. If each molecule can lie in
one of two directions, and there are N molecules in the sample,
then D = 2 N . Even in 1 g of solid carbon monoxide there are
2 × 1022 molecules, so this degeneracy is far from negligible! (Try
calculating the value of D.) The value of the residual entropy is
k log D, which works out to 0.21 J K−1 for a 1 g sample, in good
agreement with the value inferred from experiment.
Solid carbon monoxide might seem to be a somewhat rarefied
example and of little real interest except as a simple illustra-
tion. There is one common substance, though, of considerable
importance that is also highly degenerate in its ground state: ice.
We do not often think—perhaps ever—of ice being a degenerate
solid, but it is, and the degeneracy stems from the location of the
hydrogen atoms around each oxygen atom.
Figure 13 shows the origin of ice’s degeneracy. Each water
molecule is H2 O, with two short, strong O–H bonds at about
104◦ to each other. The molecule is electrically neutral overall,
but the electrons are not distributed uniformly, and each oxygen
atom has patches of net negative charge on either side of the
molecule, and each hydrogen atom is slightly positively charged
on account of the withdrawal of electrons from it by the electron-
hungry oxygen atom. In ice, each water molecule is surrounded
by others in a tetrahedral arrangement, but the slightly positively
charged hydrogen atoms of one molecule are attracted to one
of the patches of slight negative charge on the oxygen atom of
a neighbouring water molecule. This link between molecules
is called a hydrogen bond, and is denoted O–H· · ·O. The link
is responsible for the residual entropy of ice, because there is a
13. The residual entropy of water, reflecting its ‘degeneracy’ at T = 0,

arises from the variation in the locations of hydrogen atoms (the small
white spheres) between oxygen atoms (the shaded spheres). Although each
oxygen atom is closely attached to two hydrogen atoms and makes a more
distant link to a hydrogen atom of each of two neighbouring water mole-
cules, there is some freedom in the choice of which links are close and
which are distant. Two of the many arrangements are shown here.
randomness in whether any particular link is O–H· · ·O or

O· · ·H–O. Each water molecule must have two short O–H bonds
(so that it is recognizable as an H2 O molecule), and two H· · ·O
links to two neighbours, but which two are short and which two
are long is almost random. When the statistics of this variability
is analysed, it turns out that the residual entropy of 1 g of ice
should be 0.19 J K−1 , in good agreement with the value inferred
from experiment.
The concept of entropy is the foundation of the operation of

heat engines, heat pumps, and refrigerators. We have already
seen that a heat engine works because heat is deposited in a

cold sink and generates disorder there that compensates, and in
general more than compensates, for any reduction in entropy
due to the extraction of energy as heat from the hot source.
The efficiency of a heat engine is given by the Carnot expres-
sion. We see from that expression that the greatest efficiency is
achieved by working with the hottest possible source and the
coldest possible sink. Therefore, in a steam engine, a term that
includes steam turbines as well as classical piston-based engines,
the greatest efficiency is achieved by using superheated steam.
The fundamental reason for that design feature is that the high
temperature of the source minimizes the entropy reduction of
the withdrawal of heat (to go unnoticed, it is best to sneeze in
a very busy street), so that least entropy has to be generated in
the cold sink to compensate for that decrease, and therefore that
more energy can be used to do the work for which the engine is
intended.
A refrigerator is a device for removing heat from an object and
transferring that heat to the surroundings. This process does not
occur spontaneously because it corresponds to a reduction in
total entropy. Thus, when a given quantity of heat is removed
from a cool body (a quiet library, in our sneeze analogy),
there is a large decrease in entropy. When that heat is released
into warmer surroundings, there is an increase in entropy, but
the increase is smaller than the original decrease because the
temperature is higher (it is a busy street). Therefore, overall
there is a net decrease in entropy. We used the same argument in
the discussion of Clausius’s statement of the second law, which
applies directly to this arrangement. A crude restatement of
Clausius’s statement is that refrigerators don’t work unless you

turn them on.
In order to achieve a net increase of entropy, we must release
more energy into the surroundings than is extracted from the
cool object (we must sneeze more loudly in the busy street). To
achieve that increase, we must add to the flow of energy. This
we can do by doing work on the system, for the work we do
adds to the energy stream (Figure 14). When we do work the
original energy extracted from the cool body is augmented to
heat + work, and that total energy is released into the warmer
surroundings. If enough work is done on the system, the release
of a large amount of energy into the warm surroundings gives a
large increase in entropy and overall there is a net increase in
14. The processes involved in a refrigerator and a heat pump. In a refrig-

erator (left), the entropy of the warm surroundings is increased by at
least the amount by which the entropy of the system (the interior of the
refrigerator) is decreased; this increase is achieved by adding to the flow
of energy by doing work. In a heat pump (right), the same net increase in
entropy is achieved, but in this case the interest lies in the energy supplied
to the interior of the house.
entropy and the process can occur. Of course, to generate the

work to drive the refrigerator, a spontaneous process must occur
elsewhere, as in a distant power station.
The efficiency of refrigeration is reported as the ‘coefficient of
performance’ of the arrangement. This quantity is defined as the
ratio of the heat removed from a cool object to the work that
must be done in order to achieve that transfer. The higher the
coefficient of performance the less work we have to do to achieve
a given transfer—the less power we have to draw from the supply,
so the more efficient the refrigerator. By a calculation very similar
to that in footnote 1, 5 we can conclude that the best coefficient of
performance that can be achieved by any arrangement when the
object (the food) to be cooled is at a temperature Tcold and the
surroundings (the kitchen) is at Tsurroundings is
1
Coefficient of performance (refrigerator) = Tsurroundings
Tcold
−1
◦
For instance, if the cold object is cold water at 0 C (273 K) and
the refrigerator is in a room at 20◦ C (293 K), then the coefficient
of performance is 14, and to remove 10 kJ of energy from the
freezing water, which is enough to freeze about 30 g of the water
5. Suppose we want to extract an amount q of energy as heat from a cold

object at a temperature Tcold and deposit it into the surroundings at a temper-
ature Tsurroundings . The entropy of the object falls by q /Tcold . The entropy of the
surroundings increases by q /Tsurroundings . However, if we do an amount of work
w on the system, it joins the energy stream and the heat released is increased to
q + w. As a result, entropy of the surroundings increases by (q + w)/Tsurroundings .
For this change to cancel the decrease in entropy of the cold object we arrange
the value of w so that (q + w)/Tsurroundings = q /Tcold , or w = (Tsurroundings /
Tcold – 1)q . Therefore, as the coefficient of performance, c , is defined as c = q /w,
we arrive at the result quoted. Note that to calculate the work required to
transfer a given amount of heat, we use w = q /c .
to ice, under ideal conditions we need to do about 0.71 kJ of

work. Actual refrigerators are much less efficient than this ther-
modynamic limit, not least because heat leaks in from outside
and not all the energy supplied to do work joins the energy
stream. Air conditioning is essentially refrigeration, and this cal-
culation indicates why it is so expensive—and environmentally
damaging—to run. It takes a lot of energy to fight Nature when
she wields the second law.
When a refrigerator is working, the energy released into the
surroundings is the sum of that extracted from the cooled object
and that used to run the apparatus. This remark is the basis of
the operation of a heat pump, a device for heating a region (such
as the interior of a house) by pumping heat from the outside into
the interior. A heat pump is essentially a refrigerator, with the
cooled object the outside world and the heat transfer arranged to
be in the region to be heated. That is, our interest is in the back
of the refrigerator, not its interior. The coefficient of performance
of a heat pump is defined as the ratio of the total energy released
as heat into the region to be heated (at a temperature Tinterior ), to
the work done in order to achieve that release. By the same type
of calculation as already done in footnotes 4 and 5 ( and which
in this case is left to the reader), it turns out that the theoretical
best coefficient of performance when the region from which the
heat is extracted is at a temperature Tsurroundings is
1
Coefficient of performance (heat pump) = Tsurroundings
1− Tinterior
Therefore, if the region to be heated is at 20◦ C (293 K) and the

surroundings are at 0◦ C (273 K), the coefficient of performance
is 15. Thus, to release 1000 J into the interior, we need do only

67 J of work. In other words, a heat pump rated at 1 kW behaves
like a 15 kW heater.
We began this chapter by asserting that we are all steam engines.

With a sufficiently abstract interpretation of ‘steam engine’, that
is most definitely true. Wherever structure is to be conjured
from disorder, it must be driven by the generation of greater
disorder elsewhere, so that there is a net increase in disorder
of the universe, with disorder understood in the sophisticated
manner that we have sketched. That is clearly true for an actual
heat engine, as we have seen. However, it is in fact universally
true.
For instance, in an internal combustion engine, the com-
bustion of a hydrocarbon fuel results in the replacement of a
compact liquid by a mixture of gases that occupies a volume
over 2000 times greater (and still 600 times greater if we allow
for the oxygen consumed). Moreover, energy is released by the
combustion, and that energy disperses into the surroundings.
The design of the engine captures this dispersal in disorder and
uses it to build, for instance, a structure from a less ordered pile of
bricks, or drive an electric current (an orderly flow of electrons)
through a circuit.
The fuel might be food. The dispersal that corresponds to
an increase in entropy is the metabolism of the food and the
dispersal of energy and matter that that metabolism releases. The
structure that taps into that dispersal is not a mechanical chain

of pistons and gears, but the biochemical pathways within the
body. The structure that those pathways cause to emerge may
be proteins assembled from individual amino acids. Thus, as we
eat, so we grow. The structures may be of a different kind: they
may be works of art. For another structure that can be driven
into existence by coupling to the energy released by ingestion and
digestion consists of organized electrical activity within the brain
constructed from random electrical and neuronal activity. Thus,
as we eat, we create: we create works of art, of literature, and of
understanding.
The steam engine, in its abstract form as a device that gen-
erates organized motion (work) by drawing on the dissipation
of energy, accounts for all the processes within our body. More-
over, that great steam engine in the sky, the Sun, is one of the
great fountains of construction. We all live off the spontaneous
dissipation of its energy, and as we live so we spread disorder into
our surroundings: we could not survive without our surround-
ings. In his seventeenth meditation, John Donne was unknow-
ingly expressing a version of the second law when he wrote, two
centuries before Carnot, Joule, Kelvin, and Clausius, that no man
is an island.
4. F R E E E N E RG Y
The availability of work
F ree energy? Surely not! How can energy be free? Of course,
the answer lies in a technicality. By free energy we do not
mean that it is monetarily free. In thermodynamics, the freedom
refers to the energy that is free to do work rather than just tumble
out of a system as heat.
We have seen that when a combustion occurs at constant
pressure, the energy that may be released as heat is given by
the change of enthalpy of the system. Although there may be a
change in internal energy of a certain value, the system in effect
has to pay a tax to the surroundings in the sense that some of
that change in internal energy must be used to drive back the
atmosphere in order to make room for the products. In such
a case, the energy that can be released as heat is less than the
change in internal energy. It is also possible for there to be a
tax refund in the sense that if the products of a reaction occupy
less volume than the reactants, then the system can contract. In
this case, the surroundings do work on the system, energy is
transferred into it, and the system can release more heat than
is given by the change in internal energy: the system recycles the
incoming work as outgoing heat. The enthalpy, in short, is an
accounting tool for heat that takes into account automatically
the tax payable or repayable as work and lets us calculate the heat
output without having to calculate the contributions of work in
a separate calculation.
The question that now arises is whether a system must pay a
tax to the surroundings in order to produce work. Can we extract
the full change in internal energy as work, or must some of that
change be transferred to the surroundings as heat, leaving less to
be used to do work? Must there be a tax, in the form of heat, that
82 Free Energy: The availability of work
a system has to pay in order to do work? Could there even be a

tax refund in the sense that we can extract more work than the
change in internal energy leads us to expect? In short, by analogy
with the role of enthalpy, is there a thermodynamic property that
instead of focusing on the net heat that a process can release
focuses on the net work instead?
We found the appropriate property for heat, the enthalpy, by
considering the first law. We shall find the appropriate property
for work by considering the second law and entropy, because a
process can do work only if it is spontaneous: non-spontaneous
processes have to be driven by doing work, so they are worse than
useless for producing work.
To identify spontaneous processes we must note the cru-
cially important aspect of the second law that it refers to the
entropy of the universe, the sum of the entropies of the sys-
tem and the surroundings. According to the second law, a
spontaneous change is accompanied by an increase in entropy
of the universe. An important feature of this emphasis on
the universe is that a process may be spontaneous, and work
producing, even though it is accompanied by a decrease in
entropy of the system provided that a greater increase occurs
in the surroundings and the total entropy increases. Whenever
we see the apparently spontaneous reduction of entropy, as
when a structure emerges, a crystal forms, a plant grows, or a
thought emerges, there is always a greater increase in entropy
elsewhere than accompanies the reduction in entropy of the
system.
To assess whether a process is spontaneous and therefore capa-
ble of producing work, we have to assess the accompanying
Free Energy: The availability of work 83
entropy changes in both the system of interest and the surround-

ings. It is inconvenient to have to do two separate calculations,
one for the system and one for the surroundings. Provided we are
prepared to restrict our interest to certain types of change, there
is a way to combine the two calculations into one and to carry
out the calculation by focusing on the properties of the system
alone. By proceeding in that way, we shall be able to identify the
thermodynamic property that we can use to assess the work that
can be extracted from a process without having to calculate the
‘heat tax’ separately.
The clever step is to realize that if we limit changes to those taking

place at constant volume and temperature, then the change in
entropy of the surroundings can be expressed in terms of the
change in internal energy of the system. That is because at con-
stant volume, the only way that the internal energy can change
in a closed system is to exchange energy as heat with the sur-
roundings, and that heat can be used to calculate the change in
entropy of the surroundings by using the Clausius expression for
the entropy.
When the internal energy of a constant-volume, closed system
changes by U , the whole of that change in energy must be due
to a heat transaction with the surroundings. If there is an increase
in internal energy of the system (for instance, if U = +100 J),
then heat equal to U (that is, 100 J) must flow in from the sur-
roundings. The surroundings lose that amount of energy as heat,
and so their entropy changes by −U/T , a decrease. If there is

a decrease in internal energy of the system, U is negative (for
instance, if U = −100 J) and an equal amount of heat (in this
case, 100 kJ) flows into the surroundings. Their entropy therefore
increases by −U/T (this is a positive quantity because U is
negative when U decreases). In either case, therefore, the total
change in entropy of the universe is S(total) = S − U/T ,
where S is the change in entropy of the system. This expression
is in terms of the properties of the system alone. In a moment
we shall use it in the form −T S(total) = U − T S, which
is obtained by multiplying both sides by −T and changing the
order of terms on the right.
To tidy up the calculation, we introduce a combination of
the internal energy and the entropy of the system called the
Helmholtz energy, denoted A, and defined as A = U − TS. The
German physiologist and physicist Hermann von Helmholtz
(1821–1894), after whom this property is named, was responsible
for formulating the law of conservation of energy as well as mak-
ing other major contributions to the science of sensation, colour
blindness, nerve propagation, hearing, and thermodynamics in
general.
At constant temperature a change in the Helmholtz energy
stems from changes in U and S, and A = U − T S, exactly
as we have just found for −T S(total). So, a change in A
is just a disguised form of the change in total entropy of the
universe when the temperature and volume of the system are
constant. The important implication of this conclusion is that,
because spontaneous changes correspond to positive changes
(increases) in the total entropy of the universe, provided we limit
our attention to processes at constant temperature and volume,

spontaneous changes correspond to a decrease in Helmholtz
energy of the system. The restriction of the conditions to constant
temperature and volume has allowed us to express spontaneity
solely in terms of the properties of the system: its internal energy,
temperature, and entropy.
It probably seems more natural that a spontaneous change
corresponds to a decrease in a quantity: in the everyday world,
things tend to fall down, not up. However, don’t be misled by the
seductions of familiarity: the natural tendency of A to decrease is
just an artefact of its definition. Because the Helmholtz energy
is a disguised version of the total entropy of the universe, the
change in direction from ‘total entropy up’ to ‘Helmholtz energy
down’ simply reflects how A is defined. If you examine the
expression for A without its derivation in mind, you will see
that a negative value will be obtained if U is negative (a low-
ering of internal energy of the system) and S is positive. You
might then jump to the conclusion that systems tend towards
lower internal energy and higher entropy. That would be a wrong
interpretation. The fact that a negative U favours spontaneity
stems from the fact that it represents the contribution (through
−U/T ) of the entropy of the surroundings. The only criterion
of spontaneous change in thermodynamics is the increase in total
entropy of the universe.
As well as the Helmholtz energy being a signpost of spon-
taneous change it has another important role: it tells us the
maximum work that can be extracted when a process occurs at
constant temperature. That should be quite easy to see: it fol-
lows from the Clausius expression for the entropy (S = q rev /T
rearranged into q rev = T S) that T S is the heat transferred to

the surroundings in a reversible process; but U is equal to the
sum of the heat and work transactions with the surroundings,
and the difference left after allowing for the heat transferred,
the value of U − T S, is the change in energy due to doing
work alone. It is for this reason that A is also known as the
‘work function’ and given the symbol A (because Arbeit is the
German word for work). More commonly, though, A is called a
free energy, suggesting that it indicates the energy in a system that
is free to do work.
The last point becomes clearer once we think about the mo-
lecular nature of the Helmholtz energy. As we saw in Chapter 2,
work is uniform motion in the surroundings, as in the moving
of all the atoms of a weight in the same direction. The term TS
that appears in the definition of A = U − TS has the dimensions
of an energy, and can be thought of as a measure of the energy
that is stored in a disordered way in the system for which U is
the total energy. The difference U − T S is therefore the energy
that is stored in an orderly way. We can then think of only the
energy stored in an orderly way as being available to cause orderly
motion, that is, work, in the surroundings. Thus, only the differ-
ence U − T S of the total energy and the ‘disordered’ energy is
energy that is free to do work.
A more precise way of understanding the Helmholtz energy is
to think about the significance of changes in its value. Suppose a
certain process occurs in a system that causes a change in internal
energy U and happens to correspond to a decrease in entropy,
so S is negative. The process will be spontaneous and able to
produce work only if the entropy of the surroundings increases
15. On the left a process occurs in a system that causes a change in internal
energy U and a decrease in entropy. Energy must be lost as heat to the
surroundings in order to generate a compensating entropy there, so less
than U can be released as work. On the right, a process occurs with an
increase in entropy, and heat can flow in to the system yet still correspond
to an increase in total entropy; as a result, more than U can be released
as work.
by a compensating amount, namely S (Figure 15). For that

increase to occur, some of the change in internal energy must
be released as heat, for only heat transactions result in changes
in entropy. To achieve an increase in entropy of magnitude S,
according to the Clausius expression, the system must release
a quantity of heat of magnitude T S. That means that only
U − T S can be released as work.
According to this discussion, T S is a tax that the surround-
ings demand from the system in order to compensate for the
reduction in entropy of the system, and only U − T S is left
for the system to pay out as work. However, suppose the entropy
of the system happens to increase in the course of the process.
In that case the process is already spontaneous, and no tax need
be paid to the surroundings. In fact, it is better than that, because
the surroundings can be allowed to supply energy as heat to
the system, because they can tolerate a decrease in entropy yet
the entropy of the universe will still increase. In other words,

the system can receive a tax refund. That influx of energy as heat
increases the internal energy of the system and the increase can
be used to do more work than in the absence of the influx. That
too, is captured by the definition of the Helmholtz energy, for
when S is negative, −T S is a positive quantity and adds to
U rather than subtracting from it, and A is bigger than U .
In this case, more work can be extracted than we would expect if
we considered only U .
Some numbers might give these considerations a sense of real-
ity. When 1 L of gasoline is burned it produces carbon dioxide
and water vapour. The change in internal energy is 33 MJ, which
tells us that if the combustion takes place at constant volume
(in a sturdy, sealed container), then 33 MJ will be released as
heat. The change in enthalpy is 0.13 MJ less than the change in
internal energy. This figure tells us that if the combustion takes
place in a vessel open to the atmosphere, then slightly less (0.13
MJ less, in fact) than 33 MJ will be released as heat. Notice that
less heat is released in the second arrangement because 0.13 MJ
has been used to drive back the atmosphere to make room for the
gaseous products and so less is available as heat. The combustion
is accompanied by an increase in entropy because more gas is
produced than is consumed (sixteen CO2 molecules and eighteen
H2 O molecules are produced for every twenty-five O2 molecules
that are consumed, a net increase of nine gas molecules), and
it may be calculated that S = +8 kJ K−1 . It follows that the
change in Helmholtz energy of the system is −35 MJ. Thus, if
the combustion took place in an engine, the maximum amount
of work that could be obtained is 35 MJ. Note that this is larger
than the value of U because the increase in entropy of the

system has opened the possibility of heat flowing into the system
as a tax refund and there being a corresponding decrease in the
surroundings yet leaving the change in total entropy positive. It
is, perhaps, refreshing to note that you get a tax refund for every
mile you drive; but this is Nature’s refund, not the Chancellor’s.
The discussion so far refers to all kinds of work. In many cases we

are not interested in expansion work but the work, for example,
that can be extracted electrically from an electrochemical cell
or the work done by our muscles as we move around. Just as
the enthalpy (H = U + pV) is used to accommodate expansion
work automatically when that is not of direct interest, it is pos-
sible to define another kind of free energy that takes expansion
work into account automatically and focuses our attention on
non-expansion work. The Gibbs energy, which is denoted G ,
is defined as G = A + pV. Josiah Willard Gibbs (1839–1903),
after whom this property is named, is justifiably regarded as
a founding father of chemical thermodynamics. He worked at
Yale University throughout his life and was noted for his public
reticence. His extensive and subtle work was published in what
we now consider to be an obscure journal (The Transactions of the
Connecticut Academy of Science) and was not appreciated until it
was interpreted by his successors.
In the same way as A tells us the total work that a process
may do at constant temperature, the change in the Gibbs energy,
G , tells us the amount of non-expansion work that a process

can do provided the change is taking place at constant tem-
perature and pressure. Just as it is not really possible to give
a molecular interpretation of the enthalpy, which is really just
a clever accounting device, it is not possible to give a simple
explanation of the molecular nature of the Gibbs energy. It is
good enough for our purposes to think of it like the Helmholtz
energy, as a measure of the energy that is stored in an orderly way
and is therefore free to do useful work.
There is another ‘just as’ to note. Just as a change in the
Helmholtz energy is a disguised expression for the change in total
entropy of the universe when a process takes place at constant
volume (remember that A = –T S(total)), with spontaneous
processes characterized by a decrease in A, so the change in
Gibbs energy can be identified with a change in total entropy for
processes that occur at constant pressure: G = –T S(total).
Thus, the criterion of spontaneity of a process at constant pres-
sure is that G is negative:
at constant volume, a process is spontaneous if it corresponds to a

decrease in Helmholtz energy.
at constant pressure, a process is spontaneous if it corresponds to
a decrease in Gibbs energy.
In each case, the underlying origin of the spontaneity is the

increase in entropy of the universe, but in each case we can
express that increase in terms of the properties of the system
alone and do not have to worry about doing a special calculation
for the surroundings.
The Gibbs energy is of the greatest importance in chemistry

and in the field of bioenergetics, the study of energy utilization
in biology. Most processes in chemistry and biology occur at
constant temperature and pressure, and so to decide whether
they are spontaneous and able to produce non-expansion work
we need to consider the Gibbs energy. In fact, when chemists and
biologists use the term ‘free energy’ they almost always mean the
Gibbs energy.
There are three applications that I shall discuss here. One is the
thermodynamic description of phase transitions (freezing and
boiling, for instance; a ‘phase’ is a form of a given substance, such
as the solid, liquid, and vapour phases of water), another is the
ability of one reaction to drive another in its non-spontaneous
direction (as when we metabolize food in our bodies and then
walk or think), and the third is the attainment of chemical equi-
librium (as when an electric battery becomes exhausted).
The Gibbs energy of a pure substance decreases as the
temperature is raised. We can see how to draw that conclusion
from the definition G = H – TS, by noting that the entropy of a
pure substance is invariably positive. Therefore, as T increases,
TS becomes larger and subtracts more and more from H, and
G consequently falls. The Gibbs energy of 100 g of liquid water,
for instance, behaves as shown in Figure 16 by the line labelled
‘liquid’. The Gibbs energy of ice behaves similarly. However,
because the entropy of 100 g of ice is lower than that of 100 g of
Gibbs energy Gas
Liquid
Solid
0°C 100°C
Temperature
16. The decrease in Gibbs energy with increasing temperature for three
phases of a substance. The most stable phase corresponds to the lowest
Gibbs energy; thus the solid is most stable at low temperatures, then the
liquid, and finally the gas (vapour). If the gas line falls more steeply, it
might intersect the solid line before the liquid line does, in which case
the liquid is never the stable phase and the solid sublimes directly to a
vapour.
water—because the molecules are more ordered in a solid than

the jumble of molecules that constitute a liquid—the Gibbs
energy does not fall away as steeply, and is shown by the line
labelled ‘solid’ in the illustration. The entropy of 100 g of
water vapour is much greater than that of the liquid because
the molecules of a gas occupy a much greater volume and are
distributed randomly over it. As a result, the Gibbs energy of
the vapour decreases very sharply with increasing temperature,
as shown by the line labelled ‘gas’ in the illustration. At low
temperatures we can be confident that the enthalpy of the solid
is lower than that of the liquid (because it takes energy to melt a
solid) and the enthalpy of the liquid lies below that of the vapour
(because it takes energy to vaporize a liquid). That is why we

have drawn the Gibbs energies starting in their relative positions
on the left of the illustration.
The important feature is that although the Gibbs energy of the
liquid is higher than that of the solid at low temperatures, the two
lines cross at a particular temperature (0◦ C, 273 K, as it happens,
at normal atmospheric pressure) and from there on the liquid
has a lower Gibbs energy than the solid. We have seen that the
natural direction of change at constant pressure is to lower Gibbs
energy (corresponding, remember, to greater total entropy), so
we can infer that at low temperature the solid form of water is the
most stable, but that once the temperature reaches 0◦ C the liquid
becomes more stable and the substance spontaneously melts.
The Gibbs energy of the liquid remains the lowest of the three
phases until the steeply falling line for the vapour intersects it. For
water, at normal atmospheric pressure that intersection occurs
at 100◦ C (373 K), and from that temperature on, the vapour is
the most stable form of water. The system spontaneously falls to
lower Gibbs energy, so vaporization is spontaneous above 100◦ C:
the liquid boils.
There is no guarantee that the ‘liquid’ line intersects the ‘solid’
line before the ‘vapour’ line has plunged down and crossed the
‘solid’ line first. In such a case, the substance will make a direct
transition from solid to vapour without melting to an inter-
mediate liquid phase. This is the process called sublimation. Dry
ice (solid carbon dioxide) behaves in this way, and converts
directly to carbon dioxide gas.
All phase changes can be expressed thermodynamically
in a similar way, including melting, freezing, condensation,
vaporization, and sublimation. More elaborate discussions also

enable us to discuss the effect of pressure on the temperatures at
which phase transitions occur, for pressure affects the locations
of the lines showing the dependence of Gibbs energy on temper-
ature in different ways, and the intersection points move accord-
ingly. The effect of pressure on the graph lines for water accounts
for a familiar example, for at sufficiently low pressure its ‘liquid’
line does not intersect its ‘solid’ line before its ‘vapour’ line has
plunged down, and it too sublimes. This behaviour accounts for
the disappearance of hoar frost on a winter’s morning, when
actual ice is truly dry.
Our bodies live off Gibbs energy. Many of the processes that
constitute life are non-spontaneous reactions, which is why we
decompose and putrefy when we die and these life-sustaining
reactions no longer continue. A simple (in principle) example
is the construction of a protein molecule by stringing together in
an exactly controlled sequence numerous individual amino acid
molecules. The construction of a protein is not a spontaneous
process, as order must be created out of disorder. However, if the
reaction that builds a protein is linked to a strongly spontaneous
reaction, then the latter might be able to drive the former, just
as the combustion of a fuel in an engine can be used to drive
an electric generator to produce an orderly flow of electrons, an
electric current. A helpful analogy is that of a weight which can
be raised by coupling it to a heavier weight that raises the lighter
weight as it falls (Figure 17).
In biology a very important ‘heavy weight’ reaction involves
the molecule adenosine triphosphate (ATP). This molecule con-
sists of a knobbly group and tail of three alternating phosphorus
17. A process that corresponds to a large increase in total energy (repre-

sented here by an increase in disorder on the left) can drive a process in
which order emerges from disorder (on the right). This is analogous to a
falling heavy weight being able to raise a lighter weight.
and oxygen groups of atoms (hence the ‘tri’ and the ‘phosphate’
in its name). When a terminal phosphate group is snipped off by
reaction with water (Figure 18), to form adenosine diphosphate
(ADP), there is a substantial decrease in Gibbs energy, arising in
part from the increase in entropy when the group is liberated
from the chain. Enzymes in the body make use of this change
in Gibbs energy—this falling heavy weight—to bring about the
linking of amino acids, and gradually build a protein molecule.
It takes the effort of about three ATP molecules to link two
amino acids together, so the construction of a typical protein of
about 150 amino acid groups needs the energy released by about
450 ATP molecules.
The ADP molecules, the husks of dead ATP molecules, are
too valuable just to discard. They are converted back into ATP
molecules by coupling to reactions that release even more Gibbs
energy—act as even heavier weights—and which reattach a phos-
phate group to each one. These heavy-weight reactions are the
18. A molecular model of adenosine triphosphate (ATP). Some of the

phosphorus (P) and oxygen (O) atoms are marked. Energy is released
when the terminal phosphate group is severed at the location shown by
the line. The resulting ADP molecule must be ‘recharged’ with a new
phosphate group: that recharging is achieved by the reactions involved in
digestion and metabolism of food.
reactions of metabolism of the food that we need to ingest

regularly. That food may be the material that has been driven
into existence by even heavier reactions—reactions that release
even more Gibbs energy, and ultimately off the nuclear processes
that occur on the Sun.
Our final illustration of the utility of the Gibbs energy is one
of crucial importance in chemistry. It is a well-known feature of
chemical reactions that they all proceed to a condition known
as ‘equilibrium’ in which some reactants (the starting materials)
are present and the reaction has appeared to have come to a
stop before all the reactants have been converted into products.
In some cases the composition corresponding to equilibrium is
virtually pure products and the reaction is said to be ‘complete’.

Nevertheless, even in this case there are one or two molecules
of reactants among the myriad product molecules. The explosive
reaction of hydrogen and oxygen to form water is an example. On
the other hand, some reactions do not appear to go at all. Never-
theless, at equilibrium there are one or two product molecules
among the myriad reactant molecules. The dissolution of gold in
water is an example. A lot of reactions lie between these extremes,
with reactants and products both in abundance, and it is a matter
of great interest in chemistry to account for the composition
corresponding to equilibrium and how it responds to the con-
ditions, such as the temperature and the pressure. An important
point about chemical equilibrium is that when it is achieved the
reaction does not simply grind to a halt. At a molecular level all is
turmoil: reactants form products and products decompose into
reactants, but both processes occur at matching rates, so there
is no net change. Chemical equilibrium is dynamic equilibrium,
so it remains sensitive to the conditions: the reaction is not just
lying there dead.
The Gibbs energy is the key. Once again we note that at
constant temperature and pressure a system tends to change in
the direction corresponding to decreasing Gibbs energy. When
applying it to chemical reactions, we need to know that the Gibbs
energy of the reaction mixture depends on the composition of
the mixture. That dependence has two origins. One is the differ-
ence in Gibbs energies of the pure reactants and the pure prod-
ucts: as the composition changes from pure reactants to pure
products, so the Gibbs energy changes from one to the other.
The second contribution is from the mixing of the reactants and
products, which is a contribution to the entropy of the system

and therefore, through G = H − TS, to the Gibbs energy too.
This contribution is zero for pure reactants and for pure products
(where there is nothing to mix), and is a maximum when the
reactants and products are both abundant and the mixing is
extensive.
When both contributions are taken into account, it is
found that the Gibbs energy passes through a minimum at
an intermediate composition. This composition corresponds
to equilibrium. Any composition to the left or right of the
minimum has a higher Gibbs energy, and the system tends
spontaneously to migrate to lower Gibbs energy and attain the
composition corresponding to equilibrium. If the composition
is at equilibrium, then the reaction has no tendency to run
in either direction. In some cases (Figure 19), the minimum
Gibbs energy
Equilibrium
Equilibrium
Equilibrium
Progress
of reaction Pure products
Pure reactants
19. The variation of the Gibbs energy of a reaction mixture as it changes

from pure reactants to pure products. In each case, the equilibrium com-
position, which shows no further net tendency to change, occurs at the
minimum of the curve.
lies far to the left, very close to pure reactants, and the Gibbs
function reaches its minimum value after only a few molecules
of products are formed (as for gold dissolving in water). In
other cases, the minimum lies far to the right, and almost all the
reactants must be consumed before the minimum is reached (as
for the reaction between hydrogen and oxygen).
One everyday experience of a chemical reaction reaching equi-
librium is an exhausted electric battery. In a battery, a chem-
ical reaction drives electrons through an external circuit by
depositing electrons in one electrode and extracting them from
another electrode. This process is spontaneous in the thermo-
dynamic sense, and we can imagine it taking place as the
reactants sealed into the battery convert to products, and the
composition migrates from left to right in Figure 19. The Gibbs
energy of the system falls, and in due course reaches its minimum
value. The chemical reaction has reached equilibrium. It has no
further tendency to change into products, and therefore no fur-
ther tendency to drive electrons through the external circuit. The
reaction has reached the minimum of its Gibbs energy and the
battery—but not the reactions still continuing inside—is dead.
5. THE THIRD LAW
The unattainability of zero
I have introduced the temperature, the internal energy, and
the entropy. Essentially the whole of thermodynamics can be
expressed in terms of these three quantities. I have also intro-
duced the enthalpy, the Helmholtz energy, and the Gibbs energy;
but they are just convenient accounting quantities, not new fun-
damental concepts. The third law of thermodynamics is not
really in the same league as the first three, and some have argued
that it is not a law of thermodynamics at all. For one thing, it does
not inspire the introduction of a new thermodynamic function.
However, it does make possible their application.
Hints of the third law are already present in the consequences
of the second law, where we considered its implications for refrig-
eration. We saw that the coefficient of performance of a refriger-
ator depends on the temperature of the body we are seeking to
cool and that of the surroundings. We see from the expression
given in Chapter 3 1 that the coefficient of performance falls to
zero as the temperature of the cooled body approaches zero.
That is, we need to do an ever increasing, and ultimately infinite,
amount of work to remove energy from the body as heat as its
temperature approaches absolute zero.
There is another hint about the nature of the third law in
our discussion of the second. We have seen that there are two
approaches to the definition of entropy, the thermodynamic, as
expressed in Clausius’s definition, and the statistical, as expressed
by Boltzmann’s formula. They are not quite the same: the ther-
modynamic definition is for changes in entropy; the statisti-
cal definition is an absolute entropy. The latter tells us that a
1. To save you the trouble of looking back: c = 1/(Tsurroundings /Tcold − 1), and
Tsurroundings /Tcold → ∞ (and therefore c → 0) as Tcold → 0.
104 The Third Law: The unattainability of zero
fully ordered system, one without positional disorder and with-

out thermal disorder—in short, a system in its nondegenerate
ground state—has zero entropy regardless of the chemical com-
position of the substance, but the former leaves open the possibil-
ity that the entropy has a value other than zero at T = 0 and that
different substances have different entropies at that temperature.
The third law is the final link in the confirmation that
Boltzmann’s and Clausius’s definitions refer to the same prop-
erty and therefore justifies the interpretation of entropy changes
calculated by using thermodynamics as changes in disorder of
the system, with disorder understood to have the slightly sophis-
ticated interpretation discussed in Chapter 3. It also makes it
possible to use data obtained by thermal measurements, such as
heat capacities, to predict the composition of reacting systems
that correspond to equilibrium. The third law also has some
troublesome implications, especially for those seeking very low
temperatures.
As usual in classical thermodynamics, we focus on observations

made outside the system of interest, in its surroundings, and
close our minds, initially at least, to any knowledge or precon-
ceptions we might have about the molecular structure of the
system. That is, to establish a law of classical thermodynamics,
we proceed wholly phenomenologically.
Interesting things happen to matter when it is cooled to
very low temperatures. For instance, the original version of
The Third Law: The unattainability of zero 105
superconductivity, the ability of certain substances to conduct

electricity with zero resistance, was discovered when it became
possible to cool matter to the temperature of liquid helium
(about 4 K). Liquid helium itself displays the extraordinary
property of superfluidity, the ability to flow without viscosity
and to creep over the apparatus that contains it, when it is
cooled to about 1 K. The challenge, partly because it is there,
is to cool matter to absolute zero itself. Another challenge, to
which we shall return, is to explore whether it is possible—
and even meaningful—to cool matter to temperatures below
absolute zero of temperature; to break, as it were, the temperature
barrier.
Experiments to cool matter to absolute zero proved to be very
difficult, not merely because of the increasing amount of work
that has to be done to extract a given amount of heat from an
object as its temperature approaches zero. In due course, it was
conceded that it is impossible to attain absolute zero using a
conventional thermal technique; that is, a refrigerator based on
the heat engine design we discussed in Chapter 3. This empirical
observation is the content of the phenomenological version of
the third law of thermodynamics:
no finite sequence of cyclic processes can succeed in cooling a body

to absolute zero.
This is a negative statement; but we have seen that the first and
second laws can also be expressed as denials (no change in inter-
nal energy occurs in an isolated system, no heat engine operates
without a cold sink, and so on), so that is not a weakening of its
implications. Note that it refers to a cyclic process: there might be
other kinds of process that can cool an object to absolute zero,

but the apparatus that is used will not be found to be in the same
state as it was initially.
You will recall that in Chapter 1 we introduced the quantity
‚ as a more natural measure of temperature (with ‚ = 1/kT),
with absolute zero corresponding to infinite ‚. The third law as
we have stated it, transported to a world where people use ‚ to
express temperature, appears almost self-evident, for it becomes
‘no finite sequence of cyclic processes can succeed in cooling a
body to infinite ‚’, which is like saying that no finite ladder can
be used to reach infinity. There must be more to the third law
than appearances suggest.
We have remarked that thermodynamicists become excited when

nothing at all happens and that negations can have seriously
positive consequences, provided we think about the conse-
quences carefully. The pathway to a positive implication in
this case is entropy, and we need to consider how the third
law impinges on the thermodynamic definition of entropy.
To do so, we need to think about how low temperatures are
achieved.
Let’s suppose that the system consists of molecules that each
possess one electron. We need to know that a single electron has
the property of spin, which for our purposes we can think of
as an actual spinning motion. For reasons rooted in quantum
mechanics, an electron spins at a fixed rate and may do so either
clockwise or anticlockwise with respect to a given direction.

These two spin states are denoted ↑ and ↓. The spinning motion
of the electron gives rise to a magnetic field, and we may think of
each electron as behaving like a tiny bar magnet oriented in either
of two directions. In the presence of an applied magnetic field,
the two orientations of the bar magnets arising from the two spin
states have different energies, and the Boltzmann distribution
can be used to calculate the small difference in populations for
a given temperature. At room temperature there will be slightly
more lower energy ↓ spins than higher energy ↑ spins. If some-
how we could contrive to convert some of the ↑ into ↓ spins, then
the population difference will correspond to a lower tempera-
ture, and we shall have cooled the sample. If we could contrive
to make all the spins ↓, then we shall have reached absolute
zero.
We shall represent the sample at room temperature and in
the absence of a magnetic field by . . . ↓↓↑↓↑↑↓↓↓↑↓ . . . with
a random distribution of ↓ and ↑ spins. These spins are in
thermal contact with the rest of the material in the sam-
ple and share the same temperature. Now we increase the
magnetic field with the sample in thermal contact with its
surroundings. Because the sample can give up energy to its sur-
roundings, the electron spin populations can adjust. The sam-
ple becomes . . . ↑↓↓↑↓↓↓↑↑↓↑ . . . with a small preponderance
of ↓ spins over ↑ spins. The spin arrangement contributes to the
entropy, and so we can conclude that, because the spin distribu-
tion is less random than it was initially (because we can be more
confident about getting a ↓ in a blind selection), the entropy of
the sample has been reduced (Figure 20). That is, by turning up
M
Entropy
Entropy
Low D
field
High
field
0 Temperature 0 Temperature
20. The process of adiabatic demagnetization for reaching low tem-

peratures. The arrows depict the spin alignment of the electrons in
the sample. The first step (M) is isothermal magnetization, which
increases the alignment of the spins, the second step (D) is adiabatic
demagnetization, which preserves the entropy and therefore corre-
sponds to a lowering of temperature. If the two curves did not meet
at T = 0, it would be possible to lower the temperature to zero (as
shown on the left). That a finite sequence of cycles does not bring the
temperature to zero (as shown on the right) implies that the curves
meet at T = 0.
the magnetic field and allowing energy to escape as the electron

spins realign, we lower the entropy of the sample.
Now consider what happens when we isolate the sample ther-
mally from its surroundings and gradually reduce the applied
field to zero. A process that occurs without the transfer of energy
as heat is called adiabatic, as we saw in Chapter 1, so this step
is the ‘adiabatic demagnetization’ step that gives the process its
name. Because the process is adiabatic the entropy of the entire
sample (the spins and their immediate surroundings) remains
the same. The electron spins no longer have a magnetic field to

align against, so they resume their original higher entropy ran-
dom arrangement like . . . ↓↓↑↓↑↑↓↓↓↑↓ . . . . However, because
there is no change in the overall entropy of the sample, the
entropy of the molecules that carry the electrons must be low-
ered, which corresponds to a lowering of temperature. Isother-
mal magnetization followed by adiabatic demagnetization has
cooled the sample.
Next, we repeat the process. We magnetize the newly cooled
sample isothermally, isolate it thermally, and reduce the field
adiabatically. This cycle lowers the temperature of the sample a
little more. In principle, we can repeat this cyclic process, and
gradually cool the sample to any desired temperature.
At this point, however, the wolf inside the third law hurls off
its sheep’s clothing. If the entropy of the substance with and
without the magnetic field turned on were to be like that shown
in the left-hand half of Figure 20, then we could select a series of
cyclic changes that would bring the sample to T = 0 in a finite
series of steps. It has not proved possible to achieve absolute
zero in this way. The implication is that the entropy does not
behave as shown on the left, but must be like that shown on
the right of the illustration, with the two curves coinciding at
T = 0.
There are other processes that we might conceive of using to
reach absolute zero in a cyclic manner. For instance, we could
take a gas and compress it isothermally then allow it to expand
adiabatically to its initial volume. The adiabatic expansion of a
gas does work, and as no heat enters the system, the internal
energy falls. As we have seen, the internal energy of a gas arises
largely from the kinetic energy of its molecules, so adiabatic

expansion must result in their slowing down and therefore to
a lowering of the temperature. At first sight, we might expect
to repeat this cycle of isothermal compression and adiabatic
expansion, and hope to bring the temperature down to zero.
However, it turns out that the effect of adiabatic expansion on
the temperature diminishes as the temperature falls, so the pos-
sibility of using this technique is thwarted.
An even more elaborate technique would involve a chemical
reaction in which the process involved using a reactant A to
form a product B, finding an adiabatic path to recreate A, and
continuing this cycle. Once again, careful analysis shows that the
technique will fail to reach absolute zero because the entropies
of A and B converge on the same value as the temperature
approaches zero.
The common feature of this collective failure is traced to the
convergence of the entropies of substances to a common value
as T approaches zero. So, we can replace the phenomenological
statement of the third law with a slightly more sophisticated
version expressed in terms of the entropy:
the entropy of every pure, perfectly crystalline substance ap-

proaches the same value as the temperature approaches zero.
Note that the experimental evidence and the third law do not
tell us the absolute value of the entropy of a substance at
T = 0. All the law implies is that all substances have the same
entropy at T = 0 provided they have nondegenerate ground
states—no residual order arising from positional disorder of the
type characteristic of ice. However, it is expedient and sensible
to choose the common value for the entropy of all perfectly

crystalline substances as zero, and thus we arrive at the conven-
tional ‘entropy’ statement of the third law:
the entropy of all perfectly crystalline substances is zero at

T = 0.
The third law does not introduce a new thermodynamic

function, and is therefore not the same type of law as the other
three: it simply implies that the entropy can be expressed on an
absolute scale.
At first sight, the third law is important only to that very tiny sec-
tion of humanity struggling to beat the low-temperature record
(which, incidentally, currently stands at 0.000 000 000 1 K for
solids and at about 0.000 000 000 5 K for gases—when molecules
travel so slowly that it takes 30 s for them to travel an inch ). The
law would seem to be irrelevant to the everyday world, unlike the
other three laws of thermodynamics, which govern our daily lives
with such fearsome relevance.
There are indeed no pressing consequences of the third law
for the everyday world, but there are serious consequences for
those who inhabit laboratories. First, it eliminates one of science’s
most cherished idealizations, that of a perfect gas. A perfect gas—
a fluid that can be regarded as a chaotic swarm of independent
molecules in vigorous random motion—is taken to be the start-
ing point for many discussions and theoretical formulations in
thermodynamics, but the third law rules out its existence at

T = 0. The arguments are too technical to reproduce here, but
all stem from the vanishing of entropy at T = 0. 2 There are
technical salves to what might seem fatal injuries to the fabric
of thermodynamics, so the subject does survive this onslaught
from its own laws. Another technical consequence is that one
major application of thermodynamics to chemistry lies in the
use of thermal data, specifically heat capacities measured over
a range of temperatures, to calculate the equilibrium composi-
tion of reactions and thus to decide whether a reaction is likely
to be successful or not and to optimize the conditions for its
implementation in industry. The third law provides the key to
this application, which could not be done if the entropies of
substances were different at absolute zero.
Absolute zero is unattainable—in a sense. Too much should not

be read into the third law, because in the form that expresses
the unattainability of absolute zero it concerns processes that
maintain thermal equilibrium and are cyclic. It leaves open
the possibility that there are non-cyclic processes that can
reach absolute zero. The intriguing consequential question
that might occur is whether it is possible to contrive spe-
2. The third law implies that the thermal expansion coefficient—a measure
of how the volume of a substance responds to a change of temperature—must
vanish as T → 0, but the thermodynamic properties of a perfect gas imply that
the thermal expansion coefficient becomes infinite as T → 0!
cial techniques that take a sample to the other side of zero,

where the ‘absolute’ temperature is negative, whatever that
means.
To understand what it means for a body to have a temperature
below zero, below, paradoxically, its lowest possible value, we
need to recall from Chapter 1 that T is a parameter that occurs in
the Boltzmann distribution and which specifies the populations
of the available energy levels. It will be simplest, and in practice
most readily realizable, to consider a system that has only two
energy levels, a ground state and a second state above it in energy.
An actual example is an electron spin in a magnetic field, of
the kind already mentioned in this chapter. As we have already
remarked, because these two spin states correspond to opposite
orientations of the bar magnet, they have two different energies.
According to the Boltzmann distribution, at all finite tempera-
tures there will be more electrons in the state of lower energy
(the ↓ state) than of higher energy (the ↑ state). At T = 0,
all the electrons will be in the ground state (all will be ↓) and
the entropy will be zero. As the temperature is raised, electrons
migrate into the upper state, and the internal energy and the
entropy both increase. When the temperature becomes infinite,
the electrons are distributed equally over the two states, with half
the electrons ↓ and the other half ↑. The entropy has reached
its maximum value, a value which according to Boltzmann’s for-
mula is proportional to log 2.
Note in passing that an infinite temperature does not mean
that all the electrons are in the upper state: at infinite tempera-
ture, there are equal populations in the two states. This is a
general conclusion: if a system has many energy levels, then when

the temperature is infinite, all the states are equally populated.
Now suppose that T is negative, such as –300 K. When
T is given a negative value in the Boltzmann distribution we find
that the population of the upper state is predicted to be greater
than that in the lower state. For instance, if it happens that at
300 K the ratio of populations upper : lower is 1 : 5, then setting
T = −300 K gives a ratio of 5 : 1, with five times as many electron
spins in the upper energy state than in the lower state. Setting
T = −200 K gives a ratio of 11 : 1, and with T = −100 K the ratio
is 125 : 1. At −10 K the population of the upper state is nearly
1 000 000 000 000 000 000 000 times greater. Notice how, as the
temperature approaches zero from below (−300 K, −200 K,
−100 K, . . . ), the population migrates almost exclusively into the
upper state. In fact, just below 0, the population is entirely in the
upper state. Immediately above zero the population is entirely in
the lower state. We have seen that as the temperature is raised
from zero to infinity, the population migrates from the lower
state and the two states become equally populated. As the tem-
perature is lowered from zero to minus infinity the population
migrates from the upper state into the ground state, and at minus
infinity itself the populations are again equal.
We saw in Chapter 1 that the inverse temperature, specifically
‚ = 1/kT, is a more natural measure of temperature than
T itself. That it is to humanity’s regret that ‚ has not been
adopted becomes very clear when instead of plotting the energy
against T as shown in Figure 21, we plot it against ‚, for in
Figure 22 we get a nice smooth curve instead of the unpleasant
jump in the first graph at T = 0. You should also note that
21. The variation of (left) the internal energy and (right) the
entropy for a two-level system. The expressions for these
two properties can be calculated for negative temperatures,
as shown on the left of the illustration. Just above T =
0 all the molecules are in the ground state; just below
T = 0 they are all in the upper state. As the temperature becomes
infinite in either direction, the populations become equal.
there is a long expanse of ‚ at high ‚, corresponding to very low

temperatures, and it should not be surprising that there is plenty
of room for a lot of interesting physics as T approaches zero. We
are stuck, however, with the inconvenience of T in place of the
smooth convenience of ‚.
If we could contrive a system in which there are more ↑ (high
energy) electrons than ↓ (low energy) electrons, then from the
Boltzmann distribution we would ascribe it a negative tempera-
ture. Thus, if we could contrive a system in which there are five
22. The same system as in Figure 21 but plotted against ‚ instead of

T . The internal energy varies smoothly across the range.
times as many ↑ electrons as ↓ electrons, then for the same

energy separation as we assumed in the preceding discussion,
we would report the temperature as −300 K. If we managed to
contrive a ratio of 11 : 1, then the temperature would be reported
as −200 K, and so on. Note that it is easier to contrive extremely
low temperatures (those approaching minus infinity) because
they correspond to very tiny imbalances of populations whereas
large imbalances correspond to temperatures just below zero.
If the temperature is −1 000 000 K, the population ratio is only
1.0005 : 1, a difference of only 0.05 per cent.
The entropy tracks these changes in the distribution of popu-
lations. Thus, whereas S increases from zero to log 2 (in suitable
units) as T rises from zero to infinity, so too does it increase
from zero to log 2 at infinitely negative temperature. On either

side of zero we know precisely which state every electron is in
(as ↓ just above zero and as ↑ at just below zero), so S = 0.
At either extreme of infinity, the two states are equally popu-
lated, so a random selection gives equal chances of drawing ↑
and ↓. You should reflect on these figures in terms of ‚ instead
of T .
The big question is whether the inversion of a thermal equi-
librium (that is, Boltzmann) population can be contrived. It
can, but not by thermodynamic procedures. There are a vari-
ety of experimental techniques available for polarizing, as it is
called, a collection of electron or nuclear spins that use pulses
of radiofrequency energy. In fact, there is an everyday device
that makes use of negative temperatures: the laser. The essen-
tial principle of a laser is to produce large numbers of atoms
or molecules in an excited state and then to stimulate them
to discard their energy collectively. What we have referred to
as the ↓ and ↑ states of an electron can be regarded as the
analogues of the lower and upper energy states of the atom or
molecule in the laser material, and the inverted population on
which the laser effect depends corresponds to a negative absolute
temperature. All the laser-equipped devices we use around the
home, as in CD and DVD players, operate at temperatures below
zero.
∗
The concept of negative temperature really applies in practice

only to systems that possess two energy levels. To achieve a dis-
tribution of populations over three or more energy levels that
can be expressed as a Boltzmann distribution with a formally
negative value of the temperature is much more difficult and
highly artificial. Moreover, negative temperatures effectively take
us outside the domain of classical thermodynamics because
they have to be contrived and in general do not persist for
more than very short periods. Nevertheless, it is possible—and
interesting—to reflect on the thermodynamic properties of sys-
tems that have formally negative temperatures.
The first law is robust, and independent of how populations
are distributed over the available states. Therefore, in a region
of negative temperature, energy is conserved and the internal
energy may be changed by doing work or making use of a tem-
perature difference.
The second law survives because the definition of entropy
survives, but its implications are different. Thus, suppose energy
leaves a system as heat at a negative temperature, then according
to Clausius’s expression the entropy of the system increases: the
energy change is negative (say −100 J) and so is the temperature
(say −200 K), so their ratio is positive (in this case (−100 J)/
(−200 K) = +0.5 J K−1 ). We can understand that conclusion at
a molecular level by thinking about a two-level system: think
of the inverted population, which has a high energy but low
entropy, losing some of its energy and the population returning
towards equality, a high entropy (log 2) condition, so the entropy
increases as energy is lost. Similarly, if energy as heat enters

a system of negative temperature, the entropy of the system
decreases (if 100 J enters a system as −200 K, the change in
entropy is (+100 J)/(−200 K) = −0.5 J K−1 , a decrease). In this
case, the upper state becomes more highly populated as energy
floods in, so the population moves towards a greater imbalance,
towards the entire population being in the upper state and the
entropy close to zero.
The second law accounts for the ‘cooling’ of a system with a
negative temperature. Suppose heat leaves the system: its entropy
increases (as we have just seen). If that energy enters the sur-
roundings at a positive temperature, their entropy also increases.
Therefore, there is an overall increase in entropy when heat is
transferred from a region of negative temperature to one of ‘nor-
mal’ positive temperature. Once the populations of the first sys-
tem have equalized, we can treat the system as having a very high
positive temperature—one close to infinite temperature. From
this point on, we have an ordinary very hot system in contact
with a cooler system, and the entropy continues to increase as
heat flows from the former to the latter. In short, the second law
implies that there will be a spontaneous transfer of heat from a
system of negative temperature in contact with one of positive
temperature and that the process will continue until the temper-
atures of the two systems are equal. The only difference between
this discussion and the conventional one is that, provided one
system has a negative temperature, the heat flows from the system
with the lower (negative) temperature to the one with the higher
(positive) temperature.
If both systems have a negative temperature, heat flows from

the system with the higher (less negative) temperature to the sys-
tem with the lower (more negative) temperature. To understand
that conclusion, suppose a system at −100 K loses 100 J as heat:
the entropy increases by (−100 J)/(−100 K) = 1 J K−1 . If that
same heat is deposited in a system at −200 K, the entropy changes
by (+100 J)/(−200 K) = −0.5 J K−1 , a decrease. Therefore, overall
the total entropy of the two systems increases by 0.5 J K−1 and the
flow of heat from −100 K (the higher temperature) to −200 K is
spontaneous.
The efficiency of a heat engine, which is a direct consequence
of the second law, is still defined by the Carnot expression
(p. 52). 3 However, if the temperature of the cold reservoir is
negative, the efficiency of the engine may be greater than 1. For
instance, if the temperature of the hot source is 300 K and that
of the cold sink is −200 K, then the efficiency works out as 1.67:
we can expect to get more work from the engine than the heat
we extract from the hot source. The extra energy actually comes
from the cold sink, because, as we have seen, extracting heat from
a source with a negative temperature increases its entropy. 4 In a
sense, as the inverted population in the cold (negative) sink tum-
3. For your convenience again: ε = 1 − Tsink /Tsource .

4. Suppose we extract heat q from a source at a temperature 300 K, the
entropy decreases by q /(300 K). If we also withdraw heat q from the sink
at −200 K, its entropy increases by q /(200 K). The total change is positive
provided that q /(200 K) is at least equal to q /(300 K), or q = (200 K/300 K)q .
Both contributions can be converted into work without changing the entropy,
so the work we can get is equal to q + q . The efficiency is (work done)/(heat
absorbed from the hot source), or (q + q )/q = 1 + (200 K/300 K) = 1.67.
bles back down towards equality, the energy released contributes

to the work that the engine produces.
If both the source and the sink of a heat engine are at neg-
ative temperatures, the efficiency is less than 1, and the work
done is the conversion of the energy withdrawn as heat from the
‘warmer’, less negative, sink.
The third law requires a slight amendment on account of the
discontinuity of the thermal properties of a system across T = 0.
First, on the ‘normal’ side of zero, we simply have to change the
law to read ‘it is impossible in a finite number of cycles to cool
any system down to zero.’ On the other side of zero, the law takes
the form that ‘it is impossible in a finite number of cycles to heat
any system up to zero.’ Not, I suspect, that anyone would wish to
try!
C O N CLU SI ON
We are at the end of our journey. We have seen that thermo-

dynamics, the study of the transformations of energy, is a subject
of great breadth and underlies and elucidates many of the most
common concepts of the everyday world, such as temperature,
heat, and energy. We have seen that it emerged from reflections
on measurements of the properties of bulk samples, but that
the molecular interpretation of its concepts enriches our under-
standing of them.
The first three laws each introduce a property on which the
edifice of thermodynamics is based. The zeroth law introduced
the concept of temperature, the first law introduced internal
energy, and the second law introduced entropy. The first law
circumscribed the feasible changes in the universe: those that
conserve energy. The second law identified from among those
feasible changes the ones that are spontaneous—which have a
tendency to occur without us having to do work to drive them.
The third law brought the molecular and empirical formulations
of thermodynamics into coincidence, uniting the two rivers.
Where I have feared to tread is in two domains that spring
from or draw analogies with thermodynamics. I have not
touched on the still insecure world of non-equilibrium thermo-
dynamics, where attempts are made to derive laws relating to
124 Conclusion
the rate at which a process produces entropy as it takes place.

Nor have I touched on the extraordinary, and understandable,
analogies in the field of information theory, where the content
of a message is closely related to the statistical thermodynamic
definition of entropy. I have not mentioned other features that
some regard as central to a deep understanding of thermo-
dynamics, such as the fact that its laws, especially the second law,
are statistical in nature and therefore admit to brief failures as
molecules fluctuate into surprising arrangements.
What I have sought to cover are the core concepts, concepts
that effectively sprang from the steam engine but reach out to
embrace the unfolding of a thought. This little mighty handful
of laws truly drive the universe, touching and illuminating every-
thing we know.
F U RT H E R R E A D I N G
If you would like to take any of these matters further, then here are some
suggestions. I wrote about the conservation of energy and the concept
of entropy in my Galileo’s Finger: The Ten Great Ideas of Science (Oxford
University Press, 2003), at about this level but slightly less quantitatively.
In The Second Law (W. H. Freeman & Co., 1997) I attempted to demon-
strate that law’s concepts and implications largely pictorially, inventing
a tiny universe where we could see every atom. More serious accounts
will be found in my various textbooks. In order of complexity, these
are Chemical Principles: The Quest for Insight (with Loretta Jones, W.
H. Freeman & Co., 2008), Elements of Physical Chemistry (with Julio de
Paula, Oxford University Press and W. H. Freeman & Co., 2006), and
Physical Chemistry (with Julio de Paula, Oxford University Press and
W. H. Freeman & Co., 2006).
Others, of course, have written wonderfully about the laws. I can
direct you to that most authoritative account, Thermodynamics, by
G. N. Lewis and M. Randall (McGraw-Hill, 1923; revised by K. S.
Pitzer and L. Brewer, 1961). Other useful and reasonably accessible texts
on my shelves are The Theory of Thermodynamics, by J. R. Waldram
(Cambridge University Press, 1985), Applications of Thermodynamics,
by B. D. Wood (Addison-Wesley, 1982), Entropy Analysis, by N. C. Craig
(VCH, 1992), Entropy in Relation to Incomplete Knowledge, by K. G.
Denbigh and J. S. Denbigh (Cambridge University Press, 1985), and
Statistical Mechanics: A Concise Introduction for Chemists, by B. Widom
(Cambridge University Press, 2002).
INDEX
absolute entropy 68, 103 Carnot efficiency 52, 64

absolute zero 105 at negative temperatures 120
unattainability of 109 Celsius, A. 9
abstraction, power of 50 Celsius scale 9, 59
acceleration of free fall 24 change in entropy 60
adenosine diphosphate 95 chemical equilibrium 96
adenosine triphosphate 94 civilization, rise of 34
adiabatic 9 classical thermodynamics 11
adiabatic demagnetization 108 Clausius, R. 54
ADP 95 Clausius definition 60
air conditioner 76 Clausius statement 55, 63
amino acid 94 closed system 4
ATP 94 coefficient of performance
average speed 17 heat pump 76
refrigerator 75, 103
battery 99 combustion 39
‚ (beta) 13, 16, 106, 114 condensation 93
natural parameter 15 conservation, symmetry and 45
bioenergetics 91 constant pressure spontaneous
boiling 93 process 90
Boltzmann, L. 12, 68 constant volume spontaneous
Boltzmann distribution 13, 113 process 90
Boltzmann’s constant 14, 16, 68 cooling to low temperatures 108
Boltzmann’s formula 68
boundary 4 dead battery 99
degeneracy 70
carbon monoxide solid 70 degenerate solid 70, 71
Carnot, S. 51 demagnetization 108
128 Index
diathermic 8 first law 29, 123

disorder 66 at negative temperatures 118
state occupancy 67 flat battery 99
Donne, J. 78 fluctuation–dissipation
dry ice 93 theorem 42
food 77
efficiency 51, 73 free energy 81, 86, 91
Carnot 52, 64, 120 freezing 93
heat engine 73 fusion 40
electric battery 99
electron spin 107 gasoline combustion 88
energy 23, 25 Gibbs, J. W. 89
quality of 50 Gibbs energy 89
quantity of 50 chemical reaction 97
energy level 67 mixing 97
separation 43 phase transition 91
enthalpy 38, 81 temperature dependence 92
enthalpy of fusion 40 ground state 68
enthalpy of vaporization 40 heat 28
entropy 49, 60 molecular nature 32
absolute 68, 103 heat capacity 41
and disorder 66 temperature dependence 42
ice 71 heat engine 51
at negative temperatures efficiency 73
116 heat pump 76
at T = 0 70, 111 heat tax 81
residual 70 heater 24
equilibrium 96 heater as worker 34
chemical 96 Helmholtz, H. von 84
mechanical 6 Helmholtz energy 84
thermal 7 molecular interpretation 86
expansion coefficient 112 hoar frost 94
extensive property 4 hydrogen bond 71
Fahrenheit, D. 9 ice
Fahrenheit scale 9 residual entropy 71
feasible change 65 structure 72
Index 129
ideal gas 17 natural process 54

imponderable fluid 51 negative temperature 113
infinite temperature 113 Noether, E. 45
information theory 124 Noether’s theorem 45
intensive property 4 non-cyclic process 112
internal combustion engine 77 non-equilibrium
internal energy 26 thermodynamics 123
temperature dependence 41
as total energy 44 open system 4
isolated system 4
particle in a box 67
perfect gas 17
Joule, J. P. 26
nonexistence of 111
joule 14, 24
perfect gas temperature 60
perpetual motion 29
Kelvin, Lord 53
phase 91
kelvin 11
phase transition 91
Kelvin scale 10
piston 4, 50
Kelvin statement 53, 63
Planck, M. 69
polarized spins 117
laser 117
power 62
latent heat 40
pressure, molecular
law of conservation of energy 23
interpretation 12, 17
liquid helium 105
property 4
low temperature record 111
protein 94
low temperatures 104
quality of energy 50, 61
maximum work 36 quantity of energy 50
Maxwell–Boltzmann
distribution 17 Rankine scale 11
mechanical equilibrium 6 refrigerator 55, 73
melting 93 relative populations 15
mixing 97 residual entropy 70
molecular interpretation reversible process 35
heat 32
Helmholtz energy 86 scalding 40
pressure 12, 17 second law 49, 123
work 32 Clausius statement 55
130 Index
second law (cont.) tax refund 89

entropy statement 62 temperature 7
Kelvin statement 53 temperature scale
at negative temperatures in terms of work 58
118 thermal equilibrium 7
sink 51 thermal expansion coefficient 112
sneeze analogy 61 thermodynamic scale 10
Snow, C. P. 49 thermometer 9
source 50 third law 103, 105, 123
spin 107 entropy statement 111
spontaneous change 65 Thomson, W. 53
spontaneous process 54, 82 triple point 59
constant pressure 90
constant volume 90 universe 3, 62, 82
state function 27
vaporization 40, 94
statistical thermodynamics 11
steam engine 50 watt 62
abstraction 78 weight analogy 94
sublimation 93 weight of an arrangement 69
superconductivity 105 work 23
superfluidity 105 capacity to do 25
superheated steam 52 maximum as reversible 36
surroundings 3 molecular nature 32
symmetry and conservation
laws 45 zeroth law 7, 123
system 3 as afterthought 3
SYMBOL AND UNIT INDEX
A Helmholtz energy 84 k Boltzmann’s constant 13

K kelvin 11
‚ (beta) = 1/kT 13 kg kilogram 14
m mass 24
c coefficient of performance 75
m metre 14
C heat capacity 41
p pressure 38
D degeneracy 70
Delta X, X = X final − X initial 39 q energy transferred as heat 64
E energy 13 S entropy 49, 60

ε (epsilon) efficiency 58 s second 14
g acceleration of free fall 24 T temperature 10

G Gibbs energy 89
U internal energy 26
H enthalpy 38 W watt 62
W weight of arrangement 69
J joule 14 w energy transferred as work 75

Peter Atkins - Four Laws That Drive The Universe (2007)

Uploaded by

Copyright:

Available Formats

Peter Atkins - Four Laws That Drive The Universe (2007)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Peter Atkins - Four Laws That Drive The Universe (2007)

Uploaded by

Copyright:

Available Formats

What are the four laws of thermodynamics?

What are the four laws of thermodynamics?

What is entropy and how is it related to the second law of thermodynamics?

What is entropy and how is it related to the second law of thermodynamics?

     

This page intentionally left blank

Among the hundreds of laws that describe the universe, there

The mighty handful consists of four laws, with the numbering

In conclusion, and before you roll up the sleeves of your mind

1. THE ZEROTH LAW: The concept of temperature 1

2. THE FIRST LAW: The conservation of energy 21

3. THE SECOND LAW: The increase in entropy 47

4. FREE ENERGY: The availability of work 79

5. THE THIRD LAW: The unattainability of zero 101

Further reading 125

A system is deﬁned by its boundary. If matter can be added to

thermodynamics and it is helpful to keep the distinction in

So much for these slightly dusty deﬁnitions. Now we shall use

change. For instance, their pressures might change or we could

if A is in thermal equilibrium with B, and B is in thermal equilib-

The zeroth law implies that just as the pressure is a physical

–300 –200 –100 0 100

100 200 300 400

3. Three common temperature scales showing the relations between them.

scale, which uses degrees (‘kelvins’, K) of the same size as the

In each of the ﬁrst three chapters I shall introduce a property

from the fact that in the discussion of bulk properties we don’t

populations, and although atoms continue to jump between

Population of state of energy E

We see that for states of progressively higher energy, the popu-

1. Energy is reported in joules (J): 1 J = 1 kg m2 s−2 . We could think of 1 J

4. The Boltzmann distribution is an exponentially decaying function of

are occupied. High temperatures (low ‚) corresponds to our

as its measure of temperature, Boltzmann’s constant would not

average speeds and low temperatures to lower average speeds

5. The Maxwell–Boltzmann distribution of molecular speeds for mole-

A word or two of summary might be appropriate at this

there will be a consequent change of state—their temperatures

when you stretch or compress a spring, and the amount of work

once we have introduced the concept of heat. That concept hasn’t

Now we extend these concepts from dynamics to thermodynam-

experiment was performed by J. P. Joule (1818–1889), one of

6. The observation that diﬀerent ways of doing work on a system and

diﬀerence of the initial and ﬁnal values of the internal energy,

7. When a system is adiabatic (left), a given change of state is brought

release of energy as heat. Its internal energy falls. We then leave

the internal energy of an isolated system is constant.

That is the ﬁrst law of thermodynamics, or at least one statement

perpetual motion has never been achieved. There have been

We have a variety of cleaning-up exercises to do before we leave

coincidence, it is not an indicator that heat is actually a ﬂuid any

There has probably been detected a slipperiness in the preceding

inside there is no distinction between the type of funds in which

8. The molecular distinction between the transfer of energy as work (left)

is increased. In more pictorial terms, a block of iron at high

undergoes collisions with the other molecules in the system, and

The originators of thermodynamics were subtle people, and

the block, and an inﬁnitesimal change in the temperature of the

Thermodynamicists are also subtle in their discussion of the