Anda di halaman 1dari 55

Construction

Materials
Their nature and
behaviour
Fourth edition
Edited by
Peter Domone and
John I IIston
9 ~ ~ ~ r ~ l ~ ~ ; o ~ ~ francis
LONDON AND NEW YORK
hrst published as Concrete Timber and Metals 1979
hy Cha pma n a nd Ha ll
Second editi o n publ ished 1994
by Cha pman and Ha ll
Third ed iti o n publi shed 200 I
by Spo n Press
Th is edition publi shed 2010
by Spon Press
2 Pa rk Squa re, Milton Pa rk, Abingdon, Oxon OX 14 4RN
Simulta neously publi shed in the USA a nd Canada
by Spon Press
270 Madi son Avenue, New Yo rk, NY 100 16, USA
Span Press is an imprint of the Taylor 6 Francis Croup, an informa business
20 10 Spon Press
T ypeser in Sa bon by
Cra phi craft Li mited, Hong Kong
Printed and bo und in Crea r Brita in by
MPC Boo ks C ro up, UK
All ri gh ts rese rved. No pa rr of thi s book ma y be reprinted o r reproduced o r
utili sed in a ny form or by any electro ni c, mecha ni ca l, o r other means, now
known or herea fter in vented, includi ng photocopying and recording, o r in
a ny in fo rma ti o n storage or ret ri eval system, without per missio n in writi ng
from the publishers .
T hi s publ ica ti on presents materi a l of a broa d scope a nd applicahi lit)'.
Des pi te stringe nt efforrs by al l conce rned in the publi shing process , some
t ypographi ca l o r editori a l erro rs may occur, a nd rea ders are encouraged to
br ing these ro o ur a ttenti on whe re they reprcsenr erro rs of substa nce. The
publisher a nd a uthor di scl a im a ny lia bility, in whole o r in pa rr, a ri sing fro m
informat ion conta ined in thi s publi ca tion. The reader is urged to co nsult
wit h an a ppro priate li censed professional pri or to raking any action or
mak ing a ny inte rpretat io n th at is within the rea lm of a li censed
professio nal practi ce.
Bri tish Library CatalrJguing in Publication Data
A ca ta logue record fo r thi s book is a vail able fro m t he Br iti sh Libra ry
Library of Congress Cataloging-in- P11hlication Data
Constructi o n ma teri a ls : their na ture and behavio ur / ledircd hyl
Peter Domone a nd J. M. !I Iston. - 4th ed.
p. em.
Includes bibli ographica l references.
I . Building ma terial s. I. Domone, P. L. J. II. lll ston, J. M.
TA403.C636 2010
624 .1 '8-dc22 2009042708
ISBN 10: 0-41 5-465 15-X (hbk )
ISBN I O: 0-4 15-465 16-8 (pbk )
ISBN1 0: 0-203- 92757-5 (ebk )
ISBNl3: 978-0-41 5-465 15- 1 (hbk )
ISBN I 3: 978-0-41 5-465 16-8 (pbk )
ISBN 13: 978-0- 203-92757- 1 (ebk )
NAZARBAYEV
UNIVERSITY
LIBRARY
Introduction
We conventi onall y think of a materi al as being either
a solid or a fluid. These states of matter are con-
venientl y based on the response of the mater ial to
an appli ed force. A solid will maintain its shape
under its own weight, and resist appli ed forces with
little deformation.' An unconfined fluid will flow under
its own weight or applied force. Fluids can be divided
into liquids and gases; liquids are essenti all y incom-
press ibl e and maintain a fi xed volume when pl aced
in a container, whereas gases are greatl y compressibl e
and will also expa nd to fill the volume ava il abl e.
Although these di visions of materials are often con-
veni ent, we must recogni se that they are not di stinct,
and some materials di splay mi xed behaviour, such
as gels, which can va ry from near solids to near
liquids.
In construct ion we are for the most part con-
ce rned with solids, since we use these to ca rry rhe
appli ed or self-weight loads, but we do need to
understand some aspects of fluid behaviour, for
exa mpl e when dea ling wit h fresh concrete or the
flow of water or gas into and through a mater ial.
In termed ia te viscoelasti c behaviour IS a I so
Important.
PART 1
FUNDAMENTALS
Revised and updated by Peter Domone,
with acknowledgements to the previous
authors, Bill Biggs, ian McColl and
Bob Moon
This first part of the book is aimed at both describ-
ing and explaining the behaviour of materi als in gen-
eral, without specifi call y concentrating on any one type
or group of materials. That is the purpose of the
later secti ons. This part therefore provides the bas is
for the later parts, and if you get to grips with the
principl es then much of what follows will be cl ea rer.
In the first chapter we start with a descript ion of
the building blocks of all material s - aroms - and
how they combine in single elements and in com-
pounds to form gases, liquids and solids. We then
introduce some of the pri nci pies of thermodynami cs
and the processes involved in changes of state, with
an emphasis on the change from liquid to solid. In
the next two chapters we desc ribe the behaviour of
solids when subj ected to load and then consider the
structure of the various t ypes of solids used in con-
structi on, thereby giving an expl anation for and an
understanding of their behaviour.
This is foll owed in subsequent chapters by con-
sidera ti on of rhe process of fracture in more dera il
(including an int roduction to the subj ect of fracture
mechani cs), and then by bri ef di sc ussions of the
behaviour of liquids, viscoelastic material s and gels,
the nature and behaviour of surfaces and the electrical
and ther mal properties of materi als.
1
Bur note rhar the deformation may sri!! be signifi ca nt on an engineering sca le, as we shall see extensively in rhis book.
As engineers we are primaril y concerned with the prop-
erti es of materi a ls at the mac rostructural leve l, but
in order to understa nd these properti es (which we will
introduce in Cha pter 2 ) a nd to modify them to o ur
advantage, we need a n understa nding of the structure
of materi als at the atomi c level through bonding forces,
mo lecul es a nd mo lecul a r a rra ngement. Some kno wl -
edge of the processes involved in c ha nges o f sta te,
pa rti cul a r ly fro m liquids to solids, is a lso va lua bl e.
The concept o f 'ato mi sti cs' is not new. The ancient
Greeks - a nd es pecia ll y Democritus (ca. 4 60 Bc) -
had the idea o f a single elementa ry particl e but their
science did not extend to observati on and experiment.
Fo r that we had to wa it nea rl y 22 centuri es unril
Da lton, Avogadro a nd Ca nni zzaro fo rmul ated atomic
theory as we know it today. Even so, ver y ma ny
mysteri es still rema in unresolved. So in rrea ting the
subj ect in thi s way we a re reaching a lo ng way back
into the develo pment o f tho ught a bo ut the uni verse
a nd the way in whi ch it is put together. Thi s is
covered in the fi rs t pa rt of thi s c ha pte r.
Concepts o f cha nges o f sta te a re mo re recent.
Engineering is much concerned with cha nge - the
cha nge fr o m the unl oaded to the loaded sta te, rhe
consequences o f cha nging tempera ture, environment,
etc. The first scientifi c studi es o f thi s can be a ttributed
to Ca rnot ( 1824 ), la ter extended by such g ia nts as
Cla usius, J o ul e a nd others to pro duce ideas s uch as
the conserva ti o n o f energy, mo mentum, etc. Since
the ea rl y studi es we re ca rri ed o ut o n hea t engines
it became known as t he science of thermodyna mi cs,
1
1
In ma ny enginee ring courses ther modyna mi cs is trea rcd
as a se para te topic, o r nor conside red a t a ll. Bur, beca use
irs a ppli cati ons set rul es rha r no engineer ca n ignore, a
bri ef discuss ion is included in thi s chapter. Wha t are these
rul es? Succinctl y, they a re:
Yo u cannot win, i.e. you cannot get mo re o ur of a
system th a n you pur in .
You ca nn ot brea k even- in a ny cha nge something will
be lost or, to be mo re precise, it will be useless fo r the
purpose you have in mind.
Chapter 1
Atoms, bonding,
energy and
equilibrium
but if we ta ke a broader view it is rea ll y the a rt
a nd science o f managing, contro lling a nd using the
t ra nsfer o f energy- whether the energy o f rhe atom,
the energy of the rides o r the energy o f, say, a lifting
ri g. The second pa rt o f thi s cha pter therefore dea ls
with the conce pts o f energy as a ppli ed to changes
of sta te, fro m gases to li quid, bri efl y, a nd from liquid
ro solid, mo re extensive ly, including considera ti o n
of equilibrium a nd equilibrium di agra ms. If these
a t first seem da unting, yo u may skip past these
secti o ns on first read ing, but come back ro them,
as they a re impo rta nt.
1.1 Atomic structure
Atoms, the bui lding bl oc k o f elements, consist of a
nucleus sur ro unded by a cl o ud o f o rbiting electro ns.
T he nucl eus consists o f positi ve ly cha rged protons
a nd neutral ne utro ns, a nd so has a net pos iti ve
cha rge tha t ho lds the nega ti vely cha rged electro ns,
whi ch revolve a round it, in positi on by a n electro-
sta ti c a ttracti o n.
2
The cha rges on the proto n a nd
electron a re eq ual a nd Oppos it e (1. 602 X ] o - JY COU-
lombs) a nd the number of elect rons a nd prot o ns
are equa l a nd so the a to m overa ll is electri ca ll y
neutra l.
Prot ons a nd neutrons have a pproxima tely the
sa me mass, 1.67 x 1 o-n kg, whereas a n electron
has a mass o f 9. 11 X 1 o - l l kg, near ly 2000 times
less. These relati ve densit ies mea n that the size of
the nucl eus is very sma ll compa red to the size of
the atom. Altho ugh the na ture o f the elect ron clo ud
ma kes it di ffi cult to defin e the size of atoms precisely,
heliu m ha s the sma ll est atom, with a radius of a bo ut
2
Pa rti cle ph ys icists have di scove red or postul a ted a con-
siderable number of other sub-atomi c pa rti cles, such as
qua rks, muons, pi ons and neutrinos. It is however sufficient
fo r o ur pur poses in thi s book fo r us to consider o nl y
electrons, protons a nd neutrons.
3
Fundamentals
Table 1.1 Avai labl e electron st ates in the fir st fo ur shell s and sub-shell s of electrons in the Bohr a tlll ll
(a fter Ca lli ster, 2007)
Principal quantum
number (11 ) Shell Sub-shell ( /)
j
K s
2 L s
p
3 M s
p
d
4 N
p
d
f
0.03 nanometers, whil e caesium has one of the largest,
with a radius of about 0.3 nanometres.
An element is characteri sed by:
the atomic number, whi ch is the number of pro-
tons in the nucleus, and hence is also the number
of electrons in orbit;
the mass number, wh ich is sum of the number of
protons and neutrons. For many of the li ghter ele-
ments these numbers are simil ar and so the mass
number is approximately twi ce the atomi c number,
though this relationship breaks down with increas-
ing atomic number. In some elements the number
of neutrons can vary, leading to isotopes; the atomic
weight is the weighted average of the atomi c masses
of an element 's naturally occurring isotopes.
Another usefu l quantity when we come to con-
sider compounds and chemi cal reactions is the mole,
which is the amount of a substance that contains
6.023 x 10
23
atoms of an element or mol ecul es of
a compound (Avogadro's number). Thi s number has
been chosen beca use it is the number of atoms that
is contained in the atomi c mass (or weight ) expressed
in grams. For example, carbon has an atomi c we ight
of 12.011 , and so 12.011 grams of ca rbon contain
6.023 x 10
23
atoms .
The manner in wh ich the orbits of the electrons
are di stributed around the nucl eus con tro ls th e
characteristics of the element and the way in which
atoms bond with other atoms of the sa me element
and with atoms from different elements.
For our purposes it will be suffi cient to describe
the structure of the so-call ed Bohr atom, which
arose from developments in quantum mechanics in
4
Maximum number of elecrrons
Number of
energy states (nr
1
) Per sub-shell Per shell
3
3
5
I
3
5
7
2 2
2 8
6
2 18
6
10
2 32
6
10
14
the earl y part of the 20th century. This overcame
the probl em of ex plaining why nega tively charged
electrons wou ld not coll apse into the posit ively
charged nucleus by proposing that electrons revolve
around the nucl eus in one of a number of di sc rete
orbita ls or shells, each with a defin ed or qua nti sed
energy level. Any electron moving between energy
levels or orbitals would make a quantum jump with
either emi ssion or absorpti on of a di sc rete amount
or quantum of energy.
Each e lecrron is characteri sed by four qua nrum
numbers:
the principal quantum number (n = 1, 2, 3, 4 . .. ),
whi ch i the quantum shell ro whi ch rhe electron
belongs, also denoted by K, L, M, N ... , cor-
responding to n = 1, 2, 3, 4 ... ;
the secondar y quantum number (I = 0, I , 2 ...
n- 1 ), which is the sub-shell to which the electron
belongs, denoted by s, p, d, f, g, h for I = 1, 2,
3, 4, 5, 6, according to its shape;
the third quantum number (m
1
), whi ch is the
number of energy states within each sub-shell ,
the total number of which is 2/ + I ;
the fourth quantum number (m,) which desc ribes
the electron 's direct ion of spin and is either +
1
/1
or -
1
1!.
The number of sub-s hell s that occur within each shell
therefore increases wit h an increase in the principal
quantum number (n), and the number of energy states
within each subshell (m
1
) increases with an increase
in the secondary quantum number(/ ). Table 1.1 shows
how thi s leads to the ma ximum number electrons
in each shell for the first four shells.
<@. 8 \)
' '
' '
' '
' '
.(@. 8 .... )
' '
' '
' '
' '
', _ _Helium __ --
:::::
Fig. 1. 1 The atomic structure of the first three elements
of the periodic table and sodium.
Each electron has a unique set of quantum numbers
and with increasing atomi c number, a nd hence
inc reasing number of electrons, the shell s a nd sub-
shell s fill up progressively, starting with the lowest
energy state. The one electron of hydrogen is therefore
in the onl y sub-s hell in the K shell (denoted as 1 s
1
),
the two electrons of helium are both in thi s sa me
shell (denoted as 1 s
2
) and in lithium, which has
three electrons, two are in the ls
1
shell and the third
is in the 2s
1
shell. By convention, the confi gurati on
of li thium is written as l s
2
2s
1
The confi gura ti on
of subsequent elements foll ows logicall y (for ex-
a mpl e, sodium with 11 electrons is 1 s
2
2s
2
2p
6
3s
1
).
The structures of these elements a re illustrated in
Fig. 1.7 .
An extremely impo rta nt factor gove rning the
properti es o f an element is the number of electrons
in the outermost shell (known as the valence elec-
trons), since it is these that are most readily ava il a bl e
to form bonds with other atoms. Groups of elements
with simil a r properti es are obta ined with va rying
atomic number but with the sa me number o f outer
shell elect rons. For exa mpl e, the 'alkali meta ls'
lithium, sod ium, potass ium, rubidium and caesium
a ll have o ne electron in their outermost shell , and
a ll a re ca pabl e of forming strong a lkalis.
A further factor relating t o thi s is that when the
outermost electron shell is completely fill ed the elec-
tron configurati on is stabl e. This normall y corresponds
to the s a nd p st ates in the outermost shell being
Atoms, bonding, energy and equilibrium
fi lled by a t ot a l of eight electrons; such octets are
found in neon, argon, kr ypton, xenon etc., and these
' noble ga es' form very few chemi ca l compounds
for thi s reason. The excepti on to the octet ru le for
sta bility is helium; the outermos t (K) shell onl y has
roo m for its two electrons.
The li sting of the elements in o rder o f increasing
atomi c number a nd a rra nging them into groups o f
the sa me va lence is the basis of the periodic tabl e
of the elements, which is an extremely conveni ent
way of categori sing the elements and predicting
their li kely properties and behaviour. As we will see
in the next secti o n, the number of va lence electrons
st rongly influences the nature of the interatomic
bonds.
1.2 Bonding of atoms
1.2. 1 IONIC BONDING
If a n a t o m (A) with one electron in the outermost
shell reacts with a n atom (B) with seven electrons
in the o utermost shell, then both ca n atta in the
octet structure if a tom A donates irs va lence electron
to atom B. Howeve r, the electri ca l neutra lity of rhe
atoms is disturbed a nd B, with an extra electron,
becomes a negati vel y charged ion (a n ani on), whereas
A becomes a pos iti vely charged ion (a ca ti on). The
rwo io ns a re then a ttracted to each other by the
electrostati c force between them, and a n ioni c com-
po und is fo rmed.
The number o f bo nds rhat ca n be for med with
other atoms in this way is determined by the valency.
Sodium has one electron in its outer shell ; it is a bl e
t o give this up to form the cati on whereas chl o rine,
whi ch has seven electrons in its ourer shell , ca n
acce pt o ne t o form the anion, thus sodium chloride
has the c hemi ca l fo rmula NaCI (Fig. 1. 2). Oxygen,
however, has six va lence electrons a nd needs to
______ ...
Sodium
2-8-1
Fig. 1.2 Ionic bonding.
Chlorine
2-8-7
5
Fundamentals
(a) Between chlorine atoms
Fig. 1.3 Covalent bonding.
' borrow' or 'share' rwo; since sodium ca n onl y donate
one electron, the chemi ca l formula for sodium oxide
IS a
2
0. Magnesium has two va lence elect rons and
so the chemi ca l formula for magnesium chl oride is
MgCI
2
and for magnesium oxide MgO. Thus, the
number of va lence electrons determines the relati ve
proporti ons of elements in compounds.
The strengt h of the ioni c bond is proporti onal to
eAe
1
;1r where eA and e ~ are the charges on the ions
and r is the interatomi c separa ti on. The bond is
strong, as shown by the hi gh melting point of ioni c
compounds, and its strength increases, as mi ght be
expected, where two or more electrons are donated.
Thus the melting point of sodium chl oride, aCI,
is 80l C; that of magnesium oxide, MgO, where
two elect rons are involved, is 2640C; and that of
zirconium ca rbide, ZrC, where four electrons are
involved, is 3500C. Although ionic bonding involves
the transfer of electrons between different atoms, the
overa ll neutralit y of the materi al is maintained.
The ioni c bond is always non-directi onal; that is,
when a crys tal is built up of large numbers of ions,
the electrostati c charges are ar ranged symmetri ca II y
around each ion, with the result that A ions surround
themse lves with B ions and vice versa, with a solid
being formed. The pattern adopted depends on the
charges on, and the relative sizes of, the A and B
ions, i. e. how many B ions ca n be comfortably
accommodated around A ions whilst prese rving the
correct ratio of A ro B ions.
1.2.2 COVALENT BONDING
An obvious limitati on of the ioni c bond is that it ca n
onl y occur between atoms of different elements, and
therefore it ca nnot be res ponsibl e for the bonding
of any of the solid elements. Where both atoms are
of the electron-acceptor type, i.e. with cl ose to 8
outermost electrons, octet structures ca n be built
up by the sharing of two or more va lence electrons
between the atoms, forming a covalent bond.
For exa mple, two chl orine atoms, which each have
seven va lence electrons, ca n achi eve the octet struc-
6
(b) Between oxygen atoms
ture and hence bond together by contributing one
electron each to share with the other (Fig. 1.3a).
Oxygen has six va lence elect rons and needs to
share two of these with a neighbour to form a bond
(Fig. 1. 3b). In both cases a molecul e with two a roms
is formed (CI
2
and 0
2
), which is rhe normal state
of these two gaseous elements and a few others.
There are no bonds between the molecul es, whi ch is
why such elements are gases at norma l temperature
and pressure.
Covalent bonds are very strong and directional; they
ca n lead to ve ry strong two- and three-dimensional
structures in elements where bonds ca n be formed
by sharing electrons with more than one adj acent
atom, i. e. which have four, five or six va lence elec-
trons. Ca rbon and sili con, both of whi ch have four
va lence electrons, are two importa nt exa mpl es. A
struct ure ca n be built up with each atom forming
bonds with fo ur ad jacent atoms, thus ach1eving the
required electron octet. In practice, the atoms arrange
themselves with eq ual angles between all the bonds,
which produces a tetrahedral structure (Fig. 1.4 ).
Ca rbon atoms are arranged in thi s way in di amond,
which is one of the hardest materials kn own and
also has a very hi gh melting point (3500C).
Cova lent bonds are a I so formed between a roms
from different elements to give compounds. Methane
( H
4
) is a simpl e exa mpl e; eac h hydrogen atom
achi eves a stabl e helium electron confi gurat ion by
sharing one of the four atoms in car bon's outer shell
and the carbon atom achi eves a stabl e octet figuration
by sharing the electron in each of the four hydrogen
atoms (Fig. 1.5) . It is also possibl e for carbon atoms
to form long chains to which other atoms ca n bond
along the length , as shown in Fig. 1. 6. Thi s is the
basis of many polymers, whi ch occur extensively in
both natural and manufactured forms.
A large number of compounds have a mi xt ure
of cova lent and ioni c bonds, e.g. sulphates such as
Na
2
S0
4
in which the sulphur and oxygen are cova-
lentl y bonded and form sulphate ions, which form
an ioni c bond wi th the sodium ions. In both the
0
Outer shell of
a si ngle carbon
or sil icon atom
t
Atoms, bonding, energy and equilibrium
- --+
Fig. 1.4 Covalent bonding in carbon or silica to form a continuous stmcture with (our bonds orientated at equal
spacing giving a tetrahedron-based structure.
H
Fig. l. S Covalent bonding in methane, C l - 1 ~ .
t t t
-
Fig. 1.6 Covalent bonding in carbon chains.
ionic and cova lent bonds the elect rons are held fairl y
strongly and a re not free to move far, whi ch accounts
for the low electrical conducti vit y of mater ia ls co n-
ta ining such bonds.
1.2.3 METALLIC BONDS
Metallic atoms possess few va lence electrons and thus
ca nnot form covalent bonds between each other;
instead they obey what is termed the free-electron
theory. In a metalli c crysta l the valence electrons
are detached from their atoms and ca n move freely
bet ween the positi ve meta lli c ions (Fig. 7. 7). T he
positive ions a re a rranged regul ar ly in a crysta l
latti ce, a nd the electrost ati c at tract ion between the
positive ions and the free negative electrons provides
the cohes ive strengt h of the meta l. The meta lli c
bond may thus be regarded as a ve ry spec ial case
t t
7
Fundamentals
Fig. 1. 7 The free electron system in the metallic bond
in a monovalent metal.
of cova lent bond ing, in which the octet struct ure is
sa ti sfi ed by a generali sed donati on of the va lence
electrons to form a 'cloud' that permeates the whole
crystal latti ce, rather than by electron shar ing berween
specifi c atoms (true covalent bonding) or by donat ion
to another atom (ioni c bonding).
Since the electrostati c attracti on between ions and
electrons is non-direct ional, i.e. the bonding is not
locali sed berween indi vi dual pairs or groups of atoms,
metalli c crystals ca n grow easil y in three dimensions,
and the ions ca n approach all neighbours eq uall y
to give maxi mum structu ra I densit y. The resulting
structures are geometricall y simple by comparison
with the structures of ioni c compounds, and it is
this simplicit y that accounts in part for the ductility
(a bility to deform non-reversibl y) of the metallic
elements.
Metalli c bonding also expl ains the hi gh thermal
and electri cal conducti vity of metals. Since the va lence
electrons are not bound to any pa rt icular atom, they
ca n move through the latti ce under the appli cation
of an elect ri c potenti al, ca using a current fl ow, and
ca n also, by a se ri es of colli sions with neighbouring
electrons, tra nsmit thermal energy rapidl y through
the latt ice. Optical properties ca n also be expl ained.
f or example, if a ray of light fall s on a metal, the
electrons (being fr ee) ca n absorb the energy of the
light bea m, thus preventing it from passing through
the crystal and rendering the metal opaque. The
electrons that have absorbed the energy are excited
to hi gh energy levels and subsequentl y fal l back to
their ori ginal va lues with the emi ssion of the li ght
energy. In other words, the li ght is refl ected back
fr om the surface of the metal, whi ch when poli shed
is hi ghl y refl ecti ve.
The ability of metals to for m all oys (of extreme
importance to engineers) is also explained by the
free-electron theory. Since the electrons are not bound,
when two metals are all oyed there is no question
8
of elect ron exc hange or sharing between atoms in
ion ic or cova lent bond ing, and hence t he ordinary
va lence laws of combination do not apply. The prin-
cipal limitati on then becomes one of atomic size,
and providing there is no grea t size difference, two
metals may be abl e to form a continuous se ries of
all oys or solid soluti ons from I 00% A to I 00% B.
The rules governing the composition of these soluti ons
are di scussed later in the chapter.
1. 2.4 VAN DER WAALS BONDS AND THE
HYDROGEN BOND
Ioni c, cova lent and metal li c bonds all occ ur beca use
of the need for atoms ro achi eve a sta bl e electron
confi guration; they are strong and are therefore
sometimes known as primary bonds. However, some
form of bondi ng force berween the res ulting molecul es
must be present since, for exa mpl e, gases will all
li quefy and ultimately soli dify at suffi cient ly low
temperatures.
Such secondary bonds of forces are known as
Wan der Waals bonds or Wan der Waals forces and
are universal to all atoms and mol ec ul es; they are
however sufficiently weak that their effect is often
overwhelmed when primary bonds are present. They
arise as fo ll ows. Although in Fig. 1.1 we represented
the orbiting electrons in di sc rete shell s, the true
pi ct ure is that of a cloud, the densit y of the cloud
at any poi nt being related to the probabilit y of
finding an electron there. The elect ron charge is thus
'spread' around the atom, and, ove r a peri od of
time, the charge may be thought of as symmetri call y
distributed with in its parti cul ar cloud.
However, the electroni c charge is moving, and thi s
mea ns that on a scale of nanoseconds the electrostati c
fi eld around the atom is continuously fluctuating,
resulting in the formation of a dynamic electri c dipole,
i.e. the centres of positi ve charge and negati ve charge
are no longe r coincident. When another atom is
brought into proximit y, the dipol es of the two atoms
may interact co-operatively with one another (Fig. 1.8)
and the result is a weak non-directi onal electrostati c
bond.
As well as thi s fluctuating dipole, many mol ecul es
have permanent dipoles as a result of bonding between
different types of atom. These ca n pl ay a consider-
able part in t he structure of polymers and orga ni c
compounds, where side-chains and radical groups
of ions ca n lead ro points of predomi nantl y positi ve
or nega tive charges. These will exert an electrostati c
attract ion on other oppositely charged groups.
The strongest and most important exa mple of
dipole interacti on occ urs in compounds between
hydrogen and nitrogen, oxygen or fluorine. It occurs
Momentary dipoles
Attraction
~ ~
Fig. 1.8 Weak Van der Waa ls linkage between atoms
due to fluctuating electrons fi elds .
@--------8
resulting
dipole
(a) The water molecule (b) The struct ure of water
Fig. 1.9 The hydrogen bond between water molecules.
beca use of the sma ll and simpl e structure of the
hydrogen ato m and is known as the hydrogen bond.
When, for exa mpl e, hydrogen links covalently with
oxygen to form water, the electron contributed by
the hydrogen a tom spends the greater pa rt of its
time between the t wo atoms. The bond acq uires a
definite dipol e with the hydrogen becoming virtuall y
a positively cha rged ion (Fig. J .9a).
Since the hydrogen nucl eus is not screened by any
other elect ron shell s, it ca n attract t o itself other
nega ti ve ends o f dipoles, and the result is the hydrogen
bond. It is considera bl y stronge r (a bout 10 rimes)
tha n other Van der Wa a ls linkages, bur is much
weaker (by 10 to 20 rimes) than a ny of the primary
bonds. Figure 1. 9b shows the res ultant structure of
water, where the hydrogen bond forms a seconda ry
link between the wa ter mo lecul es, and acts as a
bridge between two electronegati ve oxygen ions.
Thus, this relati vely insi gnificant bond is one of the
most vita l factors in the evoluti on a nd su rviva l of
life on Ea rth . It is responsible for the abnorma ll y
Atoms, bonding, energy and equilibrium
hi gh melting and bo iling points of water a nd for its
hi gh specifi c hear, which provides an essential globa l
tempera ture contro l. In the absence of the hydrogen
bond, wa ter would be gaseous at a mbi ent tempera-
tures, like a mmoni a a nd hydrogen sulphide, a nd we
wou ld not be here.
The hydrogen bond is a lso res ponsibl e fo r the
unique property of water of expansion during freezing
i.e. a density decrease. In solid ice, the combina ti on
of covalent and strongish hyd rogen bonds result in a
three-dimensional ri gid but relati vely open structu re,
but o n melting thi s structure is partially destroyed
and the water molecul es become more cl osely packed,
i.e. the densit y increases.
1.3 Energy and entropy
The bonds that we have just described ca n occur
between atoms in gases, liquids and solids and to
a large extent a re res ponsibl e for their ma ny a nd
var ied properti es. Although we hope construction
ma ter ia ls do not cha nge sta te whilst in service, we
a re very much concerned with such changes during
their manufacture, e.g. in the cooling of metal s fr om
the mo lten to the solid state. Some knowledge of the
processes and the rul es governing them are therefore
useful in understanding the structure and properti es
of the materi a ls in their 'ready-to use' state.
As engineers, a ltho ugh we conventi ona ll y express
our findings in terms of force, defl ect ion, stress, stra in
a nd so on, these are simpl y a conventi on. Fundamen-
t a ll y, we a re rea ll y dea ling with energy. Any cha nge,
no marter how simpl e, involves an exchange of energy.
The mere act o f lifting a bea m involves a change in
the potential energy of the bea m, a change in the
strain energy held in the lifting ca bl es and a n input
of mecha ni ca l energy from the lifting device, which
is itself tra nsforming electri ca l or other energy into
kineti c energy. The ha rnessing and control of energy
a re at the heart of a ll engineer ing.
Thermodynamics teaches us a bout energy, a nd
draws a ttenti on to the fact that every materi a l
possesses a n internal energy assoc iated with its
structure. We begi n this secti on by discussing some
of the thermodyna mi c principl es that are of impo rt-
a nce t o underst a nding the behaviour patterns.
1.3. 1 STABLE AND METASTABLE EQUILIBRIUM
We sho uld recogni se that a ll systems a re a lways
seeking to minimi se their energy, i. e. t o become more
stabl e. However, although thermodynamically correct,
some changes toward a more stable conditi on pro-
ceed so slowl y that the system a ppea rs t o be stable
9
Fundamental s
>-
Ol
Q;
c
w
P,
Act1vat1on
energy
________ ! ________ _
---- -- ------ r -
Free
energy
--- -------
Fig. 1. 10 Illustration of activation and free energy.
eve n th ough it is no t . For exa mpl e, a sma ll ball
sitting in a ho ll ow at the top of a hill wi ll remai n
there until it is lifted o ut and rolled down the hill.
The ba ll is in a metastable st a te a nd requires a sma ll
input of energy to st a rt it on its way down the main
slope.
Figure 1.10 shows a ball sitting in a depression wi th
a potenti al energy of P
1
. It will roll to a lower energy
st a te P
2
, but onl y if it is fir st li fted t o the top of the
hump between the t wo ho ll ows. Some energy has
to be lent to the ba ll to do thi s, which the ba ll returns
when it roll s down the hump t o its new positi o n.
T hi s borrowed energy is known as the activation
energy for the process. Thereafter it possesses free
energy as it roll s down to P
2
Howeve r, it is losing
potenti a l energy a ll the time a nd eventua ll y (say, a t
sea level) it wi ll achi eve a st a bl e equilibrium. However,
note t\-vo things. At P
1
, P
2
, etc. it is a ppa rentl y stable,
but actuall y it is metasta bl e, as the re a re o ther more
stable sta tes ava il able to it , give n the necessary
act iva ti on energy. Where does the activation energy
come from? In material s science it is extracted mostl y
(but not exclusively) from heat. As things a re hea ted
to hi gher temperatures the a to mi c particl es react
mo re rapidl y a nd can break o ut of their metastable
sta te into one where they ca n now lose energy.
1.3.2 MIXING
If whisky and water a re pl aced in the same conta iner,
they mi x spontaneously. The interna l ene rgy o f the
res ulting soluti o n is less tha n the sum of the two
interna l energies before they were mi xed. There is no
way tha t we can separate them except by distill ation,
i. e. by hea ting them up a nd coll ecting the vapours
a nd separa ting these into alcoho l and wa ter. We
must, in fact , put in energy t o sepa rat e them. But,
since energy ca n be neither be created nor destroyed,
the fact that we must use energy, a nd quite a lot
of it, to restore the st a tus quo must surely pose
the question ' Where does the energy come from
initi a ll y?' The a nswer is by no means simple but, as
10
we sha ll see, every particle, whether of wat er or
whi sky, possesses kinetic energies of motion and of
1nteracnon.
When a sys tem such as a li q uid is left to it se lf,
its interna l energy remains constant, but when it
interacts with anot her sys tem it wi ll either lose or
ga in energy. T he tra nsfer may in volve work or hea t
or both and the first law of ' the
conserva ti on of energy and hea t', requires rh.1r:
dE= d Q- dW ( 1.1 )
where E =internal energy, Q = hea t and W = work
done by the system on the surro undings. Wh;Jt thi s
tell s us is that if we ra ise a cupfu l of wate r from
20C to 30C it does not ma tter how we do it. We
ca n hea r it, stir it with paddl es or even pur in a
whole army of gnomes each equipped with a hot
wa ter bottl e, but the internal energy a t 30C will
a lways be above that at 20C by exactl y the sa me
a mo unt. Note that the first law says not hing .1bour
the sequences of changes tha t a re necessa ry to br ing
about a change in interna l energy.
1.3.3 ENTROPY
Class ica l thermodyna mi cs, as normall y taught to
engineers, regards entropy, S, as a capacity propert y
of a system whi ch increases in proportion ro the
heat a bsorbed (dQ) at a given temperature (T). Hence
the well known relationship:
dS;? dQJT ( 1.2)
whi ch is a perfectl y good defi niti on but does nor
give any sort of pi cture of the mea ning of entropy
and how it is defin ed. To a materials scienti st entropy
has a rea l phys ica l mea ning, it is a meas ure of the
state of disorder or chaos in the system. Whi sky a nd
water combine; thi s simpl y says that, stat istica ll y,
there a re many ways that the atoms can get mi xed
up and on ly one possible way in which the whisky
ca n st ay on top of, o r, depending on how you pour
it, at the bottom of, the water. Boltzmann showed
tha t the entropy of a system could be represented
by:
S = k In N ( 1.3)
where N is the number of ways 111 which the
particles ca n be di stributed and k IS a const a nt
(Boltzmann 's consta nt k = 1.38 x 10-
23
}/K). The
logar ithmi c relat ionship is important; if the mo l-
ecul es of water can adopt N
1
configurati ons and those
of whisky N
2
the number of possibl e confi gurations
open to the mi xture is not N
1
+ N
2
but N
1
x N
2
It
follows fr om thi s that the entropy of any closed
system not in equilibrium will tend t o a ma ximum
since thi s represe nt s the rn osr proba bl e a rray o f
confi gurati ons. Thi s is the second law o f thermo-
dyna mi cs, fo r whi ch yo u should be very grateful.
As yo u read these words, yo u are keeping a li ve by
brea thing a ra ndoml y di stributed mi xture o f oxygen
a nd nitrogen. Now iri s sra ri sri ca ll y possibl e that at
some instant all the oxygen atoms wi ll coll ect in
one corner o f the room whil e yo u try to ex ist on
pure nitrogen, bur onl y srari sri ca ll y possible. There
a re so ma ny o ther poss ibl e di stributi ons involving
a mo re ra ndom a rra ngement o f t he two gases that
it is most li kely that you will continue to brea the
the norma l ra ndom mi xture.
1.3.4 FREE ENERGY
It must be cl ea r t hat the fund a ment al tendency for
entro py to increase, that is, fo r systems to become
mo re ra ndomi sed, must stop somewhere a nd some-
how, i.e. the system must reac h equi li brium. If nor,
the entire uni ve rse would brea k down into chaos .
As we have seen in t he first pa rr of thi s cha pter, the
reason for the existence o f liquids and solids is that
their atoms a nd mo lecul es a re nor totall y indifferent
to each other a nd, under cert ain conditi ons and
with certain limitati ons, wil l associ ate or bond wit h
each other in a non-random way.
As we stared a bove, from the fir st la w o f thermo-
dyna mi cs the cha nge in interna l energy 1s given
by:
d E= dQ - dW
From the second law of thermodyna mi cs the entropy
cha nge in a reversibl e process is:
TdS = d Q ( 1.4)
Hence:
d E= TdS - dW ('I .5 )
In discussing a system subj ect to change, it is con-
veni ent to use the concept of free energy. For irre-
versibl e changes, the change in free energy is always
nega ti ve and is a measure of the dri ving force leading
ro equilibrium. Since a sponta neous change must
lead to a more pro ba bl e stare (or else it would nor
happen) it foll ows that, at equ ilibri um, energy IS
minimi sed wh il e ent ropy is ma ximi sed.
The Helmho ltz free energy is defin ed as :
H = E- TS (1.6)
a nd the Gibbs free energy as:
G = pV + E- TS (1. 7)
a nd, ar equilibrium, both must be a m1n1mum.
Atoms, bond ing, energy and eq ui librium
1.4 Equilibrium and equilibrium
diagrams
Most of the ma teri als that we use a re not pure bur
consist of a mi xture o f o ne o r mo re constituents.
Each o f the three materi a l sta res o f gases, liquids a nd
solids may consist o f a mi xture o f different compo n-
ents, e.g. in all oys of t wo meta ls. These compo nents
a re ca ll ed phases , with each phase being ho mogen-
eous. We need a scheme that a ll ows us to summa ri se
rhe influences o f temperature a nd press ure on the
relative sta bilit y o f each sta te (a nd, where necessa ry
its compo nent phases) a nd o n the tra nsiti ons that can
occur betwee n these. The time- honoured approach
ro thi s is with equilibrium diagrams. Note the wo rd
equilibrium. T he rmo dyna mi cs te ll s us tha t thi s is
the condit io n in whi ch the materi a l has minimum
interna l energy. By defin iti o n, equilibrium di agra ms
tell us a bout thi s minimum energy state that a system
is t ry ing to reach, but when using these we sho uld
bea r in mind tha t it will a lways ta ke a finite time
fo r a tra nsiti o n fr o m o ne sta te ro a nother to occur
o r fo r a chemica l reaction ro ta ke pl ace. So metimes,
thi s rime is va ni shingly sma ll , as when dyna mite
ex pl odes. At other times, it ca n be a few seconds,
days o r even centuri es. Glass made in the Middl e
Ages is still glass a nd shows no sign o f cr ys ta ll ising.
So, no t eve ry s ubstance or mi xture that we use has
reached t hermodyna mic equ ilibrium.
We onl y ha ve space here to int roduce some of the
elements o f the grea t wea lth o f fund a menta l theory
underl ying the fo rms o f equi librium diagra ms.
1.4.1 SINGLE-COMPONENT DIAGRAMS
The temperature-pressure di agram for water (Fig. 1.1 1)
is an impo rta nt exa mpl e of a single-component
diagra m, and we can use thi s to esta bli sh some ground
r ul es a nd language for use later.
The d iagram is in ' tempera ture-press ure space'
a nd a number of lines are marked which represent
bo unda ry conditi o ns between differing phases, i. e.
states o f H
2
0 . The line AD represents combinat io ns
o f temperature a nd pressure at whi ch liquid wa ter
a nd solid ice a re in equi li brium, i.e. ca n coexist . A
sma ll hea t input wi ll a lter t he proporti o ns o f ice
a nd wa ter by melting some o f the ice. However, it
is a bsorbed as a cha nge in interna l energy o f the
mi x ture, the la tent hea t of melting. The temperature
is not a ltered, bur if we put in la rge a mo unts o f
hea r, so that a ll the ice is melted a nd there is some
heat left over, the temperature rises and we end
up with sli ghtl y wa rmed wa ter. Similarl y, line AB
represents the equi li brium between li quid wa ter a nd
11
Fundamental s
1000
100
Ul
0
10
E
I
Liquid
:_ x:
y
<ll 1.0
__ _,
:;
Ill
I
Ill I
~ 0.1 Ice'
a_
I
I
0.01
0.001
- 20 0 20 40 60 80 100 120
Temperature (C)
Fig. 1. 11 Pressure-temperature diagram for wat er
(from Kingery er al., 1976).
gaseous stea m, and line AC the equilibrium between
solid ice and rather cold wa ter vapo ur.
It is helpful to consider what happens if we move
around within the di agra m. First, let us start at
po int X, represent ing -5C a t a tmospheri c pressure.
We know we should have ice a nd, indeed, the point
X lies within the phase field labell ed ice. Adding
heat at constant pressure takes the temperature a long
the broken line. This crosses the phase boundary,
AD, at 0C (po int Y) a nd the ice begins t o melt as
further heat is added. Not until a ll the ice has melted
does the temperature continue to ri se. We now have
liquid water until we reach 100C (po int B). Now,
again, heat has to be added t o boil the water but there
is no tempera ture increase until a ll the liquid wa ter
has gone. We now have stea m a nd irs tempera ture
ca n be increased by further hea t input.
Next think of keeping temperature const a nt a nd
increasing press ure, aga in st arting at point X. If the
press ure is raised enough, to about 100 at mospheres
(=: 10 MPa, point D) we reach the ice-water equili briw11
and the ice ca n begin t o melt . Thi s accounts for the
low frict ion between, for exa mpl e, an ice ska te a nd
the ice itself: local press ures ca use local melting. It
is a factor that engineers need to consider when
contemplat ing the use of loca lly refri ge ra ted a nd
frozen ground as coffer da ms o r as fo unda ti o ns for
oil ri gs in Alaska.
The Gibbs phase ru le is a forma l way o f sum-
ma rising the relati onship between the number of
phases (P) tha t ca n coexist at a ny given point in
the di agram and the changes bro ught about by sma ll
changes in temperature or pressure. This states t hat:
P +F=C+2 (1.8)
12
Here, C is the number of components in the system;
in thi s case we have only H
2
0 so C = I. F is the
number of degrees of freedom a ll owed to cha nge.
To illustrate, at point X in Fig. 1.11 there is just
one phase, ice, soP= 1 and F = 2. Th is means that
both temperature a nd press ure can be cha nged
independentl y wit hout bringing about a signifi ca nt
change to the material. At Y both solid a nd liquid
ca n coexist , so P = 2 and F = 1. To maintain the
equilibrium, temperature and pressure must be changed
in a co-ordinated way so that the point Y moves
a long the bo unda ry AD. At A, a ll th ree phases can
coexist so P = 3, therefore F = 0, i. e. a ny change at
a ll will di sturb the equilibrium.
1.4.2 TWO-COMPONENT DIAGRAMS
We now go on to look at two-component diagrams,
such as we get with a lloys between two meta ls o r
between iron and carbon. We now have a further
var iable, compositi on a nd, stri ctl y, we shou ld con-
sider the joint influences of thi s va ri a bl e in add iti on
to temperature a nd press ure. We would therefore
need three-dimensional diagrams, but to simplify
things we usua lly take press ure to be constant . After
all, most engineering materia ls are prepared and used
at atmospheric pressure, unl ess yo u work for NASA!
This leaves us with a compositi on-tempera ture
diagram, the lifeblood of mater ia ls sc ienti sts.
The a ll oys formed bet ween copper, Cu, and ni ckel,
Ni (Fig. 1.12 ) produce an exa mple of the simples t
form of two-component di agra m. Thi s is drawn
wit h compositi on as the hori zont a l axis, one end
representing pure (100%) Cu, t he ot her pure (100%)
Ni. T he verti ca l axis is temperat ure.
Let us think a bout an a ll oy t hat is 50%Cu:50%Ni
by mass. At hi gh temperatures, e.g. at A, the a lloy
is tot a lly mo lten. On cooling, we move down the
Solid
Cu
Cu/Ni :50/50
Ni
Composition (mass%)
Fig. 1. 12 Equilibrium phase diagram for copper-
nick el.
compositi on line until we a rri ve at B. At thi s tem-
pera ture, a tin y number of sma ll crystals begin to
fo rm. Further reducti on in temperature brings about
a n increase in the a mount of solid in equili brium
with a dimini shing a moun t of liquid. On a rri ving at
C, a ll the liquid has gone a nd the materi a l is totall y
solid. Further cooling brings no further changes .
Note that there is an importa nt difference between
thi s all oy and the pure metals of whi ch it is composed.
Both Cu and Ni have we ll defin ed unique melting
(or fr eezing) temperatures bur the all oy solidifi es
over the temperature range BC; metallurgists oft en
spea k o f the ' pas t y ra nge'.
We now need to exa mine seve ral matters in more
deta il. First, t he solid crystals that form a re what is
known as a 'solid soluti on' . Cu and Ni are chemi ca ll y
simil a r elements a nd both, when pure, fo rm face-
centred cubi c crystals (see Chapter 3) . In thi s case,
a 50:50 all oy is also composed of face-centred cubi c
crys tals but each la tti ce sire has a 50:50 chance of
being occupi ed by a Cu atom o r a Ni atom.
If we appl y Gibbs's phase rul e at po int A, C = 2
(two components, Cu & Ni) and P = 1 (one phase,
liquid ) a nd so F = 3 (i. e. 3 degrees o f freedom). We
ca n therefore independentl y alter compositi on, tem-
perature and pressure and the structure remains liquid.
But remember, we have taken press ure to be constant
and so we are left with 2 practical degrees of freedom,
compos iti on a nd temper ature. The same argument
ho lds at point D, but, of course, the structure here
is the cr ystalline solid soluti on of Cu a nd Ni.
At a point bet ween 8 a nd C we have liquid and
solid phases coexisting, so P = 2 a nd F = 2. As
before, we must di scount one degree of fr eedom
beca use press ure is ta ken consta nt. Thi s lea ves
us with F = 1, whi ch mea ns tha t the status quo ca n
be maintained onl y by a coupl ed change in both
compos iti on a nd temperature. Therefore, it is not
onl y tha t the structure is two phase, but also that
the proporti ons o f liquid a nd solid phases remain
unaltered.
We ca n fin d the propo rti ons o f liquid and solid
corresponding to a ny point in the two phase fi eld
using the so-ca ll ed Lever rule. The fir st step is to
draw the constant temperature line through the point
X, Fig. 1.1 2. Thi s intersects the phase boundari es
at Y and Z. The solid line containing Y represents
the lower limit of ] 00% liquid, a nd is known as the
liquidus . The solid line containing Z is the upper
limit of 100% solid and known as the solidus.
Neither the liquid no r solid phases corres ponding
to point X have a compositi on identica l with that
of the a ll oy as a whole. The liquid contains mo re
Cu a nd less Ni, the solid less Cu a nd mo re Ni. The
Atoms, bonding, energy and equilibrium
compos iti o n o f eac h phase is given by the points Y
a nd Z, respecti ve ly. The proporti ons of the phases
ba la nce so tha t the weighted average is the sa me as
the overa ll compos iti o n o f the a ll oy. It is easy to
show tha t:
(Weight o f liquid o f compositi on Y) x YX =( Weight
o f solid o f compositi o n Z) x X Z (1.9)
Thi s is simil a r ro wha t would be expected o f a
mec ha ni ca l leve r ba la nced a bout X, hence the na me
Lever rul e.
O ne consequence o f a ll thi s ca n be seen by re-
examining the cooling o f the 50:50 a ll oy from the
liquid phase. Consider Fig. 1.1 3. At point X
1
on
the liquidus, soli d ifica ti on is a bout to begin. At a
temperature infin itesima ll y below X
1
there will be
some crysta ls solidi fy ing o ut of the liquid; their
composition is given by Z
1
At a temperature a bout
halfway between solidus a nd liquidus {X
2
), we have
a mi xture of solid a nd liquid of compositi ons Z
2
a nd Y
2
. In genera l, the propo rti on o f liquid to solid
ha lfway thro ugh the freez ing ra nge need not be
""50:50, but in thi s case it is. Fina ll y, at a temperature
infinites ima ll y a bove x3, which is on the solidus,
we have nea rl y 100% solid o f compos iti on Z
3
rogether with a va ni shingly small a mount o f liquid
o f compositi on Y
1
When the temperature fall s to
just below X
3
, the all oy is tota ll y solid and Z
1
has
become identica I with X
1
.
Note two impo rta nt First, Z
3
is the sa me
as the ave rage compositi o n we sta rted with, X
1

Second, solidifi ca ti o n t a kes pl ace ove r a ra nge o f
tempera tures, a nd as it occ urs the compositi ons o f
liquid a nd solid phases cha nge continuously. For thi s
to actuall y ha ppen, substanti a l amounts of diffusion
Solid
Cu
Composition
Fig. 1. 13 Equilibrium phase diagram for Cu-Ni
(Fig. 1.1 2 redrawn to show composition variations
with temperature) .
Ni
13
Fundamentals
1400
1200
G
e..... 1000
Ql
:s
~
Ql
0..
E
Ql
f-
800
400
y
, A
........ .. ......... ........... ~ E . l ..
AI + Si
Silicon (mass%)
Si
Liquid + Si
Fig. 1.14 Equilibrium phase diagram for aluminium-
silicon.
must occ ur in both liquid and solid. Diffusion in
solids is ve ry much slower than that in liquids and
is the source of some practi ca l di ffi cult y. Either
solidifi cati on must occur slowly enough for diffusion
in the solid to keep up or stri ct equilibrium conditi ons
are not met. The kineti cs of phase transformati ons
is therefore of interest, but for the moment, we will
continue to di scuss ve ry slowl y formed, equilibrium
or nea r equilibrium structures.
1 .4. 3 EUTECTIC SYSTEMS
Ler us now exa mine another di agram, that for
aluminium-sili con (AI-Si) all oys (Fig. 1.14 ). Pure
AI fo rms face-centred crys tals (see Chapter 3) bur
Si has the sa me crys tal structure as di amond. These
are incompat ibl e and extensive solid soluti ons like
those for Cu:Ni ca nnot be fo rmed. Si crys tal s ca n
di ssolve onl y tiny amounts of AI. For our purposes,
we ca n ignore this solubilit y, alth ough we mi ght
recogni se rhar the semi conductor industry makes
great use of it, small as it is. AI crys tals ca n dissolve
a littl e Si , bur aga in nor ve ry much, and we will
ignore it. Thus, two solid phases are poss ibl e, AI
and Si. When liquid, rhe elements di ssolve readil y
in the melt in any proporti ons.
Consider rhe compos iti on Y. On cooling to the
liquidus linear A, pure (or nea rl y pure) crystals of
Si begin ro form. At B we have solid Si coex isting
with liquid of compositi on LB in proporti ons given
by the Lever rul e. At C we have solid Si in equilibrium
with liquid of compositi on close to E.
14
Now consider all oy X. The sequence is much rhe
same except the first solid to form is now AI. When
the temperature has fa ll en to almost T ~ we have
soli d AI in eq uilibrium with liquid of compos iti on
close to E. ore rhar borh all oy X and all oy Y,
when cooled to TF, contain substanti al amounts of
liquid of compositi on E. An infinitesimal drop in
temperature below T
1
causes thi s liquid to solidify
into a mi xture of solid AI and solid Si. At E we
have 3 phases whi ch ca n coex ist; liquid, solid AI
and solid Si. The system has two components and
thus the phase rul e gives us no degrees of freedom
once we have di scounted pressure. E is an in va ri ant
point; any change in temperature or composition will
di sturb the equilibrium.
The point E is known as the eutecti c point and we
spea k of the eutecti c compositi on and the eutecti c
temperature, T
1
, . Thi s is the lowest temperature ar
whi ch liquid ca n exist and the eutecti c all oy is thar
whi ch remains liquid down to T
1
. It solidifi es at a
unique temperature, quite unlike Cu- i or AI-Si
all oys of other compositi ons. All oys close to the
eutecti c compos iti on ("" 13%Si) are widely used be-
ca use they can be eas il y cast into compl ex shapes,
and the Si di spersed in the AI strengthens it. Eutecti c
all oys in other systems find simil ar uses (cast-iron
is of nea r eutecti c compos iti on) as well as uses as
brazing all oys etc.
1.4.4 INTERMEDIATE COMPOUNDS
Oft en, the bas ic components of a system ca n form
compounds. In metals we have CuAI
2
, Fe
1
C and many
more. Some other relevant exa mpl es are:
Si0
2
and corundum, Al
2
0
1
, whi ch form mullite,
3(AI
2
0
1
)2(Si0
2
), an important constituenr of fired
cl ays, pottery and bricks. Figure 1.1 5 shows the
SiOcAI
2
0
1
diagram. It can be thought of as two
di agrams, one for ' Si 0
2
-mullite' and the other
fo r ' mullire-AI
2
0,' , joined together. Eac h part
di agram is a simpl e eutecti c system like AI-Si;
lime, CaO, and silica , Si0
2
, which form the
compounds 2(Ca0 )Si0
2
, 3(Ca0)Si0
2
and others,
whi ch have grea t technological signifi ca nce as
acti ve ingredi ents in Portland cement (to be di s-
cussed in detail in Chapter 13). In a simil ar way
to mullite, the lime (CaO)-sili ca (Si0
2
) di agram
(Fig. 1.1 6) ca n be thought of as a seri es of joined
together eutecti c systems.
In many cases we do not have to think about the
whole di agram. Figure 1. 17 shows the Al-CuAI
2
di agram, aga in a simpl e eutecti c system. A notabl e
fea ture is the so-ca ll ed solvus line, AB, whi ch
represents the solubilit y of CuAI
2
in solid crystals
2200
Liquid
2000

Mulli te + Liquid
Corundum

+ liquid
::l
1800 Crystobalite
"'
(jj
+ li quid
a.
E
Mull ite + Li quid
Q)
Corundum 1-
1600
+ mull ite
Crystobalite + Mullite
1400 L-----L-----'----'---'---'--------'-----LI>r'----'---'----'
Si0
2
20 40 60 80 Al,0
3
Al
2
0
3
(mass%)
Mull ite 3AI
2
0
3
:2Si 0
2
Fig. 1.15 Equilibrium phase diagram for silica (SiO
2
) -
alumina (AI
2
0
1
) .


::J
'
Q)
o._
E
Q)
1-
2600
2200
1800
F
1400
1000
600
Si0
2
[ C = CaO l C
3
S--v
s = sio, c
2
s . _ :
Liqcid c,s,:l"
:
I
20 40 60 80 CaO
CaO (mass%)
Fig. 1.1 6 Equilibrium phase diagram fo r lime (CaO)-
silica
Atoms, bonding, energy and equilibrium
700
Liquid
600

500
::J
CuAI
2
co
Qj
solvus o._
400
E
a+ CuAI
2
Q)
1-
300
B
200
AI 10 20 30 40 50
Copper (mass%)
Fig. 1. 17 Equilibrium phase diagram for AI-CuAI
2

o f AI. Thi s curves sha rpl y, so tha t very much less
CuAI
2
will di ssolve in AI a t low tempera tures tha n
wi ll a t hi gh t emperatures. Thi s is a fortuna te fact
tha t underli es o ur a bility to a lter the mi crostructures
o f some a ll oys by suita bl e hea t-trea tments, di scussed
in mo re det a il la te r.
We ha ve no t yet cons idered the iro n-carbo n
di agra m, whi c h is pe rha ps the most impo rt a nt
di agra m for nea rl y al l engineers. T hi s is of pa rti cul a r
relevance in civil a nd structu ra l engineering s ince
st eel in a ll its fo rms is used extensively. We wil l
leave di scussing thi s until Cha pter 11.
References
Kinge ry WD, Bowen HK a nd Uhlma nn DR (1976). Intro-
duction to Ceramics, 2nd ed iri on, J ohn Wil ey a nd Sons,
New York.
Ca lli srer WD (2007). Mareria ls science and engineering. An
inr roducri on 7rh edn., John Wil ey and Sons, New York.
15
Chapter 2
Mechanical properties
of solids
We have seen in Chapter 1 how bonds are formed
between atoms to for m bulk elements and com-
pounds, and how changes of state occur, with an
emphasis on the formation of solids fr om molten
liquids. The behaviour of solids is of parti cul ar
interest to constructi on engineers for the obvious
reason that these are used to produce load- bea ring
structures; in th is chapter we define the properti es
and rules used to quantify the behaviour of solids
when loaded. To understand thi s behaviour and there-
fore to be able to cha nge it to our advantage we
need to consider some other aspects of the st ructure
and nature of the mater ials beyond those discussed
in Chapter 1; we wi ll do thi s in Chapter 3.
You wil l find it necessa ry to refer to the defi nitions
etc. given in thi s chapter when readi ng the sub-
sequent on indi vidual materia ls. Alt hough we will
include here some exa mpl es of the behaviour of
construction materials, all of t he definiti ons and
explanati ons are applica bl e to any materia ls being
used by engineers of any di sci pline.
2.1 Stress, strain and
stress-strain curves
Load ing ca uses materi als to deform and, if high
enough, to brea k cl own and fai l. All load ing
on materi als ca n be considered as combinat ions of
three basic types - tension, compression and shea r.
These are normally shovvn di agra mmati ca lly as in
Fig. 2.1.
-
-- LI _ __ T_e_ns_io_n __ __,l -
_. I Compression I --
-
Fig. 2.1 Basic types of load.
16
Clea rl y the deformat ion from loading on an
element or test specimen will depend on both its
size and the properties of the material from which
it is made. We ca n eliminate the effect of size by
converting:
the load to stress, cr, defi ned as load, P, divided
by the area, A, to which is applied, i. e.
cr = PIA (2.1 )
and
the deformation to strain, , defin ed as change in
length, t,.l, divided by original length, I, 1. e.
= t:,./11 (2.2)
These definiti ons are ill ustrated for simpl e tension
in Fig. 2.2a. Compressive stress and strain are in
Cross-sectional
1 area A
Initial length, I Extension, 6 /
Tensile stress, o = PI A Tensile strain, c = 6111
(a) Tension
p
Displacement , x
f.-1
Loogth] .____ -=---'
shear force, P
Loaded
area, A
Shear stress, T = PIA Shear strain, y = xl l
(b) Shear
Fig. 2.2 Definitions of tensile and shear stress and strain.
th e oppos ite direct ions. The equi va lent definiti ons
of shea r stress, T, and shear strain, y, which are not
quite so obvious, are shown in Fig. 2.2b.
As wi th all quantiti es, the dimensions and units
must be considered:
stress = load/a rea and therefore its dimensions
are [ForceJ!J Length]
2
. Typical units are N/mm
2
(or MPa in rhe Sl system), lb/i n
2
and tonf/ft
2
in
the Imperi al system.
strain = change in length/ori ginal length and there-
fore irs dimensions are [Lengt h] /[Lengt hJ, i.e. it
is dimension less.
However, strain va lues ca n be very small and it is
often convenient to use either:
percentage st ra in (or % strain )= strain x 100
or mi crostra in (Jl s) = stra in x 10
6
As we ll as the linea r strain, we ca n also simil arl y
define
volumetric strain (EJ
= change in volume(Ll V)/ori ginal volume( V) (2.3)
The relat ionship between stress and strain IS an
extremely important character isti c of a materi al. It
va ri es with the rate of appli cati on of srress (or load);
we wi II consider four cases:
a) steadily increasing - zero to failure in a few
minutes, e.g. as in a laboratory test
b) permanent or static - constant with time, e.g.
the self we ight of the upper part of a structure
acting on the lower part
c) impact or dynamic - very fast, las ting a few
microseconds, e.g. the impact of a vehicle on a
crash barrier, or an expl osion
d) cyclic - va ri able wi th load reversals, e.g. ea rth-
quake loading- a few cycles in a few minutes,
and wave load ing on an offshore structure -
many cycles over man y yea rs.
For the moment, we will confine ourselves to case
(a): steady loading to fa il ure in a few minutes. This
is what is used in the most common types of labora-
tory tests that are used to meas ure or characteri se
a mater ial's behaviour.
There are a wide variety of different forms of
stress-strain behaviour for different materi als;
Fig. 2.3 shows those for some common mater ials.
Mosr have at least two di stinct regions:
An initi al linea r or nea r-linear region in which
1000
800
ro-
600
Cl.

<Jl
<Jl

(jj
400
200
0
0
Mechanical properties of sol ids
Mild steel
1
Aluminium alloy
,........___ Unloading behaviour
I
Cast iron,'
I
I
5 10 15 20 25 30
Strain(%)
Fig. 2.3 Typical tensile stress-strain curves (or some
structural materials.
and thi s porti on of the graph the elastic region;
if the behavi our is also linea r, we ca ll thi s linear
elasticity. The strains involved are usuall y sma ll.
An increasingly non-linear region in which the
strains can increase significantl y with progressively
small er increments of stress. Un loading at any
point in thi s region will result in t he strain
reducing along a line para ll el to the initi al elasti c
region, and hence there is a permanent deforma-
ti on at zero load (as shown on the graph for an
aluminium all oy). This is known as the plastic
region, and the permanent deformation as plastic
deformation.
Eventuall y, of course the material breaks, whi ch
may occur at the end of either the elasti c or the
plastic region, sometimes after an apparent reduc-
ti on of st ress (as shown for mi ld steel).
We now take a cl ose look at each of these regi ons
in turn, starting with the elastic region, defining
some of the material constants that are used to
quantify the behaviou r as we go.
2.2 Elastic behaviour and the
elastic constants
In
the stra in is full y recove red when
removed, i. e. the material returns to its ini ri
and dimensions. Thi s is call ed elastic deft r .
' @ LIBRARY
l_ ..,.. __ --"-------'
17
Fundamentals
rJ)
rJ) cr ,
~
iJj
Tangent modulus at
cr , = tan e
Strain
Secant modulus between
cr , and cr
2
= tan e
Fig. 2.4 Definitions of tangent and secant 111 oduli of
elast icit y.
defined as follows, are used ro calcul ate defl ections
and movement under load.
2.2.1 THE ELASTIC MODULI
For linear elastic materi als, stress is proportiona l
to strain (Hooke's law) and for uni ax ial tension or
compression we ca n defin e:
Young's modulus (E)= slope of stress-strain
graph = srress/stra in = a! ( 2.4)
[E is also known as the modul es of elasti cit y,
the E-modulus or simpl y the stiffness . I
Since strain is dimension less, the dimensions of E are
the sa me as those of stress i.e. I Force ]/!Length 1
2
. Con-
ven ient SI units to avoid large numbers are kN/ mm
2
or CPa.
For materi als that have non-linear elasti c behav-
iour (quire a few, particularl y non- metals) a modulus
va lue is still useful and there are some alternati ve
definiti ons, illustrated in Fig. 2.4:
The tangent modulus is the slope of the tangent
to the curve at any stress (which should be
quoted) . A special case is t he tangent modulus at
the origin i.e. at zero stress.
The secant modulus is the slope of the straight
line joining two points on the curve. Note that
stress levels corres ponding to the two points must
be given. If onl y one stress is given, then it is
reasonabl e to assume that the other is ze ro.
E-va lues for construction materi als range from
0.007 CPa for rubber to 200 CPa for steel (di amond
is stiffer still at 800 CPa, bur this is hardl y a con-
structi on materi a I). Va I ues therefor e va ry ve ry
widely, by more than 4 orders of magnitude from
rubber to steel.
1
1
We di scuss va lues for the major groups of materi als in
the relevant pa rts of the book, and then make compa ri-
sons of thi s and other key properti es in Chapter 61.
18
For shea r load1ng and deformati on, the equivalent
to E is the
shear modulus (G)
=shea r stress( r )/shea r strain (y) (2.5)
G, which is sometimes ca ll ed t he modulus of rigid-
ity, is another elastic constant for the mate rial, and
it has a different numerical va lue to E.
The bulk modulus is used when estimating the
change in volume of a materi al under load. In the
case of uniform st ress on a material in all directi ons
i. e. a pressure ({J) as mi ght be found by submergence
of the specimen to some depth in a liquid:
The volumetric strain (_.) = change
(reduction ) in volume/ori ginal volume (2.6)
and
the bulk modulus ( K) = pi, (2. 7)
2.2.2 POISSON'S RATIO
When a materi al is loaded or stressed in one direc-
ti on, it will deform (or st rain ) in the direction of
the load, i.e. longitudinall y, and perpendicular
to the load i.e. laterall y. The Poisson 's ratio is the
rat io of the strain in the direction ro. Thus in
Fig. 2.5:
, = x/L (extension)
,. = - y/a and ~ . = -zlh
(The - ve sign indicates contract ion.)
The y and z direct ions are both perpendi cu lar to
the directi on of load ing, x,
Poisson's ratio (u) = -/, = -), (2.8 )
The Poisson's rati o is another elasti c constan t for
the materi al. The minus sign ensures that it is a
positi ve number. Va lues for common material s vary
from 0. 15 to 0.49 (see Table 6 1.1 in Chapter 6 1 ).
)-X
z
- tio ( Q' ;;,-
i:=====L=====f r;
Longitudinal strain,,= x/ L Lateral strain Ey = - yl a = r, = - z/ b
Poisson's ratio, u = -/, = - ,/f,
Fig. 2.5 Definiti on of Poisson '.s ratio.
W'e should note that the above definiti ons of
E G and u ass ume that the materi al has simil ar
p;operri es in all directions (i.e. it is isotropic) and
rherdore there is a single va lue of each elasti c
con t.lnt fo r any directi on of loading. Anisotropic
mate ri als, i.e. those which have different properti es
in dii fferent directi ons, e.g. timber, will have differ-
ent v.1lues of , C and u in each di rect ion, and
cl ea r- lv the di recti on :111d well as the va lue itself must
then be stared.
2.2.3 RELATIONSHIPS BETWEEN THE
ELASTIC CONSTANTS
The fo ur elasti c constants that we have now defin ed,
E, C, u and K, mi ght at first glance seem to describe
different aspects of behaviour. It is possibl e to prove
that rhey are nor independent and that they are
related by rhe simpl e express ions:
and
= 2C( I + u)
K = /3( 1 - 2u)
(2.9)
(2. 10)
The proof of these expressions is nor undul y diffi cult
(see fo r exa mpl e Case, Chil ver and Ross, 1999) but
what is more important is the consequence that if
you know, or have measured, any two of the con-
stants then you ca n calcul ate the va lue of the others.
Many materi als have a Poisson's rati o between 0.25
and 0.35, and so the shea r modulus (C) is oft en
about 40 % of the elasti c modu lus () .
Equati on 2. 10 tell s us something about the limits
to rhe va lue of Poisson's rati o. We have defined
the hul k modulus, K, by considering the case of
the change in volume of a specimen under press ure
(equati on 2.7) . Thi s change must always be a reduc-
ti on, as it would be inconceiva bl e for a material to
expand under pressure - i. e. in the sa me directi on
as the pressure. Therefore K must always be positi ve
and since E is also positi ve (by definiti on) then
( 1 - 2u) must be positi ve, and so
u ~ 0. 5 ALWAYS! (2.11 )
A material with u > 0. 5 ca nnot ex1st; if you ha ve
carri ed out some tests or done some calculati ons
that give such a value, t hen you must have made a
mistake. It also foll ows that if u = 0.5 then K is
zero and the materi a I is i ncom pressi bl e.
2.2.4 WORK DONE IN DEFORMATION
The work done by a load when deforming a mater-
ial, although not an elastic constant, is another useful
value. Work is force x di stance, and so
W = J: Pde
(2.12)
Mechanical properties of soli ds
where W =work done by the load Pin ca using an
extension e.
The wo rk done on unit volume of the materi al
of length I and cross-secti on A is:
W = Pcl e/AI = - - = ad
J
,. j" P de J[
o o A I o
(2. 13)
whi ch is the a rea under rhe stress-strain curve.
Thi s work must go somewhere, and it is stored
as inte rnal strain energy within rhe materi al. With
elasti c deformati on, it is ava il abl e to return the
materi al to its zero state on unl oading; in pl as ti c
deformati on, it permanentl y deforms the mater ial
a nd, eventuall y, it is suffi cient to ca use fracture. We
will expl ore the relationship between thi s energy
and fracture in more deta il in Cha pter 4.
2.3 Plastic deformation
As we have sa id, deformati on is plastic if it results
in permanent deformation after load removal.
In ve ry broad terms, materi als ca n be di vided into
those that are:
Ductile - whi ch have large pl as ti c deformati on
before failure (say strains> 1 %)
a nd those that are:
Brittle - with little or no pl as ti c deformati on
before failure (say strains < 0.1 %)
Some exa mpl es of stress- strain curves of each
type of materi al have been shown in Fig. 2.3. It
foll ows fr om equati on (2.1 3) that duct il e materi als
require much greater amount s of work and have
much grea ter amounts of internal strain energy
befo re failure. There are clea rl y some intermedi ate
materi als, but engineers generall y prefe r to use
ductil e materi als that give warning of di stress before
failure in the event of overl oad. Brittl e materi als fail
suddenl y without warning - oft en catastrophicall y.
Significant pl as tic deformati on obviously occurs
onl y in ductil e materi als. We ca n use the ideali sed
stress-strain curve for mild steel shown in Fig. 2.6
to illustrate common features of thi s behaviour.
There is a sharp and di stinct end to the linea r
elas tic behaviour (point A), ca ll ed the limit of
elasticity or the yield point .
There is a region of increasing st ra in with littl e
or no increase in stress (AB), often very short .
Un loading in the plasti c region (sa y from a
point x ) produces behaviour parall el to the initi al
(linear ) elasti c behavi our. Reloading produces
19
Fundamentals
(f)
(f)

Ui
0
D
Fai lure
Y'
:; y
Strain
Pl asti c
Fig. 2.6 Stress-strain wrue for mild steel.
simil ar elasti c behaviour up to the unl oad point,
and the defor mation then continues as if the un -
load/reload had nor occ urred, i. e. rhe materi al
' remembers' where it was.
Another feature of pl astic behaviour, not apparent
fr om Fig. 2.6, is that the deformati on rakes pl ace
at constant volume, i. e. the Poisson's rati o is 0.5
for deformation beyond the yield point.
Two important implicati ons for engineers are:
l. The stress at the yield point A, ca ll ed rhe yield
stress (crJ , is an important propert y for design
purposes. Working stresses are kept safely below
thi s.
2. If, before use, rhe materi al is loaded or strained
to say a point x beyond rhe constant stress
region, i.e. beyond B, and then unl oaded, it
ends up at point y. If it is then used in t his stare,
the yield stress (i. e. at x ) is grea ter than the
ori ginal va lue (at A) i. e. rhe material is 'stronger'.
Thi s is known as work hardening or strain
hardening (or sometimes cold working) to di s-
tingui sh it from other methods of strengthening
that involve hea t treatment (which we will di s-
cuss in Chapter 8). The working stresses ca n
therefore be increased. The drawbac k is that the
failure strain of the work- hardened materia l
(fr om y to fa ilure) is less than that of the ori-
ginal materi al (fr om 0 to failure) and so therefore
is the total work ro fracture. The work -hardened
material is therefore more brittle.
If there is no dist inct end to the elasti c behaviour
i. e. the graph graduall y becomes non-linea r, then
an alternative to the yield stress call ed the proof
stress is used instead. This is defin ed and obtained
as shown in Fig. 2.7:
1. A tangent is drawn to the stress- strain cur ve at
the ori gin .
20
(f)
(f)

Ui
0.1% proof
stress
0.1%
(2)
Fig. 2. 7 Determination of proof stress.
Strain
2. A low va lue of strain is selected - norma ll y either
0. 1% (as in the fig ure) or 0.2%.
3. A line is drawn through thi s point para ll el to the
tangent at the or igin.
4. The stress va lue at the point where thi s intersects
the stress-strain curve is the 0. 7% proof stress.
(If a strain va lue of 0.2% is chosen, then the
result is the 0.2 % proof stress.)
2.4 Failure in tension
The form of fa ilure in uni ax ial tension depends on
whet her a materi al is brittl e or ductil e. As we have
already said, brittl e materi als fail wit h littl e or no
pl as ti c deformati on; failure occurs suddenl y without
warni ng, and the fracture surface is perpendi cu lar
to the directi on of loading (Fig. 2.8).
Ductil e mater ials nor on ly undergo large strains
before fai lure, but often have an apparent reduc-
ti on of stress before failure (i. e. beyond point D in
Fig. 2.6) . Up to the maximum stress (D), the elong-
ation is uniform, but after this, as the load starts to
decrease, a loca li sed narrowing or necking can be
seen somewhere along the length (Fig. 2.9a).
As t he stress continues to fall (bur still at increas-
ing strain ) the di ameter at the neck also decreases,
unti l, wit h very ductil e materi als, it reaches almost
zero before fa ilure, whi ch rakes the for m of a sharp
- .._I ...._
Fig. 2.8 Brillle fai lure in tension.
(a) Necki ng in ductile materi als in the reducing stress
region of the stress-strain curve
(b) Chisel-point failure in very ductile materials
cone cup
(c) 'Cup and cone' failure in medium-strength metals
Fig. 2.9 Necking and fa ilure in ductile mat erials in
tension.
po int (Fig. 2.9b) . Thi s form of failure is extreme,
a nd occurs onl y in very ductil e materi a ls such as
pure met als, e.g. lea d and gold, o r chewing gum
(t ry it for yourself). These materi a ls tend to be wea k,
and most ductile structural materi a ls fail at a stress
a nd stra in some way down the fa lling pa rt of the
curve but with the stress well a bove ze ro. Necking
still occurs after the maximum stress and fa ilure
occurs at the na rrowest secti on in the fo rm of a
' cup and cone' (Fig. 2. 9c). The inner part of the
fa ilure surface is perpendi cul a r to the a ppli ed load,
as in a britt le fai lure, a nd the cracks first form here.
The outer rim, at a bout 45 t o thi s, is the fin a l ca use
of the fa ilure.
2.5 True stress and strain
The behaviour shown in Fig. 2.6 shows the fai lure
occurring at a lowe r st ress than the maximum, i.e.
the ma teri a l seem to be getting wea ker as it ap-
proaches fa ilure. It fact, the opposite is occurring,
and the reason why the stress appea rs to fa ll is
beca use of the way we ha ve ca lcul ated it. We ha ve
defin ed stress as load/area and Fig. 2.6 has been
o bta ined by dividing the load (P) by the ori gi nal
a rea before loading (A
0
) . The stress that we have
obtained should stri ctl y be ca ll ed the nominal stress
( crnonJ, i. e.
(2. 14)
In fact , the cross-sectional a rea (A) is reducing
throughout the loading i. e. A < A
0
. At any load,
Mechanical properties of solids
True
fai lure
stress

U5
Nominal
failure st ress
True ---.__ ,. /
stress : .... ....
Nominal
stress
/.
Strain
Fig. 2.10 True and nominal stress/strain behaviour.
the true stress (cr".,,.) wi ll therefore be hi gher than
the nomina l stress, i. e.
(2. 15)
In the elasti c and pl ast ic regions the reducti on is
uniform a long the length (the Poisson's ratio effect )
but the magnitude of the stra ins in volved are such
that the difference between the no mina l and the
actua l a rea is very sma ll. However, once necking
sta rts the a rea of the neck reduces at a rate such that
the true stress continues to increase up to fai lure,
as shown in Fig. 2. 1 0.
In the case of stra in, the relati on between the
increment of cha nge in length (de), the increment
of stra in (d) a nd the length (/) is, by definiti on
d = dell (2.16)
a nd so the true strain () is
- J'
1., I /
0
(2.17)
where /
0
is the initi a l length
True st rain is not difficult to calculate, but measure-
ment of the cr oss-sectional a rea throughout the
loading, and hence ca lculati on of the true stress, is
more diffi cult, and therefore true stress-stra in gra phs
are ra rely o btai ned, except perha ps for research
purposes. Howeve r, meas urement of the size o f the
neck a fter fr acture is easy, whi ch ena bl es the true
fracture stress to be readil y o bta ined.
2.6 Behaviour in compression
The elastic behaviour and constants di scussed 111
secti on 2.2 appl y equa ll y to tensil e and compressive
loading. There are however differences in the observed
behaviour during pl as tic deformati on and failure.
21
Fundamentals
Test machine
platens
Frictional
restraint
forces
Fig. 2.11 Non-uni(or111 plastic de(onnation in
a COIIIfJression test.
2.6.1 PLASTIC DEFORMATION OF
DUCTI LE MATERIALS
Va lues of yield stresses for ducti le materi als are
similar to those in tension, but the subsequent
behaviour in a laboratory test is influenced by the
loading system. Test machines apply the load
through large blocks of steel ca ll ed platens, whi ch
bea r on the specimen. These are stiffer than the
specimen and therefore the latera l expansion of
the specimen is opposed by friction at the platen/
specimen interface. This ca uses a confining force
or restraint at either end of the specimen. The effect
of thi s fo rce reduces with dist ance from the platen
i. e. towa rds the centre of the specimen, with the
result rhar a cylindri cal specimen of a duct il e mater-
ial of say, mild steel will plastically deform into a
barrel shape, and the sides will not stay strai ght, as
in Fig. 2. 11 .
Continued loading of ductil e mater ials to hi gher
and hi gher stress will simpl y resu lt in a flatt er and
fl atter disc i. e. more and more pl as ti c deforma ti on,
bur no failure in rhe sense of cracking or breakdown
of st ructure. In fact the area is increasing, and there-
fore very hi gh loads are req uired to keep the true
stress (see secti on 2.5) increasing. Tests ca n therefore
easil y reac h t he ca pacity of the test machine.
2.6.2 FAILURE OF BRIDLE MATERIALS
Fa ilure stresses of brittle materia ls in compression
are much hi gher than those in tension- up to twenty
times higher for some materials, e.g. concrete. This
results from a very different cracking and failure
mechani sm. Cracking is a pu lling apart of two sur-
faces, and therefore occurs by the acti on of a tensil e
strain . In uni-axial compressive loading, the strains
22
l
t
Compressive
stress and strain
Lateral tensi le
c::=:> strain
Stable crack
development and
growth
Fig. 2.12 Multiple crack pall em in a bril! le
in compression leading to higher strength than in
tension.
in the direction of loading are obviously compres-
sive and it is the lateral strains that are tensil e
(Fig. 2. 12). The cracks are formed perpendicular to
rhese strains, i.e. parall el to the load direction. A
single small crack will not immediately grow ro
cause fai lure, :111d a whole network of cracks needs
to be formed, grow and intersect before complete
mater ial brea kdown occurs. Thi s requires a much
hi gher stress than that necessa ry to ca use t he si ngle
fa ilure crack under tensile loading.
There is a further effect resu lting from rhe fric -
tion restraint of the platens discussed above that
ca uses the fa ilure stress (i. e. the apparent compres-
sive strength ) to be dependent on the specimen
geometry, pecifi ca ll y the height/width rati o. In the
p:n t of the specimen nea r the platen, thi s restraint
opposes and reduces rhe latera l tensil e st rain. Thi s
increases the load req uired for compl ete brea k-
cl own, i. e. failure (in effect, thi s part of the specimen
is under a tri-axial compress ive st ress system).
The effect of the restraint reduces with di stance
from the platen (fig. 2.13). Short fat specimens
wi ll have most of their volume exper iencing
hi gh restraint, whereas the central part of longer,
thinner specimens will be nearer to a uni -axial stress
system, and will therefore fail at a lower average
appli ed st ress.
The typical effect of t he height/width ratio is
shown in Fig. 2. 14 from tests on concrete; the
strength (i.e. the failure st ress) expressed relative to
that at a height/width rati o of 2. We will di scuss
measurement of the compressive strength of concrete
in more detail in Chapter 21.
l
t
~ } . . Reg1on of h1gh
~ restraint from
platen friction
}
Region of lower
::l restraint from platen
friction
Fig. 2.13 Variation of restraint from platen friction in
a compression test leading t o the size effect on
compressive fai lure stress.
1.8
1.6
~
0:0
c
1.4
2:'
u;
<!>
. ~
1.2
;o
Width
a;
a::
1
0.8
0 1 2 3 4
Height: width rat io
Fig. 2.14 The effect of height/width ralio on !he
compressive strength of brittle materials.
2. 7 Behaviour under constant
load - creep
Constant load or stress is a very common occur-
rence, e.g. the stress due to the self-weight of a
struct ure. Materia ls respond to thi s stress by an
i m mediate strain defor mation, norma II y elastic,
foll owed by a n increase in strain with time, ca ll ed
creep. Typica l behaviour is ill ustrated in Fig. 2.15.
A stress appl ied at time t
1
and ma inta ined a t a
const ant level unti l remova l a t time t
2
res ults in:
an initi al elasti c strain on stress application (related
to the stress by the modu lus of elastici t y)
an increase in t hi s stra in due to creep during the
peri od of consta nt stress- fairl y rapid at first but
then at a decreasing rate
Mechanical properties of solids
Elasti c recovery
__ _._....; C_reep recovery
)-o<:::...__L__ _ _ _____ _:_ ____ __. Time
Initial elastic
strain
Fig. 2.15 Schematic of creep behaviour due to
a constant applied st ress.
a n immediate elastic recovery on stress remova l,
oft en simi lar in magnitude to t he initi a l elast ic
stra in
furth er recovery with time (ca ll ed creep recovery )
again at decreasing rate. This is norma ll y less
than t he creep strain, so t hat t he materi a l does
not return to zero positi on, i. e. there IS some
permanent deformation.
For ca lcu lation purposes, we defin e:
creep coefficient
=creep strain/initia l elast ic strain
effective elastic modulus
= stress/( tota l strain )
(2.18 )
= stress/(elast ic +creep strai n) (2 .19 )
Bot h o f these will obviously vary wit h time.
Creep increases with time, with the appl ied stress
(sometimes values of the s p e c i ~ c creep = creep/unit
stress a re quoted ) a nd with temperature. The mag-
nitude of creep var ies widely in different materia ls.
For exa mple, most meta ls and meta ll ic a ll oys onl y
start to creep at temperatures approachi ng ha lf of
their melting point (expressed in degrees Kelvin ),
whereas with concrete and many pol ymers the creep
strain can be as grea t or greater than the initia l
elast ic stra in at roo m temperature.
Creep curves typica ll y have three parts, illustrated
in Fig. 2. 16 for two different leve ls of stress :
primary creep: init ia ll y rapid, but at a red uc-
ing rate with t ime
secondary creep: a steady rate of stra in (d/dt )
often expressed as
d1dt = Ca" (2.20)
where 0 = applied stress, C is
a constant and n is the creep
23
Fundamentals
c
- ~
U5
Primary
creep
Secondary
creep
Failure, creep
rupture
""" Tertiary
creep
Time
Fig. 2.16 Sub-divisions of creep curves.
tertiary creep:
exponent, wh ich usuall y lies
between 3 and 8
at hi gh stress levels, afte r a
peri od of time (whi ch can be
very length y) there may be an
increasing rate wit h time lead-
ing to fa ilure, a process known
as creep rupture. Thi s onl y
occ urs if the stress is hi gh,
t ypica ll y more t han 70-80%
of the failure stress meas ured
in a short -term test.
In some situati ons, the strain is constant e.g. a
ca ble stretched betwee n two fi xed supports or a
tensioned bolt cl amping two metal plates together.
The stress reduces with time, as shown in Fig. 2. J 7,
a process known as stress relaxation. In extreme
cases, the stress reduces to ze ro, i. e. the ca bl e or
bolt become slack.
During creep and the stress relaxa tion the mater-
ial is, 111 effect, fl owing, albeit at a ver y slow rate.
c
t
I
"
U5
t,
Time
en
en
Stress rel axation
~
U5
t,
Time
Fig. 2.1 7 Schematic of stress relaxation at constant
st rain.
24
It therefore appea rs to be behav ing somewhat li ke
a liquid. Such mi xed solid/li quid behaviour is ca ll ed
viscoelas ti cit y; we will be discussing thi s more detail
in Chapter 5.
2.8 Behaviour under cyclic
loading - fatigue
2.8. 1 FATIGUE LIFE AND S/ N CURVES
Cycli c loading is ve ry common, e. g. wind and wave
load ing, vehi cle loading on roads and bridges. We
can defi ne the characteri stics of the loading as shown
in Fig. 2. 18, in whi ch:
p = peri od of loading
fr equency = lip (e.g. in cycles/sec or Hertz)
crlll.IX = maximum appli ed stress
<Jill"' = minimum appli ed stress
<Jill = mea n stress = (<Jill." + cr"' "J/2
S = stress range = (<J ill." - <J """)
Repeated cycli c loading to st ress levels where <Jill"
is less that the ultimate (or even t he yield stress )
ca n lead to fa ilu re (think of bending a wire back-
wa rds and forwards - the fir st bending does not
break the wire, but several more wi ll ). This is ca ll ed
fatigue failure.
2
Thi s ca n be sudden and br ittl e, and
ca n occ ur aft er many yea rs of sati sfa ctory se rvice.
It is therefore potent iall y very dangerous. Fatigue life
is defin ed as the number of cycl es (N) to fai lure. It is
nor the ti me to fa ilure, al though thi s can of course
be calcul ated if the fr equency of loading is known.
I ~
p
~ I
s
Time
Fig. 2.18 Characteristics of cyclic /oadi11g.
2
Bur do nor make t he common mi st ake of ca lling t hi s
'meta l fati gue'. All st ructural mat eri al s, nor jusr met al s,
are subj ect ro fati gue fa ilure under appropriat e combin-
ati ons of st ress range and rime.
From res ring, it has been found that:
N is independent of frequency (except a r very
hi gh frequencies, above 1 kHz)
N is dependent on t he stress ra nge (5) rather than
the individual va lues o f 0
111
.\X o r 0
11
,
11
, provided 0
111
_, ,
o r 0
11
, ;
11
does nor approach t he yie ld strength . For
exa mple, t he fati gue li fe under a st ress cycling
between -50 a nd +50 MPa is the sa me as t hat
under stress cycling between +25 and + 125 MPa
a hi gher stress range res ults in a shorter fat igue
li fe, i.e. N increases wit h decreasing S. The relation-
s hip is often o f the form
S.N' = C (2.21 )
where a and C are constants (a is between 0.12
a nd 0.07 for most materi a ls).
The fatigue perfo rmance of materi a ls is norma ll y
given as SIN curves i.e. graphs of stress range (5)
vs. fat igue li fe (N). Typica l SIN curves for mi ld steel
and a copper a ll oy a re shown in Fig. 2.19. Fatigue
li ves a re o ft en very long (e.g. tho usa nds, tens o f
t ho usa nds, or hundreds of thousa nds of cycles) so
a log sca le is norma ll y used for N.
The indi vidua l da ta points shown for mi ld steel
illustra te the considerable scatter that is obtained
from res t progra mmes . Apart fr om t he obvious su-
perior performance of steel, t he best-fit line t hrough
the data shows a disconti nui ty ar about 240 MPal10
7
cycl es where it becomes parallel to the x-axis for
higher fa ti gue lives. This mea ns that at va lues of S
below 240 MPa t he fatigue life is infinit e, which is
very useful for design purposes . T his stress range is
ca ll ed the fati gue endurance limit, a nd is a t ypica l
cha racteri sti c of ferrous meta ls. Non-ferrous met a ls
500
Mild steel
-ro
400
0..

VJ
300
Ql
_ __ _________ _
Ol
c

200
en
en

<n
100
Non-ferrous metal
e.g. copper al loy
10
4
10
5
10
6
10
7
Cycles to fai lure - N
Pig. 2.19 Fatigue life data (SIN curves) for mild steel
and a copper alloy.
Mechanical properties of solids
such as copper do no t show such a limit i.e. there
is no 'sa fe' stress ra nge below which fatigue failure
will not occ ur eve ntua ll y.
2.8.2 CUMULATIVE FATIGUE DAMAGE:
MINER'S RULE
SIN curves define t he fat igue li fe a t any given st ress
ra ngeS. Ho wever, in most situa ti o ns in practi ce the
ma teri a l o r structura l component will no t be sub-
jected to cyc les of a single stress ra nge, but t o dif-
fer ing numbers of cyc les o f d ifferent stress ra nges.
For exa mpl e wind acti on will res ult in a few cycles
at a hi gh stress ra nge from severe storms a nd a
la rge number of cyc les at lower stresses from lesser
strength winds . To est imate t he effect o f this cumu-
lat ive fatigue da mage Miner's rule is used. Thi s
accounts for the parti a l effect of the number of cycles
a t eac h pa rt icu la r stress range by considering tha t,
if t he ma teri a l is stressed for n
1
cycl es at a stress
range that wi ll ca use fa il ure in a t o tal o f N
1
cycles,
then a fraction n/N
1
of the fa t igue life is used up;
fa ilure occurs when the sum of a ll the fracti o ns,
In,I N
1
, reaches 1, irres pect ive of the sequence of
applicati o n of t he va rious cycles o f load ing.
Figure 2.20 illustrates the case o f three stress ranges,
5
1
, 5
2
a nd S
1
, being appl ied fo r n
1
, n
2
a nd n
3
cyc les
res pecti ve ly, a nd fo r whi ch t he tota l fa tigue lives
are N
1
, N
2
and N
1
T he n
1
cycles a t the stress range
5
1
use up n/N
1
o f t he tota l fati gue life. T he same
a ppli es for the n
2
cyc les ar 5
2
a nd the n
3
cycles a t
S
1
. Therefore the pro porti o n o f the total fatigue life
used up by a ll three stress ra nges is:
n/N
1
+ nzf N
2
+ n;IN
1
Applied
loads
I
s,
I
I
G)
Ql 52 -- . -+- : -----,.-------'"'-.
Ol I
C I
: :
53 - - -:- - - -- -
I I
en : :
I
I
I
I
'
I I
I 1 I
+ +
n
3
N, N
2
Number of cycles (N)
Fig. 2.20 Example of the application of Miner's rule of
cumulative fatigue damage.
25
Fundament als
Failure will occur when thi s sum reac hes I. T hi s
will be achi eved wit h cont inued cyc li c loading,
in whi c h o ne o r a ll of n
1
, n
2
o r n
1
may increase,
depend ing o n t he na ture o f the loading. T he general
ex pressiOn of Miner's ru le is th a t fo r fa ilure:
(2.22)
2.9 Impact loading
Str uctu res a nd components of struct ures can be
subjected to ve ry rapid rates of appli cat ion of
stress a nd stra in in a number of circ umsta nces,
such as explosions, mi ssil e o r vehi cle impact, a nd
wave sla m. Ma teri a ls ca n respond to such impact
loading by:
a n apparent inc rease in elas ti c mod ulus, bur thi s
is a third or fou rt h o rder effect onl y- a 10
4
rimes
increase in loadi ng ra te gives onl y a 10 % inc rease
in elasti c mod ulus
a n increase in br ittl e behavio ur, lead ing to fas t
brittl e frac t ure in nor ma ll y ducti le ma ter ia ls. Thi s
ca n be very da ngero us - we thin k we a re using
a d uct ile mater ial tha t has a hi gh work to fracture
a nd gives wa rning of fai lure, but thi s react s to
impact load ing like a brittl e mate ria l. T he effect
is enha nced if the materi al conta ins a pre-existing
defect s uch as a crack.
T he lat ter effect ca nnot be predi cted by extra po la t-
ing the resu lts of la bo ra to ry tensil e or compression
tests such a t hose descr ibed ea rli er, a nd impact tes t
proced ures have been deve lo ped to assess t he be-
havio ur of specimens conta ining a machined notch,
which acts a loca l stress raiser. T he Cha rpy test fo r
meta ls is a good example of such a tes t . In thi s, a
heavy pendu lum is released a nd strikes t he sta nda rd
specimen a t the bottom o f its swing (Fig. 2.2 1 ). The
spec imen brea ks a nd t he energy needed fo r the frac-
ture is dete rmined fro m t he difference between the
starting a nd fo ll ow- thro ugh pos iti o ns o f the pendu-
lum. The energy a bsor bed in the fracture is ca lled
the Cha rpy impact va lue. As we have di sc ussed
ea rli er in the chapter, br ittl e ma ter ia ls req uire less
energy fo r fa ilure tha n ductil e materia ls, a nd a n
impact va lue of 15J is no rma ll y used as a somewhat
a rbitrary di vision bet ween the t wo, i.e. br itt le mater-
ia ls have a va lue below t his, a nd ducti le mater ia ls
a bove.
An exa mple o f the use o f the rest is in determin ing
t he effect o f temperature on duct il e/brittl e behavio ur.
Ma ny ma ter ia ls that a re ducti le a t no rma l tempe ra-
tures have a tendency to br it tl eness a t red ucing
26
Fig. 2.2 1 Charpy impact test specimen ((rum
dimensions specified in BS eN ISO 148-3 :2008).
5
>-
Ol
Qj
c
Q)
t5
<1l
a.
- ~
>-
a.
(U
.r:::
0
150
Transit ion
temperature
Brittle
4
100
50
15 -- - - -- ---
I
o + - - - ~ - - ~ ~ T ~ ~ - - ~ - - - - ~ - - ~
- 75 - 50 - 25 0 25 50 75
Temperature (deg C)
Fig. 2.22 Varia tion of the Char{J)' impact energy of
a steel with temperature (afi er Rollason, / 9(, 1 ).
temperatures. This effect fo r a part ic ul a r steel is
shown in Fig. 2.22. The decrease in ductilit y with
fa lling temperatu re is ra pi d, with the 15J division
occurring at about -20C, whi ch is ca lled rhe transi-
tion temperature. It would, fo r exampl e, mean tha t
th is steel should not be used in such st ructu res as
oil producti on insta llations in Arcti c conditi ons.
Impact behaviour and fast fracture a re an import -
ant pa rt o f the subj ect ca lled fract ure mechanics,
whi ch seeks to describe a nd predi ct how a nd why
cracking and fracture occur. We will consider thi s
in more derail in Cha pter 4.
2.10 Variability, characteristic
strength and the Weibull
distribution
Enginee rs are continuall y faced with uncertaint y.
Thi s may be in the estimati on of the loading on a
structure (e .g. what is the design load due to a hur-
ri cane that has a small but finit e chance of occurring
sometime in the next 100 yea rs?), analys is (e.g. what
ass umpti ons have been made in the computer model-
ling and are they val id ?) or with the constructi on
materi als themsel ves . When dealing with uncertainty
in materi als, with natural materi als such as timber
we have to cope wit h nature's own va ri ati ons, which
ca n be large. With manufactured materia ls, no matter
how well and carefull y the producti on process is
cont roll ed, they all ha ve some inherent var iability
and are therefore not uni fo rm. Furthermore, when
ca rrying out tests on a set of sa mpl es to assess thi s
va ri ability there wi ll also be some una voidabl e
va ri ati on in the testing procedure itself, no matter
how carefull y the test is carried out or how skilful
the operati ve. Clea rl y there must be procedures to
dea l with thi s uncertainty and to ensure a satisfac-
tory balance between sa fety and economy. Structural
failure ca n lead to loss of li fe, but the constructi on
costs must be acceptabl e.
In thi s secti on, after some basic stati sti ca l consid-
erati ons for describing variabi lity, we will di scuss
two approaches to coping with variati ons of strength
- characteristic strength and the Weibull di stribu-
ti on. We will take strength as being the ultimate or
failure stress of a material as measured in, say, a
tension, compression or bending test (a lthough the
arguments appl y equall y to other properti es such
as the yield or proof stress of a materi al).
2. 10. 1 DESCRIPTIONS OF VARIABILITY
A seri es of tests on nominall y identi ca l specimens
fr om either the sa me or successive batches of mater-
ial usuall y gives va lues of strength that are equall y
spread about the mean value wi th a normal or
Ga ussian distr ibuti on, as shown in Fig. 2.23.
The mea n va I ue, 0 "" is cl efi ned as the a ri th me tic
ave rage of all the res ults, i.e.:
<J"' = (I cr )/n
(2.23 )
where n is the number of results.
The degree of spread or vari at ion about the mea n
IS given by the standard deviation, s, where
s
2
= I (cr - a"/l(n- 1) (2.24)
s
2
is called the vari ance, and s has the same units as cr.
c
'iii
c
Q)
"0
Q)
(.)
c
~
::::J
(.)
(.)
0
> - ~ -
"" 0 0
i5
ro
.0
e
(l_
>-
(.)
c
Q)
::::J
o-
~
LL
Mechanical properties of soli ds
Mean: am
Failure stress (a)
Fig. 2.23 Typi cal normal distribution of fai lure stresses
(rom successive tests on samples of a construction material.
c
'iii
c
Q)
"0
.
i5
ro
.0
0
ct
Material 2
high strength, low
variability and
standard variation
Failure stress (a)
Fig. 2.24 Two combinations of mean strength and
variabilit y.
Materia ls ca n have any combinati on of mea n
strength and va ri abilit y (or standa rd deviati on)
(Fig. 2.24).
For compari son between mater ials, the coeffi cient
of variation, c, is used, where
c = s/cr"'
(2. 25 )
c is non-dimensional, and is normall y expressed as
a percentage. Typi ca l va lues of care 2% for steel,
which is produced under carefull y controll ed condi-
ti ons, 10-15% for concrete, whi ch is a combination
of different components of different particl e sizes,
and 20- 30% fo r timber, whi ch has nature's own
variat ions. Steel and timber are at the two ends of
the va riabilit y sca le of construct ion mater ials.
For structural use of a materi al, we need a 'safe'
stress that takes into account of both the mea n
failure stress and the variability. Thi s is cl one by
considering the normal di stributi on curve (Fig. 2.23)
in more detail. The equati on of the curve is
1 [ (cr - cr,,i ]
y = - -exp
s fiit 2s
2
(2.26)
27
Fun damen tals
.c
"U)
c
<!)
""0
.
:.0
ell
.0
0
a:
- 3s
2.1 %
- $
0 .1%
s 2s
Failure stress (cr)
Fig. 2.25 Proportion of results in the regions of the
normal distribution curve.
Some impo rt a nt properties of thi s eq ua ti o n a re:
The c urve encloses the whole population of da ta,
a nd therefore not surpri singly, integrat ing the
a bove eq uation between the limits of -oo a nd
+co gives a n answer o f 1, o r 100 if the pro babi li ry
density is expressed as a percent age.
50% of the res ult s fa ll below the mea n a nd 50%
above, but a lso, as shown in Fig. 2.25:
68. 1% of results li e within o ne sta ndard
devia ti o n of the mea n
95 .5% of results li e within two st a ndard
deviati o ns o f the mea n
99.8% of res ults li e within three sta ndard
devia ti ons of the mea n.
2.10.2 CHARACTERISTIC STRENGTH
A gua ra nteed minimum va lue of stress below which
no sa mpl e will ever fai l is impossibl e to define- the
na ture of the no rma l di stributi o n curve mea ns tha t
there wi ll a lwa ys be a chance, a lbeit very small, o f
a failure below a ny stress va lue. A va lue o f stress
ca li ed the characteristic strength is therefo re used,
which is defined as the stress be low which a n ac-
ceptably sma ll number of results will fa ll. Enginee r-
ing judgement is used to define 'acceptabl y sma ll '.
If thi s is ve ry small , then there is a ve ry low ri sk of
failure, but the low stress wil l lea d to increased
cross-sectional a rea a nd hence grea ter cost . If it is
hi gher, then the structure ma y be c hea per but there
is an increased ri sk of fa ilure.
Clea rl y a bal a nce is therefore required between
safety a nd economy. For ma ny ma ter ial s a stress
below which 1 in 20 of the res ult s occurs is con-
sidered acceptabl e, i.e. there is a 5 % fa ilu re ra te.
Analysis of the normal di stributi on curve shows
tha t thi s stress is 1.64 sta ndard devia ti ons below
28
5% of
result s
Margin
Ochar Om
Failure stress (cr)
Fig. 2.26 Definiti on of characteristic strength (a,.
1
._,)
and margin for a 1 in 20 (5%) fai lure rate crit eri on.
the mea n. This distance is ca ll ed the margin and so,
as shown in Fig. 2.26:
cha racteri stic st rength = mea n strength - ma rgin
(2.27)
where k, the standard deviati on multipli cati on factor,
is 1.64 in thi s case.
The value of k varies according to the c hosen
failure rate (Table 2.1), and, as we sa id above, judge-
ment and consensus are used to arrive at a n acceptable
failure rate. In practice, thi s is not a lways the sa me
in all circumstances; for exa mpl e, .5 % is t ypi ca l
for concrete (i. e. k = 1. 64 ), and 2% for timbe r (i. e.
k = 1.96 ).
There is a further step in determining a n a ll ow-
able stress for design purposes. The strength data
used to determine the mea n and sta nda rd dev ia tion
for the above analys is will norma ll y ha ve been
obta ined from laboratory tests on small specimens,
which genera ll y will have no apparent defects or
damage. They therefo re represent the best that can
be expected fr om the materi al in idea l o r near idea l
circumstances. In practice, str uctu ra I e Iemen ts and
members contai n a large volume o f materi al, which
Table 2.1 Va lues of k, the standard deviati o n
multiplicati on factor, for various failure rates
Failure rate (%)
50
16
10
5
2
k
0
1
1.28
1.64
1.96
2.33
has a grea ter chance of containing man ufacturing
and handli ng defects. Thi s size effect is taken into
account by reducing the characteristic strength by
a partial materials' safety factor, Ym
It i nor mal practi ce for Ym to be given as a va lue
grea ter than one, so the characteristic strength has
to be divided by y"' to give the all owable stress.
Hence:
all owable design stress
= characteristi c strength/y"'
= (mea n strength - margin )/y"' (2.28)
As with the failure rate, the va lue of Ym is based on
knowledge and experi ence of the performa nce of
th e mater ial in practi ce. For exa mple, typical values
recommended in the Europea n standard for struc-
tural concrete des ign (Eurocode 2, BS EN 1992) are
1.15 for reinforcing steel and 1. 6 for concrete.
2. 10.3 THE WEIBULL DISTRIBUTION
An alternati ve stati sti ca l approach to the di stribu-
tion of strength, particularly for brittl e materials,
was developed by the Swedi sh engineer Waloddi
Weibul l. As we have di scussed in secti on 2.8 (and
wil l consider further in Chapter 4) brittl e fracture
is initiated at flaws or defects, which are present to
a grea ter or lesser extent in all material s. Therefore
the variations of strength can be attributed to vari -
ati ons in the number and, more parti cularl y, the
maximum size of defect in a test specimen. Larger
specimens have a hi gher probability of containing
large r defects and therefore ca n be expected to have
a lower mea n strength (as just discussed in relati on
to the partia l materia ls safet y factor, YnJ
Wei bull defined the surviva l probability, PJ V
0
),
as the fraction of identi cal sampl es of volume V
0
that sur vive aft er appli cation of a stress cr. He then
proposed that:
(2.29)
where 0
0
and m arc constants. Pl ots of thi s equation
for three va lues of m are shown in Fig. 2.27. In
each case, when the stress is low, no specimens
fail and so the su rvival proba bility is l, but at
increasing stress more and more sampl es fail until
eventuall y, when they have all failed, the surviva l
probabilit y is zero. Putting a = a
0
in equat ion (2.29)
gives P,( V
0
) = 1/e = 0.37, so 0
0
is the stress at whi ch
37% of the sampl es survive. The va lue of m, which
is called the Weibull modulus, is a measure of the
behaviour on either side of cr
0
, and therefore indi-
0.8
-0.6
~
Q . ~ 0.4
0.2
1/ e = 0.37
/
Mechanical properties of solids
m = 5 10 100
cr
0
Stress ( cr)
Fig. 2.27 The Wleibu/1 distribution (or three values of
the Wleibu /1 modulus, m.
ca tes the degree of var iability of the res ults (a nd in
this se nse has a similar role to the coeffici ent of
va ri ati on as defi ned in equati on 2.25). Lower va lues
indi cate greater variability; m for concrete and
bricks is t ypica ll y about 10, whereas for steel it is
about 100.
We ca n extend thi s analysis to give an estimate
of the volume dependence of surviva l probability.
P,( V
0
) is the probability that one specimen of volume
V
0
will survive a stress cr. If we test a batch of n
such specimens, then the probability that they will
all survive thi s stress is lf,( V
0
) ) " . If we then test a
volume V = nV
0
of the materi al, which is the equi va -
lent of combining all the small er specimens into a
single large specimen, then the surviva l probability,
P,(V), is still (P_( V
0
) ) " .
P,(V) = [P,(V
0
) ]" = [f,(V
0
)j'
1
v" (from eqn (2. 29))
= (exp(- (cr/cr
0
)}"')"
1
v"
which gives
(2.30)
So, hav ing determined cr
0
and m fr om tests on
sa mpl es of volume V
0
and selected an acceptabl e
va lue for P,( V), the design stress for structural element
of volume V ca n be ca lcul ated.
References
Case .L Chil ve r H a nd Ross C ( 1999). Strength of Mater-
ials and Structures, Elsevier Science & Techno logy,
Lo ndo n, p. 720.
Ro ll ason EC ("196 1 ). Metallurgy for Engineers, Edward
Arno ld, Lo ndon.
29
Chapter 3
The structure of
sol ids
In Chapter J we discussed the va ri ous ways in whi ch
atoms bond together to form soli ds, liquids and
gases, and some of the principles involved in the
changes between these states . In Chapter 2 we
desc ribed the behaviour of solids when subj ected
to load or stress and the var ious rules and con-
stants used to characteri se and quantify thi s. We
now go on to consider the structu re of solids in
more derail , whi ch will provide an explanat ion
fo r much of rhe behaviour described in Chapter 2.
Although the type of bonding between atoms goes
some wa y towa rds ex pl aining the properti es of
the resulting elements or compounds, it is equal ly
important to understand the ways in whi ch the
atoms are arranged or packed together. We start
by considering the relatively ordered structure of
crystalline solids, and then di scuss some as pects
of the less ordered structures of ce rami cs and
polymers.
3.1 Crystal structure
Ma ny construction materials, particul arl y metals and
some cerami cs, consist of small crystals or grains
within whi ch the atoms are packed in regul ar,
repeating, three-dimensional patterns giving a long-
range order. The grains are 'glued' together ar the
gra in boundari es; we wil l consider the importance
of these later, but first we will di scuss the possibl e
arrangements of atoms within the grains. For thi s,
we will ass ume that atoms are hard spheres - a
consi derable but conveni ent simplificati on.
It is also conveni ent to start with the atomic
st ructure of elements (whi ch of course consist of
single-sized atoms) that have non-directi onal bond-
ing (e.g. pure metals with metalli c bonds). The
simpl est structure is one in which the atoms adopt
a cubi c pattern i. e. each atom is held at t he corner
of a cube. For obvi ous reasons, thi s is ca ll ed the
simple cubic (SC) structure. The atoms touch at
30
(a) Hard sphere model (b) Reduced sphere model
Fig. 3. 1 The simple wbic stmcture of ato111s in crystals
and the unit cell.
Fig. 3.2 The coordination number = 6 (or the simple
cu/;ic st ructure.
the centre of each edge of the cube (Fig. J. la ). The
structure is sometimes more conven ientl y shown as
in Fig. J .lb. We ca n use thi s fi gure to define some
properti es of crystalline structures:
the unit cell: the small est repeating unit within the
structure, in this case a cube (Fig. 3. 1)
the coordination number: the number of atoms
touching a parti cul ar atom or the numbers of its
nea rest neighbours, in thi s case 6 (Fig. 3.2)
the closed-packed direction: the directi on along
whi ch atoms are in continuous cont act, in thi s
case any of the sides on the unit cell
Fig. 3.3 Unit cell dimensions and at omic radii in the
simple cuhic structure.
Fig. 3.4 The body-centred wbic structure, unit cell and
close-packed direction.
the atomic packing fact or (A PF): t he volume of
atoms in the unit cell /volume of the unit cell ,
whi ch therefore represents the effi ciency of packing
of the atoms.
The APF ca n be calculated from simpl e geometry.
In thi s case:
there are eight corner atoms, and each is shared
between eight adjoining cell s
. . each unit cell contains 8 x 1/8 = 1 atom
the atoms are touching along the sides of the
cube {the close- packed directi on)
radius of each atom, r = 0.5a (F'ig. 3.3)
when a = length of the side of the unit cube
the volume of each atom = 4/3n:r
1
= 4/3n: (0.5a)
3
. . APF = [atoms/cell].[ volume each atom)/
volume of unit cell
= Ill x [ (4/3n: (0.5a)
1
l/ l a
3
] = 0.52.
There are two other crystal structures that have
cubi c structures with atoms located at the eight
corners bur whi ch have additional atoms:
the body- centred cubic (BCC) structure, which also
has an atom at the centre of the cube (Fig. 3.4)
the face-centred cubic (FCC) structure, which
also an atom at the centre of each of the six
faces (Fig. 3.5) .
With the body-centred cubic structure, the coord in-
ati on number is 8 {the atom in the cell centre touches
The structure of soli ds
Fig. 3.5 The (ace-centred cubic structure, unit cell and
close-packed direction.
the eight corner atoms) and the close-packed di recti on
is the cell di agona l. It should be apparent fr om
Fig. 3.4 that:
each unit cell cont ains 8/8 + 1 = 2 atoms
consider ing the close- packed directi on gives:
4r = -.J 3a or r = -.J 3a/4.
APF = [2] x [(4/3n: (-.J 3a/4)' ]/ [a l ] = 0.68.
With the face-centred cubi c structure, a little thought
should convince you that the coordinati on number
is 12 and the close- packed directi on is the face
di agonal. From Fig. 3. 5:
each unit cell contains 8/8 + 612 = 4 atoms
considering the close-packed directi on gives:
4r = -.J2a or r = -.J 2a/4.
. . APF = [4] x [(4/3n:( -.J 2a/4)
1
]/[a
1
j = 0. 74.
Moving from the SC to the BCC to the FCC struc-
ture therefore gives an increase in the coordinati on
number (from 6 to 8 to 12) and in the effici ency of
packing (from an APF of 0.52 to 0.68 to 0. 74).
One further struct ure that mi ght be expected to
have effi cient packing needs considerati on - the
hexagonal close-packed (HCP) structure. If we start
with a single plane, then the most effici ent packing
is a hexagonal layout, i.e. as the atoms labell ed A
in Fig. 3.6. In adding a second laye r the atoms
{labell ed B) place themselves in the holl ows in the
first layer. There are then two possibl e pos iti ons for
the atoms in the third layer, either directl y above
the A atoms or in the positions labell ed C. The first
of these opti ons gives the structure and unit cell
shown in Fig. 3. 7. Two dimensions, a and c, are
requ ired to defin e the unit cell , wit h cia = 1.633.
The coordination number is 12 and atomi c packing
factor 0. 74, i. e. the same as for the face-centred
cubic st ructu re.
If we know the crystal structure and the atomic
weight and size of an element then we ca n make
an estimate of its density. For exa mple, take copper,
wh ich has a face-centred cubi c structure, an atomi c
31
Fundamentals
Fig. 3.6 Arrangement of atoms in successive layers of
the hexagonal close-packed structure.

Fig. 3. 7 Atomic arrmrgement and unit cell of the
hexagonal close-packed structure.
c
weight of 63.5, and an atomic radius of 0. J 28 nm
(atomic weights and sizes are readil y avai lable from
tabl es of properti es of elements).
atomi c weight = 63 .5, therefore 63.5 g
1
of cop-
per contain 6.023 x lOLl atoms
2
.. mass of one atom = 10. 5 x 10-
21
g
In the FCC structure there are 4 atoms/ unit cell
. . mass of unit ceJJ = 4.22 X 1 o-n g
1
The atomi c weight in grams is normall y ca ll ed the molar
mass, and is the weight of one mole.
2
This is Avogadro's number, which we defined in Chapter 1
as the number of atoms in one mole. By definition it has
the sa me va lue fo r all elements.
32
As above, in the FCC structu re, length of side
of unit cell (a) = 4rl 2
a = 4 x 0. 128/ 2 = 0.362 nm
unit cell volume= 4. 74 x 10-
2
nm
1
density = weight/volume
= 4.22 X 10-
22
g/4.74 X I 0
2
llnl
1
= 8900 kg/rn
1
A typical measured value of the density of copper is
8940 kg/m \ so our estimate is close.
We genera ll y expect that elements that adopt one
of the crystal structures descri bed above will prefer
to adopt rhe one that has the lowest internal energy.
The efficiency of packing (i.e. the APF) is an import-
ant, but not the sole, factor in this . ln practice, no
common metals adopt t he simpl e cubi c structure,
but the energy difference bet ween the other three
structures is often small , and the structures adopted
by some common metals are:
FCC - aluminium, copper, nicke l, iron (;1bove
9l 0C), lead, si lver, gold
HCP - magnes ium, zinc, cadmium, coba lt,
titanium
BCC - iron (below 9l0C), chromium, mol yb-
denum, ni obium, va nadium.
The two structures for iron show that the crystal
st ructure ca n have different minimum energies ar
different temperatures. These var ious forms are
call ed allotropes . Changes from one structure to
another brought about by changes of temperat ure
are of fundament al importa nce to metal lurgica l
practice. For exa mpl e, as we wi ll see in Chapter I I,
the change from FCC to BCC as the
of iron is reduced through 91 0C forms the founda -
ti on of the metal lurgy of steel.
3.2 Imperfections and impurities
In practi ce it is impossibl e for a perfect and uniform
atomic structure to be formed throughout the material
and there will always be a number of imperfections.
Point defects occur at di screte sites in the atomic
latt ice and ca n be either mi ssing or ex tra atoms,
ca ll ed vacancies and interstitial atoms respectively,
as shown in Fig. 3.8. A linea r dislocati on is a one-
dimensional defect; an exampl e is when part of a plane
of atoms is mi ssing and ca uses an edge dislocation,
as shown in Fig. 3.9. The result of all such defects
is that the surrounding atomi c structure is di storted
and so is not in irs preferred lowest energy state.
Thi s has important consequences during loading;
when the internal strain energy is suffi cient to locall y
rearrange the structure a di sloca ti on is, in effect,
Vacancy Interstitial atom
Fig. 3.8 Vacancy and int erstitial defects in a crystal
latt ice.
Fig. 3.9 Edge dis location.
moved. The di slocation does not move back to its
or iginal positi on on unl oading, and so the resulting
deformation is irreversibl e i.e. it is plastic, as shown
in Fig. 2.6. If the required internal energy needed
to tri gger the dislocati on movement is sharpl y
defin ed then thi s gives ri se to a di stinct yield point
(point A in Fig. 2. 6) . We will discuss thi s dislocat ion
movement in metals in more detai l in Chapter 8.
It is also impossible to produce a completely pure
materi al, and some foreign atoms will also be present,
thus producing a solid soluti on. A substitutional
impurity occurs when the foreign atoms take the place
of the parent atoms, resulti ng in a substitutional
solid solution (Fig. 3. 1 Oa) . If the atoms of the two
materia ls are of a simil ar size then there will be
littl e di storti on to the atomi c latti ce, but if their size
differs significantl y then some di stortion will occur.
An interstitial imfJUrity, as the name impli es, occurs
when the fore ign atoms are forced between the parent
atoms, resulting in an interstitial solid solution; aga in
the degree of di storti on depends on the relati ve sizes
of the atoms involved (Fig. 3.10b).
The impuriti es may occur by chance during manu-
facture, but nea rl y all metal s used in constructi on
are in fac t alloys, in wh ich controll ed quantiti es of
The structure of solids
(a) (b)
Fig. 3. 10 Dist ortions from (a) a substitutional impurity
and (b) an interstitial impurity.
carefull y selected impuriti es have been deliberately
added to en hance one or more properti es. We wi ll
discuss some important exa mpl es of all oys in Part 2
of thi s book.
3.3 Crystal growth and grain
structure
Crys tals are formed in a cooling liquid. In the liquid
the atoms are in a state of constant moti on and
change positi ons freque ntl y. During cooling thi s
moti on becomes more sluggish until , sooner or later,
the atoms arrange themselves loca ll y into a pattern,
often one of those described above, that forms the
nucl eus of the solid materi al. The kineti cs of nucl e-
ati on are qui te complex, but it almost always begins
from an impurit y particl e in the melt. The nucleus
will have the form of the unit cell , whi ch is often
a cube. As the li qu id solidi fi es it gives up its latent
hea t of solidifi ca ti on. The corners of the cu be lose
hea t faster than the edges so that atoms from the
melt att ach themselves to the corners first, then tO
the edges and las t of all to the faces of the elemen-
tary cube. Thus a branching or dendritic pattern
is bui lt up from each nuclea tion site (fig. 3.11a)
and dendrites wi ll grow fr om each site until they
are stopped by inter ference fro m other dendrites
(Fig. 3.1Jb and c) .
Eventuall y all the liquid is used up by infilling
between the arms of the dendrites and the character-
istic polycrystalline grain structure results (fig. 3.11d).
There are three important fa cts to note here:
1. With in each grain the atoms are arra nged in a
regu lar latti ce, albeit containi ng some or all the
defects and imperfecti ons described above.
2. The ori entati on of the crystal lattice differs from
grain to grain.
3. At eac h grain boundary there is a line of
mi smatch in the atomic arrangement.
33
Fundamentals
(a)
Fig. 3. 11 Schemat ic of dendritic crystal grotuth a11d
result ing grain boundaries.
The size of the individual grains depends on the type
of materi al and more signifi ca ntl y, the cooling rare;
larger ga ins are for med with a slower cooling rare. In
many metals, the grains are large enough to be viewed
with opti ca l mi croscopy, whi ch is extremely useful
to metallurgists. As we shall see in Chapter 8, the grain
size and the grain boundari es have important influ-
ences on the mechani ca l properti es of the metal.
3.4 Ceramics
Most cerami cs are compounds of metalli c and non-
metalli c elements e.g. silica, whi ch is sili con ox ide,
Si0
2
, or alumina, aluminium ox ide, Al
2
0
1
The
atomic bonding ranges from ioni c to cova lent and
indeed, many ce ramics have a combinati on of the
two types (e.g. about half the bonds in sili ca are ioni c
and half covalent ). Covalent bonds in parti cul ar are
hi ghl y directi onal and therefore the structure of
ce rami cs is more compl ex that of the single-element
metalli c solids desc ribed above.
Silicon and oxygen are the two most abundant
elements in the ea rth's crust and so it is not surprising
that sili ca and sili cates are both important and wide-
spread. Sili ca in va ri ous forms is a major component
of many constructi on materi als, including concrete,
aggrega tes, bri cks and glass, and so we will use it
to illustrate the type of structures that cerami cs ca n
adopt.
34
(a) Sil ica tetrahedron, SiO.'-
(b) Sili ca chain
(c) Sili ca ring
Fig. 3. 12 Ionic and st ructural arrangeme11ts of silica,
Si0
2

Sili con is tetravalent and ca n fo rm four equa ll y
spaced cova lent bonds by sharing each of its va lence
electrons with one of the va lence electrons of a
divalent oxygen atom. In the res ulting tetrahedron
each of the four oxygen atoms requires an extra
electron to achi eve a stabl e confi gurati on and there-
fo re thi s is, in effect, an SiO/ - ion (Fig. 3.12a ). Thi s
bas ic unit of sili ca has the abilit y to combine with
other units and with other elements in a wide vari ety
of ways of varying compl exit y, giving ri se to an
enormous number of sili ca- based materials with a
wide range of properti es.
For exa mpl e, if two of the oxygen atoms are
shared with other tetrahedra, then either a chain or
ring structure ca n be formed, as shown in Fig. 3.1 2b
and c, respecti vely, with the overall composition
Si0
1
. If the ring structure has long- range order then
a regul ar, crys talline materi al is obtained, but if it
has a more random, non-ordered structure then an
amorphous, non-crystalline or glassy materi al res ults
(Fig. 3.13 ). In general glassy structures are produced
by rapid cooling from the molten liquid, as a res ult
of whi ch the basic units do not have rime to ali gn
themselves in their preferred ordered state.
The oxygen atoms that are not part of the chain
or ring bonds are availabl e to form ionic or coval ent
bonds with other atoms or atomi c groups. For
exa mpl e, there is a seri es of compounds between
sili ca and va rying rati os of the ox ides of calcium,
magnesium and sodium to give, among others, calcium
sili cates with overall compositi ons of CaSi0
1
, Ca
1
Si 0 ,,
Ca
2
Si0
4
, and Ca
1
Si
2
0
7
, magnes ium silicates such as
talc, Mg
3
Si
4
0
10
(0H)
1
, and as bestos, Mg
1
Si
1
0 ,(0H)
4
,
and sodium sili ca te or water glass ( a
2
Si0
1
) .
The st ructure of soli ds
(a) Ordered, crystalline (b) Amorphous, glassy, non-crystalline
Fig. 3. 13 Two-dimensional views of the forms of silica, SiO
2

The strong, directi onal cova lent bonds give ri se
to the brittl e nature of most ce rami cs, with failure
often initi ated at a defect in the structure. We will
di sc uss the mechani sms of such failure in some
detail when considering the subj ect of fracture mech-
ani cs in the next chapter.
3.5 Polymers
Polymers contain long-chain or string- like molecul es
of very hi gh molecul ar weight. They occur naturall y
in pl anrs and animals or can be synthesised by
polymerisation of the small molecul es in a monomer.
In construction timber and rubber have for many
centuri es been the most widely used naturall y oc-
CLIITing polymers, bur syntheti c polymers such as
pl asti cs, polyester and epoxy resins, and many types
of rubber are of increasing importance.
The backbone of the chain normall y consists of
covalentl y-bonded ca rbon atoms (a lthough some
sili con-based polymers, known coll ectively as silicones,
are made). The monomer mol ecul e typicall y contains
a doubl e bond between carbon atoms, whi ch reduces
to a si ngle bond on polymeri sati on. The monomer
therefore provides the repea t unit in the chain; two
exa mpl es, polyethylene and polyvinyl chl oride (PVC),
are shown in Fig. 3.14. Polymeri sing a mixture of more
that one type of monomer produces a copolymer.
In linear polymers, the repeat units are joined in
single chains, whi ch intertwine like a mass of string,
as illustrated schematicall y in Fig. 3.1 5a. The cova-
lent honds in the chains are strong, bur the bonding
between the chains is due to secondary, Van der
Waa ls bonds (see Chapter 1), whi ch are wea ker but
in many cases suffi cientl y strong for the polymer to
exist as a solid at normal temperatures .
If the chain has side- branches then a branched
polymer is formed (Fig. 3.1 5 b), often with a lower
packing effici ency and hence a lower density than
H H
C = C Q
H H
Monomer
H H
C = C
I I
H Cl
Monomer
.- --- - --
H H: H H :H H H H H H
I I: I I :I I I I I I
- c - c ~ c - c ~ c - c - c - c - c - c -
~ : I I :I I
' '
H H: __ ~ - - ~ - ~ H H H H H H
Repeat unit
(a) Polyethylene
H H:-H- - H -:H H H H H H
I 1: I I :I I I I I I
- c - c ~ c - c ~ c - c - c - c - c - c -
1 1: I I : I I I I I I
H q H Cl :H Cl H Cl H Cl
~ - - - - - - _,
Repeat unit
(b) Polyvinyl chloride
Fig. 3. 14 Monomer and polymer molecules (or two
common polymers.
a linea r pol ymer. It is al so possibl e for the chains
to be linked by other atoms or molecul es that are
cova lentl y bonded to adjacent chains, thus forming
a cross-linked polymer (Fig. 3. 15c) . With suffi cient
cross-linking then a networked polymer results
(Fig. 3. 15d). Cross- linked or network polymers have
a more rigid structure than linea r polymers, and
are often therefore stronger but also more brittl e.
A polymer is one of two types depending on its
behaviour with ri sing temperature. Thermopl asti c
polymers soften when heated and harden when cooled,
both processes being totall y reversibl e. Most linea r
polymers and some branched polymers with fl exibl e
chains fall into thi s group. Common examples are poly-
ethylene, polystyrene and pol yvinyl chl oride. Thermo-
setting pol ymers, which harden during their formati on,
do not soften upon heating; these are mostly cross-
linked and networked pol ymers, and they are
generall y harder and stronge r than thermopl asti cs.
Exa mpl es are vulcani sed rubber and epoxy resins.
35
Fundamental s
(a) Linear
Repeat
units
(b) Branched
(c) Cross-linked (d) Networked
Fig. 3. 15 Schematics of molecular struct ures of polymers (after Callister, 2007) .
Crystalline region
70
60
50
ro-
o..
40
U)
U)
30
Ui
B: Cross- li nked, crystall ine - britt le
C: Mi xed
Amorphous 20
regi on
Fig. 3. 16 Schematic of molecular arrangement in
polymers.
The polymer chains can pack together either in
a ' random-walk' di sordered manner, giving an amor-
phous structure, or in regul ar repea ting patterns,
giving a crystalline structure. Oft en both types of
structure will occur at in different regions of the
sa me polymer, as illustrated in Fig. 3. 16.
The stress- strain behaviour of polymers is dependent
on the extent of the crystallinity and cross- linking.
Figure 3.1 7 shows three possibl e fo rms of stress- strain
curve. Curve A is typi cal of a linea r polymer; there
are large recovera bl e strains at low st resses while
the intertwined long molecul ar chains are pull ed
stra ighter, foll owed by an increase in the sti ffness
as the chains become ali gned. Materi als exhibiting
thi s type of behaviour are known as elastomers .
36
10
A: Linear, amorphous - elastomeri c

0 2 4
Strain (%)
6 8
Fig. 3. 17 Typical stress- strain behauiour of polymer
types (aft er Callister, 2007) .
Polymers that are heavil y cross- linked and crystal-
line ca n have a hi gh ela sti c modulus and be ve ry
brittl e with low failure strains, as in curve B.
polymers with a mi xture of crystalline and amorphous
regions and an inter medi ate level of cross- linking
behave as in curve C, i. e. with di stinct elast ic and
pl asti c behaviour ve ry simil ar in nature to that of
mild steel, whi ch we di scussed in Chapter 2.
Reference
Ca lli ster WD (2007). Materials science and engineering.
An introduction, 7rh editi on, john Wi ley and Sons,
New York.
An important consequence of the structures o f solids
described in Chapter 3 is the nature of the fracture
a nd crac king processes that occur when they a re
subj ected to suffi cientl y hi gh stress. This is the sub-
ject of the branch of materi als science called fracture
mechanics; we will introduce some of the concepts of
thi s in thi s chapter, including the importan t property
of toughness.
The terms ' fracture' and ' failure ' a re o ften used
synonymo usly but they are not necessa ril y describing
the sa me process. In its broadest sense, failure means
that a structure or component of a structure becomes
unfit for further service; thi s can be due, fo r exa mpl e,
to excessive defor mati on or by reduction of area
owing to corrosion or abrasion as well to loca l break-
down or fracture. Fracture is the sepa ration of a com-
ponent into two o r more pi eces under the action of
a n imposed load, at temperatures low compared to
rhe melting temperature of the material. As we have
seen in Chapter 2, thi s sepa rati on ca n occur under a
graduall y increasing load, a permanent or static load,
leading to creep rupture, a fluctua ting load, leading
to fat igue, or a n impact load. We have also seen that
fracture ca n be of a brittle or ductil e nature, depend-
ing on the a mo unt of strain before fracture.
We need to be able to predict and analyse fracture,
a nd so we wi ll sta rt by considering predi cti ons of
strengt h from a knowl edge of the bonding forces
between a toms. This is a logica l place to sta rt , but
as we will see these estimates turn out to be wildly
inaccu ra te and so we then need to turn to fracture
mecha nics.
4.1 Theoretical strength
To fracture a ma teri al, we need to brea k the bonds
between the individual atoms and ma ke sure that
rhey do not refo rm. It is therefore instructive to
start by considering the energies a nd forces within
the bonds. As we ll as lea ding to a n estimate o f the
Chapter 4
Fracture and
toughness
tensil e strength, we will establi sh the theoretica l basis
for some of the o bserved deformation behaviour
a long the wa y.
There are both attracti ve a nd repul sive forces
between atoms, which bal a nce one another when
the atoms are in equi librium. The causes of these
forces a re somewhat compl ex, but in si mpl e terms
they a re mainly due to the gravitati onal attraction
between the t wo masses (whi ch is concentrated in
the nucl eus) and the repul sive force between the
similarl y (negatively) charged elect ron clouds as they
sta rt to overlap. However, whatever the ca use, the
energies tend to vary as the inverse of the di stance
between the atoms, raised to some power. So, if the
dista nce between the atoms is r, a nd A, B, m a nd
n are constants that va ry with the material and its
st ructure then:
the atttracti ve energy is Ar-"
the repul sive energy is Br-"'
the res ulta nt energy is U = Br-"' - Ar-"
Pl otting these energies as a function of interatomic
spacing gives the Condon-Morse curves, shown
schematica ll y in Fig. 4. l a. Figure 4.lb presents the
sa me information , but in terms of the fo rce between
ad jacent atoms, F, which is the differenti a l of the
energy wit h respect to di sta nce. There a re th ree
things to note:
1. The bond energy U is a continuous function of r.
Thus we ca n express the energy as a series:
U (r ) = U (rO) + r( dU/ dr ),o + (r
2
/2)(d
2
Ufd,2),o + ...
(4.1 )
where U (rO) is the energy at r = Yo, i. e. the inter-
atomic separati on at which the attractive and
repul sive forces balance, a nd the differenti al is
ta ken at r = r
0

2. The minimum in the curve at r
0
allows the second
term to be eliminated, since (dU/ dr ) = 0 at a
m1n1mum.
37
Fundamentals
Br m
'
- -------
Net energy,."
/
/
"
"
"
"
(a) Energy
Ar "
Q)
u
0
lL
'
'
' Attraction
" /
I
Repulsion
(a) Force
Fig. 4. 1 Condon- Morse curues of uariation of energy and int eratomic force with r. the interatomic spacing.
3. The di splacement from r
0
is small , so 1gnonng
terms hi gher than r
2
we obtain:
U(r) = U(r
0
) + (r
2
12 )( d
2
Ui dr
2
)r
11
(4.2 )
and hence
F = dU/ dr = r(d
2
Uidr
2
),
0
(4. 3)
i.e. the force is proporti onal to di spl acement via
a constant (d
2
U/dr
2
) .
In other words, the constant of proporti onalit y
is the slope of the F-r graph at the equilibrium
position where r = r
0

We can use these mathemati cal facts about these
graphs to predi ct some conseq uences and try and
relate them to the rea l world. There are in fact a
great many consequences but those most releva nt
to the subj ect of t hi s book are as fo ll ows.
1. When a materi al is extended or compressed a
littl e, the force is proporti onal to the extension
(equation 2.4). Th is is Hooke's law. The slope
of the F-r cur ve at r = r
11
is th e funda ment a l
ori gin of the elas tic consta nt E (or stiffness).
2. Since the F-r curve is nea rly symmet rica l about
the equi li brium position, the stiffness of a materi al
will be nearl y the sa me in tension and compres-
sion. Thi s is, in fact, the case.
3. At large strains, greater than about 10%, the F-r
curve ca n no longer be considered straight and
so Hooke's law shou ld brea k down . It does.
4. There should be no possibility of failure in com-
press ion since the repul sive force between the
atoms increases ad infinitum. Thi s is so.
5. There shoul d be a limit to the tensile strength,
since the attracti ve force between the atoms has
a maximum va lue. Thi s is so.
We can make a theoreti ca l estimate of thi s tensil e
strength (orr ) by ass uming that, on fracture, the
38
internal strain energy due ro the loading goes to
crea ting new surface. We wi ll discuss surfaces Ill
more detail in Chapter 6, but fo r t he moment we
need to use the concept of surface energy. A toms
at the surface are bonded to other surfa ce atoms
and atoms furth er into the materi al; t hey therefore
have asymmetri c bonding forces leading to a hi gher
energy state than that of atoms with in the material,
whi ch have uniform bond ing in all directions
(Fig. 6. 1 ). Thi s excess energy is the surface energy (y)
of the materi al; it gives ri se to the surface tension of
li quids, but is perhaps not quite so obvi ous in solids.
Analys is equa ting es timates of the surface energy
to the internal strain energy immediately before
fracture gives, after making some simplifying assump-
ti ons about the Condon-Morse cur ves:
(4.4)
For many materi als, thi s gives a value for 0'
1
, of
about /10. On thi s basis we would expect the
strength of steel to be approximately 20,000 MPa,
whi ch is about :I 0 times higher th an the st rongest
steel that we are c:tpabl e of producing- a probl em!
4.2 Fracture mechanics
Clea rl y some other explanati on than that above is
required ro explain the va lues of tensi le strength that
we obtain in practi ce. Thi s is provided by fr acture
mechani cs, whi ch arose fr om the studi es of A. A.
Griffith in the 1920s on the brittl e fractur e of glass .
Griffith recognised that all materi als, no matter how
ca refull y made and how uni form in appeara nce,
contain defects and fl aws and it is the propagation
or growth of these defects that leads to fr acture.
These may be mi croscopic e.g. as in the case of a
metal that is ma de up of fine grains or crysta ls (as
described in Chapter 3) or macroscopi c as in the

Fig. 4.2 Surface a11d internal crack geometry.
case of concrete with large aggregate particl es
bonded imper fec;ly together by the surrounding
hardened cement. So, in the above analys is we have
therefore made some incorrect ass umpti ons about
both the stresses within a materi al and the nature
of the fracture process.
First, consider the stresses within the materi al.
We have assu med that the stress acts uni for ml y
across a secti on, and is therefore simpl y the load
di vided by the cross-secti onal area over whi ch it is
acting. We can thin k of defects and fl aws as cracks,
which may be either at the surface or contamed w1thm
the materi al. Cracks are usuall y long and narrow
wi th a sharp tip, and so we can draw them as shown
in Fig. 4.2, with a length a for a surface crack and
2a fo r an internal crack, and a ti p rad1us r 111 each
case. The cracks act as local stress ra isers, with
rhe stress at the crack tip being many times greater
than the average st ress in the materi al. It is possibl e
with stress analys is techniques ro show that the
loca l stress (a
10
J is hi ghest at the crack tip and is
given by
(4.5)
= 2a(a/r)
01
for small r (4. 6)
We ca n also define the stress concentration fact or,
k,, as
k = a
1
.Ia = [1 + 2(alr)
0

5
] or 2(alr)
05
for r a
I <>< . (4.7)
For a circle, a = r and so k, = 3, but for a small
sharp crack, say a = I mm and r = 1 j..lm then k, =
63 . It is therefore easy to see how the local st ress
ca n reach the theoreti ca l strength rhar we estimated
above, and hence t he atoms will be pull ed apart.
Thi s is, however, onl y parr of the picture. Further
stress analys is shows that that the local stress a loe
quickl y dimini shes with di stance from the crack,
as shown in Fig. 4. 3. The atoms will be torn apart
nea r the crack rip, bur the crack will onl y grow (or
propaga te) through the materi al if there is also
Fract ure and toughness
Distance from crack tip
a
Fig. 4.3 Variati on in local internal stress with distance
from crack tip (from Ashby and jones, 2005).
suffi cient energy in the system ro keep dri ving ir.
Thi s energy is, of course, the internal strain energy
ca used by the loading. In ductil e materia ls, as well
as creating new surface thi s energy is consumed
in pl asti c deformati on of the maten al 111 the reg1on
where the local st ress is hi gher than the y1eld stress,
whi ch may be some di stance in adva nce of the
crack tip. Even in brittl e mater ials w1th no pl asti c
deformati on some locali sed mi crocracking may
occur in thi s region.
Therefore when considering the balance between
the energy consumed by rhe crack propagati on that
is avail abl e from the internal strain energy we need
to consider more than just the energy of the new
crack sur face. We do this by defining the total energy
consumed when a unit area of new crack is for med
as rhe toughness (GJ of the materi al. G, is a
property, with units of energy/a rea e.g. Ji m- .
In brittl e materi als, such as glass, the surface
energy is a signifi ca nt part of the total energy, G,
is low and crack propaga ti on occurs readil y at low
stra ins. In ductil e mater ials, such as mild steel,
rhe opposite holds; the energy requi red for pl asti c
deformati on as a crack adva nces is ma ny orders of
magnitude hi gher than the new surface energy, G,
is hi gh and fa ilure requi res much h1 gher strams.
Analys is of the energy balance gives the va lue of
the stress ro cause fracture (a r) as
a r = (Gc/rra)
05
(4. 8)
for the conditi on of pl ane stress, whi ch occurs when
rhe material is relati vely thin in the directi on per-
pendi cul ar to the appli ed load, and
a r = [ GcE/rr (l - u
2
)a]
01
(4.9)
(where u = Poisson's rati o)
for rhe conditi on of pl ane strain in thi cker secti ons.
39
Funda menta ls
We therefore now have equati ons for the failure
stress, 0
1
, in terms of three mate ri al properti es (Ge,
E and u ) and a defect size, a. A grea ter defect size
would lead to a lower strength , whi ch is logica l,
but thi s is a fundamental difference to the previous
idea that the strength of a material is a constant
when defin ed in terms of stress.
Rearranging equat ion (4.8 ) shows that failure will
occ ur when the combinati on of appli ed stress and
crack size sa ti sfy
(4.1 0)
For any combinati on of appli ed stress and crack
length the term o- (anf; is ca ll ed the stress intensity
fac tor, denoted by K. At the combinati on of stress
and crack length to ca use fracture the va lue of K
is ca ll ed the critical stress intensity factor, K, . K, is
more commonl y known as the fractu re toughness,
and it fol lows that:
(4. 1 L)
Thi s choice of name is a littl e confusing, but you
must remember that roughness, G" and fracture
ro ughness, K_, are different propert ies with different
va lues and different units. K, has units of force/
length
112
, e.g. M / m
312
The va lues of G,. and K, ca n
va ry widely for different mater ials, and ranges of
both properti es fo r the most common constructi on
materi als are shown in Table 6J.1.
Substituti on of K, into equation 4.10 and re-
arranging shows that for any parti cul ar applied stress
there is criti ca l crac k length, a,, , that will res ult in
fracture, given by
(4.12)
When equati on 4.12 is sati sfi ed, the crack wi ll propag-
ate rapidl y to fa ilure i. e. there will be a fas t fracture.
With britt le materi als, without the capacit y for
plasti c deformati on and which consequentl y have
low va lues of G, and hence K, small defects, of the
sa me order of size as mi ght occur during manu-
facture, are suffi cient for fas t fracture. The result is a
so-call ed cleavage type of facture. Conversely, ductil e
materi als, whi ch have the ca pacit y for significant
pl asti c fl ow and yielding, have hi gher G, and K, values,
and therefore defects from manufacturing are not
suffi cient for fast fr acture to occur before excess ive
yielding and a tearing form of facture occur. How-
ever, it fo ll ows that, with large cracks (e.g. fro m
some previous damage) or stress concentrat ions
from poor des ign detailing, fa st fracture may occur
before yielding, a potenti all y dangerous situati on.
The crack size at which failure in a ductil e materia l
changes fr om yielding and rea ring into fa st fract ure
40
15
ui
Ul
Yield stress ov
~
u; Failure by
~ yielding
~
iii
LL
: Failure by fast fracture
Defect size, a
Fig. 4.4 Variation of failure stress with defect si::,e (or
a ductile material.
and cleavage, the critical yield crack length, a"'"' is
obtained by substituting the yield stress 0 , into eq ua-
ti on 4.12, giving
(4.13 )
A plot of failure stress aga inst defect size for ductil e
materi als is of t he form shown in Fig. 4.4. The va lue
of a""" is a signifi ca nt property, since if the material
contains cracks larger than thi s fa ilure wi ll nor onl y
occur at a stress lower than expected but also without
warn ing; in ot her words a ' sa fe' ductil e mater ial
behaves as a ' dangerous' brittl e material. It is there-
fore useful to know a>.,,;, not onl y for design bur a lso
when inspecting structures during their servi ce li fe.
Toughness and fracture ro ughness are sometimes
described as the ability of a materi al to tolerate cracks.
Lack of toughness ca n lead to fast fr acture, which is
extremely dangerous as it ca n occ ur without warni ng
and will often res ult in catastrophi c failure and loss
of life. Enginee rs therefore need to take great pains
to avoid it occur ri ng. Clear ly one way of doing thi s
is to avoid using material s with low G, or K, va lues.
However, thi s is not always possibl e, e.g. construction
without using any concrete is diffi cult to contempl ate,
and so we should then either:
ensure that the loading wil l not ca use hi gh tensil e
stress and/or
rein force aga inst fast fracture (or at leas t reduce
its ri sk) by forming a composite, e.g. by adding
steel reinforcement to concrete or fibres to resins.
Reference
Ashby MF and Jones DRH (2005) . Engineering Mat erials,
Vol '/ : An introduct ion to properties, applications and
design, 3rd editi on, Elsevier Butterwort h Heineman,
London.
5.1 Liquids
Liquids are effecti vely incompressibl e when subjected
to direct stress (hydrau li c power systems depend on
rhi s property), which impli es that the elementary
particles (atoms or molecules) are in direct contact.
However, liquids obviously fl ow under rhe acti on
of the shear stress, which shows rhar the parti cles
are able to move relative ro each ot her i. e. there are
no primary bonds between them. In describing their
behaviour we are therefore concerned with the
relati onship between the appli ed shear stress, T, and
the rate of shear strain, dy/dt (Fig. 5.1 ). Most simpl e
liquids, such as wa ter, white spirit, pet rol, lubricating
oil ere., and many true soluti ons, e.g. sugar in water,
show ' idea l' or Newtonian behavi our where the rwo
are directl y proportional i. e.
T = 11dy/dt (5.1)
where 11 = coeffi cient of viscos ity (or, stri ctl y, the
dynamic viscosity) which, as strain is dimensionl ess,
has units of stress x rime e.g. Pa .s.
Thi s definition also appli es to gases, but as might
be ex pected, the viscositi es of gases and liquids differ
markedl y. At 20C, 11 for air is about 1.8 x 10-
1
Pa.s
and for water it is about I x 10-
1
Pa.s. In both types
of fluid ar hi gher temperatures the particles possess
more energy of their own and the st ress required ro
U'_j
Force P 11
/::?-Are- a A - ~ - - - ~ ____ ..,. :r '7
]
I j ... ...,. .. I
r - - - - - - - - :::...-( .. -- 1
,.... --------+ t ..... v
1- - - - - - - - - - ('' ... . -: .. -
~ - - - - - - - - - ... .... .,
L---------V;----
Shear stress 1 = PI A
Shear strain rate = dyldt = d(xl y)dt
Fig. 5.1 Shear stress and shear strain rate for fluid fl ow.
Chapter 5
Liquids, viscoelasticity
and gels
move them is reduced, i.e. viscosity reduces rapid ly
as temperature is increased (think of asphalt road
surfaces on very hor da ys) .
We are often faced wirh liquids that contai n disper-
sions of solid particles. These disturb and effecti vely
increase rhe viscosit y; for sma ll volume fractions of
particles rhe viscosit y is given approx imately by
11 = 11 o[l + exVr I
(5.2)
where 11 o = rhe viscosit y of the pure fluid , Vr = the
volume fracti on of particles and ex is a constant.
The value of ex va ri es with rhe shape of rhe particles;
Einstein showed rhar ex = 2.5 for spheres, bur ir is
hi gher for irregu lar particles.
This eq uati on breaks down when the volume frac-
tion of rhe particles increases ro rhe point where
the perturbed regions in rhe li quid begin to overlap,
and terms in V/ appea r. Materia ls such as pastes,
clays and freshly mixed mortar and concrete, which
have soli ds contents in excess of, say 70%, can
deform more or less elasricall y up to a certain yield
stress and can preserve their shape against gravit y.
Above this stress, however, they behave like liquids
and deform rapidl y, e.g. toothpaste does not fl ow
off your toothbrush bur you ca n brush it around
your teet h, clays can be moulded to the shape desired
by the potter, concrete can be shovell ed and vibrated
into the formwork. The shea r st ress-stra in rate curve
(known as the flow curve ) for these mater ial s can
take a variet y of forms, as shown by rhe solid lines
in Fig. 5.2. The genera l equati on for rhe three types
of behavi our that have a positive va lue of the yield
stress IS:
T = T
0
+ a(dy/dtt (5.3)
where ' " is rhe intercept on rhe shear stress axis
and a and n are constants. The three curves have
different va lues of n. In shear thinning behaviour,
the curve is convex to rhe shear stress axis and n < l ;
in shear thickening, the curve is concave to the shear
stress ax is and n > 1. The part icular case of a
41
Fundamentals
Shear thinning
: ~ Ltnear (Btngham)
~ ~ / ~ ' : ' th>ekeomg
tiJ ~ / ~ Newtontan
/
/
/
/
Rate of shear strain, dy/dl
Fig. 5.2 Different f orms of shear stress-shear strain
rat e flow curves .
stra ight -line relati onship is cal led Bingham behav-
iour, for whi ch n = 1. The equati on for thi s is
normall y written as:
1 = 1, + 1-l. (dy/dt ) (5 .4)
where 1, is the yield stress, and 1-l is the plastic
viscosity.
This is of parti cul ar interest for concrete tech-
nologists, as fresh concrete has been shown to
confo rm reasona bl y wel l to thi s model. We wi ll
di scuss this further in Chapter 1 8.
5.2 Viscoelastic behaviour
In man y cases it is not possibl e to draw a sharp
dividing line between the mechanica l behaviour of
liquids and solids; there is a large group of mater-
ials, known as viscoelastics, whose behavi our is
part liquid and pa rt solid. Many natural material s,
e.g. tendons, pl ant fibres and wood, behave in thi s
way. Of engineering matenal s, rubbers, many soft
pol ymers and substances like tar and as phalt are
exa mpl es .
We have already briefl y discussed such behaviour
in Chapter 2 when defi ning the two separate but alli ed
cases of creep and stress relaxa ti on. Under constant
stress, a mater ial res ponds by steadil y increasing
stra in; under constant strain, stress relaxation occurs
wit hout dimensional change (Figs 2. 15 and 2.7 7).
Some of the mi crostructural mechanisms of thi s
behaviour in different materi als will be di scussed in
later parts of thi s book, but here we introduce how
the behavi our ca n be modell ed by using mechanica l
analogues consisting of arrays of springs that behave
according to Hooke's law i.e. stress oc strain, and
42
Spring, 5:
CT
5
= E
5
-- Dashpot , fluid F:
CTF = ll Fd
5
/df
CT
(a) Maxwell
(b) Voigt-Kelvin
Fig. 5.3 Viscoelastic models.
viscous elements that behave as an idea l Newtoni an
liquid, i. e. st ress oc rate of stra in.
One such ar ray, known as the Maxwell model,
is shown in Fig. 5.3a. It consists of a n elas ti c spring,
S, of mod ulus E in ser ies wi th a das hpot, i. e. a
piston moving in a fluid, F, of viscosity 11 contained
in a cylinder. Now think of suddenl y a ppl yi ng a
constant strain. At first al l the stra in is ta ken up by
stretching the spring and the load required to do thi s
is calculated from the strain in the spring. Later, the
spring shortens by pulling the pi ston up through the
fluid in the das h pot. Some of the tota l stra in is now
taken up by the movement of the pi ston and less
by the stretch in the spring. The load required is now
less than before, and thus the system is ex hibiting
stress relaxa tion . Mathemati ca l a na lys is gives :
a, = a" exp(-t/1) (5.5)
where a" is the initi al appli ed stress, a, is the stress
sustained at time t and 1 = 11/ E is the so-ca ll ed
relaxation time. Under consta nt stra in the stress
decays exponentiall y, which is reasonabl y close to
observed behaviour. In fact, 1 is the time taken for
the stress to decay to 1/e of its init ia l va lue.
Now take the case of applying a constant load
or stress. The spring stretches and rema ins ar that
strain as long as the load remains. At the sa me time
the dashpot slowly extends as the piston is pull ed
through the fluid in it. The tota l extension therefore
increases linea rl y with time, which is not typi ca l
creep behaviour. Thi s model therefore represents
stress relaxa ti on very well , but is less successful at
representing creep.
For modelling creep, we ca n use the so-ca ll ed
Voi gt-Kelvin model in whi ch the spring and clashpot
are arranged in parallel (Fig. 5.3b). Both elements
must exper ience the sa me strain at any given time
but load ca n be transferred over time from one
element to the other. Analysis of the model gives:
Maxwell
Voigt- Kelvin
Fig. 5.4 Four element viscoelastic model.
(5.6)
where , = stra in a t time t, 0 = appli ed stress a nd
E and 1 are as before.
This gives a good representation of creep behav-
io ur, but not of relaxation. To get out o f these
diffi c ulti es, the two t ypes of models a re combined
into what is known as the four-element model
(Fig. 5.4). This gives a reasonable representation of
both creep and relaxation in many cases, but where
the viscoelastic material is a polymer consisting of many
molecul es and particles of varying size and properties,
ma ny elements with different relaxation times (i. e.
a relaxation spectrum) need to be combined.
Nevertheless, the concept of relaxati on times is
important fo r two reasons. First, it helps us to di s-
tinguish between solids and liquids. A perfect solid
will s upport the stress indefi nitely, i. e. 1 = =, but
fo r a liquid, relaxa ti on is virtually instantaneous
(for water 1 "" 10-
11
s). In between there is a grey
a rea where stress relaxation may occur over a few
seconds or centuri es.
Second, we have the relationship between the
relaxa ti on time and the rime sca le of the loading t.
If the load is appli ed so fast that relaxation ca nnot
occur (t << 1) the materi al will effecti ve ly behave
elasti ca ll y, but under slow loading (t >> 1) it will
fl ow. This was one o f the effects that we menti oned
when considering impact loading in Chapter 2. An
extreme case is the well-known ' potty putty', which
bounces when dropped or thrown aga inst a wa ll but
coll apses into a puddl e under its own weight when
left a lo ne. Potty putty is a silicone- based inorgani c
polymer, and ma ny other po lymeric materi als also
show marked sensiti vity to loading speed.
There are two important consequences of viscoelas-
ti city. The first is tha t the stress-strain relationship
is non-linea r. We noted that in a n elastic or Hookean
solid the st ra in energy stored on loading is completely
recovered when we unload. Figure 5.5 shows that
for a viscoelasti c materi al the energy recovered on
unl oadi ng is less tha n that stored during loading.
Liquids, viscoelasticity and gels
Strain
Fig. 5.5 Loading/unloading behaviour for a viscoelastic
material.
..--- Loading 3
Loading 1
Strain
Fig. 5.6 Boltzmann's superposition principle.
This energy must go somewhere, a nd norma ll y this
is into hea t, whi ch explains why car t yres get hot
after a few mil es in which they are repeatedl y loaded
and unloaded.
The second consequence is known as Boltzma nn's
superposition theory. This states that each increment
of load makes an independent a nd additive con-
tribution to the total deforma ti on. Thus, under the
loading programme shown in Fig. 5.6, the creep
res ponse is additive a nd the total creep is the sum
of all the units of incremental creep. This is useful
in the a nal ys is of varying load levels on the creep
behavio ur of concrete and soil s.
5.3 Gels and thixotropy
There is a group of materi als tha t show a mi xture
of solid a nd liquid behaviour beca use they are just
that- a mi xture of a solid a nd a liquid. One of the
most familiar of these is the gel, known to most of us
from childhood in the form o f jelli es and pastilles.
Gels are formed when a liquid contains a fairly
concentrated suspension of very fine particl es, usuall y
43
Fundamentals
of colloidal dimensions(< I ~ m ) . The pa rti cles bond
into a loose srrucrure, trapping li quid in irs intersti ces.
Depending on rhe number of links formed, gels ca n
va ry from ve ry nea rl y fluid structures to almost ri gid
solids. If the links are few or wea k, the indi vidual
particles ha ve considerabl e freedom of move ment
a round thei r points of contact, and the gel defo rms
eas il y. A hi gh degree of linkage gives a st ructure that
is hard and ri gid in spite of all its internal pores.
The most impo rta nt engineer ing gel is undoubtedl y
hardened cement, whi ch develops a hi ghl y ri gid but
permea ble structure of compl ex ca lei um s i I ica res by
the chemica l react ion between rhe fres h, powdered
ce ment and water. We discuss thi s in some detail in
Chapter 13.
A fea ture of ma ny gels is their ve ry hi gh specifi c
surface area; if the gel is permeable as well as porous,
t he surface is ava il able for adsorbing la rge amounts
of water va pour, a nd such a gel is an effecti ve drying
agent . Adsor pti on is a reversibl e process (see sec-
tion 6.4); when the gel is saturated it may be hea ted
to dri ve off the water and its drying powers regained.
Si li ca gel is a n exa mpl e of thi s.
If a gel sets by the formation of rather weak links,
the linkages may be broken by vigorous stirring so
that the gel liquefies aga in. When t he stirring ceases,
rhe bonds will gradual ly li nk up a nd rhe gel will
t hi cken and return ro its original set. Reve rsible
44
behaviour of thi s sort , in whi ch an increase in the
appli ed shea r rate causes rhe material ro act in a more
flu id manner, and vice-versa, is known as thixotropy.
A fa mili ar a ppli ca tion of thi s is in non-drip paints,
whi ch li quefy when stirred and spread eas il y when
being brushed on, but vv hi ch ser as a gel as soon as
brush1n g is compl eted so that dripping and strea ks
on verti ca l surfaces a re avoided. Clays ca n also
exhibit thi xotropy. This is rurned to adva ntage on
a porter's wheel and in rhe mi xing of dril ling muds
fo r oi l expl oration. The thi xotropi c mud serves to
line rhe shafr wi th a n impermeable layer, whil st in
rhe centre ir is kepr fluid by the movement of the
drill and acrs as a med ium for removing rhe rock
dril lings. On the orher hand, a thi xotropic clay
underl ying maj or civil engineering works could be
high ly haza rdo us.
The reverse effecr to thi xotropy occurs when an
increase in rhe a ppli ed shear rate ca uses a viscous
materi al to behave mo re in the manner of a solid,
and is known as dilatancy. Ir is a less famili a r bur
rather more spectacu lar phenomenon. Cornfl our-
wa ter mi xtures demonstrate the effect over a rat her
na rrow range of compos iti on, when the viscous
liquid will fracture if stirred vigorously. It is of short
durati on, however, since fracture reli eves the stress,
and the fractured surfaces immediately liquefy and
run together aga in .
All mate ri als a re bounded by s urfaces, whi ch a re
interfaces of va r ying na ture. Fo r the enginee r the
most importa nt are the li q uid-va po ur, solid-vapour,
so li d-li qu id a nd solid-solid interfaces. The las t o f
these can be the boundary between t\vo differing solid
phases in a ma teri a l, e.g. cement gel a nd aggrega te
in concrete, o r betwee n two simil a r crysta ls tha t
differ onl y in orienta tion, e.g. the gra in bo unda ri es
in a pure metal o r, at the macroscopi c level, as
the interface between structura l compo nents, e.g.
concrete and steel. Surfaces owe t heir interes t a nd
importance to two simpl e fea tures:
they a re a reas o f a bno rma li t y in rela ti on to the
structure tha t they bo und, a nd
they a re the o nl y part of the ma teri a l access ibl e
to chemi ca l cha nge, i. e. all chemi cal cha nges and,
for th at matter, most tempera ture cha nges ta ke
place at o r thro ugh surfaces.
The influence of sur faces o n the bulk behaviour
o f materi als depends o n the ra ti o of sur face a rea to
the total mass. Thi s in turn depends partl y on the
size and part ly on the shape of the indi vidual
pa rti cles making up the bu lk ma ter ia l. An extreme
exa mpl e is cla y, which is composed o f platelets
t ypi ca ll y 0.01 IJm thi ck by 0.1 !Jill across; they a re
therefore very sma ll with a hi gh surface area/volume
ratio. One gram of montmorill onite clay, rather small er
tha n a suga r cube, may cont a in a tota l surface area
of over 800 squa re metres! Porous structures such as
ha rdened cement a nd wood a lso conta in eno rmous
interna l surface a reas that exert a considera bl e effect
on their enginee ring properti es.
6.1 Surface energy
As we bri efl y di scussed in Cha pter 4, a ll sur faces
have o ne thing in common: the atoms, mo lecul es
o r io ns a t the surface a re subj ected to asymmetric
or unsaturated bonding forces (Fig. 6.1 ). Since bond-
Chapter 6
Surfaces
Fig. 6. 1 Asymmetric and symmetric forces in surface
and internal at oms, molecules or ions.
tng 1 ta ken to lower their energy (Fig. 4.1) the
s urface atoms or ions wi ll be in a sta te of hi gher
energy than interi or ones. This excess energy is known
as the surface energy o f t he mate ria l. In soli ds the
presence of the surface energy is no t immed iately
a ppa rent, since the a toms in the s ur face a re held
firmly in position. However, wi th li quids the mobil e
structure per mits the individua l atoms to respond,
and the resul t is the well -known surface- tension effect.
Si nce s ur faces a re hi gh-energy regions they will
a lways act to minimi se their a rea, a nd t hus lowe r
their energy, when possibl e. If a soa p film is stretched
across a frame wit h a movable wire as in Fig. 6.2,
the force required to hold the wire 111 place is:
F = 2y/ (6.1 )
where I is the length of the wire, y is the surface
tensio n o f the soa p film/air inter face a nd the factor
2 is int roduced beca use the film has two surfaces.
The surface tension is in the plane o f the soa p-air
interface a nd has units of force per unit length (N/m).
It is impo rta nt to note that surface tensio n di ffers
from a n elas ti c fo rce act ing between the surface
a to ms in tha t it rema ins constant whet her the film
is forced to expa nd o r a ll owed to contract. Thi s is
beca use the work done in expand ing the fi lm is used
to bring add itiona l atoms to the surface rather tha n
45
Fundamental s
Soap film
0 . 0
. . .
. . . . . . . . .
. . . . . . . . . . ...
F
~ Sliding wire
Fig. 6.2 Equilibrium between soap fi lm and applied
for ce.
to increase the interatomic spac ing in the surface.
Only when the fi lm has become so thin that the
two surfaces interact with eac h ot her will the force
show partial elasti c behaviour, by whi ch time the
film is on the point of rupture.
If the fi lm is stretched by pulling the movable
wire through a di sta nce d, the work done o n it,
2/yd, is stored as surface energy of the newly crea ted
surface of a rea 2/d, therefore th e surface energy per
un it area is 2/yd/2/d = y. Thus surface tension and
surface energy are numeri ca ll y equal with units of
Jm-
2
(= Nm-
1
). It sho uld be noted that y tends to
be used interc hangea bl y for both surface tension
and energy.
6.2 Wetting
We are all famili a r with liquids wett ing a solid
surface. Clea rl y intermol ecul a r forces are involved,
and the beha viour is another example of a system
seeking to minimi se its total energy. The degree of
wetting ca n be defined by the contact angle (8)
between the li quid-va pour interface a nd the solid-
liquid interface at the edge of a droplet ( Fig. 6.3) .
If the conditi ons for wetting are favoura bl e, then the
contact angle is low, and the liquid wi ll spread over
a la rge area. If the conditi ons are less favourab le
then the contact angle wi ll be hi gher and the liquid
will form droplets on the surface. If 8 is grea ter than
90 then the surface is oft en sa id to be unwettabl e
by the liqui d.
The behaviour depends on the relative magnitudes
of three surface tensions or energies: liquid-solid, y
1
,,
liquid-va pour, Yt.o and solid-vapour, Yw At the edge
of the dropl et the three tensions will act as shown
in Fig. 6.3 a nd the eq uilibrium conditi on, resolved
parall el to the solid surface is:
46
Vapour
Ysv
Soli d
Fig. 6.3 Surface {orces at the edge o{ a droplet .
(6.2)
Since the limits to cos8 are 1, a restr icti on of thi s
equation is that it does not appl y if either y" or y
1
,
is larger th an the sum of the other two surface
energies. If either of these is not the case, then:
if y,, ~ Y1, + y
1
,, then there is compl ete wett ing (in
effect, 8 = 0)
if y
1
, $ Yw + y
1
,, then there is no wetting (in effect,
8 = 180).
The quantity Yw- {y
1
, + y
1
J is ca ll ed the spread ing
force or spread ing parameter and the behav1our
depends on whether it is positive or nega ti ve:
If it is posi ti ve then there is compl ete wett ing
{the fir st case a bove) of the soli d sur face; clea rl y
the energy of such a system is lowe red when the
solid-vapour interface is repl aced by a soli d-
liquid and a liquid-vapour interface.
If it is nega ti ve then 8 > 0 and pa rti al or littl e
wetting occurs.
if y" > y
1
, then Yt- COS 8 (equati on 6.2) is positi ve
and 8 < 90, g1ving pa rti al wetting
if y,, < y
1
, (which is compa ra ti ve ly rare, pro-
vided the surfaces a re clea n) then y
1
,, cos 8
is nega ti ve a nd 8 > 90, giving littl e or no
tendency to wetting.
The ri se of wa ter in a capill a ry tube IS a consequence
of the a bilit y of water to wet glass . If, in Fig. fi.4,
8 is the angle of contact between water and glass,
the water is drawn up the tube by a circumferenti a!
force 2rrry
1
, cos 8, so that:
(6.3)
where rr r
2
hp is the weight of water in the capill ary
(p = unit weight of wa ter ), neglecting the we ight
of wa ter cont ained in the curve of the meni scus. It
fol lows that the height of the water in the capill ary
IS:
h = 2y
1
, cos 8/pr (6.4)
If r is sma ll , h will be la rge. This gives rise to the
phenomenon with the general name of absorption,
Vapour
Ysv
Liquid
h
i
2r
I
Fig. 6.4 Capillary rise of liquid up a tube.
where water (or a ny other liquid ) is sucked into the
continuous ca pill a ri es within a porous ma teri al. Two
exa mpl es of such ma ter ia ls are brick a nd conc rete
in both of these the pores are sma ll a nd if they
all continuous then h could reach 10 m- extreme
rising da mp! In practice, the pores are not continu-
ous and eva po rati on keeps the level lower than thi s
but it is still a signifi ca nt problem. '
6.3 Adhesives
The a bilit y o f adhesives to sprea d and thoroughl y
wet surfaces is cr iti ca l. The ad hes ion of a liqui d to
a so lid surface is clea rl y relevant a nd the liquid may
also have to penetrate a thin joint e.g. when repairing
cracks with a resin of when soldering or brazing
meta ls. The work needed to break away the adhesive
(whi ch may be considered as a viscous liquid ) from
the solid is the work required to create a liquid-vapour
and a solid-va pour interface from a n equi va lent a rea
?f liqui?-solid inter face, i. e. it is the work to totally
de-wet the sol1d surface. Hence the work to cause
breakage at the inter face, per unit ar ea, is given by:
W = Y1 v + Yw- Y1, (6.5)
But fro m equati o n 6.2:
(6.6)
a nd therefore:
W = Y1 v(1 + cos 8 ) (6.7)
Thus, the liquid-solid adhesion increases with the
abilit y of the ad hesive to wet the solid, reaching a
maXImum - when 8 = 0 and wetting is complete -
g1ven by:
(6.8)
Surfaces
Film of liquid
Fig. 6.5 Adhesive effect of a thin fi lm of liquid between
flat plat es.
For thi s to be the case Yw > y
1
" (equati on 6.6) and
under these conditi o ns fracture will occur within
the adhes ive, since the energy necessary to form two
liquid-va po ur interfaces is less than that needed to
form a liquid-va po ur a nd a solid-vapour interface.
Surface tensio n is a lso the ca use of the adhes ion
between two fl at surfaces separated by a thin film
of liquid. Where the surface of the liquid is curved
(as for exa mpl e in Fig. 6.5) there will be a pressure
difference p across it; if the curvature is spher ical
o f radius r, then:
p = 2ylr (6.9)
In the case o f two circular di scs, however, the surface
of the film has t wo radii of curvature, as shown in
Fig. 6.5; r
1
is approximately eq ual to the radius of
the discs a nd presents a convex surface to the atmos-
phere whilst r
2
"' d/2, where d is the thickness of
the film between the pl ates, and presents a concave
surface to the a tmosphere. The pressure difference
between the liquid film and its surroundings is now
given by:
If d << r
1
, then:
2y
p = --
d
(6. 10)
(6. 11 )
the negati ve sign indicating that the press ure is lower
within the liquid than outside it . Since the pressure
acts over the whole surface o f the di scs, the force
needed to overcome it a nd sepa rate the di scs is
given by:
F =
d
(6.12)
The magnitude ofF thus depends on the factor r
1
2
/d,
and 1t IS therefore importa nt that surfaces to be jo ined
sho uld be as flat a nd cl ose ly spaced as possible. If
47
Fundamentals
you have tri ed to pull apart two wet glasses you
will know how tenaciously they cling to each other;
by contrast, however, they can easil y be slid apart
since liquid films have littl e resistance to shear. For
exa mpl e, If d = 0.01 mm, r = 100 mm and y(water)
= 0.073 Nm-
1
then F "" 460 N. Thi s va lue of F for
a liquid film gives some idea of the potenti al of
adhesives that ga in additi onal strength and ri gidit y
by setting to hi ghl y viscous materi als on polymeri sa-
ti on or solve nt eva porati on.
6.4 Adsorption
The ability of liquids to wet solids depends very much
on the cleanliness of the solid, as anyone with any
experi ence of soldering wi ll appreciate. The presence
of any dirt, such as oxide or grease films, will totall y
alter the balance of surface tensions di sc ussed above
and usuall y prevents wetting.
Clea n surfaces, in fact, are so rare as to be virtuall y
non-existent, since the broken surface bonds wi ll
readily attract to themselves any for eign atoms or
molecul es that have a sli ght affinit y fo r the surface
materi al. Thi s effect is known as adsorption and
by sati sfying or partiall y sa ti sfying the unsaturated
surface bonds it serves to lower urface energy.
Adsorpt ion is a dynami c process, i.e. molec ul es are
constantl y ali ghting on and taking off fr om th e
surface. Different molecul es adsorb with va rying
degrees of intensit y, depending on the nature of the
bond that they are abl e to form at the interface.
The strength of the bond may be expressed in terms
of <!>, the energy of adsorption. As in the case of
interatomi c bonds, a negati ve va lue of <!>., is taken
to indi cate pos itive adsorpti on, i.e. the molecul es
are attracted to the interface, and the sur face energy
(tension) is thereby lowered. A positive va lue of <1> "
indi ca tes a repul sive interacti on and the molec ul es
avoid the surface. Typi ca l pl ots of <!> ., aga inst the
di stance of the adsorbed layer fr om the surface are
given in Fig. 6.6. They closely resembl e the Condon-
Morse curves (Fig. 4.1) and their shape is due to the
same circumstance of equilibrium between attractive
and repul sive forces, although the attracti on is fa r
weaker than that of the principal interatomi c bonds.
If the molec ul e being adsorbed is non-polar and
does not react chemi ca ll y with the surface, absorp-
ti on, if it occ urs, wi ll be by Van der Waa ls bonds,
and the minimum va lue of <!> ., is small (curve 2 in
Fig. 6.6). If on the other hand the molec ul e is
strongly polar, as is the case with water or ammonia,
the electrostati c forces between t he surface and the
charged porti on of the mol ecule give ri se to stronger
48
B.
c
Q
0.
0
({)
"0
"'
0
>-
CJ)
(ij
c
w
1. Chemi sorption
2. Physical adsorpti on
Di stance of adsorbed
Fig. 6.6 Energies of adsorption for different adsorption
mechanisms.
bonding. If a chemi ca l reacti on occ urs as part of
the bonding mechani sm, e.g. when fa tt y acid in a
lubrica nt fo rms an adsorbed layer of metalli c soa p on
a metal surface, the bonding is still stronger (curve I)
and the effect is referred to as chemisorption.
The behaviour of wa ter is of parti cul ar import-
ance in thi s context. Beca use of irs abilit y to form
hydrogen bonds with neighbouring molecules, water
adsorbs rapidl y and strongly on most solid surfaces.
Des pite the tenacity with which such a laye r is held
(cl ay does not lose all its adsorbed water until ht:ated
to 300C) , the interacti on ca nnot be thought of as
chemi sorpti on; rather it is, in a sense, a halfwa y
stage to soluti on or alternat ive ly to the taking up
of wa ter of hydrati on (see below) . Bonding is strong
enough ro maintain a sur face layer perhaps seve ra l
mol ecul es thi ck, but the affinit y is not suffi cient for
the molec ul es ro penetrate into the inrc: rsti ces of the
structure.
The phys ica l nature of such a film is diffi cult to
visuali se; it cannot be thought of as a fluid in the
accepted sense of the term even when more than one
molecul e thi ck, as in the case of the cl ays and cements.
Yet the molecul es are mobil e in thi s situati on. They
will not desorb readil y, but they ca n diffuse along
the surface under the impetus of pressure gradi ents.
Such movements, occurring over the vast internal
surface area of cement gels, are primaril y responsi bl e
for rhe slow cree p of concrete under stress (see
secti on 20.6). The ability of wa ter molecul es to
penetrate solid-solid interfaces in clays and build
up thi ck adsorbed layers results in the swelling of
clays, and has ca used considerabl e struct ural damage
to buildings erected on clays that are li abl e to behave
in thi s manner. The readiness with which water will
adsorb on surfaces is adva ntageous in the case of
porous silica gel and mol ecul ar sieves being used as
drying agents.
6.5 Water of crystallisation
As well as water being assoc iated wi th a materi al
by absorption into capillaries or adsorpti on by the
surface, many ioni c crystals contain water molecul es
locked up in their struct ure as water of crystalli sa-
tion. Such crystals are known in general as hydrates,
and their formation ca n be very important in the
development of bulk strength. Both cement and
' pl as ter of Pari s' owe their import ance to their
ability to take up water and form a rigid mass of
interlocking crystals. However, in the case of cement
the crys tal s are so small that it is difficult to decide
whether to cl ass ify the st ructure as a crystalline
hydrate or a hydrated gel.
Water and ammonia - both of which are small
and strongly polar - are the only two molecul es
that can be taken up as a structural part of crystals.
Their small size permits them to penetrate into the
interstices of crystal structures where close packing
Surfaces
of ions is not possible. Thi s is parti cul arl y the case
where the negative ion is large, such as 50
4
(sulphate)
or Si0
4
(s ili ca te). It must be emphas ised that thi s
process is nor to be thought of as a capi ll ary action,
as in the take-up of large amounts of water by clays.
The water molecules are bonded into definite sites
with in the crysta l structure, and the crystal will only
form a stabl e hydrate if ions of the appropriate signs
are ava il abl e and correct ly pl aced to form bonds
with the positi vely and nega tively charged regions
of the water molecul e. We have already mentioned
(secti on 1.2) the abnormal properti es of water arising
from the ability of the mol ecul es to link up by means
of hydrogen bonds; the formation of crystalline
hydrates is an extension of the same behaviour.
Normall y the water molecul es cluster round the
positive ions in the crystal , forming hydrated ions.
This has the effect of making the sma ll positi ve ion
behave as if it were a good deal large r. As a result,
the size difference between the positive and negati ve
ions is effectively reduced, thus making possibl e
simpler and more closely packed crysta l structures .
Water bonded in thi s manner is very firml y held in
many instances, so that hydrati on becomes virtuall y
irreve rsibl e. Cement, for example, retains its water
of crystallisation up to temperatures of -900C.
49
Chapter 7
Electrical and thermal
properties
Elect r ical conducti vit y is not no r ma ll y a const ra mt
in st r uctura l des ign, but ther ma l conducti vit y a nd
ther ma l expa nsio n a re impo rta nt, fo r exampl e in
the fa bri c of a building a nd when esti ma ting diurna l
or seasona l expa nsio n a nd contracti o n. We sha ll
nevertheless fir st bri efl y cons ider electri ca l conduc-
ti vity since it provides a bas is fo r t he mo re complex
ideas of t hermal pro perti es.
7.1 Electrical conductivity
Electri cal conducti vit y is defin ed as t he cur rent per
unit cross-secti ona l area of a conduct or per unit
voltage gra di ent a long t he conductor, a nd hence has
units of reciproca l ohm.m or (ohm.mr
1
Current fl ow
involves the fl ow of elect rons, a nd the conducti vit y
depends on the ease of thi s fl ow. When an elect rostatic
fi eld is appli ed, the e lectrons ' drift ' preferenti a ll y
a nd, being nega ti vely cha rged, thi s d rift is towa rds
the pos iti ve pole of the fi eld. T he resist a nce to d rift
is provided by the 'st a ti ona ry' io ns. T he fo rce F
tendi ng to accelerate each electron IS:
F = Ee (7.1 )
where e is t he cha rge on t he electron a nd E is the
electrostat ic fie ld. The bodil y movement or flu x o f the
electrons J, i.e. the current per unit cross-secti ona l
a rea, ca n be expressed as:
(7.2)
where 11 is the concentra ti on o f free elect rons a nd
V:, is the drift velocity.
In meta ls the met alli c bond (see secti o n 1.2) res ults
in a large concentrati on o f free electro ns and hence
the flu x and the conducti vit y a re hi gh, being in the
ra nge of 0 .6 x 10- to 6 x 10
7
(ohm.mt
1
(bur not e
that thi s is still a ra nge o f a n o rder o f magnitude).
In materi a ls with stro ng io nic or cova lent bo nds,
e.g. ma ny po lymers, the electro ns a re much less
mo bil e and hence the conducti vit y is up to 27 orders
50
o f magni t ude lower; va lues range from l 0 II) to I o-
2
il
(ohm. mt
1
e.g. fo r polyerhelene it is 10-
1
"' (ohm.mt
1

For most appli ca ti ons these conducti viti es a re
effecti vely zero and hence such mat eri a ls a re used
as insul ators . T here is an intermedi a te group of
mate ri a ls with wea ker cova lent bonds a nd therefore
more mo bil e elect rons, fo r exa mple sili con a nd
germa nium. T hese a re the semi conductors that are
so important electroni c appli cati ons, with conducti v-
iti es ra nging from 10-
6
to 10
4
(ohm.mt
1

7.2 Thermal conductivity
T herma l conducti vit y is defi ned as the rate of hea t
transfer across a unit area of a materi al due a unit rem-
perature gradi ent. It therefore has units o f wm- IK-
1

In meta ls, heat t ransfer foll ows much the sa me
genera l principl e as electri ca l conducti vit y a lthough
it is not the bodil y movement o f electrons hut rather
the tra nsference of energy between them by colli sion.
The ana lys is is, however, much more compli ca ted
but it is intui t ively obvious that t he hi gher the tem-
pera ture, the grea ter the excitati on of the electrons
and t he larger the number of colli sions. Typi ca l
va lues ra nge fr om a bout 20 to 400 wm- IK-
1
Since
both therma l and electri ca l conducti viti es have their
origins in the same structural fea tures they are roughl y
pro porti ona I.
T he therma l conducti vit y o f non-meta ls is more
compl ex still , since it involves energy transfer between
the atoms that make up the mater ial. Hearing res ults
in increasing vibra ti on of the nomina ll y sta ti onary
atoms, increasing energy tra nsfe r bet ween them.
Va lues fo r polymers, in whi ch the atoms are rigidl y
held, are t ypi ca ll y in the ra nge 0.2 to 0.5 Wm- IK-
1

Cera mi cs a lso have therma l conducti viti es genera ll y
lower tha n meta ls, but the va lues, whi ch va r y ove r
a wide range, a re dependent on the mi crostructure.
Where thi s is cryst alline, with the atoms ti ghtl y
packed and therefore readil y abl e to t ransmit energy,
the conduct ivit y can be rela t ively high, for example
90 Wm-
1
K-
1
for si li con carbide, but glasses, whi ch
have an a morphous, loosely packed structure, have
va lues of the order of l wm- IK-
1
Not surpri singly,
thermal conductivity is also related to densit y, with
li ght ma terials with an open structure, such as cork
and many timbers, being better insul ators than heavy
more compact materials.
Moisture also has a significant effect on the therma l
conductivit y of porous materials. If the pores are
filled, the water acts as a bridge and since the con-
ductivity of water is many times greater than that
of air, the res ulting conductivity is greater.
7.3 Coefficient of thermal
expans1on
Thermal expansion of a material res ult s from in-
creased vibration of the atoms when they gai n thermal
Electrical and thermal properties
energy. Each atom behaves as though it has a larger
atomic radius, caus ing an overa ll increase in the
dimensions of the material. The linear coefficient of
thermal expansion is defined as cha nge in length
per unit length per degree, and hence has units of
K-
1
The volume coefficient of thermal expa nsion,
defined as the change in volume per unit volume per
degree, is sometimes used; for an isotopic material
thi s is three times the linear coefficient.
As with the thermal conductivity, the coefficient
of thermal expansion depends on the ease with
which the atoms can move from their equ ilibri um
position. Values for metals range from about 1 x 10-
5
to 3 x 10-
5
K-
1
(but are significantly less for some
all oys); most ceramics, which have strong ion ic or
cova lent bonds, have lower coefficients, typically of
the order of 10-
7
K-
1
Alt hough the bonding within
the cha ins of polymers is covalent and therefore
strong the secondary bonds between the chains are
often weak, leading to higher coefficients, typically
of the order of 10-
5
K-
1

51
Further reading for
Part 1 Fundamentals
Ashby MF and Jones DRH. Engineering Materials,
Elsevier Butterwort h Heineman, London
Vol 1 (3 rd editi on, 2005 ). An introduction to proper-
ties, applications and design
Vol 2 (3rd edition, 2005). An introduction to micro-
structures, processing and design
Vol 3 ( '1 993 ), Materials failure analysis: case studies
and design implications
Certainly books to be dipped into. Very thorough
with much detail; as well as the basics and theory,
some ex cellent case studies throughout illustrate the
engineer's approach.
Calli ster WD Jr (2007) . Materials Science and Engineer-
ing, An introduction, 7th edi ti on, .J ohn Wi ley & Sons
Inc, New York.
A comprehensive introduction to the science and
engineering of materials.
Gordon J E ( 1976) . The new science of strong materials
- or Why you don't fall through the fl oor, Penguin
Books, Harmondsworrh, Middl esex.
52
Gordon J E ( 1978). Structures - or Why things don't fall
down, Penguin Books, Harmondsworrh, Middlesex.
Ex cellent and very readable. Read them in bed, on
the bus or on the train. Despite being more than
thirty years old now, they will tell you more than
many hours of library study.
Petrowski H ( 1982). To Engineer is Human, Macmillan,
London.
Petrowski considers what it is like to be an engineer
in the twentieth century and lays some emphasis on
the things that have gone wrong. Not a book for
those lacking in self-confidence but good (a nd easy)
reading.
Cott rell AH ( 1964). The Mechanical Properties o{ Matter,
John Wi ley, New York .
First class, scientifi c and of much wider covaage
than the title suggests. F.s sential reading for any
student wishing to follow up the concepts herein and
highly desirable reading for all students of all branches
of engineering.

Anda mungkin juga menyukai