t24 tN't'R()t)trc't'toN't'() CttRAMtcs

Problems

3. l. Lead orthosilicate forms a glass of density 7.36 g/cc. What is the oxygen density of this
glass? How does this compare with fused silica (density 2.2elcc)2 Where would you
predict the lead ions are going?
3.2. The structure of many multicomponent oxide gÌasses has been found to be heterogene-
ous on a scale of 50 to 500 Â.
(a) How might such heterogeneities be detected and observed?
(b) Discuss the relation between such heterogeneities and the random-network
model of gÌass structure.
(c) How might such heterogeneous structures be explained in terms of liquidìiquid
immiscibility? Include a hypothetical temperature-composition diagram, as well
as free energy-composition diagrams for several temperatuÍes.
3.3. Explain how you can experimentally distinguish among crystalline SiO., SiO, glass,
silica gel, and liquid silica. Explain in terms of the structures of these different forms
of the same composition.
3.4. A certain glass found in recent geological formations is high in SiO,(70+%) and Al.O,
(11.5+%). The alkalis (Na:O and K:O) are minor constituenïs (5%), as are the alkaline
earths (MgO + CaO : 2%). There is almost a l07o weighï loss on heating above about
900"C, and the volatile constituent associated with the weight loss is water. This glass
has the unusual and irreversable property of softening at 850"C, becoming hard at 950
to I 100'C, and then softening at about I 150"C. On cooling the glass has only one fictive
temperature range at 1150'C similar to but higher than most commercial glasses.
Explain the irreversible softening at about 850'C. For your reference, a sodaìime glass
has 727o SiO,, 17a AlzO3,9Vo CaO, 4% MgO, 13% NauO, 17a K.O.
3.5. (a) What is the fictive temperature of a glass?
(b) Why do glasses have coefficients of expansion more like liquids above the fictive
temperature and more like crystalline solids below the fictive temperature.
(c) What would you predict would happen to the fictive temperature of a silica if a
function of Na'O content, NaF content?
3.6. (a) If you were asked to obtain a glass (liquid) at 800"C with the highest moÌe percent
silicon dioxide and you were restricted to only one other oxide in addition to
silica, what additional material would you choose? Give an explanation.
(b) Why does quartz melt at a temperature below the melting point of crystobalite?
3.'1. (a) Sketch a possible bond arrangement in B,O, which follows Pauling's rules and
has CNB : 3'
(b) Enumerate Zachariasen's rules for glass formation. Does the structure you have
sketched fuìfill these requirements?
(c) Mention two distinctly different ways that structure could adapt to addition of
alkali oxide, say, Na'O.
ì,t1, A typical sodalime-silica window glass has a composition of NarO, Vo
('rrO, 7o SiO:.
ì i,) ('l:rssify the following elements as modifiers, intermediates, or-Vo - (glass
network formers
lorrncrs) in connection with their use in forming oxide type glasses: Si, Na, B, Ca, Al,
l'. K. lì4.
4
Structural
Imperfections
In the previous two chapters we have considered structures of ideal
crystals and also structures of some noncrystalline glasses. There are
many properties that are markedly dependent on small deviations from
ideal structures, and we can best approach these structure-sensitive
properties on the basis of departures from the ideal structures.
If we consider a perfect crystal as one with a completely ordered
structure having its atoms at rest (except for zero-point oscillation at the
absolute zero temperature) and with the electrons distributed in the
lowest-energy states, there are several types of deviations or imperfec-
tions which may occur. The first is increased amplitude of vibration of the
atoms about their equilibrium rest positions as the temperature is in-
creased. (These elastic vibrations are nearly harmonic, since the forces
between atoms nearly obey Hooke's law; wavelike solutions for the
motions of the atoms can be quantized and amounts of energy hv, called
phonons, associated with unit quantum excitation of the elastic vibration.
The relationship between phonons and the vibrational frequency is the
same as that between light photons and the vibrational frequency of light
waves.) Imperfections also occur in the electronic energy levels; electrons
may be excited into higher energy levels, leaving vacant positions in the
normally filled electronic energy-level bands, called electron holes. If the
excited electron remains closely associated with the electron hole, the
electron-electron-hole pair is called an exciton, which also may be looked
on as an exçited state of an atom or ion. Finally there are a number of
atomíc defects, including substitution of a wrong atom or a foreign atom
for a normal one, interstitisl atoms, Dacant atom sites, and line imperfec-
tions called díslocations. Finally we might also consider the crystal
surfaces or boundaries between crystals as imperfections, but these are
disçussed separateÌy in the next chapter.
At the outset we must admit that there are a large number of
combinations, permutations, and interactions among solutes, atomic
clefects, electronic defects, dislocations, and surfaces. In the present
125
 )r)UC't't()N't'() ( :liRAMtCs S'l'lll l( "1't I R A l, IM l'l'llÌ ltl'l( "1'l( )NS t27
126 tN'l'rr(

clrlrptcr wc rlcscribc some of these structural imperfections individually

irntl hricfly indicate how and when they are likely to occur. Later, as it
@oWo
bccomes important for the disçussion of particular ceramic processes or
properties, we consider their characteristics more fully. There is no
o @ o W o (.ih. Çn
ceramic material for which our knowledge is complete. More extensive
discussions of structural imperfections are available in the references @ o I lo (ih o (:) o
cited.
o%.Wol o (,,,)

@ o % o Alh o w o
,/,',:!\
4.1 Notation used for Atomic Defects
Several types of structural imperfections are believed to occur in
ceramic materials. One departure from ideality involves the motion of an
atom from a normal site to an interstitial position, as illustrated in Fig.
o % rì@ o @ Lr %
4.1a. This type of disorder, which results in equal concentrations of
@o@o@o% o (b)

o@o@o@o@
@u,@ o @ o %.
o @' ,.o o@o @
@ @, r!_ @.:@ o
O

o % o @ o % o @
@ o @ o @o@ o
(a)
Iìig. 4.1. (a) Frenkel disorder. Ion leaving normal site forms interstitial ion and leaves a
vircancy.
vircant lattice sites and interstitial atoms, is called Frenkel disorder.*
Arrother kind of disorder, which involves the simultaneous production of
lrolh cation and anion vacancies, as illustrated in Fig.4.Íb, is referred to
irs SclroÍÍky disorder.l Ceramic systems are rarely if ever without
irrrprrrilics, and solute atoms may either substitute for host atoms on
rrorrn:rl l:rtlice sites, as in substitutional solid solution, or incorporate on
rrorrrr:rlly rrrroccupied interstitial sites in the host lattice, as in interstitial
solirl srrluliorr.'l'hcse two arrangements are illustrated in Fig. 4.1 c and d.
lrr :rrltlilion lo:rloln locations one must also describe their valence state or,
Io lrt'rrrorc pree isc, the electronic energy levels in the crystal which also
+l lirrrthcl. /. l'1rt,.s.,.t5, ír52 (1926).
l{'. Wrrgrrr'r ;urrl W Stlrottky. Z. Phys. Chem.,Bll, 163 (1931).
(c) (d)
Fig.4.l.(contd.\(b)Schottkydisorder'Equa|numbersofanionandcationvacancies
occur. (c) Substitutional and (d) interstitial solid solutions'
tend to depart from complete order at any temperature above absolute
zero. Such deviations are also affected by the presence of atomic
imperfections and solute atoms.
In keeping track of the various defects which may exist simultaneously
in a single ceramic material, an essential requirement is an adequate
system of notation for describing point defects in ceramics. The Kroger-
Vink notation (Refs. 1 and 2) is most widely used.'In this notation, when
we add or subtract elements from the crystal, we do so by adding or
subtracting electrically neutral atoms and thus avoid making judgments
and decisions about bond types; for application to ionic systems this
t28 tN't'R()t)t r("1'l()N'l'() ('l,lRAMl('s
lcquircs thirt wc scparatcly add or subtract electrons. Let us considcr thc
virritlus conceivable defects and notations for the imaginary binary
c<lmpound MX.
l. Vacant Lattice Sites (Vacancies). When empty lattice sites occur,
they are indicated by the symbols Vu and V' for the M and X sites,
respectively; in the atomic notation, the subscript M indicates a miSsing
M atom. In an ionic lattice such as sodium chloride this would mean the
removal of a Na* ion together with an electron; similarly, Vcr would
indicate the removal of a Cl- ion with the addition of an electron.
2. Interstitial Atoms. In addition to the normally occupied lattice sites
of a crystal structure, there are also interstitial sites. When atoms occupy
these interstitial positions, they are denoted by M' and X.
3. Misplaced Atoms. It is also possible in some compounds for M
atoms to be on X sites (M'); the subscript always indicates the position
of each particular atom in the lattice.
4. Associated Centers. In addition to single defects, it is also possible
for one or more lattice defects to associate with one another, that is, to
cluster together. These are indicated by bracketing the components of
such a cluster; for example, (V'V*) or (XX-).
5. Solutes. These, if present, are coded as to"lattice position in the
same way as native defects; Lw and S* indicate a solute atom L on an Àí
site and S on an X site. L' indicate that the solute L is on an interstitial
site.
6. Free Electrons, Electron Holes. In strongly ionic materials, elec-
trons are normally localized at a particular atom site in a way which can
be described in terms of the ion valence. However, as discussed in
Chapter 2, this is not always the case, and some fraction of the electrons,
denoted e' , may not be localized at a particular site; similiarly there may
be missing electrons, denoted electron holes h ', which are not localized at
a particular atom site.
7. Charged Defects. In insulators and semiconductors we usually
think of the respective species as ions; for example, sodium chloride is
lnlrclc up of Na* and Cl- ions. If we conceive of removing a positively
chirlgcd Na* sodium ion from the NaCl structure, we remove the sodium
irlorn without one of its eleçtrons; as a result, the vacancy has associated
with it irn extra electron with a negative charge which we write as e',
whcle lltc superscript refers to a unit negative charge. If this excess
clcclron is krcalized at the vacant site, as would normally be the case in
Nir('1, wc writc V'(.,,. Similarly, if we conceive of removing a negatively
chrlgcrl ('l clrkrrinc ion, we remove the chlorine atom plus an associated
clcclrorr, lcirving ir positive electron hole, which we write h', where the
strpe rscripl rcl'cls lo u unit positive charge. If this excess positive charge
S'l'Rll( "1't jRAl, I MI'l,lRl,'1,:( "1'l()NS I29
is locirlizc{ irt thc vacirnt silc, lrs wor-rld norntally bc llrc c:tsc lor Nir('1. rvc
writc Vi,,. tn some materi4ls Iess strongly i6nic lhirn Nir('l llrcsc ('x('('s5 (ll
missing electrons, e' or h', may not remain localizctl ltl llte vltç;rrrl rilr'.
the separation being represented by the reaçtions
Y;": \,2*. * e' (,1.1)
Y;,: \/"1* h' (,1..Ì)
By separating the notation for atoms from the notation for electric cltlrt gc.
we avoid the prospect of unintentionally making a priori assutnplions
about the nature of the defects.
With each of the other defect symbols-Vu, V", M,, Mr, (VuVx)-illr
effective charge relative to the host lattice is also possible. Thus Zni'
would indicate a Zn'* ion at an interstitial site which is normally
unoccupied and without an effective change. Substituting a divalent Ca'*
ion for monovalent Na* on a sodium site gives a local electronic struçture
augmented by one extra positive charge and is represented as CaN.. Note
that the superscripts * and - are used to indicate real charged ions,
whereas the superscripts ' and ' indicate effective positive and negative
charges with respect to the host lattice. Other possibilities arise from
nonstoichiometry. In FeO, for example, it is possible to have Fe'* ions in
addition to the normal Fe'* ions. ín this case, the Fe'* ions are indicated
as Fei".
4.2 Formulation of Reaction Equations
As each type of defect and its concentration in a material can be
described in terms of associated energies of formation and other ther-
modynamic properties, it is possible to treat all imperfections as chemical
entities and treat them in a manner referred to as defect chemístry. Defect
interactions may be conceptualized in terms of mass-action equilibria,
thus enabling the representation of such interactions by means of defect
equations. The following rules must be observed:
1. Site Relation. The number of M sites in a compound M"X, must
always be in correct proportion to the number of X sites (l : 1 in MgO, I :2
in UOz, etc.). In maintaining this proportion, however, the total number of
each type of site may change.
2. Site Creation. Some defect changes such as introducing or
eliminating a vacant M site Vw corr€spond to an increase or decrease in
the number of lattice sites. It is important that this be done in a way that
does not change the site relation described in rule l. Defects indicating
 t Jlt lN'l'R()l)tl("1'l()N'l'() ('l,lR^Ml('S
silc crcirtion illc Vv, Vr, M., M*, X., Xx, and so on" Nonsilc-crcating
cntitics are et, h', M,, Li, and so on.
3. Mass Balance. As in any chemical equation, a mass balance must
be maintained. Here it is helpful to remember that the subscript in the
defect symbol indicates the site under consideration and is of no
significance for the mass balance.
4. Electrical Neutrality. The crystal must remain electricaÌly neutral.
Only neutral atoms or molecules are exchanged with other phases outside
the crystal under consideration; within the crystal neutral particles can
yield two or more oppositely charged defects. The condition of electrical
neutrality requires that both sides of a defect-reaction equation have the
same total effective charge, not necessarlly zero.
5. Surface Sites. No special indication of surface sites is used. When
an atom M is displaced from the bulk of the crystal to its surface, the
number of M sites increases.
At this point we anticipate the next section, in which we discuss the
solution of CaClu in KCI to see how these rules apply.
In KCI there are equal numbers of cation and anion sites;to introduce
the two chlorine atoms in CaClz on anion sites, we must use potassium
sites as well as two chlorine sites; since we have only one Ca, we may
tentatively assume that the second K site required for the proper site
relation may be vacant. Considering only atomic substitutions, a possible
solution process is
KCI
CaCL(s) ----------) Ca^ * Vx +zclcl (4.3)
For a strongly ionic material such as CaClz, we may further assume that
the substitutions are fully ionized, which gives us an alternate and more
realistic solution process:
Kcr >
CaCL(s) Cax+ Vl +zclcl (4.4)
wlrich also conserves electrical neutrality, mass balance, and site relation.
s. rrlsu a third possibility, the formation of charged ca'* interstitials,
t'hlolirrc ions on chlorine sites, and potassium ion vacancies, for which we
wrilc
KCÌ
CaCL(s ) -------> Cai' +2V k -t 2Clct (4.5)
wlrich rrlso corrscrvcs electrical neutrality, mass balance, and site relation.
l)ecitlirrg iun()rìlÌ thcsc and other possibilities is a matter for defect
clrcrrrislry iurtl is lhc rnain subject of the following sections.
s't'Rl l("t't rRAt, tMt'ttRtrt,l( "t't( )Ns l_lt
4.-Ì Solid Solrrlions
Of the many types of departure from an ideal cryst:rl, tlrc otrc clrstr.sl lo
visualize is the inclusion of foreign atoms in the host cryslirl. irs illrrslrirlarl
in Fig. 4. lc and d. lf a material crystallizes in the prescncc ol' I'orcig,rr
atoms, they may be almost completely rejected by the crysllrl il'tlrcv
appreciabÌy increase the energy of the crystalÌine form. On tlrc otlrcr
hand, if building them into the host structure in an ordered way lcirtls to ir
large lowering of the system's energy, a new crystalline form dcvclops. lrr
intermediate cases foreign atoms fit into the structure in a random wuy its
the crystal is built up. When this occurs, there is usually a change in thc
cell size with composition in accordance tiith Vegard's law that tlrc
lattice-cell dimensions vary linearly with the concentration of solutc
added.
Solid solutions are stable when the mixed crystal has a lower free
energy than the alternative-building up two crystals of different compos-
ition or building up a new structure in which the foreign atoms are put on
ordered sites. As previously discussed, the free energy is given by the
relation
G:E+PV-T*ç (4.6)
where E is largely determined by the structural energy and the entropy is
a measure of the randomness (probability) of the structure. If an atom
added at random greatly increases the structure energy, the solid solution
is unstable and two crystal structures result. On the other hand, if the
addition of a foreign atom greatly lowers the structure energy, the system
tends to form an ordered new phase. If the energy is not much changed,
the entropy is increased by random additions so that the solid solution has
the lowest energy and is the stable configuration. Different rules for the
likelihood of solid solutions being stable in particular systems are
applications of these general principles to specific cases.
Two examples of stable solid solution are shown in Figs. 4.2 and 4.3.
For the MgO-NiO system, both end members have the sodium chloride
crystal structure, and a complete series of solid solutions occurs. The
schematic free-energy-composition diagram for 1500"C is shown in Fig.
4.2b. The lowest-free-energy phase for all compositions at 1500"C is the
solid solution rather than any ordered structure or the liquid. For the
MgO-ALO: system shown in Fig. 4.3, the end members have different
crystal structures, and a third possibility is the formation of an inter-
mediate compound. Experimentally it is found that for a 50:50 mixture of
MgO and AlzO,, the spinel phase has the lowest free energy and at 1750"C
in the intervals q-b and c-d two solid solutions of fixed composition but
 172 tN't'R()t)UCï'toN't'() (:liRAMlcs S'l'Rtl(l'I'tlR^1. lMIll{R1'lil(l'l'l()NS Lì.1

3000

Liquid
ô 2500
CJ \
I o
Periclase'SS Spinel
'/ and liquid
-
Corundum
and \ \' and
(a) =
E @)Fo
f
liquid \ Z- --\----\-.lliquid
8_ zooo o
E Ë Periclase SS \ Spinel
o o
F F
b cd.
Periclase SS Spinel
and and
ipinel corundum
0 u.z 0.4 0.6 1.0
Meo Nio
Mole Íraction NiO 0 0.2 0.4 0.6 0.8 1.0
Meo 41203
Mole Íraction Al2O3

ü
o
(b) c
o U
o o
r (b) 5
o
o
L Spinel SS

Meo Nio
Mole fraction Nio
lììg.4.2. (c) of MgO-NiO. (b) Schematic representation of free-energy-
Phase diagram
0 0.2 0.4 0.6 0.8 1.0
Mso Al2o3
cornposition diagram of MgO-NiO system for ?<2000'C. Mole fraction AÌzo:
Fig.4.3. (a) Phase diagram of the MgO-AÌzO. system. (ó) Schematic representations of
vlrrilrble amounts are in equilibrium with each other (the compositions in free-energy-composition diagram of MgO-Al,O, for 1750'C. I:
e rlrrilibrium have the same values of chemical potential, - : (#)ro. ,hu,
is, ;rrr cc;ual slope, a common tangent, for the free-energy-composition shown in Figs. 4.2 and 4.3 represent this kind of substitution. For
t'rrvcs). In each of the regions MgO-a, b-c, and d-AI:O: a different example, crystals of
magnesium oxide frequently contain appreciable
vrrr irrhle composition solid solution is the lowest-free-energy equilibrium amounts of NiO or FeO with a random distribution of Ni'z* or Fe'* ions
slt uclltlc. replacing Mg'* in the crystals so that the final composition of the crystal
Sorrrt' r'rrlcs irnd generalizations useful for predicting solid-solution can be read Mg',Ni,O, as shown in the phase diagram in Fig. 4.2. A
bcltirvior :rrc tlisctrssed in subsequent sections. similar series of 'solid solutions exists in the system AlzO,-Cr,O, (rubies
Sullslllulftrrrtl Solid Solution. Substitution of one ion for another is are 0.5 to 2Vo CrrO. in Al'O'), ThOz-UOz, albite-anorthite, and many of
conun()n irr tlrc l'olrnrrtion of ceramic crystals. The solid-solution phases the spinels. In some systems there is a complete series of solid solutions
 l.l4 lN'l'R()l)ll("1'l()N 'l'() (ll,lltAMt('S
l'olrne rl bclwccn cnd nìcnìbcr-s ( I'ig. 4.2). Howevcr, in most syslcms only a
lirrritctl nrlnber of foreign atoms can be added to the substitutional solid
solution (Fig. a.3); an excess above the soÌubility limit at a given
tenrperuture results in the formation of a second phase.
'['here are several factors determining the extent of substitution that can
take place in solid solutions, and a number of rules expressing these
factors have been devised. These rules reflect variations in the free
energy, which is made up of several terms. Since the free energy is a
function of temperature, a series of free-energy vs. composition curves
for each temperature can be drawn similar to the ones shown in Figs. 4.2
and 4.3. Lowering of the free energy resulting from an increase in entropy
requires that there be at least a small solubility of foreign atoms in any
structure. Factors which allow extensive substitution are as follows.
1. Srze Fecron. If the sizes of two ions differ by less Íhan about l5Vo
conditions are favorable for the formation of substitutional solid solu-
tions. If the relative ion sizes differ by more than 157o, substitution is
generally limited and is usually less than a fraction of 17o. This factor is by
far the most important for ionic compounds.
2. Ver-eNcy Facron. If the added ion has a valençe different from
that of the host ion, substitution is limited. It can occur, as indicated
below, but other structural changes are also required to maintain overall
electrical neutrality.
3. Cnnlrrcet- AFFrNrry. Tht greater the chemical reactivity of the
two crystalline materials, the more restricted is solid solubility, since a
new phase is usually more stable. For oxides this restriction is usualÌy
implicit in factors''of ion valency and size.
4. SrnucruRE TypE. For complete solid solubility the two end
members must have the same type of crystal structure. For example, TiO,
could obviously not form a complete series of solid solutions with SiO.
(see Chapter 2 for descriptions of these structures). This does not,
however, restrict limited solid solution.
On the basis of these factors, an estimate of the extent of substitutional
solitl solr.rtion to be expected can usually be obtained. For oxides, the
rrr:r.jor l'lctors are the relative ion sizes and valencies. Although different
iorr sizcs rlcfinitely preclude extensive solid-solution formation, valence
rlill'clcrrccs cirn frequently be made up in other ways. For example,
cxlcrrsivc solitl solution among Mg'*, Alt*, and Fe2* ions is common in
cliry nrirrcr':rls hirving the montmorillonite structure. The charge deficiency
tlrrc lo lerplircing lriv:rlcnt Alt* with divalent Mgt* or Fe'* is made up by
cxchirngcrrble iorrs irtlsrlrhed on the surface of the small clay particles.
Similarly, srrbslitrrtiorr ol'lhc Sio. ions by Al'" ions in tetrahedral coordina-
tion in kaolirritc lctrtls lo u charge difference which is made up by
s't'Rt t(''t't rRAt. I MI't,tRtrt,t( "t't( )Ns l.Ì5
exchangeablc ions adsorbccl ()n thc pirl lic'lc rtttl:rr'r^s. 'l'hcsc
substitutional-solid solution-effects itt'c lirrgcly tc'spottrtl'lr lor tlte oh-
served base-exchange propertics anrl lltc lrbility lo lot trt 'ìl;rl)l(' suslìcn-
sions with the clay minerals. In tttttclt lltt sluttc wirv, nr:rrv ol lltc
aluminosilicate structures iÌrc tlcrivirlivc slluclurt's ttr wlrtt lt ;rrr Al''
replaces an Sio" in the crystirl lrrrrl lrrt lrlklrliol lrlkrrlrtrt't';ullr ton lrlr rr{o lur
expanded interstitial positiorr. Solirl solrrlions lrcr;rrcrrlly or't'tn rrr lvlur'lr rr
number of different ion srrbslilrrliotts lr:rvt' l:rkt'rt plrttt'.
There is yet another way in wlriclr tlre t't;rtrrl virlt'rrey rc(lrrt'rìrt'rrl t:rn
be circumvented, and this is by lcaving irrr occ:rsiorrrl :rlonr silr'v:rt':url.
The magnesium aluminate spinel structurc, Mg,Al:().', Ilrs lrr:t'rt tlt'st'ril,r'tl
in Chapter 2. Extensive substitutional solid solution ()ccurs bclwccn llris
material and AlrO.. This corresponds to the substitution ol'Al'' iotts l'ot'
some of the Mg'" ions. In order to maintain electrical neutrality, cach lwrr
Al'* ions added must replace three Mgt* ions, leaving one vacant latticc
site. The end member of this complete series is y-AlrOr, which has a
face-centered cubic packing of oxygen ions as in spinel in a structure
corresponding to Alsr:O, with one-ninth of the total cation sites vacant.
This kind of solid solution in which different ion and vacant atom sites
are added in the right proportions to give electrical neutrality is not
uncommon. For example, additions of CaO ÍoZrOz form a solid solution
with the cubic fluorite structure in which Ca'* are substituted for Zro*.
Each time this is done, an oxygen ion site is left vacant to maintain the
cation-anion site relationship of I :2.
Cao(s) ''o' , Cal,+ oo + Yõ (4.7)
Similarly, additions of LazO: to CeOz or ZrO, and CdO with BirO3 give
rise to substantial numbers of vacant sites in the anion array. In contrast,
additions of MgCL to LiCl, AlzO' to MgAl,Oo, and FezO: to FeO lead to
vacant sites in the cation array. Although in the systems just mentioned
these effects are large, similar effects occur in samples for whiçh solid
solution is extremely limited; for example, CaClz dissolved in KCl. In this
system the solid solubility is less than 17o.
In refractory oxides with close-packed crystal structures, the tempera-
ture dependence of the solubility is often large (see Fig. 4.3). Although the
solubility of Al,O' in MgO is several percent at 2000'C, it decreases to
only 0.0lVo at a temperature of 1300'C. At high temperatures the TS
product dominates the free energy for solution; but as the temperature is
lowered, the large enthalpy term (-E + PV) for the formation of vacant
lattice sites dominates (Eq. 4.6).
The most direct evidence for the formation of vacancies is obtained by
t.tó tN'l'R()t)tJ("t'toN't'o cl:R^Mtcs

tlclcrrninirrg the latticc constant for a structure in order to comparc

lhcorctical with measured crystal density. Several examples follow. tta?sk,,*,
A crystal with the composition Zro,rrCao.,rOr.ss which crystaÌlizes in the
fluorite structure has an X-ray pattern indicating that in a unit celÌ having
an edge of 5.131 Â there are four cation sites and eight anion sites. If 5.8
the cation sites are all filled, and oxygen ion vacancies occur to give
the proper site relationship, there are 4x0.15x40.08/6.03x l0'?3gCa, 4Ò.

4x0.85x91.2216.03 x10"g Zr, and 8x1.85/2x16.0/6.03x10'zrg O in

\'% Lo^
È 5.6
135.1 A'for a total of 5.480 g/cm'. [The weight of each ion per unit cell is
given by (number of sites) (fraction occupied) (atomic weight)/(Avoga-
dro's number).1 This is in excellent agreement with the directly measured
value of 5.477 glcm'.
H

&"
1600"C isotherm in
cubìc field (ZtOz-CaO)

Figure 4.4a shows the change in density calculated from the X-ray I

lattice parameter as compared with the directly measured density; the o = Calculated from X-ray data
upper curve is calculated for another possible model in which the extra r = forcnometrically determined

cation occupies an interstitiaÌ position. The anion vaçancy structure (Eq. 51015202530
4.7) is found to be in accordance with these calculations. However, for Mole fi CaO
(a)
samples quenched from 1800'C, the data in Fig. 4.4b show that the higher
temperature equilibrium corresponds to a different sort of defect struc- 6.2
ture. (This is perhaps our first clear warning of the possible divergence
between observations at room temperature and the equilibrium situation
at high temperatures.) When CaCL is added to KCl, the density change is i"tbn rlur.titu, roou,
found to be in accordance with cation-vacancy formation, as indicated in !

Eq. 4.4 and Fig. 4.54. When Al.O. is added to MgO, the density change is
also found to be in accordance with cation-vacancy formation. These data 5.8 I
are shown in Fig. 4.5b.
!o,^
A charge balance can also be achieved by changes in the electronic
structure, as discussed in Sections 4.8 and 4.9. ü
Interstitial Solid Solutions. If added atoms are small, they can go on
{^ 5.tt
\
interstitial sites in the crystal to form solid solutions. This type of solution
1800'C isotherm in
\% I

is particularly common with metallic bonding, in which added H, C, B, 5.4 cubic Íield (ZrOz-CaO)
rrrrrl N fit easily into interstitial sites. I

'l'he ability to form interstitial solid solutions depends on the same

l':rclors, cxcept for structure type, that apply for substitutional solid o = Calculated from X-ray data
5.2
solrrlions--siz-e, valency, and chemical affinity. The size effect depehds on I = Pycnometrically determined
lhe oligirrlrl h<lst crystal structure. In face-centered cubic structures, such
;rs Mg(), llrc only available interstitial sites are tetrahedral sites sur-
51015202530
Mole ft CaO
rourrrlctl by l-orrr oxygen ions. In contrast, in TiOz there are normally (b)
vrrr'rurl octrrlrctlnrl interstices; in the fluorite structure there are larger
Fig. 4.4. Change in density on the addition of CaO to ZrOz in solid solution quenched from
irrlo'sÍiccs willr eighlfold coordination; in some of the network silicate (a ) 1600"C and (b ) 1800'C. At 1600"C each Ca'?* addition is accompanied by the formation oÍ
slnrclrrres srrclr trs thc zcolites the interstitial positions are very large. a vacant lattice site. At 1800'C there is apparently a change in defect type with composition.
'l'hcrcl'olc, wc cxpccl thc order of ease for forming interstitial solid From A. Diness and R. Roy, Solid State Communication,3, 123 (1965).

r37
 tN't'R()t)t r("1't()N'l'() ('t,tRAMl('s S'l'Rl l( "1'l I RAl, I MlDl,:lÌl'l :( "1'l()NS l.ì.,

+0.000.l A A(\

4.20 5.10

E 4.00 5.08 Oxygen interstrtials

U
o
u0
ç fi:so ^ 5.06
o
! @
E > 5.04
:c o
ô
3,60 t_
É
oa 3.40 Ô 5.02

3.20 5.00

3.00 4.98
010 20 40 o24 681012
(o) CaFz Mole %YFt Yzo: Mole ?6 ZrOz
(a) (b)
3.60

Fig.4.6. Additions of (a) YF, to CaF, and (á) ZrO,to Y,O. give rise to the formation of
Calculated for interstitial anions.
AÌ2O3'-241's+ Yús+3Oo
U -----^ --_<-
t\_

'. ---r--.- :--_

-----:-
c 3.56 Experimental
o -r---
Õ density data are given in Fig. 4.6b and favor Eq. 4.8 (oxygen interstitials)
3.54
MgOI234
Mole fi Al2Os
(b)
F'ig.4.5. When (a)CaCl: is added to KCI from H. Pick and H. Weber, Z. Physìk,128,409
(1950) and (b) AlrO. is added to MgO, vacant cation lattice sites are created. From V.
Srrrbican and R. Roy, J. Phys. Chem. Solids,26, 1293 (1965).
solutions to be zeolite >ThOr>TiOz>MgO, and this is found to be the
cll|ic.
'l'lrc :rdclition of ions on interstitial sites requires some associated
clìiu'lÌc birlirnce to maintain electrical neutrality. This can be accomplished
by vlrcrrrcy formation, substitutional solid solution, or changes in the
clectronic slructure. All occur. For example, when YF. or ThFo is added
to ('irl;,, ir s()li(l solution is formed in which Tho* or Y'* substitute for Cat*
and iÌt thc slrrrrc tinrc F ions are placed on the interstitial sites so that
electrical ncrrtlirlity is rnarintained (see Fig. 4.6a).
Likewisc, thc ldtlition <tf ZrOz to Y'O' creates oxygen interstitials to
maintain the proper site relation and electrical neutrality. The measured
rather than Eq. 4.9 (yttrium vacancies).
Density inçrease: 2ZrOr(s) "'o' , 2Zr* + 3Oo + O,í (4.8)
Density decrease: 3ZrOr(s) "'o' , 3Zrl + 6C,o+ vi @9)
In many silicate structures the additional charge due to an interstitial
Bet*, Li", or Na* is balanced by the replacement of some of the Sio" by
Al' in solid solution.
4"4 Frenkel Disorder
Several different types of disorder can occur in crystals without the
addition of any foreign atoms. The particular type of disorder in which
equal numbers of vacant lattice sites and interstitial atoms occur is called
Frenkel disorder* and is illustrated in Fig. 4.1a. As the extent of this kind
of disorder is increased, the structural energy is increased, but at úe same
time the entropy (randomness of the structure) increases. At higher
+Frenkel, op. ciÍ.
t40 tN't'R()txt(:'t't()N 't'o ('lil{AMt(:s S'l'Rll("1'tlR^1, lMl,l,lRl,'1,:("1'l()NS ldl

tcnrpcraturcs thc higher entropy form, the disordered form, is favored to e llrrrgc irr lrtltlition lo thc corrlìgrrrirtiorrirl crrtropy is rrcgligiblc, srtclr llrirl
givc the minimum free energy required for thermodynamic stability. n /
The free energy of the crystal can be written as the free energy of the (.1, l'l)
perfect crystal AGo plus the free energy change nÂg necessary to create n
^ft\
N:e*P\-z,r)
interstitials and vacancies less the entropy increase AS. that accrues from 'l'he added entropy As results mainly from lattice strains ancl accornpiury
the different possible ways in which the defects can be arranged: ing changes in vibrational frequencies resulting from the introduclion ol'
the defect. There is less than full agreement on theoretical estimâtcs, irrrrl
AG:AGo+nlg-TAS. (4.10)
experimental observations indicate that although values of exp (As/2h)
o
The configurational entropy AS. is proportional to the number of ways between l0 and 100 are most common, values as small as 10 and as largc
in which the defects can be arranged, W, and is given by as l0o have been reported. That is, estimates of the absolute values ol'
defect concentrations are subject to much uncertainty. In contrast, the
klnW (4.t1) relative change in defect concentration with temperature, and therefore
^S.: Ah, is much more amenable to measurement.
For a perfect crystal, the N atoms which are indistinguishable can only be
placed in one manner on the N lattice sites which are distinguishable. The Silver bromide has been well studied and forms Frenkel defects on the
configurational entropy is cation sublattice at moderate temperatures:
Ago"sAgi +Vl" (4. l8)
Âs.:t(Ìn#j:khr:o (4.12)
By defining IVl"] as the fraction of charged vacant silver sites IVl*]:
However, if there are N normal sites and an equal number of interstitial n,/N and the fraction of silver interstitials as [Agi]: n,/N,
sites, the interstitial atoms /li câÍì be arranged in N!/t(N - n,)tn,!l ways,
and the vacant sites n, can be arranged in N!/[(N -n")ln"l] ways. The IVÁ"ltAe;l- "^o (-#) (4.1e)
configurational entropy for these randomly (noninteracting) arranged
defects is or since IyÁ*]: [Agi] for stoichiometric AgBr,

^s":rr.[*4;r"L] t*+;] (4.13) tAe;l:".0(-#)

Another general way in which we can look at the formation of a small
(4.20)

Since Stirling's approximation for large numbers yields ln N!:

N ln N - N and since n' : n, : n, the entropy is fraction of defects is by application of the law of mass action. Creation of
a vacancy and an interstitial ion in an ionic crystal can be written as a
: 2klN ln N -(N - n) ln (N - n) - n In nl (4.14)
chemical equation:
^S.
rrrrrl the total free-energy change is (normal ion) + (interstitial site) : (interstitial ion) + (vacant site)
aG : + n!,g -2kr (Y)] Ago"ÌV,:Agl +Vks (4.21)
aGo
[r r" ffi)* n rr' (4.15)
For this equation the mass-action constant is
At e <prrilibrium the free energy is a minimum with respect to the number
lAeillVÁ_l
.l rlt'lt'cts: rrrt,, ({$) :0. By differentiating Eq.4.l5 and setting it ^':1v,][.AeoJ
(4.22)
\ ôn /,.,
crlttirl lo zcro irnd then taking N - n - N, we have Ag :2kT ln (N/rz), or for small concentrations of defects, IV,]: [Ago"]: 1; thus

il "*o (- #):".0 (#) .-r(-#) (4.16)

lAg,llVi"l: Kp or tAg;l: VK (4.23)

In lltc rrpl,licirtiorr ol'Flq. 4. l6 it is sometimes assumed that the entropy The concentration of Frenkel defects is determined by the energy of
t42 lN'l'R()l)tl("1'l()N'l'O ('1,:RÀMl('S S'l'Rl l( "1'l IRAl, I MPl,lR l,'l,l( "1'l()NS l.t r

l'orrnirrg it v1Ìcitncy itnd interstitiitl ion and by the tenlpcraturc us givotl In lrigh pol:rrizirbility irnd conscqttcnlly irre bcttcr lrblc lo [rc rrccorrrnrotlrrlt'rl
t,,cl.4.l 6. F'or energies of formation in the range of I to 6eV and irtlo intcrstitial sites.'lhis is true for AgBr, l'or cxiurrplc, in wlrt'lr
temperatures between 100 and 1800'C, concentration of defects may strbsterntial numbers of interstitial Ag* ions occur along with irssociirlcrl
range between a few percent to 1 part in 10o', as shown in Table 4'1. The vacancies. The Frenkel formation energy Lh : 1.1 eV; the p,re cxporre rrl ilrl
equilibrium concentrations at room temperature are always small. They tcrm exp(As/2k) is in the range of 30 tq 1500.
can become appreciable at higher temperatures if the energy of formation An oxide system in which Frenkel defects are formed is YrO,, l'or'
for the defects is not too large. which we can write
Oo+ \4 : O't' +K" : [O;'][Vil
V,ó (4.24)
Table 4.1. Defect Concentrations at Different Temperatures
When this relationship is combined with the solid-soÌution behavior
: rl- #]:".0 #)
# - .*o l-#l "*o ll*'1"*
(- shown for solid solutions of zirconia in yttria (Eqn. 4.8), that is, the
formation of oxygen interstitials in a concentration determined by the
solute concentration, we see that the concentration of vacant oxygen sites
Defect must be simultaneously diminished. That is, the Frenkel equilibrium, the
Concentration l eV' 2eV 4eV 6eV 8eV
product of interstitial and vacancy concentrations, remains in force.
n/N at 100'C 2xtoI 3xl0'a 1x10'?7 3xl0o' lx105o
n/N at 500'C 6x l0 o
3x107 1xl0'' 3 x l0-to 8x10'?7 4.5 Schottky Disorder
n/N at 800"C 4x l0 l 2xl0-r 4xl0-'o 8x10'5 2x10'e
n/N at 1000'C I x 10-2 lx10-o lx10-8 lx10r'? lx1016 Another particular kind of disorder which occurs in ionic crystals is the
n/N at 1200'C 2x 1O-' 4x100 1x107 5x10" 2 x 10-l'g presence of both cation and anion vacancies in thermal equilibrium (Fig.
n/N at 1500"C 4 x 10-'? 1x100 2x106 3x10' 4x10't 4.lb). Just as for Frenkel disorder, energy must be expended to form
n/N at 1800'C 6 x lO-'z 4x103 1x105 5x108 2x10 ro
vacant sites, but the increased entropy makes a finite vacancy concentra-
n/N at 2000"C 8x10'z 6x10r 4x10-5 2x 10 7
lx10' tion favorable for a minimum free energy as the temperature is raised.*
: If we consider Schottky disorder in a crystal such as NaCl, we can
'1 eV 23.05 kcal/mole. derive the concentration of vacant sites in exactly the same way as was
done for Frenkel disorder in Eqs. 4.10 to 4.16. If Âg is the energy required
Although the configurational entropy change for forming FrenkeÌ to form a pair of vacant sites by moving two ions to the surface,
defects can be calculated from statistical mechanics, the energy change in
putting an atom into an interstitial position depends to a great extent on
the structure and ion characteristics, as discussed in the last section. #:.*o (-#):..0 (#).*o (-#):.-o ( :r+) A2s',
Calculation of this energy is difficult because there is a large correction The defect conçentration increases exponentially with temperature, and
tcrm required for the ion polarizabilities, making calculations for alkali the vacancy concentration is given as shown in Table 4. l.
halides difficult and for oxides nearly impossible. An example of such a In principle, the enthalpy to create point defects in ioniç crystals can be
cirlculation is discussed in the next section. calculated from a Born-Haber cycle, which involves (1) the creation of
F,-or alkali halide crystals with a sodium chloride structure the energy charged imperfections in the crystal by removal of ions to the vapor, (2)
Icc;r-rired to form an interstitial ion plus a vacancy is of the order of 7 to the transformation of the gaseous ions into atoms, and (3) the formation
tl t'V, so that they do not occur in measurable numbers. For crystals with of the compound from the gaseous atoms. The results for NaCl are in
llrc Íìrrorite structure there is alarge interstitial position in the structure, good agreement with experiment:ï
thc rrrnorrnt of energy necessary for forming interstitials is lower, Ah:
:
,_

L,h*rc:2.12 eY Ah"o. 2.02-2.19 eY

Lt e V t'or caFz and exp (#) t- about 10a, rhus Frenkel defects are
*Wagner and Schottky, op. ciÍ.
c()rÌrìì()rì. 'l'hcy are only prominent in crystals containing ions that have a ÌF. G. Fumi and M. P. Tosi, Discnss. Faraday Soc., 23,92 (1957).
 I.t4 lN'l'R()t)f("t't()N't'() ('t,tRAMI('S S'l'Rll("1'tlRAl, IMI'l,lRl'l,l("1'l()NS r.t5

Schotlky tlisorclcr c()rÌlrn()nly ()ccurs in the alkali halicles at elevatccl for thcir fornritliott ltl tlccttt.
pilirirr borrrrtlirry, tlisloc:rlíon, or frcc sttrfitcc
tcIttpcratures. lior cxirrnplc. in MgO, magnesium itlns must leave their latticc posiliorrs
ljor oxides the calculations which involve Coulomb interactions, Born :urtl rrriglatc to a surface or grain boundary; thus,
rcpulsions, and polarization effects are subject to substantial uncertain-
ties. However, the energy for formation of vacancies in oxides is two to Mgu"* Oo ê Vi.l"+ Và' + Mg"'.'* O""" (4.26)
three times as large as for the alkali halides, which means that equilibrium Sincc magnesium ions and oxygen ions on the surface form a layer over
Schottky disorder does not become important in oxide crystals until very othor ions previously located at the surface, this equation is equivalent to
high temperatures are reached. Thus, the intrinsic number of defects the usual form of the Schottky equation:
caused by thermal effects is almost always smaller than those from
solutes, as discussed in the sections on solid solutions and non- null ê ViL,+ Và' (4.27)
stoichiometry. Table 4.2 contains some experimental data and estimates and affects only the kinetics, not the equilibrium state.
for Schottky and Frenkel defect formation energies. As is true for Frenkel defects, it is the product of the vacancy
concentrations that is fixed by the Schottky equilibrium. FromEq.4.27,
Table 4.2. Some Defect Energies of Formation K" : IVil"][vô'] (4.28)

Energy of When AlrO, is added to MgO as a solute, cation vacancies are created, as
Formation,, Preexponential Term - shown in Fig. 4.5b. This effect combined with the Schottky equilibrium
Compound Reaction Àh (eV) exp (As/2k) (Eq. a.28) requires that the concentration of anion vacancies be simul-
taneously diminished.
AgBr Agr* -+ Agl + V|, 1.1 30-1500
BeO null ê V'i,.+ V;,' -6 ?
Mgo null ê V{Ls+ V;' -6 ? 4.6 Order-Disorder Transformations
NaCl null È Y("+ Vil ',)_) a 5-50
LiF null c Viì+ V; 2.4-2.7 r 00-500
In an ideal crystal there is a regular arrangement of atom sites with a
null ê V'é^+ V;,' periodic arrangement of atoms on all these positions. In real crystals,
CaO -6 7
however, we have seen that foreign atoms, vacant sites, and the presence
CaF: F' <2 Y;+Fi 2.3-2.8 100
Ca.. ê V3"+Cai' -7 ,Ì of interstitial atoms disturb this complete order. Another type of depar-
null r= V'é"+2V; - 5.5 ,!
ture from order is the exchange of atoms between different kinds of
UOr oo ê yô'+ oï 3.0 ? positions in the structure, leading to a certain fraction of the atoms being
Uu è V'J" +U;"' - 9.5
,!
on "wrong" sites. This disorder is similar to the other kinds of structural
null a Vli'+2Vò' - 6.4
,ì
imperfections we have disçussed, in that it raises the structure energy but
also increases the randomness or entropy so that disorder becomes
increasingly important at high temperatures. This leads to an order-
A Íìnal but significant principle must be remembered when considering disorder transition between the low-temperature form which is mostly
conccrrlrations of interstitial ions and lattice vacancies at a particular ordered and the high-temperature form which is disordered. This kind of
lelrrpt'r;rlrrre. Since the equilibrium defect concentrations were derived transition is commonly observed in metal alloys. It also occurs for ionic
lor t'r[rilibrium conditions, sufficient time must be allowed for equilibrium systems, but these are more likely to be either completely ordered or
I'lrt' r.t'rrclrctl. Since this usually involves diffusional processes over many completely disordered, and transitions are only infrequently observed.
irtornic tlirncnsions, equilibrium at low temperatures may in practice never There are some similarities and also differences between order-disorder
hc rr'irt'hr'tl. 'l'hrrs the high-temperature defect çoncentrations may be transitions and high-low polymorphic transitions (Section 2.10).
rqrrt'rrt'lrt'tl in wlrerrr the crystal is cooled, as illustrated in Fig. 4.4. The degree of order can be described on a long-range basis as the
Arr irrrPoll;rrrl tlistinction between Schottky defects and Frenkel defeçts fraction of atoms on "wrong" sites, or on a short-range basis as the
is llrirl Schottky rlcÍ'ccls require a region of lattice perturbation such as a fraction of "wrong" atoms in a first or second coordination ring. For our
 I 'lír lN'l'R()l)tl("1'l()N'l'() ('t,ltÌ^Mr('s
l)il1'p():ics iì (lcsclipti()n ol'l()ng-rangc ordcr is suflìcient. [,et us consider
lwo kincls ()f atorns, A and B, in a lattice having two kinds of sites, cv and
lì' with the total number of atoms equal to the number of sites N. If R. is
thc fraction of a sites occupied by the "right" A atoms and Rp is the
fraction of B sites occupied by the B atoms in a perfectly ordered crystal,
all the atoms are on the "right" sites and R. : RB : l. If there are an equal
number of A and B atoms and a and B sites, then for a completely
random arrangement only half the Á atoms are on a sites, R" : ll2, and
only half the B atoms are on B sites, RB:112. We can define an order
parameter S, whiçh is a measure of how completely the a sites are filled
with Á atoms, in such a way that for complete order, S equals one, and
for complete disorder, S equals zero, as
R"-+ \-w"
!ì:-
t-; t-+
where w- is the fraction of a sites containing the wrong B atoms.
If only a small degree of disorder occurs, we can derive the dependence
of order on temperature and on the energy Eo required for the exchange
of a pair of atoms in exactly the same way as was done for Frenkel
disorder in Eqs. 4. l1 to 4.16, with the result
w" wa / E"\
R":&:exp\-2kT/
For increasing amounts of disorder, however, more of the neighbors of a
"wrong" atom will also be "wrong," so that there is an increasìn g ease of
disordering (a lower value for E) as the amount of disorder increases. In
the simplest and nearly satisfactory theory of disorder,* it is assumed that
the energy required for disorder of a pair of ions is directry proportional
to the amount of order, that is,
Eo : EoS
'l'lris is an oversimplification because
the value of Eo depends on the
slr'r'l-rirnge order even when the long-range order is constant. More
slrlisl'ircl.ry relationships may be derived by considering the effect of
slror't-r'irrrgc <lrdering on the energy of disorder.ï In either case,
as the
tlis.r'tle r irrcrcases with temperature, owing to the cooperative nature
of
lhc plrerrorrìcnon, the rate of disorder also increases until complete
*W. L. lìr'rrgg:rnrl Ii. .t. Williams, proc.
R. Soc. (London), 145A, 699 (1934).
l'H. A. Ilcthc, /,rrrr'. /ì. Soc. (l,ontlon), l5A. 552 ú935).
S'l ll I r( "1't I l{ Al, IM l'l,lRI,'l,l( "1'l()NS t47
disorder is tcn, lrcrl ;rl :otttc ltitt.tsition tcmpcrature (Fig. 4.7). Generally
the num$ct'ç1 ,'\ ;rrrrl ll rtlottts itre not equal, so that relationships derived
must inclrrtlc llrrr vlrn:rltlc:rs well. An excellent review of the entire
subject is givrrr lry lr. (1. Nix and W. Shockley.*
DisOrrle r lrrrrrslonttittions are common in metals in which the nearest
neighbgrs irr:rrr Áli alloy can be ordered or disordered without a large
changc irr e rrcrgy. In ionic materials exchanging a cation with one of its
coortlirtrrtiorr polyhedra of anions is so unfavorable energetiEally thirl it
nevcr ()ccll's; all order-disorder phenomena are relatctl t., c*irrlso,r pttsiliotts
in thc cirtign substructure or anion positions in thc itttitln sttl.rsllttclttlc. ltt
this clse the energy change is one of thc sccrttttl cttttt'tlittltliotr; llrc litsl
cg1;r{ination remains unchanged. lf thc ittttrtts itt'c itbtttll lltt' sittttc si/t' :ttttl
charge, the energy from the secontl crxrt'tlirtitliott littg ol'likc cltlrtgctl iotls
is almost entirely coulomhic. If rrll thc cirliott silcs ilr lltt'sltttclttt'e ittr.'
equivalent, the energy chirngc ol'tlisoltle:r'is snlrll, irrrtl lltc tlisot'tletetl
(4.29)
form is the only 6nc th:rt ()ccurs: this is lruc, for cxitnrplc, ip sglid
solutions of Nio-Mgo ancl Al:o,-(lr-:or. (tsut at sulÌìciently low tempcra-
.I'S product of Eq. 4.6 is sufficiently small, phase
tures at which thc
separation is to be expected in almost all systems, as discussed in chapter
8.) In addition, there are many materials which are almost completely
disorclered, even though the valencies are different as long as only one
kind of ion site is involved. For example, both Li'FezOa and LizTiO, have
the sodium chloride structure with random distribution of the cations on
(4.30) the cation sites. These two compounds also form a continuous series of
solid solutions not only with each other but also with Mgo. In a similar
way, in the compound (NH.).MoO.F. it is impossible to distinguish
between the positions of the O'- and F- ions; that is, in this compound
there is disorder on the anion sites. No ordered form of these compounds
is known.
The most important examples of order-disorder transformaìion in
ceramic systems occur in materials having two different kinds of cation
sites, for example, the spinel structure in which some cations are on
(4.31)
octahedral sites and some are on tetrahedral sites (see Fig. 2.25); various
degrees of order in the cation positions occur, depending on the heat
treatment. It has been found in almost all ferrites having the spinel
structure that the cations are disordered at elevated temperatures and the
stable equilibrium low-temperaturefor.m is ordered. The change of order
with temperature follows a relation such as that illustrated in Fig. 4.7.
Another kind of disorder may result when there are unoccupied sites
available in the ordered structure. This is the case for AgzHgIo. In the
+Rea. Mod. Phys., 10, I (1938).
 t.ilt lN'l'lt()l)ll("1'l()N'l'() ('l,IÌ^Mt('S S'l'Rtl("1'llRAl, lMPl,ll{1,'l,l("1'l()NS t.t.,

'l'lte l'r'rrctiorrirl rnol:rr corrccrrtnrliorr oÍ' vlrclrrrcy p:rils is

liivt'rr lry
: (,li:")
,".x]!#jj z exp(#') - z exp(*)".n
('r,ì)
cooooaco ooco cooc Where Z is the distinct number of orientations ol' thc pirir wlrrr..lr
aoaoacoc ocao ocoo contribute to the configurational entropy (Z :6 for V.'., Vi,,, plrir.s). Sirrt't.
coooooco oooo oaoo the product of sodium ion and chlorine ion vacancies is íixctl hy tlrr.
ocococoo ocooooac
I cocoooco cooo cooo Schottky equilibrium,
V) ocaoococ aoooooca
o
õ
E t Yl.ltYá,1 :
"^o
(?) "-o (#) (4. ì4)

and t(ví"vó)r : Z exp(*)".n (*)".0 (-44#&) ro,'r

0
T" the concentration of vacancy pairs is a thermodynamic characteristic of
Temperature + the crystal (a function of temperature) and independent of solute concen-
Fig. 4.7. Disorder as a function of temperature. Complete disorder is reached at a critical trations.
temperature T.. The coulombic energy of attraction between oppositely charged defects
is

ordered low-temperature form three-quarters of the available sites are

filled in an ordered way. A typical order-disorder transformation occurs at
- Lh,,:yR (4.36)

a temperature of about 500'c. Above this temperature there is complete
disorder, with one Hg and two Ag ions randomly arranged on the four
cation sites available,
4.7 Association of Defects
when schottky or Frenkel defects are present in an ionic crystal, there
is a coulomb force of attraction between the individual defects of
opposite effective charge. The electrostatic interaction between defects
oí' opposite charge can be described by the Debye-HuckeÌ theory of
t'lcclrolytes (Refs. 2 and 7). However, within the precision of availabÌe
lhcory to take into account repulsive force, rearrangements of nearby
:rlorrrs rrnrl polarization effects (and considering the paucity of experimen-
l:rl tlrrt:r) it is preferable to focus on the major contribution of the
t'lcclr oslrrlic interaction at small distances and consider the association as
rt'srrlling in thc formation of a complex defect, for example, a vacancy
lrrrir corrsistirrg Of an anion vacancy and a cation vacancy on nearest-
rrciglrlror silt's irr ir material containing Schottky defects. we can write for
lltt'Iolrrr:rliorr ol'srrch a vacancy pair
Y,!. + Vó,: (V("Vè,)
where eiei are the effective charges (electronic charge x valence), rc is the
static dielectric constant, and R is the separation between defects. This
relationship is clearly very approximate, but it gives about the right values
and leads to useful insights. For sodium chloride the cation-anion
separation is 2.82Â, the dielectric constant is 5.62 such that the energy
required to separate a vacancy pair is
- Lh,o: :0.9 eVl (4.37)
where 4.8 x 10-t0 esu is the electronic charge. A more precise calculation*
gives a somewhat lower value than this, 0.6 eV. For the combination of
two vacant sites to form a vacancy pair it is a reasonable assumption,
supported by some experimental data, that the preexponential term,
exp (As"o/k) is near unity.
For oxide materials, in which vaca+cies have larger effective charges,
the energy gained by the formaticin of vacancy pair's is larger, as
illustrated in Table 4.3. Hence vacancy pairs in ceramic oxides should be
more important than for the better-studied alkali halides. In Fig. 4.8 we
have calculated on a speculative basis the expected concentrations of
(4.32) *lìumi and Tosi, op. ciÍ.
 r50 tN'l'Rot)u("t't()N't'o cliRAMt('s S'l'Rt l( "1't I RAl, IMPl,lRl,'l,l( "1'l()NS Itt
'l'ablc 4.3. Approximate Coulombic Defect Association Energies
Calculated from Eq. 4.36
(This simple calculation overestimates the correct value by an
uncertain amount, perhaps 50 Ío 100%)

R (A) - Ah* - qrqj /xR (eV) c

Yô,l
NaCl 5.62 c
Vk^-Vt, 2.82 0.9 P
Cari. - V(" 3.99 0.6
o -rn
_'

c ,-
CaFr 8.43 6
-l)
Fi- Vi 2.74 0.6
Yò"- V',é^ ú
3.86 0.9 I -14
Yc"- Y6.-Yò" 3.86 0.4

Mgo 9.8 2.8

V{1s- V'o' 2.tl 2.8
mpMgO
Feú, - Vü, 2.98 1.0
Feür- VíÍr-Feü* 2.98 0.5
-18
1000/? cK)
Nio 12.0
yií,- vó' 2.09 2.3 Iìig. 4.8. Calculated (for NaCl) and estimated (for MgO) individual and associated defect
YKr - NiÌìr 2.95 0.8 concentrations for samples containing I ppm aliovalent solute.
Niir - yilt-NiÌir 2.95 0.4
Li*r - Nii' 2.95 0.4

UOz -15 where Z is t"he distinct number of solute-vacancy-pair orientations

oif - vô' 2.09 0.5 (Z : 12 in the NaCl lattice for neighboring cation sites), and it is
reasonable to assume that the preexponential term involving vibrational
*esu'/cm x6.242x 10tt: eV. entropy is near unity. An estimation of the energy of association based on
the coulombic attraction (Eq. a.36) gives results for a number of systems,
Schottky defects and vacancy pairs in sodium chloride and in magnesium as illustrated in Table 4.3.|n contrast to the intrinsic nature of vacancy
oxide. pairs, the concentration of solute-vacancy associates depends strongly on
The electrostatic attraction of oppositely charged defects also leads to the solute concentration.
irssociation between solutes and lattice defects. For the incorporation of As the temperature of a solute-containing crystal is further lowered, a
clrlcium into sodium chloride we have the reaction temperature is reached which corresponds to the soÌubility limit where
precipitation of the solute occurs. At temperatures below this level, the
CaCL(s) _x:95 Cai"+ Yi," +2cl3l (4.38) solute concentration remaining in solid solution in the crystal is deter-
mined by the free energy of the precipitation reaction. For sodium
l'he l lcc cncrgy of the system is decreased by the association reaction chloride containing CaCl', we can write
CaN.* Vi'": (Ca;.V(,^) (4.3e) CaN"+ V&"+2clc, (4.4t)
= CaCL(ppt)
I'or wlriclr wc cirn write a mass-action constant
i"#Htrh:.^o (-#) *
"*n
(-ïF) G 42)

or [vi,.l* "^o (*#) (4.43)

 r52 lN'l'lÌOl)tl("1'l()N'l'O ('l,lR^Ml('S S'l'Rl I( "l't IRAl. t Ml,lLRt,'l,l( "t't( )NS I i.ì

lrig.4.l wc scc that thc solubility of aluminum oxidc in

Sirrriltrlly, l'ronr
Mg() dccrcirses from arlmost lOTo at 2000"C to less than 0. lVo at |5OO"C. Oorrrlrrclron lr,rrrrl
corrcsponding to a heat of solution of about 3 eV. For the precipitation
reaction we can write
I
Mgv" * 2Alús + \/ ii, r 4oo MgALO.(ppt) (4.44) )) irI'
=
I vii"][Alú*]' - "-n (*ff) (4.4s)
since [AIú"] :2[Ví,"] (4.46)

rvii"r o (1)"'""0 (.#Ë) (4.47)

such that the defect concentration in the crystal is approximately deter-
mined by the heat of precipitation as defined by these reactions. Since the
total solubility also includes defect associates, Âhoo, given in Eqs. 4.43 and
4.47 is not equal to the negative of the heat of solution.
4.8 Electronic Structure
In our ideal crystal, in addition to all atoms being on the right sites with
all sites filled, the electrons should be in the lowest-energy configuration.
Because of the Pauli exclusion principle, the electron energy levels are
limited to a number of energy bands up to some maximum cutoff energy at
0'K which is known as the Fermi energy -EÍ(0). At higher temperatures
thermal excitation gives an equilibrium distribution in some higher energy
states so that there is a distribution about the Fermi level Ey(T) which is
the energy for which the probability of finding an electron is equal to
one-half. Only a small fraction of the total electron energy states are
affected by this thermal energy, depending on the electron energy band
scheme.
The different temperature effects observed for metals, semiconductors,
iurtl insulators are related to the electronic energy band levels (Fig. 4.9). In
rnclirls, these bands overlap so that there is no barrier to excite electrons
I. highcr energy states. In semiconductors and insulators a completely
íille rl crrcrgy band is separated from a completely empty conduction band
.l' lriglrcr clectron energy states by a band gap of forbidden energy levels.
In intlirrsic scmiconductors the energy difference between the filled and
e nrl)t y brrrrtls is not large compared with the thermal energy, so that a few
e lcclrons irlc thcrmally excited into the conduction band, leaving empty
Metal semiconductor
lntrinsic lnsulator
lrig.4.9. Electron energy band levels for metals with partly filled conduction band,
irrlrinsic scmiconductors with a narrow band gap, and insulators with a high value for E..
the conduction band is completely devoid of electrons and the next lower
bzrnd of energy is completely full, with no vacant states.
In an intrinsic semiconductor, each electron whose energy is increased
so that it goes into the conduction band leaves behind an electron hole, so
that the number of holes equals the number of electrons, p : n. 'lhe
nomenclature usually employed is to indicate the positive electron-hole
concentrationby p, that is, p:lh'1, and the negative excess electron
concentration by n, that is, n : fe'1. In this case the Fermi level Ei is
halfway between the upper limit of the filled band and the lower level of
the conduction band.
The concentrations of the intrinsic electronic defects can be calculated
in a manner analogous to that described for Frenkel and Schottky defect
concentrations. In this calculation the thermal randomization of electrons
is related to the probability of a valence electron in the full band having
enough energy to jump across the energy Eap E" into the conduction
band. Because of the Pauli exclusion principle, Fermi statistiqs are
required to calculate the distribution. The concentration of free electrons
is
n : [s'): (4.48)
N. 1*expt(8.-Et)lkTl
where n. is the number of electrons, per cubic çentimeter, N. is the
ilensity of available states in the conducton band,

clccll.rr posilions (electron holes) in the normally filled band. In perfect *- _ rl2rmïkTl"': 1g,"7sm, at T:300.K (4.4e)
insrrlirlors thc grrp hctween bands is so large that thermaÌ excitation is h, l-
instrÍÌicicnl to clrirngc lhc electron energy states, and at all temperatures E is the energy level at the bottom of the conduction band, and Ey is the
l5.t lNï'ROl)ll('ï'l()N 'l'() (llilRAMl(lS S'l'Rl l( "1'1,RAl, IMPl,lRl,'l,l( "l'l()Nl; t5t

liclrrri cnclgy, irs illustrittcd in tsig.4.9. The Fermi energy represents the 'l'rrblr .1.4. llllrrl (Jl;r" ulrrl Agrproxilrrtlt ('oltt't'ttlt'ttllolts of l'llct'lrottl trlltl llok'ri irr
chcrnicirl potential of the electron and at 0"K is at the center of the band I'urc, Skrichft trnclrlt' Solirls
gitp.
A similar relationship holds for electron holes in the valence band, and rr lr)"'t'tt,
| ,l'i I
I
t'lt'r'l t otts/r'ttt'
when the concentration of electrons and electron holes is smaÌl, these Mt'llrng
expressions reduce to ('ryst:rl ti, (cV) lìoottt l'ctrtlr t(xxl" l( l'oittl 'l'r'rrrlr ("li )

:fe'):
n
#: "*o [-tEï-E ] (4.50) K('l
Nrr('l
7
L\
IO "'
lo"
,t0
'l
Iro
/(l
I0,t1,
l() /.1
l0 '' l0 l()' l(rll
p:th l:ff-"*o[-@-ۓ] (4.51)
('tt I ir
U( )r
l0
5,1 l(ì '' r (I'
"
l(|r' Ìt5()
()l{0
Nio ,1 .) l(ì r" l(Í' l()r' I

where N, is the density of eÌectron-hole states in the valence band, Al,( ), / .,1 lo " .lo I0" I Ì02
Mg{ ) tt l0 ''' 0.0 I I0" I l7l
N, : rlu#u'l'/': 1s,n7cm, at T: 3oo.K (4.s2) lii(), tt
l.l{
I0
t0'
n"
0.0I
l0''
l0' 1941
705
Agllr 10"

The product of the electron concentration per cubic centimeter times the
('tlS+ 2. 13 l0' l0'' l0't 1773
('rl( )+ 2.t 20 10'' l0'u 1750
hole concentration per cubic centimeter is given by 10" 10t' 1750
7,rr( )* 3.2 10"'
to
(iarO, 4.6 l0 10t 10" 2000
,.r, : (ry)' e"p ( - tt43
^f )"'
<^ t l0-t'
4 t.i F- t2 l0-8
-&) FerOr* 3.1 10t 10t' l0to 1733
: lo" exp (--t4)cm-6 at 3oo'K (4.53) Si 1.1 l0'o l0'u l0tt 1693

where E" = E. - Et, h is Planck's constant, and mt, mt are the effective
masses of free electrons and electron holes in the crystal lattice, usually
somewhat larger than the mass of a free electron (in oxides nr * is
approximately 2 to 10m and in alkalide halides n * approximately equals
ll2m). In a pure crystal the concentration of electrons equals the
concentration of electron holes. tü/hen solutes or nonstoichiometry
affects the electron energy levels, the ratio of electrons to holes changes
brrt, as is the case for Frenkel and Schottky equilibrium, their product
rcmains constant.
Thc magnitude of the energy band gap covers a wide range, varying
Í'rorrr irs small a value as 0.35 eV for PbS to a value of about 8 eV for
slrrhlc oxides such as MgO and AlzO.. In Table 4.4 some characteristic
vlrlrrcs of the band gap and the resulting concentrations of electrons and
holcs irr prrrc materials are illustrated.
l.irllicc tlcfccts, atom vacancies, interstitial atoms, and solute atoms are
silcs oÍ' pcrlrrrltations to the energy states respresented in the band
schcrrrc in lrig. 4.9 and result in localized energy states in the band gap.
lf an ircltlctl clcctron or hole is loosely associated with an impurity site,
we can apploxirnatcly calculate the energy to add or remove an electron
by assuming that thc clcctron is bound to the defect in a way similar to the
'Most of the data are based on the optical band gap, which may be larger than the
electronic band gap.
*Sublimes or decomposes.
hydrogen atom, except that it has an effective electron mass m ï and is
immersed in a medium with a dielectric constant x. The energy is assumed
to be proportional to the first excited level in the hydrogen atom:
E : B 6(YlX;)'"u (4.s4)
where z is the ionization state of the defect. For the alkali halides m ï is
aboú ll2m and the dielectric constant is about 5. so that the energy
required to ionize a sodium or chlorine atom vacancy in NaCl or to excite
an electron from a neutral calcium solute atom iS estimated as about
0.3 ev. If we assume that the excess electron is located at the nearest-
neighbor distance, we can calculate'that the energy of ionization as
calculated for ion associates (Eq. 4.36) is
QtQz
rR
 r56 lN't'R()Dtr('ï'toN't'o (:llRAMt(.s S'l'lÌl l( "l'l lÌ A l,
I I M lf l,llÌ l,'l,l( "1'l( )NS t57

which givcs lhc cnergy rcqr,rircd to ionize a sodium or chlorine ion

viìcancy in NaCl as :rbout 0.9 eV. Finally, in the case in which the electron lV'. ll V, 'l cxl')
is bound within a narrow orbit, interactions between the valence electron
and the impurity center are decisive and the ionization energy of the
center is determined by specific quantities such as the ionization energy
: exP
t-i#l
or electron affinity, polarization terms, the local electrostatic potential, and Yiltv;,1:".0
and so forth. In general it is a priori unknown which of these cases [Ë*]
applies. l- 2.2 eV I
--.^pl-
In showing the electron energy levels at defects within the band gap, we l',r I

always follow the convention of indicating the nature of the level by That is, for the pure materiul thc rlrtio oÍ'uniorrizcd to ionizctl vucanclcs rs
labeling it as if occupied. Neutral levels near the çonduction band may be given by
ionized to free an electron and are called electron donor levels. Neutral
levels near the valence band may be ionized by accepting electrons lY*l lV,,l [ 4.] cV ì
and are called acceptors. In a sample of potassium chloride (Fig.4. l0), IvilItz,',| t*PI t'r I
(4. s6)

vacant chlorine sites may be ionized with the expenditure of about l.g ev;
calcium atoms substituted on potassium sites may be ionized with the
expenditure of about I ev. when a neutral site such as a potassium atom
vacancy is ionized, approximateÌy I ev is required. Potassium chloride is
a wide-band-gap material with E*:7 eV. The difference in energy
between the lowest donor level and the highest acceptor level is 4.2ey
this is the energy gained from the ionization of a neutral chlorine atom
vacancy and the transfer of its electron to the potassium vacancy, which
then has an effective negative charge. Thus we can write for KCI that the
Schottky equilibrium for atomic unionized defects is given by
As a conscqucncc, in widc-band-gap materials the concentration of
neutral dcfccts is nrany orders of magnitude smaller than the concentra-
tion of ionized defects, a fact which we have assumed in Sections 4.2 to
4.6. For materials which have a narrower band gap, particularly the
transition elements with unfilled d orbitals and the higher atomic weight
elements, the defect energy levels approach the center of the band gap,
are near the Fermi level, unionized or partially ionized defects occur, and
the electron energy leveÌs are both more complicated and frequently more
controversial.

4.9 Nonstoichiometric Soli{s

Conduction band (K 4s") Conduction band (Mg 3s") In elementary chemistry and in many analytiçal çhemical techniques
we rely on the idea that chemical compounds are formed with constant
yi.8 1eV oïtuuJ ar il,
fixed proportions of constituents. From a consideration of structure
eV
-Pcal Do-w2 eY vacancies and interstitial ions we have already seen that this is only a
J-u; I special case and that compounds without simple ratios of anions to
/eV
4
I

2eV=8. - En- E. ,Y8 tV

rI
'o
cations, that is, nonstoichiometric compounds, are not uncommon. An
example for which the stoichiometric ratio does not even exist is wüstite,
Eo 4.5 eY:= tE.
"_E^ -EA
i
I having an approximate composition of Feo."rO. This material has the
[ ,;,
I
sodium chloride structure; samples of different compositions were
+,
I

t:. lcV " ';"

v;;s
v;;, studied by E. R. Jette and F. Foote,* with the results shown in Table 4.5.
__l
l"/,%', Eo;v I .5 eV
0.45eV
,t p
Vfic For samples of different composition, Íhe unit-çell size and the crystal
density were determined. The departule from stoichiometry might be
I rlllrl lr,rrtrl ({lì 3pc1
Filled band (O 2p6l
accounted for either by oxygen ions in interstitial positions (to give
úr) l(( il
(ó) Mgo
FeO'.or, for example) or by vacant cation sites. Since the density inçreases

l;rr:. ,l 10. l,,stnrr:rtcrl e lt'clrolr cncrgy levels in KCI and MgO. *E. R. Jette and F. Foote, J. Chem. P/rys., l, 29 (1933).
 I
lN'l'R()l)Ll("1'l()N'l'() ('1,:RAMI('S I
S'l'lll l( "1't I lÌ /\ l, IM l'l lR l,'1,:( "1'l( )NS l5(,
I

'l'ablc 4.-5. Composition and Structurc of Wüstitc" i

I
I
O,, V;; I ;O'(s) I lt" L
(.1 rH )

Density
Composition AtomVo Fe Edge of Unit Ceil (Â) G/cm') where c' is an addecl electron in the structure. Similarly, llte ithst'ttt't'rtl rrtt
clcctron normally present in the stoichiometric structurc cottcsgrottrls lrt
FeosrO 47.68 4.290 5.613 tn electron hole or a missing electron h'
Feo.mO 47.85 4.293 5.624
Feo.s:O 48.23 4.30r 5.658 2Fer. * !rOXs>: 2Fei. + Oo + Vi. (,1.5())
Feo.q+sO 48.65 4.310 5.728
I
Source. E. R. Jette and F. Foote, J. Chem. Phys., l, 29 (1933). ;o^c): oo + v'í.+2h' (4.60)

as the oxygen-to-iron ratio decreases, the changing structure must be due

Oxides in general show a variation of composition with oxylÌcrì
pressure, owing to the existence of a range of stoichiometry. Stablc
to cation vacancies. As more iron vacancies are created, the density
oxides having a cation with a preference for a single valence state (a high
decreases, as does the size of the unit cell.
ionization potential) such as AlzO. and MgO have very limited ranges of
To compensate for the smaller number of cations and consequent loss
of positive charge, two Fe2* ions must be transformed into Fe'* ions for nonstoichiometry, and in these materials observed nonstoichiometric
effects are very often related to impurity content. Oxides of cations
each vacancy formed. From a chemical point of view, we may consider
having a low ionization potential can show extensive regions of non-
this simply as a solid solution of Fe,O, in FeO in which, in order to
stoichiometry. For reactions such as those illustrated in Eqs. 4.57 to 4.60
maintain electrical neutrality, three Fe'* ions are replaced by two Fe'" and
a vacant lattice site, that is, Fer'* Vr.O: replaces Fe3O3, in which V.. we can write mass-action expressions and equilibrium constants and
represents a vacant cation site. To a first approximation the Fe'* ions may
relate the atmospheric pressure to the amount of nonstoichiometry
observed. For example, cobaltous oxide is found to form cation vacan-
be considered as distributed at random. Similar structures are observed
cies:
for FeS and FeSe, in which ranges of stoichiometry occur corresponding
to vacancies in the cation lattice. Other examples are Co, I
):
"O, Cur--O, tOíC
O" + V'é"+2h' (4.6t)
Ni' ,O, 7-AlzO., and y-FezO.. Similarly, there are compounds with
vacancies in the anion lattice such as ZrOz-, and TiOz Also oxides For this equation the equilibrium constant is given by
".
occur in which there are interstitial cations such as Znrt,O, Cr:**O:, and
Vé"lttr'l'
Cd'*,O. Compounds with interstitial anions are less common, but UOz*"
is one.
Ìr: tO"ltPa]- (4.62)

AII these structures can be considered, from a chemical point of view, Since the concentration of oxygen ions in the crystal is not significantly
solid solutions of higher and lower oxidation states, that is, FezO. in FeO, changed ([O"] : 1) and the concentration of electron holes equals twice
[.] ,O* in UO', and Zr in ZrOr. However, the electrons associated with the the concentration of vacancies, ZfV'c"] : lh'1,
virlcncy differences are frequently not fixed at one specific ion site but
lcirtlily migrate from one position to another. The idea that this electron is IV'é"]- P"i'" (4.63)
intlcpcnrlcnt of any fixed ion position can be indicated by representing it Similarly, when ZnO is heated in zinc vapor, we obtain a nonstoichiomet-
scplrrrrlcly in the reaction of formation of the nonstoichiometric com- ric composition containing excess zinc,Znt*"O, for which we iânrwrite
porlrtl. lrol thc reaction of TiO, to form TiOz,- ptus Zn(g): Zni + 3 (4.&)
JOr(S)

: Á:
[Zn,lÍe'l (4.65)
2Ti.,.i + Oo 2Tii; + Và' +:O,G) (4.s7) o
rZ^
is equivlrlcn( to lZn',1- P'/3 (4.66)
t60 tN't'tÌ()t)u("1't()N 't'() ('t,:RAMt(,S S'l'Rtl("1'tlRAl, lMl'l'lRl.'l'l("1'loNS lítl

( )r' sirrrilirrly l'or thc oxygcn prcssurc dcpcndencc (7n(g) + I l2Oz <t ZnO) it liltt'itt ttl;tltott',ltt;t
ctlrrirliorrs irr lr loglrlillrrrric l'olrtt sttclt lltirl tlrt'r't' is
t'o
(4.67) bctwccn tcrnrs lrnd ntirkc lhe :rssttrngrliott lltitl ott r':tt'lt sttlt'ol lltr'
lZn,)- Prt, onc oÍ'thc conccrrtrirliorts is so tlotttittitttl itr lo ttt;tht'
neutrality cquation
An essential consideration in each case is the nature of the defect the othcrs negligible. At a givcn lctttpcrittttlc wt' ('itll llrclr pt clìill (' ;r
(substitutional, interstitial, vacancy) and the degree of ionization. For diagram of the Iog concentrirtion ol'eirch spccit's irs;r l'ttttt'liolt ol llrr' lop',
example, the zinc interstitials in ZnO might be doubly ionized: oxygen pressure; the log concctttrttliotì ol t':tc'lt s;rt't'it's ;llìl)(':ll \ ;15 ;l
Zn(e): Zni' *2e' (4.68) straight line with a slopc crlrrcspotttlirtg lo lltt'()xVH('tt lìl('\\llrt'rlcpctl
dence within a given nctrlltrlity cotltliliott.
which would give a diff erent concentration-partial-pressure relationship: Let us consider an oxitlc rrrirtcliirl in wlrit'lr otyp.r'rr lirt'rrkt'l rlt'lt't'ls
occur, the oxygen c()nlct'ìt vtrlit's ovcl lr uurg('ol sloit'lttotttt'ltV. ttrttl lltr'
tzn;'l x fe'f x P'.(g)''' o Pn. ''u (4.6e)
electron and electrotr-lrolc corrccnllirliort is ;rpprce irrltlt'. Wc t:ttt wt llt'
The correct model choice requires experimental data. Since the electrical
conductivity is proportional to the concentration of free electrorrs and O,, - Oï l. V"' l()1llv;l Ki1 (.1.70)
therefore to the concentration of charged zinc interstitials, the electrical-
conductivity data in Fig.4.l1 support our choice of singly charged zinc o,,: jo,{r)+vi'+2c' LVò'lle'l'P,,,''t - Kt (4.1 l)
interstitials (Eqs. 4.64 to 4.67) as the actual defect mechanism.
null: e '+ h' [e'llh'l: K (4.12)
-2.7 toittft'l'
lo,,rr:o'!+zh' ffi: K, @.13)

;--650"
Actually only three of these four equations are required, since the two
representations of the oxygen addition are equivalent; that is, K'Ku :
b
U
o
,\ K,'K'i. The neutrality equation is
^\, 2lo'!l + [e' l : 2[V )+ [h']. (4.74)
'\- à'

but if the energy gap is such that the concentration of electronic defects at
the stoichiometric composition is substantially greater than that of
-2.7
0.6 r.0 1.4 1.8 2.2 2.6 3.0 Frenkel defects, we can replace this representation with the simpler
log po, (mm) requirement that n : p. When the concentration of electrons is fixed, the
t''
l;ig. .
4. I I Conductivity oï ZnO as a function of oxygen pressure at 650"C. From H. H. oxygen vacancy concentration is proportional to Po, according to Eq'
l|:rurrrhrrch and C. Wagner, Z. Phys. Chem.,B22, 199 (1933). 4.71. Similarly with the electron-hole concentration fixed, the oxygen
interstitial concentration is proportional to Po.n'''; at the stoichiometric
So Í'irr we have only considered the major species present over a limited composition the oxygen interstitial and oxygen vacancy concentrations
r:rrrgc oÍ'stoichiometry. For a more complete description of the defect are equal.
slrrrctrnc it is necessary to write down all the equilibria expressions At a sufficiently high oxygen pressure the concentration of oxygen
invrlvirrg intcractions among vacancies, interstitials, electron energy interstitials increases to a point at which the neutrality condition can be
k'vt'ls, rrrrrl chcmical composition, including the influence of solutes and approximated by tOïl : ll2p.At a sufficiently low oxygen piessure the
irrrpruitics, :rrrtl solve this set of equations together with relations expres- concentration of oxygen vacancies with a positive effective charge
sirrg t'lt'clric:rl,cllrlge balance, site balance, and mass balance. In an increases to a point at which the neutrality condition can be approximated
:rpl.rr()xirrrrrle rrrclhorl proposed by Brouwer* we can write the mass-action by tVô'l : U2n.
An alternate possibility occurs when the concentration of Frenkel
+l'ltilip:; llr..r. /{r7r., 9. ì66 ( l9-54). defects is substantially greater than the concentration of intrinsic elec-
 lNï'ROl)lrc't'toN't'o (:F)RAMtcs S'l'Rll("1'l I RAL lMPl'Rl,'1,)(:'l'l()NS I ír.l

hy lhcrrnorlynirrnics.'l'hcy ntay bc forn.rcd in vitriotts witys httl rrtt'lrctlrrlrr

to':t= rp ,, Neutralrty condition lrcst visr.ralized l'ty considering the plastic dcformatiort ttl :r r'tyslttl,
Ivït = in lvi'ì = toiì toi't= ip illustrated in Fig.4.l4. Deformation occurs by relativc shcitrirtg rtl'lwtt
\"*o;ll o"rlti-l parts of a crystal with respect to each other along a pkÌnc, tltc slilr lrlirttc,
I I Ioíl parallel to a plane in the lattice. If it were necessary to citrry orrl tltis
j j shearing process by one simultaneous jump of all the atoms on lhc slilr
o tvi;l
?
;l -- plane, an excessively large amount of energy would be rcqttit'ctl lrttrl
plastic deformation would need much higher stresses (about 10" psi) lhirrr
are actually observed. Instead, it is believed that deformation occttrs by lr
wavelike motion (Fig. 4.14), with the lattice distortion limited to a narrow

logpo, +
logpo, (b)

(a) - 2[U['l = [e'l th.l = 2ív;;l

Fig.4.l2. Schematic representation of concentration of oxygen point defects and elec-
tronic defects as a function of oxygen pressuÍe in an oxide which, depending on the partial
pÍessure of oxygen, may have an excess or deficit of oxygen. In (a) K' > K'j;in (b) K'i> K,
(reference 9).

tronic defects; that is, Agí<E-. The relative defect concentrations as a

function of oxygen pressure are illustrated in Fig. 4.12.
Relative to the actual situation existing in real ceramic materials, Fig.
4.12 has been simplified by ignoring the presence of associates and the
influence of impurities, which are often decisive. For an oxide MO in
which Schottky equilibrium is predominant for the pure material, we have
shown in Fig. 4.13 a schematic representation of the substantial changes
which result from the introduction of impurities. It will be well worth- lvifl,ln'la, pfi
while for the reader to apply our earlier discussion to the careful
interpretation of Figs. 4.12 and 4.13. Although these Brouwer diagrams
clearly indicate the strong influence of nonstoichiometry and the expected
oxygen pressure dependence of the defect structure, we should warn
rrgirin that they are largely schematic; precise values for all the necessary
cqrrilibrium constants are not available for any oxide system.

4. lll l)lslocations
All thc imperfections we have considered thus far are point defects. 1ss
"'Qz
p! _
Anothcr kintl oÍ imperfection present in real crystals is the line defect Fig.4.l3. Schematic representation of defect concentrations as a function of oxygen
calletl ir dislocitlion. These are unique in that they are never present as pressurefor (a) a pure oxide which forms predominantly Schottky defects at the
equilibrium impcrfcctions for which the concentration can be calculated stoichiometric composition.
 lír4 tN't'Rot)lr("t'toN't'() (,t,tRAMt(.s

l?frv,çFfrl* o o o o U
(_)
olo o o
x? o o (, o o o (J o/o o o
exl
C.

o o o o oc o \_, o o
c
\
lo o o o o o o o o o \J
o
c
o IFfrVííFrt]*
ú
o
tro oo aì o o o o o o o
\t LJ o o o o o o o o
Electrical neutrality conditions
(a)
(o) le') = 21Y6'1
(b) te'l = IFfil
@ Frtl=zÍvMl
(d) 2\ví;l = lh'l
tog pë,_
Fig.4.13(contd.) (á)AnoxidewhichformsSchottkydefectsbutcontainscationimpurities
[F.l> K""' lmodified from (reference 8)].

region. The boundary between the slip and unslipped parts of a crystal is
called the dislocation line. The dislocation line can be perpendicular to the
direction of slip, an edge dislocation, or parallel to the direction of slip, a
screw dislocation. The structure of an edge dislocation is equivalent to
the insertion of an extra plane of atoms into the crystal. This can be
illustrated by means of a soap-bubble raft (Fig. a.l5).
A characteristic of the dislocation is the Burgers vector b, which is a
rrnit slip distance for the dislocation and is always parallel to the direction
ol' slip. The Burgers vector can be determined by carrying out a circuit
corrnt of iìtoms on lattiçe positions around the dislocation, as illustrated
l'rrr thc two disloçations in Fig. 4.16. If we start at a point Á and count a
givcn nrrnrbcr of lattice distances in one direction and then another
nunrhcr ol' llrttice distances in another direction, continuing to make a
complctc circuit, we end up at the starting point for a perfect lattice. If Fig.4.14 (a) Pure edge and (à) pure screw dislocations occurring during plastic deforma
therc is a tlisloc:irtion present, we end up at a different site. The vector tion.
betwecn thc stirrting point and the end point of this kind of circuit is the
Burgers veçtor. l"or :rn edge dislocation the Burgers vector is always

165
 tNï'R()t)U("t't()N ï'() ('tcRAMt('s
N
Fig. 4.15. Dislocation in a raft of soap bubbles. From W. L Bragg and J. F. Nye, Proc. R
Soc. (London), Al9O,474 (1947).
perpendicular to the dislocation line. For a screw dislocation the Burgers
vector is parallel to the dislocation line.
In general, however, a line defect or dislocation is not restricted to
these two types but can be any combination of them (Fig.4. 16). Any
dislocation in which the Burgers vector is neither parallel nor perpendicu-
lrrr to the dislocation line is called a míxed dislocation and has both edge
irnrl screw characteristics. Dislocations can terminate at crystaÌ surfaces
brrl ncver inside the crystal lattice. Thus they must either form nodes with
olhcr dislocations or form a closed loop within the crystal. Such loops and
rrrrtlcrs irrc often observed (Fig. 4.17). At a node the vector sum of the
lìrrrgels vcctors must be zero.
'l'he originul source of dislocations in crystals is not completely clear.
No tlislocirli()ns iÌre present at equilibrium, since their energy is much too
grcut in cornplrrison with the increase in entropy they produce. They must
bc introducctl in ir nonequilibrium way during solidification, cooling, or
handling. Possiblc s()urces include thermal stresses, mechanical stresses,
S'l'Rl l("f I I RAl, IM I'l,lRl,'1,)( "Il()NS
Fig. 4.16. Combination edge and screw dislocation. Burgers vector b shown for pure screw
and for pure edge. Dislocation line connecting these is shown.
precipitation of vacancies during cooling, and growth over second-phase
particles.
Crystal dislocations, which were first postulated independently to
account for plastic deformation by Orowan, by Taylor, and by Polanyi in
1934, were not directly observed in real crystals until 1953. In that year
precipitates formed along the dislocations (this technique is called deco -
ration) were observed in silicon under infrared lighting. Etch pits formed
by chemical etching where the dislocation lines touch the crystal surface
were also first used to study dislocations in 1953. In the late 1950s various
X-ray topographic techniques (Lang and Berg-Barrett) were developed.
Transmission-electron-microscope techniques which were developed in
the late 1950s provide perhaps the best means for observation. In the
transmission electron microscope if the wave vector g of the electron
beam and the Burgers vector b are such that g. b:0, one ob5erves the
dislocation lines disappear and thereby determines the Burgers vector.
Transmission-electron-microscope and etch-pit techniques have been
used to characterize dislocations and to measure their velocities resulting
from applied stresses.
The concentration of dislocations is measured by the number of
dislocation lines which intersect a unit area. Carefully prepared crystals
 (rll lN'l'R()l)tl("1'l()N'l'() ('l,lRÀMl('S
I
Íìig. 4.17 . X-ray topographs of sapphire samples representing (a ) a node formed by three
brrsal dislocations. The directions of Burgers vectoÍs are denoted by arrows- (b) Several
sirrglc hclical turns indicated by the arrows. (c) A single spiral turn shown around the letter
r. 'l'his type of turn is usually larger than single helical turns shown in (b). (d) A dislocation
Ioop and cusp dislocation formed by closing a single helical turn. 2Ï10 reflection; CuKa
rrtli:rtiorr; lr:rces of (2110) planes are vertical;thickness of sample: (a),(b),and (d) l85 prn;
(r') 125 rr rr. F'rom J. L. Caslavsky and C. P. Gazzara, Philos. Mag., 26,961 (1972).
nliry c()illlrirì 1O'? dislocation lines per square centimeter, and some bulk
cryslirls :rrrrl cl'ystal whiskers have been prepared nearly free of all
dislocrrtions; itÍ'tcr plastic deforrnation the çoncentration of dislocations
inclcirscs trcrncrrtlorrsly, to 10to to 10tt per square centimeter for some
heavily tlcÍorrncrl rnclals.
S'l'Rt l("1'tIRAl, IMl'llRl'll("1'l()NS 169
Final
Fig.4.l8. Frank-Read mechanism for multiplying dislocations. Successive stages are
sh<rwn for the generation of a dislocation loop by the pinned segment J-J of a dislocation
line. This process can be repeated indefinitely.
Multiplication occurs when dislocations are made to move during
deformation. For example, the dislocation in Fig.4.l8, pinned at two
points by impurities, boundaries, or other dislocations, may be caused to
move out to form a loop by an applied stress and eventually breakaway,
forming a new dislocation and the original pinned segment. A segment
such as this, pinned at its ends, is called a Frank-Read source.
Another multiplication mechanism, multiple cross glide, assumes that
Frank-Read sources are generated from cross slip. This is represcnted
schematically for a face-centered cubic crystal in Fig. 4. 19. This process
assumes that a screw dislocation lying along AB can cross glicle onto
position CD on a parallel glide plane. The composite jogs AC and BD are
relatively immovable; however, the segments lying in the two slip planes
are free to expand and can operate as a Frank-Read source. Multiple
cross glide is a more effective mechanism than a simple Frank-Read
source, since it results in more rapid multiplication of dislocations.
Just as we associate an excess energy per unit area with surfsces, an
excess energy per unit length can be used to describe dislocations.
Analogous to the behavior of soap bubbles in which the total surface area
and thus its surface energy are reduced as much as possible, a dislocation
containing a bulge straightens out and minimizes its length if free to
move; a dislocation loop tends to decrease its radius and ultimately
disappear. A dislocation may be considered to have a line tension equal to
its energy per unit length.
t70 tN't'R()t)tI("t'l()N't'o ('t,lR^Mt('s
Fig. 4.19. Cross slip in a face-centered cubic crystal. The t1011 direction is common to (111)
and (l I1) close-packed planes. A screw disloc ationat z is free to glide in either of these planes.
Cross slip produces a nonplanar slip surface. In (c) cross slip has caused a dislocation
generation source at C-D. (Compare with Fig. 4.18.)
At the center of a dislocation the crystal is highly strained with atoms
displaced from their normal sites. This is true to a lesser degree even
some distance away from the dislocation center. At distances of more
than a few interatomic distances from the dislocation center, elasticity
theory can be used to obtain some useful properties of dislocations. We
can consider a screw dislocation such as that illustrated in Fig. 4.20 as a
distortion of a cylinder of radius r. The shear strain y is approximately
equal to the tangent of 7 which is equal tobl2rr, as illustrated in Fig. 4.20.
If Hooke's law for elastic shear is obeyed, r equals Gy, where G is the
modulus of elasticity in shear; the shear stress is given by r : Gbl2rr.
That is, the magnitude of the shear stress is proportionalto llr, where r is
the distance from the dislocation center. Inside some limiting value ro,
1fig.4.20. Iil:rstic tlistoltion around a screw dislocation with Burgers vector
S'l'RI I("1'tIlÌAl, IMl'l,lRl'l,l("1'l()NS t7)
'l'lrt'
Ho<lkc's law clocs not hold, and thc shcirr strcss citnnot bc cirlcttlitlctl.
strain energy associate<J with the straine<J region is eqtrll ltt l12(ì7'.'l'lrlrt
is, the strain energy per unit volume is given by E: ll2G$lTnr)'' ll'llt"
distorted cylindrical shell has a thickness dr and a length /, its vçlrrrrrc is
2rr dr l, and
+:i.(*)= x2nrdr:#i (4.75 )
and E : l" 4E-: Çb',n r' (4.16\
J,n I 4r ro
where E is the strain energy per unit length. calculations of the struin
energy for edge dislocations or mixed dislocations (edge and screw
components) yield essentially the same functional dependence. T'hus, an
approximate relationship for the strain energy per unit length can be
written:
E: oGb' (4.77)
where a :0.5-1.0.
One important result is that the strain energy of a dislocation is
proportional to the square of the Burgers vector b. This is important
because it provides a criterion for what dislocations can be formed in a
given crystal. Those with the smallest Burgers vector have the lowest
strain energy and consequently are the most likely to form. Similar
relationships hold for edge dislocations but are somewhat more compli-
cated, since an edge dislocation is unsymmetrical. In an edge dislocation,
as is clear from considering the added layer of atoms, there is a
compressive stress above and a tensile stress below the dislocation line.
Many common ceramic systems contain a cÌose-packed array of
oxygen atoms. Slip in these oxide systems is usually observed in one of
these close-packed directions. This is consistent with the energy required
to cause strain (Eq. 4.76) beçatse the Burgers vector in a close-packed
direction is smaller; b' is smaller and therefore the strain energy, also.
Dislocations in ionic materials are more complex than in ielemental or
metallic systems. compare the edge dislocations for a mbtal and for
sodium chloride in Fig.4.21a. Note that in order to maintain the regularity
of ions above and below the glide plane, two extra half planes of atoms
are required for sodium chloride. Dislocations may also have an effective
charge just as do point defects (vacancies, interstitials, impurities). This is
illustrated in Fig. 4.21b. A jog in the dislocation results in incomplete
bonding for the negative ion in the case illustrated and results in an
effective charge of - e 12.
b
One place in which dislocation theory has been particularly successful
 172 tN't'R()l)lr("1't()N't'() (,t,Ï{AMt(,s

Lxtnr hall plarres

oí atoms "{17.',11i
^riri;lil;l;*tlft
? Ç'ç
óò3
lll
ï

tô', -cf
T

-ï- f Glide plane -

t
?f
Òô I
Metal Sodium chloride
hl
Extra half planes
of atoms

(a) (b)
Fig.4.22. Dislocation strdcture of (a)a low-angle grain boundary and (à) a bubble modeÌ.
From C. S. Smith, Metal. Prog.,58,478 (1950).
tog

Metal Sodium chloride

(b)
l;ig. 4.21. (a) Schematic representation of an edge dislocation in sodium chloride; (b)
rlernonstration of how dislocation jogs in ionic crystals can have efiective charges.

is irr rlcscribing the structure of low-angle grain boundaries. Just above an

crlgt' tlislocirtion, as illustrated in Fig. 4.22, where an extra plane of atoms
rs insr'r'lerl. thcre is a compressive stress, and below the dislocation there
rs rr lt'rrsilc slrcss. consequently, dislocations of the same sign (positive
l.r llr.sc witlr the extra plane inserted above the slip plane) in the slip
;rl:rrre tcrrrl lo rcpcl one another. Similarly, dislocations of the same sign in
tlillcrt'nr sli;r plirrrcs lcnd to line up above each other to form low-angle
grrrirr lrtrrrntlrlit's ( lìig. 4.22). After annealing, dislocations line up to form
nclworks ol' low-:rnglc g,ririn boundaries. A mosaic structure results (Fig.
4.23\. Fig. 4.23. Three-dimensional dislocation network in KCI decorated with silver particles
(495x). Courtesy S. Amelinckx.

173
 s't'Rt l("1'URAl, IMlDl,)Rl'11('ï'loNs t75

Whcn ir crystlrl is plastically deformcd rìnd then unncrlcd. s()rììc ol'llìc

tliskrcations introduced by the deformation process tend to lirrc trp irt
Iow-angle grain boundaries in a process called polygonization which hirs
been observed for AlzOr, H'O, and many metals. In Fig. 4.24 the restllt of
bending a single crystal of aluminum oxide (sapphire) at high tempera-
tures, which forms a greater number of positive than negative disloca-
tions, is illustrated. Annealing leads to the lining up of the excess positivc
dislocations above each other in the form of low-angle grain boundaries
which can be seen either with the etch-pit technique or by the different
optical properties illustrated by observation of the bent crystal in polar-
ized light.
Dislocations are particularly important in connection with plastic
deformation (Chapter 14) and also in connection with crystal growth
(Chapter 8) and are considered in somewhat more detail with these
phenomena.

Suggested Reading

l. F. A. Krôger and V. J. Vink, "Relations between the Concentrations of

Imperfections in Crystalline Solids," Solid State Plrysics, Vol. 3, F. Seitz and
D. Turnbull, Eds., Academic Press, Inc., New York, 1956, pp.307435.
2. F. A. Krôger, The Chemistry of Imperfect Crystals, North-Holland Publishing
Company, Amsterdam, 1964.
3. N. F. Mott and R. W. Gurney, Electronic Processes in Ionic Crystals,2d ed.'
Clarendon Press, Oxford, 1950.
4. D. Hull, Introductíon to Dislocatíons, Pergamon Press, New York, 1965.
5. F. R. N. Nabarro, Theory of Crystal Díslocatíons, Clarendon Press, Oxford,
1967.
6. H. G. Van Bueren, Imperfectíons in Crystals, North-Holland Publishing
Company, Arnsterdam, Interscience Publishers, Inc., New York, 1960.
7. L. W. Barr and A. B. Lidiard, "Defects in Ionic C.vttutí" in Physical
Chemistry, Vol. 10, W. Jost, Ed., Academic Press, New York, 1970.
8. R. J. Brook "Defect Structure of Ceramic Materials," Chapter 3 in Electrical
Conductiuity in Ceramics and Glass, Part A, N. M. Tallen, Ed., Marcel
Dekker, Inc., New York, 1974.
9. P. Kofstad, Nonstoichiometry, Electrical Conductiuity, and Diffusion ín
Binary Metal Oxiiles, John Wiley & Sons, Inc., New York, 1972-

Fig-4.24. I)olygonizirtion of Al,o.. (c) Etch pits at dislocations in bent rod. courtesy p.
Gibbs. (r) I)islocations lincd up in polygon boundaries after annealing. courtesy p. Gibbs.
(c) Polygons in hcnt cryslal vicwed in polarized light. Courtesy M. Kronberg.

174
 176 lN'l'R()ltlJ("1'l()N'l'O ('l,lRAMl('S

l)roblems
5
4.1. Assunring no lattice relaxation around vacancies, what would you predict as the p,,,
rrntl 'l'clependence of the density of (a) Fe, (b) UOz*,, and (.c) Zn,t,O.
"O,
4.2. ljstimate the concentration of associates at 1000"C inZrOz doped with 12m/o CaO
(rc - 30).
4.3. Al,O. will
form a limited solid solution in MgO. At the eutectic temperature (1995"C),
approximately 18üf%o of AI:O. is soluble in MgO. The unit-cell dimensions of MgO
decrease. Predict the change in density on the basis of (c ) interstitial Al3* ions and (á )
Surl'irccs.,
substitutional Al3* ions.
4-4. Make a table listing the structural imperfections that occur in crystalline solids. Do I nlcrl'ttccs.,
not consider secondary imperfections that result from the interaction of two or more
basic imperfections such as F centers. In your table make a one-sentence definition so irn(l (irtrirr
that the instructor can evaluate your understanding of the imperfection. In a third
column designate whether or not the imperfection is thermodynamically stable. Bou nditrtcs
4.5. (a) If two parallel edge dislocations of the same sign lie on the same slip plane, that
is, their half planes of atoms are parallel and terminate on the same plane
perpendicular to the half planes, would there be a force of attraction or
repulsion between them? The surfaces and interfaces between different grains and phases arc
(.b) Would this be the result of compressive, tensile, or shear forces? important in determining many properties and processes. From one point
(c) If the distance between them increased 10 times, how much would the force of
interaction decrease or increase?
of view these may be regarded as two-dimensional imperfections or
(d) If one of the edge dislocations were replaced by a screw dislocation, describe departures from the ideal crystal lattice structure. or the surfaces of a
the interaction between the two line defects. crystal may equally well be included in defining the ideal crystal structure
4-6. Estimate the number of free vacancies, interstitials, and associates in I cm'at 500"C for any particular environment. In either case it is desirable to understand
of (a) pure AgBr and (á) AgBr+ 10-'m/o CdBra the structure, composition, and properties of the boundaries of a solid or
4.7. Construct a diagram similar to Fig.4.8 for the AgBr data in Problem 4.6. liquid and the interfaces between phases, for they have a strong influence
4.8. Estimate the electron binding energy for the reaction Ti,t e,:Tii, in rutile, TiO: on many mechanical properties, chemical phenomena, and electrical
(x - 100).
properties.
4.9. To the schematic data in Fig. 4.13b add the curve for the nonneutrâl associate
(V kF;). Assuming that cation interstitials form, will (V hF;) associates be larger in
number than (V,i,Mi)? 5.1 Surface Tension and Surface Energy
4. 10. The common edge dislocation in rock salt structures is shown in Fig. 4.21 . Sketch the
edge dislocation in CaFr. It is observed experimentally that a force is required to extend a liquid
'1.I l. Í)islocations are observed in most simple oxides, and in fact crystals are di{ücult to surface. The surface or interfacial tension 7 is thus {efi,ned as the
prcpirre without large concentrations, >l0o/cm'. In more complex oxides, such as reversible work lr. required to increase the surface of the liqúid by a unit
g:rlrcts (e.g., yttrium aluminum garnet, Y.AlrOr:, gadolinium gallium garnet, area:
(itl,(ì:r.O,,) single crystals are easily grown dislocationfree. Why?
dw,: y dA (5.1)

Specification of the conditions under which this surface work is done

allows the surface tension to be related to other thermodynamic proper-
ties of the system. Consider the two-phase multicomponent system
shown in Fig.5.l. According to the first and second laws of ther-
modynamics, the variation in the internal energy E or the Gibbs free
energy G of this system, in changing from one equilibrium state to
another, is expressed by
r77
1r

ili
)l

il