Revista Latinoamericana de Metalurgia y Materiales, Vol 5, N° 1, 1985

ARTICULO INVITADO / INVITED PAPER

The Factors Influencing Solid Solubili ty in Metallic Alloys

Julio A. Alonso
Departamento de Física Teórica, Universidad de Valladolid, Valladolid, Spain

Solid solubility is a phenomenon of metallurgical interest. The extent of equilibrium solid solubility of a metal A into a host metal B..
depends on several factors which control the free energy of formation of the solid solution and competing intermediate phases. One of
those factors is the entropy of formation. Then, there are several elemental properties of the components which control the enthalpy of
formation: atomic size, electronegativity, generalized 'valence, electro n density at the boundary of bulk atomic cells, and structure.
Thosefactors are discussed separately and then integrated into a semiempirical theory ofsolid solubility. The theory is applied to seve-
ral types of alloys and the comparison with experiment is satisfactory.

1. INTRODUCTION
The solid part of the phase diagram of a binary sys-
tem formed by two metal s A and B normally contains
(although not necessarily) one or several intermediate
phases, in addition to terminal solid solutions of A in B
and oí B in A. In a solid solution, the solute atoms enter
substitutionally into the host lattice preserving the crys-
talline structure of the host, Interstitial solid solutions
also exist but those are less common than substitutional
solutíons. On the other hand, an intermediate phase nor-
maJly has a narrow composition range and a crystal
structure different from those of the component metals.
Solid solutions and intermediate phases are competitive.
Progress in the understanding of the factors governing
the occurrence of the different alloy phases has been grow-
ing in recent years, but a complete theoretical descrip-
tion of phase diagrams, and in particular of solid
solutions is a very difficult task. In fact, the general opi-
nion of workers in this field is that "the prediction ofter-
minal solid solubilities is a formidable problem and it
remains a rnajor unsolved question in metallurgy" [1].
For this reason, semiempirical treatments, focussing on
a few fundamental factors, have been traditionalIy
employed, and in order to do this one must know the fac-
tors influencing solid solubility.
In this review I consider the different íactors which
influence solid solubility. It will become clear that a sin-
gle factors is not enough and that progress can only be
achieved when the different factors are considered toge-
ther. The conclusion is, as I hope to make clear along the
paper, that substantial solid solubility only occurs when
the components of the alloy are chemically similar. Thus
the objective oí the paper is to explain how this concept oí
chemical similarity can be made quantitative,
2. THERMODYNAMIC DESCRIPTION
The extent oí equilibrium solid solubility of a solute
metal A in a solvent metal B at a temperature T can be
calculated from the free energy offormation LlGs (xA• xB; TI
of the substitutional solid solution as a function of
composítíon
LlGs (XAf XB; T) = Gs (xA, XB; T)-
- XA GA (T) - XB GB(T), (1)
where GA and GB are the free energies of the pure metals
and Gs (xA, xB; T) is the free energy of the solid solution
with atornic concentrations XA and XB = 1 - XA. A neces-
sary condition for the thermodynamic of the solid solu-
tion is
(2)
Nevertheless. (2) is not a sufficient condition, because of
the possible existence of competing phases, normally an
ordered intermetallic compound of fixed composition
A"BJJ with free energy of formation AGc (a, f3; T). Con-
sequently, the extent of solid solubility at a temperature
T is calculated by computing LlGs (xA, XB; T) as a func-
tionof concentration and comparing it with LlGc (a, (3; T)
using the tangent method [2]. If intérrnediate phases do
not exist, then the competing phase is a solid solution of
B in A (íf the crystal structures of A and B are dífferent).
The free energy of formation can be separated as
AG= AH - TAS, (3)
where AH is the enthalpy of formation and AS is the
entropy of formation.
The entropy of formation of a substitutional solid
solution is positive [3]. Its calculation is a difficult mat-
ter. The simplest approximation is to consider ideal
mixing (when the enthalpy of formation is nearly zero).
Therí the entropy of random mixing can be used to
write [4]
(4)

3
 LatinAmerican Journal of Metallurgy and Materials, Vol 5, N° 1, 1985
where K is the Boltzmann constant. More sophisticated
approximations can also be used [5]for departures from
ideal behaviour. In contrast, the entropy of formation
aSe of ordered intermetallic compounds is normally zero
or negative [3, 6-9] (magnetic effects sometimes induce
exceptions to this rule [7]. Therefore, the entropic term
favors the existence of solid solutions.
Most efforts to understand solid solubility have
been devoted to investigate the factors which control the
enthalpy of formation aR..
3. SIZE FACTOR (FIRST HUME-ROTHERY
FACTOR)
In the early 1930's Hume-Rothery proposed several
factors controllíng solid solubility, which form the basis
of our present understanding [lO, 11]. The first one is the
famous size rule, stating that solid solutions are not to be
expected ifthe atomic sizes of solvent and solute differ by
more than 15%,and that solid solutions may form if the
size difference is less than 15%, provided other factors
are favorable. To apply this rule, the atomic size of a
metallic element can be measured by the radius r of a
sphere with a volume V equal to the volume per atom of
the pure metal. Waber and coworkers [12] have applied
the size rule to 1423 terminal solid solutions. Of the sys-
tems predicted to be insoluble, 90%were found to exhíbit
Jimited solid solubility (the distinction between a limited
and an extensive terminal solid solution alloy was taken
at5 atomic %).However, ofthe systems predicted to have
extended solid solutions only 50%were so found. In other
words, a favorable size factor is a necessary but not a suf-
ficient condition for the formation of extensive solid solu-
tions. Clearly other factors also playa roleo Since the
solid solution preserves the crystal structure of the
parent metal it is understandable that atoms with a size
very different from the size of the host atoms introduce
large lattice distortions, with a corresponding high elas-
tic energy cost in aHs, which oposes the formation of a
substitutional solid solution.
In 1954 Friedel [13] used the eleastic continuum
theory to derive a quantitative expression for the size
factor. He calculated the elastic energy needed to intro-
duce a sphere of radius rA (representing the solute atom)
into a spherical hole of radius rB (with rB = (3V¡J47T)1/3,
the atomic radius of the host). Then assuming (a) com-
plete disorder and (b) that the enthalpy offormation of a
solid solution is dorninated by the elastic energy effect,
Friedel found that there is a large solid solubility only if
the following condition is fulfilled
I~rl <
r
(KT BXB)1/2
7TJ\
(5)
In this expression TBand XBare the melting tempe-
rature and the compressibility of the host respectively,
.ó.r = rA- rB'and f is theaverage atornic radius, Ther. h. s
of (5) defines a specific size factor for each host metal, in
contrast with the Hume-Rothery rule. A reasonable ave-
rage value oí TBXBr~3 = 1014C.G.S. leads to the condi-
4
tion I dr I / r < 0.15, in· agreement with the Hume-
Rothery size rule. Independently, Eshelby [14] proved
that the contribution to the heat of solution from size mis-
match can be written
where JlB is the shear modulus of the host and XAis the
compressibility of the solute metal. The heat of solution
Ah, (A in B) is related to the heat of formation
.ó.Hs (x¿ xB) of a concentrated solid solution by
From (6) Eshelby then derived another expression
for the size factor (SF)
(8)
where aB is the Poisson ratio of the host and R is the gas
'constant, This size factor was also used to justify Hume-
Rothery's size rule. Values of the individual size factors
(5) and (8) have benn tabulated by Gschneidner [15].
Those values areclose to 15%for simple metals and be-
tween 10% and 15% for transition metals.
The size factor has been further discussed by Egami
and Waseda [16] using elasticity theory [14]. These au-
thors point out that a substitutional impurity introduces
a volume strain on the host which increases Iinearly with
solute concentration. If the str.ain exceeds a certain point
then the solid solution becomes topoligically unstable
and the system transforms to a different crystalline
structure (or upon rapid quenching freezes into a glass).
The reason for this topoligical transformation is that
local pressure and local coordination number are di-
rectly related [17]. Then, Egami and Waseda have shown
that, at a critical composition, the local elastic energy
due to the local pressure can be reduced by a change in
the local coordination number. This critical concentra-
tion has been found to correlate with the minimum solute
concentration necessary to obtain a stable amorphous
phase by rapid quenching from the melt [16].
4. ELECTROCHEMICAL FACTOR (SECOND
HUME-ROTHERY FACTOR)
The second Hume-Rothery rule states that the elec-
trochemical nature of the two elements must be similar if
solid solution formation is to be expected. If, however,
their electrochemical natures are different, compound
formation is likely to occur. A measnre of the electroche-
mical nature of the elements is provided by the "electro-
negativity" 0, a concept introduced by Pauling [18].This
author defined the electronegafu - as he power of an
atom in a molecule or compo d aí:tl"act electrons to
itself. In the modern view, electro ega - -ty is minus the
 Revista Latinoamericana de Metalurgia y Materiales. Vol 6. N° 1. 1986

2.6
denivative dE(N) / dN ofthe energy ofthe atom wíth res-
DARKEN· GURRY MAP FOR Pb
pect to the total number of electrons N [191..Different
empirical electronegativity scales exist, derived from 2.2
• SOLUBILI TY ~ 0.5 %

thermochemical data, like those due to Pauling [18] or .0.01'1..$ SOLUBILlTY s 0.5%
a SOLUBILITY i 0.01%
Miedema [20],orfrom free atom data. For instance Mulli- >-
~ Ag O NOT MEASURED
ken proposed [21] >
.... 1.6' PbD HOST
~
us' Hg Bi
Z
IP+EA o
<r Nb
Ta
cP = 2 ' (9) f-
e>
1.4 o <, 1'_ -- Sn"
w
...J se Sn ;' cd
2

where IP is the first ionization potential and EA is the 1.0

sr
o
electro n affinity of the atom. In fact, Mulliken's formula
is a finite difference approximation to - dE(N) / dN.
oBO
. Rb
o ecs
The physical significance of the electrochemical factor 0.6 L_---L __ --.L_---;..L.- __
K
.l.-_--L __ ~-____=_'
0.6 1.0 1.4 i.a 2.2 2.6 3.0 3.4
for alloy formation is the following one: a difference
RADIUS (A)
dcp = cPA - CPBin the electronegativity of the two atoms Fig. L Darken-Gurry map for various solutes disolved in Lead, as
induces a transfer of electronic charge from the less elec- constructed by Chelikowsky [23]. An encIosed region oí high
tronegative to the 'more electronegative atom until the solubility is indicated, centered near the Pb host. Electrone-
electronegativities are equalized in the alloy. This gativities are from Gordy and Thomas [24] and atornie rad.ii
charge transfer (Ch T) contributes to the enthalpy of Pauling [18]. The measured solubilities correspond to their
room temperature values.
alloy formation with a term

(10) dicting limited solid solutions, and an improvement in

proportional to - (dCP? [18,20]. This term then favors
the formation of the alloy. The proportionality factor is a
.function ofthe area of contact between the atomic cells of
solute and solvent [20].The total area of contact between
disimilar atomic cells is larger in ordered compounds
than in random substitutional solutions. Then the elec-
trochemical factor favors the formation of ordered com-
pounds. Only if dcp is small the entropy of random
mixing is able to overcome the tendency to form ordered
compounds and to stabilize the solid solution. This is the
reason for the need of similar electronegativities for solid
solution formation.
5. DARKEN-GURRY MAPS
In the early 1950's Darken and Gurry (DG) took an
important step in the predietion of solid solubilit~ when
they were able to use simultaneously both the size and
the electronegativity factors [22]. These authors eons-
tructed a map where the two coordinates are the electro-
negativity and ·the atomic radius. In this map all the
elements whieh líe close to the point representing the
host exhibit high solid solubility in that host. They made
this eorrelation more quantitative by constructing an
ellipse centered on the host and with an axis of± 15%of
the solvent's size and another axis of ± 0.4 of a Pauling
electronegativity unit of the solvent's electronegativity
value. The ellipse then encloses most ofthe solutes which
have extensive solubility in the host and exeludes those
having limited solid solubility. A DG plot for Pb alloys is
shown in Figure 1. Waber et al [121made an extensive
analysis of the DG method, eoncluding that this method
eorrectly prediets limited solid solubility in 85% of the
cases and extended solubility in 62% of the cases. Thus
the DG method yields an improvement in predictingsolid
solutions, with a slight deerease in the reliability of pre-
the overall reliability over the simple size rule. Gschneid-
ner and eoworkers have used the DG method to predict
the observed behaviour of actinide and rare-earth metal
alloys [25-27].Others have used the DG method for speci-
fie elements. Reference to those works can be found in a
review paper by Gschneidner [28]. It is clearfrom the dis-
cussion in Seeti-ons 3 and 4 why only elements near the
host in the DG map show large solid solubility. A large
size differenee oposes the formation of solid solutions
and a large electronegativity differenee favors the for-
mation of ordered intermetallic compounds.
An interestingobservation from Figure 1 is that ato-
mic radius and electronegativity show some degree of
correlation, namely, a large electronegativity roughly
corresponds to a small atomie size. But there is a lar~e
seatter about an "average" correlation,'and those devia-
tions are precisely the reason why the two-coordinate
map does a better job than the simple rule based on ato-
mie size only.
Sood has recently applied the DG method to metas-
table alloys obtained by ion implantation in metal s [29,
30]. After low dose irnplantation, some elements adopt
substitutional positions in the host while others become
interstitial. Sood then formulated an empirical rule for
the substitutionality as follows: "A metastable substitu-
tional solution will be formed if the implanted specíes
have the atomic radius within - 15%to + 40%ofthe host
atomie radius, and the Pauling eleetronegativity wit.hin
± 0.7 units ofthatofhost atorns". The region of substítu-
tionality permitted by these rules is then a rectangle. The
validity of this rule was tested for implantation in Cu, Fe,
Ni, V and Be. Figure 2 shows sueh a plot for implantation
in Cu.
Finally we notiee that the separation, in a DG plot,
between soluble and insoluble impurities can be impro-
ved by drawing, instead of the DG ellipse, a relaxed

5
 LatinAmerican Journal of Metallurgy and Materials, Vol 5, N° 1, 1985

details. Goodman and coworkers observed that the trend

in Figure 3 may be rationalized in terms of band theory
>- concepts: if the boundary region corresponds to the dia-
esubstitutional
f-J
•....• onon su bs.
gonal involving the dotted, rather than the cross-hat-
ched squares, then the observed rule could be sum-
> marized by stating that whatever transition metal has a
f-
-<i: more nearly half-filled d-band, that metal prefers being
<..9 Do
the solute rather than the solvent. The actual boundary
w2. is one set of squares to the left. Thus, the trend is descri-
z bable as obeying a "Ieft-shifted" half-filled d-band
o
o: rule.
f-
u
w 1. Rb ELEMENT B
....J O O
d7 d6 d5 d'f
W es ---- ----
Ni Pd Pt eo Rh Ir Fe Ru Os Mn Te Re er Mo W V Nb To

Ti • O • O O:'00 O O O O •
3
O • • :.:~Q.~.Q.O O O O O O O O
0.5 to 1.5 2.0 2.5
d Zr
IHf •• :0:.0:':> O O O O • O. O
o v •••••• : :.."0:·0 O O
A TOMIC RADIUS (A) d
4
Nb •• • • • • :":'.">::':.9: O O O
Fig.2. Darken-Gurry plotfor implanted alloys in Copper (29).Ce and
Eu are plotted twice according to two valence values. Circle is
Darken-Gurry solubility zone. Dashed rectangle represents
metastable substitutional solubility zone as proposed by
Sood. Electronegativities and atomic radii are from a compi-
E
L d5
I To
¡er •••
Mo ••••
• •.•

O....••

O ••••
.:::.0 ..9. O • O

,,-,,,,-/','-'1
~ W •••••••••
lation of Teatum et al [31J. E Mn ••• f-'o-'>""':>"".>....:>J

~ d61 Te O O. O • A MORE SOLUBLE IN B

ellipse with axes of optimized direction and length [23] as

A Re
Fe
O • O
•• O B MORE SOLUBLE IN A

in Figure 1.
6. RE LATI VE VALENCE EFFECT
(THIRD HUME-ROTHERY FACTOR)
The third Hume-Rothery rule states that a higher
valent metal is more soluble in a lower valent metal than
viceversa. Darken and Gurry [22] have presented this
rule in a slightly broader framework, namely they sta-
ted: "a disparity in valence is conductive to low solubility
and this disparity has an especially pronounced effect
when the valence of the solute is less than that of the sol-
vent". In our opinion, two aspects of the rule must be
separated. The first one is a statement about absolute
solubilities, that is, a difference in valence leads to low
solubility. On the other hand the second is a statement
about the relative solubilities of A in B and B in A. The
work of Gschneidner [28] and of Goodman and coworkers
[32-34] has clarified the situation with respect to the rela-
tive solibilities. Gschneidner examined 300 systems for-
med by two metals of different valence andfor which the
terminal solid solubilities are known at both ends of the
phase diagram. The result was that 55% of the systems
do not obey the rule. The work of Goodman et al. deals
with the relative solubilities in transition metal alloys.
Figure 3 serves to illlustrate the discovery of these au-
thors. The chart can be divided in two regions separated
by a diagonal boundary (the cross-hatched región). Only
in one ofthe regions the relativevalence rule i obeyed. In
contrast, in the other región, the rule obeyed is tbat "the
lower valent metal is more soluble in the higher alent
one than viceversa". Some exceptions actually appear in
each region and the reader should consult refs [32-34] for
d71RU O •
Fig. 3. Experimental relative solubilities of transition metal alloys.
The dotted region indicates systems where the d bands ofthe
alloy' constituents are equally far from half-filled. The hat-
ched region indicates a diagonal boundary regio n in which no
clear bias appears in the relative solubilities. From Watson et
al [34].
In conclusion, Hume-Rothery's rule about relative
solubilities is definitely wrong, although some trends in
relativa solubilities have been identified in the particular
class of transition metal alloys. N evertheless, Gschneid-
ner's opinion is that Hume-Rothery's third rule should
not be abandoned, but rather reformulated by broade-
ningthe meaning ofvalence to indicate, notjust the num-
ber of outer electrons involved in bonding, but also to
include the kind of electrons, s, p or d, that are involved.
The elements were then divided into two groups. a) The
d-group elements are those which have d electrons or
readily available empty d orbitals. This group includes
the transition metals, noble metals, lanthanides, actini-
des and the alkali and alkaline-earth metals, except Li,
a, Be and Mg. b) The rest are the sp group of elements,
which have only unpaired s and/or p electrons. If, instead
of disparity in valence one considers the disparity in elec-
tronic character (or group), it can be concluded that low
solubility is to be expected for alloys formed by one d-
metal and one sp-metal(absolute solubility statement),
and as a reminiscence ofthe relative valence rule, one can
also conclude that the d-metal is more tolerant to an sp
solute tan viceversa. Some insigth into the reasons for
these effects is left to Section 8.

6
 Revista Latinoamericana de Metalurgia y Materiales. Vol 5. N° 1. 1985

7. ELECTRON DENSITY AT THE ATOMIC an intermetallic cornpound AoBfJ can be written

CELL BOUNDARY AND MIEDEMA MODEL (a = XM f3 = xB)
OF ALLOY FORMATION

In Sections 3 to 6 we have presented the traditional

Hume-Rothery factors, that have served as a basis to dis-
where
cuss solid solubility. Other factors which have been
introduced recently are now discussed.
In 1979 Chelikowsky [23] presented solubility plots
in which the two coordinates are the electronegativity (j)

and the electron density n at the atomic cell boundary.

Chelikowsky applied the new coordinates to the case of (12)
divalent hosts and he demonstrated that those coordina-
tes give more accurate predictions than the Darken-
Gurry coordinates. He was able to draw an ellipse x~ and x~ are atomic-cell-surface area concentrations,
containing most metal s which are soluble in a given host.
The position of the host metallies inside the sllipse, al- defined
though its precise location varies from host to host, Á
solubilityplot for Fe-alloys i~~iven in Fi81lre4.-(j) and n
(13)
6 ,-------,--------,--------,--,
e SOLUBILlTY) 15% I Os I
• 15%> SOLU,BILlTY > 5% ?t.;í~r u Miedema's model assumes that the heat of formation is
o
° 5%) SOLUBILlTY)
1%>SOLUBILlTY
1°;';° PdRh e
eTe e
proportional to the area of contact between di similar ato-
b. HOST Ni¡e ) mic cells in the alloy. This is the reason for the prefactor
A¡jl Co· XA x~ [1 + 8(xÁ X~)2]in (11). For a random distribution of
5- b. -
As. B Fe .W atoms, as in a liquid alloy, the factor is simply XA x~. This
G~ 5;·° ee'M'o is corrected by the factor [1 + 8(xÁ X~)2] which has been
•• Cu Cr / introduced by Miedema to account for the fact that in an
5bo / °Ag M~ /
Si Hg Al Be eV ordered solid compound the total area of contact be-
P b8s~4:.0.-0~TO tween disimilar atomic cells is larger than for a random
4'- oCd o UDNb
-

alloy, P and Q in (12) are positive constants. The first

In Puo
.Ti termin (12), proportional to - P(~(j)2, is a chargetraris-
el> (V) Mg °Hf fer term which accounts for the electrochernical factor of
°
YooOTh
°Zr Section 4. On the other hand, the second contribution,
Loo Se proportional to (~nli:l)2, takes into account the increase
3f- -
Lio in electronic energy needed to smooth out the difference
oNo in electron density at the interfaces between A and B ato-
Coo mic cells. The recommended values of n and (j), as well as
oSr a fuI! discussion of the model, are given in Miedema's
oK °Bo
o
papers [20,37]. A third electronic contribution to ~ho has
2 "-Rb - been neglected in writing (12). We have left it out for the
moment since this contribution is not needed .to discuss
Chelikowsky's plots, That contribution, only present
for alloys of a transition metal and a polivalent non-
transition metal, will be considered in Section 8.
1 I I
The success of Chelikowsky's plots can now be
0.5 1.0 1.5 2.0 understood from eqns (11) and (12). The two coordinates
n 1/3 in the plots control the heat of formation of ordered com-
pounds, which are competitors of solid solutions. If
l"ig'-.4. Chelikowsky's plot for so lid solubility in Iron alloys [35J. The
two coordinates are the electronegativity 1> and the electron
P(~(j)2 » Q(~nI!3)2, then ~Hc « O and urdered com-
density parametern ofMiedema's theory. An ellipse has been pounds are likely to form, On the other hand, if
drawn separating soluble and insoluble elements, The solubi- P(~(j)l « Q(~nl/3)2, then ~Hc » ()/and alloys do not
lit Y refers to the maximum solid solubility (at a temperature form at al!. Finally, if P(~(j)}2 :::,,:Q(~nl/3)2 then .ó.Hc:::":O
not higher than the melting temperature of the pure solute). and a disordered solid solution can be the stable phase
See ref [35] for details. cp is given in Volts and nl/:l in (density
uníts)':".
due to entropy eontributíons.
Figure4 also shows that a rough correlatíon exists
between (j) and n1/:1. The points are scattered about a
are fundamental ingredients in a serniernpirical model of straight line with slope (Q/P)I/2, which is the value of
heats of alloy formation developed by Miedema and ~(j) / ~n1ll dividing compounds with positive heats of
coworkers [20,36]. In this model, the heat of formation of formation from those with negative ones. If the values of

7
 LatinAmerican Journai of Metallurgy and Materials, Vol 5, N° 1, 1985

rP lie exactly on the line (Q / p)l/2 nl/3, all energies of corn- A 501.>0.5
pound formation would be exactly zero. Therefore. the .0.5 >501. > 0.0.1
K
deviation of the parameter cp from this line is the signifí- 5 /' O 00..0.1> sol.
. cant feature [38] and it means that cp and nlf3 are not Vl I + Host
¡e Bo
equivalent coordinates. e
\
:J
Plots using cp and n as coordinates have also been -, eCo
:§ 1,
'\
employed in other problems. In fact, the first application -,
Ul <,
was to metastable alloys produced by low dose implanta- :J <,
tion in Berillium [39, 40]. A perfect separation was found '6 -,
~ <,
between substitutional and interstitial implants in Be. 3 O O
Other recent applications to implantation and to amorp- u Mo W
.-
hous alloys produced by rapid quenching have al so be en E
.8
performed [41-45]. a
2

8. TERMOCHEMICAL PLOTS -20.0. -10.0. o. 10.0.

11h~ in Pb ( kj I mole solu!e l

A point to notice is that the Darken-Gurry plots, alt-
Fig. 5. Solubility map for Lead alloys using the thermochernicat
hough less successful than-Chelikosky's plots, provide a coordinates [49). Data correspond to room ternperature solu-
good first approximation for the prediction of solid solu- bilities. Contours of increasing solubility have been drawn.
bilities. Both maps have a coordinate in common, the
electronegativity 1>. The fact that Pauling electronegati-
vity [18] is normally used in the DG plots, whereas Mie- applied to study solid solubility in metals [35, 42, 48-52]
dema electronegativity was used by Chelikowsky, is of and semiconductors [53] (Figures 5, 6 and 7 show the
minor importance. There is a good correlation between corresponding plots for Pb, Cu and Fe alloys respecti-
both scales [36]. The second coordinate is different: ato- vely). Other applications have been to alloys formed by
mic size V in one case and cell boundary electron density ion implantatíon in metals and semiconductors [50, 53]
n in the other. Noticing that V and n are not equivalent and to the prediction of amorphous alloys obtained by
coordinates [35, 36, 46] Alonso and Simozar 135] propo- rapid quenching from the melt [47, 51, 54, 55] or formed
sed in 1980 a new scheme that takes into account the by ion beam mixing of multilayers [56].
three coardinates 1>, n and V. To preserve the simplicity
of two-dimensional map these authors combined 1> and n oCa ¿ sol.> 15
1,
in a new coordinate: the chemical part of the heat of solu- ol.c e 15 > sol. > 1
Vl y
tion as given in Miedema's theory. That chemical part, o 01 > sol.
·c
which excludes size mis match effects and peculiar struc- :J + Host
Se
tural effects (tú be considered in Section 10) is given by ..9 O Hf
~ho of eqn (12). In writting (12) a term was neglected. °Zr o
Ul
Then we now write the complete expression for the che- :J
'O 3
mical part of Ah, a
~
u
• ·Eo
o
ON
(14) 2
-150 -80 o 80. 150.

I1h; in Cu (kj/moI8solut8l
where the new term - R is different form zero only for
alloys formed by a transition metal (TM) and a polivalent Fig. 6. Solubility map Ior Copper alloys using thermochemical COOl'-
non-transition metal (PM). Miedema interprets this term dinates [49[. Data correspond to room temperature solubili-
tieso Contours of increasing solubility have been drawn.
as due to hybridization between the d-electrons of the
transition metal and the p-electrons of the polivalent
neighbours. In first order, this (negative) contribution The success ofthe thermochemical plots [49] and the
depends not much on which d-metal is alloyed with which form of the boundaries separating soluble and insoluble
p-metal partner. The values of Rhave been given by Mie- impurities is easily understood, because the V axis gives
dema for the different TM-PM combination. This means a measure of the size mismatch contribution to the heat
that the thermochemical coordinates (V and Llh~)contain of solution and the Ll~ axis accounts for the chemical
much more than the combined effect ofV, n and 1>. They part of ~hs' These are the two main contributions to Llh,.
also contain hybridization effects. Only if those two contributions are not too different from
Aetually, the thermochemical coordinates were zero a solid solution can exis since:
simultaneously and independently praposed by Giessen
and Whang [47] for the prediction of amorphous alloy a) if ~~» O, then o do not form at all,
formation. The thermochemical coordinates have been b) if~« 0.0 compounds are expected,

8
 Revista Latinoamericana de Metalurgia y Materiales, Vol 5, N° 1, 1985

e) if Ll~ > O (positive, but small), then the deci- f'l.H~ (x\, XI;) = X,\ x~ Llh~ (A in. B). (16)
ding factor is Llh:!ze. Only if Llh:!ze is small
(Llh:ize ~ O) can a substitutional solution exist On the other hand, if the solute concentration is srnall,
due to the stabilizing effect of the entropy of Eshelby [14] has shown that LlH:¡'''(xAo xu) is given by
random mixing, (17)
d) if Ll~ < O (not too negative), the deciding fac- (y - l)y ] ~h"'''(A in B)
tor is again Llh:!ze. If this term is small, the (y* - l)y*' ,
entropy of random mixing again favors a sub s-
titutional solid solution. But if Llh!ize »
O,orde- where
red compounds are expected (except if Llb; is
very nearly zero). This is because in an ordered
compound the crystal structure and the concen- y=l+~. (18)
3XJ:
tration of the components are optirnized in such
a way that elastic size mismatch effects are
vitually eliminated.
v" = 1 + 41-'-1: (19)
.• sol. > 15 3 K\
.-15>501.>1
I/! 01 > sol. oCa Adding to (15) the entropy part of eqn (4) an expression
'2~ 4 ~----------~
+ Host No°- o/ for the free energy of formation of dilute solid solu-
K tions results,
a
This theory has been applied to a very simple class
Zr Hf
of alloys, those formed by metals with very low mutual
solid solubility and without intermediate compounds.
OAg AlIoys ofthis kind are: Se W,Se Ta, Se Nb, Sc Cr, Se V, V
u Th. Vi Th. Ti Y and Ti La 157J. In these cases, both f'l.h:""
E and Ah; are large and positive. Figure 8 is a plot of the
~oo calculated free energy of forrnation of dilute solid solu-
- ---~ tions ofNb in Se and Se in Nb, at three different tempera-
tures: 1;100K. 1400 K and 1500 K. At these temperatures,
-80 O 80 160 t-.G is negutive at very low solute concentration. This is
lthc in F," (kj / mal," 50Iut,") due to the entropy of mixing sinee AH, (XI' XI) is posi-
s
Fig. 7. Solubility map for Iron alloys using thermochemical coordi-
tive, Then. after reaching a minimum. f'l.G, raises to posi-
nates [491. Data eorrespond to room temperature solubilities. tive values at an still low solute concentration. Figure 8
Contours of inereasing solubility have been drawn. shows that the solubility of Sc in Nh 01' Nb in Sc is very
low. The extcnt of solubility evidently increases with T.
The c.ilculated solubilities of these and other alloys are
We close this Seetion with sorne cornrnents about given in Table 1. where the predictions are eompared to
che relative valence effect (see Section 6). First we recal! experiment. The expeJ'imental values are not very pre-
that Gscheneidner discovered 1281 that low solubility cise in so me cases ancl calculated solubilities are very
occurs when one eomponent of the alloy is a d-metal and small, in :tgl'eement with measured limits.
the other is an sp-metal. We observe that this effect is
reflected in the ter-m - R of eqn (14). This contribution is
negntive and leads to substantially negative values of
I O f----f------f------------+---____\_______\_::\-1
1300
Ah". As already discussed, this favors the existenee of § 1400
ordered eompounds. t::)-

~ -001' 1500
h
9. SIMPLE SEMIEMPIRICAL THEORY G
~
OF SOLID SOLUBILITY
~~
':;,g...• -002
The ingredients presented so far allow us to esta- ~~--~~~~~~·~--~~--~~~
blish a simple theory of solid solubility [57]. The heat of O O997 Oggg 1.0
solution Ah, (A in B) is obtained by adding the chemical CNb(af%}~
part of eqn (14) to the size rnismateh part of eqn (6). Then, Fig. 8. Calculated free energy of formation ~G, of dilute solid solu-
for a concentrated solid solution the enthalpy of forma- tions of Nb in Se and of Se in Nb at several temperatures [57].
tion can be written, . ~G, has been calculated as explained in Section 9.

Due to the limited validity of eqn (17) the theory has

been applied to dilute alloys. Recently, the size mismatch
The chemical part is calculated from Miedema's theory contribution has been extended to the eoneentrated
(see Sections 7 and 8) regime 162J. Here we simply outline the method. The for-

9
 LatinAmerican Journai 01Metallurgy and Mtüerials, Vol 5, N' 1, 1985

mation of a eoneentrated alloy is divided into severa] res, Niessen and Miedema derived the variation of the
steps. In the first step a dilute alloy is formed and eqn (17) absolute stability Eu(z) (O' = bcc, fcc, hep) of eaeh of the
is used to calculated the elastic size mismatch contribu- main erystal structures. Their results are shown in
tion. At this point, the dilute alloy is considered as an Figure 9, where the referenee state is the average of the
effective pure metal with atoms having the average ato- elose-paeked struetures and the looser paeked bee lat-
mic volume and average elastic constants. Then, a few tiee, that is
more atoms are added to this effective host. The elastic
energy cost in this second step is again computed from
eqn (17), noticing that the host is now the effective host. (~) ---bcc
The process continues in this way and any concentration 20 ---- fcc
can be reached after a sufficient number of steps. The

t .'..•..•
hcp
theory was applied to the alloys Se Y, Se Zr, Se Rf, Ti Rf,
~\
VCr, VNb, VMo andVWatT = 1000 K, and itwasfound :J
[62] that those eight alloys have a negative near-zero
:,
,¡
\~
\.
value of ~G, over the whole concentration range at that /¡ ,':
ternperature. This faet is eonsistent with the experimen- I
J ".\
tal phase diagrarns ofthose alloys [6]. which show conti- ! \
nuous solid solubility at that temperature, O~----~~----~r-----------
10. CRYSTAL STRUCTURE FACTOR

The erystal structure of the two pure metal s also -10

influenees the extent of solid solubility. A substitutional
solid solution preserves the crystal structure ofthe host.
Then, it is clear that complete solid solubility (from
x" = Oto x,\ = 1) can only oeeur ifthe erystal structure of -20
the two pure components is the same. Gsehneidner [28]
has analyzed the influenee of crystal strueture in an
empirical way. His diseussion is based on the distinction
between a) eommon metallie structures (fcc, bcc, hcp,
double hcp, Sm-type and fct-In type structures) and b)
Fig. 9. Absolute stability ofthe bcc, fcc and hcp crystal structures of
other structures. Gschneidner observed that solid solu- 4d ánd 5d transition metal s as a function of the number of
bility is low when the crystal structures of the eompo- (s + d) electrons. The reference state in defined by eqn (20).
nents are different. In other case, structural effeets do From reference [641.
not play an important role and the other factors are the
deciding ones. For the purposes of applying this rule al!
the eommon metal!ic structures are considered to be l!;hCJ\'(~) + Efcc(z) + E () = O
the same. 2 I~c Z • (20)

Even if structural effects are smal! when the two

eomponents have one of the common metallic structures,
those effects are not negligible. This has been demons-
trated reeently by Miedema and Niessen [63.64]. These
authors have studied the structural contribution to the
enthalpy of solution of alloys of two transition metals.
This contribution reflects the fact that there exists a pre-
ferenee for the transition metals to crystallize in one of
the main crystallographic struetures bcc, fcc or hcp,
dependingon thenumberofvalenee (s + d) eleetrons per
atorn. It is then reasonable to assume thatalso structure
dependent energies in transition metal solid solutions
will vary systematically with the average number of
valenee electrons per atom (z) if the two metals form a
eommon band of d-type electronie states. Consequently,
disolving metal A in the host B changes the energy that
stabilizes the crystal structure of the matrix metal, This
energy ehange will be proportional to the difference
(z" - zB)' For a more quantitative description, the varia-
tion of stability of the particular crystal struetures with
respect to z is needed. Combining experimental [65] and
theoretical data [66-69] for the energy differences
E¡,.e- Ereeand E¡,ep- Ercebetween bcc, fce and hcp structu-
Solid metals with z = 5,6,8 and 9 have a struetural stabi-
lit y which is large compared to that for solid metal s with
z = 3,4. 7 01' 10. This suggests that the entropy of fusion
of the former will be higher than the average value for
metals, Sueh an increase has indeed been observed for
Nb, Ta, Mo, W and Ru [64, 70].
Using these structural enthalpy differences, the
ehange in structural stability of the matrix upon solution
can be predicted. The structural eontribution to the ent-
halpy of solution of A in B can be written
i1h:"(A(O'A) in B(O'B» = ((z - z ) dEuA(z)
A Il dz
+
This equation contains two contributions, One of them
(EaB(B) - EaA(A) comes from thedifference in structural
enthalpy of the two pure metals. The second term
(z, - Zll) dEuB(z) / dz is the change of struetural stability
ofthe O'B structure due to thechange in averagevalence.
For example, for hcp-Hf (z = ) and hcp-Re (z = 7). the

10
 Revista Latinoamericana de Metalurgia y Materiales, Vol 5, N° 1, 1985

matrix is destabilized upon dissolving either Hf in Re or
Re in Hf (see Fig. 9). Niessen and Miedema have calcula-
tr
ted the structural contribution L1h: to the heat of solu-
tion for all combinations of two transition metal s [64,71]
showing that L1h:tr represents a substantialcontribution
to .ó.h,; in some alloys.
The theory presented in Section 9 can now be impro-
ved by adding the structural contribution L1H~t. to the
heat of formation .ó.Hs of a solid solution
alloys, with the structural contribution considered, are
siven in Table 1. Finally, we have computed solid solubi-
Iity limits in other class of alloys: Those having interme-
diate compounds. Figure 11 illustrates the method. The
free energy of formation of the solid solution is first com-
puted as a function of concentration. On the other hand,
the free energy of formation L1G(, of the stoichiometric
compound AOB¡i (with composition taken from the expe-
rimental phase diagram) is computed as

(22) (24)

For a very dilute alloy (x, « 1) , L1H:tr may be written where L1H(' is given by eqn (11). In writing (24) the
entropy offormation is assumed to be zero (strictly spea-
king, very small with respect to the entropy of formation
of disordered solid solutions). Then the tangent rule is
applied to compute the extent of solid solubility. Results
Evidently this formula is not valid in the concentrated for several alloys are given in Table 2 and the agreement
regime. Then we apply a similar idea to that used in Sec- with experiment is remarkable.
tion 9. The concentrated alloy is formed in several steps.
In the first step a small amount of solute is added to the
host and eqn (23) can be used to compute the structural TABLA 1
contribution. At this point the host is considered as an
effective pure host with atoms having the concentration THEORETICAL AND EXPERIMENTAL SOLID
averaged valence. Then a few more atoms are added to SOLUBILITIES (IN AT %) AT THE INDICATED
this effective host and the corresponding structural con- TEMPERATURE
tribution is cornputed again from (22). The process is con-
tinued and any concentration can be reached after a Alloy T(K) Theoret (1) Theoret (2) Exp
sufficient number of steps,
The applications ofthis theory to différentclasses of Se in Nb 1500 0.07 0.015 < 0.1 [581
Nb in Sc 1500 0.18 0.025 < 0.3 [58]
alloys will be published elsewhere [72]. Here we only Se in Cr 1350 < 0.001 0.14 < 0.1 [58]
show a sample of results. Figure 10 is a plot of the free Cr in Se 1350 0.22 0.31 < 1 [58]
Se in V 1570 0.024 < 0.3 [58]
V in Se 1570 0.45 < 0.5 [58]
Th in V 1600 < 0.001 < 0.1 [59]
ZrSc V in th 1600 0.012 0.05 [59]
Th in Ti 1000 0.008 O [60]
Ti in Th 1000 0.004 O [60]
-O Ti in Y 1140 0.005 0.0003 O [59]
~ 1. Y in Ti 1150 < 0.001 0.05 [59]
---, f La in Ti 1100 < 0,001 0.07 [58];' 1.4 [61]
~
~
1 (1) Without struetural contribution.
V)
'\ I (2) With structural eontribution.
<.é)
<l 2. -. <, /
/

11. SUMMARY
----- . -._.- »:

The different factors influencing the extent of equi-

o 02 04 0.6 08 librium solid soluhility-of a solute metal in ahost metal
have been discussed. Since the stability region of a subs-
titutional solid solution is obtained by comparing the
Fig. 10. Calculated free energy of formation of a ZrSc solid solution
free energy of formation of the solid solution alloy to
at T = 1000 K, as a function oí eoneentration. Continuous those of competing phases (intermediate phases) one is
line has been computed taking into account the structura' led to study the different contributions to the free energy
contribution to the heat of solution. In contrast, this contri- of formation of alloy phases. The entropy of formation
bution was neglected in constructing the dashed curve.
(generally zero for intermediate compounds) is one ofthe
key stabilizing factors of a substitutional solid solution.
On the other hand, there are several important contribu-
energy of formation of a ZrCr solid solution as a function tions to the enthalpy of formation of alloy phases arising
of concentration. The dashed curve was calculated as from several intrinsic elemental properties characteri-
explained in Section 9. On the other hand, the continous zing the chemical similarity between the components:
tt
curve contains the structural contribution L1H: , which atomic size, electronegativity, electron density at the
isª, sizable part of L1G,. Calculated solubilities for several boundary of bulk atornic cells, generalized valence (to

11
 LatinAmerican Joumal oi Metallurgy and Materials, Vol. 5, N' 1, 1985

o ACKNOWLEDGEMENTS
Pd V
The writing of this review was started during a stay
(Summer 1983) at the University of Pennsylvania
(Department of Materials Scienee and Engineering) sup-.
ported by a grant frorn the Committee for Scientific and
-10. Technological Cooperation between Spain and USA. The
work has been supported by CAICYT of Spain. The aut-
o
hor is greatful to J. M. López and to Dr. S. Simozar for
<,
E collaboration in several topies covered in this review.

--
}
::::s:::
-20. REFERENCES

~ 1. . See for instance ref'erences 23 and 34.

<J 2. P. Haasen, Physical Metallurgy. Carnbridge University Press.
Cambridge (~978).
3. R. Hultgren, P. D. Desui, D. T. Hawkins, M. Gleiser and K. K.
-30. Kelley, Selected values ofthe tbermodynamic properties ofbinary
alloys. Arner. Soco Met. Ohio (1978).
4. A. E. Guggenheim. Mixtures. Oxford University Press. Oxford
(l952)

o 01 0.2 5. R. B. Me Lellan, Mater. SeL Eng. 9(1972) 121.

6. F. R. de Boer, R. Boom and A. R. Miedema, Physíca 101B(1980)
294.
7. F. R. de Boer, R. Boom and A. R. Miederna, Physica 113B(1982)
18.
,Fig. 11. Computation ofthe solubility limits oí Pd in V by the tangent
method at two temperatures. Plotted are the curves repre- 8. F. R. de Boer, R. Boom and A. R. Miedema, Physica 115B(1983)
senting the free energies of the Pd, V1- x solid solution, and 285.
the point representing the free energy of the PdV3 com- 9. B. M. Mogutnov and L. A. Shvartsman, Russian J. Physical Chem.
pound. The point of tangency marks the solubility limito 54(1980) 328.
10. W. Hume-Rothery, G. W. Mabbottand K. M. Channel-Evans, Phil.
Trans. Royal Soco London, 233A(1934) 1.
TABLA 2
11. W. Hume-Rothery, R. E. Smallman and C. W. Haworth, Structure
oí metals and alloys. Institute of Metals. London (1969).
THEORETICAL AND EXPERIMENTAL SOLID
12. J. T. Waber, K. Gschneidner,
A. C. Larson and M. Y. Princ~. Trans.
SOLUBILITIES (IN AT %). AT THE INDICATED Met. Soco A1ME 227(1963) 717. '
TEMPERATURE. IN ALLOYS HAVING 13. J. Friedel, Advane. Phys, 3(1954) 446.
INTERMEDIATE'COMPOUNDS
14. J. D. Eshelby, So lid State Phys. vol. 3 (Eds. F. Seitz and D. Turn-
bull). Academic Press, New York (1956) 79.
Solubility Solubility
Alloy Compound T(K) (theoret) (exp) [59J 15. A. Gscheneidner, Solid State Phys. Vol. 16 (Eds. F. Seitz and D.
TumbulI). Academic Press. New York (1964) 275.
16. T. Egami and Y. Waseda, J. Non Crystalline Solids, 64(1984)
Pd in V Pd V3 900 17 15 113.
Hf in V Hf V2 1273 0.5 0.6
Hf in V Hf V2 1473 1 1.2-1.5 17. T. Egami and V. Vitek, Computar modelling of structure and pro-
Hf in V Hf V2 1773 2.8 2.1-2.4 perites of amorphous metal s (Ed. V. Vitek). The Metallurgical
Ti in ir Ti 1r3 1500 14 13 Society of A1ME, Warrendale (1983) 127.
Ti in Ir Ti Irs 2000 16 14 18. L. Pauling, The nature oí the chemical bond. Cornell University
Press. Ithaca (1960).
19. R. P. Iczkowski and J. L. Margrave, J. Amer. Chem. Soc.
83(1961) 3547.
indicate, not just the number but also the kind of elee-
trons involved in bonding) and crystal structure. All 20. A, R. Miederna, P. F. de Chátel and F, R. de Boer, Phyaica,
100B(1980).
those factors have been first discussed separately and
21. R. S. Mulliken, J. Chem. Phys. 2(1934) 782; 3(1935) 573.
then have beenput together to develop a semiempirical
theory of solid solubility, based on the work of Miedema 22. L. S. Darken and R. W. Gurry, Physical Chemistry of metals. Me
Graw Hill. New York (1953).
and coworkers. This theory has been applied to several
23. J. R. Chelikowsky, Phys. Rev. 19B(1979) 686.
particular classes of alloys and good results have
24. W. Gordy and W. J. O. Thomas, J. Chem. Phys. 24(1956) 439.
been obtained.
25. K. A. Gscheneidner and J. T. Waber. The Rare Earths (Eds. F. H.
In conclusion, solid solutions occur when the two Spedding and A. H. Daane). John Wiley, New York (1961)
components of the alloy are chernically similar. Other- 386.
wise, phase separation or formation of intermediate pha- 26. K. A. Gscheneidner, Rare Earth Alloys. Van N ostrand. Princeton
ses will occur (1961) 69.

12
 Revista Latinoamericana de Metalurgia y Materiales. Vol. 5. N° 1. 1985
27. K. A. Gschneidner, J. Less-Cornrnon Metals, 4(1962) 107.
28. K. A. Gschneidner, Theory oí Alloy Phase Formation (Ed. L. H.
Bennett). The Metallurgical Society of AIME. Warrendale
(1980) 1.
29. D. K. Sood, Phys. Letters, 68A(1980) 469.
30. D. K. Sood. Radiat. Effects. 63(1982) 141.
31. E. Teatum. K. Gscheneidner and J. T. Waber. Report LA-2345
(1960).
32. D. A. Goodman. L. H. Bennett and R. E. Watson, Scripta Metall.
17(1983) 91.
33. D. A. Goodman, L. H. Bennett and R. E. Watson, Alloy Phase Dia-
grams (Ed. L. H. Bennett. T. B. Massalsli and B. C. Giessen). Else-
vier. Amsterdam (1983) 43.
34. R. E. Watson. L. H. Bennett and D. A. Goodman. Acta Metall.
31(1983) 1985.
35. J. A. Alonso and S. Simozar, Phys. Rev. 22B(1980) 5583.
36. A. R. Miedema, F. R. de Boer and P. F. de Chátel, J. Phys. F: Metal
Phys. 3(1973) 1558.
37. A. K. Niessen, F. R. de Boer, R. Boom. P. F. de Chátel, W. C. Mat-
tens and A. R. Miedema. CALPHAD. 7(1983) 51.
38. C. H. Hodges, J. Phys. F: Metal Phys. 7(1977) L247.
39. E. N. Kaufman, R. Vianden, J. R. Chelikowsky and J. C. Phillips.
Phys. Rev. Letters, 39(1977) 167.
40. R. Vianden, E. N. Kaufman and J. R. Rodgers. Phys. Rev.
22B(1980) 63.
41. B. D. Sawicka, Nucl. Instrum. and Methods. 182/183(1981)
1030.
42. B. D. Sawika andJ. A. Sawicki, Nuc!. Instrum. and Methods, 209/
210(1983) 799.
43. B. Rauschenbach and K. Hohmuth, Phys. Status Sol. 72a(1982)
667.
44. K. Hohmuth, B., Rauschenbach, A. Kolitsch and E. Richter, Nucl,
Instrum. and Methods, 209/210(1983) 249.
45. Z. Bangwei, Physica, 12IB(1983) 405.
46. J. A. Alonso and M. P. Iñiguez. Phvsica, 103B(1981) 333.
47. B. C. Giessen and S. Whang, J. de Physique, 41(1980) C8-95.
48. J. A. Alonso, J. M. López, S. Simozar and L. A. Girifalco, Acta
Metall., 30(1982) 105.
49. J. M. López and J. A. Alonso, Physica, 113B(1982) 103.
50. J. A. Alonso and J. M. López, Philos. Mag. 45A(1982) 713.
13
51. S. Simozar and J. A. Alonso, Phys, Status Sol., 81a(1984) 55.
52. H. Jones, Mater. Sci. Eng., 57(1983) L5.
53. J. M. López and J. A. Alonso, Phys. Status Sol. 72a(1984)
777.
54. B. C. Giessen. Proc. 4th Internat. Conf. on Rapidly Quenched
Metals. VoI.1(Ed. T. Masumoto and K. Suzuki). Japan Instituteof
Metals, Sendai, 213.
55. B. C. Giessen and S. Whang, Mater. Res. SocoSymp. Proc. Vol. 19,
(1983) 289.
56. J. A. Alonso and S. Simozar. Solid State Commun .. 48(1983)
765.
57. J. M. López and J. A. Alonso, Phys. Status Sol. 76a(1983)
675.
.58. W. G. Moffat. Binary Phase Diagrams Handbook. General Elec-
tríe Company, Schenectady (1976).
59. R. P. Elliot, Constitution of Binary Alloys, First Supplement, Me
Graw-Hill. New York (1965).
60. M. Hansen and K. Anderko. Constitution of Binary Alloys, Me
Graw-Hill. New York (1958).
61. E. M. Savitskii and G. S. Burkhanov, J. Less-Common Met.,
4(1962) 301.
62. J. M. López and J. A. Alonso, Phys, Status Sol., 85a(1984)
423.
63. A. R. Miedema and A. K. Niessen, CALPHAD, 7(1983) 27.
64. A. K. Niessen and A. R. Miederna, Ber. Bunsenges. Phys. Chem ..
87(1983) 717.
65. R. Hultgren. P. D. Desay, D. T. Hawkins, M. Gleiser. K. K. Kelley
and D. D. Wagman, Selected values ofthe thermodynamic proper-
ties of the elements. Amer. Soc. Metals. Ohio (1973).
66. A. R. Williams and J. F. Janak, Private communication to A. K.
Niessen and A. R. Miedema (1982).
67. O. K. Andersen. J. Madsen, U. K. Poulsen.J. Jepsen and J. Kollar,
Proc. Int. Conf. Magnet. (Eds. P. F. de Chátel andJ. J. M. Franse).
North Holland Publ. CO.Amsterdam (1977).
68. D. G. Pettifor, CALPHAD, 1(1977) 305.
69. D. G. Pettifor, J. Phys. C.: Solid Sto Phys. 3(1970) 347.
70. A. K. Sheindlin, S. A. Kats, B. Y. Berezin, V. y. Chekhovskoy and
M. M. Kenisarin, Rev. Int. htes Temp. Refract., 12(1975) 12.
71. A. K. Niessen and A. R. Miedema, Private cornmunication
(1984).
72. J. M. López and J. A. Alonso, manuscript in preparation.