Villars Combined

Copyrighted Materials
Copyright", 2000 John Wiley & Sons Retrfeved from YiYiv/.knovel.com
Chapter 1
Factors Governing Crystal Structures

Pierre Villars
Intermetatllc Phases Data Bank (IPDB) and Materials Phases Data System (MPDS), Postal Box L, CH6354 Vitznau, Switzerland
1. Introduction The fundamentals of the constitution of an alloying system are determined by the crystal structure of its intermetallic compounds and its phase diagram. Knowing these fundamentals enables scientists to solve many problems in materials science, and therefore it is important to have easy access to the experimentally determined data, especially as the experimental work to determine such information is very time- and costintensive. Recently, scientists working in this field have had the advantage of access to comprehensive, up-to-date handbooks for crystal structures as well as for phase diagrams. Pearson's Handbook of Crystallographic Data for Intermetallic Phases, second edition (Villars and Calvert, 1991) contains critically evaluated crystallographic data for over 25000 distinctly different intermeralhc compounds (over 50000 binary, ternary, etc., entries) covering the world literature from 1913 to 1989. The A tlas oj Crystal Structures jor Intermetallic Phases (Daams et al., 1991), a companion of Pearson's Handbook, contains for most of the intermeta1lic compounds in Pearson's Handbook graphical representations of the crystal structures. The handbook Binary Alloy Phase Diagrams, second edition (Massalski et 01., 1991) contains about 4000 phase diagrams, most of them critically evaluated phase diagrams. In 1994 ASM International will publish the Handbook of Ternary Alloy Phase Diagrams (Villars et ai., 1993), which comprehensively covers the world literature from 1900 to 1989 and will contain information for over 8800 ternary systems including over 15 000 isothermal
sections, liquidus and solidus projections, as well as vertical sections. For the literature up to the year 1977, there also exists the very comprehensive Multicomponent Alloy, Constitution Bibliography (Prince, 1956, 1978, 1981). Looking at the research activities in the last 20 years with respect to the number of systems investigated. grouped according to binary. ternary. and quaternary systems, a shift in the ongoing research from binary to ternary and quaternary systems is clearly seen. The research field of multinary systems is a huge field for potential novel materials with optimized known physical properties (e.g, tP68 BFel4Nd2-typepermanent magnets and, in the inorganic field, the new high-Z] superconductors) as well as 'new' physical properties, as the recent past has proven with the quasicrystals (see Chapter 20 by Kelton in this volume). All this was achieved by moving from the binary to the ternary and quaternary systems, but this trend confronts us with many additional 'difficult-to-handle' problems. We have to be aware that in the future most research will be done on rnultinary systems; therefore, in this chapter we will first discuss the binary intermetallic compounds (systems), also called binaries, and whenever possible extend the discussion to ternary intermetallic compounds (systems), called ternaries. We will try to show the main problems by extending to ternaries our knowledge gained by investigation of binaries and propose ideas which, according to the author, have a high probability of success in the quantitative extension to multinaries of regularities found in binaries. The experimental variables we have are the selection of the chemical elements, their possible combinations, their
Crystal Structures of Intermetallic Compounds. Edited by J_ H. Westbrook and R. 1. Fleischer 1995, 2000 John Wiley & Sons Ltd
Crystal Structures of Intermetallic Compounds be checked with the available experimentally determined data). To increase the efficiency in the successful search for 'new' interrnetallic compounds, the main efforts should go toward creating an internationally accessible information-prediction system incorporating all databases (experimentally determined facts) as well as generally valid principles and the 'highest-quality' regularities. The problems involved in such a project have been reviewed by Westbrook (1993). The combination of the experience and intuition of the experimentalist together with easy access to the data already experimentally determined in the form of upto-date handbooks as well as access to the envisioned information-prediction system would very much help to coordinate world research activities. Furthermore, it would reduce the number of unwanted duplications as well as increase the. probability of investigating first the most promising systems and not, as in the past, provide us with the systems in a random statistical sequence. Otherwise we will have to wait until the year 5500 for the next 140 generations of scientists to investigate the ternary systems, assuming an activity rate similar to that in this century.
concentrations, the temperature, and the pressure. Of enormous consequence from the practical point of view is the variety of combination possibilities of the chemical elements and the increase in the possible number of combinations on going from binaries to multinaries accompanied by a much larger available concentration range. Unfortunately, the number of available experimentally determined data is in inverse proportion to these opportunities. We have a relatively robust database for binaries, but a sparse database for ternaries, and almost no data for quaternaries. This is quantitatively demonstrated by looking at the relevant numbers of binary, ternary, and quaternary systems. Tn the binary systems, taking 100 chemical elements into account, there exist (lOOx 99)/ (1 x 2) = 4950 binary systems, a large number, but still, with united. coordinated international research efforts, all these systems could be experimentally investigated by the end of this century. Therefore, one would inevitably have found among these the most interesting and economically important binary materials. Massalski et 01. (1991) and Villars and Calvert (1991) contain information on about 4000 systems, so already we have a very robust database with about 80010 of the possible systems fully or partly investigated experimentally. With the ternary systems the situation looks completely
different. There exist (J 00 x 99 x 98)/(1 x 2 x 3) = 161 700

systems. Villars and Calvert (1991) and Villars et at. (1994) contain experimentally determined data for only about 8800systems, most of which have only been partly investigated. So here we have a sparse database with only about 5010 of the potentially available systems having been partly investigated. In addition, in order to establish structures and phase relationships, one has to prepare and investigate at least 10times more samples per system in the ternary case compared to the binary case. In the quaternary case with (100 x 99 x 98 x 97)/ (J x 2 x 3 x ,,) :::3".921225 potential systems, less than 0.1010have been partly investigated. Without having very good guidelines, it is a hopeless situation to search systematically for novel materials with an adequate success rate in multinary systems. Therefore, the only practicable way to go is, with the help of the robust binary database, to find regularities. such as laws. rules. principles, factors. tendencies, and patterns valid within the binaries, and then to extend these to the ternaries and quaternaries in such a way that the binaries and multinaries can be treated together; otherwise the regularities would be based on too few data sets and therefore would not be trustworthy. In addition. the regularities should show an accuracy clearly above randomness to be of practical use (these can easily
2. Strategy to Find the Factors Governing Crystal Structure .

When we taJk about factors governing crystal structures we intend to reduce these to atomic property expressions (APEs) of the constituent chemical elements, so that the existing experimental data can be systematized. For the purpose of this chapter, we consider that the only regularities of real practical value are those able to systematize a large group of data with an accuracy in the range of at least 95010. For the not yet experimentally investigated systems, one goes in exactly the reverse way, assuming the validity of the regularity found for a welldefined group of data. The more data considered, the more trustworthy is a prediction based on such regularities. To make predictions one starts from the chemical elements, their concentrations. and the tabulated atomic properties (APs) of the elements, and calculates the APEs for a system of interest. Therefore, APs are only of practical interest where they are known for most chemical elements with adequate accuracy. From the APEs, in the context of the considered regularity, one can predict how the systems of interest most probably (with an accuracy of at least 95070) will behave. Then scientists can decide which systems are promising to investigate experimentally.

In principle, it would be sufficient to use as input only the atomic numbers (ANs) and the compositions of the intermetallic compounds of the systems under consideration. Slater (I 956) once made a comment as follows: 'I don't understand why you metallurgists are so busy in working out experimentally the constitution [crystal structure and phase diagram] of multinary systems. We know the structure of the atoms (needing only the AN] , we have the laws of quantum mechanics, and we have electronic calculation machines, which can solve the pertinent equation rather quickly!' Some 35 years later Chelikowsky (1991) writes in an excellent review the foUowing: 'Although the interactions in intermetallic compounds are well understood, it is not an easy task to evaluate the total energy of solids, even at absolute zero. As the energy of an isolated atom is in the order of about io-ev, but the cohesive energy
only in the order of about 1-10 eV/atom, one must have a method that is accurate to one part on 106, or better.' The other fact that greatly complicates evaluating the cohesive energy by theoretical methods is the number of particles involved. Given that a macroscopic solid may contain 1023 nuclei and electrons, it is impossible to determine the total energy of the crystal structure without some approximations. Within the last 15 years, two advances have made it possible to predict the cohesive energy of solids by numerical solutions of the quantum-mechanical equations of motion, e.g. the Schrodinger equation: the invention of high-speed computers and the device of one-electron potentials, which greatly simplifies many-body interactions. The accuracy of these computations is usually not at the same level as experiment. Nonetheless, it is now possible, for chemical elements and simple intermetallic
::::::I
~
(/)
c. +
ct
Fr Cs
Aa
Ac Hf TaW
Re Os Ir Pt Au Hg
Rb
K
Sa La Zr Nb Mo Tc Au Rh Pd Ag Cd Sr Y Ti V Cr Mn Fe Co Ni Cu In Ga
Sc AI
TI In
Pb
Bi
Po
At
Ca Mg Be
Sn Sb Te Ge As Se Si P S C
AN
Figure 1. An atomic property, AP, versus a function of atomic number, AN, plot 0 f the pseudo potential radii after Zunger, Rl ,representing an example of the size factor. In such diagrams the chemical elements are ordered by increasing group number each group is ordered by increasing quantum number. This special atomic-number (AN) scale was used instead of a linearly increasing atomic-number scale because it gives units of comparable size for distance between the atoms within any group as wen as along any period. The scale chosen for the comparison, of course, does not affect the number of factors (groups), only the appearance of the patterns changes
a~~t
Crystal Structures of Intermetallic Compounds

to describe the alloying behavior. Villars (1983) conducted a survey of 53 different APs as a function of the AN (182 variables in all when the different methods of determination are taken into account), and it was found that there were only five main groups, here called factors. The results are best seen in AP versus AN plots. as shown in Figure 1 for the pseudopotential radii after Zunger (1981) for s, p, and d elements. The f ele~ents have been left out because in most cases only incomplete data are available. In all but a few cases (19 out of 182) very regular symmetric patterns were obtained in such plots; this means a regular behavior within a group with increasing quantum number (QN) as wen as along a period with increasing AN. In these diagrams the chemical elements are ordered by increasing group number and each group is ordered by increasing quantum number. This special 'AN' scale was used instead of a linearJy increasing AN scale because it gives units of comparable size for distance between atoms within any group as well as along any period. The scale chosen for comparison does not affect the number of factors (groups), of course; only the appearance of the patterns changes. It should be stressed that the equivalence of APs belonging to the same group is of a qualitative, not a quantitative nature. Adherence to one or another of the five factors is very obvious. Table 1summarizes the idealized characteristics
compounds, to predict whether a given crystal structure is the most stable one at 0 K and 1 atm. We have to stress that the crystal structures (and from Pearson's Handbook (Villars and Calvert, 1991) we know there are at least 2750 different ones) have to be given as input for firstprinciples calculations, and this for each potential intermetallic compound. Assuming the potential intermetallic compound crystallizes in one of the 2750 known crystal structures and knowing its nominal stoichlometry=All, A~> etc.-it may still, in some cases, require a few hundred first-principles calculations for each potential intermetaUic compound; even with high-speed computers, this is not yet workable. But the larger problem is that the differences of the cohesive energies of those few hundred calculations will be so small that the accuracy would have to be in the range of one part in 109 to determine the most stable crystal structure. The complexity of the above-mentioned problem shows that one cannot expect that within the next decade the constitution (crystal structure and phase diagram) of multi nary systems will be calculated from first principles. Meanwhile it is therefore sensible to adopt semiempirical approaches based on the experimentally known data to search for the most reliable regularities. As at the moment it is impossible to start only from the atomic property AN, it is obvious to try to find out which other APs of the chemical elements are needed
R;+
Table 1. Idealized characteristics of the 'patterns' in the atomic property (AP) versus atomic number (AN) plots of the five factor (groups)
Long periods
Factors (groups) Size factor Atomic-number factor Cohesive-energy factor First Second Third s elements Straight lines with negative slopes and a slight maximum around the Ni and Cu group elements Straight lines with positive slopes
Groups d elements p elements
Straight lines with positive slopes Straight lines with positive slopes
The first two
sand d elements Straight lines Line with two Line with a clear Line with a clear with negative slopes maxima at the maximum at maximum at V group and the the V group the Mn group
Co group
and the last two are irregular; the rest are straight lines with positive slopes
Straight lines with negative slopes
p elements Straight lines with negative slopes Electrochemical factor sand d elements Lines with a maximum around V to Ni group p elements Straight lines with positive slopes Straight lines with negative slopes The first two Straight and the last two lines with are irregular; . negative the Test are slopes straight lines with positive slopes For every group, horizontal straight lines
Valence-electron factor
Straight lines with positive slopes and a discontinuity at Cu

Table 2. Atomic properties (APs) of the chemical elements grouped according to the five factors (groups)
Factors (groups) Size factor
Element property determined experimentally or derived from a model Classical crossing points of the self-consistently screened. non-local atomic pseudopotentials Radius of maximum radial electron density for outer orbitals from Hermann- Skillman calculations" Renorrnalized orbital radius Radius calculated by Hartree-Fock-Slater method Ionic radius Softness parameter Atomic volume Metallic radius Covalent radius Reduced thermodynamic potential at 298 K Electrochemical weight equivalent Entropy of solid elements at 298 K Density Atomic number Atomic weight Principal quantum number Atomic electron scattering factor Bond energy of deep-lying electrons Specific beat Wavelength of K and L series Maximum number of electrons in the solid element Melting point Boiling point Heat of fusion Heat of vaporization Heat of sublimation Energy for atomization of I mol of the solid element at 0 K Bulk modulus Young's modulus Torsion modulus Compression modulus Crystal lattice energy Dimer dissociation energy Surface tension Liquid-solid interfacial energy Enthalpy of formation or monovacancies Cohesive energy Solubility parameter Compressibility modulus Linear coefficient of thermal expansion at 273 K Electronegativity Chemical potential (after Miedema)" ~3 (after Miedema)" s e'ectron binding energy s-p parameter Positron annihilation rate Electron affinity Hardness Normal electrode potential First ionization potential Term value (after Herman and Skillman)" Number of valence electrons (corresponding to group number) Number of vacancies or holes in the d bands above the Fermi level
Number of data sets
15
6 6 3 3
2 6
3
Atomic-number factor
3 4 2
Cohesive-energy factor
2
2
7
6
5
4
3
4 4
3 3 3
2 3
4
2
13
Electrochemical factor
2 3 5 3 3
Valeace-etectron factor
"Herman and Skillman (1963).

bde Boer et al, (1988).
6 10
7
Crystal Structures of Intermetallic Compounds energodynarnic potential (Volchenkova, (989), and vibrational frequencies and dissociation energies of homonuclear diatomic molecules (Suffczynski, 1987), etc.: however, all of them could be assigned to one of the five factors chosen. In Section 4 we will always check to determine to which factor(s) the AP and respectively the APE belong that are successful in organizing large groups of experimentally determined data. At the end we will find out which of those five factors are governing crystal structures.
-sa
E c Q)
Q)
5 3 2
~ > .!.
:>-
2' Q)
> 'u,
.c.
Q)
0 o
0.7
3. Compound-Formation Diagrams
A fundamental question that has to be answered first
before discussing crystal structures is: 'Which systems form at least one "new" intermetallic compound?'
figure 1. Plot for the chemical elements of the relationship of two atomic properties, APs, belonging to the cohesive-energy factor (group), namely cohesive energy and melting point T (Chelikowsky, 1979)
3.1 Binary System The most outstanding model for binary systems is Miedema's model, which is excellently explained in the series Cohesion and Structure, volume I (de Boer et al., 1988). Two APs of the constituent chemical elements enter the description of enthalpies of formation AJlfor: the chemical potential for electronic charge (electronegativity) _xM and the electron density at the boundary of the Wigner-Seitz atomic cell, Nws. Both APs belong to the electrochemical factor group. Table 2.4 of de Boer et al. (1988) gives the most recent recommended values for _xM and nws for most chemical elements. The key expression for the sign of tbe enthalpy of formation tJIfor of binary alloys is
of those five factors in AP versus AN diagrams, and Table 2 lists the 53 APs grouped by those five factors. Figure 2 shows two APs belonging to the same factor plotted against each other. the cohesive energy versus melting point T of the chemical elements. As a first approximation a linear dependence may be seen to exist. The five factors (groups) are:
Size factor
Atomic-number factor Cohesive-energy factor Electrochemical factor Valence-electron factor In Table 3 we have given, in periodic table representation for the chemical elements, an AP for each factor for which accurate and complete data are available as wen as (and this is very important for practical use) values that are independent of the constituent chemical elements of a compound. These are: pseudopotential radius after Zunger (1981), ~+p; atomic number. AN; melting point, T; eleetronegativity after Martynov and Batsanov (1980), XM&B; and group number = number of valence electrons, V. It is clearly shown in Table 3 that there exist no overall tendencies between those five factors. so they represent the most independent factors. In recent years some other APs have come to the attention of the author like, for example. average electron distances in the structures of the individua1 elements (Schubert, 1990), the
P and Q are constants for certain groups of systems,

e.g. systems containing only transition elements. The concept that the energy effect upon alloying is generated solely at the contact surfaces between dissimilar WignerSeitz atomic cells results consequently in the fact that AHor does not vary with the concentration as long as the basis atomic cell A remains fully surrounded by dissimilar atomic cells B. For intermetallic compounds the elastic mismatch effect will be of no practical importance, since only crystal structures that are favorable for the given constituent chemical element sizes will be realized. Therefore, relation (1) directly applies to intermetallic compounds ABy if they are sufficiently rich in B such that A atoms are completely
Ln
0..
"0
co
~
U1
N
t1}
..
:r.
CD
II
Crystal Structures of Intermetallic Compounds nws diagram corresponds to a particular binary system. Since information on phase diagrams is easier to retrieve than numerical values of in assigning' + ' or '.' to the points in this plot, the following criteria are used: '.' in the binary systems with one or more intermetalIic compounds that are stable at low temperature, indicating f).}lOf..<_ 0; and' +' if there is no intermetallic compound in the binary system or if both terminal solid solubilities are smaller than 10 at .010, indicating /).J{for > o. There are complications in binary systems combining non-transition metals with transition metals. If the analysis is performed in the same way as in Figure 3, the demarcation line deviates from a straight line, and there appears to be a systematic deviation, which can be resolved more favourably by recognizing a dependence on the number of conduction electrons of the non-transition metal. Another semiempirical approach by Villars has also proven to be successful and is described in Villars (1985). The demarcation lines (surfaces) are much more complicated and consequently less trustworthy; therefore we recommend the use of Miedema's model. Nevertheless in Villars' model one gets, for the case where Interrnetallic compounds are absent in binary systems, additional information as to the following four phase diagram types: complete solid solubility, complete solid insolubility, simple eutectic, and simple peritectic systems. Figure 4 shows a part of a chemical element versus chemical element plot (Villars et al., (989). Each field represents a binary system. for the systems where experimentally determined phase diagrams show the absence of intermetallic compounds, 'x' is given. Where no experimental data are available, predictions are given indicated by , " ' (after Miedema) and C /' (after Villars, only given in the cases where Miedema gave no prediction). Figure 4 shows that, upon alloying, the s-s, d3-7_dl-1, f-f, s_d3-7, s-f and dJ-7_f chemical element combinations almost always result in the absence of intermetallic compounds; see clement designations in Table 3.
surrounded by B neighbors, and therefore the enthalpy of formation IlHfor of the intermetallic compound AB,)' per mole of A is equal to
Il.}{for _ /1lfnterflJ/:,C
.... in B AB)' -
su=.
(2)
Assuming that Q and Pare truly constant for arbitrary choicesof the constituent chemica]elements.,..tbc.sign of the enthalpy of formation is simply determined by the ratio
(3)
For W> QIP the Il.}{'or value is negative, while in the opposite case the value is positive. The analysis of the sign of the predicted and experimental sn= <0 is demonstrated for binary systems including two transition metals (Figure 3). Each symbol in the _xM versus
sn=
". 'j, .
,
+ +
,..
_:
."
.... . .. .. .. .
.
.+
..
.'
'. .. '. -I"
:. ':, ,"
... .. ..... .'..I. .
.:' : ..:: ...... ~.: + + .. ~ + + + +

..... "
'1."
"~""~. ;,.;r
' ... +
. ':.:-. ., .......... + ........... +
'...."
...... .'
, "...., 'J:
1./
,J/
..
+ +
~'" ~
+
'i~: W++ +
3_2 . Ternary Systems

Attempts to extend Miedema's model to ternary systems failed. The extension of compound-formation maps to ternary systems is described in Villars (1986). In the ternary systems we face two additional major problems: only very few ternary systems are fully determined, and for many of them only a part of an isothermal section has been experimentally determined. During an extensive literature search (Villars and Calvert, 1985) we found in 1984 information on 5598
+r
++
...+4-#
++ +. +~
0.5
jnnl13'NS I (au.)
1.0
F" ... e 3. Miedema's two-dimensionalcompound-formation map for binary systems containlng two transition metals. For the meaning of (he' + 'and '.' symbols,seetext; each symbol
represents one binary system
Element B
f
(I lib It No
LlI80
Sf Co Vb b k
Y L.. T", , H9 0, h &4 ~
N~ p, Co LA,., "'
d7 I ~ V!/VVYl/l/ V ~7~VV/VV/ d6 MoIX x X x x x X IX
I',
X~iX~~~X~
X IX X X X X IX X IX
x l/iX
II Til I, Hf Ti T~ ltD V W ~
Crl"'l
~V~~XX~
X IX
x~~
)<
X~
wVl/Vl/Vl/VV
IX~~
IVVVVVVVVV
d5
NbVVVVV'/VV ~VV////VV
TiVV
Hi [/
,'''xx,~
/1'\1\ ~ XVV
XX X '\~
/~~~~~~~~,
/~
V~~XX
'xx vx~x~
IX, IX
dIt j
X
)c
I /VVrx
<
nV/~Y~VVVX IXX f II VVV!XVVVVI'\)c 1/ 1/ 1X)c ~~~~~~ ~"I'\c\''\' ~I'\~/~~//V~)c ~~~~~X~ d3 I Lo VVVI/I/ )C Vv IX X X iX~~)o
~~~~~~)C~~~)~ ~VIX
~Vv
VVVV~
x r-.,['\C\"
~I'
~X~~V)(V
~
I)( )C
)(X~
!"\ /1\ ,
'\)<
~
V
~
;>c
~ )(~X f'\.
~)()(
~)(
IX
IXIX
VV
",'~I',~
')( 1)<
x 1'\ ><
IX
XIXX
~
IX
E
~
I~
E
~ W
~!XI'\~~~><I'\I'\I'\I'\V I'\I~I'\~ ~~I'\~~' IXIXIXI'~~I~ ~ :~~~~~~><~~I'\I'\/I\~~I"I'I'\~~ ~~I'\~I/V)cVl/><f'\.x IX n~I'\~~ ~ )c1'\1'\ I'. IX :, >< )<
;~~I'\~~'\)C~I'\
~
~~~
XIX
11'\)(
Hcli'I',~Ki\.><f'\.:".
[, ~~~~~
d31
f
S
X~i\. ~IX > 1'\1'\. 'h~~~~~X~i'.'I'\f'.I'\~ LIIi"i'\['\.~I'\Xl'\iI'\ ~r-.,I'\
I'\')C~I,\ >< XX "r-.,>
21
I YbI\""~~><~D<;
Co $I
I/I/I/I/I/I/v IX Scl/VI/VI/VVI/><!X Ell ~ '" 1\ 1'\ I' X1'\ IX

y
IX IX
)c
I)c IX
IX
XIX
110
Li
rx
1'\
sl
1/ ~ D< '" Wal/I><~ K IXI>< ~~ ~
Figure 4. Part of a chemical element versus chemical element plot containing the experimental data and predictions for compound formation or respectively its absence in binary systems. For the meaning of the symbols, see text
ternary systems, which represents only 3.601. of the possible 161 700 systems. Furthermore. looking at the alloying situation from the constitutional point of view, the situation becomes definitely more problematic. Having in mind the two additional difficulties. which cannot be avoided, one can only have confidence in a prediction for ternary systems if it is possible to separate the-two groups (compound formation and respectively its absence) in one diagram with relatively simple demarcation lines in a two-dimensional diagram or respectively demarcation surfaces in a three-dimensional diagram. The absence of compound formation is experimentally established for 550 ternary systems (0.30/0).Based on the experimental fact that no ternary
system with compound formation is known whose binary systems are all of the type that do not form intermetallic compounds, we included all combinations (ternary systems) that are surrounded by the 583 experimentally estabhshed binary phase diagrams showing no compound formation and ended up with 1602additional ternary systems where the absence of intermetallic compounds is expected (assuming the correctness of the above-mentioned statement). All together we thus know 2152 ternary systems in which no <new' ternary intermetallic compound is formed (1.3'70). To assign a ternary system to the compound-forming group, we do Dot need to have a complete isothermal section, since knowledge of a single ternary compound with a fully determined or
10
Crystal Structures of Intermetattic Compounds
2.5
."
..
(5)
and the sum of the ratios of the melting temperatures in kelvin,

(6)
ci +
N
I/)
IX:
<I
o
-+
"II/)
1.5
<
ct
<I
+
al
-c
ci +
(/)
ct
<I
Fagure5. ViUars' three-dimensional compound-formation plot
of the section
fOT
VB = 2 for ternary systems. For the meaning
of' 11 '. '0' and '.' symbols, see text; each symbol represents one ternary system
assigned structure is sufficient. For all such ternary
systems we checked where the nearby binary intermetallic compounds have the same crystal structure. Where we found this situation we assumed a solid solubility range between the binary and ternary intermetallic compounds and therefore excluded those systems from the new-compound-formation group. By adjusting our coordinates used for the binary systems (Villars, ]985) to the ternary systems, we came to the following coordinates: .the magnitude of Zunger's pseudopotentiaJ sums,
(IM;+Ll.ABI
where TA> Ts> Te. The values for R;+P' V, and Tare given In Table 3. In one three-dimensional diagram we plotted the 2152 systems where the absence of intermetallic compounds is observed and 5048 systems where compound formation is observed. For the user's convenience we made our plots on 12 sections with various isovalenceelectron difference (IAVASI + IAVAd + l..i Vsd)/3 values. Figure 5 shows such a section for (I..i vABI + l..i V"c! + !AVBc!)/3 = 2. The symbol 11 ' stands for no compound formation determined by experiment; '0' stands for no compound formation by extrapolation from experimentally established binary phase diagrams, all showing absence of intermetallic compounds; and '.' stands for compound-forming systems. With a relatively simple demarcation surface it was possible to separate these two groups satisfactorily. Figure 6 shows the schematic three-dimensional formation diagram. The separation surface is based on 2152+ 5048 ternary systems and is accurate to 940'/0, which might be in the range of experimental accuracy. It is nicely seen from Figure 6 that the absence of 'new' intermetallic compounds is found along the three coordinates; this means that the magnitude of any one of the factors, i.e. the pseudopotential radii difference expression. or the valence-electron difference expression, the melting-point ratio expression of the constituent chemical elements, has to be small. In the cases where the values of any two or all three expressions are small, we also have absence of intermetallic compounds.
4. Regularities in IntermetaUic Compoonds We have tried to summarize the most outstanding publications about regularities and gave special preference to the ones which included in their investigation large groups of well-defined data sets, in the range of hundreds to thousands of data sets. Again, whenever possible, we discuss first the binaries followed by the ternaries (multinaries) in the subsequent approaches.
radii
+ IM~+P'ACI + IM~+P'BCI)l3
(4)
the magnitude of the difference in the number of valence electrons,
11
Compound formation
X;(18RZs+P,ABI
+ I.dRZs+p,ACI +18RZs+p,BCI)13
Y= (TA'TS + TA'Te + TS'Tc)13

Z=(jt1VABI
+ It1VACI + I~VBCI)l3
Figure 6. Schematic three-dimensional representation of Villars' compound-formation diagram for ternary systems
4.1 Atomic Environment Approach 4.1.1 Binary, Ternary, and Multinary Intermetallic Compounds
crystal-structure definition based on the space-group theory there exist at least 2750 different crystal structures; this number increases at a rate of about 2070 per year. From the practical point of view this high number makes it almost impossible to find rigorously valid, general regularities within crystal structures. The atomic-environment approach described here drastically reduces the number of distinctly different crystal structures by looking at the gross features of the crystal structures (see also Daams, Chapter 15 in this volume). The lattice symmetry and space group of intermetallic compounds are important for certain physical properties such as piezoelectricity and ferroelectricity, but for crystal chemistry they are less important in that a minor change in the position of the atoms in a crystal structure can often reduce its symmetry, therefore leading to a new mathematical description and a 'new' crystal structure. The resulting deformation of the atomic environment (AE) (for the cases where those atoms form a convex polyhedron, this is also called the coordination polyhedron), however. is often negligible, so that the AE does not change significantly.
We know that with the traditional
The aim of the AE concept is to define the AE as clearly as possible, so that we will ultimately be able to group these environments into distinct atomic environment types (AETs). With the help of the maximum-gap rule after Brunner and Schwarzenbach (1971), we define the AEs. All interatomic distances (d) between a central atom and its neighbors are plotted in a histogram as shown in Figure 7(a). The height of the bars is proportiona1 to the number of neighbors (n), and it is convenient to express all distances (d) relative to the shortest distance (dmin). In most cases a clear gap is revealed, as can be seen in Figure 7(a) for the cF4 Cu crystal structure. The AE, the cubooctahedron of Figure 7(b), is constructed with the atoms to the left of this gap. For cases where no clear gap is revealed, clearly defined rules are given in Daams and Villars (1992). It is important to stress that the method for defining the AE is not important, aslong as one finds all the 'idealized'. distinctly different AETs realized in intermetallic compounds. It is important also to realize that the AETs are not isolated building units; crystal structures can consist of interpenetrating AETs. Daams and Villars (1993)and Daams et at. (1992) investigated all cubic and rhombohedral interrnetallic compounds and found that the 14 AETs given in Figure 8 represent over 800/0 of all positions; this means 15500 positions from 19146 investigated point sets. Figure 9 shows their frequency within cubic crystal structures. The
12
(a)
25
(b) n
12-b
20
roo
dmin .= 0.255 nm
24
15 10
Maximum ~
r
II
5
gap
.
I
1.0
1.5
2.0
2.5
122.2(C) Cubooctahedron
Figure 7. (a) A typical example of a next-neighbor histogram and (b) the atomic environment, AE, a cubooctahedron (l2-b), constructed with the atoms before the maximum gap in this histogram for the cF4 Cu Crystalstructure. 122~c)is the polyhedron code, based on the number of triangles, squares, pentagons. hexagons, etc., that join each other in the different vertices. The code gives the number of equivalent vertices (with the number of faces in the above-mentionedsequence, separated by a period) as an exponent
tetrahedron, octahedron, cube, icosahedron, cubooctahedron and rhombic dodecahedron are strongly preferred within the 14 AETs. It might be noteworthy to realize that all of them are highly symmetric, but coordination polyhedra do not necessarily have to be symmetric. Preliminary results in the remaining crystal systems indicate that aU 2750 crystal structures can be described with the use of less than 20 frequently occurring AETs and the addition of some rarely occurring AETs (40-60). This enables us to reduce the 2750 crystal structures to about 300 coordination types, i.e. crystal structures having the same number and kinds of AETs. Crystal structures belonging to a specific coordination type have very similar gross features and differ only in details. Table 4 lists some crystal structures belonging to the 'tetrahedron' coordination type, crystal structures having just tetrabedron(a) as AE(s)- As one can see from this example, it brings together crystal structures from different crystal systems (cubic-monoclinic) and the number of atoms per unit cell varies between 2 and 72_ The results of this investigation strongly indicate tbat in intennetallic compounds the short-range order is very dominant and the long-range order is of much less importance. In addition, the short-range coordination is very much angularly directed, so that the 14 AETs of Figure 8 are strongly preferred. Another advantage
of this approach is that one can treat binary, ternary, quaternary, etc., intermetallic compounds as well as chemical elements in exactly the same way. When we consider the strong tendency of the atoms in crystal structures of intermetallic compounds to realize one of the 14 different preferred AETs, we are not able to associate this tendency with anyone of our five factors. Pettifor (1988b) has mentioned in a review about structure maps an additional factor, which he named the <angular valence-orbital factor'. The AE approach proves the importance of this factor on a very comprehensive scale. This factor has always been hidden by the fact that prior investigations were made on restricted classes of intermetal1ic compounds, such as for example '8 - N' compounds, sp compounds, interstitial compounds, etc., during the search for regularities. Mooser and Pearson (1959) used the average principal quantum number (QN) as a measure of the directionality of the bonds: the smaller the value of QN, the more directional the bonds. The QN belongs to the atomicnumber factor, but the angular valence-orbital factor in the context of the AET is much more precise and quantitative. It is worthwhile mentioning the work by Brunner (1988) in wbich be developed a simple iterative program to calculate the final position of N points from a starting position. considering only the well-defined
13
..,;
o
~ o
cti
o
"'IIi
(")
'i:,
c:\J ,...
..0
,....
C\I
"0
Q)
c o .....
L: S <h
!:l
.0 N ,...
14
l! CD
...
."
2500
..... 0
0-
...
CD
2000
.c
::::J
1500
1000
500
to I
0')
--------N N
I
to I
to I
.c o
I
to I
('I')
Observed AET
Fagore 9. A frequency plot of the 14 most frequently occurring AETs found in 5086 cubic intermetallic compounds
Table 4. Some of the crystal structures of the 'tetrahedron' coordination type 4-a (eN 4), together with some relevant details Pearson notation cF8 Space group
- -.. o__ ___ _. __ _.
.. .....
Number of compounds _ _,----_ .._ .. _Ternary 2

75 30
Prototype
-
_.
____
Binary
39
Quarternary
Total 2
156
Remarks
.'_"-.'-_
cPS
hP4 hR2 tP8 tP24 tl12
C SZn SZn BiO
F~3m
F43m
42
18
BecY'
CdP2 ReP2 CuZFeS4Sn CuFeS2 OPb AsCuS AsCu]S. MnN2Si AsCuSe Cu2GeSeJ P2Zn
uis
tl16 oPS oP12 oPl6 oPl6 opn

0112
P63mc R3m P4/mnm P432,2 I~/amd 1~2m J42d Pbcm Pna2, Pmn21 Pna21 Pbcn
Imm2
28
I 1 I 1
76
I 1 I
1 3 43 2 3 8
I
23
47
26
90
I
13
2 16 8
}
2 2
mP24
P2/c
Ideal Ideal Slightly distorted Elongated by 15010 Ideal Slighlly distorted Ideal Slightly distorted Nearly ideal Distorted Distorted Slightly distorted Slightly distorted Distorted Slightly distorted Distorted
"Thin films. IlHigh-temperature phase.
Factors Governing Crystal Structures Coulomb repulsion (N is a given number from 2 , ... , which is equal to the coordination number CN -1). It is interesting that, without knowledge of the actual bonding situation and therefore the appropriate potential (because it is only known in very exceptional cases), with such a simple potential function he gets seven of the 14 AETs. It would be interesting to go the reverse way and look at how the potentials have to be adjusted to get the 14 frequently occurring AETs as wen as the Jess-frequently occurring AETs. From that result it should be possible to get quantitative relations between bonding type (metallic, ionic, and covalent), AET, and the constituent chemical elements. 4.2 Structure Map Approach As Pettifor is writing a separate chapter in this volume (Chapter 18) about structure maps, we give here only a very abbreviated summary in the context of our chapter. pointing out some principal difficulties and future developments. 4.2.1 Binary Intermetallic Compounds Table 5 shows in summary for AB compounds the published structure maps together with some significant numbers: class of compound, map coordinates used, crystal structure (number of compounds included), total number of compounds included, and number of violaticns. Most of the above-mentioned structure maps were also done for AB2, AB). A2B3, and A3B5 compounds. It is also worthwhile mentioning the work of Watson and Bennett (1984), who discussed the occurrence or absence of Laves phases AB2 in terms of atomic volumes and d-band hole counts. Analysing the APEs of all coordinates listed in Table 5 that are successful in separating experimentally determined crystal structures of intermetallic compounds into distinct domains (including also the coordinates used for ternary intermetallic compounds, see below) shows that three of our five factors are involved: .Size factor (covalent radii Re, ionic radii Ri, metallic radii Rm, and pseudopotential radii Rs+p. Electrochemical factor (electronegatlviues X, Herman-Skillman term values, and s-p parameters). Valence-electron factor (number of electron vacancies per atom, number of valence electrons = group numbers, number of valence s + d electrons, and Mendeleev numbers), The principal quantum number QN belongs to the group of atomic-number factors and was only used by Mooser
15
and Pearson (1959) with a moderate separation effect. Mendeleev's number MN. introduced by Pettifor (1984), is a combination of the atomic-number and valenceelectron factors (equal to a one-dimensional representation of the periodic tabJe). 4.2.2 Ternary Intermetallic Compounds The extension of the discussion from binary to ternary (multinary) systems leads to two serious additional problems: (i) While about 9000 distinctly different intermetallic compounds (about 80010 of those possible) are known in the binary case, we are confronted with 25000 distinctly different intermetallic compounds as soon as we include ternary and quaternary intermetallic compounds. Although the number of 25000 intermetallic compounds looks overwhelming, it is in fact only a small fraction of the still unknown intermetallic compounds. If we consider 100 chemical elements we can define 161 700 ternary systems, and assuming three intermetallic compounds per system, we end up with about 500 000 ternary compounds (only 3% of which are now known). (ii) In the binary case we could restrict ourselves to fewer than 100 crystal structures and still include over 90% of all investigated intermetallic compounds, but the inclusion of ternary and quaternary intermeta1lic compounds requires that about 1500 of the most populous crystal structures be included to treat 90% of all intermetallic compounds. The past has shown us that it is impossible to reduce this number significantly by taking only 'ternary' crystal structures, as illustrated by the following example. In the 'old' Pearson's Handbooks (Pearson, 1958, 1%7) we find the tetragonal tIlO Al4Ba type with 14 exclusively binary representatives. Its successor (Villars and Calvert, 1985),20 years later, lists 17 binary and 450 ternary, mainly 1:2:2 representatives. e.g. BaMn2Sb2, NdRh2Si2 and LaNi2Sn2' Assuming one would need four different structure maps (e.g. each treating one class of isostoichiometric 1:1:1, 1:1:2, 1:1:3, and 1:2:3 intermetallic compounds), then each structure map would have to be able to separate 375 different crystal structure domains within the limited two-dimensional (three-dimensional) APE area (respectively space) available. It is very unlikely that it would be possible to separate correctly those inter metallic compounds, based on so few experimentally determined data, into 37.5 distinctly different crystal structure domains bounded by demarcation lines (surfaces). According to the experience of the author with binaries, at least 50% of all ternary information should be
16
Tallie S. Summary or the results of AB structure maps chat used APE as axes
Total number of compounds 157 Number
of
Reference Mooser and Pearson (1959)
Class of compound XY normal-valence compounds XZ normal-valence compounds XT normal-valence compounds XY normal-valence QN
Coordinates
VS.
Crystal structure (number of intermetallic compounds)
violations
9 3 3
Pearson (1%2) St John and Bloch (1974) Machlin et 01. (1977)
Watson and Bennett

(I978a)
Watson and Bennett (I978b)
Watson and Bennett (1978b)
Andreoni el al. (1979) Bloch and Schatteman (1981)
Zunger (1981) Zunger (1981)
Villars (1983)
cF8 CINa (l15), bP4 SZn (23), cF8 SZn (19) QN VS. I~AllIinlJl cF8 CINa (33), hP4 SZn (5), cF8 SZn (7), cP2 CICs (4) QN vs. 1.6....yi>8,tlill'l cF8 ClNa (22), hP4 AsNi (28), oP8 MoP (22) QN VS. I~aullllsi cF8 CINa (liS), hP4 SZn (23), cF8 SZn (19) R.:;,xIRc.Y _RJaB I vs. cF8 C (2), cP2 CICs (1), cF8 XNy8-N compounds IRJ~ p .. s.X p+.. Y IRJ&B +RJ~B CINa (27), hP4 SZn (8), cFS p- ',X p-s,Y SZn (24) RJ4B VS. cP4 AuCu (2) oC8 BCr (7), cP2 Fractionally bounded IRJ&B p+s,X p+"Y sub-octet XNyP-N IRIAB + R'l<B I CICs (16), oPS GeK (6), tll64 p-s.x p-.,y compounds NaPb (7), cP16 NaTI (7) (3~P~6) laxw.tBI VS. Nv TU compounds oP4 AuCd (I), tP4 AuCu (14), oC8 BCr (7), cP2 CICs (43), tP4 CuTi (6), tP30 CrFe (18) Non-octet XN-DyD 1~.tfI1 VS. Sx+Sy mPI6 AsLi (4), hP4 AsNi (2), tP4 compounds (N < 8) AuCu (3), oC8 BCr (12), oP16 CdSb (2). eP2 CICs (15), cFS CINa (5), cP64 GeK (6). tl64 NaPb (1), cF16 Nan (7) XU compounds axwe vs. s; bP4 AsNi (30), oCS BCr (I8), cP2 ClCs (43), cP2 OCs + another structure (15), cF8 CINa (45). cP8 FeSi (10), oPS MnP (22) RA -RA vs. cP2 CICs (18), cF8 ClNa (15), hP4 XY normal-valence IH',x P+',Y compounds RA SZn (4), cF8 SZn (3), cF8 C (1), +RAP-$,Y P-S,X hP4 C (I) XNya-N octet (RB&.~ + R8&S_ 1 oP4 AuCd (1), cF8 C (2), oP8 GeS o I T compounds (R~S)x - (R~+ (8), cP2 CICs (2). cF8 CINa (58), hP6 HgS (1), tP4 OPb (2), cF8 SZn i- R~s - R~S)y (27), hP4 SZn (8) VS. t (R~RUS) + (RB&'S + o X I R!.t:s)y X"'Y-N octet IRkH.x-R~H.yl vs. cFS C (4), cP2 ClCs (3), cF8 CINa compounds I p Z +RzP-s.Y I (65) cF8 SZo (29). bP4 SZn (II) -s.x MN non-octet AuCd (U). tP4 AuCu (27), IR_k+S.M - ~H.NIvs. oP4 compounds mP16 AsLi (4) bP4 AsNi (62), oC8 I Z + z I p- ,,111 p-,N BCr (41), oP8 BFe (16), cP2 CJCs (122). cFS ClNa (33), hP6 CoSn (3), tP4 CuTi (7), cP64 GeX (6), oP8 GeS (6), cPS FeSi (17), oCI6 KO (1), oP8 MnP (30), oC16 HgNa (I), oPl~ NaP (6), tl64 NaPb (7). mC32 NaSi (1), cF16 NaTl (7) AD compounds 20 structure types with at least six l:VAll VS. I~I (excluding hP4 representati yes vs. 1t.RZ. ...8 S"'p, AsNi type
1~aati031
49
72 157
9
0 4
62 45
89
63
15 4
183
24
42
J09
112 408
5 32
99
22
compounds
Burdett and McLaran (1984) TU transitiontransition metal compounds
Nru
VS.
4E~
cP2 CICs (45), tP4 CuT; (5). oP4 AuCd (3), hP2 CW (4), tP4 AuCu (3), tP30 CrFe (20), oC8 BCr (10)
90
(continued opposite)

Table 5 (continued) -----------------------------------_
.. ..
17
_ _-------Total
number Crystal structure (number of

Reference Pettifor (1988) Walzer (1990) Class of compound AB compounds XT simple metaltransition metal compounds Coordinates interrnetallic compounds) 52 structure types cF8 C1Na (72), cP2 CICs (43) of compounds
Number
of violations ?
-1000
115
Abbreviations used X, Y are non-transition elements Z are A group cations T. U are transition elements M,N s, p, and d elements A,B s, p, d. and f elements QN average principal quantum number _xPwliJli. W&l!. MItB electronegativity after Pauling; Watson and Bennett (t978a); Martynov and Batsanov (1980) Rp_,' ~+sI R~,A.z pseudopotential radius after sr John and Bloch (1974); Andreoni et al, (1979); Zunger (1981) OI,D2.,.1l,22 pseudopotential radius after Walzer (1990) ~~l renormalized orbital radius after Bloch and Schatteman (1981) N; number of electrons vacancies per atom S s-p factor N average number of valence s+ d electrons per atom V number of valence electrons (group number) EHliS d-orbital energy after Herman and Skillman (1963) MNP Mendeleev number after Petti for (1988a)
R:
experimentally known (- 20 x what we now know) to be abJe to draw such 'finely tuned' complex demarcation lines (surfaces) that the results would be trustworthy. To 'overcome' this principal problem one has to develop structure maps that include all data, which means binary. ternary, and quaternary intermetallic compounds (da1tonide and berthollide), to increase the number of included data sets. and to try to reduce the number of 2750 crystal structures. One way to reduce this number is to group crystal structures with similar gross features into coordination types, and then to try to separate the different coordination types into distinct domains. Within a specific coordination-type domain would be located all crystal structures having very similar geometrical arrangements of the atoms. To our knowledge there exist five 'ternary' structure maps, of which the first three have restricted themselves to a very small subgroup of ternary intermetallic compounds. These are briefly mentioned below: Kugimiya and Steinfink (1968) were very successful in separating ternary AB2C4 (C = S. Se, Te) compounds in a K AS versus RCAI RCB plot into 12 different crystal structure domains. Here KAB = X~ X~/(Rte)2 where X~ and X~ are the electronegativities after Gordy (1946) of the chemical elements A and B. (R~)2= I,e (RCd+Rtc)2+(R~+R~c>2-i'1.155(RtA +Rt) is an ionic radius expression containing RCA' and
the ionic radii after Ahrens (1952) of the chemical elements A, B, and C. The electronegativities and ionic radii used are listed in Kugimiya and Steinfink (1968). Burdett et al. (1981) separated 172 AB2C4 spinels into normal and inverse spinels using an RHPB versus R;+P,A plot, where RHP are the crossing points of the non -local pseudopotentials. Hoverstreydt (1988) separated 202 equiatomic rareearth (A) transition-metal (B) silicides, germanides, and gallides (C) (ABC) with the help of a three-dimensional structural stability diagram. The three axes are: R m.e
T&G&W -
T&G&W m,A
~&B B
difference in metallic radii of C and A atoms as given by Teatum et 01. (1972) sum of group number V and periodic number PN of Band C atoms Martynov and Batsanov's (1980) electro negativity of the atom B
Res.
Rtc.
Pettifor (1988a) has extended his MN versus MN structure maps to pseudobinary and ternary phases. He could achieve this by introducing for pseudobinaries an average Mendeleev number MN. It is assumed that ternary and quaternary additions C and D go preferentially to the
18
A and B sites. respectively. Thus, the alloy (Axel -x) (ByDl_ yh may be regarded as the pseudobinary AB3 characterized by the average Mendeleev numbers: MNA =xMNA +(I-x)MNc and MNB=yMNB+(Iy) MNn. He did this successfully for AB and AB3 compounds. For mainly ternary borides, sulfides. seIenides, and tellurides he produced seven structure maps treating the following different groups separately: AB(C = s,Se, Te)~ AxBy(C = S,Se, ro,
M; = 2x(R;
+p
+ p,A -
R; P,B)+ 2x(R;
+
+ p.A -
R;
p,e)
+2y(R~+P,B-R~+p,c)
(12)
with x<y<z
and x+ y+z= 1.
= boron), AxBy(C = boron),

AB(C
with x> y with v> 1 with x<y
A....ByCz
and another two groups. The separation achieved for the most populous crystal structures in these ternaries is quite satisfactory, even though he had to subdivide the sparse database nine times. This was possible because ternary intermetallic compounds containing S, Se, Te, and B are overrepresented compared to the others. Villars et al. (1989) investigated the intermetallic compounds belonging to the single-environment types crystallizing in one of 47 crystal structures comprising 2511 binary , ternary, and quaternary intermetallic compounds. As a result it was found that one threedimensional structure stability diagram separates 2483 intermetaJHc compounds (99070)into the following five different coordination-type domains: 'tetrahedron type' (4-a), 'octahedron type' (e-a), 'cubooctahedron type' (12-b), 'twinned cubooctahedron type' (12-d), and 'rhombic dodecahedron type' (14-b), with an accuracy of 97%. To treat binary. ternary, quaternary, pseudobinary, and pseudoternary interrnetallics in the same diagram, one had to introduce the concentration as a variable in the following way: for binary intermetallic compounds, AxBy
(7) (8)
Figure 10 shows a section for the V range < 2.74. The extension of this approach to two-, three- and multienvironment types was tested with 5500 binary intermetallic compounds (Villars et al., 1989) and the results showed that, with the proposed demarcation surfaces in four such three-dimensional structure maps, it is possible to separate the intermetallics into different coordination-type domains with an overall accuracy of 980/0. Investigations are under way to include all 25 000 distinctly different intermetallic compounds in those four three-dimensional structure maps.
4.3 Solid-Solubility Map Approach

Under Section 4.2 we showed what factors have to be varied to change the crystal structure. Here we investigate respectively how little or how much the factors can be varied without changing the crystal structure. One would expect the factors found in structure maps to be the same relevant factors here. In addition, one would expect that the APs of the constituent chemical elements should be of comparable magnitude to get extended solid solubility.
4.3.1 Binary Systems

Alonso (1985) gives a beautiful review of the present knowledge, the important points of which are summarized here. In the early t930s Hume-Rothery proposed that three factors control the solid solubility of the chemical elements:
First Hume-Rothery rule (size factor)

Solid solutions are not to be expected if the atomic sizes of the solvent and solute differ by more than 15%. Second Hume-Rothery rule (electrochemical factor) The electrochemical nature of the two chemical elements must be similar if solid-solution formation is to be expected, otherwise compound formation is likely to occur . Third Hume-Rothery rule (valence-electron factor) A higher-valence metal is more soluble in a lowervalent metal than vice versa. These three rules form the basis of our present understanding, although they are only of a qualitative nature. Below, we summarize in short the six most
tiRs+p = 2x(R s+p.A
-z
R s+p,B)
(9)
for ternary intermetallic compounds, AxByCZ
.6.XM&8 = 2x(X~B
- X~&B)
- X:&B)
+ 2x(X~&B
(11) - X~&B)
+ 2y(X~&8
19
2.0
1.5
. \
Octahedron ,
.
~
o
o
14-b Rhombic dodecahedron
<J
1.0
6-a
I I
I
o0c>
o.~
,
I
0 0 o
-:0.5
o
o o o
-1.0
I
I I
,
6-a
-1.5
14b
Rhombic dodecahedron
l
I I I
I I I
Octahedron
-2.0
I
I
.5
~XM&B
Figure 10. A section of the three-dimensional plot: t.]i7 versus AX'M&B versus V<2.74 for 2486 single-environment-type compounds (including binary, ternary, and quaternary dall~riide and berthollide interrnetalhc compounds). Each symbol indicates a crystal structure, '~a' stands for the octahedron type, 'l2-b' for the cubooctahedron type. and 'l4-b' for the rhombic dodecahedron type. Each symbol indicates a crystal structure within the coordination-type domain: '.'cF8 CINa; '0' cP2 CICs. '<l' cF16 BiF3 'I>' cF44 CeD3 '0' cF16 NaTI, 'v ' cF16SuHg2Ti,_:D' el2 ~ '+ teN AuCu3 x' tiS AI)Tl, ',,"' tP4 AuCu. Berthollide phases are given as lines (projections of the Von the t:.R s+p Z vs. ~M&B plot) as the three coordinates are concentration-dependent
important improvements toward realization of a quantitative rationalization with satisfactory accuracy. They demonstrate how much improvement can be achieved by refining the APE of those factors found by Hume-Rotherv.
(i) Darken and Gurry plots
In the early 1950s Darken and Gurry (1953) took an important step in the prediction of solid solubility when they were able to incorporate simultaneously the atom size factor R~ and electrochemical factor after Gordy and Thomas
20
22
, ,
0
Extensive solubility limited or no solubility
1.8
t0.5
iii
* Host: iron
o No information
)(
1.4
o.
1.0
.
0
1.0 1.4 1.8
RPauling

2.6
_0
2.2
(A)
Figure 11. A Darken and Gurry solubility plot for iron as solvent; Gordy and Thomas" eleetronegativities XG.tT versus Pauling's metallic radii R: (after Waber et al., 1963) (1956) XG&T by using two coordinates. They made this correlation more quantitative by constructing an ellipse centred around the solvent and with one axis of 150/0 of the solvent's size R~ and the other axis of 0.4 of XO&T of the solvent. Waber et al. (1963) made an extensive analysis of 1423terminal solid solubilities with the Darken and Gurry method, concluding that this method correctly predicts 85% of the instances of 1imited solid solubility. and extended solid solubility in 62010 of the cases where this is observed. Finally we notice that the separation of Darken and Gurry plots between soluble and insoluble solid solubility can be improved by drawing relaxed ellipses with axes of optimized direction and length. Figure 11gives a Darken and Gurry plot for iron: 'e' stands for extensive solid solubility ,t. ' for limited or no solid solubility, and '0' for no information available. For those iron-based solid solutions, the use of the first Harne-Rothery rule' leads in 90070 of the cases to a correct prediction for limited solid solubility (the distinction between a limited and extensive terminal solid-solubility alloy was taken as 5 at. 0/0). but only in 50010of cases to a correct prediction for extended solid solubility. Favorable size factor is a necessary but not a sufficient condition for extensive solution. (iij Chelikowsky plots
Solubility> 0.5% Solubility 0.01-0.5% ... Solubility < 0.01%
o
6
c Host: cadmium
o
Not measured
.. .. ..
...
o
00
0 000
o
o o
3
o o
..o
00
0.5
1.0
1.5
n113wS (a.u.)
12. Chelikowsky's plot for cadmium as solvent; Miedema's electronegativities XM versus electron density at Wigner-Seitz atomic cell boundaries nws Figure
Chelikowsky (1979) presented solid-solubility plots in which the following two AP

0
21
4
0
0 0
.5
"3
t1I CL.
:)
0 0
Ol
"
.,.,. .. -
It' ~
.
,,
0 0 0
0 0
.. Solubility> 15% Solubility 1-15% o Solubility < 1% Host: copper
*"
.......
o o 00
a::
,,
,, ,, ,
0
......~
.~-'
, ,
o 2~~---------~---------L
-160
-80
6Hsol
(kJ/mol solute)
80
160
Figure 13. Alonso's thermochemical plot for copper as solvent; Pauling's atomic radii for eN 12 R~ versus Miedema's thermochemical expression J1HSOI containing Miedema's electronegativities XM and electron densities at Wigner-Seitz atomic cell boundaries nws
coordinates are used: XM and n~~. Chelikowsky applied the new coordinates to the divalent solvents and demonstrated that they are superior to the Darken and Gurry approach. Figure 12 shows such a plot for cadmium: '.' stands for solid solution >0.5 at.OJo,'.' stands for limited solubility of 0.01-0.5 at. 0/0, , '" for no solid solubility (less than 0.01 at.%), and '0' for no information available. Alonso and Simozar (1980) proposed a new scheme that takes into account the coordinates XM, nws. and R!:. of the Darken-Gurry and Chelikowsky plots. As a good correlation exists beween XG&.T and XM, Alonso decided just to use XM. To preserve the simplicity of two-dimensional maps they combined XM and nws in one new coordinate: the chemical part of the heat of solution sn= as given in Miedema's theory; the other coordinate is R~, adjusted for CN 12. Figure 13 shows an Alonso plot for copper: '.6.' stands for a solid solubility of> 15 at.07o, '.' for a solid solubility between 1 and 15 at.OJo,and '0' for a limited solid solubility with less than 1 at. 0J0.
(iv) Zhang Bang wei plots {ii} Alonso plots
fourth, fifth, and sixth long periods and on 18 nontransition metals were investigated, For every solvent, the soluble chemical elements are separated from the insoluble by a parabola y = a + bil with the criterion of ~O.5 at.OJosolubility at room temperature (RT). The overall reliability was 90'10 for 2460 alloys, which is quantitatively the best one. Figure 14shows such a plot for titanium as solvent: '0' stands for solid solubility > O.S at. 070, ' I:. ' for 0.01-0.5 at. %, , x ' for < 0.01 at. 0J0 at 25C, and '.' for no information available. (v) Goodman's extension oj ,he third Hume-Rothery role Goodman et 01. (1983)observed a trend in relative solubilities summarized in Figure )5. This trend may be rationalized in terms of the band-theory concept: If the boundary region corresponds to the diagonal involving the dotted area, rather than the hatched squares, then the observed rule could be summarized by stating that whichever transition metal has more nearly half-filled d bands, that chemical element prefers being the solute rather than the solvent. The actual boundary is one set of squares to the left. The trend is describable as obeying a 'left-shifted' half-filled dband rule.
Solid solubility was studied by constructing plots with an APE y containing the valence, R ~, and _xl' versus an APE x containing the R~ilhellg of the solute with respect to the solvent (Zhang, 1985). The kernel radius R ~ equals approximately the positive ionic radius not including the valence electrons; its values are taken from Chen (1976). Some 2460 binary alloys based on transition metals of the
(vi) Crystal-structure condition The crystal structure of the two chemical elements also influences the extent of solid solubility. A substitutional solid solution preserves the crystal structure of the solvent. Complete solid solubility can only occur if the crystal structure of the two chemical elements is the same. Gschneidner
22
1.2
Crystal Structures of Intermetallic Compounds only significant work is that from Fleischer (1988), . who tested the following hypothesis: significant solid solubility of an element C in an intermetal1ic compound AxB)' is likely to exist if the intermetallic compound Axey has the same crystal structure as AxBr He took from Pearson's Handbook (Villars and Calvert, 1985) 28 AxB)'compounds with known crystal structure and melting point above J400C and searched for all AxCy compounds that have the same crystal structure (Axey were not screened as to the melting point). He found 170 such combinations; for 37 of those systems isothermal sections or quasibinary sections were available (Prince, 1956, 1978, 1981): 19 showed complete solid solubility, 28 showed extensive solid solubility >25 at.~o, and in only one system was the solid solubility less than 5 at. 070. These results indicate that the existence of quasibinary systems of isostructural binary intermetallic compounds with one common chemical element is a useful indication that chemical elements in the same crystallographic position can be substituted by one another. It would be interesting to do a complete systematic investigation without any restrictions. To the knowledge of the author the solid-solubility behavior of ternary intermetallic compounds has never been systematically investigated. However, instances are known 0f ternary solid solubility between berthollide compounds, between daltonide compounds, and even between berthollide and daltonide compounds (see Chapter 30 by Matveeva in this volume). The regularity discussed above belongs clearly to the influence discussed in binaries under (vi) 'crystal structure condition' and therefore belongs to the angular valence-orbital factor.
o Solubility>0.5% t:. Solubility O.Ol~.5% x Solubility < 0.01% "*' No information Host: titanium
0.8
O.L.
ox
a. a.6
X
t:.
*
<6X
60
X 0
a.
*
X
*-
*
20
i-:: x ....
40
Figure 14. Zhang Bangwei's plot for titanium as solvent; expression y containing valence, kernel radii, and Pauling's eleetronegativities versus expression x containing relative radii of solute to solvent
(1980) made the distinction between (a) common metallic structures (f.c.c., b.c.c., h.c.p., La, Sm and In) and (b) other structures. He observed that solid solubility is low when the crystal structures of the chemical elements are different. In other cases, structural effects do not play an important role and the other factors are then the deciding ones. It is important to realize that all the observations in the discussion above hold for the solid solubilities of the chemical elements and not for the solid-solubility behavior of berthollide intermetallic compounds within binary systems. Coming back to our five groups of factors, the three Hurne-Rothery rules and the improvements (i)-(v) can be grouped as to size, electrochemical, and valenceelectron factors. These factors are the same as the ones found with the structure map approach, which makes it nicely consistent. The crystal-structure condition belongs to the angular valence-orbital factor,
4.4 Space Filling Approach 4.4.1 Binary Intermetallic Compounds Laves (1967) formulated that, when metals crystallize, their individual bonding tendencies must adjust themselves to the properties of space, and this factor is primarily responsible for the formation of the crystal structures of metals. If we consider the atoms as spheres in contact (pure metal bonding), space will be filled in the best way when the atoms are arranged in the cubic or hexagonal closed-packed structures (c.e.p. and h.c.p.) with the AETs cubooctahedron (12-b) and twinned cubooctahedron (12-d) respectively (CN 12) and a space-filling ratio of 0.73. He called this tendency the space principle.
4.3.2 Ternary Systems

Very little work has been done in this area. The
23
Ni Pd Pt Co Rh Ir Ft> Ru Os Mn Tc Rt> Cr Mo W
---- ----
Ti
HI 'II d'
ooo 0 0 .~d.:'(S0
0 0
0 0 0 0
Ta...
Nb
Cr
W
Mn
0.. ..

0 0 ,0:".0:':": 0 0 0 . . .. :-':0:'0' :',>.: 0
:::'0','00.0 .. . ...
:'.:9:
0 0 0 0 0 0 0 0
'II Nb Ta
Re
00. 0 0
~~:L.-lo.-"'"
o o
A more solubl
e in B
in A
FE>
d7 Ru
mort> soluble
Figure 15. Experimental relative solubilities of transition-metal alloys. The dotted region indicates systems where the d bands of the alloy constituents are equally far from half-filled. The hatched region indicates a diagonal boundary region in which no dear bias appears in the relative solubilities
As most intermetallic compounds do not have purely metallic bonding, but in general an unknown mixture of metallic. covalent, and ionic bonding. this principle in his rigid form is not obeyed; as a consequence of this fact we gel the rich variety of 27.50 different crystal structures. The space-filling ratio. calculated as a first approximation with hard spheres and the appropriate radii sets (Ren Rj, Rc), ranges for intermetallic cornpounds between 0.65 and 0.85, thus not strongly violating the space principle of Laves, valid for metals, There exist three groups of quantitative indicators of the space principle, as follows.
(i) Geometrical stability plots These plots are also a kind of structure map using just traditional radii (Rm Ri, RJ as variables for the AP or APE coordinates. Compared to the highest developed structure maps organizing thousands of intermetallic compounds, they are inferior, but still for some specific questions are quite helpful. Geometrical stability plot after Villars et al. (J 982). They calculated, for the most populous crystal struc-
tures, generalized space-filling ratio plots as a function of "jfT&G&W and RT&GItW / RTItO&W where jiTltG&W = m m.A m,B, m + yRT&G&W)/(x+y) for A B where x<y. ( XRTltO&W m,A m.B x .Y R m,A T&G&W and RT&O&.w are the metallic radii after m,B Teatum et 01. (1972). Figure 16 shows such a plot for the tP30 CrxFey type (also called sigma' phase); all experimentally known compounds are found to be narrowly located in a restricted range of the space-filling factor RT&G&w / RT&G&W Jir&.G&W m.A m,B' m Geometrical stability plot after Ballestracci and Bertaut (1964). Figure 11 shows such a geometrical stability plot for ternary ABe2 compounds (A = Li, Na, K ; B::; Y and rare earths; C = sulfur) plotting R~A versus R~B' using ionic cell radii for sixfold coordination as given by Shannon (J976). As one chemical element is the same for all compounds, we classed this plot with the binaries.
cia (b/a) range within a crystal structure Villars and Girgis (1982) observed by examination of J 06 crystal structures containing more than five binary representatives as compiled in Pearson (1967) that over
(ti) Narrow
24
1.6
1.4
C(I
'-T .-
,
0.84 0.82 0.80 0.76 -
._ .._. __ .
..--_.-'-
3:
~
E 1.2 -
........
076
"< E-
3:
" 'a:
1.0 0.8
0.6
r-
t P30 CrFe
*' .~ .-.. .1~ .i'.

,/ .a.
At. ....
0.71. 0.74 -
r..
0.76
0.78 -
0.80 -
0.8
1.0
1.2
1.4
RTGWm
1.6
(A)
1.8
2.0
Figure 1'. Geometric structure field map, K~~~&W IR~~~w versus R~&G&W plot with generalized space-filling ratio isocontours for tP30 Cr_.Fey-typeintermetallic compounds
9S0J0exhibit very narrow ranges of unit-cell dimension ratios cia or bla for the representatives of a given crystal structure. Brunner (1977) investigated the two crystal structures that most violate the above-mentioned cia restraint, the hP3 AlB! and hP4 AsNi types, both crystal structures with fixed atomic positions, but very variable c/ a values of the compounds crystallizing in these two crystal structures. Figure 18 shows cia versus RT&G&'wIRT&.G&w plots in connection m,A rn.B with the change of the atomic environments (AEs) of both point sets of both crystal structures. From these it is clear that these are not just two crystal structures, but in fact five different crystal structures (coordination types, all two-environment types). Thus, these two crystal structures are one of the very few examples where the atomic-environment approach as compared to the space-group theory reveals a more complex structural situation; most often it is the opposite way round. Of course, from the geometrical point of view, these are five distinctly different arrangements (see also Daams, Chapter 15 in this volume).
(iii) Unit-cell parameter(s}-radii
geometrical relationships between unit-cell parameter expressions and radii expressions of the constituent chemical elements for intermetallic compounds crystallizing in a certain crystaJ structure. Table 6 lists some of the unit-cell parameter (expression) versus radii (expression) dependencies together with the crystal structures where each has been found. Figure 19 shows an example for intermetallic compounds crystallizing in the cF24 CU2Mg type (Edwards, 1972) and Figure 20 for intermetallic compounds crystallizing in the tP30 Cr xfey type (sigma phase), with x<y (Villars et 01., 1982). The aim is to predict dAB (bond length) distances of an intermetallic compound with the aid of normalized radii R and suitable functions. These functions should be as simple as possible and the dAB distances derived should have the same accuracy as those found by experiment. As tabulated metallic, ionic, and covalent radii (Rm, R, and Rc) are already normalized to a constant eN (in general eN 12 for a metallic bond, CN 6 for an ionic bond, and CN 4 for a covalent bond), one should know: CN of the atoms in the intermetallic compound.
dependence There exist a whole range of publications dealing with such
25
1.5
hR4CrNaS2 cF8 GINa D. cI28 P4Th3
.<Ii(
00Q:
1.3
1.1
0.9 0.9 1.0

RS.I,B (A) Figure 17. Geometric structure field map. an R~". versus R~1jplot for ABC2 compounds, where C is sulfur hP4 AsNi (SA) (a)
1.1
cia
1.2-2.2
cia
06-1.2
0.5-0.8
(b)
~ lSI [J $~ tf; !EJ

AI
17-24
1.0-17
Ni
As
1.9
cia
cia
1.7
1.5
1.3
1.0
1.2
1.["
I.S
1.6
0.6
0.6
1.0
1.2
RTGWm.AIRTGWm.B
Fi2ure 18. (a) Atomic environments (AEs) in the hP3 AlB2 and hP4 AsNi types as a function of the unit-cell dimension ratio cia. (b) Plot of cia versus R~?WIR~~;&.w (A;: AI, As; B;: B, Ni) of the representatives of these structures, showing three (AlBJ and two (AsNi) distinct groups, each' being a different coordination type
26
Unit-cell dimensionts) _ _._-,
.. ..
Table 6. Summary of unit-cell dimension(s)-radii dependences studied by various authors versus versus versus versus versus versus versus versus versus versus versus Radii dependence"
Rm,A Rm,A + Rm,B RW&.c/~&.G
m,A m,lI
Crystal structure t132 si,W s (A1Bs) tI6 MoSi2 (AB2) hP12 MgZnz (ABz) (in total over 20 crystal structures) tI48 BaCdll, cFl12 NaZnl3 cP36 BaHgIJ hRJ9 ThzZnl1 tP30 CrFe tP30 CrFe cF24 Cu2Mg (ABz) 106 crystal structures tII0 Al4Ba (ABzGez) cF24 Cu2Mg, hPl2 MgZn2, cP8 CrlSi and tP30 CrFe cF8 C1Na, cP2 CICs, hP4 SZn and cF8 SZn (270 compounds in one plot) cP8 CrlSi
References Wang and Taylor (1978) Raynor (1974) Pearson (1968) Pearson (1980)
Stuwe (1959)
c
unit-cell diagonal (RW&G_d )/d m.A A II a
a(c) a(c)
RW&G/d
m,A
RW&G
m,A
at constant
RW~G
m,9
jiW&O
m
jiW&O m
"I,A jiWI!i.O m
RW&GIRw&O
01,8
dAB
o(c) o(e)
R m,A W&C at constant

jiWI!i.O m
RW&C
rn.B
Wilson and Spooner (1973) Edwards (1972) Villars and Girgis (1982) Pearson and Villars (1984) Shoemaker and Shoemaker (1964)
d"'8
versus
RZlkc s
Zhang et al. (1988)
versus
6V~.A
+ 2V~,B
Paufler (1977)
"V atomic volume of Cr ~Si.type phases.
Bonding type realized in the intermetallic compound, to choose the appropriate radii set. Radius as a function of the presence of a second constituent chemical element.
1.0
1.2
1.4
1.6
RTGW rn,AIRTGW m, B
Figure 19 grKi&.w / [dobs + 02] (XGl!J.T - XGI!J.T\2] versus _Rf&1JAW/ m,A A 1\ B 'I m,A R~~&W correlation of the representatives crystallizing in the cF24 C~Mg type, where a';.b' = 0.4330 (a = unit-cell dimension), ~?W,~?W are metallic radii after Teatum et ai., and ~&T, X~&T are 'the electronegativities after Gordy and
Thomas
But these points are almost never known exactly, especially the bonding type. The bonding type can vary strongly between different intermetallic compounds having the same crystal structure, as for example in the cF8 CINa type. Therefore, one has to make the best choice between the available tables of radii. Achievement of a good dependence can then give us an indication of which bonding type is dominant in such intermetallic compounds. Unfortunately a comprehensive study searching for such quantitative relations does not yet exist. The first lOdependencies in Table 6 use such kinds of radii (metallic, ionic, covalent). The dependence from Cohen's group (Zhang et 01., ]988) is completely different; it uses pseudopotentials derived from first principles, R~c, and contains no chemical-elementspecific adjustable parameters; therefore it is independent of the three bulleted items listed. The Zunger and Cohen (1978) s-radius R~ is determined by the balance between the Pauli repulsive and Coulomb attractive potentials screened by the valence electrons. Figure 21 shows for 270 AB compounds
27
tP30
0
CrFe
CD
3.00
E 0
"'0
2.75
Figure 20. Linear dAB VS. ii~~w correlation of the representatives crystallizing in the tP30 Cr.Fe, type (correlation factor 0,992); dAU =(0552a2+006252)112 RT&CkW=(xRT&G&w+yRT&G&W)/(x+y) en tn,.'\. m~8 ' where a , c are the experimentally determined unit-cell dimensions, and R:~&w. R:~~;&.w are the metallic radii after Waber and Gschneidner
4 ,.
9
a
cP2 CICs
cF8 CINa
.<
1,5
2,0
RZCs (a.u.)
2,5
3.0
Figure 21. Measured distance
dAB (bond length) versus renormalized radii expression R~c after Zunger (1981) correlation for 270 AB intermetallic compounds with cFS CINa and cP2 CICs crystal structures
28
Crystal Structures of Intermetallic Compounds 4.5 Symmetry and Simplicity Approach

4.5.1
having cF8 CINa and cP2 C1Cs structures a plot of the bond length dAB. calculated from the experimentally determined unit-cell dimensions, vers~s a radius expression R;a:c containing _xP and N. By comparing the accuracy between the calculated and observed distances (unit-cell dimensions) in Figures 19 and 20 with Figure 21, we find approximately O.OO3nm and O.OlO nm respectively. The three indicators of the space principle belong to the size factor and the angular valence-orbital factor.
Binary, Ternary, and Multinary Intermetatlic Compounds
Laves (1%7) observed that 23 metals crystallize in the el2 W type instead of, as would be expected from the space principle, in either the cF4 Cu (c.c.p.) or hP2 Mg (h.c.p.) types, hence realizing as the AE a rhombic dodecahedron (14-b) with CN 14 (12-b) instead of a cubooctahedron or respectivelya twinned cubooctahedron (12-d)'WitheN 12. Recognizing this to be a consequence of the tendency to form arrangements of high symmetry, he called this tendency the symmetry principle. To analyse the validity
10
--.. -
---
....
n .~
-.._;
-------..
-------_.
_.
.... I
--
..
..
Vl
"0
c:
(l
--_.-
.. _ 0#._
----_
..
::I
0!j
E
.2
..
--;
'-
... ~
E 8
c:
...
__ ..
-~
:\
2
._....
..
_--_ ..
_.
u.. .J
o
,Uti ,J.
t_
1til LL.
70 80
J.
...
l...1 U ..
_I.
IJ
.1
10 20 30 40
50 60
90 100 110 120 130 140 150 160 170 180 190 200210 Z20 230
Space-group number
Figure 22. Number of intermetallic compounds. distributed by space-group numbers
29
(/) Q)
150 I-
_ 30
-
Q)
CL :>.
r--
--(/)
-;;!( o
r--
"0
c:
:::::I
..... ::l t3
0
......
:::I ,_ .(/)
100 I-
a ~
20 Ir-
8
.S:?
J
.!
E
:::I
-a; 10 I50 I-
~
r--
... c
E
Q)
r-
t--
I 2 3 t; 5 6 7 8 9 10II 1213II. 15 16
Number of point sets
~ Figure 24. The percentage of binary, ternary, and quaternary inter metallic compounds having various numbers of point sets within their crystal structures. We consider 650 crystal structures including 16500 inter metallic phases. The 100 most common crystal structures contain about 72% of all known intermetallic compounds
2 3 4 5 6 7 8 9 10II 1213141516
Number of point sets
Figure 23. The number of crystal structure types versus the number of point sets within each type
of the symmetry principle for intermetallic compounds, Rodgers and Villars (J993) searched CRYSTMET (1992) for the frequency of the space-group numbers. Figure 22 shows that over 90010 of all intermetallic compounds crystallize in one of the following 11 out of 230 space groups: no. 12, no. 62, no. 63, no. 139, no. 166, no. 191, no. 194, no. 216, no. 221, no. 225, and no. 227. These space groups are for each crystal system the most symmetrical ones. This is quantitative proof that intermetaJlic compounds strongly prefer the highest symmetry in contrast with, for example, organics, which most often crystallize in the ac1inic or monoclinic crystal systems. An analogous principle is the simplicity principle, which means intermetaUic compounds strongly prefer crystal structures that are simple. A first measure for simplicity is the number of point sets within a crystal structure. Figure 23 shows the number of crystal structures versus the number of point sets in a crystal structure based on Pearson's Handbook (Villars and Calvert, 1985). The vast majority of the crystal structures have between one and six point sets per crystal structure. Figure 24 shows the percentage of binary, ternary, and quaternary intermetallic compounds versus the number of point sets per crystal structure, and demonstrates even more clearly that tbe majority (over 800'/0) of a11intermetallic compounds have between one and four point sets per crystal structure. Here we
3000
.....
~
1.0
C\I
2500
0 '0
c:
2000
a..
::::I 0
8
....
E
::::I
1500
1000
500
1 2 3 4
5 6 7 8 9 10
Number of AETs
Figure 25. Number of intermetallic compounds versus the number of AETs per crystal structure for the 5521 cubic intermetallic compounds
30
700 -r- - 00
~
600
,_"_o___
Vl
"'0
0
0
C :::l
500
._._ ..
0.
CTl
~
u 0
U 'U
40U
..
E ... . c ._ 300
....
E
:::l
200
100
l-r-
II~
20
40
1.1 I
60
I II
80
I'
120
I
140
J
(
100
160
180
.J
I
L
240 260
2RO
200
220
Number of atoms per unit cell

ltigure lii. Number of intermetallic compounds versus their number of atoms per unit cell for the 'tP' Bravais-type crystal structures
considered the 650 most populous crystal structures including 16500 intermetallic compounds. As the number of point sets per crystal structure is just the upper limit of the number of possible different AETs. Daams and Villars (1993) investigated the number of different AETs within a crystal structure, which is the real measure for simplicity, for all cubic Intermetallic compounds. Figure 25 shows that the situation is even simpler: of 5521 investigated intermetallic compounds, 5086 or 920/0 crystallize in crystal structures with three or fewer different AETs within a given crystal structure; 2561 lnterrnetallic compounds have just one AE within the crystal structure. 2035 belong to the two-environment types, and only 490 to the three-environment types (see Figure 25). A similar behavior was also observed for the rhornbohedral and hexagonal crystal structures (Daams et al.,
}992). The simplicity principle can also be shown by plotting the number of Bravais-type intermetallic compounds, e.g, '(p' versus the number of atoms per unit cell. Figure 26 shows dearly that the vast majority of all 'tP'-type intermetallic compounds have less than 24 atoms per unit cell, although there are crystal structures known with up to 1000 atoms per unit cell. Analogous behavior is observed for all Bravais types. Both the symmetry and simplicity principles can be considered as a consequence of the angular valence-orbital factor. In Figure 8, the 14 AETs most often realized are very highly symmetrical. and because these AETs are not building units, but interpenetrating, it is not likely that crystal structures with many different AETs within a crystal structure are realized.
Factors Governing Crystal Structures 4.6 Active Concentration Range Approach 4.6.1 Binary and Ternary Systems There is a regularity that has been observed by the author to be valid for binary as well as ternary systems, namely that at least 5 at. % of a second or a third chemical element is needed to be alloyed to a unary or binary system to get a 'new' intermetallic compound(s). There exist only a few exceptions, like e.g. B66Y, BwGe, Ca33Ge, Zn22Zr, and AgMS16Zn3J' This condition is necessary but not sufficient to get 'new' interrnetallic compound(s). In the binary A-B systems this means that one will not usually find inter metallic compounds ABy with y> 19. Figure 27 shows a part of the ternary concentration triangle. The dark area is the concentration range where no 'new' ternary intermetaIlic compound(s) occur; it is seen that near the A, B, and C comers even 15 at. OJo of the other two chemical elements are needed. This has quite a consequence for experimentalists searching for 'new' intermetallic compounds as well for the determination of phase diagrams. In the binary systems 100/0and in the ternary systems even 28.50/0of the available concentration area does not need to be investigated for 'new' intermetallic compounds. To the knowledge of the author there exist but
31
very few established exceptions to this rule. Nevertheless, in some cases it is very difficult to distinguish between ternary solid solubility of the nearby binary intermetallic compound and a 'new' ternary intermetallic compound. This rule has the following impact on our five factors: any APE will come into effect only when at least 5 at. 010 of the second or third chemical element is added. The same holds for the angular valence-orbital factor; this means an AE within a crystal structure will not change, rather it will buffer the produced 'mismatch' caused by the additional chemical element, at least for the first 5 at. 070.
4. 7 Stoichiometric Restraint Approach

4.7. J Binary Intermetallic Compounds
As shown in Section 3 one is able to predict compound formation or respectively its absence fairly accurately for both ternary as well as binary systems. In the case where compound formation is predicted, one would be very much interested to know how many intermetallic compound(s) occur and at which stoichiometries. Unfortunately, very little work has been done in this field. Miedema and co-workers (de Boer et al., 1988)
B
Concentration area where no 'ternary intermetallic compounds Occur
80
60
A (at.%)
40
20
Figure 27. Part of the concentration area (dark), showing where no 'new' ternary intermetallic compound(s) occur
32
5 > llHfof>
c:
.$2-,
~I mE 0 .... E
0.-0 c: to :J .; 0
0-
0 > llHfor> -4
'-4 > ll.Hfor~-IO
are much less frequent, but are still not randomly distributed over the whole concentration range. As the majority of all binary systems have been investigated, Figure 29, it is very representative. 4.7.2 Ternary Intermetattic Compounds In ternaries the situation is even more restricted, although one would expect many more possible stoichiometric ratios because of the much larger available concentration area. Figure 30 shows an analogous plot for the ternary systems; again the seven simplest stoichiometric ratios, I: 1: 1, 1:1:2, 1: I :3, 1:2:3, 1:2:2, 1: 1:4 and 1:2:4, cover 90070 of all known ternary dalton ide intecmetallic compounds. It is very interesting that almost all realized stoichiometric ratios are along five directions in the concentration plane. For example, two of these directions are AfiC. X= 1-5, and A)3yC, x=y= 1-5. The knowledge of this experimental fact drastically reduces the number of samples to be investigated per system from about 200 samples to 25 samples (only 12.5070), with the chance of finding at least 90010 of all existing daltonide intermeta11ic compounds. The Ni-Si- Ti system determined by Westbrook et al. (1958)represents a nice example having six ternary intermetallic compounds, a11with simple atomic ratios (Figure 31). No comprehensive approach until now has been able to define 'new' structure types and therefore predict the most preferred stoichiometric ratios of binary and ternary intermetalJic compounds starting from a simple approach. According to the author, two approaches might be able to predict most of the unknown, yet possible, crystal structures, as well as describe which stoichiometric ratios occur and the rules by which those ratios are highly preferred.
(i) Space-group approach A crystal structure is defined by the formula, the unit-cell dimension(s), the
~ 0>.(/)
-10 ~ llH'or> - 20
c a.
Q) "'0 0 Q) (.)
um :o~
-20 ~ ll.Hfor> -40 -40 ~ llHfor> -75

-75 > llHfO(
a..-
Q) .... .... 0
Number of intermetallic compounds per system

Figure 28. The relation between the average number of stable intermetailic compounds per system and the enthalpy of formal ion HJo~ predicted for equiatomic interrnetallic compounds. The histogram summarizes the experimental phase diagram information for systems containing two transition metals
found for binary systems a quantitative relation between the average number of stable intermetallic compounds and the enthalpy of formation AH~~~(;predicted for equiatomic intermetallic compounds. Figure 28 shows a histogram that summarizes the results from experimentally determined phase diagrams and the calculated enthalpy of formation .Mf~~~for the equiatomic intermetallic compounds containing transition metals. For equiatomic intermetallic compounds can be calculated for most binary systems with Miedema's model, essentially starting from XM and nws of the constituent chemical elements. No attempts have been made to extend this to ternary systems. Rodgers and Villars (1993) made some statistical investigations on 25000 distincrly different intermetallic compounds with the help of CR YSTMET (1992) and found some interesting and useful results. Looking at daltonide intermetallic compounds, one observes, not a random statistical distribution of the experimentally determined intermetallic compounds over the available concentration range, but a strong preference for five particular stoichiometric ratios, 1:I, 1:2, 1:3, 2:3, and 3:5, as shown in Figure 29 (only those stoichiometric ratios are shown where at least five compounds are known to have this composition). The other stoichiometric ratios
tlH~:rc
space group, and its occupied point sets. As 230 space groups exist, one could generate within each space group all possible point-set combinations assuming an arbitrary upper limit of e.g. 100 atoms per unit cell. Villars (1981) generated such point-set combinations for space group P421m and got 67072 possible combinations (crystal structures). Knowing the experimental fact that the great majority of all daltonide intermetallic compounds have each point set occupied by just one kind of chemical element, one can in principle also generate for all binary and ternary intermetallic compounds all possible stoichiometric ratios. We observed three restraints that are obeyed by all now known intermetallic compounds:
33
.-'- - '--.'-' ._ .... --
5000
1:2
1: I
4000
(J)
.-...
"0
t:
a E
... ~
c
~ ~
::I
3000
0 Q ;:...
1:3
... .8 s
Z
:::I
2000
...-
2:3
1000 1:0
;j~v
..
I
10
II
20
I
A (at%)
I
30
40
50
Figure 29. Number of daltonide intermetallic compounds versus available concentration range for binary systems (A..B)', x<y,
x+ y=]) .The c/a (bla) value is smaller than 4.5. An intermetallic compound does not have any 4holes' within its crystal structure that are ]arger than the largest atom witbin that intermetallic compound . No intermetallic compound is known to be built up of densely packed layers each lying exactly above the others. Applying these three restraints to those 67072 pointset combinations of space group P421m reduces them to 499 remaining combinations, still too high a number to be of practical use taking all 230 space groups into account. Meanwhile Rodgers and Villars (1993) realized that over 900/0of all known intermetallic compounds crystallize in one of the 11 most common space groups (see Figure 22), This experimental fact brings the number to about 5000 potential crystal structures (assuming that in each of the 11space groups application of the above empirical restraints reduces the number of potential structures to a level similar to that of the investigated space group) in comparison with the 2750 experimentally found crystal structures.
Atomic environment/rsimulated annealing' approach From the investigations of Daams and Villars (1992) and Daams et 01. (1992) of all cubic,
(ii)
rhombohedral, and hexagonal crystal structures, we know that the number of atomic environment types (AETs) in realized intermetallic compounds is limited to a total of 20 most-often occurring AETs and a group of 2-3 times less-often occurring AETs. Preliminary results on the remaining crystal classes indicate that the numbers will not be increased significantly by including all crystal structures. This experimental fact together with 'simulated annealing' gives us the possibility to predict as yet unknown crystal structures.
34
Crystal Structures of lntermetallic Compounds Concentration area where no 'ternary' compounds occur
1139
65
2552
ABC2
~---.,~~r-4~-___;
1545
AS C3 -..:----4IH1n~:--____,
5S9
ABC4
50
40
30
20
10
c
.
Figure 30. Number of daltonide intermetallic compounds versus available concentration range for ternary systems (AxByCt,
x<y<z. x+ y+z= I) 'Simulated annealing' (Pannier et al., 1990) is an attempt at solving the following problem: Given the chemical composition of an intermetallic compound and the values of its unit-cell dimensions. predict its crystal structure by optimizing the arrangement of the atoms in accordance with a set of prescribed rules. The procedure uses simple, empirical crystal-chemistry arguments (Pauling's principles for ionic compounds) and a powerful stochastic search procedure, known as 'simulated annealing'. Combining the AET restraint with 'simulated annealing' by varying the unit-cell dimensions and chemical compositions in a systematic way might also result in the prediction of the most probable 'new' crystal structures as well as their stoichiometric ratios. The stoichiometric restraint is directly connected with the occupation of the point sets within a crystal structure. and therefore is a consequence of the angular valence-orbital factor.
intermetallic compounds, systems or crystal structures there exist very strict structural relation restraints. This can be done by investigation of the experimentally known polymorphic modifications within an intermetallic compound (normal condition, high temperature, and high pressure), investigation of the realized crystal structure combinations within binary systems, and investigation of the experimentally found interrnetallic compounds within a certain crystal structure.
{i) Polymorphic modification
4.8 Structural Relation Approach

4.8.1 Binary Intermeta/lic Compounds There exist three main ways to show that, within
combinations restraint Villars (1983) showed, in context with the structure maps looking at around 1000 binary AB compounds crystallizing in one of the 22 most populous crystal structures, that from the possible (22 x 21)/( I x 2) = 231 crystal-structure combinations only 13 are realized (see Table 7). Figure 32 shows the structure map (Rodgers and Villars, 1993) of the three-dimensional plot: sum of valence electrons at r;v1\8 = 7 versus magnitude of Martynov and Batsanov's electronegativity difference I~BI versus magnitude of Zunger's pseudopotential radii difference It&_Rz Asl for binary 1; t interrnetallic s+p, compounds. It is nicely seen that there exist domains
35
Si
90
80
60
NiSi
50
SiTi
30
_ 20
,
\
,
\
10
Ni
10
20
40
60
Ti
Figure 31. Isothermal section at 1273 K of the Ni-Si- Ti system showing six ternary intermetallics, all with simple atomic ratios (after Westbrook et al., 1958)
where two modifications are realized and domains where only one modification is found. Exactly the same is shown in a chemical element versus chemical element diagram for AB compounds (see Figure 33). The crystal structure{s) are indicated in each field with the notation used in Villars et al. (1989). This experimental fact enables us to predict both the possible crystal-structure modifications in polymorphic intermetallic compounds and which intermetallic compounds are likely to show polymorphism.
(ii) Companion crystal-structure restraint Belov and co-workers (Smirnova et al., 1983) traced rules governing the coexistence of crystal structures with each other in binary systems with the help of experimentally determined phase diagrams and then constructed a system of interrelations of crystal structures. An example of such a crystal-structure sequence in binary systems containing transition elements is: cP8 Cr3SirI6 MoSi2-tP4 AuCu-hP3 AlB2
Table 7. Binary AB compounds that are experimentally found to crystallize in two or three modifications Structure types related through polymorphism AuCd (oP4)-CuTi (tP4) Bfe (oP8)-BCr (oC8) BFe (oP8)-BCr (oC8)CICs (cP2) BPe (oP8)-CICs (cP2) BCr (oC8)-CICs (cP2) MnP (oP8)-FeSi (cP8) GeS (oP8)-C1Na (cPS) AuCu (tP4)-CuTi (tP4) AuCu (tP4)-ClCs (cP2) SZn (hP4)-SZn (cPS) HgS (hP6)-SZn (cP8) AsNi (hR6)-CINa (cFS) CICs (cP2)-FeSi (cPS) Representatives AuTi DySi, EuSi, GdNi, GePr, HoNi, HoSi, LuNi. NiTm NiYb, PrPt AuCe, AuGd. AuNd, AuPr, AuSm, AuTb AgYb, AuPm, AuYb AuDy, AuEr, AuRo, AuTm RhSi PbS, PbSe, PbTe, SnTe AgTi, CuTi AuMn, JrMn, MnRh Agl, BrCu, CdS, CdSe, ClCu, CuI, OZn, PoZn, SZn, SeZn HgS, HgSe PdSb, PdTe OsSi, RuSi
36
cF8 CINa cF8 CINa

0
og
o ()
0
()
3
........
0<>
o (10 o
0 0 () 0 () o()~()
C) c(
;:, t"O
hP12 NaO
Q. N
!/l
2
'6
e
0
ct
<l
0)
e
0
cF6 CINa
0
0 0
00 e ~ "" a o. "
0
0 0
\
0 000 0
<>
o ~
0 o~
o~,~ o
e
<>
0 0 0
0
2
I~XMBABI
Figure 32. Villars' structure map for EVAO=7 of the three- dimensional plot of valence electrons EVAB versus magnitude of Martynov and Batsanov's electronegativity difference IAXM&:BI versus magnitude of Zunger's pseudopctential radii difference I.6R~+pl. showing domains where two modifications can occur oP8 BFe/oeS BCr and oP8 BFe/cF8 CINa
(iii) Chemical element/point-set restraint In 90 of the 106 investigated, most populous, binary crystal structures (Villars, 1981) we found a very strict regularity between the position of the chemical element in the periodic table (s.p.d.f elements) and its point-set occupation within a crystal structure. This can be best explained with e.g. the tI32 Si, W s type. Its binary representatives are only found in the following chemical element combinations: p elements with d elements, and p elements with f elements. In addition. the ratio between p and d or f elements is always 3;5. This means in this example that p elements always occupy the point sets 4(a) and 8{h), and d or f elements the point sets 4(b) and 16(k). Looking at the AETs of those positions, this can be
easily explained: the 4(a) and 8(h) positions have AETs with eN 10, whereas the 4(b) and 16(k) positions have AETs with CN J4 and 15 respectively. It is well-known that p elements prefer low CN in their elemental state, whereas d or f elements prefer a higher CN. Compounds with 46 crystal structures are formed only by one combination group (e.g. p with d elements), 34 crystal structures are formed by two different combination
groups (e.g, p-d and p-f), and the 10 structures remaining are formed from three or more groups. All three restraints have only been partially investigated and only for binaries. The most practical impact of this approach will be extension to ternaries. We are convinced
37
(/)
~.E F
I I
vvvv
'C~"'O'C
CD
38
Crystal Structures of Intermetallic Compounds Symmetry principle The vast majority of all intermetal1ic compounds crystallize in one of the following 11 space groups: nos. 12.62,63, 139, 166, 191,194,216, 221, 225, and 227. Simplicity principle The vast majority of all crystal structures have three or fewer different atomic environment types (AETs) within the crystal structure (single-, two- or three-environment types). In addition, the vast majority of al1intermetallic compounds have less than 24 atoms per unit cell. A tomic-environment principle The vast majority of aU atoms (point sets) in intermetallic compounds realize as their atomic environment one of the 14 atomic environment types shown in Figure 8. Active composition range principle In binary (ternary) systems at least 5 at. 0'/0is needed of the second (third) chemical element to form 'new' intermetallic compound(s). This reduces the available concentration range for the occurrence of 'new' intermetallic compounds by 100'/0 for binary systems and by 28.5070 for ternary systems. Stoichiometric restraint principle The vast majority of daltonide intermetaUic compounds have each point set(s) occupied by just one kind of chemica) element. The following stoichiometric ratios are highly preferred: for binary intermetallic compounds, 1:1, 1:2. 1:3, 2:3, and 3:5; for ternary interrnetallic compounds, 1:1:1, 1:1:2, 1:1:3, 1:2:2, 1:1:4, 1:2:3 and 1:2:4. Chemistry principle The vast majority of crystal structures show a very strict regularity between the position of the chemical elements in the periodic table (s, p. d, and f elements) and its point-setfs) occupation within a crystal structure.
that most of these relations can be understood by considering the AETs. The comprehensive investigation of the connection between the above-mentioned restraints and the AETs will reveal many new regularities and enable us to understand polymorphic transformations, closely related crystal structures, and the role of the position of the chemical elements in the periodic table. Analogous to the stoichiometric restraints, the structural restraints are a consequence of the angular valence-orbital factor.
s.
Nine Quantitative Principles
Here we formulate nine principles which, among the presently known regularities, are the most quantitative ones. The validity of these principles was tested on a fair number of compounds in a well-defined group of experimentally determined data sets and showed accuracies in the range of 90-1000'/0. We can therefore rely on predictions based on those principles with considerable confidence, and can say that these apply 'for the vast majority'. Applying these nine principles will both drastically reduce the number of systems (or respectively, samples) to be investigated and remarkably increase the success rate of finding novel materials. Compound-formation map principle The size, electrochemical, valence-electron, and cohesive-energy factors are the factors governing compound formation. The compound-formation maps do quantitatively predict the compound formation in binary systems (after Miedema) and in ternary systems (after Villars). Structure map principle The size, electrochemical, valence-electron, and atomic-number factors are the factors governing crystal structures of intermetallic compounds. Structure maps do quantitatively separate intermetallic compounds into distinct crystal-structure domains. There exists a whole range of different structure maps for binary intermetallic compounds as listed in Table 5 and for ternary intermetallic compounds (see Section 4.2.2, Ternary Intermetallic Compounds). Solid-solubility map principle The size, electrochemical, and valence-electron factors control solid solubility. Solid-solubility maps do quantitatively separate regions of limited and extended solid solubility for a given chemical element solvent.
6. Quantibltive Relations Between Crystal Structures and Physical Properties of Inlermetallic Compounds In the previous sections we showed which factors govern crystal structures and formulated some generally valid principles. all this based on experimentally determined data, with the main aim to .predict 'new' intermetallic compounds. The question arises: "Why should we be interested in predicting and verifying "new" ternary and quaternary intermetallic compounds']' Most scientists, in universities and especially in industry, are mainly
Factors Governing Crystal Structures interested in specific physical properties of intermetallic compounds. So an obvious question is: 'Does a quantitative connection exist between intermetallic compounds crystallizing in a certain crystal structure and a physical property of interest?' Here we just mention briefly the most outstanding correlations, which should stimulate motivation for future research in that area. The following examples show definitely that such relations exist. It is important to realize that the crystal structure of an intermetallic as well as the APEs of its constituent chemical elements are necessary conditions, but not necessarily sufficient for prediction of a particular physical property. Below we Jist some physical properties for which it has been possible to show such relations: electrical conductivity. high- T; superconductivity, highT; ferroelectricity, stable quasicrystals, higb-weldability materials. high-melting low-density materials, and materials with tailored melting point. Kiang and Liu (1980) found a connection between the electrical conductivity of a binary compound and its fractional number of bonds m versus Pauling's electronegativity difference X" calculated from its constituent chemical elements. Figure 34 shows a projection of the electricalconductivity (isoconrours). The symbols indicate the crystal structures of those intermetallic compounds. Villars and Phillips (1988) and Phillips (1989) used the three-dimensional V versus llR;+ p versus MM&B structure maps, also called quantum structure diagrams (QSD). and they have proven to be very successful for
39
systematization of the occurrence of high-T, superconductors (see Figure 35) and high-Z; ferroelectrics (see Figure 36). They are presented in two-dimensional Ll,i?';+ p versus MM&B plots together with plots of V versus Te. The superconductors with T, higher than 10K are clustered around three domains A, B, and C, each having a very limited number of crystal structures. The recently found high-T, compounds are located in C. Even more restricting are the conditions for highT, ferroelectrics with T; 500 K. These are located in a very small V versus + P versus UM&B volume (about 10/0 of the available volume). Again, this property is found in only about a dozen crystal structures. The same QSD was also successful in the systematization and prediction of stable ternary quasicrystals. Figure 37 (Villars et al., 1986) shows the domain in which the known stable and metastable ternary quasicrystals are located, together with the stable intermetallic compounds crystallizing in the cI162 A~MgllZnll type. The experimentally known quasicrystals have in addition to fulfil the condition that they are located at the boundary between compound formation and its absence in the compound-formation diagram (also called QFD, see Figure 6). The use of QSD and QFD in the context of the global structural multinary chemistry of high- T, superconductors, ferroelectrics, and stable quasicrystals 1S very comptehensively and self-consistently discussed in a publication by Rabe et at. (1992).
aR;-
LO
'\r V 6
--.--,-
0 6 0.
cP2 CICs
cl2
.. .. ,v
t.
... ...
6
I!>.
cF4 Cu (Fe CINa hPi. SZn cFS SZn
as
- -..
,.
____ ~__ v.v'" _ .4b. ~i
\
6
I
66
I!>.
.. ..
oi.,
T
I
"'-V-t ~" ..
v.
,
E 0.6 -, Z... " .. ~ __ .-..

-,.
~_v_
v~
t:J.
Ie
e..L."
~'.
~~~.'t...',_.
'" 02 . .," ,,"6

-\,'
'. .. o ,..
o
.. ,.
-~
,
, ,
.. ' C "
,.,'
6
.
\'
,.,
V v
t1. .. 6
t.~
, , ,
a
I!>.
'
'
,~'6
.6
0.8
0.4
1.2
c;). ,
" .'
"
1.6
2.0
~xPauling Figure 34. Isoelectrical conductivity plot for the fractional number of bonds m versus Pauling's electronegativity difference sx: Each compound is represented by a symbol indicating its crystal structure (broken curves are numbered in units lif a cm ")
40
..
B
.. ..
f-
1-2.5 ~ ~ ... ...
20- B
Structure types
..
...
f-
I-
2.0
1510
~
_j
...
.to
-C
...
~
+" ++. +
1.5
.-.. 20~ .._. I-u15-
3996
..
+
...cF8 ... oP16 cI40 hP2

+
CINa
B2LuRu C3Pu2 CW
- 1.0
- 0.5
I
~O
I
_L
-2.0
-1.5
-1.0
<P~o 'e
~
~c
I-
0.5
20-
rI . -/V\
~
-t
+t:
hR15 MOGPbSe " cP5 Ca03Ti (Lo,Sr)2Cu04 .L tl* 'f cF56 At2,Mg04 - tP 14 B02,Cu30SY - oP14 B02Cu307Y
o
9
cP8
Cr3Si AI2Cu CrFe AuCu3 B4CeC04
~
@
~ cI58 Mn (l
8 tI12 tP30 D cP4 e tP18
4
~~
1'1
151.0 10
1"1
~ A
& ~
(0)
(b)
Flgnre 35. (a) Domains of superconductors with Tc> 10 K projected (with respect to V). The coordinates aXM4:B and are explained in the text. (b) Matthias profiles of T, versus V for islands A, B, and C. The crystal structures are designated on the extreme right
tJi;~
2.0
Structure types
..::i
as
let
N
tn
1500
~i1.5
<I
1.0
1000
+ cPS CaTi03 hP:30 LuMn03 ..L oP20Gd~ T oP24 Sb{Nb,Ta)04 c tI24 COW04 Others
.to
Quasiternaries and quaternaries
-1.5
-1.0
-0.5
500
4
v
Figure 36. Quantum structural diagram for high- T, ferroelectrics (Te> 500 K) in an analogous representation to Figure 35
Pettifor (l988a) shows a Mendeleev number MN

versus Mendeleev number MN weld ability plot (Figure 38) for metal pairs to be joined (differentiating only between excellent, good, poor, and unsatisfactory welds). It is definitely seen that excellent weld ability is only found in a specific domain.
Fleischer (1987) constructed compounds,
a whole series of specific
gravity versus melting point plots for intermetallic

each plot for a certain crystal structure. compounds crystallizing in a certain crystal structure one can tailor compounds with e.g. high melting point but low specific
Figure 39 shows that by using intermetallic
41
Stable quasicrystal cI 162 AIsMg,1Zn,1

o Metastable
quasiCfYSlaI
m
::l
-1.0 - 1.0 -0.5
1.0
0.5
Figure 31. Quantum structural diagram for icosahedral intermetallic compounds and quasicrystals. The insert shows the domain spanned by dsp and psd combinations (s, p. and d elements) with average valence-electron concentration Vbetween 1.3 and 2.5
00
00
80
7S
oc
00
00
oc
00
co 00 00 00
00 00 00 00 00 00 00 00 00 !Xl 00 00 00 00 00 CD
oooeo
OCOCO
00000 00000 00
cococco
I:IJCOOOO
0 0 0
C 0 0
00 00
00
88 88 00 00
00
88 00
gg
00
885
0000 0000
.... 1
It-
0000000
0000000 .... 00
z
~
co
00 00
88 00
00 00 00 00
00 00
8888 o... 81 "

o o. ..
.0... .0
Weldability .. Excellent
00 00 00
co
o.
Good o Poor o Unsatisfactory
00 00
go
o
55
60
GS
70
75
80
MNA
Figure 38. Weldability property plot for metal pairs (after Pettifcr)
42
Crystal Structures of IntermetaJlic Compounds 20 15 ~

I
I I
7 10
E
o 8 0)
::>.
o
c Q)
I~~
o c:P
~DC
IJ 0 0
0
P ~ 6
{!
n
oB
OCbo
0
4 b
3 (a) cF8 CINa

I I
(b) cP4
I
AuCu)
1500
2000
2500 3000 1500

T(OC)
2000 2500
3000
Filure 39. Melting point T versus density plots (after Fleischer) (a) for cF8 CINa type compounds and (b) cP4 AuCu, type compounds
gravity (aerospace industry). From Figure 39 it is clearly seen that most cFS CINa-type intermetallic compounds are better in that respect than cP4 AuCu} intermetallic compounds. Finally Hulliger and Villars (1993) show a chemical element versus chemical element melting point Tplot for AB intermetallic compounds. A part of plot (see Figure 40) shows that the behavior is very complex, but definitely not a random statistical distribution. It is also observed that there is no significant difference between congruent and incongruent melting compounds. The crystal structure is not of first priority for prediction of the melting point; much more relevant are the constituent chemica1 elements. This very short summary shows that there exist quantitative connections between the physical properties of an intermetallic compound. its crystal structure, and the APE of its constituent chemical elements. Such connections enable the experimentalist to reduce the enormous number of possible systems to the systems most similar to the a1ready known intermetallic compounds having the physical property of interest, and therefore to the systems with the highest probability of finding novel materials.
7. Conclusion From the investigations of cubic, rhombohedral, and
hexagonal crystal structures and some cross-checks, we know that the gross features of the 2750 crystal structures (based on the space-group theory) can be described with fewer than 20 often-occurring AETs and some 2-3 times Iess-frequently occurring AETs, which simplifies the description of crystal structures drastically. Based on those AETs one expects that the 2750 crystal structures can be reduced to around 300 different coordination types that are crystal structures with the same number and kind of AETs, i.e. crystal structures having very similar gross features. Crystal structures belonging to two different coordination types are therefore distinctly different crystal structures; crystal structures belonging to the same coordination type are very similar. This conclusion is highly supported by the results revealed with the symmetry and simplicity approaches; in other words this means that the short-range order of atoms is very dominant, and long-range order is of much less importance in crystal structures. Of great practical impact are the nine quantitative principles as well as the restricted number of crystal-structure combinations found in intermetallic compounds (polymorphism) and Belov's observed companion crystal-structure combinations within binary systems. One of the main goals in the future will be to find quantitative relations between the bonding types (metallic, ionic, and covalent) of intermetallic compounds, and the AETs and APE of the constituent chemical elements.
43
Element A IL'
x '~r
NG
p,
(, He 'I' fIlo C. C. "lb. ~t St 80 $c Lv ...... T... Or TO GIl E." 50ft r.d ... ~ Lo [X x x x X IX IX' Y IX' X X IX' IX' IX' X IX X X X IX X X / 4,. ~ LX ~ X X Lx: X II X X [>< X [X X X X X V X X [X X r-,/ [>< -, [>< IX ~ I~ >< -,r-,X x e ~ X lX :\,X X Jot' I:' )< K. )< IXIX I" IXIX 1"- r-, 1"- X X V IX <,r-, Ul)< <, [X[X r-,[X[X '/ ~ )< r-, IX )< V r-,X
I(
lint Pv
II II II
,..
IJ Ttl Ti II~
Z, HI
ND To t<
Mo 11 MIl Tt Rr
... :-
"
u.. C
v: X lX x = - ':' X X X X LX X X X :'X LX - ': r-, "-r-, r-, r-,

t:I
-. :.....
e e
-, x
I'":
II II
v rx
X ix
e- ex: -,
IX j',
II
,,,-
""
x
-. """ -,
II
X IXIX tx X [X
[X
~ ~
I~ .';[X r-, [X [X -, r-, x ~
X
C
X
1:1
<,
~
-e
IXIXIXlX X IXIXIX X X [X X X [X[X ,.,

II It..
" X 1"- IXIX [2([>( 0.. X r-, r-, /

.
IX
[X r-,lX X
lX X
'" 1:1 ~' I~ :'

~ C
co
Q)
c:
.... i'~'EI '7i I':' I'~ I':;' ",. .."I;;;r;112 .:lEI ,.,
,C
~. ~. r.;:.
I';;'(~.
....
!~ ~
UIEr
I~'
I".;lEI 1'"" EI I':' I~'

II
X -,[X[X[X x X- X lX X X X IX [>< IX / H. l6 :~ x X [XlX [X X X [X LX V /r.lib u r-, / V IXIXIXIXIX V X X IX 1/ 1/ IX 1"II r-, -, '/ X [X lX X [X '/ x .xLX. [7V / clX u X V 6lX [X X [X V X X [X LX V u CG u X r-,X [X X V IX 'X .v- A X [X [X V '/ ~r X r-, X IX .x IXIXX IX X X iXIX IX 1/ V ~.. X IX V )< V [X ~V L / [XIX [X 1/ 1/ 1st X lXlX V X V V x,.1J<\ V X [X IXIXV V 8. r;let ;s,~ IXX IXIX [X IX Sc ~ r-, I)< I':' r~t I>< r;[X~ r-,X IX [X X [XIX rx [X Lo.
U II
0..IX
[X[X X [X X LX ~ [X ~
LX D<
li
II
II
II II II
U U
II II
U~
"h !I:I
II
~
E Q)
W
~
rt;' I'~ 'as r.;o u X r-,lX I';' lX [X 1"'- IX [X X X rx >< rx x I"':: u X r-,lX It: lXlX ex:IX[X ex: [X ex: )< rx )< E, it:r: I:" I;;'I;' lO. ' , )< r;' I:' I: II:!I: u X r-, IX I"': ; D< IX e- IX IXX IX IX u )< 1';,..17'. I':' I'": I':"1:ra- 1;"1:-r;: ex: r-,[X rJ ex:I)< IX :;;x IX IX [X rx I>< ~ u )< r _HI" lEI lEI ~ I: !;;- u!,: u IX r-,IXr;IXD< [X r- IX X lX X I)< lX. )< Or )< fI> c I':' ~ I"';'.';fa' r~ IX r-,IXr;IXD< r-, ><IX X X X D< ) ~ rD lEI r;.:0' I~' [X r-, 171= IX )< IX ><D< lX X X [)< i5< [X
II
lX lX lX [X lX [X !XlX [X
[X
'ttl
T",
II
t .1:1
... -
~~.
c I:
C
Ito
II
II
II
II
II
Gel
... ..r.;o r~ .. -, r:
AS compound (I)
II X r-, )< r; X IX )< ex )< i)< [)< r.;, D< X IX i)( IX' >< !X )< [)< )< II U i'~ lID 111:1 X i"... ,.... ..... II IX )( IX' )< X ")( ')( "X' II IX II C lEI :trl , .... X i"... X 11:1 D< )< )< IX II IX t It! UIX f'., [X r:l)< IXIXIX[X I> X , I)< II II i)< c ui>< ra lX D< lX lX D< X X IX II IX .... II U II II II II II .. II >< I> II II II U C I';: ... ulX c I'; 11:1 IXIX I';;IXIXIX IX x LX.IXIX II II c t':i ["\ r'-- f"-, ,,_, f"-, i"["\ I'IX rx UIX II ,';;! [X IXLXI>< rx ~"lX 1-:: 11:1' I':!IX '7IX 'oU IX' IX ~ IX IXIXD< IX u
lX. )'<
.. ., ...
..
u u
vxv
x )<IX
D< x
)<
tx
x IX IX .,J )< r-, Sa

~
,..
~.
c.
.'.
-'
t
~ ..I':;: r~' 1"'"

t
tot)
.... 1-; I':l ~ ,.~ n ":' r,:: T, r:;lXr'P! I~ IX X IX [X u IX z. ...

jt~. fCI"
""
on
rx
"" H9
IJ
n.
c16 b:!- ID' )<,. X IXIXr-< ,

I'~ c It~)
.. ~ V i': -;;- ~ ~ n m ", ~C I':' IX IX IX Nt>

n HI
~ [X
:x IX t.. ~I" ~.; c ~ I':!IX ex:: t< 'irar.;' [XIX >< z,_;!
1:1
-:
ID"
Il1O 'II
7 '/
I." II In C
.... , '"
H'
':
..
10lio Tt
1>..0: Fl. i...Li'--'
Figure 44). Part of a melting point T{K) contour plot for AB intermetalliccompounds and alloys plotted as elementA vs element B. Melting temperatures TD1(K). Lower limits (e.g. >9(0) are deduced from inf()f'mationabout sample preparation (annealing temperatures. etc.). Crossed fields. nOD-existence according to experimental phase diagrams; half-crossedfields, non-existence predicted by Miedema ('\, ) or Villars (/); u, existence unknown; c, congruentmelting; (c), the given Tvalue is the solidus temperature of an alloy; n, non-congruent melting (peritectic, peritectoidal); d. the given Tvalue is tbe decomposition temperature at nonnal pressure; s, sublimation; ., decomposition at a lower temperature
With the help of the regularities
found semi-
empirically within the experimentally established intermetallic compounds, four factors are shown to govern
crystal structure from the initially suggested five factors
with the help of AP versus AN plots. Petti for formulated an additional factor, the angular valenceelectron factor, which is fully supported by the results of the atomic-enviromnent, space, synunetry, simplicity.
44

information system in material science. This system should always be extended by newly found 'highest'quality regularities. The benefits would be: (i) Refinement of the already known regularities with exact definition of their validity and boundary conditions. (ii) Opportunity to search in a systematic way, with a high probability of success, for regularities until now 'missed'. (iii) Verification of all regularities against the databases and derivation of an 'accuracy level' for each regularity. (iv) Reduction of the number of unwanted duplications of experimental work. (v) Provision of a highly developed tool for the experimentalist working in material science. This tool, together with the intuition and experience of the scientists, should reduce the 'unlimited' large number of potential systems to a limited, though large, number of systems. The result would be the synthesis of an potential novel materials with desired physical properties with the highest probability of success. (vi) The interaction of the results of first-principles calculations would be very stimulating for the improvement of such an information-prediction system and vice versa. (vii) Enormous economic benefits to those wishing to maintain their competitiveness in the area. This approach, combined with effective materials processing techniques, should show new results in materials science and engineering. Savitskii (1976) was one of the first to stress the need for information-prediction systems in materials science. He was also the first (Savltskii et al., 1984) to develop in that context a database on the physical properties of ternary inorganic phases. The logical structure of his file is shown in Figure 41. Kiselyova (1993) uses that database together with cybernetic-statistical methods to develop an information-prediction system. This is a development of the idea of machine learning to recognize patterns for prediction, first published by Savitskii (1976). Figure 42 shows the scheme of their system. Zhou et 01. (1989) developed such a system for retrieval and prediction of binary intermetallic compounds. and the scheme of this system is given in Figure 43. The need for an international information-prediction system has been strongly evident for the last 10 years as most research activities moved from binaries to ternaries and multinaries, opening an unlimited
stoichiometric restraint, and structural relation restraint approaches. So we end up with the following five factors governing crystal structures: .Size factor Atomic-number factor Electrochemical factor Valence-electron factor Angular valence-orbital factor
The cohesive-energy factor seems not to be governing

crystal structure, but does determine compound formation or respectively its absence. The size, electrochemical, and valence-electron factors govern both crystal structure as well as compound formation or respectively its absence. The relations between these factors and the crystal structures is very complex, but the many regularities found prove that these relations are not only of a qualitative nature, but already in many cases exhibit a quantitative nature with an acceptable accuracy, It is very interesting that by comparing the present situation with the 'classical' regularities found SOor so years ago by Hume-Rothery, Pauling. Laves. Goldschmid, Kasper, and Mooser and Pearson, only one of the six factors mentioned is new: the cohesive-energy factor. From the nine quantitative principles the following are also completely new: compound-formation principle. atomicenvironment principle. structure map principle, simplicity principle. active composition range principle, stoichiometric restraint principle, and chemistry principle. One great improvement made in the last 50 years is tbe optimization of APEs of the constituent chemical elements within those factors in choosing the optimal AP for our five groups of distinctly different APs. A major improvement was made for the size factor in using pseudopotential radii, which are independent of the constituent chemical elements, in contrast to the classical metallic, ionic, and covalent radii. Another major improvement is the beginning of the extension of regularities from binary to ternary (multinary) intermetallic compounds/systems with an acceptable quantitative accuracy. Forced by the almost unlimited number of available chemical element combinations in multinary systems, the author is completely convinced that, if in the future one wishes to find, efficiently and systematically, novel ternary (multinary) materials, thi.swill only be possible by linking databases (CRYSTMET, 1992; CRYSTIN, 1992; AET. thermodynamic databases, physical property databases. AP databases, etc.) in a clever way with the nine quantitative principles and thus creating a prediction-

Atomic numbers of oomponents
45
. Syst em
~
Degree of previous study Temperature of isothermal sections Composition of oompounds
l
QuaSibinary sections
l
Refer~ numbers
Number of compounds
Com pounds
I
~
Type of melting Melting point Types of crystal structure
~
Temperature of dissociation or decomposition
~
Boiling
point
Prope rties
i
PolymorphOus transition pressure
PQ4ymorphous transition temperature
t
Crystal system Symmetry
group
Number of formula per unit
~uperoonducting transition temperature
Upper critical magnetic field at 4.2 K
Figure 41. Logical structure of Savitskii's file of inorganic phases

I
I
User
+f
Conversational processor
+
f
J
tt
I
L
Monitor
l_
L";
There is no regularity
There are data about the phase
......,
Elements database There is regularity
r--i Compounds database I

.....___
+
Knowledge base
-I
...
Prediction system I
..
.....
C
0
p u t
r
I
--I
Learning subsystem
Predicting subsystem
FigUAl 42. Scheme of the Russian system
46
Conversation module
Explanation module
Knowledge acquiring and processing module
Knowledge base
Dynamic database monitor module Reasoning and predicting module
End
ClassificatiOn module
Interface
Graphic display
Query module Figure 43. Scheme of the Chinese system
'reservoir' of potential novel materials. Only if such a long-term project on a large scale could be started soon, would it be available 10 years from now. Without such an information-prediction system the number of unwanted duplications will increase to an unaffordable magnitude, and the probability of finding novel materials with highest-quality physical properties will remain very low. Unfortunately, to the knowledge of the author, such an activity does not yet exist. despite the fact that costs of such an informationprediction system would be negligible compared to the money spent worldwide by the different countries for unwanted duplications.
8. Appendix
8.1 Glossary
Intermetallic compound Intermetallic compounds are binary. ternary. quaternary, etc., compounds containing any chemical elements except oxygen, the halides, and noble gases. Also excluded are such compounds with typical inorganic 'groups' like -NH, -NH2 -N2 etc. This definition has been used for Pearson's Handbook (Villars and Calvert. 1991) and the Atlas of Daams et aJ. (1991). This definition includes therefore also sulfides, selenides, carbides, and nitrides, which most
Factors Governing Crystal Structures material scientists would not consider to be intermetallic compounds, but because of their structural similarity we included them. <New' intermetallic compound Such an intermetallic compound has a different crystal structure than the crystal structure of its constituent chemical elements in the binary case or of its binary boundary intermetallic compound in the ternary case. Crystal structure (also called structure type or prototype) Based on the space-group theory, a crystal structure is completely determined by the following data: Chemical formula Crystal system and unit-cell dimension(s) Space group Occupation number and coordinates of the occupied point sets
47
Atomic environment type The atomic environment types (AETs) are the distinctly different idealized atomic environments realized in intermetallic compounds. Coordination type Coordination types are crystal structures with the same number and kind of distinctly different AETs. Single-environment types Single-environment types are crystal structures in which each atom has the same AET. environment types Two- (three-) environment types are crystal structures in which there exists two (three) different AETs.
Two(three-}
8.2 Abbreviations and Symbols Used
Crystal structures are named by the first intermetallic compound found to be unique in respect to the third and fourth items and are represented by the Pearson symbol followed by the formula of the prototype, e.g. hP3 AlB2 The first two letters of the Pearson symbol are identical to the Bravais type, and the digits give the number of atoms per unit cell, Daltonide intermetallic compounds Binary daltonide intermetaUic compounds" are intermetalIic compounds that occur only at a fixed stoichiometry, which means that each point set(s) of such a crystal structure can only be occupied by one chemical element.
a.c
Berthollide intermetallic compounds Binary berthollide

intermetallic compounds" are intermetallic compounds that allow deviations from stoichiometry, which means that at least one point set of such a crystal structure has to be occupied by two or more chemical elements, or in some exceptional cases at least one point set is not fully occupied, and the degree of occupancy varies. Atomic environment The atomic environment (AE) is defined by the atoms surrounding a centra) atom. The surrounding atoms to the left of the maximum gap in a plot of the number of surrounding atoms (n) versus the distance (d) normalized relative to the shortest distance (dmin) belong to the AE (see Figure 7). In the cases where the atomic environment can be described by a convex polyhedron, it is often also called the coordination polyhedron.
Definitions for ternary compounds are more complex.
Unit-cell dimension(s) AJ3yCz Chemical formula (A,B,C = chemical elements, x,y .z= stoichiometric ratio) Atomic environment AE AET Atomic environment type AN Atomic number AP Atomic property APE Atomic property expression CN Coordination number d Distance between atoms dmin Shortest distance between central atom and surrounding atoms tJ{for Enthalpy of formation Al-PI Enthalpy of solution n Number of surrounding atoms Electron density at the boundary of the nws Wigner-Seitz atomic cell N Number m Fractional number of bonds MN Mendeleev number PN Periodic number QFD Quantum formation diagram QN Quantum number QSD Quantum structural diagram R, Covalent radius (subscript: chemical element designation; superscript: radii set after e.g. Zunger, Pauling, etc.) R, Ionic radius (see Rc) s; Metallic radius (see Rc) Rnp Pseudopotential radius (see Rc) T Melting point (see Rc) V Valence-electron number (see Rc) X Electronegativity (see R,;;)
48
Crystal Structures of Interrnetallic Compounds 9. References

Hovestreydt, E. (1988). J. Less-Common Met . 143, 25. Hulliger, F., and Villars. P. (1993). J. Alloys and Compounds, 197, 197. Kelton. L. L., and Liu, T. L. (1980). Phys. Lett., 76A(2), 181. Kiang, L. L., and Uu, T. L. (1980). Phys. 76A(2), 181. Kiselyova, N. N. (1993). J. Alloys and Compounds, 197, 159. Kugirniya, K., and Stein fink. H. (1968). Inorg. Chem., 7(9),
J\hrens, L. H. (1952). Geochim. Cosmochim. Acta, .2. 155. Alonso, J. A. (1985). Rev. Latinoam. Metal. Mater., 5(1),3. Alonso, J. A., and Simozar, S. (1980). Phys. Rev., 822. 5583. Andreoni. W. Baldereschi, A. Biemont, A., and Phillips, J. C. (1979). 20(12),4814. Ballestracci, R., and Bertaut, E. F. (1964). Bull. Soc. Fr. Miner. Cristallogr., 81,512. Bloch, A. N., and Schatteman, G. C. (t981). In Structure and Bonding in Crystals (eels M. O'Keeffe and A. Navrotsky), Academic Press, New York. Brunner, G. O. (1971). Acta Crystallogr., .33A, 226. Brunner, G. O. (1988). 2. Kristallogr., 182, 49. Brunner, G. 0., and Schwarzenbach, D. (1971). Z. Kristallogr., 1.33, 127. Burdett, J. K., and McLaran, T. I. (1984). Solid State Chem., 53, 382. Burdett. J. K. Price, G. D., and Price, S. L. (1981). Phys. Rev., B24(6), 2903. Chelikowsky, 1. R. (1979). Phys. nev., BI9(2). 686~ Chelikowsky, J. R. (1991). In Encyclopedia 0/ Applied Physics. Vol. 3, p. 553. VCH Publishers, Weinheim. Chen Nianyi (1916). Bond-Parametric Function and its Application. Scientific Publishing House, Beijing, pp. 15-16. CRYSTlN (1992). CAN/SND, Canadian Institute for Scientific and Technical Information, NRCC, Ottawa. Canada. CRYSTMET (1992). CAN/SND, Canadian Institute for Sciemific and Technical Information, NRCC, Ottawa, Canada. Daams, J. L. C., and Villars, P. (1993). J. Alloys and Compounds. 197, 243. Daams, J. L. C, Villacs. P . and van Vucht, J. H. N. (1991). Atlas of Crystal Structures/or Intermetailic Phases, Vols 1-4. ASM International, Metals Park. OH. Daams, J. L. C., Villars. P., and van Vucht, J. H. N. (1992). J. Alloys and Compounds. 182, I. Darken, L. S., and Gurry, R. W. (1953) Physical Chemistry of Metals. McGraw-Hill, New York. de Boer, F. R., Boom, R., Martens, W. C. M., Miedema, A. R., and Niessen, A. K. (1988). In Cohesion in Metals, Transition Metal A Boys (Cohesion and Structure, Vol. I} (eels F. R. de Boer and D. O. Pettifor). North-Holland, Amsterdam. Edwards. A. R. (1972). Metall. Trans. J. 1365. Fleischer. R. L. (1987). J. Mater. Sci., 22, 2281. Fleischer, R. L. (1988). J. Mater. Sci. Leu., 7, S25. Goodman, D. A., Bennett, L. H., and Watson, R. E. (1983). Scripta Metall., 17, 91. Gordy. W. (1946). J. Chern. Phys . 14. 305. Gordy, W., and Thomas, W. J. O. (1956). J. Chem. Phys., 24,439. Gschneidner. K. A. (1980). In Theory oj Alloy Phose Formation (ed. L. H. Bennett). Metallurgical Society of AlME, Warrendale, PA. Herman, F., and Skillman, S. (1963). Atomic Structure Calculations. Prentice-Hall, Englewood Cliffs, NJ.
u,
1762.
Laves, F. (1967). In Intermetallic Compounds (ed, J. H. Westbrook). Wiley, New York. Machlin, E. S., Chow. T. P., and Phillips, J. C. (1977). Phys. Rev. Leu., 38, 1292. Martynov, A. J., and Batsanov, S. S. (1980). Russ. J.lnorg. Chem., 25, 1737. Massalski, T. B. Okamoto, H., Subramanian, P. R., and Kacprzak, L. (1991). Binary Alloy Phase Diagrams, 2nd edn. ASM International, Metals Park, OH. Moeser, E., and Pearson, W. B. (l9S9). Acta Crystallogr. n, 1015. Pannier, J., Bassas-Alsina, J., Rodriguez-Carvajal, J., and Caignaert, V. (1990). Nature, 346, 343. Pautler, P. (1977). Kristall Technik, 12, 667. Pearson. W. B. (1958, 1967). A Handbook 0/ Lattice Spacings and Structures of Metals and Alloys, Vols. 1 and 2. Pergamon, Oxford. Pearson, W. B. (1962). J. Phys. Chem, Solids, 23, 103. Pearson, W. B. (1968). Acta Crystallogr., 24B, 1415. Pearson, W. B. (1980). J. Less-Common u, 72, 107. Pearson, W. 8., and Villars, P. (1984). J. Less-Common Met., 97, 119. Petti for , D. G. (1984). Solid State Commun., 51, 31. Pettifor, D. G. (J988a). Mater. Sci. Technol., 4, 675. Pettifor, D. O. (1988b). In Encyclopedia oj Materials Science and Engineering (eds R. W, Cahn and M. 8. Bever), Vol. 1. Pergamon, Oxford. Phillips, 1. C. (1989). Physics Of High r, Superconductors. Academic Press, New York, Prince, A. (1956). The Constitution Diagrams of Alloys: A Bibliography, 2nd edn. Institute of Metals, London. Prince, A. (1978). Multicomponent Alloy Constitution Bibliography /955-/973, Metals Society. London. Prince, A. (1981). Multicomponent Alloy Constitution Bibliography /974-1977, Metals Society, London. Rabe, K. M., Phillips. J. C., Villars, P., and Brown, L D. (1992). Phys. Rev., B4S, 7650. Raynor, G. V. (1974). J. Less-Common Met., 37, 247. Rodgers, J., and ViUars, P. (1993). J. Alloys and Compounds, 197, 167. Savitskii, E. M. (1976). In Intermetallische Phases (ed. O. Henkel). VEB Deutscher Verlag fur Grundstoffindustrie.
Leipzig.
Savitskii, E. M., Kiselyova, N. N., Pischik, B. N., Kravchenko, N. V., and GoJikova, M. S. (1984). SOy. Phys. Dokl., 29(11), 971. Schubert. K. (1990). Private communication. Shannon, R. D. (1976). Acta Crystallogr., 32A, 751.

Shoemaker, C. B., and Shoemaker, D. P. (1964). Trans. Metall. Soc. AlME, 230. 486. Slater. J. C. (1956). Theory of Alloy Phases. American Society for Metals. Cleveland, OH. Smirnova, N. L., Berzon, E. M., Konyaev, Yu. S., and Belov, N. v. (1983). SOy. Phys. Dolk., 28(5), 359. St John, 1., and Bloch, A. N. (1974). Phys. Rev. Lett., 33, 1095. Stuwe, H. P. (1959). Trans. Melall. Soc. A/ME, 21S, 4()8. Suffczynski, M. (1987). In Computer Handling and Dissemination of Data (ed. P. S. Glaeser). North-Holland, Amsterdam. Teatum, E., Gschneidner, K., and Waber, J. (1972). Cited in Pearson, W. B. (1972). The Crystal Chemistry and Physics 0/ Metals and Alloys. Wiley, New York, p. 151. Villars, P_ (1981). Dissertation No. 6787, ETH Zurich, Switzerland. Villars. P. (1983). J. Less-Common Met., 92, 215. Villars, P. (1985). J. Less-Common Met., 109, 93. Villars, P. (1986). J. Less-Common Met., 119, 175. ViUars, P., and Calvert, L. D. (1985). Pearson's Handbook oj Crystallographic Data for Intermetallic Phases, Vols 1-3. American Society for Metals, Metals Park, OH. Villars, P., and Calvert, L. D. (199). Pearson's Handbook oj Crystallographic Data for Intermetallic Phases, Vols 1-4, ASM International, Materials Park, OH. Villars, P., and Girgis, K. (1982). Z. Metallk.; 73(7), 45.5_ Villars, P., and Phillips, 1. C. (1988). Phys. Rev., 37(4), 2345. Villars, P., Girgis, K., and Hulliger, F. (1982). J. Solid State Cbem., 42, 89. Villars, P., PhiUips, J. C., and Chen, H. S. (1986). Phys. Rev.
49
u, 57(24),
3085.
Vol. 2) (eds F. R. de Boer and D. G. Petti for). NorthHolland. Amsterdam. Villars, P., Prince, A., and Okamoto, H. (1994). Handbook oj Ternary Alloy Phase Diagrams, Vols 1-8. ASM International, Metals Park, OH. Volchenkova, R. A. (1989). Russ. J. Phys. Chern. 63(9).1349. Waber, J. T., Gschneidner, K., Larson. A. C, and Prince, M. Y. (1963). Trans. Me/all. Soc. AIME, 227,717. Walzer, W. (1990). Phys. Status Solidi B, 1628, 75. Wang, E., and Taylor, J. B. (t978). Trans. Meta/I. Soc. A/ME, 234, 252. Watson. R. E., and Bennett, L. H. (197Sa). J. Phys. Chern. Solids, 39, 1235. Watson, R. E., and Bennett, L. H. (l978b). Phys. Rev., B18, 6439. Watson, R. E., and Bennett, L. H. (1984). Acto Metall. 32(4), 477. Westbrook, J. H. (1993). MRS Bulletin, February, p. 44. Westbrook, J. H., DiCerbo, R. K., and Peat, A. J. (1958). General Electric Report 58-RL-2117,Research Information Section, Schenectady, NY. Wilson, C. G . and Spooner, F. J. (1973). Acta Crystaltogr., 29A.342. Zhang Bangwei, (1985). Z. Metallk., 76(4), 264. Zhang, S. B., Cohen, M. L., and Phillips, J. C. (1988). Phys. Rev . 838(17), 12085. Zhou Ban, Jin Shirning, Shao Jun, and Chen Nianyi (1989). Acta Metatl. Sin. (English edition), 28(6). 428. Zunger, A. (1981). In Structure and Bonding in Crystals (eds M. O'Keeffe and A. Navrotsky), Academic Press, New
York.
Villars, P., Mathis, K., and Hulliger , F. (1989). In The Structures of Binary Compounds (Cohesion and Structure,
Zunger, A., and Cohen, M. L. (1978). Phys. Rev. 8188, 5449.
This chapter was originally published in 1995 as Chapter 11 in Intermetallic

Compounds, Vol. 1: Principles, edited by 1. H. Westbrook and R. L. Fleischer.

Villars Combined

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Villars Combined

Uploaded by

Copyright:

Available Formats

Copyrighted Materials

Copyright", 2000 John Wiley & Sons Retrfeved from YiYiv/.knovel.com

Factors Governing Crystal Structures

different. There exist (J 00 x 99 x 98)/(1 x 2 x 3) = 161 700

2. Strategy to Find the Factors Governing Crystal Structure .

Factors Governing Crystal Structures

Crystal Structures of Intermetallic Compounds

Groups d elements p elements

Straight lines with negative slopes

Straight lines with positive slopes and a discontinuity at Cu

Factors Governing Crystal Structures

Factors (groups) Size factor

Number of data sets

"Herman and Skillman (1963).

P and Q are constants for certain groups of systems,

'. .. '. -I"

... .. ..... .'..I. .

.:' : ..:: ...... ~.: + + .. ~ + + + +

3_2 . Ternary Systems

Factors Governing Crystal Structures

Y L.. T", , H9 0, h &4 ~

d7 I ~ V!/VVYl/l/ V ~7~VV/VV/ d6 MoIX x X x x x X IX

X~i\. ~IX > 1'\1'\. 'h~~~~~X~i'.'I'\f'.I'\~ LIIi"i'\['\.~I'\Xl'\iI'\ ~r-.,I'\

I'\')C~I,\ >< XX "r-.,>

I/I/I/I/I/I/v IX Scl/VI/VI/VVI/><!X Ell ~ '" 1\ 1'\ I' X1'\ IX

1/ ~ D< '" Wal/I><~ K IXI>< ~~ ~

Crystal Structures of Intermetattic Compounds

and the sum of the ratios of the melting temperatures in kelvin,

Fagure5. ViUars' three-dimensional compound-formation plot

VB = 2 for ternary systems. For the meaning

assigned structure is sufficient. For all such ternary

the magnitude of the difference in the number of valence electrons,

Factors Governing Crystal Structures

Y= (TA'TS + TA'Te + TS'Tc)13

Crystal Structures of Intermetallic Compounds

Crystal Structures of Intermetallic Compounds

Number of compounds _ _,----_ .._ .. _Ternary 2

C SZn SZn BiO

tl16 oPS oP12 oPl6 oPl6 opn

"Thin films. IlHigh-temperature phase.

Crystal Structures of Intermetallic Compounds

Reference Mooser and Pearson (1959)

Class of compound XY normal-valence compounds XZ normal-valence compounds XT normal-valence compounds XY normal-valence QN

Crystal structure (number of intermetallic compounds)

Pearson (1%2) St John and Bloch (1974) Machlin et 01. (1977)

Watson and Bennett

Watson and Bennett (I978b)

Watson and Bennett (1978b)

Andreoni el al. (1979) Bloch and Schatteman (1981)

Zunger (1981) Zunger (1981)

Factors Governing Crystal Structures

number Crystal structure (number of

Crystal Structures of Intermetallic Compounds

= boron), AxBy(C = boron),

with x> y with v> 1 with x<y

4.3 Solid-Solubility Map Approach

4.3.1 Binary Systems

First Hume-Rothery rule (size factor)

tiRs+p = 2x(R s+p.A

for ternary intermetallic compounds, AxByCZ

Factors Governing Crystal Structures

14-b Rhombic dodecahedron

(i) Darken and Gurry plots

Crystal Structures of Intermetallic Compounds

__ ~ v.v'" _ .4b. ~i