a

Review Article http://dx.doi.org/10.4322/2176-1523.1048

AN INTRODUCTION TO THE CALPHAD METHOD AND THE

COMPOUND ENERGY FORMALISM (CEF)
Gabriele Cacciamani 1

Abstract

The CALPHAD (Calculation of phase diagrams) method and the Compound Energy Formalism (CEF) are briefly
introduced and a few selected examples of their application are reported.
Keywords: CALPHAD; Thermodynamics; Compound energy formalism; Alloys; Intermetallics.

UMA INTRODUÇÃO AO MÉTODO CALPHAD E AO FORMALISMO DA

ENERGIA DOS COMPOSTOS (CEF)

Resumo

O método CALPHAD (Calculation of Phase Diagrams) e o Formalismo da Energia dos Compostos (CEF) são
apresentados e alguns exemplos de suas aplicações são discutidos.
Palavras-chave: CALPHAD; Termodinâmica; Formalismo da energia dos compostos; Ligas metálicas; Compostos
intermetálicos.

1 INTRODUCTION

Design and preparation, use, and eventually, recycling subsequent modeling of kinetic processes which determine
and disposal of materials require a detailed knowledge of behavior and properties of real materials.
their constitutional properties among which, in particular, The CALPHAD approach was introduced in the early
phase stabilities and equilibria as a function of the conditions 70s of the past century by L. Kaufman and H. Bernstein [1]
the materials are subjected. and later developed by several authors: see for instance P.J.
Such kind of information is provided, in a complete Spencer [2] for a brief history of CALPHAD. Moreover, the
and concise way, by phase diagrams, which are widely used, CALPHAD method and its theoretical as well as practical
especially in the case of non-stoichiometric materials such aspects have been comprehensively treated in a number of
as alloys, ceramics, salt mixtures, etc. books: we may mention N. Saunders and A.P. Miodownik [3],
M. Hillert [4], H.L. Lukas et al. [5].
Traditionally, phase diagrams have been determined
Nowadays CALPHAD is a very complex process
by means of experimental investigations which become more
where not only experiments but also quantum mechanical
and more difficult and time consuming with the increasing
and statistical thermodynamic calculations contribute to
number of components. Moreover, phase equilibria and
the determination of the parametrised expressions of the
thermodynamic properties, unless strictly correlated, are thermodynamic functions of the phases. Then, any calculation
generally determined by independent experiments and may or further modelling based on these parametrised functions
result inconsistent with each other. (e.g. phase field modelling of crystal growth, simulation of
The CALPHAD approach helps to overcome these diffusion-controlled transformations, etc.) is considered part
difficulties because it reduces the number of experiments (or an extension) of the CALPHAD approach. Nevertheless,
needed to investigate a system, assures the self-consistency the core of the method is still represented by the block
between phase equilibria and thermodynamics, makes easy diagram shown in Figure 1.
to calculate any kind of sections in multi-component phase It consists in the determination of the Gibbs energy
diagrams, and provides the thermodynamic basis for the function of all the phases involved in the system under

1
Università degli Studi di Genova – UNIGE, Genova, Liguria, Italy. E-mail: [email protected]. Alloy Phase Diagram International Commission
– APDIC, Orlando, Florida, United States of America.

2176-1523/© 2016 Associação Brasileira de Metalurgia, Materiais e Mineração. Published by ABM. This is an open access paper, published
under the Creative Commons CC BY-NC-ND license (Attribution-NonCommercial-NoDerivs) - https://creativecommons.org/licenses/
by-nc-nd/4.0/.

16 Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016
 An introduction to the calphad method and the compound energy formalism (CEF)

1.2 Selection and Evaluation of Input Data

All the data which may be related to the Gibbs

energy of the phases must be used to evaluate the empirical
parameters ak. To this end, several input data may be
considered:

- Experimental results (e.g. from DTA, DSC, calorimetry,

EMF, vapor pressure, LOM, SEM, X-ray diffraction,
etc.);
- Calculation results (e.g. from ab initio and first
principles calculations, from semi-empirical models,
etc.);
- Estimated data (e.g. by extrapolation / interpolation
procedures, or by applying chemical criteria based
on the similarity between elements, etc.);

A preliminary critical evaluation of the available

literature is mandatory to select the most reliable data to
be used.
Figure 1. Flowchart representing the core of the CALPHAD method.
1.3 Optimization of Model Parameters
consideration. They are parametrized functions in that
Empirical parameters ak are evaluated by an error
they depend on the state variables (pressure, temperature,
minimization procedure such as non-linear least squares
composition, etc.) and on a series of empirical parameters
(different software packages use different algorithms).
characteristics of a specific phase in a specific system.
An appropriate weight factor must be associated to each
Once Gϕ(P, T, x) functions have been determined for each
input datum. Correlation between input data and model
phase ϕ, they are collected in a thermodynamic database
parameters must be checked. When necessary, constraints
which can be used to calculate phase equilibria, phase diagram
and approximations may be introduced to reduce the number
and thermodynamic properties of complex multi‑component
of independent parameters to be evaluated.
systems.
More in detail, the steps of the optimization process 1.4 Calculations and Comparisons
include:
The Gibbs energy functions containing the newly
1.1 Selection of Models for the Gibbs Energy optimized ak parameters are used to calculate phase and
Functions property diagrams which may be compared to known
equilibria or data.
The Gibbs energy of each phase competing for Metastable equilibria and thermodynamic functions
equilibrium in a given system is expressed as a function of in metastable conditions must also be checked.
the state variables (xi) and a number of empirical parameters Extrapolation to higher order systems and comparison
(ak) (Equation 1): with known data is also recommended.
Unsatisfactory comparisons may lead to the adoption
=Gϕ Gϕ ( x1, …, x i , a1, ..., a k ) (1) of more appropriate thermodynamic models for selected
phases or to the selection of new or different input data.
The G functions assume different forms depending
on the nature of the phase ϕ: this can be a stoichiometric 1.5 Applications
compound, a metallic liquid, a disordered solid solution, an
ordered metallic solution, an ionic liquid, etc. The optimized interaction parameters are generally
The composition dependence of G is usually expressed used for the implementation of new databases or the
according to the Compound Energy Formalism (CEF), a improvement of existing ones. Then, databases for specific
general formalism applicable to different kind of phases. materials may be used to predict phase equilibria and
Notice that Gϕ must be defined and evaluated in a thermodynamic properties in higher order systems, or, in
wide range of conditions, generally much wider than the combination with other data, to predict various materials
stability range of ϕ in the system under consideration. properties such as microstructure and kinetic-related
Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016 17
Cacciamani

properties, mechanical properties, corrosion resistance, In the simplest case modeling is just the selection
etc. or to simulate materials behavior during processing. of a mathematical form which has proved useful,
whether it is based on some physical model or
One of the most attractive aspects of the CALPHAD
approach is its increasing predictive ability. For instance, not. However experience shows that a model is
as the number of system components increases, the usually more powerful if it is based upon physically
number of new empirical parameters needed to obtain a sound principles. With such a model one can hope
satisfactory description of the Gibbs energy of the phases to make predictions outside the tested range with
tend to decrease. In other words, if the optimization of some confidence.
binary systems may require the evaluation of a relatively
If the Gibbs energy dependence on temperature
high number of parameters (and, consequently, the need of
and pressure, though complicated, may be represented by
a large number of experimental data), the optimization of
a general expression, applicable to all kind of phases, the
a ternary system may require the introduction of a limited
dependence on composition is strongly affected by nature
number of new parameters (in addition to those already
and strength of the atomic interactions. In this case it is not
evaluated for the binary sub-systems), and consequently
possible to adopt a single model for all phases.
a relatively limited number of new experimental data.
The best solution proposed until now is the “Compound
In practice, once the binary and ternary interaction parameters
have been evaluated, quaternary and higher order systems Energy Formalism” (CEF) [6], a general formalism which
may be reliably calculated without need of new empirical allows to select the most appropriate model for each phase
parameters. Thanks to this behavior it can be said that the in the framework of the same formalism. According to the
Calphad approach is essentially an optimization method CEF, the different constituents of a phase (neutral atoms, ions,
for 2-3 component systems, while, for multi-component vacancies, etc.) are distributed in one or more thermodynamic
systems it assumes, more and more, the characteristics of sublattices where they mix according to the classical solution
a prediction method. This makes Calphad particularly useful theory (adopting ideal, regular, subregular, etc. behavior)
for the development of technologically relevant materials with different interaction parameters in different sublattices.
which are typically multi-component. The Gibbs energy of the phase will then depend on the site
On the contrary, the main limitation of the Calphad fractions (i.e. the mole fractions relative to each sublattice)
approach is the impossibility to predict the stability of a instead of the overall composition.
multi-component phase not already present in lower-order Thermodynamic sublattices can have different
systems. In the more recent years, this limitation has partially physical meanings in different phases: in ionic compounds, for
been overcome by coupling thermodynamic modelling and example, they can be used to separate anions and cations; in
ab initio calculations. In fact, the stability of phases which alloys they can be identified with crystallographic sublattices.
may be expected to compete for equilibrium in a specific In this case, for instance, it is clear that atoms in the same
systems can be calculated (typically by DFT method) and crystallographic site have the same interaction with their
included in the Calphad equilibrium calculations. surroundings, while atoms in different sites generally have
different coordination and, then, different interaction with
the rest of the structure.
2 COMPOUND ENERGY FORMALISM Formally, the Gibbs energy of a phase ϕ can be
expressed as the sum of three main terms as shown in
The thermodynamic modeling of the phases, i.e. the Equation 2: the reference term, one ideal mixing term and an
way we use to express their thermodynamic properties as a excess term. They can be followed by further terms in case
function of the state variables, is certainly one of the crucial further interactions (such as magnetic, electrical, etc.) occur:
points in the CALPHAD approach. The importance of the
modeling was well expressed by Mats Hillert (p. 400)[4]: Gϕ = ref Gϕ + id Gϕ + ex Gϕ + mag Gϕ (2)
The purpose of modeling is two-fold. From a In case of a pure element, second and third terms
scientific point of view one likes to learn how nature
are missing and any dependence on composition of course
functions. One way to gain knowledge is to define
disappears. Equation 2 becomes Equation 3:
some hypothesis resulting in a model and test it by
comparing the predictions from the model with
Gϕ = ref Gϕ + mag Gϕ (3)
experimental information. […]
From a more technological point of view one likes to Where the terms in Equation 3 are given by Equations 4 and 5:
predict the properties of a particular system in order ref
Gϕ ( T ) = A ϕ +B ϕ T +Cϕ TlnT +Dϕ T 2 + ... (4)
to put it to efficient use in some practical construction
or operation. Then it is often most convenient to
have a model which yields an analytical expression. mag
Gϕ =RT ⋅ f ( τ ) ⋅ ln ( β ( x ) +1) (5)
18 Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016
 An introduction to the calphad method and the compound energy formalism (CEF)

where A, B, C, D,… are the empirical parameters which For a binary phase with magnetic ordering the
must be evaluated for any specific element, on the basis composition dependence of mag Gϕ (Equation 5) results
of the experimental information. In the magnetic term from the composition dependence of TC and β, which are
(Equation 5), based on the model introduced by Inden [7] expressed by a Redlich–Kister series expansion (Equation 11):
and subsequently adapted by Hillert and Jarl [8], β is the
TCϕ ( x ) = x A TCϕ ( A ) +x B TCϕ (B ) +x A x B ∑ TCϕ ⋅ ( x A − x B )
i i
average magnetic moment per mole of atoms in Bohr
magnetons, τ is the ratio T/TC (TC= critical temperature i=0..n
(11)
βϕ ( x ) = x A βϕ ( A ) +x B βϕ (B ) +x A x B ∑ βϕ ⋅ ( x A − x B )
i i
for magnetic ordering), and f(τ) is a polynomial expression
i=0..n
obtained by expanding Inden›s description of the magnetic
heat capacity into a power series of τ. Temperature and where i TCϕ and i βϕ in Equation 11 are expansion parameters
magnetic parameters for pure elements are available in the to be evaluated on the basis of the experimental information
PURE database by SGTE [9]. available.
If two or more components are present, mixing can
occur and all terms of Equation 2 are significant. Depending 2.2 Stoichiometric Phases (or Compounds)
on the mixing (or non mixing) mechanism a few typical
situations may be distinguished: Stoichiometric phases or compounds are modelled
by an n-sublattice model with each sublattice occupied by
2.1 Substitutional Random Solutions only one element. As in the case of pure elements, mixing
terms of the Gibbs energy are null. In this case, Equation 6
In the simplest case of random mixing, as in the reduces to Equation 12:
liquid or in the terminal solid solutions, only one sublattice
is needed and the three main terms of Equation 2 are given
ref
Gϕ = ∑ x i GiSER ( T ) + form Gϕ ( T ) (12)
i
in Equation 6:
where GiSER ( T ) is the Gibbs energy of the pure component i
ref
Gϕ = ∑ x i Giϕ ( T ) in its standard element reference state (SER) and form Gϕ ( T ) is
i
the Gibbs energy of formation of the compound expressed,
id
G =RT ∑ x i lnx i
ϕ (6)
i
as a function of temperature, according to Equation 4.
ex
Gϕ = ex 2 Gϕ + ex 3 Gϕ + ...
2.3 Ordered Solutions
where: xi is, at the same time, the site fraction in the unique Ordered solutions are intermetallics with a reference
sublattice and the overall mole fraction of component i. stoichiometry, but characterised by more or less extended
Giϕ ( T ) is the Gibbs energy of the pure component i in the solubility ranges. The solubility mechanism is generally
ϕ phase (Equation 3), ex 2 Gϕ and ex 3 Gϕ are the binary and due to the reciprocal substitution between components
ternary excess terms, respectively. They are expressed in their crystallographic sites. In this case each component
according to Equations 7 and 8: can occupy two or more (possibly all) sublattices. A typical
example may be an ordered binary solution represented
n −1 n
Gϕ = ∑ ∑ x i x j ∑ ν Lϕi,j ( T ) ( x i − x j )
ν
ex 2
(7) by a two-sublattice model where each sublattice is mainly
i=1 j=i+1 ν occupied by one component but its substitution by the
n − 2 n −1 n
other element is allowed. Then both elements can occupy
ex 3
Gϕ = ∑ ∑ ∑ x x x (u L ( T ) +u L ( T ) +u L ( T ) )
i j k
ϕ
i i
ϕ
j j k
ϕ
k
(8) both sublattices. When a new component is added, it will
i=1 j=i+1 k=j+1 preferentially occupy one of the sublattices, but it will also
be allowed to occupy the other one.
respectively, with the coefficients given by Equation 9:
Whenever more components are allowed to mix
1− xi − x j − xk 1− xi − x j − xk in two or more sublattices the Gibbs energy is expressed
ui = x i + , uj = x j + , as a function of the site fractions y (s) , the mole fractions of
3 3 i
(9) each component i in the sublattice s. Site fractions obey the
1− xi − x j − xk
uk = x k + conditions given by Equation 13:
3
∑y (s)
i =1
Lϕ ( T ) in Equations 7 and 8 are empirical parameters whose i

temperature dependence is similar to that of Gϕ ( T ) given ∑n (s)

y i(s) (13)
in Equation 4, as presented in Equation 10: s
= xi
∑ n (1 − y )
(s)
VA
Lϕ ( T ) = A ϕ +B ϕ T +Cϕ TlnT +Dϕ T 2 + ... (10) s

Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016 19
Cacciamani

where n( s ) are the stoichiometric coefficients relating the Among them, only the second one corresponds
sublattices. Notice that Equation 13 has been formulated to a stable phase and can be evaluated on the basis of
assuming that vacancies (VA) can also be included in one experimental information. However the Gibbs energies of
ϕ ϕ ϕ ϕ
or more sublattices. This accounts for a second solubility formation of all of them, GMg :Mg, GMg :Cu, GCu :Mg, GCu :Cu, appear

mechanism, namely vacancy formation in one or more in Equation 14. The three values which correspond to
sublattices. metastable or even unstable compounds cannot be derived
Equation 6 becomes Equation 14: from experimental data, but, in many cases, can be directly
calculated by ab initio methods, or indirectly evaluated on
ref
Gϕ = ∑∑ ...∑ y i(1) y (j 2) ...y k( s ) Giϕ:j:...:k the basis of experimental phase equilibria.
i j k

1 2.4 Ordered Phases Related to Random Solutions

RT ∑∑ n( ) y i( ) lny i( )
s s s
id
Gϕ =
∑n s
(s)
s i (14) A special case, relatively frequent in intermetallic
ex ϕ ex 2 ϕ ex 3 ϕ systems, are the ordered phases which may be related to
G = G + G + ... =
disordered ones, as in the case of the AuCu3 type phase
∑∑∑ y (s) (s)
i yj ∑∑ y ( )L k
r ϕ
i,j :...:k ( T ) + ...+ ex 3 Gϕ + ... (Pearson symbol tP2, strukturbericht designation L1 2)
s i j r ≠s k
related to the disordered fcc structure (Cu type, Pearson
symbol cF4, strukturbericht designation A1), or the
where Giϕ:j:...:k in ref Gϕ are the Gibbs energies of all the
CsCl type phase (Pearson symbol cP2, strukturbericht
so-called “end members”, the stoichiometric compounds
designation B2) related to the disordered bcc structure
(either stable, metastable or unstable) formed when only
(W type, Pearson symbol cI2, strukturbericht designation
one constituent is present in each sublattice. Lϕi,j:...:k ( T ) in
ex 2 ϕ
A2). In some cases ordered and disordered structures are
G are binary interaction parameters corresponding to stable in the same system and can transform to each other
the mixing of components i and j on the sublattice s while by a first order as well as a second order transformation.
the other sublattices are singly occupied. More terms can A typical example of this behaviour can be found in the
be added to ex 2 Gϕ , corresponding to simultaneous mixing Al-Fe system. In this system, when the transformation is
on two sublattices while the remaining sublattices are singly second order, A2 and B2 are in equilibriun at the same
occupied. ex 3 Gϕ accounts for the possible interactions of composition (no two-phase field between them can exist)
three elements in a given sublattice. and this implies that there is no discontinuity between
A simple case is represented by the Laves phase MgCu2 the Gibbs energy of the two phases (or between their
type (Pearson symbol cF24, strukturbericht designation composition derivatives) at the equilibrium composition.
C15). It is described in the space Group Fd-3m (num. 227). Such a behaviour may be modelled by introducing one single
Mg occupies the 8a Wickoff position, while Cu is in 16d, but Gibbs energy expression for both ordered and disordered
both elements are allowed to occupy the other position, phases. The ordering state will depend on the occupation
at least partially. Then the phase may be modelled by a of the different sublattices. This means that, in the case of
2-sublattice model: A2/B2 ordering in Al-Fe, we can assume a two-sublattice
model, (Al,Fe)0.5 (Al,Fe)0.5, which become equivalent to
(8a)1 (16d)2 → (Mg, Cu)1 (Mg, Cu)2 (Al,Fe)1 when the two sublattices have identical occupation
and, consequently, become indistinguishable. This is realised
where each sublattice corresponds to a crystallographic
by introducing the expression for the Gibbs energy given
site and the 1:2 stoichiometry correspond to the ratio 8:16
by Equation 15:
between the site multiplicities. Each sublattice is typically
occupied by the element reported in bold, but all the G A 2 /B 2 = G A 2 ( x ) +GB 2 ( y1 , y 2 ) − GB 2 ( x, x ) (15)
elements are allowed to enter all sublattices to reflect the
substitution mechanism responsible for the homogeneity
where G A 2 ( x ) is the Gibbs energy of the disordered phase
range of the phase. calculated at the concentration x according to the one sublattice
It is important to notice that the model adopted model (Equation 6), while GB 2 ( y1 , y 2 ) and GB 2 ( x, x ) are the
generates four so called end-members, i.e. four stoichiometric Gibbs energies of the ordered phase (Equation 14) calculated
compounds which correspond to all the possible combinations first at the site fractions y1 , y 2 (ordered state) and then at
with each sublattice occupied by a single element: the site fractions x, x (disordered state). So, the difference
GB 2 ( y1 , y 2 ) − GB 2 ( x, x ) is the Gibbs energy of ordering.
(Mg)1 (Mg)2 = 3 Mg
If a 4 sublattice model is introduced, a larger number
(Mg)1 (Cu)2 = MgCu2 of ordered phases related to the same bcc lattice can be
described. This is illustrated in Figure 2.
(Cu)1 (Mg)2 = Mg2Cu
The correspondence between site occupancies and
(Cu)1 (Cu)2 = 3 Cu structures is the following:
20 Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016
 An introduction to the calphad method and the compound energy formalism (CEF)

Figure 2. Ordered structures based on the bcc lattice which may be represented by a four sublattice model.

Sublattice occupation Corresponding structure can be used to understand and/or predict the behaviour of
(A)1/4(A)1/4(A)1/4(A)1/4 A2 cI2-W real systems and how it can be used to solve application
problems.
(A)1/4(A)1/4(B)1/4(B)1/4 B2 cP2-CsCl
(A)1/4(B)1/4(A)1/4(B)1/4 B32 cF16-NaTl 3.1 CALPHAD assessment supported by quantum
DFT calculations: the Fe-Ni system
(A)1/4(B)1/4(C)1/4(C)1/4 cF16-AlScAg2
(A)1/4(B)1/4(B)1/4(B)1/4 D03 cF16-BiF3 Fe-Ni is a key system for a large number of
technological alloys. From a scientific point of view it shows
(A)1/4(B)1/4(C)1/4(D)1/4 cF16-AlLiMgSn an interesting combination of structural and magnetic
(A)1/4(B)1/4(C)1/4(A)1/4 L21 cF16-AlCu2Mn ordering in conjunction with very experimental difficulties
due to the very slow kinetics at temperatures lower than
All of them are special cases of the general model: about 500 °C. For all these reasons it is a very interesting
(A,B,C,D)1/4(A,B,C,D)1/4(A,B,C,D)1/4(A,B,C,D)1/4 playground for CALPHAD modelling. Recently [10] we
Which is used to express thermodynamic functions assessed the system by using four sublattice models for the
of this group of phases. fcc- and bcc-based ordered phases and used DFT atomistic
Similar 4-sublattice models may be used to describe calculation to evaluate the enthalpy of formation of all ordered
the ordered structures related to the fcc or hcp lattices. end-members. As a result we were able: 1) to confirm
To summarise, a selection of examples of CEF modelling the presence of the tricritical point (“Nishizawa horn”)
of different kind of phases is reported in the Table 1 below: related to the magnetic ordering of the fcc solid solution
(Figure 3a); 2) to confirm the low temperature stability of
the L10 fcc ordered phase (experimentally supported by
3 SELECTED EXAMPLES the analysis of meteoric samples), as well as the metastable,
low temperature miscibility gap in the fcc solid solution,
In the following a few examples of CALPHAD 3) to underline the role of the magnetic interactions in the
applications are reported, taken from recent results obtained establishment of the Fe-Ni phase equilibria; 4) to calculate
in the author’s lab. These examples show how CALPHAD the thermodynamic stability of the stable and/or metastable
Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016 21
Cacciamani

Table 1. Examples of CEF modelling

Phase type Typical sublattice model Example phase Example sublattice model
Substitutional disordered solutions (A,B) liquid-(Cu,Ni) fcc-(Co,Ni) (Cu,Ni) (Co,Ni)
Stoichiometric compounds (A)u(B)v Al3Ni (Al)3(Ni)1
Substitutional solutions with interstitials (A)u(B,Va)v bcc-(Fe-N) (Fe)1(N,Va)3
Substitutional ordered solutions (A,B)u(A,B)v C15-Cu2Mg (Cu,Mg)2(Cu,Mg)1
Ionic solids (Cations)u(Anions)v NaCl+CaCl2 (Ca2+,Na+)1(Cl–,Va)2
spinel (Mg2+,Al3+)1(Mg2+,Al3+)2(O2–)4
Semiconductors with point defects (A,B,B+) (A,A–,B) (Va,e-) (Va,h+) GaAs (As,Ga,Ga+) (As,As–,Ga) (Va,e–) (Va,h+)
Order/disorder relations (A,B) and (A,B)u(A,B)v fcc-(Au,Cu) and (AuCu3) (Au,Cu) and (Au,Cu)1/4(Au,Cu)3/4

Figure 3. (a) the calculated Fe-Ni phase diagram; (b) the calculated enthalpy of mixing of the ordered and disordered fcc and bcc Fe-Ni phases
compared with results of ab-initio calculations.

ordered and disordered phases (Figure 3b). This is especially surface and promotes the transition of carbon from the sp2
important for the modelling of higher order systems. to the sp3 hybridisation states. Once the hydrogen is released
as H2 molecules, the sp3 hybridisation is maintained and a
3.2 CALPHAD simulation of a non-equilibrium new layer of diamond structure has grown.
process: diamond synthesis by micro-plasma This non-equilibriun non-dissipative process may
CVD be modelled according to the CALPHAD approach by
introducing a hypothetical new phase, called “activated
It is well known that the equilibrium synthesis of diamond graphite” (G*), which represents graphite excited by the
crystals requires very high pressures and temperatures. By a interaction with atomic hydrogen. Its thermodynamic
non-equilibrium process, however, small diamond crystals can properties then depend on a combination of the properties
be obtained at low pressure and relatively low temperature. of regular graphite at the temperature of the substrate
When a high temperature plasma jet formed by a mixture of and the properties of atomic hydrogen at the temperature
hydrogen and methane impinges on a metal substrate at a of the plasma. According to this model, first introduced
temperature of about 500-1000 °C various carbon structures by Wang et al. [11], diamond can grow when it is more
(graphite, diamond, graphene, nanotubes...) may grow on stable than the activated graphite. In a recent paper [12]
the metal surface, depending on the process parameters, we calculated, in a pressure vs. temperature diagram, the
namely total pressure, metal substrate temperature, plasma diamond growth region and verified the good agreement
temperature, gas composition. A possible explanation of of the model calculations with the experimental results,
the diamond formation is that the highly energetic atomic as shown in Figure 4. Notice that exotic carbon forms
hydrogen present in the plasma is adsorbed on the carbon are obtained in the region where activated graphite is
22 Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016
 An introduction to the calphad method and the compound energy formalism (CEF)

more stable than diamond, while, on the opposite side Wetting and joining properties of these materials
of the diamond region, regular graphite is obtained, as with metals and alloys are scarcely known. So they have
expected. been recently investigated, especially by means of wetting
experiments carried out using the sessile drop technique
3.3 CALPHAD simulation of wetting experiments: and assisted by CALPHAD calculations [13-17].
Ni alloys in contact with UHTC In one of the mentioned works [16], for example,
phase diagram calculations easily explained the observed
UHTCs (Ultra High Temperature Ceramics) are sequence of unusual isothermal transformations occurring
materials characterised by melting temperatures in excess during a wetting experiment of a Ni-B alloy in contact with
of 3000°C. Among them diborides of the group 4 metals a ZrB2 substrate.
(Ti, Zr, Hf) are candidate materials for application in extreme As shown in Figure 5, the Ni-17at%B alloy in contact
conditions such as leading edges of vehicles for hypersonic with ZrB2 at 1130°C melts and reacts with the ceramic
flight, atmospheric re-entry vehicles, rocket propellers, substrate. During the isothermal reaction the composition of
refractory linings, electrodes, cutting tools, etc. the liquid drop moves along the line shown in the Figure 5a

Figure 4. The pressure vs. temperature phase diagram of carbon where the stability areas of diamond and activated graphite are superimposed
and compared with experimental results. The coloured area represent the range where diamond can be obtained by micro-plasma CVD process.

Figure 5. Evaluation of the reaction between a Ni-B alloy drop and a ZrB2 substrate. a) and b) Equilibrium during wetting. c) Equilibrium
during cooling.

Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016 23
Cacciamani

until it reaches the saturation composition, in equilibrium 4 CONCLUSION

with ZrB2. Doing so it crosses the two-phase field where the
ternary compound Zr3Ni20B6 becomes stable. Accordingly, CALPHAD is a very useful and powerful method
during the isothermal wetting experiment precipitation in materials science and processing. Thank to its
and subsequent re-dissolution of a solid phase is observed, ability to predict phase equilibria and thermodynamic
as evidenced in Figure 5b. Once the equilibrium between functions in multi-phase and multi-component systems
ZrB2 and the drop is reached, the system is cooled down. it has been successfully applied to the description and
In Figure 5c the solid phases formed during cooling are simulation of very complex technological materials and
shown. They correspond to the phases actually identified processes. Thanks to the integration of CALPHAD
in the solidified drop. modelling with DFT and/or molecular dynamics
The same procedure has been successfully used to calculations, on one side, and kinetic simulations, on
predict and/or interpret the results of many more wetting the other side, an “all scales” modelling of materials
experiments. may be achieved.

REFERENCES

1 Kaufman L, Bernstein H. Computer calculation of phase diagrams. New York: Academic Press; 1970.
2 Spencer PJ. A brief history of CALPHAD. Calphad. 2008;32:1-8.
3 Saunders N, Miodownik AP. CALPHAD (calculation of phase diagrams): a comprehensive guide. Oxford: Pergamon;
1998.
4 Hillert M. Phase equilibria, phase diagrams and phase transformations: their thermodynamic basis. 2nd ed.
Cambridge: Cambridge University Press; 2008.
5 Lukas HL, Fries SG, Sundman B. Computational thermodynamics: the CALPHAD method. Cambridge: Cambridge
University Press; 2007.
6 Hillert M. The compound energy formalism. Journal of Alloys and Compounds. 2001;320(2):161-176.
7 Inden G. The role of magnetism in the calculation of phase diagrams. Physica B+C. 1981;103(1):82-100.
8 Hillert M, Jarl M. A model for alloying in ferromagnetic metals. Calphad. 1978;2:227-238.
9 Dinsdale AT. SGTE data for pure elements. Calphad. 1991;15:317-425. http://dx.doi.org/10.1016/0364-
5916(91)90030-N.
10 Cacciamani G, Dinsdale A, Palumbo M, Pasturel A. The Fe–Ni system: thermodynamic modelling assisted by
atomistic calculations. Intermetallics. 2010;18:1148-1162.
11 Wang J-T, Wei Zhang D, Ding S-J, Wang P-F. A new field of phase diagrams of stationary nonequilibrium states.
Calphad. 2000;24:427-434.
12 Ghezzi F, Cacciamani G, Caniello R, Toncu DC, Causa F, Dellasega D, et al. Carbon structures grown by direct
current microplasma: diamonds, single-wall nanotubes, and graphene. The Journal of Physical Chemistry C.
2014;118:24714-24722.
13 Kaufman L, Cacciamani G, Muolo ML, Valenza F, Passerone A. Wettability of HfB2 by molten Ni(B) alloys
interpreted by CALPHAD methods, Part 1: definition of the B–Hf–Ni system. Calphad. 2010;34:2-5.
14 Cacciamani G, Riani P, Valenza F. Equilibrium between MB2 (M=Ti,Zr,Hf) UHTC and Ni: a thermodynamic database
for the B–Hf–Ni–Ti–Zr system. Calphad. 2011;35:601-619.
15 Valenza F, Muolo ML, Passerone A, Cacciamani G, Artini C. Control of interfacial reactivity between ZrB2 and
Ni-based brazing alloys. Journal of Materials Engineering and Performance. 2012;21:660-666.
16 Artini C, Muolo ML, Passerone A, Cacciamani G, Valenza F. Isothermal solid–liquid transitions in the (Ni,B)/ZrB2
system as revealed by sessile drop experiments. Journal of Materials Science. 2013;48:5029-5035.
17 Artini C, Muolo ML, Passerone A, Valenza F, Manfrinetti P, Cacciamani G. Experimental investigations and
thermodynamic modeling in the ZrB2Ni section of the BNiZr system. Journal of Alloys and Compounds.
2014;592:115-120.

Received: 27 Nov. 2015

Accepted: 18 Dez. 2015
24 Tecnol. Metal. Mater. Miner., São Paulo, v. 13, n. 1, p. 16-24, jan./mar. 2016