The CALPHAD Method

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

1001

The CALPHAD
20. The CALPHAD Method

20.1 Outline of the CALPHAD Method.............. 1002


Phase diagrams offer various areas of materials
20.1.1 Description of Gibbs Energy ........... 1002
science and technology indispensable informa-
20.1.2 Equilibrium Conditions ................. 1004
tion for the comprehension of the properties of
20.1.3 Evaluation
materials. The microstructure of solid materials of Thermodynamic Parameters....... 1005
is generally classified according to the size of
the constituents – for example, at the electron, 20.2 Incorporation of the First Principle
atomic, or granular level (see Chap. 3). Accordingly, Calculations into the CALPHAD Approach . 1006
fundamental principles like quantum mechan- 20.2.1 Outline
ics, statistical mechanics, or thermodynamics of the First Principle Calculations ... 1006
20.2.2 Gibbs Energies of Solution Phases
are applied individually to describe the physi-
Derived by the First Principle
cal properties. Phases are important features of
Calculations................................. 1007
material because they characterize homogeneous
20.2.3 Thermodynamic Analysis
aggregations of matter with respect to chemical of the Gibbs Energies Based

Part E 20
composition and uniform crystal structure. The on the First Principle Calculations .. 1011
various functions of a material are closely related 20.2.4 Construction of Stable
to the phases and structures of the material’s and Metastable Phase Diagrams..... 1011
composition. Therefore, to develop a material with 20.2.5 Application
a maximum level of desired functions, it is essential to More Complex Cases.................. 1013
to undertake design of the structure in advance.
20.3 Prediction of Thermodynamic Properties
Phase diagrams are composed by means of
of Compound Phases with
experimental measurements, as well as statisti-
First Principle Calculations..................... 1019
cal thermodynamic analysis. The construction of
20.3.1 Thermodynamic Analysis
phase diagram calculations based on experiments of the Fe–Al–C System .................. 1019
and thermodynamic analysis are generally referred 20.3.2 Thermodynamic Analysis
to as the calculation of phase diagrams (CALPHAD) of the Co–Al–C and Ni–Al–C
approach [20.1]. This method provides a very ac- Systems ...................................... 1023
curate understanding of the properties originating
in the macroscopic character of the material under References .................................................. 1030
study.
This chapter is organized in three parts:
for metastable phase equilibria in the Fe−Be,
• In the first part, a brief outline of the CALPHAD and Co−Al binary systems are shown.
method is summarized. • In the third part the application to predict ther-
• In the second part, the method for deriving the modynamic properties of compound phases is
Gibbs free energies incorporating the ab initio discussed. The thermodynamic modeling for
calculations is presented in order to clarify the the Perovskite carbide with an E21 -type struc-
uncertainty of thermodynamic properties for ture in the Fe−Al−C, Co−Al−C and Ni−Al−C
metastable solution phases, taking theFe−Be- ternary systems is illustrated, and constructions
based bcc phase as an example. Some results of phase diagrams are performed.
1002 Part E Modeling and Simulation Methods

20.1 Outline of the CALPHAD Method


Phase diagrams provide basic and important informa- to an attractive term in (20.1). The constants P and Q
tion especially for the design of new materials. However, are proportionality constants, derived empirically. Equa-
by using experimental techniques only, a lot of time tion (20.1) does not contain any parameters that refer to
and labor is required in construction of even a partial the crystal structure, and therefore, the heats of solu-
region of a phase diagram, because the practical mater- tion of face-centered-cubic (fcc), body-centered-cubic
ials are composed of multicomponent alloys, more than (bcc), and hexagonal-close-packed (hcp) structures, as
ternary systems. In order to break this difficult situa- well as liquids, are predicted. However, it is known that
tion, the method of calculation of phase diagrams was the absolute values are not always predicted accurately,
advocated and is referred to as the CALPHAD method. although this model predicts the correct sign for the for-
CALPHAD was originally a name for this researcher’s mation enthalpy in various alloy systems. For example,
group, however it has recently become the name for the thermodynamic analysis of the Fe−Pd system predicts
technique by which a phase diagram is calculated on the enthalpy of formation for the L10 and the L12 ordered
the basis of thermodynamics. In this method a variety phases to be −14 kJ/mol and −22 kJ/mol, respectively,
of experimental values concerning the phase boundaries while the predicted values from Miedema’s model are
and the thermodynamic properties is analyzed according −6 kJ/mol for the L10 phase and −4 kJ/mol for the
to an appropriate thermodynamic model and the inter- L12 phase [20.3]. The large difference between these
action energies between atoms are evaluated. By using values means that incorporating Miedema’s model with
this technique, phase diagrams outside the experimen- a thermodynamic analysis is rather difficult.
Part E 20.1

tal range can be calculated based on thermodynamic Such serious problems, which are intrinsic to the
proof. Difficulty in extension of the calculated results CALPHAD approach, should be solved with the assis-
to higher-order systems is much less than that in the tance of the first principle energetic calculation method.
case of experimental work, since the essence of the cal- Thus, in the present chapter, a new approach to in-
culation does not change so much between a binary troduce some thermodynamic quantities obtained by
system and a higher-order system. This method provides the ab initio band energy calculations in the conven-
a very accurate understanding of the properties originat- tional CALPHAD-type analysis of some alloy systems
ing in the macroscopic character of the material under is presented.
study. However, a shortcoming of this approach is that
it is hard to obtain information on metastable equilib- 20.1.1 Description of Gibbs Energy
ria, or on undiscovered phases, since the thermodynamic
parameters from this method can only be evaluated from Gibbs Free Energy
experimental data. Using the Regular Solution Approximation
The empirical model of de Boer et al. [20.2] has of- The selection of a thermodynamic model by which
ten been used to deduce the thermodynamic quantities of Gibbs free energy of an alloy system is described is
systems where the experimental values do not exist. In the most important factor when using the CALPHAD
this model, generally called Miedema’s model, the crys- method. In a system in which interaction between alloy-
tal lattice is divided into Wigner–Seitz cells. The simple ing elements is not so strong, the regular solution model
expression for the formation energy in alloys is derived is well known to describe the thermodynamic properties
by considering the change in the electron density states of the alloys comparatively well. For instance, the free
at the interface between the cells, as shown in (20.1) energy of the A−B binary alloy is expressed as (20.2)
  G m = xA 0G A + xB 0G B + xA xB L AB
1/3 2
ΔE ∝ −P (ΔΦ)2 + Q Δn WS . (20.1)
+ RT (xA ln xA + xB ln xB ) , (20.2)
Here, Δn WS is the difference in electron density based 0G
where i is the Gibbs energy of the pure component i
on the volume of the Wigner–Seitz cell between differ- in the standard, and xi is the mole fraction of i. L AB
ent species of atoms. This term always leads to local shows the interaction energy for A atoms and B atoms,
perturbations that give rise to a positive energy con- and is given by
tribution in (20.1). On the other hand, ΔΦ presents  
1
a difference between the chemical potentials of the dif- L AB = NZ εAB − (εAA + εBB ) , (20.3)
ferent species of atoms at the cell surfaces, and leads 2
The CALPHAD Method 20.1 Outline of the CALPHAD Method 1003

where N denotes the so called Avogadro’s number, Z is basic end members Aa Cb , Aa Db , Ba Cb , and Ba Db , as
the coordination number, and εAB , εAA and εBB show the shown in Fig. 20.1a. The symbols yB1 and yD 2 denote the

cohesion energies between A–B, A–A, B–B atomic pair, mole fractions, respectively, for the sublattices 1 and 2,
respectively. The ordering between unlike atoms occurs and they are represented as follows:
for L AB < 0 because the A–B atom pair is more stable
than the average of the A–A and B–B atom pairs. On the n 1B n 2D
yB1 = , 2
yD = . (20.4)
other hand, the clustering of the like atom is caused in n 1A + n 1B n 2C + n 2D
case of L AB > 0. For L AB = 0, no atomic interaction ex-
ists, and then the solution shows the random mixing of j
In this equation n i denotes the number of i atoms in
atoms. This interaction parameter should not be constant the j sublattice. The nonplanar surface composed of
but a function of the temperature and the chemical com- four reference states, i. e., one each for the compounds
positions of alloys. From a qualitative point of view, the Aa Cb , Aa Db , Ba Cb , and Ba Db , is called the surface of
first term of (20.2) represents the energies of mechan- reference as shown in Fig. 20.1b, and represented by
ical mixing, the second term stands for excess energy
m = yA yC G Aa Cb + yA yD G Aa Db
G ref
which shows deviation from the ideal solution, and the 1 20 1 20
third is ideal mixing entropy. + yB1 yC2 0G Ba Cb + yB1 yD
20
G Ba Db . (20.5)

Gibbs Free Energy Using the Sublattice Model For the entropy term, the following Gibbs energy of mix-
For the simplest case of the sublattice model [20.4], ing can be obtained, assuming the atoms mix randomly
let us consider a phase represented by the formula

Part E 20.1
within each sublattice
(A, B)a (C, D)b , containing two elements A and B in one   
mix = RT a yA ln yA + yB ln yB
G ideal
sublattice “1” and two elements C and D in the other sub- 1 1 1 1
lattice “2”. The coefficients a and b show the number of  
sites in each sublattice and the size of the sublattices may + b yC2 ln yC2 + yD2 2
ln yD . (20.6)
be generally chosen as a + b = 1 for convenience. There
are four elements in the system, however the degrees of It seems difficult to accept that the interaction between
freedom in varying in composition are two, because the A and B atoms should be quite independent of whether
total numbers of atoms in each sublattice are fixed. Thus the other sublattice is occupied by C and D atoms. Thus
these two composition parameters are conveniently plot- the regular solution model should be defined as the
ted on a square where the corners represented the four power series representation of excess Gibbs free energy

a) AaDb b)
BaDb
yD2

0
GAaDb
0
GBaDb
0
GBaCb

0
GAaCb

AaCb BaCb
yB1

Fig. 20.1 (a) Composition square in a quaternary system (A, B)a (C, D)b and (b) surface of reference for the free energy
1004 Part E Modeling and Simulation Methods

by the following equation the CrC and FeC compounds, where the interstitial sites
are completely filled with C. The second line shows the
G xs
mix = yA yB yC L A,B:C + yA
1 1 20 1 1 20
yB yD L A,B:D
ideal entropy term, while the residual lines represents
+ yC2 yD yA L A:C,D + yC2 yD
2 10
yB L B:C,D ,
2 10 the excess terms of the Gibbs energy. When yC2 = 0 in
(20.7) the equation, the Gibbs energy equation coincides with
the equation for the simple substitutional mixing of Cr
where L i, j:k (or L i: j,k ) is the interaction parameter
and Fe.
between unlike atoms on the same sublattice. Conse-
quently the Gibbs energy for one mole of the phase,
Magnetic Contribution to Gibbs Energy
denoted as (A, B)a (C, D)b , can be represented by the
In magnetic materials containing Fe, Ni and Co, there
two-sublattice model as shown in (20.8):
is polarization of electron spins, and it is necessary to
G m = G ref
m + G mix + G mix .
ideal xs
(20.8) consider the magnetic contribution to the Gibbs energy,
by adding the extra term to the paramagnetic Gibbs
energy G m as follows:
Gibbs Energy for Interstitial Solutions
Represented by Sublattice Model G = G m + G mag . (20.10)
The Gibbs energy for interstitial phases is very impor-
tant in steels and ferrous alloys, where the elements The magnetic contribution to the Gibbs free energy,
such as C, N, S, or B occupy the interstitial sites of G mag , is given by the expression
solid solutions. The structure of such a phase is consid- G mag = RT f (τ) ln(β + 1) , (20.11)
Part E 20.1

ered as consisting of two sublattices, one occupied by


substitutional elements and the other occupied by the where
interstitial elements as well as vacancies. Let us con-  
1 79τ −1 474 1
sider the austenite phase for the Fe−Cr−C system, for f (τ) = 1 − + −1
instance. The occupation of the sublattices is shown as A 140 p 497 p
 3 
(Cr, Fe)1 (C, Va)1 , since the number of sites in each sub- τ τ9 τ 15
× + + , for τ < 1
lattice is 1 : 1 in the case of an fcc structure. The Gibbs 6 135 600
energy is given by the following equation: (20.12)
Gm = y1 y2 0G Cr:Va + y1 y2 0G Fe:Va and
Cr Va Fe Va

+ y1 y2 0G Cr:C + y1 y2 0G Fe:C 1 τ −5 τ −15 τ −25
Cr C Fe C
f (τ) = − + + , for τ ≥ 1 ,
 A 10 315 1500
(20.13)
+ RT y1 ln y1 + y1 ln y1
Cr Cr Fe Fe
  and

+ y2 ln y2 + y2 ln y2 518 11 692 1
C C Va Va A= + −1 . (20.14)

 v 1125 15 975 p
v
+ y1 y1 y2 L Cr,Fe:Va y1 − y1 The variable τ is defined as T/TC , where TC is the Curie
Cr Fe Va Cr Fe
v
 v temperature and β is the mean atomic moment expressed

in Bohr magnetons μB . The value of p depends on the
v
+ y1 y1 y2 L Cr,Fe:C y1 − y1
Cr Fe C Cr Fe structure, and p = 0.28 for the fcc phase.
v

 v
v
+ y1 y2 y2 L Cr:C,Va y2 − y2 20.1.2 Equilibrium Conditions
Cr C Va C Va
v

 v
v
Figure 20.2 shows the so-called Gibbs energy-
+ y1 y2 y2 L Fe:C,Va y2 − y2 . composition diagram for the A−B binary system. This
Fe C Va C Va
v figure schematically shows the relationship between
(20.9)
a eutectic type of binary phase diagram and the Gibbs
The first two terms represent the Gibbs energy of fcc Cr energy curves of each phase. Let us consider the case in
and Fe, because the second sublattice consists of vacan- which the microstructure of the alloy is composed of the
cies. The next two terms represent the Gibbs energy of α and β phases. When the Gibbs energies of the α and
The CALPHAD Method 20.1 Outline of the CALPHAD Method 1005

The line ae be is a common tangent of Gibbs energy G α


Temperature
and G β , and the equilibrium composition of each phase
β
is given as xBα and xB . In the calculation of the phase
equilibrium, it is convenient to use these chemical po-
Liq tentials as the basic equations. For instance, to obtain the
equilibrium between phase α and phase β, (20.4) should
be calculated by using the numerical analytic method
β
μαA = μA ,
α β β
μαB = μB . (20.16)
T1 The chemical potential μ of each element is obtained by
using

2
∂G m ∂G m
μi = G m − xj + (20.17)
Gibbs energy ∂x j ∂xi
j=1
GLiq G
α and hence the chemical potentials of A and B species in
β the φ phase are expressed as:
G
 2
φ φ φ φ φ
μA = 0G A + RT ln xA + xB L AB ,

Part E 20.1
 2
φ φ φ φ φ
μB = 0G B + RT ln xB + xA L AB . (20.18)
μ Bα
b1
G1
be μBβ
a1 20.1.3 Evaluation
Ge of Thermodynamic Parameters
μ Aα
ae The interaction parameters included in the free energy
expression can be directly evaluated from the experi-
μAβ
mental values on activity. For example the activity of
Pb over the Pb−Sb binary liquid have been determined
xB1
as shown in Table 20.1 [20.5]. The liquid state of the
β α β
A x Bα f f xB B elements is adopted as its standard. If the composition
dependency of the interaction parameter is expressed
Content
as (20.19), the chemical potential of Pb in the binary
Fig. 20.2 Eutectic type of phase diagram and correspond- system can be derived as (20.20)

ing Gibbs energy-composition diagram for hypothetical L PbSb = 0 L PbSb + xPb − xSb 1 L PbSb , (20.19)
A−B binary system 2
β μPb = 0G Pb + RT ln xPb + 1 − xPb
β phases are given by the points a1 G α1 and b1 G 1 , re-  
1
spectively, the total energy of the solution is represented × L PbSb + 4xPb − 1 L PbSb . (20.20)
0
by the line a1 b1 in the figure. At the alloy composition
xB1 , the Gibbs energy is G 1 and each phase has the vol-
ume fraction of f α and f β . Then the following relation Table 20.1 Experimental data on activity of Pb in the
holds: Pb−Sb binary liquid
β
G 1 = G α1 f α + G 1 f β . (20.15) xPb 0.05 0.1 0.2 0.3 0.4 0.5
The thermodynamic equilibrium is achieved under the aPb 0.04 0.08 0.162 0.247 0.338 0.433
condition of the lowest energy of the alloy. According to xPb 0.6 0.7 0.8 0.9 0.95
this criterion, the point G 1 is not the lowest energy state,
aPb 0.555 0.674 0.785 0.897 0.946
but decreases further to G e by changing f α and f β .
1006 Part E Modeling and Simulation Methods

This is straightforwardly arranged as


Activity of Pb
  1.0
RT ln aPb /xPb 
 2 = 0 L PbSb − 1 L PbSb 0.9
1 − xPb Oelsen et al. (903 K)
0.8
+ 4xPb 1 L PbSb (20.21)
0.7

since the relation μPb = 0G Pb + RT ln aPb holds. There- 0.6


fore, if the left-hand side of (20.21) is calculated from
0.5
the activity data shown in Table 20.1 and the each value
is plotted against xPb , the interaction parameters 0 L PbSb , 0.4
1L
PbSb can be obtained, respectively, from the intercept
of the ordinate and the slope of the straight line as: 0.3

0
L PbSb = −2900 J/mol , 1
L PbSb = −200 J/mol . 0.2

(20.22) 0.1

The calculated activity of Pb in the liquid phase 0


0 0.2 0.4 0.6 0.8 1.0
was compared with the experimental values in Fig. 20.3.
Part E 20.2

Sb / atom fraction
Considering the other observed data set at a different
temperature yields the dependence on the temperature Fig. 20.3 Calculated activity of Pb in the Pb−Sb binary
for the interaction parameters 0 L PbSb , 1 L PbSb . liquid compared with experimental values [20.5]

20.2 Incorporation of the First Principle Calculations


into the CALPHAD Approach
In this section, in order to clarify the uncertainty of ther- Minimizing the grand potential with respect to all the
modynamic properties for metastable solution phases, correlation functions allows for the Gibbs energy of
the method for deriving the Gibbs free energies in- mixing to be obtained as a function of composition at
corporating the first principle calculations is presented, a constant temperature.
taking the Fe−Be- based bcc phase as the example.
First, a set of some superstructures is selected to be rep- 20.2.1 Outline
resentative of a series of bcc-based ordered phases, and of the First Principle Calculations
the total energies are calculated using the first princi-
ple calculations. Next, the effective cluster interaction First principle calculations using density functional
energies can be extracted from these formation ener- theory (DFT) have proved to be quite reliable in
gies using the cluster expansion method (CEM). This condensed matter physics. There still remains a bar-
leads to a set of composition-independent parameters, rier to overcome in application to materials science,
from which the energy of the set of superstructures can for instance, stoichiometric deviations, surfaces, im-
be reproduced in terms of a set of correlation func- purities, and grain boundaries, however, for certain
tions. Once we know the cluster interaction energies materials, the direct application of this technique has
for the alloy system, the enthalpy term is expressed by been attempted in studying the various properties us-
the combination of effective cluster interaction and the ing 100 or more atoms in a unit cell. According to
correlation functions for the clusters. The entropy term the theorem based on DFT, the total energy E tot of
in the Gibbs energy expression is calculated using the a nonspin-polarized system of interacting electrons in
cluster variation method (CVM) with the tetrahedron an external potential is given as a function of the
approximation to calculate the configurational entropy. ground state electronic density ρ as in the following
The CALPHAD Method 20.2 Incorporation of the First Principle Calculations into the CALPHAD Approach 1007

equation: and the other using much more complex basis sets such

as the full potential linearized augmented plane wave
E tot (ρ) = V (r)ρ(r) d r + T [ρ]
3
(FLAPW), which gives accurate results on formation
energies for metals. The FLAPW method, as embodied
e2 ρ(r)ρ(r  ) 3 3 
+ d r d r + E XC [ρ] . in the WIEN2k software package [20.7], is one of the
2 |r − r  |
most accurate schemes for electronic calculations, and
(20.23)
 allows for very precise calculations of the total energies
Here the first term V (r)ρ(r) d3r
denotes the Coulomb in a solid, and will be employed in the present ener-
interaction energy between the electrons and nu- getic calculations. The FLAPW method uses a scheme
clei, T [ρ] is the single-particle kinetic energy,
 ρ(r)ρ(r for solving many-electron problems based on the lo-
)
e2 3 3 
2 |r−r  | d r d r is the Hartree component of the cal spin density approximation (LSDA) technique. In
electron–electron energy, and E XC [ρ] is the exchange- this framework, a unit cell is divided into two regions:
correlation functional. nonoverlapping atomic spheres and an interstitial re-
Kohn and Sham [20.6] showed that the correct den- gion. Inside the atomic spheres, the wave functions of
sity in the equation is given by the self-consistent the valence states are expanded using a linear combina-
solution of a set of single particles following tion of radial functions and spherical harmonics, while
Schrödinger-like equations as a plane wave expansion is used in the interstitial region.

The LSDA technique includes an approximation for both
εi Ψi (r) = − ∇ 2 + V (r) + e2 the exchange and correlation energies, and it has been
2m
 recently enhanced by the addition of electron density
ρ(r  ) 3 

Part E 20.2
× d r + VXC (r) Ψi (r) . gradient terms to the exchange-correlation energy. This
|r − r  | has led to a generalized gradient approximation (GGA),
(20.24) as suggested by Perdew et al. [20.8], and we used this
In (20.24), VXC represents the exchange correlation improved method rather than the LSDA approach.
potential and VXC (r) = ∂E XC [ρ]/∂ρ(r). Thus we can
obtain an appropriate electronic density in much easier 20.2.2 Gibbs Energies of Solution Phases
way as compared with solving a many-body Schrödinger Derived by the First Principle
equation. In calculations, the single particle Kohn–Sham Calculations
equation in (20.24) is solved separately on a grid of sam-
pling points in the irreducible part of the Brillouin zone, The formation enthalpies of some binary ordered struc-
and the obtained orbitals are used to construct the charge tures derived by the first principle calculations are
density. compared with experimental data in Table 20.2 [20.9].
The first principle calculations based on DFT may In this comparison, alloy systems where the thermody-
be, at this time, divided into two methods; one employ- namic analysis had already been completed, are selected.
ing pseudopotentials and relatively simple basis sets, The bcc- and fcc-based ordered phases are included

Table 20.2 Comparison of the estimated formation enthalpies for some ordered structures with the appropriate evaluated
data
Alloy Structures Temperature Experimental values Calculated values
systems (K) (kJ/mol) (kJ/mol)
Al−Ni B2 298 − 71.7 − 69.5
L12 298 − 41.0 − 40.5
Al−Ti L10 298 − 39.8 − 39.6
D019 (Ti3 Al) 298 − 27.5 − 27.3
Be−Cu B2 298 − 21.4 − 19.0
Fe−Pd B2 298 − 15.6 − 13.2
Fe−Ti B2 298 − 51.6 − 46.5
Ni−Ti B2 999 − 34.2 − 34.1
D024 (Ni3 Ti) 827 − 54.0 − 51.4
Al−Li B32 298 − 22.8 − 23.1
1008 Part E Modeling and Simulation Methods

in the calculations. As can be seen in this table, the where φ denotes the type of superstructure and V is the
calculated values and evaluated data are reasonably volume.
consistent, and this close agreement encourages us to Then, the effective cluster interaction energies,
proceed to the next step to evaluate the Gibbs free {νi (V )}, can be extracted from these formation energies
energies of solution phases. using the cluster expansion method (CEM) developed
by Connolly and Williams [20.10]. This leads to a set of
First Principle Calculations composition-independent parameters, from which the
of the Gibbs Free Energy energy of the set of superstructures can be reproduced
The bcc phase in the Fe−Be binary system is illustrated φ
in terms of a set of correlation functions {ξi },
as an example to derive the Gibbs energy on the basis
γ

of ab initio method, since the phase is metastable and φ φ


located in the central part of the phase diagram, the ΔE form (V ) = νi (V ) × ξi , (20.26)
thermodynamic properties in this region are unknown. i=0
An outline for deriving the Gibbs free energies of where νi (V ) is the effective interaction energy of the
φ
the bcc-based structures incorporating the ab initio cal- i-point cluster, and ξi is the correlation function for clus-
φ
culations is as follows. First, a set of superstructures ter i in the phase φ; ξi is defined as the ensemble average
{A-A2, A3 B–D03 , AB-B2, AB-B32, AB3 –D03 , B-A2} of the spin operator σ( p), which takes values of ±1, de-
are selected to be representative of a series of bcc-based pending on the atom occupancy of the lattice site p.
ordered phases, the total energies were calculated us- φ
The values of ξi for the superstructures considered in
ing the FLAPW method. With the known total energies, this study are summarized in Table 20.3. The formation
Part E 20.2

the formation energy of the bcc-based superstructures, energies to the segregation limit for the Fe−Be binary
φ
ΔE form , is defined by averaging the total energy of the system in the D03 , B2, B32, and A2 structures in the
elements with chemical composition to the segregation ground state are summarized in Table 20.4 [20.11].
limit, as shown in the following equation The upper limit of the summation in (20.26) γ speci-
fies the largest cluster that participated in the expansion.
  In the case of the Fe−Be system, a tetrahedron cluster is
φ φ φ
ΔE form (V ) = E tot (V ) − 1 − x bccFe
E tot VFe considered, as schematically illustrated in Fig. 20.4, as
Be being the largest cluster, since in the bcc structure, the
φ
− x E tot
bccBe
VBe , (20.25) tetrahedron forms an irregular shape containing both the
Be first- and second-nearest neighbor distances. The second
φ
Table 20.3 The values of ξi for the bcc superstructures {A-A2, A3 B–D03 , AB-B2, AB-B32, AB3 –D03 , B-A2}
Ordered structures ξ0 ξ1 point ξ2 n.n.pair ξ3 n.n.n.pair ξ4 triangle ξ5 tetrahedron
A-bcc 1 1 1 1 1 1
A3 B–D03 1 1/2 0 −1 −1 −1
AB-B2 1 0 −1 1 −1 1
AB-B32 1 0 0 −1 0 1
AB3 –D03 1 −1/2 0 −1 1 −1
B-bcc 1 −1 1 1 −1 1

Table 20.4 The formation energies to the segregation limit for the Fe−Be binary system in the D03 , B2, B32, and A2
structures in the ground state
Basic lattice Composition of Be Strukturbericht Formation enthalpy (kJ/mol)
bcc 0 Fe A2 0.0
bcc 0.25 Fe3 Be D03 − 12.3
bcc 0.5 FeBe B2 − 31.4
bcc 0.5 FeBe B32 − 9.3
bcc 0.75 FeBe3 D03 − 18.6
bcc 1 Be A2 0.0
The CALPHAD Method 20.2 Incorporation of the First Principle Calculations into the CALPHAD Approach 1009

Fig. 20.4 Various clusters consisting of ordered structures

pair interaction was also taken into account, which leads where B, B  , and V0 are the bulk modulus, its pressure
to the consequence that mathematically, five γ should derivative, and the equilibrium volume, respectively,
represent six types of cluster. From Table 20.3, we can at normal pressures. Table 20.5 shows the coefficients

Part E 20.2
φ
see that {ξi } is a 6 × 6 matrix and that it is a regular ma- in (20.28) for each ordered structure. In this table,
trix. The matrix inversion of (20.26) yields the effective E(V ) is the total energy of each ordered structure
interaction energies as in the equilibrium volume of the B2 structure; i. e.,
V0 = 124.0438 a.u.3 . The formation energy of each
5   φ

φ −1 φ structure is represented as ΔE form , by defining the
νi (V ) = ξi ΔE form (V ) . (20.27) average concentration of the total energy of the bcc
i=0 Fe and the bcc Be phases at the segregation limit.
By applying (20.28) to these formation energies, the
Thus the interatomic interactions can be estimated by
expanding the total energies of the ordered structures
obtained from the band-energy calculations. Table 20.6 The effective interaction {νi (V )} corresponding
Using the Murnaghan equation of states [20.12], as to the clusters
shown in (20.28), the total energies of the A2, B2, B32, Effective cluster interactions (kJ/mol)
and the D03 structures were expressed as a function of Tetrahedron CVM
the volume
ν1 − 4.0
   B 
BV  V0 V0 ν2 28.8
E(V ) =   B 1− + −1 ν3 − 4.8
B (B − 1) V V
ν4 4.0
+ E V0 , (20.28)
ν5 6.2

Table 20.5 The coefficient of the Murnaghan equation of states for the bcc superstructures
Structures B B E(V0 ) V0
(GPa) (Ry/a.u.3 ) (Ry) (a.u.3 )
Fe (bcc) 334.2450 5.5525 − 5091.1796 155.5298
Fe3 Be (D03 ) 75.3890 0.7851 − 3833.1698 144.3524
FeBe (B2) 180.8774 3.6243 − 2575.1689 124.0438
FeBe (B32) 87.6360 5.9438 − 2575.1367 122.6823
FeBe3 (D03 ) 81.5496 3.6914 − 1317.1225 112.6662
Be (bcc) 244.0350 3.0866 − 59.0657 105.6658
1010 Part E Modeling and Simulation Methods

effective interaction, {vi (V )}, corresponding to each tropy formula is


cluster is calculated, as shown in Table 20.6, where
ΔS = kB ×
the positive numbers define the attractive force work-  12  
ing between unlike atoms. The value of ν2 represents  
the interaction energy between the nearest neighbor Nz ijk ! (Nxi )!
i, j,k i
atoms, while ν3 denotes the next-nearest neighbor ln  6  4  3 ,
interaction.   
The Gibbs free energies of the metastable bcc-based Nwijkl ! Nyij ! Nyij !
i, j,k,l i, j i, j
phase in the Fe−Be binary system are evaluated us-
ing the effective cluster interaction energies up to the (20.31)
tetrahedron cluster, including a second pair interaction where xi , yij , yij , z ijk , and wijkl are the cluster proba-
as bilities of finding the atomic configurations specified
by the subscript(s) at a point, the nearest neighbour

5 pair, the second-nearest neighbour pair, the triangle,


ΔE = νi ξi . (20.29) and the tetrahedron clusters, respectively, and N is
i=0 the number of lattice points. Minimizing the grand
potential with respect to all the correlation functions
At a finite temperature T the free energy of a phase allows for the Gibbs energy of mixing to be obtained
of interest ΔG can be obtained by adding a config- as a function of composition at a constant tempera-
urational entropy term ΔS to the internal energy as ture, T .
Part E 20.2

follows [20.13, 14]: The compositional variation of the formation en-


thalpy derived by the cluster expansion and the cluster
ΔG = ΔE − T ΔS . (20.30) variation methods is shown in Fig. 20.5 [20.11]. The
solid lines show the results of thermodynamic analyses
The cluster variation method (CVM) with the tetra- described in the following section. The open squares de-
hedron approximation are used to calculate the note the corresponding Gibbs free energy derived by the
configurational entropy. For the bcc structure, the en- method shown in this section.

Gibbs free energy ⌬G (kJ mol) Gibbs free energy ⌬G (kJ mol)
0 0
227°C 727 °C
ab initio calculation
–5 –5

–10 –10

bcc–A2
–15 –15
bcc–A2
–20 –20

bcc–B2
–25 –25
bcc–B2

–30 –30

–35 –35
Fe 0.2 0.4 0.6 0.8 Be Fe 0.2 0.4 0.6 0.8 Be
Molar fraction xB Molar fraction xB

Fig. 20.5 The Gibbs energies of mixing in the α bcc solid solution at 227 ◦ C and 727 ◦ C based on ab initio energetic
calculations
The CALPHAD Method 20.2 Incorporation of the First Principle Calculations into the CALPHAD Approach 1011

20.2.3 Thermodynamic Analysis β


G − 0.50G bcc − 0.50G bcc
of the Gibbs Energies Based Fe:Be Be Fe
on the First Principle Calculations = −37 100 + 9T (J/mol) ,
0 β β
L = 0L = −4T (J/mol) ,
The Gibbs free energy of the bcc phase derived using Be,Fe:Be Be:Be,Fe
the first principle calculations is analyzed according to 0 β β
L = 0L = −380 − 4T (J/mol) .
the two-sublattice model. According to Table 20.4, the Fe:Be,Fe Be,Fe:Fe
most stable ordered structure in the Fe−Be bcc phase
The calculated Gibbs energies of mixing in the α bcc
is recognized as a B2 structure, and hence the Gibbs
solid solution at 227 and 727 ◦ C are drawn in Fig. 20.5
free energy for this simple structure is described in this
using the solid lines for the ordered state (bcc−B2) and
section. The Gibbs energy for one mole of φ phase,
the disordered state (bcc−A2), respectively. The con-
denoted as (Fe, Be)m (Fe, Be)n , is represented by the
vex curvature of the free energy in the vicinity of the
two-sublattice model as described in Sect. 20.1.1 as the
equiatomic composition corresponds to the formation
following equation:
of the B2 structure.
φ φ
G φ = y1 y2 0G + y1 y2 0G
Fe Fe Fe:Fe Fe Be Fe:Be 20.2.4 Construction of Stable
φ φ and Metastable Phase Diagrams
+ y1 y2 0G + y1 y2 0G
Be Fe Be:Fe Be Be Be:Be
  The information on the experimental data on the phase
m boundaries and other thermodynamic quantities are ther-
+ RT y ln y + y ln y
1 1 1 1

Part E 20.2
m +n Fe Fe Be Be modynamically analyzed together with the estimated
 
n 2 metastable quantities of the bcc phase described in the
+ y2 ln y2 + y2 ln y foregoing section.
m +n Fe Fe Be Be
The calculated results of the Fe−Be phase diagram
+ exG φ + G mag . (20.32)
are compared with the experimental data in Fig. 20.6.
The term yis denotes the site fraction of element i in the The shaded area shown in this figure is the metastable
sublattice s. The terms m and n are variables denoting (bcc+B2) two-phase region, which is accompanied by
the size of the sublattice s, and straightforwardly, the the ordering of the bcc structure on formation. The dot-
relationships of m = 0.5 and n = 0.5 hold for the B2 ted line shows the order–disorder transition line, along
structure. 0G i: j denotes the Gibbs energy of a hypothet- which the two-phase field expands into the higher tem-
ical compound i 0.5 j0.5 , and terms relative to the same perature range. The age hardening of this alloy has been
stoichiometry are identical, whatever the occupation of investigated experimentally and the results are summa-
the sublattice. The excess Gibbs energy term, exG φ , con- rized in [20.15], and a brief outline of the ageing process
tains the interaction energy between unlike atoms, and of this alloy is as follows. The disordered bcc structure
is expressed using the following polynomial forms in the initial stage, and consequently, the B2-type
ordered structure separates in the bcc phase. A 100
ex
G φ = y1 y1 y2 L Fe,Be:Fe +y1 y1 y2 L Fe,Be:Be modulated structure with changes in concentration was
Fe Be Fe Fe Be Be observed in some samples using electron microscopy.
+ y2 y2 y1 L Fe:Fe,Be +y2 y2 y1 L Be:Fe,Be , This ordering behavior of the bcc structure is possibly
Fe Be Fe Fe Be Be
explained by the metastable equilibria in Fig. 20.6.
(20.33) It is well known that the solubility of Be in bcc
where L i, j:k (or L i: j,k ) is the interaction parameter be- Fe (α) deviates significantly from the Arrhenius equa-
tween unlike atoms on the same sublattice. The magnetic tion; i. e., a proportional relationship exists between the
contribution to the Gibbs free energy G mag was given logarithm of the solubility and the reciprocal of the tem-
by (20.11). perature. The solubility of Be in the α phase is shown
The thermodynamic parameters obtained by fitting in Fig. 20.7. The solubility would be denoted by the
the ab initio values to (20.32) are shown as follows: broken line if there were neither an order–disorder tran-
sition nor a magnetic transition in the bcc Fe phase.
β
G − 0.50G bcc − 0.50G bcc This is approximated by the straight line following the
Be:Fe Be Fe Arrhenius law for dilute solutions. Thus, the deviation
= −37 100 + 9T (J/mol) , of the solubility from the ideal Arrhenius law is repre-
1012 Part E Modeling and Simulation Methods

Temperature T(°C)
1600
L

1400

1200
δ
α
γ
1000

ζ
800

600
ε

400

bcc + B2
Part E 20.2

200
Fe 10 20 30 40 50 60 70 80 90 Be
Oesterheld Gordon Geles Hammond Oesterheld Be (mol. %)
Wever Teitel Heubner Ko Wever
Fig. 20.6 Calculated Fe−Be phase compared with experimental data

Solubility of Be in α-Fe (mol. %) sented by the shaded areas. The dashed lines show the
102 order–disorder transition temperature and the magnetic
transition temperature of the bcc Fe phase. The solubility
of Be decreases in the lower temperature region below
650 ◦ C, while it increases in the higher temperature re-
Order–disorder transition gion. The decrease in solubility in the lower temperature
region can be explained from the viewpoint of the change
in magnetism [20.16]. On the other hand, the solubility
increases owing to the order–disorder transition of the
101
bcc phase in the higher temperature range. Figure 20.8
shows the change in the order parameter along a solubil-
ity line for the α phase. The order parameter was defined
by
η = y1 − y2 . (20.34)
Be Be

Magnetic The solubility line in Fig. 20.8a intersects the order–


transition
disorder transition line indicated by the dotted line at
100 about 650 ◦ C. As can be seen in Fig. 20.8b, the order
6 7 8 9 10 11 12 13
parameter along the solubility line increases at a higher
Reciprocal temperature 104 / T (K–1)
temperature than 650 ◦ C, yielding the progress of the B2
Fig. 20.7 Solubility of Be in the α phase ordering in the bcc disordered phase.
The CALPHAD Method 20.2 Incorporation of the First Principle Calculations into the CALPHAD Approach 1013

Temperature T(°C) Order parameter


1400 0.6
1190 °C
1200 0.5 1000 °C
γ
α
1000 0.4
830 °C
800 0.3

600 0.2

400 0.1
bcc + B2 650 °C
200 0
Fe 5 10 15 20 25 30 35 Fe 10 20 30 40 50 60
Be (mol. %) Be contenet in α phase (mol. %)

Fig. 20.8 (a) Enlarged Fe-rich portion of the calculated Fe−Be binary phase diagram. The dotted line shows the order–
disorder transition line. (b) Change in the order parameter along the solubility line for the α phase

Part E 20.2
20.2.5 Application to More Complex Cases Temperature T (K)
2000
In the foregoing Sect. 20.2.3, two-sublattice formalism L
was applied to describe the Gibbs energies derived 1800
from the first principle calculations. However, it is
1600
generally true that a more complex thermodynamic for- γ(Co) β
γ(Al)
malism such as a four-sublattice model is appropriate 1400
so as to reflect directly the results of the ab initio
values in phase diagram computations, since several 1200

CoAl3
stoichiometric compositions are selected for superstruc-

Co4Al13
1000
tures, i. e., A3 B–D03 , AB-B2, AB-B32, AB3 –D03 etc. Co2Al5
in the bcc phase. Then in this section, an analysis of 800
Co2Al9
metastable phase separation of bcc phase in the Co−Al ε (hcp)
and Ni−Al binary systems is presented [20.17] using 600
Co 20 40 60 80 Al
the four-sublattice model. The more complex thermo- Al (mol. %)
dynamic modeling brings a detailed knowledge about
the metastable phase equilibria in these binary systems. Fig. 20.9 Outline of the Co−Al binary phase diagram
The Co−Al phase diagram consists of liquid, two fcc
solid solutions γ (Co) and γ (Al), hcp Co(ε), β with CsCl Formation Energies of Superstructures
structure, and the intermetallic phases Co2 Al5 , CoAl3 , for the Co–Al System
Co4 Al13 , and Co2 Al9 , as shown in Fig. 20.9. There ex- The formation energies to the segregation limit for the
ists one eutectic reaction concerning the liquid phase Co−Al system in the D03 , B2, B32 structures in the
and six invariant reactions between the solid phases. The ground state are summarized in Table 20.7, where the
B2 (β) phase has a large homogeneity range at higher corresponding values for the Ni−Al system are included.
temperatures, but the range decreases remarkably in the The Gibbs free energies of the metastable bcc-based
lower temperature region. It has been confirmed experi- phase are evaluated using the effective cluster interac-
mentally that the two-phase region for the A1(γ )/B2(β) tion energies up to the tetrahedron cluster, including
phases extends over a wide composition range with de- the second pair interactions. Using the same procedure
creasing temperature. Owing to a sudden drop in the as in Sect. 20.2.2, the Gibbs energy of mixing can be
homogeneous region, these experimental values were obtained as a function of composition at a constant
debatable. temperature T .
1014 Part E Modeling and Simulation Methods

Table 20.7 Formation energies to the segregation limit for Co−Al and Ni−Al in the D03 , B2, B32, and A2 structures in
the ground state
Alloy system Molar fraction of Al Structure Formation energy (kJ/mol)
Co−Al 0.25 D03 − 12.7
0.5 B2 − 62.3
0.5 B32 − 23.7
0.75 D03 − 18.3
Ni−Al 0.25 D03 − 40.7
0.5 B2 − 69.5
0.5 B32 − 34.0
0.75 D03 − 18.2

Expression of Gibbs Energy of the B2 Phase Calculated Phase Equilibria


Using the Four-Sublattice Model in the Co–Al Binary System
The β phase was described using a four-sublattice The calculated Gibbs free energy of the bcc phases at
model, so as to reflect the results of the ab initio cal- 727 ◦ C was compared with the results from the ab ini-
culations on the phase diagram computation as tio energetic calculations in Fig. 20.10. The calculated
    Co−Al binary phase diagram is compared with the ex-
Al1y1 M y11 Al2y2 M y22 perimental data in Fig. 20.11. The characteristic features
Part E 20.2

Al M 0.25 Al M 0.25 of the binary phase diagram are well reproduced by the
    calculations. The dotted line shows the metastable two-
Al3y3 M y33 Al4y4 M y44 . (20.35)
Al M 0.25 Al M 0.25 phase separated region between the A2 (bcc-Co) and the
B2 (β) phases. The two-phase separation, based on the
The number of each sublattice site is 0.25, and the bcc structure, closely relates the anomaly in the phase
four sites are therefore equivalent. The disordered state boundaries in the Co−Al binary system. This situation is
is described when the site fractions of the different schematically illustrated in Fig. 20.12, where the Gibbs
species are the same in the four sublattices. For the free energies of the B2 (β) and the A1 (γ Co) phases are
ordered structures, if two sublattices have the same drawn. The two-phase separation of the β phase does
site fractions, as do the two others, but are differ-
ent, then the model describes the B2 and B32 phases.
If three sublattices have the same site fractions, and Gibbs free energy ΔG (kJ/mol)
0
are different from the fourth, then D03 ordering is ab initio calculation
described. Thermodynamic analysis
–10
The Gibbs energy of the β phase is expressed by

–20
Gm = yi1 y2j yk3 yl4 0G i: j:k:l
i j k l
–30
RT

4
 
+ yis ln yis –40
4
s=1 i

–50
+ yis ysj ykr ylt ym
u
L i, j:k:l:m ,
s=1 i j>i k l m –60
(20.36)
–70
where 0G i: j:k:l denotes the Gibbs energy of a compound Co 0.2 0.4 0.6 0.8 Al
Molar fraction xAl
i 0.25 j0.25 k0.25 l0.25 , and terms relative to the same stoi-
chiometry are identical, whatever the occupation of the Fig. 20.10 The calculated Gibbs free energy of the bcc
sublattice. L i, j:k:l:m is the interaction parameter between phase in the Co−Al binary system at 727 ◦ C compared
unlike atoms on the same sublattice. with the results from the ab initio energetic calculations
The CALPHAD Method 20.2 Incorporation of the First Principle Calculations into the CALPHAD Approach 1015

Temperature T(K)
2000
Gwyer
L
Fink
1800 Koester
Schramm
1600 Ettenberg
γ(Co) β Takayama

1400

1200
CoAl3
1000
γ(Al)
800
Co4Al13
Co2Al5
600
ε Co2Al9

Part E 20.2
400
Co 20 40 60 80 Al
Al (mol. %)
Fig. 20.11 A comparison of the calculated Co−Al binary phase diagram with previous work

Gibbs free energy Temperature T (K)


2000

γ
G 1800
β
β 1600
G γ(Co)

1400
T = T1 T = T1

Gibbs free energy Al (mol. %) 1200

1000
γ T = T2
G
800
β
G
600
ε
T = T2
400
Al (mol. %) Co 20 40 60 80 Al
Al (mol. %)
Fig. 20.12 The schematic Gibbs free energies of the B2 (β) and A2 (γ Co) phases in the Co−Al system
1016 Part E Modeling and Simulation Methods

not occur at higher temperatures T1 mainly owing to the the B2 structure, the interaction between unlike atoms
effect of the entropy of random mixing on the atomic strengthens around the 50% Al composition, since the
arrangement. On the other hand, the tendency towards degree of order reaches a maximum at the equiatomic
ordering becomes stronger at lower temperatures T2 . In composition. The decrease in the enthalpy term due to
the attractive interaction results in the acute point in the
free energy curve in the vicinity of the 50% Al compos-
Gibbs free energy ΔG (kJ mol) ition, and consequently, a two-phase separation in the
0
β phase forms. This situation yields a significant shift of
–10 ab initio calculation the phase boundary for the (γ (Co) + β)/β composition
Thermodynamic analysis to the equiatomic composition.
–20 The procedure used for the Co−Al binary system
is applied to the Ni−Al binary system. The calculated
–30 Gibbs free energy of the bcc phase is compared with
the results of the ab initio energetic calculations in
–40 Fig. 20.13. The calculated Ni−Al binary phase diagram
is compared with the experimental data in Fig. 20.14.
–50 The phase separation of the A2 and B2 structures can
not be observed in this binary system.
–60
Origin of Phase Separation in the Co–Al
Part E 20.2

–70 and Ni–Al Systems


Ni 0.2 0.4 0.6 0.8 Al The formation energies to the segregation limit for
Molar fraction xAl
Co−Al and Ni−Al in the D03 , B2, B32, and A2 struc-
Fig. 20.13 Calculated Gibbs free energy of the bcc phase tures in the ground state, as listed in Table 20.7, are
in the Ni−Al binary system at 727 ◦ C compared with the plotted versus Al concentration in Fig. 20.15. In com-
results from the ab initio energetic calculations paring the two systems, it can be seen that in the Co−Al

Temperature T(K)
2000
Fink
L Alexander
1800 Phillips
Taylor
Nash
1600
Robertson
γ(Ni) β Hilpert
1400 Verhoeven
γ Jia

1200

1000

800
γ(Al)
Ni5Al3 Ni2Al3 NiAl3
600

400
Ni 20 40 60 80 Al
Al (mol. %)

Fig. 20.14 A comparison of the calculated Ni−Al binary phase diagram with previous work
The CALPHAD Method 20.2 Incorporation of the First Principle Calculations into the CALPHAD Approach 1017

and 50% Al compositions. This energetic analysis sug-


Formation energy ΔE (kJ mol)
10 gests that a two-phase separation of the bcc-Co and B2
bcc-Co bcc-Al (β) phases occurs on the Co rich side in Co−Al system,
0 while such a separation is not realized in the Ni−Al
system.
–10
Co3Al-D03
CoAl3-D03 Figure 20.16 shows the density of states (DOS) of
–20 the B2 (β) structure in the Co−Al and Ni−Al alloy
CoAl-B32 systems. In this figure, E F denotes the Fermi energy,
–30
where no electrons occupy the electronic states above
–40 this energy level. It can be seen that the distribution of the
DOS of the B2 phase in the two systems is similar, with
–50 both showing a low DOS at the Fermi level. However,
–60 when observing the DOS of the D03 phase for these two
systems, as shown in Fig. 20.17, we observe a marked
–70 CoAl-B2 difference between these two systems, in that the Fermi
–80 level is located near the peak of the DOS for the Co3 Al–
Co 0.2 0.4 0.6 0.8 Al D03 phase but decreases in a region with a very low DOS
Molar fraction xAl
in the Ni3 Al–D03 phase. This fact indicates that from
Formation energy ΔE (kJ mol)
10 an energetic point of view, the stable D03 structure of
bcc-Ni bcc-Al the Ni3 Al phase is highly preferred, while in the Co3 Al–

Part E 20.2
0 D03 phase is unstable with respect to separation of the
–10
A2 and B2 phases in the Co−Al system.
NiAl3-D03
–20
Total density of states
–30
8
NiAl-B32 CoAl (B2)
–40
Ni3Al-D03
6
–50

–60 4

–70
2
–80
Ni 0.2 0.4 0.6 0.8 Al 0
Molar fraction xAl –16 –14 –12 –10 –8 –6 –4 –2 0 2 4
EF Energy
Fig. 20.15 Variation of the formation energies to the segre- E (eV)
gation limit for Co−Al and Ni−Al with concentration of Total density of states
12
Al
10 NiAl (B2)
system, the energy of the D03 structure at the 25% Al
8
composition is located slightly above the straight line
connecting the energy values at the 0% Al and 50% Al 6
compositions, which results in a two-phase separation of
4
the A2 (bcc-Co) and B2 (β) structures in Co−Al that is
more stable than the formation of the Co3 Al–D03 struc- 2
ture in the ground state. On the other hand, the energy
plot of the Ni−Al system shows the stabilization of the 0
–16 –14 –12 –10 –8 –6 –4 –2 0 2 4
Ni3 Al–D03 structure compared with the two-phase sep- EF Energy
E (eV)
aration of the A2 (bcc-Ni) and B2 (β) structures, since
the formation energy of the D03 phase is lower than the Fig. 20.16 The density of states of the B2 (β) structure in
straight line connecting the energy values at the 0% Al the Co−Al and Ni−Al alloy systems
1018 Part E Modeling and Simulation Methods

Fig. 20.17 The density of states of the D03 structure in the


Total density of states Co−Al and Ni−Al alloy systems
25
Co3Al (DO3)
This difference in the density of states at the Fermi
20
level might be considered to be the result of the ex-
tra d-electron in Ni versus Co. From the point of view
15
of the rigid band approximation, this difference in the
10
number of electrons for two neighboring elements shifts
the Fermi level towards the higher energy side in the
5 Ni−Al system versus the Co−Al system. This raises
the different relative positions of the Fermi level in the
0 DOS curve, which consequently leads to the different
–20 –15 –10 –5 0 5 structural stabilities.
EF Energy
E (eV) The ground state analysis of the Ni−Al system
Total density of states
35 suggests that the phase separation concerning the D03
Ni3Al (DO3) structure forms at absolute zero. When assuming that
30 such a two-phase separation forms at a finite tempera-
25 ture, however, the agreement between the calculated
phase boundaries and experimental data is insufficient.
20
For example, Fig. 20.18 shows the phase diagram in the
Part E 20.2

15 case that a two-phase separation between the metastable


10 D03 and B2 structures occurs around 500 ◦ C. The ho-
mogeneity range of the B2 single phase is comparatively
5
Fig. 20.18 The calculated Ni−Al phase diagram assuming
0
–20 –15 –10 –5 0 5 that the critical temperature of the two-phase separation
EF Energy
E (eV) between the metastable D03 and B2 structures was around
500 ◦ C

Temperature T (K)
2000
Fink
L Alexander
1800 Phillips
Taylor
Nash
1600
Robertson
Á(Ni) ‚ Hilpert
1400 Verhoeven
Á Jia

1200

1000

800
Ni5Al3
Á(Al)
Ni2Al3 NiAl3
600
DO3 + B2
400
Ni 20 40 60 80 Al
Al (mol. %)
The CALPHAD Method 20.3 Prediction of Thermodynamic Properties of Compound Phases with First Principle Calculations 1019

reduced owing to the influence of the metastable misci- imental data. Therefore, the phase separation of the D03
bility gap with decreasing temperature, and accordingly, structure is not likely in the experimentally observable
the calculated phase boundaries deviate from the exper- temperature range.

20.3 Prediction of Thermodynamic Properties of Compound Phases


with First Principle Calculations
In an analysis of phase equilibria containing compound the formation energy of Fe3 C, by considering the spin
phases, physical properties of metastable structures are polarization, is 8.1 kJ/mol. This result shows the ef-
often required. The necessity appears clearly for instance fect of the ferromagnetism of the Fe3 C phase in the
in the following case. Figure 20.19 shows the isother- lower temperature region. Furthermore, because the for-
mal phase diagram for the Fe−Cr−C ternary system. mation energy from bcc-Fe and graphite is positive,
In this ternary system, several types of carbides form the Fe3 C structure is less stable than graphite at ab-
in which some amount of alloying element is soluble. solute zero. The calculated formation energies for the
If we consider the cementite, Cr substitutes more than Cr7 C3 and Cr23 C6 phases show a reasonable agreement
10% of Fe, and it forms the ternary line compound. In with the thermodynamic data reported in the litera-
such a case, the Gibbs energy of the cementite phase ture [20.19]. From consideration of the data shown in
is usually described by using sublattice model as the Table 20.8, the thermodynamic properties for metal-
(Fe, Cr)3 C formula. Then if we want to evaluate the lic carbides evaluated by the first principle calculations

Part E 20.3
thermodynamic function for this phase, we need the can be applied to the general procedures used in the
formation energy of Cr3 C, which is metastable in the CALPHAD method.
Cr−C binary system. In the procedure of CALPHAD
approach, this parameter is usually determined on the 20.3.1 Thermodynamic Analysis
basis of the experimental data in the Fe-rich side. How- of the Fe–Al–C System
ever, it could be easily understood that this technique
follows large amount of errors in estimation. Applying The Perovskite carbide in this ternary system, Fe3 AlC
the first principle calculation may possibly solve this dif- (κ), is an fcc-based ordered phase with an E21 -type
ficulty. Thus in the present section, some examples for
application to predicting thermodynamic properties of
C
compound phases and phase diagram calculations will
be illustrated.
90
Carbides and nitrides play a key role in the micro-
structure control of steels, due to a fine dispersion of 80
these precipitates. The effectiveness of the first principle
calculations to the analysis of thermodynamic proper- C (at. %) 70
ties of these compounds might be an interesting issue.
60
A comparison of the calculated formation energies with
the experimental values is attempted for some typic- 50
al carbides observed in steels to clarify the validity
of the FLAPW method. Table 20.8 [20.18] shows the 40
M7C3 + C
φ
formation energy, ΔE form , defined by averaging the to- 30 M7C3
tal energy of the constituent elements with chemical Fe3C
Cr3C?
composition up to the segregation limit, as follows 20 γ + M7C3 M23C6
  C 10
φ φ φ M φ α + M23C6
ΔE form = E tot − x M E tot − 1 − x M E tot , (20.37)
γ
where φ denotes the type of carbide, and M and C rep- Fe 20 40 60 80 Cr
resent a metallic element and graphite, respectively. For Cr (at. %)
example, the formation energy of Fe3 C in the para- Fig. 20.19 The isothermal phase diagram for the Fe−Cr−C ternary
magnetic state is calculated to be 17.9 kJ/mol, while system
1020 Part E Modeling and Simulation Methods

Table 20.8 Comparison of the calculated formation energies with the experimental values for some typical steel carbides
Carbides Space Calculated lattice Observed lattice Calculated formation Observed formation
group parameter (nm) parameter (nm) enthalpy in the ground enthalpy at 25 ◦ C
state (kJ/mol) (kJ/mol)
Fe3 C Pnma a = 0.4871 − + 17.9 −
(paramagnetic) b = 0.6455 − −
c = 0.4330 − −
Fe3 C Pnma a = 0.5018 a = 0.5078 + 8.1 + 6.3
(ferromagnetic) b = 0.6650 b = 0.67297
c = 0.4460 c = 0.45144
Cr7 C3 Pnma a = 0.4373 a = 0.4526 − 19.8 − 22.8
b = 0.6772 b = 0.7010
c = 1.1730 c = 1.2142
Cr23 C6 Fm-3m a = 1.0475 a = 1.06595 − 21.8 − 19.7

structure. The application of this material as a heat a) Al


resistant alloy from the formation of a coherent fine Fe
microstructure consisting of an fcc solid solution and
Part E 20.3

the κ phase is attracting great attention [20.20]. Regard- Fe3Al-L12


less of such a promising potential for this new material,
little information on the thermodynamic properties of
this carbide phase is known. Thus, an attempt to calcu-
late the full phase equilibria of the Fe−Al−C ternary
system is made, introducing the first principle val-
ues for Fe3 AlC into a CALPHAD-type thermodynamic
analysis.
In the Fe3 AlC structure, the Fe and Al atoms are ar- b) Al
ranged in an L12 -type superlattice, in which the C atoms Fe
occupy interstitial sites, resulting in an E21 superstruc-
ture. Figures 20.20a, 20.20b show schematic diagrams
of the Fe3 Al–L12 and Fe3 AlC–E21 structures, respec-
tively. If the C atom occupies only the body-centered
Fe3AlC-E21 C
sites, then it is surrounded by six Fe atoms, and this is
preferable from an energetic point of view. Occupation
of the other fcc interstitial sites enhances the tetragonal-
ity of the L12 crystal structure, yielding a larger strain
energy. The only difference between these two structures
is the existence of C atoms in the octahedral interstitial Fig. 20.20a,b Crystal structures for (a) Fe3 Al–L12 and
sites. Therefore, the Gibbs free energies of these two (b) Fe3 AlC–E21
ordered structures should be described by the same ther-
modynamic model. The two-sublattice model denoted is applied to the κ phase. The Gibbs free energy for the
by the formula, κ phase is calculated using (20.38):
   
Fe y(1) Al y(1) Fe y(2) Al y(2) ,
Gκ = y
(1) (2) (3) 0 κ
+y y y 0G κ
3 1 (1) (2) (3)
Fe Al Fe Al y y G
Al Al C Al:Al:C Al Al Va Al:Al:Va
was generally applied to the L12 structure, and the three-
(1) (2) (3) 0 κ
+y y y 0G κ
(1) (2) (3)
sublattice model, +y y y G
      Al Fe C Al:Fe:C Al Fe Va Al:Fe:Va
Fe y(1) Al y(1) Fe y(2) Al y(2) C y(3) Va y(3) , +y
(1) (2) (3) 0 κ
y y G +y y y G κ
(1) (2) (3)
Fe Al 3 Fe Al 1 C Va 1 Fe Al C Fe:Al:C Fe Al Va Fe:Al;Va
The CALPHAD Method 20.3 Prediction of Thermodynamic Properties of Compound Phases with First Principle Calculations 1021

(1) (2) (3) 0 κ


+y y y 0G κ
(1) (2) (3) Calculation of the Thermodynamic Properties
+y y y G
Fe Fe C Fe:Fe:C Fe Fe Va Fe:Fe:Va of the κ Phase
 The thermodynamic parameters required by the model in
(1) (1) (1) (1)
+ 3RT y ln y + y ln y the (Fe, Al)3 (Fe, Al)1 (C, Va)1 form are evaluated using
Al Al Fe Fe
 first principle calculations and the results are listed in
(2) (2) (2) (2) Table 20.9. The calculated values denote the formation
+ RT y ln y + y ln y
Al Al Fe Fe enthalpies based on the stable structure of the pure elem-
 ent in the ground state. The ferromagnetic state of the
+ RT
(3)
y
(3) (3)
ln y + y ln y
(3) κ phase is calculated to be almost 10 kJ/mol more stable
C C Va Va than its paramagnetic state, and the magnetic moment
+y
(1) (1) (2) (3) κ
y y y L of the phase was calculated to be 3.05 μB . Besides these
Al Fe Al C Al,Fe:Al:C cohesive energies for the stoichiometric components,
(1) (1) (2) (3) κ the interactions between atoms on the same sublattice
+y y y y L are defined in the same way as for the thermodynamic
Al Fe Al Va Al,Fe:Fe:Va
parameters, and these interaction parameters, as well as
(1) (1) (2) (3) κ
+y y y y L the entropy term of the formation energy, are estimated
Al Fe Fe C Al,Fe:Fe:C
using the experimental phase boundaries.
(1) (1) (2) (3) κ
+y y y y L
Al Fe Fe Va Al,Fe:Fe:Va Electronic Structure and Phase Stability
of the κ Phase
+ y y y y Lκ
(1) (2) (2) (3)

Part E 20.3
Al Al Fe C Al:Al,Fe:C As shown in Fig. 20.20, a crystallographic similarity ex-
(1) (2) (2) (3) κ
ists between the Perovskite carbide κ and the Fe3 Al–L12
+y y y y L structure, i. e., a C atom is placed in the center of an
Al Al Fe Va Al:Al,Fe:Va
octahedron composed of six Fe atoms occupying the
(1) (2) (2) (3) κ
+y y y y L face-centered positions in the L12 structure. The actual
Fe Al Fe C Fe:Al,Fe:C
calculated equilibrium lattice constant of the Fe3 Al–
+y
(1) (2) (2) (3) κ
y y y L L12 structure is 0.3502 nm, while that of the κ phase is
Fe Al Fe C Fe:Al,Fe:Va 0.3677 nm, which correspond well with the experimen-
(1) (2) (3) (3) κ tal results of Palm and Inden [20.21]. This fact implies
+y y y y L
Al Al C Va Al:Al:C,Va that occupation by the C atoms in the octahedral in-
(1) (2) (3) (3) κ
terstitial sites causes an expansion of the L12 lattice.
+y y y y L Since the calculated enthalpy of formation of the κ phase
Al Fe C Va Al:Fe:C,Va
( − 27.9 kJ/mol of atoms) is much lower than that of the
(1) (2) (3) (3) κ
+y y y y L Fe3 Al–L12 structure ( − 8.8 kJ/mol of atoms), it is con-
Fe Al C Va Fe:Al:C,Va
cluded that the interstitial C atoms enhance the stability
(1) (2) (3) (3) κ of the L12 structure. Thus, the role of the C atoms will
+y y y y L . (20.38)
Fe Fe C Va Fe:Fe:C,Va be discussed in the context of their electronic structure.

Table 20.9 Calculated thermodynamic parameters required by the (Fe, Al)3 (Fe, Al)1 (C, Va)1 -type three-sublattice model
Structure Strukturbericht Magnetism Calculated lattice Observed lattice Calculated formation
symbol parameter (nm) parameter (nm) enthalpy in the ground state
(kJ/mol of compound)
Fe3 AlC E21 Paramagnetic 0.3677 − − 128.5
Ferromagnetic 0.3677 0.3781 − 139.5
Al3 FeC E21 Paramagnetic 0.3890 − + 190.5
Fe3 Al L12 Paramagnetic 0.3502 − − 35.2
FeAl3 L12 Paramagnetic 0.3734 − − 67.6
Fe4 C − Paramagnetic 0.3645 − + 88.4
Al4 C − Paramagnetic 0.4057 − + 160.5
1022 Part E Modeling and Simulation Methods

Figures 20.21 and 20.22 show the total density of these diagrams, we noticed a marked difference between
states (total DOS) and the angular-momentum-resolved these two structures, in that the Fermi level is located
density of state for each element (p-DOS) for the near the peak of the total DOS for the Fe3 Al–L12 struc-
Fe3 AlC–E21 (κ) and Fe3 Al–L12 structures, respectively. ture, but decreases in a region with a very low DOS
The term E F denotes the Fermi energy, and no elec- in the Fe3 AlC–E21 structure. This fact indicates that
trons occupy electronic states above this energy level. In from an energetic point of view, the stable E21 structure
both structures, the DOS mainly consist of the contribu- of the κ phase is highly preferred, compared with the
tion from the Fe d-electrons. However, while observing Fe3 Al–L12 structure.

a) Total DOS Fe3AlC-E21 a) Total DOS Fe3Al-L12


14 14
12 12
10 10
8 8
6 6
4 4
2 2
Part E 20.3

0 0
–15 –10 –5 0 5 –15 –10 –5 0 5
(eV) (eV)
b) p-DOS (Fe) Fe3AlC-E21 b) p-DOS (Fe) Fe3AlC-E21

d-xy 12 d-xy
4
10
3 8

2 6
4
1
2

0 0
–15 –10 –5 0 5 –15 –10 –5 0 5
(eV) (eV)
c) p-DOS (Al) Fe3AlC-E21 c) p-DOS (Al) Fe3AlC-E21
0.16
s s
p p
0.3
0.12
EF
EF
0.08 0.2

0.04 0.1

0 0
–15 –10 –5 0 5 –15 –10 –5 0 5
(eV) (eV)
Fig. 20.21 (a) Total density of states, and angular- Fig. 20.22 (a) Total density of states, and angular-
momentum-resolved density of states for: (b) Fe and (c) Al momentum-resolved density of states for: (b) Fe and (c) Al
for the Fe3 AlC–E21 (κ) structure for the Fe3 Al–L12 structure
The CALPHAD Method 20.3 Prediction of Thermodynamic Properties of Compound Phases with First Principle Calculations 1023

Figures 20.23a and 20.23b show the calculated tour plots, it can be seen that bonding between the Fe
electron charge density plots of the Fe3 AlC–E21 (κ) and C atoms occurs in the Fe3 AlC–E21 structure, since
and Fe3 Al–L12 structures in the (00 12 ) plane, where a finite charge density between these atoms can be ob-
the contour lines correspond to an electron density of served. This interaction between the atoms enhances the
100 e/nm3 . The Fe atoms can be seen in the middle of energetic stability of the Fe3 AlC–E21 structure.
the horizontal and vertical axes, while the C atoms are
located in the center of the Fig. 20.23a. From these con- Comparison of the Calculated Phase Equilibria
with the Experimental Data
The calculated Fe−Al−C ternary phase diagrams are
a) Fe3AlC-E21 shown in Fig. 20.24 for temperatures, T = 800, 1000,
and 1200 ◦ C. The enlarged portion of the isothermal sec-
Fe
tion diagrams is compared with the experimental phase
boundaries determined using X-ray diffraction and
metallographic observation [20.21]. From EPMA meas-
urements, they reported that the chemical composition
of the κ phase shifted remarkably from its stoichiomet-
ric composition. The homogeneity range of the κ phase
extends from Fe2.9 Al1.1 C0.7 to Fe2.8 Al1.2 C0.7 according
Fe C Fe to the calculations.

Part E 20.3
20.3.2 Thermodynamic Analysis
of the Co–Al–C and Ni–Al–C Systems

The microstructures of Ni-based superalloys contain the


Ni3 Al–L12 phase, which shows an anomalous flow–
stress dependence on temperature [20.22]. In addition,
Fe
because they exhibit very high melting temperatures and
have good resistance to oxidation, alloys with complex
b) Fe3Al-L12 phase structures containing the NiAl–B2, fcc-Ni, and
Ni3 Al–L12 phases have been investigated for techno-
Fe logical applications. On the other hand, it is difficult to
produce Co−Al-based superalloys with a microstructure
consisting of fcc-Ni and Ni3 Al–L12 , because of the ab-
sence of a stable strengthening L12 phase in the Co−Al
binary system. However, the addition of carbon to this
alloy stabilizes the formation of the Perovskite type car-
bide (E21 ) with the composition, M3 AlC (κ phase), as
seen in Fig. 20.18, and this carbide is anticipated to form
Fe Fe
a fine coherent microstructure in a Co-based solid solu-
tion. Then the entire phase equilibria of the Co−Al−C
ternary system is attempted to be clarified by coupling
the CALPHAD and ab initio calculations. The same pro-
cedure is applied to the Ni − −Al − −C system, and the
results are compared to the Co−Al−C system.
Fe
Calculation of the Co–Al–C Phase Diagram
The same thermodynamic model as (Fe, Al)3 (Fe, Al)1
(C, Va)1 is applied to the Co−Al−C and Ni−Al−C
Fig. 20.23a,b Calculated electron charge density plots of: based κ phase, and the parameters necessary for this
(a) the Fe3 AlC–E21 (κ) and (b) the Fe3 Al–L12 structures model are evaluated using first principle calculations,
in the (001/2) plane as listed in Table 20.10. The calculated Co−Al−C
1024 Part E Modeling and Simulation Methods

a) 800 °C C Palm, Inden b) 1000 °C C Palm, Inden


γ + (C) γ γ+κ
90 β + (C) 90 γ + (C) γ +β
80 γ +β 80 β + (C) β+κ
C content β +κ C content
γ + κ + (C)
(mol. %) 70 (mol. %) 70
γ + κ + (C) β + κ + (C)
60 β + κ + (C) 60 γ + β +κ
50 50
Al4C3 Al4C3
40 40
30 30
20 20
κ κ
10 10
γ γ
Fe 20 β 40 60 80 L Al Fe 20 β 40 60 80 L Al
FeAl2 Fe2Al5 Fe4Al13 FeAl2 Fe2Al5 Fe4Al13
Al content (mol. %)

20 κ 20 κ
Part E 20.3

mol. % C
mol. % C
10
10
γ γ

Fe 20 β 40 Fe 20 β 40 mol. % Al
mol. % Al

c) 1200 °C C Palm, Inden


γ γ+κ
90 γ + (C) γ +β
80 β + (C) β+κ
C content
(mol. %) 70 γ + κ + (C)
β + κ + (C)
60
γ + β +κ
50
Al4C3
40
30

L 20
κ
10
γ
Fe 20 β 40 60 80 L Al
Fe4Al5 Al content (mol. %)

20 κ
mol. % C
10
γ
Fig. 20.24a–c Isothermal section diagram calculated
at: (a) 800 ◦ C, (b) 1000 ◦ C, and (c) 1200 ◦ C with an Fe 20 β 40 mol. % Al
enlarged portion on the Fe-rich side
The CALPHAD Method 20.3 Prediction of Thermodynamic Properties of Compound Phases with First Principle Calculations 1025

Table 20.10 The calculated thermodynamic parameters required by the (M, Al)3 (M, Al)1 (C, Va)1 -type three sublattice
model
Structure Strukturbericht Calculated lattice Observed lattice Calculated formation
symbol parameter (nm) parameter (nm) enthalpy in the ground state
(kJ/mol of compound)
Co3 AlC E21 0.3675 0.3700 − 179.0
Al3 CoC E21 0.3909 − + 113.5
Co3 Al L12 0.3515 − − 79.6
CoAl3 L12 0.3726 − − 90.4
Co4 C − 0.3621 − + 98.0
Al4 C − 0.4057 − + 160.5
Ni3 AlC E21 0.3713 − 141.0
Al3 NiC E21 0.3905 + 64.5
Ni3 Al L12 0.3504 0.3572 − 169.5
NiAl3 L12 0.3774 − 91.6
Ni4 C − 0.3645 + 61.0

ternary phase diagrams are shown in Fig. 20.25 for Phase Separation of the κ Phase
temperatures of 900 ◦ C, 1100 ◦ C, and 1300 ◦ C. The in the Co–Ni–Al–C Quaternary System

Part E 20.3
enlarged portion of the isothermal section diagrams According to the first principle calculations, there was
at 1100 ◦ C was compared with the experimental a large difference in the phase stability between the
data [20.21]. Co3 Al–L12 and Ni3 Al–L12 phases, compared to the
The κ phase only appears at the stoichiometric com- difference between the Ni3 AlC–E21 and Co3 AlC–E21
position in our results, while a small homogeneity range phases. In such an energetic situation, a two-phase sep-
is exhibited by this phase in the experimental phase dia- aration should occur, depending on the difference in
grams. This aspect is closely related to the large, positive the formation energies of the compounds, given by the
formation energies in the metastable ordered structures, following expression
such as Al3 CoC and Al4 C. As the formation energy of
Co3 AlC shows an extremely large negative value when ΔG = 0G Ni + 0G Co − 0G Co − 0G Ni .
3 Al 3 AlC 3 Al 3 AlC
compared to these structures, the κ phase only forms at (20.39)
the stoichiometric position. The homogeneity can be ex-
pressed by introducing interaction parameters between This miscibility gap originates in the energy differ-
unlike atoms in the same sublattice. However, this treat- ence between the terminal compounds, and as the
ment is not applied in the present analysis, because the absolute value of ΔG increases, then the critical
experimental phase boundaries of the κ phase are still temperature of the miscibility gap increases. Such
uncertain. a phase separation is often observed in some com-
The calculated vertical section diagrams at constant plex carbonitrides or alloy semiconductor systems.
10 mol % C and 30 mol % Al are shown in Fig. 20.26. Figure 20.28 shows the calculation of a miscibility gap
The calculated values agree well with the experimen- in the Co3 AlC−Ni3 AlC−Co3 Al−Ni3 Al pseudoquater-
tal results, and hence, the new type of approach based nary system at T = 1000 ◦ C. In this model calculation,
on the incorporation of the CALPHAD method into ab only the formation enthalpies of the stoichiometric com-
initio calculations has proven to be applicable to phase pounds are used, designated by the four vertices of the
diagram calculations for higher-order systems. composition square. One can see that a two-phase sep-
aration forms in the direction of the Co3 AlC and Ni3 Al
Calculation of the Ni–Al–C Phase Diagram diagonal. Therefore, a two-phase separation between
In the Ni−Al−C system, on the other hand, it has been these terminal compounds will be involved in the phase
experimentally verified that the κ phase does not appear equilibria of the quaternary system.
in the vicinity of the stoichiometric composition. Fig- Using the thermodynamic description of the four
ures 20.27a–c show the calculated isothermal section ternary systems that comprise the Co−Ni−Al−C qua-
diagrams at 900, 1000, and 1300 ◦ C, respectively. ternary system, a vertical section diagram is calculated
1026 Part E Modeling and Simulation Methods

a) 900 °C C b) 1100 °C C Kimura β + κ + (C)


et al.
γ+κ
90 90
β+κ+γ
80
C content 80 C content
β + (C)
(mol. %) (mol. %) 70
70 β +κ
60
60
50
Al4C3
50 40
Al 3C3
4

40 30
20 κ
30
10
20 κ

10 Co 20 40 60 80 L Al
γ β Co2A5 Co4Al13
CoAl3 Al content
Co 20 40 60 80 L Al (mol. %)
Part E 20.3

γ(Co) β Co2Al5 Co4Al13 Co2Al9


CoAl3
C (mol. %) 20 κ
Al content (mol. %)
c) 1300 °C C
10

90
Co γ(Co) 20 β 40 Al
80
C content (mol. %) Al (mol. %)
70

60

50
Al4C3
40

30

20 κ
L
10

Co Fig. 20.25a–c Isothermal section diagram of the Co−


20 β 40 60 80 L Al
γ(Co) Al−C system calculated at: (a) 900 ◦ C, (b) 1100 ◦ C, and
Al content (mol. %)
(c) 1300 ◦ C

at a constant C content of 4 mol %, and an Al content The κ phase gradually changes its character from the
of 23 mol %, as shown in Fig. 20.29. In this figure, the Co3 AlC-based carbide to the Ni3 Al-based intermetal-
term κ1 denotes the Ni-based L12 structure, while the lic compound with increasing Ni content in the alloy.
Co-based Perovskite structure is represented by κ2 . At The precipitates decompose into almost stoichiomet-
higher temperatures, the E21 structure forms a homo- ric Co3 AlC and Ni3 Al phases in the lower temperature
geneous solution, and this phase is designated as κ. range.
The CALPHAD Method 20.3 Prediction of Thermodynamic Properties of Compound Phases with First Principle Calculations 1027

a)
Temperature T(°C)
1700
Kimura et al. Co-Al-10 mol % C
1600 L + (C)

1500 L
L + β+ (C)
1400
L+γ
1300 β+ (C)
L + γ + (C)
L +κ L+ β+κ
1200
γ + (C) β + κ + (C)
1100
γ + (C) + κ γ + β +κ β+κ
1000

900
Co 5 10 15 20 25 30 35 40 45

Part E 20.3
90 mol. % Al content (mol. %)

b)
Temperature T (°C)
1700
Kimura et al. Co-30 mol % Al-C
1600
L L + (C)
1500
L + β + (C)
1400
L+β
L+ β +κ
1300
β + κ + (C)
1200
β +κ
1100
γ +β+κ
1000

900
Co 5 10 15
70 mol. % C content (mol. %)
Fig. 20.26a,b Calculated vertical section diagrams at constant (a) 10 mol % C and (b) 30 mol % Al
1028 Part E Modeling and Simulation Methods

a) 900 °C C c) 1300 °C C

90 90

80 80
C content (mol. %) C content (mol. %)
70 70

60 60

50 50
Al4C3 Al4C3
40 40

30 30

20 20

10 10

Ni γ 20 κ1 40 β 60 Ni2Al3 80 L AlNi γ 20 κ1 40 β 60 80 L Al
Part E 20.3

Al content (mol. %) Al content (mol. %)

b) 1000 °C C

90

80
C content (mol. %)
70

60

50
Al4C3
40

30

20

10

Ni γ 20 κ1 40 β 60 Ni2Al3 80 L Al
Al content (mol. %)
Fig. 20.27a–c Isothermal section diagram of the Ni−Al−C system calculated at (a) 900 ◦ C, (b) 1000 ◦ C, and (c) 1300 ◦ C
The CALPHAD Method 20.3 Prediction of Thermodynamic Properties of Compound Phases with First Principle Calculations 1029

Fig. 20.28 Calculated miscibility gap in the Co3 Al


C content (mole fraction)
Co3AlC Ni3AlC C−Ni3 AlC−Co3 Al−Ni3 Al pseudoquaternary system at
0.20 1000 ◦ C
0.18

0.16

0.14

0.12

0.10

0.08

0.06

0.04

0.02

0
0 0.2 0.4 0.6 0.8 1.0

Part E 20.3
Co3Al Ni3Al
Ni content (mole fraction)

Temperature T (°C)
1800

1600
L
L +κ
1400
β+L
κ+β +L
1200
κ
1000
γ+κ+ β γ+κ
800

600
γ + κ1 + κ2 + β
400
γ + κ1 + κ2
200

0
73 mol. % Co 10 20 30 40 50 60 70
Ni content (mol. %)
Fig. 20.29 Vertical section diagram calculated at a constant C content of 4 mol % and an Al content of 23 mol %
1030 Part E Modeling and Simulation Methods

References

20.1 N. Saunders, A. P. Miodownik: CALPHAD (Calcula- 20.13 M. H. F. Sluiter, Y. Watanabe, D. de Fontaine,


tion of Phase Diagrams): A Comprehensive Guide Y. Kawazoe: First-principles calculation of the
(Elsevier, Oxford 1998) pressure dependence of phase equilibria in
20.2 F. R. de Boer, R. Boom, W. C. M. Mattens, the Al−Li system, Phys. Rev. B 53, 6137–6151
A. R. Miedema, A. K. Niessen: Cohesion in Metals, (1996)
Transition Metals Alloys (North-Holland, Amster- 20.14 T. Mohri, Y. Chen: First-principles investigation of
dam 1988) L10 -disorder phase equilibrium in the Fe−Pt sys-
20.3 G. Ghosh, C. Kantner, G.B. Olson: Thermodynamic tem, Mater. Trans. 43, 2104–2109 (2002)
modeling of the Pd−X (X = Ag, Co, Fe, Ni) systems, 20.15 H. Ino: A pairwise interaction model for decom-
J. Phase Equil. 20, 295–308 (1999) position and ordering processes in b.c.c binary
20.4 M. Hillert, L.-I. Staffansson: The regular solution alloys and its application to the Fe−Be system, Acta
model for stoichiometric phases and ionic melts, Metall. 26, 827–834 (1978)
Acta Chem. Scand. 24, 3618–3626 (1970) 20.16 T. Takayama, M. Y. Wey, T. Nishizawa: Effect of
20.5 W. Oelsen, F. Johannsen, A. Podgornik: Erzmetall magnetic transition on the solubility of alloying
9, 459–469 (1956) elements in bcc iron and fcc cobalt, Trans. Japan
20.6 W. Kohn, L. J. Sham: Self-consistent equations in- Inst. Met. 22, 315–325 (1981)
cluding exchange and correlation effects, Phys. 20.17 H. Ohtani, Y. Chen, M. Hasebe: Phase separation
Rev. 140, A1133–1138 (1965) of the B2 structure accompanied by an ordering in
20.7 P. Blaha, K. Schwarz, G. K. H. Madsen, D. Kvasnicka, Co−Al and Ni−Al binary systems, Mater. Trans. 45,
J. Luiz: WIEN2k, An Augmented Plane Wave and 1489–1498 (2004)
Part E 20

Orbitals Program for Calculating Crystal Properties 20.18 H. Ohtani, M. Yamano, M. Hasebe: Thermodynamic
(Karlheinz Schwarz, Tech. Universität Wien, Vienna analysis of the Fe−Al−C ternary system by in-
2001) corporating ab initio energetic calculations into
20.8 J. P. Perdew, K. Burke, Y. Wang: Generalized gra- the CALPHAD approach, ISIJ Intern. 44, 1738–1747
dient approximation for the exchange-correlation (2004)
hole of a many-electron system, Phys. Rev. B 54, 20.19 R. Hultgren, P. D. Desai, D. T. Hawkins, M. Gleiser,
16533–16539 (1996) K. K. Kelley: Selected Values of the Thermodynamic
20.9 H. Ohtani, M. Hasebe: Thermodynamic analysis of Properties of Binary Alloys (ASM, Metals Park 1973)
phase diagrams by incorporating ab initio ener- 20.20 Y. Kimura, M. Takahashi, S. Miura, T. Suzuki,
getic calculations into CALPHAD approach, Bull. Iron Y. Mishima: Phase stability and relations of multi-
Steel Inst. Japan 9, 223–229 (2004) phase alloys based on B2 CoAl and E21 Co3 AlC,
20.10 J. W. D. Connolly, A. R. Williams: Density-functional Intermet. 3, 413–425 (1995)
theory applied to phase transformations in 20.21 M. Palm, G. Inden: Experimental determination of
transition-metal alloys, Phys. Rev. B 27, 5169–5172 phase equilibria in the Fe−Al−C system, Intermet.
(1983) 3, 443–454 (1995)
20.11 H. Ohtani, Y. Takeshita, M. Hasebe: Effect of the 20.22 H. Ohtani, M. Yamano, M. Hasebe: Thermodynamic
order–disorder transition of the bcc structure on analysis of the Co−Al−C and Ni−Al−C systems by
the solubility of Be in the Fe−Be binary system, incorporating ab initio energetic calculations into
Mater. Trans. 45, 1499–1506 (2004) the CALPHAD approach, Comp. Coupling Phase Diag.
20.12 F. D. Murnaghan: The compressibility of media un- Thermochem. 28, 177–190 (2004)
der extreme pressures, Proc. Nat. Acad. Sci. USA 30,
244–247 (1944)

You might also like