You are on page 1of 10

CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

Contents lists available at SciVerse ScienceDirect

CALPHAD: Computer Coupling of Phase Diagrams and


Thermochemistry
journal homepage: www.elsevier.com/locate/calphad

Atomistic modeling of pure Co and CoAl system


Wei-Ping Dong a,1, Hyun-Kyu Kim a, Won-Seok Ko a, Byeong-Moon Lee a, Byeong-Joo Lee a,b,n
a
b

Department of Materials Science and Engineering, Pohang University of Science and Technology (POSTECH), Pohang 790-784, Republic of Korea
Division of Advanced Nuclear Engineering, Pohang University of Science and Technology (POSTECH), Pohang 790-784, Republic of Korea

a r t i c l e i n f o

abstract

Article history:
Received 30 November 2011
Received in revised form
22 March 2012
Accepted 6 April 2012
Available online 8 May 2012

Interatomic potentials for pure Co and the CoAl binary system have been developed based on the
second nearest-neighbor modied embedded-atom method (2NN MEAM) potential formalism. The
potentials can describe various fundamental physical properties of the relevant materials in good
agreement with experimental information. The potential is utilized to an atomistic computation of
interfacial properties between fcc-Co (g) and Co3Al (g0 ) phases. It is found that the anisotropy in the g/g0
interfacial energy is relatively small and leaves a room for further modication by alloying other
elements. The applicability of the atomistic approach to an elaborate alloy design of advanced Co-based
superalloys through the investigation of the effect of alloying elements on interfacial and elastic
properties is discussed.
& 2012 Elsevier Ltd. All rights reserved.

Keywords:
Modied embedded-atom method
Atomistic simulation
Co
CoAl
Interfacial energy

1. Introduction
The development of superalloys has been driven by the demand
to increase the operating temperature of gas turbines serving in
power plants and aircraft engines. Nowadays, various classes of
superalloys are widely used including Fe-based, Co-based and Nibased superalloys, among which the Ni-based superalloys strengthened with the L12 compound (g0 phase) have been regarded as those
with the highest heat resistance. Even though the melting point of
the face-centered cubic (fcc) cobalt, the high temperature structure, is
higher than nickel by 40 1C, the traditional Co-based superalloys
exhibit a lower strength at high temperature than Ni-based superalloys. Recently however, Sato et al. [1] found a Co-based superalloy
with outstanding high-temperature strength. Similar to Ni-based
superalloys, the coexistence of g-Co (disordered fcc structure) solidsolution phase and the g0 -Co3(Al,W) (L12 structure) phase, and the
small difference in the lattice parameter between them resulted in a
microstructure of regularly aligned, coherent cuboidal g0 precipitates
on a sub-micrometer scale. The superior strength of the g/g0 superalloy at elevated temperatures is generally attributed to a high
volume fraction of g0 precipitates that inhibit the dislocation glide
in g channels.
Sato et al.0 s work [1] aroused great attention to the study of the
Co-based superalloys with the two coherent phases, g and g0 .
Many studies have been carried out on microstructures [2,3],
n

Corresponding author. Tel.: 82 54 2792157; fax: 82 54 2792399.


E-mail address: calphad@postech.ac.kr (B.-J. Lee).
1
On leave from State Key Laboratory of Solidication Processing
Northwestern Polytechnical University (NPU), Xian 710072, China.
0364-5916/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.calphad.2012.04.001

mechanical properties [48], phase equilibria [9], structural stability and elastic properties [10], and on the effect of a replacement of
W by other elements, for example, Mo, resulting in the precipitation
of g0 -Co3(Al,Mo) [11]. Most materials processes (annealing, hot
rolling, extrusion, forging, super plastic deformation, etc.) modify
the grain structure and the distribution of interfaces, especially in
high temperature superalloys. It should be emphasized here that
the microstructure evolution is strongly affected by the interfacial
properties during recrystallization and grain growth. Therefore,
information on the g/g0 interfaces in Co-based superalloys, particularly their structure, energy and dynamics behavior, is highly
required to gain a better understanding of the strengthening effect.
The most fundamental property would be the interfacial energy and
its anisotropy that strongly affect the size of the critical nuclei, the
nucleation rate and the shape of g0 precipitates. A probable solute
segregation on the g/g0 interface, which may signicantly inuence
coarsening and precipitatedislocation interactions, would be also
of interest. However, all those interfacial energy and solute segregation are quantities hard to measure experimentally. This difculty in experimental works for interfacial properties presents an
opportunity for atomistic simulations that provide information on
atomic scale structural evolution. One can calculate the interfacial
energy and solute segregation rather easily using atomistic simulations such as molecular dynamics (MD), molecular statics (MS) or
Monte Carlo (MC). The atomistic simulations can also be used to
search for candidate alloying elements that modify the g/g0 interfacial energy, mist strain or change the volume fraction of g0 for an
improvement of mechanical properties.
The most accurate way of atomistic simulations for the investigation of atomic level materials properties would be to use

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

rst-principles. However, rst-principles calculations are not


appropriate to investigate large-scale materials behavior because
of its size (or number of atoms) limit. An alternative can be to use
(semi-)empirical interatomic potentials, which can handle more
than a million atoms. In this case, accurate interatomic potentials
which can describe various fundamental physical properties (structural properties, elastic properties, thermal properties, etc.) of
relevant materials are important. Because of the importance of
(semi-)empirical atomistic simulation approaches, many studies
had been performed to develop an interatomic potential for pure
Co: the embedded atom method (EAM) potential [12,13], modied
EAM potential [14] and modied analytic EAM potential [15,16]. All
those potentials, except one EAM [13] which includes no calculation result to compare with, reproduce fundamental physical
properties (at 0 K) of pure Co reasonably well, even though it is
not known how those potentials can describe nite temperature
thermal properties as well as 0 K properties. One of the abovementioned interatomic potentials, the EAM [12], was further used
to develop a potential of the CoAl binary system [17]. However, to
the knowledge of the present authors, the formalism [12] used for
pure Co has never been applied to bcc elements. Since the alloying
of W or Mo is frequently carried out, it would be necessary to
perform atomistic simulations for the CoAlW or CoAlMo ternary
system. For those simulations, one would need a potential formalism
that can deal with hcp, fcc and bcc elements simultaneously.
The second nearest-neighbor modied embedded-atom method
(2NN MEAM) formalism, which was proposed by Lee and Baskes
[18,19] by generalizing the MEAM [20], is a such potential formalism, and has been used to describe a wide range of elements and
their alloy systems [21]. The 2NN MEAM has already been applied
to develop interatomic potentials for Al [22], W [19] and Mo [19] as
well as some hcp elements, Ti and Zr [23], and has also been
successfully used for exploring many aspects of solid interfacial
properties, especially the interfacial energy [24,25]. Therefore, one
can say that the 2NN MEAM can be a suitable potential formalism
to investigate the g-Co/g0 -L12 interfacial properties of Co-based
superalloys. As a start point to investigate the effect of elements
on interfacial properties of g-Co/g0 -L12 in Co-based superalloys
(CoAlW or CoAlMo) on an atomic scale, the purpose of the
present work is to develop 2NN MEAM interatomic potentials for
pure Co and the CoAl binary system.
This article presents a brief description of the 2NN MEAM
formalism and the procedure for the determination of potential
parameters. The reliability of the developed potentials is examined by comparing calculated fundamental physical properties
with available experimental or other calculations data. Then, the
potentials are used to calculate interfacial properties (interfacial
energy and work of separation) of CoAl binary alloys.

rah R r0 expbh R=re 1,

2
(h)

where r0 the atomic electron density scaling factor and b the decay
lengths are adjustable parameters, and re is the nearest-neighbor
distance in the equilibrium reference structure. A specic form is
given to the embedding function Fi, but not to the pair interaction fij .
Instead, a reference structure where individual atoms are on the
exact lattice points is dened and the total energy per atom of the
reference structure is estimated from the zero-temperature universal
equation of state of Rose et al. [26]. Then, the value of the pair interaction is evaluated from the known values of the total energy per
atom and the embedding energy, as a function of the nearestneighbor distance. In the original MEAM [20], only rst nearestneighbor interactions are considered. Neglecting the second and
more distant nearest-neighbor interactions is performed by the use
of a strong, many-body screening function [27]. The consideration of
the second nearest-neighbor interactions in the modied formalism
is effected by adjusting the screening parameters, Cmin, so that the
many-body screening becomes less severe. In addition, a radial cutoff
function [27] is applied to reduce the calculation time. Details of the
(2NN) MEAM formalism have been published in the literature
[1820,27] and will not be repeated here.
To describe binary alloy systems, the pair interaction between
different elements should be determined. For this, a similar
technique that is used to determine the pair interaction for pure
elements is applied to binary alloy systems. For the CoAl system,
the B2-CoAl ordered structure was chosen as a reference structures. In the B2-CoAl structure, the total energy per atom (for
1/2Co atom 1/2Al atom) is given as follows:
EuCoAl R



1
Z2
F Co rCo F Al rAl Z 1 fCoAl R
SCo fCoCo aR SAl fAlAl aR ,
2
2

3
where Z1 and Z2 are the numbers of rst and second nearestneighbors in the B2-CoAl structure, respectively. In the present
case, Z1 and Z2 are 8 and 6, respectively. SCo and SAl are the
screening functions for the second nearest-neighbor interactions
between Co atoms and between Al atoms, respectively, and a is
the ratio between the second and rst nearest-neighbor distances
in the reference structure. The pair interaction between Co and Al
can now be obtained in the following form:
fCoAl R



1
Z2
2EuCoAl RF Co rCo F Al rAl  SCo fCoCo aR SAl fAlAl aR :
2
Z1

4
The embedding functions FCo and FAl can be readily computed.
The pair interactions fCoCo and fAlAl between the same types of
atoms can also be computed from the descriptions of individual
elements. To obtain EuCoAl R, the universal equation of state [26]
should be considered again for the B2-CoAl as follows:

2. Interatomic potential
2.1. Potential formalism
In the MEAM, the total energy of a system is given by
2
X
_
1X
E4
F i ri
S f R ,
2 j a i ij ij ij
i

for different angular contributions with weight factors t(h) (h13).


Each partial electron density is a function of atomic conguration
and atomic electron density. The atomic electron densities ra(h)
(h04) are given as

Eu R Ec 1 an dan ea ,


1

where Fi is the embedding function for an atom i embedded in a


background electron density ri , Sij and fij Rij are the screening
function and the pair interaction between atoms i and j separated
by a distance Rij . For energy calculations, the functional forms for
Fi and fij should be given. The background electron density at
each atomic site is computed considering the directionality of
bonding, i.e. by combining several partial electron density terms

where d is an adjustable parameter,


an aR=r e 1

and


9BO 1=2
:
Ec

re is the equilibrium nearest-neighbor distance, Ec the cohesive


energy, B the bulk modulus and O is the equilibrium atomic
volume of the reference structure. The parameters Ec, re (or O),

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

2.3. Determination of potential parameters for the CoAl binary


system

B and d of the B2-CoAl in the universal equation of state are


determined from experimental data or from high-level calculations. Then the pair interaction between Co and Al is determined
as a function of the interatomic distance R.
2.2. Determination of potential parameters for pure Co
The 2NN MEAM formalism gives 14 independent model parameters for pure elements: four (Ec, re, B, d) for the universal
equation of state, seven (b(0), b(1), b(2), b(3), t(1), t(2), t(3)) for the
electron density, one (A) for the embedding function and two
(Cmin, Cmax) for many-body screening. Out of the 14 parameters,
Ec, re and B are the cohesive energy, the nearest-neighbor distance
and the bulk modulus of the reference structure, respectively.
When a real structure is selected as the reference structure, these
parameters become material properties and the 0 K rst-principles or room temperature experimental values are assigned to
them. In the present case, the hcp structure was selected as the
reference structure for the MEAM Co, and the values of the above
three parameters were obtained from experimental information
on hcp Co. The parameter d is given a value of either 0 or 0.05
according to the (@B=@P) value from experiments or rst-principles calculations [19]. The (@B=@P) value from a rst-principles
calculation for Co is 5.07 [26]. According to the present calculation, these values were 4.49 with d 0, and 5.01 with d 0.05.
However, a previous study [23] mentioned that rst-principles
calculations overestimate the (@B=@P) values for some elements
(Nb, Ta, Mo, W) by more than 15% compared to experimental
values. Therefore, a nal decision was made in the present study
to give the value of zero to the d for Co.
After the d value was determined, all the other parameters
(A, b(0), b(1), b(2), b(3), t(1), t(2), t(3), Cmin, Cmax) were determined by
tting to physical properties of Co. A value of 2.80 which is the
same as in the original [14,20] MEAM was initially given to Cmax
and the parameter optimization was started by giving a specic
value to Cmin. A and b(0) values were given considering elastic
constants and also roughly considering energy differences among
hcp, bcc, and fcc structures. Then, the b(1), b(2), b(3) parameter
values were adjusted so that elastic constants are reproduced
more accurately. After that, the t(1), t(2), t(3) parameter values were
determined tting to surface energies, vacancy formation energies and structural energy differences. The above procedure was
repeated changing Cmin value until most of the target properties
were correctly reproduced. The thermal expansion coefcient was
also considered when adjusting the Cmin value. When all target
property values were satisfactorily reproduced, a nal check was
made on the stability of the hcp structure at nite temperatures.
However the calculated energy difference between the bcc and
hcp structure was too high compared to a target value, and the
situation was not improved no matter how to change the Cmin
value. This difculty could be removed by lowering the Cmax
parameter value. In this way, the above procedure was repeated
again changing Cmax value until a parameter set that can describe
all the target properties satisfactorily was obtained. Table 1 shows
the nally determined 2NN MEAM parameters for the pure Co.

The MEAM for an alloy system is based on the MEAM potentials of


the constituent elements. In the present work, the MEAM parameters
for Al were taken from Lee et al. [22] without any modication. The
2NN MEAM potential parameters for pure Al are also listed in Table 1.
As described in the previous section, the extension of the
MEAM to an alloy system involves the determination of the pair
interaction between different types of atoms. The main task is to
estimate the potential parameters (Ec, re (or O), B and d) of the
universal equation of state for the reference structure. The rst
three are material properties if the reference structure is a real
phase structure that exists on the phase diagram of the relevant
system. Experimental data for that phase can be used directly.
Otherwise, the parameter values should be optimized so that
experimental information for other phases or high-level calculations results can be reproduced, if available, or assumptions
should be made. The fourth parameter, d, is a model parameter.
The value can be determined by tting to the (@B=@P) value of the
reference structure. When the reference structure is not a real phase,
it is difcult to estimate a reasonable value. For such alloy systems, d
is given an average value of those for the pure constituent elements.
In addition to the parameters for the universal equation of state,
two more model parameters, Cmin and Cmax, must be determined to
describe alloy systems. As can be seen in Table 1, each element has
its own value of Cmin and Cmax. Cmin and Cmax determine the extent
of screening of an atom (k) from the interaction between two
neighboring atoms (i and j). For pure elements, the three atoms are
all the same type (ijkAAA or BBB). However, in the case of
alloys, one of the interacting atoms and/or the screening atoms can
be different types (there are four cases: ikjABA, BAB,
AAB and ABB). Different Cmin and Cmax values may be given
in each case. Another model parameter is the atomic electron
density scaling factor r0. For an equilibrium reference structure
(Rre), the values of all atomic electron densities become r0. This is
an arbitrary value and does not have any effect on calculations for
pure elements. This parameter is often omitted when describing the
potential model for pure elements. However, for alloy systems,
especially for systems where the composing elements have different coordination numbers, the scaling factor (relative difference)
has a great effect on calculations.
The 13 model parameters discussed above, Ec, re, B, d, Cmin,
Cmax and r0 (there are four binary Cmin and Cmax parameters),
must be determined to describe an alloy system. The optimization
of the model parameters is performed by tting to known
physical properties of the alloy system using a trial and error
approach. Several sets of parameters that equally reproduce the
target property values are obtained. Those parameter sets are
used to calculate thermal properties or properties at nite
temperatures such as stability of equilibrium phases, thermal
expansion coefcients, etc. The best set is nally selected considering calculated thermal properties and 0 K properties.
As already mentioned, the B2 ordered CoAl phase was chosen
as the reference structure for the CoAl system. The experimental

Table 1
2NN MEAM potential parameters for pure Co and Al. The units of the cohesive energy Ec, the equilibrium nearest-neighbor distance re and bulk modulus B are eV, A and
1012 dyn cm  2, respectively. The reference structures are hcp Co and fcc Al.

Co
Ala
a

Ec

re

b(0)

b(1)

b(2)

b(3)

t(1)

t(2)

t(3)

Cmin

Cmax

4.41
3.36

2.50
2.86

1.948
0.794

0.9
1.16

3.50
3.20

0.0
2.6

0.0
6.0

4.0
2.6

3.00
3.05

5.00
0.51

 1.0
7.75

0.49
0.49

2.00
2.80

0.00
0.05

Ref. [22].

10

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

Table 2
2NN MEAM potential parameters for the CoAl system. The units of the formation
energy of the reference phase DEc, the equilibrium nearest-neighbor distance re
and bulk modulus B are eV, A and 1012 dyn cm  2, respectively. The reference
structure is B2-CoAl.
Selected value
Al
DEc(EBc 2  0.5ECo
c  0.5 Ec )
re
B
d
Al
rCo
0 /r0
Cmin (CoAlCo)

Cmin (AlCoAl)
Cmin (CoCoAl)

 0.565
2.4768
1.62
0.5dCo 0.5dAl
1:1
0.49( C Co
min )
1.10
1/2
1/2 2
) 0.5(C Al
]
[0.5(C Co
min
min )

Cmin (CoAlAl)

1/2
1/2 2
) 0.5(C Al
]
[0.5(C Co
min
min )

Cmax (CoAlCo)

2.0( C Co
max )

Cmax (AlCoAl)

2.8( C Al
max )
2.8 (default value)
2.8 (default value)

Cmax (CoCoAl)
Cmax(CoAlAl)

physical properties of the CoAl alloys, which are available in


literature and can thus be used to determine potential parameter
values, were lattice parameter [17,28,29], bulk modulus [30,31],
elastic constants [17] and melting point [28,31] of the abovementioned ordered B2 phase. Lattice parameters and enthalpy of
formation for L12 Co3Al [3436] and CoAl3 [3537] phases, and
enthalpy of mixing of liquid [35] calculated using a CALPHAD
method, were also available and could be considered for further
adjustment of potential parameters.
The Ec, re, and B parameter values were directly determined
from corresponding experimental values for CoAl. The d value
of the reference structure was approximated as a weighted
average (1:1) of those for pure Co and Al due to the lack of
information, as frequently had been done [32,33]. Most of the
eight Cmin and Cmax parameters were given default assumed
values, while only one of Cmin was optimized to better describe
the elastic constants for CoAl and the enthalpy of formation for
Co3Al and CoAl3 phases. The atomic electron density scaling factors
(r0) for Co and Al were assumed to be equal to each other, i.e.,
the ratio is 1, because a reasonable composition dependency of
enthalpy of mixing of liquid could be obtained without adjusting
the ratio. In addition to the above-mentioned experimental information, atomic volume [28], enthalpy of formation [37,38] and
lattice parameters [29,35,3942] of various Al-rich intermetallic
phases (mP22-Al9Co2, mC93-Al13Co4 and hP28-Al5Co2) were available. All those information was used for comparison, as a means to
evaluate the transferability of the developed potential. Table 2
shows the nally determined 2NN MEAM potential parameters for
the CoAl binary system.

3. Calculation of physical properties


In this section, fundamental physical properties of pure Co and
CoAl binary alloys were calculated using the present 2NN MEAM
potentials, and were compared with experimental data or other
calculations in order to evaluate the reliability of the potentials.
The 2NN MEAM formalism includes up to second nearest-neighbor
interactions. Therefore, the radial cutoff distance during atomistic
simulations should be larger than the second nearest-neighbor
distance in structures under consideration. In the present study,
a value of 4.5 A which is larger than the second nearest-neighbor
distance of aluminum was chosen as the radial cutoff distance.

3.1. Physical properties for pure Co


The calculations for pure Co were performed for bulk properties (lattice constants, elastic constants, structural energy differences), point defect properties (vacancy, divacancy and interstitial
formation energy, activation energy of vacancy diffusion), planar
defect properties (surface energy, stacking fault energy) and
thermal properties (thermal expansion coefcient, specic heat,
melting point, enthalpy and volume change on melting). All
calculations except thermal properties were performed at 0 K,
allowing full relaxation of individual atoms. Comparisons between
the present calculations and experimental data or other calculations
are presented in Table 3. Here, properties marked with a * are
those used for tting during the parameter optimization.
As mentioned earlier, the values for cohesive energy Ec,
nearest-neighbor distance re and bulk modulus B of the reference
structure (hcp in the present case) were given as input data in the
MEAM. Therefore, it is not surprising to see that those quantities
for the hcp-Co are correctly reproduced, even though there can be
small deviations from the given values because of the relaxation
(non ideal value of c/a ratio). Among the other target properties,
elastic constants were given the highest weight during the tting
procedure, because it was thought that elastic constants can be
measured most accurately than any other properties. The biggest
effort was made to improve the structural energy difference
between hcp and bcc Co, DEhcp-bcc, because the original MEAM
[14] yielded a value (0.241 eV/atom) far from a thermodynamically assessed value using the CALPHAD method (0.022 eV/atom)
[47]. Compared to the previous MEAM value [14], the present
value, 0.0472 eV/atom, is much closer to the CALPHAD value.
Other structural energy differences, DEhcp-fcc, DEhcp-sc and
DEhcp-dia, are also comparable with the CALPHAD [47] and MEAM
results [14]. It is shown that the present potential can describe
bulk properties of Co as satisfactorily as any other empirical
potentials.
In case of point defect properties, the present potential reproduces the vacancy formation energy in good agreement with
experimental data [4851]. For the vacancy migration energy,
activation energy of vacancy diffusion and the divacancy formation
energy, two different values are given, one for in-basal plane
migration or formation (designated as In) and the other for outof-basal plane migration or formation (designated as Out). Out of
several probable interstitial sites, the octahedral site is calculated to
be energetically most favorable for a self-interstitial atom, and the
corresponding self-interstitial formation energy is included in
Table 3. For surface energy, only average values (the values given
in parentheses) for polycrystalline solids are available in literatures
[48,52], while calculations can be performed for several different
planes such as basal, prism and pyramidal planes. On the basal
plane, there are two intrinsic and one extrinsic stacking fault, where
I1 represents ABABCBCB, I2 ABABCACA, and E ABABCABAB. I2 is the
most important stacking fault in connection with plastic deformations (dislocation splitting into Shockley partials). Therefore, the I2
stacking fault energy was calculated and compared with experimental data [53]. The present potential shows a good agreement
with experimental data for the stacking fault energy as well as the
surface energy.
The next properties calculated using the present 2NN MEAM
potential were the thermal properties such as thermal expansion
coefcient, specic heat, melting point, enthalpy of melting and
volume change on melting. The results are compared with available
experimental data [53,54] in Table 3. The thermal expansion
coefcient and specic heat were calculated in a temperature range
of 0600 K. The melting point was calculated using an interface
method. The enthalpies of melting and the volume change were
those calculated at the calculated melting point. Here the values in

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

11

Table 3
Calculated physical properties of hcp Co using the present 2NN MEAM potential, in comparison with experimental data and other calculations. Values listed are the
the elastic constants B, C11, C12, C44, C33, C13 (1012 dyn/cm2), the structural energy difference DE (eV/atom), the
cohesive energy Ec (eV), the lattice parameter a and c (A),
f
vacancy formation energy EfV , vacancy migration energy Em
V , activation energy of vacancy diffusion QD and divacancy formation energy E2V (eV), the self-interstitial
formation energy QUOTE (eV), the surface energy of basal (expt. values are average values), prism and pyramidal planes (erg/cm2), the stacking fault energy Esf (erg/cm2),
thermal expansion coefcient e (10  6/K), specic heat Cp (J/mol K), melting point m.p. (K), enthalpy of melting DHm (kJ/mol) and the volume change on melting DVm/Vsolid
(%). Properties marked with * are those used for tting during the parameter optimization. Values in parentheses are for metastable melting of hcp Co.

Co

Expt.

 4.41a,  4.39b,  4.39c,  4.39d

Ec
Lattice parameter

a
c
c/a ratio (relative to ideal)

B
C11
n
C12
n
C44
n
C33
n
C13
n
DEhcp-bcc
n
DEhcp-fcc
DEhcp-sc
DEhcp-dia
n

n f

EV

Em
V
QD
Ef2V
EfI
n

Esurf

2NN MEAM

MEAM
Ref. 14

EAM
Ref. 12

AEAM
Ref. 15

AEAM
Ref. 16

FP calculations

 4.41

 4.41

 4.39

 4.39

 4.39

 4.386p,  4.39q

2.51 , 2.50
4.069c
0.994b, 0.998e, 0.994f
1.878a, 1.898b
3.195b
1.661b
0.824b
3.736b
1.021b
0.022g
0.004g

2.502
4.075
0.998
1.948
3.22
1.402
0.688
3.39
1.226
0.0472
0.0036
0.668
1.224

2.495
4.075
1.000
1.863
3.193
1.389
0.705
3.256
1.176
0.241
0.005
0.59
1.2

2.507
4.069
0.994
1.773n
2.721n
1.302n
0.809n
3.413n
0.736n
0.061n
0.007
0.65n
1.37n

2.507
4.069
0.994
1.948
3.275
1.340
0.728
3.569
1.178
0.0356
0.0029

2.497
4.069
0.998

2.507p, 2.494r

2.95
1.59
0.71
3.35
1.11
0.0164
0.0062

2.039p, 1.954q
3.142p, 3.1445r
1.408p, 1.4513r
0.667p, 0.6131r
3.991p, 3.4384r
1.223p, 1.2193r
0.0852n
0.0231n

1.35h, 1.35i, 1.46j, 1.41k

1.46

1.48

1.46

1.46

1.40

1.56p, 1.5754r

0.72
0.72
2.12
2.12
2.61
2.59
3.48

0.89s
0.89s
2.301s
2.30s

In
Out
In
Out
In
Out
Octahedron
Basal (0001)

0.86
0.81
2.32
2.27
2.78
2.68
3.67
2550h, 2160l

Prism (1 1 0 0)

1934
1957

Prism (1 1 2 0)

2057

Pyramidal(1 1 0 1)

2023

e(0600 K)
Cp(0600 K)
n
m.p.
D Hm
DVm/Vsolid
n

0.36
0.42n

2.54
2.50

2775s, 2148t

2347
3454

1172

2124

Pyramidal(1 1 0 2)
Esf(I2)

27m, 42n
m

13.8
27.30o
1768g, (1700)g
16.20g
3.84o

33
11.03
26.04
1600 (1585)
17.1 (16.39)
5.08 (4.16)

30

41

37
17.1p
32.95p
1950p
14.74p
3.17p

Calculation by the present authors using lammps.


Ref. [43].
b
Ref. [44].
c
Ref. [45].
d
Ref. [46].
e
Ref. [14].
f
Ref. [29].
g
Ref. [47].
h
Ref. [48].
i
Ref. [49].
j
Ref. [50].
k
Ref. [51].
l
Ref. [52].
m
Ref. [53].
n
Ref. [16].
o
Ref. [54].
p
Ref. [55].
q
Ref. [56].
r
Ref. [57].
s
Ref. [58].
t
Ref. [59].
a

parentheses are for metastable hcp Co. According to the present


2NN MEAM calculation the melting point of fcc Co is higher than
that of hcp Co by about 15 K. This means that fcc Co is more stable
than hcp Co near melting point and that the present potential
describes the experimentally known hcp-fcc transformation.
Finally, in order to further conrm the reliability of the present
potential, the lattice expansions in a and c directions were calculated

at various temperatures and compared with available literature


data [60]. By this, it could be examined whether the present 2MM
MEAM reproduces well the temperature dependence of the c/a
ratio which is believed to be closely related with the slip behavior
of hcp alloys. The results are presented in Fig. 1, where the lattice
expansion represents fractional changes (%) relative to lattice
parameters at 293 K. It is shown that the thermal expansions in a

12

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

and c directions are excellently reproduced by the present


potential.

3.2. Physical properties of CoAl binary system


The fundamental physical properties of the CoAl alloys calculated
using the present 2NN MEAM potential listed in Tables 1 and 2 are
presented in this section and compared with experimental data and
other calculations. An ideal interatomic potential for an alloy system
would be the one that can correctly reproduce physical properties
(thermodynamic, structural and elastic properties, etc.) of all solution
and intermediate phases relevant to the system. Therefore, an
attention was paid to whether the present potential can describe
the above-mentioned alloy properties correctly.

Fig. 1. Thermal linear expansion of hcp Co in a and c directions calculated using


the present 2NN MEAM potential. Literature data is from Ref. [60].

In the present work, some rst-principles calculations were


also performed to provide necessary information for comparisons.
Those calculations were based on the density functional theory
using the ab initio total energy and molecular dynamics program
VASP (Vienna ab initio simulation program) developed at the
Material physik of the Universitat
Wien [6163].
Institut fur
The calculations were done in a plane-wave basis and the
exchange-correlation functional is expressed by the generalizedgradient approximation using the projector augmented wave
approach (PAW-PBE). All pseudopotentials were taken from the
VASP library.
Since the B2-CoAl is the reference structure and also the most
stable structure in the CoAl binary system, the properties calculated
rst using the present 2NN MEAM potential are the cohesive energy,
lattice parameter, elastic constants, surface energy and thermal
properties of the B2-CoAl compound (see Table 4). Again, the target
values for the cohesive energy (enthalpy of formation), lattice
parameter (nearest-neighbor distance) and bulk modulus which
were given as input data are correctly reproduced. The other elastic
constants, C11, C12 and C44, and melting point are also reproduced in
good agreement with experimental information [17,28,31].
Besides the B2-CoAl, four more intermediate phases (L12CoAl3, mP22-Al9Co2, mC93-Al13Co4 and hP28-Al5Co2) appear on
the phase diagram and one more metastable phase, L12-Co3Al, is
also reported [1] for the CoAl alloy system. It could be better if
the present potential could reproduce physical properties of all
those phases correctly. Although the crystallographic structures
of the above phases are very complicated to be reproduced using
an empirical potential, the calculated lattice parameters (Table 5)
and the enthalpy of formation (Table 6) of those structures are
comparable with experimental data. Fig. 2 compares the calculated atomic volume of individual phases with corresponding
experimental data [28], and Fig. 3 shows an enthalpy vs. composition diagram according to the present calculation. It should be
noticed here that the present potential predicts most of the
Al-rich intermediate compounds (mP22-Al9Co2, mC93-Al13Co4
and hP28-Al5Co2) as metastable phases and the metastable

Table 4
Calculated physical properties of B2-CoAl using the present 2NN MEAM potential, in comparison with experimental data and other calculations. Values listed are the
the elastic constants B, C11, C12, C44 (1012 dyn/cm2), the surface energy (erg/cm2) (eV), thermal expansion coefcient e
cohesive energy Ec (eV), the lattice parameter a (A),
(10  6/K), specic heat Cp (J/mol K), melting point (K), enthalpy of melting DHm (kJ/mol) and the volume change on melting DVm/Vsolid (%). Properties marked with n are
those used for tting during the parameter optimization.
CoAl

Expt.

2NN MEAM

EAM

EDM

 4.45a,  4.445b
2.86a, 2.863b 2.86c
1.62d, 1.62 70.03e
2.69a
1.07a
1.39a

 4.45
2.86
1.62
2.659
1.099
1.391
2095
1771
2027
15.76
26.8
1900
29.47
13.36

 4.47a
2.86a
2.04a
2.26a
1.46a
1.15a
1799f
1574f
1863f

 4.47
2.86a
1.99a
2.31a
1.43a
1.59a

Ec
Lattice parameter, a
n
B
n
C11
n
C12
n
C44
n
Esurf
n

e(3001200 K)
Cp(3001200 K)
n
Melting point
D Hm
DVm/Vsolid

(100)
(110)
(111)

1921c, 1700e

Calculation by the present authors using VASP.


Ref. [17].
b
Ref. [29].
c
Ref. [28].
d
Ref. [30].
e
Ref. [31].
f
Ref. [64].
g
Ref. [35].
h
Ref. [36].
i
Ref. [65].
a

FP calculations
a

 4.393g,  4.505h,  4.393n


2.862b, 2.861c, 2.862e, 2.8672i, 2.85n
2.077 0.02c, 1.51d, 1.57 7 0.01e, 1.592i, 1.78n
2.56a, 2.51 7 0.04c, 2.57e, 2.95n
1.07a, 1.017 0.03c, 1.07e, 1.19n
1.30a, 1.34 70.01c, 1.3e, 1.39n

15.82i
11.5i
2040c, 1645c, 2070 7310e, 1865i

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

13

Table 5

Calculated lattice parameters of intermetallic phases in comparison with experimental data and other calculations. The unit of the lattice parameter is A.
Phase

Expt.

2NN MEAM

L12-Co3Al
L12-CoAl3

Lattice parameter
Lattice parameter

a
a

3.63
3.81

mP22-Al9Co2

Lattice parameter

a
b
c

6.213a, 6.18b
6.290a, 6.24b
8.5565a, 8.59b

mC93-Al13Co4

Lattice parameter

a
b
c

8.14c, 8.158c, 8.1593d


12.72c, 12.342c, 12.3586d
14.79c, 14.452c, 14.4547d

hP28-Al5Co2

Lattice parameter

a
b
c

7.6715a, 7.6581d, 7.6717e


7.656e
7.6715a, 7.6044d, 7.6052e

EAM Ref. [17]

FP calculations

3.56*
3.80*

3.655f, 3.52g, 3.59*


3.73g, 3.79*
6.089h

6.13
6.46
8.42
7.79
12.14
15.72

7.524h

7.79
7.79
7.539

Calculation by the present authors using lammps or VASP.


Ref. [29].
b
Ref. [39].
c
Ref. [40].
d
Ref. [41].
e
Ref. [42].
f
Ref. [34].
g
Ref. [35].
h
Ref. [66].
a

Table 6
Calculated enthalpy of formation of intermetallic phases, in comparison with literature data. The unit of the enthalpy of formation is kJ/gram-atom.
Phase
L12-Co3Al
L12-CoAl3
mP22-Al9Co2
mC93-Al13Co4
hP28-Al5Co2

Expt.

 0.40a
 0.31a,  0.31b
 0.43a

2NN MEAM

EAM Ref. [17]

FP calculations

 0.34
 0.28
 0.16
 0.23
 0.26

 0.41*
 0.24*

 0.20c,  0.18*
 0.402a,  0.43c,  0.21*
 0.306a,  0.34c,  0.31b,  0.32b,  0.46d
 0.41c
 0.374a,  0.48c,  0.69d

Calculation by the present authors using lammps or VASP.


Ref. [37].
Ref. [38].
c
Ref. [36].
d
Ref. [66].
a

Fig. 2. Atomic volume of various intermetallic phases in the CoAl binary system
calculated at 0 K using the present 2NN MEAM potential. Literature data is from
Ref. [28].

Fig. 3. Enthalpy of formation of various intermetallic phases in the CoAl binary


system calculated at 0 K using the present 2NN MEAM potential.

L12-Co3Al phase as a stable phase, erroneously, even though the


size of error is small as shown in Fig. 3. Fig. 4 shows that the
enthalpy of mixing for liquid is in good agreement with a CALPHAD
calculation [35].

As a further means to examine the reliability of the present


potentials, the elastic constants of fcc-Co, L12-Co3Al and L12-CoAl3
were calculated and compared with available literature data [67]
(see Table 7). In the case of fcc-Co, a rst-principles calculation

14

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

was available [67], of which values are comparable with the


present calculation. On the other hand, no direct information was
available for the elastic constants of L12-Co3Al and L12-CoAl3.
Therefore, a rst-principles calculation was performed for those
quantities in the present study, and the results are inserted in
Table 7.
It has been shown that the present 2NN MEAM interatomic
potential describes fundamental physical properties of pure Co and
CoAl binary system reasonably well. Especially, thermodynamic

Fig. 4. Enthalpy of mixing of the liquid CoAl alloys calculated at 2500 K using the
present 2NN MEAM potential, in comparison with CALPHAD assessments [35].

Table 7
Calculated elastic constants C11, C12, C44, B for fcc-Co, L12-Co3Al and L12-CoAl3, in
comparison with literature data. The unit of elastic constant is 1012 dyn/cm2.
Structure

C11

C12

C44

fcc-Co

Present work
FP calc.a
FP calc.

2.80
3.00
2.96n

1.52
1.92
1.76n

1.10
1.59
1.36n

1.95
2.28
2.07n

L12-Co3Al

Present work
EAM [17]
FP calc.

3.13
2.80n
2.02n

1.44
1.84n
1.78n

1.42
1.31n
0.95n

2.00
2.16n
1.86n

L12-CoAl3

Present work
EAM [17]
FP calc.

1.82
2.51n
2.00n

0.85
1.51n
0.86n

0.69
0.94n
0.68n

1.17
1.84n
1.26n

Calculation by the present authors using lammps or VASP.


Ref. [67].

and elastic properties of fcc-Co (g) and L12-Co3Al (g0 ) are also
reasonably described, which indicates that the present potential
should be suitable for an atomistic investigation of interfacial
properties of g/g0 coherent interfaces in Co-based alloys.

4. Calculation of interfacial properties


As mentioned earlier, interfacial energy and its anisotropy are
crucial parameters that control the size distribution and the shape
of precipitates. Therefore, it would be necessary to know those
quantities and the effect of alloying elements on them for the g/g0
interface in order to guide an alloy design of Co-based superalloys. Calculation of interfacial energy is performed by computing the total energy of a supercell that involves an interface
between the two phases and the summation of total energies of
individual pure phase samples of the same size [24]. The total
energy of the supercell Esupercell is calculated by a molecular statics
simulation applying a three-dimensional periodic boundary condition to remove any surface effects and allowing full relaxations
of individual atom positions and sample dimensions. This energy
is compared with the summation of total energies of g-Co (ECo )
and g0 -Co3Al (ECo3 Al ) samples of the same size, calculated maintaining the same lattice parameters with those of the supercell in
the two directions parallel to the interface but allowing a relaxation into the direction perpendicular to the interface. Then, the
interfacial energy s is dened by the following equation:



1
s Esupercell  ECo ECo3 Al =2A,
8
2
where A represents the area of the interface (two interfaces
should be considered due to the periodic boundary condition).
In the present study, samples with three typical low-index
coherent interfaces: the (100), (110) and (111) interfaces, were
prepared by combining equal amounts of g-Co and g0 -Co3Al. Fig. 5
shows part of those samples. Initially, pure Co was used for the gCo sample. The calculated interfacial energies were of negative
values for all the three interfaces as listed in Table 8. A similar
result had been obtained in a previous study on the interfacial
energy between g-Ni and g0 -Ni3Al [24]. A negative value of
interfacial energy could be interpreted as that dissolution of the
compound phase would be energetically favorable in the given
condition (pure Co or pure Ni). It should be reminded here again
that at a typical annealing temperature, 1300 K of Co-based
superalloys, the solubility of Al in Co is around 15%. Therefore,
the calculation of interfacial energy should be performed using a
binary solution sample for g-Co, instead of pure Co. A reasonable

Fig. 5. Supercells used to calculate the (001), (011) and (111) interfacial energies between g0 -Co3Al and g-Co. White and red (or dark) circles denote Co and Al atoms,
respectively.

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

Table 8
Calculated coherent interfacial energy at 0 K between g-Co and Co3Al at various Al
contents in g-Co. The unit of the interfacial energy is J m  2.
at% Al in g-Co

0%
5%
10%
15%
20%

(100)

(110)

(111)

 0.101
 0.075
 0.039
0.002
0.016

 0.066
 0.043
 0.014
0.008
0.019

 0.032
 0.006
0.011
0.026
0.042

value could be obtained for the g/g0 interfacial energy when


a solid solution with 15% Al was used for the g-Ni sample in the
NiAl system [24]. Similar solid solution samples with various
Al contents were used for the g-Co phase in the present study, and
the results are listed in Table 8. Positive values are obtained for all
the three interfaces when the Al content is over 15%.
When a random solid solution is used in an atomistic computation, it is necessary to pay attention to the probable effect of
solute distribution on computation results, because there can be a
statistical error due to the limited size (number of atoms) of
samples. To reduce the probable statistical error, relatively large
samples with 24000  36000 atoms were used and ve independent calculations were performed for individual interfaces. The
values listed in Table 8 are averaged values of the ve independent calculations. It is widely accepted that the (100) interface
has the lowest energy for the g/g0 interface in Co-based superalloys [5,6]. As expected, the present calculation (see Table 8)
predicts the (100) interface as the lowest energy interface
between Co3Al (g0 ) and Co-rich fcc solution phase (g).
The g/g0 interface is known to be the weakest region in superalloys [5,6]. To a certain degree, the binding strength of g/g0
interface can be regarded to represent the rupture strength of
superalloys. Note that the critical stress required to crack propagation is given by the Grifth equation

sF W se Y=cp1=2 ,

0%
5%
10%
15%
20%

Work of separation
(100)

(110)

(111)

4.37
4.35
4.30
4.27
4.19

4.34
4.29
4.26
4.22
4.17

3.99
3.96
3.93
3.89
3.84

wide possibility to change the anisotropy in the interfacial energy


and the shape of the g0 precipitates by alloying other elements. It
is believed that the size distribution can also be changed by
alloying and thus changing the lattice mismatch or mist strain
energy at the interface. Indeed, one can notice that different
shapes and size distributions of g0 -Co3(Al,W) precipitates have
been reported depending on the further alloying elements, Ta, Ni,
V or Mo [1,57].
The strongest advantage of the present 2NN MEAM approach
is that all the above-mentioned properties can be investigated for
practical Co/Co3(Al, W, Ta or Mo) multicomponent alloys since
interatomic potentials are already available for all the additional
alloying elements, W [19], Ta [19] and Mo [19]. One can expect
that a more elaborate (computational) materials design toward
advanced Co-based superalloys would be feasible. It can be
further mentioned here that a home-made molecular dynamic
simulation code based on the present 2NN MEAM interatomic
potentials is available on-line (as a supplemental online material
of an article [21] and on the website: http://cmse.postech.ac.kr)
together with a potential parameter database. The 2NN MEAM
potential formalism was recently implemented to the lammps
([68], /http://lammps.sandia.gov/S), one of the most widely used
molecular dynamics code, which means that all the 2NN MEAM
potential parameter sets can also be used with the lammps code.

where sF is the critical stress for the crack propagation, Y is the


Youngs modulus, c is the length of a surface crack. Therefore, for a
given size of a crack and material type, one can see that W se is the
governing factor that determines the size of sF . W se , the work of
separation is dened as the reversible work required to separate a
supercell along the interface into two phase samples, creating a free
surface in each sample, and is employed to evaluate the binding
strength of the corresponding interface. For the calculation of
the work of separation for the g/g0 interface, the same supercells
as in the calculation of interfacial energy are used. However,
when calculating the total energy of individual phase samples
(E0Co and E0Co3 Al ), the side of each sample that composed the interface
is left as a free surface without applying the periodic boundary
condition. Again, the same lattice parameters as those of the supercell in the two directions parallel to the interface are maintained. The
work of separation is then obtained from the following equation:
W se E0Co E0Co3 Al Esupercell =2A:

Table 9
Calculated work of separation at 0 K between g-Co and Co3Al at various Al
contents in g-Co. The unit of the work of separation is J m  2.
at.% Al in g-Co

Interfacial energy

15

10

The results of the calculations are listed in Table 9. It is shown


that the (100) interface has the highest value of the work of
separation even though the differences from other interfaces
considered are within 10%.
What should be further discussed at this point is that even
though the (100) interface has the lowest energy as shown in
Table 8, the absolute value is quite small and the difference from
those of other interfaces is also small. This means that there is a

5. Conclusion
An interatomic potential that can describe various fundamental physical properties of pure Co and the CoAl binary system in
good agreement with experimental information is now available.
The newly developed 2NN MEAM potential was used to investigate interfacial properties (interfacial energy and work of separation) between g-Co and g0 -Co3Al. Even though the (100) interface
has the lowest interfacial energy and the biggest work of separation, the differences from the other interfaces ((110) and (111))
considered are small, especially for the interfacial energy. These
small differences provide a large space for alloying other elements
to change its anisotropy and the shape of the g0 precipitates. Since
the 2NN MEAM potential is already available for various elements
and multicomponent alloys, the potential developed in the present work can be easily extended to multicomponent CoAlW
Metal systems for an elaborate alloy design of advanced Co-based
superalloys.

Acknowledgments
This work was supported by WCU (World Class University)
program through the National Research Foundation of Korea
funded by the Ministry of Education, Science and Technology
(R31-30005) and was supported by a grant from the National

16

W.-P. Dong et al. / CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 716

Natural Science Foundation of China (Grant nos. 10902086,


51075335 and 51174168), the Basic Research fund of the Northwestern Polytechnical University (Grant nos. JC201005 and
201109), the Doctoral Foundation of Northwestern Polytechnical
University (CX201103) and the Ministry of Education Fund for
Doctoral Students Newcomer Awards of China (W-P Dong). W-P
Dong would like to thank all the members of the CMSE Lab at
POSTECH for their assistance.

Appendix A. Supplementary material


Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.calphad.2012.04.001.

References
[1] J. Sato, T. Omori, K. Oikawa, I. Ohnuma, R. Kainuma, K. Ishida, Science 312
(2006) 9091.
[2] D.H. Ping, C.Y. Cui, Y.F. Gu, H. Harada, Ultramicroscopy 107 (2007) 791795.
[3] T. Oohashi, N.L. Okamoto, K. Kishida, K. Tanaka, H. Inui, Mater. Res. Soc.
1128C (2009) 16.
[4] S. Miura, K. Ohkubo, T. Mohri, Mater. Trans. 48 (2007) 24032408.
[5] A. Suzuki, G.C. DeNolf, T.M. Pollock, Scr. Materialia 56 (2007) 385388.
[6] A. Suzuki, T.M. Pollock, Acta Materialia 56 (2008) 12881297.
[7] T.M. Pollock, J. Dibbern, M. Tsunekane, J. Zhu, A. Suzuki, JOM 62 (2010)
5863.
[8] M. Chen, C.Y. Wang, Phys. Lett. A 374 (2010) 32383242.
[9] S. Kobayashi, Y. Tsukamoto, T. Takasugi, H. Chinen, T. Omori, K. Ishida,
S. Zaefferer, Intermetallics 17 (2009) 10851089.
[10] Q. Yao, H. Xing, J. Sun, Appl. Phys. Lett. 89 (2006) 161906.
[11] H. Li, J.B. Sha, S.S. Li, Acta Aeronaut. Astronaut. Sin. 32 (2011) 11391146.
[12] R. Pasianot, E.J. Savino, Phys. Rev. B 45 (1992) 1270412710.
[13] X.W. Zhou, R.A. Johnson, H.N.G. Wadley, Phys. Rev. B 69 (2004) 144113.
[14] M.I. Baskes, R.A. Johnson, Modelling Simulation Mater. Sci. Eng. 2 (1994)
147163.
[15] R.F. Zhang, Y. Kong, B.X. Liu, Phys. Rev. B 71 (2005) 214102.
[16] W.Y. Hu, B.W. Zhang, B.Y. Huang, F. Gao, D.J. Bacon, J. Phys.: Condens. Matter
13 (2001) 11931213.
[17] C. Vailhe, D. Farkas, J. Mater. Res. 12 (1997) 25592570.
[18] B.J. Lee, M.I. Baskes, Phys. Rev. B 62 (2000) 85648567.
[19] B.J. Lee, M.I. Baskes, H. Kim, Y.K. Cho, Phys. Rev. B 64 (2001) 184102.
[20] M.I. Baskes, Phys. Rev. B 46 (1992) 27272742.
[21] B.J. Lee, W.S. Ko, H.K. Kim, E.H. Kim, CALPHAD 34 (2010) 510522.
[22] B.J. Lee, J.H. Shim, M.I. Baskes, Phys. Rev. B 68 (2003) 144112.
[23] Y.M. Kim, B.J. Lee, M.I. Baskes, Phys. Rev. B 74 (2006) 014101.

[24] A.C. Silva, J. Agren,


M.T. Clavaguera-Mora, D. Djurovic, T. Gomez-Acebo,
B.J. Lee, Z.K. Liu, P. Miodownikh, H.J. Seifert, CALPHAD 31 (2007) 5374.
[25] H.K. Kim, W.S. Jung, B.J. Lee, Acta Materialia 57 (2009) 31403147.
[26] J.H. Rose, J.R. Smith, F. Guinea, J. Ferrante, Phys. Rev. B 29 (1984) 29632969.

[27] M.I. Baskes, Mater. Chem. Phys. 50 (1997) 152158.


[28] A.F. Voter, Intermetallic Compounds: Principles and Practice, Wiley, New
York, 1993.
[29] P. Villas, L.D. Calvert, Pearsons Handbook of Crystallographic Data for
Intermetallic Phases, ASM, Materials Park, OH, 1991.
[30] R.L. Fleischer, Acta Metall. Materialia 41 (1993) 863869.
[31] M.J. Mehl, J.E. Osburn, D.A. Papaconstantopoulos, B.M. Klein, Phys. Rev. B 41
(1990) 1031110323.
[32] Y.M. Kim, B.J. Lee, J. Mater. Res. 23 (2008) 10951104.
[33] Y.M. Kim, Y.H. Shin, B.J. Lee, Acta Materialia 57 (2009) 474482.
[34] S.R. Broderick, H. Aourag, K. Rajan, Stat. Anal. Data Min. 1 (2009) 353360.
[35] H. Ohtani, M. Yamano, M. Hasebe, CALPHAD 28 (2004) 177190.
[36] M. Mihalkovic, M. Widom, Phys. Rev. B 75 (2007) 014207.
[37] M. Widom, J.A. Moriarty, Phys. Rev. B 58 (1998) 89678979.
[38] J.A. Moriarty, M. Widom, Phys. Rev. B 56 (1997) 79057917.
[39] X.L. Ma, K.H. Kuo, Metall. Mater. Trans. A 23 (1992) 11211128.
[40] J. Grin, U. Burkhardt, M. Ellner, K. Peters, J. Alloys Compd. 206 (1994)
243247.
[41] B. Grushko, R. Wittenberg, K. Bickmann, C. Freiburg, J. Alloys Compd. 233
(1996) 279287.
[42] U. Burkhardt, M. Ellner, Y. Grin, B. Baumgartner, Powder Diffr. 13 (1998)
159162.
[43] E.A. Brandes, Smithells Metals Reference Book, Butterworth-Heinemann,
London, 1983.
[44] G. Simmons, H. Wang, Single Crystal Elastic Constants and Calculated
Aggregate Properties, MIT Press, Cambridge, 1971.
[45] C. Kittel, Introduction to Solid State Physics, Wiley, New York, 1976.
[46] C.J. Smithells, Metals Reference Book, Butterworks, London, 1976.
[47] A.T. Dinsdale, CALPHAD 15 (1991) 317425.
[48] F.R. Boer, R. Boom, W.C.M. Mattens, A.R. Miedema, A.K. Niessen, Cohesion in
Metals: Transition Metal Alloys, North-Holland, Amsterdam, 1988.
[49] H.J Wollenberger, Physical Metallurgy, Elsevier, Amsterdam, 1983.
[50] E.A. Brandes, G.B. Brook, Smithells Metals Reference Book, ButterworthHeinemann, London, 1992.
[51] J.R. Beeler, M.F. Beeler, Interatomic Potentials and Crystalline Defects, AIME,
New York, 1981.
[52] W.R. Tyson, W.A. Miller, Surf. Sci. 62 (1977) 267276.
[53] A. Korner, H.P. Karnthaler, Philos. Mag. A 48 (1983) 469477.
[54] R. Hultgren, Selected Values of the Thermodynamic Properties of Binary
Alloys, ASM, Metals Park, Ohio, 1973.
[55] F. Cleri, V. Rosato, Phys. Rev. B 48 (1993) 2233.
[56] J.H. Li, S.H. Liang, H.B. Guo, B.X. Liu, Appl. Phys. Lett. 87 (2005) 194111.
[57] H.B. Guo, J.H. Li, B.X. Liu, Phys. Rev. B 70 (2004) 195434.
[58] L. Vitos, A.V. Ruban, H.L. Skriver, J. Kollar, Surf. Sci. 411 (1998) 186202.
[59] E. Aghemenloh, J.O.A. Idiodi, S.O. Azi, Comput. Mater. Sci. 46 (2009) 524530.
[60] /http://www.diracdelta.co.uk/science/source/l/i/linear%20thermal%20expan
sion/source.htmlS.
[61] G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251.

[62] G. Kresse, J. Furthmuller,


Comput. Mater. Sci. 6 (1996) 15.

[63] G. Kresse, J. Furthmuller,


Phys. Rev. B 54 (1996) 11169.
[64] S.P. Kim, S.C. Lee, K.R. Lee, Y.C. Chung, Acta Materialia 56 (2008) 10111017.
[65] M.F. Grosso, H.O. Mosca, G. Bozzolo, Intermetallics 18 (2010) 945953.
[66] G. Trambly de Laissardiere, D. Nguyen Manh, L. Magaud, J.P. Julien, F. CyrotLackmann, D. Mayou, Phys. Rev. B 52 (1995) 79207933.
[67] Q. Yao, Y. Wang, Y.H. Zhu, Physica B 405 (2010) 27532756.
[68] G.J. Wagner, Sandia National Lab, (2011), private communication.

You might also like