Beruflich Dokumente
Kultur Dokumente
F UYANG T IAN
Doctoral Thesis
School of Industrial Engineering and Management, Department of
Materials Science and Engineering, KTH, Sweden, 2013
ISBN 978-91-7501-899-7
Materialvetenskap
KTH
SE-100 44 Stockholm
Sweden
skap fredagen den 29 Nov. 2013 kl 10:00 i B2, Kungliga Tekniska Hogskolan,
Brinellvagen
23, Stockholm.
c Fuyang Tian, 2013
Tryck: Universitetsservice US AB
Abstract
Ab initio theory provides a powerful tool to understand and predict the behavior of
materials. This thesis contains both of these aspects. First we use ab initio alloy theory
to investigate a new kind of complex alloy (high-entropy alloy). Second we introduce a
novel potential (interlayer potential), which can be extracted from ab inito total energy
ii
calculated single- and polycrystal elastic parameters exhibit strong composition and
crystal structure dependence. Based on the present theoretical findings, it is concluded
that alloys around the equimolar NiCoFeCrAl composition have superior mechanical
performance as compared to the single-phase regions.
Many modern materials and material systems are layered. The properties related to
layers are connected to interactions between atomic layers. We introduce the interlayer
potential (ILP), a novel model potential which fully describes the interaction between
layers. The ILPs are different from the usual interatomic potentials which present inter
action between atoms. We use the Chen-Mobius
inversion method to extract the ILPs
from ab initio total energy calculations. The so obtained ILPs can be employed to investigate several physical parameters connected with the particular set of atomic layers,
e.g. surface energy, stacking fault energy, elastic parameters, etc.
As an application, we adopt the supercell method and the axial interaction model in
connection with the ILPs to calculate the stacking fault energy along the fcc 111 direction, including the intrinsic stacking fault energy, extrinsic stacking fault energy and
twin stacking fault energy as well as the interactions between the intrinsic stacking
faults. We find that the data derived from ILPs are consistent with those obtained in
direct ab initio calculations. Along the fcc 111 direction, we study the surface energy
and surface relaxation using the ILPs. The phonon dispersions are also described.
Our conclusions are as follows
the EMTO-CPA ab initio alloy theory can be used to understand and predict the
fundamental properties of multicomponent alloys.
inversion method may pro the interlayer potentials based on the Chen-Mobius
vide a new way to investigate the properties related to layers in layered materials,
Preface
List of included publications:
I Empirical design of high-entropy alloys with optimal properties
Fuyang Tian, Lajos Karoly Varga, Nanxian Chen, Jiang Shen, and Levente Vitos,
Journal of Alloys and Compounds (under review) (2013).
II Ab initio investigation of high entropy alloys of 3d elements
Fuyang Tian, Lajos Karoly Varga, Nanxian Chen, Lorand Delczeg, and Levente
Vitos,
Physical Review B 87, 075144 (2013).
III Ab initio design of elastically isotropic TiZrNbMoVx high-entropy alloys
Fuyang Tian, Lajos Karoly Varga, Nanxian Chen, Jiang Shen, and Levente Vitos,
Journal of Alloys and Compounds (under review) (2013).
IV Structural stability of NiCoFeCrAlx high-entropy alloy from ab initio theory
Fuyang Tian, Lorand Delczeg, Nanxian Chen, Lajos Varga, Jiang Shen, and Levente Vitos,
Physical Review B 88, 085128 (2013).
V A novel potential: the interlayer potential for the fcc (111) plane family
Fu-Yang Tian, Nan-Xian Chen, Jiang Shen, and Levente Vitos,
Journal of Physics: Condensed Matter 24, 045001 (2012).
VI Interlayer potentials for fcc (111) planes of Pd-Ag random alloys
Fu-Yang Tian, Nan-Xian Chen, Lorand Delczeg, and Levente Vitos,
Computational Materials Science 23, 045006 (2012).
Comment on my own contribution
I All experimental collections, 80% data analysis; the manuscript was written jointly
(70%).
II All calculations, data presentation, literature review; the manuscript was written
jointly (75%).
III All calculations, data presentation, literature survey; the manuscript was written
jointly (80%).
IV All calculations, data presentation, literature survey; the manuscript was written
jointly (90%).
vi
V All calculations, data presentation, literature survey; the manuscript was written
jointly (50%).
VI All calculations, data presentation, literature survey; the manuscript was written
jointly (75%).
Contents
Preface
Contents
1
vi
Introduction
1.1
Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2
High-entropy alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3
Interlayer potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 Ab initio theory
2.1
Many-particle problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2
2.2.1
Exchange-correlation approximations . . . . . . . . . . . . . . . . .
2.2.2
2.3
3
14
3.1
General considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2
3.3
3.4
3.5
Elastic anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
19
CONTENTS
vii
4.1.1
4.1.2
4.1.3
4.2
4.3
4.4
Hardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.2
5.3
5.4
28
Assessing CPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.1.1
CPA versus SC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.1.2
Partial CPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
CuNiCoFeCrTix HEAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.2.1
Equilibrium volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.2.2
Magnetic structures . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.2.3
Elastic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
TiZrNbMoVx HEAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.3.1
Structural properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.3.2
Ductile behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.3.3
Elastic isotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
NiCoFeCrAlx HEAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.4.1
Phase transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.4.2
Elastic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Interlayer potentials
50
6.1
6.2
Process of inversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.3
6.4
6.3.1
6.3.2
CONTENTS
viii
7
59
Equilibrium properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
7.1.1
7.1.2
7.2
7.3
7.4
Surface energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
7.5
Phonon dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
70
Acknowledgements
71
Bibliography
72
Chapter 1
Introduction
1.1 Overview
In this thesis, we discuss the structural and elastic properties of high-entropy alloys
1.2
High-entropy alloys
The design of conventional alloys is based on one or two principal elements, and some
minor elements are added to modify their microstructure and properties, for examples,
steel and aluminum alloys.
In the early 2000s, a simple solid-solution phase was discovered in the equimolar multicomponent alloys [1, 2]. Yeh et al. named the new type of metals high-entropy alloys (HEAs). These metastable solid solutions are composed of at least four equimolar
or near-equimolar metallic elements and are stabilized by the mixing entropy Smix
= kB lnn (kB is the Boltzmann constant and n is the molar fraction of equimolar alloy
components) [1]. Due to the unique microstructures and special properties, such as
high strength, good wear, corrosion resistance, oxidation, hydrogen storage, diffusion
barrier and even biomedical applications, HEAs have attracted a rapidly increasing attention in the scientific community. The amorphous phase and nanostructure have also
been reported [3]. Nevertheless, our discussions of HEAs are based on the single-phase
solid solutions, i.e. bcc or fcc crystallographic structure.
1
CHAPTER 1. INTRODUCTION
In the experimental area, researchers used different tools, including traditional casting, mechanical alloying, sputtering, splat-quenching, etc., to obtain the HEAs with
different alloying elements and then to investigate the corresponding microstructures
and mechanical, thermal, and electronic performances [325]. With the appearance of a
large number of different-type of HEAs, one could establish general rules of physics of
HEAs. Such empirical assessments were based on the regular solid solution theory, and
made use of the atomic mismatch, mixing entropy and enthalpy, electronegativity, and
valence electron concentration [2631].
On the other hand, atomistic simulation methods were applied to study the features
connected to the microstructure of the HEAs. For instance, del Grosso et al. employed
the Bozzolo-Ferrante-Smith (BFS) method for alloys to investigate the transition to the
high entropy regime for alloys with refractory elements [32]. Based on ab initio simulation combined with the supercell technique, it was reported that the bcc phase has
partial ionic bonding between Al and other transition metals for the series of HEAs
Alx Cry CoNiFe (x = 1, 1.5, 2, 2.5, and 3, y = 1 and 2) [33]. Zhang et al. used the sublattice
model, supported by first-principles total energy calculations, to explore the possibility
of forming fcc HEAs of CoFeMnNiM and CoFeMnNiSmM (M=Cr, Zn, Ru, Rh, Pd, Re,
Os, Ir and Pt) [34].
At present, single solid-solution HEAs are mainly divided into three types. Namely,
the HEAs with fcc structure are composed of 3d transitional elements, for example
NiCoFeCrx (x = 0.5 1.15) [35], Ni-Co-Fe-Cr-Mn [2], and Cu-Ni-Co-Fe-X (X=Cr, V,
Ti, Mn) [1, 26, 36, 37] etc.; the HEAs with bcc structure are mainly consist of refractory elements, for instance, Ti-Zr-Hf-Nb(V)-Ta(Nb), Ti-Zr-V-Nb-Mo or Ti-Zr-Nb-Mo-V
etc [10,1315,3843]; for the HEAs with the fcc+bcc phase, the main reason of the duplex
phase is the addition of the bcc (fcc) stabilizer Al (Ni) or the increase of the bcc stabilizer
Cr (Co) [1, 25], for instance CuNiCoFeCrAlx (x = 0 3.0) [1, 4, 44], Cu0.5 NiCoFeCrAlNix
[25], NiCoFeCrAlx (x = 0 3.0) [10, 1820, 4552], CuNi2 FeCrAlx [28, 53], etc.
Despite of the single solid-solution phase characteristic to HEAs, it is particularly difficult to use conventional ab initio atomistic simulation methods to investigate these
systems. That is because HEAs are chemically and often also magnetically disordered
multicomponent extended solid solutions. To assess the performance of standard alloy theory based on ab initio DFT [54, 55] for the case of HEAs first we employ the
EMTO method [5659] in combination with the CPA [6063] and investigate the basic bulk properties of the HEA based on 3d transition metals. Using NiCoFeCr as a
four-component model system and NiCoFeCrAl0.2667 alloys we establish the accuracy
of the single-site mean-field approximation by comparing the CPA results with those
generated by the supercell technique.
Then the three-types HEAs are investigated. We select CuNiCoFeCr, NiCoFeCrTi and
CuNiCoFeCrTix (x = 0 0.5, 1.0) HEAs to study the bulk properties of the 3d-HEAs
with the fcc structure. Taking TiZrVNb and TiZrNbMoVx (x = 0 1.5) as examples, we
study the equilibrium volumes and elastic properties of refractory HEAs adopting the
bcc phase. NiCoFeCrAlx (x = 0 2.0) HEAs are used to study the HEAs with the phase
transformation. We focus on the equation of state, electronic structure, magnetic state,
and elastically properties.
CHAPTER 1. INTRODUCTION
Mobius
inversion method [70]. Previously, the Chen-Mobius
inversion has successfully been applied to extract the interatomic potentials from ab inito calculations for
rare earth-transition intermetallic compounds [71], ionic crystals [72], compound semiconductors [73], transition metal carbides or nitrides [74, 75] and metal/ceramic interfaces [76] as well as metal/SiC interface [77]. Recently, Yuan et al. proposed the lattice
inversion embedded-atom-method (LI-EAM) with Chen-Mobius
inversion method and
improved effectively the embedded-atom-method (EAM) potentials of metals [78].
Although one can in principle extract the model potentials from the experiments, such
data are usually obtained at different experimental conditions and thus includes some
degree of uncertainty. On the other hand, modern ab initio calculations can provide
total energies for complex alloys and thus can be used to derive reliable ILPs. In the
present thesis, we use the CASTEP [79] and EMTO [63] computational tools to obtain
the interlayer energy versus the interlayer distance for metals and alloys.
In fcc metals and alloys, there are several physical properties associated with the fcc
(111) planes, such as the elastic constant c44 , stacking fault energy, and surface energy
or the phonon dispersion, which are all basic and important quantities in materials science. Taking pure metals Al, Ni, Cu, Ag, Au and Pd as well as random alloy Pd-Ag as
examples, we discuss the properties related to layers, including the stacking fault energy (intrinsic fault, extrinsic stacking fault and twin fault), the interaction between two
intrinsic stacking faults, surface energy, surface relaxations, and the phonon dispersion.
Chapter 2
Ab initio theory
In the field of computational materials science, electronic structure calculation has become a very useful tool. It is often called ab initio (first principles) calculation. Many
physical or chemical properties of materials can be predicted directly from the solution
of the fundamental equations for the electrons.
The electronic and nuclei system in a material can be described the many-particle Schrodinger
equation
b = E,
H
(2.1)
where is the many-particle wave function, = (r1 , r2 , rN , R1 , R2 , , RM ) which
is a function of all the positions (rN for electrons, RM for ions) in the system. E is the
eigenvalue of the Hamiltonian
2
elec
nucl
elec
nucl
~2 2
1 e2
1 e2 Zi Zj
~2 Rj
b
H =
ri
+
+
2me i
2 j Mj
2 i=j |ri rj | 2 i=j |Ri Rj |
elec
nucl
e2 Zj
,
|ri Rj |
(2.2)
describing electrons with mass me and ions with mass Mj . Here ~ = h/2, h is Planck
constant, Zi and Zj are the nuclear charges. The first two terms represent the kinetic
energy of nuclei and electrons, respectively. The last term is the Coulomb interaction
between the electrons and the nuclei, while the other two terms account for the electronelectron and the nucleus-nucleus interactions, respectively.
5
For solid systems composed of thousands of nuclei and electrons we have to make
approximations for Eq. 2.2 in order to find the eigenvectors and eigenvalues for this
b Firstly the nuclei are far more massive than the electrons and their
Hamiltonian H.
velocities are therefore relatively low in comparison. Therefore, one may neglect the
second term in Eq. 2.2. This is called the Born-Oppenheimer (BO) approximation. We
can use BO approximation to separate the time scales of electron and atom motions and
thus to treat the terms in dealing with the electron states separately from the ones dealing with the atomic nuclei. Further, the BO approximation allows us to recast the last
term in Eq. 2.2 describing the Coulomb interaction between the atomic nuclei and the
electrons as an external potential acting on the electrons. Hence Eq. 2.2 can be simplified
as
elec
elec
elec
nucl
1 e2
~2 2
e2 Zj
b
ri +
H =
2me i
2 i=j |ri rj |
|ri Rj |
i
j
(2.3)
= Te + Vee + Vext ,
where Te is the kinetic energy of electron, Vee for electron-electron interactions, and the
last term Vext is the external potential. So within the framework of BO approximation,
the many-particle electron-nucleus problem has been reduce to n electron-electron problem. In the following sections, we will use atomic Rydberg units (~ = 2me = e2 /2 = 1).
Etot = Ts +
1
vext (r)n(r)d r +
2
3
n(r)n(r ) 3 3
d rd r + Vnn + Exc ,
|r r |
(2.4)
where Ts is the kinetic energy of a system of noninteracting electrons, the three middle terms represent the electron-nucleus, electron-electron, nucleus-nucleus electrostatic
energies, and the last term Exc is the exchange-correlation energy which can be decomposed into its exchange and correlation parts (Exc =Ex +Ec ). The corresponding
(2.5)
n(r )
dr + xc (r),
|r r |
(2.6)
(2.7)
ki are the one electron energies and ik the corresponding Kohn-Sham orbitals. For
solving Eq. 2.6 one can use fully-relativistic, scalar-relativistic or non-relativistic techniques. Because the exact form of xc is not known, we have to adopt different approximations. The local density approximation (LDA) [80] and the generalized gradient approximation (GGA) [81] are the two most widely applied approximations for the
exchange-correlation terms. There serval methods for solving the Kohn-Sham equations, of which full-potential, pseudopotential, and muffin-tin methods are extensively
applied in the calculations of materials .
In our work, we use the EMTO and the CASTEP computational tools. In following
sections, we introduce the two versions of GGA (Perdew-Burke-Ernzerhof (PBE) [81]
and PBEsol [82]) as well as the ultrasoft pseudopotentials (USP) and the EMTO method.
LDA
3
Exc (n) = n(r)LDA
(2.8)
xc (n(r))d r,
where LDA
= LDA
+ LDA
, in which, LDA
is the exchange-correlation energy of homogexc
x
c
xc
is the exchange energy per electron
neous electron system expressed per electron, LDA
x
LDA
and c is the correlation energy per electron.
In order to better describe an inhomogeneous system, the gradient of the electron density should also be taken into account. This results in the so-called GGA functional
family. Compared to the LDA, the exchange energy of GGA is
GGA
(2.9)
(n)FxGGA (s),
Ex (n) = d3 rnLDA
x
where the enhancement factor FxGGA has the asymptotic behavior
(2.10)
|n|
,
2kF n
(2.11)
kF =(3 2 n)1/3 is the Fermi wave vector and is the parameter of gradient expansion.
The GGA correlation term is
EcGGA (n)
d3 rn(r){LDA
(n) + t2 (r) + },
c
(2.12)
|n|
,
2kT F n
(2.13)
GGA
Exc
(n , n )
where
=
d3 rnLDA
xc (n)Fxc (rs , , s),
(2.14)
n n
.
n + n
(2.15)
The PBE functional is nowadays the most commonly used functional for solid-state calculations. There are no empirical parameters in PBE. The PBEsol has the same analytical
form as the PBE. The differences between PBE and PBEsol are the values of and .
In the USP plane-wave total energy method, the accuracy of the pseudopotential is of
particular importance. Vanderbilt introduced USPs in 1990 [83]. USPs are optimized to
the plane-wave solid-state calculations and needs the lowest possible cutoff energy for
the plane-wave basis set.
The general form of the pseudopotential is expressed as
VNL =
|lmVl lm|,
(2.16)
lm
where Vl is the pseudopotential for angular momentum l, |lm are the spherical harmonics.
In the USP plane-wave method, the nonlocal potential VNL is given as
(0)
I
VNL =
Dnm
|nI m
|,
(2.17)
n,m,I
where the projectors and coefficients D(0) characterize the pseudopotential and differ
for different atomic species. For the derivations of and D(0) , one can refer to Ref.
[83, 84]. The index I refers to an atomic site. The valence charge density is defined as
n(r) =
|i (r)|2 +
I
QInm (r)i |nI m
|i ,
(2.18)
nmI
where are the pseudo-wave functions and Q(r) are the augmentation functions that
are strictly localized in the core regions, viz.
Qnm (r) = n (r)m (r) n (r)m (r),
(2.19)
where (r) are all-electronic wave functions. The are constructed from (r) and satisfy
i |S|j = ij ,
(2.20)
I
qnm |nI m
|,
(2.21)
nmI
(2.22)
10
H = Ts + Vef f (r) +
I
I
Dnm
|nI m
|,
(2.23)
nm
(0)
I
(2.24)
Dnm = Dnm + drVef f QInm (r).
In the USP plane-wave method, the eigenvalues from the pseudo potentials are equal
to that from all-electron potentials. Their corresponding orbitals match exactly outside
the core radii. Because the scattering properties are correct at each reference energy, the
transferability can be systematically improved by increasing the number of such energies. The valence charge density is precisely equal to the all-electron valence density in
the reference configuration [79].
|vR (r R) v0 |,
(2.26)
RL
a
a
.
(j , rR )vRL,j
RL
(2.27)
11
a
The expansion coefficients, vRL,j
are determined in a way that j (r) is a solution for Eq.
2.6 in the entire space.
The exact muffin-tin orbitals are constructed using different basis functions inside the
potential sphere (r R < sR ) and in the interstitial region (r R > sR ), herein sR is
the radius of the potential sphere centered at site R. The basis functions used are called
partial waves (aRL ) inside the sphere and screened spherical waves (aRL ( V0 , r-R)) in
the interstitial zone.
Inside the potential sphere, the basis functions aRL are constructed from solutions of
(2.28)
a
The normalization factor NRL
() assures a proper matching at the potential sphere boundary to the basis function outside of the potential sphere.
In the interstitial region the screened spherical waves are solutions of the free electron
Schrodinger
equation. The boundary conditions for the free electron Schrodinger
equation are given in conjunction with non-overlapping spheres, called hard spheres, centered at lattice site R with radius aR . The screened spherical waves are just defined as
being free electron solutions which behave as real harmonics on their own a-spheres
centered at site R and vanish on all the other sites.
These basis functions, partial waves and screened spherical waves, must join continuously and differentiable at aR . This is implemented using additional free electron wave
a
functions (RL
(, aR )), by which the connection between the screened spherical waves
and the partial waves is obtained. It joins continuously and differentiable to the partial
wave at sR and continuously to the screened spherical wave at aR . Because of aR < sR ,
the additional free-electron wave function should be removed, which is realized by the
so-called kink-cancelation equation.
In the calculation of the total energy, the EMTO method employs the Full Charge Density (FCD) technique, which not only improves the calculation efficiency but also ensures total energies with an accuracy similar to that of the full-potential methods. The
total charge density is obtained by summations of the one-center densities, which may
be expanded in terms of real harmonics around each lattice site
n(r) =
nR (r R) =
(2.29)
RL
The total energy of the system is obtained via the FCD technique using the total charge
density. The space integrals over the Wigner-Seitz cells in Eq. 2.4 is solved via the shape
function technique. The FCD total energy is expressed as
12
Etot = Ts [n] +
(2.30)
where Ts [n] is the kinetic energy, FintraR is the electrostatic energy due to the charges
inside the Winger-Seitz cell, ExcR is the the exchange-correlation energy, and Finter is the
electrostatic interaction between the cells (Madelung energy).
Coherent potential approximation
The Coherent Potential Approximation (CPA) was introduced by Soven for the electronic structure problem and by Taylor for phonons in random alloys [85, 86]. Later,
Gyorffy formulated the CPA in the framework of the multiple scattering theory using
the Green function technique [87]. In this theory, the real atomic potential is replaced
by an effective (coherent) potential constructed from real atomic potentials of the alloy
components. The impurity atoms/alloy components are then embedded into this effective potential. Two main approximations are applied. First, it is assumed that the local
potentials around a certain type of atom from the alloy are the same, i.e. the effect of
local environments is neglected. Second, the system is replaced by a monoatomic set-up
described by the site independent coherent potential P .
Disordered local magnetic moment model
Disordered local magnetic moment (DLM) model [88, 89] is a approach to simulate the
paramagnetic state of random alloys using CPA . The DLM model is expected to correctly account for the random distribution of the local magnetic moments of the PM
state well above the magnetic transition temperature, where the local magnetic moments show negligible short range order. According to that model, an alloy component
M of concentration m is presented by its spin-up () and spin-down () counterparts
assumed to be distributed randomly on the underlying sublattice, i.e. each magnetic
alloy components are treated as Mm Mm/2 Mm/2 . For example, NiCoFeCr is described
as a quasi-eight component random solid solutions, viz. Fe0.125 Fe0.125 Cr
Ni0.125 Ni0.125 Co0.125 Co0.125 .
0.125 Cr0.125
EMTO-CPA
Today, CPA represents the most efficient alloy theory for the electronic structure calculations in multicomponent random solid solutions. The CPA has been successfully
implemented in the EMTO method [57]. The single-site nature of the CPA limits its
applicability to systems with negligible short-range order and local lattice relaxation effects. Nevertheless, it turned out that the EMTO-CPA method can accurately capture
13
the structural energy differences and trace energy changes related to lattice distortions
in complex alloys.
The above details about the EMTO-CPA method are from Ref. [63].
Chapter 3
Elastic properties of solids
The shape of materials changes when the external stress is applied. This deformation
is often described in terms of stress or force per unit area and strain or displacement
per unit distance. In studying of elastic properties, the stress-strain relation follows the
Hookes law, i.e. the deformation of materials is reversible.
For a single crystal, the elastic constant c or elastic compliance s are used to describe the
elastic properties. For an isotropic and quasi-isotropic materials in a statistical sense, the
elastic properties can be completely described by any two of the polycrystalline elastic
moduli, the bulk modulus (B), the shear modulus (G), and the Youngs modulus (E) as
well as the Possion ratio (v). Using averaging methods, we may obtain the polycrystalline elastic moduli from the single crystal elastic constants.
ij =
cijkl kl ,
(3.1)
k,l=1
where i, j, k, and l are indices running from 1 to 3. cijkl are called elastic constants.
Eq. 3.1 indicates that the number of coefficients cijkl is 81. Because cijkl = cklij = cjikl =
cijlk , there are at most 21 different elastic constants, which can be arranged in a 6 6
matrix. This matrix is symmetric, i.e. c =c . The relations between c and cijkl are
listed in Table 3.1.
Eq. 3.1 can be rewritten as
=
=1
14
c .
(3.2)
15
i, j or k, l
or
11 22
1 2
33
3
23 or 32
4
13 or 31
5
12 or 21
6
(3.3)
s ,
=1
3.2
The adiabatic elastic constants are the second order derivatives of the internal energy
with respect to the strain tensor e, viz.
c =
1 2E
,
V e e
(3.5)
where and are listed in Table 3.1, E represents the internal energy.
At volume V , we may obtain the elastic constants by straining the lattice and evaluating the total energy changes due to the stain as a function of its magnitude. Since the
change of total energy is usually more larger with the volume than with a general strain,
we choose the applied volume conserving, except for the bulk modulus. By choosing
volume conserving strains we can separate two contributions to the total energy. Using isochoric strains we assure the identity of our calculated elastic constants with the
stress-strain coefficients. We denote by e1 , e2 , ..., e6 the elements of strain matrix, i.e.
e1
D(e) = 12 e6
1
e
2 5
1
e
2 6
e2
1
e
2 4
1
e
2 5
1
e
2 4
e3
(3.6)
16
(3.7)
where E(0) is the energy of the undistorted lattice and O(e3 ) stands for the terms proportional to ek with k 3.
For the volume conserving deformation, the criterion is det(D(e) + I) = 1, where I is the
3 3 identity matrix. As a result, the distortion matrix can be rewritten as a function of
a single parameter, and result in a particular combination of the elastic constants [63].
s11 =
s11
s44
(3.8)
The two cubic elastic constants (c11 and c12 ) are derived from the bulk modulus
B=
c11 + 2c12
3
(3.9)
c11 c12
.
2
(3.10)
The bulk modulus was extracted from the equation of state described by a Morse function [91] fitted to the total energy calculated for seven different volumes around the
equilibrium, i.e.
P
2 E(V )
=V
,
(3.11)
V
V 2
where P is the applied hydrostatic pressure, V represents volume. In this thesis, the
equilibrium volume is often expressed in terms of average Wigner-Seitz radius w.
B = V
17
In the calculations of two cubic shear elastic moduli, c and c44 , we make use of the
following volume conserving orthorhombic and monoclinic deformations
1 + o
0
0
1 o
0
0
0
1 m
0 and m 1
1
0
0
1 2
o
0
0 ,
1
2
1m
(3.12)
2
+
which lead to the energy change E(o ) = 2V c o2 + O(o4 ) and E(m ) = 2V c44 m
4
O(m ). Both energies can be computed for six distortions, = 0.00, 0.01, , 0.05.
For the cubic crystal, the mechanical stability criteria on the elastic constants are
c > 0, c44 > 0, and B > 0.
(3.13)
The Cauchy pressure (c12 c44 ) is used as an indicator of the brittle-ductile behavior.
Negative (c12 c44 ) has been associated with the covalent nature of the metallic bond and
is characteristic to brittle alloys, whereas positive (c12 c44 ) indicates metallic character
and enhanced ductility [92].
3.4
An isotropic polycrystalline system is described by the bulk modulus B and the shear
modulus G. For a cubic lattice, the polycrystalline bulk modulus is identical with the
single-crystal bulk modulus. For the shear modulus, we used the arithmetic Hill average
(GR + GV )
G=
,
(3.14)
2
where the Voigt and Reuss bounds are given by [90]
GR =
and GV =
(3.15)
The Youngs modulus E and Poisson ration are connected to B and G by the relations
E=
9BG
3B + G
and =
3B 2G
.
2(3B + G)
(3.16)
The B/G ratio is used as an indicator of the brittle-ductile behavior. High values of
B/G (>1.75) indicate that a material is ductile, while low values are associated with
brittleness. The may also be used to predict the brittle-ductile behavior [93]. It has
been reported that bulk metallic glasses with > 0.31 are ductile [94].
18
c44
.
c
(3.17)
GV GR
,
GV + GR
(3.18)
which is used as another measure of the elastic anisotropy. Elastically isotropic materials
have AVR = 0 and AZ = 1. We note that large AVR indicates large uncertainty in the
predicted shear and Youngs moduli and Poisson ration.
In an arbitrary direction [hkl], the E can be given in thems of sij [90]
E[hlk]1 = s11 + (2s11 2s12 s44 )N 4 ,
(3.19)
and for the shear modulus G(hkl) referring to torsion around [hkl], shear occurs in all
directions in the plane (hkl), then the general expression of G can defined as
G(hlk)1 = s44 + 2(2s11 2s12 s44 )N 4 ,
(3.20)
(3.21)
where
and n1 , n2 , n3 are direction consines for the direction [hlk], i.e.
h
,
+ k 2 + l2
l
=
,
2
h + k 2 + l2
k
=
.
2
h + k 2 + l2
n1 =
n2
n3
h2
(3.22)
Chapter 4
Single phase high-entropy alloys
As single-phase solid solutions, most of HEAs has a single crystallographic structure
(bcc or fcc) or the mixture of bcc and fcc phases. In order to distinguish the crystal
structures (bcc and fcc) and understand the effect of different phases on the mechanical
properties, especially on hardness, in this chapter we show the structural properties
and hardness of single-phase HEAs via the standard solid-solution theory including
the average equilibrium volume (wmix ), atomic size difference (), the average valence
electron concentration (VEC), and hardness (Hv).
We may estimate the equilibrium volume of an alloy using Vegards rule. In order to
be able to compare different lattices and account for elements adopting different parent
lattices, here we use the Wigner-Seitz radius rather than the lattice parameter. Accordingly, the average Wigner-Seitz radius of the solid solutions is given by
wmix =
ci wi ,
(4.1)
where n stands for the number of alloy components, ci is the atomic percent of the
ith component, wi is the Wigner-Seitz radius of the ith component obtained from the
experimental data collected, e.g., in Ref. [63].
The average valence electron concentration (VEC) is calculated as
VEC =
ci (VEC)i ,
(4.2)
where (VEC)i is the valence electron concentration of the ith alloy element. The VEC
counts for the total electrons including the d-electrons in the case of transition metals.
The atomic size difference () is defined coventionally as
19
20
v
u n
u
ri
ci (1 )2 ,
= 100t
r
i=1
where r =
4.1
i=1 ci ri ,
(4.3)
According to the available experiments, we divide the one-phase HEAs into three types.
One type is about the 3d HEAs with fcc structure, the second is the refractory HEAs with
bcc structure, and the last is about the HEAs with the duplex fcc-bcc phase.
4.1.1
From Table 4.1, we can see that HEAs with the fcc phase are mainly composed of late
3d transition elements. The calculated wmix is consistent with the experimental w. The
largest VEC is 9.50 for CuNiCoFe HEA [95]. The atomic size difference is smaller than
5%, except for CuNiCoFeTi with 1 = 6.50% (2 = 6.13%) [51], Cu0.75 NiCoFeCrTi0.5 Al0.25
with 1 = 5.40% (2 = 5.03%) [96], and NiCoFeMnCrNb with 1 = 5.49% (2 = 6.03%) [2].
According to Senkovs data [97], 1 is the smallest for the ternary equimolar NiFeCr,
whereas the smallest 2 obtained from [98] is 0.92 for the equimolar CuNiCoFeMn and
NiCoFeMnCr HEAs. Note that Vickers hardness in dendrite are very different from that
in interdendritic for NiCoFeMnCrNb and NiCoFeMnCrV HEAs [2].
21
Table 4.1. Structural properties for HEAs with a fcc structure, w is the Wigner-Seitz radius
(Bohr) converted from experimental lattice parameters via 4 43 w3 = a3 for fcc structure
and 2 43 w3 = a3 for bcc structure, a is the cubic lattice parameter; wmix is the average
Wigner-Seitz radius (Bohr); VEC stands for the average valence electron concentration
(e/a); is the atomic size difference (%), 1 is calculated based on the data of Ref. [97], 2
is calculated from Ref. [98]; Vickers hardness Hv (HV). The corresponding sources are
listed in the last column.
HEAs
CuNiCoFeCr
CuNiCoFeCrTi0.5
CuNiCoFeCrAl0.5 V0.2
CuNiCoFe
CuNiCoFeMn
CuNiCoFeV
CuNiCoFeTi
CuNiFeCr
CuNiFeCrMo
CuNi2 FeMn2 Cr
Cu0.75 NiCoFeCrTi0.5 Al0.25
Cu0.75 NiCoFeCrAl0.25
Cu0.5 NiCoFeCrAl0.3
Cu0.5 NiCoCrAl0.5 Fe2
Cu0.5 NiCoCrAl0.5 Fe3
Cu0.5 NiCoCrAl0.5 Fe3.5
NiCoFe
NiCoFeCr0.5
NiCoFeCr0.6
NiCoFeCr0.7
NiCoFeCr0.8
NiCoFeCr0.9
NiCoFeCr0.95
NiCoFeCr1.05
NiCoFeCr1.10
NiCoFeCr1.15
NiCoFeCr
NiCoFeCrMo0.3
NiCoFeCrMo0.1 Al0.3
NiCoFeCrTi0.1 Al0.3
NiCoFeCrTi0.3
NiCoFeMnCr
NiCoFeMnCrNb
NiCoFeMnCrV
NiCoFeCrPd
NiCoFeCrPd2
Ni1.5 Co1.5 FeCrTi0.5
Ni1.5 Co1.5 FeCrTi0.5 Mo0.1
NiFeCr
w
2.643
2.667
2.648
2.651
2.667
2.676
2.653
2.650
2.638
2.639
2.639
2.640
2.641
2.641
2.642
2.642
2.642
3.577
2.659
2.659
2.658
2.651
2.673
2.644
2.693
2.738
2.651
wmix
2.647
2.670
2.683
2.638
2.673
2.721
2.656
2.710
2.699
2.663
2.666
2.677
2.676
2.675
2.627
2.709
2.637
2.638
2.639
2.640
2.641
2.642
2.643
2.643
2.642
2.661
2.672
2.675
2.670
2.681
2.708
2.673
2.677
2.651
VEC
8.80
8.36
8.16
9.50
9.00
8.60
8.40
8.75
8.20
8.43
8.00
8.40
8.21
8.00
8.00
8.00
9.00
8.57
8.50
8.43
8.37
8.31
8.28
8.22
8.20
8.17
8.25
8.09
7.84
7.80
7.88
8.00
9.17
9.17
8.60
8.83
8.09
8.05
8.00
1
1.07
4.82
4.15
1.14
3.18
2.20
6.50
1.15
3.58
3.57
5.40
3.43
3.58
4.08
3.84
3.74
0.33
0.31
0.31
0.31
0.30
0.30
0.30
0.30
0.30
0.30
0.30
2.38
3.90
4.40
3.76
3.27
5.49
3.29
4.04
4.69
4.89
4.92
0.26
2
hardness
Ref.
1.07
133
[1, 5]
4.46
[5, 96]
3.87
202
[99]
1.03
[95]
0.92
[36]
2.88
[26]
6.13
[51]
0.96
[100]
4.10
263
[6]
0.99
[101]
5.03
[96]
3.00
[27]
3.30
[102]
3.67
[25]
3.41
[25]
3.30
[25]
0.75
124
[37, 103]
0.95
[35]
0.97
[35]
0.99
[35]
1.00
[35]
1.02
[35]
1.02
[35]
1.03
[35]
1.04
[35]
1.04
[35]
1.03
116
[9, 45]
2.92
210
[104]
3.74
[48, 105]
4.06
[105]
3.49
350
[106]
0.92 300/290
[2]
6.03 1031/399
[2]
2.70 1007/287
[2]
3.76
[9]
4.33
[9]
4.60
[107]
4.72
[107]
0.98
[100]
22
Table 4.2. Structural properties for HEAs with the bcc structure. See caption for
Table 4.1.
HEAs
Cu0.5 NiCoCrAl
Cu0.25 NiCoFeCrAl
WMoTaNb
WMoTaNbV
TaNbHfZrTi
TaNbVTi
TaNbVTiAl0.25
TaNbVTiAl0.5
TaNbVTiAl1.0
MoNbZrTi
MoNbZrTiV0.25
MoNbZrTiV0.5
MoNbZrTiV0.75
MoNbZrTiV
MoNbCrTiAl0.5
MoNbCrVAl0.5
MoNbCrVTiAl0.5
NbVZrTi
NbVHfZrTi
NbHfZr
w
2.683
2.990
2.962
3.167
3.094
3.150
3.244
wmix
2.717
2.709
3.000
2.964
3.169
3.003
3.002
3.001
3.000
3.090
3.074
3.059
3.046
3.035
2.940
2.887
2.917
3.066
3.115
3.239
VEC
6.09
7.38
5.50
5.40
4.40
4.75
4.65
4.56
4.40
4.75
4.76
4.78
4.79
4.80
5.00
5.22
5.00
4.50
4.40
4.33
1
2.81
5.65
2.31
3.15
4.99
3.93
3.83
3.73
3.57
5.99
6.31
6.55
6.73
6.85
5.88
5.40
5.52
7.05
7.07
5.00
2
5.44
5.13
2.27
3.21
4.01
3.53
3.42
3.33
3.16
4.92
5.24
5.48
5.65
5.77
5.11
4.90
4.90
6.78
6.75
3.69
hardness
496
445.5
525
382.6
328.5
380
Ref.
[3]
[26]
[12]
[12]
[108]
[15]
[15]
[15]
[15]
[109]
[109]
[109]
[109]
[109]
[110]
[110]
[110]
[42]
[111]
[43]
23
2.80
2.80
2.78
2.76
2.74
2.78
w
w
w
w
(fcc)
mix
2.76
(fcc)
(bcc)
mix
2.74
(bcc)
2.72
2.72
2.70
2.70
2.68
2.68
2.66
2.66
2.64
2.64
2.62
2.62
fcc
bcc
Figure 4.1. The comparison between the experimental w and the calculated wmix
for HEAs with the fccbcc transformation.
4.2
The atomic size difference is basic and important quantity for solid solution state. For
this reason, in Eq. 4.3 we use two different sets of to study distortion of structure.
One is from Ref. [97], which is applied to study glass forming ability of amorphous
metallic alloys. Another is metallic radius from Ref. [98] which is derived from the
twelve coordinated metal. Note that the large difference of between Ref. [97] and
Ref. [98] mainly results from atomic radius of Mn. Interestingly, these of HEAs with
the fcc phase are small, compared to the HEAs with the bcc structure.
24
Table 4.3. Structural properties for HEAs with the fccbcc phase transformation.
See caption for Table 4.1.
HEAs
CuNiCoFeCrAl0.3
CuNiCoFeCrAl0.5
CuNiCoFeCrAl0.8
CuNiCoFeCrAl1.0
CuNiCoFeCrAl2.3
CuNiCoFeCrAl2.5
CuNiCoFeCrAl2.8
CuNiCoFeCrAl3.0
CuNi2 FeCrAl0.2
CuNi2 FeCrAl0.4
CuNi2 FeCrAl0.5
CuNi2 FeCrAl0.7
CuNi2 FeCrAl0.8
CuNi2 FeCrAl1.0
CuNi2 FeCrAl1.2
CuNi2 FeCrAl1.5
CuNi2 FeCrAl1.6
CuNi2 FeCrAl1.8
CuNi2 FeCrAl2.0
CuNi2 FeCrAl2.2
CuNi2 FeCrAl2.5
Cu0.5 CoFeCrAl
Cu0.5 CoFeCrAlNi0.5
Cu0.5 CoFeCrAlNi1.0
Cu0.5 CoFeCrAlNi1.5
Cu0.5 CoFeCrAlNi2.0
Cu0.5 CoFeCrAlNi2.5
Cu0.5 CoFeCrAlNi3.0
Cu0.5 NiFeCrAl
Cu0.5 NiFeCrAlCo0.5
Cu0.5 NiFeCrAlCo1.5
Cu0.5 NiFeCrAlCo2.0
Cu0.5 NiFeCrAlCo3.0
Cu0.5 NiFeCrAlCo3.5
Cu0.5 NiCoFeAl0.5 Cr
Cu0.5 NiCoFeAl0.5 Cr2
Cu0.5 NiCoFeAl0.5 Cr3
Cu0.5 NiCoFeAl
Cu0.5 NiCoFeAlCr0.5
Cu0.5 NiCoFeAlCr1.0
Cu0.5 NiCoFeAlCr1.5
Cu0.5 NiCoFeAlCr2.0
phase
fcc
fcc
fcc+bcc
fcc+bcc
fcc+bcc
fcc+bcc
bcc
bcc
fcc
fcc
fcc
fcc
fcc+bcc
fcc+bcc
fcc+bcc
fcc+bcc
bcc
bcc
bcc
bcc
bcc
bcc
bcc
bcc
bcc+fcc
bcc+fcc
bcc+fcc
fcc
bcc
bcc
bcc+fcc
bcc+fcc
fcc
fcc
fcc
bcc+fcc
bcc
fcc
fcc+bcc
bcc
bcc
bcc
w
2.646
2.644
2.700
2.696
2.661
2.648
wmix
2.666
2.667
2.755
2.762
2.770
2.776
2.658
2.670
2.676
2.687
2.729
2.736
2.744
2.751
2.760
2.731
2.718
2.708
2.679
2.729
2.717
2.682
2.678
2.679
2.681
2.645
2.706
2.704
VEC
8.47
8.27
6.97
6.87
6.72
6.63
8.77
8.56
8.45
8.26
7.55
7.41
7.29
7.17
7.00
7.00
7.30
7.55
8.20
7.22
7.40
7.93
8.00
8.00
7.43
8.56
7.55
7.42
7.31
1
3.42
4.17
6.40
6.48
6.57
6.61
2.94
3.84
4.20
4.74
6.02
6.17
6.30
6.40
6.52
5.86
5.68
5.51
4.93
5.92
5.71
4.88
4.75
4.37
3.77
0.84
5.34
5.18
2
3.15
3.82
5.84
5.91
5.99
6.09
2.69
3.48
3.82
4.30
5.46
5.60
5.71
5.80
5.91
5.21
5.12
5.02
4.60
5.21
5.12
4.60
4.51
4.00
3.38
1.06
4.82
4.64
hardenss
Ref.
18020
[1, 112]
20820
[1, 112114]
27120
[1, 112]
40620
[1, 112]
600
[1]
620
[1]
655
[1]
635
[1]
160
[53]
172
[53]
270
[53]
290
[53]
320
[53]
390
[53]
520
[53]
550
[53]
557
[53]
550
[53]
570
[53]
580
[53]
600
[53]
[25]
[25]
458
[25, 114]
[25]
[25]
[25]
[25]
[25]
[25]
[25]
[25]
[25]
[25]
[25]
[25]
[25]
174
[25, 115, 116]
[25]
[25]
[25]
[25]
HEAs
Cu0.5 NiCoFeCr
Cu0.5 NiCoFeCrAl0.5
Cu0.5 NiCoFeCrAl1.0
Cu0.5 NiCoFeCrAl1.5
Cu0.5 NiCoFeCrAl2.0
NiCoFeCrAl0.25
NiCoFeCrAl0.3
NiCoFeCrAl0.375
NiCoFeCrAl0.5
NiCoFeCrAl0.75
NiCoFeCrAl0.875
NiCoFeCrAl
NiCoFeCrAl1.25
NiCoFeCrAl1.5
NiCoFeCrAl2.0
NiCoFeCrAl2.5
NiCoFeCrAl3.0
phase
fcc
fcc
bcc
bcc
bcc
fcc
fcc
fcc
fcc+bcc
fcc+bcc
fcc+bcc
fcc+bcc
bcc
bcc
bcc
bcc
bcc
2.708 7.55
2.731 7.17
2.751 6.92
2.653 2.662 7.94
2.659 2.666 7.88
2.656 2.671 7.80
25
1
0.84
4.37
5.51
6.11
6.46
3.47
3.76
4.08
5.95
6.16
6.41
6.53
6.56
2
hardenss
Ref.
1.06
174
[25, 115, 116]
4.00
[25, 27]
5.02
458
[25, 114]
5.56
[25]
5.87
[25]
3.25
1102
[45]
3.49
[48]
3.80 130.80.5
[45]
159.02
[45]
388.05
[45]
538.012
[45]
484.026
[45]
5.55
487
[45]
5.77 403/580
[45, 49]
6.04 432/615
[45, 49]
6.19 487/690
[49]
6.26 506/740
[49]
4.4 Hardness
The hardness Hv shows a maximum value as a function of the VEC. For the singlephase HEAs a similar behavior is expected. In Table 4.1-4.3, we list the data of Vickers
26
7.80
BCC
FCC
4.33
4.5
5.0
9.50
7.55
5.5
6.0
6.5
7.0
7.5
8.0
8.5
9.0
9.5
VEC
Figure 4.2. The crystal structure versus the average valence electronic concentration (VEC) for the single solid solution HEAs.
hardness of the available single-phase HEAs, but some are from macro measures, while
others are micro values. Here we employ the values of macro hardness. As shown in
Figure 4.3, the hardness as a function of VEC follows Gauss type distributions. The
maximum value is obtained at VEC 6.8. Note that the data of hardness are particularly limited under VEC = 6.20 and some hardness values were not included in this
correlation as they show large deviations for the same HEAs.
The solid solution hardening mechanism originates from the distortion of the crystalline
cell preventing the easy motion of the dislocations necessary for the plastic deformation.
This is why one can expect a monotone increase of the hardness (and of the ultimate
tensile stress , Hv 3) with the lattice deformation expressed through . Figure 4.4
shows that hardness as a function of . The scatter around an average linear relationship
is smaller for Pearsons 2 [98], compared to Senkovs 1 [97]. The increase of hardness
as a function of atomic size difference is about 107 HV/% and 90 HV/% for Pearsons
[98] and Senkovs data [97], respectively. Despite of the scattering of the hardness data,
one can approximate 100 HV/ % for the effect of atomic size difference. For the high
atomic mismatch, 6%, which is similar with the maximum (about 650 HV) shown
in Figure 4.4. Applying Hv 3, the maximal strength for one phase bcc HEA can be
estimated as 200 HV, which is a value comparable to the best known steels value.
In order to make a more fundamental assessment between the atomic size and hardness
one may make use of classical hardening theories based on the size and elastic misfit
parameters. Such approach was successfully combined with ab initio calculations in the
case of steel and aluminum alloys [118, 119].
4.4. HARDNESS
27
700
600
Hardness (HV)
500
6.80
400
300
200
100
6.0
6.4
6.8
7.2
7.6
8.0
8.4
8.8
VEC
700
600
Hardness (Hv)
500
400
300
200
100
1
Chapter 5
Structural and elastic property of HEAs
5.1 Assessing CPA
5.1.1 CPA versus SC
In order to assess the performance of CPA in the case of HEAs, we select NiCoFeCr and
set up a simple supercell (SC) with fcc underlying lattice. To mimic a homogeneous
solid solution, we distribute the four alloying elements so that they are neighbors to
each other within a conventional fcc unit cell, as shown in Figure 5.1. Similar structure for different HEAs are suggested in Refs. [120, 121]. We notice that NiCoFeCr is
found to show no tendency for long-range ordering [121]. Taking into account that the
long-range order has a rather small effect on the elastic properties of alloys [122], a direct comparison between the CPA and the SC results calculated for the present ordered
structure seems to be well founded.
The results obtained for the supercell from Figure 5.1 and those calculated for the corresponding Ni0.25 Co0.25 Fe0.25 Cr0.25 random solid solution are listed in Table 5.1. The
average SC equilibrium Wigner-Seitz radius is 2.601 Bohr, which is rather close to 2.607
Bohr obtained for solid solution using CPA. The agreement between the SC (207 GPa)
and CPA (208 GPa) bulk moduli is also excellent. For all theoretical parameters, we
find a good consistency between the CPA and SC results. In particular, the three cubic
elastic constants, c11 , c12 and c44 , obtained with the two methods differ on the average by 4%. The somewhat larger relative errors in the Zener anisotropy (c44 /c ) and
the Cauchy pressure (c12 c44 ) are still acceptable, especially if we consider that the
present supercell is the simplest periodic approximant of the four-component random
alloy considered in the CPA calculations. The good agreement seen for the shear and
Youngs modulus, Poisson ratio and polycrystalline anisotropy ratio AVR indicate that
the CPA is an efficient and accurate method to investigate the bulk properties of these
multicomponent alloys.
28
29
Figure 5.1. The supercell used to model the NiCoFeCr alloy [121].
Table 5.1. Theoretical bulk parameters for fcc NiCoFeCr alloy calculated using the CPA and
SC approximations. Listed are the equilibrium Wigner-Seitz radius w (Bohr), cubic elastic constants c11 , c12 , and c44 , and c (GPa), the Zener anisotropy AZ , the
Cauchy pressure (c12 c44 ) (GPa), the bulk modulus B (GPa), the shear modulus
G (GPa), the Young modulus E (GPa), the Poisson ratio , and the polycrystalline
elastic anisotropy ratio AVR .
Method
c11
c12
c44
c
AZ
CPA
271.0 175.0 189.3 48.0
3.9
SC
257.1 183.5 193.9 36.8
5.2
w
B
G
E
CPA
2.607 207
110
280 0.275
SC
2.601 208
101 262 0.290
(c12 -c44 )
-14.3
-10.4
AVR
0.21
0.29
30
HEAs
NiCoFeCr
CuNiCoFeCr
NiCoFeCrTi
CuNiCoFeCrTi0.1
CuNiCoFeCrTi0.2
CuNiCoFeCrTi0.3
CuNiCoFeCrTi0.4
CuNiCoFeCrTi0.5
CuNiCoFeCrTi1.0
Element
wc
we
str.
fcc
fcc
fcc
fcc
fcc
fcc
fcc
fcc
fcc
Ti (hcp)
3.04
3.053
w(EMTO) w(Expt.)
2.607
2.632 [9]
0.95%
2.628
2.643 [5]
0.57%
2.682
2.650 [7]
1.21%
2.635
2.643
2.651
2.655
2.663
2.694
Cr (B2)
Fe (bcc) Co (hcp)
2.65
2.64
2.60
2.684
2.667
2.613
wc
2.623
2.636
2.706
2.644
2.652
2.659
2.666
2.673
2.703
Ni (fcc)
2.60
2.602
we
2.642
2.647
2.724
2.655
2.663
2.670
2.677
2.684
2.715
Cu (fcc)
2.69
2.669
In Table 5.2, we list the PBE-level results from Ref. [63] as well as some experimental
data for the elementary solids in their low-temperature crystallographic phases. It has
been found that density functional theory (PBE) underestimates the hcp Ti, antiferromagnetic B2 Cr, ferromagnetic body centered cubic (bcc) Fe, ferromagnetic hcp Co and
fcc Ni, whereas for fcc Cu a weak underbinding is observed [63]. Using the Wigner-Seitz
radii for the alloy constituents, we may estimate the equilibrium volume of the HEAs
via Vegards rule. In Table 5.2, wc stands for the estimated volume based on the previous PBE-level theoretical data, and we the one obtained from the experimental data.
Rather interestingly, we gives an excellent estimate for the equilibrium radius of the
equimolar NiCoFeCr, CuNiCoFeCr, and CuNiCoFeCrTix (x = 0.1 0.5, 1.0) alloys. As
31
2.70
2.68
2.66
2.64
2.62
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
The content of Ti
shown in Figure 5.2, we can see that with increasing Ti content the equilibrium volume
of CuNiCoFeCrTix linearly increases.
32
4.0
3.5
-0.6
-0.4
-0.2
0.0
0.2
-0.6
CuNiCoFeCr
-0.4
-0.2
0.0
0.2
-0.6
-0.4
-0.2
0.0
0.2
NiCoFeCrTi
NiCoFeCr
3.0
2.5
2.0
1.5
1.0
Total
0.5
Fe
Fe
Fe
5.0
Cr
Cr
Cr
4.5
Ni
Ni
Ni
4.0
Co
Co
Co
3.5
Cu
0.0
Ti
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.6
-0.4
-0.2
0.0
0.2
-0.6
-0.4
-0.2
0.0
0.2
-0.6
-0.4
-0.2
0.0
0.2
Energy (Ry)
Figure 5.3. Total (upper panels) and partial (lower panels) density of states (DOS)
for the nonmagnetic fcc CuNiCoFeCr, NiCoFeCr and NiCoFeCrTi alloys.
The position of the Fermi level is marked by vertical dashed lines.
-0.6
-0.4
-0.2
0.0
33
0.2
-0.6
-0.4
-0.2
0.0
0.2
-0.6
-0.4
-0.2
0.0
0.2
0.0
0.2
2.0
CuNiCoFeCr
NiCoFeCrTi
NiCoFeCr
1.5
1.0
0.5
Total
0.0
2.0
1.5
Fe
Cr
Spin up
Co
1.0
0.5
0.0
-0.5
-1.0
-1.5
Spin down
-2.0
-0.6
-0.4
-0.2
0.0
0.2
-0.6
-0.4
-0.2
0.0
0.2
-0.6
-0.4
-0.2
Energy (Ry)
Figure 5.4. Total (upper panels) and Fe, Co and Cr partial (lower panels) density of states (DOS) for the paramagnetic fcc CuNiCoFeCr, NiCoFeCr and
NiCoFeCrTi alloys. In the lower panels, only the Fe , Co and Cr partial
DOS are shown. Apart from a sign difference, the partial DOS for Fe , Co
and Cr are identical to those shown in the figure.
34
For all alloys, Fe has a moderate pDOS peak located very close to the Fermi level. It
is found that DFe (EF ) is the largest among all pDOS at EF , followed by DCo (EF ) and
DCr (EF ). This distinct Fe peak at EF leads to magnetic instability in Fe sublattice. Indeed, as shown in Figure 5.4, the spin-polarized pDOS of Fe has two separate peaks:
one above the Fermi level and one below the Fermi level. These two Fe peaks hybridize
with the Cr and Co peaks, respectively. As a result of the magnetic splitting, the total
D(EF ) drops significantly in all three alloys. In nonmagnetic CuNiCoFeCr, we have
DFe (EF ) DCo (EF ) and the Co peak is close to EF . This may explain the appearance of
the local magnetic moments on the Co sites in the paramagnetic state.
Additional total energy calculations performed for the hypothetical nonmagnetic fcc 3dHEAs give 2.592 Bohr, 2.610 Bohr, and 2.675 Bohr for the equilibrium Wigner-Seitz radii
of NiCoFeCr, CuNiCoFeCr, and NiCoFeCrTi, respectively. These values are smaller
than those listed in Table 5.2, which were obtained for the paramagnetic state. Taking
into account the uncertainty associated with the experimental equilibrium volume of
NiCoFeCrTi [7, 52], we may conclude that neglecting the paramagnetism in the present
3d-HEAs worsens the agreement between theory and experiment.
227.8
219.7
207.6
198.4
174.3
2.628
2.635
2.655
2.663
2.694
2.682
0
0.1
0.4
0.5
1.0
184.5 170.9
154.6
152.6
151.7
151.0
148.6
271.0 175.0
2.607
c12
c11
127.0
165.3
160.2
150.8
142.7
125.0
189.3
c44
6.8
36.6
33.5
27.9
23.7
12.8
48.0
c
(c12 -c44 ) B
G B/G
NiCoFeCr
3.9
-14.3
207 110 1.88
CuNiCoFeCrTix
4.6
-10.7
179 91 1.97
4.8
-7.5
175 87 2.01
5.4
2.0
171 78 2.19
6.0
8.3
169 71 2.38
9.8
23.6
157 54 2.91
NiCoFeCrTi
18.7
43.9
175 47 3.72
AZ
0.67
0.25
0.26
0.31
0.34
0.48
0.21
AVR
234.0
223.1
200.4
187.1
145.4
0.376 130.3
0.282
0.288
0.303
0.313
0.346
0.275 280.0
134 [7]
55.6 [5]
98.6 [5]
76.5 [5]
E(Expt.)
Table 5.3. Theoretical Wigner-Seitz radius w (Bohr), elastic constants c11 , c12 and c44 as well as c , c44 /c and
(c12 -c44 ); the elastic moduli B, G and E as well the Poisson ratio v for NiCoFeCr, CuNiCoFeCrTix
(x = 0.0 0.5, 1.0) and NiCoFeCrTi. For reference, the available experimental Youngs moduli are
also listed.
0.2
0.0
0.2
0.4
0.6
0.8
1.0
0.4
0.6
0.8
1.0
221
11
204
187
170
156
154
12
152
150
148
168
44
154
140
126
c'
37.5
30.0
22.5
15.0
9.8
12
44
44
/c'
8.4
7.0
5.6
4.2
24
16
8
0
-8
The content of Ti
Figure 5.5. Elastic constants (unit of GPa) as a function of the content of Ti x for
CuNiCoFeCr and CuNiCoFeCrTix (x = 0.1 0.5, 1.0).
450
400
NiCoFeCr
CuNiCoFeCr
CuNiCoFeCrTi
NiCoFeCrTi
350
300
E (GPa)
36
250
200
150
100
50
001
111
110
Figure 5.6. Theoretical Youngs modulus for NiCoFeCr, CuNiCoFeCr, CuNiCoFeCrTi and NiCoFeCrTi alloys as a function of direction including the three
main cubic directions.
37
Table 5.4. Wigner-Seitz radius w (Bohr) and bulk modulus B (GPa) for the refractory elements (M) and HEAs with bcc structure. wt represent the theoretical
(EMTO) Wigner-Seitz radius of pure elements, we stands for the experimental or extrapolated radius (those for Ti and Zr were extrapolated to 0 K,
Ref. [125, 126]), wt and we are the average Wigner-Seitz radii of the HEAs
estimated from the wt and we values of pure elements, respectively, according to Vegards rule.
M
Ti
we
Be
HEAs
wt
wt
we
a
3.038
TiZrVNb
3.054 3.057 3.062
3.043 96.7b TiZrNbMo
3.075 3.100 3.090
a
Zr 3.327 75.4 3.324
Bt
105.8
tured material theory would predict the lowest E to be around 100 GPa, which is still
almost double of the experimental value. For the two Ti-containing CuNiCoFeCrTi0.5
and CuNiCoFeCrTi alloys the calculated Youngs moduli differ from the reported experimental values by 90%. On the other hand, the agreement between theory and
experiment is almost perfect for NiCoFeCrTi. Such good agreement is rather unexpected since for this alloy we obtained very large anisotropy ratio. The single-crystal
Youngs modulus of NiCoFeCrTi changes significantly with direction (Figure 5.6), the
lowest value being close to 20 GPa (for <001>) and the largest around 307 GPa (for
<111>). Moreover, recent experiments show that NiCoFeCrTi is not a single fcc phase
alloy [124].
38
3.08
3.06
3.04
3.02
3.00
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
The content of V
to be the most stable one among the three close-packed lattices (fcc, bcc and hcp).
First we study the atomic radius of the refractory elements and then turn to the corresponding HEAs. In Table 5.4, we list our theoretical results and the experimental
data, along with a few former theoretical values for the refractory elements in the bcc
phase. In order to estimate the Wigner-Seitz (WS) radius (w) of bcc Ti and Zr at 0 K,
we extrapolate the high temperature data assuming a linear thermal expansion, i.e.
w(T ) = w(0 K)(1 + T), where w(0 K) is the WS radius at 0 K, w(T ) at temperature
T, and is the linear thermal expansion coefficient. Using the experimental values
wTi (1155 K) = 3.077 Bohr and wZr (1140 K) = 3.358 Bohr [125] together with the reported
thermal expansion coefficients for bcc Ti (10.9 106 K1 ) and Zr (9 106 K1 ) [126],
we get 3.038 Bohr and 3.324 Bohr for w(0 K) of hypothetical bcc Ti and Zr, respectively.
Considering these estimated and the quoted experimental values for V, Nb and Mo
(Table 5.4), we conclude that the present theory correctly describes all five refractory elements. We also notice the reasonable agreement between the present theoretical values
and those from Ref. [128].
We show the calculated equilibrium WS radii for bcc TiZrVNb and TiZrNbMoVx (x =
0 1.50) HEAs in Table 5.4 and in Figure 5.7. The only experimental equilibrium radius we could find in the literature is 3.094 Bohr [42] reported for TiZrVNb. To further
assess the theoretical volumes predicted for the present HEAs, we make use of Vegards rule and estimate the mean equilibrium WS radii for alloys using those of the
alloy components. In Table 5.4, wt stands for the estimated radius based on the present
theoretical values, and we for the one obtained from the experimental data. The rather
good agreement between wt and we confirms the high performance of the present density functional approach for the alloy constituents. It is interesting to note that for all
HEAs considered here, the calculated equilibrium WS radius wt is slightly smaller than
wt . Hence all alloys show a small but systematic negative deviation relative to Vegards
rule. Similar to CuNiCoFeCrTix , the WS radius of TiZrNbMoVx has a linear change
shown in Figure 5.7 as a function of the content of Vanadium x (x = 0 1.50).
Alloy
TiZrVNb
TiZrNbMo
TiZrNbMoV0.25
TiZrNbMoV0.50
TiZrNbMoV0.75
TiZrNbMoV1.00
TiZrNbMoV1.25
TiZrNbMoV1.50
TiZrNbMo0.8
TiZrNbMo0.8 V0.2
TiZrNbMo0.9
TiZrNbMo0.8 V0.5
c11
166.4
209.9
211.0
212.2
213.2
213.7
218.0
219.3
199.0
200.8
204.3
203.7
c12
94.7
101.0
100.6
100.3
100.3
100.7
101.9
102.2
98.7
99.0
99.5
100.0
c44
53.8
52.6
52.1
51.6
51.2
50.9
50.0
49.8
52.8
52.5
52.6
51.9
c
35.9
54.4
55.7
55.9
56.4
56.5
58.0
58.5
50.1
50.9
52.5
51.9
AZ
1.500
0.966
0.944
0.923
0.908
0.900
0.861
0.850
1.054
1.031
1.004
1.000
(c12 c44 )
41.0
48.4
48.6
48.7
49.1
49.8
51.9
52.1
45.9
46.5
46.8
48.1
VEC
4.50
4.75
4.76
4.78
4.79
4.80
4.81
4.82
4.68
4.70
4.72
4.72
B
118.6
137.3
137.4
137.6
138.0
138.5
140.6
141.2
132.2
132.9
134.4
134.6
G
45.70
53.33
53.31
53.30
53.25
53.17
53.09
53.10
51.71
51.82
52.56
51.88
E
121.1
141.7
141.6
141.7
141.5
141.1
141.4
141.6
137.2
137.6
139.5
137.9
0.330
0.328
0.328
0.328
0.329
0.330
0.332
0.334
0.327
0.327
0.327
0.329
B/G
2.604
2.575
2.579
2.581
2.591
2.608
2.648
2.658
2.556
2.565
2.557
2.594
AVR
0.01958
0.00014
0.00041
0.00076
0.00111
0.00140
0.00268
0.00317
0.00033
0.00011
0
0
Table 5.5. Theoretical elastic constants c11 , c12 , c44 and c (GPa), the Zener anisotropy AZ (c44 /c ) and Cauchy pressure (c12 c44 )
(GPa); valence electronic concentration (VEC) (e/a); polycrystalline elastic moduli B, G, E (GPa), and Poissons ratio v, the
B/G ratio and the elastic anisotropy ratio AVR of the presently considered HEAs.
40
0.3306
0.3287
2.673
B/G
2.646
2.619
2.592
51.6
50.4
49.2
12
-c
44
(GPa)
52.8
48.0
0.00
0.25
0.50
0.75
1.00
1.25
1.50
V content ( )
Figure 5.8. Theoretical Poissons ratio (v), Pugh ration (B/G), Cauchy pressure
(c12 c44 ) as a function of V content x for TiZrNbMoVx (x = 0 1.5)
alloys.
41
V content x
0.00
0.25
0.50
0.75
1.00
1.25
1.50
53.35
142
2.66
53.30
141
B
G
2.64
B/G
53.25
53.15
2.62
B/G
53.20
139
G (GPa)
B (GPa)
140
2.60
138
53.10
2.58
53.05
2.56
137
4.75
4.76
4.78
4.79
4.80
4.81
4.82
VEC (e/a)
Figure 5.9. Correlation between the bulk (B), shear (G) moduli and Pugh ratio B/G
of TiZrNbMoVx (x = 01.5) and the valence electron concentration (VEC).
42
TiZrNbMoy Vx as a function of x and y (keeping the Ti, Zr, and Nb atomic fractions to
1). Some of the results of this additional study are shown in the lower part of Table
5.5. We find that TiZrNbMoy Vx becomes almost perfectly isotropic for (x, y) = (0, 0.9)
or (x, y) = (0.5, 0.8). Taking and TiZrNbMo0.8 V0.5 as example, we show the perfect
isotropy in Figure 5.10.
Very interestingly, for both isotropic TiZrNbMo0.9 and TiZrNbMo0.8 V0.5 HEAs, the VEC
is about 4.72. On this ground, we suggest that VEC=4.72 is an important criterion for
the isotropic HEAs. For comparison, Li et al. reported that the Ti-V alloys (Gum Metals)
become elastically isotropic for VEC 4.7 [130]. Since the present results correspond
to static conditions, the above conclusion is valid only at low temperatures. Due to the
particular electronic structure of the refractory HEAs alloys, the elastic anisotropy is
expected to depend strongly on temperature.
43
80
Exp. a
Exp. b
Exp. c
Exp. d
Cal.
78
76
74
0.0
0.1
0.2
0.3
0.4
0.5
1.0
1.2
1.4
Exp. a
Exp. b
Exp. c
Exp. d
Exp. e
Cal.
1.6
1.8
3.6
bcc
82
2.0
80
2.4
78
1.2
x=1.11
76
-1.2
74
0.0
Energy (mRy)
fcc
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
Al content ( )
Figure 5.11. Upper panels: comparison between the theoretic and experimental
equilibrium volumes for fcc (x = 0.0 0.5, upper left panel) and bcc
(x = 1.0 2.0, upper right panel) FeCrNiCoAlx alloys. The quoted experimental data are a Ref. [47], b,c Ref. [45], d Ref. [51], and e Ref. [49]. Lower
panel: theoretical fcc and bcc equilibrium volumes and structural energy
difference for 0 x 2.5.
In the lower panel of Figure 5.11, we show the theoretical equilibrium volume (V ), bulk
modulus (B), and structural energy difference Et Etbcc Etfcc for the NiCoFeCrAlx
alloys as a function of Al content. The theoretical equilibrium volumes of the fcc and
bcc alloys are compared to the experimental data in the upper panels of Figure 5.11. Experimental volumes are available for the single fcc phase for x 0.5, and for single bcc
phase for x 0.9 [45,47,49,51]. Taking into account the large scatter in the experimental
data, we conclude that the present theory reproduces well the experimental trends of
V (x) for both fcc and bcc structures. Aluminum addition is found to increase the equilibrium volume of the solid solution, which is consistent with the fact that w of Al is
larger than those of the other alloy components. The slopes of w(x) and B(x) versus x
are larger in the fcc phase than in the bcc phase. This is likely to be due to the enhanced
flexibility of the bcc structure to incorporate a large substitutional element as compare
to the close-packed fcc lattice.
x=0.851
2
x=0.545
x=1.181
-2
-4
600 K
G (mRy)
0.0
44
0.1
0.2
0.3
0.4
0.5
0.4
0.5
x=0.982
2
x=0.597
x=1.229
-2
-4
300 K
0.0
0.1
0.2
0.3
x=1.11
bcc
x=0.651
fcc
x=1.277
-2
-4
0 K
0.0
0.1
0.2
0.3
0.4
0.5
Al content ( )
Figure 5.12. Comparison of the Gibbs free energies for bcc and fcc
(NiCoFeCr)1y Aly (y = 0.0 0.5) for T = 0, 300 and 600 K. Note
that y = x/(4 + x), where x is the atomic fraction of Al in NiCoFeCrAlx .
x
w(fcc) B(fcc) w(bcc)
0
2.607
207
2.626
0.10
2.611
200
2.629
0.25
2.619
197
2.634
0.30
2.622
196
2.636
0.375 2.626
194
2.639
0.50
2.632
190
2.644
B(bcc)
199
197
193
192
189
187
Et
3.937
3.408
2.523
2.380
2.021
1.566
x
1.00
1.25
1.30
1.50
2.00
2.50
B(bcc)
178
171
170
167
159
153
Et
0.228
-0.236
-0.239
-0.449
-0.773
-0.831
Table 5.6. Theoretical Wigner-Seitz radius w (Bohr), bulk modulus B (GPa), and energy difference between the bcc and fcc phases Et (mRy/atom) for NiCoFeCrAlx high entropy alloys calculated
for the fcc and bcc phases as a function of Al fraction x.
46
According to Table 5.6, the fcc structure is predicted to be more stable than the bcc one
for x = 0.0 1.0, and the bcc structure becomes stable from x = 1.25. Using a cubic
spline fit for the calculated energy points, we find that the structural energy difference
between ideal bcc and fcc lattices vanishes at 1.11 Al fraction (Figure 5.11, lower panel).
Due to the large atomic volume of Al, one may anticipate that the interatomic distance
between Al and the other elements is larger than the average bulk value. We estimated
the size of the local lattice relaxation (LLR) around the Al atoms in NiCoFeCrAlx alloys
by making use of the above 2 2 2 supercells, each of them containing one single
Al atom. We relaxed the first 12 nearest neighbor NiCoFeCr sites in the fcc supercell
and the first 8 nearest neighbor NiCoFeCr sites in the bcc supercell. For the energy
gain upon the LLR, we obtained Ebcc = 0.17 mRy and Efcc = 0.32 mRy. The larger
relaxation effect in the fcc lattice is in line with our previous observation that the bcc
lattice can accommodate the large substitutional Al easier than the fcc lattice. Then
we consider Etr x(Ebcc Efcc ) as the measure of the LLR effect on the structural
energy difference per Al fraction. Adding Etr to Et from Table 5.6, we obtain that
the total structural energy difference vanishes around x = 1.2, i.e. at only slightly larger
Al content than the one predicted from the total energies obtained for rigid underlaying
lattices.
Two phases arrive at equilibrium when their chemical potentials become equal. Here
we consider the NiCoFeCrAlx system as a pseudo-binary (NiCoFeCr)1y Aly alloy (with
y = x/(4 + x)) and compute the relative formation energy according to G (y) =
G (y) (1 2y)Gfcc (0) 2yGfcc (0.5), where stands for fcc or bcc, and G (y) is the
Gibbs free energy per atom for (NiCoFeCr)1y Aly in the phase. This is approxi
mated as G (y) E (y) T Smix (y) T Smag
(y), where E (y) is the total energy per
atom for (NiCoFeCr)1y Aly in the phase, and T is the temperature. The two entropy terms are estimated within the mean-field
5 approximation. Namely, the mixture entropy
5of ideal solutions is Smix = kB i=1 ci lnci , and the magnetic entropy
Smag = kB i=1 ci ln(1 + i ), where ci is the concentration and i the magnetic moment
of the ith alloying element. Accordingly, all chemical and magnetic short range order
effects and the longitudinal spin fluctuations are neglected (i.e. for each alloy composition we assume constant local magnetic moments with temperature). The above
phenomenological approximation for the magnetic entropy was previously used to estimate the free energy of paramagnetic Fe [131] and Fe-based alloys [132, 133] having
non-integer magnetic moments. The present Gibbs free energies at different temperature are plotted in Figure 5.12. According to the rule of common tangent line, we find
that at room temperature NiCoFeCrAlx has single fcc phase for x . 0.597 (y . 0.130),
single bcc phase for x & 1.229 (y & 0.235), and two phases (duplex) between the above
limits. In terms of valence electron concentration, the present theory predicts that at 300
K the fcc phase is stable for VEC & 7.57 and the bcc one for VEC . 7.04. These theoretical solubility limits should be compared to 8.0 and 6.87 estimated by Guo et al. [28] and
7.67 7.88 and 7.06 7.29 observed in experiments [45, 51].
47
1.229
0.597
0.32
0.31
0.30
0.29
fcc
bcc
B/G
12
-c
44
(GPa)
0.28
8
4
0
-4
-8
-12
2.4
2.3
2.2
2.1
2.0
1.9
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Al content ( )
Figure 5.13. Theoretical Poissons ratio (upper panel), Cauchy pressure (middle
panel) and Pugh ratio (lower panel) for NiCoFeCrAlx as a function of
Al fraction. Results are shown for the fcc phase for x 1 and for the bcc
phase for x 1. The two single phase regions (fcc or bcc) as well as the
two-phase (duplex) region (fcc and bcc) corresponding to Figure 5.12 (300
K) are marked by different (colored) areas.
48
Table 5.7. Theoretical single-crystal elastic constants (cij , c = (c11 c12 )/2, AZ = c44 /c ), Cauchy
pressure (c12 c44 ), and polycrystal elastic moduli (B, G, E, , AVR and B/G) for fcc
and bcc NiCoFeCrAlx alloys as a function of Al content. The unit is GPa except for the
dimensionless AZ , AVR , and B/G.
x
0
0.3
0.5
1.0
1.0
1.3
1.5
2.0
fcc
fcc
fcc
fcc
bcc
bcc
bcc
bcc
c11
271
246
233
214
214
208
205
197
c12
175
171
169
167
160
151
148
140
c44
189
177
171
158
152
150
149
147
c
48.0
37.3
32.2
23.5
27.2
28.1
28.5
28.3
AZ
3.94
4.75
5.29
6.85
6.72
5.59
5.34
5.26
(c12 -c44 )
-14.3
-6.12
-2.13
9.00
7.84
0.80
-1.66
-6.56
B
207
196
190
183
178
170
167
159
G
110
96
89
76
78
78
78
77
AVR
280 0.275 0.209
248 0.289 0.262
231 0.297 0.295
201 0.317 0.369
204 0.309 0.311
203 0.301 0.298
202 0.297 0.293
199 0.291 0.289
B/G
1.88
2.04
2.13
2.40
2.29
2.17
2.13
2.06
the fcc lattice and slightly increases that of the bcc lattice. At the same time, Al stabilizes thermodynamically the bcc structure relative to the fcc one (Figure 5.11 and 5.12).
Combining these two effects, we obtain that around the duplex region (Figure 5.13),
the NiCoFeCrAlx system has two very similar distinct local minima within the Bain
configurational space (described by c/a and volume) with a clear barrier between them
(Figure 5.14). One local
minimum corresponds to the bcc phase (c/a = 1) and another to
the fcc phase (c/a = 2). This situation is rather unusual for elemental cubic transition
metals and their alloys [134], for which the thermodynamically unstable cubic structure
is usually also dynamically unstable (or barely stable).
Negative Cauchy pressure has been associated with the covalent nature of the metallic
bond and is characteristic to brittle alloys [135]. According to Pugh [93] materials with
B/G ratio above 1.75 are ductile. For isotropic materials, the Pugh criteria for ductility implies > 0.26, which has been confirmed for bulk metallic glasses [94]. In the
case of NiCoFeCrAlx , alloys outside of the two phase region possess small or negative
Cauchy pressure, and their Pugh and Poissons ratios are also the lowest (Figure 5.13).
On the other hand, close to x = 1 both phases have large positive Cauchy pressure, and
large B/G and , indicating strong metallic character and enhanced ductility for these
systems. Our calculated Youngs moduli are very close for the fcc (201 GPa) and bcc
(204GPa) NiCoFeCrAl. In Figure 5.15, we show the change of Youngs modulus is from
70 to 370 GPa (77 to 355 GPa) for the fcc (bcc) NiCoFeCrAl along the different directions.
The experimental value of E (127 GPa) [20] is within this scope of our calculations.
49
4.00
2.72
3.42
2.83
2.70
2.25
1.67
1.08
2.66
0.50
2.64
-0.08
(Bohr)
2.68
-0.67
2.62
-1.25
-1.83
2.60
-2.42
2.58
-3.00
0.96
1.04
1.12
1.20
1.28
1.36
1.44
1.52
c/a
Figure 5.14. Energy contour (in mRy) for paramagnetic NiCoFeCrAl as a function
of the tetragonal ratio (c/a) and the Wigner-Seitz radius w.
350
bcc
300
fcc
250
200
150
E_exp.=127.0 GPa
100
50
001
111
110
Figure 5.15. Youngs modulus of NiCoFeCrAl (bcc and fcc structure) as a function
of direction including the three main cubic directions.
Chapter 6
Interlayer potentials
Within the theory of effective interaction potentials, the energy of different structural
models is usually expressed in terms of the model potentials. If the energy of a structural model is obtained by ab initio calculations or experiments, one can use different
f (d),
(6.1)
d|n
(d)F (n/d),
(6.2)
d|n
d = 1,
1
r
(1) d has r distinct prime divisors,
(d) =
0
others.
(6.3)
51
Figure 6.1. The stacking arrangement for ABC, AB and ABAC structural models.
and for the dhcp (0001) layers ABACABAC . By the ABC structural model we mean
a stacking sequence as in the case of the fcc (111) lattice and with the interlayer distance
as the only free lattice parameter. Similarly, the structural models AB and ABAC correspond to the hcp (0001) and dhcp (0001) lattices with variable interlayer distances. For
the ABC model, the ILPs are denoted by AB , BC , CA , AC , BA , CB , AA , BB , and
CC . Similarly, the potentials are AB , BA , AA , BB in the AB structural model, and
AB , BA , AC , CA , BC , CB , AA , BB , and CC in the ABAC structural model. Due to
the symmetry, we have AB =BC =CA , AC =BA =CB , and AA =BB =CC .
By definition, the ILP ab (d) is between neighboring layers, i.e. between the A-B, A-C
and B-C layers, and aa (d) between aligned layers, i.e. between the A-A, B-B and C-C
layers. Here d represents the coordinate perpendicular to the layers.
E ABC (d) =
(6.4)
n=1
where d is the interlayer distance (separation between adjacent atomic layers) and the
energy is per 3 atoms. For the AB and ABAC models, the corresponding total energy
(per 2 and 4 atoms, respectively) may be written in the following forms
52
AB
(d) =
(6.5)
n=1
and
ABAC
(d) =
(4ab ((2n 1)d) + 2ab ((4n 2)d) + 2aa ((4n 2)d) + 4aa (4nd)). (6.6)
n=1
In the present thesis, we only use the ABC and AB structures as the calculation models
for the ILPs inversion.
We can solve ab and aa if E ABC and E AB are known. The energy difference ab (d) is
defined as
3
ab (d) = E ABC (d) E AB ( d),
2
then we obtain two following formulas
ab (d) =
ab ((3n 2)d)
n=1
or
ab (d) =
(6.7)
3
ab ((3n )d) +
ab ((3n 1)d),
2
n=1
n=1
(6.8)
(6.9)
r(n)ab (b(n)d).
n=1
b(n)
r(n)
1 2 3 4 5
1 23 2 4 92
1 -1 1 1 -1
6 7
5 7
1 1
8
15
2
-1
9 10
8 10
1
1
ab (d) =
n=1
R(n)ab (B(n)d).
(6.10)
{
R(n) =
53
(6.11)
According to Chen-Mobius
inversion method [70], we can get
ab (d) =
(6.12)
J(n)ab (B(n)d),
n=1
J(n)R(B 1 (
B(n)|B(k)
B(k)
)) = k,1 .
B(n)
(6.13)
The coefficients B(n), R(n) and J(n) are listed in Table 6.2.
Table 6.2. The data list for the coefficients B(n), R(n) and J(n) in Eq. 6.13.
1
B(n) 1
R(n) 1
J(n) 1
3
2
-1 1
1 -1
3
2
9
4
5
3
0 0
1 -2
27
8
7 8
4 92
0 1 -1
1 0 -2
9 10
5 81
16
1
0
-1
1
aa (d) = E
AB
d
1
( )
ab ((n )d),
2
2
n=1
(6.14)
aa (d) =
(n)aa (nd),
(6.15)
n=1
n = 1,
1
s
(1) n = p1 p2 . . . ps (p1 = p2 = . . . = ps ),
(n) =
0
others.
where the coefficient (n) are listed in Table 6.3.
(6.16)
54
(n)
1
1
2
-1
3 4
-1 0
5 6
-1 1
7 8 9 10
-1 0 0
1
6.3
a1
b
e 1 (rij c1 )
a2
b
e 2 (rij c2 )
a3
.
b
e 3 (rij c3 )
(6.17)
1+
1+
1+
Here rij represents the interlayer distance. Note that there are totally 12 adjustable parameters in this function (D0 , y, R0 , a1 , a2 , a3 , b1 , b2 , b3 , c1 , c2 , and c3 ) and thus it offers a
rather flexible form to describe the pair interactions between layers. In Table 6.4, we list
the parameters. R0 is often set to 1.0.
ij = D0 e
The ILP ab is similar to the interatomic pair potential. At the equilibrium interlayer
distance, the ILP ab takes its minimum value. Note that ILP aa is plotted only for large
interlayer distances, since Eqs. 6.4 and 6.5 involve aa (d) only for d 2d1 , where d1
stands for the shortest interlayer distance defined.
55
0.2
0
0.2
0.4
Inversion
Fitting for
0.6
ab
0.8
1
0.1
0.2
Inversion
0.3
Fitting for aa
0.4
0.5
0.6
3.5
4.5
5.5
Figure 6.2. The CASTEP ILPs ab (a) and aa (b) of fcc (111) planes for Al, the dis
crete points are the ab initio data obtained by Chen-Mobius
inversion and
the lines represent the fitting potentials.
Table 6.4. The RSL2 parameters of the ILPs ab and aa from Eq. 6.17. The units
are indicated.
Parameter
ab
D0 (eV)
-7.372035
R0 (A)
1.000000
y
1.412776
a1 (eV)
8381.2168
1 )
b1 (A
3.150760
c1 (A)
1.567630
aa
-302.853273
1.000000
2.588845
-0.243731
365.406083
3.698911
Parameter
ab
a2 (eV)
-602.0528
1 )
b2 (A
2.819443
c2 (A)
1.766391
a3 (eV)
-7620.8720
1 )
b3 (A
3.23390
c3 (A)
1.565706
aa
-69.426275
7.853272
0.000432
0.651101
9.464628
1.409775
56
0
ab potential
0.02
0.04
0.06
0.08
0.1
3
2
x 10
0
10
aa potential
5
10
15
20
4
10
12
14
Figure 6.3. The EMTO ILPs (ab upper panel and aa lower panel) for Pd fcc (111)
planes plotted as a function of the interlayer distance.
(6.18)
written in terms of the average interlayer distance d. Here , a, b and c are the four
independent Morse parameter. In order to clearly express the ILP ab , we can write the
Morse function in another form
( Rd 1)
ab (d) = D0 e
2 ( Rd 1)
2D0 e
+ E0 .
(6.19)
aa (d) = Y0 + A1 e
Td
+ A2 e
(6.20)
57
0
0.2
0.4
Interlayer
potential
Ab initio
0.6
0.8
2
0
0.2
2.5
3.5
4.5
b
Interlayer
potential
Ab initio
0.4
0.6
0.8
2
2.5
3.5
4.5
Figure 6.4. Comparison between the ab initio total energies of the structure model
ABC (panel a) and AB (panel b) (symbols) and the corresponding total energies obtained from the ILPs (lines). The energies are plotted as a function
of interlayer separation.
6.4
By contrasting the energies of the ABC and AB models as derived from the ILPs and the
original ab initio total energies, we learn about the accuracy of the inversion and fitting
procedures. Figure 6.4 shows this comparison taken Al as example. We find that the
ILPs reproduce well the results of the ab initio calculations from CASTEP method.
For the IILPs derived from the EMTO-CPA method, we compared the energies of the
ABC and AB model derived from the interlayer potentials (i.e. Eqs. 6.4 and 6.5) with the
ab initio total energies, At the same time, we also employed the ABAC model to check
the ILPs (in Eq. 6.6). A comparison is shown for Ag in Figure 6.5 for the ABC model
(upper panel), AB model (middle panel) and ABAC model (lower panel). We find that
the present interlayer potentials reproduce well the results of the ab initio calculations
including the ABAC model system which was not used to derive the potential.
The good agreement between the ILPs and ab initio calculations demonstrates the robustness of our approach.
58
ABC
0.02
0.04
E (Ry/layer)
0.06
2
0
AB
Ab initio data
Interlayer potentials
0.02
0.04
0.06
2
0
ABAC
0.02
0.04
0.06
2
d ()
Figure 6.5. Comparison between the ab initio total energy and the one calculated
with the ILPs for the Ag. Results are shown for the ABC model (upper
panel), AB model (middle panel) and ABAC model (lower panel).
Chapter 7
Properties related to atomic layers
7.1 Equilibrium properties
In order to get ILPs, we need to calculate equilibrium lattice parameter and cohesive
energy. Choosing the proper exchange-correlation approximation functional is particulary important. Different exchange-correction functionals have been compared in
Refs. [142146]. It was reported that PBE often overestimates the lattice constants and
underestimates the cohesive energy. On the other hand, PBEsol may accurately describe
the equilibrium properties of dense solids and their surfaces [147]. In the present thesis,
we employed the PBE generalized gradient approximation for Al, Ni, Cu, Ag and Au
in CASTEP and the modified PBE (PBEsol) for Pd-Ag random alloy in EMTO.
units of A) for five fcc metals. For comparison, some experimental lattice constants and the deviations between theory and experiment (a/a, in %) are also
shown.
Metal
Al
a (PBE)
4.0492
a (Exp.) 4.0495 [149]
a/a
0.01
a0
2.8634
d0
2.3381
Ni
3.5294
3.5240 [150]
0.15
2.4918
2.0346
Cu
3.6261
3.6147 [151]
0.32
2.5560
2.0871
59
Ag
4.1384
4.0857 [152]
1.29
2.8890
2.3589
Au
4.1811
4.0789 [153]
2.51
2.8838
2.3548
60
61
1,792
1,790
Madelung constant
0.8165
1,788
1,786
1,784
1,782
ABC
1,780
AB
1,778
1,776
1,774
1,772
0,60
0,65
0,70
0,75
0,80
0,85
0,90
0,95
1,00
1,05
d/a
Figure 7.1. The Madelung constant versus d/a0 value for the ABC and AB structural models.
Table 7.2. Equilibrium lattice constants (a) for fcc Pd, Ag and Pd0.5 Ag0.5 . The experimental
results corrected for the zero-point phonon term are shown in parentheses [154,155].
The quoted experimental values for Pd0.5 Ag0.5 were obtained for Pd0.5087 Ag0.4913
and Pd0.462 Ag0.538 (3.9850 A)
[137]. Previous theoretical values were ob(3.9766 A)
tained using full-potential methods [143, 144] as well as the EMTO-CPA method
[138]. In the lower panel of the table, a0 and d0 stand for the equilibrium nearestneighbor distance in the fcc (111) plane and the equilibrium interlayer distance perpendicular to the (111) plane as determined from the corresponding calculated a
values. dABC and dAB are the calculated equilibrium interlayer distance for fcc (111)
and hcp (0001) models, respectively. All parameters are expressed in units of A.
a (DFT)
a (Expt.)
a (Error)
a (Theo.)
a0
d0
dABC
dAB
Pd(EMTO) Pd(CASTEP)
3.8908
3.8795
3.881 (3.877) [154]
3.879 (3.875) [155]
+0.4%
-0.06%
3.876 [143], 3.882 [144]
2.7512
2.7432
2.2464
2.2389
2.2460
2.2419
2.2666
2.2659
62
51.1 [156]
a
results obtained for the ideal interlayer distance.
63
Figure 7.2. The stacking arrangement for fcc (111) planes, the perfect stacking arrangement (a), one dimension arrangement of viewing each layer as a
unity (b) and (d), the stacking arrangement with an intrinsic fault (c).
.
fcc (111) arrangement and can be expressed as ABC ABC ABC .. BC ABC ABC
.
(Figure 7.2d), where .. marks the missing layer.
The formation energy per area of a stacking fault in an otherwise perfect crystal is determined by the energy difference between structures with (Esf ) and without (Et ) the
stacking fault, viz.
sf =
Esf Et
,
A
(7.1)
where Et refers to the same number of layers (N ) as the faulted structure and A is the
area of the stacking fault plane.
In ab initio calculations of the stacking fault energy, the supercell approach and the axial
interaction model (AIM) have been applied [136, 139]. A comparison of these two approaches has been reported [140]. In the supercell model, Esf represents the total energy
of a supercell containing N layers and one stacking fault and Et = N Efcc , where Efcc
is the energy per atom of the ideal fcc stacking. Within the AIM approach including
interactions up to the third order, the ISF energy may be expressed as
sf =
(7.2)
64
In this expression all total energies are per atom. The total energies from Eqs. 7.1 and
7.2 can be determined either by direct DFT calculations or via Eqs. 6.4, 6.5 and 6.6 using
the ILPs.
For intrinsic stacking faults, the stacking fault energy is expressed directly with ILPs as
following
sf =
n=1 (3n
aa (3nd)]
.
(7.3)
The extrinsic stacking fault (ESF) is obtained by inserting an extra layer in the ideal
. .
fcc (111) arrangement. Its stacking is ABC ABC.. B .. ABC ABC , which is also
viewed as two next-nearest intrinsic stacking faults. While the twin stacking fault is
a plane boundary between two fcc stacking sequences with opposite orientation, i.e.
ABCABACBA .
The values of stacking fault energy are usually very small (order of 10-200 mJ/m2 ). Thus
obtaining accurate stacking fault energy is a rather difficult task in the experiments. Due
to the experimental uncertainties involved (e.g., temperature and alloying effects), the
reported values show large scatter listed in Table 7.4. In our calculations of the stacking
fault energy for Al, Ni, Cu, Ag and Au, we constructed the supercell with 12 layers
vacuum layers for the perfect stacking and the stacking with intrinsic stacking
and 14 A
fault. In the supercell method, the intrinsic stacking fault is the middle of 12 layers, the
same surface layers are constructed for the two stackings. So the total energy difference
between two stackings is the intrinsic fault energy. We found a good agreement between
the stacking fault energies from the ILPs and ab initio calculations (Table 7.4).
For Pd-Ag alloys and Pd and Ag, in addition to the comparisons between the supercell
and AIM results, we also contrast the values from EMTO and CASTEP.
The ILPs may be used to establish the interaction between two intrinsic stacking faults
located within the same plane family. To this end, we considered an ideal fcc supercell ABCABCABCABC from which we removed two layers separated by a specific
.
.
.
number of atomic layers (Ns ). For instance, in ABC..BCA..CABC (where .. stands
for the missing layer) the two ISFs are separated by three atomic layers (Ns = 3). The
EMTO-ILP formation energies for the above double-ISF in Pd, Ag and Pd0.5 Ag0.5 are
plotted in Figure 7.3 as a function of the number of layers separating the two faults. It
is expected that with increasing Ns , the energy of double-ISF should converge to twice
of that of the ISF from Table 7.3 (ILP supercell results). Indeed, for all three systems the
double-ISF energy reaches with good accuracy the converged value already for Ns 3.
For smaller separations, there is a clear attractive interaction between the ISFs. To understand this, we should realize that when the two ISFs become next-nearest neighbors
(Ns = 1), then the double-ISF reduces to an extrinsic stacking fault, viz. configuration
. .
ABC..B..ABCABC . The calculated (EMTO-ILP) ESF energies are 192.4, 20.6 and 89.6
65
mJ/m2 for Pd, Ag and Pd0.5 Ag0.5 , respectively. These values are marginally larger than
the corresponding ISF energies, meaning that energetically it is slightly more favorable
to have one ESF than two ISFs. It is noticeable that the double-ISFESF reconstruction
happens without a sizable energy barrier (except the one associated with the shift of the
atomic layers parallel to the stacking fault).
Using the ILPs derived from EMTO calculations, we also determined the formation
energy of a twin stacking fault described by the configuration ABCABCBACBA
for Pd, Ag and Pd0.5 Ag0.5 . The theoretical TSF energies are 155.7, 9.9 and 52.0 mJ/m2 ,
respectively. The small values of TSF indicate that these faults are created much easier
than the intrinsic and the extrinsic stacking faults.
Table 7.4. Stacking fault energies (sf , in units of mJ/m2 ) for Al, Ni, Cu, Al, and Au. The displayed data are as follows: ILP present results obtained using the interlayer potentials;
DFT present results obtained from direct DFT calculations; TBP, TB: calculations based
on tight-binding method; FLMTO: results of full potential linear-muffin-tin orbitals
calculations; EAM: calculations from the embedded-atom-method potentials.
Metal
ILP
DFT
TBP [157]
TB [158]
FLMTO [159, 160]
Potential [161]
EAM [162]
EAM [163]
Exp. [164]
Exp. [165]
Exp. [166]
Al
Ni
Cu
Ag
Au
180
213
33
16
26
180
183
34
22
41
305/265
21/15
1/0
96
18
29
50
164
46
146.0/95.4
304.4/120.3 20.6/33.5
352/361
104/109
119/121
67/73
41/44
146.47
90.99
74.59
44.19
61.58
170/200/240 64-140/410
40/169
21/26-58
24-47
166
64-140/410 40/45/169
24-47
16/43/58
166
128/250
48/55/78
16/22
Pd(E)
Pd(C)
PdAg(E)
Ag(E)
Ag(C)
a
sf (ILP)
185.6
199.3
81.1
15.0
24.7
sf (ILP)b
186.9
227.0
87.9
16.8
26.9
a
c
sf (DFT)
185.0
178.5
76.8
26.6
26.0/28.4c
b
c
sf (DFT)
192 [167]
164.3
83 [167]
23 [167] 19.5/20.3c
sf (Expt.)
180 [168]
a
b
c
AIM results; supercell results; results obtained for the ideal interlayer distance;
E(C) stands for the EMTO (CASTEP) method.
66
35
30
25
Ag
20
200
150
2Eist
Ag0.5Pd0.5
100
400
300
Pd
200
1
Figure 7.3. (Color online) The formation energy for two parallel intrinsic stacking
faults for fcc Pd, Ag and Pd0.5 Ag0.5 plotted as a function of the number of
atomic layers between the two faults. The dashed lines show the energy
corresponding to two non-interacting intrinsic stacking faults.
Alternatively, in the case of the ILP-based description of the fcc (111) surface, we may
consider a semi-infinite surface geometry CBACBAVacuum , so that there is only
one surface layer. Then the surface energy is the sum of the ILPs between the atomic
layers and the removed layers, viz.
s =
k=1 [(3k
(7.5)
where k represents a layer in the surface model and also indicates the number of inter-
67
Figure 7.4. Schematic plot of the structures used to calculate the surface energy.
Panel (a): perfect fcc (111) stacking arrangement; panel (b): 12-atomic layers slab geometry separated by vacuum.
7.5
Phonon dispersion
The vibration of atoms in the crystal is correlated and the collective vibration forms a
wave of allowed wavelength and amplitude. The quantum of such lattice vibration is
68
and
Table 7.5. Surface energy s (in mJ/m2 ), bulk equilibrium interlayer distance db12 (in A),
For ILP, the numbers in parenthesis
surface equilibrium interlayer distance ds12 (in A).
are the surface energies obtained for the semi-infinite surface model. sr (in mJ/m2 )
represent the surface energy after surface relaxation.
s (ILP)
s (DFT)
db12
ds12
sr (ILP)
called phonon. The phonon dispersion describes the wave vector dependence of the
phonon frequencies.
To establish the phonon spectra for a system, we first get the force constant matrix,
given by the second order derivative of the internal energy with respect to the Cartesian
coordinates. The force constant matrix, F , between two atoms i and j is given by
Fij (k) =
2U
(
)eik(rij +R) ,
(7.6)
where and are position vectors in the Cartesian coordinates. The dynamical matrix
D is
Dij =
1
(mi mj )1/2
Fij (k),
(7.7)
where mi and mj are the masses of atom i and j. Further details about the phonon
dispersion from the force constant matrix are discussed as in Ref. [141].
Possessing the phonon spectrum, one can predict a series of thermo-physical properties.
The phonon dispersions are represented by the interatomic forces derived from the interatomic potentials. Here we calculated the phonon dispersion for the fcc (111) planes
using the ILPs.
the whole fcc (111) layer as a unit, we consider the mass of the
Viewing
2
atoms in the 3a /4 area as the mass of the layer (a is the fcc lattice constant from Table 7.3). For each layer, we use the equivalent-atom model to represent the total layer
and the masses of these equivalent-atoms are identical.
Although the three layers in the fcc (111) planes are only different because of the relative
position of atoms, the one-dimension primitive cell contains three different atoms in
our phonon calculations, so there are six branches in the phonon dispersion. Note that
69
Frequency (THz)
Pd
Pd-Ag
Ag
Figure 7.5. The phonon dispersion for Pd and Ag metals and Pd0.5 Ag0.5 alloy.
the optical branches from Figure 7.5 are in fact the folded-in acoustic branches (there
is no gap at the zone boundary between the branches). In the fcc with one atom per
primitive cell, the real phonon dispersion only has two acoustical branches along the
fcc 111 direction, with degenerated transversal branches. Since we investigate the
interactions along one direction, the distinct number of phonon branches is reduced
along the principal symmetry direction G-Z as shown in Figure 7.5. Here we would like
to make an Erratum to the Supplements: in the figure of the dispersion phonon from the
published papers there is a small mistake in the transversal phonon branch as a result
of the employed phonon-calculation software.
Furthermore, the phonon density of states from the Supplements represent the density
of states of a hypothetical one-dimensional lattice and thus should not be compared to
the real three-dimensional results.
The interlayer interaction for Pd is stronger than that for Ag, and thus the vibrational
frequencies versus k for Ag are weaker than those for Pd. The vibrational frequency of
the Pd0.5 Ag0.5 alloy is in between those obtained for Pd and Ag metals. To our knowledge, this is the first phonon spectra presented for completely random alloy and calculated using the mean-field coherent potential approximation.
Chapter 8
Concluding remarks and Future work
Based on the available experiments, the structural properties and hardness of singlephase high entropy alloys were discussed. Using the exact muffin-tin orbitals method
in combination with the coherent potential approximations (EMTO-CPA), we investigated the structural, electronic and elastic properties of single-phase high-entropy alloys. Our ab initio predicted structural properties are consistent with the experimental
results, especially for the equilibrium volume and phase transformation. According to
the calculated elastic properties from ab initio alloy theory, we find that: 1) NiCoFeCr
and CuNiCoFeCr are more isotropic and less ductile than the Ti-containing single phase
alloys (CuNiCoFeCrTix with x & 0.4 and NiCoFeCrTi) in CuNiCoFeCrTix HEAs, 2) for
refractory- HEAs, we may design the elastically isotropic alloys according to the average
valence electron concentration VEC 4.72, 3) for the NiCoFeCrAlx HEAs with fccbcc
phase transformation, the HEAs abound the equimolar NiCoFeCrAl composition have
superior mechanical performance as compared to the single-phase regions.
Interlayer Potentials suit to describe the properties related to layers in layered materials. We should mention that due to the complex magnetic interactions along the closepacked layers, there are some limitations for using the ILPs to investigate the related
properties of magnetic materials. For the present systems, the high accuracy of the ILPs
originates from the fact that the total layer is considered as a unit and in a layered materials the interlayer interactions are usually much weaker than the interaction within
the layer. For some layered materials, for example graphite and BN, ab initio tools do
not accurately describe the interlayer cohesive energy. In such cases, one may use the
Chen-Mobius
inversion to get accurately ILPs starting from the experimental data.
The present results obtained for the random solid solutions show that the EMTO-CPA
method is a powerful tool to investigate complicated alloys and is able to provide accurate ILPs. For future works, I will continue to use the EMTO-CPA method to study
the high entropy alloys (HEAs). After performing an intensive theoretical investigation of the HEAs, we will use the above ILPs technique to describe the layer dependent
properties of this complex and promising family of engineering alloys.
70
Acknowledgements
First, I am grateful to my supervisors Prof. Levente Vitos and Prof. Nanxian Chen for
giving me the opportunity to learn the EMTO-CPA method and the Mobius
inversion,
for always showing a great interest to my work and for professional guidance.
Prof. Lajos Karoly Varga, at the Institute for Solid State Physics and Optics, Wigner
Research Center for Physics, Hungarian Academy of Sciences, Budapest, thank you for
providing the chance to know the experimental process of high-entropy alloys. I also
appreciate Prof. Jiang Shens help, at the Institute for Applied Physics, University of
Science and Technology, Beijing.
Lorand, Erna, B. Narisu, Hualei, Song, Wei, Guisheng, Xiaoqing, Stephan, Andreas,
Dongyoo: many thanks for helpful discussions. I would like to thank all the members
of the AMP group for creating an inspiring environment for my learning and for help.
I am grateful to Qingmiao Hu, Xiaojian Yuan, Ping Qian, Yudong Wang, Gang Liu, and
Hubin Luo for their great support.
The Swedish Research Council, the European Research Council, the China Scholarship
Council, National 973 Project of China, Swedish Foundation for International Cooperation in Research and Higher Education (STINT), and the Hungarian Scientific Research
Fund are acknowledged for financial support.
Last but not least, I would like to express my deepest thanks to my wife and daughter,
for their support and love. I also acknowledge my parents and my brother for great
support and encouragement.
71
Bibliography
[1] J. W. Yeh, S. K. Chen, S. J. Lin, J. Y. Gan, T. S. Chin, T. T. Shun, C. H. Tsau, S. Y.
Chang, C. H. Tsau, and S. Y. Chang, Adv. Eng. Mater. 6, 299 (2004).
[2] B. Cantor, I. T. H. Chang, P. Knight, and A. J. B. Vincent, Mater. Sci. Eng. A 375, 213
(2004).
[3] J. W. Yeh, S. K. Chen, J. Y. Gan, S. J. Lin, T. S. Chin, T. T. Shun, C. H. Tsau, and S. Y.
Chang, Metall. Mater. Trans. A 35, 2533 (2004).
[4] C. J. Tong, M. R. Chen. S. K. Chen, J. W. Yeh, T. T. Shun, S. J. Lin, and S. Y. Chang,
Metall. Mater. Trans. A 36, 1263 (2005).
[5] X. F. Wang, Y. Zhang, Y. Qiao, and G. L. Chen, Intermetallics 15, 357 (2007).
[6] C. Li, J. C. Li, M. Zhao, and Q. Jiang, J. Alloys Compd. 475, 752 (2009).
[7] K. B. Zhang, Z. Y. Fu, J. Y. Zhang, W. M. Wang, H. Wang, Y. C. Wang, Q. J. Zhang,
and J. Shi, Mater. Sci. Eng. A 508, 214 (2009).
[8] S. Singh, N. Wanderka, B. S. Murty, U. Glatzel, and J. Banhart, Acta Mater. 59, 182
(2011).
[9] M. S. Lucas, L. Mauger, J. A. Munoz, Y. M. Xiao, A. O. Sheets, S. L. Semiatin, J.
Horwath, and Z. Turgut, J. Appl. Phys. 109, 07E307 (2011).
[10] K. B. Zhang and Z. Y. Fu, Intermetallics 28, 34 (2012).
[11] S. Varalakshmi, G. A. Rao, M. Kamaraj, and B, S. Murty, J. Mater. Sci. 45, 5158 (2010).
[12] O. N. Senkov, G. B. Wilks, D. B. Miracle, C. P. Chuang, and P. K. Liaw, Intermetallics
18, 1758 (2010).
[13] Y. P. Wang, D. Y. Li, L. Parent, and H. Tian, Wear 271, 1623 (2011).
[14] O. N. Senkov, J. M. Scott, S. V. Senkova, D. B. Miracle, and C. F. Woodward, J. Alloys
Compd. 509, 6043 (2011).
72
BIBLIOGRAPHY
73
[15] X. Yang, Y. Zhang, and P. K. Liaw, Procedia Engineering 36, 292 (2012).
[16] W. Y. Tang, M. H. Chuang, S. J. Lin, and J. W. Yeh, Metall. Mater. Trans. A 43, 2390
(2012).
[17] A. Cunliffe, J. Plummer. I. Figueroa, and I. Todd, Intermetallics 23, 204 (2012).
[18] S. K. Chen and Y. F. Kao, AIP Advances 2, 012111 (2012).
[19] Y. P. Wang, B. S. Li, M. X. Ren, C. Yang, and H. Z. Fu, Mater. Sci. Eng. A 491, 154
(2008).
[20] Y. J. Zhou, Y. Zhang, Y. L. Wang, and G. L. Chen, Appl. Phys. Lett. 90, 181904 (2007).
[21] Y. F. Kao, S. K. Chen, J. H. Sheu, J. T. Lin, W. E. Lin, J. W. Yeh, S. J. Lin, T. H. Liou,
and C. W. Wang, Int. J. Hydrogen Energy 35, 9046 (2010).
[22] I. Kunce, M. Polanski, and J. Bystrzycki, Int. J. Hydrogen Energy 38, 12180 (2013).
[23] S. Y. Chang and D. S. Chen, Appl. Phys. Lett. 94, 231909 (2009).
[24] V. Baric, M. Balaceanu, M. Braic, A. Viadescu, S. Panseri, and A. Russo, J. Mechl
Behav, Biomed. Mater. 10, 197 (2012).
[25] G. Y. Ke, S. K. Chen, T. Hsu, and J. W. Yeh, Annales De Chimie-Science Des Materiaux
31, 669 (2006).
[26] Y. Zhang, Y. J. Zhou, J. P. Lin, G. L. Chen, and P. K. Liaw, Adv. Eng. Mater. 10, 6
(2008).
[27] Y. J. Zhou, Y. Zhang, F. J. Wang, and G. L. Chen, Appl. Phys. Lett. 92, 241917 (2008).
[28] S. Guo, C. Ng, J. Lu, and C. T. Liu, J. Appl. Phys. 109, 103505 (2011);
[29] S. Guo and C. T. Liu, Prog. Nat. Sci.: Mater. Int. 21, 433 (2011).
[30] X. Yang and Y. Zhang, Mater. Chem. Phys. 132, 233 (2012).
[31] S. Guo, Q. Hu, C. Ng, and C. T. Liu, Intermetallics 41, 96 (2013).
[32] M. F. del Grosso, G, Bozzolo, and H. O. Mosca, J. Alloys Compd. 534, 25 (2012);
Physica B 407, 3285 (2012).
[33] C. Li, M. Zhao, J. C. Li, and Q. Jiang, J. Appl. Phys. 104, 113504 (2008).
[34] C. Zhang, M. Lin, and B. Wu, et al., J. Shanghai Jiaotong Univ. (Sci.) 16, 173 (2011).
[35] M. S. Lucas, D. Beelyea, C. Bauer, N. Bryant, E. Michel, Z. Turgut, S. O. Leontsev,
J. Horwath, S. L. Semiatin, M. E. McHenry, and C. W. Miller, J. Appl. Phys. 113,
17A923 (2013).
74
BIBLIOGRAPHY
[36] E. L. Liu, J. B. Zhu, L. Li, L. C. Li, and Q. Jiang, Mater. Des. 44, 223 (2013).
[37] C. X. Wang, H. Xie, L. Jia, and Z. L. Lu, Mater. Sci. Forum 724, 335 (2012).
[38] O. N. Senkov, G. B. Wilks, J. M. Scott, and D. B. Miracle, Intermetallics 19, 698 (2011).
[39] O. N. Senkov, J. M. Scott, S. V. Senkova, F. Meisenkothen, D. B. Miracle, and C. F.
Woodward, J. Mater. Sci. 47, 4062 (2012).
[40] O. N. Senkov and C. F. Woodward, Mater. Sci. Eng. A 529, 311 (2011).
[41] O. N. Senkov, S. V. Senkova, C. Woodward, and D. B. Mircale, Acta Mater. 61, 1545
(2013).
[42] O. N. Senkov, S. V. Senkova, D. B. Mircale, and C. Woodward, Sci. Eng. A 565, 51
(2013).
[43] W. Guo, W. Dmowsk, J. Y. Noh, P. Rack, P. K, Liaw, and T. Egami, Metall. Mater.
Trans. A 44, 1994 (2013).
[44] C. J. Tong, M. R. Chen, S. W. Yeh, T. T. Shun, S. J. Lin, and S. Y. Chang, Metall. Mater.
Tran. A 36, 1263 (2005).
[45] Y. F. Kao, T. J. Chen, S. K. Chen, and J. W. Yeh, J. Alloys Compd. 488, 57 (2009).
[46] Y. F. Kao, S. K. Chen, T. J. Chen, P. C. Chu, J. W. Yeh, and S. J. Lin, J. Alloys Compd.
509, 1607 (2011).
[47] H. P. Chou, Y. S. Chang, S. K. Chen, and J. W. Yeh, Mater. Sci. Eng. B 163, 184 (2009).
[48] T. T. Shun, C. H. Hung, and C. F. Lee, J. Alloys Compd. 493, 105 (2010).
[49] C. Li, J. C. Li, M. Zhao, and Q. Jiang, J. Alloys Compd. 504S, 1607 (2010).
[50] Y. F. Kao, T. D. Lee, S. K. Chen, and Y. S. Chang, Corros. Sci. 52, 1026 (2010).
[51] W. R. Wang, W. L. Wang, S. C. Wang, Y. C. Tsai, C. H. Lai, and J. W. Yeh, Intermetallics 26, 44 (2012).
[52] C. Zhang, F. Zhang, S. L. Chen, and W. S. Cao, JOM, 64, 7 (2012).
[53] S. Guo, C. Ng, and C. T. Liu, Gu J. Alloys Compd. 557, 77 (2013).
[54] P. Hohenberg and W. Kohn, Phys. Rev. 136, B864 (1964).
[55] W. Kohn and L. J. Sham, Phys. Rev. 140, A1133 (1965).
[56] O. K. Andersen, O. Jepsen, and G. Krier, in Lectures on Methods of Electronic Structure Calculation (World Science, Singapore, 1994).
BIBLIOGRAPHY
75
76
BIBLIOGRAPHY
[79] http://accelrys.com/products/materials-studio.
[80] J. P. Perdew and Y. Wang, Phys. Rev. B 45, 13244 (1992).
[81] J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996).
[82] J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou, and K. Burke, Phys. Rev. Lett. 100, 136406 (2008).
[83] D. Vanderbilt, Phys. Rev. B 41, 7892 (1990).
[84] D. R. Hamann, M. Schluter, and C. Chiang, Phys. Rev. Lett. 43, 1494 (1979).
[85] P. Soven, Phys. Rev. 156, 809 (1967).
[86] D. W. Taylor, Phys. Rev. 156, 1017 (1967).
[87] B. L. Gyorffy, Phys. Rev. B 5, 2382 (1972).
[88] A. J. Pindor, J. Stauton, G. M. Stocks, and H. Winter, J. Phys. F: Met. Phys. 13, 979
(1993).
[89] B. L. Gyorffy, A. J. Pindor, J. Stauton, G. M. Stocks, and H. Winter, J. Phys. F: Met.
Phys. 15, 1337 (1995).
[90] Grimvall G. Thermophysical Properties of Materials, (Amsterdam: North-Holland
1999).
[91] V. L. Moruzzi, J. F. Janak, and K. Schwarz, Phys. Rev. B 37, 790 (1988).
[92] D. Nguyen-Manh, M. Mrovec, and S. P. Fitzgerald, Mater. Trans. 49, 2497 (2008).
[93] S. F. Pugh, Philos. Mag. 45, 823 (1954).
[94] X. J. Gu, A. G. McDermott, S. Joseph Poon, and G. J. Shiflet, Appl. Phys. Lett. 88,
211905 (2006).
[95] F. L. Liu, J. B. Zhu, C. Zhang, J. C. Li, and Q. Jiang, Mater. Sci. Eng. A. 548, 64 (2012).
[96] F. J. Wang, Y. Zhang, and G. L. Chen, J. Alloys Compd. 478, 321 (2009).
[97] O. N. Senkov and D. B. Miracle, Mater. Res. Bull. 36, 2183 (2001).
[98] W. B. Pearson, Crystal chemistry and physics of metals and alloys, Wiley (1972).
[99] M. R. Chen, S. J. Lin, J. W. Yeh, M. H. Chuang, S. K. Chen, and Y. S. Huang, Metall.
Mater. Tran. A 37, 1363 (2006).
[100] D. A. K. Singh and A. Subramanian, Adv. Mater. Res. 585, 3 (2012).
BIBLIOGRAPHY
77
[101] B. Ren, Z. X. Liu, D. M. Li, L. Shi, B. Cai, and M. X. Wang, J. Alloys Compd. 493, 148
(2010).
[102] M. H. Tsai, Y. Hao, G. M. Cheng, W. Z. Xu, K. Y. Tsai, C. W. Tsai, W. W. Juan, C.
C. Juan, W. J. Shen, M. H. Chuang, J. W. Yeh, and Y. T. Zhu, Intermetallics 32, 329
(2013).
[103] Y. Zhang, T. T. Zuo, Y. Q. Cheng, adn P. K. Liaw, Scientific reports 3, 1455 (2013).
[104] T. T. Shun, L. Y. Chang, and M. H. Shiu, Mater. Charact. 70, 63 (2012).
[105] T. T. Shun, C. H. Hung, and C. F. Lee, J. Alloys Compd. 495, 55 (2010).
[106] T. T. Shun, L. Y. Chang, and M. H. Shiu, Mater. Sci. Eng. A 556, 170 (2012).
[107] Y. L. Chou, J. W. Yeh, and H. C. Shih, Corros. Sci. 52, 2571 (2010).
[108] O. N. Senkov, J. M. Scott, S. V. Senkova, D. B. Miracle, and C. F. Woodward, J.
Alloys Compd. 509, 6043 (2011).
[109] Y. Zhang, X. Yang, and P. K. Liaw, JOM 64(7), 830 (2012).
[110] C. M. Liu, H. M. Wang, S. Q. Zhang, H. B. Tang, and A. L. Zhang, J. Alloys Compd.
583, 162 (2014).
Fazakas, V. Zadorozhnyy, L. K. Varga, A. Inoue, D.V. Louzguine-Luzgin, F. Y.
[111] E.
Tian, and L. Vitos (unpublished).
[112] H. Y. Chen, C. W Tsai, C. C. Tung, J. W. Yeh, T. T. Shun, C. C. Yang, and S. K Chen,
Annales De Chimie C Science des Materiaux 31, 685 (2006).
[113] C. W. Tsai. Y. L. Chen, M. H. Tsai, and J. W. Yeh, J. Alloys Compd. 486, 427 (2009).
[114] C. C. Tung, J. W. Yeh, T. T. Shun, S. W. Chen, Y. S. Huang, and H. C. Cheng, Mater.
Lett. 61, 1 (2007).
[115] Y. J. Hsu, W. C. Chiang, and J. K. Wu, Mater. Chem. Phys. 92, 112 (2005).
[116] C. M. Lin and H. L. Tsai, J. Alloys Compds. 489, 30 (2010).
[117] F. Y. Tian, L. Delczeg, N. X. Chen, L. K. Varga, J. Shen, and L. Vitos, Phys. Rev. B
88, 085128 (2013).
[118] H. L. Zhang, B. Johansson, R. Ahuja and L. Vitos, Comp. Mater. Sci. 55, 269272
(2012).
and L. Vitos, Comp. Mater. Sci. 41, 86 (2007).
[119] J. Zander, R. Sandstrom,
[120] J. W. Yeh, Y. L. Chen, S. J. Lin, and S. K. Chen, Adv. Eng. Mater. 560, 1 (2007).
78
BIBLIOGRAPHY
[134] P. Soderlind,
O. Eriksson, J. M. Wills, and A. M. Boring, Phys. Rev. B 48, 5844
(1993).
[135] D. Nguyen-Manh, M. Mrovec, and S. P. Fitzgerald, Mater. Trans. 49, 2497 (2008).
[136] P. J. H. Denteneer and W. van Haeringen, J. Phys. C: Solid State Phys. 20, L883
(1987).
[137] B. R. Coles, J. Inst. Met. 84, 346-348 (1955-56).
[138] E. K. Delczeg-Czirjak, L. Delczeg, M. Ropo, K. Kokko, M. P. J. Punkkinen, B. Johansson, and L. Vitos, Phys. Rev. B 79, 085107 (2009).
[139] C. Cheng, R. J. Needs, and V. Heine, J. Phys. C: Solid State Phys. 21, 1049 (1988).
[140] M. Chandran and S. K. Sondhi, J. Appl. Phys. 109, 103525 (2011).
BIBLIOGRAPHY
79
80
BIBLIOGRAPHY
[174] V. Zolyomi,
J. Kollar, and L. Vitos, Phil. Mag. 88, 2709 (2008).
[175] S. K. Kwon, Z. Nabi, K. Kadas, L. Vitos, J. Kollar, B. Johansson, and R. Ahuja, Phys.
Rev. B 72, 235423 (2005).