First-principles modeling of energetic and mechanical properties of

Ni-Cr, Ni-Re and Cr-Re random alloys

A. Breidi,1, ∗ S. G. Fries,1 M. Palumbo,1 and A. V. Ruban2, 3

1
ICAMS, Ruhr Universität Bochum, Universität Straße 150, 44801 Bochum, Germany
2
Applied Materials Physics, Department of Materials Science and Engineering,
KTH Royal Institute of Technology, SE-100 44 Stockholm, Sweden
3
Materials Center Leoben Forschung GmbH, A-8700 Leoben, Austria
(Dated: January 10, 2016)
We apply the exact-muffin-tin-orbitals (EMTO) method to investigate structural properties, for-
mation enthalpies, mechanical stability and polycrystalline moduli in Ni-Re, Ni-Cr and Cr-Re dis-
ordered fcc, bcc and hcp phases. Substitutional disorder is treated by using the coherent potential
approximation (CPA). We predict the alloy lattice parameters in good agreement with the exper-
iment. We find a continuous softening, as a function of Cr composition, of the tetragonal shear
modulus C 0 in fcc Ni-Cr phase indicating mechanical instability in Cr-rich Ni-Cr alloys. On the
other hand, we show that the mechanical stability of fcc Ni-Re alloys persists through the whole
composition range. We observe an intriguing behaviour of the Young’s modulus vs. the intrinsic
ductility curve in Ni-rich Ni-Re fcc phase.
Keywords: DFT, Random alloys, Enthalpies, Elastic constants

I. INTRODUCTION heating elements and thermocouples, etc4 .

Levy et al.5 conducted a high-throughput DFT-
Enhancing the industrial-relevant properties based calculations on Re-transition metal sys-
of Ni-base superalloys, such as the service life- tems, predicting the stoichiometric compound
time and the operating temperature, is the Ni4 Re (prototype D1a ) to be stable. Recently,
keen interest of materials scientists and en- Maisel et al.6 did a combined theoretical and
gineers. These improvements are achieved experimental study on Ni-Re. They reproduced
through the development of more complex, theoretically the stability of Ni4 Re compound at
multi-component alloys with very specific com- 0 K and predicted it to be stable up to ≈930 K
positions. A major part in this process is to according to Monte Carlo simulations. Addi-
understand how refractory elements affect the tionally, they6 argued that the experimental ob-
phase stability of these alloys. Among the re- servation of Ni4 Re precipitates depends on sam-
fractory elements, Re and Cr are known to play ple cooling conditions.
crucial roles. Re low-temperature ductility and Palumbo et al.7 calculated the Cr-Ni-Re phase
high melting point makes it an attractive solute diagram where the main topological features of
for high-temperature structural materials, par- the experimental isothermal section were repro-
ticularly when added to Ni and Cr. Moreover, duced. However, the range of solubility of bcc,
Re is known to induce a significant improve- fcc, and hcp solid solutions appears to be ei-
ment in the mechanical properties (specifically ther underestimated or overestimated, which is
strength and plasticity), the so-called rhenium attributed to the improper description of the
effect1 , when added with a concentration close configurational disorder in the solid solutions.
to the solubility limit. On the other hand, the Despite the importance of the alloys Ni-Re, Ni-
Ni-Cr2 alloys form solid solutions over a wide Cr and Cr-Re, the study of their energetic, elas-
range of composition and temperature on the tic and mechanical properties is either absent or
fcc lattice3 . They are considered important in scarce at best. Hence, it was necessary to in-
industry as they are used for oxidation resis- vestigate extensively the compositional depen-
tance, high-temperature corrosion resistance, in dence of these properties through employing a
 2

proper theoretical approach that accounts for 1. Effective medium approach

the chemical random nature of these alloys.
The paper is organized as follows: the follow- The electronic structure calculations reported
ing section presents the first-principles details here have been performed by means of the exact
and theoretical formalism for the elastic con- muffin-tin orbitals (EMTO) method8–10 within
stants and moduli calculations. In Sec. III we the context of density functional theory13,14
present and discuss the structural and energetic (DFT). In the EMTO calculations, the effect of
properties. In Sec. IV the mechanical stability substitutional disorder on the electronic struc-
of the fcc phases for the Ni-M alloys is inves- ture has been taken into consideration using
tigated, while in Sec. V we study in-depth the the coherent potential approximation15 (CPA).
mechanical properties of Ni-rich Ni-M fcc alloys. The on-site contribution from the screened
In Sec. VI we conclude, highlighting the signifi- Coulomb interactions to the electrostatic poten-
cance of the new results delivered by this work. tial and electrostatic energy of a random alloy in
the DFT self-consistent CPA calculations16,17
has been determined using EMTO locally
II. COMPUTATIONAL METHOD self-consistent Green’s-function (ELSGF)18–20
method. This method allows accurate calcula-
A. Ab initio calculations tions of random alloys with proper treatment of
the local environment effects, due to the use of
We employ two different approaches in our a local interaction zone at every site. This pro-
investigation, the effective medium and the lat- vides accurate account of the interaction of the
tice representation. The calculations are mainly central atom with its several nearest neighbor-
done using the effective medium approach while ing coordination shells. The electrostatic prob-
the lattice representation is used, whenever pos- lem in this case is treated exactly for the whole
sible, merely for the sake of comparison. Nowa- supercell modeling random alloy, which allows
days, it is a routine task to calculate the elas- calculations of the screening effects. This has
tic constants of chemically ordered phases using been determined for fcc, hcp, and bcc random
available electronic structure methods. How- alloys modeled by 256 (4×4×4 translations of
ever, the calculation of the elastic properties of the 4-atom fcc cubic cell), 288 (6×6×4 trans-
low-symmetry systems, such as random alloys, lations of the 2-atom hcp unit cell) and 256
is less obvious. In fact, a proper reproduction (8×4×4 translations of the 2-atom bcc cubic
of the chemical disorder in real alloys repre- cell)-atom supercells, respectively. These super-
sents a serious challenge, since a simple atom- cells allowed us to vary the alloy composition in
istic description of random alloys based on a steps of 6.25 at.%.
statistical distribution of atoms demands large The basis functions were expanded up to
configurational space, which renders the elec- lmax = 4 in the self-consistent ELSGF and
tronic structure calculation impracticable. Al- EMTO-CPA calculations. Multiple moments
M ad
ternatively, we mainly apply the coherent po- of the electron density, up to lmax = 6, were
tential approximation as implemented in the ex- determined during the self-consistent calcula-
act muffin-tin orbitals (EMTO) method8–10 be- tions of the multipole moment corrections to
cause it provides homogenity and preserves the the Madelung potential and energy. Total ener-
point symmetry of the system. Moreover, it has gies were calculated in the generalized gradient
been proven to be quite accurate in the calcu- approximation (GGA) for the exchange corre-
lations of elastic properties of high symmetry lation energy with parameterization of Perdew,
structures11,12 . Burke, and Ernzerhof(PBE96)21 . The inte-
gration over the Brillouin zone was performed
using a 31×31×31 grid of special k-points
determined according to the Monkhorst-Pack
 3

scheme22 which guaranteed the convergence of i.e. the translation vectors of the supercells
the total energy, equilibrium lattice parameters remained unchanged (no lattice distortions),
and elastic moduli. We should mention here thus retaining the symmetry of the underlying
that the screening parameters extracted from lattice. This was done in order to enforce
the electrostatic potential and electrostatic en- the ‘macroscopic’ symmetry persisting in
ergy, determined with ELSGF, were then used alloy, which would be otherwise gone once the
as input parameters in the EMTO-CPA. The ‘microscopic’ (local site) symmetry is revoked
Lyngby23 version of the EMTO code has been in the calculation. Hence, only the volume
used, which accurately treats electrostatics in relaxations are allowed. The c/a ratio was set
random alloys in contrast to other existing ver- to Rhenium’s equilibrium value 1.615 for the
sions. alloys Ni-Re and Cr-Re. We should mention
here that the hcp Cr is mechanically unstable
and the hcp Ni (FM and NM) is metastable30 .
2. Supercell Method Consequently Cr-rich Cr-Re hcp alloys are
most probably mechanically unstable. Fixing
The projector augmented wave (PAW) the c/a ratio to the Re value is justified by the
method24 implemented in the Vienna ab initio fact that the c/a equilibrium ratio does not
simulation package (VASP)25–27 was used in deviate significantly from 1.615, at least for the
order to study supercells. The exchange cor- relevant Re-rich Ni-Re and Cr-Re.
relation energy was treated in the GGA with
the PBE96 functional21 . After necessary tests
to control the stability of energy differences 3. Magnetic states
between phases, the energy cut off was set
to 350 eV for Ni-Re and Cr-Re alloys and All the calculations are done either ne-
400 eV for Ni-Cr alloys. These alloys in the glecting spin-polarization, assuming that the
random distribution of alloy components were state is non-magnetic (NM), or in the
modeled by supercells of different sizes. As ferromagnetic/anti-ferromagnetic (FM/AFM)
in the previous Subsec. II A 1, fcc alloys were state. In the considered Ni-rich alloys, at low
modeled by a 256-atom supercell formed by temperatures, spin-polarization effects become
4×4×4 translations of the 4-atom fcc cubic cell; very weak at relatively small compositions of
hcp random alloys were modeled by a 288-atom the other alloy component. However, most
supercell, built by 6×6×4 translations of the probably this is not anymore true for the high
2-atom hcp unit cell, while bcc alloys here temperature states, where local magnetic mo-
are modeled by a 128-atom supercell formed ment on Ni and Cr can survive due to longitu-
by 4×4×4 translations of the 2-atom bcc dinal spin fluctuations31 . Nevertheless, we ne-
cubic unit cell. For each composition, the glect these effects, as we do not include thermal
alloy species have been distributed in a way excitations in the present investigation.
to minimize the Warren-Cowley short-range
order (SRO) parameters28,29 at several nearest
neighbor coordination shells. We found the B. Elastic constants calculations
SRO parameters to be less than 0.002 at the
first eight coordination shells, depending on Cubic lattices have three independent elastic
the crystal structure and alloy composition. stiffness constants: C11 = C22 = C33 , C12 =
Accordingly, in the self-consistent calcula- C13 = C23 , and C44 = C55 = C66 . C11 and C12
tions, we have used 4×4×4 Monkhorst-Pack include large contributions from the variation
k-point mesh22 for the fcc and bcc supercells, of volume. A practical way to reduce numerical
while 2×2×2 mesh for the hcp supercells. noise in their calculation is to separate calcula-
The shape of the supercell was kept fixed, tion of the bulk modulus B, which in this case
 4

is However, in a polycrystalline material, the

C11 + 2C12 monocrystalline grains are randomly oriented.
B= , (1) On a large scale, such materials can be consid-
3
ered quasi-isotropic or isotropic in a statistical
0
and the tetragonal shear modulus C = sense. An isotropic system is completely de-
(C11 − C12 ) /2 related to the volume-preserving scribed by the bulk B and the shear G moduli,
strain32 . To determine the bulk modulus, B0 , which can be obtained by averaging the single-
0
and its pressure derivative, B0 , at the equilib- crystal elastic constants. There is no exact ex-
rium volume V0 , we fitted the energy-volume pression for the polycrystal-average shear mod-
curve to the Murnaghan equation of state33,34 ulus in terms of the Cij , but it can be estimated
 0  using various homogenization schemes.
E(V ) = E(V0 ) +
B0 V

(V0 /V )B0
+ 1 −
B0 V 0
.(2) The upper bound due to Voigt37 is calculated
0 0 0
B0 B0 − 1 B0 − 1 as
The tetragonal shear modulus can be de- C11 − C12 + 3C44
GV = , (8)
termined by applying a volume-conserving or- 5
thorhombic strain tensor (ε) to the fcc primitive
and the lower bound due to Reuss38
vectors:
  5(C11 − C12 )C44
x 0 0 GR = . (9)
4C44 + 3(C11 − C12 )
ε =  0 −x 02  (3)
x
0 0 1−x 2 Hill39 has shown that the Voigt and Reuss
bounds are rigorous upper and lower bounds.
for which the total energy is an even function The average shear modulus can be estimated
of the distortion parameter x for mechanically
stable systems: from these bounds
2 4
GV + GR
∆E(x) = ∆E(−x) = V (C11 − C12 )x + O[x ]. (4) G= . (10)
2
C44 is readily obtained by applying a volume- Alternatively, instead of the arithmetic aver-
conserving monoclinic strain tensor(ε) to the fcc age, one might prefer to use the geometric or
primitive vectors: harmonic means. In weakly anisotropic materi-
 x
0 2 0
 als, expectedly, the both bounds for G converge.
x The Young’s modulus E is connected to B and
ε = 2 0 0  (5)
x2 G through
0 0 4−x2
9BG
where the total energy dependence has the form E= . (11)
3B + G
1 2 4
∆E(x) = ∆E(−x) =
2
V C44 x + O[x ]. (6) Compositional dependency of both elastic con-
stants and polycrystalline moduli of Ni-Cr and
The distortion parameter x varies between Ni-Re alloys are obtained by the above men-
0 and ±0.05, in accordance with Mehl’s tioned methods. The use of the CPA-based
prescription32 . It is well known that cubic sin- methods allows us to follow the composition
gle crystals have elastic anisotropy35 , i.e. the dependence of these properties in random al-
Young modulus E and the shear modulus G loys keeping the homogeneity of the systems and
may depend on direction. The shear modulus preserving their point symmetry11,12 .
anisotropy or the Zener anisotropy index 36 for
cubic crystals is usually expressed as the ratio
between
0
C44 and the tetragonal shear modulus
C = (C11 − C12 ) /2:
2C44
AG = . (7)
(C11 − C12 )
 5

III. STRUCTURAL AND ENERGETIC

PROPERTIES

A. Lattice parameters

Lattice parameters of random fcc Ni-Re, Ni-

Cr and bcc Cr-Re alloys have been studied
experimentally40–46 and theoretically47,48 . In
Fig. 1, we compare our results for the 0 K lattice
constant, a0 , with other theoretical and experi-
mental investigations. It is clear that the com-
position dependence of the lattice parameter in
all alloys is in good agreement with the room
temperature (RT) experimental data. The LDA
full potential results by Mei et al.47 are substan-
tially lower, consistently with the known ten-
dency of local density approximation (LDA) to
underestimate the lattice constants of the 3d-
metals. On the other hand, the GGA some-
what overestimates the lattice constant of Ni in
the FM state. Our calculated value at zero K,
3.525 Å , is actually very close to the RT ex-
perimental data, while the experimental zero K
lattice parameter of Ni is about 3.515 Å49 . So,
the expected DFT value without a contribution
from zero-point vibrations should be 3.508 Å 50 .
However, the error still is only about 0.4 %, and
consequently the the variation of a0 under al-
loying is more significant than the temperature-
induced drift between zero and room tempera-
tures
The composition dependence of the lattice
constant in Ni-Cr random alloys exhibits some
deviations from the linear behaviour in the FM
FIG. 1. EMTO-calculated zero-K lattice constants
state where the slope changes from 8.32·10−4 in (a) Ni-Re, (b) Ni-Cr and (c) Cr-Re alloys com-
Å/(at.%Cr) in the composition interval of 0 – pared with other theoretical (Ref. 47 and 48) and
6.25 at.%Cr, to 2.08·10−4 Å/(at.%Cr) between experimental (Refs. 40–46) studies. The inset in the
6.25 and 18.75 at%.Cr, and finally to 9.56·10−4 middle panel shows the difference between FM and
Å/(at.%Cr) above 18.75 at.%Cr where the al- NM lattice constants in relative units.
loys become fully non-magnetic. To better un-
derstand this variation, we plot the relative
difference of the spin-polarized and the non- in particular, to the finite local magnetic mo-
spin polarized calculated lattice parameters as ments on Ni atoms, which create additional
a function of Cr composition. It exhibits strong magnetic pressure. The lattice parameter for
nonmonotonous variation peaking up at 6.25 the bcc Cr-Re alloys increases with the Re com-
at.%Cr and then gradually decreasing and it position exhibiting a smooth linear variation fol-
levels off to 0 at 25 at.%Cr. The variation of the lowing Vegard’s law. Our NM and AFM results
slope in the FM state is related to magnetism, are in excellent agreement with the FLMTO
 6

calculations48 and experimental data44,45 . that there is also a local minimum in Re-rich
hcp Ni-Re alloys. Unfortunately, it is very lit-
tle known about atomic configuration of these
B. Formation Enthalpies alloys.
As is seen in Fig.2 (a,b), the enthalpy as a
The formation enthalpies H of Ni1−x Mx function of concentration in Ni-Cr alloys passes
(M=Re,Cr) random alloys in the φ structure, through a maximum. This is consistent with ex-
which can be fcc, bcc or hcp, are defined as: perimentally known limited solubility of Ni in
bcc-Cr3,52,53 . The formation enthalpies of the
N i1−x Crx N i1−x Crx
Hφ = Eφ − (12) whole set of phases of Cr-Re alloys are positive
and vary almost linearly over the whole compo-
(1 − x)EfNcci − xEbcc
Cr
;
sition range, implying a phase separation.
N i1−x Rex N i1−x Rex
Hφ = Eφ − We must emphasize here that fcc Cr, bcc Ni, bcc
Re, hcp Cr are dynamically or elastically unsta-
(1 − x)EfNcci Re
− xEhcp ,
ble according to our elastic study, see Sec. IV,
and those of Cr-Re random alloys: and to other theoretical works30 . Moreover, the
alloys of these phases in Ni, Cr and Re-rich sides
Cr1−x Rex Cr1−x Rex
Hφ = Eφ − (13) are possibly unstable. In fact, we do find in
Cr Re Sec. IV Cr-rich Ni-Cr fcc phase elastically un-
(1 − x)Ebcc − xEhcp .
stable.
where Eφ are the total energies in the φ struc- Recently55 , the enthalpies of fcc and bcc Ni-Cr
ture. In the case of Ni-M alloys, the calcula- alloys were calculated using cluster expansion
tions have been done in the FM and NM states; (CE) method. However, we could not comapre
however, bcc Ni-Cr enthalpies have been addi- them with our data because we can not distin-
tionally calculated in the AFM state. We found guish which of their results correspond to or-
the energy difference between NM and AFM bcc dered and which to random structures.
Cr to be as small as ∼ 1 kJ/mol. We have set
the AFM (NM) bcc, FM (NM) fcc, hcp to be
the ground states of Cr, Ni, Re, respectively, IV. MECHANICAL STABILITY AND
ELASTIC MODULI
corresponding to low (high) temperature range
T < TC or TN (T > TC or TN ). The enthalpies
of formation in the NM state can be roughly Having investigated the thermodynamic
compared with experimental ones in the para- properties of the different solid solutions in the
magnetic state; qualitatively, they demonstrate previous section, we examine here the compo-
the energetic effect of magnetism when com- sitional dependency of elastic properties. Since
pared to their ferromagnetic counterparts. In the realistic alloys are polycrystalline, we com-
Fig. 2 (b,d), we show the formation enthalpies bine ab initio calculations with convenient av-
of NM Ni-M alloys, which correspond to the eraging techniques, introduced earlier in Sub-
high temperature paramagnetic state. One can sec. II B. The computed single-crystal elastic
see that in fcc and hcp Ni-Cr and Ni-Re al- constants C11 , C12 , C44 related to fcc lattice,
loys, there is a negative local minimum of the Born’s conditions of stability, the derived poly-
enthalpies of formations in the Ni-rich region, crystalline bulk modulus B, shear modulus G,
which is absent in the case of FM alloys, shown and Young’s modulus E are shown in Fig. 3.
in Fig. 2 (a,c). These results are in accordance We apply the Born’s criteria68 to explore the
with the high-temperature experimental solu- mechanical stability of alloys. It is the criterion
bility of Re41 and Cr3,52,53 in the Ni matrix. that a cubic crystal can be stable against small
These local minima indicate the existence of homogeneous deformations if the energy is pos-
a pronounced ordering tendency in these sys- itive for all combinations of strain components.
tems. In this respect, it is interesting to note
 7

FIG. 2. Formation enthalpies at 0 K of completely random fcc, hcp, and bcc solid solutions without
contribution from local atomic relaxations. Available theoretical51 and experimental (1583 K52 , 1538 K53 )
enthalpies are shown for the system Ni-Cr. The unary enthalpies are from Ref. 54. The VASP-calculated
fcc enthalpies, for Ni-rich Ni-Cr alloys, are not visible because they coincide with those of EMTO.

Thereby, the restrictions imposed on the elastic ble through the whole compostion range, in-
stiffness constants are69 cluding fcc Re. Our prediction of mechanical
instability and stability of fcc Cr and Re, re-
C11 > 0 ; (14a) spectively, is in accordance with former theo-
C44 > 0 ; (14b) retical reports71,72 . Because of the mechanical
C11 − |C12 | > 0 ; (14c) instability of Cr-rich Ni-Cr fcc alloys, we do not
plot the elastic moduli for the unstable compo-
C11 + 2C12 > 0 . (14d)
sitions in Fig. 3.
We find that Born’s68 criteria is not fully sat- One can notice that all the elastic properties
isfied, namely C 0 is ≤ 0, for a wide compo- of Ni-Cr and Ni-Re displayed in Fig. 3 change
sitional range of fcc Ni-Cr phase (62.5 up to qualitatively in a similar way below ∼20 at. %Cr
100 at. %Cr). In fact, C 0 is known as Born or Re. Notably, there is an increase in the
instability30 . It represents the stiffness associ- bulk (B), tetragonal shear (C 0 ) and Young’s (E)
ated with a shear in <110> direction on a plane, moduli73 ; in fact, it is well-known that TM ele-
and in certain martensitic transformations, the ments, in their equilibrium structure, feature a
vanishing of this modulus constitutes a direct concave and roughly parabolic trend as a func-
trigger of a phase transition70 . In contrast, we tion of their d-electron number where elements
predict fcc Ni-Re alloys to be mechanically sta-
 8

fcc fcc
350 600
Elastic Constants Cij (GPa)

300 (a) (d)

500
250
200 400
150
300
100
C11
50 200
C12
0
C44 100
-50
0
(e)
1000
Born’s Stability (GPa)

600
(b)
400

200 500

0
C11
C44
-200 C11 - |C12| 0
C11 + 2C12
500
(c) (f)
Elastic Moduli (GPa)

200 400
B
150 G 300
E
100 200

50 100

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Ni Cr Ni Re
FIG. 3. Single-crystalline elastic constants (a,d) C11 , C12 and C44 related to fcc lattice, Born’s criteria
of stability (b,e), polycrystalline (c,f) bulk (B), shear (G) and Young’s (E) moduli for NM fcc Ni-Cr and
Ni-Re, calculated by the EMTO method. The blue dashed curve in panel (b) is an adaptation of the C 0
tetragonal shear modulus as calculated in Ref. 51.
 9

TABLE I. Structural and elastic properties of fcc and bcc alloys and pure elements, based on the computa-
tional and experimental (Exp.) methods. The unit of lattice parameters is Å and that of elastic constants
is GPa.
Element Phase Method a C0 C11 C12 C44 B AG
Ni fcc-FM Exp.59 3.524 50.2 252 152 123 185.87 2.46
Ni fcc-FM Exp.60 0 K N/A 55 261 151 132 187.6 2.4
Ni fcc-FM Exp.60 300 K N/A 50.4 250.8 150 123.5 183.6 2.45
Ni fcc-FM Our Work (EMTO) 3.525 49 266 167 155 200 3.13
Ni fcc-FM PAW61 3.508 58 276 159 131 198 0.84
Ni fcc-FM FLAPW62 3.520 66 287 155 150 199 2.27
Ni fcc-NM Our Work (EMTO) 3.518 30 242 181 168 201 5.45
Ni fcc-NM FLAPW62 3.51 83 323 162 39 207 0.48
Ni-3.125 at. %Re fcc-NM Our Work (EMTO) 3.5381 35 253 183 159 206 4.54
Ni-3.125 at. %Re fcc-NM PAW-VASP63 3.5333 59 281 162 126 201 2.11
Ni-3.125 at. %Re fcc-FM Our Work (EMTO) 3.5442 92 263 171 153 201 3.33
Ni-3 at. %Re fcc-FM PAW-VASP64 3.53 46 218 126 131 157 2.85
Ni-3.125 at. %Cr fcc-FM Our Work (EMTO) 3.5286 47 254 159 157 190 3.29
Ni-3.125 at. %Cr fcc-FM PAW-VASP63 3.5202 60 283 162 131 199 2.16
Ni-3.125 at. %Cr fcc-AFM PAW-VASP63 3.5205 60 281 161 130 196 2.16
Cr bcc-AFM Exp.57 (77 K) 2.8792 151 391 90 103 190 0.68
Cr bcc-AFM Exp.65 (150 K) N/A 145 375 85 101 182 0.69
Cr bcc-NM Our Work (EMTO) 2.85 193 509 121 88 251 0.46
Cr bcc-NM PAW61 2.849 180 499 139 102 259 0.39
Cr bcc-NM FLMTO48 2.85 170 501 161 N/A 274 N/A
Cr bcc-NM FP-LAPW66 2.85 169 508 170 127 283 0.75
Cr bcc-NM PAW67 2.8792a 165 446 117 101 226 0.62
Cr fcc FP-LAPW66b 3.6200 -141 89 371 -76 277 —
Cr fcc Our Work (EMTO) 3.6286 -156 25 337 -72 233 —
a Although Ref.67 found abcc
N M = 2.851 Å, they computed the elastic constants , non-magnetically, at the
experimental lattice parameter abcc
AF M = 2.8792 Å.
b We have re-evaluated C11 and C12 starting form the calculations of Ref. 66, where we found C11 (C12 ) to be 89
(371) GPa in contrast to those reported in their paper66 , 371 (87) GPa.

with half d-band filling having the maximum site behavior beyond ∼20 at. %. While they
cohesive energies74–76 and the maximum elas- increase in Ni-Re, they decrease in Ni-Cr sys-
tic moduli(B 77,78 , C 079 ). Moreover, recently tem (vanishing at 62.5 at. %Cr). Though both
de Jong et al.80 have shown that alloying hcp bcc Cr and hcp Re are half d-band elements, Cr
Re with low composition of 3d, 4d and 5d so- is mechanically unstable (C 0 < 0) in fcc struc-
lutes induces a parabolic (concave) trend of B ture while Re is metastable. Therefore, when Cr
and G moduli with maximum values occuring at starts to populate the fcc Ni-matrix, its effect as
solutes with approximately half d-band filling. a solute changes completely from enhancing the
On the ground of the above mentioned obser- mechanical properties (G, E increase) to dete-
vations, the qualitatively similar change of the riorating these properties, causing instability of
whole elastic properties, shown in Fig. 3, and the fcc lattice.
particularly the increase in the moduli (B,C 0 ,E) We found for bcc Cr-Re alloys (cf. Fig. 4), that
can be understood, as Cr and Re have half d- the bulk modulus increases with Re composi-
band filling. Shear and Young’s moduli vari- tion in both AFM and NM calculation setups.
ation as function of Cr or Re manifest oppo- It is clear from Fig. 4 that the AFM calculation
 10

tively, they differ significantly as the solute com-

position increases. While there is much more
resistance for uniaxial (C11 ) and biaxial (C12 )
strains with Re as solute (cf. Fig. 5 (a,b)),
Cr addition makes Ni-matrix more resistive to
shear strain (C44 ) (cf. Fig. 5 (c)). On the other
hand, the magnetism in Ni-rich Ni-Re alloys has
a qualitative impact on the elastic constants. As
shown in Fig. 5 (a,b), ferromagnetic alloys re-
sist much more (less) a uni(bi)-axial strain.
Fig. 6 (a),(b) and (c) depict the change of the
Nickel polycrystalline moduli, B, G and E,
upon alloying with Re and Cr. The difference in
G and E between NM Ni-rich Ni-Re and Ni-Cr
is small, while it is highly pronounced in B and
FIG. 4. Bulk Modulus (B) for Cr-Re alloys, cal- reaches as far as 20 GPa at 20 at. %, indicat-
culated by the EMTO and VASP48 methods. The ing higher resistance to uniform compression,
symbols mark the experimental results56,57 . which is attributed to Re larger bulk modulus.
Since G and E are calculated using the elas-
tic constants Cij ’s, the effect of magnetism is
of bcc Cr-Re alloys matches the experimental prominent on their compositional dependency.
data obtained for paramagnetic alloys56 , par- As can be seen in Fig. 6 (a,b,c), FM alloys re-
ticularly for Cr-rich side. Note that these ex- sponse is more resistive than NM ones, to shear
perimental data56 were extrapolated from the (G) and uniaxial (E) stresses, but less resistive
low-temperature measured values down to 0 K, to uniform stress (B).
for the sake of comparison with our 0 K calcu- Based on the computed elastic constants and
lations. polycrystalline moduli one can derive indicative
mechanical characteristics, such as the ductil-
ity ratio(B/G)81 , Cauchy pressure (C12 − C44 ),
V. ELASTIC MODULI AND and elastic anisotropy constant (AG = C44 /C 0 ).
MECHANICAL PROPERTIES OF In fact, the Cauchy pressure parameter82 and
NICKEL-RICH ALLOYS B/G81 ratio are proposed as criteria to gauge
the intrinsic ductility, solely due to alloy chem-
In the previous section we examined the elas- istry, and they are regarded as approximative
tic stability and the polycrystalline moduli com- indicators of ductility/brittleness. In reality,
positional dependency through the whole range. extrinsic effects such as impurity, grain size
In this section we report results on the elastic distribution, interface/grain boundary, disloca-
and on some indicative mechancial properties of tions, Rice-Thompson ratio83 , and solute soft-
the Ni-rich Ni-M alloys, within their solubility ening effect84 , etc. also impact ductility signif-
limits3,41 . We perform EMTO calculations for icantly, but they lie beyond the scope of the
the composition range 0 - 20 at. % with 1 at. % present study.
step. Within this approximative approach, B/G ra-
Despite the modest wideness of the composi- tio higher than 1.7581 implies that the material
tion range studied here, there are important is ductile. According to this criterion, we find
differences in the elastic constants (cf. Fig. 5 (cf. Fig. 6 (d)) that Ni-rich alloys have B/G
(a-c)), to be noted, between NM Ni-rich Ni-Re ratio greater than 1.75, which means that they
and Ni-Cr alloys. Qualitatively, the variations behave in a ductile manner.
in Cij ’s are reasonably similar, while, quantita- Pettifor82 suggested that the Cauchy pressure
 11

FIG. 5. Elastic constants (a-c) C11 , C12 and C44 related to fcc lattice, (d) shear elastic constant C 0 , and
(e) anisotropy constant AG , calculated for the fcc Ni-rich alloys by the EMTO method.

= C12 − C44 parameter can be used to describe metallicity of the bonds of the Ni-rich matrix.
the angular character of atomic bonding in met- Other critical elastic moduli considered to have
als and intermetallics, which can be responsible significant implication in engineering science is
for the brittle or ductile behavior of materials. the elastic anisotropy AG of crystals, which is
It is typically negative for directional bonding, reported to highly correlate with the possibil-
i.e. the material in this case resists against the ity to induce micro-cracks in material.85 . For
shear strain (C44 ) more strongly than with the a completely isotropic material, AG takes the
volume change (C12 ), whereas it is positive for value of 1, while values smaller or greater than
metallic bonding. Indeed the Cauchy pressure unity measures the degree of elastic anisotropy.
is positive for most metals, since they are rela- Whilst Ni displays a fair degree of anisotropy,
tively ductile. As it is clearly shown in Fig. 6 it is not the most anisotropic of fcc metals. It is
(e), the value is positive and increases upon interesting to note, see Fig. 5 (e), that there is
alloying, pointing out an enhancement in the a continuous drop in the anisotropy of NM Ni
 12

FIG. 6. Polycrystalline moduli (a,b,c) bulk (B), shear (G) and Young’s (E), (d) ductility B/G, and (e)
Cauchy pressure calculated for the Ni-rich alloys. The inset in panel (c) compares our 0 K and the 77 K
experimental58 Young’s modulus variation as a function of Ni composition.

upon alloying with relatively low content of Re certain set of prescribed criteria are to be ful-
and Cr. filled and compromised86 . Depending on its ap-
It is known and commonly observed that an im- plication, an alloy, in general, is preferred to be
provement in a certain property of an alloy86 , lightweight, stiff and reasonably ductile. In or-
for instance stiffness, often comes at the expense der to compare the stiffness/ductility of Ni-rich
of a deterioration in another property, such as Ni-Cr and Ni-Re alloys, we plot in Fig. 7 the
ductility. Therefore, when designing alloys, a Young’s modulus vs ductility (B/G) ratio. The
 13

stability, of the solution phases fcc, hcp

and bcc in Ni-M(Cr,Re) and Cr-Re ran-
dom alloys, using effective medium and
supercell approaches within the context
of DFT, which helped us to monitor their
compositional dependency at infinitesimal
composition change. Our results are in
good agreement with the available exper-
imental values.
• The compositional variation of the en-
thalpies is nontrivial, it changes dra-
matically throughout the compositional
FIG. 7. Young’s modulus E variation as a function range and hence can not be captured
of B/G ratio for Ni-rich alloys 0 - 20 at.%. with a naı̈ve consideration of few highly-
stoichiometric compositions. Thereby,
the reported enthalpies are stimulating to
FM Ni-Re curve follows a typical behaviour, construct new phase diagrams of the sys-
an increase of ductility accompanied with a re- tems studied here.
duction of the stiffness, while Ni-Cr curve ex-
hibits an increase of stiffness at constant duc- • A combined first-principles – homogeniza-
tility. Surprisingly, there is an overall increase tion scheme, offered us an insight into the
of the NM Ni-rich matrix stiffness as a func- polycrystalline shear and Young moduli.
tion of B/G ratio and Re content, which we We found an interesting increase of both
find tempting to say that it seems it is possi- strength and B/G ratio – so called intrin-
ble to increase both stiffness and ductility by sic ductility – in NM Ni-rich Ni-Re fcc al-
changing only the composition of a binary alloy, loys. Experimentalists are invited to ex-
as far as, merely, alloy chemistry is concerned. plore this possibility.
This is due to the continuous decrease/increase
of E for FM/NM Ni-rich alloys, as shown in
Fig. 6(c). The decrease/increase in E for Ni- ACKNOWLEDGMENTS
rich alloys can be traced back to the decrease of
the response of these alloys to axial stress (see AB, SGF and MP acknowledge funding by
Fig. 5(a)). However, it remains an open ques- the Deutsche Forschungsgemeinschaft (DFG)
tion whether the simultaneous increase of stiff- through project C6 as the Collaborative Re-
ness and ductility can be realized at high tem- search Centre SFB/TR 103.
perature where the alloy is not simply nonmag-
netic and the disordered Ni magnetic moments
exist due to temperature induced longitudinal
spin fluctuations31 .

VI. CONCLUSIONS

• We have modeled the lattice parameters,

enthalpies, elastic constants and elastic
 14

∗ 15
corresponding author; [email protected]; B. L. Gyorffy, Phys. Rev. B 5, 2382–2384 (Mar
Moved to: School of Metallurgy and Materials 1972).
16
- University of Birmingham. A. V. Ruban and H. L. Skriver, Phys. Rev. B 66,
1
in Rhenium and Rhenium Alloys, edited by 024201 (Jun 2002).
17
D. Bryskin (Warrendale, PA, 1997) pp. 629–728. A. V. Ruban, S. I. Simak, P. A. Korzhavyi, and
2
M. Rahaman, B. Johansson, and A. V. H. L. Skriver, Phys. Rev. B 66, 024202 (Jun
Ruban, Phys. Rev. B 89, 064103 (Feb 2014), 2002).
18
http://link.aps.org/doi/10.1103/PhysRevB. I. A. Abrikosov, A. M. N. Niklasson, S. I. Simak,
89.064103. B. Johansson, A. V. Ruban, and H. L. Skriver,
3
P. Nash, Bulletin of Alloy Phase Diagrams 7, Phys. Rev. Lett. 76, 4203–4206 (May 1996).
19
466–476 (1986), http://dx.doi.org/10.1007/ I. A. Abrikosov, S. I. Simak, B. Johansson, A. V.
BF02867812. Ruban, and H. L. Skriver, Phys. Rev. B 56,
4
Metals Reference Book, fifth edition ed., edited 9319–9334 (Oct 1997).
20
by C-J. Smithells (Butterworth-Heinemann, O. E. Peil, A. V. Ruban, and B. Johansson, Phys.
1976). Rev. B 85, 165140 (Apr 2012).
5 21
O. Levy, M. Jahnátek, R-V. Chepulskii, G-L. W. J. P. Perdew, K. Burke, and M. Ernzerhof, Phys.
Hart, and S. Curtarolo, Journal of the American Rev. Lett. 77, 3865–3868 (Oct 1996).
22
Chemical Society 133, 158–163 (2011), http:// H-J. Monkhorst and J-D. Pack, Phys. Rev. B 13,
dx.doi.org/10.1021/ja1091672. 5188–5192 (Jun 1976), http://link.aps.org/
6
S. B. Maisel, N. Schindzielorz, A. Mottura, doi/10.1103/PhysRevB.13.5188.
23
R-C. Reed, and S. Müller, Phys. Rev. B 90, The code can be obtained from A. V. Ruban
094110 (Sep 2014), http://link.aps.org/doi/ ([email protected]) upon request.
24
10.1103/PhysRevB.90.094110. P. E. Blöchl, Phys. Rev. B 50, 17953–
7
M. Palumbo, T. Abe, S. G. Fries, and A. Pas- 17979 (Dec 1994), http://link.aps.org/doi/
turel, Phys. Rev. B 83, 144109 (Apr 2011), 10.1103/PhysRevB.50.17953.
25
http://link.aps.org/doi/10.1103/PhysRevB. G. Kresse and J. Hafner, Phys. Rev. B 47, 558–
83.144109. 561 (Jan 1993), http://link.aps.org/doi/10.
8
L. Vitos, H.L. Skriver, B. Johans- 1103/PhysRevB.47.558.
26
son, and J. Kollár, Computational Ma- G. Kresse and J. Hafner, Phys. Rev. B 49,
terials Science 18, 24 – 38 (2000), 14251–14269 (May 1994), http://link.aps.
http://www.sciencedirect.com/science/ org/doi/10.1103/PhysRevB.49.14251.
27
article/pii/S0927025699000981. G. Kresse and J. Furthmüller, Computa-
9
L. Vitos, Phys. Rev. B 64, 014107 (2001). tional Materials Science 6, 15 – 50 (1996),
10
L. Vitos, I. A. Abrikosov, and B. Johans- http://www.sciencedirect.com/science/
son, Phys. Rev. Lett. 87, 156401 (Sep article/pii/0927025696000080.
28
2001), http://link.aps.org/doi/10.1103/ J. M. Cowley, Journal of Applied Physics 21,
PhysRevLett.87.156401. 24–30 (1950), http://scitation.aip.org/
11
V. I. Razumovskiy A. V. Ruban and P. A. content/aip/journal/jap/21/1/10.1063/1.
Korzhavyi, Phys. Rev. Lett. 107, 205504 (Nov 1699415.
29
2011), http://link.aps.org/doi/10.1103/ B.E. Warren, X-ray diffraction (New York,
PhysRevLett.107.205504. Dover, 1990).
12 30
V. I. Razumovskiy, A. V. Ruban, and P. A. G. Grimvall, B. Magyari-Köpe, V. Ozoliņš,
Korzhavyi, Phys. Rev. B 84, 024106 (Jul 2011), and K. A. Persson, Rev. Mod. Phys. 84, 945–
http://link.aps.org/doi/10.1103/PhysRevB. 986 (Jun 2012), http://link.aps.org/doi/10.
84.024106. 1103/RevModPhys.84.945.
13 31
P. Hohenberg and W. Kohn, Phys. Rev. 136, A. V. Ruban, S. Khmelevskyi, P. Mohn, and
B864–B871 (Nov 1964). B. Johansson, Phys. Rev. B 75, 054402 (Feb
14
W. Kohn, Rev. Mod. Phys. 71, 1253–1266 (Oct 2007).
32
1999), http://link.aps.org/doi/10.1103/ J. M. Mehl, Phys. Rev. B 47, 2493–2500 (Feb
RevModPhys.71.1253. 1993), http://link.aps.org/doi/10.1103/
 15

PhysRevB.47.2493. http://www.sciencedirect.com/science/
33
C.-L. Fu and K.-M. Ho, Phys. Rev. B 28, 5480– article/pii/0011227577901308.
50
5486 (Nov 1983), http://link.aps.org/doi/ P. Haas, F. Tran, and P. Blaha, Phys. Rev. B 79,
10.1103/PhysRevB.28.5480. 085104 (Feb 2009), http://link.aps.org/doi/
34
F. D. Murnaghan, Proc. Nat. Acad. Sci. 30, 244– 10.1103/PhysRevB.79.085104.
51
247 (1944). P.J. Craievich and J.M. Sanchez, Computa-
35
S. I. Ranganathan and Martin Ostoja- tional Materials Science 8, 92 – 99 (1997),
Starzewski, Phys. Rev. Lett. 101, 055504 http://www.sciencedirect.com/science/
(Aug 2008), http://link.aps.org/doi/10. article/pii/S0927025697000219.
52
1103/PhysRevLett.101.055504. A. Watson and F.H. Hayes, Journal of Alloys
36
C. Zener, Elasticity and Anelasticity of Metals and Compounds 220, 94 – 100 (1995), ISSN
(University of Chicago, Chicago, 1948). 0925-8388, proceedings of the 5th Interna-
37
W. Voigt, Annalen der Physik 274, 573–587 tional Meeting on Thermodynamics of Alloys,
(1889), ISSN 1521-3889, http://dx.doi.org/ http://www.sciencedirect.com/science/
10.1002/andp.18892741206. article/pii/0925838894060088.
38 53
A. Reuss, ZAMM - Journal of Applied Math- W. A. Dench, Trans. Faraday Soc. 59, 1279–
ematics and Mechanics / Zeitschrift fr Ange- 1292 (1963), http://dx.doi.org/10.1039/
wandte Mathematik und Mechanik 9, 49–58 TF9635901279.
54
(1929), ISSN 1521-4001, http://dx.doi.org/ Y. Wang, S. Curtarolo, C. Jiang, R. Ar-
10.1002/zamm.19290090104. royave, T. Wang, G. Ceder, L.-Q. Chen,
39
R. Hill, Proceedings of the Physical Society. Sec- and Z.-K. Liu, Calphad 28, 79 – 90 (2004),
tion A 65, 349 (1952), http://stacks.iop.org/ http://www.sciencedirect.com/science/
0370-1298/65/i=5/a=307. article/pii/S0364591604000392.
40 55
A. S. Pavlovic, V. Suresh Babu, and M. S. Jan S. Wróbel, Duc Nguyen-Manh, Mikhail Yu.
Seehra, Journal of Physics: Condensed Mat- Lavrentiev, Marek Muzyk, and Sergei L. Du-
ter 8, 3139 (1996), http://stacks.iop.org/ darev, Phys. Rev. B 91, 024108 (Jan 2015),
0953-8984/8/i=18/a=007. http://link.aps.org/doi/10.1103/PhysRevB.
41
K. Yaqoob and J.-C. Joubert, Journal of 91.024108.
56
Solid State Chemistry 196, 320–325 (2012), H.L. Alberts, Physica B: Condensed Matter 161,
http://www.sciencedirect.com/science/ 87 – 90 (1990), http://www.sciencedirect.
article/pii/S0022459612004112. com/science/article/pii/0921452689901105.
42 57
S. A. Khan, A. B. Ziya, S. Atiq., and A. Aziz, D. I. Bolef and J. de Klerk, Phys. Rev. 129,
Phys. Scr. 84, 065603 (2011), http://stacks. 1063–1067 (Feb 1963), http://link.aps.org/
iop.org/1402-4896/84/i=6/a=065603. doi/10.1103/PhysRev.129.1063.
43 58
L. Karmazin, Czech. J. Phys. B 29, 1181 (1979). Y. Mishima, S. Ochiai, N. Hamao, M. Yodogawa,
44
Y. Nishihara, Y. Yamaguchi, T. Kohara, and T. Suzuki, Transactions of the Japan Insti-
and M. Tokumoto, Phys. Rev. B 31, 5775– tute of Metals 27, 656–664 (1986).
59
5781 (May 1985), http://link.aps.org/doi/ J. R. Neighbours, F. W. Bratten, and C-S.
10.1103/PhysRevB.31.5775. Smith, Journal of Applied Physics 23, 389–393
45
M. Garfinkle, Metallurgical Transactions 1, 1062 (1952), http://scitation.aip.org/content/
(1970). aip/journal/jap/23/4/10.1063/1.1702218.
46 60
A. Nash and P. Nash, Bull. Alloy Phase Diagr. C. Kittel, Introduction to Solid State Physics,
6, 348–350 (Nov 1985). 7th ed. (Wiley, New York, 1996, 1996).
47 61
J. Mei, B.R. Cooper, Y.G. Hao, and F.L. Van M. de Jong, W. Chen, T. Angsten, A. Jain,
Scoy, in Alloy Modeling and Design, edited by R. Notestine, A. Gamst, M. Sluiter, C. Ande
G.M. Stocks and P.E.A. Turchi (TMS, 1993) pp. Krishna, S. van der Zwaag, J. J Plata, C. To-
165–174. her, S. Curtarolo, G. Ceder, K. A Persson, and
48
N. I. Medvedeva, Yu. N. Gornostyrev, and A. J. M. Asta, Scientific data 2 (2015), http://dx.
Freeman, Phys. Rev. B 67, 134204 (Apr 2003), doi.org/10.1038/sdata.2015.9.
62
http://link.aps.org/doi/10.1103/PhysRevB. G. Y. Guo and H. H. Wang, Chin. J. Phys. 38,
67.134204. 949 (2000b).
49 63
J. Bandyopadhyay and K.P. Gupta, Cryogen- D. Kim, S-L. Shang, and Z-K. Liu,
ics 17, 345 – 347 (1977), ISSN 0011-2275, Computational Materials Science 47,
 16

80
254 – 260 (2009), ISSN 0927-0256, M. de Jong, D. L. Olmsted, A. van de
http://www.sciencedirect.com/science/ Walle, and M. Asta, Phys. Rev. B 86,
article/pii/S0927025609003048. 224101 (Dec 2012), http://link.aps.org/doi/
64
N. I. Medvedeva and A. L. Ivanovskii, Nanosys- 10.1103/PhysRevB.86.224101.
81
tems: Physics, Chemistry, Mathematics 5, 486– S. F. Pugh, Philosophical Magazine Series 7 45,
493 (2014). 823–843 (1954), http://dx.doi.org/10.1080/
65
K. W. Katahara, M. Nimalendran, M. H. 14786440808520496.
82
Manghnani, and E. S. Fisher, Journal of Physics D. G. Pettifor, Materials Science and Technology
F: Metal Physics 9, 2167 (1979), http:// 8, 345–349 (1992), http://www.maneyonline.
stacks.iop.org/0305-4608/9/i=11/a=008. com/doi/abs/10.1179/mst.1992.8.4.345.
66 83
G. Y. Guo and H. H. Wang, Phys. Rev. B 62, C. B. Geller, R. W. Smith, J. E. Hack, P. Saxe,
5136–5143 (Aug 2000), http://link.aps.org/ and E. Wimmer, Scripta Materialia 52, 205
doi/10.1103/PhysRevB.62.5136. – 210 (2005), http://www.sciencedirect.com/
67
“Chromium: Structure and elastic properties,” science/article/pii/S1359646204005536.
84
Application Note, Materials Design, inc. 2002- D. R. Trinkle and C. Woodward, Science 310,
2008. 1665–1667 (2005), http://www.sciencemag.
68
M. Born and K. Huang, Dynamical Theory of org/content/310/5754/1665.abstract.
85
Crystal Lattices (Clarendon Press, Oxford, UK, V. Tvergaard and J. W. Hutchinson, Journal
1954) p. 141. of the American Ceramic Society 71, 157–166
69
G. A. Korn and T. M. Korn, Mathematical Hand- (1988), ISSN 1551-2916, http://dx.doi.org/
book for Scientists and Engineers (McGraw-Hill, 10.1111/j.1151-2916.1988.tb05022.x.
86
1968). M. F. Ashby, Materials Selection in Mechanical
70
Y. Tsunoda, N. Oishi, and N. Kunitomi, Phys- Design, 4th ed. (Butterworth-Heinemann, 2011).
ica B+C 119, 51–55 (1983), ISSN 0378-4363,
http://www.sciencedirect.com/science/
article/pii/037843638390164X.
71
P. J. Craievich, M. Weinert, J. M. Sanchez,
and R. E. Watson, Phys. Rev. Lett. 72, 3076–
3079 (May 1994), http://link.aps.org/doi/
10.1103/PhysRevLett.72.3076.
72
K. Persson, M. Ekman, and G. Grimvall,
Phys. Rev. B 60, 9999–10007 (Oct 1999),
http://link.aps.org/doi/10.1103/PhysRevB.
60.9999.
73
This can be seen more clearly in Fig. 5 and Fig. 6.
74
M. Cyrot and F. Cyrot-Lackmann, Journal of
Physics F: Metal Physics 6, 2257 (1976), http:
//stacks.iop.org/0305-4608/6/i=12/a=012.
75
F. Cyrot-Lackmann, Journal of Physics and
Chemistry of Solids 29, 1235 – 1243 (1968).
76
J. Friedel, in Physics of Metals, edited by J. M.
Ziman (Cambridge University Press, New York,
1969).
77
D. G. Pettifor and R. Podloucky, Journal of
Physics C: Solid State Physics 19, 315 (1986),
http://stacks.iop.org/0022-3719/19/i=3/
a=003.
78
J. H. Rose and H. B. Shore, Phys. Rev. B 49,
11588–11601 (May 1994).
79
J. M. Wills, Olle Eriksson, Per Söderlind, and
A. M. Boring, Phys. Rev. Lett. 68, 2802–
2805 (May 1992), http://link.aps.org/doi/
10.1103/PhysRevLett.68.2802.