Available online at www.sciencedirect.

com

Acta Materialia 61 (2013) 2919–2925

www.elsevier.com/locate/actamat

Large-scale first-principles determination of anisotropic

mechanical properties of magnetostrictive Fe–Ga alloys
Hui Wang a,b,c,d, Z.D. Zhang a,b, R.Q. Wu d, L.Z. Sun c,⇑
a
Shenyang National Laboratory of Materials Science, Institute of Metal Research, Chinese Academy of Sciences,
Shenyang 110016, People’s Republic of China
b
International Centre of Materials Physics, Chinese Academy of Sciences, Shenyang 110016, People’s Republic of China
c
Department of Civil and Environmental Engineering, University of California, Irvine, CA 92697-2175, USA
d
Department of Physics and Astronomy, University of California, Irvine, CA 92697-4575, USA

Received 29 December 2012; received in revised form 21 January 2013; accepted 22 January 2013

Abstract

Magnetostrictive Fe1xGax alloys, as a new class of smart materials, have great potential in sensing and actuator applications. How-
ever, the fundamental understanding of the anisotropic elastic responses at high Ga concentration remains one of the most challenging
problems for the binary alloys. Here, we apply the density functional theory and large-scale ab initio molecular dynamics simulation to
investigate the effect of high Ga concentration on the elastic anisotropy of the Fe–Ga alloys with supercell models obtained by non-linear
and non-uniform annealing processes. It is demonstrated that the formation of D03-like structures has an important effect on the softness
of the tetragonal shear modulus and a negligible influence on the rhombohedral shear modulus. Meanwhile, the Fe dangling bond to its
nearest Ga atoms results in a decrease in the Young’s modulus and the negative Poisson’s ratio in the [1 1 0] direction. The improved
Young’s modulus in the [1 1 0] direction compared to that in the [1 0 0] direction is attributed to the different arrangement of the pure
Fe layer and the Fe–Ga mixed layer along the [1 1 0] and [1 0 0] axes. Furthermore, the ductility of Fe1xGax alloys is enhanced at high
Ga content, playing a key role in the enhanced magnetostriction.
Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Fe–Ga alloys; Density functional theory; Mechanical property; Anisotropy; Magnetostrictive alloys

1. Introduction MnGa Heusler shape memory alloys [6,7] exhibit large

strains up to 6%, which primarily stems from a magnetic-
Materials exhibiting a coupling between magnetization field-induced martensitic phase transition. However, the
and strain are magnetostrictive materials, which play an applications of these alloys have been limited owing to
increasingly important role in various applications ranging their mechanical brittleness and their use of rare-earth ele-
from stress- and torque-sensing devices to actuator and ments (in the case of Terfenol-D). Exploration of a new
energy harvesters [1–3]. Such adaptive materials can class of materials with combined high magnetostriction
change their dimensions under an external magnetic field, and mechanical ductility is indeed needed.
and conversely, the magnetic state can be tailored by the In view of their electronic structure, ductile materials
application of a mechanical load. Terfenol-D (TbxDy1x- usually consist of metallic elements that form metallic
Fe2) alloys are well known for their giant magnetostriction bonds with delocalized electronic states. As one of the
up to a few thousands ppm, and have already been used in Fe-based magnetostrictive materials, Fe1xGax alloys
practical applications in the past few decades [4,5]. Ni2- (Galfenol) exhibit desirable mechanical properties and
large tetragonal magnetostriction under a lower magnetic
⇑ Corresponding author. Tel.: +1 949 824 8670; fax: +1 949 824 2117. field (150–250 Oe), ideal for sensor and actuator applica-
E-mail address: lsun@uci.edu (L.Z. Sun). tions [8]. Interdisciplinary research efforts have been dedi-

1359-6454/$36.00 Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.actamat.2013.01.046
2920 H. Wang et al. / Acta Materialia 61 (2013) 2919–2925

cated to understand the composition dependence of phase tion interaction of electrons, using the Perdew–Burke–Ern-
stabilities [9] and the magnetoelastic coupling mechanism zerhof (PBE) functional [21]. We treated Fe-3d4s4p and
of these alloys [10–14]. In particular, the weak elastic stiff- Ga-3d4s4p as valence states and adopted the projector-
ness at high Ga concentrations is assigned as one of the pri- augmented wave (PAW) method [22] to represent the effect
mary reasons for the enhancement of magnetostriction of of the ionic cores. The energy cutoff for plane-wave expan-
Fe1xGax [15]. Furthermore, due to the different atomic sion was set as 400 eV, which is sufficient for Fe–Ga sys-
arrangement in crystallographic axes, single-crystalline tems according to test calculations. At the end of AIMD
Fe1xGax alloys demonstrate a strong elastic anisotropy steps, we optimize the atomic structure further, with a cri-
[16]. This leads to auxeticity, i.e. negative Poisson’s ratio, terion that the atomic force on each atom becomes weaker
which is associated with indentation resistance, fracture than 0.01 eV Å1 and the energy convergence is better than
toughness and damping/sound absorption. A thorough, 105 eV.
systematic investigation of the anisotropic mechanical For high Ga concentration (15.6% < x < 22.7%) cases,
properties of Fe1xGax alloys is essential to understand based on our recent ab initio molecular dynamics simula-
the deformation mechanism responsible for the magneto- tion [12], we set up 10 initial structures randomly with
striction at high Ga concentration, especially near the first 128 Fe and Ga atoms taking the bcc lattice sites of the
magnetostriction peak. However, it is still a challenge to 4  4  4 cubic supercell for each Ga concentration. We
simulate actual materials around x = 19% through density heated the system up to 2000 K, which is far above the
functional theory (DFT) calculations, due to the size limit melting point of Galfenol (1500 K) for 15 ps. The conver-
of unit-cell models and complex chemical ordering of Ga gence of the total free energy (770 eV per cell) after
atoms in the body-centered cubic (bcc) lattice. 0.5 ps and the calculated Ga–Ga radial pair distribution
In this work, we report the results of anisotropic elastic function (RDF) demonstrates the liquid phase at 2000 K.
properties of Fe1xGax alloys using the DFT-based ab ini- We further cooled the system down to 500 K with non-uni-
tio molecular dynamics (AIMD) simulation. We observe form and non-linear cooling rates (especially about
that the formation of D03-like structures has an important 1012 K s1 near the melting point where the total free
effect on the softness of tetragonal shear modulus and a energy decreases abruptly) for the recrystallization. The
negligible influence on the rhombohedral shear modulus. calculated Ga–Ga RDF and the snapshot of atomic
The decrease in the Young’s modulus and the negative arrangement exhibit the crystallized phase at 500 K. The
Poisson’s ratio in the [1 1 0] direction can be attributed to structures were fully relaxed at 0 K. We eventually chose
the induced Fe dangling bond to its nearest Ga atoms. the most energetically favorable atomic configurations for
The calculated anisotropic elastic constants agree well with each Ga concentration. As an example, a supercell with
experimental data reported by different groups. Further- 128 atoms for Ga concentration only at 20.3% was illus-
more, we demonstrate that the enhanced ductility of the trated, shown in Fig. 1.
alloys with increased Ga content can be explained in terms For the integrity of this work, various sized unit cells
of the Cauchy pressure and the Pugh’s modulus ratio. were applied for Fe1xGax alloys at different Ga composi-
tions. For low Ga concentration (x 6 12.5%), the 16-atom
2. Computational methodology unit-cell model has proved to be accurate enough to char-
acterize the Ga distribution in Fe1xGax alloys [10]. Specif-
2.1. Density functional theory and ab initio molecular ically, only one Fe site at the center position (1/2, 1/2, 1/2)
dynamics

The molecular dynamics (MD) approach allows simula-

tion of dynamic processes in the framework of Newton’s
equation of motion. One of the most important aspects
of MD simulation is the calculation of the potential energy
surface or atomic forces. DFT-based electronic structure
calculations can accurately describe the multi-body interac-
tion in condensed matter systems by solving the Kohn–
Sham equations self-consistently [17]. Ab initio molecular
dynamics, unifying Newton’s and Schrödinger’s equations,
may simulate the finite-temperature dynamic process by
using forces computed from DFT calculations without
relying on any adjustable parameters [18,19].
We conducted the electronic-structure calculations and
Fig. 1. Atomic configurations for Fe1xGax alloys at x = 20.3%. The blue
structural relaxations of Fe1xGax alloys using the Vienna
and red balls represent Fe and Ga atoms, respectively. The green balls
Ab-initio Simulation Package (VASP) [20]. The spin-polar- depicted by Fe1 and Fe2 represent the Fe atoms in mixed Fe–Ga layer and
ized generalized gradient approximation (GGA) was pure Fe layer, respectively. (For interpretation of the references to color in
employed to describe the electronic exchange and correla- this figure legend, the reader is referred to the web version of this article.)
 H. Wang et al. / Acta Materialia 61 (2013) 2919–2925 2921

and two Fe sites at the original position (0, 0, 0) and center with V0 being the lattice volume at equilibrium stage. With
position (1/2, 1/2, 1/2) were replaced by the Ga atoms in the combination of Eqs. (3)–(6), the three independent elas-
the unit cell with 16 atoms, corresponding to Ga-6.25% tic constants C11, C12 and C44 can be determined.
and Ga-12.5%, respectively, while for high Ga concentra- Since C11, C12 and C44 compromise a complete set of
tion (12.5 < x 6 22.7%), the 128-atom supercell was elastic constants for a cubic system, the direction-depen-
selected from MD results. A 13  13  13 Monkhost–Pack dent Young’s modulus E and Poisson’s ratio m can be
and 4  4  4 gamma-centered k-point mesh in the Brillou- derived by the following equations:
in zone [23] was used to evaluate integrals in the reciprocal C 11 R
space for the 16-atom unit cell and 128-atom supercell, E½100 ¼ ð7Þ
C 11 þ C 12
respectively. Test calculation of K-point for Fe1xGax
alloys at x = 20.3% (128-atom supercell) indicates that 4C 44 R
E½110 ¼ ð8Þ
4  4  4 gamma-centered K-point mesh is accurate 2C 44 þ R
enough for the total energy and magnetic property calcula- C 12
m½010 ¼ ð9Þ
tions with the 128-atom supercell model. C 11 þ C 12
R  2C 44
2.2. Determination of elastic constants m½110 ¼ ð10Þ
R þ 2C 44
In the case of cubic crystalline alloys, symmetry permits where R = (C11  C12)(C11 + 2C12)/C11.
the relationship between the stresses and strains to be com-
pletely defined with the three elastic constants C11, C12 and 3. Results and discussion
C44, which can be determined based on the strain-depen-
dent strain energy [24]. In general, with infinitesimal strains One important factor responsible for the magnetostric-
(all six strain components e1, e2, e3, e4, e5, e6) applied to dis- tion of Fe1xGax alloys can be attributed to the effect of
tort the lattice within the volume V, the total strain energy additional Ga atoms on the mechanical properties of these
U is defined as: alloys, such as the tetragonal shear modulus C0 (C0 = (C11 -
 C12)/2) and the rhombohedral shear modulus C44 for
VX 6 X 6
tetragonal and rhombohedral magnetostriction, respec-
U¼ C ij ei ej ð1Þ
2 i¼1 j¼1 tively. In order to obtain these shear constants, the depen-
dence of the total energy on lattice volume, and the strain
and can be explicitly expressed as: energy density as a function of either tetragonal strain in
V the [1 0 0] direction or rhombohedral strain in the [1 1 1]
U¼ C 11 e21 þ C 11 e22 þ C 11 e23 þ 2C 12 e1 e2 þ 2C 12 e1 e3
2 direction, need to be determined first, as demonstrated in

þ2C 12 e2 e3 þ C 44 e24 þ C 44 e25 þ C 44 e26 ð2Þ Fig. 2. The smoothness of data indicates the high quality
and reliability of our calculations, and the anisotropic elas-
for a cubic symmetric system. tic constants can be acquired by fitting the data with the
Specifically, for rhombohedral distortion, the six strain quadratic polynomials based on Eqs. (3)–(5).
components are (0, 0, 0, d, d, d) and the strain energy The dependence of the calculated tetragonal and rhom-
can be simplified as: bohedral shear constant on Ga concentrations demon-
3 strated a very different behavior, as illustrated in Fig. 3a
U¼ C 44 d2 ð3Þ and b, which generally agrees well with the available exper-
2V
imental data in the literature [14,15,25,26]. Further, C0
Similarly, the tetragonal distortion (d, 0, 0, 0, 0, 0) in the showed a strong linear dependence of Ga concentration
[1 0 0] direction can yield the corresponding strain energy and decreased almost by a factor of five in the range stud-
as: ied here, while C44 exhibits weak dependence of Ga concen-
tration, which is also consistent with the form effect
1
U¼ C 11 d2 ð4Þ corrected results by using the electron-per-atom ratio (e/
2V a) as the common variable [27]. More importantly, to
Furthermore, the bulk modulus B of the lattice is associ- uncover the underlying reasons, detailed studies of the
ated with the elastic constants as: dependence of local chemical structures on shear constants
are indeed needed. Previous investigations of the micro-
C 11 þ 2C 12
B¼ ð5Þ structure of Fe1xGax alloys have been performed close
3
to the first peak in the magnetostriction [28], which reveal
which can be calculated from the second derivative of the the formation of D03-like structures due to the coherent
volume-dependent total energy Etotal as: nature of the A2 matrix and D03-like structures. Mean-
 while, recent theoretical calculations demonstrated that
d 2 Etotal 
B¼V ð6Þ the instability of the D03 structure can be removed by
dV 2 V¼V 0 applying a hydrostatic-type strain of 0.3% [11] or by
2922 H. Wang et al. / Acta Materialia 61 (2013) 2919–2925

Fig. 2. Quadratic dependence of total energy Etotal on lattice volume (a) and strain energy density U/V on tetragonal strain (b) and rhombohedral strain
(c) for Fe1xGax at x = 20.3%.

tent and almost kept a constant value between 125 and

130 GPa, which indicate that D03-like structures have neg-
ligible influence on the rhombohedral shear modulus.
To aid engineering application, a more detailed investi-
gation of the mechanical properties of Fe1-xGax alloys,
such as Young’s modulus and Poisson’s ratio, in certain
crystallographic axes is needed. Based on Eqs. (7)–(10),
the Young’s modulus E and Poisson’s ratio m can be deter-
mined by the anisotropic elastic constants obtained above.
As shown in Fig. 4, the AIMD simulation results qualita-
tively match all the experimental results [22,24] very well,
indicating the consistent electronic origin of Fe1xGax
microstructures. As demonstrated in Fig. 4a, the Young’s
moduli E[100] and E[110] in the directions of [1 0 0] and
[1 1 0] decrease with the increased Ga-composition while
E[110] is greater than E[100] indicating an anisotropic elastic
response. From Fig. 4b, the Poisson’s ratio m[010] in the
Fig. 3. DFT simulation and experimental data of dependence of (a) [0 1 0] direction almost keeps as constant, while m[110] in
tetragonal shear constant C0 , (b) rhombohedral shear constant C44 and (c)
the [1 1 0] direction drops from positive to negative at
local chemical structures, D03-like structures on the Ga concentration for
Fe1xGax alloys. The vertical dash line indicates the boundary between x = 7% and reaches a negative value of 0.67 at
low and high Ga concentration. x = 23%, indicating an auxetic response of Fe1xGax alloys
at high Ga concentration.
The anisotropy of the Young’s modulus comes from the
including small pocket of A2 phase throughout the struc- different arrangement of a pure Fe layer and a Fe–Ga
ture [29]. In this work, we calculated the radial distribution mixed layer in crystallographic axes. As Ga concentration
function (RDF) and integrated it over the whole supercell increases, Ga atoms prefer to stay away from each other
to obtain the number of D03-like structures for each Ga in the bcc-Fe matrix. Previous theoretical results demon-
concentration, as depicted in Fig. 3c. As we can see at strate that Fe atoms nearest to Ga atoms (Fe2 atoms shown
low Ga concentration (the left side of the vertical dash line in Fig. 1) provide with the largest contribution to magneto-
in Fig. 3c), no D03-like structures were observed, indicating elastic coupling in Fe1xGax alloys [10,11]. In Fig. 5a, the
that the randomly distributed Ga atoms in the bcc-Fe calculated density of states shows that there is a pro-
matrix is probably the main structure in this range, where nounced non-bonding peak of Fe2 atom near around
C0 almost decreased to the half value of pure bcc-Fe and Fermi level in contrast to the valley at the same energy
C44 increased by 15–20%. With the continuous increasing range for Fe1 atoms (Fe1 atom shown in Fig. 1), indicating
of Ga content, D03-like structures increased significantly the broken Fe bond toward nearest Ga atoms. Due to the
and gradually increased at higher Ga concentration, as weak hybridization of Fe and Ga atoms, Fe bonds toward
shown in the right side of the vertical dash line of Ga atoms become dangling bonds and dxz,yz states shift in
Fig. 3c. The linear decreasing of C0 is continuous until it energy from the edge to the center of the 3d band, which
reaches non-zero minimum in this range; this is probably reduces the total number of interlayer Fe–Fe bonds. As
due to the fact that the formation of D03-like structure seen in Fig. 5b, for the schematic illustration of the atomic
induced more dangling bonds for Fe atoms adjacent to arrangement of Fe1xGax alloy at x = 20.3% in the ½110 
Ga atoms. Differently, C44 is almost insensitive to Ga con- direction, The Fe–Fe bond is separated by a Fe–Ga mixed
 H. Wang et al. / Acta Materialia 61 (2013) 2919–2925 2923

Fig. 4. Comparison between DFT simulation (filled symbols) and experimental data (open symbols) of (a) Young’s modulus (E[100] and E[110]) and (b)
Poisson’s ratio (m[010] and m[110]) for Fe1xGax alloys.

Fig. 5. (a) Partial density of states (PDOS) of Fe1 and Fe2 atoms for Fe1xGax alloys at x = 20.3% and (b) atomic configuration of Fe1xGax alloy at
x = 20.3% viewed from ½110  direction. The green arrow indicates the non-bonding states induced by Fe2 atom.

layer in the [0 0 1] direction, which decreases the interaction low Young’s modulus at high Ga concentrations
between Fe–Fe layers. The Fe–Fe bond can hardly with- (Fig. 4a). Meanwhile, with the introduction of the elastic
stand stress if the stress is in the [0 0 1] direction, and will anisotropy constant A (A = 2C44/(C11  C12)) [30], the
also cause a significant drop for the tetragonal shear mod- Poisson’s ratio m[110] in the [1 1 0] direction can be rewritten
ulus C0 (Fig. 3a), while the ability to withstand stress in the as:
[1 1 0] direction is obviously stronger because of the exis-
1  A C3B11
tence of more Fe–Fe bonds, as illustrated in Fig. 5b. This m½110 ¼ ð13Þ
is also the reason why the Young’s modulus in the [1 1 0] 1 þ A C3B11
direction is larger than that in the [0 0 1] direction. The Since the anisotropy constant A demonstrates a gradual
dependence of the Young’s moduli in the [1 1 0] and [0 0 1] increment at low Ga concentration and a burst at the Ga
directions on the tetragonal shear modulus C0 can be concentration x = 15.6% shown in Fig. 6 consistent with
derived from Eqs. (7) and (8) as: the experimental data [24,26], we may note that the sign
6C 0 B change of the Poisson’s ratio m[110] occurs when the anisot-
E½100 ¼ ð11Þ ropy constant A is greater than 2, which is in agreement
C 11 þ C 12
with the suggestion of Jain and Verma [31], who attributed
and the negative Poisson’s ratio to materials exhibiting anisot-
4C 44 ropy constant larger than 2. Eq. (13) shows that the Pois-
E½110 ¼ ð12Þ
C 44
1 þ 3C 0
son ratio in the [1 1 0] direction is negative when A > 3B/
B
C11 or C0 < C11C44/3B. Therefore the drop of tetragonal
Since the rhombohedral shear modulus C44 (Fig. 3b) and shear modulus C0 plays an important role for the decreased
bulk modulus B (doesn’t show here) are almost constant Poisson’s ratio and final reach of large magnitude of the
for Fe1xGax at 15% 6 x 6 22.7%, the sharply decreased negative Poisson’s ratio in the [1 1 0] direction (Fig. 4b).
tetragonal shear modulus can be a primary factor for the Such a large negative Poisson’s ratio can exhibit a signifi-
2924 H. Wang et al. / Acta Materialia 61 (2013) 2919–2925

ing mechanism. The Pugh’s modulus ratio G/B reflects the

competition between the shear and cohesive strength at the
crack tip of fracture. It is shown that all calculated G/B val-
ues are smaller than 0.57, demonstrating a ductile property
from the Pugh’s criterion. In order to clarify how good the
ductility of Fe1xGax alloys is, we illustrated a few typical
ductile materials of (Au, Ag), brittle material (Si), and
super-hard materials (cubic BN and diamond) for compar-
ison [40–42]. As shown in Fig. 7a, we can see that ductile
Au and Ag locate in the upper left corner in relationship
of (C12  C44) against G/B, while hard cubic BN and dia-
mond almost lie in the bottom right corner. Unlikely, Fe
and Si are almost close to the transitional edge of ductile-
to-brittle transition. Fe1xGax alloys locate right between
Fig. 6. DFT simulation and experimental data of dependence of aniso- pure Ag and Fe, indicating an improved ductility with
tropic constant on the Ga concentration for Fe1xGax alloys.
additional Ga in Fe. The inset of Fig. 7a explicitly shows
a gradually increasing ductility of Fe1xGax alloys with
cant in-plane auxetic response if the alloys are designed un- increasing Ga concentration. Meanwhile, the relationship
der specific loading conditions. between (C12  C44) and G/B can be roughly fitted by qua-
It is worth noting that the excellent ductility of Fe1x- dratic curve for various Ga concentrations. To quantify a
Gax alloys can be further supported with both the classic relationship of the ductile-to-brittle transition, we renor-
criterion of the Cauchy pressure, defined as (C12  C44) malized the Cauchy pressure (C12  C44) by dividing the
[32], and the Pugh’s modulus ratio G/B [33], where G and Young’s modulus E, and re-plotted the relationship of
B are shear and bulk modulus, respectively. Previous stud- (C12  C44)/E vs. G/B based on the data in Fig. 7a. It is
ies also demonstrate the validity of studying ductile-to-brit- shown in Fig. 7b that Fe1xGax alloys at different Ga con-
tle-transition in intermetallic compounds based on the tent can be fitted by a quadratic curve while the typical
above two criteria [34–39]. For metallic bonding, the Cau- ductile and hard materials can be fitted by a hyperbola
chy pressure is typically positive [32] and the Pugh’s mod- curve. As indicated by the arrow in Fig. 7b, the closer to
ulus ratio is less than 0.57 [33]. However, for directional the upper left corner, the more ductile and stronger metallic
covalent bonding with angular character, the Cauchy pres- the bonding; inversely, the closer to the bottom right cor-
sure is negative and the Pugh’s modulus ratio is larger than ner, the more brittle and stronger the covalent bonding.
0.57. These correlations have been verified and validated One may note that the value (C12  C44)/E of Ag is greater
[40–42] for typical metallic bond-based inductile materials than that of Fe1xGax alloys (the highest Ga content stud-
(e.g. Au, Ag) as well as directional covalent bonding-based ies here, depicted by the vertical black dot line in Fig. 7b),
brittle materials (e.g. semiconductor Si, super-hard cubic which implies that the metallic bonding of Ag is stronger
BN, and diamond). From Fig. 7a, the Cauchy pressure than Fe1xGax alloys. It is also noted that the ductility
(C12  C44) for Fe1xGax at all Ga concentration remains of Fe1xGax alloys increases gradually with the Ga concen-
almost the same positive value, indicating a metallic bond- tration and almost reaches the same G/B value as Ag, indi-

Fig. 7. (a) Correlation between the Cauchy pressure (C12  C44) (vertical axes) and the Pugh’s modulus ratio G/B (horizontal axes) represented by blue
ball for Fe1xGax alloys, along with typical ductile and hard materials for comparison depicted by red balls. The magnified correlation for Fe1xGax
alloys at different Ga content was shown in the left bottom inset. (b) Renormalized hyperbolic correlation derived by dividing the Young modulus E from
(C12  C44) for all the data of (a). The horizontal dashed line of (C12  C44) denotes the critical zero Cauchy pressure defined by Pettifor [32], whereas the
vertical dashed line of G/B = 0.57 corresponds to the critical Pugh’s modulus ratio defined by Pugh [33]. (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article.)
 H. Wang et al. / Acta Materialia 61 (2013) 2919–2925 2925

cating an excellent ductility as Ag, which plays a significant [2] Clark AE, Wun-Fogle M, Restorff JB, Lograsso TA. IEEE Tran
role in the enhancement of magnetostriction in these Fe– Magn 2001;37:2678.
[3] Cullen JR, Clark AE, Wun-Fogle M, Restorff JB, Lograsso TA. J
Ga alloys. It is expected that the ductility of Fe1xGax Magn Magn Mater 2001;948:226.
alloys can be further improved at even higher Ga concen- [4] Clark AE. In: Wohlfarth EP, editor. Ferromagnetic materials, vol.
tration following the trend of quadratic relationship as 1. Amsterdam: North-Holland; 1980. p. 531.
depicted by the green arrow in Fig. 7b. [5] Andreev AV. In: Buschow KHJ, editor. Handbook of magnetic
materials, vol. 8. Amsterdam: Elsevier; 1995. p. 59.
[6] Kainuma R, Imano1 Y, Ito W, Sutou Y, Morito H, Okamoto S, et al.
4. Concluding remarks Nature (London) 2006;439:957.
[7] Chmielus M, Zhang XX, Witherspoon C, Dunand DC, Müllner P.
Based on the ab initio molecular dynamics (AIMD) Nature Mater 2009;8:863.
results, different size-scaled cell models were constructed [8] Atulasimha J, Flatau AB. Smart Mater Struct 2011;20:043001.
at different Ga concentration in order to systematically [9] Ikeda O, Kainuma R, Ohnuma H, Fukamichi K, Ishida K. J Alloys
Compd 2002;347:198.
evaluate the anisotropic mechanical properties at Ga con- [10] Zhang YN, Cao JX, Wu RQ. Appl Phys Lett 2010;96:062508.
centration, especially near the first magneostriction peak [11] Wang H, Zhang YN, Yang T, Zhang ZD, Sun LZ, Wu RQ. Appl
(x = 19%). The calculated anisotropic shear constants Ć Phys Lett 2010;97:262505.
and C44 agree well with experimental data from different [12] Wang H, Zhang YN, Wu RQ, Sun LZ, Xu DS, Zhang ZD.
groups, which prove the validity and high quality of our Manuscript in preparation.
[13] Xing Q, Du Y, McQueeney RJ, Lograsso TA. Acta Mater
calculations based on the supercell model. Meanwhile, 2008;56:4536.
the formation of D03-like structures has an important effect [14] Huang M, Mandru AO, Petculescu G, Clark AE, Wun-Fogle M,
on the softness of tetragonal shear modulus and negligible Lograsso TA. J Appl Phys 2010;107:09A920.
influence on rhombohedral shear modulus. The decreasing [15] Wuttig M, Dai L, Cullen J. Appl Phys Lett 2002;80:1135.
Young’s modulus and the large negative Poisson’s ratio in [16] Courtney TH. In: Mechanical behavior of materials. Boston,
MA: McGraw-Hill; 2000. p. 56–61.
the [1 1 0] direction can be attributed to the Fe dangling [17] Kohn W, Sham LJ. Phys Rev 1965;140(4A):A1133.
bond to its nearest Ga atoms. Moreover, the reason that [18] Car R, Parrinello M. Phys Rev Lett 1985;55(22):2471.
the Young’s modulus in the [1 1 0] direction is greater than [19] Marx D, Hutter J. In: Grotendorst J, editor. Modern methods and
in the [1 0 0] direction can be attributed to the different algorithms of quantum chemistry, NIC series, vol. 1. Jülich: John von
arrangement of the pure Fe layer and Fe–Ga mixed layer Neumann Institute for Computing; 2000. p. 301–449.
[20] Kresse G, Hafner J. Phys Rev B 1994;49:14251.
along the [1 1 0] and [1 0 0] axes. Furthermore, the ductility [21] Perdew JP, Burke K, Ernzerhof M. Phys Rev Lett 1996;77:3865.
of Fe1xGax alloys was enhanced with increasing Ga con- [22] Kresse G, Joubert D. Phys Rev B 1999;59:1758.
tent in terms of both classic Pettifor’s Cauchy pressure [23] Monkhorst HJ, Pack JD. Phys Rev B 1976;13:5188.
C12  C44 and classic Pugh’s modulus ratio G/B, which is [24] Nye JF. Physical properties of crystals. Oxford: Clarendon; 1985.
significant for the enhanced magnetostriction at high Ga [25] Kellogg RA, Russell AM, Lograsso TA, Flatau AB, Clark AE, Wun-
Fogle M. Acta Mater 2004;52:5043.
concentration. [26] Clark AE, Hathaway KB, Wun-Fogle M, Restorff JB, Lograsso TA,
Keppens VM, et al. J Appl Phys 2003;93:8621.
Acknowledgements [27] Restorff JB, Wun-Fogle M, 1, Hathaway KB, Clark AE, Lograsso
TA, et al. J Appl Phys 2012;111:023905.
This work was supported by CAS/SAFEA International [28] Xing Q, Lograsso TA. Appl Phys Lett 2008;93:182501.
[29] Petculescu G, Ledet KL, Huang M, Lograsso TA, Zhang YN, Wu
Partnership Program for Creative Research Teams. HW RQ, et al. J Appl Phys 2011;109:07A904.
and LZS acknowledge the support from NSF (Grant No: [30] Zhang YN, Wu RQ, Schurter HM, Flatau AB. J Appl Phys
CMMI-0800417) and RQW acknowledges the support 2010;108:023513.
from ONR (Grant No: N00014-08-1-0143). Furthermore, [31] Jain M, Verma MP. Indian J Pure Appl Phys 1990;28:178.
ZDZ thanks the support from the National Natural Sci- [32] Pettifor DG. Mater Sci Technol 1992;8:345.
[33] Pugh SF. Philos Mag 1954;45:823.
ence Foundation of China (No. 50831006) and from the [34] Cottrel AH. Proc. European Conf. advanced materials and processes.
Ministry of Science and Technology of China National Ba- Cambridge; 1991.
sic Research Program (No. 2012CB619404). Computer [35] Johnson RA. Phys Rev B 1998;37:3924.
simulations were conducted with the Shenyang Supercom- [36] Pettifor DG, Cottrell AH. Electronis theory in alloy design farn-
puter Facility of Institute of Metal Research, Chinese ham. UK: Ashgate Publishing; 1992.
[37] Chen K, Zhao L, Tse JS. J Appl Phys 2003;93:2414.
Academy of Sciences and the US Department of Energy [38] Chen K, Zhao L, Tse JS. Phys Lett 2004;331:400.
Supercomputer Facility. [39] Niu HY, Chen XQ, Liu PT, Xing WW, Cheng XY, Li DZ, et al. Sci
Rep 2012;2:718.
References [40] Frederikse HPR. CRC Handbook of Chemistry and Physics. Elastic
constants of single crystals. Boca Raton, FL: CRC Press; 2006.
[1] Clark AE, Restorff JB, Wun-Fogle M, Lograsso TA, Schlagel DL. [41] Connétable D, Thomas O. Phys Rev B 2009;79:094101.
IEEE Trans Magn 2000;36:3238. [42] Yao HZ, Ouyang LZ, Ching WY. J Am Ceram Soc 2007;90:3194.