Effect of alloying elements on the elastic properties of Mg from first-principles calculations

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Available online at www.sciencedirect.

com

Acta Materialia 57 (2009) 3876–3884


www.elsevier.com/locate/actamat

Effect of alloying elements on the elastic properties of Mg from


first-principles calculations
S. Ganeshan *, S.L. Shang, Y. Wang, Z.-K. Liu
Department of Materials Science and Engineering, The Pennsylvania State University, University Park, PA 16802, USA

Received 10 March 2009; received in revised form 20 April 2009; accepted 22 April 2009
Available online 21 May 2009

Abstract

The influence of different alloying elements on the lattice parameters and elastic properties of Mg solid solution has been studied using
first-principles calculations within the generalized gradient approximation. The solute atoms employed herein are Al, Ba, Ca, Cu, -
Ge, K, Li, Ni, Pb, Si, Y and Zn. A supercell consisting of 35 atoms of Mg and one solute atom is used in the current calculations. A good
agreement between calculated and available experimental data is obtained. Lattice parameters of Mg–X alloys are found to be dependent
on the atomic radii of the solute atoms. A correlation between the bulk modulus of Mg–X alloys and the nearest-neighbor distance
between Mg and X is shown. Addition of solute atoms belonging to the s-block and p-block of the periodic table results in a lower bulk
moduli than d-block elements. A strong dependence of the elastic modulus of Mg–X alloys on the elastic properties of the solute atoms is
also observed. Using the bulk modulus/shear modulus ratio (B/G), the change in the ductility of Mg due to the addition of the solute
atom is briefly described. Linear regression coefficients for the elastic constants of each of the alloys are obtained as a tool for predicting
the trend in the elastic properties of Mg as a function of concentration of the solute atoms.
Ó 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Magnesium alloys; Elastic behavior; Density functional theory

1. Introduction alloys based on their industrial importance (X = Al, Ba, -


Ca, Cu, Ge, K, Li, Ni, Pb, Si, Y, and Zn). Out of the 12 sys-
Magnesium alloys find applications in several fields, par- tems studied, experimental data were available for four of
ticularly in the automobile and aerospace industries owing them (X = Al, Li, Pb and Zn), and are compared with the
to their high strength-to-weight ratio, specific strength, stiff- predicted values in this study. The present work, together
ness and low density. Over the last decade, Mg alloys have with the previous work [2] on the elastic constants of com-
been experiencing a tremendous increase in use [1], inspiring pounds, forms a basis for predicting the elastic properties
us to develop a better understanding of the effects of alloying of Mg alloys.
elements on their properties in order to sustain this growth. The contents of this paper are organized as follows. In
Due to the low solubility of alloying elements in Mg, most Section 2, the methodology and computational details for
Mg alloys consist of a solid solution matrix with binary or calculating lattice constants and elastic stiffness of hexago-
ternary compounds. To understand the effect of alloying ele- nal Mg–X alloys are explained. In Section 3, evidence for
ments on mechanical properties, we recently predicted the the accuracy of our calculated results is presented by com-
elastic constants of 25 binary Mg compounds from first- paring these with available experimental data. Further, an
principles calculations [2]. The current work presents the in-depth analysis of the results is also provided. Finally in
elastic stiffnesses (Cijs), elastic moduli (bulk, shear and Section 4, a summary of our work is given.
Young’s) and lattice parameters (a and c) of 12 Mg–X dilute
2. Methodology

*
Corresponding author. Tel.: +1 814 232 0854. Herein, first-principles calculations based on density
E-mail address: sxg319@psu.edu (S. Ganeshan). functional theory [3], within the generalized gradient

1359-6454/$36.00 Ó 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2009.04.038
S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884 3877

approximation [4] as incorporated in the Vienna Ab Initio For the hexagonal structure studied in the present work,
Simulation Package [5,6], are employed. The ion–electron the number of independent components of elastic stiffness
interaction is described using the projector augmented wave tensor decreases to five (i.e. C11, C12, C13, C33 and C44).
method (PAW) [7]. A supercell consisting of 36 atoms (35Mg Eq. (2) can thus be written in a simplified form as:
and 1X atom, corresponding to 2.77 at.% X) was employed 0 1
C 11 C 12 C 13 0 0 0
for the current calculations. An energy cut-off of 1.25 times B C 12 C 22 C 13
the maximum potential energy of either Mg or X (whichever B 0 0 0 C C
B C
element has the higher default cut-off energy) as given in B C 13 C 13 C 33 0 0 0 C
B C
their pseudo-potential file is used. With the structure of the B 0 0 0 C 44 0 0 C
B C
alloy being hexagonal type, a C point centered k-mesh of B C
@ 0 0 0 0 C 44 0 A
6  6  5 is used [8]. The atomic arrangements are relaxed
using the Methfessel–Paxton technique [9] for the recipro- 0 0 0 0 0 C11 C2
12

0 11 0 1
cal-space integration. The lattice parameters listed in this e1;1    e1;6 r1;1    r1;6
paper correspond to those obtained after performing static B e2;1    e2;6 C B r2;1    r2;6 C
B C B C
calculations for the equilibrium volumes of each of the B C B C
B e3;1    e3;6 C B r3;1    r3;6 C
Mg–X alloys. Their corresponding first nearest-neighbor ¼B
Be
C B
C B
C: ð3Þ
distances have been analyzed using the CONVASP code B 4;1    e4;6 C B r4;1    r4;6 C
C
B C B C
[10]. Elastic constants at the equilibrium volumes are calcu- @ e5;1    e5;6 A @ r5;1    r5;6 A
lated using the stress–strain method [11]. The methodology e6;1    e6;6 r6;1    r6;6
used involves applying a set of strains (e = e1, e2, e3, e4, e5
and e6) where e1, e2, e3 refer to normal strains and e4, e5, e6 The bulk (B), shear (G), Young’s (E) modulus and Pois-
refer to shear strains, respectively. The crystal lattice vectors son’s ratio (ˆ) are computed using Voigt’s [12] method as
before (Q) and after ðQÞ deformation are related as follows: follows:
0 1
1 þ e1 e6 =2 e5 =2 B ¼ ðC 11 þ 2C 12 Þ=3; ð4Þ
B C
Q ¼ Q@ e6 =2 1 þ e2 e4 =2 A: ð1Þ G ¼ ðC 11  C 12 þ 3C 44 Þ=5; ð5Þ
e5 =2 e4 =2 1 þ e3 E ¼ ð9BGÞ=ðG þ 3BÞ; ð6Þ
In the current study, the following linearly independent m ¼ ð3B  2GÞ=2ð3B þ GÞ; ð7Þ
set of strains is applied:
2 3 where
x 0 0 0 0 0
60 x 0 0 0 07 C 11 ¼ ðC 11 þ C 22 þ C 33 Þ=3; ð8Þ
6 7
6 7 C 12 ¼ ðC 12 þ C 13 þ C 23 Þ=3; ð9Þ
60 0 x 0 0 07
6 7
60 0 0 x 0 07 C 44 ¼ ðC 44 þ C 55 þ C 66 Þ=3: ð10Þ
6 7
6 7
40 0 0 0 x 05
0 0 0 0 0 x 3. Results and discussion
with x = ±0.01. The corresponding stresses (r = r1, r2, r3,
Calculated lattice parameters of pure Mg and the 12 bin-
r4, r5 and r6) for the deformed crystals due to applied
ary Mg–X alloys along with the available experimental
strains are obtained from first principles. From the n set
data are shown in Table 1. Out of the 12 alloys studied
of strains (e) and the resulting stresses (r), elastic stiffnesses
here, experimental data were available for Mg–Al [13–
(Cijs) are then calculated based on Hooke’s law as follows:
15], Mg–Li [14,16], Mg–Pb [14,15,17] and Mg–Zn [14,18],
0 1
C 11 C 12 C 13 C 14 C 15 C 16 measured by X-ray diffraction. The lattice spacing quoted
B C 21 C 22 C 23 C 24 C 25 C 26 C in Raynor’s [15] work corresponds to a common tempera-
B C
B C 31 C 32 C 33 C 34 C 35 C 36 C ture of 20 °C, while others [13,14,16–19] correspond to that
B C
B C 41 C 42 C 43 C 44 C 45 C 46 C at room temperature. A good agreement between calcu-
B C
@C C 52 C 53 C 54 C 55 C 56 A lated and experimental data is obtained; with a difference
51
C 61 C 62 C 63 C 64 C 65 C 66 between the two of less than 1%. The percentage differences
0 11 0 1 between calculated and experimental data of lattice param-
e1;1    e1;n r1;1    r1;n eters and c/a ratio are also given in Table 1.
Be C B C
B 2;1    e2;n C B r2;1    r2;n C As shown in Table 1 and Figs. 1 and 2, the lattice
B C B C parameters a and c of Mg tend to decrease on the addition
B e3;1    e3;n C B r3;1    r3;n C
¼BB C B C: ð2Þ of 2.77 at.% Al, while the c/a ratio is increased. In the case
C B C
B e4;1    e4;n C B r4;1    r4;n C of Mg–2.77 at.% Pb, both the lattice parameters as well as
B C B C
@ e5;1    e5;n A @ r5;1    r5;n A c/a ratio are larger than those of pure Mg. Kashyap et al.
e6;1    e6;n r6;1  r6;n [20] also observed a similar influence of Al and Pb on the
3878 S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884

Table 1
Calculated and experimental lattice parameters of Mg–X alloys.
X Calc. a Exp. a % diff. Calc. c Exp. c % Diff. Calc. Exp. % Diff. % c/aa db Atomic Block
(Å) (Å) (Å) (Å) c/a (Å) volume (Å3)
Al 3.183 3.199 [13] 0.505 5.177 5.196 [13] 0.373 1.627 1.625 [13] 0.131 0.347 3.130 45.413 p
3.198 [14] 0.479 – 5.196 [14] – – 1.625 [14] 0.119 – – –
3.192 [15] 0.287 – 5.184 [15] – – 1.624 [15] 0.150 – – –
Ba 3.252 – – 5.231 – – 1.609 – – 0.766 3.367 47.905 s
Ca 3.221 – – 5.211 – – 1.618 – – 0.201 3.276 46.822 s
Cu 3.181 – – 5.123 – – 1.610 – – 0.657 3.060 44.909 d
Ge 3.183 – – 5.172 – – 1.625 – – 0.251 3.134 45.381 p
K 3.235 – – 5.230 – – 1.616 – – 0.290 3.332 47.410 s
Li 3.194 3.208 [14] 0.438 5.150 5.203 [14] 1.028 1.613 1.622 [14] 0.587 0.525 3.139 45.500 s
3.207 [16] 0.410 5.202 [16] 1.004 1.622 [16] 0.587 – – 0.000
Ni 3.178 – – 5.096 1.603 1.094 3.008 44.568 d
Pb 3.202 3.205 [15] 0.110 5.208 5.214 [15] 0.104 1.627 1.627 [15] 0.006 0.353 3.236 46.232 p
3.212 [14] 0.328 5.224 [14] 0.301 – 1.626 [14] 0.027 – – 0.000
3.212 [17] 0.313 5.224 [17] 0.296 – 1.626 [17] 0.017 – – 0.000
Si 3.177 – – 5.168 – – 1.627 – – 0.355 3.111 45.167 p
Y 3.214 – – 5.205 – – 1.620 – – 0.081 3.249 46.559 d
Zn 3.184 3.196 [14] 0.377 5.148 5.188 [14] 0.777 1.617 1.623 [14] 0.399 0.259 3.111 45.209 d
3.195 [18] 0.336 – 5.186 [18] -0.731 – 1.623 [18] 0.399 – – –
Mg 3.195 3.210 [13] 0.451 5.178 5.211 [13] 0.628 1.621 1.624 [13] 0.175 0.000 3.179 –
3.209 [13] 0.444 – 5.211 [13] 0.626 – 1.624 [13] 0.175 – – – s
3.210 [14] 0.463 – 5.211 [14] 0.630 – 1.623 [14] 0.169 – – –
3.203 [15] 0.234 – 5.200 [15] 0.416 – 1.624 [15] 0.181 – – –
3.209 [18] 0.448 – 5.211 [18] 0.624 – 1.624 [18] 0.175 – – –
a
Percentage difference in the c/a ratio between Mg–X and Mg.
b
Nearest-neighbor distance between Mg and X.

3.26 5.25
Lattice constant 'a' of Mg-X alloys (Å)

Lattice constant 'c' of Mg-X alloys (Å)

K K
3.24 Ba Ba
Pb Y
5.2 Ca
Ca
3.22 Y Pure Mg
Al
Si Ge Zn
5.15
3.2 Li Pb Li
PureMg
Cu
Ge Zn
Al
3.18 Cu 5.1
Si Ni
Calc. Ni
Calc.
3.16 Exp. Exp.
5.05
1.2 1.4 1.6 1.8 2 2.2 2.4 1.2 1.4 1.6 1.8 2 2.2 2.4
Calculated atomic radius of pure elements X (Å) Calculated atomic radius of pure elements X (Å)

Fig. 1. Influence of atomic radius of solute atom (X) on the lattice Fig. 2. Influence of atomic radius of the solute atom (X) on the lattice
constant a of Mg–2.77 at.% X solution. constant c of Mg–2.77 at.% X solution.

lattice parameters of Mg. Li and Zn are observed to ing axial ratio [21]. Although Becerra and Pekguleryuz. [16]
decrease the lattice parameters as well as the c/a ratio of also observed a decrease in the lattice parameters of Mg–
Mg (see Figs. 1 and 2 and Table 1). The decrease in the Zn as in the present study, no significant change in the
c/a ratio of Mg–Zn as compared to that in Mg–Li is less c/a ratio of Mg–Zn has been reported. As for some of
due to the larger decrease in the lattice parameter c when the other elements in the current work, Ba, Ca, K and Y
Li is added to Mg. This is in agreement with the data increase the lattice parameters a and c of Mg, but decrease
reported by Becerra and Pekguleryuz [16]. This decrease the c/a ratio. Busk [22] in his study on the effect of solute
in the c/a ratio of Mg due to the addition of Li increases additions on the lattice parameters of Mg stated that Ba
the room temperature formability of Mg. The correlation was found to increase the lattice parameters of Mg due
between formability and axial ratio was attributed to the to its large atomic radius. Busk [22] also reported an
change in the critical resolved shear stress on the basal increase in the lattice parameters of Mg due to the addition
plane of a hexagonal close-packed (hcp) lattice, with vary- of Ni even though the atomic radius of Ni is smaller than
S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884 3879

that of Mg. Busk was unable to explain the reason for this ments to 16% Li, considering it to be the maximum
increase in the lattice parameters of Mg–Ni alloys. In the amount of Li that dissolves in hcp Mg. Fig. 4 shows a plot
present work, Ni decreases both the lattice parameters as of calculated vs. experimental elastic stiffness for Mg–Li at
well as the c/a ratio of Mg, as expected. The results in 2.77 at.% Li. In Fig. 5, our calculated data at 2.77 at.% Li
Busk’s [22] paper were only reported as a set of summa- is plotted along with the measured data at various composi-
rized data calculated from an empirical equation. Since tions. From both Figs. 4 and 5, it can be seen that the differ-
the exact composition of the alloys is not known, his data ence between the two sets of data is not very significant,
has not been shown in Table 1. Calculations pertaining to except for an error (13%) in the C33. For alloy systems con-
Mg–Ni in the present study have been performed both with taining Al, Zn and Pb, though measurements for their indi-
and without magnetic spin on Ni. The net magnetic vidual elastic stiffness were unavailable, their Young’s
moment in both cases remained zero. modulus measured using the high-frequency free–free trans-
Figs. 1 and 2 depict the change in the lattice parameter verse vibration technique were obtained from Ref. [24].
of Mg, due to the addition of solute elements. These figures Measurements were available for Mg–Al and Mg–Pb alloys
are plotted against the atomic radius of the solute atoms, at several compositions, which have been extrapolated to
which have been calculated for their stable structures at Mg–2.77 at.% Al and Mg–2.77 at.% Pb for comparison
room temperature. The atomic radii of pure elements cor- herein. In the case of Mg–Zn, measurements of its Young’s
respond to half of the nearest-neighbor distance between modulus were available at different c/a ratios. In order to
the atoms and are consistent with those calculated by maintain an appropriate base for comparison, we have
Wang et al. [23]. The change in the lattice parameters of extrapolated the measured values to the equilibrium c/a ratio
the Mg–X alloys are proportional to the radius of the sol- of Mg–2.77 at.% Zn calculated in this work. Data thus
ute atom. Similar observations for Mg alloys were also obtained at 2.77 at.% of the alloying element is shown in
made in Ref. [22]. The trend seen in the first nearest-neigh- Table 2. The average difference between calculated and
bor distance between Mg and X (see Table 1) (which in experimental data lies within 8%, thus validating the reason-
increasing order is Ni, Cu, Si, Zn, Al, Ge, Li, Pb, Y, Ca, K able level of accuracy in the calculations carried out in this
and Ba) also further supports the correlation between study.
atomic radius of the solute atom and the lattice contrac- The maximum change in the elastic modulus of Mg is
tion/expansion of Mg. This increasing trend in the first observed for K, whereas the minimum change is observed
nearest-neighbor distance with respect to the atomic radius when Al is added. The largest values of bulk, shear and
of the solute atom is shown in Fig. 3. On comparing Figs. Young’s moduli pertained to Mg–Ni, and the lowest values
1–3, we notice that the trends are similar in all the three corresponded to Mg–K. One of the reasons for these obser-
cases. vations is the size of the alloying element, which tends to
Elastic stiffness coefficients (Cijs) of pure Mg and dilute affect the volume of Mg–X, and thus the elastic properties.
Mg–X alloys that have been calculated in this study are A more clear relation between atomic volume and elastic
shown in Table 2. Experimentally measured elastic stiffness modulus is given below. All the alloys systems under study
coefficients of Mg–X alloys were available, to the best of satisfy the Born criteria for mechanical stability, i.e.
the author’s knowledge, only for Mg–Li at compositions C11 > 0, (C11  C12) > 0, C44 > 0 and (C11 + C12)C33
up to 15.94 at.% Li and 298 K using the pulse-echo tech- 2C212 > 0 for hexagonal structure. Meaning that on the addi-
nique [19]. The authors of Ref. [19] restricted the measure- tion of 2.77 at.% X element to Mg, the Mg alloy is within the
limit of mechanical stability.
It is interesting to note the influence of the alloying ele-
3.4 ment from different groups on the elastic modulus of Mg.
The elements chosen in the current work (Al, Ca, Ba, Cu,
3.35 K Li, Ni, K, Pb, Si, Ge, Y, Zn), belong either to the s-block,
Nearest-neighbor distance

3.3 Ba p-block or d-block of the periodic table, i.e. the valence


between Mg-X (Å)

3.25 Ca shell electrons pertain to the s-orbital, p-orbital or d-orbital


Y (see Table 1).
Pb
3.2
Pure Mg From the results in Table 2 it is apparent that the addi-
Ge Li tion of s-block elements to Mg results in a lower bulk than
3.15 Al
Si the addition of p- and d-block elements. Between p- and d-
3.1
Zn block elements, the former have a lower bulk modulus
3.05 Cu value except in the case of element Y. The reason for Y fall-
Calc.
3 Ni ing under the category of the p-block elements could be due
1.2 1.4 1.6 1.8 2 2.2 2.4 to the number of d electrons being only 1, whereas for
Calculated atomic radius of pure elements X (Å) other transition elements, such as Ni, Cu and Zn, the num-
Fig. 3. Correlation between the atomic radius of the solute atom (X) and ber of d electrons is significant. Furthermore, Ni, Cu and
the nearest-neighbor distance between Mg and X in Mg–2.77 at.% X Zn are 3d metals while Y is a 4d metal. In a study by Shein
solution. and Ivanovskii [25], it was observed that the bulk moduli of
3880 S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884

Table 2
Calculated and experimental elastic properties (GPa) of Mg–X alloys.
X C11 C12 C13 C33 C44 B G E k ˆ B/G
Al 65.6 25.9 19.3 69.0 13.6 36.6 18.5 47.4 40.47 0.28 1.98
Al [24] – – – – – – – 45.0 – – –
Ba 54.8 24.5 20.8 59.3 16.0 33.4 16.8 43.2 39.78 0.28 1.98
Ca 56.4 27.6 22.3 60.4 14.6 35.2 15.4 40.4 40.82 0.31 2.28
Cu 61.6 24.5 25.7 62.7 15.9 37.5 17.4 45.2 40.59 0.30 2.16
Ge 64.5 26.3 20.2 67.6 14.2 36.6 18.2 46.7 40.58 0.29 2.02
K 54.7 23.1 21.7 56.4 13.8 33.2 15.3 39.7 39.11 0.30 2.17
Li 58.9 24.5 23.2 54.0 15.0 34.84 16.17 42.00 38.62 0.30 2.15
Li [19] 59.0 25.9 21.7 61.0 16.2 35.3 17.1 44.1 – 0.29 2.06
Li [24] – – – – – – – 45.8 – –
Ni 63.5 25.9 23.8 69.6 19.2 38.2 19.6 50.3 43.70 0.28 1.94
Pb 63.1 24.9 19.7 67.3 13.6 35.8 17.9 46.0 40.91 0.29 2.00
Pb [24] – – – – – – – 43.1 – – –
Si 65.0 27.3 19.7 70.1 14.5 37.0 18.5 47.5 40.73 0.29 2.01
Y 59.5 27.3 21.6 64.5 19.0 36.1 18.3 47.1 41.43 0.28 1.97
Zn 62.3 25.5 23.1 66.2 14.1 37.1 17.3 44.8 40.80 0.30 2.15
Zn 48.0 [24]
Mg 63.5 24.85 20.0 66 19.3 35.83 18.5 47.4 40.27 0.28 1.93
Mg [19] 59.5 25.9 21.8 61.6 16.4 35.6 17.3 44.6 – – 2.05
Mg [35] 63.5 25.9 21.7 66.5 18.4 36.9 19.4 49.5 – – 1.9
Mg [24] 45.2 – – –
B, bulk modulus; G, shear modulus; E, Young’s modulus; k, spring constant; ˆ, Poisson’s ratio.

60
alloys decreased from 3d to 4d metals. In the case of shear
C11
moduli the results were a little different. When elements
Calculated Cij of Mg-Li (GPa)

55
C33 from the s-block were added to Mg, the shear modulus still
50
remained lower than the cases when p- and d-block ele-
45 ments were added. But between the p- and d-block ele-
40 ments, the results were mixed. Mg–Zn and Mg–Cu
35 possessed lower shear modulus than all the p-block ele-
30 ments, but Mg–Ni is observed to have the highest shear
25 C13 C12 modulus. Mg–Y again showed similar results to that of
the p-block elements. Thus, although there is an apparent
20
C44 C66 dependence of the bulk modulus on the valence electrons
15 of the element added to Mg, such dependence is absent
15 20 25 30 35 40 45 50 55 60
Experimental Cij of Mg-Li (GPa) in the case of shear modulus.
Another significant factor that corresponds directly to the
Fig. 4. Comparison of calculated and measured Cij of Mg–2.77 at.% Li.
bulk modulus of these alloys is the nearest-neighbor distance
between Mg and the alloying element X. The inverse depen-
dence of bulk modulus on the nearest-neighbor distance of
different Mg–X alloys that is observed here is analogous to
70
the results reported by Cohen [26]. We notice, as shown in
60 Fig. 6, that with the increase in the distance between Mg
Elastic stiffness Cij (GPa)

and X, the bulk modulus decreases owing to the lattice


50
expansion. Alloying elements such as Ni and Cu that have
40 very short nearest-neighbor distances from Mg, have greater
bulk modulus than elements such as K, Ca and Ba, whose
30
nearest-neighbor distance from Mg is large. This relation
20 can be additionally extended to saying that elements that
cause a decrease (increase) in the lattice constants results in
10 an increase (decrease) in the bulk modulus. With the bulk
C11 C12 C13 C33 C44 Exp.

0
C11 C12 C13 C33 C44 Calc. modulus (B) and nearest-neighbor distance (d) combined
0 2 4 6 8 10 12 14 16 together, the magnitude of bond strength between the
at% of Li in Mg atoms, in terms of its spring constant (k), may also be esti-
p
Fig. 5. Cij of Mg–Li, with calculated data at 0% and 2.77 at.% Li, and mated. Using the equation k = 2Bd/4 given in Ref. [27],
experimental data [19] with increasing composition of Li. the spring constant for various Mg–X alloys is calculated
S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884 3881

38 Ni
the change of volume. In this connection, we see that ˆ for the
Bulk modulus of Mg-X alloys (GPa)
Exp.
Calc.
systems in our work ranges between 0.28 and 0.31, thus indi-
Cu Si cating a considerable change in the volume due to elastic
37 Zn
Al Ge deformation. The only prominent exception in the trends
Y of Figs. 6 and 7 is Mg–Li. Noting that our calculated lattice
36 Pure Mg
Pb parameters, as well as elastic stiffness, are nowhere near the
Ca experimental values, it is possible that the low value of the
35 Li lattice parameter c of Mg–Li, which plays a key role in the
nearest-neighbor distance and atomic volume, is the reason
34 for this discrepancy.
Ba In the earlier part of this section, the influence of the
K
3 3.05 3.1 3.15 3.2 3.25 3.3 3.35 3.4 atomic radius of the pure element X on the overall lattice
First nearest neighbor distance in Mg-X (Å) constants of Mg–X alloys was projected. A similar corre-
spondence between the bulk modulus of pure elements and
Fig. 6. Correlation between bulk modulus and nearest-neighbor distance
observed in Mg–2.77 at.% X solution. that of the alloys is seen in Fig. 8. Both the ordinate as well
as abscissa coordinates are calculated from this work. Ele-
ments having higher bulk modulus, such as Ni, Cu, Si, Ge,
and shown in Table 2. The highest value of ‘‘k” corresponds are those which also result in higher bulk modulus when
to that of Mg–Ni, which possesses the highest bulk modulus added to Mg. On the other hand, pure elements with lower
and shortest nearest-neighbor distance with Mg. Mg–Ba, bulk modulus, such as Ba, Ca, Li, K, result in a lower bulk
Mg–Li and Mg–K come under the category of having a com- modulus when added to Mg. Fig. 9 shows the shear modulus
paratively low spring constant. The connection between of Mg–X alloys plotted against the shear modulus of the sol-
bulk modulus, lattice constants and nearest-neighbor dis- ute atoms. Although the trend between the Mg–X alloys
tance projected here is in accordance with the definition of containing the above-mentioned elements is the same, some
bulk modulus, and is directly related to the external force exceptions are noted. Mg–Al and Mg–Y have a higher shear
required to compress or extend interatomic distances in modulus than Mg–Cu and Mg–Zn despite the pure elements
opposition to the internal forces that seek to establish an (Al, Y) having a low shear modulus compared to Cu and Zn.
equilibrium, or undistorted, interatomic distance [28]. The direct influence of the pure element’s shear modulus on
Fig. 7 validates the interrelation between bulk modulus that of Mg–X alloy is less obvious than in the case of bulk
and lattice parameters, which shows a strong dependence modulus.
of the bulk modulus on the atomic volume of the alloys, So far, the influence of the alloying elements on the elas-
due to the addition of the alloying element. The atomic vol- tic stiffness and elastic moduli (B, G and E) of Mg were dis-
ume of Mg–X alloys, plotted on the abscissa of the figure, is cussed. Elastic constants can also act as a tool in
obtained from the lattice constants calculated in this study. understanding a given material’s mechanical and physical
To better highlight the change in the volume of the Mg–X properties, such as machinability [30], ductility [31] and
alloys, due to uniaxial deformation, we make use of Pois- bond characteristics [28]. Mg is known for its excellent
son’s ratio (ˆ) (see Eq. (7)). Table 2 provides the value of ˆ machinability, while being limited in terms of its ductility.
for the alloy systems studied. Based on the results of Ravin- Hence, it is essential to know what alloying elements can
dran et al. [29], a ˆ of 0.5 indicates that the change in the vol- improve the ductile properties of Mg. The results obtained
ume associated with elastic deformation is zero. Ravindran from the current calculations can help one estimate the
et al. [29] also report that the lower the value of ˆ, the greater influence of the alloying elements on the ductility of Mg.

Ni
Bulk modulus of Mg-X alloys (GPa)

38 Calc. 38
Bulk modulus of Mg-X alloys (GPa)

Ni
Cu Exp.
Zn Si Cu
37 Si 37
Zn
Ge Al
Y Al Ge
36
Pb 36 Y
Pb
Ca Ca
35 Li 35
Li
34 34
Ba Calc.
K K Ba Exp.
44 44.5 45 45.5 46 46.5 47 47.5 48 20 40 60 80 100 120 140 160 180 200
3
Calculated Atomic volume of Mg-X alloys (Å ) Calculated bulk modulus of pure elements (GPa)

Fig. 7. Correlation between bulk modulus and the atomic volume of Mg– Fig. 8. Influence of bulk modulus of solute atom (X) on the bulk modulus
X alloys. of Mg–2.77 at.% X solution.
3882 S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884

20 At this juncture, it becomes important to understand the


Shear modulus of Mg-X alloys (GPa)

Ni change in the elastic properties of Mg with the change in


19 the composition of the solute atom. Based on the results
Al from this study, we have calculated the linear regression
Si
coefficient (d) for elastic constants of the Mg–X alloys con-
18 Pb Y Ge
sidered herein, using the following equation:
Zn Cu XMg  XMg–X
17
Ba d¼ ; ð11Þ
Dx
Li
16 where XMg represents the elastic constants of pure Mg,
Ca Calc. XMg–X are the elastic constants of Mg–X alloys, and Dx
K Exp.
15 is the composition of the solute atom, i.e. 2.77 at.%.
10 20 30 40 50 60 70 80 90 100
Calculated shear modulus of pure elements (GPa)
Coefficients (d) for each of the alloys are shown in Table
3. Fig. 5 shows the calculated elastic stiffness using the lin-
Fig. 9. Dependence of the shear modulus of Mg–X alloys on the shear ear regression coefficients for Cijs of Mg–Li along with the
modulus of solute atoms (X).
measured values at various compositions. As mentioned
earlier, the data are close to one another, while a prominent
The ductility of a metal as given by Pugh [31] can be esti- difference exists in the case of C33. The reason for this dif-
mated from the bulk modulus/shear modulus ratio, B/G. A ference is not known, while the calculated C33 value for
critical value that separates brittleness and ductility is given pure Mg is close to that measured from experiments (see
at B/G = 1.75. When materials have a B/G ratio above Table 2). The decreasing trend in the Young’s modulus
1.75, they are considered ductile, and below B/G = 1.75 of Mg–Li with the increasing composition of Li as evident
they are considered brittle. The B/G ratio of pure Mg is from the negative coefficient of linear regression in our cal-
about 1.92. Fig. 10 shows the B/G ratio of Mg–X alloys culations is consistent with the negative slope obtained
plotted against the B/G ratio of pure X. All the alloy sys- from the work of Wazzan and Robinson [19] (see Table
tems invariably have a B/G ratio not only greater than 3). In another set of experimental measurements [24], it is
the critical value of 1.75 but also greater than that of pure seen that trend in the Young’s modulus of Mg–Li follows
Mg. Based on Ref. [31], all the alloy systems can be classi- a positive slope, in contrast to the results reported in Ref.
fied as primarily exhibiting ductile properties at 2.77 at.% [19]. This discrepancy between the two experimental data
X. Mg–Ca, Mg–K and Mg–Li come under the category sets leaves less room for a precise explanation for the differ-
of high B/G ratio, while Mg–Ni and Mg–Y alloys have a ences in the C33 value, which further effects the overall
relatively low B/G ratio when compared to the other Young’s modulus value of Mg–Li. In the case of Mg–Al
Mg–X alloys. The positive influence of Li on the ductility and Mg–Pb alloys, the trends shown from our calculated
of Mg was earlier reported by the authors of Refs. Young’s modulus and those from experiments complement
[32,33]. Hassan and Gupta [34], in their study on the each other. Fig. 11 shows the calculated Young’s modulus
mechanical properties of Mg–Ni composites, observed that of these two alloy systems along with the measured data at
while the elastic modulus of Mg–Ni increased, the ductility various compositions. The line joining the data points
was adversely affected. Also, in the case of Mg–Y, the low highlights the trend in which the Young’s modulus is
B/G value of Mg–Y alloy observed in this study strongly expected to change with the increasing composition of
correlate to the low B/G ratio of Mg–Y compounds as cal- the solute atom. From Fig. 11, it is seen that the Young’s
culated in Ref. [2]. modulus of Mg decreases significantly with increase in
the concentration of Pb, while this decrease is less in the
2.4 Ductile region case of Al. In both cases the coefficient of linear regression
Ca for the calculated Young’s modulus lies close to the value
2.2 Zn K obtained from experiments. Hence, the coefficients (d)
Cu
B/G ratio of Mg-X alloys

Li
Si obtained in this work can act as a tool for approximately
2 Ba Al Pb
Ge
Y Ni predicting the trends in the elastic properties of Mg with
1.8 B/G = 1.75 respect to the composition of solute atoms.
1.6

1.4
4. Summary
Brittle region
1.2 Lattice parameters and elastic constants of 12 Mg–X
Calc.
Exp. (X = Al, Ba, Ca, Cu, Ge, K, Li, Ni, Pb, Si, Y, and Zn) solu-
1
1 1.5 2 2.5 3 3.5 4 tions have been calculated from first-principles calculations.
Calculated B/G ratio of solute atoms (X) Reasonable agreement between calculated and measured
Fig. 10. Influence of the solute atoms on the ductility of Mg, based on data is observed, thus validating the first-principles method-
B/G ratio. ology employed. The effect of the solute atom on these prop-
S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884 3883

Table 3
Calculated coefficient of linear regression, d (GPa/at.%) for the elastic constants of Mg–X alloys along with available experimental data.
X d-C11 d-C12 d-C13 d-C33 d-C44 d-B d-G d-E
Al 0.74 0.36 0.24 1.06 0.60 0.27 0.03 0.01
Al [24] – – – – – – – 0.07
Ba 3.16 0.14 0.27 2.42 0.25 0.87 0.61 1.51
Ca 2.58 0.98 0.81 2.04 0.27 0.21 1.12 2.54
Cu 0.69 0.14 2.06 1.19 0.22 0.61 0.41 0.80
Ge 0.34 0.51 0.05 0.56 0.41 0.28 0.14 0.25
K 3.18 0.63 0.61 3.47 0.56 0.95 1.18 2.77
Li 1.65 0.13 1.16 4.35 0.11 0.36 0.85 1.96
Li [19] 0.25 0.12 0.10 0.25 0.07 0.15 0.07 0.18
Li [24] – – – – – – – 0.22
Ni 0.02 0.39 1.37 1.30 1.41 0.84 0.40 1.03
Pb 0.15 0.00 0.12 0.45 0.61 0.03 0.24 0.53
Pb [24] – – – – – – – 0.66
Si 0.52 0.88 0.12 1.48 0.29 0.43 0.03 0.02
Y 1.44 0.88 0.59 0.54 1.33 0.09 0.07 0.13
Zn 0.43 0.23 1.10 0.05 0.43 0.46 0.46 0.93

48 supported by the Materials Simulation Center and the


47
Research Computing and Cyber Infrastructure Unit at
Pennsylvania State University. We would like to thank
Young's modulus (GPa)

46 Dr. Manjeera Mantina from the Phases Research Lab at


45 the Pennsylvania State University for her valuable
suggestions.
44

43 References
42
[1] Yang Z, Li JP, Zhang JX, Lorimer GW, Robson J. Acta Metall Sin
41 Mg-Pb Exp. Calc. from coefficient of linear regression (this work) 2008;21:313.
Mg-Al Exp. Calc. from coefficient of linear regression (this work)
[2] Ganeshan S, Shang SL, Zhang H, Wang Y, Mantina M, Liu ZK.
40
0 1 2 3 4 5 6 7 Intermetallics 2009;17:313.
at% X (X=Al, Pb) in Mg [3] Kohn W, Sham LJ. Phys Rev 1965;140:1133.
[4] Perdew JP, Burke K, Ernzerhof M. Phys Rev Lett 1996;77:3865.
Fig. 11. Predicted trends in the Young’s modulus of Mg–Al and Mg–Pb [5] Kresse G, Furthmuller J. Phys Rev B 1996;54:11169.
alloys based on the calculated coefficient of linear regression, along with [6] Kresse G, Furthmuller J. Comput Mater Sci 1996;6:15.
experimental data [24]. [7] Kresse G, Joubert D. Phys Rev B 1999;59:1758.
[8] Van de Walle A, Asta M, Ceder G. Calphad 2002;26:539.
[9] Methfessel M, Paxton AT. Phys Rev B 1989;40:3616.
erties of Mg has been studied. It is believed that alloying ele- [10] Morgan D. CONVASP—a computer program for analyzing VASP
ments with large size cause an increase in the lattice param- input and output files. MIT; 2003.
eter of Mg when doped. A dependence of the bulk modulus [11] Shang SL, Wang Y, Liu ZK. Appl Phys Lett 2007;90.
on the valence electrons of the solute atom is observed; while [12] Simmons G, Wang H. Single crystal elastic constants and calculated
aggregate properties: a handbook. Cambridge (MA): MIT Press;
such a trend is absent with respect to the shear modulus. A 1971.
correlation between the nearest-neighbor distance of Mg [13] Batchelder FWV, Raeuchle RF. Phys Rev 1957;105:59.
and X, the volume of Mg–X alloys and their bulk modulus [14] Hardie D, Parkins RN. Philos Mag 1959;4:815.
is reported. The elastic properties of solute atoms are also [15] Raynor GV. Proc Roy Soc London, Ser A 1942;180:0107.
found to play a key role in the elastic behavior of Mg–X [16] Becerra A, Pekguleryuz M. J Mater Res 2008;23:3379.
[17] Walker CB, Marezio M. Acta Metall 1959;7:769.
alloys. The influence of the alloying addition on the ductility [18] Barret C, Massalski TB. Structure of metals. Oxford: Pergamon
of Mg has been described. The coefficient of linear regression Press; 1987.
for each of the elastic constants has been obtained, and pro- [19] Wazzan AR, Robinson LB. Phys Rev 1967;155:586.
vides a means to estimate the trend in the elastic properties of [20] Kashyap KT, Ramachandra C, Sujatha M, Chatterji B. Bull Mater
Mg with respect to the concentration of the solute atom. Sci 2000;23:39.
[21] Raynor GV. The physical metallurgy of magnesium and its
alloys. New York: Pergamon Press; 1959.
Acknowledgments [22] Busk RS. JOM 1950;2:1460.
[23] Wang Y, Curtarolo S, Jiang C, Arroyave R, Wang T, Ceder G, et al.
Calphad 2004;28:79.
This work is funded by the National Science Founda-
[24] Hardie D. Acta Metall 1971;19:719.
tion (NSF) through Grant DMR-0510180. First-principles [25] Shein IR, Ivanovskii AL. J Phys – Condes Matter 2008;20.
calculations were carried out on the LION clusters [26] Cohen ML. Phys Rev B 1985;32:7988.
3884 S. Ganeshan et al. / Acta Materialia 57 (2009) 3876–3884

[27] Ledbetter H, Migliori A. J Appl Phys 2006;100. [32] Yan YD, Zhang ML, Han W, Cao DX, Yuan Y, Xue Y, et al.
[28] Courtney TH. Mechanical behaviour of materials. New York: McGraw- Electrochim Acta 2008;53:3323.
Hill; 1990. [33] Fedor M, Elkin VGD. Russian ultralight constructional Mg–Li alloys.
[29] Ravindran P, Fast L, Korzhavyi PA, Johansson B, Wills J, Eriksson Their structure, properties, manufacturing, applications. In: Kainer K,
O. J Appl Phys 1998;84:4891. editor. Magnesium. Materials Park (OH): TMS; 2005. p. 94.
[30] Sun ZM, Music D, Ahuja R, Schneider JM. Phys Rev B 2005;71. [34] Hassan SF, Gupta M. J Alloys Compd 2002;335:L10.
[31] Pugh SF. Philos Mag 1954;45:823. [35] Slutsky LJ, Garland CW. Phys Rev 1957;107:972.

You might also like