Journal of Non-Crystalline Solids 321 (2003) 120–125

www.elsevier.com/locate/jnoncrysol

Relationship between the widths of supercooled liquid

regions and bond parameters of Mg-based bulk metallic glasses
Shoushi Fang *, Xueshan Xiao, Lei Xia, Weihuo Li, Yuanda Dong
Institute of Materials, Shanghai University, Box 269, Yanchang Road 149, Shanghai 200072, China
Received 18 January 2002; received in revised form 28 January 2003

Abstract

The relationship between the widths of supercooled liquid regions (DTx ) of Mg-based bulk metallic glasses (BMG)
and bond parameters (i.e., electronegativity difference and atomic size parameters) was investigated. From the available
experimental data in the literatures, it seems that for Mg-based BMG DTx has a close correlation to bond parameters. It
increases with the atomic size parameter (d) and electronegativity difference (Dx) of Mg-based alloys, which are defined
by atomic radius and electronegativity. An equation was deduced for the first time by a linear regression analysis
program in order to judge whether Mg-based multicomponent alloys can form metallic glasses.
Ó 2003 Elsevier Science B.V. All rights reserved.

1. Introduction vention of crystallization [3,4]. In the past decades,

Bulk metallic glasses (BMG) have been inten-
sively studied due to their excellent mechanicals,
good magnetic properties and good corrosion re-
sistance since the late 1980s. However, the question
of what families and ranges of alloy compositions
are easy of preparation in BMG remains an issue
[1,2]. It is evident that the width of supercooled
liquid region (SLR) DTx can be used to represent
the glass forming ability for BMG; and most im-
portantly, a larger DTx provides a larger tempera-
ture scale to measure the physical properties of the
alloy in supercooled liquid state, permitting the
measurements to be made continuously between
the amorphous and the liquid state without inter-
*
Corresponding author.
E-mail address: ssfang@eastday.com (S. Fang).
the different widths of SLRs of Mg-, Ln-, Zr-, Ti-,
Fe- and Pd-based BMG were presented. It is ur-
gently necessary for scientists to find BMG with
large glass forming ability or to predict the width of
SLR simply by the structural information (e.g.
atomic size, electronegativity, valence electrons)
using a universal predictive model. There are some
common empirical rules of wide SLR such as sig-
nificant atomic size ratios above 12% among mul-
ticomponent alloys and large negative heat of the
mixture that are closely related to atomic packing
states [5]. Zhang [6] proposed an empirical criterion
by means of bond parametric functions that com-
prises of electronegativity difference, atomic size
parameter and electron to radius ratio. But this
qualitative model can only be used to judge whe-
ther a binary alloy made by rapid quenching can
form metallic glass. Shek et al. [7] proposed an-
other empirical criterion for high glass forming

0022-3093/03/$ - see front matter Ó 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0022-3093(03)00155-8
 S. Fang et al. / Journal of Non-Crystalline Solids 321 (2003) 120–125 121

ability by Hume–Rothery rules that the composi- Table 1

tion of the amorphous alloy and its crystalline The covalent radius and electronegativity of elements in the
Periodic Table
counterparts should be near a constant VEC (va-
Name of element Abbrevia- Electro- Covalent
lence electron concentration) ÔsurfaceÕ in the qua-
tion negativity radius
ternary Zr–Al–Ni–Cu systems. Senkov and (nm)
Miracle [8] suggested a model that the alloys with
Hydrogen H 2.2 0.037
best glass-forming ability have a concave upward Helium He – 0.14
shape of atomic size distributions versus the atomic Lithium Li 0.95 0.123
concentration after analyzing the Zr-, Cu-, Ln- and Beryllium Be 1.57 0.090
Pd-based bulk amorphous alloys. From topologi- Boron B 2.04 0.082
Carbon C 2.55 0.077
cal and chemical points of view, it is clear that two
Nitrogen N 3.04 0.075
critical parameters, i.e., atomic size ratios and Oxygen O 3.44 0.073
electronegativity difference among elements in Fluorine F 3.98 0.072
multicomponent alloys, played an important roll in Sodium Na 0.93 0.154
affecting the width of SLR. However, few papers Magnesium Mg 1.31 0.130
Aluminum Al 1.61 0.118
were published dealing with them and it is not clear
Silicon Si 1.9 0.117
for multicomponent alloys to mention the differ- Phosphor P 2.19 0.106
ence in atomic size and electronegativity though it Sulfur S 2.58 0.102
is simple in a binary alloy. In this paper, we try to Potassium K 0.82 0.196
investigate the relationship between the widths of Calcium Ca 1.00 0.174
Scandium Sc 1.36 0.144
supercooled regions of Mg-based metallic glasses
Titanium Ti 1.54 0.132
and bond parameters, i.e., the electronegativity Vanadium V 1.63 0.125
difference and atomic size parameter for multi- Chromium Cr 1.66 0.127
component alloys. Manganese Mn 1.55 0.146
Iron Fe 1.83 0.120
Cobalt Co 1.88 0.126
Nickel Ni 1.91 0.12
2. Relationship between DTx and bond parameters Copper Cu 2.00 0.138
Zinc Zn 1.65 0.131
The electronegativity difference Dx among the Gallium Ga 1.81 0.126
elements of a multicomponent alloy is defined by Germanium Ge 2.01 0.122
Arsenic As 2.18 0.120
Eq. (1) according to the theory of statistics pre-
Selenium Se 2.55 0.116
sume that the atoms among elements were ran- Rubidium Rb 0.82 0.211
domly packed [9]: Strontium Sr 0.95 0.192
Yttrium Y 1.22 0.162
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X n Zirconium Zr 1.33 0.148
2 Niobium Nb 1.6 0.137
Dx ¼ Ci  ðxi  xÞ ; ð1Þ
i¼1 Molybdenum Mo 2.16 0.145
Technetium Tc 1.9 0.156
Ruthenium Ru 2.2 0.126
where n is the number of components of the alloy;
Rhodium Rh 2.28 0.135
xi is the Pauling electronegativity of element i, Palladium Pd 2.20 0.131
which is obtained from literature and given in Silver Ag 1.93 0.153
Table 1 [10,11]; Ci is the atomic percentage of el- Cadmium Cd 1.69 0.148
ement i in the alloy; and x is the arithmetical mean Indium In 1.78 0.144
Tin Sn 1.96 0.141
value of electronegativity for a compound, which
Antimony Sb 2.05 0.140
can be calculated as follows: Tellurium Te 2.1 0.136
Cesium Cs 0.79 0.225
X
n
Barium Ba 0.89 0.198
x ¼ Ci  xi : ð2Þ Lanthanum La 1.1 0.169
i¼1 (continued on next page)
122 S. Fang et al. / Journal of Non-Crystalline Solids 321 (2003) 120–125

Table 1 (continued)
Name of element Abbrevia- Electro- Covalent
tion negativity radius
(nm)
Cesium Ce 1.12 0.165
Praseodymium Pr 1.13 0.165
Neodymium Nd 1.14 0.164
Promethium Pm 1.2 0.167
Samarium Sm 1.17 0.166
Europium Eu 1.2 0.185
Gadolinium Gd 1.2 0.161
Terbium Tb 1.2 0.159
Dysprosium Dy 1.22 0.159
Holmium Ho 1.23 0.158
Erbium Er 1.24 0.157
Thulium Tm 1.25 0.156
Ytterbium Yb 1.1 0.170 Fig. 1. Relationship between DTx and electronegativity differ-
Lutetium Lu 1.27 0.156 ence of Mg-based BMG.
Hafnium Hf 1.3 0.144
Tantalum Ta 1.5 0.134
Tungsten W 2.36 0.130
dency from Figs. 1 and 2 that DTx increases with
Rhenium Re 1.9 0.128 the electronegativity difference and atomic size
Osmium Os 2.2 0.125 parameter. We assume that DTx is proportional to
Iridium Ir 2.2 0.126 the two bond parameters Dx and d, which can be
Platinum Pt 2.28 0.129 expressed by an equation as follows:
Gold Au 2.54 0.134
Mercury Hg 2.00 0.144 DTx ¼ a0 þ a1  Dx þ a2  d; ð5Þ
Thallium Tl 2.04 0.155
Plumbum Pb 2.33 0.154 where a0 is constant that stands for the tempera-
Bismuth Bi 2.02 0.152 ture interval of pure metal, a1 and a2 are propor-
Polonium Po 2.0 0.153 tional coefficient, which can be obtained from
Table 2 using linear regress program analysis by a
And the atomic size parameter is defined as fol- computer. For Mg-based BMG system (not in-
lows [9]: clude binary alloys from which BMG are not
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi produced [8]), DTx can be calculated as follows:
u n  2
uX ri
d¼t Ci  1  ; ð3Þ DTx ¼ 13:52254 þ 113:35386  Dx
i¼1
r
þ 264:37921  d: ð6Þ
where ri is the covalent atomic radius of element i; The relative coefficient of Eq. (6) is 0.91. Here,
r is the arithmetical mean value of covalent atomic a0 is equal to )13.52254, less than zero, which
radius for a compound. The covalent atomic ra- means that pure metal (Dx ¼ d ¼ 0) cannot form
dius of each element is obtained from the literature BMG. Fig. 3 shows both the experimental and
and listed in Table 1 [12,13]. r can be calculated as calculated values of DTx for Mg-based BMG sys-
follows: tems, which indicates that the proposed model is
X n rational. It is reasonable from Eq. (6) that the SLR
r ¼ C i  ri : ð4Þ of metallic glass (DTx ) becomes wider with the in-
i¼1
crease of electronegativity difference as well as the
Fig. 1 shows the relationship between DTx and atomic size difference. The variable range of DTx
electronegativity difference of Mg-based BMG caused by Dx is 20.698–40.161 with Dx increasing
listed in Table 2 whereas; Fig. 2 shows the rela- from 0.1826 to 0.3543 whereas; the variable range
tionship between DTx and atomic size parameter of DTx caused by d is 6.345–21.626 with d in-
defined by Eqs. (1) and (3). There is a clear ten- creasing from 2.4% to 8.18% in this paper. It seems
 S. Fang et al. / Journal of Non-Crystalline Solids 321 (2003) 120–125 123

Table 2
Bond parameters and DTx in Mg-based BMG
Components Dx d (%) DTx (experimental) DTx (calculated) Reference
Mg85 Ni10 Ca5 0.1970 3.41 17 17.83 [14]
Mg87:5 Ni10 Ca2:5 0.1889 2.75 13 15.15 [14]
Mg82:5 Ni15 Ca2:5 0.2228 3.11 15 19.96 [14]
Mg79:4 Ni10:4 Nd10:2 0.1899 4.56 21 20.06 [15]
Mg75:8 Ni14:6 Nd9:6 0.2186 4.47 22 23.07 [15]
Mg65:3 Ni19:9 Nd14:8 0.2506 6.06 34 30.91 [15]
Mg80 Ni10 Nd10 0.1866 4.49 16.3 19.50 [16]
Mg75 Ni15 Nd10 0.2213 4.68 20.4 23.94 [16]
Mg70 Ni15 Nd15 0.2245 6.01 22.3 27.83 [16]
Mg85 Ni10 Y5 0.1826 3.17 15 15.56 [17]
Mg80 Ni15 Y5 0.2170 3.47 16 20.26 [17]
Mg80 Ni10 Y10 0.1850 4.39 23 19.04 [17]
Mg75 Ni10 Y15 0.1872 5.71 26 22.81 [17]
Mg75 Ni15 Y10 0.2197 4.58 28 23.49 [17]
Mg70 Ni20 Y10 0.2459 4.74 22 26.88 [17]
Mg70 Ni15 Y15 0.2222 5.84 31 27.11 [17]
Mg65 Ni20 Y15 0.2487 5.95 28 30.40 [17]
Mg60 Ni25 Y15 0.2694 6.03 37 32.96 [17]
Mg50 Ni30 Y20 0.2887 7.33 41 38.58 [17]
Mg75 Cu20 Ca5 0.2916 2.92 24 27.24 [14]
Mg55 Cu40 Ca5 0.3569 3.23 35 35.46 [14]
Mg82:5 Cu15 Ca2:5 0.2543 2.40 20 21.63 [14]
Mg80 Ni15 La5 0.2233 3.60 25 21.30 [18]
Mg75 Ni20 La5 0.2494 3.83 20 24.87 [18]
Mg80 Ni10 La10 0.1972 4.76 28 21.41 [18]
Mg75 Ni15 La10 0.2316 4.94 27 25.79 [18]
Mg70 Ni20 La10 0.2581 5.09 32 29.18 [18]
Mg65 Ni25 La10 0.2789 5.20 26 31.85 [18]
Mg70 Ni15 La15 0.2392 6.45 28 30.63 [18]
Mg65 Ni20 La15 0.2661 6.55 33 33.94 [18]
Mg50 Ni35 La15 0.3174 6.72 45 40.22 [18]
Mg55 Ni25 La20 0.2952 8.12 34 41.42 [18]
Mg45 Ni35 La20 0.3265 8.18 48 45.12 [18]
Mg80 Cu15 Y5 0.2490 2.57 19 21.51 [17]
Mg80 Cu10 Y10 0.2117 2.99 21 18.37 [17]
Mg60 Cu30 Y10 0.3232 3.39 35 32.07 [17]
Mg70 Cu15 Y15 0.2540 3.78 33 25.26 [17]
Mg60 Cu25 Y15 0.3082 3.89 38 31.69 [17]
Mg70 Cu10 Y20 0.2159 4.40 23 22.59 [17]
Mg60 Cu20 Y20 0.2871 4.49 31 30.90 [17]
Mg40 Cu40 Y20 0.3543 4.44 42 38.37 [17]
Mg50 Cu30 Y20 0.3297 4.50 29 35.76 [17]
Mg65 Cu10 Y25 0.2179 5.12 22 24.72 [17]
Mg65 Cu25 Mm10 0.3159 3.53 35 31.62 [19]
Mg65 Y10 Cu15 Al10 0.2564 8.09 43 36.93 [20]
Mg57:6 Y10 Cu30 Al2:4 0.3220 7.27 44 42.21 [20]
Mg57 Y10 Cu30 Al3 0.3217 7.34 38 42.34 [19]
Mg65 Y10 Cu15 Ni10 0.2904 7.88 35 40.23 [19]
Mg65 Y10 Cu15 Zn10 0.2594 7.13 36 34.72 [19]
Mg65 Y10 Cu15 Ag10 0.2936 8.01 41 40.94 [21]
Mg55 Y10 Cu25 Li10 0.3372 6.94 49 43.06 [22]
Mg50 Y10 Cu25 Li15 0.3510 6.87 42 44.44 [22]
Mg65 Y10 Cu15 Ag5 Pd5 0.3196 7.63 37 42.89 [23]
124 S. Fang et al. / Journal of Non-Crystalline Solids 321 (2003) 120–125
Fig. 2. Relationship between DTx and atomic size parameter of
Mg-based BMG.
Fig. 3. A comparison between the calculated and experimental
values of DTx of SLR in Mg-based BMG.
that electronegativity difference is as important as
the atomic size parameters in dominating the
widths of SLRs for Mg-based metallic glasses.
3. Discussion
3.1. The necessary condition for Mg-based alloys to
form metallic glasses
As we know, the temperature interval of SLR in
BMG is greater than zero. It can be obtained from
Fig. 4. Relationship between DTx and bond parameters for Mg-
based BMG (d–Dx–Tx curve).
Eq. (6) that the necessary condition for Mg-based
alloys to form BMG is as follows:
113:35386  Dx þ 264:37921  d > 13:52254: ð7Þ
Or approximately
8:38  Dx þ 19:55  d > 1: ð8Þ
Fig. 4 illustrates the relationship between DTx
and bond parameters. Fig. 4 also illustrates the
metallic glass forming range of Mg-based alloys.
The temperature interval DTx of SLR for Mg-
based BMG can be plotted as a series of lines,
whose slop is )0.429. The slop would be different
for different type BMG. Generally, it is of impor-
tance to denote that the relationship between DTx
in BMG and bond parameters given by Eq. (5) can
apply to other type BMGs with the constant a0
and proportional coefficient a1 , a2 changing a little,
which would be published later.
3.2. The covalent atomic radius of elements in bulk
metallic glasses
The atomic covalent radius of element would
change a little with different coordination numbers
and different bond type in different alloys [24]. This
could explain a few exceptions founded from Eq.
(6) such as Mg60 Ni25 La15 (43.5 K by calculation, it
was 58 K in literature [18]). If the atomic radius of
Mg and Ni changed from 0.13 and 0.12 nm [12] to
0.137 and 0.115 nm in literature [13] (changing 5%
 S. Fang et al. / Journal of Non-Crystalline Solids 321 (2003) 120–125
and 4% respectively), the calculated DTx would be
52 K, which is closer to the value in literature. If
the actual atomic radius were used, the results
would be better. Nevertheless, it is very useful and
convenient to use the Pauling electronegativity and
covalent atomic radius to make some predictions
for the widths of SLRs of BMG because the real
atomic radii are difficult to be measured.
4. Conclusions
The relationship between the widths of SLRs of
Mg-based BMG and bond parameters was inves-
tigated. There exists a clear tendency that the
widths of supercooled regions increase with the
atomic size difference (d) and electronegativity
difference (Dx) defined in this paper. It seems that
electronegativity difference is as important as
atomic size parameter in dominating the widths of
SLRs for Mg-based BMG. An equation was de-
duced for the first time by linear regress program
analysis to evaluate the widths of SLRs by elec-
tronegativity difference and atomic size parameter,
which is also suitable for Mg-based quaternary or
quinary systems.
Acknowledgements
The authors would like to acknowledge the
support of the National Natural Science Founda-
tion of China under grant numbers 59971028 and
50031010. This work was financially supported by
the developing foundation of Shanghai municipal
educational committee (203020).
125
References
[1] R.J. Gottschall, Mater. Trans. 42 (4) (2001) 548.
[2] W.L. Johnson, MRS Bull. 24 (1999) 42.
[3] A. Inoue, T. Zhang, T. Masumoto, J. Non-Cryst. Solids
156–158 (1993) 473.
[4] W.-H. Wang, Q. Wei, S. Friendrich, Phys. Rev. B 57 (14)
(1998) 8211.
[5] A. Inoue, T. Zhang, J. Non-Cryst. Solids 250 (1998) 552.
[6] B.W. Zhang, ACTA Metall. Sinica 17 (3) (1981) 285.
[7] C.H. Shek, Y.M. Wang, C. Dong, Mater. Sci. Eng. A 291
(2000) 78.
[8] O.N. Senkonv, D.B. Miracle, Mater. Res. Bull. 36 (2001)
2183.
[9] S.S. Fang, Z.Q. Zhou, J.L. Zhang, et al., J. Alloys Compd.
293–295 (1999) 10.
[10] L. Pauling, The Nature of the Chemical Bond, 3rd Ed.,
Cornell University Press, Ithaca, 1960.
[11] A.L. Allred, J. Inorg. Nucl. Chem. 17 (1961) 215.
[12] L. Liao (Ed.), Crystalline Chemistry and Crystalline
Physics, Geological Publishing House, Beijing, 2000,
p. 177 (in Chinese).
[13] Z. Pan (Ed.), Crystallography and Mineralogy, Geological
Publishing House, Beijing, 1984, p. 133 (in Chinese).
[14] T. Shibata, A. Inoue, T. Masumoto, J. Mater. Sci. 28
(1993) 379.
[15] Y. Li, H. Jones, H.A. Davies, Scripta Metall. Mater. 26
(1992) 1371.
[16] Y. Li, Mater. Trans. 42 (4) (2001) 556.
[17] S.G. Kim, A. Inoue, M. Kohinata, T. Masumoto, Mater.
Trans., JIM 31 (11) (1990) 929.
[18] A. Inoue, M. Kohinata, T. Masumoto, et al., Mater.
Trans., JIM 30 (5) (1989) 378.
[19] B.S. Murty, K. Hono, Mater. Trans. 41 (11) (2000) 1538.
[20] S. Linderoth, N.H. Pryds, A. Inoue, et al., Mater. Sci. Eng.
A 304–306 (2001) 656.
[21] H.G. Kang, E.S. Park, W.T. Kim, et al., Mater. Trans.,
JIM 41 (7) (2000) 846.
[22] W. Liu, W.L. Johnson, J. Mater. Res. 11 (9) (1996) 2388.
[23] K. Amiya, A. Inoue, Mater. Trans. 41 (11) (2000) 1460.
[24] J. Xiao (Ed.), Alloy Energetics, Shanghai Science and
Technology Publishing House, Shanghai, 1985, p. 294.