Extreme Mechanics Letters (

Contents lists available at ScienceDirect

Extreme Mechanics Letters

journal homepage: www.elsevier.com/locate/eml

Thermally induced failure mechanism transition and its correlation

with short-range order evolution in metallic glasses
Mehdi Jafary-Zadeh a , Rouhollah Tavakoli b , David J. Srolovitz c, , Yong-Wei Zhang a,
a

Institute of High Performance Computing (IHPC), A*STAR, Singapore 138632, Singapore

Department of Materials Science and Engineering, Sharif University of Technology, Tehran 113659466, Iran

Departments of Materials Science and Engineering & Mechanical Engineering and Applied Mechanics, University of Pennsylvania, Philadelphia,
PA 19104, USA

graphical

article

abstract

info

Article history:
Received 13 February 2016
Received in revised form
23 June 2016
Accepted 26 July 2016
Available online xxxx
Keywords:
Metallic glasses
Failure mechanisms
Temperature effect
Molecular dynamics
Atomistic structure
Mechanical properties
Brittle
Ductile
Plastic flow

abstract
The effect of temperature on the short-range order (SRO) structures, deformation mechanisms and
failure modes of metallic glasses (MGs) is of fundamental importance for their practical applications.
However, due to lack of direct structural information at the atomistic level from experiments and the
absence of previous molecular dynamics (MD) simulations to reproduce experimental observations over
a wide range of temperature, this issue has not been well understood. Here, by carefully constructing
the atomistic models of Cu64 Zr36 and Fe80 W20 MGs, we are able to reproduce the major deformation
modes observed experimentally, i.e. single shear banding (SB) at low temperatures, multiple shearbandings at intermediate temperatures and homogeneous plastic flow at elevated temperatures. By
examining the evolution of SRO, we find that the different failure modes exhibit distinctively different
full-icosahedron (FI) evolution pathways at different temperatures. Specifically, at low temperatures, the
FI concentration first deceases to a minimum, then recovers slightly, and finally comes to a plateau; at
intermediate temperatures, it first decreases and then reaches a plateau; while at elevated temperatures,
it decreases simply monotonically. These different pathways arise from the dynamic competition between
the destruction and recovering of FI clusters. We further show that the local softening caused by the
destructions of FI clusters is crucial for the formation of localized shear planes and further shear bands.
Since our simulations exhibit the same trend for both MG systems, it is expected that these findings may
be generic for a wide range of MGs.
2016 Elsevier Ltd. All rights reserved.

1. Introduction

Corresponding authors.
E-mail addresses: srol@seas.upenn.edu (D.J. Srolovitz),
zhangyw@ihpc.a-star.edu.sg (Y.-W. Zhang).
http://dx.doi.org/10.1016/j.eml.2016.07.009
2352-4316/ 2016 Elsevier Ltd. All rights reserved.

With increasing demands to develop advanced structural

materials which are stiffer and tougher so as to be used under
extreme conditions (e.g. high temperature and/or high applied

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

stress), metallic glasses (MGs), which combine the metallic

bonding and amorphous atomic structure, are at the cutting edge
of materials research for such endeavours [1,2]. Here, amorphous
refers to lack of long-range order (translational symmetry) in
material structures, whereas short-to-medium-range order (on the
length scale of a few to tens of atoms) still exists [3,4]. Due to
the absence of long-range order, conventional structural defects,
such as dislocations, are absent in metallic glasses. Consequently,
there is no dislocation-mediated slip in MGs, and hence, compared
to their crystalline counterparts, MGs offer several superior
mechanical properties, such as high yield strength (close to the
theoretical limit), high elastic limit and high hardness [1,2,5,6].
However, MGs are in disrepute for their highly localized plastic
deformation at low temperatures via shear banding. As a result,
MGs often exhibit nearly zero ductility and catastrophic failure [1].
So far, extensive research has been performed to understand their
deformation mechanisms and failure modes under a variety of
environmental conditions (see the review paper [5] and references
therein), aiming to enhance their ductility and suppress their
catastrophic failure.
It is well-known that temperature is an important environmental parameter that is able to significantly affect the deformation
mechanism and failure mode of MGs [6]. Experimental studies
revealed that the deformation mechanisms of MGs vary greatly
with temperature [69]. At low temperatures (typically, well below 0.6Tg , where Tg denotes the glass transition temperature),
MGs are in the brittle regime [6,9,10]. In this regime, failure is
catastrophic through single shear band (SB) formation along the
principal shear direction of the samples [11]. At intermediate temperatures, MGs initially exhibit a certain inhomogeneous plastic
deformation characterized by the appearance of multiple shear
bands, followed by a dominant single shear banding failure [8]. In
contrast, at elevated temperatures (typically, greater than 0.6Tg ),
metallic glasses undergo a homogeneous deformation [6,7,9]. Such
a temperature-dependent mechanical behaviour is not only scientifically fascinating but also technologically important. For example, it allows hot-forming operation at high temperature and
achieving of high strength and hardness after cooling/annealing.
In fact, it recently found applications in the development of microand nano-electromechanical systems (MEMS/NEMS) [1214]. To
expedite the applications of MGs, it is crucial to gain a solid understanding of their temperature-dependent mechanical behaviour.
Moreover, from the scientific point of view, since MGs have relatively simple atomic structures compared to other types of glasses,
such as polymeric and ceramic ones [15,16], they also offer an
opportunity to understand the fundamental behaviour of noncrystalline solids [17,18].
For crystalline materials, owing to their periodic lattice structures and well-defined structural defects, their plastic deformation mechanisms are well established. In contrast, even modern
advanced microstructural characterization techniques cannot
readily evaluate short- and medium-range orders in amorphous
MGs [4]. In this regard, molecular dynamics (MD) simulations
provide a powerful tool to analyse the 3-dimensional atomistic
structure of MGs, and reveal the correlations between their mechanical behaviour and structural features, which otherwise are
not readily achievable experimentally. Indeed, MD simulations
have been extensively employed to study the mechanical behaviour, such as shear banding, viscosity and crack initiation/
propagation in a variety of MG systems (like monolithic MGs and
nanoglasses) [1926]. For example, MD simulations were used to
study the steady-state flow of a MG sample under pure shear stress
condition, and it was revealed that there is a scaling relation between stress and temperature [27]. In addition, MD simulations
were performed to study the effect of configurational potential energy and temperature on the elastic properties of CuZr glass [28].

The calculated temperature-dependence of Youngs modulus of a

Cu64 Zr36 MG sample using MD simulations showed a good agreement with experimental measurements [29]. Besides, MD simulations were used to study the homogeneous deformation of MG
samples at elevated temperatures and found that the pronounced
tensile ductility is due to the creation of high content of free volume
in the atomic structure, which leads to the decrease in viscosity
[29,30]. It should be noted that, however, these simulations were
performed either in a narrow temperature range or at a temperature which is well below room temperature. It is worth noting
that the concept of localization of plastic deformation in metallic glasses and its transition to a global (homogeneous) plastic
deformation is a ubiquitous phenomenon in MGs. This transition
has been studied experimentally or by simulations for a variety of
parameters such as strain rate [31,32], cooling rate [3236], sample size and/or geometrical constrains [34], chemical composition [37], etc. However, to our knowledge, there is still lack of
coherent MD simulations, which are able to reproduce the various
deformation mechanisms and failure modes observed experimentally in a broad range of temperature. By examining the methodologies of previous MD simulations, we found that in order to
capture the temperature effect in a wide range of temperature, several parameters must be carefully taken into account in the simulations setup, such as appropriate glass preparation (e.g. not using
too fast quench rate), large enough sample, and proper boundary
conditions.
In the present study, we employ large-scale MD simulations
with carefully constructed simulation models to explore the
mechanical behaviour of two typical binary metallic glasses,
Cu64 Zr36 and Fe80 W20 , under uniaxial tension at a wide range of
temperature. We aim to:
1- perform MD simulations to reproduce the various deformation
mechanisms and failure modes observed experimentally at a
broad range of temperature;
2- correlate different deformation mechanisms and failure modes
at different temperatures to the atomistic structural evolution
of MGs.
In principle, the collective structure of the entire MG would
be responsible for its macroscopic mechanical behaviour. Hence,
medium range order (MRO), i.e. the structural features beyond
the nearest neighbour shell of atoms, may also have an important
role in mechanical response of the glass. While in the literature,
there are a variety of definitions for MRO in MGs (e.g., connectivity
schemes of SRO clusters [16], Bergman superclusters [38], fractallike structures [39], etc.), our focus in the current work is on
the SRO structures. As widely accepted in the literature [3,16,40],
the SRO in MGs can be characterized by the Voronoi tessellation
method [41,42]. From this perspective, naturally a next stage for
the progression of this study would be examining the effect of
temperature on the structural organizations beyond SRO and its
correlations with mechanical response.
2. Methods and model
Our molecular dynamics (MD) simulations were performed
using Large-Scale Atomic/Molecular Massively Parallel Simulator
(LAMMPS) [43], which is a well-known and widely accepted open
source software package. The many-body interatomic interactions
in our simulations were calculated using the embedded atom
model (EAM) potentials for both CuZr [44] and FeW [45] metallic
system. Cu64 Zr36 MG has a simple atomic structure. In addition, a
well-established empirical EAM potential is available for this MG.
As a result, this MG is one of the most well-studied prototypical
MGs employed to gain insights into their structural and mechanical
properties. We choose this MG since a rich database obtained from

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

previous studies allows us to make comparison. To examine the

generality of the temperature-induced transitions in deformation
modes observed in Cu64 Zr36 MG, we have also chosen another
binary MG, Fe80 W20 , which also has a relatively simple atomic
structure, but a much higher glass transition temperature and
higher mechanical strength, and available EAM potential for
MD simulations. To carefully construct the atomistic models of
Cu64 Zr36 and Fe80 W20 MG samples, we followed the melting
and quenching technique using a representative cubic sample
containing about 10,000 atoms. Periodic boundary conditions
(PBCs) were applied along all the three directions. We started with
the solid solutions of the desired alloy compositions in which the
solute atoms were randomly substituted in the solvent lattice. In
the case of Cu64 Zr36 (Fe80 W20 ) the solid solution was melted at
2300 K (3500 K) and the melt was equilibrated for 2 ns. Then,
the liquid alloy was quenched to 50 K with the cooling rate of
10 K/ns, which was slower than that used in all previous MD
simulations. The simulation time step during our melting-andquenching procedure was 0.002 ps and the isothermalisobaric
(NPT) ensemble was used to keep the pressure of the system at
zero. The glass transition temperature, Tg , of the MGs is determined
by analysing the evolution of average potential energy per atom as
a function of temperature during the simulated quenching process
of their liquids [46,47]. The samples for the uniaxial tensile test
were then constructed by replicating the representative MG cube.
The geometry of the tensile sample is shown in Fig. 1 which is a
rectangular slab with the dimensions of 40 5 120 nm in the
x, y and z directions, respectively. PBCs were applied along the
z direction (the loading direction) and the y direction, while the
sample had free surfaces in the x-direction, i.e. along its thickness.
Applying PBCs along z and y axes resembles the bulk material
in these directions. Besides, the selected width of 40 nm along x
direction (free surface) is large enough to avoid undesirable sizeinduced transition in deformation mode from shear banding to
necking, which typically occurs in MD simulations by reducing the
width to the range from 7.5 nm to 20 nm, depending on the sample
preparation [48,49]. Before loading, the sample was equilibrated
at desired test temperature. During loading, the temperature was
kept constant and a zero stress condition was maintained along
the y direction to mimic the uniaxial stress state. The loading
was applied by rescaling the simulation box in the z direction
with the strain rate of 107 s1 . The Verlet algorithm [50] was
used to integrate the equations of motion and the NoseHoover
thermostat/barostat [51] were used to control the temperature
and pressure of the system. The time step during the loading
process was 0.001 ps. The engineering stress was calculated from
the normal component (along the z direction) of the atomic stress
tensor based on the Virial stress theorem [52,53]. The atomic
von Mises strain was adopted as a measure of local (atomistic)
deformation in the sample [54]. To analyse the evolution of SRO
in the MG sample during deformation, we used our in-house
code which was developed based on the Voro++ package [55].
To generate the atomistic visualization, we employed the OVITO
open source package [56]. We emphasize that a combination of
the large-scale sample sizes, the slow quenching rate and the
slow strain rate is crucial to reproduce the failure modes observed
experimentally at low, intermediate and high temperatures as
shown below.
3. Results and discussion
The engineering stressstrain curves and deformation behaviour of the Cu64 Zr36 glass at different temperatures are presented in Fig. 2. The temperatures used to perform the uniaxial
tensile tests on the Cu64 Zr36 MG sample are 50 K (0.06Tg ), 300 K
(0.38Tg ), 480 K (0.6Tg ) and 600 K (0.77Tg ), where the glass transition temperature (Tg ) of this glass sample obtained from our

Fig. 1. Geometry of the MG sample used in the uniaxial tensile test simulations. The
dimensions of the sample were 40 nm, 5 nm and 120 nm in x, y and z directions,
respectively. PBCs were applied in y and z directions. A uniaxial tensile strain was
applied in z direction.

MD simulations is about 780 K. From Fig. 2(a), it can be seen that

at the low temperature of 50 K (0.06Tg ), the stress level drops
abruptly after the stress peak, i.e. the ultimate tensile strength
(UTS), which is characteristic of the brittle failure of MGs as a result of the severely localized plastic deformation [11]. By increasing
the temperature to 300 K (0.38Tg ), the sharp falling of the stress
becomes relatively smoother, and the material strength decreases.
With further increasing the temperature to 480 K (0.6Tg ) and 600 K
(0.77Tg ), the deformation behaviour clearly alters towards an increasingly more ductile behaviour. To further probe this transition,
we study the atomic von Mises shear strain, vM , with respect to the
load-free samples at temperatures of 50 K (0.06Tg ), 300 K (0.38Tg ),
480 K (0.6Tg ), and 600 K (0.77Tg ) as shown in Fig. 2(b), (c), (d), (e),
respectively. The snapshots in each panel present the status of the
sample at different stages of applied tensile strain, .
Fig. 2(b) shows that the failure mechanism at the cryogenic
temperature of 50 K is through the formation and propagation
of a single shear band (SB) along the principal shear direction
of the sample (45 with respect to the loading direction). This
clearly illustrates a brittle failure mode, which is in agreement
with the experimental observation of single shear banding and
brittle behaviour of MGs at cryogenic temperatures [57,58].
According to Fig. 2(c), at 300 K (0.38Tg ), the failure of the
sample is still in a brittle mode through the formation of a
dominant SB along the whole sample width. However, before
the formation of the dominant SB, there is an enhanced plastic
deformation occurring in the sample, which is manifested as the
formation of multiple SBs as shown in Fig. 2(c). The observation
of multiple SBs at room temperature in our simulations is
also in good agreement with experimental results [58,59]. In
contrast to the inhomogeneous plastic deformation at low and
intermediate temperatures, the deformation behaviour at elevated
temperatures changes drastically. As can be seen in Fig. 2(d, e),
at the temperatures of 480 K (0.6Tg ) and 600 K (0.77Tg ),
the severe deformation localization disappears. Fig. 2(e) illustrates
that at the elevated temperature of 600 K (0.77Tg ), the local

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

Fig. 2. (a) Engineering stressstrain curves for the Cu64 Zr36 MG at different temperatures. Snapshots showing different deformation mechanisms of the Cu64 Zr36 MG at
these temperatures: (b) failure through a single shear banding at 50 K (0.06Tg ), (c) failure through initially multiple shear bandings and subsequent dominant shear banding
at 300 K (0.38Tg ), (d) largely homogeneous plastic deformation at 480 K (0.6Tg ), and (e) homogeneous plastic deformation at 600 K (0.77Tg ). The colour bar denotes the
atomic shear strain. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

atomic strains are almost uniformly distributed across the whole

sample up to high applied strains. The observed transition of
deformation mode in our MD simulations from localized shear
banding at low temperatures (below 0.6Tg ) to homogeneous
deformation at elevated temperatures (above 0.6Tg ) is consistent
with experimental observations [6,7,9].
To further quantify the degrees of localization, we adopt
the standard deviation of atomic shear strain, vM , as the
degree
of strain localization parameter as the following [37]:

1
N

i=1

vM 2
vM
(ivM ave
) , where ave
is the average atomic

shear strain, and N is the total number of atoms in the

system. evaluates the deviation of strain distribution from the
homogeneous behaviour: A larger indicates larger fluctuations
in the local atomic strain and a more localized deformation mode.
Fig. S5 in the Supplementary Material (see Appendix A) shows
the evolution of parameter during uniaxial tensile loading of
Cu64 Zr36 MG sample in the temperature range of 50 K (0.06Tg )
up to 600 K (0.77Tg ). It can be seen that at the low temperature
of 50 K (0.06Tg ), the atomic deformation is homogeneous in the
elastic regime while there is a very steep increase in values
at applied strain of > 8%, which is attributed to the formation
of shear band and severe localization of plastic deformation.
By increasing the temperature to 300 K (0.38Tg ), the strain
localization corresponding to the formation of a dominant shear
band delays to a higher applied strain ( > 9%) with smoother
slope compared to that of 50 K, implying a slower rate of strain
localization at this temperature. In contrast, at the elevated
temperatures of 480 K (0.6Tg ) and 600 K (0.77Tg ), the values of
parameter remain small up to very high stages of applied strain,
indicating almost a homogeneous atomistic plasticity at elevated
temperatures.
To further examine the generality of the temperature-dependent
transitions observed in the Cu64 Zr36 MG, we also conducted simulations using the Fe80 W20 MG. The results are shown in Fig. 3,
in which the test temperatures are 50 K (0.03Tg ), 300 K (0.2Tg ),
900 K (0.6Tg ) and 1200 K (0.8Tg ), where Tg 1500 K for this glass.
Hence, the normalized test temperatures (T /Tg ) for the Fe80 W20
MG are comparable to those for Cu64 Zr36 MG, which allows us to
directly compare the deformation modes between these two MG
systems (see Figs. 2 and 3).

As can be seen clearly in Fig. 3, the temperature effect on the

overall mechanical behaviour of the Fe80 W20 MG is identical to that
of Cu64 Zr36 MG. More specifically, at the low temperature of 50 K
(0.03Tg ), there is a sharp stress drop (after the UTS) as the result of
the severely localized plastic deformation through the formation
and propagation of a single shear band (see Fig. 3(b)). At 300 K
(0.2Tg ), the post-yield stress drop becomes smoother, but the
failure mode is still brittle, resulting from the initial formation of
multiple shear bands and the subsequent formation of a dominant
shear band (see Fig. 3(c)). Similar to the Cu64 Zr36 sample, a further
increase of the test temperature to 0.6Tg (900 K) and above (see
Fig. 3(a), (d)) also leads to a progressively uniform plastic flow
in the Fe80 W20 MG. Fig. 3(e) clearly demonstrates that at the
temperature of 1200 K (0.8Tg ), there is no sign of inhomogeneous
plastic localization; instead, the sample undergoes a homogeneous
deformation similar to that of the Cu64 Zr36 MG at the same
normalized temperature (see Fig. 2(e)). A comparison between
the strength of the Fe80 W20 sample (Tg 1500 K) and that of the
Cu64 Zr36 sample (Tg 780 K) demonstrates that MGs with higher
Tg offer better mechanical properties at elevated temperatures,
and thus may be considered as high-strength thermal resistance
materials for structural applications in extreme environments.
Here, it would be instructive to recall that in MD simulations,
the strength of a given material is typically higher than its
experimental one. This is mainly attributed to the strain rate
of MD simulations, which is orders of magnitude higher than
that of experiments, due to the temporal limitations of MD
technique. Typically, a higher strain rate in MGs leads to a higher
yield strength of the material [60]. Besides, the samples in MD
simulations are free of inclusions, microscopic cracks, and surface
flaws such as roughness. Such unavoidable extrinsic flaws in
experimental MGs may lead to stress concentration, which in
turn reduces the materials strength. However, as discussed in
the Introduction section, MD simulations can still successfully
capture the major aspects of the mechanical behaviour of MGs
in different scenarios. Consequently, to interpret our simulation
results, we focus on the major mechanistic features and trends in
the behaviour of the samples.
As shown in Figs. 2 and 3, with increasing temperature, the
distribution of localized plastic zones becomes increasingly more
uniform and their local shear strain intensity is increasingly

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

Fig. 3. (a) Engineering stressstrain curves for the Fe80 W20 MG at different temperatures. Snapshots showing different deformation mechanisms of the Fe80 W20 MG at these
temperatures: (b) failure through a single shear banding at 50 K (0.03Tg ), (c) failure through initially multiple shear bandings and subsequent dominant shear banding at
300 K (0.2Tg ), (d) largely homogeneous plastic deformation at 900 K (0.6Tg ), and (e) homogeneous plastic deformation at 1200 K (0.8Tg ). The colour bar denotes the atomic
shear strain. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

reduced and dispersed. Hence, an increase in temperature is able

to postpone or even suppress the formation of the dominant shear
band. Regarding the formation of the dominant shear band, our
simulations show that, in general, there are four major steps:
(1). the nucleation and growth of individual localized shear zones
independently, (2). the linkup and coalescence of localized shear
zones along the principal shear direction to form shear planes,
which may be considered as the precursors for shear bands, (3).
thickening and expansion of shear planes, and (4). formation of a
single shear band at low temperatures or formation of multiple
shear bands first and the subsequent formation of the dominant
shear band at intermediate temperatures. These formation steps
are consistent with previous studies [4,11,54].
A salient feature of our results is that the experimentally
observed temperature-dependent transitions in the mechanical
response of MGs from localized shear banding at low temperatures
to homogeneous deformation at elevated temperatures [7,8] have
been captured using a unified MD framework. This offers an
opportunity to shed light on the detailed characteristics of this
transition at the atomistic level, which would not be readily
possible by current experimental techniques. As can be seen in
Fig. 4, the temperature-dependent transitions in the mechanical
response of the Cu64 Zr36 sample are strongly correlated with the
change in the sharpness of the shear band boundaries. Fig. 4(a)(c)
show the deformation behaviour of Cu64 Zr36 sample at 50 K
(0.06Tg ) and = 0.09, 300 K (0.38Tg ) and = 0.10, and 480 K
(0.6Tg ) and = 0.23, respectively. The shear band regions in
the upper frames of Fig. 4 are highlighted at the bottom frames
of the figure. At the low temperature of 50 K, a single shear band
occurs with sharp interfaces (also see Supplementary Fig S2 for
the Fe80 W20 MG, Appendix A). With increasing the temperature,
the boundaries of the shear bands become increasingly smeared
and diffusive (see Fig. 4(b), (c)). This remarkable change at the
boundaries of shear bands implies a strong temperature effect on
the structural evolution of the material [37].
In order to elucidate the effect of temperature on the structural
evolution of MGs at the atomistic level, we analyse SRO clusters,
i.e. the local atomic packing, using the Voronoi tessellation
method [41,42]. Such analysis is able to offer valuable information
about the nanoscale structures of MGs, such as the topology of
nearest neighbour shell, the atomic coordination number (CN), and
the corresponding statistical distributions [16,40].

We monitor the evolution of certain SRO clusters in terms

of their fractional variation as a function of applied strain, , at
different temperatures. Fig. 5(a)(d) show the evolution of the 10
most populous polyhedra in the Cu64 Zr36 MG (which are either Cuor Zr-centred) at the temperatures of 50 K (0.06Tg ), 300 K (0.38Tg ),
480 K (0.6Tg ) and 600 K (0.77Tg ), respectively. The common
nomenclature to identify a given Voronoi polyhedron in MGs is
given by Ai j k l, where A is the central atom of the polyhedron
consisting of i, j, k and l faces with 3, 4, 5 and 6 edges, respectively.
In general, four indexes are sufficient to describe the prominent
polyhedra, i.e. the SRO clusters, in MGs [3,19]. Summation of all
these indexes (total number of faces) gives rise to CN of the central
atom, i.e. CN = i + j + k + l, which is also the number of
its nearest neighbours. For example, in Fig. 5, the Voronoi index
Cu0 0 12 0 indicates a category of Cu-centred cluster, in which
the central Cu atom has 12 nearest neighbours (CN = 12), and
all the bonds between the central atom and its nearest neighbours
are five-fold bonds (local five-fold symmetry) [61,62]. This class
of clusters is often called full icosahedron (FI). As can be seen in
Fig. 5, FI is the dominant SRO in the Cu64 Zr34 MG, which is in
agreement with previous studies on the same material [19,35,40].
Fig. 5 shows that during deformation at a given temperature,
there is a certain correlation between the applied strain and the
concentration of the SRO clusters. In particular, considering the
concentrations of Cu0 0 12 0 or Zr0 1 10 5 in Fig. 5(a), it can
be seen that at the low temperature of 50 K, there is a decay in
the fractions of these clusters down to their minimum at about
= 9%. Afterwards, a limited recovery occurs in their fractions
and eventually these curves come to a plateau. At the temperature
of 300 K, the fractions of these clusters decrease first and then
come to plateau without noticeable recovery as shown in Fig. 5(b).
In contrast, at the elevated temperatures of 480 K and 600 K
(see Fig. 5(c) and (d)), the decays are monotonic. Thus, Fig. 5
demonstrates that temperature can affect the pathways in the
evolution of SROs in MGs. Our simulation results also show that
the Fe80 W20 MG follows the identical behaviour, indicating that
these observed features are common among MGs. Furthermore,
Fig. 5 also shows that at all the temperatures, the FI is the
dominant SRO in the structural evolution of the Cu64 Zr36 MG.
Moreover, we find the same trend in the Fe80 W20 MG (see Fig. S3
in the Supplementary Materials, Appendix A); the dominant SRO

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

Fig. 4. Effect of temperature on the sharpness of shear band (SB) boundaries of the Cu64 Zr36 MG. (a) 50 K (0.06Tg ) and = 0.09, (b) 300 K (0.38Tg ) and = 0.10, and (c)
480 K (0.5Tg ) and = 0.23. A comparison of (a), (b) and (c) shows that with increasing the temperature, the SB boundaries change from a sharp at low temperatures to a
more smeared interface at elevated temperatures.

Fig. 5. Structural evolution of short range order (SRO) clusters in the Cu64 Zr36 MG by applying tensile strain at different temperatures. These results indicate the key role of
full icosahedra (FI) clusters, i.e. Cu0 0 12 0, in the mechanical behaviour of the MG sample in the whole temperature range.

is Fe0 0 12 0, i.e. Fe-centred FI. These observations are consistent

with previous studies [19,63] that FI is the most prominent SRO in
MGs with different chemistries and compositions.
Fig. 5 shows that in all the populous SROs presented here,
the k index showing pentagonal faces has the highest value
among all indexes. Obviously, the FI cluster has the highest
number of pentagonal faces among all the SRO clusters. Hence, to
further characterize the temperature-dependent transitions in the
deformation mechanism of MGs, we focus on the evolution of FI
clusters as a function of applied strain at different temperatures.
Fig. 6 shows the variation of the FI concentration as a function
of applied strain (i.e. the FI evolution curves) at different
temperatures.

From Fig. 6, it can be seen that at the very low temperature

of 50 K (0.06Tg ), the concentration of FI reaches a minimum at
about = 9% and after a marked recovery, it comes to a plateau
at the applied strain of about = 10%, which corresponds to the
propagation of the dominant SB across the width of the sample
(see Fig. 2(a)). At the temperature of 300 K (0.38Tg ), the decay
of the FI evolution occurs till = 10% and then it comes to
a plateau at 11%, which corresponds to the appearance of
the dominant shear band in the sample (see Fig. 2(c)). At the
elevated temperatures of 480 K (0.6Tg ) and 600 K (0.77Tg ), the
concentration of FI clusters monotonically decreases up to high
values of applied strain. The above analyses demonstrate that
the evolution of FI clusters is closely correlated with different
deformation mechanisms. By comparing the two curves at 50 K

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

central atom A (with the radius of RA ) in a cluster is embedded in

the shell of atoms B (with the radius of RB ), c is given by [73]:

c =

15 1xc
2

where 1xc =

Fig. 6. Evolution of full icosahedra (FI) clusters in the Cu64 Zr36 MG as a function of
applied strain at different temperatures.

and 300 K, we find that in the regime of single SB, there is

a clear recovery in the concentration of the FI, however, this
recovery is absent in the regime of multiple SBs. By comparing
the curves at four temperatures, we find that the reach of the
plateau in the FI evolution signifies the formation of the dominant
SB in the sample. Indeed, such variation in the evolution of FI
clusters with applied strain at different temperatures is correlated
with different deformation mechanisms; primarily elastic at lower
temperatures and increasingly plastic (irreversible) deformation
at elevated temperatures, as discussed in detail in Section A
(Different deformation mechanisms vs. the evolution of SRO) in the
Supplementary Materials (see Appendix A). Hence, our simulations
reveal that the pathway of decayrecoveryplateau indicates
the formation of single SB, that of decayplateau indicates the
formation of multiple SBs initially and subsequent formation
of the dominant SB, and that of monotonic decay indicates a
homogeneous deformation. It is noted that a decrease in the
concentration of FI clusters (and also other populous SROs) are
equivalent to the increases in fragmented polyhedra, which are
composed of those low-populated clusters with an unfavourable
CN and a lower local symmetry. Our simulation results show that
Fe-centred FI clusters in the Fe80 W20 MG also exhibit a similar
trend during deformation at different temperatures compared
with Cu64 Zr36 MG (see Fig. S3(d) in the Supplementary Materials,
Appendix A), indicating that such correlation is, in fact, a general
behaviour at least, in this family of MGs. It is noteworthy that the
average degree of local five-fold symmetry has been suggested as a
universal structural indicator to explain the underlying dynamics
of metallic glasses [64]. Among the structural motifs, FI has the
highest degree of five-fold symmetry, and numerous experimental
and computer simulations support that FI clusters play a unique
role in the dynamics and mechanical properties of a wide range of
pure metallic liquids, as well as multicomponent MGs especially
composed of transition metals [62,6470]. On the other hand,
while FI may not be the dominant motif in some glasses such as
those containing metalloids, similar to the current work, still other
structural motifs with a high degree of five-fold symmetry (e.g. the
Voronoi index of 0, 2, 8, 0 in NiP alloys [47]) could be selected
as their structural indicator.
Our above structural analyses reveal the differences in the
FI decay (or the destruction of FI clusters) during deformation
at different temperatures. To shed light on the deformation
mechanisms at the atomistic level, we monitor the evolution
of spatial distribution of the destructed FI clusters at different
temperatures. To this end, we adopt the Egamis theory of local
topological instability [71,72] as a criterion for the destruction
of the FI clusters. According to this theory, there is a critical
shear strain, c , beyond which the local atomic structure becomes
unstable and would like to change its topology [72,73]. If the

))2 x(x+2)
(x+1 x(x+2
,
4 (2 3)

(1)
and x is the radius ratio, i.e.

x = RA /RB . For the Cu-centred FI (Cu0 0 12 0), the effective

radius ratio is x 0.9 [74]. Hence, Eq. (1) gives the value of
c-FI 0.12. Therefore, if the local atomic shear strain reaches this
threshold, the FI clusters will distort and undergo bond breaking,
that is, a plastic deformation. Fig. 7(a), (b) present the snapshots of
the Cu64 Zr36 MG at 50 K (0.06Tg ) and 600 K (0.77Tg ), respectively,
at different stages of deformation. In these snapshots, only those
central atoms in FI clusters whose shear strain have exceeded the
threshold, 0.12, are presented. Our structural analysis confirms
that those FI clusters have indeed been destroyed, and as a result,
they have been converted to other unfavourable (fragmented)
clusters. At the low temperature of 50 K and the applied strain of
= 7% (see Fig. 7(a)), the density of the localized plastic zones
in which the FI clusters are destroyed is very low, and they are
mostly isolated from each other. Since the FI is the most stable
SRO cluster against shearing [75], their destructions may cause the
local softening, leading to the formation of new plastic zones or
growth of the existing ones. As can be seen in Fig. 7(a), these plastic
zones occasionally percolate along planes parallel to the principal
shear direction, forming shear planes, which can be considered as
the precursors for the formation of SBs. Further development of
these shear planes through their propagating along the shear plane
direction and thickening along the lateral direction leads to the
formation of shear bands, causing catastrophic failures.
Fig. 7(b) also shows that the distribution of destroyed FI clusters
at the elevated temperature of 600 K (0.77Tg ) is much more
dispersed than that at 50 K (0.06Tg ) at the same applied strain.
This is due to the thermally activated nature of bond-breaking
and destruction of atomic clusters. Consequently, the plastic zones,
which are distributed more uniformly at elevated temperatures,
readily interact and interlink, forming a widely spread network
of shear planes (see Fig. 7(b) at = 8% and = 9%). Clearly,
it is this network of shear planes which uniformly decorates the
whole sample that leads to a homogeneous plastic flow (also see
Supplementary Fig. S4 for the Fe80 W20 MG, Appendix A).
To explain what causes the recovery of the FI clusters, we select
two small cubic boxes with the dimensions of 5 5 5 nm3 located
inside and outside of the SB (see Point 1 and Point 2 in Fig. 7(a),
respectively), and trace the variation of FI concentration within
these boxes as a function of applied strain at 50 K (0.06Tg ). The
resulted FI evolution curves for these two regions are shown in
Fig. 7(c). For comparison, the same curve obtained for the whole
sample is also presented. Fig. 7(c) shows that the concentration
of FI inside the SB region drops abruptly from 20% (at the initial
state) to 12% upon the formation of the SB at 9%. After the
SB formation is completed, the concentration remains unchanged.
We note that such a low concentration of FI inside the SB is
similar to that in the super-cooled liquid of this alloy above its
Tg 800 K, which is in agreement with previous studies [75,76].
However, outside the SB region, the concentration of FI exhibits
a minimum at 9%, after which, interestingly, a marked
recovery is observed. Remarkably, this recovery is corresponding
to the formation of the dominant SB. This implies that, at low
temperatures, once the dominant SB forms, the matrix outside the
SB region unloads. Consequently, the distorted SRO (e.g. FI) clusters
in the isolated shear zones, which are embedded in the elastic
matrix, are pushed back partially to their original topologies. This
leads to the recovery in the concentration of the FI clusters in
the whole sample as discussed in Fig. 6. Since the distribution of

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

mechanisms of MGs might be useful for controlling the failure

behaviour and developing precise fabrication techniques for MG
materials. Here, it is interesting to note that a direct observation
of local structural order (SROs) in MGs has been recently realized
experimentally [77]. Hence, research like the current work would
be insightful and inspiring for experimentalists to further develop
imaging techniques and evaluate theoretical predictions.
4. Summary
We performed MD simulations to investigate the effect of
temperature on the deformation mechanisms and failure modes of
two typical MGs. By carefully constructing the atomistic models,
we reproduced the experimental observations of failure modes
for a wide range of temperature, that is, a failure through
a single shear banding at low temperatures, a semi-brittle
failure through multiple shear bandings initially and subsequent
dominant shear banding at intermediate temperatures, and a
homogeneous plastic flow at elevated temperatures. Our structural
analyses clearly demonstrated the strong correlation between
the evolution of SRO clusters, especially the FI clusters, and
the deformation modes. More specifically, at low temperatures
under which the failure is through the single shear-banding,
the FI concentration first deceases to a minimum (after the UTS
point), then recovers markedly, and finally comes to a plateau.
At intermediate temperatures the failure is through the multiple
shear bandings first and then subsequent dominant shear banding,
the FI concentration first decreases and then reaches a plateau,
without a noticeable recovery in this regime. In contrast, at
elevated temperatures the deformation is through homogeneous
plastic flow, the FI concentration decreases monotonically. We
further showed that at elevated temperatures, the FI clusters are
destroyed more readily by forming a fine network of shear planes,
which are uniformly distributed within the MG samples. Since
our simulations show the same trend for both CuZr and FeW
MG systems, it is expected that our findings may be generic and
applicable to many other MG systems.
Acknowledgements
Fig. 7. Evolution of destroyed FI clusters during the deformation at temperatures
of (a) 50 K (0.06Tg ) and (b) 600 K (0.77Tg ). (c) Evolution of FI clusters as a function
of applied strain during deformation at 50 K. The curves show the evolution of FI
clusters from the initial state (zero strain) at Point 1 in (a), i.e. inside the SB, and
Point 2 in (a), i.e. outside the SB, as well as the overall variation within the whole
sample (All Sample curve).

destroyed FI clusters at 50 K (0.06Tg ) is much more concentrated

than that of 480 K (0.6Tg ) at a given applied strain (e.g., = 9%), to
accommodate the same amount of applied strain, the deformation
intensity of the plastic zones at 50 K should be much higher
than that at elevated temperatures. This implies that at 50 K, the
deformation contrast between the plastic zone and its surrounding
matrix (e.g. at the SB boundaries) is much sharper, which is in
agreement with our observation that sharper SB boundaries occur
at lower temperatures (See Fig. 4).
Due to some of their superior mechanical and chemical properties, MGs are promising functional and structural materials for
developing MEMS/NEMS [12,14]. However, how to prevent their
catastrophic failure while still maintaining their superior properties is still an active research topic. In addition, a precisely
controlled fabrication of MG nanostructures is still a daunting task.
A promising fabrication approach is to use their very good formability at elevated temperatures [14]. However, how to achieve
desired mechanical behaviour after the forming procedure via
controlled cooling remains a challenging issue. The new insights
revealed here into the temperature effects on the deformation

The authors gratefully acknowledge the financial support

from the Agency for Science, Technology and Research (A*STAR),
Singapore and the use of computing resources at the A*STAR
Computational Resource Centre, Singapore. M.J.Z. thanks to Dr.
Zachary Aitken for his constructive discussions and comments.
Appendix A. Supplementary material
The Supplementary Material includes five figures and their
corresponding explanations. Supplementary material related to
this article can be found online at http://dx.doi.org/10.1016/j.eml.
2016.07.009.
References
[1] M. Ashby, A. Greer, Metallic glasses as structural materials, Scr. Mater. 54
(2006) 321326.
[2] A. Greer, E. Ma, Bulk metallic glasses: at the cutting edge of metals research,
MRS Bull. 32 (2007) 611619.
[3] Y. Cheng, E. Ma, Atomic-level structure and structureproperty relationship in
metallic glasses, Prog. Mater. Sci. 56 (2011) 379473.
[4] S. Takeuchi, K. Edagawa, Atomistic simulation and modeling of localized shear
deformation in metallic glasses, Prog. Mater. Sci. 56 (2011) 785816.
[5] C.A. Schuh, T.C. Hufnagel, U. Ramamurty, Mechanical behavior of amorphous
alloys, Acta Mater. 55 (2007) 40674109.
[6] C. Suryanarayana, A. Inoue, Bulk Metallic Glasses, 2011.
[7] S. Song, J. Jang, J. Huang, T. Nieh, Inhomogeneous to homogeneous transition in
an Au-based metallic glass and its deformation maps, Intermetallics 18 (2010)
702709.

M. Jafary-Zadeh et al. / Extreme Mechanics Letters (

[8] Z. Wang, B. Sun, H. Bai, W. Wang, Evolution of hidden localized flow during
glass-to-liquid transition in metallic glass, Nature Commun. 5 (2014) 5823.
[9] T. Nieh, J. Wadsworth, C. Liu, G. Ice, Plasticity in the Supercooled Liquid Region
of Bulk Metallic Glasses, Lawrence Livermore National Laboratory (LLNL),
Livermore, CA, 2000.
[10] A. Argon, Plastic deformation in metallic glasses, Acta Metall. 27 (1979) 4758.
[11] A. Greer, Y. Cheng, E. Ma, Shear bands in metallic glasses, Mater. Sci. Eng. RRep. 74 (2013) 71132.
[12] Y. Liu, J. Liu, S. Sohn, Y. Li, J.J. Cha, J. Schroers, Metallic glass nanostructures of
tunable shape and composition, Nature Commun. 6 (2015) 7043.
[13] C. Pan, T. Wu, Y. Chang, J. Huang, Experiment and simulation of hot embossing
of a bulk metallic glass with low pressure and temperature, J. Micromech.
Microeng. 18 (2008) 025010.
[14] P. Sharma, N. Kaushik, H. Kimura, Y. Saotome, A. Inoue, Nano-fabrication
with metallic glassan exotic material for nano-electromechanical systems,
Nanotechnology 18 (2007) 035302.
[15] D.B. Miracle, A structural model for metallic glasses, Nature Mater. 3 (2004)
697702.
[16] H. Sheng, W. Luo, F. Alamgir, J. Bai, E. Ma, Atomic packing and short-tomedium-range order in metallic glasses, Nature 439 (2006) 419425.
[17] A.L. Greer, Metallic glasses. . . on the threshold, Mater. Today 12 (2009) 1422.
[18] H.-B. Yu, W.-H. Wang, K. Samwer, The relaxation in metallic glasses: an
overview, Mater. Today 16 (2013) 183191.
[19] S. Adibi, P.S. Branicio, Y.-W. Zhang, S.P. Joshi, Composition and grain size effects
on the structural and mechanical properties of CuZr nanoglasses, J. Appl. Phys.
116 (2014) 043522.
[20] Z. Sha, L. He, S. Xu, Q. Pei, Z. Liu, Y. Zhang, T. Wang, Effect of aspect ratio on the
mechanical properties of metallic glasses, Scr. Mater. 93 (2014) 3639.
[21] P. Murali, T.F. Guo, Y.W. Zhang, R. Narasimhan, Y. Li, H.J. Gao, Atomic scale
fluctuations govern brittle fracture and cavitation behavior in metallic glasses,
Phys. Rev. Lett. 107 (2011) 215501.
[22] X.W. Gu, M. Jafary-Zadeh, D.Z. Chen, Z. Wu, Y.-W. Zhang, D.J. Srolovitz, J.R.
Greer, Mechanisms of failure in nanoscale metallic glass, Nano Lett. 14 (2014)
58585864.
[23] Z. Sha, Q. Pei, Z. Liu, Y. Zhang, T. Wang, Necking and notch strengthening
in metallic glass with symmetric sharp-and-deep notches, Sci. Rep. 5 (2015)
10797.
[24] Z. Sha, L. He, Q. Pei, H. Pan, Z. Liu, Y. Zhang, T. Wang, On the notch sensitivity
of CuZr nanoglass, J. Appl. Phys. 115 (2014) 163507.
[25] Z. Sha, L. He, Q. Pei, Z. Liu, Y. Zhang, T. Wang, The mechanical properties of a
nanoglass/metallic glass/nanoglass sandwich structure, Scr. Mater. 83 (2014)
3740.
[26] S. Adibi, P.S. Branicio, R. Liontas, D.Z. Chen, J.R. Greer, D.J. Srolovitz, S.P.
Joshi, Surface roughness imparts tensile ductility to nanoscale metallic glasses,
Extreme Mech. Lett. 5 (2015) 8895.
[27] P. Guan, M. Chen, T. Egami, Stress-temperature scaling for steady-state flow in
metallic glasses, Phys. Rev. Lett. 104 (2010) 205701.
K.
[28] G. Duan, M.L. Lind, M.D. Demetriou, W.L. Johnson, W.A. Goddard III, T. agin,
Samwer, Strong configurational dependence of elastic properties for a binary
model metallic glass, Appl. Phys. Lett. 89 (2006) 151901.
[29] M. Mendelev, R. Ott, M. Heggen, M. Feuerebacher, M. Kramer, D. Sordelet,
Deformation behavior of an amorphous Cu64.5 Zr35.5 alloy: A combined
computer simulation and experimental study, J. Appl. Phys. 104 (2008)
123532.
[30] X. Wang, H. Lou, J. Bednarcik, H. Franz, H. Sheng, Q. Cao, J. Jiang, Structural
evolution in bulk metallic glass under high-temperature tension, Appl. Phys.
Lett. 102 (2013) 051909.
[31] Y. Shi, M.L. Falk, Strain localization and percolation of stable structure in
amorphous solids, Phys. Rev. Lett. 95 (2005) 095502.
[32] Y. Shi, M.L. Falk, Stress-induced structural transformation and shear banding
during simulated nanoindentation of a metallic glass, Acta Mater. 55 (2007)
43174324.
[33] Y. Shi, M.L. Falk, Atomic-scale simulations of strain localization in threedimensional model amorphous solids, Phys. Rev. B 73 (2006) 214201.
[34] Y. Shi, M.L. Falk, Simulations of nanoindentation in a thin amorphous metal
film, Thin Solid Films 515 (2007) 31793182.
[35] J. Ding, Y.-Q. Cheng, E. Ma, Full icosahedra dominate local order in Cu64 Zr34
metallic glass and supercooled liquid, Acta Mater. 69 (2014) 343354.
[36] J. Ding, S. Patinet, M.L. Falk, Y. Cheng, E. Ma, Soft spots and their structural
signature in a metallic glass, Proc. Natl. Acad. Sci. 111 (2014) 1405214056.
[37] Y. Cheng, A. Cao, E. Ma, Correlation between the elastic modulus and the
intrinsic plastic behavior of metallic glasses: The roles of atomic configuration
and alloy composition, Acta Mater. 57 (2009) 32533267.
[38] Y. Zhang, F. Zhang, C. Wang, M. Mendelev, M. Kramer, K. Ho, Cooling rates
dependence of medium-range order development in Cu64.5 Zr35.5 metallic
glass, Phys. Rev. B 91 (2015) 064105.
[39] D. Ma, A.D. Stoica, X.-L. Wang, Power-law scaling and fractal nature of
medium-range order in metallic glasses, Nature Mater. 8 (2009) 3034.
[40] N. Mattern, P. Jvri, I. Kaban, S. Gruner, A. Elsner, V. Kokotin, H. Franz, B.
Beuneu, J. Eckert, Short-range order of CuZr metallic glasses, J. Alloys Compd.
485 (2009) 163169.
[41] W. Brostow, J.-P. Dussault, B.L. Fox, Construction of Voronoi polyhedra,
J. Comput. Phys. 29 (1978) 8192.
[42] M. Tanemura, T. Ogawa, N. Ogita, A new algorithm for three-dimensional
Voronoi tessellation, J. Comput. Phys. 51 (1983) 191207.
[43] S. Plimpton, Fast parallel algorithms for short-range molecular dynamics,
J. Comput. Phys. 117 (1995) 119.

[44] M. Mendelev, M. Kramer, R. Ott, D. Sordelet, D. Yagodin, P. Popel, Development

of suitable interatomic potentials for simulation of liquid and amorphous
CuZr alloys, Phil. Mag. 89 (2009) 967987.
[45] X. Zhou, H. Wadley, R.A. Johnson, D. Larson, N. Tabat, A. Cerezo, A. PetfordLong, G. Smith, P. Clifton, R. Martens, Atomic scale structure of sputtered metal
multilayers, Acta Mater. 49 (2001) 40054015.
[46] G. Duan, D. Xu, Q. Zhang, G. Zhang, T. Cagin, W.L. Johnson, W.A. Goddard,
Molecular dynamics study of the binary Cu46-Z54 metallic glass motivated
by experiments: Glass formation and atomic-level structure, Phys. Rev. B 71
(2005) 224208.
[47] H. Sheng, E. Ma, M.J. Kramer, Relating dynamic properties to atomic structure
in metallic glasses, JOM 64 (2012) 856881.
[48] Q.-K. Li, M. Li, Assessing the critical sizes for shear band formation in metallic
glasses from molecular dynamics simulation, Appl. Phys. Lett. 91 (2007)
231905.
[49] Y. Shi, Size-independent shear band formation in amorphous nanowires made
from simulated casting, Appl. Phys. Lett. 96 (2010) 121909.
[50] H.C. Andersen, RATTLE: A Velocity version of the SHAKE algorithm for
molecular dynamics calculations, J. Comput. Phys. 52 (1983) 2434.
[51] D.J. Evans, B.L. Holian, The nosehoover thermostat, J. Chem. Phys. 83 (1985)
40694074.
[52] P.S. Branicio, D.J. Srolovitz, Local stress calculation in simulations of
multicomponent systems, J. Comput. Phys. 228 (2009) 84678479.
[53] D. Tsai, The virial theorem and stress calculation in molecular dynamics,
J. Chem. Phys. 70 (1979) 13751382.
[54] F. Shimizu, S. Ogata, J. Li, Theory of shear banding in metallic glasses and
molecular dynamics calculations, Mater. Trans. 48 (2007) 29232927.
[55] C. Rycroft, Voro++: A Three-Dimensional Voronoi Cell Library in C++, Lawrence
Berkeley National Laboratory, 2009.
[56] A. Stukowski, Visualization and analysis of atomistic simulation data with
OVITOthe open visualization tool, Modelling Simul. Mater. Sci. Eng. 18 (2010)
015012.
[57] R. Maa, D. Klaumnzer, E. Prei, P. Derlet, J. Lffler, Single shear-band
plasticity in a bulk metallic glass at cryogenic temperatures, Scr. Mater. 66
(2012) 231234.
[58] S.-W. Lee, M. Jafary-Zadeh, D.Z. Chen, Y.-W. Zhang, J.R. Greer, Size effect
suppresses brittle failure in hollow Cu60Zr40 metallic glass nanolattices
deformed at cryogenic temperatures, Nano Lett. 15 (9) (2015) 56735681.
[59] H. Yu, P. Yu, W. Wang, H. Bai, Thulium-based bulk metallic glass, Appl. Phys.
Lett. 92 (2008) 141906.
[60] J. Lu, G. Ravichandran, W.L. Johnson, Deformation behavior of the Zr 41.2 Ti
13.8 Cu 12.5 Ni 10 Be 22.5 bulk metallic glass over a wide range of strain-rates
and temperatures, Acta Mater. 51 (2003) 34293443.
[61] G.E. Martin, Transformation Geometry: An Introduction to Symmetry, Springer
Science & Business Media, 2012.
[62] X. kui Xi, L. long Li, B. Zhang, W. hua Wang, Y. Wu, Correlation of atomic cluster
symmetry and glass-forming ability of metallic glass, Phys. Rev. Lett. 99 (2007)
095501.
[63] Y. Cheng, A.J. Cao, H. Sheng, E. Ma, Local order influences initiation of plastic
flow in metallic glass: Effects of alloy composition and sample cooling history,
Acta Mater. 56 (2008) 52635275.
[64] Y. Hu, F. Li, M. Li, H. Bai, W. Wang, Five-fold symmetry as indicator of dynamic
arrest in metallic glass-forming liquids, Nature Commun. 6 (2015).
[65] S. Hao, C.-Z. Wang, M. Li, R.E. Napolitano, K.-M. Ho, Dynamic arrest and glass
formation induced by self-aggregation of icosahedral clusters in Zr1x Cux
alloys, Phys. Rev. B 84 (2011) 064203.
[66] K. Kelton, G. Lee, A.K. Gangopadhyay, R. Hyers, T. Rathz, J. Rogers, M. Robinson,
D. Robinson, First X-ray scattering studies on electrostatically levitated
metallic liquids: demonstrated influence of local icosahedral order on the
nucleation barrier, Phys. Rev. Lett. 90 (2003) 195504.
[67] N. Jakse, A. Pasturel, Ab initio molecular dynamics simulations of local
structure of supercooled Ni, J. Chem. Phys. 120 (2004) 61246127.
[68] P. Ganesh, M. Widom, Ab initio simulations of geometrical frustration in
supercooled liquid Fe and Fe-based metallic glass, Phys. Rev. B 77 (2008)
014205.
[69] Y. Shen, T. Kim, A. Gangopadhyay, K. Kelton, Icosahedral order, frustration,
and the glass transition: Evidence from time-dependent nucleation and
supercooled liquid structure studies, Phys. Rev. Lett. 102 (2009) 057801.
[70] J. Saida, M. Matsushita, T. Zhang, A. Inoue, M. Chen, T. Sakurai, Precipitation of
icosahedral phase from a supercooled liquid region in Zr65 Cu7.5 Al7.5 Ni10 Ag10
metallic glass, Appl. Phys. Lett. 75 (1999) 34973499.
[71] T. Egami, Nano-glass mechanism of bulk metallic glass formation, Materials
transactions 43 (2002) 510517.
[72] T. Egami, Atomic level stresses, Prog. Mater. Sci. 56 (2011) 637653.
[73] J. Park, Y. Shibutani, S. Ogata, M. Wakeda, Effects of atomic deviatoric distortion
on local glass transition of metallic glasses, Mater. Trans. 46 (2005) 28482855.
[74] L. Ward, D. Miracle, W. Windl, O.N. Senkov, K. Flores, Structural evolution and
kinetics in Cu-Zr metallic liquids from molecular dynamics simulations, Phys.
Rev. B 88 (2013) 134205.
[75] A. Cao, Y. Cheng, E. Ma, Structural processes that initiate shear localization in
metallic glass, Acta Mater. 57 (2009) 51465155.
[76] Y. Tong, T. Iwashita, W. Dmowski, H. Bei, Y. Yokoyama, T. Egami, Structural
rejuvenation in bulk metallic glasses, Acta Mater. 86 (2015) 240246.
[77] A. Hirata, P. Guan, T. Fujita, Y. Hirotsu, A. Inoue, A.R. Yavari, T. Sakurai, M. Chen,
Direct observation of local atomic order in a metallic glass, Nature Mater. 10
(2011) 2833.