Fe65.5Cr4Mo4Ga4P12 C5B5.5 BMGs: Sample preparation, thermal stability and mechanical properties

Article abstract of DOI:10.1016/j.jallcom.2006.08.188

Bulk amorphous Fe-based alloys with the nominal composition Fe_65.5Cr₄Mo₄Ga₄ P₁₂C₅B_5.5 have been obtained by copper mold casting in different shapes: cylindrical rods with diameters up to 3 mm, rectangular bars of 2 mm × 2 mm and discs of 10 mm diameter and 1 mm thickness. These alloys exhibit good soft magnetic properties, characterized by low coercivity and high saturation magnetization. Besides the magnetic properties, the Fe_65.5Cr₄Mo₄Ga₄ P₁₂C₅B_5.5 bulk metallic glass (BMG) shows a high glass transition temperature T_g, as well as a high crystallization temperature T_x, with an extension of the supercooled liquid region of around 65 K. The mechanical behavior was investigated by compression and Vickers hardness tests. The fracture strength for the as-cast samples σ_f is 2.8 GPa and the fracture strain ε_f is 1.9%. Upon annealing at 715 K for 10 min, i.e. at a temperature below the calorimetric glass transition, the fracture strain drops to 1.6% and no plastic deformation is observed. The Vickers hardness H_V for the as-cast samples is about 885, and increases to 902 upon annealing. The fracture behavior of these Fe-based bulk glassy alloys is significantly different in comparison with the well-studied Zr-, Cu- or Ti-based good glass-formers. The fracture is not propagating along a well-defined direction and the fractured surface looks irregular. Instead of veins, the glassy alloy develops a high number of microcracks.

Full text of DOI:10.1016/j.jallcom.2006.08.188

Journal of Alloys and Compounds 434–435 (2007) 171–175
Fe65.5Cr4Mo4Ga4P12C5B5.5 BMGs: Sample preparation,
thermal stability and mechanical properties
M. Stoicaa,b,∗, J. Eckertc, S. Rotha, A.R. Yavarib, L. Schultza
a IFW Dresden, Institute for Metallic Materials, P.O. Box 270016, D-01171 Dresden, Germany
b LTPCM-CNRS UA29, ENSEEG, Institut National Politechnique de Grenoble, Domain Universitaire,
1130 Rue de la Piscine, BP 75 Saint Martin d’He`res Campus, 38402 Grenoble, France
c TU Darmstadt, Department of Materials and Geo Sciences, Physical Metallurgy Division,
Petersenstraße 23, D-64287 Darmstadt, Germany
Available online 6 October 2006
Abstract
Bulk amorphous Fe-based alloys with the nominal composition Fe65.5Cr4Mo4Ga4P12C5B5.5 have been obtained by copper mold casting in
different shapes: cylindrical rods with diameters up to 3 mm, rectangular bars of 2 mm × 2 mm and discs of 10 mm diameter and 1 mm thickness.
These alloys exhibit good soft magnetic properties, characterized by low coercivity and high saturation magnetization. Besides the magnetic
properties, the Fe65.5Cr4Mo4Ga4P12C5B5.5 bulk metallic glass (BMG) shows a high glass transition temperature Tg, as well as a high crystallization
temperature Tx, with an extension of the supercooled liquid region of around 65 K. The mechanical behavior was investigated by compression and
Vickers hardness tests. The fracture strength for the as-cast samples σf is 2.8 GPa and the fracture strain εf is 1.9%. Upon annealing at 715 K for
10 min, i.e. at a temperature below the calorimetric glass transition, the fracture strain drops to 1.6% and no plastic deformation is observed. The
Vickers hardness HV for the as-cast samples is about 885, and increases to 902 upon annealing. The fracture behavior of these Fe-based bulk glassy
alloys is significantly different in comparison with the well-studied Zr-, Cu- or Ti-based good glass-formers. The fracture is not propagating along
a well-defined direction and the fractured surface looks irregular. Instead of veins, the glassy alloy develops a high number of microcracks.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Amorphous materials; Thermal analysis
1. Introduction
The critical cooling rate of about 100 K/s necessary for glass
formation of magnetic BMGs is higher than the value of about
1–10 K/s characteristic for non-magnetic alloys with very good
glass-forming ability [3,4]. Thus, the maximum achievable
diameter of these Fe-based alloys is limited to only a few mil-
limeters [5], in contrast to Zr- or Cu-based glassy alloys, which
can easily reach 10 mm diameter [6]. Only very recently, Pon-
nambalam et al. [7] and, independently, Lu et al. [8] succeeded to
cast Fe-based BMGs with a thickness larger than one centimeter
(using some small additions of Y, Ln and/or Er), but their glasses
are paramagnetic at room temperature, with a Curie temperature
of around 55 K. The difficulty to cast such ferrous glasses in
bulk form is one of the reasons for which, despite an increasing
number of published papers on the mechanical properties of
BMGs, most of these reports deal only with non-ferrous Zr-, Ti-
or Cu- based alloys [6,9,10]. The mechanical properties of bulk
Fe-based glasses started to be investigated only in the past few
years [7,8,11–13]. The known strength values for non-ferrous
glassy alloys are in the range of: 1.5–1.8 GPa for Zr-based alloys,
1.7–1.9 GPa for Ti-based alloys, 1.9–2.5 GPa for Cu-based
Because of the absence of crystalline anisotropy, Fe65.5Cr4-
Mo4Ga4P12C5B5.5 amorphousalloysexhibitgood softmagnetic
properties, characterized by a low coercive force and a high per-
meability. Nevertheless, residual anisotropies may be present,
such as shape anisotropy or stress-induced anisotropies, caused
by internal mechanical stresses induced during the preparation
procedure [1,2]. Using the copper mold casting method, such
alloys can be cast in form of bulk specimens directly suitable for
use as magnetic sensors, magnetic valves or magnetic clutches.
But for possible use in magnetic devices, the Fe-based BMGs
also require good mechanical strength besides the good soft
magnetic properties.
∗
Corresponding author. Present address: Leibniz Institute for Solid State and
Materials Research (IFW) Dresden, PO Box 270016, D-01171 Dresden,
Germany. Tel.: +49 351 4659 644; fax: +49 351 4659 452.
E-mail addresses: m.stoica@ifw-dresden.de, mstoicaro@yahoo.com
(M. Stoica).
0925-8388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2006.08.188

172
M. Stoica et al. / Journal of Alloys and Compounds 434–435 (2007) 171–175
Table 1
alloys or 2.7–3.1 GPa for Ni-based alloys (for an overview, see
[12]). For our Fe65.5Cr4Mo4Ga4P12C5B5.5 BMG, the fracture
strength exceeds 2.8 GPa. In the case of Zr- or Cu- based bulk
glassy alloys, the fracture usually proceeds along a shear plane,
which is declined by ∼45◦ to the direction of the applied load,
and the fracture surface exhibits a well-developed vein pattern
[6,10]. In contrast, the studied Fe65.5Cr4Mo4Ga4P12C5B5.5
samples develop a cleavage-like fracture surface with a high
number of microcracks, which finally destroy the sample com-
pletely. The fracture surface appears to contain a high number
of small fracture zones, which are probably generated at the
same time due to the very high stress level upon deformation.
Glass transition temperatures Tg, crystallization temperatures Tx and liquidus
temperatures Tliq, as well as reduced glass transition temperatures Trg = Tg/Tliq
γ parameter, extension of the supercooled liquid region ꢀTx and crystallization
enthalpy ꢀHx for as-cast Fe66.5Cr4Mo4Ga4P12C5B5.5 glassy alloys measured
at 40 K/min heating rate
,
Sample
Rod (Ø
1.5 mm)
Rod (Ø
2 mm)
Rod (Ø
2.5 mm
Rod (Ø
3 mm)
Tg (K)
Tx (K)
Tliq (K)
Tg/Tliq
γ
746
812
1344
0.55
0.388
66
749
814
1347
0.55
0.388
65
750
814
1346
0.55
0.388
64
751
815
1346
0.55
0.388
64
ꢀTx (K)
ꢀHx (J/g)
69.6
67.8
65.1
56.1
2. Experimental
were studied by scanning electron microscopy (SEM), using a JEOL JSM 6400
microscope operated at 25 kV.
The preparation of the amorphous Fe65.5Cr4Mo4Ga4P12C5B5.5 BMGs was
done in several steps. First, master alloy ingots were obtained by induction
melting using Fe–B, Fe–C, Fe–Ga, Fe–P pre-alloys and pure elements as Mo
(99.4% purity), Cr (99.95% purity), Fe (99.9% purity) and crystalline B (99.99%
purity). Induction melting of Fe with B, Fe with C (99.9% purity) and Fe with
Ga (99.7% purity) allowed to produce the respective pre-alloys. The FeP pre-
alloy was obtained by induction melting of consolidated powder resulted upon
milling Fe powder (99.9% purity, less than 10 ␮m particle size) with amorphous
red P powder (99% purity, less than 100 ␮m particle size)
3. Results and discussion
The glass-forming ability (GFA) can be evaluated from ther-
mal stability measurements. Usually, high values of the exten-
sion of the supercooled liquid region ꢀTx and the reduced glass
transition temperatures Trg = Tg/Tliq indicate a good GFA [9].
The values for Tg, Tx and Tliq as a function of geometrical dimen-
sions of the as-cast rods are given in Table 1. Using these values,
ꢀTx and Trg were calculated. Recently, Lu and Liu [18,19] pro-
posed a new parameter γ defined as Tx/(Tg + Tliq), to predict
GFA for various glass-forming systems. Usually, the γ values
of BMGs range between 0.35 and 0.50 [18].
Amorphous rods with diameters of 1.5, 2, 2.5 and 3 mm, length of 70 mm,
rectangular bars of 2 mm × 2 mm with length of 30 mm as well as discs of 10 mm
diameter and 1 mm thickness were prepared from the master alloy with nom-
inal composition Fe65.5Cr4Mo4Ga4P12C5B5.5. The samples were obtained by
induction melting under argon atmosphere at a pressure of 80 KPa and sub-
sequent injection into a copper mold under an applied pressure of 3 × 105 Pa.
Because the presence of oxides can have a negative influence in order to prepare
bulk amorphous alloys, we checked the oxygen content of the pre-alloys and of
the master alloy. This revealed very low values: 180 ppm for Fe-P, 130 ppm for
the other pre-alloys and 50 ppm for Fe65.5Cr4Mo4Ga4P12C5B5.5, respectively.
Thethermalstability, i.e. theglasstransition, theextensionofthesupercooled
liquid region and the crystallization, was examined by differential scanning
calorimetry (DSC), using a Netzsch DSC 404 under argon flow. The glass tran-
sition temperature Tg and the crystallization temperature Tx were measured
as the onset temperatures of the glass transition and the crystallization events,
respectively, during heating with a constant rate of 40 K/s. The extension of
the supercooled liquid region, defined as the difference between the glass tran-
sition temperature and the crystallization temperature, ꢀTx = Tx–Tg was also
calculated. Additionally, the melting temperature Tm defined by the liquidus
temperature Tliq at the onset of melting upon heating with the same constant rate
of 40 K/s was measured. The amorphous structure as well as the crystallization
behavior and the magnetic properties of these glassy samples have already been
investigated and published previously [14–17].
In order to investigate the mechanical behavior, different techniques were
used. First, room temperature compression tests using an electromechanical
Instron 8562 testing device were performed for as-cast rods of 2 and 2.5 mm
diameter, respectively, as well as for as-cast and annealed 2 mm × 2 mm rect-
angular bars. The length of the samples was between 4 and 5 mm and the
machine was operated in the constant position rate mode, with a displacement
of 10−3 mm/s. The corresponding strain rate was evaluated as 10−4 s−1. From
the compression tests, the fracture strength σf, the fracture strain εf, the yield
strength σy, the yield strain εy and the Young’s modulus E were derived.
The Vickers hardness was measured for the same kind of bar samples using a
computercontrolledStruersDuramin5hardnesstester. Thetestswereperformed
using a typical diamond indenter in the form of pyramid with square base and
an angle of 136◦ between opposite faces, applying a load of 1.96 N for 10 s. The
diagonal of the imprints as well as the hardness were calculated using a Digital
VideoMeasuringSystem. Forindentations, thesampleswereembeddedinepoxy
resin and the measured surface was carefully polished with a paste containing
diamond particles with a diameter smaller than 0.25 ␮m. The characteristics of
the fractured surface as well as the features of the indents after the hardness tests
The glass transition temperature and the crystallization tem-
perature slowly increase with increasing rod diameter. The
differencesinthermalstabilitybetweenrodswithdifferentdiam-
eters are caused by a different degree of relaxation as a result of
the different cooling rates reached during solidification. Another
reason can be a slightly variation in actual composition of the
glasses. Such compositional variations may arise if at least the
rods with larger diameters are not fully amorphous but contain
some (nano)crystalline phase(s), which may form (i)- due to a
possible appearance of crystalline nuclei in the liquid state, or
(ii)- due to an insufficient cooling rate for complete glass forma-
tion upon casting. Some crystalline inclusions can form from the
molten state and, upon casting, the already formed crystalline
clusters may or may not act as seeds for further nucleation. Thus,
thecompositionoftheremainingmatrixisslightlydifferentfrom
the starting overall composition and this can modify the thermal
stability data. Concerning the second hypothesis, the cooling
rate decreases with increasing diameter of the cast rods and thus
the maximum achievable diameter for which the sample is still
amorphous is limited. Hence, the rod most susceptible to con-
tain some crystalline inclusions should be the rod with the largest
diameter. The presence of crystalline inclusions was observed
in the case of 3 mm diameter rod; in a previous work [14] we
have described in detail the formation of such crystalline clus-
ters by means of time-resolved X-ray diffraction in transmission
configuration using a high-energy high intensity monochromatic
synchrotron beam.

M. Stoica et al. / Journal of Alloys and Compounds 434–435 (2007) 171–175
173
Fig. 1. Compressive stress–strain curves for as-cast Fe65.5Cr4Mo4Ga4P12
C5B5.5 2 and 2.5 mm diameter rods.
-
Fig. 2. Compressive stress–strain curves for Fe65.5Cr4Mo4Ga4P12C5B5.5 as-
cast and annealed 2 mm × 2 mm rectangular bars.
A small decrease of the crystallization enthalpy ꢀHx with
increasing rod diameter can be observed. This can also be an
indication that the rods with larger diameters are susceptible
to contain some frozen-in crystalline nuclei. However, the dif-
ferences between the values are relatively small and within the
measurements errors.
The extension of supercooled liquid region of around 65 K
and the reduced glass transition temperature of around 0.55 is
characteristic for good glass forming systems. It is, however,
in the lower limit when compared with the ratio 0.57–0.72
measured in the (Pd,Ni,Cu)80P20 and (Pd,Ni,Fe)80P20 glass sys-
tems, which are two of the best glass forming systems known
[4,20,21].
Lu and Liu [18] linked the new dimensionless γ parame-
ter to the critical cooling rate Rc as well as to a critical cross
section Zc by studying the data available in literature for repre-
sentative non-ferrous BMGs. Using ␥ parameter equal to 0.388
(Table 1), the calculated critical cooling rate is 91 K/s and the
maximum achievable diameter 2.9 mm. According to the data
reviewed by Inoue [6], the Fe-based BMGs of the composition
Fe–Cr–Mo–Ga–P–C–B should achieve a thickness of a few mm
and the critical cooling rate is of the order of hundreds K/s. The
calculated values are somehow lower, but the errors are accept-
able.
Fig. 1 shows the compressive stress–strain curves for the
as-cast Fe65.5Cr4Mo4Ga4P12C5B5.5 cylindrical rods with 2 and
2.5 mm diameter, respectively. Both samples exhibit similar fea-
tures, i.e. an elastic deformation regime followed by a small
compressive plastic strain. The yield stress σy, measured at the
offset yield, is 3.19 GPa in the case of the 2 mm diameter sample
and 3.27 GPa for the 2.5 mm diameter sample. The correspond-
ing elastic strain εy is 2.03% for the 2 mm diameter rod and
1.91% for the 2.5 mm diameter rod, respectively. Young’s mod-
ulus is determined as 160 GPa for the 2 mm diameter sample and
170 GPa for the 2.5 mm diameter sample. The fracture of the
samples occurs at a fracture stress σf of 3.23 GPa for the 2 mm
diameter sample. σf reaches 3.27 GPa for the 2.5 mm diameter
sample. The fracture strain εf is 2.33 and 2.03%, respectively.
The pure compressive plastic strain is 0.30 and 0.12% for the two
samples, respectively. Hence, this Fe-based bulk glass exhibits
a very high strength, higher than that observed in the case
of non-ferrous bulk amorphous alloys [12], but close to that
measured for Fe–B–Si–Nb (3.25 GPa) and Fe–Ga–P–C–B–Si
systems (3.16 GPa) [11].
The differences between the two described samples regarding
their mechanical data are quite small and are within the limits of
error of the measurement. The error in the measurement of the
small cross-section of the sample is 5% and is comparable with
the difference between the two calculated values for Young’s
modulus (160 GPa for the 2 mm and 170 GPa for the 2.5 mm
diameter sample, respectively).
Fig. 2 shows the compressive stress–strain curves for as-
cast and annealed rectangular bars with the cross section
2 mm × 2 mm. The samples were annealed for 10 min at 712 K,
i.e., at a temperature equal to Tg—15 K, using a heating and
cooling rate of 5 K/min (for rectangular bars, Tg = 727 K, mea-
sured in DSC at 5 K/min heating rate). It can be noted that the
fracture stress retains almost the same value after annealing, but
the sample becomes more brittle and no plastic deformation is
observed. The yield stress is 2.82 and 2.84 GPa for the as-cast
and for the annealed sample, respectively. The corresponding
compressive elastic strains are 1.76 and 1.60%, respectively.
Young’s modulus is 161 GPa for the as-cast state and 177 GPa
after annealing. The fracture of both samples occurs at nearly the
same value of compressive stress, 2.84 GPa, but the correspond-
ing fracture strain is different: 1.91% for the as-cast sample and
1.63% for the annealed one. The as-cast sample shows a small
plastic deformation of 0.15%, but in the case of the annealed bar
the plastic regime extends only over 0.03%.
The appearance of the fracture surface was investi-
gated by SEM. Fig. 3(a and b) show micrographs of the
Fe65.5Cr4Mo4Ga4P12C5B5.5 as-cast rectangular bar after frac-
ture at different magnifications. The observed fracture behavior
is totally different from that observed for Zr- or Cu- based bulk
glassyalloys, wherethefractureusuallyproceedsalongthemax-
imum shear stress plane, which is declined by about 45◦ to the
direction of the applied load, and the fracture surface contains a
number of a well-developed vein patterns [6,10]. Fig. 3(a) shows
a side view of the fractured sample. There are no distinct veins
and dimples, at least not over the whole fracture surface but

174
M. Stoica et al. / Journal of Alloys and Compounds 434–435 (2007) 171–175
high value found for Tg, because is considered that Tg reflects
the bonding nature between the constituent elements [24].
The Vickers hardness was measured for the same kind of
as-cast and annealed rectangular bar samples with the cross sec-
tion 2 mm × 2 mm as used for compression tests. For the as-cast
bars, HV is 885 (8.68 GPa) with a typical standard deviation of
5. The hardness of the annealed bars increases up to HV = 902
(8.84 GPa) with a standard deviation of 2.1. The smaller value
of the standard deviation in the case of the annealed samples
indicates a more homogeneous behavior of the annealed spec-
imens in comparison with the as-cast state. Upon indentation,
some slips generated by the Vickers diamond indenter and the
absenceofanyvisiblecrackcanbenoticed(picturenotpresented
here). These features are in accordance with the elastic-plastic
behavior found from the compression test.
The ratio σf/E attains values of around 0.017 and HV/3E takes
values of 0.018 and 0.016 for the as-cast and annealed sam-
ples, respectively. Hence, the relationship between the fracture
strength and the hardness σf ≈ HV/3 is verified [2]. Consider-
ing that a number of amorphous alloys reported up to date have
nearly fixed values of about 0.02 for σf/E and HV/3E [2], it is
concluded that the studied Fe-based bulk glassy alloys follow
the same trend. The agreement also suggests that the Fe-based
bulk glassy alloys have an elastic-plastic deformation mode sim-
ilar to other non-ferrous bulk glassy alloys, the main difference
being a much higher strength level, which can be achieved in
compression tests.
4. Conclusions
Fig. 3. SEM micrograph (topological view) showing the fracture surface after
compression test: (a) side view and (b) detail at higher magnification.
The bulk amorphous Fe65.5Cr4Mo4Ga4P12C5B5.5 samples
with various shapes produced by copper mold casting exhibits a
good GFA and a high thermal stability. A high strength and some
compressive plastic strain was observed upon compression test
performed at room temperature. A structural relaxation treat-
ment at elevated temperature does not affect the stress level,
which remains almost the same, but the plasticity disappears.
The fracture mechanism is different in our Fe-based bulk glassy
alloy in comparison with other non-magnetic metallic glasses.
The actual failure mode of metallic glassy sample is a competi-
tion result between shear fracture and distensile fracture.
only in some regions. Recent systematic investigations on the
glasses with different alloy systems indicate that the shear frac-
ture always deviates from the maximum shear stress plane either
under compression or under tension [22]. For the samples stud-
ied here, the fracture surface appears to consist of a high number
of small fracture zones, which leads to breaking of the samples
into many small parts, as indicated in Fig. 3(b). It indicates that
the fracture of metallic glasses can occur either in a shear mode
or in a break mode, depending on the constituent elements and
microstructure in detail. The actual failure mode of a metallic
glassy sample is a competition result between shear fracture and
distensile fracture as reported by Zhang et al. [23]. For the Fe-
based glasses, it has quite high fracture strength, indicating that
the critical shear fracture stress must be high enough in compar-
ison to the bonding force among the constituent elements in the
present glass. The brittleness can be also enhanced by the pres-
ence of metalloids, i.e. C, B and P [2]. Inoue et al. [11] supposed
that such cracks as presented in Fig. 3(b) are relatively easy to be
initiated nearly simultaneously because the stress level is very
high, around 3 GPa, and the shock wave caused by initiation
of one crack induces the generation of other cracks at differ-
ent sites. The high strength of the Fe-based bulk glassy alloys
seems to originate from the high bonding forces between the
constituent elements. This is in agreement with the relatively
Acknowledgements
¨
The authors thank H. Klauß and Dr. A. Guth for their help
with compression tests and SEM investigations, and S. Kuszin-
¨
ski, H. Schulze and S. Muller-Litvanyi for technical assistance.
Partial financial support from the German Science Founda-
tion under grant number EC 111/12-1 and from the EU within
the framework of the RTN-networks on bulk metallic glasses
(HPRN-CT-2000-00033)andductileBMGcomposites(MRTN-
CT-2003-504692) is also acknowledged.
References
[1] A. Inoue, J.S. Gook, Mater. Trans. JIM 36 (1995) 1180.
[2] H.S. Chen, Rep. Prog. Phys. 43 (1980) 353.

M. Stoica et al. / Journal of Alloys and Compounds 434–435 (2007) 171–175
175
˚
[3] A. Inoue, A. Kato, T. Zhang, S.G. Kim, T. Masumoto, Mater. Trans. JIM
32 (1991) 609.
[14] M. Stoica, J. Eckert, S. Roth, L. Schultz, A.R. Yavari, A. Kvick, J.
Metastable Nanocryst. Mater. 12 (2002) 77.
[4] Y. He, T. Shen, R.B. Schwarz, Metall. Mater. Trans. A 29 (1998) 1795.
[5] A. Inoue, T. Zhang, A. Takeuchi, Appl. Phys. Lett. 71 (1997) 464.
[6] A. Inoue, Acta Mater. 48 (2000) 279.
[7] V. Ponnambalam, S.J. Poon, G.J. Shiflet, J. Mater. Res. 19 (2004) 1320.
[8] Z.P. Lu, C.T. Liu, J.R. Thomson, W.D. Porter, Phys. Rev. Lett. 92 (2004)
245503.
[15] M. Stoica, J. Degmova´, S. Roth, J. Eckert, H. Grahl, L. Schultz, A.R. Yavari,
˚
A. Kvick, G. Heunen, Mater. Trans. 43 (2002) 1966.
[16] M. Stoica, J. Eckert, S. Roth, L. Schultz, Mater. Sci. Eng. A 375–377 (2004)
399.
[17] M. Stoica, S. Roth, J. Eckert, L. Schultz, M.D. Baro´, J. Magn. Magn. Mater.
290–291 (2005) 1480.
[9] A. Inoue, Bulk Amorphous Alloys, Trans Tech Publications, Uetikon-
Zuerich Switzerland, 1999.
[10] A. Inoue, W. Zhang, T. Zhang, K. Kurosaka. Acta Mater. 49 (2001) 2645.
[11] A. Inoue, B.L. Shen, A.R. Yavari, A.L. Greer, J. Mater. Res. 18 (2003)
1487.
[12] A. Inoue, B.L. Shen, H. Koshiba, H. Kato, A.R. Yavari, Acta Mater. 52
(2004) 1631.
[13] A. Inoue, X.M. Wang, Acta Mater. 48 (2000) 1383.
[18] Z.P. Lu, C.T. Liu, Acta Mater. 50 (2002) 3501.
[19] Z.P. Lu, C.T. Liu, Phys. Rev. Lett. 91 (2003) 115505.
[20] R.B. Schwarz, Y. He, Mater. Sci. Forum 235–238 (1997) 231.
[21] T.D. Shen, Y. He, R.B. Schwarz, J. Mater. Res. 14 (1999) 2107.
[22] Z.F. Zhang, J. Eckert, L. Schultz, Acta Mater. 51 (2003) 1167.
[23] Z.F. Zhang, G. He, J. Eckert, L. Schultz, Phys. Rev. Lett. 91 (2003) 045505.
[24] F.R. De Boer, R. Boom, W.C.M. Mattens, A.R. Miedema, A.K. Niessen,
Cohesions in Metals, North-Holland, Amsterdam, The Netherlands, 1989.