[Gabe Newell]

Fe-based bulk metallic glasses with diameter thickness larger
than one centimeter

V. Ponnambalam and S. Joseph Poona
Department of Physics, University of Virginia, Charlottesville, Virginia 22904-4714

Gary J. Shiflet
Department of Materials Science and Engineering, University of Virginia, Charlottesville, Virginia 22904-4745

(Received 15 December 2003; accepted 17 February 2004)

Fe?Cr?Mo?(Y,Ln)?C?B bulk metallic glasses (Ln are lanthanides) with maximum
diameter thicknesses reaching 12 mm have been obtained by casting. The high glass
formability is attained despite a low reduced glass transition temperature of 0.58. The
inclusion of Y/Ln is motivated by the idea that elements with large atomic sizes can
destabilize the competing crystalline phase, enabling the amorphous phase to be
formed. It is found that the role of Y/Ln as a fluxing agent is relatively small in terms
of glass formability enhancement. The obtained bulk metallic glasses are
non-ferromagnetic and exhibit high elastic moduli of approximately 180?200 GPa and
microhardness of approximately 13 GPa.

Iron-based bulk metallic glasses have been reported to
exhibit high yield strengths two to three times those of
high-strength steels and elastic moduli comparable to
those of super-austenitic steel alloys.1,2 These bulk metallic
glasses alloys are beginning to gain recognition as
a new class of structural materials with certain superior
properties, such as strength. In particular, bulk amorphous
Fe?Mn?Cr?Mo?C?B alloys, which can be cast
into 4 mm-diameter rod-shaped samples, are called nonferromagnetic
amorphous steel alloys.2 To date, the most
formable Fe-based bulk metallic glasses are reported to
form 5 to 6-mm-diameter samples.3,4 Because of the ability
to process them is limited, the potential for Fe-based
bulk metallic glasses as structural materials has yet to be
realized.
In this paper we report a significant increase in the
formability of Fe?Cr?Mo?C?B bulk metallic glasses
when alloyed with few atomic volume fractions of Y and
Ln (lanthanides). The improvement on glass formability
upon adding Y/Ln is realized. The maximum attainable
diameter of glassy Fe50Cr15Mo14C15B6 samples is
only 1.5 mm, but is increased to 9?12 mm with only
2 at.% Y/Ln addition. Although the study focused mainly
on the Er-containing alloys, additions of other Ln such as
Dy, Yb, and Gd produce essentially similar results. Preliminary
measurements of mechanical and magnetic
properties are also reported. While it has recently been
shown that the glass formability of Fe?Co?Mo?Zr?B alloys
can be improved by adding 2 at.% Y,4 the idea of
adding Y/Ln to Fe?Mn?Cr?Mo?C?B alloys has been independently
pursued by our group. The key results presented
herein were reported in a program review conducted
by DARPA in February 2003; a patent disclosure
on our alloys was filed by the University of Virginia
Patent Foundation in April 2003. The utilization of Y/Ln
to enhance the glass formability is motivated by the idea
that if the atomic-level stress due to large atom solutes
has become too large for the crystalline state to remain
stable, the system will be left in the vitrified state upon
cooling from the melt.5 The Y/Ln to Fe atomic size ratios
of approximately 1.4 are practically the largest values
attainable in Fe-based alloys. Since Fe?Cr?Mo?C?B is
found to devitrify into a single crystalline phase, the present
alloy provides an ideal system for applying the idea
of vitrification via destabilization of the crystalline state.
In this work, the roles of Y/Ln on glass formability are
investigated.
Alloy ingots were prepared by melting appropriate
amounts of Fe (99.9%), Cr (99.99%), Mo (99.9%), Y/Ln
(99.9%), C (99.99%), and B (99.9%) in an arc furnace
under an argon gas. The total weight loss due to melting
was found to be less than 0.2% of the starting materials.
Samples were prepared by injecting the molten alloys
contained in a quartz tube into the cylinder-shaped cavity
of a copper mold. The prepared samples were sectioned
and metallographically examined, using an optical
microscope to explore the inhomogeneity across the
fractured surface. Using a Scintag x-ray diffractometer
and copper target, x-ray diffraction (XRD) was
performed to examine the amorphicity of the inner parts
of the samples. Thermal studies were performed using
differential thermal analysis (DTA) and differential scanning
calorimetry (DSC) techniques. The heating rate applied
was 10 ?C per minute. Oxygen contents in the
samples were analyzed at Wah Chang, an Allegheny
Technologies company. Samples were fused with graphite
crucibles in an impulse-type fusion furnace heated to
approximately 3000 ?C, followed by infrared detection of
carbon dioxide. Preliminary measurements of mechanical
and magnetic properties were performed. The nonferromagnetic
nature of the samples was tested with a
strong bar magnet. Magnetic transitions of some samples
were detected by using a Quantum Design magnetic
property measurement system. Elastic moduli were determined
by employing the resonant ultrasound spectroscopy
technique.6 The ultrasound spectroscope was manufactured
by Quasar International. Measurements were
performed on rectangular parallelepiped with dimensions
2.0 ? 3.0 ? 3.5 mm3. Hardness measurements were made
using a Vickers microhardness tester.
The XRD obtained for a 10-mm-diameter sample is
shown in Fig. 1. Similar patterns that indicate amorphicity
of the samples are also obtained for the 9-mm
and 12-mm-diameter rods. A camera photo of two
glassy rods of 10 mm and 12 mm in diameter is shown
in the inset of the figure. The fractured segment of a
12 mm-diameter sample is also included in the photo. A
shiny appearance typical of a bulk glassy alloy can be
seen on the fractured surface. In casting the samples with
large diameters, the casting conditions must be optimized
to achieve a more uniform flow so that a large size
whole-piece product can be obtained. In addition to the
XRD, isothermal DSC experiments are also performed in
the supercooled liquid region. The isothermal transformation
curves obtained for glassy Fe48Cr15Mo14Er2C15B6
shown in Fig. 2 follow those expected in a truly amorphous
phase that crystallizes via a nucleation-and-growth
process.7 This analysis is based on the Johnson?Mehl?
Avrami transformation theory8 in which an effective activation
energy for the nucleation-and-growth process is
obtained as approximately 400 kJ/mol, or about 4.2 eV.
The alloys exhibit non-eutectic melting with a liquidus
region of 70?90 ?C, as shown in Fig. 3. Results obtained
from several typical alloys are listed in Table I. The
supercooled liquid regions are only approximately 40 ?C,
and the changes in glass transition temperature Tg and
crystallization temperature Tx resulting from Y/Ln additions
are small. Despite the decrease in liquidus temperature,
Tl of up to 30 ?C, given the high Tl, the reduced
glass transition temperature Trg increases only from 0.57
in Fe50Cr15Mo14C15B6 to 0.58 in the large size amorphous
samples.
Some preliminary measurements of properties are that
the alloys exhibit a spin-glass-like magnetic transition at
approximately 30 K. The elastic constants of glassy alloys
of composition Fe48Cr15Mo14Er2C15B6 and similar
glassy alloys but with slightly different Mo and Y/Ln
contents exhibit basically similar compressional modulus
C11 and shear modulus C44 values of approximately 280
and approximately 75 GPa, respectively. Measurements
made on different samples differ by less than approximately
5%. The Young?s and bulk moduli are found to be
approximately 200 GPa and about 180 GPa, respectively.
The bulk modulus is approaching that of super-austenitic
steel.9 The microhardness of these alloys are typically in
the range 1200?1300 DPN, based on which a tensile
strength of approximately 4 GPa can be estimated.1
Comparing with those alloys that do not contain Y/Ln,
it is suggested that the remarkable enhancement in glass
formability observed cannot be attributed to the relatively
low Trg ∼ 0.58. If a different parameter Tx/(Tg + Tl)
were used,4 the reduced temperature value would be approximately
0.39. In comparison, previous reports of
5-mm-diameter amorphous iron alloys showed appreciably
higher Trg ∼ 0.61 to 0.62 and Tx/(Tg + Tl) ∼ 0.41.3,4
Further studies will be needed to fully understand the
significant enhancement in glass formability despite the
moderate Trg value. One plausible mechanism is that the
competing single metastable Fe23C6 phase10 that forms
upon the devitrification of Fe?Cr?Mo?C?B becomes less
stable when alloyed with Y/Ln, as found in recent cohesive
energy calculations.11 Meanwhile, it is observed that
upon alloying with Y/Ln, the growth of Fe23C6 phase
during devitirification is drastically impeded. These findings
taken in sum suggest that while the large atoms can
be accommodated in the melt, their presences in the crystalline
structure decrease the stability of the Fe23C6