Production of high-strength Al-based alloys by consolidation of amorphous and partially amorphous powders D I S S E R T A T I O N zur Erlangung des akademischen Grades Doctoringenieur (Dr.-Ing.) vorgelegt der Fakultät Maschinenwesen der Technische Universität Dresden von Kumar Babu Surreddi geboren am 1 Febraur 1977 in Visakhapatnam (Indien) Gutachter: ......................................................... ......................................................... ......................................................... Eingereicht am: .............................................. Tag der Verteidigung: ......................................
124
Embed
Production of high-strength Al-based alloys by ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Production of high-strength Al-based alloys by
consolidation of amorphous and partially amorphous powders
D I S S E R T A T I O N
zur Erlangung des akademischen Grades
Doctoringenieur (Dr.-Ing.)
vorgelegt
der Fakultät Maschinenwesen
der Technische Universität Dresden
von
Kumar Babu Surreddi
geboren am 1 Febraur 1977 in Visakhapatnam (Indien)
involves an increase in the energy of the starting crystalline material by the addition of
some externally provided energy, and the storage of this energy in the crystal up to a point
at which it becomes unstable with respect to the amorphous state. The highly energized
material then lowers its energy by transforming into a different atomic structural
arrangement, i.e., the glass [Schultz 1994].
1.2.1 Rapid solidification
Rapid solidification (RS), which includes techniques such as melt spinning and gas
atomization, involves high velocity of propagation of the advanced solidification front
[Jones 1999] and cooling rates from 103 to 109 K/s [Anantharaman 1987]. Rapid
solidification results from the rapid extraction of the heat of transformation from the mass
of molten metal or alloy either directly by the external heat sink and /or internally by the
undercooled melt. When extraction of heat is so rapid, the liquid undergoes a significant
undercooling [Anantharaman 1987] and, as a result of the limited atomic mobility, the
structural disorder of the liquid phase liquid is retained (quenching-in) in the glassy
material. Therefore, as a first approximation, the structure of a glass can be considered as a
frozen liquid.
The main features of this process can be understood by considering the changes in
Gibbs free energy G, viscosity η and density ρ, which occur when a glass-forming system
9
is cooled from the liquid into the crystalline or glassy states (Figure 1.1 free energy,
viscosity and density diagram). At high temperatures, above Tliq, the liquid has a lower G
than the crystalline solid. Therefore, it is the thermodynamically stable phase at that
temperature. Below Tliq the undercooled melt is thermodynamically less stable than the
crystalline phase and may crystallize if a critical nucleus is provided, giving rise to the
discontinuous increase of viscosity shown in Figure 1.1.
Figure 1.1 Schematic representation of the variation in Gibbs free energy G, viscosity η and density ρ as a function of temperature occurring when a glass-forming system is cooled from the liquid state into the crystalline or glassy states (after [Cahn 1996, Davies 1983, Greer 1993]).
On the other hand, crystallization can be avoided if the cooling rate is high enough
to prevent nucleation. In this case, the system continues to follow the liquid Gibbs free
energy curve without any change at Tliq. The viscosity continuously increases and its
10
equilibrium values can be well described by the Vogel-Fulcher-Tamman (VFT) empirical
Figure 1.3 Ball-powder-ball collision of powder mixture and formation of layered powder particles during mechanical alloying ([Suryanarayana 2001]).
Alloying begins to occur at this stage due to the combination of decreased diffusion
distances (interlayer spacing), increased lattice defect density, and any heating that may
have occurred during the milling operation [Suryanarayana 2001]. Therefore,
amorphization during MA is not a purely mechanical process, but involves an
interdiffusion process, driven by the negative heat of mixing, of the thin layers in a similar
way as observed in diffusion couples [Schwarz 1983].
Mechanical milling: Amorphization by MM consists of energizing the equilibrium
crystalline compound by the severe cyclic deformation provided by the milling process.
The advantage of MM over MA is that since the powders are already alloyed and only a
reduction in particle size and/or other transformations need to be induced mechanically, the
time required for processing is generally shorter than that for MA [Suryanarayana 2001].
An interesting aspect of MM is that instead of lowering the Gibbs free energy of the
system, in this process the free energy of the equilibrium crystalline compound is raised to
a level equal to or larger than that of the amorphous phase. The mechanical treatment
increases the Gibbs free energy of the intermetallic compound by the generation of
chemical disorder, point defects, such as vacancies, and lattice defects (e.g. dislocations)
[Schwarz 1988]. In addition, an important contribution to the energy increase most likely
comes from the reduction of grain size to a nanometer level and the consequent storage of
energy in the grain boundaries, which constitute an appreciable fraction of the material
volume [Bakker 1995].
14
1.2.3 Crystallization of metallic glasses
Regardless of the processing route used for their production, metallic glasses are
not in a state of internal equilibrium and, when heated to a sufficiently high temperature,
they tend to a more stable condition. Upon annealing below the glass transition
temperature, the glass initially relaxes towards a state corresponding to the ideal frozen
liquid with lower energy [Cahn 1993]. The structure evolves to one with higher density,
which could be considered characteristic of glass formation at a slower cooling rate [Cahn
1993] and finally, above the glass transition temperature, the glass crystallizes.
Crystallization studies of metallic glasses are of primary importance not only in
order to analyze their thermal stability against crystallization but also to investigate the
fundamental aspect of the processes of nucleation and growth. Metallic glasses crystallize
by a nucleation and growth mechanism [Köster 1981] in a similar way as for solidification
of liquids below their liquidus temperature; however, since the crystallization process is
much slower than solidification of liquids, it is relatively easier to investigate the
crystallization in glasses than in liquids. Metallic glasses can be considered as deeply
undercooled liquids, therefore, their crystallization behavior may be analyzed in a similar
way.
When a liquid is cooled below the liquidus temperature, it is energetically less
stable than the crystalline phase and tends to transform to the more stable crystalline solid
(Figure 1.4). The difference in Gibbs free energy between these phases provides the
driving force for the nucleation process. However, the crystallization does not start
immediately after the system has reached the range of parameters where the new phase has
the lowest free energy. The liquid must be undercooled below Tliq before crystallization
can occur, due to the existence of an energy barrier to nucleation [Herlach 1997, Kelton
2004]. The crystallization of a liquid is not a transformation that occurs in the entire
volume at once, but it starts and progressively extends from discrete centers throughout the
material [Christian 2002a]. These centers are aggregates of atoms characterized by an
atomic configuration similar to the lattice of the product phase. However, not all these
aggregate s of the new phase, called embryos, are stable. In fact, embryos below a critical
15
Figure 1.4 Schematic free energy diagram as a function of temperature for a liquid undercooled below the liquidus temperature (Tliq). Gliq and Gsol refer to the Gibbs free energies of the liquid and solid phases, respectively (after [Porter 2009]).
minimum size are associated with an increase in Gibbs free energy, are instable, and thus
quickly disintegrate [Burke 1965]. The reason for this is that the formation of an embryo
within the parent phase is accompanied by the creation of an interface [Burke 1965,
Christian 2002a]. Due to the different structures between the liquid and the crystal, a
mismatch along the interface arises. The positive energy associated to this interface has to
be supplied by the Gibbs free energy of the transformation and thus disfavors the
crystallization of the liquid [Burke 1965]. On the other hand, aggregates larger than the
critical size are stable and capable of continuous existence. Such stable structures are
called nuclei and their formation is termed nucleation [Porter 2009]. Nucleation that occurs
randomly throughout a system in the absence of foreign bodies that can catalyze
crystallization is said to be homogeneous. In contrast, nucleation at preferred sites is named
heterogeneous [Turnbull 1950].
The resistance of liquids to nucleation can be better understood in the framework of
the classical theory of nucleation [Christian 2002a] for vapor condensation, where it is
assumed that the embryos have uniform structure, composition and properties. These
assumptions leave the shape and size of the embryo or nucleus as the only variable
parameters. For homogeneous nucleation, ΔG of formation of a spherical embryo of radius
r within the liquid phase is given by [Burke 1965, Christian 2002a, Fisher 1948, Porter
2009, Turnbull 1969]:
16
ΔG = 3
4π r3 ΔGv + 4π r2 σ , (1.4)
where ΔGv is the Gibbs free energy difference between unit volume of the crystal and
liquid, and σ is the interfacial energy per unit area of the solid/liquid interface. Figure 1.5
shows ΔG as a function of r at a temperature below Tliq, where the crystal is
thermodynamically more stable than the liquid, i.e. ΔGv is negative below Tliq (ΔGv and σ
are assumed to be independent of r). There is a clear competition between the interfacial
energy and the ΔGv terms. In fact, the interfacial energy is always positive and, therefore,
is opposed to ΔGv. ΔG passes through a maximum, denoted ΔG#, at a radius r# (the critical
nucleus size). In the case of embryos with radii smaller than r#, the interfacial energy is
greater than the volume free energy, with the result that there is a net increase in ΔG upon
growth and the embryos have the tendency to shrink rather than to grow. In contrast, for
values greater than r#, the volume free energy term dominates on the surface term because
it is proportional to r3. In this case, the net free energy change accompanying the
transformation is negative, with the result that large embryos (nuclei) are stable. Embryos
of radius r# have an equal possibility to shrink or to grow [Burke 1965, Porter 2009].
Figure 1.5 Free energy of formation of a spherical embryo as a function of the radius r (after [Herlach 1993]).
The critical nucleus size, r#, is defined by the condition 0#
=∂Δ∂
=rrrG , which gives
17
r# = GΔ
−σ2 . (1.5)
The critical value of ΔG corresponding to r# is equal to
ΔG# = 2
3
316
vGΔπσ , (1.6)
which corresponds to the activation energy for homogeneous nucleation, i.e. the barrier to
nucleation that has to be overcome in order to form a nucleus of critical size [Burke 1965,
Porter 2009].
Metallic glasses can be used as precursors for nanocrystalline materials, perhaps the
most attractive microstructure from the point of view of the functional properties. For
example, glass-matrix composites consisting of nanosized particles embedded in a glassy
matrix can be produced by controlled devitrification (crystallization) of metallic glasses
[Chen 1999, Inoue 1997b, Inoue 1999]. This technique has been used for long time for
conventional glasses [Holand 2001] in order to produce composite materials with a wide
variety of microstructures and advantageous properties.
The basic principle for the production of glass-matrix composites by crystallization
of a glassy precursor is to control the crystallization kinetics by optimizing the annealing
conditions (annealing temperature and time, heating rate, etc.) and chemical composition in
order to obtain a glassy phase that partially or completely transforms into a nanocrystalline
material with the desired microstructure [Cahn 1996, Köster 1981].
Controlling the microstructure development from amorphous precursors requires
detailed understanding of the specific mechanisms influencing structural transformations.
Thermal analysis, in particular differential scanning calorimetry (DSC), has been
successfully employed for studying the phase transformations involving nucleation and
growth, continuous grain growth of pre-existing nuclei and in general for investigating the
crystallization kinetics of glass-forming liquids and metallic glasses [Scott 1977, Weinberg
1996, Yinnon 1983].
Crystallization is a thermally activated reaction. A general objective of the
modeling of thermally activated reactions is the derivation of a complete description of the
progress of a reaction that is valid for any thermal treatment, be it isothermal or by linear
heating (isochronal). However, this is a difficult task because any reaction might progress
through a range of mechanisms and intermediate stages, all of which can have a different
18
temperature dependency. To come to terms with this potentially very complicated problem,
most researchers attempt to achieve this objective by making a few judiciously simplifying
assumptions. A simplifying assumption that is encountered in numerous publications is the
hypothesis that the transformation rate during a reaction is the product of two functions,
one depending solely on the temperature, T, and the other depending exclusively on the
fraction transformed, α [Starink 2004].
( ) )(Tkfdtd αα
= . (1.7)
The temperature dependent function is generally assumed to follow an Arrhenius-
type dependency
⎟⎠⎞
⎜⎝⎛−=
RTEkk exp0 . (1.8)
Thus, to describe the progress of the reaction at all temperatures and for all
temperature-time programs, the function f(α), the reaction constant, k0 and the activation
energy, E need to be determined. In general, the reaction function f(α) is unknown at the
outset of the analysis. From the above equations, it follows that for transformation studies
performed at a constant temperature, T, E can be obtained from the equation [Christian
2002b]:
( ) ii
f cRTEt +⎟⎟
⎠
⎞⎜⎜⎝
⎛=ln , (1.9)
where tf is the time needed to reach a certain fraction transformed and ci is a constant,
which depends on the reaction stage and on the kinetic model. Thus, E can be obtained
from two or more experiments at different T. For isothermal experiments, k(T) is constant,
the determination of f(α) is relatively straightforward, and is independent of E.
For non-isothermal experiments, the reaction rate at all times depends on both f(α)
and k(T), and the determination of f(α), k0 and E (the so-called kinetic triplet) is an
interlinked problem. A deviation in the determination of any of the three will cause a
deviation in the other parameters of the triplet. Over the past decades a variety of non-
isothermal methods have been proposed. Though most of them are used for oxide-glasses,
in case of metallic glasses the most widely used non-isothermal methods are the Kissinger
[Matusita 1984], Gao and Wang method [Gao 1986], Augis and Bennet’s method [Augis
1978], and Lasocka’s method [Lasocka 1984]. While isothermal analyses are in most
cases more definitive, it has been shown that the non-isothermal technique also has several
advantages, in particular that experiments can be performed quite rapidly [Henderson
1979]. Additionally, many phase transformations occur too rapidly to be measured under
isothermal conditions because of transients associated with the experimental apparatus
[Henderson 1979]. In this thesis, the isochronal method employed for calculating the
activation energy for the crystallization has been the Kissinger method. Although, the
Kissinger analysis was not originally developed for solid-state transformations, Henderson
has shown that it is applicable to these transformations [Henderson 1979]. The activation
energy calculated using the Kissinger approach depends on the temperature dependences
of the nucleation and growth rates and on any transient events, which they may exhibit.
Despite difficulty in interpretation, this approach has been widely used for comparing the
stability of metallic glasses [Louzguine 2002a, Stoica 2009, Venkataraman 2007a,
Venkataraman 2005]. A high value of the activation energy is generally interpreted as a
measure of the high stability and resistance of the glass towards crystallization.
The activation energy (Ea) for the crystallization processes can be evaluated from
constant-rate heating DSC scans taken at different heating rates using the Kissinger method
[Kissinger 1957]. The method is based on the assumption that if a reaction proceeds at a
rate varying with temperature i.e. possesses activation energy, the position of the
calorimetric DSC peak, Tp, varies with the heating rate if the other experimental conditions
are maintained fixed [Kissinger 1957]. The variation of the peak temperature can be used
to determine the activation energy of the reaction. According to the Kissinger method, the
crystallization peak temperature, Tp, in the DSC scan depends on the heating rate, φ, as
follows:
⎟⎟⎠
⎞⎜⎜⎝
⎛−=⎟
⎟⎠
⎞⎜⎜⎝
⎛
P
a
P RTE
T 2ln φ+ Constant . (1.10)
By plotting ln(φ/T2P) versus (1/TP), a straight line with slope Ea / R is obtained, where Ea is
the crystallization activation energy and R is the gas constant.
On the other hand, kinetic data on first-order transformations are often obtained by
isothermal analysis. One of the legacies of the classic work done by Kolmogorov
20
[Kolmogorov 1937], Johnson and Mehl [Johnson 1939] and Avrami [Avrami 1939, 1940,
1941] concerning the kinetics of phase transformations involving nucleation and growth
under isothermal conditions is the Johnson-Mehl-Avrami (JMA) [Yavari 1999]
transformation equation.
The crystallized volume fraction during isothermal annealing can be determined
accurately by measuring the partial area of the exothermic signal [Málek 2000, Peng 2005]
assuming that the volume fraction of the transformed material (X) at any given time (t) is
directly proportional to the fractional area of the exothermic peak [Scudino 2008,
Venkataraman 2007b]. According to the JMA equation [Christian 2002b]:
[ ]( )nT tKtX )(exp1)( τ−−−= , (1.11)
where X is the volume fraction of the crystallized phase(s), t the annealing time, KT is a
kinetic constant dependent on the temperature, n is the Avrami exponent and τ the
incubation time for the process. The incubation time is the time period that must elapse
prior to formation of nuclei.
The kinetic constant KT (which can be used to estimate the activation energy for the
transformation, e.g. the devitrification process) is a function of the annealing temperature
and, assuming it to be described by an Arrhenius-type equation, it can be written as
⎟⎠⎞
⎜⎝⎛ −
=RTEKK A
T exp0 , (1.12)
where K0 is a constant and EA is the activation energy for crystallization. The most
important use of this equation has been in the determination of the Avrami exponent. The
Avrami exponent n can vary from 1 to 4 and it is used to describe the transformation
mechanism, such as the nucleation and growth behavior [Christian 2002a]. Based on the n
values, valuable information about the phase transformation can be obtained especially
regarding the nucleation and growth processes as a function of time [Christian 2002b].
Avrami exponent greater than 2.5 implies increasing nucleation rate of all shapes growing
from small dimensions. Constant nucleation rate occurs when n is 2.5 while decreasing
nucleation rate takes place when n is in between 1.5 and 2.5. Zero nucleation rate occurs
when n is 1.5. When Avrami exponent n is 1, it represents formation of needles and plates
of finite long dimensions and thickening of long needles.
21
The values of KT and n can be calculated by using the relation [Avrami 1939, 1940,
1941]
)ln(ln1
1lnln τ−+=⎥⎦
⎤⎢⎣
⎡⎟⎠⎞
⎜⎝⎛
−tnKn
X T . (1.13)
Plotting ln[ln(1/(1-X))] against ln(t-τ) for different annealing temperatures, JMA plots are
obtained. The Avrami exponent n is the slope of the JMA plots.
1.3 Powder metallurgy
Powder metallurgy (P/M) may be defined as a near-net or net-shape manufacturing
process that combines the features of shape-making technology for powder compaction
with the development of the desired microstructures and properties (physical and
mechanical) through subsequent densification or consolidation processes (e.g., sintering)
[Sanderow 1997]. In this process, parts can be produced from metal powders without
passing through the molten state [Dowson 1990]. This process is highly cost effective in
producing simple or complex parts at, or close to, final dimensions with respect to other
fabrication methods like casting, stamping or machining. P/M is the best choice when
requirements for strength, wear resistance or high operating temperatures exceed the
capabilities of die casting alloys [Upadhyaya 2002]. In addition, P/M offers greater
precision, eliminating most or all of the finish machining operations required for castings,
and it avoids casting defects such as blow holes, shrinkage and inclusions [Upadhyaya
2002]. This method is used for several groups of important materials, such as refractory,
composite, porous and glassy materials [Dowson 1990].
P/M processing provides the following advantages over other processing routes
[Trudel 1998]:
Production of complex shapes to very close dimensional tolerances, with minimum
scrap loss and fewer secondary machining operations.
Physical and mechanical properties of the components can be tailored through close
control of starting materials and process parameters.
Particular properties can be improved through secondary processing operations
such as heat treating and cold/hot forming.
22
The main steps in P/M are the powder production and its consolidation. The
sequence of operations to obtain the final product is threefold [Dowson 1990]:
Mixing or blending the powders according to the desired composition and structure.
Loading the powder mix or blend into a suitable die or mould followed by the
powder consolidation through the application of pressure with or without heat in
either controlled or open atmosphere.
Heating the compacts, generally in protective atmosphere, to cause the particles to
bond together. This process is called sintering. The sintering temperature is
normally below, and many cases significantly below, the melting point of the metal
or alloy compact.
1.3.1 Powder production
There are numerous methods of powder production [Dowson 1990], which involve
mechanical (e.g. ball milling) and chemical methods (e.g. reduction of oxides). Mechanical
methods for powder production consist of mechanical comminution, milling and grinding.
Powders produced by chemical methods involve chemical reduction and decomposition of
compounds. Another procedure for powder production is atomization, in which a stream of
molten metal is forced through a small nozzle and then, depending upon the metal involved,
is disintegrated by a jet of water or gas [Dunkley 1986]. The selection of the processing
route for metal powder production is based on the raw material available and the desired
end product and its structure [Upadhyaya 2002]. Suitable methods for powder production
depend on required production rates, powder properties, and the physical and chemical
properties of the material. Chemical and electrolytic methods are useful for producing
high-purity powders [Trudel 1998]. Mechanical milling is the most widely used method of
powder production for hard metals and oxides [Suryanarayana 2001]. Additional milling of
atomized or electrolytic powders is also a very common and economical practice to
produce uniform particle size and shape [Trudel 1998]. Among the different processing
routes, atomization is perhaps the most versatile method that produces metal powders over
a wide range of production rates (from 1 to 105 tons/yr) and a wide variety of powder sizes
from 10 to 1000 µm [Yule 1994].
The most important parameters of any particulate product are particle shape and
size, particle size distribution, purity and apparent density [Dowson 1990]. These
parameters are strictly linked to the powder production process used and they significantly
23
influence the properties of the final material. For example, large particle sizes give rise to
porous end products compared to fine particles. Therefore, a fine grain size is generally
preferred as it also gives greater particle strength, which helps to prevent particle fracture
during compacting [Hirschhorn 1969]. Particle shape is also an important factor as
irregularly shaped particles are required to ensure that the as-pressed component has a high
green strength from the interlocking and plastic deformation of individual particles with
their neighbors. In addition, powder densification can also be remarkably influenced by
particle size distribution [Ferguson 1998].
1.3.2 Metal powder compaction
The compaction of metallic powders has two major functions: to consolidate the
metal powders into desired shape and to sinter the compacts to obtain desired structure and
density [Upadhyaya 2002]. Die compaction process is the one of the most used compaction
method. In this method, shown schematically in Figure 1.6, a die cavity of the desired
shape is filled with the metal powder. Pressure is applied by the axial movement of one or
both punches. The pressure causes the metal particles to mechanically interlock and cold
weld together into a porous mass of the approximate shape and dimensions desired for the
final component. This as-pressed shape, commonly referred to as a green compact, is then
heated to elevated temperatures to achieve full density [Ferguson 1998].
Although most of the sintered parts are made by pressing the powder mix at
ambient temperature followed by sintering, hot pressing is used in certain cases, such as for
hard and brittle materials, where pressure and heat are applied simultaneously. At elevated
temperatures metals are softer and, therefore, it is usually possible to achieve higher
density without increasing the applied pressure [Dowson 1990]. Hot pressing is a suitable
method for densifying materials with poor sintering behavior. This technique, which
combines powder pressing and sintering into one single operation, offers many advantages
over conventional powder consolidation. By the simultaneous application of temperature
and pressure, it is feasible to achieve near theoretical density in a wide range of hard-to-
work materials. As the resistance of metal particles to plastic deformation decreases
rapidly with increasing temperature, much lower pressures are required for consolidation
by hot pressing. In addition, densification by hot pressing is relatively less sensitive to
powder characteristics, shape, size and size distribution, which are critical in cold pressing
and sintering [Upadhyaya 2002].
24
Figure 1.6 Schematic diagram of cross sectional view of uni-axial hot pressing. The die containing the powder is externally heated while pressure is applied through the upper and lower punches (after [German 1996]).
The parameters controlling hot pressing (pressure, temperature, time and the
working atmosphere) largely control the properties of the compacts. The various steps
involved in the hot pressing procedure are the following [Upadhyaya 2002]:
1. Powder or a cold compacted preform is placed into the die mould.
2. The mould is heated either by resistance or by induction to a predetermined temperature.
3. The powder in the die cavity is then pressurized.
4. The temperature is steadily increased during compacting until a maximum required
temperature is reached.
5. Compacting pressure and temperature are maintained for a dwell time and
6. The mould is cooled slowly, under pressure, to a temperature at which oxidation of the
material would not occur.
There are many variations on the general procedure given above. In many cases, it
is preferable to apply a nominal pressure or even the maximum required compacting
pressure before the initiation of consolidation cycle. In place of inert gas, vacuum for hot
pressing offers additional advantages of removing air from the powder body thus
eliminating the possibility of air entrapment [Upadhyaya 2002]. The use of elevated
temperatures and long dwell times allows densities of >95% to be achieved at compaction
pressures that are one third to one half those needed for cold pressing to lower density
levels [Ferguson 1998]. Higher densities can be achieved by extrusion. Extrusion is a
plastic deformation process to produce highly dense bulk samples in which a pre-
25
compacted sample or a billet is forced to flow by compression through the die orifice of a
smaller cross-sectional area than that of the original sample (Figure 1.7).
Figure 1.7 Schematic illustration of a vertical uni-axial hot extrusion process.
Because of the large forces required in extrusion, most metals are hot extruded
under conditions where the deformation resistance of the metal is low. Depending on the
material being extruded, hot extrusion is done at relatively high temperatures. Besides the
working temperature, other important parameters of hot extrusion are the extrusion ratio,
the working pressure, the speed of deformation and the frictional conditions and
lubrication. Among these parameters, the extrusion ratio (the ratio of the initial cross-
sectional area of the sample to the final cross-sectional area after extrusion) is the main
factor for achieving the desired density.
1.3.3 Sintering
The ISO (International Organization for Standardization) definition of sintering is:
“The thermal treatment of a powder or compact at a temperature below the melting point
of the main constituent, for the purpose of increasing its strength by bonding together of
the particles” [Dowson 1990]. Bonding together of the particles implies the formation of
bonds in the areas where neighboring particles are deformed at their points of contact by
the applied pressure. During sintering these areas of metallurgical contact grow and the
strength of the sintered body progressively increases [Dowson 1990].
Sintering is a complex process and for any given metal and set of sintering
conditions there are different stages, driving forces and material transport mechanisms
26
associated with the process. The various stages of sintering can be grouped in the
following sequence [Exner 1979, Hirschhorn 1969, Upadhyaya 2002]:
(1) Initial bonding among particles
(2) Neck growth
(3) Pore channel closure
(4) Pore rounding
(5) Densification or pore shrinkage
(6) Pore coarsening
Figure 1.8 Schematic illustration of two-particle model for initial stage of sintering (a) without shrinkage (b) with shrinkage (after [Exner 1979, Kang 2005]).
Bonding takes place very early in the sintering process as the materials is heated up. The
bonding process involves diffusion of atoms leading to the development of grain
boundaries. This takes place at sites where intimate physical contact between adjacent
particles occurs.
Neck growth is the second stage of sintering and is closely related to the first stage of
initial bonding. The newly formed bonded areas are termed necks, which grow in the
second stage of sintering. The neck growth requires the transport of materials within the
sintered mass but does not imply any decrease in the amount of porosity, i.e. no shrinkage
of the material [Exner 1979, German 1996]. Neck growth takes place rather rapidly in the
early stages of the sintering process and continues in the following stages. Figure 1.8
shows the schematic illustration of two-particles sintering together and defining geometry
of neck growth. Initially, the contacts between the spheres are point contacts. After some
27
sintering, due to necking, the contacts become more planar in nature. Neck growth also
results in growth of the initial grain boundaries associated with stage one.
Pore channel closure represents a rather major change in the nature of the porosity in the
sintered mass. Closing off of the tortuous and interconnected pore channels leads to the
development of isolated or closed porosity. Pore channel closure stage may proceed for
some time and overlap stages four and five. The change from interconnected to isolated
porosity can usually be observed microscopically. In particular it is noted that: (a) with
porosities greater than about 10 vol.% most of the porosity is in an interconnected form;
(b) with porosities less than about 5 to 10 vol.% most of the porosity is of the closed or
isolated type [German 1996].
Figure 1.9 Schematic illustration of three particle sintering model: (a) original point contacts, (b) neck growth, (c) and (d) pore rounding (after [Exner 1979, Hirschhorn 1969]).
Pore rounding may be considered as natural consequence of neck growth. When material
is transported to the neck regions from the pore surfaces, the pores themselves become
more rounded as shown in Figure 1.9. With sufficient time at temperature is possible to
achieve almost perfectly spherical pores. Pore rounding is promoted by high sintering
temperatures. This stage of sintering is particularly important with respect to the influence
of porosity on the mechanical properties of the sintered materials [Exner 1979, German
1996].
Pore shrinkage and eventual pore elimination is often considered as the most important
stage of sintering. Only with sufficient time at temperature may it evidence itself by
densification of the sintered mass. It is important to realize that the process of pore
28
shrinkage, leading to a decrease in the volume of the sinter mass, must involve movement
of the solid into the porosity.
Pore coarsening usually takes place after most of the other stages of sintering have
occurred. The process simply consists of the shrinkage and elimination of small isolated
pores and the growth of larger ones. The total amount of porosity associated with all these
pores remains the same, but the number of pores decreases and the average size increases.
Hence, no densification of the material is associated with this stage [German 2008].
Driving force for sintering
At elevated temperatures, the loose powder or the compact is not at equilibrium and
it is prone to substantial changes in its internal structure towards a more stable state [Tilley
2004]. The driving force for the change is the reduction in free energy of the system. In the
sintering process, the necessary reduction of free energy is associated with the decrease in
internal surface area of the sintered mass. A decrease in surface area corresponds to a
decrease in the surface free energy contribution to the total free energy of the system (i.e.
area multiplied by specific surface or interfacial free energy). The total free energy of a
powder compact is expressed as γA, where γ is the specific surface (interface) energy and A
the total surface (interface) area of the compact. The variation of the total free energy can
be expressed as [Exner 1979, Kang 2005]:
AAA Δ+Δ=Δ γγγ )( . (1.14)
Here, the change in interfacial energy (Δγ) is due to densification and the change in
interfacial area ΔA is due to grain coarsening. For solid state sintering, Δγ is related to the
replacement of solid/vapor interfaces (surface) by solid/solid interfaces [Kang 2005].
In other words, the sintered mass undergoes changes that tend to eliminate the
internal surface area. For example, pore rounding reduces the surface area while
maintaining the amount of porosity at a constant level. This is because the ratio of surface
area to volume is reduced when the shape of the pore approaches a sphere. The surface
area can then be reduced by the pore shrinkage and the densification stage of sintering. The
ratio of surface area to volume is also decreased by increasing the average size of the pores
while maintaining the total volume constant (i.e. pore coarsening). As well, pore channel
closure decreases the surface/volume ratio. Hence, the greater the amount of surface area
in the original materials, the greater the driving force for sintering [Tilley 2004].
29
Transport mechanisms
Sintering can be considered as “a thermally activated material transport in a powder
mass or a porous compact, decreasing the specific surface by growth of particle contacts,
shrinkage of pore volume and change of pore geometry” [Thümmler 1993]. Accordingly,
most sintering theories are based on transport phenomena associated with a particular stage
of sintering. In the following, the major mechanisms of material transport are presented
[Kang 2005]. Different sintering mechanisms are illustrated in two-particle model in the
Figure 1.10
Figure 1.10 Schematic illustration of different material transport paths during sintering as applied to the two-sphere sintering model (E-C, evaporation-condensation; SD, surface diffusion; VD, volume diffusion; GBD, grain boundary diffusion;) (after [Exner 1979, Kang 2005, Schatt 1987, Schatt 1985a, Schatt 1985b]).
Evaporation and condensation: because of the higher vapor pressure over convex surfaces
as compared to the neck regions, it is possible in some systems for material to be
transported as vapor to the neck region. Neck growth, pore rounding and pore channel
closure can be accomplished by this mechanism. This mechanism is important for
materials with relatively high vapor pressure, so that significant amount of material can be
transported [Exner 1979, German 1996].
Volume (lattice) diffusion: This transport mechanism is widely accepted for the sintering of
metallic materials. Volume or lattice diffusion refers to the movement of atoms within the
solid crystalline material. The most prevalent specific type of atomic motion is the
“vacancy exchange” mechanism. This process involves the movement of atoms into vacant
lattice sites (i.e. vacancies). If there is a directionality associated with a substantial amount
30
of such atomic motion, then there is a net transport of material in a specific direction.
Diffusion of atoms along a specific direction (rather than a random motion) is a
consequence of chemical potential gradients existing within the solid. Material transport by
volume diffusion is due to the existence of vacancy concentration potentials (differences)
in the solid, and the movement of vacancies from regions of high concentration to regions
of low vacancy concentration [Brand 1993, German 1996].
Surface diffusion: Atomic transport in the solid state can also occur by surface diffusion, i.e.
the motion of atoms on external surface. In analogy with volume diffusion, the most
probable mechanism of surface diffusion is the exchange between surface atoms and
surface vacancies. This process is particularly important in the first sintering stages, when
the specific surface is still high. It is generally accepted that surface diffusion does not
cause pore shrinkage and hence densification, however, it can promote neck growth
[Hirschhorn 1969].
Creep deformation: This process, which occurs at elevated temperatures and under
constant load, is a consequence of the repetition of the following steps: (1) generation of
dislocations activated by the applied stress, their movement and their arrest at some
obstacles; (2) dislocation climb and generation of new dislocations. The rate controlling
step is the diffusion dependent climb. There is a stress concentration at the head of the
dislocation pile-up which can lead to the production or annihilation of vacancies. The
driving force for this material transport mechanism is the presence of shear stresses in the
solid. Those factors that promote stresses in the solid, such as a large curvature in the neck,
small pore radius and high surface tension, would tend to increase the probability of having
this mechanism be the dominant one for sintering [German 1996, Hirschhorn 1969, Schatt
1983].
Transformations during sintering
In this section, the main changes or transformations that may take place during
sintering are considered [Hirschhorn 1969].
Grain growth: Grain growth is the most important transformation that occurs during
sintering. In normal sinter mass there is a very large amount of grain boundary areas
because of the small sized particles. Grain boundaries represent a positive contribution to
the free energy of the material; therefore, a large driving force exists for removing the
grain boundaries and for grain growth to reach the lowest energy state (higher degree of
stability). This is why there is substantial grain growth during sintering. Almost any
31
deviation from pure single−phase material will reduce the tendency for grain growth. This
includes (i) grain boundary pinning to decrease grain boundary mobility through residual
pores [Hahn 1990], impurities and solutes [Averback 1993] and second phase particles
[Hillert 1988], and (ii) reduction of the driving force for grain growth by lowering the grain
boundary energy through the addition of solute atoms that segregate at the grain
boundaries [Koch 2009].
Alloying: Very often the sintering operation is used to produce homogeneous alloys from
the original mixture of two or more elemental powders. Alloying during sintering is due to
diffusion. Although most of the available experimental evidence indicates that volume
(lattice) diffusion is the main mechanism for alloying, surface diffusion might become the
dominant mechanism, particularly at low temperatures.
Phase transformations: Many types of phase transformations may occur in the solid state
during sintering at a constant temperature or during the cooling of the material from the
sintering temperature. In some cases such transformations would follow sufficient alloying.
Probably the best example of a phase transformation associated with sintering is the
production of sintered steels. Plain carbon steels would be made by mixing graphite and
iron powder; during sintering the iron and graphite would alloy to form the high
temperature austenite phase (a solid solution of carbon in fcc iron). Once the austenite is
formed then the desired pearlitic structure can be obtained upon cooling.
Influence of material and process parameters
The major variables which determine sinterability and the sintered microstructure
of a powder compact may be divided into two categories: material variables and process
variables. The variables related to raw materials (material variables) include chemical
degree of powder agglomeration, etc. These variables influence the powder compressibility
and sinterability (densification and grain growth). The process variables involved in
sintering are mostly thermodynamic variables, such as temperature, time, atmosphere,
pressure, heating and cooling rate [Exner 1979, German 1996, Kang 2005, Schatt 2007].
Particle size: Decreasing particle size leads to improved sintering. With a smaller particle
size there would be greater inter-particle contact (number of necks) and, hence, more paths
for volume diffusion. Also, a small particles size may correspond to a smaller grain size,
promoting transport by grain boundary diffusion.
32
Particle shape and surface morphology: The optimization of these parameters may lead to
improved intimate physical contact among particles in the sinter mass and increased
internal surface area, promoting sintering. For example, irregularly shaped particles lead to
higher density of the compacts than spherical powders. As well, increasing micro and
macro-surface roughness may assist sintering.
Particle structure: A fine grain structure of the original particles can promote sintering
because of its favorable effect on several material transport mechanisms. The presence of a
large amount of lattice imperfections, such as dislocations, usually resulting from plastic
deformation, can promote sintering because such defects enhance sintering by diffusional
processes.
Particle composition: The driving force for sintering may be either increased or decreased
by alloying elements or impurities in the material. Diffusional mass transport may also be
affected by the presence of alloying or impurity atoms in the lattice.
Temperature: Increasing the sintering temperature greatly increases the rate and magnitude
of any changes occurring during sintering. Many investigations of sintering have indicated
that the precise material transport mechanism, which controls the rate of sintering,
remarkably changes with varying temperature.
Time: Although the degree of sintering increases with increasing time, the effect is small in
comparison to the temperature dependence. The rate of sintering decreases with increasing
time.
Pressure: Combination of temperature and pressure induce accelerated densification
process and elimination of residual pores. Pressure assisted sintering adds stress to
accelerate the material flow during sintering [German 1996].
33
Chapter 2: Sample preparation and characterization
2.1 Preparation of amorphous and partially amorphous powders
The samples investigated in this work were produced through two ways: solid-to-
solid transformation, i.e. by mechanical alloying of elemental powder mixtures, and
through liquid-to-solid transformation, i.e. by melt spinning and gas atomization.
The starting material for mechanical alloying consisted of a mixture of highly pure
elements (Table 2.1) of the given nominal compositions. Differently, pre-alloyed materials
were used as starting materials for the melt spinning experiments. The pre-alloyed
materials were prepared by mixing the appropriate weights of each chemical constituent, in
the form of small lumps, for the desired atomic composition. The lumps were mechanically
cleaned to remove any possible surface oxide layer. The elements were then melted in an
induction furnace in an argon gas atmosphere purified with titanium getter. The rods were
remelted several times to ensure homogeneity in composition. The molten alloy was then
cast into cylindrical rods with 10 mm diameter and 100 mm length by copper mold casting
under argon atmosphere.
Table 2.1. Starting materials for mechanical alloying experiments.
Element Supplier Purity Particle size ≤
Al Alfa Aesar 99.5% 45 µm
Y Mateck 99.9% 420 µm
Ni Alfa Aesar 99.5% 300 µm
Co Alfa Aesar 99.9% 300 µm
2.1.1 Melt spinning
Glassy ribbons were prepared in a single-roller melt spinner (Edmund Bühler D-
7400) under argon atmosphere. In this apparatus (Figure 2.1) a piece of the pre-alloyed rod
(5-7 g) was molten inductively in a quartz tube having a rectangular slit, at the end of the
nozzle, of 3 mm length and 0.7 mm width. The position of the nozzle tip can be adjusted
with respect to the wheel surface so that the molten alloy was perpendicularly ejected onto
the wheel surface from a distance of about 0.3 mm. The whole unit is enclosed in a
34
chamber connected with vacuum pumps and an argon inlet. The chamber was evacuated to
10-3 Pa and rinsed with argon two times. The temperature of the liquid metal was
monitored by an optical pyrometer and when the temperature rises about 150-200 K above
the melting point of the alloy (typically 1200-1400 K), an overpressure of about 250 mbar
of pure argon was applied from
Figure 2.1 (a) Schematic diagram of melt spinning process [Suryanarayana 2001]
(b) single roller Edmund Bühler melt spinning device.
an external reservoir to eject the molten alloy out of the quartz tube onto the external
surface of a copper wheel rotating at a velocity of 40 ms-1. The melt-spun ribbon, which
detached from the wheel surface during melt spinning, was guided towards a collecting
box. The resulting ribbons had typical widths of 3 mm and thicknesses of 40-50 µm.
2.1.2. Ball milling
Milling experiments, starting from elemental powder mixtures with nominal
composition Al85Y8Ni5Co2, were performed using a Retsch PM400 planetary ball mill and
hardened steel balls and vials (Figure 2.2(a)). The schematic illustration of ball milling is
shown in Figure 2.2(b). In this type of mill, four vials are arranged eccentrically on the
supporting disc (sun wheel) of the planetary mill. While operating, the rotation of the
supporting disc is accompanied by a rotation of the vials around their own axes, in the
opposite direction. As a result of the superposition of opposite rotating motions of
supporting disc and vials, the steel balls in the grinding vials are subjected to superimposed
rotational movements, the so-called Coriolis forces. The difference in speeds between the
balls and grinding vials produces an interaction between frictional and impact forces,
35
which releases high energies. The rotational velocity of the supporting disc can be
considered as a rough estimate of the milling intensity. This can be controlled and kept
constant at values ranging from 60 to 400 revolutions per minute (rpm). In the present
work, the rotational velocity was set to 150 rpm for the milling experiments. Milling was
carried out as a sequence of 15 min milling intervals interrupted by 15 min breaks to avoid
a strong temperature rise until no reflections from the starting metallic elements were
detected in the X-ray diffraction pattern. The starting materials for the milling experiments
(typically 30 g) were charged in the milling vials equipped with a flexible “O”-ring,
together with 10 mm-diameter steel balls to give a ball-to-powder mass ratio (BPR) of 10:1.
To avoid any possible atmosphere contamination during milling, vial charging and any
subsequent sample handling was carried out in a Braun MB 150B-G glove box under
purified argon atmosphere (less than 1 ppm of O2 and H2O).
Figure 2.2 (a) Retsch PM400 planetary ball mill chamber showing the vials and balls and (b) schematic illustration of the ball milling principle [Suryanarayana 2001].
Additionally, Al-based glassy powders were produced by controlled milling at
cryogenic temperature of melt-spun glassy ribbons with nominal composition
Al85Y8Ni5Co2. The ribbons were milled for 5 h at a ball-to-powder mass ratio of 10:1 and
at rotational velocity of 150 rpm. The milling was carried out with a BPR of 10:1 as a
sequence of 15 minutes milling intervals interrupted by 15 minutes breaks.
2.1.3 Gas atomization
Gas atomization is one of the most used processing routes for the production of
metallic powders due to its advantages, such as homogeneous and uniform size distribution
36
of particles and possibility to scale up the process to tonnage quantities. Figure 2.3 shows a
schematic representation of a gas atomizer. Atomization consists of breaking up of bulk
liquid metal or alloy into fine droplets and allowing them to solidify as powder particles
[Lawley 1977]. Mostly, these particles are round due to the liquid surface tension. This
property causes thin ligaments of liquid to be unstable; that is, they break up into droplets,
or atomize. The main fluid properties which affect the size of the droplets are surface
tension, viscosity and density. The average droplet size is larger with higher density,
viscosity and surface tension [Lawley 1978]. Atomization is a popular route for large scale
powder production. There are different atomization processes in industrial practice to
produce different metal or alloy powders. Most common atomization processes can be
mainly classified into gas, water and centrifugal atomization processes [Lawley 1978]. In
gas atomization a continuous stream of liquid metal is broken down into droplets by means
of a subsonic or supersonic gas jet. Atomization occurs by the kinetic energy of the
atomizing medium, typically nitrogen, argon, or air. Various atomization geometries are
used in commercial practice. The process is governed by a number of interrelated operating
parameters. Controllable variables include jet distance, jet pressure, nozzle geometry,
velocity of gas and metal, metal superheat, angle of impingement, metal surface tension
and the melting range in case of alloy powder production [Lawley 1978].
Figure 2.3 Schematic representation of a gas atomization unit ([German 1984]).
37
In gas atomization, the nozzle geometry is a critical aspect to govern particle size,
shape, distribution and yield of the powder [Dunkley 1986]. The solidification rate in gas
atomization depends on the partic
Co
GAP were prepared by high pressure Ar ga
5
m
pure elem
le size, with higher rates (104-106 K/s) associated with
smaller particle sizes (~20-40 μm) [Lawley 1977], as well as on the type of the atomization
medium. Higher solidification rates are achieved with smaller particle sizes and lighter
gases. Additionally, it is possible to increase the solidification rate by cooling the
atomizing gas, or even by pumping additional pre-cooled gas from the bottom of the
chamber. If sufficient superheating is provided to the alloy and the atmosphere being
neutral, the final powder product is spherical. The range of powder sizes is broad with the
mean particle diameter of around 100 μm, even though a mean diameter of 12 to 15 μm
has been also reported [Lawley 1977]. Typically, the yield of powders in gas atomization
can go up to as high as 80% of the starting material. Typical solidification rates achievable
are 102- 106 K/s, especially for powder particles smaller than 30 μm [Lawley 1977].
In this work, Al-based gas-atomized powders (GAP) with different compositions
have been obtained from different sources. The Al Gd Ni Co and Al Y Ni84 6 7 3 90.4 4.3 4.4 0.9
s atomization at the Materials Processing
Center, Ames Laboratory, Ames (USA). This was achieved by mixing the appropriate
weights of each chemical constituent in the form of small lumps, for obtaining the desired
composition and heating them in a graphite crucible. The elements used had purities
ranging from 99.9% to 99.999%. Prior to heating, the lumps were mechanically cleaned to
remove any possible surface oxide layer. Subsequently, the melt was cast in a water-chilled
Cu mold. The powders were then produced by high pressure Ar gas atomization using a
close coupled annular nozzle having a melt delivery inner diameter of 3.2 mm. Following
atomization, the powders were screened using sieves to a size below 106 μm. The particle
size analysis reveals that the median size (X50) of the as-atomized powders Al84Gd6Ni7Co3
and Al90.4Y4.3Ni4.4Co0.9 are 23 µm and 21 µm respectively.
The Al87Ni8La GAP were prepared using the “Nanoval” process [Gerking 1993] at
NANOVAL GmbH & Co. KG, Berlin (Germany). The master alloy was prepared fro
ents Al (99.999%), Ni (99.95%) and La (99.5%) by induction melting in a high
frequency electromagnetic levitation furnace under a purified argon atmosphere. The
levitated melt was kept at high temperatures until the surface oxide skin broke up. After 2
to 7 min of holding, the high frequency current was switched off and the levitated melt fell
down into a water chilled copper crucible, where it solidified forming a crystalline ingot.
This procedure was repeated several times to ensure the homogeneity of the alloy. The
38
melt of the Al87Ni8La5 alloy was then heated up to approximately 1350 K, and was
subsequently gas-atomized using argon gas. The gas-pressure was controlled manually and
could reach a maximum value of 30 bar. Atomization yielded two powder fractions: a
coarser major powder fraction collected in the atomizer recipient and a finer minor powder
fraction collected in the cyclone. A slow oxidation of the as-atomized powder was allowed
by flooding the atomizer with air.
2.2 Powder consolidation
methods, such as hot pressing, hot extrusion and spark
plasma sintering were used in this work to consolidate the powders produced by milling
.2.1 Hot pressing and extrusion
Uni-axial hot pressing is one of the simplest powder consolidation processes where
to the powder placed in a die–punch setup at high
temper
ity. An electro-hydraulic universal axial pressing
machin
Different consolidation
and gas atomization processes into highly dense bulk samples.
2
high pressure is applied uniaxially
ature to induce particle sintering.
The use of elevated temperatures and pressures along with long dwell times allows
densities of > 95% of the theoretical dens
e made by WEBER PWV 30 EDS, Germany, with a capacity of 350 kN maximum
load was used to consolidate the powders by hot pressing. Before starting hot pressing all
the parts such as compaction die and punches are cleaned and sprayed with a thin layer of
boron nitride for lubrication purpose. The whole setup is placed inside the closed chamber.
Approximately 2 to 3 g of powder is placed in the 10 mm diameter of die. The temperature
was measured by a thermocouple of Pr/Rh Pt which was fixed in a dedicated cavity within
the die, ensuring a continuous monitoring of the operating temperature. The chamber is
evacuated to 1×10-3 Pa before starting hot pressing for degassing and to minimize the
oxidation during hot pressing. Careful selection of pressure, temperature and dwell time
has been chosen according to the desired microstructure after hot pressing. Desired
pressure (typically 500 MPa) is applied, then die and punch setup is heated to the desired
temperature with an inductive coil. The compaction temperature and dwell time was
chosen from the thermal studies performed using differential scanning calorimeter. After
39
finishing the heating cycle the chamber is left in vacuum to cool down and Argon was
purged to remove the sample from the chamber. Full density is usually not achieved, and 2
to 4% porosity remains in consolidated sample. For this reason hot-pressed samples are
often hot extruded to obtain highly dense samples.
Hot extrusion is a process by which the billet or pre-compacted samples are forced
through the die cavity to obtain long straight samples by applying high pressure (about
500 M
2.2.2 Spark plasma sintering
Spark Plasma Sintering (SPS) is also known as pressure-assisted resistance or
This section focuses on the production of Al85Y8Ni5Co2 glassy powder by
mechanical alloying. The Al85Y8Ni5Co2 alloy has been chosen because of its relatively
good glass-forming ability among the different Al-based glass-forming systems. The
presence of the rare earth element Y and transition element Ni in Al85Y8Ni5Co2 increases
the glass-forming ability of the alloy [Zhang 2007]. In addition, nickel has, among all the
transition metals, the largest negative heat of mixing with aluminum (-22 kJ/mol) [Zhang
2007], which may assist the glass-forming ability of the alloy. Furthermore, addition of
cobalt improves the mechanical properties of the alloy, namely hardness and strength
[Zhang 2007].
Börner et al. [Börner 2001] successfully obtained complete amorphization of the
Al85Y8Ni5Co2 alloy by mechanical alloying and no indications of remaining crystalline
phases were found by X-ray diffraction and TEM. Complete amorphization can be
achieved in this system by employing a systematic variation and optimization of the
milling parameters, i.e. using proper milling conditions such as interval milling at a low
intensity corresponding to a rather low kinetic energy [Börner 2001, Suryanarayana 2001].
Similarly, in this work mechanical alloying of the Al85Y8Ni5Co2 alloy was carried out at
47
low milling speed as a sequence of milling intervals of 15 minutes interrupted by breaks of
15 minutes to avoid a strong temperature rise.
Figure 3.1 XRD patterns (Co-Kα radiation) for the mechanically alloyed Al85Y8Ni5Co2 powder: as-milled, heated to 653 K and heated to 748 K.
Figure 3.1 shows the XRD pattern of the Al85Y8Ni5Co2 powder mechanically
milled for 500 h. The pattern shows the presence of a broad diffuse halo in the range 40 −
50° (2θ), characteristic for amorphous material. Superimposed are broad and low intensity
diffraction peaks, which indicate the presence of a small amount of a crystalline phase,
most likely the Al3Y (space group R-3m).
Figure 3.2 DSC scan (heating rate 20 K/min) for the mechanically alloyed Al85Y8Ni5Co2 powder.
48
Figure 3.2 presents the constant-rate heating DSC scan (20 K/min) of the
mechanically milled powder heated to 823 K. The DSC curve is characterized by the
presence of two sharp exothermic events with onset temperatures Tx1 = 613 K and Tx2 =
669 K, followed at higher temperature by a broad exothermic peak at about 740 K. No
distinct glass transition can be observed.
In order to study the structural evolution during heating, samples of the
Al85Y8Ni5Co2 powder were annealed in the DSC by continuous heating at 20 K/min up to
different temperatures and then were cooled to room temperature at 100 K/min. The phases
formed were identified by X-ray diffraction and their patterns are shown in Figure 3.1. The
XRD pattern after heating up to the completion of the first exothermic DSC peak (653 K)
reveals the diffraction peaks from fcc Al (space group Fm-3m). An overlapping broad
diffraction peak at about 43° (2θ) can also be observed, which reveals the presence of a
residual glassy phase after the first exothermic event. When the sample is heated to 748 K,
i.e. above the second exothermic event, the XRD pattern displays the presence of
diffraction peaks from fcc Al along with the peaks from the intermetallic phases Al3Y,
Al4Ni3 (space group Ia-3d) and Al9Co2 (space group P21/a).
Although the DSC scan in Figure 3.2 shows no clear glass transition prior to
crystallization, glass transition may, nonetheless, occur. A better insight into the flow
behavior of the supercooled liquid can be derived from viscosity measurements [Busch
1998, Deledda 2004]. Accordingly, the influence of temperature on the viscosity of the as-
milled powder was studied by parallel plate rheometry (Figure 3.3).
Figure 3.3 Temperature dependence of the viscosity for the mechanically alloyed Al85Y8Ni5Co2 powder (heating rate 20 K/min).
49
The viscosity η was measured from the change of the height of the sample versus
time according to the Stefan’s equation shown in Figure 2.5. As the glass transition
temperature is reached (Tg1 at about 600 K) and the glassy solid transforms into the super-
cooled liquid (SCL), the curve in Figure 3.3 displays a strong viscosity drop. At 625 K
(Tx1) the alloy starts to crystallize, leading to a strong viscosity increase, which indicates
the loss of liquid-like behavior. A second drop of viscosity (Tg2), probably related to the
glass transition of the residual glassy phase, is visible at temperatures corresponding to the
second exothermic DSC peak (650 K), where the residual glass crystallizes into
intermetallic compounds. The values of Tx1 and Tx2 evaluated from the viscosity
measurements (623 and 670 K) are in good agreement with the data determined from the
constant-rate heating DSC scans (613 and 669 K) given the different instruments used for
the experiments. This promises that by utilizing the viscous flow of the supercooled liquid
(SCL) above the glass transition temperature, bulk amorphous samples can be produced by
consolidation of the MA powder.
In order to understand the mechanism underlying the first crystallization event,
Johnson-Mehl-Avrami (JMA) analysis was done by carrying isothermal DSC
measurements at different annealing temperatures at 583, 588 and 593 K. The
corresponding DSC curves are shown in Figure 3.4.
Figure 3.4 Isothermal DSC scans for the mechanically alloyed Al85Y8Ni5Co2 powder performed at different annealing temperatures.
All the DSC traces show a single exothermic peak with a symmetric bell shape
after a certain incubation period (t0) which decreases with increasing the annealing
50
temperature. A prepeak can be observed during the incubation period. This might be
related to the growth of pre-existed nuclei. The presence of the exothermic peak suggests a
crystallization mechanism consisting of nucleation from the supercooled liquid rather than
a simple grain growth of already present nuclei. In fact, as reported by Chen et al. [Chen
1988], the isothermal calorimetric signal for a nucleation and growth process is an
exothermic peak with a maximum at a non-zero time whereas a grain growth process is
characterized by a monotonically decreasing signal.
As already mentioned in section 1.2.3, the crystallized volume fraction is directly
proportional to the fractional area under the exothermic peak [Málek 2000, Venkataraman
2007b]. Figure 3.5(a) shows the typical sigmoidal curves for different annealing
temperatures derived from Figure 3.4, which represent the crystallized volume fraction (X)
as a function of the annealing time. The curves become steeper with increasing annealing
temperature, indicating that the transformation proceeds faster as the temperature is
increased.
Figure 3.5 (a) Crystallized volume fraction (X) versus time (t) for the isothermal annealing of the mechanically alloyed Al85Y8Ni5Co2 powder and (b) corresponding Avrami plots (0,15 < X < 0.85) for different annealing temperatures.
According to the Johnson-Mehl-Avrami equation (equation 1.11), JMA plots have
been obtained by plotting ln[ln (1/(1-X))] versus ln(t-t0) for the different annealing
temperatures. Figure 3.5(b) shows the JMA plots for the Al85Y8Ni5Co2 powder. The
transformation range under consideration is 15 – 85 vol.% of crystallized material (0.15 <
X < 0.85). The data do not lie on a straight line and, instead, describe curves with a slope
(i.e. the Avrami exponent n) that is continuously changing in the entire transformation
range considered. Such a variation of n does not allow the use of the JMA analysis for the
51
modeling of the crystallization behavior of the Al85Y8Ni5Co2 powder. However, it suggests
that different mechanism may operate during the devitrification process.
Although amorphization by mechanical alloying was successfully achieved for the
Al85Y8Ni5Co2 powder, two main disadvantages preclude the use of this alloy for the
consolidation into bulk specimens. The amorphization of the Al85Y8Ni5Co2 powder by MA
requires an extremely long milling time (~ 500 h), which makes this processing step not
satisfactory, slowing down the entire production process. In addition, due to such a high Al
content, the powder yield is very low, generally not exceeding 5%. This is due to the soft
aluminum, which sticks to the walls of the vials and on the surface of the balls during
milling. Also, an important effect related to the preferential sticking of the pure Al is the
possible change in the chemical composition of the final alloy [Zhang 2007].
Alternatively to mechanical alloying of elemental powder mixtures, Al85Y8Ni5Co2
glassy powders can be produced by controlled milling of melt-spun glassy ribbons.
Accordingly, in the next paragraph results concerning the production and characterization
of Al-based glassy powders by milling of Al85Y8Ni5Co2 ribbons is presented. The
pulverization of the melt-spun ribbons was achieved by using proper milling conditions, i.e.
interval-milling at low intensity, corresponding to a rather low kinetic energy, and
performed at cryogenic temperature in order to retain their glassy structure and to avoid
sticking of the material to the milling tools due to the high ductility of the ribbons.
3.2 Al85Y8Ni5Co2 glassy powders by milling of melt spun ribbons
The constant-rate heating DSC scan (20 K/min) of the as-spun Al85Y8Ni5Co2
ribbon is shown in Figure 3.6. The curve is characterized by a distinct glass transition (Tg)
before three exothermic heat flow events indicating the transformation from the solid-state
glass into the supercooled liquid, followed by the supercooled liquid (SCL) region (ΔTx =
Tx1 - Tg) which is about 22 K.
These three exothermic heat flow events occur due to crystallization during heating.
The onsets of Tg and of the crystallization events (Tx1, Tx2 and Tx3) for as-spun ribbon are
538, 560, 602 and 656 K, respectively. The enthalpies of crystallization of as-spun ribbon
related to the exothermic DSC peaks are ΔH1 = 32.1 J/g, ΔH2 = 36.4 J/g and ΔH3 = 47.7 J/g.
52
Figure 3.6 DSC scans (20 K/min ) for the as-spun Al85Y8Ni5Co2 ribbon and ribbon ball milled for 5 h.
The DSC scan of the Al85Y8Ni5Co2 ribbon is rather different with respect to the MA
powder with the same nominal composition (Figure 3.2). This is most likely due to a
change in the chemical composition in the MA process, resulting from the preferential
sticking of Al during milling.
Figure 3.7 XRD patterns (Cu-Kα) for the as-spun Al85Y8Ni5Co2 glassy ribbon and the ribbon heated at 20 K/min to completion of the first (580 K ) and second (630 K ) crystallization events.
The structural evolution during heating of as-spun ribbon is shown in Figure 3.7.
The XRD pattern of the as-spun ribbons shows the typical broad maxima characteristic for
amorphous materials together with a broad diffraction peak at about 2θ 44°, most likely ≈
53
due to the formation of a small amount of fcc Al during melt spinning. When the sample is
heated to 580 K, i.e. above the first crystallization peak, the XRD pattern displays the
formation of fcc Al (space group Fm3m) [Villars 1985]. Additionally, an overlapping
broad diffuse maximum due to the residual amorphous phase can be observed. XRD
pattern after heating up to the completion of the second exothermic DSC peak (630 K)
leads to formation of crystalline peaks the formation of the Al3Y phase (space group R-3m)
[Villars 1985].
The melt-spun ribbons were then ball-milled to produce glassy powders. The
pulverization of the ribbons was performed at cryogenic temperature (about 77 K) in order
to retain their glassy structure and to avoid sticking of the material to the milling tools
[Calin 2004]. Due to the low milling temperature, the melt-spun ribbons can be easily
pulverized. The yield of the powder obtained by this method is rather high, exceeding 90%.
Figure 3.8 XRD patterns (Cu-Kα radiation) for the as-spun Al85Y8Ni5Co2 glassy ribbon and ribbon ball milled for 5 h.
The effect of milling on the thermal stability and the microstructure of the melt-
spun Al85Y8Ni5Co2 ribbon are shown in Figure 3.6 and Figure 3.8, respectively. Milling for
5 h does not change the multi-step crystallization behavior characterizing the as-spun
ribbon. The values of Tg, Tx1, Tx2,, and Tx3 for the milled ribbon are 543, 562, 604 and
653 K and are, therefore, only slightly changed with respect to the as-spun ribbon. The
enthalpies of crystallization were found to be ΔH1 = 30.4 J/g, ΔH2 = 27.2 J/g and ΔH3 =
41.9 J/g.
54
This indicates that the first crystallization event is marginally affected, whereas the
subsequent events are much more influenced by the milling treatment. The mechanical
deformation does not induce crystallization of the glass, as illustrated by the XRD pattern
of the ribbon milled for 5 h (Figure 3.8), which, besides the broad diffraction peak already
observed in the as-spun ribbon, does not show additional crystalline precipitates. These
results indicate that Al85Y8Ni5Co2 glassy powders displaying strikingly similar structure
and crystallization behavior in comparison to the parent as-spun sample can be produced
by pulverization of glassy precursors by carefully controlling the milling conditions.
The thermal stability investigations of the milled powders reveal a distinct glass
transition followed by a supercooled liquid region. In this region the powders exhibit a
deformation regime characterized by a viscous flow behavior [Eckert 1997] that may allow
the production of bulk samples by hot consolidation at temperatures within the range of the
supercooled liquid region [Eckert 1997]. The viscosity of milled ribbons is shown in
Figure 3.9 evaluated by parallel plate rheometry according to the Stefan’s equation shown
in Figure 2.5.
Figure 3.9 Temperature dependence (heating rate 20 K/min) of the viscosity of the supercooled liquid for the single-phase Al85Y8Ni5Co2 glassy ribbon milled for 5 h.
The curve displays a decrease of viscosity with increasing temperature from about
1.4 x 1010 Pa s at about 400 K to 2.1 x 109 Pa s at 520 K, most likely due to structure
relaxation. As the glass transition temperature is reached and the glassy solid transforms
into the SCL (above 520 K) the viscosity displays a stronger decrease to the minimum
55
value of 2.5 x 108 Pa s at 565 K. At about 565 K the crystallization sets in and the viscosity
abruptly increases with increasing temperature, indicating the loss of liquid-like behavior.
The Al85Y8Ni5Co2 glassy powders from the milling of melt-spun ribbons were then
consolidated by hot pressing (HP) into cylindrical samples of 10 mm diameter and
5–10 mm length. The viscosity data together with the results from DSC were used to select
the proper consolidation parameters. Hot pressing was performed at 550 K using a pressure
of about 500 MPa. No extrusion of the hot-pressed specimens was possible due to the
insufficient viscous flow. However, the bulk HP samples are characterized by a density of
3.162 g/cm3, which is 96.7% of the density of the starting cylindrical rods of the pre-alloy.
The Al85Y8Ni5Co2 hot-pressed bulk specimens produced from milled ribbons are thus
characterized by a relatively high porosity, as shown by the SEM image in Figure 3.10.
Figure 3.10 SEM image of the hot-pressed single phase Al85Y8Ni5Co2 glassy ribbon.
In order to reach a higher density and, consequently, to improve the ductility of the
samples, the Al85Y8Ni5Co2 glassy powders from milled ribbons were blended with 50 and
70 vol.% of Al in order to produce glass-reinforced metal matrix composites (MMCs). To
obtain a homogeneous dispersion of the glass reinforcement, the blended powders were
milled for 10 minutes and then consolidated by hot pressing followed by hot extrusion.
Consolidation was performed at 520 K, in order to take advantage of the viscosity drop in
the supercooled liquid regime, using a pressure of 500 MPa. By using these consolidation
parameters, extrusion was performed in 10 minutes, a sufficiently short consolidation time
in order to avoid crystallization of Al-based glassy powders [Calin 2004]. For comparison
purposes, a bulk specimen was produced by extrusion of pure Al powders using the same
consolidation parameters as used for the MMCs.
56
Figure 3.11 DSC scans (20 K/min) for the as-spun Al85Y8Ni5Co2 ribbon, ribbon ball milled for 5 h and composite samples with 50 and 30 vol.% of glass reinforcement.
The DSC scans of the MMCs with 50 and 30 vol.% of glassy phase are shown in
Figure 3.11. The presence of pure Al does not change the overall crystallization behavior.
In fact, the values of Tg, Tx1, Tx2, and Tx3 were found to be 269, 286, 329 and 377 K for the
sample with 50 vol.% of glass and 272, 287, 330 and 378 K for the sample reinforced by
30 vol.% of glassy phase. These values are remarkably similar to the as-spun as well as to
the milled ribbons. The enthalpies of crystallization are ΔH1 = 13.6 J/g, ΔH2 = 11.5 J/g and
ΔH3 = 18.1 J/g for the sample with 50 vol.% glass and ΔH1 = 9.9 J/g, ΔH2 = 8.1 J/g and
ΔH3 = 12.7 J/g for the sample containing 30 vol.% of glassy phase, which, after
normalization by the vol.% of glass reinforcements, gives similar values with respect to the
single-phase milled ribbons.
As a typical example of the structure of the consolidated composites, Figure 3.12
shows the XRD pattern of the MMC with 50 vol.% of Al85Y8Ni5Co2 milled ribbon. The
pattern is characterized by few narrow diffraction peaks belonging to the fcc Al phase
together with the broad maximum belonging to the glassy phase at about 2θ ≈ 39°. This
indicates that no crystallization of the glass occurred during consolidation of the
composites.
57
Figure 3.12 XRD pattern (Cu-Kα radiation) for the hot-pressed and hot extruded composite with 50 vol. % of Al85Y8Ni5Co2 glass reinforcement
Figure 3.13 Relative density of the consolidated samples as a function of the volume of glass reinforcement.
Figure 3.13 shows the relative density of the consolidated samples as a function of
the volume fraction of glass reinforcement. The relative density of the specimens with
respect to the density of the starting materials used for the melt spinning experiments (i.e.,
the cylindrical rods of crystalline intermetallic compounds) decreases from 99.2% for the
58
sample with 30 vol.% of glass reinforcement to 97.4% for the specimen with 50 vol.% of
glassy phase and, finally, to 96.7% for the single-phase glass (100 vol.%).
A similar behavior was reported for Al-based MMCs reinforced with SiC particles
[Chung 1999, Slipenyuk 2006] and was ascribed to clustering of the reinforcing particles
[Slipenyuk 2006]. Figures. 3.14(a) and 3.14(b) show SEM images of the composites with
50 and 30 vol.% glass reinforcement, respectively. The images display a homogeneous
distribution of flake-shaped particles (the glassy phase) dispersed in the fcc Al matrix. No
porosity is visible, further corroborating the high density of the samples.
Figure 3.14 SEM micrographs for the hot extruded composites with (a) 50 vol.% and(b) 30 vol.% glass reinforcement.
On the other hand, the SEM image of the single-phase Al85Y8Ni5Co2 glassy
specimen produced by hot pressing of the milled ribbons (Figure 3.10) displays a large
number of pores. This indicates that incomplete bonding between the particles has
occurred during consolidation of the single-phase glass, leading to a rather poor
densification of the material.
Figure 3.15 shows a typical room temperature uni-axial compression true stress-
strain curves under quasistatic loading for the single-phase glass produced by HP of the
milled ribbons and for the composite materials along with the curve for the hot extruded
pure Al. The HP specimen of the single-phase glass exhibits an elastic regime of 1.45% up
to a stress of about 400 MPa but no appreciable ductility, most likely due to the residual
porosity of the sample that may initiate cracks leading to the early failure of the material.
The observed fracture strains exceed 40% for all the composite materials. However, due to
the strong softening characterizing the composite specimens after reaching the compressive
strength (the maximum compressive stress which the material is capable of sustaining
59
[ASTM 2004], the compression tests shown in Figure 3.15 were stopped after reaching the
maximum stress and before fracture occurrence.
Figure 3.15 Room temperature compression stress-strain curves (strain rate of 1x10-4 /sec) for the hot-pressed and hot extruded pure Al, hot pressed and hot extruded composite with 30 vol.% of Al85Y8Ni5Co2 glass reinforcement, hot-pressed and hot extruded composite with 50 vol.% glass reinforcement and hot-pressed single-phase Al85Y8Ni5Co2 glassy powder.
Pure Al exhibits an elastic regime of 0.2% before yielding, which occurs at about
75 MPa. After yielding the stress increases with increasing strain and the sample exhibits
work-hardening up to the compressive strength of 155 MPa, reaching a strain at maximum
stress of about 25%. The mechanical properties of pure Al are remarkably increased by the
addition of the glass reinforcement. The specimen containing 30 vol.% glass displays an
elastic regime of 0.3% and a yield strength of 120 MPa. The compressive strength is raised
to 255 MPa while retaining a strain at maximum stress of about 10%. When the amount of
glassy phase is further increased to 50 vol.% the elastic range is still 0.3% while the yield
and compressive strength are further increase to 130 and 295 MPa, respectively, and the
strain at maximum stress is found to be about 7%.
The prediction of the overall mechanical properties of a composite from the
properties of the single constituents is an important prerequisite for the material design and
application. Among the different methods for estimating the mechanical properties of a
composite, the rule of mixtures (ROM) is the simplest and most intuitive [Kim 2000, Kim
2001]. The ROM considers the properties of the composite as volume-weighted averages
of the components properties and assumes that the components are non-interacting during
60
deformation (Behavior of glass in the sintering of metal-glass materials). This approach
has been extensively used to model the mechanical properties of fiber-reinforced matrix
composites [Chawla 1998, Kelly 1972]. Two ROM methods have been widely employed
to predict the mechanical properties of composites [Kelly 1972, Kim 2000]: (i) the Voigt
model, based on the equal strain assumption, and (ii) the Reuss model, based on the equal
stress assumption. Although these models have been derived for the elastic properties of
composites, they have been also used for the overall plastic regime [Bruck 1999, Kim 2000,
Louzguine 2002b, Mileiko 1969].
The Voigt or iso-strain model assumes that the two components, matrix and
reinforcement, experience the same strain during deformation [Chawla 1998]. For the
stress of the glass-Al matrix composite [Kim 1999], this can be written as:
σc = Vgl·σgl + VAl·σAl , (3.1)
where V is the volume fraction, σ is the strength and the subscripts c, gl and Al indicate the
composite, the glass reinforcement and the Al matrix, respectively. It is often observed that
the strength of a composite is lower than predicted by the Voigt model [Sarkar 1982]. This
is generally attributed to (i) inadequate bonding, (ii) porosity and (iii) inherent material
defects, e.g. cracks [Sarkar 1982]. Therefore, the iso-strain treatment represents the upper
bound.
The lower bound is given by the Reuss or iso-stress model, which assumes that the
composite exhibits equal stress in the two components [Chawla 1998]. For the stress, this
can be written as:
σc = 1−
⎟⎟⎠
⎞⎜⎜⎝
⎛+
Al
Al
gl
gl VVσσ
. (3.2)
The effect of the porosity on the mechanical properties of a composite can be taken
into account by considering the volume of the composite material to be made up of three
different volumetric components, i.e. reinforcement, matrix and porosity [Madsen 2003]:
Vc = Vgl + VAl·+ Vp , (3.3)
where the subscript p denotes the porosity.
61
Figure 3.16 (a) Compressive strength and (b) strain at maximum stress (evaluated from Figure 3.15) as a function of volume percent of glass reinforcement for the samples: (■) hot extruded, (□) hot-pressed, (○)single-phase Al85Y8Ni5Co2 melt-spun glassy ribbon from reference [Louzguine 2002b], and (∇) calculated from Equation. 3.2.
The values of the maximum stress and the strain at maximum stress as a function of
the amount of the Al85Y8Ni5Co2 glass reinforcement is shown in Figures. 3.16(a) and
3.16(b) together with the values of the glassy ribbon [Louzguine 2002b] and of the single-
phase glass consolidated by hot pressing (present work). The strength of the samples
strongly deviates from the Voigt model (dotted line) and, instead, can be fitted well by
using the Reuss model (corrected for porosity by equation (3.3)) (dashed line). This
behavior indicates that the compressive strength obeys the iso-stress model. The significant
difference in strength observed between the single-phase Al85Y8Ni5Co2 glass consolidated
by hot pressing (400 MPa) and the melt-spun glassy ribbon with the same composition
(1250 MPa) [Louzguine 2002b] cannot be exclusively ascribed to the effect of porosity. In
fact, the maximum stress of a melt-spun ribbon with the density of the HP glass calculated
by equation (3.2) should exceed 1000 MPa (open triangle in Figure 3.16(a)). Most likely,
the considerably low strength of the HP sample is due to inadequate bonding between the
particles.
Figure 3.16(b) shows the strain at maximum stress for the different composite
materials. No values for the strain are available for a fully dense glassy specimen (such as
for the melt-spun glassy ribbon in Figure 3.16(a)). Therefore, the strain value of the low-
density single-phase Al85Y8Ni5Co2 glass consolidated by HP was used in the data fitting.
Similarly, to the maximum stress values in Figure 3.17(a), the corresponding strain at
62
maximum stress can be well fitted using the iso-stress Reuss’s model. Although lower than
predicted by the Reuss model, the strain of the HP specimen follows the iso-stress
treatment within the experimental error. This indicates that the poor particle bonding is less
significant in affecting the strain of the single-phase glass.
The validity of the Reuss model for the description of the composites studied in the
present work is justified by the following considerations. The milled ribbons used as
reinforcements are in the form of flake-shaped particles. Therefore, they can be treated as
short fibers as a first approximation. Although not aligned along the same direction, they
tend to lie on the same plane (see Figures. 3.16(a) and 3.16(b)), which is normal to the
applied stress during the compression test. Therefore, such composites can be considered
as a random fiber array in a matrix deformed perpendicularly to the fiber direction, the
latter being a requirement for the application of the iso-stress model [Chawla 1998]. In
addition, it has been reported that the Voigt model fits the data well for a high volume
fraction of reinforcement [Kim 2001]. In this case, the deformation affects the
reinforcements as a consequence of the smaller distance between the particles/fibers [Kim
2001]. On the other hand, the Reuss model works well for small volume fractions and
longer distances between the reinforcements, where the deformation mainly occurs in the
soft matrix [Kim 2001]. In the current work, the amount of reinforcements is relatively low
(≤ 50 vol.%) and the distance between the particles is large (> 50 μm). Therefore, most
likely the plastic deformation occurs mainly in the matrix and the mechanical properties
obey the iso-stress model.
63
Chapter 4: Synthesis and characterization of high-strength Al-
based alloys by consolidation of gas-atomized powders
The results shown in Chapter 4 clearly demonstrate that powder metallurgy
methods are suitable for the production of Al-based glassy and glass-composite materials
with high strength combined with considerable ductility. However, the production of Al-
based glassy powders by mechanical alloying of pure elements yields poor output, which
limits the use of MA for the production of Al-based powders with high Al contents. In
order to circumvent this limitation, Al-based amorphous powders can be produced by
pulverization of the melt-spun glassy ribbons. Although this technique allows the
production of larger quantities of high quality material (i.e. constant composition and
microstructure combined with low contamination levels) compared to MA, the procedure
is rather complex and requires the detailed characterization of every single melt-spun
ribbon. On the other hand, gas atomization offers the possibility to easily produce large
quantities of powders with homogeneous properties (e.g. structure and thermal stability)
along with a uniform size distribution of particles. Accordingly, in this chapter amorphous
and partially amorphous Al-based gas-atomized powders (GAP) with different
compositions have been used as precursors for the production of high-strength bulk
samples. The chapter is arranged in three sections. The first section deals with the detailed
study of Al84Gd6Ni7Co3 powders and consolidated samples with special emphasis given to
the crystallization behavior of the powders, their consolidation into bulk specimens and the
corresponding mechanical properties. The second section focuses on the consolidation and
mechanical behavior of Al87Ni8La5 produced by SPS. Here, a detailed study has been
carried out to understand the remarkable deformation behavior of the bulk samples. Finally,
the third part shows the characterization, synthesis and mechanical properties of the
Al90.4Y4.3Ni4.4Co0.9 alloy, which displays promising tensile properties comparable to
commercial high-strength Al-based alloys.
4.1 Gas-atomized Al84Gd6Ni7Co3 powder
Structural and thermal characterization. Figure 4.1.1 shows typical SEM
micrographs of the as-atomized Al84Gd6Ni7Co3 powder. The morphology of the powder is
smooth and round. Small-sized (< 1 µm) satellites are found around large particles (Figure
64
4.1.1(b)). Due to turbulent atomization conditions, both coarse and fine particles, arising
from efficient secondary breakup of particles during atomization. During in flight
collisions, the interaction results in welding of finer size particles to larger particles
[Özbilen 1999]. Smaller size powder particles with dimensions below 10 µm do not show
any feature, whereas larger particles display the presence of precipitates, suggesting that
partial crystallization occurred during gas atomization.
Figure 4.1.1 (a) and (b) SEM micrographs of the as-atomized Al84Gd6Ni7Co3 powder.
The XRD pattern of as-atomized Al84Gd6Ni7Co3 powder (Figure 4.1.2) exhibits a
broad maximum at 2θ angles between 35 and 55° which is typical for glassy materials
along with tiny Bragg peaks due to the presence of small amounts of fcc Al and
orthorhombic Al19Gd3Ni5 (space group Cmcm) [Gladyshevskii 1992] phases.
Figure 4.1.2 XRD patterns (Co Kα) for the gas-atomized Al84Gd6Ni7Co3 powder, in the as-atomized state and after heating to 633 and 873 K.
65
This might be due to the cooling rate during gas atomization (about 102 - 103 K/s)
[Suryanarayana 1991], which is not sufficiently high to suppress the formation of
crystalline phases, as already observed for gas-atomized Al-Ni-Mm (Mm = misch metal)
and Al-Ni-La powder [Ohtera 1992]. This is in agreement with the SEM micrographs in
Figure 4.1.1, which show the presence of precipitates for particles with larger size.
Figure 4.1.3 (a) Isochronal DSC scans taken at different heating rates and (b) corresponding Kissinger plots for the evaluation of the activation energies related to the first and second crystallization events.
Figure 4.1.3(a) presents the isochronal DSC scan for the Al84Gd6Ni7Co3 glassy
powder as a function of temperature taken at different heating rates (φ). The curves exhibit
a distinct glass transition with onset temperature Tg, followed by the supercooled liquid
region ΔTx before two crystallization events with onset and peak temperatures Tx and TP,
respectively, occur at higher temperature. The values TP at different heating rates are
summarized in Table 4.1.1.
Table 4.1.1. Summary of the results from the isochronal DSC experiments.
The peak temperatures TP1 and TP2 shift towards higher temperatures as the heating
rate is increased from 10 to 80 K/min. The supercooled liquid region ranges between 15
and 20 K, which is similar to what was observed for other Al-based amorphous systems,
such as Al-Y-Ni-Co [Inoue 1998].
66
The XRD pattern (Figure 4.1.2) of the sample annealed up to the completion of the
first exothermic DSC peak (633 K) reveals diffraction peaks from fcc Al and a broad peak
from the Al19Gd3Ni5 intermetallic phase together with the broad maxima of the residual
amorphous phase. The diffraction peaks of fcc Al are remarkably broad, which indicates
that this phase has nm-sized dimensions. A similar crystallization behavior has been
observed for different Al-based amorphous alloys, such as Al-Ni-RE and Al-Nd-Ni-Co
[Gao 2008, Huang 2008, Ye 2000]. When the sample is heated up to 873 K, i.e. far above
the second crystallization peak, the XRD pattern (Figure 4.1.2) shows the diffraction peaks
from fcc Al, Al19Gd3Ni5 and Al9Co2 (space group P21/a) [Douglas 1950] intermetallic
phases. No amorphous phase is visible at this stage, indicating that complete crystallization
of the glass occurs during the heat treatment.
Figure 4.1.4 Viscosity curve of as-atomized the Al84Gd6Ni7Co3 powder at a heating rate 10 K/min.
10 K/min) of the as-atomized powder. At the glass transition temperature (~550 K), where
the gla
scans taken at different heating rates using the
Kissing
Figure 4.1.4 shows the temperature dependence of the viscosity (heating rate
ssy solid transforms into the SCL, the curve displays a strong viscosity drop. At 590
K crystallization sets in and the viscosity abruptly increases with increasing temperature,
indicating the loss of liquid-like behavior.
The activation energy (Ea) for the crystallization process in Figure 4.1.3(a) was
evaluated from constant-rate heating DSC
er method (equation 1.10). By plotting ln(φ/T2P) versus (1/Tp), a straight line with
slope Ea / R is obtained (Figure 4.1.3(b)). The activation energies corresponding to the
67
crystallization events are Ea1 = 283 ± 2 kJ/mol and Ea2 = 210 ± 4 kJ/mol. These values are
comparable to the values attributed to self diffusion of aluminum [Volin 1968] suggesting
that the nucleation is diffusion controlled process.
In order to understand the mechanism underlying the first crystallization event,
isothermal DSC measurements were carried out at different annealing temperatures
ranging from 568 to 578 K and the corresponding curves are shown in Figure 4.1.5(a). All
curves show a single exothermic peak with an almost symmetric bell shape and an
incubation time (τ) that decreases with increasing annealing temperature. The inset in
Figure 4.1.5(a) shows the typical sigmoidal curves for different annealing temperatures
derived from time Figure 4.1.5(a), which represent the crystallized volume fraction as a
function of annealing time (see section 1.2.3). The curves become steeper with increasing
annealing temperature, indicating that the transformation proceeds faster as the annealing
temperature is increased.
Figure 4.1.5 (a) Isothermal DSC scans taken at different annealing temperatures and (inset) crystallized volume fraction (X) vs. time (t) and (b) Avrami plots (in the range 0.10 < X0.85) calculated from the isothermal DSC scans in Figure 4.1.4(a).
ing to the analysis
describ
<
Figure 4.1.5(b) shows the JMA plots for the Al84Gd6Ni7Co3 as-atomized powder in
the transformation range 10 – 85 vol.% (0.10 < X < 0.85) accord
ed in section 1.2.3. The Avrami exponent decreases from 3.0 ± 0.07 for isothermal
annealing at Ta = 568 K to 1.6 ± 0.04 for Ta = 578 K. Values of n of about 3.0 may be
related to a transformation mechanism characterized by diffusion controlled three
dimensional growth and increasing nucleation rate, whereas n = 1.6 suggests almost zero
nucleation rate [Avrami 1939, 1940, 1941]. This behavior can be understood by
considering the phase formation during crystallization. The first crystallization event is
68
characterized by the formation of nanocrystalline fcc Al (Figure 4.1.2), most likely through
a nucleation and growth mechanism, as suggested by the isothermal bell-shape peak in
Figure 4.1.5(a). Primary formation of nanocrystalline fcc Al from the amorphous phase is
characterized by an extremely high nanocrystal density of about 1021 – 1023 m-3 [Perepezko
2007], which may explain the high value of n = 3.0 observed for low annealing
temperatures. However, each fcc Al nanocrystal formed rejects the solute elements Gd, Ni
and Co into the residual amorphous matrix, thus reducing the driving force for the
formation of additional fcc Al and significantly reducing the nucleation rate [Allen 1998],
in accordance with the small value of the Avrami exponent (n = 1.6) observed for high
annealing temperatures.
Figure 4.1.6 Arrhenius plot for the isothermal activation energy of the Al84Gd6Ni7Co3 powder.
calculated from the intercept of the JMA plots (Figure 4.1.5(b)) (see equation 1.11). Under
isother
The reaction rate constant KT is a function of annealing temperature can be
mal conditions, the Arrhenius equation (equation 1.8) is often used to calculate the
activation energy for crystallization of an amorphous alloy [Scott 1977]. Figure 4.1.6
shows the plot of ln(KT) versus (1000/T), which yields a straight line whose slope gives the
activation energy of crystallization. The activation energy calculated by this method is
271 ± 3 kJ/mol. This value is remarkably similar to the activation energy calculated by the
Kissinger method (283 ± 2 kJ/mol).
69
Consolidation and mechanical properties. In order to study the influence of
temperature on densification and microstructural evolution, cylindrical samples (10 mm
diameter and 10 mm length) were produced from gas-atomized Al84Gd6Ni7Co3 powder by
hot pressing at 573, 623, 673 and 723 K with 3 minutes dwelling time. During hot pressing,
the change in the sample dimensions (shrinkage) occurs in the axial direction with
increasing the temperature. The corresponding lateral expansion does not occur due to the
constraints imposed by the wall of the die. The height variation of the powder bed in the z-
direction during hot pressing allows to understand the densification and the active sintering
temperatures [Garay 2010, German 1996]. For the present HP experiments, the
instantaneous variation of the powder bed (corrected for the dimensional changes of the
HP equipment) has been continuously measured. The shrinkage in the sample during HP at
constant pressure is given by Δh/ho, where ho is the initial height and Δh is the change in
height during the process. The shrinkage rate is the first derivative of the shrinkage
⎟⎟⎠
⎞⎜⎛ Δhd
rate versus temperature plots are useful in identifying the most active sintering temperature
[
⎜⎝ ohdt
, and since time and temperature are linearly related during constant rate heating,
this is derived from the shrinkage versus temperature plot. The shrinkage and shrinkage
German 1996].
Figure 4.1.7 Shrinkage data for the hot-pressed Al84Gd6Ni7Co3 samples
The shrinkage and shrinkage rate plots for hot pressing of the Al84Gd6Ni7Co3
powder are shown in Figure 4.1.7. The shrinkage increases sharply with increasing
temperature up to 573 K. For temperatures above 573 K, the slope decreases remarkably
70
and finally, above 650 K, the shrinkage curve displays a plateau, which indicates that full
densification is reached. In the first steps of hot pressing, the densification progresses
through particle rearrangement, particle bonding, necking and neck growth along with
plastic flow. As temperature increases, the shrinkage occurs due to mass transport by grain
boundary, surface and volume diffusion processes [German 1996]. The shrinkage rate
curve reveals that the maximum sintering rate is at about 573 K, which is within the super-
cooled liquid region. This implies that the shrinkage rate is maximum above the glass
transition region, where the sample experiences viscous flow behavior.
The density of the Al84Gd6Ni7Co3 samples hot-pressed at 573, 623, 673 and 723 K
is shown in Figure 4.1.7. The results reveal that the density is directly related to the
shrinkage curve and increases with increasing sintering temperature, reaching a plateau for
sintering at temperatures ≥673 K.
Figure 4.1.8 XRD patterns (Co-Kα radiation) of Al84Gd6Ni7Co3 powder: as-atomized, and
The XRD patterns of the HP samples are shown in Figure 4.1.8. The XRD pattern
of the
bulk sample consolidated by hot pressing at 573, 623, 673 and 723 K.
sample HP at 573 K reveals a broad maximum at 2θ ≈ 43° due to the residual
amorphous phase together with the broad diffraction peaks from fcc Al. Besides the
amorphous maximum and the diffraction peaks from fcc Al, less intense peaks from the
Al19Gd3Ni5 and Al9Co2 phases can be observed in the XRD pattern for the sample HP at
623 K. The XRD patterns for the samples hot-pressed at 673 and 723 K show the
diffraction peaks from fcc Al, Al19Gd3Ni5 and Al9Co2 phases. No broad maximum due to
the amorphous phase is visible at this stage, which indicates complete crystallization of the
71
glass. These diffraction peaks are rather broad, suggesting that the phases formed are of
nano or ultra-fine dimensions. The diffraction peaks are sharper for the samples HP at
723 K than the peaks observed in the samples HP at 673 K, which implies that larger grain
growth occurs at 723 K.
Figure 4.1.9 (a) OM and (b)-(d) SEM micrographs of sample hot-pressed at 573 K.
In order to analyze the effect of the densification on the microstructure of the
Al84Gd
e observed at the particle interfaces along with
particle deformation to an irregular “polygonal” shape due to the viscous flow behavior
6Ni7Co3 specimens, the microstructure of the bulk samples hot-pressed at different
temperatures have been investigated by OM, SEM and TEM and the corresponding
micrographs are shown in Figures 4.1.9 – 4.1.12. Figure 4.1.9 shows the OM and SEM
micrographs of the sample hot-pressed at 573 K, where the highest shrinkage rate occurs
(Figure 4.1.7). The OM micrograph (Figure 4.1.9(a)) reveals that densification occurs at
this sintering temperature. However, a large amount of porosity (≈2%) is clearly visible
(Figure 4.1.9(b)), which indicates that full densification is not achieved, corroborating the
shrinkage results shown in Figure 4.1.7.
Initiation of necking can also b
72
(Figure
sitional contrast
(Figure
compared to the material HP at 573 K. Figures 4.1.10(b) and 4.1.10(c) clearly
display
4.1.9(c)). In addition, a concentration gradient is visible as dark regions formed at
the particle boundaries near the pore regions (Figure 4.1.9(c)). The SEM-EDX results
reveal that these dark areas are rich in aluminum, which is in agreement with the XRD
pattern in Figure 4.1.8 showing the formation of Al for sintering at 573 K. Therefore, the
initial sintering stages discussed in section 1.3.3, i.e. initial bonding between the particles,
necking and concentration gradient, all occurs during sintering at 573 K.
The Al-rich concentration gradient (≈10% more than centre) mainly occurs around
the pores, whereas the necking areas do not show significant compo
4.1.9(d)). This implies that preferential diffusion of Al towards the pores takes
place during sintering. A possible explanation for this behavior might be related to the
evaporation-condensation transport mechanism. Vapor transport during sintering leads to
repositioning of atoms located on the particle surface. Evaporation occurs from a surface
and transport takes place across the pore space, leading to condensation on a nearby
surface. This causes mass transport towards the pore area [German 1996]. This type of
mass transport may be assisted by the viscous flow occurring during hot pressing. The
viscosity of the Al84Gd6Ni7Co3 powder drastically drops at temperatures above the glass
transition (Figure 4.1.4). The viscosity is inversely related to diffusivity [Cahn '96] and,
therefore, the low viscosity of the SCL at 573 K may enhance the diffusivity of Al leading
to an improvement of the mass transport towards the pore space. As a result, in the material
sintered at 573 K, viscous flow combined with evaporation-condensation of Al may lead to
the significant Al-rich concentration gradient around the pore space, as observed in the
Figures 4.1.9(c) and 4.1.9(d). However, these factors are not sufficient to obtain full
densification, as it is evident from the shrinkage curve for hot pressing at 573 K (Figure
4.1.7).
The SEM image of the sample HP at 623 K (Figures 4.1.10(a)) shows reduced
porosity
that necking at the particle interfaces is much more developed as compared to the
sample HP at 573 K (Figure 4.1.9). In addition, the powder particles maintained a
spherical/elliptical shape, which is in contrast to the “polygonal” shape characterizing the
sample HP at 573 K. This is most likely related to the different periods spent by the
samples at temperatures within the SCL region, where the material is characterized by a
viscous flow. While the sample HP at 573 K was kept isothermally within the SCL region
for 3 minutes, the sample HP at 623 K (which is above the minimum viscosity in Figure
73
4.1.4) passed through the SCL region without dwelling time, spending about 30 seconds at
temperatures within the SCL region.
Figure 4.1.10 SEM micrographs of the sample hot-pressed at 623 K.
es and in
the cen
(a)) shows few
small-s
The neck growth visible in Figure 4.1.10 results in particle shape chang
tre-to-center approach of the powder particles [German 1996]. In addition, neck
growth occurs by particle coalescence, as shown in Figure 4.1.10(c). Beside neck growth,
nc/UFG dendrites with size of 50 to 200 nm can also be observed inside the power
particles (Figure 4.1.10(d)). The formation of small dendrites is due to the high processing
temperature (623 K), which causes crystallization of the amorphous phase.
The OM micrograph of samples hot-pressed at 673 K (Figure 4.1.11
ized pores together with bright regions at the particle interface, revealing good
bonding of the particles. The black interface between the particles visible in the SEM
micrographs (Figures. 4.1.11(b)–4.1.11(d)), corresponding to the bright areas in Figure
4.1.11(a), is not due to porosity.
74
Figure 4.1.11 (a) OM and (b)-(d) SEM micrographs of sample hot-pressed at 673 K.
SEM-EDX analysis reveals that the particle interface consists of fcc Al. Pore
rounding occurs at this stage, as shown in Figure 4.1.11(d). In addition to the improved
densification, HP at 673 K induces extended crystallization of the glassy phase, as
demonstrated by the formation of several rod-like nano-sized features, which can be
identified as the Al19Gd3Ni5 and Al9Co2 intermetallic phases observed by XRD (Figure
4.1.8).
The OM image of the sample hot-pressed at 723 K (Figure 4.1.12(a)) shows no
visible porosity. The SEM micrographs (Figures 4.1.12(b) and 4.1.12(c)) reveal that the
particles consist of agglomerates of bright rod-like particles and black regions of nm-scaled
dimensions. The TEM-BF images in Figure 4.1.12(d) shows the triangular area between
particles. TEM-EDX analysis reveals that the triangular area consists of UFG Al of about
200 to 300 nm size. The regions surrounding the triangular area consist of nano-sized Al
grains (black particles) and rod-like nano-sized features rich in Gd, Ni and Co of about
50 nm thickness and 200 nm length, most likely corresponding to the Al19Gd3Ni5 and
Al9Co2 compounds observed by XRD (Figure 4.1.8). This is corroborated by the work of
75
Gao et al. [Gao 2003] who observed a rod-like morphology for the ternary Al19Gd3Ni5
phase.
Figure 4.1.12 (a) OM, (b) and (c) SEM, (d) TEM-BF micrographs of sample hot-pressed at 723 K.
The hardness of the samples HP at 573, 623, 673 and 723 K reveal Vickers
microhardness of about 289 ± 15, 542 ± 10, 535 ± 8, 432 ± 10 HV, respectively. The
lowest hardness values observed in the samples HP at 573 K can be ascribed to the poor
bonding and presence of large pores characterizing this sample. On the other hand, for the
samples HP at higher temperatures, which display good bonding between the particles and
small residual porosity, the hardness values are rather high and show a decreasing trend
with increasing HP temperature. This may be related to the crystallization of the hard
amorphous phase and the formation of softer phases (e.g. Al). However, the hardness of
the present consolidated samples is much higher than for conventional Al-based alloys
(60 – 250 HV [Davis 1993]) due to their nanocrystalline/ultra fine-grained microstructure.
76
Figure. 4.1.13 Room temperature compression true stress-true strain curves for the Al84Gd6Ni7Co3 gas-atomized powder hot-pressed at different temperatures.
Typical room temperature uni-axial compressive true stress-strain curves for the HP
samples are shown in Figure 4.1.13. The samples HP at 623 K fractured in a brittle manner
at about 700 MPa without any yield or plastic strain. This early fracture might be due to the
presence of large pores and inadequate particle bonding, as observed in Figure 4.1.10. The
sample HP at 673 K exhibits a yield stress (0.2% offset) of about 1250 ± 10 MPa followed
by strain hardening up to the maximum stress of 1560 ± 5 MPa where fracture occurs at
3.5 ± 0.2%. The sample HP at 723 K exhibits lower yield and compressive strengths
(1150 ± 10 MPa and 1440 ± 7 MPa, respectively) with respect to the sample HP at 673 K,
however, the fracture strain slightly increases to 4 ± 0.4%. The stress-strain curves for the
samples HP at 673 and 723 K exhibit strain hardening behavior. The decreased values of
yield and compressive strengths for the samples HP at 723 K as compared to the samples
HP at 673 K may be linked to the larger grain size occurring at 723 K, as shown by the
XRD patterns in Figure 4.1.8. This may also explain the larger plastic deformation shown
by the sample HP at 723 K.
The strength levels of the samples HP at 673 and 723 K are three times larger than
for conventional Al-based high strength alloys [Davis 1993]. Such high strength levels are
most likely due to the multi-phase microstructure consisting of a uniform distribution of
high-strength nanocrystalline rod-shaped intermetallic phases and nanocrystalline Al
particles along with areas of deformable ultra fine-grained Al at the triangular areas.
77
Figure 4.1.14 SEM micrographs of fractured samples after compression testing of samples HP at 723 K.
Plastic deformation during compression testing of ultra fine-grained Al at the
triangular areas is corroborated by the fracture surface of the sample HP at 723 K (Figure
4.1.14(a)), revealing the formation of dimples at the triangular areas, indicative of ductile
fracture. On the otherhand, the areas with high density of intermetallic compounds display
brittle fracture characterized by intra-granular rupture as well as decohesion of the particles.
Although the Al regions at the triangular areas do not form a continuous network
throughout the specimen, they may nevertheless allow significant movement of
dislocations. This, together with the obstacle to the dislocation movement due to the
intermetallic phases, may explain the high strength levels combined with a limited but
distinct good plastic deformability of the present HP samples.
78
4.2 Gas-atomized Al87Ni8La5 powder
The results presented in the previous section indicate that consolidation of the
Al84Gd6Ni7Co3 powder into highly-dense bulk samples cannot be achieved without
extended crystallization of the material. Nevertheless, crystallization during consolidation
is not detrimental and leads to bulk samples with a remarkably high strength of about
1550 MPa. However, such a high strength is accompanied by a relatively limited room
temperature plastic deformation of about 3.5 to 4%. To further test the effectiveness of
powder consolidation as a method for the production of Al-based materials with enhanced
mechanical properties through the combined crystallization and consolidation of glassy
precursors, in this section Al87Ni8La5 gas-atomized powders are consolidated by SPS
above their crystallization temperature and their mechanical properties are investigated in
detail.
Figure 4.2.1 XRD pattern (Co-Ka radiation) of Al87Ni8La5 powders: as-atomized, after heating (heating rate 40 K/min) to different temperatures in the DSC, and bulk sample consolidated by SPS at 713 K.
The structure of the as-atomized powder, investigated by XRD, is presented in
Figure 4.2.1. The pattern displays the broad diffuse maxima, characteristic of amorphous
materials, together with few broad diffraction peaks corresponding to fcc Al and Al11La3
(space group - Immm) phases. This implies that the structure of the as-atomized Al87Ni8La5
powder is not completely amorphous, as already observed for the Al84Gd6Ni7Co3 powder
79
(section 4.1, Figure 4.1.2), and further confirms the limited glass-forming ability of Al-
based alloys with high Al content.
Figure 4.2.2 shows the constant-rate heating DSC scan (20 K/min) of the as-
atomized powder. The DSC curve exhibits two exothermic events due to crystallization
with onset temperatures Tx1 = 445 K and Tx2 = 612 K. This type of thermal behavior can be
observed in many Al-based glassy alloys with high Al contents above about 85 at.%, which
crystallize through two stages upon heating to elevated temperatures [Inoue 2001b]. The
enthalpies of crystallization related to the first and second exothermic DSC peaks are ΔH1
= 19 ± 3 J/g and ΔH2 = 77 ± 5 J/g, respectively. These values are very similar to those
reported for Al-Ni-La alloys with similar composition (22 and 80 J/g, respectively) [Ye
2000], suggesting that only a small fraction of material crystallized during gas atomization.
Figure 4.2.2 DSC scan and (inset) temperature dependence of the viscosity for the as-atomized Al87Ni8La5 powder.
Figure 4.2.1 shows the XRD patterns of the Al87Ni8La5 powder obtained after
continuous heating at 20 K/min to 573, 673 and 773 K. When the sample is heated up to
573 K, i.e. above the first crystallization peak, the XRD pattern displays the presence of
fcc Al and the simultaneous decrease of the diffuse amorphous maximum. Heating above
the second exothermic DSC peak (673 and 773 K) leads to the decomposition of the
residual glassy phase into the intermetallic compounds Al11La3 and Al3Ni.
Similarly to the MA Al85Y8Ni5Co2 (section 3.1), no clear glass transition prior to
crystallization can be observed in the DSC scan in Figure 4.2.2. In order to verify the
occurrence of the glass transition, the influence of temperature on the viscosity of the as-
80
atomized powder was studied by parallel plate rheometry (inset in Figure 4.2.2). The
viscosity decreases with increasing temperature from about 3 x 108 Pa s at 400 K to
5 x 106 Pa s at 470 K, which indicates the occurrence of the glass transition and the
transformation of the glassy solid into the supercooled liquid (SCL). At about 470 K
crystallization sets in and the viscosity abruptly increases with increasing temperature,
indicating the loss of liquid-like behavior.A second drop of viscosity is visible at
temperatures corresponding to the second exothermic DSC peak (610 K), where the
residual glass crystallizes into intermetallic compounds (Figure 4.2.1). It has been reported
for Al-based glasses that Tg and Tx increase significantly with increasing solute
concentration [Inoue 1998]. Although for the present Al87Ni8La5 powder the composition
of the residual glassy phase after the first crystallization event is not known, it is most
likely depleted in Al due to the primary formation of fcc Al, as shown in Figure 4.2.1. It is
thus plausible that the residual Al-poor glassy phase undergoes the glass transition at
higher temperature, explaining the second drop of viscosity.
The occurrence of two viscosity drops may be a considerable advantage for the
consolidation of the gas-atomized powder in assisting interparticle bonding and
densification. In order to take benefit of the double viscous flow behavior, during
consolidation by SPS a constant pressure of 500 MPa was applied from room temperature
through the crystallization events to the final sintering temperature (713 K), which was
held for about 5 minutes. The consolidation of the Al87Ni8La5 powder at high temperatures
can thus be considered as a combined in-situ devitrification and densification of the
powders.
The XRD pattern of the Al87Ni8La5 powder sintered at 713 K is also shown in
Figure 4.2.1. The structure consists of fcc Al together with two intermetallic compounds,
i.e. Al11La3 and Al3Ni. No trace of the amorphous phase is visible, indicating that complete
crystallization occurred during sintering. The diffraction peaks in Figure 4.2.1 are rather
broad, indicating that the phases formed are of nanoscale or ultra-fine dimensions. Indeed,
Rietveld structure refinement reveals an average grain size for the different phases ranging
between 100 and 200 nm. This is in agreement with the results reported for the
devitrification of melt-spun Al87Ni8La5 glassy ribbons [Sahoo 2005], which show the
formation of similar intermetallic phases with an average grain size of less than 140 nm.
81
Figure 4.2.3 (a) OM, (b) SEM, (c)-(e) TEM micrographs, and (f) EDX elemental mapping of the Al87Ni8La5 powder consolidated by SPS at 713 K.
The microstructure of the consolidated bulk material was studied by OM, SEM and
TEM, and the corresponding micrographs are shown in Figure 4.2.3. Optical microscopy
investigations (Figure 4.2.3(a)) reveal the formation of a bright interface layer between the
particles, which indicates that particle bonding has taken place during consolidation. The
SEM micrograph in Figure 4.2.3(b) shows that the particles mostly retain their original
spherical shape with a neck geometry characterized by center approach and particle
penetration [Exner 1979], as shown in the inset in Figure 4.2.3(b). TEM investigations of
the sintered sample (Figures. 4.2.3(c)–4.2.3(e)) show that the microstructure of the
particles consists of black areas, often continuously connected, with dimensions of about
200 – 300 nm along with bright areas with dimensions in the range 100 – 200 nm. The
TEM micrograph in Figure 4.2.3(e) and the corresponding EDX elemental mapping in
Figure 4.2.3(f) reveal that the black areas consist of fcc Al whereas the bright areas
82
comprise two phases: a Ni-rich phase with size below 100 nm and a La-rich phase with
dimensions of about 200 nm, which most likely correspond to the intermetallics Al3Ni and
Al11La3 observed by XRD (Figure 4.2.1). The black interface between the particles
(triangular areas) visible in the SEM and TEM micrographs (Figures. 4.2.3(b)–4.2.3(d)),
corresponding to the bright areas in Figure 4.2.3(a), is not due to porosity. TEM and EDX
analysis reveals that the particle interface consists of an fcc Al matrix along with several
bright particles with dimensions below 50 nm, most likely Al11La3 and/or Al3Ni phases.
A similar microstructure (i.e. Al at the triangular areas) has been observed for the
HP Al84Gd6Ni7Co3 sample (Figure 4.1.12). However, for the Al87Ni8La5 powder
consolidated by SPS the fcc Al phase at the triangular areas is continuously connected to
the particles through fcc Al channels (see arrows in Figure 4.2.3(d)) with thickness of
about 200 nm. This microstructure promises improved plastic deformation with respect to
the HP Al84Gd6Ni7Co3 sample as a result of the enhanced dislocation activity. Only few
pores are visible (indicated by arrows in Figure 4.2.3(c)), corroborating the high density of
the sintered samples evaluated by density measurements (~98 %).
Figure 4.2.4 (a) Compression true stress-true strain curves for the Al87Ni8La5 bulk material consolidated by SPS at 713 K (present work) and Al85Ni10La5 consolidated by SPS at 753 K [Sasaki 2008]. (b) Compression true stress-true strain curves for the Al87Ni8La5 material and pure Al: experimental data (lines) and values calculated by using the effective medium approach (points).
A typical room temperature uni-axial compression stress-strain curve for the
sintered sample is shown in Figure 4.2.4(a). The material exhibits an elastic regime of
0.7% and a yield strength (0.2% offset) of about 740 MPa followed by a region with strain
hardening up to the maximum stress of 930 MPa. After reaching the maximum, the stress
gradually decreases with increasing strain to about 640 MPa and fracture occurs at 27%
strain. Similar features have been recently reported for nanocrystalline Al-5 at.% Fe
83
consolidated by SPS [Sasaki 2007]. The sintered Al-Fe exhibits a compressive strength of
1045 MPa followed by a pronounced work softening-like behavior and a plastic strain of
about 30% [Sasaki 2007]. Plastic deformation of the sintered Al87Ni8La5 sample does not
lead to further densification, as demonstrated by the density of the specimen after
compression test that is reduced by about 1% with respect to the as-sintered specimen
(98% of the theoretical density).
Spark plasma sintering of the Al87Ni8La5 powder leads to a highly dense specimens
displaying high strength combined with remarkable plastic deformation. Such a behavior is
presumably due to the multi-phase microstructure consisting of soft fcc Al and high-
strength intermetallic compounds [Ohtera 1992]. However, the observed room temperature
plastic deformation is in contrast to what was reported for other Al-based alloys produced
by consolidation of gas-atomized powders [Inoue 2001a, Kawamura 2001], which display
a similar microstructure. For example, although fully crystallized Al-Ni-Y-Co samples
exhibit an extremely high compressive strength of 1420 MPa [Inoue 2001a, Kawamura
2001], the plastic strain is only about 1% [Inoue 2001a]. This is similar to recent results of
Sasaki et al. [Sasaki 2008] on fully crystallized nanocrystalline Al85Ni10La5 samples
produced by SPS of gas-atomized amorphous powders, which show a compressive strength
exceeding 1200 MPa, but no plastic deformation (Figure 4.2.4(a)).
Figure 4.2.5 SEM micrographs of a consolidated Al87Ni8La5 specimen: (a) as-sintered and (b) polished fracture surface after the compression test.
A possible explanation for the larger plastic deformation of the present Al87Ni8La5
with respect to the Al85Ni10La5 sample of Sasaki et al. [Sasaki 2008] related to the different
microstructures of the sintered samples. The sample of Sasaki et al. [Sasaki 2008] sintered
at 753 K (above the crystallization temperature) displays a microstructure consisting of fcc
Al regions surrounded by a large amount of intermetallic particles [Sasaki 2008]. The
regions of fcc Al appear to be confined and constrained by the intermetallic phases, which
84
presumably leads to the observed lack of deformation capability as a result of the limited
dislocation nucleation and movement. In contrast, in the present sample the fcc Al regions
at the particle interface (inter-particle) are continuously connected to the particles through
intra-particle fcc Al channels (Figure 4.2.3), giving rise to a network of ultra fine-grained
(UFG) Al reinforced with nm-scale intermetallic particles, which extends over the entire
specimen. Within this structure, the fcc Al regions are not confined and, as a result, the
continuous network of fcc Al may allow the movement of dislocations, explaining the
remarkable plastic deformation. This is corroborated by the comparison between the
particle morphology in the sintered samples before and after compression (Figure 4.2.5).
Before testing, the particles have a regular spherical shape (Figure 4.2.5(a)), whereas after
compression they collectively assume a squeezed elliptical shape (Figure 4.2.5(b)) with
major axis perpendicular to the compression direction (indicated by arrows in Figure
4.2.5(b)). Assuming that the average local deformation (εp) of the particles is given by εp =
(l - r)/r,, where l is the major axis of the ellipse after compression and r is the radius of the
original undeformed spherical particle, the value of εp is about 17%. This deformation can
be ascribed to the intra-particle Al regions because intermetallic phases are typically brittle
at room temperature [Koch 1998] and, therefore, are not able to deform plastically.
Similarly to the surrounding particles, the inter-particle Al regions at the interface also
deform plastically in order to keep the geometrical integrity (Figure 4.2.5(b)).
The uniaxial strain-stress curve of deformable metallic materials can be expressed
by the Ramberg-Osgood equation [Clyne 1993]
n
y
y
EE
/1
⎟⎟⎠
⎞⎜⎜⎝
⎛+=
σσσ
ασε , (4.2.1)
where E is the Young’s modulus of the metallic material, n is the strain hardening
exponent, α a dimensionless constant (α = 3/7 is usually taken for Al-based alloys
[Wilkinson 2001], and the yield strength. Based on Equation (eq. 4.2.1), the strain-
stress curve can be successfully modeled by using the effective medium approach (EMA)
[
yσ
Kouzeli 2002, Nan 1996, Scudino 2009]. For the present Al-Ni-La sample, calculations
using Equation (equation 4.2.1) and EMA are performed to determine n of the fcc Al
(Figure 4.2.4(b)). The calculations are in good agreement with the experimental results in
the strain hardening part of the curve before the critical strain of structural instability,
giving a value of n of 0.16. On the other hand, the value of n for pure aluminum evaluated
85
from the strain-stress curve in Figure 4.2.4(b) is only 0.04. This indicates that the
Al87Ni8La5 sample is characterized by a more intense dislocation storage and interaction
compared to pure aluminum [Hull 2001]. Two main reasons may be responsible for this
behavior: (i) extensive generation of dislocation and (ii) dislocation movement limited by
constraint effects [HullBacon 2001]. When a crystal is plastically deformed, dislocations
are generated, moved and stored. Dislocation storage, which causes the material to work-
harden, occurs by mutual trapping or by accommodating the deformation incompatibility
between various parts in the deformed materials. The dislocations that are mutually trapped
are referred to as statistically stored dislocations [Ashby 1971] and their density Sρ is
difficult to estimate. The dislocations that are stored due to deformation incompatibility are
called geometrically necessary dislocations, Gρ . Gρ is mainly related to the thermal
mismatch strains imposed upon cooling down from the processing temperature ( ) and
to the strain gradient present during deformation ( ), i.e., [
thGρ
sgGρ sg
GthGG ρρρ += HullBacon
2001]. In the case of the present sample consisting of fcc Al and rigid intermetallics,
and can be expressed respectively as [
thGρ
sgGρ Arsenault 1986, Nix 1998]
⎟⎟⎠
⎞⎜⎜⎝
⎛++
−
Δ×Δ=
321
111)1(
4DDDbV
TCTEV
p
pthGρ , (4.2.2)
λερ
bsgG =
4 , (4.2.3)
where is the difference in the coefficient of thermal expansion between fcc Al and
the intermetallics,
CTEΔ
TΔ is the temperature differential upon cooling (about 700 K), is the
volume fraction of intermetallic phase (about 0.60), b is the magnitude of the Burger’s
vector (0.28 nm [
pV
Rooy 1979]), , and are the three-dimensional sizes of the
intermetallic phase (about 1 μm × 1 μm × 0.5 μm), ε is the strain and λ is the local length
scale of the deformation field or the size of fcc Al (about 500 nm, see Figure 4.2.3(c)). By
using of about 5 × 10-6 K-1, is estimated as 3.0 × 1014 m-2. From Equation (4.2.2)
follows that increases with the strain. At the applied strain of 2%, can reach a
value of about 5.7 × 1014 m-2. This indicates that a high density of dislocations nucleate in
the fcc Al during deformation, explaining why the present sample has a strain hardening
exponent much larger than that of pure aluminum.
1D 2D 3D
CTEΔ thGρ
sgGρ sg
Gρ
86
Figure 4.2.6 SEM images of a consolidated Al87Ni8La5 specimen: (a) as-sintered, (b) after 4% and 10% plastic deformation.
Although the Al network is able to accommodate a large amount of plastic
deformation, deformation of fcc Al through dislocation activity is nevertheless limited by
the presence of the rigid intermetallics, which most likely makes further deformation
increasingly difficult. The storage of dislocations then may cause stress concentration in
the constrained Al, resulting in the formation of microcracks between the particles. In
order to clarify this aspect, the microstructure of the Al87Ni8La5 consolidated samples at
different stages of plastic deformation (4 and 10%) was analyzed using SEM and the
corresponding images are shown in Figures. 4.2.6(b) and 4.2.6(c), respectively, together
with the microstructure of the as-sintered sample (Figure 4.2.6(a)). The SEM micrographs
in Figure 4.2.6 reveal interesting features. The contraction of the powder particles parallel
to the compression direction (indicated by arrows in Figure 4.2.6), already observed in
Figure 4.2.5(a), starts to be visible at a plastic deformation of 4% and becomes evident for
the sample deformed up to 10% (see for example the particles labeled 1, 2 and 3). At the
same time, microcracks are formed at the interface between the particles (particles labeled
4 and 5).
87
Only few cracks are clearly visible in Figure 4.2.6 most likely because the SEM
investigation was carried out on the sample surface and, therefore, the inter-particle Al is
only partially constrained. In the inner part of the sample a larger number of microcracks
may form even at lower strains due to the giant hydrostatic stress and the three dimensional
constraints [Goods 1979]. Nevertheless, Figure 4.2.6 clearly shows that microcracking
occurs during deformation. As a consequence of crack formation and resulting stress
relaxation, softening-like behavior can be observed in the strain-stress curve (Figure 4.2.4
(a)). For large deformations, microcracks readily coalesce to form a main crack that rapidly
propagates through the sample, finally leading to fracture. Figure 4.2.7 shows the fracture
surface of the compressed sample, where inter-particle fracture can be clearly observed.
Besides particle deformation and microcracking, the sample deformed up to 10%
displays profuse shear banding, as shown by the OM and SEM images in Figure 4.2.8.
Most likely shear banding accommodates the strain when the dislocation movement is
limited. The shear bands form an angle of about 45° with the compression direction
(indicated by arrows in Figure 4.2.8). Multiple shear bands are often observed during the
deformation of metallic glass composites [Hays 2000, He 2003]. The difference between
the present material and metallic
Figure 4.2.7 SEM micrograph of the fracture morphology after compression test for the consolidated Al87Ni8La5 specimen.
glass composites is that shear banding is typically the only deformation mechanism of
glassy composites, while dislocation motion is also responsible for the plastic deformation
of the present composite, as shown in Figure 4.2.8(d), where, within the shear bands, both
particles and inter-particle matrix are plastically deformed. The activation of multiple shear
bands together with the dislocation-associated deformation results in a composite having a
fracture strain of about 27%.
88
Figure 4.2.8 (a)-(b) OM and (c)-(d) SEM images of the consolidated Al87Ni8La5 sample plastically deformed up to 10%, displaying profuse shear banding.
89
4.3 Gas-atomized Al90.4Y4.4Ni4.3Co0.9 powder
As shown in the previous two sections, the consolidation of Al-based gas-atomized
powders yields highly-dense bulk samples characterized by high strength levels. The HP
Al84Gd6Ni7Co3 powder exhibits remarkably high strength of about 1550 MPa as compared
to about 930 MPa for the Al87Ni8La5 sample consolidated by SPS. However, the plastic
strain is significantly improved for the Al87Ni8La5 material. This is related to the higher Al
content of the Al87Ni8La5 alloy with respect to the Al84Gd6Ni7Co3 powder and to the
corresponding microstructural features: increasing Al content gives rise to enhanced plastic
strain along with a decrease of strength. To further investigate the effect of Al content on
the mechanical behavior of the consolidated samples, in this section, Al90.4Y4.4Ni4.3Co0.9
gas-atomized powders are consolidated by hot pressing and their mechanical properties are
investigated in detail. Additionally, samples were consolidated by hot pressing followed by
hot extrusion in order to test the tensile mechanical properties of the bulk sample and to
compare the tensile mechanical behavior with the results obtained in compression.
Figure 4.3.1 XRD pattern (Co-Kα radiation) of Al90.4Y4.4Ni4.3Co0.9 powder: as-atomized, after heating to 640 and 873 K in the DSC, bulk sample consolidated by hot pressing at 673 and 723 K and samples hot pressed and hot extruded at 723 K.
The microstructure of the as-atomized Al90.4Y4.4Ni4.3Co0.9 powder investigated by
XRD is shown in Figure 4.3.1. The pattern exhibits broad maxima typical for an
amorphous phase together with broad diffraction peaks of fcc Al. This indicates that the as-
atomized powder is a mixture of an amorphous phase and nanocrystalline Al, which is
90
similar to the structure observed for the Al87Ni8La5 gas-atomized powder (Figure 4.2.1).
The DSC scan (40 K/min) of the as-atomized Al90.4Y4.4Ni4.3Co0.9 powder is shown in
Figure 4.3.2. The DSC curve exhibits two exothermic peaks due to the crystallization of
the glass without any clear glass transition region. The onset temperatures of first and
second exothermic peaks are Tx1 = 600 K and Tx2 = 650 K, respectively.
Figure 4.3.2 DSC scan (40 K/min) of as-atomized Al90.4Y4.4Ni4.3Co0.9 powder.
The XRD patterns of the as-atomized powder after heating to different
temperatures are shown in Figure 4.3.1. The XRD pattern after heating to 640 K (above the
first crystallization event) shows broad diffraction peaks belonging to Al along with
diffraction peaks from the Al3Y and Al4Ni3 intermetallic compounds. In addition, an
extremely weak and broad maximum due to the residual amorphous phase can be observed.
The XRD pattern of the powder heated to 873 K (far above the second crystallization
event) shows the peaks from Al and intermetallic phases Al3Y and Al4Ni3 and Al9Co2. No
amorphous phase is visible at this stage, indicating that complete crystallization of the
glass occurs during the heat treatment at 873 K.
According to the results presented in section 4.2, excellent mechanical properties
can be obtained for the Al84Gd6Ni7Co3 powder by HP at 673 and 723 K. Therefore, the
same sintering temperatures (673 and 723 K) have been used to consolidate the
Al90.4Y4.4Ni4.3Co0.9 powder by HP. The XRD pattern of the Al90.4Y4.4Ni4.3Co0.9 sample hot-
pressed at 673 K (Figure 4.3.1) reveals the formation of the phases already observed in the
sample heated to 640 K in the DSC, i.e. Al, Al3Y and Al4Ni3 intermetallic compounds. As
well, the week broad maximum due to the presence of a residual amorphous phase can be
91
observed. The presence of residual amorphous phase is due to the sintering temperature
(673 K), which is below the second crystallization event. The XRD pattern of the sample
HP at 723 K (above the second crystallization peak in Figure 4.3.2) reveals the formation
of Al, Al4Ni3, Al3Y and Al9Co2 phases without any visible residual amorphous phase
(Figure 4.3.1).
Figure 4.3.3 (a)-(b) SEM micrographs (c)-(d) bright-field TEM micrographs of the Al90.4Y4.4Ni4.3Co0.9 samples hot-pressed at 673 K
Figures 4.3.3(a) and 4.3.3(b) show the SEM micrographs of the sample hot-pressed
at 673 K. The micrographs reveal the formation of a bright interface between the particles
along with black areas at the triangular areas, as already observed for the Al84Gd6Ni7Co3
and Al87Ni8La5 consolidated samples (Figures 4.1.12 and 4.2.3). The particles are not
homogeneous and can be considered as a composite microstructure consisting of grains of
about 1-3 μm size separated by a matrix with rod-like morphology (Figure 4.3.3(b)). EDX
analysis indicates that the bright interfaces are rich in Y, Ni and Co (≈ 40±10% rich than
the original atomic percent), which may correspond to the intermetallic compounds
92
observed by XRD (Figure 4.3.1). A similar composition has been observed for the rod-like
matrix between the particles. On the other hand, the 1-3 μm sized grains in Figure 4.3.3(b)
show a higher Al content (of about 5% of that of nominal Al content) that, considering the
XRD results in Figure 4.3.1, suggests a mixed microstructure consisting of Al and the
residual amorphous phase.
Figures 4.3.3(c) and 4.3.3(d) show the TEM-bright field micrographs
corresponding to the black regions between the particles (triangular areas) observed in
Figure 4.3.3(a). The figures reveal that the triangular areas are made of Al grains of about
100 – 300 nm and are surrounded by the rod-like features (the intermetallic compounds)
with 40 nm thickness and 300 to 500 nm length.
Figure 4.3.4 shows the SEM micrographs of the sample hot-pressed at 723 K.
Similarly to the sample HP at 673 K (Figure 4.3.3), the micrographs show the formation of
bright interfaces between the particles, indicative of good particle bonding, along with the
presence of the rod-like structures. The amount of rod-like structure in the sample hot-
pressed at 723 K is higher (≈ 30 ± 10 % from the image analysis) as compared to the
sample HP at 673 K. This is accompanied by the disappearance of the 1-3 μm sized grains
visible in Figure 4.3.3(b) and to the formation of a uniform microstructure throughout the
particle consisting of rod-like intermetallics and ultra fine-grained Al grains. The formation
of such a microstructure can be related to the high sintering temperature (723 K) which
leads to the full crystallization of the amorphous phase and to the formation of additional
fcc Al and intermetallic compounds from the residual amorphous phase (Figure 4.3.1).
Figure 4.3.4 SEM micrographs of hot-pressed Al90.4Y4.4Ni4.3Co0.9 samples at 723 K
Figure 4.3.5 shows the room temperature compressive stress-strain curves of the
samples consolidated by hot pressing at 673 and 723 K. The sample HP at 673 K exhibits a
93
yield and compressive strength of about 880 ± 10 MPa and 925 ± 2 MPa, respectively.
With increasing stress, the curve displays a softening behavior up to fracture, which occurs
at 850 ± 4 MPa stress and 14 ± 1% strain. The softening-like behavior is similar to that
observed for the Al-Ni-La alloy discussed in section 4.2. The stress-strain curve of the
sample HP at 723 K exhibits lower yield and compressive strengths (780 ± 10 MPa and
820 ± 2 MPa) as compared to the sample hot-pressed at 673 K. However, the fracture
strain is remarkably larger, reaching a value of 30 ± 2% at 690 ± 5 MPa stress.
Figure 4.3.5 Compression true stress-true strain curves of samples of as-atomized Al90.4Y4.4Ni4.3Co0.9 powder consolidated by hot pressing at 673 and 723 K.
The higher stress level of the sample hot-pressed at 673 K with respect to the
sample HP at 723 K is most likely due to the presence of the residual amorphous phase, as
shown by the corresponding XRD pattern in Figure 4.3.1 and the SEM micrograph in
Figure 4.3.3(b). Due to the incomplete crystallization of the amorphous phase, the sample
consolidated at 673 K displays a lower amount of crystallized fcc Al. As a result, the
fracture strain of the sample HP at 673 K is reduced compared to the sample HP at 723 K
that, in contrast, is fully crystallized and is not characterized by the hindrance of the
dislocations movement induced by the residual amorphous phase.
The current results clearly demonstrate that Al-based materials characterized by
high strength combined with considerable plastic strain can be produced through the
combined devitrification and consolidation of glassy precursors. The room temperature
mechanical properties of the materials investigated in the previous sections have been
tested in compression because the limited dimensions of the HP and SPS samples do not
94
permit accurate tensile tests. However, for the full evaluation of the mechanical behavior
of a material, tensile stress-strain data are necessary. Accordingly, preliminary tensile
results have been obtained for Al90.4Y4.4Ni4.3Co0.9 samples consolidated by hot pressing
followed by hot extrusion at 723 K. Through the extrusion of the HP samples, specimens
with dimensions suitable for standard tensile tests can be produced.
Figure 4.3.6 SEM micrographs of the Al90.4Y4.4Ni4.3Co0.9 samples consolidated by hot pressing followed by hot extrusion at 723 K.
Similarly to the sample HP at 723 K, the XRD pattern of the sample hot-extruded at
723 K (Figure 4.3.1) reveals diffraction peaks from Al, Al4Ni3, Al3Y and Al9Co2 phases
without any visible residual amorphous phase. The relative amounts of the phases is
slightly different for the sample hot-extruded at 723 K, which shows a larger amount of
Al3Y and a lower amount of Al4Ni3 compared to the sample that was hot-pressed at 723 K
(Figure 4.3.1). In addition, the diffraction peaks belonging to fcc Al are sharper, indicative
of a larger grain size. The grain size is about 150 nm for extruded sample as compared to
≈ 120 nm for the HP samples according to Scherer formula. This is due to the longer
dwelling time at 723 K of the double-step hot-pressing/hot-extrusion process (20 minutes)
compared to the single hot-pressing process (3 minutes). Figure 4.3.6 shows SEM
micrographs of the sample hot-extruded at 723 K, which reveals a microstructure
consisting of bright rod-like particles (the intermetallic compounds) distributed in dark-
gray background (fcc Al). In contrast to the material HP at 723 K (Figure 4.3.4), the
particles do not maintain the spherical shape. Also, no particle boundaries can be observed.
Due to the extrusion step, the nanocrystalline intermetallic particles are well-distributed
within the fcc Al giving rise to a homogeneous composite microstructure. In hot pressing
the powder particles shows spherical shape consists of nanocrystalline particles in Al
matrix, bright interfaces and congregation of Al at the triangular areas. This type of mixed
95
structure is completely sheared due to extrusion and results two phase structure of
nanocrystalline particles distributed in Al matrix without any boundaries.
Figure 4.3.7 shows the compressive and tensile stress-strain curves of the sample
hot extruded at 723 K. The hot-extruded Al90.4Y4.4Ni4.3Co0.9 material exhibits compressive
yield and fracture stresses of about 490 ± 5 MPa and 645 ± 3 MPa combined with 15 ± 1%
plastic strain. These values are smaller than those observed for the sample HP at 723 K
(Figure 4.3.5), probably because of the different microstructure induced by the additional
extrusion step. When tested in tension, the extruded specimen exhibits yield and fracture
stresses of about 480 ± 5 MPa and 565 ± 2 MPa, which are rather similar to the
compressive results. On the other hand, the tensile ductility compared to compression is
remarkably reduced (4 ± 1%), presumably because of the residual porosity (~ 1%) which is
more critical in the tensile mode for reducing the plastic strain through crack formation and
propagation than in the compressive mode [German 2008].
Figure 4.3.7 Room temperature tensile and compressive true stress-true strain curves for the samples consolidated by hot extrusion carried at 723 K of as-atomized Al90.4Y4.4Ni4.3Co0.9 powder.
The small difference between compressive and tensile strength observed in this
work clearly indicates that compression tests can be successfully used for a preliminary
evaluation of the strength of the consolidated materials. In addition, these preliminary
tensile results further demonstrate the validity of the combined devitrification and
consolidation of glassy precursors as a suitable method for the production of Al-based
materials characterized by high strength and considerable plastic deformation.
96
Chapter 5: Conclusions and outlook
Amorphous, partially amorphous and nanocrystalline Al-based alloys have been
attracting widespread attention as potential candidates for structural as well as functional
applications due to their high strength combined with low density. Although these
materials exhibit improved mechanical properties compared to conventional Al-based
crystalline alloys, the maximum scale of the products is limited to a thickness of less than
100 micrometers due to their relatively low glass-forming ability. In general, Al-based
metallic glasses and nanostructured materials with high Al content can only be obtained by
melt-spinning in the shape of ribbons or by gas atomization in the form of powder. This
limitation has prevented a wide extension of application fields of the Al-based amorphous
and nanocrystalline alloys even despite their excellent mechanical properties. To overcome
this limitation, powder metallurgical methods, such as gas atomization or mechanical
alloying followed by powder consolidation, can be employed to create bulk Al-based
samples with the desired microstructure. Accordingly, in this work novel bulk Al-based
alloys with high content of Al have been produced by powder metallurgy methods from
amorphous and partially amorphous materials. The present work focused on three specific
aspects:
(1) Production and characterization of Al-based amorphous and partially
amorphous powders with high Al content (> 80 at.%).
(2) Consolidation of the powder precursors into bulk samples with the
desired microstructure by different techniques.
(3) Microstructural characterization and mechanical property evaluation of
the consolidated bulk specimens.
Different processing routes, including mechanical alloying of elemental powder
mixtures, pulverization of melt-spun glassy ribbons and gas atomization, have been used
for the production of the Al-based powders. Although the mechanically-alloyed
Al85Y8Ni5Co2 powders reveal promising results in terms of glass formation and stability,
the milling time needed for amorphization is extremely long and the production yields poor
output, which drastically limits the use of mechanical alloying for the production of Al-
based powders with high Al contents. A better approach in terms of output of the powders
is the production of glassy powders by controlled pulverization of melt-spun ribbons. In
order to retain their glassy structure and to avoid sticking of the material to the milling
97
tools due to the high ductility of the ribbons, proper milling conditions have to be used (i.e.
interval-milling at a low intensity, corresponding to a rather low kinetic energy, performed
at the cryogenic temperature). Although this technique allows the production of larger
quantities of high quality material (i.e. constant composition and microstructure combined
with low contamination levels) compared to mechanical alloying, the procedure is rather
complex and requires the detailed characterization of every single melt-spun ribbon. On
the other hand, gas atomization offers the possibility to easily produce large quantities of
powders with homogeneous properties (e.g. structure and thermal stability) along with a
uniform size distribution of particles. Therefore, gas atomization is the best choice for the
production of Al-based amorphous and partially amorphous powders as precursors for the
subsequent consolidation step.
Materials in powder form have to be consolidated to achieve dense bulk specimens.
Consolidation of metastable phases, such as amorphous and nanocrystalline materials, is
not a trivial process and often results in undesirable microstructural transformations (e.g.
crystallization and grain coarsening), or insufficient particle bonding. These characteristics
severely limit the consolidation parameters that can be used and, as a result, temperature,
pressure and the time span of the consolidation process have to be adjusted carefully in
order to achieve a balance between good densification and desired microstructure. For that
reason, the crystallization behavior and the temperature dependence of the viscosity have
been studied in detail in order to optimize the processing conditions and to select the
proper consolidation parameters.
Following their characterization, the Al-based powders have been consolidated into
bulk specimens by hot pressing (HP), hot extrusion and spark plasma sintering (SPS) and
their microstructure and mechanical properties have been investigated. The results indicate
that the mechanical properties of the consolidated samples can be varied within a wide
range of strength and ductility depending on the microstructure and the consolidation
techniques used.
Single-phase amorphous bulk Al85Y8Ni5Co2 specimens were produced by hot
pressing of the pulverized ribbons. Room temperature compression tests of the single-
phase glass reveal low strength and no ductility due to the residual porosity of the
consolidated specimen. No extrusion of the single-phase glass was possible at temperatures
below the crystallization temperature due to the insufficient viscous flow of the SCL in the
present consolidation conditions. In order to reach a higher density and, consequently, to
improve the ductility of the samples, the milled amorphous ribbons were blended with
98
different volume fractions of elemental fcc Al to produce glass-matrix composites. The
resulting powders were then consolidated by hot pressing followed by hot extrusion. When
50 vol.% Al is added, the maximum stress is found to be 295 MPa, therefore, decreased
with respect to the single-phase amorphous specimen (400 MPa). However, the strain at
maximum stress is about 7 %. The material containing 70 vol.% Al exhibits a maximum
stress of 255 MPa and a strain at maximum stress is about 10 %. These results indicate that
glass-reinforced Al-based composites with high strength combined with considerable
ductility can be produced by powder metallurgy methods. The mechanical properties of the
glass-reinforced composites can be modeled by using the iso-stress Reuss model, which
allows the prediction of the mechanical properties of a composite from the volume-
weighted averages of the components properties.
Higher strength levels combined with good plastic deformation at room
temperature can be achieved by the combined devitrification and consolidation of gas-
atomized amorphous and partially amorphous precursors. For this purpose, Al-based gas-
atomized powders (GAP) with compositions Al84Gd6Ni7Co3, Al87Ni8La5 and
Al90.4Y4.3Ni4.4Co0.9 have been used as precursors for the production of high-strength bulk
samples.
The results on hot-pressing of the Al84Gd6Ni7Co3 powder indicate that
consolidation into highly-dense bulk samples cannot be achieved without extended
crystallization of the material. Nevertheless, crystallization during consolidation is not
detrimental and leads to bulk samples with a remarkably high strength of about 1500 MPa,
which is three times larger than the conventional high-strength Al-based alloys.
Investigation of the sintering behavior reveals that preferential diffusion of Al toward the
pore regions occurs during hot pressing, leading to the filling of the triangular areas
between the particles by ultra fine-grained Aluminum. This, together with the formation of
rod-like intermetallics, leads to bulk samples characterized by high strength combined with
a limited but distinct plastic deformability (3.5 - 4%).
To further test the effectiveness of powder consolidation as a method for the
production of Al-based materials with enhanced mechanical properties through the
combined crystallization and consolidation of glassy precursors, Al87Ni8La5 gas-atomized
powders have been consolidated by SPS above their crystallization temperature. Spark
plasma sintering leads to highly dense bulk specimens with a multi-phase structure
consisting of fcc-Al together with Al11La3 and Al3Ni intermetallic compounds. The
consolidated bulk material exhibits high compression strength of 930 MPa together with
99
plastic strain exceeding 25 %. The high deformation capability is most likely due to the
formation of a microstructure consisting of a network of ultra fine-grained Al reinforced
with nm-scale intermetallic particles. Within this structure, the fcc Al regions are not
confined and, as a result, the continuous network of fcc Al may allow the movement of
dislocations, explaining the remarkable plastic deformation with respect to the
Al84Gd6Ni7Co3 bulk sample.
In order to investigate the effect of high Al content on the mechanical behavior of
the consolidated samples, Al90.4Y4.4Ni4.3Co0.9 gas-atomized powder have been consolidated
by hot pressing. The bulk samples display remarkable mechanical properties, namely, high
compression strength ranging between 820 and 925 MPa combined with plastic strain in
the range 14 – 30%. Strength and plastic strain of the hot-pressed samples are strictly
linked with their microstructure. Higher strength and reduced plasticity are related to the
presence of a residual amorphous phase, which may hinder the dislocations movement
within the Al phase. On the other hand, reduced strength but enhanced plastic deformation
is a result of the complete crystallization of the glass and of the formation of additional fcc
Al from the residual amorphous phase. In addition, preliminary tensile tests for the
Al90.4Y4.3Ni4.4Co0.9 alloy consolidated by hot pressing followed by hot extrusion reveal
promising tensile properties (tensile strength 565 MPa and 4% ductility) comparable to
commercial high-strength Al-based alloys.
The results presented in this thesis clearly indicate that powder metallurgy, i.e.
powder synthesis and consolidation, is a particularly suitable method for the production of
Al-based materials characterized by high strength combined with considerable plastic
strain. The combined devitrification and consolidation of glassy precursors into high-
strength deformable bulk samples promise a new route for the development of novel and
innovative high-performance Al-based materials for eco-friendly transport applications.
However, some aspects have still to be investigated. For example, the mechanism
responsible for the preferential diffusion of Al toward the pore regions has to be fully
clarified and supplementary investigations, such as additional studies on the effect of the
glass crystallization and related atomic diffusion on the sintering behavior and resulting
microstructure, are required to clarify this aspect. Also, systematic investigations on the
tensile properties of these novel Al-based alloys have to be carried out. In addition, fatigue,
wear and corrosion properties of the consolidated materials, which are crucial aspects for
any potential commercial application, have to be fully evaluated.
100
Bibliography
Afonso, C.R.M., de Oliveira, M.F., Bolfarini, C., Botta, W.J. and Kiminami, C.S.,
Microstructure of spray formed Fe83Nb4Zr1Ti1B9Cu2 alloy, Materials Science Forum,
416-4 (2003) 388-394.
Allen, D.R., Foley, J.C. and Perepezko, J.H., Nanocrystal development during primary
crystallization of amorphous alloys, Acta Materialia, 46 (1998) 431-440.
Anantharaman, T.R. and Suryanarayana, C., Rapidly solidified metals - a technological
overview, Trans Tech Publications, Aedermannsdorf, Switzerland 1987.
Arsenault, R.J. and Shi, N., Dislocation generation due to differences between the coefficients
of thermal expansion, Materials Science and Engineering, 81 (1986) 175-187.
Ashby, M.F., The deformation of plastically non-homogeneous alloys, Strengthening Methods
in Crystals, (1971) 137-192.
ASTM, E9 -09 Standard Test Methods of Compression Testing of Metallic Materials at Room
Temperature in: Metals Test Methods and Analytical Procedures, ASTM International,
ASTM, West Conshohocken, PA, U.S., 2004, pp. 10.
Augis, J.A. and Bennett, J.E., Calculation of the Avrami parameters for heterogeneous solid
state reactions using a modification of the Kissinger method, Journal of Thermal
Analysis and Calorimetry, 13 (1978) 283-292.
Averback, R.S., Hofler, H.J. and Tao, R., Processing of Nano-Grained Materials, Materials
Science and Engineering A, 166 (1993) 169-177.
Avrami, M., Kinetics of phase change I - General theory, Journal of Chemical Physics, 7
(1939) 1103-1112.
Avrami, M., Kinetics of Phase Change. II Transformation-Time Relations for Random
Distribution of Nuclei, The Journal of Chemical Physics, 8 (1940) 212-224.
Avrami, M., Granulation, Phase Change, and Microstructure - Kinetics of Phase Change. III,
Journal of Chemical Physics, 9 (1941) 177-184.
Bakker, H., Zhou, G.F. and Yang, H., Mechanically driven disorder and phase
transformations in alloys, Progress in Materials Science, 39 (1995) 159-241.
Börner, I. and Eckert, J., Phase formation and properties of mechanically alloyed amorphous
Al85Y8Ni5Co2, Scripta Materialia, 45 (2001) 237-244.
Brand, K. and Schatt, W., Solid-Phase Sintering of Powder Mixture Compacts with Complete
Solubility of the Components, Zeitschrift Für Metallkunde, 84 (1993) 893-898.
100
101
Bruck, H.A. and Rabin, B.H., Evaluating microstructural and damage effects in rule-of-
mixtures predictions of the mechanical properties of Ni-Al2O3 composites, Journal of
Materials Science, 34 (1999) 2241-2251.
Burke, J., The kinetics of phase transformation in metals, Pergamon Press, Oxford, 1965.
Busch, R., Bakke, E. and Johnson, W.L., Viscosity of the supercooled liquid and relaxation at
the glass transition of the Zr46.75Ti8.25Cu7.5Ni10Be27.5 bulk metallic glass forming alloy,
Acta Materialia, 46 (1998) 4725-4732.
Cahn, R.W., Metallic Glasses, in: Cahn, R.W., Haasen, P., Kramer, E. J. (Ed.) Materials
Scince and Technology: A comprehensive treatment, VCH, Weinheim, 1993, pp. 493.
Cahn, R.W. and Greer, A. L., Metastable States of Alloys, in: Robert, W.C. and Peter, H.
(Eds.) Physical Metallurgy (Fourth Edition), North-Holland, Oxford, 1996, pp. 1723-
1830.
Calin, M., Grahl, H., Adam, M., Eckert, J. and Schultz, L., Synthesis and thermal stability of
ball-milled and melt-quenched amorphous and nanostructured Al-Ni-Nd-Co alloys,
Journal of Materials Science, 39 (2004) 5295-5298.
Chawla, K.K., Composite Materials : Science and Engineering, 2 ed., Springer New York,
1998.
Chen, H.S., Thermodynamic Considerations on Formation and Stability of Metallic Glasses,
Acta Metallurgica, 22 (1974) 1505-1511.
Chen, H.S., Turnbull, D., Formation, Stability and Structure of Palladium-Silicon Based
Alloy Glasses, Acta Metallurgica, 17 (1969) 1021.
Chen, L.C. and Spaepen, F., Calorimetric Evidence for the Microquasicrystalline Structure of