Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using

the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

Focused Development of Magnesium Alloys using

the Calphad Approach
Rainer Schmid-Fetzer, Joachim Gröbner

Technical University of Clausthal, Institute of Metallurgy,

Robert-Koch-Str. 42
D-38678 Clausthal-Zellerfeld, Germany

Abstract
In traditional alloy development, experimental investigations with many different alloy
compositions are performed. The selection criteria for multicomponent alloying elements and
their compositions become diffuse in a traditional approach. Computational thermochemistry
as used in the Calphad approach can provide a clear guideline for such selections and helps to
avoid large scale experiments with less promising alloys. Thus, it is a powerful tool to cut
down on cost and time during development of Mg-alloys. An overview of the Calphad
method is given. As an example of application, recent developments of new creep resistant
magnesium alloys that show about 100 times less creep than the best commercial alloys are
reported. Also outlined are the methods used in our long-term project of construction of the
necessary thermodynamic magnesium alloy database for several alloying elements like Al, Li,
Si, Mn, Ca, Sc, Y, Zr and Rare Earths (Ce, Gd, Nd), using the Calphad method combined
with key experiments.

1 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

1. Introduction example the heat content which is

important for die casting, or the chemical
The current revival of magnesium potentials of individual alloy components,
as a structural material is carried by a which are important for selective
relatively small number of traditional Mg- vaporization or oxidation.
alloys. Compared to that, a large number of These are the basic data to
highly specialized and sophisticated understand and control the behavior of any
aluminum alloys, not to speak of steel, was novel or modified Mg-alloy. Large-scale
developed in an ongoing effort over the experiments for new multicomponent
past decades. It is evident that there is an alloys can then be focused on most
urgent need for the development of new or promising alloys identified in that
improved magnesium alloys if we want to approach. Long-term experiments with less
fully exploit the potential of this promising alloys can be avoided. Thus, it is
fascinating lightweight material that also a powerful tool to cut down on cost and
offers excellent castability, machinability time during development of magnesium
and bio-compatibility. Experiments on a alloys.
technological scale for preparation and Quantitative access to all these
testing of new alloys are very expensive basic data in a consistent, numerical and
and time consuming. In view of the huge easily readable form is – at least for
number of possible alloy components, multicomponent alloys – virtually
compositions and processing parameters, impossible without the use of
one would like to have at least an computational thermochemistry. The
“educated guess” in which direction to go. fundamental idea goes back to the
In this report we want to show that introduction of the Calphad method by
thermodynamic calculations can provide Larry Kaufman in 1970 [3]. In a nutshell it
much more than that. Computational is to calculate all these data, ranging from
thermochemistry based on the Calphad stable and metastable phase equilibria
method is a modern tool that supplies down to the chemical potentials, from a
quantitative data to guide the development unique set of thermodynamic model
of alloys or the optimization of materials parameters of the alloy system.
processing [1]. In the following the foundations,
It enables the calculation of applications and recent advancements of
multicomponent phase diagrams and the the Calphad approach are briefly reviewed,
tracking of individual alloys during heat emphasizing the predictive capability to
treatment or solidification by calculation of multicomponent alloys. The ongoing long-
phase distributions and phase term effort in developing a thermodynamic
compositions. It also allows the simulation magnesium alloy database by modeling
of phase transformations during and key experiments is outlined. And, as
solidification of Mg-alloys, which are an example of its application, the focused
responsible for the development of as-cast development of new creep resistant Mg-
microstructures. That can be done either by alloys is illustrated. Two generations of
using the two simple extreme models of these new alloys, the first with high and the
equilibrium solidification or Scheil second with low scandium content, have
solidification, or by more sophisticated been developed using computational
models that require additional kinetic thermochemistry as a powerful tool. Their
material parameters [2]. Other quantities of creep properties are superior compared to
multicomponent Mg-alloys, important for the best commercial alloys.
materials processing, are also provided by
computational thermochemistry, for

2 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

2. The Calphad Approach More advanced methods to model these

“unary” data for elements and
2.1 Foundations and current state stoichiometric end-members of solutions
have also been developed, addressing the
Phase diagrams provide the heat capacity of crystalline phases [9],
graphical presentation of the equilibrium liquid and amorphous phases [10],
state of a material as function of enthalpy estimations for stable and
temperature, pressure and composition of metastable states [11], enthalpies and
the components. This is why they are entropies of transition [12], λ-transitions
frequently used as roadmaps for alloy [13], and periodic system effects [14].
design or a better understanding of the Highlights of the developments of models
processing of materials. The for phases with a range of composition will
thermodynamic properties of materials, be detailed in section 2.3.
like the heat of solidification or the This progress in thermodynamic
chemical activity of components, are also modeling together with the capabilities of
frequently used to understand, for example, modern computational technology led to
metallurgical reactions of materials. These the current state of the Calphad approach
two aspects, phase diagrams and which is characterized by the following
thermodynamic properties, have been highlights:
treated separately for a very long time
despite the fact that their fundamental • A predictive capability allows the
interrelations had been established more extrapolation of thermodynamic
than a century ago by J.W. Gibbs, whose descriptions and phase equilibrium
work has been summarized by Hertz [4]. calculations from assessed binary
Eventually the mathematical systems to ternary, quaternary and
calculation of phase diagrams arose, some higher order systems. This feature is
early examples between 1908 and 1970 are indispensable for an application to real
summarized by Kattner [5]. In 1970 life alloys or processes, it will be
Kaufman et al. [3] initiated the Calphad described further in section 2.3.
method (calculation of phase diagrams) • Identification of key experiments
with a detailed description of procedures drastically reduces the necessary
together with a listing of computer experimental effort in multicomponent
software. The subsequent meetings of the systems. Key experiments can be
ever growing Calphad group, organized by selected to give the maximum amount
Larry Kaufman, soon attained the level of of information, either by verification of
annual international conferences. These extrapolated descriptions or in
meetings together with the establishment assessing higher order interactions. The
of the international Calphad journal helped simultaneous evaluation of all
in the rapid development of sophisticated experimental data irrespective of their
thermodynamic modeling procedures and type (phase equilibrium,
their coupling to phase equilibrium thermodynamic data) and location in
calculations. One important development the multicomponent system is only
concern the establishment of an possible with this approach and results
acknowledged standard for the Gibbs in an internally consistent database of
energies of the pure elements in all their the entire system.
stable and metastable states [6], based on • Stable and metastable phase equilibria
the concept of lattice stabilities [3, 7, 8]. can be calculated. By suspending
This standard is essential to enable individual phases, which may not form
combination of separate assessments of under specific conditions due to kinetic
binary systems to higher order systems. reasons, the complete metastable phase

3 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
relations can be given quantitatively. thermodynamic equilibrium calculations
• The driving forces for all phase started about a century ago with the first
transformations are available. understanding of alloying behavior,
• Local phase equilibria can be microstructure development and
calculated, providing a numerical input metallurgical reactions. These applications
to materials processing software, for are far from being "old-fashioned", though.
example in solidification simulation or Today we are able to apply quantitative
reactor modeling. phase equilibrium calculations in
• The reading of multicomponent phase multicomponent multiphase materials to
diagrams is drastically simplified by the development of technical alloys and
the fact that all the interesting two- also to the optimization of near-
dimensional sections in temperature, equilibrium processing conditions [1]. This
composition or even chemical potential goes as far as setting the electronic carrier
can be readily calculated, plotted and concentration in CdTe semiconductors by
read directly, instead of trying to annealing at appropriate partial pressures
envisage or construct them from an of Cd [15]. Many successful applications
only graphically available set of phase rely on the fact that the actual process
diagrams and their complex conditions are close to equilibrium and a
projections. Moreover, for any selected large number of examples covering metals,
individual alloy, the amounts of the alloys, intermetallics, ceramics,
phases present and their compositions semiconductors, geological materials,
can be calculated as function of nuclear core materials or even waste and a
temperature. It may be rather variety of processes were given [1]. The
cumbersome to extract that information calculations of solidification paths under
if only a printed set of ternary phase equilibrium conditions or even with the
diagrams is given, and it is virtually popular Scheil model are also considered
impossible for quaternary or higher as very useful direct applications, since no
order systems. kinetic data are needed when completely
blocking the solid state diffusion in the
This powerful tool in materials research Scheil model [2]. An application to the
goes beyond a mere “calculation of phase magnesium alloys AZ91 and AZ62 was
diagrams”. This is why the term given together with different experimental
“computational thermochemistry” is methods [16].
frequently used to describe the current However, even materials
state of the Calphad approach. It is processing far from equilibrium (such as
depicted in Fig. 1 and will be detailed splat cooling, mechanical alloying,
below. annealing of reactive thin film multilayers)
requires a sound knowledge of
2.2 Applications and software thermodynamic data of equilibrium and
metastable phases [17,18,19]. Using the
Applications of the Calphad Calphad approach the driving forces for
approach can be classified in four metastable phase formation or even
categories: direct applications, coupling complete metastable phase diagrams can
with steady state reactor modeling, be calculated to provide "road maps" for
coupling with micro-kinetics, and coupling the engineering of these advanced
with macro-kinetics. Only the latter two materials.
require the knowledge of kinetic data of A coupling with steady state reactor
the materials system. modeling is done as follows. A reactor like
The direct applications of an electric arc furnace can be modeled by
information read from phase diagrams or splitting it up to a certain number of

4 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
reactor stages. Within each stage a (local) numerical methods [1, 2].
equilibrium calculation can be performed Thermodynamic software and data
using the thermodynamic data set of the sets are called as a "subroutine" to provide
materials system. In addition to the number local equilibrium calculations and the
of stages one has to specify the feed phases driving forces for diffusion. These
(amounts, compositions, temperatures, calculations yield quantitative data at
input stages), the enthalpy locally prevailing temperature, pressure,
supplied/evolved or the temperature composition or chemical potentials. It is
imposed on each stage, pressure of each advisable to use the "subroutine" approach
stage and optionally distribution and not to fully integrate the
coefficients of phases which may simulate thermodynamic and diffusion software.
kinetic inhibitions. Steady-state The thermochemical and phase equilibrium
equilibrium is attained by equilibrium calculations in real systems may become
calculations where the output of one stage rather complex and should be thoroughly
is used as input of the next stage in co- and tested separately for internal consistency
counter-current flow reactors. and accuracy. A prominent example of this
Convergence is usually achieved with approach is the DICTRA software [22],
reasonably estimated initial values for each just a few applications are mentioned here
stage obtained by separate equilibrium for the multicomponent diffusion in steel
calculations. A nice example is the [22,23].
production of metallurgical grade silicon in Coupling computational
an electric arc furnace, where the thermochemistry with macro-kinetics is at
ChemSage software was used [20]. It the front end of integration in materials
should be noted that actual kinetic data processing software. Comprehensive
(diffusion, fluid flow, ...) of the materials simulation of dynamic, macro-scale
system are not necessary in this approach. processes involving fluid flow constitute
The optionally imposed phase distribution the highest level of complexity and
coefficients are not considered as detailed requirement of additional system data. An
kinetic data, for example the imposed example is the simulation of technical
fraction of NH3 in MOCVD growth of alloy casting, where in up to three-
nitride semiconductors in a simple one- dimensional simulation of the moving
stage reactor [21]. mushy solidification front, void formation
A coupling of the Calphad at hot spots and macro-segregation may be
approach with micro-kinetics concerns covered. The other aspect of
mostly solid state transformations, growth microsegregation, dendrite shape
or dissolution of precipitates, reactions at formation and ultimately the development
solid/solid interfaces and so on. Fluid flow of microstructure is still mostly treated
and heat transfer problems do not occur or separately. Both aspects require
are irrelevant. Thermal equilibrium is often thermodynamic equilibrium calculations at
assumed since thermal diffusion is much the core of appropriate process simulation
faster compared to species diffusion in software that tries to solve the
solids. The framework of multicomponent simultaneous mass- heat- and momentum-
(multiphase) diffusion is used, which transfer with moving phase boundaries [2].
requires the full data set of diffusion Such process simulation software is
coefficients of all species in all phases. working in the framework of
Possibly even the nucleation may be computational fluid dynamics (CFD) or
included with an appropriate model. The flow-sheeting. Depending on the
phase growth is usually assumed to be complexity, the full set of data for the
diffusion controlled. Appropriate software macro-kinetic description of the system
is designed with analytical and/or may be needed, possibly including the

5 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
micro-kinetic data. The current challenge aiming at an automatic calculation of
is to incorporate exact thermodynamic multicomponent phase diagrams without
phase equilibrium calculation into the need to provide user knowledge of the
commercial software packages like system.
Phoenics, Fluent, Aspen, Process and so
on. To this aim a general software interface
TQ (ChemApp) has been developed that 2.3 Thermodynamic modeling of
allows to call the most widely used Gibbs multicomponent multiphase equilibria
energy minimizers as a subroutine [1, 24].
An example for the casting process, where The basic principle to find the equilibrium
the link to thermodynamic software was state of a multicomponent alloy (or
programmed with another software system) is to minimize its total Gibbs
interface, is given by [25]. If the simulation energy, G, at constant temperature T,
of mushy zone is also included in an pressure P, and overall composition:
attempt to couple macro- and micro-
kinetics, the computation time becomes a G = ∑ nφ Gφ = min
φ
major issue [26]. Another problem is that
the thermodynamic software subroutine (1)
φ
has to be 100% reliable and convergent for where G is the molar Gibbs energy of any
each individual complex equilibrium possible phase φ in the system and nφ is its
calculation even if starting from unrealistic amount. Minimization is done by
initial values set by the process software. distributing the given amount of alloy
Envisaging about 105 calls of the components onto the various available
thermodynamic subroutine, a single failure phases, subject to the overall material
return would obstruct the process balance. The stable equilibrium is
simulation. described by a phase assembly of all
The early versions of software to phases with nφ > 0, whereas the metastable
calculate multicomponent phase equilibria, phases (nφ = 0) do not materialize. Each
Solgasmix by Eriksson [27], and to phase has its own composition which is
calculate (mainly binary) phase diagrams, generally different from the fixed overall
the Lukas-programs [28], were given away alloy composition.
free as source code, which substantially We have thus to supply the
fostered the development of the field. More functions Gφ for all the phases. For the
general mathematical relations to calculate pure element i in the φ phase the Gibbs
phase diagrams were laid out by Hillert 0 ,φ
energy function Gi (T ) = Giφ (T ) − H iSER
[29] which were eventually programmed in
is described by the equation
Thermocalc [30]. Currently a number of
software packages, sometimes combined 0 ,φ
with databases, are commercially Gi (T ) = a + b ⋅ T + c ⋅ T ⋅ ln T + d ⋅ T 2
available. Frequently used are ChemSage + e ⋅ T 3 + f ⋅ T −1 + g ⋅ T 7 + h ⋅ T − 9
and Fact, recently merged to become (2)
FactSage, MTDATA, Thermocalc, and a
second generation of software, Pandat 1,

1
More and updated information is found at the MTDATA
corresponding websites: http://www.npl.co.uk/npl/cmmt/mtdata
ChemSage http://gttserv.lth.rwth- Pandat http://www.computherm.com/
aachen.de/gtt/ Thermocalc http://www.thermocalc.se/
Fact: http://www.crct.polymtl.ca/fact/fact.htm More Web Sites
http://www.crct.polymtl.ca/FACT/web
FactSage http://www.factsage.com/ sites.htm

6 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
c − 2 c −1 c
where H iSER is the molar enthalpy of the E
G tern,φ = ∑∑∑ xi x j xk {L1ijk,φ ( xi + δ ijk )
stable element reference (SER) at 298.15 K i =1 j > i k > j
and 1 bar. The values of the parameters a, + L ( x j + δ ijk ) + L3ijk,φ ( xk + δ ijk )}
2 ,φ
ijk
b, c,... are usually taken from the SGTE
compilation by Dinsdale [6]. (7)
c − 3 c − 2 c −1 c
E
G quat ,φ = ∑∑∑∑ xi x j xk xl {L1ijkl
,φ
( xi + δ ijkl )
Stoichiometric phases like the binary i =1 j > i k > j l > k
compound AmBn can be described by the
+ L ( x j + δ ijkl ) + L3ijkl,φ ( xk + δ ijkl ) + L4ijkl,φ ( xl + δ ijkl )}
2 ,φ
relation ijkl

(8)
m n where
G1A:2m Bn = ⋅ G10,φ + ⋅ G20,φ δijk = (1-xi-xj-xk)/3
m+n m+n
(9)
+ amn + bmn ⋅ T + ...
(3) δijkl = (1-xi-xj-xk-xl)/4
(10)
where the last two terms stand for the
Gibbs energy of formation of AmBn from It is noted that in a ternary system (c=3) δijk
the pure components 1(=A) and 2(=B) in = 0 whereas in a quaternary system with xl
the specified reference state φ. ≠ 0 the same term δijk ≠ 0. Instead of Eqs.
The molar Gibbs energy, Gφ, of a (7,8) one might have used seemingly
binary disordered solution phase φ, simpler equations where the δijk and δijkl
applicable for most liquids or substitutional terms would have been omitted. The main
solid solutions, is generally described as: advantage of the given formulation is that
Eq. (7) always reduces to the regular
Gφ = x1G10,φ + x2G20,φ + RT ( x1 ln x1 + x2 ln x2 ) solution contribution if all the three L-
+ x1 x2 L12 (
0 ,φ
+ L112,φ ( x1 − x2 ) + L12
2 ,φ
( x1 − x2 ) 2 + ... ) parameters are identical:
(4)
if
Hillert [31] pointed out the advantages of L1ijk,φ = L2ijk,φ = L3ijk,φ = Lφijk
this specific formulation [32] of the power (11)
series for the excess term to the ideal then
c − 2 c −1 c
solution when this equation is used for
extrapolation to ternary systems.
E
G tern , φ = ∑∑∑ xi x j xk Lφijk
i =1 j > i k > j
For a multicomponent solution
(12)
phase with c components the following
equation is used:
c c
even in a quaternary or higher order
G = ∑ xiG
φ
i
0 ,φ
+ RT ∑ xi ln xi + G E bin ,φ
system. Without using the δijk term in Eq.
i =1 i =1 (7) a residual quantity xixjxkLijk(1-xl) would
tern ,φ quat ,φ
+ GE
+ G
+ ...
E
be left in quaternary. This occurrence of xl
(5) would be incompatible with the regular
The excess contributions originating from interaction of the three species i-j-k,
all the binary interactions (EGbin) or ternary correctly given in Eq. (12). Such a regular
interactions (EGtern) or quaternary interaction can be derived from
interactions (EGquat) are: thermodynamic principles assuming
c −1 c n different pairwise i-j bond energies and an
E
G bin ,φ = ∑∑ xi x j ∑ Lvij,φ ( xi − x j ) v ideal entropy term [33]. The same
i =1 j >i v =0
reasoning is valid for the quaternary
(6)

7 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
interaction2. and G(Mg) can be calculated from Eq. (5),
This has been clearly pointed out setting EGtern = EGquat = 0. This enables
by Hillert [31, 33] and in fact the Eqs. (5- equilibrium calculations in the entire
10) are just extensions of his equations. quaternary alloy system. It is a general
The formulation is slightly different, the experience that the binary interactions are
present δijk (and δijkl) variables are invariant most prominent, that is, the higher order
to a permutation of indices whereas interactions like EGtern are usually small.
Hillert’s original variables (vijk) are not. In the next level of refinement these
It should be also noted that the calculations can be used to identify key
ternary and quaternary interaction terms in experiments at decisive conditions in the
Eqs. (7,8) provide a subregular behavior four ternary edge systems, either to verify
with a linear composition dependence that EGtern ≅ 0 or to determine quantitative
inside the {}-bracket. No higher terms values. In addition, solid solutions of
have yet been used in practice. In a binary compounds into the ternary systems
predictive approach from the known binary and true ternary compounds have to be
systems to a c-component system one modeled.
would not use any ternary or quaternary Using both EGbin and EGtern in Eq.
interaction terms. If those are used at all it (5) results not only in experimentally
certainly does not make sense to envisage a supported calculations in the ternary edge
higher than the subregular order given in systems, it also improves the reliability of
Eqs. (7,8). calculations in the quaternary alloy system.
Reviewing Eqs. (5-10) it becomes In the next level key experiments in the
evident what is really meant by the quaternary system can be identified.
predictive capability of this approach. Let Experience shows that generally EGquat ≅ 0
us, for simplicity, assume that we deal with if EGbin and EGtern are described properly.
a c=4 component magnesium alloy system This interactive approach
which comprises, in addition to “extrapolative calculations” – “key
stoichiometric compounds, just two experiments” – “supported modeling” is
solution phases, the liquid phase L and the also useful during assessment of the EGbin
(Mg) solid solution. Once EGbin of the six of individual binaries. In ternary or higher
binary edge system have been assessed, GL order systems, however, it is indispensable
considering the huge number of possible
2
It should be noted that the binary interaction of
experimental conditions that cannot be
Eq.(6) could equivalently be written in the same handled by a plain trial and error approach,
style as Eqs. (7,8): limited resources given.
c −1 c More complex solution phases
E
G bin ,φ = ∑∑ xi x j require models that take the specific
i =1 j > i
(13) physical nature of the interactions between
n
the components and/or the crystalline
∑A
v =0
v +1,φ
ij ( xi + δ ij ) n−v
( x j + δ ij ) v

structure of the phase into account. This

reduces the number of adjustable
where δij = (1-xi-xj)/2 parameters and can provide realistic
(14) functional behaviors which are out of the
The subregular behavior is given for the order n=1.
scope of Eq. (4), like a sharp bend in the
The conversion between the parameters Aijv +1,φ and Gφ-x curve. A classical example is the
Lvij,φ has been given by Hillert [1980Hil]. Since Wagner-Schottky model [34] for ordered
virtually all binary assessments are published in the intermetallic phases, which derives the
form of Eq. (6) instead of Eq. (13) it is reasonable composition dependence of the Gibbs
to keep this form in the general Gibbs energy energy from the generation and interaction
equation for a c-component solution phase. of point defects in the perfect crystal.

8 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
Substantial progress had been made on the thermodynamic data sets are marked with
thermodynamic modeling of solutions [35], grey background in Fig. 2. Two main paths
specifically for liquids [36], selected for the current Mg-alloy development are
intermetallic compounds [37], order- indicated by arrows. The first group of
disorder phase transitions [38], oxides [39], alloys are Li-containing, aiming at a
electronic materials [40], and more further density reduction and improved
advanced models for the composition ductility. The second group with Sc and
dependence of intermediate phases [41]. RE-additions aims at increased creep
resistance. In this paper the second group
is discussed as an example of focused alloy
3. Magnesium Database development.
Development To create the thermodynamic phase
descriptions the Calphad method is used as
3.1 Thermodynamic Modeling given above and depicted in Fig. 1. After
critically reviewing the experimental data
In order to use the Calphad and selecting the appropriate models for all
approach as a tool in magnesium alloy phases, the actual determination of the
design, a high quality thermodynamic thermodynamic parameters (like the amn,
database for Mg-alloys is needed. bmn, or Lvij,φ ) is done by multidimensional
Development of a reliable thermodynamic least square optimization. This
database for multicomponent alloys optimization is, depending on the
requires a combination of experiments and complexity of the system, done stepwise
computational thermochemistry with a until finally all parameters of that system
feedback of data from alloy application. are optimized simultaneously taking all the
Since numerous binary and ternary experimental information (phase equilibria
subsystems have to be treated and and thermodynamic quantities) into
validated with key experiments before account.
multicomponent alloys can be calculated
reliably, this development becomes a long- 3.2 Key Experiments coupled with
term project. In our group a modeling and applications
thermodynamic database for several
alloying elements like Al, Li, Si, Mn, Ca, In Mg-systems the experimental
Sc, Y, Zr and Rare Earth elements has literature data were very sparse in both,
been under construction for more than six phase diagram data and thermodynamic
years and is still ongoing. The base of this properties. Only a few ternary system were
13-component system are the binary investigated in more detail and calculated
systems. Figure 2 gives an overview of the thermodynamically [42,43,44]. An attempt
necessary systems and the current state of to apply these calculations for alloy
assessment. Highlighted in the first column development was published [45].
are the included elements. Also shown are Several experimental methods are
the current quaternary aim systems Al-Mg- being applied in our group to produce a
Li-Si and Mg-Mn-Sc-RE (with RE = Ce, sufficient experimental basis for a
Gd, Sc and Y).78 independent binary multicomponent Mg-database. For
systems are necessary as basis for a 13- preparation of highly reactive and
component system. 26 of them could be evaporating alloys containing Mg, Li and
taken from literature and included in our Ca the levitation melting technique was
database after checking the consistency. adapted. The main advantages are crucible
For 23 binary systems the thermodynamic free melting, suppressing of evaporation
descriptions were created by our group. with protective gas overpressure and visual
These system with existing consistent control of energy supply to avoid

9 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
overheating. Other preparation methods Investigations started with binary
include arc and electron beam melting for Mg-Sc alloys. Scandium was chosen for
extremely high melting intermetallic precipitation hardening because of its large
phases, see Table 1. solubility in (Mg) and the retrograde
During thermal analysis (DTA) the solubility at lower temperatures. The
alloys were protected in a sealed tantalum binary phase diagram had to be re-
crucible, Fig. 3. The development of investigated [61], it is shown in Fig. 4.
techniques for producing these crucibles However, binary MgSc precipitates form
and sealing the lids free of micro-cracks very slowly during ageing and improve the
were found to be essential for reproducible mechanical properties only slightly
DTA measurements of these highly because of their incoherent interface.
reactive and easily vaporizing alloys. For Therefore, Mn was added as second
phase and crystal structure analysis X-ray alloying element. The precipitation of
diffractometry at room and high Mn2Sc was predicted by thermodynamic
temperature, electron microscopy and calculations. Mn2Sc precipitations form
metallography are also used in a combined coherently and were found very efficient
approach. for improving creep resistance and
These techniques have been applied hardness. New MgSc15Mn1 or
in our work on ternary and quaternary MgSc6Mn1 alloys show about 100 times
alloy systems for improvement of creep better creep resistance than the best
resistance [46], thermal stability, spray commercial WE43 alloy at 350°C and 30
forming (Mg-Mn-Sc-RE, Mg-Al-Ca-RE) MPa [46].
and density reduction (Mg-Li-Al-Si, Mg- In spite of the good properties of
Al-Ca-Si) [47,48]. Thermodynamic this first generation of Mg-Mn-Sc alloys,
descriptions supported by experimental the high cost of Sc addition (6 wt.% Sc or
work of several subsystems were published more) initiated a search for a second
for Mg-Sc [49], Mn-Sc [50], Al-Sc [51], generation by investigating quaternary
Al-Gd-Mg [52], Gd-Li-Mg [53] , Ca-Li- systems.
Mg [54], Al-Li-Si [55,56] and Al-Ca [57]. Additional alloying elements Ce,
The thermodynamic and technical Gd, Y and Zr were considered for this
application of the ternary Al-Mg-Sc system purpose to achieve a sufficient quantity of
was also discussed [58,59]. suitable precipitates to improve mechanical
In the following the new cerium properties using a minimum of expensive
containing quaternary system Mg-Mn-Sc- alloy element addition. These element
Ce together with Mg-Mn-Sc-Gd, Mg-Mn- combinations Mg-Mn-(Sc, Ce, Gd, Y, Zr)
Sc-Y and Mg-Mn-Sc-Zr are shown as form a variety of quaternary systems and
example for the selection of new creep within those there is a huge amount of
resistant alloys using computational possibilities to select alloy compositions.
thermochemistry. More details of alloy Therefore phase diagram and other
selection in the quaternary systems Mg- calculations were performed to identify
Mn-(Sc, Gd, Y, Zr) will be published soon promising candidates.
[60].
4.2 Alloy selection in the Mg-Mn-Ce-Sc
system
4. First generation of creep
resistant ternary Mg-Mn-Sc-(Ce) For a first alloy selection several
vertical sections of phase diagrams in
alloys
various ranges of Mn and Ce were
calculated. For practical reasons the Mn
4.1 Ternary Mg-Mn-Sc alloys
content was limited at 1.5 wt.%. First

10 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
alloys were cast with 3 to 5 wt.% Ce and disregarded because of the very slow
up to 15 wt.% Sc. The reason of these high kinetics and the incoherence of this
alloying elements addition was the lack of precipitate. In other words the higher Sc-
knowledge about the amount and the content does not substantially improve the
mechanical effect of the precipitations that calculated amount of mechanically
are possibly formed during heat treatment. efficient precipitates. In fact, the
The calculated quaternary phase equilibria mechanical properties of this alloy with 9
are presented in two-dimensional sections wt.% Sc show no remarkable difference to
with constant Mn and Ce content. Fig. 5 that with 6.3 wt.% Sc [62].
shows a T-x section with constant 1.5 wt.% The amount of the favored Mn2Sc
Mn, 4.2 wt.% Ce from 0- 15 wt.% Sc. particles is mainly controlled by the Mn-
Primary crystallization of Mn2Sc phase can content. Therefore the Sc-content could be
be observed for a wide range of higher Sc decreased to economically and technically
contents. The L + Mn2Sc phase field and suitable amounts and possibly be supported
several different solid phases stable at by precipitates stemming from cheaper
lower temperatures can be seen. Rare Earth additions. In a second
A large variety of phase diagram generation of creep resistant alloys,
sections can be calculated using the promising quaternary candidates were
thermodynamic database. At this point the selected primarily by computational
question how to identify promising alloy thermochemistry.
candidates from all these calculated
diagrams becomes important. What phase
diagram features are related to what alloy 5. Second generation of creep
processing steps? What is needed is a list
of beneficial phase diagram features, resistant, low-scandium alloys
derived from the relevant alloy processing
steps. The most important of these points 5.1 Alloy selection in the Mg-Mn-Sc-Gd
are given in Table 2, they are applied in the system
following discussion.
Fig. 6 shows phase amounts during Several vertical sections in the
equilibrium solidification for an alloy with quaternary Mg-Mn-Sc-Gd system were
1.5 wt.% Mn, 4.2 wt.% Ce and 6.3 wt.% studied in the ranges of 0-1.5 wt.% Mn, 0-
Sc, indicated by dotted line in Fig. 5. Small 10 wt.% Sc and 0-10 wt.% Gd. The
amounts of primary Mn2Sc and calculated quaternary phase equilibria are
significantly larger amounts of CeMg12 are presented in two-dimensional sections with
expected in that alloy. In fact, in the constant Mn and Gd content. Figure 9
micrograph of this alloy after T5 heat shows a T-x section with constant 1 wt.%
treatment (aged after casting) fine grained Mn, 5 wt.% Gd from 0- 1 wt.% Sc. A
Mn2Sc associated with larger crystal of preferable primary solidification of (Mg), a
CeMg12 can be seen in Fig. 7 [62]. It will large one-phase field of (Mg) and several
be shown later that the finely distributed different solid phases stable at lower
Mn2Sc particles have a very positive effect. temperatures can be seen.
One alloy with higher Sc content was For an alloy with 1 wt.% Mn, 5
studied to promote the favored Mn2Sc wt.% Gd and 0.8 wt.% Sc (indicated by the
precipitations. The calculated phase right dotted line in Fig. 9) equilibrium
amounts in Fig. 8 for an alloy with 1.5 phase amounts during solidification and
wt.% Mn, 3.2 wt.% Ce and 9.0 wt.% Sc heat treatment are given in Fig. 10. At the
show no remarkable differences to the liquidus temperature of 650°C, primary
alloy with 6.3 wt.% Sc in Fig. 6. The low- (Mg) is formed and consumes the melt
temperature formation of MgSc should be totally down to the solidus point at 619°C.
At 590°C the first precipitate Mn2Sc starts

11 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
forming, which can be seen in the enlarged is formed and consumes the melt totally
insert in Fig. 10. Large amounts of the down to the solidus point at 613°C. The
second precipitate GdMg5 start forming at Mn12Y phase is only stable in the solid
425°C. Similar positive features are state in a high temperature range from 605
observed for an alloy with only 0.3 wt.% to 322°C and may not form at all during
Sc. The equilibrium phase amounts during fast cooling. The established Mn2Sc and
solidification and heat treatment for that the new Mg24Y5, even in very large
alloy with 1 wt.% Mn, 5 wt.% Gd and 0.3 amount, could form by ageing in a
wt.% Sc (indicated by the left dotted line in favorable temperature range. This makes
Fig. 9) are given in Fig. 11. The Mn23Sc6 MgMn1Y5Sc0.8 also a promising alloy.
phase is formed as first precipitate in the In order to check if the amount of
solid state. Both alloys fulfill all features Mn2Sc could be raised by increasing the
illustrated in Table 2 and were classified as manganese content to 1.5 wt.%, we can
very promising for further large scale creep examine Fig. 14 for the alloy
experiments. MgMn1.5Y5Sc0.8. This alloy is
In fact, first results of these alloys disqualified since Mn2Sc forms as the
show a creep resistance similar to the primary phase from the liquid, clearly seen
previous ternary high-scandium Mg-Mn-Sc in the inset of Fig. 14. (Mg) forms only
alloys, about 100 times better than best secondary, and even Mn12Y forms as a
commercial WE43 alloy at 350°C and 30 tertiary phase during solidification. Such a
MPa, as detailed later. microstructure cannot be “repaired” by
heat treatment. It would be difficult to
envisage such a drastic change caused by
5.2 Alloy selection in the Mg-Mn-Sc-Y raising the Mn-content from 1.0 to 1.5
system wt.% without the thermodynamic
calculations.
In the Mg-Mn-Sc-Y system several
vertical sections in the ranges of 0-1.5
wt.% Mn, 0-10 wt.% Sc and 0-10 wt.% Y 5.3 Alloy selection in the Mg-Mn-Y-Zr
were studied. Figure 12 shows a T-x system
section of a calculated quaternary phase
diagram with constant 1 wt.% Mn, 5 wt.% For complete substitution of
Y from 0- 1 wt.% Sc. A large one-phase expensive scandium, zirconium was
field of (Mg) and several different solid checked as a possible alloying element.
phases stable at lower temperatures can be The impact of such a substitution can be
seen, like in the corresponding diagram for studied in Fig. 15 for scandium-free alloys
Gd. Differences compared to the Gd with 1 wt.% Mn, 4.5 wt.% Y and 0-1 wt.%
system can be observed concerning the Zr. This phase diagram section shows a
stable solid phases and the formation very steeply rising liquidus line, 1000°C
temperatures. In the whole range of are reached for less than 0.1 wt.% Zr.
primary crystallization of (Mg) the Moreover, and actually the reason for the
secondary phase is Mn12Y. The desired steep liquidus line, a huge primary
Mn2Sc precipitate forms only at crystallization field, L + Mn2Zr, stretches
temperatures below 500°C or even lower over the entire composition range in Fig.
with decreasing Sc content. 14. Only for extremely small Zr-additions,
For an alloy with 1 wt.% Mn, 5 wt.% Y and not discernible in Fig. 15, is a primary
0.8 wt.% Sc, indicated by dotted line in (Mg) solidification expected. The reason
Fig. 12, phase amounts during equilibrium for this destructive phase diagram feature
solidification are given in Fig. 13. At the is the extremely high thermodynamic
liquidus temperature 644°C, primary (Mg) stability of Mn2Zr in comparison to the

12 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
other phases. Since yttrium does not play a thermochemical calculations were prepared
significant role in that part, the only way to by squeeze casting by our project partners
diminish the L+Mn2Zr primary field would within the Thrust Research Project SFB
be a drastic reduction of the manganese 390, “Magnesium Technology”, at the TU
content. But this would also drastically Clausthal. Yield strength and creep rates of
reduce the amount of beneficial Mn- these alloys were measured in the as cast
containing precipitates. As a result, the condition and also after heat treatment
entire quaternary Mg-Mn-Y-Zr alloy [62]. It is clearly seen from Fig.17 that the
system is disqualified. secondary creep rate at 350°C of our first
generation (high-scandium) alloys is better
by a factor of 100 compared to the WE43
5.4 Defining the commercial creep benchmark alloy.
resistant benchmark alloy The first generation of alloys
showed a strong annealing response due to
In order to assess the results on the the formation of the Mn2Sc precipitates.
creep resistance measurements of the The existence of the Mn2Sc precipitates
selected new alloys a commercial alloy has was confirmed by X-ray diffraction, SEM
to be chosen as a benchmark. The history and TEM investigations and energy
of commercially developed creep resistant dispersive X-ray microanalysis. A
alloys is briefly summarized in Tab. 3. micrograph showing finely dispersed
These alloys usually contain about 0.7 Mn2Sc precipitates in a MgSc6Mn1 alloy
wt.% Zr as a grain refiner. The influence of after T5 treatment is given in Fig. 18 [62].
the Rare Earths (RE) on the creep resistance At 400°C an increase of the
generally decreases in the series Nd (Pr) > secondary creep rate can be observed for
Ce> La [63]. the Ce-containing alloys, Fig. 19. This
The current end point of disadvantage is partly compensated by
commercial alloys is given by the WE higher creep ductility caused by significant
series containing Y and Rare Earths. In grain boundary gliding in the Ce-
fact, the maximum stress that can be containing alloys [62]. Comparing Ce-
tolerated at 200°C for 100 h and 0.2% containing alloys with different Sc-content,
elongation is highest for the WE series as no significant change in the creep rate can
seen in Fig. 16. The alloy WE43 with T6 be observed, Fig. 20. The result is in line
heat treatment was chosen as the with the calculations in section 4.2 and the
benchmark alloy. Die casting alloys like minimized Sc-contents of the second
AE42 or AS21, known to be more creep generation alloys.
resistant as standard AZ or AM alloys, are
not included in the comparison since they
rank even below the ZE alloys in Fig. 16. 6.2 Second generation of creep resistant,
low-scandium alloys

6. Alloy preparation and The second generation (low-

scandium) alloys were selected from the
creep resistance most promising candidates identified by
measurements the thermodynamic calculations given in
section 5 from the cerium-free alloys. The
6.1 First generation of Mg-Mn-Sc-(Ce) creep curves for these low-scandium
alloys quaternary alloys are shown in Fig. 21
[62]. Again, an almost 100 times lower
The most promising alloy creep rate is achieved at 350°C and 30
compositions identified by the MPa compared to the best commercial

13 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
alloy WE43. The best alloy be recognized before starting large-style
MgGd5Mn1Sc0.3 with the smallest experiments, thus reducing the
elongation after 15⋅104 s contains only 0.3 experimental effort to a reasonable
wt.% of expensive Sc. It is exactly one of volume. The next step, the experimental
the few most promising alloys indicated by study of mechanical properties of
the thermodynamic calculations. The identified promising alloys, has shown
calculated equilibrium phase amounts excellent results both in the first generation
during solidification and heat treatment for (ternary and high-scandium) and second
this alloy was shown in Fig. 11. This is generation (quaternary, low-scandium) of
considered to be a substantial progress new creep resistant alloys. Obviously,
compared to the first generation of alloys these experiments cannot be replaced by
with much higher scandium contents of 6 thermodynamic calculations. However,
or 15 wt.% Sc. considering the huge number of less
promising alloy combinations that could
have been selected by trial and error from
7. Conclusions multicomponent systems, the focused alloy
development following this approach
Focused magnesium alloy avoids a waste of time and effort.
development is possible using the powerful
tool of thermodynamic calculations. Alloy
compositions with promising possibilities Acknowledgement
of alloy microstructure design can be
selected by means of thermodynamically Part of this work is supported in the
calculated phase diagrams, phase amount »Thrust Research Project SFB 390:
charts and solidification curves. Most Magnesium Technology« by the German
importantly, element combinations and Research Concil (DFG).
compositions with unwanted properties can

References

[1] R. Kattner, G. Eriksson, I. Hahn, R. Schmid-Fetzer, B. Sundman, V. Swamy, A.

Kussmaul, P.J. Spencer, T.J. Anderson, T.G. Chart, A. Costa e Silva, B. Jansson, B.J.
Lee, M. Schalin, Calphad 2000, 24, 55-94.
[2] J Ågren, F. H. Hayes, L. Höglund, U.R. Kattner, B. Legendre, R. Schmid-Fetzer
Applications of Computational Thermodynamics, Z. Metallkd. 2001 (submitted).
[3] L. Kaufman, H. Bernstein, Computer Calculation of Phase Diagrams with Special
Reference to Refractory Materials, Academic Press, New York, 1970
[4] J. Hertz, J. Phase Equilibria 1992, 13, 450-458.
[5] U.R. Kattner, J. Metals 1997, (12), 14-19.
[6] A.T. Dinsdale, Thermochemical data of the elements, Calphad 1991, 15, 317-425.
[7] L. Kaufman, Calphad 1977, 1, 7.
[8] L. Kaufman, Calphad 1978, 2, 55.
[9] M.W. Chase, I. Ansara, A. Dinsdale, G. Eriksson, G. Grimvall, L. Hoglund, H.
Yokokawa, Calphad 1995, 19, 437-447.
[10] J. Ågren, B. Cheynet, M.T. Clavaguera-Mora, K. Hack, J. Hertz, F. Sommer, U.
Kattner, Calphad 1995, 19, 449-480.
[11] A. Chang, C. Colinet, M. Hillert, Z Moser, J.M Sanchez, N. Saunders, R.E Watson, A.
Kussmaul, Calphad 1995, 19, 481-498.

14 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
[12] D. de Fontaine, S.G. Fries, G. Inden, P. Miodownik, R. Schmid-Fetzer, C. Shuang-Lin,
Calphad 1995, 19, 499-536.
[13] B. Burton, T.G Chart, H.L. Lukas, A.D Pelton, H. Seifert, P. Spencer, Calphad 1995,
19, 537-553.
[14] F. Aldinger, A.F. Guillermet, V.S Iorich, L. Kaufman, WA. Oates, H. Ohtani, M. Rand,
M. Schalin, Calphad 1995, 19, 555-571.
[15] Q. Chen, M. Hillert, B. Sundman, W.A. Oates, S.G. Fries, and R. Schmid-Fetzer, J.
Electron. Mater. 1998, 27, 961.
[16] D. Mirkovic, J. Gröbner, R. Schmid-Fetzer, An Approach to determine Solidification
Curves of Commercial Magnesium Alloys in Magnesium Alloys and their Applications,
(Ed: K.U. Kainer), DGM, Wiley-VCH, Weinheim, 2000, 783-788.
[17] C. Gente, M. Oehring, and R. Bormann, Phys. Rev. B 48, 1993, 13244.
[18] R. Bormann, "Kinetics of Interface Reactions in Polycrystalline Thin Films," in:
Polycrystalline Thin Films, K. Barmak, M.A. Parker, J.A. Floro, R. Sinclair, and D.A.
Smith (Eds.), Mat. Res. Soc. Symp. Proc. 343, 1994, p. 169.
[19] R. Busch, F. Gaertner, C. Borchers, P. Haasen, and R. Bormann, Acta metall. mater.
1995, 43, 3467.
[20] G. Eriksson and K. Hack, "Production of Metallurgical Grade Silicon in an Electric Arc
Furnace," in: The SGTE Casebook, (Ed: K. Hack), The Institute of Materials, London,
1996, p. 200.
[21] A. Koukitu, N. Takahashi, T. Taki, and H. Seki, Jpn. J. Appl. Phys. 1996, 35, L673.
[22] J. Ǻgren, ISIJ International 1992, 32, 291.
[23] J. Ǻgren, "Multicomponent Diffusion in Compound Steel," in: The SGTE Casebook,
(Ed: K. Hack), The Institute of Materials, London, 1996, p. 209.
[24] G. Eriksson, H. Sippola, and B. Sundman, "A Proposal for a General Thermodynamic
Software Interface," in: Proceedings of the Colloquium on Process Simulation, Helsinki
University of Technology, Report TKK-V-B99, 1994, p. 67.
[25] D.K. Banerjee, M.T. Samonds, U.R. Kattner, and W.J. Boettinger, "Coupling of Phase
Diagram Calculations for Multicomponent Alloys with Solidification Micromodels in
Casting Simulation Software," In: SP97 Proc. 4th Intl. Conf. Solidification Processing
(Eds: J. Beech and H. Jones), University of Sheffield, UK, 1997, p. 354.
[26] M.C. Schneider, J.P. Gu, C. Beckermann, W.J. Boettinger, and U.R.Kattner, Metall.
Mater. Trans. A 1997, 28A, 1517.
[27] G. Eriksson, Chemica Scripta 1975, 8, 100-103.
[28] H.L. Lukas, E.Th. Henig, B. Zimmermann, Calphad 1977, 1, 225-236.
[29] M. Hillert, Physica 1981, 103B, 31-40.
[30] B. Sundman, B. Jansson, J.O. Andersson, Calphad 1985, 9, 153-190.
[31] M. Hillert, Calphad 1980, 4, 1-12.
[32] O. Redlich, A.T. Kister, Indust. Eng. Chem. 1948, 40, 345-348.
[33] M. Hillert, Phase Equilibria, Phase Diagrams and Phase Transformations: Their
Thermodynamic Basis, Cambridge University Press, Cambridge, UK 1998, 480-481.
[34] C. Wagner , W. Schottky, Z. Phys. Chem. 1930, B11, 163-210.
[35] M. Hillert, Calphad 1997, 21, 143-153.
[36] A.D. Pelton, M. Blander, M.T. Clavaguera-Mora, M. Hoch, L. Hoglund, H.L. Lukas, P.
Spencer, B. Sundman, Calphad 1997, 21, 155-170.
[37] I. Ansara, T.G. Chart, A.F. Guillermet, F.H. Hayes, U.R. Kattner, D.G. Pettifor, N.
Saunders, K. Zeng, Calphad 1997, 21, 171-218.
[38] G. Cacciamani, Y.A. Chang, G. Grimvall, P. Franke, L. Kaufman, P. Miodownik, J.M.
Sanchez, M Schalin, C. Sigli, Calphad 1997, 21, 219-246.
[39] M. Hillert, B Burton, SK Saxena, S Degterov, KCH. Kumar, H Ohtani, F Aldinger, A

15 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
Kussmaul, Calphad 1997, 21, 247-263.
[40] T. Andersson, D. de Fontaine, S.G. Fries, B. Legendre, W.A. Oates, R. Schmid-Fetzer,
H-J Seifert, Q. Chen, Calphad 1997, 21, 265-285.
[41] Ibrahim Ansara, Ben Burton, Qing Chen, Mats Hillert, Armando Fernandez-Guillermet,
Suzana G. Fries, Hans Leo Lukas, Hans-Jürgen Seifert, W. Alan Oates, Calphad 2000,
24, 19-40.
[42] S.-L. Chen, Y. Zuo, H. Liang, Y.A. Chang, Metall. Mater. Trans. 1997, 28A, 435-446.
[43] H. Liang, S.-L. Chen, Y.A. Chang, Metall. Mater. Trans. 1997, 28A, 1725-1734.
[44] H. Feufel, T. Gödecke, H.L. Lukas, F. Sommer, J. Alloys and Compounds 1997, 247 (1-
2), 31-42.
[45] Z.-K. Liu, Proceedings Magnesium Technology 2000, Kaplan, H.I., Hryn, J, Clow, B.
(eds.), The Minerals, Metals & Materials Society (TMS), Nashville, March 12-16, 2000.
[46] F. von Buch, B.L. Mordike, A. Pisch, R. Schmid-Fetzer, Material Scie. & Eng. A 1999,
263A, 1-7.
[47] U. Draugelates, B. Bouaifi, S. Steinborn, R. Schmid-Fetzer, A. Pisch, First Israeli
International Conference on Magnesium Science&Technology, Dead Sea, Israel,
November, 1997, 10-12.
[48] A. Pisch, R. Schmid-Fetzer, F. v. Buch, B.L. Mordike, P. Juchmann, F.-W. Bach, H.
Haferkamp,: Magnesium Alloys and their Applications, (B.L. Mordike, K.U. Krainer
(eds.), International Magnesium Conference - Wolfsburg, April 28-30, 1998, Werkstoff-
Informationsgesellschaft mbH, Frankfurt), 181-186.
[49] A. Pisch, R. Schmid-Fetzer, G. Cacciamani, P. Riani, A. Saccone, R. Ferro, Z. Metallkd.
1998, 89 (7), 474-477.
[50] A. Pisch, R. Schmid-Fetzer, Z. Metallkd. 1998, 89 (10), 700-703.
[51] G. Cacciamani, P. Riani, A. Saccone, R. Ferro, A. Pisch, R. Schmid-Fetzer,
Intermetallic, 1999, 7, 101-108.
[52] J. Gröbner, D. Kevorkov and R. Schmid-Fetzer, Z. Metallkd. 2001, 92, 22-27.
[53] D.G. Kevorkov, J. Gröbner, R. Schmid-Fetzer, V.V. Pavlyuk, G.S. Dmytriv, O.I,
Bodak, J. Phase Equilibria 2001, 22 (1) 34-42.
[54] J. Gröbner, R. Schmid-Fetzer, A. Pisch, C. Colinet, V.V. Pavlyuk, G.S. Dmytriv, D.G.
Kevorkov, O.I, Bodak, Metall. Mater. Trans. A 2001 under review.
[55] D.G. Kevorkov, J. Gröbner and R. Schmid-Fetzer, J. Solid State Chemistry 2001, 156,
500-505.
[56] J. Gröbner, D. Kevorkov and R. Schmid-Fetzer, J. Solid State Chemistry 2001, 156, 506-
511.
[57] D.G. Kevorkov, R. Schmid-Fetzer, Z.Metallkd. 2001, under review.
[58] J. Gröbner, R. Schmid-Fetzer, A. Pisch, G. Cacciamani, P. Riani, N. Parodi, G. Borzone,
A. Saccone, R. Ferro, Z. Metallk. 1999, 90 (8), 72-880.
[59] A. Pisch, J. Gröbner, R. Schmid-Fetzer, Material Scie. & Eng. 2000, A289, 123-129.
[60] J. Gröbner, R. Schmid-Fetzer, J. Alloys Comp. 2001, under review.
[61] A. Pisch, R. Schmid-Fetzer, G. Cacciamani, P. Riani, A. Saccone, R. Ferro, Z.
Metallkd. 1998, 89 (7), 474-477.
[62] F.v.Buch, “Entwicklung hochkriechbeständiger Magnesiumlegierungen des Typs Mg-
Sc(-X-Y), Mg-Gd und Mg-Tb“; Thesis TU Clausthal 1999.
[63] M.M. Avedesian, H. Baker (Eds), ASM Specialty Handbook Magnesium and
Magnesium Alloys, ASM International, Materials Park, Ohio 1999

16 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
Tab. 1: Methods for key experiments in Mg-database development
Sample Preparation Levitation Melting
Arc Melting
Electron beam Melting
Reaction Sintering
Sample Analysis Thermal Analysis (DTA, STA)
X-ray Diffractometry (XRD, 25-1500°C)
Electron Microscopy (SEM/EDX/WDX)
Metallography, Optical Microscopy

Table 2: Beneficial phase equilibrium features and their relevance for (Mg)-alloy
processing

Phase diagram feature Relevance for alloy processing

Liquidus temperature << 800°C Simple and cheap melting process
Primary phase field L + (Mg) Primary solidification of (Mg) matrix,
avoid primary growing precipitates
(Mg) single-phase field Enables homogenization annealing
Solid state (Mg) + precipitate(s) field with Enables microstructural engineering
suitable temperature range and amounts

Table 3: Commercial creep resistant Mg-alloys

Alloy Composition (mass%) Year
Zn Ag Y RE
ZE41 4.2 1.2 1947
EZ33 2.7 3.3
EQ21 1.5 2.1 1959
QE22 2.5 2.1
WE43 4.0 3.4 1979
WE54 5.2 3.0

17 of 26
 Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

900
842°C [3, 4]
800

700
Temperature in °C

600
(βCa)
Liquid
500

400 369°C

300
233°C 242°C

200 181°C
145°C
(αCa)
CaLi2

100 (Li)

0
0 20 40 60 80 100
Ca at.% Li Li

Fig. 1: Calculated Ca-Li phase diagram

compared to experimental literature data [3, 4]

18 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
Mg
100
(Mg) + (Mg)
Ca(Li,Mg)2 Three-phase
150°C
80 Two-phase
One-phase
CaMg2
Mg

60
%
at.

40 (Li)
(αCa) + (Li) +
Ca(Li,Mg)2 Ca(Li,Mg)2
20

0
0 20 40 60 CaLi2 80 100
Ca (αCa) Liquid Li
Liquid +
at.% Li Ca(Li,Mg)2

Fig. 2: Ca-Li-Mg phase diagram at 150°C:

Calculated isothermal section including the
investigated sample compositions.

19 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

Mg
250°C 100
(Mg)
(Mg) +
Ca(Li,Mg)2
80

CaMg2
Mg

60
(Li)
%
at.

(Li) +
Ca(Li,Mg)2
40
(αCa) +
Ca(Li,Mg)2

20

Liquid
0
0 20 40 60 80 100
Ca Liquid + (αCa) Liquid + Li
at.% Li
(αCa) + Ca(Li,Mg)2 Ca(Li,Mg)2

Fig. 3: Ca-Li-Mg phase diagram at 250°C:

Calculated isothermal section

20 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
Mg
500°C 100
(Mg)
(Mg) + Ca(Li,Mg)2
Liquid + (Mg)
80
Mg
CaMg2 (Li)
%
at.

60

Liquid + Ca(Li,Mg)2

40

Liquid + (βCa)
20
Liquid

0
0 20 40 60 80 100
Ca (βCa) Li
at.% Li

Fig. 4: Ca-Li-Mg phase diagram at 500°C:

Calculated isothermal section

21 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
Mg
100

60
0 (Mg)
80 YY
E

YY
CaMg2 YY (Li)
Mg

60
%
at.

40 600

Ca(Li,Mg)2
500
YY

20
50

400
0
60

U
70

300
YY
YY
0

0
0 CaLi2
(βCa) 20 40 (αCa) 60 80 100
Ca Li
at.% Li

Fig. 5: Calculated Ca-Li-Mg liquidus surface

22 of 26
 Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)
800
[DTA, this work]
710°C
700

Liquid
Temperature in °C

600

500 Liquid +
Ca(Li,Mg)2

400

Ca(Li,Mg)2
300

242°C
200
0 10 20 30 40 50 60

at.% Li
CaMg2 CaLi2

Fig. 6: Calculated pseudobinary section

CaMg2 - CaLi2 including own DTA
results of liquidus and solidus.

23 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

700 [DTA, this work]

Liquid
600
Liquid
Liquid + + (Mg)
Liquid
Temperature in °C

500 Ca(Li,Mg)2
+ (β Ca)
(Mg)+
400 Ca(Li,Mg)2
(Li)+
Ca(Li,Mg)2 (Mg)+(Li)+
300 Ca(Li,Mg)2

200 (αCa) + Ca(Li,Mg)2

100
0 10 20 30 40 50 60 70 80
Ca 87 Ca 0
at.% Mg
Li 13 Li 13
Mg 0 Mg 87

Fig. 7: Calculated ternary phase diagram

section at 13 at.% Li

24 of 26
Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

Liquid + (Mg)
Liquid
+ (Mg) + (Li) Liquid
Temperature in °C

(Mg)
(Li) Liquid
+ (Li)

(Mg) +
Ca(Li,Mg)2

(Li) +
Ca(Li,Mg)2

20 40 60 80 Ca 0.1
Ca 0.1
Li 0 at.% Li Li 99.9
Mg 99.9 Mg 0

Fig. 8: Calculated ternary phase diagram

section Mg-Li with constant 0.1 at.% Ca.

25 of 26
 Rainer Schmid-Fetzer, Joachim Gröbner: Focused Development of Magnesium Alloys using
the Calphad Approach. Adv. Eng. Mater. 3, 947-961 (2001)

Liquid + (Mg)
Liquid Liquid
+ (Mg) + (Li)
Temperature in °C

Liquid +
(Mg) + Liquid
Ca(Li,Mg)2 + (Li)

Liquid +
(Li) +
Ca(Li,Mg)2
(Mg) +
Ca(Li,Mg)2

(Li) +
Ca(Li,Mg)2

20 40 60 80
Ca 0.5 Ca 0.5
Li 0 at.% Li Li 99.5
Mg 99.5 Mg 0

Fig. 9: Calculated ternary phase diagram

section Mg-Li with constant 0.5 at.% Ca.

26 of 26