Materials Science and Technology

ISSN: 0267-0836 (Print) 1743-2847 (Online) Journal homepage: https://www.tandfonline.com/loi/ymst20

Understanding the yield behaviour of L12-ordered

alloys

Y. M. Wang-Koh

To cite this article: Y. M. Wang-Koh (2017) Understanding the yield behaviour of L12-ordered
alloys, Materials Science and Technology, 33:8, 934-943, DOI: 10.1080/02670836.2016.1215961

To link to this article: https://doi.org/10.1080/02670836.2016.1215961

Published online: 31 Aug 2016.

Submit your article to this journal

Article views: 1864

View related articles

View Crossmark data

Citing articles: 10 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ymst20
 Understanding the yield behaviour of L12-
ordered alloys
Y. M. Wang-Koh∗
Nickel superalloys exhibit a remarkable characteristic. Their yield stress that required to cause the
onset of plastic deformation increases with temperature. This typically occurs up to a temperature
of around 800°C. This effect is thought to originate from the precipitates of the microstructure, which
have an L12-ordered crystal structure. A number of other L12-based alloys exhibit similar yield
properties. It is generally accepted that this is caused by the exhaustion of dislocations by
cross-slip from {111} glide planes to {010} planes on which they are sessile. However, the
underlying mechanisms that control this cross-slipping process are yet to be fully understood,
with little consistency between empirical results and theory. A critical review of the various
theories surrounding nickel superalloys is offered.
Keywords: Anomalous yield, Dislocations, Superalloys, Intermetallics, Cross-slip, Anti-phase boundary

This review was submitted as part of the 2016 Materials Literature Review Prize of the Institute of Materials, Minerals and Mining run by the Editorial
Board of MST. Sponsorship of the prize by TWI Ltd is gratefully acknowledged.

Introduction decade since the inception of the first engine in 1940.

Ni-based superalloys are the industry standard
materials for applications in both marine and aerospace
turbine engines, where resistance to creep, fatigue defor-
mation, corrosion and oxidation, is required. These
materials possess a unique ability to maintain material
properties within the harsh environment and tempera-
tures experienced during operation. This is thought to
be due to precipitates present in their microstructure.
The peak temperature within the turbine of an aircraft
engine exceeds that in which titanium alloys are useful
due to their poor oxidation resistance. Conversely,
excellent oxidation and creep resistance of ceramics is
offset by poor toughness and ductility. Ni-based super-
alloys have emerged as the choice material for such
applications.
A gas turbine is composed of four sections: the com-
pressor, combustion chamber, turbine and exhaust noz-
zle. Hot gases drawn in by the compressor are ignited
in the combustion chamber and expand within the tur-
bine, which extracts mechanical work in order to drive
the compressor. The gas turbine is an example of a
heat engine, whose efficiency is given by: η = 1 − (T1/
T2), where T1 and T2 are the initial and final tempera-
tures, respectively. The most practical way to increase
engine efficiency is to try and raise T1, the temperature
at which gases enter the turbine, known as the turbine
entry temperature (TET).
Great emphasis has been placed on increasing the TET,
which has led to improvements of around 100°C every
Department of Materials Science and Metallurgy, University of Cambridge,
27 Charles Babbage Road, CB3 0BN, UK
∗
Corresponding author, email ymw24@cam.ac.uk
The highest TET achieved by a gas turbine is around
1600°C.1 This has placed great demand on the turbine,
and it is here that yield properties are most critical for
high stress applications. To further improve strength, the
volume fraction of precipitates in the superalloys has
increased in both the turbine discs and blades.
As the volume fraction of γ′ of these alloys is increased,
deformation mechanisms that were thought to be only a
feature of turbine discs have begun to appear in the blades.
Although not a primary limiting factor in blade alloy
design, yield is becoming increasingly important,
especially where stress concentrations result in low cycle
fatigue challenges, such as at the turbine blade root.
There are many observations of dislocation pairs in two-
phase alloys. Through transmission electron microscopy
(TEM), the anti-phase boundary (APB) energies on the
active slip plane and cross-slip plane have been measured
in an attempt to deduce the deformation mechanism.
However, the mechanism governing the locking and
unlocking processes identified in single-phase intermetal-
lics is yet to be confirmed nor translated across to two-
phase superalloys.
The L12 phase
Nickel superalloys consist primarily of two phases: a Ni3Al
γ′ -phase is present in the microstructure as precipitates,
often cuboidal (Fig. 1), surrounded by a pure Nickel matrix
γ phase. This γ phase has a face-centred cubic (f.c.c.) struc-
ture, compared to the Ni3Al, γ′ phase displays the primitive
cubic, ordered, L12 crystal structure, with Al atoms at the
cube corners and Ni atoms at the centre of each face (Fig.
2). The properties of the superalloys are thought to depend
on the coherency of the γ/γ′ interface, as will be shown in
‘Effect of lattice misfit’. A distinct cube–cube orientation

© 2016 Institute of Materials, Minerals and Mining

Published by Taylor & Francis on behalf of the Institute
Received 29 January 2016; accepted 15 July 2016
934 DOI 10.1080/02670836.2016.1215961 Materials Science and Technology 2017 VOL 33 NO 8
 Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys

has consequences on the deformation mechanisms of

Ni3Al.
As the dislocation penetrates further into the γ′ precipi-
tate, it experiences resistance as the width of the APB
grows. The passage of a second dislocation with the
same Burgers vector would remove the APB, restoring
the structure to an undefected state. Therefore, dislo-
cations move in pairs through γ′ , as supported by images
from TEM.

The cross-slip process

The yield stress anomaly
1 The microstructure of a single-crystal superalloy after
Unlike in most alloy systems, the Ni superalloys and a
standard heat treatment showing cuboidal γ′ precipitates
select group of L12-based compounds exhibit a remark-
encompassed within a pure Ni γ-matrix2
able characteristic: their yield stress, defined here as the
stress required to initiate plastic deformation by dislo-
relationship exists between γ and γ′ , which can be described cation glide, increases or shows minimal variation with
as follows: increasing temperature, from room temperature up to
around 800°C, before dropping off rapidly. Pure nickel,
{100}g //{100}g′
like most materials, exhibits a decrease in yield strength
with increasing temperature, suggesting the remarkable
, 010 .g // , 010 .g′ mechanical properties stem from the L12-ordered phase.
However, a comparison between alloyed-γ′ and a two-
phase Ni superalloy, Mar-M200, shows the increase in
If the lattice misfit between the γ and γ′ phases is not too
strength is much more substantial in the former (Fig. 3).
large, the γ/γ′ interface remains coherent, minimising the
Through the experimentation of various L12 alloys, the
interfacial energy. The γ′ phase may deform while remain-
yield stress anomaly is principally characterised by the
ing geometrically compatible with the γ-phase.
following features:
In an f.c.c. material, the slip systems are of the type
(i) Below the peak temperature, where the yield stress is
a/2<110>{111}. The Burgers vector of a dislocation
increasing with increasing temperature, the active slip
must be a lattice vector and is normally the shortest
system is predominantly <110>{111}. Beyond the
vector. However, a/2<110> is not a lattice vector in
peak temperature, the active slip system predominantly
the ordered L12 structure of Ni3Al. If a dislocation
becomes <110>{010}.
with said Burgers vector were to glide through, it
(ii) The yield stress usually begins to increase beyond
would move Ni atoms onto Al sites, and Al atoms
room temperature, though exceptions do exist; the
onto Ni sites, resulting in the formation of ‘forbidden’
yield stresses of Ni3Ge5 and Ni3(Si, Ti)6 are known to
Ni–Ni and Al-Al bonds. The slip plane becomes a
increase from liquid helium temperatures (77 K).
boundary between two γ′ structures. This boundary is
(iii) The peak position is dependent upon load
known as an APB. With higher energy Ni–Ni and
orientation.
Al–Al bonds replacing favourable Ni–Al bonds, an
(iv) The yield stress is reversible; prior deformation at
APB requires energy to form. In Ni3Al, APBs on the
T1 and then at T2 < T1, exerts only a modest effect
{111} plane have an energy of around 0.1 J m−2.3 This
higher than a virgin sample deformed at T2.
(v) The yield stress is strain rate insensitive.

2 The crystal structures of (left) fcc γ-phase and (right) L12

γ′ -phase. Substitution of Ni for Al in the L12 crystal struc-
ture reduces the shortest Burger’s vector from [110] to 1/2 3 A comparison of the CRSS of the Ni-based superalloy Mar-
[110]. Two shears are required to put the atoms back to M200 and the individual constituent phases of a superal-
their correct atomic positions loy. Adapted from Pollock and Tin4

Materials Science and Technology 2017 VOL 33 NO 8 935

Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
(vi) The work hardening rate is abnormally high, peaks
within the temperature range where anomalous yield is
observed, and is orientation-dependent.
(vii) The yield stress anomaly is not a feature of all L12
alloys.
It is thought that all of these characteristics must be
included into any mechanism describing the family of
L12 alloys. This is reflected in the significant effort spent
iterating various models to account for each principal
characteristic. However, recent experiments have shown
that for stoichiometric single-crystal Ni3Al the yield stress
is independent of alloy stoichiometry and orientation-
independent.7–9 This demonstrates complexities involved
in producing a coherent, adaptable mechanism to explain
such atypical phenomenon.
Kear–Wilsdorf locking
One common agreement between all theories is the cause of
the positive temperature dependence is due to the cross-slip
of dislocations from octahedral {111} planes onto cubic
{100} planes. Since {100} is not a glide plane, dislocations
in this cross-slipped configuration become immobile, form-
ing Kear–Wilsdorf locks.10 Dislocations find it favourable
to cross-slip onto {100} planes because the APB energy is
lower on these planes compared to the active {111} slip
plane. With increasing temperature, a greater proportion
of dislocations become immobile by cross-slip, resulting
in an increase in strength. Slip activation on the cube planes
is thought to lead to the decline in stress past the tempera-
ture corresponding to peak stress in L12 alloys.
However, this explanation for the deformation behav-
iour is incomplete and unable to explain some character-
istics thought to be principle to the yield stress anomaly of
L12 alloys. Namely, the anomalous increase in stress
begins as low as −196°C for some alloys,5,6,11 far too
low for a thermally activated mechanism.
Modelling
Pinning models
Various models have attempted to explain the above
characteristics, all of which are thought to be principal
features of anomalous yielding, but have been met with
difficulty in relation to experimental observations. The
earliest model proposed a mechanism where screw dislo-
cations cross-slip from {111} planes, on which they are
mobile, to {100} planes, becoming immobile; the driving
force being the differential in APB energy between {111}
and {100} planes.12
The positive temperature dependence is attributed to an
increased propensity for {010} slip with increasing
4 Schematic of the successive positions for a dislocations
moving on the (111) plane by bowing between pinning
points14
936 Materials Science and Technology 2017 VOL 33
temperature, causing a transition in the mechanism con-
trolling flow stress: from exhaustion hardening at low
temperature, where mobile dislocations are immobilised
by interactions with any forest dislocations, debris and
solute impurities, to debris hardening at high tempera-
ture; dislocations are mobilised by multiplication from
old sources and the activation of fresh ones. No transition
of mechanisms was found to account for the increase in
yield stress and an alternate theory was proposed by
Takeuchi et al., which has now become known as the
TK model, in which segments along the length of a
screw dislocation cross-slip, becoming pinning points
(Fig. 4).13 The temperature dependence is accounted for
by defining the probability of cross-slip as a function of
temperature:


Ht
f = f0 exp −
kT
where f is the probability of cross-slip, f0 is the probability
constant, Ht is the activation energy of cross-slip, k is
Boltzmann’s constant and T is the temperature. This
suggests a greater number of pinning points along a dislo-
cation length are generated at decreasing pinning point
spacing as the temperature increases, and a higher level
of effective stress is required to keep dislocations in a
steady-state motion.
These pinning points unlock when forces either side of
the kink cause the segments of dislocations adjacent of the
pinning point to collapse. At a critical angle, unpinning
would occur athermally, with the dislocation free to
glide again until further cross-slip occurs. Paidar et al. 15
developed this model further, accounting for why the dis-
location segments become immobile, proposing the APB
anisotropy originated from the spread of the dislocation
core into a non-planar configuration. They were then
able to explain the tension/compression asymmetry and
orientation dependence, though adapting the unpinning
mechanism to a thermally activated process.
The tension–compression asymmetry of the critically
resolved shear stress (CRSS) was justified by the Escaig
5 Schematic of compact cross-slip mechanism, also known
as Escaig mechanism. The dislocation immediately re-dis-
sociates onto the cross-slip plane, represented by the
darkly shaded region bound by points A and B17
NO 8

mechanism for cross-slip, shown schematically in Fig. 5.16
A pair of partial dislocations is compressed against each
other on the primary slip plane. The Escaig mechanism
analyses the dislocation core width of this pair before
and after cross-slip. If it is energetically favourable for
the faulted ribbon between the two partials to expand
more widely on the cross-slip plane than the slip plane,
the total superpartial pair constricts, before spon-
taneously cross-slipping and dissociates on the cross-
slipped plane. The difference between the normalised
stresses on the slip and cross-slip planes gives rise to the
tension/compression asymmetry. This asymmetry, a key
characteristic of the anomalous yield phenomenon,
would have been missed by Takeuchi and Kuramoto
since their experiments involved only compression tests.
A further contribution to the energy balance of cross-
slip was made by Yoo and co-workers.18 The elastic aniso-
tropy exerts an additional torque on a dislocation pair. In
the case of two parallel screw dislocations, this torque
facilitates the cross-slip from {111} onto {010}, where
the torque disappears due to symmetry.
Both the TK and PPV models have been adopted to
explain anomalous yield because of their strong agree-
ment with the macroscopic, mechanical observations
associated with anomalous yield behaviour. However,
currently no TEM investigations – neither in situ nor
post-mortem samples have succeeded in producing
further evidence in support of the pinning process.
Instead, the microstructures of L12-based alloys exhibit
long, screw dislocations adopting either incomplete or
complete Kear–Wilsdorf (KW) locking configurations.19
Furthermore, although the model of Pope et al. does a
good job rationalising the sensitivity of flow stress to crys-
tallographic orientation, it struggles to account for the
independence of strain rate to the yield stress. A balance
in activation energy between the formation and unlocking
of dislocation-locks needs to be established because it is
the activation energy of the unlocking mechanism that
is responsible for the strain rate sensitivity. Microstruc-
tural observations of dislocation features that are left
behind after flow show small, elongated dipole loops.20
Explaining their formation led to the development of an
alternative unlocking mechanism and different set of
models.
Kink expansion models
The presence of long, straight screw superdislocations is
one of the most significant microscopic observations in
L12 compounds. Edge dislocations mainly act as links
between long screw dislocations and their cross-slipped
parts, and are mobile.21 These step segments have been
termed ‘superkinks’.19 These superkinks can then shuffle
along the dislocation, forming long segments of screw
orientation cross-slipped on (010) planes, which are
locked by KW locks.
6 The steady-state configuration for the unlocking–locking
sequence. Unlocking occurs at A by superkink AB and
progresses through the sequence 1 to 6. The screw is
eventually locked again along CD21
Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
The steady-state configuration for the unlocking–
locking sequence is shown in Fig. 6. The locks are
considered to be stronger, and harder to unlock, than the
pinning points of the PPV model. Unlocking occurs by a
process that is thermally activated but also involves a
large athermal component, to account for the small strain
rate sensitivity.14 A sudden increase in strain rate decreases
the critical length, producing a large number of mobile
superkinks, rapidly increasing the number of mobile super-
dislocations. These superdislocations would then move
rapidly to accommodate the new strain rate level.
Hirsch’s model is able to explain the orientation
dependence and tension/compression asymmetry and
non-Schmid effects by adopting the activation enthalpy
and Escaig mechanisms of the PPV model, despite the dif-
fering origins of the locked screws configuration between
the two models. Furthermore, the Escaig model fails to
quantitatively explain the orientation effects because
Escaig himself admits that the empirical data used to sup-
port the model is unreliable.16 Nevertheless, it must be
emphasised that this is not the focus of Hirsch’s model,
rather it does a good job of explaining the small strain
rate dependence, another key feature of yield behaviour
in L12-based compounds.
Hirsch’s model is deemed one of the most successful
superkink unlocking models, due to the depth of detail
describing the locking and unlocking mechanisms. Unlike
the PPV model where unlocking is a purely athermal
process, unlocking occurs via a hybrid thermal/athermal
process. Furthermore, PPV assumed that jogs move
slowly and screws are pinned only locally compared to
the fast movement of jogs on {010} planes, which requires
a finite, albeit unknown, amount of activation energy and
thus can only occur above a certain temperature.
However, both the PPV pinning model and Hirsch’s
superkink model possess two underlying assumptions
that hinder their ability to account for all empirical fea-
tures of yield behaviour. First, both models assume a
characteristic length from which no basis is explained;
for the PPV model, this is the characteristic distance
between pinning points, whereas for superkinks, it is
the length of the superkinks that is assumed to be con-
stant. Second, the models are based on a steady-state
fluctuation. This means the density of obstacles along a
superdislocation remains constant, and only upon reach-
ing a critical stress, can these superdislocations become
unlocked. Beyond this stress, all obstacles become
mobile and are eliminated simultaneously. This would
translate to a sharp discontinuity in the flow stress at
the yield point. However, yielding in L12-based com-
pounds occurs gradually. Furthermore, since all the
cross-slipped segments dissolve at a critical stress, flow
continues indefinitely at a uniform stress. Empirically,
single crystals exhibit significant work hardening beyond
yield.
As a result, the superkink model was revised.14,22 The
adapted model assumes the superkink height is statisti-
cally distributed and must lie between upper and lower
bound values to become mobile. Above the upper
bound critical height, the superkinks not only become
mobile but can also generate new superkinks. Below the
lower value, the superdislocation is locked. An increase
in temperature shifts these critical values so that a greater
portion of dislocations are locked than mobile upon
increasing temperature. The mechanism and causation
Materials Science and Technology 2017 VOL 33 NO 8 937
Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
for the superkink height variation with temperature how-
ever is not clear.
The mobility of a screw dislocation is proportional to
the average height of the superkinks on it. If the average
superkink height falls below a critical value, a mobile
dislocation will experience ‘exhaustion of mobility’
and cease. The mobile dislocation density and kink
height evolution are dependent on one another as well
as temperature and applied stress. Including these revi-
sions, the prediction of L12 compounds’ yield behaviour
improves. Some discrepancies between experimental and
theoretical results still exist. Since the density of mobile
superdislocations can change rapidly, the dynamic
changes in superdislocation mobility needs to be fac-
tored into the model. However, at the time of design,
this could not occur due to the limited experimental
understanding of the dynamical process of exhaustion
in L12 compounds.
This section highlights the inconsistencies present
between proposed models and experimental observations.
The origin of the yield anomaly by exhaustion of dislo-
cation motion remains to be elucidated. In particular,
the specific mechanisms and causality of dislocation lock-
ing and unlocking. Significant work is required, both
experimentally and theoretically before an appropriate
model should be proposed.
Anomalous yield in intermetallic L12
alloys
Stoichiometric, single crystals of Ni3Al are difficult to syn-
thesise due to the peritectic reaction during solidification,
described by the reaction: liquid + NiAl → Ni3Al, because
grains that nucleate during the peritectic reaction are diffi-
cult to suppress.23 Elemental additions to Ni3Al or devi-
ations from stoichiometry towards the Ni-rich side avoid
this reaction,24 and alloy designers have considered repla-
cing some high-temperature, high-strength alloys with
intermetallics due to the similarities between the γ′ -phase
7 Schematic of the possible core configurations. The oval
shape represents the core of the ½[−101] superpartial.
When the APB lies on the (111) plane, the core also lies
on that plane (top position); here the superpartial is glis-
sile. When the APB lies on the (010) plane, the core
spreads onto the (1−11) or (111) plane (middle and bottom
ovals, respectively), depending upon the position of the
core. In both of these configurations, the superpartial is
sessile15
938 Materials Science and Technology 2017 VOL 33
and pure Ni3Al. As a result, the flow stress behaviour of
many Ni3(Al, X) alloys has been analysed.25–29
Various hypotheses relating the mechanical properties
to the yield stress anomaly have been proposed, though
the wide variety of factors thought to affect the anomaly
makes forming a coherent argument difficult. Neverthe-
less, it is commonly agreed that the yield stress anomaly
is a result of competition between the exhaustion (lock-
ing) and multiplication of mobile dislocations. The most
commonly adopted theories are outlined below, with
accompanying experimental justifications. It should be
noted here that for alloys in which the yield stress increase
in accompanied by a change on long range order par-
ameter, such as Cu3Au30 and β-brass,31 the following
mechanisms and discussion do not apply and have been
omitted from the current discussion.
The Kear–Wilsdorf locking mechanism is the most
commonly quoted explanation for observed, anomalous
yield behaviour in L12-based alloys. Post-mortem TEM
studies of dislocations lead to quick judgement calls on
what controls anomalous yield. Ni3Ga is the intermetallic
from which the TK model is founded. Similar to Ni3Al,
the yield stress of Ni3Ga under compression reaches a
maximum around 425°C, though the increase in yield
stress is steeper than Ni3Al.32 This behaviour was
accounted for by a higher {111} to {010} APB energy
ratio in Ni3Ga compared to Ni3Al, though no measure-
ments or theoretical calculations were conducted and
the justification seems dubious.
APB energies are notoriously difficult to calculate, both
theoretically and experimentally. The most common
method experimentally is to measure the dissociation
widths between partial dislocations by TEM under
weak-beam condition; this technique is thought to give
the best compromise between adequate resolution and
comfortable working conditions. However, sources of
error make it difficult to ascertain exact values for APB
energies. One problem arises from the fact that any
observed dislocation configurations may not be under
complete equilibrium. Furthermore, since images are
taken under diffracting conditions, a dislocation is shifted
with respect to its line and the shift is unique to each par-
tial, the resultant separation does not correlate directly
with the actual separation in the crystal.33 Baluc et al.
have shown the APB energy ratio is often overestimated
prior to considering contributions from elastic
anisotropy.34
Dislocation cores
The immobilisation of screw dislocations, observed by
TEM, was thought to be analogous to the plastic defor-
mation of body-centred cubic (b.c.c.) metals.35,36 It is
now commonly accepted that low temperature plastic
deformation of b.c.c. metals is controlled by screw dislo-
cations which can cross-slip very easily but are simul-
taneously sessile due to the non-planar structure of the
dislocation cores.37,38 It was thought that screw dislo-
cations behave in a similar manner in the L12 alloys, lead-
ing to work analysing the nature of dislocation cores in
model L12-ordered alloys by computer simulation
techniques.
In the L12 structure, two {111} planes and one {010}
plane share a common line along any <110> direction,
providing an opportunity for screw dislocations to spread
NO 8

into several non-parallel crystallographic planes. Two dis-
sociation configurations are possible: a glissile one on
{111} planes and a sessile one on {010} planes. When
the dislocation core is planar and spread in the plane of
the APB, as shown in Fig. 7, the dislocation is glissile.
The core can also be non-planar, extending in both the
(111) and (1−11) planes; the dislocation possessing this
core configuration is sessile. Two further symmetry-
related configurations are also possible, where the APB
is on the (010) plane with the core spread onto (111)
and (1−11) planes, respectively. Both of these configur-
ations are sessile.
Before cross-slip, the APB lies entirely on the (111)
plane and the core is coplanar with the APB. After
cross-slip, the APB lies on the (010) plane, the core
spreads on the (1−11) or (111) plane. In both cases,
after cross-slip onto (010), the dislocation core is not con-
fined to the plane determined by the APB resulting in a
sessile dislocation with respect to glide on the (010).15 It
was proposed that [−101] superdislocations dissociate
into two ½[−101] superpartials separated by an APB
(the case for Ni3Al, Ni3Ga and Ni3Ge), or into ⅓[−211]
and ⅓[−1−12] superpartials separated by super intrinsic
stacking fault, the case for Pt3Al and Pt3Ga, which
show a dramatic flow stress increase at low temperatures
(see below). The anomalous increase in flow stress is
then explained by an increasing quantity of core trans-
formations from the glissile to sessile forms.
Phase stability
The phase stability of the L12 crystal structure with
respect to isomorphous crystal structures is also thought
to affect anomalous yield behaviour. This in particular
has gained support following analysis of the yield stress
upon systematic stoichiometric deviations where it is
hypothesised that increasing levels of non-stoichiometry
affect the dislocation cores.
The phase stability theory was first proposed to account
for why the yield stress anomaly was not a feature of all
L12-ordered alloys.39,40 It was shown that a lower L12
phase stability with respect to the isomorphous ordered
f.c.c. D022 phase (Fig. 8) reduced the APB energy on
cube planes to almost zero, facilitating cross-slip from
{111}. In general, substitution of ternary elements away
from the stoichiometric composition reduces the compo-
sition region in which the L12 structure is stable by
8 The crystal structure of L12 (left), and D022 (right)
Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
reducing the average electron atom ratio or increasing
the atomic radius ratio.5,39–42 The effects of compo-
sitional deviations from stoichiometry32,43 and ternary
additions44–46 also fit the predicted behaviour of this
concept.
Pt3Al and Pt3Ga have an L12 crystal structure yet pos-
sess drastically differing flow behaviour.47–50 These alloys
exhibit an increase in flow stress with decreasing tempera-
ture (see Fig. 9); a phenomenon termed the low tempera-
ture anomaly (LTA). In the L12 structure, Pt atoms form
an octahedron within the unit cell as shown in Fig. 10.
The D0c and D0c′ structures are obtained by distorting
or rotating this Pt octahedra, respectively.49 Both trans-
formations are accompanied by a tetragonal distortion
of the lattice. The D0c′ structure was found to be more
energetically favourable than the L12 structure in the
temperature region where the LTA is observed, while
the D0c structure was unstable across the whole tempera-
ture range.
This instability has electronic origins. A transition from
L12 to D0c′ alters the charge distribution in the D0c′ phase
to become more covalent. Additional electron density
bridges form between Pt and Al atoms, enhancing chemi-
cal bonding, stabilising the D0c′ structure. The stability
ratio between the D0c′ and L12 structures is also sensitive
to stoichiometry. Iterative additions of Al atoms to Pt3
−xAl1+x cause the Pt octahedra to rotate back to the origin
position of the L12 structure. The L12 structure was found
to become stable at compositions in excess of 6 at.-% Al.48
Despite the consistency of argument to justify anoma-
lous yield through phase stability, further calculations
based on phase stability for Ni3(Al, X) single crystals
and other L12-based alloys are yet to be fulfilled.
9 Temperature dependence of CRSS for three different
orientations under compression for single crystals of
Pt3Al. All three orientations show evidence of the LTA47
Materials Science and Technology 2017 VOL 33 NO 8 939
Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
10 From top left, clockwise: 3D crystal structure of L12 Pt3Al, {001} projection of L12, D0c, D0c′ crystal structures49
Extending towards two-phase
superalloys
Despite the abundant source of information regarding
how specific elemental additions to Ni3Al affect
flow stress behaviour, the behaviour of Ni3(Al, X) and
Ni3−xAlx compounds is considerably different from that
of a two-phase Nickel superalloy. First, the yield stress
increase with increasing temperature is not as steep, and
occurs over a much narrower temperature range in a
two-phase superalloy. This is shown in Fig. 3, a
11 The deformation structure of CMSX-4 subject to creep
conditions shows stacking faults accompanying dislo-
cations in γ and dislocation pairs in γ′ precipitates51
940 Materials Science and Technology 2017 VOL 33
comparison between the constituent phases of a superal-
loy, γ and γ′ , and a commercial superalloy Mar-M200.
Furthermore, upon reaching the peak temperature, the
sharp drop-off in yield stress is thought to occur via differ-
ent mechanisms; in Ni3Al-base alloys, there is a transfer
from octahedral slip to cube slip as more cubic slip sys-
tems are activated, while in a two-phase superalloy the
decrease is attributed to a decrease in γ′ volume fraction.
The dislocation structures observed through TEM are
also quite different. Instead of the long lengths of screw
dislocations, divided by superkinks, in a two-phase super-
alloy, the dislocations grow out from built-in dendritic
cores in the γ-phase and are constrained within the narrow
γ-matrix channels. Upon application of sufficient stress,
the dislocations enter in the γ′ precipitates, where they
split into partial dislocations. Dislocations may therefore
possess a different Burgers vector since they originate as
non-APB-bound dislocations. The configurations
adopted by partial dislocations within the γ′ precipitates
vary in shape, and are sometimes accompanied by stack-
ing faults, as shown in Fig. 11.
To further develop the performance and usefulness of
superalloys, our understanding of anomalous yielding
needs to extend to Nickel-based superalloys. Upon shift-
ing to a two-phase γ/γ′ system, the additional effects on
the strength must be taken into account: the volume frac-
tion and size of γ′ , the misfit between γ and γ′ and the APB
energy of γ′ . Clear relations describing those effects, theor-
etical or empirical, are still yet to be fully understood and
significant data supporting them are still lacking.
NO 8

Shah and Duhl52 attempted to relate the precipitate size
to yield strength. The yield strength has been hypothesised
to be controlled by the distance between cube cross-slip
events and γ′ precipitate size in the superalloy PWA-
1480 by the following relation:
1 1
s= + analysis as TMS-75 is found to have a smaller misfit
R l
where sigma is the yield strength, R is the precipitate size
and λ is the distance between cube cross-slip events within
the precipitates upon shearing. In the temperature region
room temperature to peak yield temperature, the γ′ size
was thought to control strength since the distance
between cross-slip events is the smaller of the two par-
ameters. Around the peak temperature, the distance
between cross-slip events becomes very small, in turn
becoming the dominant factor controlling strength. It is
yet to be determined what controls a specific segment of
a dislocation line cross-slipping. This theory breaks
down when stacking faults become the dominant shearing
mechanism.
Effect of lattice misfit
In addition to the deformation of both the γ and γ′ phases,
the γ/γ′ interface is thought to play a crucial role in the
deformation mechanism of two-phase superalloys. Poly-
crystalline Ni-based superalloys usually have low lattice
misfit values and a γ′ volume fraction of around 50%.53
This is in stark contrast to the grain-boundary-free micro-
structures of single-crystal superalloys with up to 70% γ′
volume fraction and, relatively, much higher lattice misfit.
In both cases, the misfit between matrix and precipitates
creates significant internal stresses which are thought to
be a contributing factor to the mechanical high-tempera-
ture behaviour, though the resultant, specific effect on dis-
location behaviour is contentious.
A coherent interface exists between the γ and γ′ phases
within a Ni-based, single-crystal superalloy. Nevertheless,
the small difference in their respective lattice parameters
results in a lattice misfit, denoted δ, and defined as
2(ag′ − ag )
d=
a g ′ + ag
where aγ and aγ′ are the lattice parameters of the γ and γ′
phases, respectively. For all commercial Ni-based superal-
loys, the misfit value is negative at engine-operating
temperature.54,55
The effect of lattice misfit on the dislocation motion in
single-crystal superalloys is currently unclear because it is
difficult to isolate and test the lattice parameters against
the strength of superalloys in situ; lattice misfit can be
changed by altering the alloy composition that concomi-
tantly modifies a number of other properties. Neverthe-
less, two lines of argument exist: the first suggests
keeping misfit as small as possible to minimise the coher-
ency stresses, producing a more stable microstructure.56
Alternatively, Harada and co-workers57 propose a larger
misfit favours the formation of denser dislocation net-
works at the γ/γ′ interfaces, protecting the shearing of γ′
precipitates, in turn stabilising the microstructure.
Experiments attempting to elucidate the effect lattice
misfit has on creep strength have shown that a combi-
nation of the sign of the lattice misfit and the direction
of applied stress are both factors in high-temperature
Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
creep. For alloys with a negative lattice misfit, the precipi-
tates inhibit deformation in adjacent vertical γ channels,
while horizontal γ channels are deformed quite easily.
Zhang et al. 56 attempted to clarify the effect of lattice mis-
fit on dislocation motion during creep in TMS-75 and
TMS-138A. The misfit values were determined by X-ray
than TMS-138: −0.16 to −0.33%, respectively. TEM
observations during primary creep showed dislocation
loops gliding between narrow matrix channels during pri-
mary creep.
Effects of elemental partitioning
The effect of Re and Ru on the creep strength of Ni-based
superalloys has been extensively studied, due to the ben-
eficial additions each element has provided to superal-
loys.58 Re partitions to the γ-matrix, hardening the γ-
phase by solid solution strengthening.55,59 The effect of
Ru however is still contentious.60 It has been reported
that additions of Ru homogenise the distribution of Re
between the γ and γ′ , known as reverse partitioning, in
turn slightly reducing the strength of the γ-phase.61 How-
ever, there has been significant disagreement on this
effect.62,63
Additions of Re have been shown to make the lattice
misfit more negative, while additions of Ru were found
to have little effect. Re-containing alloys have been
thought to improve creep strength. High lattice misfit,
associated with the presence of Rhenium, leads to the for-
mation of fine, elongated rafted microstructures and
dense interfacial dislocation networks during creep.
These aid creep strength at elevated temperatures.
Whether these produce a similar effect on the yield stress
remains to be investigated.
Conclusion
It is known that the Ni3Al, L12-phase imparts strength to
a Nickel superalloy and allows anomalous yield
behaviour to be exhibited. Therefore, the study of yield
behaviour has up to now focused predominantly on
L12-ordered alloys and ternary Ni3(Al, X) compounds.
However, the mechanism controlling anomalous yielding
is still to be elucidated. The most commonly adopted
mechanism involves the cross-slip of dislocations from
{111} planes to {100} planes, where they become locked.
The locking and unlocking processes and detailed mech-
anisms that govern such processes are also still unclear.
Various models have been proposed, with increasing com-
patibility with the various macroscopic mechanical
characteristics observed in L12-ordered alloys. Both still
lack the versatility to explain all characteristics thought
to be principal features of the anomalous yield behaviour.
Though present, the anomalous yield effect is not as
pronounced in two-phase γ/γ′ systems. Factors such as
volume fraction and size of γ′ , the misfit between γ and
γ′ and the APB energy of γ′ additionally need to be
accounted for.
Acknowledgements
The author gratefully appreciates the support of his super-
visor Professor Cathie Rae of the University of Cam-
bridge, and Dr Neil Jones of Rolls Royce plc. along
Materials Science and Technology 2017 VOL 33 NO 8 941
Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
with the EPSRC and Rolls Royce plc. for financial
support.
ORCiD
Y. M. Wang-Koh http://orcid.org/0000-0003-4094-6434
References
1. Mitsubishi heavy industries, ltd. ‘Archiveorg’, 2016, available at
https://web.archive.org/web/20131113162749/http://www.mhi.co.jp/
en/news/story/1105261435.html, (accessed 19 June 2016).
2. Z. X. Shi, S. Z. Liu and J. R. Li: ‘Rejuvenation heat treatment of the
second-generation single-crystal superalloy DD6’, Acta Metall. Sin.
(Engl. Lett.), 2015, 28, (10), 1278–1285.
3. J. Douin, P. Veyssière and P. Beauchamp: ‘Dislocation line stability in
Ni3AI’, Philos. Mag. A, 1986, 54, (3), 375–393.
4. T. M. Pollock and S. Tin: ‘Nickel-based superalloys for advanced
turbine engines: chemistry, microstructure and properties’, J. Propul.
Power, 2006, 22, (2), 361–374.
5. T. Suzuki, Y. Oya and D. M. Wee: ‘Transition from positive to nega-
tive temperature dependence of the strength in Ni3Ge-Fe3Ge solid
solution’, Acta Metall., 1980, 28, (3), 301–310.
6. T. Y. Takasugi and M. Yoshida: ‘Strength anomaly and dislocation
structure at 4.2 K in Ni3(Si, Ti) single crystals’, Philos. Mag. A,
1992, 65, (3), 613–624.
7. D. Golberg, M. Demura and T. Hirano: ‘Effect of Al-rich off-
stoichiometry on the yield stress of binary Ni3Al single crystals’,
Acta Mater., 1998, 46, (8), 2695–2703.
8. D. Golberg, M. Demura and T. Hirano: ‘Single crystal growth and
characterization of binary stoichiometric and Al-rich Ni3Al’,
J. Cryst. Growth, 1998, 186, (4), 624–628.
9. T. Hirano, M. Demura and D. Golberg: ‘Compliance to Schmid’s
law in the stress anomaly regime of binary stoichiometric Ni3Al’,
Acta Mater., 1999, 47, (12), 3441–3446.
10. B. H. W. Kear and H. G. Wilsdorf: ‘Dislocation configurations in
plastically deformed polycrystalline Cu3Au alloys’, Trans. TMS-
AIME, 1962, 224, 382–386.
11. K. Suzuki, M. Ichihara and S. Takeuchi: ‘Dissociated structure of
superlattice dislocations in Ni3Ga with the L12 structure’, Acta
Metall., 1979, 27, (2), 193–200.
12. P. H. Thornton and R. G. Davies: ‘The temperature dependence of
the flow stress of gamma prime phases having the Ll2 structure’,
Metall. Trans., 1970, 1, (2), 549–550.
13. S. Takeuchi and E. Kuramoto: ‘Temperature and orientation depen-
dence of the yield stress in Ni3Ga single crystals’, Acta Metall.,
1973, 21, (4), 415–425.
14. M. J. Mills and D. C. Chrzan: ‘Dynamical simulation of dislocation
motion in Ll2 alloys’, Acta Metall. Mater., 1992, 40, (11), 3051–
3064.
15. V. Paidar, D. P. Pope and V. Vitek: ‘A theory of the anomalous yield
behaviour in L12 ordered alloys’, Acta Metall., 1983, 32, (3),
435–448.
16. B. Escaig: ‘Sur le glissement dévié des dislocations dans la structure
cubique à faces’, J. Phys., 1968, 29, (2–3), 225–239.
17. J. Bonneville and B. Escaig: ‘Cross-slipping process and the stress
orientation dependence in pure copper’, Acta Metall., 1979, 21,
(9), 1477–1486.
18. C. L. Fu, Y. Y. Ye and M. H. Yoo: ‘Theoretical investigation of the
elastic constants and shear fault energies of Ni3Si’, Philos. Mag.
Lett., 2006, 67, (3), 179–185.
19. Y. Q. Sun and P. M. Hazzledine: ‘A TEM weak-beam study of dis-
locations in γ′ in a deformed Ni-based superalloy’, Philos. Mag.
A, 1988, 58, (4), 603–617.
20. C. T. Chou, P. B. Hirsch, M. Mclean and E. Hondros: ‘Anti-phase
domain boundary tubes in Ni3Al’, Nature, 1982, 300, 621–623.
21. P. B. Hirsch: ‘A new theory of the anomalous yield stress in L12
alloys’, Philos. Mag. A, 1992, 25, (7), 1725–1730.
22. F. Louchet: ‘Dislocation exhaustion and stress anomaly in
L12 alloys: the ‘E.L.U.’ model’, J. Phys. III, 1995, 5, (11), 1803–
1807.
23. R. W. W. Guard and J. H. Westbrook: ‘Alloying behaviour of Ni3Al
(γ′ phase)’, Trans. AIME, 1959, 215, 807–814.
24. B. R. McDonnell, R. T. Pascoe, G. F. Hancock and C. W. A. Newey:
‘The growth of single crystals of the intermediate phases NiAl and
Ni3Al’, J. Mater. Sci., 1967, 2, (4), 365–370.
942 Materials Science and Technology 2017 VOL 33 NO 8
25. S. Takeuchi and E. Kuramoto: ‘Anomalous temperature dependence
of the yield stress in Ni3Ga’, J. Phys. Soc. Jpn., 1971, 31, (4), 1282–
1282.
26. T. Saburi, T. Hamana, S. Nenno and H.-R. Pak: ‘Temperature and
orientation dependence of the yield strength of Ni3(Al, W)’,
Jpn. J. Appl. Phys., 1977, 16, (2), 267–272.
27. J. Fang, E. M. Schulson and I. Baker: ‘The dislocation structure
in L12 ordered alloy Ni3Ge’, Philos. Mag. A, 2006, 70, (6),
1013–1025.
28. C. Lall, S. Chin and D. P. Pope: ‘The orientation and temperature
dependence of the yield stress of Ni3(Al, Nb) single crystals’,
Metall. Trans. A, 1979, 10, (9), 1323–1332.
29. R. D. Rawlings and A. E. Staton-Bevan: ‘The alloying behaviour
and mechanical properties of polycrystalline Ni3Al (γ′ phase) with
ternary additions’, J. Mater. Sci., 1975, 10, (3), 505–514.
30. D. P. Pope: ‘The flow stress of Cu3Au’, Philos. Mag. 1972, 25, 917–
925.
31. G. W. Ardley and A. H. Cottrell: ‘Yield points in brass crystals’,
Proc. R. Soc. London Ser. A – Math. Phys. Sci., 1953, 219, (1138),
328–341.
32. O. Noguchi, Y. Oya and T. Suzuki: ‘The effect of nonstoichiometry
on the positive temperature dependence of strength of Ni3AI and
Ni3Ga’, Metall. Trans. A, 1981, 12, (9), 1647–1653.
33. P. Veyssière and D. G. Morris: ‘Comment on ‘Transmission
electron microscopy of dislocation structures in iron-doped
Al3Ti with the L12 structure’’, Philos. Mag. A 1993, 67, (2), 491–495.
34. N. Baluc, R. Schäublin and K. J. Hemker: ‘Methods for determining
precise values of antiphase boundary energies in Ni3Al’, Philos.
Mag. A, 1991, 64, (5), 327–334.
35. M. Yamaguchi, V. Paidar, D. P. Pope and V. Vitek: ‘Dissociation and
core structure of <110> screw dislocations in L12 ordered alloys I.
Core structure in an unstressed crystal’, Philos. Mag. A, 1982, 45,
(5), 867–882.
36. V. Paidar, M. Yamaguchi, D. P. Pope and V. Vitek: ‘Dissociation and
core structure of <110> screw dislocations in L12 ordered alloys II.
Effects of an applied shear stress’, Philos. Mag. A, 1982, 45, (5),
883–894.
37. J. W. Christian: ‘The interaction between dislocations and point
defects’, Proc. 2nd Int. Conf. on ‘Strength of metals and alloys’,
Metals Park, OH, 1970, American Society of Metals, 31.
38. V. Vitek: ‘Theory of core structures of dislocations in body-centred-
cubic metals’, Cryst. Lattice Defects, 1974, 5, (1), 1–34.
39. D. M. Wee, O. Noguchi, Y. Oya and T. Suzuki: ‘New L12 ordered
alloys having the positive temperature dependence of strength’,
Trans. Jpn. Inst. Met., 1980, 21, (4), 237–247
40. D. M. Wee and T. Suzuki: ‘The temperature dependence of hardness
of Ll2 ordered alloys’, Trans. Jpn. Inst. Met., 1979, 20, (11),
634–646.
41. T. Suzuki and Y. Oya: ‘The temperature dependence of the strength
of pseudo-binary platinum-based L12 alloys with B-subgroup
elements’, J. Mater. Sci., 1981, 16, (10), 2737–2744.
42. A. Fujita, Y. Mishima and T. Suzuki: ‘Temperature dependence of
strength in a Cu3Au-5% Ni alloy and its relevance to the APB mor-
phology’, J. Mater. Sci., 1983, 18, (6), 1881–1886.
43. T. Suzuki, Y. Oya and S. Ochiai: ‘The mechanical behavior of non-
stoichiometric compounds Ni3Si, Ni3Ge, and Fe3Ga’, Metall.
Trans. A, 1984, 15, (1), 173–181.
44. Y. Kuriki, S. Ochiai, M. Yodogawa and T. Suzuki: ‘The effect of zinc
addition on the mechanical properties of Ni3Al’, Trans. Jpn. Inst.
Met., 1985, 26, (3), 213–214.
45. Y. Mishima, S. Ochiai, M. Yodogawa and T. Suzuki: ‘Mechanical
properties of Ni3Al with ternary addition of transition metal
elements’, Trans. Jpn. Inst. Met., 1986, 27, (1), 41–50.
46. S. Ochiai, Y. Mishima, M. Yodogawa and T. Suzuki: ‘Mechanical
properties of Ni3AI with ternary addition of B-subgroup elements’,
Trans. Jpn. Inst. Met., 1986, 27, (1), 32–40.
47. D. M. Wee, D. P. Pope and V. Vitek: ‘Plastic flow of Pt3Al single crys-
tals’, Acta Metall., 1984, 32, (6), 829–836.
48. N. L. Okamoto, Y. Hasegawa and H. Inui: ‘Plastic deformation of
single crystals of Pt3Al with the L12 structure having a far Al-rich
off-stoichiometric composition of Pt-29at.%Al’, Philos. Mag.,
2014, 94, (12), 1327–1344.
49. Y. N. Gornostyrev, O. Y. Kontsevoi, A. F. Maksyutov, A. J.
Freeman, M. I. Katsnelson, A. V. Trefilov and A. I. Lichtenshtein:
‘Negative yield stress temperature anomaly and structural instability
of Pt3Al’, Phys. Rev. B, 2004, 70, 014102.
50. Y. Oya-Seimita, T. Shinoda and T. Suzuki: ‘Low temperature
strength anomaly of L12 type intermetallic compounds Co3Ti and
Pt3Al’, Mater. Trans., JIM, 1996, 37, (9), 1464–1470.

51. V. Sass, U. Glatzel and M. Feller-Kniepmeier: ‘Anisotropic creep
properties of the nickel-base superalloy CMSX-4’, Acta Mater.,
1996, 44, (5), 1967–1977. 58.
52. D. M. Shah and D. N. Duhl: ‘The effect of orientation, temperature
and gamma prime size on the yield strength of a single crystal nickel
base superalloy’, Superalloys, 1984. doi:10.7449/1984/Superalloys_ 59.
1984_105_114.
53. R. C. Reed: ‘The superalloys: fundamentals and applications’, 2006,
New York, Cambridge University Press. 60.
54. H. Mughrabi: ‘The importance of sign and magnitude of γ/γ′ lattice
misfit in superalloys—with special reference to the new γ′ -hardened
cobalt-base superalloys’, Acta Mater., 2014, 81, 21–29.
55. T. M. F. Pollock and R. D. Field: ‘Chapter 63: dislocations and high- 61.
temperature plastic deformation of superalloy single crystals’, 2002,
Amsterdam, Elsevier. 62.
56. J. X. Zhang, T. Murakumo, Y. Koizumi and H. Harada: ‘The influ-
ence of interfacial dislocation arrangements in a fourth generation
single crystal TMS-138 superalloy on creep properties’, J. Mater. 63.
Sci., 2003, 38, (24), 4883–4888.
57. J. Zhang, J. Wang, H. Harada and Y. Koizumi: ‘The effect of
lattice misfit on the dislocation motion in superalloys during high-
Wang-Koh Understanding the yield behaviour of L1 2 -ordered alloys
temperature low-stress creep’, Acta Mater., 2005, 53, (17), 4623–
4633.
A. Heckl, R. Rettig, R. F. Singer: ‘Modeling of precipitation kinetics
of TCP-phases in single crystal nickel-base superalloys’, Adv. Mater.
Res., 2010, 278, 339–344.
X.-X. Yu and C. Y. Wang: ‘The effect of alloying elements on the
dislocation climbing velocity in Ni: a first-principles study’, Acta
Mater., 2009, 57, (19), 5914–5920.
S. Neumeier, F. Pyczak and M. Göken: ‘The influence of ruthenium
and rhenium on the local properties of the γ - and γ′ -phase in nickel-
base superalloys and their consequences for alloy behaviour’, 2008.
doi:10.7449/2008/Superalloys_2008_109_119v.
K. W. O’Hara, W. Ross and R. Dorolia: ‘Nickel base superalloy’,
US Patent, 1995.
J. Chen, Q. Feng and Z. Sun: ‘Topologically close-packed phase pro-
motion in a Ru-containing single crystal superalloy’, Scr. Mater.,
2010, 63, (8), 795–798.
R. A. Hobbs, L. Zhang, C. M. F. Rae and S. Tin:
‘Topologically close-packed phase promotion in a Ru-contain-
ing single crystal superalloy’, Mater. Sci. Eng.: A, 2008, 489,
(1–2), 65–76.
Materials Science and Technology 2017 VOL 33 NO 8 943