ArticlePDF Available

Planar defect formation in the γ′ phase during high temperature creep in single crystal CoNi-base superalloys

Authors:

Abstract and Figures

The structure and formation mechanism of extended planar defects in the γ/γ′ microstructure of creep deformed CoNi-base single crystal superalloys have been studied by conventional and advanced transmission electron microscopy (TEM). Planar defects in numerous isolated as well as contiguous γ′ precipitates on {111} planes reveal a characteristic configuration whereby superlattice intrinsic stacking faults (SISF) are fully embedded within antiphase boundaries (APB). Detailed analysis revealed that a leading 1/3[ 2] superpartial dislocation first creates an SISF. The SISF is then transformed into an APB by a trailing 1/6[ 2] partial dislocation. The partial is left inside the precipitate and remains as a dislocation loop. Thus, the entire shearing process constitutes a crystallographic slip of type 1/2[ 2]. A force balance analysis indicates that the initial APB energy exceeds the SISF energy. However, energy-dispersive X-ray spectroscopy (EDXS) indicates pronounced local reordering and diffusion processes near both types of planar defects. The APB qualitatively adopts the composition of the γ phase whereas the SISF locally changes its composition towards that of the Co3W–D019 phase. We propose that these atomic diffusion processes determine the formation and shrinkage of the loops. A post mortem in situ TEM heating experiment shows that with increasing temperature the APBs exhibit complete faceting into {100} planes followed by coarsening, eventually leading to disintegration of the γ′ precipitate. This indicates a detrimental impact of APBs as potential nuclei for fragmentation of the γ/γ′ microstructure in CoNi-base superalloys.
Content may be subject to copyright.
Full length article
Planar defect formation in the
g
0
phase during high temperature creep
in single crystal CoNi-base superalloys
Y.M. Eggeler
a
, J. Müller
a
, M.S. Titus
b
, A. Suzuki
c
, T.M. Pollock
b
, E. Spiecker
a
,
*
a
Institute of Micro- and Nanostructure Research &Center for Nanoanalysis and Electron Microscopy (CENEM), Friedrich-Alexander-Universit
at Erlangen-
Nürnberg (FAU), Cauerstraße 6, 91058, Erlangen, Germany
b
Materials Department, University of California Santa Barbara, Santa Barbara, CA 93106, USA
c
GE Global Research Center, One Research Circle, Niskayuna, NY 12309, USA
article info
Article history:
Received 24 January 2016
Received in revised form
30 March 2016
Accepted 30 March 2016
Available online 8 May 2016
Keywords:
Cobalt-base superalloys
Creep
Antiphase boundary
Stacking faults
Transmission electron microscopy
abstract
The structure and formation mechanism of extended planar defects in the
g
/
g
0
microstructure of creep
deformed CoNi-base single crystal superalloys have been studied by conventional and advanced trans-
mission electron microscopy (TEM). Planar defects in numerous isolated as well as contiguous
g
0
pre-
cipitates on {111} planes reveal a characteristic conguration whereby superlattice intrinsic stacking
faults (SISF) are fully embedded within antiphase boundaries (APB). Detailed analysis revealed that a
leading 1/3[112] superpartial dislocation rst creates an SISF. The SISF is then transformed into an APB
by a trailing 1/6[1 12] partial dislocation. The partial is left inside the precipitate and remains as a
dislocation loop. Thus, the entire shearing process constitutes a crystallographic slip of type 1/2[112]. A
force balance analysis indicates that the initial APB energy exceeds the SISF energy. However, energy-
dispersive X-ray spectroscopy (EDXS) indicates pronounced local reordering and diffusion processes
near both types of planar defects. The APB qualitatively adopts the composition of the
g
phase whereas
the SISF locally changes its composition towards that of the Co
3
WeD0
19
phase. We propose that these
atomic diffusion processes determine the formation and shrinkage of the loops. A post mortem in situ
TEM heating experiment shows that with increasing temperature the APBs exhibit complete faceting into
{100} planes followed by coarsening, eventually leading to disintegration of the
g
0precipitate. This in-
dicates a detrimental impact of APBs as potential nuclei for fragmentation of the
g
/
g
0microstructure in
CoNi-base superalloys.
©2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
1. Introduction
The discovery of a stable
g
/
g
0
two phase eld in the ternary
Co-Al-W system at 900
C opened new pathways for the devel-
opment of high temperature creep resistant superalloys [1]. These
Co alloys exhibit a similar microstructure as the extensively studied
and widely used Ni-base superalloys, where cubes of ordered
g
0
phase, possessing the L1
2
crystal structure, are coherently
embedded in a solid-solution face-centered cubic (fcc)
g
phase
matrix [2]. The lattice constant of the
g
0
precipitates exceeds the
lattice constant of the
g
matrix, resulting in a positive lattice mist
contrary to the negative mist that is usually observed in Ni-base
superalloys [3,4]. Adding Ni to the ternary system widens the
g
0
phase eld and enhances the
g
0
solvus temperature [5]. Thus,
further additions of higher order alloying elements becomes
possible with Ni additions, and the extent of alloy design space
improves. In high-temperature-deformed microstructures of alloys
based on the ternary system, a high density of stacking faults (SFs)
extending across the
g
/
g
0
microstructure has been observed, which
indicates that the stacking fault energy is lower compared to the
APB energy [6,7]. Tensile creep experiments performed at 900
Con
single crystalline Co- and CoNi-based superalloys have demon-
strated a creep resistance comparable to 1st generation Ni based
superalloys [8,9]. Recently, the deformation mechanisms in crept
CoNi-base alloys have been investigated, and the analyses revealed
shearing of the
g
0
precipitates, leaving antiphase boundaries (APB)
in the wake of the dislocations [10]. This is in contrast to Ni-base
superalloys, where a second coupled partial dislocation follows
and reestablishes the order [11].
The research presented here builds on the ndings reported in
*Corresponding author.
E-mail address: erdmann.spiecker@fau.de (E. Spiecker).
Contents lists available at ScienceDirect
Acta Materialia
journal homepage: www.elsevier.com/locate/actamat
http://dx.doi.org/10.1016/j.actamat.2016.03.077
1359-6454/©2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Acta Materialia 113 (2016) 335e349
Ref. [10] and unambiguously identies a new creep-induced
microscopic deformation mechanism in CoNi superalloys using
conventional and high resolution transmission electron microscopy
(CTEM, HRTEM), geometric phase analysis (GPA), large angle
convergent beam electron diffraction (LACBED), as well as local
chemical analysis by energy dispersive X-ray spectroscopy (EDXS)
in combination with high resolution scanning transmission elec-
tron microscopy (HRSTEM). Within this mechanism, a superlattice
intrinsic SF (SISF) is formed on a {111} plane by glide of a 1/3<112 >
leading partial dislocation across multiple contiguous
g
0
pre-
cipitates. A second 1/6<112 >trailing partial dislocation follows and
transforms the SISF into an antiphase boundary (APB). While the
leading dislocation continues to glide, the trailing dislocation forms
a closed loop inside the precipitate, which separates the SISF (inside
the loop) from the APB (outside the loop). This conguration con-
rms a microscopic 1/2<112 >shear mechanism. Enrichment of
alloying elements at APBs and SISFs as well as the migration of APBs
towards energetically favorable {100} orientations clearly indicate
that local reordering and atomic diffusion at and along the planar
faults play a key role for the observed shearing mechanism and,
more generally, the creep behavior of CoNi superalloys in the high
temperature/low stress regime. Possible energetic and kinetic
reasons in determining the observed shear mechanism and the
relationship to a similar mechanism in Ni-based superalloys will be
addressed.
2. Experimental procedures
A detailed description of the single crystal casting procedure,
the tensile creep experiments and the composition of the CoNi
superalloy CoNi-A(Co 45.9 at%, Ni 29.2 at%, Al 9.8 at%, W 6.3 at%,
Ta 2.4 at%, Cr 6.4 at%) studied in this work was reported by Titus
et al. [8]. The as aged state of the CoNi-A alloy exhibits a cuboidal
precipitate morphology with a
g
0
area fraction of 70e80%, a mean
precipitate size of ~ 410 nm along the cube edge, and a mean
channel width of 46 nm [10]. In short, the single crystal sample was
creep deformed along the [001] crystallographic direction at 900
C
under an initial tensile stress of 310 MPa up to 0.5% strain. Electron
backscatter diffraction (EBSD) revealed that the tensile axis was
actually 6.2
off from [001]. TEM samples were prepared from the
deformed gage zone of the creep specimen. Using a wire saw, the
specimens were cut longitudinal and perpendicular with respect to
the tensile axis. The discs with a diameter of about 3 mm were then
mechanically ground and polished to a thickness of 120
m
m before
nal thinning with a Fischione twin-jet electropolisher in a solution
of 80% ethanol, 5% perchloric acid and 15% water (volume %). The
electrolyte was cooled with liquid nitrogen to a temperature
of 5
Cto20
C, and polishing was performed at 50 V and
170 mA. Some samples were additionally polished using a Gatan
precision ion polisher system (PIPS) at 1e2 keV with beam angles
down to 4
in order to remove surface oxides and contamination.
Comprehensive defect characterization by CTEM and LACBED
was carried out with a Phillips CM30 T/STEM operating at an ac-
celeration voltage of 300 kV. Annular Dark Field Scanning TEM
(ADF-STEM) and HRTEM were performed on an image side
aberration-corrected FEI Titan
3
80e300 operating at an accelera-
tion voltage of 300 kV and 200 kV, respectively. As recently re-
ported by Phillips et al. [12], in order to obtain diffraction contrast
from crystal defects, ADF-STEM was performed on low indexed
Kikuchi bands. In our case, a medium camera length of 360 mm was
used. The intrinsic nature of SFs has been conrmed by applying
geometric phase analysis (GPA) to HRTEM images [13]. High reso-
lution STEM imaging and EDXS mapping at 200 kV were carried out
using a double-aberration-corrected Titan Themis
3
300 equipped
with a Super-X detector incorporating the Bruker ChemiSTEM
system. The Super-X detector comprises four silicon drift detectors
(SDD) symmetrically placed around the optical axis, close to the
sample area. All four signals are combined into one spectrum to
improve the collection efciency. More details explaining this
system can be found in Refs. [14e16]. The STEM-EDXS maps of the
APB and the SISF were collected within 15 min using drift
correction.
The behavior and stability of planar defects at high temperatures
was studied in situ using a conventional double-tilt heating holder
(GATAN, Model 652) in the Phillips CM30 transmission electron
microscope operated at 300 kV. In situ imaging of defects, in
particular APBs, was carried out in centered dark eld (CDF) mode
with a superlattice reection. The in situ TEM heating experiment
was performed by heating from room temperature (RT) to 1050
C
in 50
C steps, with 15 min holds at each temperature in order to
subsequently record an image.
The deformation defects observed in TEM samples cut perpen-
dicular to the [001] tensile axis were shown and discussed previ-
ously [10]. For a more detailed analysis of the defects, additional
TEM samples oriented parallel (foil normal [010] and foil normal
[110]) to the tensile axis were prepared and studied with advanced
TEM techniques in the present work.
3. Results
3.1. Conventional transmission electron microscopy analysis
In all projections, planar defects were frequently observed to
extend on the same plane over several contiguous precipitates,
with an absence of any defect contrast in the
g
matrix channels
between the precipitates. TEM images of a typical collection of
planar defects are shown in Fig. 1. To summarize the main char-
acteristics, the extended defects formed over multiple and adjacent
g
0
precipitates, whereby in each precipitate an APB/SISF/APB
conguration was observed inwhich a partial dislocation separated
the two types of planar defects. Near the
g
/
g
0
interface, the APBs
tend to deviate from the (111) glide plane towards energetically
favorable orientations. Interestingly, no planar defects were
observed in the
g
channel under various beam directions. Thus the
whole extended defect has been formed by glide of a perfect matrix
dislocation or an array of dislocations summing up to a perfect
matrix dislocation.
TEM images were taken under different imaging conditions in
order to conrm the APB/SISF/APB sequence of planar defects. Fig. 1
(a) shows a Bright Field (BF) image taken near the [010] zone axis
under a g(200) two beam condition. The image clearly reveals SF
contrast (red dotted arrows) on the inclined (111) plane inside the
g
0
precipitates whereas no SF contrast is observed in the
g
channels.
Moreover, the SFs do not cover the full area inside the
g
0
pre-
cipitates, but, in each individual precipitate, they are terminated by
APB segments on both sides near the
g
/
g
0
interface, as revealed by
residual bright contrast (blue solid arrows) in Fig. 1 (a). Centered
dark eld (CDF) images taken with superlattice reections g(100)
and g(101) revealed clear APB contrast, as shown by the blue ar-
rows in Fig. 1 (b) and (c). In the CDF image taken with g(100) the
APB and SF segments are visible, as expected from the simulta-
neous excitation of g(200). No SF contrast is observed in the CDF
image taken with g(101), which is consistent with the fact that the
gvector is contained in the (111) glide plane and is thus perpen-
dicular to the SF translation vector, i.e. g$R¼0. However, a (partial)
dislocation located at the boundaries (or contact lines) between
APB segments and SFs become visible. Finally, from the CDF images
it can be seen that the APB segments are not fully conned to the
(111) plane, but they tend to deviate from the glide plane near the
g
/
g
0
interface, consistent with the observations in Ref. [10].For
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349336
instance, the two APB segments in the rst precipitate (from left) in
Fig. 1 (b) are clearly seen to bow upwards when approaching the
g
/
g
0
interface. This APB migration effectively reduces the total APB
area, and, moreover, the APB approaches the energetically favor-
able {100} orientation [17e20].
As the SFs are bounded by APBs on both sides in the
g
0
precip-
itate as shown in Fig. 1, it may be envisaged that the SFs are
enclosed by dislocation loops which are surrounded by APBs. In
order to prove this, the planar defect congurations were further
studied in a TEM foil whose normal was approximately parallel to
the [110] direction. In this orientation, the (111) habit plane of the
planar defects is at a considerably smaller inclination angle (~35
)
compared to that of Fig. 1 and thus larger areas of the slip plane can
be inspected.
Fig. 2 shows three ADF-STEM images of rather thick sample
regions revealing the characteristic defect features observed
throughout the specimen. Pronounced defect contrast was ob-
tained by orienting the sample on a {002} Kikuchi band (close to
the [110] zone axis) and choosing a medium camera length of
360 mm. A high number of cropped stacking fault loops, as repre-
sented by Fig. 2 (a), are seen all over the sample. Quite frequently
full loops are also observed, as shown in Fig. 2 (b) and (c). (A proof
that the planar defects correspond to stacking faults will be given in
Fig. 3.) Outside of each stacking fault loop, a faint supplementary
contrast is observed, which is due to the presence of an APB in
agreement with Fig. 1.
Fig. 3 (a) shows a TEM BF image of a thicker sample area (cor-
responding to the area in Fig. 2 (b)) imaged under a two beam
g(002) condition. In the central part, a stacking fault loop
completely embedded in an APB (revealed by its residual contrast)
can be unambiguously identied. Furthermore, the
g
0
precipitates
on the right and left exhibit the same APB/SF/APB conguration
conrming the extended dimension of the planar defects across the
g
/
g
0
microstructure as already observed in Fig. 1. The stacking fault
loops to the left and right in the image are, however, cropped due to
the limited thickness of the TEM foil. It should be noted that, in
contrast to Fig. 1, the individual
g
0
precipitates and
g
channels are
more difcult to discern in Fig. 3 (a) (and also in Fig. 2) because the
vertical
g
/
g
0
interfaces are inclined to the electron beam, which is
oriented close to [110] foil normal. Fig. 3 (b) shows a BF image
taken near the [11 1] zone axis under g(202) two beam condition.
While the dislocation loop is clearly revealed, indicating g.b
loop
s0,
the stacking fault remains invisible as expected for a diffraction
vector contained in the fault plane (g·R
SF
¼0). Invisibility of both
the stacking fault and the partial dislocation loop is obtained under
the g(220) two beam condition, as shown in Fig. 3 (c). In this case,
g·b
loop
and g·R
SF
are both equal to zero. Finally, the BF image in
Fig. 3 (d), taken with the g(111) reection, shows pronounced
stacking fault contrast and only residual dislocation contrast. The
stacking fault habit plane was determined by tilting the sample
towards the [111] and [1 11] zone axes. The projected loop area was
largest near the [11 1] zone axis, which indicates a SF habit plane of
(11 1) or, equivalently, (111).
Subsequent analysis of an APB of the same defect conguration
in thinner sample regions enabled an unambiguous determination
of the APB displacement vector, R
APB
¼1/3[112], as shown in Fig. 4.
For the analysis, it should be recalled that the displacement vector
must be contained in the (11 1) slip plane because the APB origi-
nated from shear along this plane. In the CDF image Fig. 4 (a) taken
with superlattice reection g(101) the APB shows pronounced
contrast indicating g·R
APB
sn (n: integer). In contrast, the stacking
fault is invisible as expected from the fact that g(101) is contained
in the (11 1) slip plane. The partial dislocation separating APB and
SF is again visible because b
loop
¼1/6[112] (see below) and thus
g·b
loop
s0.
Invisibility of the APB is achieved for g(001) and g(110) under
CDF conditions, Fig. 4 (b) and (c), indicating g·R
APB
¼0 or n (n:
integer) for both cases. Potential APB displacement vectors con-
tained in the slip plane are therefore 1/2[110] or 1/2[112]. It should
be noted that these two displacement vectors are equivalent from a
crystallographic point of view because the difference vector [101] is
a lattice translation of the ordered structure (see supplementary
material for details). However, because the APBs are generated by
shear, the displacement vector is likely created by a glide system
which experienced resolved shear stress during creep. The 1/2
[110](111) glide system exhibits a Schmid factor equal to zero since
the Burgers vector is perpendicular to the [001] tensile axis. In
contrast, for the 1/2[112](1 11) glide system both the Burgers
vector and the glide plane are inclined by 35.26
from the tensile
axis. This results in the highest possible Schmid factor (0.471) for
Fig. 1. TEM micrographs are shown for a typical planar defect conguration in the CoNi-A alloy creep deformed at 900 C and 310 MPa to a strain of 0.5%. The [010] foil normal is
perpendicular to the tensile axis, as shown in the schematic. (a) TEM BF image with g(20 0) excited, (b) CDF image taken with g(100), (c) CDF image of one precipitate taken with
g(101). A SISF and two APBs are marked by red dotted and blue solid arrows, respectively. Each
g
0precipitate along the extended planar defect exhibits the same characteristic APB/
SISF/APB arrangement. The APBs bend away from the (111) glide plane close to the
g
/
g
0interface towards {100} planes. (For interpretation of the references to colour in this gure
legend, the reader is referred to the web version of this article.)
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349 337
slip on {111}, which corresponds, in the present case, to a resolved
shear stress of about 146 MPa. This indicates that 1/2[112](1 1 1) is
the active glide system by which the APBs in Figs. 1e4have been
generated. The drawing in Fig. 3 illustrates the orientation of the
[112] slip direction (red arrow) and its (111) glide plane relative to
the [001] tensile direction. Note that the absence of APB contrast in
Fig. 1 (a) and the presence of APB contrast in Fig. 1 (b) and (c) are
also consistent with an APB displacement vector R
APB
¼1/2[112]
or, equivalently, 1/2[110] because g·R
APB
is equal to an integer for
g(200) whereas it differs from an integer for g(100) and g(101).
Stacking fault shear by <112 >dislocation ribbons is well known
for Ni-base Superalloys that are creep deformed at low tempera-
tures and high stresses (750
Ce850
C, 500 MPa) [21e23]. For
these conditions researchers have shown that dislocation ribbons
with total Burgers vector of type <112 >are nucleated from
1/2<110 >dislocations active in the matrix channels [22e25]. Upon
cutting of the
g
0
phase these ribbons are fragmented into a leading
partial dislocation, 1/3<11 2 >, creating an SISF, which is followed by
two 1/6<112 >partial dislocations enclosing an APB and trans-
forming the SISF into a super extrinsic stacking fault (SESF). A nal
trailing partial dislocation with b ¼1/3<11 2 >reestablishes the L1
2
stacking order [23].
In contrast to the observations in Ni base superalloys, the CoNi
superalloy under investigation reveals only 1/2<112 >shear that
leaves an APB in the
g
0
precipitate. No SESFs were found at this
stage of creep. However, in many
g
0
precipitates residual SISF loops
remain embedded in the APB as discussed above (see also Fig. 7 and
Fig. 8 for a direct proof of the intrinsic character of these stacking
faults). From the invisibility of the loop in the BF image Fig. 3 (c), it
can be concluded that the Burgers vector of these (Shockley) partial
dislocation loops is ±1/6[112], which is parallel to the total Burgers
vector 1/2[112] of the ribbon. We therefore anticipate that the
dislocation loops have been created by the trailing partial of the
ribbon which transforms the SISF into an APB and propose the
following shear process illustrated by the model in Fig. 5: (1) A
band of two closely spaced Shockley partials, a leading 1/3[1 12]
and trailing 1/6[112] dislocation, approach the
g
0
precipitate on the
(111) glide plane. (2) The 1/3[112] leading partial enters the
g
0
precipitate forming a SISF, whereas the trailing partial is left at the
interface. (3) Both partials continue to shear, such that the trailing
partial also enters the precipitate transforming the SISF into an APB.
(4) While the leading partial cuts the entire
g
0
precipitate the
Fig. 2. HAADF STEM micrographs taken on the {002} Kikuchi band show three
exemplarily regions with extended ribbons of faults in the
g
/
g
0microstructure in a
sample with foil normal close to [110]. The ribbons consist of cropped stacking fault
loops (a) and entire loops (b) and (c) in the precipitates embedded in APBs. This type of
defect structure is a common feature after 0.5% strain in the creep deformed CoNi-A
superalloy.
Fig. 3. BF images of an extended planar defect in a [1 10]-oriented TEM foil, showing unambiguously a faulted loop fully embedded in an APB. (a) BF image taken with g(002) excited
showing the APB band with weak residual contrast and embedded dislocation loops with pronounced SF contrast in each precipitate. (b) BF image with g(202) excited of central
dislocation loop tilted to the [1 11] zone axis exhibiting maximum projected size and no SF contrast. (c) BF image with g(220) excited showing absence of both SF and dislocation
contrast. (d) Image with g(111) excited showing pronounced stacking fault and residual dislocation contrast. (e) Schematic of slip systems looking from the [1 10] direction. Light
blue triangle symbolizes the inclined (1 11) plane of the shear band with its <110 >and <11 2>slip directions. The red arrow represent the experimentally evaluated Burgers vector
direction of the dislocation loop. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349338
trailing partial forms a closed loop which remains inside the pre-
cipitate. Simultaneously the APB migrates from the (111) glide
plane towards energetically favorable {100} orientations near the
g
/
g
0
interface (not shown in this projection, cf. Fig. 13). In the nal
conguration (5) the 1/6[112] partial dislocation loop surrounds a
SISF and is embedded in an APB, as experimentally observed in
Figs. 1e3.
In order to conrm this hypothesis the sign and magnitude of
the Burgers vector of the partial dislocation constituting the SISF
loop will be explicitly determined by large-angle convergent beam
electron diffraction (LACBED, see next section). Based on the sign of
the Burgers vector, it will be shown that the external load acts to-
ward shrinkage of the dislocation loop corresponding to SISF /
APB transformation.
3.2. Large-angle convergent beam electron diffraction analysis
To conrm the Burgers vector of the partial dislocation loop and
directly determine its absolute sign and magnitude, we apply a
more advanced Burgers vector analysis technique based on LAC-
BED. Detailed descriptions of the use of LACBED for the character-
ization of superdislocations in the
g
0
phase and of dislocation loops
are given in Ref. [26] and Refs. [27,28], respectively. In short, a
convergent electron beam with a large convergence angle is
focused at the eucentric height (Fig. 6 (a)), which results in signif-
icant overlap of diffraction discs in the back focal plane (not
shown). While maintaining the beam crossover in this plane, the
sample is lowered (or raised [27]) from the eucentric height, which
creates a point pattern of Bragg diffracted beams in the rst image
plane. Using a small selected area aperture, either the direct beam
or a diffracted beam can be selected. This enables observation of
individual diffraction discs (bright eld or dark eld LACBED disc)
in the nal diffraction pattern. Since the sample position deviates
from the eucentric height and is no longer in exact reciprocal
relationship to the back-focal plane of the objective lens (magnied
onto the screen), the LACBED disc comprises not only a pattern of
Bragg lines, like in CBED, but also a shadow image of the illumi-
nated sample area. As rst demonstrated by Cherns and Preston
[29] and described in more detail in the book by Morniroli [30], the
Burgers vector of dislocations can be unambiguously determined
by evaluating the contrast of Bragg lines intersecting the trace of
the dislocation line. Using the Cherns and Preston rule [29,30], the
value n ¼g·b, where gis the diffraction vector of the Bragg line and
bis the Burgers vector of the dislocation, can be deduced from the
twisting of the Bragg line approaching the intersection and the
number of subsidiary maxima/minima (in bright/dark eld LAC-
BED) at the intersection. In general, intersections of three linearly
independent Bragg lines with the dislocation line are required to
determine the full Burgers vector. However, since only three
Shockley partial dislocations are contained within a single (111)
plane (six including their sign), evaluation of a single intersection
turned out to be sufcient for determination of the full Burgers
vector. Since b
loop
is a partial dislocation, n ¼g·b
loop
is not neces-
sarily an integer which can complicate the analysis [31,32.] In order
to circumvent any problems we chose the Bragg line g¼(006) for
the LACBED analysis because n is limited to integer values, ±1or±2,
for all the possible aforementioned Burgers vectors.
Fig. 6 (b) summarizes the results of the LACBED analysis. At the
top of Fig. 6 (b) the central part of a bright eld LACBED disc con-
tains the stacking fault loop from Fig. 3 as a shadow image. The loop
is intersected by the Bragg deciency line, g¼(006), which clearly
exhibits splitting into two subsidiary intensity maxima (marked by
arrows) upon crossing the dislocation loop in the left and right loop
regions. In the corresponding g(006) dark eld LACBED pattern
(Fig. 6 (b) center), the Bragg line appears bright on dark back-
ground. The contrast of the splitting into two subsidiary intensity
minima (marked with two arrows at each intersection) is more
discernable compared to the bright eld disc and conrms the
number of splittings obtained in bright eld LACBED. A schematic
of the Bragg line splitting is depicted in the bottom part of Fig. 6 (b)
and includes the sign of the excitation error, which changes from
negative to positive upon crossing the Bragg line. Applying the
evaluation rules by Cherns and Preston [29,30] the observed
splitting yields n ¼g.b
loop
¼2. Because the habit plane of the
SF loop was identied to be (111), only 6 possible Burgers vectors
of type 1/6<211>must be considered. These include ±1/6[112],
±1/6 [121] and ±1/6[211], as already discussed in connection
with Fig. 3 (a). With the information from the LACBED analysis,
the Burgers vector of the dislocation loop is unambiguously
identied as b
loop
¼þ1/6[112], in agreement with the contrast
behavior observed from the conventional diffraction analysis (Fig. 3
(a)-(d)).
An important advantage of LACBED over conventional Burgers
vector analysis is that the absolute sign (and the magnitude) of the
Fig. 4. Dark-eld TEM analysis of the same defect band studied in Fig. 3, however in
thinner sample regions, enabling characterization of the APB. (a) The CDF image taken
with g(101) close to the [010] zone axis clearly reveals the APB (blue solid arrow)
which is connected to the right to a cropped invisible SF loop. The partial dislocation
separating the two types of planar faults is visible. (b) CDF imaging with g(001)
superlattice reection close to the [1 10] zone axis results in only residual contrast for
the APB whereas the SF is clearly visible (red dotted arrow). (c) The SF-loop and the
APB as well as the dislocation loop are all invisible when performing CDF imaging with
g(110). (For interpretation of the references to colour in this gure legend, the reader is
referred to the web version of this article.)
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349 339
Burgers vector can be determined. Therefore, one can discriminate
whether the resolved shear stress (resulting from the tensile load
along [001]) acts towards contraction or expansion of the disloca-
tion loop. In order to draw this conclusion the nish-to-start right-
hand (FS-RH) convention for the denition of the Burgers vector
must be taken into account because the Cherns and Preston rules
are based on this convention. As described in detail in the
supplementary information (SI), the resolved shear stress acts to-
wards contraction and, eventually, annihilation of the SISF loop.
This is important information which must be taken into account
considering how the SISF loops formed during creep (cf. our model
in Fig. 5 and related discussion).
Fig. 5. Model of the shear process and planar fault formation observed in this work. (1) A band of two closely spaced Shockley partials, a leading 1/3[112] and trailing 1/6[112]
dislocation, approaches the
g
0precipitate on the (111) glide plane. (2) The 1/3[1 12] leading partial enters the
g
0precipitate forming a SISF, whereas the trailing partial is left at the
interface. (3) Both partials continue to shear, such that the trailing partial also enters the precipitate transforming the SISF into an APB. (4) While the leading partial cuts the entire
g
0
precipitate the trailing partial forms a closed loop which remains inside the precipitate. Simultaneously the APB migrates from the (111) glide plane towards energetically favorable
{100} orientations near the
g
/
g
0interface (not shown in this projection, cf. Fig. 13). In the nal conguration (5) the 1/6[112] partial dislocation loop surrounds an SISF and is
embedded in an APB, as experimentally observed in Fig. 3.
Fig. 6. (a) Illustration of the optical path for Large Angle Convergent Beam Electron Diffraction (LACBED), in particular showing that the sample with dislocation loop was lowered
from the object plane. For simplicity, only the direct beam is shown. (b) BF (top) and DF (middle) LACBED micrographs showing the g(006) Bragg line passing the dislocation loop of
Fig. 3. The double splitting (arrows) of the Bragg line at the intersection with the dislocation loop is clearly revealed as illustrated in the schematic (bottom). For more details see
text.
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349340
3.3. HRTEM and geometric phase analysis (GPA)
Because the displacement vector of the APBs is R
APB
¼1/2[112]
and the Burgers vector of the remaining loop is b¼1/6[112], the
total displacement vector of the area inside the dislocation loop
must be equal to R¼1/3[112]. Therefore, the SF inside the dislo-
cation loop is expected to be a superlattice intrinsic SF (SISF)
because it formed by glide of a single leading partial dislocation of
type 1/3[112]. In order to prove the intrinsic nature, SFs in edge-on
orientations were analyzed by HRTEM in thin parts of the [1 10]-
Fig. 7. Proof of intrinsic nature of a stacking fault in edge on orientation by HRTEM (a) and geometric phase analysis (GPA) (b, c). The phase image in (c) was obtained by applying
the GPA to the {111} reection shown in the power spectrum (b). The phase shift of þ2
p
/3 upon crossing the fault from bottom left to top right is indicative of an intrinsic SF (see
Supporting Information for details).
Fig. 8. HRSTEM analysis of SISF and APB: (a) Low-mag STEM image of
g
0precipitate with characteristic APB/SISF/APB defect conguration. The approximate position of the partial
dislocation loop b
loop
is indicated. (b) and (c) High-resolution STEM images of SISF and APB at the sample areas indicated by boxes in the low-mag STEM image. The bright and dark
Z contrast of the planar faults indicates accumulation/depletion of high Z alloying elements at the SISF and APB, respectively, as conrmed by EDXS (see Fig. 9). While the large-area
HRSTEM images in the back show unltered images the enlarged images shown as insets were obtained by periodic averaging of the fault contrast along the respective fault plane.
For the SISF the ABCABjABCAB stacking is directly revealed conrming the intrinsic nature of the stacking fault. Both images clearly show the L1
2
superstructure by alternating
contrast of (002) lattice planes as exemplarily indicated for the enlarged APB image (inset) where two atomic layers with bright contrast are marked. Across the APB there is no
relative displacement of these layers (dotted lines) which is in agreement with the expected APB displacement vector (see text for details).
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349 341
oriented TEM sample already studied in Figs. 2e4.Fig. 7 (a) shows a
typical HRTEM image taken with an image side aberration
corrector. The limited surface quality of the TEM sample makes
direct evaluation of the HRTEM contrast inside the SF difcult.
Nevertheless, the SF character can be unambiguously determined
by studying the relative shift of the crystal lattices on both sides of
the SF. For this, we employed the geometric phase analysis (GPA)
introduced by Hÿtch et al. [33]. Masking a reciprocal lattice point in
the FFT power spectrum and applying the GPA to the {111} reec-
tion inclined to the SF plane normal (Fig. 7 (a) and (b)) results in a
phase image (Fig. 7 (c)), which reveals the relative shift of the
corresponding lattice planes across the SF. The shift corresponds
to þ2
p
/3 or 2
p
/3 depending on the intrinsic or extrinsic nature of
the SF, respectively (see SI for further details). An integrated line
scan of the geometric phase shift across the SISF from the bottom
left to the top right is depicted on the right of the phase image
(Fig. 7 (c)). The phase image clearly reveals a phase shift of þ2
p
/3
which proves the intrinsic nature of the SF. We note here that the
HRTEM analysis alone cannot discriminate between a SISF and a
complex SF (CSF), which is also of intrinsic nature, since the pro-
jected structure looks identical. However, the SISF nature is obvious
from the formation mechanism (Fig. 5) and is in agreement with
the fact that CSFs possess relatively high energies in L1
2
ordered
structures, compared to SISFs and APBs [2,34].
3.4. HRSTEM and EDXS analysis of planar faults
A deeper insight into the structure and local chemistry of planar
faults can be obtained by aberration-corrected HRSTEM and local
chemical analysis exploiting the advanced capabilities of EDXS with
large-area silicon drift detectors (SDD) available within the
ChemiSTEMtechnology. This technique has recently been uti-
lized to investigate segregation at SISFs, SESFs, and dislocations in
some Coe, CoNie, and Ni-based superalloys [35e37]. Here we
conrm their results and extend the analysis to an adjacent APB in a
characteristic APB/SISF/APB conguration.
Fig. 8 shows a low magnication HAADF-STEM image of a
typical APB/SISF/APB arrangement in edge-on orientation ob-
tained from the (110) TEM sample oriented parallel to the [001]
load axis (longitudinal cut). The SISF and APBs are clearly visible
as bright and dark lines, respectively, crossing the
g
0
particle. The
intersections of the partial dislocation loop with the thin foil are
indicated by circles and correspond to the transitions between
the two types of planar faults. The APB in the top part of the
image clearly deviates from the {111} slip plane towards a {100}
orientation upon approaching the
g
channel. Fig. 8 (b) and (c)
show (aberration corrected) HRSTEM images of the two faults at
atomic resolution, the images were approximately taken from
the two areas marked with boxes in Fig. 8 (a). It should be
mentioned that the large area HRSTEM images shown in the
background correspond to raw data, i.e. only contrast and
brightness has been adjusted and no ltering has been applied.
For better visibility, and in order to improve the signal-to-noise
ratio, the enlarged HRSTEM images shown as insets were ob-
tained by periodic averaging along the fault planes and repetition
of the resulting units. The intrinsic nature of the SISF, already
conrmed by HRTEM/GPA (Fig. 7), is further corroborated by
HRSTEM, as can be seen from the arrangement of atomic columns
in the inset (Fig. 8 (b)). In agreement with Titus et al. [35] the SISF
shows pronounced bright Z-contrast indicating segregation of
heavy alloying elements. In contrast, the APB shows dark Z-
contrast (Fig. 8 (c)) similar to the disordered
g
phase indicating
enrichment of light elements (or depletion of heavy elements).
Both conclusions will be conrmed by local EDXS analyses
(Fig. 9). As can be seen from the enlarged insets in Fig. 8 (b) and
(c), the L1
2
ordering of the
g
0
precipitate is clearly revealed in the
HRSTEM images (doubling of (002) period). However, there is no
relative shift of the (projected) superlattice structure across the
APB, as indicated by dashed lines in Fig. 8 (c). This is consistent
with the deformation mechanism proposed in Fig. 5.As
explained in the supporting material such slip creates APBs with
1/2<110 >displacement vector perpendicular to the tensile axis.
For the present <110>TEM sample the displacement vector
therefore points into the foil which explains why no relative shift
of the projected superlattice structure upon crossing the APB is
visible.
Fig. 9 presents results of local chemical analyses across the SISF,
the APB, and a
g
channel (see red, blue and black box in Fig. 8 (a)) by
exploiting the improved detection efciencies of EDXS with large
silicon drift detectors (SDD) as implemented in the ChemiSTEM
technology. In agreement with former measurements by Titus et al.
[35] on the CoNi-C alloy Co, Cr and W are enriched at the SISF in
comparison with the adjacent
g
0
phase whereas Ni and Al are
depleted. Only the Ta concentration seems to be rather homoge-
neous across the SISF. The elemental maps across the
g
channel,
Fig. 9 right bottom, and the corresponding concentration prole
conrms that Co and Cr partition to the
g
phase whereas the re-
fractory elements W and Ta partition to the
g
0
phase. The same
partitioning behavior is found in the vicinity of the APB (middle)
indicating that the local composition of the APB approaches that of
the disordered
g
phase, which stays in line with the similar dark Z-
contrast of APB and
g
channel in the STEM images. Segregation
along the APB is generally more pronounced with the distinction
that W tends to segregate to the SISF and away from the APB, while
Cr is highly enriched at the APB. This indicates that a greater extent
of diffusion processes have occurred near the APB compared to the
SISF.
Despite the high degree of segregation that occurs within the
APB, the intensity of L1
2
atomic ordering is merely reduced but not
completely suppressed inside the several-atomic-layer-thick APB,
as can be seen from the double period contrast in the dark contrast
region in Fig. 8 (c). Because this residual ordering contrast is
observed along the full length of the APB, it is rather unlikely that
this phenomenon is a pure projection effect resulting from an in-
clined APB. Rather this observation indicates that the crystal lattice
in the broadened APBs still retains a certain degree of L1
2
atomic
ordering, even though the chemical composition approaches that
of the disordered
g
phase.
3.5. In situ heating experiment
In order to better understand the high temperature stability of
the planar defects, the structural evolution of an APB/SISF/APB
arrangement was investigated in situ during heating in the TEM
(Fig. 10). The planar defect conguration remains stable up to a
temperature of 850
C and retains its original structure that was
established during creep at 900
C. The APB/SISF/APB arrangement
is conrmed by corresponding BF and g(100) CDF images, as shown
in Fig. 10 (a) and (b). At 1000
C, Fig. 10 (c), the APB deviates from its
original (close to) {111} orientation and develops a step-like
structure consisting of short alternating segments of (100) and
(010) facets. This can be attributed to the expected anisotropic APB
energy showing a pronounced minimum for the {100} orientation
[20,38,39]. The checkered-like pattern covering the whole
g
0
pre-
cipitate is attributed to surface effects resulting from remains of the
sample preparation by ion milling or from local oxidation. Further
increasing temperature to 1050
C, Fig. 10 (d), the APB exhibits
lateral coarsening that eventually leads to complete fragmentation
of the
g
0
precipitate. The center of the APB has now visibly trans-
formed into the disordered
g
phase, bordered by two new
g
/
g
0
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349342
interfaces. It is worth noting that the equilibrium volume fraction of
the precipitates will decrease as the temperature increases and
approaches the solvus temperature of approximately 1140
C. It
should also be noted that the SISF does not show comparable
broadening nor does it act as preferred nucleation site for dis-
ordering. This demonstrates that it is the APBs and not the SISFs
that have a strong destabilization effect on the
g
/
g
0
microstructure
at elevated temperatures.
4. Discussion
The formation of stacking faults upon shear of <112>dislocation
ribbons in
g
0
hardened Ni-base superalloys was rst suggested by
Kear et al. [24] and plays an important role in the high stress and
low temperature creep regime. The nucleation of these ribbons as
well as the type of faults and dislocations were investigated in
detail by Rae et al. [22,23]. This deformation mechanism involves
the reaction of two 1/2<110 >{111} dislocations in the
g
channels
followed by dissociation into Shockley partials during penetration
of the
g
0
phase in accordance to:
1=2h101iþ1=2h011i/1=3h112iþ1=6h1 12i(1)
The leading 1/3[112] partial creates the SISF and the trailing 1/6
[112] partial remains at the
g
/
g
0
interface. Together, this constitutes
a 1/2[112] extended band of concentrated shear. However, it was
shown that in order to shear through a
g
0
precipitate, the rst shear
band has to be followed by a second band of the same character
which reestablishes the previously faulted
g
0
phase. This leads to
the following overall arrangement of dislocations and planar faults:
1/3[112] þSISF þ1/6[1 12] þAPB þ1/6[112] þSESF þ1/3[112].
It should be mentioned that in this notation the whole shear
band moves to the left, i.e. the 1/3[1 12] dislocation on the left
(right) corresponds to the leading (trailing) partial dislocation of
Fig. 9. HRSTEM images (top) and corresponding EDXS element maps (below) of a SISF, an APB and a narrow
g
channel in between
g
0precipitates, obtained in edge on orientation in
the areas marked by red, blue and black boxes in Fig. 7 (a). The elemental maps and the one-dimensional concentration proles below clearly reveal accumulation and depletion of
alloying elements in the vicinity of the planar faults and in the
g
channel with respect to the surrounding
g
0phase. The concentration prole across the
g
channel (bottom right)
conrms that Co and Cr partition to the
g
phase whereas Ni, Al and the heavy refractory elements W and Ta partition to the
g
0phase. The APB (middle) shows the same partitioning
behavior indicating that the local composition at the APB approaches that of the disordered
g
phase. This is in agreement with the similar Z contrast of APB and
g
channel (both
appearing dark) in the STEM images. The SISF shows less pronounced and partly different segregation behavior. Like for the APB and the
g
channel Co and Cr are slightly enriched
whereas Ni and Al are slightly depleted at the SISF with respect to the surrounding
g
0phase. However, the heavy refractory elements W and Ta are not depleted along the planar
fault but are slightly enriched (W) or show almost no segregation (Ta). This is likely the reason for the bright Z contrast of the SISF in the HRSTEM imagesofFig. 8 and 9.
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349 343
the shear band. Separated in the center, this <112>shearing
mechanism comprises a leading half and a trailing half ribbon of
type 1/2<112 >with an APB in between each half ribbon.
4.1. The 1/2<112 >shear mechanism
In the present analysis we investigated the elementary dislo-
cation processes which create the observed APB/SISF/APB
arrangement of faults in a CoNi superalloy creep deformed at high
temperatures (900
C) and low stresses (310 MPa). We found that
analogies exist to the classical <11 2 >shearing mechanism
described above. However, instead of two 1/2<112 >bands only one
such band shears through the
g
/
g
0
microstructure, leaving behind
an APB that extends across contiguous
g
0
precipitates. In addition,
and most interestingly, isolated SISF loops were observed in many
of these precipitates on the {111} slip plane. The SISFs do not span
across the full precipitate but remain embedded in the APB, which
tends to deviate from the {111} plane towards the energetically
favorable {100} orientation. With these results, we propose the
following dislocation shear mechanism which is illustrated in
Fig. 5:
(1) Two closely coupled Shockley partials, a leading 1/3[1 12]
and a trailing 1/6[112], approach the
g
0
precipitate on a (111)
plane.
1
Instead, a complex dislocation core dissociation likely
occurs [24]. The pair of Shockley partials likely formed by a
reaction of two 1/2<110 >(111)
g
matrix channel dislocations
with different Burgers vectors, 1/2[101] and 1/2[011], as
proposed by Rae et al. [23] (cf. Eq. (1), above). However, we
do not have direct experimental evidence for this.
(2) Upon entering the
g
0
precipitate, the leading 1/3[112] partial
creates a SISF on the glide plane. The trailing partial is left at
the interface, cutting in at a later stage.
(3) Once the trailing partial dislocation penetrates the
g
0
pre-
cipitate, it begins to transform the SISF into an APB.
(4) While the leading partial shears through the entire
g
0
pre-
cipitate, the trailing partial begins to form a loop embedded
in the APB. Simultaneously, the APB migrates from the (111)
plane towards energetically favorable {100} planes (not
shown in Fig. 5, cf. Fig. 13).
(5) The trailing partial closes by reaction with itself (like in an
Orowan looping process) and forms a complete loop on the
(111) plane, which is fully contained in the
g
0
phase and
surrounded by an APB that has now partially migrated onto a
{100} plane. This arrangement of SISF loops that are fully
embedded in APBs is experimentally observed in Figs. 2 and
3and extends over multiple, contiguous
g
0
precipitates as
shown by Figs. 1e3.
The schematic in the bottom right of Fig. 5 shows the
arrangement of Burgers vectors involved in the shearing mech-
anism. In order conrm this mechanism, evidence of a leading
1/3<112 >Shockley partial dislocation was sought. As the TEM
sample is very thin, observing this conguration is very rare and
only one instance was observed during analysis from a [001]
sample cut perpendicular to the tensile axis. Fig. 11 (a) shows a
BF image taken under a two beam condition using g(200). Here
the stacking fault is visible lying in the center of the
g
0
precipi-
tate. Comparison with the corresponding CDF image using the
superlattice reection g(100) (Fig. 11 (b)) shows that an APB is
connected to the SISF, however only from one side. This conrms
that the leading partial has not yet cut through the entire
g
0
precipitate, as in step (3) of our model in Fig. 5. The BF image in
Fig. 11 (c) obtained under a two beam condition for g(131) (close
to the [013] zone axis) reveals the two partial dislocations
bounding the SISF. Under these imaging conditions, both planar
faults are invisible. The stronger contrast of the leading partial
(with Burgers vector b
1
¼1/3[112]) compared to the trailing
partial (with Burgers vector b
2
¼1/6[112]) can be attributed to
the different magnitude of the Burgers vectors resulting in
g.b
1
¼2andg.b
2
¼1, respectively. Fig. 11 (d) serves as illustration
for a better understanding of the experimentally observed defect
arrangement in Fig. 11 (a) e(c). In order to directly conrm the
identical sign but different magnitudes of the Burgers vectors of
the leading partial terminating the SISF and the trailing partial
separating the SISF from the APB the unique capabilities of
LACBED have been exploited once more, as presented in Fig. 12.
BF- as well as DF-LACBED was performed for both dislocations
using the identical g(600) Bragg line. The Bragg line splits twice
for the leading partial and only once for the trailing partial, as
schematically visualized on the right of Fig. 12. A careful analysis
of the sign of the Burgers vectors indicated by the direction of
twisting of the Bragg line when approaching the respective
dislocation shows that g.b
1
¼þ2andg.b
2
¼þ1 for the choice of
the dislocation line directions indicated in the schematics on the
right of Fig. 12 (compare the corresponding analysis of the
dislocation loop in the Supporting Information). Thus the
diffraction contrast analysis (Fig. 11) in combination with LACBED
(Fig. 12) unambiguously shows that the leading and trailing
partials possess Burgers vectors b
1
¼þ1/3[1 12] and b
2
¼þ1/6
Fig. 10. In situ TEM heating experiment showing the structural evolution of an APB/
SISF/APB ribbon with increasing temperature. Blue arrows mark the APB and red ar-
rows mark the SISF. At 850 C (a) and (b) the planar faults are still in their original state
after creep. The APB/SISF/APB arrangement of faults is conrmed by corresponding BF
(a) and g(100) CDF (b) images revealing only the SISF and both, SISF and APBs,
respectively. (c) At 1000 C the APB shows pronounced faceting into energetically
favorable {100} facets. The chess-like pattern covering the
g
0phase is attributed to
surface effects. (d) With increasing temperature, at 1050 C, the APB shows lateral
broadening indicating that transformation of the ordered
g
0phase into the disordered
g
phase preferentially takes place at APBs eventually leading to complete disintegra-
tion of the
g
/
g
0microstructure. (For interpretation of the references to colour in this
gure legend, the reader is referred to the web version of this article.)
1
The leading partial with Burgers vector along [1 12] cannot have a magnitude of
1/6 because this would cause two adjacent atomic (111) planes to lie exactly on top
of each other, which is not possible (see schematic in the bottom right of Fig. 5).
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349344
[112], respectively.
As expected, the glide forces acting on the dislocations as a
result of the externally applied load (resolved shear stress) point in
the forward direction (see SI). In summary, our experimental ob-
servations clearly demonstrate that the observed fault congura-
tion forms upon 1/2[112] shear. Additionally, the mechanism is not
only active in individual
g
0
precipitates, but generally occurs in
many contiguous
g
0
precipitates along the (111) slip plane.
4.2. Balance of forces acting on dislocation loop
It is worth considering the stability of the loop conguration, as
observed in Figs. 2 and 3 and the conditions under which it is
supposed to shrink and annihilate. In order to answer this question
we consider the glide forces (per unit length of the dislocation)
acting on the dislocation loop (see Fig. 5, step 5):
1. The glide force resulting from the resolved shear stress is F
t
¼
t
res
$b, and, as a rst order approximation, the dislocation line
tension F
l
¼Gb
2
=Racts towards the center of the loop, which
supports loop shrinkage and annihilation, as illustrated by the
two black arrows F
t
and F
l
in Fig. 5 step 5.
2. The force acting in the opposite direction must originate from a
difference in planar fault energies, such that the APB energy,
g
APB
, is larger than the SISF energy,
g
SISF
, resulting in a net force
F
f
¼
g
APB
g
SISF
, depicted as the opposite pointing black arrow F
f
in
Fig. 5 step 5.
At force balance, F
l
þF
t
¼F
f
, a critical radius of curvature can be
derived as:
R
0
¼G$b
2
ð
g
APB
g
SISF
t
res
$bÞ(2)
where Gis the shear modulus, bthe magnitude of the Burgers
vector b
loop
of the loop, and
t
res
the resolved shear stress for the
experimentally evaluated glide system [112](111). We note here
that even though force balance is achieved, a loop of radius R
0
constitutes an unstable conguration as the only force that de-
pends on the loop radius R, the line tension F
l
¼Gb
2
=R, increases
with decreasing R. This means that SISF loops with radii R>R
0
are
expected to expand whereas loops with radii R<R
0
shrink and
annihilate.
The minimum radius that we observed for SISF loops in our
sample was approximately 50 nm. This means that the critical
radius R
0
<R
min
z50 nm as, otherwise, such loops would immedi-
ately shrink and annihilate. Based on that we can derive a lower
limit for the difference in planar fault energies according to
Dg
¼
g
APB
g
SISF
>G$b
2
R
min
þ
t
res
$b(3)
With R
min
z50 nm,G¼89 GPa [40e43],
t
res
¼0.47$
s
app
¼0.47$310 MPa ¼146 MPa and b¼0.147 nm,
Dg
can be estimated to
be at least 60 mJ/m
2
.
The question arises why SISF loops enter the
g
0
precipitates at all
and become surrounded by an APB of higher fault energy rather
than remaining at the interface. If the intersection of the (111) glide
plane with the precipitate is small enough such that a loop with
radius R>R
0
is larger than the precipitate (e.g. close to the corner of
Fig. 11. CTEM analysis of terminating planar faults in a 1/2[1 12] shear band, as pro-
posed in step (3) of the model in Fig. 5. The BF image obtained with g(200) in two
beam condition reveals the presence of an SISF in the center of the
g
0precipitate, not
reaching to the
g
/
g
0interface (a). The corresponding CDF image with g(100) shows the
existence of an APB only on one side of the SISF (b). The BF image taken under two
beam condition for g(131) close to the [013] zone axis (c) clearly reveals both Shockley
partials (arrows), however with the upper partial (1.) showing stronger contrast than
the lower one (2.) as expected from the twice as large Burgers vector. For a better
understanding the illustration (d) summarizes the conguration of dislocations and
planar faults observed in the experiment.
Fig. 12. LACBED analysis of the two partial dislocations investigated by conventional
TEM in Fig. 11. (a) Leading partial: Two splittings of the g(600) Bragg line are observed
in BF- and DF-LACBED. Application of the Cherns and Preston [29] rule results in
g$b
1
¼þ2 which is consistent with a Burgers vector þ1/3[112]. (b) Trailing partial: The
same Bragg line g(600) reveals only one splitting upon crossing the dislocation line
resulting in g$b
2
¼þ1 which corresponds to a Burgers vector þ1/6[1 12]. The sche-
matics on the right illustrate for both cases (a) and (b) the Bragg line splitting with
respect to the sign of the excitation error s and the dislocation line direction u.
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349 345
the
g
0
cube), the applied stress pushes the dislocation into the
precipitate, but line tension forces the SISF loop to immediately
shrink and annihilate. Indeed, we occasionally observed
g
0
pre-
cipitates along a 1/2[112] shear band showing only an APB without
a SISF loop.
Let us now assume that the shear band intersects the
g
0
pre-
cipitate further away from the corner resulting in a larger inter-
section which provides enough space to t in a loop of radius R>R
0
.
Line tension still forces the loop to cut in at the edges of the pre-
cipitate where otherwise small radii of curvature would result.
However, after cutting in at the edges there is no further driving
force for separating the partial from the
g
/
g
0
interface. However,
this is in contrast to our observation that SISF loops are always
embedded in the
g
0
precipitate fully surrounded by an APB. Our
microscopic observations that (1) APBs migrate towards energeti-
cally favorable {100} orientations and (2) APBs as well as SISFs
show pronounced segregation of alloying elements directly indi-
cate that local reordering and atomic diffusion of alloying elements
at/along planar faults take place during creep. Driven by energy
minimization such processes act towards decreasing the planar
fault energies. In order to explain the frequent occurrence of SISFs
in CoNi superalloys Titus et al. [35] proposed, based on experi-
mental observations and complementary atomistic simulations,
that the energy of SISFs is considerably reduced by segregation of
alloying elements accompanying shear, corresponding to a dis-
placiveediffusive L1
2
/D0
19
phase transformation. In the present
work we conrmed the experimental ndings but observed an
even more pronounced segregation of alloying elements at APBs
which may be described as wettingof the APBs by the
g
phase.
Wetting of APBs by the disordered phase is expected to be driven by
a strong reduction of the APB energy, as discussed by Kikuchi et al.
and other authors [20,44e48]. Therefore, the loop will begin to
shrink only when the difference between the APB and SISF energies
decreases such that the expanding force arising from the formation
of a high energy APB is decreased to a level comparable to that of
the current shrinkage force from the dislocation loop with radius R.
Thus gradual shrinkage of the loop is expected to be governed by
atomic diffusion and reordering processes preferentially behind the
trailing partial constituting the loop until the loop radius ap-
proaches R
0
, below which the loop contracts and annihilates by
pure glide. Within this scenario the observed APB/SISF/APB
conguration, whereby SISF loops are embedded in an APB (Figs. 2
and 3), corresponds to a snapshot of a transient phenomenon
which eventually results in a conguration with only APBs and no
SISFs. We would like to point out that the force balance discussed
above does not consider all possible forces. Additional forces are
expected to result from internal stresses due to the
g
/
g
0
lattice
mist and due to channel dislocations introduced upon creep.
4.3. Diffusion assisted 1/2<11 2 >{111} shear
To better understand the diffusion processes involved in the 1/2
[112]{111} shearing mechanism (described in section 4.1), Fig. 13
provides a schematic model of the same shear process as in
Fig. 5. However, here the planardefects are oriented edge on, as it is
the case in a [110] projection (see Fig. 8 (a). In the following we
comment on the diffusion-assisted shear process with respect to
Fig. 13:
(1) The leading and trailing partial dislocation (red and blue)
approach the
g
0
precipitate on a {111} glide plane (dashed
black line).
Fig. 13. Schematic model of elementary diffusion processes that are proposed to take place durin g/after cutting of
g
0precipitates by a 1/2[1 12] ribbon to explain the observed
segregation of alloying elements at SISFs and APBs (Fig. 8) and the coarsening of APBs in in situ heating experiments (Fig. 9). For details on the cutting process the reader is referred
to Fig. 5 which illustrates the process in a [111] projection. The moderate segregation of Cr, Co and W and depletion of Ni and Al at SISFs, also observed by Titus et al.[35], is proposed
to result (mainly) from diffusional exchange within the
g
0precipitate whereas the pronounced segregation of
g
channel elementsCr and Co as well as depletion of
g
0elements
Ni, Al, W and Ta at APBs requires extensive diffusional exchange with the
g
channel. While the SISF stays on its (111) slip plane and develops a constant (chemical) thickness of a few
atomic layers (cf. Fig. 8 and 9) the APB segments migrate towards energetically favorable {100} orientations and broaden as result of diffusional exchange with the
g
channel. In
other words, the APBs, rst created by glide, enable the
g
phase to enter and spread into the
g
0particles eventually leading to disintegration and segmentation of the
g
/
g
0
microstructure as observed in the in situ heating experiment (Fig. 9).
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349346
(2) The leading partial enters the precipitate by forming a SISF
on the {111} plane (red) while the trailing partial remains at
the
g
/
g
0
interface. Penetration of the leading partial requires
a sufciently high resolved shear stress to generate a SISF,
which is assisted by the repulsive force from the trailing
partial. Segregation of alloying elements at the SISF as pro-
posed by Titus et al. [35] and consistently detected by EDXS
in the present work (Fig. 9) gradually decreases the SISF
energy and facilitates further penetration of the leading
partial.
(3) Due to line tension the trailing partial enters the
g
0
precipi-
tate preferentially at edges of the
g
0
cube (cf. Fig. 5, steps (2)
and (3)) even if the energy of the freshly formed APB is
higher than that of the SISF. The APB habit plane immediately
begins to migrate (by atomic reordering) towards the {100}
planes, whereby reducing its energy in two ways: rst, by
reduction of the total APB area and, secondly, by minimiza-
tion of the specic (anisotropic) APB energy, which, for many
L1
2
ordered structures, is known to be lowest for {100}
orientation [20,38,39].
(4) Further reduction of the APB energy is achieved by pro-
nounced segregation of
g
channel elements Cr and Co as well
as depletion of
g
0
forming elements Ni, Al, W and Ta. The
extensive diffusional exchange with the
g
channel is
accompanied by lateral broadening of the APB as observed by
HRSTEM (Fig. 8 (c)) and chemical analysis (Fig. 9, center
column). Kikuchi and Cahn [20] proposed that in two phase
alloys the disordered phase will coat all APBs near the
congruent temperature. In this regard they introduced the
term of perfect wettingof the APB by the disordered phase.
By analogy in our case, perfect wetting of the APB would
occur by the
g
phase. Owing to the reduction of the APB
energy, glide of the trailing partial can proceed step by step,
always controlled by the migration and diffusion processes
which reduce the defect energy. Once the leading partial
exits the
g
0
precipitate, the remaining trailing partial, now
fully surrounding the precipitate (see step 4 of Fig. 5), forms a
closed loop (by mutual annihilation of dislocation segments
with opposite line direction) which enters the precipitate
from all sides and shrinks according to the mechanism
described above.
(5) The nal conguration is achieved as experimentally
observed in Fig. 8. On both sides of the SISF, the APB has
partially migrated to {100} planes and has coarsened.
The proposed mechanism is supported by the in situ heating
experiment, Fig. 10. At elevated temperatures, the APB reduces its
energy by migrating towards the {100} orientation. Its energy is
further reduced by wetting the APB with the disordered
g
phase,
which leads to APB broadening. Finally, complete fragmentation of
the
g
0
precipitates occurs at high temperatures after an extended
period of time. In the in situ experiment, temperatures larger than
the creep temperature have been used in order to make these
processes visible. C. Leroux et al. [26] showed that the L1
2
/A1
transformation in L1
2
ordered Co
30
Pt
70
alloys near the congruent
temperature is mainly achieved by wetting of the APB by the
disordered A1 fcc phase. Our microscopic analysis of the APB by
EDX and HRSTEM indicates that not only the local composition of
the APB adopts the composition of the disordered
g
phase, but also
the degree of ordering is reduced in this region (even though not
fully suppressed, see Figs. 8 and 9). This suggests that upon
coarsening the APB gradually develops into the
g
phase, both
structurally and chemically. In summary, we have experimentally
observed two mechanisms for lowering the APB energy: First, the
migration to {100} planes, and, secondly, the wetting by and,
eventually, transformation into the
g
phase. In contrast to the APB,
the SISF remains stable inside the
g
0
phase during the in situ heating
experiment. The splitting of the
g
0
precipitate clearly originates
from the APBs, which act as preferred nucleation site for the
g
phase, rather than from the SISF, indicating that the SISF contrib-
utes less to weakening and destabilization of the
g
0
phase.
After 1/3<112>shear, the local structure of the SISF is an or-
dered hexagonal structure, D0
19
. Combined with the detected local
composition along the fault, Titus et al. proposed a displacive-
diffusive phase transformation, whereby the SISF assists the crea-
tion of the hexagonal Co
3
W phase [23] which stands in excellent
agreement with our ndings. Only recently T.M. Smith et al. [37]
investigated the local structure and chemical composition of
SESFs and of their related dislocations in creep deformed Ni-base
disk superalloys. Similar to Titus et al. they proposed a shear
induced phase transformation of L1
2
which in their case transforms
into the D0
24
h
phase. Additionally they observed a solute atmo-
sphere of
g
forming elements around the pair of leading partial
dislocations forming the SESF, indicating that faults trailing such
partials would correspond to simpleintrinsic or extrinsic stacking
faults in a local
g
phase environment. Unlike faults in the perfect
L1
2
structure such faults do not show nearest-neighbor violations
and, therefore, possess lower energies [37]. Both authors proposed
that the movement of the leading partial may be constrained by the
diffusive motion of the solute atmosphere, similar to the classical
Cottrell cloud, and that this mechanism is the rate limiting process
during creep deformation [35,37]. Taking into account their results,
our investigated deformation mechanism, see Figs. 5 and 13, may
suggest a solute atmosphere in the vicinity of the trailing Shockley
partial dislocation differing from the original
g
0
phase composition.
In fact, because the trailing dislocation glides on the habit plane of
the SISF, it locally encounters the D0
19
crystal structure and
chemistry along the SISF (enrichment of W, Co, Cr, depletion of Ni,
Al) rather than entering the undisturbed
g
0
phase. This probably
reduces, but denitely alters, the energy required for penetrating
the
g
0
precipitate by the trailing partial in the rst place. In other
words, following the concept of displacive-diffusive phase trans-
formation ([35,49,50]), our investigated shear mechanism com-
prises initially a L1
2
/D0
19
transformation followed by a D0
19
/
A1 transformation which destabilizes the
g
0
phase.
4.4. Comparison of shear mechanisms in Ni-, CoNi- and Co-base
superalloys
The elementary shear mechanisms of
g
0
precipitates in tensile
[001] creep at temperatures exceeding 850
C have been compared
for Ni-, CoNi- and Co-base single crystal superalloys in ([42,51]). In
Ni-base superalloys shearing of the
g
0
phase is dominated by APB-
coupled 1/2<101>matrix dislocations [2,26,52]. In contrast, for
CoNi-base superalloys crossing of
g
0
precipitates by single matrix
dislocations associated with the formation of APBs has been
observed to be active [10]. In Co-base superalloys 1/3<112 >partial
dislocations produced by reaction of two dissimilar 1/2<101 >
matrix dislocations cross the precipitates creating a SISF which is
terminated by a 1/6<112 >trailing partial left behind at the
g
/
g
0
interface [42]. The new 1/2<112 >shearing mechanism in CoNi-
base superalloys identied in this work can be viewed as an
extension of the mechanism observed in Co-base superalloys.
Assisted by local reordering and chemical segregation the
1/6<112 >trailing partials enter the
g
0
precipitates as dislocation
loops gradually transforming the SISF into an APB. Eventually this
may result in only APBs being present as in the case of shearing by
single 1/2<110 >matrix dislocations. However, we emphasize that
the APBs produced by these two mechanisms are different, as are
the forces acting on the dislocations and the efciencies by which
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349 347
the two shearing mechanisms contribute to elongation of the
sample and global strain. While the displacement vectors of APBs
produced by a single 1/2<101>matrix dislocation are inclined from
the tensile axis by 45
the one produced by 1/2<112>shear is
perpendicular to the tensile axis (see supporting information for
details). Thus the two mechanisms can be discriminated even if all
the dislocations are absent from the planar defects and only APBs
remain inside the precipitates. Finally, the Schmid factor is much
higher for the 1/2<112 >shearing mechanism as the Burgers vec-
tors involved, 1/3<112 >and 1/6<112>have an optimum alignment
in the glide plane with respect to the load axis. Correspondingly,
the contribution to global strain in loading direction is much higher
as well, namely a factor of four per APB produced.
Concerning the relative importance of the two mechanisms,
1/2<101>shear and 1/2<11 2 >shear, observed in CoNi-base su-
peralloys, additional quantitative work is required, including a
statistical evaluation of the APBs left behind in the sample, as a
function of strain. It is worth noting that the two mechanisms for
APB formation are competitive, in terms of the processes that occur
in the matrix prior to insertion of dislocations into the precipitate.
Based on the Schmid factor and contribution to creep, 1/2<112>
shear is expected to be dominant. However, it requires a sufcient
density of (dissimilar) matrix dislocations to form the partial dis-
locations in the rst place. On the other hand, 1/2<101 >shear could
reduce the density of
g
channel matrix dislocations, thus reducing
the probability of reactions between dissimilar matrix dislocations.
Finally, we emphasize again the similarities between the
1/2<112 >shearing mechanism observed in the present work for a
CoNi superalloy under conditions of high temperature and low
stresses and the <112 >mechanism observed earlier [23] for Ni-base
superalloys in the low temperature and high stress regime. So far
there is no indication for the CoNi superalloys that deformation
occurs by formation of<112 >ribbons that extend across multiple
precipitates; this process results in high levels of creep strain in Ni-
base alloys. Obviously, the lower APB energy in CoNi superalloys
facilitates the formation of individual 1/2<112 >half ribbons by
reducing the driving force for re-establishing the order by a second
1/2<112 >half ribbon.
5. Conclusions
This study investigated an interesting new deformation mech-
anism dominating the
g
/
g
0
microstructure of single crystal CoNi-
base superalloys creep deformed at high temperatures (900
C)
and low stresses (310 MPa). The major conclusions are:
Planar defects in numerous isolated as well as contiguous
g
0
precipitates on {111} planes reveal a characteristic APB/SISF/APB
conguration whereby superlattice intrinsic stacking fault (SISF)
loops are fully embedded within an antiphase boundary (APB)
that typically extends over multiple contiguous
g
0
precipitates.
The pure shear process creating the APB/SISF/APB conguration
involves a leading Shockley partial of type 1/3[112] forming the
SISF followed by a trailing Shockley partial of type 1/6[1 12]
forming the APB. Thus, in total the shear band constitutes a
crystallographic slip of type 1/2[112] with residual dislocation
loops of type 1/6[112] inside the
g
0
precipitates.
A simple force balance calculation indicates that the initial APB
energy (without chemical redistribution, see below) must be
larger than the SISF energy by at least 60 mJ =m
2
. Below a critical
radius, dislocation loops are unstable due to line tension being
dominant whereas loops with larger radii would be expected to
expand.
The local chemistry and crystallography of both planar faults
differ from the surrounding
g
0
phase composition. Elemental
segregation of Co, Cr, W and depletion of Ni, Al was found along
the SISF which locally adopts the Co
3
W-(D0
19
) structure, in
agreement with [32]. Distinct enrichment of Co and Cr, deple-
tion of Al, Ni, W, Ta, and a (partially) disordered A1 structure
were observed in the vicinity of the APB. Thus the APB locally
approaches the structure and composition of the
g
phase, which
may be interpreted as wettingof APBs by the disordered phase
[20,53]. Indeed, the changes in chemical composition of the APB
is much more pronounced (compared to the SISF) indicating
that prominent atomic diffusion of alloying elements from the
g
phase along the planar fault is involved, driven by reduction of
the APB energy.
The APB originally formed on a {111} plane further reduces its
energy by migrating towards {100} planes, as explicitly
conrmed by in situ TEM heating experiments. The atomic
diffusion processes involved in the reduction of APB energy
through (1) wetting of the APB by the disordered
g
phase and (2)
migration of the APB plane towards energetically favorable
{100} orientations are proposed to be rate limiting for disloca-
tion loop formation resulting in the nal APB/SISF/APB cong-
uration observed in the precipitates.
Within the context of displacive-diffusive phase trans-
formations [35,54], this new shearing mechanism can be described
as L1
2
/D0
19
/A1 transformation that is associated with slip of
1/3[112] and 1/6[1 12] Shockley partial dislocations, which corre-
sponds to the formation of a SISF and its transformation into an
APB. In terms of destabilization of the
g
/
g
0
microstructure, the APBs
appear to aid in the degradation of the
g
/
g
0
microstructure at
elevated temperatures.
Acknowledgements
Y.M.E., J.M., and E.S. gratefully acknowledge nancial support by
the German Research Foundation (DFG) via the Collaborative
Research Center SFB/Transregio 103 From Atoms to Turbine Blades
ea Scientic Approach for Developing the Next Generation of
Single Crystal Superalloys(project A7), the Research Training
Group GRK1229 Stable and Metastable Multiphase Systems at
High Temperatureand the Cluster of Excellence EXC315 Engi-
neering of Advanced Materials. M.S.T., A.S., and T.M.P. gratefully
acknowledge nancial support by the NSF DMREF Grant DMR
1534264. The authors thank C. Ophus (NCEM, Berkeley) for periodic
averaging of HRSTEM contrast along the SISF and APB planes
(Fig. 8).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.actamat.2016.03.077.
References
[1] J. Sato, T. Omori, K. Oikawa, I. Ohnuma, R. Kainuma, K. Ishida, Cobalt-base
high-temperature alloys, Science 312 (2006) 90e91.
[2] R.C. Reed, Superalloys, rst ed., Camebridge University Press, New York, 2006.
[3] M. F
ahrmann, J.G. Wolf, T.M. Pollock, The inuence of microstructure on the
measurement of mismatch in single-crystal Ni-base superalloys, Mater. Sci.
Eng. A 210 (1996) 8e15.
[4] H.-Y. Yan, V.A. Vorontsov, J. Coakley, N.G. Jones, H.J. Stone, D. Dye, Quaternary
alloying effects and the prospects for a new generation of Co-base superalloys,
in: Superalloys 2012, John Wiley &Sons, Inc., Hoboken, NJ, USA, 2012, pp.
705e714.
[5] K. Shinagawa, T. Omori, J. Sato, K. Oikawa, I. Ohnuma, R. Kainuma, K. Ishida,
Phase equilibria and microstructure on
g
0
phase in Co-Ni-Al-W system, Mater.
Trans. 49 (2008) 1474e1479.
[6] A. Suzuki, High-temperature strength and deformation of
g
/
g
0
two-phase
CoeAleW-base alloys, Acta Mater. 56 (2008) 1288e1297.
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349348
[7] F. Pyczak, A. Bauer, M. G
oken, S. Neumeier, U. Lorenz, M. Oehring, N. Schell,
A. Schreyer, A. Stark, F. Symanzik, Plastic deformation mechanisms in a crept
L1
2
hardened Co-base superalloy, Mater. Sci. Eng. A 571 (2013) 13e18.
[8] M.S. Titus, A. Suzuki, T.M. Pollock, Creep and directional coarsening in single
crystals of new
g
e
g
0
cobalt-base alloys, Scr. Mater. 66 (2012) 574e577.
[9] M.S. Titus, A. Suzuki, T.M. Pollock, High temperature creep of new L1
2
-con-
taining cobalt-base superalloys, in: Superalloys 2012, 2012, pp. 823e832.
[10] Y.M. Eggeler, M.S. Titus, A. Suzuki, T.M. Pollock, Creep deformation-induced
antiphase boundaries in L1
2
-containing single-crystal cobalt-base superal-
loys, Acta Mater. 77 (2014) 352e359.
[11] A. Kostka, G. M
alzer, G. Eggeler, A. Dlouhy, S. Reese, T. Mack, L1
2
-phase cutting
during high temperature and low stress creep of a Re-containing Ni-base
single crystal superalloy, J. Mater. Sci. 42 (2007) 3951e3957.
[12] P.J. Phillips, M.C. Brandes, M.J. Mills, M. De Graef, Diffraction contrast STEM of
dislocations: imaging and simulations, Ultramicroscopy 111 (2011)
1483e1487.
[13] M.J. Hÿch, L. Potez, Geometric phase analysis of high-resolution electron mi-
croscopy images of antiphase domains: example Cu
3
Au, Philos. Mag. A 76
(1997) 1119e1138.
[14] A.J. D'Alfonso, B. Freitag, D. Klenov, L.J. Allen, Atomic-resolution chemical
mapping using energy-dispersive x-ray spectroscopy, Phys. Rev. B 81 (2010)
100101.
[15] L.J. Allen, A.J. D'Alfonso, B. Freitag, D.O. Klenov, Chemical mapping at atomic
resolution using energy-dispersive x-ray spectroscopy, MRS Bull. 37 (2012)
47e52.
[16] M.-W. Chu, S.C. Liou, C.-P. Chang, F.-S. Choa, C.H. Chen, Emergent chemical
mapping at atomic-column resolution by energy-dispersive x-ray spectros-
copy in an aberration-corrected electron microscope, Phys. Rev. Lett. 104
(2010) 196101.
[17] P. Beauchamp, J. Douin, P. Veyssi
ere, Dependence of the antiphase boundary
energy upon orientation in the L1
2
structure, Philos. Mag. A 55 (1987)
565e581.
[18] N. Baluc, R. Sch
aublin, K.J. Hemker, Methods for determining precise values of
antiphase boundary energies in Ni
3
Al, Philos. Mag. Lett. 64 (1991) 327e334.
[19] N.L. Okamoto, T. Oohashi, H. Adachi, K. Kishida, H. Inui, P. Veyssi
ere, Plastic
deformation of polycrystals of Co
3
(Al,W) with the L1
2
structure, Philos. Mag.
91 (2011) 3667e3684.
[20] R. Kikuchi, J.W. Cahn, Theory of interphase and Antiphase boundary in F.C.C.
alloys, Acta Metall. 27 (1979) 1337e1353.
[21] B.H. Kear, J.M. Oblak, A.F. Giamei, Stacking faults in gamma prime Ni
3
(Al,Ti )
precipitation hardened nickel-base alloys, Metall. Trans. I (1970) 2477e2486.
[22] C.M.F. Rae, N. Matan, R.C. Reed, The role of stacking fault shear in the primary
creep of [001] -oriented single crystal superalloys at 750
C and 750 MPa,
Mater. Sci. Eng. A 300 (2001) 125e134.
[23] C.M.F. Rae, R.C. Reed, Primary creep in single crystal superalloys: origins,
mechanisms and effects, Acta Mater 55 (2007) 1067e1081.
[24] B.H. Kear, A.F. Giamei, J.M. Oblak, On the origin of stacking faults in plastically
deformed Ni
3
Al (gamma'-Phase), Scr. Metall. 4 (1970) 567e574.
[25] V. a. Vorontsov, C. Shen, Y. Wang, D. Dye, C. M. F. Rae, Shearing of
g
0
pre-
cipitates by a112dislocation ribbons in Ni-base superalloys: a phase eld
approach, Acta Mater. 58 (2010) 4110e4119.
[26] J. Müller, G. Eggeler, E. Spiecker, On the identication of superdislocations in
the
g
0
-phase of single-crystal Ni-base superalloys ean application of the
LACBED method to complex microstructures, Acta Mater. 87 (2015) 34e44.
[27] J.P. Morniroli, R.K.W. Marceau, S.P. Ringer, New developments in the char-
acterization of dislocation loops from LACBED patterns, J. Microsc. 223 (2006)
246e248.
[28] C. J
ager, E. Spiecker, J.P. Morniroli, W. J
ager, Analysis of dislocation loops by
means of large-angle convergent beam electron diffraction, J. Phys. Condens.
Matter 14 (2002) 12777e12782.
[29] D. Cherns, a R. Preston, Convergent beam diffraction studies of interfaces,
defects, and multilayers, J. Electron Microsc. Tech. 13 (1989) 111e122.
[30] J.P. Morniroli, Large Angle Convergent Beam Electron Diffraction, Institute of
Physics Publishing, Paris, 2004.
[31] M. Tanaka, M. Terauchi, T. Kaneyama, Identication of lattice defects by
convergent-beam electron diffraction, J. Electron Microsc. 40 (1991) 211e220.
[32] J.P. Morniroli, R.K.W. Marceau, S.P. Ringer, L. Boulanger, LACBED character-
ization of dislocation loops, Philos. Mag. 86 (2006) 4883e4900.
[33] M.J. Hÿtch, E. Snoeck, R. Kilaas, Quantitative measurement of displacement
and strain elds from HREM micrographs, Ultramicroscopy 74 (1998)
131e146.
[34] K.J. Hemker, M.J. Mills, Measurements of antiphase boundary and complex
stacking fault energies in binary and B-doped Ni
3
Al using TEM, Philos. Mag. A
68 (1993) 305e324.
[35] M.S. Titus, A. Mottura, G. Babu Viswanathan, A. Suzuki, M.J. Mills, T.M. Pollock,
High resolution energy dispersive spectroscopy mapping of planar defects in
L1
2
-containing Co-base superalloys, Acta Mater. 89 (2015) 423e437.
[36] G.B. Viswanathan, R. Shi, A. Genc, V. a. Vorontsov, L. Kovarik, C.M.F. Rae,
M.J. Mills, Segregation at stacking faults within the
g
0
phase of two Ni-base
superalloys following intermediate temperature creep, Scr. Mater. 94 (2015)
5e8.
[37] T.M. Smith, B.D. Esser, N. Antolin, G.B. Viswanathan, T. Hanlon, A. Wessman,
D. Mourer, W. Windl, D.W. McComb, M.J. Mills, Segregation and
h
phase
formation along stacking faults during creep at intermediate temperatures in
a Ni-based superalloy, Acta Mater. 100 (2015) 19e31.
[38] J.A. Horton, C.T. Liu, Anisotropic antiphase boundaries in rapidly solidied
Ni
3
Al, Acta Metall. 33 (1985) 2191e2198.
[39] R. Yang, G.A. Botton, R.W. Cahn, The role of grain-boundary disorder in the
generation and growth of antiphase domains during recrystallization of cold-
rolled Cu
3
Au, Acta Mater. 44 (1996) 3869e3880.
[40] K. Tanaka, T. Ohashi, K. Kishida, H. Inui, Single-crystal elastic constants of
Co
3
(Al,W) with the L1
2
structure, Appl. Phys. Lett. 91 (2007) 181907.
[41] W. Walter Milligan, D. Stephen Antolovich, The mechanisms and temperature
dependence of superlattice stacking fault formation in the single-crystal su-
peralloy PWA 1480, Metall. Trans. A 22 (1991) 2309e2318.
[42] M.S. Titus, Y.M. Eggeler, A. Suzuki, T.M. Pollock, Creep-induced planar defects
in L1
2
-containing Co- and CoNi-base single-crystal superalloys, Acta Mater. 82
(2015) 530e539.
[43] D. Sieb
orger, H. Knake, U. Glatzel, Temperature dependence of the elastic
moduli of the nickel-base superalloy CMSX-4 and its isolated phases, Mater.
Sci. Eng. A 298 (2001) 26e33.
[44] C. Leroux, A. Loiseau, M.C. Cadeville, D. Broddin, G. Van Tendeloo, Order-
disorder transformation in Co
30
Pt
70
alloy: evidence of wetting from the
antiphase boundaries, J. Phys. Condens. Matter 2 (1990) 3479e3495.
[45] L. Potez, A. Loiseau, High-resolution electron microscopy studies of antiphase
boundaries in Cu
3
Au, Interface Sci. 2 (1994) 91e103.
[46] M. Chandran, S.K. Sondhi, First-principle calculation of APB energy in Ni-based
binary and ternary alloys, Model. Simul. Mater. Sci. Eng. 19 (2011) 025008.
[47] Y. Le Bouar, A. Loiseau, A. Finel, F. Ducastelle, Wetting behavior in the Co-Pt
system, Phys. Rev. B 61 (2000) 3317e3326.
[48] P.R. Swann, W.R. Duff, R.M. Fisher, The electron metallography of ordering
reactions in Fe-Al alloys, Metall. Trans. 3 (1972) 413e423.
[49] Q.Z. Chen, D.M. Knowles, Mechanism of 112/3 slip initiation and anisotropy
of
g
0
phase in CMSX-4 during creep at 750
C and 750 MPa, Mater. Sci. Eng. A
356 (2003) 352e367.
[50] P.A. Carvalho, P.M. Bronsveld, B.J. Kooi, J.T.M. De Hosson, On the fcc/D0
19
transformation in CoeW alloys, Acta Mater. 50 (2002) 4511e4526.
[51] A. Suzuki, H. Inui, T.M. Pollock, L1
2
-Strengthened cobalt-base superalloys,
Annu. Rev. Mater. Res. 45 (2015) 345e368.
[52] T.M. Pollock, A.S. Argon, Creep resistance of CMSX-3 Nickel base superalloy
single crystals, Acta Metall. Mater. (1992) 1e30.
[53] S.M. Allen, J.W. Cahn, A microscopic theory for antiphase boundary motion
and its application to antiphase domain coarsening, Acta Metall. 27 (1979)
1085e1095.
[54] Q.Z. Chen, A.H.W. Ngan, B.J. Duggan, in-situ TEM study of the L1
2
to DO
19
phase transformation in the intermetallic compound Fe
3
Ge, Intermetallics 6
(1998) 105e114.
Y.M. Eggeler et al. / Acta Materialia 113 (2016) 335e349 349
... Some of these alloys can achieve similar creep strengths as 1st generation single-crystalline Ni-base superalloys [2][3][4][5]. However, in contrast to Ni-base superalloys, the γ' precipitates in these novel alloys are already sheared in the early creep stages by various shearing mechanisms [6][7][8][9][10][11]. The resulting defect structures consist of either extended anti phase boundaries (APB) created by single superpartial dislocations [8,9,12], superlattice intrinsic and/or extrinsic stacking faults (SISFs and SESFs) [6,9,11,13], the APB-SISF-APB configuration [10,11], microtwinning [11,14,15] or a combination of multiple mechanisms [11]. ...
... However, in contrast to Ni-base superalloys, the γ' precipitates in these novel alloys are already sheared in the early creep stages by various shearing mechanisms [6][7][8][9][10][11]. The resulting defect structures consist of either extended anti phase boundaries (APB) created by single superpartial dislocations [8,9,12], superlattice intrinsic and/or extrinsic stacking faults (SISFs and SESFs) [6,9,11,13], the APB-SISF-APB configuration [10,11], microtwinning [11,14,15] or a combination of multiple mechanisms [11]. While the nucleation mechanisms of these defect structures differ considerably, segregation of alloying elements to these defects and their leading dislocation(s) is required in all cases to lower the intrinsic planar defect energies and facilitate the propagation of the leading dislocation(s) through the γ' precipitates [7,10,14,[16][17][18]. ...
... The resulting defect structures consist of either extended anti phase boundaries (APB) created by single superpartial dislocations [8,9,12], superlattice intrinsic and/or extrinsic stacking faults (SISFs and SESFs) [6,9,11,13], the APB-SISF-APB configuration [10,11], microtwinning [11,14,15] or a combination of multiple mechanisms [11]. While the nucleation mechanisms of these defect structures differ considerably, segregation of alloying elements to these defects and their leading dislocation(s) is required in all cases to lower the intrinsic planar defect energies and facilitate the propagation of the leading dislocation(s) through the γ' precipitates [7,10,14,[16][17][18]. ...
Article
Stacking fault shearing is commonly observed in Co-base and CoNi-base superalloys in the intermediate temperature regime between 850 °C to 1000 °C in contrast to Ni-base superalloys. In this work, we report on the significance of recovery processes after aging and creep experiments on predeformed specimens with extreme stacking fault densities. Above 950 °C, superlattice stacking faults start to recover rapidly, whereby intrinsic ones recover more quickly than extrinsic ones. Our results also reveal that the significance of stacking fault shearing in the high temperature creep regime could be higher than that indicated by traditional postmortem investigations due to recovery processes.
... Furthermore, extensive stacking fault formation and frequent interactions between these planar defects also cause high work hardening rates [3,11] and this is also assumed to cause strengthening during creep [13][14][15][16][17][18]. Since stacking fault shearing is a segregation-assisted process [14,[19][20][21][22][23][24][25][26], these strengthening effects, however, are diminished if the temperature becomes too high as their propagation velocity increases and extrinsic stacking faults evolve into microtwins. ...
Article
Full-text available
The defect evolution associated with an anomalous work hardening behavior of a single crystalline quaternary Co-Al-W-Ta superalloy at 950 °C was investigated by transmission electron microscopy. As plastic deformation is initially confined to the γ matrix channels, a plateau arises in the stress-strain curve after yielding. At about 1% plastic strain, extensive shearing of the γ′ precipitates under superlattice stacking fault formation occurs leading to extreme work hardening rates up to 12 GPa and a total increase in stress of about 200 MPa. Additional investigations on the temperature and strain-rate dependence of the anomalous work hardening behavior reveal the significance of diffusion and segregation processes on the stress-strain curve and the work hardening behavior.
... Additionally, the energy barrier for cross-slip at the primary plane adjacent to that of CA should be smaller than elsewhere because it transforms the APB into a SISF of lower energy [37], facilitating this mechanism at this particular location. The concerns of Sass et al. [6,24] regarding the character of the dislocation that cross-slips can also be addressed in this scenario. ...
Article
Full-text available
Superlattice intrinsic stacking faults (SISF) are the main culprit for the low temperature creep deformation of modern nickel-based superalloys used in jet engines. While these faults were identified over fifty years ago, their nucleation mechanism remains unclear. This work provides the first ever experimental evidence, via transmission electron microscopy, of a SISF nucleating from a cross-slip event in a polycrystalline alloy. Such an instance was identified in a grain with a near-〈001〉 tensile loading orientation. In the nucleation mechanism proposed, cross-slip allows the two dissimilar a2〈110〉 dislocations required to form a SISF to meet on adjacent planes at a precipitate interface. The concept of a nascent fault is introduced: the initial stacking fault that forms on a crystallographic plane and the dislocations of which continue to form coplanar faults as they glide away. This nucleation mechanism and the subsequent dislocation evolution are detailed taking into consideration the shear stresses on the individual Shockley partials and the full dislocations involved, as well as the stress orientation dependence of the energy barrier for cross-slip. These findings will guide future characterisation efforts in the field and inform the modelling of more realistic predictive models of creep behaviour.
... With boron addition, these alloys also outperform Haynes-188, a solid-solution and carbide strengthened commercial cobalt-base superalloy [23,24]. Detailed studies revealing the deformation mechanisms and the arising defect structures within these alloys (both in polycrystalline and single-crystalline forms) are also available [28,[41][42][43][44][45][46][47]. ...
Article
While boron is added to many structural alloys primarily as a grain-boundary strengthener, it has myriad other implications on the microstructure and thermophysical properties of these alloys. In the present study, we report these effects by varying boron addition in the range of 0.02at.% to 0.8at.% on the microstructure, thermo-physical properties and the creep resistance of tungsten free, γ/γ′ based Co–30Ni–10Al–5Mo–2Ta–2Ti (at.%) polycrystalline superalloy. Without boron addition, the plastic strain during creep at 850 °C is primarily accommodated through grain-boundary decohesion, often enhanced by severe oxygen attack along the grain boundaries. The addition of boron alleviates the poor grain-boundary cohesivity while inhibiting internal oxidation. An increase in the solvus temperature with boron addition was found, leading to an increase in the high-temperature capability of this alloy (∼40 °C). However, a concurrent decrease in the alloy's melting point has been observed. Coupled with improved grain-boundary cohesivity and larger grain size, improvement in creep resistance by almost an order of magnitude could be achieved. While at low amounts of boron (0.02–0.08 at.%) borides do not form, indicating solid solubility of boron within the bulk alloy, at high boron contents (0.2–0.8 at.%), Co and Mo rich boride forms primarily along the grain boundaries. These borides are surrounded by a distribution of fine secondary γ′ precipitates around them. In contrast, borides in Co–Al–W alloys promote precipitate free zones (PDZ), causing deterioration of mechanical properties. The beneficial effect of boron along grain-boundaries, both in elemental form and in the form of borides, has been highlighted in the present study which paves way for optimizing boron addition in other alloys of this class.
Article
TO further understand the creep performance of γ′-strengthened Co-based superalloys at low temperatures, the creep behaviors and mechanisms of a Co-Al-W-based and a CoNi-based single crystal superalloys (alloys 5Co and 7CoNi) were investigated at 1033 K/800 MPa and 1033 K/550 MPa based on their yield stresses, respectively. The creep behavior of alloy 5Co is characterized by forming a slow acceleration creep stage instead of a second deceleration creep stage in alloy 7CoNi after the rapid increase of creep rate in the acceleration creep stages of both alloys. The rapid shearing of multiple γ′ precipitates by the leading partial dislocations of individual stacking faults (SFs) in alloy 5Co are responsible for the rapid increase of creep rate, while the formations of anti-phase boundary (APB)-coupled dislocation pairs and SF-APB-SF structures in γ′ precipitates dominate the acceleration creep stage in alloy 7CoNi. The appearance of the slow acceleration creep stage in alloy 5Co can be ascribed to the generation of dislocations in the γ matrix and γ/γ′ interface. In contrast, the interactions of individual SFs, APBs and SF-APB-SF structures in alloy 7CoNi decrease the creep rate in the second deceleration creep stage. In both alloys, the W segregation at the SF improves the creep resistance by preventing the subsequent formation of APB in the same {111} plane, and contributes to the dissociation of the superdislocation(s) of APB into SF-APB-SF structures in the γ′ precipitates of alloy 7CoNi. This work will be beneficial for understanding the creep mechanism in different temperatures and further compositional optimization of γ′-strengthened Co-based superalloys.
Chapter
In recent times, applying correlative experimental techniques using electron microscopy (EM) and atom probe tomography (APT) to interpret several materials science problems have shown promising results. Especially, getting structural and three-dimensional compositional information from the same region of interest on a material led to understanding mechanisms occurring at the atomic scale. Direct integration of these mechanisms with material properties can open avenues for new material development. Some recent correlative experimental methods and protocols will be discussed in this chapter with relevant examples.KeywordsElectron-backscattered diffraction (EBSD)Transmission Kikuchi diffraction (TKD)Electron channeling contrast imaging (ECCI)Electron microscopy (EM)Atom probe tomography (APT)
Chapter
High-temperature technology is of major importance in many industrial applications, such as aircraft engines and land-based power generation gas turbines. Demands for high fuel and engine efficiency require the increasing service temperature of superalloys. The further development of widely used Ni-base superalloys hits a bottleneck due to the limitation of the melting point of Ni. In 2006, the discovery of γ/γ′ Co–Al–W alloys began a new era in the development history of high-temperature materials. Compared with Ni-base superalloys, the higher melting temperature by 50~150 °C, the greater creep resistance, and the comparative mechanical properties spotlight the research of novel γ/γ′ Co–Al–W-based alloys as one of the candidates for high-temperature materials for future generations of advanced propulsion systems. Despite the extraordinary improvement achieved in the various aspects of novel γ/γ′ Co–Al–W-based superalloys, several drawbacks still restrict the wide applications, such as metastable nature of γ′ precipitates, narrow γ/γ′ composition range, overhigh mass density, inferior medium–low-temperature strength. In this chapter, we review the current exploration of γ′-strengthened Co-base superalloys in view of these drawbacks, including phase stability of γ′ precipitates, γ′-solvus temperature, development of low-density γ/γ′ Co-base superalloys and CoNi-base superalloys, and high-temperature mechanical capability of γ/γ′ Co-base superalloys. Finally, the challenges and future research needs in the development of novel Co-base superalloys are prospected.KeywordsCo-based superalloysPhase stabilitySolvus temperatureLow-densityMechanical properties
Article
A novel phenomenon, dynamic localized phase transformation (LPT) at stacking faults during deformation, has been observed in various Ni-base and Co-base superalloys and found to play a critical role in determining the creep performance of these alloys. In this article, we review recent experiment observations of LPT, thermodynamic analysis on the fundamentals of LPT, first-principles calculations to link LPT to mechanical deformation, and computational tools and databases required to predict LPT. We discussed the generality of the LPT phenomenon, various challenges in quantitative predictions of LPT, LPT-strengthening and LPT-softening, and opportunities offered by the LPT mechanism in the design of the next generation of LPT-strengthened superalloys.Graphical abstract
Article
The content and the existing form of refractory elements significantly affect the performance in service of superalloys, which has gained extensive attention in this field. In this study, basic influence rule of diverse Ta and Ti contents on the microstructure and properties of the novel γ′ phase strengthened Co–Ni-based superalloys is clarified by thermodynamic calculation and experimental research. Under the premise of a total Ta and Ti content of 5 at%, increasing Ta content elevates the solvus temperature of γ′ phase and volume fraction of Laves phase, and also reduces the volume fraction of β phase in the as-cast alloys. Moreover, the average size of γ′ phase goes down and the coarsening of γ′ phase is inhibited simultaneously in the alloy after aging treatment. Consequently, increasing the content of Ta can enhance the microstructure stability and elevates the Vickers hardness and yield strength of the alloy. However, the density and cost of the alloy have to be raised. Meanwhile, the variation of Ta and Ti contents affects the oxidation performance of the alloys by influence the volatilization of W and Mo oxides. 2Ta3Ti and 3Ta2Ti alloys possess favorable oxidation resistance at 1050 °C. Considering the effect of Ta and Ti elements and normal performance requirements of Ni-based superalloys, 3Ta2Ti alloy has more impressive engineering application prospects. This work provides a solid foundation for optimizing the composition and improving the properties of Co–Ni-based superalloys.
Article
The L12-strengthened heterostructure Co40Cr20Ni30Mo2Al4Ti4 (at%) high-entropy alloys (HEAs) exhibit excellent tensile mechanical properties over a wide temperature range. In particular, the yield strength of this alloy reaches up to about 1400 MPa from room temperature to 600 °C, which is much higher than that of most alloys. The reasons for the different mechanical behaviors of this alloy above and below the transformation temperature are discussed. Unique super-lattice stacking faults appear in the high-temperature deformed alloys, which may be attributed to the decrease of super-lattice stacking fault energy when the temperature increases. In the present work, the deformation mechanisms of precipitation-strengthened HEAs at high temperatures are preliminarily explored, which has guiding significance for the development of advanced high-temperature structural materials.
Article
Full-text available
In this paper, the local compositional and structural changes occurring in association with stacking faults in a Ni-base superalloy are characterized and related to the possible rate-controlling processes during creep deformation at intermediate temperatures. These rate-controlling processes are not presently understood. In order to promote stacking fault shearing, compression creep tests on specially prepared single crystals of an exploratory Ni-base superalloy were conducted at 760 °C in the [0 0 1] orientation. Scanning transmission electron microscopy (STEM) imaging was coupled with state-of-the-art energy dispersive X-ray (EDX) spectroscopy to reveal for the first time an ordered compositional variation along the extrinsic faults inside the γ′ precipitates, and a distinct solute atmosphere surrounding the leading partial dislocations. The local structure and chemistry at the extrinsic fault is consistent with the η phase, a D024 hexagonal structure. Density Functional Theory (DFT) and high angle annular dark field (HAADF)-STEM image simulations are consistent with local η phase formation and indicate that a displacive-diffusive transformation occurs dynamically during deformation. © 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Article
Full-text available
Scanning transmission electron microscopy (STEM) at atomic resolution is in widespread use, Z-contrast imaging and electron energy-loss spectroscopy (EELS) being well established. We discuss an extension of STEM to obtain atomic-resolution elemental maps using energy-dispersive x-ray (EDX) spectroscopy. The first such results were published by D'Alfonso and co-workers [1] followed by the results in Ref. [2]. There has since been considerable progress from an experimental point of view [3,4]. Detectors enabling count rates an order of magnitude better than those used in the earlier work have been developed and installed on microscopes which are corrected for spherical aberration C s , allowing for the use of finer probes and higher currents. Here, first principles calculations are presented which confirm and facilitate the interpretation of such experimental maps. Simulations allow us to discuss issues such as probe channelling and spreading, the delocalization of the images and the contribution from thermal scattering (electrons which have excited phonons). Figure 1 shows a comparison of experimental and calculated elemental maps taken on SrTiO 3 orientated along <001> [4]. The data were obtained on an FEI Titan G2 equipped with a Super-X detector. The probe C s -corrector and a high brightness XFEG gun were operated at an accelerating voltage of 200 kV. The probe semi-angle was around 20 mrad and the beam current of the order of 100 pA. Maps were acquired on a 64×64 pixel array using iterative drift correction and with a 25 µs dwell time. The sum of the Sr L shell and K shell, Ti K shell and O K shell signals were monitored. A composite of the Sr and Ti maps is shown in Fig. 1(a). The top half of the figure shows raw experimental data. At the bottom left we show filtered chemical maps (using a moving average filter) and to the right of that a simulation is inlaid. The component Sr and Ti maps and the corresponding inlays are shown in Fig. 1(b) and Fig. 1(c) respectively. The results for oxygen are shown in Fig. 1(d). The averaged experimental result (bottom left) and the simulation (inlaid to the right of that) are seen once again to be in good qualitative agreement. However, it should be noted that there is considerably more signal measured on Ti/O columns than on pure O columns despite the linear density of oxygen atoms in the two types of atomic column being the same. This apparently anomalous behaviour has also been noted in electron energy-loss spectroscopy (EELS) [5,6] but can be understood using image simulations carried out from first principles. Simulations were performed using a model for thermal diffuse scattering based on a Born-Oppenheimer (BO) type approximation [7] and extended to incorporate ionization. This model is a quantum-mechanical approach to solving the many-body Schrödinger equation which models the distribution of both elastically scattered electrons and those which have been inelastically scattered (perhaps multiple times) due to the excitation of a phonon. It is assumed that the cross section for x-ray emission, which occurs when a hole is filled post ionization, is simply proportional to the cross section for inner-shell ionization in EELS where a detector spanning the whole solid angle is assumed and all energy losses above threshold are taken into account. The contribution to the ionization cross section for a particular edge from both elastically and thermally scattered electrons can then be calculated. Simulated elemental maps of the oxygen K-shell signal in <001> SrTiO 3 across a unit cell are shown in Figure 2. The total signal is shown in (a), the signal for ionization by elastically scattered electrons in (b) and that due to electrons which have first been thermally scattered is in (c). The increased signal on the Ti/O column is explained by considering the contribution from the thermally scattered electrons. Further examples of EDX elemental mapping will be presented and a brief comparison between elemental maps obtained in EDX and EELS will be given, illustrated by data taken simultaneously for the case of <001> SrTiO 3 .. [4] LJ Allen,