In Situ Straining Experiments: Examples (Vesely 1968, Furubayashi 1969, Saka et al.

1972,
Louchet et al. 1979). Screw dislocations were indeed
of Results observed moving as very straight segments at a low
and constant velocity. On the contrary, nonscrew ones
Several types of in situ straining experiments have appeared curved and much more mobile. Such a ki-
been developed in different scientific instruments (see netics has been ascribed to a Peierls mechanism,
In Situ Deformation Experiments: Technical Aspects). where only screw dislocations are subjected to a fric-
Examples described in this article concern in situ tion force on account of their core splitting along
straining experiments performed by the transmission three intersecting planes. Such a process has been de-
electron microscope to understand the mechanical scribed theoretically by Vitek (1966), Escaig (1968),
properties of crystalline materials. This technique is and Duesberry and Hirsch (1968) just a few years
indeed the oldest, the most advanced one, and that before in situ observations were made.
which has yielded the highest number of results. It is The first dislocation sources were also observed
thus the best illustration of the capability of in situ and described, and the influence of extrinsic obstacles
experiments. In the author’s opinion, other in situ such as ‘‘forest’’ dislocations intersecting the glide
approaches will be more and more advanced in future plane of moving dislocations was discussed.
in a spirit similar to that described here. More quantitative analyses were made later on com-
mercial 200 kV electron microscopes. The most com-
plete set of quantitative results has been obtained in
1. Peierls Forces magnesium single crystals deformed in a prismatic slip
(Couret and Caillard 1985a, 1985b). The orientation of
Peierls (1940) and Nabarro (1947) were the first ones the straining axis was chosen so as to inhibit easy glide
to point out that the energy of moving dislocations in the basal plane. Under such conditions, straight
has necessarily the same periodicity as that of the screw dislocations are seen to move viscously in the
crystal lattice, and that dislocation motion through prism plane (Fig. 1). This motion is however mixed with
such a potential must be submitted to some friction oscillations in the basal plane that yield the wavy
forces. These forces, commonly called ‘‘Peierls forc- slip traces depicted by arrows in Fig. 1 (Movie clip 1).
es,’’ have been extensively studied in metals, alloys, Oscillations are presumably due to an easy drift in the
and covalent materials such as semiconductors. They basal plane, under weak fluctuations of internal stress
are the strongest when dislocations are parallel to
dense crystallographic directions. Under such condi-
tions, dislocation motion takes place by a series of
jumps over interatomic distances between adjacent
positions of lowest energy, also called ‘‘Peierls val-
leys.’’ However, instead of overcoming intermediate
positions of maximum energy as a whole, which
would be too difficult a process, dislocations move by
nucleation of a small bulge till they reach the neigh-
boring Peierls valley. This bulge subsequently widens,
and transforms into a pair of opposite kinks moving
apart along the dislocation line, thus transferring the
whole dislocation in the adjacent valley. Difficult
kink-pair nucleation and easy kink motion give rise
to a slow motion of rectilinear segments parallel to
dense crystallographic directions.
The study of Peierls forces is very well adapted to
in situ experiments. Provided the stress is set to con-
venient values, dislocation movements are indeed
slow enough to be recorded, and surface effects are
usually fairly weak as friction forces are generally
much higher than image forces.

1.1 Metals
Many of the first in situ experiments were carried out
in body-centered cubic (b.c.c.) metals (Fe, Fe–3%Si,
Nb) strained at and below room temperature in a
high-voltage electron microscope (HVEM). They pro- Figure 1
vided the first direct evidence of a Peierls mechanism Slip traces in prism slip of Mg (Movie clip 1).

1
In Situ Straining Experiments: Examples of Results

Figure 2
Schematic representation of the oscillations between the basal and prismatic planes.

(Fig. 2). They yield important information because they strain rate that also depends on an unknown mobile
prove that dislocation cores remain spread in the basal dislocation density.
plane during the motion in the prism one. This spread-
ing, which is unfavorable for prism slip, is
at the origin of the Peierls force in this plane. This 1.2 Semiconductors
illustrates how observing the geometry and the kinetics A different kind of Peierls force controls the motion
of motions can yield information on dislocation core of dislocations in semiconductors. It exhibits many
structures and Peierls friction stresses. similarities but also several differences with respect to
Dislocations subjected to a Peierls stress are the- that in metals and alloys:
oretically predicted to move at a velocity: (i) Peierls valleys correspond to dislocation cores

 where covalent bonds are not much disturbed. Deep
Ekp ðtÞ
v ¼ nL exp
ð1Þ valleys are along screw and 601 directions in diamond-
kT
type cubic lattices.
(ii) The motion of kinks along dislocations is
where n is a vibration frequency (B1012 s–1) of the slower than in metals because it proceeds by cutting
dislocation line, L is the dislocation length, Ekp is and rotating covalent bonds.
the activation energy of kink-pair nucleation, t is the For a small dislocation length, the theoretical ex-
stress, k is the Boltzmann constant, and T is the pression for the velocity of dislocations is
temperature. In order to determine the parameters


used in this equation, dislocation velocities have been Ekp ðtÞ þ Em
measured as a function of L, t, and T. The disloca- vp tL exp

kT
tion velocity v has been plotted as a function of its

length L, at constant stress, in Fig. 3(a). It increases Ekp
LoX ¼ 2b exp ð4Þ
linearly with L, in excellent agreement with theory, 2kT
considering that L–L0 is the length of the screw seg-
ment. This confirms that the number of nucleation where b is the Burgers vector, and Em is the activation
sites of kink-pairs is actually proportional to the energy of kink motion along dislocations. For longer
length of the moving dislocation. Obviously, such a dislocations, the expression becomes
result could be obtained only by an in situ straining

ð1=2ÞEkp ðtÞ þ Em
experiment. vptb exp
with L4X ð5Þ
The microscopic activation area A which corre- kT
sponds to the area swept by the kink-pair at its saddle
position, that is, at the position that makes the In contrast with Ekp, which describes only the elastic
movement irreversible, as well as the microscopic ac- distortion of the crystal around dislocations, Em de-
tivation energy Ekp, have been measured at the scale scribes properties of covalent bonds in their core.
of individual dislocations (Fig. 3(b)), following the The plot of v(L) is expected to be linear for LoX
relations: and to reach a constant value for L4X. Thus, since
the experimental determination of v(L) yields values
kT @ln v
A¼ ð2Þ of Ekp þ Em, the experimental determination of X can
b @t yield values of Ekp and therefore Em, and gives in-
formation on the respective contributions of elastic
@ln v distortions around dislocations and the strength of
Ekp ¼ k ð3Þ
@ð1=TÞ covalent bonds in their core.
Dislocation velocities have been measured in III–V
The value obtained (A ¼ 9b2) is about half the value compounds such as GaAs and InSb (Louchet et al.
deduced from macroscopic tests. It is the most sig- 1993). In these non-centrosymmetrical structures, 601
nificant one because it relies on the direct measure- dislocations with opposite signs have different atoms
ment of the velocity of dislocations. On the contrary, in their cores (e.g., Ga or As). They are referred as
the macroscopic activation area is deduced from the 601a and 601b, respectively. Dislocation sources have

2
 In Situ Straining Experiments: Examples of Results

been observed with screw and 601b segments sub-

jected to high Peierls forces, and much more mobile
601a ones (Fig. 4 and Movie clip 2). The close values
0.2 of the velocities of screw and 601b dislocations in
III–V compounds are in fairly good agreement with
recent theoretical developments (Caillard and Martin
2003). The velocity of screw segments increases lin-
v (µm s−1)

early up to the maximum observable dislocation

lengths (3 mm in HVEM, from Fnaiech et al. 1987and
0.1 Caillard et al. 1989), which allows one to determine a
lower limit for X and Ekp, and an upper limit for Em.
Em is accordingly fairly small in agreement with a
weak bonding between type-III and type-V atoms
(Vanderschaeve et al. 2001).
Similar plots have been made in germanium by
two groups with substantially different results. One
L0 experiment yields a ratio of the velocities of screw and
(a) L (µm) 601 segments, vs/v60, to be B1.7–1.8, (Dupas 2003)
which is in fair agreement with the theoretical value
of 5/3 given by Caillard and Martin (2003). The other
yields the opposite results (Louchet et al. 1988).
T = 373 K There is a tendency for a saturation of the v(L)
curves toward a maximum value, especially for 601
segments, according to Louchet et al. (1988), whereas
the same plots are found linear by Dupas (2003).
More experiments are needed to determine values of
the saturation length X.
All these examples show that a complete set of re-
sults can be extracted from in situ experiments. As in
metals, the data can be compared with theoretical
estimates with a high degree of confidence because
T = 300 K they correspond to well-defined properties of indi-
ln v (µm s−1)

H
vidual dislocations, in contrast with macroscopic re-
sults that average various dislocation mechanisms
and include collective effects.
Results of in situ transmission electron microscopy
(TEM) experiments can be compared with results of
other in situ techniques, for example, double-etch pits
and X-ray topography. Usually, they correspond to
different ranges of dislocation lengths and velocities,
9b 2

so that they appear complementary rather than com-

A=

petitive. Double-etch pits must however be interpret-

ed carefully because the character of the moving
dislocations (screw/601) as well as their mode of mo-
tion cannot be controlled. For instance, double-etch
pits in III–V compounds yield a higher velocity for
601b dislocations than for screw ones (Yonenaga
1997), whereas these velocities are found similar in in
situ TEM, and simple theoretical considerations show
(b) τ (MPa) that they can only be equal or in the opposite ratio
(Caillard and Martin 2003). In situ X-ray topography
Figure 3 exhibits the same advantages as in situ TEM, but the
Microscopic measurements in Mg: (a) dislocation respective magnifications and resolutions differ by
velocity as a function of length and (b) dislocation three orders of magnitude (George et al. 1972). Fig. 5
velocity as a function of local stress and temperature. shows how velocities versus stresses measured by
TEM and XRT correspond to each other (Louchet
1981). The two curves join fairly well, which allow
one to determine activation areas related to individ-
ual dislocation motions over a wide stress range.

3
In Situ Straining Experiments: Examples of Results

Figure 4
Dislocation source in GaAs (Movie clip 2).

4
 In Situ Straining Experiments: Examples of Results

1
600 °C
615 °C

10−1 650 °C 540 °C

520 °C
v (µm s−1)

10−2

580 °C

10−3
550 °C 520 °C

10 100
τ (MPa)

Figure 5
Comparison between dislocation velocities in Si
measured in situ TEM and in situ XRT.

2. Influence of Local Chemistry on Glide

Mechanisms
In the preceding section, some aspects of the disloca-
tion mobility in crystalline materials have been de-
scribed. The present one is focused on the influence of
local chemical composition on this mobility. Two ef-
fects are presented. In the first case, fixed obstacles
corresponding to randomly distributed clusters are Figure 6
considered. On the contrary, the second one involves Ordinary dislocations in TiAl.
dynamic processes concerning both dislocations and
obstacles, which occur under the effect of temperature.
Examples, which are described in detail, concern hand, on the basis of the measurement of an increase
ordinary dislocations of the g phase of intermetallic of their density with increasing temperature, two
TiAl alloys. This g phase has an L10 ordered face- groups have initially concluded that they result from
centered tetragonal structure. The term ‘‘ordinary’’ intrinsic processes involving dislocation cross-slip
dislocation is commonly used to specify that their (Viguier et al. 1995, Sriram et al. 1997). The higher
Burgerspvector
ffiffiffi joins two neighboring atoms separat- the temperature, the easier the cross-slip and the
ed by a 2=2 (a is the lattice parameter) as in an f.c.c. larger the pinning-point number. On the other hand,
metal. As illustrated in Fig. 6, after deformation up these pinning points have been ascribed to small
to 600 1C, ordinary dislocations are elongated along extrinsic obstacles such as oxygen atoms or Al2O3
their screw orientation and anchored at many points. precipitates (Messerschmid et al. 1995, Morris 1996).

2.1 Anchoring Points and Glide Mechanism at Low

Temperatures (To400 1C) (a) Origin of the pinning points. In situ experiments
performed at room temperature by Halle and Tou-
The nature and origin of these pinning points have louse have produced a set of experiments leading to a
been the subject of strong controversies. On the one better understanding of these pinning points.

5
In Situ Straining Experiments: Examples of Results

(a) (b)

t=0s t = 15.44 s

(c) (d)

t = 16.44 s t = 22.24 s
Figure 7
Glide of ordinary dislocations in TiAl (Movie clip 3).

* Ordinary dislocations gliding in the same plane contrary, the observation of the anchoring of succes-
are anchored at the same place (Zghal et al. 1998). sive dislocations at the same place is in consistency
Figure 7 (Movie clip 3) shows the motion of two with extrinsic obstacles in nature, due to some chem-
ordinary dislocations marked 1 and 2 in a (1 1 1) ical heterogeneity. On the basis of atom probe anal-
plane. As pointed out by the black triangles (Fig. yses (Menand et al. 1996), they were attributed to the
7(d)), their slip traces appear almost similar indicat- segregation of part of the interstitial oxygen atoms
ing that their slip planes are identical or at least very which may be correlated to a local enrichment in ti-
close. The dynamic sequence indeed shows clearly tanium atoms and/or which may induce a small or-
that these two dislocations anchor at the same P1, P2, dered region (Zghal et al. 1998). This interpretation is
and P3 points. in good consistency with the decrease of the density
* Pinning points are also anchoring nonscrew or- of cusps and debris after annealing treatments that
dinary dislocations (Zghal et al. 1998, Haussler et al. precipitate interstitial atoms in excess (Gr!egori 1999).
1999) and superlattice dislocations (Haussler et al. Note that Furibayashi et al. (1969) have previously
1999). presented similar results and interpretations for Fe–
* In situ experiments have shown that Shockley 3% Si b.c.c. metals in which they have shown that the
dislocations moving in twins are also pinned at many probability for forming a kink at the obstacle is par-
points and that the overcoming processes are qual- ticularly high because an obstacle reduces the kink
itatively similar for these two types of dislocations. mobility.
Moreover, the linear densities of the pinning points
anchoring the Shockley dislocations and the ordinary
dislocations are of the same order of magnitude (b) Measurement of the anchoring strength. The
(Zghal et al. 1998). strength of pinning points can be evaluated from in
The Burgers vector of Shockley dislocations being situ experiments, provided individual overcoming
parallel to o1 1 24 directions, these dislocations are processes can be observed (Vivas et al. 1997). The
thus not able to cross-slip, and so, a cross-slip cannot last video frame before unpinning is taken as the
be at the origin of these pinning points. On the critical configuration at which the obstacle breaks,

6
 In Situ Straining Experiments: Examples of Results

and this strength is considered equal to the force ex- procedure, the elastic interaction between the two an-
erted by the two anchored segments at this instant. chored segments is not taken into account.
This method has been developed in the frame of the In the case of ordinary dislocations in TiAl alloys,
theoretical development proposed by Hazzledine the situation is simplified because we are always
et al. (1984), considering that clusters are pinpoint dealing with elongated screw dislocations moving
sized. The force exerted by the dislocation on each of in the direction perpendicular to their length, which
its pinned extremities is the resultant of two compo- makes the determination of the force orientation
nents (Fig. 8(a)). The first, E, tends to shorten the easier.
dislocation and is tangent to the curved dislocation. An example of such an analysis in a single g-phased
The second, E0 ¼ @E/@y, tends to rotate the disloca- Ti46Al54 alloy is displayed in Figs. 8(b) and 8(c). Fig-
tion toward the screw direction with lower energy ure 8(b) shows the last video frame before the unpin-
and is perpendicular to E. Here, E is given by ning of the dislocation. In this picture, the glide plane
is projected onto the picture plane in order to perform


 measurements with ease and better accuracy. Figure
mb2 sin2 y L
EðyÞ ¼ cos2 y þ ln ð6Þ 8(c) repeats the same picture on which geometrical
4p 1
n r construction and measurement are performed. b is the
direction of the Burgers vector. The first step consists
where b ( ¼ 0.28 nm) is the Burgers vector of ordinary in superimposing calculated dislocation loops in equi-
dislocations, m ( ¼ 65.2 GPa) is the shear modulus, n librium under an applied stress following the Wulff
( ¼ 0.236) is the Poisson’s ratio, y is the angle between construction proposed by Hazzledine et al. (1984).
the dislocation line and the Burgers vector, and L and r Then, it is easy to plot the tangents to the anchored
are the outer cut-off distance and inner cut-off radius, segments and to measure their angle with the Burgers
respectively, of the elastic strain field of the dislocation vector direction. For this example, y1 ¼ 251 and
(the value taken for L is the average distance between y2 ¼ 151, where F ¼ 2.74 nN. In the same Ti46Al54
pinning points, i.e., 0.15 mm, and that for r is b). In this alloy, 15 measurements have been performed leading

FT
F1
F F2
E E1
E2′ E1′ E2

1
E′
2
(a) (b)

2

D
L

1
bp
0.1 µm

(c) (d)

Figure 8
Analysis of the strength of obstacles.

7
In Situ Straining Experiments: Examples of Results

to an average value of 1.8 nN (Couret 1999). Since the based superalloys. Contrary to pinning points of TiAl
average separating distance between obstacles can al- alloys, in this case, dislocations are in front of diffuse
so be measured from in situ experiments (d ¼ 0.15 and continuous obstacles. Experiments have shown
mm), the resolved stress (t) corresponding to the glide that the deformation occurs by dislocation pile-ups.
of dislocations with shear obstacles of strength Careful analysis of dislocation distribution evidenced
F ¼ 1.8 nN can be calculated with the relation given that the two leading dislocations are paired. The for-
by Fleischer (1961): t ¼ F/db. This yields t ¼ 43 MPa. mation of diffuse antiphase boundaries (APBs) by
This value is significantly lower than the yield stress, leading dislocation is the driving force for this pair-
which indicates that the pinning points do not control ing. The size of short-range order domains has been
the ordinary dislocation glide. As previously observed evaluated from the number of paired dislocations and
in b.c.c. metals (Louchet et al. 1979), this glide mech- it has been found to be 0.7 nm, in agreement with
anism is indeed controlled by the frictional force diffuse neutron scattering experiments.
which is at the origin of the elongation of the dislo-
cation along their screw direction and which results
from the core spreading out of the glide plane (Sect. 1) 2.3 Dynamic Strain Aging
(Couret 2001). Note that this method was previously
used in the 6056 aluminum alloy (Vivas et al. 1997) In TiAl alloys, in situ experiments reveal a specific
and in that case, a good consistency was found be- behavior of ordinary dislocations at moderately high
tween the resolved stress calculated from in situ ex- temperatures (400–600 1C). As illustrated in Fig. 9,
periments and the macroscopic yield stress, which is the glide of dislocations occurs by burst with fast and
easily explained by the lack of friction force in {1 1 1} heterogeneous movements. Between views (a) and (b),
planes of aluminum alloys. the strain was increased and suddenly a new burst of
This kind of work based on in situ experiments dislocation was activated. Further increase of the
provides orders of magnitude of the obstacle strain does not induce any motion of locked disloca-
strength. Such an evaluation can only be obtained tions, but activates new bursts of freshly emitted dis-
by this method since it is absolutely necessary to di- locations. These dislocations appear elongated along
rectly observe dislocations overcoming obstacles. their screw orientation and anchored at many pinning
points, just as at room temperature. It follows that
at a microscopic scale, only dynamics allows discrim-
2.2 Local Order ination between room and high-temperature be-
haviors. Note that this specific high-temperature
In situ experiments also allow one to study short- behavior has been also evidenced by plastic instabil-
range order. One example can be found in the work ities measured during straining macroscopic tests at
of Pettinari et al. (2001, 2002) in the g-phase of nickel- the same temperature when loading axis is favorable

(a)
0.8 µm

(b)

Figure 9
Burst of ordinary dislocations in TiAl.

8
 In Situ Straining Experiments: Examples of Results

to the activation of ordinary dislocations (Gre! gori able or stress-stabilized configurations. Two exam-
1999). Finally, this high-temperature glide has been ples of such an approach are given below.
attributed to dynamic strain aging, for which the
velocity of dislocations is reduced by diffusion of sol-
ute atoms, and it has been correlated to yield stress 3.1 Metastable–Glissile Core Configurations
anomalies. The same behavior has been observed for
/1 1 1S dislocations in Fe70Al30 at 370 1C (Mol!enat Dislocations often have several possible core struc-
et al. 1997). tures. For instance, screw dislocations can dissociate
Our study is worthy of comparison with the in situ in two equivalent intersecting {1 1 1} planes in f.c.c.
results obtained by the German group since slightly metals, and in two equivalent configurations deduced
different observations have led to similar conclusions from each other by a 1801 rotation in b.c.c. metals
(Haussler et al. 1999).This group has indeed observed (Duesberry and Vitek 1998). Nonequivalent core
ordinary dislocations at 525 1C moving either in an structures can be found as well, the one of lower en-
unstable or a viscous manner, and has interpreted these ergy being stable and the others being metastable.
results by a diffusion-controlled formation of atmos- Direct evidence of a metastable core structure has
pheres of impurities, which causes a strain aging effect. been obtained for the first time in beryllium deformed
in situ in a prism slip (Couret and Caillard 1989a,
1989b). The experiment was similar to that described
3. Dislocation Core Structures under Stress previously in the case of magnesium (Sect. 1.1). In
the latter case, dislocations move by series of jumps
Dislocation structures observed in deformed samples between adjacent Peierls valleys, which results in a
by conventional post mortem TEM are always re- steady motion at the scale of TEM observations. In
laxed, namely, they have enough time to evolve to the beryllium, the same kind of motion takes place by
nearest local minimum of energy. Relaxed structures series of very fast jumps over variable distances
may then be substantially different from those con- B0.1 mm, separated by waiting times of B0.1–1 s.
tributing to plastic deformation. Figure 10 (Movie clip 4) is an example of such a jerky
‘‘Living’’ dislocation structures can be frozen using motion. The very fast jumps are imaged as double
several expedients such as precipitation during cool- contrasts corresponding to the superimposition of the
ing (Morris and Martin 1984) or irradiation under starting and final positions, as a result of the reman-
stress (Mugrabi 1968). However, observing plasticity ence of the fluorescent screen. Slip traces at the foil
in real time is certainly the best way to detect metast- surfaces (marked by black triangles in Fig. 11(a)) are

Figure 10
Jerky motion in Be (Movie clip 4).

9
In Situ Straining Experiments: Examples of Results

Tr. P

Y
Tr. B

]
00
0]
00

1
[1
[1
b

300 nm
0.01µm

(a) (b)
Figure 11
Steps on dislocations in Be (a) in situ with associated steps on traces and (b) post mortem.

made of straight segments, corresponding to alternate there is no chance to observe metastable configura-
slips in prism and basal planes. They result from the tions with such a short lifetime in relaxed specimens.
emergence of the steps seen on the dislocations Indirect proofs can however be found in bulk mate-
(marked by white triangles). They prove that slip is rials deformed in the same conditions by inspecting
stable in the prism plane during the jumps, and that the shape of nonscrew dislocations. Indeed, the latter
dislocations are dissociated in the basal plane—where are not smoothly curved, in contrast with those ob-
they tend to move under fluctuating internal stress- served in case of a Peierls mechanism, but made of
es—between the jumps. This behavior can be readily segments contained alternately in prism and basal
explained by the existence of a core structure extend- planes (Fig. 11(b) from Jonsson and Beurs (1987)).
ed in the prism plane, allowing for a fast motion with These segments are similar to those already observed
negligible friction force. However, since the corre- in situ, emerging at the foil surfaces and forming the
sponding flight times are very short, this configura- stair-shaped slip traces in Fig. 11(a). They are ac-
tion appears to be metastable. The stable–metastable cordingly the direct result of the screw motion de-
(sessile–glissile) transition takes place after a sub- scribed above. Note however that this interpretation
stantial waiting time with the help of stress and tem- would not have been given in the absence of in situ
perature, and the reverse metastable–stable (glissile– experiments showing the corresponding dislocation
sessile) transition takes place more rapidly with the dynamics.
help of temperature only. The same behavior has been observed in titanium
The frequencies of jump lengths and waiting times with different purities (Farenc et al 1995). The aver-
have been measured during deformation. They ex- age jump length has been measured as a function of
hibit exponential variations, which shows that dislo- temperature (Fig. 12). It decreases rapidly with in-
cations have constant probabilities of locking and creasing temperature, in agreement with an increas-
unlocking per unit time, as expected from ordinary ing probability of locking per unit time, and falls to
thermally activated processes. The overall dislocation interatomic distance at 370 1C. Above this tempera-
behavior has been modeled (Couret and Caillard ture, it reduces to the classical Peierls mechanism
1989a, 1989b, Caillard and Martin 2003). discussed in Sect.1.1. The transition between both
It is obvious that this mechanism could not be mechanisms has been modeled. It readily explains
identified by conventional observations, because discontinuities in the activation parameters of the

10
 In Situ Straining Experiments: Examples of Results

cross-slip of screw superdislocations onto the {1 0 0}

‘‘cube’’ plane.
250 Different types of Kear–Wilsdorf locks have been
observed: complete ones, dissociated in the {1 0 0}
200 cube cross-slip plane, and incomplete ones, almost
800 dissociated in the primary {1 1 1} ‘‘octahedral’’ plane,
A
with only a small extension in the cube plane. The

A (b )
2
yg 150
600 former were abundantly observed and studied by
conventional electron microscopy observations of re-
yg (nm)

100 laxed samples. The latter were reported much less

400
frequently; as a result, their role was considered un-
50 important for the mechanical properties of Ni3Al.
200
In situ experiments revealed that the density of in-
complete locks is fairly high under stress. Moreover,
0
0 100 200 300 400 500 600 700 their motion often appears as series of jumps over
T (K) distances scaling with their dissociation width, called
‘‘APB jumps’’ (Mole! nat and Caillard 1991; Fig. 13).
Figure 12 This motion produces steps on dislocations (just as
Jump length (and macroscopic activation area) as a those observed in beryllium in Fig. 10), of height
function of T in Ti. equal to this peculiar jump length (Fig. 13(a) and
Movie clip 5). Then, macroscopically deformed sam-
plasticity of titanium, measured in the same temper- ples were re-analyzed in order to detect these steps.
ature range (Farenc et al. 1995). Results showed that they do exist in a fairly high
Jerky motions of rectilinear screw and edge dislo- density (Couret et al. 1993; Fig. 13(c)). Their presence
cations have been observed during the in situ defor- escaped notice because the step height was not con-
mation of other metals and alloys, in particular in sidered as an important parameter. This observation
intermetallic alloys as TiAl (Couret 1999), Ni3Al de- has been considered as an indirect proof of the exist-
formed in cube slip (Mole! nat and Caillard 1992), and ence of the mechanism revealed by in situ experiments.
Fe3Al (Mol!enat et al. 1997). The existence of distinct The motion of incomplete locks has been subse-
glissile and sessile configurations with different ener- quently modeled, and their critical unlocking stress
gies has been confirmed by atomistic calculations in
Ti (Vitek and Igarashi 1991) and TiAl (Mahapatra
et al. 1995, Simmons et al., 1997).
In Ti3Al, in situ experiments have shown that dis-
locations in prism planes have different core properties
according to their local environment. Superdisloca-
tions can indeed glide either between dense planes
consisting of Ti atoms exclusively, or between planes
consisting of Ti and Al atoms. The different glide
planes can be identified easily because the correspond-
ing dissociation widths are different on account of
different APB energies. In situ experiments showed
that dislocations gliding in Ti planes behave as those
in pure titanium, whereas those gliding in Ti–Al planes
move much more easily on account of a single glissile
core structure (Legros et al. 1996a, 1996b). In situ
experiments also showed that, in contrast to previous
expectations, most dislocations do not move in the Ti
planes where they have the lowest energy (lower APB
energy) but they do so in the Ti–Al planes where their
mobility is the highest.

3.2 Incomplete Kear–Wilsdorf Locks in NI3Al

Ni3Al has been extensively studied during the last 30
years because of its surprisingly large increase of
strength with increasing temperature. This anoma- Figure 13
lous behavior is due to the thermally activated for- APB jumps (a) in situ and (b) post mortem (Movie
mation of the so-called ‘‘Kear–Wilsdorf’’ locks by clip 5).

11
In Situ Straining Experiments: Examples of Results

has been found to be directly related to the macro- The first in situ observations were carried out by
scopic strength of Ni3Al (Caillard 2001). Halle on AlPdMn single grains (Wollgarten et al.
As a conclusion, these two examples show how in 1995). Only high-temperature deformation can be in-
situ experiments can initiate substantial progress in vestigated by this method because of the high brit-
the understanding of dislocation behavior and me- tleness of quasicrystals. Perfect dislocations were
chanical properties. As dislocation structures can be observed to move viscously in fivefold, threefold, and
observed under stress and as a function of time, new twofold planes, often perpendicular to the straining
dislocation properties and new mechanisms can be axis. They have polygonal shapes with edges parallel
revealed, and their contribution to plastic deforma- to twofold directions. In fact, both the shape of dis-
tion can be estimated. Then, the characteristic features locations and their kinetics of motion are very similar
of these mechanisms can be determined, and their to those observed in elementary semiconductors (see
signature can be found in macroscopically deformed Sect. 1). The Burgers vectors were unfortunately not
(but relaxed) samples by conventional TEM observa- determined, which did not allow one to check the
tions. This ensures that in situ observations were not exact mode of motion (glide or climb).
disturbed by thin foil artifacts. The existence of such More recent post mortem observations in AlPdMn
new mechanisms can eventually be supported by at- deformed at intermediate temperatures showed that
omistic calculations. Lastly, they can be modeled and dislocation motion involves at least a large compo-
compared with macroscopic mechanical properties. nent of climb (Caillard et al. 2000, Mompiou et al.
2003). In fact, no evidence of glide could be obtained.
This analysis was possible because the plane of mo-
4. Quasicrystals tion was unambiguously identified as the phason
plane left in the wake of dislocations.
Quasicrystals are binary or ternary alloys with a very The same procedure cannot be used at high tem-
narrow composition range, in which the arrangement peratures, because the planes of motion are more
of atoms has no long-range periodicity, but exhibits a difficult to determine. Then, new in situ experiments
high degree of symmetry including the unusual have been carried out in order (i) to determine if the
fivefold rotational one. Among several unusual phys- dominant mode of dislocation motion is the same as
ical properties, they are highly brittle at low and at lower temperatures (i.e., climb) and (ii) to tenta-
intermediate temperatures, but become fairly ductile tively isolate pure glide (Mompiou et al. 2003). They
at high temperatures. confirmed the results of the first investigations by
The first TEM observations of deformed icosahe- the German group. The same extensive dislocation
dral AlPdMn samples revealed that plasticity is motion could be activated either in unaxial straining,
achieved by motion and multiplication of disloca- or under thermal stresses, by just heating the sample.
tions, as in crystals (Wollgarten et al. 1993). Two Contrast analyses could be performed on mobile dis-
types of dislocations have been observed as a function locations, showing that climb prevails as at interme-
of the deformation temperature: (i) imperfect ones, diate temperatures. The modes of motion are thus
trailing planar defects called ‘‘phason planes,’’ or fundamentally different in quasicrystals and in sem-
‘‘phason walls,’’ at intermediate temperature and (ii) iconductors, in spite of their high degree of resem-
perfect ones at high temperature. More details on the blance. Figure 14 shows an exceptional event which
structure and the contrast of these dislocations can be definitely proves that glide is much more difficult
found in the references cited (Wollgarten et al. 1991). than climb at high temperature (Mompiou et al.

(a) (b) (c) (d) (e)

d1
D
u1 u2

d2
tr. PA

T
g5a
T =740 °C, t =0 s 0.5 µm t =15 s t =19.5 s t = 27 s t =131 s

Figure 14
Dipole formation in AlPdMn.

12
 In Situ Straining Experiments: Examples of Results

2004). Two dislocations with opposite Burgers vec- dislocations, even if moving dislocations cannot be
tors (determined by contrast analysis) move toward directly observed.
each other in closely parallel fivefold planes. They are For plasticity studies, in situ experiments have al-
parallel to twofold directions and their motion takes ways to be combined with other experimental inves-
place by pure climb. Now, as they reach their min- tigations performed at various scales and with
imum distance, they rotate towards a pseudo-twofold theoretical approaches. Under such conditions, com-
direction and form a dipole which remains stable plementary information can be obtained, leading to a
during more than 30 mn. This behavior is very sur- better understanding of the investigated phenomena.
prising because annihilation should a priori take place As illustrated by the above examples, it should how-
under high attractive stress between the two opposite ever be emphasized that, in several cases, in situ re-
dislocations. In fact, annihilation is impossible be- sults have been at the origin of new explanations.
cause the plane which contains the dipole (a twofold Complementary in situ techniques using different
one) also contains the Burgers vector of the pair. In types of instruments as RX generators, scanning
other words, annihilation is impossible because it electron microscopes, or atomic force microscopes
would require dislocation motion by glide. This ex- have been developed to study mechanical properties.
periment shows that glide is at least 1000 times slower The main advantage of the former is the observation
than climb, under similar stresses. This behavior is at of bulk samples whereas its main limitation is the low
variance from that of crystals where dipoles usually dislocation density that can be observed. For the
form by glide and annihilate by climb. scanning and atomic force microscopies, only surfac-
This experiment has yielded an important result es can be imaged and so studies are reduced to the
that could not be obtained by other means. Indeed, it analysis of traces left by the moving dislocations. One
is always difficult to prove by ordinary static obser- of the reasons for the success of in situ straining ex-
vations that something does not exist: if only one periments in the TEM is probably the high capacity
observation is sufficient to prove that something does of this instrument for dislocation analysis. Converse-
exist, a large number of observations are usually ly, the advantage of this technique is to provide a
necessary to prove the reverse. The in situ experiment direct and unique observation of moving dislocations
reported above proves unambiguously that glide is under stress and temperature.
‘‘almost’’ impossible because the kinetics of the two
competing processes have been compared in the sim- See also: In Situ Deformation Experiments: Techni-
ple and well-defined situation of two interacting op- cal Aspects.
posite dislocations.
Bibliography
Caillard D Mat. Sci. Eng. A 74, 319–21
5. Concluding Remarks Caillard D, Martin J L 2003 In: Cahn R W (ed.) Thermally
Activated Mechanisms in Crystal Plasticity. Pergamon Mate-
In this article, results of straining in situ experiments
rials Series, Pergamon, Oxford
have been presented concerning several types of Caillard D, Cl!ement N, Couret A, Androussi Y, Lefebvre A,
materials, as metals, intermetallic alloys, aluminum Vandershaeve G 1989 Inst. Phys. Conf. Ser. 100, 403
alloys, superalloys, compound semiconductors, and Caillard D, Vanderschaeve G, Bresson L, Gratias L 2000 Phil.
quasicrystals. They have led to the study of various Mag. A 80, 237
elementary deformation mechanisms responsible for Couret A 1999 Intermetallics 9, 1977
the mechanical properties. Other examples concern- Couret A 2001 Phil. Mag. A 79, 1977
ing, for example, multiphased alloys are not present- Couret A, Caillard D 1985a Acta Metall. 33, 1447
ed here due to lack of space. It should however be Couret A, Caillard D 1985b Acta Metall. 33, 1455
emphasized that in situ experiments sometimes fail Couret A, Caillard D 1989a Phil. Mag. A 59, 783
Couret A, Caillard D 1989b Phil. Mag. A 59, 801
because of limitations, generally related to the mate- Couret A, Sun Y Q, Hirsch P B 1993 Phil. Mag. A 67, 29
rial properties. The characteristic dimensions of the Dupas C 2003 Ph.D. Thesis, EPFL
system to be studied can be larger than the observable Duesberry M S, Hirsch P B 1968 In: Rosenfield A R, Hahn G
thickness of the sample. This is the case for some T, Bement A L, Jaffee R I (eds.) Dislocation Dynamics.
composite materials containing very large fibers. A McGraw-Hill Book Company, New York, 57pp
fast sample degradation under combined effects of Duesberry M S, Vitek V 1998 Acta Metall. 46, 1481
temperature and vacuum can also limit the time of Escaig B 1968 Phys. Status Solidi 28, 463
observation. For instance, in situ experiments were Farenc S, Caillard D, Couret A 1995 Acta Metall. Mater. 43,
impossible in TiAl alloys above 800 1C. On the con- 3669
Fleisher R L 1961 Acta Metall. 9, 996
trary, as illustrated by the study of dynamic strain Fnaiech M, Reynaud F, Couret A, Caillard D 1987 Phil. Mag.
aging occurring in TiAl alloys at intermediate tem- A 55, 405
perature (Sect. 2), quick events can be studied Furubayashi E 1969 J. Phys. Soc. Japan 27, 130
because it is always possible to extract informa- George A, Escaravage C, Champier G, Schroter. W 1972 Phys.
tion from the analysis of slip traces and locked Status Solidi B 53, 483

13
In Situ Straining Experiments: Examples of Results

Gr!egori F 1999 Ph.D. Thesis, Universit!e de Paris VI Mugrabi H 1968 Phil. Mag. 18, 1211
Haussler D, Bartsch M, Aindow M, Jones I P, Messerschmidt Nabarro F R N 1947 Proc. Phys. Soc. 54, 256
U 1999 Phil. Mag. A 79, 1045 Peierls R E 1940 Proc. Phys. Soc. 52, 34
Hazzledine P M, Karnthaler H P, Korner A 1984 Phys. Status Pettinari F, Prem M, Krexner G, Caron O, Coujou A, Kirchner
Solidi 81, 473 H O K, Cl!ement N 2001 Acta Metall. Mater. 49, 2549
Jonsson S, Beuers J 1987 Mater. Sci. Eng. 91, 111 Pettinari F, Jouiad M, Kirchner H O K, Cl!ement N, Coujou A
Legros M, Couret A, Caillard D 1996a Phil. Mag. A 73, 61 2002 Phil. Mag. 82, 3045
Legros M, Couret A, Caillard D 1996b Phil. Mag. A 73, 81 Saka H, Doi M, Imura T 1972 J. Phys. Soc. Japan 29, 803
Louchet F 1981 Phil. Mag. 43, 1289 Simmons J P, Rao S I, Dimiduk D M 1997 Phil. Mag. A 75,
Louchet F, Kubin L P, Vesely D 1979 Phil. Mag. A 39, 433 1299
Louchet F, Cochet-Muchy D, Brechet Y, Pelissier J 1988 Phil. Sriram S, Dimiduk D M, Hazzledine P, Vasudevan V K 1997
Mag. A 57, 327 Phil. Mag. A 76, 965
Louchet F, P!elissier J, Caillard D, Peyrade J P, Levade C, Van- Vanderschaeve G, Levade C, Caillard D 2001 J. Microsc. 203,
derschaeve G 1993 Microsc. Microanal. Microstruct. 4, 199 72
Mahapatra R, Girshick A, Pope D P, Vitek V 1995 Scr. Metall. Vesely D 1968 Phys. Status Solidus 29, 675
Mater. 33, 1921 Viguier B, Hemker K J, Bonneville J, Louchet F, Martin J L
Menand A, Huguet A, N!erac-Partaix A 1996 Acta Metall. 1995 Phil. Mag. A 71, 1295
Mater. 44, 4729 Vitek V 1966 Phys. Status Solidus 18, 687
Messerschmidt U, Bartsch M, Haussler D, Aindow M, Hat- Vitek V, Igarashi M 1991 Phil. Mag. A 63, 1059
tenhauer R, Jones I P 1995 Mater. Res. Soc. Symp. Proc. 364, Vivas M, Lours P, Levaillant C, Couret A, Casanove M J,
47 Coujou A 1997 Phil. Mag. A 76, 921
Mol!enat G, Caillard D 1991 Phil. Mag. A 64, 1291 Wollgarten M, Gratias D, Zhang Z, Urban K 1991 Phil. Mag.
Mol!enat G, Caillard D 1992 Phil. Mag. A 65, 1327 A 64, 919
Mol!enat G, Couret A, Caillard D 1997 Mater. Sci. Eng. 234, Wollgarten M, Beyss M, Urban K, Liebertz H, Kocster U 1993
660 Phys. Rev. Lett. 71, 543
Mompiou F, Caillard D 2004 (to be published) Wollgarten M, Bartsch M, Messerschmidt U, Feuerbacher M,
Mompiou F, Bresson L, Cordier P, Caillard D 2003 Phil. Mag. Rosenfeld R, Beyss M, Urban K 1995 Phil. Mag. Lett. 71, 99
83, 3133 Yonenaga I 1997 J. Phys. III France 7, 1435
Mompiou F, Feuerbacher M, Caillard D 2003 Phil. Mag. (to Zghal S, Menand A, Couret A 1998 Acta Metall. Mater. 46,
appear) 5899
Morris M A 1996 Intermetallics 4, 417
Morris M A, Martin J L 1984 Acta Metall. 32, 1609 A. Couret and D. Caillard

Copyright r 2006 Elsevier Ltd.

All rights reserved. No part of this publication may be reproduced, stored in any retrieval system or transmitted
in any form or by any means: electronic, electrostatic, magnetic tape, mechanical, photocopying, recording or
otherwise, without permission in writing from the publishers.
Encyclopedia of Materials: Science and Technology
ISBN: 0-08-043152-6
pp. 1–14

14